Expression of Osteoprotegerin, Receptor Activator of NF-κB Ligand (Osteoprotegerin Ligand) and Related Proinflammatory Cytokines During Fracture Healing

Authors

  • Tamiyo Kon,

    1. Department of Orthopedic Surgery, Boston University Medical Center, Boston, Massachusetts, USA
    2. Department of Orthopedic Surgery, Chiba University School of Medicine, Chuo-ku, Chiba, Japan
    Search for more papers by this author
  • Tae-Joon Cho,

    1. Department of Orthopedic Surgery, Boston University Medical Center, Boston, Massachusetts, USA
    2. Department of Orthopedic Surgery, Seoul National University College of Medicine, Chongno-gu, Seoul, Korea
    Search for more papers by this author
  • Toshimi Aizawa,

    1. Department of Orthopedic Surgery, Boston University Medical Center, Boston, Massachusetts, USA
    2. Department of Orthopedic Surgery, Tohoku University School of Medicine Seiryo-machi, Aoba-ku, Sendai, Japan
    Search for more papers by this author
  • Masashi Yamazaki,

    1. Department of Orthopedic Surgery, Chiba University School of Medicine, Chuo-ku, Chiba, Japan
    Search for more papers by this author
  • Nasser Nooh,

    1. Department of Periodontology and Oral Biology, Boston University School of Dental Medicine, Boston, Massachusetts, USA
    Search for more papers by this author
  • Dana Graves,

    1. Department of Periodontology and Oral Biology, Boston University School of Dental Medicine, Boston, Massachusetts, USA
    Search for more papers by this author
  • Louis C. Gerstenfeld,

    1. Department of Orthopedic Surgery, Boston University Medical Center, Boston, Massachusetts, USA
    Search for more papers by this author
  • Thomas A. Einhorn

    Corresponding author
    1. Department of Orthopedic Surgery, Boston University Medical Center, Boston, Massachusetts, USA
    • Address reprint requests to: Dr. Thomas A. Einhorn, Boston University School of Medicine, Doctors Office Building, Suite 808, 720 Harrison Avenue, Boston, MA 02118, USA
    Search for more papers by this author

Abstract

Fracture healing is a unique biological process regulated by a complex array of signaling molecules and proinflammatory cytokines. Recent evidence for the role of tumor necrosis family members in the coupling of cellular functions during skeletal homeostasis suggests that they also may be involved in the regulation of skeletal repair. The expression of a number of cytokines and receptors that are of functional importance to bone remodeling (osteoprotegerin [OPG], macrophage colony-stimulating factor [M-CSF], and osteoprotegerin ligand [receptor activator of NF-κB ligand (RANKL)]), as well as inflammation (tumor necrosis factor α [TNF-α] and its receptors, and interleukin-1α [IL-1α] and -β and their receptors) were analyzed over a 28-day period after the generation of simple transverse fractures in mouse tibias. OPG was expressed constitutively in unfractured bones and elevated levels of expression were detected throughout the repair process. It showed two distinct peaks of expression: the first occurring within 24 h after fracture and the second at the time of peak cartilage formation on day 7. In contrast, the expression of RANKL was nearly undetectable in unfractured bones but strongly induced throughout the period of fracture healing. The peak in expression of RANKL did not correlate with that of OPG, because maximal levels of expression were seen on day 3 and day 14, when OPG levels were decreasing. M-CSF expression followed the temporal profile of RANKL but was expressed at relatively high basal levels in unfractured bones. TNF-α, lymphotoxin-β (LT-β), IL-1α, and IL-1β showed peaks in expression within the first 24 h after fracture, depressed levels during the period of cartilage formation, and increased levels of expression on day 21 and day 28 when bone remodeling was initiated. Both TNF-α receptors (p55 and p75) and the IL-1RII receptor showed identical patterns of expression to their ligands, while the IL-1R1 was expressed only during the initial period of inflammation on day 1 and day 3 postfracture. Both TNF-α and IL-1α expression were localized primarily in macrophages and inflammatory cells during the early periods of inflammation and seen in mesenchymal and osteoblastic cells later during healing. TNF-α expression also was detected at very high levels in hypertrophic chondrocytes. These data imply that the expression profiles for OPG, RANKL, and M-CSF are tightly coupled during fracture healing and involved in the regulation of both endochondral resorption and bone remodeling. TNF-α and IL-1 are expressed at both very early and late phases in the repair process, which suggests that these cytokines are important in the initiation of the repair process and play important functional roles in intramembraneous bone formation and trabecular bone remodeling.

