Fracture healing is accelerated in the absence of the adaptive immune system

Abstract

Fracture healing is a unique biologic process starting with an initial inflammatory response. As in other regenerative processes, bone and the immune system interact closely during fracture healing. This project was aimed at further elucidating how the host immune system participates in fracture healing. A standard closed femoral fracture was created in wild-type (WT) and recombination activating gene 1 knockout (RAG1^−/−) mice lacking the adaptive immune system. Healing was investigated using micro–computed tomography (µCT), biomechanical testing, and histologic and mRNA expression analyses. Biomechanical testing demonstrated a significantly higher torsional moment on days 14 and 21 in the RAG1^−/− mice compared to the WT group. µCT evaluation of RAG1^−/− specimens showed earlier mineralization and remodeling. Histologically, endochondral ossification and remodeling were accelerated in the RAG1^−/− compared with the WT mice. Histomorphometric analysis on day 7 showed a significantly higher fraction of bone and a significantly lower fraction of cartilage in the callus of the RAG1^−/− mice than in the WT mice. Endochondral ossification was accelerated in the RAG1^−/− mice. Lymphocytes were present during the physiologic repair process, with high numbers in the hematoma on day 3 and during formation of the hard callus on day 14 in the WT mice. Expression of inflammatory cytokines was reduced in the RAG1^−/− mice. In contrast, expression of anti-inflammatory interleukin 10 (IL-10) was strongly upregulated in RAG1^−/− mice, indicating protective effects. This study revealed an unexpected phenotype of enhanced fracture healing in RAG1^−/− mice, suggesting detrimental functions of lymphocytes on fracture healing. The shift from proinflammatory to anti-inflammatory cytokines suggests that immunomodulatory intervention strategies that maximise the regenerative and minimize the destructive effects of inflammation may lead to enhanced fracture repair. © 2011 American Society for Bone and Mineral Research.

Introduction

Fracture healing is a unique physiologic process resulting in unscarred restoration of bone tissue.1 Sequential and overlapping stages of fracture healing can be distinguished.2 The initial stage after injury is characterized by hematoma formation and subsequent inflammation.3, 4 Growing evidence indicates that immune cells and their secreted factors are crucial for the physiologic response following injury.5 During this inflammatory response, cells and soluble factors of the hematoma interact with putative regenerative progenitor cells.6 Putative progenitor cells are likely located in the periosteum/endosteum, the bone marrow, and the surrounding soft tissues, including muscle tissue.7 These cells, together with chondroinductive and osteoinductive growth factors, promote formation and maturation of the callus.8 This callus growth recapitulates embryonic bone formation.9 However, one major difference is the strong inflammatory response following injury. The interplay between the immune and skeletal systems is in the focus of the emerging field of osteoimmunology.10–14

Progress has been made in understanding the link between the adaptive and innate immune system in bone, with emphasis on the osteoclast.15, 16 One experimental approach to improve fracture healing in an osteoimmunologic context is the application of anti–receptor activator of NF-κB ligand (anti-RANKL) molecules.17 In contrast, the function of the adaptive immune system in fracture healing is less understood. Lymphocytes are the main effector cells of the adaptive immune system and have been shown to contribute to regenerative processes.18 It was shown that CD8⁺ cytotoxic T cells have a counterregulatory role in wound healing because depletion of CD8⁺ T-lymphocytes has a positive influence on wound healing.19 In contrast, ablation of major histocompatability complex type II (MHC-II) and subsequent lack of T-helper cell–dependent immune reactions lead to delayed wound healing.20 Furthermore, arterial injuries in lymphocyte-deficient mice showed increased scar formation that has been rescued by adoptive transfer of B-lymphocytes.21 Lack of B-lymphocytes can lead to higher bone formation.22 A role of lymphocytes in fracture healing has been suggested23 but has not been characterized in depth yet.

Previous studies using pharmacologic approaches based on immunosuppression with, for example, cyclosporine or tacrolimus to dissect the role of T cells on fracture healing24 resulted in cross-reactivity and thereby also influenced mesenchymal cells. To gain further insight into the contribution of lymphocytes to fracture healing, absence of mature lymphocytes can be modeled by recombination activating gene 1 knockout (RAG1^−/−) mice that can be described as nonleaky SCID mice.25 The underlying defect is the loss of the recombination activating gene, which encodes for a recombinase that is crucial for the somatic VDJ recombination in the development of B- and T-cell receptors. The aim of this study was to use this highly specific system with respect to lack of lymphocytes. The hypothesis of this study was that fracture healing would be delayed in the absence of lymphocytes as effector cells of the adaptive immune system.

