Hydrogels derived from demineralized and decellularized bone extracellular matrix

Acta Biomaterialia 9 (2013) 7865–7873

Contents lists available at SciVerse ScienceDirect

Acta Biomaterialia

journal homepage: www.elsevier .com/locate /actabiomat

Hydrogels derived from demineralized and decellularized boneextracellular matrix

1742-7061/$ - see front matter � 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.actbio.2013.04.029

⇑ Corresponding author. Tel.: +44 115 82 31232.E-mail address: [email protected] (L.J. White).

M.J. Sawkins a, W. Bowen a, P. Dhadda a, H. Markides a, L.E. Sidney a, A.J. Taylor a, F.R.A.J. Rose a,S.F. Badylak b, K.M. Shakesheff a, L.J. White a,⇑a School of Pharmacy, University of Nottingham, Nottingham NG7 2RD, UKb McGowan Institute for Regenerative Medicine, University of Pittsburgh, Pittsburgh, PA, USA

a r t i c l e i n f o

Article history:Received 10 December 2012Received in revised form 15 April 2013Accepted 16 April 2013Available online 25 April 2013

Keywords:Extracellular matrixDemineralized bone matrixHydrogelBone graftDecellularization

a b s t r a c t

The extracellular matrix (ECM) of mammalian tissues has been isolated, decellularized and utilized as ascaffold to facilitate the repair and reconstruction of numerous tissues. Recent studies have suggestedthat superior function and complex tissue formation occurred when ECM scaffolds were derived fromsite-specific homologous tissues compared with heterologous tissues. The objectives of the present studywere to apply a stringent decellularization process to demineralized bone matrix (DBM), prepared frombovine bone, and to characterize the structure and composition of the resulting ECM materials and DBMitself. Additionally, we sought to produce a soluble form of DBM and ECM which could be induced to forma hydrogel. Current clinical delivery of DBM particles for treatment of bone defects requires incorporationof the particles within a carrier liquid. Differences in osteogenic activity, inflammation and nephrotoxi-city have been reported with various carrier liquids. The use of hydrogel forms of DBM or ECM mayreduce the need for carrier liquids. DBM and ECM hydrogels exhibited sigmoidal gelation kinetics consis-tent with a nucleation and growth mechanism, with ECM hydrogels characterized by lower storage mod-uli than the DBM hydrogels. Enhanced proliferation of mouse primary calvarial cells was achieved onECM hydrogels, compared with collagen type I and DBM hydrogels. These results show that DBM andECM hydrogels have distinct structural, mechanical and biological properties and have the potentialfor clinical delivery without the need for carrier liquids.

� 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction

The extracellular matrix (ECM) of mammalian tissues can beisolated, decellularized and utilized as a biological scaffold [1]. Bio-logical scaffolds derived from tissues such as the small intestine,urinary bladder or dermis have been shown to facilitate functionalrestoration of different tissues, including heart and vascular struc-tures [2,3], oesophagus [4] and musculo-skeletal tissues [5,6]. Themechanisms by which constructive remodelling occur are welldocumented [7,8] and include the recruitment of progenitor cells[9], promotion of cell migration and proliferation [10,11], regionalangiogenesis [12] and promotion of a favourable M2 macrophagephenotype at the interface of the host tissue and biological scaffold[13]. Although these tissue-derived biological materials have beensuccessfully used in non-homologous sites, recent studies havedemonstrated ‘‘tissue specificity’’, with the occurrence of addi-tional functions [14] and complex tissue formation [15–17] when

biological scaffolds were derived from site-specific homologoustissues.

Musculo-skeletal conditions are the most common cause of se-vere long-term pain and physical disability worldwide [18], withmore than 3 million musculoskeletal procedures performed annu-ally in the USA [19]. Degenerative disease, severe infection, traumaand the excision of tumours can result in large non-healing defectsin bone and other integrated tissues [20]. Current treatment op-tions for bone have limited effectiveness. Although autologousbone grafts are considered to be the gold standard with the bestclinical outcome, significant limitations include restricted avail-ability of donor tissue and morbidity at the harvest site. Shortcom-ings of allografts comprise issues of processing, sterilization,disease transmission and potential immunogenic response [21],with high rates of fractures and complications, attributed to theirlimited ability to revascularize and remodel [22].

Bone graft substitutes, such as demineralized bone matrix(DBM), have been developed to overcome the limitations of bothautografts and allografts [23]. Osteoconductive DBM is producedby the acid extraction of the mineral content from allogeneic boneand contains growth factors, non-collagenous proteins and type I

7866 M.J. Sawkins et al. / Acta Biomaterialia 9 (2013) 7865–7873

collagen. Whilst the osteoinductive effect of DBM has been welldocumented in animal studies, albeit with variability [24,25], thereis a paucity of similar information for human clinical studies [26],despite a robust clinical demand for DBM products. Differences inthe preparation and processing methods and donor age all have animpact on DBM properties and clinical performance [27]. The endproduct of the demineralization process is a DBM powder. To facil-itate handling, formulation and reliable delivery clinically theseparticles are usually incorporated in a carrier. For example, themost common clinical form of DBM is a mouldable putty, which in-volves formulation with a biocompatible viscous carrier that pro-vides a stable suspension of DBM powder particles [27]. Theviscous carriers are generally either water-soluble polymers, suchas sodium hyaluronate or carboxymethylcellulose, or anhydrouswater-miscible solvents, such as glycerol. Studies designed to testthe effectiveness of various carriers on DBM efficacy are limited.One study reported nephrotoxicity [28] amidst speculation regard-ing glycerol as a carrier. Differences in osteogenic activity have alsobeen observed [24,25,29] which may be related to different carri-ers, the amount of DBM in the carrier and ability of the carrier tolocalize the DBM particulates to the bone defect site for a sufficientperiod of time to promote bone regeneration [29]. Additionally, arecent study characterized an inflammatory response to four com-mercial bone graft substitutes and found that the three DBM mate-rials produced more inflammation than a synthetic hydroxyapatitecompound. It was undetermined whether the DBM material or car-rier provoked the inflammatory response [30].

