Dr Jing Yu, Laboratory of Physiology, Anatomy and Genetics, Sherrington Building, Parks Road, University of Oxford, Oxford OX1 3PT, UK. T: +44 (0)1865 272479/272431; F: +44 (0)1865 272469; E: email@example.com
The distribution of microfibrils was studied immunohistochemically in intervertebral discs taken from young normal human surgical cases and from the bovine tail. Co-localization of microfibrils and elastin fibres was examined by dual immunostaining of fibrillin-1 and elastin. Collagen fibre network orientation was studied by using polarized filters. A similar microfibrillar network was seen in both bovine and human discs with network organization being completely different from region to region. In the outer annulus fibrosus (OAF), abundant microfibrils organized in bundles were mainly distributed in the interterritorial matrix. In addition, the microfibril bundles were orientated parallel to each other and co-localized highly with elastin fibres. Within each lamella, co-localized microfibrils and elastin fibres were aligned in the same direction as the collagen fibres. In the interlamellar space, a dense co-localized network, staining for both microfibrils and elastin fibres, was apparent; immunostaining for both molecules was noticeably stronger than within lamellae. In the inner annulus fibrosus, the microfibrils were predominantly visible as a filamentous mesh network, both in the interterritorial matrix and also around the cells. The microfibrils in this region co-localized with elastin fibres far less than in the OAF. In nucleus pulposus, filamentous microfibrils were organized mainly around the cells where elastin fibres were hardly detected. By contrast, sparse elastin fibres, with a few of microfibrils, were visible in the interterritorial matrix. The results of this study suggest the microfibrillar network of the annulus may play a mechanical role while that around the cells of the nucleus may be more involved in regulating cell function.
The intervertebral discs (IVDs) lie between adjacent vertebrae in the spine. Their role is primarily mechanical. They withstand multiple forces including compressive and tensile loading and torsion during daily activity and, together with the spinal ligaments and facet joints, provide the spine with flexibility. There are 23 IVDs in the human spinal column, occupying about one-third of its total length. The unique mechanical properties of the IVD are essential for the correct mechanical functioning of the spinal column.
The mechanical properties of the IVD arise from the composition and network architecture of its extensive extracellular matrix (ECM). This structure is also important in ensuring transmission of the appropriate mechanical signals to the disc cells (Bruehlmann et al. 2004), and thereby influencing disc cell activity. The disc is mainly composed of collagen and proteoglycans. Anatomically, the disc is constructed with a central nucleus pulposus (NP) surrounded laterally by the annulus fibrosus (AF), and above and below by the endplates (Roberts et al. 1989). The proteoglycan concentration is highest in the NP and gradually falls towards the periphery of the AF; the collagen concentration gradient is in the opposite direction, concentration being greatest in the outer AF (OAF) and gradually decreasing towards the NP.
Collagen forms the main structural network of the disc. Bundles of collagen fibres in the AF form concentric lamellae. Within each lamella the collagen fibrils are arranged parallel to each other, but lie obliquely to the spinal axis, the direction reversing in adjacent lamella. Within the NP, the collagen forms a loose network of fine fibrils (Inoue & Takeda, 1975; Inoue, 1981). In addition to the collagen network, there is a network of elastin fibres which is an integral part of the structure of the IVD. As the disc deforms under various loads, the elastin fibre network is thought to play a significant role in maintenance of collagen organization and in the recovery of the disc size and shape after deformation (Yu et al. 2002).
