Study Animals
A total of 192 male Sprague-Dawley rats (Harlan, Indianapolis, Indiana), weighing 300 to 350 g, underwent anterior cruciate ligament reconstruction in the right knee with use of an autologous ipsilateral flexor digitorum longus tendon graft. The animals were randomly assigned to one of two groups: the control group (ninety-six rats) or the liposomal clodronate-injected group (ninety-six rats). Animal care was provided according to guidelines established by the Center for Laboratory Animal Services at our institution. Injections were performed without animal sedation. All protocols were approved by the Institutional Animal Care and Use Committee.
Macrophage Depletion by Means of Liposomal Clodronate
To evaluate the role of macrophages in tendon-to-bone healing following anterior cruciate ligament reconstruction, liposomal clodronate was administered to the experimental group to selectively deplete macrophages. Clodronate is a bisphosphonate (Cl2MBP) commonly used to treat metabolic or malignant bone diseases. When encapsulated within liposomes, it is actively and exclusively phagocytosed by macrophages and induces apoptosis without macrophage stimulation32. Liposomes containing clodronate were obtained from the Clodronate Liposomes Foundation (Vrije Universiteit, VUmc, Amsterdam, The Netherlands). In brief, a mixture of phosphatidylcholine (Lipoid, Ludwigshafen, Germany) and cholesterol (Sigma Chemical, St. Louis, Missouri) was suspended in 0.6-M clodronate (a gift of Roche Diagnostics, Mannheim, Germany) and sonicated to produce multilamellar liposomes33. The liposomes were washed twice by ultracentrifugation to remove nonencapsulated clodronate and then were resuspended in phosphate buffered saline solution for injection. Two days prior to surgery, rats in the experimental group were administered 1.33 mL/100 g of liposomal clodronate by intraperitoneal injection. Animals then received weekly injections of the same dose until the time when they were killed34. An additional 0.1 mL of liposomal clodronate was injected directly into the knee joint at the time of surgery. Animals were carefully monitored for signs and symptoms of local or systemic infection. Prior work in our laboratory with use of a similar experimental design at a lower dose of liposomal clodronate did not demonstrate an increased incidence of infection10. The control animals did not receive injections of liposomes without clodronate, because there is no evidence that the liposomes alone would have any effect on healing.
Surgical Procedure
We used a rat model of anterior cruciate ligament reconstruction that we developed in our laboratory11. Animals were sedated by intraperitoneal injection of a mixture of 80 mg/kg of ketamine hydrochloride and 5 mg/kg of xylazine and were maintained under sedation with inhaled isoflurane as necessary. The right hind limb was prepared and draped in sterile fashion. Two 5-mm incisions were made on the medial aspect of the distal part of the leg and the plantar surface of the foot. Dissection of the flexor digitorum longus tendon was carried out at both sites, and the full length of the tendon was harvested. A third incision was made over the knee, and a medial arthrotomy was performed. The native anterior cruciate ligament was excised. With use of an 18-gauge needle (1.27-mm diameter), a bone tunnel was drilled in the proximal aspect of the tibia and the distal part of the femur at the insertion sites of the anterior cruciate ligament. The harvested tendon graft was passed through both bone tunnels. The ends of the tendon graft were secured to the periosteum of the distal aspect of the femur and proximal part of the tibia as well as the collateral ligaments with use of 4-0 Ethibond suture (Ethicon, Somerville, New Jersey). The deep and superficial incisions were closed in routine fashion with use of 4-0 Vicryl (polyglactin) and 4-0 Ethilon suture (Ethicon), respectively. Animals were allowed ad libitum postoperative activity.
Fluorochrome Labeling
Twelve animals in each group were injected with serial fluorochrome labels before they were killed, at seven, fourteen, twenty-eight, and forty-two days, in order to label newly mineralized bone along the bone tunnels. These animals received intraperitoneal injections of 25 mg/kg of calcein (Sigma-Aldrich, St. Louis, Missouri) and 90 mg/kg of xylenol orange (Sigma-Aldrich) at eight and three days, respectively, before they were killed. The seven-day animals received calcein seven days before they were killed and xylenol orange three days before they were killed.
