Intestinal SIRT1 Deficiency-Related Intestinal Inflammation and Dysbiosis Aggravate TNFα-Mediated Renal Dysfunction in Cirrhotic Ascitic Mice

In advanced cirrhosis, the TNFα-mediated intestinal inflammation and bacteria dysbiosis are involved in the development of inflammation and vasoconstriction-related renal dysfunction. In colitis and acute kidney injury models, activation of SIRT1 attenuates the TNFα-mediated intestinal and renal abnormalities. This study explores the impacts of intestinal SIRT1 deficiency and TNFα-mediated intestinal abnormalities on the development of cirrhosis-related renal dysfunction. Systemic and renal hemodynamics, intestinal dysbiosis [cirrhosis dysbiosis ratio (CDR) as marker of dysbiosis], and direct renal vasoconstrictive response (renal vascular resistance (RVR) and glomerular filtration rate (GFR)) to cumulative doses of TNFα were measured in bile duct ligated (BDL)-cirrhotic ascitic mice. In SIRT1IEC-KO-BDL-ascitic mice, the worsening of intestinal dysbiosis exacerbates intestinal inflammation/barrier dysfunction, the upregulation of the expressions of intestinal/renal TNFα-related pathogenic signals, higher TNFα-induced increase in RVR, and decrease in GFR in perfused kidney. In intestinal SIRT1 knockout groups, the positive correlations were identified between intestinal SIRT1 activity and CDR. Particularly, the negative correlations were identified between CDR and RVR, with the positive correlation between CDR and GFR. In mice with advanced cirrhosis, the expression of intestinal SIRT1 is involved in the linkage between intestinal dysbiosis and vasoconstriction/hypoperfusion-related renal dysfunction through the crosstalk between intestinal/renal TNFα-related pathogenic inflammatory signals.


