Scanning electron microscopy-based quantification of keratin and hyaluronic acid microstructure in electrospun scaffolds

Abstract

Background

The extracellular matrix (ECM) structural deficiencies in chronic wounds prevent the wounds from healing through natural physiological processes. Electrospun biocompatible polymers offer a platform to produce microstructure wound dressing materials that mimic the ECM containing various bioactives to address the deficiencies in the chronic wound healing process. Quantitative characterization of the electrospun fiber microstructure could provide valuable information on using fiber constructs to facilitate wound healing. This work employed a validated image analysis tool to quantitatively explain various parameters for the microstructure of six electrospun fiber constructs, D1{Polycaprolactone (PCL), Polyvinyl alcohol (PVA), Keratin}, D2{PCL, PVA, keratin, Co-enzyme Q10 (CoQ10)}, D3 (PCL, PVA, keratin, mupirocin), D4 (PCL, PVA, keratin, CoQ10, mupirocin, valsartan), D5 {PVA, Hyaluronic acid (HA)}, and D6 (PVA), using scanning electron microscopy imaging modality.

Results

The fiber intersection density (FID) parameter was quantified in the formulations, e.g., 0.272% for D5 and 0.416% for D4. Orientation histograms for D1 and D6 are characteristic of isotropic materials, while orientations for D2 and D3 indicate anisotropy with 2 preferred orientations in each formulation. D4 and D5 present orientations characteristic of transversely isotropic materials. The tortuosity for D2 and D4 indicates almost straight fiber segments, in contrast with undulated fiber segments in all other formulations. Furthermore, the mean fiber diameter was quantified, e.g., 1.414 and 1.630 mm for D3 and D4, respectively.

Conclusion

Co-electrospun PVA/PCL microfibers offer great potential for controlled delivery of bioactives needed to accelerate the healing of chronic wounds. This image-based analysis technology quantitatively characterized different formulations of electrospun fiber scaffolds. This analysis sets the stage for future study that utilizes microstructural information in finite element biomechanical modeling, to investigate possible influence of structure-based mechanical factors on the ECM restorative potential of wound dressings. Adjustment of electrospinning conditions could produce fabricated constructs like the native ECM structural components with a functional role in wound healing.

1 Background

The extracellular matrix (ECM) forms a gel-like structural framework that is found abundantly in tissues. There, it functions as a scaffold for the cells in the development of their functions and performs a crucial function in various processes, including proliferation and migration. Two key structural components of the ECM are glycosaminoglycan (GAG) hyaluronic acid (HA) and the proteoglycan Versican (Ver). The presence of HA and Ver in the skin ECM has a great impact in processes such as differentiation and wound healing [1]. Keratin is another major ECM structural component of the human skin. The effectiveness of keratin wound care technologies in clinical settings is substantiated by the capacity of topically applied keratins to stimulate keratinocytes. This stimulation leads to concurrent enhancements in the production of basement membrane proteins like collagen IV and collagen VII [2].

Based on the duration and progression of the healing process, wounds can be termed as either acute or chronic. An acute wound typically occurs due to an accident or surgical procedure, and its healing process follows a predictable timeline, typically spanning 8–12 weeks. This timeframe varies depending on factors such as the wound’s size, depth, and the extent of damage to the epidermis and dermis layers of the skin. In contrast, chronic wounds deviate from the normal healing stages and cannot be repaired in a systematic and timely manner [3]. They often develop in individuals living with conditions such as vascular insufficiency (either venous or arterial), pressure necrosis, prolonged inflammation, severe infections, cancer, and diabetes. Under these circumstances, the injured skin is unable to attain a normal healthy state, with an additional risk of infection and/or inflammation extending to an otherwise healthy surrounding skin [4]. Moreover, chronic wounds exhibit a temporal failure, making a comprehensive restoration of skin structure and function unsuccessful in such cases either.

Chronic wounds such as pressure ulcers are associated with patients with prolonged long-term immobility. The skin damage may progress to injuries affecting the muscle, tendon and bone, leading to infections (e.g., septicaemia, osteomyelitis) and, even death. Chronic wounds like leg or venous ulcers arise from venous hypertension or the malfunction of venous valves [5]. Foot ulcers and related wounds represent a significant and debilitating complication of diabetes, frequently linked to injuries, blister formation, or cut [6]. Infection of such wounds may result in severe complications, potentially necessitating amputation.