Tissue Preparation
Eight rats in each group were killed by carbon dioxide inhalation at three, seven, ten, fourteen, twenty-eight, and forty-two days following surgery. The involved hind limbs were dissected to the joint capsule, and the femora and tibia were detached. Five specimens from each group were prepared for histologic analysis after decalcification. These samples were fixed in 10% neutral buffered formalin at 4°C for three days. Following fixation, specimens were cut in cross-section perpendicular to the axis of the tendon graft in the bone tunnel with use of a cooled diamond wheel saw and then were decalcified overnight at 4°C in Immunocal (Decal Chemical, Tallman, New York). The tissue was dehydrated in increasing concentrations of alcohol (70% to 100%) and was cleared in three changes of xylene. The tissue was then embedded in paraffin at 60°C. Serial sections that were 5 µm thick were prepared for hematoxylin and eosin staining and immunohistochemistry.
The other three specimens at each time point were prepared for histologic analysis without decalcification. Samples were fixed in 70% ethanol, and coronal sections along the long axis of the tendon graft were cut. The sections were further dehydrated in increasing concentrations of alcohol, were fixed in acetone, and then were embedded in methylmethacrylate. Serial sections that were 5 µm thick were prepared for Goldner-Masson trichrome staining and fluorescence microscopy.
Immunohistochemistry
Immunohistochemistry was used to identify macrophages at the tendon-bone interface. Two distinct subpopulations of macrophages that selectively express ED1 lysosomal glycoprotein antigen or ED2 membrane glycoprotein antigen have been identified18,35,36. ED1+ macrophages represent the monocyte-derived population recruited from the circulation to the site of wound-healing, while ED2+ macrophages are those activated and derived from the local microenvironment.
Hydrated serial sections were treated with 3% H2O2 to quench endogenous peroxidase activity, and nonspecific antibody binding was blocked with 5% goat serum. Mouse anti-rat antibodies to ED1-macrophages (CD68) and ED2-macrophages (CD163) (Serotec, Raleigh, North Carolina) were used to localize macrophages in the healing tendon-bone interface. We also localized TGF-ß using a rabbit polyclonal antibody (Abcam, Cambridge, Massachusetts). Primary antibodies were applied to separate serial sections for sixty minutes at room temperature. Bound antibodies were visualized with use of a goat avidin-biotin peroxidase system with 3,3'-diaminobenzidine (DAB; Dako, Carpinteria, California) as a chromagen. The number of ED1, ED2, or TGF-ß-positive cells were counted in ten high-power fields (50 µm × 50 µm at 160× magnification) per specimen.
Histologic Analysis
Decalcified histologic sections were stained with hematoxylin and eosin for routine histomorphologic analyses with use of a light microscope (Nikon Optiphot; Nikon USA, Melville, New York) connected to a charge coupled device (CCD) camera (Nikon DXM1200; Nikon USA). The histologic sections were reviewed by two authors (P.L.H. and S.A.R.) who were blinded to the group, and the final grade was determined by consensus. Locations along the perimeter of the tendon-bone interface were systematically selected, and the width of fibrovascular tissue formation between tendon and bone was measured (160× magnification). All histomorphometric assessments were made with use of ImageJ software (National Institutes of Health, Bethesda, Maryland). We first examined the tibial and femoral tunnel data separately and found no differences between the two tunnels. The data for the tibial and femoral tunnels were thus combined.
Decalcified sections were also stained with picrosirius red staining and then were examined under polarized light to optimize visualization of collagen fiber continuity between graft and bone37,38. The polarizer was rotated to the position of maximum birefringence (brightness) on each slide. The light intensity on the microscope was kept constant for all specimens to standardize the measurements and reduce the potential for error. Ten randomly selected digital images were collected for each specimen. With use of ImageJ software, the images were converted to 8-bit grayscale on which the brightness of individual pixels was represented on a scale from 0 to 255, with 255 indicating maximum pixel brightness. Average brightness per high power field (50 µm × 50 µm at 320× magnification) was measured at several randomly chosen locations along the tendon-bone interface for each image. The brightness is proportional to collagen fiber density and organization.
