|Home | About | Journals | Submit | Contact Us | Français|
Deterioration of the intervertebral discs is an unfortunate consequence of aging. The intervertebral disc in mammals is composed of three parts: a jelly-like center called the nucleus pulposus, the cartilaginous annulus fibrosus and anterior and posterior endplates that attach the discs to vertebrae. In order to understand the origin of the disc, we have investigated the intervertebral region of chickens. Surprisingly, our comparison of mouse and chicken discs revealed that chicken discs lack nuclei pulposi. In addition, the notochord, which in mice forms nuclei pulposi, was found to persist as a rod-like structure and express Shh throughout chicken embryogenesis. Our fate mapping data indicates that cells originating from the rostral half of each somite are responsible for forming the avian disc while cells in the caudal region of each somite form vertebrae. A histological analysis of mammalian and non-mammalian organisms suggests that nuclei pulposi are only present in mammals.
An unfortunate consequence of aging is the degeneration and failure of many of the body’s tissues, which often leads to pain and the loss of mobility. In humans, the intervertebral discs (IVD) often deteriorate and are believed to cause most cases of lower back pain. In the United States up to 85% of people are affected by back pain at some point in their lives, a prevalence that corresponds to back pain-related health care costing approximately 100 billion dollars annually (Andersson, 1999; Katz, 2006; Smith et al., 2011). Current treatments for the deterioration of the discs include anti-inflammatory pain management and disc arthroplasty (artificial disc replacement)(Mirza and Deyo, 2007). Though implants help to restore mobility, they are subject to failure through wearing (Hanley et al., 2010).
The mammalian IVD functions to resist compressive loads placed on the vertebral column while also providing it with structural support and flexibility. In mammals, the disc is composed of two distinct parts, the nucleus pulposus and annulus fibrosus. The central nucleus pulposus is a hydrogel-like structure made primarily of proteoglycans. The annulus fibrosus surrounds the nucleus pulposus and contains layers of collagen I fiber bundles arranged in alternating directions and a less organized layer of collagen II fibers (Humzah and Soames, 1988; Smith et al., 2011). The breakdown of proteoglycans in the nucleus pulposus and tears in the annulus fibrosus can lead to disc deterioration.
In mice, the embryonic notochord is present from E7.5 until E12.5. Previous work from our laboratory has demonstrated that the embryonic mouse notochord forms all cell types in the nucleus pulposus (Choi et al., 2008; Choi and Harfe, 2011b). The nucleus pulposus persists throughout adult life in both mice and in humans. Given the established role of the notochord as a signaling center during embryonic formation of the vertebral column, it is possible that notochord derived cells in the nucleus pulposus continue to regulate disc morphology into adulthood (Hunter et al., 2003; Smith et al., 2011). These cells may serve as a stem cell population to replace damaged cells in the discs or they may secrete proteins that have the potential to induce cell division in damaged nuclei pulposi.
The molecular pathways responsible for the formation of the intervertebral discs has been postulated to be similar in all mammals, an assumption based on observations that the IVD structures in organisms such as humans, canines, rabbits, and pigs are similar (Walmsley, 1953; Butler, 1989; Alini et al., 2008; Kong et al., 2008). In the mouse system it has been demonstrated that cells from the notochord contribute to the nucleus pulposus and that the notochord disappears during later embryonic development (Walmsley, 1953; Choi et al., 2008).
The cellular origin of the annulus fibrosus is less clear. It has been proposed that the annulus fibrosus is derived from the sclerotome, the ventral portion of each of the somites (Bagnall and Sanders, 1989; Goldstein and Kalcheim, 1992; Huang et al., 1994). At embryonic day 12 in the mouse, sclerotome cells migrate toward the midline and condense around the notochord (Christ and Wilting, 1992). These forming condensations have regions that are variable in cell density. (Rufai et al., 1995). The more compact condensations later give rise to the vertebral bodies, while those that are less dense have been observed to adopt fibroblast morphology. It is this region of the vertebral column upon which the lamellae of the annulus fibrosus form (Christ and Wilting, 1992; Rufai et al., 1995).
