Male Sprague–Dawley (SD) rats were sourced from Beijing Viton Lever (SCXK jing 2021–0006) with an average weight of 310 ± 20 g. The rats had unrestricted water and food access. Control appropriate room temperature and relative humidity. Light and darkness were alternated every 12 h. The experimental procedures received approval from the Institutional Animal Care and Use Committee of Nanjing University of Chinese Medicine (202306A074). At the end of the experiment, each rat was humanely euthanized via intraperitoneal injection with a 200 mg/kg dose of sodium pentobarbital solution.

Embolic Stroke Modeling Methods were described by Zhang et al. (18). Rats were anesthetized with isoflurane (5% for induction and 2% for maintenance). A PE-50 tube is inserted into the donor rat’s femoral artery to procure arterial blood, which is subsequently maintained in a petri dish at 37°C for 2 h and at 4°C for 22 h. Transfer the clot to the PE-10 tube and aspirate and rinse the clot 10–15 times to eliminate red blood cells. The cleaned clot is gathered in a revised PE-50 catheter. The cervical vessels of rats were exposed after anesthesia. A modified PE-50 catheter that contained an embolus was inserted 19–22 mm from the external carotid artery to the internal carotid artery. Upon the catheter tip arriving at the source of the middle cerebral artery (MCA), it is retracted by 1–2 mm, and the thrombus is gradually injected into the MCA using 5–10 μL of saline. Extract the catheter 5 min post-injection. The laser speckle imaging system (RWD, China) was employed to assess brain blood flow, with successful occlusion indicated by a marked decrease in perfusion (Supplementary Figures 1A,B).

The overall experiment consisted of the first and second experiments, with the rats randomized into multiple groups in each experiment.

The rats in the A + 6 h rt-PA and A + 6 h rt-PA + Mdivi-1 groups were treated with acupuncture at 6 h post-stroke. The rats in the A + 4.5 h rt-PA group were treated with acupuncture at 4.5 h post-stroke. Acupuncture points were selected as Shuigou (GV26, situated at the intersection of the superior and middle thirds of the human median sulcus) and bilateral Neiguan (PC6, positioned on the anterior forearm, 2 cm above the transverse line of the distal palmar wrist, between the palmar longus tendon and the radial carpal flexor tendon). Firstly, the bilateral PC6 acupoints were pricked straightly at a depth of 3 mm, and then the lift-insertion-twisting-laxation method was performed for a duration of 1 min; then the GV26 acupoint was pricked obliquely in the direction of the nasal septum for 2–3 mm, and then the sparrow-pecking needling was applied, and stimulation lasted for 1 min. Leave the needle in place for 30 min.

Neurological deficits were evaluated using a modified 6-point scale, with assessments conducted at 2 h and 24 h following stroke induction (22, 23). The criteria are the following: 0, no notable functional impairment; 1, flexion of the opposite forelimb; 2, diminished grip of the opposite forelimb during tail traction; 3, spontaneous, non-directional motion, exhibiting circling to the opposite side solely during tail traction; 4, spontaneous circling to the opposite side; 5, mortality. Neurologic deficit scores were required to reside between 2 and 4.

24 h post-stroke, the rat brain tissues were immediately excised and subsequently fast frozen at −20°C refrigerator for 15 min. Frozen brain tissue was placed in a mold, and eight consecutive coronal brain slices, each approximately 2 mm thick, were cut 2 mm back from the frontal pole at equal intervals, followed by incubation with 2% TTC reagent (Sigma, USA) for approximately 15 min at 37°C in a dark environment. Normal regions are depicted in red, while ischemic regions are represented in white. The cerebral infarct volume percentage was ultimately computed using ImageJ software (version 1.5.4). The formula for calculating the volume of cerebral infarction is as follows: infarcted brain volume (%) = (total brain volume − stained brain volume) / total brain volume × 100%.

The rats were given 2% Evans Blue (EB, Sigma, USA) at a rat weight of 4 mL/kg via the tail vein for 2 h prior to euthanasia. The rats were perfused using saline to remove all EB from the organism. The brain was extracted from the cut skull, with the right brain soaked in formamide (1 mL/100 mg), and subsequently, the brain tissue was homogenized. Water bath at 60°C for 24 h, with subsequent centrifugation at 13000 rpm for 20 min. The absorbance was determined by measuring the optical density (OD) at a 620 nm wavelength with a spectrophotometer. The EB content was calculated using the formula: EB concentration (μg/mL) × volume of formamide (mL) / mass of brain tissue (g).

HT following delayed rt-PA thrombolysis was assessed by quantifying hemoglobin levels in the ischemic brain hemisphere by spectrophotometry (24). Anesthetized rats underwent fast perfusion with saline, followed by excision of brain tissue from the ischemic hemisphere. Subsequent to weighing the tissue, 2 mL of pre-chilled 1 × PBS was included and homogenized. Utilize a centrifuge to separate and remove the supernatant. The hemoglobin concentration was tested using the QuantiChrom™ Hemoglobin Assay Kit (Hayward, USA). The OD value was measured using a microplate reader. The hemoglobin concentration (mg/dL) was calculated using the formula: (OD of Sample − OD of Blank) / (OD of Calibrator − OD of Blank) × 100.

