miR-NPs-RVG promote spinal cord injury repair: implications from spinal cord-derived microvascular endothelial cells

Abstract

Background

Spinal cord injury (SCI) often leads to a loss of motor and sensory function. Axon regeneration and outgrowth are key events for functional recovery after spinal cord injury. Endogenous growth of axons is associated with a variety of factors. Inspired by the relationship between developing nerves and blood vessels, we believe spinal cord-derived microvascular endothelial cells (SCMECs) play an important role in axon growth.

Results

We found SCMECs could promote axon growth when co-cultured with neurons in direct and indirect co-culture systems via downregulating the miR-323-5p expression of neurons. In rats with spinal cord injury, neuron-targeting nanoparticles were employed to regulate miR-323-5p expression in residual neurons and promote function recovery.

Conclusions

Our study suggests that SCMEC can promote axon outgrowth by downregulating miR-323-5p expression within neurons, and miR-323-5p could be selected as a potential target for spinal cord injury repair.

Graphical Abstract

Introduction

Spinal cord injury (SCI) is an extremely destructive disorder of the central nervous system (CNS) that always leads to severe locomotor functional impairment. Axon regeneration and growth are the basis for functional recovery after injury [1]. A combination of exogenous injury and inadequate intrinsic stimulation results in failure of axonal regeneration and growth [2]. Activation of axon-associated endogenous regeneration genes may contribute to axon regeneration [3, 4]. Manipulating the expression of these genes may promote axonal outgrowth, which in turn promotes spinal cord injury repair.

Axons exhibit vigorous growth capacity during the developmental stage, which might provide clues associated with intrinsic genes on axon regeneration and growth [5, 6]. During development, nerves and microvascular pervade the mesenchymal tissue in a high degree of orderly pattern to ensure a well-innervated and perfused functional system in the central nervous system. It is well known that the vasculature system supplies oxygen, nutrients, and hormonal information, removes local tissue metabolic waste products, and facilitates cell circulation [7, 8]. Research shows that nerves and microvessels guide and promote each other during development, and studies have revealed that the vascular system can influence the nervous system’s development and maturity by secreting neurotrophin-3 [8,9,10,11]. Microvascular endothelial cells (MECs), as the main component of vascular systems in the CNS, might play an important role in the process of the vascular-guided nervous system’s development. When co-cultured with neural stem cells, MECs can enhance the proliferation and migration of neural stem cells in vitro [12, 13]. While the effect of MECs on neurons has rarely been investigated, whether MECs can influence the behaviour of neuronal axons remains to be studied. Once the relationship between MECs and neurons was deciphered, intrinsic genes on axon regeneration and growth might be clear.

Changes in cellular phenotype are usually due to the switch of transcriptome [14]. miRNAs regulate a variety of physiological processes by manipulating target gene stability and expression [15]. The fact that the manipulation of one miRNA can affect many genes makes them suitable as intervention targets for the regulation of endogenous regenerative genes [16]. At present, interventions based on miRNA mainly include inhibiting intracellular miRNA by antagomirs and upregulating expression with mimics [17]. While the application of miRNA in vivo still faces challenges, such as off-target effects and instability in blood [18].

In this study, we initially validated that SCMECs could promote axon growth. RNA-seq was performed on neurons from the co-culture system to screen the intrinsic genes of axon growth. At the transcriptomic level, we found that miR-323-5p associated with axon growth was significantly downregulated in the neurons cocultured with SCMECs. In vitro neurons, reducing the expression of miR-323-5p could impel endogenous axon outgrowth. To down-regulate miR-323-5p in vivo and boost endogenous axon regeneration ability after SCI, PLGA nanoparticles encapsulating miR-323-5p inhibitor modified with RVG (miR-NPs-RVG) were employed to repair spinal cord injury. Behavioural and electrophysiological studies show recovery of motor and sensory functions.

Materials and methods

Animals

All Wistar rats were purchased from the Vital River Laboratory (Beijing, China). Animals are maintained in an environment with suitable temperature and humidity, and have free access to water and food. The animal room was illuminated with a 12-h-on/12-h-off cycle. All rats were removed from the study and euthanized with an overdose of pentobarbital sodium. The best efforts were made to reduce the suffering of rats in the experiment. All animal experiments were performed according to protocols supervised by the Ethics Committee of Tianjin Medical University (Tianjin, China, IRB2021-DW-74).

