FTO-dependent m6A modification of Plpp3 in circSCMH1-regulated vascular repair and functional recovery following stroke

Abstract

Vascular repair is considered a key restorative measure to improve long-term outcomes after ischemic stroke. N⁶-methyladenosine (m⁶A), the most prevalent internal modification in eukaryotic mRNAs, functionally mediates vascular repair. However, whether circular RNA SCMH1 (circSCMH1) promotes vascular repair by m⁶A methylation after stroke remains to be elucidated. Here, we identify the role of circSCMH1 in promoting vascular repair in peri-infarct cortex of male mice and male monkeys after photothrombotic (PT) stroke, and attenuating the ischemia-induced m⁶A methylation in peri-infarct cortex of male mice after PT stroke. Mechanically, circSCMH1 increased the translocation of ubiquitination-modified fat mass and obesity-associated protein (FTO) into nucleus of endothelial cells (ECs), leading to m⁶A demethylation of phospholipid phosphatase 3 (Plpp3) mRNA and subsequently the increase of Plpp3 expression in ECs. Our data demonstrate that circSCMH1 enhances vascular repair via FTO-regulated m⁶A methylation after stroke, providing insights into the mechanism of circSCMH1 in promoting stroke recovery.

Fig. 1. CircSCMH1 promoted vascular repair during stroke recovery.

a Schematic of EV-circSCMH1 administration and staining analysis in monkeys. b, c Representative images with CD31 staining showing blood vessels in the peri-infarct cortex at day 28 after PT in monkeys, followed by the analysis of vascular area fraction, total vascular length, and the number of branches. Scale bars, 100 μm. n = 4 monkeys/group. ^*P = 0.0337, ^**P = 0.0018, ^***P = 0.0004 versus PT + EV-Vector. d Schematic of EV-circSCMH1 administration and staining analysis in mice. e, f Representative images with CD31 staining showing blood vessels in the peri-infarct cortex at day 28 after PT in mice, followed by the analysis of vascular area fraction, total vascular length, and the number of branches. Scale bars, 100 μm (overview), 20 μm (insets). n = 6 mice/group. ^***P = 0.0008 (vascular area), ^***P = 0.0004 (vascular length, numbers of branches) versus sham; ^#P = 0.0253 (vascular area), ^##P = 0.0041 (vascular length), ^##P = 0.0043 (numbers of branches) versus PT + EV-Vector. g Representative images and quantification of newly generated BrdU⁺/CD31⁺ endothelial cells in the peri-infarct cortex at day 28 after PT. Scale bars, 20 μm. n = 6 mice/group. ^**P = 0.0018 versus sham; ^###P < 0.0001 versus PT + EV-Vector. h Representative images and quantification of CD13⁺ pericyte coverage on CD31⁺ microvessels in the peri-infarct cortex at day 28 after PT. Scale bars, 20 μm. n = 6 mice/group. ^***P < 0.0001 versus sham; ^##P = 0.0048 versus PT + EV-Vector. i Schematic of long-term optical imaging platform. j, k Representative images obtained by LEDs for HBT after PT, followed by reconstruction and analysis of branch area fraction using Imaris x64 9.0.0. Scale bars, 100 μm (overview), 20 μm (insets). n = 4 mice/group. A indicates anterior; L indicates lateral. ^*P = 0.0110 (14d), ^*P = 0.0212 (21d, 28d) versus PT + EV-Vector. The data in c were expressed as mean ± SEM; Student’s t-test (two-sided). The data in f–h were expressed as mean ± SEM; one-way ANOVA followed by Holm–Sidak post hoc multiple comparison test. The data in k were expressed as mean ± SEM; two-way repeated-measures ANOVA followed by Holm–Sidak post hoc multiple comparison test. Components of this figure were created using Servier Medical Art templates, which are licensed under a Creative Commons Attribution 3.0 Unported License; https://smart.servier.com. Source data are provided as a Source Data file. BrdU 5-bromo-2’-deoxyuridine, CD13 Aminopeptidase N, d day, ibz ischemic border zone, Pre pre-injury.