Total protein was obtained from the tissue of the cerebral ischemia penumbra (IP). Protein concentration was calculated using the BCA Protein Assay Kit (Beyotime, China). The supernatant was combined with 5 × SDS-PAGE protein loading buffer and 1 × PBS to prepare the sample storage solution. Equal amounts were electrophoresed on an 8–12% SDS-PAGE gel and transferred to PVDF membranes (Millipore, USA). The following particular primary antibodies were added to the membrane after it had been blocked with 5% BSA for 2 h at room temperature and then incubated for the entire night at 4°C: Pink1 (1:1000, ab186303, Abcam, UK), Parkin (1:1000, ab77924, Abcam, UK), LC3B (1:2000, ab192890, Abcam, UK), P62 (1:2000, ab109012, Abcam, UK), COX4I1 (1:1000, ab14744, Abcam, UK), NLRP3 (1:1000, ab263899, Abcam, UK), Caspase-1 (1:1000, ab286125, Abcam, UK), and ZO-1 (1:2000, 21,773-1-AP, Proteintech, China). Following three washes, the membrane was incubated for 1 h at room temperature with HRP-conjugated goat anti-rabbit (1:10000, RS0001, Immunoway, China) or anti-mouse IgG (1:10000, RS0002, Immunoway, China). The bands were ultimately identified using improved chemiluminescence. β-tubulin (1:10000, 30302ES20, Yeasen, China) and GAPDH (1:10000, 10,419-AP, Proteintech, China) were used as internal controls. ImageJ software was used to quantify the grayscale values of the target bands and the internal reference bands. The ratios of the internal reference bands to the destination bands were normalized with a sham group mean of 1 to improve the accuracy and comparability of the data.

Total mRNA was extracted from brain tissue with Trizol and subsequently reverse transcribed into cDNA following the kit instructions (Yeasen, China). SYBR Green Master Mix was used for real-time PCR. A three-step PCR amplification program was selected: 95°C for 5 min of pre-denaturation, 95°C for 10 s of denaturation, 55°C for 10 s of annealing, and 72°C for 20 s of extension, comprising a total of 40 cycles. The 2^-ΔΔCt experiment was performed to statistically ascertain the levels of mRNA expression. Additionally, β-tubulin was utilized as an internal reference. The primers utilized are displayed in Table 1.

IL-18 and IL-1β levels were measured by ELISA. Rat cerebral cortex tissue was homogenized on ice and centrifuged at 3000 rpm for 20 min at 4°C, followed by extraction of the supernatants. An enzyme calibrator (Elabscience, China) was used to assess the OD of each well at 450 nm. The level of the target molecule was measured based on the standard curve derived from the diluted standards.

The cortex IP area of rats, perfused through the heart, was sectioned into 3 small cubes measuring 1 mm × 1 mm × 1 mm and promptly preserved with 2.5% glutaraldehyde for 24 h at 4°C. The slice thickness was determined to be 60 nm. Subsequently, double staining was conducted utilizing 3% uranyl acetate and lead citrate. TEM was used to observe the architecture of the BBB as well as neuronal mitochondria.

The data were indicated as mean ± standard deviation (mean ± SD). The analysis was performed using GraphPad Prism version 8.0.2. When the data for each group followed a normal distribution, the one-way ANOVA was employed for multiple group analyses. When the data did not meet the criteria for distribution as normal, a nonparametric test was utilized. Significant differences were deemed to be p < 0.05.

We evaluated the efficacy of acupuncture on delayed rt-PA-induced complications by evaluating HT 24 h post-stroke. The hemoglobin levels in both the 4.5 h and 6 h rt-PA groups exceeded those of the model group, with a particularly notable increase in the 6 h rt-PA group beyond the time window. However, in comparison with the 6 h rt-PA group, the hemoglobin content was markedly reduced in the A + 6 h rt-PA group (Figure 1A).

In addition, we assessed the impact of acupuncture on delayed rt-PA treatment for acute cerebral ischemia by measuring the neurological impairment score at 2 and 24 h post-stroke, as well as evaluating the volume of cerebral infarct volume at 24 h after stroke. The neurologic impairment score was reduced in the A + 6 h rt-PA group relative to the model and 6 h rt-PA groups at 24 h after stroke (Figure 1B). Compared to the model group, the volume of cerebral infarction was decreased in the 4.5 h rt-PA group and increased in the 6 h rt-PA group. However, the volume of cerebral infarction in the A + 6 h rt-PA group was considerably reduced relative to the 6 h rt-PA group (Figures 1C,D).

