Preferential delivery of lipid-ligand conjugated DNA/RNA heteroduplex oligonucleotide to ischemic brain in hyperacute stage

Abstract

Antisense oligonucleotide (ASO) is a major tool used for silencing pathogenic genes. For stroke in the hyperacute stage, however, the ability of ASO to regulate genes is limited by its poor delivery to the ischemic brain owing to sudden occlusion of the supplying artery. Here we show that, in a mouse model of permanent ischemic stroke, lipid-ligand conjugated DNA/RNA heteroduplex oligonucleotide (lipid-HDO) was unexpectedly delivered 9.6 times more efficiently to the ischemic area of the brain than to the contralateral non-ischemic brain and achieved robust gene knockdown and change of stroke phenotype, despite a 90% decrease in cerebral blood flow in the 3 h after occlusion. This delivery to neurons was mediated via receptor-mediated transcytosis by lipoprotein receptors in brain endothelial cells, the expression of which was significantly upregulated after ischemia. This study provides proof-of-concept that lipid-HDO is a promising gene-silencing technology for stroke treatment in the hyperacute stage.

Graphical abstract

Takanori Yokota and colleagues demonstrated that systemic administration of α-tocopherol-ligand conjugated DNA/RNA heteroduplex oligonucleotide is selectively delivered into the ischemic brain region, even in the hyperacute phase after ischemic stroke, achieved significant gene-silencing efficacy and changed stroke phenotype in mice.

Introduction

Stroke remains a major cause of mortality and long-term disability worldwide. Over the past few decades, major advances in acute ischemic stroke therapy using intravenous (i.v.) recombinant tissue plasminogen activator (rt-PA) or an endovascular procedure have increased the chances of stroke recovery and survival.^1,2 However, because the use of these therapies is severely restricted by their short therapeutic time windows,^1,2 there is an urgent need to develop other drug options. The molecular cascade of pathophysiological events occurring after ischemic stroke has been researched extensively, and several genes of interest have become available as therapeutic targets for improving ischemic recovery.^3,4 Nonetheless, our ability to modulate the target gene in ischemic brain remains limited by the poor delivery of therapeutic agents into the brain parenchyma, especially during the hyperacute phase of a stroke.

The brain parenchyma is distinctively separated from the peripheral blood circulation by the blood-brain barrier (BBB), and the movement of molecules through the endothelial cell (EC) layer is strictly restricted under heathy conditions.⁵ The passage of molecules through the BBB is controlled by transcellular and paracellular pathways, and the paracellular barrier starts to break down 24 to 48 h post stroke, a relatively late time point, by which time, in mice, most neurons are irreversibly damaged.⁶ Consequently, during the acute phase (<24 h) of an ischemic stroke, when many salvageable neurons still remain, a drug administered systemically will not be delivered into the ischemic brain region via paracellular pathways. Rather, the movement of molecules via transcellular pathways, including endocytosis and transcytosis (e.g., receptor-mediated transcytosis of brain ECs), increases as early as 3 to 12 h after a stroke following middle cerebral artery occlusion (MCAO).^6,7 We recently found that, during the hyperacute phase (as early as 3 h) following MCAO in mice, the expression of lipoprotein receptors involved in receptor-mediated transcytosis across the BBB and endocytosis in neurons, including low-density lipoprotein receptor (LDLR), scavenger receptor class B member 1 (SRB1), and LDLR-related protein 1 (LRP1), clearly increased on brain ECs and neurons in the ischemic cortex.⁸ These findings suggest that lipid components in these lipoproteins, such as α-tocopherol (Toc, a natural isomer of vitamin E)^9,10 and cholesterol (Cho), could be suitable ligands for drug delivery via blood vessels into the ischemic brain in the hyperacute phase.

We recently developed a short DNA/RNA heteroduplex oligonucleotide (HDO; Figure S1; Table S1), composed of DNA oligonucleotides flanked by locked nucleic acid (LNA) oligonucleotides as the ASO strand and complementary RNA oligonucleotides as the complementary RNA (cRNA) strand.^{11,12,13,14,15,16} In previous studies, we showed that vitamin-E-conjugated HDO, a molecule in which α-tocopherol is covalently conjugated to the cRNA strand of HDO (Toc-HDO), binds to serum lipoproteins via the hydrophobic property of tocopherol in the circulating blood and is distributed along the physiological transport pathway of α-tocopherol.^13,14 In the liver, i.v. administration of Toc-HDO is up to 20 times more efficient in silencing genes than administration of the parent ASO, and the uptake of Toc-HDO into hepatocytes is mediated by LDLR.¹³

Therefore, here, in hyperacute ischemic stroke, we tested the use of i.v.-administered lipid-HDO (either Toc-HDO or Cho-HDO; Figure S1; Table S1) for gene regulation in the ischemic area of the brain.

Results

Malat1 RNA levels increase in neurons and brain ECs after ischemia in vitro and in vivo

We selected the long non-coding RNA (lncRNA), metastasis-associated lung adenocarcinoma transcript 1 (Malat1), as a target gene of ischemia-inducible RNA, because Malat1 is highly upregulated in the ischemic brain after MCAO and has been shown by using Malat1 knockout (KO) mice to modulate angiogenesis, inflammation, and apoptosis.^{17,18,19,20,21}

To evaluate Malat1 RNA levels after ischemia in vitro, Neuro2A cells (mouse neurons) and bEnd3 cells (mouse ECs) were exposed to oxygen-glucose deprivation (OGD) for 2 and 4 h. Quantitative reverse transcription (qRT)-PCR assays showed that Malat1 RNA levels were significantly greater after OGD treatment than in untreated controls. At 2 and 4 h, the levels in neurons were about 3.9 and 3.2 times, respectively, that in controls (Figure 1A); the corresponding levels in brain ECs relative to controls were about 1.4 and 1.3 times at 2 and 4 h (Figure 1B). We then conducted in vivo experiments exploring mouse brains following permanent MCAO (pMCAO) surgery, in which the ischemic region was confined to the ipsilateral cerebral cortex in the territory of the middle cerebral artery (MCA).²² Four days after pMCAO, Malat1 RNA levels in the ipsilateral (ischemic) and contralateral (non-ischemic) cortexes were respectively about 3.7 and 2.3 times those in sham-operated controls (Figure 1C). Images obtained by using the ViewRNA (Affymetrix, Tokyo, Japan) in situ hybridization (ISH) technique further showed that, 4 days after pMCAO, Malat1-RNA-positive signals were markedly increased relative to those in controls (Figure 1D) in both the ipsilateral ischemic (Figure 1E) and the contralateral non-ischemic (Figure 1F) cortexes, consistent with the results of the qRT-PCR analysis. Malat1-RNA-positive signals were observed in cells, including neurons in which the nucleolus was clearly visualized and the ECs lining microvessels.

Figure 1.

Increased expression of Malat1 in neurons and ECs exposed to OGD and in the brain after pMCAO

Measurement of Malat1 levels with qRT-PCR analyses in Neuro2A (A) and bEnd3 (B) cells 2 or 4 h after oxygen–glucose deprivation (OGD) (n = 3, mean values ±SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 versus controls; one-way ANOVA, followed by Dunnett’s post hoc test). (C) Malat1 levels increased in both the ipsilateral (ischemic) cortex and the contralateral non-ischemic cortex 4 days after pMCAO (n = 3, mean values ±SEM. ∗p < 0.05, ∗∗p < 0.01 versus sham-operated controls; two-way ANOVA, followed by Bonferroni’s post hoc test). ViewRNA in situ hybridization (ISH) for Malat1 RNA (red) in the cerebral cortexes of sham-operated controls (D, ipsilateral side) and pMCAO animals 4 days after pMCAO (E, ipsilateral ischemic cortex; F, contralateral non-ischemic cortex). ISH images are representative of three independent experiments. Scale bar in (D)–(F), 50 μm.