Material and Methods

Experimental design

Unilateral closed fractures were produced in the left femur of isoflurane-anesthesized 8-week-old male RAG1^−/− mice and wild-type (WT) mice (C57BL/6N background; Bundesinstitut für Risikobewertung, Berlin, Germany) according to Bonnarens and Einhorn.26, 27 The mice were evaluated at 3, 7, 14, 21, and 28 days. All bones were assessed at surgery and at harvesting by radiography. Fractures that did not meet standard criteria were not included in the subsequent analyses.

Animal experiments were carried out in accordance with the Care and Use of Animals and the national welfare guidelines. The study was approved by the local legal representative (Landesamt für Gesundheit und Soziales, Berlin, Germany, Registration Number G 0206/08).

Micro–computed tomography (µCT)

For characterization of 3D callus properties, including size, geometry, structure, and mineralization, the newly formed mineralized callus tissue was analyzed using high-resolution µCT. At days 7, 14, 21, and 28, fractured and contralateral femurs were batch scanned with a fixed isotropic voxel size of 10.5 µm (70 kVp, 114 µA; Viva40 µCT, Scanco Medical AG, Brüttisellen, Switzerland). The scan axis coincided nominally with the diaphyseal axis of the femurs. A minimum of 630 slices was chosen such that it included the fracture callus in all dimensions. The cortical bone was excluded from the volume of interest (VOI) manually. A fixed global threshold of 190 mg hydroxyapatite (HA)/cm³ was selected that allowed the rendering of mineralized callus only. This threshold was verified by manually evaluating 10 single tomographic slices from 4 samples per group to isolate the mineralized tissue and preserve its morphology while excluding unmineralized tissues. Standard µCT measures were calculated for each sample. All analyses were performed on the digitally extracted callus tissue using 3D distance techniques (Scanco Medical AG software).28

Biomechanical testing

After µCT analysis, fractured and nonfractured femurs were subjected to torsional testing to assess biomechanical properties. Proximal and distal ends were placed into device pots and embedded using polymethyl methacrylate. The femurs were aligned relative to the loading axis using a slot laser and centred in the device pots. Specimens were kept humidified during the whole procedure. Bones were loaded to failure in torsion (0.5 degree/sec, axial preload 0.3 N; BOSE ElectroForce 3200 test system, Eden Prairie, MN, USA). Ultimate torque, torsional stiffness, and energy to failure were calculated from torque-rotation curves.

Histologic analyses

For histologic analysis of fractures, femurs were harvested at 3, 7, 14, 21, and 28 days. All histologic analyses were conducted according to the ASBMR standards adapted to the study of bone repair.29, 30 Bones were excised with little surrounding soft tissue, fixed with 4% paraformaldehyde (PFA) at 4°C for 48 hours and decalcified in a 1:1 solution of 4% PFA and 14 % EDTA at 4°C for 3 weeks. Bones were dehydrated, embedded in paraffin, and cut into 4-µm sagittal slices. Movat-pentachrome staining was performed for histomorphometry.31 Computerized histomorphometric analysis was performed with an image-analysis system (KS400 3.0, Zeiss, Eching, Germany).

For immunofluorescent staining and confocal microscopy, histologic samples were fixed in 4% PFA and equilibrated in 30% sucrose/PBS. Cryostat sections of adult femurs were stained and mounted with fluorescent mounting medium (DakoCytomation, Glostrup, Denmark). Stainings were carried on using Cy5-labeled monoclonal antibodies against B220 (RA3.6B2) and biotinylated anti-CD3 (500A2; eBioscience, San Diego, CA, USA). For secondary reagent streptavidin-Alexa555 (Invitrogen Life Technologies, Carlsbad, CA, USA) was used. Cell nuclei were visualized by labeling with cytox green. All confocal microscopy was carried out on a DM IRE2 microscope (Leica, Wetzlar, Germany).

For osteoblast counting, Movat-pentachrome staining and osteocalcin immunohistochemistry were applied. Cells were counted as osteoblasts if they had contact with bone and displayed the typical palisade-like morphology. Three regions of the callus were selected in the lateral periosteal callus: a proximal region, a region at the level of the gap, and a distal region. For osteocalcin immunohistochemistry, paraffin sections were deparaffinized, washed, and hydrated. Antigen retrieval was performed using 0,18 % trypsin for 10 minutes at room temperature. All solutions were prepared in Tris-buffered saline (TBS). Samples were blocked with normal goat serum (S-1000, Vector Laboratories, Burlingame, CA, USA). Primary mouse monoclonal antibody to osteocalcin (Cat. no. LAX-210-333-C100, Enzo Life Sciences, Lörrach, Germany) was diluted in DAKO-Diluent (S-00809, Glostrup, Denmark) 1:4000. Samples were incubated at 4°C overnight and incubated with a secondary antibody (biotinylated antirabbit, made in goat, BA-1000) followed by incubation with AB- complex (alkaline phosphatse universal, AK-5200, Vectastatin ABC Kit). After incubation with chromogen buffer and AP substrate (Alkaline Phosphatase Substrate Kit 1, SK-5100, Vector Laboratories), sections were counterstained with methyl green.

