Controlled deposition of human neural stem cells on fiber substrates using gel encapsulation

TL;DR

This paper introduces a low-cost method to precisely seed human neural stem cells onto fiber substrates using gel encapsulation, enabling controlled cell distribution and long-term viability.

Contribution

A bioprinting-inspired device for controlled cell seeding on 3D substrates, offering a low-cost alternative to traditional bioprinting systems.

Findings

Gel-encapsulated human neural stem cells were seeded with spatial control on fiber substrates.

Cells remained viable and retained normal differentiation potential over long-term development.

The device enabled multiple isolated cell populations on a single substrate, suitable for 3D electrospun fibers.

Abstract

Electrospinning is a versatile technique for manufacturing micro-nano diameter fibers and has been used extensively for tissue engineering in vivo and advanced cell culture in vitro. Standard means of seeding cells onto such substrates typically offer no control over cell distribution, yielding dispersed, heterogeneous and low concentrations of cells. In this article, we investigate the viability of using a simple bioprinting-inspired device for seeding gel-encapsulated cells onto fiber substrates. Using human neural stem cells, we were able to consistently seed cells with spatial control. We examined their long-term development, showing viable cells and normal differentiation potential. Furthermore, this device was able to seed on multiple sites within a single substrate, creating isolated populations and demonstrating the potential for this approach as a low-cost alternative to bioprinting systems, which is also applicable to somewhat challenging 3D substrates like electrospun fibers.

Figures

Overview of seeding setup, microcapillary delivery device (MCDD) and experimental set-up.A) Seeding cells onto fiber scaffolds with a standard hand-held pipette typically produces low-density dispersed cultures, often with non-uniform distribution across a large area. By using a microcapillary to seed gel-encapsulated cells onto the substrate, it is possible to seed the same number of cells onto a much smaller area and with precise localization. B) Photograph of microcapillary and polyethylene tube, as it was set up and attached to a 3-axis micro-manipulator. C) Microscope image of the polyimide capillary, with an inner diameter of 0.1650 mm, and outer diameter of 0.3050 mm. D) Testing was done using a fluorescence dissection microscope where gel-medium-mixture was supplemented with fluorescent particles to visualize the gel. Here, four images from a video capture of the injection process is shown. The faintly visible microcapillary is outlined and the polyimide capillary has been outlined for visibility, which is retracted in the last image when injection is complete. E) A summary of the experimental plan: We used standard pipette seeding as a reference for our microcapillary delivery device, while we utilized the MCDD in two separate ways. In one case we seeded alginate-suspended cells onto electrospun fibers, while in the other case we further applied alginate lyase, which slowly digests the alginate and releases the cells from the gel. In all cases, we performed our analysis at three time points: 4, 14 and 21 days after seeding. F) Schematic of the seeding procedure. Cell suspension is loaded into the microcapillary, together with alginate solution and medium, while the fiber substrate is soaked in a small volume of medium containing 20 mM Ca2+. A small volume is injected onto the substrate, and gelation occurs to limit cell dispersion. After a few minutes, medium is added, which optionally contains alginate lyase to degrade the gel and release the cells onto the fiber substrate.

Seeding areas achieved with microcapillary injection.A) With hand-held pipette seeding, the cells typically spread unevenly over the majority of the substrate area, which in this case was 800 mm2. A to-scale version of a stitched fluorescence micrograph of DAPI-stained cells is shown on the circle, which represents the entire substrate area B) Stitched fluorescence micrograph of DAPI-stained cells, showing the distribution of cells on a fiber scaffold, with typical low-density and heterogeneous distribution of cells. C) Stitched fluorescence micrograph of DAPI-stained cells seeded by microcapillary seeding, and subsequently released with alginate lyase, showing the controlled distribution achieved through this technique. D) Stitched fluorescence micrograph of DAPI-stained cells seeded by microcapillary seeding, without alginate lyase addition, showing highly similar distribution characteristics to those treated with alginate lyase. E) Quantification of area occupied by cells seeded through microcapillary injection, and either released by alginate lyase (Alg + Aly) or not (Alginate). No significant changes are seen, which indicates that cells remain attached over time, and during treatment of alginate lyase. We also get an idea of the variability of areas; most of the groups show the 50th quantiles within a range that doesn’t exceed 5mm2. From left to right: N=(17,17,18,17,13,12), Boxes: Interquartile range; Whiskers: Min/Max. P-values estimated by t-test with Bonferroni correction.

Estimated respiration by Alamar blue assay.Alamar blue assay of cells seeded by hand-held pipette (Fiber Mesh), or with microcapillary injection, either with (Alg + Aly) or without (Alginate) alginate lyase release. Cultures seeded through MCDD seem show reduced of respiration. Its important to note that this may be due to several factors other than cell count, see results and discussion for further discussion. N=14-18, Boxes: Interquartile range; Whiskers: Min/Max. P-values estimated by t-test with Bonferroni correction.

