房颤（Atrial fibrillation，AF）是一种常见的持续性心律失常，全球约有1%的人口受其影响。AF常导致心脏快速而不规则的心律反应，不仅降低患者的生活质量和增加医疗负担，还显著增加并发症的发病率和死亡率。尽管过去几十年的研究揭示了AF发生和发展的一些机制，例如离子通道重构和纤维化等，但是AF的治疗仍然面临重大挑战。这主要是因为我们对AF心房重构的潜在机制尚不十分清楚。因此，寻找针对心房重构的新治疗策略对于AF的预防和治疗可能提供新的思路和方法。

心房的结构重构对于房颤的发生与维持中起着至关重要的作用，尤其是纤维化过程。心房纤维化时细胞外基质蛋白如胶原蛋白和纤连蛋白的过度积聚导致纤维化组织形成，占据了正常心房肌细胞的空间，使心房的收缩与舒张功能减退，干扰心肌细胞之间的电传导耦联，并形成电传导的解剖屏障。这种纤维化解剖屏障的形成进一步减慢了心电传导与扩布速度，形成无规则的阻滞区域，增加了折返环路和异位电信号活动，从而成为房颤的触发器和维持因素。尽管心房结构重构在房颤的发生和发展中扮演着重要角色，但由于心房组织的复杂性，目前对于其结构重构机制的认识仍然有限，这一限制阻碍了人们对房颤机制的深入探索。

单细胞测序的发展为我们提供了独特的机遇，使我们能够从单个细胞的分辨率揭示心房组织的复杂性。通过单细胞测序，我们对心房的细胞组成进行深入探索，并确定其中的关键细胞类型。针对关键细胞类型，我们进一步研究了房颤心房结构重构过程中的细胞特征以及细胞之间的相互作用。这些研究发现将使我们对房颤的机制有更加深入的了解，并为房颤治疗提供更多的新思路。

本研究主要内容包含如下三个部分：

第一部分 成人心房单细胞图谱及房颤心房结构重构过程中的关键细胞

背景：心房颤动是临床上最常见的持续性心律失常，心房结构重构是促进房颤发生并维持的关键过程，然而该过程的作用机制尚不明确。构建正常至房颤过程中的心房单细胞图谱，寻找关键细胞亚群，对揭示房颤心房结构重构的深层机制具有重要意义。

目的：构建成人心房单细胞图谱，揭示房颤心房结构重构过程中的关键细胞。

方法和结果：本研究提取公共数据库中所有涉及心房样本的数据集进行单细胞整合分析，揭示了心房主要由8种细胞构成，包括心肌细胞（CMs，31.0%），成纤维细胞（FBs，26.1%），周细胞（PCs, 13.0%），内皮细胞（ECs，11.3%），髓系免疫细胞 （6.2%），平滑肌细胞（4.0%），淋系免疫细胞（3.2%），以及脂肪细胞（2.3%）。ECs在正常及潜在重构心房中起主要调控作用，其中与FBs的相互作用最为显著。为进一步分析ECs在房颤心房中的作用，我们收取3例行外科房颤消融患者的左心耳以及3例窦性心律供心的对应部位进行单细胞测序。拟时序分析结果表明，房颤心房FBs显著激活，发生典型的结构重构即纤维化，该过程受ECs的主要调控，TGFB1为ECs作用于FBs的最重要细胞因子。通过构建FBs示踪鼠Col1a2-Cre;R26RGFP进一步验证了该相互作用的空间可及性。另外，在房颤心房结构重构过程中，ECs显著上调间质标志物，出现间质激活的可塑性。

结论：成人心房由众多细胞组成，其中ECs具有重要调控作用，是促进房颤心房结构重构的关键细胞亚群。单细胞测序提示ECs通过TGFB通路激活FBs参与房颤心房结构重构，该过程中ECs表现为显著的间质激活可塑性。

第二部分 谱系示踪验证房颤心房结构重构过程中内皮细胞具有间质激活的可塑性

背景：房颤单细胞图谱揭示了内皮细胞（ECs）具有间质激活的可塑性，是ECs参与心房结构重构的重要机制。然而，关于内皮间质转化是否存在及其对纤维化的贡献意义始终存在争议。为明确房颤发生过程中ECs的间质激活特性以及其对心房结构重构的贡献，需联合房颤动物模型，谱系示踪技术和单细胞测序，揭示ECs在房颤心房结构重构过程中的最终转归。

