SHED were obtained from donors. The morphology of cells was irregular, mostly spindle or polygonal shapes. The cell bodies were small, the cytoplasm was abundant and transparent, and there were almost no protrusions (Figure 1B). The identity of SHED was confirmed by evaluating expression of specific marker proteins by flow cytometry. Percentages of cells expressing positive markers CD44, CD73, CD90, and CD105 were 99.57%, 97.27%, 99.54%, and 99.91%, respectively. Furthermore, the percentage of cells expressing the negative marker human leukocyte antigen was 0.10%, indicating successful isolation of SHED (Figure 1A). In a colony-formation assay, SHED proliferated after 10 days and cell clone formation was visible under light microscopy. After crystalline violet staining, multiple cell colonies were visible (Figure 1C). Following induction of osteogenic differentiation of SHED for 21 days, calcium nodule deposition was visible and Alizarin Red staining was positive (Figure 1D). Following induction of lipogenic differentiation of SHED for 28 days, lipid droplets were observed in the cytoplasm, which stained positive for Oil Red O (Figure 1E). In summary, the extracted cells were SHED with the potential for unlimited proliferation, osteogenic differentiation, and lipogenic differentiation.

To optimize the formation of SHED sheet, we assayed the effect of varying AA concentrations on the proliferation of SHED. The results show that 240 μg/mL AA obviously inhibited proliferation, and while 120 μg/mL AA promoted cell proliferation and facilitated cell sheet formation. Thus, we chose 120 μg/mL AA for cell sheet construction (p < 0.001, Figure 1F). After SHED were cultured with complete medium containing AA for 1 week, SHED sheet was incubated with homogenate protein of spinal cord for 1 week to differentiate into neural cells. Finally, hp-SHED sheet peeled off the culture dish and self-assembled into a spinal cord-like shape, forming the bioactive filling material (Figure 1G).

First, hp-SHED sheet can serve as a biological scaffold, providing a structural basis for cell growth, differentiation, migration, and axonal extension. Cross and vertical sections of hp-SHED sheet revealed a highly porous three-dimensional inner structure, which facilitates nerve cell attachment, migration, and nerve axon regeneration and extension (Figure 1H). In addition, our results show that both collagens Ⅰ and Ⅲ were positively expressed in hp-SHED sheet. Cells in the hp-SHED sheet were tightly arranged, had a “honeycomb” structure, and regular ECM distribution (Figure 1I). High expression of collagen provides stability to the ECM microenvironment, enhancing intercellular interactions within hp-SHED sheet. Stable ECM is essential for sustained and stable cytokines and neurotrophins production in hp-SHED sheet.

To systematically and comprehensively explore the biological functions of hp-SHED sheet, 20 rats were divided into four groups: Sham, Control (Ctrl), SHED, and hp-SHED sheet (n = 5 per group). The spinal cord at T9–T10 was cut in rats to establish a complete SCI model. After modeling, the lesion cavity was filled with SHED or hp-SHED sheet in experimental groups, and PBS in the Ctrl group (Supplementary Figure S1). During the 60-day recovery period, all rats were subjected to body weight measurements, Von Frey testing, BBB scoring and grip strength test every 20 days. Before surgery, rats in each group were weighed. After surgery, rats lost weight because of the effects of spinal cord trauma. However, several days later, the weight of rats started to recover and gradually increased. Weight recovery was faster and more evident in the hp-SHED sheet group, followed by the SHED group, and poorest in the Ctrl group. However, differences between groups were not statistically significant. The rats in the hp-SHED sheet group recovered better and had better appetite, but with an increase in locomotion. The rats in the Ctrl group recovered less and had a poor appetite, but with less locomotion due to poor recovery of the hind limbs. The SHED group was in between. Eating and energy expenditure were in balance, so there was no significant difference in body weight (Figure 2A).

Von Frey filaments were used to detect the sensory recovery of rat hindlimbs. On the second day after surgery, all three groups of SCI rats had complete sensory deficits and negative hindlimb sensory ratings; whereas, rats in the Sham group had sensitive hindlimb sensation, with all rats responding positively. Twenty days after surgery, 20% of Ctrl rats recovered sensation, 40% of SHED rats recovered sensation, and 80% hp-SHED sheet rats recovered sensation (Figure 2B). Thus, the hp-SHED sheet group had a higher proportion of rats with sensory recovery compared with the SHED group, was both were significantly higher than Ctrl group. Rats in the hp-SHED sheet group recovered earlier and more completely.