Introduction
Renal failure is a challenge in cirrhotic patients because the likelihood of mortality occurring increases with worsening renal function [1]. One of the most severe forms of renal dysfunction in advanced cirrhosis is hepatorenal syndrome (HRS), which is accompanied by systemic, intestinal, renal inflammation as well as renal vasoconstriction [2][3][4][5]. At present, the search for new potential agents to treat patients with cirrhotic HRS whose responses to standard treatments are poor is ongoing [6,7].
In advanced cirrhosis, persistent increased circulating tumor necrosis factor-α (TNFα)related bacterial translocation and systemic (hepatic, intestinal, renal) inflammation are involved in the development of intestinal/renal dysfunction and HRS [3,5]. In cirrhotic patients at diagnosis and resolution of the infection, the development of renal impairment is associated with significantly high plasma and ascitic fluid TNFα levels [8]. Agents with anti-TNFα effects can prevent the development of systemic inflammation and renal dysfunction in cirrhotic rats with portal hypertension [9,10].
In healthy rats, the infusion of high doses of TNFα lowers blood pressure, glomerular filtration rate and renal blood flow as well as directly induces renal vasoconstriction and increases renal vascular resistance [11,12]. Advanced cirrhotic patients with HRS are characterized by marked renal vasoconstriction, reduced renal blood flow and perfusion [2,4]. Increased circulating TNFα is involved in the pathogenesis of HRS, and higher plasma TNFα levels were observed in cirrhotic patients with HRS than in those without HRS [2,3,9,10,13]. TNFα inhibitors, such as pentoxifylline, can prevent and improve HRS in advanced cirrhotic patients [9,10].
In cirrhotic rats with ascites, anti-TNFα monoclonal antibody administration directly suppresses intestinal inflammation, reduces intestinal barrier dysfunction, and decreases the incidence of bacterial translocation [14,15]. Increased renal TNFα expression indicates the priming of cirrhotic kidneys by chronic intestinal inflammation, barrier dysfunction, and bacterial translocation [5,16]. Chronic intestinal decontamination with rifaximin, nonabsorbed gut-directed antibiotic, decreases the severity of acute kidney injury and HRS by reducing serum TNFα levels in patients with advanced cirrhosis [16]. However, side effects, such as nausea, vomiting, flatulence and abdominal pain, high cost, resistance, and the hepatotoxicity of anti-TNFα antibody limit the application of agents with anti-TNF effects in advanced cirrhotic patients with circulatory and renal dysfunction. Therefore, for cirrhotic patients with HRS, the identification of the upstream origin of systemic and local TNFα, to simultaneously treat intestinal inflammation, intestinal barrier dysfunction, intestinal dysbiosis, bacterial translocation, systemic inflammation and renal dysfunction is urgently needed.
In rats with DDS-induced colitis, SIRT1 activator pretreatment corrects colonic dysbiosis and reduces systemic and colonic mucosa inflammation [17]. SIRT1 RNA and protein expression are reduced in whole intestinal biopsies and the lamina propria mononuclear cells of patients with inflammatory bowel disease (IBD) [18]. In patients with IBD, treatment with infliximab, a chimeric monoclonal antibody against TNFα, restores the intestinal mucosal expression of SIRT1 [18]. Mice with intestinal deletion of SIRT1 (SIRT1 IEC-KO ) had abnormal activation of the TNFα pathway, gut dysbiosis and severe intestinal inflammation [19]. Pharmacological activation of intestinal SIRT1 attenuates TNFα-mediated intestinal barrier dysfunction, inflammation, and dysbiosis [20,21]. In Toxoplasma gondiiinfected mice, pretreatment with SIRT1 activator significantly suppresses TNFα levels in the ileum, mesenteric lymph nodes and spleen, preventing intestinal barrier dysfunction, reducing bacterial translocation and improving intestinal dysbiosis [21].
A recent study reported an increased risk of chronic kidney disease (CKD) in cases with inflammatory bowel disease [22]. In CKD, the suppression of intestinal bacteria dysbiosis significantly attenuates the severity of renal dysfunction [23]. In acute kidney injury model, anti-TNFα, antioxidant activities and acute pharmacologic activation of SIRT1 induces reno-vasodilatation, increases renal blood flow (RBF) and decreases RVR [24][25][26]. In diabetic nephropathy animals, chronic pharmacological activation of SIRT1 improved renal dysfunction by decreasing plasma and renal TNFα [27]. Systemic sirtuin 1 (SIRT1) activation reduces portal pressure by downregulating hepatic TNFα expression, inhibiting hepatic inflammation, and suppressing intrahepatic vasoconstriction [28]. Nonetheless, the contribution of the crosstalk between intestinal and renal TNF and SIRT1 signals in the severe renal dysfunction of cirrhosis has not yet been explored.
Accordingly, this study comprehensively explores the impacts of intestinal SIRT1 deficiency on TNFα-mediated intestinal inflammation, intestinal barrier dysfunction, intestinal bacterial dysbiosis, bacterial translocation on the development of severe renal dysfunction in cirrhotic mice with ascites.

Intestinal SIRT1 Deficiency Aggravates Severity of Renal Dysfunction in SIRT1 IEC-KO -BDL-Cirrhotic Mice with Ascites
Compared to WT-sham mice, the WT-BDL mice showed typical cirrhotic livers, ascites, upregulated hepatic TNFα-TNFR1 signals, significant circulatory dysfunction (reduced MAP, decreased CO and CI), elevated serum ALT, bilirubin and creatinine levels and hyponatremia, decreased urine output and decreased body weight ( Figure 1 and Table 1). In SIRT1 IEC-KO BDL-cirrhotic mice with ascites, the deletion of intestinal SIRT1 expression ( Figure 1B and Table 1) aggravated the abovementioned abnormalities observed in WT-BDL mice ( Figure 1 and Table 1). However, the serum ALT level and hepatic TNFR2 signals were not different between WT-BDL and SIRT1 IEC-KO BDL mice. Notably, the greater increase in jaundice was not associated with a greater increase in serum ALT level in SIRT1 IEC-KO -BDL mice compared to those in WT-BDL mice. The heart rate was not different among the four groups (Table 1). improved renal dysfunction by decreasing plasma and renal TNFα [27]. Systemic sirtuin 1 (SIRT1) activation reduces portal pressure by downregulating hepatic TNFα expression, inhibiting hepatic inflammation, and suppressing intrahepatic vasoconstriction [28]. Nonetheless, the contribution of the crosstalk between intestinal and renal TNF and SIRT1 signals in the severe renal dysfunction of cirrhosis has not yet been explored. Accordingly, this study comprehensively explores the impacts of intestinal SIRT1 deficiency on TNFα-mediated intestinal inflammation, intestinal barrier dysfunction, intestinal bacterial dysbiosis, bacterial translocation on the development of severe renal dysfunction in cirrhotic mice with ascites.

