Disc Tissue Samples and Immunohistochemical Preparation
Intervertebral discs, resected from the L4-L5 and L5-S1 spinal segments of fourteen (seven male and seven female) cadavers, were drawn from autopsy samples acquired in a previous study16. The mean period (and standard deviation) between the death of the donor and tissue harvest was 13 ± 5 hours, with a maximum of twenty-two hours. The cadaver donors ranged in age from thirty-two to seventy-eight years, with a mean age of 59 ± 14 years, at the time of death. An unpaired Student t test showed no difference in mean age between the male and female donors. The causes of death of the donors were myocardial infarction (three individuals), sepsis or bacterial infection (five), malignant disease (two), cardiac arrhythmia (one), cerebral vascular accident (one), pulmonary embolism (one), and respiratory failure (one). Three of these individuals had a history of back pain. One reported back pain several months before death, but no pathological changes were noted. Another individual reported back pain within four years before death. None of the individuals had been previously treated for spinal disorders. The original sample collection was approved by the Brigham and Women's Hospital institutional review board.
Macroscopic evaluation was originally performed on removed disc sections, which were assessed with use of the Thompson grading scheme, in which grade-I discs are juvenile and grade-V discs are severely degenerated17. Two investigators independently assigned the Thompson grades to the discs, and discrepancies were resolved through conference16. Discs were originally fixed for a minimum of forty-eight hours in formalin and decalcified in 0.1-M EDTA for up to five weeks. Axial tissue samples from the disc were embedded in paraffin. Sections were cut to a 6-µm thickness with a microtome (Shandon Finesse Model ME+; Thermo Fisher Scientific, Waltham, Massachusetts) and were transferred to positively charged slides and warmed at 60°C for a minimum of two hours to ensure adherence to the slide prior to immunohistochemical staining in an autostainer (Universal Staining System; Dako, Carpinteria, California). Samples of normal caprine articular cartilage were similarly prepared to serve as controls for lubricin staining.
Immunohistochemical Staining for Lubricin
The human disc sections were deparaffinized in xylene and rehydrated in reagent alcohol. The following steps were performed automatically in the autostaining process, separated by rinses with Tris-buffered saline solution containing Tween (TBS, S3006; Dako). Samples were treated with 0.1% protease (type XIV, P5147; Sigma, St. Louis, Missouri) for forty-five minutes to enable antibody penetration as previously described18. Endogenous peroxidase activity was quenched by applying hydrogen peroxide to the samples for ten minutes, and nonspecific binding was blocked with 5% horse serum for thirty minutes. Samples were then incubated with a primary anti-lubricin monoclonal antibody (S6.79, provided by T.M. Schmid at Rush University Medical Center, Chicago, Illinois) at 0.96 mg/L or with a negative mouse immunoglobulin-2b (IgG2b) control (X0944; Dako) at 0.96 mg/L for forty-five minutes; the concentrated primary antibody (4.6 mg/mL) and negative control (100 mg/L) were diluted at ratios of 1:4800 and 1:104, respectively, to achieve the same concentrations of 0.96 mg/L. The S6.79 anti-lubricin monoclonal antibody is an IgG2b immunoglobulin that was developed in the mouse against human lubricin19. The antibody targets the amino terminus of lubricin, and unpublished studies (by T.M. Schmid) suggest that the binding epitope is located within the region of the protein coded by exon 3. Thus, the antibody should recognize the four previously identified human lubricin splice variants, as each includes the region coded by exon 315.
After application of primary antibody or negative mouse immunoglobulin, a labeled streptavidin-biotin horseradish peroxidase system was used with 3-amino-9-ethylcarbazole (AEC) chromogen to visualize lubricin (LSAB2 system; Dako). In this process, a biotinylated secondary antibody was applied for fifteen minutes, followed by a peroxidase-labeled streptavidin treatment for another fifteen minutes. AEC was then applied to the sections for ten minutes to allow detection of lubricin as indicated by the presence of red chromogen at the antigen site in the samples. Samples were counterstained with Mayer hematoxylin and coverslips were applied to the sections with an aqueous mounting medium (Faramount; Dako).
