Endothelial Cell‐Based Vascular Bandages for Blood–Brain Barrier Repair and Targeted siRNA Delivery
Yaosheng Li, Yunfei Dong, Yaode He, Juanjuan Zheng, Hui Liu, Lu Li, Xiaoxu Hao, Yanli Zhao, Zefeng Yang, Yuankai Sun, Zhiwei Du, Bo Zhao, Weihang Zhou, Honghui Wu, Tianyuan Zhang, Jiahe Wu, Xiangrui Liu, Xianzhen Yin, Zhicai Chen, Rui Xue, Min Lou, Zhen Gu, Xinchi Jiang
TL;DR
This paper introduces a new therapy using brain endothelial cells to repair the blood-brain barrier and deliver siRNA to treat cerebral ischemia/reperfusion injury.
Contribution
A novel brain-targeted delivery platform using vascular bandages of endothelial cells for BBB repair and siRNA delivery.
Findings
mECs target damaged cerebral vessels via VLA-4 and support BBB integrity by forming new junctions.
OGD-treated mECs enhance siRNA delivery to damaged endothelium via CX43-mediated communication.
Vascular bandage treatment rescued BBB function and reduced infarct area in ischemia/reperfusion injury.
Abstract
Changes in the blood‐brain barrier (BBB) are key targets for mitigating cerebral ischemia/reperfusion injury. The rapid progression of reperfusion injury necessitates the development of carriers that target and regulate early BBB disruption, while supporting its structure and function during BBB recovery. This study proposes the use of brain microvascular endothelial cell (mECs)‐based vascular bandages carrying siRNAs to simultaneously target, support, and regulate the damaged BBB. Specifically, mECs can target damaged cerebral blood vessels after intravenous injection by interacting with the highly expressed very late antigen ‐ 4 (VLA‐4) in the vessels. Furthermore, by covering the cerebral blood vessels and forming new junction proteins with the vascular endothelium, mECs support the permeability and structural integrity of the vasculature. Additionally, oxygen‐glucose…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6
FIGURE 7
FIGURE 8- —Natural Science Foundation of Zhejiang Province10.13039/501100004731
- —National Natural Science Foundation of China10.13039/501100001809
- —Dr. Li Dak Sum & Yip Yio chin Development Fund for Regenerative Medicine, Zhejiang University
- —Ningbo Top Medical and Health Research Program
- —Shanghai Excellent Academic Leader (Youth)
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsBarrier Structure and Function Studies · RNA Interference and Gene Delivery · Endoplasmic Reticulum Stress and Disease
Introduction
1
Ischemic stroke is the third leading cause of death globally [1]. Thrombolysis and thrombectomy are preferred treatments for ischemic stroke [2, 3, 4]. However, reperfusion injury can occur after recanalization, exacerbating brain tissue damage and leading to infarct expansion in some patients, which significantly limits the prognosis of ischemic stroke [5, 6]. Dysfunction of the blood‐brain barrier (BBB) is involved in reperfusion injury [7, 8, 9]. Patients with increased BBB permeability had a higher risk of developing reperfusion injuries (Figure S1). Further analysis confirmed that elevated BBB permeability was an independent risk factor for reperfusion injury (Tables S1 and S2). An increasing number of preclinical studies have indicated that active functional and structural changes occur in the BBB, particularly in endothelial cells, during the early stages of reperfusion [5, 10, 11], and modulation of the status and function of vascular endothelial cells alleviates reperfusion injury [11, 12, 13, 14]. Therefore, there is an urgent clinical need to develop therapeutic strategies that can rapidly restore the BBB integrity and endothelial function, thereby preventing reperfusion injury and improving the prognosis of ischemic stroke.
We previously found that stem cells and stem cell biomimetic liposomes could promote BBB repair by eliminating adverse factors through anti‐inflammatory and microenvironmental modulation [15, 16, 17, 18, 19]. Moreover, directly modulating endothelial cells with nucleic acid drugs also contributes to BBB restoration, as evidenced by studies showing that the downregulation of p66^Shc^ levels in endothelial cells via siRNA enhances BBB restoration [14]. However, interventions in endothelial cells for reperfusion injury therapy still face two main challenges: (1) targeting: In current stroke therapies, brain‐targeted delivery strategies are primarily designed to bypass the BBB and reach the brain parenchyma, where they regulate glial and immune cells involved in inflammation [20, 21], whereas delivery systems that specifically localize to and remain at the BBB itself have received limited attention; and (2) immediacy. Approaches based on microenvironmental modulation or endothelial cell regulation cannot provide immediate support for BBB function and require time to exert their effects. Given the rapid progression of reperfusion injury, substantial tissue damage may occur during the time required for BBB restoration [10, 14, 19]. These challenges highlight the need for biomaterials or delivery systems that, in addition to targeting the BBB itself, not only function as drug carriers but also actively support the structural and functional integrity of the BBB during its recovery.
Based on observations that supplementing cellular components to in vitro BBB models can enhance BBB integrity [22], this study proposes the design of vascular bandages based on brain microvascular endothelial cells (mECs) to address the above challenges in reperfusion injury therapy within a single system. First, mECs can home to abnormal BBB regions through interactions with adhesion molecules upregulated on injured cerebral vessels [23]. Second, as the fundamental structural and functional units of the BBB, mECs could contribute to barrier integrity once integrated into the vasculature [24, 25]. Third, extensive intercellular material exchange among endothelial cells further enables mECs to serve as carriers for therapeutic delivery [26]. Building on these features, this study utilized mECs as vascular bandages to deliver p66^Shc^ siRNA, thereby achieving targeted delivery, providing structural support, and functional regulation of the injured BBB to mitigate reperfusion injury.
In this study, we evaluated the ability of intravenously administered mECs to target cerebral vessels using fluorescence micro‐optical sectioning tomography (fMOST) and explored the mechanisms underlying vascular tropism. The supportive effects of mECs on BBB structure and function were evaluated in middle cerebral artery occlusion/reperfusion (MCAO/R) mice, and the possible mechanisms behind their supportive effect were preliminarily explored. Moreover, we examined the ability of mECs to deliver therapeutic siRNA to vascular endothelial cells through intercellular communication, and evaluated the synergistic protective effects of mECs and siRNA on BBB function and ischemia‐reperfusion injury. Finally, we conducted a preliminary safety assessment of the mECs‐siRNA delivery system
Results
2
Intravenous mECs Targeting and Covering the Damaged Cerebral Vasculature
2.1
Based on optimal yield and purity, we selected the mouse brain microvascular endothelial cell line bEnd.3 (Research Resource Identifier: CVCL_0170) as the mEC source. The STR profiling and cell contamination testing results are provided in Supporting Information 2, showing that the bEnd.3 cells used were free of mycoplasma contamination and had genotypes consistent with the National Infrastructure of Cell Resources (NICR) database. Flow cytometry showed that bEnd.3 cells and primary endothelial cells (BMECs) exhibited comparable expression to that of the endothelial markers CD31, VCAM‐1, and VE‐cadherin, as well as the negative marker CD45 (Figure S2A).
Next, we explored the biodistribution of DsRed‐labeled mECs in a mouse MCAO/R model. The reperfusion time point in MCAO/R is designated as +0 h, and mECs are administered via tail vein injection at +4 h. The results revealed that mECs signals could be detected in the brain from +24 to +72 h (Figure 1A,B). During this same period, mECs were rapidly cleared from other organs (Figure S2B), resulting in a significant increase in the ratio of the brain fluorescence signal to the total fluorescence signal intensity (Figure S2C). Additionally, we observed a significantly higher fluorescence signal from mECs in the lesioned than in the contralateral hemisphere (Figure 1C). These findings demonstrated that mECs can be distributed in the injured hemisphere of the MCAO/R model following intravenous injection and can remain there for several days.
mECs target and cover the blood vessels of the injured hemisphere. (A) The distribution of DsRed labeled mECs in MCAO/R mice brain from +24 to +72 h, the unit of radiant efficiency is p·s−1·cm−2·sr−1·µW−1·cm2. (B) Quantification of the DsRed signal intensity in the brain of MCAO/R mice (n = 3, biologically independent samples). (C) The ratio of fluorescent signals in the ipsilateral and contralateral hemispheres of the infarction (n = 3, biologically independent samples). (D) Flow cytometry analysis for the percentage of DsRed+ cells in MCAO/R mice at +24 or +72 h (n = 3, biologically independent samples). (E) Distribution of DsRed labeled mECs in brain section at different parts of brain (I: cortex; II: hippocampus; III and IV: perilesional area; ROI area 600 × 600 × 600 µm3). (F) The spatial relationship between mECs and brain vasculature in brain sections. White arrow indicates mECs covering on vessels. (G) Scheme of ROIs selected for fMOST analysis I: contralateral side of lesion; II: cortex; III: hippocampus; IV: perilesional area. (H) Distribution curves of mECs in the lesioned hemisphere within 1 to 5 µm from the vessel. (I) Distribution of mECs and Ves‐mECs within 4 ROIs. White arrows indicate Ves‐mECs and yellow arrows indicate for mECs. (J) The spatial relationship between mECs and Ves‐mECs with vessels from perilesional area. White arrows indicate for Ves‐mECs; yellow arrows indicate for mECs, the dashed line indicates the location of the cross section selection. Data were demonstrated as mean ± SEM. P was calculated by Two‐way ANOVA with Sidak test for multiple comparisons in (B–D).
