Previously, we have reported that nitric oxide synthases (NOSs), which generate NO, modulate fracture healing. However, the cellular sources of the NOS isoforms during the course of fracture healing have not been studied systematically. The purpose of this study was to localize the cellular distribution of NOS isoforms (inducible NOS [iNOS], endothelial NOS [eNOS], and neuronal NOS [bNOS]) by in situ hybridization and immunohistology after femoral fractures in rats. The iNOS signal was detected during the initial stages (on day 4 and day 7) of fracture healing in 52 ± 2% (mean ± SE, n = 7) of cells within the intramembranous region, along the edge of the periosteal callus. The iNOS signal in callus cells declined to an undetectable level on day 14. eNOS was detected during the middle stages (on day 7 and day 14) of fracture healing in cells lining the blood vessels and also in 49 ± 3% of cells in the chondral region. The bNOS signal was found to be increased at the later stages (day 14 and day 21) of fracture healing in 51 ± 3% of cells at the junction between fibrous tissue and cartilage within the fibrochondral region. In summary, the expression of NOS isoforms during fracture healing was time dependent and cellular distinctive.
FRACTURE HEALING is a sequential and coordinated physiological process, which is regulated by numerous local and systemic molecules. Nitric oxide (NO), a small uncharged diffusible gas, has a modulatory role in bone repair.(1)
NO synthases (NOSs) are a family of enzymes that generate NO from L-arginine in a reaction yielding citrulline as a coproduct.(2) Endothelial NOS (eNOS) and neuronal NOS (bNOS) activities are calcium dependent and naturally constitutive.(3,4) The inducible iNOS, which generally is seen in inflammatory conditions, is high output and calcium independent.(5,6)
Fracture healing occurs through a complex cellular cascade. After a fracture, hematoma is formed around the fracture ends. Mesenchymal cells stimulated by growth factors migrate and differentiate to osteoblasts.(7,8) Bone formation then occurs by intramembranous and endochondral ossification. During intramembranous bone formation, new bone matrix forms adjacent to the periosteum and becomes mineralized. In endochondral ossification, bone forms via a step involving an interposition of chondrocytes during the healing process in a manner similar to bone formation in the growth plate, where the matrix around the hypertrophic chondrocytes is mineralized.(9) Subsequently, during the process of remodeling, the mineralized matrix is replaced by lamellar bone.(10)
We have reported that all three NOS isoforms were induced during rat and human fracture healing,(1) and showed a temporal expression of NOS in rat fracture healing using the techniques of NOS activity assay, competitive polymerase chain reaction (PCR), and Western blot.(1,11) In addition, we found in a rat femoral fracture model that inhibition of NOS inhibited the healing process and addition of NO to the fracture site via a donor reversed this impairment.(1)
Although there are several reports about expression of NOS isoforms in cultures of osteoblasts,(12,13) osteoclasts,(14,15) and whole bone sections,(16,17) little is known about the cellular localization of NOS isoforms in bone-healing models. Hence, the purpose of this study was to test the hypothesis that different NOS isoforms may have distinctive spatial and temporal cellular expression patterns in the callus of healing rat femur. This was studied by in situ hybridization and immunohistology at different stages of rat fracture healing.
MATERIALS AND METHODS
Mouse anti-rat bNOS and eNOS monoclonal antibodies were purchased from Transduction Laboratory (Lexington, KY, USA). Rabbit anti-rat inducible NOS (iNOS) polyclonal antibody was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA). All reagents for immunohistology study were obtained from DAKO, Inc. (Carpinteria, CA, USA). Other chemicals were purchased from Sigma Chemical (St. Louis, MO, USA).
Animal experiments and design
The Animal Care and Ethics Committee of the University of New South Wales approved all animal experiments. Twenty-eight male Sprague-Dawley rats (Biological Resources Center, Sydney, NSW, Australia) with a mean weight of 320 g had an open midshaft right femoral fracture created. The right femur was fractured as described previously(1) and the intact left femur was used as an internal control. Anesthesia was achieved by intraperitoneal (ip) injection of 300 μg/kg of fentanyl and 5 mg/kg of midazolam. Postoperative pain relief was obtained by preoperative injection of 2 ml of 0.5% Marcaine (Astra Pharmaceuticals, North Ryde, NSW, Australia) around the fracture site and an intramuscular injection of 0.05% mg/kg of buprenorphine (Reckitt and Coleman, West Ryde, NSW, Australia) at the end of the operation.
