On how nucleus-endplate integration is achieved at the fibrillar level in the ovine lumbar disc


Neil D. Broom, PhD, Department of Chemical and Materials Engineering, University of Auckland, Auckland, New Zealand. T: + 64 9 9238974; F: + 64 9 3737463; E: nd.broom@auckland.ac.nz


The intervertebral disc nucleus has traditionally been viewed as a largely unstructured amorphous gel having little obvious integration with the cartilaginous endplates (CEPs). However, recent work by the present authors has provided clear evidence of structural cohesion across the nucleus-endplate junction via a distinctive microanatomical feature termed insertion nodes. The aim of this study was to explore the nature of these insertion nodes at the fibrillar level. Specially prepared vertebra-nucleus-vertebra composite samples from ovine lumbar motion segments were extended axially and chemically fixed in this stretched state, and then decalcified. Sections taken from the samples were prepared for examination by scanning electron microscopy. A close morphological correlation was obtained between previously published optical microscopic images of the nodes and those seen using low magnification SEM. Progressively high magnifications provided insight into the fibrillar-level modes of structural integration across the nucleus-endplate junction. The closely packed fibrils of the CEP were largely parallel to the vertebral endplate and formed a dense, multi-layer substrate within which the nodal fibrils appeared to be anchored. Our idealised structural model proposes a mechanism by which this integration is achieved. The nodal fibrils, in curving into the CEP, are locked in place within its close-packed layers of transversely aligned fibrils, and probably at multiple levels. Secondly, there appears to be a subtle interweaving of the strongly aligned nodal fibrils with the multi-directional endplate fibrils. It is suggested that this structural integration provides the nucleus with a form of tethered mobility that supports physiological functions quite distinct from the primary strength requirements of the disc.


The intervertebral disc, consisting of three primary functional regions – the nucleus, annulus, and cartilaginous endplates (CEPs) – provides a strong but flexible linkage between adjacent vertebrae. The nucleus is surrounded by the concentric annular layers or lamellae, with each comprising parallel arrays of collagen fibres crossing obliquely at alternating angles between adjacent layers.

The annulus and nucleus are contained superiorly and inferiorly between the CEPs, the latter in turn being structurally linked to the vertebral bodies via the vertebral endplates (VEPs). Under compression the nucleus, enclosed by the annulus, is loaded hydrostatically and by expanding against the walls of the annulus transfers the compressive load into the annulus (Coventry et al. 1945; Nachemson, 1975; Humzah & Soames, 1988; Cassidy et al. 1989; Adams et al. 1996; Hukins & Meakin, 2000; Urban et al. 2000; Adams, 2004).

Our recent investigation (Wade et al. 2011) into the microstructural and micromechanical properties of the nucleus has shown that the highly convoluted fibrosity within the nucleus actually exhibits endplate-to-endplate continuity. This continuity was revealed using a novel ring-severing technique to dislocate the influence of the annular fibres and then applying a tensile load to suitably prepared vertebra-nucleus-vertebra samples. The effect was to draw out the normally convoluted nuclear fibres into alignment and thus enhance their optical imagability. We also showed that this nuclear fibrosity is structurally integrated with the CEPs and is capable of withstanding significant tensile forces. Although the optical imaging technique (differential interference contrast) that we used in this recent study clearly demonstrated that the fibres of the nucleus anchor into the endplate via discrete, clearly defined insertion nodes, light microscopy cannot reveal the details of this anchoring system at the fibrillar level.

Ultrastructural studies that have looked at the fibrous architecture in detail have largely concluded that there is no significant attachment between the nucleus and the CEP (Inoue & Takeda, 1975; Takeda, 1975; Inoue, 1981). These authors also described the fibres of the CEPs as running parallel with the VEPs, with the nucleus containing a loose disorganised meshwork of fibres. More recent ultrastructural investigations have tended to focus largely on the overall organisation of the disc matrix and the properties of the fibres of the matrix, or on the integration of proteoglycans with the fibres (Buckwalter et al. 1976, 1985; Ishii et al. 1991; Akhtar et al. 2005; Aladin et al. 2010).

