ZnO nanoparticles-modified polycaprolactone-gelatin membranes for guided/bone tissue regeneration, antibacterial and osteogenic differentiation properties

Periodontitis is a highly prevalent infectious disease that causes the progressive destruction of the periodontal supporting tissues. If left untreated, it can lead to tooth loss impairing oral function, aesthetics, and the patient’s overall quality of life. Guided and Bone Tissue Regeneration (GTR/BTR) are surgical therapies based on the placement of a membrane that prevents epithelial growth into the defect, allowing the periodontal/bone cells (including stem cells) to regenerate or restore the affected tissues. The success of these therapies is commonly affected by the local bacterial colonization of the membrane area and its fast biodegradation, causing postoperative infections and a premature rupture of the membrane limiting the regeneration process. This study presents the antibacterial and osteogenic differentiation properties of polycaprolactone-gelatin (PCL-G) electrospun membranes modified with ZnO nanoparticles (ZnO-NPs). The membranes´ chemical composition, surface roughness, biodegradation, water wettability, and mechanical properties under simulated physiological conditions, were analyzed by the close relationship with their biological properties. The PCL-G membranes modified with 1, 3, and 6% w/w of ZnO-NPs showed a significant reduction in the planktonic and biofilm formation of four clinically relevant bacteria; A. actinomycetemcomitans serotype b, P. gingivalis, E. coli, and S. epidermidis. Additionally, the membranes presented appropriate mechanical properties and biodegradation rates to be potentially used in clinical treatments. Notably, the membranes modified with the lowest concentration of ZnO-NPs (1% w/w) stimulated the production of osteoblast markers and calcium deposits in human bone marrow-derived mesenchymal stem cells (BM-MSC) and were biocompatible to human osteoblasts cells (hFOB). These results suggest that the PCL-G membranes with 1% w/w of ZnO-NPs are high-potential candidates for GTR/BTR treatments, as they were the most effective in terms of better antibacterial effectiveness at a lower NPs-concentration while creating a favorable cellular microenvironment for bone growth.


Introduction
The World Health Organization reports that around 14% of the global adult population suffers from severe periodontitis, representing over one billion cases worldwide [1,2]. Periodontitis is an infectious disease associated with dysbiotic oral biofilms, causing the progressive destruction of the supporting periodontal tissues (including cementum, ligament, and alveolar bone). If left untreated, it can lead to loss of tooth impairing oral function, aesthetics, and the patient's overall quality of life [3,4]. Guided Tissue Regeneration (GTR) and Bone Tissue Regeneration (BTR) treatments are the current gold-standard therapeutic techniques to promote the regeneration or restoration of periodontal tissues [5,6] and stabilize loosening teeth, or prepare the bone tissue for dental implants. The principle of both treatments is the placement of membranes overlying the periodontal/osseous defect, acting as a barrier to exclude epithelial down-growth and allowing periodontal cells to repopulate the defect site for restoring the tissues [5,6]. Periodontal regeneration predominantly relies on the regenerative capacity of endogenous cells, mainly stem cells recruited from the surrounding tissues that differentiate into specialized cells and produce cementum, periodontal ligament, and bone [7,8]. There are two main types of commercial regeneration membranes, resorbable and non-resorbable. One disadvantage of the non-resorbable membranes is that they must be removed in a second-stage surgical procedure about six months after placement [9]. Resorbable membranes (mainly composed of xenogenic collagen) must not be removed but frequently exhibit poor mechanical properties and fast biodegradation, which can cause a lack of stability, complicating surgical handling and premature rupture during the treatment, negatively affecting the periodontal regeneration process [10]. Another critical challenge is the control of local infections caused by bacterial colonization on the membrane, which is favored by the physiological fluids in the oral cavity, such as blood and saliva [11][12][13]. Microbial studies from retrieved membranes have reported the presence of periodontopathogens such as Aggregatibacter actinomycetemcomitans (A. actinomycetemcomitans) and Porphyromonas gingivalis (P. gingivalis) at high levels when treatment failures occur [14][15][16]. In this respect, local or systemic antibiotic therapy is usually administrated for preventing or treating bacterial contamination; however, the growing appearance of antibiotic-resistant bacteria demands exploring alternative antibacterial compounds [17].
The ideal regeneration membrane must be biocompatible, not elicit any adverse response, have antibacterial properties, and promote the regeneration of periodontal tissues. Additionally, the membrane should be mechanically stable for comfortable surgical handling, flexible to conform to the defect site, and biodegradable with a biodegradation rate that allows restoration of the periodontal tissues. A feasible approach to achieve these desirable properties is to develop biocompatible and biodegradable polymeric membranes functionalized with non-antibiotic, antimicrobial nano-compounds. In this respect, metallic and metal oxide nanoparticles whit antibacterial properties have received considerable attention as antimicrobial agents. Particularly, zinc oxide nanoparticles (ZnO-NPs) have gained interest by their effective microbicidal properties against wide pathogen microorganisms, including some antibiotic resistant-bacteria [18,19]. Moreover, Zn is an essential trace element in human beings [20], involved in essential cellular processes such as DNA synthesis, enzyme activity, and cell division [21,22]. It has been reported that Zn has an essential effect on bone formation and mineralization, stimulating osteoblast cells proliferation [23,24] and increasing the expression of osteoblast gene markers and calcium deposition in human bone marrow-derived mesenchymal stem cells (BM-MSC) [25,26]. Then, releasing Zn ions from GTR/ BTR membranes could promote osteoblast proliferation and osteogenic differentiation and, consequently, enhance periodontal regeneration. Taking this into account, we hypothesize that an adequate concentration of ZnO-NPs incorporated into PCL-G membranes can lead to important functional properties that would enhance the features of the membranes, such as an antibacterial effect against various strains, the induction of a favorable environment for bonelike tissue regeneration, and desired mechanical and biodegradation properties to be clinically used.
