Defective bone repair in mast cell-deficient Cpa3Cre/+ mice

In the adult skeleton, cells of the immune system interact with those of the skeleton during all phases of bone repair to influence the outcome. Mast cells are immune cells best known for their pathologic role in allergy, and may be involved in chronic inflammatory and fibrotic disorders. Potential roles for mast cells in tissue homeostasis, vascularization and repair remain enigmatic. Previous studies in combined mast cell- and Kit-deficient KitW-sh/W-sh mice (KitW-sh) implicated mast cells in bone repair but KitW-sh mice suffer from additional Kit-dependent hematopoietic and non- hematopoietic deficiencies that could have confounded the outcome. The goal of the current study was to compare bone repair in normal wild type (WT) and Cpa3Cre/+ mice, which lack mast cells in the absence of any other hematopoietic or non- hematopoietic deficiencies. Repair of a femoral window defect was characterized using micro CT imaging and histological analyses from the early inflammatory phase, through soft and hard callus formation, and finally the remodeling phase. The data indicate 1) mast cells appear in healing bone of WT mice but not Cpa3Cre/+ mice, beginning 14 days after surgery; 2) re-vascularization of repair tissue and deposition of mineralized bone was delayed and dis-organised in Cpa3Cre/+ mice compared with WT mice; 3) the defects in Cpa3Cre/+ mice were associated with little change in anabolic activity and biphasic alterations in osteoclast and macrophage activity. The outcome at 56 days postoperative was complete bridging of the defect in most WT mice and fibrous mal-union in most Cpa3Cre/+ mice. The results indicate that mast cells promote bone healing, possibly by recruiting vascular endothelial cells during the inflammatory phase and coordinating anabolic and catabolic activity during tissue remodeling. Taken together the data indicate that mast cells have a positive impact on bone repair.

Introduction high levels only in the mast cell lineage, results in the complete absence of mast cells in connective and mucosal tissues by genotoxicity [16]. This model of mast cell deficiency has been used by many investigators to probe roles of mast cells under physiological and pathological conditions. Of note, in almost all instances, the suggested roles of mast cells could not be reproduced, thus contesting previous work in Kit mutant mice [11,15]. Because Cpa3 Cre/+ mice have no defects in the immune system other than the complete absence of mast cells and a partial reduction in basophils, we consider these mice an appropriate model to investigate the potential contribution of mast cells to bone repair in the absence of confounding factors arising from alteration in other cell lineages.

Mouse model of mast cell deficiency
Animal procedures were conducted in accordance with a protocol approved by the Facility Animal Care Committee of McGill University (AUP-7016), in keeping with the guidelines of the Canada Council on Animal Care. Animal surgery and post mortem analyses were performed essentially as described previously [17][18][19]. Founder mice heterozygous for insertion of Cre recombinase in the gene encoding mouse mast cell Cpa3 (Cpa3 Cre/+ mice on the C57BL/ 6 background) were obtained from the German Cancer Research Center, DKFZ, Heidelberg, Germany. A colony was established by mating with WT C57BL/6 mice (Charles River Laboratories, Senneville, Qc H9X 3R3, Canada) and the offspring genotyped as described [16] using PCR of DNA isolated from ear punch biopsies. Cpa3 Cre/+ mast cell-deficient male and female offspring were separated and 3-4 mice/cage maintained with free access to food and water from 4.5 to 8 months prior to surgical intervention.