The objectives of the present study were to apply a stringentdecellularization process to DBM prepared from bovine bone andto characterize the structure and composition of the DBM andresulting ECM materials. To our knowledge this is the first timethat a demineralized ECM material has been produced from bone.Additionally, we sought to produce a soluble form of the DBM andECM materials which could be induced to polymerize into a gel.The rationale was to provide enhanced clinical utility of thesematerials without the inclusion of a carrier. The long-term objec-tive of this work was to develop a gel form of DBM/ECM biologicalmaterial that retains osteoconductivity and osteoinductivity. Thepresent study describes the first steps towards this goal with char-acterization of the gelation kinetics, rheological properties andin vitro cytocompatibility of the gels.

2. Materials and methods

2.1. Bone preparation

Fresh bovine tibiae were harvested from cattle aged 12–24 months, slaughtered by an EU certified butcher (J. BroomhallLtd, Eastington, UK). Bones were received in segmented form andwere separated into cancellous and cortical groups, with thecancellous group used in this study. Bones were either used as

Fig. 1. Production of ECM hydrogel from bone. Bovine tibiae were processed to form (produce (B) decellularized bone (bECM) prior to pepsin digestion and solubilization to f

received or stored at �20 �C in order to preserve their osteoinduc-tive potential [31] and were processed using a modification of pre-viously reported methods [32]. Cancellous segments were cleanedof residual tissue and washed with phosphate-buffered saline(PBS) containing 0.1% w/v Gentamicin (Invitrogen, Paisley, UK).Washed segments were frozen in liquid nitrogen and sectionedto produce fragments no greater than 4 � 4 � 4 mm. Fragmentswere washed in distilled water, immersed in liquid nitrogen andground in a commercial coffee mill (Krups F203) (Fig. 1A).

2.2. Demineralization and decellularization

Cancellous bone granules were demineralized using an adapta-tion of previously reported methods [33]. In brief, the granuleswere demineralized under agitation in 0.5 N HCl (25 ml g�1 bone)at room temperature for 24 h. Stirred beakers of bone granulesand acid were agitated at 300 r.p.m. to generate a small vortex;particles were suspended in motion in the acid and did not settleduring the process.

After demineralization the resultant material, referred to as bo-vine demineralized bone matrix (bDBM), was filter separated un-der vacuum from the acid and rinsed with distilled water. Thelipid in the demineralized powder was then extracted with a 1:1mixture of chloroform (Fisher Scientific, Loughborough, UK) andmethanol (Fisher Scientific) for 1 h and then repeatedly rinsed,firstly in methanol and then distilled water. The bDBM was thensnap frozen, lyophilized overnight and stored at �20 �C untilrequired.

An enzymatic decellularization protocol, adapted from previ-ously reported methods [34], was applied. Briefly, lyophilizedbDBM was rinsed with distilled water and decellularized in a solu-tion of 0.05% trypsin (Sigma–Aldrich, Poole, UK) and 0.02% ethy-lenediamine tetraacetic acid (EDTA) (Sigma–Aldrich) at 37 �C and5% CO2 under continuous agitation for 24 h.

The resultant material, referred to as bovine decellularizedmatrix (bECM) was rinsed in PBS supplemented with 1% w/vpenicillin/streptomycin under continuous agitation for 24 h at4 �C to remove residual cellular material. The bECM was then snapfrozen, lyophilized overnight and stored at �20 �C until required(Fig. 1B).

2.3. Digestion and solubilization

A previously reported pepsin digestion and solubilization tech-nique was employed [35]. Lyophilized bDBM and bECM were sep-arately added to 1 mg ml�1 pepsin in 0.01 N HCl for a finalconcentration of 10 mg matrix per ml suspension, i.e. 1 g dry ma-trix was mixed with 100 mg pepsin in 100 ml of 0.01 N HCl. Thesuspension was mixed on a stirrer plate at room temperature for96 h, until no visible pieces of matrix remained. The resultant

A) fragments and then subjected to mineral, lipid and cell removal procedures toorm an ECM hydrogel (C).

M.J. Sawkins et al. / Acta Biomaterialia 9 (2013) 7865–7873 7867

bDBM and bECM digests were aliquoted and stored at �20 �C untilrequired.

2.4. Assessment of cellular content

Representative samples of lyophilized bDBM and bECM werefixed in 10% neutral buffered formalin and embedded in 3% agarosegel prior to paraffin embedding. Sections were cut at 5–7 lm thick-ness and stained with haemotoxylin and eosin (H&E) to identifythe presence of any visible intact nuclei.

Quantification of DNA content was conducted by an adaptationof previously reported methods [36,37]. DNA was extracted frompepsin digests of lyophilized bDBM and bECM (10 mg ml�1 con-centration) using 50:48:2 (vol.%) phenol/chloroform/isoamyl alco-hol (Sigma–Aldrich, Poole, UK). DNA was precipitated from theaqueous phase at �20 �C by the addition of 0.1 volume of 3 M so-dium acetate (pH 5.2) (Sigma–Aldrich, Poole, UK) and 2 volumes ofethanol and was then frozen. The frozen DNA was then centrifugedat 10,000g for 10 min to form a DNA pellet. The pellet was washedwith ethanol, dried at room temperature and resuspended in 1 mlof TE buffer.

The concentration of each extracted DNA sample was deter-mined using a Quant-iT™ PicoGreen dsDNA assay kit (Invitrogen,Paisley, UK) following the manufacturer’s protocol. A standardcurve was constructed by preparing samples of known DNA con-centration from 0 to 1000 ng ml�1. Extracted DNA samples werediluted to ensure their absorbencies fell within the linear regionof the standard curve. Samples were read using a Tecan InfiniteM200 plate reader (Tecan UK, Reading, UK).

2.5. Assessment of collagen content

The hydroxyproline content of pepsin digests of bDBM andbECM was determined by hydrolysing with concentrated HCl(1 ml of each solution) at 120 �C overnight. Samples were incu-bated uncapped at 90 �C until dry and then 4 ml of 0.25 M sodiumphosphate buffer (pH 6.5) (Sigma–Aldrich, Poole, UK) was added toeach sample. Blank pepsin solution was hydrolysed and used as acontrol and diluent for the assay. 50 ll of each sample was reactedwith 50 ll of chloramine T solution (Sigma–Aldrich, Poole, UK) andallowed to oxidize at room temperature for 20 min. The sampleswere then mixed with 50 ll of p-dimethylaminobenzaldehyde(p-DAB) (Ehrlich’s reagent, Sigma–Aldrich, Poole, UK) and incu-bated at 60 �C for 30 min. A standard curve was constructed bypreparing samples of known hydroxyproline concentrations from0 to 30 lg ml�1. The colorimetric change at an absorbance of540 nm was detected using a Tecan Infinite M200 plate reader (Te-can UK, Reading, UK). The total collagen content of the digests wasdetermined using the relationship that hydroxyproline forms14.3% of total collagen [38].

