The ultrastructure of cartilage tissue and its swelling response in relation to matrix health

Abstract

This multi‐length scale anatomical study explores the influence of mild cartilage structural degeneration on the tissue swelling response. While the swelling response of cartilage has been studied extensively, this is the first study to reveal and correlate tissue microstructure and ultrastructure, with the swelling induced cartilage tissue strains. Cartilage sample strips (n = 30) were obtained from the distal‐lateral quadrant of thirty mildly degenerate bovine patellae and, following excision from the bone, the cartilage strips were allowed to swell freely for 2 h in solutions of physiological saline and distilled water successively. The swelling response of this group of samples were compared with that of healthy cartilage, with (n = 20) and without the surface layer (n = 20). The subsequent curling response of cartilage showed that in healthy tissue it was highly variable, and with the surface removed some samples curved in the opposite direction, while in the mildly degenerate tissue group, virtually all tissue strips curved in a consistent upward manner. A significant difference in strain was observed between healthy samples with surface layer removed and mildly degenerate samples, illustrating how excision of the surface zone from pristine cartilage is insufficient to model the swelling response of tissue which has undergone natural degenerative changes. On average, total tissue thickness increased from 940 µm (healthy) to 1079 µm (mildly degenerate), however, looking at the zonal strata, surface and transition zone thicknesses both decreased while deep zone thickness increased from healthy to mildly degenerate tissue. Morphologically, changes to the surface zone integrity were correlated with a diminished surface layer which, at the ultrastructural scale, correlated with a decreased fibrillar density. Similarly, fibrosity of the general matrix visible at the microscale was associated with a loss of later interconnectivity resulting in large, aggregated fibril bundles. The microstructural and ultrastructural investigation revealed that the key differences influencing the tissue swelling strain response was (1) the thickness and extent of disruption to the surface layer and (2) the amount of fibrillar network destructuring, highlighting the importance of the collagen and tissue matrix structure in restraining cartilage swelling.

This multiscalar investigation identifies key structural differences influencing the bovine articular cartilage free swelling response. These are (1) the thickness and extent of disruption of the surface layer, and (2) the amount of fibrillar network destructuring. Findings highlight the importance of the collagen matrix structure in restraining cartilage swelling.

1. INTRODUCTION

Articular cartilage is a complex tissue consisting mostly of water (60%–80%) contained by hydrophilic proteoglycan molecules trapped in a low permeability dense collagen fibril network. Due to this structural and compositional make‐up, cartilage displays a unique swelling capacity that arises from an interstitial hydrostatic pressure acting against a physical restraint of the collagen network (Fry & Robertson, 1967; Maroudas, 1976; Ogston, 1970). When the tissue is subject to compression, the swelling pressure withstands loads, dependent upon the frictional resistance of displacing water through the collagen network remaining very high (Linn & Sokoloff, 1965; Mow et al., 1992). The swelling pressure arises from the fixed charge nature of large aggregates of proteoglycans. This induces a Donnan osmotic pressurisation of the interstitial fluid that is constrained by the considerable elastic forces provided by the collagen fibre network (Eisenberg & Grodzinsky, 1985; Maroudas, 1976; Myers et al., 1984). The proteoglycan molecules have negatively charged glycosaminoglycan chains that tend to attract mobile cations from the bathing solution, into the tissue, to create a Donnan osmotic pressure from the ionic imbalance (Maroudas, 1976; Wan et al., 2010). Such a fixed charged density in the cartilage tissue can be confirmed from studying the influence of bathing solution molarity on tissue swelling behaviour (Eisenberg & Grodzinsky, 1985; Myers et al., 1984), where it was shown that the magnitude of swelling in free swelling cartilage samples is greater in solutions of lower molarity (i.e., hypotonic) (Setton et al., 1998). This occurs because in solutions of higher molarity, the concentration of the bath exceeds the average fixed charge density of the tissue, and hence water tends not to move into the tissue (Chahine et al., 2005; Eisenberg & Grodzinsky, 1985).