We identified newly mineralized bone along the bone tunnels in the fluorochrome-labeled sections. Unstained coronal sections were observed under fluorescent light, and digitized images were recorded along the length of the tendon graft-bone tunnel interface (320× magnification). The distance between fluorochrome labels was measured, and the mineral apposition rate was determined. Serial coronal sections were also stained with Goldner-Masson trichrome stain for quantitative analysis of new osteoid formation at the tendon-bone interface. The thickness of the "seam" of newly deposited osteoid along the bone tunnel was measured with use of ImageJ software (320× magnification). Because the precision of these measurements requires accurate coronal plane sections, we ensured reproducible sectioning by identifying the tunnel entrance and exit, which facilitates accurate sectioning in the coronal plane.
Biomechanical Testing
Twelve rats in each group were killed for biomechanical testing at seven, fourteen, twenty-eight, and forty-two days following surgery. The femur-anterior cruciate ligament graft-tibia construct was harvested and frozen at -80°C until testing. Prior to testing, all soft tissue was dissected leaving the anterior cruciate ligament graft. The tibiae and femora were potted in cement to ensure secure fixation. Specimens were mounted on a materials testing machine (MTS Systems, Eden Prairie, Minnesota) in a specially designed jig so that tension was aligned along the long axis of the tendon graft. A preload of 0.2 N was applied, and then the specimens were preconditioned for five cycles. The specimens were then loaded in uniaxial tension to graft failure. Load to failure (N) and stiffness (N/mm) were determined from the linear portion of the load-deformation curve with use of SigmaPlot 8.0 (SPSS, Chicago, Illinois). The site of graft failure (pullout from bone tunnel compared with midsubstance rupture) was also recorded.
Statistical Methods
The data were compared between groups with use of the Wilcoxon rank-sum test and unpaired t tests. Biomechanical failure mode was analyzed with use of the chi-square test for independence. Differences were considered significant at p < 0.05. All data are presented as the mean and the standard error of the mean.
Animal Model
The surgical procedure was well tolerated with few complications. Three intraoperative deaths (two in the control group and one in the experimental group) and one early postoperative death (control group) occurred secondary to complications with anesthesia. Two animals (both in the experimental group) were killed early because of dehiscence of the surgical incision secondary to the animal chewing at the site. All prematurely killed animals were replaced as per protocol. No adverse clinical observations were noted as a result of the liposomal clodronate injections. There were no deep infections.
Presence of ED1+ and ED2+ Macrophages at the Healing Tendon-Bone Interface
ED1+ macrophages were observed as part of the early inflammatory cell infiltrate at the tendon graft-bone interface at three days following surgery among control specimens (5.1 ± 1.8 cells per high-power field; Fig. 1). By seven days, the number of ED1+ macrophages had doubled to 10.7 ± 1.6 cells per high-power field. Over the remaining time points, the presence of ED1+ macrophages at the tendon-bone interface among control specimens slowly declined to 3.9 ± 1.1 cells per high-power field at forty-two days. In contrast, very few ED2+ cells were observed throughout the early postoperative time points among control animals, ranging from 0.3 ± 0.3 cells per high-power field at three days to a peak of 1.5 ± 0.5 cells per high-power field at twenty-eight days.
There were significantly fewer ED1+ and ED2+ macrophages in the specimens from the macrophage-depleted limbs compared with the control limbs (p < 0.05 for ED1+ at all time points; p < 0.01 for ED2+ at three days, and p < 0.05 at all other times). Following serial injections of liposomal clodronate, the number of ED1+ macrophages at the tendon-bone interface of experimental animals remained constant, averaging 1.8 ± 0.6 cells per high-power field (range, 1.6 ± 0.9 to 2.1 ± 0.9 cells per high-power field) over all time points. This reflected a relative reduction in ED1+ cells of 82% at seven days, the point of peak ED1+ macrophage infiltration among control animals. Similarly, the number of ED2+ interface macrophages observed among treated specimens was constant at 0.1 ± 0.1 cells per high-power field (range, 0.04 ± 0.1 to 0.2 ± 0.1 cells per high-power field) over all time points. The reduction in ED2+ macrophages relative to control specimens ranged from 63% at three days to 85% at twenty-eight days.
TGF-ß Staining at the Healing Tendon-Bone Interface
TGF-ß-positive cells were found in the early inflammatory cell infiltrate at the tendon graft-bone interface at all time points. There was positive staining for TGF-ß in vascular cells and fibroblasts in the tendon-bone interface. Compared with the control limbs, the macrophage-depleted specimens had significantly less TGF-ß staining at every time point except seven days (6.2 ± 1.5 cells per high-power field and 2.1 ± 0.6 cells per high-power field, respectively, at fourteen days; 8.1 ± 0.9 cells per high-power field and 3.1 ± 1.9 cells per high-power field at twenty-eight days; and 6.9 ± 0.9 cells per high-power field and 4.2 ± 1.2 cells per high-power field at forty-two days; p < 0.01 for all) (Fig. 2).