Differentiation, thickness and density of cultures.A) Example micrograph of cells seeded by microcapillary injection and released with alginate lyase, antibody labelled for DCX and stained with DAPI, here fixed 4 days after seeding. B) Same as A, but after 14 days, showing the increase in DCX+ cells. C) 3D rendering of a stitched image stack from a culture seeded by microcapillary injection. D) Same image stack as in C, but projected along the x-axis, to show the thickness of culture. E) Quantification of fraction doublecortin positive cells, showing that doublecortin (DCX) was not significantly different between the three groups. F) Quantification of culture thickness, which on day 4 and day 21 were higher in cells seeded with microcapillary injection and not released with alginate lyase (Alginate), compared to the two other groups. Note that fibers are not visible in fluorescence microscopy. G) Quantification of cell density. Medians are always higher in microcapillary seeding examples, but only significantly so at 14 days after seeding in cultures seeded with microcapillary and not released with alginate lyase (Alginate). N=9 for each group, Boxes: Interquartile range; Whiskers: Min/Max. P-values estimated by t-test with Bonferroni correction.

Seeding of multiple cultures on a single substrate.Stitched images of DAPI expressing cells, showing distributions of cells seeded on multiple locations. Circles outline the average area of cultures seeded in this manner as single cultures.

Full text

1 Introduction

Electrospun fibers are widely used as a versatile and inexpensive material for neural tissue engineering and organ-on-a-chip models. Such fibers offer unique benefits over other popular materials for these purposes, as they can be tailored to resemble a range of target tissues in terms of surface roughness [1], porosity [2], fiber dimensions [3], fiber alignment [4], elasticity [5], or layer formation [6,7]. Probably the most common materials to use for electrospinning are synthetic polymers like PCL or PLLA [8]. These can in turn be surface modified to alter their interactions with living cells [9]. However, the range of materials that can be spun into fibers are extremely diverse, and use of biological materials like collagen [10], or even complex mixtures of decellularized tissue have been documented [11].

The use of electrospun fibers for in vitro models has been largely justified by a growing recognition of the influence that mechanical and topographic features have on cell behavior. When cells are cultured outside their native tissue, there is always a risk that they exhibit behaviors which they never do otherwise, and the ability to better replicate the micro-environment of tissues has been a topic of great focus for many decades. Naively, it may seem that as long as the right chemicals and bio-molecules are present, then cells can not distinguish between dish and tissue. As it turns out, topographical and mechanical properties also have a strong influence on cell behavior, both directly by the action of cellular mechanosensing, but also indirectly as it affects how cells distribute themselves, how they interface with each other, how nutrients and bio-molecules flow et cetera. A vast body of research documents the use of electrospun fibers as a form of artificial extracellular matrix for neurons, showing to have a strong influence on neuronal cell differentiation [12], axonal growth [13], migration [14], network formation and electrophysiological function [15]. Applications of electrospun fibers range from basic research on neuron-ECM interactions [16], to advanced in vitro disease models [17], or for production of therapeutic grafts [18] or other tissue repair applications.

In the same way that the application of electrospun fibers aims to improve the accuracy of in vitro models, appropriate cell tools such as neural stem cells can be used to derive naturalistic populations of neuronal cells in vitro [19]. Many physiological processes involve interactions between multiple cell types, and producing appropriate cell populations has been a major focus in the past decades [20]. Together with the inaccessibility of human cells and tissues, and the questionable relevance of non-human animal cells for in vitro models, this has led to a growing interest in using human neural stem cells as a cell source. Neural stem cells have been shown to be capable of self-renewal and differentiation in vitro [21]. They can be used to generate all major cell types of the brain, including a diverse range of neuronal types [22]. However, neuronal stem cells are critically sensitive to their chemical, mechanical and topographic micro-environment [23], and to their interaction with neighboring cells. Using the potential of stem-cell-derived cultures and electrospun fibers to create effective in vitro models is an active area of research, which could potentially accelerate drug discovery, elucidate disease mechanisms, and reduce dependence on animal models [24].

We have previously worked on developing novel solutions for electrospun fiber-based in vitro models [25]. When cells are seeded on fiber substrates, they typically spread out across the entire scaffold, and unless a large number of cells are used, the cell concentration will be low. In this study, we wanted to explore the ability to control cell deposition on such electrospun fibers. Furthermore we were interested in the potential to sequentially seed controlled populations of neurons, which in turn could be used as a generic tool for seeding different cell types as spatially controlled populations in order to model particular inter-cellular interactions. We hence took inspiration from bioprinting to explore these questions and constructed a microcapillary delivery device (MCDD), with which we could seed small populations of alginate suspended cells onto electrospun fiber substrates at high density.

Bioprinting is a highly diverse field in which various techniques are employed to deposit cells onto a culture substrate. Popular and conceptually simple approaches include extrusion printing and inkjet printing, where cells are often suspended in a mixture which provides stability to the printed cells, such as a gel like alginate.

More advanced bioprinting systems can print with single-cell resolution [26], but this is typically limited to small volumes as print resolution typically trades off of print speed [27]. A typical extrusion-based printing system reliably prints down to 100 μm [28]. Many of these systems, however, require specialised and expensive equipment, and are often designed to print on flat substrates. Developing a robust low-cost system which can print on more challenging surfaces, such as fiber scaffolds, would therefore be an interesting solution. Several existing protocols utilize hybrid printing techniques, for example, by combining a thermoplastic polymer extruder with a cell-alginate extruder [29], by using electrospun fibers as structural support [30], or by supplementing the cell suspension with fiber fragments [31]. However, these techniques often preclude some of the unique benefits which comes when electrospun fibers themselves serve as the cell-supporting substrate. The ability to control cell alignment and neurite growth orientation using electrospun fibers should ideally be preserved even when using hybrid approaches.

