Adenoviral-mediated transfer of human BMP-6 gene accelerates healing in a rabbit ulnar osteotomy model

ELSEVIER

Journal of Orthopaedic Research

Journal of Orthopaedic Research 22 (2004) 1261-1270

www.elsevier.com/locate/orthres

Adenoviral-mediated transfer of human BMP-6 gene accelerates healing in a rabbit ulnar osteotomy model A.L. Bertone a,b3*, D.D. Pittman ', M.L. Bouxsein ', J. Li ', B. Clancy ', H.J. Seeherman a

Department of Veterinary Clinical Sciences, College of Veterinary Medicine, 601 Tharp St., The Ohio State University. Columbus, OH 43210, USA Women's Health and Bone/ Wyeth Research, Cambridge, MA, USA Cardiovascular and Metabolic Diseased Wyeth Research, Cambridge, MA, USA

Abstract This study evaluated healing of rabbit bilateral ulnar osteotomies 6 and 8 weeks after surgery in response to percutaneous injection of transgenic adenoviral (Ad) bone morphogenetic protein-6 (BMP-6) vector or green fluorescent protein vector control (Ad-GFP) administered 7 days after surgery compared to untreated osteotomy controls. The amount, composition and biomechanical properties of the healing bone repair tissue were compared among groups and to historical data for intact rabbit ulnae obtained from similar studies at the same institution. Quantitative computed tomography was used to determine area, density and mineral content of the mineralized callus in the harvested ulnae. Maximum torque, torsional stiffness, and energy absorbed to failure were determined at 1.5"/s. Calcified sections of excised ulnae (5 pm) were stained with Goldner's Trichrome and Von Kossa, and evaluated for callus composition, maturity, cortical continuity, and osteotomy bridging. Radiographic assessment of bone formation indicated greater mineralized callus in the ulnae injected with Ad-hBMP-6 as early as 1 week after treatment (2 weeks after surgery) compared to untreated osteotomy ulnae (p < 0.006) and Ad-GFP treated osteotomy ulnae (p < 0.002). Quantitative computed tomography confirmed greater bone area and bone mineral content at the osteotomy at 6 weeks in Ad-BMP-6 treated osteotomy as compared to untreated osteotomy ulnae (p < 0.001) and Ad-GFP treated osteotomy ulnae 0, < 0.01). Ad-BMP-6 treated osteotomy ulnae were stronger (p < 0.001 and 0.003) and stiffer (p = 0.004 and 0.003) in torsion at 6 weeks than untreated osteotomy ulnae or Ad-GFP treated osteotomy ulnae, respectively. Maximum torque, torsional stiffness, and energy absorbed to failure were greater in Ad-BMP-6 treated osteotomy ulnae compared to their respective untreated contralateral osteotomy ulnae at 8 weeks [p < 0.031. Maximum torque and torsional stiffness in the Ad-BMP-6 treated osteotomy ulnae were not different to intact ulnae values at 6 and 8 weeks. These experiments confirm that BMP-6 can be potently osteoinductive in vivo resulting in acceleration of bone repair. 0 2004 Orthopaedic Research Society. Published by Elsevier Ltd. All rights reserved. Keywords: Gene therapy; Fracture healing; B M P Rabbit osteotomy

Introduction

Approximately 6.5 million fractures occur annually in the US [62], and 10-15% result in delayed or nonunions [24], some of which may respond to treatment with bone morphogenetic proteins [BMPs]. The use of biologic agents, such as growth and differentiation factors, to enhance fracture repair has been extensively investigated [8,12]. Bone morphogenetic proteins are one family of *Corresponding author. Tel.: +1-614-292-6661~2;fax: +1-614-6885642. E-mail address: bertone. [email protected] (A.L. Bertone).

such differentiation factors, capable of inducing bone formation by recombinant protein application [2,4,13, 41,42,47,61,66,72] or gene delivery [ 1,3,5,7,16,18-20, 26,30,33,37,39,40,46,53,57,56,58,63-65,69-711. The first reports of clinical trials using BMP-2 and BMP-7 in human patients have appeared in the orthopedic literature, documenting their clinical efficacy in spinal fusion [9,34], repair of open tibial fractures [27,48], and healing of recalcitrant tibial fractures [22]. Investigation of new BMPs that may offer greater potency or altered mechanisms of bone formation may expand the field of responsive cases and reduce cost of treatment of this widespread problem.

0736-0266/$ - see front matter 0 2004 Orthopaedic Research Society. Published by Elsevier Ltd. All rights reserved. doi:lO. 1016/j.orthres.2004.03.014

1262

A.L. Bertone et al. / Journal of Orthopaedic Research 22 (2004) 1261-1270

This study focuses on one of the lesser characterized members of this protein family, BMP-6 [17]. BMP-6 regulates the differentiation of osteochondral progenitor cells during skeletal development [52] through autocrine feedback with other growth and differentiation factors [32,59]. During fracture healing, BMP-6 expression is induced in both soft and hard callus stages, indicating a role in bone healing [38]. BMP-6 appears to have a more potent [5,31] and yet, nonessential and redundant role in skeletal development as compared to BMP-2 [50,68]. BMP-6 may function through an alternate mechanism is compared to BMP-2 or BMP-4 [31,33]. Correspondingly, BMP-6 has induced direct muscle mineralization by gene delivery in mice as compared to no bone formation with BMP-2 or endochondral ossification with BMP-4 [33]. Transfer of the gene encoding BMP-6 may be a mechanism to facilitate bone healing through the direct mineralization process. BMP-6 supplementation has not been reported to promote bone repair in animal models. Gene delivery was selected for this study of BMP-6 bone forming activity in vivo in immunocompetent animals as several in vivo studies have evaluated the nature of fracture healing and the utility of gene transfer in this process [7,24,44,55,64]. Gene transfer by percutaneous injection leading to endogenous bioactive peptide production offers the potential advantages of a simple direct delivery without a requirement for a carrier or surgery and could be used to treat closed fractures in people. Percutaneous direct adenoviral delivery of BMP-9 and BMP-2 genes and retroviral delivery of BMP-4 gene has been successful in inducing bone formation in immunocompromized animals [4,30], and in a rat fracture model, respectively [64]. Gene transfer also provides the ability to study genes that have limited recombinant protein available, such as BMP-6, by sustained release of peptide, and greater potency due to endogenous post-translational modification [21,2534, 58,671. Adenoviral (Ad) vectors can efficiently deliver transgenes to repair cells, induce protein production in vivo, and be used for screening of functional biologic activity. The overall goal of this study was to test whether percutaneous injection of Ad-human (h)BMP-6 would accelerate bone healing in immunocompetent rabbits using an osteotomy model that heals spontaneously. This model was selected to reflect healing of closed long bone fractures in humans.

