Gene Therapy (2010) 17, 733–744 & 2010 Macmillan Publishers Limited All rights reserved 0969-7128/10 $32.00 www.nature.com/gt ORIGINAL ARTICLE Comparative efficacy of dermal fibroblast-mediated and direct adenoviral bone morphogenetic protein-2 gene therapy for bone regeneration in an equine rib model

A Ishihara1,2, LJ Zekas3, SE Weisbrode2 and AL Bertone1,2,3 1Comparative Orthopedic Research Laboratories, Department of Veterinary Clinical Sciences, The Ohio State University, Columbus, Ohio, USA; 2Department of Veterinary Biosciences, The Ohio State University, Columbus, Ohio, USA and 3Department of Veterinary Clinical Sciences, The Ohio State University, Columbus, Ohio, USA

Cell-mediated and direct adenoviral (Ad) vector gene were evident in the cortical and cancellous bone directly therapies can induce bone regeneration, including dermal adjacent to the healing drill defects treated with either fibroblasts (DFbs). We compared two effective therapies, DFb-BMP2 or Ad-BMP2. Using our cell/vector dosage and DFb-mediated and direct Ad vector delivery of bone model, BMP2, whether delivered by the DFb vector or direct morphogenetic protein-2 (BMP2), for relative efficacy in Ad vector, induced greater and robust bone regeneration. bone regeneration. Equine rib drill defects were treated by DFb-mediated BMP2 therapy promoted greater cortical percutaneous injection of either DFb-BMP2 or an Ad-BMP2 bone regeneration than did direct gene delivery, possibly vector. At week 6, both DFb-BMP2- and Ad-BMP2-treated rib because of an increased cellularity of the bone healing defects had greater bone filling volume and mineral density, site. BMP2 delivery, regardless of gene delivery method, with DFb-BMP2 inducing greater bone volume and maturity increased the mineral density of the neighboring bone, which in the cortical bone aspect of the defect than Ad-BMP2. may be beneficial clinically in repairing or weak bone. The transplantation of DFb alone induced modest bone Gene Therapy (2010) 17, 733–744; doi:10.1038/gt.2010.13; formation. Increased mineral density and bone turnover published online 11 March 2010

Keywords: bone morphogenetic protein; dermal fibroblasts; bone healing; cell-based gene therapy; adenoviral vectors

Introduction relative potency between gene- and cell-mediated BMP2 deliveries is unknown. Regenerative strategies for inferior bone repair are a Although embryonic and adult mesenchymal stem growing need in aging people, in which our immediate cells have been extensively studied for cell-mediated population demands include slow healing fractures gene therapy,8,9 other cell sources, such as dermal (estimated 1 million patients annually), spine fusions fibroblasts (DFbs) may be an alternative candidate (300 000 surgeries and $18 billion annually) and osteo- because of their excellent reprogramming capacity.10 In porotic fractures (704 000 cases annually) (http:// fact, DFbs have been differentiated into bone-forming www.boneandjointdecade.org). At present, pharmaceu- cells by the transduction of osteogenic genes including ticals (such as osteoclast inhibitors) and bioactive BMP2,11–15 and such genetically modified DFbs have proteins (such as bone morphogenetic protein-2 been autologously transplanted to induce ectopic bone (BMP2)) are recognized advancements in bone repair; formation14,15 and promote bone healing in rodent however, the use of living cell vectors as regenerative models.12,13,15 In contrast to the use of stem cells, skin treatment for bony disorders offers the advantages of cell-mediated therapy is attractive for clinical use molecular engineering to release both paracrine and because of the relatively less painful harvest technique, autocrine bioactive factors, direct integration of the cells low risk of donor site infection or morbidity, less into the regenerative process and a broad diversity of fastidious culture procedure and rapid high cell medical applications.1–4 Although direct BMP2 gene yield.12–15 Therefore, the use of DFb may be considered delivery is shown to promote bone regeneration,5–7 the as one choice of cell type for ex vivo gene therapy, although the osteoinductive capacity may not be superior to stem cells. Correspondence: Dr AL Bertone, Department of Veterinary Clinical The authors’ laboratory has demonstrated in separate Sciences, The Ohio State University, 1900 Coffey Road, Columbus, studies that DFb-mediated16 and direct adenoviral (Ad) OH 43210, USA. 7 E-mail: [email protected] BMP2 gene therapy significantly accelerated cortical Received 20 July 2009; revised 13 September 2009; accepted 2 long-bone healing in an equine metacarpal/tarsal model, November 2009; published online 11 March 2010 with DFb functioning through endochondral ossification Cell- and gene-based therapy for bone healing A Ishihara et al 734 and direct Ad-BMP2 delivery functioning through direct Ad-GFP, Ad-BMP2, DFb alone, DFb-GFP and DFb-BMP2 bone formation. Equine bone healing models allowed (Figure 2a). All injections were administered at 2 weeks the comparison of the efficacy of therapeutic methods after rib drilling, and all rib specimens were harvested at at multiple sites within the same animal to reduce 6 weeks after the injections (Figure 2b). During this individual variation, and mimic human osseous injuries period, all six horses showed good incisional and and demonstrate the practicality in cost and time surgical site healing, and there were no differences in efficiency for cell transplantation using large numbers healing of skin, edema and inflammation in subcuta- of genetically engineered cells. The study reported herein neous or muscular layers between the treatment groups. was designed to use drill hole defects within equine ribs to directly compare the efficacies of these two strategies Computed tomography that were successful in the long bone. Unlike other Quantitative computed tomography (qCT) at 6 weeks animal models such as slit osteotomy and critical-size after injection showed an induction of bone regeneration gap ostectomy,17–19 the rib drill defect model provides a in equine rib drill defects by direct gene delivery and well-contained area for percutaneous injection, a well- DFb-mediated BMP2 therapy (Figure 3a). In the cortical confined filling defect to quantify bone formation filling zone, bone volume was significantly greater in the with high resolution, both cortical and cancellous Ad-BMP2, DFb-GFP and DFb-BMP2 groups (Figure 3c), bone healing environments, and permits multiple and cortical mineral density was significantly greater in bone defects per animal for appropriate control groups the Ad-BMP2 group (Figure 3d). In the medullary filling within the same animal. In addition, the rib drill defect zone, bone volume was significantly greater in the model with a rich cancellous bone environment may be a DFb-BMP2 groups (Figure 3e). Bone density was not more receptive environment to exhibit differences in these changed in any group (Figure 3f). In the adjacent cortex two methods of BMP2 gene delivery mechanisms, zone, mineral density was significantly greater in the because of the potentially greater numbers of osteopro- Ad-BMP2 and DFb-BMP2 groups (Figure 3g). In the genitor cells as receptive targets for paracrine molecular adjacent medullary zone (Figure 3b), mineral density was signals of BMP2. significantly greater in the DFb-BMP2 group (Figure 3h). The goal of our study was to use an equine rib Of the 12 ribs used in 6 horses (3 holes per rib, 36 total drill defect model to evaluate the relative efficacy of holes), 5 ribs were fractured at 1 of the 3 drill holes. The DFb-mediated BMP2 and Ad-BMP2 gene delivery to treatment assignments of these five fractured drill holes promote bone regeneration. We compared the effects were GBSS (2), Ad-GFP (2) and Ad-BMP2 (1), and none of cell-mediated and direct BMP2 gene delivery using of the drill holes with DFb injections were fractured a percutaneous injection of DFbs or viral vectors regardless of the GFP/BMP2 gene transduction. Thus, the with and without BMP2 gene transduction. In this risk of fractures in DFb-injected defects (0 out of 18 model, we demonstrated successfully that an autologous defects) was significantly lower than in non-DFb-injected transplantation of BMP2-expressing DFbs and direct defects (5 out of 18 defects) (P ¼ 0.045). Owing to the Ad-BMP2 induced bone regeneration, and DFb-BMP2 fracture confounding our bone regeneration outcomes, vectors were more potent than a direct vector injection these five fractured drill defects were not included in for cortical bone regeneration. Cell- and gene-mediated qCT and histological evaluation and were treated as BMP2 delivery showed equivalent increased bone missing data points in data analysis. Despite this data regeneration in the cancellous bone region and increased exclusion, the computed powers were 40.8 in all bone density in adjacent osseous tissues. outcomes of qCT and histology.

