Exosomal shuttle-mediated intraneural revascularization after PNI.
According to the window period of polarized vessel formation on the 1st to 5th day reported in previous studies10, the day 5 injury-site nerve segments were harvested for detection. In the present study, GW4869 (2.0µg/g body weight) was intraperitoneally administered daily from 5 days before the operation to after 5 days to inhibit the intraneural exosome paracrine. This showed that the CD31+ blood vessels were significantly diminished by GW4869 administration in the sciatic nerve, which means remarkably restrained intraneural revascularization (Fig. 1A-1B). Consistent with the studies of exosome involvement in regulating nerve regeneration and repair, GW4869 administration significantly impeded axonal (anti-NF-200) and myelin regeneration(anti-MBP) (Fig. 1C-1E). Further, paw print and sciatic nerve function index (SFI) results showed that the neural function recovery was also remarkably delayed (Fig. 1F-1G). These data suggested that the intrinsic exosomal shuttle in the post-injury sciatic nerve was intimately linked to intraneural revascularization, post-injury axon regeneration, and functional recovery after SCI.
SCs-Exos facilitated intraneural revascularization, and hypoxia precondition enhanced this process.
To explore whether Schwann cells-derived exosomes affected these post-injury intraneural angiogeneses, in vitro, we collected the culture supernatants of normoxia or hypoxia-treated SCs to co-culture with HUVECs. This showed that supernatants from normoxia-cultured Schwann cells (N-SCs-supernatants) facilitated HUVECs tube formation (Fig.S1A and S1D), invasion (Fig.S1B and S1E), proliferation (Fig.S1C and S1F). Interestingly, the supernatants from hypoxia-cultured Schwann cells (H-SCs-supernatants) presented a more eminent pro-angiogenic role than N-SCs-supernatants (Fig.S1A-S1F). To ascertain the role of exosomes on pro-angiogenesis, GW4869 was used to inhibit exosome secretion. This showed that the above pro-angiogenic roles of both N-SCs-supernatants and H-SCs-supernatants were counteracted by GW4869 treatment (Fig.S1A-S1F). To further explore the effect of ECs treated with the above SCs-supernatants respectively on dorsal root ganglion neurons (DRGns), a co-culture system was established (Fig.S2A). This showed that H-SCs-supernatants-treated ECs performed more significant promotion on axonal growth than N-SCs-supernatants, but both of them were counteracted by GW4869 administration (Fig.S2B and S2C). In summary, our data suggested that exosome-relevant paracrine of SCs mediated its pro-angiogenic role on endothelial cells further promoting neural axonal growth in vitro, where hypoxia precondition could enhance this positive process.
Further, we extracted the exosomes from SCs culture supernatants via ultracentrifugation as the procedures reported in previous studies 23,24. Transmission electron microscopy (TEM) showed that both N-SCs-Exos and H-SC-Exos were bilayer round-shaped membrane vesicles (Fig. 2A). The diameters of N-SCs-Exos and H-SCs-Exos ranged from 30nm to 212 nm (average 101.25nm) and 40nm to 225nm (average 109.25nm), respectively, and with a concentration at 4.56×109 and 5.32×109 particles/ml, respectively (Fig. 2B); Western blotting showed that exosomal specific markers were positive expressions like CD9, CD63, and TSG101 but cytoplasmatic protein Calnexin was negative expression (Fig. 2C); These data suggested that exosomes from SCs were successfully purified.
In vitro, the two purified exosomes were co-cultured with HUVECs, and the results showed that H-SCs-Exos significantly promoted HUVECs tube formation (Fig. 2D and 2G), invasion (Fig. 2E and 2G), proliferation (Fig. 2F and 2G). Further, a co-culture system of ECs and DRGns was established (Fig.S2D), and showed that H-SCs-Exos-treated ECs more eminent promoted DRGns’ axonal growth (Fig.S2E).
