Three-dimensional bioprinting collagen/silk fibroin scaffold combined with neural stem cells promotes nerve regeneration after spinal cord injury

Abstract

Many studies have shown that bio-scaffolds have important value for promoting axonal regeneration of injured spinal cord. Indeed, cell transplantation and bio-scaffold implantation are considered to be effective methods for neural regeneration. This study was designed to fabricate a type of three-dimensional collagen/silk fibroin scaffold (3D-CF) with cavities that simulate the anatomy of normal spinal cord. This scaffold allows cell growth in vitro and in vivo. To observe the effects of combined transplantation of neural stem cells (NSCs) and 3D-CF on the repair of spinal cord injury. Forty Sprague-Dawley rats were divided into four groups: sham (only laminectomy was performed), spinal cord injury (transection injury of T10 spinal cord without any transplantation), 3D-CF (3D scaffold was transplanted into the local injured cavity), and 3D-CF + NSCs (3D scaffold co-cultured with NSCs was transplanted into the local injured cavity. Neuroelectrophysiology, imaging, hematoxylin-eosin staining, argentaffin staining, immunofluorescence staining, and western blot assay were performed. Apart from the sham group, neurological scores were significantly higher in the 3D-CF + NSCs group compared with other groups. Moreover, latency of the 3D-CF + NSCs group was significantly reduced, while the amplitude was significantly increased in motor evoked potential tests. The results of magnetic resonance imaging and diffusion tensor imaging showed that both spinal cord continuity and the filling of injury cavity were the best in the 3D-CF + NSCs group. Moreover, regenerative axons were abundant and glial scarring was reduced in the 3D-CF + NSCs group compared with other groups. These results confirm that implantation of 3D-CF combined with NSCs can promote the repair of injured spinal cord. This study was approved by the Institutional Animal Care and Use Committee of People’s Armed Police Force Medical Center in 2017 (approval No. 2017-0007.2).

Chinese Library Classification No. R459.9; R651.2; R641

Introduction

Spinal cord injury (SCI) has become a global health problem and major focus in the field of neuroscience. Currently, there are few effective measures to treat SCI. Indeed, efficacies of salvage surgery, drug therapy, and rehabilitation training are limited (Kadoya et al., 2009; Liu et al., 2017a). In addition to lethal primary damage, many severe secondary factors such as inflammation, glial scarring, and cystic cavity formation also inhibit axonal regeneration (Chen et al., 2017; Liu et al., 2017b; Fakhri et al., 2018). In this situation, creating a mild microenvironment and providing necessary supports are two critical steps for neural regeneration (Chen et al., 2018; Yin et al., 2018).

Biological scaffolds no longer represent entirely new formulations nowadays, as they have been widely used in tissue engineering for years. First and foremost, the safety of scaffolds is the most important consideration. Some substances like polylactic acid, polyglycolic acid, and allografts cause strong rejection after entering the body (Badhe et al., 2008; Bond et al., 2008; Hammond et al., 2008; Lee, 2008); whereas, materials such as polypropylene and chitosan are hard to degrade, which directly leads to long-term retention and residual toxic degradation products in vivo (Dominkus et al., 2006; Jones et al., 2007; Ibrahim et al., 2008). In addition, fabrication techniques, controlled microarchitecture, and biomaterial composition remain unresolved but critical issues.

Collagen, a structural extracellular matrix protein that does not elicit immunogenicity after pretreatment and purification, exhibits good biocompatibility and degradation rates, which makes it a preferred biomaterial (Simionescu et al., 2006). However, poor physical and mechanical properties, and poor thermal stability are weaknesses of using collagen (Cornwell et al., 2007). Thus, it is advised to cross-link collagen with other kinds of materials. Silk fibroin is a tough and flexible protein with good thermal stability. After mixing and cross-linking, the advantages of both materials can be preserved.

Transplantation of cells and biomaterial scaffolds are considered to be promising therapies for nerve tissue repair (Shrestha et al., 2014). Indeed, many studies have combined biomaterials with cells to address issues in SCI (Liu et al., 2017; Zheng et al., 2017). Stem cells have been recognized as a potential source of alternative cells at the SCI lesion site, with mesenchymal stem cells (MSCs) being the most commonly transplanted (Zurita and Vaquero, 2006; Li et al., 2016). Although MSCs are multipotent cells, it is challenging to induce their differentiation into nerve cells. Moreover, the local environment of the injured site is harsh for MSCs to survive (Ma et al., 2018). Thus, we attempted to transplant neural stem cells (NSCs) into the injured site.

In addition to the matters above, the scaffold fabrication process is also a vital factor. Traditional methods primarily include fiber bonding, electrospinning (Jeffries et al., 2015), and gas foaming (Garg et al., 2015). However, there are many shortcomings in the degree of fineness and bionics achieved by these methods. As a result of the complex anatomical structure of the spinal cord, appropriate structure within the scaffold is essential for SCI repair. Damaged axon tracts in the white matter are the most direct and critical reason for the loss of sensory-motor function after SCI (Wu et al., 2014; Fakhri et al., 2019). Thus, fabrication of a scaffold that emulates the anatomical structure of white matter is essential for longitudinal axonal guidance during repair. The recent emergence of three-dimensional (3D) printing, a type of solid free-form fabrication, permits construction of the scaffold internal structure according to specific requirements (Cheng et al., 2019; Huang et al., 2019); thus, it is an ideal choice for the fabrication of biomimetic scaffolds.

To optimize the efficacy of neural regeneration, collagen and silk fibroin were mixed at a 4:2 ratioand shaped into a porous 3D collagen/silk fibroin scaffold (3D-CF) according to general anatomy of the spinal cord using a 3D printer. This 3D-CF exhibited stable properties, provided space for the survival and proliferation of NSCs, filled the injured cavity, and offered guidance for the regenerating nerve.

