Age-Related Changes in the Gut Microbiota Promote Atrial Fibrillation

Background: Aging is the most signicant contributor to the increasing prevalence of atrial brillation (AF). The gut microbiota dysbiosis is involved in age-associated diseases. However, whether age-associated gut microbial dysbiosis contributes to AF is still unknown. The aim of this study was to evaluate the effect of gut microbiota on the susceptibility of aging-induced AF and to elucidate the underlying mechanisms. Results: The gut microbiota proling of fecal samples in aged (22-24 months old) and young (2-3 months old) rats was performed by 16S ribosomal RNA gene analysis. A rat model of fecal microbiota transplantation (FMT) was established for analyzing the possible role of age-associated gut microbial dysbiosis in AF. Here, we found that aging process led to marked shift of the microbiota spectrum. The microbiota in young rats following FMT resembled that of aged microbiota and transmitted the increased AF susceptibility by elevation of circulating lipopolysaccharide (LPS). The mechanism for LPS-driven atrial pro-arrhythmic action was dependent on the activation of atrial nucleotide binding and oligomerization domain-like receptor family pyrin domain-containing (NLRP3)-inammasome. Inhibition of inammasome by MCC950 resulted in lower atrial brosis and AF susceptibility. In addition, atrial brosis, plasma LPS concentration, plasma glucose to oral glucose tolerance test (OGTT), intestinal permeability, and gut pathology were signicantly increased in elderly human subjects. Finally, altering the microbiota in aged recipient to resemble that of young restored the intestinal barrier dysfunction and LPS and impaired glucose tolerance, and the worse outcomes of aged dysbiosis on atrial brosis and AF susceptibility were abrogated. Conclusions: These data suggest that age-associated microbial dysbiosis induces circulating LPS and impairs glucose tolerance, and thereby promotes AF susceptibility through enhanced activity of atrial NLRP3-inammasome. is the rst to demonstrate a mechanistic link between age-associated gut microbiota and the pathophysiology of AF. Our results suggest that manipulation of the gut microbiota composition to block activation of NLRP3-inammasome is a potential novel anti-AF approach, especially in aged subjects.


Introduction
Atrial brillation (AF) is the most common cardiac arrhythmia in clinical practice and it can cause substantial morbidity and mortality [1,2]. The incidence of AF rises steadily with increasing age. However, the complex complications and high recurrence rate in aged patients often hinder clinical AF treatment and prevention. Better understanding of aging-related AF and discovery of novel therapeutic strategies are urgently needed.
Gut microbiomes and their metabolites including the lipopolysaccharide (LPS), bile acids, trimethylamine N-oxide (TMAO), and short-chain fatty acids, have been implicated in aging-related cardiovascular diseases [3][4][5]. It has been reported that the gut microbiota composition of older human differs from that of young one with decreased Bi dobacterium and Firmicutes and increased Bacteroidetes and Enterobacteriaceae, which is referred to gut dysbiosis in elders [6]. The age-associated gut microbiota dysbiosis could accelerate the rate of aging leading to the "in amm-aging" state in the host [7,8]. Besides, the increased levels of in ammatory cytokines have been found in AF patients [9], indicating that in ammation is closely linked to the development of AF [10,11]. However, how age-associated microbial dysbiosis and the associated "in amm-aging" status promote AF remains unclear and merits experimental elucidation.
Lipopolysaccharides (LPS) are consisted of a hydrophobic lipid A and a hydrophilic carbohydrate core with polysaccharide O-antigen, and which belongs to cell wall structural component of Gram-negative (e.g. E. coli) [12]. LPS is a potent trigger of systemic in ammation in host and increased with aging process, suggesting that LPS is an important mediator in the development of "in amm-aging" [5]. An increased serum LPS levels are prediction of adverse events in AF patients has been reported [4]. Indeed, LPS can translocate from the intestine by either paracellular or transcellular mechanisms to enter circulation and quickly recognized by the toll-like receptor (TLR)4 [13]. TLR4 activates intracellular signaling cascades that results in the translocation of nuclear factor kappa B (NF-ĸB) and mediation of numerous pro-in ammatory genes expression. Thus, we speculated that gut derived-LPS had an implication on increased of age-related AF susceptibility, which may cause by alterations of aged dysbiosis-mediates systemic in ammation.
In the current study, we explored the causal relationship between age-associated microbiota dysbiosis and AF in a rat model of FMT. Our results showed that aged dysbiosis elevated LPS level and impaired glucose tolerance, thereby promoting aging-related AF through activating atrial NLRP3-in ammasome.