All rats were killed by carbon dioxide inhalation. Healing callus were harvested on 4, 7, 14, and 21 days after fracture, n = 7 for each time point.
The femoral fracture callus was fixed in 4% (wt/vol) paraformaldehyde in PBS (pH 7.6) for 6 h followed by 2 weeks of decalcification in 0.34 M of EDTA solution containing 0.31 M of sodium hydroxide. Specimens were cut in a sagittal plane and paraffin-embedded. Sections from the tissue blocks were cut at a thickness of 4-6 μm and mounted on silane-coated slides (Lancaster Synthesis, Eastgate, White Land, Morecambe, UK) and dehydrated. The slides were deparaffinized with xylene (8 minutes) and rehydrated in three sequential ethanol baths from 100% to70% ethanol (each for 2 minutes) before immunohistology and in situ hybridization studies.
In situ hybridization
Specific oligonucleotide cRNA probes and probe labeling
Both rat iNOS and eNOS cDNA clones were generated by reverse transcription PCR from 1 μg of rat RNA and used pGEMT-Easy plasmid (Promega, Madison, WI, USA) as vector. The amplified iNOS cDNA fragment was complementary to nucleotides 3191-3584 of the published sequence (GeneBank accession no. U03699). The amplified eNOS cDNA fragment corresponded to nucleotides 1-691 of the published sequence (GeneBank accession no. RNU 02534). The rat bNOS clone was made from a 544-bp EcoRI-KpnI fragment of full-length rat bNOS cDNA clone (a kind gift from Dr. A.M. Snowman of Johns Hopkins University, Baltimore, MD, USA) and subcloned into the Bluescript II KS+ (Stratagene, La Jolla, CA, USA) vector. Sense and antisense cRNA probes were synthesized by in vitro transcription with the relevant RNA polymerase and labeled with digoxigenin-uridine triphosphate (DIG-UTP; Boehringer Mannheim, Mannheim, Germany). Each labeled probe was purified by ethanol precipitation. The degree of labeling was estimated by spot test on a nylon membrane with a digoxigenin-labeled control (Boehringer Mannheim).
In situ hybridization
After deparaffinizing, tissue sections were placed in PBS and then deproteinized with 10 μg/ml of proteinase K (Boehringer Mannheim) for 10 minutes at 37°C. Sections were postfixed with 4% (wt/vol) paraformaldehyde in PBS (pH 7.4) for 5 minutes and acetylated by 0.25% (vol/vol) acetic anhydrate containing 0.1 M of triethanolamine for 10 minutes at room temperature. The sections then were prehybridized for 30 minutes at 55°C and then hybridized with 5 ng/μl of digoxigenin-UTP-labeled sense or antisense probes for 12 h at 55°C. After washing with serial SSC (3 M of sodium chloride solution containing 0.3 M of trisodium citrate, pH 7.0) baths from 2× to 0.1×, the immunologic detection of digoxigenin-labeled transcripts was performed according to the manufacturer's protocol (Boehringer Mannheim) using alkaline phosphate-conjugated antidigoxigenin polyclonal serum. Finally, the sections were counterstained with 1% (wt/vol) neutral red and mounted. Cells positive for NOS mRNA stained a red or purple color.
Immunohistology was performed by the biotin/streptavidin method. After deparaffinizing using xylene, the sections underwent antigen retrieval by 3 minutes microwave at 75% power then 2 minutes microwave at 25% power and then were blocked by 3% (vol/vol) hydrogen peroxide in methanol for 15 minutes, followed by 10% (wt/vol) skim milk in Tris-buffered saline (TBS) solution (0.15 M of sodium chloride and 0.05 M of Tris, pH 7.6). Then, sections were incubated with anti-iNOS antibody (1:250 dilution), anti-eNOS antibody (1:400 dilution) or anti-bNOS antibody (1:300 dilution) for 1 h at room temperature. After washing in TBS, the sections were incubated with secondary antibody (biotinylated anti-mouse immunoglobulins in PBS) followed with streptavidin peroxidase and visualized using diaminobenzidine (DAB) as the chromogen. Next, slides were counterstained with Mayer's hematoxylin and mounted. Negative control sections without primary antibodies were used in each run. Cells positive for NOS protein presence stained a red-brown color.