To determine the exact effects of N-SCs-Exos and H-SCs-Exos on intraneural endothelial cells in vivo, PKH-67-labeled N-SCs-Exos, and H-SCs-Exos was locally injected into the sciatic nerve injury site. The co-localized expression of PKH-67+CD31+ was detected to evaluate the internalized capacity of vascular endothelial cells in the sciatic nerve injury site to two-type exosomes. The observed internalization of exosomes revealed that both N-SCs-Exos and H-SCs-Exos were internalized by intraneural vascular endothelium but the intensity of H-SCs-Exos is slightly higher than that of N-SCs-Exos (Fig. 2H and 2I). Consistent with that in vitro, H-SCs-Exos facilitated intraneural revascularization (CD31 + blood vessels) more remarkably than N-SCs-Exos (Fig. 2H and 2J). To further ascertain the post-injury structural construction of sciatic nerve, NF-200+ neurofilament and MBP+ myelin were detected by IF, and this showed that the fluorescent intensity of both was remarkably enhanced by N-SCs-Exos and H-SCs-Exos, where H-SCs-Exos performed more eminent (Fig. 2K-2M). To detect SFI at certain time points, we found that N-SCs-Exos significantly improved the sciatic nerve function; meanwhile, H-SCs-Exos demonstrated larger facilitation in improving sciatic nerve function than N-SCs-Exos (Fig. 2N and 2O).
These data suggested hypoxia precondition enhanced the promotion of SCs-Exos on intraneural revascularization, axon regeneration, and functional recovery.
SCs-Exos reprogram endothelial energy metabolism to facilitate intraneural revascularization.
Endothelial glycolysis level was intimately linked to post-injury angiogenesis. To explore the role of endothelial metabolic changes in facilitating post-injury intraneural revascularization, the metabolites in the injury site of the sciatic nerve were assayed. The results showed a noticeably increased lactate (Fig.S3A) and pyruvate production (Fig.S3B) and decreased acetyl-coenzyme A (acetyl-CoA) production (Fig.S3C) after SCI. Above metabolic reprogramming was counteracted by GW4869 administration (Fig.S3A-3C). Further, immunohistochemistry (IHC) showed the level of lactate dehydrogenase A (LDHA) that converted pyruvate to lactate both in the whole nerve section and intraneural blood vessels was increased (Fig.S3D-3F), while pyruvate dehydrogenase-E1-alpha subunit (PDH-E1α) that converting pyruvate to acetyl-CoA were decreased (Fig.S3G-3I). Consistently, the GW4869 administration reversed the above performances (Fig.S3G-3I). These data suggested that crush injury shifts the metabolic phenotype of local sciatic nerves in favor of glycolysis. To further determine the effects of this metabolic shift on post-injured nerve repair, an IF assay was performed after 3PO administration, a specific inhibitor of the glycolytic pathway, and this showed that glycolysis inhibition by 3PO significantly hampered the nerve fiber (anti-NF200) and myelin reconstruction (anti-MBP) (Fig.S3J). SFI measurement showed that 3PO delayed the automatic repair process of sciatic nerve function (Fig.S3K and S3L). These data unveiled the changes in further favor of glycolysis after PNI initiate the automatic repair process.
Collectively, our data indicated that exosomal shuttling is involved in injury-triggered metabolic reprogramming of endothelial cells in favor of glycolysis to facilitate intraneural revascularization and further functional recovery.
