Strain measurements in the cellular and pericellular matrix of chondrocytes

Transcription

1 Strain measurements in the cellular and pericellular matrix of chondrocytes BMTE Bart van Dijk August 007 Internship at Columbia University, New York City, NY Supervisor Eindhoven University of Technology: René van Donkelaar Supervisors Columbia University: Gerard A. Ateshian Mike Albro Nadeen Chahine Eindhoven University of Technology Department of Biomedical Engineering

2 Abstract Chondrocytes are responsible for the production and maintenance of extracellular matrix in articular cartilage. Mechanical loading influences the biosynthetic response of chondrocytes. The microenvironment of articular cartilage plays an important role in this mechanoregulation of chondrocytes. The objective of this study was to investigate the behaviour of the chondrocyte and its microenvironment, under static loading, in order to address the relationship between the applied deformation and cellular strain. In-situ strain measurements were performed on cells in the middle zone (MZ) under compressive loading, using a custom loading device. Miniature cartilage samples were compressed and images were made before loading and at equilibrium and digital image correlation (DIC) was used to determine the strains in the cells, their pericellular matrix (PCM) and the extracellular matrix (ECM). Also, different regions of the PCM have been compared. The experiments indicate that a strain amplification mechanism is present in the microenvironment of the cell. The maximum (compressive) principal strain (e ) in the cells is in the loading direction and larger than the applied strain on the sample. In the ECM, e is smaller than the applied strain and significantly smaller than e in the cells and the PCM. Furthermore, e is significantly larger in the cells than almost all of the PCM. The minimum (tensile) principal strain (e 1 ) at equilibrium is much smaller than the maximum principal strain in magnitude and is positive and perpendicular to the loading. There are no significant differences between the different regions, and most of them are not even significantly different from zero. We can conclude that it is possible to measure differences in strains using DIC and a microscopy based loading device. There is strain amplification in the cells, due to the difference in mechanical properties of the cells, the PCM and the ECM. Mechanotransduction of chondrocytes may be mediated by this strain amplification mechanism during loading. 1

3 Contents 1 Introduction 3 Materials and Methods 4.1 Sample preparation Experimental setup Image Analysis Statistical Analysis Results 7 4 Discussion 9 5 Appendix 13

4 1 Introduction Articular cartilage consists of extracellular matrix components, like proteoglycans and type II collagen, water and chondrocytes. Chondrocytes are the sparsely distributed cells in cartilage, responsible for the production of these extracellular matrix components. The proteoglycans form large aggregates, which contribute to the compressive stiffness [1]. Collagen forms a highly cross-linked meshwork that is resistant to tensile forces [10]. At the tissue level, articular cartilage has complex mechanical properties such as depth-dependent inhomogeneity and tension-compression nonlinearity []. Consequently, articular cartilage is divided into three different zones, the superficial, middle and deep zone. Collagen in cartilage appears to be a random-oriented network, but preferred fibril patterns are evident [3]. Collagen is oriented perpendicular to the articular surface in the deep zone, and parallel in the superficial zone [9]. The proteoglycan density is different in the different zones as well, lower in the superficial zone than in the middle and deep zones [10]. The ultra structure and alignment of matrix molecules such as collagen and proteoglycans result in this directional dependence in the mechanical properties, known as anisotropy. Additionally, the proteoglycans in the tissue attract water into the cartilage matrix, resulting in a dense hydrated environment. When loaded, the tension in the collagen fibers limits the lateral expansion of the tissue and the interstitial fluid pressurizes and experiences resistance to flow out of the tissue. The applied load is sustained by both the extracellular matrix and the pressurized fluid, increasing the ability of the tissue to bear load [7]. Cartilage is avascular, so chondrocyte activity is predominantly regulated by local factors in the cellular microenvironment [4]. Mechanical signals imparted by stretch, pressure, tension, fluid flow, or shear stress rapidly lead to the activation of multiple intracellular signaling molecules and pathways. The pericellular matrix, the region surrounding the chondrocytes, is believed to play an important role in regulating this biochemical and biomechanical environment of the chondrocyte [5]. The goal of this study is to determine the deformation behaviour of the chondrocyte and its microenvironment under static loading, in order to address the relationship between the applied deformation and the cellular strain. In situ strain measurements are done on cells, their pericellular matrix and the extracellular matrix in the middle zone (MZ) of bovine articular cartilage, using a custom microscopy loading device. Digital image correlation (DIC) is used to characterize the behaviour of cartilage at the cellular level. 3