Histological evaluation Results Histomorphometry at 6 weeks after treatments showed denser and more mature bone formation and greater In vitro gene transduction and osteogenic differentiation mineralizing activity in equine rib drill defects by direct Equine DFbs were shown to be permissive to Ad gene gene delivery and DFb-mediated BMP2 therapy (Figures transduction, efficiently induced BMP2 protein and 4 and 5). In the cortical filling zone (Figure 4a), the successfully differentiated into bone-forming cells in vitro. %porosity and %mineralizing area was significantly In six horses, DFbs isolated from skin punch biopsy greater in the DFb-BMP2 group (Figures 4b and c), and were successfully transfected by Ad-encoding green the %lamellar bone composition was significantly greater fluorescent protein (GFP) or BMP2 gene at 200 MOI in the Ad-BMP2 and DFb-BMP2 groups (Figure 4d). In (multiplicity of infection) to achieve 480% transduction the medullary filling zone (Figure 4e), the %porosity was efficiency (Figures 1a and b) and efficient BMP2 protein significantly greater in the Ad-BMP2, DFb, DFb-GFP and production. In the Von Kossa staining performed at 7 DFb-BMP2 groups (Figure 4f), the %mineralizing area days after gene transduction, BMP2-expressing DFb was significantly greater in the Ad-BMP2 and DFb-BMP2 showed mineralized nodule formation (Figures 1d and groups (Figure 4g), and the %lamellar bone composition f) and positive alkaline phosphatase stain (Figures 1e and was significantly greater in the Ad-BMP2 and DFb-BMP2 g). There was no nodule formation in DFb-GFP and DFb groups (Figure 4h). In the adjacent cortex zone (Figure 5a), alone (Figure 1f), and significantly greater numbers of the mineral apposition rate (MAR), osteon filling period DFb-BMP2 were positive on the alkaline phosphatase and osteon activation frequency was significantly greater staining than DFb-GFP and DFb alone (Figure 1g). in the Ad-BMP2 and DFb-BMP2 groups (Figures 5b–d). In the adjacent medullary zone (Figure 5e), the MAR was Clinical assessment significantly greater in the Ad-BMP2, DFb and DFb-BMP2 Equine rib drill defects were treated by subcutaneous groups (Figure 5f), the %mineralizing surface was injection of Gey’s balanced salt solution (GBSS: control), significantly greater in the Ad-BMP2, DFb, DFb-GFP and

Gene Therapy Cell- and gene-based therapy for bone healing A Ishihara et al 735

Figure 1 In vitro gene transduction and osteogenic differentiation of equine dermal fibroblast (DFbs). Equine DFbs were seeded in 48-well plates at a density of 5 Â 104 cells per well and, 24 h later, the monolayer cells were transfected with adenoviral (Ad) vector encoding bone morphogenetic protein-2 gene (Ad-BMP2) or green fluorescent protein gene (Ad-GFP) at 200 MOI. Successful transgene GFP expression was achieved within 2 days (a) with 84% efficiency (b). Efficient BMP2 production in the culture medium was confirmed by ELISA, with the peak protein secretion at 7 days after gene transduction (c). The Von Kossa staining at day 7 demonstrated the mineralized bony nodule formations in DFb-BMP2, but DFb and DFb-GFP showed no nodule formations (d, f)(*P ¼ 0.001). The alkaline phosphatase (ALP) staining at day 7 demonstrated the significantly greater ALP uptake in DFb-BMP2 compared with DFb and DFb-GFP (#P ¼ 0.007) (e, g). Data were expressed as mean+s.e.m.

DFb-BMP2 groups (Figure 5g), and the surface activation In vivo tracking of transplanted DFb frequency was significantly greater in the Ad-BMP2, DFb Equine ilium drill defects were treated by subcutaneous and DFb-BMP2 groups (Figure 5g). injection of GBSS, DFb alone, DFb-GFP and DFb-BMP2

Gene Therapy Cell- and gene-based therapy for bone healing A Ishihara et al 736

Figure 2 Experimental models and assignments (a) and protocol (b) of the in vivo study. The rib drill defects were used to demonstrate an acceleration of bone healing by dermal fibroblast (DFb)-based bone morphogenetic protein-2 (BMP2) delivery, and the ilium drill defects were used for an in vivo tracking of transplanted DFb. Using six healthy adult horses, the 8 mm diameter 10–12 mm deep drill hole defects were created bilaterally in the tenth or eleventh ribs (three holes per rib, six holes per horse) and, 2 weeks later, treated by percutaneous injections of Gey’s balanced salt solution (GBSS: saline control, n ¼ 6 defects), adenoviral vector encoding green fluorescent protein (GFP) gene (Ad-GFP: 5 Â 109 IFU per defect, n ¼ 6), adenoviral vector encoding BMP2 gene (Ad-BMP2, n ¼ 6), DFb with no genetic modification (DFb: 5 Â 107 cells per defect, n ¼ 6), DFb with GFP gene transduction (DFb-GFP, n ¼ 6) and DFb with BMP2 gene transduction (DFb-BMP2, n ¼ 6). For fluorescent labeling of bone mineralization activity, calcein was administered intravenously at 3 and 4 weeks after the cell/vector injections. The 8 mm diameter 10–12 mm drill holes were also created in the tuber coxae region of ilium (two holes per side, four holes per horse) at days 40 and, 2 weeks later which was 2 days before the euthanasia, treated by percutaneous injections of GBSSS (n ¼ 6), DFb (n ¼ 6), DFb-GFP (n ¼ 6) and DFb-BMP2. All six horses were euthanized at 6 weeks after the cell/vector injection and the rib specimens and ilium drill hole tissues were harvested for analyses.

