Role of donor and host cells in muscle-derived stem cell-mediated bone repair: differentiation vs. paracrine effects

Abstract

Murine muscle-derived stem cells (MDSCs) have been shown capable of regenerating bone in a critical size calvarial defect model when transduced with BMP 2 or 4; however, the contribution of the donor cells and their interactions with the host cells during the bone healing process have not been fully elucidated. To address this question, C57/BL/6J mice were divided into MDSC/BMP4/GFP, MDSC/GFP, and scaffold groups. After transplanting MDSCs into the critical-size calvarial defects created in normal mice, we found that mice transplanted with BMP4GFP-transduced MDSCs healed the bone defect in 4 wk, while the control groups (MDSC-GFP and scaffold) demonstrated no bone healing. The newly formed trabecular bone displayed similar biomechanical properties as the native bone, and the donor cells directly participated in endochondral bone formation via their differentiation into chondrocytes, osteoblasts, and osteocytes via the BMP4-pSMAD5 and COX-2-PGE2 signaling pathways. In contrast to the scaffold group, the MDSC groups attracted more inflammatory cells initially and incurred faster inflammation resolution, enhanced angiogenesis, and suppressed initial immune responses in the host mice. MDSCs were shown to attract macrophages via the secretion of monocyte chemotactic protein 1 and promote endothelial cell proliferation by secreting multiple growth factors. Our findings indicated that BMP4GFP-transduced MDSCs not only regenerated bone by direct differentiation, but also positively influenced the host cells to coordinate and promote bone tissue repair through paracrine effects.—Gao, X., Usas, A., Proto, J. D., Lu, A., Cummins, J. H., Proctor, A., Chen, C.-W., Huard, J. Role of donor and host cells in muscle-derived stem cell-mediated bone repair: differentiation vs. paracrine effects.

The treatment of bone fractures, nonunions, and large bone defects remains a tremendous clinical challenge. Recently, innovative therapies that promote bone repair have been successfully explored, such as the combinatorial use of scaffolds, growth factors, and adult stem cells. Muscle-derived stem cells (MDSCs), isolated via the modified preplate technique from skeletal muscle, represent a population of adult-derived stem cells that possess the ability to differentiate into multiple cell lineages, including osteogenic cells. We have shown that murine MDSCs transduced with bone morphogenetic protein 2 (BMP2) or BMP4 are capable of differentiating toward an osteogenic lineage and promoting bone healing in both ectopic bone formation and cranial defect models (1, 2). Our group and others have also demonstrated that human muscle-derived cells, isolated by different techniques, could undergo osteogenesis in vitro and promote bone formation in vivo (3–5). Moreover, we recently demonstrated that human MDSCs transduced with lenti-BMP2 could undergo osteogenesis in vitro and heal a critical size bone defect in vivo (6).

Angiogenesis plays an important role in MDSC-mediated bone regeneration, and it has been shown that the implantation of murine MDSCs expressing both BMP4 or BMP2 and VEGF, a proangiogenic protein, could increase angiogenesis and enhance bone regeneration. Conversely, blocking angiogenesis by implanting MDSCs that express the VEGF antagonist, soluble fms-like tyrosine kinase-1(sFlt1) reduces the process of bone formation (7, 8).

Despite the progress that has been made in understanding the role that MDSCs play in the bone regeneration process, it remains largely unknown to what degree the donor MDSCs directly contribute to the regenerated bone structure, as well as the mechanisms by which the donor MDSCs interact with the host cells to promote bone healing. Until now, it remained unclear what roles the transplanted adult stem cells and host cells played in stem cell-mediated bone repair. The implantation of mesenchymal stem cells (MSCs) has been shown to promote bone repair by enhancing the migration of CD31⁺ and CD146⁺ cells (9), while another study found that the MSCs enhanced the recruitment of inflammatory cells (10). Therefore, a more detailed investigation into the role that the donor and host cells play during the process of adult stem cell-mediated bone regeneration is important to understand the mechanism by which bone repair occurs after injury.

In this study, we investigated the roles that both the donor MDSCs and the host cells played in promoting bone repair, as well as the involvement that certain molecular pathways had in the regeneration processes. We hypothesized that BMP4/green fluorescent protein (BMP4GFP)-transduced MDSCs transplanted into a critical size calvarial bone defect could differentiate into an osteogenic lineage and contribute to bone formation, while simultaneously chemoattracting and influencing the host cells to coordinate the formation of functional bone.

MATERIALS AND METHODS

Isolation of murine MDSCs, vector construction, and retroviral transduction

Murine MDSCs were isolated from the hind-limb muscle of 3-wk-old C57BL/10J mice via the modified preplate technique (11). A retroviral vector containing human BMP4 and GFP separated by an internal ribosome entry site (IRES) and under the control of the human CMV promoter, which allowed for the expression of BMP4 and GFP as individual proteins, was constructed as described previously (7, 8). The addition of the GFP tag allowed us to track the donor cells in vivo. Retro-GFP using the same vector served as a control. MDSCs were transduced with retro-GFP or retro-BMP4/GFP at a multiplicity of infection of 5 in the presence of 8 μg/ml polybrene (Sigma-Aldrich, Milwaukee, WI, USA). The transduction was repeated 3 times over a 24-h period. The cells were then expanded for 4 passages, after which the GFP⁺ cells were selected by fluorescence-activated cell sorting (FACS; BD FACSAria IIu; BD Biosciences, Bedford, MA, USA). The secretion of BMP4 in the supernatant of the BMP4GFP-transduced MDSCs and GFP-transduced MDSCs was measured with the Human BMP-4 Quantikine ELISA kit (DBP400; R&D Systems, Minneapolis, MN, USA), and the amount of BMP4 released was calculated (ng/million cells/24 h). Alkaline phosphatase (ALP) staining was performed at 8, 12, and 16 passages post-transduction, while the cells were cultured in nonosteoinductive proliferation medium (PM), using a Sigma-Aldrich 86C kit according to the manufacturer's instructions.

Semiquantitative RT-PCR, Western blot analysis, and ELISA

GFP and BMP4GFP-transduced MDSCs were harvested at different passages and lysed with Qiagen RLT lysis buffer (Qiagen, Valencia, CA, USA) supplemented with 10 μM β-mercaptoethanol. Total RNA was extracted using the RNeasy minikit (Qiagen), according to the manufacturer's protocols. Reverse transcription was performed using a Superscript III reverse transcriptase kit (Invitrogen; Life Technologies, Carlsbad, CA, USA). The cDNA was diluted with DNase- and RNase-free water and stored at −20°C for further PCR amplification. The PCR primers were designed using Primer3 (12), and the PCR was performed as a 25-μl reaction with the Gotaq PCR system (Promega, Madison, WI, USA). The PCR products were electrophoresed on a 1% agarose gel, and the images were captured using GelDoc with QuantOne software (GelDoc system; Bio-Rad, Hercules, CA, USA). The primer information is listed in Table 1.

Table 1.

