ARTICLE IN PRESS

0142-9612/$ - se

doi:10.1016/j.bi

�Correspond+90 312 210 12

E-mail addr

Biomaterials 26 (2005) 4023–4034

www.elsevier.com/locate/biomaterials

Collagen–chondroitin sulfate-based PLLA–SAIB-coated rhBMP-2delivery system for bone repair

Dilek Sendil Keskina, Aysen Tezcanera, Petek Korkusuzb, Feza Korkusuzc, Vasif Hasircid,�

aDepartment of Engineering Sciences, Middle East Technical University, 06531 Ankara, TurkeybHacettepe University Faculty of Medicine, Department of Histology and Embryology, 06100 Ankara, Turkey

cMedical Center and Department of Physical Education and Sports, Middle East Technical University, 06531 Ankara, TurkeydDepartment of Biological Sciences, Middle East Technical University, Biotechnology Research Unit, Inonu Bulvari, 06531 Ankara, Turkey

Received 15 July 2004; accepted 28 September 2004

Available online 13 November 2004

Abstract

Bone morphogenetic proteins (BMPs) are osteoinductive proteins used intensively in clinical investigations involving various

bone-related treatments. Owing to their high potential in new bone formation they require local application at the treatment site.

For this purpose various controlled delivery systems with BMPs as the excipients have been prepared in recent years. Focusing on

this clinical need a disc-shaped BMP carrier was designed as a local delivery system using soluble collagen and chondroitin sulfate.

In situ release studies carried out with a model protein (FITC-labeled Protein A) presented a very high rate of release; with most of

the protein content being released within 24 h. This rate could be decreased by providing a poly(L-lactide) (PLLA) and sucrose

acetate isobutyrate-based (SAIB-based) coat around the release system, applied after BMP loading. In this way, it was possible to

extend the release period from 24 h to about 12 days. In situ release of BMP from the same carriers, as quantitated using an ELISA

kit, was even slower, with 50% of the protein being released in 15 days.

In order to be able to secure the BMP delivery system at the bone defect site and to provide support a mesh knitted using Vicryl

sutures and bonded with poly(L-lactide-co-glycolide) (PLGA) was tested in in vivo. Two time periods, 1 and 3 weeks, were used to

evaluate the healing process. Osteoinduction by the BMP carrier system was assessed by histology-based bone scoring and X-ray

examinations. PLLA–SAIB-coated collagen discs containing BMP presented good biocompatibility and optimum osteogenic

stimulation. Structural changes in histological micrographs at week 1 indicated dose-dependent periosteal ossification. At the end of

week 3 histological findings with both BMP (1 and 2mg) doses were almost the same.r 2004 Elsevier Ltd. All rights reserved.

Keywords: Collagen; Chondroitin sulfate; PLLA; SAIB; rhBMP-2; Controlled drug delivery

1. Introduction

Bone morphogenetic proteins (BMPs) have beenshown to elicit new bone formation both at bone defectsite and also at heterotropic sites in a large number ofspecies [1–8]. Two proteins currently in clinical trials forstimulation of local bone regeneration, namely recom-binant human BMPs 2 (rhBMP-2) and 7 (rhBMP-7),were both formulated and employed with absorbable

e front matter r 2004 Elsevier Ltd. All rights reserved.

omaterials.2004.09.063

ing author. Tel.: +90 312 210 51 80; fax:

89.

ess: vhasirci@metu.edu.tr (V. Hasirci).

collagen sponges (Helistats, InfuseTM) [9,10] and ademineralized bone matrix [11].The success of local application of BMP depends very

much on the delivery system used. BMP delivery systemshave received considerable interest in hard tissueengineering [7,12]. There are several criteria that aBMP delivery system should fulfil: biocompatibility,ease of handling, sterilizability, and ability to act as astructural support for new bone formation [7]. Naturalorigin materials such as human demineralised bonematrix, collagen, [5,13,14] and synthetic materials likeb-tricalcium phosphate [3,15], poly(lactic-co-glycolicacid) [2,16,17] have been tested as BMP carriers for

ARTICLE IN PRESSD.S. Keskin et al. / Biomaterials 26 (2005) 4023–40344024

bone regeneration. The important problem to beaddressed with current BMP carriers is that of optimumdelivery of the mediator to allow osteoinduction,differentiation, and extracellular matrix synthesis.This study was designed to investigate the effect of

long-term local availability of rhBMP (1 and 2 mg/implant) on osteoinduction. A BMP delivery systembased on poly(L-lactide)–sucrose acetate isobutyrate-coated (PLLA–SAIB-coated) collagen–chondroitin sul-fate (COL–CS) discs was, therefore, prepared andstabilized at the defect site using a biodegradable mesh.The delivery system was intended to provide two weeksof delivery of BMP in vivo by using a hydrogel core andhydrophobic coat approach as PLLA–SAIB-coatedCOL–CS disc. The effect of provision of two differentdoses were tested in vivo. The level of new boneformation was evaluated with standardized radiogra-phy, histology, and scanning electron microscopy.

2. Materials and methods

2.1. Isolation of collagen type I from rat tail

Isolation of collagen type I from rat tail was carriedout according to Franke et al. [18]. Briefly, rat tails wereincubated in 70% ethanol for 2h and the skin coveringthe tendon fibers was removed. The tendon fibers wereseparated from each other and incubated in 0.1% aceticacid (100mL/g tendon) for 48h at 4 1C. The solution wascentrifuged at 10,000 rpm for 30min. The resultingsupernatant free of tissue remnants was poured into Petriplates and frozen at �40 1C, and freeze-dried for 48h.

