Identification of circulating murine CD34+OCN+ cells

ARTICLE IN PRESS Cytotherapy, 2018; 000: 1 10

Identification of circulating murine CD34+OCN+ cells

RYAN R. KELLY1,2, LINDSAY T. MCDONALD1,2, VINCENT D. PELLEGRINI3, JAMES J. CRAY5 & AMANDA C. LARUE1,2,4 1

Research Services, Ralph H. Johnson Department of Veterans Affairs Medical Center, Charleston, SC, USA, Department of Pathology and Laboratory Medicine, Medical University of South Carolina, Charleston, SC, USA, 3 Department of Orthopedics, Medical University of South Carolina, Charleston, SC, USA, 4Hollings Cancer Center, Medical University of South Carolina, Charleston, SC, USA, and 5Division of Anatomy, The Ohio State University, Columbus, OH, USA 2

Abstract Background aims: Previous studies identified a circulating human osteoblastic population that expressed osteocalcin (OCN), increased following fracture and pubertal growth, and formed mineralized colonies in vitro and bone in vivo. A subpopulation expressed CD34, a hematopoietic/endothelial marker. These findings led to our hypothesis that hematopoietic-derived CD34+OCN+ cells exist in the circulation of mice and are modulated after fracture. Methods: Flow cytometry was used to identify CD34+OCN+ cells in male B6.SJL-PtprcaPepcb/BoyJ and Vav-Cre/mTmG (VavR) mice. Non-stabilized tibial fractures were created by three-point bend. Fractures were longitudinally imaged by micro-computed tomography, and immunofluorescent staining was used to evaluate CD34+OCN+ cells within fracture callus. AMD3100 (10 mg/kg) was injected subcutaneously for 3 days and the CD34+OCN+ population was evaluated by flow cytometry. Results: Circulating CD34+OCN+ cells were identified in mice and confirmed to be of hematopoietic origin (CD45+; Vav1+) using two mouse models. Both circulating and bone marrow-derived CD34+OCN+ cells peaked three weeks post-non-stabilized tibial fracture, suggesting association with cartilage callus transition to bone and early mineralization. Co-expression of CD34 and OCN in the fracture callus at two weeks post-fracture was observed. By three weeks, there was 2.1-fold increase in number of CD34+OCN+ cells, and these were observed throughout the fracture callus. AMD3100 altered CD34+OCN+ cell levels in peripheral blood and bone marrow. Discussion: Together, these data demonstrate a murine CD34+OCN+ circulating population that may be directly involved in fracture repair. Future studies will molecularly characterize CD34+OCN+ cells, determine mechanisms regulating their contribution, and examine if their number correlates with improved fracture healing outcomes.

Key Words: circulating osteoprogenitor, fracture, hematopoietic stem cell, mesenchymal stromal cell, mobilization

Introduction Current paradigms suggest that mesenchymal stromal cells (MSCs) are bone marrow (BM) precursors to osteoblasts, whereas hematopoietic stem cells (HSCs) generate osteoclasts. However, this dogma has been challenged by findings that HSCs exhibit multilineage plasticity [1] and studies demonstrating that MSCs act in supportive roles after injury, prompting a renaming of MSCs to “medicinal signaling cells” [2,3]. Thus, stem cell populations besides MSCs may facilitate osteogenesis. Previous studies in our laboratory have demonstrated the ability of the HSC to generate osteochondrogenic cells [4]. Therefore, examination and characterization of

HSC-derived osteoprogenitor populations, including circulating osteoprogenitors, warrant study for contributions during bone repair. Circulating osteoprogenitors are receiving increased attention for potential in bone healing applications. Current evidence for circulating osteoprogenitors is primarily based on human studies. Human circulating cells expressing osteocalcin (OCN; an osteoblast protein that serves as a marker for bone turnover) were identified in low concentrations in peripheral blood (PB). OCN+ cells increased with fracture and pubertal growth and formed mineralized nodules in vitro and ectopic bone in vivo [5]. A subset of OCN+ cells (»40%) expressed CD34 and exhibited a small, round phenotype, indicative of

Correspondence: AMANDA C. LARUE, PH.D., 109 Bee Street, Research Services (151), Ralph H. Johnson VAMC, Charleston, SC, 29403. E-mail: [email protected] (Received 1 January 2018; accepted 17 July 2018) ISSN 1465-3249 Copyright Published by Elsevier Inc. on behalf of International Society for Cellular Therapy. https://doi.org/10.1016/j.jcyt.2018.07.004

