Human intervertebral disc cell morphology and cytoskeletal composition: a preliminary study of regional variations in health and disease

Abstract

Classically, intervertebral disc cells have been described as fibrocytic in the anulus fibrosus and chondrocytic in the nucleus pulposus. Recent animal studies, however, have suggested that disc cell morphology may be more complex than previously considered. Here, by utilizing labelling of components of the cytoskeleton in combination with confocal microscopy, we have examined the detailed morphology of human intervertebral disc cells in pathological and non-pathological tissue. Filamentous-actin- and vimentin-positive cells that appeared either fibrocytic or chondrocytic were observed in all intervertebral discs. However, in localized areas of the disc, stellate cells that extended multiple, branching cytoplasmic processes into their surrounding matrix were also seen. This stellate appearance formed a marked feature of disc cells regionally in certain pathologies, i.e. in cells of the outer anulus fibrosus in scoliotic discs and in inner anulus/nucleus pulposus cells in one spondylolisthetic disc. We conclude that the phenotypic variation of human intervertebral disc cells should be extended to include cells with a stellate appearance, which may be more prevalent in tissue that has been subjected to abnormal load or tension.

Introduction

Mature human intervertebral disc cells have been described as being fibrocytic (or fibroblast-like) in the outer anulus fibrosus or chondrocytic (or chondrocyte-like) in the inner anulus and nucleus pulposus (Walmsley, 1953; Roberts et al. 1991; Chelberg et al. 1995). That is to say, in histological studies using thin tissue sections (mostly 4–6 µm), they have appeared as elongated and bipolar cells (fibrocytes) or as oval/spherical cells (chondrocytes). However, there are indications that disc cell morphology may be more complex than previously considered. A preliminary study of mature human discs described cell processes of up to 80 µm in length extending from stellate cells in the subchondral region of the nucleus pulposus (Johnson et al. 1986). These long processes contained varicosities and appeared to have bulbous, vesicle-filled endings. Shorter cytoplasmic projections have been noted also forming connections between disc cells of the rat anulus fibrosus (Postacchini et al. 1984), whereas in bovine discs cytoplasm-filled processes are a feature of some disc cells throughout the tissue and in the anulus fibrosus, at least, contain filamentous actin (F-actin) and vimentin (Errington et al. 1998; Bruehlmann et al. 2002). In the study by Errington et al. (1998), it was suggested that these cell processes might have a role to play in ‘sensing’ mechanical strain. Mechanotransduction pathways that might function in such ‘mechanosensing’ may involve specific cytoskeletal components. Furthermore, specific cytoskeletal features may reflect different mechanical influences on cells. For example, the organization of vimentin intermediate filaments within chondrocytes, as opposed to actin microfilament structures, has been shown to respond to mechanical stimuli in explant cultures of rat articular cartilage (Durrant et al. 1999), whereas increased vimentin content is observed in vivo in the greater rather than the lesser weight-bearing regions of cartilage (Eggli et al. 1988).

The behaviour of mature intervertebral disc cells is likely to be dependent upon their morphology and hence cytoskeletal composition. Such disc cell features may also reflect changes in their surrounding matrix. However, the extent to which cytoplasmic processes may be a feature of human disc cells throughout the tissue is not clear and their cytoskeletal properties have not been reported. In this study therefore we have examined ‘normal’ and pathological human intervertebral discs for the presence of cytoskeletal components, including F-actin and vimentin, to reveal their detailed cellular morphology. In addition, we have determined the presence of focal adhesion complexes, which are indicative of intercellular connection or cell–matrix adhesion, through vinculin immunolocalization.

Materials and methods

Specimens of 13 human intervertebral discs were obtained from archival material (details in Table 1). The tissues examined included three specimens of no known disc pathology (obtained at post-mortem, ‘normal’), with all other discs being removed by discectomy from patients with disc disorders: four patients with low back pain and evidence of disc degeneration, three patients with spondylolisthesis (slippage of the vertebrae), including one patient with evidence of disc degeneration at another disc level (Table 1, specimens 6 and 10), and three patients with scoliosis (lateral curvature of the spine, where specimens were taken from both the convex and the concave sides of discs at or nearest to the apex of the scoliotic curve). Surgical specimens were received and snap frozen within 1 h of discectomy and post-mortem specimens within 48 h of death. Thirty-micrometre-thick tissue sections were fixed in either 10% neutral buffered formalin or acetone and methanol (1 : 1 ratio) prior to histochemical staining and immunolabelling.