The results obtained from plasma samples and atria and intestine tissues of human subjects con rmed the main ndings of our animal study.

Methods
Human Studies. Electrophysiological study.
Atrial brillation (AF) was induced with essentially the same protocol as described previously in detail [14]. In brief, rats were anesthetized with 1% sodium pentobarbital (30 mg/kg) through peritoneal injection.
After open-chest surgery, a 1.9-F octapolar catheter (Transonic Systems Inc., New York, USA) was placed on the right atrium to deliver programmed stimuli. To assess the inducibility of atrial arrhythmias, 50-Hz burst pacing was applied for 3 s with 12 bursts separated by a 2-s interval. AF was de ned as >1 s of irregular atrial electrograms (>800 bmp) with irregular ventricular response. AF duration was de ned as the mean duration of all AF episodes within 60 s in each rat.
The sequencing of 16S ribosomal RNA (rRNA)-encoding gene of fecal samples was performed as described previously [15]. Microbiota in fecal samples, representative of that in the proximal colon, was collected from rats and immediately frozen and stored. Bacteria taxa were analyzed by amplifying the V3 to V4 hypervariable regions of the 16S rRNA gene and sequenced with Illumina HiSeq 2500 platform (Illumina, San Diego, CA). Quality lter of 16S rRNA sequences was conducted by using Quantitative Insights Into Microbial Ecology (QIIME) and custom scripts. Resulting sequences were then clustered into Operation Taxonomic Unit (OTU) by searching against Greengenes reference database of 16S rRNA gene sequences and clustered at 97% by Uclust algorithm [16]. The differences of Genus level and Family level between treatment groups were calculated by using pairwise Wilcoxon rank-sum tests.
Collect Fecal Sample and Microbiota Transplantation.
Feces from 3 aged rats (22-24 months old) with increased AF susceptibility residing and from young rats (2-3 months old) were collected and the samples were pooled by group. Each feces sample was diluted in sterile PBS (100 mg feces in 2 ml buffer) and homogenized for 5 min. After vortexed for 1 min and centrifuged at 800g for 3 min, the supernatant was collected.
The protocol of fecal microbiota transplantation (FMT) was performed as described by García-Lezana et al. [17] In order to facilitate re-colonization of transplanted gut microbiota, rats were administrated with omeprazole (50 mg/Kg/d) for 3 days before FMT for decontamination. To facilitate intestinal emptying, 1 ml and 2 ml of CitraFleet (sodium picosulfate, 0.16 mg/ml and magnesium oxide 51.2 mg/ml) along with 2 ml of water were administrated to the animals 24h and 12h prior to FMT, respectively. A study reported by Manichanh et al. [18] showed that the transplanted microbiota can be acquired following a single fecal transplant and can last for at least three months. We therefore adapted a single oral gavage for 6 weeks for re-colonization of transplanted gut microbiota. The young rats gavaged with autogenous fecal were referred to the Young-FMT group and the young rats gavaged with fecal from aged rats to the Aged-FMT group.
For the experiments involving long-term young FMT rescuing study, we collected feces from 3 young without AF rats (2-3 months old). The aged recipient rats (16-18 months old) were administrated by autologous fecal microbiota transplantation (Aged+AgedFMT group) and microbiota transplantation from young rats (Aged+YoungFMT group), respectively, once a week for 6 months.
Nuclear and Cytoplasmic Protein Extraction.
Pre-frozen atrial tissue was cut into small pieces and washed with PBS, followed by centrifugation at 10000r/min for 5 min, according to the subcellular structure nuclear and cytoplasmic protein extraction kit protocol (Wuhan, China). Homogeneous cell suspension was collected by vortex after mixed with cytoplasmic protein extraction reagent A in a tube. Cytoplasmic protein extraction reagent B was then added into the tube on ice. Finally, the tube was centrifuged at 16000g for 5 min, and the supernatant was removed, followed by addition of the nucleoprotein extraction reagent to the remaining insoluble cell debris containing nuclei. After 40 min vortex and 5 min centrifugation, the supernatant was transferred into a clean tube containing nuclear proteins.
Isolation of Rat Cardiomyocytes.
