Interactive Fibroblast-Keratinocyte Co-cultures: An In Vivo -Like Test Platform for Dental Implant-Based Soft Tissue Integration

Abstract

For clinicians, soft connective tissue integration (STI), one of the critical issues for dental implant success, is usually tested using the fibroblasts monolayer regime. Therefore, we aimed at an extension of this regime by employing interactive gingival fibroblast-keratinocyte cocultures (CCs) as an in vivo-like test platform. In the extended regime, 13 STI-relevant genes were analyzed in response to five different titanium implant biomaterial surfaces. The genes quantitated by real-time polymerase chain reaction were categorized as pro supportive or contra supportive, that is, nonsupportive for cell growth on an engineered surface. Monocultures had higher levels of contra supportive gene expression, but the fibroblast-keratinocyte CC had two out of five of the titanium surfaces with more pro supportive gene expression than contra supportive gene expression. We defined this change from contra supportive gene expression to pro supportive gene expression by developing the “relative supportive difference” index. Hence, interactive CCs exhibit valuable supportive effects on the expression of STI-relevant genes, possibly via physiological cell-to-cell-interactions. Our results render interactive gingival CCs suitable as a test platform for dental implant-related STI under more in vivo-like conditions.

Introduction

In dentistry, the engineering of advanced functional, that is, tissue convenient, dental implants paves the way for the progressive improvement of implant success, which is substantiated by proper osseo- and, particularly, soft-tissue integration.

For transcutaneous implants, such as dental implants, bacterial penetration can cause implant failure. Therefore, soft connective tissue integration (STI), comprising gingival-epithelial and gingival-connective tissue implant attachment, is pivotal. After implantation, epithelial cells, once having reached the implant surface, attach via basement membrane-inserting hemidesmosomes.^1,2 Due to the implants' lack of cementum, dento-gingival collagen fibers are not inserted into the implant surface. Consequently, the barrier-to-junctional epithelial depth migration and the prevention of bacterial invasion are impaired. This, in turn, can lead to gingival recession or pocket formation and inflammation-driven bone resorption,³ that is, periimplantitis with a clinical prevalence of 44.9%⁴ or even a range from 11.3%–47.1%,⁵ as a subsequent causative for bone loss with 43% prevalence or even implant loss with a prevalence up to 6%.⁶ With regard to the implant abutment, STI is mainly governed by two parameters: (1) the material's physicochemical characteristics and (2) surface topography. While physicochemical properties determine the propensity of cells to adhere to the implant surface through the surface's free energy, that is, wettability,⁷ surface topography affects cell adhesion and behavior by its form as well as texture, that is, waviness and roughness.^8,9 Common treatment processes that create various titanium surfaces include polishing, machining, plasma spraying, blasting, etching, and anodization.^10,11

With regard to STI and the heterogeneity of the conducted in vivo and in vitro investigations, studies conducted on humans and that evaluated the peri-implant soft tissues reported a lower inflammatory level in tissues surrounding zirconia compared with matched titanium implants.¹² Moreover, results from human case reports, though showing a satisfactory clinical outcome, were overall estimated as being nonconclusive.³ Studies conducted in vivo are frequently based on animal models that employ either dogs^13–17 or monkeys.¹⁸ Studies carried out under in vitro conditions address cell behavioral features such as adhesion/attachment, morphogenesis, and/or proliferation. With regard to the cellular test system, all these studies employ monocultures (MCs) of one single cell type involved in STI, that is, monolayers of either gingival epithelial keratinocytes^19–21 or fibroblasts,^22–26 respectively.

Despite the advantage of such simple cell culture devices that allow for a direct evaluation of biomaterial-related effects on cell behavior, the translation of the results obtained from conventional monolayers to the in vivo situation is limited. This is mostly due to the lack of physiological interactions of the solitary grown cell type in a monolayer with its natural counterparts in the human body. In case of STI-relevant gingival fibroblasts (GFs), these counterparts are gingival keratinocytes (GK). In previous studies with the aforementioned interactive partners, GF/GK, our group has noted a discriminating expression of keratinocyte-derived growth factors for GK cultured in monolayers versus in cocultures (CCs) with GF.²⁷ The discriminative expression of genes, essentially those contributing to periodontal tissue homeostasis, has been also reported for interactive CCs of periodontal ligament fibroblasts when grown in conjunction with their natural partner, namely osteoblasts of the alveolar bone.²⁸ These paradigms provide evidence that complex interactive CCs allow for physiological cell-to-cell-interactions, which are most likely mediated by diffusible growth factors. Hence, the results obtained from such cell systems resemble the in vivo situation more closely than those obtained from monolayers. Therefore, interactive CCs render precious test platforms for behavioral cell studies on gene expression under more in vivo-like conditions. As already mentioned, not only the genes encoding for constituents of the extracellular matrix (ECM), such as fibrillar collagen type I,²⁹ but also other collagens³⁰ and ECM molecules, such as versican,³¹ are pivotal for STI, due to their potential ligand role in cell adhesion. This also holds true for cell adhesion-mediating integrins,³² which, as ECM receptors, are involved in attachment of the gingival soft tissue to a natural tooth or optionally an implant-innate abutment surface. On the other hand, bacteria-driven inflammation compromises periodontal STI, for example, by the induction of inflammatory cytokines such as interleukin 6 (IL6) in GF.³³ In the context of the respective biological gene function of genes under study in terms of their involvement in periodontal tissue homeostasis, genes were annotated in the categories supportive or nonsupportive with regard to the dental implant-associated process of STI. This STI gene function-related annotation accounted for the scientific findings in early peri-implant tissue reactions in response to different implant topographies,³⁴ and their involvement in either promoting or destructive processes in the context of soft tissue regeneration.^35–37 Based on this molecular background of STI and the knowledge that gene expression is governed by the complexity of the cellular test system, the present gene expression study focuses on genes of (1) the ECM and its turnover; (2) cell adhesion; and (3) inflammation in GF and GF/GK interactive CCs in response to various titanium surfaces, representing advanced dental implant materials.

Materials and Methods

Cell culture

Primary cultures of GFs were established using the explant technique.³⁸ For tissue harvest, informed consent was obtained from the patients according to the Helsinki Declaration, and the protocol was approved by the institutional ethics committee. Cultures were maintained for routine cell culture in Dulbecco's Modified Eagle's Medium (DMEM) (PAA) containing 10% fetal calf serum (FCS) (Seromed; Biochrom) and 50 μg/mL kanamycin (Roche Diagnostics). GFs were cultured in passages 8–12, and comparable passages were used for seeding on titanium test specimens. To exclude cell-senescence-caused alterations of cell morphology and the expression of tissue-specific biomarkers, periodontal cells were previously tested on the consistent expression of respective biomarkers and stable morphology over time.

