Improved Cellular Infiltration in Electrospun Fiber via Engineered Porosity

Abstract

Small pore sizes inherent to electrospun matrices can hinder efficient cellular ingrowth. To facilitate infiltration while retaining its extracellular matrix-like character, electrospinning was combined with salt leaching to produce a scaffold having deliberate, engineered delaminations. We made elegant use of a specific randomizing component of the electrospinning process, the Taylor Cone and the falling fiber beneath it, to produce a uniform, well-spread distribution of salt particles. After 3 weeks of culture, up to 4 mm of cellular infiltration was observed, along with cellular coverage of up to 70% within the delaminations. To our knowledge, this represents the first observation of extensive cellular infiltration of electrospun matrices. Infiltration appears to be driven primarily by localized proliferation rather than coordinated cellular locomotion. Cells also moved from the salt-generated porosity into the surrounding electrospun fiber matrix. Given that the details of salt deposition (amount, size, and number density) are far from optimized, the result provides a convincing illustration of the ability of mammalian cells to interact with appropriately tailored electrospun matrices. These layered structures can be precisely fabricated by varying the deposition interval and particle size conceivably to produce in vivo-like gradients in porosity such that the resulting scaffolds better resemble the desired final structure.

INTRODUCTION

Tissue engineering has emerged as a promising means of replacing damaged organs. Cells are seeded in vitro in/on a scaffold, supplied with adequate nutrients and benefit from active removal of waste products. Such scaffolds should promote appropriate cellular adhesion, proliferation, and specific levels of function. Materials used are typically biodegradable and are resorbed either after or during successful tissue regeneration. In addition, a high level of porosity is desired to enable both the efficient influx of anabolic nutrients and the outflow of catabolic wastes.

Methods such as solvent casting/particulate leaching,¹ gas forming,² emulsion lyophilization,³ and phase separation⁴ have been used to generate 3-dimensional scaffolds from natural and synthetic biodegradable polymers including collagen, poly(lactic acid) (PLA), poly(glycolic acid) (PGA), poly(lactic-co-glycolic acid) (PLGA), and poly(ε-caprolactone) (PCL). These scaffolds show consistent promise in promoting overall levels of cellular proliferation.

Another method producing such highly interconnected porous structures is electrospinning. First patented by Formhals in 1934,⁵ this technique yields a nonwoven fiber mesh that greatly resembles extracellular matrix (ECM). Typical fiber diameters range from 10 nm to 10 μm,⁶ providing a substantial surface area per unit volume, a property promoting cellular adhesion. Electrospinning allows limited control of the pore sizes found between interfiber contacts by selecting an average diameter⁷ via control of spinning parameters such as solution viscosity, distance, and voltage.

PCL is a good candidate material when scaffolds require short-term load-bearing ability owing to its relatively slow degradation rate in vivo.⁸ Scaffolds made of PLA, PGA, or PLGA exhibit shrinkage and substantial chemical degradation shortly after biological exposures involving hydrolysis.⁹ PCL, in contrast, is relatively inert; phagocytosis of PCL by macrophages and giant cells occurs only once the molecular weight of the polymer is reduced to 3000 or less by nonenzymatic bulk hydrolysis of the ester linkages.^10,11 Electrospun PCL has, logically, been selected as a scaffolding for bone^12–15 and cartilage^16–18 to support cell proliferation and ECM deposition.

Paradoxically, however, the small fiber size intrinsic to electrospinning can hinder efficient cellular infiltration. Eichhorn et al.⁷ have shown that the mean pore radius of electrospun matrices varies with fiber diameter. For example, a 100-nm fiber diameter yields a mean pore radius less than 10 nm at a relative density of 80%. The comparative size of a rounded cell—ranging from 5 to 20 μm—shows that such small pore sizes will obstruct cellular migration. For a scaffold that requires minimal cellular infiltration (e.g., a vascular graft) proliferation limited to the surface may be acceptable or even desirable. However, the thickness of human articular cartilage in the knee has been observed to range from 0.5 to 7.1 mm.¹⁹ To achieve uniform cellular proliferation throughout such thick scaffolds, both relatively large pore sizes and extended culture periods are necessary. Further, articular cartilage can be naturally anisotropic, being composed of superficial, middle, deep, and calcified zones,²⁰ which will require scaffolds possessing variable porosity and architecture.