SCI model

Female Wistar rats, weighing 210 ± 10 g, were randomized into three groups as follows: Sham group (n = 10), in which rats received laminectomy only; SCI group (n = 10), in which rats received SCI without intervention; miR-NPs-RVG group (n = 10), rats received SCI and injected with miR-NPs-RVG (2 mg/kg, 500 µg/mL) via the tail vein on days 1, 3 and 7after injury. Surgical procedures: rats were anesthetized with 3% isoflurane mixed with oxygen gas, and maintained at 1% isoflurane throughout the surgical procedure. Lamine of T10 of the spinal cord was removed, and a pair of toothed retaining rods fixed the bilateral pedicles, ensuring that no vertebral movement was interfered with by respiratory movement. The impact bar was placed on the spinal cord, and a 10 g node (a diameter of 2.5 mm) was freely dropped from a height of 2.5 cm to create contusion injury with NYU Impactor Model III (W.M. Keck Center for Collaborative Neuroscience, Rutgers, the State University of New Jersey, United States). Successful establishment of the SCI contusion model was indicated by hind limb spasms and tail fluttering. After the surgery, the rats were put on a heating pad until they regained consciousness. Bladder care was given 2 times daily. Tissue processing: at indicated time points, mice were deeply anesthetized with isoflurane (5% isoflurane mixed with oxygen gas) and perfused transcardially with 4% paraformaldehyde or PBS. For WB, rats were only perfused with PBS. For immunohistochemistry, rats were perfused with PBS and 4% paraformaldehyde.

Preparation of primary cortical neurons and spinal cord microvascular endothelial cells (SCMEC)

Primary neurons were isolated from embryonic day 14–16 Wistar rat cortical tissue. Briefly, the pregnant rat was sacrificed following isoflurane (RWD Life Science, Shenzhen, China) anaesthesia and sterilized by soaking in 70% pre-cooled ethanol. The embryos were removed in aseptic operating table. Brains of embryos were harvested and the parietal cortical layer was dissected with an anatomical microscope. After the removal of blood vessels and pia mater, the parietal cortex was digested with papain in a 2 mg/mL density (Solarbio, China) and 1000 U/mL DNase I (Sigma-Aldrich, USA) at 37 °C for 15 min. All 24-well plates were covered with poly-d-lysine (PDL). DMEM complete medium was used for the initial seeding plate. After four hours, replace the DMEM medium with a fresh Neurobasal medium (Gibco, USA).

Spinal cord microvascular endothelial cells (SCMECs) were isolated from 2-week-old Wistar rats. In brief, rats were sacrificed with deep anaesthesia and sterilized with 70% precooled ethanol. The spinal cord was dissected from canalis vertebralis and placed in a pre-cold dissection buffer. The mater was carefully removed with general micro equipment under the anatomical microscope and the tissue was scissors minced as possible. Then, the tissue was enzymatic dissociated with collagenase II in a 1 mg/mL working density for 1 h. The digestion was terminated by adding complete media. To abolish myelin debris, centrifuge and resuspend the pellet in 20% BSA and centrifuge without break (1000 rpm, 5 min). After discarding the supernatant, the pellet was digested with collagenase/dispase (5 mg/mL, Sigma-Aldrich, USA) and DNase I (1000 U/mL, Sigma-Aldrich, USA) for 1 h at 37 °C. After the termination of digestion, the cells were collected and cultured in an EGM-2 medium (Lonza, Belgium) added with puromycin for purity. Replace with fresh medium after two days to remove unadherent cells.

Coculture

To determine the influence of microvascular endothelial cells on neurons, SCMECs were co-cultured with neurons directly and indirectly. Neurons were cultured separately as control (C). Direct coculture (D-CO) systems were constructed by mixing SCMECs and neurons at a 1:1 ratio. Indirect coculture (ID-CO) systems were constructed by employing a 0.4-μm-pore size transwell chamber system (Corning, Beijing, China) without Matrigel. Neurons were plated into the low chamber immediately after isolation. The next day, the upper chambers with SCMECs were placed. Cocultures were fed with a mixed medium consisting of Neurobasal medium and EGM-2 medium without FBS.

RNA sequencing

To evaluate the differential gene expression of neurons between C and ID-CO culture systems, RNA-seq was performed. Total RNA of the neurons of C and ID-CO culture systems were extracted using the TRIzol Reagent (Invitrogen, USA). After constructing cDNA libraries, cDNAs were sequenced using Illumina HiSeq2500. Single-end sequencing reads had a length of 1 × 50 bp.

Luciferase reporter assay

The pGL3 plasmid encoding a luciferase reporter gene was purchased from Promega (Madison, WI, USA). Full-length 3′UTR of genes were inserted into the MCS site. HEK293T cells were co-transfected transiently with constructed reporter plasmid and miR-323-5p mimics.36 h later, cells were collected in lysis buffer. Luciferase activity was performed using the Dual-Luciferase Reporter Assay System (Promega).