The establishment, characterization, and serial cultivation of immortalized human gingival keratinocytes (IHGKs) have been reviewed by Roesch-Ely and coworkers.²⁷ IHGKs were maintained in a keratinocyte growth medium (basal keratinocyte medium, KGM2, with provided supplements, Promocell), containing 50 mg/mL kanamycin (Sigma).

Test specimens

All test specimens under study were made of titanium at a diameter of 5 mm and a thickness of 1 mm. The specimens were modified with regard to their surface topography/chemistry as follows: (1) polished titanium (polished); (2) machined titanium (machined); (3) machined titanium, plasma cleaned and stored in a 0.15 M sodium chloride solution under nitrogen (plasma modM); (4) titanium acid etched with a mixture of hydrochloric and sulfuric acid (surface A); and (5) titanium acid etched with a mixture of hydrochloric and sulfuric acid and stored in a 0.15 M sodium chloride solution under nitrogen (mod A). After physical surface modifications, titanium specimens were packaged and gamma sterilized.

Establishment of GF-MCs and GF/GK CCs on titanium specimens

To ensure the reproducibility of the obtained data, the experiments were conducted in three independent biological replicates, respectively. The test specimens were carefully placed into 96-well plates (Falcon, BD Biosciences). To avoid drying of the chemically surface-modified specimens plasma mod M and mod A, titanium discs were maintained in phosphate-buffered saline (PBS; Invitrogen) before the seeding of the cells. Culture dish surfaces without specimens were used as controls. For MCs, a cultivation period of 7 days, usually yielding 80% of confluency, was conducted, and 2×10⁴ cells were seeded on the specimens. The cells were maintained under standard cell-culture conditions: 37°C, 97% humidity, and 5% CO₂.

For GF/GK CCs, GFs were precultivated on the test specimens for 24 hours in the 96-well plate format. Cell-culture dish surfaces without specimens were again used as controls. In the control setup, 2×10⁴ GFs were directly seeded in the 12-well plate format. For the CC setup, the GFs' preseeded specimens with 2×10⁴ cells were transferred, due to the lack of space, from the 96-well plate to the 12-well plate format, containing 2 mL FAD-medium each (Ham's F12/DMEM: mixing ratio 1:3, Biochrom), 5% FCS, 50 mg/mL Kanamycin (Sigma), and the supplements of KGM2 (Promocell). After GF precultivation, porous cell-culture inserts with pore sizes of 3 μm (Falcon, BD Biosciences) were placed in the 12-well plates, and 1×10⁵ GKs suspended in 1 mL of FAD medium were seeded on the porous membrane. This setup was cultivated for the aforementioned time period of 7 days. All cells were maintained under standard cell-culture conditions: 37°C, 97% humidity, and 5% CO₂.

Scanning electron microscopy

After the respective culture periods of the test specimens, GFs were fixed with 3.8% formaldehyde in PBS for at least 1 h and rinsed thrice with PBS buffer. Then, the specimens were dehydrated by rinsing through graded ethanol/water mixtures (50%, 70%, 80%, 90%, and 100%; each step for 10 min at room temperature). Thereafter, ethanol was slowly exchanged by liquid CO₂. Finally, the samples were dried using the critical-point method³⁹ and sputtered with a thin layer of gold of approximately 10 nm in thickness.

Cytoskeleton staining

GFs were fixed after the 7 days with 3.8% formaldehyde in PBS for 10 min and rinsed thrice with PBS buffer. Subsequently, the cells were permeabilized with 0.2% TritonX-100 (Roche) for 5 min and washed thrice with PBS followed by incubation with the Image-iT FX signal enhancer (Invitrogen) for 30 min and rinsing with PBS for 3 times each. After that, the actin filaments were stained by incubating the cells with Phalloidin-Alexa488 (Invitrogen; wd=1:100) for 45 min. Thereafter, the cells were washed thrice with PBS for 15 min, respectively. Then, the cell nuclei were counterstained with a 300 nM DAPI solution for 10 min. The cells were washed thrice with PBS followed by rinsing once with water. Finally, the cells were embedded in a mounting medium (SouthernBiotech) and observed by fluorescent microscopy (BZ-6000; Keyence).

RNA-isolation and quantitative real-time–polymerase chain reaction analysis of GFs derived from monocultures and CCs

After the cultivation period of 7 days, GFs were lysed with RLT buffer (Qiagen) and homogenized with QIAshredder columns according to manufacturer's instructions (Qiagen), followed by total RNA isolation using the RNeasy mini kit (Qiagen). The specimens' derived RNA was analyzed by determination of the RNA concentration and integrity using the automated electrophoresis system (Experion BioRad). First-strand cDNA was synthesized from 500 ng of total RNA aliquots in the reaction mixture using the C-03-first-strand kit (SABiosciences, Qiagen), according to the manufacturer's protocol. For normalization, cDNA concentration was adjusted to 5 ng/μL for each polymerase chain reaction (PCR). The qPCR analysis was performed with the CFX96 real-time PCR detection system (BioRad). Quantitative amplification detection was achieved using the SABiosciences qPCR SYBR Green Master Mix (SASiosciences, Qiagen). For the screening and determination of relative gene expression modulations, which appears relevant in the soft tissue integration (STI) context, a customized PCR Array for the detection of 86 different biomarkers was performed (SABiosciences, Qiagen), using the total RNA derived from the GFs cultivated only on the test specimen surface A and polished. qPCR of the arrays was performed according to the manufacturer's instructions. Following the identification of 13 relevant biomarkers in the STI context, qPCR of the controls and the remaining test specimens machined, plasma mod M and mod A were performed with the commercially available primers of Collagen 12 alpha 1 (COL12A1), Collagen 14 alpha 1 (COL14A1), Collagen 4 alpha 2 (COL4A2), Collagen 5 alpha 1 (COL5A1), Collagen 1 alpha 1 (COL1A1), Secreted protein-acidic and -rich in cysteine/osteonectin (SPARC), Integrin alpha 2 (ITGA2), Integrin alpha 8 (ITGA8), Integrin beta 3 (ITGB3), Versican (VCAN), Matrix-Metalloproteinase 1 (MMP1), Matrix-Metalloproteinase 3 (MMP3), Interleukine 6 (IL6), and the housekeeping genes β-Actin (ACTB) and Ribosomal protein L13A (RPL13A) from SABiosciences (SABiosciences, Qiagen; see Table 3). The standard temperature profile included an initial denaturation step for 10 min at 95°C, followed by 40 cycles of denaturation at 95°C for 15 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s.

The relative expression levels of each mRNA were analyzed using a modification of the ΔΔC_T equation, which allows the counting for differences in efficiencies (E=10–^1/slope) between the PCRs.⁴⁰ The data were calculated using the Excel sheet template provided with the PCR arrays from SABiosciences (SABiosciences, Qiagen). The data were normalized to the index C_T of the ACTB and/or RPL13A nonmodulated housekeeping genes and always referred to the relative gene expression values of the cell's RNA derived from standard cell-culture plastic dishes. The data presented reflect the means (n=3,±SD) of three independent experiments. By using this modus operandi, the genes exhibiting a threshold of±2 were considered to be significantly modulated in response to surface exposition. The significance of gene expression values was evaluated by performing the Student's t-test, and significant values were annotated by an asterisk in Table 3. The fold difference factors, comparing CC with monoculture relative expression values, were calculated in case of fold down-regulation by using the respective reciprocal values. In case of fold up-regulation, the direct listed gene expression values were used for factor calculation. The calculated factor values of±3 were determined as significant and transferred to the relative supportive difference (RSD) summary in Table 4.