In this study, we designed and fabricated an electrospun scaffold providing localized, controllable macroscopic porosity by combining electrospinning with the well-established technique of salt leaching.^1,21,22 Specific characteristics of the two processes form a useful synergy, producing a more uniform scaffold than would normally be expected from a salt-based technique. The result facilitates anisotropic cellular infiltration while retaining the highly porous, ECM-like nature of electrospun scaffolds.

MATERIALS AND METHODS

Electrospinning

A 12 wt.% solution of PCL (M_w 65,000, Sigma-Aldrich, St. Louis, MO) in acetone (Mallinckroff Chemicals, Phillipsburg, NJ) was prepared and electrospun at 23 kV using a high voltage DC power supply (Model FC50R2, Glassman high voltage, High Bridge, NJ) with a 20-cm tip-to-substrate distance and a flow rate of 18 mL/h onto aluminum foil wrapped on a 4- ×4-in. steel plate. Salt crystals previously sieved to sizes between 90 and 106 μm were introduced into the Taylor Cone and the falling fiber beneath it as shown in Fig. 1. The amount of NaCl in each layer was approximately 0.75 g and covered approximately 10% of the surface area of the aluminum foil. Deposition of each allotment of NaCl required 30–60 s during electrospinning; the interval between each allotment was either 5 or 10 min. Electrospinning under these conditions required approximately 90 min to make approximately 5 mm thick sheets. For specific comparisons of cellular penetration, approximately 4 mm thick sheets were spun without salt incorporation utilizing otherwise identical conditions.

FIG. 1.

Schematic illustration of an electrospinning setup showing the mechanical introduction of salt crystals (gray particles): (a) syringe pump, (b) extension tubing, (c) sheath surrounding the needle into which the crystals are added, (d) needle through which voltage is applied, (e) electrospun fiber interacting with falling NaCl crystals, and (f) grounded collector.

In specific cases, fluorescein isothiocyanate (FITC) (Sigma-Aldrich) was added (0.1 mg/mL) to the polymer–acetone solution before electrospinning to fluorescently label electrospun PCL fiber. This enabled us to observe both the electrospun fiber and the infiltrating cells in sectioned samples via fluorescent microscopy.

The deposited sheet was carefully punched with a dermal biopsy punch (Miltex, York, PA) to generate cylindrical “plugs” 3 or 6 mm in diameter and approximately 5 mm in height. The plugs were then inserted into a sealed plastic bag and this bag placed into a 45°C water bath for 10 min to achieve the partial sintering needed to prevent the extensive delamination otherwise observed after exposure to aqueous solution. The embedded salt crystals were then leached out by exposure to DI water at 37°C for 3 days; the DI water was replaced every 24 h.

Cell culture

The CFK2 cell line (obtained from Dr. Henderson’s lab, Department of Medicine, McGill University, Montreal, Canada) having the phenotypic characteristics of chondrocytes derived from fetal rat calvariae²³ was used. Approximately 650,000 cells per sample placed in a 12-well tissue culture plate (Falcon, Franklin Lakes, NJ) were gravitationally seeded onto salt-leached plugs lying on their sides in each well. The seeded cells were cultured in Ham’s F-12 medium (Mediatech, Herndon, VA) containing 10% fetal bovine serum (FBS; Mediatech, Herndon, VA), 1% penicillin–streptomycin (Fisher Scientific, Fair Lawn, NJ), and 1% L-glutamine (Mediatech, Herndon, VA). The culture medium was changed every other day. The samples were harvested at day 3 and week 3 before subsequent characterization.