Whether SCs-Exos promoted endothelial angiogenesis through regulating metabolic reprogramming was further determined in vitro. Firstly, HUVECs were treated with DiR-labeled N- and H-SCs-Exos to detect the internalization capacity of HUVECs to them, and the result showed that both of them were successfully internalized by HUVECs without significant differences (Fig. 3A and 3B). According to the slight predominance of internalization of H-SCs-Exos in vivo, the sufficient concentration of exosomes masked such difference in vitro may be the main reason. Next, extracellular acid rate (ECAR) and oxygen consumption rate (OCR) were detected by the Seahorse XF96 energic analyzer, and this showed that H-SCs-Exos more significantly enhanced the ECAR (Fig. 3C) and inhibited the OCR of HUVECs (Fig. 3D), performing increased ECAR/OCR ratio (Fig. 3E), which meaning upregulated glycolytic metabolism and downregulated mitochondrial respiration. Further, the metabolites including pyruvate, lactate, and acetyl-CoA in HUVECs were detected. The results showed that compared to N-SCs-Exos, H-SCs-Exos more eminent increased the production of pyruvate (Fig. 3F) and lactate (Fig. 3G), but decreased the production of acetyl-CoA (Fig. 3H) in HUVECs. To delve deeper into the changes of related enzyme proteins catalyzing this process, the western blotting results showed that H-SCs-Exos performed more evidential than N-SCs-Exos in upregulating the expression of glycolytic enzyme protein that mediated lactate production, including glucose transporter (GLUT1), hexokinase2 (HK2), 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase 3 (PFKFB3) and LDHA (Fig. 3I), while downregulated PDH-E1α that mediated acetyl-coA production more remarkable than N-SCs-Exos (Fig. 3I).
In vivo, after N- or H-SCs-Exos locally administrated, western blot results showed that H-SCs-Exos was more significantly increased the overall level of lactate-produced protein enzyme including CLUT1, HK2, PFKFB3, and LDHA at the injury site after SCI than that of N-SCs-Exos (Fig. 3J), while PDH-E1α was more remarkable diminished by H-SCs-Exos (Fig. 3J). The metabolites assays showed that H-SCs-Exos more remarkably raised the production of pyruvate (Fig. 3K) and lactate (Fig. 3L) than N-SCs-Exos. In contrast to that, acetyl-CoA production was more significantly diminished by H-SCs-Exos administration (Fig. 3M). Furthermore, the percentage of LDHA+CD31+ /CD31+ area and PDH-E1α+CD31+ cells/CD31+ area was determined respectively in intraneural vascular endothelial cells of the injured sciatic nerve sites. These results showed that H-SCs-Exos more significantly increased the expression of LDHA both in the whole injured nerve (Fig. 3N and 3O) and intraneural ECs (Fig. 3N and 3P), but decreased that of PDH-E1α than N-SCs-Exos (Fig. 3Q-3S). In sum, these results suggested that hypoxia precondition enhanced the function of SCs-Exos in upregulating glycolysis and downregulating OXPHOS of intraneural vascular endothelial cells to facilitate intraneural revascularization and nerve repair.
Transferred miR-21-5p was responsible for the metabolic regulation of SCs-Exos on endothelial cells
MicroRNAs (miRNAs) are enriched in exosomes and perform various biological functions25. According to our and others’ previous studies, Schwann cells and its-released exosomes contain a large number of miR-21-5p 26–28. Interestingly, in the current study, H-SCs-Exos had a larger number of miR-21-5p than N-SCs-Exos (Fig.S4A), in agreement with the characteristics of miR-21-5p as a hypoxia-responsive miRNAs as previous evidence29. The amount of pri-miR-21-5p and pre-miR-21-5p in the injury site after SCI was assayed by qRT-PCR, and both showed a remarkable raise (Fig.S4B and S4C). Further, after being locally administrated with N-SCs-Exos or H-SCs-Exos, the amount of pre-miR-21-5p in the injury site of the sciatic nerve was more remarkably increased than N-SCs-Exos, while neither of them affected pri-miR-21-5p amount (Fig.S4D and S4E). Similarly, the level of pre-miR-21-5p in HUVECs treated with H-SCs-Exos was higher than that of those treated with N-SCs-Exos or PBS, likewise, pri-miR-21-5p have not been changed (Fig.S4F and S4G). Our data suggest that miR-21-5p could be transferred by both N-SCs-Exos or H-SCs-Exos to the ECs rather than stimulated by them, where H-SCs-Exos enriched more miR-21-5p.