Intestinal SIRT1 Deficiency Aggravates Severity of Renal Dysfunction in SIRT1 IEC-KO -BDL-Cirrhotic Mice with Ascites
Compared to WT-sham mice, the WT-BDL mice showed typical cirrhotic livers, ascites, upregulated hepatic TNFα-TNFR1 signals, significant circulatory dysfunction (reduced MAP, decreased CO and CI), elevated serum ALT, bilirubin and creatinine levels and hyponatremia, decreased urine output and decreased body weight ( Figure 1 and Table 1). In SIRT1 IEC-KO BDL-cirrhotic mice with ascites, the deletion of intestinal SIRT1 expression ( Figure 1B and Table 1) aggravated the abovementioned abnormalities observed in WT-BDL mice ( Figure 1 and Table 1). However, the serum ALT level and hepatic TNFR2 signals were not different between WT-BDL and SIRT1 IEC-KO BDL mice. Notably, the greater increase in jaundice was not associated with a greater increase in serum ALT level in SIRT1 IEC-KO -BDL mice compared to those in WT-BDL mice. The heart rate was not different among the four groups (Table 1).

Genetic Deletion of Intestinal SIRT1 Exacerbates TNFα-Mediated Intestinal Inflammation and Barrier Dysfunction in SIRT1 IEC KO -BDL Mice
In WT-BDL mice with a higher intestinal TNFα level, higher ileal mucosal injury score, more fecal albumin loss, and worsening Evans blue-assessed barrier dysfunction were observed compared to WT-sham mice. Additionally, the upregulation of intestinal lipocalin-2, TNFR1, TNFR2, p-MLCK was associated with the downregulation of intestinal SIRT1, catalase, CuZnSOD, MnSOD, occludin, ZO-1 and E-cadherin expressions in WT-BDL mice ( Figure 2). Compared to WT-BDL mice, intestinal deletion of SIRT1 aggravates the abovementioned intestinal abnormalities in SIRT1 IEC KO -BDL-cirrhotic mice ( Figure 2). Nonetheless, the expression levels of intestinal TNFR2 and E-cadherin were not different between WT-BDL and SIRT1 IEC KO -BDL mice ( Figure 2E).

Genetic Deletion of Intestinal SIRT1 Exacerbates TNFα-Mediated Intestinal Inflammation and Barrier Dysfunction in SIRT1 IEC KO -BDL Mice
In WT-BDL mice with a higher intestinal TNFα level, higher ileal mucosal injury score, more fecal albumin loss, and worsening Evans blue-assessed barrier dysfunction were observed compared to WT-sham mice. Additionally, the upregulation of intestinal lipocalin-2, TNFR1, TNFR2, p-MLCK was associated with the downregulation of intestinal SIRT1, catalase, CuZnSOD, MnSOD, occludin, ZO-1 and E-cadherin expressions in WT-BDL mice ( Figure 2). Compared to WT-BDL mice, intestinal deletion of SIRT1 aggravates the abovementioned intestinal abnormalities in SIRT1 IEC KO -BDL-cirrhotic mice (Figure 2). Nonetheless, the expression levels of intestinal TNFR2 and E-cadherin were not different between WT-BDL and SIRT1 IEC KO -BDL mice ( Figure 2E).