Materials and Methods

Animals

Female specific-pathogen-free adult Sprague-Dawley rats aged 12 weeks weighing 200–220 g [Animal License No. SCXK (Jun) 2016-0002] were provided by the Military Academy of Medical Sciences of the People’s Liberation Army (Beijing, China). Sprague-Dawley rats of 14–15 days gestational age were sacrificed for NSC extraction. The animal experiment was approved by the Institutional Animal Care and Use Committee of People’s Armed Police Force Medical Center in 2017 (approval No. 2017-0007.2). The experimental procedure followed the United States National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85-23, revised 1996).

Collagen and silk fibroin preparation

Collagen was produced from bovine tendon according to previously published research (Shreiber et al., 2003). Briefly, fresh bovine tendons were washed and stripped of their adventitia. After cleaning and crushing thoroughly, bovine tendon was soaked in 0.05 M Tris buffer (Yacoo Science Co., Ltd., Suzhou, China) for 24 hours to remove soluble impurities. After centrifugation, precipitates were collected and placed in acetic acid solution containing pepsin. The supernatant was collected by centrifugation at 2683 × g for 10 minutes at 4°C. After full dissolution, NaCl solution was added and salting-out sediments were collected by centrifugation. Sediments were dialyzed against deionized water at 4°C for 5 days, and deionized water was replaced every day to obtain a collagen gel. Silk fibroin (Kaidi Silk Co., Ltd., Jiangsu, China) was also prepared according to a previously described method (Ruan et al., 2011). Briefly, silk was boiled three times in a 0.5% Na₂CO₃ (Solarbio Science & Technology Co., Ltd., Beijing, China) solution at 90°C for 30 minutes each, air dried, dissolved in CaCl₂·CH₃CH₂OH·H₂O (mass concentration ratio, 1:2:8) (Solarbio Science & Technology Co., Ltd.) solution, stirred at 60°C for 2 hours, and centrifuged at 2683 × g for 10 minutes at 4°C to obtain a supernatant. The supernatant was dialyzed for 72 hours in a 3500D dialysis bag, which was transferred to 40% polyethylene glycol (Solarbio Science & Technology Co., Ltd.) to concentrate for 7 hours to obtain silk fibroin solution.

3D-collagen/silk fibroin scaffold preparation and properties

First, to guide the growth of key anatomical structures, we referred to cross-sectional anatomy of the Sprague-Dawley rat spinal cord (n = 10) and designed the scaffold structure using Solid Works charting software (Dassault Systemes, Suresnes Cedex, France). Four passes were set to provide guidance for the growth of gracile, cuneate fasciculus, corticospinal, and spinal cord thalamus tracts.

As described in our previous study, a 4:2 ratio of collagen/silk fibroin is a good choice (Xu et al., 2016). Thus, scaffolds were prepared at this ratio using a 3D bioprinter (Regenovo Bio Technology Co., Ltd., Hangzhou, China). Parameters of 3D bioprinting were set (210-µm diameter nozzle, 9-mm/s printing speed, 2-mm/min extrusion speed, 0.1-mm thickness, –20°C platform temperature). Collagen was blended with silk fibroin (quality ratio 4:2). Blended creatures were squeezed out through the print head into a casting mold (2-mm length) and rapidly solidified into a solid state. Scaffolds were collected and placed in a –20°C refrigerator for 24 hours, and then sterilized thoroughly by ⁶⁰Co (Jinpengyuan Irradiation Technology Co., Ltd., Tianjin, China) after freeze-drying treatment.

An in vivo degradation experiment was carried out. Briefly, after general anesthesia, rats were fixed on the operating table in the prone position. Skin of the thoracolumbar back was prepared and disinfected. Next, skin and subcutaneous tissue were cut along the midline of the back, and two saccular gaps were made where two 3D-CF scaffolds were implanted. After implantation, rats were sacrificed each week. The remaining scaffold material was washed with distilled water and dried under vacuum to a constant weight. The degradation rate of the scaffold was calculated as (W₀ –W_T)/W₀ × 100%, where W₀ represents mass of the scaffold material before degradation, and W_T represents mass of the residual scaffold material after degradation.

Compression stress, another mechanical property of scaffolds, was also examined. The 3D-CF was immersed in 0.01 M phosphate-buffered saline (PBS) solution at pH 7.4 for 24 hours at 37°C. Mechanical compression properties of scaffolds were tested on an Instron 5865 mechanical test machine (Instron, Norwood, MA, USA) with a 0.5-Hz sinusoidal waveform, preload of 0.1 N, and a maximum compressive strain of 60%. The stress-strain curve was displayed on the computer screen, and parameters such as compressive elastic modulus, compression displacement, and compressive strain of scaffolds were obtained. Mechanical properties of scaffolds (n = 3) were statistically analyzed.

Fourier transform infrared-attenuated total reflection was also examined. Briefly, 3D-CF was ground into a powder, sampled by KBr tableting, and measured on a Nicolet 870 infrared spectrometer (Nicolet, MA, USA) with a scan range of 400–4000/cm and resolution of 4/cm.

