Male mice (C57BL/6) aged 6~8 weeks and weighing 21~23 g (n= 10/group) were purchased from Beijing SPF Laboratory Animal Technology Co., Ltd. (Beijing, China). The mice were housed in a 12 h light/dark cycle in a temperature-controlled room (22 ± 2°C) and provided access to food and water. The animal treatment protocols were conducted according to the guidelines of the National Institutes for Animal Research and approved by the Ethics Committee for Animal Experimentation of Peking Union Medical College Hospital of the Chinese Academy of Medical Sciences (reference no. XHDW-2021-037).

A mouse model of MCAO was established as previously described (29). The MCAO mice were successfully established by inserting a 7-0 monofilament (Doccol Corporation, Sharon, MA, USA) via the right external carotid artery (ECA) across the right internal carotid artery (ICA) to the origin of the middle cerebral artery (MCA) 1 h after withdrawing the monofilament.

MCAO in mice was established using a laser speckle contrast imager (PeriCam PSI HR System, Jarfalla-Stockholm, Sweden) while supervising regional cerebral blood flow (rCBF). Animals were excluded that died or failed to show a ≤50% or ≥80% rCBF reduction for further analysis (39–41). Sham group mice underwent the same procedure but without MCAO. After the operation, the mice were housed individually and euthanized by cervical dislocation after 72 h, and brain tissue, colon tissue and colon contents were collected for further experiments. The model success rate was 87.72%.

The iPS cells were amplified and cultured to logarithmic growth stage and were passaged at a ratio of 1:4~1:6 every 4~5 days as described previously (29). In brief, cell coverage reached 60-70% after adherent, and then the cells were cultured overnight. The culture medium was changed to GDEV medium for 24h, and the supernatant of cell culture was collected when the cell coverage reached 80-90%. The collected cell supernatant was ultrafiltrated to harvest exosomes. The purified exosomes were stored in a -80°C refrigerator for subsequent experiments.

The mice were randomly divided into five groups, including the sham operation group (Sham), model group (MCAO), electroacupuncture group (EA), iPSC-EVs group (EV), and EA combined with iPSC-EVs group (EA+EV). EA and EV treatments were performed four times at 0 h, 24 h, 48 h, and 70 h after reperfusion. In the EA group, acupuncture needles (ANDE, Guizhou, China) were inserted into the mice at the “Baihui” acupoint (GV20) and left the “Zusanli” acupoint (ST36) at a depth of 2–3 mm. The stimulation was carried out for 30 min at 2 Hz (intensity, 1 mA) frequency with continuous waves using an electroacupuncture device (KWD-808 II, Great Wall Brand, Baoding, China). Additionally, in EV group, iPSC-EVs were injected into the tail vein of mice, 20 μg per injection in each mouse was performed for a total of four injections. EA combined with iPSC-EVs treatment (EA+EV) utilized the same EA and EV treatment methods described above.

Neurological function and deficits were evaluated 72 h after reperfusion by the Clark score, which is closely related to cerebral infarction. Using this test, the general functional impairment score (0-28 points) and focal functional impairment score (0-28 points) were applied to evaluate the injury of mice. The lowest number of points (0) indicated normal function, and the highest score (28 points) represented the most severe level of functional impairment.

For histopathological staining (72 h postreperfusion), frozen brain sections (12 μm thick) were used for Nissl staining, and colon tissues were embedded in paraffin (4 μm thick) with hematoxylin and eosin (H&E) to reveal histopathological lesions. Images of the Nissl staining and H&E staining results were captured using a fluorescence microscope (Carl Zeiss, Germany).

Mouse brain tissue and colon tissue were collected from each group. Protein samples were separated on SDS−PAGE gels then electrotransferred to nitrocellulose membranes (Millipore, Burlington, MA, USA). The primary antibodies were hybridized with membranes. The primary antibodies included anti-IL-10 (1:200, ab9969, Abcam, Cambridge, MA, USA) and anti-IL-17 (1:200, ab79056, Abcam, Cambridge, MA, USA). The secondary antibodies incubated with membranes. The ImageJ software were used to scan and analyze protein bands. The bands of brain tissue were normalized to β-tubulin (ab 7291, Abcam, Cambridge, MA, USA), and colon tissue were normalized to β-actin (ab8227, Abcam, Cambridge, MA, USA).