Surgical model
Adult male and female mice were used for all experiments. Bilateral 1 mm x 2 mm defects were generated on the anterolateral aspect of the femora using the third trochanter as an anatomical landmark. Mice were anesthetized with isoflurane before shaving both hind limbs, disinfecting the skin with 70% ethanol and exposing the anterolateral aspect of the femora through a 3-mm skin incision extending from the third trochanter down the diaphysis. A 1 mm x 1 mm x 2 mm rectangular cortical window defect was generated with a 1 mm burr on Stryker drill (50,000 RPM, Hamilton, ON, Canada). After gentle irrigation to remove bone shards the muscle and skin layers were re-apposed and sutured with PDS-II 4-0 thread. For pain control, an IP injection of 10 mg/kg carprofen with 0.1 mg/kg buprenorphine in 0.5 mL of sterile saline was administered immediately after wound closure, and 5 mg/kg carprofen injected for 3 days post-operative. Cohorts of mice were euthanized by CO2 asphyxiation under anesthesia from 5-56 days post-operative. Femora were carefully dissected free of soft tissue before fixing for 24 hours in 4% paraformaldehyde. The bones were then rinsed x3 with sterile PBS and stored at 4o C until micro computed tomographic (micro CT) imaging.

Micro CT analysis
Scans were performed on a Skyscan 1172 instrument (Bruker, Kontich, Belgium) with a 0.5 mm aluminium filter at a voltage of 50kV, a current of 200μA and a resolution of 5 μm/pixel. 2D images were reconstructed into 3D models using NRecon software v. 1

Histological analysis
Bones were either left un-decalcified and embedded in poly-methyl methacrylate (PMMA) plastic or decalcified and embedded in paraffin using established methodology. Skin and bone tissues were stained for mast cells using acidified toluidine blue (aTB) and anti-tryptase antibody (Abcam, Cambridge MA, USA ab 151757 1:300). CD34 (Abcam ab23830 1:300) immunohistochemistry was used to identify vascular endothelial cells in bone and soft tissue in regenerating bone. Sections of PMMA embedded bones were stained with von Kossa and counterstained with toluidine blue (VK/TB) to distinguish mineralized from soft tissue. VK/TB stained sections were compared with 2D micro CT images from the same region, or with sections of decalcified, paraffin embedded bone stained to identify alkaline phosphatase (ALP) activity in osteogenic cells. Tartrate resistant acid phosphatase (TRAP) histochemical staining was used to identify osteoclasts and F4/80 (Abcam ab6640 1:200) immunohistochemistry to identify macrophages in decalcified bone sections. Microscopic images were captured with a Zeiss Axioskop 40 microscope (Carl Zeiss, Toronto, ON, Canada) and stain intensity expressed as % of the ROI using ImageJ v.1.6.0 software (NIH, Bethesda, MD, USA).

Statistical analysis
Quantitative data is expressed as mean ± SD and the open source statistical program R v.3.3.0 (R Core Team, 2015) used for Wilcoxon sum-rank tests. Analysis of variance (ANOVA) followed by Tukey post-hoc analysis were used for longitudinal and multiple comparisons between the different time points in the WT and Cpa3 Cre/+ mice. Differences were considered significant at p <0.05.

Distribution of mast cells during bone repair
Mice were genotyped into Cpa3 +/+ (mast cell proficient; WT) and Cpa3 Cre/+ (mast cell deficient) littermates by PCR (Fig 1, panel A). Mast cells are normally distributed throughout the dermis [20]. Skin biopsies harvested from the backs of adult WT and Cpa3 Cre/+ mice were stained with aTB and for MC tryptase. Numerous aTB (Fig 1, panel B) and MC tryptase positive (Fig 1, panel D) cells were seen in the dermis, usually around hair follicles, in all WT (Fig  1, panel A) but not in Cpa3 Cre/+ (Fig 1, panels C and E) mice. The same strategy was used to identify mast cells in sections of regenerating bone harvested from mice euthanized from 5d-56d PO. At 5d PO aTB positive cells were occasionally seen in WT mice on the periosteal surface of the proximal femur outside the region of the defect. At 14d PO they were seen in residual connective tissue in the defect/medulla of WT mice (Fig 1, panel B2) and in marrow, usually adjacent to vascular channels at 28d and 56d PO (Fig 1, panels B3 and B4). aTB positive cells were never seen in any of the bones harvested from Cpa3 Cre/+ mice (Fig 1, panels C1 to C4). Quantitative data for mast cell staining, expressed as the number of aTB positive cells/ mm 2 , is shown in Table 1. Non-granular, anti-tryptase antibody reactive cells of undefined