2.6. Gelation

Rat tail collagen type I, bDBM and bECM gels were formed usinga previously described method [39,40]. Briefly, gelation was in-duced by neutralizing the salt concentration and pH of the pepsindigests or collagen solution at 4 �C followed by warming to 37 �C.Neutralization of the required digest volume occurred by additionof one tenth of the digest volume of 0.1 N NaOH, one ninth of thedigest volume of 10� PBS and by then diluting to the desired finalECM concentration with 1� PBS on ice. Gelation of this pre-gelsolution occurred after 1 h at 37 �C (Fig. 1C). Concentrations of 3and 6 mg ml�1 bECM, bDBM and collagen were prepared.

Turbidimetric gelation kinetics of collagen type I (Coll I), bDBMand bECM hydrogels were determined spectrophotometrically aspreviously described [39]. Pre-gel solutions were kept at 4 �C and

transferred to cold 96-well plates (100 ll). The plates were placedin a pre-warmed (37 �C) Tecan Infinite M200 plate reader and theturbidity of each well measured at 405 nm every 3 min for 1.5 h.Absorbance values for each well were recorded; six individual(n = 6) measurements of each hydrogel type and concentrationwere performed and the results averaged. These readings werethen scaled from 0 (at time 0) to 1 (at maximum absorbance) toprovide a normalized absorbance (NA) as shown in Eq. (1).

NA ¼ A� A0

Amax � A0ð1Þ

where A is the absorbance at a given time, A0 is the initial absor-bance and Amax is the maximum absorbance. The lag time (tlag)was defined as the intercept of the linear region of the gelationcurve with 0% absorbance.

2.7. Rheological characteristics

The rheological characteristics of bECM, bDBM and collagentype I hydrogels were determined using a Physica MCR 301 rheom-eter (Anton Paar, Hertford, UK). Pre-gel solutions at 4 �C wereplaced between 50 mm parallel plates separated by a 0.2 mmgap. The plates were pre-cooled in a humidified chamber to 4 �Cand were then warmed to 37 �C during the first 75 s of each mea-surement run. Initially a 60 min time course experiment was per-formed during which the samples were subjected to anoscillatory strain of 1% at a constant angular frequency of 1 rad s�1

with readings taken every 30 s. Immediately following this thesamples were subjected to an amplitude sweep covering the range0.1–200% strain at the same constant angular frequency.

2.8. Gel morphology

Surface morphology of the bDBM, bECM and collagen type Ihydrogels was examined by scanning electron microscopy (SEM).Gel specimens (400 ll per well) were fixed in 1 ml of 3% glutaral-dehyde and then rinsed in PBS, followed by dehydration through agraded series of ethanol (30–100%). Subsequently the hydrogelswere critically point dried in a Samdri pvt-3 critical point dryer(Tousimis, Rockville, MD). The samples were then attached to alu-minium mounting stubs and sputter coated with platinum using aPolaron SC7640 (Quorum Technologies, Ashford, UK) sputter coat-er at a voltage of 2.2 kV and plasma current of 15 mA for 90 s.Hydrogels were then examined using a Phillips XL30 FEG SEM(FEI, Eindhoven, The Netherlands) and images were obtained at8000� and 16,000� magnification.

2.9. In vitro cell proliferation

Mouse primary calvarial cells (mPCs), an osteogenic populationof cells comprised predominantly of osteoblasts, were obtainedfrom 1- to 3-day-old mouse calvaria by sequential enzymaticdigestion. Briefly, the calvaria were dissected from CD1 neonatesand digested using a solution of 1.4 mg ml�1 collagenase type IAand 0.5 mg ml�1 trypsin II S (Sigma–Aldrich, Poole, UK). Cells re-leased in the first two populations (10 min each digestion) werediscarded and the population of cells from the next three diges-tions (20 min each digestion) were plated in tissue culture flasksat a density of 6.6 � 103 cells cm�2. All digestions were performedon rollers set to 30 r.p.m. at 37 �C. Cells were cultured in a-minimalessential medium (Lonza, Slough, UK) containing 10% fetal calf ser-um (FCS) and 2 mM L-glutamine (Sigma–Aldrich, Poole, UK) and100 U ml�1 penicillin and 100 lg ml�1 streptomycin (Invitrogen,Paisley, UK).

In vitro cell proliferation on the surface of 3 and 6 mg ml�1

bECM, bDBM and collagen type I hydrogels was characterized

Fig. 3. Turbidimetric gelation kinetics of bECM and bDBM hydrogels comparedwith collagen at (A) 3 and (B) 6 mg ml�1 concentrations. Pre-gel solutions wereneutralized and added to the wells of a 96-well plate at 37 �C to induce gelation. Theabsorbance at 405 nm was measured at 3 min intervals and normalized between 0(the initial absorbance) and 1 (the maximum absorbance). Data representmeans ± standard deviation for n = 6.

7868 M.J. Sawkins et al. / Acta Biomaterialia 9 (2013) 7865–7873

using the CellTiter96� Aqueous Non-radioactive MTS colorimetricassay (Promega, Southampton, UK). Briefly, pre-gel solutions keptat 4 �C and transferred to cold 96-well plates (100 ll). Once thehydrogels had formed (1 h at 37 �C) mPCs were added to the sur-face of the gels and cultured for 48–72 h. Proliferation was as-sessed following the manufacturer’s instructions; the CellTiter96� MTS solution is bioreduced by cells to a formazan product, sol-uble in tissue culture medium. Briefly, 20 ll of CellTiter 96� AQue-ous One Solution was added to each well, incubated for 3 h and theabsorbance of the formazan product at 490 nm measured directlyusing a Tecan Infinite M200 plate reader. The conversion of MTSto the aqueous soluble formazan product is accomplished by dehy-drogenase enzymes found in metabolically active cells. Thus thequantity of formazan product measured as the 490 nm absorbanceis directly proportional to the number of living cells in culture. Thebackground absorbance of each distinct hydrogel type and concen-tration was subtracted from the absorbance of mPCs on the corre-sponding hydrogel to provide a normalized absorbance. Allconditions were assessed in sextuplicate.