Immunocytochemistry and immunohistochemistry

For immunocytochemistry, the supernatants of cells were removed. After being washed three times with washing buffer, 4% paraformaldehyde was used to fix cells for 10 min. Cells were blocked and permeabilized in 5% goat serum supplemented with 0.25% Triton-100. Primary antibodies are incubated for 2 h at room temperature; secondary antibodies are incubated for 1 h. Cells were viewed under a Leica DMi8 fluorescence microscope (Leica, Germany).

For immunohistochemistry, the anaesthetized rats were perfused with precooled PBS and 4% paraformaldehyde sequentially. Then, 0.5 cm segments containing the epicenter of the injured spinal cord were cut and dehydrated in sucrose gradient buffer. Dehydrated spinal cords were embedded in the TissueTek OCT compound (Sakura, Japan). Tissues were sliced into 10 μm sections. The OCT was washed with TBST solution. Samples were outlined using a pap pen and penetrated and blocked at the same time with a blocking solution containing 5% BSA and 0.25% Triton-100 for 1 h. Primary antibodies were incubated overnight. Secondary antibodies were applied for 1 h at room temperature. Panoramic Scan Whole Slide Scanner (3DHISTECH, Hungary) was used to scan slides were scanned. All antibodies and dilution ratio used can be found in Supplementary Table 1.

In vitro miRNA transfection

The miR-323-5p NC, mimics and inhibitors were generated in GenePharma (Shanghai, China). Transfection was carried out by mixing the reagent with Lipofectamine RNAiMax (Invitrogen, USA). The synthesized miRNAs (20 nM) were mixed with Lipofectamine RNAiMax followed the manufacturer’s guidance. After 24 h, the medium was replaced with a fresh medium without antibiotics.

Preparation of miR-NP-RVG

miR-NPs composed of miR-323-5p inhibitor and PLGA, including PLGA and maleimide-PEG-PLGA (Chongqing Yusi Pharmaceutical Technology, China), were synthesized by using the double emulsion solvent diffusion method. Briefly, miR-323-5p inhibitor was mixed with spermidine (Solarbio, China) at an N/P ratio of 8:1. To form the miRNA/spermidine complex, the mixture was stood for 15 min. 2 mL of dichloromethane were used to dissolve 2 mg of maleimide-PEG-PLGA and 18 mg of PLGA separately (Aladdin, China). Then add the miRNA/spermidine complex. To acquire a primary oil phase emulsion, sonicate the mixed solution at 42 kHz ± 6% (2 min, ice bath).

The oil phase emulsion was carefully induced to 12 mL of 0.1% w/v poloxamer solution. Then emulsify the suspension by sonication at an amplitude of 70% for 40 s. Rotary evaporation was used to evaporate the DCM in vacuo at 45 °C slowly. The nanoparticles were cleaned by adding ddH20 and centrifugation (13,000 rpm, 30 min, 4 °C).

To obtain the ability to target neurons, the nanoparticles were modified with rabies virus glycoprotein-cys (RVG29-cys, sequence: YTIWMPENPRPGTPCDIFTNSRGKRASNGC, Shanghai Apeptide Co. Ltd, China) via a thiol-maleimide Michael addition reaction. Briefly, nanoparticles were incubated with excess RVG29-cys for 48 h. Nanoparticles were washed by centrifugation to remove unlinked peptides. The nanoparticle solutions were used for characterization and immediate use for in vivo and vitro studies.

Electrophysiological test

Electrophysiological study was conducted for each group at week 7 following SCI (n = 5 rats/group). Before the examination, the animals were anesthetized with 1% pentobarbital (50 mg/kg). Motor evoked potentials (MEP) and sensory evoked potentials (SEP) were measured with electrophysiological device (YRKJ-G2008; Zhuhai Yiruikeji Co, Ltd, Guangdong, China).

For MEP, the stimulating electrode (needle electrodes) was placed under the scalp behind the ear and attached to the skull. The reference electrode was placed 1 mm beside the stimulation electrode. The recording electrode was inserted into the contralateral gastrocnemius muscle, 1 mm apart inserted the reference electrode, and the ground electrode was placed subcutaneously in the back of the rat. For SEP, the stimulation electrode was inserted in the posterior tibial nerve of the hindlimb. The reference electrode was placed 1 mm beside the stimulation electrode. The recording electrode was placed subcutaneously on the gastrocnemius head, 1 mm apart inserted the reference electrode, and the ground electrode was placed subcutaneously in the back of the rat. The latency and amplitude of MEP were recorded and analyzed in the experiment.

