Human Umbilical Cord Blood-Derived CD34+ Cells Reverse Osteoporosis in NOD/SCID Mice by Altering Osteoblastic and Osteoclastic Activities

Background Osteoporosis is a bone disorder associated with loss of bone mineral density and micro architecture. A balance of osteoblasts and osteoclasts activities maintains bone homeostasis. Increased bone loss due to increased osteoclast and decreased osteoblast activities is considered as an underlying cause of osteoporosis. Methods and Findings The cures for osteoporosis are limited, consequently the potential of CD34+ cell therapies is currently being considered. We developed a nanofiber-based expansion technology to obtain adequate numbers of CD34+ cells isolated from human umbilical cord blood, for therapeutic applications. Herein, we show that CD34+ cells could be differentiated into osteoblastic lineage, in vitro. Systemically delivered CD34+ cells home to the bone marrow and significantly improve bone deposition, bone mineral density and bone micro-architecture in osteoporotic mice. The elevated levels of osteocalcin, IL-10, GM-CSF, and decreased levels of MCP-1 in serum parallel the improvements in bone micro-architecture. Furthermore, CD34+ cells improved osteoblast activity and concurrently impaired osteoclast differentiation, maturation and functionality. Conclusions These findings demonstrate a novel approach utilizing nanofiber-expanded CD34+ cells as a therapeutic application for the treatment of osteoporosis.


Introduction
Osteoporosis is a systemic bone disorder, affecting more than 200 million people worldwide [1]. Bone is a dynamic organ that undergoes constant remodeling via cycles of bone formation and resorption, by osteoblasts and osteoclasts [2]. Imbalance of osteoclastic and/or osteoblastic activities generally results in low bone mineral density (BMD), loss of bone mass and mechanical strength, leading to increased risk of fractures, typical of osteoporosis [3]. Impaired osteoblastic differentiation of bone marrow progenitor cells may also play a significant role in developing osteoporosis. Age, endocrine malfunction or deficiency, nutrition, or lack of physical activity, all can imbalance the osteoblasts and osteoclasts activities, affecting both trabecular and cortical bone at molecular, cellular and structural levels [4,5]. It has been shown that reduction in trabecular bone in osteoporosis is associated with increased adiposity in bone marrow, which could be due to transcriptional switch in favor of adipogenesis instead of osteoblastogenesis of bone marrow precursor cells [6,7].
The mesenchymal progenitor cells in the bone marrow give rise to osteoblasts under the influence of multiple osteogenic signals specific for their proliferation and differentiation [8]. Osteoblastic differentiation is initiated by binding of bone morphogenetic proteins (BMPs) to their receptors that activate transcription factors, Runx2 and Osterix, and subsequent expression of downstream osteoblast specific genes such as alkaline phosphatase, collagen type 1, osteonectin, osteocalcin and bone sialoprotein [9,10,11]. BMPs upregulate osteoblastic genes via activation of Smad1/5/8 signaling molecules and regulate mineralization of osteoblastic cells via Wnt in an autocrine signaling loop [12]. Runx2 is a potent inhibitor of adipogenesis, and is required for the differentiation of adipocytes to osteogenic lineage [13]. Additionally, balance of osteoprotegrin (OPG): receptor activator of nuclear factor kappa-B ligand (RANKL) ratio, osteocalcin and cytokines such as interleukin (IL)-1 IL-4, IL-6, monocyte chemotactic protein (MCP)-1 and granulocyte macrophage colony stimulating factor (GM-CSF) have been shown to regulate the activities of osteoblastic and osteoclastic cells [14,15].
Although, associated with side effects, anti-resorptive and anabolic therapies are currently available for osteoporosis [3,16]. Furthermore, these therapies have temporary effects, and the decrease in fracture incidences in long-term is debatable [17,18]. Recently, much effort has been expended to understand the therapeutic effectiveness of CD34+ cells in various degenerative diseases. However, the major hurdles are the unavailability of sufficient number of biologically functional CD34+ cells and maintaining their regenerative potential for therapeutic applications.
We previously reported that human CD133 + /CD34 + cells could be expanded in vitro up to 250-fold in a serum-free medium on aminated poly-ether sulfone (PES) nanofiber coated plates within 10 days, while preserving stem cell phenotype and biological functionality [19]. These cells are considered biologically superior as they exhibit better engraftment capabilities, express homing markers (CXCR4 and LFA-1) towards bone marrow and maintain their multipotency. This allows them to differentiate into multiple lineages such as endothelial, and hematopoietic lineages. Here we show that nanofiber-expanded CD34 + cells could be differentiated towards osteoblastic lineage, in vitro. Furthermore, CD34 + cell transplantation into an osteoporotic NOD/SCID murine model augments bone formation rate, bone mineral density and improves bone micro-architecture. These improvements correlate with the elevated serum levels of osteocalcin, interleukin (IL)-10 and granulocyte-macrophage colony stimulating factor (GM-CSF), and decreased level of monocyte chemotactic protein-1 (MCP-1). CD34 + cell transplantation not only improved osteoblast functionality but also concurrently impaired differentiation and maturation of osteoclasts, thereby reducing osteoclast activity in osteoporotic mice. The findings demonstrate a novel potential of nanofiber-expanded CD34 + cells in reverting osteoporosis.

