Bone marrow cell homing to sites of acute tibial fracture: 89Zr-oxine cell labeling with positron emission tomographic imaging in a mouse model

Background Bone fracture healing is dependent upon the rapid migration and engraftment of bone marrow (BM) progenitor and stem cells to the site of injury. Stromal cell-derived factor-1 plays a crucial role in recruiting BM cells expressing its receptor CXCR4. Recently, a CXCR4 antagonist, plerixafor, has been used to mobilize BM cells into the blood in efforts to enhance cell migration to sites of injury presumably improving healing. In this study, we employed zirconium-89 (89Zr)-oxine-labeled BM cells imaged with positron emission tomography (PET)/computed tomography (CT) to visualize and quantitate BM cell trafficking following acute bone injury and to investigate the effect of plerixafor on BM cell homing. Unilateral 1-mm incisions were created in the distal tibia of mice either on the same day (d0) or 24 h (d1) after 89Zr-oxine-labeled BM cell transfer (n = 4–6, 2–2.3 × 107 cells at 9.65–15.7 kBq/106 cells). Serial microPET/CT imaging was performed and migration of 89Zr-labeled cells to the bone injury was quantified. The effects of three daily doses of plerixafor on cell trafficking were evaluated beginning on the day of fracture generation (n = 4–6). The labeled cells localizing to the fracture were analyzed by flow cytometry and immunohistochemistry. Results In d0- and d1-fracture groups, 0.7% and 1.7% of administered BM cells accumulated within the fracture, respectively. Plerixafor treatment reduced BM cell migration to the fracture by approximately one-third (p < 0.05 for both fracture groups). Flow cytometry analysis of donor cells collected from the injured site revealed a predominance of CD45+ stem/progenitor cell populations and subsequent histological analysis demonstrated the presence of donor cells engrafted within sites of fracture repair. Conclusion 89Zr-oxine labeling enabled visualization and quantitation of BM cell recruitment to acute fractures and further demonstrated that plerixafor plays an inhibitory role in this recruitment. Electronic supplementary material The online version of this article (10.1186/s13550-018-0463-8) contains supplementary material, which is available to authorized users.


Background
Bone healing is a multistep physiological process that begins immediately following injury. Bone fracture triggers an immunological cascade beginning with signaling by pro-inflammatory cytokines such as interleukin 1 (IL-1), IL-6, and tumor necrosis factor-α (TNF-α), secreted by macrophages and periosteal cells [1][2][3], which, in turn, recruit inflammatory and fibrogenic cells to the injury site [4]. Bone remodeling and healing are preceded by angiogenesis, chondrogenesis, and osteogenesis. These processes are contingent on the migration and proliferation of hematopoietic stem cells (HSCs), endothelial progenitor cells (EPCs), and periosteum-derived mesenchymal stem cells (MSCs) from the bone marrow (BM) or circulation [5]. The blood vessel walls and nearby soft-tissues also act as sources of MSCs [6][7][8]. For the migration of HSC, MSC, and EPC and subsequent homing to the fracture region, stromal cell-derived factor-1 (SDF-1) has been reported to play a crucial role. The interaction of SDF-1 in the injured stroma with CXCR4 expressed on these stem cells results in the accumulation of these cells at the fracture site [5,9,10].
Investigators have examined blockade of CXCR4 to mobilize stem and progenitor cells to enhance their recruitment to the fracture and to promote fracture healing. Some studies report that increasing the number of mobilized stem and progenitor cells facilitates the repair process [11][12][13]. Others, however, report opposite results [14,15]. For example, mice treated with either SDF-1 neutralizing antibody or TF14016, a CXCR4 antagonist, demonstrated decreased stem cell engraftment and bone formation [15]. These data suggest that CXCR4 inhibition and subsequent mobilization of CXCR4 + cells may, or may not, improve bone injury healing by releasing high number of stem/progenitor cells into the circulation, thus making more available for localization within the injury site. To address this question, radiolabeled cells can be used to track and quantify recruitment of stem/progenitor cells to the bone injury with and without CXCR4 blockade. We have previously reported a method of cell labeling using 89 Zr-oxine followed by positron emission tomography (PET) that can label various cell types without altering their cellular viability or function [16]. Using this technology, we have successfully visualized BM cell migration in murine models of BM transplantation [17]. Additionally, it was demonstrated that a CXCR4 inhibitor, plerixafor, mobilized these 89 Zr-oxine-labeled BM cells into the circulation [17]. To explore the migration of BM cells to sites of bone injury, we employed 89 Zr-oxine cell labeling of BM cells and PET imaging in bone injury model. This method enabled the quantification of the number of cells migrating to the fracture site. We employed two tibial injury/BM cell transfer experimental schemes; in one model, mice received a radiolabeled BM cell transfer followed 1 day later by incisional injury of the tibia. In the second model, BM cell transfer was performed immediately after the tibial injury. Using these two schemes, we examined the trafficking of BM cells and the effect of plerixafor on the short-term BM cell homing to the bone injury.