Microfibrils are a structural element of the ECM. They have been found ubiquitously distributed in connective tissues and are reported to be organized in tissue-specific architectures (Sakai et al. 1986; Charbonneau et al. 2004). In elastic tissues such as blood vessels, ligament and lung, microfibrils are one of the components of elastic fibres, forming a meshwork around a central elastin core. But in some other tissues, such as the ciliary zonules of the eye, microfibrils form bundles without elastin. In primitive organisms microfibrils may be the only elastic element (Davison et al. 1995; Thurmond & Trotter, 1996). The importance of microfibrils in human tissue function has been emphasized by the clinical manifestations of Marfan syndrome (MFS) and congenital contractural arachnodactyly (CCA). These conditions are associated with mutations of the fibrillin-1 and fibrillin-2 genes, respectively, where fibrillins are the main structural components of microfibrils. MFS patients have abnormalities in the cardiovascular, ocular and skeletal systems (Pyeritz, 2000) and CCA patients have abnormalities mainly in the skeletal systems (Kitahama et al. 2000; Gupta et al. 2002; Kielty et al. 2002). Skeletal abnormalities include spinal deformities (kyphoscoliosis), joint laxity and contractures, overgrowth of the long bones and osteopenia (Kohlmeier et al. 1995).
Although immunohistochemical staining has demonstrated that elastin fibres are abundant and regularly organized in human IVDs and bovine tail discs (Yu et al. 2002, 2005), little is known about microfibril organization in the IVD. Our aim was to investigate the organization of microfibrils in human IVDs, particularly their relationship with elastin fibres, collagen fibres and cells. We anticipate that this work will provide a basis for investigations into the function of microfibrils in IVD physiology.
Materials and methods
‘Normal’ human disc samples were collected from two young patients. Patient no. 1 was 17 years old and underwent spinal surgery after an accident (discs of T12/L1 and L1/2). Patient no. 2 was a 12-year-old with a spinal tumour, a fibrillary astrocytoma extending from the pituitary to the sacrum, and in whom the surgery was intended to stabilize a secondary postural kyphosis (disc of L3/4).
We were unable to obtain large numbers of young and normal human discs because of constraints on the use of cadaveric human tissue in the UK. We were therefore forced to examine bovine caudal discs, which are accepted as a relatively good model for human IVD. They resemble human IVD in many respects, such as their cell phenotype (Horner et al. 2002) and their appearance and composition (Oshima et al. 1993). Importantly, the elastin fibre network organization has been found to be similar in both species (Yu et al. 2002, 2005). Therefore, bovine discs were used to provide further information on the organization of microfibrils and their relationship with elastin fibres, collagen fibres and cells in the IVD.
Specimen preparation for histology
Human disc specimens were collected at the operating theatre, wrapped to retain moisture, and kept at 4 °C for less than 4 h before dissection. Bovine tail discs were obtained from three adult steers (18–24 months old) within 5 h of slaughter, and the two uppermost discs were dissected from the vertebral bodies. Strips of 5–10 mm in thickness were dissected around the circumference from each specimen and snap frozen and stored at −70 °C until used.
Frozen sections 20 µm in thickness were cut with a cryostat microtome with transverse or tangential orientation in two planes as shown in Fig. 1. The sections were mounted on polylysine-coated microscope slides (VWR International Ltd) and stored at −20 °C for further analysis.
No specimens could be obtained from the endplate region because insufficient human material was available and the adult bovine endplate is too thin to excise. However, there was a clear distinction in structure between the inner region of the annulus fibrosus (IAF), with wide-spaced lamellae, and the less heavily hydrated and denser OAF. The regions selected for analysis are illustrated in Fig. 2 by using a bovine sample.
Dual immunostaining of microfibrils and elastin fibres
In the histochemical literature the distinction between elastin and elastic fibres, which are either composite structures of elastin and other glycoproteins or even composed entirely of microfibrillar glycoproteins, is sometimes not clearly made. Here we define elastin fibres as those fibres immunostained with an antibody against elastin. Microfibrils are defined as the fibrils immunostained with an antibody again fibrillin-1. Elastic fibres are defined as composite fibres composed of co-localized elastin fibres and microfibrils. Microfibrils and elastin fibre organization were revealed by immunostaining fibrillin-1 and elastin. All sections were fixed with 10% formalin for 10 min and then pretreated with hyaluronidase (Sigma H6254, concentration of 4800 U mL−1) overnight followed by collagenase (Sigma C5138, concentration of 30 U mL−1) for 30 min at room temperature to reduce the masking effect of glycosaminoglycan (GAG) and collagen.