WB

Western blot analysis was performed according to the procedure described before [19]. Briefly, total protein from the cocultured primary neurons was extracted using RIPA reagent supplemented with protease and phosphatase inhibitors (Roche, Switzerland). Protein was separated on a 10% SDS-PAGE gel, followed by transfer to polyvinylidene fluoride membranes. After being blocked with 5% BSA, the membranes were incubated with primary antibodies overnight at 4 °C. The next day, after washing away the unbound primary, the membranes were incubated with the second antibody linked with horseradish peroxidase (HRP) for 60 min and then visualized using an ECL chemiluminescence kit (Thermo Fischer, USA). β-Actin and GAPDH were used as internal reference. Antibodies and dilution ratio are listed in Supplementary Table 1.

Reverse transcription and quantitative real-time PCR

The extraction of total RNA was performed using TRIzol Reagent (Invitrogen, USA) according to laboratory practice. For mRNA, complementary DNA (cDNA) was generated from total RNA using a FastKing RT kit (Tiangen, China). UltraSYBR One-Step RT-qPCR Kit (Cwbio, China) was employed to perform Quantitative real-time PCR on Roche LightCycler 96. For miRNA, cDNA was reverse-transcribed by using a stem-loop method. All samples were run at least in triplicate. Primers are listed in Supplementary Table 2.

Statistical analysis

GraphPad Prism 9.2.0 software (GraphPad Software, SanDiego, CA, USA) was used for statistical analysis. The homogeneity of variances was tested with the Levene test. Normal distribution was performed using the Kolmogorov–Smirnov test of normality. Data analysis between different groups was carried out via Student’s t-test and one-way analysis of variance (ANOVA), which were followed by Tukey multiple comparison post hoc test. The level of significant difference between groups was defined as: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Each experiment was repeated at least three times, and the results were shown as the standard error ± mean (SEM).

Results

SCMECs could promote neurite outgrowth in a noncontact fashion

The special anatomical position of blood vessels and nerves suggests a tight communication between SCMECs and neurons which are the main constituent cells of two systems. To explore the effect of SCMECs on the biological behaviour of neurons, primary SCMEC and neurons were isolated and cocultured (Fig. 1A and Fig. S1A). Cell crosstalk encompasses three main pathways: direct cell–cell crosstalk, cell-ECM crosstalk, and paracrine signaling [20]. Thus, co-culture systems are designed to include both direct and indirect coculture (the D-CO group and ID-CO group) (Fig. 1A). Our results show that neurons co-cultured with SCMECs exhibit longer axons as well as more neurites compared to neurons cultured alone (C group) (Fig. 1B–D). At DIV7, the axons grew further, but there was no difference in the number of neurites (Fig. S2A–C). Interestingly, there were no significant differences observed in the length of axons and the number of neurites between the D-CO and ID-CO groups. This suggests that the direct cell–cell contact crosstalk between SCMECs and neurons is not a contributing factor (Fig. S2B, C).

Fig. 1

SCMECs promote axon growth in vitro. A Timeline and experimental paradigm. Primary SCMECs and neurons were cocultured using the D-CO and ID-CO methods. Neurons cultured alone (C group) were used as a control. B Representative fluorescence images and 8-bit grayscale images of neurons co-cultured with SCMEC indirectly compared with neurons cultured alone (C group) at DIV(days in vitro) 3 and 7, Scale bar = 25 µm. C, D The neurite number and average axonal length were calculated based on all neurons from three different in vitro culture systems. Results are shown as box plots (n = 30, ns: non-significant **p < 0.01, ***p < 0.001). E Protein expression level of GAP43 in neurons was measured by Western blot. GAPDH protein was used as a loading control. G Quantification of GAP43 expression in the ID-CO and C group. GAP43 protein is expressed at higher levels in ID-CO group (***p < 0.001)

Growth cone is a specialized structure of axon terminals which guides axons toward their targets [21]. Growth-associated protein 43 (GAP43) is a protein found in axonal growth cones located in the presynaptic terminal [22]. Growing axons usually appear as active growth cones [23]. By detecting the expression of GAP43, the growth state of neuronal axons can be determined. It was shown that GAP43 expression increased dramatically in the ID-CO group at DIV 3 and 7 (Fig. 1E, F), which indicated that the growth of the axon is exuberant. These results indicate that SCMEC could promote neuronal axon growth through a noncontact fashion rather than direct contact.