The contents of the colon were harvested after 72 h, immediately frozen in liquid nitrogen and stored at -80°C. Total genomic DNA samples were extracted using the OMEGA Soil DNA Kit (M5635-02) (Omega Bio-Tek, Norcross, GA, USA). The V3-V4 region of the bacterial 16S rRNA gene was amplified by PCR. The Illumina MiSeq/NovaSeq platform was used to perform double-end sequencing of community DNA fragments. Microbiome bioinformatics was performed with QIIME 2 2019.4 with slight modification according to the official tutorials. Sequence data analyses were mainly performed using QIIME2 and R packages (v3.2.0).

The data were analyzed with SPSS 24.0 and were presented as the means ± standard deviations (SD). Comparisons among groups involved the use of one-way analysis of variance (ANOVA) and the Tukey–Kramer post hoc multiple comparisons test. Differences among groups were considered significant when the P value was < 0.05.

To determine the effects of EA and iPSC-EVs on mice subjected to cerebral ischemia−reperfusion injury, a mouse MCAO model was established according to the procedure (29). The acupoints and iPSC-EVs injection points were shown in Figure 1A . The mice were randomly divided into five groups, including the Sham operation group (sham), model group (MCAO), electroacupuncture group (EA), iPSC-EVs group (EV), and EA combined with iPSC-EVs group (EA+EV) ( Figure 1B ). Laser speckle cerebral blood flow was monitored before the operation, during the operation, and after reperfusion to ensure successful model establishment ( Figure 1C ). Cerebral blood flow in the MCAO, EA, EV, and EA+EV groups was significantly lower than that in the Sham group (P < 0.001, Figure 1D ). There was no significant difference among the MCAO, EA, EV, and EA+EV groups, and there was no difference in cerebral blood flow after recovery ( Figure 1D ). The Clark neurological function score was determined 72 h after EA and EV treatments ( Figure 1E ). The score in the MCAO group was significantly higher than that in the Sham group (P < 0.001, Figure 1E ). The Clark scores were decreased significantly after EA, EV, and EA+EV treatment 4 times compared with those of the MCAO group (P < 0.001, Figure 1E ). However, there was no statistical significance among the EA, EV, and EA+EV groups ( Figure 1E ). Moreover, Nissl staining showed that unlike in the Sham group, the structure of the cortex on the injured side in MCAO mice was incomplete, the number of Nissl bodies was reduced or they were destroyed, and the arrangement was irregular (P < 0.05, Figures 1F, H ). EA, EV, and EA+EV significantly alleviated neuronal injury (P < 0.05, Figures 1F, H ). There was no significant difference between the EA and EV groups, but significant differences were observed between the EA and EA+EV groups and the EV and EA+EV groups (P < 0.05, Figures 1F, H ). EA, EV, and EA+EV reduced injury after cerebral ischemia−reperfusion. Intriguingly, H&E staining showed that unlike in the Sham group, according to the Chiu’s score the colon tissue in the MCAO group showed obvious pathological changes (P<0.01, Figures 1G, I ), including intestinal mucosal edema, integrity destruction, villi damage, breakage and stripping, and a reduction in the number of goblet cells. The EA, EV, and EA+EV treatment groups exhibited significantly fewer pathological changes of intestinal mucosal edema, villus damage, and goblet cell depletion in colon tissue than the MCAO group (P < 0.05, P < 0.05, P < 0.01, Figures 1G, I ).

To confirm the protective effects of EA and iPSC-EVs in the brain and colon tissues in MCAO mice, the expression levels of the inflammatory cytokines IL-17 and IL-10 on the brain and colon tissues were measured by western blot. The intestinal flora was dysregulated after cerebral ischemia. IL-17 and IL-10, which were proinflammatory and anti-inflammatory factors, respectively, played an important role in intestinal injury and brain injury. In the brain tissue on the side of the injury, IL-17 expression was significantly increased in the MCAO group (P < 0.05, Figure 2A ) and was significantly decreased in the EA, EV, and EA+EV groups (P < 0.05, Figure 2A ). IL-10 expression increased in the MCAO group (P < 0.05, Figure 2A ) and significantly increased after EA, EV and EA+EV treatments (P < 0.05, P < 0.05, P < 0.01, Figure 2A ), and EA+EV treatment was more effective than EA and iPSC-EVs treatments alone.