For these reasons, we used enzymatic digestion of alginate to slowly release cells from the alginate so that they would be exposed to the fiber scaffolds. We estimated basic characteristics of cells seeded in this manner, and confirm cell survival, neurite extension and normal differentiation potential. We furthermore demonstrate sequential seeding of multiple such populations of cells. We believe that this technique could be used to model how brain cell type composition, and neuronal micro architectures influences brain function. This technique could be used as an expendable bioprinter, serving as a low-cost alternative to applications where high throughput isn’t a concern. With minor modifications, we expect our protocol to be applicable to a wide range of culture substrates.

2 Materials and methods

2.1 Fiber preparation

Electrospun PCL fiber substrates (3D NanoMatrix; PCL 700) were provided by Cellevate (Cellevate AB, Sweden). These fibers have a reported diameter of 700 nm, 100 μm thickness, 15-50 μm pore pores and approximately 85% porosity (S1 Fig). Fibers were delivered as attached to plastic membranes in the bottom of standard 24 well plates. Each individual fiber mesh covered the entire bottom of each well, to a total of about 800 mm^2^. One day prior to cell seeding, fibers were coated with 0.1% polyethylenimine (PEI) (Sigma-Aldrich, USA) in borate buffer (Sigma-Aldrich, USA) for one hour, then rinsed in calcium- and magnesium-free dPBS (Sigma-Aldrich, USA), and finally in water and dried over night in room temperature. Immediately prior to seeding, fibers were immersed in 10 μg/ml laminin-2020 (Sigma-Aldrich, USA) and incubated in 37 °C for one hour.

2.2 Cell expansion

The human neural cell line used for this study was originally established by Lars Wahlberg, Åke Seiger, and colleagues at the Karolinska University Hospital, Stockholm, Sweden (original work with the cell line is described in [32], with informed verbal consent of patients. In turn, the cell line was kindly provided to us via Prof. Anders Björklund (Dept. Exp. Med. Sci., Lund University, Sweden). Ethical permission from local ethical committee at Huddinge Hospital. In brief, the cell line was established from forebrain tissue, isolated and obtained from one 7-week (post conception) human embryo.

Cells were expanded in DMEM/F12 (Invitrogen, UK), with 2.0 mM L-glutamine (Sigma-Aldrich, USA), 0.6% glucose (Sigma-Aldrich, USA), N2-supplement (Invitrogen, UK), 2.0 μg/mL heparin (Sigma-Aldrich, USA), 20 ng/ml human basic fibroblast growth factor (Invitrogen, UK), 20 ng/ml human epidermal growth factor (PROSPEC, Israel), and 10 ng/mL human leukemia inhibitory factor (PROSPEC, Israel).

Cells were passaged roughly every 14 days by dissociation with accutase (Thermo Fisher Scientific, USA) where cell count was assessed by trypan blue (Sigma-Alrich, USA) exclusion using an automated cell counter (TC20, BioRAD). After dissociation, cells were either seeded as per given experiment, or reseeded for further expansion. Passages between 14 and 16 were used in our experiments.

2.3 Microcapillary delivery device and seeding

The Microcapillary Delivery Device (MCDD) used for seeding alginate suspended cells consisted of a syringe pump (Legato 111; kdScientific), micro manipulator, a 100 μl Hamilton syringe, poly-ethylene tubing and a polyimide 165 μm tip.

The seeding apparatus was sterilised with alcohol, assembled inside a laminar flow cabinet, and further sterilised with UV-light. The syringe pump unit was kept outside the laminar flow cabinet to allow for easy control and power access at the expense of relatively long tubes that fed into the capillary inside the cabinet. The syringe, tubing and capillary was backfilled with 70% ethanol for sterilisation. Sterile water was then used to thoroughly rinse out the ethanol. Bubbles were very carefully avoided and rinsed out when necessary to avoid dampening of the syringe pumps pressure.

Just prior to MCDD seeding, the electrospun fibers were soaked in differentiation medium with 20 mM Ca^2+^ by adding roughly 300 μl to each well, and after 5 minutes removing most of this solution to leave the fibers just slightly covered in calcium-containing differentiation medium.

Single cell suspensions were obtained by passaging as described above. After determining the cell concentration, cells were centrifuged and resuspended into their respective cell concentrations for MCDD seeding and reference seeding.