miRNA-323-5p plays a pivotal role in the regulatory effect of SCMEC on neurons

Alterations of phenotypes in neurons usually result from changes in the transcriptional program. miRNA sequencing (miRNA-seq) on neurons from the C group and ID-CO group was performed to investigate more deeply the mechanisms and pathways by which SCMEC promotes axon growth. Unsupervised hierarchical clustering analysis was performed to highlight the transcriptomic difference of miRNA in neurons from the C and ID-CO groups (Fig. 2A). A total of 378 differentially expressed miRNAs were detected with 147 downregulated and 231 upregulated in ID-CO group compared to the C group (Fig. 2B and Fig.S3A). The downregulated miRNAs with a mean log2-fold change higher or lower than 0 and a p-value < 0.05 were chosen for further analysis. To identify putative targets of downregulated miRNAs, we employed four different prediction tools, miRDB, Elmo, microcosm, and miRanda and identified 289 common genes (Fig. 2C). The predicted genes were significantly enriched in axon guidance, PI3K-Akt and regulation of actin cytoskeleton signalling pathways, as revealed by the Metascape enrichment analysis. The above results indicate that the downregulated miRNAs in neurons are closely related to the growth of axons regulated by SCMEC. To clarify the exact changes in mRNA in neurons, we performed mRNA-seq, and a total of 3170 upregulated DEG were used for further analysis (Fig. 2E and Fig. S3B). Further KEGG analysis showed that PI3K-Akt and axon guidance signalling pathways were highly enriched (Fig. 2F). GO enrichment analysis showed axonogenesis was the top pathway (Fig. 2G).

Fig. 2

Transcriptome analysis of neurons in co-culture systems. A The heatmap of representative differential expressed miRNAs between the C group and ID-CO group. Each row represents a miRNA; each column represents one sample tested. B Volcano plot of differentially expressed miRNAs in neurons. The X-axis displays the miRNA fold change and the Y-axis gives the statistical significance of the change. The red (or blue) color indicates up-regulated (or down-regulated) genes. C The Venn diagram illustrates the predicted overlapping target mRNAs of significantly downregulated miRNA in miRDB, elmmo, microcosm, and miRanda databases. D The metascape enrichment analysis of the predicted target genes of significantly downregulated miRNAs. E Volcano plot of differentially expressed mRNAs between the C group and ID-CO group (Red, significantly upregulated; blue, significantly downregulated). F KEGG pathway enrichment analysis upregulated genes in neurons from mRNA-seq. G Biological process (GO‐BP) analysis of upregulated genes in neurons from mRNA-seq. H KEGG pathway enrichment analysis of the common genes from predicted target genes of miRNAs and genes generated from mRNA-seq. I Veen diagrams of the intersection of upregulated mRNAs relating to Axonogensis, Axon guidance, and Regulation of actin cytoskeleton. J Veen diagrams of the intersection of downregulated miRNAs relating to Axonogensis, Axon guidance, and Regulation of actin cytoskeleton. K Visualization of miRNA-323-5p interacting with predicted targeted genes (pink: involved in Axonogensis, Axon guidance, regulation of actin cytoskeleton)

Subsequently, a combined analysis of differentially upregulated expressed mRNAs and downregulated miRNAs was performed to screen out target genes. KEGG analysis showed enrichment in axon guidance and PI3K-Akt signalling pathways for common genes in upregulated mRNAs and target mRNAs of downregulated miRNAs (Fig. 2H). There are 15 miRNAs associated with axonogenesis, regulation of actin cytoskeleton and axon guidance signalling pathways in the downregulated miRNAs. The critical component genes of the three signalling pathways described above are targeted by miR-323-5p (Fig. 2I, J). The predicted secondary structure of miR-323-5p resembles a short hairpin RNA which can silence gene expression (Fig. S3D). Additionally, predicted target sites in the 3′-UTR sequence of downstream genes suggest a negative regulatory effect which was subsequently confirmed by a dual luciferase reporter assay (Fig. S3E–I). In conclusion, transcriptome analysis showed that SCMEC could promote axon growth by downregulating miRNA-323-5p expression via the PI3K-Akt signalling pathway in neurons.

Inhibition of miRNA-323-5p can promote axon growth in vitro

The transcriptome sequencing result revealed that the downregulating miR-323-5p may positively affect axon growth. RT-qPCR result showed miR-323-5p expression was lower in the ID-CO group (Fig. 3A), and its target genes are higher than neurons from the C group (Fig. 3B), which is also compatible with the classic theory that miRNAs can act as negative regulators in gene expression by inhibiting target gene translation [24]. In addition, Unc5b is the receptor of netrin 1, which is thought to be involved in axon guidance and cell migration during development [25]. GSEA enrichment analysis also suggested that the central gene was involved in the netrin signalling pathway (Fig. S3C). The high expression of netrin1 in the medium of the ID-CO system further confirmed our results (Fig. 3C). To gain further insight into the effect of miR-323-5p in the progression of neurite outgrowth, miR-323-5p inhibitors and mimics were transiently transfected into rat primary cortical neurons in vitro. At 48 h post‐transfection, miR-323-5p inhibitor treatment increased the axon length, compared with neurons cultured alone (Fig. 3D, E). Overall, these results indicate that SCMEC promotes neuronal axonal growth by downregulating miR-323-5p.