Differentiation of nanofiber-expanded CD34 + cells towards osteoblastic lineage
Our previous studies showed that nanofiber expanded human umbilical cord blood (hUCB) CD34 + cells retain multipotency as evident by their ability to differentiate into endothelial or smooth muscle cells [19,20]. Here we sought, whether these cells could also be differentiated towards osteoblastic lineage in vitro. To test that, CD133 + cells were isolated from hUCB and expanded on nanofiber coated plates in serum-free expansion medium with supplements for 10 days, in vitro [19]. The cell phenotype was confirmed by the expression of CD34, CD45, CXCR4, LFA-1, MHC-I & II, CD14, CD11a and absence of CD69, CD117, CD105 using flowcytometric analyses (data not shown). CD34+ cells were then induced to osteoblastic differentiation in the presence of ascorbic acid and b-glycerophosphate, in vitro. Remarkable cellular changes in shape and size were observed within 7 days and mineralized nodules were apparent at 21 days in more than 95% of the cells ( Figure 1A, upper right panel). The mineral deposition by induced CD34 + cells was confirmed by the presence of calcium in Alizarin red stained cells. CD34 + cells cultured in tissue culture media (TCM) lacking ascorbic acid and b-glycerophosphate, also showed few randomly scattered cells with faint intracellular alizarin red staining ( Figure 1A, lower left panel). However, the intensity of stain was markedly higher in cells induced to differentiate into osteoblastic lineages ( Figure 1A, lower right panel).

Molecular evidences for osteoblastic differentiation of CD34 + cells
To further investigate the differentiation of CD34 + cells into osteoblastic lineage, the expression of osteoblast specific genes and proteins were analyzed. Semi quantitative RT-PCR analysis revealed that differentiated nanofiber-expanded cells upregulated expression of transcription factor Runx2, and bone associated proteins such as alkaline phosphatase, osteocalcin, osteonectin and collagen type 1A1 at various time points during the course of differentiation ( Figure 1B). The expression of osteocalcin in freshly isolated CD133 + cells was consistent with earlier reports [21]. Fold increase for the gene expressions at mRNA level compared to undifferentiated control was analyzed, as follows: Runx2; Day Figure 1C). Western blots showed an increase in protein levels of BMP2, BMP4, dishevelled protein (DVL) 2, DVL3 and Smad1/5/8 in differentiated cells at all time points during the course of differentiation. GAPDH was used as loading control and MC3T3 cells differentiated into osteoblasts were used as a positive control ( Figure 1D). Fold increase of the protein level compared to undifferentiated control was analyzed, as follows BMP2; Day 7,

Nanofiber-expanded CD34 + cells induce bone formation in a murine model of osteoporosis
We further investigated the therapeutic potential of nanofiberexpanded human CD34 + in a mouse model of dexamethasoneinduced osteoporosis in immunocompromised NOD/SCID mice. Mice (n = 6/group) injected with dexamethasone for 21 consecutive days, followed by withdrawal for 5 days with tapering dose, and subsequently were either not treated and sacrificed (Op), treated with CD34 + cells via intra-cardio-ventricular injection (Op+ Cells; 0.5610 6 /mouse), or treated with DMEM alone (Op+Med). Mice in all groups did not exhibit weight loss or gain during the course of experimentation. Hematoxylin and eosin (H & E) staining of longitudinal sections of femurs showed a decrease in the number of trabecular bone spicules in Op and Op+Med mice, as compared to untreated controls. However, after 28 days of CD34 + cell transplantation, an increase in the numbers of trabecular bone spicules was observed ( Figure 2B, arrows). The number of adipocytes was significantly increased in Op and Op+Med mice, as compared to untreated control ( Figure 2B, arrowheads). Further, the number of adipocytes was evaluated in each group and a marked reduction in adipocytes was observed in Op+Cells mice, as compared to Op mice. The number of adipocytes/high power field (HPF) was: control, 9.562.7; Op, 34.2567.5; Op+Media, 29.7563.9; Op+Cells, 13.7561.7 ( Figure 2C, lower panel).

CD34 + cells home to the bone marrow
The chemokine receptor, CXCR4 binds to SDF-1, a chemotactic ligand expressed by bone marrow cells. Additionally, the presence of lymphocyte adhesion molecules (LFA-1) is required for the bone marrow homing of CD34 + cells. Since, we observed that both CXCR4 and LFA1 are highly expressed on CD34 + cells after 10 day of expansion on nanofibers, we next sought to examine the homing of these cells in the osteoporotic mice. Although delivered systemically, via intra-cardio-ventricular injection, CD34 + cells home to bone marrow ( Figure 2D) as well as other organs such as lung, liver and spleen (data not shown). In the bone marrow, CD34 + were detected near the endosteal sites and around the bone marrow sinusoids, as well as at the surface of trabecular bone and at day 14 after differentiation of cells with osteoblast specific stimulants (b-glycerophosphate and ascorbic acid) in DMEM complete medium cultured on a 24-well tissue culture plate. Arrowheads indicate the clusters that formed after differentiation. Alizarin Red S staining (lower panel) was performed to the cells cultured for 21 days either with osteoblast specific differentiation medium (differentiated cells) or DMEM complete medium only (control). Red stains in the control image indicate the intracellular calcium in random differentiated cells and arrowheads indicate higher level of mineral depositions in differentiated cells. The experiment was repeated at least three times and representative images are shown. (B). RNA was isolated from differentiated cells during the course of osteoblastic differentiation at various time points as stated, and one microgram of RNA was used to make cDNA. One micro liter of cDNA was used to perform semi-quantitative RT-PCR analysis for Runx2, alkaline (Alk) phosphate, osteocalcin, osteonectin, collagen Type 1A1 and GAPDH as a loading control. Nanofiber expanded cells (10 days) were used as a control.