Having recently demonstrated that a substantial degree of structural integration does actually exist between the nuclear fibres and the disc endplate (Wade et al. 2011), the aim of this new investigation was to explore this integration at an ultrastructural level.

Materials and methods

Ovine lumbar spines dissected from freshly slain mature animals (ewes) were wrapped in plastic film and stored at −20 °C for no longer than 3 months. The ovine lumbar disc was chosen for this study as it has been shown to be similar to the human lumbar disc with respect to biochemical, structural and biomechanical properties (Wilke et al. 1997a,b; Alini et al. 2008; Wilke, 2008). We are aware that ovine discs do not possess a ring apophysis as is present in the human disc (Pfeiffer & Pfeiffer, 2006). However, we make the assumption that the mechanics of the nucleus are similar enough between the species to justify our use of this model system to explore the primary structural features of the disc.

In preparation for testing, the extraneous soft tissues and posterior elements were sawn off each motion segment. The spine was securely clamped while these elements were being removed to avoid any accidental overloading and thus damage to the discs. The vertebrae were then bisected centrally to isolate the discs and their adjacent vertebrae.

A sagittal slab of tissue consisting primarily of vertebra-disc-vertebra and anterior and posterior annular segments was carefully sawn from each motion segment while in its frozen state (Fig. 1A,B). Using a scalpel blade, both annular elements were severed (see dashed arrows in B), their effective dislocation being indicated by a sudden large extensibility of the remaining nuclear material (Fig. 1C). We termed this the ‘ring-severed’ state (Wade et al. 2011).

Figure 1.

 (A–C) Schematics illustrating the stages involved in preparing and stretching a ring-severed sample. The whole disc is shown in (A) with the approximate location from which the sagittal slab was obtained. Schematic (B) shows the extracted sagittal slab with the dashed arrows indicating the sites of ring-severing of the posterior and anterior annulus to isolate the nucleus. The resulting ring-severed sample is shown in its axially extended state in (C) (arrows indicate direction of loading).

Each ring-severed sample was then statically loaded and stretched to around four to five times its original disc height, this amount of axial stretch having been shown previously to be optimal for drawing the heavily convoluted nuclear fibres into alignment without causing their rupture (Wade et al. 2011). A force of around 10–20 N was required to stretch the samples to this extent, depending on the sample dimensions. While held in this extended state by maintaining a minor load of about 2 N, the sample was chemically fixed in 0.1 m formalin for 3 days followed by decalcification in 10% formic acid for 14 days, and then rinsed thoroughly in cold running water.

To aid in the appropriate choice of location for the SEM studies 30-μm-thick sagittal sections were first obtained by cryosectioning. These were then wet-mounted on slides and examined using differential interference contrast optical microscopy (DIC). Once a suitable view of the nucleus-endplate junction region was obtained under DIC from these thin sections, immediately adjacent thick slices (∼1 mm) in the same orientation were removed with a sharp blade for SEM. To enhance fibrillar clarity at their exposed cut surfaces, these thick slices were digested in bovine testicular hyaluronidase (Sigma Type I-S, 400–1000 units mg−1, at a ratio of 1.25 mg mL−1 in 0.1 m sodium acetate and 0.1 m sodium chloride pH 5 buffer solution, giving approximately 875 units mL−1) for 3 days at 37 °C to remove the proteoglycan component. They were then dehydrated in ethanol, critical point-dried, vacuum-coated with platinum and then examined by SEM.

The idealised structural model was drawn using a computer-aided design graphics package creo elements/pro 5.0 (PTC Needham, MA 02494).

A total of 38 ring-severed and stretched samples from 14 lumbar spines were used in the study. Twenty-three normal discs from 10 spines were processed for analysis by optical microscopy. Another 15 normal discs from four spines were processed for analysis by SEM only.