(Figure 2). An in vivo tracking of transplanted DFb was contain the DFb in the healing site, accumulate secreted successful, using the drill hole tissues harvested 2 days growth factors in the local region, target rich host cells by after injection. There was significant BMP2 gene up- BMP2 and potentially provide a healthy environment for regulation in the drill hole tissues with DFb-BMP2 transplanted cells. injection, but there was no significant GFP gene Bone regeneration has been induced by both cell and upregulation in the drill hole tissues with DFb-GFP gene vectors, but the potency between two effective injection (Figure 6a). The drill hole tissues were digested dosages and strategies have not yet been compared and seeded in culture dishes. The BMP2 protein con- within the same animal. Our model showed that centration in the culture medium was significantly cell-mediated BMP2 delivery appeared to promote bone higher in the DFb-BMP2 group than in the GBSS, DFb regeneration more effectively and extensively than did or DFb-GFP groups (Figure 6b), and the concentrations direct BMP2 gene delivery, at least in this model and at were less than the detection limit in all of GBSS, DFb the particular cell/vector dosages used in this study. It is and DFb-GFP-treated drill hole tissues and the 3/6 of not entirely clear whether the efficacy of DFb-BMP2 is DFb-BMP2-treated tissues (Figure 6b). primarily due to the secreted BMP2 affecting endo- genous progenitors or a direct contribution of BMP2- expressing DFb for bone regeneration. However, our Discussion study identified the bmp2-gene-expressing (Figure 6a) and BMP2-protein-secreting DFbs (Figure 6b) in the Skin cell-mediated gene therapy using a single percuta- injected drill defects at 48 h after transplantation neous injection of DFb genetically modified to secrete supporting a likely paracrine effect by the secreted BMP2 BMP2 can accelerate bone healing, which was at least protein. In addition, the in vitro results of our previous equally and potentially more effective than direct BMP2 study16 and the study reported herein (Figures 1d and e) gene delivery. In our study, DFb-BMP2-treated defects have shown that DFb directly formed mineral and an had greater amount of cortical bone filling and more osteoblast phenotype after Ad-BMP2 transduction, and mature lamellar bone formation than did Ad-BMP2- therefore may directly contribute to bone formation treated defects. For the same reasons delayed direct in vivo. Future study using an in vivo cell-tracking vector injection was performed,17,20 delayed cell injection technique to verify integration of the transplanted DFb into a solid granulation tissue may be advantageous to within the regenerated bone is warranted. Moreover, the

Gene Therapy Cell- and gene-based therapy for bone healing A Ishihara et al 737

Figure 3 Computed tomographic (CT) images and quantitative analysis of bone regeneration in equine rib drill defect model. Six skeletally mature horses had surgically created three drill hole defects in the tenth or eleventh ribs bilaterally (three defects per rib; six rib defects per horse). Two weeks after surgery, the six defects were treated with ultrasound-guided percutaneous injections of autologous dermal fibroblasts (DFbs) after BMP2 gene transduction (DFb-BMP2; n ¼ 6 defects), GFP gene transduction (DFb-GFP; n ¼ 6), DFb alone (DFb; n ¼ 6), Ad-BMP2 (n ¼ 6), Ad-GFP (n ¼ 6) or Gey’s balanced salt solution (GBSS; n ¼ 6). Quantitative CT of the ribs were performed at 6 weeks after injections. (a) Cross-sectional (left) and three-dimensional images (right) of the representative rib drill hole defects in each treatment group (scale bars: 1 cm). (b) Schematic images of the four regions evaluated by quantitative CT. In the cortical filling zone, bone volume was significantly greater in the Ad-BMP2, DFb-GFP and DFb-BMP2 groups (c), and cortical mineral density was significantly greater in the Ad-BMP2 group (d). In the medullary filling zone, bone volume was significantly greater in the Ad-BMP2 and DFb-BMP2 groups (e). Bone density was not changed in any group (f). In the adjacent cortex zone, mineral density was significantly greater in the Ad-BMP2 and DFb-BMP2 groups (g). In the adjacent medullary zone, mineral density was significantly greater in the DFb-BMP2 group (h). Data were expressed as mean+s.e.m. abcDifferent letters differ significantly (Po0.05). NS: no significant differences among groups. use of different osteogenic genes such as Runx2,11 the manipulation and control of many steps that occur BMP713 and Lim mineralization protein-315 may further in vivo with direct delivery. First, conditions can be improve the osteoinductive capacity of DFb-mediated controlled to optimize cell transduction efficiency, plus gene therapy. injecting cells increases cellularity at the repair site. Cell-mediated gene delivery method offers many Many viral vectors, and certainly the Ad vector, induces advantages, because using cells as vectors can permit a local tissue inflammation,21 although the cell injection

Gene Therapy Cell- and gene-based therapy for bone healing A Ishihara et al 738

Figure 4 Histological evaluation of the cortical and medullary bone filling regions of the rib drill hole defects. Six skeletally mature horses had surgically created three drill hole defects in the tenth or eleventh ribs bilaterally (three defects per rib; six rib defects per horse). Two weeks after surgery, the six defects were treated with ultrasound-guided percutaneous injections of autologous dermal fibroblasts (DFbs) after BMP2 gene transduction (DFb-BMP2; n ¼ 6 defects), GFP gene transduction (DFb-GFP; n ¼ 6), DFb alone (DFb; n ¼ 6), Ad-BMP2 (n ¼ 6), Ad-GFP (n ¼ 6) or Gey’s balanced salt solution (GBSS; n ¼ 6). Histological evaluations were performed at 6 weeks after injections. In the cortical filling zone (a) (scale bars: 0.5 mm), the %porosity and %mineralizing area was significantly greater in the DFb-BMP2 group (b, c), and the %lamellar bone composition was significantly greater in the Ad-BMP2 and DFb-BMP2 groups (d). In the medullary filling zone (e) (scale bars: 0.5 mm), the %porosity was significantly greater in the Ad-BMP2, DFb, DFb-GFP and DFb-BMP2 groups (f), the %mineralizing area was significantly greater in the Ad-BMP2 and DFb-BMP2 groups (g), and the %lamellar bone composition was significantly greater in the Ad-BMP2 and DFb-BMP2 groups (h). Data were expressed as mean+s.e.m. abcDifferent letters differ significantly (Po0.05).

might also induce an inflammatory response. In addition vector delivery would be previous exposure to wild-type to inflammation, activation of the innate immune adenovirus, the consequence of which would be limited response by viral vectors is well confirmed,22,23 and efficacy of the direct vector injection. In total, these indeed in horses injected directly with the Ad vector advantages of cell-mediated gene therapy may be develop detectable serum antibody within weeks.7 responsible for the relatively greater bone regeneration Complicating the potential effectiveness of direct Ad capacity shown in DFb-BMP2-treated rib drill defects.

