**Design of Materials for Bone Tissue Scaffolds**

Editor

**Antonio Boccaccio**

MDPI ' Basel ' Beijing ' Wuhan ' Barcelona ' Belgrade ' Manchester ' Tokyo ' Cluj ' Tianjin

*Editor* Antonio Boccaccio Department of Mechanics, Mathematics and Management Politecnico di Bari Bari Italy

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Materials* (ISSN 1996-1944) (available at: www.mdpi.com/journal/materials/special issues/bone tissue scaffolds).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Volume Number*, Page Range.

**ISBN 978-3-0365-3954-6 (Hbk) ISBN 978-3-0365-3953-9 (PDF)**

© 2022 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

### **Contents**



### **About the Editor**

#### **Antonio Boccaccio**

Antonio Boccaccio holds an MS in Mechanical Engineering (graduated cum laude) (Politecnico di Bari, Italy, 2002) and a PhD in Bioengineering (Politecnico di Milano, Italy, 2006). Since December 2019, he has been the Associate Professor (ING-IND/15 Design Methods for Industrial Engineering) at the Politecnico di Bari. In 2005, he was a Visiting Research Scholar at the Centre for Bioengineering, Trinity College, Dublin. Professor Boccaccio's research interests are morphological optimization of biomaterials, modeling and simulation of biomedical devices and mechanobiological processes, and optical techniques for reverse engineering. He has authored more than 120 publications (56 peer-reviewed ISI journal papers, 3 editorials, 1 letter, 2 national journal papers, 10 book chapters, 3 invited lectures, 49 conference papers) and 1 European patent. He is a member of the Editorial Board of three internationally reputed journals and serves as a reviewer for more than 40 ISI journals. In 2012, he has been awarded the Fylde Electronics Prize for the best paper published in 2010 in the journal *Strain* by the British Society for Strain Measurement. He also ranked first (among 40 applicants) in a university competition for post-doctoral fellowships held in the Politecnico di Bari. In 2021, he received the Progetto Ingegneria award as the "Professor Most Voted" by the students of the degree course in Industrial Design, Politecnico di Bari.

### **Preface to "Design of Materials for Bone Tissue Scaffolds"**

The strong growth recently experienced by the manufacturing technologies, along with the development of innovative biocompatible materials, has allowed the fabrication of high-performing scaffolds for bone tissue engineering. The design process of materials for bone tissue scaffolds presently represents an issue of crucial importance and is being studied by many researchers throughout the world. A number of studies have been conducted, aimed at identifying the optimal material, geometry, and surface that the scaffold must possess to stimulate the formation of the largest amounts of bone in the shortest time possible. This book presents a collection of 10 research articles and 2 review papers describing numerical and experimental design techniques definitively aimed at improving the scaffold performance, shortening the healing time, and increasing the success rate of the scaffold implantation process.

> **Antonio Boccaccio** *Editor*

### *Editorial* **Design of Materials for Bone Tissue Scaffolds**

**Antonio Boccaccio**

Dipartimento di Meccanica, Matematica e Management, Politecnico di Bari, 70125 Bari, Italy; antonio.boccaccio@poliba.it; Tel.: +39-080-5963393

**Abstract:** The strong impulse recently experienced by the manufacturing technologies as well as the development of innovative biocompatible materials has allowed the fabrication of high-performing scaffolds for bone tissue engineering. The design process of materials for bone tissue scaffolds represents, nowadays, an issue of crucial importance and the object of study of many researchers throughout the world. A number of studies have been conducted, aimed at identifying the optimal material, geometry, and surface that the scaffold must possess to stimulate the formation of the largest amounts of bone in the shortest time possible. This book presents a collection of 10 research articles and 2 review papers describing numerical and experimental design techniques definitively aimed at improving the scaffold performance, shortening the healing time, and increasing the success rate of the scaffold implantation process.

**Keywords:** bone tissue engineering; porous materials; bone regeneration

Scaffolds for bone tissue engineering are porous materials that are used to reconstruct large dimensions bone defects. The ideal scaffold should satisfy to the following three principal requirements: (1) it should exhibit a structural response that is adequate and as close as possible to that of the tissues adjacent to the fracture site; (2) it should be biocompatible and biodegradable; (3) it should possess adequate surfaces capable of promoting the adhesion of mesenchymal stem cells, their proliferation and their subsequent osteogenic differentiation [1]. It is commonly known that the rate of bone tissue regeneration and the cellular response is significantly influenced by: (a) the scaffold mechanical behavior, which is, in turn, a function of the scaffold micro-architecture and of the mechanical properties of the material it is made from [2,3]; (b) the surface roughness status and the biological/chemical response of the scaffold/tissue interface surfaces to external factors [4]. The adhesion of stem cells to the scaffold surface as well as the tissue differentiation process occurring in the scaffold pores are regulated by very complex mechanobiological mechanisms taking place at both the micro- (i.e., some micrometers, approximately the dimension of a stem cell) and macro- (i.e., some hundreds of micrometers, corresponding to the typical dimensions of scaffold pores) levels, respectively [5–9]. The scaffold surface must be adequately structured to favor the adhesion of stem cells and their consequent differentiation. Similarly, the scaffold architecture must be properly shaped, and the scaffold material must be adequately designed to trigger favorable biophysical stimuli, leading to the formation of the bony tissue.

Many studies have recently been conducted to investigate the optimal manufacturing technologies that can be used to fabricate "smart and custom" scaffolds capable not only of guaranteeing the above-mentioned requirements, but also of satisfying the specific requests of the specific patient in whom it will be implanted [5]. One of the most recent research lines, in fact, has been focused on the design of "personalized" scaffolds that better suit the anthropometric features of the patient, thus allowing to achieve a successful follow-up in the shortest possible time [10]. Different studies have recently been published with the aim of better understanding the relationship between the scaffold geometry/material properties and the consequent mechanobiological phenomena taking place inside the scaffold

**Citation:** Boccaccio, A. Design of Materials for Bone Tissue Scaffolds. *Materials* **2021**, *14*, 5985. https://doi.org/10.3390/ma14205985

Received: 17 September 2021 Accepted: 6 October 2021 Published: 12 October 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

during the regeneration process. However, no clear explanations are yet available on the relationship existing between the mechanical/chemical environment and the consequent biological response of tissues occupying the scaffold pores. This Special Issue attempts to bridge the gap and to give a possible response to the open questions.

Most of the studies of the Special Issue developed innovative materials favoring the formation of new bone in the fracture site where the scaffold is implanted [11–16]. Three papers investigate the issues related to the geometry/dimensions that the scaffold pores must possess to guarantee an adequate mechanobiological response [10,17,18]. Finally, three articles deal with more clinical/applicative aspects [19–21].

The studies investigating innovative materials concern not only the material the scaffold is made from, but also all the materials in presence of which mesenchymal stem cells can be put to favor their adhesion, proliferation, and differentiation. In detail, Nicoara et al. [12], synthesized and characterized two types of materials—with antibacterial properties provided by silver nanoparticles (AgNPs)—based on hydroxyapatite and bacterial cellulose, that are known to possess excellent biocompatibility and bioactivity properties and are, hence, particularly suited to be used in the field of bone tissue engineering. The obtained composite materials were found to have a homogenous porous structure, a high water absorption capacity, and a considerable antimicrobial effect due to silver nanoparticles embedded in the polymer matrix. The fabrication of a composite bone cement made of graphene oxide and poly(methyl methacrylate) was described by Krukiewicz et al. [14], who investigated the potential of this cement to enhance the osteogenic differentiation of human primary mesenchymal stem and progenitor cells. Bastos et al. [15] developed an advanced three-dimensional (3D) biomaterial by integrating bioactive factors, such as lactoferrin and hydroxyapatite, within gellan gum spongy-like hydrogels. The authors demonstrated that that gellan gum spongy-like hydrogels gathered favorable 3D bone-like microenvironment with an increased human adipose-derived stem cells viability. Ishida et al. [16], evaluated starfish-derived β-tricalcium phosphate obtained by phosphatization of starfish-bone-derived porous calcium carbonate as a potential bone substitute material. They concluded that starfish-derived β-tricalcium phosphate may be effective for bone regeneration applications, such as in the treatment of fractures and bone loss. The osteoblastic features of adult mesenchymal stem cells integrated with 3D-printed polycarbonate scaffolds differentiated in the presence of oligostilbenes, such as resveratrol and polydatin, were investigated by Posa et al. [13]. They found that both resveratrol and polydatine stimulate the adhesion of the mesenchymal stem cells to the bone matrix protein osteopontin via αVβ<sup>3</sup> integrin and, specifically, polydatine treatment prompted a greater reorganization of this integrin in focal adhesion sites. The effects of a titanium surface coated with polylysine homopolymers on the cell growth of dental pulp stem cells and keratinocytes was investigated by Contaldo et al. [11]. They found an increase in cell growth for both cellular types cultured with polylysine-coated titanium compared to cultures without titanium and those without coating.

Very interesting are also the studies investigating the geometry of the scaffold pores, as well as the issues related to the structural response to mechanical loads and the scaffold porosity. Percoco et al. [17] and Rodríguez-Montaño et al. [10], using the mechanoregulation model by Prendergast et al. [22], determined the optimal dimensions that the pores of scaffolds 3D printed with the FDM technique and including spherical pores, respectively, must possess. In this model, the fracture site is modelled as a biphasic poroelastic material, and the biophysical stimulus that triggers the osteogenic differentiation of the mesenchymal stem cells is hypothesized to be a function of the octahedral shear strain and of the interstitial fluid flow measured in the regenerating tissue. The authors, by using this model, defined, via an optimization algorithm, the optimal dimensions of pores for different load values acting on the scaffold [10,17]. Martinez-Marquez et al. [18], in their review paper, used the quality by design system to explore the quality target product profile and ideal quality attributes of additively manufactured titanium porous scaffolds for bone regeneration with a biomimetic approach. The systematic literature

review presented an overview of the reported properties in research studies of fully porous titanium bone implants fabricated with additive manufacturing published in the last two decades. Unit cell geometry, porosity, elastic modulus, compressive yield strength, ultimate compressive yield strength, and compressive fatigue strength were systematically reviewed and benchmarked against the proposed ideal quality attributes.

The studies dealing with applicative/clinical aspects investigate very wide and interesting topics. The effects of chronic alcoholism on the repair of bone defects associated with xenograft was investigated by German et al. [21]. The interesting review paper by Stokovic et al. [19] summarizes the bone regeneration strategies and the animal models used for the initial, intermediate, and advanced evaluation of promising therapeutical solutions for new bone formation and repair. Dentistry issues were investigated by Grassi et al. [20], who evaluated the clinical success of horizontal ridge augmentation in severely atrophic maxilla using freeze-dried, custom-made bone harvested from the tibial hemiplateau of cadaver donors.

All the papers of the Special Issue were submitted to peer review, and thanks to the help of the reviewers, the quality of all the manuscripts was significantly improved. My special thanks go, therefore, to the authors for their excellent contribution, to the reviewers, for their invaluable help, as well as to the editorial staff of *Materials*, in particular to Ariel Zhou, Section Managing Editor for her kind assistance, competence and patience.

**Funding:** I thank the Italian Ministry of Education, University and Research under the Programme: (1) PON R&I 2014–2020 and FSC (Project 'CONTACT', ARS01\_01205); (2) 'Department of Excellence' Legge 232/2016 (Grant No. CUP - D94I18000260001), for the funding received.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


*Communication*

## **An Algorithm to Optimize the Micro-Geometrical Dimensions of Sca**ff**olds with Spherical Pores**

**Óscar Libardo Rodríguez-Montaño 1,2 , Carlos Julio Cortés-Rodríguez 1 , Antonio Emmanuele Uva 2 , Michele Fiorentino 2 , Michele Gattullo 2 , Vito Modesto Manghisi <sup>2</sup> and Antonio Boccaccio 2, \***


Received: 14 August 2020; Accepted: 11 September 2020; Published: 13 September 2020

**Abstract:** Despite the wide use of scaffolds with spherical pores in the clinical context, no studies are reported in the literature that optimize the micro-architecture dimensions of such scaffolds to maximize the amounts of neo-formed bone. In this study, a mechanobiology-based optimization algorithm was implemented to determine the optimal geometry of scaffolds with spherical pores subjected to both compression and shear loading. We found that these scaffolds are particularly suited to bear shear loads; the amounts of bone predicted to form for this load type are, in fact, larger than those predicted in other scaffold geometries. Knowing the anthropometric characteristics of the patient, one can hypothesize the possible value of load acting on the scaffold that will be implanted and, through the proposed algorithm, determine the optimal dimensions of the scaffold that favor the formation of the largest amounts of bone. The proposed algorithm can guide and support the surgeon in the choice of a "personalized" scaffold that better suits the anthropometric characteristics of the patient, thus allowing to achieve a successful follow-up in the shortest possible time.

**Keywords:** geometry optimization; computational mechanobiology; bone tissue engineering; python code; parametric CAD (Computer Aided Design) model

#### **1. Introduction**

One of the main issues recently investigated in the field of bone tissue engineering and that has received substantial attention is the identification of the optimal geometry of bony tissue scaffolds to support the numerous cellular activities involved in bone formation and regeneration [1]. Scaffolds are porous structures that mainly perform a dual function: transporting nutrients, waste, and oxygen, and a structural function consisting of transferring the load to the cells and regenerated tissues occupying their pores and to the adjacent tissues where they are implanted [2,3]. A large number of porous topologies have been studied from both the theoretical and the experimental point of view, but there is not yet a consensus between researchers regarding the geometry that the "optimal" scaffold should possess to maximize the amounts of regenerated bone [4]. However, some "general" guidelines are commonly accepted in the literature such as the range of the dimensions that pores have to possess to favor the regeneration process [5].

In general, bone tissue scaffolds can be classified into two principal categories: irregular and regular. Regular scaffolds are fabricated using advanced manufacturing processes such as additive layer manufacturing (ALM) that allow controlling with high precision the specific dimension of the single unit

cell the scaffold is made from. The irregular scaffolds are fabricated with conventional physical-chemical processes that allow controlling the average dimensions of the scaffold microarchitecture only on a statistical base [6]. A typical advantage of regular structures is the regularity of the scaffold domain that implies the regularity of the physical environment and hence the regularity of the mechanical stimulus acting on the regenerating tissue.

A very interesting scaffold topology is that including spherical pores. It is commonly known that the adhesion and differentiation of stem cells take place more easily on curved surfaces, especially on concave surfaces [5,7]. Scaffold topologies including spherical pores were recently produced with ALM techniques [8]. Spherical pores are also included in previously explored scaffold geometries such as FCC (face-centered cubic), BCC (body-centered cubic) [9,10], and Schwartz-P primitives [11–13]. However, no studies are reported in the literature optimizing the geometry of scaffolds with spherical pores, with the scope of maximizing the amounts of neo-formed bone. Here we aim to bridge this gap. We modeled the scaffold and the tissues occupying it as biphasic poroelastic materials and computed the biophysical stimulus acting on the tissue inside the scaffold pores according to the model of Prendergast et al. [14], as a function of the octahedral shear strain and the interstitial fluid flow. The objective of this study was to identify the optimal geometrical parameters of a regular scaffold with spherical pores and cylindrical interconnections that maximize the amounts of neo-formed bone. We found that this scaffold topology is particularly suited to bear shear loads. The proposed model fits well the requirements of so-called Precision Medicine (i.e., the branch of Medicine that studies personalized medical solutions for the specific requirements of the patient) and tries to answer the question about the optimal scaffold micro-geometry to achieve a successful follow-up in the shortest possible time.

#### **2. Materials and Methods**

#### *2.1. Unit Cell Geometry*

The parametric model of a scaffold occupying a cubic volume of side *L* = 2.548 mm and including 4 × 4 × 4 = 64 unit cells was developed. The same scaffold dimensions were utilized in previous studies [15,16]. The general purpose software Abaqus (version 6.12, Dassault Systèmes, Vélizy-Villacoublay, France) was utilized for both the parametric geometry modeling and the finite element analysis. Each unit cell is a hexahedron with a spherical cavity and cylindrical interconnections oriented along the orthogonal directions of the coordinate axes. It can be obtained as a Boolean subtraction of the volume of a sphere with cylinders from a cubic volume with the side *Luc* = *L*/4 (Figure 1). Depending on the diameter of the spherical surface *D<sup>s</sup>* , two different unit cell topologies can be designed: a "small" (S) topology where 0 < *D<sup>s</sup>* ≤ *Luc* and a "large" (L) topology where *Luc* < *D<sup>s</sup>* < *Luc* × √ 2 (Figure 2). Obviously, spherical diameters *D<sup>s</sup>* > *Luc* × √ 2 are not allowed, as the geometry deriving from such an assumption would lead to a scaffold unit cell completely different with respect to that hypothesized. Regarding the diameters of cylinders *Dc*, other constraints must be respected depending on the specific topology. In the case of Topology (S), the diameter of cylinders must satisfy the following inequality:

$$0 < D\_{\mathfrak{c}} \le D\_{\mathfrak{s}} / \sqrt{2},\tag{1}$$

In the section views obtained with a plane cutting the unit cell in half (Figure 3a), the figure of a square (represented with a dashed line, Figure 3) can be traced as the intersection of the edges of the cylinders. If this square is included within the edge of the spherical surface (highlighted in blue, Figure 3), the inequality (1) is verified. Inside the unit cell, a unique spherical surface can be identified that is interrupted by the cylindrical surfaces (Figure 3b). When the vertices of the square touch the spherical edge, the condition

$$D\_{\mathfrak{c}} = D\_{\mathfrak{s}} / \sqrt{2},\tag{2}$$

is reached. Finally, when the vertices of the square go beyond the spherical edges, only isolated (i.e., *D<sup>c</sup>* > *D<sup>s</sup>* / √ 2) or no (i.e., *D<sup>c</sup>* >> *D<sup>s</sup>* / √ 2) portions of spherical surface can be identified, and the geometry of the unit cell changes completely with respect to that hypothesized, which leads to the change in the scaffold connectivity.

