Single‐Cell RNA Sequencing and Spatial Transcriptomics Reveal Pathogenesis of Meningeal Lymphatic Dysfunction after Experimental Subarachnoid Hemorrhage
单细胞 RNA 测序与空间转录组学揭示实验性蛛网膜下腔出血后脑膜淋巴功能障碍的发病机制

2.1. Single‐Cell Transcriptomic Atlas of mLVs Alterations at Different Time Courses after SAH
2.1. SAH 后不同时间进程下 mLVs 变化的单细胞转录组图谱

To investigate whether mLVs are impaired after SAH, we used experimental SAH models by injecting autologous blood into the prechiasmatic cisternae. This model is ideal for exploring CNS lymphatic function because it does not demand complete exposure to neck anatomy, which can avoid direct impact on mLVs (Figure S1A, Supporting Information). Also, fluorescent beads were injected intra‐cisterna magna (i.c.m.) at various time courses after SAH modeling for further assessment of the drainage function of mLVs (Figure S1A, Supporting Information). The results presented a substantial decrease in beads drainage to the dCLNs as early as 3 h after SAH, and the drainage impairment remained until 72 h post‐injury (Figure  1B and Figure S1B, Supporting Information). As for the upstream path of dCLNs, decreased drainage and wounded structure also presented in mLVs at 24 and 72 h after SAH (Figure 1I and Figure S1C, Supporting Information).
为探究蛛网膜下腔出血（SAH）后脑膜淋巴管（mLVs）功能是否受损，我们采用向视交叉前池注入自体血液的方法建立实验性 SAH 模型。该模型因无需完全暴露颈部解剖结构，可避免对 mLVs 的直接损伤，是研究中枢神经系统淋巴功能的理想选择（图 S1A ，支持信息）。此外，在 SAH 建模后的不同时间点，通过小脑延髓池（i.c.m.）注射荧光微珠以进一步评估 mLVs 的引流功能（图 S1A ，支持信息）。结果显示，SAH 后最早 3 小时即可观察到微珠向深部颈淋巴结（dCLNs）的引流显著减少，且这种引流障碍持续至损伤后 72 小时（图 1B 和图 S1B ，支持信息）。在上游路径方面，SAH 后 24 小时和 72 小时，mLVs 也表现出引流减少及结构损伤（图 1I 和图 S1C ，支持信息）。

To comprehensively elucidate the cellular composition alteration of mLVs after SAH, we performed scRNA‐seq (10× Genomics) on mice in different groups (Sham group, SAH 24 h group, SAH 72 h group, ten mice per group) (Figure 1A). After excluding damaged or dead cells and putative cell doublets out of the original scRNA‐seq data and adjacent sequent analysis, a total of 25 536 cell transcriptomes of 30 mice were retained, in which 9891 cells were originated from the sham group, 8262 cells were from SAH 24 h group, and 7383 were cells from SAH 72 h group (Figure 1C). Then we normalized the gene expression referring to sequencing depth and mitochondrial read count and applied principal component analysis (PCA) based on highly variably expressed genes across cells. To correct the batch effect, we integrated scRNA‐seq data with the Seurat integration method.^{[ 12 ]} Further, a unified uniform manifold approximation and projection (UMAP) embedding space and graph‐based clustering approach was used to identify cell clusters. Clusters were annotated with well‐known markers. The cells were classified into 16 major cell types (Figure 1C–E) including B cells marked by Cd79A, Cd79B, and Ms4a1,^{[ 13 ]} NKs marked by Ncr1, Klrb1c, and Nkg7,^{[ 14 ]} macrophages marked by C1qb, Cd68, Csf1r, and Mrc1,^{[ 15 ]} neutrophils marked by S100α9, IL1β, and Csf3r,^{[ 16 ]} monocytes marked by Ccr2 and Ly6c2,^{[ 17 ]} T cells marked by Cd3d, Cd3e, and Cd247,^{[ 18 ]} pDCs marked by Cd209a and Cd74,^{[ 19 ]} mast cells marked by Fcer1a, Kit, and Ms4a2,^{[ 20 ]} erythroblasts marked by Ttr, Enpp2, and Ptgds, erythrocyte marked by Car2, Hba‐a1, and Hbb‐bs,^{[ 21 ]} mesenchymal cells marked by Col1a1, Col1a2, Col3a1, and DCN,^{[ 13 ]} epithelial cells identified by the expression of Birc5 and Ube2c,^{[ 22 ]} endothelial cell marked by Pecam1, Cdh5, and Kdr,^{[ 13 ]} forebrain neuron marked by Gng13 and S100α5,^{[ 23 ]} glial cell marked by S100b and Cdh2,^{[ 24 ]} and neuralprecursor cells marked by Cdh2, Pde6g, and Gnb3.^{[ 24 ]} Although all 16 major cell types were presented in mLVs both in sham and SAH groups (Figure 1C), the dynamic changes of each cell type were different. Meanwhile, given that mLECs originated from endothelial cells, we further performed clustering analysis for endothelial cells and obtained three sub‐clusters. This result showed that mLECs proportion became the least at 24 h after SAH (Figure 1F), indicating that at this time course, mLVs injury might be the severest.
为了全面阐明蛛网膜下腔出血（SAH）后 mLVs 的细胞组成变化，我们对不同组别（假手术组、SAH 24 小时组、SAH 72 小时组，每组十只小鼠）的小鼠进行了单细胞 RNA 测序（scRNA-seq，10× Genomics）（图 1A ）。在排除原始 scRNA-seq 数据中的受损或死亡细胞及潜在的双细胞后，通过后续分析保留了 30 只小鼠共计 25,536 个细胞转录组，其中假手术组来源细胞 9,891 个，SAH 24 小时组 8,262 个，SAH 72 小时组 7,383 个（图 1C ）。随后，我们根据测序深度和线粒体读数对基因表达进行标准化，并基于细胞间高变表达基因进行主成分分析（PCA）。为校正批次效应，采用 Seurat 整合方法对 scRNA-seq 数据进行整合 ^{[ 12 ]} 。进一步利用统一流形逼近与投影（UMAP）嵌入空间及基于图的聚类方法识别细胞簇，并通过已知标记物对细胞簇进行注释。 细胞被分类为 16 种主要类型（图 1C–E ），包括由 Cd79A、Cd79B 和 Ms4a1 标记的 B 细胞 ^{[ 13 ]} ，由 Ncr1、Klrb1c 和 Nkg7 标记的 NK 细胞 ^{[ 14 ]} ，由 C1qb、Cd68、Csf1r 和 Mrc1 标记的巨噬细胞 ^{[ 15 ]} ，由 S100α9、IL1β和 Csf3r 标记的中性粒细胞 ^{[ 16 ]} ，由 Ccr2 和 Ly6c2 标记的单核细胞 ^{[ 17 ]} ，由 Cd3d、Cd3e 和 Cd247 标记的 T 细胞 ^{[ 18 ]} ，由 Cd209a 和 Cd74 标记的 pDC 细胞 ^{[ 19 ]} ，由 Fcer1a、Kit 和 Ms4a2 标记的肥大细胞 ^{[ 20 ]} ，由 Ttr、Enpp2 和 Ptgds 标记的成红细胞，由 Car2、Hba‐a1 和 Hbb‐bs 标记的红细胞 ^{[ 21 ]} ，由 Col1a1、Col1a2、Col3a1 和 DCN 标记的间充质细胞 ^{[ 13 ]} ，通过 Birc5 和 Ube2c 表达鉴定的上皮细胞 ^{[ 22 ]} ，由 Pecam1、Cdh5 和 Kdr 标记的内皮细胞 ^{[ 13 ]} ，由 Gng13 和 S100α5 标记的前脑神经元 ^{[ 23 ]} ，由 S100b 和 Cdh2 标记的胶质细胞 ^{[ 24 ]} ，以及由 Cdh2、Pde6g 和 Gnb3 标记的神经前体细胞 ^{[ 24 ]} 。尽管在假手术组和 SAH 组的 mLVs 中均存在这 16 种主要细胞类型（图 1C ），但每种细胞类型的动态变化各不相同。 同时，考虑到 mLECs 来源于内皮细胞，我们进一步对内皮细胞进行了聚类分析，并获得了三个亚群。结果显示，SAH 后 24 小时时 mLECs 的比例降至最低（图 1F ），表明在这一时间点，mLVs 的损伤可能最为严重。

Similarly, flow cytometric analysis showed the population of CD45^–CD31⁺LYVE1⁺ cells (mLECs) declined after SAH, and that the most serious injury happened 24 h post SAH (Figure 1H). Moreover, injured mice got lower modified Garcia scores, prolonged T _turn and T _total, prolonged deficits, and delayed recovery compared to the sham group in several neurological function tests, respectively (Figure S1D, Supporting Information). Taken together, these results demonstrated that SAH caused the injury of mLVs structure and drainage function, and elicited loss of neurological function in mice in the acute phase.
同样，流式细胞术分析显示，CD45 ^– CD31 ⁺ LYVE1 ⁺ 细胞（mLECs）群体在 SAH 后减少，且最严重的损伤发生在 SAH 后 24 小时（图 1H ）。此外，与假手术组相比，损伤小鼠在多项神经功能测试中改良 Garcia 评分较低，T _turn 和 T _total, 延长，恢复延迟（图 S1D ，支持信息）。综上所述，这些结果表明 SAH 导致小鼠急性期 mLVs 结构和引流功能损伤，并引发神经功能丧失。