ARTICLE IN PRESS 2 R. R. Kelly et al. hematopoietic/endothelial origin [6]. Similarly, »20% of human PB CD34+ cells express OCN mRNA [7]. Human circulating CD34+ cells have also been shown to express multiple bone-related genes, including b-catenin, BMP-2, OCN and RUNX2 [8]. In vitro, BM-derived CD34+ cells rapidly lose CD34 expression during osteoblast differentiation, suggesting they may be a progenitor population [9]. Due to need for rapid efflux from BM to PB following injury, a major proportion of circulating osteoprogenitors, including CD34+OCN+ cells, likely arise from non-adherent BM, which contains hematopoietic/endothelial lineages. These cells may be exported to the circulation through sinusoids adjacent to bone trabeculae or via capillaries at sites of bony remodeling [10]. Long et al. identified nonadherent BM cells with osteogenic potential that expressed OCN and/or alkaline phosphatase (ALP) [11,12]. There is also evidence to suggest myeloid cells have calcifying potential. CD34+Collagen II+ (COL) cells were observed in intimal lesions of atherosclerotic plaques, with more than 50% of CD34+COLII+ cells co-expressing CD13 [13]. CD45+OCN+COLI+ cells were identified within calcified human aortic leaflets, localized to regions of heterotopic ossification, and accounted for 1.1% of circulating mononuclear cells (MNCs) [14]. Freshly isolated human OCN+ALP+ cells formed calcifications when implanted subcutaneously into nude mice. These cells were of myeloid origin and negative for common MSC markers (CD90, CD44 and CD29) [15]. In contrast, MSCs are defined, in part, by plastic adherence, and it is estimated that circulating osteoprogenitors able to adhere to plastic exist in exceedingly low numbers (>1 in 108) in adults [16]. Other studies did not find evidence of circulating MSCs from normal donors [17]. Thus, the idea of circulating MSCs is controversial, and circulating osteoprogenitors likely originate from the non-adherent (HSC-enriched) BM fraction. Therapeutically, HSC-derived circulating osteoprogenitors are attractive targets. Cell-based therapies typically involve ex vivo expansion of stem/ progenitor cells. Stem cell transplantation therapies, most notably MSC-based, often result in pulmonary cellular entrapment and poor homing to the desired site. An alternative strategy is to increase endogenous progenitors through mobilization agents, such as AMD3100 (CXCR4 antagonist), which may lead to increased numbers of cells homing from the BM to the injury site. In this regard, recent studies have demonstrated AMD3100 enhances fracture repair [18,19]. Further, AMD3100 treatment increased HSC and endothelial progenitor cell (EPC) populations, but not MSCs, suggesting pro-osteogenic effects of AMD3100 are MSC-independent [20].

In this study, we sought to determine whether CD34+OCN+ cells existed in mice similar to humans and if these cells expressed hematopoietic markers. Further, we sought to define the temporal participation of these cells to fracture, detect cells expressing CD34 and OCN within the fracture callus during early fracture consolidation and mineralization, and determine whether CD34+OCN+ cells could be pharmacologically increased using AMD3100. Our findings have the potential to affect the field by allowing studies of circulating CD34+OCN+ osteoprogenitors in murine models, which come with numerous experimental advantages for more mechanistic analyses compared to human cohorts.

Methods Animal models Mice were bred and maintained in the Ralph H. Johnson VA Medical Center (VAMC) Animal Research Facility. Research was conducted in accordance with US Public Health Service Policy on Humane Care and Use of Laboratory Animals and VAMC Institutional Animal Care and Use Committee (IACUC) guidelines. B6.SJL-PtprcaPepcb/BoyJ (C57Bl/6/CD45.1) breeders were obtained from Jackson Laboratories (Bar Harbor, ME). A double transgenic strain (Vav-Cre/mTmG; VavR) was generated in which all hematopoietic cells driven by Vav1 (global hematopoietic gene) are permanently labeled with green fluorescent protein (GFP) after Cre-mediated excision of floxed red fluorescent protein (RFP), whereas non-hematopoietic cells remain RFP-labeled [21]. Briefly, B6.129(Cg)-Gt(ROSA) 26Sortm4(ACTB-tdTomato,-EGFP)luo/J mice containing a CMV beta-actin enhancer-promoter driving a loxPflanked RFP preceding GFP (mTmG, Jackson Laboratories) were crossbred with B6.Cg-Tg(Vav1-cre) A2Kio/J mice that express Cre recombinase driven by Vav1 (Vav-Cre, Jackson Laboratories). Genotype was confirmed based on expression of GFP in PB. Male mice aged 11 24 weeks were used.

Flow cytometry BM cells were flushed from tibiae and femurs of euthanized mice. MNCs were isolated using Lympholyte-M (Cedarlane Laboratories Limited, Burlington, Canada). PB cells (50 mL) were harvested by retroorbital bleed. Red blood cells were lysed (Pharm Lyse lysing buffer; BD Biosciences, Franklin Lakes, NJ; 15 min, room temperature [RT]) and PB washed in phosphate-buffered saline (PBS) containing 0.1% bovine serum albumin (BSA). BM MNCs or PB cells were incubated with mouse FcR blocking

ARTICLE IN PRESS Murine CD34+ OCN+ cells reagent (Miltenyi Biotec, San Diego, CA). Staining was performed (30 min, 4˚C) using the following fluorescent-conjugated antibodies in 100 mL of PBS/ 0.1% BSA: 1 mL rat anti-mouse CD34 (FITC or AlexaFluor-700; RAM34; 0.5 mg/mL stock) (eBioscience, Thermo Fisher Scientific, Waltham, MA), 1:5 stock dilution of rabbit anti-OCN (AB10911) (EMD Millipore, Billerica, MA) conjugated to AlexaFluor 647 via APEX AlexaFluor 647 antibody labeling kit (Thermo Fisher Scientific, Waltham, MA), and 1:3 stock dilution of rat anti-mouse CD45 (PE-Cy7; 30-F11; 0.2 mg/mL stock) (BioLegend, San Diego, CA). Cells were washed and incubated with 0.5 mL LIVE/DEAD Violet Dead Cell Stain (30 minutes, 4˚C; Thermo Fisher Scientific, Waltham, MA). Cells were then washed, resuspended in PBS/0.1% BSA, and 100 mL of 123count eBeads Counting Beads (Thermo Fisher Scientific, Waltham, MA) were added to PB samples prior to analysis to allow for absolute counting of cell number. All samples, other than those used for ImagesStream analysis (described subsequently), were analyzed using the LSRFortessa X-20 (BD Biosciences, Franklin Lakes, NJ). Fluorescence minus one (FMO) controls were used to set gates. A minimum of 100,000 cell events (based on FSC/SSC gating) was recorded for each sample. Data were analyzed using FlowJo v10 (TreeStar, Ashland, OR).