Fig. 2. Increased m⁶A modification and decreased FTO in AIS patients and PT mice.

a, b Total RNA was extracted from the somatosensory cortex of AIS or nonstroke patients. The m⁶A level was determined as the percentage of all adenosine residues in RNA (a). The level of circSCMH1 was measured by qPCR (b). There were six individuals/group. ^*P = 0.0183 (a), ^*P = 0.0184 (b) versus nonstroke. c Western blotting analysis of FTO in the somatosensory cortex of AIS and nonstroke patients. There were six individuals/group. ^*P = 0.0105 versus nonstroke. d Total RNA was extracted from the peri-infarct cortex of PT mice, m⁶A levels were determined as the percentage of all adenosine residues in RNA. n = 6 mice/group. ^**P = 0.0018 (28d), ^***P = 0.0002 (21d), ^***P < 0.0001 (1d, 3d, 7d, 14d) versus sham. e–g Western blotting analysis of FTO in the peri-infarct cortex of mice after PT (e), dMCAO (f), and tMCAO (g) stroke. Two representative immunoblots were presented from 6 mice/group. ^*P = 0.0186 (21d), ^*P = 0.0408 (28d), ^***P < 0.0001 (1d, 3d, 7d), ^***P = 0.0001 (14d) versus sham in e. ^**P = 0.0012 (21d), ^**P = 0.0057 (28d), ^***P < 0.0001 (1d, 3d, 7d), ^***P = 0.0002 (14d) versus sham in f. ^**P = 0.0064 (21d), ^**P = 0.0070 (28d), ^***P < 0.0001 (1d, 3d, 7d, 14d) versus sham in g. The data in a–c were expressed as mean ± SEM; Student’s t-test (two-sided). The data in d–g were expressed as mean ± SEM; one-way ANOVA followed by Holm–Sidak post hoc multiple comparison test. Source data are provided as a Source Data file. AIS acute ischemic stroke, d day, dMCAO distal middle cerebral artery occlusion, FTO fat mass and obesity-associated protein, m⁶A N⁶-methyladenosine, PT photothrombotic, tMCAO transient middle cerebral artery occlusion.

Fig. 3. CircSCMH1 decreased m⁶A modification and bound with FTO.

a Total RNA was extracted from the peri-infarct cortex of PT mice, m⁶A levels were determined as the percentage of all adenosine residues in RNA. n = 6 mice/group. ^**P = 0.0072 (28d), ^***P < 0.0001 (3d, 14d) versus the sham; ^##P = 0.0056 (3d), ^##P = 0.0097 (14d), ^##P = 0.0024 (28d) versus the PT + EV-Vector. b Total RNA was extracted from the primary mouse brain microvascular ECs treated with circSCMH1 plasmid at 12 h after OGD, and m⁶A levels were determined as the percentage of all adenosine residues in RNA. Data were presented by three independent experiments. ^*P = 0.0231 versus Con+Vector; ^##P = 0.0051 versus OGD + Vector. c Interaction between circSCMH1 and FTO was detected by RNA-binding immunoprecipitation in the primary brain microvascular ECs. Data were presented by three independent experiments. ^***P < 0.0001 versus FTO pull-down of circHECW2; ^###P < 0.0001 versus FTO pull-down of Gapdh mRNA. d Interaction between circSCMH1 and FTO was measured by RNA pull-down assay in the primary mouse brain microvascular ECs. Data were presented by three independent experiments. ^**P = 0.0085 versus the circCon probe. e Prediction of circSCMH1-FTO interaction by catRAPID algorithm. f The interaction between circSCMH1 and FTO was validated by RNA immunoprecipitation in bEnd.3 cells with WT FTO and mutant FTO. Data were presented by three independent experiments. ^**P = 0.0022 versus circSCMH1 in FTO-WT. g Western blot analysis of FTO expression in lysates of bEnd.3 cells with circSCMH1 or mutated circSCMH1 (Δ426–477) overexpression following biotinylated circSCMH1 probe pull-down assay. Data were presented by three independent experiments. ^**P = 0.0069 versus WT. The data in a, c, f were expressed as mean ± SEM; one-way ANOVA followed by Holm–Sidak post hoc multiple comparison test. The data in b were expressed as mean ± SEM; two-way ANOVA followed by Bonferroni’s post hoc multiple comparison tests. The data in d and g were expressed as mean ± SEM; using the Student t-test (two-sided). Source data are provided as a Source Data file. Con control, d day, IgG immunoglobulin G, IP immunoprecipitation, WT wild type. Δ151–202: lacking region 151 to 202 amino acids; Δ351–402: lacking region 351 to 402 amino acids; Δ426–477: lacking region 426 to 477 amino acids or nucleic acids.