Gene Therapy Cell- and gene-based therapy for bone healing A Ishihara et al 739

Figure 5 Histological evaluation of the adjacent cortex and medulla regions of the rib drill hole defects. Six skeletally mature horses had surgically created three drill hole defects in the tenth or eleventh ribs bilaterally (three defects per rib; six rib defects per horse). Two weeks after surgery, the six defects were treated with ultrasound-guided percutaneous injections of autologous dermal fibroblasts (DFbs) after BMP2 gene transduction (DFb-BMP2; n ¼ 6 defects), GFP gene transduction (DFb-GFP; n ¼ 6), DFb alone (DFb; n ¼ 6), Ad-BMP2 (n ¼ 6), Ad-GFP (n ¼ 6) or Gey’s balanced salt solution (GBSS; n ¼ 6). Histological evaluations were performed at 6 weeks after injections. In the adjacent cortex zone (a) (scale bars: 0.5 mm), the mineral apposition rate, osteon filling period and osteon activation frequency was significantly greater in the Ad-BMP2 and DFb-BMP2 groups (b–d). In the adjacent medullary zone (e) (scale bars: 0.5 mm), the mineral apposition rate was significantly greater in the Ad-BMP2, DFb and DFb-BMP2 groups (f), the %mineralizing surface was significantly greater in the Ad-BMP2, DFb, DFb-GFP and DFb-BMP2 groups (g), and the surface activation frequency was significantly greater in the Ad-BMP2, DFb and DFb-BMP2 groups (h). Data were expressed as mean+s.e.m. abcDifferent letters differ significantly (Po0.05).

In equine metacarpal/tarsal bone, DFb-BMP2-treated secreted BMP2 from the transplanted DFb to induce bone osteotomies were composed of mature cartilage that did regeneration. not fully convert to bridging woven bone.16 The authors It is important to point out that direct BMP2 gene speculated that incomplete endochondral ossification delivery was still an effective strategy for bone regenera- may have been caused by the limited number of tion. In agreement with our previous studies,7,17 delayed osteoprogenitor cells in the dominantly cortical bone vector injection (2 weeks after surgery) into the mature healing site. Although BMP2 is a key factor for granulation tissue seemed to be supportive of an chondrocyte maturation during endochondral ossifica- effective outcome, presumably by viral containment, tion, multiple signaling molecules from osteoprogenitor growth factor accumulation and gene transduction to a cells such as BMP2, BMP4, BMP6 and BMP7 are involved large receptive host cell population.7,17 There is always a in the apoptosis of hypertrophic chondrocytes.24 We dilemma in accurately calculating an in vivo transduction speculate that the intramedullarly delivered DFb-BMP2 efficiency (that is, number of viral particles per cell) and may work synergistically with host osteoprogenitor cells vector/cell dissemination (that is, local tissue distribu- present in the cancellous bone environment and tion). The numerous physical factors in vivo, including efficiently complete the endochondral ossification pro- regional blood flow or lymphatic drainage, tissue cess. The rib also has such a large population of host solidity, as well as type and concentration of native cells osteoblastic lineage that may serve as a target for the at the treatment site, make an in vivo transduction

Gene Therapy Cell- and gene-based therapy for bone healing A Ishihara et al 740 Our results indicated that placing DFbs in the medullar cavity by itself may be beneficial to facilitate bone regeneration. In this study, the volume of medul- lary filling was greater in all DFb-treated bone defects, regardless of the BMP2 gene transduction (that is, all of DFb, DFb-GFP, DFb-BMP2) compared with saline control (Figures 4e and f). This result may suggest that the cancellous bone environment may provide the differ- entiation factors and three-dimensional architecture, such that DFbs could sufficiently undergo osteogenic differentiation. Possibly, the transplanted DFbs synthe- sized the extracellular matrix to support the migration of host osteoprogenitor cells. DFbs have been shown to produce cartilage-like dense extracellular matrix when cultured in osteogenic-induction medium,25,26 which may serve as a biological scaffold to stimulate recruit- ment of osteoprogenitor cells from the surrounding cancellous bone or promote endochondral ossification. In this regard, DFb-mediated BMP2 therapy may be considered to have a dual osteoinductive and osteocon- ductive effect; therefore, possibly, the simultaneous injection of DFbs and stem cells would promote greater bone regeneration than the use of an individual cell type. To the authors’ knowledge, this is the first study reporting the increased osteon/surface activation frequency by gene- or cell-mediated BMP2 delivery (Figures 5d and h). Activation of bone remodeling is initiated as an osteoclastogenesis, not osteoblastogenesis, and therefore is not considered as a primary function of Figure 6 In vivo tracking of transplanted dermal fibroblasts (DFbs). BMPs. However, in recent years, new roles of BMPs Six skeletally mature horses had surgically created two drill hole associated with osteoclasts have been discovered, in- defects in the tuber coxae aspect of the ilium bilaterally (two defects cluding a recruitment of monocytic precursor cells,25 per side; four ilium defects per horse). Two weeks after surgery, the stimulation of osteoclastic differentiation27 and facilitat- four defects were treated with ultrasound-guided percutaneous ing the transition from bone resorption to bone formation injections of autologous DFbs after BMP2 gene transduction (DFb- 24 BMP2; n ¼ 6 defects), GFP gene transduction (DFb-GFP; n ¼ 6), DFb phases as coupling factors. In addition, BMP2 has been alone (DFb; n ¼ 6) or Gey’s balanced salt solution (GBSS; n ¼ 6). The shown to activate hematopoietic cells of murine bone drill hole tissues were harvested at 2 days after injections. There was marrow resulting in an upregulation of osteoclast significant BMP2 gene upregulation in the drill hole tissues with differentiation factor and enhanced osteoclast-like multi- DFb-BMP2 injection (*P ¼ 0.003) (a), but there were no significant nucleated cell formation.28 Recruitment of hematopoietic GFP gene upregulation in the drill hole tissues with DFb-GFP cells and conversion of these precursors into osteoclasts injection (NS: no significant differences) (panel a). The drill hole are important processes in the initiation phase of bone tissues were digested and seeded in culture dishes. The BMP2 29 protein concentration in the culture medium was significantly remodeling. Therefore, in our study, the intra- higher in the DFb-BMP2 group than in the GBSS, DFb or DFb-GFP medullarly delivered BMP2 may stimulate the hematopoietic groups (#P ¼ 0.001) (panel b), and the concentrations were less than precursors present in the rib marrow cavity and activate the detection limit in all of GBSS, DFb and DFb-GFP-treated drill bone turnover of the adjacent cortex and medulla. As hole tissues and the 3/6 of DFb-BMP2-treated tissues (ND: not BMP2 may have a role in coordinating the functional detected) (panel b). Data were expressed as mean+s.e.m. connection between osteoclastic and osteoblastic activi- ties as a coupling factor,29 the increase in bone resorption could be tightly orchestrated with subsequent bone efficiency, at best, a good estimate of the MOI at the formation ensuring an anabolic bone remodeling. One injection site. In addition, the well-confined drill hole of the potential clinical applications of this aspect is area should provide adequate vector containment, and bone fragility disorders such as osteoporosis, because the presence of highly cellular bone marrow tissue encouraging anabolic remodeling in specific long bones adjunct to the drill hole should be beneficial to provide may be a more physiological method to improve bone sufficient host progenitor cells for efficient gene trans- mineral density than shutting down systemic bone duction. An advantage of direct gene therapy is the less turnover. complexity of vector preparation, as opposed to cell- In vivo cell tracking successfully identified the func- mediated therapy, but can involve a risk of innate tional presence of our injected cells in the granulation immune reaction or an increase in neutralizing antibody, tissue bed. We chose to harvest the ilium drill hole tissue which may prohibit multiple vector injections in the 2 days after the DFb transplantation. Our results showed same individual. In the clinical setting, single direct that percutaneously transplanted DFb-BMP2 were pre- Ad-BMP2 treatment may be applied to an orthopedic sent at the recipient site, maintained BMP2 transgene patient without the need to wait for sufficient numbers of expression and local BMP2 production, at least when the autologous cells to be prepared for cell-mediated BMP2 cells were retrieved and re-cultured (Figure 6), The therapy. DFb-GFP-treated drill hole tissue also had higher GFP