Fibrous Tissue Formation and Remodeling at the Healing Tendon-Bone Interface
Healing between the tendon graft and bone tunnel in control specimens was initiated by the formation of a loose, poorly organized fibrous interface tissue. The interface consisted predominantly of neutrophils, fibroblasts, and mononuclear cells at the earliest time point, with only sparse collagen fibers within the matrix and distinct margins between tendon graft, interface, and bone. There were numerous islands of fragmented bone surrounding the interface along the tunnel. At seven days, the dominant cell morphology consisted of spindle-shaped fibroblasts. Osteoblasts were intermittently seen along the edge of the bone tunnel. By ten days, there was increasing evidence of newly formed trabecular bone along the tendon-bone interface, and the fibrovascular interface tissue became denser and better organized, with evidence of early remodeling around the outer edge of the tendon graft. At fourteen days, increasing numbers of collagen fibers were observed aligning circumferentially around the bone tunnel. Occasional plump, round chondroid-appearing cells were also seen among the fibroblasts. By twenty-eight days, the interface tissue had a variable width among control specimens consisting of different degrees of remodeling, from dense fibrovascular tissue to less frequent areas of early fibrocartilage and collagen fiber continuity with bone. There were also increased numbers of osteoblasts along the bone tunnel with increased trabecular bone surrounding the interface. This pattern of healing continued to progress through forty-two days with more consistent areas of tissue remodeling and the increased presence of collagen fibers extending between the interface and bone.
Tendon-to-bone healing among the macrophage-depleted specimens progressed similarly to that of the control specimens but at an accelerated rate and with more marked interface remodeling, reestablishment of collagen fiber continuity, and direct bone ingrowth into tendon. At all time points, the width of the fibrous interface tissue measured between tendon graft and bone was significantly reduced among experimental limbs compared with controls (Figs. 3-A and 3-B). Three days following surgery, the relative reduction in interface tissue observed between groups was 48% (74.5 ± 13.3 µm compared with 38.8 ± 20.6 µm; p < 0.01). At seven days, additional differences between control and experimental specimens were evident. Tendon graft, interface, and bone tissues remained distinctly separate, but a more consistent perimeter of trabecular bone surrounded the grafts in the experimental group. Randomly dispersed collagen fibers were seen within the fibrous interface tissue, and there was a significant difference in interface width between control and experimental animals (116.5 ± 29.5 µm and 23.7 ± 7.1 µm, respectively; p < 0.05). At ten days, the healing interface in the macrophage-depleted group was characterized by a dense fibrovascular tissue with occasional large, round cells resembling chondrocytes. By fourteen days, distinct areas of collagen fiber continuity between tendon and bone had developed among all macrophage-depleted specimens. Occasional zones of early fibrocartilaginous tissue were seen in addition to an increased quantity of collagen fiber bundles bridging the tendon graft and bone at 45° angles. The reduction in interface width between groups remained significant at ten and fourteen days (106.5 ± 29.8 µm compared with 30.7 ± 3.1 µm, and 124.5 ± 29.1 µm compared with 18.2 ± 5.8 µm, respectively; p < 0.01).
Healing continued to progress through twenty-eight days at which time the degree of interface remodeling among macrophage-depleted animals was significantly improved compared with controls (interface width of 135.5 ± 18.1 µm and 20.4 ± 4.9 µm, respectively; p < 0.01). Occasional areas of fibrovascular tissue were observed, but the interface morphology more commonly demonstrated a gradual transition between tendon graft and bone with early characteristics of a direct, four-zone insertion site. Increased numbers of collagen fibers resembling Sharpey fibers continued to align into the surrounding bone tunnel. By forty-two days, the quality of the tendon-bone interface among experimental sections consistently improved (interface width of 93.0 ± 11.2 µm compared with 18.4 ± 6.9 µm; p < 0.01).