For detection of collagen type II expression by chondrocytes, immunohistochemistry was applied. Sections were deparaffinized in xylene and rehydrated by an ascendant line of alcohol until hydration with PBS. For antigen retrieval, samples were incubated with hyaluronidase (2 hours at 37°C), followed by 0.1% pepsin (30 minutes at 37°C). Samples were blocked with normal horse serum (S-2000, Vector Laboratories) and diluted in PBS for 20 minutes at room temperature. Dilution of the primary antibody (mouse monoclonal to collagen type II, Cat. no. 031502101, Quartett, Berlin, Germany) was made in DAKO-Diluent (S-0809) 1:50. Samples were incubated with the primary antibody at 4°C over night. Samples were incubated with the secondary antibody (biotinylated antimouse IgG, made in horse, BA-2000) for 30 minutes at room temperature. Samples were washed two times in PBS followed by incubation with AB-complex (Alkaline Phosphatase Universal, AK-5200, Vectastatin ABC Kit) for 50 minutes. AP substrate (Alkaline Phosphatase Substrate Kit 1, SK-5100, Vector Laboratories) was added to the samples and incubated for 15 minutes.

In addition, collagen type X expressed by hypertrophic chondrocytes was detected by immunohistochemistry. Preparation of sections was similar to that for collagen II immunohistochemistry. For antigen retrieval, samples were incubated with pepsin (2 hours at 37°C), followed by incubation with hyaluronidase (30 minutes at 37°C). Samples were blocked with normal horse serum (S-2000, Vector Laboratories) and diluted in PBS for 20 minutes at room temperature. Dilution of the primary antibody (mouse monoclonal to collagen type X, Cat. no. 2031501005, Quartett) was made in DAKO-Diluent (S-0809) 1:100. Samples were incubated with the primary antibody at 4°C over night followed by incubation with the secondary antibody (biotinylated antimouse IgG, made in horse, BA-2000) for 30 minutes at room temperature. Samples then were incubated with AB-complex (Alkaline Phosphatase Universal, AK-5200, Vectastatin ABC Kit) for 50 minutes. After taking of the chromogen buffer, the AP substrate (Alkaline Phosphatase Substrate Kit 1, SK-5100, Vector Laboratories) was added to the samples and incubated for 15 minutes. Each step of incubation was followed by washing two times. After staining, the samples were counterstained with hematoxylin. To ensure specificity of antibodies, we performed negative controls that did not reveal any nonspecific staining.

Tatrate-resistant acid phosphatase (TRACP) staining was conducted at pH 5.0 in the presence of L(+)-tartaric acid using naphtol AS-MX phosphate (Sigma, St Louis, MO, USA) in N,N-dimetylformamide as a substrate. To meet the criteria of osteoclasts, cells had to have more than 2 nuclei, show TRACP⁺ staining, and be located next to the bone. The number of osteoclasts was determined.

mRNA preparation and expression analysis

Bones were excised, and the surrounding soft tissues were dissected. To standardize tissue collection, only the diaphyseal regions with or without callus were harvested. Collected tissues from 6 mice per time point were snap-frozen in liquid nitrogen and stored at −80°C. Control samples (day 0 time point) were collected from unfractured femurs of each RAG1^−/− and WT mice. Tissues were pulverized with pestle and mortar under continuous cooling with liquid nitrogen. After homogenization (T10, Ultra-turrax, IKA Werke GmbH, Staufen, Germany), total RNA was isolated from each sample using TRIzol (Invitrogen Life Technologies) according to the manufacturer's protocol. DNA was eliminated with DNAse I (Invitrogen Life Technologies).