3

in water. Sections were permeabilized in 0.02% Triton X-100/PBS, serum blocked overnight (4˚C), and incubated with anti-OCN (1:20; Abcam; ab93876; Cambridge, MA) and anti-CD34 (1:50) (Abcam; ab8158) antibodies (2 h, RT). Sections were blocked overnight (4˚C), followed by staining with fluorescence-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA; 1:100; 2 h, RT). Sections were stained with Hoechst 33342 dye (ThermoFisher Scientific, Waltham, MA; 500 ng/mL; 10 min), mounted using Fluoro-gel (Electron Microscopy Sciences, Hatfield, PA), and dried overnight. Micro-Computed Tomography (microCT) High-resolution (18 micrometer on-a-side cubic voxels; 360˚ rotation) images were obtained from anesthetized mice using the Siemens Inveon micro-CT scanner (Siemens Medical Solutions, Knoxville, TN). Axial two- and three-dimensional images were reconstructed and analyzed using Siemens software package IRW and Cobra EXXIM software (EXXIM Computing, Livermore, CA). Microscopy and Quantitative Analysis

Mice were given 100 mL of buprenorphine (0.03 mg/ mL; Reckitt Benckiser Pharmaceuticals, Hull, England) subcutaneously 1 h before being anesthetized with isoflurane. Tibial fractures were created on the right hindlimb by dropping a weight (»160 g) from a height of 14 20 cm, as previously described [4]. This model resulted in closed, unilateral, non-stabilized, mid-diaphyseal tibia fractures from which the mice recovered without complication.

Imaging was performed using a Nikon A1R confocal system equipped with differential interference contrast optics. Images were processed using NIS Elements (Nikon, Tokyo, Japan), ImageJ (National Institutes of Health, Bethesda, MD) and Adobe Photoshop CS5 (Adobe Systems). Fluorescence and differential interference contrast images of bone fracture callus sections were taken at 200 £ and 600 £ magnification. For quantification, two or three sections with a minimum of eight sections between were imaged at 200 £ from three fractured mice at 2 and 3 weeks post-fracture. Three fields per section were imaged, and the percentage of CD34+OCN+ cells compared with total Hoechst+ cells was calculated based on these cell counts.

ImageStream X Mark II

Mobilization

PB cells were harvested and stained for CD34, OCN, and LIVE/DEAD Violet, as described above. Cells were resuspended in 50 mL of PBS/0.1% BSA and analyzed and imaged using the ImageStream X Mark II imaging flow cytometer (Millipore Sigma, St. Louis, MO).

AMD3100 (10 mg/kg; Sigma-Aldrich, St. Louis, MO) or saline was injected subcutaneously once daily for 3 days. PB or BM was collected from mice 1 hour after last injection.

Fracture model

Statistical Analyses Immunofluorescence Bones were fixed in 4% paraformaldehyde in PBS (24 h), decalcified in 8% formic acid in water, and paraffin-embedded. Five-micrometer sections were cut, deparaffinized using xylene and dehydrated in alcohols. Antigen retrieval was performed using 5% formic acid

Analysis was conducted using Microsoft Excel (Microsoft Corporation, Redmond, WA) or GraphPad Prism 7 software (GraphPad Software, La Jolla, CA). Data are presented as mean § SD or SEM, as indicated. P  0.05 was considered significant. Twotailed unpaired t-test was used for comparing two

ARTICLE IN PRESS 4 R. R. Kelly et al. groups. One-way analysis of variance (ANOVA) with Bonferroni multiple comparisons post hoc correction test was used for multiple group comparison.

Results We first sought to demonstrate presence of CD34+OCN+ cells within murine PB under normal physiologic conditions by flow cytometric analysis. To determine whether events showing as CD34+OCN+ were indeed cells, we used the ImageStream X Mark II imaging flow cytometer. After gating on live, single cells, we imaged CD34+, OCN+, and CD34+OCN+ cells (Figure 1A). Next, we determined the frequency of CD34+OCN+ cells in PB under normal physiologic conditions and found an average of 0.1% CD34+OCN+ cells after gating on FSC/SSC, single, live cells (Figure 1B). Using 123count eBeads Counting Beads, we found an average of 3438 CD34+OCN+ cells per milliliter of PB under normal steady-state conditions (Figure 1B). To determine whether cells were of hematopoietic origin, PB from adult male C57Bl/6 mice was analyzed for expression of CD45, a pan hematopoietic marker. Analysis revealed >99.6% of the CD34+OCN+ population expressed CD45 (Figure 1C). To confirm hematopoietic origin, flow cytometric analysis was conducted on VavR PB-derived CD34+OCN+ cells, where 98.6% of these cells expressed GFP (Vav1+) (Figure 1D), whereas C57/Bl6 CD45.1 non-transgenic mouse PB was GFP (<0.5% GFP+), as expected (Figure 1E). We then examined the temporal pattern of the CD34+OCN+ population following fracture. Non-stabilized tibial fractures were created and frequency of CD34+OCN+ cells in PB and the BM of the contralateral limb was determined by flow cytometric analysis. The percentage of circulating CD34+OCN+ cells significantly increased at 3 weeks post non-stabilized tibial fracture (»2.6 fold over 1 week post-fracture group) and returned to baseline 4 weeks post-fracture (Figure 2A). Absolute counts of PB CD34+OCN+ cells correlated with the observed percentage increase, as we measured an »2.4-fold increase in CD34+OCN+ cell number at 3 weeks post non-stabilized tibial fracture (mean = 12 939 CD34+OCN+ cells/mL of PB) compared to 1 week post-fracture group (mean = 5419 CD34+OCN+ cells/mL of PB). Percentage of CD34+OCN+ cells in the BM compartment of the contralateral limb was also found to be elevated at 3 weeks post non-stabilized tibial fracture (»3.5 fold over 1 week post-fracture group) (Figure 2B). Elevation in the CD34+OCN+ population at the 3 weeks post-fracture time point correlated with callus consolidation and mineralization, based on micro-CT imaging of fractured mice (Figure 2C).