Collagen Fiber Continuity at the Healing Tendon-Bone Interface
There was progressive establishment of collagen fiber continuity between tendon graft and bone tunnel among both groups following anterior cruciate ligament reconstruction. Under polarized light at high magnification, the presence of collagen fibers was recorded as mean birefringence per high-power field (Fig. 4). At seven days, poorly organized, sparse collagen fibers were identified at the tendon-bone interface in the control specimens, and only limited bundles of collagen fibers were present along the perimeter of the bone tunnel by fourteen days. The degree of collagen fiber continuity among control animals continued to increase at twenty-eight and forty-two days with steady improvements in fiber organization and alignment.
In contrast, there was more rapid establishment of collagen fiber continuity in the macrophage-depleted animals compared with the controls at fourteen, twenty-eight, and forty-two days. Similar to the control specimens, the macrophage-depleted animals had birefringent areas of collagen fibers at seven days, but they remained relatively unorganized and only intermittently distributed along the interface. At fourteen days, the mean birefringence observed at the interface of experimental specimens was significantly greater than that of controls (p < 0.05). Markedly increased and more consistent collagen fiber continuity was also present at twenty-eight and forty-two days among the experimental animals compared with controls (p < 0.01).
New Bone Formation at the Healing Tendon-Bone Interface
Newly formed osteoid was observed along the bone tunnels of control and experimental animals at all time points. New bone formation remained relatively constant from three to ten days among control animals (3.1 ± 0.9 µm to 3.2 ± 1.6 µm, respectively) and liposomal clodronate-treated animals (4.4 ± 0.2 µm to 5.6 ± 2.3 µm; p < 0.05 for the difference between groups at ten days), with greater widths of new bone forming along the bone tunnels of experimental limbs (Fig. 5). At fourteen days, an increase in osteoid formation was observed among both control and macrophage-depleted groups (6.6 ± 0.2 µm and 9.3 ± 1.2 µm, respectively; p < 0.05), and a significantly greater width of osteoid was also seen among macrophage-depleted animals at twenty-eight days (4.1 ± 1.0 µm and 8.2 ± 0.9 µm, respectively; p < 0.05). By forty-two days, osteoid formation along the bone tunnel remained high (6.8 ± 1.4 µm for controls compared with 7.8 ± 1.5 µm for the experimental group), but no significant difference was observed between groups.
The mineral apposition rate was determined at seven, fourteen, twenty-eight, and forty-two days by serial fluorochrome labeling of bone observed under fluorescent light. Uptake of one or both of the fluorochromes was inconsistent at seven and fourteen days, limiting the number of specimens available for statistical analysis. Nonetheless, trends were consistent with those observed for new osteoid formation. At seven days, the rates of bone mineralization were similar between control and experimental groups (4.5 ± 0.5 µm/day and 4.9 µm/day, respectively; Fig. 6). An increased mineral apposition rate among macrophage-depleted animals was observed at all other time points, with significant differences compared with controls at twenty-eight and forty-two days (2.9 ± 0.9 µm/day and 6.8 µm/day, respectively, at fourteen days; 2.6 ± 1.1 µm/day and 8.6 ± 0.5 µm/day at twenty-eight days, p < 0.05; and 5.5 ± 1.4 µm/day and 9.4 ± 1.4 µm/day at forty-two days, p < 0.05).
Biomechanical Testing
Of ninety-six specimens, ten were discarded as a result of either graft failure discovered at the time of killing (one control and three experimental animals) or inadequate dissection leading to inaccurate load and stiffness values (three control and three experimental animals). These animals were not replaced.
Among control specimens, load to failure and stiffness increased significantly from seven to fourteen days (p < 0.01; Fig. 7) and from fourteen to twenty-eight days (p < 0.01). Between twenty-eight and forty-two days, there was no significant change in load to failure and stiffness among control animals. Similarly, a significant increase in load and stiffness was observed among macrophage-depleted limbs between seven and fourteen days (p < 0.01) and between twenty-eight and forty-two days (p < 0.01), but not between fourteen and twenty-eight days.