2.2. Preparation of collagen–chondroitin sulfate discs

Collagen (6%) was dissolved in 0.1% acetic acidsolution and chondroitin sulfate (0.6%, from bovinetrachea, Sigma) was added to the collagen solution toobtain a (COL:CS) ratio of 10:1. The resulting solutionwas cast in a Petri plate and frozen at –40 1C overnight.COL–CS was freeze dried for 16 h, and stored undervacuum overnight at 50 1C to achieve dehydrothermalcrosslinking. Discs 3mm in diameter were cut andstored in a desiccator until use. COL–CS discs werechemically crosslinked with 28mM N-(3-dimethylami-nopropyl)-N0-ethylcarbodiimide, (EDC), and 10mM N-hydroxysuccinimide, (NHS) by 2 h incubation at 37 1Cin a shaking water bath and then washing with 0.1 MPBS for 1 h and 4 times with distilled water [19].

2.3. Loading of COL–CS discs with Protein A-FITC or

rhBMP-2

Ten microliter of distilled water containing 1 or 2 mgProtein A-FITC (used as a model protein, Sigma) or

rhBMP-2 (Research Diagnostics Inc., USA) was placedon COL–CS discs and impregnation was enhanced withapplication of five vacuum-pressure cycles.

2.4. Coating of COL–CS discs

COL–CS discs were coated by immersion in PLLA(Resomer-L 210, PLLA, degree of polymerisation=3.6,Boehringer Ingelheim, Germany), (10%, w/v in chloro-form, and PLLA–SAIB (1:1) (10%, w/v each in chloro-form), dried under vacuum at room temperature andstored in a desiccator until use.

2.5. In situ release of Protein A or rhBMP-2

Uncoated, PLLA and PLLA–SAIB-coated discsloaded with Protein A-FITC or rhBMP-2 were placedin sterile PBS (1mL, 0.1 M, pH 7.4) and incubated at37 1C in a shaking water bath. Protein A release wasmonitored with a spectrofluorometer (lEx: 494.4 nm,lEm: 516.8 nm, Shimadzu RF 5000, Japan). At eachmeasurement time, the complete release medium wasreplaced with fresh PBS.rhBMP-2 release was studied by taking aliquots

(10 mL) from the release medium at different time pointsand quantitating the amount of protein with QuantikineColorimetric Sandwich ELISA method (R&D Systems,UK). Briefly, rhBMP-2 containing aliquots from therelease medium were pipetted into wells precoated witha monoclonal antibody specific for BMP. After washingaway the unbound molecules, an enzyme-linked mono-clonal antibody specific for BMP was added to the wells.Color is developed in proportion to the amount ofBMP-2 bound in the first step. The amount of BMP wascalculated using a calibration curve constructed withBMP standards.

2.6. Preparation of Vicryl mesh

The biodegradable polymeric mesh was knitted(1mm2 square pattern) using Vicryl (Ethicon Vicryl,Polyglactin 910, calcium stearate-coated, 5-0) suturesand stabilized by coating with a poly(L-lactide-co-glycolide) (PLGA 50:50 solution (0.6% in chloroform).The mesh was cut into rectangular strips (1� 10 cm) andits tensile properties were measured with a MechanicalTester (MT-LQ, Surrey, UK).

2.7. In situ changes in the mechanical properties of Vicryl

sutures and mesh

In order to study the stability of the Vicryl mesh in therelease medium, sutures used to prepare the meshes weretested. They were incubated for a month in PBS (10mM,pH 7.4) at 37 1C in a shaking water bath. At each time

ARTICLE IN PRESSD.S. Keskin et al. / Biomaterials 26 (2005) 4023–4034 4025

point their tensile properties were tested in wet stateusing the Mechanical Tester (MT-LQ, UK).

2.8. In vivo studies

2.8.1. Animals and surgical protocol

Male Sprague–Dawley rats (24) with a weight rangeof 250–300 g were used in the study. Animals werehoused 3 to a cage in the Animal Care Facility of theuniversity, maintained at 22 1C on a 12 h light-12 h darkcycle. Food and water were available ad libitum. Theywere allowed to recover from transport and handlingbefore testing. All procedures were in full compliance ofTurkish Law 6343/2, Veterinary Medicine DeontologyRegulation 6.7.26, and with the Helsinki Declaration ofAnimal Rights [20].The rats were divided into two groups for application

of rhBMP-2 loaded (1 and 2 mg/implant) PLLA–SAIB-coated COL–CS implants, 12 rats (4 from each dosegroups and 4 controls) terminated at week 1 and 12 ratsat week 3. rhBMP-2 containing disks were implantedinto the right-hind limbs (Table 1). Control groups hadtwo negative controls: (a) defects without implants atthe left-hind limbs, and (b) defects with placebo implantat the right-hind limbs.Anesthesia was induced with an intramuscular injec-

tion of ketamine (40mg/kg) and xylazine (10mg/kg).Local anesthetic, 0.5mL lidocaine, was provided routi-nely at the cutaneous site of surgery. The 1/3 centraldiaphyseal parts of the right and left femurs were usedfor the experiments. A 3.2mm cortical defect that didnot extend into the medullary area was created using acontrolled speed diamond burr. Migration of healingmediators from the bone marrow was limited by thismethod leaving mainly cortical and, to some extent,periosteal healing of the defect alone. The implant wasplaced in the defect created. The diaphysis was exposedcircumferentially and the polymeric mesh was wrappedaround the implant securing the defect site. The woundwas closed using sterile, non-absorbable suture material.Postoperatively, the animals were given i.m. injections

of Flexo (15mg/kg, Santa Farma, Istanbul) and

Table 1

In vivo experiment groups for PLLA–SAIB-coated COL–CS System applica

Group no. Implant type

1 1mg rhBMP-2-loaded PLLA–SAIB-coated COL–C2 1mg rhBMP-2-loaded PLLA–SAIB-coated COL–C3 2mg rhBMP-2-loaded PLLA–SAIB-coated COL–C4 2mg rhBMP-2-loaded PLLA–SAIB-coated COL–C5 Placebo implants (PLLA–SAIB-coated COL–CS)

6 Placebo implants (PLLA–SAIB-coated COL–CS)

7 —a

8 —a

n ¼ 4 for all each group.aNegative control groups involving neither the carrier nor the rhBMP-2.