5 Materials and Methods.1 Sample preparation Full thickness osteochondral plugs (ø 5 mm) were harvested from the carpometacarpal joint of a 3-4 month old calf under sterile conditions and cultured in high glucose Dulbecco s Modified Eagle s Medium (DMEM) supplemented with 1% L-Proline, 1% ITS, 1% PSAM, 100 mg/ml sodium pyruvate and 10% DEX. These disks were cultured in a humidified incubator at 37 C and 5% CO. All explants were allowed to acclimate to the culture environment for at least days prior to testing. On the day of testing, all subchondral bone and most of the articular layer was removed using a freezing micro-tome. Two smaller cylindrical samples (ø mm) were prepared from the disks and cut diametrically to create 4 semi-cylindrical specimens.. Experimental setup The samples were mounted in a manual controlled loading apparatus, fitted with a 50 grams load cell for load measurement. The samples are mounted with the articular surface at the loading platen and the bone side at the static platen. A loading micrometer is used to control the displacement. The setup is displayed in Figure 1 []. The samples were immersed in Phosphate Buffered Saline (PBS) while testing. A light microscope with a 0x objective was used (resolution: 0.3µm/pixel). Images were acquired with a Micromax 5-MHz interline transfer chip camera (Princeton Instruments, Trenton, NJ, US, image Figure 1: The experimental setup resolution: 1300*1030 pixels) and Metamorph Imaging system (Universal Imaging, West Chester, PA, US). A significant tear strain was applied (10%) on the samples and allowed to equilibrate for 0 minutes. Either 5 or 10% strain was applied on the specimens and allowed to equilibrate as well. Images were made before loading and at equilibrium..3 Image Analysis Digital image correlation (DIC), tracking the displacement of the cells during the deformation, was performed using VIC-D (Correlated Solutions, West Columbia, SC, US). Displacement analysis (DIC subset region =

6 pixels, step size = 9 pixels) was performed on regions of interest (ROIs) consisting of a couple of cells and their surrounding matrix. For every subset region, the corresponding subset region in the deformed image was determined, by minimizing the correlation coefficient. VIC-D uses the following correlation coefficient (r). (A mn B mn ) m n r = 1 ( ) ( ) (1) (A mn ) (B mn ) m n m n Where A and B are the intensities of the subset region of the original and the deformed image, respectively. The difference in x and y position between the deformed and the undeformed state is the displacement, respectively u and v. α 1 + α x + α 3 y = u () β 1 + β x + β 3 y = v (3) Where α i and β i (i = 1,, 3) are constants and x and y are the positions in x and y direction. Out of three points, triangular surfaces are created and used to determine the strain. Using equations and 3, the following sets of equations are used for the displacement in x and y direction of these surfaces. 1 x 1 y 1 α 1 u 1 1 x y α = u (4) 1 x 3 y 3 α 3 u 3 1 x 1 y 1 1 x y 1 x 3 y 3 And using the following definitions: β 1 β β 3 = v 1 v v 3 (5) x = x i, i = 1,, 3 (6) ȳ = y i, i = 1,, 3 (7) ū = u i, i = 1,, 3 (8) v = v i, i = 1,, 3 (9) Where x i is the position in x direction for point i, y i is the position in y direction for point i, u i is the displacement in x direction for point i and v i is the displacement in y direction for point i. The sets of equations 4 and 5 5