To verify whether metabolic regulation of SCs on endothelial cells was through transferring miR-21-5p, H-SCs-Exos were obtained from hypoxia-culture or normoxia-culture SCs that pretreated with an anti-miR-21-5p oligonucleotide (H-SCs-Exosanti-miR-21−5p) or with scrambled control (H-SCs-ExosNC) (Fig. 4A). Next, the mitochondrial metabolic program was detected by measuring ECAR and OCR. This showed that H-SCs-Exos-upregulated ECAR of HUVECs was counteracted by H-SCs-Exosanti-miR-21−5p (Fig. 4B). On the contrary, the decreased OCR of HUVECs by H-SCs-Exos was also rescued by H-SCs-Exosanti-miR-21−5p (Fig. 4C). The ECAR/OCR ratio of HUVECs showed an eminent increase after H-SCs-ExosNC, which was counteracted by H-SCs-Exosanti-miR-21−5p (Fig. 4D). Next, the metabolites assay showed that raised pyruvate (Fig. 4E) and lactate production (Fig. 4F) was also been counteracted, and reduced acetyl-CoA (Fig. 4G) was remarkably reversed by H-SCs-Exos anti-miR-21−5p. The western blot showed that H-SCs-Exos anti-miR-21−5p remarkably reversed H-SCs-Exos-upregulated HK2, PFKFB3, and LDHA proteins expression and downregulated PDH-E1α protein expression (Fig. 4H-4L).
In vivo, the sciatic nerve was locally administrated with PBS, H-SCs-Exos, and H-SCs-Exosanti-miR-21−5p at the injury site. Unsurprisingly, these results showed that H-SCs-Exos-augmented pyruvate (Fig. 5A) and lactate production (Fig. 5B) and reduced acetyl-CoA production (Fig. 5C) at injury-site nerve segments all have been reversed by H-SCs-Exosanti-miR-21−5p. LDHA+CD31+ cells and PDH-E1α+CD31+ cells in the injury site of the sciatic nerve were detected with immunofluorescent staining and showed that H-SCs-Exos-increased both the percentage of LDHA + CD31 + area (Fig. 5D and 5E) and the overall LDHA expression in injured sciatic nerve segments (Fig. 5D and 5F) have all been reversed by H-SCs-Exosanti-miR-21−5p. Meanwhile, H-SCs-Exos-decreased both the percentage of PDH-E1α+CD31+ area (Fig. 5G and 5H) and the overall PDH-E1α level have all been counteracted by H-SCs-Exosanti-miR-21−5p (Fig. 5G and 5I). Thereby, H-SCs-Exos-initiated intraneural revascularization has also been impeded by the knockdown of miR-21-5p (Fig. 5G and 5J). Finally, sciatic architectural regeneration has been evaluated. These results showed that the promotional effects of H-SCs-Exos on axonal regeneration (Fig. 5K and 5L) and remyelination (Fig. 5K and 5M) were significantly quenched by H-SCs-Exosanti-miR-21−5p. Unsurprisingly, paw print and SFI results showed that H-SCs-Exosanti-miR-21−5p remarkably restrained H-SCs-Exos-enhanced sciatic nerve function (Fig. 5N and 5O).
Above all data suggested that H-SCs-Exos regulated the further skew of endothelial energetic metabolism in favor of glycolysis to facilitate post-injury intraneural revascularization and nerve repair via transferring miR-21-5p.
miR-21-5p regulating endothelial metabolic program via targeting VHL and PDH-E1α
Hypoxia-induced factors 1 alpha (HIF-1α), a key transcriptional factor responsible for mediating glycolysis in the hypoxia microenvironment of the injury site 30,31, was assayed in injury-site sciatic nerve segments and HUVECs at the transcriptional and translational level. This showed that the mRNA and protein expression of HIF-1α was significant increase after SCI, and counteracted by the local injection of GW4869 (Fig. 6A and 6B).