While much of the development of the vertebral column has been characterized in mouse models, intervertebral development in chickens has been less thoroughly studied. During primary gastrulation, paraxial mesoderm in the segmental plate epithelializes to form the first 28 pairs of somites (Deries et al., 2010). Then, during the second day of chicken embryonic development somites undergo a craniocaudal polarization by which separate cranial and caudal compartments perform different developmental tasks; specifically, an epithelio-mesenchymal transition mediated by secreted signals from surrounding tissues gives rise to the sclerotome (Deries et al., 2010), a ventral mesenchyme, and the dermamyotome a dorsal layer of epithelial tissue (Christ and Wilting, 1992; Christ et al., 2000; Brent et al., 2003). As the vertebrae form, sclerotome resegmentation occurs along a barrier dividing each half-somite – termed the von Ebner’s fissure – such that the cranial portion of one somite and the caudal portion of the adjacent somite contribute to one vertebral body (Bagnall and Sanders, 1989).
The components of the axial skeleton, including the vertebral bodies, neural arches, spinous processes, proximal ribs, and intervertebral discs, are believed to be derived from distinct cell subpopulations within the sclerotome (Brent and Tabin, 2002; Brent et al., 2003). Currently the literature provides conflicting reports as to which half-sclerotome gives rise to the disc based on studies of the fate of half-somites in chickens. Bagnall and Sanders determined that the caudal half-somite contributes to the disc using peanut lectin staining (Bagnall and Sanders, 1989), as did Huang et al. using quail-chick grafting experiments (Huang et al., 1994). However, Goldstein and Kalcheim reported that the rostral half-somite was responsible for disc formation (Goldstein and Kalcheim, 1992).
Using the lipophilic fluorescent probes DiI and DiA, we fate mapped distinct regions of each half-somite in chickens. In mice, the anterior sclerotome was fate mapped using a Tbx18:Cre allele (Cai et al., 2008). Based on our mouse and chick lineage analysis we propose that cells from the rostral half-sclerotome contribute to the caudal half of the vertebral body and the intervertebral disc, while the caudal half-sclerotome contributes to the rostral half of the adjacent vertebra. In addition, our analysis revealed that the notochord persists throughout embryonic chicken development, resulting in an absence of nuclei pulposi within chicken intervertebral discs. Examination of additional species revealed that nuclei pulposi were only found in mammals.
To determine if the chicken intervertebral disc was similar in structure to the mammalian disc, we examined chicken vertebral columns from Hamburger Hamilton (HH) stage 19 – 44 (Hamburger and Hamilton, 1992) and adult animals using the stains picrosirius red and alcian blue. These reagents stain collagen red and glycosaminoglycans blue, respectively. From HH19 through birth the notochord was observed to persist throughout the ventral midline of chicken embryos (Figure 1). Beginning at HH 35, a “ring” of tissue expressing collagens was observed to form between adjacent vertebrae. This ring of collagen was bisected by the notochord (Figure 1D–F). The notochord was also present in the middle of all vertebrae during embryogenesis (Figure 1).
During mid-embryogenesis in mice the notochord forms all cell types present in nuclei pulposi (Choi et al., 2008). Others and we have proposed that the forming vertebrae “push” notochord cells into the forming discs where they form nuclei pulposi (Aszodi et al., 1998; Choi and Harfe, 2011b). Examination of the vertebral column of adult chickens and quail did not detect the presence of nuclei pulposi (Figure 1H–J). We have denoted the tissue between vertebrae as “IVD” for intervertebral discs. Throughout this report, the tissue located between adjacent vertebrae is referred to as the intervertebral discs, irrespective if the disc contains a nucleus pulposus.
Under polarized light different types of collagen show different colors and intensities of birefringence due to their arrangement within a tissue (Junqueira et al., 1978; Junqueira et al., 1982). An examination of adult chicken intervertebral discs using polarized light revealed that the middle region of the disc contained collagen type III (green color; Figure 1K, K’) and was surrounded by Collagen type I/II (yellow/red color; Figure 1K, K”).
To determine which part of the somite formed the intervertebral disc present in chickens, specific regions of the somites were labeled with lipophilic dyes. Initially, somites in HH16 embryos (2.5 day old embryos) were labeled with a single dye and harvested 13.5 days later (day 16). Embryos injected with DiI in the sclerotome, contained fluorescence in the vertebral bodies, medial portions of the ribs and the intervertebral tissue (Figure 2A–C). Lipophilic dyes have been shown to not diffuse between intact membranes indicting that the cells fluorescing in day 16 embryos were derived from cells originally labeled at day 2.5. The presence of DiI in the intervertebral disc of day 16 embryos indicates that the sclerotome is responsible for forming at least part of this tissue.