In a previous work [27], we developed microfibrillar nanocomposite membranes incorporating, via electrospinning, different concentrations of ZnO-NPs (1, 3, and 6% w/w) in homogeneous polymeric blends of polycaprolactone (PCL) and gelatin (G), that showed antibacterial effect against one clinically relevant bacterial strain, Staphylococcus aureus. Moreover, the physicochemical and mechanical characteristics of the membranes showed their potential to be used as GTR/BTR membranes; however, their properties were studied in a dry state, and the natural environment of the oral cavity is humid. The novelty of the present work relays on the fact that simulated physiological conditions for clinical applications were used to study complementary properties of the PCL-G membranes added with different concentrations of ZnO-NPs, finding the adequate amount of NPs to improve important functional properties of the membranes. Thus, their antibacterial effect was studied against different clinically relevant bacteria (P. gingivalis, A. actinomycetemcomitans serotype b, Escherichia coli, and Staphylococcus epidermidis), and their osteogenic differentiation capacity using human stem cells were analyzed, as well as their biocompatibility in periodontal cells. Additionally, the chemical composition, surface roughness, and wettability of the membranes were analyzed, as these characteristics are closely related to the biological properties of biomaterials.

Fabrication of the membranes
The membranes were fabricated as previously reported [27]. Briefly, blend solutions of PCL and gelatin were prepared by dissolving PCL (19% w/v) in AcAc, and adding the proper amount of gelatin (G) to obtain two solutions with a final PCL:G mass ratios of 70:30 and 55:45, respectively. Different amounts of ZnO-NPs (1, 3, and 6% w/w; relative to PCL content) were separately dispersed into the blend PCL:G solutions, obtaining eight different solutions for electrospinning; PCL:G (70:30) solutions with 0, 1, 3, and 6% w/w of ZnO-NPs, and PCL:G (55:45) solutions with 0, 1, 3, and 6% w/w of ZnO-NPs. An only-PCL solution in AcAc (19% w/v) was also prepared. The solutions were then stirred at 300 rpm for 48 h at room temperature (RT). The conductivity of the solutions was measured in a JENWAY 3540 conductivity meter at room conditions (temperature ≈ 24.5°C-27.6°C and relative humidity ≈ 46%-49%). The PCL:G and PCL: G:ZnO-NPs solutions were independently electrospun using a feed rate of 1 ml /h, 14 kV voltage, and 14 cm of distance from the needle to the static collector. The only-PCL solution was pumped at the same feed rate but increasing the needle-to-collector distance and the voltage to 15 cm and 15 kV, respectively. The electrospinning process was conducted for 35 min for each solution. After electrospinning, the electrospun fibers (membranes) were removed from the collector, double washed with double distilled water (dd H 2 O) and ethanol (EtOH; 70%), dried at RT, and sterilized under ultraviolet light (UV) on each side. The thickness of three different samples for each different membrane was measured using a digital micrometer (Mitutoyo 293-140 QuantuMike). The membranes were named accordingly to their composition as described in table 1, which also shows the conductivity of the solutions for electrospinning. The conductivity of the solutions increased as the gelatin concentration increased, the PCL:G (55:45) solution (6.38 μScm −1 ) showed higher conductivity than the PCL:G (70:30) solution (5.18 μScm −1 ), and both PCL:G solutions showed higher conductivity in comparison with the PCL solution (0.26 μScm −1 ). The same trend was observed with the increment of ZnO-NPs in the PCL:G (70:30) solution, where the conductivity increased from 5.73 to 5.96 μScm −1 for 1% to 6% w/w ZnO-NPs, and in the PCL-G (55:45) solution where the conductivity also increased from 6.66 to 7.03 μScm −1 for 1% to 6% w/w ZnO-NPs.

Physical and chemical characterization of the membranes
The micromorphology of the membranes synthesized for this work was characterized by Scanning Electron Microscopy (SEM; JEOL, JSM-7800F); micrographs were acquired from carbon-coated membranes samples at 3.0 kV. The diameter distribution of the fibers (d) was determined from two different SEM micrographs of each membrane, each micrograph was segmented with sixteen equally spaced squares, and each square was segmented with nine equally spaced squares. Then, the diameter of the fibers intersected with the sides of the squares was measured (AxioVision software; Carl Zeiss Microscopy), acquiring at least 280 measures for each membrane.
The incorporation of the ZnO-NPs in the electrospun membranes was characterized by Raman spectroscopy. The spectra were obtained over a wavenumber range of 600-2000 cm −1 with an ANTON PAAR Raman spectrometer model Cora 5000, using an excitation wavelength at 1064 nm, laser power at 300 mW, and integration time close to 5000 ms.
Mechanical properties were determined from membranes samples in fully hydrated state, resembling the conditions in which the membranes are intended to be used. Mechanical stress-strain tests were performed using a universal test machine (Shimadzu, AGS-X) with a 100 N load cell under a crosshead speed of 1 mm min −1 at ambient conditions (24°C and 45% of relative humidity). Three specimens of each membrane with rectangular dimensions (25 × 5 mm 2 ) according to the ASTM D1708-06a were tested. Sample thickness was measured with a digital micrometer (Mitotuyo 291-140 QuantuMike) at three different points of each specimen. The elastic modulus (E), the elongation at break (Ɛ), and the maximum tensile strength (σ max ) were calculated from the strainstress curves obtained.
The surface of the membranes was quantitatively evaluated by means of the Taylor-Hobson method measuring the amplitude roughness parameters, since the topographical pattern of the surfaces, such as grooves and ridges, can regulate cell growth, mobility, and fate [28]. The 3D profilometric characterization of the samples was conducted through profilometry tests using a Taylor-Hobson Phase Grating Interferometric (TH-PGI) transducer. The interferometry-based measurements provide as large as 400 mm range with resolutions of the order of 0.1 nm. The profiles resulting from this technique include both, roughness along with the waviness of the surface. To be able to use the data obtained for calculating the true roughness of the surfaces, the waviness curves were removed from the results. Data were smoothed with a moving span equal to 35 and then smoothed curves were subtracted from the raw data. For the scanning, a tip with a radius of 100 μm, a total scanning length of 3 mm, and a scanning speed of 0.1 mm s −1 were applied. The acquired 1D profiles of the samples were divided into 100 segments of 30 μm in length. Then, roughness measurement parameters were calculated for the height data along each segment as characterization measures. The amplitude roughness parameters can be defined in terms of the probability distribution of the heights throughout the surface [29,30]. The standard deviation of heights (STD) was assessed by using the following equation: where N is the number of segments (N = 100); c i , P(c i ) = h(c i )/N, and h(c i ) are central height, the probability distribution, and the histogram of the ith segment, respectively; andc is the mean value of heights.