Sephazol (25mg/kg, M.Nevzat, Istanbul) for analgesiaand prevention of infection. Rats were terminated bydiethyl ether inhalation at the end of each time point.

2.8.2. Radiological evaluation

Conventional X-rays were obtained on Agfa Mamor-ay MR3-II (Belgium) films with a Fisher Imaging HF-X(Denver, CO, USA) mammography machine. Thedistance of the X-ray source to the bones was 75 cm.The setting of the machine was 25 kV, 100mA, and160mAs. Films were developed using the Ecomat 21(ELK Medical Products, Tokyo, Japan) automaticdeveloping machine. Each bone was exposed in theanterior–posterior and the lateral directions. Defecthealing was graded from 0 to 3. In case of no healing ofthe defect, it was graded as zero. Healing of 1/3, 2/3, and3/3 of the defect was graded as 1, 2, and 3, respectively.Two radiologists with no prior knowledge about theexperiment graded the X-rays. The average was takeninto consideration when there was a difference in thescores.

2.8.3. Scanning electron microscopy

COL–CS discs were coated with gold before and afterdehydrothermal and chemical crosslinking and theirmorphology was examined with a SEM (JEOL, ModelJSM 6400, Japan). Samples containing the bone and theimplant (PLLA–SAIB-coated COL–CS discs and Vicrylmeshes) were retrieved at different time points, fixed in10% phosphate buffered formalin (pH 7.0) at roomtemperature, gold coated and examined by scanningelectron microscopy.

2.8.4. Histological evaluation

Bone specimens were retrieved, fixed in 10% phos-phate-buffered formalin (pH 7.0) at room temperature,rinsed in buffer, and decalcified in De Castro solution.They were dehydrated in a graded series of ethanolbefore embedding in paraffin. Five to seven micrometerthick sections were prepared with a rotary microtome(Microm, HM 360, Germany). Haematoxylinand Eosin, and Goldners Masson Trichrome-stained

tions

Treatment period (week) Application site

S 1 Right femur

S 3 Right femur

S 1 Right femur

S 3 Right femur

1 Right femur

3 Right femur

1 Left femur

3 Left femur

ARTICLE IN PRESSD.S. Keskin et al. / Biomaterials 26 (2005) 4023–40344026

sections were evaluated for overall morphology, newbone formation, and tissue response. Stained sections (aminimum of 10 sections obtained from different levels ofeach tissue) were examined by at least two investigatorsand the specimens were documented with an OlympusBH-2 microscope (Tokyo, Japan).Histological findings were scored in two categories.

Tissue response to the implant was scored according toRoyals et al. [21] and defect healing was scoredaccording to An and Friedman [22]. The scoring systemis summarized in Table 2. Each sample received twoseparate scores between 0 and 4 for the presence andlevel of fibrous connective tissue and the inflammatorycellular infiltration at the implantation site.

2.8.5. Statistical analysis

Radiologic data were analyzed using parametric test(two-way analysis of variance) to assess statisticalsignificance. Independent variables were the group ofthe subject and the two time points. Dependent variablewas the radiologic findings (XP).

3. Results

3.1. In situ release studies

In the initial drug release studies collagen hydogelswith or without CS (1:8, 1:4, and 1:2 CS:COL) were

Table 2

Histological scoring system

Categories Parameters Scores

Category 1 3 2

Bone defect repair New bone

formation in the

defect

Full bone

formation in the

defect

Moderate bo

formation (4

Cortex remodelling Full cortex

remodeling

Moderate

remodeling (

50)

Category 2 4 3

Tissue response Fibrous connective

tissue formation

Severe deposition

of dense

collageneous

connective tissue

around implant

Disruption o

normal tissu

architecture

presence of

moderately d

fibrous conn

tissue

Inflammatory

cellular infiltration

Severe cellular

infiltrate response

to implant or tissue

necrosis at or

around the site

Presence of l

numbers of

lymphocytes

macrophages

foreign body

cells, also no

presence of

eosinophils a

neutrophils

crosslinked by dehydrothermal treatment (48 h at 80 1C)in a vacuum oven and also chemically crosslinked withEDC–NHS. It was observed that Protein A release wastoo rapid, releasing about 90% of the protein within 1day in all samples. The collagen discs were very unstableand they lost their integrity within 2 days unless CS wasadded. It was, therefore, decided to add CS to thehydrogel composition at the initial hydrogel formationsteps. In order to decrease the release rate, degree ofcrosslinking was improved by increasing the concentra-tion of chemical crosslinkers two-fold and also bycoating the hydrogels with a highly hydrophobic andcrystalline biodegradable polymer, PLLA. Effect ofprotein loading method on the release profile was alsoinvestigated by comparing injection and impregnationmethods in terms of release behavior. Since the drug isintroduced to the center of the discs via injection, it tooksome time for solvents to reach the drug molecules at thecore of the disc so a 1 day lag phase followed by a burstwas observed. However, with the impregnation method,drug molecules were able to leave the carrier startingfrom the placement in the medium without a lag andfollowed by a burst release as observed with the othermethod. As a result, impregnation by vacuum-pressurecycles was found to be more suitable as a drug loadingmethod for controlled release purposes.It was also observed in this study that without

crosslinking, COL–CS discs were not stable and theylost their integrity and disintegrated after 1 day

1 0

ne

%50)

Mild bone

formation (o%50)No new bone

4%Mild remodeling

(o% 50)

No remodelling

2 1 0

f

e

and

ense

ective

Presence of

moderate

connective tissue

Presence of

delicate spindle

shaped cells or

mild fibroplasia

No difference from

normal control

tissue, no presence

of connective tissue

at or around

implant site

arge

,

and

giant

table

nd

Presence of several

lymphocytes,

macrophages with

a few foreign body

giant cells and a

small foci of

neutrophils

Presence of a few

lymphocytes or

macrophages, no

foreign body giant

cells, eosinophils

or neutrophils

No difference from

normal control

tissue, no presence

of macrophages,

foreign body cells,

lymphocytes,

eosinophils or

neutrophils at or

around implant

site

ARTICLE IN PRESS

0

1

2

0 63 9 12 15

Time (days)

Cum

ulat

ive

BM

P R

elea

se (

ug)

COL-CS-1 ug BMP

COL-CS-2 ug BMP

0

10

20

30

40

50

60

70

0 21 3 4

Square root time (days1/2)

Cu

mu

lati

ve B

MP

Rel

ease

(%

)

Fig. 2. Cumulative release profile of rhBMP in PBS (0.1M, pH 7.4) at

37 1C.