High gene-silencing efficacy and safety of lipid-HDO via i.v. administration

To test the gene-silencing efficacy of lipid-HDO in hyperacute ischemic stroke, 3 h after pMCAO, we gave mice i.v. injections (50 mg/kg) of phosphate-buffered saline (PBS, control), ASO, lipid-ASO (Toc-ASO), HDO, lipid-HDO (either Toc-HDO or Cho-HDO) targeting Malat1, or shuffle Toc-HDO targeting a scrambled sequence of Malat1. In the ipsilateral ischemic cortex 3 days after i.v. administration, Malat1 RNA levels were significantly lower in the Toc-HDO (69.9% reduction) and Cho-HDO (84.9% reduction) groups than in the PBS controls but not in the ASO, HDO, and shuffle Toc-HDO groups (Figure 2A). Both Toc-HDO and Cho-HDO had significant gene-silencing effects in the ischemic cortex, but there were no significant differences in gene-silencing capacity between them. Importantly, Toc-HDO showed significantly greater silencing of Malat1 than ASO, Toc-ASO, or HDO (Figure 2A). In the contralateral non-ischemic cortex, Malat1 RNA levels were significantly lower in the Cho-HDO group (52.9% reduction) than in the PBS controls, but Malat1 RNA levels did not differ significantly among the other treatment groups (Figure 2B). These results indicated that the selectivity of gene-silencing efficacy in the ischemic brain might be highest for Toc-HDO.

Figure 2.

High gene-silencing efficacy of lipid-HDO on ischemic cortex lesion and ECs in the hyperacute phase after stroke

qRT-PCR analyses of Malat1 RNA levels in the ischemic cortex (ipsilateral) (A) and the contralateral non-ischemic cortex (B) 3 days after i.v. administration (50 mg/kg) of PBS (vehicle controls), ASO, Toc-ASO, HDO, Toc-HDO, or Cho-HDO targeting Malat1, or shuffle Toc-HDO targeting a scrambled sequence of Malat1 (n = 3, mean values ±SEM. ∗∗∗p < 0.001 versus PBS controls, one-way ANOVA, followed by Dunnett’s post hoc test; †p < 0.05 versus ASO, two-tailed, unpaired t test; ‡p < 0.05 versus ASO, two-tailed, unpaired t test; §p < 0.05 versus HDO, two-tailed, unpaired t test). (C) qRT-PCR of measurement of Malat1 RNA levels in fractionated brain ECs purified from half-brain samples. Analyses were done 3 days after i.v. administration of Toc-HDO targeting Malat1 (50 mg/kg) or PBS alone, via the tail vein 3 h after pMCAO (n = 3, mean values ±SEM. ∗p < 0.05 versus PBS controls; two-way ANOVA, followed by Bonferroni’s post hoc test). (D) qRT-PCR analyses 3 days after i.v. administration of 0, 12.5, 25, or 50 mg/kg Toc-HDO targeting Malat1 shows dose-dependent reduction of gene silencing in both the ischemic cortex and the contralateral non-ischemic cortex (n = 3, mean values ±SEM. ∗∗∗p < 0.001 versus PBS controls; two-way ANOVA, followed by Tukey’s post hoc test). In situ hybridization for Malat1 RNA in the ischemic cortex (E) and the contralateral non-ischemic cortex (F) 3 days after i.v. administration of PBS (left) or Toc-HDO (right) 3 h after pMCAO. Scale bar in (E) and (F), 50 μm.

To evaluate the safety of Toc-HDO, we gave mice i.v. injections (50 mg/kg) of PBS or Toc-HDO targeting Malat1 3 h after pMCAO surgery and performed serum biochemical and histological analyses. Three days after pMCAO surgery, levels of blood urea nitrogen or serum creatinine, aspartate transaminase, alanine transaminase, total bilirubin, or lactate dehydrogenase were not elevated in the Toc-HDO group compared with the PBS group (Figure S2A). In addition, hematoxylin and eosin staining of kidney and liver sections revealed no histological abnormalities in the Toc-HDO group compared with the PBS group (Figure S2B).

On the basis of these results, we decided to test Toc-HDO as a representative lipid-HDO in the following experiments. In fractionated brain ECs, we similarly found as our first experiment that, relative to PBS controls, Malat1 RNA levels in the Toc-HDO group were significantly reduced in ischemic but not non-ischemic brain ECs relative to those in the PBS controls (Figure 2C). We also tested the dose dependence of Toc-HDO in hyperacute ischemic stroke. We found that Toc-HDO had a dose-dependent effect in the ischemic cortex (median effective dose [ED₅₀] 34.3 mg/kg) but no such effect in the contralateral non-ischemic cortex (Figure 2D). Similarly, ISH experiments performed in the brain 3 days after i.v. administration of PBS or Toc-HDO revealed that Malat1-RNA-positive signaling in the ischemic cortex was markedly lower in the Toc-HDO group than in PBS controls (Figure 2E), consistent with the qRT-PCR results (Figure 2A). Malat1-RNA-positive signals were not clearly reduced in the contralateral non-ischemic cortex (Figure 2F).

Efficient delivery of Toc-HDO into the ischemic cortex during the hyperacute phase of ischemic stroke

To confirm the delivery effects of Toc-HDO targeting Malat1, we investigated the oligonucleotide content and distributions in the cerebral cortex during hyperacute phase (3 h, 6 h) of ischemic stroke, following i.v. administration (50 mg/kg) of Alexa Fluor-labeled ASO or Toc-HDO immediately or 3 h after pMCAO surgery. The oligonucleotide content in the Toc-HDO group, but not in the ASO group, was significantly higher in the ischemic cortex than in the contralateral non-ischemic cortex at both 3 h and 6 h after pMCAO (over 9.6- and 4.3-fold, respectively; Figures 3A and 3B). In addition, the oligonucleotide content in the ischemic cortex, but not in the non-ischemic cortex, was significantly higher in the Toc-HDO group than in the ASO group at both 3 h and 6 h after pMCAO (over 16.9- and 6.7-fold, respectively; Figures 3A and 3B), indicating that hyperacute ischemia enhanced the delivery of Toc-HDO into the ischemic cortex. Next, we used confocal laser scanning to detect Alexa Fluor signals in the brain. Consistent with the quantification of oligonucleotide contents, signals in the Toc-HDO group were higher in the ischemic cortex (Figures 3C and 3D) than in the equivalent area of the contralateral non-ischemic cortex (Figure 3E). In the Toc-HDO group, signals were observed on brain parenchymal cells, leptomeninges, leptomeningeal arteries, arterioles, and microvessels in the ischemic cortex (Figures 3C and 3D). In contrast, in the ASO group, weak signals (red dots) were observed on parts of microvessels but not on brain parenchymal cells (Figures 3F–3H). These results indicate that Toc-HDO, but not ASO, could be effectively delivered into brain ECs and parenchymal cells in ischemic regions, even in the hyperacute phase of ischemic stroke.

Figure 3.