Table 1.

Intervertebral disc specimens

Specimen	Pathology	Age (years)	Sex	Disc Level
1	Cadaveric, ‘normal’	21	F	L5–S1
2	Cadaveric, ‘normal’	73	M	L5–S1
3	Cadaveric, ‘normal’	74	F	L5–S1
4	LBP, DDD	30	M	L4–L5
5	LBP, DDD, prolapse	37	F	L4–L5
6*	LBP, DDD	38	F	L4–L5
7	LBP, DDD	48	F	L5–S1
8	Spondylolisthesis	14	F	L5–S1
9	Spondylolisthesis	25	M	L5–S1
10*	Spondylolisthesis	38	F	L5–S1
11	Scoliosis	13	M	T11–T12
12	Scoliosis	14	F	T8–T9
13	Scoliosis	19	M	L1–L2

LBP; low back pain. DDD; degenerative disc disease.

^*Specimens from same patient.

The actin cytoskeleton was visualized using FITC-labelled phalloidin (Cambridge Bioscience, Cambridge, UK), which has high affinity for polymerized, filamentous actin (F-actin), but does not bind to non-polymerized actin monomers (G-actin) (Wulf et al. 1979; Weiland, 1986). Vimentin was detected using a specific monoclonal antibody (clone VIM 3B4, Novacastra, Newcastle upon Tyne, UK) followed by a secondary step of biotin-labelled antimouse antibody and a tertiary step of FITC-streptavidin (both from Vector Laboratories, Peterborough, UK). Sections were also immunolabelled for vimentin, as outlined above, with immunopositivity revealed using a commercial kit (Vectastain Elite ABC, Vector Laboratories) and 3,3′-diaminobenzidine as chromogen (DAB, Sigma, Poole, UK). Immunolabelling of parallel sections with an irrelevant primary control antibody (Dako, Cambridge, UK) and secondary and tertiary steps (Vector Laboratories) was negative. Sections labelled with fluorochromes to reveal immunopositivity were counterstained with propidium iodide to reveal cell nuclei and mounted in Vectashield (Vector Laboratories), and those labelled with DAB where counterstained with Meyer's haemotoxylin.

In addition, dual labelling procedures to demonstrate F-actin and the focal adhesion complex protein, vinculin, were performed using FITC-labelled phalloidin in combination with a vinculin-specific monoclonal antibody (clone V284, Novacastra) followed by a secondary Texas Red-labelled antimouse antibody (to reveal vinculin immunopositivity, Vector Laboratories) and DAPI (to counterstain for cell nuclei, Vector Laboratories). Serial sections of each disc were also immunolabelled with the monoclonal antibody QBend10 (SkyBio, Boston, UK) to detect endothelial cells in blood vessels, immunopositivity being revealed using a commercial kit (Vectastain Elite ABC, Vector Laboratories) and DAB. All incubations with primary antisera were performed overnight at 4 °C. Six-micrometre-thick serial sections were stained with haemotoxylin and eosin for general morphology.

Histological staining and immunolabelling were viewed using conventional and fluorescent microscopy. The proportions of cells that were positive for F-actin and vimentin were determined by scoring a minimum of 50 cells in at least five randomly selected areas of the outer anulus fibrosus (within 0.5 cm of the border between the anulus and the anterior longitudinal ligament) and the inner anulus/nucleus pulposus (further than 0.5 cm from this border into the disc). The determination of these regions and this method of scoring immunopositive cells within intervertebral discs have been described previously (Johnson et al. 2001). Confocal microscopy (LSM510, Zeiss Ltd, Welwyn Garden City, UK) was used to collect images throughout the full depth of the 30-µm-thick sections and image analysis performed to project the Z stack of those images (Z_proj).