We determined the proportion of DsRed^+^ cells in mouse brain samples to confirm the presence of mECs in the brain. Consistent with the IVIS findings, mECs were identified in the brain at +24 h, and the percentage of mECs showed a slight yet insignificant increase up to +72 h (Figure 1D; Figure S2D). These findings provide additional confirmation of the ability of mECs to reach the brain of MCAO/R mice after intravenous injection. Notably, mECs did not exhibit long‐term residence in the brain, as their presence could not be detected at +168 h (Figure S2E). Subsequently, we attempted to confirm the selectivity of mECs for the ipsilateral hemisphere of the lesion. We observed an extensive distribution of mECs in the affected hemisphere, whereas the frequency of mECs signals in the contralateral hemisphere was significantly lower (Figure S2F i,F_ii_). Furthermore, mECs were mainly concentrated in the perilesional area (Figure 1E‐III,IV), although mECs signals were also detectable in other affected areas, such as the hippocampus and cortex in relatively lower numbers (Figure 1E‐I,II). These observations confirmed that mECs were distributed in the perilesional area of the brain following intravenous injection.
We then determine whether mECs were localized to the BBB in the brain. Therefore, the spatial relationship between mECs and cerebral vessels was investigated. We labeled brain microvessels with Dylight488‐Lectin and found that mECs were closely associated with blood vessels and exhibited the morphology of bandages covering the vasculature (Figure 1F, Figure S2G). We used the fMOST technique to process brain samples from MCAO/R mice and reconstruct the cerebral vascular network to better characterize the spatial relationship between mECs and blood vessels [27, 28]. We selected a 1200 µm‐thick coronal section containing the infarct site for reconstruction and selected three regions (cortex, hippocampus, and perilesional area) with mEC distribution as regions of interest (ROIs) to observe the spatial relationship between mECs and vessels (Figure 1G).
We then quantified the distance between each mEC and the nearest blood vessel in the entire lesioned hemisphere, and observed a distinct inflection point at 1 µm on the distribution curve of mECs at different distances from blood vessels, indicating a concentrated distribution of mECs within 0–1 µm from the vessels (Figure 1H; Figure S3A). These cells were labeled as vessel‐related mECs (Ves‐mECs). We explored the detailed morphology of mECs and Ves‐mECs in the ROIs (Figure 1I). Ves‐mECs directly attached to and covered the vessels (Figure 1I,J; Figure S3B, indicated by white arrow). Meanwhile, mECs located farther than 1 µm retained a spherical shape and did not establish spatial associations with the vessels (Figure 1I,J; Figure S3B, indicated by yellow arrow). Supplementary videos have been provided to better illustrate the spatial relationship between mECs and blood vessels (Movies S1, and S2). These results confirm that mECs could co‐localize with the BBB in the brain and form a bandage‐like structure that covers the blood vessels.
Next, we examined the selective targeting of mECs to damaged blood vessels compared to normal vessels. Significant morphological changes in the vasculature of the ipsilateral hemisphere following stroke were observed, characterized by an increase in vessel diameter (Figure 2A,B) and a decrease in branching (Figure S3C). The distribution of mECs across the four ROIs correlated with vascular morphological changes, showing the highest abundance in areas with significant vessel diameter variation and branching alterations, and little to no presence in the contralateral hemisphere (Figures 1I and 2C). Notably, the number of Ves‐mECs was closely and linearly related to vessel diameter across regions (Figures 2C, S3D), suggesting that mECs preferentially accumulate in areas with pronounced vascular structural changes.
mECs exhibit selectivity for vessels with more severe structural abnormalities. (A) The morphology and diameter of cerebral microvessels in four ROIs. I: contralateral side of lesion; II: cortex; III: hippocampus; IV: perilesional area. (B) Quantitative analysis of the mean diameter of vascular segments in 4 regions (n = 3667 for ROI I; n = 2649 for ROI II; n = 1342 for ROI III; n = 1163 for ROI IV, each segment of vessels in images regarded as a sample). (C) Trend plots of vessel diameter, number of mECs, and number of Ves‐mECs within the 4 ROIs. P value indicates significance analysis of the correlation between the number of Ves‐mECs and vessel diameter. (D) Distribution of Ves‐mECs within the 4 ROIs on affected vessels (diameter greater than 7.86 µm, marked in red) as well as normal vessels (diameter less than 7.86 µm, marked in white). Blue arrows indicate Ves‐mECs around affected vessels, yellow arrows indicate Ves‐mECs around normal vessels. (E) Proportion of Ves‐mECs distributed on two groups of vessels within ROIs to the total number of mECs within the ROIs (n = 3, each ROI with a distribution of mECs is considered as a sample). Data were demonstrated as mean ± SEM. P was calculated by Student's t‐test for (E); One‐way ANOVA with Tukey post hoc test for multiple comparisons in (B); Linear correlation analyses between the two sets of data were used for (C).
We examined whether the extent of vascular damage affected the tendency of mECs to establish structural connections with the vascular system. Using the 95% confidence interval of vascular diameter in the contralateral hemisphere (Figure 2A), we set 7.86 µm as the threshold to classify blood vessels into affected (>7.86 µm) and normal (<7.86 µm) groups (Figure 2D). By calculating the proportion of Ves‐mECs around affected (Figure 2D, blue arrow) and normal (Figure 2D, orange arrow) vessels within the ROIs, we found a significantly higher proportion surrounding affected vessels (Figure 2E), indicating that mECs preferentially form structural connections with damaged vasculature.
In summary, the data indicated that mECs can target damaged cerebral blood vessels and cover the affected vessels, showing potential as a vascular bandage to support the BBB.
The Upregulated VLA‐4 on Damaged Vessels Guides the Vascular Targeting and Coverage of mECs
2.2
Next, we investigated how the mECs achieve vascular targeting. Cell adhesion molecules (CAMs) play a crucial role in mediating cellular migration and homing, and their expression is markedly altered following reperfusion [29]. Therefore, we performed a proteomic analysis to assess CAM expression profiles in mECs and changes in the cerebrovascular segments of MCAO/R mice after reperfusion. The results demonstrated that very late antigen ‐ 4 (VLA‐4) was expressed in the cerebrovascular segments and was upregulated following reperfusion, with its corresponding receptor, vascular cell adhesion molecule‐1 (VCAM‐1), detected in mECs (Figure S4A). Considering that VLA‐4 is not conventionally expressed in endothelial cells, we further performed western blot analysis to confirm its presence in primary brain microvascular endothelial cells (BMECs) and mECs (Figure S4B,C).
Changes in VLA‐4/VCAM‐1 expression in an in vitro BBB model subjected to oxygen‐glucose deprivation/reoxygenation (OGD/R) treatment were conducted to investigate whether VLA‐4 is functionally associated with reperfusion injury, the time point at the end of OGD is defined as +0 h (Figure 3A). The VLA‐4 expression continued to increase following reoxygenation, reaching an approximately five‐fold increase at +8 h (Figure 3B; Figure S4D). These results suggest a potential association between upregulated VLA‐4 expression and endothelial injury. Meanwhile, the expression of VCAM‐1 did not show a similar continuous increase (Figure 3B; Figure S4E). The lack of the involvement of other cell types may contribute to the absence of VCAM‐1 upregulation after OGD/R [30, 31].