INTRODUCTION

FRACTURE HEALING, like any other response to trauma, is accompanied by a cascade of events that begins with an inflammatory reaction.(1–3) During this inflammatory phase, macrophages and other immune cells present at the fracture site release interleukin-1 (IL-1), IL-6,(4) and other cytokines associated with the innate tissue response to injury or microbial challenge.(5) These cytokines enhance extracellular matrix synthesis, stimulate angiogenesis, and recruit endogenous fibrogenic cells to the injury site.(6–9) At the end of the fracture repair sequence, woven bone formed by both endochondral and intramembraneous pathways bridges the fracture gap and is resorbed and replaced by lamellar bone through a process of bone remodeling.(10, 11) The remodeling of fracture callus is crucial in restoring mechanical integrity and integrating and incorporating the new bone into the preexisting skeletal structure.

Studies to date have shown that IL-1 and tumor necrosis factor α (TNF-α) play important regulatory roles in bone remodeling and homeostasis.(12–17) Through a variety of mechanisms, these cytokines regulate osteoclast activity either by stimulating hemopoietic progenitor cells to differentiate along an osteoclastic lineage and fuse to form more osteoclasts, or directly stimulate existing osteoclasts enhancing their resorptive capacity. TNF-α participates in the cellular response to trauma by inducing acute phase proteins and increasing the adhesiveness of leukocytes on vascular endothelial surfaces.(18) It is a primary mediator of immune regulation, an important component of almost all inflammatory responses, and is produced by a wide variety of immune and nonimmune cells. TNF-α has been studied extensively in bone and has been implicated in the regulation of osteoclastogenesis.(19, 20) Levels of TNF-α production and IL-1 expression in monocytes have been shown to be elevated in surgically produced or naturally occurring menopausal states, and a number of studies have shown that TNF-α can induce osteoclast formation in vitro. IL-1 also has been shown to participate in osteoblast proliferation, differentiation, and collagen synthesis in cooperation with other factors such as vitamin D3 or parathyroid hormone.(21) However, some controversy exists as to whether this is a direct effect of IL-1 or if it is mediated through the actions of TNF-α on other cells in the marrow stromal environment. It remains unclear if TNF binding protein will block TNF-α stimulation of osteoclastogenesis.(17) Recently, a novel member of the TNF receptor family, osteoprotegerin (OPG), and its specific ligand receptor activator of NK-κB ligand (RANKL), in conjunction with macrophage colony-stimulating factor (M-CSF), have been shown to be key regulators in the control of bone mass through their modulation of the bone resorptive cycle.(22–24) OPG is a secreted soluble TNF receptor family member that binds to RANKL and prevents it from stimulating osteoclastogenesis.(25, 26)

Although the role of these proinflammatory cytokines in bone remodeling is well established,(12, 15-17, 20–26) their participation in the response of skeletal tissue to injury is not well understood. Previous studies on rat fracture calluses have established IL-1 and IL-6 activity by functional assay.(4) IL-1 was shown to be expressed at very low constitutive levels throughout the period of fracture healing but could be induced by lipopolysaccharide (LPS) to high levels of activity in the early inflammatory stage (day 3) after fracture. In contrast, IL-6 was found to be maximally expressed in the early inflammatory stage but required either LPS or concanavalin A induction to be detected on day 7 or day 14 fracture callous. Together, with more recent data, these findings collectively suggest that proinflammatory cytokines may carry out important regulatory functions during fracture repair. To investigate the potential roles for these cytokines in fracture healing, as well as the more recently identified TNF family members noted previously, the temporal and spatial patterns of the expression of three cytokine groups, IL-1 (-α and -β) and their receptors, TNF-α, Lt-β, and their receptors, and OPG, RANKL, and M-CSF were analyzed.

