TGFβ3 is Neuroprotective and Alleviates the Neurotoxic Response Induced by Aligned Poly-L-Lactic Acid Fibers on Naïve and Activated Primary Astrocytes

Abstract

Following spinal cord injury, astrocytes at the site of injury become reactive and exhibit a neurotoxic (A1) phenotype, which leads to neuronal death. In addition, the glial scar, which is composed of reactive astrocytes, acts as a chemical and physical barrier to subsequent axonal regeneration. Biomaterials, specifically electrospun fibers, induce a migratory phenotype of astrocytes and promote regeneration of axons following acute spinal cord injury in preclinical models. However, no study has examined the potential of electrospun fibers or biomaterials in general to modulate neurotoxic (A1) or neuroprotective (A2) astrocytic phenotypes. To assess astrocyte reactivity in response to aligned poly-L-lactic acid microfibers, naïve spinal cord astrocytes or spinal cord astrocytes primed towards the neurotoxic phenotype (A1) were cultured on fibrous scaffolds. Gene expression analysis of the pan-reactive astrocyte makers (GFAP, Lcn2, SerpinA3), A1 specific markers (H2-D1, SerpinG1), and A2 specific makers (Emp1, S100a10) was done using quantitative polymerase chain reaction (qPCR). Electrospun fibers mildly increased the expression of the pan-reactive and A1-specific markers, showing the ability of fibrous materials to induce a more reactive, A1 phenotype. However, when naïve or activated astrocytes were cultured on fibers in the presence of transforming growth factor β3 (TGFβ3), the expression of A1-specific markers was greatly reduced, which in turn improved neuronal survival in culture.

Graphical Abstract

1. Introduction

Traumatic spinal cord injury (SCI) is a devastating condition that affects approximately 18,000 people each year in the United States. The estimated number of people living with SCI in the United States is 294,000 [1]. SCI typically occurs when a severe force displaces or initiates a burst fracture of the vertebrae leading to the rapid contusion and compression of spinal cord tissue (known as primary injury). Primary injury kills neurons and glia at the injury site. Multiple secondary injury cascades are initiated within minutes of the primary injury leading to an additional loss of neurons or glia at or near the site of injury [2–4]. Further damage may occur if the blood-brain-barrier is compromised and circulating leukocytes infiltrate the spinal cord [5] where they release neurotoxic mediators that can further damage or kill cells that survive the primary injury [6]. Axons in the white matter [7] or nearby motor neurons do have limited potential to regenerate; however, optimal axon regeneration cannot occur in the central nervous system (CNS) because of both neuron-intrinsic and extrinsic inhibitors of regeneration. The formation of a “scar” at the lesion border, comprised of multiple cell types including reactive astrocytes, is believed to be a potent extrinsic inhibitor of axon regeneration [8,9].

Astrocytes, the primary glial cells of the CNS, actively participate in the regulation of the extracellular environment by maintaining ionic and neurotransmitter homeostasis. They are responsible for uptake of excess glutamate from the extracellular space, thereby preventing neuronal death via excitotoxicity. Following SCI, astrocytes at the injury site become reactive [9]; this is characterized by an increase in the expression of the intermediate filament glial fibrillary acidic protein (GFAP) [9] and chondroitin sulfate proteoglycans (CSPGs) [4] along with an increase in cell migration [10] and proliferation [11]. Although reactive gliosis plays an important role in stabilizing the CNS tissue after the injury [12,13], the formation of a glial scar limits axonal regeneration.

The morphological appearance of an astrogliotic scar has been well documented; however, the underlying mechanisms responsible for the formation of the scar have only recently been elucidated. SCI transforms native astrocytes in and around the injury site into reactive astrocytes, which are eventually transformed into scar-forming astrocytes [14]. It was also reported that environmental cues instruct astrocytic fate, and that astrocytic scar formation is driven by type 1 collagen synthesis via the integrin-N-cadherin pathway. To better understand astrocytic phenotypic and functional heterogeneity, an in-depth genomic analysis of reactive astrocytes was done by Zamanian et al. to propose a working model. This analysis has shown that the reactive astrocytes exist in two distinct reactive states depending on the nature of the insult – a neurotoxic A1 phenotype or a neuroprotective A2 phenotype [15,16]. A1 reactive astrocytes were observed when mice were exposed to a neuroinflammatory agent (lipopolysaccharide) while an ischemia-induced insult triggered an A2 phenotype. A follow-up study revealed that the resident microglia activated by the inflammatory stimulus, lipopolysaccharide, drive naïve astrocytes towards this A1 reactive state [17]. Since it has been shown recently that SCI causes an increase in the number of neurotoxic A1 reactive astrocytes [18,19], we assessed if we could alleviate this astrocytic response using a biomaterial intervention.

Research in the field of neural tissue engineering has received considerable attention in recent years [20] and various biomaterial scaffolds are being developed into nerve guidance channels or hydrogels that more effectively stimulate spinal cord tissue regeneration. However, biomaterial development for SCI has primarily focused on improving the health of neurons at the site of the injury (e.g., through local release of growth factors) [21–23] while little research is done assessing the ability of biomaterials to modulate astrocytic phenotype after an injury.