Primer information

Gene	Accession no.	Primers, 5′–3′	Product size (bp)	Annealing temperature (°C)
Sox-9	NM_011448.3	F: CGGACAAGCGGAGGCCGAAG	276	55
		R: CGTCGCGGAAGTCGATGGGG
Bmpr1b	NM_007560.3	F: ACCTGGTGCCCAGTGACCCT	171	55
		R: TCAGGGCCGTCAGCCTGGAG
Bmpr2	NM_007561.2	F: GGCGTGTGCCAAAAATCGGGC	263	55
		R: TGGAGTGAGGCCGGTGGTGT
Cox-2	NM_011198.3	F: GGGCCCTTCCTCCCGTAGCA	232	55
		R: CCATGGCCCAGTCCTCGGGT
Inos	NM_010927.2	F: CGCTGGCTACCAGATGCCCG	289	55
		R: CGAGGCCACCCACCTCCAGT
Mcp1	NM_011333.3	F: CACAGTTGCCGGCTGGAGCA	147	55
		R: TTGGGACACCTGCTGCTGGT
Vegfa	NM_001025250.2	F: CGCCGCAGGAGACAAACCGAT	151	53
		R: ACCCGTCCATGAGCTCGGCT
Hifα1	NM_010431	F: TCAAGTCAGCAACGTGGAAG	198	55
		R: TATCGAGGCTGTGTCGACTG
Fgf2	NM_008006.2	F: AGCGGCTCTACTGCAAGAAC	183	55
		R: GCCGTCCATCTTCCTTCATA
Pdgfβ	NM_011057.3	F: GATCTCTCGGAACCTCATCG	138	55
		R: GGCTTCTTTCGCACAATCTC
Tgfβ1	NM_011577.1	F: TGAGTGGCTGTCTTTTGACG	293	55
		R: TCTCTGTGGAGCTGAAGCAA
Igf1	BC012409	F: TGGATGCTCTTCAGTTCGTG	173	55
		R: GCAACACTCATCCACAATGC
Igf2	M14951.1	F: GTCGATGTTGGTGCTTCTCA	185	55
		R: AAGCAGCACTCTTCCACGAT
Gapdh	BC145812.1	F: CCGGGGCTGGCATTGCTCTC	208	55
		R: GTGTTGGGGGCCGAGTTGGG

F, forward; R, reverse.

Cell lysates were prepared in radioimmunoprecipitation assay (RIPA) buffer (no. 9806; Cell Signaling Technology, Danvers, MA, USA) supplemented with protease (P8340) and phosphatase inhibitors (P5726 and P0044, 1:100; Sigma-Aldrich). Protein concentration was quantified using a Bio-Rad protein assay kit 2 (500–0002). Western blot analysis was performed using rabbit anti-pSMAD1/5/8 (9511S; 1:1000 dilution; Cell Signaling Technology), goat anti-mouse cyclooxygenase 2 (COX-2; sc-1746; 1:50 dilution; Santa Cruz Biotechnology, Santa Cruz, CA, USA), rabbit anti-runt-related transcription factor 2 (RUNX2; 1:1000; Cell Signaling Technology), goat anti-monocyte chemotactic protein 1 [MCP1; also known as chemokine (C-C motif) ligand 2; sc-1784; 1:200 dilution; Santa Cruz Biotechnology], rabbit anti-vascular endothelial cell growth factor a (VEGFa; ab46154; 1:200 dilution; Abcam, Cambridge, MA, USA), rabbit anti-platelet-derived growth factor β (PDGFβ, sc-7878; 1:50 dilution; Santa Cruz Biotechnology), and mouse anti-β-actin (A5441; 1:8000 dilution; Sigma-Aldrich) primary antibodies to detect MCP1, VEGFa, PDGF, and β-actin, respectively. Horseradish peroxidase-conjugated rabbit anti-goat (31402; 1:5000 dilution; Pierce, Rockford, IL, USA), rabbit anti-mouse (31450; 1:10,000 dilution; Pierce), and goat anti-rabbit (31460; 1:5000 dilution; Pierce) secondary antibodies were used to detect goat, mouse, and rabbit primary antibodies. Supersignal Western Pico Chemilumnescent Substrate (Pierce) was used to reveal the target protein bands. A FOTO/analystFX (Fotodyne, Hartland, WI, USA) system was used to capture digital images. The level of MCP-1 protein in the conditioned medium was measured with a Mouse MCP-1 colorimetric ELISA kit (EMMCP1; Thermo Scientific, Waltham, MA, USA), according to the manufacturer's protocol.

Animal experimental design, creation of critical size calvarial defect, and cell implantation

All animal protocols were approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh. For this study, we utilized a previously described 5-mm critical-sized cranial defect model created in the right parietal bone of mice (13). Fibrin sealant (Tisseel; Baxter, Westlake Village, CA, USA) was used as a scaffold for all the in vivo experiments. Male C57BL/6J mice (Jackson Laboratories, Bar Harbor, ME, USA) were used for this project and were divided into 3 groups: scaffold + PBS (scaffold); scaffold + retro-GFP-transduced MDSCs (5×10⁵ cells) in PBS (MDSC/GFP); and scaffold + retro-BMP4GFP-transduced MDSCs (5×10⁵ cells) in PBS (MDSC/BMP4/GFP). Following the creation of the defect, the PBS-, GFP-, or BMP4GFP-transduced MDSCs were mixed with 15 μl of thrombin and immediately implanted into the defect. Then, 15 μl of fibrin sealant (Tisseel; Baxter) was added and allowed to solidify for 1–2 min prior to closing the scalp with sutures, and the animals were allowed to recover in an oxygen chamber. The mice were euthanized at 3, 7, 14, 21, and 28 d after transplantation (n=5 for each time point and group).

Micro-computed tomography (microCT) and biomechanical analysis

Bone regeneration within the calvarial defect was monitored via microCT (Viva CT 40; Scanco Medical, Brüttisellen, Switzerland) at 1, 2, 3, and 4 wk postsurgery. After obtaining 2-dimensional image slices, the view of interest (VOI) was uniformly delineated, and 3-dimensional reconstructions were created using an appropriate threshold that was kept constant throughout the analyses. The de novo bone volume was quantified using the software provided in the Viva CT 40 system by contouring every slice of the new bone area. Three-dimensional bone volume was measured using Gauss σ 0.8, Gauss Support 1, and threshold at 163. Bone defect coverage was measured using ImageJ software. The microCT parameters and terminology utilized followed the guidelines of the American Society of Bone and Mineral Research (14). Four skull samples harvested from the MDSC/BMP4/GFP group at 28 days postimplantation (dpi) were used for biomechanical testing using a Biodent System (Active Life Scientific, Santa Barbara, CA, USA). Contralateral normal parietal bone was used as the control. The parameters of indentation distance increase (IDI) and total indentation distance (TID) were used to represent the biomechanical (hardness) properties of the new bone.

Histological evaluation of calvarial defect bone regeneration

The skulls of the animals sacrificed at 3, 7, 14, and 21 d were dissected, embedded in NEG50 freezing medium (Fisher Scientific, Pittsburgh, PA, USA), snap-frozen in liquid nitrogen, and stored at −80°C. Skull tissues were serially cryosectioned (10 μm thickness). Alcian blue staining was performed to determine the presence of endochondral bone formation, according to an online protocol (IHC World; http://www.ihcworld.com/_protocols/special_stains/alcian_blue.htm). Von Kossa staining was performed to detect mineralization within the regenerated tissues (IHC World). Herovici's staining was used to reveal the formation of major bone matrix-collagen type I (15), and hematoxylin and eosin (H&E) staining was performed to reveal the bone morphology at the 28-d time point.