MATERIALS AND METHODS

Materials

Histological reagents and chemicals were purchased from either Sigma Chemical Co. (St. Louis, MO, USA) or Fisher Scientific (Springfield, NJ, USA). Super frost slides and immunostaining trays were obtained from Shandon Lipshaw (Pittsburgh, PA, USA). Secondary antibodies and immunohistochemical staining reagents were obtained from Vector Laboratories (Burlingame, CA, USA). Tyramide signal amplification kits came from NEN Life Sciences (Boston, MA, USA). Antibodies to mouse IL-1α and TNF-α came from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and PharMingen Corp. (San Diego, CA, USA), respectively.

Production of simple transverse fractures

Eight to 10-week-old male BALB/c mice were used for this study. Closed, transverse, middiaphyseal fractures of the tibias were generated by controlled blunt trauma using a modification of the technique developed for rats.(27, 28) The tibia was chosen for use in the mouse versus the femur in the rat, because for this species it provides better anatomic features than the femur for the performance of the fracture procedure.(27) Fracture stabilization by intramedullary fixation in the mouse was carried out using the trochar of a 25G or 23G spinal needle. The animals were permitted full weight bearing and unrestricted activity after awakening from anesthesia. Mice were killed by cervical dislocation on days 1, 3, 7, 14, 21, and 28, postoperatively.

Histology

At the time of euthanasia, the fracture site was recovered and fixed for 3 days in 4% paraformaldehyde in phosphate-buffered saline (PBS) at 4°C. Specimens were completely decalcified in Immunocal (Decal Chemical Corp., Congers, NY, USA) and embedded in paraffin. The tissue was positioned appropriately in paraffin to make sagittal sections and cut at a thickness of 6 μm. Sections were stained with hematoxylin and eosin (H&E) alcian blue or assayed for tartrate-resistant acid phosphatase (TRAP) activity.(29–31) Quantitative histological analyses were carried out on three to five H&E-stained specimens from each of five animals per data point. The microscopic images were measured using Image ProPlus software (Media Cybernetics, Silver Spring, MD, USA). The total area of the callus was measured at ×4 magnification. Measurements of cartilage, mixed trabeculae of cartilage and bone, bone, and marrow space within the callus were carried out at ×10 magnification. In all cases, each tissue also was examined at higher magnification (×30) to ensure that it was identified properly.

Immunohistochemical analysis

Immunostaining was carried out with antibodies against mouse TNF-α, IL-1, and IL-6. Immunoperoxidase staining was performed using an avidin-biotin-peroxidase method and enhanced with tyramide. Reactions were carried out with appropriate secondary and tertiary antibodies using the Vector Stain Elite Immunostaining Kit and a Tyramide Signal Amplification Kit, following the manufacturer's protocols. Uniformity of reactions was maintained by carrying out all reactions in standardized buffer conditions and volumes in immunohistology staining trays. All reactions also were carried out with controls in which the primary antibody was replaced with nonspecific antisera to validate antibody specificity.

RNA analysis

RNA was isolated and quantified in fracture calluses and control tissues in duplicate sets of pooled samples (n = 10 [day 1 and day 3] and n = 5 [days 7, 14, 21, and 28]). Bones were retrieved on days 1, 3, 7, 14, 21, and 28 after fracture and powdered under liquid nitrogen with a mortar and pestle. The total RNA was extracted from the powdered tissues with Tri-Reagent TM (Molecular Research Center, Inc., Cincinnati, OH, USA) as previously described.(32) RNA quantities were determined by optical density at 260 nm (OD260) and sample integrity was monitored by visualization of ribosomal RNAs with ethidium bromide after denaturing RNA gel electrophoresis.(33)