SEM

Samples were coated with an 8 nm thick layer of osmium (OPC-80T, SPI Supplies, West Chester, PA). Samples emerging from cell culture were fixed with 10% formalin (Richard-Allen Scientific, Kalamazoo, MI) and exposed to a graded ethanol series in DI water (50%, 70%, 85%, 90%, and 100% ethanol) to achieve dehydration that was then finalized using a graded ethanol–hexamethyldisilazane (HMDS, Electron Microscopy Sciences, Hatfield, PA) series (25%, 50%, 75%, and 100% HMDS) followed by drying under a hood overnight. The dried sample was either coated directly with osmium or gently parted along the salt-generated gaps to observe cell migration before osmium coating.

Cryosection

Samples cultured for 3 weeks were fixed with 10% formalin followed by three 10-min rinses in phosphate-buffered saline (PBS, Mediatech, Herndon, VA). The fixed samples were embedded in OCT compound (Sakura Finetek, Torrance, CA) and then frozen at −80°C. The frozen samples were cut to 12-μm sections via a cryostat (CM3050S, Leica Microsystems, Bannockburn, IL) and placed onto Super Frost Plus glass slides (Fisher Scientific). The samples were stored at –80°C until staining.

Nuclear staining

The cryosectioned samples were rinsed 4 times with PBS to remove residual OCT. The cells were then permeablized with 0.2% TritonX-100 (Sigma-Aldrich) for 30 min followed by 3 rinses with PBS. The samples were stained with 0.01% 4′,6-diamidino-2-phenyl-indole dihydrochloride (DAPI, Sigma-Aldrich) in PBS containing 0.5% bovine serum albumin (BSA) for 5 min in the dark. The stained samples were observed via an Axioplan2 microscope (Carl Zeiss, Thornwood, NY).

Histochemical staining

The cryosectioned samples were rinsed 4 times with PBS. The samples were then stained using a 1.5% Safranin O (Sigma-Aldrich) reagent for 40 min. The Safranin O solution was filtered through a 0.45-μm pore size nitrocellulose membrane (Bio-Rad, Hercules, CA) before use. The stained samples were quickly rinsed several times in DI water and observed under a microscope.

RESULTS

The interaction between the dispensed salt crystals and the fibers during electrospinning is shown in Fig. 2a. Under normal circumstances the falling salt would deposit directly below the exit of the sheath in Fig. 1. Interaction with the Taylor Cone and the solidifying fiber beneath it carried falling salt out beyond the circumference of the sheath via spirally moving fibers, resulting in a uniform distribution within the deposited layer. Uniformity was dependent on the weight, and hence the size, of the crystals. Previous use of salt particles larger than 150 μm produced visibly nonuniform salt distributions concentrated directly below the sheath. Figure 2b shows electrospun fibers wrapped around an incorporated NaCl crystal. A normal electrospun morphology (rounded fibers) is observed in spite of the presence of the salt crystal.

FIG. 2.

(a) Image of salt particles interacting with fibers during electrospinning. (b) SEM image of a salt particle entrapped by a large group of fibers.

A low-magnification scanning electron microscopy (SEM) image of a cylindrical plug containing incorporated salt crystals is shown in Fig. 3a. A partially delaminated layered structure is evident owing to the volume of incorporated particles. Gap sizes appear to range from 100 to 200 μm. Salt distribution within each delamination is relatively uniform. The side of the plug was slightly smeared because of shear forces applied during punching. Much more shallow (less than 10 μm in depth) delaminations were also produced within each layer, and these are likely a characteristic of the electrospinning process itself because no salt was being deposited at those points during the spinning. A salt crystal entrapped by fibers marked by an arrow is shown (Fig. 3b). Each delaminated layer is held together by fibers deposited during salt incorporation. These fiber meshes were strong enough, thanks in part to the limited sintering (45°C for 10 min) that caused the formation of simple “necks” between fibers, to maintain an overall monolithic structure during salt leaching and subsequent cell culture. The structure of PCL fibrous plug after salt leaching is shown in Fig. 4a. As expected, leaching did not visibly degrade the fibers but some slight swelling of the overall structure was observed. This expansion seemed to be largely due to springback resulting from the release of compressive forces applied during punching. Figure 4b shows partially delaminated layers joined by “vertically” (in this image) oriented fibers. Salt incorporation followed by leaching increased relative as-spun porosity from 79 (salt-free) to 83.2%.

