d e n t a l m a t e r i a l s 3 4 ( 2 0 1 8 ) 538–550

Available online at www.sciencedirect.com

ScienceDirect

jo ur nal ho me pag e: www.int l .e lsev ierhea l th .com/ journa ls /dema

Development of layered PLGA membranes forperiodontal tissue regeneration

Itsumi Yoshimotoa, Jun-Ichi Sasakia,∗, Ririko Tsuboia,b,Satoshi Yamaguchia, Haruaki Kitagawaa, Satoshi Imazatoa

a Department of Biomaterials Science, Osaka University Graduate School of Dentistry, 1-8 Yamadaoka, Suita, Osaka565-0871, Japanb Division for Interdisciplinary Dentistry, Osaka University Dental Hospital, 1-8 Yamadaoka, Suita, Osaka565-0871, Japan

a r t i c l e i n f o

Article history:

Received 3 October 2017

Received in revised form

30 November 2017

Accepted 21 December 2017

Keywords:

Biodegradable membrane

Barrier membrane

Guided tissue regeneration

PLGA

Bone regeneration

a b s t r a c t

Objective. Various commercial products are available for guided tissue regeneration (GTR)

therapy; however, they do not combine biosafety with the ability to control cell function.

The purpose of this study was to evaluate the physicochemical and biological charac-

teristics of the novel bilayer biodegradable poly(lactic-co-glycolic acid) (PLGA) membrane,

and to assess whether the bilayer PLGA membrane could be used for periodontal tissue

regeneration.

Methods. Bilayer biodegradable membrane was fabricated thorough a two-step freezing

and lyophilization process using PLGA solution. The characteristics of bilayer mem-

branes were evaluated with respect to surface morphology, stability, mechanical strength,

and operability for clinical use. Cell proliferation and osteogenic differentiation were

investigated on the each surface of bilayer membrane. Then, these membranes were

implanted to the rat calvaria bone defect models and evaluated their capability for tissue

regeneration.

Results. Biodegradable membranes composed of the solid and porous layer were successfully

prepared and the surface morphologies analyzed. Physicochemical analyses revealed that

the membranes possessed enough stability and mechanical properties for clinical use. It was

also confirmed that the solid layer inhibited cell proliferation and subsequent connective

tissue invasion, while the inner layer promoted proliferation and osteogenic differentiation,

thus resulting in bone regeneration in vivo.

Significance. The layering technology used to fabricate the bilayer polymer membrane could

be applied in the developing of other novel biomaterials. The present study demonstrates

that the bilayer biodegradable polymer membranes facilitate tissue regeneration in vivo, and

therefore represent a prospective biomaterial for GTR therapy.

© 2018 The Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved.

∗ Corresponding author.E-mail address: jun1-s@dent.osaka-u.ac.jp (J.-I. Sasaki) .

https://doi.org/10.1016/j.dental.2017.12.0110109-5641/© 2018 The Academy of Dental Materials. Published by Elsev

ier Ltd. All rights reserved.

3 4

1

GfloTt[(

srracib[iamcbamabri

letpWpe[tcs

orrpgtbdcp[e

aet

d e n t a l m a t e r i a l s

. Introduction

uided tissue regeneration (GTR) is a clinical approachor periodontal tissue regeneration by guiding periodontaligament-derived cells and osteoblasts following applicationf a barrier membrane to the periodontal tissue defect [1,2].he membranes used in GTR therapy are classified into two

ypes: biodegradable (including animal origin constituents3–5] and polylactic acid (PLA) [6–8]) and non-biodegradableincluding polytetrafluoroethylene (PTFE) and titanium) [9–11].

Recently, the use of biodegradable membranes is main-tream in GTR therapy [12,13], because secondary surgery toemove the remaining membranes is unfavorable after tissueegeneration [14,15]. Biodegradable membranes composed ofnimal collagen, PLA, polyglycolic acid (PGA), and poly(lactic-o-glycolic acid) (PLGA) are commercially available and aren widespread dental clinical use. Collagen-derived mem-ranes possess high biocompatibility and biodegradability

16,17]; however, they might contain infectious agents follow-ng their generation from an unspecified number of swinend cows [18,19]. Furthermore, collagen membranes have lowechanical strength [20] and their degradation rate is diffi-

ult to control [21,22]. By contrast, membranes composed ofiodegradable polymer are superior in biodegradability [23,24]nd mechanical property [25]. In addition, biodegradable poly-er membranes are eventually degraded into carbon dioxide

nd water, therefore they are believed to be safe within theody. Despite this, biodegradable polymer membranes areeported to have low cytocompatibility [24,26,27] and operabil-ty for clinical use [28].

Clinical reports indicate that GTR therapy remains prob-ematic for periodontal regeneration. For example, Tannert al.[29] reported that collagen membranes are insufficiento block epithelial tissue invasion, and consequently induceeriodontal recovery with non-native epithelial attachment.hen applying PTFE membranes, some reports show that

eriodontal tissue regeneration is impeded by penetratingpithelial cells between the membrane and regeneration site30,31]. Thus, it is believed that selective control capability,o block tissue invasion at the epithelial side and promoteell proliferation and differentiation at the tissue regenerationite, is necessary for biodegradable polymer membranes.

A number of studies have investigated cellular behaviorn biodegradable polymers [32–34]. Osteoblast attachment iseported to be good on PLGA nanofibers but not on solid mate-ials [32]. In addition, porous PLA films are reported to promoteroliferation of fibroblasts while a smooth surface inhibits cellrowth [33]. A study using epithelial 293T cells showed thatight PLA/PLGA surfaces inhibit cell invasion into the mem-rane [34]. Based on these findings, bilayer membranes withiffering surface morphologies are thought to enable betterontrol of cell function, as well as tissue behavior. In fact, arevious report has proposed the use of bilayer membranes

35]; however the effect of the bilayer structure on tissue regen-ration remains unclear.