Surface roughness measurement

The surface roughness images were acquired with a confocal microscope (μSurf, NanoFocus AG). Three measurements were performed on each test specimen, and three specimens were measured for each type of surface. The whole roughness image with a size of 798 μm×798 μm was used for the calculation of the 3D roughness parameters. The micro (roughness, low-pass filter on wavelength) and macro (waviness, high-pass filter on wavelength) roughness values were determined using a moving average Gaussian filter with a cut-off wavelength of 30 μm (x=31 μm, y=30 μm, 20×19 image points). All images were acquired with the 20×objective. The roughness S_a (arithmetic mean deviation), the maximum peak-to-valley height S_t, and the skewness S_sk were determined (mean±SD, n=3) and summarized in Table 1.

Table 1.

Measurements of Different Surface Textures

Surface	S_a [μm]	S_t [μm]	S_sk
Polished	0.006	0.05	0.224
Machined	0.119±0.012	0.75±0.05	−0.071±0.042
Surface A	0.653±0.076	4.01±0.36	0.014±0.051
Mod A	0.648±0.055	4.02±0.30	−0.052±0.021
Plasma mod M	0.118±0.011	0.76±0.06	−0.058±0.075

Data represent the mean of three independent surface texture measurements of three specimens (mean±SD, n=3).

S_a, arithmetic average of the 3D-roughness; S_t, Maximum height of the 3D-profile; S_sk, skewness of the profile.

X-ray photoelectron spectroscopy analysis

The chemical composition of the sample surfaces (outermost 5–10 nm) were determined by X-ray photoelectron spectroscopy (XPS) (performed by SuSoS AG). Test specimens were γ-sterilized, and the titanium discs stored in liquid were carefully rinsed with ultrapure water and blown dry with nitrogen before the analysis. XPS Spectra were acquired on a PhI5000 VersaProbe spectrometer (ULVAC-PHI, Inc.) equipped with a focused scanning monochromatic Al-K_α source (1486.6 eV).The photoelectrons were detected at an angle of 45° to the normal surface by means of a hemi-spherical analyzer with a multi-channel detection system with 16 channels. Each sample was analyzed at one spot with an area of 1.4×0.5 mm². A survey scan and detailed spectra of the elements observed in the survey were acquired.

The samples stored in saline solution were rinsed with ultrapure water and dried in a stream of nitrogen before the XPS measurements. Then, the samples were measured as received, and the chemical surface composition was determined from three independent measurements (mean±SD, n=3) and summarized in Table 2.

Table 2.

Measurements of Chemical Surface Compositions

Surface	CA [°]	C [%]
Polished	73°±5°	31.5
Machined	78.4°±5.7°	34.2
Surface A	127.5°±5.8°	44.9
Mod A	5.5°±5.1°	7.2
Plasma mod M	7.4°±0.8°	12.2

Data represent the mean of three independent measurements (mean±SD, n=3); chemical surface composition values evaluated by XPS analysis are given in [%], and contact angles are given in [°].

XPS, X-ray photoelectron spectroscopy.

Contact angle measurement

The contact angle measurements were performed using a sessile drop test with ultrapure water (EasyDrop DSA20E, Krüss GmbH). The water droplets were dosed using an automated unit, and the droplet size was chosen as 3 μL for the samples stored dry and as 0.3 μL for the samples stored in saline solution. The samples stored in saline solution were blown dry in a stream of Ar before the contact angle measurements, to prevent the extensive formation of salt crystals that might interfere with the measurement. The samples stored dry were measured as received. The contact angles were measured by fitting a circular segment function to the contour of the droplet placed on the surface, calculated from three independent measurements (mean±SD, n=3), and summarized in Table 2.

Results

Measurements of chemical and physical titanium implant specimen surface attributes

In the forefront of the analysis of cellular responses to different titanium implant surfaces, we investigated the physical and chemical characteristics of the test specimens under study and first determined the individual surface textures by confocal microscopy. The arithmetic average of 3D-rougness (S_a), the maximum height of the 3D-profile (S_t), and the skewness of the profile (S_sk) were used as read-out parameters for surface roughness. The measurements resulted in the strongest surface roughness for surface A 0.653 μm and mod A 0.648 μm, and in a very smooth surface topography for polished 0.006 μm, machined 0.119 μm, and plasma mod M 0.118 μm (Table 1). In addition to surface roughness measurements, we analyzed the chemical implant surface composition by focusing on the carbon portion at the surface, which is considered responsible for the degree of surface wettability and can be determined by the contact angle measurements of a water drop on the titanium-based implant surface. This surface analysis revealed increasing contact angles with concomitant increasing carbon contents at the titanium surfaces, summarized in Table 2. In detail, the three out of five surfaces under study depicted the largest contact angles of 73° polished, 78.4° machined, and 127.5° surface A (Table 2). As expected, these results corresponded with high carbon contents in the titanium test specimens of 31.5% polished, 34.2% machined, and 44.9% surface A (Table 2).

Effects of various titanium surfaces on fibroblast morphogenesis

In the context of soft tissue integration (STI), it appears likely that the establishment of proper cell morphology over time on the implant surface of choice reflects a pivotal prerequisite for adequate gene expression. After 7 days of culture, we first analyzed the effects emanating from the candidate titanium biomaterials on the morphogenesis of GFs by combined cytoskeleton staining (CS) for actin and scanning electron microscopy (SEM). Figure 1 representatively exemplifies the main characteristics in cellular actin distribution and morphogenesis by depicting cell details and overviews, shown in the respective inlays. While fibroblasts on polished and machined titanium surfaces displayed actin arranged in distinct bundles at the cell margins (Fig. 1A, B), the other 3 surfaces revealed less restrictive margin-associated actin bundling (Fig.1C: plasma modM, D: surface A, E: surface Mod A). With regard to actin being the key player in stress fiber formation and focal cell-matrix and/or cell-surface adhesions,⁴¹ the actin CS patterning found on the different implant surfaces suggests easy fibroblast access in the case of polished and machined titanium, and aggravated adhesion on plasma modM, surface A, and surface mod A, respectively. SEM clearly corroborated visible fibroblast margins with distinct cell protrusions, presumably reflecting cell-surface contact structures, when GFs were grown on polished (Fig. 1F) as well as machined titanium (Fig. 1G). By contrast, both cell margins and protrusions were remarkably less pronounced for cells cultured on plasma modM (Fig. 1H), surface A (Fig. 1I), and surface modA (Fig. 1J).