Methods
Mice C57BL/6 wild type and green fluorescence protein (GFP) transgenic mice were purchased from Jackson Laboratories (Bar Harbor, ME).

Fracture generation
All procedures were approved by the institutional animal care and use committee and were compliant with the humane use of animals. A fracture of the tibia was created as previously described [18] with minor modifications. In brief, the mice were anesthetized with isoflurane inhalation and the left hind limb surface was soaked with 10% Povidone-iodine solution (Purdue Pharma L.P., Stamford, CT). Using a surgical scalpel (WESTNET, Inc., Canton, MA), an approximately 1-mm-long incision was created in the medial side of the left distal tibia without cutting entirely through the bone. In some mice, the incision was made in the soft-tissue without reaching the tibia. Mice remained mobile after the surgery.
BM cell labeling with 89 Zr-oxine BM cells were flushed from femurs and tibias of donor mice. Red blood cells were lysed using ammoniumchloride-potassium (ACK) lysis solution (Thermo Fisher Scientific Inc., Waltham, MA) and the BM cells were washed in phosphate-buffered saline (PBS). Cell labeling with 89 Zr-oxine complex was performed to achieve the labeling doses optimized for the BM cells in the previous study [17]. Cells were incubated with 89 Zr-oxine complex at 45 kBq/10 6 cells in PBS at 30:1 cell suspensionto-89 Zr-oxine solution volume ratios for 20 min at room temperature. Labeled cells were washed twice in RPMI culture media (Life Technologies, Grand Island, NY) supplemented with 10% fetal calf serum (Gemini Bio Products, Sacramento, CA), 100 IU/mL penicillin, 100 μg/mL streptomycin (Life Technologies), and 50 μM 2-mercaptoethanol (Sigma-Aldrich, St. Louis, MO). 89 Zr-oxine-labeled BM cells were transferred to mice intravenously (2-2.3 × 10 7 cells at 9.65-15.7 kBq/10 6 cells). BM cells collected from wild type were used unless specified that GFP transgenic mice were used as donors.
MicroPET/CT imaging of transferred 89 Zr-oxine-labeled BM cells One group of mice received 89 Zr-oxine-labeled BM cell administration followed 1 day later by tibial injury (d1fracture group). MicroPET/CT imaging (BioPET, Bioscan, Washington, DC) was performed 1 day after the cell transfer, before and after the injury, and on days 2 and 5. Another group of mice received 89 Zr-oxine-labeled BM cells immediately after the tibial injury (d0-fracture group) followed by microPET/CT imaging on days 0, 1, 2, and 5. A 5-min emission PET scan per bed position was performed for a total of two bed positions on day 0, using a 400-700 keV energy window. On days 1, 2, and 5, scan time per bed position was increased to 6, 7.5, and 12.5 min, respectively, to account for radioactive decay. The acquired images were reconstructed using a three-dimensional ordered-subsets expectation maximization algorithm. VivoQuant software (inviCRO LLC, Boston, MA) was used to fuse the maximum intensity projection PET images with CT images.
Quantitation of 89 Zr-oxine-labeled BM cell migration on microPET/CT images Volumes of interest were drawn on the acquired images using VivoQuant and MIMvista (MIM software, Cleveland, OH) software. The percentage of injected cells migrated to the injury and organs were obtained from the quantitated radioactivity in the tissue corrected for decay divided by the injected dose. For the tibial fracture, corresponding contralateral tibia was used as a control. Signal-to-background ratios of the fracture or the contralateral control were calculated from maximum standardized uptake value (SUV) of the fracture or contralateral control against mean SUV of the muscle. For the bone marrow uptake, the skull, forelimbs distal to the shoulder joint, and fracture site were excluded from the quantitation.