Materials and methods Study design

Six immature male Sprague-Dawley rats and 20, skeletally mature, New Zealand White rabbits (aged: 8-1 1 months and weight 3.54.2 kg) were used in this study. Historical control rabbits for comparative

intact ulnae biomechanical values were in the same age and weight ranges. The rats were used for Ad-BMP-6 dose-response studies following intramuscular administration. Two rabbits were used to verify Ad-BMP-6 and Ad-GFP gene expression following intramuscular injection. Eighteen rabbits had surgically created bilateral ulna osteotomies and were assigned to one of three groups. Two groups (12 rabbits) had one osteotomy treated with hBMP-6 gene in an adenoviral vector (Ad-hBMP-6) and the contralateral osteotomy untreated. Six of these rabbits were euthanized at 6 weeks (one group) and six at 8 weeks (second group). The third group ( n = 6) had one osteotomy treated with firefly green fluorescent protein (GFP) gene in an adenoviral vector (Ad-GFP) and the contralateral osteotomy untreated. The adenoviral construct was injected 1 week after osteotomy. All operations and injections were performed with the animal under general anesthesia and sterile conditions. The animal housing, care, and study protocol were approved by the Institutional Animal Care and Use Committee at Wyeth Research. All procedures were carried out according to AAALAC guidelines. Outcome assessments included weekly radiographs, peripheral quantitative computer tomography, mechanical testing and histology. Specimens were processed for histology after mechanical testing. Generation of adenouiral uector construct

Human BMP-6 cDNA was cloned from human placental and brain cDNA libraries [17]. The full-length clone defined a 1539 base-pair open reading frame that encodes the 513-amino acid hBMP-6. HBMP6 cDNA was subcloned into an expression vector (Ad ori I-]), under the control of the cytomegalovirus promoter, and co-transfected with a plasmid containing the 9-36 map units of E-1 defective human Adenovirus-5 into receptive human embryonic kidney 293 cells [ATCC, Rockland, Md] to produce infectious adenoviral particles containing hBMP-6 transgene. A similar vector containing G F P was generated as a control. Adenoviral preparations were purified by two rounds of cesium chloride centrifugation. Particle titers were determined by optical density at 260 nm and diluted to a concentration of 1 x 10I2/ml in phosphate buffered saline in 10% glycerol. DNA from the purified virus was subjected to PCR amplification and sequencing to verify presence of the transgene. Expression of transgenes was verified in cell culture. Adenouiral construct dose selection

Six rats were used to determine in uiuo osteogenic dose response of the adenoviral constructs. The rats were assigned to two groups (n = 3/ group) with bilateral injections into the quadriceps muscles paired as follows: (1) Ad-hBMP-6 at 1 x 10" particles in 100 pl versus AdhBMP-6 at 3 x 10" particles in 300 p1 and (2) Ad-hBMP-6 at 2 x 10" particles in 200 p1 versus Ad-GFP at 2 x 10" particles in 200 pl. Outcome measurements included muscle palpation and weekly radiographs for the 4 week duration of the study. Following euthanasia, the quadriceps tissue was harvested and stained with hematoxylin and eosin and Von Kossa stain following calcified tissue processing. Gene expression

Two rabbits were used to document gene expression of the adenoviral constructs on days 3, 7 and 13 following intramuscular injection in the lumbar and triceps muscles of 2 x 10" particles of Ad-BMP-6 or Ad-GFP administered in a buffer solution including 0.25% carbon black to mark the site of injection. Following euthanasia and radiographs, an -1 cm area of lumbar and tricep muscles including the carbon blacklAd-treatment site was harvested and immediately snap frozen in liquid nitrogen for subsequent gene expression studies. Frozen muscle was pulverized, RNA extracted [RNeasy@ Protect Minikit, Qiagen Inc., Valencia, CAI, and trace genomic DNA removed [SV 96 Total RNA Isolation, Promega Corp, Madison, WI]. Specific primers and probes were designed to amplify a uniquely identifying hBMP-6 gene sequence, a vector specific sequence that flanks the multiple cloning site and permits amplification of all cDNA inserts, and rabbit specific primers for glyceraldehyde phosphate dehydrogenase (GAPDH) for relative gene expression quantitation by real time reverse transcription polymerase chain reaction (RT-PCR) [ABI Prism'" 7700 Sequence Detection System, Applied Biosystems, Foster

A . L. Eertone et ul. I Journal of Orthopaedic Research 22 (2004) 1261-1270

1263

Table 1 Primer and probe sequences for RT-PCR analysis of gene expression HBMP-6

Forward primer Reverse primer Probe

5'-AGAATGCTCCTTCCCACTCAAC-3' 5'-GGTGAACCAAGGTCTGCACA-3' 5'-CATGAATGCAACCAACCACGCGA-3'

Vector

Forward primer Reverse primer Probe

5'-GACATGATAAGATACATTGATGAGTTTGG-3' 5'-GCAATAGCATCAC AA ATTTC ACAAAT-3'

5'-CAAACCACAACTAGA ATGCAGTGAAAAAAATGCTT-3'

Forward primer Reverse primer Probe

5'-CTGGGCTACACCGAGGACC-3' 5'-CCCCAGCATCGAAGGTAGAG-3' 5'-CGTCTCCTGCGACTTCAACAGTGCC-3'

Rabbit GAPDH

GAPDH = glyceraldehyde phosphate dehydrogenase.