A typical DIC image of the nucleus-endplate region is shown in Fig. 2. Although fibril-level detail is not visible, a clear fibrosity resulting from the general alignment of the aggregated nuclear fibrils (i.e. fibres) is readily imaged. These fibres can be seen entering the CEP via the repeating nodal attachment points as was first described in our recent study (Wade et al. 2011). The boundary between the CEP and the VEP is also clearly visible. When a comparable region1 is imaged under SEM at low magnification these morphologically distinct attachment nodes are again readily recognised (Fig. 3). In this same SEM image there is also a clear morphological delineation between the nucleus proper, the CEP and the VEP. Importantly, the physical dimensions of the various features in the images can be seen to be similar, thus enabling us to correlate the fibrillar-level detail obtained in the present study with that of our previous microscopic study of the nucleus and endplate using DIC optical imaging (Wade et al. 2011).

Figure 2.

 Representative DIC image of the nucleus/endplate region, similar to that previously reported by Wade et al. (2011). Note how the nuclear fibres have been drawn into alignment in the direction of extension and anchored within the cartilaginous endplate (CEP) at the two discrete attachment nodes visible in this image. The boundaries between the nucleus (N), the CEP and the vertebral endplate (VEP) are indicated by arrows.

Figure 3.

 Low magnification SEM view of the nucleus endplate region. The aligned fibres of the nucleus (N) are visible, and the distinction between nucleus, cartilaginous endplate (CEP) and vertebral endplate (VEP) is indicated with arrows on the right. The two attachment nodes in this field of view are also marked with arrows.

Under SEM at progressively high magnifications (Figs 4–9) the individual fibrillar elements become clearer in both the nodal regions and the nucleus proper. The fibrils can be seen to be gathered into a single focal point within the CEP. In Fig. 4A (enlargement of boxed region in Fig. 3) as the fibrils approach the focal tip they appear to leave the sagittal plane of cutting by curving abruptly both to the right (arrows at X) and into the bulk of the matrix (arrow at Y), i.e. into the page. This suggests that the actual cutting plane of the sample section has intersected its structure to one side of the axial plane of symmetry of the node. In Fig. 4B the fibrils higher up the node lie in part within the sagittal cutting plane (see X) but then turn out of the cutting plane near the focal tip (see Y). This suggests that this particular node has been sectioned approximately through its axis of symmetry. Figure 4C illustrates again the pronounced curving of the nodal fibrils in the focal region.

Figure 4.

 (A) Enlarged view of the boxed region in Fig. 3 showing the attachment node in which the vertically aligned fibres of the nucleus can be seen inserting into the close-packed, transversely aligned fibres of the CEP (and captured in cross-section), which in this view appear to be turning into the page. (B) Example of attachment node that has been sectioned almost through its axial plane of symmetry. (C) Focal region of node in which fibrils curve into the endplate at the focal tip. Higher up, the fibrils in this nodal region have been captured more in cross-section, as indicated by X on the upper left hand side, suggesting that at this point they have deviated significantly from an axial orientation.

Figure 5.

 Enlarged view of the attachment node in Fig. 4B. The difference in alignment of the fibrils is obvious, with the out-of-plane endplate fibrils surrounding the largely in-plane nucleus fibrils at the node.

Figure 6.

 Enlargement of boxed region 1 of Fig. 5 showing the fibrous meshwork of the cartilaginous endplate (CEP). In this sagittal view the fibrils are captured in orientations ranging from near cross-section at (X) to almost transversely in-plane at (Y), confirming the multi-directional nature of the endplate fibrils.

Figure 7.

 Enlarged view of the cartilaginous endplate (CEP) in boxed region 2 of Fig. 5. At the lower edge of the image, the fibrils lie approximately parallel with the section plane. Towards the centre of the image, the fibrils are almost directly out of plane (near A). In the upper left corner of the image, the fibres appear to have been sectioned diagonally. In all cases, the fibrils can be seen to run roughly parallel to the vertebral endplate (VEP).

Figure 8.

 Enlarged view of boxed region 3 of Fig. 5 showing a more detailed view of the nodal tip. Endplate fibrils are clearly marked by X.

