Dr Neil Broom, Department of Chemical and Materials Engineering, University of Auckland, Private Bag 92019, Auckland, New Zealand. E: firstname.lastname@example.org
The purpose of this study was to investigate the structural mechanisms that create cohesion between the concentric lamellae comprising the disc annulus. Sections, 50–60 µm thick, were obtained using a carefully chosen cutting plane that incorporated the fibrous component in alternating lamellae as in-plane and cross-sectioned arrays. These sections were then subjected to microtensile stretching both across (radial) and along (tangential) the in-plane fibre direction, in their fully hydrated state. Structural responses were studied by simultaneously viewing the sections using high-resolution Nomarski interference contrast light microscopy. Additional bulk samples of annulus were fixed while held in a constant, radially stretched state in order to investigate the potential for interlamellar separation to occur in a state more representative of the intact disc wall. The study has provided a detailed picture of the structural architecture creating disc wall cohesion, revealing a complex hierarchy of interconnecting relationships within the disc wall, not previously described. Importantly, because our experimental approach offers a high-resolution view of the response of the interlamellar junction to deformation in its fully hydrated condition, it is a potentially useful method for investigating subtle changes in junction cohesion resulting from both early degeneration and whole-disc trauma.
The complex, multiply laminate architecture of the intervertebral disc wall (annulus fibrosus) presents a major challenge to investigators wanting to understand in greater depth the fundamental structural relationships that give rise to this organ's remarkable load-bearing qualities. Equally challenging is the need to have a more precise picture of the kind of structural changes occurring in the annulus that lead to its weakening with the related problems of abnormal disc bulging, nuclear displacement and prolapse.
Interpreting structural relationships in soft connective tissues such as the disc is made more difficult because of the large deformations associated with their normal functional behaviour. These tissues generally exhibit highly non-linear stress–strain responses, with the low-stress phase being a direct consequence of large-scale reversible alterations occurring in their fibrous architecture. Observation of the structural response to deformation of these tissues in their fully hydrated state is therefore a potentially valuable approach.
In an earlier study of the human disc annulus, Marchand & Ahmed (1990) employed partial dehydration (freeze dessication) to render the annular fibres more visible. Using a layer-by-layer peeling technique combined with low-resolution stereo-microscopy, these authors were able to describe a number of important structural features at the more macroscopic level; these included number of distinct layers, percentage of incomplete layers, modes of interruption within the laminate structure, and layer thickness. The main disadvantages of the approach used by Marchand and Ahmed were the limited level of resolution of the fibrous architecture and the unnatural semi-dried state of their tissues. At the whole organ level, Stokes & Greenapple (1985) used stereo-photometry to measure the strains in the outermost annular fibres of discs subjected to compression and torsion.
Although none of the above-noted studies has provided anything that represents a high-resolution observation of the annular structures in response to mechanical loading of the tissue in its fully hydrated state, Pezowicz et al. (2005) recently demonstrated that by combining Nomarski DIC imaging with a microtensile technique, such an observation of the annular architecture can be achieved. Specifically, they were able to apply tensile loading both along and transverse to the collagen alignment direction in fully hydrated, unmodified sections taken from within a single lamella. These studies revealed a complex hierarchy of interconnecting relationships within the collagenous array hitherto undescribed. The research presented in this paper uses the same hydrated tissue approach as used in the above intralamellar study but focuses instead on interlamellar structural relationships.
Materials and methods
Ox tails were collected fresh and immediately frozen for storage. Individual motion segments were sawn from the frozen tails, thawed, dissected to expose the intervertebral disc, and then blocks of annulus removed as shown in Fig. 1(A). These blocks were hand-trimmed to create a cutting plane that contained entirely within it one of the two oblique fibre bundle directions and the other approximately in cross-section, as shown schematically in Fig. 1(A). The trimmed block was then cryo-glued to a metal base and, using a freezing sledging microtome, frozen serial sections of 50–60 µm nominal thickness were obtained (Fig. 1B). The tissue had to be sectioned in this frozen state in order to obtain thin coherent slices. Small samples of outer annulus (dimensions ∼1.5 × 4 mm) suitable for microtensile testing were then trimmed from these cut, thawed sections. It should be noted that samples from this region of the annulus were chosen for testing because the outer lamellae are rather more distinct than those nearer the nucleus (Marchand & Ahmed, 1990; Tsuji et al. 1993). Small fabric tabs were glued to the samples in their long dimension so as to facilitate tensile stretching of the sections both across or along the in-plane fibre bundle direction, these being defined as radial and tangential, respectively (see Fig. 1A).