Previous work by our group has shown that aligned PLLA fiber conduits support the ingrowth and migration of astrocytes when implanted into an acute rat model of complete transection SCI [24]. Other studies have shown that astrocytes cultured on PLLA fibers increase their expression of the glutamate transporter, GLT-1 [25,26], endowing them with the ability to potentially reduce excitotoxity after SCI. We hypothesized that, since fibronectin-coated electrospun fibers can promote a neuroprotective astrocyte phenotype (increase in GLT-1 expression), PLLA fibers may reduce astrocyte expression of canonical A1 markers. We tested this hypothesis using quantitative polymerase chain reaction (qPCR) to measure the relative expression of pan-reactive, A1, and A2 specific astrocyte genes in naïve astrocytes or astrocytes exposed to media conditioned by inflammatory macrophages – a source of excitotoxins and cytokines in vivo after SCI [27]. We found that PLLA fibers induce a mild neurotoxic phenotype in both naïve and A1 reactive astrocytes. However, this response was alleviated by incubating astrocytes with the anti-inflammatory cytokine, TGFβ3 [17,28].

2. Materials and Methods

2.1. Electrospinning

Poly-L-lactic acid (PLLA) fibers were electrospun and characterized using previously described protocols [24,25,29]. Endotoxin-free PLLA (LACTEL Absorbable Polymers), suitable for animal testing, was used in this study. Briefly, a thin film of 4% w/w PLLA in chloroform (Sigma, St. Louis, MO) was drop cast onto 15 × 15 mm glass coverslips (Ted Pella, Redding, CA), and the solvent was left to evaporate. PLLA was dissolved in chloroform to make a 12% w/w polymer solution, and micron-diameter fibers were generated using a custom designed electrospinner. Glass coverslips coated with PLLA were affixed to a rotating collection wheel (1500 rpm), and electrospinning was done at 22°C and 22% relative humidity for 15 minutes at an electrical potential of 15 kV and flow rate of 2 ml/h. All polymer surfaces (film or aligned fibers) were sterilized overnight using 70% ethanol before use in experiments. Fiber diameter, alignment and density, characterization was conducted using previously published methods (Supplementary Material and Fig. S1). The fibers had an average diameter of 1.81 ± 0.26 μm and were oriented within 5° of each other with a density of 20 ± 2 fibers per 50 μm of the coverslip.

2.2. Astrocyte Isolation and Cell Culture

Astrocytes were isolated from the spinal cords of 2-day-old Sprague Dawley rat pups (Taconic Biosciences, Germantown, NY) and cultured using previously described protocols [30,31]. All procedures were approved by the Rensselaer Polytechnic Institute and the Ohio State University Institutional Animal Care and Use Committees and in accordance with the Guide for the Care and Use of Laboratory Animals from the National Institutes of Health. Briefly, after a rapid decapitation, spinal cords were isolated from the animals and treated with TrypLE express (ThermoFisher Scientific, Waltham, MA) and Deoxyribonuclease I (1 mg/ml; Sigma) for 30 min at 37°C and neutralized with astrocyte growth culture media. The resulting cell suspension was plated into poly-L-lysine (Sigma) coated 75 cm² tissue culture flasks. One day post-plating, the cell suspension was replaced with fresh astrocyte growth media containing Dulbecco’s Modified Eagle Medium supplemented with fetal bovine serum (10% v/v), glutamax (1% v/v), penicillin (100 IU/ml), and streptomycin (100 μg/ml), all obtained from ThermoFisher. Cells were maintained at 37°C in a 95% air/5% CO₂ incubator until they reached ~70% confluency after which the cell cultures were purified for astrocytes using a previously described shaking procedure [32]. The flasks were shaken on an orbital shaker for 20 h at 37 °C and 240 rpm, and the supernatant was discarded. Purified astrocytes were detached from the flasks by adding TrypLE express for 3 min and pelleted by centrifugation at 1500 rpm for 5 min. The pellet was resuspended in astrocyte growth media, and the cells were plated onto fibronectin (10 μg/ml; Sigma) coated PLLA films, PLLA fibers, or wells in a 12-well tissue culture plate at a density of 50,000 cells per scaffold and maintained in the incubator for 1 or 4 days until used for experiments. Two hours postplating, the culture media was carefully pipetted and replaced with fresh astrocyte growth media containing various stimulants. In a subset of the experiments, the astrocytes were activated in the flasks for 4 days prior to plating them onto the PLLA scaffolds. To activate the astrocytes in culture and drive them towards an A1-phenotype, we used macrophage conditioned media (MCM) which was obtained by treating rat bone marrow derived macrophages with 100 ng/ml lipopolysaccharide (O55:B5; Sigma) and 20 ng/ml interferon-γ (PeproTech, Rocky Hill, NJ) and collecting the media after 24 h. To activate the astrocytes, MCM was used at 25% (v/v) in astrocyte growth media. To test the effect of transforming growth factor β3 (TGFβ3) on astrocytes, the media with and without MCM was supplemented with 1 ng/ml recombinant human TGFβ3 (PeproTech). To test the neuroprotective effect of TGFβ3, we collected the conditioned media from the astrocytes treated with 25% MCM or 1 ng/ml TGFβ3 or both for 4 days to be used in PC12 cell culture.