Mineral apposition rate in response to CD34 + cell transplantation
To determine the effect of CD34 + cell transplantation on the mineral apposition rate (MAR), calcium binding fluorescent dye calcein (green) was injected at 17 days post CD34+ cell or medium injection. Subsequently, on day 24, fluorescent dye alizarin (red), and femurs were harvested on day 28. The inter-label distance between the two dyes was narrower at cortical and trabecular regions of the Op+Med mice compared to the Op+Cells mice indicating limited bone deposition in Op+Med mice. The values for mineral apposition rate were assessed in plastic embedded femur bone sections of Op+Med and Op+ Cells on the endosteal surface of metaphysial region of the femur, distal to the growth plate (Cortical MAR, mm/day; Op+Med, 0.4760.04; Op+Cells, 3.0760.29). Similarly, significant increase in trabecular MAR was observed in Op+Cells compared to Op+Med at the growth plate region (Trabecular MAR, mmm/day; Op+Med, 0.4960.05; Op+Cells, 1.3560.09) ( Figure 3).

Ultra structural analysis of bones after CD34 + cell transplantation
To evaluate the extent of trabecular and cortical bone repair/ regeneration and to image the differences in bone quality at the ultrastructural level, femurs from Op, Op+Med, Op+Cells were examined by micro computed tomography (MicroCT) ( Figure Figure 4A, lower right, panel). Similarly, similar pattern was observed for bone mineral density (BMD) of the trabeculae. BMD was significantly increased in Op+Cells mice compared to Op +Med (BMD, g/cm 3 ; control, 0.22360.02; Op, 0.12160.02; Op+Med, 0.12960.02; Op+Cells, 0.2160.01). The reductions in BMD in Op mice indicated that dexamethasone treatments effectively decreased mineral density, and BMD was increased after CD34 + cell transplantation indicated reversal of the osteoporotic phenotype. Similarly, significant decrease in the degree of anisotropy (DA) was observed in the Op+Cells mice compared to Op+ Med mice (DA; control, 2.260.29; Op, 360.28; Op+Med, 2.660.23; Op+Cells, 1.7560.1). Our data correlates with the previously reported results where higher degree of anisotropy was observed in osteoporotic bone compared to their healthy controls [22,23]. Similarly, structure model index (SMI) of the trabeculae bone was reported to be an important predictor of changes in micro-architecture of trabeculae in osteoporotic conditions. SMI indicates three-dimensional shape of the trabecular bone. Value of SMI for ideal plate is 0 and for ideal rod is 3 [24]. Transition from plate to rod shape has been reported in osteoporotic and aged bones when compared to the healthy controls [25]. Similarly, our data showed a transition from more rod like structures in Op, Op+Med mice and more plate like in Metaphysial bones were also analyzed for cortical porosity, ratio of total bone volume to tissue volume and bone mineral density (BMD). MicroCT analysis of cortical bones revealed significant decrease in BMD in Op+Cells mice as compared to Op+Med mice, (BMD, g/cm 3

CD34 + cell transplantation elevated serum levels of osteocalcin
As osteocalcin is the characteristic marker of osteoblast function, the levels of osteocalcin in the serum were evaluated to investigate in vivo effects of CD34+ cell therapy on osteoporotic mice. As shown in Figure 5, after osteoporosis, the serum levels of osteocalcin decreased as compared to untreated control mice. However, after CD34 + cell transplantation, the levels of serum osteocalcin (ng/ml) were elevated, as follows: control, 30  In vivo immunolabeling with calcein (green) and alizarin (red) was performed in osteoporotic mice that received CD34 + cell (Op+Cells) or medium (Op+Media) as a control 10 and 3 days respectively before sacrifice of the mice. Femurs were harvested, formalin fixed and then embedded in methylmethacrylate resin. Thirty mm thick sections were mounted on the slides and observed under a fluorescence microscope. Increased mineral apposition rate (MAR) was observed in the endosteal sites of the cortical bone as well as trabeculae, distal from the growth plate, as detected by green fluorescence in mice that received CD34+ cells compared to control (n = 3/group). The values of MAR (mm/day) for the cortical and trabecular bone are shown graphically. doi:10.1371/journal.pone.0039365.g003 increase in the activation of osteoblasts and bone formation, as a consequence of CD34 + cell transplantation in Op mice.