Immunofluorescence and immunohistochemistry

Briefly, cryosections were fixed with 4% paraformaldehyde (PFA; Sigma-Aldrich) and blocked with 5% donkey serum. Sections were subsequently incubated with primary targeted protein antibody together with GFP primary antibody overnight at 4°C. The following day, the sections were incubated with the corresponding secondary antibodies for 2 h at room temperature. Finally, sections were counterstained with 4,6-diamino-2-phenylindole (DAPI) and coverslipped with Permafluor mounting medium (TA030FM; Fisher Scientific). The primary antibodies used in this study included goat anti-mouse collagen type 2 A1 (COL2A1; sc-1747; 1:50 dilution; Santa Cruz Biotechnology), goat anti-mouse osteocalcin (OC; sc-23790; 1:50 dilution; Santa Cruz Biotechnology), rabbit anti-GFP (ab290; 1:1000 dilution; Abcam), goat anti-GFP (ab6673; 1:200 dilution; Abcam), rabbit anti-pSMAD5 (ab92698; 1:100 dilution; Abcam), goat anti-mouse COX-2 (sc-1746; 1:50 dilution; Santa Cruz Biotechnology), rat anti-mouse CD68 (ab53444; 1:100; Abcam), rat anti-Gr-1 (BD557445; 1:100 dilution; BD Biosciences), rat anti-mouse CD31 (BD553370; 1:300 dilution; BD Biosciences), rat-anti-mouse CD4 (BD 550280; 1:50 dilution; BD Biosciences), and rat anti-mouse CD8α (ab3081; 1:50 dilution; Abcam) antibodies. The secondary antibodies included donkey anti-goat-594-DyLight (705-515-147; 1:200 dilution; Jackson ImmunoResearch Laboratories, West Grove, PA, USA), donkey anti-rabbit-IgG 649 DyLight (711-495-152, 1:200 dilution; Jackson ImmunoResearch Laboratories), donkey anti-rabbit-IgG Cy2 (711-225-152; 1;100 dilution; Jackson ImmunoResearch Laboratories), and donkey anti-rat IgG-DyLight-594 (705-515-150; 1:200 dilution; Jackson ImmunoResearch Laboratories). To colocalize GFP with COL2A1, OC, COX-2, pSMAD5, CD68, Gr-1, CD31, CD4, and CD8 at the 7, 14, and 21 d time points, we used donkey anti-rabbit-649 DyLight to reveal GFP-positive donor cells, as this channel exhibited much less autofluorescence. For the 3 d time point sections, we used donkey anti-rabbit-IgG Cy2 to reveal the GFP donor cells. Calvarial bone samples harvested 28 dpi were fixed in 4% PBS buffered formaldehyde for 24 h and decalcified with 10% EDTA plus 1% sodium hydroxide (pH 7.0) for 3 wk. Skull tissues were dehydrated through an ethanol gradient, cleared with xylene, and paraffin embedded. Immunohistochemistry was then carried out on 5 μm paraffin sections. After deparaffinization, washing, and blocking with 5% donkey serum in PBS, sections were incubated with rabbit anti-GFP antibody (ab290; 1:1000 dilution; Abcam) overnight. The following day, sections were treated with 0.5% H₂O₂ in PBS for 30 min at room temperature, washed in PBS, and incubated with goat anti-rabbit-biotin (BA 1000; Vector Laboratories, Burlingame, CA, USA) for 2 h at room temperature. After 3 washes, the sections were then incubated with ABC reagents (PK 7200, Elite ABC kits; Vector Laboratories) for 2 h at room temperature. Diaminobenzidine (DAB) staining (SK-4100; Vector Laboratories) was used to visualize the GFP-positive cells. Hematoxylin (H-3404; Vector Laboratories) counterstaining was performed following the DAB reaction.

Immunofluorescent staining of the cultured cells was performed by fixing the cells in cold methanol for 2 min and then incubating the cells with the following primary antibodies: goat anti-MCP1 (sc-1784; 1:25 dilution; Santa Cruz Biotechnology), rabbit anti-VEGFa (ab46154; 1:50 dilution; Abcam), rabbit anti-PDGFβ (sc-7878; 1:10 dilution; Santa Cruz Biotechnology). The remaining steps were performed as described above for immunofluorescent staining of the cryosections.

Immunofluorescent images were acquired using Northern Eclipse (Empix Imaging, Cheektowaga, NY, USA), and bright-field pictures were acquired using Q-Capture (QImaging, Surrey, BC, Canada) on a Nikon Eclipse E800 microscope (Nikon, Melville, NY, USA). To quantify different cell numbers, a total of 6 pictures (×200) were taken for each sample using sections from 3 different axial levels (300 μm apart) and analyzed with ImageJ (U.S. National Institutes of Health, Bethesda, MD, USA) cell counter software. The number of cells within the defect area of the ×200 field was counted, and the defect area was measured on each picture. The average cell number of each sample was then calculated. The cell number in the defect region was normalized to the ×200 field area for each picture from all groups. The normalized number of CD68-, Gr-1-, CD31-, CD4-, and CD8-positive cells in each group was monitored and expressed as the mean ± sd.

Chemotaxis assays

GFP- or BMP4GFP-transduced MDSCs (1.5×10⁵) were seeded in T175 flasks and cultured in 18 ml PM for 36 h. The medium from the GFP- and BMP4GFP-transduced MDSC cultures was centrifuged at 2000 rpm for 5 min, portioned into aliquots, and stored at −80°C for future analyses. Chemotaxis assays were performed using a ChemoTx disposable chemotaxis system (116-8; NeuroProbe, Gaithersburg, MD, USA) to detect the chemoattraction of macrophages using the isolated conditioned culture medium. We placed the conditioned medium from both GFP- and BMP4GFP-transduced MDSCs in the lower chamber (well) of the plates and plated 3 × 10⁵ RAW264.7 macrophage cells in PM above the membrane. Four replicates were performed for each treatment group. After incubating the plates for 18 and 24 h, the filters were rinsed with Versene solution (Sigma-Aldrich), and the remaining cells were removed with a cell scraper. The plate was then centrifuged at 400 g for 10 min to detach any cells that had migrated through but remained attached to the underside of the filter. Finally, we measured the relative number of cells that migrated to the lower chamber of the well using an MTS assay (Promega), according to the manufacturer's instructions. To determine whether MCP-1 played a role in the chemotaxis of the macrophages, we set up an experiment in which the conditioned media from both GFP- and BMP4GFP-transduced MDSCs were first blocked with a MCP-1-neutralizing antibody (BD554440; 50 ng/ml; BD Biosciences) for 2 h at 37°C. MCP-1 protein (BD 554590; BD Biosciences) was used as a positive control, and the chemotaxis assay was performed as described above. All of the chemotaxis assays were repeated 3 times.

Endothelial cell proliferation assay

Mile Sven1 (MS1) mouse endothelial cells (CRL-2279; American Type Culture Collection, Manassas, VA, USA) were cultured at 37°C under 10% CO₂ in medium consisting of high-glucose DMEM (Invitrogen; Life Technologies) supplemented with 5% FBS and 1% penicillin/streptomycin and passaged every 2 d. The cells were trypsinized when they reached 70–80% confluency. For the proliferation experiments, cells were seeded in a flat-bottom 96-well plate (Corning, Corning, NY, USA) at a density of 4 × 10³ cells/well in 0.2 ml medium. Four replicates were performed for each treatment group. To test whether the MDSCs had a paracrine effect on the endothelial cells in vitro, we tested MS1 cell growth in conditioned medium from both GFP- and BMP4GFP-transduced MDSCs, nonconditioned MDSC PM, and unconditioned MS1 culture medium at 37°C and 5% CO₂ for 48 and 72 h. At 2 h before finishing the experiment at each time point, 100 μl of medium was removed from each well, and 20 μl of Cell Titer 96 AQueous One solution (TB 245; Promega) was added. The plates were incubated for an additional 2 h, and then the absorbance was measured using a Tecan Infinite 200 plate reader (Tecan Group Ltd., Männedorf, Switzerland) at 490 nm (A₄₉₀). The proliferation assay was repeated 4 times, and the results were reported from one experiment (mean of 4 replicates). Cell growth under each condition was compared to PM alone at each of the respective time points.