A sufficient volume of cell suspension (around 50 μl; enough to fill the polyimide microcapillary) was loaded into the microcapillary by withdrawing cells through the narrow tip of the MCDD, at the slow rate of 5 μl/min. The micromanipulator was used to position the microcapillary tip very close to the electrospun fibers, and injecting 1 μl of cell suspension at a speed of 10 μl/min. After injection was complete, 20 seconds were let to pass before retracting the microcapillary and moving on to the next fiber sample. After seeding, samples were placed in 95%RH 37 °C 5% CO_2_ for 20 minutes to fully gellate. Then, 500 μl of differentiation medium was carefully added to the samples. This medium was either supplemented with 1 mg/ml Alginate Lyase (Sigma Aldrich) or not, depending on which experiment was being performed. To avoid agitation during cell release by alginate lyase, these samples were not moved to an incubator but were left in place for 20 minutes while the alginate lyase was acting. The cultures were kept in 95%RH 37 °C 5% CO_2_ and fed every 2-3 days by replacing half of the medium with fresh differentiation medium (DMEM/F-12 medium with 2.0 mM L-glutamine (Cytiva, USA), 0.6 % glucose (VWR, USA), N2-supplement, 2.0 μg/mL heparin (Sigma-Aldrich, USA), Penicillin-Streptomycin (Thermo Fisher, USA) and 1% fetal bovine serum (Invitrogen, UK)).

When seeding cells by standard pipette seeding, cells were seeded by pipetting a 10 μl droplet containing 100,000 live cells onto fiber meshes in Differentiation medium.

2.4 Alamar blue assay

Alamar blue reagent (Thermo Fisher, USA) was prepared by mixing 1:9 with differentiation medium. We replaced the medium with 400 μl of reagent in each culture well, including blank wells with no cells, and incubated all samples in 5%CO_2_ and 37°C for 4 hours. After incubation, 100 μl from each well was transferred to a 96-well fluorescence reading plate. 530 nm emission by 485 nm excitation was measured for each well using a FLUOstar OPTIMA FL microplate reader (BMG Labtech). We subtracted the mean fluorescence intensity from each sample, and then normalized all samples to the mean of the standard pipette seeding reference, such that ‘1’ now represents an identical value as the mean of the reference.

2.5 Fixation

The samples were immediately fixed after the Alamar blue assay was performed. Importantly, the cultures had to be pre-incubated with 20 mM calcium chloride to stabilise the gel which otherwise became violently dispersed by the fixative. Samples were fixed for 10 minutes by application of 4% Paraformaldehyde solution, after which the samples were rinsed with PBS.

2.6 Immunocytochemistry

Samples were pre-incubated in blocking solution (PBS, 1% BSA (Sigma-Aldrich, USA) and 0.25% Triton X-100 (Applichem, DE) for 30 minutes. All antibodies were similarly diluted in PBS, 1% BSA and 0.25% Triton X-100. GFAP-reactive rabbit anibodies (Agilent, USA) were used at a 1:2000 the original concentration, while Doublecortin-reactive mouse antibodies (Santa Cruz Biotechnology, USA) where used at 1:200 the original concentration. The samples were incubated with both primary antibodies over night.

Rabbit-IgG-reactive donkey antibodies (Abcam, UK), with a conjugate Texas Red fluorophore (614 nm) was diluted 1:200, while mouse-IgG-reactive rat antibodies (thermo Fisher, USA) with conjugate Alexa-488 was diluted 1:400. All samples were incubated with both these antibodies for one hour.

The samples were imaged in glass-bottom microscopy wells, with Vectashield (Vector Laboratories, USA), containing 4,6-diamidino-2-phenylindole (DAPI).

2.7 Image acquisition

For imaging cell seeding area, we used a 4x magnification objective and captured tiled images that covered the entire seeding area and repeated this for every sample. Images for detailed measurements of density, thickness and ratio were acquired with a Zeiss Axio Imager M2 fluorescent microscope (Zeiss, Germany). Images were acquired from three samples from each culture condition and time point. We picked representative samples which had a normal distribution of cells. The images were acquired with a 40x magnification objective, and z-stacks were captured which spanned the entire thickness of the cultures.

2.8 Image analysis

All image analysis was done using manual measurements in imageJ. We used ImageJ’s built in ability to measure area and did this by enclosing an area which roughly seemed to contain more than 90% of all cells. Density, thickness and DCX ratio was derived from 40x magnification images, where cells were manually counted throughout the entire volume. The thickness was measured as the distance between the last and first frame of the z-stack which has a substantial number of cells in focus.

2.9 Statistical analysis

To test for statistical significance, we used t-test with Bon-Ferroni correction between all groups and time points. All measurements which were from different wells were regarded as independent measurements. In cases were multiple images had to be captured to acquire sufficient number of cells for cell counting or thickness measurement, the mean of these images was used as an independent measurement. Normality was tested using a Shapiro-Wilks test, which found that almost all groups were normally distributed.

3 Results

3.1 Cell seeding using Microcapillary Delivery Device (MCDD)

In order to control the distribution of cells on electrospun fibers, we designed a microcapillary delivery device (MCDD) (Fig 1A–1C) capable of injecting small volumes of alginate-encapsulated cells. The delivery device consisted of a 165 μm inner diameter polyimide capillary (Fig 1C), which by an intermediate polyethylene tube was coupled to a precision syringe pump capable of withdrawing and injecting volumes in the nanoliter range. The polymide capillary was seated in a larger syringe which served as a simple holder and allowed us to couple it to a micromanipulator (Fig 1B).