BDL-Cirrhotic Mice due to the Deficiency of Intestinal SIRT1
In the in situ renal perfusion study of WT-BDL mice, the lower basal MAP, GFR, RBF and the higher RVR were noted compared to those of WT-sham mice ( Figure 4). Notably, the magnitude (percentage change) of the TNFα-induced decrease in MAP was not different between the four groups (WT-BDL-cirrhotic and SIRT1 IEC-KO -BDL-cirrhotic mice) ( Figure 4A). A higher degree of TNFα-induced increase in the RVR and the decrease in GFR and RBF were observed among SIRT1 IEC KO -BDL-cirrhotic ascitic mice than in WT-BDL-cirrhotic ascitic mice ( Figure 4B-D). However, the degree of TNFα-induced changes in GFR, RVR and RBF did not differ between WT-sham and SIRT1 IEC KO -sham mice.

BDL-Cirrhotic Mice due to the Deficiency of Intestinal SIRT1
In the in situ renal perfusion study of WT-BDL mice, the lower basal MAP, GFR, RBF and the higher RVR were noted compared to those of WT-sham mice ( Figure 4). Notably, the magnitude (percentage change) of the TNFα-induced decrease in MAP was not different between the four groups (WT-BDL-cirrhotic and SIRT1 IEC-KO -BDL-cirrhotic mice) ( Figure 4A). A higher degree of TNFα-induced increase in the RVR and the decrease in GFR and RBF were observed among SIRT1 IEC KO -BDL-cirrhotic ascitic mice than in WT-BDL-cirrhotic ascitic mice ( Figure 4B-D). However, the degree of TNFα-induced changes in GFR, RVR and RBF did not differ between WT-sham and SIRT1 IEC KO -sham mice.

Genetic Deletion of Intestinal SIRT1 Aggravates the Intestinal Bacterial Dysbiosis in SIRT1 IEC KO -Mice
The Shannon and Faith's Phylogenetic Diversity (PD) index, and evenness index were significantly lower in the WT-BDL mice and SIRT1 IEC KO -BDL mice than in the WT-sham mice ( Figure 5A and supplement Figure S1A-B). These results indicate that the overall microbial species diversity was lower in the WT-BDL mice and the SIRT1 IEC KO -BDL mice than in the WT-sham mice. The UniFrac principal coordinate analysis (PCoA) is shown in Figure 5B. The WT-sham and WT-BDL groups were separated into different clusters (PER-MANOVA, P = 0.001, Q valve = 0.006). Likewise, the microbiota of the SIRT1 IEC KO -BDL group was clustered separately from that of WT BDL group (PERMANOVA, PER-MANOVA, P = 0.003, Q valve = 0.008).