FIG. 1.

Cytoskeleton staining (CS) and scanning electron microscopy (SEM) on gingival fibroblast (GFs) and the influence of different titanium surface treatments on actin stress fiber formation and cell morphology. (A–E) Representative images of actin filament staining of GFs cultured on titanium disc specimens. The different surface treatments are illustrated as (A) polished, (B) machined, (C) plasma modM, (D) surface A, and (E) mod A. (F–J) Similar details to the actin staining documented by SEM images under consideration of the focal adhesion sites of GFs to the differently treated titanium specimens. High-resolution images document the altering focal adhesion site morphology of GFs in response to the modified test specimens (F) polished, (G) machined, (H) plasma modM, (I) Surface A, and (J) mod A. Inlays always illustrate the overview of the detailed images. Bars represent 10 μm. Color images available online at www.liebertpub.com/tec

Surface impact on gene expression in monolayers and CCs

Before the exertion of biological function, the transcription of STI-promoting genes is mandatory. To reproducibly analyze the impact of the different biomaterial implant surfaces on the relative expression of genes, encoding for STI-relevant biomolecules, we employed a screening experiment for two lead implant surface modifications with customized pathway-specific PCR Arrays, comprising 86 genes, to monitor and identify the transcription of STI-relevant biomarkers in real time. For veritable evaluation, determined values of fold-up or down-regulation among the total in a previous screening study identified 13 genes, addressing (1) ECM and its turnover, (2) cell adhesion, and (3) inflammation, implying reference to control. This reference includes the respective surfaces in the respective culture modes, that is, GFs in monolayers or keratinocyte CCs, referred to as GFs established in standard culture dishes. By using this modus operandi, the genes exhibiting a threshold of±2 were considered to be significantly modulated in response to surface exposition.

To identify a study-integrated candidate gene as being supportive, that is, pro or nonsupportive, that is, contra significant with regard to STI, it appears noteworthy that pro and/or contra is per se not defined by its up- and or down-regulation, but through its biological function. For instance, the significant up-regulation of a candidate gene, encoding for a matrix or adhesion molecule, respectively, appears favorable for STI, and, thus, is defined as supportive or pro; while the considerable up-regulation of a matrix-degrading protease, such as MMP1, is considered nonsupportive, and herewith defined as contra. Hence, in light of their biological function, the up-regulation of the matrix genes, A1 collagens type-12, -14, -1, and A2 collagen 4, -5, as well as versican and SPARC/osteonectin, along with the integrin subunits α2, α8, and β3, were considered pro STI; while their down-regulation was attributed contra. Vice versa, the up-regulation of the matrix-degrading proteases MMP1 and MMP3 in conjunction with the pro-inflammatory cytokine IL6 were considered contra, while their down-regulation was assigned pro STI.

Based on this algorithm, relative gene expression revealed the predominance of contra significant differences in GF monolayers, irrespective of the chemical or topographical properties of the analyzed implant surface. In detail, among the 5 titanium-based implant biomaterials, the polished surface exhibited 7 contra out of 8 total determined significant differences. This trend of prevalence also applied to the other surfaces tested, showing 2 out of 2 contra significant differences, in case of machined, 4 out of 6 in case of surface A, 6 out of 10, for mod A, and 4 out of 6, for plasma modM. A detailed overview of the fold-up and/or down-regulation values is given in Table 3, while the total number of detected significant differences and their assignment to pro as well as contra is summarized in Table 4.

Table 3.

Fold Up- or Down-Regulation of Surface-Sensitive Genes

Surface		Polished		Machined		Surface A		Mod A		Plasma mod M
Target gene	Culture type	Mono-culture	Coculture	Mono-culture	Coculture	Mono-culture	Coculture	Mono-culture	Coculture	Mono-culture	Coculture
COL12A1(pro)	Fold up-/down-regulation with reference to control	−2.39^*±0.126	−1.29±0.235	−2.11^*±0.245	−1.5^*±0.125	−2.54^*±0.225	−2.81^*±0.144	−4.63^*±0.147	−2.13^*±0.254	−2.69^*±0.125	−3.13^*±0.245
	Fold difference coculture to mono-culture	1.85		1.41		0.90		2.17		−1.16
COL14A1 (pro)	Fold up-/down-regulation with reference to control	−2.16^*±0.252	7.36^*±0.212	−1.32±0.202	3.21^*±0.258	−1.4±0.148	−1.29^*±0.164	−1.43±0.169	1±0.142	−1.27±0.227	2.38^*±0.211
	Fold difference coculture to mono-culture	15.89		4.24		1.09		1.43		3.02
COL1A1(pro)	Fold up-/down-regulation with reference to control	2.36^*±0.321	1.19±0.358	−1.42^*±0.231	−1.35^*±0.331	−3.34^*±0.425	−2.54^*±0.197	−2.43^*±0.191	−1.62^*±0.419	−1.58^*±0.321	−1.31^*±0.351
	Fold difference coculture to mono-culture	2.81		1.05		1.32		2.09		1.21
COL4A2 (pro)	Fold up-/down-regulation with reference to control	−1.73±0.187	1.02±0.177	1.19±0.189	2.36^*±0.196	1.62^*±0.123	1.47^*±0.321	1.72^*±0.344	1.74^*±0.122	1.13±0.187	1.3^*±0.177
	Fold difference coculture to mono-culture	1.76		1.98		−1.1		1.01		1.15
COL5A1 (pro)	Fold up-/down-regulation with reference to control	−2.59^*±0.247	1.02±0.249	−1.34±0.239	−1.12±0.247	−1.59^*±0.325	−2.28^*±0.188	−2.59^*±0.171	−1.5^*±0.325	−1.84^*±0.286	−1.94^*±0.228
	Fold difference coculture to mono-culture	2.64		1.2		−1.43		1.73		−1.05
VCAN (pro)	Fold up-/down-regulation with reference to control	−2.3^*±0.155	4.95^*±0.256	−1.82±0.259	4.56^*±0.147	−2.48^*±0.118	1.83±0.278	−8.99^*±0.289	3.09^*±0.201	−3.4^*±0.155	−3.6^*±0.254
	Fold difference coculture to mono-culture	11.38		8.3		4.54		27.78		−1.06
ITGA2 (pro)	Fold up-/down-regulation with reference to control	−1.1±0.363	2.8^*±0.389	−1.02±0.377	2.97^*±0.333	1.27±0.194	−1.03±0.173	−2.13^*±0.169	1.41^*±0.189	−1.49±0.363	−1.17±0.399
	Fold difference coculture to mono-culture	3.08		3.03		−1.30		3		1.27
ITGA8 (pro)	Fold up-/down-regulation with reference to control	1.13±0.204	2.05^*±0.197	1.61^*±0.177	−1.07±0.210	2.7^*±0.278	−1.42±0.289	2.28^*±0.279	−1.37±0.278	2.39^*±0.204	35.93^*±0.198
	Fold difference coculture to mono-culture	1.81		−1.72		−3.83		−3.12		15.03
ITGB3 (pro)	Fold up-/down-regulation with reference to control	1.17±0.318	1.05±0.212	1.89^*±0.211	10.2^*±0.319	1.42±0.311	8.35^*±0.247	2.35^*±0.255	8.29^*±0.299	1.8^*±0.320	2.65^*±0.201
	Fold difference coculture to mono-culture	−1.11		5.4		5.88		3.53		1.47
MMP1 (contra)	Fold up-/down-regulation with reference to control	1.04±0.332	1.09±0.355	−1.91^*±0.327	1.44^*±0.322	−2.81^*±0.264	−1.23±0.255	−1.86±0.239	1.07±0.267	1.42±0.332	−2.17^*±0.288
	Fold difference coculture to mono-culture	1.04		2.75		2.28		1.99		−3.08
MMP3 (contra)	Fold up-/down-regulation with reference to control	3.05^*±0.206	2.53^*±0.174	4.86^*±0.169	6.98^*±0.208	11.6^*±0.111	6.39^*±0.240	12.81^*±0.241	4.85^*±0.115	7.37^*±0.210	4.55^*±0.177
	Fold difference coculture to mono-culture	−1.20		1.44		−1.82		−2.64		−1.62
SPARC (pro)	Fold up-/down-regulation with reference to control	−3.2^*±0.182	−4.96^*±0.266	−1.44±0.256	9.94^*±0.189	−1.09±0.320	10.43^*±0.188	5.21^*±0.211	6.99^*±0.322	3.32^*±0.191	22.33^*±0.249
	Fold difference coculture to mono-culture	−1.55		14.31		11.37		1.34		6.73
IL6 (contra)	Fold up-/down-regulation with reference to control	−1.97±0.179	−3.21^*±0.122	1.53±0.145	1.04±0.177	−1.03±0.311	−1.6^*±0.201	−5.23^*±0.214	1.07±0.309	−1.35±0.169	−12.19^*±0.122
	Fold difference coculture to mono-culture	−1.62		−1.47		−1.55		5.6		−9.03