To determine whether there were cells expressing CD34 and OCN within the fracture callus, cohorts of mice were euthanized at 2 and 3 weeks postfracture, followed by immunofluorescent staining of tibial sections for CD34 and OCN. At 2 weeks postfracture, OCN staining was largely restricted to cartilaginous areas within the fracture callus (Figure 2D; upper panel; first series), whereas CD34 was abundantly expressed throughout the callus (Figure 2D; upper panel; second series). Interestingly, chondrogenic cells, as defined by morphology and location, expressing both CD34 and OCN were observed (Figure 2D; upper panel; first series; arrows). By 3 weeks post-fracture, there was an increase in abundance of CD34+OCN+ cells throughout the callus area (Figure 2D; lower panel; arrows). This increase in abundance at 3 weeks post-fracture was found to be 2.1-fold over percentage of CD34+OCN+ cells identified in 2-week sections compared with total Hoechst+ cells in each field (Figure 2E). This finding correlates with the observed systemic mobilization of CD34+OCN+ cells in PB and BM from contralateral limb at 3 weeks post-fracture (Figure 2A). Lastly, we examined the potential of AMD3100 to mobilize CD34+OCN+ cells from the BM into the PB. To mobilize CD34+OCN+ cells, AMD3100 (10 mg/kg) was subcutaneously administered daily for 3 days. This resulted in an »2.5-fold increase in absolute number of PB CD34+OCN+ cells, as determined using 123count eBeads Counting Beads. This increase in cell number began to trend toward resolution 1-week postAMD3100 (Figure 3A). BM was also collected from mobilized animals and tested for CD34+OCN+ cell levels. We found that 3-day administration of AMD3100 did not increase percentage of CD34+OCN+ cells in the BM, but that CD34+OCN+ cells did increase within the BM 1 week after treatment (an approximately two-fold increase compared with 3d-AMD3100 group) (Figure 3B). As a control to determine whether stress associated with subcutaneous administration alone leads to cell mobilization, a cohort of animals (n = 3) was injected with saline for 3 days, but no statistically significant differences in CD34+OCN+ cell levels between groups was observed (data not shown). Thus, our results demonstrate CD34+OCN+ cells exist in mice, express hematopoietic markers and their number can be pharmacologically increased in PB with AMD3100. Further, CD34+OCN+ cells are elevated both in PB and systemic BM 3 weeks post-fracture, which is associated with callus consolidation, early mineralization and number of CD34 and OCN co-expressing cells within the fracture callus. Discussion Fracture repair involves recruitment of osteochondroprogenitor cells to initiate healing. Understanding the

ARTICLE IN PRESS Murine CD34+ OCN+ cells

5

Figure 1. Identification of hematopoietic CD34+OCN+ cells in the PB of mice. (A) Representative images of cells within gated areas (CD34, CD34+OCN , 2.31%; OCN, CD34 OCN+, 1.52%; DP, CD34+OCN+, 0.26%) captured via ImageStream. Images taken in Brightfield (BF), FITC (CD34), BV421 (LIVE/DEAD, L/D) and AF647 (OCN) channels are displayed, along with merged image. Scale bar = 7 mm. (B) Representative flow cytometry plot demonstrating CD34+OCN+ staining of murine PB cells using LSRFortessa X-20. Left bar graph represents mean with SD of percentage CD34+ (mean 1.83%), OCN+ (mean 1.60%), and CD34+OCN+ (mean 0.10%) murine PB cells under normal, steady-state conditions. Right bar graph represents mean with SD of absolute number of CD34+ (mean 62,762), (Continued on next page)

ARTICLE IN PRESS 6 R. R. Kelly et al. role and origin of circulating osteoprogenitors and the factors regulating their mobilization and homing is essential for development of systemic therapies to improve bony healing. In this short report, we identify murine CD34+OCN+ cells in circulation and in the BM. We have also found these cells in Sprague-Dawley rats following iatrogenic femur fracture (data not shown). CD34+OCN+ cells were of hematopoietic origin based on CD45 and Vav1 expression. The demonstration of Vav1 expression is significant in that it confirms hematopoietic origin of this population, as it has been previously suggested that a subset of BMderived MSCs may express low levels of CD45 [22]. The circulating CD34+OCN+ population was found to peak at 3 weeks post-fracture, correlating with callus consolidation and early mineralization. Importantly, CD34+OCN+ cells were not elevated during the inflammatory response and hematoma formation immediately after fracture. It is worth noting that a human study did not find increased numbers of circulating bone precursors 15 days after fracture in nine male patients; however, only one time point was examined [23]. Interestingly, we also found an elevation in CD34+OCN+ cell levels in the BM isolated from the contralateral limb at 3 weeks post-fracture, suggesting there may be a systemic cell migration mechanism governing recruitment of these cells to the fracture callus site during early callus mineralization, when osteogenic cells are typically needed. We also found that a subset of cells in cartilaginous areas of the callus 2 weeks post non-stabilized tibial fracture were CD34+OCN+. On the basis of morphology and location, these cells appear to be of chondrogenic lineage. This confirms results from our earlier studies using a clonal cell transplantation model that demonstrated a hematopoietic origin for chondrocytes following non-stabilized tibial fracture [4]. The expression of OCN by chondrocytes is not entirely surprising, as studies have demonstrated ability of hypertrophic chondrocytes to become osteoblasts [24]. Further, OCN is expressed in hypertrophic chondrocytes and posthypertrophic chondrocytes of the epiphyseal growth plate, and intracellular OCN has been detected in chondrocytes 8 days post-fracture [25 28]. Thus, identification of CD34+OCN+ chondrocytes may be suggestive of cells transitioning during fracture

