Eight shoulders were resected from eight Spanish goats, ranging from two to five years of age and weighing between 30 and 50 kg (mean [and standard deviation], 41 ± 18 kg). Five stifle joints (corresponding to the knee joint in humans) were also resected from five of these goats to yield tissues for comparative analysis to determine the distribution of lubricin in goat articular cartilage and synovial membrane for comparison with the staining of lubricin in these tissues in other species. The patellar tendon was selected for comparison of the lubricin staining in the infraspinatus tendon. After the animals were killed with use of an intravenous overdose of pentobarbital, the intact complexes of the infraspinatus tendon and its bone insertion were carefully exposed and resected from the muscle junction to the greater tuberosity. These tissue blocks contained the head of the humerus and the adjacent synovium, as well as the infraspinatus tendon-bone complex.
All tissue specimens were fixed in 10% neutral buffered formalin for at least six hours. The tissue blocks were then decalcified in 0.1-M ethylenediaminetetraacetic acid (EDTA) at pH 7.4 for up to six weeks, depending on the degree of calcification. All of the specimens were dehydrated in graded alcohol solutions and xylene and were embedded in paraffin. The paraffin-embedded tissue was sectioned along the long axis of the infraspinatus tendon in the frontal plane (Fig. 1). Six-micrometer-thick microtomed sections were deparaffinized and stained with hematoxylin and eosin for cell morphology and identification. Masson trichrome19 stain was used to reveal collagen. When collagen is in a relaxed state at the time of formalin fixation, Masson trichrome stains the collagen blue; collagen that is under tension at the time of fixation is stained red20. Histological sections were examined with normal transmitted light and under polarized light to reveal collagen organization, orientation, and crimp.
Immunolocalization of lubricin in the tissue samples was performed. After deparaffinization with xylene, microtomed sections were rehydrated in ethanol. A final wash with Tris-buffered saline solution (S3001; DakoCytomation, Carpinteria, California) was done to reduce enzymatic activity. Sections were then treated with 0.1% protease XIV (P5174; Sigma, St. Louis, Missouri) for forty-five minutes, to facilitate antibody penetration of the tissue; this procedure was used in one of our prior studies21. The sections were then treated with peroxidase-blocking reagent (S2001; DakoCytomation) for ten minutes and 5% goat serum (Sigma) for thirty minutes prior to incubation with the primary antibody. A purified monoclonal antibody to superficial zone protein (S6.79; Rush University Medical Center, Chicago, Illinois) was used at 1:1000 dilution (1 µg/mL protein concentration) for thirty minutes. The antisuperficial zone protein monoclonal antibody was made in the mouse against human superficial zone protein22. It has been found to react with a variety of mammalian proteoglycan-4-lubricin molecules, including human21,22, dog17,22, bovine21,22, guinea pig22, and rabbit22. The S6.79 antibody recognizes the amino terminus of superficial zone protein. For this paper, the protein staining positive with the antibody is referred to as lubricin. Studies (unpublished observations of one of us [T.S.]) have suggested that the epitope for the antibody is in the region of the protein coded by exon 3. Therefore, the antibody should recognize all of the four Prg4 splice variants8,11 that have been found in human tissue because they all include the region coded by exon 318.
Serial sections allocated as negative immunohistochemical controls were treated with nonspecific mouse myeloma immunoglobulin (IgG2a [catalog number 02-6200]; Zymed Laboratories, South San Francisco, California), instead of the lubricin antibody, at the same concentration. Visualization was achieved with use of biotinylated link as a secondary reagent, streptavidin-horseradish peroxidase (K0675; DakoCytomation) as a tertiary reagent, and substrate (AEC substrate chromogen; K3464; DakoCytomation). The immunohistochemical staining was performed with use of the Dako Autostainer (DakoCytomation) with the program for lubricin. At the end of the immunohistochemical procedures, the slides were counterstained with hematoxylin and coverslips were applied with Faramount mounting medium (S3025; DakoCytomation).
Digital images were acquired with use of a MicroFire camera (model S99809; Meyer Instruments, Houston, Texas) mounted on an Olympus BX51 microscope (Olympus, Tokyo, Japan). The distance to which lubricin staining was detected in the tendon, relative to the humeral insertion site, was made with use of ImageJ software (National Institutes of Health, Bethesda, Maryland). Estimates of the number, and the mean diameter, of fascicles making up the infraspinatus tendons of the eight goats were determined from light microscopy. The identification of structures as fascicles was made on the basis of the presence of continuous discrete layers bordering the fascicle, as defined by hematoxylin and lubricin staining. It was not possible to identify some features in the histological sections as fascicles because of sectioning artifacts (e.g., tears in the section). Therefore, all fascicles in a tendon may not have been counted and measured.