All reagents for quantitative real-time polymerase chain reaction (qRT-PCR) analysis were purchased from Bio-Rad (ie, iScrip cDNA Synthesis Kit, iQ SYBR Green Supermix; Paris, France), and plate assays were read in Icycler IQ5 optical system software, Version 2.0 (Bio-Rad). One microgram of total RNA was used for each preparation of cDNA. The methods of DNA amplification were as described previously.32 In qRT-PCR, all samples were run in duplicate. Each plate contained two negative controls and a positive control. Expression levels were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Sample threshold values were then normalized for each time-point: ΔC_T = C_T(exp) – C_T(GAPDH). C_T values of day 0 were used as a reference: ΔΔC_T = ΔC_T(exp day_x) - ΔC_T(exp day₀). The fold change in mRNA expression for each time point was plotted in a graph using day 0 as a reference: 2^{−ΔΔCT(day 0} = 1.

Results

Enhanced biomechanical properties, callus mineralization, and remodeling during fracture healing in RAG1^−/− mice

Radiographs revealed that both animal groups displayed secondary fracture healing with formation of a mineralized callus on day 14 (Fig. 1A, B). Biomechanical testing clearly showed a significantly higher torsional moment on days 14 and 21 in the RAG1^−/− compared with the WT group (Fig. 1C). No significant difference was detected in torsional stiffness (Fig. 1D) or energy to failure.

Figure 1. Enhanced biomechanical properties and callus development of RAG1^−/− fractures. (A, B) Representative radiographic images of WT (A) versus RAG1^−/− (B) fractures on days 7, 14, 21, and 28. (C, D) Quantitative data from biomechanical testing of fractures. Ultimate torque (C) and torsional stiffness (D) relative to the intact side. (n = 8 each, ANOVA. ^ap < .05; ^bp < .01).

Qualitative and quantitative microstructural properties of the fractures suggested more mature callus tissues in the RAG1^−/− mice (Fig. 2). From µCT-based analysis it could be appreciated that mineralized regenerative tissue was visible adjacent to the cortical bone on day 7 (Fig. 2A, B). On day 14, the callus trabecular network was denser in the RAG1^−/− mice, and on day 28, increased mineralization of RAG1^−/− mice was reflected by a thicker neocortex in the callus periphery (Fig. 2A, B).

Figure 2. Microstructural properties of callus tissue analyzed by quantitative µCT. (A, B) 3D image reconstructions (longitudinal section, entire callus, and transverse section) of WT (A) versus RAG1^−/− (B) fractures on days 7, 14, 21, and 28. Scale bar = 1 mm. (C–F) Quantitative data of the µCT analysis. (C) Bone volume fraction (BV/TV) was elevated at early and late time points in RAG1^−/− mice. (D) BMD. (E) Total callus volume (TV) decrease was detected 1 week earlier in the RAG1^−/− mice. (F) Tissue mineral density (TMD) increased over time, suggesting maturation of newly formed bone in both groups. (n = 8 each, ANOVA; ^a,bp < .05).

Quantitatively, total callus volume (TV) of RAG1^−/− mice was significantly lower than in the WT group on day 21 suggesting earlier callus remodeling (Fig. 2E). In the RAG1^−/− mice, the callus was remodeled during the third week. In the WT mice, the callus volume decreased similarly, but in the fourth week. Furthermore, BMD was significantly higher in the RAG1^−/− mice on days 7 and 28 (Fig. 2D). This was corroborated by a significantly higher bone volume fraction on day 28 (Fig. 2C). Tissue mineral density (TMD) increased over time, suggesting a subsequent maturation of newly formed bone in both groups (Fig. 2F).

Accelerated endochondral bone formation and callus remodeling during fracture healing in RAG1^−/− mice

Histologic characterization revealed that endochondral bone formation and the final remodeling of the bony callus were accelerated in the RAG1^−/− mice compared with WT mice (Fig. 3). On day 3, the fracture gap was filled with hematoma (Fig. 3A, B), and a first periosteal reaction occurred (Fig. 3C). In the RAG1^−/− mice, early mineralization and subsequent bone formation were visible on day 7, and only very few cartilage remnants remained at the center of the callus on day 14. In contrast, the WT mice showed almost no mineralization of the callus on day 7 and a less organized callus structure with large islands of cartilage on day 14. In the later fracture healing process, a neocortex was formed at the periosteal surface, growing thicker and melting down to the original shape of the bone. This remodeling appeared to be faster in the RAG1^−/− mice, as suggested by the histologic images (Fig. 3A, B). Histomorphometric analysis on day 7 revealed a significantly higher fraction of bone and, inversely, a significantly lower fraction of cartilage in the callus of the RAG1^−/− mice compared with the WT mice (Fig. 3E, F). During endochondral bone formation on day 14, the fraction of cartilage in RAG1^−/− mice was lower than in the WT mice (Fig. 3F).