2.10. Statistical analysis

All statistical analyses were performed using GraphPad Instat(Graph Pad Software Inc., La Jolla, CA). All values are reported asmeans ± standard deviation. In vitro cell proliferation values weretested for normality and statistically compared using a Tukey–Kra-mer multiple comparisons test. Significance for all statistical anal-yses was defined as p < 0.001.

3. Results

3.1. Preparation of ECM hydrogel from bone

Fresh bovine tibiae were processed into a fragmented form anddemineralized using acid extraction to remove the mineral con-tent. Lipid removal was achieved with chloroform/methanol, andan enzymatic decellularization procedure was applied to thedemineralized bone matrix (bDBM) to produce decellularized ma-trix material (bECM) (Fig. 1). The bDBM and bECM materials weredigested and solubilized with pepsin and hydrogels were success-fully prepared from both bDBM and bECM at concentrations of 3and 6 mg ml�1. The higher ECM/DBM concentration hydrogels(6 mg ml�1) had a more rigid structure compared with the lowerconcentration (3 mg ml�1) hydrogels.

3.2. Cellular and collagen content of demineralized and decellularizedmaterial

Determination of effective decellularization was based uponestablished criteria [41,42], specifically: (i) removal of nuclei as ob-served by imaging and analysis of H&E and/or DAPI stained sec-tions; (ii) samples should possess <50 ng double stranded DNA

Fig. 2. Decellularization was assessed by imaging and analysis of (H&E) stained sectionweight measured with a Quant-iT PicoGreen dsDNA assay kit.

(dsDNA) per mg initial dry weight. H&E sections clearly showedthat decellularization had removed all cell nuclei (Fig. 2B), butsome cell nuclei appeared to be present in demineralized sections(Fig. 2A). Quantification of dsDNA content showed that this wasconsiderably lower than the 50 ng threshold in both bDBM andbECM (Fig. 2C).

The soluble collagen contents of the bDBM and bECM materialwere determined to be 0.93 ± 0.06 and 0.92 ± 0.06 mgs of collagenper mg of initial dry weight respectively.

3.3. Turbidimetric gelation kinetics

The turbidimetric gelation kinetics of bDBM and bECM hydro-gels were characterized spectrophotometrically and comparedwith the same concentration (3 and 6 mg ml�1) collagen type I

s of (A) bDBM and (B) bECM and (C) quantification of dsDNA content per mg dry

M.J. Sawkins et al. / Acta Biomaterialia 9 (2013) 7865–7873 7869

hydrogels. The turbidimetric gelation kinetics for all materials andconcentrations showed a sigmoidal shape (Fig. 3) with hydrogelformation occurring after a lag period (tlag). At bothconcentrations the collagen type I hydrogel had a shorter tlag thanthe bECM hydrogel, which had a shorter tlag than the bDBM hydro-gel (Supplementary Fig. S1). Gelation kinetics appeared to be inde-pendent of concentration.

3.4. bDBM, bECM and collagen type I hydrogel rheology

The rheological characteristics of the bDBM and bECM hydro-gels were determined using a parallel plate rheometer and com-pared with the same concentration (3 and 6 mg ml�1) collagentype I hydrogels. In each case the storage (G0) and loss (G00) moduliof the hydrogels increased after the pepsin digests (or pre-gel col-lagen type I solutions) were neutralized and the temperature wasincreased from 4 �C to 37 �C. Solid-like behaviour was confirmedas the storage moduli were greater than the loss moduli by a factorof approximately 10 for the bECM and bDBM hydrogels and a fac-tor of approximately 20 for the collagen type I hydrogels (Fig. 4).The bECM, bDBM and collagen type I hydrogels showed an increasein rate of gelation with increasing concentration, with the bECM

Fig. 4. Rheological characterization of bECM, bDBM and collagen type I 6 mg ml�1

hydrogels. The gelation kinetics were determined by monitoring changes in thestorage modulus (G0) and loss modulus (G00) after inducing gelation. Data representmeans ± standard deviation for n = 6.

Fig. 5. Amplitude sweep of bECM, bDBM and collagen type I 6 mg ml�1 hydrogels.The storage modulus (G0) and loss modulus (G00) were monitored when hydrogelswere subjected to an amplitude sweep of 0.1–200% strain at a constant angularfrequency. Data represent means ± standard deviation for n = 6.

and bDBM hydrogels having storage moduli of 32.5 ± 3.8 and58.7 ± 4.3 Pa, respectively, at 3 mg ml�1 (Supplementary Fig. S2)and 143.7 ± 7.6 and 313.0 ± 31.2 Pa at 6 mg ml�1 (Fig. 4).

Immediately following the time course studies the hydrogelswere subjected to an amplitude sweep of 0.1–200% strain at a con-stant angular frequency. As expected from the collagen content re-sults above, all hydrogels exhibited strain stiffening behaviour(Fig. 5), with the most marked peaks observed for collagen type I.Both the bDBM and bECM hydrogels exhibited strain stiffening ata lower strain rate than collagen type I for both concentrations.For the 6 mg ml�1 hydrogels the maximum modulus was achievedfor collagen type I at 64% strain, whereas for bDBM and bECM thisoccurred at 20.5% strain (Fig. 5). The maximum storage moduli ofcollagen type I, bDBM and bECM 6 mg ml�1 hydrogels were892.7 ± 127.1, 604.3 ± 133.7 and 274.3 ± 36.3 Pa, respectively(Fig. 5). For the 3 mg ml�1 hydrogels the maximum moduli oc-curred at 29.2% strain for bECM, 43.8% strain for bDBM and 93.6%strain for collagen type I. The maximum modulus of the bECM3 mg ml�1 hydrogel was considerably lower (202.3 ± 14.2 Pa) thanthose of bDBM and collagen (507.3 ± 76.4 and 522.3 ± 82.5 Pa,respectively) (Supplementary Fig. S3).