Results

Phalloidin-labelled filamentous actin (F-actin) was absent (in four specimens) or detected very rarely (in two specimens) in cells of the outer anulus fibrosus in ‘normal’ discs and those from patients with discogenic low back pain. By contrast, F-actin labelling was detected more often in outer anulus cells in two of the spondylolisthetic discs and formed a marked feature of the outer anulus in all scoliotic discs (see Table 2). The F-actin-positive cells in the outer anulus of scoliotic discs appeared stellate or dendritic, rather than bipolar and fibroblastic. They extended several cytoplasmic processes, which occasionally branched and contained varicosities, into their surrounding extracellular matrix (Fig. 1). In specimens obtained from both the convex and the concave aspects of the curved, scoliotic spine, the cytoplasmic processes of these F-actin-positive cells appeared locally to form an intricate and extensive network. These processes followed the orientation of collagen fibrils within the anular lamellae, but also passed through the lamellae (Fig. 2a). In general, F-actin-positive cells were seen more frequently in disc specimens from the convex side of the scoliotic curve than the concave side, although statistical significance was not reached (see Table 2; Wilcoxin's signed rank test). Where F-actin was detected with more frequency in the outer anulus cells of spondylolisthetic discs, it appeared similarly in extended cytoplasmic branches. However, these F-actin-labelled cells localized to the interlamellae regions and were bipolar (Fig. 2b).

Table 2.

The distribution of cytoskeletal proteins and cell processes in the human intervertebral disc

		Positive cells (%)
Disc pathology	Location	F-actin	Vimentin	Cell processes
Cadaveric, ‘normal’	OA	0 (0)	11 (5–23)	–/+
	IA/NP	2 (0–5)	63 (48–93)	+
LBP, DDD	OA	1 (0–2)	22 (7–28)	–/+
	IA/NP	48 (21–74)	75 (54–92)	+
Spondylolisthesis	OA	18 (0–28)	27 (5–55)	–/+
	IA/NP	27 (10–45)	67 (52–92)	++
Scoliosis (concave)	OA	29 (2–46)	43 (24–60)	++
	IA/NP	21 (2–44)	78 (60–88)	–/+
Scoliosis (convex)	OA	60 (27–77)	26 (12–49)	++
	IA/NP	27 (11–49)	60 (40–74)	–/+

OA: outer anulus fibrosus. IA/NP: inner anulus/nucleus pulposus. LBP: low back pain. DD: degenerative disc disease. Values shown for F-actin- and vimentin-immunopositive cells are the mean and range (in parantheses) for each pathology and location. The relative distribution of cell processes was assessed semiquantitatively, where –/+ indicates that processes were rarely seen and not in all discs of each group, + indicates the presence of several processes in at least one region of all discs in each group, and ++ indicates a more widespread presence of processes in several regions of all discs of each group.

Fig. 1.

FITC-phalloidin-labelled, F-actin-positive disc cells in a specimen obtained from the outer anulus fibrosus of the concave side of a scoliotic spine. Serial images were collected using confocal microscopy (a–d; shown at 3-µm intervals) and demonstrated the presence of branching, cytoplasmic processes emanating from several disc cells (arrowed), giving these cells a stellate or dendritic appearance. Scale bar = 15 µm.

Fig. 2.

FITC-phalloidin labelling of F-actin in human intervertebral discs, revealed by Z-stacked projections following confocal microscopy. (a) The outer anulus fibrosus of the convex side of a scoliotic spine, where stellate or dendritic cells locally formed an extensive network of cytoplasmic processes within the discal matrix. (b) The outer anulus fibrosus of a spondylolisthetic disc. The anulus cells appeared bipolar with cytoplasmic processes aligned to the orientation of collagen fibrils in the anular lamellae. (c) F-actin-positive chondrocytic disc cells in the inner anulus/nucleus pulposus of a disc removed from a patient with disc degeneration and low back pain. (d) F-actin-positive inner anulus/nucleus pulposus chondrocytic disc cells in a cell cluster in degenerated tissue. (e) F-actin-positive chondrocytes in the growth plate of a disc removed from a young patient with scoliosis. (f) F-actin positivity in blood vessels in the periphery of the intervertebral disc. Strong positivity appeared to be present within endothelial cells and in the vascular smooth muscle. Scale bar = (a) 15 µm; (b)–(f) 20 µm. Z_proj = 30 µm for each image.