2.2. Deconvolution of Spatial Atlas and Visualization of Cell Type Interactions
2.2. 空间图谱反卷积与细胞类型互作可视化

As mentioned above, scRNA‐seq data lost the spatial characteristics of each cell. Therefore, we then combined ST to improve the integrity of sequencing data and reveal the spatial distribution of all cells. After hematoxylin and eosin (H&E) staining and brightfield imaging, mLVs slides were subjected to discover distinct histological features. Standard quality control and dimensionality reduction were performed using Seurat methods, and visualization was realized through t‐distributed Stochastic Neighbor Embedding (T‐SNE). Based on differentially expressed genes (DEGs) in each cluster, stepwise cell clustering was performed to construct an ST map. As was mentioned in a recent study, we similarly used RCTD software to deconvolute each tissue‐covering spot,^{[ 25 ]} and further reconstructed the layered and segmented structure of sinus anatomy. The predicted localization of the 12 immune cell clusters confirmed their enrichment in specific regions (e.g., superior sagittal sinus [SSS], transverse sinus [TS], and hotspot). We mapped the ratios of each cell cluster within each spot as representative pie charts (scatter pies) both on the SSS and TS. Interestingly, distinct changes occurred in infiltrated immune cells into mLVs after injury. Compared to normal mLVs, infiltrated monocytes significantly increased and became the main group among all cell clusters 24 h after SAH. Although infiltrated macrophages were the most at 72 h with the majority subtype of M2, there were significantly more monocytes and M1 type macrophages gathering at the hotspot site (Figure  2A).
如上所述，scRNA-seq 数据丢失了每个细胞的空间特征。因此，我们随后结合 ST 技术以提高测序数据的完整性，并揭示所有细胞的空间分布。经过苏木精-伊红（H&E）染色和明场成像后，对 mLVs 切片进行观察以识别不同的组织学特征。使用 Seurat 方法进行标准质量控制和降维，并通过 t 分布随机邻域嵌入（t-SNE）实现可视化。基于每个簇中的差异表达基因（DEGs），逐步进行细胞聚类以构建 ST 图谱。如近期研究所述，我们同样采用 RCTD 软件对每个组织覆盖点进行反卷积， ^{[ 25 ]} 并进一步重建了窦解剖的分层和分段结构。12 个免疫细胞簇的预测定位证实了它们在特定区域（如上矢状窦[SSS]、横窦[TS]及热点区）的富集。我们将每个点内各细胞簇的比例以代表性饼图（散点饼图）形式映射在 SSS 和 TS 上。 有趣的是，损伤后浸润到 mLVs 中的免疫细胞发生了明显变化。与正常 mLVs 相比，SAH 后 24 小时，浸润的单核细胞显著增加，并成为所有细胞簇中的主要群体。尽管 72 小时时浸润的巨噬细胞最多，且以 M2 亚型为主，但在热点区域聚集的单核细胞和 M1 型巨噬细胞数量明显更多（图 2A ）。

To further explore the alterations in the meningeal immune microenvironment, we attempted to map the scRNA‐seq data of each cell to ST to accurately reveal the infiltration of immune cells into mLVs after SAH. Immune cell infiltration was dominated by monocytes at 24 h and by macrophages at 72 h after SAH. Neutrophil infiltration peaked at 24 h and returned to normal level (sham) by 72 h. T cell infiltration gradually increased at different time courses after injury while B cell infiltration was relatively stable. NKs and pDC infiltration also continuously increased after injury (Figure 2B). These results reflected the cellular changes of mLVs in different stages as SAH progressed.
为深入探究脑膜免疫微环境的改变，我们尝试将单细胞 RNA 测序数据映射至空间转录组，以精确揭示蛛网膜下腔出血后免疫细胞向脑膜淋巴管的浸润情况。免疫细胞浸润以单核细胞在 24 小时占主导，而 72 小时后则以巨噬细胞为主。中性粒细胞浸润在 24 小时达到峰值，至 72 小时恢复至正常（假手术组）水平。T 细胞浸润在损伤后不同时间点逐渐增加，而 B 细胞浸润则相对稳定。自然杀伤细胞和浆细胞样树突状细胞的浸润在损伤后也持续增加（图 2B ）。这些结果反映了蛛网膜下腔出血进展过程中脑膜淋巴管在不同阶段的细胞变化。

2.3. Cell‐to‐Cell Communication Analysis Revealed THBS1‐CD47 Ligand‐Receptor Pair Activation Associated with mLVs Impairment
2.3. 细胞间通讯分析揭示 THBS1-CD47 配体-受体对激活与脑膜淋巴管损伤相关

Given that immune inflammation had long been suggested as a key pathological mechanism involved in SAH pathogenesis,^{[ 26 ]} and our results revealed the differences of infiltrated immune cells between the sham and SAH group, we wondered if the dynamic remodeling of immune microenvironment played an important role in mLVs impairment. Visualization of mLECs distribution prediction showed that mLECs damage was severest at 24 h after injury (Figure  3A). Therefore, we first investigated cell‐to‐cell communication among all cell types to visualize the overall situation. Through manifold learning and quantitative contrasts, CellChat identified differentially over‐expressed ligands and receptors between each cell cluster. In total, 4239 significant L‐R pairs were detected, which were further categorized into 2258 signaling pathways. Significant interactions were identified on the basis of a statistical test that randomly permuted the group labels of cells and then recalculated interaction probability.
鉴于免疫炎症长期以来被认为是参与 SAH 发病机制的关键病理机制， ^{[ 26 ]} 且我们的研究结果揭示了假手术组与 SAH 组之间浸润免疫细胞的差异，我们推测免疫微环境的动态重塑是否在 mLVs 损伤中起重要作用。mLECs 分布预测的可视化显示，损伤后 24 小时 mLECs 的损伤最为严重（图 3A ）。因此，我们首先研究了所有细胞类型之间的细胞间通讯，以可视化整体情况。通过流形学习和定量对比，CellChat 识别了每个细胞簇之间差异过表达的配体和受体。总共检测到 4239 个显著的 L-R 对，这些被进一步分类为 2258 条信号通路。基于对细胞群标签进行随机置换后重新计算相互作用概率的统计检验，确定了显著的相互作用。

Next, we set out to find the relationship between mLECs and other cell types. Several signals that contributed most to the outgoing or incoming signaling patterns of mLECs were identified (Figure 3B). According to the findings above, mLVs injury was most severe at 24 h after injury and monocytes were the dominant infiltrated cells at this time course, indicating that monocytes and mLECs interaction might play crucial roles in mLVs dysfunction. Therefore, by focusing on the communication between mLECs and monocytes, we found that THBS signaling pathways were especially actively involved in all signaling patterns. We further explored the potential L‐R pairs based on THBS signaling working on mLECs and calculated the contribution of these L‐R pairs. Among them, the L‐R pair “THBS1‐CD47” exhibited the strongest interaction in cell‐to‐cell communication (Figure 3C and Figure S2A,B, Supporting Information). To further assess the spatial distribution of the THBS1‐CD47 L‐R pair, we used stLearn,^{[ 27 ]} a technology especially used to analyze spatial communication, which could combine the spatial information of cells with all the predicted communication to discover the regional specificity of L‐R pairs. L‐R co‐expression enrichment among neighbor spots and cell‐type density enrichment were integrated to find hotspots within a tissue, where interactions between cell types were most likely to occur. The result showed that the THBS1‐CD47 L‐R pair significantly activated at 24 h after SAH (Figure 3D).
接下来，我们着手探究 mLECs 与其他细胞类型之间的关系。识别出对 mLECs 传出或传入信号模式贡献最大的若干信号（图 3B ）。基于上述发现，损伤后 24 小时 mLVs 损伤最为严重，此时单核细胞是主要的浸润细胞群，表明单核细胞与 mLECs 的相互作用可能在 mLVs 功能障碍中起关键作用。因此，通过聚焦 mLECs 与单核细胞间的通讯，我们发现 THBS 信号通路在所有信号模式中尤为活跃。我们进一步挖掘了作用于 mLECs 的潜在 L-R 配对，并计算了这些配对的贡献度。其中，"THBS1-CD47" L-R 对在细胞间通讯中展现出最强的相互作用（图 3C 和图 S2A,B ，支持信息）。 为进一步评估 THBS1-CD47 L-R 对的空间分布，我们采用了 stLearn 技术（ ^{[ 27 ]} ），这是一种专门用于分析空间通讯的技术，能够将细胞的空间信息与所有预测的通讯相结合，以发现 L-R 对的区域特异性。通过整合邻近点之间的 L-R 共表达富集和细胞类型密度富集，我们识别出组织内最可能发生细胞类型间互作的热点区域。结果显示，THBS1-CD47 L-R 对在 SAH 后 24 小时显著激活（图 3D ）。

2.4. THBS1 Impaired Meningeal Lymphatic Drainage and Exacerbated Neurological Dysfunction
2.4. THBS1 损害脑膜淋巴引流并加剧神经功能障碍

After discovering that the THBS1‐CD47 L‐R pair was potentially associated with mLVs injury, we then would like to assess the impact of THBS1 after SAH. We checked endogenous THBS1 alteration after SAH by using our previous mass spectrometry proteomics data from ProteomeXchange Consortium (PXD030593). Differentially expressed proteins in human CSF of SAH patients measured by mass spectrometry indicated significant up‐regulation of THBS1 protein (we used CSF samples from normal pressure hydrocephalus (NPH) patients as the control group, Figure 3E). Given that the human THBS family had three subtypes, including THBS1, THBS2, and THBS4,^{[ 28 ]} we used an ELISA kit to again test the expression of all THBS subtypes in CSF samples of SAH patients to verify if THBS1 was the key THBS member. As we expected, although THBS2 and THBS4 levels also elevated, expression of THBS1 increased most after SAH compared to the NPH group (Figure 3F and Table S1, Supporting Information). Also, higher THBS1 level was strongly connected with poorer clinical outcome in SAH patients (Figure 6F).
在发现 THBS1-CD47 配体-受体对可能与脑膜淋巴管（mLVs）损伤相关后，我们进一步评估了 SAH 后 THBS1 的影响。通过分析先前提交至 ProteomeXchange 联盟（PXD030593）的质谱蛋白质组学数据，我们检测了 SAH 后内源性 THBS1 的变化。质谱测定的人 SAH 患者脑脊液差异表达蛋白显示 THBS1 蛋白显著上调（以正常压力脑积水（NPH）患者脑脊液作为对照组，图 3E ）。鉴于人类 THBS 家族包含 THBS1、THBS2 和 THBS4 三种亚型 ^{[ 28 ]} ，我们采用 ELISA 试剂盒复测 SAH 患者脑脊液中所有 THBS 亚型的表达以验证 THBS1 是否为关键成员。如预期所示，虽然 THBS2 和 THBS4 水平亦有升高，但相较于 NPH 组，SAH 后 THBS1 的表达增幅最为显著（图 3F 及附表 S1 ，支持信息）。此外，较高的 THBS1 水平与 SAH 患者较差的临床预后密切相关（图 6F ）。

