Characterization of the annulus fibrosus–vertebral body interface: identification of new structural features

Authors

  • Y. S. Nosikova,

    1. Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, ON, Canada
    2. CIHR-Bioengineering of Skeletal Tissues Team, Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, ON, Canada
    Search for more papers by this author
  • J. P. Santerre,

    1. Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, ON, Canada
    Search for more papers by this author
  • M. Grynpas,

    1. Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, ON, Canada
    Search for more papers by this author
  • G. Gibson,

    1. Department of Orthopaedics, Henry Ford Hospital and Medical Centers, Detroit, MI, USA
    Search for more papers by this author
  • R. A. Kandel

    1. Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, ON, Canada
    2. CIHR-Bioengineering of Skeletal Tissues Team, Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, ON, Canada
    Search for more papers by this author

Dr Rita A. Kandel, Mount Sinai Hospital, 600 University Ave., Toronto, ON, Canada M5G 1X5. T: 416-586-8516 ; F: 416-586-8719; E: rkandel@mtsinai.on.ca

Abstract

Current surgical treatments for degenerative intervertebral disc disease do not restore full normal spinal movement. Tissue engineering a functional disc replacement may be one way to circumvent this limitation, but will require an integration of the different tissues making up the disc for this approach to be successful. Hence, an in-depth characterization of the native tissue interfaces, including annulus insertion into bone is necessary, as knowledge of this interface is limited. The objective of this study was to characterize the annulus fibrosus–vertebral bone (AF–VB) interface in immature (6–9 months old) and mature (18–24 months old) bovine discs, as well as to define these structures for normal adult human (22 and 45 years old) discs. Histological assessment showed that collagen fibers in the inner annulus, which are predominantly type II collagen, all appear to insert into the mineralized endplate zone. In contrast, some of the collagen fibers of the outer annulus, predominantly type I collagen, insert into this endplate, while other fibers curve laterally, at an ∼ 90 ° angle, to the outer aspect of the bone, and merge with the periosteum. This is seen in both human and bovine discs. Where the AF inserts into the calcified zone of the AF–VB interface, it passes through a chondroid region, rich in type II collagen and proteoglycans. Annulus cells (elongated cells that are not surrounded by proteoglycans) are present at this interface. This cartilage zone is evident in both human and bovine discs. Type X collagen and alkaline phosphatase are localized to the interface region. Age-associated differences in bovine spines are observed when examining the interface thickness and the matrix composition of the cartilaginous endplate, as well as the thickness of the mineralized endplate. These findings will assist with the design of the AF–VB interface in the tissue engineered disc.

Introduction

The intervertebral disc (IVD) is located between the vertebral bodies (VBs) of the spine, and is composed of three distinct tissues: annulus fibrosus (AF); nucleus pulposus (NP); and cartilage endplates (CEPs; Roberts et al. 2006). The NP is located in the center of the disc, and is made up of a network of type II collagen fibrils, embedded in a proteoglycan-rich gelatinous matrix. The AF is composed of highly organized collagen-rich lamellae that wrap around the NP. It has been suggested that the tissue composition varies across the disc as there is a decrease in type II collagen and proteoglycans, and an increase in type I collagen, going from the NP to the outer AF. The CEP confines the disc, and is located superiorly and inferiorly at the interface between the VB and IVD. The CEP is rich in type II collagen and proteoglycans (Schollmeier et al. 2000; Moore, 2006). The unique composition and structure of extracellular matrix for the disc allows the spine to resist compressive, bending and torsional loads (Broberg, 1983; Urban & McMullin, 1985; Bayliss et al. 1988). The loss of IVD tissue with age compromises disc function and has been implicated in the development of degenerative disc disease (Adams & Roughley, 2006; Richardson et al. 2007).