Neonatal rat cardiomyocytes (NRCMs) were isolated from 1 to 3-day-old Sprague-Dawley (SD) neonatal rats as previously described [19]. Brie y, neonatal rat hearts were nely minced and placed together in 0.25 % trypsin. Cell suspension was collected, centrifuged and re-suspended in DMEM supplemented with 10 % FBS and 100 U/ml penicillin and 100 μg/ml streptomycin. The re-suspension was plated onto a culture ask, and cardiac broblasts (CFs) were obtained after differential adhesion for 90 min leaving cardiac myocytes (CMs) in suspension portion. Then the culture medium was replaced to purify CFs, and the CMs were seeded at a density of 1 × 10 6 cells per well in a six-well culture plate. Cell cultures were incubated at 37°C in a humidi ed atmosphere with 5 % CO 2 and 95 % air. In vitro experiments, the CFs and CMs were cultured with LPS (100 ng/mL) [20] and different concentrations of glucose: 5.5 mmol/L glucose (control) and 30 mmol/L glucose (HG) [21]for 72 h at 37°C with 5% CO 2 , respectively. Histology.
Fresh atrial and proximal colon tissues were cut into 5-μm sections, xed in 4% paraformaldehyde and embedded with para n. Then, the tissues were stained with hematoxylin and eosin (HE), Masson's trichrome and TUNEL staining according to our previous study [22]. The HE staining was used to determine atrial and proximal colon structure. Subsequently, the brotic area was measured by Masson's trichrome staining and quanti ed by using software (Image-pro plus 6.0, Meida Cybernetics LP). Collagen volume fraction was calculated as collagen area/total area × 100%. The cell apoptosis was assessed by TUNEL staining and semi-quanti ed as the number of apoptotic cells per eld.
Following 6 h of fasting, rats received a single i.v. dose of 4kDa FITC-dextran (44mg/100g, Sigma Aldrich, catalog #FD4). After 4 h feeding, blood samples wwere collected and centrifuged. FITC uorescence was quanti ed using the standard curve method according to manufacturer's instructions.
Plasma and Fecal LPS Levels.
The levels of serum and fecal lipopolysaccharide (LPS) from rats were measured by using a quantitative chromogenic limulus amoebocyte lysate (LAL) QCL-1000 test kit (Lonza Bioscience, Switzerland) according to the manufacturer's instructions. First, the serum and fecal samples were pretreated with pyrogen-free water provided in the kit. Then, the dilute samples were deactivated in 75°C water bath for 10 min. After incubation with 50 μl of LAL reagent at 37°C for 10 min, 100 μl of LAL chromogenic substrate was added into the samples for 6 min. Finally, 100 mg/ml of SDS was added to terminate the reaction with yellow color indicating cleavage of the substrate. The OD value of LPS level was measured by spectrophotometrically at 405 nm.
After proximal colon protein extraction, the protein solution was reduced with 5 mM dithiothreitol and 11 mM iodoacetamide for digestion [23]. The tryptic peptides were dissolved in proper sequence with 0.1% formic acid (solvent A) and different concentrations of solvent B (0.1% formic acid in 98% acetonitrile), followed by centrifugation with an EASY-nLC 1000 UPLC system. The peptides were subjected to NSI source followed by tandem mass spectrometry (MS/MS) in Q ExactiveTM Plus (Thermo Fisher Scienti c, US) coupled online to the UPLC. Then, the peptides were selected for MS/MS using NCE setting of 28 and the fragments were detected in the Orbitrap. A data-dependent procedure was performed by alternating between one MS scan, which was followed by 20 MS/MS scans with 15.0s dynamic exclusion. Automatic gain control (AGC) was set at 5E4. Fixed rst mass was set as 100 m/z.

Statistical Analyses.
The statistical analyses were performed with GraphPad Prism 6.0 software (GraphPad Software, Inc, La Jolla, CA). Data are all expressed as mean ± SEM. Two-group comparisons were performed using Student's non-paired t-test or Kruskal-Wallis H test. Data with more than two groups were analyzed by one-way ANOVA, followed by Tukey tests. AF incidence rates were compared with Fisher's exact test. The statistics regarding microbiome analysis in the subsections on microbiome analysis was described in gut microbiota pro ling of methods. Differences were considered as statistically signi cant when P < 0.05.