Gene Therapy Cell- and gene-based therapy for bone healing A Ishihara et al 741 transgene expression than did DFb or DFb-BMP2. The injection. The ilium defect tissues were processed for the BMP2 gene transduction could be supportive for cell PCR to detect BMP2/GFP gene expression and the viability, so that more number of DF-BMP2 than DFb- collagenase digestion to culture BMP2-secreting or GFP GFP might survive. Cultures of stem cells engineered to expressing DFb. express BMP2 outlive the cells transduced with marker genes and appeared more robust and morphologically DFb isolation and in vitro osteogenic differentiation healthy. It is also possible that, because a significantly Full-thickness skin tissue was harvested using a 5-mm greater GFP expression was not detected in the DFb- diameter biopsy punch from the pectoral region (10–12 GFP-treated tissues, an active bone healing environment punches per horse) from each of the six horses. The may be more favorable for the DFb to sustain BMP2 dermal layer was dissected from the epidermis under a expression than GFP expression. microscope, and DFbs were isolated by type-1 collage- nase digestion (GIBCO, Grand Island, NY, USA) and cultured in Dulbecco’s modified Eagle’s medium sup- Materials and methods plemented with L-glutamine (300 mgmlÀ1), penicillin (30 mgmlÀ1), streptomycin (30 mgmlÀ1) and 10% fetal

Ad vector production bovine serum at 37 1C in a 5% CO2 atmosphere. To show Recombinant, replication-deficient, serotype-5 Ad in vitro gene transduction and osteogenic differentiation, vectors containing either a 1547 base-pair open reading DFbs were seeded in triplicate wells of 48-well plates frame segment of human BMP2 (Ad-BMP2) or GFP (Falcon, Franklin Lakes, NJ, USA) at a density of 5  104 (Ad-GFP) under the control of the cytomegalovirus cells per well and, 24 h later, transfected with Ad-BMP2 promoter were generated.16,17 The expression of trans- or Ad-GFP at 200 MOI (for example, 2  102 infectious genes was verified in cell culture.8 unit (IFU) per cell; 1 Â107 IFU per well) (Adeno-X Rapid Titer Kit; Clontech, Mountain View, CA, USA). Two and Experimental design seven days after gene transduction, gene transduction Six skeletally mature horses (weighing 445–550 kg) were efficiency was quantified by fluorescent microscopy, used to isolate DFbs by full-thickness skin punch biopsy. and BMP2 protein production confirmed with ELISA In vitro gene transduction and osteogenic differentiation (enzyme-linked immunosorbent assay; R&D Systems, were confirmed in DFbs from all six horses. The in vivo Minneapolis, MN, USA). Seven days after gene transduc- study used the same six horses in two separate tion, osteogenic differentiation of DFb was quantified by experiments; (1) rib drill hole defects for an assessment the number of Von Kossa-positive mineralized bony of bone regeneration and (2) ilium drill hole defects for nodules within the well and the intensity score of an in vivo cell tracking (Figure 2). All procedures were alkaline phosphatase (Sigma-Aldrich) staining in three approved by the Institutional Laboratory Animal Care representative  100 microscopic fields per well using and Use Committee at The Ohio State University. the following grading scheme: 0 (0% of cells in the field At day 0, the same six horses had surgically created showing stain uptake), 1 (1–25%), 2 (26–50%), 3 (51–75%) three drill hole defects in the tenth or eleventh ribs and 4 (76–100%). bilaterally (three defects per rib; six rib defects per horse) (Figure 2a). Skin biopsy was performed 2 weeks before Rib drill hole defect model the rib drilling (day 14) to culture sufficient DFbs for The rib drill hole defects were used to evaluate the injection. For each horse, the six rib drill hole defects relative efficacy of direct gene- and DFb-mediated BMP2 were treated with percutaneous injection of autologous delivery to induce bone regeneration. Skin punch biopsy DFbs after BMP2 gene transduction (DFb-BMP2; n ¼ 6 was performed at 2 weeks before the rib drilling (day 14). defects), GFP gene transduction (DFb-GFP; n ¼ 6), DFb At day 0, general anesthesia was induced by xylazine alone (DFb; n ¼ 6), Ad-BMP2 (n ¼ 6), Ad-GFP (n ¼ 6) or (Rompun; Bayer, Pittsburgh, PA, USA; intravenously GBSS (n ¼ 6) (Sigma-Aldrich, St Louis, MO, USA). The (i.v.), 1.1 mg kgÀ1), ketamine (Ketaset; Fort Dodge six defects in each horse were assigned in a block design Animal Health, Overland Park, KS, USA; i.v., to rotate these injections. All injections were adminis- 2.2 mg kgÀ1) and diazepam (Valium; Roche, Madison, tered 2 weeks after surgery (day 14). Six weeks after the WI, USA; i.v., 0.11 mg kgÀ1) and maintained by Isoflur- injection (day 56), the horses were euthanized and rib ane (IsoFlo; Abbott, Parsippany, NJ, USA; infusion, specimens were harvested. Efficacy was assessed by qCT 2–5%). After aseptic preparation of the bilateral latero- and histology. ventral thorax, 4-cm stab incisions were made through Using the same six horses, two drill hole defects were the skin and external and internal abdominal oblique also created bilaterally on the tuber coxae of the ilium muscles, and a low-speed cordless electric drill was used (two defects per side, four ilium defects per horse). to create 8-mm diameter 10-mm deep defects on the Separate skin biopsies were performed 2 weeks before tenth or eleventh ribs bilaterally. Three drill holes were the ilium drilling (day 26) to culture sufficient DFbs. For made in each rib at 10-cm apart (that is, 6 rib defects per each horse, the four ilium drill hole defects were treated horse), and the most ventral hole was created at 20-cm with percutaneous injection of DFb-BMP2 (n ¼ 6 defects), dorsal from the costochondral junction. Efforts were DFb-GFP (n ¼ 6), DFb (n ¼ 6) or GBSS (n ¼ 6). The four made to create drill holes in the middle of the ribs. The defects in each horse were assigned in a block design to three drill hole defects in each rib were created by three rotate these injections. The drill defects were created at separate stab incisions to prevent the periosteal commu- day 40 of the experiment, the injections were adminis- nication between the defects, and were thoroughly tered at day 54 (2 weeks after drilling) and the horses flushed with sterile saline solution to remove all were euthanized at day 56; therefore, the tissues within bone shards. Antibiotics were administered for 1 day the ilium drill holes were harvested 2 days after the after surgery (procaine penicillin (Crystacillin; Solvay,