2.3. Live Cell Imaging

To assess the viability and proliferation of astrocytes on the PLLA scaffolds, the cells were incubated with the vital fluorescent dye, calcein acetoxymethyl (AM) ester (1 μg/ml; ThermoFisher) with the addition of pluronic F-127 (0.25% w/v; ThermoFisher) for 15 min followed by the cell permeant nuclear stain Hoechst 33342 trihydrochloride trihydrate (10 μg/ml; Thermo Fisher) for another 15 min [33,34]. Cells were imaged at room temperature (RT) using an external imaging solution (10 mM HEPES, pH 7.4, containing 140 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 2 mM MgCl₂, 5 mM D-glucose). An Olympus IX-81 confocal microscope equipped with epifluorescence illumination (Metal Halide lamp, 120 W) driven by Metamorph Premier 7.7.3.0 imaging software (Molecular Devices, Sunnyvale, CA) was used to visualize and image the astrocytes. The fluorescence of calcein and Hoechst 33342 were visualized and imaged using fluorescein isothiocyanate (FITC) and 4′,6-diamidino-2-phenylindole (DAPI) filter sets, respectively, using a 10x LUC Plan FLN objective at 1 day and 4 days post-plating. For each condition and time point, ten coverslips were imaged with cells originating from three independent cultures. Five view fields (~0.67 mm²) per coverslip were imaged to ensure a representative population of cells. The number of cells per view field were counted using ImageJ 1.49v software (National Institutes of Health, Bethesda, MD). The cells positive for calcein and Hoechst 33342 were considered live, while the cells positive for Hoechst 33342 and negative for calcein were considered dead. The number of cells per coverslip was calculated by adding the number of cells in each of the five view fields. The number of live cells in each group and time point were divided by the average number of live cells on the PLLA films at 1-day post-plating to calculate the normalized live cell count for each coverslip. The percentage of dead cells was calculated by dividing the number of dead cells by the total number of cells for each coverslip.

2.4. RNA Isolation and Quantitative Polymerase Chain Reaction (qPCR)

To measure the levels of expression of the various astrocyte reactivity markers, the cells were lysed using TRIzol (ThermoFisher) and the RNA was isolated per manufacturer’s instructions. cDNA was prepared using M-MLV Reverse Transcriptase (ThermoFisher), and the cDNA obtained was amplified using Fast SYBR™ Green Master Mix (ThermoFisher) and the primer sets listed in Table 1. Primer sets were obtained from Integrated DNA Technologies (Coralville, IA) after conducting BLAST analyses. The reaction was run in an Applied Biosystems 7900HT Fast Real-Time PCR System. The samples were collected from at least 3 independent cultures and all the genes assessed were run in triplicate. The relative gene expression was calculated using ΔΔCt method with conditions normalized to β-actin and the untreated astrocyte controls plated in tissue culture wells.

Table 1.

Primer sets used to assess the gene expression of the astrocyte reactivity markers

	Gene	Forward Primer	Reverse Primer
Pan	GFAP	GCCAAGCACGAGGCTAATGA	GACTCAAGGTCGCAGGTCAA
	Lcn2	TGTCCGATGAACTGAAGGAGC	GGTGGGAACAGAGAAAACGATG
	SerpinA3	TGCAATAGCTCCCTTTTTGGC	ACAGACAGGCTCAATGCCTC
A1	H2-D1	CCATCCACCGACTCCAACTTA	CTCCTCCTCACAACAACCACC
	SerpinG1	AGGCTTGAGTAAGCACTGCC	AGGCAGTGGCGAGGAAATAG
A2	Emp1	ATAAAGTCCTGGTGGTGGGC	CCTTGCAGCCCCAACAATTC
	S100a10	TGCCGAGATGGAAAAGTGGG	GAGAGGACGCATCAAGGTGT
	β-Actin	GCAGGAGTACGATGAGTCCG	ACGCAGCTCAGTAACAGTCC

2.5. PC12 Cell Culture

PC12 (rat pheochromocytoma cell line) cells were obtained from the American Type Culture Collection (ATCC, Catalog No. CRL-1721) and cultured in Dulbecco’s Modified Eagle Medium supplemented with fetal bovine serum (10% v/v), glutamax (1% v/v), penicillin (100 IU/ml), and streptomycin (100 μg/ml). The cells were maintained as monolayer cultures at 37°C in a 95% air/5% CO₂ incubator and the media was changed every 2–3 days. The cells were passaged once they reached 70–80% confluency. For the experiments, the PC12 cells were plated in 24-well tissue culture plates precoated with Collagen IV (100 μg/ml; Sigma) and treated with the conditioned media obtained from astrocytes (see section 2.2) in the presence of nerve growth factor (75 ng/ml, NGF-7S, Sigma) for 7 days with fresh media changes every 48 h. To test for the residual effects of TGFβ3 and MCM in the astrocyte conditioned media on PC12 cells, parallel experiments were performed where PC12 cells plated in 24-well plates were treated with PC12 media or TGFβ3 (1 ng/ml) diluted in PC12 media or MCM (25% v/v) diluted in PC12 media.

2.6. Immunocytochemistry

To assess astrocytic proliferation and PC12 cell survival, the cells were plated in 24-well tissue culture plates and labeled for Phalloidin using immunocytochemistry. Briefly, the cells were fixed with 4% paraformaldehyde at RT for 30 min and then permeabilized with 0.2% (v/v) Triton X-100 for 10 min. The cells were then incubated with 4% (w/v) bovine serum albumin (BSA) and 0.2% (v/v) Triton X-100 in PBS for 30 min to prevent nonspecific binding followed by an overnight (>12 h) incubation of the cells at 4°C with Alexa Fluor 546 Phalloidin for astrocytes or Alexa Fluor 488 Phalloidin for PC12 cells (1:500; ThermoFisher). Cells were then washed three times with PBS and the nuclei were counterstained with DAPI. Imaging was done using an ArrayScanXTI High Content Analysis Reader (ThermoFisher) and a 20x objective. Images were taken from 36 arbitrary view fields (~ 0.23 mm²) from each well (n = 3/group) and the images were analyzed using the associated HCS Studio Cell Analysis Software to count the number of cells per view field. The number of cells per view field in all the treatment groups were divided by the average number of cells in the untreated group to calculate the normalized cell count.