VLA‐4/VCAM‐1 pathway orients the structural connection between mECs and vessels. (A) Schematic illustration of different treatments on BBB model: (1) OGD followed by various durations of reoxygenation; (2) The supernatant of OGD‐treated endothelial cells (OGD‐SN) was collected and added to other normal endothelial cells, which were subsequently incubated under normal conditions for different time periods. (B) Blots of VLA‐4 and VCAM‐1 expression after OGD and reoxygenation for different times. (C) Blots of VLA‐4 and VCAM‐1 after OGD supernatant treatment for different times. (D) Distribution of VLA‐4 knocked down and VCAM‐1 knocked down mECs in brains of MCAO/R mice at +72 h, the unit of radiant efficiency is p·s−1·cm−2·sr−1·µW−1·cm2. (E) Quantitative analysis of fluorescence intensity of normal mECs, VCAM‐1 KD mECs or VLA‐4 KD mECs in the brains of MCAO/R mice at +72 h. (n = 3, biologically independent sample). The spatial relation of (F) VLA‐4 KD mECs or (G) VCAM‐1 KD mECs with brain blood vessels, the orange arrows mark mECs covering blood vessels, whereas the white arrows highlight mECs lacking vascular colocalization. (H) Illustration of the process of blocking VLA‐4 in MCAO/R mice. At +2 h anti VLA‐4 antibody was injected through tail vein to block VLA‐4 in mice. (I) Distribution of DsRed labeled mECs in VLA‐4 blocked MCAO/R mice at +72 h, the unit of radiant efficiency is p·s−1·cm−2·sr−1·µW−1·cm2. (J) Quantification of mECs fluorescence signal intensity in the brains of MCAO/R mice with VLA‐4 blockage (n = 3, biologically independent samples),. (K) The spatial relation of mECs with VLA‐4 blocked blood vessels, the orange arrows mark mECs covering blood vessels, whereas the white arrows highlight mECs lacking vascular colocalization. (L) Counts of VLA‐4 KD mECs and VCAM‐1 KD mECs in MCAO/R mouse brain sections, and counts of normal mECs in VLA‐4‐blocked MCAO/R mouse brain sections (n = 5, biologically independent samples). (M) Proportion of VLA‐4 KD mECs and VCAM‐1 KD mECs covering on vessels in MCAO/R mouse brain sections, and proportion of normal mECs covering on vessels in VLA‐4‐blocked MCAO/R mouse brain sections (n = 5, biologically independent samples). Data were demonstrated as mean ± SEM. P was calculated by Student's t‐test for (J). One‐way ANOVA with Tukey post hoc test for multiple comparisons in (E), (L), and (M).
In addition to evaluating the expression levels of VLA‐4/VCAM‐1 in cells directly exposed to OGD/R, we collected the culture supernatant of OGD treatment (OGD‐SN) and replaced a part of the culture medium of normal endothelial cells with OGD‐SN (Figure 3A) to mimic the condition of the endothelium at the adjacent area of the infarct core, the time point when OGD‐SN is added is defined as 0 h. The results indicated that VLA‐4 expression in normal endothelial cells was significantly and rapidly upregulated after the addition of the OGD supernatant compared to that in cells directly subjected to OGD treatment (Figure 3C; Figure S4F). In contrast, VCAM‐1 showed only a slight upregulation at +1 h (Figure 3C; Figure S4G). These findings suggest that endothelial VLA‐4 upregulation is associated with reperfusion and that the interaction between vascular VLA‐4 and VCAM‐1 in mECs may represent a potential mechanism for the vascular targeting of mECs.
We knocked down VCAM‐1 in mECs (VCAM‐1 KD group) (Figure S4H,I), and knocked down VLA‐4 in mECs as a control (VLA‐4 KD group) to test the role of VLA‐4 in targeting of mECs (Figure S4H,I). The results indicated that VLA‐4 knockdown in the control group resulted in only a slight and non‐significant decrease in mEC signaling in the brain (Figure 3D,E; Figure S4J). In contrast, knockdown of VCAM‐1 resulted in a nearly five‐fold decrease in brain signals (Figure 3D,E; Figure S4J). Furthermore, brain sections revealed that VLA‐4‐KD mECs retained the ability to form bandage‐like structures that covered the vasculature (Figure 3F). In contrast, VCAM‐1 KD mECs exhibited reduced brain distribution and minimal coverage of blood vessels (Figure 3G).
We administered a VLA‐4 neutralizing antibody to MCAO/R mice at +2 h [19, 32, 33, 34], 2 h before delivery of normal mECs to block VLA‐4 in vessels to verify the importance of vascular VLA‐4 on the distribution process of mECs (Figure 3H). The blockade of VLA‐4 resulted in a reduction in the distribution of mECs within the brain (Figure 3I,J; Figure S4J). The mECs also failed to establish a spatial connection with blood vessels (Figure 3K). We quantified the number of mECs in the brain sections and the proportion of mECs covering the blood vessels relative to the total cell population. The results confirmed that the distribution and vascular coverage of mECs in the VLA‐4‐blocked MCAO/R mice exhibited a reduction similar to that observed after VCAM‐1 KD in mECs (Figure 3L,M). These findings demonstrated that mECs achieve vascular targeting and coverage through the interaction of surface VCAM‐1 with vascular VLA‐4.
mECs Maintain the Permeability of the BBB Following Reperfusion and Alleviate Reperfusion Injury in MCAO/R Mice
2.3
After confirming the ability of mECs to target damaged vessels, we investigated whether mECs covering the vessels supported the structure and function of the BBB. mECs intervention reduced Evans blue extravasation from the blood vessels (Figure 4A). The Evans blue content in the whole brain (Figure 4B), as well as in the ipsilateral hemisphere of the ischemic lesion (Figure 4C), was significantly lower in the mEC‐treated group than in the model group. Additionally, mECs intervention significantly reduced brain water content (Figure 4D), suggesting a lower degree of brain edema in the mEC‐treated group. MCAO/R modeling resulted in a five‐to ten‐fold increase in serum albumin content and distribution area on the injured side of the brain (Figure 4E,F; FigureS5A). mECs intervention successfully restricted the leakage of serum proteins (Figure 4E), both in terms of content (Figure 4Fi) and distribution area (Figure 4Fii). These results indicated that mECs could maintain BBB function in MCAO/R mice. To further clarify the specificity of mECs in supporting BBB integrity, we compared their effects with another cell type reported to have BBB‐protecting potential, human mesenchymal stem cells (hMSCs). Under identical administration time, mECs demonstrated a substantially stronger ability to maintain BBB permeability than hMSCs (Figure S5B). These findings highlight the distinct advantage of mECs in supporting BBB function.
mECs can maintain BBB function and alleviate reperfusion injury. (A) Representative images and quantification analysis of the Evans blue content in (B) whole brain or (C) ipsilateral hemisphere of the lesion derived from MCAO/R mice treated with mECs. The Evans blue was injected at +72 h (n = 8, biologically independent samples). (D) Water content of brain from mice in mECs treated group or MCAO/R model group at +72 h (n = 8, biologically independent samples). (E) The distribution of plasma protein components albumin leaked in the lesion hemisphere of mice's brain in the mECs treatment group and the MCAO/R model group after immunofluorescence staining. Quantification analysis of (Fi) amount and (Fii) distribution area of albumin in brain of mECs treatment group and the MCAO/R model group (n = 5, biologically independent samples). (G) Representative images of blots of ZO‐1, Occludin, and VE‐cadherin in the brain of MCAO/R mice receiving mECs intervention. (H) Quantitative results of Occludin and VE‐cadherin expression in the brain of MCAO/R mice receiving mECs intervention (n = 3, biologically independent samples). (I) TEM images of morphology of endothelium on brain microvascular (white arrow: caveola and vesicle; white star: tight junction; yellow arrow: base membrane). (J) Quantification analysis of infarction area of MCAO/R mice brain after TTC staining at +72 h (n = 8, biologically independent samples). (K) The changes in neurological function score (mNSS) of MCAO/R mice receiving mECs intervention at 72 h. (n = 8, biologically independent samples). L) Survival curve of MCAO/R mice receiving mECs intervention (n = 8, biologically independent samples). Data were demonstrated as mean ± SEM. P by Student's t‐test in graphs.
Next, we investigated the expression levels of related junction proteins in the brains of MCAO/R mice (Figure 4G). Intervention with mECs maintained Occludin and VE‐cadherin expression (Figure 4H), while that of ZO‐1 did not show significant improvement (Figure S5C). Furthermore, mECs intervention prevented the formation of tight junction gaps between endothelial cells (Figure 4I, indicated by asterisks) and reduced the number of caveolae and vesicles within the endothelial cells (Figure 4I, indicated by a white arrow; Figure S5D). The basement membrane also appeared relatively intact in the mECs intervention group (Figure 4I, indicated by the yellow arrow). These findings indicated that mECs contribute to maintaining BBB structural integrity after reperfusion.
We further assessed whether effective intervention in the BBB with mECs could mitigate the expansion of the infarct area induced by reperfusion injury. The 2,3,5‐Triphenyl‐2H‐tetrazolium chloride (TTC) staining showed that mECs intervention effectively restricted the expansion of the infarct area after post‐reperfusion (Figure S5E). The ischemic area in the mECs intervention group was significantly smaller than that in the model group (Figure 4J). Hematoxylin and eosin staining confirmed a smaller affected area in the mEC‐treated group (Figure S5F). These findings suggest that supporting the BBB barrier function through mECs can inhibit the expansion of the infarct area caused by reperfusion injury. Neurological functional assessment using the modified Neurological Severity Score (mNSS), which has been demonstrated to be reliable in mice [15, 35, 36], indicated that mECs prevented further neurological loss post‐reperfusion in MCAO/R mice (Figure 4K). The mEC‐treated mice were capable of feeding independently at an earlier stage, resulting in better outcomes in terms of weight change, although this change was not significant (Figure S5G). However, faster recovery of beam balance and walking gait performance was observed in the mEC‐treated mice (Figure S5H). Ultimately, mECs effectively reduced the mortality rate of MCAO/R mice (Figure 4L).
