Hepatocyte-Derived Exosomal microRNAs Orchestrate Vascular Inflammation and Endothelial Function: Insights Into Molecular Mechanisms of Trimethylamine-N-Oxide in Atherosclerosis


 BackgroundTrimethylamine-N-oxide (TMAO) has been proved to be a new proatherogenic compound for promoting vascular inflammation and endothelial dysfunction. Hepatocyte-derived exosomes played an important role in the regulation of vascular inflammation and endothelial function. Since TMAO is produced in the liver, hepatocytes may be the first potential target of TMAO. However, it is not clear whether TMAO can directly stimulate normal hepatocytes to produce exosomes so as to mediate the motivating effects of TMAO on inflammation and endothelial dysfunction. MethodsThe hepatocytes were cultured and treated with TMAO at a physiological concentration for 24 hours (TMAO-Exos). The untreated group served as the control (Control-Exos). The exosomes were isolated from the culture supernatant and then added to the human aortic endothelial cells (HAECs) for 48 hours. The mRNA expressions of inflammatory cytokines and caspase-3 were determined by qPCR and cell apoptosis was evaluated by using the Hoechst 33342 staining solution. The miRNA profile in the exosomes were detected using an RNA-sequencing strategy. The miRNA-mRNA network was predicted, and the biological functions of the target genes were annotated by using bioinformatics methods. ResultsTMAO-Exos were able to promote the expressions of inflammatory cytokines and HAECs apoptosis. Moreover, miRNAs carried by the TMAO-Exos were quite different from that in the Control-Exos, including miR-92a-3p, miR-103-3p and miR-122-5p, etc. Further analysis showed that these differentially expressed miRNAs were predicted to target genes such as Mapk8, Casp9, Mapk10, Bcl2l11, Ikbkg and Akt1, which were supposed to be involved in the signal pathways related to vascular inflammation and endothelial function. ConclusionsThese novel results provided evidence that TMAO could indirectly talk to vascular endothelial via promoting hepatocytes to secreting exosomes that carried important genetic information, which may give a new insight into the interactions between liver and vasculature in the atherogenesis caused by TMAO. New intervention targeting this cellular crosstalk may be feasible and effective in the prevention and treatment of TMAO-induced atherogenesis.


Abstract
Background Trimethylamine-N-oxide (TMAO) has been proved to be a new proatherogenic compound for promoting vascular in ammation and endothelial dysfunction. Hepatocyte-derived exosomes played an important role in the regulation of vascular in ammation and endothelial function. Since TMAO is produced in the liver, hepatocytes may be the rst potential target of TMAO. However, it is not clear whether TMAO can directly stimulate normal hepatocytes to produce exosomes so as to mediate the motivating effects of TMAO on in ammation and endothelial dysfunction.

Methods
The hepatocytes were cultured and treated with TMAO at a physiological concentration for 24 hours (TMAO-Exos). The untreated group served as the control (Control-Exos). The exosomes were isolated from the culture supernatant and then added to the human aortic endothelial cells (HAECs) for 48 hours.
The mRNA expressions of in ammatory cytokines and caspase-3 were determined by qPCR and cell apoptosis was evaluated by using the Hoechst 33342 staining solution. The miRNA pro le in the exosomes were detected using an RNA-sequencing strategy. The miRNA-mRNA network was predicted, and the biological functions of the target genes were annotated by using bioinformatics methods.

Results
TMAO-Exos were able to promote the expressions of in ammatory cytokines and HAECs apoptosis. Moreover, miRNAs carried by the TMAO-Exos were quite different from that in the Control-Exos, including miR-92a-3p, miR-103-3p and miR-122-5p, etc. Further analysis showed that these differentially expressed miRNAs were predicted to target genes such as Mapk8, Casp9, Mapk10, Bcl2l11, Ikbkg and Akt1, which were supposed to be involved in the signal pathways related to vascular in ammation and endothelial function.

Conclusions
These novel results provided evidence that TMAO could indirectly talk to vascular endothelial via promoting hepatocytes to secreting exosomes that carried important genetic information, which may give a new insight into the interactions between liver and vasculature in the atherogenesis caused by TMAO. New intervention targeting this cellular crosstalk may be feasible and effective in the prevention and treatment of TMAO-induced atherogenesis.