The skewness (SKEW) reveals the level of symmetry of the height distribution in correlation to the mean value, and is estimated as the third moment of deviation: The sharpness of height distribution throughout the surface is described by Kurtosis (Kurt), the fourth moment of deviation: The entropy is the measure of randomness of the distribution of heights: Finally, the smoothness of height variation throughout the surface is defined as follow: Water wettability of the membranes was determined by measuring the water contact angles (WCA) via the static sessile drop method using an OCA 15EC goniometer (Dataphysics Company) after deposition of 4 μL dd H 2 O drops on the surface of the membranes. To study the membranes in similar conditions to those in which they are intended to be used, and to exclude the water absorption effect, the WCA were measured on fully hydrated membranes samples (hydrated up to their water uptake plateau); supplementary figure S1(A).
To evaluate the biodegradation of the membranes, dry samples (1 cm in diameter) were weighed (W0) and individually immersed in an enzymatic solution (13 mg l −1 of lysozyme and 190 U l −1 of lipase dissolved in PBS), aiming to simulate enzymes concentrations in human serum [31]. Then, the samples were incubated at 37°C and 120 rpm. After 1, 3, 11, 22, 33 and 44 days of incubation, the samples were collected, washed with dd H 2 O, dried, and weighed (W1). Biodegradation percentage was estimated as:

In vitro antibacterial test
The antibacterial properties of the membranes were characterized by evaluating the planktonic and biofilm growth inhibition of two anaerobic bacterial strains; A. actinomycetemcomitans serotype b and P. gingivalis, and two aerobic bacterial strains; E. coli and S. epidermidis. Pure cultures of each bacterial strain were collected from agar plates; for anaerobic bacteria, enriched HK agar, prepared with TSA, BHA and Yeast, added with menadione 1% v/v and hemin 1% v/v. were used, and for aerobic bacteria, TSA medium was used. Collected cultures were resuspended in culture broth (MBB added with menadione 1% v/v and hemin 1% v/v for anaerobic bacteria and TSB for aerobic bacteria) and adjusted to an optical density (OD) of 1 at λ = 600 nm (BioPhotometer D30). Bacterial suspensions of 1 × 10 6 cells/mL of each anaerobic bacteria, and 1 × 10 5 cells/mL of each aerobic bacteria were individually inoculated on sterilized membranes samples of 8 mm in diameter and placed in 48-culture well plates. Samples were incubated at 37°C in an orbital shaker (Cleaver Scientific Ltd) at 120 rpm, under anaerobic or aerobic conditions according to the bacterial strain tested. Incubation was maintained for 7 and 14 days for each anaerobic bacteria, and 1, 3, and 7 days for each aerobic bacteria. Each experiment was carried out in triplicate.
After each incubation time, the effect of the lixiviated products of the membranes to reduce the planktonic bacterial growth in the peri-membrane area, was evaluated by measuring the optical absorption (OD) of the inoculated culture media at λ = 595 nm (FilterMaxF5 multi-mode microplate reader; Molecular Devices). Percentage of planktonic growth inhibition was calculated according to the following equation: where, OD A1 = absorbance of supernatants from bacteria incubation with the PCL-G or PCL-G-Zn membranes, OD B1 = absorbance of supernatants from bacteria incubation with the PCL membranes (negative control), OD A2 = absorbance of supernatants from the tested membranes incubated with no bacteria, and OD B2 = absorbance of supernatants from PCL membranes incubated with no bacteria. The capacity of the membranes to reduce the biofilm growth was assessed by the MTT assay. After each incubation time, the membranes were rinsed once with fresh culture broth to detach loosely attached bacteria, transferred to a new culture well plate, and incubated with a 1:10 solution of MTT:culture broth for 3 h at 37°C and 120 rpm. Then, the formazan crystals metabolized by the viable bacterial adhered on the membranes were solubilized in a ISO:DMSO solution (1:1), and absorbance was read at λ = 570 nm (Filter-MaxF5 multi-mode microplate reader; Molecular Devices). Biofilm growth inhibition (%) was estimated using equation (7), but in this case, OD A1 = absorbance of solubilized-formazan from bacteria adhered on PCL-G or PCL-G-Zn membranes, OD B1 = absorbance of solubilized formazan from bacteria adhered on PCL membranes (negative control), OD A2 = absorbance from PCL-G or PCL-G-Zn membranes incubated with no bacteria, and OD B2 = absorbance from PCL membranes incubated with no bacteria.

Viability of hFOB cells in contact with the membranes
Based on the results of our previous work [27], where hFOB viability at 24 h of culture decreased below 50% compared to the control for membranes with ZnO-NPs concentrations higher than 1% w/w, in the present work, cell viability was only evaluated for the PCL-G30-1Zn and PCL-G45-1Zn membranes. Cell viability was measured after 1, 3, and 7 days of hFOB incubation in the presence of the membranes lixiviates to evaluate the effects of the lixiviates on bone cells metabolism over time.