Gene Therapy Cell- and gene-based therapy for bone healing A Ishihara et al 742 Marietta, GA, USA; intramuscularly, 22 000 IU kgÀ1) and (Hounsfield units) was used for cortical filling and gentamicin (Gentocin; Schering, Kenilworth, NJ, USA; adjacent cortex zones to trace ‘compact’ bone but exclude i.v., 6.6 mg kgÀ1)); the anti-inflammatory drug (phenyl- ‘spongial’ bone, and fibrous, muscle and fat tissues butazone (Butazolidin; Schering; orally, 4.4 mg kgÀ1) was (Mimics software tutorial manual; Materialise). Similarly, administered for 3 days after surgery. three central coronal slices across the medullar zone were One week after the surgery (day 7), DFbs were seeded selected, and the volume and mineral density of in 18 large culture flasks (75-cm2;3 106 cells per flask). medullary bone filling was calculated within the 8-mm On day 12, 2 days before injection, the DFbs (80–90% diameter circular ROIs positioned at the center of the confluent) were treated with Ad-BMP2 (DFb-BMP2; 6 remaining defect. Using the same three coronal slices, flasks), Ad-GFP (DFb-GFP; 6 flasks) or untreated (DFb; mineral density of the adjacent medulla was calculated 6 flasks). The total cell number at this time point was within four of the 4-mm diameter circular ROIs estimated at 1 Â107 DFb per flask based on a pilot trial concentrically positioned at 8-mm away from the center and hemacytometer counting (Hausser Scientific, Hor- of the remaining defect. The threshold level of 150– sham, PA, USA). Ad-BMP2/GFP were administered at 3000 HU was used for medullary filling and adjacent 200 MOI; therefore, 2  109 IFU of vector was adminis- medulla zones to trace both ‘compact’ and ‘spongial’ tered in each flask. bone but exclude fibrous, muscle and fat tissues (Mimics Two weeks after the surgery (day 14), 48 h after DFb software tutorial manual; Materialise). All HU values gene transduction, cells were harvested by trypsinization were converted to mineral density using potassium (GIBCO), counted by hematocytometer, centrifuged and phosphate standards (CT Calibration Phantom; resuspended into a 500 ml total volume with GBSS Mindways Software Inc., San Francisco, CA, USA). containing 5  107 cells for each of DFb, DFb-GFP and DFb-BMP2. All cell injections were completed within Histological evaluation 60–90 min after trypsinization. Moreover, the Ad-GFP Calcified rib specimens were dehydrated in alcohol, and Ad-BMP2 vectors were diluted into a 500 ml total embedded in methylmethacrylate, cut into 10-mm volume with GBSS containing 5  109 IFU. sections in the transverse plane perpendicular to the Horses were placed under general anesthesia, and the longitudinal axis at the center of the drill hole and assigned rib defects were treated by percutaneous stained with Masson’s Trichrome. The rib specimens ultrasound-guided (Technos MPX, Esaote S.p.A, Genoa, were evaluated semi-quantitatively for four regions: Italy) needle injection of 5  107 cells of DFb-BMP2, cortical filling zone, medullary filling zone, adjacent DFb-GFP, DFb, or 5  109 IFU of Ad-GFP or Ad-BMP2, or cortex and adjacent medulla (Figure 2). GBSS in a 500 ml volume directly into the drill hole The cortical filling zones were evaluated for the defects. For fluorescent labeling of bone mineralization porosity (%unfilled area), maturity (%composition of activity, calcein (Sigma-Aldrich) was administered at a immature woven, mature lamellar bone and mixture rate of 20 mg kgÀ1, dissolved in 2% sodium bicarbonate within the filled area) and mineralization activity solution (Abbott Laboratories, North Chicago, IL, USA) (%calcein-labeled area within the filled area), by point i.v. at 3 and 4 weeks after the DFb injection (days 35 and counting with the microscopic grids under  200 42, respectively). magnification. The five representative areas in the cortical bone filling in o5-mm depth from the top of Quantitative computed tomography drill hole were evaluated and averaged. Six weeks after the cell/vector rib injection (day 56), Similarly, the medullary filling zones were evaluated horses were sedated with xylazine and euthanized by for the porosity, maturity and mineralization activity, lethal i.v. overdose of pentobarbital (Beuthanasia; i.v., by point counting with the microscopic grids under  200 2.2 mg kgÀ1). Immediately after the euthanasia, the ribs magnification. The five representative areas of the with drill hole defects were dissected. qCT (Picker PQS medullary bone filling in o10-mm depth from the top Helical CT Scanner; Philips Medical Systems, N.A., of drill hole were evaluated and averaged. Bothell, WA, USA) was performed in the transverse The adjacent cortices were evaluated for the MAR, plane (perpendicular to the rib axis) at 1-mm contiguous osteon filling period and osteon activation frequency. slices to evaluate bone regeneration. The entire drill hole Within the 1-mm2 cortical bone area under  200 defect, along with 1.5 cm margins dorsal and ventral, magnification, the number of single-labeled osteon was imaged. Two-dimensional images in the coronal (sL.On) and double-labeled osteon (dL.On) were counted plane (perpendicular to the drill hole axis) were created and, in the 2–3 representative dL.On within the 1-mm2 using an image analysis software (Mimics, Materialise, area, the radii of Haversin canal (Hv.Rd), inner label Ann Arbor, MI, USA). The coronal sections allowed (In.Rd), outer label (Out.Rd) and cement line (Cm.Rd) visualizing the cross-sectional area of cylinder drill hole were measured using the microscopic grids, and these defects and using the 8-mm diameter circular region of values were used for the following calculations.30,31 interest (ROI) to trace the original size of the drill hole. MARðmm per dayÞ¼ðOut:Rd À In:RdÞÄ7 Three central coronal slices across the cortical zone were selected, and the volume and mineral density of cortical bone filling was calculated within the 8-mm diameter Osteon filling periodðdaysÞ circular ROIs positioned at the center of the remaining defect. Using the same three coronal slices, the mineral ¼½lnðHv:Rd=Cm:Rdފ Ä ½lnðIn:Rd=Out:RdÞÄ7Š density of adjacent cortex was also calculated within four of the 4-mm diameter circular ROIs concentrically Osteon activation frequencyðno: per mm2 per yearÞ positioned at 8-mm away from the center of the remaining defect. The threshold level of 600–3000 HU ¼½ðsL:OnÞ=2 þ dL:OnŠÄ½ðCm:Rd À Hv:RdÞ=MARŠÂ365