Statistical analysis

All values are expressed as means ± sd. Student's t test was used for comparison of 2 groups. One-way ANOVA followed by Tukey's post hoc t test was used for multiple group analyses. For those data that did not have a gaussian distribution and unequal variances (CD31 at the 3 d time point in vivo), the Wilcoxon rank sum test was utilized. Values of P < 0.05 were considered to be statistically significant.

RESULTS

MDSCs expressed BMP receptors (BMPRs) and paracrine factors

The transduction efficiency of MDSCs with retro-GFP and retro-BMP4GFP was ∼85% (data not shown). To track the MDSCs in vivo, we utilized FACS to purify the GFP- and BMP4GFP-transduced MDSCs based on GFP expression, allowing us to obtain 100% GFP-positive cell populations (Fig. 1A). The secretion of BMP4 by the BMP4GFP-transduced cells was 19.5 ± 4.1 ng/10⁶ cells/24 h at the time of in vivo implantation. No BMP4 secretion was detected by the GFP-transduced MDSCs. ALP staining was negative for both populations of transduced cells when the cells were cultured in nonosteoinductive PM, indicating that no osteogenic differentiation occurred 8, 12, or 16 passages post-transduction (Fig. 1B). Western blot analysis showed an increase in pSMAD1/5/8 expression by the BMP4/GFP-transduced cells (Fig. 1C, D); however, RUNX2 could not be detected in either the GFP- or BMP4/GFP-transduced cells, which implies that neither of the transduced MDSC populations differentiated toward an osteogenic lineage after transduction when cultured in nonosteogenic PM (Supplemental Fig. S1A). We also found that COX-2 protein level was higher in BMP4/GFP-transduced MDSCs than GFP-transduced MDSCs, albeit with no statistical difference (Supplemental Fig. S1A, B). Semiquantitative RT-PCR demonstrated that both of the transduced populations slightly expressed Bmpr1b and inducible nitric oxide synthase (Inos); however, both populations highly expressed Bmpr2, the transcription factor Sox-9, Cox-2, Mcp1, and Vegfa (Fig. 1E). Quantification of gene expression at different passages of culturing revealed no statistically significant changes between the BMP4GFP-transduced (passages 35, 38, and 40), and GFP-transduced MDSCs (passages 27, 30, and 32) (Fig. 1F).

Figure 1.

Transduction efficiency and gene expression. A) GFP- and BMP4GFP-transduced MDSCs are 100% GFP⁺ after sorting. B) ALP staining indicated no osteogenic differentiation after 8, 12, and 16 passages post-BMP4GFP transduction and expansion in nonosteoinductive PM. C) Western blot of pSMAD1/5/8. D) Quantification of pSMAD1/5/8 indicates slightly activated BMP4-pSMAD1/5/8 signaling. *P < 0.05. E) RT-PCR demonstrated that MDSCs expressed transcripts for Sox-9 and BMP receptors, as well as Cox-2, Mcp1, Vegfa, and Inos. F) Quantification of gene expression indicated no change in BMP4GFP-transduced cells compared to GFP-transduced cells. Scale bars = 200 μm (A); 100 μm (B).

BMP4GFP-transduced MDSCs were capable of repairing a calvarial bone defect through the process of endochondral bone formation

Following cell transplantation, we monitored bone formation via microCT analysis. We observed bone formation beginning at 14 d, with complete bone closure of the defect by 28 d in the MDSC/BMP4/GFP group. There was no bone regeneration detected in the MDSC/GFP and scaffold control groups, even after 28 d (Fig. 2A). Quantification of calvarial defect coverage revealed significant differences between the MDSC/BMP4/GFP group and the other two groups (Fig. 2B). Quantification of the new bone volume in the defect area indicated significantly more bone formation in the MDSC/BMP4/GFP group than the other two groups at 14, 21, and 28 d. The average new bone volume of the MDSC/BMP4GFP group was 313-, 2986-, and 7447-fold greater than the scaffold group and 1369-, 4309-, and 7613-fold greater than the MDSC/GFP group at 14, 21, and 28 d, respectively (Fig. 2C). Biomechanical testing results indicated no statistical differences in either IDI or TID between the newly regenerated bone and the normal (uninjured) bone in the MDSC/BMP4/GFP group (Fig. 2D, E). Herovici's staining revealed red collagen type I bone matrix formation, which was similar to the contralateral normal bone (×20, ×200) in the MDSC/BMP4/GFP group and blue collagen type 3 in the scaffold and MDSC/GFP groups (Fig. 2F). H&E staining revealed typical trabecular bone formation that contained bone matrix, bone marrow, and blood vessels in the defect area of the MDSC/BMP4/GFP group, but only a thin layer of fibrotic tissue in the scaffold and MDSC/GFP groups (Fig. 2G). To determine what kind of bone was being formed, we performed alcian blue and von Kossa staining and observed alcian blue-positive chondrocyte nodules in the defect area at 7 d by the MDSC/BMP4/GFP group. By 2 wk, we observed alcian blue-positive hypertrophic chondrocytes, as well as bone matrix formation; however, we did not find any alcian blue-positive cells in the MDSC/GFP and scaffold control groups at either the 7 or 14 d time points (Fig. 3A). Von Kossa staining revealed mineralization at both 14 and 21 dpi in the MDSC/BMP4/GFP group, but not in the other two groups (Fig. 3B).

Figure 2.

Characterization of the de novo bone histomorphometry. A) MicroCT reconstruction of the calvarial defect revealed bone regeneration at 14 dpi and complete healing of the defect by 28 dpi in the transduced MDSC/BMP4GFP group, but not in the MDSC/GFP and scaffold control groups. B) Quantification of bone defect coverage in the three groups. C) Quantification of the de novo bone volume in the defect area. D, E) Biomechanical testing indicated newly formed bone in the MDSC/BMP4/GFP group (callus) at 4 wk has similar IDI (D) and TID (E) parameters to the normal contralateral calvarial bone. F) Herovici's staining to differentiate major bone matrix collagen type 1 (red) and collagen type 3. Region between black arrows indicates the defect region in each group. In the MDSC/BMP4/GFP groups, similar bone matrix formed in the defect area as the normal contralateral side (×20). High magnification (boxed area in ×20 view) demonstrated the formation of mainly collagen type 1⁺ (red) trabecular bone in the MDSC/BMP4/GFP group. Mainly collagen 3 (blue) was found in the scaffold and MDSC/GFP groups. G) H&E staining indicated the formation of trabecular bone in the MDSC/BMP4/GFP group with functional bone structure, as indicated by the appearance of bone matrix, vascularity, and bone marrow. Only fibrotic tissue was observed in the scaffold and MDSC/GFP groups. Col1, collagen 1; Col3, collagen 3; BM, bone marrow; BV, blood vessel; Br, brain; TB, trabecular bone; FT, fibrotic tissue. *P < 0.05, **P < 0.01, ***P < 0.001 vs. scaffold group; ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 vs. MDSC/GFP group.

Figure 3.