目的：寻找最适房颤结构重构小鼠模型，在体示踪ECs，明确房颤心房结构重构过程中ECs的可塑性及其对纤维化的贡献程度。

方法和结果：本研究构建了四种小鼠模型包括血管紧张素 II 诱导的心房纤维化模型（AngII）、主动脉缩窄模型（TAC）、心肌梗死模型（MI）和导丝损伤瓣膜模型（Wire injury）。相应模型分别于术后4周，8周，8周和12周进行心房大小检测，纤维化评估和电生理表型检测。心房大小由超声评估左房前后径，纤维化通过马松染色及免疫荧光判断FBs激活特征，电生理表型通过心内导管检测房颤诱发率及心房有效不应期。结果显示，除外Wire injury 模型，AngII，MI和TAC模型心房显著增大，分别为假手术组的1.3，1.2和1.4倍。纤维化显著增加，分别为假手术组的2.1，1.9和7.1倍。四种模型中仅AngII和TAC模型房颤诱发率显著提高，分别为假手术组的2.03和2.09倍。所有模型中仅TAC术后心内膜下出现FBs激活，且TAC模型的纤维化分布与人心房组织最为接近。选用TAC作为后续ECs示踪鼠（Cdh5-Cre;R26RRFP）的房颤建模术式，分选心房干预前后RPF细胞进行单细胞测序。拟时序分析及基因评分揭示了心房重构过程中ECs存在不同的分化轨迹，所有轨迹均呈现出间质激活特征。然而，免疫荧光显示蛋白层面ECs转化为间质细胞数量较少，存在转录与蛋白层面ECs间质化现象的差异。

结论：TAC模型具有显著的心房纤维化及较高的房颤诱发率，是研究房颤心房重构尤其是结构重构的最适小鼠模型。在TAC基础上，谱系示踪结合单细胞测序明确了房颤心房结构重构过程中ECs间质激活的可塑性。然而这种可塑性存在转录与蛋白层面的差异，有待后续进一步探索研究。

第三部分 具有可塑性的内皮细胞通过TGFB1通路调控成纤维细胞功能参与房颤心房结构重构

背景：谱系示踪联合单细胞测序明确了房颤心房结构重构过程中ECs间质激活的可塑性。然而这种特征出现了转录与蛋白层面的不一致性。转录层面间质激活是ECs分化轨迹中最显著的特征，而蛋白层面，完全转化为间质细胞的ECs却极为稀少，提示ECs参与心房结构重构的过程并非通过转化为间质细胞来实现。ECs间质激活的意义有待进一步探索。

目的：明确ECs间质激活的意义以及ECs参与房颤心房结构重构的主要方式。

方法和结果：本研究首先通过免疫荧光及高通量细胞表型检测明确ECs间质激活后较少转化为间质细胞。ECs示踪鼠Cdh5-Cre;R26RRFP行主动脉缩窄术（TAC）后共定位RFP与α-SMA、COL1A1，PDGFRα，DDR2，FAP和LTBP2。除外α-SMA，其余间质标志物均无ECs共定位现象。α-SMA与ECs共定位比例亦为稀少，占比仅0.1%。其余模型，包括血管紧张素 II 诱导的心房纤维化模型（AngII）、心肌梗死模型（MI）和导丝损伤瓣膜模型（Wire injury）中均未见RFP与α-SMA、COL1A1，PDGFRα，DDR2，FAP和LTBP2共定位现象。体外采用高通量细胞表型检测实时采集ECs的细胞形态、转化及迁移过程。结果显示ECs间质激活过程中细胞面积、长度和宽度显著增大，圆度和宽长比显著降低。ECs转化为间质细胞的比例约为0.3%。ECs迁移速率显著下降。除外ECs形态变化，其余参数均不符合传统内皮间质转化现象。本研究进一步分选人心房原代ECs及FBs进行细胞互作实验。结果显示间质化ECs通过异常释放TGFB1促进了FBs的激活，增殖，胶原沉积与释放。TGFB1特异性受体抑制剂SB431542可以抑制该过程。腺相关病毒RGDLRVS-AAV9-Cdh5-Tgfb1在体ECs过表达Tgfb1后小鼠心房显著扩大，纤维化程度显著增加，房颤持续时间延长。

结论：房颤心房结构重构过程中，ECs间质激活的意义不在于本身转化为间质细胞构成纤维化的一部分，而在于促进FBs的激活、增殖、胶原沉积与释放等纤维化表型。该过程主要通过TGFB1通路实现。

Atrial fibrillation (AF) is a prevalent sustained cardiac arrhythmia that affects approximately 1% of the global population. AF often results in rapid and irregular heart rhythm, leading to decreased quality of life, increased healthcare burden, as well as significantly higher incidence of complications and mortality. Despite the identification of certain mechanisms underlying the occurrence and progression of AF, such as ion channel remodeling and fibrosis, the treatment of AF remains a significant challenge. This is primarily due to our incomplete understanding of the underlying mechanisms of atrial remodeling in AF. Hence, exploring novel therapeutic strategies targeting atrial remodeling may offer new insights and approaches for the prevention and treatment of AF.