π **Figure 1.** To build the scaffold unit cell (**c**), a boolean subtraction was carried out between a cubic volume (side *Luc* = *L*/4) (**a**) and the volume of a sphere (highlighted in blue) with cylinders (highlighted in green) oriented orthogonally according to the coordinate axes (**b**). The section A-A view (**d**) with the plane π (**c**), shows how the unit cell is interiorly made.

In the case of Topology (L), the diameter of cylindrical surfaces *D<sup>c</sup>* must satisfy the following inequality

$$\sqrt{\left(D\_{\rm s}^{2} - L\_{\rm uc}^{2}\right)} < D\_{\rm c} \le D\_{\rm s} / \sqrt{2} \,\,\,\,\tag{3}$$

In fact, to guarantee the "coherence" of the hypothesized scaffold geometry, the cylindrical diameter must be greater than the length of the chord *C* obtained by the intersection of the spherical edge with the edge of the cylindrical surface (Figure 4). The length of the chord is given by

$$\mathcal{C} = \sqrt{\left(D\_s^2 - L\_{uc}^2\right)},\tag{4}$$

The considerations regarding the figure of the square that can be traced in the section view as the intersection of the cylindrical edges continue to remain valid also in the case of the Topology (L) and, consequently, lead to define the upper limit for *D<sup>c</sup>* that must be *D<sup>c</sup>* ≤ *D<sup>s</sup>* / √ 2. Table 1 summarizes the constraint equations that *D<sup>s</sup>* and *D<sup>c</sup>* must satisfy to guarantee that the unit cell geometry remains the same, thus conserving its "intrinsic" coherence, for the variable values that *D<sup>s</sup>* and *D<sup>c</sup>* can assume.

≤ ≤ √2 **Figure 2.** Two different topologies can be built for the scaffold unit cell: "small" (S) (**a**) and "large" (L) (**b**). Topology (S) includes a spherical surface with 0 < *D<sup>s</sup>* ≤ *Luc* (**c**); Topology (L) includes a spherical surface with *Luc* < *D<sup>s</sup>* ≤ √ 2 × *Luc* (**c**).

√2 √2 **Figure 3.** (**a**) Section views—in the plane x–z—of the scaffold unit cell (topology (S) with indicated edges of the primitives (cube, cylinders, and sphere) utilized. When the square obtained by the intersection of the cylinders touches with its vertices, the spherical edge (in blue), the limit condition *D<sup>c</sup>* = *Ds*/ √ 2 is reached. For *D<sup>c</sup>* > *Ds*/ √ 2, the topology of the unit cell changes. (**b**) Section views—in the three-dimensional space—of the unit cell obtained for different values of *D<sup>s</sup>* and *Dc*.

ඥሺ<sup>௦</sup> <sup>ଶ</sup> − ௨ <sup>ଶ</sup> ሻ < ≤ ௦/√2

**Figure 4.** Schematic utilized to determine the equation constraint that the diameter of cylinders *D<sup>c</sup>* must satisfy in the case of Topology (L).

= ඥሺ<sup>௦</sup>

<sup>ଶ</sup> − ௨ <sup>ଶ</sup> ሻ ,

≤ √2

**Table 1.** Constraint equations that the diameter of the sphere *D<sup>s</sup>* and the cylinders *D<sup>c</sup>* must satisfy to guarantee the coherence of the scaffold geometry.


#### *2.2. Sca*ff*old Model and Applied Boundary and Loading Conditions*

The unit cell described above was mirrored with respect to different planes and replicated 64 times to generate the geometry of the entire scaffold (Figure 5). The model includes also the granulation tissue, highlighted in red (Figure 5), occupying the scaffold pores. Both the scaffold and the granulation tissue were modeled as biphasic poroelastic materials with the same material properties (Table 1) as those utilized in previous studies [15,17,18].

A rigid plate (highlighted in blue, Figure 5d,e) was fixed at the upper face of the scaffold-granulation tissue system using a tie constraint to uniformly transfer the load. A tie constraint between the scaffold and granulation tissue was also established to prevent any relative displacement between these two materials. On the bottom surface of the model, an encastre boundary condition was fixed, while for the outer surfaces of the granulation tissue, a pore pressure equal to zero was set to allow, according to Byrne et al. [19], the free exudation of fluid. Two different loading conditions were hypothesized: a compression (Figure 5d) and a shear (Figure 5e) load. The values of load per unit area *FUA* hypothesized in this study were the same as those utilized in a previous article [16]: in the case of compression load, 0.05, 0.1, 0.5, 1.0, and 1.5 MPa, and in the case of shear load, 0.01, 0.05, 0.1, 0.2, and 0.5 MPa. C3D4P tetrahedral elements available in Abaqus® were used to discretize the model. The average element size and the maximum deviation factor were set at 50 µm and 0.01, respectively.

A python script was generated that allows automatically (i) building the scaffold and the granulation tissue geometry; (ii) applying the boundary and the loading conditions; (iii) discretizing the model into finite elements; and (iv) running the finite element analyses. This script was then incorporated within an optimization code written in Matlab (Version R2016b, MathWorks, Natick, MA, USA) that, based on mechanobiological criteria deriving from the model of Prendergast et al. [14], allows the optimal scaffold geometry to be predicted.

**Figure 5.** The CAD models of scaffold (**a**) and granulation tissue (**b**) were assembled to generate the model (**c**) utilized in the study. Two different boundary and loading conditions were hypothesized to act on the model: a compression load **F** (|**F**| = *FUA* × *L* × *L*) on the upper surface and an encastre on the lower one (**d**); a shear load **F** (|**F**| = *FUA* × *L* × *L*) on the upper surface and an encastre on the lower surface (**e**).

#### *2.3. A Brief Outline of the Mechano-Regulation Model Implemented to Determine the Sca*ff*old Optimal Geometry*

Once the scaffold is implanted in the region with bone deficiency, mesenchymal stem cells (MSCs) migrate from the adjacent tissues, thus invading the scaffold. Therefore, MSCs start their differentiation process. The model of Prendergast et al. [14] assumes that the biophysical stimulus *S* that triggers

the differentiation process in the fracture domain is a function of the octahedral shear strain and of the interstitial fluid flow acting on the mesenchymal tissue. Depending on the values that *S* assumes, differentiation into different phenotypes, such as fibroblasts, chondrocytes, or osteoblasts, will be stimulated. The ranges of the biophysical stimulus *S* that determine the fate of the MSCs are described in the following inequalities:

$$\begin{aligned} S > 3 &\rightarrow \text{Fibroblasts (Fibrous tissue)}\\ 1 < S < 3 &\rightarrow \text{Chondrecovery (Cartilage)}\\ 0.53 < S < 1 &\rightarrow \text{Osteoblats (Immuture bone)}\\ 0.01 < S < 0.53 &\rightarrow \text{Osteoblats (Mature bone)}\\ 0 < S < 0.01 &\rightarrow \text{Bone resolution} \end{aligned} \tag{5}$$

Further details on the mechano-regulation algorithm can be found in previous studies [20,21].

#### *2.4. Optimization Algorithm*

The optimization algorithm aims to identify the scaffold geometry that allows maximizing the amounts of neo-formed bone for each value of force per unit area *FUA* hypothesized in the study (Figure 6).

**Figure 6.** Schematic of the optimization algorithm implemented to determine the optimal scaffold geometry.

In detail, the algorithm, written in Matlab, employs the *fmincon* function from the Matlab optimization toolbox to determine the optimal values of the design variables *D<sup>s</sup>* and *D<sup>c</sup>* that maximize *BO%*, the percentage of the scaffold volume occupied by mature bone. In each optimization cycle, the values of *D<sup>s</sup>* and *D<sup>c</sup>* are perturbed and entered into a python script. This script is given in input to Abaqus, which builds the model, applies the boundary and loading conditions, generates the mesh, and runs the finite element analysis. Then, the algorithm reads the results of the FEM analysis, computes the biophysical stimulus *S,* and compares it with the boundary values reported in the inequalities (5). At this point, it computes *BO%*, the percentage of the scaffold volume occupied by mature bone, as the ratio between the volume of the elements with *S* that satisfy the inequality 0.01 < *S* < 0.53, and the total volume of the scaffold *L* × *L* × *L*. The algorithm perturbs so many times the values of *D<sup>s</sup>* and *D<sup>c</sup>* until the maximum value of *BO%* is determined. Once this occurs, the optimization algorithm stops and outputs the predicted optimal values of the design variables *D<sup>s</sup>* and *D<sup>c</sup>* as well as the value of the percentage *BO%*, which represents the maximum percentage of the scaffold volume that can be occupied by bone for a given load value. During the optimization process, *D<sup>s</sup>* and *D<sup>c</sup>* can assume variable values concerning both (L) and (S) Topology but must always satisfy the constraint equations summarized in Table 1.

All the optimization analyses were conducted on an HP XW6600-Intel®Xeon®DualProcessor E5-5450 3 GHz–32 Gb RAM workstation (Intel Corporation, Mountain View, CA, USA) and required approximately 1500 h of computation.

#### **3. Results and Discussion**

The optimized scaffold geometries predicted by the proposed algorithm in the case of compression load present spherical pores and cylindrical interconnections that become smaller for increasing values of the load (Figure 7). This can be explained with the argument that as the load increases, the biophysical stimulus acting on the mesenchymal tissue increases too, thus favoring the formation of soft tissues like cartilage and fibrous tissue. Hence, the algorithm to counterbalance this tends to increase the scaffold stiffness by decreasing the dimensions of the spherical pores and the cylindrical connections (Figure 7a,b). Comparing the percentages *BO%* with those predicted in a previous study [20] for regular scaffolds based on a hexahedron unit cell with elliptic and rectangular extrusions, we found that scaffolds with rectangular extrusions perform always better than those with spherical pores. Conversely, those with elliptic extrusions work better than the scaffolds with spherical pores only for high load values (Figure 7c). When the load is high, in fact, elliptic and rectangular extrusions tend to orientate according to the load direction, which makes the scaffold more "suited" to bear and transfer the compression load acting on it.

The optimal geometries predicted in the case of shear load present pores with dimensions that get increasingly smaller as we move towards higher load values (Figure 8a,b). Interestingly, in this case, the scaffold with spherical pores performs, for all the hypothesized values of shear load, better than those with elliptic and rectangular extrusions (Figure 8c).

**Figure 7.** (**a**) Optimized scaffold geometries (section views A-A), (**b**) optimal values of *D<sup>s</sup>* and *Dc*, and (**c**) percentage of the scaffold volume occupied by mature bone, predicted by the optimization algorithm for different values of the compression load. The percentages of bone are compared with those predicted for scaffolds with hexahedron unit cells including elliptic and rectangular extrusions [20].

**Figure 8.** (**a**) Optimized scaffold geometries (section views A-A), (**b**) optimal values of *D<sup>s</sup>* and *Dc*, and (**c**) percentage of the scaffold volume occupied by mature bone, predicted by the optimization algorithm for different values of the shear load. The percentages of bone are compared with those predicted for scaffolds with hexahedron unit cells including elliptic and rectangular extrusions [20].

In general, the biophysical stimulus *S* acting on the mesenchymal tissue assumes higher values in the proximity of the spherical pores, while smaller values are observed in the proximity of the cylindrical interconnections (Figure 9). The regularity of the scaffold geometry leads to a regular distribution of the biophysical stimulus that is repeated with approximately the same characteristics as many times as the cells of the scaffold. Such a spatial distribution demonstrates that the biophysical stimulus depends on the scaffold geometry and on how this transfers the load to the mesenchymal tissue.

**Figure 9.** Spatial distribution of the normalized biophysical stimulus *S*/*Smax* computed for a scaffold (*D<sup>s</sup>* = 0.425 mm *D<sup>c</sup>* = 0.275 mm) subjected to the compression load of *FUA* = 0.5 MPa.

μ The proposed study has some limitations. First, the model includes a spherical pore the diameter of which was optimized based on the mechanobiological model of Prendergast et al. [14]. As demonstrated in previous studies [16,20], scaffolds oriented according to the load direction perform better than those without a specific orientation [18]. To make the proposed geometry "oriented" according to the load direction, the spherical surface should be changed with that of prolate or oblate spheroids. In this case, the number of variables to optimize are two: the minor and the major axis of the spheroid. With this strategy, the spheroidal surface would properly orient, thus making the scaffold more "suited" to bear and transfer the load acting on it [22–24]. This topic will be the objective of future studies. Second, a clear and direct experimental study that demonstrates the correctness of the predictions of the proposed model is, at the moment, lacking. In general, it is difficult to systemically study the effects of scaffold geometry on the process of bone tissue regeneration. The identification of the geometrical features that principally affect the tissue differentiation process occurring in a scaffold requires the systematic study of different scaffold geometries. However, at the moment, no such studies are available in the literature [7]. Third, a simplified hypothesis was followed regarding the diffusion of mesenchymal stem cells once the scaffold is implanted. The event in which the MSCs migrate from the adjacent tissues and invade the scaffold could not take place *sic et simpliciter*. In fact, once a scaffold is implanted, it will be most likely infiltrated with blood, which clots within a few minutes, thus clogging the pores of the scaffolds. Moreover, other cells such as connective tissue fibroblasts could compete with MSCs to colonize the scaffolds. However, in the case where MSCs are the only cells entering the scaffold, having a highly osteogenic microarchitecture, once the new bone is deposited, it will prevent further MSCs inwards migration and bone ingrowth. Studies on the transient phase of the MSCs migration and diffusion through the scaffold should be carried out in the future. Fourth, the proposed algorithm allows to determine the optimal dimensions of the spherical pores and the cylindrical interconnections. However, this poses relevant technological issues in the sense that the proposed approach requires the implementation of additive manufacturing techniques that must guarantee adequate precision for the produced scaffolds. Stereolithography is one of the most powerful and versatile additive manufacturing techniques [25]. It has the highest fabrication accuracy, which ranges from 1.2 to 200 µm [26]. Fused deposition modelling (FDM) was demonstrated to have the lowest precision [27]. The experimental tests previously conducted with FDM demonstrated that this technique is suitable to build accurate scaffold samples only in the cases where the strand diameter is close to the nozzle diameter. Conversely, when a large difference exists, large fabrication errors can

be committed on the diameter of the filaments [17]. Scaffolds fabricated with selective laser sintering (SLS) show dimensional deviations—with respect to the nominal dimensions—up to 7.5% [28]. Fifth, the scaffold model investigated has rather small dimensions with respect to those of the scaffolds commonly used in the clinical context. In principle, using a larger scaffold model is possible but poses serious issues of computational power. Sixth, the time variable was not included in the proposed algorithm, i.e., we do not simulate how the bone regeneration process takes place in the scaffold and optimize the scaffold geometry based on the "picture" taken at the instant of time zero, after its implantation. In reality, the inclusion of the time variable requires very high computational power and a computational time tremendously longer than the time required to perform the optimization analyses carried out in this study. In fact, for each candidate geometrical solution, the algorithm should ideally predict how the bony tissue growths and how the scaffold dissolves. This series of analysis cycles should be repeated as many times as the cycles required by the optimization algorithm, which leads to computational times at least two orders of magnitude larger than those required in this study. Increases in computational power will ultimately allow simulating the bone regeneration and the scaffold dissolution processes to optimize the scaffold geometry on a temporal perspective as well as modelling scaffolds with dimensions closer to those actually employed in clinical practice.

Despite these limitations, the proposed model shows a mechanical behavior consistent with that of spongy bone. In fact, if we compute the ratio *Eapp* /*E*, where *Eapp* is the "apparent" Young's modulus of the scaffold considered in its entirety and *E* = 1000 MPa is Young's modulus of the material the scaffold is made from (Table 2), we find values falling within the variability range of this ratio experimentally measured for cancellous bone (Figure 10).


**Table 2.** Material properties utilized in the model of scaffold and granulation tissue [15,17,18].

To compute the ratio *Eapp* /*E*, three different finite element models of the sole scaffold (i.e., the granulation tissue was removed) were built, with the following pairs of *D<sup>s</sup>* and *D<sup>c</sup>* values expressed in millimeters [mm]: (*D<sup>s</sup>* = 0.85; *D<sup>c</sup>* = 0.55), (*D<sup>s</sup>* = 0.75; *D<sup>c</sup>* = 0.5), (*D<sup>s</sup>* = 0.65; *D<sup>c</sup>* = 0.45), which are close to the typical dimensions of pores commonly adopted in scaffolds for bony tissue [29,30]. These models were clamped on the lower base and subjected to a compression load of *FUA* = 0.1 MPa. The displacement *u*<sup>2</sup> (Figure 10a) produced by the load was computed with Abaqus and used to determine the apparent Young's modulus as:

$$E\_{app} = F\_{\text{ULA}} \times L / \mu\_{\text{2}} \tag{6}$$

Interestingly, the values of the ratio predicted numerically are consistent with those measured experimentally [31,32] on samples of human spongy bone (Figure 10b). Furthermore, if we compute for the three models described above the scaffold volume fraction *V<sup>f</sup>* , i.e., the ratio between the volume of the scaffold *V<sup>s</sup>* and the total volume of the model *Vtot* = *L* × *L* × *L*, we find values that are consistent with those experimental reported by Snyder and Hayes [33] and measured for human spongy bone (Figure 10c).

**Figure 10.** (**a**) *u*<sup>2</sup> displacement field of the scaffold models subjected to a compression load of 0.1 MPa. (**b**) Values of the ratio *Eapp* /*E* computed for the three models and compared with those experimentally measured (represented with the red lines) for cancellous bone. (**c**) Scaffold volume fraction values compared with the volume fraction of human spongy bone.