RNAse protection analysis

Messenger RNA (mRNA) expression during fracture healing was assessed quantitatively by ribonuclease protection analysis (RPA). Linearized plasmids for mouse collagen types I, II, and X; osteocalcin; TNF-α; lymphotoxin-α (LT-α; TNF-β); LT-β; TNF receptors (p55 and p75); IL-1α, IL-1β, and IL-1 receptors (IL-1RI and IL-1RII); OPG; RANKL; M-CSF; L32; and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were purchased from PharMingen Corp. (San Diego, CA, USA). Single-stranded [32P]-labeled complimentary RNA (cRNA) probes were generated using RNA transcription kits purchased from PharMingen Corp. as per the manufacturer's instructions. RNAse protection products were fractionated on a denaturing 6% acrylamide gel.(34) The autoradiographic bands of the RNAse protection products were quantified using an Alpha Innotech Image Analysis System (Alpha Innotech Inc., San Heandro, CA, USA) Underexposed autoradiographic images were used for all analyses and the autoradiographic densities of the protected bands from each probe were measured. The quantities of individual mRNAs for specific cytokines and extracellular matrix proteins were determined in the following manner. It is assumed that each probe had the same specific activity (i.e., the same ratio of radiolabeled nucleotides are incorporated for each runoff product). The second assumption was that the annealing efficiencies between all the probes and their target mRNA sequences were the same under the conditions used in these studies. All data were normalized to a so-called housekeeping gene such as GAPDH or L32 mRNA. Data were expressed as either a percentage of the highest level seen at single time point across the time course for a given mRNA or as a value in comparison to the housekeeping mRNA species and corrected for variations in probe length (Table 1).

Table Table 1.. Comparison of Cytokine mRNA Levelsa
original image

RESULTS

Histomorphological analysis of fracture healing

Calluses were examined to establish the temporal and spatial profiles of the endochondral and remodeling phases of bone repair. Representative histological sections are presented of the fracture sites on days 1, 3, 7, 14, 21, and 28 after fracture (Fig. 1). Immediately after fracture, only slight swelling in the subperiosteal areas of the fractures and development of a hematoma were observed (day 1 and day 3). However, extensive callus formation was evident by day 7, and there was initiation of both chondrogenesis near the interface between the soft (cartilaginous) and hard (bony) callus and direct outgrowth of new bone from the periosteal surface. The initiation of chondrogenesis was particularly evident when the acidic Alcian blue-stained sections were compared with H&E-stained sections (Fig. 1B). By day 7, cartilage formation was maximally induced and small areas of hypertrophic chondrocyte development were evident in the subperiosteal areas of the callus tissue immediately adjacent to the fracture site. By day 14, the cartilaginous callus had become ossified and extensive tissue resorption was apparent. This was accompanied by the initiation of extensive bone replacement in these areas. At day 21 and day 28, the cartilage had largely been resorbed and replaced with new trabecular bone and bone marrow (Fig. 1A).

Figure FIG. 1..

Temporal histological analysis of fracture repair on days 1, 3, 7, 14, 21, and 28. (A) Panoramic view of histological features in standard transverse fractures (H&E, ×40). (B) Comparison of standard H&E and acidic Alcian blue-stained sections on day 7 and day 14 postfracture to define specifically the extent of cartilage formation. Note the intense blue staining of the cartilage on day 7 and day 14 and the extent of cartilage resorption on day 14 (×100).

To develop quantitative temporal relationships between the expression of the various cytokines and the overall changes in the cellular composition of the fracture calluses, five parameters were assessed within the callus: (a) the total area of the callus, (b) the marrow space within the new bone of the callus, (c) the area of cartilage, (d) the area of mixed bone and cartilage, and (e) the area of new bone (Figs. 2A and 2B). The tissue mass within the calluses increased rapidly between day 7 and 14, whereupon it reached a maximum size. Thereafter, the calluses gradually became smaller. On day 7, the predominant tissue present was cartilage. However, small amounts of new endochondral bone were already present in the form of mixed trabeculae. On day 14, the area that was strictly encompassed by cartilage decreased, while the area that contained mixed trabeculae of cartilage and bone increased rapidly so that it represented the predominant type of tissue present. By day 21, the cartilage and mixed trabeculae were replaced almost entirely by bone and marrow spaces within callus tissue. A similar histological picture was obtained on day 28 with continued shrinkage in the callus size.

Figure FIG. 2..

Quantitative histomorphological analysis of fracture tissues. (A) Total size of the fracture calluses size and marrow space on days 7, 14, 21, and 28 after fracture. (B) Tissue composition on days 7, 14, 21, and 28. Figure keys are depicted in the graphs. Error bars denote one SD from the mean.