City, CAI (Table 1). Gene expression was quantified by both gene specific (hBMP-6) and vector specific primers (hBMP-6 and GFP) [Taqman EZ RT-PCR Core Reagents, Applied Biosystems] in all muscle samples and normalized to rabbit GAPDH expression. All Taqman probes were labeled at the 5 end with 6-carboxy-fluorescein (FAM) and at the 3' end with a nonfluorescent dark quencher molecule 6-carboxytetramethylrhodamine(TAMRA). The thermal cycle protocol was 50 "C for 2 min, 60 "C for 30 min, 95 "C for 5 rnin, then 40 cycles of 95 "C for 15 s and 60 "C for 1 rnin. Surgical procedure and specimen harvest

The surgical procedure to create the bilateral osteotomies has been reported in detail elsewhere [13,42]. In brief, following general anesthesia and aseptic preparation of the forelimb, a high speed oscillating saw was used to create a blade width (0.5-1 mm) defect in the middiaphysis of the ulna. One week after osteotomy, the rabbits were sedated and the randomly assigned ulna osteotomy was treated by percutaneous fluoroscopic-guided needle injection of 2 x 10" AdhBMP-6 or Ad-GFP particles in 200 p1 volume directly into the osteotomy gap (Fig. 1). Radiographs were taken immediately after surgery, following treatment and weekly thereafter for the duration of the study. The rabbits were sedated with acepromazine (IM, 0.5 ml) and given a lethal dose of euthanasia solution (Euthasia V, 0.22 mVkg) at the time of sacrifice. Both forelimbs were carefully harvested, cleaned of soft tissue, wrapped in saline-soaked gauze, and frozen (-20 "C) in airtight containers for subsequent biomechanical testing. Radiographic evaluation

Radiographs were evaluated without knowledge of treatment for mineralization at the osteotomy and in soft tissues. Mineralized callus area was estimated by multiplying the length ( I ) and width (w)of the

callus measured on the lateral surface of the ulnae in the craniocaudal view (mm2) to assess change in mineralized callus size with time. Morphology of osteotomy mineralized callus

Excised forearms were scanned using peripheral quantitative computed tomography (pQCT, XCT3000, Stratec/Norland, Pforzheim, Germany) to determine the area, density and mineral content of the mineralized callus. Contiguous images (1 mm separation) perpendicular to the diaphyseal axis were obtained, starting at the distal extent of the fracture callus and extending proximally for 15 mm. The slice closest to the middle of the osteotomy was identified and crosssectional area (mm2) and total bone mineral content (BMC,g) of the mineralized portion of the fracture callus was computed for the middle slice and for one slice proximal and distal to the middle slice. Biomechanical testing and post-torsional testing fracture patterns

The forelimbs were thawed to room temperature, disarticulated at the elbow and carpus, and the ends of each forearm were potted in square molds using polymethylmethacrylate (PMMA). Following embedding, a scalpel blade was placed between the radius and ulna at the proximal and distal ends, and a high-speed dremel saw was used to cut through the radius at those points. The radius was then gently pried away and removed from the ulna. The ulnae were tested counterclockwise quasi-statically to failure in torsion (1.S"/s) using a servohydraulic materials testing system (Model 8500, Instron, Canon, MA). Left limbs were tested proximal end up and right limbs tested proximal end down to ensure similar directional rotation for both sides. The testing was stopped when the specimen broke, or when 30" of rotation was reached, whichever occurred first. The torque-rotation data were used to compute the maximum torque, torsional stiffness, and energy absorbed to failure. These values are relevant quantifiers of the structural integrity of the mineralizing callus. After testing, faxitron radiographs of the limbs were obtained to identify the location of the fractures [42]. An individual blinded to the time point and treatment group classified the fractures as occurring either (1) through the osteotomy site only, (2) through both the osteotomy site and the intact bone, or (3) through the intact bone only. Histologic processing

Fig. 1. Fluoroscopic-guided needle injection of adenoviral preparations directly into the ulnar osteotomy.

After mechanical testing, the calcified specimens were placed in 70% ethanol, dehydrated in graded concentrations of alcohol, and embedded in methylmethacrylate. During embedding, the positioning of the ulna was standardized in an attempt to ensure that the same region of the callus was evaluated in all specimens. A Reichert Jung Polycut was used to create 5 pm thick sections in the sagittal plane and stained with Von Kossa and Goldner's Trichrome. The slides were evaluated qualitatively for the presence and amount of mineralized tissue, fibrous tissue, and/or cartilage tissue at the osteotomy bridging point. Mineralized tissue at the osteotomy bridging point was assessed for callus maturity, estimated by lamellar pattern, cell type and overall porosity. Bridging of the osteotomy and cortical remodeling were recorded for each ulnae.

1264

A.L. Bertone et al. I Journal of Orthopaedic Research 22 (2004) 1261-1270

Data analysis

A two-factor analysis of variance (ANOVA), with group (factor 1) and treatment versus no treatment (factor 2) as the variables, was used to evaluate the effect of Ad-hBMP-6 treatment on osteotomy healing for objective data. When there was a significant interaction between factors, a Fischer's protected least squares difference post-test was conducted to identify significant comparisons. Chi-square analysis was used to compare frequency data. Differences were considered significant at p < 0.05.

2.5 I

I

6n 2J

Results Preliminary studies

Bone induction was observed in all rat quadriceps injected with 200 or 300 p1 of Ad-BMP-6 (Fig. 2 ) . Bone did not form in the quadriceps of any rat injected with 100 p1 of Ad-hBMP-6 (1 x 10l2particledml) or 200 p1 of Ad-GFP. Rabbit muscle injected with 2 x 10" adenoviral particles in 200 pl did not induce soft tissue mineralization based upon palpation and radiographs. The 200 p1 dose was selected for the rabbit osteotomy studies. Muscle cell expression of GFP and hBMP-6 persisted for 13 days at 1-2 fold over rabbit GAPDH expression (Fig. 3). Endogenous BMP-6 expression was not detected in Ad-GFP injected muscle sites. Expression of GFP and hBMP-6 was undetectable in uninjected triceps or lumbar muscle. Main study

All rabbits survived for the duration of the study and the osteotomy was well tolerated. Animals were freely ambulating by post-operative day 3. The total adenoviral vector dose ( 2 x 10" particles) delivered within the osteotomy was well tolerated. Minimal or no redness, swelling or lameness was noted following treatment.

TransgenelDay After Injection Fig. 3. Vector (bars) andor hBMP-6 (black) specific primers detected hBMP-6 or GFP expression for 13 days after injection into the lumbar and triceps muscle of rabbits. Endogenous BMP-6 expression was not detected in Ad-GFP injected sites.