We then performed in vivo experiments to explore the role of THBS1 in mLVs injury after SAH. First, low‐dosage mouse recombinant THBS1 protein was delivered into mice cisterna magna to observe mLVs function changes. Compared to negative control (NC) mice, recombinant THBS1 (rTHBS1) protein treatment caused less beads aggregation in mLVs and dCLNs (Figure 3G,I and Figure S3A, Supporting Information) and decreased mLECs number of mLVs (Figure 3H). Second, we constructed THBS1‐overexpressing mice through the delivery of AAV‐THBS1 into mice cisterna magna, along with cultivating THBS1‐KO mice for further research. Compared to AAV‐control, mLVs injury was even worse with AAV‐THBS1 treatment after SAH (Figure  4A–C and Figure S3B, Supporting Information). Conversely, THBS1‐KO partly reversed mLVs damage caused by SAH (Figure 4H–J and Figure S3C, Supporting Information). Also, neurological function tests presented similar results. Overexpression of THBS1 aggravated neurological deficits caused by SAH, while THBS1 knockout improved neurofunction (Figure 4D–G,K–N). What is more, consistent with the results of THBS1 knockout, mLVs and neurofunction impairment was reversed as well after blockage of THBS1 with THBS1‐antibody (Figure  5A–G and Figure S3D, Supporting Information). Collectively, all these results proved that up‐regulated THBS1 caused mLVs structure and drainage function damage after SAH, thereby exacerbating neurobehavioral dysfunction.
随后，我们进行了体内实验以探究 THBS1 在 SAH 后 mLVs 损伤中的作用。首先，将低剂量小鼠重组 THBS1 蛋白注入小鼠小脑延髓池，观察 mLVs 功能变化。与阴性对照（NC）小鼠相比，重组 THBS1（rTHBS1）蛋白处理导致 mLVs 和 dCLNs 中珠子聚集减少（图 3G,I 和图 S3A ，支持信息），并降低了 mLVs 中 mLECs 的数量（图 3H ）。其次，我们通过向小鼠小脑延髓池递送 AAV-THBS1 构建了 THBS1 过表达小鼠，同时培养 THBS1-KO 小鼠用于进一步研究。与 AAV 对照相比，SAH 后 AAV-THBS1 处理使 mLVs 损伤更为严重（图 4A–C 和图 S3B ，支持信息）。相反，THBS1-KO 部分逆转了 SAH 引起的 mLVs 损伤（图 4H–J 和图 S3C ，支持信息）。此外，神经功能测试也呈现相似结果。THBS1 的过表达加剧了 SAH 导致的神经功能缺损，而 THBS1 敲除则改善了神经功能（图 4D–G,K–N ）。 此外，与 THBS1 敲除结果一致，使用 THBS1 抗体阻断 THBS1 后，脑膜淋巴管（mLVs）和神经功能损伤同样得到逆转（支持信息中的图 5A–G 和图 S3D ）。综上所述，这些结果均证实，SAH 后上调的 THBS1 导致了脑膜淋巴管结构和引流功能损伤，从而加剧了神经行为功能障碍。

2.5. Disturbing THBS1‐CD47 Interaction Promoted mLVs Restoration via Inhibiting STAT3/BCL2‐Mediated Apoptosis in mLECs
2.5. 干扰 THBS1-CD47 相互作用通过抑制 STAT3/BCL2 介导的 mLECs 凋亡促进 mLVs 修复

Previous studies reported that THBS1 could inhibit endothelial cell proliferation and chemotaxis by engaging its receptors CD47.^{[ 29 ]} Similarly, our results predicted that the THBS1‐CD47 L‐R pair played an important role in mLVs impairment and proved the injurious effect of THBS1 after SAH. We therefore moved on to discover CD47 function. Blockage of CD47 with antibody attenuated meningeal lymphatic dysfunction compared to igG treatment at 24 h post‐SAH (Figure 5A–C and Figure S3D, Supporting Information). Accordantly, the neurological function also improved significantly (Figure 5D–G).
先前研究报道，THBS1 可通过与其受体 CD47 结合抑制内皮细胞增殖和趋化作用 ^{[ 29 ]} 。类似地，我们的结果预测 THBS1-CD47 配体-受体对在 SAH 后脑膜淋巴管损伤中起重要作用，并证实了 THBS1 的损害效应。因此，我们进一步探究了 CD47 的功能。与 IgG 处理相比，SAH 后 24 小时用抗体阻断 CD47 可减轻脑膜淋巴功能障碍（图 5A–C 和图 S3D ，支持信息）。相应地，神经功能也显著改善（图 5D–G ）。

Next, to further investigate the underlying mechanism of THBS1‐CD47 interaction‐induced injury, GSEA analysis was performed.^{[ 30 ]} The result indicated that the apoptosis pathway in mLECs was significantly activated 24 h after injury (Figure 5H). Tunel staining indicated that apoptosis was significantly activated after SAH. Overexpression of THBS1 further promoted the apoptosis of mLVs while blockage of THBS1 showed decreased Lyve1⁺Tunel⁺ cells (Figure S5A, Supporting Information). Co‐cultivation of human monocytes and LECs in vitro confirmed that the content of THBS1 in the supernatant of the co‐culture system was significantly increased under the stimulation of hemoglobin (150 µm), indicating that monocytes are one of the sources of THBS1 secretion after SAH (Figure S5B, Supporting Information). Flow cytometric analysis revealed that the apoptosis of LECs in the co‐culture system pretreated with THBS1 neutralizing antibody is significantly reduced compared to the control group (Figure S5C,D, Supporting Information). However, the precise mechanism of mLVs injury remains unknown.
接下来，为进一步探究 THBS1-CD47 相互作用诱导损伤的潜在机制，我们进行了 GSEA 分析。 ^{[ 30 ]} 结果显示，损伤后 24 小时 mLECs 中的凋亡通路被显著激活（图 5H ）。Tunel 染色表明 SAH 后凋亡显著激活。THBS1 的过表达进一步促进了 mLVs 的凋亡，而阻断 THBS1 则显示 Lyve1 ⁺ Tunel ⁺ 细胞减少（图 S5A ，支持信息）。体外共培养人单核细胞与 LECs 证实，在血红蛋白（150 µm）刺激下，共培养系统上清液中 THBS1 含量显著增加，表明单核细胞是 SAH 后 THBS1 分泌的来源之一（图 S5B ，支持信息）。流式细胞术分析显示，与对照组相比，经 THBS1 中和抗体预处理的共培养系统中 LECs 的凋亡显著减少（图 S5C,D ，支持信息）。然而，mLVs 损伤的确切机制仍不清楚。

Studies had demonstrated that THBS1‐CD47 interaction induced endothelial cell apoptosis through disturbing VEGF/VEGFR2 signaling, which was essential for endothelial development.^{[ 29 ]} Also, VEGFR2 inhibition was found to promote apoptosis via suppressing STAT3/Bcl‐2 signaling in osteosarcoma.^{[ 31 ]} Therefore, we hypothesized if THBS1‐CD47 interaction could medicate mLECs apoptosis via inhibiting STAT3/Bcl‐2 signaling. First, co‐immunoprecipitation (COIP) confirmed the interaction between THBS1 and CD47 in LECs (Figure S4A, Supporting Information). Second, along with increased expression of Bax, expression of pSTAT3 and Bcl‐2 were significantly decreased after rTHBS1treatment in primary mLECs in a dose‐dependent manner (Figure 5I), indicating that  rTHBS1 treatment promoted mLECs apoptosis by repressing the activation of the STAT3/Bcl‐2 signaling pathway. Finally, in SAH mice mLVs samples, the level of pSTAT3 and Bcl‐2 peaked at 24 h and then decreased at 72 h after SAH, which was consistent with GSEA results (Figure 5J). THBS1 knockout or blockage of THBS1/CD47 could all significantly increase pSTAT3 and Bcl‐2 expression (Figure 5K,M), whereas THBS1‐overexpression presented the opposite effect (Figure 5L). Taken together, all these findings verified that the THBS1‐CD47 L‐R pair promoted mLECs apoptosis via STAT3/Bcl‐2 signaling, and inhibition of this interaction might help alleviate mLVs injury.
研究表明，THBS1-CD47 相互作用通过干扰 VEGF/VEGFR2 信号传导诱导内皮细胞凋亡，而该信号对内皮发育至关重要。 ^{[ 29 ]} 此外，发现抑制 VEGFR2 可通过抑制 STAT3/Bcl-2 信号通路促进骨肉瘤细胞凋亡。 ^{[ 31 ]} 因此，我们推测 THBS1-CD47 相互作用是否可能通过抑制 STAT3/Bcl-2 信号通路介导 mLECs 凋亡。首先，共免疫沉淀（COIP）证实了 LECs 中 THBS1 与 CD47 的相互作用（图 S4A ，支持信息）。其次，在原代 mLECs 中，随着 Bax 表达的增加，rTHBS1 处理后 pSTAT3 和 Bcl-2 的表达呈剂量依赖性显著下降（图 5I ），表明 rTHBS1 处理通过抑制 STAT3/Bcl-2 信号通路的激活促进了 mLECs 凋亡。最后，在 SAH 小鼠 mLVs 样本中，pSTAT3 和 Bcl-2 的水平在 SAH 后 24 小时达到峰值，随后在 72 小时下降，这与 GSEA 结果一致（图 5J ）。 THBS1 基因敲除或阻断 THBS1/CD47 相互作用均能显著增加 pSTAT3 和 Bcl-2 的表达（图 5K,M ），而过表达 THBS1 则呈现相反效果（图 5L ）。综上所述，这些发现证实 THBS1-CD47 配体-受体对通过 STAT3/Bcl-2 信号通路促进 mLECs 凋亡，抑制该相互作用可能有助于减轻 mLVs 损伤。