Figure 3. Histologic analysis of fracture healing demonstrates earlier mineralization and remodeling in RAG1^−/− mice. (A, B) Representative histologic images of WT (A) versus RAG1^−/− (B) fractures ont days 3, 7, 14, 21, and 28. Scale bars = 2 mm; Movat-pentachrome staining. (C) Detail of periosteal reaction on day 3 in WT (left) versus RAG1^−/− (right) mice. Note the thicker layer in RAG1^−/− mice. Scale bars = 100 µm. (D) Detail of hypertrophic cartilage formation on day 7 in WT (left) versus RAG1^−/− (right) mice. Note the enhanced mineralization in the RAG1^−/− mice. Scale bars = 100 µm. (E, F) Graphic presentation of quantitative data from the histomorphometric analysis of more mature callus tissues. (E) Total osseous tissue (TOT) of callus is higher in RAG1^−/− mice on day 7. (F) Cartilage fraction (Cg). Lower cartilage volume ratio on days 7 and 14 indicates earlier endochondral ossification (n = 8 each, ANOVA; ^ap < .05; ^bp = .01).

Since RAG1^−/− mice displayed faster bone formation than WT mice, we decided to assess the presence of lymphocytes in the fracture callus by confocal microscopy. We assessed whether lymphocytes migrated within the callus of WT mice and whether this was related to the observed differential pattern of fracture repair.

Three days after fracture, B and T cells were detected within the hematoma near the fracture gap and infiltrating the callus from the surrounding soft tissues (Fig. 4A, B). Remarkably, B cells constituted the relative majority of these cells, and they were found to localize also in the areas of the callus that would be later on populated by hypertrophic chondrocytes. At this time point, lymphocytes could not be detected nearby or between periosteal cells. After 7 days, lymphocytes became very rare in the callus. They could not be found within areas of hypertrophic chondrocytes (Fig. 4C) but rather at their boundary (Fig. 4D). Two weeks after fracture, the callus was found to be infiltrated by large numbers of lymphocytes. They did not reside near the central zone where cartilage was still present but rather peripherally (Fig. 4E), concentrating in areas of newly formed woven bone (Fig. 4E–G), as identified previously by Movat-pentachrome staining (Fig. 3). Also at this late time point, the majority of callus infiltrating lymphocytes consisted of B cells (Fig. 4F, G). Interestingly, T and B cells could be detected in close proximity or in contact with bone lining cells (Fig. 4G).

Figure 4. T- and B-lymphocytes are present in the fracture callus during healing. Longitudinal sections of fractured femurs from WT mice were taken from animals euthanized after 3 (A, B), 7 (C, D), and 14 (E–G) days. Sections were stained for the presence of B cells (B220) and T cells (CD3). Cell nuclei were visualized by labeling with cytox green. Arrows indicate stained cells (magnification ×200). Panels B, F, and G were obtained by confocal scanning. Panels B and F represent smaller areas of panels A and E. Representative sections from one of three similar experiments are shown. H = hematoma next to fracture gap; ST = soft tissue (muscle); C = cartilage; WB = woven bone.

Immunohistochemistry of collagen types II and X further revealed the distribution of cartilage on day 7 (Fig. 5A, B). In the WT animals, larger callus regions were collagen type II⁺ than in the RAG1^−/− mice (Fig. 5A). This observation corroborates the result of the Movat-pentachrome staining showing mineralized tissue in the central callus region that lacked immunostaining for both collagen types. Presence of collagen type X appeared more accentuated in the RAG1^−/− mice, indicating a higher amount of hypertrophic cartilage on day 7 (Fig. 5B).

Figure 5. Patterns of cartilage and bone formation. (A) Collagen II immunohistochemistry. Scale bar = 2 mm. Note the lower collagen II content on day 7 in RAG1^−/− fractures. (B) Collagen X immunohistochemistry. Scale bar = 2 mm. The transition zone between bone and cartilage is similar in both groups. (C, D) Movat-pentachrome analysis of osteoblast number (N.Ob); arrows indicate typical osteoblasts. (E, F) Osteocalcin immunohistochemistry analysis of osteoblast number (N.Ob); arrows indicate typical osteoblasts (n = 9 each, ANOVA; ^ap < .05; ^bp < .01; scale bar = 20 µm). Relative mRNA expression of genes involved in bone formation (G–I): collagen I (G), osterix (H), and osteopontin (I) (n = 6 each, t test; ^ap < .05; ^bp < .01).