Endochondral bone formation and donor and host cell contribution during bone regeneration. A) Alcian blue staining at 7 and 14 d indicated chondrocytes and cartilage formation (endochondral bone formation) in the MDSC/BMP4/GFP group, but not in the other two groups. B) Von Kossa staining of the tissues at 14 and 21 d indicated the presence of mineralized tissue (brown/black) in the defect area in the MDSC/BMP4/GFP group but not in the other groups. Scale bars = 100 μm. C) Dual immunofluorescent staining using COL2A1 and OC and GFP at different stages of bone formation (7, 14, 21, 28 dpi) demonstrated that COL2A1 and OC colocalized with GFP at different time points, as shown in the bottom panel (boxed area) in the MDSC/BMP4/GFP group, but no COL2A1 or OC expression was observed in the other two groups at any time point. Immunohistochemistry at 28 d indicated donor GFP⁺ cells differentiated into trabecular bone osteocytes and bone surface osteoblasts (red arrows indicate osteocytes; green arrows indicate osteoblasts). Host cells mainly constitute the bone marrow between trabecular bone (shown as blue nuclei cells). Scale bar = 100 μm. De, defect; Br, brain; Chon, chondrocytes; BM, bone marrow; TB, trabecular bone.

BMP4GFP-transduced MDSCs differentiated into chondrocytes, osteoblasts, and osteocytes during the bone repair process

The participation of donor BMP4GFP-transduced MDSCs in the endochondral bone formation process was determined by colocalizing chondrogenic and osteogenic cell markers with GFP (donor cells) at different time points postimplantation. At 7 dpi, the majority of cells in the MDSC/BMP4/GFP group expressed COL2A1 (chondrocytes) and were also GFP⁺, identifying them as donor cells. At 14 dpi, in addition to COL2A1/GFP-double-positive donor cells, we also observed the appearance of OC/GFP-double-positive cells, which identified them as osteoblasts. At 21 dpi, we still observed COL2A1/GFP-double-positive cells near the edge of the regenerated bone or in the cartilage area. At this time point, we observed OC⁺ cells throughout the regenerated bone that colocalized with GFP⁺ cells, indicating that they were donor derived. At 28 d, immunohistochemical staining revealed that the majority of osteoblasts (green arrows) lining the trabecular bone matrix as well as the osteocytes (red arrows) embedded inside the trabecular bone matrix, were GFP⁺ (brown color). The host cells constituted the newly formed supportive stroma (nuclei shown in blue only at 28 d). As expected, we did not detect any COL2A1/GFP- or OC/GFP-double-positive cells at any of the time points in the MDSC/GFP or scaffold groups (Fig. 3C).

BMP4GFP-transduced MDSCs regenerated bone in the defect area through the activation of the COX-2–prostaglandin E₂ (PGE2) and BMP4-pSMAD5 signaling pathways

Dual color immunofluorescent staining of COX-2 and GFP indicated that COX-2 was expressed in both the donor and host cells in the MDSC/GFP and MDSC/BMP4/GFP groups at d 3. At 7 d, there was no COX-2 expression in the scaffold control group, while some cells in the MDSC/GFP group still expressed COX-2, which colocalized with GFP. Strikingly, we found COX-2 to be highly expressed in the MDSC/BMP4/GFP group, where it colocalized with GFP. At 14 d, COX-2 was expressed in the MDSC/GFP group, which colocalized with GFP. At this time point, COX-2 was still highly expressed by cells in the chondrocyte stage in the MDSC/BMP4/GFP group. At 21 d, few cells still expressed COX-2 in the MDSC/GFP and MDSC/BMP4/GFP groups (Fig. 4A, enlarged views in bottom panels). Dual-color immunofluorescence of pSMAD5 and GFP indicated pSMAD5 was expressed by some donor and host cells at 3 d in the defect area of both the MDSC/GFP and MDSC/BMP4/GFP groups. At 7 d, pSMAD5 was expressed in some of the chondrocytes and undifferentiated donor MDSCs in both the MDSC/GFP and MDSC/BMP4/GFP groups. At 14 d, at the initiation of bone formation, pSMAD5 was still expressed in the early undifferentiated stage of the cells in both of these groups. At 21 d, when the bone tissue became more mature, pSMAD5 was highly expressed in the periphery of the regenerated bone and was also expressed in the osteoblast cells lining the bone matrix, but not by the osteocytes (Fig. 4B).

Figure 4.

Activation of COX-2 and BMP4-pSMAD5 signaling during MDSC-mediated bone regeneration. A) Dual-color immunofluorescence of COX-2 and GFP indicated that the COX-2 signaling pathway is activated during MDSC-mediated bone regeneration primarily in the donor cells. Bottom panels are high-magnification views of boxed areas and colocalization of COX-2 with GFP in the MDSC/BMP4/GFP group images. B) Dual-color pSMAD and GFP immunofluorescence indicated SMAD5 is activated in the donor cells at different time points. Bottom panels are high-magnification views of boxed areas in the MDSC/BMP4/GFP group images. Scale bar = 100 μm. De, defect, Br, brain; TB, trabecular bone; Chon, chondrocytes.

BMP4GFP-transduced MDSCs triggered rapid inflammatory cell infiltration and earlier inflammatory resolution during the bone repair process

Under typical injury conditions, the infiltration of inflammatory cells into the injury site occurs very rapidly; therefore, we evaluated the inflammatory reaction in the defect area of the mice by characterizing macrophage and neutrophil infiltration at multiple time points postimplantation. At 3 dpi, we found significantly more cells that expressed the macrophage marker CD68 in the MDSC/BMP4/GFP and MDSC/GFP groups compared with the scaffold only group. We also observed macrophage migration into the defect site from the surrounding tissues. At 7 dpi, a large number of macrophages were found in all groups. Notably, at this time point, the number of macrophages was seen to decrease in both the MDSC/BMP4/GFP and MDSC/GFP groups compared with the scaffold group. The MDSC/GFP group was significantly different from the scaffold control group; however, no difference was noted between the MDSC/BMP4/GFP and MDSC/GFP groups. Typically, acute inflammation subsides 14 d after injury, and indeed, we found a reduction in the number of macrophages in all groups; however, the number of macrophages was significantly lower in both the MDSC/BMP4/GFP and MDSC/GFP groups compared to the scaffold group. Moreover, the MDSCBMP4/GFP group possessed significantly fewer macrophages than the MDSC/GFP group at this time point (Fig. 5A, B). The scaffold group is representative of what we would expect to occur in normal tissue healing because fibrin is a normal component of mammalian wound healing and is typically absorbed by 7 dpi. The number of macrophages in the MDSC/BMP4/GFP group was also significantly lower than the MDSC/GFP group, indicating an even earlier resolution of inflammation in this group. At 21 dpi, we observed significantly fewer macrophages in the MDSC/BMP4/GFP and MDSC/GFP groups compared to the scaffold group (Fig. 5A, B). Similarly, we observed significantly more cells expressing the neutrophil marker Gr-1 in the MDSC/BMP4/GFP and MDSC/GFP groups compared to the scaffold group at 3 dpi, with the MDSC/BMP4/GFP group being significantly higher than the MDSC/GFP group (Fig. 5C, D). There were significantly fewer neutrophils in the MDSC/BMP4/GFP and MDSC/GFP groups compared with the scaffold control group at 7 dpi. At 14 and 21 dpi, there were fewer neutrophils present in all of the groups, with no differences observed between the 3 groups (Fig. 5C, D). We were also unable to colocalize CD68 or Gr-1 with GFP at any of the time points, which indicates that these inflammatory cells were all host derived.

Figure 5.

BMP4GFP-transduced MDSCs affect host inflammation in vivo. A) Infiltration of macrophages at 4 different time points postimplantation in the 3 groups. B) Quantification of macrophage number at different time points indicated an early enhanced onset and accelerated resolution of inflammation in the MDSC/GFP and MDSC/BMP4/GFP groups. C) Infiltration of neutrophils in the defect area at different time points in the three groups. D) Quantification of infiltration of neutrophils at different time points indicated an early onset at 3 d and resolution of acute inflammation at 7 d in MDSC/GFP and MDSC/BMP4/GFP groups. Scale bar = 100 μm. De, defect; Br, brain; Chon, chondrocytes; TB, trabecular bone. *P < 0.05, **P < 0.01 vs. scaffold control group; ^##P < 0.01 vs. MDSC/GFP group.