F-actin was frequently seen, albeit variably, in cells of the inner anulus and nucleus pulposus in all surgical specimens, but was rarely seen in ‘normal’ tissue (Table 2). The majority of these F-actin-positive cells in the disc's inner regions appeared chondrocytic in shape, whether present as single cells (Fig. 2c) or in cell clusters (Fig. 2d). In localized areas of all pathological tissue, however, inner anulus/nucleus pulposus cells with F-actin-labelled cytoplasmic processes were also occasionally seen. In the scoliotic discs examined, F-actin was seen in chondrocytes within the growth plates of adjacent vertebral bodies and, occasionally, in the cartilaginous endplates (Fig. 2e). There were no apparent differences between the frequency of F-actin-positive growth plate or endplate chondrocytes from either side of the scoliotic curve or in the caudal and cranial regions (data not shown). At the periphery of all types of intervertebral disc, including ‘normal’ tissue, F-actin was consistently detected in blood vessels at the periphery of the disc (Figs 2f and 3). Taking into account all disc specimens examined, there was a negative correlation between the frequency of F-actin-positive disc cells with age of −0.51 in the outer anulus and −0.25 in the inner anulus/nucleus pulposus. However, no significant difference was seen in F-actin positivity in patients/donors who were 30 or more years of age compared with those less than 30 years of age, regardless of pathology.

Fig. 3.

Fig. 3 FITC-phalloidin labelling of F-actin was particularly strong in blood vessels at the periphery of ‘normal’ and pathological intervertebral discs. Images of F-actin labelling in the anterior region of the anulus fibrosus, collected at low magnification (×10 objective, top) and high magnification (×40 objective, bottom) using confocal fluorescence microscopy (Z_proj = 30 µm) either alone (left) or in combination with transmitted polarized light (right), also demonstrate how F-actin-positive disc cell cytoplasmic processes (white arrows) orientated to collagen fibrils within lamellae (lamellar edge indicated by open arrowheads). Yellow arrows indicate blood vessels at the edge of this surgically disrupted tissue. Scale bar = 100 µm (top panels) and 15 µm (bottom panels).

Vimentin immunopositivity was clearly detected in cells of the inner anulus and nucleus pulposus of all disc specimens. In discs from patients with low back pain or scoliosis, and in ‘normal’ discs, the majority of these vimentin-immunopositive cells were chondrocytic in shape, but occasionally appeared to emanate short cytoplasmic processes (Fig. 4a–c). However, in localized areas of the inner anulus/nucleus pulposus of all types of tissue, cells with extensive, vimentin-immunopositive cytoplasmic processes were seen, giving these disc cells a stellate or dendritic appearance (Fig. 4d–f; Table 2), akin to that observed in the F-actin-labelled cells in the outer anulus fibrosus of scoliotic discs. In one of the spondylolisthetic discs (specimen 8, 14-year-old female), these stellate, vimentin-immunopositive cells were the major phenotype in the inner anulus fibrosus and nucleus pulposus, forming more than 70% of the cell population (Fig. 4e). Weakly vimentin-immunopositive cells were detected in the outer anulus fibrosus of all pathological types of disc and in ‘normal’ discs, but in the most peripheral regions of the tissue, i.e. within 0–2 mm of the anterior longitudinal ligament, this vimentin immunopositivity was either very weak or the disc cells appeared immunonegative. Thus, there appeared to be a gradual increase in vimentin content in disc cells from the outer lamellae of the anulus fibrosus towards the nucleus pulposus (Fig. 5). In the outer anulus cells in scoliotic discs in which vimentin was detected, weak immunopositivity localized in cell processes (Fig. 5c), but these processes were not as strongly labelled or clearly observed as those seen by phalloidin-labelling of F-actin in the same disc regions. Vimentin immunopositivity was seen in some chondrocytes in the cartilaginous endplate and in blood vessels, but was generally either weak or absent in these areas.