Radiographic analysis

Rapid bone induction and a greater area of mineralized callus was radiographically apparent within two weeks after surgery (1 week after injection) in the Ad-BMP-6 injected osteotomies compared to the contralateral osteotomy control and the Ad-GFP treated osteotomies (Fig. 4). By 4 weeks the Ad-BMP-6 treated osteotomies appeared bridged with bone. At 6 weeks after surgery, the osteotomy line was no longer visible in most of the Ad-BMP-6 treated limbs. The osteotomies were not fully bridged in, and were not different between, the untreated and GFP treated limbs. The radiographic mineralized callus area stimulated by AdBMP-6 was maximal between weeks 3 and 5, and was greater than untreated osteotomies for each time point throughout the duration of the study (p < 0.0004 among groups; Fig. 5). As the study progressed, these differences between the Ad-BMP-6 treated limbs and the untreated osteotomy control limbs became smaller. Quantitative computed tomography

Fig. 2. Soft tissue mineralization at the origin (black arrow), insertion (white arrow) and injection site (open arrow) of rat quadriceps muscle in limbs injected 4 weeks previously with 200 pl containing 2 x 10" Ad-BMP-6 particles.

The mean mineralized pQCT cross-sectional callus area and total bone mineral content were significantly greater in the Ad-BMP-6 osteotomies at 6 weeks compared to their untreated contralateral osteotomies (169% [p < 0.00051 and 174% [p < 0.00011, respectively) and to Ad-GFP treated osteotomies (p < 0.01) (Table 2 ) . There was a tendency for Ad-GFP treated osteotomies to have less mean cross-sectional callus and bone mineralized content than contralateral control osteotomies (p < 0.07) (Table 2). At 8 weeks, there was no significant difference in mean cross-sectional callus or bone mineral content between the paired limbs in the Ad-BMP-6

A.L. Bertone et al. I Journul of Ortliopaedic Research 22 (2004) 1261-1270

6 Week

1265

8 Week

Fig. 4.A time course collage of healing osteotomies with the untreated osteotomy left and the Ad-BMP-6 treated osteotomy right. Greater osteotomy callus, healing, and bone remodeling is apparent in the Ad-BMP-6 treated ulnar osteotomies in this representative animal.

-*Ad-BMP Untreated Control Ad-GFP

-

0 1

2

I

3

I

I

I

4 5 6 Time (Weeks)

I

I

7

8

0

-

Ad-GFP Untreated Control

Fig. 5. Graph depicting the effect of adenoviral-mediated transfer of hBMP-6 or GFP gene on radiographic callus size as a function of time after surgery (differences among groups p < 0.0004). At each time point, groups with different letters are different.

treated rabbits 0) > 0.4). There was no significant difference in bone density within the callus. Torsional biomechanics

Maximum torque, torsional stiffness, and energy absorbed to failure were -2-fold greater in Ad-BMP-6 treated osteotomy ulnae compared to their respective untreated contralateral osteotomy ulnae at 6 weeks [p < 0.011, and -1.5-fold greater at 8 weeks Ir, < 0.031

(Table 3). Failure torque and torsional stiffness for AdBMP-6 treated ulnae were not different from reported historical data for intact ulnae [13] at both time points (Table 3). Fracture patterns post-torsional testing

All of the ulnae in the Ad-BMP-6 treated 6 and 8 week groups failed through the host bone outside the osteotomy during mechanical torsional testing (median

A.L. Bertone et al. I Journal of Orthopaedic Research 22 (2004) 1261-1270

1266

Table 2 Effect of Ad-hBMP-6 on properties of the mineralized callus, assessed by peripheral quantitative computed tomography (mean f SD) Adenoviral treated

Untreated

24.62 f 4.8*.** 16.39f 5.0 14.92 f 6.3

14.63f 3.8 16.36f5.1 18.97k 3.0

14.32 f 2.3:'" 9.50f 3.1 8.46 f 3.6

8.24f 2.9 9.72f3.1 10.86f 3.5

Adenoviral treated

Untreated

3 (3)* 3 (3) 1 (1-3)

l(1-3) 2 (1-3) 1 (1-3)

1 =original osteotomy; 2 = osteotomy and host bone; 3 =through host bone. * Significantly different than 6 week untreated and GFP-treated ulnar osteotomy (p < 0.03).

Mineral content (rng)

B M P - 6 6 weeks B M P - 6 8 weeks GFP-6 weeks

~

BMP-6-6 weeks BMP-6-8 weeks GFP-6 weeks

Area ( m d j

B M P - 6 6 weeks B M P - 6 8 weeks GFP-6 weeks

Table 4 Location of fractures following torsion testing [median (range)]

*Significantly greater than 6 or 8 week untreated ulnae (p < 0.0001). I* Significantly greater than ulnae treated with Ad-GFP (p < 0.01).

mode of failure=3; Table 4). All of the 6 week AdBMP-6 group untreated osteotomy ulnae failed through the original osteotomy. The 8 week Ad-BMP-6 group untreated osteotomy ulnae failed through a combination of the host bone and the osteotomy. The mode of failure for the Ad-GFP treated 6 week osteotomy ulnae was through the original osteotomy site (Table 4). Histologic evaluation

A robust and greater rate of healing was observed in 6 of 6 Ad-BMP-6 treated osteotomy ulnae at 6 weeks and 6 of 6 Ad-BMP-6 treated osteotomy ulnae at 8 weeks compared to 3 of 12 untreated osteotomy ulnae at 6 weeks and 1 of 6 Ad-GFP treated osteotomy ulnae at 6

weeks lp < 0.051. Complete bony bridging was observed in 6 of 6 Ad-BMP-6-treated osteotomy ulnae at both 6 and 8 weeks. In contrast the untreated contralateral osteotomy ulnae in the Ad-BMP-6 group were bridged with a combination of fibrous tissue and cartilage in 4 of 12 limbs. Similar incomplete or no bony bridging was observed for 5 of 6 Ad-GFP treated and 5 of 6 untreated osteotomy ulnae at 6 weeks. All Ad-BMP-6 treated osteotomy ulnae (12 of 12) had a more mature callus and cortical remodeling compared to their untreated osteotomy ulnae (n = 12) and the Ad-GFP treated (n = 6) and their contralateral untreated osteotomy ulnae (n = 6) (Fig. 6).