Confluent cell cultures were treated with 0.25% v/ v trypsin-EDTA at RT, collected by centrifugation, seeded at a density of 1 × 10 4 cells/well on 24-well culture plates with DMEM supplemented with 10% v/v FBS and 3% v/v geneticin, and incubated at 37°C and 5% CO 2 during 1, 3, and 7 days. Independent cell cultures were used for each evaluation time point. At the same time, sterilized independent membrane samples (diameter = 6 mm) were incubated with culture medium to obtain the lixiviate products after 1, 3 and 7 days of incubation. Cells cultured with fresh culture media were used as control samples (Ctrl). After 1, 3 or 7 days of cell culture with the lixiviated products of the membranes, the cell viability was assessed by MTT assay using the same MTT protocol described in section 2.3. Cell viability in percentage was estimated according to:

Osteogenic microenvironment induced by the membranes
To investigate the potential of the membranes to induce an osteogenic microenvironment, bone marrow-derived mesenchymal stem cells (BM-MSC) where incubated with the lixiviate products of the PCL-G membranes with 1% w/w of ZnO-NPs. The expression of the osteoblast markers, osteocalcin and osteoprotegerin, and the calcium synthesis were evaluated after 14 days of incubation. The BM-MSC were cultured using Mesenchymal Stem Cell Basal Medium supplemented according to the manufacturer instructions. Then, confluent cell cultures were treated with trypsin-EDTA at RT, collected by centrifugation, seeded at a density of 1 × 10 4 cells/well on 24-well culture plates with DMEM supplemented with 1% v/v of antibiotic-antimycotic and 10% v/v FBS, and incubated at 37°C and 5% CO 2 . At the same time, sterilized independent membrane samples (diameter = 6 mm) were incubated with DMEM supplemented to obtain the lixiviate products every three days. For this, conditioning media from the membranes were collected, replenished with fresh medium, and then, collected media were transferred to the cell cultures, repeating the process every 3 days. Cells cultured with fresh media were used as control samples (Ctrl). After 14 days of cell culture with the lixiviated products of the membranes, culture media were replaced with fresh media with no FBS and incubated for 24 h. Then, culture media were collected and the levels of osteocalcin and osteoprotegerin were analyzed using ELISA Kits according to the manufactureŕ s instructions, and normalizing the values obtained to the number of cells in each well. The number of cells after 1 and 14 days of exposure to the lixiviate products of the membranes was estimated by the MTT assay, using a calibration curve. The initial number of cells (1 day of incubation) was also calculated, to corroborate the homogeneity of cultures before their exposure to the lixiviate products of the membranes.
Additionally, BM-MSC cells' calcium deposition was measured using alizarin red staining. The cell cultures used to obtain the supernatants for the ELISA assays were fixed with PFA (4% v/v in PBS) for 15 min at RT and washed twice with PBS (1X). Then, cultures were stained with alizarin red solution (2% v/v, pH 4.1 to 4.3) for 15 min at RT in darkness. Two independent sets of stained cultures were prepared for quantitative and qualitative analysis of calcium deposits. For quantitative analysis, after alizarin red staining, deionized water was used to wash the cultures for 3 times and the stained calcium deposits were dissolved using a solution of AcAc (20% v/v) and methanol (10% v/v), and the optical density was measured at λ = 405 nm using a Multi-Mode Microplate Reader (BioTek Synergy™ HTX). For the qualitative analysis, alizarin red stained cultures were counterstained with CAT hematoxylin (cell nuclei staining), and micrographs were acquired by optical microscopy (Axioinvert25, ZEISS).

Statistical analysis
The interferometry-based measurements were statistically analyzed using MATLAB ® . The results of the physical and chemical characterization were expressed as the mean values (ME) ± the standard deviation (SD). The biological experiments were performed in triplicated and repeated at least two times, and results were expressed as the mean values (ME) ± the standard error of the mean (SEM). All statistical analysis of data were performed using the GraphPad PISM v7.0 software. Data were analyzed using the oneway analysis of variance (ANOVA) followed by a Tukey's multiple comparison test, considering p < 0.05 as statistically significant. All the results were plotted using Origin 9.0 software.

Physical and chemical characterization of the membranes
The PCL, PCL-G, and PCL-G-Zn membranes were white in color, flexible, and soft to the touch with ≈ 0.48 ± 0.11 mm in thickness. SEM micrographs ( figure 1(A)) show that the electrospun membranes used in the present studies were constituted by randomly oriented fibers free of defects (beads), with a fiber average diameter (d) ranging from 1.143 to 0.711 μm. The addition of gelatin to the PCL membranes decreased the fiber diameter of the PCL-G membranes, and the incorporation of ZnO-NPs to the PCL-G membranes generated narrower fibers in the PCL-G-Zn membranes (supplementary figure S1). Moreover, it was observed that the fiber average diameter tended to decrease as solution conductivity increased, however, this correlation was not directly proportional (supplementary figure S2).
The Raman spectra of the membranes are shown in figure 1(B). All samples exhibited the characteristic bands associated with the main structure of PCL [32]. The main signal between 1400 and 1500 cm −1 can be ascribed to the C=O stretching (Signal 1b), the two bands between 780 and 900 cm −1 can be associated with the C-COO stretching (Signal 1c), and the two bands between 1100 and 1280 cm −1 can be attributed to C-O-C asymmetric stretching (Signal 1a). On the other hand, the multiple bands observed in the region from 900 to 1100 cm −1 can be attributed to the skeletal C-C vibrations (Signal 2), and two additional bands between 1300 and 1400 cm −1 can be attributed to the CH 2 scissoring (Signal 3). The spectra did not show any clear signal corresponding to gelatin, probably because the bands were weak, and can be easily masked by the stronger PCL bands at similar wavenumbers, plus the fact that the main and strongest bands of gelatin are expected to appear at higher wavenumbers than those measured in the present study (2500 and 4000 cm −1 ) [33]. It seemed that the presence of ZnO-NPs in the membranes did not modify the main skeletal structure of the polymeric blend, and consequently, no new bands nor modifications of the Signals 2 and 3 were observed [34]. Nevertheless, the ZnO-NPs incorporation produced a shift to lower values in Signal 1b (C=O stretching), and an increase in the relative intensity of Signal 1c (-C-COO stretching, inset in figure 1(B)), indicating that the Zn atoms somehow interacted with the polymeric fibrillar matrix to form a complex fiber-metal composite, most probably through the oxygen-rich groups.