One of the most striking changes to occur over the last 2 weeks of the experimental period was the formation of the marrow space and the high levels of resorption and remodeling that were observed in the callus tissue. However, the development of marrow was not observed until day 14 between the second and third week of the experimental period, a 2- to 3-fold increase in the area of marrow space was seen (Fig. 2A). Subsequently, as the original cortical structure of the bone began to be reestablished, the size of the marrow space slowly shrank during the last week of the experimental period. The development of this space may be correlated to increased remodeling within the callus tissue (Fig. 3), based on the numbers of TRAP-staining multinucleated cells present in the tissue. Initially, multinucleated TRAP-positive (TRAP+) cells were seen in areas of both calcified cartilage and mixed bone and cartilage trabeculae at day 14 (Fig. 3, top panel). Trabecula of newly woven bone was the primary tissue seen in the calluses by day 21. Large numbers of TRAP+ multinucleated cells, having typical morphology of active osteoclasts as characterized by the many resorption pits (Fig. 3, bottom panel), also were seen on the surfaces of this callus bone tissue. These data show the extensive remodeling that this tissue continued to undergo during the later stages of bone repair.

Figure FIG. 3..

Analysis of TRAP+ cells in fracture calluses. Top panel shows a representative area of cartilage bone and mixed trabeculae seen on day 14 after fracture in the callus tissues. The bottom panel depicts the callus tissue on day 21. The left panels depict ×200 magnification of a typical bone area while the right panels depicts either H300 or ×400 magnification. Arrows indicate typical TRAP+ cells.

Analysis of bone, cartilage, cytokine, and cytokine receptor mRNA expression during fracture healing

RPA was carried out to assess the expression of various genes that define early and late stages of chondrogenic and osteogenic differentiation (Fig. 4). Chondrogenic differentiation was initiated sometime after day 3 and occurred in a tight window of time between day 7 and 14 with maximal levels of type II collagen mRNA expression observed at day 7. Between day 7 and 14 there was progressive differentiation of the chondrocytes to form hypertrophic cells, as shown by the increased ratio of type X collagen (1:3-1:1) relative to the expression of type II collagen mRNA. Osteogenesis was initiated as early as day 3 after fracture, as evidenced by elevated levels of type I collagen mRNA expression in comparison to that observed in unfractured bones. Maximal levels of type I collagen expression were not reached until day 14, and these levels were maintained throughout the remainder of the time period of fracture healing while the peak in osteocalcin mRNA expression occurred at 21 days.

Figure FIG. 4..

Quantitative analysis of mRNA expression for selected bone and cartilage matrix genes. The upper panels present representative autoradiographic images of the RPA products for collagen types I, II, and X and osteocalcin, as resolved on an 8% polyacrylamide gel electrophoresis (PAGE) sequence gel. Days after fracture are denoted. The lower panels present the graphic analysis of the relative mRNA levels. Band densities were determined from several different radiographic exposures and taken from the linear range of the film exposure. Band densities where normalized to the internal standard L32 and expressed as a relative value to the greatest band intensity (100%) over the time course.

The second group of mRNAs examined consisted of the primary regulatory cytokines that control osteoclast differentiation and activity (M-CSF, OPG, and RANKL [OPGL]). These results are depicted in Fig. 5. OPG, RANKL, and M-CSF were all maximally induced within 24 h after fracture. By day 3, the levels of OPG had decreased rapidly, while RANKL and M-CSF levels remained elevated. At the time when cartilage formation was highest (day 7), the levels of M-CSF and RANKL had decreased, while the levels of OPG had again increased to very high levels. Subsequently, during the period when endochondral tissue began to be actively resorbed and the number of osteoclasts in the tissue was increasing (Fig. 3B), the levels of both M-CSF and RANKL showed striking increases in their mRNA expression. Concurrent with the increased mRNA expression of RANKL and M-CSF was the relative decreased expression of OPG. As resorption of calcified cartilage progressed and reached its completion, the levels of M-CSF and RANKL fell and the relative ratio of OPG increased again.

Figure FIG. 5..

Quantitative analysis of mRNA expression for OPG, RANKL, and M-CSF. The left panels present representative autoradiographic images of the RPA products as resolved on an 8% polyacrylamide gel electrophoresis (PAGE) sequence gel. Days after fracture are denoted. Right panels present the graphic analysis of the relative mRNA levels. Band densities were determined from several different autoradiographic exposures and taken from the linear range of the film exposure. Band densities where normalized to the ratio of the internal standard GAPDH and expressed as a relative value to the greatest band intensity (100%) over the time course.