Results
Aging Altered the Gut Microbiota Composition and Promoted AF Susceptibility Firstly, we found that the aged rats (22-24 months old) had an increased atrial brillation (AF) susceptibility than young rats (2-3 months old) did (Additional File 1:FigureS1A-B) [24,25], while the atrial effective refractory period (ERP) were no changes (Additional File 1:FigureS1C). To investigate the relationship between gut microbiota and aged-related AF, we further pro led the fecal microbiota from aged and young rats by analyzing the DNA sequence encoding the 16S rRNA gene. The PCoA of Unweighted UniFrac distances (β-diversity) among aged and young rats (Figure1A) showed that microbiota community structures were signi cantly separated by aging at the rst principal coordinate (PC1 axis) and at the second principal coordinate (PC2 axis). On the family level, the abundance of Lactobacillus spp. was higher, whereas the abundances of members in family Bacteroidaceae were clearly lower in young rats than in aged rats (Figure1B). These differences could be traced to α-diversity at both the genus level and operational taxonomic unit (OTU) level (Figure1C), which showed distinct patterns in young as compared to those of aged rats, suggesting a signi cant shift in gut microbiota composition during aging.
Age-Associated Microbial Transplantation Increased AF Susceptibility and Atrial Fibrosis Next, we performed fecal microbiota transplantation (FMT) in young rats by utilizing fecal samples collected from aged AF rats (Aged-FMT) or young rats themselves (Young-FMT) for 6 weeks. Analysis of fecal microbiota composition revealed that aged-related FMT successfully changed the gut microbial ecology of young rats. Unweighted UniFrac distances of 16S rRNA gene sequences showed that the microbiota clustering between Young-FMT group and Aged-FMT group were separated (Additional File 1: FigureS2A). The top 15 differential microbiotas in genus level suggested that the gut microbiota community of recipient rats was successfully transmitted towards to aging phenotype (Additional File 1: FigureS2B). These differences were also traced to abundances of OTUs at the species level (Additional File 1:FigureS2C-D).
Interestingly, burst pacing induced AF in Aged-FMT rats, but not in Young-FMT rats ( Figure 1D). Moreover, Aged-FMT had a marked increase in both AF inducibility ( Figure 1E) and AF duration (Additional File 1:FigureS2E) as compared with those in Young-FMT rats. However, the AERP were no changes between two groups ( Figure 1F). In addition, Masson's trichrome staining showed that atrial brosis was markedly increased in Aged-FMT rats ( Figure 1G) and the expression levels of TGF-β1 and α-SMA in atria, two brosis-related marker proteins, were also markedly increased ( Figure 1H). Taking together, these data suggest that the age-associated microbial dysbiosis causally increased both AF susceptibility and atrial brosis.

Aging Human with AF Elevated Circulating LPS Levels
Gut-derived lipopolysaccharides (LPS) are produced by the outer membrane vesicles of most Gramnegative bacteria and plays an important role in the in ammatory process of aging-related diseases [5,26]. To determine the role of LPS in aging-related AF, we collected 1,152 plasma samples of human donors for LPS levels detection. Herein, we found that the plasma LPS concentrations were progressively higher in human donors with advancing age (Figure 2A), while Elderly+AF patients having the highest levels ( Figure 2B). Besides, the circulating LPS had good ability (AUC, 0.922; 95% CI, 0.867-0.977) to distinguish the elderly patients who presents AF ( Figure 2C), suggesting that the alteration of circulating LPS during aging had a closely relationship with development of AF.

Inhibition of LPS Prevented Aged Dysbiosis-induced AF Susceptibility and Atrial Fibrosis
The increased circulating LPS levels were also found in aged rats ( Figure 2D) and Aged-FMT rats ( Figure  2E), suggesting that aged microbiota transfer the LPS of aging phenotype to recipient rats. LPS was exclusively mediated via activation of cell surface TLR4 to trigger MyD88 expression, which leads to the secretion of downstream proin ammatory cytokines [27]. Interestingly, the expression of TLR4/MyD88 pathways were higher in atrial tissue than other organs of aged rats ( Figure 2F). Moreover, LPS-RS (1 mg/kg, once per week for 6 weeks), a potent LPS antagonist, prevented the increased AF inducibility and AF duration in Aged-FMT rats (Figure2G-I). Masson's trichrome staining showed that atrial brosis and the expression levels of TGF-β1 and α-SMA in atria were also markedly reduced by LPS-RS (Additional File 1:FigureS3A-B). Taking together, these data indicated that aged microbiota could elevate circulating LPS and promoted AF susceptibility.