Quantification and distribution of i.v. administered oligonucleotides in the hyperacute phase after ischemic stroke

(A and B) Alexa Fluor 647-labeled ASO or Toc-HDO targeting Malat1 (50 mg/kg) was injected into the tail vein immediately or 3 h after pMCAO surgery. (A) Concentrations of brain oligonucleotide content in the ischemic and contralateral non-ischemic cortexes 3 h after pMCAO surgery (n = 3, mean values ±SD. ∗∗∗p < 0.001 versus contralateral non-ischemic cortex; †††p < 0.001 versus ASO; two-way ANOVA, followed by Bonferroni’s post hoc test). (B) Concentrations of brain oligonucleotide content in the ischemic and contralateral non-ischemic cortexes 6 h after pMCAO surgery (n = 3, mean values ±SD. ∗∗∗p < 0.001 versus contralateral non-ischemic cortex; †††p < 0.001 versus ASO; two-way ANOVA, followed by Bonferroni’s post hoc test). Immunofluorescence images in the ischemic cortex (C, D, F, and G) and the contralateral non-ischemic cortex (E and H) following i.v. administration of Alexa Fluor 568-labeled ASO or Toc-HDO 6 h after pMCAO. Red, Alexa Fluor 568-labeled oligonucleotides (ASO or Toc-HDO). Scale bar in (C)–(H), 100 μm. Toc-HDO targeting Malat1 (50 mg/kg) was injected via the tail vein 3 h after pMCAO surgery. Shown are concentrations of Toc-cRNA (I) and DNA/LNA gapmer (ASO) (J) in the ischemic cortex and the contralateral non-ischemic cortex 6 h after pMCAO surgery, as measured by HELISA (n = 3, mean values ±SEM.∗p < 0.05 versus ASO; Student’s two-tailed t test).

To better understand the kinetics of the drug, we used hybridization-based enzyme-linked immunosorbent assay (HELISA) to evaluate the contents of heteroduplex (Toc-HDO) and single-stranded ASO derived from Toc-HDO in the cerebral cortexes and the liver during the hyperacute phase (6 h) of ischemic stroke, following i.v. administration (50 mg/kg) of Toc-HDO 3 h after pMCAO surgery.

After the distribution of Toc-HDO into the tissues, the α-tocopherol-conjugated cRNA strand (Toc-cRNA) of Toc-HDO is cleaved immediately into short cRNA fragments by a nuclease. Through this process, single-stranded ASO is released and acts as an antisense oligonucleotide in target cells for gene regulation.¹³ Therefore, we measured the content of uncleaved, full-length Toc-cRNA and used it to represent Toc-HDO in the HELISA assay. In this context, cleaved, short Toc-cRNA fragments were not expected to be detectable in HELISA.

We found that the contents of both uncleaved Toc-RNA and single-stranded ASO were higher in the ischemic cortex than in the contralateral non-ischemic cortex (2.2- and 5.8-fold, respectively; Figures 3I and 3J). In addition, over 99% of the drug existed as single-stranded ASO in the brain cortexes and the liver (Figure S3; Table S2) These results indicated that Toc-HDO was delivered preferentially into the ischemic brain region rather than the non-ischemic brain and processed immediately, releasing the active ASO component, which regulated the target gene in the hyperacute phase.

Toc-HDO markedly transported into brain ECs and neurons

To examine the cell types into which Toc-HDO was transported in the hyperacute phase of ischemic stroke, we stained Toc-HDO-injected brains with cell-specific markers. z stack images in which signals of Toc-HDO and CD31 (also known as PECAM-1, which detects ECs) or anti-neuronal nuclei (NeuN, which detects neurons) were colocalized, indicating that Toc-HDO could be transported into brain ECs (Figure 4A) and neurons (Figure 4B) in the ischemic cortex, even in hyperacute ischemic stroke. The signals, however, were not clearly colocalized with other brain-cell markers, such as anti-ionized calcium-binding adapter molecule 1 (IBA1; a microglial marker) or anti-glial fibrillary acidic protein (GFAP; an astrocyte marker) (Figures S4A and S4B).

Figure 4.

Distribution of i.v.-administered Toc-HDO on brain ECs and neurons in ischemic cortex

(A and B) Orthogonal reconstruction from a confocal z series represented as if viewed in the x–z (top) and z–y (right) panels, showing colocalization of Alexa Fluor 568-labeled Toc-HDO (red, 6 h after pMCAO) and CD31 (green) or NeuN (green) signals in the ischemic cortex. Blue, Hoechst 33342; scale bar, 100 μm. (C) In the protocol, 50 mg/kg Toc-HDO targeting Malat1 was injected i.v. 3 h after pMCAO. (D) PBS control image was obtained 3 h after i.v. administration. (E–H) Representative images show phosphorothioate-positive signals in the ischemic cortex 3, 9, 24, or 72 h after i.v. administration of 50 mg/kg Toc-HDO targeting Malat1 3 h after pMCAO. ECs lining microvessels, green arrows; neuron-like cells, blue arrows. Scale bars in (D)–(F), 10 μm.

To further confirm the temporal profiles of Toc-HDO transported to the brain after ischemic stroke, we injected Toc-HDO targeting Malat1 (50 mg/kg) without Alexa Fluor labeling 3 h after pMCAO, and 3, 9, 24 or 72 h after i.v. administration we collected the brains for histological analyses (Figure 4C). Toc-HDO expression, as obtained by immunohistochemical staining with phosphorothioate, which chemically modifies the internucleotide linkage of Toc-HDO, showed that phosphorothioate-positive signals were stronger in the ischemic cortex (Figures 4E–4H) than in the contralateral non-ischemic cortex (Figures S5B–S5E). In the hyperacute phase of ischemic stroke, phosphorothioate-positive signals in the ischemic cortex were strong, primarily in the ECs lining microvessels, 3 and 9 h after i.v. administration (Figures 4E and 4F); they became stronger by 9 h after i.v. administration but were weak in neuron-like cells. In addition, in the late phase, phosphorothioate-positive signals in the ischemic cortex remained strong, but mainly in neuron-like cells, especially in the nucleus, 24 and 72 h after i.v. administration (Figures 4G and 4H); they became stronger by 72 h after i.v. administration. However, in the contralateral non-ischemic cortex, very weak phosphorothioate-positive signals were detected in the ECs lining microvessels and only in the hyperacute phase of ischemic stroke (e.g., 3 h after i.v. administration) (Figures S5B and S5C). These data indicated that, starting from the hyperacute phase, Toc-HDO accumulated mainly in the ECs lining microvessels (Figures 4E and 4F), whereas they were delivered chiefly into neuron-like cells, and especially into the nucleus, in the late phase (Figures 4G and 4H). These data suggested that Toc-HDO could be successfully transported into ECs and neurons in ischemic regions as early as 3 h after ischemic stroke, and that it was further transported mainly into brain parenchymal cells (e.g., neurons) in the late phase.

