Electrospun Nanofibrous Membranes for Guided Bone Regeneration: Fabrication, Characterization, and Biocompatibility Evaluation—Toward Smart 2D Biomaterials

Featured Application

This research focuses on the development of electrospun nanofibrous membranes incorporated with bioactive nanoparticles for guided bone regeneration (GBR). The findings indicate their potential for application in bone tissue engineering, particularly in dental and orthopedic surgery.

Abstract

Electrospun nanofibrous membranes have gained considerable attention in bone tissue engineering due to their ability to mimic the extracellular matrix and provide a suitable environment for cell attachment and proliferation. This study investigates the fabrication, characterization, and biocompatibility of poly(L-lactic acid) (PLA)-based membranes enhanced with periclase (MgO) and gold nanoparticles (AuNPs). The membranes were fabricated using an optimized electrospinning process and subsequently characterized using scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDS), Fourier-transform infrared spectroscopy (FT-IR), and contact angle measurements. Additionally, in vitro biodegradation studies in simulated body fluid (SBF) and cytocompatibility tests with osteoblast-like cells were conducted. The results demonstrated that the incorporation of MgO and AuNPs significantly influenced the structural and chemical properties of the membranes, improving their wettability and bioactivity. SEM imaging confirmed uniform fiber morphology with well-distributed nanoparticles. FT-IR spectroscopy indicated successful integration of bioactive components into the PLA matrix. Cytocompatibility assays showed that modified membranes promoted higher osteoblast adhesion and proliferation compared to pristine PLA membranes. Furthermore, biodegradation studies revealed a controlled degradation rate suitable for guided bone regeneration applications. These findings suggest that electrospun PLA membranes enriched with MgO and AuNPs present a promising biomaterial for GBR applications, offering improved bioactivity, mechanical stability, and biocompatibility.

1. Introduction

Guided bone regeneration (GBR) represents a well-established and evolving strategy for the treatment of alveolar ridge defects and critical-sized bone injuries in both dental and orthopedic contexts [1,2,3,4,5,6,7]. This approach utilizes barrier membranes to isolate the bone defect, guiding bone regeneration while preventing soft tissue infiltration. Various materials have been developed for this purpose, including natural polymers like collagen and synthetic polymers such as polylactic acid (PLA) and polycaprolactone (PCL) [2,3,4,8,9]. Recent advancements include biofunctionalization techniques and 3D printing technologies that improve the regenerative performance of membranes [1,5,6,7,10].

Clinical trials, in vivo studies, and retrospective analyses have confirmed the effectiveness of GBR with different materials, often enhanced with bioactive agents or cellular factors [4,5,6,7,11,12,13]. Among these, PLA has gained significant attention for its biodegradability, favorable mechanical properties, and processability [14,15,16,17,18]. However, pristine PLA lacks sufficient hydrophilicity and intrinsic bioactivity, and it generates acidic degradation byproducts, which may impair bone regeneration [18,19,20,21,22]. To address these limitations, composites incorporating nanomaterials like magnesium oxide (MgO) and gold nanoparticles (AuNPs) have been developed [8,9,10,11,12,13,23,24]. MgO has demonstrated osteoinductive and antibacterial properties, while AuNPs promote angiogenesis, osteoblast adhesion, and matrix mineralization [10,11,12,13,23,24]. Both types of nanoparticles serve to buffer acidic degradation products and support osteogenesis. Meanwhile, bone’s hierarchical structure and mineral composition guide the design of biomimetic scaffolds [25,26,27,28]. Materials such as bioceramics, calcium phosphates, and collagen–mineral composites have also been shown to enhance cellular responses in bone regeneration models [29,30,31,32].

Electrospinning technology enables the production of fibrous PLA membranes that mimic the extracellular matrix and support nutrient diffusion and cell proliferation [14,15,16,17,33]. Functionalization via electrospinning with bioactive agents like MgO, AuNPs, and calcium phosphates enhances both biological and mechanical characteristics [13,16,17,34,35]. Recent work has also explored novel fiber structures, including multilayered membranes and memory-shape materials that adapt to defect contours [19,36,37].

Numerous studies have explored membrane bioactivity and mechanical stability using metrics such as contact angle, swelling index, degradation rate, and cytotoxicity assays [18,19,20,21,38,39,40]. In this context, our study proposes the fabrication and evaluation of PLA membranes modified with MgO and AuNPs using electrospinning. We anticipate that the simultaneous incorporation of MgO and AuNPs will positively influence surface and structural properties, including porosity and support for osteoblast activity. [18,20,21,38,39,40,41,42]. Characterization techniques such as SEM, FT-IR, and EDS are employed to analyze morphology and chemical composition, while porosity and density measurements provide insight into structural behavior during degradation [28,30,31,43,44]. Performance is benchmarked against contemporary materials used in bone tissue engineering and regenerative medicine [45,46,47,48,49]. This study also integrates findings from the scaffold mineralization, collagen interaction, and mechanical reinforcement literature to optimize membrane functionality [34,35,47,48,49,50,51,52,53,54,55]. Moreover, recent developments in biomaterials research emphasize the relevance of machine learning (ML) to predict nanoparticle toxicity, optimize material design, and improve clinical translation [53,56,57,58,59,60]. These computational approaches can support the evaluation of PLA-based membranes and expand their applicability across diverse regenerative contexts. promising antibacterial and bioactive effects of PLA membranes reinforced with MgO and AuNPs, indicating thermal stability and shape-memory behavior, which are relevant for guided bone regeneration (GBR) applications.

Prior experimental studies have demonstrated vant for GBR [54,55]. These reports form the foundation for the current investigation.

In this study, we propose a dual-functionalization approach using MgO and AuNPs to create electrospun PLA membranes with enhanced regenerative properties for GBR applications. Our null hypothesis is that the incorporation of these nanoparticles does not significantly affect key functional parameters compared to pristine PLA. To test this, we formulate the following objectives: (1) to fabricate and characterize PLA-based membranes modified with varying concentrations of MgO and AuNPs, (2) to evaluate their surface properties (e.g., wettability via static contact angle), (3) to assess their degradation behavior in simulated body fluid (mass loss %), and (4) to determine their biocompatibility and osteogenic potential, particularly by quantifying osteoblast-driven calcium deposition via Alizarin Red S (ARS) staining. Our working hypothesis is that dual-functionalization will enhance calcium deposition, slow degradation in physiological conditions, and improve membrane wettability, thereby increasing clinical suitability for GBR.