Mobilization of BM cells by plerixafor
D1-fracture mice received an intravenous injection of 5 mg/kg of plerixafor (Adooq Bioscience, Irvine, CA) at days 1-3 after the BM cell transfer (d1-fracture-plerixafor group). For d0-fracture mice, plerixafor was injected on day 0 at the time of BM cell transfer, and on days 1 and 2 for a total of 3 doses (d0-fracture-plerixafor group).

Flow cytometry analysis of BM cells migrated to the fracture sites and spleen
Thirty-three million BM cells collected from GFP transgenic mice were transferred to wild-type hosts, in the d1-and d0-fracture models. Two days after cell transfer, BM cells were flushed and collected from the fractured section of tibia and the corresponding section of the contralateral non-fractured tibia. A 70 μm filter was used to create a single cell suspension. The femur of the contralateral side was also isolated and BM cells were collected. Spleen was harvested, homogenized, and passed through a 70 μm filter; red cells were lysed using ACK lysis solution. Collected cells were stained with antibodies against CD45, CD29, CD105, CD11b, Ly6C, and Ly6G or with lineage markers (CD3, NK1.1, Ly6G, CD2, CD5, B220). All antibodies were purchased from eBioscience (San Diego, CA). Data collected by a flow cytometer (FACSCalibur, Becton Dickinson, San Jose, CA) was analyzed using FlowJo software (Tree Star, Inc., Ashland, OR).

Histological analysis
89 Zr-oxine-labeled GFP + BM cells were transferred to wild-type recipient mice and fracture was generated 1 day later. Tibias were removed from both fracture and non-fractured contralateral control sites at 1.5 and 7 weeks after the cell transfer and placed in 4% paraformaldehyde in PBS. Tibias were then embedded in paraffin and sectioned into 5-μm-thick slices. Following deparaffinization and rehydration, the slides were stained with hematoxylin and eosin (H&E). Immunohistochemistry staining was performed with an anti-GFP antibody conjugated with horseradish peroxidase (Life Technologies) followed by 3,3′-diaminobenzidine staining (GFP-IHC). Tibias from non-fractured wild-type and GFP transgenic mice were processed as controls.

Statistical analysis
Two-way repeated measure analysis of variance (ANOVA) with Sidak's multiple comparisons test was used to analyze the difference in BM cell accumulation at the fracture site compared to non-fractured tibia. Mann-Whitney test was used for comparing the peak BM cell accumulation values at the fracture with and without mobilization, and for analyzing the flow cytometry results between fractured-and non-fractured-tibias. P value, with adjustment for multiple comparison when comparing more than two groups, less than 0.05 was considered significant.

BM cells labeled with 89 Zr-oxine show rapid homing to bone marrow and bone injury site
We first tracked 89 Zr-oxine-labeled BM cells transferred to mice without a fracture, as a control, by microPET/CT imaging beginning 1 day after the cell transfer (n = 4). In normal mice, 89 Zr-labeled BM cells were observed in the BM, spleen, and liver (Fig. 1a). When a group of mice underwent bone incision 1 day after the labeled BM cell transfer (d1-fracture group, n = 4), microPET/CT imaging showed redistribution of 89 Zr-labeled BM cells; the cells accumulated at the injury site within 24 h of the injury (2 days after cell transfer). The cells remained at the site until the end of the 5-day scanning period (Fig. 1b). To ascertain how the bone fracture affected the trafficking of circulating BM cells, a fracture was introduced and BM cells were immediately transferred (d0-fracture, n = 4). The BM cells homed to the BM, spleen, and liver, as well as to the fracture site starting 1 day after the fracture/cell transfer (Fig. 1c). Again, the BM cells remained at the injury at least until the end of the 5-day imaging period. Quantitation of the radioactivity at the fracture site (Fig. 2a) revealed that approximately 1.7 ± 0.5% and 0.7 ± 0.04% of the injected labeled cells had migrated into the bone injury on day 2 for d1-and d0-fracture groups, respectively (Fig. 2b,  c). 89 Zr-oxine-labeled cell accumulation at the injury was significantly higher than that at the non-fractured contralateral tibia in both groups (p < 0.0001 and p < 0.001 on days 2 and 5, respectively, in d1-fracture group; p < 0.0001 on days 1, 2, and 5 in d0-fracture group).
Due to the very low background signals in the recipients, bone injury-to-background SUV ratios reached as high as 345 ± 114 and 250 ± 86 on day 2 in the d1and d0-fracture groups (Additional file 1: Figure S1a, b). Cell distribution in the bone marrow, liver, and spleen remained fairly plateaued during the 5-day imaging period (Additional file 1: Figure S1c, d).

