Homogeneous osteogenesis and bone regeneration by demineralized bone matrix loading with collagen-targeting bone morphogenetic protein-2

ARTICLE IN PRESS

0142-9612/$ - se

doi:10.1016/j.bi

�CorrespondE-mail addr

1These autho

Biomaterials 28 (2007) 1027–1035

www.elsevier.com/locate/biomaterials

Homogeneous osteogenesis and bone regeneration bydemineralized bone matrix loading with collagen-targeting bone

morphogenetic protein-2

Bing Chena,1, Hang Lina,1, Jianhua Wangb,1, Yannan Zhaoa, Bin Wanga,Wenxue Zhaoa, Wenjie Suna, Jianwu Daia,�

aKey Laboratory of Molecular and Developmental Biology, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, 3 Nanyitiao,

Zhongguancun, Beijing 100080, ChinabStomatological Hospital, Shandong University, Jinan 250012, China

Received 27 July 2006; accepted 9 October 2006

Available online 13 November 2006

Abstract

Considerable research has been focused on the development of bone morphogenetic protein-2 (BMP-2) delivery system for

homologous and efficient bone regeneration. The aim of the present study was to develop a collagen-based targeting bone repair system.

A collagen-binding domain (CBD) was added to the N-terminal of native BMP-2 to allow it bind to collagen specifically. We showed that

the collagen-binding bone morphogenetic protein-2 (named bone morphogenetic protein2-h, BMP2-h) had maintained the full biological

activity as compared to rhBMP2 lacking the CBD. In vitro functional study also demonstrated that collagen matrix could maintain

higher bioactivity of BMP2-h than native BMP-2. When demineralized bone matrix (DBM) impregnated with BMP2-h was implanted

subcutaneously in rats, homogeneous bone formation was observed. Moreover, in a rabbit mandible defect model, surgical implantation

of collagen matrix loaded with BMP2-h exhibited remarkable osteoinductive properties and excellent homogeneous bone formation. Our

studies suggested that this novel collagen-based BMP-2 targeting bone repair system induced better bone formation not only in quantity

but also in quality. Similar approaches may also be used for the repair of other tissue injuries.

r 2006 Elsevier Ltd. All rights reserved.

Keywords: Bone morphogenetic protein; Collagen; Bone repair; Ectopic bone formation; Rabbit mandible defect

1. Introduction

Segmental bone loss or non-union results in the vastdemand for new bone to replace and restore the function ofthe lost bone [1]. Repair of bone defects is one of the majortherapeutic goals in various clinical fields. Traditionally,autologous bone grafts have been considered as the ‘‘goldstandard’’ [2]. However the major drawbacks of thismethod are donor site morbidity and donor availabilities[3]. Recent progress in regenerative medicine raises thehope of repairing bone defects with the combination ofbiomaterials and growth factors [4,5].

e front matter r 2006 Elsevier Ltd. All rights reserved.

omaterials.2006.10.013

ing author. Tel./fax: +86 010 82614426.

ess: [email protected] (J. Dai).

rs contributed equally to this work.

Osteogenesis, osteoinduction, and osteoconduction arethree essential elements of bone regeneration along withthe final bonding between host bone and grafting materialwhich is called osteointegration [6]. Many types of bonefilling materials have been developed and have playedcritical roles in bone repair [7–10]. Collagen is one of themost widely used bone-filling biomaterial in present bonetissue engineering [11–13]. However studies have shown thefunctions of collagen-based biomaterials alone are limitedin bone repair. It cannot stimulate bone formation incritical sized defects by itself. Therefore, growth factorssuch as TGF-b [14,15], bFGF [16], and bone morphoge-netic proteins (BMPs) [17,18] are often employed topromote bone formation. Among them, BMP-2 is a potentbone stimulator and it plays key roles in many steps duringbone morphogenesis [17,19]. After bone trauma, BMP-2

ARTICLE IN PRESSB. Chen et al. / Biomaterials 28 (2007) 1027–10351028

accelerates the ossification of extensive bone lesions [20,21].However, in practice, the therapeutic concentrations ofBMP-2 are difficult to be maintained at wound sites due toits rapid diffusion by body fluid. Periodic addition ofBMP-2 often requires invasive procedures such as injectionor infusion, which could be painful and troublesome [22].The large amount and multiple administrations of BMP-2may be clinically impractical and expensive as well. Inaddition, its excessive doses, especially to the normaltissues, may have undesirable systemic side effects.

A variety of carriers have been developed to restrict thefree diffusion of BMP-2, and they often provide a slowreleasing system [23]. Our goal here is to construct acollagen-based BMP-2 targeting repair system, which willattach BMP-2 to the scaffolds not only to reduce theloading dose of BMP-2 but also to achieve a homogeneousrepair in the whole defect site. In this work, a collagen-binding domain (CBD) of ‘‘TKKTLRT’’ [24] was added tothe N-terminal of recombinant human BMP-2 (namedrhBMP2-h) for its binding specially to collagen (Fig. 1A).We hypothesized more rhBMP2-h was maintained incollagen matrix, like demineralized bone matrix (DBM)than native one when washed by body fluids (Fig. 1B).