Immunostaining procedures were conducted at room temperature, unless otherwise indicated. PBS (phosphate-buffered saline) was used for washing between each incubation and treatment. Sections were blocked with 6% normal donkey serum (NDS) in Tris buffer (50 mm Tris-HCl plus 10 mm calcium acetate) for 30 min before fibrillin-1 staining. An anti-human fibrillin-1 polyclonal antibody raised in sheep against the proline-rich region of fibrillin-1 was used as primary antibody (Kettle et al. 1999). After blocking, sections were incubated with anti-fibrillin-1 antibody in Tris buffer overnight at 4 °C, then incubated with cyanine Cy3-conjugated donkey anti-sheep IgG (1 : 100 dilution, Stratech Scientific Ltd, UK, cat. no. 713-165-003) for 30 min. Sections were then blocked again with 6% NDS (in PBS) for 30 min, then incubated with rabbit polyclonal anti-alpha human-elastin antibody (Biogenesis, UK, cat. no. 4060-1054) for 3 h. Finally, sections were incubated with FITC-conjugated donkey anti-rabbit IgG (1 : 100 dilution, Stratech Scientific Ltd, UK, cat. no. 711-095-152) for 30 min. Mounting medium containing DAPI (Vector Laboratory, cat. no. H-1500) was used.
Cell nuclei were not well stained by DAPI in sections pretreated with hyaluronidase and collagenase, suggesting that pretreatment can remove cells. In order to investigate microfibril organization in relation to cells and collagen fibres, some additional sections were double stained (as described above) without pretreatment, which gave a weak stain of both microfibrils and elastin fibres.
Negative control sections were incubated with normal sheep serum instead of fibrillin-1 antibody and normal rabbit serum instead of elastin antibody during all the staining procedures and showed no cross-reactivity.
Collagen fibres were visualized from their birefringence under crossed polarizing filters.
A fluorescent microscope with a polarized light filter combination was used to visualize and study microfibrils, elastin fibres, collagen fibres and cells in 2D images. For 3D imaging, confocal laser scanning microscopy (Zeiss Axiovert 200 with Microradiance LSM) and near-infrared femtosecond multiphoton microscopy (MPM) were used.
Single photon confocal laser scanning microscopy (CLSM)
Before collecting images, the iris was set to nearly the minimum aperture. The laser power, gain and offset were then adjusted according to the background of the different specimens. Images were collected as stacks of z series with a step size of 0.5 µm.
We used a modified femtosecond two-/multiphoton laser scanning system (Mulholland, 2004) configured for the non-invasive 3D imaging of thick biological specimens as recently described and detailed by Tirlapur et al. (2006). A 800-nm near-infrared laser wavelength was used for excitation of the fluorescent probes (Cy3 and FITC). Images were collected as stacks of z series with a step size of 0.15 µm.
Image processing and analysis
LaserSharp 2000, Imaris and Adobe Photoshop software were used to process images. Quantitative co-localization analysis for dual stained microfibrils and elastin was performed using the Imaris Co-Loc function.
In general, fibrillin-1 microfibrils were found in all the regions of the IVD, but the organization was very different in the OAF to the NP region. The organization was similar in human and bovine tissues.
Microfibrils in the OAF
Microfibrils were numerous in the OAF and we discuss their distribution with respect to the principal components of the tissue.