Degeneration of the IVD can be associated with low-back pain (LBP; Freemont, 2009; Cheung, 2010; Chen et al. 2011). The healthcare and related costs for LBP exceed $100 billion per year in the USA alone (Katz, 2006). Current non-surgical and surgical treatments for LBP, which are directed towards relieving pain, are not entirely successful, and all except for disc replacement fail to restore disc height and normal spinal movement (Javedan & Dickman, 1999; Okuda et al. 2004; Putzier et al. 2005; Mirza & Deyo, 2007; Hanley et al. 2010). Thus, tissue engineering a functional IVD replacement may be one way to circumvent these limitations (O’Halloran & Pandit, 2007; Kandel et al. 2008). Considerable progress has been made in generating NP tissues (Seguin et al. 2004; Yang & Li, 2009). However, due to the highly oriented lamellar structure of the AF, there have been greater challenges associated with regenerating its complex organization (Mizuno et al. 2004; Wilda & Gough, 2006; Nerurkar et al. 2009; Bowles et al. 2010; Attia et al. 2011). In addition, there is the need to generate a proper force distribution within the engineered tissue, which will require recreating the interfaces between the different tissues. Hamilton et al. have shown that it is possible to regenerate the NP–cartilage interface (Hamilton et al. 2006), but generating the AF–VB interface has yet to be accomplished. Histological studies of in vivo disc have shown that the NP merges with the CEP, which is integrated to bone and can develop calcified deposits after 20 years of age (Coventry et al. 1945a; Oda et al. 1988). After the third decade, part of the mineralized endplate is slowly replaced by bone (Coventry et al. 1945a; Bernick & Cailliet, 1982; Oda et al. 1988; Grignon et al. 2000). However, there is less known about the AF–VB interface. The CEP at the vertebral rim in the outer AF has been called the ring apophysis (RA) in humans (Grignon et al. 2000). It has been shown to calcify as early as 6 years old, ossify at about 13 years old, and merge with the bone during adolescence (Bick & Copel, 1951). It has been suggested that the AF fibers pass through the RA and anchor to the VB via Sharpey’s fibers (Hashizume, 1980; Inoue, 1981; Kazarian, 1981). This description is controversial, as others have shown that the AF anchors to the calcified cartilage, formed during the development of the VB (Francois, 1982). The aim of this study was to more thoroughly characterize the bovine AF–VB interface, which has been proposed by some to be suitable as a model for the human disc under certain conditions (Wilke et al. 1996; Demers et al. 2004; Alini et al. 2008; Beckstein et al. 2008), as well as to compare it with the interface in human discs. The use of bovine discs also provides the opportunity to obtain sufficient numbers of healthy discs to generate robust data, which is not possible for human discs given the limited availability of normal human samples.

Main body

Bovine and human disc harvest

The IVD from the bovine caudal (C3–C5) spines of 6–9 (immature) or 18–24 (mature) months (= 9 at each age per experiment) were harvested, fixed in 10% buffered formalin or acetone and sliced coronally. In total, 72 bovine specimens were used in this study. Four discs were obtained: one from a 22 year old (L4 and 5); and three from a 45 year old (T10–T12 and L1–L2) from human cadavers, in accordance with policy and procedures that comply with Health Canada Regulations and The American Association of Tissue Banking Standards. The tissue use was approved by the Mount Sinai Hospital Institutional Review Board. Both human spines were handled similarly to the bovine samples.

Histological evaluation

Selected samples were fixed in formalin, decalcified in 0.5 m ethylenediaminetetracetic acid [EDTA (pH 7.4); Sigma Chemical, St Louis, MO, USA] at 4 °C and embedded in paraffin. To examine the mineral at the interface, other disc samples were processed undecalcified, dehydrated in graded acetone concentrations, followed by embedding in Spurr’s epoxy media (Electron Microscopy Sciences, Hatfield, PA, USA). Five-micron-thick decalcified or undecalcified sections were cut and stained with either toluidine blue to assess the presence of sulfated proteoglycans, or hematoxylin and eosin (H&E) to visualize cells at the AF–VB interface.

Prior to sectioning the Spurr-embedded tissues, the AF–VB interface region was visualized by backscatter scanning electron microscopy. Undecalcified sections were then cut and stained with von Kossa to visualize mineral distribution. The sections were examined by light and polarized light microscopy.

Alkaline phosphatase activity

Azo dye histochemical staining (Sigma Chemical) was used to demonstrate the presence of alkaline phosphatase activity. A sample of the AF adjacent to the VB was taken, fixed in acetone, and infiltrated sequentially with an increasing sucrose gradient. The samples were rapidly frozen in OCT (Tissue-Tek, VWR) and stored at −80 °C until further analysis. Eight-micron sections were cut at −20 °C, air-dried and stored at −80 °C. The sections were fixed in 10% buffered formalin for 5 min and incubated at 37 °C in azo dye solution [3 mg naphthol AS-MX phosphate, dissolved in 20 μL dimethyl sulfoxide and 10 mg of Fast Blue BB salt, and brought to a final volume of 10 mL with 0.2 m Tris buffer (pH 9.2)] for 3 min at room temperature. This optimal time and temperature were determined experimentally by testing a range of time points (between 30 s and 15 min at room temperature and 37 °C). The filtered solution was used immediately.