3.5. Hydrogel morphology

SEM of the surface of bECM, bDBM and collagen type I hydro-gels showed qualitatively that both the bDBM and bECM hydrogelspossessed a randomly oriented fibrillar structure, similar to colla-gen type I (Fig. 6). At both 3 (Fig. 6A, C and E) and 6 mg ml�1

(Fig. 6B, D and F) all three types of hydrogels were nanofibrouswith what appeared to be interconnecting pores. The organizationof the collagen fibres in the bECM (Fig. 6A and B) and bDBM (Fig. 6Cand D) hydrogels appeared visually similar to that of collagen typeI (Fig. 6E and F).

3.6. In vitro cell culture

mPCs were cultured on the surface of 3, 4, 6 and 8 mg ml�1

bECM, bDBM and collagen type I hydrogels. Proliferation of mPCsupon the hydrogels was assessed after 2 and 3 days. Increased pro-liferation of mPCs occurred between days 2 and 3 on bECM gels.Cell number was significantly greater (p < 0.001) on the bECM gelscompared with collagen type I and bDBM hydrogels on day 3. Sim-ilar trends were observed at all concentrations, with the resultsshown in Fig. 7 for 3 mg ml�1 gels. Proliferation of an immortalizedcell line on 4 mg ml�1 hydrogels is included as SupplementaryFig. S4.

4. Discussion

DBM was prepared from bovine bone (bDBM) using an adapta-tion of well-documented acid extraction and lipid removal pro-cesses [33]. The bDBM was then decellularized (bECM) and bothmaterials were solubilized and subsequently induced to form ahydrogel under physiologically relevant conditions (pH, salt con-centration and temperature). The bDBM and bECM hydrogels wereevaluated for structural, mechanical and in vitro cell responsecharacteristics.

Recognition of the deleterious effects in vivo of residual cellularcontent upon constructive remodelling [13] has led to a recent def-inition of minimum criteria for decellularization [42]. The residualDNA content of both the bDBM and bECM materials (Fig. 2C) wasdetermined to be considerably lower than the upper limit of50 ng dsDNA content recommended for complete decellularization[42]. In addition, H&E staining clearly showed the qualitative ab-sence of nuclei in the bECM sections, whereas some nuclei can

Fig. 6. Scanning electron micrographs of (A) bECM gel 3 mg ml�1, (B) bECM gel 6 mg ml�1, (C) bDBM gel 3 mg ml�1, (D) bDBM gel 6 mg ml�1, (E) collagen type I gel 3 mg ml�1

and (F) collagen type I gel 6 mg ml�1. All images 8000� magnification.

Fig. 7. Proliferation of mPCs on Coll 1, bDBM and bECM hydrogels. ⁄, significance between D2 and D3 for bECM; �, significance between bECM and collagen type I at the sametime point; �, significance between bECM and bDBM at the same time point.

7870 M.J. Sawkins et al. / Acta Biomaterialia 9 (2013) 7865–7873

be observed in the bDBM sections. This is in keeping with reportedobservations that DBM may retain a small percentage of cellulardebris [27].

The soluble collagen content of the bDBM and bECM materialswas determined to be very similar (0.93 ± 0.06 and 0.92 ± 0.06 col-lagen mg initial dry weight�1, respectively). This was expected,

M.J. Sawkins et al. / Acta Biomaterialia 9 (2013) 7865–7873 7871

since bDBM (and consequently bECM) material directly derivedfrom bone is known to be a composite of collagens (mostly typeI), non-collagenous proteins and growth factors, residual calciumphosphate and retained cellular debris [27]. The collagenous nat-ure of the bDBM and bECM prompted the use of collagen type Ias a comparator material in the evaluation of the structural,mechanical and in vitro cell response characteristics.

The in vitro self-assembly of collagen monomers into fibrils hasbeen well studied turbidimetrically [43–45], with the mechanismsof fibre network formation clearly elucidated [46]. Turbidity mea-surements employed in this work showed that both bDBM andbECM hydrogels exhibited sigmoidal gelation kinetics consistentwith a nucleation and growth mechanism. Although the compo-nents responsible for gelation are unknown, the high soluble colla-gen content of the material would suggest that gelation is largelydue to the presence of self-assembling collagen molecules. How-ever, the turbidimetric gelation kinetics of both bECM and bDBMhydrogels were slower than collagen type I at the same total pro-tein concentrations. This is most likely due to the presence of gly-cosaminoglycans (GAGs), different types of collagen (III and IV) andother molecules which can modulate collagen self-assembly. It haspreviously been demonstrated that collagen type I and interstitialECM (derived from porcine small intestine submucosa) possess dif-ferent kinetic parameters of assembly, in particular the length ofthe lag phase [45]. A decrease in turbidimetric absorbance andchange in gelation kinetics of collagen type I when mixed withGAGs has also been observed, although the precise rationale forthese changes has not been described [45]. Nonetheless, these find-ings suggest that the gelation behaviour of the bDBM and bECMhydrogels results from complex interactions between the differentcomponents retained in the materials rather than being purely dic-tated by collagen self-assembly.

A complex gelation behaviour of the bDBM and bECM hydrogelswas also indicated by rheological assessments. Although the gela-tion kinetics followed a sigmoidal shape, the storage moduli ofboth bDBM and bECM continued to increase throughout the exper-imental period; two phase gelation occurred at both hydrogel con-centrations. As reported above, this rheological behaviour is likelydue to the presence of other molecules, such as GAGs, which havebeen shown to directly affect gel mechanical properties [47]. Asimilar outcome was reported for a hydrogel derived from urinarybladder matrix (UBM) [39].

Although the storage modulus of the 6 mg ml�1 bECM hydrogel(143.7 ± 7.6 Pa) was similar to that recently reported for a hydrogelderived from decellularized dermal ECM [40], the bECM gels hadsignificantly lower storage moduli than the bDBM or collagen typeI hydrogels at both concentrations. The decellularization processdiscussed herein comprised the use of an enzymatic agent (trypsin)combined with a chelating agent (EDTA). Whilst trypsin is effectivein the removal of cell nuclei even from dense tissues [42], cleavageof proteins such as collagen and elastin and consequent ECM dis-ruption can occur, and this is correlated with changes in themechanical properties [48]. This effect is also evident in the strainstiffening behaviour, where the maximum modulus of the bECMhydrogel was significantly lower than that of bDBM or collagentype I at both concentrations. The maximum modulus for bECMalso occurred at a lower strain rate, indicating possible disruptionof the hydrogel structure.