2.7. Statistical Analysis

All data are reported as medians with interquartile range or mean ± standard error of mean. Statistical analysis was performed using the GraphPad Prism 8 Statistical Software (GraphPad Software, San Diego, CA). The number of subjects required for each of the assays was estimated using power analysis (set at 80% and α = 0.05). Nonparametric statistics were used for all the subgroups that deviated from normality based on the Shapiro-Wilk test. To test the differences between the two independent groups, Mann-Whitney U-test (Fig. 1) was used. To test the difference between the 1 day and 4-day time points, the two correlated groups were compared using Wilcoxon Signed-Rank test (Fig. 1). To test the differences between the normally distributed multiple independent groups were first analyzed using one-way ANOVA followed by post hoc Tukey’s test for multiple comparisons (Figs. 2, 3, 6–9). For all others, the independent groups were first analyzed using the Kruskal–Wallis one-way ANOVA followed by the Dunn’s test for multiple comparisons (Figs. 4, 5). Significance was established at p ≤ 0.05.

Fig. 1.

Rat spinal cord astrocytes show enhanced adhesion on fibronectin-coated PLLA fibers in culture. (A) Representative images of spinal cord astrocytes in culture plated onto PLLA films (top) or fibers (bottom) and loaded with Calcein (left) and Hoechst 33342 (middle), 4 days post-plating. The right column shows a merge of the images. Scale bar, 100 μm. (B) Summary graphs showing the median (with interquartile range) normalized live cell count (left) and the percentage of dead cells (right), 1 day and 4 days post-plating. Pound symbols (#) indicate a statistical difference compared to the PLLA films at the corresponding time point. Time-dependent differences within the groups are marked by the horizontal bars and asterisks. ^#p < 0.05, ^## and **p < 0.01.

Fig. 2.

PLLA scaffolds induce a reactive response in cultured spinal cord astrocytes. Summary graph showing the relative gene expression of the various astrocyte reactivity markers when naïve astrocytes were plated onto PLLA films or fibers, 4 days post-plating. All gene expression data were normalized to the housekeeping gene, β-actin, and the astrocyte controls in tissue culture wells, represented by the dashed horizontal line. Pound symbols (#) indicate a statistical difference compared to the astrocyte controls in tissue culture wells. Other differences are marked by the horizontal bars and asterisks. ^## and **p < 0.01.

Fig. 3.

TGFβ3 alleviates the A1-reactive astrocyte phenotype induced by MCM in culture. Summary graph showing the relative gene expression of the various astrocyte reactivity markers when naïve astrocytes were plated into tissue culture wells and treated with TGFβ3. All gene expression data were normalized to the housekeeping gene, β-actin, and the untreated astrocyte controls, represented by the dashed horizontal line. Pound symbols (#) indicate a statistical difference compared to the untreated astrocyte controls. Other differences are marked by the horizontal bars and asterisks. ^# and *p < 0.05, ^## and **p < 0.01.

Fig. 6.

The reactivity of naïve spinal cord astrocytes on PLLA fibers is alleviated by TGFβ3. Summary graph showing the relative gene expression of the various astrocyte reactivity markers when naïve astrocytes were plated into tissue culture wells and treated with TGFβ3 or 25% MCM or both. All gene expression data were normalized to the housekeeping gene, β-actin, and the untreated astrocyte controls in tissue culture wells, represented by the dashed horizontal line. Pound symbols (#) indicate a statistical difference compared to the untreated astrocyte controls in tissue culture wells. Other differences are marked by the horizontal bars and asterisks. ^#p < 0.05, ^## and **p < 0.01. Note that the Fibers data presented here is the same as in Fig. 2.

Fig. 9.

The reactivity of continually MCM activated spinal cord astrocytes on PLLA fibers is alleviated by TGFβ3. Summary graph showing the relative gene expression of the various astrocyte reactivity markers when continually MCM activated astrocytes were plated into tissue culture wells or onto PLLA fibers and treated with MCM and TGFβ3. All gene expression data were normalized to the housekeeping gene, β-actin, and the naïve astrocyte controls in tissue culture wells, represented by the dashed horizontal line. Pound symbols (#) indicate a statistical difference compared to the naïve astrocyte controls in tissue culture wells. Other differences are marked by the horizontal bars and asterisks. *p < 0.05, ^## and **p < 0.01.

Fig. 4.

TGFβ3 reduces the cell count of naïve and MCM activated spinal cord astrocytes in culture. (A) Representative images of spinal cord astrocytes in culture plated into tissue culture wells and treated with TGFβ3 or 25% MCM or both and immunolabeled for Phalloidin (red) and DAPI (blue), 4 days post-plating. Scale bar, 100 μm. (B) Summary graph showing the median (with interquartile range) cell count of the astrocytes, normalized to the untreated (control) group. Pound symbols (#) indicate a statistical difference compared to the control. Other difference is marked by the horizontal bar and asterisks. ^## and **p < 0.01.

Fig. 5.