The expression of IL-1, TNF-α, and their receptors also were examined over this 28-day period of fracture healing. The profiles of several of the members of the TNF family (TNF-α, TNF-β, and LT-β) and TNF-α receptors (p55 and p75) are presented in Fig. 6. Both TNF-α and LT-β showed biphasic expression during fracture healing, first being elevated on the day 24 and decreasing to baseline levels between day 3 and 14. Thereafter, their levels increased for the remainder of the experimental period. TNF-β mRNA expression was not detectable in either control or callus tissues, although it was easily detected in control mRNAs isolated from activated T cells (data not shown). However, at very long periods of autoradiographic exposure TNF-β mRNA expression was seen in the fracture calluses at 14 days and 21 days (data not shown). The TNF-α p55 receptor was expressed at very high levels in unfractured bones, and both receptors were induced in response to fracture (Fig. 6). The mRNA expression of the receptors showed a similar and parallel biphasic pattern in comparison to TNF-α. The mRNAs for the p75 receptor were expressed at much lower levels in unfractured bones and showed a much greater induction of expression in response to fracture.

Figure FIG. 6..

Quantitative analysis of mRNA Expression for TNF-α, LT-β, TNF-α receptor p55, and TNF-a receptor p75. The left panels show representative autoradiographic images of the RPA products as resolved on an 8% polyacrylamide gel electrophoresis (PAGE) sequence gel. Days after fracture are denoted. Right panels show the graphic analysis of the relative mRNA levels. Band densities were determined from several different radiographic exposures taken to obtain the linear range of the film exposure. Band densities where normalized to the ratio of the internal standard GAPDH and expressed as a relative value to the greatest band intensity (100%) over the time course.

The expression of IL-1 and its receptors is shown in Fig. 7. The expression of both IL-1 cytokines parallels that of TNF-α (Fig. 6). However, the late peak of expression subsequently shown by these cytokines is less robust than that of TNF-α. It also is interesting to note that the expression of the IL-1RI receptor appears to be restricted to the early inflammatory phase after injury, while that of the soluble IL-1RII form shows a second peak of expression at later time points.

Figure FIG. 7..

Quantitative analysis of mRNA expression for IL-1α, IL-1β, IL-1 receptor I, and IL-1 receptor II. The left panels present representative autoradiographic images of the RPA products as resolved on an 8% polyacrylamide gel electrophoresis (PAGE) sequence gel. Days after fracture are denoted. Right panels present the graphic analysis of the relative mRNA levels. Band densities were determined from several different autoradiographic exposures taken to obtain the linear range of the film exposure. Band densities where normalized to the ratio of the internal standard L32 and expressed as a relative value to the greatest band intensity (100%) over the time course.

The data presented in Figs. 5, 6, and 7 provide only a relative analysis of the temporal expression of the various cytokines over the time course of fracture healing. Two other aspects of these data that are informative about the role of these cytokines, in both noninjured and fractured bones, are the total-fold induction of each of these cytokines in response to fracture and the relative levels of the cytokine mRNAs to each other. These data are summarized in Table 1. As can be seen from these data, LT-β and IL-1β mRNAs are expressed at the highest levels in unfractured bones while RANKL is the least prevalent of the mRNAs examined. In terms of fold induction, RANKL shows the strongest induction followed by M-CSF and OPG, while TNF-α and IL-1α show the weakest levels of overall induction.