Immunohistochemical studies

To examine for the presence of types I, II and X collagen at the AF–VB interface, decalcified sections samples were digested sequentially with 2.5 mg mL−1 trypsin [Tris-buffered saline (TBS)] for 30 min at room temperature, 25 mg mL−1 hyaluronidase [phosphate-buffered saline (PBS)] for 30 min at 37 °C, and 2.5 mg mL−1 pepsin (PBS at pH 2) for 10 min at room temperature (Sigma Aldrich), blocked with 20% goat serum for 30 min at 37 °C (Sigma Aldrich). The sections were incubated with a cocktail of two antibodies, one reactive with type I collagen (1 : 100; polyclonal T59103R; Meridian Life Sciences) and the other reactive with type II collagen (1 : 500; monoclonal, clone 6B3; Thermo Scientific, Labvision, Freemont, CA, USA; TBS with 1% Triton X-100) for 1 h at room temperature. Samples were washed three times in 1% Triton X-100 (PBS), incubated with a cocktail of both Alexa Fluor® 594 conjugated-goat anti-rabbit and Alexa Fluor® 488 conjugated-goat anti-mouse IgG antibodies (both 1/1000 dilution; Invitrogen, Eugene, OR, USA) for 1 h at room temperature in the dark, followed by three washes in TBS with 1% Triton X-100. Slides were coverslipped and examined by fluorescent microscopy. Due to the autofluorescence of human bone, staining for collagens was not done.

For type X collagen staining, the sections were digested with 25 mg mL−1 hyaluronidase for 30 min at 37 °C (Sigma Aldrich), blocked with 20% goat serum for 30 min at 37 °C (Sigma Aldrich) and then incubated with antibody reactive with type X collagen (1/500; Dr Gibson, Department of Orthopaedics, Henry Ford Hospital and Medical Centers, Detroit, MI, USA). Reactivity was detected by incubation with Alexa Fluor® 488 conjugated-goat anti-mouse IgG antibody (1/500 dilution; Invitrogen) for 1 h at room temperature. To visualized nuclei, the tissue was also stained with 4′,6-diamidino-2-phenylindole (DAPI; 4 μg mL−1; Pierce Biotechnology, Rockford, IL, USA) for 5 min at room temperature.

Backscatter and secondary electron mode of scanning electron microscopy

To evaluate the mineralized tissues, the AF–bone interface was visualized by backscatter electron imaging (BSE; solid state BSE detector; FEI, Hallsboro, OR, USA) using scanning electron microscopy (FEI XL30). Prior to imaging, undecalcified Spurr-embedded tissues were cut, polished and carbon-coated. The interface was imaged, photomicrographs taken and stitched in sequence. The thickness of the hypermineralized zone was measured by determining the distance from the soft tissue insertion on one side (appears black) to the bone, which appears gray due to lower mineral content than the adjacent hypermineralized zone (which appears white; Ferguson et al. 2003). The hypermineralized zone, which was continuous with bone, was measured across the IVD with a line tool and ImageJ Software. The secondary images (SE) were sectioned into three areas: outer and inner AF and NP, and were viewed on the corresponding BSE image. One-hundred thickness measurements in total were taken across the disc, with at least 20 measurements for each of the regions. For bovine samples, two or three discs were examined from three to five independent tails. For human spines, one disc from a 22 year old, and three discs from a 45-year-old spine were evaluated. These blocks were then processed for histology and stained with toluidine blue for proteoglycans and von Kossa for mineralized tissue.

Transmission electron microscopy (TEM)

To examine the outer annulus fibrosus (OAF)-calcified region interface ultrastructurally, tissue from this area was sampled from all the discs on samples directly adjacent to those used for the backscatter images, fixed in 2% glutaraldehyde in 0.1 m sodium cacodylate buffer, rinsed in buffer, postfixed in 1% osmium tetroxide, dehydrated in a graded ethanol series followed by propylene oxide, and embedded in Spurr epoxy resin (Electron Microscopy Sciences). Sections (100 nm) were then cut (RMC MT6000 ultramicrotome), stained with uranyl acetate and lead citrate, and viewed using TEM (FEI Tecnai 20). In addition, sections (500 nm) were cut and stained with toluidine blue to assess the presence of sulfated proteoglycans and cells at the AF and VB interface.