TGFβ3 conditioned astrocyte media rescues PC12 cells from MCM induced neurotoxicity in culture. (A) Representative images of PC12 cells in culture plated into tissue culture wells and treated with the conditioned media obtained from astrocytes treated with TGFβ3 (A-TGFβ3) or 25% MCM (A-MCM) or both (A-MCM+TGFβ3) or none (ACM), and immunolabeled for Phalloidin (green) and DAPI (blue), 7 days post-plating. Scale bar, 100 μm. (B) Summary graph showing the median (with interquartile range) cell count of the PC12 cells, normalized to the ACM group. Control, TGFβ3 and MCM represent the PC12 cells in tissue culture wells treated with PC12 media, TGFβ3 and MCM, respectively. Pound symbols (#) indicate a statistical difference compared to ACM. Other differences are marked by the horizontal bars and asterisks. *p < 0.05, ^## and **p < 0.01.

3. Results

3.1. Effect of PLLA scaffolds on the adhesion and proliferation of spinal cord astrocytes

We first assessed the effect of the PLLA scaffolds on the adhesion and proliferation of spinal cord astrocytes in culture. We plated astrocytes onto fibronectin-coated PLLA films and fibers then after 1 or 4 days in culture, cells were loaded with calcein and their nuclei labeled with Hoechst 33342 after which cultured cells were imaged to estimate the number of live and dead cells. All cells positive for calcein and Hoechst 33342 were considered live while cells negative for calcein and positive for Hoechst 33342 were considered dead [33,34]. We found that the astrocytes respond to the topography of the PLLA fibers as evidenced by their directed growth along the length of the fibers (Fig. 1A). We normalized the live cell count in all the groups to the number of live cells on PLLA films at 1 day post-plating and found that the normalized live cell count at the 1-day post-plating time point on the PLLA fibers was significantly higher than that on the PLLA films implying that the PLLA fibers cause an enhanced adhesion of astrocytes compared to PLLA films in culture. We also observed that the astrocytes on both the substrates showed significant time-dependent increase in the normalized live cell count, which we used as an indirect measure of proliferation, implying that the cells are proliferating on both the PLLA scaffolds (Fig. 1B). There was also an increase in the normalized cell count at the 4-day post-plating time point on the PLLA fibers compared to the films. However, we did not observe any significant difference in the rate of proliferation (~30% on both the substrates) implying that this increase is caused by the increased adhesion of astrocytes onto the fibers. In addition, we also assessed the percentage of dead cells and found that the astrocytes on both substrates showed a significant decrease in the percentage of dead cells over time. However, the percentage of dead cells on the PLLA fibers was higher than that on PLLA films at the 4-day time point, albeit the value is less than 1.5%. Taken together, these results show that the spinal cord astrocytes respond to the morphology of the PLLA fibers and show increased adhesion in culture.

3.2. Effect of PLLA scaffolds on the reactivity of spinal cord astrocytes

A working model has been proposed to characterize astrocyte heterogeneity, both phenotype and function, based on expression of specific genes/proteins – a neurotoxic A1 phenotype and a neuroprotective A2 phenotype [15,17]. We determined whether astrocytes cultured on PLLA fibers, in addition to changing their morphology, would also adopt one of these phenotypes. To test this, we plated astrocytes onto fibronectin-coated PLLA films or fibers and then performed quantitative polymerase chain reaction (qPCR), 4 days post-plating; cells plated in tissue culture wells were used as a control. We assessed the level of expression of several astrocyte genes which we segregated into 3 categories –pan-reactive genes expressed by both A1 and A2 astrocytes (GFAP, Lcn2 and SerpinA3), A1 specific genes (H2-D1 and SerpinG1), and A2 specific genes (Emp1 and S100a10) [17]. We found that naïve astrocytes on PLLA fibers increased their expression of all pan-reactive genes (except GFAP) and A1-specific genes but not A2-specific genes, compared to control astrocytes (Fig. 2). The astrocytes on PLLA films also show an upregulation in Lcn2 and H2-D1, although to a lesser extent compared to the PLLA fibers. Together, these data indicate that the PLLA substrate activates astrocytes, promoting the development of a neurotoxic A1 phenotype but that this phenotype is exacerbated when astrocytes elongate along PLLA fibers.

3.3. Effect of TGFβ3 on the reactivity of naïve and activated astrocytes

Since the goal of our work is to alleviate astroglial reactivity and provide a scaffold to enhance axon regeneration, we used a combinatorial approach to help achieve both of these goals. A previous report showed that activating the fibroblast growth factor (FGF) signaling pathway could alleviate the neurotoxic A1 phenotype in brain astrocytes [17,28]. Since the anti-inflammatory cytokine TGFβ3 has been shown to activate this pathway, we tested the effects of TGFβ3 on reactive spinal cord astrocytes. First, we plated naïve astrocytes in tissue culture wells and treated them with TGFβ3 (1 ng/ml). After 4 days in culture, we evaluated gene expression and found that TGFβ3-treated cells showed a decrease or no change in pan-reactive, A1-specific, and A2-specific markers compared to the untreated cells (Fig. 3). To further study this effect in the context of SCI, we drove astrocytes towards a neurotoxic A1 phenotype in culture using a biologically-relevant stimulus found in the lesion site, then determined whether this response could be alleviated with the addition of TGFβ3. Specifically, we treated astrocytes with media conditioned by inflammatory macrophages [hereafter referred to as macrophage conditioned media (MCM)]. We found that after 4 days in culture, astrocytes treated with MCM showed a significant increase in their expression of the pan-reactive markers Lcn2 and SerpinA3 (Fig. 3). They also showed an increase in the A1-specific markers H2-D1 and SerpinG1 and a decrease in the A2-specific markers Emp1 and S100a10 implying that the astrocytes in culture can be successfully driven to an A1 phenotype using MCM. We then tested whether TGFβ3 could antagonize the effects of MCM by analyzing astrocyte gene expression after treating astrocytes with a combination of MCM and TGFβ3 for 4 days. Notably, expression of Lcn2, SerpinA3, H2-D1 and SerpinG1 were significantly reduced in cells treated with MCM and TGFβ3 when compared to the cells treated with MCM alone (Fig. 3). We also found that the inclusion of TGFβ3 further decreased the expression of the A2 markers. Taken together, these data suggest that TGFβ3 can alleviate the induction of an A1 astrocyte phenotype provoked by inflammatory stimuli.