Similarly, in colonic tissue, IL-17 expression was significantly increased in the MCAO group (P < 0.05, Figure 2B ) and significantly decreased in the EA, EV and EA+EV groups (P < 0.01, P < 0.05 and P < 0.01, Figure 2B ). IL-10 levels increased in the MCAO group (P < 0.05, Figure 2B ) and were significantly increased after EA, EV and EA+EV treatments (P < 0.05, Figure 2B ). Together, these results suggested that EA, EV, and EA+EV treatments regulated inflammatory factors to exert a protective effect on the brain and intestinal tissue, and EA+EV treatment displayed a more obvious effect.

To further explore the possible protective mechanism of EA, EV, and EA+EV treatments, we performed stool 16S rRNA gene sequencing analysis to distinguish between the intestinal microbiota composition in normal mice and MCAO mice. QIIME 2 2019.4 software was applied for microbiome bioinformatics with slight modification according to the official tutorials. Sequence data analyses were mainly performed using QIIME2 and R packages (v3.2.0) as shown in Figure 3 .

Regarding the number of microorganisms, the abundance of intestinal microbial species in the MCAO group was decreased significantly compared with that in the Sham group. EA, EV, and EA+EV treatments restored the abundance of intestinal microbial species ( Figures 4A, B ). It was very interesting that the number of taxonomic units in the MCAO group was significantly lower than that in the Sham group. EA treatment significantly improved the number of taxonomic units in the intestinal microbiota (P < 0.05, Figures 4C, D ), and the effect was more significant in this group than in the EV and EA+EV groups (P <0.01, Figures 4C, D ; Figure S1 ). Based on the statistics obtained from the flattened amplicon sequence variant (ASV)/operating taxonomic units (OTU) table, the specific taxonomic composition of microbial communities and the number of taxa contained in each sample at each taxonomic level were calculated. The phylum level analysis showed that the abundance of Proteobacteria in the MCAO group was significantly higher than that in the Sham group ( Figure 4E ), indicating the dysregulation of intestinal microorganisms. EA, EV, and EA+EV treatments decreased their levels, and the effect of EA was more significant than those of EV and EA+EV treatments ( Figure 4E ). Bacteroidetes was a phylum involved in metabolic activities in the colon. EA+EV treatment was associated with a high proportion of Bacteroidetes compared with that of the Sham group ( Figure 4E ). Firmicutes and Verrucomicrobia were the most beneficial bacteria, and a high proportion was observed in the Sham and EA groups ( Figure 4E ). In particular, Akkermansia was associated with cardiovascular and cerebrovascular diseases and decreased blood flow after cerebral ischemia, promoting the balance of metabolism in the body (42, 43). The abundance of Akkermansia was significantly increased in the EA group ( Figures 4F, G ). These findings indicate that EA has a more substantial effect on regulating the balance of metabolism. EV and EA+EV treatments had more prominent effects on the overall number of gut microbes, and EA was potentially more involved in regulating the specific taxa and the taxonomic composition.

The alpha diversity index represents the diversity of species and the beta diversity index represents the diversity of species between habitats to comprehensively evaluate the overall diversity of species. Alpha diversity refers to indexes including the richness, diversity, and evenness of the species in partially uniform habitats and is also known as within-habitat diversity. Here, we used richness (Chao1, Observed Species), diversity (Shannon, Simpson), diversity based on evolution (Faith’s PD), evenness (Pielou’s Evenness) and coverage (Good’s coverage) to comprehensively analyze alpha diversity.

Regarding the alpha diversity index, there were significant differences between the MCAO and Sham groups in richness and diversity and in diversity based on evolution, evenness, and coverage (P <0.05) ( Figures 5A, B ). The Chao1 value and number of observed species (the richness sparse curve) were more significant in the EV group; the Shannon, Simpson’s and Pielou’s evenness (the diversity and evenness) values were more significant in the EA+EV groups; the Faith’s PD (the diversity based on evolution) value was more significant in the EA group; and Good’s coverage index represented coverage with almost no difference ( Figure 5A ).

