Comparison of osmotic swelling influences on meniscal fibrocartilage and articular cartilage tissue mechanics in compression and shear

Authors

  • An M. Nguyen,

    1. Department of Mechanical Engineering, Stanford University, Stanford, California
    2. School of Medicine, Stanford University, Stanford, California
    Search for more papers by this author
  • Marc E. Levenston

    Corresponding author
    1. Department of Mechanical Engineering, Stanford University, Stanford, California
    • Department of Mechanical Engineering, Stanford University, Stanford, California. T: 650-723-9464; F: 650-725-1587.
    Search for more papers by this author

Abstract

Although the contribution of the circumferential collagen bundles to the anisotropic tensile stiffness of meniscal tissue has been well described, the implications of interactions between tissue components for other mechanical properties have not been as widely examined. This study compared the effects of the proteoglycan-associated osmotic swelling stress on meniscal fibrocartilage and articular cartilage (AC) mechanics by manipulating the osmotic environment and tissue compressive offset. Cylindrical samples were obtained from the menisci and AC of bovine stifles, equilibrated in phosphate-buffered saline solutions ranging from 0.1× to 10×, and tested in oscillatory torsional shear and unconfined compression. Biochemical analysis indicated that treatments and testing did not substantially alter tissue composition. Mechanical testing revealed tissue-specific responses to both increasing compressive offset and decreasing bath salinity. Most notably, reduced salinity dramatically increased the shear modulus of both axially and circumferentially oriented meniscal tissue explants to a much greater extent than for cartilage samples. Combined with previous studies, these findings suggest that meniscal proteoglycans have a distinct structural role, stabilizing, and stiffening the matrix surrounding the primary circumferential collagen bundles. © 2011 Orthopaedic Research Society Published by Wiley Periodicals, Inc. J Orthop Res 30:95–102, 2012

The knee menisci are compositionally, structurally, and mechanically inhomogeneous fibrocartilages that transmit substantial loads between the femur and tibia across widely varying activities.1, 2 Compared to articular cartilage (AC), meniscal fibrocartilage (MFC) has a lower water content (60–70% vs. 68–85%), higher collagen content (15–25% vs. 10–20% by mass) and lower proteoglycan content (1–2% vs. 5–10% by mass).3 Type I collagen is the predominant collagen type in the meniscus, and is primarily found in large, circumferentially oriented bundles2 that enhance tensile stiffness in the circumferential direction.4 The radially inner “compression” region is generally described as more hyaline-like, with relatively high contents of collagen II and the large proteoglycan aggrecan.2 Perlecan, decorin, biglycan, and other small proteoglycans have also been identified in the meniscus.5, 6 As in AC, an osmotic swelling stress arising from interactions between the negatively charged proteoglycans and the ionic interstitial fluid is thought to maintain tension in the collagen, thus contributing both directly and indirectly to meniscal tissue behavior in different stress states.7

While most descriptions of meniscal organization focus on the primary collagen architecture and macroscopic distribution of constituents, recent reports have more fully described the intricate, contiguous secondary fiber network (Fig. 1) that surrounds and subdivides the circumferential bundles.8 While the functional importance of this secondary network has not been widely studied, it appears to facilitate rearrangement of the primary collagen bundles under load.9 This secondary network is a compositionally distinct matrix compartment, containing colocalized collagens I, II, and VI,10, 11 as well as aggrecan and other proteoglycans.5, 12 Consequently, the majority of sulfated glycosaminoglycans (sGAG) in the middle and outer regions are concentrated in this compartment.5 While the contents of both cartilage and meniscus vary at the tissue length scale, the meniscal structure results in a strong heterogeneity in glycosaminoglycan content at the 100-µm length scale, much stronger than that seen in cartilage. Meniscal static and dynamic compressive moduli are correlated with glycosaminoglycan content,13 and cell-mediated proteoglycan degradation substantially reduces both the compressive and shear moduli of meniscal explants from the middle region.14 These findings suggest that, despite their relatively low bulk concentration, proteoglycans play important structural roles outside of the compression region.

Figure 1.

Cross-section of a bovine lateral meniscus showing the secondary fiber network (light regions due to autofluorescence) surrounding and subdividing the circumferential collagen bundles (dark regions). Scale bar 5 mm. (Figure provided by Janice Lai.)