Biological scaffolds prepared from decellularized tissue havebeen shown to promote and facilitate constructive tissue remodel-ling in pre-clinical studies [6,41]. Whilst the mechanisms of thisphenomenon are not fully elucidated [40], modulation of the hostimmune response, recruitment of endogenous stem and progenitorcells and complete scaffold degradation play important roles. Deg-radation of intact ECM promotes the release of matricryptic mole-cules that possess bioactive properties, including chemoattractant

effects and antimicrobial activity [9]. Additionally, molecules re-leased from in vitro pepsin-degraded and solubilized ECM scaffoldshave been shown to affect the timing and nature of recruitmentand proliferation of appropriate cell types [10]. In the present workwe assessed the mitogenic capacity of the bECM and bDBMhydrogels, using calvarial cells, a relevant cell type for bone. Thein vitro pepsin-digested and solubilized bECM hydrogel induceda higher proliferation rate compared with the similarly treatedbDBM hydrogel or collagen, with cell numbers significantly greater(p < 0.001) on the bECM hydrogels on day 3. DeQuach et al. [49] re-cently reported a similar increased proliferative effect of the degra-dation products of a skeletal muscle ECM hydrogel upon smoothmuscle cells and skeletal myoblasts.

Interestingly, the proliferation of mPCs was not only signifi-cantly increased compared with collagen but also to the pepsin-di-gested and solubilized bDBM hydrogel. We postulate that thepresence of cellular debris in the bDBM material may interferewith the activity of matricryptic molecules. Recent studies of hostremodelling outcomes in response to biological scaffolds with andwithout cellular components may provide insights into this. Rapiddegradation of acellular scaffolds was followed by replacementwith site-appropriate functional host tissue, whereas the presenceof cellular remnants shifted the macrophage polarization profile topredominantly M1, pro-inflammatory phenotype, and was associ-ated with deposition of dense connective tissue [13]. This poorremodelling outcome observed for cellular content in vivo mayalso be potentially caused by reduced proliferative capacityin vitro.

Cellular content may also be associated with the presence of thecell surface a-Gal epitope (galactose-a1,3-galactosyl-b1,4-N-acetylglucosamine) [50]. The a-Gal epitope is naturally produced on gly-colipids and glycoproteins in non-primate mammals, includingpigs and cows [51]. However, it is absent in humans and insteada natural antibody, the anti-Gal antibody is produced. Duringtransplantation of xenografts from pigs to humans anti-Gal anti-body binds to the a-Gal epitope causing graft rejection. The roleof a-Gal in modulating immune responses to xenogeneic ECM ismore ambiguous. ECM materials derived from porcine small intes-tine submucosa contain a-Gal epitopes. In studies in both mice[52] and non-human primates [53] the presence of a-Gal did notadversely affect the immune and remodelling response to SIS-ECM implants. More recently decellularized bovine anterior cruci-ate ligament (ACL) tissues were treated with a-galactosidase to re-move a-Gal epitopes [54]. No significant difference was seenbetween a-galactosidase-treated and decellularized bovine ACLs,which suggests that the decellularization process itself may haveremoved a-Gal epitopes [54] and that the lower number of remain-ing a-Gal epitopes is insufficient to cause an adverse host response.

The mechanical environment of the substrate significantly af-fects in vitro cell behaviour and thus the bECM and bDBM hydrogelstructure and mechanical properties may also influence the cell re-sponse. A recent study characterized the cell infiltration and con-traction of a porcine-derived dermal ECM hydrogel comparedwith a UBM hydrogel [40]. UBM hydrogels possess a lower storagemodulus and larger pore size and are thus more readily infiltratedby fibroblasts at the same ECM concentration. The bECM hydrogelsstudied in this work possess lower moduli than both the collagenand bDBM hydrogels, and this may have contributed to thein vitro cell response. However, fibroblast proliferation has alsobeen shown to increase with increased collagen hydrogel stiffness[55]. It is thus logical to consider that the cellular responses tobECM, bDBM and collagen hydrogels represent the overall effectof hydrogel structure, mechanical properties, constitutive mole-cules and the biological activity of degradation products.

The objectives of this study were to apply a stringent decellular-ization process to DBM, prepared from bovine bone, and to charac-

7872 M.J. Sawkins et al. / Acta Biomaterialia 9 (2013) 7865–7873

terize the structure and composition of the bDBM and resultingbECM materials. To our knowledge this is the first time that anacellular matrix material has been produced from demineralizedbone. Additionally, we have produced hydrogel forms of bDBMand bECM that possess distinct structural, mechanical andbiological characteristics. Rheological characterization demon-strated that the rheological properties varied as a function ofhydrogel concentration, thus ensuring that the properties of thehydrogels can be tailored for specific applications. The long-termobjective of this work is to develop a gel form of DBM/ECM biolog-ical material that retains osteoconductivity and osteoinductivity.We have described the first steps towards this goal through thedevelopment of tissue-specific hydrogel scaffolds.

5. Conclusion

Bone graft substitutes, such as DBM, are usually incorporatedwithin a carrier liquid. However, carrier liquids have been impli-cated in unreliable clinical delivery, including issues relating toinflammation and differences in osteogenic activity. Demineralizedand decellularized bone matrices, prepared from bovine bone, canbe solubilized and induced to form hydrogels. These bDBM andbECM hydrogels have distinct structural, mechanical and biologicalproperties and have the potential for clinical delivery without theinclusion of a carrier. The biological properties of the bECM mate-rial suggest that the constituent molecules released during in vitroscaffold degradation enhance cell proliferation.

Acknowledgements

The authors would like to thank Dr Alexander Huber for assis-tance with the digestion and solubilization, Janet Reing for assis-tance with the DNA quantification and Dr Hassan Rashidi forprovision of the green fluorescent protein labelled mesenchymalstem cells. This research was supported by the EPSRC DoctoralTraining Centre in Regenerative Medicine, the EPSRC Centre forInnovative Manufacturing in Regenerative Medicine (GrantECP038/0512), the Biotechnology and Biological Sciences ResearchCouncil (BBSRC), UK through a LoLa grant (BB/G010617/1) and byfunding from the European Research Council under the EuropeanCommunity’s Seventh Framework Programme (FP7/2007-2013)/ERC grant agreement 227845.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.actbio.2013.04.029.