Genetic Deletion of Intestinal SIRT1 Aggravates the Intestinal Bacterial Dysbiosis in SIRT1 IEC KO -Mice
The Shannon and Faith's Phylogenetic Diversity (PD) index, and evenness index were significantly lower in the WT-BDL mice and SIRT1 IEC KO -BDL mice than in the WT-sham mice ( Figure 5A and Supplement Figure S1A,B). These results indicate that the overall microbial species diversity was lower in the WT-BDL mice and the SIRT1 IEC KO -BDL mice than in the WT-sham mice. The UniFrac principal coordinate analysis (PCoA) is shown in Figure 5B. The WT-sham and WT-BDL groups were separated into different clusters (PER-MANOVA, p = 0.001, Q valve = 0.006). Likewise, the microbiota of the SIRT1 IEC KO -BDL group was clustered separately from that of WT BDL group (PERMANOVA, PERMANOVA, p = 0.003, Q valve = 0.008).
Notably, the Taxa bar plot (Supplementary Figure S1D-H) of the 10 most abundant taxa shows the same trends as the chord diagram. The chord diagram indicates that the intestinal microbiota of the WT-BDL mice at the phylum level were characterized by a decrease in anti-inflammatory Firmicutes bacteria and an increase in proinflammatory Deferribacteres and Epsilobacteraeota bacteria compared to those in WT-sham mice ( Figure 5C). In SIRT1 IEC KO -BDL mice, an increase in proinflammatory Bacteroidetes bacteria compared to in WT-BDL mice was observed. The decreased in abundance of Firmicutes was associated with an increase in proinflammatory Proteobacteria bacteria in SIRT1 IEC KO -sham mice.
At the class and order levels, the intestinal microbiota of WT-BDL mice were characterized by a decrease in the abundance of anti-inflammatory bacteria Clostridia and Clostridiales (belong to Firmicutes phylum) and the increased proinflammatory bacteria Campylobacteria, Deferribacteres, Defferibacteres (belong to Deferribacteres phylum) compared WT-sham mice (Supplement Figure S2). Compared to WT-BDL mice, a further increase in the abundance of Bacteroidia, Bacteroidales (belong to Bacteroidetes phylum) and Defferibacterales was observed in SIRT1 IEC-KO -BDL mice. Notably, the presence of proinflammatory Gammaproteobacteria and Enterobacteriales (belong to proteobacteria phylum) bacteria in intestines of SIRT1 IEC KO -sham mice was observed (Supplement Figure S2). At the family and genus levels, in comparison with WT-sham mice feces, the absence of anti-inflammatory bacteria [Lactbacillaceae (family)/Lactobacillus (genus)] was associated with increase in proinflammatory bacteria (Muribaculaceae (family, belong to Deferribacteres phylum), Mucispinillum (genus, belong to Deferribacteres phylum), Helicobaraceae) in WT-BDL mice feces (Supplement Figure S2). Notably, SIRT1 IEC KO -BDL mice were characterized by the presence of proinflammatory bacteria (Peptostrepococcaceae (family, belong to Clostridiaes order), Bacteroidaceae (family), Romboutsia (genus, belong to Peptostreptococcaceae family), Bacteroides (genus), and Parabacteroides (genus), Muribaculaceae (family, belong to Deferribacteres phylum)) compared to the lack of these in WT-BDL mice feces (Supplement Figure S2). In comparison with WT-sham mice, SIRT1 IEC KO -sham mice were characterized by higher levels of intestinal Enterbacteriaceae and lower levels of Lactbacillaceae at the family level; as well as the presence of Escherichia-shigella, and Parabacteroides, and Citrobacter at genus level (Supplement Figure S2). Notably, the Taxa bar plot (Supplementary Figure S1D-H) of the 10 most abundant taxa shows the same trends as the chord diagram. The chord diagram indicates that the intestinal microbiota of the WT-BDL mice at the phylum level were characterized by a decrease in anti-inflammatory Firmicutes bacteria and an increase in proinflammatory Deferribacteres and Epsilobacteraeota bacteria compared to those in WT-sham mice ( Figure 5C). In SIRT1 IEC KO -BDL mice, an increase in proinflammatory Bacteroidetes bacteria compared to in WT-BDL mice was observed. The decreased in abundance of Firmicutes was associated with an increase in proinflammatory Proteobacteria bacteria in SIRT1 IEC KO -sham mice.
At the class and order levels, the intestinal microbiota of WT-BDL mice were characterized by a decrease in the abundance of anti-inflammatory bacteria Clostridia and Clostridiales (belong to Firmicutes phylum) and the increased proinflammatory bacteria Campylobacteria, Deferribacteres, Defferibacteres (belong to Deferribacteres phylum) compared WT-sham mice (supplement Figure S2). Compared to WT-BDL mice, a further increase in the abundance of Bacteroidia, Bacteroidales (belong to Bacteroidetes phylum) and Defferibacterales was observed in SIRT1 IEC-KO -BDL mice. Notably, the presence of proinflammatory Gammaproteobacteria and Enterobacteriales (belong to proteobacteria phylum) bacteria in intestines of SIRT1 IEC KO -sham mice was observed (supplement Figure S2). At the family and In WT-sham mice, both the cladogram (Supplement Figure S3) and the linear discriminant analysis (LDA) effect size method (LAD score > 3; Figures 6 and 7) revealed an increase in beneficial bacteria (Clostridia, Clostridiales, Lachnospiraceae, Lactobacillale, Lactobacillaceae (belong to Defirmicutes), short chain fatty acid producer Prevotellaceae (belong to order bacteroidales), Ruminococcaceae UCG and Ruminococcus1 and Ruminiclostridium 6) that have involvement in the integrity of intestinal epithelial barrier. In contrast, the intestinal microbiota of WT-BDL mice were characterized by an increase in pathogenic bacteria (Rikenellaceae (belong to Bacteroidetes rhylum), Alistipes (belong to Rikenellaceae family), Parasutterella (belong to Proteobacteria phylum), Defferibacterales, Deferribacteres, Deferribacteraceae, and Mucispirillum (belong to Deferribacteres phylum) that compromise barrier function (Figures 6 and 7). long to Bacilli class), Streptococcus (belong to Streptococcaceae family and Bacilli class), Actinobacteria, Eggerthellaceae, Coriobacteriales, Coriobacteria (belong to Actinobacteria phylum) that compromise the barrier function (Figures 6 and 7).