Serum levels of cytokines and growth factors
It was reported that in vitro cultures of osteoblasts produce factors such as granulocyte-macrophage colony stimulating factor (GM-CSF) and interleukin-1 (IL-1) [26]. Multiplex ELISA was performed to analyze the levels of serum cytokines and growth factors implicated to play significant role in bone homeostasis (from Quansys Biosciences, Logan Utah). Out of 20 markers tested (n = 4 mice/group) four (MCP-1, GM-CSF, and IL-10) showed marked changes in all groups ( Figure 5). CD34 + cell transplan- . Interestingly, we observed that levels of IL-10 (pg/ml) were suppressed in Op and Op+Med mice but were dramatically upregulated in Op+Cells mice (IL-10 in control, 2.92560.8; Op, 1.36 0.1; Op+Med, 1.97560.5; Op+Cells, 12.562.3). As there was a high variation among the animals and number of animals was small, results were not statistically significant when posthoc analysis was performed by using Bonferroni correction.

Impaired osteoclast differentiation following CD34 + cell transplantation
Our observations regarding altered levels of cytokines and growth factors and increased bone formation suggested possible effects of CD34 + cell transplantation on differentiation and/or function of bone resorbing osteoclasts. Thus, we investigated any changes in differentiation or function of osteoclasts in Op mice with or without CD34 + cell transplantation. Bone marrows harvested from mice from all groups were allowed to adhere to the plastic overnight in the presence of M-CSF. Non-adherent cells were then induced to osteoclastic differentiation in the presence of M-CSF and RANKL. During the course of differentiation, cells were harvested at various time points (day 3 and 6) and osteoclast specific tartrate resistant acid phosphatase (TRAP) staining was performed to detect cells differentiating into mature osteoclasts. Osteoclasts were defined multinucleated TRAP+ cells. At day 3 and 6, significant numbers of cells were positive for TRAP staining in Op+Med mice. By day 6, extensive numbers of large multinucleated cells were positive for TRAP staining in the osteoporotic animals ( Figure 6A, left panel). However, significantly less numbers of TRAP positive multinuclear cells were observed in animals that received CD34 + cells ( Figure 6A, right panel). Quantification of TRAP+ cells at both days 3 and 6 is shown graphically ( Figure 6A).

Reduced functionality of osteoclasts after CD34 + cell transplantation
Osteoclasts use actin rich, ring-shaped attachment structures called sealing zone instead of focal adhesions for adherence to mineralized substrate. Generation of the sealing zone is requirement for osteoclasts to resorb bone. Therefore, to evaluate the osteoclast function, cells were immunostained for F-actin ring formation using anti-F-actin specific antibody. The cells from bone marrow of Op mice (Op+Med), demonstrated formation of clear F-actin rings ( Figure 6B, upper left panel). Contrarily, clear F-actin ring in the bone marrow cells from Op+Cells mice was not observed. We next assessed whether F-actin ring lacking bone marrow cells from Op+Cells mice could resorb bone, using ivory slices as substrate. As shown in Figure 6B