The recovery of sensation was always accompanied by the recovery of motor ability. At 60 days, average grip strengths were 107.80 g ± 14.81 g in the Ctrl group, 173.00 g ± 16.70 g in the SHED group, and 235.40 g ± 27.93 g in the hp-SHED sheet group (p < 0.05, Figure 2C). At 60 days, average BBB locomotor scores were 3.20 ± 0.84 in the Ctrl group, 6.40 ± 1.14 in the SHED group, and 8.20 ± 0.84 in the hp-SHED sheet group (p < 0.001, Figure 2D). After 60 days of treatment, Ctrl rats were still unable to crawl on their hind legs and had only slight movement of the hindlimb joints. While SHED rats showed extensive movement of the three joints in the hindlimbs. Hp-SHED sheet rats could land on the plantar placement of the paw with no weight support (Figure 2E; Supplementary Video S1). Compared with the Ctrl group, the joint movement of both hp-SHED sheet and SHED groups showed significant improvement, although the treatment effect of hp-SHED sheet was more significant. In conclusion, hp-SHED sheet produced more significant neurorestorative effects in terms of both sensory and joint movement recovery.

We analyzed the recovery of spontaneous urination in rats from histological characteristics of the bladder. Compared with the Sham group, the Ctrl group had overly enlarged bladders due to urinary retention and loss of medial bladder folds, along with abnormally large bladder volumes and increased bladder weights. In contrast, following treatment with SHED or hp-SHED sheet, there were significant improvements in various aspects of bladder size, fold texture, bladder volume, and bladder weight. Notably, for all these observations, hp-SHED sheet was more effective (p < 0.05, Figures 3A–C). The analysis of bladder mucosal folds, weight, and capacity revealed that both hp-SHED sheet and SHED had therapeutic effects compared with the Ctrl group, although bladders in the hp-SHED sheet group were more like the Sham group. These results demonstrate that rats in the hp-SHED sheet group recovered spontaneous urination function earlier.

After spinal cord sampling at 60 days postoperatively, it was found that the recovery at the spinal cord lesion differed significantly between the groups (Figure 3D). To compare groups, pathological analysis of the spinal cord was performed. Specifically, the spinal cord lesion, as well as rostral and caudal stumps, were stained for markers of neurons (NF), astrocytes (GFAP), myelin (MBP), and regenerating peripheral nerves (CGRP) to detect neuronal regeneration, glial scar production, and neuronal axonal remyelination. The results revealed an absence of NF staining at the site of SCI in Ctrl rats, and there were obvious voids at the damage site. In SHED rats, there was a small amount of positive NF staining at the damage site. In hp-SHED sheet rats, a large amount of NF positive staining was clearly observable at the injury site, and there was no obvious cavity at the damage site (Figure 3E). Quantification of NF-positive areas at the lesion site shows that the hp-SHED sheet group had the largest area of positive neuron staining (3.43% ± 0.04%) compared to the Ctrl (0.21% ± 0.17%) and SHED groups (1.52% ± 0.16%), indicating that more neurons regenerated, matured, and formed nerve fibers in the SCI lesion this group (p < 0.01, Figure 3I).

High numbers of astrocytes promote glial scar deposition, which in turn prevents nerve fiber growth into and extension of nerve axons. To detect astrocyte formation, GFAP staining was performed. In Ctrl rats, the least GFAP staining was observed at the SCI lesion site because of the massive cell death (0.26% ± 0.08%) (Figure 3J). In SHED rats, positive GFAP staining was increased at the damage site as treatment inhibits cell death. In contrast, only weak GFAP-positive staining was observed in hp-SHED sheet rats because the treatment allowed more cells to survive and to differentiate towards non-astrocyte (Figure 3E). Quantification of GFAP-positive areas at the lesion site shows that hp-SHED sheet rats had less GFAP staining (1.15% ± 0.12%) than SHED rats (2.46% ± 0.31%) (Figure 3J). As shown by percentages of GFAP-positive cells, at the rostral and caudal site, astrocyte formation and glial scar deposition were reduced in SHED and hp-SHED sheet rats compared with Ctrl rats, with hp-SHED sheet having the most evident inhibitory effect. At the lesion site, besides the massive cell death in Ctrl rats, hp-SHED sheet rats effectively inhibited astrocyte production compared to SHED rats. Thus, hp-SHED sheet reduced glial scar deposition to mitigate its inhibitory effect on nerve regeneration by decreasing astrocyte production.