At seven days, the control and experimental groups had similar findings with respect to load to failure (1.5 ± 0.7 N and 1.6 ± 1.0 N, respectively) and stiffness (1.1 ± 0.5 N/mm and 1.6 ± 1.1 N/mm). At fourteen days, a significant increase was observed among macrophage-depleted limbs compared with the controls with regard to both load to failure (7.9 ± 2.1 N and 6.1 ± 2.6 N; p < 0.05) and stiffness (8.0 ± 4.4 N/mm and 4.5 ± 3.4 N/mm; p < 0.01). At twenty-eight days, the control specimens exhibited significant increases compared with macrophage-depleted limbs with respect to mean load (10.5 ± 2.4 N and 9.2 ± 2.8 N, respectively; p < 0.05) and mean stiffness (8.7 ± 4.1 N/mm and 6.5 ± 2.0 N/mm). By forty-two days, experimental animals again had significant increases compared with the controls with respect to mean load to failure (13.5 ± 4.2 N and 9.7 ± 3.9 N, respectively; p < 0.05) and mean stiffness (11.5 ± 5.0 N/mm and 7.5 ± 3.2 N/mm; p < 0.05).
The failure mode during biomechanical testing was recorded as tunnel pullout, tendon midsubstance failure, or combined pullout and midsubstance failure (Table I). The pattern of graft failure was similar between the groups. The mode of failure observed in both groups at seven and fourteen days, which was predominantly by graft pullout, was significantly different from that at twenty-eight and forty-two days, which more commonly involved either complete or partial failure at the midsubstance of the tendon (chi-square test; p < 0.05).
In this study, we used a rat model of anterior cruciate ligament reconstruction to study healing of a tendon graft in a bone tunnel. We developed this model to take advantage of the availability of rat-specific molecular and immunohistochemical reagents. Our previous work with this model characterized the inflammatory cell infiltrate that is present in the early healing period and suggested the important role of macrophages in the early healing process between tendon and bone11. Since studies by us and others have shown that there is formation of a "scar tissue" interface between tendon and bone rather than reformation of a normal insertion site, we hypothesized that the presence of abundant macrophages plays an important role in this scar formation. In other experimental models, the presence of macrophages and overexpression of TGF-ß has been linked to states of chronic inflammation and tissue fibrosis14,15,21,25,28,30,31. In contrast, embryonic wounds heal in the absence of an inflammatory cell infiltrate with recapitulation of normal, "scar-less" tissue morphology. Studies of macrophage depletion have demonstrated substantial reductions in the degree of fibrosis as well as accelerated tissue healing14,19,39,40. Moreover, inhibition of TGF-ß in wounds has been shown to reduce tissue fibrosis14,27,28, while overexpression results in a chronic fibrotic state15,21,31.
We selectively depleted macrophages using a liposome-encapsulated bisphosphonate, liposomal clodronate. Liposomal clodronate has been used by other investigators to demonstrate the antifibrotic and regenerative effects of macrophage depletion in models of organ fibrosis32,40,41. The effects of liposomal clodronate on macrophages have been shown to last for two weeks following administration42. To our knowledge, this is the first study to examine the effect of macrophage depletion on tendon-to-bone healing in a ligament reconstruction model.
Healing at the tendon-bone interface in this model begins with an early inflammatory cell infiltrate, followed by formation of a loose, poorly organized fibrous tissue interface. As healing proceeds, there is gradual bone ingrowth into tendon and progressive reestablishment of collagen fiber continuity between interface and bone. We did not find consistent reformation of a native ligament insertion morphology, with a fibrocartilage transition zone between ligament and bone. The pattern of macrophage infiltration at the tendon graft-bone tunnel interface in our control specimens was similar to that in a previous study in our laboratory11. ED1+ macrophages, recruited from circulating blood monocytes, were present by three days following surgery, peaked around seven days, and subsequently declined in number. ED2+ cells, derived from the local tissue environment, accumulated in smaller numbers by seven days and peaked by twenty-eight days. Among animals administered liposomal clodronate, the numbers of ED1+ and ED2+ macrophages were significantly reduced at all time points through forty-two days (p < 0.05), with a decrease in total cell number ranging from 56% to 87% for ED1+ cells and 63% to 85% for ED2+ cells.