2. Materials and Methods

2.1. Materials

All chemicals and reagents used were of analytical grade. Linear PLA (poly(L-lactic acid), average Mw 120–150 kDa) was purchased from GoodFellow (Cambridge, UK). HAuCl₄ was sourced from Sigma-Aldrich (Poznań, Poland). Phosphate-buffered saline (PBS), dimethylformamide (DMF), chloroform, sodium citrate, cetylpyridinium chloride, paraformaldehyde, dioxane, acetone, ethanol (96%), and cell culture reagents (Dulbecco’s Modified Eagle Medium, fetal bovine serum, trypsin/EDTA, antibiotic/antimycotic) were acquired from Gibco (Thermo Fisher Scientific, Waltham, MA, USA). The MG-63 osteosarcoma cell line was purchased from Sigma Aldrich (Poznań, Poland), from the European Collection of Authenticated Cell Cultures (ECACC). Alizarin Red S dye was purchased from PolAura (Dywity, Poland). Magnesium oxide (MgO) nanoparticles gold nanoparticles (AuNPs) were synthesized in-house in the way described in the Section 2.2. Magnesium hydroxide was purchased from Warchem, Zakręt, Poland.

2.2. Methods

To obtain the nanofibrous matrix, a 150 mL flask was prepared, followed by the addition of 15 g of PLA and 15 g of dioxane, which was then supplemented with acetone. The prepared solution was placed on a magnetic stirrer with temperature control until a homogeneous solution was obtained. The solution was subsequently drawn into a 5 mL syringe (needle: 1.2 × 40 mm) and electrospun at a flow rate of 80 µL/min. To prepare the gold nanoparticle solution, 50 mL of water was heated to 90 °C, followed by the addition of 0.5 mL of sodium citrate solution (2.2 g sodium citrate/50 mL) and subsequently 0.5 mL of HAuCl₄ solution (50 mg HAuCl₄/10 mL). The solution was placed on a heated magnetic stirrer for one hour. In subsequent samples, the composition was modified by incorporating periclase and gold nanoparticles, and the flow rate was increased to 150 µL/min. Successfully obtained samples are listed in

Table 1. Samples description.

Sample	Composition			Flow Rate, [µL/min]
1	5 g PLA			150
2	5 g PLA	0.05 g MgO		80
3	5 g PLA	0.25 g MgO		80
4	5 g PLA	0.25 g MgO		80
5	5 g PLA	0.5 mL nano Au		80
6	5 g PLA	1 mL nano Au		80
7	5 g PLA	2 mL nano Au		80
8	5 g PLA	0.05 g MgO	0.5 mL nano Au	80
9	5 g PLA	0.25 g MgO	1 mL nano Au	80
10	5 g PLA	0.5 g MgO	2 mL nano Au	80

. Magnesium oxide (MgO) was synthesized from reagent-grade Mg(OH)₂ by thermal dehydration above 300 °C. Calcination was conducted at 500–1200 °C (4 h per sample). To prevent rapid dehydration and powder ejection, ~10–15 g of Mg(OH)₂ was heated in quartz crucibles with a ≤5 mm layer thickness.

MgO obtained below 900 °C was highly reactive, while above 900 °C, it exhibited a periclase-like crystalline structure, enhancing chemical resistance and suitability for refractory materials. X-ray diffraction confirmed increased crystallinity with temperature, with the highest peak intensities observed above 1000 °C, indicating well-formed periclase-phase MgO.

Electrospinning was performed using a custom, self-built setup consisting of a high-voltage power supply, a syringe pump, and a rotating drum collector housed in a temperature- and humidity-controlled chamber. Temperature and relative humidity were digitally monitored. The flow rate (1 mL/h) was controlled via the syringe pump and was selected based on prior studies reporting optimal uniform fiber formation [41]. The electrospinning setup consisted of a high-voltage power supply, a syringe pump, a metal needle (21G), and a grounded rotating drum collector covered with aluminum foil. The electrospinning process was conducted under the following optimized conditions: applied voltage: 18–22 kV; flow Rate: 0.8–1.2 mL/h; needle-to-collector distance: 15 cm; and rotating drum speed: 300 rpm. The electrospinning was carried out at room temperature (25 °C) with 50% relative humidity, and the nanofibers were collected for 6 h to obtain membranes with a sufficient thickness for further characterization. After electrospinning, the membranes were vacuum-dried for 24 h to remove residual solvents. The resulting membranes were cut into circular samples (diameter: 15 mm) for subsequent characterization, degradation studies, and biological evaluation.

2.2.1. Morphological Analysis

The surface morphology and fiber diameter of the electrospun membranes were analyzed using Apreo 2 S LoVac Scanning Electron Microscope (SEM/EDS) (Thermo Fisher Scientific, Waltham, MA, USA). Prior to imaging, the samples were sputter-coated with a thin layer of gold to enhance conductivity. SEM images were captured at magnifications ranging from 500× to 10,000×, and fiber diameter distributions were analyzed Fiji (Fiji Is Just ImageJ/ImageJ 1.54f) software (National Institutes of Health, Bethesda, MD, USA). To further verify the elemental composition, Energy-Dispersive X-ray Spectroscopy (EDS) was conducted on selected SEM images. The presence and distribution of Mg and Au elements in the polymer matrix were mapped using EDS spectra.

2.2.2. Chemical Composition and Functional Group Analysis

Fourier Transform Infrared Spectroscopy (FT-IR) (Nicolett Toledo, Waltham, MA, USA) was performed to confirm the incorporation of MgO and AuNPs in the PLA matrix. FT-IR spectra were recorded in the range of 4000–500 cm⁻¹, and peaks corresponding to PLA ester bonds, MgO metal-oxygen stretching, and AuNP-associated interactions were analyzed.

2.2.3. Wettability

The hydrophilicity of the membranes was assessed by measuring the static water contact angle using the Angle Meter app. A droplet of deionized water (5 µL) was placed on the membrane surface, and the contact angle was recorded after 10 s.

2.2.4. Porosity and Swelling Behavior

The porosity and density of the membranes was evaluated using the isopropanol displacement method (Equations (1) and (2)), where samples were immersed in isopropanol, and their volume change was recorded. Swelling behavior was analyzed by immersing membranes in phosphate-buffered saline (PBS, pH 7.4) at 37 °C and monitoring their weight changes at specific time intervals. These properties are critical for assessing membrane permeability and the ability to support nutrient diffusion in GBR applications.

$$Porosity (p)={\displaystyle \frac{V_1-V_3}{V_2-V_3}}\times 100\%$$

(1)

• V₁ = Volume of liquid before immersion of the sample.
• V₂ = Volume of liquid after immersion of the sample.
• V₃ = Volume of residual liquid after removing the sample.

$$Density (d)={\displaystyle \frac{W}{V_2-V_3}}$$

(2)

where

• W = weight, g;
• d = density [g/cm³];
• V₂ = volume of isopropanol with the sample [cm³];
• V₃ = volume of isopropanol after sample removal [cm³].