Discussion This study demonstrates that a single injection of the hBMP-6 gene in an adenoviral-mediated delivery system

Table 3 Effect of Ad-hBMP-6 on torsional biomechanics (mean f SD) Adenoviral treated

Untreated

Nm

(YOintact"

Nm

'YOintacta

0.63 +0.19*,** 0.67 f 0.19' 0.32 f 0.02

95.1 101.2 48.3***

0.29 f 0.23 0.43 f 0.16 0.34 f.0.12

43.8*** 65.0*** 5 I.@**

N mldeg 100

YOintact"

N mldeg 100

'%I

3.5 f 1.3*3** 3.8 f 1.6' 1.7f1.2

105.1 114.1 5 1.o'**

1 . 8 f 1.5 2.4 f 1.2 1.9f0.9

54.0"' 72.1**' 57.0"'

N m deg

% intacta

N m deg

% intact"

Failure torque

B M P - 6 6 weeks B M P - 6 8 weeks GFP-6 weeks

Torsional srixness (Nmldeg) BMP-6-6 weeks B M P - 6 8 weeks GFP-6 weeks

intact"

Energy absorbed to failure

B M P - 6 6 weeks 5.85 f 2.4',** 72.0 2.96 f 2.0 36.4"' 89.9 4.95 2.1 61.0"' B M P - 6 8 weeks 7.3 2.1' 40.6"' 4.69 ? 2.1 57.8"' GFP-6 weeks 3.3 f 2.0 Ad-BMP-6-treated osteotomies were greater in torque, stiffness and energy absorbed to failure than untreated or GFP-treated osteotomies. AdBMP-6 treatment resulted in a return of biomechanical properties to that of intact ulnae by 6 weeks. * Significantly greater than similar week contralateral untreated ulnar osteotomy (p range < 0.001-0.03). Significantly greater than ulnar osteotomy treated with Ad-GFP (p range < 0.003-0.02). *** Significantly lower than intact ulna values [I31 (p < 0.01). a Comparison to historical data [13].

*

..

*

A.L. Bertone et ul. I Journul of Orthopaedic Research 22 (2004) 1261-1270

AdBMP-6

Untreated

6 Week

8 Week

Fig. 6. Representative longitudinal histologic sections of ulnar osteotomy site at 6 weeks (top row) and 8 weeks (bottom row). Ulnar osteotomy treated with Ad-BMP-6 are shown on the left and untreated ulnar osteotomy are on the right. The sections were stained with Von Kossa for mineral (top row) and Goldner’s Trichrome (bottom row) and were photographed at 0.5 times magnification. Cartilage and fibrous callus (pink stain) are evident in the osteotomy of untreated ulnae. Remodeling bone is evident bridging the osteotomy in AdBMP-6 treated ulnae.

accelerated healing of rabbit ulnar mid-diaphyseal osteotomies. Progression of radiographic healing, maximum torque, torsional stiffness and energy to failure, as well as mineralized callus area, total mineral content and histologic assessment of osteotomy bony bridging and callus maturation were all greater in Ad-hBMP-6 treated osteotomies as compared to osteotomies without hBMP-6 gene. These results confirm not only a greater amount of callus of normal mineral density with AdBMP-6 administration, but an earlier and complete return of structural integrity of the ulna. Our pQCT data confirmed a greater amount of mineralizing callus of a density similar to untreated control callus with AdBMP-6 treatment which functionally produced a stronger ulna construct in torsion. These data in concert confirm acceleration of osteotomy repair in this model with Ad-BMP-6 treatment. Gene expression in nonimmunocompromized rabbit muscle was variable, but confirmed with our adenoviral preparations for at least 13 days. Gene expression most likely persisted longer than 13 days as expected for

1267

adenoviral gene delivery [6,23]. Endogenous BMP-6 expression was undetectable in untreated and Ad-GFP treated muscle. This duration of expression would appear to be adequate in this model to stimulate mineralization of callus when delivered locally to the site of bone repair. Presumably, BMP-6 is acting on cells that have migrated into the early fibrous callus in the osteotomy site [13]. Direct delivery of Ad-BMP-2 into the soft tissues [5] or spine [4] of immunocompromized animals has produced mineralization. However, even in immunocompromized animals, percutaneous injection of AdBMP-2 into soft tissues has failed to form bone [33]. This may be a dose effect or immunologic reaction to the adenoviral vector proteins [1,5,16,36]. Immunologic reaction may induce selective attrition of transfected cells or directly reduce local adenoviral titers at the injection site. The result in either case would be reduced transgene expression. Despite these findings, Ad-BMP-2 was administered in one report percutaneously at a fracture site and stimulated callus formation in immunocompetent rabbits [7]. Percutaneously injected retrovirus, an integrating virus with limited immune response by the host, has been able to promote bone healing in immunocompetent animals (BMP-4) [64,71]. The local environment of bone repair may be more sensitive to BMP augmentation than soft tissues resulting in mineralization even in the presence of a developing immune reaction. It is also conceivable that mineralization of soft tissues in response to BMPs may be a speciesspecific and dose related phenomenon that may be variable, particularly when gene delivery is used and local transgene concentrations are unknown. Gene delivery of BMPs by ex vivo cell transfection and subsequent cell injection has successfully formed bone in many sites and species of immunocompetent and immunoincompetent animals. Studies using bone marrow [19,20,63], muscle [40,71], or periosteal [14] derived mesenchymal stem cells transduced ex vivo with AdBMP-2 [40,45,63], retroviral-BMP-4 [71], retroviralBMP-7 [14], or Ad-BMP-9 [19,20] have been delivered by percutaneous injection into muscle [1,19,45], or sites of bone repair [20,40,70,71} and successfully formed bone in soft tissues [1,19,45,63,71] or healing bone [14,20,40,70,71].Use of viral delivery of transgenes into cells ex vivo and subsequent delivery of these cells reduces the load of viral protein at the site as compared to direct injection and likely reduces the immune response. This is most relevant to adenoviral vector use. Our study is the first report of successful direct percutaneous injection of adenoviral-BMP-6 in immunocompetent animals resulting in mineralizing callus at the site of bone repair. Rapid bone induction and greater callus was evident as early as 2 weeks after surgery (1 week after treatment) in the Ad-BMP-6 treated osteotomy ulnae compared to the contralateral untreated osteotomy ulnae and