Discussion
Despite the hematopoietic origin of nanofiber expanded CD34 + cells, herein we show that these cells could be differentiated towards osteogenic lineage. Both hematopoietic and osteogenic cells express similar families of transcription factors such as Cbfa/ Runx with some differences in expression levels of their sub family members [27,28]. These observations led us to believe that hUCB To determine osteoclast functionality on day 4 of differentiation, osteoclasts were harvested and plated on thin ivory slices. Bone resorption assays were performed using osteoclasts from all four groups of animals and to analyze the formation of F-actin rings on ivory bone slices (arrow, upper; left panel). Ivory slices were stained with hematoxylin and analyzed for the resorbed pits on day 10 of differentiation. The formation of F-actin rings was assessed by F-actin specific antibody (green fluorescence). F-actin rings were prominent in osteoclasts derived from Op+Med animals (arrow, upper; left panel), however, osteoclasts derived from Op+Cells animals stained negative for F-actin ring (arrow, upper; right panel). Fewer and smaller resorption pits formed by osteoclasts from Op+Cells animals (arrow, lower; right panel) compared to those formed by animals from Op+Med (arrow, lower; left panel). Evaluated values of pit area/ high power field (HPF) and number of pits/ high power field (HPF) are shown graphical. doi:10.1371/journal.pone.0039365.g006 derived CD34 + cells might also have potential to differentiate in the osteoblastic lineage. Indeed, we have observed that CD34 + cells in response to osteogenic signals, ascorbic acid and bglycerophosphate, expressed Runx2, osteocalcin, osteonectin, collagen1A1, alkaline phosphatase, and BMP, Smad and Wnt signaling molecules, which are characteristically expressed by osteoblasts [29]. Runx2 is required for osteogenic differentiation of mesenchymal cells, whereas osteocalcin, osteonectin and collagen type1A1 are integral to bone matrix [9,10,30]. BMPs expressed by osteoblasts, bind and transduce signals via receptors to activate Smad 1/5/8 signaling pathway, to induce gene expression for bone-associated proteins [10]. Wnt signaling associated proteins Dvl2 and Dvl3 are required to transmit signals during osteoblastic differentiation [31,32]. These findings provide evidence that hUCB derived CD34 + cells are multipotent in nature and microenvironmental cues could drive their differentiation towards osteoblastic lineages.
Characteristic hallmark of osteoporosis is the loss of bone. We have observed that, following dexamethasone treatment mice exhibit significant decreases in trabecular and cortical BMDs compared to controls. Trabecular bone micro architectural parameters such as anisotropy and structure model index also provide strong evidences that dexamethasone was able to induce osteoporosis. This bone loss and changes in bone microarchitecture were almost restored within 28 days following application of a single bolus of hUCB derived CD34 + cells. In osteoporosis, the balance of osteoblast and osteoclast cells is severely compromised. The number of osteoclasts is increased and they obliterate bone micro-architecture, which ultimately results in loss of bone mass [33,34]. This is further accompanied by the bone marrow stromal cells differentiation towards adipogenesis and inhibition of osteoblastogenesis [35]. In the current study, we found the loss of trabecular bone spicules and loss of cortical bones in the osteoporotic mice and concurrent increment of adipocytes. However, application of nanofiber-expanded CD34 + cells induced significant reversion of osteoporosis, by increment of bone formation presumably by osteoblasts and concomitant decrease in number of adipocytes and osteoclasts. In fact, circulating CD34 + cells have been shown to be recruited to the skeletal defect sites and heal non-union fractures in murine models [36,37]. The chemokine receptor, CXCR4 binds to stromal derived factor (SDF)-1 expressed highly in bone marrow stromal cells, and is required for efficient homing of cells to the bone marrow [38]. Since, nanofiber-expanded CD34 + cells constitutively express high levels of CXCR4, this may be critical for homing of these cells to bone marrow. In a preclinical model of osteoporosis, adenoviralmediated overexpression of CXCR4 gene was shown to be necessary for bone marrow derived mesenchymal cells, to home in the bone marrow [39]. However, advantages of using the nanofiber-expanded cells are that we do not need to induce CXCR4 expression using any viral methods, which has potential to integrate to the host genome [40]. Although nanofiberexpanded cells home to the other organs such as lung or spleen besides bone marrow, no bone formation was observed in other organs. Also, recently it was shown that CXCR4 expression on the precursor cells was important for bone formation [41]. Additionally, hUCB derived CD34 + cells have greater advantage in that these cells have not been reported to induce oncogenic transformation in experimental model systems [20].
During osteoporosis, osteoblasts undergo apoptosis, thereby, altering the bone formation and lowering bone mineral density. This is paralleled by lower levels of serum osteoclacin, a marker for bone turnover [42]. Current findings show that CD34 + cell transplantation decreased the numbers of adipocytes in bone marrow, increased trabecular numbers and thickness, and increased BMD, thereby indicating induction of osteoblast in bone. An increase in the osteocalcin levels coupled with an increase in vivo mineral apposition rate in the mice that received CD34 + cell transplantation, confirms the regenerative potential of CD34 + cells in bone formation.
Elevated levels of GM-CSF, and IL-10 and decreased level of MCP-1 in serum were also evident with the CD34 + cell transplantation. Previous in vitro studies have shown that in the presence of CD34 + cells, osteoblasts secrete GM-CSF and cytokines such as IL-6, IL-4 and IL-1 involved in bone homeostasis [15,43,44]. Current findings suggest that CD34 + cells likely modulate cytokine secretions during induction of osteogenesis. In this respect, we have observed high level of IL-10 in serum of the mice that received CD34 + cells. The importance of IL-10 in bone metabolism was demonstrated in IL-10 deficient mice, which develop osteopenia and decreased bone formation [45,46]. IL-10 exhibits anti-osteoclastic activity and may directly inhibit osteoclast precursor cell differentiation [47]. CD34 + cell transplantation thus acts by upregulating the levels of IL-10 that may in turn regulate the bone remodeling by impairing osteoclastogenesis and restoring the bone formation [46,47].
Above findings thus demonstrate a novel therapeutic potential of nanofiber-expanded CD34 + cells in resolving osteoporosis. These cells provide several advantages over other cell types as they are easily accessible, less immunogenic and maintain multipotency. They can be available in unlimited numbers via expansion on nanofibers; constitutively express CXCR4 for efficient homing to the bone marrow; and differentiate into osteogenic lineage in vitro. These cells are biologically functional as they significantly improve bone mineral density, bone formation and bone micro-architecture in osteoporotic murine model. These characteristics make them an attractive choice of cells for bone induction in patients with severe bone loss due to osteoporosis.

Isolation of CD133 + hematopoietic cells
Fresh human umbilical cord blood was obtained from The Ohio State University Medical Center with written consent from donors and prior approval from the internal review board. Briefly, heparinized cord blood was diluted with PBS (1:1) and carefully layered over 10 ml of Ficoll-Paque (GE Health Care-Biosciences, USA). After 30-min centrifugation in a swinging bucket rotor at 1400 rpm, the upper layer was aspirated and the mononuclear cell layer (buffy coat) was collected. Following labeling with magnetic bead conjugated anti-CD133 monoclonal antibody (Miltenyi Biotec Inc, Bergisch Gladbach, Germany), two cell separation cycles (with different columns) were performed using the AutoMACS cell sorter (Miltenyi Biotec) following manufacturer's protocol and reagents. After separation, purity of the cell product was determined by flow cytometry (more than 95% cells were CD133 + ). Total 4-6 samples of cord blood were procured, isolated and used nanofiber mediated expansion and in vitro and in vivo experiments.