The contributions of proteoglycans to AC mechanics have been probed with a number of approaches, including manipulation of the proteoglycan-associated osmotic swelling stress by altering the bath solution. While such tests involve conditions outside of the normal physiologic range, they allow manipulation of interactions between tissue constituents without altering the tissue composition. Decreasing the bath salinity increases the osmotic swelling stress and increases static and dynamic compressive, tensile, and shear moduli in AC.15–18 We hypothesized that, despite the substantially lower proteoglycan content, altered osmotic environments would similarly affect MFC tissue mechanics. The objectives of this study were to compare effects of altered osmotic environments on compressive and shear behaviors of MFC and AC tissue explants, and to further compare these effects on axially and circumferentially oriented meniscal explants.

MATERIALS AND METHODS

Sample Preparation

To compare responses between tissues, a total of 50 samples per tissue were isolated (Fig. 2A) from MFC and AC of seven immature bovine stifles (Research 87, Marlborough, MA). Using a 6-mm biopsy punch, full thickness cores were extracted from the middle zone of medial and lateral menisci (perpendicular to the distal surface and avoiding the horns) and from both femoral condyles. The superficial 2 mm was discarded and 2-mm thick discs of midsubstance tissue were isolated using a microtome. Samples were removed from the discs with a 4-mm biopsy punch and pre-equilibrated at 4°C for 4 h in 1× phosphate-buffered saline (PBS; Invitrogen, Carlsbad, CA) supplemented with protease inhibitors (PI, Protease Inhibitor Cocktail Set I; EMD Biosciences, San Diego, CA). Sample masses and heights were measured after pre-equilibration at 1×, and this initial thickness was subsequently used as the basis for strain calculations. Samples were randomly distributed to five groups (n = 10/group/tissue) and equilibrated at 4°C for 4 h in 0.1×, 0.5×, 1×, 2×, or 10× PBS supplemented with PI. Sample masses and heights were measured again and samples were frozen at −20°C in their equilibration solutions.

Figure 2.

(A) Preparation and equilibration of 4 mm × 2 mm meniscal fibrocartilage and articular cartilage samples for tissue comparison. (B) Preparation and equilibration of axial and circumferential 4 mm × 2 mm meniscal fibrocartilage samples for orientation comparison. (C) Mechanical testing and biochemical analysis of tissue samples.

To examine the responses of meniscal samples with different orientations, axially and circumferentially (aligned with the primary collagen architecture) oriented samples were isolated (Fig. 2B) from the lateral menisci of five additional immature bovine stifles (3/orientation/stifle). Axial samples were isolated as described above. Circumferential samples were prepared by extracting radial slices, trimming them to 2 mm with a microtome, and removing discs with a 6-mm biopsy punch. Following equilibration at 4°C for 4 h in 1× PBS with PI, samples were isolated with a 4-mm biopsy punch. For each meniscus, the three discs from each orientation were randomly allocated to 0.1×, 1×, or 10× PBS treatment groups (n = 5/group/orientation) and treated as described above.

Sample Testing

On the day of testing, samples were thawed at 37°C for 30 min and tested immersed in the prescribed equilibration solution (Fig. 2C). Samples were first tested at 37°C in oscillatory torsional shear using an AR-2000ex torsional rheometer (TA Instruments, New Castle, DE). Sequential ramps to 5%, 10%, 15%, and 20% compressive offsets were applied at 0.1%/s followed by stress relaxation for 20 min. At each compressive offset, the dynamic shear modulus magnitude G* was determined by applying ±0.5% nominal shear strain at 0.01–1 Hz. All strains were based on the thickness measured after the initial equilibration in 1× PBS. Note that this shear strain magnitude was small enough that volumetric strains and the associated flow-dependent viscoelastic phenomena were negligible. Samples were removed from the rheometer and allowed to recover for 20 min at room temperature in the equilibration solution before being tested in unconfined compression using an Instron 5848 MicroTester (Instron, Norwood, MA). Sequential ramps to 5%, 10%, 15%, and 20% compressive offsets were applied at 0.1%/s followed by stress relaxation for 20 min. At each compressive offset, the dynamic compressive modulus magnitude E* was determined by applying ±1.5% compressive strain at 0.01–1 Hz. As with the shear testing, all strains were based on the thickness measured after the initial equilibration in 1× PBS. The equilibrium compressive modulus Eeq was determined via linear regression of the relaxed stress against the nominal strain. Both shear and compressive tests were conducted using custom, fluid-filled testing chambers, with fine grit wet-dry sandpaper used to prevent slippage between the samples and the contacting surfaces.