Background
Ischemic heart disease remains a major long-term public health challenge around the world [1]. Chronic vascular in ammation and endothelial dysfunction clearly represent the characteristics of atherosclerosis [2]. In recent years, trimethylamine-N-oxide (TMAO) has been proved to be a proatherogenic compound, for exerting its pathogenic effects via promoting vascular in ammation and endothelial dysfunction [3][4][5].
However, the molecular mechanisms have not been completely explained.
Exosomes are nanosized membrane particles with a 50 to 100 nm size range, which are secreted by various types of cells and transmit information from cells to cells. The functions and characteristics of exosomes mainly depend on the types and states of the host cells from which they are originated. There are many kinds of proteins, lipids, microRNAs (miRNAs) and other non-coding RNAs carried by exosomes, which were considered as the key materials for intercellular communications and play a crucial regulatory role in many biological processes such as immune response, cardiovascular disease, tumor, and neurodegenerative disease [6][7][8].
The latest studied indicated that hepatocyte-derived exosomes played an important role in the regulation of vascular in ammation and endothelial function [9,10]. Since TMAO is produced in the liver [11], hepatocytes may be the rst potential target of TMAO. In fact, recent research has indicated that TMAO could work directly on hepatocytes and thus exerting its in uence over metabolic syndrome [12]. However, it is not clear whether TMAO can directly stimulate hepatocytes to produce exosomes so as to mediate the motivating effects of TMAO on in ammation and endothelial dysfunction.
In the present research, TMAO was found to directly stimulate normal hepatocytes to release exosomes and thus inducing in ammation activation and cell apoptosis. Next, we utilized an RNA-sequencing strategy to characterize the miRNAs candidates contained in the exosomes, and further analysis revealed that the differentially expressed miRNAs were predicted to target the potential genes which were involved in vascular in ammation and endothelial dysfunction. These novel results provided evidence that TMAO could indirectly talk to vascular endothelial via promoting hepatocytes to secreting exosomes that carried important genetic information, which may give a new insight into the interactions between liver and vasculature in the atherogenesis caused by TMAO. Therefore, new intervention targeting this cellular crosstalk may be feasible and effective in the prevention and treatment of TMAO-induced atherogenesis.
The untreated group served as control (Control-Exos). After 24 hours, exosomes were isolated and puri ed from the culture supernatant using differential centrifugation. Brie y, the medium was collected and centrifuged at 300×g for 10min, 2000×g for 10min at 4℃ and then again at 10000×g for 30 min at 4℃. The supernatant was then passed through a 0.22-mm lter (Millipore) and ultracentrifuged at 110000×g for 70 min at 4℃. The pellets were then washed with phosphate-buffered saline (PBS) followed by a second ultracentrifugation at 110000×g for 70 min at 4℃ and then resuspended in PBS. The protein levels of the exosomes were measured using a BCA protein assay kit (23228, Thermo Scienti c). The ultrastructure and size distribution of the exosomes were identi ed by transmission electron microscopy (JEM1200-EX, Japan) and nanoparticle tracking analysis (Nanosight NS300, Malvern, UK) respectively. Protein marker of CD81 (Servicebio, Wuhan, China) was detected by western blotting. Exosomes were labelled with DiI (Beyotime Biotechnology) for in vitro tracer experiment.

Western blot
The procedure was performed by standard protocols as previously described [13]. The exosomal protein concentration was determined using a BCA protein assay kit (23228, Thermo Scienti c). Then the samples were separated by SDS-PAGE and transferred onto the Millipore polyvinylidene di uoride membranes. Primary antibody of CD81 was purchased from Servicebio, and the nal antibody concentration was 1:1000. The expressions of CD81 were detected with enhanced chemiluminescence reagent (Millipore, USA).

Quantitative polymerase chain reaction
Quantitative polymerase chain reaction (qPCR) was performed by standard protocols as previously described [14]. Brie y, total RNA was extracted from treated HAECs using TRIzol reagent (Invitrogen, USA) and concentration was measured using the NanoDrop 2000 spectrophotometer (Thermo Fisher Scienti c, MA, USA). Then RNA was reverse transcribed into cDNA using the Color Reverse Transcription Kit (EZBioscience, USA). qPCR was performed on Bio-Rad CFX-96 (Bio-Rad, USA) with Color SYBR Green qPCR Master Mix (EZBioscience, USA). The mRNA expressions of interleukin-6 (IL-6), monocyte chemotactic protein-1 (MCP-1), tumor necrosis factor-α (TNF-α) and Caspase-3 were normalized to glyceraldehyde phosphate dehydrogenase (GAPDH) by using the 2^-△△CT method. The qPCR primers used in the study were listed in Table 1.