Culture and identification of NSCs by immunofluorescence staining

Sprague-Dawley rats at 14–15 days of gestational age were sacrificed. The cerebral cortex tissues of embryos were removed under sterile conditions, then cut and digested for 5 minutes with 0.25% trypsin (Solarbio Science & Technology Co., Ltd.) at 37°C. The supernatant was discarded after filtration and centrifugation at 157 × g at 4°C for 5 minutes. A total of 2 mL of 0.25% trypsin was added to obtain a single cell suspension. Cells were inoculated in culture flasks at a density of 5 × 10⁶/L. Serum-free Dulbecco’s Modified Eagle Medium (DMEM/F12; Thermo Fisher Scientific Co., Ltd., Shanghai, China) was added to cells, which were incubated at 37°C with 5% CO₂ for 7 days. According to the growth of cells, medium was replaced every 3 days. When the cells grew to occupy 80% or more of the culture flask, they were digested with 0.25% trypsin and subsequently sub-cultured. When cells were sub-cultured to the third generation, cell morphology was observed under a light microscope (DWI4000B, Leica, Germany). Next, cells were digested with 0.25% trypsin and inoculated evenly on coverslips pretreated with 4% poly-lysine (Sigma, St. Louis, MO, USA) at a density of 1 × 10⁵/mL. DMEM/F12 medium containing 10% fetal bovine serum (FBS; MRC Biotechnology Co., Ltd., Jiangsu, China) was added and coverslips were incubated at 37°C with 5% CO₂ for 1 day. Subsequently, well-adhered cells were fixed with paraformaldehyde (Aladdin Biotechnology Co., Ltd., Shanghai, China), permeabilized with PBS solution containing 5% Triton X-100, and incubated with 3% hydrogen peroxide for 15 minutes followed by 5% bovine serum albumin solution. A primary rabbit polyclonal antibody against nestin (1:1000; Chemicon, Temecula, CA, USA), a specific marker of NSCs, was added and incubated at 4°C overnight. The following day, a secondary goat polyclonal antibody (1:1000; Abcam, Cambridge, UK) and DAPI (fluorescent nuclear dye; Solarbio Science & Technology Co., Ltd.) were added and incubated at room temperature for 2 hours. Samples were observed under a fluorescence microscope (DMI4000B, Leica, Germany). In addition, cells were sub-cultured to the fourth generation in culture medium containing 10% FBS, B27, glutamine, penicillin, streptomycin, basic fibroblast growth factor, epidermal growth factor, and heparin (Cyagen, Jiangsu, China) to induce differentiation. Subsequently, cells were fixed with paraformaldehyde for 30 minutes and washed three times with PBS solution for 5 minutes each. A primary rabbit polyclonal antibody against microtubule-associated protein-2 (MAP2, specific marker of neurons) and mouse monoclonal antibody against β-Tubulin-III (specific marker of neurons; Abcam) were added at 1:100 and incubated overnight at 4°C. After a wash with PBS, secondary antibodies [goat polyclonal antibodies conjugated to FITC (1:100) or Cy3 (1:100); Abcam] and DAPI were added and incubated for 1 hour. After the removal of liquids, samples were observed under a fluorescence microscope (TCS SP5; Leica, Mannheim, Germany).

Observation of scaffold and NSC coculture, and MTT assay

A 100-μL volume of second-generation rat NSCs was seeded into the 3D-CF at a density of 10⁷ cells/mL in six-well plates (3.5-cm diameter). Numbers and growth state of cells were observed under an optical microscope 7 days after starting co-culture on the scaffold. Moreover, 3D-CF was examined by scanning electron microscopy (SEM; QUANTA200, FEI, Eindhoven, the Netherlands), with and without co-culture of NSCs. After 72 hours of co-culture, 2% glutaraldehyde (Aladdin Biotechnology Co., Ltd., Shanghai, China) was added to immobilized scaffolds co-cultured with NSCs. Scaffolds co-cultured with NSCs were then fixed with 1% osmium tetroxide, dehydrated with a gradient of acetone, rapidly frozen by liquid nitrogen, and dried at a critical point. After scaffolds co-cultured with NSCs were coated with gold, cell morphology on the scaffold was observed by SEM (Sigma 300, ZEISS, Germany). Cell compatibility of the 3D-CF scaffold was evaluated and scaffold ultrastructure was observed. Scaffolds without NSCs were also observed by SEM.

MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide, Solarbio Science & Technology Co., Ltd.) was applied to detect the proliferation of NSCs co-cultured with 3D-CF. The 3D-CF was cut into 5-mm × 5-mm specimens and cells were seeded on the scaffold for co-culture. The experiment was divided into co-culture and blank scaffold groups. For the co-culture group, each scaffold was inoculated with 100 μL of NSCs at 2 × 10⁷ cells/mL in 96-well plates (0.64-cm diameter), and placed in a 37°C incubator containing 5% CO₂ for 4 hours. After 1, 3, 5, and 7 days of co-culture, the cell-scaffold complex was transferred to a 5-mL centrifuge tube. Next 500 μL of MTT solution (2 g/L) was added to each centrifuge tube. After incubating for 4 hours at 37°C, each scaffold was transferred to a culture well containing 1 mL of dimethyl sulfoxide. After full dissolution was achieved, 150 μL of the solution was transferred to a 96-well plate. Optical density values of each well (representing a single scaffold) were measured at 490 nm. In the blank scaffold group, scaffolds were incubated without cells as a control. Considering experimental error, ten samples were set in each group to repeat the experiment.