The proposed model fits well the requirements of so-called Precision Medicine. The optimization algorithm presented in this article represents a possible approach to try to identify, given the specific patient with her/his specific anthropometric characteristics (i.e., macroscopic characteristics of the patient, such as weight, height, and geometric parameters of posture, that is, all the characteristics that allow identifying the boundary and loading conditions that act on a given anatomical region when a specific activity is performed), which are the optimal dimensions of the scaffold micro-geometry to achieve a successful follow-up with the formation of the largest amounts of bone in the shortest possible time? In fact, if one knows the anthropometric characteristics of the patient, they can hypothesize the possible value of load acting on the scaffold that will be implanted, and through diagrams such as those shown in Figures 7b and 8b, they can determine the optimal dimensions of the scaffold that favor the formation of the largest amounts of bone (Figure 7b). Furthermore, the proposed approach can support the surgeon in the choice of the best scaffold to implant in the specific fracture site of the patient. In fact, the surgeon has nowadays a very large range of scaffold geometries available on the market and hence has to choose the most suitable one for the specific requirements of the patient. For example, if, based on the anthropometric characteristics and the anatomical region of the fracture site, it is found that the scaffold will be subjected mainly to compression loading, the surgeon will choose the scaffold with rectangular extrusions (Figure 7c). If, on the other hand, it is found that the acting load will be mainly shear, then the surgeon will choose the scaffold with spherical pores (Figure 8c).

#### **4. Conclusions**

In this study, using a mechanobiology-based optimization algorithm, we computed the optimal dimensions of the micro-architecture of scaffolds including spherical pores and cylindrical interconnections. The optimization algorithm perturbs the scaffold geometry until the specific dimensions that favor the formation of the largest amounts of bone are identified. The proposed algorithm can guide and support the surgeon in the choice of a "personalized" scaffold that better suits the anthropometric characteristics of the patient, thus allowing to achieve a successful follow-up in the shortest possible time.

**Author Contributions:** Conceptualization, Ó.L.R.-M. and A.B.; methodology, Ó.L.R.-M. and A.B.; software, Ó.L.R.-M.; validation, A.B.; formal analysis, Ó.L.R.-M. and A.B.; writing—original draft preparation, Ó.L.R.-M. and A.B.; writing—review and editing, Ó.L.R.-M., A.B., C.J.C.-R., A.E.U., M.F., M.G., and V.M.M.; visualization, C.J.C.-R., A.E.U., M.F., M.G., and V.M.M.; project administration, C.J.C.-R. and Ó.L.R.-M.; funding acquisition, C.J.C.-R. and Ó.L.R.-M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research has been made possible by the collaboration between the Universidad Nacional de Colombia and Polytechnic University of Bari and was supported by the grant 647-2015 from the Colombian Ministry of Science, Technology and Innovation (MINCIENCIAS).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Titanium Functionalized with Polylysine Homopolymers: In Vitro Enhancement of Cells Growth**

**Maria Contaldo 1, \* ,† , Alfredo De Rosa 1,† , Ludovica Nucci 1 , Andrea Ballini 2,3 , Davide Malacrinò 4 , Marcella La Noce 5 , Francesco Inchingolo 6 , Edit Xhajanka 7 , Kenan Ferati 8 , Arberesha Bexheti-Ferati 8 , Antonia Feola 9,‡ and Marina Di Domenico 3,10, \* ,‡**


**Abstract:** In oral implantology, the success and persistence of dental implants over time are guaranteed by the bone formation around the implant fixture and by the integrity of the peri-implant mucosa seal, which adheres to the abutment and becomes a barrier that hinders bacterial penetration and colonization close to the outer parts of the implant. Research is constantly engaged in looking for substances to coat the titanium surface that guarantees the formation and persistence of the peri-implant bone, as well as the integrity of the mucous perimeter surrounding the implant crown. The present study aimed to evaluate in vitro the effects of a titanium surface coated with polylysine homopolymers on the cell growth of dental pulp stem cells and keratinocytes to establish the potential clinical application. The results reported an increase in cell growth for both cellular types cultured with polylysine-coated titanium compared to cultures without titanium and those without coating. These preliminary data suggest the usefulness of polylysine coating not only for enhancing osteoinduction but also to speed the post-surgery mucosal healings, guarantee appropriate peri-implant epithelial seals, and protect the fixture against bacterial penetration, which is responsible for compromising the implant survival.

**Keywords:** cell growth; titanium; polylysine; dental implants; implantology; biomaterials; epithelial growth

#### **1. Introduction**

Dental implants are multi-material prostheses that replace tooth roots with screwlike metal fixtures surgically inserted into the edentulous bone that are connected by the

**Citation:** Contaldo, M.; De Rosa, A.; Nucci, L.; Ballini, A.; Malacrinò, D.; La Noce, M.; Inchingolo, F.; Xhajanka, E.; Ferati, K.; Bexheti-Ferati, A.; et al. Titanium Functionalized with Polylysine Homopolymers: In Vitro Enhancement of Cells Growth. *Materials* **2021**, *14*, 3735. https:// doi.org/10.3390/ma14133735

Academic Editors: Javier Gil and Paolo Cappare

Received: 30 May 2021 Accepted: 1 July 2021 Published: 3 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

abutment with an artificial crown that replaces the missing tooth, looking and acting identical to the real one (Figure 1).

**Figure 1.** Schematic representation of a dental implant (**a**) and the oral structures (peri-implant mucosa and peri-implant bone) surrounding the fixture (**b**), surgically inserted in the bone. Original figures made by D.M. with SOLIDWORKS ® (CSWP-MBD Version, 2021, SolidWorks, Dassault systems, Waltham, MA, USA).

Dental implants fixtures are generally composed of biomedical titanium and its alloys [1], as they are biocompatible as well as resistant to corrosion and strength [2].

Numerous surgical protocols and variables may affect dental implant placement, and, over the years, novel implantology procedures have been constantly proposed [3–6].

The main sign of the success of a dental implant is its capability to integrate its shape with the bone and to induct the formation of novel bone around it; these properties are defined as "osseointegration"—"the close contact between the bone and an implant material in histological sections" [7,8]—and "osteoinduction"—the ability to induce the osteogenesis of new mineralized bone around the implant surfaces, thus firmly blocking the fixture within the bones of the jaws [9].

In addition to different surgical protocols [10], geometry modifications [5,11] and various surface treatments for increasing surface roughness [1,12], such as acid-etching, grit-blasting, titanium plasma-spraying, or anodization [13], as well as the use of various coatings to make the titanium surface bioactive [14–17] are responsible for empowering the wettability, bone anchoring, and biomechanical stability between the implant–bone interfaces [3,6,10,12,18], thus increasing osteoinduction and osteointegration.

Among the coating substances, the polyaminoacid poly-L-lysine has been reported to be able to bridge the cell-adhesion trough covalent attachments to cysteine in the bone [19–21].

A model of study on the osteogenic effects of substances is the use of human dental pulp stem cells [22–24], which previously has been proven to be involved in bone–implant osseointegration [25–28]. In details, the role of induced pluripotent stem cells in dentistry has been recently discussed and the use of autologous dental-derived stem cells has been proposed for bone tissue regeneration, as less invasive and more predictable alternative to conventional tissue regenerative procedures [29].

Furthermore, the mechanisms underlying the potential effects of poly-L-lysine on these kinds of cells have been reported both in vitro [20] and in vivo on sheep animal models [21].

Bacterial-induced inflammation of the soft tissues surrounding the abutment is the main cause of failure of the osteointegration immediately after the fixture placement and during the years. To avoid bacterial penetration and contamination of the peri-implant bone, which is responsible for inflammation and bone loss, the integrity of the peri-implant seal is crucial [30]. Otherwise, peri-implant inflammation occurs, and the implant survival is compromised. [31,32].

Therefore, a good epithelial attachment between the implant and the peri-implant mucosa is fundamental to achieve and maintain the osteointegration [30,33–35], and it is essential to maintain an intact oral epithelial barrier, with no local and systemic risk factors, as bacterial plaque, to offer good resistance to mechanical stress that is both physiological and pathological.

The present work aimed to confirm the in vitro effects of titanium functionalized with a poly-L-lysine coating on human dental pulp stem cells (hDPSCs), which are responsible for osteogenesis, and evaluate analogues effects on keratinocyte cell lines (HaCaT), which are responsible for epithelial attachment of the mucosa surrounding the abutment, to hypothesize a potential improvement of implant osteointegration and the potential use of poly-L-lysine for rapid mucosal healing after the implant placement and during years to preserve the health of peri-implant mucosa.

#### **2. Materials and Methods**

Machined clean square plates (1 cm × 1 cm in size; 0.2 mm thick) made up of 5-Ti-6Al-4V ELI alloy (Klein s.r.l., Milan, Italy) (Figure 2) were sterilized with ethanol 70%, dried under a fume hood, and used in six types of experiments: hDPSCs cultures alone (standard condition), hDPSCs cultures with titanium, and hDPSCs cultures poly-L-lysinecoated titanium (Figure 3a); HaCaT immortalized human keratinocyte line cultures alone, HaCaT cultures with titanium, and HaCaT cultured with poly-L-lysine-coated titanium (Figure 3b). In each experiment, cell viability and proliferation were assessed, as reported below. Sterilized titanium plates were coated with poly-L-lysine incubating at 37 ◦C for 30 min with a solution containing 0.01% poly-L-lysine and then dried and washed twice with sterile water. After this, cells were cultured on the disks.

#### *2.1. hDPSCs Culture and Growth Curve*

Experimental procedures were conducted following our previous experience in the field and according to the manufacturer's specifications [22,23,25,36–40].

Each patient or guardian gave informed consent to tooth extraction obtained with piezo-surgery technology, which was in accordance with the Declaration of Helsinki, for re-use of biospecimens in research applications. Moreover, the study was approved by the Independent Ethical Committee of University Hospital of Bari, Italy (protocol number 155/2021, 27 January 2021). With the purpose to preserve dental tissues for consequent cell isolation and expansion, piezo-surgery technology enables selective tissue cutting, and consequently, tooth buds or embedded third molars can effortlessly be removed from bones with slight wound to periodontal fibers or bud follicles.

In addition, tooth extraction, especially by piezosurgery technique, can be considered less invasive in comparison to bone marrow or other tissues biopsy [22].

Briefly, the pulp was removed and immersed for 1 h at 37 ◦C in a digestive solution of 3 mg/mL of type I collagenase and 4 mg/mL of dispase in PBS (phosphate buffered saline) containing 40 mg/mL of gentamicin. Once digested, the solution was filtered through 70 µm Falcon strainers (Becton & Dickinson, Franklin Lakes, NJ, USA). Cells were cultured in standard medium consisting of Dulbecco's modified Eagle's medium (DMEM) with 100 units/mL of penicillin, 100 mg/mL of streptomycin, and 200 mM l-glutamine (all purchased from Gibco), supplemented with 10% fetal bovine serum (FBS) (Invitrogen, Waltham, MA, USA). Cells were maintained in a humidified atmosphere under 5% CO<sup>2</sup> at 37 ◦C, and the media were changed twice a week.

At first passage of culture, cells were seeded at a density of 150.00 cells/titanium implant—with and without poly-L-lysine homopolimers coating—and in standard condition. After 1 h of incubation in 100 µL of culture medium to allow cell attachment, the cell implants and cells cultured without implants were incubated in DMEM at 10% of FBS (fetal bovine serum) into an incubator at 37 ◦C in a humidified atmosphere consisting of 5% CO<sup>2</sup> and 95% O<sup>2</sup> for 24, 48, and 72 h. For each time, an aliquot of cell suspension was diluted with 0.4% trypan blue (Sigma Aldrich, St. Louis, MO, USA), pipetted onto a haemocytometer, and counted under a microscope at 200× magnification. Live cells excluded the dye, whereas dead cells admitted the dye and were consequently stained intensely with trypan blue. The number of viable cells for each experimental condition was counted and represented on a linear graph.

#### MTT Analyses

In order to evaluate the cytotoxicity of titanium implants on cells, MTT assay (3- (4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) was used [41,42]. Cells, at a density of 300.00 cells/implant with and without poly-L-lysine coating and cells cultured in standard condition (hDPSCs cultured in tissue culture polystyrene (TCP) without titanium and polylysine) were plated in DMEM at 10% FBS for 24, 48, and 72 h. After each time point, the medium was removed, and 200 µL of MTT (Sigma, Milan, Italy) solution (5 mg/mL in DMEM without phenol red) and 1.8 mL of DMEM were added. Four hours later, the formazan precipitate was dissolved in 100 µL dimethyl sulfoxide, and then, the absorbance was measured in an ELISA reader (Thermo Molecular Devices Co., Union City, NJ, USA) at 550 nm. The mean and the standard deviations were obtained from three different experiments of the same specimen.


**Figure 2.** On top, the machined clean titanium plate at SEM. FOV: 134 µm, Mode: 15 kV—Point, Detector: BSD Full. On bottom, the chemical composition analysis of the titanium surface, in spot 1, pointed by a cross in the figure.

**Figure 3.** Schematic representation of the experiments. (**a**) hDPSCs cultured alone, on titanium plates, and on titanium plates coated with poly-L-lyisine. (**b**) HaCaT cells cultured alone, on titanium plates, and on titanium plates coated with poly-L-lyisine.

#### *2.2. HaCaT Cells Culture and Growth Curve*

HaCaT were cultured in complete culture medium consisting of DMEM (Sigma D5796, Sigma Aldrich, St. Louis, MO, USA) with 1% penicillin/streptomycin (Sigma P0781, Sigma Aldrich, St. Louis, MO, USA), 2 mM glutamine (Sigma G7513, Sigma Aldrich, St. Louis, MO, USA), and supplemented with 10% fetal bovine serum (Sigma F7524, Sigma Aldrich, St. Louis, MO, USA) [43]. All procedures were performed under sterile conditions under a NuAire laminar flow biological hood.

The cultures were expanded in plates every three days in an incubator under 5% CO<sup>2</sup> at 37 ◦C (RH = 95%), until the required number of cells was reached. Then, 5 × 10<sup>5</sup> cells were subsequently transferred to plates containing titanium alone and titanium with poly-L-lysine coating to promote engraftment.

In order to highlight cell clones adhering to titanium, after 48 h, the titanium plates with and without poly-L-lysine coating were removed from the culture and, after suitable washing with PBS twice, they were placed in plates containing only fresh culture medium to observe cell viability.

The cytotoxicity check was performed by culturing the cells in the absence and presence of the titanium plate (with and without poly-L-lysine coating) and by evaluating their viability after replacement in a new fresh medium.

After 72 h, all cells were trypsinized, collected, and evaluated for viability according to tryp blue method using the Burker chamber count (Invitrogen, Milan, Italy).

#### *2.3. Statistical Analyses*

Student's test was used for statistical evaluation. A *p*-value < 0.05 was considered significant.

#### **3. Results**

#### *3.1. hDPSCs Growth Curves Analyses*

Cell growth analysis and viability staining with trypan blue showed that hDPSCs cultured in standard condition and on titanium with and without poly-L-lysine showed the same trend in growth; however, while the titanium alone slightly negatively affects the viability for cells (*p* < 0.01), the cell growth on the poly-L-lysine coated titanium was noticeably increased (*p* < 0.001) (Figure 4).

**Figure 4.** Cell growth analyses. Although hDPSCs cultured in standard condition and on titanium with and without poly-L-lysine showed the same trend in growth, in the culture with titanium coated with poly-L-liysine, the cell growth was higher than the hDPSCs alone and hDPSCs with only titanium. \*\* *p* < 0.01, \*\*\* *p* < 0.001 compared to the hDPSCs.

#### *3.2. MTT Evaluation in hDPSCs*

To evaluate how the titanium affected the viability and proliferation of hDPSCs, MTT analyses were performed. hDPSCs were cultured on titanium with and without poly-Llysine coating for 24, 48, and 72 h. Results showed that titanium was not cytotoxic. In addition, there were no changes in terms of proliferation between cells cultured in standard condition and cells seeded on titanium, while the cells seeded on titanium coated with poly-L-lysine showed higher proliferation (*p* < 0.001) (Figure 5).

**Figure 5.** MTT evaluation. Titanium did not show cytotoxicity, and the proliferation of cells seeded on titanium was similar to those of cultured in standard condition, but in titanium coated with poly-L-liysine, cell proliferation was higher at 24, 48, and 72 h. \*\*\* *p* < 0.001 compared to the hDPSCs.

#### *3.3. HaCaT Viability and Proliferation: Mucoproliferative Effects of Titanium*

To evaluate the viability, HaCaT cells were cultured in the presence of titanium precoated with poly-L-lysine. As shown in Figure 6, the mere presence of titanium coated with poly-L-lysine did not affect HaCaT cell viability, as the cells at the bottom of the plate showed normal health morphology [44]. Moreover, cells cultured in the presence of titanium showed an increase in cell proliferation of 40% after 72 h compared to plates containing cells in the absence of titanium. To further demonstrate the presence of live cells adherent on titanium, after 48 h of culture, the titanium plate was placed in a new well with only fresh medium, and even here, adherent keratinocytes were still appreciated. This evidence demonstrates that titanium was found to be a suitable substrate for the viability and growth of keratinocytes in the presence of a culture medium (Figure 6).

**Figure 6.** (**a**) HaCat cells culture on the titanium plate coated with poly-L-lysine. (**b**) After a further 48 h, the titanium plate with the adherent HaCaT cells was placed in a new well with a fresh medium, where vital and growing cells were appreciated adhering on the bottom of the well, as shown in figure (**b**) (optical microscopy, original magnification ×10).

#### **4. Discussion**

The titanium alloys used in dentistry are biocompatible and not cytotoxic, but their surface is also inert, thus not affecting positively the osteoinduction. To empower dental implant osteoinduction, the titanium surface can be functionalized by coating it with a series of bioactive compounds and substances. For this purpose, various coatings have been proposed: nanoparticles of silver, copper, and zinc, as sanitizing agents, and antibacterial and bioactive substances [14], such as quaternary ammonium ions and chlorhexidine, antibiotics, or antimicrobial peptides [15]; calcium-phosphate alone [16] or hydroxyapatite or octacalcium phosphate complexes [17]. These substances are used to make the titanium surface bioactive to improve osteoinduction, by adding, in some cases, antibacterial properties.