Another influencing factor on cartilage mechanics is the structural and compositional variation in cartilage tissue depth. Besides the arrangement of collagen fibrils in the network from surface to deep resembling an arcade of tangentially aligned to radially aligned fibrils (Benninghoff, 1925; Clark, 1990, 1991), the content of proteoglycans and collagen ranges from 3%–10% and 10%–30% respectively, depending on tissue depth (Mow et al., 1994). There are therefore depth‐dependent tissue swelling responses that show full thickness cartilage when removed from bone and would tend to curl upwards towards the surface zone (Brown et al., 2019; Setton et al., 1994). Such a depth‐dependent swelling gradient (Setton et al., 1994) was deemed to be due to two factors: the variation in the fixed charge density from having less proteoglycans in the surface relative to the deep zones, and similarly the larger density of collagen fibrils, aligned tangentially, in the surface zone (Maroudas, 1979; Setton et al., 1994). However, a later study on the curling behaviour of cartilage showed that rather than fixed charge density variation, it was the layered collagen ultrastructural variation that played the dominant role (Wan et al., 2010). Specifically, the high stiffness at the articular surface with its dense formation of collagen fibrils was the reason full thickness cartilage strips removed from bone would curl upwards following immersion in saline solution (Wan et al., 2010). Thus, the upward “curling” response of cartilage removed from bone and subject to swelling (Brown et al., 2019; Setton et al., 1994), implies that the surface layer contains a stiffer collagen network compared with the deeper layers. Indeed, mechanical testing of isolated layers of SZ and DZ cartilage show such a variation in tensile stiffness (Verteramo & Seedhom, 2004). It has been proposed that cartilage structural “inhomogeneity” explains the curling behaviour of cartilage swelling when removed from bone, where such an inhomogeneity can be modelled as a layer‐wise differential in stiffness (Wan et al., 2010). However, such an interpretation misses the significance of the collagen network structural reality, and how it contributes mechanically across the length scales. For example, while it was shown that surface layer removal lead to an increased swelling‐induced stretch, the overall tissue still curled to the extent of those tissues with an intact surface layer (Setton et al., 1998). Setton et al. (1998) suggest that neither fixed‐charge density variations or layer‐wise anisotropy was able to explain this behaviour. Recently, by studying the collagen fibrillar network architecture, we were able to show how transverse fibrillar network interconnectivity and density was a significant influencing factor governing cartilage swelling behaviour (Brown et al., 2019).

That the collagen network microstructure and ultrastructure may be of at least equal or more importance than proteoglycans and fixed charge density in understanding the swelling mechanism of cartilage, has been relatively understated, given the current lack of data in the literature. The ultimate influence of the collagen network structure on the swelling behaviour of cartilage was emphasised in the concluding remarks of Bank et al. (2000) and also Basser et al. (1998) stating that the extent of volumetric change in the tissue from swelling is largely controlled by the mechanical properties of the collagen network, in providing a stiffness that limits hydration. It was thus revealed that with tissue degradation, the relative failure of the elastic restraint of the collagen network became the enabling factor for increased tissue swelling observed in degenerate tissue compared with healthy (Bank et al., 2000). Yet many of the past cartilage swelling studies that investigate collagen do so in terms of collagen compositional and degradation quantities (Bank et al., 2000; Basser et al., 1998), and hardly on the fibrillar network architecture and integrity. The length scale for cartilage swelling requires a consideration of the hierarchically relevant structural organisation of cartilage tissue, especially because changes to this structure have been shown to occur in the early stages of joint degeneration. For example, with ageing and degeneration, the collagen fibrillar network in articular cartilage loses lateral or transverse interconnectivity, such that the bearing of axial loads on the tissue would be compromised (Broom, 1982; Chen & Broom, 1998). Such a loss of microscale lateral interconnectivity in the collagen network also allows for a volumetric increase in hydration during free‐swelling, contributing to the macroscale softening of the articular cartilage tissue (Brocklehurst et al., 1984; Broom, 1982; Maroudas & Venn, 1977).

Acknowledging the mechanically significant age‐related collagen network structural changes in articular cartilage, comprising of an altered and loosely packed fibril arrangement (Broom, 1982; Thambyah et al., 2012), and surface layer disruption culminating in full depth fissuring of the tissue (Thambyah & Broom, 2010), we seek to understand how such changes would influence the tissue swelling response. Importantly such changes in the collagen fibrillar network structure have been associated with early tissue degeneration (Hargrave‐Thomas et al., 2013), and found to be consequential to the tissue response to mechanical loading, both under static and dynamic conditions (Thambyah & Broom, 2007, 2010; Thambyah et al., 2012; Workman et al., 2020; Workman et al., 2017). For the present study, three collagen network structure related aspects are investigated in relation to the biomechanics of swelling strain and these are: (1) degenerative changes in the surface, (2) loss of fibrillar matrix interconnectivity and (3) change in deep zone tissue microstructure.

2. Methods

2.1. Tissue harvesting

Adult bovine patellae from freshly slaughtered dairy cows were used in this study. Obtained directly from a meat processing plant, patellae were sorted according to gross appearance of cartilage such that they were either healthy appearing (G0), had some softening and swelling (G1) or some fissuring (G2) (Outerbridge, 1961). For this study, the aim was to compare the effects of mild cartilage degeneration with healthy tissue. In earlier studies, we had shown how G1 patellae of mature dairy cows consistently contain such a mild level of degeneration in the distal‐lateral quarter of the patella articular cartilage surface (Hargrave‐Thomas et al., 2013; Thambyah & Broom, 2007; Thambyah et al., 2012). Mild degeneration is associated with subtle softening of the cartilage tissue, and at the microstructural and ultrastructural scales, relatively mild but significant fibrillar network destructuring and reaggregation into large bundles (Thambyah et al., 2012; Joshua Workman et al., 2017). Hence 12 patellae classified as having G1 regions were selected for this study. Tissue was obtained from the distal‐lateral quadrant of affected bovine patellae and tests were carried out as has been carried out in our earlier study on G0, healthy, tissue (Brown et al., 2019). Thus the full method for harvesting and testing of sample strips was outlined earlier (Brown et al., 2019), but in summary: sample strips were excised from the distal‐lateral region of the patella (where split lines are oriented predominantly from base to apex). They were then allowed to equilibrate in two different solutions for a period of 3 h each. At 2 h, it was assumed that the sample had reached equilibration after which, no significant swelling and or morphological alteration was observed. The first solution was physiological saline (0.15 M), and the second was distilled water.