Mobilization of BM cells reduces cell migration to the fracture
To explore whether the SDF-1-CXCR4 system plays an important role in fracture healing by helping to mobilize BM cells from their niche into the peripheral blood, we investigated the effect of plerixafor on the accumulation of labeled cells at the injury site. In the d1-fracture-plerixafor group, microPET/CT images indicated accumulation of 89 Zr-oxine labeled-BM cells at the bone injury site under plerixafor injections (Fig. 3a, n = 4). By contrast, plerixafor administration reduced the accumulation of the circulating BM cells to the bone injury (d0-fracture-plerixafor groups, n = 4) and a very low 89 Zr signal was observed at the site (Fig. 3b). Quantitation of the images indicated that 89 Zr-labeled BM cells accumulated at the injury significantly more than the contralateral tibia in the d1-fracture-plerixafor group (Fig. 3c, p < 0.001 on day 2, p < 0.01 on day 5), but in d0-fracture-plerixafor group, BM cells accumulation at the injury was only slightly higher than the contralateral side (Fig. 3d. p < 0.05 on days 1 and 2). Plerixafor administrations significantly reduced the peak BM cell accumulation at the injury by 40% and 33% in d1-and d0-fracure-plerixafor groups, respectively: the peak was observed on day 2 in d1-fracture-plerixafor group (Fig.  3e, p < 0.05, n = 6) and on day 1 or day 2 in d0-fractureplerixafor group (Fig. 3f, p < 0.05, n = 6).

Soft tissue injury alone does not induce BM cell accumulation at the fracture site
We then evaluated the effect of a soft tissue injury alone on the distribution of BM cells. 89 Zr-oxine-labeled BM cells were transferred to the mice immediately after injury (n = 4). As shown in Fig. 4, the labeled BM cells did not accumulate in the soft tissue injury site. This suggests that injuries to the bone are required to induce BM cell migration to the injured site.