Table 3

Mechanical properties of Vicryl (4-0 and 5-0) sutures and meshes (in

situ)

Fiber Ultimate

tensile

strength (N)

Young’s modulus

(N/mm)

Toughness

(Nmm)

4-0

Time 0 12.3070.35 2.4170.22 23.9070.34Day 1 12.5171.08 1.6670.07 54.0071.56Week 1 5.0871.53 2.2170.34 5.6374.61Week 2 0.7070.41 — —

5-0

Time 0 6.9770.42 1.4870.23 13.9071.36Day 1 5.3572.51 1.3670.67 18.27716.50Week 1 2.3970.67 1.8770.31 1.5070.16

D.S. Keskin et al. / Biomaterials 26 (2005) 4023–4034 4027

incubation in the release medium. Protein A loadedCOL–CS discs were, therefore, covalently crosslinkedwith carbodiimide and N-Hydroxysuccinamide and thencoated with PLLA to test their suitability for construct-ing a controlled release system. As compared touncoated COL–CS discs the release was slower–ex-tended to 2 days. However, upon swelling of thehydrogel core in the release medium cracks wereobserved in the polymer coat of COL–CS discs. SAIB,was, therefore, blended with PLLA as a plasticizer tointroduce flexibility and resilience to the coat. Discscoated in this way did not show any cracks and releasedhalf their Protein A content in 5 days and all of it in 11days (Fig. 1). In situ release of rhBMP-2 of two differentdoses (1 and 2 mg/disc) presented a release profile similarin trend but significantly slower than that of Protein A(Fig. 2).

3.2. Mechanical properties of Vicryl sutures and knitted

meshes

When the single suture fibers of (4/0) and (5/0) Vicrylwere compared thicker ones (4/0) had a higher ultimatetensile strength than the thinner ones (5/0) while bothhad similar strain values.In situ incubation of Vicryl sutures were used to

estimate the durability in vivo of the knitted meshesconstructed from them. In situ incubation of the suturesin PBS resulted in roughly a 8.54% decrease in theultimate tensile strength of the sutures within 1 day(Table 3). During tensile testing the multifilaments inVicryl suture ruptured one by one until complete failure.At the end of the first week, a total of 70% decrease inthe ultimate tensile strength was observed. The 5/0Vicryl sutures lost their integrity after 2 weeks in situincubation and no mechanical test could be conducted.However, 4/0 Vicryl sutures maintained their form and

0

20

40

60

80

100

120

0 3 6 9 12 15

Time (days)

Cum

ulat

ive

Pro

tein

A R

elea

se (

%)

Fig. 1. Cumulative release profile of Protein A in PBS (0.1M, pH 7.4)

at 37 1C.

Mesh

4-0 Time 0 118.63717.51 9.5972.87 672.06789.955-0 Time 0 78.3271.14 6.9673.03 713.307145.9

mechanical properties measured were higher than theother group.Similar tests were carried out on meshes

(1 cm� 10 cm) with eight fibers/mesh along the axis.The ultimate tensile strength of knitted meshes at timezero was approximately 10 fold higher than that of asingle suture, thus the formation of mesh structure fromthe sutures created a structure that would give mechan-ical support to the damaged bone (Table 3).Although the thicker 4/0 sutures were found to be

better in terms of mechanical properties, considering theapplicability in rat bone defect model operations thethinner 5/0 sutures were chosen for in vivo studies.

ARTICLE IN PRESS

OL–CSweek3

n¼1)

Untreated

controlweek1

(n¼1)

Untreated

controlweek3

(n¼1)

.070.0

0.070.0

0.070.0

.070.0

0.070.0

0.070.0

.070.0

0.070.0

0.070.0

.570.7

0.070.0

0.070.0

D.S. Keskin et al. / Biomaterials 26 (2005) 4023–40344028

3.3. In vivo studies

3.3.1. Radiological findings

Radiological results show that there was a significantinteraction effect between group of subjects and timevariable (F (3, 19)=3.9, po0:05) (Table 4). Addition-ally, the main effects of time and group variables wasalso found to be significant (F (3, 19)=3.8, po0:05; F (1,19)=8.3, po0:05; respectively). Radiological defecthealing was significantly better in the 2 mg BMP carriergroup at 1 week than all the other groups (Table 4). At 3weeks, the radiological score of the 1 mg BMP carriergroup had reached that of the 2 mg BMP carrier groupand scores were substantially higher than those of theunloaded COL–CS implant alone and untreated defectgroups.

yofimplants

COL–CS1mg

BMPweek1

(n¼2)

COL–CS1mg

BMPweek3

(n¼2)

COL–CS2mg

BMPweek1

(n¼2)

COL–CS2mg

BMPweek3

(n¼4)

COL–CSweek1

(n¼2)

C (

1.070.0

1.070.0

1.070.0

0.370.5

0.570.7

1

1.070.0

1.070.0

1.070.0

1.070.0

1.070.0

1

2.070.0

2.070.0

2.070.0

2.070.0

2.070.0

2

1.570.7

1.070.0

1.570.7

1.370.5

2.070.0

1

3.3.2. Histological findings

Histological scores are presented as average andstandard deviation in Table 5. Defect area was limitedto the outer part of the cortex of bone in allexperimental and control groups. Specimens werestabilized with the polymeric mesh in all but theuntreated defect.BMP-loaded coated discs and the negative control