Age-Associated Microbiota Transplantation Increased Gut Permeability
There were no differences in the abundance of Gram-negative bacteria and fecal LPS levels between Aged-FMT and Young-FMT ( Figure 3A-B), suggesting that the elevated circulating LPS of aging phenotype may be due to the dysfunction of gut barrier. Therefore, we investigated whether the integrity of the intestinal barrier is altered by aged FMT. Firstly, the intestinal permeability was tested with oral uorescein isothiocyanate (FITC)-dextran. We found that the level of serum FITC-dextran in Aged-FMT rats was signi cantly higher than that in Young-FMT rats ( Figure 3C). Then, we observed that the levels of tight junction proteins including Zona occluding-1 (ZO-1), occludin (Ocln) and claudin4 (Cldn4), were remarkably decreased in the proximal colon of Aged-FMT rats ( Figure 3D). The similar results of tight junction proteins were also found in colon tissues of elderly human ( Figure 3E). Moreover, HE and Masson staining of proximal colon revealed a shorter in villi length and an increased brotic area in Aged-FMT rats (Figure3F and G) and elderly human (Figure3F and H) [28]. Taking together, these observations suggest that loss of integrity of intestinal barrier and alterations of gut pathological are transferable by aged FMT, which may result in transport of a large amount of gut-derived LPS into circulation.

Enhanced Intestinal In ammation Underlies the Dysfunction of Gut Barrier
To uncover the mechanisms underlying the microbiota-epithelium crosstalk responsible for the observed gut phenotype, we deeply pro led the proteomics of proximal colon tissue samples. A total of 179 proteins (124 up-regulated and 55 down-regulated) were found to be signi cantly and aberrantly expressed in proximal colon tissue samples of Young-FMT as compared with those in Aged-FMT (foldchange > 2.0, P < 0.05). A volcano plots illustrated the pro les of the differentially expressed proteins ( Figure 4A) between Young-FMT and Aged-FMT rats. The numbers of differentially expressed proteins were shown in Figure 4B. The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of Q1-Q4 proteomics data showed that these differentially expressed proteins were enriched in 33 pathways and that were mainly involved in in ammatory-related pathways ( Figure 4C). So, we further detected the expression levels of representative proteins of enrichment analysis pathways including ISG15, S100A8 and S100A9 of proximal colon tissue in Aged-FMT and Young-FMT rats, to verify the accuracy of proteomic analysis results. The expression levels of in ammation-related markers were remarkably increased in Aged-FMT rats ( Figure 4D) and elderly human donors (Figure 4E), suggesting a mechanistic explanation for the increased intestinal permeability and the elevated circulating LPS with increased age, which can be transferred by the aged-associated microbial transplantation.

LPS Promoted AF via Activating Atrial NLRP3-In ammasomes
LPS-induced activation of TLR4 as priming signal for engaging NF-κB-mediated expression of NLRP3in ammasome, and the later was involved in occurrence of AF [29]. To test whether NLRP3in ammasomes was activated in aged dysbiosis-induced AF, we measured protein levels of NLRP3, ASC, and active Casp1-p20 in atria samples collected from elderly patients (over 65 years) who had a history of AF. The expression levels of NLRP3-in ammasomes in atria of Elderly and Elderly+AF patients were markedly elevated, while Elderly+AF patients having the highest levels ( Figure 5A), suggesting that NLRP3-in ammasome of aging speci c phenotype is related to the pathogenesis of aged-related AF.
Then, we detect the activation of NLRP3-in ammasomes in aged microbiota recipient rats. While protein expression levels of ASC remained unchanged, the levels of NLRP3 and active Casp1-p20 were signi cantly increased in atria of Aged-FMT rats (Figure5B). Moreover, the upregulation of the NLRP3in ammasome components in Aged-FMT rats was associated with an enhanced activation of TLR4/MyD88/NFκB pathways (Additional File 1:FigureS4A) and an increased nuclear localization of phosphorylated NFκB (pNFκB) (Additional File 1:FigureS4B), suggesting that both "priming" and "assembly" of NLRP3-in ammasome are activated by aged microbiota transplantation. However, the protein levels of CD68, a macrophage marker, were unchanged in both Aged-FMT and Young-FMT groups (Additional File 1:FigureS4C), suggesting that the initial activation of atrial NLRP3-in ammasomes in Aged-FMT rats may be due to the systemic in ammation, but not macrophage in ltration.