Expansion of CD133 + hematopoietic cells
Electrospinning, surface grafting, and amination of PES nanofibers were carried out according to the procedure described earlier [48]. All chemicals were purchased from Sigma-Aldrich (USA) unless otherwise stated. PES granules were purchased from Goodfellow Cambridge Limited, UK. CD133 + cells were expanded on nanofiber-coated plates using serum free expansion medium (SFEM) as previously described [19]. In brief, purified recombinant human stem cell factor (SCF), Flt-3 ligand (Flt3), thrombopoietin (TPO), and IL-3 were purchased from Peprotech Inc. (Rocky Hill, NJ). The StemSpan SFEM medium was purchased from StemCell Technologies (Vancouver, BC, Canada). Nanofiber meshes were securely glued to the bottoms of wells of a 24-well tissue culture plate. Eight hundred CD133 + cells were seeded onto each scaffold in 0.6 ml StemSpanTM serum-free expansion medium, which consists of 1% BSA, 0.01 mg/ml recombinant human insulin, 0.2 mg/ml human transferrin, 0.1 mM 2-mercaptoethanol, and 2 mM L-glutamine in Iscove's MDM, supplemented with 0.04 mg/ml low-density lipoprotein (Athens Research and Technology Inc., USA), 100 ng/ml SCF, 100 ng/ml Flt3, 50 ng/ml TPO, and 20 ng/ml IL-3. Cells were cultured at 37uC in an atmosphere containing 5% CO 2 for 10 days without medium change. Cells were harvested after 10 days of expansion. All wells were washed once with nonenzymatic cell dissociation solution and twice with 2% FBS containing Hanks' buffer at 5 min intervals between each wash. The collected cells were then concentrated through centrifugation at 3006 g for 5 min. Aliquots of the cells were then used for cell counting by a hemocytometer, flowcytometric analysis, as well as for further studies.

Flowcytometry
Flowcytometric analysis was performed by using standard two colors staining with a FACS Calibur flowcytometer (Becton Dickinson, Heidelberg, Germany) as described earlier [19]. Nonspecific Fc-receptors were blocked with FcR-blocking reagent (Miltenyi Biotec Inc.) prior to adding primary antibodies. Primary antibody was incubated for 30 min at 4uC with the aliquots of expanded cells. Antibodies used were anti-CD34-PE, anti-CD133/2-FITC (all from Miltenyi Biotec Inc), PE labeled CXCR4, von Willebrand Factor, CD31, CD45, MHC class I, MHC class II, CD69, CD3, Mac-I, LFA-1, CD86, CD14 and isotype controls were purchased from BD Biosciences (USA) After incubation cells were washed with MACS buffer and resuspended in MACS buffer. Dead cells were excluded via propidium iodide staining. Data analysis was performed with BD Cell Quest software. The Milan-Mulhouse gating method was used for cell enumeration, where a double gating (CD133 + and CD34 + ) strategy was used to identify the primitive hematopoietic progenitor cell populations. At least 20,000 events were acquired.
Osteoblastic differentiation of nanofiber-expanded CD34 + cells After 10 days of CD133 + cell expansion on nanofiber, cells were collected and reseeded in DMEM supplemented with 10% fetal bovine serum (FBS), penicillin, streptomycin, and glutamate (PSG) for three days in a 24-well plate or 4-chamber slides (Lab-Tek II Chamber slide System, Nalge Nunc International Corp., Naperville, IL, USA). The cultured cells were induced by specific osteoblast differentiating factors (60 mM ascorbic acid and 10 mM b-glycerophosphate) dissolved in fresh DMEM complete medium. The differentiation process was continued for 3 weeks with a change of fresh medium containing specific osteoblast inducing factors in every 3rd day. During the process of differentiation, some of the wells of differentiated cells were harvested at specific time points for further studies (n = 4).

Alizarin red S staining
To examine in vitro mineralization in osteoblastic differentiated cells, Alizarin Red S staining was performed in wells of a 24-well plate. Briefly, after 21 days of differentiation cells were fixed in 70% ethanol for one hour at room temperature followed by a wash with water (10 minutes, 2 times) and incubated with 1% Alizarin Red S solution for 30 minutes at room temperature (Chemicon International, USA). The red stain was washed with water three times and cells were mounted with mounting solution. Images were obtained by using a digital camera attached to the microscope (Axioplan2; Carl Zeiss) using Axio-vision software (Carl Zeiss, NY, USA). Nanofiber-expanded CD34 + cells cultured in DMEM complete medium were used as a control.

Immunocytochemical staining
One quarter of a million nanofiber-expanded CD34 + cells were induced to differentiate in each well of 4-chamber slides for three weeks. After three weeks of culture immunocytochemistry was performed following standard protocol to assess osteoblastic differentiation using osteoblast specific markers such as Smad 1/ 5/8, osteocalcin and IgG as an isotype control. Nuclear stain was performed with DAPI. Slides were observed under a fluorescence microscope (Axioplan2; Carl Zeiss) and images were captured with Zeiss Axiovision imaging software (Carl Zeiss, NY, USA).