2.2.5. In Vitro Biodegradation Studies

The degradation rate of the membranes was examined by immersing the samples in simulated body fluid (SBF, pH 7.4, 37 °C) for different time intervals (1, 7, 14, and 21 days). At each time point, the membranes were removed, dried, and weighed to calculate the percentage of mass loss over time (Equation (3)). Additionally, changes in pH were monitored using a pH meter to evaluate the effect of PLA degradation on the local microenvironment. Since PLA degradation typically leads to acidification, the incorporation of MgO was hypothesized to buffer this effect, stabilizing pH levels over time.

$$Biodegradation \left(BD\right)={\displaystyle \frac{W_0-W_t }{W_0}}\times 100\%$$

(3)

where

• W₀ = initial sample weight [g];
• W_t = sample weight at a specific time [g].

2.2.6. In Vitro Cytotoxicity Evaluation

To determine the biocompatibility and osteogenic potential of the electrospun membranes, a series of in vitro biological assays were conducted using osteoblast-like cells (MG-63 cell line). Osteoblast-like cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin (100 U/mL, 100 µg/mL), and 1% L-glutamine. Cells were incubated at 37 °C, 5% CO₂, and 95% humidity, with medium changes every 48 h. Before cell seeding, membranes were sterilized using UV irradiation (30 min per side) and rinsed with phosphate-buffered saline (PBS). Cells were seeded at a density of 1 × 10⁵ cells/cm² onto the membrane surfaces and incubated for 2 days. Cytotoxicity was evaluated using the XTT assay, which measures mitochondrial metabolic activity as an indicator of cell viability. After 48 h, XTT reagent was added to the wells and incubated for 3 h at 37 °C. Absorbance was measured at 450 nm using a microplate reader, and cell viability was expressed as a percentage relative to control (tissue culture polystyrene, TCPS).

2.2.7. Parametric Predicative Modeling

The following research employs a random forest regression model, an ensemble learning method that combines multiple decision trees to create a robust and accurate prediction system. The choice of the random forest was motivated by several key factors: its ability to handle non-linear relationships, robust performance with smaller datasets, and inherent feature importance analysis capabilities. The random forest regression model used in this study was implemented using the Scikit-learn 1.4.2 library. After rigorous testing, we established an optimal configuration utilizing 200 decision trees, with each tree maintaining a minimum of 2 samples per leaf and requiring at least 5 samples for node splitting. This configuration strikes a balance between model complexity and generalization ability.

The model optimization was performed using GridSearchCV 1.4.2, a grid search method with 5-fold cross-validation to tune hyperparameters. The parameters optimized include

• n_estimators: [50, 100, 200] (number of trees in the forest);
• min_samples_split: [2, 5] (minimum number of samples required to split a node);
• min_samples_leaf: [1, 2] (minimum number of samples required in a leaf node).

The best combination of parameters was selected based on negative mean squared error (MSE) as the scoring metric. No Bayesian optimization tools like Optuna were used; instead, GridSearchCV systematically evaluated all parameter combinations.

2.2.8. Bioactivity Study (Alizarin Red S Staining)

To assess the mineralization potential of the electrospun membranes, Alizarin Red S (ARS) staining was performed to detect calcium-rich deposits formed by MG-63 osteoblast-like cells. Membrane samples were first sterilized using UV irradiation (30 min per side) and placed into 24-well plates. MG-63 cells were seeded onto the membranes at a density of 1 × 10⁵ cells/membrane and allowed to adhere for 24 h in standard culture medium.

After cell attachment, the medium was replaced with osteogenic differentiation medium, consisting of DMEM supplemented with 10% fetal bovine serum (FBS), 10 mM β-glycerophosphate, 50 µg/mL ascorbic acid, and 100 nM dexamethasone. The medium was refreshed every 2–3 days. On day 21, samples were washed with PBS and fixed with 4% paraformaldehyde for 20 min at room temperature. Fixed samples were incubated with 2% (w/v) Alizarin Red S solution (pH 4.2) for 30 min to stain calcium deposits. Excess dye was removed by washing with distilled water until the background was clear. Stained samples were photographed under a bright-field microscope for qualitative assessment. For quantitative analysis, bound dye was solubilized using 10% cetylpyridinium chloride (Sigma-Aldrich, Poznań, Poland) for 60 min at room temperature. The resulting solution was transferred to a 96-well plate, and absorbance was measured at 540 nm using a microplate reader (Rayto RT6900, DRG MedTek, Warszawa, Poland).

The general idea of the research pathways, biomaterial composition, and application is given in Figure 1 and Figure 2.

3. Results and Discussion

3.1. SEM Analysis

Scanning electron microscopy (SEM) combined with energy-dispersive X-ray spectroscopy (EDS) was employed to examine the surface morphology and elemental composition of electrospun PLA membranes with varied concentrations of MgO and AuNPs (Figure 3, Figure 4, Figure 5, Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11 and Figure 12). The analysis focused on fiber diameter, nanoparticle distribution, and surface topography to evaluate how different modifications influence scaffold architecture and potential bioactivity in guided bone regeneration applications. The pristine PLA membrane (sample 1,

Table 2. Elemental composition of sample 1 (pristine PLA).

Element	Atomic %	Atomic % Error	Weight %	Weight % Error	Net Counts
C	42.9	0.2	36.1	0.2	117,741
O	57.1	0.4	63.9	0.4	72,433

) displayed smooth, bead-free nanofibers with a uniform diameter averaging 450 ± 50 nm, reflecting an optimized electrospinning process. This morphology is typical of well-controlled PLA electrospinning, consistent with previous studies demonstrating smooth fiber surfaces with minimal topographical interference [15,16]. EDS spectra confirmed the exclusive presence of carbon and oxygen, validating the chemical purity of the unmodified PLA. This sample served as the control benchmark for assessing the morphological and compositional changes induced by nanoparticle incorporation.

The introduction of a small amount of MgO (sample 2: 0.05 g) resulted in a slight increase in fiber diameter (~470 ± 40 nm) and subtle surface roughening. EDS spectra showed detectable magnesium peaks, confirming MgO integration within the PLA matrix

Table 3. Elemental composition of sample 2.

Element	Atomic %	Atomic % Error	Weight %	Weight % Error	Net Counts
Mg	3.5	0.2	6.0	0.3	2703
C	55.0	0.3	46.9	0.2	59,347
O	41.5	0.4	47.2	0.5	20,084

). Increasing the MgO content further (sample 3: 0.25 g,

Table 4. Elemental composition of sample 3.

Element	Atomic %	Atomic % Error	Weight %	Weight % Error	Net Counts
Mg	6.0	0.3	10.4	0.5	1871
C	59.2	0.4	50.3	0.3	24,317
O	34.8	0.6	39.4	0.7	6480

) led to more pronounced surface texture, with bulging regions and a fiber diameter of ~510 ± 45 nm. These findings are in line with Hosseini et al. [9], who reported that MgO enhances nanofiber surface area and roughness, promoting osteoblast activity.

Sample 4 (0.5 g MgO) exhibited moderately coarse fibers (~530 ± 50 nm), with visible crystallite exposure. Although slight micro-clustering was observed, EDS spectra showed intensified and well-dispersed Mg signals (

Table 5. Elemental composition of sample 4.