After crush injury, sciatic nerve locally treated with Saline (Control), H-SCs-Exosanti-NC, and H-SCs-Exosanti-miR-21−5p, and the results showed that H-SCs-Exos administration enhanced the mRNA and protein expression of HIF-1α when compared to the control group, while the H-SCs-Exosanti-miR-21−5p counteracted that of HIF-1α rise (Fig. 6C and 6D). In vitro, this also showed a higher HIF-1α expression in HUVECs treated with H-SCs-Exos when compared to the PBS-treated Control group, while these effects were counteracted when HUVECs were treated with H-SCs-Exosanti-miR-21−5p (Fig. 6E and 6F). To determine whether miR-21-5p regulated HIF-1α-mediated energic metabolism to facilitate post-injury intra-neuro-revascularization, miR-21-5p mimic or inhibitor were transfected to respectively achieve the overexpression or knockdown of miR-21-5p in vivo and vitro, which was confirmed by qRT-PCR (Fig.S5A and S5B). Additionally, PX-478, a HIF-1α inhibitor that could cross the neurovascular barrier, was administrated to inactivate HIF-1α activity in the injured sciatic nerve, and this showed PX-478 significantly restrain the CD31 + revascularization of the post-injury sciatic nerve and diminished the pro-intra-neuro-revascularization effects of miR-21-5p (Fig. 6G and 6H). Metabolic level measurements showed that PX-478 administration significantly abated pyruvate (Fig. 6I) and lactate production (Fig. 6J), but increased acetyl-coA production (Fig. 6K). Overexpression of miR-21-5p performed the opposite tendency. Interestingly, PX-478 reversed the effects of miR-21-5p partly. In vitro, HUVECs co-handled with miR-21-5p and PX-478 found that increased pyruvate (Fig.S6A) and lactate production (Fig.S6B but decreased acetyl-coA (Fig.S6C) production of HUVECs. Likewise, miR-21-5p overexpression enhanced the ECAR (Fig.S6D) but inhibited the OCR of HUVECs (Fig.S6E), resulting in the raised ECAR/OCR ratio (Fig.S6F), while PX-478 reversed these effects of miR-21-5p on HUVECs in certain potent. All data suggested that miR-21-5p promoted the further tilt of ECs’ energy metabolism further toward glycolysis through enhanced HIF-1α signal to facilitate intra-neuro-revascularization after PNI.
To further determine the downstream targets of miR-21-5p, we employed three independent online bioinformatic databases, including TargetScan, miRanda, and miRTarbase. Targets of miR-21-5p contributed to HIF-1α activity prediction focused on VHL, an upstream gene mediating the ubiquitinated degradation of HIF-1α. The local administration of H-SCs-Exos further aggravated the injury-stimulated downregulation of VHL mRNA and protein, while H-SCs-Exosanti-miR-21−5p reversed this downregulation (Fig. 7A-7B). In vitro, the protein and mRNA expression of VHL was lower in HUVECs co-incubated with H-SCs-Exos, while H-SCs-Exosanti-miR-21−5p reversed H-SCs-Exos-decreased VHL expression (Fig. 7C-7D). Next, dual-luciferase report assays were performed to confirm whether miR-21-5p directly binds to the predicted target region of the VHL mRNA by co-transfecting HUVECs with a plasmid containing wild- or mutant-type VHL 3'UTR and miR-21-5p mimic or mimic NC. The predicted potential binding sites for miR-21-5p on the 3’-UTR of PDH-E1α (Fig. 7E). There was lower luciferase activity in the co-transfected wild-type and mimic group, and there was no appreciable change in the mutated group (Fig. 7F). Consistently, H-SCs-Exos decreased the luciferase activity of pVHL 3'UTR, but not that for mutant pVHL 3'UTR (Fig. 7G). Interestingly, H-SCs-Exosanti-miR-21−5p could not decrease the luciferase activity (Fig. 7G).
Further, knock downing VHL of HUVECs with si-VHL found that HIF-1α expression was remarkably upregulated at the transcriptional and translational levels (Fig.S7A and S7B). Metabolites detection found that VHL depletion with si-VHL increased the pyruvate (Fig. 7H) and lactate production (Fig. 7I), but decreased acetyl-coA production (Fig. 7J). Unsurprisingly, PX-478 reversed these phenomena (Fig. 7H-7J). VHL depletion enhanced the ECAR (Fig. 7K) and weaken the OCR (Fig. 7L) of HUVECs, increasing the ECAR/OCR ratio, while PX-478 diminished these effects. Finally, These data indicated that exosomal miR-21-5p enhanced HIF-1α-mediated endothelial glycolysis by targeting VHL.