To determine which specific part of the somite contributed to the intervertebral disc, two different dyes were used to label different regions of the same somite. DiI was injected into the rostral region and DiA was injected into the caudal region of the same HH20 somite. The placement of the dye resulted in the labeling of distinct regions of the sclerotome within the somite. In these double-labeled somites, DiI labeling of the rostral region of the somite resulted in labeling of the intervertebral disc as well as the caudal region of the adjacent somite (Figure 2D–F and Supplemental Figure 1). DiA labeling of the caudal region of a somite only labeled rostral vertebrae (Figure 2D–F). Fate mapping of the caudal region of the somite did not result in detection of labeled cells between vertebrae. These data indicate that during resegmentation of the somites, the rostral half-sclerotome in each somite forms part of the intervertebral discs.
During mouse embryogenesis the nucleus pulposus forms from the embryonic notochord (Choi et al., 2008). In chickens, our data indicates that nuclei pulposi do not form (Figure 1). To determine whether cells from the notochord contribute to the development of the intervertebral disc in chickens, we labeled notochord cells at HH16 using DiI. Analysis of these labeled cells at day 16 revealed that DiI labeled cells remained in the notochord (Figure 2G–I). No labeled cells were observed in the intervertebral discs or vertebral bodies. These data suggests that cells in the notochord are not responsible for producing any part of the intervertebral discs or any other structures in the chicken vertebral column.
During mouse development, the notochord secretes SHH, which plays an essential role in patterning the overlying neural tube and adjacent somites (Chiang et al., 1996; Litingtung and Chiang, 2000). To determine if the notochord continued to express Shh during late embryogenesis, well after the notochord disappears during mouse development, RNA in situ hybridizations were performed. In chicken embryos, Shh mRNA was expressed in notochord cells until birth (Figure 2J).
The transcription factor T (Brachyury) is expressed in nuclei pulposi of the mouse intervertebral disc. Expression in nuclei pulposi was observed as this structure was forming and persisted throughout early postembryonic development (Choi and Harfe, 2011a). To determine if chicken intervertebral discs, which by morphological analysis appeared to lack nuclei pulposi (Figure 1), expressed T (Brachyury) an immunological analysis was performed. At all stages examined, no staining was observed in the chicken intervertebral disc (Figure 2K, L). However, strong expression was observed in the mouse nucleus pulposus (Figure 2K). These data are consistent with the observation that chicken intervertebral discs lack nuclei pulposi.
Previously, we demonstrated that in mice all cell types located in the nucleus pulposus were derived from the embryonic notochord (Choi and Harfe, 2011a). Chickens do not contain a visible nucleus pulposus. Instead, their intervertebral discs are composed, at least in part, of cells derived from the sclerotome (Figure 2). To determine if the mouse annulus fibrosus, which surrounds the nucleus pulposus, is derived from sclerotome a Tbx18:Cre allele was used. The Tbx18:Cre allele expresses Cre in all cells that express Tbx18 (Cai et al., 2008). In the somites Tbx18 is expressed in the anterior region of the sclerotome and is not expressed in the intervertebral discs (Figure 3A;(Kraus et al., 2001)). To fate map these cells, Tbx18:Cre mice were crossed to the CRE-inducible lacZ allele R26R (Soriano, 1999). LacZ positive cells were found throughout the anterior annulus fibrosus but were excluded from the nucleus pulposus (Figure 3B–D). These data indicated that in mice at least part of the annulus fibrosus is composed of sclerotome cells.