2.6. Pseudotime Analysis Revealed mLECs Alteration Trace after SAH
2.6. 伪时序分析揭示 SAH 后 mLECs 的变化轨迹

Additionally, we performed pseudotime analysis of all mLECs in different samples to reveal mLECs change trajectory during the EBI stage after SAH (Figure  6A). After labeling mLECs with real grouping information, we found that the damage pattern of mLECs 24 h after SAH was consistent with the findings above. However, unexpectedly, mLECs at 72 h seemed to possess completely different cell states or characteristics compared to the sham group and 24 h group (Figure 6B). Considering that mLVs impairment was partly improved at 72 h after SAH (Figure 1I), we wondered if self‐repairing mechanisms of mLECs or other cell‐to‐cell interactions functioned at this time course, which was worth the further efforts and exploration. DEG analysis revealed the most over‐expressed genes after SAH (Figure 6C,D). Along the trajectory, the expression of S100α6 and Cldn5 gradually increased (Figure 6E). Previous research reported that S100A6 impaired lymphatic endothelial cells tight junction and increased the transendothelial migration of neutrophils in breast cancer,^{[ 32 ]} indicating that S100α6 might be a potential mLECs damage biomarker after SAH.
此外，我们对不同样本中的所有 mLECs 进行了伪时间分析，以揭示 SAH 后 EBI 阶段 mLECs 的变化轨迹（图 6A ）。在用真实分组信息标记 mLECs 后，我们发现 SAH 后 24 小时 mLECs 的损伤模式与上述发现一致。然而，出乎意料的是，与假手术组和 24 小时组相比，72 小时的 mLECs 似乎具有完全不同的细胞状态或特征（图 6B ）。考虑到 SAH 后 72 小时 mLVs 损伤部分改善（图 1I ），我们推测此时可能有 mLECs 的自我修复机制或其他细胞间相互作用在发挥作用，这值得进一步研究和探索。差异表达基因（DEG）分析揭示了 SAH 后表达最上调的基因（图 6C,D ）。沿轨迹方向，S100α6 和 Cldn5 的表达逐渐增加（图 6E ）。先前研究报道，S100A6 会损害淋巴管内皮细胞的紧密连接并增加中性粒细胞在乳腺癌中的跨内皮迁移， ^{[ 32 ]} 这表明 S100α6 可能是 SAH 后 mLECs 损伤的潜在生物标志物。

We then performed immunofluorescent staining of mLVs to identify S100α6 changes after SAH. As expected, there was a marked increase in meningeal S100α6 expression and more Lyve1⁺S100α6⁺ positive cells 24 h after injury while Lyve1⁺ cells decreased compared to the uninjured group (Figure 6I). Conversely, the expression of S100α6 was also significantly reduced when THBS1 was knocked out (Figure S6, Supporting Information). Also, through the ELISA assay, we confirmed increased expression of S100A6 in CSF samples of SAH patients compared to the control. Prognostic analysis suggested higher expression of S100A6 was related to a worse prognosis (Figure 6G and Table S1, Supporting Information). Meanwhile, the expression of S100A6 and THBS1 had a linear relationship (Figure 6H), which indicated that both of them might be appropriate predictors after SAH. In summary, our data illustrated that S100α6 might be a novel biomarker of meningeal lymphatic injury, which further supplemented the gaps in the field of mLVs annotation.
随后，我们对脑膜淋巴管（mLVs）进行了免疫荧光染色，以观察 SAH 后 S100α6 的变化。如预期所示，损伤 24 小时后，脑膜中 S100α6 表达显著增加，且 Lyve1 ⁺ 与 S100α6 ⁺ 双阳性细胞增多，而未损伤组的 Lyve1 ⁺ 细胞则有所减少（图 6I ）。相反，当 THBS1 基因敲除时，S100α6 的表达也显著降低（图 S6 ，支持信息）。此外，通过 ELISA 检测，我们证实 SAH 患者脑脊液样本中 S100A6 的表达较对照组升高。预后分析显示，S100A6 的高表达与不良预后相关（图 6G 及表 S1 ，支持信息）。同时，S100A6 与 THBS1 的表达呈线性关系（图 6H ），表明二者可能均为 SAH 后合适的预测指标。总之，我们的数据表明 S100α6 可能是脑膜淋巴损伤的新型生物标志物，进一步填补了 mLVs 注释领域的空白。

Clinical Sample Collection and Patient Information
临床样本采集与患者信息

Patients with aneurysmal SAH were diagnosed by computed tomography (CT) and digital subtraction angiograph (DSA) performed within 24 h after admissions. The inclusion and exclusion criteria were consistent with the previous study.^{[ 51 ]} All CSF samples of SAH patients were collected before coiling or clipping through a lumbar puncture or lumbar drainage (LD) or external ventricular drainage (EVD) within 48 h after SAH. All CSF sample collection and patient information were consistent with the previous report.^{[ 3a ]}
动脉瘤性蛛网膜下腔出血（SAH）患者通过入院 24 小时内进行的计算机断层扫描（CT）和数字减影血管造影（DSA）确诊。纳入与排除标准与先前研究一致。 ^{[ 51 ]} 所有 SAH 患者的脑脊液（CSF）样本均在出血后 48 小时内通过腰椎穿刺、腰椎引流（LD）或脑室外引流（EVD）在弹簧圈栓塞或夹闭手术前采集。所有 CSF 样本采集及患者信息均与先前报告一致。 ^{[ 3a ]}

Mass Spectrometry of Human CSF
人类脑脊液的质谱分析

The previous mass spectrometry proteomics data from ProteomeXchange Consortium (PXD030593) was used for downstream analysis. Used the hiplot tool to present relevant gene expression in the form of a volcano plot.^{[ 52 ]}
先前来自 ProteomeXchange 联盟（PXD030593）的质谱蛋白质组学数据被用于下游分析。使用 hiplot 工具以火山图形式展示相关基因表达。 ^{[ 52 ]}

ELISA Assay  ELISA 检测

The concentrations of human CSF THBS1 (EK0899, Boster Biotech, Wuhan), THBS2 (EK0642, Boster Biotech, Wuhan), THBS4 (CSB‐EL023490HU, Cusabio Biotech, Wuhan), and S100A6 (CSB‐EL020634RA, Cusabio Biotech, Wuhan) were determined using ELISA kits. Human CSF samples (100 µL) were used for the analysis. All procedures were performed according to the manufacturer's instructions.
采用 ELISA 试剂盒测定人脑脊液中 THBS1（EK0899，Boster Biotech，武汉）、THBS2（EK0642，Boster Biotech，武汉）、THBS4（CSB‐EL023490HU，Cusabio Biotech，武汉）和 S100A6（CSB‐EL020634RA，Cusabio Biotech，武汉）的浓度。分析使用 100 微升人脑脊液样本，所有操作均按照制造商说明书进行。

Animals  动物

Specific pathogen‐free, C57BL/6 male mice (6–8 weeks old) were purchased from Shanghai SLAC Laboratory Animal Co. Ltd. For AAV injection, C57BL/6 mice were injected with AAV‐Ctrl, AAV‐THBS1 at 6 weeks of age, when the weight was ≈20 g. SAH modeling and behavioral tests were performed 4 weeks after the AAV injection. C57BL/6 mice were injected into with recombinant mice THBS1 protein or vehicle (PBS) via cisterna magna 1 day in advance before the sacrifice. THBS1^−/‐ mice were established on a C57BL/6 background and purchased from Gempharmatech Co., Ltd (#KO‐T014586, NCBI:21 825, MGI:98 737). Mice were housed in the animal facility with controlled habituation and temperature, on 12‐h light versus dark cycles, and fed with regular rodents' chow and sterilized tap water ad libitum. Mice were allowed to accommodate for 2 weeks before experimental procedures.
无特定病原体（SPF 级）的 C57BL/6 雄性小鼠（6-8 周龄）购自上海斯莱克实验动物有限责任公司。对于腺相关病毒（AAV）注射，C57BL/6 小鼠在 6 周龄、体重约 20 克时分别注射 AAV-Ctrl 或 AAV-THBS1。AAV 注射 4 周后进行蛛网膜下腔出血（SAH）模型构建及行为学测试。在处死前 1 天，C57BL/6 小鼠通过小脑延髓池注射重组小鼠 THBS1 蛋白或对照溶剂（PBS）。THBS1 ^−/‐ 小鼠基于 C57BL/6 背景构建，购自集萃药康生物科技有限公司（品系编号 KO-T014586，NCBI:21825，MGI:98737）。小鼠饲养于环境温湿度可控的动物设施中，采用 12 小时明暗循环，自由摄食常规啮齿类饲料及灭菌自来水。所有小鼠在实验操作前均适应环境 2 周。