Transferred BM cells with hematopoietic cell markers are detected at the site of fracture
Both hematopoietic and mesenchymal cell types are reported to be involved in fracture healing [15,19,20]. Therefore, we examined which subset of BM cells had localized to the fracture using GFP + donor cells in both d1-and d0-fracture groups (n = 4). Almost all of the GFP + BM cells detected at the fractured area, as well as contralateral control tibia, femur, and spleen, were hematopoietic (CD45 + ) cells (Fig. 5a), containing undifferentiated (lineage marker negative: Lin − ) cells (Fig. 5b). Neither the CD45 + cell fraction nor the Lin − cell fraction showed statistical differences between fractured-vs Fig. 2 Quantitation of microPET imaging reveals BM trafficking kinetics to the bone injury site. The percentage of 89 Zr activity accumulated at the fracture site was quantified by setting a volume of interest on the acquired PET/CT images (a) and correcting for radioactive decay. As an internal control, 89 Zr activity in the contralateral tibia was quantified. The BM cells highly accumulated at the fracture 1 day after the fracture generation in both the day 1-fracture (b) and day 0-fracture (c) schemes and gradually decreased thereafter. The BM cells preferentially accumulated at the fracture site compared to the contralateral control in both fracture models 1 day after the fracture generation (n = 4 in each group, ****p < 0.0001 and ***p < 0.001 with two-way repeated measure analysis of variance (ANOVA) with Sidak's multiple comparisons test. The arrows indicate the timing of bone injury generation) Fig. 3 Mobilization of BM cells using plerixafor reduces cell migration to the site of fracture. a The regimen of d1-fracture with a 3-day administration of plerixafor starting on day 1 and microPET/CT images are shown. 89 Zr-oxine-labeled BM cells trafficked to the fracture site and the accumulation was observed on day 2 through day 5 (n = 4). b The regimen of d0-fracture with a 3-day administration of plerixafor starting on day 0 and microPET/CT images are shown. The 89 Zr-oxine-labeled BM cells were transferred immediately after the fracture generation and administration of the first dose of plerixafor. Weak 89 Zr signal was observed at the injury site from day 1 through day 5 (n = 4). c Quantitation of the d1-fracture-plerixafor PET images indicate that the percentage of 89 Zr-labeled BM cells accumulated at the fracture was significantly higher than in the contralateral non-fractured tibia (n = 4). d Quantitation of the d0-fracture-plerixafor PET images show a slightly higher 89 Zr-labeled BM cells accumulation at the fracture compared to the contralateral tibia (n = 4). (ns not significant, *p < 0.05, **p < 0.01, ***p < 0.001, two-way repeated measure ANOVA with Sidak's test. The arrows indicate the timing of bone injury generation). e The peak accumulation on day 2 was reduced by plerixafor compared to non-mobilized mice in the d1-fracture-plerixafor scheme (n = 6). f Plerixafor significantly reduce the peak accumulation, on day 1 and 2, compared to the non-mobilized mice in the d0-fracture-plerixafor scheme (n = 6). (*p < 0.05 with Mann-Whitney test) non-fractured tibias (Fig. 5c, d). In the analysis of non-hematopoietic cells including MSCs, d1-fracture site showed slightly higher CD29 + CD105 + cells within CD45 − donor cells than the contralateral control (Fig.  5e), but this was not statistically significant. As fracture is known to attract various inflammatory cells during the acute phase, we evaluated for monocytic and granulocytic myeloid cells in the injury region. Interestingly, we found that CD11b + Ly6C intermediate Ly6G + granulocytic myeloid cells rather than CD11b + Ly6C + Ly6G − monocytic myeloid cells migrated into the fracture (Fig. 5f, p < 0.05 in d1-fratured tibia vs control tibia, and Fig. 5g).

BM cells accumulated at the bone injury engraft and participate in fracture healing
Finally, we asked if the BM cells migrated to fracture were involved in fracture healing. To this end, histological sections of fractured and non-fractured contralateral tibias were prepared from fractured mice transferred with 89 Zr-oxine-labeled GFP + BM cells. H&E staining indicated inflammatory cells and fibrous tissue surrounding the fracture, and GFP-IHC demonstrated accumulation of GFP + cells in the fracture callus at 1.5 weeks (Fig.6a, b). At 7 weeks, the hard callus was formed and GFP + cells were present (Fig. 6c, d). In the non-fractured contralateral side, we observed fat marrow with negligible GFP + cells at day (Fig. 6e, f). A non-fractured tibia of GFP-transgenic and wild-type mouse that did not receive BM cell transplant is shown as controls (Fig. 6g, h, respectively). These histological observations support the concept that BM cells recruited to sites of injury are incorporated into the repair process.