Relationships between microfibrils and elastin fibres
Figure 3 shows typical images of dual stained microfibrils and the elastin fibre network within a lamella in the human OAF. Microfibrils appeared abundant and relatively long (more than 80 µm) and also were well organized (Fig. 3a,d). The diameter of microfibrils has been previously indicated to be 10–12 nm (Greenlee et al. 1966; Bruns & Palade, 1968; Charbonneau et al. 2004). But the diameter of the microfibrils in Fig. 3(a,d) was 320–480 nm, suggesting that microfibrils actually were organized in bundles of primitive units. The microfibril bundles were orientated approximately parallel to each other (Fig. 3a,d). Immunofluorescent dual immunostaining of fibrillin-1 and elastin revealed that microfibril bundles and elastin fibres were similar both in morphology and in location (Fig. 3). Red dots (in Fig. 3a,d) and green dots (in Fig. 3b,e) represent microfibrils and elastin fibres orientated perpendicularly or obliquely to the plane of view. Thicker elastin fibres were found within a lamella (white arrows in Fig. 3e), which may have been formed from a bundle of elastin fibres organized in a ribbon structure. Merged images of microfibrils and elastin fibres clearly indicated that microfibrils highly co-localized with elastin fibres (Fig. 3c,f). By using the Imaris Co-Loc function, the degree of co-localization of microfibrils and elastin fibres can be more accurately ascertained. Results indicated that about 75% of microfibrils co-localized with elastin fibres (scatter plots not shown).
Figure 4 shows representative images of dual stained microfibrils (Fig. 4a) and elastin fibre (Fig. 4b) networks within a lamella of OAF in a bovine disc. Figure 4(c) is the merged image of Fig. 4(a,b). In bovine tissues, microfibril bundles also organized parallel to each other and co-localized with elastin fibres. Such characters of microfibrils closely resemble that seen in the human IVD, as discussed above (e.g. Fig. 3).
We defined the interlamellar space as the area between adjacent collagen lamellae and the intralamellar space as the area within an individual collagen lamella. Figure 5 shows typical images of dual stained microfibrils and elastin fibre network in the interlamellar space (yellow arrows in Fig. 5) of the human IVD. Both microfibrils and the elastin fibre network were more densely organized in this region than those seen in the intralamellar space (yellow arrowheads in Fig. 5). It was also noticeable that the size of the interlamellae space varies between adjacent lamellae, e.g. interlamellae spaces in Fig. 5(a,b) are narrower than those in Fig. 5(d,e). Merged images of microfibrils and elastin fibres also indicated a high degree of co-localization of microfibrils and elastin fibres in the interlamellar space (Fig. 5c,f) in general. However, in the relatively wider interlamellar space (Fig. 5d,f), microfibrils appeared to form a network linking adjacent lamellae together (white arrows in Fig. 5d.f). In addition, thicker elastin fibres can be seen both in interlamellar (white arrowheads in Fig. 5b) and intralamellar space (white arrows in Fig. 5b).
Relationships between microfibrils and the collagen fibre network
Figure 6 shows typical images of the microfibril network of human (Fig. 6a–d) and bovine (Fig. 6e–h) IVDs compared with the collagen fibre network observed under crossed polarizers. Within a lamella, parallel microfibril bundles (Fig. 6a,c,e) were aligned in the same direction as collagen fibres (Fig. 6b,d,f) in both human and bovine IVDs. The crimped morphology of microfibrils closely resembled the crimped morphology of collagen fibres (Fig. 6a–f), suggesting a strong association of the microfibril network with the primary structural network of collagen fibres. In the interlamellar space (yellow arrows in Fig. 6e–h) of the bovine discs, microfibrils were also more densely packed in the region, which was similar to that seen in human tissues as reported above (Fig. 5). In addition, microfibrils were found to be densely distributed in the bridges crossing the intralamellar spaces (white arrows in Fig. 6g). The orientation of collagen fibres in these crossing bridges appeared to be different (white arrows in Fig. 6h) from that seen within the same lamella. The association between the microfibril network and collagen fibres was very similar to that previously reported of elastin fibres and collagen fibres in the OAF region of the IVDs (Yu et al. 2002).