After testing, samples were re-equilibrated in 1× PBS at 4°C for 4 h to ensure a consistent hydration state before biochemical analysis. Samples were frozen, lyophilized, weighed dry, and digested overnight at 60°C using 125 µg Proteinase K (Invitrogen) per sample. DNA contents were measured using the Hoechst 33258 fluorescence assay19 with calf thymus DNA (Invitrogen) standards, sGAG contents were measured using the 1,9-dimethylmethylene blue colorimetric assay20 with chondroitin sulfate C (Sigma–Aldrich, St. Louis, MO) standards, and hydroxyproline (HP) contents were measured using the p-dimethylaminobenzaldehyde assay21 with trans-4-hydroxy-L-proline (Sigma–Aldrich) standards. Data were expressed as a fraction of sample wet mass for statistical analyses.

Evaluation of Equilibration Time

To verify that 4 h equilibration was sufficient, the G* of five tissue samples (three AC, two MFC) were monitored following a single step change (1–2 orders of magnitude) in bath concentration (three increased, two decreased). Following initial equilibration, each sample was compressed by 10% on the rheometer and allowed to stress relax in the initial solution for approximately 2 h. Following three consecutive exchanges with an excess volume at the new PBS concentration, samples were maintained in the new solution for at least 3 h. G* was measured continuously by applying ±0.5% nominal shear strain at 0.1 Hz, and the time constant was evaluated by an exponential fit to the shear modulus versus time.

Statistical Analysis

All data were analyzed using general linear models (GLMs) following optimal Box-Cox transformation (Minitab 16; Minitab Inc., State College, PA). Biochemical contents were compared between tissues and among PBS concentrations, equilibrium compressive moduli were compared between tissues or orientations and among PBS concentrations, and dynamic moduli were compared between tissues or orientations and among PBS concentrations, compressive offsets, and testing frequencies. The donor stifle was treated as a random factor in all GLMs. Significance was set at p < 0.05 and Bonferroni's test was used for pairwise planned comparisons. Data are presented as mean ± standard error of the mean (SEM).

RESULTS

The time constant for mechanical equilibration to a concentration change was 29.2 ± 3.5 min, supporting the use of a 4 h equilibration in these studies. Note that equilibration of G* was evaluated in unconfined compression with restricted diffusion across the top and bottom surfaces, and thus overestimates the time required for free swelling equilibration.

Compared to AC samples, the water and sGAG contents of MFC samples were significantly lower while the HP content was significantly higher (Table 1, all p < 0.0001). The DNA content did not significantly differ between tissues. Neither water, DNA nor sGAG contents varied significantly among treatment groups for either tissue. The HP content was significantly lower in 0.1× samples than in 10× samples for MFC samples (p = 0.035) with no other significant differences among PBS concentrations for either tissue. Comparing orientations for MFC samples, the water content was significantly lower for circumferential samples (p = 0.001) but did not significantly vary among PBS groups. The sGAG content did not significantly differ between orientations, but was significantly lower for the 10× groups than for 0.1× or 1× groups (p = 0.0005). Collectively, these results indicate that the multiple equilibrations and mechanical testing had relatively minor effects on tissue composition, minimizing concerns about indirect influences on measured properties.

Table 1. Biochemical Contents of Articular Cartilage and Meniscus Samples
 Tissue ComparisonOrientation Comparison
CartilageMeniscusAxialCircumferential
  1. Mean ± SEM. Significance differences between tissues (*) or orientations (+), p < 0.05.

Water (w/w, %)75.5 ± 0.271.8 ± 0.4*71.3 ± 0.469.8 ± 0.4+
DNA per wet mass (µg/mg)1.84 ± 0.161.61 ± 0.098*  
sGAG per wet mass (µg/mg)61.9 ± 1.54.29 ± 0.26*7.57 ± 0.636.35 ± 0.29
HP per wet mass (µg/mg)15.2 ± 0.4630.9 ± 1.2*  

Tissue Comparison

For both tissues, all measured moduli increased significantly with increasing frequency, increasing compressive offset and decreasing PBS concentration, with some distinct differences between the tissues (Figs. 3 and 4). Across all combinations of frequency, compressive offset and concentration, all moduli were significantly greater for AC than for MFC (all p < 0.0001). To facilitate visual comparisons of trends between tissues, moduli were normalized to the mean value for each tissue in 1× PBS and (for dynamic moduli) at 10% offset and 0.1 Hz. Under these reference conditions, Eeq, E*, and G* were 52-, 52-, and 148-fold higher, respectively, for AC (Fig. 4 inset).