Induction of SAH  SAH 诱导

The prechiasmatic SAH mice model was created as previously described.^{[ 53 ]} In brief, after induction of anesthesia with 1% pentobarbital, the head was fixed in a stereotactic apparatus (RWD Life Science Co., Ltd). With a midline incision, the skin overlying the anterior skull was opened. A burr hole was drilled into the skull 4.5 mm anterior to the bregma with a caudal angel of 40° using a 0.9‐mm drill. Blood (60 µL) was withdrawn from a C57BL/6 WT blood donor and injected over a 10‐s period with a 26‐gauge needle advanced through the burr hole at a 40° angle until reaching the base of the skull. For antibody injection, a mixture of 10 µg mL⁻¹ CD47 antibody (Invitrogen, 16‐0479‐85) or THBS1 antibody (Genetex, GTX21823) or IgG (Invitrogen, 14‐4714‐85) with 60 µL autologous blood was injected into the prechiasmatic cisternae. The needle was left in place for 5 min before retraction to avoid backflow. Mice that underwent the same procedure, without blood and antibody/IgG injections, served as the sham group.
视交叉前蛛网膜下腔出血（SAH）小鼠模型的建立方法如前所述 ^{[ 53 ]} 。简而言之，使用 1%戊巴比妥麻醉后，将小鼠头部固定于立体定位仪（RWD Life Science Co., Ltd）。沿中线切开前颅骨表面皮肤，用 0.9 毫米钻头在距前囟 4.5 毫米处以 40 度尾倾角钻孔。从 C57BL/6 野生型供血鼠抽取 60 微升血液，通过 26 号针头以 40 度角经骨孔推进至颅底，在 10 秒内完成注射。抗体注射时，将 10 微克/毫升的 CD47 抗体（Invitrogen, 16-0479-85）、THBS1 抗体（Genetex, GTX21823）或 IgG（Invitrogen, 14-4714-85）与 60 微升自体血液混合后注入视交叉前池。针头留置 5 分钟后缓慢退出以防止反流。假手术组小鼠接受相同操作但不注射血液及抗体/IgG。

Intra‐Cisterna Magna Injections
小脑延髓池内注射

Mice were anesthetized by intraperitoneal (i.p.) injection with 1% pentobarbital. The skin of the neck was shaved and cleaned with iodine and 70% ethanol, and an ophthalmic solution (erythromycin ointment) was placed on the eyes to prevent drying. Stereotaxic frames were used to secure the mouse's head, and a midline incision was made in the skin. The muscle layers were retracted and the cisterna magna exposed. Using a Hamilton syringe (coupled to a 33‐gauge needle), the volume of the desired solution was injected into the CSF‐filled cisterna magna compartment. For THBS1 or Control AAV delivery experiments, 2 µL of artificial CSF containing 10¹³ genome copies per mL of AAV9‐THBS1, or AAV‐Con (AAV9, adeno‐associated virus serotype 9; purchased from PackGene Biotech, Guangzhou), were injected into the cisterna magna CSF at a rate of 2 µL min⁻¹. For recombinant protein experiments, recombinant mice THBS1 protein (150 µg kg⁻¹, R&D Systems, 7859‐TH‐050, USA) or vehicle (PBS) were injected into cisterna magna CSF at a rate of 2 µL min⁻¹. For the beads experiments, 2 µL of FluoSpheres carboxylate 0.5 µm‐beads 505/515 (Invitrogen, F8813) were injected at a rate of 2 µL min⁻¹. The needle was inserted into the cisterna magna through the retracted muscle in order to prevent backflow upon needle removal. The neck skin was then sutured and allowed to recover on a heating pad until fully awake.
小鼠通过腹腔注射 1%戊巴比妥钠麻醉。颈部皮肤剃毛并用碘伏和 70%乙醇消毒，眼部涂抹眼科用药（红霉素软膏）以防干燥。使用立体定位仪固定小鼠头部，沿中线切开皮肤。分离肌肉层暴露小脑延髓池。采用汉密尔顿注射器（连接 33 号针头）将指定体积的溶液注入充满脑脊液的小脑延髓池腔。在 THBS1 或对照 AAV 递送实验中，以 2 微升/分钟的速率向小脑延髓池脑脊液内注射 2 微升含 10 ¹³ 基因组拷贝/毫升的 AAV9-THBS1 或 AAV-Con（AAV9，腺相关病毒血清型 9，购自广州派真生物）的人工脑脊液。重组蛋白实验中，以 2 微升/分钟的速率注射重组小鼠 THBS1 蛋白（150 微克/千克 ⁻¹ ，R&D Systems 货号 7859-TH-050，美国）或载体（PBS）。微球实验中，以 2 微升/分钟的速率注射 2 微升 0.5 微米羧基荧光微球 505/515（Invitrogen 货号 F8813）。 将针头通过回缩的肌肉插入小脑延髓池，以防止拔针时回流。随后缝合颈部皮肤，并在加热垫上恢复直至完全清醒。

Neurological Scores and Behavioral Testing
神经学评分和行为测试

Neurological scores and all behavioral assessments of each group were carried out at similar and relatively fixed time during daylight hours in a blinded fashion.
各组神经学评分及所有行为评估均在白天相似且相对固定的时间内以盲法进行。

Neurological Scores  神经学评分

The Modified Garcia (ranging from 0 to 18) was conducted to evaluate short‐term neurological function. The scoring system was comprised of six sub‐tests including response capacity, alertness, coordination, motor skills, complex movements, and coordination.^{[ 54 ]} A blinded investigator performed the neurological examination, which was scored 24 and 72 h after SAH. A higher score represented a better neurological function.
采用改良 Garcia 评分（范围 0～18 分）评估短期神经功能。该评分系统包括反应能力、警觉性、协调性、运动技能、复杂动作和协调性六个子测试。 ^{[ 54 ]} 由盲法研究者进行神经学检查，分别在 SAH 后 24 小时和 72 小时评分。评分越高，神经功能越好。

Pole Test  杆测试

A pole test was performed 24 and 72 h postoperatively. Animals are placed on top of a 50‐ to 55‐cm vertical pole with a diameter of 8 to 10 mm and trained to turn around and descend the pole. The scoring started when the animal initiated the turning movement. The time to make a complete 180° turn (i.e., T _turn) and latency to reach the ground (i.e., T _total) were recorded. If the animal cannot turn but instead descends with a lateral body position, then T _total was usually attributed to T _turn. When an animal makes a turn, descends halfway, and falls, the recorded time stops when the animal reaches the floor. The maximum time to perform the test was set to 60 s. Mice were trained three times daily for 3 days before the start of testing.
术后 24 小时和 72 小时进行杆测试。将动物置于直径 8 至 10 毫米、高 50 至 55 厘米的垂直杆顶端，训练其转身并沿杆下行。评分自动物开始转身动作时启动。记录完成 180 度转身所需时间（即 T _turn ）及到达地面的延迟时间（即 T _total ）。若动物无法转身而以侧身姿势下行，则通常将 T _total 归为 T _turn 。当动物完成转身、下行中途坠落时，记录时间止于其触地瞬间。测试最长时限设为 60 秒。正式测试前，小鼠每日训练三次，持续三天。

Wire Hanging Test  钢丝悬挂试验

Wire Hanging is a simple test that evaluates grip strength, balance, and endurance. Mice were trained to suspend their body by holding on to a single wire stretched between two posts 50 to 60 cm above the ground. To prevent the animal from using all four paws, the hind limbs were gently covered with adhesive tape. Between the two posts was a pillow to avoid injury when the mice fell. “Latency to fall” was the primary end point used to assess motor performance. The maximum time to perform the test was set to 60 s. Mice were trained three times daily for 3 days before the start of testing.
钢丝悬挂是一项评估抓握力、平衡性和耐力的简单测试。训练小鼠通过抓住一根离地 50 至 60 厘米、横跨两柱之间的单根钢丝来悬吊身体。为防止动物使用所有四爪，后肢被轻柔地用胶带覆盖。两柱之间放置了一个垫子，以防小鼠跌落时受伤。“跌落潜伏期”是评估运动表现的主要终点指标。测试的最长时间设定为 60 秒。在测试开始前，小鼠每天接受三次训练，持续三天。

Tissue Collection  组织采集

Mice were euthanized with isoflurane and dCLNs were dissected under the microscope. Subsequently drop‐fixed in 4% paraformaldehyde (PFA) for 2 h at 4 °C and then the CUBIC clearance protocol was performed as described below in the dCLN clearance methods sections.^{[ 55 ]} For meningeal whole‐mount collection, mice were transcardially perfused with 20 mL cold 1× PBS, then skin and muscle were stripped from the outer skull, and the skullcap was removed with surgical scissors and fixed in 2% PFA for 12 h at 4 °C. Then the meninges (dura mater and arachnoid mater) were carefully dissected from the skullcaps with Dumont #5 forceps (Fine Science Tools). Meningeal whole‐mounts were then moved to PBS and 0.05% azide at 4 °C until further use.
小鼠通过异氟烷安乐死后，在显微镜下解剖出 dCLNs。随后用 4%多聚甲醛（PFA）在 4°C 下滴固定 2 小时，然后按照下文 dCLN 透明化方法部分所述的 CUBIC 透明化流程进行处理。 ^{[ 55 ]} 对于脑膜全片采集，小鼠经心脏灌注 20 mL 冰冷的 1×PBS，然后剥离头皮和肌肉，用手术剪取下颅盖，置于 2% PFA 中 4°C 固定 12 小时。随后使用 Dumont #5 镊子（精细科学工具）从颅盖上小心分离脑膜（硬脑膜和蛛网膜）。脑膜全片随后转移至含 0.05%叠氮化钠的 PBS 中，4°C 保存待用。

dCLN Clearance  dCLN 透明化

Consistent with the previous report,^{[ 41 ]} dCLN clearance was performed following the published CUBIC protocol with modifications.^{[ 55 ]} In brief, nodes were incubated in 50% reagent 1 (prepared 1:1 with dH2O) for 1 day at 37 °C, shaking, with DAPI (1:1000). Nodes were then transferred to reagent 1 for 1 day at 37 °C, shaking, with DAPI (1:1000). Nodes were washed two times in PBS + 0.01% sodium azide for 2 h and overnight with DAPI (1:1000) at 37 °C. Then nodes were incubated with 50% reagent 2 (prepared 1:1 with dH2O) for 1 day at 37 °C with DAPI (1:1000). Finally, nodes were incubated with reagent 2 for 1 day at 37 °C. Nodes were placed in eight well chambers (155 411, Thermo Fisher) with mineral oil and imaged with confocal microscopy.
与先前报告一致， ^{[ 41 ]} dCLN 清除按照已发表的 CUBIC 方案及修改步骤进行。 ^{[ 55 ]} 简言之，淋巴结在 50%试剂 1（以 dH2O 按 1:1 配制）中于 37°C 震荡孵育 1 天，加入 DAPI（1:1000）。随后将淋巴结转移至试剂 1 中，于 37°C 震荡孵育 1 天，加入 DAPI（1:1000）。淋巴结在 PBS + 0.01%叠氮化钠中洗涤两次，每次 2 小时，并于 37°C 与 DAPI（1:1000）过夜孵育。接着，淋巴结在 50%试剂 2（以 dH2O 按 1:1 配制）中于 37°C 与 DAPI（1:1000）孵育 1 天。最后，淋巴结在试剂 2 中于 37°C 孵育 1 天。将淋巴结置于含矿物油的八孔室（155 411，赛默飞世尔）中，使用共聚焦显微镜成像。