Microfibrils in the pericellular matrix
Figure 7 shows typical images of microfibrils within a lamella of the human IVD in relation to the cells and collagen fibre network. Bovine tissue showed a very similar pattern (not shown). Cells in the OAF region were elongated (Fig. 7a,f), as found in earlier reports of disc cell morphology (Bruehlmann et al. 2002; Johnson & Roberts, 2003). However, cells in Fig. 7(a) appeared to be different in size and orientation. Figure 7(b) and (d) are, respectively, a micrograph of collagen fibres and an image of immunostained microfibrils on the same plane as the cells in Fig. 7(a); Fig. 7(c) and (e) are the merged images of collagen network (Fig. 7b) and microfibrils (Fig. 7d) with the cells (Fig. 7a), respectively. It is clear that cells in the OAF were aligned in the same directions as collagen fibres and microfibrils. This orientation gave rise to the impression of heterogeneity of cells apparent in Fig. 7(a). In contrast to other regions, no microfibrils were present in the pericellular matrix in the OAF (Fig. 7e,f).
Microfibrils in the IAF
Typical images of dual stained microfibrils and elastin fibre network in the IAF region of the human disc are shown in Fig. 8 and bovine disc in Fig. 9. Generally, there is a transition in the appearance of the microfibrils across this region from that described above in the OAF to fine and filamentous structures (Figs 8a,b and 9a). Filamentous microfibrils are distributed not only in the ECM (Fig. 8b) but also around cells (yellow arrows in Figs 8a and 9a), although cell nuclei can barely be stained by DAPI after hyaluronidase pretreatment (see Methods). The white dashed lines in Figs 8 and 9 indicate the border of adjacent lamellae. Within a lamella, microfibrils formed a spider web-like network (Figs 8b and 9a). By contrast, the elastin fibres were still orientated nearly parallel to each other (Figs 8c and 9b) but at an angle (about 60°) to the fibres in adjacent lamella. Merged images (Figs 8d and 9c) indicated the microfibrils were less co-localized with the elastin fibre network compared with that seen in the OAF region as reported above (e.g. Figs 3c,f and 4c). In addition, some microfibrils appeared to link the elastin fibre network (Fig. 8d) within a lamella, as observed in the intralamellar space of the OAF (Fig. 5f).
Microfibrils in the NP region
Data on the NP region were obtained only in bovine tissue, as usually little human NP tissue can be removed from the surgery; even when present its original orientation can easily be distorted during surgical removal. Figure 10 shows typical images of immunostained microfibrils in relation to cells and elastin fibres in the NP region of a bovine disc. In this region, most of the microfibrils were clearly organized around the cells. A merged image (Fig. 10b) of triple stained microfibrils, elastin fibres and cell nuclei indicated that microfibrils were largely distributed in the pericellular territory where only few elastin fibres were detected (white arrowheads in Fig. 10b). Conversely, elastin fibres were located mainly in the interterritorial matrix, with a few microfibrils (white arrows in Fig. 10b).
The results presented here show that there is a well-organized microfibril network in both human lumbar and in adult bovine caudal intervertebral discs. The organization of microfibrils in bovine tail discs is very similar to that in the human discs. Taken together with the previous finding that the organization of the elastin fibre network is similar in the two tissues, this work supports the use of bovine tail discs as a reasonable model for the study of elastic fibre organization.
There is a striking difference in microfibril network organization between the AF and the NP. In the OAF, the microfibril network is predominantly in the interterritorial matrix. It consists of abundant and well-organized bundles of microfibrils which are relatively long (up to 80 µm) and aligned parallel to each other (Figs 3a,d and 4a). This organization is very similar to that seen for the elastin fibres (Figs 3b,e and 4b) and analysis of merged images of microfibrils and elastin fibres indicated a high degree of co-localization (Figs 3c,f and 4c). In addition, both networks are aligned in the same direction as the collagen fibre bundles (Fig. 6), to the extent that even the crimped morphologies are similar. The close association of microfibrils and elastin fibres with the primary structure of collagen fibres suggests a close functional relationship.