For all samples, once the soft tissue was excised from the bone—retained bone blocks were fixed in a 10% formalin solution followed by decalcification in a 10% formic acid solution. Thus, in the event that we wished to examine the microstructure at the cartilage‐bone junction we could prepare samples for microscopy accordingly. This method of mild sample decalcification has been used in previous studies by this lab group to great effect (Brown et al., 2019; Hargrave‐Thomas et al., 2013; Thambyah & Broom, 2013; Joshua Workman et al., 2017).

2.2. Test groups

A total of 30 sample strips were obtained from the G1 bovine patellae and underwent the above described swelling tests. Morphological, strain and micro‐to‐ultra structural analysis was carried out and compared with a control, the healthy tissue data reported in our earlier study (Brown et al., 2019). In that earlier study, 40 sample strips from healthy G0 patellae were obtained, of which 20 were tested intact and the remaining 20 were tested with the surface layer removed. Thus, for the present study we compare three groups: G1 (new data presented), and G0‐intact and G0‐surface layer removed (data published earlier in (Brown et al., 2019)).

2.3. Strain measurements

All samples were imaged using a stereomicroscope and a calibration frame at three time points: (I) immediately removed from the bone, (II) following equilibration for 2 h in saline and (III) 2 h in distilled water. Using the stereomicroscope images and ImageJ software, the sample dimensions were obtained at stages I‐III (Figure 1). From these dimensions, linear strain measurements were calculated for the following scenarios:

$$\frac{\left(L_{\text{III}}‐L_{\mathrm{I}}\right)}{L_{\mathrm{I}}}$$ (1)

$$\frac{\left(L_{\text{II}}‐L_{\mathrm{I}}\right)}{L_{\mathrm{I}}}$$ (2)

where L _I is sample length immediately removed from the bone, L _II is sample length following equilibration in saline for 2 h and L _III is sample length following equilibration in distilled water for 2 h. Importantly, calculations of strain considered the change in SZ and DZ length separately such that we could understand the relative change in length across the full sample depth. As such, Equations 1 and 2 were used to calculate values of strain based on surface zone (labelled as L1 in Figure 1) and DZ (labelled as L2 in Figure 1) dimensional changes.

FIGURE 1.

Exemplary images of (a) ‘straight’, (b) ‘bend’ and (c) ‘arch’ shape categories. The “Arch” sample is marked up to indicate areas of measurement for the calculation of strain [Colour figure can be viewed at wileyonlinelibrary.com]

Similarly, axial strains were calculated for each sample. Average sample thickness (initial and final) was calculated from three points across sample length. Axial strain was then calculated using the same simple linear equation.

$$\frac{h_{\left(\text{avg},\text{III}\right)}‐h_{\left(\text{avg},\mathrm{I}\right)}}{h_{\left(\text{avg},\mathrm{I}\right)}}$$ (3)

2.4. Morphology

Three levels of imaging were employed to study the tissue at different length scales. The stereomicroscope images obtained were used to study the samples at the macro‐scale. Images of the samples post‐equilibration were classified according to the degree of curvature from mild to moderate to severe, corresponding to the following shape names: bend, arch and spiral/twist respectively (Brown et al., 2019).

Following equilibration and imaging with the stereomicroscope, samples were chemically fixed in a 10% formalin solution for approximately 24 h. Liquid nitrogen was then used to snap freeze samples for storage, until when they were cryosectioned into thin 30 µ sections.

For examination at the microscale, the microsections were wet saline‐mounted on a glass slide and secured under a cover slip, allowing the tissue to be examined in its fully hydrated state via differential interference contrast (DIC) optical microscopy. Images captured via DIC provided the viewer structural detail of chondrocyte morphology and the extra cellular matrix texture. Extensive work carried out by this lab previously have shown how DIC imaging, validated by scanning electron microscopy (SEM), is able to discern in cartilage, densely woven collagen networks from those that have undergone fibrillar scale network destructuring (Nickien et al., 2017; Thambyah et al., 2012; Workman et al., 2017).

For examination at the ultrastructural scale, selected 30‐µ sections from representative groups were further processed. This involved tissue defatting by submerging samples in hexane, removal of proteoglycan content through exposure to bovine testicular hyaluronidase followed by progressive dehydration via exposure to a series of ethanol solutions of increasing concentration (Brown et al., 2019). Sections were then critical point dried, mounted onto SEM stubs and sputter coated with platinum. Images captured via SEM provided high resolution details of the fibrillar arrangement and orientation of the collagenous network.