Beta diversity, also known as between-habitat diversity, refers to the dissimilarity of species composition or the replacement rate of species along environmental gradients among different communities that change along environmental gradients. Using nonmetric multidimensional scaling analysis (NMDS) unconstrained ranking, dimensionality reduction of multidimensional microbial data can be carried out. Finally, the main trends in the data can be observed by analyzing the distribution of samples in the continuous ranking axis.

When the NMDS value is less than 0.2, the results of the NMDS analysis are more reliable. The weighted unifrac (Stress=0.0902, P=0.012, R2=0.177) and unweighted unifrac (Stress=0.116, P=0.001, R2=0.143) results in the Sham, MCAO, EA, EV, and EA+EV groups were statistically significant ( Figures 5C, D ). Sham and MCAO resulted in significantly weighted unifrac (Stress=0.0625, P=0.025, R2=0.172) and unweighted unifrac (Stress=0.0804, P=0.002, R2=0.146) values ( Figures 5E, F ). The stress values in the MCAO and EA, EV, EA+EV groups calculated in weighted unifrac (Stress=0.0823, P=0.054, R2=0.138) and unweighted unifrac (Stress=0.113, P=0.029 R2=0.108) stress assessments were less than 0.2 and were statistically significant ( Figures 5G, H ).

In summary, these results suggest that EA, EV, and EA+EV could improve alpha and beta diversity, especially with EV and EA+EV treatments, which exhibited advantages in the alpha diversity index. Regarding improvements in the beta diversity index, EA was better than EV and EA+EV treatments.

After exploring the differences in microbial community composition through assessments of beta diversity, we investigated which species were responsible for these differences. LEfSe analysis was applied for this analysis. To further compare the differences in species composition among samples and show the distribution trend of species abundance in each sample, heatmaps were used for species composition analysis.

The LEfSe method was used to identify the differences between the Sham and MCAO groups ( Figure 6A ) and among the MCAO groups subjected to EA, EV, and EA+EV treatments at different classification levels ( Figure 6B ). Based on the comparisons of the MCAO and Sham groups, Proteobacteria may be a pathogenic bacterium related to the mechanism of postischemic injury ( Figure 6A ). Based on comparisons between the EA, EV, and EA+EV treatment groups, EA improved intestinal microbiota imbalance by affecting Actinobacteria, Coriobacteria, and Verrucomicrobia and EA+EV through Verrucomicrobia ( Figure 6B ). It can be concluded that EA resulted in the most significant improvement in intestinal microbiota ( Figure S2A–E ).

Heatmaps were generated to show the distribution and abundance of each species at the phylum level ( Figure 6C ). To further compare the differences in species composition among samples and show the distribution trend of species abundance in each sample, heatmaps were used for phylum composition analysis. By default, we used abundance data from the top 20 genera with average abundance for heatmapping. The EA group was significant at the phylum level (Firmicutes, Tenericutes, Chloroflexi, Actinobacteria, Cyanobacteria, Verrucomicrobia, Gemmatimonadetes, Rokubacteria, Bacteroidetes). The MCAO group was significant at the phylum level (Proteobacteria, Acidobacteria, Patescibacteria, Epsilonbacteraeota, Halanaerobiaeota, GAL15). The EV group exhibited Deferribacteres at the phylum level (Mucispirillum schaedleri, Parabacteroides goldsteinii). The Sham and EA+EV groups exhibited specific regulation at the phylum level.

The above analysis focused on the diversity and species composition of the flora. For the study of microbial ecology, we also paid attention to the functional potential of the microbiota. The functional units obtained were used to obtain the abundance value of metabolic pathways according to the metabolic pathway database and certain calculation methods. Metabolic pathways were divided into six broad categories, including metabolism, genetic information processing, environmental information processing, cellular process processes, organismal systems, and human diseases. Each of these was further divided into multiple hierarchies ( Figure 7A ).

Having obtained the functional components of the samples/groups, especially some of the differential pathways, we next investigated which species encoded the genes with these functional potentials. A stratified sample metabolic pathway abundance table was generated for species composition pathway analysis ( Figure 7B ). It was confirmed that all or most of the functions involved in metabolic pathways could be independently performed by a certain bacterial species.