Non-healing wounds constitute favorable sites for bacteria and other pathogens, which may result in increased inflammation. The presence of inflammatory factors is a major contributory condition to the slow healing of chronic wounds. Inflammatory factors cause the immune system to recruit more macrophages and neutrophils to the wound bed and subsequently release inflammatory cytokines leading to mass production of metalloproteinases. The presence of metalloproteinases in the wound micro environment ultimately leads to the disorganization of the wound healing process [7].

Appropriate care and treatment of chronic wounds involves a clear understanding of the key role ECM structural components play in normal healing, recognition of how this ECM role is altered or becomes non-functional in chronic wounds, and the inclusion of appropriate clinical treatments in the wound dressing material to reestablish ECM structural components with functional role in healing. This study seeks to employ a previously described, custom, matrix laboratory (MATLAB), and image-based analysis tool [8] for quantifying microstructure of ECM components with great impact in wound care, such as keratin and HA, in electrospun fiber constructs (scaffolds), as a surrogate marker of the ECM restorative potential in wound dressings. The validated image analysis tool was employed to quantitatively explain various parameters for the microstructure of six fabricated electrospun fiber constructs D1–D6 using scanning electron microscopy (SEM) imaging modality.

2 Methods

2.1 Materials

The following materials and solvents were used: Mupirocin (5,9-Anhydro-2,3,4,8-tetradeoxy-8-[[3-(2-hydroxy-1-methylpropyl)oxiranyl]methyl]-3-methyl-[2E,8[2S,3S(1S,2S)]]-L-talonon-2-enonic acid 8-carboxyoctyl ester, BRL 4910A, Pseudomonic acid. Chemical abstracts service (CAS) number 12650-69-0 Sigma–Aldrich St. Louis, USA); mupirocin (Shaanxi Top Pharmaceutical Chemical Co. Limited, Shaanxi, China); Valsartan (Analytical Standard N-(1-Oxopentyl)-N-[2′-(2H-tetrazol-5-yl)[1,1′-biphenyl]-4-yl]methyl]L-valine, Sigma-Aldrich, St. Louis, USA); Valsartan Pharmaceutical Grade (CAS Number 137862-53-4 Shanghai Macklin Biochemical Co., Ltd, China); polyvinyl alcohol (PVA) (Degree of hydrolysis DH = 95, Degree of polymerization = 1700. Hebei Addtie Biotechnology Co., Limited China); Keratin and polycaprolactone (PCL) (Well-green Technology Co., Limited, Shaanxi, China); Acetic acid (Merck KGaA, Darmstadt, Germany); Co-enzyme Q10 (CoQ10) (Ubiquinone 50 CAS number 303-98-0 Sigma–Aldrich St. Louis) USA; Milli-Q water was used in all the tests (Millipore, USA). Analytical grade was used for all other chemical reagents and solvents. The latter were employed without additional purification.

2.2 Design and fabrication of D1–D6

Each of the six electrospinning solutions was prepared with a 10% w/v PVA concentration in deionized water/acetic acid solvent system (1:4), using a modification of previous method [9]. The blend was moved into a 200 mL conical flask and stirred using a magnetic stirrer at 300 rpm at a temperature of 80 °C for 4 h. PCL (2%) was added with continuous stirring, followed by the addition of Keratin (0.05%) (Table 1). Following a 30-min interval, mupirocin, valsartan, and/or CoQ10 (Table 1, D1–D4), as per the specifications for each scaffold formulation, were dissolved in a 2 mL solution of deionized water and acetic acid, and then added to the PVA-PCL mixture. The blend was agitated at a speed of 300 rpm for 1 h to ensure complete blending and subsequently electrospun. D5, containing HA in PVA (no addition of PCL, Keratin/CoQ10/mupirocin, and valsartan) was electrospun under same conditions as the D1–D4 (Table 1, D5). D6 was obtained by electrospinning only PVA (Table 1, D6). Homogeneity was ensured by keeping the solutions for 24 h. Subsequently, the solutions underwent electrospinning, conducted following process optimization under specific constant ambient conditions (temperature: 25 °C ± 1.56 °C and relative humidity: 40%). The electrospinning solution for each D series (Table 1) was filled into a 50 mL plastic syringe possessing an internal diameter of 0.8 mm and a length of 20 mm, equipped with a gauge 19G needle. The solutions were administered at a rate of 4.0 mL/h via gravitational force. The collector was positioned 15 cm away from the tip of the needle. Applying a voltage of 25 kV at the needle resulted in the collection of electrospun fiber mats. D1–D6 were stored in hermetic containers, and periodic samples were extracted for analysis as needed.