Upregulation of lipoprotein receptors in hyperacute ischemic stroke

We used qRT-PCR analyses to quantify messenger RNA (mRNA) levels of the lipoprotein receptors LDLR, SRB1, and LRP1 immediately after sham operation and 3 h, 6 h, and 3 days after pMCAO. LDLR mRNA levels in the ischemic cortex were significantly greater than sham control levels in the hyperacute phase both 3 and 6 h after pMCAO, but they were significantly lower than the sham levels in the late phase 3 days after pMCAO (Figure 5A). Similarly, mRNA levels of both SRB1 and LRP1 were significantly greater than sham control levels 6 h after pMCAO (Figures 5B and 5C). In addition, mRNA levels of LDLR, SRB1, and LRP1 6 h after pMCAO were significantly higher in the ischemic cortex than in the contralateral non-ischemic cortex (Figures 5A–5C). Next, we immunohistochemically examined the expression patterns of LDLR, SRB1, and LRP1 in the ischemic and non-ischemic cortexes 6 h after pMCAO, which was when the corresponding mRNAs were upregulated. We detected strong LDLR and SRB1 signals in microvascular cells in the ischemic cortex, but not in the non-ischemic cortex (Figures 5D and 5E). LRP1 signals detected in the ischemic cortex were strong and were colocalized with those from the neuronal cell marker NeuN, whereas those detected in the contralateral non-ischemic cortex were weak (Figure 5F). These results indicated that, soon after its onset, brain ischemia enhanced LDLR and SRB1 expression primarily in brain ECs and LRP1 expression primarily in neurons.

Figure 5.

Upregulation of lipoprotein receptors during the hyperacute phase after ischemic stroke

Measurement of LDLR (A), SRB1 (B), and LRP1 (C) mRNA levels by qRT-PCR analyses in the ischemic cortex and contralateral cortex in sham-operated mice and 3 h, 6 h, and 3 days after pMCAO (n = 3, mean values ±SEM. ∗∗p < 0.01, ∗∗∗p < 0.001 versus sham controls in the ischemic cortex; ‡p < 0.05 versus sham controls in the contralateral non-ischemic cortex; two-way ANOVA followed by Dunnett’s post hoc test; ††p < 0.01, †††p < 0.001 compared between the ischemic cortex and its contralateral non-ischemic cortex; two-way ANOVA, followed by Bonferroni’s post hoc test). Confocal immunofluorescence double-labeling images with antibodies against LDLR (red, D), SRB1 (red, E), and the endothelial-specific marker CD31 (green) in the ischemic cortex and contralateral non-ischemic cortex 6 h after induction of pMCAO in mice. (F) Confocal immunofluorescence double-labeling images with antibodies against LRP1 (red) and the neuronal cell marker NeuN (green) in the ischemic cortex and contralateral non-ischemic cortex 6 h after induction of pMCAO in mice. Scale bars in (D)–(F), 50 μm. qRT-PCR analyses of Malat1 RNA levels in the ischemic cortex (G) and in the contralateral non-ischemic cortex (H) in WT mice or LDLR KO mice 3 days after i.v. administration. PBS (vehicle controls) or 50 mg/kg of Toc-HDO targeting Malat1 was injected into WT or LDLR KO mice via the tail vein 3 h after pMCAO (n = 3, mean values ±SEM. ∗∗p < 0.01 versus PBS controls; †p < 0.05 compared between WT mice and LDLR KO mice; two-way ANOVA, followed by Bonferroni’s post hoc test).

To further elucidate the role of LDLR as a critical transporter of Toc-HDO in the hyperacute phase of ischemic stroke, we evaluated the gene-silencing effect of Toc-HDO in LDLR KO mice. In the ischemic cortex, Malat1 RNA levels were significantly lower (by 73.7%) in the Toc-HDO group than in PBS controls in wild-type (WT) mice, but they were only 36.8% lower in the LDLR KO mice; thus, the silencing of Malat1 in the Toc-HDO group was much stronger in WT mice than in LDLR KO mice (Figure 5G). In contrast, in the contralateral non-ischemic cortex, there were no significant differences among treatment groups (Figure 5H). These data suggest that LDLR may serve as a transporter in the ischemic cortex in hyperacute ischemic stroke.

Silencing Malat1 increases infarct lesions, exacerbates neurological deficits, and reduces cerebral blood flow

To examine the effect of silencing Malat1 on ischemic stroke, we evaluated cerebral blood flow (CBF; measured as a percentage of the pre-surgery value) and motor function during the 4-day period after pMCAO and quantified the infarction volume 4 days after pMCAO. Cresyl violet staining (Figures 6A–6C) showed that, 4 days after ischemic stroke, the total infarction volume was significantly greater than that in PBS controls (21.2% ± 2.4%) in the Toc-HDO group (33.8% ± 3.6%) but not in the shuffle Toc-HDO group (20.6% ± 1.6%) (Figure 6D), whereas CBF was significantly lower than in the controls in the Toc-HDO group but not the shuffle Toc-HDO group (Figure 6E). The observation that the shuffle Toc-HDO had no effect on ischemic damage (Figures 6D and 6E) indicated that the gene-silencing effect, rather than the toxic effect of Toc-HDO, contributed to the expansion in infarct volume. These results suggest that ischemia-inducible Malat1 RNA exerts a protective effect against ischemic damage in hyperacute ischemic stroke.

Figure 6.

Toc-HDO targeting Malat1 exacerbated ischemic damage and neurological functions after pMCAO

Representative cresyl violet staining of serial coronal sections 4 days after pMCAO in the PBS (A), shuffle Toc-HDO (B), and Toc-HDO (C) groups. The infarction area is traced by a red line. (D) Total infarction volume measured in eight coronal sections in each group 4 days after pMCAO (n = 4, mean values ±SEM. ∗∗p < 0.01 versus PBS controls, one-way ANOVA followed by Dunnett’s post hoc test). (E) CBF in the ischemic cortex in each group 4 days after pMCAO (n = 4, mean values ±SEM. ∗p < 0.05 versus PBS controls, one-way ANOVA followed by Dunnett’s post hoc test). (F) As assessed by the EBST, animals displayed more frequent turns toward the contralateral side (right) after pMCAO. Animals in the Toc-HDO group showed significantly less recovery of right-biased body swing rate than did the PBS controls (∗p < 0.05 versus PBS controls; repeated-measures ANOVA, followed by Tukey’s post hoc test).

We then determined whether silencing Malat1 affected neurological deficits after ischemic stroke. Following pMCAO surgery, mice showed a strong tendency to turn toward the contralateral side.²² Consistent with previous studies,^22,23 the right-biased body swing rate was significantly higher in the Toc-HDO group than in PBS controls (Figure 6F); in contrast, the shuffle Toc-HDO group showed no significant difference in this behavior relative to PBS controls.

Silencing Malat1 inhibits angiogenesis and promotes ischemic neuronal death after ischemic stroke

Because Malat1 RNAs were identified primarily in neurons and ECs in the ischemic cortex, we investigated endothelial and neuronal cell function 4 days after i.v. administration of Toc-HDO targeting Malat1 (50 mg/kg). First, we evaluated stroke angiogenesis by quantifying the areas of microvessels and proliferative ECs in the ischemic cortex by using the endothelial-specific marker CD31 and the cell mitosis marker Ki-67, respectively. In the ischemic cortex, the average area of CD31-positive microvessels per section in the Toc-HDO group (13,985 ± 1,386 μm²) was significantly lower than that in PBS controls (31,498 ± 2,618μm²) (Figures 7A and 7D); the values in the shuffle Toc-HDO group (33,109 ± 2,079 μm²) were similar to those in the PBS controls. Immunofluorescence staining revealed the presence of proliferating ECs in the ischemic cortex (data not shown). The average number of CD31+/Ki-67+ cells in the ischemic cortex was significantly lower in the Toc-HDO group (14.2 ± 1.2 cells per section) than in the PBS controls (22.4 ± 2.1 cells per section) (Figure 7C); there was no significant difference in numbers between the shuffle Toc-HDO (25.9 ± 7.7 cells per section) and PBS controls. Taken together, these results indicated that Malat1 RNA plays an important role in regulating the angiogenesis of microvessels.