FIG. 3.

(a) Side view of an as-punched 37 mm diameter PCL plug containing salt crystals. (b) A salt particle (arrow) entrapped by large groups of fibers within a delamination.

FIG. 4.

SEM images of (a) PCL plug after NaCl leaching showing approximately 200-μm pores and (b) fibers holding two partially delaminated layers together.

The distance between each delamination can be controlled by the intervals between additions of salt crystals. A 10-min interval resulted in approximately 450 μm thick layers while 5-min intervals resulted in approximately 230 μm thick layers. The shorter (5-min) time intervals also increased the overall porosity, to 86.5%.

Figure 5 documents the various morphologies adopted by CFK2 cells seeded on the salt-leached plug at day 3. Some cells are rounded (Fig. 5a) while others display a slightly elongated morphology (Fig. 5b). ECM accumulation on the plasma membrane is apparent in Fig. 5a and b.

FIG. 5.

SEM images of the seeded cells showing either rounded (a) or slightly elongated (b) shapes on the salt leached PCL plug at day 3.

Figure 6 provides images taken from a stained plug cryosection after 3 weeks of culture. Cellular nuclei are blue and the FITC-labeled PCL fibers are green. The presence of cells along the deposition plane shows that they clearly infiltrated along the pores formed by the salt crystals. A few cells infiltrate “vertically” into the electrospun mesh (see the circle in Fig. 6a) but the majority appears to be enclosed within the cavities formed by salt dissolution. Figure 6b shows a magnified image of the cells (marked with a circle in Fig. 6a) infiltrating both horizontally and vertically. Compared to the salt-leached samples, the cellular infiltration into electrospun structure that did not contain salt generated pores was minimal over the same culture duration (Fig. 6c and d). The maximum infiltration we were able to observe was approximately 160 μm (see Fig. 6d).

FIG. 6.

Nuclear stained (DAPI; blue stains) sample cryosectioned to 12 μm in thickness after 3 weeks of culture in the samples with (a, b) and without (c, d) prior salt incorporation. (a) Cellular infiltration through salt-generated pores (×100) and (b) cellular infiltration into the PCL fiber meshes in some area (magnified image of the circled area in Fig. 6a) (×200). (c) Outer surface cellular coverage without infiltration (×100) and (d) maximum cellular penetration (approximately 160 μm) observed in a sample spun without any incorporated salt. Color images available online at www.liebertpub.com/ten.

Horizontal cross-sections of the cell-seeded plug cultured for 3 weeks were observed (Fig. 7). Even though the samples were 6 mm in diameter these cross-sections were infiltrated in the range from 35% to 70% (Fig. 7a and b). The average (from 18 layers) and standard deviation of the cell coverage was determined to be 59.5 ± 9.2% (see Table 1). Lighter colored regions (marked as regions 1 and 3 in the picture) contain cells while no cells were present in the darker colored region (marked as region 2). High-resolution images of regions 1 and 3 reveal infiltrating cells along with the presence of accumulated ECM. Cellular penetration began at the edge of the salt-generated porosity and progressed inwards (in some cases substantially).

FIG. 7.