healing. Although CD34 and OCN co-expression within these cartilaginous areas represents a novel finding, future studies are needed to further define these cells and determine the mechanism and purpose of this co-expression. By 3 weeks post non-stabilized tibial fracture, CD34+OCN+ cells were found throughout the fracture callus, not just in cartilaginous areas. This is intuitive within the context of this fracture model, as we found that callus mineralization began at 3 weeks post-fracture. Thus, there should be an increase in OCN staining at this time point because OCN is thought to regulate bone matrix formation and would correlate with osteoblast presence. Further, CD34+OCN+ cells were physiologically mobilized around 3 weeks post-fracture, both in PB and BM from the contralateral limb. Thus, the observed increase in CD34+OCN+ cell number in the callus at 3 weeks compared with 2 weeks (»2.1-fold) is in line with this result. This increase was calculated as the percentage of CD34+OCN+ cells compared with total Hoechst+ cells per field. It is worth noting that we found an »55% reduction in total Hoechst+ cells per field from 3 weeks post-fracture sections compared with 2 weeks post-fracture sections, but higher numbers of CD34+OCN+ cells at 3 weeks post-fracture. This finding is not surprising given that this reduction in cell number is expected to occur during the callus consolidation and early mineralization we observed at 3 weeks as part of the fracture repair process when osteogenic cells begin to lay down extensive matrix and most non-osteogenic cells are cleared. However, it remains to be determined whether the observed increase in circulating CD34+OCN+ cells at 3 weeks post-fracture is associative or causative of the increased number of CD34+OCN+ cells incorporating into the fracture callus. Lastly, we found that the CD34+OCN+ cell count was significantly increased in PB after 3 days of AMD3100 administration, suggesting a pharmacologic means of targeting these cells. However, we did not observe a percentage increase in CD34+OCN+ cells in PB compared with total white blood cells following AMD3100 administration (data not shown). This finding is not unexpected because AMD3100 is known to mobilize the majority of white blood cells out of the BM and into circulation. In line with this, we found that AMD3100

OCN+ (mean 57,013) and CD34+OCN+ (mean 3438) murine PB cells calculated per milliliter of PB with 123count eBeads Counting Beads under normal, steady-state conditions; n = 15. (C) Representative plots of CD45-PE-Cy7 expression on CD34-FITC+OCN-AF647+ cells; mean 99.6% CD45+ expression; n = 10 (gray histogram) compared with PE-Cy7 FMO control (green histogram). (D) Representative plots showing CD34-AF700+OCN-AF647+ cells (gray gated area) analyzed for GFP expression in VavR PB, mean 98.6% GFP+ (histogram), n = 5. (E) Representative plots showing CD34-AF700+OCN-AF647+ cells (gray gated area) isolated from C57Bl/6 CD45.1 male mice examined for GFP expression as a negative control, <0.5% GFP+ (histogram), n = 4.

ARTICLE IN PRESS Murine CD34+ OCN+ cells

7

Figure 2. CD34+OCN+ cells peak in the PB and BM at 3 weeks post non-stabilized tibial fracture during callus consolidation and are associated with increased CD34 and OCN expression within the fracture callus. (A) PB CD34+OCN+ cells from a cohort of mice (n = 5) were analyzed by flow cytometry before fracture and then 1 week, 2 weeks, 3 weeks and 4 weeks post-fracture. A significant increase in % CD34FITC+OCN-AF647+ cells was observed 3 weeks post-fracture, as calculated using one-way ANOVA with Bonferroni multiple comparisons post-test, *P = 0.0006 compared with pre-fracture time point (Pre-Fx) &P = 0.0002 compared with 1 week post non-stabilized fracture time point (1wk Post-Fx), #P < 0.0001 compared with 2 week post non-stabilized fracture time point (2wk Post-Fx), and @P = 0.0248 compared to 4 week post non-stabilized fracture time point (4wk Post-Fx); mean § SEM. (B) Cohorts of mice were euthanized before fracture and 1 4 weeks post non-stabilized tibial fracture and bone marrow from contralateral hindlimb was analyzed for CD34+OCN+ cells (n = 4 5/group). A significant increase in % CD34-FITC+OCN-AF647+ cells was observed at 3 weeks based on one-way ANOVA with Bonferroni Multiple Comparison post-test, *P = 0.0328 compared with pre-fracture time point (Pre-Fx), &P = 0.0072 compared to one week post non-stabilized fracture time point (1wk Post-Fx), #P = 0.0012 compared with 2 week post non-stabilized fracture time point (2wk Post-Fx); mean § SEM. C) Representative microCT images from a cohort of non-stabilized tibial fracture mice at 1 4 weeks postfracture. Note callus transformation and mineralization at week 3. (D) Immunofluorescent images of paraffin-embedded 5-mm sections of fracture calluses. At 2 weeks post-fracture, co-expression of CD34 and OCN was found to be mainly restricted to cartilaginous areas (first series, upper panel; white arrows) and not found within non-cartilaginous regions of the callus (first series, lower panel). By 3 weeks postfracture, CD34+OCN+ cells are present throughout the callus area, both in cartilaginous regions (second series, upper panel, white arrows) and mineralized regions (second series, lower panel, white arrows). Insets demonstrate double positive cells; asterisks are for reference. Secondary only controls display minimal background staining. Scale bar = 50 mm for non-inset images, scale bar = 25 mm for inset images. (E) Quantification of number of CD34+OCN+ cells within 2-week compared with 3-week fracture callus sections show a 2.1-fold increase in 3week calluses. Percentage change is shown based on number of counted CD34+OCN+ cells in proportion to total Hoechst+ cells (mean 4.9% 2-week post-fracture group versus mean 10.4% 3-week post-fracture group). *P = 0.0022, two-tailed unpaired t-test.

ARTICLE IN PRESS 8 R. R. Kelly et al.