Statistical analysis

For the hypermineralized zone thickness measurements, data were pooled and expressed as mean ± standard error of the mean (SEM). Results were analyzed using a one-way analysis of variance (anova), and all pair-wise comparisons between groups were conducted using the Tukey’s post-hoc test. A P-value ≤ 0.05 was considered to be statistically significant.

Results

Morphology of bovine IVD

The bovine IVD is comprised of a centrally placed NP surrounded by the AF, which could be divided into inner and outer zones (Fig. 1A). The lamella of the OAF is composed of distinct aligned layers of collagen. Small amounts of proteoglycans can be seen between the OAF lamellae with toluidine blue staining (Fig. 1B). The inner (I)AF is composed of collagenous lamellae separated by an abundant proteoglycan-rich matrix (Fig. 1B). The transition from the OAF to IAF is distinct (Fig. 1A,B). With age, proteoglycans are less apparent between the OAF lamellae (Fig. 1D). The extracellular matrix within the mature IAF contains less proteoglycans when compared with the immature discs, as demonstrated by the presence of less toluidine staining (Fig. 1D). However, the interface between IAF and OAF is still distinct when stained by toluidine blue.

Figure 1.

 Photomicrographs of coronal sections of a representative bovine immature (A, B) and mature (C, D) IVD [(A and C): hematoxylin and eosin (H&E); (B and D): toluidine blue and von Kossa (TB/VK)]. Arrowheads indicate the proteoglycans in the outer annulus interlamellar space. Arrow indicates the chondroid matrix at the AF–VB interface (scale bar: 250 μm; IAF, inner annulus fibrosus; OAF, outer annulus fibrosus; *AF–VB interface).

Morphology of bovine AF–VB interface

The AF passes through chondroid tissue and inserts into the calcified region of the AF–VB interface (Figs 1B,D and 2A–D). Polarized light microscopy demonstrates that this insertion differs somewhat between the IAF and OAF (Fig. 2B,D). In the IAF, all collagen fibers appear to insert into the mineralized zone (Fig. 2B). However, in the OAF, some of the collagen fibers insert into the mineralized area (Fig. 2A,B) and other fibers in the lamella curve laterally (Fig. 2D) towards the outer aspect of the VB, at an ∼ 90 ° angle (Fig. 2B,D), and merge with the cartilage (Fig. 3A,B).

Figure 2.

 Photomicrographs of coronal sections of representative bovine immature (A–D) and mature (E–H) discs at the inner and outer (A, B, E, F), and outer aspect of the outer (C, D, G, H) annulus insertion into the calcified region [hematoxylin and eosin (H&E) light (LM) (A, C, E, G) and polarized light (PLM) (B, D, F, H)] (scale bar: 125 μm; IAF, inner annulus fibrosus; OAF, outer annulus fibrosus; *AF–VB interface; ↑indicating direction of collagen fibers).

Figure 3.

 Photomicrographs of coronal sections of a representative bovine immature outer annulus lamellae insertion into the outer aspect of VB distant to the disc [H&E light microscopy (LM) (A) and polarized light microscopy (PLM) (B)]. Arrows indicate the outer annulus collagen merging with the cartilage. *Growth plate within the VB (scale bar: 250 μm; OAF, outer annulus fibrosus).

There is a mixed cell population at the OAF–VB interface, as both AF cells and chondrocytes are present (Fig. 4). The annulus cells are elongated, and aligned parallel to the collagen fibers of the AF that pass through the CEP (Fig. 4); these cells are not surrounded by proteoglycans. The chondrocytes, in contrast, are round and are localized in clusters surrounded by proteoglycan-rich matrix (Fig. 4). In the IAF some chondrocytes are randomly distributed in the CEP, some of which abut the mineralized tissue (Fig. 5A). In contrast, most of the chondrocytes in the outer annulus present within the CEP show a more columnar arrangement (Fig. 5B). The chondroid region at the interface is a distinct layer in immature calves and decreases in thickness with age (Fig. 5C,D).