Overview of seeding setup, microcapillary delivery device (MCDD) and experimental set-up.A) Seeding cells onto fiber scaffolds with a standard hand-held pipette typically produces low-density dispersed cultures, often with non-uniform distribution across a large area. By using a microcapillary to seed gel-encapsulated cells onto the substrate, it is possible to seed the same number of cells onto a much smaller area and with precise localization. B) Photograph of microcapillary and polyethylene tube, as it was set up and attached to a 3-axis micro-manipulator. C) Microscope image of the polyimide capillary, with an inner diameter of 0.1650 mm, and outer diameter of 0.3050 mm. D) Testing was done using a fluorescence dissection microscope where gel-medium-mixture was supplemented with fluorescent particles to visualize the gel. Here, four images from a video capture of the injection process is shown. The faintly visible microcapillary is outlined and the polyimide capillary has been outlined for visibility, which is retracted in the last image when injection is complete. E) A summary of the experimental plan: We used standard pipette seeding as a reference for our microcapillary delivery device, while we utilized the MCDD in two separate ways. In one case we seeded alginate-suspended cells onto electrospun fibers, while in the other case we further applied alginate lyase, which slowly digests the alginate and releases the cells from the gel. In all cases, we performed our analysis at three time points: 4, 14 and 21 days after seeding. F) Schematic of the seeding procedure. Cell suspension is loaded into the microcapillary, together with alginate solution and medium, while the fiber substrate is soaked in a small volume of medium containing 20 mM Ca2+. A small volume is injected onto the substrate, and gelation occurs to limit cell dispersion. After a few minutes, medium is added, which optionally contains alginate lyase to degrade the gel and release the cells onto the fiber substrate.

Alginate forms a viscous fluid in the absence of multivalent cations such as $[eqn]$ , meaning that gelation can be controlled by addition of compounds such as calcium chloride. This means that an alginate solution can remain liquid in the micro-capillary, and form a gel when injected into a medium containing calcium chloride. By using a syringe pump, the microcapillary could be loaded with alginate solution by withdrawal, and subsequently injected into a solution containing calcium chloride to encapsulate cells in alginate gel. By submerging the electrospun fiber mesh with medium containing calcium chloride, cells could be blotted onto the mesh to produce clusters of high cell density. Rather than continuously extruding alginate solution, as commonly done in other bioprinting approaches, we opted for a simpler approach where the nozzle remained stationary and only injected a single population of cells.

To facilitate testing of the microcapillary delivery device, we began by using alginate solution with fluorescent microparticles, and assembled our system under a large focal length fluorescence microscope to visualize gel behavior (Fig 1D). The use of fluorescent particles, as opposed to a fluorescent dye, allowed us to observe the process of gelation, as individual particle movements were visible and could indicate if the alginate has fully gelled. From our initial testing, it was apparent that the main challenge was to ensure that the gel attached firmly to the fiber mesh, rather than the microcapillary tip. We reasoned that gelation should be slow enough that the gel has time to flow into the mesh, while also rapid enough to prevent uncontrolled dispersion of cells. We tuned our parameters iteratively, and from initial testing of a few different parameters, we obtained the existing protocol. One important finding was that a slightly wetted (Fig 1F), rather than a fully submerged mesh, reduces the risk of the gel adhering to the capillary rather than to the fiber mesh.

Our goal was to use alginate encapsulation as an intermediate step, followed by a subsequent release of cells by slow alginate digestion. Ideally cells would not remain in the alginate, which could defeat the purpose of using electrospun fibers, but instead migrate or be released from the alginate to the fiber mesh in a controlled manner. We initially attempted the calcium chelator citric acid, but found that this approach dispersed the gel uncontrollably. Changing to enzymatic digestion by alginate lyase, we could achieve slower and more controlled gel digestion.

We applied our protocol to human neural stem cells (hNSCs) which were seeded with or without gel release on electrospun fibers (Fig 1E). As reference, we seeded hNSCs by standard pipette seeding on electrospun fibers. These groups are hereafter referred to as “alginate”, “alginate lyase cell release” (Alg+Aly) and “reference”.

All three groups were studied at three different time-points after seeding: 4, 14 and 21 days (Fig 1E), in order to track long-term viability and differentiation, and the stability of seeding area, culture thickness and cell density. We aimed at seeding areas below 10 mm^2^, as this is an ideal area for interfacing the fibers with micro electrode arrays, or electrophysiological probes. We wanted to make sure that cultures are stable for a minimum of 21 days, as this is roughly the time it takes for maturation of neuronal networks in our prior work [33]. Each time point and group had composed of three independent cultures, and the entire experiment was replicated on three separate instances.

3.2 Seeding area and viability

The aim of the established protocol was to control the distribution of cells on a fiber substrate and by using fluorescence microscopy, we could characterize the distribution of cells on fiber substrates. Standard pipette seeding onto fibers always resulted in a disperse and heterogeneous cell distribution across the entire 800 mm^2^ fiber mesh (Fig 2A–2B). By seeding cells with our microcapillary delivery device, we could consistently reduce the seeding area to less than 10 mm^2^ (Fig 2C–2E). There was no clear difference in seeding area between cells which were released by alginate lyase and those retained in alginate, which indicated that the cell release did not significantly disperse the cells. In some cases, most of the cell population was missing. This only occurred in microcapillary-seeded cultures, and given that a common complication during prototyping this technique was gel detachment, a potential explanation is that these cultures did not successfully attach to the fiber substrate.