In this study, we fabricate a bilayer membrane using PLGA,

commonly used and well known biomaterial [36–38], andvaluate its usability as a bilayer PLGA membrane for GTRherapy in vitro and in vivo. We also tested the null hypothesis:

( 2 0 1 8 ) 538–550 539

that there would be no difference in tissue regeneration usingvarious thicknesses of each layer. The purpose of this studywas to evaluate the physicochemical and biological character-istics of bilayer biodegradable PLGA membrane, and to assesswhether the membrane could be used for periodontal tissueregeneration.

2. Materials and methods

2.1. Preparation of bilayer PLGA membrane

PLGA (PLA:PLGA = 75:25; Molecular weight: 25 kDa) was dis-solved in 1,4-dioxane (Wako, Osaka, Japan) at 2.7 wt%. Twotypes of layered-membrane (LM-1 and LM-2) were preparedas follows: PLGA solution was poured into the fluoroplastic-coated mold. Then, the base of the plate was cooled toeither −30 ◦C (LM-1) or −80 ◦C (LM-2) for 10 s and the PLGAsolution was frozen partially. The mold was then coveredwith a brass plate (cooled to −80 ◦C) and the PLGA solutionwas slowly and completely frozen. The frozen PLGA solutionwas then lyophilized using a freeze-dryer (FDU-1110, EYELA,Tokyo, Japan), and then formed as membrane by pressing at300 kgf/cm2 for 15 s using a hydraulic press machine (JP-1T,Nissin Kagaku, Osaka, Japan). The obtained PLGA membraneswere then treated with gamma sterilization and kept in anitrogen atmosphere at 4 ◦C prior to experimentation.

The solid layer formed by prior freezing and its oppos-ing side were referred as the outer layer and internal layer,respectively. Commercial PLGA membrane (GC membrane, GC,Tokyo, Japan), which was the same in composition to bilayermembrane fabricated, was used as a control.

2.2. Characteristics of the bilayer PLGA membrane

2.2.1. Surface structureThe surface and cross-sectional surface of each membranewere coated with gold and observed using a scanning elec-tron microscope (SEM; JSM-6390, JEOL, Tokyo, Japan) at 5 kV.Surface morphology and roughness were evaluated using alaser microscope (VK-X250, Keyence, Osaka, Japan). Rough-ness average (Ra) was calculated from the three areas ofarbitrary surface (n = 7).

2.2.2. Stability analysesTo evaluate the water contact angle, each membrane wascut into 5.0 × 5.0 mm samples. Ten microliters of phosphate-buffered saline (PBS; Nissui, Tokyo, Japan) was then droppedonto the membrane, and a picture was immediately takenusing a single-lens reflex camera (D5500, Nikon, Tokyo, Japan)in a horizontal direction. Contact angle was measured as thetangent line from the end of the droplet (n = 5).

To investigate dimensional change, each membrane was◦

cut into 10 × 10 mm samples and immersed into PBS at 37 C.

After 1, 3, 7, and 21 days, margin length and thickness weremeasured using a digital caliper (CD-15CP, Mitutoyo, Kana-gawa, Japan) (n = 4).

l s 3

540 d e n t a l m a t e r i a

2.2.3. Mechanical strengthMembrane mechanical strength was evaluated by conven-tional tensile test and suture pull-out test using a tensile tester(EZ-Test, Shimadzu, Kyoto, Japan). For the tensile test, eachmembrane was cut into 20 × 3.0 mm samples and clamped3.0 mm across both ends of the long side. At room temper-ature, the uniaxial tensile force was loaded with a cross-headspeed of 1.0 mm/min until failure (n = 14).

For the suture pull-out test, each membrane was cutinto 40 × 10 mm samples. Using a grasping jig, membraneswere clamped 5.0 mm from one end, and surgical sutures(6-0.VICRYL, Ethicon, Somerville, NJ, USA) were passed5.0 mm from the center of the short side of the otherend. The uniaxial tensile force was applied to the sutureas with the conventional tensile test (n = 8). Tensile/pull-out strength and breaking strain were calculated asfollows:

Tensile/pull-out strength (N/mm2)

= maximal load (N)/sectional area (mm2)

Breaking strain (%) = maximal stroke (mm)/sample span

Fig. 1 – Surface (A) and sectional (B) morphology were observed bbilayer membranes. Scale bar: 50 �m. (B) Arrows indicate solid (uthat the solid layer of LM-2 is thicker than that of LM-1. Asterisk

4 ( 2 0 1 8 ) 538–550

2.2.4. Compatibility assessmentMembrane compatibility to the model was measured for eval-uating operability in clinical use. An epoxy model comprisingfive landmarks was prepared using a three-dimensionalprinter (EDEN260, Stratasys, Eden Prairie, MN, USA), in accor-dance to the periodontal disease model (x-444, Nissin, Kyoto,Japan). Next, the membranes were applied to the epoxy modeland sewn with surgical suture (6-0.VICRYL) 2.0 mm from bothends of the membrane (Supplemental Fig. 1A and B).

The model with membrane was then observed by micro-computed tomography (micro-CT; R mCT2, Rigaku, Tokyo,Japan) across a 10 × 10 mm scanning field with 20 �m of res-olution. The obtained images were overlapped according tothe landmarks, and nine measurement points were indi-cated, each 1.0 mm apart. The shortest distance was measuredfor each measurement point to indicate compatibility of themembrane (n = 3) (Supplemental Fig. 1C and D).