7 are then reduced to the following: α 1 + α x + α 3 ȳ = ū (10) β 1 + β x + β 3 ȳ = v (11) To determine ɛ xx, ɛ yy, ɛ xy out of equation 10 and 11, the following relations were used: ɛ xx = δū δ x = δ(α 1 + α x + α 3 ȳ) = α (1) δ x ɛ yy = δ v δȳ = δ(β 1 + β x + β 3 ȳ) = β 3 (13) δȳ ɛ xy = ( δū δȳ + δ v δ x ) = δ(α 1 + α x + α 3 ȳ) + δ(β 1 + β x + β 3 ȳ) = α 3 + β δȳ δ x (14) In this experiment, we were interested in the principal strains. The principal strains are determined out of ɛ xx, ɛ yy, ɛ xy, using the following equation: or: e 1, = ɛ xx + ɛ yy e 1, = α + β 3 (ɛxx ) ɛ yy ± + (ɛ xy) (15) (α ) β ( ) 3 α3 + β ± + (16) We report the minimum (e 1 ) and the maximum (e ) principal strains, averaged over multiple points from within the cells (IC), in the surrounding pericellular matrix (PCM) and in the extracellular matrix (ECM). To look at differences between different parts of the PCM, it is divided into four different regions. Considering the point of view during the experiment, the four different regions are bone side, articular side, front and back of the cell. In the experiments, compression is applied from the articular side to the bone side. To account for the differences in overall strain in the ROIs, the strains are normalized by the average of the maximum principal strain of the ROI. This gives the normalized minimum (ē 1 ) and maximum (ē ) principal strains..4 Statistical Analysis Statistical analysis was performed on strain measurements. To determine the effect of the different regions on the strain, an analysis of variance (ANOVA) was used. To look at the differences between groups, an LSD-post hoc test was used with p < 0.05 considered significant. A T-test was used to test if groups were statistically different from zero. 6

8 3 Results For one region of interest (ROI), a representative maximum principal strain (e ) contour map at equilibrium is presented. Figure (a) shows a filled contour map and the original image with the same contour lines is presented in Figure (b). (a) Filled contour map of e (b) Original image with e contour lines Figure : Contour maps of the maximum principal strain (e ) of a ROI. In total, 10 cells or groups of cells are investigated in 6 different ROI s. The average normalised maximum principal strains (ē ) are reported in Figure 3. For all the regions, ē is positive, which is in the direction of the applied load. In the cells, ē is greater than 1 (ē = 1. ± 0.1), representing a more negative strain than the average of the ROI. In the ECM, ē is smaller than 1 (ē = 0.75 ± 0.1), representing a less negative strain than the average of the ROI, significantly different from the cells. In the PCM, ē is between 0.90 and 1.1, depending on the region. The four distinct Figure 3: Normalised maximum principal strains (ē ) 7

9 regions are not significantly different, except bone side and back of the cell (p < 0.05). Bone side (ē = 0.90±0.18), articular side (ē = 0.99±0.17) and front (ē = 1.05 ± 0.1) are significantly different from the cell. The region back of the cell is not significantly different from the cell (ē = 1.1 ± 0.3, p < 0.7). All of the PCM regions are significantly different from the ECM. Figure 4: Normalised minimum principal strains (ē 1 ) The average normalised minimum principal strains (ē 1 ) are reported in Figure 4. For all the regions, ē 1 is negative, representing elongation, perpendicular to the loading direction. The magnitude of ē 1 is ten times smaller than ē. The analysis of variance shows there is no significant difference among the means of the different regions (p < 0.69). Also, only two regions are significantly different from zero; the matrix in front of the cell (ē 1 = 0.11±0.1, p < 0.03) and the extracellular matrix (ē 1 = 0.05±0.06, p < 0.0). All the data is summarized in Table 1. Table 1: The normalised strains IC PCM-L PCM-R PCM-F PCM-B ECM N ē 1. ± 0.90 ± 0.99 ± 1.05 ± 1.1 ± 0.75 ± ē ± ± ± ± ± ±