Figure 7.

Modulation of proliferating ECs and degenerating neurons in ischemic cortex under silencing of Malat1

Representative images show microvessel staining with anti-CD31 (A) and FluoroJade-C (FJ-C; green) (B) in the ischemic cortex 4 days after i.v. administration of PBS, shuffle Toc-HDO, or Toc-HDO. Scale bars, 50 μm. Bar graphs show average number of Ki-67+/CD31+ cells per section (C, ∗p < 0.05 versus PBS controls), average area of microvessels per section (D, ∗∗∗p < 0.0001 versus PBS controls), and average number of FJ-C-positive cells per section (E, ∗∗∗p < 0.0001 versus PBS controls) in the ischemic cortex in each group (n = 4, mean values ±SEM; one-way ANOVA followed by Dunnett’s post hoc test).

To further explore the effect of silencing Malat1 RNA function on neurons, we performed Fluoro-Jade C (FJ-C) staining and quantified neuronal degeneration in the Toc-HDO, shuffle Toc-HDO, and PBS control groups (Figure 7B). In the ischemic cortex, the average number of FJ-C-positive cells was significantly higher in the Toc-HDO group (557 ± 32.7 cells per section) than in the PBS controls (364 ± 34.4 cells per section) (Figure 7E). In contrast, the number in the shuffle Toc-HDO group (328 ± 13.5 cells per section) did not differ significantly from that in the PBS controls, indicating that Malat1 RNA also plays a major role in regulating neuronal degeneration.

Altered mRNA profiles in ischemic cortex after i.v. administration of Toc-HDO targeting

Malat1

To explore the mechanism underlying the increased infarction volume after Malat1 silencing in the Toc-HDO group, we performed mRNA microarray analyses. The results revealed 348 mRNAs of which the expression changed significantly in the PBS group 3 days after pMCAO relative to that in the sham-operated controls: 340 were upregulated (fold change ≥2) and eight downregulated (fold change ≤0.5). Moreover, the expression of 623 mRNAs was changed significantly in the Toc-HDO group relative to that in PBS controls 3 days after pMCAO: 170 were upregulated (fold change ≥2) and 453 were downregulated (fold change ≤0.5) (data not shown). Among the significantly upregulated mRNAs, 82 were upregulated in the PBS group relative to sham-operated controls; these were further upregulated in the Toc-HDO group relative to PBS controls after pMCAO (Figure 8A). Hierarchical clustering analysis yielded a detailed heatmap of the 82 genes (Figure 8B). In contrast, among eight mRNAs significantly downregulated in the PBS group relative to sham-operated controls, none was further downregulated in the Toc-HDO group relative to PBS controls after pMCAO.

Figure 8.

Microarray analysis expression profiles of mRNAs in mice from the sham group and from the ischemic groups given i.v. PBS or Toc-HDO targeting Malat1

(A) Venn diagram of significantly upregulated mRNAs (fold change ≥2) in the sham-operated control group and in the ischemic groups given i.v. PBS or Toc-HDO. (B) Heatmap obtained by hierarchical clustering analysis shows differential expression of mRNAs; n = 3 per group. Blue and red represent low and high expression levels, respectively. Each column represents a single group, and each row represents a single mRNA.

We further investigated the microarray data by using the GeneCodis, a Web-based tool, to reveal Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and Gene Ontology (GO) term analysis to describe the biological processes. A GO term analysis for the 170 mRNAs significantly upregulated in the Toc-HDO group relative to the PBS controls suggested that silencing of Malat1 could aggravate the expression of pro-inflammatory (e.g., Stab1, Ncf1, Ly86, Ccl6, Hmgb2), pro-apoptotic (e.g., Ncf1, Btk, Pycard, Naip2, Prkcd), and anti-angiogenesis (e.g., Stab1, Cd37) factors (Table S3). In contrast, a GO term analysis of the 453 mRNAs significantly downregulated in the Toc-HDO group relative to the PBS controls revealed decreased expression of neuroprotection genes such as Gfra2 and Mef2c (Table S4). The KEGG analyses for the 82 upregulated mRNAs also indicated that these genes were mainly enriched in inflammation-associated pathways, such as the Fc epsilon RI signaling pathway, chemokine signaling pathway, and leukocyte transendothelial migration (Table S5). Taken together, our mRNA microarray results suggest that the increase in infarct volume after silencing of Malat1 by Toc-HDO may be related to inflammation, apoptotic processes, or angiogenesis via altered regulation of mRNAs.

Discussion

Here, we demonstrated that Toc-HDO is much more efficiently and selectively delivered to the ischemic region than to the non-ischemic brain. In acute MCAO, pial collaterals that connect the MCA branches with the other two major cortical arteries (i.e., the anterior cerebral and posterior cerebral arteries) act as alternative sources of blood supply into the occluded MCA territory.^24,25 In a hyperacute ischemic stroke, collateral-based retrograde blood flow occurs to preserve the blood supply to the ischemic lesion as a physiological response.²⁶ However, this response is not enough to supply sufficient oxygen and glucose to ischemic brain regions.²⁷ It has been reported that, following pMCAO, CBF decreases by about 90% relative to normal in ischemic core regions.²⁷ In patients within 4.5 h after the onset of hyperacute ischemic stroke, computed tomography (CT) perfusion analysis has revealed that CBF in the ischemic core decreases markedly, by more than 95.5%.²⁸ Given that CBF is a major factor in pharmacokinetics and pharmacodynamics, the cerebral uptake of anesthetic drugs or brain-imaging radiotracers is very limited in the ischemic brain.^{29,30,31,32,33} Here we found, however, that the uptake of Toc-HDO in the ischemic cortex paradoxically increased relative to that in the contralateral non-ischemic cortex, to over 9.6- and 4.3-fold at 3 and 6 h, respectively, after pMCAO surgery. A similar phenomenon has been reported in which cerebral uptake of a drug carrier such as lipid particles has paradoxically increased; specifically, nano-sized liposomes composed of distearoylphosphatidylcholine and cholesterol accumulated in the ischemic region when they were administered i.v. during the hyperacute phase of stroke in a pMCAO model.^34,35

We found here that i.v. administered ASO was poorly detected in brain parenchymal cells in the ischemic cortex, whereas Toc-HDO was robustly observed in neurons as well as brain ECs in the hyperacute ischemic cortex, even though the molecular weight of Toc-HDO is larger than that of ASO. This suggests that the drug delivery system that delivers Toc-HDO into the brain depends largely on lipoprotein-receptor-mediated transcellular pathways—most likely transcytosis on brain ECs and endocytosis on neurons—rather than on paracellular pathways. In the process of cellular uptake, phosphorothioate-modified ASO, which is included in our lipid-HDO, can bind to plasma albumin because of its backbone; in contrast, Toc-HDO is bound strongly to serum lipoproteins high-density lipoprotein (HDL) and low-density lipoprotein (LDL).¹³ In this study, we demonstrated that expression of the lipoprotein receptors LDLR, LRP1, and SRB1 could be significantly increased on both brain ECs at the BBB and neurons in the hyperacute phase of ischemic stroke, as we previously reported.⁸ Importantly, the gene-silencing effect of Toc-HDO in LDLR KO mice was significantly reduced relative to that in WT mice. In contrast, a reduction in gene silencing by LDLR KO was not clear in intact mice brains. These findings indicate that upregulated LDLR participates in the delivery of Toc-HDO, at least in part, via receptor-mediated transcytosis for crossing the BBB and via receptor-mediated endocytosis for uptake by neurons (Figure S6). Although Cho-HDO is also assumed to be involved in lipoprotein-receptor-mediated transcellular pathways, having shown significant gene-silencing efficacy in the ischemic cortex, delivery of Cho-HDO into the brain parenchyma may involve other mechanisms. We recently found that Cho-HDO targeting Malat1 had greater gene-silencing efficacy than Toc-HDO in the brain cortexes of WT mice,³⁶ where lipoprotein receptors are not robustly expressed. Non-lipoprotein-receptor-mediated pathways of Cho-HDO entry into the brain parenchyma may have led to the significant gene-silencing effect of this oligonucleotide in the non-ischemic cortex of pMCAO mice, although the mechanisms of Cho-HDO delivery into the brain parenchyma across the BBB have not yet been fully elucidated. Finding a suitable and specific ligand for ischemic stroke regions would be ideal to expand our therapeutic options.