Table 1 Formulations of electrospun fiber constructs, D1–D6

2.3 Morphological characterization of D1–D6

2.3.1 SEM

The surface morphology of D1–D6 was ascertained using a Phenom ProX SEM from Phenom World Eindhoven, operating at an accelerating voltage of 15 kV. D1–D6 were sliced into small segments measuring 5 mm × 5 mm. Before examination, each sample was subjected to sputter coating with a thin gold layer using a quorum-Q150R plus E. The samples, cut into 5 mm × 5 mm pieces, were subsequently affixed to the studs using carbon tape and examined at 15 kV. The analysis was conducted in triplicate for each sample.

2.3.2 Image-based analysis

This was carried out by repurposing of a MATLAB analysis tool based on images [8], which was utilized to characterize various parameters related to the microstructure of collagen and elastin fibers in human aortic tissue samples based on images obtained through multiphoton microscopy. Here, an alternative microscopy imaging technique, SEM, was employed to quantify the microstructure of D1–D6. Briefly, the SEM micrographs were converted into tag image file format (TIFF) images and used for the analysis. In the analysis, the algorithm begins by assigning a ‘fiber skeleton’ that is one pixel thick to the image. The points where the fibers intersect are identified and marked as intersections (cross-links). The calculation of fiber intersection density (FID) in percentage involves determining the number of intersections as a proportion of the entire image. Mean fiber diameter is derived by extending the fiber skeleton, which computes the diameter of individual fiber segments in pixels and the value was converted to mm. Tortuosity, which takes values greater than unity, characterizes the undulation of each fiber segment between intersections by determining the ratio of the curve length of the fiber segment to the distance between its ends (intersections). Every fiber segment is sampled at intervals equivalent to its diameter in pixels spanning its extent. The orientation in degrees (°) of the subsegment is established by the tangent line angle connecting its initial and final points (horizontal, 90°; vertical upward, 180°; vertical downward, 0°).

The mean values of the fiber orientation, tortuosity, and diameter for all D1–D6 were presented as mean ± standard deviation (SD). Statistical differences between the mean values of the characterization parameters for D1–D6 were determined by one-way analysis of variance (ANOVA) followed by Bonferroni post hoc test using the GraphPad® Prism software (GraphPad Software, La Jolla, CA). A p-value < 0.05 was considered significant.

3 Results

TIFF images, obtained from the SEM images of D1–D6 (Fig. 1, Column I) were analyzed by the custom MATLAB image-based analysis tool to obtain fiber skeletons (blue lines) with intersection points (white circles) (Fig. 1, Column II), respectively. The quantitative results of the image analysis in terms of FID, orientation, tortuosity, and diameter are presented in Table 2, where orientation, tortuosity and diameter are expressed as mean ± SD.

Fig. 1

SEM images (column I) and fiber skeleton with intersections (column II) of electrospun fiber constructs, D1–D6. D1 (PCL—polycaprolactone, PVA—polyvinyl alcohol, Keratin), D2 (PCL, PVA, keratin, CoQ10—Co-enzyme Q10), D3 (PCL, PVA, keratin, mupirocin), D4 (PCL, PVA, keratin, CoQ10, mupirocin, valsartan), D5 (PVA, HA—hyaluronic acid), D6 (PVA)

Table 2 Quantitative results of the image-based analysis

3.1 FID

FID is a microstructural parameter that causes an increase in the effective stiffness of the electrospun scaffold. FID was quantified in all of D1–D6. For example, it was found to be 0.272% and 0.416% in D5 and D4, respectively.

3.2 Fiber orientation of D1–D6

The image-based analysis of D1–D6 generated fiber orientation histograms as shown in Fig. 2. The orientation was measured in (°), where 90° corresponded to horizontal orientation, 180° corresponded to vertical upward orientation, and 0° corresponded to vertical downward orientation. Also, the orientation as expressed in mean ± SD for each of D1–D6, respectively, is presented in Table 2.