Asterisks in fold up- or down-regulation of genes indicate significant modulation in response to surface exposition, analyzed by a Student's t-Test for equal variance (p<0.01; n=3±SD). Data are normalized (ΔC_t) to the reference which includes the respective surface in the respective culture mode, that is, gingival fibroblasts in monolayers or keratinocyte cocultures, referred to gingival fibroblasts established in standard culture dishes. Moreover, values represent ΔΔC_t normalization considering the unaffected gene expression of the housekeeping genes ACTB and RPL13A.The fold difference factors, comparing cocultures with monoculture relative expression values, were calculated in case of fold down-regulation by using the respective reciprocal values. In case of fold up-regulation, the direct listed gene expression values were used for factor calculation. The calculated factor values of±3 were given as bold values and transferred to relative supportive difference (RSD) summary in Table 4. The terms pro and contra in brackets annotate the assignment of the respective gene as pro-or contra-significant in terms of soft-tissue integration (STI).

Table 4.

Number of Significant Folds Up- or Down-Regulations and Relative Supportive Difference Values

Culture type		Mono-culture		Coculture		RSD
Surface specimen	Indication	Pro	Contra	Pro	Contra	Pro	Contra
Polished	Number of significant differences coculture to mono-culture	8		7		3
	Number of indications	1	7	5	2	3	0
Machined	Number of significant differences coculture to mono-culture	2		7		5
	Number of indications	0	2	6	1	5	0
Surface A	Number of significant differences coculture to mono-culture	6		6		4
	Number of indications	2	4	2	4	3	1
Mod A	Number of significant differences coculture to mono-culture	10		5		5
	Number of indications	4	6	3	2	3	2
Plasma mod M	Number of significant differences coculture to mono-culture	6		9		5
	Number of indications	2	4	6	3	5	0

Number of significant up- or down-regulations for mono-culture and coculture and within the number of pro and contra indications for the support of soft tissue integration. The RSD indicates the number of significant differences between co- and mono-culture and within the pro and contra indications for the supportive effect of the coculture with reference to the respective mono-culture. RSD, relative supportive difference.

With regard to the fibroblast monolayer cultures, a striking finding among the predominances of contra significant differences was the level of relative gene expression detected for the matrix metalloproteinase MMP3. This protease was significantly increased at any titanium surface tested, thereby displaying a clear distinction in its fold expression (Table 3). This distinction was substantiated by fold expression values of 3.05 (polished, Table 3), 4.86 (machined, Table 3), 11.6 (surface A, Table 3), 12.81 (mod A, Table 3), and 7.37 (plasma modM, Table 3). An inter-surface comparison of elevated expression clearly revealed higher MMP3 values for the latter 3 surfaces (Table 3), suggesting that such an inter-surface difference reflects an inter-surface ranking concerning the surface-innate support of STI.

While we have observed the prevailing existence of STI-assigned contra significant differences in case of GFs established as monolayers on the various titanium biomaterial surfaces, switching of the experimental setting to interactive CCs via inclusion of GKs reduces the number of contra superiority (Table 3). Considering the absolute numbers of significant differences found in CCs, surface A and plasma modM displayed 6 and 9, respectively, among which 4 were assigned as contra in case of surface A, while 3 were detected for plasma modM (Table 4). Despite these STI-related contra indications, the CC system intriguingly proved the presence of 3 pro out of 5 total significant differences in case of the mod A surface, and remarkable 6 out of 7, and 5 out of 7 pro significant differences for the machined and polished titanium biomaterials (Table 4).

Relative supportive difference mirrors the CC effect on gene expression

While in the monolayers an inter-surface comparison regarding the expression of STI-relevant genes revealed the uniform prevalence of contra significances, the CC system yielded the dominance of pro significant differences for 2 of the tested surfaces. To confirm this hint of compensatory effects of the interactive CC with regard to the absolute number of detected significances, we carried out a direct comparison between both culture systems for each single gene included in the study.

Independent from the expression level found for a gene of interest by qPCR in the monolayer situation, support by the CC was defined as existent, when the benefit of the CC has reached or exceeded the threshold value of ±3 (bold values given in Table 3). Such a threshold appears reliable, as it is beyond the level of ±2, indicating significance in relative gene expression analysis.

This means, for instance, that even if an STI-supportive gene, such as COL1A1, exhibited down-regulation, that is, a negative expression level in monolayers, a weakening in its down-regulation of three or more in the CC system indicates its supportive effect. A vice versa situation held true for genes unfavorable for STI, for example, MMP3. Considering the threshold, here, a weakening of its up-regulation indicated support of the interactive CC.