Model construction of transected SCI and scaffold implantation

A total of 40 Sprague-Dawley rats were used for the following experiments. All animals were randomly divided into four groups: sham (n = 10; only laminectomy was performed), SCI (n = 10; SCI without any transplant), 3D-CF (n = 10; 3D scaffold transplant after SCI), and 3D-CF + NSCs (n = 10; 3D scaffold co-cultured with NSC transplant after SCI). Rats were anesthetized by intraperitoneal injection of 5% chloral hydrate (7 mL/kg). After shaving, the skin at T9, T10, and T11 levels was completely sterilized and incised along the midline. Vertebral bodies from T9–11 were exposed absolutely by blunt separation of paraspinal muscles. The surgical microscope was used to remove the T10 vertebral lamina. The spinal cord was exposed and a 1.5-mm segment of spinal cord was resected. Rats exhibited bilateral hind limb twitching, stretching of the back of the foot, and weak muscle relaxation, indicating successful model establishment. When bleeding was completely controlled, the damage cavity expanded to 2 mm in width because of tissue contraction. Immediately, scaffolds were transplanted into the injured cavities according to the anatomy of the transected spinal cord. Incisions were then sutured layer by layer and all rats were placed on a constant temperature blanket until recovery from anesthesia. Rats were subsequently sent back to cages and carefully monitored along with bladder massage until normal voiding function returned.

Behavioral observation

Basso-Beattie-Bresnahan (BBB) open-field locomotor scoring (Du et al., 2019) and an inclined plane test (Zhang et al., 2018) were used to assess behavioral changes at 1, 2, 3, 4, 6, and 8 weeks after surgery. For BBB scoring, rats were placed in an open environment and the hindlimbs were evaluated. Additionally, rats were placed on an inclined plate and the angle was gradually increased until an individual could not maintain balance on the plane for 5 seconds. Scores and angles were evaluated and recorded by two staff members who were blinded to experimental conditions.

Electrophysiological detection

Electrophysiological detection was conducted with Viking Quest (Thermo Nicolet Corporation, Madison, WI, USA) at 1 and 2 months after surgery. Motor evoked potential (MEP) detection was mainly performed under a constant voltage. Administration of general anesthesia was the same as model construction (5% chloral hydrate, 7 mL/kg, intraperitoneal injection), and then two needle-like electrodes were inserted into the projection area of the motor area as stimulating electrodes, while another two needle-like electrodes were inserted into the hindlimb muscles as recording electrodes. In addition, the rat tail was connected to a ground wire. As for parameters, the stimulating voltage was set at 50 V, pulse width at 0.2 ms, and stimulus frequency at 1 Hz. Apart from the typical peak of MEP, latency and amplitude of MEPs on bilateral hindlimbs were recorded.

Magnetic resonance imaging scan and diffusion tensor imaging construction

MRI was performed under general anesthesia 2 months after surgery using a Siemens MAGNETOM Verso 3.0T magnetic resonance imaging (MRI) system (A Tim system, Siemens, Munich, Germany). The main sequences were T2 and diffusion tensor imaging (DTI). T2 was utilized to observe the general condition of spinal cord regeneration and distance between the broken ends. DTI was applied to evaluate neurofilament connections. In T2 phase, repetition time was set to 4000 ms, with an echo time of 74 ms, slice gap of 0.15 mm, slice thickness of 1.5 mm, scan matrix of 224 × 320 seconds. Field of view read was set as 100 mm × 100 mm. In DTI, repetition time was set to 4800 ms, with an echo time of 120 ms, slice gap of 0.12 mm, slice thickness of 3 mm, and scan matrix of 128 × 128 seconds. Field of view read was set as 107 mm × 107 mm, with a b-value = 1000 s/mm².

Histology examination

All rats were sacrificed at 2 months after surgery. In total, 200 mL of normal saline was used to drain the blood, and 200 mL of 4% paraformaldehyde solution was used to fix the tissue. Subsequently, after removal of the spinal cord at T7–12, gross morphology was observed. Next, spinal cord sections were sliced longitudinally into 5-µm-thick slices on an automatic paraffin slice system (MEDITE, Burgdorf, Germany). Hematoxylin-eosin staining was conducted to observe damaged cavities and local tissue repair. Cavity area was calculated with ImageJ 1.4 software (NIH, Bethesda, MD, USA). Moreover, argentaffin staining was used to evaluate nerve fiber regeneration.

Immunofluorescence staining and western blot assay

Two months after surgery, slices were prepared as described for histological examination. To analyze axonal regeneration and glial scar formation of the spinal cord, neurofilament H (NF-H, a specific marker of nerve fiber and glial fibrillary acidic protein (GFAP, a specific marker of glial fibers) were adopted. Procedures for immunofluorescence staining were similar to those stated above for nestin staining to identify NSCs. Two months after surgery, western blot assay was performed according to standard procedures. In short, protein concentration was determined by bicinchoninic acid protein assay kit (Thermo Fisher Scientific Co.). Equal amounts (30 μg) of proteins were isolated with sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Solarbio Science & Technology Co., Ltd.) and then transferred to a polyvinylidene difluoride membrane (Millipore, USA). After blocking with 0.1% bovine serum albumin, the membrane was incubated with a primary rabbit monoclonal antibody against NF-H (1:1000; Abcam) and rabbit monoclonal antibody against GFAP (1:1000; Abcam) at 4°C overnight. The following day, membranes were incubated with a secondary antibody (goat anti-rabbit lgG; 1:2000; Invitrogen, Carlsbad, CA, USA) at 37°C for 50 minutes. A enhanced chemiluminescence kit (Solarbio Science & Technology Co., Ltd.) was used to complete exposure and development. Images were analyzed with ImageJ software, with the relative density of NF-H/GFAP represented by relative gray values.

Statistical analysis

Statistical analyses were performed with SPSS 22.0 software (IBM, Armonk, NY, USA). All data are presented as mean ± standard deviation (SD). Statistical comparisons of optical density values, behavioral scores, amplitude and latency of MEPs, cavity area, and relative gray values were conducted using two-sample t test, or one-way analysis of variance followed by Student-Newman-Keuls post hoc test and repeated measurement analysis of variance. A value of P < 0.05 was considered statistically significant.