In summary, mECs effectively supported the function and structure of the BBB, thereby alleviating lesion expansion and neurological deficits resulting from reperfusion injury.
mECs Support the BBB through the Formation of Junction Proteins with Cerebral Vasculature
2.4
To investigate the mechanisms by which mECs support BBB integrity, an in vitro BBB model was constructed using endothelial cells. Subsequently, the mECs were introduced into the BBB after OGD/R treatment of BBB model. Similar to the coverage observed in vivo, mECs structurally integrated into the BBB model (Figure 5A), and OGD/R treatment further facilitated this integration (Figure 5A,B). The addition of mECs (Figure 5C) significantly reduced the permeability of the BBB model to dextran and Evans blue (Figure 5D) while maintaining the expression of junction proteins in the BBB (Figure S6A,B). These results indicated that mECs demonstrated a comparable ability to support the BBB in vitro, suggesting that the therapeutic effect of mECs relies primarily on their interaction with vascular endothelial cells. Notably, physically separating mECs from the BBB model (Sep‐mECs, Figure 5C) compromised their ability to reduce small molecule permeability (Figure 5E) and upregulate junction protein expression of the BBB model (Figure S6C). Similarly, treatment of the BBB model with mEC‐conditioned medium (Med‐mECs, Figure 5C) had no significant effect on the permeability or junction protein levels of the BBB models (Figure 5E; Figure S6C,D). These results indicated that the BBB‐supporting ability of mECs is contact‐dependent rather than paracrine.
mECs exert therapeutic effects through direct structural connections and the formation of junction proteins with vascular endothelium. (A) Representative images demonstrate the integration of mECs into the in vitro BBB model with or without OGD/R treatment. White arrow indicates integrated mECs. (B) The proportion of mECs integrated into the BBB model relative to the total cell count (n = 8, biologically independent samples). (C) Illustrations depict the treatments in groups of OGD/R + mECs, OGD/R + Sep‐mECs, and Med‐mECs. The amount of (Di) 4 kD FITC‐Dextran and (Dii) Evans blue permeating through OGD/R treated BBB model with or without mECs replenished at +2 h (n = 6, biologically independent samples). The amount of (Ei) 4kD FITC‐Dextran and (Eii) Evans blue permeating through OGD/R treated BBB model treated by Sep‐mECs or Med‐mECs (n = 6, biologically independent samples). (F) Blots of ZO‐1, VE‐Cadherin and Occludin expression in mECs following knockdown of ZO‐1, Occludin or VE‐Cadherin (n = 3, biologically independent samples). (G) Illustrations depict the investigation of the ability of mECs to form endothelial barriers following knock down, as well as their capacity to maintain the endothelial barrier suffered OGD/R treatment. (H) The permeability of 4 kD FITC‐Dextran through endothelial barriers formed by mECs subjected to various protein knockdowns at +2 h (n = 6, biologically independent samples). (I) The permeability of 4 kD FITC‐Dextran through a BBB model replenished with mECs subjected to different protein knockdowns following OGD/R treatment at +2 h (n = 6, biologically independent samples). (J) Distribution of ZO‐1, Occludin or VE‐Cadherin knockdown mECs in brains of MCAO/R mice at +72 h, the unit of radiant efficiency is p·s−1·cm−2·sr−1·µW−1·cm2. (K) Quantification of fluorescence signal intensity of ZO‐1, Occludin or VE‐Cadherin knocked down mECs in the brains of MCAO/R mice (n = 3, biologically independent samples). (L) Quantification analysis of the Evans blue content in the whole brain of MCAO/R mice treated with ZO‐1, Occludin or VE‐Cadherin knocked down mECs. The Evans blue was injected at +72 h (n = 8, biologically independent samples). (M) Representative image of Evans Blue leakage in brains of MCAO/R mice treated with ZO‐1, Occludin or VE‐Cadherin knocked down mECs. (N) Co‐localization of ZO‐1, Occludin, or VE‐cadherin knockdown mECs with cerebral vessels and the formation of intercellular junction proteins between mECs and vascular structures, white arrow indicates mECs. Data were demonstrated as mean ± SEM. P was calculated by Student's t‐test for (B) and (D); One‐way ANOVA with Tukey post hoc test for multiple comparisons in (E), (H), (K), (I), and (L).
To further determine whether the BBB‐supporting effect of mECs in vivo also depends on direct contact, it was noted that the VCAM‐1/VLA‐4 pathway mediates interactions between mECs and blood vessels, prompting us to investigate the BBB‐protective effects of VCAM‐1 knockdown in mECs, a condition under which these cells fail to provide vascular coverage. VCAM‐1 KD mECs could not protect BBB permeability (Figure S7A) and did not prevent neurological damage or reduce mortality associated with reperfusion injury (Figure S7B,C). In contrast, the VLA‐4 KD control group exhibited BBB support similar to that of the normal mECs (Figure S7A). Additionally, in vivo blockade of VLA‐4 resulted in the loss of the BBB‐supporting ability of mECs (Figure S7D) and diminished the effect of mECs in alleviating reperfusion injury (Figure S7E,F). These results suggested that mECs require structural interactions with the BBB in vivo to exert their supportive effects.
Given the crucial role of junction proteins in maintaining vascular permeability, we knocked down the expression of tight junction proteins ZO‐1 and Occludin, as well as the adherens junction protein VE‐cadherin in mECs, to investigate whether mECs directly participate in barrier function to support the BBB. The results indicated that the expression levels of the corresponding proteins were significantly downregulated in the knockdown mECs (Figure 5F; Figures S7G–I). Remarkably, occludin knockdown also led to a decrease in the expression of ZO‐1 and VE‐cadherin (Figure 5F; Figure S7G–I). These knockdown mECs failed to form an effective in vitro endothelial barrier and showed significantly increased permeability to dextran compared to normal mECs (Figure 5G left panel; Figure 5H). In the OGD/R‐treated BBB model, the addition of mECs with knockdown of junctional proteins also did not effectively maintain the permeability of the BBB model, exhibiting a significantly higher permeability to dextran than the normal mEC‐treated group (Figure 5G right panel; Figure 5I). These findings demonstrated that mECs lacking barrier function failed to support the OGD/R‐treated BBB model.
In the MCAO/R model, edited mECs were still distributed in the brains of MCAO/R model mice (Figure 5J,K). Knockdown of junctional proteins prevented mECs from maintaining BBB permeability (Figure 5L,M), particularly Occludin and ZO‐1 knockdown, which resulted in a significant increase in Evans Blue leakage in the brain (Figure 5L). We subsequently examined brain sections from MCAO/R mice and confirmed that edited mECs were still able to form structural connections with the vasculature (Figure 5N, indicated by a white arrow); however, compared to normal mECs, there was a deficiency in the corresponding junctional proteins between the edited mECs and the vasculature (Figure 5N). Therefore, mECs depend on the formation of junctional proteins with the vascular endothelium to maintain BBB integrity.
mECs as siRNA Carriers Enable CX43‐Dependent Vascular Endothelial Delivery
2.5
Considering the unique vascular‐targeting ability of mECs, we next investigated the feasibility of using them as drug delivery vehicles capable of intercellular material exchange with injured vascular endothelial cells, aiming not only to support BBB but also to modulate the BBB status through direct delivery of therapeutic agent. Fluorescent protein transfer between mECs and the in vitro BBB model was observed (Figure 6A), and this transfer was significantly enhanced when the BBB model involved the OGD/R treatment (Figure 6A). However, when direct contact between mECs and the BBB model was prevented using the transwell system, material exchange was markedly suppressed under both normal and OGD/R conditions (Figure 6B,C). These findings suggest that the intercellular material exchange between mECs and the BBB model is contact‐dependent and sensitive to OGD/R stimulation.
SN‐mECs are capable of carrying siRNA and delivering it to cerebral blood vessels via CX43. (A) Transfer of fluorescent protein from DsRed‐labeled mECs to normal or OGD/R‐treated BBB models (n = 4, biologically independent samples). Transfer of fluorescent protein from DsRed‐labeled mECs to (B) normal BBB models or (C) OGD/R‐treated BBB models after direct contact blocking by Transwell co‐culture system (n = 3, biologically independent samples). (D) Effect of CBX‐mediated CX43 blockade on fluorescent protein transfer from GFP‐labeled mECs to mCherry‐labeled BBB models (n = 3, biologically independent samples). (E) Changes in CX43 expression in BBB models at different time points after OGD followed by reoxygenation. (F) Changes in CX43 expression in mECs after stimulation with OGD‐SN for varying durations. (G) Proportion of siRNA loaded in mECs or OGD‐SN‐preconditioned mECs (SN‐mECs) after incubation with Cy5‐siRNA (80 nm transfection concentration) for different durations (n = 3, biologically independent samples). (H) Intracellular distribution of siRNA and colocalization of siRNA with lysosomes in mECs and SN‐mECs at 24 h post‐transfection, white arrows indicate siRNA colocalized with lysosomes. (I) Release curve of Cy5‐siRNA from mECs or SN‐mECs into OGD/R‐treated BBB models (n = 3, biologically independent samples). The preceding P value indicates the difference between the mECs and SN‐mECs groups, while the subsequent P value reflects the difference between the SN‐mECs and SN‐mECs+CBX groups. (J) Changes in GFP fluorescence intensity in GFP‐labeled BBB models after 24 h of co‐culture with mECs or SN‐mECs loaded with GFP siRNA (n = 5, biologically independent samples). (K) Biodistribution of Cy5‐siRNA in major organs of MCAO/R mice following intravenous injection of mECs or SN‐mECs loaded with Cy5‐siRNA. mECs and SN‐mECs were injected 24 h after reperfusion, the unit of radiant efficiency is p·s−1·cm−2·sr−1·µW−1·cm2. (L) Localization of Cy5‐siRNA‐loaded mECs or SN‐mECs and their associated siRNA signals in cerebral vasculature. Data were demonstrated as mean ± SEM. P was calculated by Student's t‐test for (A–C,G); One‐way ANOVA with Tukey post hoc test for multiple comparisons in (D), (I), and (J).