Figure 3. CD34+OCN+ cell numbers are altered in PB and BM upon AMD3100 treatment. (A) PB was collected and analyzed for number of CD34-FITC+OCN-AF647+ cells before AMD3100, 1 h after last injection of 3-day regimen of daily 10 mg/mL subcutaneous AMD3100 injections, and 1 week after AMD3100 injections (n = 4 8/group). A transient »2.5-fold increase in number of CD34+OCN+ cells was observed after 3 days of AMD3100 treatment; mean § SEM, *P = 0.0043, one-way ANOVA with Bonferroni multiple comparisons post-correction test. (B) BM was collected and analyzed for % of CD34-FITC+OCN-AF647+ cells before AMD3100, 1 h after last injection of 3-day regimen of daily 10 mg/mL subcutaneous AMD3100 injections, and 1 week after AMD3100 injections (n = 4/group). An approximately twofold increase in CD34+OCN+ cells was observed within the BM 1 week after AMD3100-treatment compared with 3dAMD3100; mean § SEM, *P = 0.0102, one-way ANOVA with Bonferroni multiple comparisons post-correction test.

resulted in an approximately 3-fold increase of total single, live PB cells that resolved back to baseline 1 week after AMD3100 (data not shown). The finding that CD34+OCN+ cells are also increased following AMD3100 administration is significant because it demonstrates that, unlike MSCs [20], CD34+OCN+ cells are targeted by AMD3100. This provides a pharmacologic means to increase their number in circulation, where they could then travel to sites where they are needed, including the fracture callus. We predict, on the basis of our findings of temporal modulation of the CD34+OCN+ population following fracture, that both percentage and absolute number of CD34+OCN+ cells would increase in PB in fractured mice with AMD3100 administration. In the BM, the pool of CD34+OCN+ cells trended toward a decrease after 3 days of AMD3100, although this was not statistically significant. However, the percentage of CD34+OCN+ population was slightly elevated 1 week after AMD3100 administration, possibly as a means to replenish these cells back to a baseline level. The significance of this process and the purpose and function of these cells in the bone marrow during normal steady-state conditions is, at present, unknown. It should also be noted that, because of technical limitations of the 123count eBeads Counting Beads and the methods used to harvest and process BM compared with PB cells, it is not possible to determine absolute cell counts from BM like it is from PB. Thus, only percentages of CD34+OCN+ cells were reported for BM measurements. This study weakness will need to be

addressed in future studies to completely understand the regulation and homeostatic feedback of the CD34+OCN+ cell population (both by percentage and number) between BM and PB during AMD3100 treatment and/or fracture repair. Ultimately, the goal of future studies is to test the functional impact of AMD3100 mobilization and/or delivery of pro-osteogenic factors for treating atrophic fracture nonunion. This may serve to mitigate the decrease in circulating osteoprogenitors found in patients with fracture nonunion and potentially be used as a treatment option for fractures at high risk of nonunion development [29]. It has previously been shown that, in a parabiotic mouse model, circulating osteoblasts are physiologically mobilized to contribute to callus formation during fracture repair [30]. Further, circulating osteoblasts were shown to originate in the BM, based on GFP-transgenic BM transplantation and subsequent examination of GFP+ osteoblasts at sites of ectopic bone formation [31]. Mobilization is clinically attractive because it mitigates need for additional surgical interventions, does not have limitations associated with bone grafts, does not require costly ex vivo stem cell expansion and may serve as an adjunct systemic therapy following fracture reduction and stabilization. However, excess mobilization may be detrimental to fracture healing, due to inability of SDF-1 to chemoattract CXCR4-expressing osteogenic stem cells. In this short report, we found that a 3-day treatment of AMD3100 increased number of circulating CD34+OCN+ cells and that this effect is likely transient, as CD34+OCN+ cell number trended toward a return

ARTICLE IN PRESS Murine CD34+ OCN+ cells to baseline 1 week post-treatment, which would mitigate the possible negative effect associated with prolonged SDF-1 inhibition. These findings correlate well with a study by Toupadakis et al., who reported enhancement of fracture healing after 3-day AMD3100 treatment, although the authors did not specifically examine osteoprogenitor levels [32]. Thus, increasing the CD34+OCN+ population may accelerate healing, enhance quality of recovery and prevent long-term complications associated with “difficult-to-treat” fractures, including atrophic nonunions. Regarding broad clinical implications, circulating osteoprogenitors are likely effectors for pathologic ectopic calcification, as observed in vascular disorders, heterotopic ossification, diabetes and renal dysfunction [33]. Because non-osseous tissues do not harbor native osteoblasts, the source for ectopic calcification and accompanying osteoblasts is likely non-adherent circulating osteoprogenitors. Understanding the origin of these cells may elucidate molecular targets to regulate or negate these deleterious processes. G€ ossl et al. found that number of CD34+OCN+VEGFR2+CD133 cells was a marker for early coronary atherosclerosis and suggested that these cells may directly contribute to vascular calcification [8]. Fadini et al. showed circulating myeloid cells mineralized when cultured in Matrigel, which might explain the severe calcification found in cardiac tissue after injection of MNCs into infarcted rat hearts [15]. Myeloid circulating cells have been termed either hemosteoblasts or mono-osteoblasts to describe their origin [34,35]. It remains to be seen how these cells relate to the circulating CD34+OCN+ cells identified herein. Understanding mobilization, homing, proliferation, and signaling mechanisms governing circulating hematopoietic osteoprogenitor survival and function has obvious clinical utility. We provide initial evidence of a murine CD34+OCN+ circulating population similar to that observed in humans. Because this is the first time, to our knowledge, these hematopoietic CD34+OCN+ cells have been reported to exist in mice, this study opens the door for more in depth mechanistic analyses and allows for the use of complex murine models to study their impact in bone remodeling and repair. It is noteworthy that there is present difficulty in isolation, expansion and survivability of murine CD34+OCN+ cells ex vivo, which is partially based on significantly lower cell yield than that obtained from human PB. These technical challenges will need to be overcome before elucidating molecular mechanisms regulating CD34+OCN+ cell function. Thus, this initial short report has focused on short-term and in vivo experiments that do not require clonal expansion of these cells. Intriguing questions arise from this study: What is the functional role of circulating cells that express a