Element	Atomic %	Atomic % Error	Weight %	Weight % Error	Net Counts
Mg	0.9	0.1	1.6	0.2	1365
C	41.0	0.2	34.1	0.2	85,372
O	58.1	0.4	64.3	0.4	57,792

). This supports previous observations by Zhang et al. [22] that higher MgO concentrations can promote both osteoinductivity and localized pH buffering due to nanoparticle crystallinity.

Gold nanoparticle (AuNP) modification introduced different structural features. In sample 5 (0.5 mL AuNPs), the fiber diameter remained similar to that of the control (~460 ± 40 nm), but localized granules became visible, likely representing superficial AuNPs (

Table 6. Elemental composition of sample 5.

Element	Atomic %	Atomic % Error	Weight %	Weight % Error	Net Counts
Au	8.6	0.1	56.5	0.7	30,925
C	38.5	0.2	15.4	0.1	83,567
O	52.9	0.4	28.1	0.2	58,833

). Sample 6 (1 mL AuNPs) showed rougher surfaces (~470 ± 35 nm) with dispersed AuNP aggregates, which is consistent with previous reports on AuNP clustering at moderate loading levels (

Table 7. Elemental composition of sample 6.

Element	Atomic %	Atomic % Error	Weight %	Weight % Error	Net Counts
Au	10.7	0.1	62.4	0.7	30,799
C	39.0	0.2	13.9	0.1	65,940
O	50.3	0.4	23.8	0.2	43,710

) [10,11]. Sample 7 (2 mL AuNPs) demonstrated the most irregular topography and slight fiber thickening (~490 ± 45 nm), accompanied by strong EDS gold signals and visible nanoparticle clusters. These results reflect a saturation threshold, where excess AuNPs may aggregate and impair uniform surface biofunctionality (

Table 8. Elemental composition of sample 7.

Element	Atomic %	Atomic % Error	Weight %	Weight % Error	Net Counts
Au	10.0	0.1	60.7	0.7	26,501
C	39.5	0.2	14.6	0.1	61,828
O	50.4	0.4	24.8	0.2	40,453

) [12].

Dual modification with MgO and AuNPs yielded the most balanced morphological and compositional profiles. Sample 8 (0.05 g MgO + 0.5 mL AuNPs) showed uniform fibers (~500 ± 40 nm) with minimal protuberances and co-localized Mg and Au peaks, indicating effective nanoparticle incorporation without agglomeration (

Table 9. Elemental composition of sample 8.

Element	Atomic %	Atomic % Error	Weight %	Weight % Error	Net Counts
Mg	0.9	0.1	0.8	0.1	1280
Au	7.6	0.1	53.0	0.8	19,359
C	39.0	0.2	16.5	0.1	60,629
O	52.5	0.4	29.7	0.2	41,859

). Sample 9 (0.25 g MgO + 1 mL AuNPs) had smooth fibers (~510 ± 35 nm) and optimal nanoparticle distribution. EDS analysis confirmed a well-dispersed presence of both Mg and Au, correlating with the highest calcium deposition observed in biomineralization assays (

Table 10. Elemental composition of sample 9.

Element	Atomic %	Atomic % Error	Weight %	Weight % Error	Net Counts
Mg	7.5	0.1	7.0	0.1	11,087
Au	6.1	0.1	45.6	0.7	16,412
C	34.3	0.2	15.7	0.1	55,297
O	52.0	0.4	31.7	0.2	45,821

). These findings are supported by Qiao et al. [10], who reported enhanced osteoconductivity in dual-functionalized scaffolds.

Sample 10 (0.5 g MgO + 2 mL AuNPs) maintained the fiber diameter (~520 ± 40 nm) but exhibited increased heterogeneity and nanoparticle clustering (

Table 11. Elemental composition of sample 10.

Element	Atomic %	Atomic % Error	Weight %	Weight % Error	Net Counts
Mg	10.1	0.1	8.9	0.1	20,135
Au	6.7	0.1	47.8	0.6	24,555
C	32.3	0.2	14.0	0.1	67,453
O	50.8	0.3	29.3	0.2	59,616

). While bioactivity remained high, irregularity in nanoparticle distribution could lead to spatially variable cellular responses, a phenomenon also noted in studies by Yang et al. [11].

Overall, SEM and EDS analyses confirmed that MgO and AuNPs were successfully incorporated into the electrospun PLA membranes, with each component influencing specific aspects of surface morphology and elemental distribution. MgO primarily contributed to increased texture and potential pH buffering, while AuNPs enhanced surface complexity and introduced biofunctional metallic cues. These architectural modifications are essential for mimicking the extracellular matrix, as noted by other researchers [24,29], and for promoting osteoblast adhesion and mineralization. Among all formulations, sample 9 demonstrated the most favorable structural features for guided bone regeneration, representing an optimal balance between surface uniformity, nanoparticle distribution, and expected bioactivity.

3.2. FT-IR Analysis

The FT-IR spectrum analysis (Figure 13) of polylactic acid (PLA) nanofibers enables the identification of characteristic absorption bands associated with specific functional groups in the polymer matrix. A prominent band in the range of 1750–1740 cm⁻¹ corresponds to the stretching vibrations of ester bonds (-C=O), a defining structural component of PLA. Additionally, distinct bands in the regions of 1180–1080 cm⁻¹ and 1450–1350 cm⁻¹ are attributed to the stretching and deformation vibrations of the C–O–C backbone and methyl (-CH₃) groups, respectively. These features confirm the presence of the characteristic aliphatic polyester structure, consistent with the prior literature on PLA infrared spectra [13,17]. Upon incorporation of periclase (MgO), new vibrational signatures emerge in the 1000–800 cm⁻¹ region, indicating metal–oxygen (Mg–O) interactions. These peaks are attributed to symmetric and asymmetric stretching modes of Mg–O bonds, which signify successful nanoparticle dispersion within the polymer matrix. Similar spectral features have been reported by Proniewicz et al. [8], who observed analogous Mg–O bands in magnesium-based composites for bone regeneration. The presence of these signals confirms that MgO nanoparticles are retained post-electrospinning and contribute to chemical heterogeneity within the scaffold.

Moreover, the integration of MgO induced minor shifts in the PLA’s primary absorption bands, particularly a downward shift in the ester C=O band. This is indicative of potential intermolecular interactions, such as hydrogen bonding or dipolar interactions between MgO and PLA functional groups, consistent with observations made by Leonés et al. [54] in composite PLA systems.