Tendon-to-bone healing among experimental (macrophage-depleted) animals proceeded at an accelerated rate compared with untreated control specimens, with significantly reduced formation of loose, fibrovascular tissue and earlier tissue organization; closer apposition of tendon graft to bone; and an increase in the number and size of collagen fiber bundles resembling Sharpey fibers that were perpendicularly aligned at the tendon-bone interface as early as fourteen days following surgery. Bone formation as measured by newly forming osteoid along the perimeter of the bone tunnel was significantly increased in macrophage-depleted animals at fourteen and twenty-eight days, suggesting increased activity at the bone tunnel with enhanced bone ingrowth into tendon. Furthermore, by twenty-eight days, areas of remodeling fibrocartilage extending from the bone tunnel to the graft were observed among macrophage-depleted animals, suggesting partial reestablishment of a healing morphology more similar to that of a native enthesis. At forty-two days, distinct areas of fibrocartilage transition from tendon to bone similar to the zonal pattern of native attachment sites were more often observed with consistent reestablishment of collagen fiber continuity between tendon and bone. These morphologic changes resulted in significant improvements in biomechanical load to failure and graft stiffness compared with controls by forty-two days.
The formation of fibrous scar tissue at the healing tendon-bone interface is the result of extracellular matrix production by activated fibroblasts. The extent of extracellular matrix deposition and fibrosis is largely under the influence of macrophage-mediated cytokine production. Following tissue injury, platelet degranulation releases macrophage chemoattractants including TGF-ß. This early population of macrophages recruited from circulating blood monocytes undergoes "classical activation"39,40,43,44. These macrophages function in débridement of the local tissue environment and phagocytosis of apoptotic inflammatory cells. Classically activated macrophages release proinflammatory cytokines such as interleukin (IL)-1, IL-6, and TNF-a as well as matrix proteases, which aid in cell chemotaxis through the inflammatory milieu. Prior studies have shown that macrophage phagocytosis of apoptotic cells results in a phenotypic transition from "classical" to "alternative" activation39. Alternatively activated macrophages, in contrast, release anti-inflammatory and profibrogenic cytokines. These cells produce TGF-ß and platelet-derived growth factor and contribute to myofibroblast activation and proliferation, resulting in increased extracellular matrix synthesis, remodeling, and scar formation.
The balance between matrix remodeling and excessive fibrosis and the roles of macrophages in these processes is complex and incompletely understood. We found diminished TGF-ß expression in concert with diminished macrophage numbers. The role of cytokines in tissue healing is complex and dependent on numerous factors, including the timing of application, concentration, and the cell types present. For example, a number of studies have suggested that exogenously applied TGF-ß may enhance healing, extracellular matrix synthesis, and remodeling15,23,45,46. Yamazaki et al. recently described the effects of exogenously applied TGF-ß on tendon-bone healing in a canine model of anterior cruciate ligament reconstruction46. In that study, the TGF-ß-treated grafts demonstrated increased load and stiffness over the fibrin glue and negative controls as well as increased collagen fiber synthesis and new bone formation along the tunnel. These effects were due to increased formation of fibrous scar tissue in the tendon-bone interface. These results support our hypothesis that macrophage-derived cytokines, such as TGF-ß, lead to reactive scar-tissue formation at the healing tendon-bone junction. Even though the addition of TGF-ß enhanced tendon-bone healing in the described study, healing was "improved" only by formation of a greater quantity of scar tissue, which likely has inferior material properties compared with a normal insertion site. Our hypothesis is that blocking of the excessive inflammatory response allows expression of "innate" healing mechanisms and regeneration of a more normal insertion site rather than reactive scar formation.
The role of macrophages in wound-healing is complex, and we acknowledge that alteration of macrophage numbers is likely to have myriad effects on healing. Macrophages present during the early inflammatory phase of healing that produce proinflammatory mediators may have a negative effect on tissue regeneration. In the present study, we suggest that macrophage depletion led to abrogation of the early "destructive" phase of inflammatory healing, perhaps allowing expression of genes appropriate for tissue regeneration rather than reactive scar formation. Activated myofibroblasts and neutrophils likely compensated for the phagocytic responsibilities typically performed by macrophages29.
However, macrophages play an important role in the resolution of inflammation by phagocytosis of apoptotic neutrophils before these cells lyse. Resolution of inflammation is necessary for a healing wound to progress to matrix synthesis and maturation, and it contributes to the activation of macrophages with a reparative phenotype47-50. The cytokine profile in the healing wound also changes as inflammation resolves. For example, Dahlgren et al. studied growth factor expression in tendon healing and found an early peak of TGF-ß1 at one week, coinciding with the early proinflammatory period51. Of particular interest, this group suggests that inhibition of early TGF-ß1 expression could result in downregulation of the inflammatory cytokine environment, allowing earlier expression of insulin-like growth factor (IGF)-I (which has a strong positive role in tendon healing) in its restorative and proregenerative capacities. A similar mechanism may explain our results, as macrophage depletion diminished TGF-ß1 expression, which may permit earlier IGF-I activity and accelerate regenerative healing of tendon to bone.