On the basis of this algorithm, the number of “RSD,” detected by comparing mono- with CCs is summarized in Table 4, while Table 3 illustrates the calculation of the RSD derivation in detail. With reference to the control, the detected fold difference from CC to the monoculture model yielded remarkable results for collagen/COL14A1 (15.89, Table 3) in case of “polished” titanium, while for “machined” the highest fold difference in gene expression was found for osteonectin/SPARC (14.31, Table 3), which also applied to surface A (11.37, Table 3). Tremendous fold differences in gene expression values were seen for versican/VCAN (27.78, Table 3) for mod A, and integrin/ITGA8 (15.03, Table 3) for plasma modM.

Across the board, RSD demonstrated hegemony of pro significant differences for each of the 5 titanium biomaterials, irrespective of their surface topography or chemical nature. Interestingly, the mod A surface, though featuring the lowest carbon content (7.2%, Table 2) and contact angle (5.5°±5.1°, Table 2), exhibited the lowest rate of pro significant differences by a ratio of 3:2 in relation to the total number of significant differences (Table 4). In light of the clear pro supremacy, detected for the other implant biomaterials (for pro/contra ratio see Table 4: polished 3:0; machined 5:0; surface A 3:1; plasma modM 5:0, Table 4), this may suggest that the feeling of comfort on a surface, which is mirrored by the overall STI-supportive gene expression situation, is more governed by the CC-mediated interplay of GFs with their keratinocyte counterpart, rather than surface-innate chemical issues, such as the carbon content.

Discussion

While in the past, research and development conducted on titanium-based dental implant biomaterials aimed at the improvement of the osseointegration, soft tissue integration (STI) is presently gaining more consideration. From the tissue viewpoint, the expression of genes encoding for certain biomolecules that initiate and maintain the adhesion of gingival soft tissue to the transcutaneous part of the implant is of particular importance. In this context, STI-supporting genes address the ECM of gingival soft connective tissue fibroblasts, as their constituents form the basis for primary as well as long-term cell adhesion.⁴² This adhesion, facilitated by integrins,⁴³ in turn, is critical to avoid bacterial infiltration into the deeper implant regions, which involve the alveolar bone as being the main anchor of the implant. Bacterial infiltration beyond the soft tissue barrier, therefore, opens the road for periimplantitis,⁴⁴ one of the main causatives for implant loss. At the molecular level, periimplantitis, being an inflammatory process, includes protease-driven matrix degradation,^45,46 induced by pro-inflammatory mediators.⁴⁷

Thus, the biological response of GFs to the implant and/or abutment biomaterial-innate properties, comprising surface texture and chemical composition as well as additional surface treatments, is a pivotal research issue regarding STI. Therefore, our study included a panel of 13 genes, analyzed in their relative gene expression, which are assigned to the STI-relevant aspects just mentioned: (1) ECM (2) cell adhesion, (3) protease driven ECM turn over, and (4) inflammation. To acquire more tissue-relevant results with regard to the transcription of STI-associated genes, we extended, for the first time, the study on GF monolayers, exposed to 5 various titanium-based dental implant biomaterials, by including GF- GK -CCs as an interactive culture regime in the experimental setup. These more tissue-relevant results were substantiated by the fact that the cells used for the study originated from biopsies resembling very closely the soft tissue integration site of the implant/soft tissue interface, and exhibit a similar gene expression to the in vitro culture model.²⁷

As summarized in Table 1, surface texture as visualized by scanning electron microscopy (SEM) unraveled striae for machined and plasma modM titanium surfaces, while a smooth surface was visible for polished, and caldera surfaces were seen for both A and mod A, respectively. SEM pointed to the dependency of regular GF morphogenesis on surface texture, presumably by affecting the presence of adhesion structures. This assumption is backed up by the conjunction of SEM with actin fluorescence presentation, which showed not only less cell margin-localized actin bundling and its presence at presumptive adhesion sites in case of the caldera surfaces, but also plasma modM. Here, the less clear actin bundling compared with matched machined surfaces may result from the additional treatment of this titanium surface by plasma and sodium chloride storage under nitrogen. Preferential fibroblast adhesion onto polished titanium, as also seen in our study, became obvious in a report describing the supermacy of the polished surface over etched or sandblasted surfaces.⁴⁸ On the other hand, the presence of actin-associated adhesion sites and the development of regular morphology, observed on polished and machined titanium, suggest that these STI-critical issues do not necessarily depend on surface wettability, which is defined by carbon content and contact angle. Both surfaces exhibit intermediate ranges for these parameters (contact angle: 73±5° polished, 78.4±5.7° machined; carbon content: 31.5% polished, 34.2% machined; see Table 2), when compared with the others (contact angle, carbon content: 127.5°±5.8°, 44.9% surface A; 5.5°±5.1°, 7.2% mod A; 7.4°±0.8°, 12.2% plasma modM, see Table 2). This suggests that for proper GF adhesion, the carbon content with its resulting consequences is not an absolute critical issue. The relativization of carbon as an adhesion-limiting parameter is supported by a study on cell adhesion on artificial materials for tissue engineering. Here, various cell types have been shown to be supported in their adhesion and spreading when the artificial materials have been enriched with carbon.⁴⁹ Based on these findings, it can be hypothesized that in conjunction with other surface issues, a certain carbon content with its yielding contact angle is not necessarily counterproductive for successful cell adhesion.

With regard to the gene expression analysis, GF displayed constitutive up-regulation for relative MMP3 gene expression in the monolayer situation, regardless of the type of titanium surface. However, the most striking matrix-degrading protease MMP3 up-regulation was seen for surface A (11.6-fold, see Table 3) and mod A (12.81-fold, see Table 3). These striking levels point to an inter-surface ranking with regard to STI support, suggesting that the surface A/mod A-innate caldera topography represent an environment for GF elevating matrix-degrading proteases, thereby counteracting STI more, rather than the other surfaces under study. Such inter-surface variances in protease expression appear possible, as for another protease, namely MMP2, distinct relative gene expression levels have been described for GFs on titanium substrates, varying in their surface topography.⁵⁰

Based on the general prevalence of contra significances, detected for GF monolayers on all titanium biomaterials, it appears likely that the cells were less capable to properly express STI-relevant genes when they were solely growing on the surfaces. Nearly the total dominance of pro significances, as detected in the interactive CC situation for polished and “machined” titanium implant biomaterials, leads to the assumption that the expression of STI-supportive genes is not per se or exclusively governed not by the total of the surface features, but also by the complexity of the culture system, that is, whether physiological cell-to-cell-interactions are existent or not. Although not being the focus of the present study, in previous investigations on interactive CCs, comprising cells derived from periodontal tissues, we showed that these physiological interactions are substantiated by diffusible growth factors.²⁸

Moreover, the presence of pro significant differences in case of the CC situation seen in our study indicates the compensatory effects on STI with matched monolayers. In concordance with these findings, we have received first evidence for the existence of CC-mediated compensatory effects in a former study on dental resins. Herein, harmful effects were detected after the exposition of GF and keratinocyte monolayers to resins based on poly-methylmethacrylates. By contrast, corresponding interactive CCs of the mentioned periodontal cell types were devoid of these effects,⁵¹ thereby strongly suggesting keratinocyte-fibroblast interactions being the key to this compensation. In the present study, this compensatory effect is apparent by the number of RSD-, resulting from the direct comparison of monolayers with CCs. Since RSD revealed the hegemony of pro significant differences for all 5 titanium biomaterials under study, interactive CCs of GFs and keratinocytes prove to be an adequate in vitro system for analyzing the relative expression of STI-relevant genes under more tissue-like conditions.