9

mature osteoblast marker (OCN) and a hematopoietic progenitor marker (CD34)? How does the function of these cells compare to MSCs or other sources of osteogenic cells (e.g. periosteal stromal cells or bone lining cells)? Are these cells progenitor cells that are capable of proliferation? Do these cells have additional paracrine roles in fracture healing? What mechanisms regulate their systemic migration to enhance fracture repair? Lastly, can these cells serve as a marker to aid clinicians in determining therapeutic efficacy following intervention to stimulate bony repair (i.e., does number correlate with healing)? Therapeutically, these cells may serve as an autologous cell replacement therapy that could be manipulated with mobilization strategies (e.g., AMD3100) or be a target cell for gene therapy to inhibit pathological ectopic calcification. Future studies will examine functional and therapeutic impacts of temporally modulating HSC-derived osteoprogenitors in cases of complex fracture repair and in fracture models using different methods of fracture fixation that promote the two pathways of bone healing (i.e., endochondral and intramembranous ossification).

Acknowledgments This work was supported by the Biomedical Laboratory Research and Development Program of the Department of Veterans Affairs (VA Merit Award to ACL, BX000333), the National Heart, Lung, and Blood Institute of the National Institutes of Health (HL007260) and the MUSC T32 Dental Training Grant (T32DE017551). The authors acknowledge the Hollings Cancer Center Cell Evaluation and Therapy Shared Resource (Flow Cytometry and Cell Sorting Unit) and the Hollings Cancer Center/ MUSC Center for Biomedical Imaging Small Animal Imaging Core (P30 CA138313). All authors read and approved of the final manuscript before submission. Disclosure of interest: The authors have no commercial, proprietary, or financial interest in the products or companies described in this article.

References [1] Kelly RR, LaRue AC. Stem Cell-Based Therapies for Bony Repair. Frontiers in Stem Cell and Regenerative Medicine Research, 1. Bentham Science; 2017. p. 3–67. https://doi. org/10.2174/97816810825781160301. [2] Caplan AI, Correa D. The MSC: an injury drugstore. Cell Stem Cell 2011;9:11–5. https://doi.org/10.1016/j. stem.2011.06.008. [3] Caplan AI. Adult Mesenchymal Stem Cells: When, Where, and How. Stem Cells Int 2015;2015:628767. https://doi.org/ 10.1155/2015/628767. [4] Mehrotra M, Williams CR, Ogawa M, LaRue AC. Hematopoietic stem cells give rise to osteo-chondrogenic cells. Blood

ARTICLE IN PRESS 10

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

R. R. Kelly et al. Cells Mol Dis 2013;50:41–9. https://doi.org/10.1016/j. bcmd.2012.08.003. Eghbali-Fatourechi GZ, Lamsam J, Fraser D, Nagel D, Riggs BL, Khosla S. Circulating osteoblast-lineage cells in humans. N Engl J Med 2005;352:1959–66. https://doi.org/10.1056/ NEJMoa044264. Eghbali-Fatourechi GZ, M€ odder UIL, Charatcharoenwitthaya N, Sanyal A, Undale AH, Clowes JA, et al. Characterization of circulating osteoblast lineage cells in humans. Bone 2007;40:1370–7. https://doi.org/10.1016/j.bone.2006.12.064. Matsumoto T, Kawamoto A, Kuroda R, Ishikawa M, Mifune Y, Iwasaki H, et al. Therapeutic potential of vasculogenesis and osteogenesis promoted by peripheral blood CD34-positive cells for functional bone healing. Am J Pathol 2006;169:1440–57. https://doi.org/10.2353/ajpath.2006.060064. G€ ossl M, M€ odder UI, Atkinson EJ, Lerman A, Khosla S. Osteocalcin expression by circulating endothelial progenitor cells in patients with coronary atherosclerosis. J Am Coll Cardiol 2008;52:1314–25. https://doi.org/10.1016/j.jacc.2008.07.019. Chen JL, Hunt P, McElvain M, Black T, Kaufman S, Choi ES. Osteoblast precursor cells are found in CD34+ cells from human bone marrow. Stem Cells 1997;15:368–77. https://doi.org/10.1002/stem.150368. Eriksen EF, Eghbali-Fatourechi GZ, Khosla S. Remodeling and vascular spaces in bone. J Bone Miner Res 2007;22:1–6. https://doi.org/10.1080/14041040701482935. Long MW, Williams JL, Mann KG. Expression of human bone-related proteins in the hematopoietic microenvironment. J Clin Invest 1990;86:1387–95. https://doi.org/ 10.1172/JCI114852. Long MW, Williams JL, Mann KG. Expression of human bone-related proteins in the hematopoietic microenvironment. J Clin Invest 1990;86:1387–95. https://doi.org/ 10.1172/JCI114852. Doehring LC, Heeger C, Aherrahrou Z, Kaczmarek PM, Erdmann J, Schunkert H, et al. Myeloid CD34+CD13+ Precursor Cells Transdifferentiate into Chondrocyte-Like Cells in Atherosclerotic Intimal Calcification. Am J Pathol 2010;177:473–80. https://doi.org/10.2353/ajpath.2010.090758. Egan KP, Kim J-H, Mohler ER, Pignolo RJ. Role for Circulating Osteogenic Precursor (COP) Cells in Aortic Valvular Disease. Arterioscler Thromb Vasc Biol 2011;31:2965–71. https://doi.org/10.1161/ATVBAHA.111.234724. Fadini GP, Albiero M, Menegazzo L, Boscaro E, Vigili de Kreutzenberg S, Agostini C, et al. Widespread increase in myeloid calcifying cells contributes to ectopic vascular calcification in type 2 diabetes. Circ Res 2011;108:1112–21. https://doi.org/10.1161/CIRCRESAHA.110.234088. Kuznetsov SA, Mankani MH, Gronthos S, Satomura K, Bianco P, Robey PG. Circulating skeletal stem cells. J Cell Biol 2001;153:1133–40. Roufosse CA, Direkze NC, Otto WR, Wright NA. Circulating mesenchymal stem cells. Int J Biochem Cell Biol 2004;36:585– 97. https://doi.org/10.1016/j.biocel.2003.10.007. Kumar S, Ponnazhagan S. Mobilization of bone marrow mesenchymal stem cells in vivo augments bone healing in a mouse model of segmental bone defect. Bone 2012;50:1012–8. https://doi.org/10.1016/j.bone.2012.01.027. Toupadakis CA, Wong A, Genetos DC, Chung D-J, Murugesh D, Anderson MJ, et al. Long-term administration of AMD3100, an antagonist of SDF-1/CXCR4 signaling, alters fracture repair. J Orthop Res 2012;30:1853–9. https://doi. org/10.1002/jor.22145. Pitchford SC, Furze RC, Jones CP, Wengner AM, Rankin SM. Differential mobilization of subsets of progenitor cells