Figure 4.

 Photomicrographs of undecalcified sections (500 nm thick) of the OAF–VB interface of a representative immature (A) and mature (B) cow (toluidine blue). The arrowheads indicate chondrocytes and arrows indicate annulus cells (scale bar: 35 μm).

Figure 5.

 Photomicrographs of coronal undecalcified sections of representative bovine immature (A, B) and mature (C, D) discs at the inner and outer (A, C), and the most outer aspect of the outer (B, D) annulus insertion into the calcified region (toluidine blue and von Kossa). Arrowheads indicate the location of chondrocytes (scale bar: 125 μm; IAF, inner annulus fibrosus; OAF, outer annulus fibrosus; *AF–VB interface). The inserts (scale bar: 200 μm) represent the corresponding interface region as visualized by BSE. The hypermineralized zone appears more white than the bone.

Ultrastructurally, collagen fibers can be seen inserting into amorphous tissue at the OAF–VB interface (Fig. 6). At the insertion site, there is electron-dense material suggestive of a tidemark, which is present in both immature and mature discs (Fig. 6A,B).

Figure 6.

 Electron micrographs of sections of the OAF–VB interface following decalcification of a representative immature (A) and mature (B) cow. The arrows indicate the insertion of collagen fibers into electron-dense material (*tidemark).

Morphology of the mineralizing bovine AF–VB interface

The calcified endplate in calves is thin and porous, as seen in the von Kossa-stained sections and backscatter images (Figs 5 and 7). The endplate is composed of mineralizing cartilage (hypermineralized zone) and bone that form a continuous tissue. The chondroid matrix is undergoing mineralization as bone can be seen in this tissue (Figs 2E and 5C). The hypermineralized zone is seen best by BSE, as it appears brighter than the bone of the VB. The thickness of the hypermineralized zone is significantly more in the mature disc when compared with the immature disc, at the IAF–VB interface (immature: 158 ± 21 μm; mature: 348 ± 35 μm; < 0.05; Fig. 7C). No significant change in thickness with age was detected in the OAF–VB (immature: 152 ± 19 μm; mature: 120 ± 27 μm; Fig. 7C).The AF tissue in areas appears to be in direct continuity with marrow cavity of the bone in immature bovine discs (Fig. 5A). With age the bone in this zone appears denser and less porous across both the IAF and OAF (Fig. 5C,D).

Figure 7.

 Backscatter images of immature (A) and mature (B) bovine disc interface with the vertebral bone. Arrows indicate hypermineralized zone at the interface. (C) Quantification of the thickness of the hypermineralized zone across each disc region. The data are presented as the mean ± SEM (= 6 and 15 discs of immature and mature cows, respectively; scale bar: 500 μm; IAF, inner annulus fibrosus; NP, nucleus pulposus; OAF, outer annulus fibrosus).

Characterization of the bovine AF–VB interface

Given the unique insertion of collagen into the calcified endplate, the distribution of collagen types I and II, the major collagens of the AF, at this interface was evaluated. The chondroid matrix above the bone (corresponding to the AF–VB interface) is rich in type II collagen (Fig. 8A,B). Little type I collagen is detected in this region (Fig. 8A,B). Type I collagen was detected in OAF lamellae and bone. Type II collagen is also present in the osseous endplate, as there is residual cartilage that has not yet fully ossified (Fig. 8A,B). With age, less type II collagen is detected in the chondroid region of the OAF–VB interface (Fig. 8D), and more type I collagen appears to be present at the OAF insertion site (Fig. 8D).

Figure 8.

 Distribution of types I (red) and II (green) collagen in representative images of immature (A, B) and mature (C, D) bovine samples at the inner and outer (A, C), and the most outer aspect of the outer (B, D) annulus insertion into the calcified region, as visualized by immunostaining. Cell nuclei are blue (DAPI; scale bar: 125 μm; IAF, inner annulus fibrosus; OAF, outer annulus fibrosus; *AF–VB interface).

As the collagenous fibers of the AF insert into the calcified region, the interface was examined for the presence of molecules known to be involved in the mineralization process. Immunostaining showed that type X collagen is present at the AF–VB interface only in both immature and mature discs (Fig. 9); it was not seen in the AF, away from the interface. Type X collagen surrounds selected cells, which appear round and chondroid-like (Fig. 9A,B).