Seeding areas achieved with microcapillary injection.A) With hand-held pipette seeding, the cells typically spread unevenly over the majority of the substrate area, which in this case was 800 mm2. A to-scale version of a stitched fluorescence micrograph of DAPI-stained cells is shown on the circle, which represents the entire substrate area B) Stitched fluorescence micrograph of DAPI-stained cells, showing the distribution of cells on a fiber scaffold, with typical low-density and heterogeneous distribution of cells. C) Stitched fluorescence micrograph of DAPI-stained cells seeded by microcapillary seeding, and subsequently released with alginate lyase, showing the controlled distribution achieved through this technique. D) Stitched fluorescence micrograph of DAPI-stained cells seeded by microcapillary seeding, without alginate lyase addition, showing highly similar distribution characteristics to those treated with alginate lyase. E) Quantification of area occupied by cells seeded through microcapillary injection, and either released by alginate lyase (Alg + Aly) or not (Alginate). No significant changes are seen, which indicates that cells remain attached over time, and during treatment of alginate lyase. We also get an idea of the variability of areas; most of the groups show the 50th quantiles within a range that doesn’t exceed 5mm2. From left to right: N=(17,17,18,17,13,12), Boxes: Interquartile range; Whiskers: Min/Max. P-values estimated by t-test with Bonferroni correction.

One potential concern when seeding cells through a small aperture is the risk of cell death due to shear forces, which are worsened by viscosity. Elevated calcium concentrations can also have negative impact on cell survival. To estimate viability, we performed Alamar blue assays at every time point just prior to fixation of cells (Fig 3). Both MCDD seeded groups consistently showed lower respiration than the standard pipette seeding. However, it is important to recognise that Alamar blue measures total amount of respiration, which can be due to changes in number of cells as well as changes in cell behavior (see discussion for more detail). In some cases, the seeding area was virtually 0, where no substantial cell population was seen, although some small number of cells were found on the mesh. A likely scenario here is that the gel detached, or that something went wrong during seeding.

Estimated respiration by Alamar blue assay.Alamar blue assay of cells seeded by hand-held pipette (Fiber Mesh), or with microcapillary injection, either with (Alg + Aly) or without (Alginate) alginate lyase release. Cultures seeded through MCDD seem show reduced of respiration. Its important to note that this may be due to several factors other than cell count, see results and discussion for further discussion. N=14-18, Boxes: Interquartile range; Whiskers: Min/Max. P-values estimated by t-test with Bonferroni correction.

For instance, the effect of gel and 3D culture conditions has a documented effect on metabolism [34]. We also can’t rule out that the difference in cell distribution, MCDD cultures being denser, and spread out over a much smaller area, could have some influence on this result, given that Alamar blue is a dye that needs to diffuse passively throughout the culture on order to have its effect, an effect that has been reported in other studies [35]. Furthermore, we can’t at this point establish whether this apparent decrease in respiration is due to any of the effects above, or simply by a decrease in total number of cells that was indeed seeded. In summary, it is clear that the use of our MCDD yielded substantial concentrations of surviving and healthy looking cells, as seen by the subsequent antibody labelling and fluorescence imaging.

3.3 Cell density, differentiation and culture thickness

We were interested in examining if the microcapillary seeded cells were still potent to differentiate into glial- and neuronal cells as previously demonstrated [25]. It is crucial for any model system which relies on differentiation of specific cell types to not only survive, but also to keep their potential to differentiate.

Doublecortin (DCX) and Glial Fibrillary Acidic Protein (GFAP) was labelled by immunocytochemistry and served as markers for neurons and glia respectively. Overall, in all three groups and at all time-points, we found expression of both GFAP- and DCX-positive cells, indicating maintained capacity for the hNSC to differentiate into both glia and neurons when seeded by the MCDD.

By using structured illumination or deconvolution microscopy, z-stacks were captured in selected areas (Fig 4A,B) where in each image stack, every DAPI-stained nuclei and DCX-positive was counted manually. The ratio of these two were examined across the three time points. Throughout the course of culture, we detected no significant differences in differentiation between MCDD cultures and reference culture conditions.

Differentiation, thickness and density of cultures.A) Example micrograph of cells seeded by microcapillary injection and released with alginate lyase, antibody labelled for DCX and stained with DAPI, here fixed 4 days after seeding. B) Same as A, but after 14 days, showing the increase in DCX+ cells. C) 3D rendering of a stitched image stack from a culture seeded by microcapillary injection. D) Same image stack as in C, but projected along the x-axis, to show the thickness of culture. E) Quantification of fraction doublecortin positive cells, showing that doublecortin (DCX) was not significantly different between the three groups. F) Quantification of culture thickness, which on day 4 and day 21 were higher in cells seeded with microcapillary injection and not released with alginate lyase (Alginate), compared to the two other groups. Note that fibers are not visible in fluorescence microscopy. G) Quantification of cell density. Medians are always higher in microcapillary seeding examples, but only significantly so at 14 days after seeding in cultures seeded with microcapillary and not released with alginate lyase (Alginate). N=9 for each group, Boxes: Interquartile range; Whiskers: Min/Max. P-values estimated by t-test with Bonferroni correction.