Discussion
The ability to radiolabel hematopoietic cells without interfering with their viability, chemotaxis, or ability to differentiate is a powerful method to understand the migration of such cells in vivo under a variety of conditions [16,17]. Because PET cameras are more efficient than gamma cameras, it is possible to reduce the radioactivity needed to visualize hematopoietic cells in the body thereby causing less damage to the cells. In this study, 89 Zr-oxine BM cell labeling was used to examine the migration of BM cells to an acute tibial injury. Based on the previous extensive 89 Zr-oxine radiotoxicity evaluations on the BM cells [17], 89 Zr-oxine labeling doses were carefully controlled to be within the optimized range to avoid altering viability, proliferation, and more importantly, differentiation function of the cells. 89 Zr was stably retained with the BM cells [17]. Two scenarios were employed: d1-fracture in which BM cell transfer preceded the fracture by 1 day and d0-fracture in which the fracture preceded the transfer of BM cells. With the d1-fracture regimen, we examined the effect of the fracture on the bone marrow cells already distributed to the organs and reached the steady state. The d0-fracture regimen, on the other hand, addressed the effect of fracture on the circulating bone marrow cells. In both conditions, 89 Zr-oxine-labeled cells accumulated at the site of tibial injury, peaking by approximately 24 h at 0.7-1.7% of the administered cells. While representing a minority of injected cells, this small number of cells is readily detected on PET imaging (Figs. 1b, c and 3a, b) with extremely high signal-to-background SUV ratios (Additional file 1: Figures S1a, b and S2a, b). As about 17% of the transferred cells migrated to the bone marrow (Additional file 1: Figures S1c, d and S2c, d), the percentage of cells accumulated at the injury was equivalent to as high as 4-10% of the cells in the bone marrow. After reaching the peak accumulation, the BM cells at the fracture gradually decreased. This is likely because the death of some cells that failed to engraft, even if they migrated to the injury sites. The distribution of the cells to the BM, spleen, and liver, observed within 24 h in Fig. 4 Soft-tissue injury does not attract the BM cells. a A soft-tissue injury in the distal lower leg overlying the distal tibia was created and 89 Zr-labeled BM cells (2 × 10 7 cells at 9.65-21.3 kBq/10 6 cells) were transferred intravenously. MicroPET/CT images were acquired on days 0, 1, 2, and 5 (n = 4). b Soft-tissue incision without bone fracture did not recruit BM cells to the injury site. The arrow indicates the timing of soft-tissue injury generation both fracture models, was almost identical to that in non-fractured control mice.
It has been shown that the BM stromal cells and bone periosteal cells release SDF-1 from the periosteum in response to bone injury [9,[21][22][23]. SDF-1, along with other key cytokines, play an important role in recruiting various hematopoietic and mesenchymal stem/progenitor cell types to the injury site in order to facilitate tissue repair and regeneration [24][25][26][27][28][29][30]. SDF-1 expression increases shortly after injury [9,29,31,32], leading to the recruitment of cells expressing CXCR4. Not surprisingly, in both the d1-and d0-fracture groups, we observed the rapid migration of BM cells to the tibial injury within 24 h (Fig. 1b, c) and the radioactivity remained relatively stable thereafter for up to 5 days of observation. The distal tibia was chosen as the site for injury because it is not associated with hematopoietic marrow, making it less likely a homing region for transplanted BM cells. This BM cell accumulation shown at the bone injury site was not observed in similar contralateral non-injured Flow cytometry analysis demonstrates that the donor BM cells homing to the fracture site are mostly hematopoietic cells stem/progenitor cells and inflammatory cells. a BM cells from GFP transgenic mice were used as donor cells in d1-and d0-fracture models. One day later, cells were harvested from the fractured section of the tibia, corresponding contralateral tibia section, BM from non-fracture side femur, and spleen (n = 4). Flow cytometry analysis indicated that almost all the GFP + cells found in these tissues expressed CD45 (representative data). b The fraction of Lin − cells in GFP + donor cells is indicated (representative data). c-g Comparison of cell fractions found in the sections of fractured vs nonfractured tibia is shown. CD45 + cells (c) and Lin − cells (d) within the donor cells, CD29 + CD105 + cells within non-hematopoietic donor cells (e), Ly6C intermediate (int) Ly6G + cells (f), and Ly6C + Ly6Gcells (g) within CD11b + donor cells were analyzed. Ly6C int Ly6G + cells migrated to the fracture significantly higher than the non-fractured tibia (n = 4 in each group, *p < 0.05 with Mann-Whitney test) areas nor in mice that received only superficial injury to the soft-tissue overlying the tibia (Fig.4), supporting the model of SDF-1 production by periosteum after the bone injury [21,22]. This suggests that bone injury induces a unique cell signaling cascade that recruits bone marrow cells to the injury to promote healing. In this regard, the amount of the bone marrow cells recruited to the injury may be affected by the size and depth of the bone injury.