During the early stages of SCI, axonal demyelination occurs. Regeneration of myelin can be determined by assessing MBP expression. We did not observe MBP staining in spinal cords of Ctrl rats, while SHED rats displayed small amounts of MBP-positive staining. In hp-SHED sheet rats, large amounts of MBP-positive staining could be observed (Figure 3F). Quantification of MBP-positive areas at the lesion site shows that myelin-positive areas were significantly larger in SHED rats (2.22% ± 0.09%) than Ctrl rats (0.51% ± 0.25%), and significantly higher in hp-SHED sheet rats (3.99% ± 0.28%) than SHED rats, with hp-SHED sheet rats exhibiting the densest myelin structures (p < 0.01, Figure 3K).

CGRP is a marker of nerve regeneration, especially early nerve regeneration, that is often used as a rating index for spinal cord repair and regeneration. Our results show that Ctrl rats had almost no CGRP-positive cells, while CGRP-positive staining was increased in the SHED group and highest in the hp-SHED sheet group (Figure 3G). When areas of CGRP-positive staining in each group were quantified, the hp-SHED sheet group had the highest CGRP expression (4.18% ± 0.49%) compared to the Ctrl (0.25% ± 0.15%) and SHED groups (2.98% ± 0.92%) at the lesion site (Figure 3L). Increased numbers of CGRP-positive neurons in hp-SHED sheet rats suggests that hp-SHED sheet can promote neural regeneration earlier and better.

NeuN is a neuron-specific nuclear protein. NeuN staining was performed on the spinal cord, and it was found that there were a large number of neurons at the lesion in the hp-SHED sheet group. In SHED group, there were a few neurons in the lesion. The Ctrl group saw almost no neurons (Supplementary Figure S2A). When areas of NeuN-positive staining in each group were quantified, the hp-SHED sheet group had the highest NeuN expression (2.96% ± 0.69%) compared to the Ctrl (1.01% ± 0.22%) and SHED groups (1.79% ± 0.67%) at the lesion site (Supplementary Figure S2B). IBA1, a 143 amino acid cytoplasmic, inflammation response scaffold protein, is the microglia-specific marker. We stained the spinal cord for IBA1 and found a small distribution of microglia at the lesion in the hp-SHED sheet group. There were more microglia at the lesion in the SHED group. In the Ctrl group, microglial cell expression varied greatly, with some rats having significant microglia expression at the spinal cord lesions, while others had almost no positive expression due to massive cell death (Supplementary Figure S2A). When the Iba1-positive staining area was quantified for each group, the hp-SHED sheet group had the lowest Iba1 expression at the lesion site (0.68% ± 0.28%) compared to the Ctrl group (1.42% ± 1.14%) and the SHED group (1.89% ± 0.29%) (Supplementary Figure S2C).

Persistent inflammatory cell infiltration after SCI leads to further cell death and destruction of cellular and extracellular structures, resulting in the formation of cystic cavities at the lesion. After 60 days of treatment in SCI rats, spinal cord lesions displayed different degrees of cavities among the three groups. Quantification of these cavity areas reveals that hp-SHED sheet rats had the smallest average cavity area (15.49% ± 4.09%), followed by SHED rats (24.64% ± 4.58%) and Ctrl rats (38.23% ± 4.37%) which had the largest cavity area (Figure 3H). Following hp-SHED sheet treatment, more neurons, Schwann cells, and oligodendrocytes regenerated, and regenerated nerve fibers and nerve axons could extend to the lesion area, reducing the cystic area of the spinal cord lesion and promoting reconstruction of spinal cord neural circuits.