The z-stacks captured to examine DCX expression also allowed us to measure the thickness and cell density of cultures (Fig 4C,D). Note that fibers are not visible in fluorescence microscopy images. We observed that alginate suspended cells were significantly thicker at all time points, and that the difference appears to decrease over time. This result may suggest a gradual ‘compression’ of alginate suspended cultures occurring over time. Cultures where alginate was enzymatically digested only showed a significantly increased thickness at 4 days, which is expected as the structurally intact gel is likely what is supporting the increased thickness in the respective cultures. Similarly, the cell density was significantly higher only in cases where alginate was not digested. This result was lower than expected, and points to a significant dispersion of cells, corresponding to more than a 100-fold decrease in cell concentration compared to the density of cells in the original alginate-cell suspension.

3.4 Seeding spatially separate populations

In tissues, different cell types typically occupy defined structures, such as the distinct localization of dopaminergic neurons to certain regions in the brain. We wanted to test the ability to place hNSCs on multiple sites using our MCDD and thereby the potential to recreate tissue cell localization in vitro. This would require the initial seeding location to be fully terminated (i. e not ‘leaking’ cells after initial seeding location is completed), and that the cells are not dispersed by the movement of the microcapillary. By using the micromanipulator, we deposited the same number of total cells as two cell populations 3 mm apart (Fig 5A–5C). These preliminary results, shows the viability of using such a simple approach to produce spatially defined cultures. These are probably no way near the smallest controllable area possible to seed with our design, which in the bioprinting literature has been reported to be close to the nozzle radius under optimal conditions [36].

Seeding of multiple cultures on a single substrate.Stitched images of DAPI expressing cells, showing distributions of cells seeded on multiple locations. Circles outline the average area of cultures seeded in this manner as single cultures.

4 Discussion

In this study, we describe and evaluate a microcapillary delivery device (MCDD), which was used to seed gel encapsulated hNSCs on electrospun PCL fiber substrates. With our protocol, it was possible to deposit cells in controlled areas on the fiber substrate. We were also able to show that subsequent cell release by enzymatic gel digestion resulted in consistent cell deposition, and that both released cells and gel encapsulated cells retained viability and the ability to differentiate into neurons, as demonstrated by their expression of the neuronal marker doublecortin. Moreover, this approach allowed us to seed cells on multiple discreet locations on a single substrate, showing the feasibility to use this technique to produce spatially controlled co-cultures for organ-on-a-chip models or for transplant constructs.

It’s important to point out that the extent of cell-fiber integration is unknown. Although enzymatic digestion of alginate would ideally digest the entire gel, leaving only the fibers as the supportive substrate, this would have to be validated experimentally. An efficient way of testing this would be to use aligned fibers, which are known to influence neurite extension. This would allow a good quantitative comparison between released, encapsulated and standard fiber cultures in terms of fiber integration. Similarly, it’s unclear whether or not cells take up residence ‘inside’ the mesh or simply grow on top of it. In the latter case, some of the benefits of 3D cell culture are precluded, such as mechanical recapitulation of living tissue. Given that the fiber meshes we used are stated to be up to 100 μm and we only measured culture thickness of up to 30-40 μm, it’s at least safe to assume that cells are not evenly distributed throughout the fiber mesh.

Another important limitation to our study is the phenotypical assay. DCX expression is a marker for early neuronal lineage commitment, but is also highly variable and challenging to estimate objectively in high density cultures. Much of the diversity and complexity of the differentiation spectrum is not captured when using a single marker, and it’s possible that there are differences between the three groups which are not reflected in our analysis. The cell line we used have been shown to only develop spontaneous electrical activity on very few occasions [25], and therefore we did not anticipate functional maturation. In brief, many other phenotypical and functional tests would have to be conducted in order to further validate our method.

We managed to reduce the area where cells resided from a chaotic and heterogeneous distribution across the entire fiber substrate, which was around 800 mm^2^, down to less than 10 mm^2^. This is probably far from the smallest theoretical seeding area, but we chose to aim at seeding areas around 5 mm^2^ for a few reasons. Primarily, a larger area would be easier to target with electrodes, should the time arise that we would want to examine the electrophysiology of cultured neurons. Seeding a relatively large area also simplified evaluation of cell viability and behavior. While we believe that the success of our protocol would transfer down to smaller volumes and areas, eventually one would probably have to deal with the issue that the cell readily sticks to the capillary that the cells are being injected from. Small volumes probably suffer more form this as the contact area between the injected gel and the substrate decreases in relation to that of the capillary. Potentially, this could be solved simply by seeding into a bath of soft gel of some sort, as done in supported bioprinting [37].

Our primary concern when developing the MCDD was if cells would survive the stress of being pushed out through such a small capillary. The viscosity introduced by alginate would also increase shear forces, which could damage the cells and decrease viability. Our Alamar blue assay showed the presence of viable cells, although the results suggested a lower total respiration, this result is not conclusive. It is known that the presence of hydrogel and cell density can have various effects on the diffusion rate and availability of dye [35]. In 3D cultures, special considerations has to be made in order to obtain reliable results, as dye availability and secondary reactions can lead to non-linear behaviors [38]. In spheroids, this can be partially alleviated by reducing cell-cell junctions [39].