Fig. 4. CircSCMH1 altered the subcellular localization of FTO via ubiquitination.

a, b FTO expression in peri-infarct tissue’s cytoplasm (a) and nucleus (b) at day 28 after PT. Three representative immunoblots were presented from 6 mice/group. ^***P < 0.0001 (a, b) versus sham, ^###P < 0.0001 (a), ^###P = 0.0008 (b) versus PT + EV-Vector. c, d Representative western blotting of FTO expression in the cytoplasm (c) and nucleus (d) of the primary mouse brain microvascular ECs at 12 h after OGD. Data were presented by three independent experiments. ^**P = 0.0012 (c), ^**P = 0.0029 (d) versus Con+Vector; ^#P = 0.0235, (d) ^##P = 0.0039 (c) versus OGD + Vector. e Immunoprecipitation detected Ub-K63 modification of FTO in primary mouse brain microvascular ECs at 12 h after OGD. Data were representative of three independent experiments. f Interaction between circSCMH1 and UBC13 was detected by RNA-binding immunoprecipitation in bEnd.3 cells. Data were presented by three independent experiments. ^**P = 0.0076 versus UBC13 pull-down of circHECW2; ^##P = 0.0011 versus UBC13 pull-down of Gapdh mRNA. g Immunoprecipitation showed the binding of FTO with UBC13 in bEnd.3 cells. Data were presented by three independent experiments. ^*P = 0.0210 versus vector. h The bEnd.3 cells were transfected with siUBC13, the level of Ubc13 mRNA was measured by qPCR. Data were presented by three independent experiments. ^**P = 0.0020 versus siCon. i, j Western blot analysis of FTO expression in the cytoplasm (i) and nucleus (j) of bEnd.3 cells with LV-circSCMH1 and siUBC13 at 12 h after OGD. Data were presented by three independent experiments. ^***P = 0.0004 (i), ^***P = 0.0007 (j) versus Con+LV-Vector+siCon; ^##P = 0.0073 (i), ^##P = 0.0019 (j) versus OGD + LV-Vector + siCon; ^†P = 0.0184 (j), ^††P = 0.0018 (i) versus OGD + LV-circSCMH1+siCon. k Proposed model of the regulatory role of circSCMH1 and UBC13 for FTO translocation into the nucleus. The data in a, b, f, i, j were expressed as mean ± SEM; one-way ANOVA followed by Holm–Sidak post hoc multiple comparison test. The data in c, d were expressed as mean ± SEM; two-way ANOVA followed by Bonferroni’s post hoc multiple comparison tests. The data in g, h were expressed as mean ± SEM; using the Student t-test (two-sided). Source data are provided as a Source Data file. Con control, IB immunoblot, siUBC13 Ubc13 siRNA, Ub-K63 lysine 63-linked ubiquitination.

Fig. 5. Plpp3 was one of the critical target genes of circSCMH1-regulated m⁶A modification in the peri-infarct cortex of PT mice.