Fig. 1. The sketch map of the growth factor/scaffold system. (A)

Schematic drawing of the genetically engineered rhBMP2-h expression

constructs. The expressed protein contains a 6�histidine purification tag,

a protease site, an auxiliary collagen binding domain, and the 114-amino-

acid sequence of the mature human BMP-2 protein. (B) The schematic

drawing explaining the hypothesis of the collagen-based BMP-2 targeting

bone repair system. It could maintain higher BMP-2 concentration when

implanted than the traditional BMP-2/scaffolds.

We hoped that this system could promote better boneformation in a homogeneous manner.

2. Materials and methods

2.1. The processing of DBM

DBM scaffold was made from the spongy bone of bovine as described,

which was mainly composed with collagen [25]. Briefly, the spongy bones

were separated from the head of a long bone and cut into appropriate size.

Then the samples were soaked in acetone for 48 h to remove the fatty

composition. Subsequently, the 0.6M HCl was utilized to demineralize the

spongy bones followed ddH2O washing completely, enzyme treated and

ddH2O washing again, then freeze-dried. The sample’s surface character

was visualized with a HITACHI Model S-2500 SEM.

2.2. Production of genetically engineered BMP-2

The fusion protein rhBMP2-h (Fig. 1A) consists of a 6�His

purification tag, a collagen binding domain (TKKTLRT) and the mature

human BMP-2 fragment, the molecular weight of its monomer form is

17.3 kd. rhBMP-2 with no CBD (rhBMP-2) contains 6�His tag and

mature human BMP-2 fragment, the molecular weight of its monomer

form is 15.2 kd. Vector constructions and inclusion body preparation were

performed as described with slightly modification [11]. The refolding

protocol of inclusion body was the same as the Refs. [26,27].

2.3. The biological activity of the BMP-2 and BMP2-h

Mouse C2C12 cells were inoculated at a density of 1� 104 cells/well in

48-well plates and maintained in DMEM-F12 (Biofluids, USA) containing

10% fetal bovine serum (FBS, Gibco, USA) at 37 1C in a humidified

atmosphere of 5% CO2 in air for 24 h. They were stimulated for 3 days

with 10–320 nM of rhBMP2-h or rhBMP-2 variant in 400ml of DMEM-

F12 containing 0.5% FBS. After 3 days stimulation, cells were washed

with PBS and lysed with 0.1% TritonX-100/PBS and repeatedly frozen/

thawed for three times to disrupt the cell membranes. Alkaline

phosphatase (ALP) activities were determined by an established technique

using p-nitrophenyl-phosphate as the substrate [22]. Protein concentra-

tions were determined using Bradford method [28]. A commercially

available CHO cell produced rhBMP-2 was used as a standard for this test

(Sigma, USA).

2.4. Collagen-binding assay

Acid soluble Type I collagen prepared from rat tail tendon as described

previously was added to 48-well plates and air-dried in a fume hood

overnight [29,30]. The proteins solutions with increasing quantity from 0.5

to 4mM were added to those 48-well plates and incubated at 37 1C for 1 h.

Then all the wells were extensively washed with PBS, and the remaining

proteins were measured by a modified ELISA assay [22]. Briefly, the wells

were blocked with 5% BSA/PBS for 2 h on a platform shaker followed by

the addition of 200ml (1:1000 dilution) anti-polyHistidine antibody

(Sigma, USA) and incubation at room temperature for 2 h. The unbound

primary antibodies were removed by three washes with PBS. ALP-

conjugated goat-anti-mouse IgG (1:10000 dilution, Sigma, USA) was used

as a secondary antibody. The ALP reaction product was developed by

incubation with para-nitrophenylphosphate (pNPP) (Sigma, USA) for

10min at room temperature, and the results were quantified at 405 nM

using plate reader (TECAN, SUNRISE, Austria).

2.5. The biological activities of BMP-2 on collagen

The acid soluble Type I collagen coated 48-well plates were prepared

as previously. The proteins solutions with increasing concentration from

ARTICLE IN PRESSB. Chen et al. / Biomaterials 28 (2007) 1027–1035 1029

200 to 800 nM were added to those 48-well plates and incubated

at 37 1C for 2 h followed by extensively washing with PBS. Then

mouse C2C12 cells were inoculated at a density of 2� 104 cells/well

and maintained in DMEM-F12 (Biofluids, USA) containing 0.5%

FBS (Gibco, USA) at 37 1C in a humidified atmosphere of 5%

CO2 in air for 3 days. The ALP activities were determined as previously

described.