Immunohistological analysis of proinflammatory cytokine expression

The final goal of these studies was to examine the spatial localization of the cells that were expressing the major proinflammatory cytokines during the fracture healing process. Representative micrographs of the immunohistological localization are presented for TNF-α (Fig. 8). A summary of the cell types and the relative levels of expression that were seen in these analyses are presented in Table 2. At the initial time of fracture, the highest levels of TNF-α expression were seen in the marrow space and within the periosteum nearest to the fracture. This localization persisted throughout the fracture repair and is seen in both low and higher magnifications of the periosteum and marrow space at both 14 days and 21 days (Fig. 8). Immunoreactivity for IL-1α was much lower and appeared to be more restricted to the periosteum. Strong reactions were still seen in the marrow space for TNF-α antibody at day 21. The general intensity of the reactions was consistent with the overall differences seen in the levels of expression for these mRNAs. Higher magnification micrographs are presented depicting the types of cells that were expressing TNF-α within the fracture callus tissues on day 14 and day 21. Nondifferentiated mesenchymal cells expressed these cytokines in the periosteum. On day 14, hypertrophic chondrocytes within the callus tissues were distinctly reactive, and the mesenchymal cells invading the endochondral tissue were positively stained for TNF-α, although the overall expression of the TNF-α mRNA was very low. On day 21, the lining cells on the new bone surfaces also were reactive, and a strong reactivity was seen within the osteoblasts in the new trabecular bone. General reactivity also was seen in the hematopoietic cells throughout the forming marrow cavity and in particular in megakaryocytes. Qualitatively, these studies show very comparable results to the relative levels of mRNA expression seen for these cytokines over the time course of the repair process as is summarized in Table 2.

Table Table 2.. Cytokine Production as Determined by Immunocytochemistry
original image
Figure FIG. 8..

Immunohistological analysis of TNF-α expression during fracture repair. Upper panel depicts TNF-α localization on day 14 (×100). Left panel shows localization in the periosteum and right panel depicts specificity in hypertrophic chondrocytes. Middle panels depict ×400 magnification of matched fields to the top panel. Bottom panel depicts micrographs of the specific cells that are expressing TNF-α on day 21 within areas of trabecular bone and the marrow space. Immunohistological staining is brown. Specific staining is indicated with arrows.

DISCUSSION

Fracture healing is a unique response to injury in which bone regeneration involves processes associated with normal bone development, including not only cartilage and bone formation, but also endochondral resorption and bone remodeling. The results presented here confirm previous reports showing that cartilage development occurs in a discrete window of time during the first 2 weeks of fracture healing.(1, 10 11 35) In contrast, osteogenesis appears to be induced almost immediately after injury as evidenced by the strong induction of type I collagen mRNA expression on day 3, followed by that of osteocalcin on day 7. Thereafter, the expression of both mRNAs continues to increase showing elevated levels throughout the healing period. The increasing levels of type I collagen and osteocalcin mRNAs during the early time points (days 3-14) most likely is associated with the intramembranous bone formation that is occurring in the periosteum. However, at the later times, the majority of new bone formation occurs as part of the endochondral process. It is interesting to note, when comparing the expression of mRNAs for type I collagen and osteocalcin, that at early time points osteocalcin is induced 4 days later than that of type I collagen (day 7 vs. day 3) and reaches peak in expression a week later than type I collagen (day 21 vs. day 14). These results suggest that, once initiated, terminal osteogenic cell differentiation takes approximately 4-7 days to complete. This in vivo time frame is consistent with those that have been observed during in vitro studies of osteogenic differentiation.(32, 36, 37)

Proinflammatory cytokines known to regulate immune function and inflammation have long been known to be involved in bone remodeling. Now, they are shown to be involved in fracture healing. A previous report from one of our laboratories showed that IL-1 is expressed at very low constitutive levels throughout the period of fracture healing but is able to be induced by LPS treatment to very high activities in the early inflammatory phase (day 3). IL-6 showed high constitutive activities early in fracture healing but required induction as healing proceeded.(4) These results in a rat model appear to follow those that have been observed in mice, suggesting that there is considerable cross-species conservation in the initial inflammatory response to injury. In this report, the strong induction of TNF-α and IL-1 that occurred within the first 24 h after skeletal injury was accompanied by a large influx of immune cells to the injury site. This is very similar to the inflammatory response observed in soft tissue wounds. It is interesting to note that RANKL levels peaked and were maintained slightly later than the immediate early inflammatory burst. Such findings suggest that both IL-1 and TNF-α may be involved in the induction of its production; however, it is unclear if RANKL is produced only by activated immune cells or if its production also is stimulated within the local populations of osteoblasts and stromal cells at the fracture site.(12–17) Certainly, it is intriguing to speculate that the temporal induction of resorption of both the cartilage and the bone may be part of an overall cascade of events that initially is triggered, and to some degree coordinated, by the network of cytokines that is induced during the early inflammatory phase.