Oligonucleotide therapy has already been applied in clinical settings to central nervous system diseases such as spinal muscular atrophy, but this procedure requires intrathecal administration.³⁷ For clinical application of oligonucleotide therapy in ischemic stroke, the intrathecal approach is a contraindication for patients treated with tissue plasminogen activator or an antithrombotic in the hyperacute phase owing to the potential bleeding side effects. Our technology, which enables genetic regulation of the ischemic brain by i.v. administration of Toc-HDO, has great advantages for the treatment of acute ischemic stroke. Within 24 h after onset, acute ischemic stroke results in cellular bioenergetic failure, followed by multifactorial pathogenic processes, including excitotoxicity, oxidative stress, inflammation, BBB disfunction, microvascular injury, and finally the death of neurons and ECs.³⁸ The use of many potential target RNAs is limited, because the relatively long half-lives of many proteins means that 24 h is often too short to achieve downregulation of an encoded protein. In contrast, non-coding functional RNAs are better targets, as they are cleaved directly by HDO in the cytosol and nucleus. Expression changes in hundreds of lncRNAs have been shown to arise in brain tissue soon (within 12 h) after acute ischemic stroke, and many lncRNAs have been demonstrated to be potential target RNAs for the regulation of pathophysiological processes after stroke.^{17,39,40,41,42,43,44,45,46}

In conclusion, we demonstrated here that the efficacy of an i.v. lipid-HDO in silencing Malat1 in a dose-dependent manner in the ischemic cortex far exceeded that of the parent ASO or HDO without the lipid ligand. Furthermore, we showed that a lipid-HDO, specifically Toc-HDO, could be selectively delivered to the ischemic cortex—mainly into brain ECs and neurons—even in the hyperacute phase of ischemic stroke, in part via the transcellular pathways of ECs, such as lipoprotein receptor-mediated transcytosis. By downregulating Malat1 by using Toc-HDO technology, we were able to alter the stroke phenotype through the inhibition of angiogenesis and inflammation, consistent with previous studies in Malat1 KO mice.¹⁷ These data indicate that our lipid-HDO technology is useful for gene regulation in the ischemic brain in the hyperacute stage.

Materials and methods

Experimental mice

Male C57Bl/6NCrSlc mice (Japan SLC) aged 8 to 12 weeks (n = 146) were used. Mice were kept on a 12-h light/dark cycle in a pathogen-free animal facility with free access to food and water. The study was performed in strict accordance with the NIH guidelines for the care and use of animals in research. All animal experiments were approved by the Institutional Animal Care and Use Committee of Tokyo Medical and Dental University (permit number: 0170179A).

Mouse model of pMCAO surgery

An investigator blinded with respect to the treatment group performed all assessments. Mice were anesthetized, their body temperature was monitored, and the left middle cerebral artery (MCA) was visualized and occluded as described previously.⁸ After replacement of the temporal muscle, the skin was sutured and the mice were kept in a recovery cage until they had recovered from anesthesia. Sham surgery was performed by using the same process but without occlusion.

Measurement by laser Doppler flowmetry

A laser Doppler flow meter (TBF-LN1; Unique Medical Company, Tokyo, Japan) and probe (Type-CS; Advance, Tokyo, Japan) were used to monitor relative changes in CBF. CBF was recorded over the left lateral skull by placing the laser Doppler probe at a position 2.0 mm posterior and 2.5 mm lateral to bregma. Mice were placed on a heating pad to maintain a constant body temperature of 36.5°C ± 0.5°C for the entire procedure. Baseline CBF values were recorded just before surgery and 4 days after surgery. CBF values were expressed as percentages of the pre-surgery value.

Oligonucleotide drugs

PBS (Sigma-Aldrich) control solution, ASO, Toc-ASO, HDO, Toc-HDO, Cho-HDO, shuffle Toc-HDO, or Alexa Fluor-labeled ASO, and Toc-HDO (50 mg/kg, in PBS) targeting Malat1 was injected into the tail vein immediately (0 h) or 3 h after pMCAO and allowed to circulate for 3 h, 3 days, or 4 days before the animal was sacrificed. A series of gapmers were synthesized by Gene Design (Osaka, Japan). The sequences of ASO and cRNA, Toc-RNA, or Cho-RNA targeting Malat1 RNA are shown in Table S1, as follows: 16-mer gapmer ASO, 5′-C∗T∗A∗g∗t∗t∗c∗a∗c∗t∗g∗a∗a∗T∗G∗C-3′, where lowercase letters represent DNA, uppercase letters represent LNA (capital C denotes LNA methylcytosine), and asterisks represent phosphorothioate linkages; and 16-mer cRNA (or Toc-RNA, or Cho-RNA), 5′-g∗c∗a∗UUCAGUGAAC∗u∗a∗g-3′, where uppercase letters represent RNA, lowercase letters represent 2′-O-methyl sugar modification, and asterisks represent phosphorothioate linkages. The shuffle sequences of the ASO and cRNA targeting Malat1 RNA were as follows: 16-mer gapmer Malat1 ASO, 5′-T∗A∗C∗a∗t∗a∗t∗g∗c∗g∗c∗t∗a∗C∗T∗G-3′, where lowercase letters represent DNA, uppercase letters represent LNA (capital C denotes LNA methylcytosine), and asterisks represent phosphorothioate linkages; and 16-mer Malat1 cRNA, 5′-c∗a∗g∗UAGCGCAUAU∗g∗u∗a-3′, where uppercase letters represent RNA, lowercase letters represent 2′-O-methyl sugar modification, and asterisks represent phosphorothioate linkages. Alexa Fluor 568/647 fluorophores were covalently bound to the 5′ ends of DNA/LNA gapmers, and the lipid ligand, α-tocopherol, or cholesterol was bound directly to the 5′ ends of cRNA by a phosphodiester bond, with on-column oligonucleotide synthesis. For generation of HDO, Toc-HDO, or Cho-HDO, equimolar amounts of ASO and cRNA, Toc-cRNA, or Cho-cRNA strands were heated in PBS at 95°C for 5 min and slowly cooled to room temperature (RT).