10 4 Discussion The goal of this study was to asses the in-situ loading environment of chondrocytes before loading and after equilibrating, using an experimental approach. The maximum principal strains in the cells are significantly higher than in the ECM. Furthermore for all the regions in the PCM, the maximum principal strain is also higher than in the ECM. The compressive stiffness is therefore higher in the ECM than in the cells and the PCM. The maximum principal strain in the cell is significantly higher than in three of the four regions of the PCM, so the cell has a significantly lower compressive stiffness than the PCM in these regions. Although the minimum principal strains (e 1 ) are in magnitude ten to fifteen times as small as the maximum principal strains (e ), their variance is in the same order of magnitude. This is an indication that the large variance is not caused by the strains, but by the quality of the images. Analysis on digitally strained images from the experiments shows that the variance is small, namely ɛ xx = 0.1 ± and ɛ yy = 0.1 ± 0.004, when doing 10% compression in x direction and 10% elongation in y direction. This shows that the texture of the images is good enough for DIC. But during the loading, the intensity of the light changes and cells move out of the plane of view. To determine the strength and weaknesses of the analysis program, preliminary analysis was done on the effect of light intensity changes and out of plane motion. Although small intensity changes do not induce incorrect strains, an out-of-plane motion of 1 µm induces strains up to 5%, so outof-plane movement is the foremost reason for incorrect displacements and therefore incorrect strains. Although there is no significant difference between the regions when looking at ē 1, increasing the image quality will decrease the variance and will certainly give better results. Obviously, this will reduce the variance for ē as well, making it possible to see more significant results and maybe see differences among the different regions of the PCM. In the following paragraphs a number of improvements are suggested to obtain better results. The average maximum principal strains (e ) in the different experiments are in the order of % to 14%. There are two causes for this difference in average strain. The experiments were done at two different platen-toplaten strains, namely 5% and 10% on full thickness tissue samples. But for all the experiments done at 10% platen-to-platen strain, still an average strain of 8.7% to 13.7% was measured, for the different ROIs. The cause for this difference is that the compressive mechanical properties of cartilage are depth-dependent, even in a single zone [8]. To compensate for the differences, e 1 and e are normalised by the average maximum principal strain of the ROI. In this study, a manually operated loading device was used. The advan- 9

11 tages of the manual device are its accuracy, stability and relatively small compliance. An automatic loading device will be more accurate in applying the desired strain and will make dynamic loading possible, instead of only static loading, which is more representative of physiological conditions. Furthermore it will be possible to make images during loading. A second shortcoming of the used loading device, was its inability to record the loadcell data, limiting the experimental setup to strain measurements only. Without any stress data, it is not possible to determine mechanical properties of the different regions or do any numerical experiments to validate the data. A third improvement on the experimental setup is using a loading device that compresses the samples from both sides, a dual loader. With a dual loader, the displacements in the ROIs can be minimized when a ROI is selected in the center of the sample. When the displacements are small, the out of plane movements are smaller as well. Due to the low amount of useable experiments, data of different platento-platen strains are used. It would be preferable to compare experiments of the same platen-to-platen strain. Furthermore comparing the results of different platen-to-platen strains will give some insight into the non-linearity of the compressive properties of the tissue. In this study only the middle zone of the cartilage is examined. Due to the different material properties of the different cartilage zones, their compressive behaviors are very different. It is therefore recommended to look at the deep and superficial zones in future studies as well. In order to reduce the variance of the strains, the quality of the images has to be improved. Although the texture in the images seems to be good enough, increasing it will improve the ability of the DIC program to track the displacement in the images and thus the results will be more accurate. For instance, using stains for gag s or collagen type II in low concentrations can improve the texture, because this will show these matrix components, but will not saturate the whole image. Using a camera with a higher resolution will increase the amount of pixels per µm. Using a higher magnification objective will not only increase the amount of pixels per µm, but will also increase the texture of the images. A disadvantage of using a higher magnification is that the effect of the out-of-plane movement is higher. The correlation in the images varies, so not every part of the image is usable for DIC. The selection of the region of interest (ROI) is now based on the value of the correlation coefficient, which is a reliable method, but it is preferable that there is good correlation throughout the image. In the images, there is no distinction between PCM and ECM. The location and size of the PCM has to be determined using histology, to validate the results. For instance using DTAF (6-([4,6-dichlorotriazin--yl]amino)- 10