Next, we administrated aged microbiota recipient rats with a selective in ammasome-inhibitor MCC950 (15mg/kg i.p. once per week for 6-weeks) [30] to interrupt assembly of the NLRP3-in ammasome complex. Indeed, MCC950 signi cantly reduced the expression level of atrial NLRP3 pathway in Aged-FMT rats (Additional File 1:FigureS5A). Moreover, LPS-RS also signi cantly reduced activation of atrial NLRP3-in ammasome in aged microbiota recipient rats (Additional File 1:FigureS5B). Burst pacing rarely induced AF in MCC950 rats, whereas it commonly induced AF in Aged-FMT rats ( Figure 5C). Both AF inducibility and AF duration were also reduced by MCC950 in Aged-FMT rats ( Figure 5D-E). We found that Aged-FMT+MCC950 group rats exhibited less atrial collagen deposition and lower collagen volume fraction ( Figure 5F), and lower expression levels of TGF-β and α-SMA ( Figure 5G). These results suggest that activation of NLRP3-in ammasome is critical for age-associated microbial dysbiosis-induced AF.

LPS and Excess Glucose Synergistically Enhanced NLRP3-In ammasome Activity
The NLRP3-in ammasomes were activated in responding to excess extracellular glucose [31] or elevated LPS [32], all of these were accompanied with increased aging [5]. To determine the role of glucose metabolism on aged microbiota-induced AF, the alteration of glucose tolerance both in human and rats were tested. Here, we collected 12,012 oral glucose tolerance test (OGTT) results of clinical individuals and found that the plasma glucose levels in responding to OGTT rise consecutively for every decade of age until was peaked at the range of 81~90 age (Figure 6A), and the results of peak blood glucose is delayed with increased aging, suggesting that aging leads to prolonged glucose load time. Moreover, the peak glucose levels ( Figure 6B) and 2hG levels (Additional File 1:FigureS6A) in responding to OGTT were also progressively higher with aging. Interestingly, 2hG levels rise with increasing ages in both sexes, reaching higher levels in the men than women with same age (Additional File 1:FigureS6B), which is consistent with the gender difference of AF incidence [33]. The similar trend of aging phenotype still exists in DM and NDM patients, but 2hG levels were always higher in DM than NDM (Additional File 1: FigureS6C). More importantly, we found that the 2h OGTT results in Aged-FMT rats were signi cantly higher than those in Young-FMT rats (Figure6C), while LPR-RS could prevent it (Figure6D). These data suggesting that aged microbiota dysbiosis causes impaired glucose tolerance, and which may result from increased circulating LPS levels.
To explore the underling mechanisms of LPS and high glucose (HG) on atrial NLRP3-in ammasome activation during aging, we assessed the expression levels of NLRP3, ASC, and active Casp1-p20 induced by LPS and HG in vitro. While protein levels of ASC remained unchanged, LPS+HG signi cantly increased the levels of both NLRP3 and active Casp1-p20 in cardiac broblasts (CF) ( Figure 6E) and cardiac myocytes (CM) (Additional File 1:FigureS7A), respectively. Interestingly, neither LPS nor HG alone could activate NLRP3-in ammasome in CF, but only when being incubated with both of LPS and HG, can NLRP3-in ammasome be activated ( Figure 6E). More importantly, the expression levels of TLR4 and MyD88 were signi cantly higher in the CF compare with that in CM induced by LPS+HG ( Figure 6F), suggesting the TLR4/MyD88 pathways in CF was predispose to LPS and HG. Besides, LPS+HG remarkably increased the expression levels of TGF-β and α-SMA in CF ( Figure 6G), but not affect the expression levels of apoptosis markers including Bcl-2 and Bax in CM (Additional File 1:FigureS7B).
Taken together, these results indicating that LPS and HG synergistically enhance atrial NLRP3in ammasome activity and leads to atrial brosis in aged-related AF.

Long-term Young Microbiota Transplantation Prevented Aged-related AF and Atrial Fibrosis
We next tested the therapeutic potential of microbiota re-colonization on aged-related AF in a long-term FMT rat model. A young microbiota model was established in aged rats (16-18 months old) by FMT for 6 months from young rats (2-3 months old). Firstly, we observed that the expression levels of tight junction proteins were increased in aged recipient rats with young fecal (Aged+YoungFMT) compared with aged recipient rats with autologous fecal (Aged+AgedFMT) (Additional File 1:FigureS8A). In addition, the Aged+YoungFMT rats had a longer villi length in the proximal colon by HE staining (Figure 7A and B). A decrease in brotic area of proximal colon was also observed in Aged+YoungFMT rats by Masson staining (Figure 7A and C).