Poly-L-lysine is a polyaminoacid carrying positive charges, which increase cellular adhesion on different substrates, and it has been variously reported as an additional coating to titanium surfaces [19–21]. In 2005, Spoerke et al. [19] first reported that a nanotextured hybrid titanium coating made up of poly-L-lysine 14% by weight added to calcium phosphate was able to enhance the surface area of the implant and to potentiate the bioactivity of the calcium phosphate alone, by the presence of poly-L-lysine in bridging the cell-adhesion through covalent attachments to cysteine in the bone.

Different studies have reported the osteogenic effects of poly-L-lysine on dentalderived stem cells [22–24] and their involvement in bone–implant osteointegration [25–28]. In 2011, Galli et al. [20] described the potential mechanisms by which poly-L-lysine can enhance osteogenesis, thus reporting that hMSCs (human mesenchymal stem cells) and hDPSCs cultured on poly-L-lysine-treated titanium (Ti6Al4V) showed significantly higher expression of bone marker genes, produced a higher quantity of calcium deposits, and showed higher cell viability after 12 h of culture in comparison with the cells on the untreated titanium. These effects were allowed both by the poly-L-lysine positive charges and its interaction with β-integrin and other molecules from the extracellular matrix (as collagen I, fibronectin, and vitronectin) and their adhesion receptors on the studied cells, thus activating the intracellular signaling cascade responsible for the upregulation of osteogenic markers genes. Among the osteogenic markers activated, alkaline phosphatase is responsible for focal adhesion kinase (FAK) phosphorylation. While the unphosphorylated FAK is capable of blocking the mineral deposition, conversely, phosphorylated FAK (p-FAK) increased in the presence of titanium treated with poly-L-lysine and promoted calcium deposition, osteogenic differentiation, and bone maturation. In support of this mechanism, the same authors reported the presence of p-FAK only in cells treated with titanium-poly-L-lysine and from a twofold (at day 6) to eightfold (at day 25) increase of osteogenic differentiation markers in hMSCs and hDPSCs grown on titanium and poly-Llysine compared to untreated hMSCs and hDPSCs [20]. In conclusion, poly-L-lysine seems to increase the p-FAK form, thus limiting its capability to block the mineral deposition and hence promoting the osteoblastic differentiation pathway and initiating mitogen-activated protein kinases, leading to osteogenic differentiation and bone maturation.

Four years later, in 2015, Varoni et al. confirmed the effect of poly-L-lysine coatings on titanium osseointegration by in vivo studies on sheep animal models [21]. Their results showed that cortical bone microhardness significantly improved in the presence of the poly-L-lysine coating by enhancing calcium deposition and implant early osseointegration in animals.

Little literature exists about the proliferative effects of poly-L-lysine on HaCaT cells; the work closest to highlighting this effect was the study by Renò et al. [45], but a complete and exhaustive explanation of the underlying mechanisms has not been reported yet. Renò et al. tested the efficacy of two different hydrogels synthesized by crosslinking gelatin with polylysine (positively charged) (HG1) and gelatin with polyglutamic acid (negatively charged) (HG2) as scaffolds for immortalized human keratinocytes (HaCaT) growth. They found that keratinocytes adhered both onto the HG1 and HG2 surface and were capable of proliferating, without toxicity, even if the cells displayed higher adhesion and proliferation

onto HG2, forming a continuous and stratified epithelium after 7 days [45]. Further studies are necessary to elucidate the poly-L-lysine effects on epithelial cells and wound-healing processes in depth.

To prevent bacterial infections and facilitate the bone mineralization around the dental implants, recently, Guo et al. reported the synergistic effect of a composite coating made up of poly-L-lysine/sodium alginate and nanosilver [46], while Zhang et al. coated the titanium surfaces with a multilayer biofilm of ε-polylysine and arabic gum [47].

The present work tested the effects of poly-L-lysine-coated implant plates on the cell growth and cytotoxicity both on epithelial cells and dental-derived stem cells, in order (i) to confirm any proliferative effects on mesenchymal cells responsible for osteogenesis and (ii) to establish whether it exerts a potential similar muco-proliferative effect on cells of epithelial origin. For these purposes, a series of experiments were conducted on two different cell lines: epithelial (HaCaT) and mesenchymal (hDPSCs) cells.

Results unanimously have reported cell viability, lack of cytotoxicity, and a statistically significant improvement of the cell growth both for hDPSCs and HaCaT when cultured on poly-L-lysine-coated titanium plates, when compared with the cultures of the cells alone and those of the cells with uncoated titanium plates.

#### **5. Conclusions**

The oral cavity is always challenged by mechanical, chemical, and biological stimulations throughout life [48], and because the oral mucosa represents a protective barrier between the soft tissues and the external environment [49], it is essential to preserve its integrity and resistance to mechanical stress, both physiological and pathological, and to reduce irritating local factors such as bacterial plaque [30,50]. Oral dysbiosis and poor oral hygiene compromise the health of the peri-implant soft tissues. Furthermore, as in gingivitis and periodontitis, which are diseases responsible for gingival inflammation and bone loss strictly associated with bacterial plaque composition and bone diseases such as osteoporosis [51,52], peri-implant sites can be equally affected by their counterparts as well. These counterparts are called mucositis and peri-implantitis [31,32] which, respectively, lead to inflammation of the mucosa surrounding the abutment and the loss of bone around the fixture, thus compromising the stability of the implant in the bone, which is resorbed and decreased [53–56]. Furthermore, dysmetabolic diseases such as chronic hyperglycemia have been associated with periodontitis and peri-implantitis due to delayed and/or impaired wound healing for the activation of pathways linked to inflammation, oxidative stress, and cell apoptosis [57–59].

The present work is an exploratory study to confirm the bone proliferative effects of a poly-L-lysine coating on titanium and to establish analogue proliferative effects on keratinocytes and lack of cytotoxicity.

The results have confirmed the positive effects of poly-L-lysine on osteoinduction [20,21,28,42,53] and demonstrated a novel potential role also in promoting epithelial cell growth. It means that in clinical practice, a poly-L-lysine topic administration on the surgical mucosal site of a dental implant, could promote, accelerate, and ameliorate the formation of epithelial tissue around semi-submerged and on submerged implants, to favor more rapid healing of the surgical site after the fixture placement and to reinforce the epithelium surrounding the abutment during the remaining life of the implant, thus preventing mucositis and peri-implantitis arising from a loose gingival–implant thigh contact.

However, further in vivo studies are required to confirm the effects of titanium functionalized with a poly-L-lysine coating to improve implant osteointegration and to elucidate the mechanisms of action on keratinocytes and the in vivo efficacy of polylysine compounds in promoting epithelial cell growth and wound healing, as well as after the implant placement and during years to preserve the health of peri-implant mucosa, with particular attention to the aesthetic area [60].

Furthermore, the additional in vivo studies could be supported by non-invasive imaging techniques [61–65] as well as classical procedures, which could highlight and quantify the real histological and cytologic effects of poly-L-lysine on epithelial cell growth to enhance and/or support the wound healing not only at peri-implant sites but also for the treatment of oral lesions and injuries requiring the re-establishment of a healthy mucosal barrier [66–69] and the reduction of biofilm formation around the teeth and implants [70].

**Author Contributions:** Conceptualization, M.D.D., D.M. and A.D.R.; methodology, M.D.D., M.L.N. and A.F.; formal analysis, A.B.-F. and E.X.; data curation, M.D.D., M.C. and K.F.; writing—original draft preparation, M.C., L.N. and A.B.; writing—review and editing, M.C., M.L.N., D.M., A.F. and M.D.D.; supervision, F.I. and M.D.D.; funding acquisition, M.D.D. and A.D.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partly funded by the SANIDENT S.R.L. Via Settembrini, Milan, Italy, in term partnership (from 29/01/2020 to 29/07/2020) with the Department of Precision Medicine, University of Campania "Luigi Vanvitelli", Naples, Italy.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Independent Ethical Committee University Hospital of Bari, Italy, protocol number 155/2021.

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** Data sharing is not applicable to this article.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


#### *Article*

## **In Situ and Ex Situ Designed Hydroxyapatite: Bacterial Cellulose Materials with Biomedical Applications**

#### **Adrian Ionut Nicoara 1,2 , Alexandra Elena Stoica 1,2, \*, Denisa-Ionela Ene 3 , Bogdan Stefan Vasile 1,2 , Alina Maria Holban <sup>4</sup> and Ionela Andreea Neacsu 1,2**


Received: 22 September 2020; Accepted: 23 October 2020; Published: 27 October 2020

**Abstract:** Hydroxyapatite (HAp) and bacterial cellulose (BC) composite materials represent a promising approach for tissue engineering due to their excellent biocompatibility and bioactivity. This paper presents the synthesis and characterization of two types of materials based on HAp and BC, with antibacterial properties provided by silver nanoparticles (AgNPs). The composite materials were obtained following two routes: (1) HAp was obtained in situ directly in the BC matrix containing different amounts of AgNPs by the coprecipitation method, and (2) HAp was first obtained separately using the coprecipitation method, then combined with BC containing different amounts of AgNPs by ultrasound exposure. The obtained materials were characterized by means of XRD, SEM, and FT-IR, while their antimicrobial effect was evaluated against Gram-negative bacteria (*Escherichia coli*), Gram-positive bacteria (*Staphylococcus aureus*), and yeast (*Candida albicans*). The results demonstrated that the obtained composite materials were characterized by a homogenous porous structure and high water absorption capacity (more than 1000% *w*/*w*). These materials also possessed low degradation rates (<5% in simulated body fluid (SBF) at 37 ◦C) and considerable antimicrobial effect due to silver nanoparticles (10–70 nm) embedded in the polymer matrix. These properties could be finetuned by adjusting the content of AgNPs and the synthesis route. The samples prepared using the in situ route had a wider porosity range and better homogeneity.

**Keywords:** bacterial cellulose; hydroxyapatite; nanoAg; tissue engineering; antimicrobial composite

#### **1. Introduction**

One of the most significant advances in the field of tissue engineering is the development of a porous three-dimensional matrix [1]. In order to act as an optimal bone support, the synthetic matrix must have a series of properties, including biocompatibility, biodegradability, appropriate porosity (similar to the replaced tissue), antimicrobial activity, and production reproducibility [2,3]. In addition to these requirements, it is also recommended that they have mechanical properties similar to natural bone, such as compressive strength, fatigue resistance, and high Young's modulus [4].

Such characteristics allow cell penetration, vascularization, and adequate nutrient and oxygen diffusion to cells and to the unformed extracellular matrix, which ensures cells viability. The pore size is, in fact, a key element for material efficiency. The pores must be large enough to allow cells to enter and move into the framework of the scaffold, while a small dimension allows the attachment of essential cell number at the same level [5]. Depending on the type of host tissue, all the support materials used in tissue engineering may have a macroporous structure with a particular pore size. For example, researchers suggest a pore size of 200–400 microns is optimal for bone tissue engineering [5,6].

The inorganic phase of the composites designed for bone replacement is usually hydroxyapatite (HAp) [7,8]. HAp is an essential element required for tissue regeneration, with the advantages of great biocompatibility, high plasticity, and remarkable mechanical properties because its chemical and crystalline structure is similar to natural bone apatite [2,9]. It also has an ultrafine structure and a large surface area that is advantageous for cell–biomaterial interactions and has been widely studied in applications for bone engineering [8,10].

For the organic phase of natural bone, replacement with bacterial cellulose (BC) has been attempted [11]. Even though the BC structure is chemically equivalent to plant cellulose (β-D-glucopyranose units linked by β-1,4 glycosidic bonds), it is free of by-products, such as lignin, pectin, hemicellulose, and other constituents of lignocellulosic materials. BC is a biodegradable polymer consisting of nanofibrillar structures, which determine a high specific surface area and a microporous structure. The unique 3D structure of BC is the main reason for its excellent retention and osteoinductivity, properties that make it a highly desirable substitute for collagen extracellular matrix in hard tissue engineering applications [12]. However, insufficient mechanical strength of the polymer restricts its direct application in vivo [13].

Many studies [14–17] have shown that BC could provide tissue regeneration and substitution, thus being used for bioengineering of hard, cartilaginous, and soft tissues. Bacterial cellulose is widely used as a wound dressing material, and nanomaterials obtained from BC show great antimicrobial properties [18–21]. BC can be combined with polymeric and nonpolymeric compounds to acquire or enhance antimicrobial, cell adhesion, and proliferation properties [13,22–25].

Scaffolds embedded with antimicrobial agents, antibiotics, or several forms of silver nanoparticles, which are known antimicrobial agents, are attracting great interest in biomedical research. Metallic silver and silver nanoparticles (AgNPs) have been reported to provide a wide variety of antimicrobial activities [12,26–28].

AgNPs are more toxic compared to bulk silver but they have a strong anti-inflammatory impact during tissue healing and can be integrated into composite materials to obtain antibacterial properties [29–31]. The human dietary intake of silver, owing to the widespread use of silver compounds, is estimated at 70–90 µg per day [32]. One of the main risk factors in tissue engineering and implant development is microbial infections. Bacterial colonization and the development of multicellular attached communities, called biofilms, are responsible for the high rate of failure in tissue engineering [33].

The purpose of this study was to develop a composite material based on hydroxyapatite, bacterial cellulose, and silver nanoparticles with biomedical applications. The material was obtained by the coprecipitation technique, which is a reliable, simple, economic, fairly rapid, and precise method that allows the synthesis of homogenous structures and favorable pore dimensions [10]. Studies have described the synthesis of bacterial cellulose/hydroxyapatite composites for bone healing applications using different methods [34–39]. In this work, AgNPs were integrated in the composite system in order to induce antibacterial properties.

#### **2. Materials and Synthesis Methods**

The chemical reagents were calcium nitrate tetrahydrate (Ca(NO3)2•4H2O, >99%), ammonium phosphate dibasic ((NH4)2HPO4, 99%), ammonium hydroxide (NH4OH, 99%), sodium hydroxide (NaOH, 98%), silver nitrate (AgNO3, >99%), sodium citrate (C6H5O7Na, >99%), polyvinylpyrrolidone ((C6H9NO)n), and sodium borohydride (NaBH4, >99%), purchased from Sigma-Aldrich (St. Louis, MO, USA). The solvents were American Chemical Society (ACS, Washington, DC, USA) purity. Bacterial cellulose membrane was produced in the laboratory by *Gluconacetobacter* sp. strain isolated from traditionally fermented apple vinegar in the Microbiology Laboratory of the Chemical and Biochemical Engineering Department, University Politehnica of Bucharest, based on a protocol previously described [40].

In order to obtain 500 mL colloidal silver (100 ppm concentration), an aqueous silver nitrate solution (AgNO3) was used as silver precursor, to which 30 mL sodium citrate (0.3 M) was added. After 12 min, 30 mL of Polyvinylpyrrolidone (PVP, 0.007 M) and 5 mL NaBH<sup>4</sup> (1 M) were added to reduce Ag<sup>+</sup> to Ag<sup>0</sup> nanoparticles. Finally, 5 mL of oxygenated water (30%) was added, and stirring was maintained for another 10 min approximately, until a light blue color (due to the size of the nanoparticles) was obtained (Figure 1) [41].

**Figure 1.** Synthesis of silver nanoparticles.

The bacterial cellulose synthesized by the Gram-negative bacteria (*Gluconacetobacter* sp.) was boiled at 80 ◦C in water alkalized with sodium hydroxide (pH 14, measured by colorimetric method). After purification, BC was washed in distilled water until it reached neutral pH. Afterward, it was minced using a blender (Silvercrest, Neckarsulm, Germany) and weighed according to the recipe. Previously, the amount of dry matter was determined on a quantity of BC by eliminating the humidity, and it was found that 0.25 g of dry BC can be obtained from 10.62 g of wet BC [17].

Two synthesis methods were used to obtain the bacterial cellulose and HAp-based composites. The first method involved obtaining in situ hydroxyapatite nanoparticles directly on cellulose fibers and subsequently adding the AgNPs solution, followed by homogenization using an ultrasound probe (composites further referenced as BC1, BC2, BC3, and BC4).

For the synthesis of 2 g HAp, the amount of precursors required to obtain the material with different concentrations of AgNPs (0, 1, 2, and 5 wt %) was calculated.

− The Ca2<sup>+</sup> and PO<sup>4</sup> <sup>3</sup><sup>−</sup> precursors, Ca(NO3)2•4H2O, and (NH4)2HPO<sup>4</sup> were solubilized in distilled water, and bacterial cellulose was added in the calcium nitrate solution (see Figure 2). The mixtures were homogenized by magnetic stirring, and the ammonium phosphate solution was added dropwise. After homogenization, the pH was adjusted to 10.5 with an ammonium hydroxide solution. The obtained precipitates were aged for 24 h, then washed with distilled water until pH 7 was achieved. After washing, the appropriate amount of silver colloidal solution that had been previously obtained was added to each composition according to the centralizing table (Table 1). The obtained mixture was mixed for 3 min in the presence of ultrasound to ensure the best possible homogeneity and then poured into Petri dishes (d = 54 mm), frozen, and subsequently subjected to the freeze-drying process (freezing at −55 ◦C for 12 h, vacuum at 0.001 mbar for 12 h, and heating under vacuum for 24 h to 35 ◦C) in order to obtain porous composite materials [42]. −

**Figure 2.** Synthesis of BC<sup>1</sup> , BC<sup>2</sup> , BC<sup>3</sup> , and BC<sup>4</sup> composites; BC, bacterial cellulose.


**Table 1.** Bacterial cellulose, HAp, and AgNP content in the final composites.

The second method of synthesis involved the separate synthesis of HAp by the coprecipitation method, followed by its addition to bacterial cellulose gel in the presence of ultrasound for 3 min, as described in Figure 3.

**Figure 3.** Synthesis of BC<sup>5</sup> , BC<sup>6</sup> , BC<sup>7</sup> , and BC<sup>8</sup> composites.

After homogenization, the required amount of silver colloidal solution was added, followed by the steps previously described in the in situ method. The composites thus obtained by the ex situ method were noted as BC5, BC6, BC7, and BC8, and their composition is presented in Table 1.