[21]

[22] [23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

[33] [34]

[35]

from the bone marrow. Cell Stem Cell 2009;4:62–72. https://doi.org/10.1016/j.stem.2008.10.017. Georgiades P, Ogilvy S, Duval H, Licence DR, CharnockJones DS, Smith SK, et al. VavCre transgenic mice: a tool for mutagenesis in hematopoietic and endothelial lineages. Genesis 2002;34:251–6. https://doi.org/10.1002/gene.10161. Pignolo RJ, Shore EM. Circulating osteogenic precursor cells. Crit Rev Eukaryot Gene Expr 2010;20:171–80. D’Amelio P, Cristofaro MA, Grimaldi A, Ravazzoli M, Pluviano F, Grosso E, et al. The role of circulating bone cell precursors in fracture healing. Calcif Tissue Int 2010;86:463–9. https://doi.org/10.1007/s00223-010-9362-3. Zhou X, von der Mark K, Henry S, Norton W, Adams H, de Crombrugghe B. Chondrocytes transdifferentiate into osteoblasts in endochondral bone during development, postnatal growth and fracture healing in mice. PLoS Genet 2014;10:e1004820. https://doi.org/10.1371/ journal.pgen.1004820. Hughes SS, Hicks DG, O’Keefe RJ, Hurwitz SR, Crabb ID, Krasinskas AM, et al. Shared phenotypic expression of osteoblasts and chondrocytes in fracture callus. J Bone Miner Res 1995;10:533–44. https://doi.org/10.1002/ jbmr.5650100405. Gerstenfeld LC, Shapiro FD. Expression of bone-specific genes by hypertrophic chondrocytes: implication of the complex functions of the hypertrophic chondrocyte during endochondral bone development. J Cell Biochem 1996;62:1–9. https://doi.org/10.1002/(SICI)1097-4644(199607) 62:1<1::AID-JCB1>3.0.CO;2-X. Bianco P, Cancedda FD, Riminucci M, Cancedda R. Bone formation via cartilage models: the “borderline” chondrocyte. Matrix Biol 1998;17:185–92. Pullig O, Weseloh G, Ronneberger D, K€ak€ onen S, Swoboda B. Chondrocyte differentiation in human osteoarthritis: expression of osteocalcin in normal and osteoarthritic cartilage and bone. Calcif Tissue Int 2000;67:230–40. Seebach C, Henrich D, Tewksbury R, Wilhelm K, Marzi I. Number and proliferative capacity of human mesenchymal stem cells are modulated positively in multiple trauma patients and negatively in atrophic nonunions. Calcif Tissue Int 2007;80:294–300. https://doi.org/10.1007/s00223-007-9020-6. Kumagai K, Vasanji A, Drazba JA, Butler RS, Muschler GF. Circulating cells with osteogenic potential are physiologically mobilized into the fracture healing site in the parabiotic mice model. J Orthop Res 2008;26:165–75. https://doi.org/10.1002/ jor.20477. Otsuru S, Tamai K, Yamazaki T, Yoshikawa H, Kaneda Y. Bone marrow-derived osteoblast progenitor cells in circulating blood contribute to ectopic bone formation in mice. Biochem Biophys Res Commun 2007;354:453–8. https://doi. org/10.1016/j.bbrc.2006.12.226. Toupadakis CA, Granick JL, Sagy M, Wong A, Ghassemi E, Chung D-J, et al. Mobilization of endogenous stem cell populations enhances fracture healing in a murine femoral fracture model. Cytotherapy 2013;15:1136–47. https://doi.org/ 10.1016/j.jcyt.2013.05.004. Canalis E. The fate of circulating osteoblasts. N Engl J Med 2005;352:2014–6. https://doi.org/10.1056/NEJMe058080. Tintut Y, Demer L. The hemosteoblast: friend or foe? Circ Res 2011;108:1038–9. https://doi.org/10.1161/CIRCRESAHA.111.244665. Zhang Z, Shively JE. Generation of novel bone forming cells (monoosteophils) from the cathelicidin-derived peptide LL37 treated monocytes. PLoS ONE 2010;5:e13985. https:// doi.org/10.1371/journal.pone.0013985.