Evaluation of Lubricin in Disc Cells and Tissues
Light microscopy was used to visualize labeled lubricin in the disc sections. When sections contained both nucleus pulposus and anulus fibrosus tissues and a distinction was not readily apparent, polarized light microscopy was used to verify the borders of the tissues by identifying the collagen fiber organization distinct to the anular lamella. The disc anulus fibrosus tissues were divided in half for the purpose of evaluating extracellular matrix and cellular staining for lubricin in the inner as compared with the outer anular regions.
To evaluate the extent of extracellular matrix staining for lubricin, samples were graded semiquantitatively on a scale of 1 to 4, with grade 1 indicating that <10% of the total tissue area stained positively for lubricin, 2 indicating that 10% to 33% stained positively, 3 indicating that 34% to 66% stained positively, and 4 indicating that 67% to 100% stained positively. The intensity of staining was qualitatively observed but not categorically assessed.
To evaluate the percentage of cells expressing lubricin, a cell-counting protocol was employed in which the morphology and intracellular presence of lubricin were assessed for approximately 100 cells per region (i.e., the nucleus, inner anulus, and outer anulus). For consistency, the sample counting areas were taken to be the midpoint of the tissue specimens. A 200× field of view within the selected region (area = 0.9 mm2) was evaluated. If less than the desired number of cells was present in this field of view, the field was moved radially from the initial position and additional cells were evaluated. An average (and standard deviation) of 100 ± 20 cells were evaluated in the nucleus, 103 ± 7 cells were evaluated in the inner anulus, and 120 ± 27 were evaluated in the outer anulus (twenty-eight samples per region). The corresponding average areas in which cell counts were performed were 4.6 ± 3.0, 4.4 ± 2.1, and 1.2 ± 0.7 mm2 (twenty-eight samples per region). Cell morphology was evaluated on the basis of shape, and the presence or absence of an associated lacuna was noted. A cell was considered to be elongated if its nuclear length-to-width aspect ratio was >2; otherwise, it was considered to be rounded. The lubricin-staining status was identified as either negative or positive. Positive staining was defined as the presence of the red detection chromogen at the border of the hematoxylin-stained cell nucleus, distributed within the cytoplasm, or in the immediate lacunar/pericellular space.
Statistical Methods
Multifactor analysis of variance and the paired Student t test were used to evaluate continuous data, whereas chi-square analysis was performed for evaluation of categorical data. To assess the correlation between continuous and categorical data, ordinal logistic regression analysis was used. The nonparametric Spearman rank correlation test was used to obtain the correlation coefficient between the lubricin staining of the extracellular matrix and the percentage of cells staining for lubricin. The standard significance criterion of a = 0.05 was employed for all statistical tests.
Source of Funding
This work was supported in part by the Office of Research and Development, Rehabilitation Research and Development Service, Department of Veterans Affairs. K.M. Shine obtained support from a National Science Foundation Fellowship, and M. Spector derived support from a Veterans Administration Research Career Scientist Award.
Disc Sample Characterization
Of the twenty-eight intervertebral disc specimens, four were Thompson grade II, thirteen were grade III, seven were grade IV, and four were grade V16. For the purposes of facilitating analysis in some portions of this study, samples were grouped into one of two categories: low-grade (Thompson grade II or III) or high-grade (Thompson grade IV or V). Grouping samples in this manner retained the trends observed in the original Thompson scale data.
Lubricin in the Disc Extracellular Matrix and at Surfaces
Samples of caprine articular cartilage, included as control sections, demonstrated lubricin staining as a discrete layer at the articular surface (Fig. 1, a) and in some cases distributed through the extracellular matrix of the superficial zone, as should be expected. A little nonspecific lubricin staining was seen at the edges of freshly cut surfaces of the tissue made during trimming for embedment (arrowheads in Fig. 1, a), indicating that the possibility of edge-artifact staining cannot be excluded when considering the results. None of the representative articular cartilage samples prepared as negative immunohistochemical control sections (treated with mouse IgG2b instead of the lubricin antibody) exhibited any evidence of lubricin staining.