(rhBMP-2-free coated discs) degraded during the fixa-tion and decalcification stages of the histologicalprocess. Their locations were, therefore, observed asvoids. BMP-loaded implants did not cause a foreignbody reaction in any one of the experimental groups.Coated discs were in close contact with periostealossification areas in all specimens and both the discsand the Vicryl mesh were found to induce no adversereactions. Bone marrow adjacent to the defect area wasintact and healthy. Bone marrow cells were not affectedby the healing process of the defect.At the end of week 1, the bottom of the defect area

that was in contact with the 1 and 2 mg BMP containingPLLA–SAIB-coated COL–CS discs were partially filledwith highly cellular dense connective tissue (Figs. 3a, b).This dense connective tissue was infiltrating into theimplant in some locations. Connective tissue cells ofosteogenic potential were forming well organized youngbone trabecules in continuity with mature cortical bone

Table 4

Radiological scores of groups and time variables

Implant type Radiological scores

Week 1 (mean7SD) Week 3 (mean7SD)

COL–CS 1 mg BMP 1.070.0b 2.370.5c

COL–CS 2 mg BMP 2.271.6a 2.370.6a

Unloaded (control) 1.370.6b 1.070.0b

Untreated (control) 1.370.3b 1.670.6b

a–bo0.5; b–co0.5. Table5

Histologicalscoresofbiocompatibilit

Implant

Parameter

COL–CS

Presenceof

fibrous

connectivetissue

Infiltrationof

inflammatory

cells

Vicrylmesh

Presenceof

fibrous

connectivetissue

Infiltrationof

inflammatory

cells

ARTICLE IN PRESSD.S. Keskin et al. / Biomaterials 26 (2005) 4023–4034 4029

with the 2 mg BMP containing PLLA–SAIB-coatedCOL–CS discs (Fig. 3b). On the other side of theimplant that was in contact with the periosteum, highcellularity was observed. The cells were also mitotic andremnants of the polymeric mesh stabilizing the implantsat site were visible.The thickness of the highly cellular connective tissue

in between the bottom of the cortical defect and theimplant with the discs that contained no BMP (negativecontrol) was the same as those containing BMP at week1. But, the level of organization of that tissue with thenegative control was poorer when compared to thosewith BMP. Cartilage tissue islands were present in theneighbourhood of cortical bone (Fig. 3c). In thenegative control group with the untreated defect, theconnective tissue was thinner when compared to all theother groups (Fig. 5d).At the end of 3 weeks, the cortical bone at the bottom

of the defect adjacent to the 1 and 2 mg BMP containingdiscs were circumferentially filled with thick and regularspongy bone (Figs. 3e,f). This spongy bone with its bonetrabecules, surrounded by active osteoblasts and re-sorptive osteoclasts presented a picture of a well-organized bone tissue. The periosteum neighboring thepolymer was observed as a thin fibrous layer, as itsoriginal appearance. The periosteum was mitoticallyactive and thickened including new bone trabecularislands (Fig. 3e,f) at the periphery of the polymericmatrix holding the implant in place. The thickenedconnective tissue presented encapsulation of the poly-meric matrix in many locations.With the negative control at week 3, loose connective

tissue was found at the defect area. Young bonetrabeculae were present in this connective tissue. Thedeveloping spongy bone with the negative control wasthinner and premature when compared to the BMP-loaded implants. Bone trabecules at the periphery of thepolymeric mesh were not seen. Instead, polymeric meshsections were covered with fibrous capsules at this region(Fig. 3e–g). In the untreated controls, both at thebottom of the defect and at the periosteal level, aconnective tissue that was comparatively more denseand cellular than that of week 1 was observed (Table 5).Cells were mitotically active at this time point. Newbone formation, however, was not present (Fig. 3(h)).

3.3.3. Scanning electron microscopy

At the end of week 1 the mesh around the implant atthe defect site was intact and in place. However, afibrous tissue began to cover the surface of the mesh.PLLA–SAIB-coated COL–CS disc loaded with 1 mgrhBMP-2 was clearly visible under the mesh (Figs. 4aand b). The appearance of the defect sites were similiarfor both doses of rhBMP-2 and placebo group (Figs. 4a,5a, and 6). The mesh and implant with 2 mg rhBMP-2loading for 1 week implantation were removed before

gold coating to observe the defect site (Fig. 5b). As seenin the figure there was a growth of fibrous tissue towardsthe center of the defect. The implants of placebo groupwas also removed (Fig. 6) and a fibrous tissuegrowth was not observed. The defect site of untreatedcontrols was partially filled with connective tissueat the end of week 1 and completely at the end of week3 (Figs. 7a and b).At the end of 3 weeks both 1 and 2 mg rhBMP-2

implants together with the mesh was extensively coveredwith the fibrous tissue. Among the fibrous tissue thefragments of degraded mesh were also recognizable(Figs. 8a and b).