BMP4GF-transduced MDSCs attracted host vascular endothelial cells, promoting angiogenesis during bone regeneration

Osteogenesis and angiogenesis are tightly coordinated during the formation of bone (16). To characterize the role of vascular endothelial cells in BMP4GFP-transduced MDSC-mediated bone repair, we performed immunofluorescent staining for the endothelial cell marker CD31. At 3 dpi, we found CD31⁺ vascular endothelial cells at the skull defect border with the brain (dura mater) and within the defect in the MDSC/BMP4/GFP and MDSC/GFP groups (Fig. 6A). No CD31⁺ cells were found in the defect area of the scaffold group (Fig. 6A). This finding suggests that the vascular endothelial cells that migrated into the defect area were recruited by the donor cells irrespective of BMP4 expression. At 7 dpi, we found CD31⁺ vascular structures in the defect area in all groups. At 14 dpi, when endochondral bone formation occurred, we observed vascular structures and cells between the trabecular bony structures forming what appeared to be a vascular network in the MDSC/BMP4/GFP group, but not in the MDSC/GFP and scaffold groups. At 21 dpi, we observed a CD31⁺ vascular network begin to form between the bony structures in the MDSC/BMP4/GFP group only (Fig. 6A). Quantification of the CD31⁺ cells revealed significantly more vascular endothelial cells in the defect area of the MDSC/BMP4/GFP group than the scaffold group at 3, 7, 14, and 21 d. We also found significantly more CD31⁺ cells in the defect area of the MDSC/GFP group than the scaffold group at 3 d, indicating this effect was likely mediated by the MDSCs, irrespective of BMP4 transgene expression (Fig. 6A, B). At later time points, the MDSC/GFP group formed soft tissues similar to what was observed in the scaffold group and exhibited no enhanced angiogenesis compared to the scaffold control. The lack of colocalization between the CD31⁺ cells and GFP (green) at all the time points indicated that these cells were host derived.

Figure 6.

BMP4GFP-transduced MDSCs enhanced the participation of vascular endothelial cells to promote the healing of the bone defect. A) Immunofluorescent labeling of vascular endothelial cells with CD31 and GFP at different time points indicated more vascular endothelial cells in the MDSC/GFP and MDSC/BMP4GFP groups. B) Quantification of CD31⁺ cells at different time points indicated that MDSCs enhanced host angiogenesis. Scale bar = 100 μm. De, defect; DM, dura mater; Br, brain; Chon: chondrocyte; TB, trabecular bone. *P < 0.05, **P < 0.01 vs. scaffold group; ^#P < 0.05, ^##P < 0.01 vs. MDSC/GFP group.

MDSCs influenced the T-cell response during bone regeneration

At 3 d after surgery, we found very few CD4⁺ T lymphocytes in the defect area of all groups. At 7 dpi, an increased number of CD4⁺ T lymphocytes were found in the defect area of all groups; however, there were significantly fewer CD4⁺ cells present in the MDSC/BMP4/GFP and MDSC/GFP groups compared to the scaffold group. We found it interesting that at 14 dpi, the number of CD4⁺ lymphocytes in the scaffold group decreased, while there was a significant increase in the number of CD4⁺ cells in the MDSC/BMP4/GFP and MDSC/GFP groups, which we inferred was caused by the transplantation of allogenic cells. At 21 dpi, the number of CD4⁺ T lymphocytes further increased in the MDSC/GFP group and was significantly higher than the other groups (Fig. 7A, B). We observed a similar change in the number of CD8⁺ T lymphocytes at all time points, with significantly fewer CD8⁺ cells present in the MDSC/BMP4/GFP and MDSC/GFP groups in comparison to the control group at 7 d, and a significantly higher number at d 14 and 21 when compared to the scaffold group. In addition, at 21 dpi, the CD8⁺ lymphocytes in the MDSC/BMP4/GFP were significantly lower than that of the MDSC/GFP group (Fig. 7C, D). In fact, at 3 wk, the CD4⁺ and CD8⁺ T lymphocytes were found to reside between the newly formed trabecular bone structures in the MDSC/BMP4/GFP group, which we posit was the cell component of bone marrow. CD4⁺ and CD8⁺ cells did not colocalize with GFP, indicating they were host derived.

Figure 7.

BMP4GFP-transduced MDSCs inhibited the initial host T-cell response during the process of bone regeneration. A) CD4 and GFP immunofluorescence to identify T-helper lymphocytes at different time points. B) Quantification of CD4⁺ lymphocytes in the defect area at 4 time points postimplantation. MDSC/GFP and MDSC/BMP4GFP inhibited the CD4⁺ cell response at 7 d but elicited strong immune responses after 14 and 21 d. C) CD8 and GFP immunofluorescent staining to identify T-suppressor (cytotoxic) lymphocytes. D) Quantification of CD8⁺ T lymphocytes at multiple time points indicated similar change as the CD4⁺ T lymphocytes. Scale bar = 100 μm. De, defect; Br, brain; Chon, chondrocytes; TB, trabecular bone. *P < 0.05, **P < 0.01 vs. scaffold group; ^#P < 0.05, ^##P < 0.01 vs. MDSC/GFP group.

MCP1 secreted by the MDSCs appeared to play an important role in MDSC-mediated macrophage cell migration

To further examine the mechanism by which the transplantation of MDSC/GFP and MDSC/BMP4/GFP enhanced host macrophage migration into the site of injury, we performed chemotaxis experiments in vitro. The results indicated that conditioned medium derived from both the GFP- and BMP4GFP-transduced MDSCs could significantly increase the migratory capacity of RAW264.7 macrophage cells compared to fresh unconditioned PM (Fig. 8A). No differences were observed between the MDSC/BMP4/GFP and MDSC/GFP conditioned medium groups. MCP1 expression was detected by immunofluorescent staining (Fig. 8B). The secretion of MCP1 was detected by ELISA, and no statistical difference was observed between the GFP- and BMP4GFP-transduced MDSC groups (Fig. 8C). MCP1 expression was also confirmed by Western blot analysis (Fig. 8D). Furthermore, in order to determine whether the enhanced migration of the macrophages was mediated by MCP1, we blocked MCP1 activity in the conditioned media isolated from both the GFP- and BMP4GFP-transduced MDSC cultures using a neutralizing antibody. We found that blocking MCP1 significantly reduced the migratory effects of the conditioned media (Fig. 8E); however, blocking MCP1 did not entirely inhibit the macrophage migratory effect of the conditioned media, which suggests that there may be other factors being secreted by the MDSCs that influence macrophage migratory behavior and/or it could be due to the fact that the MCP1 protein was not completely blocked by the neutralizing antibody.

Figure 8.