The mechanical parameters obtained from the stress-strain curves of the membranes in fully hydrated state (supplementary figure S3) are reported in table 2. Gelatin addition to the PCL membranes (PCL-G membranes) increased, but not significantly, the elastic modulus (E) and significantly increased the elongation at break (ε) values, in relation to PCL membranes. The maximum tensile strength (σ max ) of the PCL-G membranes with the smallest gelatin concentration (PCL-G30; σ max = 1.48 MPa) increased in comparison with the PCL membranes (σ max = 1.23 MPa); nevertheless, with increasing gelatin concentration to 45% w/w the strength of the membranes decreased (PCL-G45 membranes; σ max = 1.07 MPa) in comparison to PCL-G30 and PCL; however, these σ max changes were statistically significant neither to the PCL-G30G membranes nor the PCL-G45 membranes. Upon the addition of ZnO-NPs, the stiffness and strength of the PCL-G membranes tended to decrease. In general, no trend was observed in the change of E and σ max the PCL-G-Zn membranes compared with those in the PCL-G membranes; only the σ max of the PCL-G45-6Zn significantly decreased compare with PCL and PCL-G45 membranes. Furthermore, the addition of ZnO-NPs did not significantly modify the ductility behavior of the PCL-G membranes with 1 and 3% w/w of NPs in comparison to the PCL-G membranes but, the ε of the PCL-G30-6Zn membranes (ε ≈ 202%) was significantly higher than the ε of the PCL-G30 membranes (ε ≈ 135%), and the ε of the PCL-G45-6Zn membranes (ε ≈ 92%) decreased significantly compared with that of the PCL-G30 membranes (ε ≈ 135%). Figure 2 shows the scatterplots of roughness parameters of PCL-G30 and PCL-G45 membranes with and without ZnO-NPs. STD values in figures 2(A) and (D) illustrate STD versus Kurt; STD and Kurt values decreased as the amount of NPs increased, therefore, the addition of ZnO-NPs reduced the surface roughness of PCL-G membranes, which gradually vary without sharp peaks or valleys. For a normal Gaussian height distribution, Kurt equals to 3.0, values greater than 3 indicate surfaces with sharp peaks or valleys, and values smaller than 3 indicate a gradually varying surface without sharp peaks or valleys.
The scatterplots of the SKEW versus smoothness in figures 2(B) and (E) show the symmetry of skewness values in all the samples, which reveals that the addition of ZnO-NPs neither increases nor decreases the predominance of peaks or valleys throughout the surfaces of PCL-G membranes, i.e., the height distribution of the surfaces is symmetrical. The smoothness parameter confirms the conclusion made by observing the Kurtosis behavior, that is, a decrease in surface roughness results from the addition of ZnO-NPS to the membranes. Higher Entropy values indicate a peaky surface with higher fluctuations in roughness, and a lower Entropy indicates a gradually varying surface. It was observed that the Entropy values of PCL-G membranes are very similar to the Entropy values of PCL-G membranes with ZnO-NPs (figures 2(C) and (F)), which shows the same randomness in height distribution in all the samples.
The WCA results of the membranes in fully hydrated state are shown in figure 3(A). The PCL membranes showed a hydrophobic character (WCA ≈ 112°), while the addition of gelatin significantly increased the hydrophilic character of the membranes, with WCA of 66.2°and 34.3°for the PCL-G30 and PCL-G45 membranes, respectively. Significantly lower WCA were observed in the PCL-G membranes containing ZnO-NPs, in comparison with their respective PCL-G membrane (PCL-G30 or PCL-G45). The PCL-G45-1Zn membranes shown lower WCA than the PCL-G45, however significant differences were not found. Hance, the hydrophilic character increased as the NPs concentration in the membranes increased.   The biodegradation process of the membranes was studied under simulated physiological conditions as a time-functions, the membranes samples were incubated in enzymatic solution at 37°C and orbital shaking (120 rpm), from 0 to 44 days, and the results are shown in figure 3(B). The PCL membranes showed a significantly slower biodegradation percentage after 44 days of incubation (<12%), in comparison with the PCL-G membranes that gradually degraded over time, reaching a maximum degradation of ≈27% for PCL-G30 and ≈43% for PCL-G45, after the same period of incubation. The degradation behavior of the PCL-G membranes was not significantly affected by the ZnO-NPs incorporation, only the degradation of the membranes with the highest NPs content, PCL-G30-6Zn and PCL-G45-6Zn, showed an significant increase at 33 and 44 days of incubation, reaching near 34% and 60% of degradation, respectively.

Antibacterial test
Inhibition (%) of the planktonic and the biofilm growth of the anaerobic strains tested after 7 and 14 days of incubation in presence of the membranes lixiviates are shown in figures 4(A) and (B), respectively. A decrease in the planktonic growth was observed upon anaerobic bacteria exposure to the lixiviate products of the PCL-G membranes with no ZnO-NPs (from 18% to 39% of planktonic growth); however, in the presence of the lixiviate products of the PCL-G membranes with ZnO-NPs, a significantly higher inhibitory effect was detected ( figure 4(A)). After 14 days of A. actinomycetemcomitans incubation with the lixiviate products of the PCL-G30-Zn membranes, inhibition of bacterial planktonic growth was ≈92 to 99%, depending on the ZnO-NPs concentration; on the other hand, planktonic growth inhibition was ≈89 to 97%, depending on the ZnO-NPS concentration, for incubation with the lixiviate products of the PCL-G45-Zn membranes. While the planktonic growth inhibitory effect of P. gingivalis was lower, displaying ≈57 to 71% of inhibition upon culture in the presence of the lixiviate products from the PCL-G30-Zn membranes, and ≈63 to 73% planktonic growth inhibition upon culture with the lixiviate products from the PCL-G45-Zn membranes. Similar results were observed for the biofilm growth ( figure 4(B)); the inhibitory effect on the biofilm formation was significantly higher when anaerobic bacteria were incubated on the PCL-G membranes with ZnO-NPs (reaching ≈ 87% inhibition), in comparison to the inhibitory effect upon culture on the PCL-G membranes without NPs (reaching ≈ 30% inhibition). The PCL-G30-Zn and PCL-G45-Zn membranes significantly inhibited the biofilm growth of A. actinomycetemcomitans on their surfaces, showing ≈74 to 87% inhibition after 14 days of bacteria culture on the membranes. In a similar way, the presence of ZnO-NPs in the PCL-G membranes significantly reduced the biofilm growth of P. gingivalis; the PCL-G30-Zn and PCL-G45-Zn membranes caused an inhibitory effect of ≈72 to 80%. Interestingly, neither the biofilm or the planktonic bacterial growth inhibition effect was related to the ZnO-NPs concentration in the membranes; no significant differences in the bacterial growth inhibition effect were detected among the PCL-G membranes with 1, 3, and 6% w/w of ZnO-NPs, and all the ZnO-NPs-containing membranes were comparably effective in reducing the planktonic and biofilm growth of A. actinomycetemcomitans and P. gingivalis. Nevertheless, A. actinomycetemcomitans was more sensitive to the presence of the ZnO-NPs than P. gingivalis.