Following cartilage hypertrophy, we observed earlier endochondral ossification and bone formation in the RAG1^−/− fractures. This could be explained either by an increased osteoblast activity or a higher proliferation of osteoblasts resulting in a higher osteoblast number. To analyze whether absence of lymphocytes influences bone formation, osteoblast number in the bony callus was determined on days 14 and 21 in Movat-pentachrome-stained tissue (Fig. 5C, D). Differences in osteoblast numbers on day 14 suggested higher osteoblastic activity in RAG1^−/− fractures. Lower osteoblast numbers on day 21 in RAG1^−/− mice, in line with the descriptive µCT results, suggested an earlier completion of hard callus formation and onset of remodeling (Fig. 5C, D). Additional immunohistochemical staining of osteocalcin⁺ osteoblasts over the healing course corroborated the result and further revealed higher osteoblast numbers on day 28 in RAG1^−/− mice (Fig. 5E, F). To further support the histologic results, relative mRNA expression levels of bone extracellular matrix genes (Col1A2), bone-specific transcription factors (osterix), and osteopontin were determined (Fig. 5G–I). Col1A2 expression was induced on day 3 and peaked on day 14, followed by a marked decrease in the RAG1^−/− mice. In the WT mice, expression peaked at day 7 but was induced later and lasted longer (Fig. 5G). A similar pattern was observed in osteopontin expression (Fig. 5I). The zinc finger transcription factor osterix/SP7 serves as an inhibitor of chondrogenesis and chondrocyte maturation, whereas it promotes osteoblast maturation.33 Relative expression of osterix in the RAG1^−/− showed a constantly low level. In the WT mice, relative expression of osterix had a biphasic pattern, with peaks on days 3 and 21, split by a marked drop on day 14 (Fig. 5H). In summary, fracture healing was completed 1 week earlier (about 25% faster) in RAG1^−/− mice.

Accelerated tissue resorption in the absence of lymphocytes

To dissect tissue resorption, we analyzed osteoclasts histologically and performed expression analysis of resorption-specific genes. On day 7, osteoclast number was significantly higher in RAG1^−/− mice than in the WT mice (Fig. 6A, B). The pattern of cathepsin K expression was similar in both groups, with higher values in RAG1^−/− mice on day 14, in accordance with histologic observation (Fig. 6C). Taken together, these results suggest that tissue resorption is accelerated in RAG1^−/− fractures. To further investigate the early onset of tissue resorption, expression of the osteoclastogenic factor RANKL and its decoy receptor OPG was analyzed. Early high levels of RANKL expression in the RAG1^−/− mice corresponded with early high resorptive activity (Fig. 6D). In line with the latter, early high expression of OPG (on day 7) in the WT mice compared with early low levels in RAG1^−/− and might have counterbalanced osteoclastogenesis (Fig. 6E). The RANKL/OPG ratio peaked markedly on day 14 in the WT mice (Fig. 6F). This peak coincides with a drop in the constantly moderate RANKL/OPG ratio of the RAG1^−/− mice. The results indicate a shifted balance between bone formation and bone resorption toward bone formation in the RAG1^−/− mice. Together with the histologic results, this suggests faster kinetics of both bone formation and bone resorption in the RAG1^−/− mice.

Figure 6. Patterns of osteoclast differentiation and activity. (A) Representative histologic images of WT (left) versus RAG1^−/− (right) fractures on day 14. Scale bar = 100 µm; tatrate resistant acid phosphatase (TRACP) staining. Note the abundant osteoclasts in RAG1^−/− fractures. (B) Number of osteoclasts/bone surfaces on days 3, 7, 14, 21, and 28 (n = 8 each, ANOVA; ^bp = .01). (C–F) Relative mRNA expression of genes involved in osteoclast differentiation and function: (C) cathepsin K, (D) RANKL, (E) OPG, and (F) RANKL/OPG ratio (n = 6 each, t test; ^ap < .05; ^bp < .01, ^cp < .001).

Balance of the inflammatory reaction is shifted toward anti-inflammation during fracture healing in RAG1^−/− mice

To further investigate the immunologic basis for the accelerated healing kinetics in RAG1^−/− mice, we measured the expression kinetics of several relevant cytokines in parallel (Fig. 7A–F). The potent proinflammatory mediator tumor necrosis factor α (TNF-α) was expressed earlier and at higher levels. Then, at later time points, this cytokine was downregulated in WT mice, whereas its levels remained relatively constant and low in RAG1^−/− mice. Other important mediators of inflammation such as lymphotoxin-β (LT-β) and interferon γ (IFN-γ) always were expressed at higher levels in WT mice. On the other hand, T-cell-specific cytokines such as interleukin 2 (IL-2) and IL-4 were upregulated in WT mice and, as expected, downregulated in RAG1^−/− mice. Interestingly, the potent negative regulator of T-cell activity, IL-10, was rapidly expressed in both types of mice. However, RAG1^−/− mice displayed a capacity to express this factor several orders of magnitude higher than in WT mice. Overall, these data suggest that in the fracture callus, the cytokine milieu of fast-healing RAG1^−/− mice contains a strong anti-inflammatory component.