Figure 9.

 Distribution of type X collagen (green) in a representative bovine immature (A, B) and mature (C, D) inner and outer (A, C), and the most outer aspect of the annulus insertion into the calcified region (B, D), as visualized by immunostaining. Cell nuclei are blue (DAPI). The inserts show cells at the interface at higher magnification. Arrowheads indicate clusters of cells surrounded by type X collagen at the AF–VB interface (scale bar: 125 μm and insert: 35 μm; *AF–VB interface; IAF, inner annulus fibrosus; OAF, outer annulus fibrosus).

Alkaline phosphatase activity is detected immediately above the calcified interface (Fig. 10). Both the AF cells and chondrocytes appear to express alkaline phosphatase as both elongated and round cells are positive (Fig. 10B). Few cells with alkaline phosphatase activity are present in the AF tissue away from the interface. With age, enzyme activity appears to increase in both the IAF and OAF (all the tissue samples were processed simultaneously under the same conditions; Fig. 10C,D).

Figure 10.

 Alkaline phosphatase activity (blue) is present in immature (A, B) and mature (C, D) discs at the inner (A, C) and outer (B, D) annulus insertion site. Inserts show cells at the interface at higher magnification. *Where tissue was removed from the calcified zone (scale bar: 125 μm; insert: 35 μm; IAF, inner annulus fibrosus; OAF, outer annulus fibrosus).

Human AF–VB interface

Two human native specimens (total of four discs) were analyzed to allow comparison between discs and the mature healthy AF–VB interface in humans. The annulus (22 years old) inserts into the calcified zone through a chondroid region, rich in proteoglycans (Fig. 11A,B,E–G). In the older discs, there appears to be less proteoglycans in the CEP, as indicated by decreased toluidine blue staining in both the inner and outer zones (Fig. 11H–J). Polarized light microscopy demonstrates that the IAF and OAF insert into the mineralized zone similar to bovine discs (Fig. 11B). In the IAF, all of the fibers insert into the calcified region (Fig. 11B). In contrast, within the OAF some of the collagen inserts into the calcified region and other collagen fibers from the lamella curve laterally at ∼ 90 °C to the outer aspect of the bone (note the angle of two arrows in Fig. 11B) and merge with the periosteum of the bone (Fig. 11B).

Figure 11.

 (A–D) Photomicrographs of coronal disc sections of representative 22- (A, B, E–G) and 45- (C, D, H–J) year-old human discs. Arrows indicate the direction of the outer annulus fibers. Arrowheads indicate the outer annulus collagen insertion to the periosteum of the bone. (A–D) Hematoxylin and eosin (H&E); (E–J): toluidine blue/von Kossa; scale bar: 250 μm). (F, G, I, J) Higher magnification images of (E) and (H) of the inner and outer (F, I) and the most outer aspect of the outer (G, J) annulus insertion into calcified zone of a 22- (F, G) and 45- (I, J) years-old human (scale bar: 125 μm). The inserts (scale bar: 200 μm) represent the corresponding interface region visualized by BSE–SEM. PLM, polarized light microscopy. *AF–VB interface. (K–L) Backscatter images of disc interface with the vertebral bone. Arrows indicate the hypermineralized zone (scale bar: 500 μm). (M) Quantification of the thickness of the hypermineralized zone of the human endplate. The data are presented as mean ± SEM (= 3) (IAF, inner annulus fibrosus; NP, nucleus pulposus; OAF, outer annulus fibrosus).

The calcified endplate of the 22-year-old individual is porous and hypermineralized, as seen in backscatter images (Fig. 11K,L). Bone can be seen extending into the IAF (Fig. 11K). The thickness of the hypermineralized zone is less in the older disc (45 years old) when compared with the younger disc (22 years) in the IAF–VB region (younger: 64 ± 19 μm; older: 22 ± 6 μm). No change in thickness with age was detected at the OAF–VB interface (younger: 40 ± 11 μm; older: 42 ± 15 μm; Fig. 11K,L); however, sample numbers were limited.