MDSCs promoted macrophage migration and endothelial cell proliferation in vitro via paracrine effects. A) Conditioned media from GFP- and BMP4GFP-transduced MDSCs enhanced RAW264.7 macrophage migration compared with nonconditioned PM, both at 18 and 24 h. *P < 0.05, **P < 0.01 vs. PM. B) MCP1 was expressed by GFP- and BMP4GFP-transduced MDSCs. C) MCP1 was shown to be secreted into the cell culture supernatant by both GFP- and BMP4GFP-transduced MDSCs, as measured by ELISA. D) MCP1 was also found to be expressed in the cell lysate, as indicated by Western blot analysis. E) Blocking MCP1 with a neutralizing antibody significantly reduced the migration of macrophages by the GFP- and BMP4GFP-transduced MDSC-conditioned media. *P < 0.05, **P < 0.01 vs. PM alone; ^#P < 0.05, ^##P < 0.01 vs. conditioned medium without neutralizing MCP-1 antibody. F) MDSC-conditioned medium promoted murine MS1 endothelial cell proliferation at both 48 and 72 h. *P < 0.05, **P < 0.01 vs. PM. G) MDSCs expressed numerous angiogenic genes; quantification indicated BMP4 transduction significantly decreased Igf2 and Hgf expression but did not alter the expression of the other genes tested. H) VEGFa and PDGFβ expression was further shown by Western blot analysis. I) VEGFa immunofluorescent staining of VEGFa indicated its expression in both GFP and BMP4GFP-transduced MDSCs. J) PDGFβ expression was also validated by immunofluorescence.

Conditioned medium isolated from both the GFP- and BMP4GFP-transduced MDSC cultures enhanced endothelial cell proliferation

Our in vitro proliferation assay demonstrated that conditioned medium from both the GFP- and BMP4GFP-transduced MDSC cultures significantly increased murine MS1 endothelial cell proliferation compared to nonconditioned fresh PM at 48 and 72 h (At 48 h, P<0.05 and P<0.01, respectively; at 72 h, P<0.01 and P<0.01, respectively; Fig. 8F). Semiquantitative RT-PCR results indicated that the MDSCs expressed numerous growth factors that could promote angiogenesis in addition to Vegfa, as shown in Fig. 1B, although the BMP4GFP-transduced MDSCs were observed to decrease Igf2 and Hgf expression compared with the GFP-transduced MDSCs (Fig. 8G). The expression of VEGFa and PDGFβ was also confirmed by Western blot and immunofluorescent staining (Fig. 8H–J).

DISCUSSION

In this study, we found that BMP4GFP-transduced MDSCs could heal a critical-size calvarial defect within 28 d through the formation of endochondral bone, which displayed similar biomechanical properties as the native calvarial bone. More important, we found that the BMP4GFP-transduced MDSCs differentiated into the majority of the chondrocytes, osteoblasts, and osteocytes that were present in the regenerated endochondral bone area, which was likely mediated through the activation of the BMP4-pSMAD5 and COX-2-PGE2 pathways. MDSC implantation also promoted bone healing by influencing the host cell response, which included the enhanced early onset of inflammation that resolved more quickly, an enhancement in angiogenesis, and the inhibition of an initial T-cell-mediated immune rejection response. These effects were at least partly mediated by paracrine factors secreted by the donor MDSCs, including MCP1 and the angiogenic factors VEGF, fibroblast growth factor 2 (FGF2), hypoxia-induced factor 1α (HIF1α), and COX-2.

Stem cell-mediated osteogenesis and bone regeneration require the cells to express osteogenic related receptors, including BMPR1b and BMPR2 so that the cells can respond to BMP osteogenic signaling. Indeed, we found that MDSCs transduced with either GFP or BMP4GFP expressed Bmpr1b and Bmpr2. MDSCs were also found to highly express Sox-9 and Cox-2. It is known that the Sox-9 gene plays an important role in skeletal development, and its expression precedes the expression of other osteogenic factors, such as Runx2 and osterix (17, 18). We demonstrated that when BMP4GFP-transduced MDSCs were implanted in a critical size calvarial defect, they regenerated the bone via the formation of endochondral bone. The BMP4GFP-transduced MDSCs differentiated into chondrocytes during the cartilage formation stage and into osteoblasts and osteocytes during the bone formation stage, which was mediated through the BMP4-pSMAD5 signaling pathway. Even though the GFP-transduced MDSCs were capable of responding to osteogenic signaling because of their expression of BMPRs, they did not form bone, which indicates that the MDSCs require strong osteogenic stimulation that we achieved through the transduction of the cells with BMP4.

Significant contributions by the transplanted donor stem cells in the bone formation process has been reported by other investigators using MSCs, in an endochondral bone defect model (19). The transplantation of MSCs subcutaneously into syngenic or immune-compromised recipient mice generated new bone mainly by recruiting host osteoprogenitors; however, no bone regeneration was observed when the cells were implanted allogeneically (20). Moreover, the transplantation of MSCs subcutaneously into immune-compromised mice was shown to mediate endochondral bone formation that consisted mainly of host-derived cells. These results imply that MSCs probably require a strong osteogenic signal from BMP, as do our MDSCs, to differentiate toward an osteogenic lineage in order to directly contribute to the formation of bone (21). The advantage of host cell-derived bone formation is that it eliminates the safety concerns about virally transducing the donor cells with BMPs. The main advantage of donor cell-derived bone regeneration, which in the current study was performed in allogenic recipients, is that there is a large contribution of donor cells that directly participate in forming the newly regenerated bone, which significantly accelerates the healing process of the defect (complete healing was observed by 4 wk). This is especially important in situations where host cells are unable to repair a critical size bone defect or a nonunion fracture, which is common among the aged and in certain disease states.

The effect that the transplanted MDSCs had on the host inflammatory response was most likely induced by the donor MDSCs' secretion of a variety of paracrine factors, other than BMP4, since similar effects were observed in both the MDSC/GFP and MDSC/BMP4GFP groups. We demonstrated that MCP1 was one of the primary paracrine factors secreted by the MDSCs, which was, at least in part, responsible for chemoattracting macrophages into the defect area. Stem cell-secreted MCPs and their paracrine modulation have recently garnered much attention. It has been reported that macrophages and monocytes play a key role in the inflammatory process by initiating tissue repair following injury, and it has been suggested that they could be utilized as a therapeutic cell source to promote angiogenesis (22). MCP2 has been shown to be expressed by murine MSCs and up-regulated when the cells are cultured with FGF, which drives the MSCs to induce the formation of endochondral bone (23). MCP1 has been shown to be expressed by human adipose-derived stem cells (ASCs), bone marrow MSCs, dermal sheath cells (DSCs), and dermal papilla cells (DPCs), and MCP2 and MCP3 have been shown to be expressed in DSCs and DPCs (24). The secretion of MCP1 and/or MCP3 by MSCs has also been shown to be involved in bone marrow stem cell-mediated distraction osteogenesis. Moreover, human MSCs or MSC-conditioned medium that contains MCP1 or MCP3 enhanced distraction osteogenesis, whereas fibroblasts and fibroblast-conditioned medium did not. When MCP1 and MCP3 are neutralized in the MSC-conditioned medium, it decreases the migration of bone marrow cells in vitro and negates the beneficial effects of the MSC-conditioned medium, when used in vivo for distraction osteogenesis (25). Furthermore, MCP1 has been shown to induce human periosteum-derived progenitor cell chemotactic migration (26) and can attract monocytes to the site of osseous injury, increasing the number of osteoblasts present at the injury site, which can improve healing (27). A deficiency in monocyte recruitment in response to inflammatory conditions has been observed in two mouse models of MCP1 deficiency, where MCP1 was directly knocked out and another model in which MCP1's receptor, CCR2, was knocked out (28, 29). It has also been recently shown that MCP1 expressed by osteoblasts and osteocytes is a downstream factor of parathyroid hormone-mediated bone anabolism (30). These findings demonstrated that the expression of MCPs plays a role in the bone repair process and that the expression of MCPs is not exclusive to MDSCs.