Image analysis of cell counting
We used serial sections from the same tissue blocks to NOS isoform mRNA and protein. The sagittal sections of fracture callus were studied under a Leica microscope (Leica Microkopie and Systems, Wetzlar, Germany) and imaged by Optimas Image Analysis Computer System (Optimas Corp., East Victoria Park, WA, Australia) using 14 μm2 × 12 μm2 quadrants at 100× magnification in a blinded fashion. Four quadrants of each callus sample were studied wherever possible for three regions of interest. The pooled average percentage of positively stained cells was taken as the reading for the sample. The three regions were (Fig. 1) (1) the intramembranous bone formation—the region of direct bone formation at the periosteal surfaces of bone, involving cells residing in the osteogenic layer of periosteum and osteoblasts, which line the new bone surface(18,19); (2) the chondral region—the region of indirect bone formation with a cartilaginous framework, involving osteoblasts and chondrocytes in lacunae, which get larger and rounder until they degenerate and the matrix calcifies(20,21); and (3) the fibrochondral region—the region at the junction of cartilage and fibrous tissue area involving chondrocytes and fibroblasts.(19,22) Finally, the percentage of positively stained cells for composite callus was a mean value of the three regions.
The cellular expression of each NOS isoform was compared during the time course. All data are presented as mean ± SEM. Differences among experimental groups were assessed by ANOVA and unpaired Student's t-tests. Because unfractured bone does not have any callus, the percentage of cells positive for NOS isoforms in each region was compared with data from day 4. The level of statistical significance was accepted at p < 0.05.
Descriptive localization of NOS in fracture callus
No NOS isoform mRNA or protein was detected in unfractured left femurs. In situ hybridization showed that iNOS mRNA signal was present in cells along the edge of the periosteal callus and close to the cortical bone (Fig. 2, IA). iNOS mRNA was found also in the area of endochondral ossification (Fig. 2, IIA). Most iNOS mRNA-positive cells were osteoblast-like cells residing in the periosteum and cells in the chondral region. Endothelial cells faintly stained for iNOS. Osteocytes showed little or no staining for iNOS. There was no signal detected in negative controls (iNOS sense probe) for iNOS mRNA in the periosteal callus (Fig. 2, Ia) and the chondral region (Fig. 2, IIa).
A similar cellular pattern of iNOS protein as that of iNOS mRNA was found by immunohistology in the intramembranous region (Fig. 2, IB) and endochondral ossification (Fig. 2, IIB) with prominent staining in osteoblast-like cells along the edge of the periosteal callus and close to the cortical bone. There was no signal detected in negative controls (without anti-iNOS antibody) for iNOS protein in the intramembranous region (Fig. 2, Ib) and chondral region (Fig. 2, IIb).
In situ hybridization for eNOS mRNA on serial sections showed that eNOS mRNA was located in endothelial cells of blood vessels (Fig. 2, IIIA), in the intramembranous region (Table 1), and in chondroid cells in the cartilaginous areas of the endochondral ossification region (Fig. 2, IVA). Osteocytes showed weak staining for eNOS. There was no signal evident in negative controls (eNOS sense probe) for eNOS mRNA in blood vessels (Fig. 2, IIIa) and in the chondral region (Fig. 2, IVa).
Table Table 1.. Percentage of Positively Stained Cells for NOS Isoform mRNA and Protein in the Intramembranous Region of Fracture Callus
The immunostaining study showed a similar cellular pattern of eNOS protein as that of eNOS mRNA with staining in blood vessels (Fig. 2, IIIB), the intramembranous region (Table 1) and cartilaginous areas (Fig. 2, IVB). There was no positive signal in negative controls (without anti-eNOS antibody) for eNOS protein in blood vessels (Fig. 2, IIIb) and the chondral region (Fig. 2, IVb).
The bNOS mRNA signal was found mainly at the junction of fibrous tissue and cartilaginous areas during endochondral bone formation (Fig. 2, VA) and also in the intramembranous region (Table 1). bNOS-positive cells were osteoblast-like cells, large round chondroid cells, spindle-shaped mesenchymal-like cells, and fibroblast-like cells. Osteocytes were rarely positive for bNOS. There was no positive signal in negative controls (bNOS sense probe) for bNOS mRNA in the fibrochondral region (Fig. 2, Va).