Gold nanoparticles (AuNPs), when embedded into PLA nanofibers, introduced additional absorbance features in the 500–1500 cm⁻¹ region. While AuNPs themselves exhibit limited direct IR activity due to their metallic nature, their presence can modulate the polymer environment and influence vibrational modes indirectly. In this study, subtle modifications in the intensity of C=O and C–O peaks were observed, likely due to changes in bond polarizability or surface interactions between the gold nanoparticles and ester groups. Such effects have been attributed to localized surface plasmon resonance phenomena and nanoscale interface interactions in earlier work by Yang et al. [11] and Qiao et al. [10]. Of particular interest is the observed shift in the C=O peak toward lower wavenumbers in dual-modified samples, which suggests stronger dipole–dipole interactions or electron delocalization at the nanoparticle–polymer interface. These findings support the hypothesis that AuNPs engage in weak bonding or electrostatic interactions with the PLA matrix, influencing the structural integrity and potential bioactivity of the fibers. This is in agreement with previous studies on AuNP–polymer hybrid systems demonstrating tunable interfacial chemistry [11,12]. The collected FT-IR data provide compelling evidence for the successful integration and distribution of both MgO and AuNPs within the PLA nanofibers. More importantly, they underscore the presence of molecular-level interactions that can potentially affect the scaffold’s mechanical behavior, degradation profile, and biological performance. In the context of biomedical engineering and guided bone regeneration, such chemical insights are crucial for designing responsive and functionalized biomaterials tailored for osteogenic stimulation and in vivo stability [34,35].

3.3. Wettability Study

The wettability of the electrospun membranes was evaluated through static water contact angle measurements, offering insights into surface hydrophilicity and potential interactions with osteoblasts. All tested samples exhibited hydrophobic characteristics to varying degrees (

Table 12. Wettability.

Sample	Droplet	Angle °
1		110°
2		107°
3		125°
4		125°
5		137°
6		120°
7		138°
8		129°
9		128°
10		59°

). While PLA is sometimes described as moderately hydrophilic due to its polar ester groups, in fibrous membrane form, it often displays hydrophobic behavior due to surface roughness and limited surface energy [19]. Hydrophobic surfaces can resist moisture absorption, helping preserve structural and mechanical integrity under physiological conditions, which may be advantageous for long-term implant stability [20]. Additionally, reduced bacterial adhesion on hydrophobic surfaces has been noted as beneficial for minimizing infection risks in orthopedic applications [21].

The pure PLA membranes (sample 1) demonstrated a contact angle of 110°, confirming their high hydrophobicity. This level of surface tension can hinder initial protein adsorption and osteoblast adhesion, consistent with earlier studies showing reduced cell attachment on untreated PLA scaffolds [18,22].

To enhance hydrophilicity, MgO nanoparticles were introduced into samples 2–4. However, no significant reduction in contact angle was observed, suggesting that MgO alone did not sufficiently alter the surface chemistry or morphology to affect wettability. Despite MgO’s inherently hydrophilic nature, its effect may have been limited by partial embedding within the PLA matrix, restricting exposure at the fiber surface—a phenomenon previously reported by Zhang et al. [22].

Gold nanoparticle (AuNP) incorporation (samples 5–7) also did not yield a reduction in contact angle, with values ranging from 120° to 138°. These results align with observations by Yang et al. [11], who noted that AuNP surface effects depend on particle morphology and distribution. The metallic nature and surface energy of AuNPs can vary depending on their stabilization and interaction with surrounding matrix components.

Notably, the PLA/MgO/AuNP hybrid membranes (samples 8–10) exhibited significantly improved hydrophilicity. Sample 10 showed the lowest contact angle of 59°, indicating a marked increase in surface energy and water affinity. This dramatic change supports a synergistic interaction between MgO and AuNPs, where their combined effects modify the surface microstructure and chemistry in a more favorable manner for cell attachment. Similar synergy has been reported by Abdelaziz et al. [18], where dual-inorganic modifications enhanced surface wettability and subsequent biological responses.

Overall, the average contact angle across all samples (excluding sample 10) was 124° with a standard deviation of 10.65, reaffirming the predominantly hydrophobic nature of the unmodified and singly modified membranes. The significant reduction in contact angle in sample 10 highlights the effectiveness of dual functionalization in creating an osteoblast-compatible microenvironment. This improved wettability is expected to facilitate protein adsorption, cellular adhesion, and early-stage osteogenic differentiation, key attributes for successful guided bone regeneration [34,35].

3.4. Swelling Ability Study

To assess the water uptake ability of the membranes, swelling behavior was examined by immersing the samples in phosphate-buffered saline (PBS, pH 7.4) at 37 °C and measuring weight changes over time (

Table 13. Swelling abilities.

Sample	Swelling Degree [%]
1	233.7
2	201.8
3	237.1
4	75
5	232.7
6	86.2
7	116.8
8	46.6
9	175.8
10	177.1

). Although pure PLA typically shows minimal swelling due to its hydrophobic nature, the exceptionally high swelling degree observed in the PLA nanofiber scaffold (233.7%) can be attributed to several interrelated factors involving its structural configuration and physicochemical interactions with water.

Electrospun nanofibers possess a high porosity and a large surface-area-to-volume ratio, facilitating rapid water penetration and fluid retention. Additionally, the amorphous regions within the PLA matrix are less densely packed than crystalline regions, allowing more space for water molecules to be absorbed. These findings are consistent with reports from Leonés et al. [54], who demonstrated that electrospun PLA mats exhibit significant hydration behavior due to microstructural heterogeneity.

Moreover, hydrolytic degradation under physiological conditions contributes to the formation of hydrophilic carboxyl and hydroxyl groups, which further enhance water uptake over time. This dynamic was also discussed by Castañeda-Rodríguez et al. [21], who emphasized the role of degradation-induced polarity in boosting scaffold swelling.

While the incorporation of MgO nanoparticles was expected to increase water uptake due to MgO’s hydrophilic character, PLA/MgO membranes (samples 2–4) did not exhibit a consistent improvement in swelling. This discrepancy may result from MgO’s encapsulation within the polymer matrix, reducing its direct interaction with water molecules. Furthermore, MgO may contribute to localized structural densification that counters the porosity-related benefits.

PLA/AuNP membranes (samples 5–7) demonstrated moderate swelling behavior. Among these, sample 5 (0.5 mL AuNPs) exhibited superior swelling capacity compared to higher AuNP-loaded variants. This observation may reflect a threshold beyond which increased AuNP concentration leads to surface aggregation and diminished fluid penetration, consistent with findings from Yang et al. [11] and Farjaminejad et al. [12].

The most pronounced swelling was recorded for PLA/MgO/AuNP hybrid membranes, particularly samples 9 and 10. These formulations benefit from the combined hydrophilic effects and microstructural modifications imparted by both nanoparticles. Their ability to maintain surface wettability and absorb significant volumes of water likely supports cellular migration and nutrient diffusion, thereby improving scaffold integration with host tissue.

These results confirm that dual functionalization with MgO and AuNPs enhances scaffold swelling capacity, creating a more favorable environment for osteoblast attachment, migration, and tissue regeneration. Enhanced hydration also supports early-stage cellular responses essential for guided bone regeneration, as emphasized in previous studies [29,31].