Results

3D-bioprinted collagen/silk fibroin scaffolds exhibit favorable characterization

The 3D-CF was a porous scaffold after 3D printing and freeze-drying treatment (Figure 1A–C). Figure 1D shows degradation curves at 1, 2, 3, and 4 weeks after implantation of the 3D-CF. This 4:2 composite scaffold completely degraded in rats by 4 weeks post-implantation (within the observation period of 8 weeks).

Figure 1.

Characteristics of 3D-bioprinted collagen/silk fibroin scaffold.

(A) Cross-sectional anatomy of the Sprague-Dawley rat spinal cord and design of 3D-CF. (A1) Anatomy of the transected spinal cord; (A2, A3) design of the scaffold according to cross-sectional anatomy; (A4) structure of the scaffold fabricated by 3D-bioprinting. (B) General view and internal structure of 3D-CF under a scanning electron microscope. Scale bar: 25 µm. (C) General view and internal structure of 3D-CF under a scanning electron microscope. Original magnification: 1500×; scale bar, 10 μm. (D) Degradation rate of 3D-CF at 1, 2, 3, and 4 weeks after implantation. (E) Elastic moduli of 3D-CF. (F) Infrared spectrum detection of 3D-CF. 3D-CF: Three-dimensional collagen/silk fibroin scaffold.

The scaffold had ductility after immersion. Mechanical properties were tested by elastic modulus, compression displacement, and compressive strain. The compressive elastic modulus of the scaffold was 60.05 ± 5.12 kPa (Figure 1E), demonstrating that it had both good ductility and compression resistance.

From the Fourier transform infrared spectrum of the scaffold, the peak at 3445.7/cm might be a hydroxyl peak (O-H) or N-H peak, while peaks at 2932.46/cm and 2866.81/cm might represent a methyl group or a C-H stretching vibration of a methylene group. The absorption peak at 1640.58/cm might be the stretching vibration of C=O and C=C. The absorption peak at 1570.06/cm might be the stretching vibration of C=C. The strong absorption peak at 1376.45/cm allowed judgement of the saturated C-H bending vibration. Furthermore, a strong absorption peak of 1088.16 allowed judgement of the presence of C-O. These data demonstrated that the 3D-CF scaffold had suitable lipid-soluble and water-soluble chemical bonds, which were beneficial for adhesion and growth of nerve cells (Figure 1F).

NSCs exhibit strong proliferation and differentiation abilities

Under a light microscope, NSCs were observed to grow well (Figure 2A), while many nestin-positive cells were discerned by fluorescence microscopy (Figure 2B–D). In addition, images of MAP2-positive and tubulin-3-positive cells indicated that NSCs have a strong ability to differentiate into neurons (Figure 2E–H).

Figure 2.

Compatibility between scaffold and neural stem cells.

(A) Representative image of NSC morphology under a light microscope. (B) DAPI staining of NSCs. (C) Nestin-positive cells under a fluorescence microscope. (D) Merged image of DAPI staining and nestin-positive cells. (E) DAPI staining of NSCs. (F) MAP2-positive cells under a fluorescence microscope. (G) Tubulin-3-positive cells under a fluorescence microscope. (H) Merged image of DAPI staining and MAP2-positive and tubulin-3-positive cells. (I–J) Co-culture of 3D-CF and NSCs under a light microscope. (K) Co-culture of 3D-CF and NSCs under an electron microscope. Red dotted frame: A part of magnification. Scale bar: 25 µm. (L) Co-culture of 3D-CF and NSCs under an electron microscope. Yellow arrows: NSCs and their pseudopodia. Scale bars: 25 µm in A–H; Scale bars: 50 µm in I–J; Scale bars: 10 µm in K–L. (M) Comparison of MTT results between blank scaffold and co-culture groups. *P < 0.05, **P < 0.01, vs. blank scaffold group (mean ± SD, n = 10, two-sample t test). 3D-CF: Three-dimensional collagen/silk fibroin scaffold; DAPI: 4′,6-diamidino-2-phenylindole; MAP2: microtubule-associated protein 2; MTT: 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide; NSCs: neural stem cells.

3D-bioprinted collagen/silk fibroin scaffolds possess good biocompatibility

NSCs could adhere, survive, and grow on the surface and pores of the 3D-CF scaffold, demonstrating that the 3D-CF scaffold had good biocompatibility (Figure 2I and J). When observed by SEM, the inside and the surface of the scaffold had a microscopic porous structure, with evenly distributed micropores and pores connected to each other (Figure 1B and C). Many NSCs grew in the scaffold pores, and cells were spherical or differentiated into a fusiform shape. NSCs grew densely on the surface of scaffolds. Some cells extended pseudopods to attach to the surface of scaffolds, which were suitable for cell adhesion, growth, and provided a carrier and channel for regenerating nerve fibers. In addition to four irregular beam-shaped channels that simulate the nerve conduction beam, the 3D-CF itself had good connectivity and uniformity. The NSCs grew well on the surface of the scaffold material and were fully stretched (Figure 2K and L). These results indicated that 3D-CF could provide a suitable microenvironment for NSC adhesion, elongation, and differentiation, and had good biocompatibility.

Figure 2M shows NSC proliferation on 3D-CF by MTT assay. After cell seeding, optical density values of the co-culture group were significantly increased compared with the blank scaffold group, indicating that NSCs on the scaffold were successfully seeded and proliferated. By analyzing cell proliferation during co-culture for different time periods, we observed that the difference between the 1-day and 3-day cultures was small. It might be that cells grow slowly at the beginning, but begin to grow faster after 5 days. Notably, values for the blank scaffold group were not significantly different, indicating that scaffolds had little effect on MTT assay results. Thus, the results of our experiments accurately reflected cell proliferation in scaffolds.