Figure 9.

 Enlargement of the boxed region of Fig. 8 showing transition from the close-packed fibrils of the cartilaginous endplate (CEP) into the less densely packed fibrils of the nodal focal region. Again, note especially the multitude of fibril directions present in the endplate proper (see X). Arrows indicate what may be node-endplate integrating regions.

The tip of the node in Fig. 4B is shown enlarged in Fig. 5 and further illustrates the complexity of the fibril morphologies in both the nodes and CEP, as well as hinting at the relationships between these distinct structures. The boxed region 1 is enlarged in Fig. 6 and shows densely packed fibrils, all lying in generally transverse planes (i.e. approximately parallel to the VEP) but varying in direction within these planes (e.g. compare regions X and Y in Fig. 6). The boxed region 2 is shown enlarged in Fig. 7 and, although a much more complex image, it does show the CEP fibrils mostly revealed in their near transverse planes and in a multitude of directions. Thus it would appear that the closely packed fibrils of the CEP are largely parallel to the VEP, forming a dense, multi-level, interwoven fibrillar mat.

The boxed region 3 in Fig. 5 is shown enlarged in Fig. 8 and further illustrates the contrasting morphologies between the node and its embedding CEP. Although we cannot exclude some artificial cleaving due to specimen preparation (critical point drying and sectioning), the nodal fibrils are clearly contained within densely packed curvilinear layers, which in this sectional view have been captured as they curve out of the section plane. The lower region (indicated by the two areas marked X) in Fig. 8 is clearly CEP, as it is continuous with the adjacent endplate structure shown in Fig. 6, but it is not immediately obvious where the nodal layers end and the endplate begins.

The boxed region in Fig. 8 is shown enlarged in Fig. 9. Importantly, this region exhibits minimal layer separation and is thus able to provide a clearer view of any transitional structure between the endplate proper (see lower regions marked by X) right up to the less densely packed fibrils of the nodal focal region. The several densely packed layers immediately above the identified endplate layers may either be endplate proper layers imaged in near cross-section or, alternatively, function as endplate-node integrating layers. The arrows in Fig. 9 indicate what are assumed to be endplate fibrils entering and branching within the focal region of the node. Figure 10 shows a view further up the node in which there is a gradual change from the strongly aligned nodal fibrils into a more irregular fibrillar morphology in the lower left of the image; again this is thought to represent the transition between node and endplate proper, suggesting that there is no discordant structural separation between the two.

Figure 10.

 Example of a region more distant from a node focal tip nearer the upper edge of the cartilaginous endplate (CEP). Note the gradual change from the strongly aligned nodal fibrils on the right of the image to a more irregular fibrillar morphology on the lower left.


The aim of this study was to investigate at the fibrillar level the nature of structural integration of the nucleus/endplate junction via insertion nodes, a microanatomical feature we have previously described at the light microscope level (Wade et al. 2011). Earlier investigators have suggested that there is little or no obvious connection between nucleus and endplate (Inoue & Takeda, 1975; Inoue, 1981). However, we suggest that this lack of evidence for such structural integration is probably due to the fact that the samples used in these earlier studies were not optimally viewed. In optical sections taken from the disc in an unstretched state, the natural convoluted state of the nuclear fibrosity obscures its structural arrangement and continuity. By contrast, the novel ring-severing, stretching and fixation methods employed in our recent study (Wade et al. 2011) had the effect of drawing this fibrosity into axial alignment from endplate to endplate, thus enabling it to be imaged in a single section plane.

Our recent study (Wade et al. 2011) also demonstrated that a substantial load could be transmitted across the nucleus-endplate junction through the nucleus after it had been completely isolated from the surrounding annulus, and that this arises from the structural integration of the nuclear fibrosity with the endplate via the insertion nodes. However, the fibrillar-level detail of these nodes is not visible at the light microscope level, hence this new SEM-based investigation.