Cell culture and OGD procedures

Mouse neurons (Neuro2A; American Tissue Culture Collection [ATCC] catalog no. CCL-131) and mouse ECs (bEnd3; ATCC catalog no. CRL-2299) were maintained at 37°C in a 5% CO₂, 95% air atmosphere in DMEM (Wako, Japan) containing glucose (Invitrogen, Carlsbad, CA), supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% antibiotics (100 μg/mL streptomycin and 100 U/mL penicillin); the medium was changed every 2 days. To mimic ischemic conditions in vitro, Neuro2A cells or bEnd3 cells were exposed to oxygen and glucose deprivation under the same conditions as described previously.⁴⁷ Briefly, cultured Neuro2A or bEnd3 cells were plated on culture dishes at a density of 3 × 10⁴ cells/cm² and incubated for 2 or 4 h at 37°C with glucose-free DMEM in a closed chamber (Billups-Rothenberg, Del Mar, CA) filled with 5% CO₂, 95% N₂, after which they were collected for qRT-PCR assay.

Preparation of microvascular fraction of the brain

The microvascular fraction (i.e., fractionated brain ECs) of the brain was prepared as described previously.^14,48 Briefly, both the ischemic and contralateral brains of mice were homogenized in PBS and centrifuged at 800 × g for 5 min at 4°C. The pellet was then suspended in a 15% dextran solution and centrifuged at 4500 × g for 10 min at 4°C. Subsequently, the pellet was resuspended in 5 mmol/L PBS and, after incubation for 10 min, was centrifuged for 5 min at 800 × g. Small RNAs containing fewer than 200 nucleotides were extracted from the above final pellet of small vessels by using an mirVana miRNA isolation kit (Thermo Fisher Scientific, Tokyo, Japan).

qRT-PCR assay

Total RNA was extracted from mouse brains, fractionated brain ECs, Neuro2A cells, and bEnd3 cells by using Isogen (Nippon Gene, Tokyo, Japan) in accordance with the manufacturer’s protocol. For brains, the ipsilateral ischemic cortex and contralateral non-ischemic cortex were separately dissected, immediately frozen in liquid nitrogen, homogenized in Isogen buffer, and subjected to RNA extraction. For cells, Isogen buffer was added directly into the culture plates after the OGD-treated cells or fractionated brain ECs, and the cells were lysed mechanically and subjected to RNA extraction. Reverse transcription of 100 ng of RNA to cDNA was performed with Transcriptor Universal cDNA Master (Roche Diagnostics, Mannheim, Germany) by using the following conditions: 25°C for 5 min, 55°C for 10 min, and 85°C for 5 min. Quantitative real-time PCR of the resultant DNA was performed by using TaqMan Probe Master and a Light Cycler 480 Real-Time PCR Instrument (Roche Diagnostics) with the primers for mouse Malat1 (NR_002847), LDLR (Mm00440169_m1), SRB1 (Mm00450234_m1), LRP1 (Mm00464608_m1), and glyceraldehyde-3-phosphate dehydrogenase (Gapdh; 4352932E) supplied in the TaqMan Gene Expression Assays (Applied Biosystems, Foster City, CA). Relative Malat1 RNA expression, as well as LDLR, SRB1, and LRP1 mRNA levels, was calculated in comparison with Gapdh mRNA levels as an internal control.

Quantification of signal intensity of Alexa Fluor-labeled Toc-HDO or ASO in the brain

Three hours after i.v. administration of Alexa Fluor 647-labeled Toc-HDO or ASO, the ischemic cortex and contralateral non-ischemic cortex were separately collected; each was then immediately frozen in liquid nitrogen and mechanically homogenized in PBS. To determine the standard curve, six standard solutions at a range were prepared. The signal intensity of Alexa Fluor 647 in the brain cortex was measured by using i-control software (infinite M1000 PRO, Tecan Group, Männedorf, Switzerland), after which the concentration of DNA/LNA gapmer was calculated on the basis of the standard curve. The signal intensity was determined to be linear in the range of 0.156 to 0.156 × 1/2⁵ μmol/L, as determined with r > 0.99.

HELISA

Single-stranded ASO, ASO derived from Toc-HDO, and Toc-cRNA derived from Toc-HDO in tissue samples were analyzed by using the hybridization-based ELISA assay HELISA, as previously reported by Wei et al.,⁵⁰ with minor modifications. The sequences of the capture probe against ASO and cRNA were as follows: 3ʹ-biotinylated 25-mer DNA oligonucleotides with 5ʹ-end overhang, 5ʹ-taactagtggcattcagtgaactag-3ʹ and 5ʹ-taactagtgctagttcactgaatgc-3ʹ, respectively. The detection probe was designed as 9-mer DNA oligonucleotides complementary to the 5ʹ-end overhang of the capture probe with 5ʹ-end phosphorylated and 3ʹ-end digoxigenin labeled, 5ʹ-cactagtta-3ʹ. The samples were homogenized with DNA/RNA shield (Zymo Research, CA, US). The homogenates were mixed with an equal volume of the hybridization buffer containing 60 mmol/L sodium phosphate, pH 7.4, 1.0 mol/L NaCl, 5 mmol/L EDTA, 0.2% Tween 20, and 200 nmol/L capture probe and incubated at 95°C for 5 min and then at 42°C for 90 min for hybridization. After incubation, 150 μL of the aliquots was transferred to a NeutrAvidin-coated 96-well plate (Thermo Fisher Scientific) and incubated at 37°C for 30 min to allow the attachment of biotin-labeled capture probe to the wells. After washing of the plate, 150 μL of ligation solution (New England Biolabs Japan, Tokyo, Japan) containing 250 U/mL T4 DNA ligase and 100 nmol/L detection probe was added to each well and incubated at 18°C overnight for ligation. After washing and addition of 60 U/150 μL S1 nuclease solution in 100 mmol/L NaCl, the plate was incubated at 37°C for 2 h to cleave the truncated duplex. The digoxigenin-labeled probe was detected and quantified by using alkaline-phosphatase-conjugated anti-digoxigenin Fab fragments (1:5,000, Sigma-Aldrich Japan, Tokyo, Japan) and an AttoPhos AP Fluorescent Substrate System (Promega, Tokyo, Japan).

ViewRNA ISH assay

Malat1 RNA expression was detected by using a QuantiGene ViewRNA tissue assay (catalog no. QVT0011; Affymetrix, Tokyo, Japan) according to the manufacturer’s instructions. The Malat1 probe for mouse was purchased from Affymetrix (catalog no. VB-11110-01/mouse). In brief, mouse brains were fixed in 10% neutral-buffered formalin, embedded in Optimal cutting temperature (OCT) block, and sectioned into 12-μm–thick sections. The tissue on the slides was soaked in 10% neutral-buffered formalin overnight, then dehydrated and treated with protease at 40°C for 20 min. The Malat1 probe was used at 1:40 dilution and was incubated with samples at 38°C for 3 h. After being washed, the Malat1 RNA/probe complex was hybridized with preamplifier, amplifier, and alkaline-phosphatase-conjugated oligonucleotides at 40°C for 25, 15, and 15 min, respectively. The dilution ratios and the preamplifier and amplifier were as recommended by Affymetrix. After the removal of free alkaline phosphatase oligonucleotides by washing in PBS, the slide was incubated with fast red substrate at RT for 30 min and counterstained with hematoxylin. The tissue images were acquired under an optical microscope.

Tissue preparation and measurement of total infarct volume

Four days after pMCAO or sham surgery, the animals’ tissues were fixed by transcardiac perfusion with 4% paraformaldehyde under deep anesthesia. The brains were removed, incubated in 20% sucrose, and then frozen rapidly on dry ice. The brains were then cut into 20-μm serial coronal sections from the level of the anterior pole of the caudate nucleus through to the cerebral hemisphere, mounted on slides, and processed for staining. To evaluate ischemic lesions, eight serial sections, spaced 400 μm apart from bregma level +1.2 mm to −2.0 mm, were stained with cresyl violet. The areas of the infarct, ipsilateral hemisphere, and contralateral hemisphere in the eight coronal sections were measured on an image of each section by using NIH ImageJ 1.47v software (https://imagej.nih.gov/ij/index.html). The total infarct volume of the ipsilateral hemisphere (percentage infarction volume) was calculated as a percentage of the volume of the contralateral hemisphere.