2.3. Cell behavior on the bilayer PLGA membrane

2.3.1. Cell cultureHuman and mouse bone marrow-derived mesenchymal stemcells (hBMSC and mBMSC), and mouse fibroblasts (L-929)

y SEM. (A) Outer (Out) and inner (In) surfaces are shown forpper) and porous (lower) layers. Sectional images show

s indicate membrane surface. Scale bar: 100 �m.

3 4

w(isciciSaci

2Mw

Ftva

d e n t a l m a t e r i a l s

ere used. These cells were provided by the Riken cell bankRiken, Saitama, Japan) and were cultured in Dulbecco’s mod-fied Eagle’s medium (Wako) containing 10% fetal bovineerum (Japan Bioserum, Hiroshima, Japan) and 1% peni-illin/streptomycin (Sigma Aldrich, St. Louis, MO, USA). Tonduce osteogenic differentiation of the BMSCs, cells wereultured in the growth medium described above contain-ng �-glycerophosphate disodium salt hydrate (1 × 10−2 mol/L;igma-Aldrich), ascorbic acid (50 �g/mL; Sigma-Aldrich), dex-methasone (1 × 10−6 mol/L; Sigma-Aldrich), and 10 mM ofalcium chloride solution. Cells were maintained in a humid-fied incubator at 37 ◦C with 5% CO2.

.3.2. SEM observationembranes (10 × 10 mm) were placed into wells of a 12-ell culture plate (Iwaki, Tokyo, Japan). For LM-1 and LM-2,

ig. 2 – (A) Laser microscopic images of membranes. The outer sheir inner surfaces showed various-sized asperities. (B) Surfacealues of LM-1 and LM-2 were significantly higher than that of thmong groups (p < 0.05, mean ± SD, n = 7).

( 2 0 1 8 ) 538–550 541

the samples were prepared with facing the internal andouter layer on top, respectively. L-929 cells were seededonto the membranes at 5.0 × 103 cells/well and cultured for1, 3, 7, and 12 days. Subsequently, samples were washedwith PBS, fixed with 4% paraformaldehyde for 24 h andthen treated with phased-dehydration. Cell-seeded mem-branes were lyophilized (JFD-320, JEOL), coated with gold, andobserved by SEM (JSM-6390) at 5 kV.

2.3.3. Cell proliferationL-929 cells and hBMSCs were cultured on the membranes aspreviously described, and cell numbers were measured using

the Cell Counting Kit-8 assay (Doujin Kagaku, Kumamoto,Japan) according to the manufacturer’s protocol. Briefly, cell-seeded membranes were washed twice with PBS and exposedto 1.0 mL of aqueous tetrazolium (WST-8) at 37 ◦C for 90 min.

urfaces of LM-1 and LM-2 had small bumps. By contrast, roughness of membranes. The inner surface roughnesse control. Same letters indicate no significant difference

l s 3

542 d e n t a l m a t e r i a

Optical density was quantified at 450 nm using a microplatereader (ARVO MX, PerkinElmer, Waltham, MA, USA) (n = 4).

2.3.4. Osteogenic differentiationTo investigate the induction efficiency of BMSCs towards theosteogenic lineage, mBMSCs and hBMSCs were cultured on themembranes in osteogenic medium. The membrane sampleswere prepared as same manner with that described in SEMobservation of L-929. The cells were seeded onto the mem-branes at 2.0 × 104 cells/well and cultured for 21 days. Thesamples were washed with PBS, fixed with 4% paraformalde-hyde, then washed with distilled water. Subsequently, vonKossa staining was carried out by immersing the samples into5% silver nitrate aqueous solution for 30 min under ultravi-olet light. Stained mineralized matrices on the membraneswere visualized and captured by a CCD camera (DS-Fi2, Nikon)equipped with a stereoscopic microscope (SMZ745T, Nikon)(n = 3).

2.4. In vivo implantation of the bilayer PLGAmembrane

2.4.1. Rat calvaria bone defect modelAll animal experiments followed a strict protocol approved bythe Institutional Animal Care and Use Committee of OsakaUniversity Graduate School of Dentistry (approval number: 26-021-0). Sixteen 10-week-old male Sprague-Dawley rats (CLEAJapan Inc., Tokyo, Japan) were used. After the induction ofgeneral anesthesia, the top of head was shaved, and a skin

incision was performed using a scalpel. A 5 mm-diameterbone defect was created using trephine bar (Micro Tech. Inc.,Tokyo, Japan) on each side of the sagittal suture, which wasthen covered by the circularly-cut membrane (7.0 mm in diam-

Fig. 3 – (A) Contact angles of membranes for PBS. The inner surfacontrol. Same letters indicate no significant difference among grlength (upper panel) and thickness (lower panel) in PBS were obs

4 ( 2 0 1 8 ) 538–550

eter) (n = 4). Bone defects without membrane were prepared asnegative controls (sham). Subsequently, the periosteum andepidermis were closed with 6-0 VICRYL and 4-0 silk sutures(Alfresa Pharma, Osaka, Japan), respectively. Thereafter, therats were housed with free access to water and food underspecific pathogen-free conditions.

2.4.2. Micro-CT and histological analysesAt 4 and 8 weeks after implantation, the rats were eutha-nized and their calvariae were harvested and fixed with 2.5%glutaraldehyde for 24 h. The calvariae were scanned by micro-CT (R mCT2) within a 20 × 20 mm scanning field and 40 �mresolution. Bone areas were identified according to the CT val-ues and phantom control of hydroxyapatite (200–800 mg/cm3).The volume ratio of regenerated bone within the defect sitewas measured using bone-specific image analysis software(TRI/3D-BON, RATOK, Tokyo, Japan).

Fixed rat calvariae, that had been observed by micro-CT,were decalcified with 10% ethylenediaminetetraacetic acid(Wako) for 30 days. The specimens were embedded in paraffinand sectioned at 9-�m thickness for hematoxylin and eosin(HE) staining. The stained sections were observed using a CCDcamera (DS-Fi2) equipped with a light microscope (ECLIPSECI-L, Nikon).