To evaluate the permeability of the BBB, we measured the amount of EB leakage. EB leakage was significantly increased in the model group in comparison to the sham group and further elevated in the 6 h rt-PA group. However, in contrast to the model group and the 6 h rt-PA group, EB levels were markedly diminished in the A + 6 h rt-PA group (Figures 2A,B).

Furthermore, considering the significant function of tight junctions (TJs) in preserving the integrity of the BBB, we studied whether acupuncture elevated the expression of ZO-1 in rt-PA-treated thromboembolic stroke rats. Compared with the sham group, the protein and mRNA levels of ZO-1 were dramatically reduced in the model group. Furthermore, these levels were further decreased in the 6 h rt-PA group compared to the model group. However, in the A + 6 h rt-PA group, the protein and mRNA expression of ZO-1 was distinctly enhanced in contrast to the model and the 6 h rt-PA groups. In addition, the inhibitor group showed the opposite trend to acupuncture (Figures 2C–E).

BBB ultrastructural morphology of the sham group was normal, and the surface of endothelial cells was smooth and intact, continuous and regular, and the tight junction structure was dense and sturdy. In comparison with the sham group, endothelial cells in the model group were swollen, deformed and irregular, with disrupted tight junctions and loose connections. The swelling and deformation of endothelial cells in the BBB of the 6 h rt-PA group were further aggravated compared with that of the model group, and the tight junctions were broken, and the ultrastructure of the BBB was severely damaged. As compared with the 6 h rt-PA group, the ultrastructural disruption of BBB in the A + 6 h rt-PA group was attenuated, and its endothelial cell morphology was regular and basically continuous and intact, and the disruption of tight junctions was significantly reduced. However, the inhibitor effectively antagonized the effect of acupuncture (Figure 3).

Inflammasome activation is a major contributor to BBB disruption, and the release of mediators can incite a profound inflammatory response. The results indicated a substantial increase in the expressions of NLRP3 and mature Caspase-1 in the model and the 6 h rt-PA group in contrast to the sham group; however, the expression of NLRP3 and mature Caspase-1 was significantly lower in the A + 6 h rt-PA group compared to the model and 6 h rt-PA groups. Meanwhile, the expression of NLRP3 and mature Caspase-1 was again enhanced in the A + 6 h rt-PA + Mdivi-1 group than in the A + 6 h rt-PA group (Figures 4A–F). Furthermore, the results for IL-18 and IL-1β showed approximately the same trend (Figures 4G,H).

The levels of mitophagy were evaluated using western blot and real-time PCR. Compared to the sham group, the levels of Pink1, Parkin, and LC3 protein and mRNA were significantly elevated in both the model and 6 h rt-PA groups. Furthermore, these levels were further increased in the A + 6 h rt-PA group compared to the model and 6 h rt-PA groups. However, there was a decreasing trend in the levels of Pink1, Parkin, LC3 protein and mRNA in the A + 6 h rt-PA + Mdivi-1 group relative to the A + 6 h rt-PA group (Figures 5A–D, H–J). Furthermore, P62 expression was markedly diminished in both the model and 6 h rt-PA groups compared to the sham group. P62 expression was further diminished in the A + 6 h rt-PA group. However, the P62 expression in the A + 6 h rt-PA + Mdivi-1 group was opposite to that in the A + 6 h rt-PA group (Figures 5E,K).

As a mitochondrial intima protein, COX4I1 can effectively reflect mitochondrial damage after acute cerebral infarction. The expression of COX4I1 was increased in the model and 6 h rt-PA groups compared to the sham group and was higher in the 6 h rt-PA group with a significant difference in the model group. While the expression of COX4I1 decreased in the A + 6 h rt-PA group compared with the model and 6 h rt-PA groups. It was also observed that the COX4I1 expression level in the A + 6 h rt-PA + Mdivi-1 group was again enhanced compared with that in the A + 6 h rt-PA group (Figures 5F–G,L).

We used TEM to examine the morphological alterations of the mitochondria in the neurons of each group. The mitochondrial structure of the sham group was normal, with a clear double-layered membrane and mitochondrial cristae. Compared with the sham group, the mitochondrial structure of the model group was damaged, the double-layered membrane structure was unclear or even disappeared, and the mitochondrial cristae were broken. In comparison with the model group, the mitochondrial cristae of the 6 h rt-PA group were broken or even completely disappeared, the mitochondria were swollen into vacuoles and a small number of mitochondrial autophagic vesicles could be seen. However, relative to the 6 h rt-PA group, the mitochondria in the A + 6 h rt-PA group were less disrupted, the bilayer membrane structure and cristae were clear and intact, and a greater number of mitochondrial autophagosomes were seen. The mitochondrial structure of the A + 6 h rt-PA + Mdivi-1 group was still disrupted, the cristae integrity was missing and the bilayer membrane structure was blurred when compared with the A + 6 h rt-PA group (Figure 6).