RT-PCR and western blot analyses
Nanofiber-expanded CD34 + cells were induced to osteoblastic differentiation in vitro. During the course of osteoblastic differentiation, total RNA and proteins were isolated at various time points using RNeasy Kit (Qiagen, USA) and RIPA, cell lysis buffer respectively. One microgram of RNA was used for cDNA synthesis using oligo dT (Invitrogen, USA) primer. Semi quantitative PCR was performed using one micro liter of cDNA for the gene specific primers such as Runx2, alkaline phosphatase, osteocalcin, osteonectin, collagen 1A1 and GAPDH. Fold increase for expression of each gene at mRNA level was analyzed using UN-SCAN-IT software (Silk Scientific, Inc. Similarly, Western blot was performed for the level of BMP2, BMP4, Smad-1/5/8 (all from Santa Cruz Biotechnology, Inc., CA), Dvl2, Dvl3, and GAPDH (all from Cell Signaling Technology, Inc. MA) protein. Nanofiber-expanded cells were used as a control. Induced differentiated MC3T3 cells were also used as a positive control. Fold increase for expression of each gene at protein level was analyzed using UN-SCAN-IT software (Silk Scientific, Inc.).

Generation of osteoporosis
Seven month-old female NOD/SCID mice, retired breeder (body weight approximately 25 g) were obtained from Jackson Laboratory, Bar Harbor, ME and used for osteoporosis induction. The mice were housed in sterile IACUC-approved facilities at the Biomedical Research Tower at The Ohio State University. After a week of acclimatization, mice were intraperitoneally (i.p.) injected with 5 mg/ kg body weight (b. wt.) of dexamethasone (American Regent, Inc. Shirley, NY) or with saline (as a control) for consecutive 21 days as described before [35]. Tapering doses of dexamethasone were given for 5 days for withdrawal. The mice were weighed every week during the injection to record any significant weight loss.

CD34+ cell transplantation
The mice, that received glucocorticoid injections were either sacrificed after 26 (21+5) days as an osteoporotic control or were used for CD34+ cell / medium injections (6 mice/group). The osteoporotic mice were anesthetized and injected with nanofiberexpanded CD34+ cells (0.5 million in 300 ml of serum free medium) via intra-cardio-ventricular injection or 300 ml of serum free medium (Op+Med) as a control for CD34+ cell therapy group. The osteoporotic mice of CD34+ cell therapy (Op+Cells) group received only one dose of CD34+ cells and were sacrificed after 28 days of injection. Osteoporotic mice receiving medium (Op+Med) were also sacrificed at the same time point.

CD34 + cell homing
To assess the homing of CD34+ cells, GFP overexpressed nanofiber-expanded CD34 + cells (0.5 million cells/mouse) were injected into the mice via intra-cardio-ventricular injection after induction of osteoporosis. Forty-eight hours and two weeks after cell injection, mice were sacrificed and various organs were harvested. Tissues were fixed in 10% formalin solution and subsequently embedded in paraffin block. Five-micron sections were cut and stained using anti-GFP primary antibody (Zymed Laboratories, Inc, CA, USA) and detected with 3,39-Diaminobenzidine (DAB) and mounted. Slides were observed under a microscope (Axioplan2; Carl Zeiss) and images were captured with Zeiss, Axiovision imaging software (Carl Zeiss).

Mineral apposition rate (MAR)
A separate set of seven months old NOD/ SCID mice (3 mice / group) were subjected to generation of osteoporosis (21+5 days) and followed by CD34 + cell transplantation or medium (as a control) for 28 days as described above. Mice were intraperitoneally (i.p.) injected with calcium binding dye, calcein (green flourescent dye; 10 mg/kg b.wt., Sigma); 10 days prior to sacrifice. Alizarin Red Dye (red fluorescent dye; 50 mg/kg b.wt., Sigma) was also injected to the same mice via i.p. route three days prior to sacrifice. Upon sacrifice, femurs were harvested and formalin fixed for 24 hours and then washed with 1x PBS, dehydrated with 70% ethanol, 95% ethanol, 100% ethanol, infiltrated and embedded in methylmethacrylate and polyester resin. Thirty-micrometer thick sections were cut with a diamond blade using Saw Microtome Leica SP1600 (Leica Microsystem, Wetzlar, Germany). The sections were mounted on the slides using Vecta Mount (Vector Labs, Inc. Burlingame, CA) and observed under a fluorescence microscope (Carl Zeiss, NY, USA) and images were captured with Zeiss Axiovision imaging software (Axioplan2). Images were analyzed to assess the mineral apposition rate (MAR) by using NIH Image J software. The mineral apposition rate (MAR, in mm per day) was determined by dividing the mean width of the double labels by the inter-label time as described previously [49].

Hematoxylin and eosin staining
All mice were euthanized after collecting blood and urine, and femurs were removed by surgical dissection and were preserved in 10% neutral buffered formalin. After decalcification, paraffinblocked tissues were sectioned and stained with standard hematoxylin and eosin (H&E) staining protocol at the same depth from the growth plate.