2.4.3. Biodegradability of bilayer membranesTo evaluate biodegradability in vivo, the membranes were cutinto 10 × 10 mm samples. The membranes were then subcu-taneously implanted into the back of the same rats used in

the bone defect model previously described. At 4 and 8 weeksafter implantation (at the same time as calvariae harvesting),implanted membranes were excised, washed with PBS, anddried at room temperature. The membranes were dissolved in

ce contact angles of LM-1 and LM-2 were greater than theoups (p < 0.05, mean ± SD, n = 5). (B) The alteration of marginerved for 21 days (mean ± SD, n = 4).

3 4

c(wTS

2

Onttp

3

3

A2p

Fcnsassm

d e n t a l m a t e r i a l s

hloroform at 0.6 wt% and filtered through a 0.2 �m mesh filterIWAKI) to remove impurities. The molecular weight of PLGAas measured using a separation column (GPC-104, Shodex,

okyo, Japan) equipped with liquid chromatography (LF-604,hodex) (n = 4).

.5. Statistical analysis

ne-way analysis of variance (ANOVA) with a Tukey’s or Dun-ett’s post hoc test was used for comparisons of more than

wo groups. Student’s t-test was used for comparisons ofwo groups. A significant difference was defined for values of

< 0.05.

. Results

.1. Surface structure of bilayer PLGA membrane

s expected, SEM observation showed that LM-1 and LM- presented a solid morphology at the outer surface and aorous morphology at the inner surface (Fig. 1A). Sectional

ig. 4 – Tensile test (A) and suture pull-out test (B) werearried out on the membranes (*: p < 0.05, mean ± SD, A:

= 14, B: n = 8). Left graphs show (A) tensile and (B) pull-outtrengths, and right graphs show the breaking strain. LM-1nd LM-2 had lower mechanical strength and breakingtrain compared with the control. The suture pull-out testhowed no differences in breaking strain among allembranes.

( 2 0 1 8 ) 538–550 543

SEM images revealed that the solid LM-2 layer was thicker thanthat of LM-1 (Fig. 1B). Comparatively, the control membraneshowed a solid surface and uniform layer. Laser microscopyimages of the surface morphology are shown in Fig. 2A. Theouter surfaces of LM-1 and LM-2 had small bumps. By con-trast, their inner surfaces showed various-sized asperities.The inner surface roughness values of LM-1 and LM-2 were3.84 ± 0.54 �m and 3.12 ± 0.40 �m, respectively, which weresignificantly higher than that of the control (Fig. 2B). The outersurface roughness values of LM-1 and LM-2 were significantlylower than that of the control. There were no significant dif-ferences between LM-1 and LM-2.

3.2. Physicochemical property of bilayer membrane

The inner surface contact angles of LM-1 and LM-2 were117.5 ± 4.6◦ and 107.5 ± 1.5◦ in LM-1 and LM-2, which weregreater than the control (107.5 ± 1.5◦) (Fig. 3A). By contrast,the outer surface contact angles were significantly lower com-pared to the control.

For all membranes, margin length shrunk following immer-sion in PBS for 1 day, but length remained unchanged over asubsequent 21 days (Fig. 3B). Meanwhile, LM-1 and LM-2 thick-ness increased by 60–80 �m after 1 day, then decreased fromday 3 to 7, and was maintained thereafter. Comparatively,control membrane thickness gradually increased throughout

the observation period. There were no significant differencesbetween LM-1 and LM-2 with respect to contact angles anddimensional changes.

Fig. 5 – Membrane compatibility was evaluated bymicro-CT. (A) Between measuring points (8 and 9) andmembrane were connected with straight lines. Scale bar:5 mm. (B) Distances of gaps were measured in eachmeasuring point. LM-1 and LM-2 displayed a shorterdistance comparing with the control, indicative of bettercompatibility in bilayer membranes (*: p < 0.05, mean ± SD,n = 3).

544 d e n t a l m a t e r i a l s 3 4 ( 2 0 1 8 ) 538–550

Fig. 6 – (A) SEM images of membranes cultured with L-929 cells. The outer surface of LM-1 was altered in roughness by day12 of cell culture. The surface morphology of LM-2 remained unchanged throughout the observation period. Arrows indicateattached cells on the membrane. Scale bar: 100 �m. (B) Proliferation of L-929 (left graph) and hBMSC (right graph) weremeasured on each membrane surface. At day 12, the number of cells was significantly increased on both LM-1 surfaces andthe inner surface of LM-2 (*: p < 0.05, mean ± SD, n = 4).

3 4 ( 2 0 1 8 ) 538–550 545

hwetab

3

Rdsdtd

3

Tbaicnfom(

1mr

dddLSL

3m

Rwmttl(hgLwu(a

4

Fig. 7 – Deposition of cell-derived mineralized matriceswere visualized by von Kossa staining. hBMSC (left panels)and mBMSC (right panels) were cultured on eachmembrane surface for 21 days. Mineralized matrixdeposition was enhanced on the both surfaces of LM-1 andinner surface of LM-2 compared with the outer surface of

d e n t a l m a t e r i a l s

The conventional tensile test showed that LM-1 and LM-2ad lower mechanical strength and breaking strain comparedith the control (Fig. 4A). There were no significant differ-

nces between LM-1 and LM-2 in this test. The suture pull-outest showed a significantly lower pull-out strength for LM-1nd LM-2 compared with the control, with no differences inreaking strain among all membranes (Fig. 4B).

.3. Compatibility of bilayer PLGA membrane

epresentative images (points 8 and 9) of micro-CT and theistance from these measuring points to the membrane arehown in Fig. 5. Compared with the control, LM-1 and LM-2isplayed a shorter distance at the majority of point, indica-ive of better compatibility (Fig. 5B). There were no significantifferences between LM-1 and LM-2.