Connexin 43 (CX43) is a key protein involved in intercellular communication. Treatment of both mECs and the BBB model with the CX43 inhibitor carbenoxolone (CBX) abolished the OGD‐induced enhancement of intercellular material exchange (Figure 6D). Subsequent analyses revealed that CX43 expression was significantly upregulated in both BBB models under OGD/R conditions and in mECs stimulated with OGD‐SN (Figure 6E,F; Figure S8A,B), indicating that the enhanced material exchange between mECs and the BBB following OGD stimulation is associated with CX43 upregulation.
As peak CX43 expression in mECs occurred at +48 h, we preconditioned mECs with OGD‐SN for 48 h to generate SN‐mECs with enhanced delivery capacity. We further examined the capacity of SN‐mECs to carry siRNA. Intracellular siRNA levels in mECs peaked 10 h after incubation with 80 nm siRNA, with SN‐mECs showing significantly higher siRNA loading efficiency than unconditioned mECs (Figure 6G). The siRNA remained within the mECs for at least 72 h, and the degradation rate in SN‐mECs was not significantly different from that in normal mECs (Figure S8C). We further examined the intracellular localization of the siRNA. At 24 h post‐transfection, the siRNA was predominantly retained in the cytoplasm rather than co‐localized with lysosomes in both mECs and SN‐mECs (Figure 6H). Notably, SN‐mECs exhibited a stronger intracellular siRNA signal than mECs (Figure 6H), suggesting that SN‐mECs are capable of retaining siRNA intracellularly and may possess an enhanced potential for siRNA loading.
Next, we investigated the ability of SN‐mECs to transfer siRNAs into a BBB model. The time point at which SN‐mECs were added was defined as 0 h. After co‐culturing with Cy5‐siRNA‐loaded SN‐mECs, intracellular siRNA levels in the BBB model reached a plateau at +24 h and remained stable up to +48 h. (Figure 6I). Moreover, the ratio of siRNAs delivered to the BBB model by SN‐mECs was significantly higher than that delivered by unconditioned mECs (Figure 6I). CBX treatment markedly reduced the proportion of siRNA transferred from SN‐mECs to the BBB model and delayed the time required to reach a plateau (Figure 6I). To assess whether siRNAs delivered by SN‐mECs could exert biological regulatory effects, we loaded SN‐mECs with GFP siRNA sequence and monitored its effect on GFP protein expression in a GFP‐labeled BBB model. SN‐mECs carrying GFP siRNA induced a more pronounced downregulation of GFP protein than mECs, and this effect was abolished by CBX treatment (Figure 6J). These findings indicated that SN‐mECs can deliver siRNA to the BBB model in a CX43‐dependent manner and exhibit enhanced delivery efficiency compared to normal mECs. We further compared the siRNA‐transfer efficiency of SN‐mECs with that of hMSCs. When loaded with equal amounts of Cy5‐siRNA, SN‐mECs delivered a higher level of siRNA to the OGD‐injured BBB model than hMSCs (Figure S8D,E). These findings further highlight the advantage of SN‐mECs in transferring siRNA through intercellular material exchange.
Subsequently, we investigated the ability of SN‐mECs to deliver siRNA to MCAO/R mice. After OGD‐SN treatment, SN‐mECs were still distributed at the brain lesion site, and there were no significant differences in the cellular signals of mECs and SN‐mECs in the brain (Figure S8F). Moreover, the fluorescent signals of the siRNA carried by both mECs and SN‐mECs were predominantly concentrated in the brain lesion area, with minimal distribution in other organs (Figure 6K). The SN‐mEC group exhibited a slightly higher siRNA signal in the brain than the mEC group (Figure 6K). These results suggested that SN‐mECs retained the brain‐targeting ability of unmodified mECs and could effectively deliver siRNAs to the brain.
To investigate the localization of siRNA delivered by SN‐mECs in the brain, particularly in relation to the cerebral vasculature, we examined the distribution of siRNA and mEC signals in brain sections. Both mECs and SN‐mECs covered the brain vasculature; however, in the mEC group, siRNA primarily resided within the cells and exhibited minimal co‐localization with blood vessels (Figure 6L). In contrast, siRNA delivered by SN‐mECs not only remained within the SN‐mECs but also demonstrated extensive co‐localization with cerebral vessels (Figure 6L). These findings indicated that SN‐mECs possess a superior capacity to transfer siRNA to the brain vasculature in vivo.
Delayed SN‐mECs‐p66Shc Treatment Modulates BBB Function and Alleviates Reperfusion Injury
2.6
We then explored whether SN‐mECs can regulate BBB function by delivering siRNA, and whether this effect synergizes with the supportive role of mECs in the BBB. Given the role of p66^Shc^ in endothelial injury and BBB dysfunction following reperfusion, we designed a siRNA targeting p66^Shc^ (Figure S8G,H). SN‐mECs loaded with p66^Shc^ siRNA (SN‐mECs‐p66^Shc^) effectively downregulated p66^Shc^ expression in the OGD/R‐treated in vitro BBB model (Figure 7A; Figure S8I). Compared to SN‐mECs loaded with negative control siRNA (SN‐mECs‐NC) and mECs loaded with p66^Shc^ siRNA(mECs‐p66^Shc^), SN‐mECs‐p66^Shc^ restored BBB integrity more efficiently, as reflected by reduced dextran permeability (Figure 7B). Notably, CBX adversely affected the regulation of BBB proteins and functions by SN‐mECs‐p66^Shc^, suggesting a relationship between permeability improvement and CX43‐mediated substance transfer (Figure 7A,B). These results indicated that mECs can regulate BBB model function by delivering siRNA, thereby improving the BBB model permeability.
SN‐mECs loaded with p66Shc siRNA regulate BBB function and mitigate reperfusion injury under delayed administration. (A) Expression of p66Shc on the BBB model after co‐culture of SN‐mECs loaded p66Shc siRNA (SN‐mECs‐p66Shc) with OGD/R‐treated BBB models. (B) Dextran permeability of the OGD/R treated BBB following co‐culture with SN‐mECs‐p66Shc (n = 3, biologically independent samples). (C) Expression of p66Shc in the brains of MCAO/R mice at +72 h when treated with SN‐mECs‐p66Shc at +24 h. (D) Amount of Evans blue extravasation in the brains of MCAO/R mice at +72 h when treated with SN‐mECs‐p66Shc at +24 h (n = 8, biologically independent samples). (E) Cerebral infarct volume in MCAO/R mice at +72 h when treated with SN‐mECs‐p66Shc at +24 h (n = 8, biologically independent samples). (F) mNSS scores changes of MCAO/R mice from +24 h to +72 with SN‐mECs‐p66Shc therapy at +24 h (n = 8, biologically independent samples). (G) Survival curve of MCAO/R mice after treatment with SN‐mECs‐p66Shc (n = 8, biologically independent samples). (H) p66Shc expression level in MCAO/R brain at +72 h after treated with p66Shc‐siRNA‐loaded liposomes, co‐injection of SN‐mECs with p66Shc‐siRNA‐loaded liposomes or SN‐mECs‐p66Shc. (I) Evans blue leakage in the brains of MCAO/R mice at +72 h after treated with p66Shc‐siRNA‐loaded liposomes, co‐injection of SN‐mECs with p66Shc‐siRNA‐loaded liposomes or SN‐mECs‐p66Shc (n = 5, biologically independent samples). (J) Evans blue leakage in MCAO/R mice after treatment with different doses of SN‐mECs‐p66Shc (LD: 5 × 106 cells for each mouse; MD: 1.5 × 106 cells for each mouse; HD: 3 × 106 cells for each mouse) and at different SN‐mECs‐p66Shc administration time points (HD at +24 h, HD at +36 h, and HD at +48 h (n = 5, biologically independent samples). Data were demonstrated as mean ± SEM. P was calculated by One‐way ANOVA with Tukey post hoc test for multiple comparisons in (B), (D–F), (I), and (J).