Immunofluorescence, Imaging, and Quantification
免疫荧光、成像与定量分析

For immunofluorescence staining, meningeal whole‐mounts in PBS and 0.05% azide were blocked with either 2% donkey serum or 2% goat serum, 1% bovine serum albumin, 0.1% triton, 0.05% tween‐20, and 0.05% sodium azide in PBS for 1.5 h at room temperature. This blocking step was followed by incubation with appropriate dilutions of primary antibodies: Rat anti‐Lyve1(Santa Cruze, sc‐65647, 1:200), Goat anti‐CD31 (R&D Systems, AF3628, 1:100), and Rabbit anti‐S100α6 (Abcam, ab181975, 1:200) in the same solution used for blocking overnight at 4 °C. Meningeal whole‐mounts or brain tissue sections were then washed three times for 10 min at room temperature in PBS and 0.05% tween‐20, followed by incubation with the appropriate donkey Alexa Fluor 405 or 488 or 594 anti‐rat, ‐goat,‐rabbit (Thermo Fisher Scientific, 1:1000) for 2 h at RT in the same solution used for blocking. For Tunel staining, apoptotic cells were detected according to the manufacturer's instructions using the apoptotic kit (Beyotime Biotech, C1090, China). The whole mounts were then washed three times for 10 min at RT in PBS. The tissue was then transferred to PBS and mounted with ProLong Gold antifade reagent (Invitrogen, P36934) on glass slides with coverslips. Slide preparations were stored at 4 °C and imaged using a Lecia DMI8 confocal microscope and LAS AF software (Leica Microsystems) within 1 week of staining. Quantitative analysis of the acquired images was performed using Fiji software. For lymph nodes, the percent volume of microbead coverage in cleared dCLN was assessed by creating a 3D reconstruction of the node and then calculating the volume covered by beads divided by the total volume of the node using Fiji. The right and the left dCLN percent volume were averaged together for each mouse. For assessment of mLV coverage and complexity, images of the meningeal whole‐mounts were acquired using a confocal microscope and Fiji was used for quantifications. The entire meningeal whole‐mount was traced and used for quantification. The percent area coverage of Lyve1 was used to determine the coverage of the lymphatic vessels. When applicable, the same images were used to assess the percentage of field coverage by Lyve1⁻CD31⁺ vessels. All meningeal whole mounts used for the quantification of lymphatic morphology were imaged with identical confocal settings and Fiji parameters.
进行免疫荧光染色时，将 PBS 和 0.05%叠氮化物中的脑膜全片用 2%驴血清或 2%山羊血清、1%牛血清白蛋白、0.1% Triton、0.05% Tween-20 及 0.05%叠氮化钠的 PBS 溶液在室温下封闭 1.5 小时。封闭后，用适当稀释的一抗于 4°C 过夜孵育：大鼠抗 Lyve1（Santa Cruze, sc-65647, 1:200）、山羊抗 CD31（R&D Systems, AF3628, 1:100）及兔抗 S100α6（Abcam, ab181975, 1:200），孵育液与封闭液相同。随后，脑膜全片或脑组织切片在 PBS 和 0.05% Tween-20 中室温洗涤三次，每次 10 分钟，再以相同封闭液加入驴源 Alexa Fluor 405、488 或 594 标记的抗大鼠、山羊或兔二抗（Thermo Fisher Scientific, 1:1000）室温孵育 2 小时。对于 Tunel 染色，凋亡细胞检测按照凋亡试剂盒（碧云天生物技术, C1090, 中国）说明书操作。最后，全片在 PBS 中室温洗涤三次，每次 10 分钟。 将组织转移至 PBS 中，并用 ProLong Gold 抗褪色试剂（Invitrogen, P36934 ）固定在带有盖玻片的载玻片上。制备好的玻片在 4°C 下保存，并在染色后 1 周内使用 Lecia DMI8 共聚焦显微镜和 LAS AF 软件（Leica Microsystems）进行成像。获取的图像使用 Fiji 软件进行定量分析。对于淋巴结，通过创建淋巴结的三维重建，然后使用 Fiji 计算微珠覆盖的体积除以淋巴结的总体积，来评估清除后的 dCLN 中微珠覆盖的体积百分比。每只小鼠的左右 dCLN 体积百分比取平均值。为了评估脑膜全片中淋巴管（mLV）的覆盖率和复杂性，使用共聚焦显微镜获取脑膜全片的图像，并用 Fiji 进行量化。整个脑膜全片被追踪并用于量化。Lyve1 的面积覆盖率用于确定淋巴管的覆盖率。在适用的情况下，使用相同的图像评估 Lyve1 ⁻ CD31 ⁺ 血管的视野覆盖率百分比。 所有用于淋巴管形态学定量的脑膜全标本均采用相同的共聚焦设置和 Fiji 参数进行成像。

Flow Cytometric  流式细胞术

Mice were euthanized by Euthasol injection (i.p.) and transcardially perfused with 20 mL cold 1× PBS for 5 min. Heads were removed and skulls were quickly stripped. Mandibles were removed, as well as all skull material rostral to maxillae. Surgical scissors were used to remove the top of the skull, cutting clockwise, beginning and ending inferior to the right post‐tympanic hook. Meninges (dura mater, arachnoid, and pia mater) were carefully removed from the interior aspect of the skulls and surfaces of the brain with Dumont #5 forceps (Fine Science Tools). Meninges were gently pressed through 70‐µm nylon mesh cell strainers with a sterile plastic plunger (BD Biosciences) to yield a single‐cell suspension. For lymphatic endothelial cell isolation, meninges (along with diaphragm and ear skin) were digested for 30 min in 0.60 U mL⁻¹ of Liberase TM (Roche) and 60 U mL⁻¹ of DNase 1. Cells were then centrifuged at 300 RCF at 4 °C for 10 min, the supernatant was removed and cells were resuspended in ice‐cold FACS buffer (pH 7,4; 0.1 m PBS; 1 mm EDTA; 1% BSA). Cells were stained for extracellular marker with antibodies to anti‐CD45‐PE‐Cyanine5 (1:200, clone 30‐F11, eBiosciences), anti‐CD31‐Alexa Fluor 647 (1:200, clone 390, BD Biosciences), and anti‐LYVE1‐ Alexa Fluor 488 (1:200, clone ALY7, eBioscience). Experiments were performed on meninges from n = 2 mice in the sham group and n = 1 mice in the SAH group. For cell apoptosis analysis, the human lymphatic endothelial were plated in a 6‐well plate with 10⁶ cells. After 24 h, the cells were treated with normal medium or conditioned medium (CM) with or without THBS1 antibody (5 µg mL⁻¹, Genetex, GTX21823) for another 48 h. Cells were collected using trypsin–EDTA and resuspended in annexin buffer and incubated for 15 min with Annexin V (BD Biosciences) or PI (eBioscience) according to the manufacturer's protocol. The apoptotic cells were analyzed using a FACScan flow cytometer (BD Biosciences). FACS analyses were reproduced by three independent experiments. Data processing was done with Excel and statistical analysis was performed using GraphPad Prism.
小鼠通过腹腔注射安乐死药物 Euthasol 实施安乐死，随后用 20 mL 冰冷的 1× PBS 进行 5 分钟的心脏灌注。摘除头部并迅速剥离颅骨。去除下颌骨及上颌骨前所有颅骨组织。使用手术剪沿顺时针方向剪开颅顶，起止于右耳后钩下方。用 Dumont #5 镊子（精细科学工具）小心移除颅骨内侧面及脑表面的脑膜（硬脑膜、蛛网膜和软脑膜）。将脑膜通过 70 微米尼龙细胞筛网，用无菌塑料推杆（BD Biosciences）轻柔压过，获得单细胞悬液。为分离淋巴管内皮细胞，将脑膜（连同膈肌和耳部皮肤）置于含 0.60 U/mL Liberase TM（罗氏）和 60 U/mL DNase 1 的溶液中消化 30 分钟。细胞随后在 4°C、300 RCF 条件下离心 10 分钟，弃上清，用冰预冷的 FACS 缓冲液（pH 7.4；0.1 M PBS；1 mM EDTA；1% BSA）重悬细胞。 细胞使用抗 CD45-PE-Cyanine5（1:200，克隆号 30-F11，eBiosciences）、抗 CD31-Alexa Fluor 647（1:200，克隆号 390，BD Biosciences）和抗 LYVE1-Alexa Fluor 488（1:200，克隆号 ALY7，eBioscience）抗体进行细胞外标记染色。实验在假手术组 2 只小鼠和 SAH 组 1 只小鼠的脑膜组织上进行。对于细胞凋亡分析，人淋巴管内皮细胞以 10^2 个细胞接种于 6 孔板中。24 小时后，细胞分别用正常培养基或含或不含 THBS1 抗体（5 µg/mL，Genetex，GTX21823）的条件培养基处理 48 小时。使用胰蛋白酶-EDTA 收集细胞，重悬于 Annexin 缓冲液中，并按制造商说明书与 Annexin V（BD Biosciences）或 PI（eBioscience）孵育 15 分钟。凋亡细胞通过 FACScan 流式细胞仪（BD Biosciences）分析。FACS 分析由三次独立实验重复验证。数据处理使用 Excel 完成，统计分析采用 GraphPad Prism 软件进行。