The surprising findings that chicken and quail intervertebral discs did not appear to contain nuclei pulposi led us to investigate when during evolution visible nuclei pulposi was present. Representative organisms from the fishes, amphibians, reptiles, snakes, birds and mammals were examined. Examination of organisms from reptiles and bird species revealed the presence of a thin strip of tissue between each disc that stained with picrosirius red indicating that this tissue is rich in collagen fibers (Figure 4). In the small-spotted catshark S. canicua wide regions of collagen containing tissue were found to reside between adjacent vertebrae (Figure 4). In the Mexican axolotl A. mexicanum and Fire-Bellied Toad B. orientalis vacuolated cells that stained with alcian blue were identified suggesting that the discs of this amphibian were composed of glycosaminoglycans. Both metamorphosed (data not shown) and non-metamorphosed (Figure 4B, E) axolotls had similar intervertebral structures. In this analysis nuclei pulposi were identified only in mammals (C. perspicillata, the Seba’s short-tailed bat; Figure 4M, N). Consistent with these findings, all mammals examined to date in the published literature have been reported to contain intervertebral discs that are composed of nuclei pulposi surrounded by an annulus fibrosus. Our data suggests that nuclei pulposi are only present in mammals (Figure 4O).
The origin of the chicken intervertebral disc has been debated for decades. Using the lipophilic fluorescent probes DiI and DiA to produce a fate map of sclerotome-derived cells, we determined that the rostral half-sclerotome is the part of the somite that is responsible for forming the chicken intervertebral disc. Consistent with the chicken experiments, fate mapping of the mouse anterior sclerotome revealed that the annulus fibrosus was composed of sclerotome cells. In light of these observations and the well-established sclerotome resegmentation that precedes vertebral body formation (Christ and Wilting, 1992; Goldstein and Kalcheim, 1992; Christ et al., 2000; Brent et al., 2003), we propose a model in which cells from the rostral half-sclerotome contributes to the caudal half of the vertebral body and the intervertebral disc, while the caudal half-sclerotome contributes to the rostral half of the adjacent vertebra (this model is shown in Figure 5).
Previously, using the mouse model system we have shown that the embryonic notochord formed the nucleus pulposus of the disc (Choi et al., 2008). In chickens, our fate mapping analysis indicated that the notochord does not contribute to any vertebral structures present during embryogenesis, consistent with our histological data demonstrating that nuclei pulposi are only found in mammals. The fate of notochord cells during chicken postnatal development is unclear and we cannot rule out the possibility that some notochord cells form part of the vertebra, as has been postulated occurs during fish vertebral development (Hyman, 1992).
Our proposed model supports the data presented by Goldstein and Kalcheim (Goldstein and Kalcheim, 1992) based on their chick-quail somite-grafting experiments. However, it contradicts Bagnall and Sanders’ peanut lectin binding experiments (Bagnall and Sanders, 1989) and the quail-chick fate-mapping reported by Huang et al. (Huang et al., 1994) who proposed that the caudal half-sclerotome formed the disc. The dye labeling experiments we performed in this study did not rely on cell transplantation between different organisms and is much more sensitive than peanut lectin staining. In addition, none of the techniques used in the previous studies were capable of marking two discrete cell populations within the same somite. Our data strongly supports the hypothesis that the intervertebral disc in chickens is composed of cells that originate from the rostral half-sclerotome.
In chickens, the notochord persists throughout embryonic development and continues to express Shh. In mice, vertebrae formation is completed around the time that the notochord forms nuclei pulposi. Although both mouse and chick embryogenesis takes 21 days, the formation of organ systems varies greatly. For example, limb development in chickens begins ~2.5 days after fertilization while in mice limb buds do not form until 9.5 days after fertilization (Martin, 1990; Hamburger and Hamilton, 1992). The timing of vertebral column formation also varies greatly between mouse and chick. In mice, the vertebral column is patterned ~13 days after fertilization while chicken vertebrae formation continues almost until birth (Christ and Wilting, 1992; Hamburger and Hamilton, 1992; Tam and Trainor, 1994; Christ et al., 2000; Choi and Harfe, 2011b). Since SHH plays an essential role in the patterning and formation of vertebrae the perdurance of a notochord in chickens may be required to maintain Shh expression during late embryogenesis.
In mammals, the annulus fibrosus forms around and eventually surrounds the nucleus pulposus. However, the cellular origin of the annulus fibrosus was previously unknown. Our mouse fate mapping experiments indicate that at least part of the anterior mouse annulus fibrosus is derived from the sclerotome. Tbx18 is only expressed in anterior sclerotome. It is possible that the posterior annulus fibrosus is derived from posterior sclerotome.