Figure 7. Relative mRNA expression of cytokines. Expression of inflammatory cytokines was reduced in RAG1^−/− mice. (A-C) Inflammatory cytokines such as TNF-α (A), LT-β (B), and IFN-γ (C). (D, E) Expression of lymphocyte-specific cytokines IL-2 (D) and IL-4 (E), indicating active lymphocytes in physiologic fracture healing. Expression of anti-inflammatory cytokine IL-10 was strongly upregulated in RAG1^−/− mice, indicating protective effects (F) (n = 6 each, t test; ^ap < .05; ^bp < .01; ^cp < .001).

Discussion

This study aimed at elucidating the role of the adaptive immune system in fracture healing. Remarkably, we found that lack of lymphocytes led to accelerated bone regeneration in RAG1^−/− mice, suggesting a regulatory role of these cells in fracture repair.

In contrast to regenerative processes, the traditionally considered role of lymphocytes is controlling infection and circumventing tumorigenesis.34 Strong inflammation after injury may have evolved as a reaction directed against pathogens in cases of concomitant infection. This inflammation might have deleterious effects on regeneration. This assumption is corroborated by the report that γ/δ T cells negatively influence fracture healing.35 While it is interesting to speculate that the immune system harbors elements noxious36 and elements that are beneficial for regeneration,37 our findings indicate that lymphocytes exert a detrimental function during the early phase of fracture healing.

The key feature for accelerated fracture healing is the regain of biomechanical competence with successful callus mineralization. The processes leading to this accomplishment are the formation of a soft callus that is subsequently mineralized and remodeled. The remaining question is how lymphocytes alter these processes and in particular the balance between osteoblast and osteoclast activities.

As we show in this work, T- and B-lymphocytes were recruited from the earliest time during fracture healing. Interestingly, they migrated to tissues that were shown in lymphopenic mice to harbor bone earlier than WT controls. Thus it seems conceivable that T- and B-lymphocytes exert a role in controlling bone regeneration in the callus. It has been argued previously that during bone formation, cytokines secreted by lymphocytes have paracrine effects on bone cell differentiation and function.12 These previous works focused on the importance of T-cell activity to fracture healing. However, our data suggest that probably B cells and not T cells are predominant during immunoregulation of bone repair. In addition, it has been proposed previously that T cells can promote bone resorption directly by induction of osteoclastogenesis via RANK-RANKL interaction between T cells and osteoclasts.38 Based on this assumption, the phenotype of RAG1^−/− mice could be dependent on reduced resorption owing to lack of osteoclastogenic cells. However, our data showed unexpectedly higher than normal numbers of osteoclasts in the callus of these animals. We think that the detected higher osteoblast activity in the RAG1^−/− mice may have contributed to the observed phenotype in our study because elevated osteoclast number and activity may mirror the accelerated bone formation and higher amount of bone already existing that had to be remodeled. This increased bone formation is further corroborated by the low RANKL/OPG ratio in the RAG1^−/− mice.

Activated lymphocytes are also an abundant source of cytokines that can strongly affect bone cell activity. T helper lymphocytes (Ths) can be classified into major subsets according to the effector cytokines they produce. Th1 cells promote cellular immunity via secretion of cytokines such as TNF-α, IFN-γ, and IL-2, whereas Th2 cells foster humoral immunity by releasing mediators such as IL-4 and IL-10. Other subsets include Th17 cells, which contribute to osteoclastogenesis36 and regulatory T cells that suppress osteoclast formation and are also potent producers of IL-10.39

Faster healing in RAG1^−/− mice correlated with lower levels of expression of proinflammatory cytokines such as TNF-α and LT-β in the callus. Indeed, it can be conceived that lower levels of TNF-α may favor bone formation because it can act as a proapoptotic factor on osteoblasts,40 and its high levels have been implicated in several animal models of disease characterized by extensive bone destruction such as rheumatoid arthritis.41 In addition, it is described that injection of lipopolysaccharide (LPS) in mice induces TNF-α-dependent bone loss owing to enhanced osteoclastogenesis.42 However, it also was shown that absence of TNF-α signaling impairs fracture healing.43, 44 We therefore assume that positive and negative effects of TNF-α depend on concentration.