Primary Meningeal Lymphatic Endothelial Cells Culture
原代脑膜淋巴管内皮细胞培养

For mLEC culture, meninges were dissected from P0‐P1 C57/BL6 mice. To obtain a suspension of meningeal LECs, skull caps were quickly collected, and meninges (dura mater and arachnoid) were dissected using Dumont no. 5 and no. 7 fine forceps in a complete medium composed of Dulbecco's modified Eagle's medium (DMEM) (Gibco) with 2% FBS, 1% l‐glutamine (Gibco), 1% penicillin/streptomycin (Gibco), 1% sodium pyruvate (Gibco), 1% non‐essential amino‐acids (Gibco), and 1.5% Hepes buffer (Gibco). Meninges were then incubated with 1 mL DMEM with 0.60 U mL⁻¹ of Liberase TM (Roche) and 60 U mL⁻¹ of DNase 1 for 30 min at 37 °C. Cell suspensions were then pooled into a single tube after filtration through a 70‐µm nylon mesh cell strainer. Cells were then centrifuged at 300 RCF at 4 °C for 10 min, the supernatant was removed and cells were resuspended with 1 mL ice‐cold magnetic separation buffer. Added 5 µL rat anti‐lyve1(Santa Cruze, sc‐65647, 1:200) antibody to the cell suspension. Vortexed briefly to mix and incubate at 4 °C for 15 min. Washed cells by adding 10 mL of magnetic separation buffer to the tube and centrifuged at 600 RCF for 5 min at 4 °C. Carefully aspirated the supernatant and resuspend the cells in 1 mL of magnetic separation buffer. Added 100 µL of anti‐rat IgG beads (Miltenyi Biotec, cat.no 130‐048‐502) to the cells, vortexed briefly to mix, and incubated for 15 min at 4 °C. Added 10 mL of magnetic separation buffer to the cells and centrifuged at 4 °C, 600 RCF for 5 min. According to the manufacturer's instructions, MACS sorting was carried out by using the magnetic beads matched with the anti‐rat IgG beads. Lyve1‐positive cells were collected and resuspended in mLEC culture medium, inoculated in a culture dish, and the medium was changed every 2 days.
;; 对于 mLEC 培养，从 P0-P1 C57/BL6 小鼠中解剖出脑膜。为获得脑膜淋巴内皮细胞（LECs）悬液，迅速收集颅盖，并使用 Dumont 5 号和 7 号精细镊子在由 Dulbecco 改良 Eagle 培养基（DMEM）（Gibco）配制的完全培养基中解剖脑膜（硬脑膜和蛛网膜），该培养基含 2%胎牛血清（FBS）、1% L-谷氨酰胺（Gibco）、1%青霉素/链霉素（Gibco）、1%丙酮酸钠（Gibco）、1%非必需氨基酸（Gibco）及 1.5% Hepes 缓冲液（Gibco）。随后，将脑膜与 1 mL 含 0.60 U/mL Liberase TM（Roche）和 60 U/mL DNase 1 的 DMEM 在 37°C 下孵育 30 分钟。细胞悬液经 70 微米尼龙网细胞过滤器过滤后合并至单管中。细胞在 4°C 下以 300 RCF 离心 10 分钟，弃去上清，用 1 mL 冰磁分离缓冲液重悬细胞。向细胞悬液中加入 5 µL 大鼠抗 lyve1 抗体（Santa Cruze, sc-65647, 1:200）。短暂涡旋混匀后于 4°C 孵育 15 分钟。加入 10 mL 磁分离缓冲液洗涤细胞，4°C 下以 600 RCF 离心 5 分钟。 小心吸取上清液，将细胞重悬于 1 mL 磁珠分离缓冲液中。加入 100 µL 抗大鼠 IgG 磁珠（Miltenyi Biotec，货号 130-048-502）至细胞悬液，短暂涡旋混匀后于 4°C 孵育 15 分钟。加入 10 mL 磁珠分离缓冲液，4°C、600 RCF 离心 5 分钟。按照制造商说明，使用与抗大鼠 IgG 磁珠匹配的磁珠进行 MACS 分选。收集 Lyve1 阳性细胞，重悬于 mLEC 培养基中接种至培养皿，每 2 天更换一次培养基。

THP1 Conditioned Medium Preparation
THP1 条件培养基制备

Human monocyte THP1 cells were cultured in RPMI 1640 medium containing 10% FBS (VivaCell, Shanghai, China) and 1% penicillin‐streptomycin. For CM collection, THP1 cells were plated in a 6‐well plate with FreeStyle expression medium (12 338 026; Gibco, Thermo Fisher Scientific, USA) for 24 h. The CM were collected 24 h later, centrifuged at 15 000 × g for 5 min, and filtered through a 0.22‐mm filter. Human lymphatic endothelial cells were exposed to fresh medium mixed with CM at a ratio of 1:1 (v/v) for the following process.
人单核细胞 THP1 培养于含 10%胎牛血清（VivaCell，中国上海）及 1%青霉素-链霉素的 RPMI 1640 培养基中。收集条件培养基时，THP1 细胞接种于 6 孔板内 FreeStyle 表达培养基（12 338 026；Gibco，Thermo Fisher Scientific，美国）中培养 24 小时。24 小时后收集上清，15 000×g 离心 5 分钟，经 0.22 µm 滤膜过滤。人淋巴管内皮细胞后续处理采用新鲜培养基与条件培养基按 1:1（体积比）混合的培养基。

Co‐Immunoprecipitation  共免疫沉淀

10⁶ human lymphatic endothelial cells were lysed using 400 µL of lysis buffer (Tris‐HCL HCl 50 mm, pPH 7.4, Nacl NaCl 150 mm, sodium deoxycholate 0.25%, NP‐40 1%, EDTA 1 mm, PMSF 1 mm, Aprotinin 1 mg mL⁻¹, leupeptin 1 mg mL⁻¹, pepstatin 1 mg mL⁻¹) on ice for 30 min and centrifuged at 16 113.6 rcf for 15 min. The supernatant was incubated with CD47 (Invitrogen, MA5‐11895) or IgG (Abclone, AC005) antibodies overnight and followed by protein A/G plus agarose bead (Santa Cruz Biotech., sc‐2003) for another 4 h. After washing three times with ice‐cold PBS, parvalbumin (1 ug mL⁻¹) was added for 2 h. Agarose beads were washed three times with PBS and followed by SDS‐PAGE and immunoblot analysis with indicated antibodies.
10 ⁶ 人淋巴管内皮细胞使用 400 µL 裂解缓冲液（Tris-HCL HCl 50 mM，pH 7.4，NaCl 150 mM，脱氧胆酸钠 0.25%，NP-40 1%，EDTA 1 mM，PMSF 1 mM，抑肽酶 1 mg/mL ⁻¹ ，亮抑酶肽 1 mg/mL ⁻¹ ，胃酶抑素 1 mg/mL ⁻¹ ）在冰上裂解 30 分钟，随后以 16,113.6 rcf 离心 15 分钟。上清液与 CD47（Invitrogen，MA5-11895）或 IgG（Abclone，AC005）抗体孵育过夜，再加入蛋白 A/G 琼脂糖珠（Santa Cruz Biotech.，sc-2003）继续孵育 4 小时。用冰预冷的 PBS 洗涤三次后，加入小清蛋白（1 µg/mL ⁻¹ ）孵育 2 小时。琼脂糖珠经 PBS 洗涤三次后，进行 SDS-PAGE 及指定抗体的免疫印迹分析。

Western Blot Analysis  蛋白质印迹分析

Western blot was used to analyze the lymphoid apoptosis pathway‐associated protein changes in meninges at different time points after SAH/sham surgery. Briefly, after euthanasia and perfusion with cold PBS, the mouse meninges were harvested, and total protein was extracted with RIPA lysis buffer and 1% PMSF. The protein was separated by 10% SDS‐PAGE at 120 V for 90 min, and the separated proteins were transferred to PVDF membranes. Then, the PVDF membranes were blocked and incubated with primary antibodies (rabbit anti‐STAT3 (1:1000, Abclone, cat no. A1192, China); rabbit anti‐ Phospho‐STAT3‐Y705 (1:1000, Abclone, cat no.AP0070, China); rabbit anti‐Bcl2 (1:1000, Abcam, cat no. ab182858); rabbit anti‐Bax, (1:1000, Abcam, cat no. ab182734); and rabbit anti‐β actin pAb (1:1000, diagbio, cat no. db10001, China)) on a shaker overnight at 4 °C. After being sufficiently washed, the membranes were incubated with horseradish peroxidase‐conjugated secondary antibodies (1:5000, beyotime, cat no. A0208) for 1 h at room temperature. Experiments were performed on meninges from n = 3 mice per group. A ChemiDoc imaging system (Bio‐Rad) was used to visualize the membranes, and ImageJ software was used to calculate the grey values of the bands.
采用 Western blot 分析 SAH/假手术后不同时间点脑膜淋巴凋亡通路相关蛋白的变化。简言之，小鼠经安乐死及冷 PBS 灌注后，收集脑膜组织，使用 RIPA 裂解液和 1% PMSF 提取总蛋白。蛋白样品经 10% SDS-PAGE（120 V，90 分钟）分离后，转印至 PVDF 膜。随后，PVDF 膜经封闭处理，与一抗（兔抗 STAT3，1:1000，Abclone，货号 A1192，中国；兔抗磷酸化 STAT3-Y705，1:1000，Abclone，货号 AP0070，中国；兔抗 Bcl2，1:1000，Abcam，货号 ab182858；兔抗 Bax，1:1000，Abcam，货号 ab182734；兔抗β actin 多克隆抗体，1:1000，diagbio，货号 db10001，中国）在 4°C 摇床中孵育过夜。充分洗涤后，膜与辣根过氧化物酶标记的二抗（1:5000，beyotime，货号 A0208）室温孵育 1 小时。每组实验使用 n=3 只小鼠的脑膜样本。 使用 ChemiDoc 成像系统（Bio-Rad）观察膜，并用 ImageJ 软件计算条带的灰度值。