Exosomal miRNA expression pro ling
Total RNA was isolated using TRIzol reagent (Invitrogen, USA), RNA concentration was measured using Qubit® RNA Assay Kit in Qubit® 2.0 Flurometer (Life Technologies, CA, USA) and RNA degradation and contamination was monitored on 1% agarose gels. After RNA quanti cation and quali cation, 1 μg total RNA per sample was used as input material for the small RNA library. Sequencing libraries were generated using NEBNext® Multiplex Small RNA Library Prep Set for Illumina® (NEB, USA.) following manufacturer's recommendations and index codes were added to attribute sequences to each sample and then library quality was assessed on the Agilent Bioanalyzer 2100 system using DNA High Sensitivity Chips. After cluster generation, the library preparations were sequenced on an Illumina NovaSeq 6000 platform and 50bp single-end reads were generated. After sequencing, clean data (clean reads) were obtained by removing reads containing ploy-N, with 5' adapter contaminants, without 3' adapter or the insert tag, containing ploy A or T or G or C and low-quality reads from raw data. At the same time, Q20, Q30, and GC-content of the raw data were calculated. Then, chose a certain range of length from clean reads to do all the downstream analyses. miRNA expression levels were estimated by TPM (transcript per million) and differential expression analysis between the control-Exos and TMAO-Exos samples (three biological replicates) was performed using the DESeq (v1.22.1).

Target genes prediction and functional enrichment analysis
Potential target genes were predicted from miRDB and miRBase databases. DAVID database was used for investigating the functional annotation of the target genes. GO analysis was performed to elaborate the biological functions and KEGG pathway enrichment was used to explore the relevant signal pathways, and networks were performed on Cytoscape platform (v3.8.2) [15]. STRING database (v11.0) [16] was used for analyzing the protein-protein interactions, and networks were performed on Cytoscape platform. P-value ≤ 0.05 was considered statistically signi cant.

Statistical Analysis
Data were presented as mean ± standard error of the mean (SEM). Statistical analysis was conducted using the SPSS 20.0 software (SPSS Inc., Chicago, IL, USA). The comparisons between two groups were performed with independent t-test. P < 0.05 was considered statistically signi cant.

Results
Isolation and characterization of exosomes from hepatocytes culture supernatant Nanovesicles with diameters around 100 nm were isolated and puri ed from the cell culture supernatant, which were consistent with the characteristic size range of exosomes under electron microscopes (Fig.  1a). The size distribution of the exosomes showed no signi cant difference between Control-Exos and TMAO-Exos (Fig. 1b). The expression of exosomal identity marker CD81 was determined in Control-Exos and TMAO-Exos by western blotting, protein from 293T cells was set as negative control (Fig. 1c). Exosomes were labelled with DiI and co-cultured with HUVECs for 24 hours, and it was shown that DiIlabeled exosomes were internalized into HUVECs (Fig. 1d).
The expression pro le of miRNAs in TMAO-Exos.
An RNA-sequencing strategy was conducted to identify the differentially expressed miRNAs between the Control-Exos and TMAO-Exos, and the miRNAs with a P-value ≤ 0.05 were visualized on a heatmap (Fig.  4). Compaired to the untreated group, a total of 17 miRNAs were changed signi cantly ( log2(fold change) ≥ 1 and P-value ≤ 0.05) after exposed to TMAO, in which eight miRNAs were considered as upregulated, and nine of them were down-regulated (Table 2). Eight miRNAs labeled with a purely numerical code represented newly predicted genes, and it is worth noting that six of them are down-regulated, although the function is not yet known. Prediction of miRNA-mRNA network As we known, miRNAs have vital function for negatively regulating protein expression through destabilizing or inhibiting translation of target mRNAs [17]. Therefore, we selected the known miRNAs from Table 2 for next analysis. Thirty-one candidates predicted to interact with the nine differentially expressed miRNAs were picked out and built a miRNAs-mRNA network using Cytoscape software to exhibit the complex interaction (Fig. 5).
GO and KEGG pathway enrichment analysis GO analysis revealed that changes in the biological processes (BP) of the predicted target genes were signi cantly enriched in positive regulation of apoptotic signaling pathway, positive regulation of I-kappaB kinase/NF-kappaB signaling, and protein kinase B signaling, etc. The details were speci ed in Additional le 1. The interacting networks among these biological processes were constructed using ClueGO of Cytoscape (Fig. 6a). KEGG pathway analysis showed that the predicted target genes were strongly associated with the signal pathways that were pivotal in regulating in ammation and endothelial function, such as NF-κB signaling pathway, TNF signaling pathway and Apoptosis (Additional le 2), and the interactions among these signal pathways were constructed using ClueGO (Fig. 6b). The NF-κB signaling pathway (Fig. 7a) and Apoptosis signaling pathway (Fig. 7b) were exported from the KEGG database, the predicted target genes were highlighted in red for showing the positions in the signaling network.