In the NP region, however, the microfibril organization is completely different. Here microfibrils are mainly seen around the cells whilst elastin fibres are found primarily in the interterritorial ECM (Fig. 10). The pericellular microfibril organization resembles the capsules around chondrocytes reported in articular cartilage, even though elastin was not found in articular cartilage tissue (Keene et al. 1997). It was suggested by these authors that microfibrils around the cells may influence cell activities. Recent work on transgenic mice has suggested that microfibrils are able to regulate TGFβ activation by stabilizing growth factors (TGFβ and BMPs) within the microfibril network (Neptune et al. 2003; Ramirez et al. 2004; Dietz et al. 2005).
In the IAF region, the microfibrils are seen both in the matrix and around the cells (Figs 8 and 9). As well as forming mesh capsules around the cells (Figs 8a and 9a), fibrils are also visible in the ECM (Fig. 8b) but co-localized with the elastin fibrils to a much lesser extent than in the OAF. As in the OAF, elastin fibres are aligned parallel to each other within a lamella (Figs 8c and 9b) but the microfibrils formed a filamentous meshwork (Fig. 8b). The changes of morphology and distribution of ECM microfibrils between the IAF and OAF may be related to the differences in mechanical environment and cell population. The IAF is a transitional zone between OAF and NP. In this region, in young tissue the predominant force is an isotropic pressure (McNally & Adams, 1992). In the OAF the fibres are exposed to tensile forces directed along the collagen fibres in order to support the bulging of the disc under compression. We suggested earlier that the elastic proteins may have an indirect mechanical role in influencing the alignment of collagen fibres and in directing re-alignment after release of stress. The distributions we now report together with a recent study (Smith & Fazzalari, 2006) reinforce this view. The change of disc height under compression or extension has been thought to depend on the slip of adjacent lamellae (Szirmai, 1970). As both microfibrils and elastin fibres (Yu et al. 2002, 2005) are densely co-localized within the interlamellar space in the OAF (Figs 5 and 6e) and in the cross-bridges linking adjacent lamellae (Fig. 6e,h), our results strongly suggested that an elastic-fibre network (microfibrils together with elastin fibres) plays an important role in reinforcing the integrity of collagen lamellae and permitting recoil of the collagen lamellae after deformation. An interesting aspect of the present study is that in some locations elastin alone appears to fulfil this role, and in others microfibrils are used either alone or in combination. Given that fibrillin-rich microfibrils are suggested to have an elastic modulus two orders of magnitude higher than that of elastin (Sherratt et al. 2003) this is initially somewhat surprising and may indicate unappreciated subtleties in ECM micromechanics.
Another difference between elastin and microfibrillar glycoproteins occurs in their interactions with other components of the ECM, which may also explain their different distributions. A recent study on the collagen fibre architecture of the annulus fibrosus revealed that mono-aligned collagen fibres within a lamella are highly linked by interconnecting elements (Pezowicz et al. 2005) of unknown composition. Microfibril proteins including fibrillins and latent transforming growth factor β (TGFβ) binding proteins (LTBPs) are multidomain proteins, indicating that they have the ability to react with many different molecules in the ECM (Charbonneau et al. 2004). Indeed, fibrillin-1 has been found to react with versican (Isogai et al. 2002), fibulins (Reinhardt et al. 1996; Freeman et al. 2005), elastin (Trask et al. 2000), microfibril-associated glycoprotein (MAGP)-2 (Hanssen et al. 2004) and LTBP-1 (Isogai et al. 2003). In disc tissue, versican appears to be localized within the interlamellar space (Melrose et al. 2001), suggesting a co-localization of versican with microfibrils. Recent work has found immunolabelled keratin sulfate (composed by versican) located on microfibrils of human IVDs (Akhtar et al. 2005). Combined with our finding of the close association of microfibrils and elastin with collagen fibres (Fig. 6), we suggest that elastic fibres (consisting of elastin coupled with microfibrils) could be the unknown cross-connecting elements of aligned collagen fibres.
We are very grateful for the support of the ARC (16480), and to Mr Nick White (Pathology, Oxford University) for his technical support on confocal microscopy.