In Section 2.3, we demonstrated that administering mECs at +4 h maintained BBB permeability (Figure 4A), whereas delaying treatment at +24 h would significantly reduce their protective effect (Figure S8J). Considering the uncertainty of treatment timing in patients, extending the treatment window is crucial [2, 3, 4]. Therefore, we investigated the effects of administering SN‐mECs‐p66^Shc^ on BBB regulation at +24 h and examined whether combined siRNA‐mediated BBB regulation could produce a synergistic effect with the BBB‐supporting effects of mECs, thereby improving efficacy. SN‐mECs‐p66^Shc^ successfully suppressed cerebral p66^Shc^ levels in vivo (Figure 7C; Figure S8K) and preserved BBB integrity more effectively than unmodified mECs when administrated at +24 h (Figure 7D). TTC staining revealed that SN‐mECs‐p66^Shc^ treatment significantly reduced infarct volume relative to the other groups at +72h (Figure 7E; Figure S8L). Behavioral assessments further demonstrated that SN‐mECs‐p66^Shc^ effectively prevented neurological deterioration after reperfusion (Figure 7F). Compared to unmodified mECs‐p66^Shc^, SN‐mECs‐p66^Shc^ prevented the decline in neurological function and mortality observed at +72 h (Figure S8M). Ultimately, treatment with SN‐mECs‐p66^Shc^ at +24 h significantly improved survival rates in MCAO/R mice compared with treatment with normal mECs‐NC. (Figure 7G)
To verify the synergistic interaction between siRNA delivery via intercellular communication through SN‐mECs and their inherent supportive function, we prepared 1,2‐dioleoyl‐3‐trimethylammonium‐propane (DOTAP)‐based cationic liposomes and compared the therapeutic efficacy of SN‐mECs‐p66^Shc^ with that of SN‐mECs mixed with p66^Shc^‐loaded liposomes. At a 1:1 mass ratio of Soybean lecithin (S100) to DOTAP, the resulting liposomes exhibited an appropriate particle size (Figure S9A,E) and zeta potential (Figure S9B), effectively encapsulated siRNA (Figure S9C), and supported efficient cellular transfection (Figure S9D). p66^Shc^‐loaded liposomes alone partially suppressed the MCAO/R‐induced upregulation of p66^Shc^, but their effect remained noticeably weaker than that of SN‐mECs‐p66^Shc^, despite carrying an equivalent amount of siRNA (Figure 7H; Figure S9F). Furthermore, combining unloaded SN‐mECs with p66^Shc^‐loaded liposomes did not yield additional p66^Shc^ inhibition (Figure 7H; Figure S9F). These observations suggest that the injured‐vessel targeting capability of SN‐mECs, together with their ability to exchange materials with the BBB, enables more efficient siRNA delivery and more effective BBB regulation than liposomal formulations.
We subsequently assessed the therapeutic performance of co‐administering unloaded SN‐mECs with p66^Shc^‐loaded liposomes in maintaining BBB permeability, compared to SN‐mECs‐p66^Shc^. Liposomes by themselves showed limited ability to preserve BBB integrity, and the combined administration of unloaded SN‐mECs and liposomes failed to achieve the protective effect observed with SN‐mECs‐p66^Shc^ (Figure 7I). Collectively, these results indicate a synergistic effect between the supportive function of mECs and their siRNA‐delivery capability.
Finally, the dose‐ and timing‐dependent effects of SN‐mECs‐p66^Sh^ ^c^ were evaluated. When the therapeutic dose was reduced to half of the standard dose, SN‐mECs‐p66^Shc^ still partially preserved BBB permeability (MD group), although the protective effect was no longer statistically significant compared with the MCAO/R group (Figure 7J). Further decreasing the dose to one‐sixth of the standard dose (LD group) completely abolished the BBB protection, indicating a clear dose‐dependent therapeutic relationship (Figure 7J). In parallel, delaying administration at the normal therapeutic dose progressively weakened treatment efficacy. Postponing SN‐mECs‐p66^Shc^ delivery to +36 h partially reduced its protective effect on BBB integrity, and further delaying treatment to +48 h resulted in an even more pronounced loss of efficacy, leaving only negligible BBB protection (Figure 7J). These findings indicate that the therapeutic benefit of SN‐mECs‐p66^Shc^ is time‐sensitive and that administration within 24 h after reperfusion is essential for achieving optimal BBB protection.
Taken together, siRNA‐loaded SN‐mECs can regulate the BBB, enhance therapeutic efficacy, and extend the intervention window to +24 h when synergized with the supportive effects of mECs on the BBB.
Preliminary Safety Evaluation of SN‐mECs‐p66Shc
2.7
We conducted a preliminary safety evaluation of SN‐mECs at last. Figure S2D demonstrated that mECs became undetectable in the brain by 168 h post‐injection. To further assess safety of SN‐mECs‐p66^Shc^, we first examined immune cell infiltration in the brain after SN‐mECs‐p66^Shc^ administration. Flow cytometry analyses showed no increase in peripheral immune cell population at brain at 24 or 168 h post‐injection (Figure 8A,B), indicating that mECs would not cause immune infiltration in brain.
Preliminary safety assessment of SN‐mECs‐p66Shc. Population changes of (A) neutrophils and (B) peripherally infiltrated macrophages in the brain of healthy mice at 24 or 168 h after SN‐mECs‐p66Shc injection (n = 3, biologically independent samples). (C) Fibrinogen blockage in cerebral blood vessels at 24 and 168 h post‐injection of SN‐mECs‐p66Shc. (D) Concentration of IL‐1beta and (E) IL‐6 in peripheral plasma of healthy mice at 24 and 168 h after SN‐mECs‐p66Shc injection (n = 3, biologically independent samples). Data were demonstrated as mean ± SEM. P was calculated by two‐way ANOVA with Sidak test for multiple comparisons in (A), (B), (D), and (E).
Next, we evaluated the potential for microvascular thrombosis caused by SN‐mECs‐p66^Shc^. Fibrinogen staining revealed no evidence of thrombosis at 24 or 168 h post‐injection (Figure 8C), suggesting that mEC administration may not increase thrombotic risk.
Finally, we measured pro‐inflammatory cytokines IL‐6 and IL‐1β in peripheral plasma. ELISA results indicated no elevation of these cytokines following SN mEC treatment (Figure 8D,E), implying no systemic inflammatory response was provoked.
These findings suggest that SN‐mECs‐p66^Shc^ is unlikely to induce acute safety concerns.
Discussion and Conclusion
3
BBB dysfunction plays a crucial role in the progression of various neurological disorders, including cerebral ischemia/reperfusion injury. However, the treatment of the BBB in ischemia/reperfusion injury is challenging owing to issues of targeting and immediacy. Therefore, there is a need to develop delivery systems that can effectively target the damaged BBB while simultaneously providing structural and functional support. In this study, we propose using mECs as carriers for siRNA to target, support, and regulate the BBB (Graphic abstract).
We used the bEnd.3 cell line as a source of mECs. Although bEnd.3 is widely applied as a substitute for primary cells in in vitro models, as an immortalized line, it shows lower expression of tight junction proteins than primary cells. VCAM‐1 expression and tight junction formation are critical for vascular bandage functions of mECs. However, isolating primary brain endothelial cells is laborious, often contaminated by pericytes or smooth muscle cells, and their proliferative capacity is limited, resulting in low yield. Therefore, the source of mECs was a major limitation of this study, which may restrict its clinical translation. The use of iPS cell–derived mECs may represent a promising strategy to overcome this limitation, as iPSCs are guided through defined differentiation steps to generate endothelial cells that exhibit key BBB properties [37]. These iPSC‐derived cells express hallmark endothelial markers such as CD31 and VE‐cadherin, BBB‐specific proteins including GLUT1, tight junction components such as ZO‐1 and claudin‐5, and functional efflux transporters such as P‐glycoprotein [37, 38, 39]. Studies have shown that iPSC‐derived brain endothelial cells closely resemble primary cells in both molecular and functional characteristics, suggesting significant potential for clinical applications [24].
Intravenous injection of mECs effectively targeted damaged cerebral blood vessels in MCAO/R mice. High expression of VLA‐4 in the vascular endothelium following reperfusion may play a dominant role in targeting mECs. Notably, VLA‐4 expression is typically absent in the endothelium, and our study highlights the connection between vascular VLA‐4 and endothelial injury during reperfusion; however, the biological role of endothelial VLA‐4 requires further investigation. Moreover, the interaction between VLA‐4 on mECs and VCAM‐1 on the vasculature did not appear to be decisive in the targeting process, potentially due to the relatively low expression levels of VLA‐4 in normal mECs. Nonetheless, considering that the KD of VLA‐4 in mECs has a mild and insignificant effect on brain signaling, the interaction between VLA‐4 in mECs and VCAM‐1 in the vasculature remains a feasible mechanism for targeting. Furthermore, VLA‐4 is also expressed in other brain cell types, such as microglia [40], and the potential of mECs to target other brain cells requires further exploration.