Micro CT analysis of bone repair
2D micro CT images were reconstructed into 3D models in which new bone (white) is distinguished from pre-existing bone (dark grey) (Fig 2, panels A to H). A complex healing pattern was seen in the Cpa3 Cre/+ mice that differed from that seen in the WT mice. At 14d PO less of the periosteal surface opposite the defect was covered with new bone in WT (Fig 2, panels B and C) mice than in Cpa3 Cre/+ (Fig 2, panels F and G arrows) mice. By 56d PO, bridging of the defect was most often complete in WT (Fig 2, panel D) mice but remained incomplete with residual fibrous tissue in most Cpa3 Cre/+ mice (Fig 2, panel H arrow). Quantitative analysis of bone regeneration in the cortex and defect/medulla ROIs is shown in Table 2. By 14d PO the WT mice had less cortical bone with wider spaces between trabeculae (Tb.Sp.) and increased bone porosity (Po.Vop; Po.op) compared with Cpa3 Cre/+ mice, but with little difference in the defect/medulla. By 56d PO BV/TV was higher, with narrower spaces and less porosity, in WT than in Cpa3 Cre/+ mice in both cortex and defect/medulla. (n = 6-10) (n = 7-9) (n = 6-13) (n = 7-10) (n = 8-9) (n = 8-10) (n = 5-6) (n = 5-9) Cortex . At 5d PO vessels were starting to penetrate the repair tissue from the proximal femur in WT mice but were not yet visible in Cpa3 Cre/+ mice. By 14d PO there was a dramatic increase in vessels in both WT and Cpa3 Cre/+ bones. Whereas the distribution pattern was even in WT mice it was skewed to the proximal pole leaving a distal area with no penetration (asterix) and few cortical vessels in Cpa3 Cre/+ mice. By 56d PO the new vessels were effectively restricted to cortical bone. Quantitative analyses for new vessels is shown in Table 3. At 5d PO there were more vessels with a higher volume and better connectivity in WT than in Cpa3 Cre/+ bone, and the number, volume and thickness remained higher at 56d PO. Images are representative of N = 7 WT and N = 6 Cpa3 Cre/+ at 5d PO; N = 16 WT and N = 11 Cpa3 Cre/+ at 14d PO; N = 8 WT and N = 10 Cpa3 Cre/+ at 28d PO and N = 6 WT and N = 8 Cpa3 Cre/+ at 56d PO. images were tilted at a 45˚angle to show healing of the defect over time. At 14d PO (B, F) significant new bone (white) is seen in the medullary canal and on the periosteal surface at the level of the defect (B, F arrows). By 56d PO, bone regeneration and remodelling have effectively closed the defect in the WT femur whereas mal-union is evident in the Cpa3 Cre/+ femurs, with holes penetrating the new cortical bone (H arrow). 3D models of hemi-femora (A1-H1) show the distribution of blood vessels (white) at the same time points. Revascularization of bone reaches a peak at 14d postoperative, with a skewed distribution in Cpa3 Cre/+ mice (F1 asterix), and is restricted to cortical bone by 56d PO (D1, H1).
The localization of vessels in regenerating bone was then compared with the results of CD34 immunohistochemistry, which was used as a sensitive marker for vascular endothelial cells in soft tissue and bone (Fig 3). At 5d and 14d PO there was an extensive network of vessels in the defect, medulla and cortex of WT bones, compared with the less dense pattern of vessel distribution seen in Cpa3 Cre/+ bones. By 28d PO there were few CD34 positive cells in the medulla of WT mice compared with numerous cells embedded in fibrous tissue remaining in the defect and medulla of Cpa3 Cre/+ bones (asterix). Quantification of CD34 staining, shown in Table 1, revealed peak activity at 14d PO in the defect/medulla and at 28d PO in the cortex of both WT and Cpa3 Cre/+ bones, but with fewer vessels at 28d and 56d PO in the WT mice.