Figure 3.

Dynamic compressive (E*) and shear (G*) moduli for (A,B) meniscal fibrocartilage and (C,D) articular cartilage samples at 0.1 Hz as a function of compressive offset and PBS concentration. Mean ± SEM, n = 10/group.

Figure 4.

Tissue comparison. Variations in moduli of meniscal fibrocartilage and articular cartilage samples with testing frequency, compressive offset and PBS concentration. For presentation only, data were normalized to the respective tissue means at 1× PBS, 10% offset and 0.1 Hz (see inset). Mean ± SEM, n = 10/group. Articular cartilage samples were significantly stiffer than meniscal fibrocartilage samples for each condition. Letters indicate patterns of significance based on the overall GLM, with significant differences between groups with no letters in common.

Dynamic moduli for both tissues exhibited characteristic stiffening with increasing frequency (Fig. 4A and B) and also increased with increasing compressive offsets (Fig. 4C and D). E* for AC and G* for both tissues increased significantly with each decade increase in frequency, while E* for MFC was significantly higher at 1 Hz than at lower frequencies. E* for both tissues increased significantly between consecutive compressive offsets. G* for both tissues increased significantly with compressive offset up to 15% but did not significantly differ between 15% and 20% offsets.

For both tissues, all moduli decreased with increasing PBS concentration, consistent with a decreased osmotic swelling stress at higher bath salinities (Fig. 4E–G). All moduli for both tissues were significantly greater at 0.1× than at some or all higher concentrations, while all moduli for AC and E* for MFC were lower at 10× than in most or all lower concentrations (see significance indicators in Fig. 4). Compared to 1×, the most prominent relative increase was for MFC G* at 0.1×, which was nearly threefold greater than the relative increase for AC (Fig. 4B; MFC: 3.6 ± 0.6 times; AC: 1.3 ± 0.2 times).

Orientation Comparison

As in the tissue comparison study, moduli of MFC samples at both orientations increased with increasing frequency (Fig. 5A and B), increasing compressive offset (Fig. 5C and D) and decreasing PBS concentration (Fig. 5E–G). Across all combinations of frequency, offset and concentration, circumferentially oriented samples had significantly higher moduli than axially oriented samples (all p ≤ 0.0001). For presentation, moduli were normalized to the mean value for each orientation in 1× PBS and (for dynamic moduli) at 10% offset and 0.1 Hz. Under these reference conditions, Eeq, E*, and G* were 2.1-, 1.8-, and 1.4-fold higher, respectively, for circumferential samples (Fig. 5 inset), likely reflecting the intrinsic anisotropy of meniscal tissue. Note that the axial samples in this second study were more compliant than the meniscal samples in the tissue comparison study, perhaps due to variations among different donor animals. For both orientations, E* was significantly higher at 1 Hz than at lower frequencies and G* was significantly different among frequencies. E* for both orientations and G* for circumferential samples increased significantly with each step in compressive offset. G* for axial samples also increased significantly with compressive offset, although not between each consecutive step. Both Eeq and E* were significantly greater at 0.1× than at higher concentrations for both axial and circumferential samples. Similarly, G* was significantly greater at 0.1× than at higher concentrations for both axial and circumferential samples, with slightly greater values for samples tested at 10× than at 1×. As in the tissue comparison, the most dramatic stiffening of MFC samples at both orientations was for G* at 0.1×.

Figure 5.

Orientation comparison. Variations in moduli of axial and circumferential meniscal fibrocartilage samples with testing frequency (A,B), compressive offset (C,D) and PBS concentration (E–G). For presentation only, data were normalized to the respective orientation means at 1× PBS, 10% offset and 0.1 Hz (see inset). Mean ± SEM, n = 5/group. Circumferential samples were significantly stiffer than corresponding axial samples for each condition. Letters indicate patterns of significance based on the overall GLM, with significant differences between groups with no letters in common.

DISCUSSION

In this study, we examined the effects of altered osmotic environments on the compressive and shear properties of MFC and AC over a range of testing conditions. The results are consistent with previous observations of tissue behaviors, including increased compressive22 and shear23 moduli of meniscal tissue with increasing compressive offset, increased shear moduli of cartilage with increasing compressive offset and decreasing bath salinity,17 and greater shear modulus for circumferential samples than for axial samples.23 It should be noted that, while consistent with other values in the literature,9, 13, 22, 23 the moduli measured for meniscal tissue are quite low. This is likely due to the loss of the tension and continuity of the circumferential fiber bundles that stiffen the tissue under physiologic loading. Nevertheless, these data provide insights into aspects of material behavior relevant to the overall compliance of the meniscal tissue, particularly in the shear mode required to allow the meniscus to accommodate diverse loading conditions.