Upon reaching the cerebral blood vessels, mECs can cover the vascular endothelium and directly participate in the barrier function of the BBB, maintaining its structural and functional integrity. Targeting mECs to the cerebral vasculature via their interactions with vascular VLA‐4 may represent a preliminary step in the formation of further structural connections. Following the knockdown of VCAM‐1, both the quantity and proportion of mECs covering the vessels were significantly reduced, leading to a marked decrease in the BBB‐supporting effects of mECs. However, further investigation is necessary to identify other steps involved in the structural connection between mECs and vasculature. Additionally, given that mECs participate in barrier function, there are specific requirements regarding the source of mECs used. Compared to endothelial cells from other sites, brain microvascular endothelial cells exhibit higher expression of tight junctions and more restricted endocytic and transport behaviors [29]. The ability of endothelial cells from other tissues to support the BBB requires further investigation. Other cell types, such as human mesenchymal stem cells (hMSCs) and pericytes, have been found to play a critical role in vascular permeability in studies involving vascular organoids [22, 25]. However, these cells primarily exert their effects by modulating the endothelial cell states, and their direct involvement in barrier function remains to be validated.
Furthermore, BBB dysfunction following reperfusion injury demonstrates self‐repair capacity; both clinical and animal studies have shown that BBB function in the affected hemisphere gradually recovers after the acute phase [8, 41]. Consequently, the presence of mECs in the brain for 7 days may be sufficient to support the BBB, and the significance of extending the retention time of mECs in the brain for therapeutic purposes requires further investigation. However, it should be noted that the immunogenicity of mECs may still be an important factor limiting their retention and function at the lesion site. Although endothelial cells are not as strongly immunogenic as antigen‐presenting cells, we found that circulating mECs are cleared within approximately 2 h, and mECs trapped in off‐target tissues are largely eliminated within 24 h. In our study, a relatively high cell dose was therefore used to ensure adequate accumulation of mECs in the ischemic region. Notably, our experimental results indicated that mECs did not induce a significant immune response, and their rapid clearance after exerting therapeutic effects may be beneficial in reducing the safety risks associated with cell therapy. Future strategies such as engineered modification or surface masking of mECs to reduce rapid immune clearance may help improve their persistence at the lesion site and allow dose reduction, thereby minimizing systemic risks [20]. Nevertheless, how to attenuate immunogenicity without compromising the targeting capability or integrative function of mECs requires further investigation.
mECs could also realize material exchange with the BBB model and enable the delivery of siRNA. OGD stimulation significantly promotes the material exchange between mECs and BBB model. We observed that the upregulation of CX43 expression may play an important role in this process; however, blocking CX43 did not completely disrupt material exchange, but rather reduced it to levels comparable to those observed without OGD stimulation. This suggests the existence of other forms of inter‐endothelial communication, and the role of this exchange in drug delivery needs further investigation. Additionally, studies have shown that endothelial cells exhibit increased uptake under stress, which may explain the higher siRNA loading efficiency in SN‐mECs [40]. However, further engineering of mECs to enhance their drug loading capacity is crucial for improving therapeutic efficacy and expanding potential applications. Furthermore, in this study, we only improved siRNA stability by adding deoxythymidine dinucleotide on siRNA sequence. Introducing responsive release system to enhance siRNA stability within mECs could improve their ability to regulate the BBB.
In summary, we developed a siRNA delivery system based on mECs that demonstrates the capability to target, support, and regulate the BBB. However, further research is needed to elucidate the biological processes related to the targeting, support, and regulation of mECs as drug delivery vehicles. We believe that mECs, as a novel delivery system, could provide new insights for brain‐targeted carriers and the treatment of cerebrovascular‐related diseases.
Methods
4
In Vitro Replenishment of mECs for Treatment of BBB Model
4.1
bEnd.3 BEND3 were kindly provided by Wuhan Procell Biotechnology Co.,Ltd, Research Resource Identifiers: CVCL_0170. The bEnd.3 cells were cultured in modified DMEM medium with 1.5 mg·L^−1^ NaHCO_3_ (WISENT, 319‐007, Canada) supplemented with 10% fetal bovine serum (FBS) (WISENT, 086–150, Canada) and 1% penicillin‐streptomycin solution (Cienry, CR‐15140, China). Cells were maintained at 37°C in a humidified incubator with 5% CO_2_. When the cells reached 80%–90% confluence, they were digested with trypsin (Cienry, CR‐27250, China) for 1 minute and passaged at a 1:3 ratios, typically every 3 days. The 3rd to 12th passages of the cells were used for the experiments. After establish OGD models, the time point at which OGD ended was designated as 0 h, normal mECs (bEnd.3 cells), junction protein knockdown mECs or SN‐mECs carrying p66^Shc^ siRNA (accounting for 5% of the total original cells) were replenished to BBB model at 0 h. At +12 h or +2 h, the junction proteins expression or the permeability of BBB model was measured. The integrated condition of mECs was observed using confocal laser scanning microscope (CLSM, Zeiss LSM 800, German) at +24 h, the cells in the BBB model were pre‐labeled with 1µg·mL^−1^ DiOC18(3) (DiO, Shanghai Maokang Biotechnology, MX4032, China) or transfected with GFP.
For exploration effect of mECs without direct interact with BBB model, mECs were replenished into upper chamber of transwell which is separated from BBB in separated mECs groups at 0 h. For mECs medium group, mECs were cultured at normal condition for 24 h and the medium were collected. When OGD ended the collected medium were used for reoxygenation.
Establishment of MCAO/R Model and In Vivo Replenishment of mECs
4.2
Male C57BL/6 mice of 5–6 weeks (Slaccs, C57BL/6Slac, China) were used for the establishment of MCAO/R model. The mice were housed in a specific pathogen‐free (SPF) environment at 22 ± 2°C, with 50% ± 10% relative humidity, under a 12‐h light/dark cycle, and had free access to food and water. They were acclimated for at least one week before experiments. Prior to the model establishment, the mice were fasted for 12 h with free access to water. The procedure was conducted according to a previous paper [15]. Briefly, during the establishment of the MCAO/R model, mice were anesthetized by inhalation of 3% isoflurane. The mice were placed supine and fixed on the surgical table, and anesthesia was maintained with continuous inhalation of 1.5% isoflurane throughout the procedure. A nylon monofilament with a rounded tip (Beijing Cinontech, 1623A4, China) was used to block the origin of the middle carotid artery (MCA) of left hemisphere. The occlusion of MCA was maintained for 60 min following reperfusion by withdrawing monofilament. Reperfusion was defined as 0 h, and all subsequent time points were uniformly expressed as +4, +24, and +72 h in the manuscript. At +4 h, mice showing significant behavioral changes (mNSS > 6) were used for subsequent experiments. Cultured mECs were digested with trypsin‐EDTA (0.25%) (Cienry, CR‐25200, China). The mECs were collected and resuspended with HBSS (Solarbio, H1046, China), each MCAO/R were administrated with 3 × 10^6^ mECs in 300 µL HBSS via tail vein injection.
For administration of SN‐mECs carrying p66^Shc^ siRNA, 24 h after establishment of MCAO/R model, cultured mECs were collected and injected at same dose as normal mECs via tail vein injection.
For the dose‐response study, the high‐dose group (HD) received a tail‐vein injection of 3 × 10⁶ SN‐mECs‐p66^Shc^; the medium‐dose group (MD) received 1.5 × 10⁶ SN‐mECs‐p66^Shc^; and the low‐dose group (LD) received 5 × 10⁵ SN‐mECs‐p66^Shc^. All groups were treated at +24 h, with the cells resuspended in 300 µL HBSS.
For the therapeutic time‐window study, 3 × 10⁶ SN‐mECs‐p66^Shc^ were administered via tail‐vein injection at +24, +36, or +48 h after MCAO/R.
For hMSC administration, MCAO/R mice were injected via the tail vein with 300,000 hMSCs suspended in 300 µL HBSS at +4 h.
Preparation and Administration of siRNA‐Loaded Cationic Liposomes
4.3
For the preparation of cationic liposomes, appropriate amounts of S100 (Shanghai yuanye Bio‐Technology, B28313, China) and DOTAP (Aladdin, D351073, China) were weighed and dissolved in ethanol so that the ratios of the two phospholipids in the system were 1:1, 1:2, 1:5, and 1:10. The lipid thin film was prepared by rotary evaporation at 30°C. After resuspending the film in an isosmotic sorbitol solution, cationic liposomes were prepared by probe sonication at 30 W. The particle size and zeta potential of the liposomes were measured using a Zetasizer (Malvern, Nano‐ZS90, UK).
For siRNA loading, siRNA was dissolved in Diethyl pyrocarbonate–treated water (Biosharp, BL510B, China) to prepare a 20 µM solution, which was then slowly added dropwise into an equal volume of cationic liposome suspension with a total phospholipid concentration of 200 µg·mL^−1^. After co‐incubation at room temperature for 30 min, free siRNA was removed by centrifugation at 3000 g using a 10 kDa ultrafiltration tube (LABSELECT, UFC‐040‐010‐PES, China). The siRNA encapsulation efficiency was calculated based on the amount of free siRNA in the supernatant.