The immunostaining showed a similar cellular pattern of bNOS protein with that of bNOS mRNA within the intramembranous region (Table 1) and fibrochondral region (Fig. 2, VB). There was no positive signal found in any of the negative controls (without anti-bNOS antibody) for bNOS protein in the fibrochondral region (Fig. 2, Vb).
NOS cellular expression in intramembranous region
In the intramembranous bone formation region (Table 1), iNOS mRNA was detected on day 4 and day 7, with maximal expression on day 4 (positive signal in 52 ± 1.3% of cells). iNOS mRNA then decreased to nearly undetectable levels on day 14 and day 21. eNOS mRNA was detected at high levels throughout the time course and was maximal on day 7 (in 56 ± 1.7% of cells; 1.3 times that on day 4; p < 0.05). bNOS mRNA was maximally detected on day 4 (positive signal in 61 ± 3.5% of cells) and then declined to two-thirds of this on day 14 (p < 0.05) and rose to nearly the same level as day 4 on day 21.
The NOS isoform protein signal in cells of the intramembranous region was similar to that of their mRNA (Table 1). In the intramembranous region, 52 ± 1.7% of cells maximally stained for iNOS protein on day 4. iNOS protein decreased to undetectable levels on day 14 and day 21. eNOS protein started increasing from day 4 and reached its peak presence on day 21 (positive signal in 44 ± 2.3% of cells; 21 times that on day 4; p < 0.001). The bNOS protein signal was maximal on day 4 (in 58 ± 3% of cells) and then declined to two-thirds of this from day 7 (p < 0.05) to day 21.
NOS cellular expression in the chondral region
In the chondral region, there was minimal mRNA signal for all three isoforms on day 4 (Table 2). iNOS mRNA signal was maximally detected on day 7 (in 41 ± 3% of cells; 57 times that on day 4; p < 0.001) and then declined to undetectable levels on day 14 and day 21. eNOS mRNA was maximal on day 7 (positive signal in 51 ± 2.4% of cells; 38 times that on day 4; p < 0.001) and then declined slightly. The bNOS mRNA signal increased from day 4 and reached its peak presence on day 21 (in 62 ± 3.4% of cells; 38 times that on day 4; p < 0.001).
Table Table 2.. Percentage of Positively Stained Cells for NOS Isoform mRNA and Protein in the Chondral Region of the Fracture Callus
The NOS protein detection in cells of the chondral region was similar to that of mRNA (Table 2). There was minimal detection of NOS protein on day 4. Maximal presence of iNOS protein was found on day 7 (positive signal in 48 ± 2.7% of cells; 68 times that on day 4; p < 0.001); however, iNOS protein was nearly absent on day 14 and day 21. The eNOS protein signal was maximally detected on day 14 (in 47 ± 2% of cells; 36 times that on day 4; p < 0.001) and dropped slightly on day 21. The bNOS protein signal increased from day 4 and was maximal on day 21 (in 48 ± 2.2% of cells; 36 times that on day 4; p < 0.001).
NOS cellular expression in the fibrochondral region
In the fibrochondral region, there was minimal NOS mRNA detected for all three isoforms on day 4 (Table 3). iNOS mRNA was maximally detected on day 7 (35 ± 2.6% of cells stained positive; 18 times that on day 4; p < 0.001) and then declined to undetectable levels on day 14 and day 21. The eNOS mRNA signal was maximal on day 7 (in 51 ± 3% of cells; 49 times that on day 4; p < 0.001) and then decreased. The bNOS mRNA signal increased from day 4 and reached its peak presence on day 21 (in 58 ± 3% of cells; 30 times that on day 4; p < 0.001).
Table Table 3.. Percentage of Positively Stained Cells for NOS Isoform mRNA and Protein in the Fibrochondral Region of the Fracture Callus
The NOS protein signal in cells of the fibrochondral region was found similar to that of mRNA (Table 3). Minimal protein for all isoforms was found on day 4. iNOS protein expression was maximal on day 7 (positive signal in 32 ± 1.6% of cells; 15 times that on day 4; p < 0.001) and decreased to nearly undetectable levels on day 14 and day 21. eNOS protein signal was maximally detected on day 14 (in 43 ± 2.4% of cells; 38 times that on day 4; p < 0.001) and then declined. The bNOS protein signal increased from day 4 and was maximal on day 21 (in 43 ± 2.6% of cells; 33 times that on day 4; p < 0.001).