On the other hand, prior work has established that MSCs can stimulate tissue repair by inhibiting inflammation. The secretion of proinflammatory mediators caused by injury stimulates a negative feedback loop by the MSCs, which secrete PGE2. Subsequently, PGE2 converts proinflammatory M₁ macrophages into M₂ macrophages that secrete the anti-inflammatory mediators IL-10 and IL-1ra, which inhibit inflammation. Proinflammatory mediators also stimulate MSCs to secrete TNF-α-stimulated gene/protein 6 (TSG-6), which down-regulates NF-κB signaling and decreases the secretion of TNF-α, thereby further inhibiting inflammation (31–34). It has also been shown that the injection MSCs into the infarcted hearts of mice increases the percentage of M₂ macrophages in the hearts of these mice 3 to 4 d after injection. The transient depletion of M₁ and M₂ macrophages increases the mortality of MI animals, with early depletion having the greatest negative effects (35). These results demonstrated that macrophages recruited to the site of injury by the implanted MSCs play an important role in tissue repair. These studies are supportive of our findings that the dynamic inflammatory modulation observed during MDSC-mediated bone repair is important to properly regenerate bone tissue. The implantation of MDSCs recruited more inflammatory cells due to paracrine effects and incurred enhanced inflammation at the early time point of 3 d, initiated a tissue repair cascade and then expeditiously resolved the inflammatory process by d 7 to further promote bone tissue repair.

The coordination of osteogenesis and angiogenesis is necessary for both intramembranous and endochondral ossification and has been extensively investigated (16). In the current study, we found that the implantation of MDSCs increased the recruitment of host angiogenic cells to the defect area. This finding was supported by an in vitro proliferation assay, which indicated that the conditioned media collected from both the GFP- and BMP4GFP-transduced MDSC cultures could enhance in vitro murine vascular endothelial cell proliferation. We were surprised to observe that at d 14 and 21, only the MDSC/BMP4/GFP group demonstrated a further increase in the amount of vascularization present at the site of regeneration. This may suggest that the MDSCs in the MDSC/GFP group did not receive sufficient osteogenic signaling from the host to stimulate their differentiation into an osteogenic lineage and eventually formed similar granulation tissue, as was seen in the control scaffold group. The paracrine secretion of angiogenic factors by other stem cells has also been reported to be important for enhancing tissue repair. For example, it has been shown that the injection of serine/threonine-specific protein kinase (Akt; protein kinase B)-modified MSCs (Akt-MSCs) into the injured hearts of mice could improve cardiac function in less than 72 h, which was mediated, in part, by the paracrine secretion of VEGF, FGF2, hepatocytic growth factor (HGF), insulin growth factor 1 (IGF1), and TB4 (36). Moreover, Hsiao et al. (24) analyzed several different stem cell populations, including ASCs, BMSCs, DSCs, and DPCs, and demonstrated that all these cell populations secreted VEGF, angiogenin, and stem cell-derived factor-1 (SDF1, also known as CXCL12) but did not express IGF1 or PDGFβ. Murine MSCs cultured in medium containing FGF2 have also been shown to up-regulate IGF1 and other factors that enhance the host cells' contribution to the formation of the new bone (23). It has also been shown that the implantation of MSCs seeded in a scaffold subcutaneously, recruited host-derived CD31⁺ cells 7 d after implantation, which were capable of forming capillary-like structures in vitro. MSCs were also shown to recruit CD146+ cells to the implantation site 11 dpi, which promoted the formation of new bone (9). In another recent study, it was demonstrated that FGF2 and VEGF secreted by cortical bone stem cells were capable of enhancing angiogenesis after their transplantation into an acutely infarcted heart (37). In our current study, we showed that the donor MDSCs secreted VEGFa, FGF2, HGF, PDGFβ1, HIF1α, TGFβ1, IGF1, and IGF2, albeit BMP4 transduction decreased IGF2 and HGF; although the paracrine secretion of angiogenic factors varied slightly between the MDSCs and MSCs, both populations of stem cells shared the expression of VEGFa, FGF2, IGF1, and PDGFβ; hence, the proangiogenic effect imparted by the MDSCs in this study is, at least partially, mediated by the angiogenic factors they secreted, which have been previously shown to enhance angiogenesis and accelerate bone formation in vivo (37–41).

COX-2 is responsible for the catabolization of arachadonic acid into PGE2 and other prostaglandins and plays important roles in many physiological functions. COX-2 function is essential for postnatal bone fracture healing without affecting fetal skeletal development (42). COX-2-deficient mice demonstrated impaired osteoblastogenesis and bone fracture repair (43). It has also been reported that bone repair in a long bone fracture model is delayed in Cox-2-knockout mice due to a significant reduction in vascularization (44). Moreover, the inhibition of the COX-2 pathway using selective COX-2 inhibitors decreases blood flow within the bone fracture gap and negatively impacts fracture repair (45). Most recently, it has been shown that the injection of a lentivirus encoding for COX-2 in a long bone fracture promoted bone bridging in the fracture gap by reducing cartilage callus size and enhancing angiogenesis (46). Furthermore, COX-2 has been found to be a major mediator in MSC-mediated immune-regulatory effects (47). Therefore, we reasoned that the inhibition of the immune response early in the healing process (by d 7) and its increase later on in the healing process, in both the MDSC/GFP and MDSC/BMP4/GFP groups, when compared to the scaffold-only group, could be correlated with the dynamic expression of COX-2 during MDSC-mediated bone regeneration; however, the overall role of the COX-2 signaling pathway in MDSC-mediated bone regeneration requires further analysis.

On the basis of our findings, we propose a schematic model to explain the role that donor and host cells play in BMP4GFP-transduced MDSC-mediated bone healing (Fig. 9). MDSCs that receive BMP4 signaling and express endogenous BMP receptors and SOX-9 are capable of differentiating into chondrocytes. Donor-derived GFP⁺ chondrogenic cells continue to express BMP4, which activates pSMAD5 and COX-2 during the bone-healing process, eventually forming endochondral bone. The MDSC-secreted mediators, including MCP1 and COX-2 (PGE2; Fig. 9, red box), which attract host macrophages and neutrophils to the implantation site and initiate early enhanced inflammation, thus activating the host tissue repair cascade. Through the expression of VEGFa, FGF2, HIF1α, and COX-2 (Fig. 9, purple box), the MDSCs chemoattract more host vascular cells and/or circulating angiogenic cells to form blood vessels, which support the donor cells' survival and proliferation, and, thereby, promotes angiogenesis and osteogenesis. In addition, the MDSCs display immune-regulatory properties during bone regeneration, which appears to promote the initial survival of the donor MDSCs (Fig. 9).

Figure 9.

Schematic illustration depicting a summary of donor BMP4GFP-transduced MDSCs, their contribution to bone regeneration, and their effect on host cells during the bone formation process.

In summary, this study revealed that BMP4GFP-transduced MDSCs could differentiate into chondrocytes in the early stages of endochondral bone formation and osteoblasts and osteocytes during the later stages of bone formation. Bone regeneration was found to be mediated by the activation of the BMP4-pSMAD5 and COX-2-PGE2 signaling pathways. The donor MDSCs also initiated an early host inflammatory cell response, which was resolved expeditiously. Furthermore, the implantation of MDSCs promoted the proliferation of host vascular endothelial cells, which increased angiogenesis at the injury site and was involved in regulating the immune response of the host animals. The effects that the donor MDSCs imparted on the host cells were shown to be mediated, at least in part, by the secretion of MCP-1, PGE2, and a variety of other factors, including VEGF, FGF2, HIF1α, and PDGFβ. This study elucidated the important roles that both donor and host cells play in the process of MDSC-mediated bone repair.