.4. Cell behaviors on bilayer PLGA membrane

he SEM images and MTT assay of L-929 cell-seeded mem-ranes observed no differences in cell numbers and extensionmong all surfaces at days 1 and 3 (Fig. 6). At day 7, the LM-1nner surface had a greater number of attached and extendedells compared with LM-2 and the control. At day 12, theumber of cells was significantly increased on both LM-1 sur-

aces and the inner surface of LM-2; however the outer surfacef LM-2 had only a few rounded cells. A similar cell attach-ent and growth trend was observed at day 12 using hBMSCs

Fig. 6B).The outer surface of LM-1 was altered in roughness by day

2 of cell culture (Fig. 6A). Meanwhile the inner surface of LM-1aintained its morphology. The surface morphology of LM-2

emained unchanged throughout the observation period.hBMSC and mBMSC were induced to undergo osteogenic

ifferentiation on each membrane and mineralization wasetermined by von Kossa staining (Fig. 7). Mineralized matrixeposition was enhanced on the inner surfaces of LM-1 andM-2 compared with the outer surface of LM-2 and the control.taining intensity was comparable between both surfaces ofM-1 using both cell types.

.5. In vivo efficacy evaluation of bilayer PLGAembrane

epresentative micro-CT images of rat calvaria with orithout membranes are shown in Fig. 8A. New bone for-ation was observed from the margin of the defect site in

he membrane-implanted groups at week 4 after implan-ation. The bone defect without membrane (sham) had aower volume ratio of bone formation than the other groupsFig. 8B). At week 8 after implantation, the sham groupad minimal new bone formation, while the membraneroups had enhanced bone regeneration (Fig. 8A). Of note,M-1 and LM-2 regenerated distinct bone that covered thehole defect site, and both had a significantly increased vol-me ratio of newly formed bone compared with the control

Fig. 8B). There was no significant difference between LM-1nd LM-2.

Histological images of the implantation groups at week post-implantation showed that solid contact was made

LM-2 and the control.

between the inner surface of the membrane and the regen-erated tissue, with newly formed bone just below themembranes (Fig. 9A). A number of cells had infiltrated intothe control membrane from both sides (Fig. 9B). Connectivetissues had migrated into the inner layer of LM-1 and par-tially into the outer layer. However, only the inner layer ofLM-2 showed a similar level of connective tissue infiltrationas LM-1, with only cells attached to the outer surface (Fig. 9B).

At week 8 post-implantation, newly formed bone grew ver-tically in the LM-1 and LM-2 implanted groups (Fig. 9C). Inthe control group, both connective tissue and new bone was

546 d e n t a l m a t e r i a l s 3 4 ( 2 0 1 8 ) 538–550

Fig. 8 – (A) Bone defect models with or without (sham) membrane were scanned by micro-CT at 4 and 8 weekspost-implantation. LM-1 and LM-2 regenerated distinct bone that covered the whole defect site after 8 weeks ofimplantation. Scale bar: 1 mm. (B) Volume ratios of regenerated bone were quantitatively assessed in each group. At week 8after implantation, LM-1 and LM-2 had a significantly increased volume ratio of newly formed bone compared with thecontrol and sham. Same letters indicate no significant difference among groups (p < 0.05, mean ± SD, n = 4).

observed within the membrane from each surface (Fig. 9D). Inthe LM-1 implanted group, connective tissue infiltrated bothlayers and the membrane had partially collapsed, indicatingsome level of biodegradation. A single layer of cells was incontact with the outer surface of LM-2, but connective tis-sue did not infiltrate into the membrane, even at week 8 post-implantation.

The molecular weight of the implanted membranesdecreased over time following biodegradation (SupplementalFig. 2). There were no differences in molecular weight amonggroups at 4 and 8 weeks post-implantation. This result indi-cates that bilayer and control membranes showed similarbiodegradation behaviors in vivo.

4. Discussion

A bilayer PLGA membrane with a porous (inner) and solid(outer) layer was successfully fabricated using a two-step

freezing and lyophilization process. By altering the crystal sizeof 1,4-dioxan according to the freezing time, the bilayer mem-brane was formed. During the prior quick freezing step, tinycrystals of solvent formed, resulting in the solid PLGA layer,

while slow-freezing produced the porous layer. Furthermore,the thickness of each layer was controlled by applying variouschill temperatures to the plate. We believe this novel methodwill permit fabrication of multi-layered polymer constructsfrom a single solution in the future.

With respect to membrane characterization, the inner sur-face of LM-2 had smaller cavities compared with that of LM-1.The temperature reduction of whole PLGA solution was pro-duced by prior freezing with −80 ◦C (LM-2) more than −30 ◦C(LM-1), and clearly affected the subsequent freezing process.The inner surface of the bilayer membranes had a higher con-tact angle for PBS compared with the outer surface and thecontrol. Contact angle is known to increase with increasedsurface roughness on hydrophobic surfaces, and decreasingwith increased roughness on hydrophilic surfaces [39]. Thisis therefore a valid result because all the membranes used inthis study were comprised of a hydrophobic substrate.

Biodegradable materials have accelerated hydrolysis inthe presence of cells [40] and their enzymatic activities [41].