Discussion

This study characterizes the AF and VB interface in the bovine disc, and identifies some novel features. The IAF collagen fibers extend through the CEP to insert into the calcified endplate. The OAF insertion has a more complex organization, as some of the collagen fibers insert into calcified tissue whereas others curve laterally at an ∼ 90 ° angle towards the outer aspect of the VB and merge with the cartilage. The chondroid tissue through which the collagen fibers pass is hyaline type cartilage, as it is rich in type II collagen and no collagen type I is seen in the immature endplate. The endplate becomes less prominent with age, type I collagen is present, and it undergoes mineralization as demonstrated in light microscopic and backscatter images (hypermineralized region). Interestingly, the hypermineralized zone increases in thickness in the IAF but shows no change with age in the OAF, where its thickness is similar to that of the hypermineralized zone in the younger IAF. There is a mixed cell population at the interface, as both chondrocytes and AF cells are present. Type X collagen and alkaline phosphatase are localized to the mineralizing region of the insertion site. To date, there have been only a limited number of studies that have examined the AF–VB interface. The OAF has been described by others as anchoring directly into the VB, as Sharpey’s fibers (Hashizume, 1980; Inoue, 1981; Kazarian, 1981), whereas Francois et al. suggested that the OAF collagen fibers insert into a calcified layer of cartilage (Francois, 1982). Of note, Coventry described the merging of a portion of the AF with spinal ligaments, whereas in the current study it was clear that the AF was merging with cartilage or bone of the VB and was adjacent to the ligament (Coventry et al. 1945b). The explanation for these discrepancies is not evident.

The presence of cartilage at the interface is in keeping with the findings of Schollmeier et al. who, similar to the current study, showed the presence of the type II collagen (the predominant collagen type of hyaline cartilage) in this location (Schollmeier et al. 2000). Also, in keeping with the current observations, others have shown the presence of mineralization-associated molecules at the interface. For example, both alkaline phosphatase and type X collagen mRNA gene expression are present in the healthy endplate of beagle dogs and mice (Lammi et al. 1998; Liang et al. 2011), and type X collagen protein and mRNA expression have been detected in the scoliotic human endplate (Aigner et al. 1998; He et al. 2004).

The AF–VB interface is distinct from other bone and soft tissue interfaces, such as those occurring in the tendon, ligament and articular cartilage (Benjamin & Ralphs, 1998; Kampen & Tillmann, 1998; Claudepierre & Voisin, 2005; Wang et al. 2006; Bhosale & Richardson, 2008). The ligament/tendon to bone insertion sites can either be fibrous, where collagen fibers insert directly into bone, or fibrocartilagenous multi-tissue, where fibrous tissue merges into non-mineralized fibrocartilage, integrating with mineralized cartilage, and then bone (Benjamin & Ralphs, 1998; Claudepierre & Voisin, 2005; Wang et al. 2006). The AF insertion is very different than either of these given that the collagen fibers of the lamella run through the hyaline cartilage. The other known hyaline cartilage interface is that of articular cartilage with bone (Bhosale & Richardson, 2008). The annulus insertion site is distinguished from the latter in that the articular hyaline cartilage is contiguous with a zone of mineralized cartilage that interdigitates with bone. Collagen fibers insert into the calcified cartilage at the interface with the non-mineralizing articular cartilage, but as an integral part of the hyaline cartilage not as fibers passing through the cartilage matrix, as seen in the AF–VB.

Although it is not known why different interfaces develop where soft tissue integrates with hard tissue (calcified), mechanical forces have been suggested to play a role. The vertebral column provides support and flexibility to the entire body. Thus, the AF–VB interface experiences a wide range of complex forces, including shear, compression, tensile and torsion (Broberg, 1983; Urban & McMullin, 1985; Bayliss et al. 1988). This loading differs from that experienced at the ligament/tendon interface that experiences a more specialized type of force (e.g. either compression or tension; Benjamin & Ralphs, 1998; Claudepierre & Voisin, 2005; Wang et al. 2006; Bhosale & Richardson, 2008). It may be that some of the OAF collagen fibers curve laterally to the outer aspect of the bone and merge with the periosteum because of the multidirectional tensile loads (in particular circumferential) that the AF experiences (McNally & Adams, 1992; Edwards et al. 2001; Gregory & Callaghan, 2011). This type of anchorage may be important to resist the tensile loads experienced by the annulus in a radial direction, while converting the compressive loads exerted by the NP into traction forces, as has been described for the meniscus (Messner & Gao, 1998; Petersen & Tillmann, 1998; Villegas & Donahue, 2010).