3D-bioprinted collagen/silk fibroin scaffolds promote recovery of locomotor function

After model establishment and scaffold implantation at the injury site (Figure 3), behavioral changes were assessed. With regard to the two scoring scales shown in Figure 4A and 4B, laminectomy did not affect neurological function (sham group). BBB scores were almost 0 during the first week post-surgery, but motor function recovery of hindlimbs was observed over time in each group. Notably, the recovery speed of the SCI group was obviously slower than that of 3D-CF and 3D-CF + NSCs groups. Two key points were selected for statistical analysis. At 4 weeks after surgery, BBB scores were still fairly low among the three groups compared with the sham group (P < 0.01). However, scores for 3D-CF + NSCs and 3D-CF groups were both higher than the SCI group (P < 0.05). At 8 weeks after surgery, scores in the three groups were still significantly different from the sham group (P < 0.01). However, scores for 3D-CF + NSCs and 3D-CF groups were increased compared with the SCI group (P < 0.01, P < 0.05). In the inclined plane test, motor function recovery was also observed over time in each group, but trends of angle change in 3D-CF + NSCs and 3D-CF groups were better than the SCI group. At 4 weeks after surgery, angles of the three groups were significantly lower than the sham group (P < 0.01). However, angles of 3D-CF + NSCs and 3D-CF groups were both higher than those of the SCI group (P < 0.01, P < 0.05). Angles of the 3D-CF + NSCs group were higher than those of the 3D-CF group (P < 0.05). At 8 weeks after surgery, gaps between the three groups and the sham group were still significant (P < 0.01). However, angles of 3D-CF + NSCs and 3D-CF groups were both higher than those of the SCI group (P < 0.01). Moreover, angles of the 3D-CF + NSCs group were higher than those of the 3D-CF group (P < 0.01).

Figure 3.

Model establishment and scaffold implantation at the injury site.

(A) Diagram of the transplantation process. (B) Exposure of T9–11 vertebral lamina after tissue separation. Yellow dotted frame: Location of spinal cord segments; yellow arrow: location of T10. (C) Exposure of the spinal cord after laminectomy in yellow dotted frame. (D) Resection of a 2-mm segment of the spinal cord. Yellow dotted frame: Spinal cord at T10; blue dotted frame: injured cavity. (E) 3D-CF implantation into the injured cavity. Yellow dotted frame: Spinal cord at T10; blue dotted frame: injured cavity filled with scaffold or 3D-CF. 3D-CF: Three-dimensional collagen/silk fibroin scaffold.

Figure 4.

Detection of neurological function recovery in vivo.

(A, B) BBB scores and inclined plane test results of rats in the four groups at 1, 2, 3, 4, 6, and 8 wk (weeks) after surgery. (C) Typical images of hindlimb motor evoked potentials in the four groups at 1 and 2 mon (months) after surgery. (D–G) Statistical analysis of the amplitude and latency of motor evoked potentials in the four groups. (H) Representative images of MRI and DTI construction in the four groups at 2 months after surgery. Red dotted frame: Transection of spinal cord. Yellow arrows: Distance between two ends of the transected spinal cord. §P < 0.05, §§P < 0.01, vs. sham group; #P < 0.05, ##P < 0.01, vs. SCI group; +P < 0.05, ++P < 0.01, vs. 3D-CF group (mean ± SD, n = 10; one-way analysis of variance followed by Student-Newman-Keuls post hoc test). a: Sham group, b: SCI group, c: 3D-CF group, d: 3D-CF + NSCs group. 3D-CF: Three-dimensional collagen/silk fibroin scaffold; BBB: Basso-Beattie-Bresnahan; DTI: diffusion tensor imaging; MRI: magnetic resonance imaging; NSCs: neural stem cells; SCI: spinal cord injury.

Similar to behavioral scoring scales, electrophysiological results indicated that no effects were produced when only laminectomy was performed. The representative MEP peak nearly disappeared in the SCI group at both 1 and 2 months after surgery. In 3D-CF and 3D-CF + NSCs groups, the MEP peak regained more or less as time proceeded. Recovery was especially prominent in the 3D-CF + NSCs group (Figure 4C). Statistical comparisons of latency and amplitude at different time points were made between groups (Figure 4D–G). At 1 month after surgery, left hindlimb amplitude was obviously lower in SCI (P < 0.01), 3D-CF (P < 0.01), and 3D-CF + NSCs (P < 0.05) groups compared with the sham group. Moreover, amplitude of the SCI group was lower than 3D-CF (P < 0.05) and 3D-CF + NSCs (P < 0.01) groups. Amplitude of the 3D-CF + NSCs group was higher than the 3D-CF group (P < 0.05). Statistical differences also appeared for the right hindlimb. Compared with the sham group, amplitude was obviously decreased in SCI (P < 0.01), 3D-CF (P < 0.01), and 3D-CF+NSCs (P < 0.05) groups. Amplitude of the SCI group was lower than 3D-CF (P < 0.05) and 3D-CF + NSCs (P < 0.01) groups. Amplitude of the 3D-CF + NSCs group was higher than the 3D-CF group (P < 0.05; Figure 4D). Similar statistical differences were also reflected in 1-month latency (Figure 4E), 2-month amplitude (Figure 4F), and 2-month latency (Figure 4G). The above results indicated that 3D-CF improved motor function from the perspectives of neurological scoring and electrophysiology, consistent with MRI and DTI results.

