ACL tears are an increasingly common sports injury. If untreated, these injuries may result in knee instability and secondary damage like meniscal tears and osteoarthritis, as well as an increased risk of reinjury of the knee (Finsterbush et al., 1990). Surgical reconstruction of the ACL is required to stabilize the joint and to help protect against further injuries in young, active populations (Dunn et al., 2004).
Hamstring tendons are the most commonly used autologous grafts in ACL reconstruction because of their similar composition, both of which are composed of closely packed collagen bundles (Kannus, 2000). However, the gross morphology of the ACL is more complex than hamstring tendons. For example, some researchers have divided ACL into two bundles and suggested using double bundle hamstring tendons to anatomically reconstruct the ACL (Girgis et al., 1975). Meanwhile, others pointed out that the ACL is composed of multiple bundles instead of two bundles, and each with their own morphological patterns (Amis and Dawkins, 1991; Hollis et al., 1991; Amiri et al., 2011).
The ACL bundles are comprised of many small collagen fibers, whose arrangement has not been well described and correlated with those of the hamstring tendons. Structural differences in the tendons might be expected since the functions of the ACL and hamstring tendons are obviously different. A tendon, such as the hamstring tendon, is a tough band of fibrous connective tissue that connects muscle to bone and is capable of transmitting forces and withstanding tension during muscle contraction. In contrast, the ACL is an important restraint to the anterior translation and rotation of the tibia relative to the femur (Markolf et al., 1976; Butler et al., 1980; Welsh, 1980). The ACL is required to withstand multiaxial stresses and regionally different tensile strains, whose distribution is not uniform even along the same bundle (Hirokawa et al., 2001). As function is related to structure, it would be expected that the structure and ultrastructure of collagen fiber bundles would be different between the ACL and hamstring tendons according to their different biomechanical properties.
The structure and ultrastructure of tendon grafts and the normal ACL have previously been investigated (Yahia and Drouin, 1989; Strocchi et al., 1992, 1996), but detailed descriptions of collagen fibers, such as their diameter and density in the ACL and hamstring tendons, are inconsistent (Strocchi et al., 1992; Hadjicostas et al., 2007, 2008). We therefore first examined the morphological and ultrastructural features of ACL and semitendinosus and gracilis tendons as these are candidate graft tissues. Then we examined the differences in cell type and morphology of collagen fibers at the proximal, medial, and distal regions from a single specimen of the ACL or hamstring tendons. Lastly in normal ACL and hamstring tendons, the arrangement, diameters, and densities of collagen fibers, which are important properties of collagen structure, were analyzed. Our findings provide fundamental new information about the structure and ultrastructure of the ACL and tendons that are used as surgical substitutes. These observations may provide a standard for evaluating hamstring tendon grafts after ACL reconstruction and could facilitate the application of these grafts in clinical settings.
MATERIALS AND METHODS
Human hamstring tendons and ACL were obtained surgically. Twenty semitendinosus or gracilis tendons were harvested from patients with ACL rupture (average age of 25.3 ± 7.3). About 1 cm3 of semitendinosus or gracilis tendon was removed from the middle part, at the same time not to affect the function of grafts for ACL reconstruction. Patients with hamstring tendon injuries were excluded. Twenty whole ACL samples were harvested from patients with osteoarthritis (OA) (average age 65.3 ± 5.6) who had undergone routine total knee arthroplasty. OA patients with ACL ruptures were excluded. This study met the ethical standards suggested by Harriss and Atkinson (2009) and was approved by the Ethics Committee at Third Hospital of Peking University. All patients gave their informed consent prior to their inclusion in the study.
For light microscopy, ten of both kinds of specimen were randomly selected, fixed in formalin (4% formalin in phosphate-buffered saline) and embedded in paraffin. Six-micron thick tissue sections were cut from these paraffin blocks and stained with Hematoxylin and Eosin (H&E), toluidine blue and Mallory stains. The distribution of decorin (DCN) was evaluated using immunohistochemistry with a specific antibody to human decorin (Sigma–Aldrich). Before staining, the samples were treated with 0.25% trypsin for 3 min and then incubated with the primary antibody overnight. Following PBS washing, the peroxidase-conjugated antibody was added for 1 hr at 37°C, and then the chromogenic color was developed by incubation with 3,3-diaminobenzidine (DAB). Each specimen was viewed by light microscopy (Olympus BH-2, Tokyo, Japan) and photographs were taken from three randomly chosen representative microscopic fields at a magnification of 40–400×.