2.6. In vivo ectopic bone formation

Six Male Wistar rats (200710 g) were used in this study. The animals

were housed at 24 1C with a 12-h light/12-h dark cycle and were allowed

free access to water and a commercial diet (IGDB Laboratory Rodent

Chow; Beijing, China). Surgery was performed under conditions of

general anesthesia [intravenous injection with pentobarbitol sodium]. The

left and right dorsal regions of each rat were shaved and disinfected, and

then the skin was incised. Each rat has four implant sites. The implants

were inserted subcutaneously randomly and the surgical cut was then

closed by suturing. Eight nmol rhBMP-2 or rhBMP2-h were loaded on

each scaffold (8mm� 5mm� 4mm) (n ¼ 8, respectively). DBM loaded

only with PBS were used as negative controls (n ¼ 8). Four weeks later

the host rats were sacrificed and the implants were retrieved together

with a minimum quantity of surrounding tissue by sharp dissection. After

being fixed in 4% neutral buffered formaldehyde solution for 2 days,

the implants were dried through gradient ethanol baths and in xylene,

then embedded in paraffin and sectioned into 5mm thick slices. The

sections were stained with Haematoxylin and Eosin (H&E). And the

histomorphometrical evaluation was performed with Image-Pro systems

(Media Cybernetics. MD, USA). The width of one edge accounted for 1/4

of the sample length and 5 slides per sample randomly chose were

examined. The sections were blindly presented for measurements by one

examiner.

2.7. Rabbit mandible defect repair

A total of 16 mature White New Zealand rabbits (2.9–3.3 kg) were

used. All rabbits were randomly divided into four groups. Prior to

operation, the rabbits were anesthetized by intravenous injection with

pentobarbitol sodium (30mg/kg). Then the bone defects were created on

both mandibles with size 12mm� 5mm� 4mm, which was so-called

critical size defect. In the negative control groups, the wounds were

directly closed with sutures. Scaffolds loaded with PBS, rhBMP-2 or

rhBMP2-h, respectively were the three experimental groups (0.5 nmol

proteins/mg DBM). After the scaffolds were implanted into the

mandibular defects separately, the wounds were closed with sutures. All

rabbits were sacrificed after 4 or 12 weeks and the mandibles were

harvested intact. The mandibles underwent X-ray (SIDEXIS XG,

Germany) and histological analysis. After being fixed in 4% neutral

buffered formaldehyde solution for 2 days, the implants were dried

through gradient ethanol baths and in xylene, then embedded in paraffin

and sectioned into 5 mm thick slices. The sections were stained with

Haematoxylin and Eosin (H&E). The overall of the slides were put

together with different eyeshots. And the quantity of new-formed bone in

the defect areas was analyzed with Image-Pro systems (Media Cyber-

netics. MD, USA).

Fig. 2. The structure of DBM. (A) DBM from fetal bovine possessed a

porous structure. (B) The SEM of the DBM. The pore size of the scaffold

was measured as visualized by scanning electron microscopy. Scale

bar ¼ 10mm (A) and 1mm (B), respectively.

2.8. Statistical analysis

Data were presented as means7standard deviation. Differences of

bone formation between the various groups in the rabbit mandible defect

repair were analyzed using the one-way ANOVA test, and the others a

paired method was used. Statistically significant values were defined as

*Po0.05 and **Po0.01.

3. Results

3.1. The properties of the DBM

The DBM prepared from bovine possessed a porousstructure (Fig. 2A). A micrograph of the scanning electronrevealed the porous structure of the DBM (Fig. 2B). Theaverage pore size was 400–600 mm, which would permitcellular ingrowth and facilitate the diffusion of nutrientsinto the matrix.

3.2. The biological activities of BMP-2 and BMP2-h were

similar

The biological activities of the purified and refoldedrhBMP2-h and rhBMP-2 were tested by the ALP activityassay using the C2C12 cell line. As shown in Fig. 3A–C, inthe presence of rhBMP-2 and rhBMP2-h, cultured cellsexhibited rounded and polygonal osteoblast-like appear-ance compared to the untreated cells, which were thin, andspindle shaped. After incubating for 3 days with rhBMP2-hor rhBMP-2 over a range of 10–320 nM, ALP activitiesof mouse C2C12 cell line were increased significantly ina dose-dependent manner (Fig. 3D). The activities ofrhBMP2-h and rhBMP-2 were similar to that of rhBMP-2 produced by CHO cells (Sigma, USA), which wasserved as a positive control.

3.3. rhBMP2-h bound to collagen specifically

The collagen-binding abilities of rhBMP2-h and rhBMP-2 were tested in vitro by the gradient-bindingexperiment. As shown in Fig. 4A, the retained proteinson collagen increased in a concentration-dependent fash-ion, and the binding curve of rhBMP-2 and rhBMP2-hwere similar in shape. However, more rhBMP2-h thanrhBMP-2 was retained on collagen after washing. At 4 mM,both proteins reached saturation. Based on the bindingcurve, the Kd values of rhBMP2-h and rhBMP-2 tocollagen were calculated as 0.272 and 2.11 mM respectivelyin this test system.