3.4. Effect of TGFβ3 on the cell count of naïve and activated astrocytes

To confirm that TGFβ3 is indeed a good candidate to alleviate astrocytic reactivity and reduce the neurotoxicity in culture, we treated astrocytes in tissue culture wells with TGFβ3 or MCM or both and assessed the change in cell count of astrocytes by labeling the cells for phalloidin (Fig. 4), 4 days post-plating. We found that MCM did not affect astrocytic cell count while TGFβ3 significantly reduced the cell count of astrocytes as evidenced by the decrease in the normalized cell count, 4 days post-plating (Fig. 4). The astrocytes treated with MCM and TGFβ3 also showed a significant decrease in cell count compared to the untreated or MCM treated astrocytes (Fig. 4). Combined, this data shows that the inclusion of TGFβ3 reduces the cell count of both naïve and activated astrocytes in culture. Since astrocytes proliferate in culture in the presence of serum, this decrease in cell count could be indicative of a decrease in the proliferation potential of astrocytes in the presence of TGFβ3, which could be a bonus to limit the proliferation of reactive astrocytes when translated in vivo in the context of a spinal cord injury.

3.5. Effect of TGFβ3 conditioned astrocyte media on the survival of MCM challenged PC12 cells

Finally, to show that TGFβ3 has the potential to promote neuroprotection by alleviating neurotoxicity in culture, we treated the astrocytes with TGFβ3 or MCM or both or none and collected the conditioned media from the astrocytes after 4 days in culture to obtain A-TGFβ3 or A-MCM or A-MCM+TGFβ3 or ACM, respectively. We then plated PC12 cells in tissue culture wells coated with collagen IV and assessed the survival of PC12 cells by treating them with the different astrocyte conditioned media collected (Fig. 5), 7 days post-plating. The media was also supplemented with NGF-7S to cease proliferation of PC12 cells and promote their differentiation [35,36]. We found that A-MCM significantly reduced the PC12 cell count compared to cells treated with ACM, implying that A-MCM induces neurotoxicity in culture (Fig. 5). We found that the cells treated with the A-TGFβ3 did not show any significant differences in PC12 cell count compared to ACM-treated cells. However, the cells treated with the combined A-MCM+TGFβ3 showed a significant increase in the PC12 cell count compared to the A-MCM-treated cells (Fig. 5). Since PC12 cells do not proliferate in the presence of NGF-7S, this increase in cell count with the inclusion of TGFβ3 in A-MCM-challenged PC12 cell cultures indicate that TGFβ3 promotes neuronal survival in culture and can be used as a viable candidate in conjunction with the PLLA fibers to alleviate neurotoxicity and promote the growth of neurons.

We also performed parallel experiments in which the PC12 cells were treated with PC12 media or TGFβ3 (1 ng/ml) or MCM (25%) diluted in PC12 media to assess the effects of residual TGFβ3 or MCM in the astrocyte conditioned media on PC12 cells (Fig. 5). We found that the cells treated with PC12 media or TGFβ3 did not show any significant differences in cell counts compared to cells treated with ACM. The cells treated with MCM, however, showed a 20% reduction in cell counts compared to the ACM-treated cells. Albeit this change was significantly lower than the 40% reduction observed using A-MCM. Taken together, this data shows that TGFβ3 can be used as a viable candidate in conjunction with the PLLA fibers to alleviate neurotoxicity and promote the growth of neurons.

3.6. Effect of TGFβ3 on the reactivity of naïve astrocytes on PLLA fibers

To test if TGFβ3 can also alleviate the A1-inducing effects of PLLA fibers, we plated astrocytes onto PLLA fibers, then treated them with TGFβ3. We studied their gene expression at 4 days-post plating and found that TGFβ3 reduces the reactivity of astrocytes on PLLA fibers, as shown by the decrease in the expression of Lcn2, H2-D1 and SerpinG1 (Fig. 6). We did not, however, see any differences in the expression of the A2 markers.

3.7. Effect of PLLA scaffolds on the reactivity of pre-activated spinal cord astrocytes

PLLA scaffolds alone significantly affect astrocytes, inducing an A1 phenotype (Fig. 2). However, in the context of SCI, PLLA scaffolds would be placed into a lesion environment where astrocytes would already be exposed to A1-inducing factors. Therefore, we tested whether the A1 astrocyte phenotype, which is expected to be prevalent in vivo after SCI, would be influenced by PLLA scaffolds. To test this, we first treated astrocytes in tissue culture flasks with MCM for 4 days to activate astrocytes. These activated astrocytes were then plated into tissue culture wells or onto PLLA scaffolds. After 4 days in culture, astrocyte gene expression was analyzed. Compared to naïve astrocytes, A1 astrocytes continue to express significantly higher levels of Lcn2, SerpinA3, H2-D1 and SerpinG1, regardless of the substrate that they are exposed to in vitro (Fig. 7). In fact, both PLLA films and fibers further enhanced the expression of the Lcn2 and SerpinG1, with fibers being the most potent stimulus for augmenting SerpinA3 and H2-D1 gene expression (Fig. 7). The expression of the A2 markers, however, was not significantly different between the pre-activated astrocytes on all the substrates. Together, these data indicate that the reactivity of the pre-activated astrocytes is retained in culture and this effect is further enhanced by PLLA fibers.