Structural remodeling of the atria plays a crucial role in the initiation and maintenance of AF, particularly through the process of fibrosis. Excessive accumulation of extracellular matrix proteins, such as collagen and fibronectin, during atrial fibrosis leads to the formation of fibrotic tissue, which occupies the space of normal atrial myocardial cells. This impairs the contraction and relaxation functions of the atria, disrupts electrical coupling between myocardial cells, and forms anatomical barriers to electrical conduction. The formation of these fibrotic anatomical barriers further slows down the conduction and spread of electrical signals, creating areas of irregular conduction, promoting reentry circuits, and facilitating ectopic electrical activities, thus acting as triggers and perpetuators of AF. Despite the pivotal role of atrial structural remodeling in the initiation and progression of AF, our current understanding of the underlying mechanisms is still limited due to the complexity of atrial tissue, impeding in-depth exploration of AF mechanisms.

The advent of single-cell sequencing provides a unique opportunity to unravel the complexity of atrial tissue at the resolution of individual cells. Through single-cell sequencing, we can delve into the cellular composition of the atria and identify key cell types involved. With a focus on these key cell types, we further investigate the cellular characteristics and intercellular interactions during atrial structural remodeling in AF. These findings will provide us with a deeper understanding of the mechanisms underlying AF and offer novel insights for AF treatment.

This study comprises the following three main components:

Part I Single-cell atlas of the adult atrium and key cells involved in atrial structural remodeling in atrial fibrillation

Background: Atrial fibrillation (AF) is the most common sustained clinical arrhythmia. Atrial structural remodeling is a key process that promotes the occurrence and maintenance of AF. However, the underlying mechanism of this process remains unclear. Therefore, constructing a single-cell atlas of the atrium from normal to AF and identifying key cell subsets would be of great significance in uncovering the mechanisms underlying atrial structural remodeling in AF.

Objective: To construct a single-cell atlas of the adult atrium and identify key cell subsets involved in the process of atrial structural remodeling in AF.

Methods and results: This study aimed to construct a single-cell atlas of adult atrium and identify key cells involved in the process of atrial structural remodeling in AF. All available datasets involving atrial samples from public databases were extracted for single-cell integration analysis, revealing that the atrium is primarily composed of eight cell types, including cardiomyocytes (CMs, 31.0%), fibroblasts (FBs, 26.1%), pericytes (PCs, 13.0%), endothelial cells (ECs, 11.3%), myeloid immune cells (6.2%), smooth muscle cells (4.0%), lymphoid immune cells (3.2%), and adipocytes (2.3%). Among these, ECs play a major regulatory role in normal and potentially remodeling atria, with the interaction with FBs being the most significant. To investigate the role of ECs in AF, we collected left atrial appendages from three patients undergoing surgical ablation of AF and corresponding parts of three donor hearts in sinus rhythm for single-cell sequencing. Pseudotime analysis revealed that atrial FBs were significantly activated in AF, leading to typical structural remodeling, such as fibrosis. This process was primarily regulated by ECs, with TGFB1 being the most critical cytokine acting on FBs. The spatial accessibility of this interaction was further verified by constructing a Col1a2-Cre;R26RGFP tracer mouse for FBs. Additionally, during atrial structural remodeling in AF, ECs significantly upregulated mesenchymal markers, indicating their mesenchymal activation plasticity.

Conclusion: The adult atrium is composed of multiple cell types, with ECs playing a crucial regulatory role and serving as the key cell subpopulation driving the structural remodeling of AF. Single-cell sequencing analysis has revealed that ECs activate FBs via the TGFB pathway to participate in the atrial structural remodeling in AF, and ECs exhibit significant mesenchymal activation and plasticity during this process.

Part II Lineage tracing verifies endothelial plasticity with mesenchymal activation during atrial structural remodeling in atrial fibrillation

Background: The single-cell atlas of atrial fibrillation (AF) has highlighted the plasticity of mesenchymal activation in endothelial cells (ECs), which plays an important role in the atrial structural remodeling. Nevertheless, there is still debate over the existence of endothelial-mesenchymal transition and the significance of its contribution to fibrosis. To gain more insight into the mesenchymal activation of ECs and its contribution to atrial structural remodeling in the context of AF, it is crucial to combine animal models of AF, lineage tracing technology, and single-cell sequencing.

Objective: To identify an appropriate mouse model for AF-related structural remodeling, trace ECs in vivo, and clarify the mesenchymal activation characteristics of ECs and their contribution to fibrosis during atrial structural remodeling in AF.