Histological analysis of bone repair
To further characterize the quality of regenerated bone we stained thin sections of un-decalcified bone harvested from the mid-saggital plane of the defect with von Kossa (Fig 4) to compare with 2D micro CT images. The morphological features seen in the 2D micro CT images (Fig 4, panels A to H) were reflected in the VK/TB stained histological sections. Residual shards of old bone (Fig 4, panels A and E arrows) were seen at 5d PO but no new bone was visible until 14d PO (Fig 4, panels B and F). Bridging of the defect was evident at 28d PO in WT but not in Cpa3 Cre/+ mice (Fig 4, panel C vs G) in which significant fibrous tissue remained (Fig 4, panel G asterix). By 56D PO bone repair was effectively complete in most WT bones (Fig 4, panel D) compared with Cpa3 Cre/+ bones, where the defect was filled with thin bone interspersed with fibrous tissue (Fig 4, panel H asterix). The spaces between old and new bone seen in the cortex at 28D were retained only in Cpa3 Cre/+ bones (Fig 4, panel H arrows).  The cortical window defect is stable and therefore heals through intra-membranous bone formation, with no cartilage intermediate, so ALP is expressed only by anabolic cells of the osteoblast lineage. High magnification images of VK/TB and ALP stained sections from the corresponding region of the defect/medulla (Fig 5) revealed little bone forming activity at 5d PO in either WT (Fig 5A) or Cpa3 Cre/+ (Fig 5, panel E) bones. At 14d PO there was an extensive network of new bone trabeculae at the proximal end of the defect and in the adjacent medulla, associated with intense ALP activity in both WT and Cpa3 Cre/+ bone (Fig 5, panels B and F). At 28d and 56d PO there was more bone and less osteoid (blue) spanning the defect in WT (Fig 5, panel C) than in Cpa3 Cre/+ (Fig 5, panel G) mice, while ALP activity appeared similar. Detailed examination of VK/TB staining in the cortex (Fig 6) revealed less pronounced fibrous periosteal tissue in WT (Fig 6, panel A) than Cpa3 Cre/+ bone (Fig 6, panel E asterix) at 5d PO, prior to active bone formation at 14d PO when WT (Fig 6, panel B) and Cpa3 Cre/+ (Fig 6, panel B and F) bones looked similar. At 28d and 56d PO the cortex of WT bones (Fig 6, panels C and D) had fewer lacunae and less osteoid than of Cpa3 Cre/+ bones (Fig 6, panels G and 6).
Adjacent sections of decalcified bone were stained with TRAP to identify osteoclasts or with F4/80 antiserum to identify macrophages (Fig 7). TRAP positive cells were seen at the proximal end of the defect and in the fibrous tissue filling the defect at 5d PO in WT bones (Fig 7, panel A) but not in Cpa3 Cre/+ bones (Fig 7, panel B). Peak TRAP activity occurred at 14d PO and was more intense in WT (Fig 7, panel C) than in Cpa3 Cre/+ bones (Fig 7, panel D), whereas it declined at 28d PO in WT (Fig 7, panel E) but not in Cpa3 Cre/+ bones (Fig 7, panel  F). F4/80 staining was prominent in connective tissue of WT (Fig 7, panel A1) and Cpa3 Cre/+ (Fig 7, panel B1) bones at 5d PO. By 14d PO prominent staining was seen adjacent to bone in WT (Fig 7, panel C1) but not in Cpa3 Cre/+ (Fig 7, panel D1) bones, where F4/80 positive cells remained scattered in connective tissue. By 28d PO the F4/80 positive cells were seen in marrow adjacent to bone in WT mice (Fig 7, panel E1) and persisted in fibrous tissue in Cpa3 Cre/+ bone (Fig 7, panel F1). Quantitative analyses for ALP, TRAP and F4/80 staining (Table 1)