NOS cellular expression in composite callus
The data from all three regions (intramembranous, chondral, and fibrochondral region) were combined to give an estimate of the composite NOS expression in this composite callus (Fig. 3). The iNOS mRNA signal was detected on day 4 and day 7, with peak expression on day 7 (in 41 ± 1.9% of cells; 2.2 times that on day 4; p < 0.01) and then decreased to nearly undetectable levels on day 14 and day 21. eNOS mRNA increased from day 4 and was maximal on day 7 (positive signal in 51 ± 2.5% of cells; 3 times that on day 4; p < 0.01) and then decreased. bNOS mRNA kept rising from day 4 and was maximal on day 21 (58 ± 3.2% of cells stained positive; 2.7 times that on day 4; p < 0.01).
The NOS isoform protein signal in composite callus was similar to that of mRNA (Fig. 3B). iNOS protein was maximal on day 7 (positive signal in 38 ± 2% of cells; 2.1 times that on day 4; p < 0.01) and then fell to undetectable levels on day 14 and day 21. eNOS protein increased from day 4 and was maximal on day 14 (44 ± 2.2% of cells stained positive; 19 times that on day 4; p < 0.001). The bNOS protein signal kept increasing from day 4 and was maximal on day 21 (in 44 ± 1.9% of cells; 2.2 times that on day 4; p < 0.05).
Our study has shown that the distribution of NOS isoforms during rat femoral fracture repair is time dependent, isoform specific, and cellular distinctive.
During intramembranous bone formation, all three NOS isoforms were expressed within 4 days of fracture in periosteal cells and osteoblast-like cells. iNOS was present only in cells of the intramembranous region from day 4 to day 7; after day 14, most cells expressed eNOS and bNOS rather than iNOS (Fig. 4).
As endochondral ossification became dominant, the two constitutive NOS isoforms were expressed in the chondral and fibrochondral regions. Chondrocytes and osteoblasts act as the main cell types to carry out new bone formation within chondroid callus. Interestingly, our in situ hybridization and immunostaining studies showed that although chondroid cells expressed all three NOS isoforms on day 7, iNOS expression was transient and decreased markedly after day 7 (Fig. 4). In contrast, cells positive for eNOS were seen through day 14 and somewhat decreased in number by day 21(Fig. 4). At the final time point (day 21), bNOS was the predominant NOS isoform (Fig. 4).
Corbett et al.(23) have reported in an immunohistology study that osteoblasts and chondroblasts expressed iNOS and endothelial cells expressed eNOS in a rat tibial fracture callus. However, there are no studies addressing the cellular localization of bNOS in healing models. Our study has shown for the first time that bNOS is expressed in the early chondroid region of fracture callus.
Other authors have shown that NO affects bone-forming cell metabolism. Both iNOS and eNOS affect the cell growth and proliferation of osteoblasts,(12,24) and they also influence osteoclast replication and resorption activity.(14,25) During the initial phase of healing, the higher concentration of NO may play a role in preventing infection.(26,27) Furthermore, Corbett et al.(28) reported in a rat limb fracture model that NO mediated vasoreactivity and restored blood flow at a fracture site, investigated using laser Doppler flowmetry, during the early healing stage. eNOS as a source of NO may be responsible for vasodilation, which is necessary for the repair of fracture. The role of bNOS in fracture healing is undetermined.
Our current study has answered the question of what cells express the different isoforms of NOS as the fracture heals and describes the relationship between iNOS, eNOS, and bNOS isoforms, with dominance of eNOS and bNOS in middle and late stages of fracture repair after the decreased iNOS expression in fracture callus. This is consistent with the temporal expression pattern of NOS isoforms we have defined previously by competitive PCR and Western blot methods, which showed an increase of constitutive NOS expression in later stages of the healing process combined with a decrease of inducible NOS.(11)
This work may provide important information for a clinical strategy to modulate NOS isoform activity, either to enhance fracture repair or to suppress unwanted bone formation as in myositis ossificans.
This work was supported by St. George Hospital/South East Health, Sydney, and the Australian National Health and Medical Research Council.