Single‐Cell RNA Sequencing
单细胞 RNA 测序

Freshly prepared meningeal cell suspensions were performed immediately according to the manufacturer's protocol of 10× Chromium 3' v3 kit (10× Genomics, Pleasanton, CA). Libraries were sequenced on the Illumina NovaSeq 6000 platform at Novogene, Beijing, China. On average, each sample generated about 709 m reads (SD = 121 m).
新鲜制备的脑膜细胞悬液立即按照 10× Chromium 3' v3 试剂盒（10× Genomics，普莱森顿，加利福尼亚州）制造商的协议进行处理。文库在中国北京诺禾致源的 Illumina NovaSeq 6000 平台上进行测序。平均每个样本产生约 7.09 亿条读长（标准差=1.21 亿）。

Read Alignment, Quality Control, and Clustering Analysis
读段比对、质量控制与聚类分析

The raw gene expression matrix was generated via CellRanger. Doublets were removed through Scrublet.^{[ 56 ]} Then filtered counts matrix was converted to Seurat object using the Seurat R package (version 4.0.1). The cells with gene numbers >7000, or <200, or >20% mitochondrial counts were filtered. Each sample data was normalized via LogNormalize method and the top 2000 highly variable genes were selected for subsequent dimensional reduction. The batch effect across samples was removed via the integration method implemented in the Seurat package. PCA (n = 30) and the UMAP were used to reduce the dimensionality and visualize the cell distribution. Cluster‐specific genes were identified using the Wilcox rank‐sum test implemented in the FindAllMarkers function (fold change ≥ 2 and adjusted p ≤0.05). Cell clusters were annotated to known cell lineages using well‐recognized marker genes.
原始基因表达矩阵通过 CellRanger 生成。使用 Scrublet 去除双联体细胞。 ^{[ 56 ]} 随后，过滤后的计数矩阵利用 Seurat R 包（版本 4.0.1）转换为 Seurat 对象。筛选掉基因数>7000 或<200，或线粒体计数占比>20%的细胞。各样本数据通过 LogNormalize 方法进行标准化，并选取前 2000 个高变基因用于后续降维分析。样本间的批次效应通过 Seurat 包中的整合方法予以消除。采用 PCA（主成分分析，n=30）和 UMAP 进行降维并可视化细胞分布。通过 FindAllMarkers 函数中的 Wilcox 秩和检验（倍数变化≥2 且校正 p 值≤0.05）鉴定簇特异性基因。利用公认的标志基因将细胞簇注释至已知细胞谱系。

Cell–Cell Communication Analysis
细胞间通讯分析

To analyze cell–cell interactions in the microenvironment, CellChat was used to infer ligand–receptor pairs between mLECs and other cell types.^{[ 57 ]} In brief, normalized expression values of each sample were fed into CellChat, and identifyOverExpressedGenes, identifyOverExpressedInteractions, as well as projectData functions were used to conduct preprocessing. Core functions “computeCommunProb,” “computeCommunProbPathway,” and “aggregateNet” were applied to infer cell‐cell communication networks. Significant interactions were defined with p ≤ 0.05. Finally, “netAnalysis_signalingRole_heatmap” function was used to determine the senders and receivers in the network.
为分析微环境中的细胞间相互作用，采用 CellChat 推断 mLECs 与其他细胞类型间的配体-受体对。 ^{[ 57 ]} 简言之，将各样本的标准化表达值输入 CellChat，并运用 identifyOverExpressedGenes、identifyOverExpressedInteractions 及 projectData 功能进行预处理。核心功能“computeCommunProb”“computeCommunProbPathway”和“aggregateNet”用于推断细胞间通讯网络。显著性互作定义为 p≤0.05。最后通过“netAnalysis_signalingRole_heatmap”功能确定网络中的信号发送与接收细胞。

Spatial Transcriptomics Sequencing and Data Processing
空间转录组测序与数据处理

Used OCT‐embedded frozen meninge for ST detection. The capture of gene expression information for ST slides was performed by the Visium Spatial platform of 10× Genomics through the use of spatially barcoded mRNA‐binding oligonucleotides in the default protocol. Raw sequencing reads of spatial transcriptomics were quality checked and mapped to the pre‐built mm10 reference genome by Space Ranger v1.1.
使用 OCT 包埋的冷冻脑膜组织进行空间转录组检测。Visium Spatial 平台（10× Genomics）通过默认流程中的空间条形码 mRNA 结合寡核苷酸捕获 ST 玻片的基因表达信息。空间转录组原始测序读数经质量检查后，由 Space Ranger v1.1 比对至预构建的 mm10 参考基因组。

Filtering, Normalization, and Deconvolution of Visium Data
Visium 数据的过滤、标准化与解卷积

The gene‐spot matrices generated after ST data processing were analyzed with the Seurat package (version 4.0.1) in R. Spots were filtered for gene counts <500, mitochondrial counts >25%, or red cell counts >20%. Normalization across spots was performed with the SCTransform function. Dimensionality reduction and clustering were performed with PCA with the top 30 PCs. Spatial feature expression plots were generated with the SpatialFeaturePlot function in Seurat (version 4.0.1). RCTD was used to deconvolute each spot to predict the underlying composition of cell types.^{[ 25 ]}
ST 数据处理后生成的基因点矩阵使用 R 中的 Seurat 包（版本 4.0.1）进行分析。过滤掉基因计数<500、线粒体计数>25%或红细胞计数>20%的点位。通过 SCTransform 函数进行跨点位标准化。采用前 30 个主成分的 PCA 进行降维和聚类分析。空间特征表达图由 Seurat（版本 4.0.1）中的 SpatialFeaturePlot 函数生成。利用 RCTD 解卷积每个点位以预测潜在的细胞类型组成。 ^{[ 25 ]}

Cell Trajectory Analysis
细胞轨迹分析

The Monocle2 package was used for pseudotime trajectory construction.^{[ 58 ]} Briefly, the differentialGeneTest function was used to identify DEGs across cell types. The top 1000 genes based on q value were selected as ordering genes. DDRTree method was applied to reduce dimension. orderCells function was used to order cells with the sham group as root.
使用 Monocle2 包构建拟时序轨迹。 ^{[ 58 ]} 简要而言，通过 differentialGeneTest 函数鉴定跨细胞类型的差异表达基因（DEGs）。基于 q 值选取前 1000 个基因作为排序基因。应用 DDRTree 方法进行降维。以假手术组为根，使用 orderCells 函数对细胞进行排序。

Gene Set Enrichment Analysis
基因集富集分析

The marker genes for each group of the mLEC cluster were identified by FindMarkers from Seurat with a threshold of min.pct = 0.1 and logFC.threshold = 0 and then sorted according to avg_log2FC. The GSEA software (v4.1.0) was used to identify the Kyoto Encyclopedia of Genes and Genomes pathways.
通过 Seurat 中的 FindMarkers 功能，以 min.pct = 0.1 和 logFC.threshold = 0 为阈值，识别了 mLEC 簇各组的标志基因，并依据 avg_log2FC 进行排序。随后使用 GSEA 软件（v4.1.0）鉴定了京都基因与基因组百科全书通路。

Statistical Analysis  统计分析

Statistical analysis was performed using SPSS analysis tools (IBM Corp.) or Prism 9 software program (GraphPad Software). All data are presented as the mean ± SEM. To assess the statistical significance between the two groups, a two‐tailed unpaired Student's t‐test (for parametric analysis) or Mann–Whitney U test (for nonparametric analysis) was performed. One‐way analysis of variance (ANOVA) followed by Tukey's or Dunnett's multiple comparisons test were used to detect the differences in the results between groups. The correlation between the expression profiles of CSF biomarkers and the prognosis of SAH patients was analyzed using Spearman's rank test. p values less than 0.05 were considered to indicate statistical significance. No sample outliers were excluded. Individual in vitro experiments were performed at least three times with similar results.
统计分析采用 SPSS 分析工具（IBM Corp.）或 Prism 9 软件（GraphPad Software）进行。所有数据均以均值±标准误表示。为评估两组间的统计学显著性，采用双尾非配对 Student's t 检验（参数分析）或 Mann–Whitney U 检验（非参数分析）。组间结果差异通过单因素方差分析（ANOVA）结合 Tukey 或 Dunnett 多重比较检验进行检测。CSF 生物标志物表达谱与 SAH 患者预后的相关性采用 Spearman 秩检验分析。p 值小于 0.05 视为具有统计学意义。未排除任何样本异常值。体外实验均独立重复至少三次，结果具有一致性。

Ethics Statement  伦理声明

All studies concerning human information and tissues were approved by the Ethics Committee of the Second Affiliated Hospital of Zhejiang University School of Medicine (No.2020‐892) and is registered as an observational study in ClinicalTrail.gov (No. NCT04938414). All of the patients signed informed consent. All animal procedures were approved by the second affiliated hospital of Zhejiang university – Animal Ethics Committee (Approval no. AIRB‐2022‐0698) and were performed according to the Guiding Principles for the Care and Use of Laboratory Animals Approved by Animal Regulations of National Science and Technology Committee of China.
所有涉及人类信息和组织的研究均获得浙江大学医学院附属第二医院伦理委员会批准（编号：2020‐892），并在 ClinicalTrail.gov 注册为观察性研究（编号： NCT04938414 ）。所有患者均签署知情同意书。所有动物实验程序经浙江大学附属第二医院动物伦理委员会批准（批准号：AIRB‐2022‐0698），并遵循中国国家科学技术委员会颁布的《实验动物管理条例》中关于实验动物护理与使用的指导原则执行。