In T2 sequence, the spinal cord of the sham group was intact, while broken ends of the spinal cord in the SCI group were far apart. The gap between broken ends was distinctly reduced in the 3D-CF group, while broken ends were almost connected in the 3D-CF + NSCs group. In DTI sequence, integrity of the spinal cord in the sham group was maintained. No nerve fiber was observed crossing the gap between broken ends in the SCI group at 2 months post-surgery. However, a few fibers were observed sprouting from the lower part and connecting with the other part in the 3D-CF group at 2 months post-surgery. Surprisingly, a bundle of nerve fibers came out from the lower part and formed stable connections with the other part in the 3D-CF + NSCs group (Figure 4H). These results demonstrated that 3D-CF increased the joining of broken ends at the injury site.

3D-bioprinted collagen/silk fibroin scaffolds fill the cavity at the injury site and boost regeneration of nerve fibers

First, from the perspective of general observation, the spinal cord was intact in the sham group (Figure 5A). Tissue in the gap was very slender and brittle in the SCI group (Figure 5B), while tissue was strong in the 3D-CF group (Figure 5C). Notably, in the 3D-CF + NSCs group, the damaged cavity was completely filled and the texture was similar to the original common tissue (Figure 5D). Second, outcomes of hematoxylin-eosin staining illustrated the advantages of 3D-CF + NSCs in promoting damage repair. Except for the sham group (Figure 5E), spinal cord tissue gradually became complete Figure 5F–H. Furthermore, the cavity area was significantly larger in the SCI group compared with 3D-CF (P < 0.01) and 3D-CF + NSCs (P < 0.01) groups. Moreover, the area was larger in the 3D-CF group compared with the 3D-CF + NSCs group (P < 0.05; Figure 5I). Third, outcomes of argentaffin staining revealed the advantages of 3D-CF + NSCs in accelerating neural regeneration. Briefly, the spinal cord in the sham group was intact and nerve fibers were neatly arranged (Figure 5J). The damaged cavity was still very large and the edge of the tissue was messy with only a few nerve fibers in the SCI group (Figure 5K). However, the damaged cavity was filled to some extent in the 3D-CF group and some nerve fibers were observed (Figure 5L). Remarkably, in the 3D-CF + NSCs group, tissue regeneration was relatively complete, many nerve fibers were observed in the neoplasm, and spinal cord tissue appeared similar to the sham group (Figure 5M).

Figure 5.

Histological examinations of the graft site at 2 months after surgery.

(A–D) General views of spinal cord tissue at the injury site in the four groups. (A1–D1) Magnification of red dotted frames in A–D. (E–H) Hematoxylin-eosin staining of spinal cord tissue. Original magnification: 50×. (E1–H1) Magnification of red dotted frames in E–H. Scale bar: 50 µm. (I) Statistical analysis of the cavity area in the four groups. (J–M) Argentaffin staining of spinal cord tissue in the four groups. Original magnification: 50×. (J1–M1) Magnification of red dotted frames in J–M. Scale bar: 50 µm. A, E, J: Sham group; B, F, K: SCI group; C, G, L: 3D-CF group; D, H, M: 3D-CF + NSCs group. *P < 0.05, **P < 0.01 (mean ± SD, n = 10; one-way analysis of variance followed by Student-Newman-Keuls post hoc test). 3D-CF: Three-dimensional collagen/silk fibroin scaffold; NSCs: neural stem cells; SCI: spinal cord injury.

NF-H and GFAP were used to represent axonal regeneration and glial scar formation in the spinal cord. In the SCI group, there were very few NF-positive nerve fibers in the visual field (Figure 6A), while GFAP-positive glial scarring was abundant (Figure 6B). In the 3D-CF + NSCs group, many NF-H-positive nerve fibers were observed in the field (Figure 6E), while GFAP-positive glial scarring was extremely rare (Figure 6F). The number of NF-H-positive and GFAP-positive cells in the 3D-CF group was in between that observed in SCI and 3D-CF + MSCs groups (Figure 6C and D). Similar trends were observed by western blot assay (Figure 6G and H). Relative density of NF-H in the SCI group was lower than observed in 3D-CF (P < 0.05) and 3D-CF + NSCs (P < 0.01) groups. Relative NF-H density was higher in the 3D-CF + NSCs group compared with the 3D-CF group (P < 0.05; Figure 6I). Relative density of GFAP was higher in the SCI group compared with 3D-CF (P < 0.05) and 3D-CF + NSCs (P < 0.01) groups. Relative GFAP density was lower in the 3D-CF + NSCs group compared with the 3D-CF group (P < 0.01; Figure 6J). These results suggested that 3D-CF could help fill the injured cavity and promote nerve fiber regeneration, while inhibiting glial scar formation.

Figure 6.

Immunofluorescence staining and western blot assay of the injury site.

(A, C, E) Immunofluorescence staining of NF-H-positive nerve fibers in SCI (A), 3D-CF (C), and 3D-CF + NSCs (E) groups. (B, D, F) Immunofluorescence staining of GFAP-positive glia scars in SCI (B), 3D-CF (D), and 3D-CF + NSCs (F) groups. (A1–F1) Magnification of yellow dotted frames in A–F. Scale bar: 50 µm. (G–J) Western blot assay of NF (G, I) and GFAP (H, J) protein expression in the three groups, and respective statistical analysis. *P < 0.05, **P < 0.01 (mean ± SD, n = 10; one-way analysis of variance followed by Student-Newman-Keuls post hoc test). b: SCI group, c: 3D-CF group, d: 3D-CF + NSCs group. 3D-CF: Three-dimensional collagen/silk fibroin scaffold; GFAP: glial fibrillary acidic protein; NF-H: Neurofilament-H; NSCs: neural stem cells; SCI: spinal cord injury.