Intestinal Bacterial Dysbiosis was Associated with the Upregulation of Renal Inflammatory and Barrier-Disrupted Markers in SIRT1 IEC-KO Mice
Compared to WT groups, a significant positive correlation was noted between mRNA/protein levels of renal inflammatory mediators (lipocalin-2 and TNFR1) and the abundance of barrier-disrupted pathogenic intestinal bacteria (Epsilonbacteraeota (phylum) and Campylobacteria (class)) in KO groups ( Figure 8A,B). In comparison with WT groups, the negative correlation between mRNA/protein levels of renal inflammatory mediators Notably, SIRT1 IEC KO -BDL mice were characterized by an increase in intestinal inflammatory bacteria (Bacteroides, Bacteroidaceae, Peptostreptococcaceae (belong to Clostridiales order), Romboutsia (belong to family of Peptostreptococcaceae), Ruminococcaceae, Ruminiclostridium 6, Blautai and Ruminiclostridium (belong to Clostridia class), Delatproteobacteria, Desulfovibrio, Desulfavibrionales (belong to Deltaproteobacteria phylum), Streptococcaceae (belong to Bacilli class), Streptococcus (belong to Streptococcaceae family and Bacilli class), Actinobacteria, Eggerthellaceae, Coriobacteriales, Coriobacteria (belong to Actinobacteria phylum) that compromise the barrier function (Figures 6 and 7).

Intestinal Bacterial Dysbiosis was Associated with the Upregulation of Renal Inflammatory and Barrier-Disrupted Markers in SIRT1 IEC-KO Mice
Compared to WT groups, a significant positive correlation was noted between mRNA/protein levels of renal inflammatory mediators (lipocalin-2 and TNFR1) and the abundance of barrier-disrupted pathogenic intestinal bacteria (Epsilonbacteraeota (phylum) and Campylobacteria (class)) in KO groups ( Figure 8A,B). In comparison with WT groups, the negative correlation between mRNA/protein levels of renal inflammatory mediators (lipocalin-2 and TNFR1) and the abundance of barrier-protected intestinal bacteria (Bacilli (class), Saccharimonadia (class), Erysipelotrichia (class), Verrucomicrobia (class) was noted in KO groups. Significantly, there was positive correlation between the mRNA/protein levels of anti-inflammatory (catalase, MnSOD and CuZnSOD)/barrier (occludin and Eadhesin) markers and the abundance of barrier-protected intestinal bacteria (Bacilli (class), Erysipelotrichia (class), Verrucomicrobia (class)). Additionally, there was a negative correlation between the mRNA/protein levels of anti-inflammatory (catalase, MnSOD, and CuZn-SOD)/barrier (occludin and E-adhesin) markers and the abundance of barrier-disrupted intestinal bacteria (Saccharimonadia (class), Episolbacteraeata, Campylobacteria) in KO groups.
This correlation analysis was performed to explore the crosstalk between the abundance of intestinal bacteria (with inflammatory, anti-inflammatory, barrier-protected and barrierdisrupted effects), protein/mRNA levels of intestinal/renal injured, anti-inflammatory, barrier-protected, and barrier-disrupted markers. Notably, the trend of the correlation between the abundance of intestinal microbiota across phylum, class and order levels, and the levels of intestinal pathogenic/protective markers (Supplement Figure S4A,B) were similar to those between intestinal microbiota and renal pathogenic/protective markers ( Figure 9A,B). Moreover, the significant correlations were noted between the levels of intestinal and renal protective (ZO-1, occludin, E-adhesin, catalase, MnSOD, and CuZn-SOD) as well as pathogenic markers (TNFR1, TNFR2, MLKC, p-MLKC, lipocalin-2 and KIM-1) ( Figure 9A,B).