In addition, the strongly reduced expression of LT-β further suggests lower inflammatory potential in RAG1^−/− fracture healing. Interestingly, even if downregulated relative to controls, IFN-γ still was expressed largely during fracture healing in lymphopenic mice. This Th1 cytokine can be produced not only by T cells but also by activated natural killer (NK) cells, which are still present in RAG1^−/− mice.45 Thus our data suggest that these cells of the innate immune system rapidly produced IFN-γ in response to bone fracture and maintained its expression for a prolonged period of time. Our results point to NK cells as a major source of this cytokine during fracture repair. However, the relatively lower expression of IFN-γ in RAG1^−/− mice may lead to reduced macrophage activation, possibly reducing tissue destruction. In accordance with a mechanism of enhanced fracture healing based on dampened inflammation, lymphopenic mice also displayed earlier and strikingly higher expression of IL-10. This cytokine can be produced by either myeloid cells or regulatory T cells and is a potent inhibitor of secretion of proinflammatory cytokines such as TNF-α, IL-1, and IL-6 and T-cell activation.46 Our data suggest that during fracture healing, lymphocytes can negatively regulate the amount of IL-10 secreted in the callus. IL-10 also can control bone resorption,47, 48 and IL10 knockout mice display osteopenia, mechanical fragility of bones, and defects in their formation.49 This suggests that this anti-inflammatory cytokine may endorse a central role in promoting bone growth, and its rapid and strong upregulation in RAG1^−/− mice might explain their capacity to heal faster.

In accordance with previous works,50 the active contribution of lymphocytes to fracture healing is highlighted by the significantly elevated expression in the fracture callus of the genes for T-cell effector cytokines such as IL-2, IFN-γ, and IL-4. The role of IL-2, an important T-cell growth factor,51 has not yet been investigated systematically in the context of fracture repair, to our knowledge. Our results show that IL-2 is constitutively expressed during fracture healing, but it seems not to be essential for effective repair because RAG1^−/− mice were fully capable of recovering from injury. However, lack of IL-2 may have played a role in the enhanced bone formation that we observed in RAG1^−/− mice. In fact, previous studies show that infusion of IL-2 increased bone resorption in incisor absent (ia) osteopetrotic rats.52 This work suggests that in vivo, IL-2 can promote bone resorption; thus the enhanced healing phenotype that we observed may depend in part on the lack of IL-2 in immunodeficient mice.

The Th2 cytokine IL-4 is considered to exert anti-inflammatory functions. However, IL-4 also can be produced by non-B/T cells, for example, mast cells.53 IL-4 was discovered as a “B-cell growth factor”54 and inhibits monocyte production of IL-1, TNF-α. and PGE₂.55 Furthermore. IL-4 inhibits bone resorption56 and is a chemoattractant for osteoblasts,57 stimulating proliferation and inhibiting differentiation. During physiologic fracture healing, expression of IL-4 was upregulated, but not in RAG1^−/− mice, pointing to T cells as its producers. Overall, these data suggest that lack of Th2 cells does not impair fracture repair. There are several limitations in this study. First, mRNA expression analyses give only indirect evidence and need to be interpreted cautiously because posttranscriptional mechanisms play a crucial role. Second, we did analyses only of whole callus specimens that are composed of various types of tissues. Third, although bone properties are not influenced, RAG1^−/− knockout mice have elevated numbers of NK cells that might partly substitute the niches of depleted T-lymphocytes in nonlymphatic tissue58 and might adopt at least in part lymphocyte functions.

Among the most striking findings from this study was that early bone formation and subsequent mechanical competence of fractures were accelerated in lymphocyte-deficient RAG1^−/− mice. One underlying effect could be the shift from proinflammatory cytokines to anti-inflammatory cytokines, namely, IL-10, that is caused by the absence of lymphocytes leading to accelerated bone formation. To summarize, it is likely that cytokine effects on fracture healing can be divided into noxious and beneficial.5 In case of injury, this might reflect the inherent physiologic compromise procuring defensive mechanisms and reparative potential at the same time. New immunomodulatory intervention strategies59 to improve fracture healing could specifically target lymphocyte subsets or cytokines to eliminate detrimental action and to preserve the necessary leverage.

Disclosures

All the authors state that they have no conflicts of interest.

Acknowledgements

We acknowledge assistance with the analyses by Alexander Schill and Mario Thiele. This study was supported by the Berlin Brandenburg Center for Regenerative Therapies (BCRT) and a resident fellowship (GEROK) within Collaborative Research Center (SFB) 760 of the German Research Foundation (DT).