ARTICLE IN PRESSB. Chen et al. / Biomaterials 28 (2007) 1027–10351030

3.4. Collagen matrix maintained a higher activity of

rhBMP2-h in vitro

The biological function of the two proteins when boundto collagen was tested in vitro by culturing C2C12 cells on

Fig. 3. The biological activity of rhBMP2-h and rhBMP-2 in vitro. (A)

The C2C12 cells cultured without BMP-2. The light micrographic

observation of them exhibited the characteristic appearance of C2C12

cells, which were thin and elongated with spindle-like shape. (B) The

C2C12 cells cultured in the presence of rhBMP-2 (280 nm). (C) The C2C12

cells cultured in the presence of rhBMP2-h (280 nm). The light

micrographic observation of the C2C12 cells cultured in the presence of

rhBMP-2 and rhBMP2-h (280 nm) exhibited rounded and polygonal

osteoblast-like appearance. (D) Dose-dependent induction of ALP activity

in C2C12 cells by variant BMP-2. The application of rhBMP2-h and

rhBMP-2 to mouse C2C12 cell line significantly increased cell alkaline

phosphatase (ALP) activity when the cells were incubated for 3d with the

proteins over a range of 10–320 nm. And a dose-dependent effect was

observed. As positive control, commercially available CHO-produced

rhBMP-2 was used. Data were presented as the mean7S.D. (n ¼ 6).

0.12rhBMP-2

0.25

0.15

0.05

0.03 0.06bound

bo

un

d/f

ree

0.09 0.12 0.15

0.2

0.1

00

rhBMP 2-h

Protein concentration (μμM)

0.1

0.08

0.06

ΔOD

405

0.04

0.02

(A) (B

0

0 1 2 3 4 5

Fig. 4. The collagen binding ability and biological functional study of rhBM

rhBMP2-h and rhBMP-2 in vitro. The retained proteins on collagen increased

and rhBMP2-h were similar in shape. However, more rhBMP2-h than rhBMP

cultured on variant BMP-2 loaded collagen. After 3 days culture, the ALP activ

of rhBMP-2 loaded collagen at the same concentration. At 400 nM, there was s

mean7S.D. (n ¼ 6).

rhBMP-2 or rhBMP2-h loaded collagen gel. After 3 daysculture, the ALP activity of C2C12 cultured on rhBMP2-hloaded collagen was higher than that of rhBMP-2 loadedcollagen at the same concentration. At 400 nM, there wassignificant difference between them (**Po0.01, *Po0.05)(Fig. 4B).

3.5. Homogeneous ectopic bone formation by rhBMP2-h

loaded DBM

Four weeks after implantation, no obvious bone orcartilage formation was observed in the PBS/DBM asexpected (Fig. 5A and B). Fig. 5D, E, H and I reflected thenew bone formations at the edge of rhBMP-2/DBM andrhBMP2-h/DBM implants respectively. The bony tissuegrew along the girders of the DBM, and in these areas therewas no statistical difference between rhBMP-2/DBM andrhBMP2-h/DBM according to the histomorphometricalanalysis (Fig. 5C). In contrast, in the central area,rhBMP2-h/DBM displayed much more bone formationcompared to that of rhBMP-2/DBM. Numerous lamellabones and bone marrows were observed in the section ofrhBMP2-h/DBM implants (Fig. 5J and K), while in therhBMP-2/DBM, there was only some small and scatteredbone formation (Fig. 5F and G). The bone formationbetween the edge and central of the rhBMP-2/DBM hadstatistical difference, while the rhBMP2-h/DBM had nodifference (Fig. 5C).

3.6. Homogeneous rabbit mandible defect repair by

rhBMP2-h loaded DBM

The rabbit mandibular defect model was used and the‘‘critical size’’ was 12mm� 5mm� 4mm (Fig. 6A).Representative X-ray photographs of bone defect sites atthe 4th and 12th weeks after surgery were shown in Fig. 6B.At the 4th week, no obvious calcified shadows were seen in

1.8

0.8

0.60.40.2

1.6 rhBMP-2rhBMP 2-h1.4

1.2

AL

P a

ctiv

ity

(nm

ol/m

in/m

g p

rote

in)

1

00 200 400

Protein concentration (nM)

800

)

P-2 and rhBMP2-h on collagen gel. (A) The collagen-binding ability of

in a concentration dependent fashion, and the binding curve of rhBMP-2

-2 was retained on collagen after washing. (B) ALP activity in C2C12 cells

ity of C2C12 cultured on rhBMP2-h loaded collagen was higher than that

ignificant difference between them (**Po0.01). Data were presented as the

ARTICLE IN PRESS

Fig. 5. The ectopic bone formation of different implants. (A, B, D–K) The histological images of DBM retrieved from the 4-week implanted rats. (A, B)

DBM with PBS. (D–G) DBM with rhBMP-2. (H–K) DBM with rhBMP2-h. (D, F, H, J) The edge of these implants. (E, G, I, K) The center of these

implants. (B, E, G, I, K) (original magnification � 100) were higher magnification of (A, D, F, H, J) (original magnification � 40), respectively. The arrow

head showed the new formed bone. (C) The histomorphometrical analysis of the bone formation (*Po0.05).