The lack of nuclei pulposi in all non-mammalian organisms examined was a surprise. The synergistic functional relationship between the annulus fibrosus and nucleus pulposus enables both the uniform transmission of compressive loads between the vertebral bodies and complex motion of the intervertebral joint. The nucleus pulposus is proposed to be required for flexibility of the spinal column, to be essential for absorbing shocks, and to resist spinal compression through the distribution of hydraulic pressure (Hunter et al., 2003; Smith et al., 2011). The apparent absence of nuclei pulposi in non-mammalian organisms suggests that the vertebral columns of these organisms either do not require any of these nucleus pulposus-specific functions or that the unique composition of the disc in non-mammalian animals can replace a functional nucleus pulposus in the intervertebral space.
In all non-mammalian organism examined besides the Mexican axolotl and Fire-bellied toad, the intervertebral discs were composed of collagen protein. In this report we have shown that in chickens, the cells in the disc are derived from somites and that the notochord is excluded from the disc. In addition, we demonstrated that the mouse annulus fibrosus was at least in part, derived from the sclerotome. In the Mexican axolotl and Fire-bellied toad, the intervertebral discs were composed primarily of glycosaminoglycans and very little collagen. Unlike the cells from other non-mammals, the discs of amphibians were vacuolated and resembled notochord cells. One interpretation of these data is that amphibian discs are derived from the notochord and not the somites and may resemble a primitive nucleus pulposus. Our fate mapping experiments in mouse and chick and analysis of T (Brachyury) expression indicated that the tissue present in non-mammal discs is equivalent to the mammalian annulus fibrosus. These data suggest that the “primitive” intervertebral disc was composed of an annulus fibrosis-like structure and that formation of nuclei pulposi from the notochord occurred only in mammals.
Chicken (G. gallus) eggs were purchased commercially from Charles River Laboratories and stored at 16 ± 1 °C prior to incubation. Humidified incubators were kept at 38 ± 1 °C. Frozen adult quail and chicken were purchased frozen from a local Publix supermarket. A. laysanensis (E13; equivalent to HH38 in chickens), P. molurus bivittatus (approximately pre-hatchling), T. scripta (E30) and S. canicula (stage 34 pre-hatchling which is ~145–175 days old. The large variance in potential age is due to slow temperature-dependent development of this organism (Ballard et al., 1993)), were a gift from Dr. Martin Cohn. C. perspicillata (>250 day old adult) was a gift from Dr. Chris Cretekos. A. mexicanum (~6 month old animal that had not undergone metamorphosis) was a gift from Dr. Malcolm Maden. The age of the adult fire-bellied toad (B. orientalis) was unknown.
Mice containing the R26R (Soriano 1999) and Tbx18Cre (Cai et al., 2008) alleles were crossed and embryos harvested at E16.5. Embryos were fixed in 0.2% PFA overnight before being LacZ stained as described previously (Harfe et al., 2004). After staining, the vertebral columns were fixed and dehydrated in an ethanol series, cleared with xylenes, washed several times with Blue Ribbon (Leica) and embedded in paraffin. Slides were sectioned at 10µm, coverslipped and photographed.
The dyes DiI and DiA, which exhibit unique fluorescence upon excitation, were used in this study. Stock solutions were prepared by dissolving DiI and DiA crystals separately in dimethylformamide (DMF) at a concentration of 2.5 mg/ml while heating in a 55 °C water bath. Subsequent ultracentrifugation at 5,000 rpm removed un-dissolved dye crystals to decrease the clogging of injection needles. Labeling was detected using fluorescent microscopy: DiI and DiA were observed with RFP and GFP filters, respectively. In order to distinguish DiI and DiA labeling within a single section, fluorescent and brightfield images were captured and overlaid using the GNU Image Manipulation Program (GIMP) software (version 2.6).
Embryos were accessed by making a small opening in the shell directly above the embryo. For each set of experiments, somites in the thoracic region were injected. Needles were pulled from heating capillary tubes to create a bore <3µm in diameter. These were then back-loaded with <2µl of dye and then placed onto a micromanipulator. A small bolus of dye was pulsed through the needle using compressed air into the appropriate cell population in ovo. A period of 5–15ms of pulsation was used to label a small patch of cells. Application was performed under a dissecting microscope with fluorescent filters so that success of injections could be scored. Only embryos whose dye(s) was both targeted to the correct location and contained within a small patch of cells were scored.