Regarding the antibacterial effect of the membranes against the aerobic bacteria (figure 5), their planktonic growth was slightly reduced by the lixiviated products of the PCL-G membranes compared with the products of the PCL control membranes, while the presence of the membranes containing ZnO-NPs significantly increased the inhibition of bacterial planktonic growth for all the ZnO-NPs concentrations tested ( figure 5(A)). After 1 day of incubation, the PCL-G-Zn membranes almost completely inhibited the planktonic growth of E. coli; however, the antibacterial effect decreased after 7 days of incubation, showing ≈80% of inhibition. Conversely, the inhibitory effect against S. epidermidis planktonic growth at 1 day of incubation was ≈77%, and it increased over time reaching an inhibition of ≈92% after 7 days of incubation.
The biofilm growth of aerobic bacteria on the PCL-G membranes was also reduced in comparison with the growth observed on the PCL membranes. Nevertheless, a significant increase in the biofilm growth inhibition was observed for the PCL-G membranes with ZnO-NPs, in comparison with their corresponding PCL-G membranes, either PCL-G30 or PCL-G45 ( figure 5(B)). After 7 days of incubation on the PCL-G-Zn membranes, the inhibition of E. coli biofilm growth was ≈67% to 81% depending on the ZnO-NPs concentration, while the inhibition of S. epidermidis biofilm growth was ≈72% to 86% depending on the ZnO-NPs concentration. Similar to the results of the antibacterial tests with the anaerobic bacteria, the inhibition of planktonic and biofilm growth of the aerobic bacteria E. coli and S. epidermidis was independent of the ZnO-NPs concentration in the membranes. Although, E. coli seemed to be more sensitive to the initial exposure to the membranes with ZnO-NPs than S. epidermidis.

hFOB cell viability
The viability percentage of the hFOB cells exposed for 1, 3, and 7 days to the lixiviate products of the PCL, PCL-G, and PCL-G-1Zn membranes, is shown in figure 6. For this experiment, only the membranes containing 1% w/w of ZnO-NPs were tested, since the antibacterial results revealed adequate antibacterial effects from those membranes containing 1% w/w ZnO-NPs, and that the antibacterial effectiveness of the membranes seemed to not being correlated to their ZnO-NPs concentration, within the range of ZnO-NPs concentrations tested in the present study. At 1 day of culture, viability of hFOB cells exposed to the lixiviate products of all the membranes tested did not show significant differences in comparison to the positive control (cells cultured in fresh media). After 3 days in contact with the lixiviate products of the membranes, only the PCL-G45-1Zn membranes affected the cell viability (≈ 81%). This trend was also observed after 7 days of incubation, the hFOB cells exposed to the lixiviated products of the PCL-G45-1Zn membranes exhibited the lowest viability (≈ 78%). Even though 7 days of cell exposure to the lixiviate products of the PCL-G45-1Zn membrane significantly reduced cell viability respecting the control, it is important to note that cell viability was always higher than 78%, indicating that PCL-G-1Zn membranes did not cause a cytotoxic effect (according to the ISO 10993-5).

Osteogenic microenvironment produced by BM-MSC
To assess whether the lixiviate products of the PCL-G membranes loaded with 1% w/w of ZnO-NPs induced the differentiation of BM-MSC into an osteoblastic phenotype, the levels of osteoprotegerin, osteocalcin, and calcium deposition were measured after 14 days of BM-MSC culture in the presence of the membranes ( figure 7). All experimental cultures started with the same number of cells (10 4 cells/well), then, the cells were exposure with the lixiviate products of the membranes. After 1 and 14 days of cell contact with the lixiviate products of PCL-G and PCL-G-1Zn membranes, the number of BM-MSC cells was significantly affected in relation to the control, mainly when cells were exposed to the lixiviate products of the PCL-G45-1Zn membrane; in this case, after 1 day of exposure, the cell number decreased in 17%, while at 14 days of exposure the number of cells diminished up to 43% ( figure 7(A)). Despite this, the production of osteoprotegerin and osteocalcin was significantly higher in the BM-MSC exposed to the lixiviate products of the PCL-G30-1Zn and PCL-G45-1Zn membranes compared with the control and also compared with their respective PCL-G membrane without ZnO-NPs (PCL-G30 and PCL-G45); osteoprotegerin expression levels were 1.03, and 1.70 pg/1000 cells, respectively for PCL-G30-1Zn, and PCL-G45-1Zn ( figure 7(B)), while osteocalcin levels were 0.31, and 0.52 pg/1000 cells, respectively for PCL-G30-1Zn, and PCL-G45-1Zn ( figure 7(C)).
The quantitative and qualitative analyses of calcium deposition by the BM-MSC cells after 14 days of incubation in the presence of the lixiviate products of the PCL, PCL-G, and PCL-G-1Zn membranes are shown in figures 7(D) and (E). As shown in figure 7(D), the lixiviate products of the ZnO-NPscontaining membranes, significantly stimulated the calcium production by the BM-MSC. The amount of calcium detected in the cells cultured with the lixiviate products of the PCL-G45-1Zn membrane was considerably higher than that detected for the cells culture in the presence of the lixiviate products of the PCL-G30-1Zn membrane. In agreement with quantitative results, on the visual examination calcium deposits (dark red staining) can be observed in the BM-MSC cultures exposed to the lixiviate products of the ZnO-NPs-containing membranes, and not noticeably amount of calcium deposits in the cells cultured with fresh medium (control) or with lixiviate products of the PCL or PCL-G membranes with no ZnO-NPs ( figure 7(E)). These results suggest that lixiviate products of the PCL-G membranes with 1% of ZnO-NPs, mainly the PCL-G45-1Zn membranes, can favor the production of an osteogenic cell microenvironment.