Fig. 7.

MCM pre-activated spinal cord astrocytes retain their reactivity post-plating and this reactivity is further enhanced by the PLLA scaffolds. Summary graph showing the relative gene expression of the various astrocyte reactivity markers when pre-activated astrocytes were plated into tissue culture wells or onto PLLA films or fibers. All gene expression data were normalized to the housekeeping gene, β-actin, and the naïve astrocyte controls in tissue culture wells, represented by the dashed horizontal line. Pound symbols (#) indicate a statistical difference compared to the naïve astrocyte controls in tissue culture wells. Other differences are marked by the horizontal bars and asterisks. *p < 0.05, ^## and **p < 0.01.

3.8. Effect of TGFβ3 on the reactivity of activated astrocytes on PLLA fibers

We then tested whether TGFβ3 can alleviate the effects of the PLLA fibers on activated astrocytes. MCM-activated astrocytes were plated into tissue culture wells or onto PLLA fibers then were treated with TGFβ3 for 4 days. Gene expression analysis revealed that, in the presence of TGFβ3 the panreactive and A1-specific astrocyte phenotypic markers were decreased on activated astrocytes, even those astrocytes cultured on PLLA fibers (Fig. 8). The A2-specific marker, Emp1, also showed a decrease in its expression in the activated astrocytes treated with TGFβ3.

Fig. 8.

The reactivity of MCM pre-activated spinal cord astrocytes on PLLA fibers is alleviated by TGFβ3. Summary graph showing the relative gene expression of the various astrocyte reactivity markers when pre-activated astrocytes were plated into tissue culture wells or onto PLLA fibers and treated with TGFβ3. All gene expression data were normalized to the housekeeping gene, β-actin, and the naïve astrocyte controls in tissue culture wells, represented by the dashed horizontal line. Pound symbols (#) indicate a statistical difference compared to the naïve astrocyte controls in tissue culture wells. Other differences are marked by the horizontal bars and asterisks. *p < 0.05, ^## and **p < 0.01. Note that the Act. and Act. Fib data presented here are the same as in Fig. 7.

3.9. Effect of TGFβ3 on the reactivity of continually activated astrocytes on PLLA fibers

Since astrocytes at the injury site are chronically exposed to infiltrating immune cells and their release products, we next tested whether the efficacy of TGFβ3 in mitigating the reactivity of astrocytes would be maintained in the presence of continuous inflammatory stimuli. Astrocytes were pre-activated with MCM, plated onto the PLLA fibers then treated with TGFβ3 in the presence of MCM for 4 days in culture. Similar to results in Fig. 8, regardless of the PLLA substrate, TGFβ3 treatment significantly reduced the expression of pan-reactive, A1-specific and A2-specific astrocyte markers compared to the astrocytes stimulated by MCM (Fig. 9).

4. Discussion

In this study, we show that spinal cord astrocytes survive and proliferate on PLLA fibers in culture while extending along the aligned fiber topography. They show enhanced adhesion of the PLLA fibers compared to the films, possibly because of the increased surface area offered by the fibers compared to the films [37]. Contrary to our initial hypothesis, however, the fibers caused a mild increase in the expression of A1 reactivity markers implying that they might be driving a neurotoxic phenotype in culture. It has been shown that the physical interaction of astrocytes cultured on fibrous scaffolds could change their morphology as well as their gene expression [25,38–43]. Previous reports also have shown that PLLA fibers can increase astrocytic expression of glutamate transporters in culture [25,26]. However, to our knowledge, this is the first report to evaluate the effects of a biomaterial in evoking gene expression changes in astrocytes that are potentially associated with discrete effector functions.

We show that macrophage conditioned media (MCM) can drive an A1 phenotype in culture without effecting the proliferation of the astrocytes. This phenotype in culture was shown by Liddelow et al. to be driven by activated microglia [17]. Since microglia, the resident macrophages in the central nervous system, and bone marrow derived macrophages exert similar functions following an injury, it is plausible that they release similar cytokines that can activate astrocytes. One notable difference, however, is the lack of an increase in the expression of the glial fibrillary acidic protein (GFAP), which is considered a classic reactivity marker for astrocytes. On the contrary, we found a decrease in GFAP expression in the presence of MCM (Fig. 3). This could be because of a combination of cytokines in MCM such as interleukin 1 beta (IL-1β) and tumor necrosis factor alpha (TNFα) (both produced by activated macrophages [44,45]), which have been shown extensively to decrease astrocytic expression of GFAP [17,46–48]. Work by others have also shown that the MCM can drive a reactive phenotype in astrocytes, albeit, those studies only evaluated changes in the pan-reactive astrocytes markers [49–51].