Images from polarized light microscopy of the sections stained with hematoxylin and eosin and the lubricin antibody were used to estimate the crimp length in selected fascicles. The "peak-to-peak" distance of the crimp waveform was measured for fascicles that were cut in longitudinal section. The crimp length for a fascicle was recorded as the average value for the measurements of ten consecutive crimps along a random location in the fascicle. One-factor analysis of variance was used to determine the significance of the effect of the animal on the fascicle diameter. Two-factor analysis of variance was used to determine the effects of the animal and the location (humeral compared with joint side) on the fascicle crimp length.
There was a consistent pattern of lubricin staining in the stifle joint articular cartilage (Table I, Fig. 2). Lubricin was found as a thin, discrete layer (=5 µm thick) at the surface of the cartilage and diffusely distributed in the extracellular matrix to a depth of as much as 250 µm. The discrete lubricin layer on the surface of the cartilage appeared to contain cells at some locations. No lubricin staining was seen in the lower-middle and deep zones of the cartilage in any of the samples.
There was some indication of intracellular lubricin, particularly in cells near the surface (Fig. 2). In addition, there was only a faint chromogen labeling of the pericellular (lacunar) space of the chondrocytes in the deeper region of the superficial lubricin-positive zone (Fig. 2), indicating the presence of small amounts of lubricin.
Histological sections through the goat shoulder joint (Fig. 3, a) demonstrated the microanatomy of the infraspinatus tendon insertion into the tuberosity of the humerus, the opposing humeral head, and adjacent synovium. The serial section stained for lubricin showed the specificity of the glycoprotein for selected sites (Fig. 3, b), described below.
Lubricin was found in the cartilage on the humeral head in all of the samples (Table I, Figs. 3, b and 4, a). A thin, discrete layer of lubricin staining was found on the surface of the humeral fibrocartilage (Fig. 4, a), similar in thickness and intensity of staining to the discrete lubricin layer on the articular cartilage of the stifle joints of the animals. The diffuse staining of the cartilage matrix in the humeral cartilage, which extended to a mean depth (and standard deviation) of approximately 300 ± 60 µm (Table I), was generally less prominent and less uniform than the lubricin staining found in the stifle joint articular cartilage. Punctate chromogen staining could be seen in the cytoplasm of some of the chondrocytic cells in the region within approximately 200 µm of the surface (Fig. 4, a). There also appeared to be a pericellular concentration of lubricin around many of the chondrocytes (Fig. 4, a). The number density of the immunopositive cells in the humeral region was qualitatively markedly greater than that in the articular cartilage of the stifle joint.
The synovium in the tissue blocks from all eight goats stained intensely positive for lubricin (Table I, Fig. 4, b). The chromogen could also generally be found diffusely distributed in the subsynovial tissue. The intensity of staining precluded definitive demonstration of intracellular lubricin deposits.
The rotator cuff tendon-to-bone complex in the goat showed the typical structure, consisting of tendon, noncalcified fibrocartilage, calcified fibrocartilage, and bone (Fig. 3, a). Each infraspinatus tendon sample had the characteristic tendon architecture, consisting of rows of cells (Fig. 5, a) separated by extracellular matrix. Cells appeared in both rounded and flattened morphology, with the former often contained within lacunae. The portion of the fascicle bordering the humeral joint space (Fig. 5, b) displayed cells of chondrocytic morphology in lacunae distributed in a fibrous extracellular matrix, consistent with fibrocartilage; this region stained less uniformly with eosin than the matrix in the body of the tendon and had a mottled appearance.
Separations between units of the tendon (Fig. 5, a) were frequent findings, seen at several sites in every sample. It was not possible to determine whether these separations were present in vivo, produced during the resection of the tissue prior to fixation in formalin, or created during microtomy. It was of interest that these separations did not coincide with the rows of cells, which might have been considered to be planes of weakness through which such separations, regardless of the cause, might occur. Additional studies are required to investigate the implications of these separations.
Masson trichrome stained the collagen within the fascicles of the tendon red (Figs. 5, c through f), indicating that it may have been under tension at the time of fixation. The layers separating the subunits of the tendons stained both red (Fig. 5, d) and blue (Fig. 5, e). The blue staining of the collagen in the fibrocartilaginous portion of the fascicle that bordered the humeral joint space (Fig. 5, f) contrasted with the red staining of the underlying tendon, with a distinct separation of the regions (Fig. 5, f).