Figure 8. Crosstalk between intestinal and renal pathogenic markers in bile duct ligation (BDL) mice. (A,B) Significant correlation was noted between renal injury, renal inflammatory, anti-inflammatory, barrier markers and intestinal microbiota in WT groups (WT-BDL and WT-sham) and KO groups (SIRT1 IEC-KO -BDL and SIRT1
IEC-KO -sham). "Green" lines represent positive correlation between the abundance of pathogenic proteins and mRNA levels of corresponding pathogenic markers whereas "red" lines indicate the negative correlation between the abundance of the bacteria with protein and mRNA levels of corresponding pathogenic markers.

Figure 8. Crosstalk between intestinal and renal pathogenic markers in bile duct ligation (BDL) mice. (A,B) Significant correlation was noted between renal injury, renal inflammatory, anti-inflammatory, barrier markers and intestinal microbiota in WT groups (WT-BDL and WT-sham) and KO groups (SIRT1 IEC-KO -BDL and SIRT1
IEC-KO -sham). "Green" lines represent positive correlation between the abundance of pathogenic proteins and mRNA levels of corresponding pathogenic markers whereas "red" lines indicate the negative correlation between the abundance of the bacteria with protein and mRNA levels of corresponding pathogenic markers.

Direct Link between Intestinal Bacterial Dysbiosis and Severe Renal Dysfunction in SIRT1 IEC-KO -BDL Mice
Compared to WT-sham mice, higher serum LPS binding protein (LBP) levels (Table  1), higher MLN culture positive rates (Table 1), and lower cirrhosis dysbiosis ratios (CDR) ( Figure 10A) were observed in WT-BDL mice. Notably, the deletion of the intestinal SIRT1

Direct Link between Intestinal Bacterial Dysbiosis and Severe Renal Dysfunction in SIRT1 IEC-KO -BDL Mice
Compared to WT-sham mice, higher serum LPS binding protein (LBP) levels (Table 1), higher MLN culture positive rates (Table 1), and lower cirrhosis dysbiosis ratios (CDR) ( Figure 10A) were observed in WT-BDL mice. Notably, the deletion of the intestinal SIRT1 gene aggravated the degree of increase in serum LBP level and MLN culture positive rate, as well as the decrease in intestinal CDR in SIRT1 IEC-KO -BDL mice compared to those in WT-BDL mice. Figure 10B shows a positive correlation between data of the intestinal CDR and SIRT1 activity of WT and KO mice. The correlation was stronger between KO than WT groups, and significance was only noted between KO groups. Significantly, there was a negative correlation between the level of intestinal SIRT1 activity and baseline RVR in KO groups ( Figure 10C). In the KO groups, the significant negative correlation between data of the intestinal CDR and the data of RVR was parallel to the positive correlation between the data of intestinal CDR and the data of the GFR (Figure 10D,F). In both the WT and KO groups, the correlation between intestinal SIRT1 activity and the data of GFR did not reach statistical significance ( Figure 10E).