Each prepared sample was inserted in a microtensile device which mounted directly onto the rotating stage of an optical microscope fitted with a DIC imaging facility. This microtensile testing device has been described in detail in earlier papers (Broom, 1984, 1986). Importantly, this device incorporated a system that permitted controlled stretching of the sample within the space between two parallel optical glass surfaces separated by a spacer while bathed in physiological saline. During stretching, the progression of structural response was recorded as still images under conditions in which the tensile displacement was applied manually and discontinuously. Additional structural information was obtained by recording dynamically the tissue responses using a high-resolution video camera.
Another separate experiment involved the glutaraldehyde fixation of small blocks of annulus that had first been stretched either ∼50% or ∼100% in the radial direction. To achieve this, small flat-faced metal tabs were glued with tissue adhesive to each end of the sample, which was then mounted in a horizontal tensile test device. Following the application of the predetermined tensile strain, the grip and sample assembly was immersed in PBS containing 2.5% glutaraldehyde for a period of 2 h. This treatment ensured that the stretched state of the tissue block was permanently retained. Thin sections were then cut from the outer annular regions of these fixed stretched blocks in the same cutting plane as described in Fig. 1(A).
A total of ten annular blocks from four caudal discs taken proximally from two separate tails were prepared and sectioned in the frozen state. From these blocks 29 thin sections were cut and subjected to micromechanical testing, 11 in the tangential direction and 18 in the radial direction. A further four annular blocks were fixed while stretched, two at ∼50% and two at ∼100% radial strain. An additional 12 thin sections from the 50% stretched/fixed blocks and ten from the 100% stretched/fixed blocks were examined optically, but without further microtensile manipulation.
The unstretched state
The micrograph in Fig. 2 shows at higher resolution and in a fully relaxed state both the in-plane fibre bundles with their characteristic crimped morphology (see IP) and those incorporated as cross-sections (see CS). Whereas the in-plane fibre bundles coursed uninterrupted within the sample section, the bundles shown in cross-section were regularly compartmentalized (see sites marked in Z in Fig. 2). This general structural arrangement is depicted schematically below in Fig. 9(A). The thickness of the layers also increased from the outer to inner layers (see Fig. 1B) and in cross-section this was manifested as a distinct change in aspect ratio. It should be noted that all observations were conducted on slices obtained from the outer regions.
Response of thin sections to radial stretching
Progressive stretching of the sections in the radial direction indicated that the compartmental divisions (see Z in Fig. 2) contained radial bridging elements, e.g. BE1 and BE2 in Fig. 3(A), which passed between the bundles imaged in cross-section. The bridging element BE1 in Fig. 3(A) is shown at higher magnification in Fig. 3(B) and is clearly anchored back into the in-plane collagenous arrays, e.g. at the sites marked A. Evidence for the strength of this localized anchorage is seen from the fact that where it occurs there is a tendency for transverse separation of the in-plane collagen arrays to occur under the applied load, as demonstrated at sites Y1 and Y2 in Fig. 3(A). In some sites these radial bridging elements are seen to be a radial diversion of the in-plane parallel arrays, as shown at the arrowed site in Fig. 4(A).
Some sections revealed what appeared to be a partial or complete envelopment of the cross-sectioned bundles by these same bridging fibres, as shown in Fig. 4(B) where three enveloped sides of a bundle are marked with the arrows labelled BE. These same enveloping structures were also connected in a more distributed manner to the in-plane fibre bundles, as at Y in Fig. 4(B). Even where there was no complete envelopment of the cross-sectioned bundles, a more uniformly distributed set of interconnecting elements was present along the interface between these cross-sectioned bundles and the in-plane arrays, as is visible along the zone marked X3 in Fig. 3(A) and shown at much higher magnification in Fig. 3(C). Note that in this latter image these more distributed interconnecting elements appear to weave repeatedly into the edge of the in-plane arrays (see W in Fig. 3C). Although it is less clear from this same section how these interconnecting elements link into the cross-sectioned arrays (CS), we assume the mechanism is similar to that at W. The apparent difference will be an artefact of the plane of sectioning of the symmetrically alternating arrangement of fibre bundles comprising successive lamellae. Note also that under radial stretching these more distributed interconnecting elements also tended to create transverse clefts in the in-plane arrays, as seen at site Y3 in Fig. 3(A), possibly indicating that they penetrate to a substantial depth within the array.
Figure 5 was obtained from a ‘real-time’ video recording of radial stretching. Although partial junction separation has occurred at site X, the strong cohesion between the cross-sectioned bundles and the in-plane arrays at site Y has resulted in extensive splitting and subsplitting of the latter, rather than leading to an actual junction failure.