Micro computed tomography (MicroCT)
Formalin fixed femur bones were encased in a tight fitted plastic tube to prevent any motion during scanning. Femurs were scanned using a high-resolution Micro CT scanner (SkyScan1172-D, Kontich, Belgium) at 16mm resolution. For measurements of bone mineral density (BMD; g/cm 3 ), phantoms of 25 and 75 g/cm 3 were scanned with the same settings as applied to the mice femur bones from all groups. The scanned images were reconstructed using Skyscan Nrecon software and analyzed with the CTan software (Kontich, Belgium). Analyses of trabecular bone were carried out in distal femur, 0.1 mm away from the growth plate. A threshold was established and the same threshold value was kept constant for all samples. Similarly, the metaphysial region, 2mm away from the growth plate, was selected for the cortical bone analysis. A separate threshold was established and kept constant for all cortical bone measurements. The analyses were performed for the bone volume, bone to tissue volume ratios, trabecular number (Tb.N; 1/mm), trabecular thickness (Tb.Th; mm), trabecular bone mineral density (BMD; g/cm 3 ), degree of anisotropy (DA), structure model index (SMI), cortical porosities (%), ratio of bone volume/ tissue volume (BV/TV) and cortical bone mineral density (BMD; g/cm 3 ) in animals from all groups using the Skyscan software and following manufacturer's protocol. The results were reported in mean 6 SEM (n = 6/ per group). Three-dimensional (3D) models were reconstructed using CTvol software from SkyScan.

Serum osteocalcin assay
Sandwich ELISA was performed for the collected mouse serum using the osteocalcin ELISA kit (Biomedical Technologies, Inc, Staughton, MA). Briefly, the mouse was anesthetized and blood was collected from the descending aorta in an eppendorf tube and kept at room temperature for an hour to coagulate, then centrifuged at 14000 rpm for 20 minutes at 4uC. Serum was collected and stored at 280uC until further use. Standard curve was made and the serum samples were run in triplicates to assess the amount of osteocalcin in the serum. The samples from each group were diluted (5x) and incubated with the anti-serum osteocalcin overnight and detected using streptavidin-horseradish peroxidase (HRP) detecting system. The values were reported in mean 6 SEM (n = 4/ per group, in triplicate).

Multiplex ELISA for cytokines and growth factors
To assess the levels of various cytokines and growth factors (twenty factors from 250 ml of serum) from collected serum (from all groups of animals before sacrifice and stored at 280uC), the multiplex ELISA was performed in triplicate by Quansys Biosciences, Logan Utah (n = 4/ per group). The values were reported in mean 6 SEM.

Osteoclast differentiation
To assess the therapeutic effects of CD34 + cells on the bone resorbing cells; bone marrow cells were collected from femurs of animals of all the groups, after termination of experiments and were induced for osteoclastic differentiation in vitro. Cells were cultured overnight at 37uC incubator with 5% CO 2 in aMEM containing 10% heat inactivated fetal bovine serum in the presence of 20 ng/ml M-CSF (R & D Systems, Minneapolis, MN). The next day, same number of non-adherent cells (1.5 million for all the groups) were collected and incubated for an additional 5-8 days in aMEM medium with 20 ng/ml M-CSF, and 50 ng/ml GST-RANKL [50]. The fresh medium was replaced every third day. At day 3 and 6 of differentiation, the cells were stained for TRAP staining using an acid phosphatase, leukocyte; TRAP staining kit (Sigma Aldrich, USA) and was viewed and imaged with a fluorescence microscope (Carl Zeiss, NY, USA). TRAP-positive cells (purple) containing at least three nuclei were counted as osteoclasts.

Osteoclast cytoskeleton structure and functionality
Osteoclasts were generated on plastic dishes as described above and on 3 rd day of differentiation, osteoclasts cells were removed and equal number of osteoclasts cells were re-plated either on thinly cut ivory slices or glass cover slips. Cells were fixed at various time points of culture with 1% formaldehyde in pH 6.5 (30 minutes at room temp), stabilization buffer (127 mM NaCl, 5 mM KCl, 1.1 mM NaH 2 PO 4 , 0.4 mM KH 2 PO 4 , 2 mM MgCl 2 , 5.5 mM glucose, 1 mM EGTA, 20 mM Pipes), and subsequently fixed and permeabilized with 2% formaldehyde, 0.2% Triton X-100, and 0.5% deoxycholate in the same stabilization buffer. Cells were stained with F-actin specific Ab and visualized using a Zeiss 510 META laser scanning confocal microscope (Campus Microscopy and Imaging Facility, The Ohio State University). Actin ring and podosome thicknesses were determined by generating Z-stack images of randomly selected cells and these structures were measured at their thickest points [50]. Bone resorption was assessed using ivory slices and osteoclasts were gently removed with cotton swabs and washed with water. The ivory slices were then stained with hematoxylin stain for 5 minutes at room temperature and excess stain was removed by washing with water and pits were imaged with a confocal microscope mentioned above.

Statistical analysis
Values were expressed as mean 6 SEM and statistical analysis was performed by using JMP software (version 9, SAS Institute Inc. NC). After checking equal variance two sample 't'-test was performed for assessment of significance. Posthoc analysis was performed by using Bonferroni correction and significance was determined when p values were obtained less than 0.025.