3.5. In Vitro Biodegradation Evaluation

Biodegradation is a critical factor in guided bone regeneration (GBR) applications, as membranes must retain mechanical integrity during the healing phase and subsequently degrade to permit bone tissue infiltration. In this study, degradation behavior was assessed through two main parameters: mass loss over time and pH stability of the degradation medium (

Table 14. Incubation study.

Biodegradation
Sample	Weight Loss [%]	Initial pH	pH After 24 h	pH After 7 Days	pH After 14 Days	pH After 21 Days
1	10.8	7.4	7.4	7.49	7.60	7.38
2	11.3		7.52	7.60	7.71	7.53
3	13.3		7.56	7.73	7.87	7.54
4	17.7		7.54	7.55	7.71	7,.49
5	12.8		7.48	7.45	7.60	7.52
6	13.6		7.50	7.45	7.61	7.42
7	14.4		7.48	7.45	7.61	7.47
8	16.9		7.54	7.61	7.85	7.61
9	19		7.8	7.94	8.1	7.76
10	21.1		8.01	7.99	8.15	7.87

). Membranes were immersed in simulated body fluid (SBF, pH 7.4, 37 °C) for 1, 7, 14, and 21 days, followed by drying and weighing to evaluate their degradation profile. Pure PLA membranes (sample 1) exhibited a degradation rate of 10.8% ± 0.9, which, while lower than expected from the previous literature, still reflects the polymer’s inherent hydrolytic instability over time. This value is consistent with initial breakdown of ester bonds leading to modest mass loss.

The incorporation of MgO into the PLA matrix (samples 2–4) resulted in progressive increases in degradation: 11.3% ± 1.1 for sample 2, 13.3% ± 1.0 for sample 3, and 17.7% ± 1.3 for sample 4. These values suggest that while MgO may initially buffer degradation byproducts, at higher concentrations, it may also influence microstructural organization and porosity, ultimately accelerating fluid penetration and hydrolysis. This dual role has been described in studies where inorganic additives both stabilize local pH and alter scaffold permeability [22,34].

PLA/AuNP membranes (samples 5–7) demonstrated degradation values of 12.8% ± 0.8, 13.6% ± 1.0, and 14.4% ± 0.9, respectively. These findings confirm that AuNPs, while biologically active, do not significantly inhibit degradation. Rather, they may slightly contribute to mass loss by influencing fiber morphology and localized interactions with the aqueous medium. These results align with those of Farjaminejad et al. [12], who observed modest increases in PLA degradation with AuNP inclusion.

The PLA/MgO/AuNP hybrid membranes (samples 8–10) showed the most substantial degradation values: 16.9% ± 1.1 for sample 8, 19.0% ± 1.2 for sample 9, and 21.1% ± 1.4 for sample 10. While these values are higher than those for pure PLA, they reflect a balanced degradation profile when considering the overall scaffold performance. In these dual-modified samples, MgO provides buffering against acidic byproducts, while the inclusion of both nanoparticles modifies porosity and surface accessibility, contributing to a controlled degradation pace beneficial for GBR. Comparable synergistic effects have been demonstrated in dual-inorganic composite scaffolds designed for staged degradation and tissue remodeling [10,11,30].

To complement mass loss data, pH monitoring of the SBF medium was conducted to assess acidification trends due to PLA hydrolysis. Sample 1 showed a slight pH drop to 7.38, indicating limited acidification. In contrast, MgO-containing samples (2–4) maintained pH levels near 7.54, consistent with MgO’s role as a buffering agent. Samples with AuNPs (5–7) tracked closely with the control, showing minor pH variation. Dual-functionalized membranes (8–10) maintained the most stable pH, ranging from 7.5 to 7.87, further affirming MgO’s effectiveness in mitigating acidic degradation effects. These findings highlight that while MgO enhances pH buffering and AuNPs modulate surface chemistry, both contribute to scaffold degradation profiles in different ways. The PLA/MgO/AuNP membranes, despite showing the highest mass loss, maintain favorable degradation dynamics and pH control, reinforcing their suitability for GBR applications requiring bioactivity, stability, and eventual scaffold resorption [34,35].

3.6. Porosity and Density Determination

A porosity level of approximately 40% indicates a moderately porous scaffold architecture, which is often desirable in guided bone regeneration applications. Such porosity is sufficient to allow fluid infiltration, nutrient exchange, and cell migration while preserving mechanical integrity during the early stages of tissue healing. Numerous studies have demonstrated that porous structures in the range of 30–70% can facilitate osteoblast proliferation and neovascularization, providing an optimal microenvironment for bone regeneration [36,37]. In particular, the interconnected pore networks formed by electrospun nanofibers enable capillary flow and protein diffusion, critical for cell signaling and tissue integration [38].

The density of the studied samples, oscillating around 40 mg/mL, reflects the lightweight nature of the fabricated scaffolds (

Table 15. Porosity and density.

Sample	Porosity [%]	Density [mg/mL]
1	33.3	19
2	33.3	44.3
3	33.3	42
4	33.3	49.6
5	50.0	54.5
6	40.0	30
7	50.0	40.5
8	45.5	52.4
9	50.0	42
10	45.5	47.3

). This value is significantly lower than that of natural cortical bone, which ranges between 1.8 and 2.0 g/cm³ [39]. However, such moderate density is not inherently disadvantageous. For bioresorbable membrane applications, lower density may enhance biodegradability while minimizing mechanical mismatch with surrounding soft and regenerating tissues. Additionally, lower mass-to-volume ratios may promote gradual scaffold resorption and replacement by the native bone matrix, aligning with physiological remodeling processes [40].

The observed combination of moderate porosity and low density suggests that the scaffolds strike a functional balance between structural support and biological permissiveness. This architectural profile may contribute to improved osteointegration and cellular infiltration, particularly when combined with the biofunctional effects imparted by MgO and AuNP nanoparticles. Prior work on composite and nanostructured GBR membranes supports the notion that porosity and density are critical parameters for tuning cell–material interactions, degradation kinetics, and mechanical behavior [41,42]. Therefore, the measured values confirm that the developed membranes exhibit physical properties aligned with established design criteria for guided bone regeneration scaffolds.

3.7. In Vitro Cytotoxicity Assasement

While MgO and AuNPs individually contribute to membrane performance, their combined incorporation is expected to produce a synergistic effect, enhancing the osteogenic, antibacterial, and regenerative properties of the scaffold. The central hypothesis of this study is that integrating both nanoparticles into electrospun PLA membranes results in a multifunctional GBR scaffold with superior biomechanical performance, enhanced osteoconductivity, and improved cellular interactions. This strategy addresses the known limitations of pristine PLA membranes, including insufficient biological activity, limited osteointegration, and suboptimal surface wettability [43,44].