To further explore the potential of this bioactive filling material for clinical applications, we examined potential histological changes of several major organs of rats after transplantation. As expected, histopathological sections of the liver, spleen, lung, and kidney showed no abnormalities by HE staining, indicating that hp-SHED sheet and SHED are non-immunotoxic and preliminarily safe as human filler materials (Figure 4).

SHED were obtained from exfoliated deciduous teeth of clinical patients, and teeth were repeatedly flushed with phosphate-buffered saline (PBS). The pulp was removed under aseptic conditions, mechanically sheared, and then digested with type I collagenase (3 mg/mL; Sigma-Aldrich, St. Louis, MO, United States) for 1 h at 37°C to prepare a single-cell suspension. After centrifugation, cells were resuspended in α-Minimum Essential Media (Gibco, Grand Island, NY, United States) containing 10% fetal bovine serum (Hyclone, Logan, UT, United States) and 1% penicillin/streptomycin (Invitrogen, Carlsbad, CA, United States). Cells were incubated in T75 cm² flasks, allowed to grow to 80%–90% confluence, and then passaged normally.

Cells in logarithmic growth phase were removed and digested with trypsin into single-cell suspension. Adjust the cell concentration to 1×10⁷ cells/mL. Prepare 6 EP tubes and add 0.1 mL of cell suspensions to each tube. Under light-proof conditions, 1 μl of antibodies were added to each EP tube, Anti-Human CD73 PE, Anti-Human CD105 (Endoglin) PE, Anti-Human/Mouse CD44 APC, Anti-Human CD90 (Thy-1) PE, Anti-Human HLA-DR FITC (BioGems, United States), control group plus PBS. Mix well and incubate for 30 min at 4°C away from light. Vortex shaking was removed every 10 min for 5 s to bring the cells into full contact with the antibody. At the end of incubation, the cells were washed 2–3 times with PBS. Then 0.5 mL of PBS was added to resuspend the cells, filtered through a copper mesh and assayed by flow cytometry (Agilent, United States).

SHED in single-cell suspension were seeded into six-well plates at a density of 200 cells per well. Then the cells were cultured in α-MEM at 37°C. The medium was changed every 3 days. The cells were cultured for 10 days. Then the colonies were washed by PBS and fixed in 4% paraformaldehyde (PFA) solution 10 min. After washed by PBS the cells were incubated in 0.1% crystal violet for 20 min (Beyotime, China). Then colonies were observed. Finally, calculate the number of colonies under an inverted light microscope (Leica, Germany).

SHED were seeded into 0.1% gelatin-coated six-well plates at a density of 2×10⁴ cells/cm². The cells were cultured in α-MEM at 37°C. When the degree of cell fusion reached 70%, the cells were cultured with osteogenic induction medium (OriCell^®, China). Cells in control group were cultured in α-MEM. The medium was changed every 3 days. After 21 days of induction, alizarin red was used for staining. The osteogenesis staining was observed under the inverted light microscope.

SHED were seeded into 0.1% gelatin-coated six-well plates at a density of 2×10⁴ cells/cm². The cells were cultured in α-MEM at 37°C. When the degree of cell fusion reached 100%, the cells were cultured with adipogenic induction medium A (OriCell^®, China). The medium was changed to adipogenic induction medium B after 3 days. After 1 day of maintenance, solution B was replaced with solution A. Solution A and solution B are used alternately as above. Cells in control group were cultured in α-MEM. After 28 days of induction, Oil Red O solution was used for staining. The lipid-forming staining was observed under the inverted light microscope.

Cell proliferation was detected with the CCK-8 kit (DOjinDO, Japan). Each 96-well plate was divided into seven groups, with six replicate wells in each group. In six groups, each well was inoculated with 1000 cells in a volume of 200 μL of culture medium with ascorbic acid (AA) concentrations of 0, 15, 30, 60, 120, and 240 μg/mL. The other group was not inoculated with cells, and only the culture medium was added as a blank control. Cells were incubated in a CO₂ incubator at 37°C. A 96-well plate was removed at 0 h, 24 h, 48 h, 72 h, 96 h and 120 h, and CCK-8 solution (20 μL/well) was added to the culture medium. After incubation for 2 h in the incubator at 37°C, the absorbance was measured at 450 nm.