B. Chen et al. / Biomaterials 28 (2007) 1027–1035 1031

the defect sites of the negative control group. In the othergraft groups, calcified tissues grew from the host bones;however the increases were different in these groups.Fig. 6C presented the statistical analysis of the healedareas among different groups. The rhBMP2-h/DBM andrhBMP-2/DBM groups showed an increasing in healingpercentage of nearly 80% and 60% respectively, whereasthe PBS/DBM group showed only less than 30%. Theseresults showed the scaffold alone was limited in repairingthe defect. At the 12th week, calcification became moreevident, and newly formed bone occupied essentially theentire area of the defect sites (498%) in the rhBMP2-h/DBM group. About 90% and 60% defect areas were filledby new bones in the rhBMP-2/DBM and PBS/DBMgroups, respectively (Fig. 6C).

The photomicrographs of the histological slices of the4-week implants were shown in Fig. 7A and 8. In therhBMP2-h/DBM group, the newly formed bone grew atthe host and bone-graft interface and extended toward itscenter along the scaffold. In the same time, there wassimultaneous scaffold remodeling by osteocyte invasionand new bone formation. At the center of the tissues, somenewly formed bone had been observed and was integratedwith the scaffolds. The blood vessels were also largelyformed accompanied with the bone ingrowth. Although inthe rhBMP-2/DBM group, the bone formation was thickerand more mature than in the PBS/DBM group at the edge,the centers of both groups were still filled by blood clots,and no obvious bone formations were observed around thescaffold. The results were consistent with the X-ray studies.The quantity of the newly formed bone in the defect sites

(showed in the white broken line in Fig. 7A) was calculatedby histomorphometrical analysis (Fig. 7B). At the 4thweek, the quantities of new formed bone among PBS/DBM, rhBMP-2/DBM and rhBMP2-h/DBM groups were13.672.7%, 21.674.0% and 36.272.2%, respectively.At the 12th week, the newly formed bones at the edge

were remodeled and were histological similar in three graftgroups. For the groups of PBS/DBM and rhBMP-2/DBM,bone defect still remained in the centers, and new bonesonly formed at the peripheral area. The scaffolds in thecenter of these two groups had degraded and covered withfibrillar configuration. Despite some bone ingrowth fromthe interface region, no sufficient bone formation was seenin central area. While in the rhBMP2-h/DBM group therewas much more bone formation with better quality. Thehomogenous bone tissue appeared with excellent bloodsupply and little fibrous tissue infiltration. The defect sitehad been completely filled with new bone, and there waslarge quantity of mature bone marrow with adipose cellsand blood cells in this group. The quantities of newlyformed bone were 18.774.6% (PBS/DBM), 29.474.6%(rhBMP-2/DBM) and 45.176.7% (rhBMP2-h/DBM),respectively. The difference between each two groups wasalso statistically significant.

4. Discussion

The proliferation and differentiation of sufficient pro-genitor cells are critical for bone regeneration, and theseprocesses are regulated by growth factors such as TGF-band BMPs [31,32]. BMP-2 can efficiently induce bone

ARTICLE IN PRESS

Fig. 6. X-ray photos in the repair of the rabbit mandible defect. (A) Rabbit mandible defect model. (B) Representative X-ray photographs of bone defects

at the 4th and 12th week after surgery. (C) Histomorphometric analysis of the healed areas (**Po0.01, *Po0.05). Data were presented as the mean7S.D.

Negative control: the wounds were directly closed with sutures. DBM: implants of DBM loaded only with PBS. rhBMP-2/DBM: implants of DBM loaded

with rhBMP-2. rhBMP2-h/DBM: implants of DBM loaded with rhBMP2-h.

B. Chen et al. / Biomaterials 28 (2007) 1027–10351032

formation [33–35], and it has been shown that the doseneeded to induce bone formation can be greatly reducedwhen BMP-2 is combined with an appropriate carrier [23].On the other hand, due to the potential adverse effects, itwill be safer to limit BMP-2 strictly to the wound site.A safe and effective approach is to enhance the specificbinding ability of BMP-2 to the carrier. By doing so, thequantity of the growth factors can be reduced and it can belimited to the specific wound site, avoiding diffusing intothe blood [36,37].