All twenty-eight intervertebral discs treated with the S6.79 monoclonal antibody against lubricin showed at least some positive immunohistochemical staining for the glycoprotein, as indicated by the presence of red chromogen (Table I). Lubricin staining was evident on the surfaces of fissures in the anulus (Fig. 1, b), and in the extracellular matrix of both the nucleus pulposus (Fig. 1, c and d) and the anulus fibrosus (Fig. 1, b) tissue. In the latter, staining for lubricin was observed in the inner (Fig. 1, e) as well as the outer anulus regions. As with the control articular cartilage sections, a little nonspecific lubricin staining could be found on the cut edges created during trimming of the disc samples for histological analysis (arrowheads in Fig. 1, b). No staining was evident in the immunohistochemical negative control samples (inset of Fig. 1, b), which were treated with the mouse IgG2b instead of the lubricin antibody.
In all three disc regions, lubricin staining could be found diffusely distributed within the extracellular matrix (Fig. 1, c, d, and e) although rarely throughout the entire tissue section. Rather, matrix staining appeared to be limited to select areas within each region. In some areas, lubricin staining appeared more intense and globular in contrast to the predominant diffuse distribution. This phenomenon was particularly evident in the matrix near certain borders (i.e., tissue edges and separated interlamellar junctions) or fissures within the disc sections. Lubricin could be seen outside of the pericellular space surrounding chondrocytic cells (Fig. 1, b, c, and e) and in close apposition to fibroblastic cells (Fig. 1, d).
A discrete layer of lubricin, resembling that seen on the articular cartilage control samples, was found on some of the surfaces of separations within the anulus (Fig. 1, f). It is unclear whether these fissures or tissue separations were present in vivo or during resection or were a result of postresection handling and histological processing. Of note is the fact that little staining was observed between nonseparated anular lamellae. Lubricin-staining matrix could be seen surrounding islands of non-lubricin-staining matrix (Fig. 1, g). Lubricin was also seen to coat fragments of matrix (Fig. 1, h), which may have been the result of breakup of matrix, like that seen in Figure 1, g, during trimming.
Evaluation of the extent of lubricin staining across all samples and disc regions revealed no particularly dominant extracellular matrix staining grade; all grades were represented in the total collection, with the number of samples in which each grade was seen ranging from eighteen to twenty-five (Table I). However, sorting the data by disc region (nucleus, inner anulus, and outer anulus) revealed that the extent of lubricin staining was clearly contingent on this parameter (chi-square test, p = 0.0001). In the nucleus pulposus and inner anulus fibrosus, more than half of the samples demonstrated extracellular matrix staining grades of 3 or 4 (i.e., 34% to 100% of the tissue staining positive), whereas only 14% of the outer anulus samples showed that level of staining. Of interest was the fact that the most frequent staining grade was grade 4 (67% to 100% of the tissue staining positive) in the nucleus pulposus and grade 3 (34% to 66% staining positive) in the inner anulus, with grade 1 (<10% staining positive) and grade 2 (10% to 33% staining positive) seen with equal frequency in the outer anulus. Chi-square analysis showed that the extent of lubricin staining of the matrix was not contingent on the Thompson grade (p = 0.5), the sex of the donor (p = 0.5), or the disc level (p = 0.2) from which the specimen had been derived.
Lubricin in the Disc Cells
Cellular staining for lubricin was also noted to vary by disc region. This staining was characterized by accumulation of detection chromogen immediately bordering the cell nucleus, which was made visible by hematoxylin stain, and/or by the presence of chromogen at or within the cell border or pericellular (i.e., lacunar) space when visible. Generally, cellular staining for lubricin appeared discrete and globular in nature in contrast to the more diffuse nature of the staining evident in the extracellular matrix. Staining was noted within the intervertebral disc cell populations in the majority of the disc tissue samples. The number of samples with immunohistochemical evidence of cellular lubricin staining was greatest in the nucleus pulposus (twenty-six of the twenty-eight samples) and decreased notably in the anulus fibrosus. Also, while twenty-two of the twenty-eight inner anulus samples displayed cellular staining, only eleven of the twenty-eight outer anulus samples had visible chromogen. Interestingly, the position of lubricin staining varied among cells in all three regions. In some cells, the staining was in close proximity to the nucleus, while in other cells the chromogen was more obviously in or just outside of the lacunar space (Fig. 2, a, b, and c). Such a pattern qualitatively suggested a mechanism of glycoprotein synthesis by the disc cells followed by subsequent secretion into the surrounding tissue.