Chondrocyte morphology and orientation were used to identify and measure the tangential or surface zone (SZ), the transition zone (TZ) and the radial or deep zone (DZ) cartilage (Figure 2). Thus, observing flattened, tangentially aligned, discoidal chondrocytes confirmed that images were captured within the tissue SZ, rounded cells generally singular in arrangement identified the TZ and vertical clusters of chondrocytes confirmed DZ matrix (Pedersen et al., 2013; Thambyah & Broom, 2006; Youn et al., 2006). For G0 and G1 sample groups, where it was possible to obtain a clear, full depth DIC image post fixation, zonal measurements were recorded in an effort to establish the relative thickness of each zone (Figure 3). Further, high magnification DIC images were studied to qualitatively assess chondrocyte morphology and arrangement in order to establish some understanding of the extent to which respective zones were ‘present’. For the SZ, assessment was based on the presence of flattened, discoidal chondrocytes (Brown et al., 2019; Meachim & Stockwell, 1979). For the DZ, assessment was based on the presence of a layer where chondrocytes had adopted a rounded morphology, with a packing density greater than that observed in the middle zone (Bergholt et al., 2016; Thambyah & Broom, 2013). Finally the lines of chondrocyte continuity (Figure 3) provide the observer with an indication of the collagen fibril directionality (Thambyah & Broom, 2006; Thambyah et al., 2012).

FIGURE 2.

(a) Image taken in the surface zone of tissue, note the flattened ‘tangentially‐aligned’ chondrocytes typical in the surface layer of cartilage, where tangential is the direction parallel to the articular surface orientation depicted (white double arrow line). (b) Image taken in mid zone of tissue, with the chondron morphology showing a radial direction that is typical for this region. The radial direction is relatively perpendicular to the articular surface orientation (white double arrow line). Scale bar approximately 40 µ [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 3.

Full depth DIC image of sample that remained straight. Lines of chondrocyte continuity run approximately parallel. Scale bar approximately 50 µ [Colour figure can be viewed at wileyonlinelibrary.com]

3. RESULTS

Of the 30 G1 samples tested, 3 remained straight or showed little to no change in shape. A total of 14 G1 samples were categorised as a ‘bend’ shape, showing a small amount of curvature while the remaining 13 samples were categories as the more curved ‘arch’ shape, a total of 90% of the samples. None curved severely, that is spiral or twist. The G0 samples on the other hand showed only 9 out of 20 (45%) having the bend or arch morphology. Another seven displayed an extreme spiral or twist curvature. Without the surface layer, the G0 samples displayed a variation of bends, arches and spirals, but the interesting outcome was the degree of strain such that some of the curvatures were in the opposite direction that is towards the DZ. The strain measurements (Tables 1 and 2) in the next section allows for a better appreciation of this swelling response.

TABLE 1.

Average SZ and DZ strain, calculated according to Equation 1 and sorted according to final shape category

	G0‐intact			G0‐surface layer removed			G1‐mild degeneration
	SZ	DZ		SZ	DZ		SZ	DZ
Straight (N = 4)	0.034 SD 0.06	−0.64 SD 0.20	Straight (N = 2)	0.069 SD 0.03	0.022 SD 0.05	Straight (N = 3)	0.067 SD 0.01	0.079 SD 0.03
Bend (N = 2)	0.059 SD 0.02	0.264 SD 0.18	Bend (N = 7)	0.425 SD 0.30	0.185 SD 0.17	Bend (N = 14)	0.042 SD 0.04	0.257 SD 0.12
Arch (N = 7)	−0.056 SD 0.08	0.284 SD 0.15	Arch (N = 6)	0.270 SD 0.27	0.357 SD 0.14	Arch (N = 13)	0.030 SD 0.06	0.303 SD 0.10

Equation 1 calculates strain based on the difference in sample dimensions between ‘as‐removed’ and ‘post‐hypotonic solution equilibration’ states. G0 data obtained from earlier research by this lab group and used as a control.

TABLE 2.

Average axial strain sorted according to final shape category

Shape category	G0‐intact	G0‐surface layer removed	G1‐mild degeneration
Straight	0.079 (SD 0.03)	0.157 (SD 0.13)	0.086 (SD 0.02)
Bend	0.147 (SD 0.05)	0.110 (SD 0.09)	0.108 (SD 0.08)
Arch	0.072 (SD 0.06)	−0.009 (SD 0.13)	0.122 (SD 0.12)

3.1. Strain measurements

Independent sample Kruskal–Wallis tests were carried out (p < 0.05) for all sample groups and for all categories of shape. The null‐hypothesis was that the distribution of strain would be the same across categories. Average strain values were calculated according to Equation 1 and sorted according to shape. These are outlined in Table 1 and listed alongside control data from earlier research carried out (Brown et al., 2019). For samples from the present study (under the heading “Mature (G1)”), average SZ strain decreased with increasing curvature (i.e., from straight, to bend to arch) however, that difference in strain was not significant (p > 0.05). Conversely, average DZ strain progressively increased with increasing curvature. The difference in DZ strain was significant between Straight and Arch shape categories (p < 0.05). All strains were positive (i.e., the G1 samples did not contract) and sample curvature (calculated in earlier studies as DZ minus SZ strain) was always in the direction of the SZ.