Considerable research has been focused on the develop-ment of BMP-2 delivery system. However, to the best ofour knowledge, none of the current carriers completelymeets this expectation. Our strategy is to engineer BMP-2to bind to commonly used scaffolds such as demineralizedbone collagen matrix. Several CBD derived from vonWillebrand coagulation factor (vWF) [22,38] or collage-nase [39] were fused to the growth factors, and results

showed the modified growth factors could achieve betterrepairs compared to the native growth factors at the sameconcentration. However, these studies emphasized thecontrolled release of growth factors from scaffolds andthe tissue regenerations outside or inside the scaffold werefailed to be discussed. In our previous work about PDGFwith CBD (CBD–PDGF), effective cellularization andvascularization in collagen membrane loaded CBD–PDGFwas achieved. In this work, we added the peptide of‘‘TKKTLRT’’ to modify BMP-2, and got not only morenew bone but also a homogeneous bone formation. Thispeptide was deduced from the nucleotide sequencecomplementary to that coding for the region in interstitialcollagen surrounding the bond between Gly775 and Ile776,which was cleaved by the enzyme. Labeled collagen boundspecifically and quantitatively to this peptide [24].We proposed a model in Fig. 1 to explain the collagen-

based BMP-2 bone targeting repair system. We compared

ARTICLE IN PRESS

Fig. 7. The photomicrograph of bone formation in mandible defect model. (A) The whole images of the representative slices of the implants (original

magnification � 40). (B) The histomorphometircal analysis of the bone formation at different time point (*Po0.05 and **Po0.01). DBM: implants of

DBM loaded only with PBS; rhBMP-2/DBM: implants of DBM loaded with rhBMP-2; rhBMP2-h/DBM: implants of DBM loaded with rhBMP2-h.

B. Chen et al. / Biomaterials 28 (2007) 1027–1035 1033

it to the traditional growth factor/scaffold slow releasingsystem. The behaviors of growth factor in two systems wereshown in Fig. 1B. When the traditional growth factor slowreleasing system is implanted, there is relatively highconcentration in the entire scaffold at the beginning, themesenchymal cells surrounding the scaffold are stimulatedand some of them will differentiate into the bony cells. Dueto the weak binding ability, the growth factor releases intothe body fluid and the mesenchymal cells can not get thebone-inducing signal from the inner space of the defect site,thus the fibrous tissue invading dominates the bone tissue’sgrowth. Therefore, in the traditional growth factor/scaffoldslow releasing system, better bone formation is expected atthe edge of carriers than in the center, especially in largedefects. While in the proposed collagen-based BMP-2targeting repair system, there exists specific bindingbetween the growth factor and scaffolds. The scaffolds

will maintain the BMP-2 concentration in a relativelylonger period. Thus, a homologous bone formation isexpected.We engineered the CBD-fused protein and demonstrated

that the addition of the peptide of ‘‘TKKTLRT’’ did notaffect the biological activity of BMP-2 using C2C12 cells.Subsequently, using the binding experiment, we manifestedthe binding ability of rhBMP2-h to Type I collagen wasmuch higher than that of rhBMP-2.BMP-2 on the collagen gel could resemble the in vivo

environments. The in vitro functional study showed whenincorporated into Type I collagen, at the same concentra-tions, more rhBMP2-h could be retained on collagen andshowed much higher stimulatory effect on C2C12 cells.This supported our targeting repair system in vitro.The collagen-based BMP-2 targeting bone repair system

was then evaluated in vivo by the ectopic bone formation

ARTICLE IN PRESS

Fig. 8. The local photomicrograph of the histologic images of the implants represented the differences among them (original magnification � 100). DBM:

implants of DBM loaded only with PBS. rhBMP-2/DBM: implants of DBM loaded with rhBMP-2. rhBMP2-h/DBM: implants of DBM loaded with

rhBMP2-h. BL: blood clot; BM: bone marrow; CL: calcified sample; FT: fibrous tissue; NB: new bone; SC: scaffold. Asterisks (*) represent the blood

vessels. Arrow shows the ingrowth osteoblast.

B. Chen et al. / Biomaterials 28 (2007) 1027–10351034

experiment, which was a fast way to evaluate the new boneformation. The histological slices showed DBM alonepossessed limited osteoinductive properties, while rhBMP-2/DBM and rhBMP2-h/DBM induced apparent boneformation. In rhBMP-2/DBM groups, the bone formationin the center was significantly less than that in theperipheral area, while in rhBMP2-h/DBM, there exitedequal bone formation in both central and peripheral areas.Comparing to the poor bone formation in the center ofrhBMP-2/DBM, rhBMP2-h/DBM displayed more boneformation around the trabecular structure of DBM. Theseresults indicated that our targeting repair system stimu-lated better and homologous bone formation.