Appendix B. Figures with essential color discrimination

Certain figures in this article, particularly Figs. 1 and 2, are dif-ficult to interpret in black and white. The full color images can befound in the on-line version, at http://dx.doi.org/10.1016/j.actbio.2013.04.029.

References

[1] Badylak SF. The extracellular matrix as a biologic scaffold material.Biomaterials 2007;28:3587–93.

[2] Badylak SF, Kochupura PV, Cohen IS, Doronin SV, Saltman AE, Gilbert TW, et al.The use of extracellular matrix as an inductive scaffold for the partialreplacement of functional myocardium. Cell Transplant 2006;15:S29–40.

[3] Quarti A, Nardone S, Colaneri M, Santoro G, Pozzi M. Preliminary experience inthe use of an extracellular matrix to repair congenital heart diseases. InteractCardiovasc Thorac Surg 2011;13:569–72.

[4] Badylak SF, Vorp DA, Spievack AR, Simmons-Byrd A, Hanke J, Freytes DO, et al.Esophageal reconstruction with ECM and muscle tissue in a dog model. J SurgRes 2005;128:87–97.

[5] Turner NJ, Yates AJ, Weber DJ, Qureshi IR, Stolz DB, Gilbert TW, et al.Xenogeneic extracellular matrix as an inductive scaffold for regeneration ofa functioning musculotendinous junction. Tissue Eng Part A 2010;16:3309–17.

[6] Valentin JE, Turner NJ, Gilbert TW, Badylak SF. Functional skeletal muscleformation with a biologic scaffold. Biomaterials 2010;31:7475–84.

[7] Badylak SF. The extracellular matrix as a scaffold for tissue reconstruction.Semin Cell Dev Biol 2002;13:377–83.

[8] Wolf MT, Daly KA, Reing JE, Badylak SF. Biologic scaffold composed of skeletalmuscle extracellular matrix. Biomaterials 2012;33:2916–25.

[9] Beattie AJ, Gilbert TW, Guyot JP, Yates AJ, Badylak SF. Chemoattraction ofprogenitor cells by remodeling extracellular matrix scaffolds. Tissue Eng Part A2009;15:1119–25.

[10] Reing JE, Zhang L, Myers-Irvin J, Cordero KE, Freytes DO, Heber-Katz E, et al.Degradation products of extracellular matrix affect cell migration andproliferation. Tissue Eng Part A 2009;15:605–14.

[11] Vorotnikova E, McIntosh D, Dewilde A, Zhang J, Reing JE, Zhang L, et al.Extracellular matrix-derived products modulate endothelial and progenitorcell migration and proliferation in vitro and stimulate regenerative healingin vivo. Matrix Biol. 2010;29:690–700.

[12] Valentin JE, Badylak JS, McCabe GP, Badylak SF. Extracellular matrixbioscaffolds for orthopaedic applications. A comparative histologic study. JBone Joint Surg Am 2006;88:2673–86.

[13] Brown BN, Valentin JE, Stewart-Akers AM, McCabe GP, Badylak SF.Macrophage phenotype and remodeling outcomes in response to biologicscaffolds with and without a cellular component. Biomaterials 2009;30:1482–91.

[14] Sellaro TL, Ranade A, Faulk DM, McCabe GP, Dorko K, Badylak SF, et al.Maintenance of human hepatocyte function in vitro by liver-derivedextracellular matrix gels. Tissue Eng Part A 2010;16:1075–82.

[15] Cortiella J, Niles J, Cantu A, Brettler A, Pham A, Vargas G, et al. Influence ofacellular natural lung matrix on murine embryonic stem cell differentiationand tissue formation. Tissue Eng Part A 2010;16:2565–80.

[16] Petersen TH, Calle EA, Zhao L, Lee EJ, Gui L, Raredon MB, et al. Tissue-engineered lungs for in vivo implantation. Science 2010;329:538–41.

[17] Uygun BE, Soto-Gutierrez A, Yagi H, Izamis M-L, Guzzardi MA, Shulman C, et al.Organ reengineering through development of a transplantable recellularizedliver graft using decellularized liver matrix. Nat Med 2010;16:814–20.

[18] Woolf AD, Pfleger B. Burden of major musculoskeletal conditions. Bull WorldHealth Organ 2003;81:646–56.

[19] Jahangir AA, Nunley RM, Mehta S, Sharan AD. Bone-graft substitutes inorthopaedic surgery. AAOS 2008;2:1.

[20] Evaluation criteria for musculoskeletal and craniofacial tissue engineeringconstructs: a conference report. Tissue Eng Part A 2008;14:2089–2104.

[21] Dinopoulos H, Dimitriou R, Giannoudis PV. Bone graft substitutes: what arethe options? The Surgeon 2012;10:230–9.

[22] Guldberg RE. Spatiotemporal delivery strategies for promotingmusculoskeletal tissue regeneration. J Bone Miner Res 2009;24:1507–11.

[23] Greenwald AS, Boden SD, Goldberg VM, Khan Y, Laurencin CT, Rosier RN. Bone-graft substitutes: facts, fictions, and applications. J Bone Joint Surg 2001;83:S98–S103.

[24] Peterson B, Whang PG, Iglesias R, Wang JC, Lieberman JR. Osteoinductivity ofcommercially available demineralized bone matrix preparations in a spinefusion model. J Bone Joint Surg 2004;86:2243–50.

[25] Wang JC, Alanay A, Mark D, Kanim LE, Campbell PA, Dawson EG, et al. Acomparison of commercially available demineralized bone matrix for spinalfusion. Eur Spine J 2007;16:1233–40.

[26] De Long WG, Einhorn TA, Koval K, McKee M, Smith W, Sanders R, et al. Bonegrafts and bone graft substitutes in orthopaedic trauma surgery: a criticalanalysis. J Bone Joint Surg 2007;89:649–58.

[27] Gruskin E, Doll BA, Futrell FW, Schmitz JP, Hollinger JO. Demineralized bonematrix in bone repair: history and use. Adv Drug Deliv Rev 2012;64:1063–77.

[28] Bostrom MP, Yang X, Kennan M, Sandhu H, Dicarlo E, Lane JM. An unexpectedoutcome during testing of commercially available demineralized bone graftmaterials: how safe are the nonallograft components? Spine (Phila Pa 1976)2001;26:1425–8.