θ

λ α

ெೢ ି ெ ெೢ ି ெೢ

> ெೢ ି ெ ெ

− −

#### **3. Characterization Methods**

#### *3.1. Physicochemical Characterization*

Investigation of the crystallinity of the powders was performed by means of X-ray diffraction (XRD) technique using the PANalytrical Empyrean (Malvern, Bruno, the Netherlands) equipment in Bragg–Brentano geometry equipped with a Cu anode (λCuKα = 1.541874 Å) X-ray tube. The spectra were acquired in the range of 10–80◦ 2θ angles (Bragg angle) with an acquisition step of 0.02◦ and an acquisition time of 100 s. The scanning electron microscopy (SEM) images were performed with a FEI Inspect F50 microscope coupled with an energy-dispersive spectrometer (EDS) (FEI, Eindhoven, the Netherlands). Both secondary electron and backscattered electron detectors were used at 30 kV accelerating voltage. The TEM images of AgNPs were obtained using the high-resolution transmission electron microscope TecnaiTM G2 F30 S-TWIN equipped with selected-area electron diffraction (SAED) detector, purchased from the company FEI. This microscope operates in transmission mode at a voltage of 300 kV with a resolution of 2 Å. Research conducted by Fourier transform infrared spectroscopy (FT-IR) involved the analysis of a small amount of samples using the Nicolet iS50R spectrometer (Thermo Fisher Waltham, MA, USA). The measurements were performed at room temperature utilizing the total reflection attenuation module (ATR), and 32 scans of the samples between 4000 and 400 cm−<sup>1</sup> were performed using a resolution of 4 cm−<sup>1</sup> . The differential thermal analysis (ATD-DSC) were performed using a Shimadzu DTG-TA-50H equipment (Shimadsu, Sanjo, Japan) at 25–700 ◦C with a heating rate of 10 ◦C/min.

The open porosity of the freeze-dried composite materials was calculated with Equation (1) for each material prepared in order to observe the porosity level according to the chosen manufacturing method, while the water absorption was calculated with Equation (2):

$$\text{Open porosity} \left( \text{\textquotedblleft} \right) = \frac{M\_{\text{uv}} - M\_d}{M\_{\text{uv}} - M\_{\text{uv}}} \times 100 \tag{1}$$

$$\text{Water absorption } (\%) = \frac{M\_{\text{ue}} - M\_d}{M\_d} \times 100 \tag{2}$$

where *Mwe* is the wet sample weight, *M<sup>d</sup>* is the dry sample weight, and *M<sup>w</sup>* is the sample weight in water.

#### *3.2. Degradability*

To test their biodegradability, the samples were placed in a 12-well plate in which phosphate-buffered saline (PBS) and simulated body fluid (SBF) were added, similar to the processes involved in the human body. After immersion of the samples in fluid, their integrity was monitored for 7 days. The degradation rate was calculated with Equation (3) for each material:

$$\text{Depradation } (\%) = \frac{M\_{7day} - M\_{initial}}{M\_{initial}} \times 100\tag{3}$$

where *M*7*days* is the sample weight in SBF after 7 days of immersion in SBF, and *Minitial* is the sample weight after immersion in SBF. All the weight values were obtained at room temperature using a hydrostatic analytic balance.

#### *3.3. Antimicrobial E*ffi*ciency*

The antimicrobial behavior of the freeze-dried composite materials was qualitatively assessed by an adapted growth inhibition assay [43]. To cover a wide spectrum of clinically relevant model microbial species, one Gram-positive (*Staphylococcus aureus* ATCC 23235), one Gram-negative bacteria (*Escherichia coli* ATCC 25922), and one yeast (*Candida albicans* ATCC 10231) laboratory strain were used. The standard work protocol for the adapted version of the disc diffusion method implies the preparation of microbial suspensions of 0.5 McFarland standard density (1.5 <sup>×</sup> <sup>10</sup><sup>8</sup> colony forming units (CFU)/mL), prepared in sterile buffered saline solution. The obtained microbial suspensions were afterward used to swab inoculate the entire surface of the nutrient agar Petri dishes. After inoculation, identical size samples of the sterile coatings were aseptically placed on the inoculated agar surface, and the plates were incubated at 37 ◦C for 24 h to allow the growth of bacteria. After incubation, the growth inhibition zone diameter (mm) was measured. A wider inhibition zone suggests a higher antimicrobial effect of the fibrous dressing, reflecting the ability of AgNPs contained into the composite material to diffuse within the agar.

#### **4. Results and Discussions**

The thermal analysis corresponding to the composite samples are presented in Figure 4. The two minor weight losses that occurred at temperatures below 200 ◦C were probably related to the volatilization of solvents and physical water. The main mass loss was observed in the range 250–450 ◦C, with the corresponding exothermic effect being strong and intense and indicating burning of the organic component of the composite (bacterial cellulose). Regarding the compositional aspects, the thermal analysis allowed an accurate assessment of the loading degree depending on the material deposited on the surface or between the fibers of the bacterial cellulose. It was observed that certain changes associated with endothermic processes occurred in the thermogravimetric (TG) curve with the addition of silver nanoparticles. Hence, in the 450–600 ◦C interval, exothermic effects generated by the combustion of BC were observed (see Figure 4a), while in the 600–700 ◦C interval, it can be assumed that the oxidation of silver nanoparticles and dehydroxylation of HAp occurred [15].

**Figure 4.** Thermal analysis corresponding to the composite samples: (**a**) BC<sup>1</sup> and BC<sup>4</sup> ; (**b**) BC<sup>5</sup> and BC8.

We observed (Figure 4) significant differences regarding mass loss between the samples obtained by in situ vs. ex situ method. The total weight loss in the temperature range of 30–700 ◦C was 45% for BC1, 65% for BC4, 68% for BC5, and 70% for BC8. The composites obtained in situ had a lower weight loss, which suggests good loading of BC with calcium phosphate phases (HAp).

In order to demonstrate the composition, hydroxyapatite was analyzed by XRD technique. The diffractograms are presented in Figure 5.

**Figure 5.** XRD analysis corresponding to the composites samples BC<sup>3</sup> and BC4, and hydroxyapatite (HAp) obtained by the coprecipitation method.

Due to the fact that BC5–BC<sup>8</sup> samples were made by direct mixing of bacterial cellulose with hydroxyapatite (ex situ), the composition of this sample was not expected to change; therefore, the XRD analysis was only performed for bacterial cellulose, simple hydroxyapatite, and BC<sup>3</sup> and BC<sup>4</sup> composites (in situ) [44,45].

It was observed that, in all the analyzed samples, the existence of bacterial cellulose and HAp was obvious. In addition, the low-intensity peak around 2θ = 38◦ , which can be assigned to the (111) crystalline plane, indicated the presence of silver nanoparticles in the composite structure. Investigation of the composites BC<sup>3</sup> and BC<sup>4</sup> revealed peaks located at 2θ values of 15, 16, and 23◦ , which can be attributed to bacterial cellulose according to ICDD 00-056-1718.

As the HAp peaks were poorly visible in XRD analysis, FT-IR analyses were used to better highlight hydroxyapatite formation. The results are presented in Figure 6.

The vibrational frequencies characteristic of bacterial cellulose were observed at 3500–3200 cm−<sup>1</sup> (OH stretch vibrations) and 2958 cm−<sup>1</sup> (CH<sup>2</sup> and CH<sup>3</sup> stretch vibrations). The wide band observed in the region of 3500–3200 cm−<sup>1</sup> , attributed to the hydroxyl groups within the bacterial cellulose, increased in absorbance with higher silver content. This behavior suggested that the presence of HAp crystals affected cellulose hydroxyl groups, probably by covering them at the surface. Furthermore, the change observed for the band attributed to intramolecular hydrogen bonding (~3500 cm−<sup>1</sup> ) confirmed a strong interaction between the OH groups and calcium phosphate. The chemical interaction between HAp and BC stabilized the composite so that it could maintain its mechanical integrity, an aspect required for bone substituents [46].

The FT-IR bands observed at 1020 cm−<sup>1</sup> and 570–600 cm−<sup>1</sup> were attributed to the vibrational modes of PO<sup>4</sup> <sup>3</sup>−. Because the stretching vibration of CO<sup>3</sup> <sup>2</sup><sup>−</sup> also appeared (at 1418 cm−<sup>1</sup> ), absorption of CO<sup>2</sup> from the air is suggested [47]. This is mainly a result of the affinity for carbonate of HAp as well as the lack of heat treatment during the in situ synthesis (which favors the release of CO2). Carbonated hydroxyapatite contributes to the biomimetism increase of the obtained composites, which can promote the process of osteoregeneration. It was observed that the in situ method accelerated the nucleation of HAp crystals onto BC fibers instead of crystallization as higher absorbance values

were registered for BC5–BC<sup>8</sup> samples, which contained highly crystalized HAp. The bands observed at 1641 and 643 cm−<sup>1</sup> correspond to the stretching and deformation vibrations of AgO, respectively, thus confirming the presence of silver in the obtained materials [48]. This result supports the idea that the composite material developed here possesses essential physicochemical properties and could be very useful for biomedical applications, especially hard tissue engineering. θ θ

**Figure 6.** FT-IR analysis corresponding to the (**a**) in situ and (**b**) ex-situ composite samples.

− − Through the two analyses performed, it was possible to notice the elemental composition of materials (EDS) as well as the homogeneity of hydroxyapatite particle dispersion.

− − − − The SEM image highlighted the fibrous structure of BC (see green arrow), which were decorated with inorganic particles (see blue arrow). In the SEM images performed on the composites in which HAp was obtained in situ (Figure 7a–d), a better homogeneity was observed compared to the cases in which HAp was obtained separately and subsequently mixed with BC (ex situ) (Figure 8a–d. The interaction between HAp nanoparticles distributed in the 3D network of BC stabilized the composite so that it could maintain its mechanical integrity, an aspect required for bone substituents. In addition, EDS analysis confirmed the presence of the elements specific for hydroxyapatite (Ca, P, and O) as well as the presence of silver for the samples in which it was added (Figure 7(b2–d2) and Figure 8(b2–d2)).

− − − Transmission electron microscopy images showed the silver nanostructure (Figure 9a,b), with the dimensions of the silver particles being in the range 3–60 nm. It could be observed that the quasi-spherical morphology of nanosilver and some areas were darker while others were brighter; the darker areas indicate a higher degree of crystallinity of the material.

Figure 9d shows a SAED image with information on the crystallinity of the analyzed material. The presence of diffraction rings with higher light intensity shows a high degree of crystallinity.

− The calculated open porosity for each prepared material is presented in Figure 10a, and the calculated water absorption is presented in Figure 9b.

Figure 10a shows that the composites obtained by in situ approach had a large porosity compared to samples obtained by ex situ. This suggests that the in situ route will provide a biodegradable polymer with excellent water retention and, possibly, good osteoinductivity, which can be used as an artificial substitute for hard tissue.

**Figure 7.** High-resolution backscattered-electron (BSE) images (and EDS spectra at 100× magnification for: BC<sup>1</sup> (**a**,**a1**,**a2**), BC<sup>2</sup> (**b**,**b1**,**b2**), BC<sup>3</sup> (**c**,**c1**,**c2**), BC<sup>4</sup> (**d**,**d1**,**d2**). (where **a1**–**d<sup>1</sup>** images represent a high magnification of a-d images area; green arrow indicates the fibrous structure of BC and blue arrow indicate inorganic particles).

**Figure 8.** High-resolution backscattered electron (BSE) images and EDS spectra at 100× magnification for BC<sup>5</sup> (**a**,**a1**,**a2**), BC<sup>6</sup> (**b**,**b1**,**b2**), BC<sup>7</sup> (**c**,**c1**,**c2**), and BC<sup>8</sup> (**d**,**d1**,**d2**) (where **a1**–**d<sup>1</sup>** images represent a high magnification of a–d image area; green arrow indicates the fibrous structure of BC and blue arrow indicate inorganic particles).

**Figure 9.** TEM (**a**,**b**), HR-TEM (**c**), and SAED (**d**) images for silver.

**Figure 10.** Porosity (**a**) and water absorption (**b**) results for the obtained composites (presented as mean ± S.D. of three replicates and \* *p* < 0.005 obtained by single-factor ANOVA test.

As can be observed, when the silver nanoparticle concentration increased, the composite porosity decreased for susceptible types of composite, which resulted in lower absorption capacity. Even though

high porosity, which is associated with increased absorption capacity, is an important structural parameter for bone substituents, the registered decrease due to Ag addition is not significant in this case as the water absorption was still greater than 1000–1500%.

Visual inspection (Figure 11) is an efficient technique for investigation of simulated in vitro degradation of BC samples, and it has also been used in recent literature [49] in order to investigate the degradation of cellulose-based materials. Generally, visual inspection implies macroscopic pictures of the immersed samples while observing, in a qualitative manner, the presence of detached fragments, the apparition of denser fragments that may provide additional mechanical integrity for cell growth, and so on. As expected, the composite materials did not show major changes after immersion in PBS and SBF.

**Figure 11.** Macroscopic pictures of degradation of BC composites initially (**a1**,**a2**) and after seven days of immersion in phosphate-buffered saline (PBS) and simulated body fluid (SBF) (**b1**,**b2**).

After seven days, no major visual changes were observed, and the degradation was below 5% (see Figure 12), a sign that bacterial cellulose prevented the disintegration of the composite. A rapid biodegradation of the implanted material is not desired because it takes time for it to integrate better into the host tissue. Another problem is that by biodegradation, remnants/fragments of the material can reach the level of sensitive areas, which would be fatal.

Due to the higher homogeneity and better interactions between the phosphate phase (HAp) and BC, the composites obtained in situ had a lower degradation compared to those obtained using ex situ methods.

**Figure 12.** Degradability of composites after seven days of immersion in SBF at room temperature.

#### *Antimicrobial Potential*

The antimicrobial effect of the obtained composite materials was different among the tested samples, being influenced by the synthesis approach, AgNP content, and microbial species. Figure 13 shows the diameters of the inhibition area of the microorganisms grown in the presence of the tested materials.

**Figure 13.** The diameter of the growth inhibition zone of the tested microorganisms grown in the presence of the obtained BC samples containing various amounts of AgNPs. The plain AgNP control is represented by 10 uL of AgNPs (used at maximum equivalent amount contained in BC samples), which was added to a commercial filter paper disc of a similar size as the obtained BC samples.

Because composites BC<sup>1</sup> and BC<sup>5</sup> did not contain antibacterial AgNPs, they were used as control samples for evaluation of the other samples.

The antimicrobial characteristic of the obtained composites was clearly influenced by the concentration of AgNPs, with samples with higher content of silver exhibiting the greatest microbial growth inhibition, regardless of the synthesis approach. It was observed that the materials obtained by the in situ method had a more pronounced overall antibacterial characteristic (growth inhibition zones ranging 5–11 mm) compared to the samples obtained by the ex situ method (diameter of inhibition zones ranging 5–9 mm).

The different antimicrobial effects of the materials obtained by in situ and ex situ routes correlated with their physicochemical properties. Samples obtained by the in situ route showed a larger porosity, suggesting that the bioactive compounds (i.e., AgNPs) may be absorbed more efficiently in pores. Moreover, a higher porosity degree can be directly associated with an easier release of the bioactive agent, therefore inducing an increase in the antimicrobial activity of the final composite. This idea is supported by the results obtained with the control AgNPs utilized in equivalent amounts and added to sterile commercial filter paper, which demonstrated lower inhibition zones.

The most efficient growth inhibition was observed against the yeast strain *C. albicans* ATCC 10231, with the result being relevant for samples obtained by both in situ and ex situ methods. However, the composites obtained in situ also showed increased antimicrobial activity against the Gram-positive *S. aureus* strain. This result suggests that the obtained composite materials may act differently on microbial cells, depending on the particularities of their cellular wall. Such differences were observed before with silver nanoparticles [50–53].

#### **5. Conclusions**

In this study, we report the synthesis of hydroxyapatite–bacterial cellulose silver nanocomposites obtained by two routes using coprecipitation, namely, in situ and ex situ assembly. These materials contained an organic part (bacterial cellulose), an inorganic part (hydroxyapatite), and an antimicrobial agent (AgNPs) contained in various amounts, thereby conferring new bioactive properties on the composite materials. Physicochemical and antimicrobial studies demonstrated that the most efficient in terms of potential biomedical applications were the samples obtained by the in situ approach. The porosity range of the in situ materials was greater than the porosity of ex situ composites, while the best antimicrobial activity was observed for the material coded BC4, which had a content of 5 wt % AgNPs. Due to the physicochemical structure, together with the already demonstrated great antimicrobial properties and low biodegradability of these materials, they have potential applications as successful candidates for biomedical applications, especially in hard tissue engineering. Their current limitation relates to the fact that further tests performed on osteoblast differentiation and mineralization (e.g., alkaline phosphatase and alizarin red S) are needed.

**Author Contributions:** Conceptualization, A.I.N. and A.E.S.; methodology, B.S.V.; validation, A.I.N. and I.A.N.; formal analysis, A.I.N., D.-I.E., and B.S.V.; investigation, D.-I.E., I.A.N., and A.M.H.; writing—original draft preparation, A.E.S. and A.I.N.; writing—review and editing, A.E.S., I.A.N., B.S.V., and A.M.H.; supervision, B.S.V. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was financially supported by the project Smart Scaffolds Built on Biocellulose 3D Architecture or Artificial Electrospun Templates for Hard Tissue Engineering (ScaBiES), PN-III-P1-1.1-TE-2016-0871 (contract number 66/2018), financed by the Executive Unit for Financing Higher Education, Research, Development, and Innovation (UEFISCDI). The SEM and RAMAN analyses obtained on the samples were possible due to the EU-funded project POSCCE-A2-O2.2.1-2013-1/Priority Axe 2, project number 638/12.03.2014, ID 1970, SMIS-CSNR code 48652. The XRD analyses were financed by the European Social Fund and the Romanian Government under the contract number POSDRU/86/1.2/S/58146 (MASTERMAT).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Bioengineering Bone Tissue with 3D Printed Sca**ff**olds in the Presence of Oligostilbenes**

**Francesca Posa 1,2, \* ,**† **, Adriana Di Benedetto 1,**† **, Giampietro Ravagnan 3 , Elisabetta Ada Cavalcanti-Adam <sup>2</sup> , Lorenzo Lo Muzio 1 , Gianluca Percoco <sup>4</sup> and Giorgio Mori 1**


Received: 13 August 2020; Accepted: 3 October 2020; Published: 9 October 2020

**Abstract:** Diseases determining bone tissue loss have a high impact on people of any age. Bone healing can be improved using a therapeutic approach based on tissue engineering. Scientific research is demonstrating that among bone regeneration techniques, interesting results, in filling of bone lesions and dehiscence have been obtained using adult mesenchymal stem cells (MSCs) integrated with biocompatible scaffolds. The geometry of the scaffold has critical effects on cell adhesion, proliferation and differentiation. Many cytokines and compounds have been demonstrated to be effective in promoting MSCs osteogenic differentiation. Oligostilbenes, such as Resveratrol (Res) and Polydatin (Pol), can increase MSCs osteoblastic features. 3D printing is an excellent technique to create scaffolds customized for the lesion and thus optimized for the patient. In this work we analyze osteoblastic features of adult MSCs integrated with 3D-printed polycarbonate scaffolds differentiated in the presence of oligostilbenes.