Conclusion

We report a novel approach that is used for testing soft tissue integration (STI), one of the cornerstones of advanced dental implant material success, under more in vivo-like conditions. In the context of this approach, our results show for the first time that the extension of the experimental set up from monolayers to interactive CCs reveals compensatory effects of the CC system in relative gene expression with regard to the promotion of soft tissue integration in response to various advanced titanium dental implant materials. Thus, interactive CCs render a test platform for elaborating more in vivo relevant results regarding STI. Further, they provide a platform for the prospective assessment of STI candidate advanced functional materials, for example, also including innovative ceramics-based zirconia dental implant materials.

Footnotes

Acknowledgments

This work was supported by a grant from Institut Straumannn AG, Basel, Switzerland, to the Department of Oral Biotechnology (Thorsten Steinberg and Pascal Tomakidi; Soft Tissue Screening Model: ZVK 201-00-111), University of Freiburg. Titanium biomaterial implant surfaces were kindly provided by Institut Straumann AG. The authors are grateful to Dr. Simon Berner for measurement of surface textures and contact angles as well as XPS analyses.

Disclosure Statement

The authors confirm that there are no conflicts of interest.

References

Kawahara

, Kawahara

, Hashimoto

, Takashima

, Ong

J.L.

Morphologic studies on the biologic seal of titanium dental implants. Report I. In vitro study on the epithelialization mechanism around the dental implant. Int J Oral Maxillofac Implants, 13:457. 1998.

Kawahara

, Kawahara

, Mimura

, Takashima

, Ong

J.L.

Morphologic studies on the biologic seal of titanium dental implants. Report II. In vivo study on the defending mechanism of epithelial adhesions/attachment against invasive factors. Int J Oral Maxillofac Implants, 13:465. 1998.

Rompen

, Domken

, Degidi

, Pontes

A.E.

, Piattelli

The effect of material characteristics, of surface topography and of implant components and connections on soft tissue integration: a literature review. Clin Oral Implants Res, 17,Suppl 2:55. 2006.

Rinke

, Ohl

, Ziebolz

, Lange

, Eickholz

Prevalence of periimplant disease in partially edentulous patients: a practice-based cross-sectional study. Clin Oral Implants Res, 22:826. 2011.

Koldsland

O.C.

, Scheie

A.A.

, Aass

A.M.

Prevalence of peri-implantitis related to severity of the disease with different degrees of bone loss. J Periodontol, 81:231. 2010.

Zitzmann

N.U.

, Berglundh

Definition and prevalence of peri-implant diseases. J Clin Periodontol, 35:286. 2008.

Ponsonnet

, Reybier

, Jaffrezic

, Comte

, Lagneau

, Lissac

et al. Relationship between surface properties (roughness, wettability) of titanium and titanium alloys and cell behaviour. Mater Sci Eng, C.23:551. 2003.

Ong

J.L.

, Carnes

D.L.

, Cardenas

H.L.

, Cavin

Surface roughness of titanium on bone morphogenetic protein-2 treated osteoblast cells in vitro. Implant Dent, 6:19. 1997.

Pilliar

R.M.

Overview of surface variability of metallic endosseous dental implants: textured and porous surface-structured designs. Implant Dent, 7:305. 1998.

10.

Le Guehennec

, Soueidan

, Layrolle

, Amouriq

Surface treatments of titanium dental implants for rapid osseointegration. Dent Mater, 23:844. 2007.

11.

Taborelli

, Jobin

, Francois

, Vaudaux

, Tonetti

, Szmukler-Moncler

et al. Influence of surface treatments developed for oral implants on the physical, biological properties of titanium. (I) Surface characterization. Clin Oral Implants Res, 8:208. 1997.

12.

Degidi

, Artese

, Scarano

, Perrotti

, Gehrke

, Piattelli

Inflammatory infiltrate, microvessel density, nitric oxide synthase expression, vascular endothelial growth factor expression, and proliferative activity in peri-implant soft tissues around titanium and zirconium oxide healing caps. J Periodontol, 77:73. 2006.

13.

Abrahamsson

, Berglundh

, Glantz

P.O.

, Lindhe

The mucosal attachment at different abutments. An experimental study in dogs. J Clin Periodontol, 25:721. 1998.

14.

Comut

A.A.

, Weber

H.P.

, Shortkroff

, Cui

F.Z.

, Spector

Connective tissue orientation around dental implants in a canine model. Clin Oral Implants Res, 12:433. 2001.

15.

Jansen

J.A.

, van de Waerden

J.P.

, Wolke

J.G.

, de Groot

Histologic evaluation of the osseous adaptation to titanium and hydroxyapatite-coated titanium implants. J Biomed Mater Res, 25:973. 1991.

16.

Kurashina

, de Lange

G.L.

, de Putter

, de Groot

Reaction of surrounding gingiva to permucosal implants of dense hydroxyapatite in dogs. Biomaterials, 5:215. 1984.

17.

McKinney

R.V.

Jr. , Steflik

D.E.

, Koth

D.L.

Evidence for a junctional epithelial attachment to ceramic dental implants. A transmission electron microscopic study. J Periodontol, 56:579. 1985.

18.

Kohal

R.J.

, Weng

, Bachle

, Strub

J.R.

Loaded custom-made zirconia and titanium implants show similar osseointegration: an animal experiment. J Periodontol, 75:1262. 2004.

19.

Jansen

J.A.

, de Wijn

J.R.

, Wolters-Lutgerhorst

J.M.

, van Mullem

P.J.

Ultrastructural study of epithelial cell attachment to implant materials. J Dent Res, 64:891. 1985.

20.

Kasten

F.H.

, Soileau

, Meffert

R.M.

Quantitative evaluation of human gingival epithelial cell attachment to implant surfaces in vitro. Int J Periodontics Restorative Dent, 10:68. 1990.

21.

Raisanen

, Kononen

, Juhanoja

, Varpavaara

, Hautaniemi

, Kivilahti

et al. Expression of cell adhesion complexes in epithelial cells seeded on biomaterial surfaces. J Biomed Mater Res, 49:79. 2000.

22.

Eisenbarth

, Meyle

, Nachtigall

, Breme

Influence of the surface structure of titanium materials on the adhesion of fibroblasts. Biomaterials, 17:1399. 1996.

23.

Guy

S.C.

, McQuade

M.J.

, Scheidt

M.J.

, McPherson

J.C.