Commencing with the unloaded state (schematic Fig. 9A), schematic Fig. 9(B) illustrates most of the interconnecting structures observed in Figs 3–5. The detailed schematic in Fig. 9(B1) illustrates the bridging elements BE shown in Fig. 3(A,B). The detailed schematic in Fig. 9(B2) illustrates the complex mode of splitting and sub/sub splitting occurring within the in-plane arrays resulting from the in-pulling by the interconnecting elements (e.g. site Y3 in Fig. 3A). This repeating hierarchical splitting of the in-plane arrays is identical to that observed in intralamellar sections described in our first study (Pezowicz et al. 2005). The detailed schematic Fig. 9(B3) illustrates the more distributed interconnectivity between the cross-sectioned and in-plane bundles, as noted at site X3 in Fig. 3(A) and shown enlarged in Fig. 3(C).
More extensive radial stretching resulted in progressive fragmentation of the cross-sectioned bundles (Fig. 6) and thus provided further confirmation of the extent of interconnectivity between these and the in-plane arrays. For example, the site marked F in Fig. 6(A) identifies a series of fragments that have been partially separated from the parent cross-sectioned bundles (see CS), the forces achieving this fragmentation transmitted via their regular connectivity with the adjacent in-plane array (see IP). By contrast, in Fig. 6(B) the partially separated fragment F1 has remained closely attached to a radial bridging element BE as described previously in Fig. 3. The fact that the process of fragmentation has not led to any significant disconnection between the fragment and the bridging element suggests that it represents a significant mode of interconnection within the disc wall structure. Schematics Fig. 9(B,C) portray the process of fragmentation of the cross-sectioned bundles shown in Fig. 6.
It should be noted that the length of the collagen bundles imaged in cross-section will always be restricted to the through-thickness of the microtomed sections. Thus, the fibrous elements still connected to any fragments that have separated from the parent bundle will be of a length less than or comparable with that of the section thickness. This truncated fibre length is clearly illustrated at the sites marked X in Fig. 6(C) and is also illustrated schematically in Fig. 9(C).
Response of thin sections to tangential stretching
Stretching in the fibre alignment direction of the in-plane bundles provided further confirmation of structural principles revealed by the above radial stretching experiments and also yielded further insights into the nature of interlamellar relationships.
In Fig. 7, selective fibre bundle pullout at the gripped ends of the tissue sample induced a substantial shear displacement (shear direction indicated by half arrows) between the in-plane arrays (IP) located either side of the cross-sectioned bundles (CS). This shear was resisted by the bridging elements (BE) that separated the cross-sectioned arrays. Note especially the clear integration of these now loaded and skewed bridging elements back into the in-plane arrays (e.g. at X), as also noted in the radially stretched samples (see Fig. 3B). This mode of interconnectivity is illustrated schematically in Fig. 9(D).
Tangential stretching also confirmed the same component of interconnectivity as revealed by radial stretching (see Fig. 6B), namely that between the bridging elements and the cross-sectioned bundles. For example, in the tangentially stretched sequence shown in Fig. 8(A–C), the bridging element BE is clearly linked to the cross-sectioned bundle (CS) by taut collagenous elements (T), as revealed in Fig. 8(B,C). With increased stretching the forces transmitted by these same interconnections resulted in a progressive fragmentation of the cross-sectioned bundles involved. For example, the region marked F in Fig. 8(C) is one such fragment that has been separated from its parent cross-sectioned bundle (CS). This process is illustrated schematically in Fig. 9(E).
Structures in tissues fixed while radially stretched in bulk state
Several additional structural insights were obtained from the fixed-while-stretched tissues. Firstly, the overall morphology of the permanently stretched samples reveals a radial elongation of the cross-sectioned bundles (see CS in Fig. 10) with a related contraction of the in-plane arrays in their alignment direction (see IP in Fig. 10), the latter indicated by an increased acuteness of crimp. Secondly, in contrast to the extensive separation that developed at the junction between the cross-sectioned bundles and the in-plane arrays in the fresh hydrated sections when subjected to radial stretching (see Fig. 3), a variable level of cohesion was evident in the ∼50% stretched/fixed tissues (see region marked C in Fig. 10). In some regions there was substantial junction separation that exposed interconnecting elements (see S in Fig. 10). Figure 11 was obtained from a ∼50% stretched/fixed block and shows a severely skewed region. It reveals the dominant role of the bridging elements (BE), which, from their taut appearance and associated in-pulling of the in-plane arrays, indicates that they are firmly anchored within these arrays (see sites marked X in Fig. 11). This suggests that they play an important role in interlamellar cohesion.