1268

A.L. Bertone et al. I Journal of Orthopaedic Research 22 (2004) 1261-1270

Ad-GFP-treated osteotomy ulnae. By 8 weeks after surgery (7 weeks after Ad-BMP-6 treatment), the callus size remained significantly greater, but less so, than untreated osteotomy ulnae or Ad-GFP injected osteotomy ulnae (Fig. 5; Table 3). There was progressive reduction in absolute callus size and callus size relative to the untreated osteotomy ulnae in the Ad-BMP-6 treated group (Fig. 5). This progressive reduction in callus size most likely represents early remodeling of the callus and/or loss of BMP-6 gene expression with subsequent remodeling of the callus. The accelerated callus formation and maturation is consistent with the observation of increased torsion biomechanical properties in the Ad-BMP-6 treated osteotomy ulnae compared to contralateral untreated osteotomy ulnae and provides evidence that BMP-6 promotes the entire bone healing process, including bone formation and remodeling. The biomechanical properties of the Ad-BMP-6 treated osteotomy ulnae were equivalent to intact ulnae at least by 6 weeks after treatment, untreated osteotomy ulnae were <50% of intact strength at that time point. The ability of adenoviral-mediated delivery of hBMP6 gene to accelerate osteotomy healing is attributed to the presence of the hBMP-6 gene rather than the presence of the adenoviral preparation since local bone did not form in osteotomies treated with the same preparation containing a gene that expresses a biologically inert protein. BMP-6 gene expression is associated with bone formation in vivo in skeletal cartilage maturation, and has been associated with estrogen induced osteogenesis [60] and osteophyte formation [73]. The local bone formation is presumably due to the ability of BMP-6 to induce the differentiation of osteoprogenitor cells that would exist in the early vascular/fibrous callus at the time of osteotomy injection 1 week post-operatively. BMP-6 gene expression has been identified at sites of callus mineralization during spine fusion [51], and fracture healing [38] and is upregulated by LIM mineralization protein (named after protein domains Lin-1 1, Isl-1 and Mec-3), a factor that promotes callus mineralization [lo]. BMP-6 gene expression is colocalized with BMP-2 in skeletal mineralization and BMP-6 is a down stream nonessential regulator of endochondral ossification. In skeletal development, BMP-2 gene expression predominates [32,51,501. In soft tissues and cartilage, BMP-6 appears to promote direct mineralization [11,33], however, investigation on the role of BMP-6 function in osteotomy repair is not reported. Differences in the role or clinical impact of BMP-6 and BMP-2 in bone repair are not known. However, the ability of BMP-6 to accelerate bone healing in this model was similar to rhBMP-2 delivered in a collagen sponge [13] or as a-BSM calcium phosphate paste [42], tricalcium phosphate paste [35], and Gelfoam paste [43] in callus formation, histologic appearance, and return of biomechanical properties. In this study, rabbit ulnar

osteotomies achieved torsional biomechanical properties equivalent to intact bone within 6 weeks with our AdBMP-6 treatment. Similar results were reported by 4 weeks using rhBMP-2 protein delivery. A 4 week time point was not included in the Ad-BMP-6 study reported here. At 6 weeks, mineralized callus area was similar in the Ad-BMP-6-treated osteotomies (24.6 mm2) to rhBMP-2 delivered in a collagen sponge (24.3 mm2) [13]. Callus area was greater following delivery of rhBMP-2 delivered in a calcium phosphate paste [42]. However, although similar outcome was obtained, it is difficult to compare the results of the adenoviral delivered BMP to local protein delivery, since the local concentration of BMP-6 after gene delivery is not known. As a result, the relative potency of BMP-2 versus BMP-6 in this model cannot be speculated from these data. Peak callus formation occurred within 1-2 weeks of treatment in response to both Ad-BMP-6 or rhBMP-2 in the same rabbit osteotomy model [ 131. However, adenoviral BMP-6 induced the most rapid tissue calcification after subcutaneous injection in mice compared to Ad-BMP-2, Ad-BMP-7 and Ad-BMP-9 [33]. These results support potential increased efficacy of rhBMP-6 compared to other BMPs. In summary, a single injection of hBMP-6 gene consistently accelerated healing of ulnar osteotomies by rapid and extensive bone induction. The marked acceleration of healing with gene delivery provides strong rationale for evaluating the use of rhBMP-6 to accelerate fracture repair. A direct comparison between rhBMP-2 is warranted to identify relative potency and optimal treatment applications for each gene or protein.

Acknowledgements This work is partial fulfillment of a senior fellowship (NIHNIAMS AR08639-01) for A. Bertone. One or more of the authors received payments or other benefits or a commitment or agreement to provide such benefits from a commercial entity (Wyeth Research). The author’s thank C. Blake, A. Murphy, R. Rancourt, S . Orfao, C. Resmini, S. Rezankhah, J. Cooper, K. Backstrom, and J. Kamei for technical aspects of this study. Read in part at the Annual Meeting of the Orthopedic Research Society, Dallas, Texas, February 13, 2002.

References [l] Abe N, Lee YP, Sato M, et al. Enhancement of bone repair with helper-dependent adenoviral transfer of bone morphogenetic protein-2. Biochem Biophys Res Commun 2002;297:523-7. [2] Akamaru T, Suh D, Boden SD, et al. Simple carrier matrix modifications can enhance delivery of recombinant human bone