For all categories of shape, there was generally no significant difference in strain at the SZ or DZ across sample groups. However, two exceptions were noted and, in both instances, related to samples within the G0‐Surface Layer Removed category. This comes as no surprise as the G0‐Surface Layer Removed sample group, devoid of a strain limiting SZ, were able to swell limited only by the extent of lateral interconnectivity of the collagen network. In some instances, this resulted in SZ strains so large that tissue curvature was observed in the negative direction (i.e., samples curved towards the DZ). For the present study this demonstrates how complete excision of the SZ from pristine articular cartilage produces a different swelling response to samples whose SZ has undergone natural degenerative changes, such as the G1 sample group. For samples from the present study (i.e., G1), values of axial strain increase with increasing curvature (from straight, to bend to arch). That difference was not significant (p > 0.05).

Where it was possible to obtain a clear, full depth DIC image, zonal proportions were measured for groups G0‐Intact and G1‐mild degeneration. Independent‐Samples Mann‐Whitney U Tests were carried out for all zonal categories across both G0 and G1 categories (see Table 3). While there was a slight decrease in SZ thickness from G0 to G1, that difference was insignificant (p > 0.05). Conversely, for TZ (p < 0.05) and DZ (p < 0.05), the differences in zonal thickness between G0 and G1 were significant. Considering total thickness, there was a significant increase in tissue thickness from G0 to G1 which, reinforced the idea of increased swelling with tissue degeneration (Maroudas, 1979).

TABLE 3.

Average zonal thickness sorted according to tissue category

	SZ (µm)	TZ (µm)	DZ (µm)	Total (µm)
G0 (n = 12)	64 (SD 24)	170 (SD 42)	707 (SD 276)	940 (SD 320)
G1 (n = 17)	52 (SD 10)	124 (SD 28)	904 (SD 207)	1079 (SD 204)

Note that G0 tissue only measures G0‐intact samples as those in the G0‐Surfae Layer Removed category possess no SZ.

Strain calculations were similarly carried out according to Equation 2, comparing sample dimensions in the ‘as‐removed’ state and the ‘post‐saline’ state (see Table 4). That the tissue tends to swell and curve under physiological isotonic conditions suggests that some pre‐stress must exist that is only released upon excision from the bone.

TABLE 4.

Average SZ and DZ strain, calculated according to Equation 2 and sorted according to final shape category

	G1‐mild degeneration
	SZ	DZ
Straight (N = 3)	0.061 SD 0.011	0.071 SD 0.015
Bend (N = 14)	0.041 SD 0.041	0.151 SD 0.073
Arch (N = 13)	0.031 SD 0.049	0.167 SD 0.043

Equation 2 calculates strain based on the difference in sample dimensions between ‘as‐removed’ and ‘post‐isotonic solution equilibration’ states.

3.2. Morphology

The overall differences between the microanatomy of the healthy G0 samples versus the G1 samples were found to be as follows. For G0 samples, at the SZ, the surface was pristine and continuous and showed a thick layer of tangentially oriented discoidal shaped cells (Figure 4). With the G1 samples there was a mild disruption of the surface, and the layer of tangential cells was visibly reduced (Figure 4). The density of the fibrillar network also appeared to be greater in the G0 samples that were imaged compared with the G1 (Figure 4). In the DZ, the G0 samples compared with G1 also appeared to have a greater fibrillar network density and, accompanying this, a greater number of chondrocytes (Figure 5). The G1 samples in the mid‐to‐DZ showed a significant fibrous texture in the DIC images, while for the G0 samples the matrix texture appeared smooth (Figure 6). This fibrosity was correlated to the appearance of network destructuring and aggregation into larger less entangled bundles (Figure 6). In the underlying cartilage‐bone junction, comparing G0 samples and G1 samples, there was a major difference in the zone of calcified cartilage (Figure 7). Following the mild decalcification treatment, the G0 samples appeared to have diminished zones of calcified cartilage, whereas in the G1 samples it was more pronounced and displayed multiple mineralisation fronts, or tidemarks. For samples that remained straight (n = 3) lines of chondrocyte continuity appeared parallel (Figure 3), while for those which adopted a bend (n = 14) or arch (n = 13) morphology, these lines deviated from parallel (Figure 8). The degree of fibrosity, from assessing matrix texture (Figure 9) in the mid‐to‐DZ tissue was correlated with sample shape post‐swelling. The arch shapes, which were more intensely curved than the bend shapes, tended to exhibit more fibrosity than the bend shape samples. Insights from these qualitative assessments of morphology are discussed together with quantitative data in the next section.

FIGURE 4.

Clockwise from top left (a) DIC image of pristine articular cartilage SZ with no disruption to the surface layer. White arrow indicates thickness of SZ based on cell morphology. Scale bar approx. 40 µ. (b) DIC image of mildly degenerate articular cartilage SZ with visible disruption to the surface layer. White arrows indicate thickness of SZ based on cell morphology. SZ thickness appear less than the pristine sample and varies along the length of the image. Scale bar approx. 40 µ. (c) SEM image illustrating the fibrillar network within the SZ of pristine articular cartilage. Fibrils are highly entangled and densely packed, mostly tangential in directionality. (d) SEM image illustrating the fibrillar network within the SZ of mildly degenerate articular cartilage. Fibrils are tangential in alignment however less densely packed than the pristine cartilage [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 5.