Methods and results: In this study, we constructed four mouse models, including an angiotensin II-induced atrial fibrosis model (AngII), an aortic coarctation model (TAC), a myocardial infarction model (MI), and a guidewire injury valve model (Wire injury). Corresponding models were evaluated for atrial size, fibrosis, and electrophysiological phenotype at 4-, 8-, 8-, and 12-weeks post-operation. The size of the atrial chamber was assessed by ultrasound, and the anteroposterior diameter of the left atrium was measured. The activation characteristics of fibroblasts were assessed by Masson staining and immunofluorescence for fibrosis, and the induction rate of atrial fibrillation and the effective refractory period of the atrial chamber were measured by intracardiac catheterization for electrophysiological phenotype. The results demonstrated that, except for the Wire injury model, the AngII, MI, and TAC models exhibited significantly enlarged atria, which were 1.3, 1.2, and 1.4 times that of the sham operation group, respectively. Fibrosis was significantly increased, 2.1, 1.9, and 7.1 times that of the sham group, respectively. Among the four models, only the AngII and TAC models significantly increased the induction rate of atrial fibrillation, which were 2.03 and 2.09 times that of the sham group, respectively. In all models, only subendocardial fibroblasts were activated after TAC, and the fibrosis distribution of the TAC model was the closest to that of human atrial tissue. TAC was selected as the subsequent modeling procedure for atrial fibrillation in ECs tracer mice (Cdh5-Cre; R26RRFP), and RFP cells were sorted before and after atrial intervention for single-cell sequencing. Pseudotime analysis and gene scoring revealed distinct differentiation trajectories of ECs during atrial remodeling, all of which exhibited features of mesenchymal activation. However, immunofluorescence showed that the number of ECs transformed into mesenchymal cells at the protein level was small, and there were differences in the mesenchymal activation of ECs at the transcriptional and protein levels.

Conclusion: The TAC model has significant atrial fibrosis and a high induction rate of AF and is the most suitable mouse model for the study of atrial remodeling, especially structural remodeling in AF. On the basis of TAC, lineage tracing combined with single-cell sequencing clarified the plasticity of ECs mesenchymal activation during atrial structural remodeling in AF. However, there are differences in transcription and protein levels in this plasticity, which needs to be further explored and studied in the future.

Part III Mesenchymal-activated endothelial cells participate in atrial structural remodeling in atrial fibrillation by regulating fibroblast function through the TGFB1 pathway

Background: Lineage tracing combined with single-cell sequencing has clarified the plasticity of ECs in mesenchymal activation during atrial structural remodeling in AF. However, this feature appears inconsistent at the transcriptional and protein levels. While mesenchymal activation is the most prominent feature in the differentiation trajectory of ECs at the transcriptional level, ECs that have completely transformed into mesenchymal cells are extremely rare at the protein level. This suggests that ECs do not participate in the process of atrial structural remodeling by transforming into mesenchymal cells. The significance of mesenchymal activation of ECs requires further exploration.

Objective: To elucidate the significance of mesenchymal activation of ECs and the primary mechanism by which ECs contribute to atrial structural remodeling in AF.

Methods and results: In this study, we employed immunofluorescence and high-throughput cell phenotype detection to confirm that mesenchymal activation of ECs seldom leads to their complete transformation into mesenchymal cells. Co-localization of ECs and COL1A1, PDGFRα, DDR2, FAP, and LTBP2 with RFP was absent in ECs tracer mice (Cdh5-Cre; R26RRFP) after TAC, except for α-SMA. Only 0.1% of ECs tracer mice captured α-SMA⁺EC, which suggests that mesenchymal ECs were infrequent. Mesenchymal ECs were not observed in other models, including AngII, MI, and Wire injury. Using high-throughput cell phenotype detection, we dynamically observed the mesenchymal activation process of ECs, which showed significant changes in cell area, length, and width. The roundness and width-to-length ratio decreased significantly, and the proportion of ECs transformed into mesenchymal cells was about 0.3%. The migration rate of ECs also significantly decreased, but the other parameters did not conform to the traditional phenomenon of endothelial-mesenchymal transition. Analysis of cell communication suggested that mesenchymal activation in ECs plays a role in regulating other cells, particularly FBs. To verify this hypothesis, we sorted human atrial primary ECs and FBs for cell interaction experiments, which demonstrated that mesenchymal ECs abnormally release TGFB1 to promote FB activation, proliferation, collagen deposition, and release. The TGFB1-specific receptor inhibitor SB431542 effectively inhibited this process. In vivo, overexpression of Tgfb1 in ECs by the adeno-associated virus RGDLRVS-AAV9-Cdh5-Tgfb1 led to significant atrial enlargement, increased fibrosis, and prolonged duration of AF.

Conclusion: In the process of atrial structural remodeling in AF, the main role of ECs in mesenchymal activation is not to transform themselves into mesenchymal cells to contribute to fibrosis, but rather to promote the activation, proliferation, collagen deposition, and release of FBs and other fibrotic phenotypes. This process is primarily mediated by the TGFB1 pathway.