All of the rotator cuff samples displayed a similar distribution of lubricin (Fig. 3, b). Immunostaining for lubricin in the infraspinatus tendon was clearly present at specific locations to a distance of approximately 7.5 mm from the tendon-bone insertion in longitudinal sections (Table I; Figs. 3, b and 6, a). Moreover, there appeared to be more staining in the tissue in the region bordering the humeral joint side of the rotator cuff (location 5 in Fig. 3, b), compared with the superior aspect of the tendon (location 6 in Fig. 3, b).
Lubricin was reproducibly identified in all of the samples in the following specific locations of the infraspinatus tendon (Table I, Figs. 3, b and 6, a, b, and c): in distinct layers that appeared to separate fascicles (Figs. 6, a and b), diffusely distributed in the fascicle bordering the humeral joint (location 5 in Fig. 3, b and in Fig. 6, b), and in intrafascicular planes (Figs. 6, b and c). There was no labeling in the negative control sections (Fig. 6, d), and no lubricin was found in the noncalcified fibrocartilage, which appeared as a layer approximately 500 µm thick interposed between the tendon and the calcified fibrocartilage; in the calcified fibrocartilage layer; or in bone (Table I, Fig. 3, b).
Lubricin was found in a discrete layer (up to a few micrometers in thickness) on the surface of the fascicle bordering the humeral joint space and diffusely distributed in the extracellular matrix of the fibrocartilage to a depth of 21 to 133 µm (Table I, Figs. 6, b, e, and f) from the joint surface of the tendon. Many of the rounded cells in the region near the humeral joint space displayed the presence of lubricin intracellularly and in the lacunae (Fig. 6, e).
Within the tendon, lubricin was prominently localized to discrete layers of <10 µm in thickness separating the fascicles (Figs. 3, b and 6, a, b, and c). In some of the immunohistochemical sections, the orientation of the tendon in the paraffin block was such that the fascicles were cut in oblique section. A lubricin-positive layer was seen completely surrounding the fascicles as a sheath.
The lubricin stain appeared as a uniform distribution of the chromogen and as dappled or punctate deposits of the chromogen (Fig. 6, c). There were more lubricin-positive layers found in the humeral joint region of the tendon than in the superior aspect of the tendon (Fig. 3, b).
The identification of these distinct lubricin-rich layers as the sheaths around fascicles was suggested by three observations. First, serial sections stained with Masson trichrome (Figs. 5, c through f) showed the lubricin-rich layer (Figs. 6, b and g) to be a distinct morphological feature, different in appearance from the collagen fibrils that made up the body of the fascicle1,23. Second, lubricin was found on the surface of separations within the tendon, which may have existed in vivo or have been produced during the excision of the samples and during histological processing. Third, polarized light microscopy revealed that the crimp pattern within a subunit terminated at the lubricin-positive layer (Figs. 6, h and i) (described in a following section). No blood vessels or nerves were seen in this lubricin-positive sheath around the fascicles and rarely were cells found to be fully contained within the lubricin layer.
The prominence of the lubricin staining of the layers separating the fascicles was often highlighted by separation of the fascicles (Fig. 6, g), likely resulting from the handling of the tissue after resection and during histological processing. As the fascicles separated, the lubricin-positive layer was teased apart (Fig. 6, g). This separation facilitated the examination of the lubricin-positive layer for the presence of cells and vessels within the layer, and none were found (Fig. 6, g).
In some cases, cells within the fascicles, near the lubricin layer, could be seen to be surrounded with lubricin-staining matrix, suggesting that they may have synthesized the glycoprotein (Fig. 6, c). About 50% of such cells displayed oval-shaped nuclei and appeared to be contained within lacunae (Fig. 6, c), which suggested a chondrocyte phenotype. Many of the cells found within the lubricin-positive layer separating fascicles also had these chondrocytic features (Fig. 6, c). Some of the cells showed immunopositive dots or rings around their nuclei (Fig. 6, c).
Longitudinal sections through the tendon displayed rows of contiguous cells24, separated by about 100 to 150 µm (Fig. 6, b). These rows did not stain positive for lubricin and generally were not near the lubricin-positive layers (Fig. 6, b).
Lubricin-positive layers were also seen within fascicles (Fig. 6, c). These intrafascicular layers of lubricin were generally thinner than the layers separating fascicles, and they did not extend along as much of the length of the fascicle. Moreover, the intrafascicular layers often displayed a crimp in register with the crimp of the fascicle (Fig. 6, c), supporting their identification as being intrafascicular layers.