4. Discussion

Osteoinductive effect of BMPs has been reported bymany researchers both in vivo and in vitro [1–6]. Designof delivery systems for rhBMP-2 still receive consider-able interest because many researchers agree thatproperties of the carrier material play an important rolein bone healing [4,5,8,13,23,24]. It was observed that notonly the initial level of rhBMP-2 but also the localretention of rhBMP-2 is important for achievingadequate bone formation [4,5,13,25]. These studies pointout the need for the control of release profile of rhBMP-2 from the carriers.This study was designed to investigate the effect of

long-term release of two different doses of rhBMP-2from crosslinked (dehydrothermally and chemicallywith EDC-NHS) collagen–chondroitin sulfate discsstabilized with a biodegradable mesh on osteoinduction.In situ release studies with a model protein (FITC–Pro-tein A), however, indicated a very fast release profile byreleasing most of the protein within a day. The releaseperiod, was extended to about 12 days by coating with aPLLA–SAIB layer and a slower release profile with 50%of the protein released in 15 days was obtained. Asustained release profile for rhBMP-2 was observedfrom polymer-coated COL–CS discs. SAIB, was alsotested in constructing other drug delivery systems[26–28]. No sign of toxicity due to PLLA–SAIB coatwas observed in in vivo studies.Slower release of BMP from PLLA–SAIB-coated

COL–CS discs compared to Protein A was probably dueto a stronger and higher degree of interaction betweenBMP and collagen both of which are natural compo-nents of the bone matrix. This notion of high interactionwas also indicated by a study of Friess et al. [14] inwhich BMP-loaded collagen discs were tested for ectopicimplant model and showed significant in vivo resultswith 3 h initial retention of BMP-2. Therefore, a moresignificant interaction between collagen and BMP couldbe expected to lead to a lower rate of release in the insitu studies.

ARTICLE IN PRESSD.S. Keskin et al. / Biomaterials 26 (2005) 4023–40344030

ARTICLE IN PRESS

Fig. 4. Scanning electron micrographs of bone defect with 1mgrhBMP-loaded polymer-coated COL–CS discs and mesh after 1 week

of implantation. (a) A fibrous tissue surrounding the implant, (b)

higher magnification of the same. M—mesh, I—implant, C—

connective tissue.

Fig. 5. Scanning electron micrographs of bone defect with 2mgrhBMP-loaded polymer-coated COL–CS discs and mesh after 1 week

of implantation, (a) implant in place, (b) after removal of the implant.

I*—implant covered with connective tissue, C—connective tissue, B—

bone, D—defect.

D.S. Keskin et al. / Biomaterials 26 (2005) 4023–4034 4031

The use of these meshes was expected to hold theimplant in place and to give some support to thedefected bone. The macroscopic observation of theretrieved bone implant specimens showed that the meshwas intact and in place at the end of week 1 but by threeweeks the mesh was degraded.One of the limitations of the study was the number of

samples used for histological evaluation. Therefore, datawas presented only as standard deviation of the means

Fig. 3. Defect area (D) with (a) 1mg BMP (b) 2 mg BMP containing PLLAtrabecules at the connective tissue adjacent to the cortical bone are observed. C

Arrows: Bone trabecules. Arrow head: Cartilage islands. Micrographs of P

untreated control (d) at the end of week 1. Thinned periosteal connective tiss

BMP (f) containing PLLA/SAIB-coated collagen disc implant groups at the e

defect area. C: Cortical bone, I: Implant, (*): Thinned periosteal connective t

collagen discs implant group (g) and the untreated control group (h) at the e

mesh in (g). Adjacent to the cortical bone (C), the highly cellular thickened

thickest part of the periosteum on the section. C: Cortical bone, I: Implant, A

area.

and further statistical analysis was not performed.BMP-loaded implants did not cause a foreign bodyreaction in any one of the experimental groups. Allcoated discs were in close contact with periostealossification areas in all specimens and both the discsand the Vicryl mesh were found to induce no adversereactions. Bone marrow adjacent to the defect area wasintact. At 3 weeks, BMP (both 1 and 2 mg) inducedhistologically well-organized spongy bone formation

/SAIB-coated collagen discs implant groups at week 1. Young bone

: Cortical bone, I: Implant, (*): Thickened periosteal connective tissue.

LLA/SAIB-coated collagen discs implants with no BMP (c) and of

ue (*) adjacent to cortical bone (C) is observed. 1 mg BMP (e) and 2mgnd of week 3. Well-organized spongy bone can be seen adjacent to the

issue. Arrow: Bone trabecules. Micrographs of the PLLA/SAIB-coated

nd of 3 weeks. Fibrous encapsulation is obvious around the polymeric

periosteum (*) can be observed in (h). Micrograph (h) contains the

rrow: Premature bone specula, Arrow head: Polymeric mesh. D: Defect

ARTICLE IN PRESS

Fig. 6. Scanning electron micrographs of bone defect of the placebo

control after 1 week implantation. M—mesh.

Fig. 7. Scanning electron micrographs of bone defect of the untreated

control after (a) 1, (b) 3 week implantation. B—bone, D—defect, C*—

connective tissue filling the defect.

Fig. 8. Scanning electron micrographs of bone defect of (a) 1 mg, (b)2 mg rhBMP-loaded polymer-coated COL–CS discs and mesh after 3weeks of implantation. I*—implant covered with connective tissue,

M*—mesh covered with connective tissue, B—bone, C*—connective

tissue covering the implant.

D.S. Keskin et al. / Biomaterials 26 (2005) 4023–40344032

where in the control group this was not observed. Eventhe number of evaluated histological samples waslimited, BMP containing polymers were efficient in

inducing bone and they were highly biocompatible whenused in the musculoskeletal system.In agreement with earlier investigations, polymer-

coated COL–CS rhBMP-2 carrier system initiated anosteoinductive response on bone defects [1,10,14,26]. Adifference in the amount and time dependent occurrenceof cartilage and bone was observed. This finding wasparallel to Vehof et al’s suggestion for the presence ofosteochondral ossification in rhBMP-2 induced osteo-genesis [29]. Histological evaluation of bone healingrevealed a dose-dependent periosteal ossification at theend of week 1. Anastomosing young bone trabeculeswere higher in number and better organized for thehigher dose of BMP. They were also in continuity withthe remodelling mature cortical bone. At the end ofthree weeks, the cortical bone at the bottom of the defectadjacent to the 1 and 2 mg BMP containing PLLA–-SAIB-coated COL–CS discs were circumferentially filledwith thick and regular spongy bone. At the end of week3 histological findings with both BMP doses were almost