Then, we found that the elevated plasma LPS ( Figure 7D) were restored in young microbiota recipient rats. Intriguingly, we found that young FMT signi cantly decreased activation of atrial NLRP3in ammasomes ( Figure 7E). The atrial brosis ( Figure 7F) and expression levels of TGF-β and α-SMA (Additional File 1:FigureS8B) were also attenuated by young fecal transplantation. Moreover, the aged rats that re-colonization with young microbiota were failed in inducing AF ( Figure 7G) and inducibility ( Figure 7H) by burst electrical pacing, as well as the AF duration ( Figure 7I), suggesting that long-term young FMT can prevent the occurrence of aged-related AF by inhibiting the activity of atrial NLRP3in ammasome. Taking together, these results further clarify the role of gut microbiota in aged-related AF and provide more pieces of evidence for a therapeutic strategy targeting gut microbiota in aged-related cardiovascular diseases.

Discussion
The present study provides new pathophysiological insights into the causal relationship between aged gut microbiota and the atrial brillation and opens a new venue of therapeutic strategy to treat agedrelated AF (Figure 8). Herein, we showed that age-associated microbial dysbiosis caused gut barrier dysfunction, abnormal elevation of circulating LPS, impairment of glucose tolerance, and enhancement of the activities of atrial NLRP3-in ammasomes, leading to an increase in AF susceptibility. Such a pro-AF effect elicited by age-associated microbial dysbiosis is reversible as indicated by the abrogation of pro-AF property following long-term young microbiota transplantation. More importantly, the ndings at the molecular level in rat models of aging were reproduced with human atrial, colon and plasma samples collected from a large cohort of patients.
Age-associated microbial dysbiosis causes an increased intestinal permeability [28], which in turn promotes the transfer of harmful substances and pathogens to the bloodstream [34]. Herein, we observed that both of aged rats and elderly human donors had reduced expression levels of tight junction (TJs) proteins and increased intestinal permeability. In particularly, by using FMT approach, we found that the gut pathological phenotype including shorter villus and increased brosis area of colon were transmitted by aged microbial to young recipient. The pro-in ammatory viral and bacterial pathogens cause typically disruption of TJs and result in dysfunction of gut barrier due to the loss of enterocyte microvilli [35]. Previous studies have shown that the increased Bacteroidetes is associated with dysfunction of innate immune system, resulting in host present in a pro-inflammation state [36], while de ciency of Lactobacillus spp. is involved in gut in ammation and bacterial infection [37]. These results suggest that alteration of microbiota during aging can cause a potent pro-in ammatory effect on aggravating intraintestinal in ammation response, and it may explain why intestinal permeability is increased. More importantly, proteomic analysis showed that in ammation-related signaling in proximal colon was signi cantly increased in rats received aged microbiota. It could be speculated that if aged microbiota dysbiosis-induced AF is dependent on systemic in ammation, then, there would be some special inducers present in circulation.
The gut microbial metabolites include TMAO and SCFAs and Bile acids and LPS play a pivotal role in the onset and progression of aging-related cardiovascular diseases [3,38]. LPS was considered as the most potent pathogen association molecular pattern in human that is associated with "in amm-aging" [5].
Herein, we found that the plasma LPS was increased in aged rats and aged microbiota recipient rats, suggesting aged microbiota re-colonization induced circulating LPS. More importantly, we also found that the circulating LPS concentrations were progressively higher with increased age. Aged microbial transplantation successfully transmitted LPS of aging phenotype to young recipient rats, whereas the abundance of Gram-negative bacteria and LPS in fecal were unchanged. These ndings con rmed that age-associated microbial elevates circulating LPS by increasing intestinal permeability and resulting host presents in an "in amm-aging" state. On the other hand, LPS was referred as the potent pathogen association molecular pattern in host, which activated NLRP3-in ammasome through enhancing TLR4/MyD88/NF-κB pathway. Herein, we found that the expression of TLR4 and MyD88 were highest in atria than other organs in aged rats. Thus, we can reason that aged-related AF induced by LPS may depend on the heterogeneity of TLR4/MyD88 in host, and results in serum LPS levels could speci city predicted the disadvantage outcomes in AF patients [4].