In addition, the ratio of hydrolysis changes depending onthe polymers molecular weight and structure [42,43]. Weevaluated membrane stability without the effect of hydrol-

d e n t a l m a t e r i a l s 3 4 ( 2 0 1 8 ) 538–550 547

Fig. 9 – HE-stained images of regenerated bone at (A, B) 4 weeks and (C, D) 8 weeks post membrane implantation. (A, C)Entire and (B, D) magnified images of rat calvaria. (A, B) Connective tissues had migrated into the both layers of LM-1;however only the inner layer of LM-2 showed connective tissue infiltration as LM-1, with only cells attached to the outersurface. (C, D) In the LM-1 implanted group, connective tissue infiltrated both layers and the membrane had partiallycollapsed. In contrast, a single layer of cells was in contact with the outer surface of LM-2, but connective tissue did notinfiltrate into the membrane. Arrows indicate edges of the prepared bone defects. Scale bar: (A, C) 1 mm, (B, D) 500 �m. m:membrane; nb: new bone; st: soft tissue.

l s 3

r

548 d e n t a l m a t e r i a

ysis. Under these conditions, membrane margin length wasdecreased and thickness was increased in 24 h, and theirdimensions were maintained thereafter. Holy et al. [44]reported that PLGA swells slightly in an aqueous environment,and further swelling is induced by liquid infiltrating into theporous PLGA. In the present study, it was considered that waterinfiltration induced augmentation in thickness of the mem-branes. Interestingly, margin length of the bilayer membranesdecreased compared with the controls. This was thought toresult from the release of internal stress that was generated bycompression after lyophilization. These results indicate thatthe fabricated bilayer membranes were stable in vivo.

In the clinical situation, GTR membranes are applied bycutting into an appropriate size and suturing around thedamaged tooth. Therefore, membranes require good mechan-ical properties for these procedures [45]. The mechanical testrevealed that the bilayer membranes had a lower mechani-cal intensity compared with control. However, no significantdifference was observed between the bilayer membranes andcommercial product with respect to breaking strain, furthersupporting their clinical use. We next evaluated compatibil-ity of the membranes with respect to ease of implantationand in maintaining their intended shape. The bilayer mem-branes resulted in a reduced distance between the membraneand model. In fact, the bilayer membranes were able to obtainhigh border sealing between the membrane and periodontaltissues, clarifying that the bilayer membranes possessed highoperability compared with the commercial one.

Cell number remained unchanged among the tested groupsat days 1 and 3 of culture. Huang et al. [46] used polyvinylalcohol (PVA) and reported that there was no significant dif-ference in the number of fibroblasts between solid and porousscaffolds after 1 and 3 days of culture. Gupta et al. [47] alsoprepared PVA scaffolds and showed that rat Schwann cell pro-liferation was not promoted on the rough surface until day 3,with incremented cell number after 6 days of culture. Thesestudies support our findings that surface roughness did notaffect cell attachment and short periods of proliferation onthe polymer scaffolds. The outer surface of LM-1, originallythe solid layer, appeared rough after cell culture for 12 days.It was considered that the thin outer layer was hydrolyzedby cellular activities, exposing the rough inner layer that inturn promoted cell proliferation. Outer layer of LM-2 was alsodegraded, but it was remained solid layer due to the thick-ness even after 12 days of cell culture. Mineralized depositionby BMSCs was observed to a greater extent on both LM-1 sur-faces and the inner LM-2 surface. Sonomoto et al. [48] reportthat hBMSCs have enhanced extracellular matrix (ECM) pro-duction on rough porous PLGA scaffolds compared with on asmooth surface. In addition, they show that osteogenic differ-entiation marker expression is higher on porous scaffolds. It istherefore likely that the rough bilayer membrane surface facil-itated cell proliferation as well as osteogenic differentiationof BMSCs. Considering the cell control, LM-2 could be morefavorable for long term application of GTR therapy, thereforeour null hypothesis was rejected.

Our in vivo study has shown that the bilayer mem-branes promoted bone formation compared with the assessedcommercial product, thus supporting of the use of bilayermembranes in tissue regeneration. The three-dimensional

4 ( 2 0 1 8 ) 538–550

structure that mimics the ECM is important for bone regen-eration scaffolds [49,50]. The inner surface of the bilayermembrane should therefore conform to an ideal architecture.Connective tissue migrated into the outer surface of LM-1, butnot in LM-2 where the bilayer structure was maintained over 8weeks. These results reflect the cell behaviors observed in vitro.These findings indicate that the remaining outer layer of LM-2 inhibited cell proliferation and subsequent tissue invasion,while the inner layer promoted proliferation and differentia-tion of osteogenic cells, and subsequent bone regeneration.

In this study, bilayer PLGA membranes were successfullyfabricated by a two-step freezing and lyophilization process.These biomaterials, comprising a bilayer membrane with athick solid layer, could enable tissue regeneration in long-termGTR therapy by controlling cell and tissue behaviors. In addi-tion, the technology used to fabricate the multi-layer polymerconstructs could be applied in the preparation of other novelbiomaterials.

5. Conclusion

This study has demonstrated that a bilayer GTR membranecan be fabricated with different surface morphologies onopposing sides. This bilayer membrane possesses enoughmechanical property for clinical use and can facilitate tis-sue regeneration in vivo through the control of cell and tissuebehaviors. In particular, these membranes are considered tobe useful for periodontal tissue regeneration as biodegradableGTR membranes.

Acknowledgements

This work was supported in part by Grants-in-Aid for ScientificResearch (Nos. JP17K11778 and JP17H04383) from the JapanSociety for the Promotion of Science. The funding sourcesdid not have any involvement in the study or our decision topublish the manuscript. We appreciate the technical supportreceived from the Center for Frontier Oral Science, Gradu-ate School of Dentistry, Osaka University. We thank ElizabethFinnie, PhD, from Edanz Group (www.edanzediting.com/ac) forediting a draft of this manuscript.

Appendix A. Supplementary data

Supplementary data associated with this arti-cle can be found, in the online version, athttps://doi.org/10.1016/j.dental.2017.12.011.

e f e r e n c e s

[1] Bottino MC, Thomas V. Membranes for periodontalregeneration—a materials perspective. Front Oral Biol2015;17:90–100.