Quantitative assessment of the percentages of cells staining positive for lubricin in the samples demonstrating cellular staining revealed striking differences among disc tissue regions. Four-factor analysis of variance investigating the effects of disc region, disc level, sex, and Thompson group on the percentage of cells staining for lubricin showed that only disc region had a significant independent effect (p < 0.0001). Nearly 10% of the cells stained for lubricin in the nucleus pulposus, whereas far fewer cells stained for lubricin in the inner and outer anulus (Fig. 3). The Fisher protected least-significant-difference test revealed that the percentages of cells that were positive for lubricin differed significantly among the regions of the disc (p < 0.022 for all). Of interest was the fact that the disc regions with the most positively stained cells (the nucleus pulposus and the inner anulus fibrosus) were also the areas of the disc in which the most extensive extracellular matrix staining was observed (Table I). Ordinal logistic regression analysis identified a positive association between the percentages of lubricin-staining cells and the grade of extracellular matrix staining (p < 0.0001) (Fig. 4); the Spearman rank correlation test yielded a coefficient (rho), corrected for ties, of 0.65 (p < 0.0001).
Most of the lubricin-staining cells in each region were rounded rather than elongated. In the nucleus and the inner anulus, 96.5% ± 1.7% and 86.0% ± 5.2%, respectively, of the lubricin-staining cells were round, percentages that were significantly different (p < 0.0001 for each) from the percentages of lubricin-staining cells that were elongated in these disc regions. In the outer anulus, there was no significant difference between the percentage of lubricin-staining cells that were round (51.7% ± 13.8%) and the percentage that were elongated (p = 0.9069). Characterizing the percentages of unstained (i.e., lubricin-negative) cells that were round, rather than elongated, in the same samples revealed similar trends (93.4% ± 1.0% in the nucleus pulposus, 83.4 ± 1.2% in the inner anulus, and 44.4% ± 2.0% in the outer anulus). Paired t tests demonstrated no differences between the percentages of lubricin-positive and lubricin-negative cells that were round in each of the three disc regions (p = 0.7, 0.1, and 0.5 for the nucleus pulposus, inner anulus, and outer anulus, respectively). Evaluating lubricin-positive and lubricin-negative cells for the additional presence of lacunae similarly revealed no differences based on this cellular feature. Thus, the cellular presence of lubricin did not appear to be a direct consequence of cell shape or lacunae microenvironment, but rather seemed to be a function of disc region.
To our knowledge, this is the first report about lubricin in the human intervertebral disc. We previously reported the presence and distribution of lubricin in samples from caprine discs8, which were processed without decalcification and were stained with use of the same immunohistochemical procedure that was employed in the present study. The prominence of lubricin staining among lamellae of the outer anulus of the goat suggested that lubricin may play a role in interlamellar tribology. While no lubricin was found in the caprine nucleus pulposus, cells from both the anulus and the nucleus were found to synthesize lubricin in vitro. In the present study of human tissue, a discrete layer of lubricin between lamellae of the anulus was, as we had hypothesized, not as prominent as was found in the goat and there was a greater extent of intralamellar staining for lubricin in the human inner anulus fibrosus. Lubricin was identified both in the anulus fibrosus and nucleus pulposus tissues and in cells of the human intervertebral disc samples, whereas goat disc nucleus pulposus matrix and cells did not stain for lubricin in our previous study8.
The differences in lubricin distribution between the human and goat samples may be a result of structural and/or mechanical differences between the two species. Compared with the spine of the typical quadruped, the human lumbar spine is elongated and markedly lordotic20. Additionally, while the quadrupedal spine is similar to the human spine in that it is primarily loaded by axial compression, the compressive stress in this direction may be greater in the quadruped spine, resulting in increased vertebral bone density21. These differences between species and a myriad of other biochemical factors that have not yet been investigated may account for the interspecies variation noted between the results of our current investigation of human discs and those of our prior studies of caprine models.