Ten of each kind of specimen were trimmed and placed immediately in 3% glutaraldehyde solution (buffered with 0.1M phosphate, pH 7.4) and fixed for at least 48 hr at room temperature. Then about 1 mm3 tissues were cut from each specimen prepared for transmission electron microscopy (TEM). The remainders of the specimens were used for scanning electron microscopy (SEM).
For SEM, cryofracture (Provenzano and Vanderbyjr, 2006) was used to manually fracture the specimen along the fiber direction to expose the smooth surface of the collagen fibers. In brief, the specimens were immersed in liquid nitrogen, placed on a precooled metal block, and immediately fractured with a microsurgical scalpel blade in the sagittal and coronal planes. The tissues were then dehydrated through a series of graded ethanol/H2O solutions (30, 50, 75, 90, and 100% ethanol). The tissues were critical-point dried and were mounted on 10-mm SEM mounting blocks, coated with gold and stored in a vacuum container. The samples were imaged with a JEOL JSM 5600LV (JEOL, Tokyo, Japan) SEM. Areas for imaging were randomly selected, but particular attention was paid to those areas adjacent to the fracture surfaces.
For TEM, the tissues were postfixed in 1% osmium tetroxide for 2 hr at room temperature, dehydrated in a graded series of ethanol, treated with propylene oxide, and embedded in Epon 812. Semi-thin sections (1 μm) were cut and stained with 1% toluidine blue for light microscopy. Ultrathin sections (40–60 nm) were cut using a Leica UltraCut ultramicrotome (Leica, Wetzler, Germany). The cross-sections were contrasted with uranyl acetate and lead citrate and prior to examination by TEM (JEM-1230, JOEL, Tokyo, Japan). For each sample, five TEM microphotographs at a final magnification of 40,000× were selected using a systematic random sampling method and submitted for morphometric analysis using Image-Pro Plus 6.0 software (Media Cybernetics, Silver Spring, USA).
For each cross-sectional area, the number of collagen fibrils, area of each fibril, and long and short diameter of each fibril were quantified. The cut plane was often not perfectly perpendicular to the fibril axis, so the cross-section of the fibrils frequently showed an elliptical shape in the microphotographs. In this context, the short diameter of each ellipse was the variable closest to the real diameter of the fibril, so the number of fibrils and their short diameter were used for the final statistical analysis. For each specimen, diameters of 100 collagen fibrils were analyzed.
All histological and electronic microscopy images were analyzed independently by three different researchers. The reports were collected and summarized by an additional researcher. Data of the density of fibrils were given as mean ± SD. The short diameters of fibrils were divided into five classes based on the distribution trends reported in the literature (Hadjicostas et al., 2007, 2008): Class 1 (<50 nm), Class 2 (50–100 nm), Class 3 (100–150 nm), Class 4 (150–200 nm), and Class 5 (>200 nm). The differences in the same classes between the study groups were analyzed with the Pearson Chi Square test.
The sections of the semitendinosus and gracilis tendons stained with HE and toluidine blue showed a homogeneous arrangement of collagen fibers and cell type, while the ACL showed a different arrangement at the proximal, medial, and distal regions of a single specimen. The surface of the collagen fibers in hamstring tendons appeared smooth with planar waveform in longitudinal section. The elongated cells with negative toluidine blue staining were arranged along the fiber bundles (Fig. 1A,B,E,F). Most parts of the ACL were composed of parallel arranged fibers and spindle-shaped cells similar to that observed in tendons. However, in some areas there were also some fibers crossing each other which were rarely seen in tendons (Fig. 1C). Near the insertion sites of the ACL, the collagen fibers appeared disordered and loosely arranged with varied types of cell morphology (Fig. 1D). These cells were round or oval in shape with lacunae, resembling fibrochondrocytes. Additionally, the toluidine blue staining was positive around these cells, indicating the presence of greater amounts of glycosaminoglycans (Fig. 1G,H). Meanwhile, the expression of decorin, which plays major roles in collagen fibrillogenesis, was found in both hamstring tendons and ACL (Fig. 1I–L). Immunohistochemical staining of DCN was homogeneous positive in whole semitendinosus and gracilis tendons, but the medium region of ACL showed stronger DCN staining than the two ends (Fig. 1K,L).