[2] Nyman S, Lindhe J, Karring T, Rylander H. New attachment

following surgical treatment of human periodontal disease. JClin Periodontol 1982;9:290–6.

[3] Gielkens PF, Schortinghuis J, de Jong JR, Raghoebar GM,Stegenga B, Bos RR. Vivosorb, Bio-Gide, and Gore-Tex as

3 4

d e n t a l m a t e r i a l s

barrier membranes in rat mandibular defects: an evaluationby microradiography and micro-CT. Clin Oral Implants Res2008;19:516–21.

[4] Sela MN, Kohavi D, Krausz E, Steinberg D, Rosen G.Enzymatic degradation of collagen-guided tissueregeneration membranes by periodontal bacteria. Clin OralImplants Res 2003;14:263–8.

[5] Kim YK, Kim SG, Lim SC, Lee HJ, Yun PY. A clinical study onbone formation using a demineralized bone matrix andresorbable membrane. Oral Surg Oral Med Oral Pathol OralRadiol Endod 2010;109:6–11.

[6] Milella E, Ramires PA, Brescia E, La Sala G, Di Paola L, BrunoV. Physicochemial, mechanical, and biological properties ofcommercial membranes for GTR. J Biomed Mater Res2001;58:427–35.

[7] Thoma DS, Halg GA, Dard MM, Seibl R, Hammerle CH, JungRE. Evaluation of a new biodegradable membrane to preventgingival ingrowth into mandibular bone defects in minipigs.Clin Oral Implants Res 2009;20:7–16.

[8] Bilir A, Aybar B, Tanrikulu SH, Issever H, Tuna S.Biocompatibility of different barrier membranes in culturesof human CRL 11372 osteoblast-like cells: animmunohistochemical study. Clin Oral Implants Res2007;18:46–52.

[9] Rakhmatia YD, Ayukawa Y, Furuhashi A, Koyano K. Currentbarrier membranes: titanium mesh and other membranesfor guided bone regeneration in dental applications. JProsthodont Res 2013;57:3–14.

[10] Zwahlen RA, Cheung LK, Zheng LW, Chow RL, Li T,Schuknecht B, et al. Comparison of two resorbablemembrane systems in bone regeneration after removal ofwisdom teeth: a randomized-controlled clinical pilot study.Clin Oral Implants Res 2009;20:1084–91.

[11] Villar CC, Cochran DL. Regeneration of periodontal tissues:guided tissue regeneration. Dent Clin North Am2010;54:73–92.

[12] Gentile P, Chiono V, Tonda-Turo C, Ferreira AM, Ciardelli G.Polymeric membranes for guided bone regeneration.Biotechnol J 2011;6:1187–97.

[13] Abou Neel EA, Bozec L, Knowles JC, Syed O, Mudera V, Day R,et al. Collagen—emerging collagen based therapies hit thepatient. Adv Drug Deliv Rev 2013;65:429–56.

[14] Eickholz P, Kim TS, Holle R, Hausmann E. Long-term resultsof guided tissue regeneration therapy with non-resorbableand bioabsorbable barriers. I. Class II furcations. JPeriodontol 2001;72:35–42.

[15] Kasaj A, Reichert C, Gotz H, Rohrig B, Smeets R,Willershausen B. In vitro evaluation of various bioabsorbableand nonresorbable barrier membranes for guided tissueregeneration. Head Face Med 2008;4:1–8.

[16] Behring J, Junker R, Walboomers XF, Chessnut B, Jansen JA.Toward guided tissue and bone regeneration: morphology,attachment, proliferation, and migration of cells cultured oncollagen barrier membranes. A systematic review.Odontology 2008;96:1–11.

[17] Bottino MC, Thomas V, Jose MV, Dean DR, Janowski GM.Acellular dermal matrix graft: synergistic effect ofrehydration and natural crosslinking on mechanicalproperties. J Biomed Mater Res B Appl Biomater2010;95:276–82.

[18] Parenteau-Bareil R, Gauvin R, Berthod F. Collagen-basedbiomaterials for tissue engineering applications. Materials2010;3:1863–87.

[19] MacNeil S. Biomaterials for tissue engineering of skin. MaterToday 2008;11:26–35.

[20] Hurzeler MB, Quinones CR, Schupbach P. Guided boneregeneration around dental implants in the atrophic

( 2 0 1 8 ) 538–550 549

alveolar ridge using a bioresorbable barrier. An experimentalstudy in the monkey. Clin Oral Implants Res 1997;8:323–31.

[21] Miller N, Penaud J, Foliguet B, Membre H, Ambrosini P,Plombas M. Resorption rates of 2 commercially availablebioresorbable membranes. A histomorphometric study in arabbit model. J Clin Periodontol 1996;23:1051–9.

[22] Zhao S, Pinholt EM, Madsen JE, Donath K. Histologicalevaluation of different biodegradable and non-biodegradablemembranes implanted subcutaneously in rats. JCraniomaxillofac Surg 2000;28:116–22.

[23] Partridge K, Yang X, Clarke NM, Okubo Y, Bessho K, SebaldW, et al. Adenoviral BMP-2 gene transfer in mesenchymalstem cells: in vitro and in vivo bone formation onbiodegradable polymer scaffolds. Biochem Biophys ResCommun 2002;292:144–52.

[24] Wu YC, Shaw SY, Lin HR, Lee TM, Yang CY. Bone tissueengineering evaluation based on rat calvaria stromal cellscultured on modified PLGA scaffolds. Biomaterials2006;27:896–904.

[25] Lee SB, Lee DY, Lee KK, Kim KN, Choi SH, Kim KM. Surfacemodification of guided tissue regeneration membrane usingtetracycline-containing biodegradable polymer. SurfInterface Anal 2008;40:192–7.