Polarized light microscopy revealed the crimp pattern of the tendon (Figs. 6, h and i). Nonbirefringent seams extending longitudinally in the polarized light microscopy images served to define the boundaries of the fascicular subunits in regions of the tissue where clear separations were not present between fascicles; these nonbirefringent seams generally stained positive for lubricin (Figs. 6, h and i). The crimp pattern did not remain in register across these nonbirefringent seams. Qualitatively, the crimp of the fascicles in the tendon toward the humeral joint space (Fig. 6, h) displayed a shorter peak-to-peak distance than that seen in the fascicles on the superior side of the tendon (Fig. 6, i).
The longitudinal section through the tendon near the junction with bone consisted of ten to fifteen fascicles. Eight to nine fascicles in each tendon could be seen clearly enough to measure. The diameter of the fascicles ranged from 100 to 600 µm, with a mean value (and standard deviation) of 245 ± 108 for sixty fascicles from all seven goats. The mean values of the fascicle diameter for each of the seven goats individually ranged from 167 to 347 µm. The coefficients of variation for the measurements for each goat ranged from 26% to 40%. One-factor analysis of variance revealed a significant effect of the animal on the fascicle diameter (p = 0.004, power = 0.94). For the individual goats, there was no consistent relationship between the fascicle diameter and the location within the tendon (i.e., the humeral joint side compared with the superior aspect of the tendon).
The crimp length ranged from 9 to 100 µm. The mean values of crimp length for the seven goats ranged from 18 to 67 µm, with high coefficients of variation (17% to 100%). The crimp lengths varied among animals, but for every animal the crimp lengths in the fascicles on the superior side of the tendon were longer (Fig. 6, i) compared with those on the humeral joint side (Fig. 6, h) (two-factor analysis of variance, p < 0.0001 and power = 1 for effect of animal and p < 0.0001 and power = 1 for effect of location). The mean value of the crimp (and standard error of the mean) was 30 ± 4 µm for the humeral side fascicles compared with 52 ± 5 µm for the fascicles in the superior half of the tendon.
Analyses performed on patellar tendons from five goats demonstrated immunopositive staining for lubricin as a discrete layer, approximately 10 µm thick, on the articular surface of only two samples (Fig. 7). In these two samples, the lubricin-positive layer was on the region of the ligament near patellar bone. Although lubricin was stained on the surface of two patellar tendons, no lubricin staining was found within the ligaments (Fig. 7). No chromogen was found in the negative control slides.
The notable finding of this investigation was the prominence of lubricin in discrete layers separating the fascicles in the rotator cuff. The results support the hypothesis that lubricin serves as an interfascicular lubricant in the infraspinatus tendon. The current study employed infraspinatus tendon from the goat, in anticipation of future studies directed toward the investigation of repair procedures for rotator cuff tears in a large animal model. Prior studies have demonstrated the suitability of the goat for studies of the healing of rotator cuff injuries25, despite anatomical differences compared with humans, including the presence of two heads of the infraspinatus tendon.
While the monoclonal lubricin antibody used in this study has demonstrated cross-reactivity in many species, it had not yet been used for immunohistochemistry on goat tissue. Samples of articular cartilage from the goat stifle joint and humeral head, and humeral joint synovium, displayed a distribution of lubricin consistent with patterns found in prior studies of human tissues21,22,26. The samples of goat synovium stained intensely for lubricin, and a discrete lubricin-positive layer was observed on the surface of the patellar tendon, as has been found in canine tendon17.
Lubricin may have other functions related to its antiadhesion properties9, and this theory is supported by the observation that fascicles were often found separated from one another. Prior investigations have noted that there are no direct attachments or cellular communications between neighboring fascicles1, which may be essential to allow the separation of fascicles for interfascicular sliding27. It will be interesting in future work to determine whether the absence of lubricin in the rotator cuff is associated with degenerative changes.
Separations of the fascicles of the tendon, which may have been caused by handling after resection and histological methods, occurred along the planes containing the lubricin. This finding further supported the identity of this layer as endotenon in that it appeared to allow intrafascicular movement. The endotenon did not contain cells or blood or lymphatic vessels (Fig. 6, g), as has been described in the endotenon of other tendons28.
It is of interest that Masson trichrome stained the lubricin-positive layers separating fascicles red at some locations and blue at other sites, indicating that the collagen was present in tensioned (red) and relaxed (blue) states20, respectively. Future work is required to further investigate this finding, as it relates to the binding sites for lubricin in selected planes through the tendon and the large strains (as well as sliding) that can occur in the layers separating fascicles.