Morphology of Collagen Fibers
The collagen fibers in the ACL exhibited a combination of parallel, helical, nonlinear networks, while the hamstring tendons were composed of parallel collagen fibers with planar waveform. When viewed by SEM, hamstring tendons were composed of closely packed collagen bundles sheathed in a loose connective tissue. The bundles were further subdivided into a number of smaller fascicles by the sheath. The collagen fibrils constituting the fascicle exhibited a planar waveform parallel to its long axis at a magnification of 1,000×. When magnified to 5,000×, a smooth surface of compacted collagen fibrils could be seen (Fig. 2A–D). The collagen fibers in ACL showed a combination of parallel, helical, nonlinear networks. Two types of fibers could be found in ACL. The undulated fiber, known as “crimp,” was the main composition of ACL and tendons. The other type, the helical fiber, was unique to the ACL (Fig. 2E,F). In ACL, a high density of collagen fibers constituted a subfascicular unit. Among them there were gaps filled with tangled fibers (Fig. 2G). These fibers had no uniform pattern, and their waveforms were disordered. In sagittal sections, ACL showed a lamellar layout. The direction of fibers in different lamellae was different (Fig. 2H).
Diameter and Density of Fibrils
Gracilis and semitendinosus tendon displayed an asymmetrical distribution of fibril diameters and lower density, while in the ACL fibril diameters were smaller and more homogeneous and the density was greater. The fibrils were divided into five classes by diameter. The large fibrils (>150 nm) accounted for more than one-third of fibrils in the gracilis and semitendinosus tendons, while only a few small fibrils (<50 nm) were found (Fig. 3A,B). In the ACL, fibril diameters between 50 and 100 nm were the greatest proportion and there were fewer large fibrils in the ACL than in hamstring tendons. Two patterns of fibrils could be observed in all ACL specimens. One had uniform medium diameters and the other had nonuniform variable diameters (Fig. 3C–E). The mean density of fibrils in ACL was much higher than normal hamstring tendons (P < 0.01). All data are presented in Table 1.
Table 1. Diameter distribution and density of collagen fibrils in ACL and hamstring tendons
Hamstring tendons are commonly used as a surgical substitute for ACL reconstruction. To achieve optimal anatomical reconstruction and help to restore the function of the ACL, it is important fully understand the structure of the whole ACL and hamstring tendons, the arrangement of collagen fibers and differences in fibril diameter and density.
The ACL and hamstring tendons are composed of cells and extracellular matrix. Type I collagen is the major collagen of the extracellular matrix and is responsible for the tensile stresses of the ACL and tendons (Duthon et al., 2005). The ACL has other types of collagen, including Type III, IX, and XI. The fibroblast is the major cell type in both ACL and tendons. Previous studies and our results showed that ACL has two major cell types. One is fusiform or spindle-shaped fibroblasts with negative toluidine blue staining. The other type is round or oval, resembling fibrochondrocytes with lacunae. (Murray and Spector, 1999; Jiang et al., 2001; Wang and Ao, 2004; Duthon et al., 2005). The chondrocyte-like cell mainly located at the insertion sites and middle part of the ACL. These regions showed stronger positive with toluidine blue staining and were more solid and cartilage-like in nature, which indicates a higher expression level of GAG in ACL than hamstring tendons. Aggrecan and versican, two members of large modular PGs, and their partner hyaluronan are likely to provide a higher capacity to resist compressive forces to tendon tissues (Yoon and Halper, 2005). The differences in these components contributes to the ACL having different histological characteristics from tendons or medial collateral ligaments (MCL) (Jiang et al., 2001). It has also been observed that ACL derived cells have the character of mesenchymal stem cells and could differentiate into adipogenic, osteogenic, and chondrogenic lineages (Furumatsu et al., 2010).
The arrangement of collagen fibers is more complex in the ACL than hamstring tendons as observed by both light and electronic microscopy. Fibers in ACL have two types of wave pattern—an undulated wave pattern, known as “crimp,” and a helical wave pattern. These have been reported in canine patellar tendon and ACL (Yahia and Drouin, 1989). During tensile stretch, fibril “crimp” is first straightened out by small loads, after which larger loads are needed to elongate these fibrils (Duthon et al., 2005). Helical fibrils are presented where the tissue must resist multidirectional or unpredictable loading, or where large, reversible changes in size and shape must be accounted for. A typical example is elastic ligaments, where collagen fibrils waving along elastic fibers are progressively straightened and tensioned whenever the ligament is stretched (Ottani et al., 2001). Unlike the uniform pattern of fibers in hamstring tendons, there were more and wider “gaps” with loosely arranged fibers at the two ends of the ACL. Considering the anatomy and function of the insertion site, this may help to buffer the force from ligament to bone and give space for vessels and cells to grow in.