The temporal profiles among several proinflammatory cytokines differed with respect to their relative levels of mRNA expression in unfractured bone compared with their expression in response to injury. LT-β, IL-1α, and the p55 TNF-α receptor were constitutively expressed in unfractured bone. The very strong induction of both IL-1 receptors, IL−1β, and TNF-α after fracture suggests that these cytokines are produced by either new subsets of cells recruited to the site of injury or cells resident in the tissue and up-regulated by the injury. In both types of responses, the immunohistological findings show that IL-1α and TNF-α are synthesized not only by macrophages and inflammatory cells recruited to the site of the injury, but also by cells of mesenchymal origin present in the periosteum. TNF-α also appears to be synthesized by hypertrophic chondrocytes. At later times, both IL-1α and TNF-α are synthesized by lining cells on the newly formed trabecular bone surfaces. The diversity of cells that express TNF-α suggests that this cytokine may carry out multiple functions at different times during fracture healing. TNF-α may potentially regulate the initiation of fracture healing including mesenchymal cell proliferation and differentiation in the periosteum. The increasing levels of expression of the ligands and receptors for both IL-1 and TNF-α on day 21 and day 28 after fracture suggests that these cytokines also are most likely active in promoting bone remodeling at later times during bone healing. Finally, the identification of TNF-α expression in hypertrophic chondrocytes provides strong support for its potential role as an autocrine regulator of chondrocyte maturation or induction of apoptosis.(38)

Two features of the data pertaining to the expression of TNF-α are particularly important to note: (1) TNF-β (LT-α) is never expressed at detectable levels during fracture healing or in unfractured bones. As LT-β normally heterotrimerizes with TNF-β (LT-α)(18) and is not functional as a homodimer. Its presence at these times suggests there is a different, as yet unknown, TNF-family member that heterotrimerizes with LT-β, which makes its activity functional. (2) The p75 receptor is not constitutively seen in bone and is induced by fracture, suggesting that its appearance may correspond to a new subset of cells within the callus tissue.

OPG, RANKL, and M-CSF have been shown to be important regulators of osteoclastogenesis during bone remodeling(22–24); however, their role in endochondral ossification, specifically the resorption of calcified cartilage, has not been established. The data presented here suggest for the first time, that OPG, RANKL, and M-CSF function during endochondral resorption in fracture healing in a similar manner to that which is observed during bone remodeling.(25, 26) In this context, it is interesting to note that OPG is expressed in unfractured bones at moderately high levels while RANKL is not. This is consistent with other studies that have shown that the expression of these two molecules is not necessarily coupled.(25) OPG showed a sharp peak in expression during the phase of maximal cartilage formation and then declined as endochondral resorption progressed. This is consistent with its role as a negative regulator of cells involved in calcified tissue resorption. In contrast, RANKL was expressed at very low levels in unfractured bones but showed a dramatic up-regulation immediately after fracture in conjunction with strongly induced levels of expression of M-CSF. After their induction, both of these cytokines remain elevated throughout the entire healing period. These findings suggest that the expression of these cytokines is tightly coordinated in their regulation of endochondral resorption.

Finally, it is important to note that although both TNF-α and IL-1 have been shown to play important roles in bone remodeling, our results showing low levels of expression during the period of endochondral resorption suggests that they must have a different function in this process. In contrast, their rapidly increasing levels of expression from day 21 to 28 after fracture clearly suggest an important role in bone remodeling of woven bone. Such differences in function may reflect varying mechanisms of bone remodeling and repair between the processes that occur in the periosteal and endosteal environments and in endochondral resorption.

Acknowledgements

The authors acknowledge Christopher George and Dr. Bohus Svagr for their technical assistance in the sectioning and immunohistochemistry of the fracture specimens and Irene Simkina for her assistance with the development of the conditions for RPA analysis of the bone matrix genes. The authors also acknowledge Dr. George Barnes for his technical advice and discussions during the course of this work. This study was supported in part by the National Institutes of Child Health and Human Development (NICHD) grant HD22400 (L.C.G.). Institutional support was provided by the Boston University School of Medicine (T.A.E.). All research was done in conformity with all Federal and U.S. Department of Agriculture (USDA) guidelines and with Institutional Animal Care and Use Committee (IACUC)-approved protocol.

Ancillary