Fig. 3

Validation of RSCMECs promoting axon growth via downregulating miR-323-5p in neurons. A The relative expression of miR-323-5p in neurons from the C group and ID-CO group. B The relative expression of F2r, Nfatc2, Unc5b, and Nrp1 detected by RT-qPCR in neurons from the C group and ID-CO group on DIV 1,3,7. C Netrin 1 ELISA analysis of cell culture supernatants. Supernatants were collected from the C group and ID-CO group on DIV3. D Representative fluorescence images of neurons 48 h after miR-323-5p mimics or inhibitor treatment. Neurons were labeled with Tuj1. E Average axonal length was calculated (ns indicate no significance, *p < 0.05). F The expression of PI3K-AKT in neurons from the C group and ID-CO group were detected by western blot; Representative western blot: p-PI3K, p-Akt. G, H The western blot assay of p-Akt and p-PI3K. Protein levels are normalized to β-actin and are expressed relative to the C group. Data are mean ± SD (ns indicate no significance, *p < 0.05, **p < 0.01, ***p < 0.001)

To investigate if SCMEC functions through the PI3K-Akt signaling pathway, which is responsible for axon growth, we analyzed the expression of p-PI3K and p-Akt, the main components of this pathway. The expression of p-PI3K and p-Akt was increased in the ID-CO system compared to the C group at DIV3 and 7 (Fig. 3F–H). The evidence supports our previous RNA sequencing analysis, indicating a critical role of the PI3K-Akt signaling pathway in neuronal axonogenesis. Collectively, inhibition of miR-323-5p promotes axon growth in vitro.

Fabrication and characterization of miR-NPs-RVG

Following spinal cord injury (SCI), the inability of axons to regenerate continues to pose a major challenge [26]. Several studies have demonstrated the unity of inhibiting miRNAs via complementary anti-miR molecules [27, 28]. We have shown the axonal growth-promoting effect of deregulated miR-323-5p in vitro by using miR-323-5p inhibitor. Therefore, it is highly necessary to develop new miRNA delivery systems for in vivo targeting neurons to downregulate miR-323-5p. Considering that miR-323-5p was influenced by SCMEC and microvascular can self-repair after spinal cord injury [29]. We firstly detected microvascular changes at different times following spinal cord injury (Fig. 4A). The results showed that the microvessels at the epicentre could not be restored to the pre-injury state, which indicated the microvessels in the spinal cord had limited self-healing ability, and SCMEC had a decreased promotion effect of SCMEC on neuronal axons in vivo (Fig. 4B, C).

Fig. 4

Changes of microvessels at the center of injury after spinal cord injury. A CD31 and NF200 immunofluorescence staining of epicenter after spinal cord injury. Scale bar 100 nm (dpi: days post-injury). B, C Quantification of the vessels (CD31) and axonal area (NF200). Data is presented as mean ± SEM, one-way ANOVA with a Tukey’s post-hoc test. Scale bars: 100 μm (ns indicate no significance, ***p < 0.001, ****p < 0.0001)

Therefore, neuron-targeted miR-323-5p inhibitor@PLGA-RVG nanoparticles (miR-NPs-RVG) were employed to deliver miR-323-5p inhibitor into the neurons. miR-NPs-RVG were synthesized by the double emulsion method (Fig. 5A). RVG, as the specific ligand of the nicotinic acetylcholine receptor, could be specifically recognized by neurons and was linked to particles via Michael addition (Fig. 5B). The conjugation of RVG to the polymeric NPs was confirmed by Fourier-transform infrared radiation (Fig. 5E). As was evident by electron microscope (TEM), miR NPs displayed spherical shapes and dispersed uniformly (Fig. 5C). The average particle size of miR-NPs was 157.1 ± 2.88 nm, with a dispersity of 0.09 (Fig. 5D). A surface charge of 4.9 ± 0.6 mV will avoid clearance of macrophage [30]. The small particle size (< 200 nm) is ideal for systemic application and has superior serum stability.