a Distribution of m⁶A peaks across 5′-UTR, CDS, and 3′-UTR of mRNA at day 14 after PT. b Venn diagram showing numbers of genes with significant changes in expression (up: fold change ≥ 2, P < 0.05; down: fold change ≤ 0.5, P < 0.05, rescaled hypergeometric test). c The m⁶A level of Plpp3 transcript was regulated in PT mice after EV-circSCMH1 administration. d, e Effect of EV-circSCMH1 on Plpp3 mRNA (d) and LPP3 (e) levels in mice at day 14 after PT. n = 6 mice/group. Three representative immunoblots were presented from 6 mice/group. ^***P < 0.0001 (d), ^***P = 0.0003 (e) versus sham; ^##P = 0.0030 (e), ^###P < 0.0001 (d) versus PT + EV-Vector. f, g Effect of circSCMH1 plasmid on the expression of Plpp3 mRNA (f) and LPP3 (g) in primary mouse brain microvascular ECs after OGD. Data were presented by three independent experiments. *P = 0.0171 (g), ^***P < 0.0001 (f) versus Con + Vector; ^#P = 0.0375 (g), ^###P = 0.0003 (f) versus OGD + Vector. h Specific primers against m⁶A peak were designed to amplify m⁶A peak of Plpp3 transcript in RNA from the peri-infarct cortex of mice at day 14 after PT. n = 6/each group. ^***P < 0.0001 versus the sham; ^###P < 0.0001 versus the PT + EV-Vector. i Specific primers against the m⁶A peak were designed to amplify the m⁶A peak of Plpp3 transcript in bEnd.3 cells. Data were presented by three independent experiments. ^***P = 0.0002 versus Con + Vector; ^###P = 0.0007 versus OGD + Vector. j, k The bEnd.3 cells were transfected with LV-circSCMH1 and shFTO. The Plpp3 mRNA (j) and LPP3 (k) was detected at 12 h after OGD. Data were presented by 4 (j) or 3 (k) independent experiments. **P = 0.0020 (k), ^***P = 0.0003 (j) versus Con + LV-Vector + shCon; ^##P = 0.0012 (j), ^##P = 0.0087 (k) versus OGD + LV-Vector + shCon; ^†P = 0.0408 (k), ^††P = 0.0083 (j) versus OGD + LV-circSCMH1 + shCon. The data in d, e, h, j, k were expressed as mean ± SEM; one-way ANOVA followed by Holm–Sidak post hoc multiple comparison test. The data in f, g, i were expressed as mean ± SEM; two-way ANOVA followed by Bonferroni’s post hoc multiple comparison tests. Source data are provided as a Source Data file. CDS coding region, shRNA short hairpin RNA, 3′ UTR 3′ untranslated regions, 5′ UTR 5′ untranslated regions.

Fig. 6. Endothelial-targeted FTO overexpression promoted vascular repair in PT mice.

a Schematic of AAV-BR1-FTO administration and staining analysis. b, c Representative images with CD31 staining showing blood vessels in the peri-infarct cortex at day 28 after PT in mice, followed by the analysis of vascular area fraction, total vascular length, and the number of branches. n = 6 mice/group. Scale bars, 100 μm (overview), 20 μm (insets). ^***P < 0.0001 (vascular area, numbers of branches), ^***P = 0.0001 (vascular length) versus sham + AAV-BR1-Con; ^##P = 0.0075 (vascular area), ^##P = 0.0071 (vascular length), ^###P < 0.0001 (numbers of branches) versus PT + AAV-BR1-Con. d Representative images and quantification of newly generated BrdU⁺/CD31⁺ endothelial cells at day 28 after PT. n = 6 mice/group. Scale bars, 20 μm. ^*P = 0.0101 versus sham + AAV-BR1-Con; ^###P < 0.0001 versus PT + AAV-BR1-Con. e Representative images and quantification of CD13⁺ pericyte coverage on CD31⁺ microvessels in the peri-infarct cortex at day 28 after PT. Scale bars, 20 μm. n = 6 mice/group. ^**P = 0.0054 versus the sham+AAV-BR1-Con group; ^#P = 0.0485 versus PT + AAV-BR1-Con. f Representative images obtained by using LEDs at λ = 570 nm for the HBT at day 4, 7, 14, 21, and 28 after PT in Fto^flox/flox cKI mice with AAV-BR1-Con or AAV-BR1-Cre, followed by 2D reconstruction and analysis of branch area fraction using Imaris x64 9.0.0. n = 3 mice/group. Scale bars, 100 μm (overview), 20 μm (insets). ^*P = 0.0112 (14d), ^**P = 0.0027 (21d), ^**P = 0.0019 (28d) versus AAV-BR1-Con. The data in c–e were expressed as mean ± SEM; two-way ANOVA followed by Bonferroni’s post hoc multiple comparison tests. The data in f were expressed as mean ± SEM; two-way repeated-measures ANOVA followed by Holm–Sidak post hoc multiple comparison test. Components of this figure were created using Servier Medical Art templates, which are licensed under a Creative Commons Attribution 3.0 Unported License; https://smart.servier.com. Source data are provided as a Source Data file. AAV adeno-associated virus, cKI conditional knock-in, Con control, ITR inverted terminal repeat, LSL loxP-STOP-loxP, Pre pre-injury, WPRE woodchuck hepatitis virus post-transcriptional regulation.