Biocompatibility is a fundamental requirement for GBR membranes, as these biomaterials must promote osteoblast adhesion, proliferation, and differentiation while avoiding cytotoxic responses. To evaluate cytocompatibility, osteoblast-like MG-63 cells were cultured on the modified and unmodified electrospun membranes, and viability was assessed using the XTT assay (Figure 14).

Pure PLA membranes and those containing 0.05 g MgO exhibited the highest cell viability, exceeding 100%. This result may be attributed to the highly porous nanofibrous architecture, which mimics the extracellular matrix and allows favorable cell–surface interactions. Additionally, PLA’s moderate hydrophobicity minimizes protein denaturation while maintaining a suitable interface for cellular anchorage [48]. In MgO-modified membranes, the release of Mg²⁺ ions may further stimulate integrin-mediated signaling and enhance the expression of osteogenic markers, such as alkaline phosphatase (ALP) and osteocalcin, which contribute to early differentiation and mineralization [49,50].

Similarly, the addition of low-concentration AuNPs also enhanced cell viability, which is consistent with evidence that AuNPs can modulate cellular responses by influencing cytoskeletal arrangement, focal adhesion assembly, and gene expression related to osteogenesis [45,51]. Moreover, their known antioxidant and anti-inflammatory effects can create a favorable microenvironment for osteoblast proliferation [52]. However, increasing the concentration of MgO and AuNPs resulted in a slight decrease in viability. This trend may be explained by elevated ion release or surface heterogeneity at higher nanoparticle loads, which may disturb cellular homeostasis or induce mild oxidative stress, as previously reported by [53,56].

PLA/AuNP membranes exhibited cell viability in the range of 80–90%, which, while slightly lower than MgO-containing or hybrid membranes, still falls well within the acceptable threshold for cytocompatibility. The moderate bioactivity of AuNPs has been shown to depend on particle size, surface chemistry, and dispersion, with optimal performance typically achieved at lower concentrations [12,45].

Most notably, PLA/MgO/AuNP hybrid membranes with the lowest combined concentrations demonstrated the highest viability overall, confirming the anticipated synergistic effect. The hybrid membranes appear to combine the bioactive ionic release of MgO with the surface-modifying and signaling properties of AuNPs, resulting in a conducive environment for cell attachment and proliferation. Previous studies [7,47] support this approach, showing that multicomponent nanostructured scaffolds offer superior osteogenic capacity compared to single-phase materials.

According to ISO 10993-5 guidelines [61], a material is classified as non-cytotoxic if cell viability exceeds 70% relative to the control. In this study, all experimental groups exhibited viability greater than 80%, confirming the biocompatibility of the modified membranes. These results strongly support the potential clinical application of PLA/MgO/AuNP composites as safe and effective GBR membranes. Nonetheless, no clear linear correlation was observed between increasing nanoparticle content and cellular viability, suggesting that biocompatibility is influenced by a complex interplay of factors, including nanoparticle distribution, surface roughness, porosity, and scaffold chemistry. Future research should explore mechanistic insights using gene expression profiling, real-time imaging of cell behavior, and in vivo validation to better understand the cellular responses elicited by these multifunctional membranes.

3.8. Parametric Predicative Modeling

The study by Yu et al. [56] demonstrated that feature importance analysis in random forests can reveal dominant predictors of nanoparticle cytotoxicity, such as surface chemistry and dissolution rate. Another researcher [37] showed that the Scikit-learn library, a widely recognized framework for machine learning applications in scientific research, can be utilized. Recent advancements in machine learning have demonstrated random forest regression utility in biomaterials research, particularly for predicting cytotoxicity based on material composition and physicochemical properties [38]. Noteworthy, the previous study [56,57,58,59,60] explored the use of machine learning for biomaterial analysis, emphasizing the importance of feature selection and model interpretability in predicting biological responses. The model architecture was carefully optimized through extensive cross-validation and hyperparameter tuning according to other researchers’ data [40].

In our study, SHAP analysis indicates that MgO concentration plays a primary role in influencing cell viability, reinforcing the importance of material composition in biomaterial safety assessments (Figure 14, Figure 15 and Figure 16).

Feature engineering and data processing in the model incorporate eight critical features that characterize the biomaterial composites: PLA quantity (g), which represents the base polymer content; MgO presence, a binary indicator of MgO incorporation; MgO quantity (g), the precise amount of MgO; Au presence, a binary indicator of gold nanoparticle incorporation; Au quantity (mL), the volume of gold nanoparticle solution; porosity (%), the material porosity measurement; pH, the pH level after 21 days of incubation; and mass (mg), the sample mass after 3 days. To ensure optimal model performance, a standardization procedure using StandardScaler was implemented, normalizing the numerical features to a common scale.

The model demonstrates noteworthy predictive capabilities, achieving a root mean square error (RMSE) of 8.99 (±7.22) through five-fold cross-validation. This performance metric indicates strong predictive power while acknowledging the inherent variability in biological systems. The mean absolute error of 3.51% suggests high practical utility, with predictions typically falling within an acceptable range for biomaterial development applications. Error distribution analysis reveals a minimum error of 0.31% (exceptional accuracy for some predictions), a maximum error of 7.90% (acceptable maximum deviation), a median absolute error of 3.73% (consistent with mean performance), and a standard deviation of error of 2.30% (indicating stable predictions). SHAP (SHapley Additive exPlanations) is a model-agnostic interpretability method based on cooperative game theory. It assigns each feature in a machine learning model a SHAP value, which represents its contribution to a specific prediction. By calculating the average marginal contribution of each feature across all possible combinations, SHAP provides a quantitative and consistent explanation of how input features influence the model’s output, enabling a clearer understanding and trust in complex models like random forests. Through SHAP analysis, valuable insights into the relative importance of different material properties were gained. MgO quantity emerged as a primary determinant of cell viability [%], showing a strong negative correlation with toxicity levels, suggesting that careful control of MgO content could be used to modulate material biocompatibility. Gold nanoparticle concentration demonstrated significant influence, with higher concentrations generally associated with lower cytotoxicity levels, aligning with the existing literature on the biocompatibility of gold nanoparticles. pH and mass showed complex relationships with cytotoxicity, indicating potential interaction effects that warrant further investigation. Porosity exhibited a moderate negative correlation with prediction errors (−0.313), suggesting that higher porosity materials may be more predictable in terms of their cytotoxic behavior.

While the current model demonstrates strong predictive capabilities, several limitations and opportunities for improvement exist. The dataset size, currently limited to a training set of 10 samples, constrains the model’s ability to capture more complex patterns. Extreme value prediction remains a challenge, as the model shows slightly reduced accuracy for very high cytotoxicity values, suggesting room for improvement in handling extreme cases. Additionally, the current implementation could benefit from more sophisticated uncertainty estimation methods. Future research directions include data enhancement by expanding the training dataset with more diverse samples, focusing on collecting data for extreme cytotoxicity cases, and investigating additional material properties that might influence cytotoxicity. Model improvements can be pursued by exploring advanced ensemble methods combining multiple algorithm types and incorporating genetic algorithms from the Mealpy library, as their accuracy appears promising based on initial tests. The authors of the article [41] employed evolutionary algorithm techniques, specifically genetic algorithms (GAs), for model and band selection in hyperspectral image classification. This approach demonstrates the significant potential of such algorithms in the medical and diagnostic sectors, particularly in optimizing classifiers for forensic-inspired hyperspectral data analysis. Future directions should also include the development of specialized models for different cytotoxicity ranges.