[29] Acarturk TO, Hollinger JO. Commercially available demineralized bone matrixcompositions to regenerate calvarial critical-sized bone defects. Plast ReconstrSurg 2006;118:862–73.

[30] Markel DC, Guthrie ST, Wu B, Song Z, Wooley PH. Characterization of theinflammatory response to four commercial bone graft substitutes using amurine biocompatibility model. J Inflamm Res 2012;5:13–8.

[31] Yazdi M, Bernick S, Paule WJ, Nimni ME. Postmortem degradation ofdemineralized bone matrix osteoinductive potential. Effect of time andstorage temperature. Clin Orthop Relat Res 1991;262:281–5.

[32] Lomas RJ, Gillan HL, Matthews JB, Ingham E, Kearney JN. An evaluation of thecapacity of differently prepared demineralised bone matrices (DBM) and toxicresiduals of ethylene oxide (EtOx) to provoke an inflammatory responsein vitro. Biomaterials 2001;22:913–21.

[33] Pietrzak WS, Ali SN, Chitturi D, Jacob M, Woodell-May JE. BMP depletionoccurs during prolonged acid demineralization of bone: characterization andimplications for graft preparation. Cell Tissue Bank 2011;12:81–8.

M.J. Sawkins et al. / Acta Biomaterialia 9 (2013) 7865–7873 7873

[34] Schenke-Layland K, Vasilevski O, Opitz F, König K, Riemann I, Halbhuber KJ,et al. Impact of decellularization of xenogeneic tissue on extracellular matrixintegrity for tissue engineering of heart valves. J Struct Biol 2003;143:201–8.

[35] Hong Y, Huber A, Takanari K, Amoroso NJ, Hashizume R, Badylak SF, et al.Mechanical properties and in vivo behavior of a biodegradable syntheticpolymer microfiber – extracellular matrix hydrogel biohybrid scaffold.Biomaterials 2011;32:3387–94.

[36] Gilbert TW, Freund JM, Badylak SF. Quantification of DNA in biologic scaffoldmaterials. J Surg Res 2009;152:135–9.

[37] Keane TJ, Londono R, Turner NJ, Badylak SF. Consequences of ineffectivedecellularization of biologic scaffolds on the host response. Biomaterials2012;33:1771–81.

[38] Woessner JF. The determination of hydroxyproline in tissue and proteinsamples containing small proportions of this imino acid. Arch BiochemBiophys 1961;93:440–7.

[39] Freytes DO, Martin J, Velankar SS, Lee AS, Badylak SF. Preparation andrheological characterization of a gel form of the porcine urinary bladdermatrix. Biomaterials 2008;29:1630–7.

[40] Wolf MT, Daly KA, Brennan-Pierce EP, Johnson SA, Carruthers CA, D’Amore A,et al. A hydrogel derived from decellularized dermal extracellular matrix.Biomaterials 2012;33:7028–38.

[41] Reing JE, Brown BN, Daly KA, Freund JM, Gilbert TW, Hsiong SX, et al. Theeffects of processing methods upon mechanical and biologic properties ofporcine dermal extracellular matrix scaffolds. Biomaterials 2010;31:8626–33.

[42] Crapo PM, Gilbert TW, Badylak SF. An overview of tissue and whole organdecellularization processes. Biomaterials 2011;32:3233–43.

[43] Wood GC. The formation of fibrils from collagen solutions. 2. A mechanism ofcollagen-fibril formation. Biochem J 1960;75:598–605.

[44] Wood GC, Keech MK. The formation of fibrils from collagen solutions. 1. Theeffect of experimental conditions: kinetic and electron-microscope studies.Biochem J 1960;75:588–98.

[45] Brightman AO, Rajwa BP, Sturgis JE, McCallister ME, Robinson JP, Voytik-Harbin SL. Time-lapse confocal reflection microscopy of collagen

fibrillogenesis and extracellular matrix assembly in vitro. Biopolymers 2000;54:222–34.

[46] Y.-l. Yang, Kaufman LJ. Rheology and confocal reflectance microscopy asprobes of mechanical properties and structure during collagen and collagen/hyaluronan self-assembly. Biophys J 2009;96:1566–85.

[47] Stuart K, Panitch A. Influence of chondroitin sulfate on collagen gel structureand mechanical properties at physiologically relevant levels. Biopolymers2008;89:841–51.

[48] Yang M, Chen CZ, Wang XN, Zhu YB, Gu YJ. Favorable effects of the detergentand enzyme extraction method for preparing decellularized bovinepericardium scaffold for tissue engineered heart valves. J Biomed Mater ResB Appl Biomater 2009;91:354–61.

[49] DeQuach JA, Lin JE, Cam C, Hu D, Salvatore MA, Sheikh F, et al. Injectableskeletal muscle matrix hydrogel promotes neovascularization and muscle cellinfiltration in a hindlimb ischemia model. Eur Cell Mater 2012;23:400–12.

[50] Badylak SF, Gilbert TW. Immune response to biologic scaffold materials. SeminImmunol 2008;20:109–16.

[51] Galili U. The [alpha]-gal epitope and the anti-Gal antibody inxenotransplantation and in cancer immunotherapy. Immunol Cell Biol 2005;83:674–86.

[52] Raeder RH, Badylak SF, Sheehan C, Kallakury B, Metzger DW. Natural anti-galactose a1,3 galactose antibodies delay, but do not prevent the acceptance ofextracellular matrix xenografts. Transpl Immunol 2002;10:15–24.

[53] Daly KA, Stewart-Akers AM, Hara H, Ezzelarab M, Long C, Cordero K, et al.Effect of the aGal epitope on the response to small intestinal submucosaextracellular matrix in a nonhuman primate model. Tissue Eng Part A 2009;15:3877–88.

[54] Yoshida R, Vavken P, Murray MM. Decellularization of bovine anterior cruciateligament tissues minimizes immunogenic reactions to alpha-gal epitopes byhuman peripheral blood mononuclear cells. Knee 2012;19:672–5.

[55] Hadjipanayi E, Mudera V, Brown RA. Close dependence of fibroblastproliferation on collagen scaffold matrix stiffness. J Tissue Eng Regen Med2009;3:77–84.