Homogenate proteins of the spinal cord were harvested from Sprague-Dawley rats (aged 8–10 weeks) that had been anesthetized. Briefly, spinal cords at the T8–10 levels were resected, ground, filtered (200 mm), disrupted with an ultrasonic homogenizer, and finally centrifuged at 15,000 g for 20 min. The supernatant was collected as the protein homogenate and the protein concentration was measured. The concentration of total protein used here was 1 mg/mL.

SHED were cultured in medium containing ascorbic acid (AA), which was changed every 2–3 days. A cell sheet was formed after 1 week of culture. Then, the culture medium was changed to α-MEM containing homogenate proteins of spinal cord for 1 week. Finally, the cell sheet at the bottom of the dish was gently peeled off and curled by self-assembly into a bioactive filler shaped like the spinal cord. This filling material, termed hp-SHED sheet.

The animal study was reviewed and approved by the Animal Studies Committee of Guangxi University. Twenty female Sprague-Dawley rats, aged 8–10 weeks and weighing 180–230 g, were purchased from the Laboratory Animal Center. In this exploratory study, we utilized female rats, which are commonly used in spinal cord injury studies because of their better compliance with surgical procedures and low rate of self-mutilation compared to males. Besides female rats are superior to male rats in terms of postoperative care due to low urinary infection rate. Rats were housed in animal rooms with sterilized bedding, water, and feed, in a quiet environment with appropriate temperature and humidity. Before the experiment, all rats were subjected to routine behavioral tests to ensure that their motor functions were normal. Rats were numbered sequentially and randomly divided into four groups (n = 5 per group): 1) Sham, laminectomy only without spinal cord transection; 2) Ctrl, spinal cord transection with implantation replaced by PBS; 3) SHED, spinal cord transection with SHED implanted at the lesion gap; and 4) hp-SHED sheet, spinal cord transection with hp-SHED sheet implanted at the lesion gap. Rats were fasted for 12 h before surgery and drank water freely. Rats were weighed and anesthetized with 1% pentobarbital sodium (0.4 mL/100 g) by intraperitoneal injection. Under deep anesthesia, each rat was fixed in a prone position on the operating table, and the hair near the T8–T10 spinous process on the back of the rat was shaved, prepared, and positioned to make a posterior median incision centered on T10. At this position, the skin, fascia, and muscle tissues were separated layer by layer to expose the T10 spinous process and transverse process. The T10 spinous process was clipped, the transverse process was held in place with forceps, and the T10 lamina was bitten off with a miniature biting forceps to expose the spinal cord. Care was taken not to touch the spinal cord throughout this process. The spinal cord of the T10 segment was completely transected using ophthalmic scissors. After cutting, the severed end was trimmed slightly to regularize the cut surface for better anastomosis with the filler. The section trimming and the retraction of both rostral and caudal tissue endings produced a gap of 2.0 ± 0.5 mm. Immediately after transection, both lower limbs of the rat were observed to twitch and the rat wagged its tail and then completely relaxed. A filler was implanted into the spinal cord gap to ensure anastomosis with the severed end of the spinal cord. The muscle, fascia, and skin tissues were sutured layer by layer at the end of the operation. The principle of asepsis was strictly observed during the operation. To avoid infection, penicillin was injected intramuscularly daily for 7 days after surgery, and the duration of use was extended in the presence of hematuria and pyuria. Manual urination was performed 2–3 times a day until the urination reflex was restored.

Sensory recovery was assessed using the Von Frey test (North Coast Medical, Carlsbad, CA, United States). Briefly, filaments with stimulus force gradients (0.07, 0.4, 2, 4, 10, 60, 180, and 300 g) were applied to the paw to elicit a nociceptive response (rapid stimulus avoidance by the paw) for at least 30 s between trials. Scores were determined by a blinded method. Assessments were performed at 9:00 p.m. due to wide variations in the diurnal activity of rats. Mechanical pain thresholds were assessed using the “Up and Down” method to calculate the 50% retraction-response threshold of rats. Rats with a 50% retraction threshold of 4 g or less were counted as positive for sensory function, and the percentage of positive rats was calculated. The baseline score was measured every 2–3 days after surgery, and then every 20 days until the end of the experiment (60 days).