Multiple groups have reported stem and progenitor cell recruitment, engraftment, and differentiation into bone-forming cells following CXCR4 blockade and subsequent mobilization of CXCR4 + cells from the BM niche into the circulation [11,13,33,34]. We have previously demonstrated mobilization of 89 Zr-oxine-labeled BM cells into peripheral blood by a CXCR4 blocker, plerixafor, using both PET imaging and flow cytometry analysis [17]. In the case of acute bone injury, it is thought that stem cells mobilized from the BM niche by plerixafor migrate to the fracture site in greater numbers. In both d1-and d0-fracture models, we observed an overall inhibition in BM cell uptake at the bone injury site, by as much as 33-40% by day 2, after plerixafor (Fig. 3e, f ). It is critical that CXCR4 + cells migrate and engraft using the SDF-1 chemokine gradient secreted by stromal cells in the injured bone periosteum.
Thus, it appears that while CXCR4 blockade with plerixafor released more BM cells into the circulation, it also inhibited BM cell chemotaxis to the fracture site, leading to a net decrease in cell accumulation. These findings support previous reports suggesting that plerixafor acts to block CXCR4 + cell migration toward SDF-1 producing stromal cells during the acute phase of fracture healing and, therefore, may be detrimental [35].
Because labeled donor cells constituted a small fraction of the progenitor cell population in comparison to the native host cells, it is very likely that the donor cells in the d0-fracture mice were just beginning to make their way into the arterial circulation to home to their eventual tissue destinations, when endogenous CXCR4 + cells were mobilized to home to the fracture site. Thus, this model likely underestimates the actual accumulation of BM cells at the fracture site and Fig. 6 GFP + donor cells are detected in the bone fracture. a, b BM cells from GFP transgenic mice were transferred to wild-type mice immediately after tibial fracture and 1.5 weeks later, the fractured tibia section and corresponding contralateral tibia section were harvested. Tissues slices were stained with H&E and with GFP-IHC. Microscopic images of fractured tibia indicated that the fractured bone was surrounded by inflammatory tissue (a, b, left, H&E) and GFP + donor cells were engrafted around the fracture (a, b, right, GFP-IHC). c, d Tissue sections created at 7 weeks following the bone marrow cell transfer indicated callus formation with GFP + cells in the surrounding soft-tissue (left: H&E, right: GFP-IHC). e, f Contralateral non-fractured tibia at 1.5 weeks after the cell transfer is shown. g, h Tibia from a GFP + (g) and a wild-type (h) mice are shown as controls. Scale bar indicates 200 μm in (a, c) and (e), and 100 μm in (b, d, f, g) and (h) would explain why we observed a reduced donor cell uptake in the d0-fracture mice compared to the d1-fracture mice.
MSCs, primarily osteoblast progenitor cells, and as well as hematopoietic cells have been shown to migrate to the site of fracture during bone repair [19][20][21]36]. In order to determine which BM cell types homed to the fracture, we performed flow cytometry 2 days after the cell transfer, corresponding to the peak accumulation of cells. Cells were isolated from a small section of tibia at both the fracture site and the contralateral normal site, as well as the normal femur and the spleen. GFP served as a marker for the donor cells. Across these tissue types, almost all of the donor cells were CD45 + hematopoietic cells; however, CD45 − CD29 + CD105 + cells that are likely to be MSCs or endothelial cells were also present (Fig. 5). We also observed preferential migration of granulocytic myeloid cells to the fracture in the d1-fracture model (Fig.  5f). In these flow cytometry analyses, it is possible that the specificity of cell types migrating to the fracture was significantly underestimated due to the inclusion of surrounding non-fractured tibia, causing a large "dilution" effect. We only analyzed the presence of donor cells early after their transfer, and thus, there may be some donor cell types that migrate to the fracture later and therefore, were not captured by our analysis. In addition, flow cytometry analysis after 6-7 weeks of healing, when the cells would have been fully differentiated, may reveal greater differences in marker expression on donor cells in the fracture site vs. contralateral site.
Finally, we did confirm that donor cells successfully engrafted in the healing fractured-tibia using immunohistochemistry (Fig. 6). Ten days after the fracture, the donor cells contributed to the inflammatory tissue formation which eventually enters the soft callus phase. At 7 weeks, the GFP + cells were still detectable around the callus at the fracture site.

Conclusions
BM cell 89 Zr-oxine-labeling with microPET/CT imaging revealed that acute fracture results in the redistribution of BM cells to the fracture within 24 h. Our data strongly suggests that BM mobilization occurs rapidly after fracture and that hematopoietic cells are the first cells to arrive at the bone fracture. Furthermore, our study indicated that CXCR4 blockade negatively affected BM cell migration toward the fracture site at least in the early phase of fracture healing. PET imaging enabled visualization and quantitation of in vivo BM cell trafficking and homing to various tissues, including the bone fracture site, making it a useful tool in understanding the bone marrow response to acute bone fracture.

Additional file
Additional file 1: Figure S1. BM cells accumulated at the bone injury show high 89 Zr signals by microPET/CT imaging. Figure