A.L. Bertone et al. I Journal of Orthopaedie Research 22 (2004) 1261-1270

morphogenetic protein-2 for posterolateral spine fusion. Spine 2003;28:429-34. [3] Alden TD, Beres EJ, Laurent JS, et al. The use of bone morphogenetic protein gene therapy in craniofacial bone repair. J Craniofac Surg 2000;11:24-30. [4] Alden TD, Pittman DD, Beres EJ, et al. Percutaneous spinal fusion using bone morphogenetic protein-2 gene therapy. J Neurosurg 1999;90S109-14. [5] Alden TD, Pittman DD, Hankins GR, et a]. In uiuo endochondral bone formation using a bone morphogenetic protein 2 adenoviral vector. Hum Gene Ther 1999;10:2245-53. [6] Baltzer AW, Latterman C, Whalen JD, et al. A gene therapy approach to accelerating bone healing: evaluation of gene expression in a New Zealand white rabbit model. Knee Surg Sports Traumatol Arthosc 1999;7(3):197-202. [7] Baltzar AW, Lattermann C, Whalen JD, et al. Genetic enhancement of fracture repair: healing of an experimental segmental defect by adenoviral transfer of the BMP-2 gene. Gene Ther 2000; 7:734-9. [8] Barnes GL, Kosstenuik PJ, Gerstenfeld LC, Einhorn TA. Growth factor regulation of fracture repair. J Bone Miner Res 1999;14: 1805-15. [9] Boden SD, Kang J, Sandhu H, Heller JG. Use of recombinant human bone morphogenetic protein-2 to achieve posterolateral lumbar spine fusion in humans: a prospective, randomized clinical pilot trial. Spine 2002;27:2662-73. [lo] Boden SD, Liu Y, Hair GA, et al. LMP-1, a LIM-domain protein, mediates BMP-6 effects on bone formation. Endocinology 1998; 13915125-34. [ I l l Bosky AL, Paschalis EP, Binderman I, Doty SB. BMP-6 accelerates both chondrogenesis and mineral maturation in differentiating chick limb-bud mesenchymal cell cultures. J Cell Biochem 2002;84:509-19. [I21 Bostrom MP. Expression of bone morphogenetic proteins in fracture healing. Clin Orthop 1998;355:11&23. [I31 Bouxsein ML, Turek TJ, Blake CA, et al. Recombinant human bone morphogenetic protein-2 accelerates healing in a rabbit ulnar osteotomy model. J Bone Joint Surg 2001;83-A:1219-30. [14] Breitbart AS, Grande DA, Mason JM, et al. Gene-enhanced tissue engineering: applications for bone healing using cultured periosteal cells transduced retrovirally with the BMP-7 gene. Ann Plast Surg 1999;42:488-95. [I61 Cao B, Mytinger JR, Huard J. Adenovirus mediated gene transfer to skeletal muscle. Microsc Res Tech 2002;58:45-51. [I71 Celeste AJ, Iannazzi JA, Taylor RC, et al. Identification of transforming growth factor p family members present in boneinductive protein purified from bovine bone. Proc Natl Acad Sci USA 1990;87:9843-7. [ 181 Cheng SL, Lou J, Wright NM, et al. In uitro and in viuo induction of bone formation using a recombinant adenoviral vector carrying the human BMP-2 gene. Calcif Tissue Int 2001;68:87-94. [I91 Dayoub H, Dumont RJ, Li JZ, et al. Human mesenchymal stem cells transduced with recombinant bone morphogenetic protein-9 adenovirus promote osteogenesis in rodents. Tissue Eng 2003; 91347-56. [20] Dumont RJ, Dayoub H, Li JZ, et al. Ex uiuo bone morphogenetic protein-9 gene therapy using human mesenchymal stem cells induces spinal fusion in rodents. Neurosurgery 2002;51:1 2 3 9 4 . [21] Evans CH, Robbins PD. Potential treatment of osteoarthritis by gene therapy. Rheum Dis Clin North Am 1999;25:3334. [22] Friedlaender GE, Perry CR, Cole JD, et al. Osteogenic protein-1 (bone morphogenetic protein-7) in the treatment of tibial nonunions. J Bone Joint Surg 2001;83-A:S151-8. [23] Frisbie DD, Ghivizzani SC, Robbins PD, et al. Treatment of experimental equine osteoarthritis by in vivo delivery of the equine interleukin-I receptor antagonist gene. Gene Ther 2002;9(1):1220.

1269

[24] Gazit D, Turgman G, Kelley P, et al. Engineered pluripotent mesenchymal cells integrate and differentiate in regenerating bone: a novel cell-mediated gene therapy. J Gene Med 1999;1:121-33. [25] Ghivizzani SC, Oligino TJ, Glorioso JC, et al. Direct gene delivery strategies for the treatment of rheumatoid arthritis. Drug Disc Today 2001;6:259-67. [26] Gonda K, Nakaoka T, Yoshimura K, et a]. Heterotopic ossification of degenerating rat skeletal muscle induced by adenovirusmediated transfer of bone morphogenetic protein-2 gene. J Bone Miner Res 2000; 15:105tL65. [27] Govender S, Csimma C, Genant HK. BMP-2 evaluation in surgery for Tibia1 Trauma (BESTT) study group. Recombinant human bone morphogenetic protein-2 for treatment of open tibial fractures: a prospective, controlled, randomized study of four hundred and fifty patients. J Bone Joint Surg 2002;84-A2123-34. [30] Helm GA, Alden TD, Beres EJ, et al. Use of bone morphogenetic protein-9 gene therapy to induce spinal arthrodesis in the rodent. J Neurosurg 2000;92:1914. [31] Hughes FJ, Collyer J, Stanfield M, Goodman SA. The effects of bone morphogenetic protein-2, -4, and -6 on differentiation of rat osteoblast cells in vitro. Endocrinology 1995;136:2671-7. [32] Iwasaki M, Le AX, Helms JA. Expression of Indian hedgehog, bone morphogenic protein 6 and gli during skeletal morphogenesis. Mech Dev 1997;69:197-202. [33] Jane Jr JA, Dunford BA, Kron A, et a]. Ectopic osteogenesis using adenoviral bone morphogenetic protein (BMP)-4 and BMP-6 gene transfer. Mol Ther 2002;6:464-70. [34] Johnsson R, Stromqvist B, Aspenberg P. Randomized radiostereometric study comparing osteogenic protein- 1 (BMP-7) and autograft bone in human noninstrumented posterolateral lumbar fusion. Spine 2002;27:265461. [35] Kim HD, Li R, DAugusta DA, et a]. Injectable hyaluronic acid tricalcium phosphate composites for the delivery of rhBMP-2. In: Controlled Release of Bioactive Materials 27th International Symposium 2000, Paris, France. p. 8009. [36] Kim HS, Viggeswarapu M, Boden SD, et al. Overcoming the immune response to permit ex vivo gene therapy for spine fusion with human type 5 adenoviral delivery of the LIM mineralization protein-1 cDNA. Spine 2003;28:219-26. [37] Laurencin CT, Attawia MA, Lu LQ, et al. Poly(1actide-cogyco1ide)lhydroxyapatite delivery of BMP-2-producing cells: a regional gene therapy approach to bone regeneration. Biomaterials 2001;22:1271-7. 1381 Le AX, Miclau T, Hu D, Helms JA. Molecular aspects of healing in stabilized and non-stabilized fractures. J Orthop Res 2001; 19:78-84. f39] Lee JY, Musgrave D, Pelinkovic D, et al. Effect of bone morphogenetic protein-2-expressing muscle-derived cells on healing of critical-sized bone defects in mice. J Bone Joint Surg [Am] 200 1;83-A:1032-9. [40] Lee JY, Peng H, Usas A, et al. Enhancement of bone healing based on ex vivo gene therapy using human muscle-derived cells expressing bone morphogenetic protein 2. Hum Gene Ther 2002; 13:1201-11. [41] Li G, Bouxsein ML, Luppen C, et al. Bone consolidation is enhanced by rhBMP-2 in a rabbit model of distraction osteogenesis. J Orthop Res 2002;20:779-88. [42] Li RH, Bouxsein ML, Blake CC, et al. rhBMP-2 injected in a calcium phosphate paste (a-BSM) accelerates healing in the rabbit ulnar osteotomy model. J Orthop Res 2003;21:997-1004. [43] Li RH, Seeherman HJ, Bouxsein M, et a]. Acceleration of fracture healing with rhBMP-2 delivered in an injectable gelfoam delivery system. In: Society of Biomaterials 6th World Biomaterial Congress Transactions 2000. p. 560.1. [44] Lieberman JR, Ghivizzani SC, Evans CH. Gene transfer approaches to the healing of bone and cartilage. Mol Ther 2002; 6: 141-7.