Protein-protein interactions network analysis
To further explore the interactions among the predicted target genes, protein-protein interaction (PPI) networks analysis was processed using the STRING database. The comprehensive module with the highest score (0.998) included 24 nodes and 96 edges (Fig. 8a), which was further subdivided into two gene clusters by using MCODE of Cytoscape (Fig. 8b). Moreover, CytoHubba plug-in of Cytoscape identi ed Mcl1, Bcl2l11, Birc2, Mapk8, Mapk10, Aifm1, Akt1, Akt3, Casp9 and Map3k5 as the top 10 hub genes involved in this module, in which 10 nodes and 38 edges were included (Fig. 8c). Since other genes often interact with each other through these hub genes, it may be the key components of signaling pathways that control vascular in ammation and endothelial function.

Discussion
In recent years, numerous studies have showed that intestinal ora and metabolites could exert regulatory effects on atherosclerosis by inhibiting or accelerating the disease process [18]. Study has con rmed that dietary choline and phosphatidylcholine could be metabolized into trimethylamine (TMA) in the intestinal microbiota, and then TMA would be absorbed into the liver, under catalysis of the avin monooxygenase-3, TMA was further converted to TMAO [11]. Many clinical investigations have shown that high blood TMAO levels were an independent risk factor for atherosclerosis and serious cardiovascular events [11,19,20]. Furthermore, TMAO has been found to promote atherosclerosis via boosting in ammatory activation and impairing VECs function both in vitro and in vivo [3,21,22]. Results from LDLR -/mice demonstrated that increased blood TMAO levels could activate the NF-κB pathway and promote the expressions of in ammatory markers, thus leading to atherosclerosis formation [3]. And elevated plasma TMAO levels in rats could inhibit endothelial nitric oxide synthase (eNOS) expression and boost the production of in ammatory cytokine and superoxide, thus resulting in senescence-related endothelial dysfunction [21]. In vitro experiments showed that TMAO induced in ammation and endothelial dysfunction by inhibiting the activity of eNOS and provoked oxidative stress and activated in ammasome [22]. However, the effective concentrations of TMAO on VECs in vitro were much higher than the actual levels in body [3,4,19,20,22]. On the one hand, it may attribute to the complexity of the internal environment. But on the other hand, it may suggest that the molecular mechanism of TMAO on VECs remain incompletely understood, and the indirect factors contributing to in ammation and VECs injury might not be excluded. Since TMAO is produced in the live, hepatocytes may be the rst potential target of TMAO. The results from recent related studies support the idea that there existed some relationships between TMAO and liver. Sifan Chen and colleagues recently found that TMAO at a physiological concentration could directly bind to hepatic endoplasmic reticulum stress kinase PERK and activated the unfolded protein response, and thus promoting metabolic dysfunction [12]. Another study in animal models of atherosclerosis has found that hepatic miR-146a-5p expression was associated with blood TMAO [23].
Recent studies have shown that exosomes, including hepatocyte-derived exosomes, were involved in atherosclerosis via regulating in ammation and vascular endothelial function [9,24]. In addition, Hirsova P and colleagues have found that primary hepatocytes and Huh7 cells stimulated with lipid released more extracellular vesicles (which possessed the characteristic of exosomes), thereby enhancing the expressions of IL-1β and IL-6 in macrophages via the tumor necrosis factor-related apoptosis-inducing ligand contained in these vesicles [10]. In the present study, we provided evidence that TMAO at a physiological concentration could directly stimulate normal hepatocytes to release exosomes, which were able to promote the expressions of IL-6, MCP-1 and TNF-α, and induced VECs apoptosis. As is wellknown, IL-6, MCP-1 and TNF-α are the key in ammatory factors and play pivotal roles in regulation of in ammation, endothelial dysfunction, and atherosclerosis. IL-6 has been con rmed to exert central roles in the pathogenesis of atherosclerosis [25]. In hypertensive patients combined with coronary artery disease, blood MCP-1 levels were elevated and related to the degree of endothelial damage [26], and MCP-1 secretion from VECs was involved in atherosclerosis [27]. And the role of TNF-α in inducing endothelial dysfunction has been established [28]. VECs apoptosis re ects the damage of endothelial barrier and represents the initial event of atherogenesis [29]. Now widely accepted that VECs injury plays a key role in the development of atherosclerosis [30]. Our previous study has found that proin ammatory lipid could impair eNOS activity and induce VECs apoptosis [13].