Sections examined from the ∼100% stretched-and-fixed annular blocks revealed additional features of interest. Where junction separation had occurred (see Fig. 12) it was generally more pronounced than that observed in the ∼50% stretched blocks, as would be expected. However, the images in Fig. 13 (also ∼100% stretch) clearly show that in other highly stretched regions junction cohesion was still maintained between the in-plane and cross-sectioned bundles. An indication of the severity of loading imposed on the annular layers at these high stretches is evidenced by the degree of fragmentation of the cross-sectioned bundles, and specifically their severely riven appearance at the circled sites in Fig. 13(A). In Fig. 13(B) there is clear separation of the junction region at J3. However, at J2 in the same micrograph and at J1 in Fig. 13(A) there is little visible evidence that the in-plane arrays have been distorted by being in the direct line of the tensile forces responsible for producing the fragmentation of the cross-sectioned bundles.
As was demonstrated in our first paper dealing with intralamellar relationships (Pezowicz et al. 2005), the high-resolution visualization of fully hydrated sections of the annular wall, combined with simultaneous micromechanical stretching, provides a novel approach to investigating its complex architecture. This first study revealed a set of hierarchical interconnecting relationships within the aligned array of crimped collagen fibre bundles comprising a single lamella. The sectioning plane used in this previous study excluded any possibility of visualizing structural relationships that might be involved in creating interlamellar cohesion.
By contrast, the section plane employed in the present study has allowed interannular relationships to be investigated. We note, too, that in these sections one set of the alternating fibre bundles making up the concentric lamellar architecture of the annulus is always imaged in near cross-section. Thus, although the extended three-dimensional (3D) relationships in the annular wall are never fully revealed because of the relative thinness of the sections, information gained from them has enabled a 3D reconstruction to be made. Importantly, there was still a sufficient amount of structure contained within the fully hydrated sections to show important levels of interconnectivity exposed by tensile stretching in both the radial and tangential directions.
Although an important aspect of our research has been to investigate the biomechanical response of annular sections in their fully hydrated state it is important to emphasize that the micromechanical loading experiments cannot be used to investigate the intrinsic strength of the annular wall. The reason for this is that any sectioning of the disc wall will inevitably destroy to a substantial degree the crucial structural continuity present in the intact disc. Any mechanical contribution arising from structural continuity in the bulk tissue block and which lies outside the 50–60 µm section thickness will be lost. However, our ability to stretch the thin sections in their fully hydrated state and to view both the large-scale rearrangements occurring in the fibrous architecture and the processes of rupture provide new insights into the modes of interconnectivity within and between neighbouring lamellae. These modes of interconnectivity may well play a fundamental role in the maintenance of interlamellar cohesion.
It is also important to emphasize that although the obliquely orientated fibre bundles comprising the alternating lamellae are symmetrical with respect to the disc axis, the section plane we have chosen in this new study disguises this fact. The in-plane and cross-sectioned bundles imaged in Fig. 2 correspond to two different views of what are in fact symmetrical structures. For example, the transverse splitting under radial stretching of the in-plane arrays at Y3 in Fig. 3(A) should be equated with the fragmentation occurring in the cross-sectioned bundles shown in Fig. 6, again a direct consequence of radial stretching. Thus, we have two different views of the same process.
Several important questions relating to disc function need to be addressed. How strong are both the individual lamellar arrays and the interconnections that link concentric lamellae into a coherent wall structure in the radial direction? Although the previous intralamellar study by Pezowicz et al. (2005) demonstrated a dramatic difference in relative strength and related mechanical properties between the transverse and fibre-alignment directions, it could not address the question of absolute strength. Without preserving the in situ embedding of the lamellar fibres in the vertebral endplates, any absolute strength value remains elusive.
However, the annular blocks that were chemically fixed while simultaneously held in a state of radial stretch and then sectioned do provide an opportunity to examine the potential for interlamellar separation to occur that is more representative of the bulk wall. These experiments give some indirect indication of the degree to which interlamellar connectivity plays a role in maintaining the radial tensile strength of the disc wall. Although subsequent sectioning of these fixed blocks would still have disrupted tangential fibre continuity, the glutaraldehyde fixation ‘freezes’ in the structural rearrangement resulting from the imposed radial strain, and in our experiments these were approximately ∼50 and ∼100%. The thin sections taken from these fixed/stretched blocks have therefore enabled us to visualize those structural elements rearranged as a direct consequence of the radial strain applied to the bulk sample. Recalling that all of the observations of structure in the present study were confined to the outer layers, the considerable radial elongation of the cross-sectioned bundles observed in the stretched/fixed blocks (see Figs 10, 11 and 13) clearly confirms the in situ strength of the interconnecting elements.