Gene Therapy Cell- and gene-based therapy for bone healing A Ishihara et al 743 The five representative areas of the adjacent cortex in extracted from the half of the tissue by Trizol technique o5-mm distance from the edge of drill hole were and the other half were digested by collagenase for 6 h. evaluated and averaged. Transgene expression of human BMP2 and GFP were The adjacent medullae were evaluated for the MAR, quantified by SYBR real-time reverse transcriptase-PCR percent active bone surface (BS) and surface activation with the ABI PRISM 7000 Sequence Detection System frequency. Within the 1-mm2 medullar bone area under (Applied Biosystems, Foster City, CA, USA). Mean fold  200 magnification, the trabecula width (Tr.Wd) and change was calculated using custom-designed primers the distance between two calcein labels (Lb.Dt) were for the hbmp2 (forward: 50-AAAACGTCAAGCCAAACA measured in the 2–3 double-labeled surface and CAAA-30, reverse: 50-GTCACTGAAGTCCACGTACAA averaged; the intercept counting of total BS and AGG-30) and gfp gene (forward: 50-CATGATATAGACG single-/double-labeled MS (mineralizing surface) were TTGTGGCTGTTG-3, reverse: 50-AAGCTGACCCTGAA performed with the microscopic grids, and these values GTTCATCTGC-30) in the DFb, DFb-GFP and DFb-BMP2- were used for the following calculations.32,33 treated drill hole tissue relative to expression of GBSS- MARðmm per dayÞ¼ðLb:DtÞÄ7 treated drill hole tissue and relative to expression of the endogenous equine b-actin gene (forward: 50-GGGCAT 0 0 Percent mineralizing surface ¼ðMS=BSÞÂ100 CCTGACCCTCAAG-3 , reverse: 5 -TCCATGTCGTCCC AGTTGGT-30) using the 2ÀDDCT method.34,35 Surface activation frequencyðper yearÞ The digested tissues were centrifuged, washed with GBSS and seeded in 25-cm2 culture flasks (day 57). Four ¼½MARÂðMS=BSފ Ä Tr:WdÂ365 days later (day 61), the culture medium was collected for The five representative areas of the adjacent medulla in the detection of BMP2 protein by ELISA. The lowest o5-mm distance from the edge of drill hole were BMP2 standard (62.5 pg mlÀ1) was considered as detec- evaluated and averaged. tion limit, because the mean±2 s.d. optical density in triplicate wells did not overlap that of zero standard Ilium drill hole defect model (0 pg mlÀ1). The ilium drill hole defects were used for in vivo cell tracking. Skin punch biopsy was performed at day 26 to culture the efficient DFb. At day 40, the drill hole defects Statistical analysis were created bilaterally in the tuber coxae regions of Repeated-measure ANOVA (analysis of variance) (SAS ilium as standing surgery. The horses were restrained in Institute Inc., Cary, NC, USA) was used to evaluate the stocks, sedated with xylazine (i.v., 1.1 mg kgÀ1), and the effects of DFb-mediated BMP2 therapy with the post-test skin overlying the tuber coxae was aseptically prepared multiple comparisons between the treatment groups and injected with local anesthesia (lidocaine; subcutaneous, using Proc Mixed statistical models for continuous 100 mg). Stab incisions of 2 cm were made on the lateral outcomes (that is, Von Kossa bony nodule count, qCT, aspect of the tuber coxae, and a low-speed cordless histological evaluation, reverse transcriptase-PCR gene electric drill was used to create 8-mm diameter 10-mm expression and ELISA BMP2 concentration data) and deep defects on the tuber coxae. Two drill holes were Genmod statistical models for categorical outcomes (that made in each tuber coxae at 10-cm apart (that is, four is, alkaline phosphatase stain uptake score). Repeated ilium defects per horse). The two drill hole defects in variables were considered to be nested within horses, each side were created by the two separate stab incisions and the distribution of data was assessed by the use of to prevent the periosteal communication between the a subset of normality tests (such as the Shapiro–Wilk, defects, and were thoroughly flushed with sterile saline Kolmogorov–Smirnov, Cramer–von Mises and Anderson– solution to remove all bone shards. Antibiotics and anti- Darling tests). Fisher’s exact tests were used to compare inflammatory drugs were administered at the same dose the frequency outcomes (that is, the occurrence of fractures and length as the rib drilling. All procedures were in rib drill hole defects) between the treatment groups. approved by the Institutional Laboratory Animal Care Significant level was set at Po0.05 for all analyses. Power and Use Committee at The Ohio State University. calculations were made using means, s.d, sample sizes and The DFbs were prepared by the same protocol as for 5% a-error level for all outcomes differed significantly in the rib injection. The DFbs were seeded in 18 flasks given treatment groups. at day 47 (1 week after ilium drilling), treated with Ad-BMP2, Ad-GFP or untreated at day 52, and harvested and resuspended into a 500 ml total volume with GBSS Conflict of interest containing 5  107 cells for each of DFb, DFb-GFP and DFb-BMP2 at day 54 (48 h after gene transduction). The authors declare no conflict of interest. Horses were restrained in stocks, sedated, aseptically prepared, and the assigned ilium defects were treated by percutaneous ultrasound-guided needle injection of Acknowledgements 5  107 cells of DFb-BMP2, DFb-GFP, DFb or GBSS in a 500 ml volume directly into the drill hole defects. The drill We thank Dr Jose Mendez, Dr Phillip Lerche, Dr Turi hole size and the cell dosage of DFb-BMP2, DFb-GFP and Aarnes, Dr Euta´lio Pimenta, Dr Valerie Samii, DFb were exactly same as the rib drill hole defects Dr Matthew Allen, Rebecca Hancock, David Smith, making these two model comparable. Archna Hazelbaker and Nancy Weber for technical Immediately after the euthanasia at day 56 (2 days support. This study was supported by National Institute after the cell injection), the tissue within the ilium drill of Arthritis and Musculoskeletal and Skin Diseases, K08 hole defects were aseptically harvested. The RNA was AR049201.