Figures 6 and 7 show that the fibrils of the CEP run generally parallel with the VEP and form a dense overlay. This is consistent with the findings of earlier studies (Inoue & Takeda, 1975; Inoue, 1981). Low magnification SEM imaging (Fig. 3) shows the axially stretched fibrils of the nucleus clearly emanating from the nodes of attachment embedded within the CEP. This then raises the question as to how these nodes are connected with the endplate.

Our present ultrastructural data suggest that there are at least two mechanisms that facilitate some form of structural integration between the node and endplate. First, the nodal fibrils in curving into the endplate would presumably be locked in place within the close-packed layers of transversely arranged CEP fibrils, possibly at multiple levels. Secondly, there appears to be a subtle blending of the strongly aligned nodal fibrils with the multi-directional endplate fibrils, especially near the edges of the insertion nodes. This suggests that there is interlacing of the two populations of fibrils which would also add to the strength of integration.

Based on these results the series of schematics shown in Fig. 11A–F illustrates what we believe is an approximate structural representation of how the nodal feature integrates with the CEP. The overall configuration of the nodal fibrils is shown in schematic A. Note that these fibrils fan outwards from their focal tip as in the SEM images and have a characteristically curvilinear appearance. Three layers of fibrils are shown in A. Note, however, that for clarity the node is shown in half section only – in reality, we know that the nodes are roughly circular in cross-section.

Figure 11.

 3-D schematics showing proposed idealised fibril-level model describing nodal-endplate attachment. Note: for the purposes of clarity, only a half section of the node is presented. In schematic (A) three layers of nodal fibrils loop in and out of the endplate matrix (not shown). Schematic (B) is a reduced form of A showing the alternative transverse anchorage system. Schematics (C) and (D) show two different views of the node with just one layer of endplate fibrils to illustrate how the node and endplate fibrils interweave. In schematics (E) and (F) several layers of endplate fibrils have been added.

Based on the structural evidence available and, in particular, the sheer density of fibrillar elements we cannot determine whether the nodal fibrils loop continuously in and out of the endplate at their focal point as in schematic A, or turn out of the node and run transversely for some distance within the endplate as in the reduced schematic B. However, the basic concept of endplate anchorage would appear to be equally served by either or both structural models.

The endplate fibrils are arranged as an interwoven multi-directional mat, shown in single layer form in schematics C and D and in multi-layer form in schematics E and F. Two different viewing angles of the same model configuration are shown in these schematics to illustrate how the nodal fibrils interlock with the endplate layers. The inclusion of more densely interwoven endplate fibrils in schematics E and F obscures details of the nodal fibrils within the endplate. However, it is obvious that this interweaving could provide the required structural integrity across the node-endplate interface.


The present study has demonstrated experimentally that the optically resolvable nodes embody a fibrillar-level interconnectivity between the nucleus and endplate and are, at least in part, responsible for the previously described ability of this region of the disc nucleus to carry a substantial mechanical load (Wade et al. 2011). Because normal disc loading in vivo involves the nucleus functioning primarily in hydrostatic compression, it is unlikely that this interconnectivity contributes in any substantial way to the overall strength of the disc. Rather, having this heavily convoluted nuclear fibrosity anchored within the endplate would furnish the nucleus with a substantial degree of tethered mobility, enabling it to accommodate the significant shape changes associated with normal disc function. Further, interactions between the loosely tethered nuclear collagen fibrils and the proteoglycans would prevent the latter from ‘leaking out’ through the annular layers, thus maintaining the nucleus as a hydrostatically functioning entity fully contained by the annulus and endplate.


  • 1

    It should be noted that because DIC and SEM imaging required separate samples from any given disc (but taken from immediately adjacent regions) it was not possible to image a node optically and ultrastructurally in the same sectional view. However, the general morphological similarities of the nodes between Figs 2 and 3 are obvious.


The authors gratefully acknowledge funding support from both Medtronic Asia and the Wishbone Trust of the NZ Orthopaedic Association.

Conflict of interest

All three authors have no conflict of interest to declare.