**Keywords:** bone; mesenchymal stem cells; biomaterial; polycarbonate; resveratrol; polydatin; osteogenic differentiation; focal adhesions; bone health

#### **1. Introduction**

The global increase in average life expectancy is leading to an escalation of age-related health problems which may affect organs or tissues. Although the bone tissue is capable of self-regeneration, there are several pathological conditions, determined by serious trauma or degenerative diseases, which require appropriate medical procedures in order to realize a complete recovery of the anatomical and functional properties of the tissue. These innovative therapeutic approaches are part of the so-called regenerative medicine [1,2]. The use of adult stem cells in bone reconstructive therapies offers significant benefits [3,4].

In fact, it is known that stem cells are capable of self-renewal and can differentiate into different cell types, ensuring the repair of most tissues, thus becoming highly useful for tissue engineering. Stem cells in human adults can be isolated from various tissues, including bone marrow, nervous tissue, peripheral blood, retina, liver, pancreas, tooth, and dental bud. In particular, among adult stem cells,

mesenchymal stem cells (MSCs), originally identified in the bone marrow, which is still considered the best cell source for osteogenic differentiation, can be isolated also from several adult tissues such as adipose tissue, dental tissues, skin, brain, liver, and fetal tissues [5,6]. MSCs appropriately isolated and induced to differentiation, through integration with biocompatible scaffolds, could represent a valid therapeutic approach for the regeneration of connective tissues as bone and cartilage, healing defects of traumatic, degenerative or neoplastic origin. Unfortunately, several concerns may arise from autologous and allogeneic stem cell transplants which, in their turn, may also not be sufficient to accelerate the healing process in the case of large bone defects [7]. As a consequence, in the last decades an increasing pivotal role has been attributed to tissue-engineered bone grafts developed through a combined effort of engineering, biotechnology and biomaterial science [8]. In this perspective, especially for hard tissue regeneration, one of the most promising approaches is the use of customized scaffolds combined with factors allowing cell proliferation and osteogenic differentiation [9]. Matching tissue engineering with predictive medicine, based on mechanobiological computational models, would optimize healing processes [10]. The fabrication methods for tissue engineering are conventional methods, additive manufacturing techniques and 4D printing [11]. Among these methods, 3D printing is an encouraging technique, easily available to realize personalized scaffolds to be used for tissue regeneration [12].

Dental bud stem cells (DBSCs) are widely recognized as MSCs that can effectively differentiate into osteoblast-like cells [13] becoming a suitable candidate for bone regeneration. DBSCs express the typical mesenchymal stem markers and, as we have shown, their ability to acquire the osteoblastic features and to produce mineralized bone matrix in vitro, which is the crucial event of osteogenic differentiation [14,15], can be positively influenced by several molecules [16–19]. Polydatin (Pol) deserves a particular mention among the natural molecules capable of inducing DBSCs osteogenic differentiation [19] and opens new horizons to its possible use as a therapeutic agent, as we have exhaustively detailed in our new invention patent (patent n.16999PTIT entitled "Composizioni comprendenti o costituite da Polidatina per uso nel trattamento delle patologie ossee"—"Compositions comprising or consisting of Polydatin to be used in the treatment of bone pathologies", deposited with application number 102017000079581). Pol, that is a natural precursor of the polyphenolic compound Resveratrol (Res), is a glucoside we find in abundance in fruits and plants [20]. This glucoside shares some of the beneficial biological properties fully demonstrated for Res [21–23], but, in comparison, it also presents advantages: higher stability, significant abundance and better oral absorption [24–26]. We have previously shown that DBSCs positively responded to Res and Pol treatment increasing their osteogenic potential and, moreover, Pol appeared to be more effective than Res even when used at a very low concentration [19]. To induce bone regeneration, porous scaffolds with appropriate shape, pore size, porosity, degradability, biocompatibility, mechanical properties and desirable cellular responses are required. 3D printing has revealed to be very useful in this field, thanks to the capability to process complex shapes with a wide variety of biocompatible materials such as poly(l-lactic acid) (PLLA), poly(vinyl alcohol) (PVA), poly(lactic-co-glycolic acid) (PLGA) and polycaprolactone (PCL) starting from filaments and pellets [27]. In this study we investigate the combination of Res and Pol treatment and 3D-printed polycarbonate (PC) scaffolds to study the possible effects of this set-up on MSCs commitment into the osteoblastic lineage. The PC has been chosen to have artificial scaffolds with bio-compatible material that have strength and stiffness near to bone tissue [28].

#### **2. Materials and Methods**

#### *2.1. Ethics*

The study was conducted in compliance with the Declaration of Helsinki and the International Conference on Harmonization Principles of Good Clinical Practice. The research protocol was approved by the ethical committee, within the project BIADIDENT num. Rep 4159/2018, at the University of Foggia Ospedali Riuniti, and all participants gave informed consent allowing their anonymized information to be used for data analysis.

#### *2.2. Sca*ff*old Preparation*

The scaffolds were manufactured using an Ultimaker S5, equipped with a AA nozzle diameter equal to 0.4 mm. At the best of author's knowledge, this printer can be considered as one of the best compromise between quality, price and system flexibility. The filament was the 3 mm 3DXMAX ® , a high-heat 3D printing filament made using Lexan ®. The print temperature was set to 285 ◦C and the bed temperature was kept to 110 ◦C, print speed 30 mm/s. The bed and nozzle temperature parameters are not inside the interval suggested by the 3D printer supplier, but since manufacturing time is lower than 5 min, the machine is able to complete the workpiece without damages. The different pore sizes were obtained setting on the Cura slicing software the distance between lines equal, respectively, to 0.75 mm for small pores, 0.9 mm for medium pores and 1.15 mm for large pores [29]. Figure 1 shows the Ultimaker 3D printer and samples of the manufactured scaffolds. equal to 0.4 mm. At the best of author's knowledge, this printer can be considered as one of the best

**Figure 1.** Printer and polycarbonate (PC) scaffolds. (**a**) Ultimaker S5 3D printer. (**b**) Printed scaffolds presenting pores of small (0.75 mm), medium (0.9 mm) and large (1.15 mm) dimensions. Images of representative scaffolds were chosen for the figure. Scale bar = 5 mm.

#### *2.3. Patients and DBSCs Cultures*

subjected to the third molar extraction for orthodontic reasons; each patient's parents provided a – The dental buds were collected from ten healthy pediatric donors (eight-twelve years) who were subjected to the third molar extraction for orthodontic reasons; each patient's parents provided a written informed consent. The study was authorized by the Institutional Review Board of the Department of Clinical and Experimental Medicine, University of Foggia. The dental papilla, which corresponds to the internal section of the dental bud, and contains stem cells of mesodermal origin, was cut in small fragments and enzymatically digested. Single-cell suspensions were harvested by filtration, and seeded and expanded in vitro as already reported [30–32]. In the experiments aimed to examine Res and Pol effect on cell adhesion during the osteoblastic differentiation process, DBSCs were seeded at a density of 3 × 10 3 /cm<sup>2</sup> and cultured in an osteogenic medium consisting of α-MEM supplemented

and cultured in an osteogenic medium consisting of α

−

with 2% heat inactivated fetal bovine serum (FBS), 10−<sup>8</sup> M dexamethasone and 50 µg/mL ascorbic acid (Sigma Aldrich, Milan, Italy). DBSCs were maintained in the osteogenic medium supplemented also with 10 mM β-glycerophosphate (Sigma Aldrich, Milan, Italy), for the evaluation of Res and Pol effects on cell adhesion, proliferation, differentiation and examination of their ability in the induction of matrix mineralization in cultures on biomaterials.

#### *2.4. Res and Pol Treatment*

Res and Pol extracted from *Polygonum cuspidatum* (Japanese Knotweed), according to the procedure defined in the Patent EP1292320B1, were provided by Prof. Ravagnan. Res and Pol were dissolved in ethanol at 100 mM stock solutions [33] and then added to the culture media under low serum conditions (2% FBS) to the final concentration of 0.1 µM for both of them, as detailed in Di Benedetto et al. [19]. In the experiments control cells were not treated with Res or Pol and served as control group (Ctr), treated cells were exposed to Res or Pol (treatment group), that were added to the media at every renewal (every 3 days).

#### *2.5. Immunofluorescence*

For focal adhesion staining, DBSCs were cultured on glass coverslips for 4 days and then fixed in 4% (w/v) paraformaldehyde (PFA) in PBS. Cells were then washed with PBS and blocked in 1% BSA, 5% normal goat serum in PBS for 20 min, to avoid non-specific protein binding. The following antibodies were used: αVβ<sup>3</sup> antibody 1:100 (clone LM609 antibody, cat. MAB1976, MerckMillipore, Merck KGaA, Darmstadt, Germany), followed by fluorescently labeled goat anti-mouse secondary antibody (Alexa Fluor 488, 2 µg/mL, Invitrogen ThermoFisher Scientific, Waltham, MA, USA). Samples were embedded in Mowiol containing 0.1% (*v*/*v*) DAPI for an additional staining of the nucleus. Cells were imaged by a multispectral confocal microscope Leica TCS SP5. The images were adjusted in brightness and color with ImageJ software (Research Services Branch, Image Analysis Software Version 1.52c, NIH, Bethesda, MD, USA).

#### *2.6. Alizarin Red Staining (ARS)*

DBSCs capacity to produce mineralized matrix nodules when cultured on the scaffolds was determined by performing ARS at 28 days of culture in osteogenic conditions. After removing the culture media, cells were rinsed with PBS, fixed in 10% formalin at RT for 10 min. Then cells were washed again with deionized water and stained using a 1% ARS solution for 10 min at RT. At the end of the incubation period the ARS solution was removed and cells were washed twice with deionized water and air dried. The quantification of ARS in the red stained monolayer was performed by extracting the dye and by reading the optical density (OD) in triplicate at 405 nm.

#### *2.7. Statistical Analyses*

Statistical analyses were performed by Student's *t*-test with the GraphPad Prism version 8.0.2 for MacOS software (San Diego, CA, USA). The results were considered statistically significant for *p* < 0.05 (indicated as § *p* < 0.01, \* *p* < 0.001).

#### **3. Results**

#### *3.1. Both Res and Pol Treatments Influence Cell Spreading and Focal Adhesion Assembly via* α*V*β*<sup>3</sup> Reorganization*

To investigate the influence of Res and Pol on cell adhesion and spreading, which determine, as a consequence, DBSCs exhibition of osteoblastic features, the cells were cultured for 4 days on glass coverslips in absence of treatment (Ctr) or in presence of Res or Pol added to the media (Figure 2a–c). Such a short period of time was chosen because of DBSCs predisposition to proliferate and produce various cell layers when left in culture for a few days, a condition that would not have allowed a

clear observation of focal adhesions. We examined αVβ<sup>3</sup> integrin subcellular distribution by confocal immunofluorescence. This integrin receptor has already been shown to be crucial for the osteogenic differentiation process of MSCs [30] and Vitamin D or the supramolecular aggregate T-LysYal ® (T-Lys) can enhance its expression and clusterization leading to the induction of the differentiation process [16,17]. As observable in Figure 1, in Ctr cells αVβ<sup>3</sup> integrin clusters were hardly detectable and only few structures were present at the periphery of the cells (Figure 2a). On the other hand, the presence of the molecules in the osteogenic media clearly induced a higher expression and also a reorganization of αVβ<sup>3</sup> integrin (Figure 2b,c). In particular, Pol treatment (Figure 2c) induced the strengthening of αVβ<sup>3</sup> adhesion sites by forming more elongated and larger peripheral clusters in comparison to cells treated with Res (Figure 2b). . As observable in Figure 1, in Ctr cells α β of α β 2c) induced the strengthening of α β

treatment induces clustering of α β staining of α β on of integrin α β **Figure 2.** Polydatin (Pol) treatment induces clustering of αVβ<sup>3</sup> integrin. Indirect immunofluorescence staining of αVβ<sup>3</sup> integrin (green), detected by the antibody LM609, and nuclei (blue) in Dental bud stem cells (DBSCs). Midsection confocal microscopy images show the localization of integrin αVβ<sup>3</sup> (green) in cells maintained for 4 days in osteogenic medium and treated with Resveratrol (Res) (**b**), Pol (**c**) and Control (Ctr) (**a**). Images of a representative experiment were chosen for the figure. Scale bar = 10 µm.

#### 10 μm. *3.2. Res and Pol Treatments Prompt DBSCs Proliferation and Mineral Matrix Nodules Deposition on PC Sca*ff*olds Presenting Pores of Medium Dimension (0.9 mm)*

We analyzed the proliferation capacity of our cell model on PC scaffolds and their ability to differentiate into osteoblast-like cells producing mineralized matrix. Thus, we previously demonstrated also by FT-IR microscopic analysis that dental stem cells express osteoblastic features [34].

– DBSCs were seeded on the biomaterials, which presented pores of medium dimensions (0.9 mm), and cultured in the osteogenic media without any treatment (Ctr) or exposed to Res and Pol treatments for a period of 4 weeks. Although in the first weeks of culture it was particularly difficult to find cells visible enough to be photographed using a phase contrast microscope (data not shown), after 3 weeks of differentiation, as shown in Figure 3a–c, cells appeared numerous and established strong contacts among them. In particular, in the control (Figure 3a), cells seemed to fill the corners of the scaffold pores, leaving a hole without any cell in the center of them. On the other hand, cells treated with Res and Pol (Figure 3b,c) were able to proliferate and interact with each other to cover the scaffold pores almost totally and worked to close them practically in a uniform way. Furthermore, long term cultures of DBSCs showed that the formation of calcium-rich deposits, evaluated by using the ARS after 28 days of osteogenic differentiation, was evident in the control (Figure 3d) and strengthened in the treatments (Figure 3e,f). Interestingly DBSCs capacity of mineralized matrix production was highly promoted when the scaffolds were used in combination with Pol treatment: the ARS quantification shown in the graph (Figure 3g) revealed that the production of mineral matrix nodules was greater in cells treated with 0.1 µM Res compared to the Ctr (19.65%), and remarkably enhanced when cells were exposed to 0.1 µM Pol (37.84%) if compared to the Ctr.

– 100 μm. ( – 0.001. Student's **Figure 3.** DBSCs proliferation and mineral matrix deposition on medium pore scaffolds. (**a**–**c**) Representative phase contrast pictures of DBSCs treated with Res, Pol or Ctr for 21 days in osteogenic conditions on scaffolds presenting pores of medium dimensions (0.9 mm). Scale bar = 100 µm. (**d**–**f**) ARS (Alizarin red staining) displayed mineral matrix deposition by DBSCs after 28 days of culture. (**g**) The graph shows ARS quantification using the optical density (OD) as mean percentage ± SD and is representative for three independent experiments performed in quadruplicates. § *p* < 0.01, \* *p* < 0.001. Student's *t*-test was used for single comparisons. The biomaterial pores of a representative experiment were chosen for the figure.

#### *3.3. Combined E*ff*ect of PC Sca*ff*olds and Polydatin on DBSCs Proliferation and Mineralization*

– Since we observed a greater osteogenic potential when the treatment with Pol was present, to further explore the effect of this molecule on DBSCs osteoblastic differentiation, we cultured the cells on PC scaffolds presenting pores of two other different dimensions: small (0.75 mm) and large (1.15 mm). DBSCs were maintained in mineralizing conditions and stimulated with Pol for 28 days until the deposition of mineralized matrix. As shown in Figure 4, DBSCs proliferation on small pore scaffolds advanced with the progress of culture time (Figure 4a–d), and Pol treatment induced a substantial increase in the number of cells attached to the pores, an effect which was particularly evident after three weeks of osteogenic differentiation (Figure 4c,d). After 28 days of culture, we evaluated with

ARS how Pol stimulation influenced the mineralization capacity of our cell model and we observed that deposition of mineral matrix nodules was significantly higher in cells cultured with Pol, compared to the control (Figure 4e).

– 100 μm. ( < 0.001. Student's **Figure 4.** DBSCs proliferation and mineral matrix deposition on small pore scaffolds. (**a**–**d**) Representative phase contrast pictures of DBSCs treated with Pol or not (Ctr) for 14 days (**a**,**b**) and 21 days (**c**,**d**) in osteogenic conditions on scaffolds presenting pores of small dimensions (0.75 mm). Scale bar = 100 µm. (**e**) The graph shows ARS quantification using the OD as mean percentage ± SD and is representative for three independent experiments performed in quadruplicates. \* *p* < 0.001. Student's *t*-test was used for single comparisons. The biomaterial pores of a representative experiment were chosen for the figure.