1270

A.L. Bertone et al. I Journal of Orthopedic Research 22 (2004) 1261-1270

[45] Lieberman JR, Le LQ, Wu L, et al. Regional gene therapy with a BMP2-producing murine stromal cell line induces heterotopic and orthotopic bone formation in rodents. J Orthop Res 1998;16: 330-9. [46] Lou J, Xu F, Merkel K, Manske P. Gene therapy: adenovirusmediated human bone morphogenetic protein-2 gene transfer induces mesenchymal progenitor cell proliferation and differentiation in uitro and bone formation in uiuo. J Orthop Res 1999; 17:43-50. I471 Luppen CA, Blake CA, Ammirati KM, et al. Recombinant human bone morphogenetic protein-2 enhances osteotomy healing in glucocorticoid-treated rabbits. J Bone Miner Res 2002;17:30110. [48] Maniscalco P, Gambera D, Bertone C, et al. Healing of fresh tibia1 fractures with OP-1. A preliminary report. Acta Biomed Ateneo Parmense 2002;73:27-33. [50] Martinez G, Mishina Y, Bertram JF. BMPs and BMP receptors in mouse metanephric development: in vivo and in vitro studies. Int J Dev Biol 2002;46:525-33. [51] Minamide A, Boden SD, Viggeswarapu M, et al. Mechanism of bone formation with gene transfer of the cDNA encoding for the intracellular protein LMP-1. J Bone Joint Surg [Am] 2003;85: 1030-9. [52] Minina E, Wenzel HM, Kreschel C, et al. BMP and IhMPTHrP signaling interact to coordinate chondrocyte proliferation and differentiation. Development 2001; 128:4523-34. [53] Moustsatsos IK, Turgeman G, Zhou S, et al. Exogenously regulated stem cell-mediated gene therapy for bone regeneration. Mol Ther 2001;3:449-61. [54] Musgrave DS, Bosch P, Ghivizzani S, et al. Adenovirus-mediated direct therapy with bone morphogenetic protein-2 produces bone. Bone 1999;24:541-7. [55] Oakes DA, Lieberman JR. Osteoinductive applications of regional gene therapy: ex uiuo gene transfer. Clin Orthop 2000;379Sl0112. [56] Okubo Y, Bessho K, Fujimura K, et al. Osteoinduction by bone morphogenetic protein-2 via an adenoviral vector under transient immunosuppression. Biochem Biophys Res Commun 2000;267: 382-7. [57] Okubo Y, Bessho K, Fujimura K, et al. In uitro and in uiuo studies of a bone morphogenetic protein-2 expressing adenoviral vector, J Bone Joint Surg [Am] 2001;83-A:S99-104. [%I Olmstead EA, Blum JS, Rill D, et al. Adenovirus-mediated BMP2 expression in human bone marrow stromal cells. J Cell Biochem 2001;82:11-21. [59] Pathi S, Rutenberg JB, Johnson RL, Vortkamp A. Interaction of Ihh and BMPlnoggin signaling during cartilage differentiation. Dev Biol 1999;209:239-53.

[60] Plant A, Tobias JH. Increased bone morphogenetic protein-6 expression in mouse long bones after estrogen administration. J Bone Miner Res 2002;17:782-90. [61] Pluhar GE, Manley PA, Heiner JP Jr RV, et al. The effect of recombinant human bone morphogenetic protein-2 on femoral reconstruction with an intercalary allograft in a dog model. J Orthop Res 2001;19:308-17. [62] Praemer A, Furner S, Rice DP. Musculoskeletal conditions in the US. Rosemont: American Academy of Orthopedic Surgeons; 1999. [63] Riew KD, Lou J, Wright NM, et al. Thoracoscopic intradiscal spine fusion using a minimally invasive gene-therapy technique. J Bone Joint Surg 2003;85:866-71. [64] Rundle CH, Miyakoshi N, Kasukawa Y, et al. In vivo bone formation in fracture repair induced by direct retroviral-based gene therapy with bone morphogenetic protein-4. Bone 2003;32: 591-601. [65] Scaduto AA, Lieberman JR. Gene therapy for osteoinduction. Orthop Clin North Am 1999;30:625-33. [66] Sheehan JP, Sheehan JM, Seeherman H, et al. The safety and utility of recombinant human morphogenetic protein-2 for cranial procedures in a nonhuman primate model. J Neurosurg 2003;98: 125-30. [67] Shuler FD, Georgescu HI, Niyibizi C, et al. Increased matrix synthesis following adenoviral transfer of a transforming growth factor PI gene into articular chondrocytes. J Orthop Res 2000;18: 585-92. [68] Solloway MJ, Dudley AT, Bikoff EK, et al. Mice lacking BMP6 function. Dev Genet 1998;22:321-39. [69] Turgeman G, Pittman DD, Muller R, et al. Engineered human mesenchymal stem cells: a novel platform for skeletal cell mediated gene therapy. J Gene Med 2001;3:240-51. [70] Wang JC, Kanim LEA, Yo0 S, et al. Effect of regional gene therapy with bone morphogenetic protein-2-producing bone marrow cells on spinal fusion in rats. J Bone Joint Surg 2003;85: 905-1 1. [71] Wright V, Peng H, Usas A, et al. BMP4-expressing muscle-derived stem cells differentiate into osteogenic lineage and improve bone healing in immunocompetent mice. Mol Ther 2002;6:169-78. [72] Zabka AG, Pluhar GE, Edwards 3rd RB, et al. Histomorphometric description of allograft bone remodeling and union in a canine segmental femoral defect model: a comparison of rhBMP2, cancellous bone graft, and absorbable collagen sponge. J Orthop Res 2001;19:318-27. [73] Zoricic S, Mark I, Bobinac D, Vukicevic S. Expression of bone morphogenetic proteins and cartilage-derived morphogenetic proteins during osteophyte formation in humans. J Anat 2003; 202:269-77.