The fact that there was variability of junction separation in these fixed-while-stretched sections suggests that interlamellar connectivity is dependent more on a localized rather than a homogeneous or dispersed mode of interconnectivity. A distinctive structural feature of the unstretched samples was the regular compartmentalization of the collagen fibre bundles within each lamella (see site Z in Fig. 2). The stretching experiments demonstrated that a major part of the interconnectivity between adjacent oblique bundles and adjacent parallel bundles is provided by linking elements passing between these compartments (e.g. see Figs 3B and 7). However, it is also apparent from the highly stretched and fixed samples that those regions which maintained a tight interlamellar junction did not necessarily coincide with the presence of interconnecting elements displaying evidence of being strongly anchored back into the adjacent arrays. For example, the regions marked J1 and J2 in Fig. 13(A,B) show a tight junction with clear evidence of fibres linking into the in-plane arrays but without creating the kind of in-plane separation seen for example at site J3 in Fig. 13(B). This would suggest that the anchorage maintaining the integrity of the junction imaged in these sections is at a different level outside the section plane and supports the discrete anchorage interpretation.
The schematics in Fig. 14 attempt to reconstruct the 3D picture of interlamellar connectivity derived from our study. The schematic in Fig. 14(A) shows the symmetrically oblique arrangement of four alternating lamellae (for convenience, numbered 1–4) but without any interconnecting elements included. The schematic in Fig. 14(B) shows the enveloping type of linking bundles in layers 1 and 3 with the intervening bundle in layer 2 largely uninvolved at this location. It is still not clear to us whether this enveloping mode involves the interconnecting fibres completing a return traverse. The schematics in Fig. 14(C,D) illustrate single traverse linkages between bundles in layers 1 and 3, and in 2 and 4, with additional sublinkages to bundles in the intervening layers 2 and 3, respectively. The general 3D configuration incorporating all of the above modes of interlamellar integration is represented in the master schematic in Fig. 14(E).
Although our imaging approach has revealed the rich collagenous architecture of the annulus, the presence and role of elastin cannot be ignored. Previous investigators have shown that the collagen component is intimately associated with elastin fibres (Buckwalter et al. 1976; Johnson et al. 1982). Using both histological and immunohistochemical techniques, Yu et al. (2002) have shown that elastic fibres are widely distributed throughout the entire disc, albeit in small amounts. With regard to the annulus, Yu et al. (2002) found elastic fibres running parallel to the collagen arrays within lamellae and in the cross-bridges between lamellae, and suggest that they might assist in the recovery of annular lamellae organization after deformation. Although the overall collagen content is readily identified in our Nomarski images, an unambiguous differentiation between the major collagen component and the minor elastin elements in some of the bridging structures is far more difficult to achieve. Histological preparations conventionally used to stain for elastin, e.g. orcein, require both prior fixation as well as the collagen and glycosaminoglycan components to be removed (Yu et al. 2002). Such an approach would conflict with the principal aim of the present study, which was to examine the response of the annular structures in their hydrated state.
A much earlier study by Schmorl & Junghanns (1959) recorded inward buckling of the inner lamellae of human discs. Cassidy et al. (1990) chemically fixed healthy canine discs while under compression. They observed a variety of deformation responses of the annular wall, noting that inward bulging of the inner lamellae was a common occurrence. Yasuma et al. (1986), Gunzburg et al. (1992), Tanaka et al. (1993) and Adams et al. (2000) have similarly shown that inward bulging can occur in human discs of all ages. Such modes of deformation of the annulus will inevitably impose radial tensile stresses on the interlamellar junction with the potential to disrupt interlamellar cohesion. Furthermore, a common disruptive feature of disc degeneration is the development of concentric clefts between adjacent lamellae, in effect a localized form of delamination (Hirsch & Schajowicz, 1953; Vernon-Roberts et al. 1997; Adams et al. 2002), and thus loss of interlamellar cohesion. Application of the radial stretching approach used in the present study should provide a more detailed picture of the severity of structural disruption associated with such delamination, and particularly in its earlier stages of development.
To conclude, this investigation presents a detailed picture of the structural architecture creating disc wall cohesion. Importantly, because our experimental approach provides a high-resolution view of the response of the interlamellar junction to deformation in the fully hydrated state, it offers a constructive framework for investigating subtle changes in junction cohesion resulting from both early degeneration and whole-disc trauma.
This research was supported by a research grant generously provided by the Auckland Medical Research Foundation. C.A.P. is grateful for leave granted to her by Professor R. Bedzinski of the Wroclaw University of Technology, Poland.