Fig. 7. Endothelial-targeted FTO overexpression promoted motor functional recovery in PT mice.

a Schematic of AAV-BR1-FTO administration and behavioral analysis. b–d Endothelial-targeted FTO overexpression improved behavioral recovery at different time points after PT as measured by the grid-walking test, cylinder test, and adhesive removal test. L indicates left forepaw in cylinder test; R, right forepaw in cylinder test; B, both forepaws in cylinder test. n = 19 mice for PT + AAV-BR1-Con group or PT + AAV-BR1-FTO group. ^*P = 0.0408 (7d), ^**P = 0.0039 (4d), ^***P = 0.0005 (14d) versus the PT + AAV-BR1-Con in b. ^*P = 0.0400 (4d), ^*P = 0.0444 (7d) versus the PT + AAV-BR1-Con in c. ^*P = 0.0391 (4d), ^*P = 0.0391 (7d), ^*P = 0.0347 (14d), ^*P = 0.0391 (21d), ^*P = 0.0391 (28d) versus the PT + AAV-BR1-Con in d. e Schematic of AAV-BR1-Cre administration and behavioral analysis. f–h Fto^flox/flox cKI mice with endothelial-targeted Cre expression showed behavioral recovery at different time points after PT as measured by the grid-walking test, cylinder test, and adhesive removal test. L indicates left forepaw in cylinder test; R, right forepaw in cylinder test; B, both forepaws in cylinder test. n = 11 mice for PT + AAV-BR1-Con group or PT + AAV-BR1-Cre group. ^***P < 0.0001 (4d), ^***P = 0.0007 (7d), ^***P < 0.0001 (14d), ^***P < 0.0001 (21d), ^***P < 0.0001 (28d) versus the PT + AAV-BR1-Con in f. ^*P = 0.0442 (7d) versus the PT + AAV-BR1-Con in g. ^*P = 0.0348 (14d), ^**P = 0.0034 (4d) versus the PT + AAV-BR1-Con in h. The data in b–d, f–h were expressed as mean ± SEM; two-way repeated-measures ANOVA followed by Holm–Sidak post hoc multiple comparison test. Source data are provided as a Source Data file. AAV adeno-associated virus, circSCMH1 circular RNA SCMH1, d day, FTO fat mass and obesity-associated protein, PT photothrombotic.

Fig. 8. Schematic illustration of the enhancement effect of EV-circSCMH1 on vascular repair after ischemic stroke.

Intravenous administration of EV-circSCMH1 after ischemic stroke significantly enhanced vascular repair in mouse and monkey models. The level of m⁶A was significantly increased in the somatosensory cortex of AIS patients and the peri-infarct cortex of mouse stroke models. Endothelial-targeted FTO overexpression significantly improved functional recovery in mice by enhancement of vascular repair. In ECs, circSCMH1 bound with FTO and promoted its transfer to the nucleus by Ub-K63, which decreased the m⁶A methylation of Plpp3 mRNA. Components of this figure were created using Servier Medical Art templates, which are licensed under a Creative Commons Attribution 3.0 Unported License; https://smart.servier.com. CircSCMH1 circular RNA SCMH1, EV extracellular vesicle, FTO fat mass and obesity-associated protein, m⁶A N⁶-methyladenosine, Plpp3/LPP3 lipid phosphate phosphatase 3, Ub-K63 lysine 63-linked ubiquitination.