Clockwise from top left (a) DIC image of pristine articular cartilage DZ. Chondrocytes appear densely clustered and singularly arranged. Scale bar approx. 40 µ. (b) DIC image of mildly degenerate articular cartilage DZ. Chondrocytes arranged in well dispersed vertical columns. Scale bar approx. 40 µ. (c) SEM image illustrating the fibrillar network within the DZ of pristine articular cartilage. Fibrils are highly entangled and densely packed with ambiguous directionality. (d) SEM image illustrating the fibrillar network within the DZ of mildly degenerate articular cartilage. Fibrils are radially oriented and less densely packed than their pristine counterpart [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 6.

Clockwise from top left (a) DIC image of pristine articular cartilage Mid‐to‐DZ. Chondrocytes in vertical columns surrounded by a smooth general matrix. Scale bar approx. 40 µ. (b) DIC image of mildly degenerate articular cartilage Mid‐to‐DZ. Chondrocytes arranged in vertical column surrounded by a general matrix exhibiting clear fibrosity in the radial direction. Scale bar approx. 40 µ. This aggregation of fibrils is visible at high magnifications of DIC imaging (c) SEM image illustrating the fibrillar network within the Mid‐to‐DZ of pristine articular cartilage. Fibrils are well dispersed and radially aligned. (d) SEM image illustrating the fibrillar network within the Mid‐to‐DZ of mildly degenerate articular cartilage. Fibrils are radially oriented and clustered into aggregated bundles (highlighted in red) [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 7.

(a) Deep zone image captured from G0 control block (Brown et al., 2019). No distinctive tidemark visible in the image. (b) Deep zone image captured from G1 cartilage‐on‐bone block. Repeating tide marks observed (red arrows) with the calcified zone encroaching on the overlying soft tissue. Scalebar approximately 50 µ [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 8.

Full depth DIC image of sample which formed an arch shape. Notice lines of chondrocyte continuity depart from parallel. Scalebar approximately 100 µ [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 9.

Exemplary DIC images used to categorise the ‘degree of fibrosity’ evident at the microscale via DIC microscopy from (a) slight fibrosity, (b) moderate fibrosity, (c) significant fibrosity. The table shows the count of samples, according to their shape category, and their corresponding level of fibrosity. Scalebar approximately 50 µ [Colour figure can be viewed at wileyonlinelibrary.com]

4. DISCUSSION

This is the first study to look at the ultrastructure of cartilage tissue and its swelling response in relation to matrix health. It is important to note that previous studies on cartilage swelling have attributed the swelling response to be a function of the tissue matrix ultrastructure but did not carry out any electron microscopic investigations (Maroudas, 1976; Maroudas & Venn, 1977; Narmoneva et al., 1999; Narmoneva et al., 2001; Setton et al., 1994, 1998). Thus, we believe our data and findings contribute to this important body of knowledge on cartilage mechanics. The present study shows that firstly, the swelling response of cartilage with degenerative changes is markedly different from that when assuming a layer‐wise interpretation of cartilage structure. With degenerative changes in the surface layer, the response is different compared with when healthy cartilage has its surface layer removed. The reason for this difference is due to the second finding in this study, that fibrillar matrix ‘lateral’ interconnectivity is a significant influence on cartilage swelling mechanics. Finally, this study has shown that there are consistent differences between healthy and degenerate tissue, in the microstructures of the DZ cartilage and zone of calcified cartilage. However, for a given post‐swelling shape, for example arch, the strain measurements are the same between groups. With microstructural differences between groups, but similar strain responses, the implication is that there would have been a difference in the stiffnesses of the matrices and hence different stresses at the same strain. This would be consistent with the findings of Nickien et al. (2017), where tensile testing was conducted on healthy and degenerate cartilage tissue, showing how the latter was less stiff and had a greater propensity to swell.

In a previous study it was hypothesised that removal of the SZ would impact the swelling response of articular cartilage in two ways. Firstly, through diminishing the gradient in fixed charge density from SZ to DZ. Second through altering matrix anisotropy by removing the densely packed, highly oriented collagen fibrils comprising the tangential layer (Setton et al., 1998) equivalent to the SZ defined in the present study. Upon removal of the SZ they noted that the samples still curved to the same extent as those with surface layer intact. It has also been proposed previously that the layered collagen ultrastructure bears the most influence over the tissues swelling induced curling behaviour, not the gradient in fixed charge density (Wan et al., 2010). Interestingly, Setton et al. (1998) discuss disruption to the SZ as being an early event in the pathogenesis of the degenerative joint disease, osteoarthritis, implying that removal of the SZ is perhaps comparable to naturally occurring degenerative change. Yet, data obtained in the present study demonstrates that the swelling response in cartilage with mild degeneration is different to that of healthy cartilage whose surface layer has been removed (Brown et al., 2019). Importantly, the mildly degenerate samples from the present study show that instead of having a ‘complete removal’ of the SZ, they instead have a relatively intact SZ, albeit diminished to various extents (Figure 4). Hence this diminished SZ displayed a smaller range of strain at both SZ and DZ compared with earlier data (Brown et al., 2019). That SZ disruption, not removal, might contribute to changes in swelling behaviour is consistent with early observations by Maroudas, who noted that in surface fibrillated specimens tissue hydration increased to the extent that changes to tissue fixed charge density were insufficient to explain why (Maroudas, 1979). Importantly, these findings demonstrate that swelling induced curvature cannot be controlled simply by a layer‐wise inhomogeneity as suggested by Wan et al. (2010) as, if this was true, one might expect swelling to be more similar between healthy samples with the SZ removed, and those with mild surface degeneration. An additional observation worth noting from the present dataset is the fact that following equilibration steps, all samples deformed within a single plane. This is notably different to the swelling response observed in earlier studies by this lab group where some samples distorted out of plane, resulting in a cork‐screw type morphology (Brown et al., 2019). The key difference between samples is the age and health condition of the tissue. In the earlier study, the tissue exhibited a distinct and mechanically strong cartilage deep zone. When imaged via SEM the collagen fibrils of the deep zone were not radially oriented, instead they appeared more heterogeneous in their organisation. The authors propose that this difference in collagen ultrastructure may contribute to the differences observed in morphological response between the present study and some of the tissue samples in the earlier study.