While the nature of lubricin staining was similar within the extracellular matrices and among the cell populations of the three human disc regions in this study, the extent of matrix staining was clearly contingent on the region of the disc. Substantially less extracellular matrix stained positively for lubricin in the outer anulus than in the inner disc region. Moreover, the nucleus pulposus more frequently demonstrated a higher lubricin staining grade than did the inner anulus fibrosus. Cellular staining for lubricin also peaked in the nucleus pulposus and decreased from the inner to the outer anulus tissues. These regional differences in lubricin levels are perhaps not surprising. Prior research has established that there are regional differences in material and biochemical properties between the nucleus pulposus and the anulus fibrosus tissues22,23 and that there are differences within the anulus itself based on radial position24-27. Moreover, mechanical differences between the nucleus and anulus and among regions within the anulus tissue have been established28,29, and lubricin expression is known to respond to mechanical stimuli in other musculoskeletal tissues30-35. Positive correlation between the extracellular matrix and cellular staining features of the samples provides strong evidence that the native disc cells in each region are likely responsible for local lubricin production and maintenance in the extracellular matrix.
Although it has long been recognized that the disc cell population is heterogeneous, with some cells displaying a chondrocyte-like phenotype and others appearing to be more elongated and fibroblastic, disc cell morphology does not appear to directly affect the cellular presence of lubricin. While the majority of the lubricin-positive cells in each region were rounded, so too were the majority of the lubricin-negative cells. The presence of cellular lacunae indicative of a more chondrocytic phenotype similarly showed no association with lubricin staining. Thus, disc region appears to be more predictive of cellular lubricin than does general cell morphology. Structural, mechanical, and biochemical differences between the nucleus pulposus and anulus fibrosus tissues and between areas in the anulus itself may be mediating the presence and extent of lubricin in the various disc regions.
It was an interesting finding of this study that the extent of lubricin present in the intervertebral disc extracellular matrix and native disc cells was independent of the Thompson grade. There was significant variation in disc condition among the samples, with eleven samples demonstrating the highest grades of degeneration according to the Thompson scale. However, there was no significant difference in extracellular matrix or cellular lubricin staining between these high-grade samples and those with a lower Thompson grade. Because of the low statistical power, however, we cannot conclude that the Thompson grade has no effect on lubricin staining. Recent research on other musculoskeletal tissues suggests that lubricin content may vary with health and disease and may contribute to joint degeneration and/or disease pathogenesis. For example, lubricin deficiency is associated with the recessive disorder camptodactyly-arthropathy-coxa vara-pericarditis syndrome, which demonstrates marked non-inflammatory joint disease36,37. In an ovine osteoarthritis model, Young et al. demonstrated loss of lubricin-positive cellular staining and reduced lubricin mRNA levels in degenerative articular cartilage with the onset of osteoarthritis38. It will be an interesting future direction of this research to evaluate pathological (e.g., discectomy-derived) disc tissues and determine whether lubricin may be upregulated or downregulated in symptomatic stages of disc disease.
A notable finding in this study was the presence of a discrete lubricin layer at many, but not all, tissue surfaces in the disc. The absence of such staining from most of the freshly cut surfaces produced during trimming for histological processing indicated that the majority of the surface staining was not edge-artifact staining. The fact that the lubricin layer on the torn edge of the disc is similar to the lubricin layer on articular cartilage, and on the meniscus, raises the question of whether the molecules in the disc are responsible for binding lubricin to the surface.
Qualitatively, staining at tissue surfaces appeared to be a more extensive phenomenon in Thompson-grade-V samples than it was in samples with lower Thompson grades, although no quantitative assessment of this issue was carried out and there was only a limited number of Thompson-grade-V specimens. Although it is unclear whether these surfaces were present in vivo or were a result of postresection handling and processing, the intense presence of lubricin at such locations at least suggests that lubricin may accumulate at surfaces in vivo. If such surfaces were the result of tissue tearing or separation, the presence of lubricin at such a location may provide a potential barrier to local tissue repair. Indeed, it has been shown that the presence of lubricin on articular cartilage surfaces can inhibit integrative cartilage repair39. It will be interesting to investigate, in future studies, this surface staining phenomenon in damaged or degenerated disc tissues, in which fissures and lacerations may be more common. 
Note: The authors are grateful to T.M. Schmid, Rush University Medical Center, for providing the lubricin antibody and for helpful discussions. The technical assistance of Alix Weaver is also gratefully acknowledged.