Fig. 2

Fiber orientation (°) of electrospun fiber constructs, D1–D6: Horizontal, 90°; Vertical upward, 180°; Vertical downward, 0°. D1 (PCL—polycaprolactone, PVA—polyvinyl alcohol, keratin), D2 (PCL, PVA, keratin, CoQ10—Co-enzyme Q10), D3 (PCL, PVA, keratin, mupirocin), D4 (PCL, PVA, keratin, CoQ10, mupirocin, valsartan), D5 (PVA, HA—hyaluronic acid), D6 (PVA)

The orientation of fibers of D1–D6 was determined using a distinctive skeleton-based method. The discretization of the image during fiber skeleton generation results in a discrete orientation histogram. D1 and D6 demonstrate almost uniform distribution of fiber orientations across the whole range of angles between 0 and 180°, with an exception for D6 having some preferred orientation at around 35°. D2 and D3 present with 2 preferred orientations, approximately 30° and 120° for D2, and 60° and 150° for D3, respectively. D4 and D5 were observed to have preferred orientations toward one direction, horizontal (90°) for D4 and vertical (0° or 160–180°) for D5, respectively. The fiber orientation of D4 (p < 0.01) and D5 (p < 0.0001) was significantly different from the other formulations.

3.3 Fiber tortuosity of D1–D6

Tortuosity is a microstructural parameter that describes the undulation of fiber segments of the electrospun scaffold when the parameter is greater than 1. The tortuosity of D1–D6 was calculated using the SEM image-based analysis in the format of histograms as shown in Fig. 3. Table 2 also presents the tortuosity as mean ± (SD) for each of D1–D6 as afore-mentioned.

Fig. 3

Fiber tortuosity of electrospun fiber constructs, D1–D6. D1 (PCL—polycaprolactone, PVA—polyvinyl alcohol, keratin), D2 (PCL, PVA, keratin, CoQ10—Co-enzyme Q10), D3 (PCL, PVA, keratin, mupirocin), D4 (PCL, PVA, keratin, CoQ10, mupirocin, valsartan), D5 (PVA, HA—hyaluronic acid), D6 (PVA)

3.4 Fiber diameter of D1–D6

An additional microstructural parameter derived using this custom MATLAB image-based analysis tool was the diameter (measured in pixels) of the fibers of D1–D6. The result was calculated in the form of diameter histograms as displayed in Fig. 4. Also, the diameter as expressed in mean ± SD for each of D1–D6, respectively, is presented in Table 2.

Fig. 4

Fiber diameter (pixels) of electrospun fiber constructs, D1–D6. D1 (PCL—polycaprolactone, PVA—polyvinyl alcohol, keratin), D2 (PCL, PVA, keratin, CoQ10—Co-enzyme Q10), D3 (PCL, PVA, keratin, mupirocin), D4 (PCL, PVA, keratin, CoQ10, mupirocin, valsartan), D5 (PVA, HA—hyaluronic acid), D6 (PVA)

The results of diameter analysis show that D3 produces distribution of diameters toward the value of 5 pixels, while D4 produces distribution of diameters toward the value of 6 pixels. Evidently, the results reported in Table 2 demonstrate that the mean value of diameter in D3 is 1.414 mm, while the mean value of diameter in D4 is 1.630 mm. By visual inspection of the fiber segments of the structures of D3 and D4 in Fig. 1, it was observed that this result is expected. In between D3 and D4, the D2, D5, D1, and D6 demonstrate diameters with mean values of 1.493 mm, 1.565 mm, 1.604 mm, and 1.605 mm, respectively, as displayed in Table 2. D1 and D4 had a comparable thickness as the D6 while the thickness of D2 (p < 0.0001), D3 (p < 0.0001), and D5 (p < 0.001) is significantly lesser than D6.