Fig. 5

Characterization of miR-323-5p inhibitor @PLGA-RVG (miR-NPs-RVG) and evaluation of safety and their ability to target neurons. A Structure of miR-NPs-RVG. The particles are composed of a miR-323-5p inhibitor core and a PLGA shell linked with RVG peptide. B The connection between RVG peptide and mal-PEG-PLGA via Michael addition reaction. C Representative TEM images. D Particle size distribution of miR-NPs-RVG. E Fourier transform infrared spectra of miR-NPs-RVG. F Hemolytic activity of nanoparticles. Red blood cells from healthy Wistar rats were incubated with a series of concentration gradients of nanoparticles with PBS as negative and ddH₂O as positive controls. G Hemolytic activity of different concentrations nanoparticles. H, I Cell viability was examined by calcein-AM/ethidium homodimer-1 staining. J Nanoparticle uptake assay. miR-NPs-RVG were modified with Cy5 fluorophore

Hemolysis experiments were performed to screen for a safe concentration. There was no haemolysis observed at a concentration of 500 µg/mL (a hemolysis ratio of ≤ 2 was considered haemolysis-free) (Fig. 5F, G). After coculturing the nanoparticles with HT22 for 48 h and staining for live and dead cells, we found almost no dead cells (Fig. 5H, I). The safety of nanoparticles was assessed via intravenous injection. No apparent lesion was shown in the heart, liver, spleen and kidney (Fig. S4A). By labelling miR-NPs-RVG and neurons with Cy5 and neuro-Dil, respectively, neurons and miR-NPs-RVG were cocultured. Co-localization of nanoparticles in neurons was detected, indicating that neurons could take up the nanoparticles (Fig. 5J). In conclusion, the results showed miR-NPs-RVG have good biocompatibility and cellular uptake efficiency and were suitable for systematic application.

miR-NPs-RVG propel axon regeneration and function recovery in rats with spinal cord injury

Success in vascular remodelling is an important prerequisite for successful drug delivery via the tail vein. We have found microvascular that are reconstructed after SCI (Fig. 4A). To confirm successful spinal cord delivery of miR-NPs-RVG, miR-NPs-RVG labelled with Cy5 were injected. As expected, ex vivo imaging revealed that fluorescence intensity was significantly increased in spinal cord at 2 h post injection. In SCI group, due to the loss of neurons in the epicenter, no fluorescence was detected (Fig. 6A). Colocalization of nanoparticles labelled with Cy5 and neurons also indicated the success of delivery and neuronal targeting (Fig. 6B).

Fig. 6

The effect of targeting neurons and promoting axon growth. A Distribution of fluorescence intensity in body-wide tissues. B The neurons labelled with NeuN at the lesion periphery were co-localized with Cy5-positive miR-NPs-RVG (white arrows indicate the co-localized neurons). C Immunofluorescence staining of NF-200 (green) and CD31 (red) at 35 days after SCI. The two right-most pictures were the individual channels of magnified images. Scale bar: 500 μm. D The ratio of NF-200 positive area to spinal cord area in three groups (n = 5 rats per group). E The ratio of CD31 positive area to spinal cord area in three groups (n = 5 rats per group)

Subsequently, we examined their ability to promote axonal growth in rats with spinal cord injury. At DIV 35, the density of nerve fibres labelled with NF-200 was significantly higher in the miR-NPs-RVG group compared to the SCI group. While compared with the sham group, the nerve fibre density was still significantly reduced (Fig. 6C, D). The blood vessel density was quantified in the injured area and no significant difference was found between the miR-NPs group and the SCI group (Fig. 6E). These results show that the growth of nerve fibres in the epicentre of the spinal cord injury is not related to the restoration of microvascular and miR-NPs-RVG could promote the growth of nerve fibres. Subsequently, we measure the continuity of nerve fibres by neuroelectrophysiology and evaluate neurological reconstruction by behavioural evaluation. The BBB score showed that the miR-NPs group had better motor function recovery than the SCI group (Fig. 7A). The electrophysiological performance of the miR-NPs-RVG group exhibited shorter latency and higher amplitude (Fig. 7B–F).

Fig. 7

Functional recovery after miR-NPs-RVG treatment in SCI rats. A Basso Beattie Bresnahan (BBB) score across different groups over time. Data are presented as mean ± standard deviation (n = 5). B Neuronelectrophysical detection at week 7 post-injury. Examples of motor-evoked potential (MEP) and sensor-evoked potential (SEP) were obtained (n = 5). C–F SEP/MEP latencies and amplitudes in experiment, sham and control groups (n = 5; ns indicate non significance, *p < 0.05, **p < 0.01). G–K The CatWalk XT was used to detect subtle gait and motor recovery. Analyses of these recordings yielded many parameters. Dynamic parameters are presented as stand (H), Max contact area (I). The base of support (J) and CatWalk-duty cycle (K) implies Coordination data (n = 5 rats per group, ns indicate non significance, *p < 0.05,**p < 0.01,***p < 0.001, ****p < 0.0001)

Finally, the CatWalk gait analysis system was employed to detect more subtle alterations in motor performance. It can be intuitively found that rats from the miR-NPs-RVG group had more coordinated gait, while the SCI group did not restore the support of hindlimbs, compared with the sham group (Fig. 7G–J). The CatWalk-duty cycle (Fig. 7K) implies that the coordination data demonstrated that the miR-NPs-RVG group had recovered locomotor ability close to the sham group (Fig. 7K).