ARTICLE IN PRESSD.S. Keskin et al. / Biomaterials 26 (2005) 4023–4034 4033

the same. This could possibly be explained by thehomeostatic factors in vivo antagonizing the BMPeffect. Radiologically, 2 mg BMP containing PLLA–-SAIB-coated COL–CS discs presented the highest scoresrevealing that this dose is optimal for nearly full defectrepair in 1 week. The same score was reached at 3 weeksby the 1 mg BMP containing PLLA–SAIB-coatedCOL–CS discs but not the polymer-only and defect-only groups. The scores with both the 1 and 2 mg BMPcontaining PLLA–SAIB-coated COL–CS discs at threeweeks reveal that these implants have the potential torepair and regenerate bone rapidly.Comparison of in situ release results of rhBMP with

in vivo evaluations reveal that rhBMP released fromhigher dose (0.67 mg from 2 mg loaded implant) resultedin more new bone trabeculae formation as compared tolower dose (0.39 mg from 1 mg loaded implant) at the endof 1 week. However, this dose-dependent bone forma-tion was not observed on the 3rd week for either of thedoses and resulted in the same degree of boneformation. This might be due to the maximum possibleeffect from both doses being obtained in the 1st weekdue BMP release being higher initially and the lowerdoses at later days do not create a significant difference.

5. Conclusion

The design of the delivery system appears to be notvery simple. At the beginning of the study the use of acollagen–CS-based hydrogel type delivery system wasselected because of the suitable microenvironment itwould create for the BMP and the absence of organicsolvents in the preparation method. Another reason fortheir use was that they are natural components of thebone. A final reason was that the CS being apolyelectrolyte would help retain the protein betterthrough ionic interactions. It was, however, observedthat this sytem had a rapid rate of release and todecrease this rate a control membrane was needed. ThePLLA coat was applied to serve as that coat and inorder to improve its elastic response against the swellingof the hydrogel core a biocompatible ingredient, SAIB,was added into the coat. We believe this system satisfiesthe requirements for BMP release and a simplification ofthe design was not attempted.PLLA–SAIB-coated COL–CS discs for the controlled

delivery of rhBMP-2 presented good biocompatibilityand osteogenic stimulation in vivo. There was no loss ofrhBMP-2 activity due to preparation process of thedelivery system. Week 1 histology data indicated dose-dependent periosteal ossification. At the end of week 3histological findings with both rhBMP-2 doses werealmost the same. The use of a mesh in conjunction withrhBMP-2 delivery system helped stabilize the implant inits place next to the defect site. Initial in situ drug release

studies carried out with model protein (Protein A)showed that the system could be improved by coatingwith the PLLA–SAIB blend. However, release results ofthis model protein and rhBMP-2 were not in agreement,suggesting the influence of protein dimensions, 3-Dstructure, and functional units on properties andfunctionality of the final system. The release ofrhBMP-2 from the final construct was at a slower ratethan that of Protein A and it was possible to achieve arhBMP-2 release for at least 2 weeks. Differences in theamount of drug released due to the drug loading (1 or2 mg) also correlated well with the in vivo performanceof the system, as shown by histological findings of betterbone healing in the 1st week. This dose effect, however,was not observed in the 3rd week. At both time points,however, outcomes with the rhBMP-2 groups werebetter than the placebo and the negative control groups.It is thought that this BMP delivery system will

provide sufficiently prolonged release of this bioactiveagent to trigger bone regeneration in many clinicalapplications. This system can also be used in the deliveryof other protein-based drugs related with the bonetissue. The mesh proved to be very useful in thestabilization of the implant and preventing its disloca-tion.

Acknowledgements

This project (TBAG-2060) was supported by a grantby the Scientific and Technical Research Council ofTurkey. We are indepted to Mr. Erdinc Oran (OrhanBoz Inc.) for supplying Vicryl sutures for the project.We thank Kodak (USA) for kindly supplying us withSAIB. The authors also thank Sedat Is-ıklı and SibelUlger for their technical assistance in statistical analysisof the data and radiographic measurements, respec-tively.

References

[1] Johnson EE, Urist MR, Finerman GAM. Bone morphogenetic

protein augmentation grafting of resistant femoral nonunions: A

preliminary report. Clin Orthop 1988;230:257–65.

[2] Heckman JD, Brooks BP, Morgan T. Bone morphogenetic

protein but not transforming growth factor-b enhances bone

formation in canine diphyseal nonunions implanted with a

biodegradable composite polymer. J Bone Joint Surg

1999;81A(12):1717–29.

[3] Laffargue PH, Hildebrand HF, Rtaimate M, Frayssinet P,

Amoureux JP, Marchandise X. Evaluation of human recombi-

nant bone morphogenetic protein protein-2 loaded tricalcium

phosphate implants in rabbits’ bone defects. Bone

1999;25(2):55S–8S.

[4] Uludag H, D’Augusta, Palmer R, Timony G, Wozney J.

Characterization of rhBMP-2 pharmokinetics implanted with

biomaterial carriers in the rat ectopic model. J Biomed Mater Res

1999;46:193–202.

ARTICLE IN PRESSD.S. Keskin et al. / Biomaterials 26 (2005) 4023–40344034

[5] Uludag H, D’Augusta D, Golden J, Li J, Timony G, Riedel R,

Wozney JM. Implantation of recombinant human bone morpho-

genetic proteins with biomaterial carriers: A correlation between

pharmacokinetics and osteoinduction in the rat ectopic model.

J Biomed Mater Res 2000;50:227–38.

[6] Saito N, Okada T, Horiuchi H, Murakami N, Takahashi J,

Nawata M, Ota H, Nozaki K, Takaoka K. A biodegradable

polymer as a cytokine delivery system for inducing bone

formation. Natl Biotechnol 2001;19:332–5.

[7] Saito N, Takaoka K. New synthetic biodegradable polymers as

BMP carriers for bone tissue engineering. Biomaterials

2003;24:2287–93.