The finding of lubricin on the surface of the infraspinatus tendon (in the epitenon) in the present study is similar to the findings in prior work, with use of the same antibody, which revealed lubricin on the surface of the Achilles17, flexor digitorum profundus16-18, and patellar17 tendons, and on the surface of the anterior cruciate and lateral collateral ligaments17. It has been suggested18 that, in addition to facilitating lubrication of the structures during sliding motion17, lubricin may prevent cellular adhesion to the tendon surface16,18,29. The distribution of lubricin within the body of the infraspinatus tendon was similar to the pattern of immunolocalization of lubricin in the canine flexor digitorum profundus in a prior investigation18, with use of the same antibody, in that lubricin was found surrounding collagen bundles and in fibrocartilage in regions of the tendons exposed to compressive loading. The location of lubricin within fascicles suggested that it may be facilitating the relative movement of collagen bundles. Previously, it was suggested that proteoglycans might facilitate sliding of collagen fibrils30.
Also of note was the distribution of lubricin throughout the fibrocartilaginous portion of the fascicle bordering the humeral joint space. This fibrocartilage stained blue with Masson trichrome, which may indicate a difference in the state of strain of the collagen (relaxed at the time of formalin fixation) compared with the collagen (tensioned) in the remaining part of the fascicle and in the other fascicles of the tendon. The observation that the fibrocartilage of the tendon contained lubricin is consistent with the finding of lubricin in the meniscus31.
It was not possible to determine definitively which cells were responsible for the lubricin synthesis in the endotenon layer because cells in the endotenon were not clearly visible in tissue in which the fascicles had not separated, and they may have been dislodged in samples in which the fascicles were teased apart. It was of interest that cells near the surface of the fascicles were found in this immunohistochemical evaluation to contain immunopositive spots or rings around their nuclei, which may correspond to the intracellular presence of the lubricating protein soon after its synthesis. Moreover, cells containing lubricin could also occasionally be found in lubricin-positive layers within fascicles. Thus, it is likely that tendon cells were responsible for the synthesis of the lubricin, which then diffused to selected planes within the tissue. However, if separations of fascicles existed in vivo, it is also possible that lubricin in the humeral joint fluid, synthesized by joint surface chondrocytes and synovial cells, was able to access the surface of fascicles in the body of the tendon.
The identity of the lubricin-positive layer as endotenon could be confirmed by polarized light microscopy that demonstrated the contiguous crimp of the collagen within the structures, delineated by lubricin, as fascicles. The number and diameter of fascicles varied among the animals, with a diameter range of 100 to 600 µm compared with 80 to 320 µm for rat tail tendon1. There was no clear association between fascicle diameter and location in the infraspinatus tendon. There was, however, a notable difference in the crimp pattern when the fascicles in the humeral joint side of the tendon were compared with those in the superior region. The crimp measurements (which ranged from a mean of 20 to 70 µm for the seven goats) were less than those generally found in the rat tail tendon1. There is a question of whether the amount of lubricin in the fascicular sheath is associated with the degree to which crimp periods differ in adjacent fascicles, and therefore the degree to which interfascicular sliding could be expected.
In this study of the infraspinatus tendon and in prior investigation of the canine flexor digitorum profundus18, the amount of lubricin detected within the tissues varied with the anatomic location along the length of the tendon, as would be expected on the basis of the variation of the mechanical environment. A number of studies have shown that mechanical stimuli regulate lubricin expression14,16,18,32.
One limitation of the study concerns the question of how the findings in the goat relate to the human. Differences in the tribology and distribution of lubricin in goat and human shoulders might be expected relative to the differences in the weight-bearing function and anatomy. A related limitation concerns the age range of the animals. Future work is required to systematically determine the effects of aging on changes in the lubricin distribution in the rotator cuff. Another limitation is that the distribution of lubricin splice variants in the goat infraspinatus tendon was not investigated. Prior work has demonstrated the presence of different splice variants at different locations in the canine flexor digitorum profundus18, suggesting that the lubricin in those locations might have multiple functions, some of which were unrelated to lubrication. Now that the presence of lubricin in the infraspinatus tendon has been established, future work needs to be directed to investigating the specific splice variants that are present.
In conclusion, lubricin is present in interfascicular layers in certain regions of the goat infraspinatus tendon. The results provide a basis for the hypothesis that the episodic breakdown of lubricin by proteases, perhaps associated with transient inflammatory conditions, favors the wear of tendons during interfascicular movement. The rationale for this hypothesis is, however, predicated on a lubricating function of the lubricin found in the infraspinatus tendon, which needs to be supported by future investigations of the splice variants present. Should the lubricin in the infraspinatus tendon be serving as a lubricant, the present work may provide a rationale for the investigation of an injectable form of lubricin as a therapeutic agent to treat certain rotator cuff problems. 