– Interestingly, cells cultured on scaffolds presenting pores of large dimensions (Figure 5a–g) did not respond as well as those seeded on scaffolds with small pores. A very low number of cells were able to colonize the pores, the proliferation was not increased by the passing of time and the Pol treatment did not show any clear effect. Moreover, the ARS quantification evidenced that there was no significant difference between Ctr and Pol in the mineralization degree (Figure 5g).

pore

– 100 μm. ( independent experiments performed in quadruplicates. Student's t **Figure 5.** DBSCs proliferation and mineral matrix deposition on large pore scaffolds. (**a**–**d**) Representative phase contrast pictures of DBSCs treated with Pol or Ctr for for 14 days (**a**,**b**) and 21 days (**c**,**d**) in osteogenic conditions on scaffolds presenting pores of large dimensions (1.15 mm). Scale bar = 100 µm. (**e**,**f**) ARS (red staining) displayed mineral matrix deposition by DBSCs after 28 days of culture. (**g**) The graph shows ARS quantification using the OD as mean percentage ± SD and is representative for three independent experiments performed in quadruplicates. Student's *t*-test was used for single comparisons. The biomaterial pores of a representative experiment were chosen for the figure.

#### **4. Discussion and Conclusions**

It is well known that the tissue engineering market, which was globally worth about \$4.7 billion in 2014, is estimated to reach a value close to \$5.5 billion by 2022, considering only the US market. Adult stem cell research is today in an advanced phase of trialing and, in some diseases, cells are

already part of therapeutic protocols for the treatment of illness and disabilities [35]. The involvement of precision medicine or even customized medicine proposes the personalization of health care with therapies, practices and/or "tailor-made" medical devices for the specific patient to be treated. The availability of optimized scaffolds, with shapes perfectly matching the lesion, would further reduce tissue regeneration times, especially after highly invasive surgical procedures. 3D printing is an excellent approach to design personalized scaffolds [36,37].

A correct regeneration process of hard tissues, as bone and cartilage, needs a biocompatible scaffold able to promote MSCs differentiation and transform a tissue repair in architectural and functional recovery. Bone lesions have multiple possible shapes and dimensions depending on the trauma or on the course of the chronic degenerative process [38].

In the case of MSCs bioengineering, the grafting site shape, its environment, morphology and dimension are basic for cell engraftment and differentiation [39].

Customized scaffolds, made rapidly and efficiently by 3D printers, could easily reproduce the perfect shape for the lesion and correctly create the ideal niche for MSCs engraftment [40], and their osteogenic differentiation would be optimized by the oligostilbenes Res and Pol.

In our study we found that both Res and Pol stimulated MSCs adhesion to the bone matrix protein Osteopontin via αVβ<sup>3</sup> integrin and, specifically, Pol treatment prompted a greater reorganization of this integrin in focal adhesion sites. The elongated strings observed by immunofluorescence (Figure 2b,c) represent the classic arrangement of αVβ<sup>3</sup> implicated in focal adhesion complexes. We can speculate that the already demonstrated osteogenic effect of Res and Pol on DBSCs [19] could be also related to the reorganization of αVβ<sup>3</sup> integrin in focal adhesions. Moreover, as already known, the development of focal complexes on the surface of scaffolds is an essential event to trigger signals that stimulate MSCs proliferation and osteogenic differentiation [41,42]. Considering these two issues, we can state that oligostilbenes can be considered osteoinductive.

Furthermore, when we integrated MSCs on PC scaffolds, we found that both Res and Pol were able to induce the mineral matrix deposition. Gathering our observations, we can establish that the scaffolds were able to support the production of mineralized matrix, which is the final step and the main event of MSCs osteogenic differentiation, and, in addition, the treatment with the molecules object of our study positively assisted the mineralization process. In particular, in agreement with what we have recently demonstrated [19], Pol treatment induced an increase in the mineralization degree that was higher than the one observed in Res treatment.

Moreover, examining the structure of the scaffolds, we studied whether different pore sizes could affect MSCs acquisition of the osteogenic features. Thus we printed PC scaffolds with pores of 0.75, 0.9 and 1.15 mm; MSCs were cultured on them and induced to osteogenic differentiation. We focused on the use of Pol as treatment since we observed, in the initial experiments, that this molecule had a greater effect in the formation of calcium-rich deposits in differentiated MSCs when compared to Res treatment. The observed gathered data led us to conclude that the cell number tended to gradually decrease as the surface micropore was getting larger and subsequently also the mineralization capacity (Figures 3–5). We compared the results to detect the ideal pore size for cell proliferation and osteogenic differentiation and we deduced that the dimension of 0.75 mm represented the best size to be created with the 3D printer, among the different pore sizes analyzed; the smaller pores produce the optimal niche for MSCs to promote bone formation.

Thus, in conclusion, in this context we confirmed the osteogenic potential of Pol treatment on MSCs. Then we made a step forward by finding, in the combination of this treatment with PC scaffolds presenting small-sized pores, an optimal strategy to induce the osteogenic differentiation of MSCs and the subsequent deposition of mineralized matrix.

The results of this study suggest that the integration of the scaffolds, opportunely designed by 3D printing with MSCs, could optimize tissue regeneration; moreover Pol could be considered a promising approach to improve bone regeneration encouraging further studies for a deeper understanding of its biological mechanisms.

**Author Contributions:** Conceptualization, G.M., F.P., A.D.B. and G.P.; formal analysis, F.P., A.D.B. and G.M.; investigation, F.P. and A.D.B.; methodology, F.P. and A.D.B.; resources, G.M., G.R., L.L.M., E.A.C.-A.; writing—original draft preparation, F.P., G.M. and G.P.; writing—review and editing, F.P., G.M., G.P. and E.A.C.-A.; visualization, F.P. and A.D.B.; supervision, G.M.; funding acquisition, G.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Ministero dell'Istruzione, dell'Università e della Ricerca—PRIN 20098KM9RN, PI G.M. (Progetto di Ricerca d'Interesse Nazionale—Grant 2009). A.D.B. has received funding from the Fondo di Sviluppo e Coesione 2007–2013, APQ Ricerca Regione Puglia Programma regionale a sostegno della specializzazione intelligente e della sostenibilità sociale ed ambientale—Future In Research.

**Conflicts of Interest:** Francesca Posa, Adriana Di Benedetto, Giampietro Ravagnan, Lorenzo Lo Muzio and Giorgio Mori are name inventors of the Italian patent (16999PTIT deposited with application number 102017000079581) titled "Compositions comprising or consisting of Polydatin in the treatment of bone pathologies" related to the work described.

#### **Abbreviations**


#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Enhanced Osteogenic Di**ff**erentiation of Human Primary Mesenchymal Stem and Progenitor Cultures on Graphene Oxide**/**Poly(methyl methacrylate) Composite Sca**ff**olds**

**Katarzyna Krukiewicz 1, \* , David Putzer 2 , Nicole Stuendl 3 , Birgit Lohberger 3 and Firas Awaja 4,5, \***


Received: 4 June 2020; Accepted: 3 July 2020; Published: 5 July 2020

**Abstract:** Due to its versatility, small size, large surface area, and ability to interact with biological cells and tissues, graphene oxide (GO) is an excellent filler for various polymeric composites and is frequently used to expand their functionality. Even though the major advantage of the incorporation of GO is the enhancement of mechanical properties of the composite material, GO is also known to improve bioactivity during biomineralization and promote osteoblast adhesion. In this study, we described the fabrication of a composite bone cement made of GO and poly(methyl methacrylate) (PMMA), and we investigated its potential to enhance osteogenic differentiation of human primary mesenchymal stem and progenitor cells. Through the analysis of three differentiation markers, namely alkaline phosphatase, secreted protein acidic and rich in cysteine, and bone morphogenetic protein-2 in the presence and in the absence of an osteogenic differentiation medium, we were able to indicate a composite produced manually with a thick GO paper as the most effective among all investigated samples. This effect was related to its developed surface, possessing a significant number of voids and pores. In this way, GO/PMMA composites were shown as promising materials for the applications in bone tissue engineering.

**Keywords:** bone regeneration; graphene oxide; mesenchymal stem and progenitor cells; osteogenic differentiation; poly(methyl methacrylate)

#### **1. Introduction**

Since its discovery in 2004 [1], graphene has drawn immense attention of the scientific community and has become an object of intensive research. Due to its high planar surface area, superior mechanical strength, outstanding optical properties, as well as remarkable thermal and electrical conductivity [2–4], graphene has been widely used in a variety of applications, including transparent conductors, ultrafast transistors, precise biosensors, and tissue scaffolds [5]. The potential of graphene has been further expanded by introducing to its structure a variety of functional group, resulting in the fabrication of graphene oxide (GO). GO is usually produced by the oxidation of graphite, resulting in a partial

breaking of sp<sup>2</sup> bonds present in its structure and subsequent increase in the distance between carbon layers [3]. Therefore, GO possesses both hydrophobic (due to the presence of pristine graphite structure) and hydrophilic (due to the presence of hydroxyl, epoxy, carbonyl, and carboxyl groups) parts, and is characterized by affinity for aromatic rings, excellent aqueous processability, amphiphilicity, ease of functionalization, and biocompatibility [3,5]. Consequently, GO has been marked as an excellent material for numerous biomedical applications, including the design of biosensors [5], drug delivery systems [6], antimicrobial coatings [7], cell imaging platforms [8], and in gene therapy [9].

Due to its versatility, small size, large surface area, and ability to interact with biological cells and tissues, GO is an excellent filler for various polymers and is frequently used to expand their functionality. For instance, Wan et al. [10] reported an increase in the tensile strength, modulus, and energy at break, as well as the improvement in bioactivity during biomineralization simultaneously with maintaining high porosity when an electrospun poly(ε-caprolactone) membrane was reinforced with GO nanoplatelets. Also Baradaran et al. [11] observed the increase in elastic modulus and fracture toughness, as well as the promotion in osteoblast adhesion and proliferation when GO was used as a filler for hydroxyapatite nanosheets. On the other hand, calcium phosphate mineralized graphene oxide/chitosan scaffolds were found to express biomimicry, providing a suitable environment for cell adhesion and growth, and maintaining high mechanical strength [12]. GO was also demonstrated to act as an excellent filler for such polymer matrices as poly(vinyl alcohol) [13], poly(carbonate urethane) [14], hyaluronic acid [15], and poly(acrylic acid) [16], resulting in the formation of robust composite materials with applicability in biomedical engineering.

Poly(methyl methacrylate) (PMMA) is a non-toxic thermoplastic polymer possessing a very good toxicological safety record in biomedicine [17]. PMMA is frequently used as a screw fixation in bone, bone cement, filler for bone cavities and skull defects, as well as vertebrae stabilization in osteoporotic patients [18]. The exceptional applicability of PMMA in orthopedic and dental applications is caused by its good processability, handling properties, biocompatibility, suitable mechanical strength, and Young's modulus [19]. Despite its numerous merits, the common complications of using PMMA is bone resorption observed after implantation, which is the effect of the formation of a weak-link zone derived from not sufficient interactions between cement and a bone [20]. Therefore, the modification of PMMA has become a very active area of research. A promising way to improve the biological performance of PMMA is to blend it with an antibiotic, e.g., gentamycin [21]. This modification approach is nowadays a well established strategy that allows prevention of periprosthetic infections and osteomyelitis. Another way to enhance the performance of PMMA is to incorporate in its structure a filler with a particular functionality. For instance, loading of PMMA with multiwalled carbon nanotubes could significantly improve the mechanical properties and reduce the exothermic polymerization reaction of the bone cement [22]. The use of tri-calcium phosphate and chitosan as inorganic/organic additives to PMMA decreased polymer curing temperature, extended setting time, and increased weight loss and porosity after degradation and, among all, promoted better osteointegration than pure PMMA [23]. Also graphene and graphene oxide have been used as fillers to PMMA [24], improving the mechanical properties of PMMA, particularly its fracture toughness and fatigue performance. What was interesting, GO was found to outperform graphene and provide greater enhancements due to its high functionalization that increased the interfacial adhesion between a filler and PMMA matrix. Simultaneously, the presence of graphene or GO was not found to have any negative effect on the biocompatibility of PMMA composites, potentially allowing their further clinical progression [25].

In this paper, the potential of GO/PMMA composites for the application in bone tissue engineering is assessed by the analysis of three differentiation markers expressed by human primary mesenchymal stem and progenitor cells (hMSPCs) cultured on the top of the composites. By performing the cell culture in the presence and absence of a specific induction medium, we were able to determine the efficiency of osteogenic differentiation of hMSPCs cultured on four types of GO/PMMA scaffolds, differing in the thickness of GO paper as well as the method of fabrication of the composites. Microscopic

analysis of the surface of materials allowed investigation of the biological behavior of the materials with respect to their surface morphology.

#### **2. Materials and Methods**

#### *2.1. Fabrication of GO*/*PMMA Composites*

For the production of graphene oxide (GO) paper, 4 mg/mL suspension of GO flakes (GO monolayer content > 95%, oxygen content between 40% and 50%) in water was purchased from Graphenea (San Sebastián, Spain). GO suspension was inserted into Petri dishes with serological pipettes. The GO dispersion was dried in a shaking incubator with air fan for 48 h and inserted into an oven at 180 ◦C for 20 min. By changing the volume of GO suspension, two types of GO paper were fabricated, i.e., thin (8 mL of GO suspension) and thick (10 mL of GO suspension), designated as GO(A) paper and GO(B) paper, respectively.

For the fabrication of GO/PMMA composites, GO paper as well as SIMPLEX P (Stryker, Kalamazoo, MI, USA) radiopaque bone cement (prepared according to the manufacturer's instructions with a full dose of liquid monomer) were applied. The monomer was mixed to the polymer manually under laboratory conditions. The PMMA cement was then inserted in a metal casting form and covered with GO paper. The bone cement was kept in place for 30 min to guarantee sufficient hardening. Screws were closed after 15 min. Two different methods were used to prepare combined samples including GO paper and bone cement (Scheme 1).

**Scheme 1.** Schematic representation of a fabrication process of GO/PMMA(P) (**A**) and GO/PMMA(M) (**B**) composite materials.

In the first method, GO paper was placed on the bottom of a steel flat press (covered with poly(tetrafluoroethylene), PTFE, sheet to simplify cement detachment). The cement was spread on the upper PTFE sheet and then placed in contact with the GO paper. Then, the press was closed after 15 min to reach minimum thickness until the cement was polymerized (30 min). The samples prepared in this way were designated as GO/PMMA(P). In the second method, a compound material was produced manually: GO paper was laid down on a PTFE sheet, and then the cement was spread on the upper PTFE sheet and then placed in contact with the GO paper. An aluminum bar was used to spread the cement on GO paper within the two PTFE sheets. In this case, spreading the cement was more difficult, and led to the formation of a non-homogeneous PMMA layer with GO paper broken up into little pieces, which may have been due to the shrinking and expanding behavior of PMMA during the cement hardening phase. The samples prepared in this way were designated as GO/PMMA(M).

Consequently, four types of samples were analyzed: GO(A)/PMMA(P), GO(B)/PMMA(P), GO(A)/PMMA(M), and GO(B)/PMMA(M), all with PMMA as the surface layer. For the scanning electron microscopic (SEM) investigations, a FEI Quanta 250 field emission gun (Thermo Fisher Scientific, Hillsboro, OR, USA) was used under high vacuum conditions and 20 kV high tension. The micrographs were recorded with the Everhart–Thornley–Detector in secondary electron (SE)

mode. The surfaces were sputter coated with a 10 nm thin layer of gold in order to provide sufficient electrical conductivity.

#### *2.2. Tissue Harvest and Cell Culture*

Explant hMSPCs were established from tissue samples of spongiosa bone harvested during routine hip joint surgeries. The study protocol was approved by the local ethics committee (reference number 29-156ex16/17), and informed consent was obtained from each orthopedic surgery patient. The study included a total of three patients, aged between 25 and 35, excluding pregnant women and those suffering from local inflammatory processes, metabolic bone diseases, and impaired blood coagulation. The length of harvested bone samples was kept between 4 and 6 mm, and showed either cortical or cortical and cancellous structure. The samples were extensively rinsed with a phosphate-buffered saline (PBS; PAA Laboratory, Pasching, Austria) and transferred into 75 cm<sup>2</sup> culture flasks (TPP, Trasadingen, Switzerland) with an appropriate volume of culture medium. For cell isolation and expansion, the samples were incubated in a humidified atmosphere (5% CO2, 37 ◦C).

#### *2.3. Flow Cytometry*

For a flow cytometric analysis, a total of 1 <sup>×</sup> <sup>10</sup><sup>5</sup> hMSPCs were resuspended in 200 <sup>µ</sup>L PBS. The characterization of cells was achieved with the use of commercial monoclonal antibodies, namely CD73 PE, CD90 APC, CD105 PE, CD45 APC-Cy7, CD34 APC, CD14 FITC, CD19 APC, and HLA-DR APC (BD Bioscience, San Jose, CA, USA). Titration had previously been used to determine the optimal amount of each antibody. Subsequently, two-color staining panels were used to present a combination of antibodies with non overlapping spectra. Negative cell lines and matched fluorochrome-conjugated isotype controls were applied to perform a background staining for antibodies. FACS LSR II System (BD Bioscience), FACSDiva software (BD Bioscience), and FCS Express software (De Novo Software, Los Angeles, CA, USA) were employed to perform a flow cytometry analysis, to acquire and to analyze obtained data, respectively. Rainbow Beads (BD Bioscience) was used to check the day-to-day consistency of measurements. To exclude debris and cell aggregates, viable cells were gated on forward scatter (FSC) and side scatter (SSC). hMSPCs were defined by their phenotype and analyzed on a logarithmic scale. Data from all donors were analyzed by collecting 10,000 events under identical parameters.