Fig. 4.

Immunolocalization of vimentin in human intervertebral discs, revealed using DAB as chromogen and by immunofluorescence combined with confocal microscopy. (a) Chondrocytic disc cells in the inner anulus/nucleus pulposus of a ‘normal’ disc, obtained at post-mortem (DAB). (b) Chondrocytic disc cells, some with short cytoplasmic projections, in the inner anulus/nucleus pulposus of a degenerated disc removed from a patient diagnosed with low back pain (DAB). (c) In chondrocytic disc cells (inner anulus/nucleus pulposus of a degenerated disc), vimentin intermediate filaments formed an intracellular network from the cell membrane through the cytoplasm to the nucleus, revealed using immunofluorescence and confocal microscopy. (d) Stellate or dendritic cells in the inner anulus/nucleus pulposus of a spondylolisthetic disc, revealed using vimentin immunofluorescence and confocal microscopy. (e) Vimentin-positive, stellate cells were most frequent in one of the spondylolisthetic discs (specimen 8, revealed using DAB). (f). Vimentin-positive cytoplasmic processes (arrowed) emanating from an inner anulus/nucleus pulposus cell in a degenerated disc. Scale bars = (a,e,f) 15 µm; (b,c,d) 20 µm; (c,d) Z_proj = 30 µm.

Fig. 5.

Vimentin immunopositivity was weaker in disc cells of the outer anulus than in those of the inner anulus/nucleus pulposus. Images of the outer anulus and inner anulus/nucleus pulposus of the same degenerated disc (a,b), and the same scoliotic disc (c,d), are shown. A strongly immunopositive inner anulus/nucleus pulposus cell has been arrowed in (b), and cell processes in the outer anulus cells of the scoliotic disc that were weakly vimentin positive (black arrow) or negative (white arrow) are also indicated. Scale bar = 20 µm.

To establish whether the stability of these cytoskeletal elements could affect the amounts of positivity seen in pathological vs. cadaveric specimens, particularly because F-actin was detected at such low levels in the ‘normal’ discs, we determined positivity for F-actin and vimentin in a single pathological specimen, a part of which was flash frozen within 30 min of discectomy (T0) and another part flash frozen following storage in a sealed container at 4 °C for 96 h (T96) post-discectomy. By T96, the frequency of F-actin-positive disc cells decreased in the outer anulus and in the inner anulus/nucleus pulposus by 55% and 29%, respectively, of the initial T0 values. The frequency of vimentin-immunopositive cells did not decrease in the outer anulus fibrosus over this time course, but decreased by 13% of the T0 value in the inner anulus/nucleus pulposus. A similar trend was observed in the detection of F-actin in stored samples of bovine intervertebral discs (data not shown). Thus, some of the differences seen in F-actin and vimentin positivity may be attributable to differences in tissue storage.

Immunolabelling for the focal adhesion protein, vinculin, demonstrated a similar pattern of distribution to F-actin, in that it was clearly detected in blood vessels (Fig. 6a–c) and in some growth plate chondrocytes (Fig. 6d). However, vinculin immunopositivity was not a marked feature of the majority of disc cells. The detection of blood vessels by F-actin and vinculin labelling was confirmed using CD34 immunolocalization (an endothelial marker; Fig. 6e) and general morphological staining of serial sections (Fig. 6f).

Fig. 6.

Co-localization of F-actin and vinculin in human intervertebral discs. (a–c) Triple labelling of cell nuclei (a), F-actin (b) and vinculin (c), demonstrated the likely presence of focal adhesions in blood vessels. (d) Vinculin immunofluorescence (revealed using FITC) was also detected in growth plate chondrocytes. The presence of blood vessels was confirmed with CD34 immunolabelling of endothelial cells (e) (DAB) and histological staining (f). Scale bar = (d) 20 µm; all others 15 µm. Images (e) and (f) were collected from serial sections of the same scoliotic disc.