Further improvements in feature engineering should involve investigating interaction terms between features, exploring non-linear transformations of input variables, and considering time-dependent features for long-term toxicity prediction. These advancements will enhance the predictive power and robustness of the model in biomaterial development applications.

The scatter plot highlights the correlation between actual and predicted survivability percentages, with most data points closely following the ideal prediction line. However, deviations, particularly at higher survivability values, suggest potential limitations in the model’s generalization. Given the small dataset size (only 10 samples), the model likely struggles to capture complex patterns, leading to increased variance and reduced predictive accuracy in certain cases. The observed discrepancies indicate that while the random forest regression model effectively captures the overall trend, a larger and more diverse dataset would be necessary to improve prediction reliability and reduce errors in extreme cases.

The SHAP summary plot illustrates the impact of each feature on the model’s output, with MgO quantity (MgO_quant) having the strongest effect, where higher values (blue) reduce survivability, likely due to cytotoxic effects (Figure 15). Gold nanoparticle concentration (Au_quant) also significantly influences predictions, with high values (blue) generally increasing survivability. pH and mass exhibit moderate effects, suggesting their role in biocompatibility and material stability. Features such as binary presence indicators (MgO, Au), porosity, and PLA quantity contribute less, reinforcing that the amount of MgO and Au is more critical than their mere presence. These insights suggest a need to optimize MgO levels while leveraging Au to improve biomaterial performance.

3.9. Biomineralization Analysis

To quantitatively evaluate mineralization, Alizarin Red S (ARS) staining was performed after 21 days of MG-63 cell culture in osteogenic medium. Figure 17 presents the amount of ARS-bound dye normalized to membrane mass (mg/mg), which reflects the extent of extracellular calcium deposition on each scaffold surface, a hallmark of osteoblast differentiation.

Pristine PLA membranes (sample 1) exhibited the lowest mineralization value (0.00092 mg/mg), confirming their limited osteoinductive capacity due to a lack of bioactive sites and ionic stimulation. This finding is consistent with prior studies highlighting PLA’s bioinert surface and weak mineralization potential unless modified with bioactive components [22,25].

PLA membranes modified with MgO (samples 2–4) demonstrated a concentration-dependent increase in mineral deposition, with sample 3 (0.25 g MgO) showing the highest value among them (0.00155 mg/mg). Magnesium ions are well-known to stimulate osteogenic activity, primarily by upregulating alkaline phosphatase (ALP) and modulating intracellular calcium pathways [33,37]. Previous research has shown that MgO-containing materials enhance matrix mineralization and improve cell–matrix interactions [36,38].

Membranes modified with AuNPs alone (samples 5–7) exhibited relatively lower mineralization, though sample 7 (2 mL AuNPs) showed improved calcium deposition (0.00104 mg/mg) compared to PLA alone. This suggests that AuNPs exert a mild promotive effect on osteogenesis, likely through cytoskeletal modulation and protein adsorption rather than direct ionic bioactivity [40,43]. Still, their effect appears to be less potent than that of MgO, particularly in stimulating mineral accumulation over a 21-day culture period.

The most pronounced mineralization effects were observed in dual-functionalized membranes containing both MgO and AuNPs. Sample 9 (0.25 g MgO + 1 mL AuNPs) exhibited the highest ARS staining value (0.00195 mg/mg), suggesting a strong synergistic enhancement in calcium deposition. Sample 10 (0.5 g MgO + 2 mL AuNPs) also demonstrated elevated mineralization (0.00164 mg/mg). These results indicate that the dual incorporation of nanoparticles creates a bioactive interface that supports osteoblast maturation and mineral output. Similar synergistic effects have been described in previous studies on nanocomposite scaffolds, where ionic and structural cues together promote bone-like tissue formation [34,44,50].

These findings are consistent with earlier trends observed in this study, namely, improvements in wettability, swelling, buffering capacity, and cell viability, all of which contribute to the scaffold’s biofunctionality. Enhanced surface hydrophilicity promotes protein adsorption and cell adhesion, while Mg²⁺ release and localized AuNP cues facilitate intracellular signaling required for differentiation [25,29].

Taken together, the ARS staining results confirm the superior osteoinductive capacity of PLA/MgO/AuNP membranes, particularly at optimized concentrations of both nanoparticles. These hybrid nanocomposites represent a promising platform for guided bone regeneration, combining structural suitability with biochemical functionality supported by cumulative evidence across in vitro models [35,36,50].

4. Conclusions

This study investigated the fabrication and multi-parameter evaluation of electrospun PLA membranes functionalized with magnesium oxide (MgO) and gold nanoparticles (AuNPs) for potential use in guided bone regeneration (GBR). Through a combination of surface, structural, and biological analyses, dual functionalization was associated with improved scaffold performance compared to unmodified PLA. Key in vitro findings included enhanced surface wettability, increased swelling capacity, and improved buffering ability. SEM/EDS and FT-IR analyses confirmed the uniform incorporation of MgO and AuNPs into the PLA matrix. Biodegradation studies indicated that MgO helped stabilize acidic byproducts via pH buffering, while dual modification enabled a more controlled degradation profile. Biological assays demonstrated that all tested membranes maintained cell viability above 80%, with dual-modified samples showing the most favorable osteoblast adhesion and proliferation. Alizarin Red S staining revealed significantly increased calcium deposition in the MgO/AuNP co-modified membranes, particularly in the formulation containing 0.25 g MgO and 1 mL AuNPs (sample 9), suggesting a synergistic effect. Based on these statistically significant results, the null hypothesis was rejected. The study further suggests that MgO contributes to osteogenic stimulation through ionic release and pH modulation, while AuNPs may support mineralization and angiogenic activity. However, as the presented data are based exclusively on in vitro experiments, further in vivo validation is necessary to confirm the clinical relevance of these findings. Additionally, a random forest machine learning model was implemented to predict biological performance from material parameters, yielding high predictive accuracy and identifying MgO content and pH as key determinants of cytocompatibility. This data-driven approach supports the future application of machine learning in biomaterial optimization. In summary, this study provides promising in vitro evidence that PLA membranes functionalized with MgO and AuNPs may serve as an effective platform for GBR applications. Nonetheless, further in vivo testing and functionalization strategies, such as peptide conjugation or antimicrobial enhancement, are required to confirm their translational potential in dental and orthopedic regenerative medicine.