In some instances, there were no cells present on the fibers. A likely explanation for this is cell detachment during seeding, or other complications arising during the seeding procedure. We know from our initial testing that the most common complication is fiber detachment during seeding. Furthermore, there may be a general decrease in actual cell suspension injected by the capillary, where viscosity and the potential presence of compressible bubbles in the long tube connecting the capillary to the pump, could decrease the volume being injected.

Our approach shares some commonalities with some bioprinting systems which are worth discussing. Bioprinting is a highly diverse field with many technological approaches, applications and technical imperatives. Each system has unique sets of benefits and weaknesses in terms of resolution, 3D printability, structural integrity, and throughput. There are techniques which harness the full range from single-cell printing using microfluidic printheads [40], to large-scale printing of complex 3D shapes with supported extrusion printing [37]. Incorporating these printing techniques with electrospun fibers is mostly utilized for improving the structural integrity of printed tissue [30], as the vast majority of bioprinting systems utilize hydrogels [41] which have poor structural integrity. However, in many cases the cells remain embedded in the hydrogel, and will not integrate with the fiber mesh to a large extent, thus overlooking the advantages of fiber substrates. In some notable exceptions, fibers are fragmented and suspended in the printing solution, which can even be aligned by leveraging the shear forces during extrusion [31]. However, this approach allows only limited control of fiber topology and does not benefit from the structural strength of intact fiber meshes.

Our approach address many of the challenges mentioned above. By printing cells on fibers and subsequently releasing them by enzymatic digestion, high cell concentrations with controlled localization can be achieved on fibers. This means the electrospun fibers can be used to their full advantage, while also leveraging the control which comes with extrusion-based printing, using inexpensive materials. However, the implementation described in this article suffers from low throughput, as the injection process is entirely manual, a relatively poor resolution of about 1 mm diameter, and no 3D printability. However, all these issues are in theory addressable by using motorized stages, further optimization to lower blot radius, and potentially stacking of multiple fiber meshes after printing.

Many experimental protocols have been developed to provide localized cell deposition in order to model specific intercellular interactions. Particularly in neuroscience, there is interest in modelling how different neuronal cell types interact, as different cell types from different brain regions often engage in long-range interactions with other cell types of another brain region. Such model systems exist and are explored extensively [42], but often rely on microfabrication techniques that are not compatible with 3-dimensional culture substrates such as electrospun fibers. Our method can be applied to enable modelling of two or more different cell types by applying the sequential seeding that we tested to seed two cell populations some distance away from each other. It’s worth noting that something like a perfect model of brain tissue is still a very long way away from what current technology allows, and our approach is not an attempt of exceeding the current leading approaches to these ends. Using our technique, and applying it to seed multiple cell populations, still constitute a very simple model compared to the brain. However, these types of reductionistic models are very important and used extensively in medical research when simple cell models fail to capture more systemic interactions [43]. Using this approach could also be a more accessible approach for researchers with fewer resources, as it does not rely on clean room access, or expensive micro fabrication services.

Guiding how neurons connect, and the attempt of creating “in-vivo like” neuronal networks is an area of active research. Neurons cultured by traditional means -on flat silica glass substrates- spontaneously form synapses at random, and thus produces random and highly interconnected network architectures. While random network architectures could be a valid model in some cases, many neuronal structures exist where neurons connect in non-random ways, and several approaches for producing non-random neuronal network architectures in vitro have been developed. Two popular approaches include surface modifications of silica glass substrates [44], or use of microfabricated devices [45]. Again, both of these approaches precludes the use of 3-dimensional culture substrates. Using electrospun fibers as a means of producing non-random neuronal networks is a largely unexplored concept. To our knowledge, only a few experiments describes the network dynamics of neurons cultured on electrospun fibers [46]. One potential reason for this could be that neurons become much less accessible when grown in a 3-dimensional substrate. At least, this was a major motivation for us in creating the MCDD, as a higher density of cells could facilitate electrophysiological access.

Our results indicate many potential applications of this technique. We believe that the protocol we developed could be directly applicable to many different substrates, not just electrospun fibers. In fact, electrospun fibers are probably more difficult than hydrogels, for example, as a volume of hydrogel would provide more anchorage for the injected gel. In fact, what we have developed is not dissimilar to a conventional bioprinting device [47]. It’s encouraging that a bioprinting device with the ability to print onto a 3-dimensional substrate and with relatively high fidelity can be constructed by such simple means, and we hope to see this tool being used in surprising and innovative ways by other researchers.

5 Conclusions

Our research indicate that achieving both morphological control of culture substrate, in the form of electrospun nanofibers, and cell localisation using a simple syringe pump and a small diameter nozzle, is a relatively straight-forward concept. With the correct parameters, it is possible to seed small gel-encapsulated populations of cells, and release them by enzymatic digestion of the gel without causing dispersion of cells. We demonstrated that it’s possible to seed multiple such populations on a single substrate with no additional complexity, and our opinion is that the combined capacity of substrate morphology and cell localisation for controlling tissue microarchitecture ought to be further investigated.

Supporting information

S1 FigScanning electron micrograph of electrospun fibers.Image provided by suppliers (Cellevate AB, Sweden).(TIFF)