Immunofluorescence staining

For double immunohistochemical staining, the sections were washed with PBS and blocked with 10% normal goat serum in PBS, and then incubated overnight at 4°C with the two primary antibodies diluted with 10% normal goat serum. The primary antibodies used were as follows: rabbit anti-LDLR antibody (1:100, Abcam), rabbit anti-SRB1 antibody (1:100, Abcam), rabbit anti-LRP1 antibody (1:200, Abcam), rabbit anti-rat Ki-67 monoclonal Ab (1:100; Acris) for detection of mitotic cells, rat anti-mouse CD31 antibody (1:50, Santa Cruz Biotechnology, Santa Cruz, CA) for detection of ECs, and mouse anti-NeuN antibody (1:100, EMD Millipore, Darmstadt, Germany) for detection of neurons. Ki-67 immunostaining required antigen retrieval by heating in 10 mmol/L citrate buffer (pH 6) in a microwave oven (5 × 2 min, 750 W). Subsequently, sections were incubated with fluorescence-labeled secondary antibodies for 1 h at RT. After being washed in PBS, fluorescently stained sections were mounted with Vectorshield (Vector Laboratories, Burlingame, CA) and observed under a confocal laser-scanning microscope (LSM510; Zeiss, Oberkochen, Germany).

Immunohistochemistry of anti-PS Ab

At 3, 9, 24, or 72 h after i.v. administration of Toc-HDO, the animals’ tissues were fixed by transcardiac perfusion with 4% paraformaldehyde under deep anesthesia. The brains were removed, fixed in 10% formalin neutral buffer solution overnight, and then embedded in paraffin by using a HistoStar paraffin embedding station (Thermo Fisher Scientific, Tokyo, Japan). The brains were then cut into 5-μm serial coronal sections from the level of the anterior pole of the caudate nucleus through to the cerebral hemisphere, mounted on slides, and processed for staining. The sections were pretreated with antigen retrieval Proteinase K (Dako, Santa Clara, CA) for 5 min and incubated in endogenous peroxidase and alkaline phosphatase blocking solution (Vector Laboratories, Richmond, CA) for 10 min at 25°C. Sections were blocked for 30 min by using Background Buster (Innovex Biosciences, Richmond, CA). A polyclonal rabbit anti-ASO Ab⁴⁹ that recognizes the phosphorothioate-modified backbone containing ASO was then applied at a dilution of 1:4,000 (diluted in 10% NGS) at 4°C for two nights. Subsequently, sections were incubated with secondary goat anti-rabbit IgG Ab, biotinylated (Vector Laboratories, Richmond, CA) at 1:1,000 dilution for 2 h and then incubated by using an ABC kit (Vector Laboratories, Richmond, CA) for 1 h at RT. After washes in PBS and 0.05 M Tris buffer, the sections were mounted within 1 h after being subjected to 3,3″-diaminobenzidine (DAB)-nickel-H₂O₂ reaction. The tissue images were acquired under an optical microscope.

FJ-C staining

All generating neurons were stained with FJ- C (AG325-30MG, EMD Millipore). Briefly, sections were immersed in a basic alcohol solution consisting of 1% sodium hydroxide in 80% ethanol for 5 min, 70% ethanol for 2 min, distilled water for 2 min, and 0.06% potassium permanganate solution for 10 min, and then in a solution containing 0.01% FJ-C and 0.1% acetic acid (1:100) for 10 min. Incorporation of DAPI (Sigma-Aldrich, St. Louis, MO) as a fluorescent nuclear stain was accomplished by simply adding 0.0001% DAPI to the FJ-C staining solution. After three 1-min washes, the slides were air dried on a slide warmer at 50°C for at least 5 min and then cleared in xylene for at least 1 min; coverslips were applied with DPX nonfluorescent mounting medium (Sigma-Aldrich). Slides were imaged with a confocal laser-scanning microscope as mentioned above. The number of FJ-C+ cells in each group was counted.

Cell quantification

To estimate the total numbers of FJ-C+ cells, Ki67+CD31+ cells, and CD31+ vessel areas in the ischemic cortex, eight serial sections, spaced 200 μm apart from bregma level +1.2 mm to −2.0 mm, were collected from each mouse. Counts of FJ-C+ cells and Ki-67+CD31+ cells in the ischemic cortex are given as cells per section. The CD31+ vessel area was measured as an average per section in the ischemic areas and was determined by using a confocal laser-scanning microscope with a 40× objective and NIH ImageJ 1.47v software.

mRNA microarray experiments

Samples were harvested from normal brain cortex or ischemic cortex 3 days after injection of PBS or Toc-HDO at a dose of 50 mg/kg after pMCAO. Total RNA was extracted from the ipsilateral ischemic cortex by using Isogen (Nippon Gene) according to the manufacturer’s protocol. mRNA microarray analyses (1 μg of total RNA) were performed by using mouse Oligo chips 24k (Toray Industries, Tokyo, Japan) and were analyzed by using 3D-Gene Extraction Software (Toray Industries) and the GeneCodis 4.0 Web tool (https://genecodis.genyo.es/).

Evaluation of neurological function

To assess the effect of Toc-HDO targeting Malat1 on neurological function after stroke, PBS, Toc-HDO targeting Malat1, or shuffle Toc-HDO was administered 3 h after pMCAO, and performance on the elevated body swing test (EBST) was examined 1 day before and 4 days after surgery. The EBST was used to evaluate asymmetric motor behavior in our previous study of the same pMCAO mouse model.²² Briefly, the animals were held by the base of the tail and elevated approximately 10 cm from the tabletop. They were then held in the vertical axis, defined as no deviation of more than 10° to either side. The direction of body swing was recorded for 1 min during each of three trials per day. The numbers of left and right turns were counted, and the percentage of turns made to the side contralateral to the lesioned hemisphere (percentage of right-biased body swing) was determined. All tests were conducted blind to the groups.

Evaluation of possible adverse effects

To evaluate the safety of the drug, 3 h after pMCAO surgery, we gave each mouse an i.v. injection (50 mg/kg) of PBS (control) or Toc-HDO targeting Malat1 (n = 3 for each group) and then analyzed blood chemistry and liver and the kidney histology after 3 days. Blood chemistry was assessed at the SRL Laboratory (Tokyo, Japan). For pathological analyses, liver and kidney tissues were fixed in 10% neutral-buffered formalin solution for 24 h, embedded in paraffin, and then cut into 4-μm sections by using an ROM-380 microtome (Yamato Kohki Industrial, Tokyo, Japan). Slides were stained with hematoxylin and eosin (Muto Pure Chemicals, Tokyo, Japan).

Statistical analysis

Statistical analyses were performed by using an unpaired t test, one-way ANOVA, or two-way ANOVA followed by Tukey’s, Dunnett’s, or Bonferroni’s post hoc tests. Analyses were done by using GraphPad Prism software (version 8, San Diego, CA). Data are presented as means ± SEM unless specifically indicated in the text. p values less than 0.05 were considered to be statistically significant.

Data availability

The data that support the findings of this study are available from the corresponding author on reasonable request.