4 Discussion

Polymers, such as PVA and PCL are degradable and biocompatible materials. In combination with bioactive substances, they can be considered very good candidates to produce electrospun scaffolds for faster wound healing. The enhancement of PVA’s degradability occurs via hydrolysis due to the hydroxyl groups attached to the carbon atoms. Additionally, PVA exhibits water solubility and possesses a hydrophilic characteristic. PCL is hydrophobic with excellent film forming properties. When used in combination with PVA, the mechanical strength of the resultant fibers is enhanced, and nonstick properties are also imparted on the scaffold. Furthermore, the channels formed by the rapid dissolution of PVA allow water to penetrate the matrix, thereby promoting biodegradation process of PCL while making incorporated bioactive additives available for the wound healing process in a controlled manner. In this study, HA and keratin (Table 1) were incorporated into the electrospinning mix to produce the PVA-PCL scaffold as bioactives. Keratin promotes cell adhesion leading to improved functionality [2, 10]. HA is a major structural component of the ECM, which holds its significance in numerous processes, including proliferation and migration, during wound healing [11].

The ability of electrospun scaffolds to promote wound healing lies in their potential to mimic the native ECM of the healing tissue. The custom MATLAB image-based analysis tool used in this study gave quantitative values for keratin (D1–D4) and HA (D5) microstructure, expressed in finite values as FID, mean fiber diameter, tortuosity, and fiber orientation. Thus, in designing constructs for wound healing, determination of these values made for the native ECM, can serve as a guide to the ECM mimicking capability of the scaffolds.

FID can influence the effective stiffness of D1–D6, thereby affecting the fiber restorative potential in wound dressings in concert with the stiffness of the natural ECM of the healing tissue. Connecting fibers to each other can increase the stiffness of the fiber ensemble by altering the recruitment of fibers to lower mechanical strains (reduced mean) and compelling the engagement of fibers to happen more abruptly (reduced variance) [12] based on the mechanical environment to which the healing tissue is exposed. The ability of electrospun wound dressings to mimic the stiffness of the natural ECM would be an additional advantage in this regard.

Tissues and organs vary in stiffness as influenced by cellular stiffness, the stiffness of the ECM, as well as the incompressibility of body fluids. This stiffness has been noted to impact diverse cellular functions. For instance, cells can migrate from elastic substrates that are soft to those that are stiff, but they are unable to migrate in the opposite direction. This phenomenon is referred to as durotaxis [13]. Furthermore, the stiffness of the substrate can affect the differentiation of stem cells. Mesenchymal stem cells (MSCs) were cultured on a gel with three distinct substrates, and the cells demonstrated diverse differentiation behaviors on these elastic substrates [13]. On the soft elastic substrates, MSCs primarily underwent neuronal differentiation, while on the medium elastic substrates, they mainly differentiated into myoblasts. Furthermore, on the stiff elastic substrates, the cells mainly underwent osteoblastic differentiation, with no neuronal or myoblastic differentiation observed. In modeling electrospun fibers for improved specificity, fabrication can be guided by analysis of this nature to ensure more efficient wound repair.

In chronic wounds, a major challenge is delayed or faulty angiogenesis [14], and scaffolds deployed to manage the chronic wound must be fabricated to have a level of stiffness that will enhance both cell migration and differentiation for optimal wound repair. The results of orientation analysis showed varied results for the D1–D6 used in this study. The orientation histograms for D1 and D6 are characteristic of isotropic materials, with a small transverse isotropy attributed to D6. The orientation results for D1 and D6 are expected, as they are in qualitative agreement with the structures of D1 and D6 that are depicted in Fig. 1. The orientation histograms for D2 and D3 are characteristic of anisotropic materials, and are expected, as they are in qualitative agreement with the structures of D2 and D3 that are depicted in Fig. 1. D4 and D5 displayed orientation histograms that are characteristic of transversely isotropic materials, and are expected, being in qualitative agreement with the structures of D4 and D5 that are depicted in Fig. 1.

Fiber orientation can influence cell growth and differentiation. In a study by [10], cells were cultured on aligned and random fibrous scaffolds. The cells on aligned scaffolds exhibited an elongated shape and produced more and better organized collagen-I and GAG, but not collagen-II, compared to the cells seeded on random scaffolds. Studies need to be carried out on electrospun scaffolds to determine any possible effects of fiber orientation on the functionality of the scaffold, this in turn will guard the collection method during electrospinning.