The above results indicated that the motor function of spinal cord injury rats after treatment had partially recovered. The above results suggested that the downregulation of miR-323-5p in the neurons of spinal cord injury rats could partly repair the damaged spinal cord. This promoting effect has little relationship with microvessels in the subacute stage of SCI.

Discussion

Following a spinal cord injury (SCI), the regeneration of axons is often unsuccessful, leading to permanent disability [31]. Although great efforts have been made, the regeneration of axons after SCI remains an unresolved problem in spinal cord injury repair [32, 33]. In the peripheral nervous system (PNS), axons can regenerate successfully [6, 34], which suggests that damaged axons of the CNS have regenerative capacity but were inhibited for some reason. Therefore, the key to repairing spinal cord injury is to remove inhibiting factors and identifying intrinsic regenerative genes to promote intrinsic regeneration [3, 35]. In our study, we first discovered that SCMECs could enhance the outgrowth of axons cocultured with neurons, which suggests that the presence of SCMECs can promote axon intrinsic growth. We speculate that the mechanism by which SCMEC promotes axon growth is the removal of constraints at the transcriptional level on axonal growth.

miRNAs are single-stranded non-coding RNAs that negatively regulate gene expression post-transcriptionally without encoding proteins [36, 37]. The ability of miRNAs to target multiple pathways is attractive, as it may regulate a variety of physiological processes [38]. We propose that the suppressive miRNAs, acting as negative regulators of intrinsic gene expression associated with axon regeneration, were inhibited by SCMECs in the co-culture system. After spinal cord injury, these suppressive miRNAs lose their inhibitory factors and show high expression. To figure out the mechanism of axon intrinsic growth and promote axon regrowth after spinal cord injury, RNA-seq was performed on neurons from the co-culture system. We screened out the downregulated miRNAs and upregulated genes by performing RNA-seq on neurons from the co-culture system. Further investigation indicated that miR-323-5p is associated with axon growth, and downregulating miR-323-5p could promote axon outgrowth in vitro neurons, which illustrates that miR-323-5p could be a potential target to repair spinal cord injury.

Currently, miRNA-based therapeutic strategies include the use of oligonucleotide mimetics or antisense oligonucleotides such as miR-mimic and inhibitors, agomir-miRNA, and antagomir-miRNA [39,40,41]. However, these strategies hold some major limitations, such as oligonucleotide degradation, off-target effects, and organ toxicity [42, 43]. To avoid these problems, we employed RVG-modified miR-NPs-RVG to deliver miR-323-5p inhibitors to downregulate the miR-323-5p in neurons. PLGA nanoparticles can encapsulate oligonucleotides to prevent degradation in serum, and neurons could specifically recognize RVG to ensure that miR-NPs-RVG reach target cells. The robust axon regrowth across the lesion core and recovery of locomotor indicated that miR-NPs-RVG are beneficial for SCI.

Although we have verified the therapeutic ability of nanomaterials for miRNA delivery, there are still many problems to be solved. Although chemically synthesized miRNA inhibitors can specifically bind to mRNA [44], the multitarget genetic nature of miRNA may promote the high expression of other genes at the same time as the knockdown target gene [45]. Additionally, almost all neurons in the body recognize RVG [46], which means that all neurons are affected by miR-NPs- RVG. Although the animal in our study showed no other abnormalities, the long-term effects cannot be ignored. Next step, we will conduct experiments in non-human primates. It remains to be verified whether normal neuronal axons will be affected and whether the growth of these normal neuronal axons will cause neurological diseases, such as epilepsy. At the same time, considering the conservation of genes, there are differences between rat and human genes, and whether miR-323-5p can still play a normal regulatory function on human genes.

On the other hand, it remains to be explored whether microvascular regeneration after spinal cord injury can induce axon regeneration on residual neurons. Recently, the transplantation of certain doses of SCMECs was effective in revascularization and inflammation regulation and could improve nerve regeneration which revascularization at the epicenter is to the benefit of spinal cord injury repair [13]. Meanwhile, whether the microvessels are mature and functional cannot be ignored, and our next step will be to explore the reconstruction and maturation of microvessels in spinal cord injury repair.

In conclusion, our study deciphers the mechanism by which SCMEC promotes neuronal axon growth by downregulating the expression of miR-323-5p, and its decrease could effectively promote the increase of axon growth-related genes, which eventually led to the acceleration of axon growth. We developed a miRNA-323-5p-based therapy for SCI, which downregulates miR-323-5p in vivo, promotes local nerve fiber growth, and can repair SCI with clinical translation potential.