Further study is required to determine whether healing can be improved by selective inhibition of only the earliest proinflammatory macrophage phenotype. It is unknown whether it is possible to "uncouple" the early proinflammatory and late reparative macrophage populations. It is possible that the induction-autoinduction mechanisms are so interdependent that knocking out early macrophages would by default impair their late regenerative activity.
Although the improved biomechanical results in the controls at twenty-eight days may simply be due to biologic variability, consideration of the role of reparative macrophage populations in the later phases of healing suggests a possible mechanism to explain this reversal in the biomechanical results. The presence of such reparative macrophages and associated cytokines in the healing environment in the control animals may lead to increased fibrous tissue formation and thus improved attachment strength. However, we hypothesize that there is a limit to the ultimate attachment strength of this fibrous scar-tissue interface. The closer apposition of tendon and bone in the macrophage-depleted animals (due to less scar tissue in the interface) may allow for continued development of collagen fiber continuity and ultimately a stronger attachment. However, this is currently speculative and further study is required to test this proposed mechanism.
Alteration of macrophage numbers may also affect healing through an effect on matrix metalloproteinases (MMPs) and their inhibitors (tissue inhibitors of metalloproteinases [TIMPs]). The balance between extracellular matrix synthesis and degradation is closely related to the activities of these molecules52-54. Macrophages are known to produce MMPs and TIMPs52,53. In a recent study, Demirag et al. studied the effects of synovial matrix metalloproteinase activity on tendon-bone healing using a2-macroglobulin as an MMP-blocking agent in a rabbit model of anterior cruciate ligament reconstruction55. They found that, following treatment with a2-macroglobulin, the specimens had decreased postoperative enzymatic activity and improved graft-healing with enhanced collagen fiber continuity and biomechanical strength. Considering this evidence, it is possible in the present study that depletion of macrophages following anterior cruciate ligament reconstruction may have altered the levels of early postinjury MMPs and their relative balance with other cytokines, leading to reduced collagenase-mediated matrix destruction and enhancement of tendon graft-healing.
Our study has several limitations. We evaluated animals only to forty-two days after surgery. However, our previous work demonstrated that the early inflammatory phase and the phase of cellular and matrix proliferation are largely resolved by forty-two days in the rat model. Further study is required to examine the effect of macrophage depletion on late remodeling. Further study is also required to elucidate the complex functions of macrophages, their role in tendon healing, and the interaction between mechanical load on the healing tendon-bone interface and macrophage activity. Another important consideration is that osteoclasts are derived from cells closely related to (if not the same as) the precursors of macrophages. As a result, besides depleting macrophages, our clodronate-loaded liposomes may have also had an effect on osteoclasts. Some of the differences in remodeling and mineralization could reflect an inhibition of osteoclasts. Another limitation is that we did not use a control group that received liposomes without clodronate; however, there is no evidence that the liposomes alone would have any effect on healing.
In conclusion, the results of this study suggest that macrophage depletion enhances tendon-to-bone healing in a rat model of anterior cruciate ligament reconstruction. We found changes in the structural, compositional, and biomechanical properties of the healing tendon-bone interface. Healing of a tendon graft in a bone tunnel in macrophage-depleted animals progressed at an accelerated rate on the basis of the reestablishment of collagen fiber continuity, new bone formation, and the overall degree of interface remodeling. There was less fibrous tissue (scar) formation at the healing tendon-bone interface, perhaps because of decreased expression of TGF-ß. Likewise, biomechanical properties of load-to-graft failure and graft stiffness were significantly improved at forty-two days with macrophage depletion. Clearly, the process of tendon-to-bone healing is complex, with multiple simultaneous and compensatory pathways governing early postoperative remodeling. These findings suggest the need for further study of the role of macrophages in tendon-to-bone healing in a large animal model to further delineate the cellular and molecular mechanisms of healing. 