Discussion

The treatment of SCI remains a huge challenge. The main reason is that the central nervous system of human beings exhibits only weak regeneration ability (Wang et al., 2018). Thus, severe contusion and transverse injury result in an especially poor prognosis (Wei et al., 2019). At present, there are few effective interventions for acute or chronic SCI, although many organizations have made gratifying results in basic and clinical research (Wang et al., 2018).

SCI has its distinct characteristics in acute and chronic phases (Sharma et al., 2019). Although conditions are relatively stable in the chronic phase, a large number of local neurons have atrophied and died, and severe glial scar hyperplasia (Ruschel et al., 2015) accompanied by local chronic inflammation and nutritional deficiency form physical and chemical barriers, which may profoundly inhibit nerve regeneration (Fitch and Silver, 2008; Kadoya et al., 2009). During the acute stage, local tissue edema can exist, but the majority of the neurons remain intact; thus, there is hope for rescue if intervention is conducted during the acute stage of SCI (Höller et al., 2017).

Scaffold materials have been a major focus in the field of biological tissue engineering. Many studies have demonstrated the quality and function of bio-scaffolds in delivering cells and drugs to the central nervous system (Krishna et al., 2013; Führmann et al., 2017). In studies of SCI repair, materials such as polymers, acellular tissue matrix, and nanomaterials have their own unique advantages, but there remain deficiencies in standardization. As such, there is still a long way to go for clinical translation. The ideal spinal scaffold material needs to have the following characteristics: favorable biocompatibility, specific mechanical strength, suitable degradation rate, and proper aperture. Specifically, the scaffold should retain cells at the lesion site, support the physical matrix, fill the lesion cavity, regulate the local microenvironment, and mediate directed growth (Atala, 2000; Madigan et al., 2009; Cao et al., 2011; Luo et al., 2016; Ogle et al., 2016).

After SCI, the axon of the distal injured area can reach the other end through axonal regeneration in a proper environment. However, there are massive obstacles that can restrain nerve regeneration. A prohibitive microenvironment is one of the most fatal factors for axonal regeneration (Ahuja et al., 2017). When SCI occurs, the injury site may be flooded with inflammatory factors, oxygen free radicals, inhibitory amino acids, and neurotransmitters (Shang et al., 2017). These chemicals can easily lead to toxic reactions and exacerbate local edema, resulting in insufficient blood supply and neurotrophic substance aggregation. In addition, loss of spinal cord tissue and hypertrophic glial scar formation may also physically restrict axonal regeneration. Thus, breaking these chemical and physical barriers is of great significance for SCI healing (Straley et al., 2010; Gilbert et al., 2011). The scaffold fabricated by 3D bioprinting is a favorable substrate to provide channels and guide newborn axons; moreover, this porous scaffold can supply a good microenvironment for NSC survival (Li and Dai, 2018). Indeed, after co-transplantation of scaffold and NSCs, the microenvironment could be greatly improved and neural function may be rescued.

Porous scaffolds made by conventional manufacturing technology cannot provide precise structures for the nerve bundle (Murphy and Atala, 2014). In this study, we simulated the electrophysiological spatial structure of the rat spinal cord and designed corresponding physiological holes to guide axonal abutment of the same region (The design idea has been transferred to a patent: CN105380728A). This study combined the advantages of collagen and silk fibroin, yielding 3D-CF rich in pores (10–240 μm in diameter). In addition, the degradation time is moderate, which not only can meet the requirements of experimental observation, but also promotes axons to pass through the injury site and prevents the formation of physical barriers (Sakiyama-Elbert et al., 2012). Good compressive strength and ductility make it stable. Moreover, the large number of effective chemical bonds are suitable for cell adhesion and growth. An MTT assay revealed the excellent biocompatibility of 3D-CF. Indeed, a large number of NSCs grew in the pores and extended many projections that interacted with each other. Behavioral observation was a relatively subjective evaluation, but rats in the 3D-CF + NSCs group exhibited more obvious improvements in limb function. In in vitro experiments, the new locally organized structure was arranged in an orderly manner. In the 3D-CF + NSCs group, substantial NF-H-positive cells and very few GFAP-positive cells were observed, demonstrating that the newly formed cells in the injured sitewas bioactive.

Spontaneous neural regeneration and functional recovery after SCI are difficult. Moreover, repair of the anatomical structure does not mean improvement of neurological function. However, the irreversibility of nerve defects is no longer a mainstream view. Indeed, greater understanding of injury mechanisms and pathophysiological characteristics has highlighted the enormous potential of neural regeneration.

In previous studies, scaffolds for SCI repair have not been biomimetically constructed according to the anatomical structure of the spinal cord (Liu et al., 2017, 2019). In this respect, the current study has achieved a breakthrough. The scaffolds fabricated by 3D printing provided a structural basis for neural network reconstruction of spinal cord white matter, which not only inhibited the growth of glial scars, but also provided a good direction for axon pathfinding during SCI repair.

However, this study also has some limitations, such as the lack of NSC labeling and tracing (Kirschen et al., 2018). Moreover, as there remain some deficiencies in the assessment of neurological function, additional objective assessment methods should be applied, such as motion capture technology (Jiang et al., 2018). In addition, tracing of corticospinal tracts with biotin dextran amine (Liu et al., 2017; Steward and Willenberg, 2017) would make the results of this study more complete and convincing. Further, we may also investigate alternatives for NSCs, such as exosomes from stem cells, in future investigations.

Additional file: Open peer review report 1 ^{(107.8KB, pdf)} .