Starting 1 week prior to SCI surgery, each rat was trained for 2 min per day to familiarize itself with the grip strength meter (KEWBASIS^®, KW-ZL-1, China). After 60 days of SCI, the grip strength of the hindlimb was measured for each rat by a blinded method. If the hindlimb motor impairment was too severe for the rat to grasp the crossbar, the grip strength score was 0. The grip strength tests on the same rat were performed a few minutes apart. We made sure that each rat was performed grip test with relaxed muscles and without spasms. Each rat was tested more than three times and the average was taken. The above ensures that the test values are at normal grip levels.

The Basso-Beattie-Bresnahan (BBB) locomotion rating scale is a neurological assessment method used to evaluate motor function in the hindlimbs of rats. A score of 0 is considered complete paralysis, whereas 21 is considered normal. All animals were checked for bladder fullness prior to the assessment to prevent bladder fullness from interfering with movements. The baseline score was measured every 2–3 days after surgery. Thereafter, measurements were taken every 20 days until the end of the experiment (60 days). In addition, video was recorded for each rat. The results were analyzed and processed using GraphPad Prism 8.0 software (GraphPad, San Diego, CA, United States).

Sixty days after surgery, rats were anesthetized and systemically perfused with saline. After the saline flow was free of blood, the tissues and organs were fixed with 4% PFA. Subsequently, samples of spinal cord from approximately 2 cm around the lesion site were obtained (including the lesion, as well as the rostral and caudal stumps). Similarly, the liver, spleen, lung, kidney, and bladder were removed and placed in 4% PFA for fixation.

After self-assembly of hp-SHED sheet, frozen sections were stained with hematoxylin and eosin (HE). Fixed spinal cord samples were cryosectioned, immunofluorescently stained, and imaged by confocal laser-scanning microscope (CLSM) (Leica, Wetzlar, Germany). The bladder was weighed and its volume measured. Finally, the rat liver, spleen, lung, kidney, and bladder were subjected to HE staining. Briefly, PFA-fixed tissues were removed, washed, embedded in paraffin, sectioned, and stained with HE. Images were acquired using an inverted light microscope. Finally, histological analysis was performed.

Hp-SHED sheet staining: Hp-SHED sheet was embedded in Optimal Cutting Temperature Compound and cryosectioned at a thickness of 10 μm. Sections were fixed in acetone for 5 min and then washed three times with PBS before incubation at 4°C overnight with one of the following primary antibodies: collagen Ⅰ (Abcam, Cambridge, UK), collagen Ⅲ (Abcam). The next day, an appropriate secondary antibody was added and incubated at room temperature for 1 h, and DAPI was used to stain cell nuclei. Samples were imaged under a CLSM.

Spinal cord staining: PFA-fixed spinal cords were embedded in Optimal Cutting Temperature Compound and sectioned on ice at a thickness of 10 μm. Subsequently, sections were fixed with 4% PFA, washed three times with PBS, and subjected to antigen repair in a cassette filled with EDTA antigen repair buffer (pH 8.0) in a microwave oven. Sections could cool naturally, blocked with 3% bovine serum albumin for 30 min, and incubated overnight at 4°C with one of the following primary antibodies: neurofilament (NF; Cell Signaling Technology), glial fibrillary acidic protein (GFAP; Millipore), calcitonin gene-related peptide (CGRP; GeneTex, Irvine, CA, United States), and myelin basic protein (MBP; Signalway Antibody, Baltimore, MD, United States). The next day, sections were incubated with a secondary antibody at room temperature for 1 h and DAPI was used to restain cell nuclei. After mounting, sections were observed and photographed with a CLSM. Quantitative histomorphometric analysis was performed using Image-Pro Plus software (Media Cybernetics, Rockville, MD, United States).

Experimental data were statistically analyzed using GraphPad Prism and SPSS 23.0 software (IBM, Chicago, IL, United States). Measurement data are described as mean ± standard deviation (SD) for n ≥ 3. Differences between two groups were compared utilizing t tests, while those among multiple groups were compared by ANOVA, followed by Tukey’s test for multiple comparisons. BBB scores at different time points were analyzed by repeated measures ANOVA, followed by Tukey’s test for multiple comparisons. *p < 0.05 was considered statistically significant. Significance levels: *p < 0.05, **p < 0.01, ***p < 0.001.