Polarized light microscopy showed that the collagen fibers of the AF pass through the chondroid matrix at the interface. Thus, it was surprising that very little to no type I collagen was detected, by immunostaining, in that region. However, this was consistent with a study in mice that also did not detect type I collagen in the CEP (Liang et al. 2011). One possible explanation for this may be that the collagen fibers of the annulus are masked by the presence of excessive amounts of proteoglycan and type II collagen within the chondroid matrix (Vogel et al. 1984). Alternatively, the collagen fibers inserting into the calcified tissue at this point are not type I collagen but another type, such as types II, III, V and VI collagen, which are known to be present in the AF (Ayad et al. 1981; Wu et al. 1987; Roberts et al. 1991). Of note, different types of collagen, such as types I, II, V, IX and XI have also been identified at the insertion sites of tendons and ligaments (Sagarriga Visconti et al. 1996). Further study is required to differentiate between these.

The AF–VB interface undergoes calcification with age, as mineralization of the chondroid tissue was observed in von Kossa–toluidine blue-stained sections. Furthermore, proteins involved in mineralization, such as type X collagen and alkaline phosphatase, were localized to this interface. The mineralizing cartilage corresponds to the hypermineralized zone (higher mineral content than in the adjacent bone) seen by BSE. This zone increased in thickness with age in the NP and IAF regions but not in the OAF region, which was not expected. The reason for this is not clear but may reflect differences in loading and local stress patterns with age across the disc (Carter & Orr, 1992; Meir et al. 2008; Laffosse et al. 2010). The role of hypermineralized zone in the disc is not fully known, but is suspected to be involved in transferring forces at sites where tissues of very different mechanical properties (soft tissue vs. bone) interface, such as to limit shearing with loading (Jager & Fratzl, 2000). In support of this role, a similar zone is seen at the hyaline–cartilage–bone interface, which has been shown to have an elastic modulus with an order of magnitude lower than that of bone (Mente & Lewis, 1994; Jager & Fratzl, 2000; Zizak et al. 2003). However, this is still controversial, as others have shown that this zone has a similar modulus to bone by nano-indentation methods (Ferguson et al. 2003; Gupta et al. 2005).

Although the age differences of the samples examined for the human and bovine discs and the small number of human discs examined limits the comparisons that can be made, there are some apparent similarities and differences identified between the AF–VB interfaces of the two species. Importantly, the characteristic AF insertion was similar in humans and bovine models. The CEP was rich in proteoglycans and appeared less prominent with age in both species. A hypermineralized zone was present in both. In the human disc the hypermineralized zone in the IAF decreased with age in contrast to the bovine where it increased. This was unexpected as others have described endplate calcification with age (Urban et al. 2004; Benneker et al. 2005; Shirazi-Adl et al. 2010; Rutges et al. 2011). However, Wang et al. showed that the bone mineral density at the osseous endplate does not correlate with age or degeneration (Wang et al. 2011). Although there are some differences, there are sufficient similarities that suggest that the cow maybe is a good animal to use to study AF–VB interface development, and may be suitable to use to assess biological repair approaches.

Concluding remarks

In summary, the AF–VB interface is unique in that the annulus inserts into calcified tissue through a chondroid endplate, which contains proteoglycans and type II collagen. Both AF cells and chondrocytes can be seen in this interface. In the OAF the collagen fibers either insert directly into the intervertebral body or curve laterally to merge with the periosteum along the VB. Further analysis of this insertion site is warranted as it will assist in tissue engineering an appropriate AF–VB interface, which would be a requirement for successful integration among the components of a biological disc replacement.

Acknowledgements

This research was supported by CIHR (M0P8672). The authors would also like to thank Harry Bojarski and Ryding-Regency Meat Packers for providing tissues, Mount Sinai Services for processing tissue samples, Alan Wolff for cutting the spine tissues, and Doug Holmyard for the help in sample preparation for SEM and TEM.

Author’s contributions

Nosikova, Santerre, Grynpas and Kandel were involved in drafting the article and revising it critically for important intellectual content. All authors approved the final version to be published. Dr Kandel and Yaroslavna Nosikova had full access to all of the data in the study, and take responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design: Santerre, Kandel, Nosikova. Acquisition of data: Nosikova. Generation of the antibody: Gibson. Analysis and interpretation of data: Nosikova, Santerre, Grynpas, Kandel.

Ancillary