Our hypothesis was further tested by the rabbitmandibular defect model. This model was used in therepair of ‘‘critical size’’ defect, which implied that the defectcould not heal spontaneously. The photomicrograph of thehistological slices revealed that the bone formation effectsat the peripheral and central areas in the defect sites weresimilar with the ectopic bone formation experiment. Thiswas because the bone regeneration was very slow in PBS/DBM and rhBMP-2/DBM groups, and connective tissuesgrew fleetly and occupied the space. While in rhBMP2-h/DBM groups, homogenously distributing BMP-2 in thescaffold during a longer time continuously induced moremesenchymal cells to differentiate to osteoblasts for betterbone repair. Subsequently these osteoblasts might secretemore collagen matrix and calcium depositing on thecollagen scaffolds. The rhBMP2-h/DBM system couldmaintain a relatively higher concentration of BMP-2 toprovide better environment for bone cell ingrowth anddevelopment, and to prevent fibrous tissue infiltration andto lead to the homogenous bone formation.

5. Conclusion

We have designed and tested a collagen-based BMP-2targeting bone repair system. The cell culture experimentssuggested that the engineering of BMP-2 did not signifi-cantly reduce its bioactivity. The collagen scaffold in vitrocould retain BMP-2 effectively. This collagen-based BMP-2targeting bone repair system showed a better boneinduction effect in the in vivo experiments. Our resultsdemonstrated the feasibility of using a collagen-bindingdomain to develop a collagen-based BMP-2 targetingsystem for homogenous bone repair.

Acknowledgments

This work was supported by grants from NSFC to J. D.(30428017; 30688002) and B. C. (30600304), the authorsalso gratefully acknowledge the support of K.C. WongEducation Foundation, Hong Kong.

References

[1] Shors EC. Coralline bone graft substitutes. Orthop Clin North Am

1999;30(4):599–613.

[2] Ewers R, Goriwoda W, Schopper C, Moser D, Spassova E.

Histologic findings at augmented bone areas supplied with two

different bone substitute materials combined with sinus floor lifting.

Report of one case. Clin Oral Implants Res 2004;15(1):96–100.

[3] Damien CJ, Parsons JR. Bone graft and bone graft substitutes:

a review of current technology and applications. J Appl Biomater

1991;2(3):187–208.

[4] Calvert JW, Weiss LE, Sundine MJ. New frontiers in bone tissue

engineering. Clin Plast Surg 2003;30(4):641–8.

ARTICLE IN PRESSB. Chen et al. / Biomaterials 28 (2007) 1027–1035 1035

[5] Hardouin P, Anselme K, Flautre B, Bianchi F, Bascoulenguet G,

Bouxin B. Tissue engineering and skeletal diseases. Joint Bone Spine

2000;67(5):419–24.

[6] Giannoudis PV, Dinopoulos H, Tsiridis E. Bone substitutes: an

update. Injury 2005;36(Suppl. 3):S20–7.

[7] Shin M, Yoshimoto H, Vacanti JP. In vivo bone tissue engineering

using mesenchymal stem cells on a novel electrospun nanofibrous

scaffold. Tissue Eng 2004;10(1–2):33–41.

[8] Svensson A, Nicklasson E, Harrah T, Panilaitis B, Kaplan DL,

Brittberg M, et al. Bacterial cellulose as a potential scaffold for tissue

engineering of cartilage. Biomaterials 2005;26(4):419–31.

[9] Li RH, Bouxsein ML, Blake CA, D’Augusta D, Kim H, Li XJ, et al.

rhBMP-2 injected in a calcium phosphate paste (alpha-BSM)

accelerates healing in the rabbit ulnar osteotomy model. J Orthop

Res 2003;21(6):997–1004.

[10] Li WJ, Tuli R, Okafor C, Derfoul A, Danielson KG, Hall DJ, et al. A

three-dimensional nanofibrous scaffold for cartilage tissue engineer-

ing using human mesenchymal stem cells. Biomaterials 2005;26(6):

599–609.

[11] Hollinger JO, Schmitt JM, Buck DC, Shannon R, Joh SP, Zegzula

HD, et al. Recombinant human bone morphogenetic protein-2 and

collagen for bone regeneration. J Biomed Mater Res 1998;43(4):

356–64.

[12] Geiger M, Li RH, Friess W. Collagen sponges for bone regeneration

with rhBMP-2. Adv Drug Deliv Rev 2003;55(12):1613–29.

[13] Liu LS, Thompson AY, Heidaran MA, Poser JW, Spiro RC. An

osteoconductive collagen/hyaluronate matrix for bone regeneration.

Biomaterials 1999;20(12):1097–108.

[14] Bostrom MP, Asnis P. Transforming growth factor beta in fracture

repair. Clin Orthop Relat Res 1998(Suppl. 355):S124–31.

[15] Glowacki J, Yates KE, Maclean R, Mizuno S. In vitro engineering of

cartilage: effects of serum substitutes, TGF-beta, and IL-1 alpha.

Orthod Craniofac Res 2005;8(3):200–8.

[16] Ornitz DM. FGF signaling in the developing endochondral skeleton.

Cytokine Growth Factor Rev 2005;16(2):205–13.

[17] Yoon BS, Lyons KM. Multiple functions of BMPs in chondrogen-

esis. J Cell Biochem 2004;93(1):93–103.