Nevertheless, relying solely on orientation-based information is insufficient for adequately modeling the biomechanical response of material composed of electrospun fibers but could be used to predict the restorative potential of electrospun fibrous constructs in wound dressings [15]. In a structure-based approach, additional parameters related to the material properties of the individual fibrous constituents would be required. The ensemble of quantitative microstructural parameters produced by this custom MATLAB image-based analysis can be very useful in this regard.

One definition of tortuosity is the association between the preferred fluid flow path within the scaffold and the curvature of its porous structure. This suggests that tortuosity can impact the permeability of the structure in relation to its porosity [11]. The tortuosity analysis values show that D2 and D4 demonstrate tortuosity near the value of 1. Tortuosity histograms for D2 and D4 indicate that the fiber segments of the corresponding D2 and D4 are almost straight. This result is expected if we proceed with qualitative comparison between the histograms and the structures of D2 and D4 that are depicted in Fig. 1. All other formulations present with tortuosity augmenting beyond the magnitude of 1.2. In particular, D3 and D5 can have maximum tortuosity in the order of 1.45 and 1.65, respectively. However, D1 and D6 can contain fiber segments with tortuosity up to the order of 2.7 and 2, respectively. It was observed that the results of tortuosity histograms for D1, D3, D5, and D6 are expected, being in qualitative agreement with the structures of D1, D3, D5, and D6 depicted in Fig. 1, respectively.

Tortuosity can be applied to adjust parameters of probability density functions that characterize the gradual engagement of fibers concerning mechanical strains [16]. Tortuosity assessments can be employed in the preliminary screening of scaffold designs to obtain an initial feasibility evaluation, so that electrospun fiber formulations with unsuitable behavior for the desired application can be discarded prior to carrying out more time-consuming analysis [11]. The provision of nutrients is primarily governed by the transport characteristics of the construct, which are, in turn, contingent upon the porosity, tortuosity, and surface chemistry of the tissue construct. To model constructs with improved transport properties, electrospun scaffolds of higher tortuosity are required. This could be achieved by manipulations in the polymer blend. The design and creation of scaffolds with customized properties are therefore essential stages in facilitating tissue growth within their host environment [11].

Scaffolds with smaller diameter have been associated with enhanced cell attachment, a phenomenon that is linked to an increase in specific surface area. In a study by [17], cell attachment was not promoted in a serum-free medium, regardless of the fiber morphology or diameter, signaling the importance of serum adhesion proteins in cell adhesion to the electrospun PCL fibers used in the study. Higher specific surface area attributed to smaller-diameter fibers, enables increased absorption of adhesion proteins onto them, leading to an increase in the number of focal adhesion points accessible for cell attachment. This phenomenon, was however, only observed to expedite the initial kinetics of cell attachment on scaffolds with smaller fiber diameters, thus at saturation, the overall count of cells adhering to the scaffold might not exhibit a significant difference [17,18,19].

The custom MATLAB image-based analysis tool is presently demonstrated to be efficient in characterizing SEM images depicting the microstructure in electrospun fiber scaffolds. In the future, the tool could be employed for the quantification of microstructural parameters in both electrospun wound dressings and natural ECM of chronic wounds, having the potential to describe elements with diverse levels of tortuosity. The ultimate objective would be to create a model for the fiber microstructure of D1–D6 employing finite element techniques. Acquired microstructural data in this study (FID, orientation, tortuosity) can be integrated with the distribution of fiber diameters, which could then be utilized to define the section properties of individual finite elements, representing the fiber skeleton framework. The capabilities of the custom MATLAB image-based analysis tool could also be extended to accommodate larger fiber diameters and images with diverse ranges of fiber diameters.

5 Conclusion

In this study, a specialized image analysis tool developed in MATLAB was utilized in quantifying microstructure of co-electrospun PVA/PCL D1–D6 imaged using SEM. Co-electrospun PVA/PCL microfibers offer great potential for controlled delivery of bioactive substances needed to accelerate the healing of chronic wounds. This study demonstrates that it is possible to extract microstructural parameters of D1–D6 to better control the fabrication ability of electrospun wound dressings to mimic the natural ECM, which is an additional advantage in the healing process. Collectively, it is hoped that a biomechanical modeling simulation tool based on input microstructural parameters could be developed for enabling the prediction of ECM restorative potential in wound dressings. Such a tool could improve risk assessment for time-wise failure of chronic wounds and direct therapy for patients.