[8] Saito N, Okada T, Horiuchi H, Takhashi J, Murakami N,

Nawata M, Kojima S, Nozaki K, Takaoka K. Local bone

formation by injection of recombinant human bone morphoge-

netic protein-2 contained in polymer carriers. Bone

2003;32:381–6.

[9] Boyne PJ, Marx RE, Nevins M, Triplet G, Lazaro E, Lilly LC,

Alder M, Nummikoski PA. A feasibility study evaluating rhBMP-

2/absorbable COL–CS sponge for maxillary sinus augmentation.

Int J Periodon Rest Dent 1997;17:11–25.

[10] Howell TH, Fierellini J, Jones A, Alder M, Nummikoski P,

Lazaro M, Lilly L, Cochran DA. A feasibility study evaluating

rhBMP-2/absorbable collagen sponge device for local alveolar

ridge preservation or augmentation. Int J Periodont Restor Dent

1997;17:125–39.

[11] Cook SD, Wolfe MW, Salkeld SL, Rueger DC. Effect of

recombinant human osteogenic protein-1 on healing of segmental

defects in nonhuman primates. J Bone Joint Surg

1995;77A:734–50.

[12] Lu HH, Kofron MD, El-Amin SF, Attawia MA, Laurencin CT.

In vitro bone formation using muscle-derived cells: a new

paradigm for bone tissue engineering using polymer–bone

morphogenetic protein matrices. Biochem Biophys Res Commun

2003;13:882–9.

[13] Maeda H, Sano A, Fujioka K. Controlled release of rhBMP-2

from collagen minipellet and the relationship between release

profile and ectopic bone formation. Int J Pharm 2004;275:109–22.

[14] Friess W, Uludag H, Foskett S, Biron R, Sargeant C.

Characterization of absorbable sponges as rhBMP-2 carriers.

Int J Pharm 1999;187:91–9.

[15] Niedhart C, Maus U, Redmann E, Schmidt-Rohlfing B, Niethard

FU, Siebert CH. Stimulation of bone formation with an in situ

setting tricalcium phosphate/rhBMP-2 composite in rats. J

Biomed Mater Res 2003;65A(1):17–23.

[16] Nagao H, Tachikawa N, Miki T, Oda M, Mori M, Takahashi K,

Enomoto S. Effect of recombinant human bone morphogenetic

protein-2 on bone formation in alveolar ridge defects in dogs. Int

J Oral Maxillofac Surg 2002;31(1):66–72.

[17] Bessho K, Carnes DL, Cavin R, Ong JL. Experimental studies on

bone induction using low-molecular-weight poly (DL-lactide-co-

glycolide) as a carrier for recombinant human bone morphoge-

netic protein-2. J Biomed Mater Res 2002;61(1):61–5.

[18] Franke H, Galla HJ, Beuckmann CT. Primary cultures of brain

microvessel endothelial cells: a valid and flexible model to study

drug transport through the blood-brain barrier in vitro, Brain Res

Prot 2000;5:248–56.

[19] Pieper JS, Oosterhof A, Dijkstra PJ, Veerkamp JH, van

Kuppevelt TH. Preparation and characterization of porous

crosslinked collageneous matrices containing bioavailable chon-

droitin sulfate. Biomaterials 1999;20:847–58.

[20] Bennet BT. Regulation and requirements. In: Bennett BT, Brown

MJ, Schofield JC, editors. Essentials for animal research: a primer

for research personnel. Beltsville, MD: Maryland; 1994. p. 1–9.

[21] Royals MA, Fujita SM, Yevey GL, Rodriguez J, Schultheiss PC,

Dunn RL. Biocompatibility of a biodegradable in situ forming

implant system in rhesus monkeys. J Biomed Mater Res

1999;45:231–9.

[22] An YH, Friedman RJ. In: An YH, Friedman RJ, editors. Animal

models in orthopedic research, CRC, 1999. p. 251–259.

[23] Kokubo S, Mochizuki M, Fukushima S, Ito T, Nozaki K, Iwai T,

Takahashi K, Yokota S, Miyata K, Sasaki N. Long-term stability

of bone tissues induced by an osteoinductive biomaterial,

recombinant human bone morphogenetic protein-2 and a

biodegradable carrier. Biomaterials 2004;25:1795–803.

[24] Kuboki Y, Jin Q, Takita H. Geometry of carriers controlling

phenotypic expression in BMP-induced osteogenesis and chon-

drogenesis. J Bone Joint Surg 2001;83A(Part 2):S1105–15.

[25] Okamoto T, Yamamoto Y, Gotoh M, Huang CL, Nakamura T,

Shim Y, Tabata Y, Yokomise H. Slow release of bone

morphogenetic protein from a gelatin sponge to promote

regeneration of tracheal cartilage in a canine model. J Thorac

Cardiovasc Surg 2004;127(2):329–34.

[26] Burns PJ, Fleury JJ, Gibson J, Sullivan S, Tipton A. Evaluation

of the SABERTM delivery system for the controlled release of

deslorelin acetate for advancing ovulation in the mare: Effects of

formulation and dose. Theriogenology 1998;49(1):256.

[27] Knoll A, Schmidt S, Chapman M, Wiley D, Bulgrin J, Blank J,

Kirchner L. A comparison of two controlled-release delivery

systems for the delivery of amiloride to control angiogenesis.

Microvasc Res 1999;58(1):1–9.

[28] Okumu FW, Dao le N, Fielder PJ, Dybdal N, Brooks D, Sane S,

Cleland JL. Sustained delivery of human growth hormone from a

novel gel system:SABER. Biomaterials 2002;23(22):4353–8.

[29] Vehof JWM, Takia H, Kuboki Y, Spauwen PHM, Jansen JA.

Histological characterization of the early stages of bone-

morphogenetic protein-induced osteogenesis. J Biomed Mater

Res 2002;61:440–9.