Discussion
In this work, we investigated the antibacterial and osteogenic differentiation properties of the ZnO NPsmodified polycaprolactone-gelatin membranes. Since the end goal is to use these membranes in clinical practice, the antibacterial properties were studied using two important anaerobic periodontal Figure 6. Viability in percentage of human osteoblasts (hFOB) exposed to lixiviate products of the PCL, PCL-G, or PCL-G-1Zn membranes, estimated by the MTT assay, after 1, 3, and 7 days of incubation. * , p < 0.05 versus ctrl (cells culture with fresh media); $ , p < 0.05 versus PCL-G45 membranes without NPs; and # , p < 0.05 versus PCL-G30 membranes with 1% of ZnO-NPs. pathogens, A. actinomycetemcomitans and P. gingivalis [14][15][16], and two aerobic pathogens causing prevalent nosocomial infections, E. coli and S. epidermidis [35,36]. We also studied the ZnO-NPs-modified membranes´capability to induce a favorable environment for bone-like tissue regeneration, and their mechanical and biodegradable properties under simulated physiological conditions. This study complemented our previous work in which fibrillar membranes composed of PCL and 30 or 45% w/w of gelatin, and loaded with 1, 3, and 6% w/w of ZnO-NPs, showed physicochemical, mechanical, and biological properties adequate to be potentially used as GTR/BTR membranes [27].
Concerning the polymeric component of the membranes, gelatin is a natural polymer that has attracted attention in medical applications due to its low immunogenicity and RGD-like sequences that promote cell adhesion and migration [37]. The biomedical uses of gelatin encompass controlled drug delivery, wound dressing, and bioadhesives for closing skin wounds [38][39][40]. Despite the fast dissolution and biodegradation of gelatin in a physiological environment, its homogeneous blending with mechanically stable polymers with longer biodegradation rates, such as PCL, makes it possible to combine the intrinsic properties of both polymers and obtain blended materials with enhanced properties [41][42][43][44]. The fibrillar structure of the experimental membranes showed random fibers free of beads with a regular diameter. The addition of gelatin to the PCL-G membranes reduced the average diameter of the fibers in relation to the PCL membranes. The topography of the PCL-G-Zn membranes was affected by the increase of the number of ions and conductivity produced by the addition of the ZnO-NPs [45]. The fiber average diameter tended to decrease as solution conductivity increased, nevertheless, the correlation was not directly proportional. While the roughness parameters analysis showed that adding the ZnO-NPs into the electrospinning solutions resulted in membranes with smoother surfaces, due to a good miscibility of the solutions [46]. The higher conductivity of the ZnO-NPs electrospinning solutions can cause more elongation of the polymeric jet by the electrical forces during the electrospinning process, and this could decrease the surface roughness of the PCL-G-Zn membranes [45]. On the other hand, the Raman spectra confirmed the incorporation of the ZnO-NPs into the membranes and suggested that the interaction between the NPs and the polymeric matrix was through the Zn atoms that might interact to form a complex fiber-metal composite through the oxygenrich groups.
The mechanical parameters of the PCL-G and PCLG-Zn membranes (table 2) measured under fully hydrated conditions (simulated physiological conditions), showed higher E and ε values then those reported for three commercial collagen membranes in wet conditions (E = 10-20 MPa; ε = 25%-26%; [47]). According to the mechanical parameters obtained, the present PCL-G-Zn membranes exhibited higher flexibility than the reported commercial membranes. In general, the tensile strength values of the PCL-G-Zn membranes were lower (table 2) than those of commercially available collagen-based membranes (σ max = 1.6-4 MPa; [47]); although this should be considered when the membranes are placed in the defect site, the PCL-G-Zn membranes possess the required strength for their potential clinical use as they are not expected to be subjected to high tensile strength after being immobilized at the defect site. Furthermore, the PCL-G-Zn membranes showed higher elongation at break than the reported commercial membranes [47]. Therefore, the higher plasticity and flexibility of the PCL-G-Zn membranes could confer them better conformability to the defect site and structural integrity during the GTR/BTR treatments. This is important since during clinical GTR/BTR procedures, the membranes are in contact with physiological fluids (blood and saliva), therefore, the membranes must have adequate mechanical properties under fully hydrated conditions to maintain their structural integrity throughout the periodontal treatment, including the placement of the membranes by suturing in the defect site, and later, the possible mechanical disturbances during daily oral activities [48][49][50]. Taking this into account, the hydrophobic/hydrophilic character of the GTR/BTR membranes in hydrated conditions is also a relevant characteristic, especially because hydrophilic surfaces are expected to promote cell proliferation, and consequently enhance tissue regeneration [51,52]. In agreement with previous reports [41,53,54], the wettability of the PCL membranes in hydrated conditions, changed from hydrophobic to hydrophilic with the addition of gelatin to the membranes. This phenomenon can be mainly ascribed to the contribution of the polar nature of the gelatin molecules. The incorporation of the ZnO-NPs into the PCL-G membranes increased their wettability, probably due to the easy ionization of ZnO in aqueous media that can increase the surface energy of the membranes and consequently their interaction with water, resulting in hydrophilic membranes [55]. The hydrophilic nature and porosity of the membranes can favor their tissue integration, facilitating blood infiltration and stabilizing the blood clot, which is considered a key initiator of tissue repair [56]. Regarding this, the degradation of a biomaterial can be affected by their hydrophilic/hydrophobic character. In the present study, the degradation of the PCL-G and PCL-G-Zn membranes was significantly faster compared with the PCL membrane. The biodegradation of the PCL-G45 membranes with or without ZnO-NPs was considerably faster than that of their PCL-G30 membrane counterparts; however, in all cases the fastest degradation occurred during the first day of incubation. Although the degradation of the membranes was slightly affected by the concentration of ZnO-NPs (the higher the NPs concentration, the fastest the degradation), the gelatin component played the most important role in this regard. The partial dissolution of gelatin in aqueous medium significantly increased the weight loss of the PCL-G and PCL-G-Zn membranes compared to the PCL membranes [57]; in addition, gelatin conferred a high wettability to the PCL-G membranes, facilitating the membrane-water interactions, and in consequence, accelerating their hydrolytic degradation, whereas the low wettability of the PCL membranes slow down its degradation rate [58]. This could be useful in the GTR/BTR treatments, since the gelatin dissolved from the membranes into the defect site might enhance cell proliferation and tissue regeneration [41,59], while the PCL component in the membranes, would continue acting as barrier for longer time [60][61][62]. The biodegradation of the membranes is an important property that can determine the success of GTR/BTR treatments; a fast degradation can cause a premature loss of the membrane and in consequence the undesirable infiltration of epithelial cells into the periodontal defect, affecting its restoration [7,27,63]. The period during the membranes must remain in the defect site mainly depends on the severity of the tissue damage, in general terms, membranes are expected to function as barriers for at least 6 weeks to allow regeneration of the periodontal tissues [64]. However, in the case of most severe defects, the membranes must remain mechanically stable for 6 to 24 months after implantation [49,65].