The observed stiffening with increased compressive offset may be related to multiple compression-induced changes within the tissue. Sustained compression produces a net fluid exudation and consequently an increased resistance to fluid flow through the extracellular matrix (decreased permeability), which would be expected to increase the dynamic compressive modulus E*. The relative increase in E* with increasing compressive offset was more dramatic for meniscus than for cartilage, with (at 1× and 0.1 Hz) a 6.3-fold increase from 5% to 20% for MFC as compared to a 2.4-fold increase for AC. Due to the inhomogeneous, hierarchical structure of the meniscus, macroscopic compression might result in preferential loss of fluid (and therefore preferential decrease in permeability) from the more permeable tissue partition, thus disproportionately increasing the dynamic stiffness. Due to the net fluid exudation, sustained compression also increases the proteoglycan-associated osmotic swelling stress by effectively increasing the fixed charge density, which may offset the presumed reduction in collagen tension with increased compression. Such effects might explain the increased dynamic shear modulus G* with increasing compression, as shear deformation involves no net fluid flow and is thus insensitive to changes in tissue permeability.

Although altered osmotic environments can involve other phenomena (e.g., changing the balance between intra- and extra-fibrillar fluid,24 which may in turn alter the mechanical behavior of the collagen networks), the effects of altered bath salinity are typically attributed primarily to changes in the proteoglycan-associated osmotic swelling stress. This can directly influence tissue mechanical behaviors, as roughly half of the equilibrium compressive stiffness of AC has been attributed to osmotic resistance to compression-induced proteoglycan concentration.15 Relative changes in equilibrium modulus were comparable for meniscus and cartilage, suggesting similar phenomena in both tissues. Reducing the bath osmolarity increases the effective radius of the glycosaminoglycans (e.g., by increasing the Debye length) and consequently decreases the permeability, an effect demonstrated by experiments on charged hydrogels.25 Furthermore, the increased osmotic stress stiffens the collagen network restraining the proteoglycans. These effects are manifest in an increased dynamic compressive stiffness E*, again with comparable relative increases for both tissues.

Unlike the compressive properties, the effects of altered osmotic stress on the dynamic shear modulus G* were notably different for cartilage and meniscus (Fig. 4G). While cartilage displayed a monotonic increase in G* with decreasing bath concentration, the shear modulus of meniscal explants was insensitive to the bath concentration at or above 0.5×. The loss of shear modulus observed with the onset of glycosaminoglycan depletion in meniscal explants14 thus may not be fully attributable to a loss of swelling stress, indicating that proteoglycans may contribute in other ways to the integrity of the secondary fiber network. In contrast, reducing the bath concentration to 0.1× resulted in a substantial increase in G* for meniscal tissue, an effect seen for both axial and circumferential samples. This suggests that the secondary network is not fully “inflated” in reference conditions, but that sufficiently increasing the swelling stress stiffens the secondary network and consequently restricts relative sliding of the circumferential bundles. While this occurs under non-physiologic conditions for immature meniscal tissue, this stiffening effect may be physiologically relevant for mature menisci with higher proteoglycan contents.5

An improved understanding of normal tissue mechanics establishes a baseline for understanding the modifications in meniscal mechanics with pathology, such as changes associated with meniscal degradation during early stages in the development of knee osteoarthritis. Additionally, studies such as these may provide useful design parameters or targets for tissue engineered or synthetic biomaterial meniscal replacements, as mimicking the native tissue's interactions (such as relative sliding of circumferential bundles) may be necessary to adequately accommodate to the wide range of loading conditions applied to the native meniscus. Overall, these studies contribute to a deeper understanding of meniscal tissue behavior, particularly regarding the contributions of proteoglycan-associated osmotic swelling stress to the compressive and shear responses.

Acknowledgements

This work was supported by a grant from the National Institute of Arthritis and Musculoskeletal and Skin Diseases (R01AR052861), a National Science Foundation Graduate Research Fellowship (AMN), and a Stanford Master's of Science in Medicine Fellowship (AMN). The authors thank Patricia Ho for assistance with preliminary studies and Janice Lai for providing the image in Fig. 1.

Ancillary