Discussion

In this preliminary report, we show that some of the more complex morphological features and cytoskeletal properties recently described in bovine intervertebral discs (Errington et al. 1998; Bruehlmann et al. 2002) are also present in ‘normal’ and pathological human intervertebral discs. In particular, we have observed disc cells with a stellate morphology, i.e. with several cytoplasmic processes extending into the discal matrix, similar to that described by Bruehlmann et al. (2002), where they were identified in interlamellar regions of the anulus fibrosus. Mature human discs, even ‘normal’ ones, have a very much more heterogenous structure (even to the extent of appearing disorganized) than young bovine discs. This ‘normal’ heterogeneity, along with the nature of the pathological tissue examined, which was frequently surgically disrupted, limited the precise identification of where the stellate cells localized in human tissue. Nonetheless, within the anulus fibrosus, cytoplasmic processes emanating from the stellate cells appeared to pass through intra- and interlamellar regions. In addition, stellate cells were observed in the inner, non-lamellar, anulus fibrosus and nucleus pulposus of both pathological and ‘normal’ discs. Therefore, it is apparent that the morphology of human intervertebral disc cells encompasses some cells with a stellate appearance in various areas of the tissue.

Some cytoskeletal properties appeared to be common to cells in all areas/types of intervertebral disc, whereas others were specific or more prevalent in certain regions or pathology. F-actin labelling was not seen in cells in the outer anulus fibrosus of ‘normal’ or degenerate discs (although its absence in ‘normal’ discs needs confirmation, because a time-dependent loss of F-actin positivity was observed in stored tissue), but was observed more frequently in some spondylolisthetic discs and formed a marked feature of outer anulus cells in scoliotic discs. In addition, F-actin was seen in chondrocytic cells in the growth plate of adjacent vertebral bodies in scoliotic discs, and locally in single and clustered chondrocytic cells in the inner regions of ‘normal’ and pathological discs. The functional significance of this regional expression of F-actin is unknown and requires further investigation. For example, F-actin has been suggested to play a role in the secretion of matrix vesicles and hence in matrix mineralization in articular cartilage growth plate (Hale & Wuthier, 1987), but it remains to be seen if this is the case for the disc growth plate or in clustered disc cells. The formation of prominent F-actin stress fibres in outer anulus disc cells has also been suggested to play a role in these cells synthesizing an orientated matrix, and hence the anular lamellae, during rat development; this is in response to altered tensile stress and compressive loading patterns following expansion of the notochord to form the nucleus pulposus (Hayes et al. 1999). If this putative response to tensile stress (and compressive load) also occurs in human disc cells, it remains intriguing that F-actin appeared more prominent on the convex side of the scoliotic curve than on the concave side.

In common with other connective tissues (Eggli et al. 1988; Benjamin et al. 1994), vimentin content in disc cells appeared greatest in areas that are known to be subject to the most compressive load, i.e. towards the discs’ centre. In all discs, except one, the majority of the vimentin-immunopositive cells were chondrocytic in shape. However, stellate, vimentin-immunopositive cells were present locally in all disc types and formed the majority of cells in one of the spondylolisthetic discs. A recent study of chondrocyte morphology in non-degenerate and increasingly degenerate osteoarthritic human cartilage (Bush & Hall, 2003) demonstrated that in all grades of tissue degeneration a surprisingly high proportion of chondrocytes (∼30–40%) had cell processes, but chondrocytes similar in appearance to the stellate cells observed in the inner regions of the intervertebral disc, i.e. with several long processes, were seen in only extensively fibrillated cartilage. As was suggested in this former study, the extension of cytoplasmic processes by these stellate cells may reflect a cellular response to a weakened extracellular matrix.

In conclusion, this novel study demonstrates that stellate cells are a feature of the human intervertebral disc. Further study will determine if their presence or prevalence is ‘normal’ or associated with certain pathology (e.g. spondylolisthesis). Clearly, an increased knowledge of how disc cell morphology and function are related will enhance our understanding of the intervertebral disc in health and disease.