[26] Chen G, Xia Y, Lu X, Zhou X, Zhang F, Gu N. Effects of surfacefunctionalization of PLGA membranes for guided boneregeneration on proliferation and behavior of osteoblasts. JBiomed Mater Res A 2013;101:44–53.

[27] Xifu Z, Fei Y, Shenguo W. Fabrication and cell affinity ofbiomimetic structured PLGA/articular cartilage ECMcomposite scaffold. J Mater Sci Mater Med 2011;22:693–704.

[28] Tomita M, Lavik E, Klassen H, Zahir T, Langer R, Young MJ.Biodegradable polymer composite grafts promote thesurvival and differentiation of retinal progenitor cells. StemCells 2005;23:1579–88.

[29] Tanner MG, Solt CW, Vuddhakanok S. An evaluation of newattachment formation using a microfibrillar collagen barrier.J Periodontol 1988;59:524–30.

[30] Gottlow J, Nyman S, Lindhe J, Karring T, Wennstrom J. Newattachment formation in the human periodontium byguided tissue regeneration. J Clin Periodontol 1986;13:604–16.

[31] Selvig KA, Nilveus RE, Fitzmorris L, Kersten B, Khorsandit SS.Scanning electron microscopic observations of cellpopulation and bacterial contamination of membranes usedfor guided tissue regeneration in humans. J Periodontol1990;61:515–20.

[32] Stachewicz U, Qiao T, Rawlinson SC, Almeida FV, Li WQ,Cattell M, et al. 3D imaging of cell interactions withelectrospun PLGA nanofiber membranes for boneregeneration. Acta Biomater 2015;27:88–100.

[33] Ribeiro C, Sencadas V, Areias AC, Gama FM,Lanceros-Mendez S. Surface roughness dependentosteoblast and fibroblast response on poly(l-lactide) filmsand electrospun membranes. J Biomed Mater Res A2015;103:2260–8.

[34] Zhang E, Zhu C, Yang J, Sun H, Zhang X, Li S, et al.Electrospun PDLL/PLGA composite membranes for potentialapplication in guided tissue regeneration. Mater Sci Eng C2016;58:278–85.

[35] Hoornaert A, d’Arros C, Heymann MF, Layrolle P.Biocompatibility, resorption and biofunctionality of a newsynthetic biodegradable membrane for guided boneregeneration. Biomed Mater 2016;11:045012.

[36] Lü JM, Wang X, Marin-Muller C, Wang H, Lin PH, Yao Q, et al.Current advances in research and clinical applications ofPLGA-based nanotechnology. Expert Rev Mol Diagn

2009;9:325–41.

l s 3

[50] Sell SA, Wolfe PS, Garg K, McCool JM, Rodriguez IA, BowlinGL. The use of natural polymers in tissue engineering: a

550 d e n t a l m a t e r i a

[37] McCaul LK, Bagg J, Jenkins WM. Rate of loss of irradiatedpolyglactin 910 (Vicryl Rapide) from the mouth: aprospective study. Br J Oral Maxillofac Surg 1999;38:328–30.

[38] Deeken CR, Abdo MS, Frisella MM, Matthews BD.Physicomechanical evaluation of absorbable andnonabsorbable barrier composite meshes for laparoscopicventral hernia repair. Surg Endosc 2011;25:1541–52.

[39] Albrektsson T, Wennerberg A. Oral implant surfaces: Part1—review focusing on topographic and chemical propertiesof different surfaces and in vivo responses to them. Int JProsthodont 2004;17:536–43.

[40] Smith R, Oliver C, Williams DF. The enzyme degradation ofpolymers in vitro. J Biomed Mater Res 1987;21:1149–66.

[41] Williams DF. Mechanisms of biodegradation of implantablepolymers. Clin Mater 1992;10:9–12.

[42] Von Recum HA, Cleek RL, Eskin SG, Mikos AG. Degradationof polydispersed poly(l-lactic acid) to modulate lactic acidrelease. Biomaterials 1995;16:441–7.

[43] Grizzi I, Garreau H, Li S, Vert M. Hydrolytic degradation ofdevices based on poly(dl-lactic acid) size-dependence.

Biomaterials 1995;16:305–11.

[44] Holy CE, Dang SM, Davies JE, Shoichet MS. In vitrodegradation of a novel poly(lactide-co-glycolide) 75/25 foam.Biomaterials 1999;20:1177–85.

4 ( 2 0 1 8 ) 538–550

[45] Leal AI, Caridade SG, Ma J, Yu N, Gomes ME, Reis RL, et al.Asymmetric PDLLA membranes containing Bioglass

®for

guided tissue regeneration: characterization and in vitrobiological behavior. Dent Mater 2013;29:427–36.

[46] Huang CY, Hu KH, Wei ZH. Comparison of cell behavior onpva/pva-gelatin electrospun nanofibers with random andaligned configuration. Sci Rep 2016;5:37960.

[47] Gupta D, Venugopal J, Prabhakaran MP, Dev VR, Low S,Choon AT, et al. Aligned and random nanofibrous substratefor the in vitro culture of Schwann cells for neural tissueengineering. Acta Biomater 2009;5:2560–9.

[48] Sonomoto K, Yamaoka K, Kaneko H, Yamagata K, Sakata K,Zhang X, et al. Spontaneous differentiation of humanmesenchymal stem cells on poly-lactic-co-glycolic acidnano-fiber scaffold. PLoS One 2016;7:0153231.

[49] Polo-Corrales L, Latorre-Esteves M, Ramirez-Vick JE. Scaffolddesign for bone regeneration. J Nanosci Nanotechnol2014;14:15–56.

focus on electrospun extracellular matrix analogues.Polymers 2010;2:522–53.