From the present study, the macro to ultra‐structural data suggest that differences in the swelling induced curvature between healthy and mildly degenerate tissue are, due to two types of disruption in the integrity of the collagen network. First is by thinning of and disruption to the SZ (Figure 4) and second visible fibrosity correlated with transformation of the collagen ultrastructure resulting in a less dense fibril arrangement (Figures 5, 6 and 9). These two disruptions and their effects are illustrated in the schematic provided (Figure 10). The schematic attempts to demonstrate how the elastic collagen network provides a restraining stress (σ_x and σ_y in Figure 10) to the internal osmotic swelling pressure. These stresses are a function of the magnitude of swelling pressure in conjunction with, as depicted, the degree of structural integrity. We have shown in previous studies that for bovine patellae G0 and G1 tissues, there is a significantly reduced matrix stiffness for the latter (Nickien et al., 2017). In light of this, comparing an intact surface layer with one that is diminished (compare Figure 10A,B), we expect the stiffness would be reduced in the latter and the resulting collagen network stresses in restraining the swelling pressure would be reduced also, thus σ’ < σ. Similarly, with collagen network destructuring (compare Figure 10C with D) the stiffness of the matrix is reduced thus we expect that the resultant restraining stresses in the network will be smaller for G1 samples compared with G0. Further insight may be established by interpolating values for these restraining stresses, using literature values for matrix stiffness. For example, what would these restraining stresses amount to if the present strain measurements were applied to known values of matrix stiffness? Tensile testing of bovine cartilage, of the SZ and DZ separately, show that the Young's Modulus is approximately 9MPa and 6MPa respectively (Verteramo & Seedhom, 2004). These estimates are consistent with those obtained by Nickien et al. (2017) from lateral tensile testing of planar sections of adult bovine patella cartilage, similar to the tissue used in the present study. Their data suggests that the stiffness modulus at the DZ was 7 and 2 MPa for G0 and G1 tissue respectively. Thus, if we consider as an example the DZ strain from the present data set for an arched tissue swelling response, comparing G0 and G1 (Table 1), the strains are similar at around 0.3. However, considering the stiffness difference between G0 and G1, the stress in the matrix could possibly be around 2.1 MPa (0.3 × 7 MPa) versus 0.6 MPa (0.3 × 2 MPa) respectively.

FIGURE 10.

The inherent stresses in the collagen network, that provide restraint to swelling resulting from a given osmotic pressure (π), is depicted in the figure. The stresses are shown as a series of sigmas (σ), for the horizontal and vertical directions (x and y respectively), and the network restraint stresses are presented for four scenarios. (a) The intact G0 surface zone (SZ) is compared with that of (b) the SZ of the G1 samples. The stresses in the network shown in (b) would be less restraining due to the diminished surface layer as reported in this study. In the deep zone (DZ), for (c) the intact cartilage the dense and well interconnected matrix would provide a larger restraint to swelling, than (d) the matrix with a destructured network [Colour figure can be viewed at wileyonlinelibrary.com]

While it is important to note that these estimates of stress are unverified, the significance in making them is two‐fold. Firstly, the difference in stresses between G0 and G1, highlights the significant role of the collagen network, at the ultrastructural sub‐micron length scale, in being a primary controller of the cartilage macro‐scale swelling response. That the sub‐micron destructuring of the network can cause tissue‐level changes in stiffness, was shown in a fibre‐network computer model, where nodal connections were disrupted to simulate network destructuring (Bilton et al., 2018). Hence the restraint from the collagen network to osmotic pressure, and how network destructuring results in a greater volumetric expansion, we believe has been shown and illustrated in the present study. Further, on degeneration of the collagen network, Maroudas and Venn (1977) determined that the free swelling behaviour of cartilage was influenced by the extent of degeneration or fibrillation of the tissue. The increased swelling observed in degenerate tissue was associated with changes to the collagen matrix, rather than being determined by an alteration to the tissue fixed charge density (Maroudas, 1979). The findings of both Bank et al. (2000) and Basser et al. (1998) corroborate this suggestion with both research groups determining that stiffness of the collagen network determines propensity to swell. Importantly, Bank et al. (2000) determine that reduced stiffness of the collagen network is a consequence of degradative change to the collagen fibrils themselves. While this neglects to acknowledge the highly interconnected collagen ultrastructure, a feature that has been well studied by this lab group (Broom, 1982; Broom et al., 2001; Brown et al., 2019; Chen & Broom, 1998; Nickien et al., 2013, 2017; Thambyah & Broom, 2007; Workman et al., 2020), it does highlight the potential for samples to exhibit similar strains whilst experiencing different internal residual stresses.