It has been recognized that miRNAs carried by exosomes could be actively absorbed by both neighboring and distal cells to exert regulatory roles in the pathogenesis of cardiovascular diseases, and the speci c functions depend on the host-cell molecular architecture. Besides, the differential expression of exosomal miRNAs have been found to be promising biomarkers for early detection of cardiovascular diseases [31]. The latest research showed that steatotic hepatocyte-derived exosomes could transfer miR-1 to VECs, thus inhibiting KLF4 expression and activating NF-κB pathway, thus facilitating endothelial in ammation and atherogenesis [9]. Zheng B and colleagues also found that exosomes secreted from vascular smooth muscle cells (VSMCs) mediated the communication between VSMCs and VECs by transferring KLF5induced miR-155 to VECs, ultimately impaired endothelial function and accelerated atherogenesis progression [32]. In our study, miRNAs carried by the TMAO-treated hepatocyte-derived exosomes were quite different from that in the untreated group, including several well-known miRNAs, such as miR-92a-3p, miR-103-3p, miR-122-5p and miR-199a-3p, etc. Further bioinformatics analysis showed that these differentially expressed miRNAs were predicted to target mRNAs such as Mapk8, Casp9, Mapk10, Bcl2l11, Ikbkg and Akt1, which were supposed to be involved in the signal pathways related to in ammation and vascular endothelial function, including cell apoptosis. A large body of evidence has demonstrated that NF-κB signaling pathway, TNF signaling pathway, Apoptosis signaling pathway, Tolllike receptor signaling pathway and FoxO signaling pathway play a crucial role in in ammation, endothelial function, and atherosclerosis [28,29,[33][34][35][36]. Particularly, NF-κB signaling pathway has been at the crossroads of in ammation and atherogenesis [33], and speci c NF-κB inhibition of endothelial cell reduced proin ammatory gene expression such as IL-6 and TNF, and protected ApoE -/mice from atherosclerosis [34]. Recent studies have revealed that exosome derived from CD137-modi ed endothelial cells could promote Th17 cell differentiation and atherosclerosis progression in ApoE -/mice through NF-κB pathway mediated IL-6 expression [37], and exosomes released from mature dendritic cells carried TNF-a on exosome membrane, which was found to be e cient in activating the NF-κB pathway, thus provoking endothelial in ammation and atherosclerosis [24]. Furthermore, there exists a close relationship between the NF-κB pathway and the in ammatory cytokines of IL-6, MCP-1 and TNF-α. IL-6 and MCP-1 contain the binding elements for NF-κB, which is critical for transcriptional induction of IL-6 and MCP-1 genes, and TNF-a serves as one of the NF-kB-activating factors, inducing IL-6 and MCP-1 transcription intensely [38,39].
So far, to our best knowledge, no studies have de ned the relationship between hepatocyte-derived exosomes and the facilitating effects of TMAO on vascular in ammation and endothelial dysfunction.
The intriguing ndings demonstrated that hepatocyte-derived exosomes undertook the proatherogenic roles of TMAO in the vasculature, at least in part, via transferring a cluster of miRNAs to the endothelial cells. And further analysis indicated that the level and diversity of the exosomal miRNAs could be directly correlated with the TMAO-induced vascular in ammation and endothelial dysfunction. Future investigation of the precise mechanisms whereby circulating exosomal miRNAs interact with the corresponding target genes to irritate in ammatory response and impair endothelial function will increase our knowledge of the roles of TMAO in vascular health and atherosclerosis.

Conclusions
In the present study, we found that TMAO was able to stimulate normal hepatocytes to produce exosomes, which could be absorbed by endothelial cells, thus enhancing the mRNA expressions of in ammatory cytokines and cell apoptosis via transferring a cluster of miRNAs to the VECs and regulated signal pathways related to in ammation and vascular endothelial function. These studies may provide a new insight into the interactions between liver and vascular endothelial in the atherogenic process orchestrated by TMAO, at least in part, could be attributable to the hepatocyte-derived exosomes and the loaded miRNAs, which suggests that targeting this cellular crosstalk may provide a novel approach that can restrain the deleterious effects induced by TMAO on VECs.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for Publication
Not applicable.
Availability of data and materials