Proteoglycans represent <3% of the dry weight of ligaments and it has been confirmed that the small leucine-rich proteoglycan, decorin, is the most abundant proteoglycan presented in the matrix of ligaments (Ilic et al., 2005). We found that decorin expressed homogeneously in whole semitendinosus and gracilis tendons while inhomogeneously in ACL. It has been reported that Decorin binds to collagen fibrils and actively participates in fibrillogenesis, and it may slow lateral fibril fusion, which could result in uniformly thinner fibrils under certain conditions (Reed and Iozzo, 2002). The structure of collagen fibers supports the toughness of the ACL, and different distributions of decorin may result in the inhomogeneous diameter by affecting fibrillogenesis in different regions of ACL.
Diameter and density are important properties of collagen structure. These properties are known to correlate with fibril functions (Raspanti et al., 1990). Prior studies have shown that the ultimate tensile strength of fibrils is greater in larger fibrils, while the interfibrillar binding, which depends on their surface areas, increases in the smaller fibrils. Each fibril is therefore likely to fulfill a defined functional role (Ottani et al., 2001). Large and nonuniform fibrils are a typical component of tendons and reticular dermis, whose main function is to resist stretching. In contrast, small and uniform fibrils are often found in tissues which need to withstand multidirectional stresses, such as in vascular or intestinal walls (Raspanti et al., 1990). Both large and small fibrils exist in the ACL (Strocchi et al., 1992). Our results showed that ACL has a major proportion of small fibrils, with large fibrils (>150 nm) comprising only about 10%. In contrast, hamstring tendons have a large proportion of large fibrils, and wider distribution of fibril diameters. These results are consistent with different functions of ALC and tendons, similar to those results previously reported (Hadjicostas et al., 2007, 2008).
Collagen fibril densities are changed in response to repetitive compressive and shear stresses in skin tissue, indicating that collagen density should be another indicator for biomechanical properties of connective tissues (Sanders and Goldstein, 2001). In cross-section, the ACL has a higher but more nonuniform density of fibrils than hamstring tendons. These findings suggest that hamstring tendons used as a substitute may provide enough resistance to tensile strength but ACL has better interfibrillar binding and thus creep resistance.
After surgical implantation, the original hamstring tendon graft undergoes remodeling and neo-ligamentization. In this process, the graft regains a blood supply and cells increase and relocate. The collagen fibers synthesized by the cells remodel under the prevailing stresses and therefore the ultrastructure of the grafts change during the process (Arnoczky et al., 1982; Hunt et al., 2005). In consideration of the substantial differences in ultrastructure between these two kinds of tissues, how the tissues remodel and adapt to the new environment is a clinical concern. Several studies have been done on the ultrastructural changes during graft remodeling and differences fibril diameter between the regenerated tissue and the normal semitendinosus tendon graft have been reported (Yoshiya et al., 2004). It has been shown that the hamstring tendons undergo remodeling during the first 2 years after surgery but do not completely resemble the ultrastructure of a normal ACL up to 10 years later (Zaffagnini et al., 2009). So the ultrastructure after surgical replacement might therefore be different from both normal ACL and hamstring tendons, and this could have an effect on ultimate function of the graft. Further studies, however, will be required to define changes in ultrastructure with time in grafted tissues.
The results presented here should to be considered with some caution due to the different ages of ACL and hamstring tendons donors used in this study. Prior studies have documented age-related structural changes in these connective tissues. For example, the diameters of the collagen fibers of human Achilles' tendon were reported to be greater in the 20- to 29-year-old group compared to other groups (Sargon et al., 2005). Additionally in human ACL, fibril diameter was related to the aging process, but the differences were more obvious between young (<20 years) and adult or the elderly (Strocchi et al., 1996). Our previous study indicated that the distribution trend of fibril diameter in adult and elder patients were similar (Jiao et al., 2010).
In conclusion, we compared the ultrastructure and morphology of human hamstring tendons and the anterior cruciate ligament. We demonstrated that the ACL has different cell types, a higher expression level of GAG, a more complex morphology of collagen fibers and distribution of decorin compared to hamstring tendons. The hamstring tendons showed more large fibrils and lower fibril density than the ACL. This work may provide fundamental information on these tissues that may be useful in the application of tendons in the clinical reconstruction of the ACL following injuries.