Discussion
The goal of the current study was to characterize the impact of mast cell deficiency on the repair of cortical bone defects using adult mast cell-deficient Cpa3 Cre/+ mice. The Cpa3 Cre/+ strain is constitutively devoid of mast cells in connective and mucosal tissues and it has no known alterations in other cell lineages involved in bone repair. In WT but not in Cpa3 Cre/+ mice mast cells appeared in the repair tissue from 14d to 56d PO. Interestingly, bridging of the bone defect was complete in all (6/6) WT mice at 56d PO but only 3/8 Cpa3 Cre/+ mice. This incomplete bridging was associated with disruption of re-vascularization and impaired bone mineralization. Osteoclast activity was reduced in Cpa3 Cre/+ mice in the early phase of repair but increased at later stages, with no clear differences in macrophage activity. Taken together, the results indicate that mast cells have a positive impact on bone repair that is mediated in part by recruitment of vascular endothelial cells, as also suggested by the previous work of Boesiger et al (1998) [21], and in part by altered metabolism of newly formed bone. It was proposed more than two decades ago that mast cells are involved in tissue digestion and revascularization, which are early and essential steps in the bone healing cascade [9]. Mast cells reside over the long term in connective tissues where they are available locally and for potential trafficking via the vascular and lymphatic systems to sites of tissue injury and repair [22]. The surgical intervention used in this study would have temporarily disrupted existing hind limb vessels and caused local ischemia and hypoxia, which are the major stimuli for re-vascularization [23]. Mast cells, identified by the use of metachromatic staining with aTB, were first seen in WT bone at 5d PO, which was the earliest time at which the soft callus could be preserved intact for histological analyses. At this time they were localized adjacent to vessels in muscle, in bone marrow of the femur proximal to the site of injury, and at the periosteal junction between soft tissue and bone. The appearance of aTB positive cells at the proximal, rather than distal, end of the femur suggests they migrated to the wound from soft tissue stores via the arterial or lymphatic vessels that supply the hind limb. This conjecture was supported by micro CT analyses of vessels in regenerating bone showing initiation of re-vascularization at 5d PO at the proximal end of the defect in WT bones. Quantitative analyses revealed peak numbers of aTB positive mast cells in the defect/medulla of WT bones at 14d PO, and in the cortex at 28d PO. This timeframe was supported by qualitative data showing MC tryptase positive cells in bone marrow starting at 14d in WT but not Cpa3 Cre/+ mice. Peak numbers of mast cells in regenerating bone coincided with peak numbers of vessels visualized by micro CT and with CD34 immunostaining of vascular endothelial cells. Degranulation or secretion of factors by aTB positive mast cells would result in local release of angiogenic mediators like heparin, angiogenin, vascular endothelial growth factor and matrix metalloproteinases in WT mice [24]. Given their lack of mast cells these mediators would be absent in Cpa3 Cre/+ mice, which could have accounted for the delay and disorganization of bone re-vascularization as evidenced by micro CT. In this and other studies of bone repair we have used CD34 as a sensitive marker of endothelial cells, but it is also expressed on MSC, fibrocytes and other precursor cells capable of differentiating down the osteogenic lineage [25]. In the current work CD34 cells were clearly organized into vascular channels in the condensed mesenchyme in defect/medulla and the cortex at 5d PO prior to bone formation. Consistent with delayed healing, CD34 cells persisted in the Cpa3 Cre/+ mice in the periosteum and in residual fibrous tissue in the defect/medulla at 56d PO, when re-vascularization was effectively complete in WT mice. The absence of clearly defined channels and the location of CD34 positive cells at 56d PO suggest the cells were not endothelial cells but rather fibrocytes in the periosteum, or MSC that were unable to differentiate into bone-forming osteoblasts in the absence of mast cell mediators. The current literature identifies bone active agents like platelet derived growth factor, fibroblast