Herein, we found that the activity of atrial NLRP3 in elderly patients with or without AF was higher than that in young controls, suggesting that NLRP3 activity is closely associated with AF, especially during aging. Interestingly, inhibition of NLRP3-in ammasome ameliorates the age-related neurodegenerative disease by modulating the relationship between in ammation and gut microbiota composition, suggesting that NLRP3 activity to engaging for dysbiosis-induced aged-related in ammation diseases [39]. As both elevated LPS [5] and hyperglycemia [40] were accompanied by aging, we speculated that both LPS and HG may have promoting effect on aged microbiota-induced AF [4]. Herein, we found that the plasma LPS and glucose in responding to OGTT were increased in aged microbiota recipient.
Activation of NLRP3-in ammasome particularly relevant to aging was resulted from a two-step process: 1) "Priming", introducing NLRP3 protein expression triggered by Toll like receptor-NF-κB signaling, and 2) "Assembly", enabling activators to trigger caspase-1 cleavage. LPS recognition by TLR4/MyD88 is achieved by promoting the transfer of NF-κB into cell nucleus where it causes transcriptional activation, resulting in successive activation of NLRP3-in ammasome "priming". Moreover, excess glucose could promote "assembly" of in ammasome complex. Herein, we found that the enhanced NLRP3 was accompanied by an increased expression of TLR4/MyD88/NF-κB pathway in aged-FMT rats, suggesting that both "priming" and "assembly" processes of in ammsome complex are activated. Besides, aged microbiota FMT successfully activated atrial NLRP3-in ammasome and ultimately increased AF susceptibility and atrial brosis. In addition, a speci c NLRP3-in ammasome inhibitor, MCC950, attenuated AF susceptibility and atrial brosis in aged microbiota recipient. Conversely, young microbiota restored the LPS and impaired glucose tolerance in aged rats.
Recent study showed that the enhanced CM-speci c NLRP3-in ammasome was involved in development of AF by inducing atrial brosis and calcium handling dysregulation [41]. However, whether the LPS and HG had a synergistic effect on aged-related AF due to NLRP3 activation had never been previously reported. In the present study, we found that only simultaneous stimulation with both LPS and HG could enhance NLRP3-in ammasomes in CF. Besides, the expression levels of brosis-related proteins in CF were also increased. Moreover, recent study showed that the activation of NLRP3-in ammasome in the hepatic stellate cells (HSCs) due to LPS could promote liver brosis also con rmed our ndings [42]. The NLRP3-in ammsome is mainly distributed in macrophage and CF, but less in CM [43], and it can interact with different types of cells [44]. Thus, we speculated that NLRP3 activation could promote the release of damage associated molecules from CF into the extracellular space to activate CM in the development of aged microbiota dysbiosis-induced AF.
Impaired glucose tolerance accompanied with aging is another contributor to development of AF. Here, we collected a total of 12,012 OGTT records of individuals and found OGTT was increased consecutively with aging. Moreover, the peaked levels of 2hG and 3hG at range of 81~90 ages suggested that glucose loading duration was prolonged by aging. Interestingly, men had a higher 2hG than women did, which is consistent with the gender difference of AF incidence [33]. Previous study had shown excess LPS led to age-associated hyperglycemia by inducing mitochondrial dysfunction, cellular senescence and adipose tissue dysfunction [45]. Similarly, we speculated that the mechanism underlying aged microbiota-induced increased OGTT is insulin resistance or adipose dysfunction by elevated circulating LPS. Further investigation is required to elucidate the precise mechanisms by which the LPS alter the glucose metabolism during aging.
Our study has several limitations. There is no clear de nition of equivalent human microbiota dysbiosis, which yields therapeutic value of FMT based on animal studies [46]. Besides, there is paucity of data on long-term safety of FMT and some clinical case reports showed that FMT treatment causes serious infectious disease [47]. Thus, the safety issue of long-term FMT in elderly AF patients must be carefully evaluated in the clinical settings. Moreover, although the diet and housing conditions were carefully controlled between young rats and aged ones, it the possible in uence by unidenti ed factors on the gut microbiota of aged rats could not be ruled out. Clearly, the mechanisms for the unfavorable alterations of gut microbiota during aging needs further in-depth investigations.

Conclusions
In summary, in this study we have demonstrated that gut microbiota dysbiosis is associated with the pathogenesis of aged-related AF. Young microbiota transplantation in part attenuated AF susceptibility and atrial brosis by inhibition the activity of atrial NLRP3-in ammasome in aged rats. To the best of our knowledge, this study is the rst to demonstrate a mechanistic link between age-associated gut microbiota and the pathophysiology of AF. Our results suggest that manipulation of the gut microbiota composition to block activation of NLRP3-in ammasome is a potential novel anti-AF approach, especially in aged subjects.