12 fluorescein), which is a reactive dye that can be used to visualize the PCM under fluorescence microscopy [6]. We conclude that the different mechanical properties of cells, their surrounding pericellular matrix and the extracellular matrix can be determined using DIC and a microscopy-based loading device. The applied strains get magnified in the cells due to the difference in mechanical properties of the cells, the PCM and the ECM. These results suggest that this strain amplification could contribute to the mechanotransduction in chondrocytes. References [1] G. A. Ateshian, N. O. Chahine, I. M. Basalo, and C. T. Hung. The correspondence between equilibrium biphasic and triphasic material properties in mixture models of articular cartilage. Journal of Biomechanics, 37: , 004. [] N. O. Chahine, C. C. B. Wang, C. T. Hung, and G. A. Ateshian. Anisotropic strain-dependent material properties of bovine articular cartilage in the transitional range from tension to compression. Journal of Biomechanics, 37: , 004. [3] M. H. Chen and N. Broom. On the ultrastructure of softened cartilage: A possible model for structural transformation. Journal of Anatomy, 19:39 341, [4] E. M. Darling, S. Zauscher, and F. Guilak. Viscoelastic properties of zonal articular chondrocytes measured by atomic force microscopy. Osteoarthritis and Cartilage, 14: , December 005. [5] F. Guilak, L. G. Alexopoulos, M. L. Upton, I. Youn, and J. B. Choi. The pericellular matrix as a transducer of biomechnical and biochemical signals in cartilage. Annals of the New York Academy of Sciences, 1068:498 51, April 006. [6] G. M. Lee and R. F. Loeser. Interactions of the chondrocyte with its pericellular matrix. Cells and Materials, 8: , March [7] M. A. Soltz and G. A. Ateshian. Experimental verification and theoretical prediction of cartilage interstitial fluid pressurization at an impermeable contact interface in confined compression. Journal of Biomechanics, 31:97 934, [8] C. C. B. Wang, N. O. Chahine, C. T. Hung, and G. A. Ateshian. Optical determination of anisotropic material properties of bovine articular cartilage in compression. Journal of Biomechanics, 36: ,

13 [9] W. Wilson, N. J. B. Driessen, C. C. van Donkelaar, and K. Ito. Prediction of collagen orientation in articular cartilage by a collagen remodeling algorithm. Osteoarthritis and Cartilage, 14: , May 006. [10] W. Wilson, C. C. van Donkelaar, B. van Rietbergen, and R. Huiskes. The role of computational models in the search for the mechanical behavior and damage mechanisms of articular cartilage. Medical Engineering and Physics, 7:810 86, March

14 5 Appendix (a) Filled contour map of e1 (b) Original image with e1 contour lines (c) Filled contour map of e (d) Original image with e contour lines Figure 5: Contour maps of the maximum (e ) and minimum (e1 ) principal strains of ROI 1 (a) Filled contour map of e1 (b) Original image with e1 contour lines (c) Filled contour map of e (d) Original image with e contour lines Figure 6: Contour maps of the maximum (e ) and minimum (e1 ) principal strains of ROI

15 (a) Filled contour map of e1 (b) Original image with e1 contour lines (c) Filled contour map of e (d) Original image with e contour lines Figure 7: Contour maps of the maximum (e ) and minimum (e1 ) principal strains of ROI 3 (a) Filled contour map of e1 (b) Original image with e1 contour lines (c) Filled contour map of e (d) Original image with e contour lines Figure 8: Contour maps of the maximum (e ) and minimum (e1 ) principal strains of ROI 4

16 (a) Filled contour map of e1 (b) Original image with e1 contour lines (c) Filled contour map of e (d) Original image with e contour lines Figure 9: Contour maps of the maximum (e ) and minimum (e1 ) principal strains of ROI 5 (a) Filled contour map of e1 (b) Original image with e1 contour lines (c) Filled contour map of e (d) Original image with e contour lines Figure 10: Contour maps of the maximum (e ) and minimum (e1 ) principal strains of ROI 6