The second point of significance of estimating network stresses is that it may provide some insight into the residual stresses within cartilage, particularly in the DZ. The volumetric swelling of the DZ cartilage when detached from bone, may be interpreted as the potential strain from the residual stress in the collagen network when it is attached to bone, and subject to osmotic pressure (Fry & Robertson, 1967; Maroudas, 1976). In our own studies, this is evidenced even under both isotonic and hypotonic equilibration conditions and can be observed in the strain data presented in Tables 1 and 4. That the tissue tends to swell and curve when equilibrated in isotonic solution suggests that some pre‐stress must exist which is realised upon release from the bone and equilibration in solution. This corroborates the findings of Canal Guterl et al. (2010) who when studying the electrostatic and non‐electrostatic contributions of proteoglycans to the equilibrium modulus of bovine cartilage found that in samples where proteoglycans were depleted (90.4% −97.7% depletion), samples excised from the underlying bone still curled. The authors presume this to be evidence of residual stresses in the collagen matrix itself which, is released when cut. Hence if, as suggested, there are reduced residual stresses experienced by the collagen network at the cartilage bone junction for G1 samples compared with G0 samples, it may correlate with morphological differences seen in the zone of calcified cartilage (Figure 7). The idea that residual stress of the collagen network at the cartilage‐bone junction decreases with age agrees with data reported by Kempson (1982), suggesting that the stiffness of DZ cartilage tissue decreases continuously with age. Since for the degenerate samples, the change in morphology appears to be a build‐up of the calcified layer, such a stiffened interface (Hargrave‐Thomas et al., 2013) may be a biomechanical response to the reduced DZ cartilage internal stresses given that remodelling of the subchondral bone is associated with abnormal mechanical stresses at interface (Donell, 2019).

The following section addresses some important limitations to the study. Comparing these microstructural and ultrastructural observations with strain, SZ or DZ assessment (to assess the extent to which the respective zones were ‘present’) as well as degree of fibrosity (Figure 9) aided in clarifying the multi‐scalar implications on the swelling induced curvature. Figure 9 provides a count of sample shape according to their respective degree of fibrosity. Generally, those with a more intense curvature tended to exhibit more fibrosity. However, in some instances this was not the case. For example, three samples remained straight yet they exhibited moderate‐to‐severe fibrosity at the microscale. Of interest is how SZ strains of G1 samples that remained straight displayed an average strain of 0.067, which was greater than those categorised as bend (0.042). For these samples that remained straight, SZ thickness was less than that of those samples that adopted the bend morphology. This may suggest that for the few of the G1 samples which remained straight, it may be due to the diminished effect of the SZ in its strain limiting function, resulting in a surface layer swelling to the extent that the sample appeared straight. Similarly, two samples in the G1 group adopted a ‘Bend’ morphology, yet exhibited the most intense degree of fibrosity. Here, the SZ was present although DIC images reveal fibrillation and disruption to the SZ which, at the sub‐microscale correlated with a decreased fibrillar density (Figure 4). However, for these samples that had the bend morphology, SZ thickness was greater than those samples that remained straight, suggesting that for the former, in spite of disruptions to the tissue ultrastructure there is a greater resistance to transverse swelling due to the depth of tangentially aligned fibrils. Ultrastructural details of mildly degenerate matrix (compare Figures 5D and 6D) elucidate how structure can change in radial depth. It is possible then that when extracting a cartilage strip from the osteochondral block, the depth of the cut can influence the swelling induced curvature. However, such effects are difficult to measure, especially since the strip thickness after cutting was not different across samples.

We also acknowledge that the cellular response to the experimental methods has not been addressed. Indeed, the focus of the present study has been limited to the intricate structural detail of the collagen network. Yet knowledge of the tissue function must include an understanding of the cellular function. As such, this is an area that we hope can be explore in future studies such that a comprehensive multi‐scalar understanding of the tissue mechanics might be established from the cellular domain through to the gross tissue scale.

In conclusion, this microstructural and ultrastructural investigation revealed that the key differences influencing the tissue swelling strain response was (1) the thickness and extent of disruption to the surface layer and (2) the amount of fibrillar network destructuring, highlighting the importance of the collagen and tissue matrix structure in restraining cartilage swelling. We hope that studies like the present are able to contribute to the development of biomechanical models that aim to capture the multi‐length scale tissue mechanics of cartilage in relation to its delicate yet mechanically significant ultrastructural features.

Brown, E.T.T. , Simons, J.M.L.J.W. & Thambyah, A. (2022) The ultrastructure of cartilage tissue and its swelling response in relation to matrix health. Journal of Anatomy, 240, 107–119. 10.1111/joa.13527

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.