Previous studies demonstrated that HIF-1α also inhibits the mitochondrial TCA cycle by activating pyruvate dehydrogenase kinase-1 (PDK1) to suppress PDH-E1α32–35. In this study, we found upregulated PDK1 at mRNA and protein levels after SCI, which was counteracted by GW4869 administration (Fig. 7N-7O). Interestingly, H-SCs-Exos administration further enhanced PDK1 upregulation but also was invalidated by the depletion of miR-21-5p (H-SCs-Exosanti-miR-21−5p) in vivo and vitro (Fig. 7P-7S). To determine whether exosomal miR-21-5p-increased HIF-1α accumulation led to PDK1 activation inhibiting endothelial OXPHOS, we established a co-transfected system with miR-21-5p NC/mimic and si-NC/HIF-1α. The expression of miR-21-5p (Fig.S5B) and HIF-1α (Fig.S7C-7D) was successfully overexpressed and knockdown, respectively.
Further results unveiled, at the transcriptional translation level, a remarkable increase of PDK1 and a decrease of PDH-E1α after miR-21-5p mimic treatment, surprisingly where the knockdown of HIF-1α with si-HIF-1α counteracted PDK1 increase but did not rescue PDH-E1α expression (Fig. 7T-7V). Thus, there might be another approach to mainly inhibit PDH-E1α by miR-21-5p. Reviewing the targets of miR-21-5p in the TargetScan and miRanda bioinformatic database, we found that PDH-E1α, with a conserved binding site for miR-21-5p, might be a responsible downstream target (Fig. 8A). Dual-luciferase report assays were performed to confirm whether miR-21-5p directly binds to the predicted target region of the PDH-E1α mRNA by co-transfecting HUVECs with a plasmid containing wild- or mutant-type PDH-E1α 3'UTR and miR-21-5p mimic (Fig. 8A). There was low luciferase activity in the wild-type group transfected with miR-21-5p mimic and no appreciable change in the mutated group (Fig. 8B). qRT-PCR and western blot assay showed that miR-21-5p mimic inhibited the expression of PDH-E1α, and inhibitor increased that (Fig. 8C and 8D). Thus, our data suggested that the direct inhibition of PDH-E1α is probably the main way for miR-21-5p to inhibit the endothelial OXPHOS.
Previous studies elucidated that inhibition of PDH-E1α could vicariously enhance lactate-production glycolysis. In the present study, PDH-E1α in HUVECs was knocked down via transfecting siRNA (si-PDH-E1α), and the transfection efficiency was detected (Fig. 8E-8F). This should that downregulation of PDH-E1α enhanced the ECAR (Fig. 8G) and the attenuated OCR (Fig. 8H) of HUVECs, increasing ECAR/OCR ratio (Fig. 8I); simultaneously metabolites assays showed an increase of the pyruvate (Fig. 8J) and lactate production (Fig. 8K); but a decrease of Acetyl-CoA production (Fig. 8L). Pro-angiogenic potentiality assay showed that knockdown of PDH-E1α at certain content promoted the tube formation capacity (Fig. 8M), invasion capacity (Fig. 8N), and proliferation (Fig. 8O). These data suggest that exosomal miR-21-5p-mediated PDH-E1α inhibition weakens the OXPHOS and vicariously strengthens the lactate-production glycolysis of intraneural endothelial cells to promote angiogenesis, synchronously, after PNI.
In summary, all of the above data show that H-SCs-Exos-augmented miR-21-5p transition facilitated intraneural endothelium-initiated revascularization by targeting VHL to enhance HIF-1α-accumulation-mediated glycolysis and targeting PDH-E1α to weaken mitochondria OXPHOS. Concurrently, the two pathways also reciprocally affected each other and conjointly enhanced endothelial glycolysis.