growth factors, transforming growth factor and tumor necrosis factor, as well as matrix metalloproteinases as pre-formed components of stored granules that are released upon mast cell activation [22,26]. After 14d PO mast cells were present in close proximity to vessel in the defect/medulla of WT mice but absent from Cpa3 Cre/+ mice. The absence of mast cell derived bone active agents could have impaired mineralization of newly formed bone at 14d and 28d PO despite similar levels of alkaline phosphatase activity in osteoblasts (Figs 6 and 7). Furthermore, the absence of carboxypeptidase A, which is a major constituent of mast cell granules that targets the vasoconstrictor endothelin [27,28] could have resulted in excess endothelin-1 mediated vasoconstriction, increased oxidative stress and altered bone cell function [28].
The importance of an adequate periosteal reaction to bone healing is emphasized by the current widespread use of vascularised fibular periosteal grafts to promote healing of large bone defects [29,30] and in cell-based bone tissue engineering [31]. Prior to skeletal maturity the periosteum is thick and highly vascular, with active osteoblasts depositing intramembranous bone to increase the external diameter of growing long bones [32]. The absence of mast cells in Cpa3 Cre/+ mice resulted in thickening of the periosteum, similar to that seen in adult FGFR3-null mice [33], and suggests an imbalance in FGF signaling might be involved in impaired bone regeneration in the Cpa3 Cre/+ mice. However, the similarity in ALP activity between WT and Cpa3 Cre/+ mice supports the hypothesis that the impairment was not mediated at an early stage of osteoblast differentiation.
The two principle cells responsible for catabolic activity in bone are osteoclasts and macrophages. Apart from being phagocytes the lineages appear to share little in common. Osteoclast activity is restricted to digestion of mineralized tissue in cartilage and bone whereas macrophages have been implicated in inflammatory disorders and fibrosis of most if not all major organs, including the brain. In our previous work that characterized bone repair in Kit W-sh mast cell deficient mice osteoclast activity was shown to be elevated throughout the bone healing cascade [18]. This was not the case in the current study in which osteoclast activity was reduced in Cpa3 Cre/+ mice up to 14d PO and elevated thereafter compared with WT mice. A simple explanation for the discrepancy is that Kit is expressed on osteoclasts and their precursors, as well as mast cells, whereas Cpa3 expression is restricted to mast cells. Osteoclasts are considered to be tissue specific macrophages, distinct from bone marrow macrophages and "osteomacs" located on the periosteal and endosteal surface of resting bone [34]. Using F4/80 immunostaining to identify all macrophages in healing bone it was surprising to find the lowest levels of expression during the inflammatory phase of repair at 5d PO, with an approximate 5-fold increase thereafter in both WT and Cpa3 Cre/+ mice. The relatively low abundance of macrophages under healing conditions in the absence of mast cells in Cpa3 Cre/+ mice in the current study, and previously in Kit W-sh mice, suggests an interdependence between the two lineages. This conjecture is further supported by the increase in both mast cells and macrophages in bone defects in WT mice administered lipopolysaccharide (LPS) [35]. The timeframe and pattern of distribution of F4/80 positive cells in the current study closely resembled that of "osteomacs" in a similar model of cortical bone repair [36]. Given the proposed anabolic function of these cells in bone mineralization [37], their relative short supply in Cpa3 Cre/+ bone might explain the increase in osteoid compared with WT bone at 14d and 28d PO. Taken together the data suggest a functional relationship exists between mast cells, macrophages and MSC in the bone micro-environment that warrants further investigation. Only few suggested mast cell functions have been confirmed in studies using Kit-independent models of mast cell deficiency. Thus, the impairment of bone fracture healing in mice in the absence of mast cells appears to be a remarkable, largely unanticipated, non-immunological mast cell function.
Supporting information S1 File. Bone repair in mast cell-deficient mice.xlsx. Dataset to the article. (XLSX)