Finally, we show that the TGFβ3 alleviates the ability of PLLA fibers or MCM to promote an A1 reactive astrocyte phenotype. It is worth noting, however, that the increase in the reactivity of the astrocytes induced by the fibers alone is orders of magnitude lower than that caused by MCM (Figs. 2, 3). Several groups have shown the potential of FGF signaling in suppressing astrocytic activation or in maintaining their non-reactive state [28,52,53]. TGFβ3 is an anti-inflammatory cytokine known to signal via the FGF pathway which has been shown to reset the A1 astrocytes in culture to their non-reactive state. Although we show a decrease in the expression of the A1 markers in the presence of TGFβ3, this decrease doesn’t reach the basal level of expression as expected. This could be explained, in part, by the usage of a single dose of TGFβ3 at a concentration of 1 ng/ml over the 4-day time period. A follow up study could test whether longer duration treatment with TGFβ3 or higher concentrations of TGFβ3 would be more effective at reducing A1 gene expression profiles. Another potential limitation of this study is that all the analyses focused on measuring gene expression changes and not the actual levels of protein produced by astrocytes. However, we show using PC12 cells that the conditioned media from TGFβ3 treated MCM-challenged astrocytes drives a neuroprotective effect in culture (Fig. 5).

We also show that the inclusion of TGFβ3 reduced the expression of the neuroprotective A2 markers as well in addition to the A1 markers in the astrocytes treated with MCM (Figs 3, 8 and 9). There is no additional change in the expression of the A2 markers, however, when the astrocytes were plated on PLLA fibers (Figs 6, 8 and 9). This result contrasts with a previous report showing the efficacy of TGFβ3 in enhancing the expression of the A2 markers [17]. This discrepancy could possibly result from our use of MCM instead of a combination of interleukin 1 alpha (IL-1α), TNFα, and complement component 1, q subcomponent (C1q) to activate the astrocytes. MCM has more cytokines/chemokines that could potentially interfere with the mechanism of driving an A2 phenotype in astrocytes.

Thus, while fibers are known to stimulate the production of GLT-1 and enable neuroprotection in excitotoxic environments [27], fibers also seem to shift astrocytes towards a neurotoxic phenotype, at least at the time points tested here. A combinatorial system where the astrocyte reactivity can be mitigated (using TGFβ3) while at the same time providing a guidance scaffold for growth and neuroprotection may further enhance the potential of electrospun fibers for spinal cord injury applications. The next step would be to test this by using co-cultures of astrocytes and neurons to assess neuroprotection and neurite outgrowth in the presence of MCM and TGFβ3 on PLLA fibers. This could be done using an approach where the guidance scaffold and TGFβ3 are a part of the same system such that there is a prolonged release of TGFβ3 from the material over time. This could be achieved by incorporating TGFβ3 into the PLLA fibers during the electrospinning process as we have shown previously using riluzole and neurotrophin-3 [54]. Another approach would be to use a slow degrading biomaterial like a hydrogel [55,56] or coacervate [57,58], in conjunction with the PLLA fibers, that could preserve the bioactivity of TGFβ3 and enable a slow release over time making the approach translational for SCI applications.

Finally, since it has been shown that fiber composition, fiber diameter and its surface coating can influence the effect that fibers have on astrocytes [25,26,59], it is worth noting that this shift towards a neurotoxic phenotype is specific to the particular PLLA fibers that we used in this study and cannot be generalized to all electrospun fibers. An in-depth analysis taking these variables into consideration is warranted to get a more holistic understanding of the effect of electrospun fibers on astrocytes. It is also critical to understand that these experiments were done using immature astrocytes isolated from the spinal cords of 2-day-old pups, which are not representative of the mature astrocytes found in the injury site. Immature astrocytes have been shown to provide a favorable substrate for axonal growth while mature astrocytes do not have this ability [60,61]. Although the high yield and relative ease of isolation makes immature astrocytes a great tool to study astrocytic responses in vitro [62]. Also, studies assessing the response of astrocytes to various biomaterials have traditionally relied on the expression on GFAP and proteoglycans to understand astrocyte reactivity [43,60]. However, considering our updated understanding of astrocyte phenotypic changes, studying the effect of biomaterials on astrocytes must move beyond solely assessing the expression of these traditional markers and into performing a more comprehensive analysis, which involves characterization of astrocyte phenotype.

Conclusions

We show here that the PLLA fibers are highly biocompatible and spinal cord astrocytes respond to their surface topography in culture. Although they cause a mild increase in the neurotoxic response of the astrocytes, this response is alleviated by the inclusion of the anti-inflammatory cytokine, TGFβ3. In addition, this alleviation of astrocytic response extends to pre-activated as well as continually activated astrocytes in culture, which in turn promotes neuroprotection. Together, these findings show the potential of a combination of PLLA fibers and TGFβ3 as a translatable therapy to target and alleviate reactive astrogliosis following a SCI.

Statement of Significance.

This is the first study to assess the ability of biomaterials to influence the expression of A1/A2 astrocyte markers. Further, to our knowledge, it is the first study to analyze the ability of a biomaterial to mitigate reactivity if the astrocytes are first primed towards a reactive, A1 phenotype. Since previous studies have shown poly-L-lactic acid electrospun fibers can reduce glial fibrillary acidic protein expression and increase glutamate transporter expression, we hypothesized that fibers may also reduce the expression of A1 reactivity markers. Interestingly, fibrous scaffolds increased the expression of A1 reactivity markers slightly, but the increase in expression was orders of magnitude less than astrocytes induced towards an A1 phenotype by media conditioned by bone marrow derived macrophages. While electrospun fibers may induce astrocytes towards a neuroprotective phenotype, electrospun fiber scaffolds approaches may need to be supplemented with factors, such as TGFβ3, to reduce the expression of A1 reactivity markers. In total, a more holistic assessment (neuroprotective capability and the expression of reactivity markers) of astrocyte response to biomaterials may better predict biomaterial effectiveness in preclinical models of spinal cord injury.