Gene Therapy Cell- and gene-based therapy for bone healing A Ishihara et al 744 References 19 Zachos TA, Bertone AL, Wassenaar PA, Weisbrode SE. Rodent models for the study of articular fracture healing. J Invest Surg 1 Evans CH, Ghivizzani SC, Robbins PD. Orthopedic gene therapy 2007; 20: 87–95. in 2008. Mol Ther 2009; 17: 231–244. 20 Betz OB, Betz VM, Nazarian A, Egermann M, Gerstenfeld LC, 2 Tseng SS, Lee MA, Reddi AH. Nonunions and the potential of Einhorn TA et al. Delayed administration of adenoviral BMP-2 stem cells in fracture-healing. J Bone Joint Surg Am 2008; 90: vector improves the formation of bone in osseous defects. Gene 92–98. Therapy 2007; 14: 1039–1044. 3 Patterson TE, Kumagai K, Griffith L, Muschler GF. Cellular 21 Appledorn DM, Patial S, McBride A, Godbehere S, Van Rooijen strategies for enhancement of fracture repair. J Bone Joint Surg N, Parameswaran N et al. Adenovirus vector-induced innate Am 2008; 90: 111–119. inflammatory mediators, MAPK signaling, as well as adaptive 4 Woo SL. From ‘Gene Therapy’ to ‘Gene and Cell Therapy’: why? immune responses are dependent upon both TLR2 and TLR9 Mol Ther 2008; 16: 1899–1900. in vivo. J Immunol 2008; 181: 2134–2144. 5 Betz OB, Betz VM, Nazarian A, Pilapil CG, Vrahas MS, Bouxsein 22 Descamps D, Benihoud K. Two key challenges for effective ML et al. Direct percutaneous gene delivery to enhance healing adenovirus-mediated liver gene therapy: innate immune of segmental bone defects. J Bone Joint Surg Am 2006; 88: 355–365. responses and hepatocyte-specific transduction. Curr Gene Ther 6 Betz OB, Betz VM, Nazarian A, Egermann M, Gerstenfeld LC, 2009; 9: 115–1127. Einhorn TA et al. Delayed administration of adenoviral BMP-2 23 Tian J, Xu Z, Smith JS, Hofherr SE, Barry MA, Byrnes AP. vector improves the formation of bone in osseous defects. Gene Adenovirus activates complement by distinctly different Therapy 2007; 14: 1039–1044. mechanisms in vitro and in vivo: indirect complement activation 7 Ishihara A, Shields KM, Litsky AS, Mattoon JS, Weisbrode SE, by virions in vivo. J Virol 2009; 83: 5648–5658. Bartlett JS et al. Osteogenic gene regulation and relative 24 McCullough KA, Waits CA, Garimella R, Tague SE, Sipe JB, acceleration of healing by adenoviral-mediated transfer of Anderson HC. Immunohistochemical localization of bone human BMP-2 or -6 in equine osteotomy and ostectomy models. morphogenetic proteins (BMPs) 2, 4, 6, and 7 during induced J Orthop Res 2008; 26: 764–771. heterotopic bone formation. J Orthop Res 2007; 25: 465–472. 8 Zachos T, Diggs A, Weisbrode S, Bartlett J, Bertone A. 25 Sommar P, Pettersson S, Ness C, Johnson H, Kratz G, Junker JP. Mesenchymal stem cell-mediated gene delivery of bone Engineering three-dimensional cartilage- and bone-like tissues morphogenetic protein-2 in an articular fracture model. Mol using human dermal fibroblasts and macroporous gelatine micro- Ther 2007; 15: 1543–1550. carriers. J Plast Reconstr Aesthet Surg 2009; (e-pub ahead of print). 9 Zilberman Y, Kallai I, Gafni Y, Pelled G, Kossodo S, Yared W et al. 26 Okamoto M, Murai J, Yoshikawa H, Tsumaki N. Bone Fluorescence molecular tomography enables in vivo visualiza- morphogenetic proteins in bone stimulate osteoclasts and tion and quantification of nonunion fracture repair induced by osteoblasts during bone development. J Bone Miner Res 2006; genetically engineered mesenchymal stem cells. J Orthop Res 21: 1022–1033. 2008; 26: 522–530. 27 Itoh K, Udagawa N, Katagiri T, Iemura S, Ueno N, Yasuda H 10 Takahashi K, Tanabe K, Ohnuki M, Narita M, Ichisaka T, et al. Bone morphogenetic protein 2 stimulates osteoclast Tomoda K et al. Induction of pluripotent stem cells from adult differentiation and survival supported by receptor activator of human fibroblasts by defined factors. Cell 2007; 131: 861–872. nuclear factor-kappaB ligand. Endocrinology 2001; 142: 3656–3662. 11 Hirata K, Tsukazaki T, Kadowaki A, Furukawa K, Shibata Y, 28 Koide M, Murase Y, Yamato K, Noguchi T, Okahashi N, Moriishi T et al. Transplantation of skin fibroblasts expressing Nishihara T. Bone morphogenetic protein-2 enhances osteoclast BMP-2 promotes bone repair more effectively than those formation mediated by interleukin-1alpha through upregulation expressing Runx2. Bone 2000; 32: 502–512. of osteoclast differentiation factor and cyclooxygenase-2. Biochem 12 Hirata K, Mizuno A, Yamaguchi A. Transplantation of skin Biophys Res Commun 1999; 259: 97–102. fibroblasts expressing BMP-2 contributes to the healing of 29 Matsuo K, Irie N. Osteoclast-osteoblast communication. Arch critical-sized bone defects. J Bone Miner Metab 2007; 25: 6–11. Biochem Biophys 2008; 473: 201–209. 13 Krebsbach PH, Gu K, Franceschi RT, Rutherford RB. Gene 30 Metz LN, Martin RB, Turner AS. Histomorphometric analysis of therapy-directed osteogenesis: BMP-7-transduced human fibro- the effects of osteocyte density on osteonal morphology and blasts form bone in vivo. Hum Gene Ther 2000; 11: 1201–1210. remodeling. Bone 2003; 33: 753–759. 14 Rutherford RB, Moalli M, Franceschi RT, Wang D, Gu K, 31 McGee ME, Maki AJ, Johnson SE, Nelson OL, Robbins CT, Krebsbach PH. Bone morphogenetic protein-transduced human Donahue SW. Decreased bone turnover with balanced resorption fibroblasts convert to osteoblasts and form bone in vivo. Tissue and formation prevent cortical bone loss during disuse Eng 2002; 8: 441–452. (hibernation) in grizzly bears (Ursus arctos horribilis). Bone 15 Lattanzi W, Parrilla C, Fetoni A, Logroscino G, Straface G, 2008; 42: 396–404. Pecorini G et al. Ex vivo-transduced autologous skin fibroblasts 32 Tsangari H, Findlay DM, Fazzalari NL. Structural and remodeling expressing human Lim mineralization protein-3 efficiently indices in the cancellous bone of the proximal femur across form new bone in animal models. Gene Therapy 2008; 15: adulthood. Bone 2007; 40: 211–217. 1330–1343. 33 Chapurlat RD, Arlot M, Burt-Pichat B, Chavassieux P, Roux JP, 16 Ishihara A, Zekas LJ, Litsky AS, Weisbrode SE, Bertone AL. Portero-Muzy N et al. Microcrack frequency and bone re- Dermal fibroblast-mediated BMP2 therapy to accelerate bone modeling in postmenopausal osteoporotic women on long-term healing in an equine osteotomy model. J Orthop Res 2009; 28: bisphosphonates: a bone biopsy study. J Bone Miner Res 2007; 22: 403–411. 1502–1509. 17 Bertone AL, Pittman DD, Bouxsein ML, Li J, Clancy B, 34 Livak KJ, Schmittgen TD. Analysis of relative gene expression Seeherman HJ. Adenoviral-mediated transfer of human BMP-6 data using real-time quantitative PCR and the 2(-Delta Delta gene accelerates healing in a rabbit ulnar osteotomy model. C(T)) method. Methods 2001; 25: 402–408. J Orthop Res 2004; 22: 1261–1270. 35 Schefe JH, Lehmann KE, Buschmann IR, Unger T, Funke-Kaiser 18 Egermann M, Lill CA, Griesbeck K, Evans CH, Robbins PD, H. Quantitative real-time RT-PCR data analysis: current concepts Schneider E et al. Effect of BMP-2 gene transfer on bone healing and the novel ‘gene expression’s CT difference’ formula. in sheep. Gene Therapy 2006; 13: 1290–1299. J Mol Med 2006; 84: 901–910.

Gene Therapy