For sclerotome single-labeling experiments, embryos were both windowed and injected at HH16 in order to adequately fill a single somite with a bolus of dye. Injections were performed ventrally to label ventral sclerotome. Embryos for notochord labeling were injected within the notochord directly ventrally to the sclerotome at HH16. For double labeling experiments, embryos were both windowed and injected at HH20 so that the cranial and caudal regions of a single sclerotome could be injected. One dye was placed solely within the rostral-most third of the somite, while the other was injected within the caudal-most third of the somite. In order to avoid cross-labeling cell populations, the boluses were places sufficiently far apart so that there was no overlap between them. Refer to Figure 2 for a schematic of the injection sites. All injected embryos were collected at day 16 (+/− 1 day). The vertebral column of each embryo was dissected and fixed in a 4% phosphate-buffered paraformaldehyde solution.
Chicken embryos were harvested at Day 16, fixed, decalcified, dehydrated, and embedded in paraffin. Sections were cut at 7µm. Slides were baked for at least an hour at 65°C before being dewaxed in xylenes and dehydrated in ethanol series. Antigen retrieval was done with 10mM sodium citrate buffer (pH 6.0) for 15 minutes at 70% power in a conventional microwave. After cooling sections were outlined with PAP Pen (Research Products International Corp.), and blocked for 30 min. with 1% donkey serum. T antibody (Santa Cruz, sc-17743) was used at 1:200 and incubated overnight at 4°C. The next day slides were washed 4×10 min with PBST (phosphate buffered saline with Tween), blocked for 30 min with 1% donkey serum, and incubated for 1 hr with Alexa Fluor 568 donkey anti-goat IgG secondary antibody (Invitrogen). After binding, several washes were peformed with PBST. Tissue was counterstaining with DAPI (1:1000 in PBS), washed several times with PBST and mounted with DAKO fluorescent mounting medium.
Tissues were fixed in 4% PFA and prepared for embedding overnight in 15% sucrose solution, followed by an overnight wash in 30% sucrose solution in water. The next day, tissues were removed and placed in Optimum Cutting Material (OCT) at room temperature, positioned and frozen at −80 °C. Either 16 µm, 12 µm or 7 µm frontal sections were lifted onto glass slides and preserved at −80 °C. RNA in situ hybridization was performed as previously described (Wilkinson, 1992). A plasmid containing cDNA for mouse Tbx18 was linearized with EcoRI (New England Biolabs) and an in vitro RNA synthesis was performed using Sp6 polymerase to make antisense Tbx18 mRNA. Prior to staining, slides were dewaxed in xylenes and rehydrated through an ethanol series. Next slides were bathed for 15 minutes in Alcian blue solution (3% glacial acetic acid, 1% Alcian blue GX powder (Sigma) in de-ionized water), followed by a 10-minute tap water rinse. Slides were next bathed in Picro Sirius Red Solution (Poly Scientific) for 45 minutes then washed in 0.5% acidified water (glacial acetic acid in DI water). Finally, slides were dehydrated in ethanol and put into xylenes before being mounted (Permount, Fisher Scientific) and coverslipped. Solutions were filtered before use.
Supplemental Figure 1. The rostral half-sclerotome forms part of the intervertebral disc. This panel is identical to the panels shown in Figure 2 D–F with the exception that panel C in this figure shows that DiO labeled cells (green) formed part of the vertebra. DiI labeling (red) of the ventral-rostral somite marked cells that were observed in the intervertebral disc. . S = somite, D = dorsal, V = ventral, R = rostral, C = caudal, N = notochord, VB = vertebrae, N = notochord.
We thank Marty Cohn, Malcolm Maden and Chris Cretekos for providing tissue samples used in this study and Lachlan Smith for reading the manuscript. B.J.B and Y.L. were supported by Science for Life grants from HHMI. B.J.B. was also supported by a grant from the University of Florida Medical Student Research Program. J.A.M was supported by a University of Florida Alumni Fellowship. This work was supported by grants from the National Institutes of Health/National Institute on Aging (AG029353) and National Institutes of Health/National Institute of Arthritis and Musculoskeletal and Skin Diseases (AR055568) to B.D.H.