For cationic liposome administration, at +24 h after reperfusion, MCAO/R mice were injected via the tail vein with cationic liposomes containing 1.2 µg p66^Shc^ siRNA suspended in 300 µL HBSS.
For the combined administration of SN‐mECs and cationic liposomes, MCAO/R mice were injected at +24 h with 1.2 µg p66^Shc^ siRNA–loaded liposomes and 3 × 10⁶ SN‐mECs, resuspended separately in 150 µL HBSS and mixed immediately before injection.
Biodistribution of mECs
4.4
The DsRed (GENECHEM, hU6‐MSC‐CMV‐RFP, China), DiI (Shanghai Maokang Biotechnology, MX4007, China) labeled mECs or SN‐mECs carrying p66^Shc^ siRNA were administrated as mentioned above. At +24 or 72 h, mice were sacrificed after cardiac perfusion with PBS, and major organs (Brain, heart, lungs, liver, kidney, spleen) were collected. The fluoresce signal were detected with an in vivo imaging system (IVIS Spectrum, caliper, USA), exposure time set to automatic. For imaging of mECs, a filter pair with excitation wavelength at 535 nm and emission wavelength between 570 to 720 nm were used for fluoresce signal measurement, exposure time set to automatic. For imaging of siRNA, excitation wavelength at 620 nm and emission wavelength between 660 to 820 nm were used. The spectrum of DsRed, DiI, and free siRNA were also measured in order to perform spectrum unmix to remove the background.
In Vivo Blockage of VLA‐4
4.5
According to previously reported time points of VLA‐4 blockade in stroke studies and the effective onset time of neutralizing antibodies, at +2 h, that is, 10 mg·kg^−1^ anti‐VLA‐4 antibody (Thermo Fisher Scientific, 16‐0492‐85, USA) was injected through the tail vein. mECs were administrated at +4 h, the distribution of mECs in the brain was observed using IVIS, and the survival ratio of MCAO/R mice was monitored to evaluate the influence of VLA‐4 blockade on the therapeutic efficacy of mECs.
Investigation of Material Exchange between mECs and the BBB
4.6
In the investigation of material exchange mechanisms, mECs transfected with DsRed or GFP were added to a GFP or mCherry‐labeled BBB model at a cell number ratio of 1:10, with the BBB model subjected to 2 h of OGD/R treatment. In the transwell separation group, mECs and the BBB model were separated using transwell chambers. In the CBX (GlpBio Technology Inc, GC10624, USA) treatment group, 100 µM of CBX was added to the co‐culture system concurrently with the addition of mECs. After 24 h of co‐culture, flow cytometry was employed to assess the transfer of fluorescent proteins.
For the study of siRNA delivery, mECs transfected with Cy5‐siRNA or GFP siRNA were added to the GFP‐labeled BBB model 10 h post‐transfection. The fluorescence intensity of Cy5‐siRNA or GFP protein within the BBB model was measured at various time points after transfection using flow cytometry. In the investigation of siRNA‐mediated regulation by mECs, mECs transfected with p66^shc^ siRNA were added to the BBB model subjected to OGD/R treatment 10 h post‐transfection and incubated for 12 h, after which the expression level of p66^Shc^ in the BBB model was analyzed.
Safety Evaluation
4.7
Suspensions of 3 × 10⁶ SN‐mECs or SN‐mECs‐p66^Shc^ in 300 µL HBSS were injected into 5‐week‐old healthy male C57BL/6 mice via the tail vein.
For evaluation of peripheral pro‐inflammatory cytokines, peripheral blood was collected 24 or 168 h after injection and centrifuged at 2000 g for 10 min to separate plasma. The concentrations of IL‐1β (Boster, EK0394, China) and IL‐6 (Boster, EK0411, China) in the plasma were measured using ELISA kits.
For assessment of inflammatory cell infiltration in the brain, the mouse brains were collected 24 or 168 h after injection, digested into single‐cell suspensions, and analyzed by flow cytometry to determine changes in neutrophil and macrophage populations.
For evaluation of microvascular thrombosis in the brain, frozen brain sections were prepared 24 or 168 h after injection, and the colocalization of brain microvessels with fibrinogen was examined to determine whether microthrombi had formed.
Statistical Analysis
4.8
Data were presented as bar or line graphs, with error bars indicating means and standard error of the mean (SEM), along with overlaid individual data points. Statistical analysis was performed on GraphPad Prism 8. Student's t‐test or Welch's t‐test was used for comparisons between two groups; one‐way analysis of variance (ANOVA) with Tukey's post hoc test was used for comparisons more than three groups. Two‐way ANOVA with Sidak's post hoc test for grouped comparisons. Significance was set at P < 0.05.
Funding
This work was supported by the Natural Science Foundation of Zhejiang Province (LD22H300002 and LQ21H300002), National Natural Science Foundation of China (nos. U22A20383 and 82003668), Dr. Li Dak Sum & Yip Yio chin Development Fund for Regenerative Medicine, Zhejiang University, Ningbo Top Medical and Health Research Program (no. 2022030107), and Shanghai Excellent Academic Leader (Youth) (no. 23XD1460400).
Ethics Statement
All the animal experiments were proved by animal Experimental Ethics Committee of the Zhejiang University (ZJU20240180) and performed in accordance with the committee's guidelines on animal handling informed. Consent was obtained from all participants written informed consent was obtained from all patients. The human ethics committee of the second affiliated hospital of Zhejiang University, School of Medicine approved the protocol of this study (Approved Number: Yan‐2011‐018). All clinical investigations were conducted according to the principles expressed in the Declaration of Helsinki.
Conflicts of Interest
The authors declare no conflict of interest.
Supporting information
Supporting File 1: advs73451‐sup‐0001‐SuppMat.docx.
Supporting File 2: advs73451‐sup‐0002‐MovieS1.mp4.
Supporting File 3: advs73451‐sup‐0003‐MovieS2.mp4.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1M. Brauer , G. A. Roth , A. Y. Aravkin , et al., “Global Burden and Strength of Evidence for 88 Risk Factors in 204 Countries and 811 Subnational Locations, 1990–2021: A Systematic Analysis for the Global Burden of Disease Study 2021,” Lancet 403, no. 10440 (2024): 2162–2203.38762324 10.1016/S 0140-6736(24)00933-4PMC 11120204 · doi ↗ · pubmed ↗
- 2V. Costalat , T. G. Jovin , J. F. Albucher , et al., “Trial of Thrombectomy for Stroke with a Large Infarct of Unrestricted Size,” New England Journal of medicine 390, no. 18 (2024): 1677–1689.38718358 10.1056/NEJ Moa 2314063 · doi ↗ · pubmed ↗
- 3G. W. Albers , M. Jumaa , B. Purdon , et al., “Tenecteplase for Stroke at 4.5 to 24 Hours with Perfusion‐Imaging Selection,” New England Journal of Medicine 390, no. 8 (2024): 701–711.38329148 10.1056/NEJ Moa 2310392 · doi ↗ · pubmed ↗
- 4S. Yan , Y. Zhou , M. G. Lansberg , et al., “Alteplase for Posterior Circulation Ischemic Stroke at 4.5 to 24 Hours,” New England Journal of Medicine 392, no. 13 (2025): 1288–1296.40174223 10.1056/NEJ Moa 2413344 · doi ↗ · pubmed ↗
- 5P. Yang , Z. Zhu , Y. Zang , et al., “Increased Serum Complement C 3 Levels Are Associated with Adverse Clinical Outcomes After Ischemic Stroke,” Stroke 52, no. 3 (2021): 868–877.33517703 10.1161/STROKEAHA.120.031715 · doi ↗ · pubmed ↗
- 6Y. Zhou , Y. D. He , S. Q. Yan , et al., “Reperfusion Injury Is Associated with Poor Outcome in Patients with Recanalization After Thrombectomy,” Stroke 54, no. 1 (2023): 96–104.36367100 10.1161/STROKEAHA.122.039337 · doi ↗ · pubmed ↗
- 7R. Khatri , A. M. Mc Kinney , B. Swenson , and V. Janardhan , “Blood‐Brain Barrier, Reperfusion Injury, and Hemorrhagic Transformation in Acute Ischemic Stroke,” Neurology 79, no. 13 (2012): S 52–S 27.23008413 10.1212/WNL.0b 013e 3182697 e 70 · doi ↗ · pubmed ↗
- 8C. Liu , S. Zhang , S. Yan , et al., “Reperfusion Facilitates Reversible Disruption of the Human Blood‐Brain Barrier Following Acute Ischaemic Stroke,” European Radiology 28, no. 2 (2018): 642–649.28856409 10.1007/s 00330-017-5025-3 · doi ↗ · pubmed ↗