SEM images of the delaminated layers showing cellular infiltration after 3 weeks of culture; cells are shown covering (a) approximately 70% and (b) approximately 35% of the 6-mm cross sections in the same plug; regions 1 and 3 show cellular infiltrated areas; region 2, no cellular coverage. (c) High-resolution image of region 1 and (d) high-resolution image of region 3.

Table 1.

Variation in Fiber Diameter and Cellular Coverage for Delaminated Areas Taken from Two Specimens

Layer	Sample 1		Sample 2
	Average fiber diameter (μm)	Cellular coverage (%)	Average fiber diameter (μm)	Cellular coverage (%)
1	0.71	35.2	0.72	50.7
2	0.73	50.8	0.72	64.4
3	0.73	55.3	0.73	59.3
4	0.74	69.8	0.75	68.2
5	0.76	68.8	0.74	67.4
6	0.75	64.5	0.74	70.5
7	0.74	67.7	0.74	54.1
8	0.75	58.2	0.74	58.9
9	0.75	50.7	0.75	56.6
Average	0.74	57.9 ± 11.3	0.74	61.1 ± 6.8

The average fiber diameters and cellular coverages are included along with the standard deviation of the latter.

Finally, Fig. 8 shows glycosaminoglycan (GAG) distribution in the cryosectioned sample after 3 weeks of culture using a Safranin O (red) stain. The majority of the GAG content is observed in the gaps generated by salt leaching consistent with the cell distributions shown in Figs. 6 and 7. In Fig. 8, the vertically aligned fibers that hold the delaminated layers together, and the associated adherent cells, can be clearly observed.

FIG. 8.

Safranin O–stained sample cryosectioned in 12 μm after 3-week culture shows GAG (red stains) accumulation in the NaCl crystal-generated pores. Vertical fibers that hold partially delaminated electrospun layers are also observed (×200). Color images available online at www.liebertpub.com/ten.

DISCUSSION

Electrospinning is a promising technique allowing efficient, economical production of tissue engineering scaffolds. The process produces a unique nonwoven nano- and/ or microfibrous structure that resembles natural ECM. The influence of nano- or micro-structures on cellular migration, orientation, and cytoskeletal organization has been demonstrated,^24–27 and electrospun topographies significantly enhance cellular behavior.²⁸

However, as has been pointed out, pore size exponentially decreases with fiber size.⁷ Zhang et al. reported cellular infiltration of only 48 μm using bone marrow stromal cells.²⁹ Van Lieshout et al. also showed poor penetration of human myofibroblasts into electrospun PCL compared to a knitted equivalent.³⁰ Li et al. postulate that cells on an electrospun “surface” could penetrate by enzymatic degradation of individual fibers³¹ but this mechanism is improbable for relatively resistant synthetics such as PCL. Faster “degraders” such as PGA, PLA, and gelatin can be used but at the cost of poorer initial chemical and mechanical stability.

Within this framework, it seems clear that efficient initial seeding is critical for tailored cellular ingrowth into electrospun scaffolds to produce in vivo-like cellular distributions. Dynamic depth filtration has achieved effective seeding in other fibrous scaffolds^32,33 but would not be as successful in standard electrospun matrices owing to the small pore sizes. Stankus et al.³⁴ developed a method that simultaneously introduces cells into a scaffold simultaneous with electrospinning providing concurrent deposition of cells and fibers throughout the scaffold.

The inherent appeal of marrying salt deposition with electrospinning is that it utilizes the inherent randomness of the Taylor Cone and the falling fiber beneath it to achieve uniform results that guide subsequent cell proliferation. Simple vertical additions of salt to the surface of the collector plate would be biased based on their method of introduction. The result could not be uniform unless considerable effort went into altering the normal gravity-driven trajectory of the introduced salt particles. With the current technique, a much simpler approach succeeds in providing a uniform distribution because it interacts with the elongating, whipping fiber formed by the balance between electrostatic repulsion and solution viscosity to result in a relatively random, uniform arrangement of salt particles in the as-deposited mass. Neither solvent evaporation nor the formation of nanometer-scale fiber diameters appears to be affected by the presence of the adhering salt particles, as indicated by the insignificant changes in average fiber diameter in each delaminated layer (Table 1). In other contexts involving the use of salt as a porogen a lack of control of salt placement can result in distinct gradients in porosity and pore size. An additional consequence, poor diffusion between largely disconnected porosity, is avoided here because electrospinning is characterized by highly interconnected (albeit cell-impermeable; see Fig. 6) pores.