#### *2.4. Multilineage Di*ff*erentiation Analysis*

A seeding density of hMSPCs was established at 10<sup>4</sup> cells/cm<sup>2</sup> , and the cells were seeded in an expansion medium composed of Dulbecco's modified Eagle's medium (DMEM-F12; GIBCO Invitrogen), 10% FBS (Lonza, Braine-l'Alleud, Belgium), 1% L-glutamine, 1% penicilline/streptomycine, and 0.1% amphotericine B. Additionally, 100 nM dexamethasone, 0.1 mM ascorbic acid-2-phosphate, and 10 mM β-glycerophosphate (all Sigma Aldrich, St. Louis, MO, USA) were added to the differentiation medium to induce osteogenesis. Histochemical staining (Alkaline Phosphatase kit No. 85; Sigma Aldrich) was used to assay alkaline phosphatase (ALP) activity after 7 and 14 days of culture. According to the instructions of the manufacturer, ALP enzyme activity was calculated basing on the absorbance of p-nitrophenol phosphate (405 nm) [26]. Adipogenic differentiation was performed in a medium containing 100 nM dexamethasone, 50 µM indomethacine (Sigma Aldrich), and 0.135 IE/mL insulin (Novo Nordisk, Bagsværd, Denmark), and was detected by Oil Red O staining of the adipocyte specific fat vacuoles after 21 days of culture. Chondrogenic differentiation was initiated by culturing cells in DMEM-F12 supplemented with 10% FBS, 100 µM L-ascorbic acid, and 1 ng/mL TGF-β3 (Lonza). Alcian blue staining was applied to verify the production of glycosaminoglycans and mucopolysaccharides after 21 days of culture. Cells were then fixed with 10% formaldehyde and stained with 1% Alcian blue in 3% acetic acid solution at pH 2.5.

#### *2.5. Real-Time RT-PCR*

RNeasy Mini Kit and DNase-I treatment (Qiagen, Hilden, Germany) were used to isolate total RNA from undifferentiated and osteogenic differentiated hMSPCs cultured on different GO surface modifications (on the GO-uppermost surface of the composites) on day 21. A total of 1 µg of RNA was reverse transcribed with iScriptcDNA Synthesis Kit, (BioRad Laboratories Inc., Hercules, CA, USA) using a blend of oligo(dT) and random hexamer primers. SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) and CFX96 Touch (BioRad) were used for the amplification and measurements, respectively. A standard 3-step PCR temperature protocol (annealing temperature of 60 ◦C) was used for each qPCR, and was followed by a melting curve protocol both to confirm a single gene-specific peak and to detect primer dimerization. ∆∆Ct method was used for the relative quantification of expression levels, and was based on the geometric mean of the internal controls TBP (TATA-box binding protein), RPLP0 (ribosomal protein, lateral stalk, subunit P0), and B2M (β-2 microglobulin), respectively. The expression levels (Ct) of the target genes were normalized to the reference genes (∆Ct), and the difference between the ∆Ct value of the test sample and the ∆Ct of the control sample gave the ∆∆Ct value. Consequently, the expression ratio was calculated as 2∆∆Ct. Three QuantiTect primer assays (Qiagen) were selected for real time RT-PCR, namely ALPL, SPARC, and BMP2. ΔΔ β Δ Δ Δ ΔΔ ΔΔ

#### *2.6. Statistical Analysis*

Differences between groups were evaluated by means of a Student's unpaired *t*-test and the exact Wilcoxon test with the PASW statistics 18 software (IBM Corporation, Somers, NY, USA). Two-sided *p*-values (*p* < 0.001 \*\*\*; *p* < 0.01 \*\*; *p* < 0.05 \*) were considered statistically significant. SigmaPlot® 14.0 (Systat Software Inc., San Jose, CA, USA) was used to make graphical representations.

#### **3. Results**

#### *3.1. Surface Characterization of GO*/*PMMA*

The protocol of GO paper fabrication resulted in the formation of a self-supporting, uniform, and black material, with the average thickness of 5 ± 1 µm for GO(A) and 16 ± 1 µm for GO(B), and the average specific weight of 0.87 <sup>±</sup> 0.08 mg/cm<sup>2</sup> for GO(A) and 2.90 <sup>±</sup> 0.08 mg/cm<sup>2</sup> for GO(B). The morphology of GO(A) and GO(B), as presented in the SEM images (Figure 1), showed some wrinkles on the surfaces which were the most probably the edges of graphene oxide, revealing strong adhesion between GO platelets. Overall, the surface of GO paper was relatively smooth, and there were no obvious defects (pores or cracks) observed.

**Figure 1.** SEM micrographs presenting surface morphology of GO(A) and GO(B) paper.

As demonstrated in SEM micrographs of GO/PMMA composites (Figure 2), PMMA was covering the surface of GO paper, and more uniform surface was obtained when a thin GO(A) paper was used as a filler. The interface between GO and PMMA can be seen as a border region, particularly in the SEM image of GO(A)/PMMA(P). GO(A) paper was thickly coated with PMMA, while the presence of a thick GO(B) paper was found to introduce wrinkles to the surface of polymer composite. Moreover, GO(B) paper seemed to protrude cleanly from the fracture site. The surface of GO(B)/PMMA(M) was observed to exhibit a significant number of voids and pores.

**Figure 2.** SEM micrographs presenting surface morphology of GO(A)/PMMA(P), GO(B)/PMMA(P), GO(A)/PMMA(M), and GO(B)/PMMA(M) (arrows indicate discussed surface structures), as well as untreated PMMA, with high magnification images as the insets.

#### *3.2. hMSPC Characterization and Multilineage Di*ff*erentiation Analysis*

Cells providing morphologic characteristics of human primary MSPCs (mononuclear, fibroblast-like, spindle shaped, plastic-adherent) were isolated from all samples within 4–8 days. hMSPCs showed a positive expression of CD73 (99.8 ± 0.1%), CD90 (99.9 ± 0.1%), CD105 (69.1 ± 9.8%) of gated cells. The typical forward/side scatter characteristics of 71.5 ± 4.9% were gated. The negativity for CD14 (0.2 ± 0.2%), CD19 (0.6 ± 0.1%), CD34 (0.4 ± 0.3%), CD45 (23.9 ± 7.8%), and HLA-DR (0.5 ± 0.3%) confirmed the phenotype of MSPCs (Figure 3A).

ALP activity was measured of absorbance (optical dense, OD) of p-nitrophenol in supernatant at the wavelength of 405 nm over 14 days (Figure 3B). ALP expression was detected on day 7 and day 14, respectively, when the cells were osteogenically differentiated with a significant increase (*p* < 0.001). No expression of ALP was observed in any of the samples of undifferentiated negative controls. Due to the interaction of the cationic dye Alcian blue and acid glycosaminoglycans, augmented blue coloration was noticed for chondrogenic differentiated hMSPCs, and not for undifferentiated controls (Figure 3C). As a result of chondrogenic differentiation, a 4.7-fold increase (*p* < 0.05) was noticed for the expression of aggrecan. In order to demonstrate the multilineage ability, hMSPCs were also differentiated in the adipogenic lineage. The adipogenic cell differentiation was demonstrated with the formation of lipid vacuoles which were visualized by Oil Red O staining on day 21 (Figure 3D). These results explicitly characterized our primary cells as hMSCPs.

**Figure 3.** hMSPC characterization and multilineage differentiation analysis. The used hMSPCs were characterized according (**A**) the positive expression of CD73, CD90, CD105, and negative expression CD14, CD19, CD34, CD45, and HLA-DR using multicolor fluorescence-activated cell sorting analyses. The values indicated the percentage of positively stained cells. The capacity for multilineage differentiation potential was confirmed by (**B**) ALP staining for osteogenic differentiation, (**C**) Alcian blue staining and the expression of aggrecan for the chondrogenic differentiation, and (**D**) the Oil Red O staining of lipid droplets for the adipogenic lineage; *p* < 0.01 \*\*; *p* < 0.05 \*).

#### *3.3. E*ffi*ciency of Osteogenic Di*ff*erentiation*

ALP, SPARC, and BMP-2 assays were performed to assess the mineralization of hMSPCs cultured on the surface of GO(A)/PMMA(P), GO(B)/PMMA(P), GO(A)/PMMA(M) and GO(B)/PMMA(M), as well as untreated PMMA. Consequently, the results shown in Figure 4 describe how strongly the expression of individual markers was increased by the osteogenic differentiation medium, with the undifferentiated hMSPCs as the control. As demonstrated by ALP assay, all investigated surfaces led to a significant increase in mineral deposition, with the most pronounced effect noticed for GO(B)/PMMA(M) (8-fold increase when compared with a control). The same material was also found to lead to the significant increase in the expression of SPARC (2-fold increase with respect to control), even though the highest relative gene expression (3-fold increase with respect to control) was noticed for unmodified PMMA. All investigated surfaces, including GO/PMMA composites as well as unmodified PMMA samples, were shown to decrease the relative BMP-2 expression from 7 to 9 times when compared with a control.

The relative gene expression profiles of hMSPCs cultured in normal expansion medium were analyzed with respect to ALP, SPARC and BMP-2, and compared with an unmodified PMMA as the control. Consequently, the results presented in Figure 5 describe the expression of the individual markers by the osteogenic differentiation medium in relation to an unmodified PMMA control These results showed the unchanged SPARC expression, simultaneously with the decrease in ALP expression (approximately 2-fold) and a significant increase of BMP-2 expression (from 2-fold for GO(A)/PMMA(P) to 2.5-fold for GO(A)/PMMA(M)) for GO/PMMA composites with respect to a PMMA control.

**Figure 4.** The expression of specific osteogenic markers (ALP, SPARC, and BMP-2) of hMSPCs cultured in the presence of osteogenic differentiation medium for 21 days on the surface of GO/PMMA composites as well as untreated PMMA, compared with undifferentiated hMSPCs cultivated in expansion medium as the negative control; *p* < 0.001 \*\*\*; *p* < 0.01 \*\*; *p* < 0.05 \*.

**Figure 5.** The expression of specific osteogenic markers (ALP, SPARC, and BMP-2) of hMSPCs cultured in the absence of a specific induction medium for 21 days on the surface of GO/PMMA composites as well as untreated PMMA; *p* < 0.001 \*\*\*; *p* < 0.01 \*\*; *p* < 0.05 \*.

To estimate the efficiency of osteogenic differentiation, GO/PMMA coatings were compared with an unmodified PMMA as a control (Figure 6). This kind of evaluation was chosen to show the efficiency of the osteogenic differentiation ability of each group in relation to the PMMA control. Particularly, GO(B)/PMMA(M) composite was found to improve the expression of osteogenic differentiation markers (2-fold increase for ALP, 1.5-fold increase for SPARC, and 2-fold increase for BMP-2), while GO(A)/PMMA(M) was found to decrease ALP expression (2-fold). All other effects were assessed as not significant.

**Figure 6.** The expression of specific osteogenic markers (ALP, SPARC, and BMP-2) of hMSPCs cultured for 21 days on the surface of GO/PMMA composites with respect to a PMMA control; *p* < 0.001 \*\*\*; *p* < 0.01 \*\*; *p* < 0.05 \*.

#### **4. Discussion**

The greatest challenge of modern nanotechnology is to expand its scope from nanoscale into macroscale. Therefore, the possibility of fabricating free-standing, paper-like materials basing on nanoscale components is a subject of intensive research [27]. The protocol of GO paper fabrication, introduced by us, allowed formation of self-supporting, uniform films composed of stacked platelets of graphene oxide. The irregularity of surface of GO paper seemed to increase with its thickness, which is of special importance for further biological studies since higher roughness promotes cell adhesion [28].

GO paper is known to possess many functional groups, and could easily form hydrogen bonds with hydrophilic polymers. The presence of hydrophobic methacrylate groups in PMMA, however, discourages these interactions to occur. Therefore, since PMMA chains would rather remain in a coiled conformation, they are supposed to fit well into wavy structures of the nanosheets of GO paper leading to more efficient packing within intersheet gallery [29]. Consequently, SEM micrographs demonstrated a good interlocking between GO paper and PMMA matrix. Particularly GO(A) paper was found to be thickly coated with PMMA, suggesting strong adhesion between this polymer and GO. GO(B) paper, on the other hand, seemed to protrude from the fracture site, suggesting a weak interfacial bonding between GO(B) and PMMA. This is consistent with some previous studies indicating that small sizes of GO sheets promote the formation of homogeneous composites [30]. Moreover, the presence of voids and pores on the surface of GO(B)/PMMA(M) could be associated with the presence of unreacted residual monomer, which is volatile and is supposed to be released after polymerization [31]. This would mean that some double bonds present in GO paper could be attacked by the radical species formed during MMA polymerization, retarding or inhibiting the reaction of polymerization [24].

As a biological model, cells providing morphologic characteristics of human primary MSPCs (mononuclear, fibroblast-like, spindle shaped, plastic-adherent) were isolated and cultured on the surface of GO/PMMA composites. The phenotype of hMSPCs was confirmed according to the criteria of the International Society for Cellular Therapy [32] for defining multipotent mesenchymal stromal cells. In addition, hMSPCs were successfully differentiated towards the osteogenic, chondrogenic, and adipogenic lineage, which was confirmed by ALP expression, Alcian blue staining, aggrecan expression, and Oil Red O staining. To assess the efficiency of osteogenic differentiation of hMSPCs cultured onto PMMA as well as GO/PMMA composites, three differentiation markers were analyzed, namely alkaline phosphatase (ALP), secreted protein acidic and rich in cysteine (SPARC), and bone morphogenetic protein-2 (BMP-2). ALP is a metalloenzyme playing an important role in the mineralization of tissue cells [33]. ALP is found to act as both a mineralization promoter by increasing the local concentration of phosphate, as well as an inhibitor of mineral formation by decreasing the concentration of extracellular pyrophosphate. Since ALP is observed to be highly expressed in mineralized cells, it can be used to predict their bone forming capacity under different conditions. Osteonectin (SPARC), on the other hand, is the most abundant non-collagenous extracellular matrix protein present in bone [34]. SPARC gene dosage has a dramatic effect on bone volume and is one of critical regulators of bone remodeling, calcium turnover and an initiator of mineralization [34,35]. Therefore, SPARC can be used for the examination of osteogenic differentiation. Another marker for osteogenic differentiation is bone morphogenetic protein-2 (BMP-2), which is known as a potent osteogenic factor with roles in both normal bone healing and pathological bone formation in soft tissues [36]. BMP-2 is found to facilitate osteogenic differentiation through inducing ALP activity, promoting mineralization and enhancing adherence of cells [37].

As demonstrated by ALP assay, all investigated surfaces led to a significant increase in mineral deposition. This effect was accompanied with a decrease in BMP-2 expression, which might suggest that the presence of a differentiation medium had a stronger effect on the osteogenic differentiation than BMP-2, resulting in down regulation of the latter. Nevertheless, the expression of specific osteogenic markers, such as ALP and SPARC, was found to be significantly increased by the osteogenic differentiation medium. In the absence of a specific induction medium, on the other hand, the cells cultured on GO/PMMA composites were found to be able to induce an increase in BMP-2. In this way, GO/PMMA composites were shown to be able to drive cellular differentiation without any addition of osteogenic supplements, just as reported for hMSCs in collagen matrices subjected cyclic tensile strain [38], even though the process of osteogenic differentiation was found to be much more effective when cells were culture in the presence of a medium trigger.

The comparison of GO/PMMA coatings with an unmodified PMMA as a control suggested that among all investigated GO/PMMA composite materials, GO(B)/PMMA(M) composite was the most efficient inducer of osteogenic differentiation, particularly basing on the relative expression of ALP (other changes were found to be statistically non-significant). This effect should be assigned to its surface morphology, as exhibiting a significant number of voids and pores. As presented by Abagnale et al. [39], specific patterns present on the surface can boost differentiation of MSCs towards specific cell types. These patterns should be in the micrometer range to be able to support the differentiation processes initiated by induction media, and this requirement is met particularly by GO(B) paper and GO(B)/PMMA(M) composite. All these data suggest that GO/PMMA composites, particularly GO(B)/PMMA(M) produced manually with a thick GO paper, may direct hMSPCs toward osteogenic differentiation and can serve as promising materials in bone tissue engineering. The properties of GO/PMMA make this material advantageous for potential applications as screws or implants fixation. Previous studies [25,40] showed that PMMA filled with GO conforms to physicochemical and mechanical demands of these clinical applications. In this study, we have shown that incorporation of GO into PMMA matrix may provide an additional functionality to the resulting composite material, enhancing its biocompatibility through facilitating osteogenesis. Still, before the introduced material could be considered as a bone cement, further studies should include a comprehensive biomechanical characterization of GO/PMMA.

#### **5. Conclusions**

In this paper, we present the potency of GO/PMMA composites to induce osteogenic differentiation of hMSPCs. Through the analysis of three differentiation markers, namely ALP, SPARC, and BMP-2 in the presence and in the absence of a specific induction medium, we were able to assess the efficiency of osteogenic differentiation of hMSPCs cultured on four types of GO/PMMA composites, differing in the thickness of GO paper as well as the method of fabrication of the composite. All investigated GO/PMMA composite materials were found to effectively induce osteogenic differentiation, and to outperform both unmodified PMMA and a negative control (undifferentiated hMSPCs). Among GO/PMMA composite materials, a composite produced manually with a thick GO paper (GO(B)/PMMA(M)) acted as the most efficient inducer of osteogenic differentiation, particularly basing on the relative expression of ALP (other changes were found to be statistically non-significant). Since GO(B)/PMMA(M) was the composite surface possessing a significant number of voids and pores, its developed surface morphology was supposed to be responsible for directing hMSPCs toward osteogenic differentiation. In this way, GO/PMMA composites, and particularly GO(B)/PMMA(M), were shown as promising materials in bone tissue engineering.

**Author Contributions:** Conceptualization, D.P., B.L., and F.A.; Investigation, D.P., N.S., B.L., and F.A.; Methodology, D.P., B.L., and F.A.; Supervision, F.A.; Writing—original draft, K.K.; Writing—review and editing, D.P., B.L., and F.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** The cell studies described in this work were supported by the FFG Bridge program (Grant No. 861608). K.K. would like to thank the National Science Center, Poland (SONATA program, Grant No. 2016/23/D/ST5/01306) and the Silesian University of Technology, Poland (04/040/BK\_20/0113).

**Acknowledgments:** The authors thank Sandra Ngo and Mariangela Fedel for their support in fabricating the GO, and Heike Kaltenegger for her support in isolating and characterizing the primary hMSPCs.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