[18] Fromigue O, Modrowski D, Marie PJ. Growth factors and bone

formation in osteoporosis: roles for fibroblast growth factor and trans-

forming growth factor beta. Curr Pharm Des 2004;10(21):2593–603.

[19] Reddi AH. Cell biology and biochemistry of endochondral bone

development. Coll Relat Res 1981;1(2):209–26.

[20] Bouxsein ML, Turek TJ, Blake CA, D’Augusta D, Li X, Stevens M,

et al. Recombinant human bone morphogenetic protein-2 accelerates

healing in a rabbit ulnar osteotomy model. J Bone Joint Surg Am

2001;83-A(8):1219–30.

[21] Boden SD, Zdeblick TA, Sandhu HS, Heim SE. The use of rhBMP-2

in interbody fusion cages. Definitive evidence of osteoinduction in

humans: a preliminary report. Spine 2000;25(3):376–81.

[22] Han B, Perelman N, Tang B, Hall F, Shors EC, Nimni ME.

Collagen-targeted BMP3 fusion proteins arrayed on collagen

matrices or porous ceramics impregnated with Type I collagen

enhance osteogenesis in a rat cranial defect model. J Orthop Res

2002;20(4):747–55.

[23] Becerra J, Andrades JA, Santamaria JA, Cifuentes M, Guerado E.

Bone regeneration, cell therapy and tissue engineering. Med Clin

(Barc) 2001;116(1):23–34.

[24] de Souza SJ, Brentani R. Collagen binding site in collagenase can be

determined using the concept of sense-antisense peptide interactions.

J Biol Chem 1992;267(19):13763–7.

[25] Schouten CC, Hartman EH, Spauwen PH, Jansen JA. DBM induced

ectopic bone formation in the rat: the importance of surface area.

J Mater Sci Mater Med 2005;16(2):149–52.

[26] Vallejo LF, Brokelmann M, Marten S, Trappe S, Cabrera-Crespo J,

Hoffmann A, et al. Renaturation and purification of bone

morphogenetic protein-2 produced as inclusion bodies in high-cell-

density cultures of recombinant Escherichia coli. J Biotechnol

2002;94(2):185–94.

[27] Vallejo LF, Rinas U. Optimized procedure for renaturation of

recombinant human bone morphogenetic protein-2 at high protein

concentration. Biotechnol Bioeng 2004;85(6):601–9.

[28] Bradford MM. A rapid and sensitive method for the quantitation of

microgram quantities of protein utilizing the principle of protein–dye

binding. Anal Biochem 1976;72:248–54.

[29] Miller EJ, Rhodes RK. Preparation and characterization of the

different types of collagen. Methods Enzymol 1982;82(Pt A):33–64.

[30] Nimni ME. The molecular organization of collagen and its role in

determining the biophysical properties of the connective tissues.

Biorheology 1980;17(1–2):51–82.

[31] Sykaras N, Opperman LA. Bone morphogenetic proteins (BMPs):

how do they function and what can they offer the clinician? J Oral Sci

2003;45(2):57–73.

[32] Vladimirov BS, Dimitrov SA. Growth factors—importance and

possibilities for enhancement of the healing process in bone fractures.

Folia Med (Plovdiv) 2004;46(2):11–7.

[33] Wozney JM, Rosen V, Celeste AJ, Mitsock LM, Whitters MJ, Kriz

RW, et al. Novel regulators of bone formation: molecular clones and

activities. Science 1988;242(4885):1528–34.

[34] Urist MR. Bone: formation by autoinduction. Science 1965;150(698):

893–9.

[35] Hirota S, Takaoka K, Hashimoto J, Nakase T, Takemura T, Morii

E, et al. Expression of mRNA of murine bone-related proteins in

ectopic bone induced by murine bone morphogenetic protein-4. Cell

Tissue Res 1994;277(1):27–32.

[36] Hall FL, Kaiser A, Liu L, Chen ZH, Hu J, Nimni ME, et al. Design,

expression, and renaturation of a lesion-targeted recombinant

epidermal growth factor-von Willebrand factor fusion protein:

efficacy in an animal model of experimental colitis. Int J Mol Med

2000;6(6):635–43.

[37] Nimni ME. Polypeptide growth factors: targeted delivery systems.

Biomaterials 1997;18(18):1201–25.

[38] Andrades JA, Han B, Becerra J, Sorgente N, Hall FL, Nimni ME.

A recombinant human TGF-beta1 fusion protein with collagen-

binding domain promotes migration, growth, and differentiation of

bone marrow mesenchymal cells. Exp Cell Res 1999;250(2):485–98.

[39] Nishi N, Matsushita O, Yuube K, Miyanaka H, Okabe A, Wada F.

Collagen-binding growth factors: production and characterization of

functional fusion proteins having a collagen-binding domain. Proc

Natl Acad Sci USA 1998;95(12):7018–23.