Another important aim of this study was to analyze the antibacterial properties of the ZnO-NPs incorporated in the PCL-G membranes using clinically relevant bacteria. The results showed that even though the gelatin component in the membranes seemed to slightly reduce the planktonic and biofilm growth of the different bacteria tested compared to the PCL membrane, a significantly higher antibacterial effect was observed for the membranes loaded with ZnO-NPs. The slight antibacterial effect induced by the PCL-G membranes could be explained by the gelatin dissolution from the membranes, and the consequent presence of different peptidic hydrolysates in the lixiviates, which have been previously shown to possess antimicrobial properties [66,67]. Nevertheless, a predominant and significant antibacterial activity was only observed for the ZnO-NPs-containing membranes. Interestingly, the PCL-G membranes loaded with 1, 3, and 6% w/w of ZnO-NPs produced a significant and sustained over time antibacterial effect against all Gram-negative and Gram-positive bacteria tested, independently of the NPs concentration.
The PCL-G membranes with 1% w/w of ZnO-NPs significantly inhibited the planktonic and biofilm growth against two important anaerobic periodontal pathogens, A. actinomycetemcomitans and P. gingivalis [14][15][16], and two aerobic pathogens causing prevalent nosocomial infections, E. coli and S. epidermidis [35,36]. The Gram-negative P. gingivalis was the least sensitive to the PCL-G-Zn membranes; however, their planktonic and biofilm growth inhibition were always higher than 56% and the antibacterial effect was sustainable for 14 days. The inhibition of the planktonic bacterial growth must be directly related with the release of ZnO-NPs or Zn 2+ ions from the membranes into the broth culture media, mainly as a result of the gelatin dissolution and the membranes degradation; once bacteria are in contact with solid NPs, a mechanical disruption of cell membranes can occur [68], while Zn 2+ ions can destabilized the bacterial membrane permeability by electrostatic interactions [69]. Zn 2+ ions can internalize the bacteria membranes and bind to proteins and DNA structures, affecting the processes of DNA amplification, Zn 2+ ions can also provoke the excessive generation of reactive oxygen species (ROS), causing damage to the internal components of bacteria [70][71][72][73]. Another mechanism attributed to the antibacterial effect of ZnO-NPs is the oxidative stress generated by the production of reactive oxygen species (ROS) through photocatalysis [69,74]. Nevertheless, in this study the membranes were not illuminated by UVA radiation (ZnO-NPs energy gap value is in the UVA range); hence, the antibacterial effect must be predominantly a concomitant effect of the Zn 2+ , ZnO-NPs, and peptidic hydrolysates in the medium. On the other hand, the inhibition of the biofilm growth on the PCL-G-Zn membranes can be mainly attributed to the action of the ionized ZnO-NPs that remain embedded in the membranes, and to the peeling effect of the membranes surface attributed to the gelatin dissolution that can destabilize the solid-liquid interface between the membrane and the aqueous media affecting the bacterial adhesion [27,75]. To a lesser extent, biofilm growth inhibition might also be related to the smoother surface of the PCL-G-Zn membranes compared to the PCL and PCL-G membranes, since smoother surfaces reduce the available contact area between bacteria and the surface, and consequently decrease the initial bacterial colonization [76].
Regarding biocompatibility and osteogenic potential, the lixiviate products of the PCL-G membranes loaded with 1% w/w of ZnO-NPs allowed more than 70% of viability and proliferation of bone cells (hFOB), indicating that PCL-G-1Zn membranes did not cause a cytotoxic effect (ISO 10993-5); therefore, the membranes have the potential to allow the bone cell migration to the defect site [77]. Even more, the lixiviate products of the PCL-G45-1Zn membranes favored the osteogenic differentiation of BM-MSC as they expressed significantly higher levels of osteoprotegerin, osteocalcin [78], and calcium deposits compared with the other experimental membranes. Osteoprotegerin is a glycoprotein produced by osteoblast cells that plays an important role in bone remodeling, and osteocalcin is the most abundant noncollagenous bone matrix protein [79,80]. These results suggest the potential of the PCL-G-1Zn membranes, mainly PCL-G45-1Zn, for generating a favorable cellular microenvironment to promote bone regeneration [78,81]. This cellular behavior can be attributed to the presence of ZnO-NPs; zinc is one of the most abundant trace elements in the human body, and plays an essential role in the proper function of many macromolecules and enzymes [20][21][22]. Our study and other studies have shown that Zn ions at certain concentrations and under some environmental conditions can lead to osteogenic differentiation of stem cells [25,26].
In summary, the PCL-G45-1Zn membranes showed the best antibacterial effectiveness at biocompatible ZnO-NPs concentrations. They presented appropriate mechanical properties and biodegradation rates for GTR/BTR clinical use. These membranes are biocompatible to human gingival fibroblast [27], osteoblasts, and BM-MSC and generate a favorable cellular microenvironment for bone regeneration, allowing osteoblasts proliferation and promoting the osteoblastic differentiation of BM-MSC. Therefore, they can potentially reduce the risk of infection and stimulate the differentiation of endogenous stem cells into osteoblasts, promoting the regeneration of bone in clinical therapies.

Conclusion
This study provides valuable information for developing advanced, functionalized membranes to improve the success of GTR/BTR clinical treatments. These biocompatible membranes, with simultaneous osteoinductive and antibacterial properties, would reduce the probability of infections and increase the potential of bone tissue regeneration, enhancing the success of the GTR and BTR regeneration treatments. Furthermore, the use of non-antibiotic compounds such as ZnO-NPs, represents a promising option to overcome the world-health problem of increasing microbial resistance to antibiotics.