Figure 5 shows cells in various shapes in which they exhibit different levels of mobility.^35–37 Most surprising, given the relative inefficiency of the simple gravity-based seeding employed here, was that the nearly complete (in some cases) penetration of the salt-generated gap observed (Figs. 6 and 7). Figure 8 validates the observations of both horizontal and vertical penetration while showing that the cells have apparently retained an appropriate phenotype capable of producing glycosaminoglycan. Given that the details of salt deposition (amount size, number density) are far from optimized, the result provides a clear illustration of the ability of mammalian cells to infiltrate appropriately engineered electrospun matrices.

To our knowledge, this represents the first observation of extensive cellular infiltration of electrospun matrices. Thus it is important to examine and understand the nature of infiltration. Figure 9 shows that cell clustering can take several forms. Relatively “thick” cell populations are observed at points of initial cell seeding (Fig. 9a) at the edge of the cylinders. These represent the initial gravity-driven cellular deposition into the edge of the gap (Fig. 4b) followed by proliferation leading to the observed mass of cells. Internal to the plug we see evidence for both large numbers of cells (Fig. 9b) as well as more scattered, isolated cell populations (Fig. 9c). Infiltration was highly variable; Fig. 7a shows nearly complete penetration of a delaminated cross-section and Fig. 7b shows cell populations scattered in between areas in which fiber density apparently remains high enough such that no cells are present. The origin of these disparate populations merits discussion. Distances of 3000–6000 μm were clearly spanned by these cells. Given the apparent good adhesion of the cells to the electrospun fiber, it seems likely that infiltration is primarily driven by localized proliferation rather than cellular locomotion. The latter, at roughly 100 μm/day,³⁸ would not be sufficient to allow for the cross-sectional coverage observed after only 3 weeks of culture. It would be interesting to determine how much culture time or initial porosity would be needed for proliferation to lead to physical delamination of these structures. The given method of salt incorporation used here leads to variations in porosity substantial enough to allow easy access of cells in some cases (Fig. 6a and b) while minimizing access in others (Table 1). Areas in between salt deposition appear to be no more cell permeable than the as-electrospun fiber (Fig. 6c and d).

FIG. 9.

SEM images of (a) dense cellular population near the edge of the plug, (b) dense cellular population, and (c) sparse population in the middle of the plug.

The method developed in this study provides large pores into which cells can infiltrate without first requiring fiber degradation. We showed that cells can migrate up to approximately 4 mm and can cover approximately 70% of the cross-sectional area of these 6 mm diameter plugs using simple gravitational seeding. If a dynamic seeding—like the depth-infiltration seeding developed by Li et al.³³—is applied to this scaffold infiltration would likely be even more effective. This technique clearly provides considerable flexibility in scaffold design and in promoting cellular ingrowth along specific directions. Both the amount of salt and the particle size will doubtless be important. Further, the layered structure can be precisely tailored by varying the deposition interval to produce controlled gradients in pore distribution. This gradient can lead to cellular density gradient in a scaffold that better resembles the in vivo equivalent, for example, the zonal anisotropic structure of cartilage.

CONCLUSIONS

By combining salt leaching with electrospinning, we demonstrate a method that produces a delaminated structure in an electrospun scaffold. With this method, not only can cellular infiltration into a thick, 3-dimensional electrospun scaffold be facilitated, but scaffolds having designed, anisotropic structures can also be produced to guide cellular proliferation to better approximate targeted tissues.