3rd , Rossmann

J.A.

, Van Dyke

T.E.

In vitro attachment of human gingival fibroblasts to endosseous implant materials. J Periodontol, 64:542. 1993.

24.

Hakkinen

, Yli-Urpo

, Heino

, Larjava

Attachment and spreading of human gingival fibroblasts on potentially bioactive glasses in vitro. J Biomed Mater Res, 22:1043. 1988.

25.

Kokoti

, Sivropoulou

, Koidis

, Garefis

Comparison of cell proliferation on modified dental ceramics. J Oral Rehabil, 28:880. 2001.

26.

Sauberlich

, Klee

, Richter

E.J.

, Hocker

, Spiekermann

Cell culture tests for assessing the tolerance of soft tissue to variously modified titanium surfaces. Clin Oral Implants Res, 10:379. 1999.

27.

Roesch-Ely

, Steinberg

, Bosch

F.X.

, Mussig

, Whitaker

, Wiest

et al. Organotypic co-cultures allow for immortalized human gingival keratinocytes to reconstitute a gingival epithelial phenotype in vitro. Differentiation, 74:622. 2006.

28.

Winter

, Kohl

, Huppertz

, Herold-Mende

, Wiest

, Komposch

et al. Expression of mRNAs encoding for growth factors, ECM molecules, and MMP13 in mono-cultures and co-cultures of human periodontal ligament fibroblasts and alveolar bone cells. Cell Tissue Res, 319:467. 2005.

29.

Huang

M.C.

, Seyer

J.M.

, Thompson

J.P.

, Spinella

D.G.

, Cheah

K.S.

, Kang

A.H.

Genomic organization of the human procollagen alpha 1(II) collagen gene. Eur J Biochem, 195:593. 1991.

30.

Lukinmaa

P.L.

, Vaahtokari

, Vainio

, Thesleff

Expression of type I collagen pro-alpha 2 chain mRNA in adult human permanent teeth as revealed by in situ hybridization. J Dent Res, 71:36. 1992.

31.

Nakamura

, Sone

, Takahashi

, Mizoguchi

, Echigo

, Sasano

Expression of versican and ADAMTS1, 4, and 5 during bone development in the rat mandible and hind limb. J Histochem Cytochem, 53:1553. 2005.

32.

Van Der Pauw

MTM

, Everts

, Beertsen

Expression of integrins by human periodontal ligament and gingival fibroblasts and their involvement in fibroblast adhesion to enamel matrix-derived proteins. J Periodontal Res, 37:317. 2002.

33.

Bartold

P.M.

, Haynes

D.R.

Interleukin-6 production by human gingival fibroblasts. J Periodontal Res, 26:339. 1991.

34.

Yamano

, Al-Sowygh

Z.H.

, Gallucci

G.O.

, Wada

, Weber

H-P

, Sukotjo

Early peri-implant tissue reactions on different titanium surface topographies. Clin Oral Implants Res, 22:815. 2011.

35.

Chanut-Delalande

, Bonod-Bidaud

, Cogne

, Malbouyres

, Ramirez

, Fichard

et al. Development of a functional skin matrix requires deposition of collagen V heterotrimers. Mol Cell Biol, 24:6049. 2004.

36.

Konttinen

Y.T.

, Kaivosoja

, Stegaev

, Wagner

H.D.

, Levón

, Tiainen

V-M

et al. Extracellular Matrix and Tissue Regeneration Regenerative Medicine. Netherlands: Springer, 2011; 21–80.

37.

Parks

W.C.

, Wilson

C.L.

, Lopez-Boado

Y.S.

Matrix metalloproteinases as modulators of inflammation and innate immunity. Nat Rev Immunol, 4:617. 2004.

38.

Tomakidi

, Breitkreutz

, Fusenig

N.E.

, Zoller

, Kohl

, Komposch

Establishment of oral mucosa phenotype in vitro in correlation to epithelial anchorage. Cell Tissue Res, 292:355. 1998.

39.

Yeong

W.Y.

, Chua

C.K.

, Leong

K.F.

, Chandrasekaran

, Lee

M.W.

Comparison of drying methods in the fabrication of collagen scaffold via indirect rapid prototyping. J Biomed Mater Res B Appl Biomater, 82:260. 2007.

40.

Livak

K.J.

, Schmittgen

T.D.

Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods, 25:402. 2001.

41.

, Wu

, Shi

, Chen

, Wu

Migfilin and Mig-2 link focal adhesions to filamin and the actin cytoskeleton and function in cell shape modulation. Cell, 113:37. 2003.

42.

Bornstein

, Sage

E.H.

Matricellular proteins: extracellular modulators of cell function. Curr Opin Cell Biol, 14:608. 2002.

43.

Juhasz

, Murphy

G.F.

, Yan

H.C.

, Herlyn

, Albelda

S.M.

Regulation of extracellular matrix proteins and integrin cell substratum adhesion receptors on epithelium during cutaneous human wound healing in vivo. Am J Pathol, 143:1458. 1993.

44.

Mombelli

, Lang

N.P.

The diagnosis and treatment of peri-implantitis. Periodontol 2000, 17:63. 1998.

45.

Borsani

, Salgarello

, Mensi

, Boninsegna

, Stacchiotti

, Rezzani

et al. Histochemical and immunohistochemical evaluation of gingival collagen and metalloproteinases in peri-implantitis. Acta Histochem, 107:231. 2005.

46.

Kuula

, Salo

, Pirila

, Hagstrom

, Luomanen

, Gutierrez-Fernandez

et al. Human beta-defensin-1 and −2 and matrix metalloproteinase-25 and −26 expression in chronic and aggressive periodontitis and in peri-implantitis. Arch Oral Biol, 53:175. 2008.

47.

Konttinen

Y.T.

, Lappalainen

, Laine

, Kitti

, Santavirta

, Teronen

Immunohistochemical evaluation of inflammatory mediators in failing implants. Int J Periodontics Restorative Dent, 26:135. 2006.

48.

Kononen

, Hormia

, Kivilahti

, Hautaniemi

, Thesleff

Effect of surface processing on the attachment, orientation, and proliferation of human gingival fibroblasts on titanium. J Biomed Mater Res, 26:1325. 1992.

49.

Bacakova

, Filova

, Rypacek

, Svorcik

, Stary

Cell adhesion on artificial materials for tissue engineering. Physiol Res, 53,Suppl 1:S35. 2004.

50.

Chou

, Firth

J.D.

, Uitto

V.J.

, Brunette

D.M.

Effects of titanium substratum and grooved surface topography on metalloproteinase-2 expression in human fibroblasts. J Biomed Mater Res, 39:437. 1998.

51.

Tomakidi

, Schuster

, Breitkreutz

, Kohl

, Ottl

, Komposch

Organotypic cultures of gingival cells: an epithelial model to assess putative local effects of orthodontic plate and occlusal splint materials under more tissue-like conditions. Biomaterials, 21:1549. 2000.