Discussion
Pharmacologic and genetic depletion of systemic TNFα can suppress aging-associated intestinal and systemic inflammation and normalize intestinal dysbiosis [19]. The preservation of the gut microbiota in older TNF −/− mice indicates the important pathogenic roles of TNFα in age-related intestinal dysbiosis, systemic inflammation and intestinal barrier dysfunction [29]. Increased abundance of Rikenellaceae family is associated with the exacerbation of intestinal inflammation [30]. In cirrhotic patients, intestinal inflammation and barrier dysfunction, overgrowth of the proinflammatory intestinal bacteria, such as Enterobacteriaceae, Veillonellaceae, and Streptococcaceae) and decreased anti-inflammatory

Discussion
Pharmacologic and genetic depletion of systemic TNFα can suppress aging-associated intestinal and systemic inflammation and normalize intestinal dysbiosis [19]. The preser-vation of the gut microbiota in older TNF −/− mice indicates the important pathogenic roles of TNFα in age-related intestinal dysbiosis, systemic inflammation and intestinal barrier dysfunction [29]. Increased abundance of Rikenellaceae family is associated with the exacerbation of intestinal inflammation [30]. In cirrhotic patients, intestinal inflammation and barrier dysfunction, overgrowth of the proinflammatory intestinal bacteria, such as Enterobacteriaceae, Veillonellaceae, and Streptococcaceae) and decreased anti-inflammatory Lachnospiraceae bacteria abundance are correlated with the severity of cirrhosis [29][30][31][32]. The exogenous administration of Akkermansia attenuates metabolic endotoxemia-induced inflammation through restoration of the gut barrier [33]. In this study, the anti-inflammatory intestinal bacteria, including Lachnospiraceae and Akkermansia, were dominant in sham mice, whereas the proinflammatory intestinal bacteria, including Streptococcaceae and Rikenellaceae, were dominant in WT-BDL and SIRT1 IEC KO -BDL mice with severe intestinal barrier dysfunction. The cirrhosis dysbiosis ratio (CDR), which represents the balance between proinflammatory and anti-inflammatory intestinal bacteria, is important for prediction of cirrhosis-related intestinal and renal complications [31,32].
In particular, this study revealed a significant positive correlation between the CDR and intestinal SIRT1 activity, between the CDR and glomerular filtration rate, as well as a negative correlation between CDR and renal vascular resistance. This emphasizes the involvement of intestinal proinflammatory and anti-inflammatory bacteria in the intestinal and renal dysregulation of cirrhosis (Supplement Figure S5). TNFα causes intestinal tight junction disruption and subsequent intestinal barrier dysfunction via myosin light chain kinase (MLCK) activation [34,35]. The upregulation of intestinal TNFα-TNFR signals and phosphorylation of intestinal MLCK is involved in the pathogenesis of intestinal barrier dysfunction and disease progression in cirrhosis [36]. In SIRT1 IEC-KO -BDL-cirrhotic mice, the depleted intestinal SIRT1-related upregulation of TNFα-TNFR signals are associated with the upregulation of intestinal p-MLCK expression and disruption of intestinal barrier integrity. In the cirrhotic mice, there is a positive correlation between the expression levels of intestinal p-MLCK and renal lipocalin-2 (inflammatory and renal injury marker) as well as intestinal p-MLCK and renal KIM-1.
Upon inflammation and tissue damage, in addition to crosstalk in the intestine, activation of renal SIRT1 attenuates the detrimental effects of TNFα on renal mesangial and tubular epithelial cells [19][20][21][24][25][26][27]. The deficiency of intestinal SIRT1 exacerbates TNFαmediated renal damage in mice with cholestasis [37]. In a nephropathy model, systemic SIRT1 activation suppresses intestinal/renal TNFα expression and ameliorates renal dysfunction [27]. In this study, the elevation of serum bilirubin was associated with increased creatinine level in SIRT1 IEC-KO -BDL-cirrhotic mice compared to WT BDL-cirrhotic mice ( Figure 1E,F). In addition to supporting the previous observation regarding the crosstalk between hepatic SIRT1 and TNFα [28], this study suggests a link between SIRT1−TNFα crosstalk and intestinal SIRT1-related effects and renal dysfunction of cirrhosis.
In response to inflammation and tissue injury, TNFα directly induces the production of lipocalin-2 from intestinal and renal epithelial cells [38]. Lipocalin-2 is an inflammatory mediator for intestinal and renal inflammation and is positively correlated with inflammatory disease severity [38][39][40]. Serum lipocalin-2 are positively correlated with serum levels of soluble TNFR and negatively correlated with glomerular filtration rate [40]. Renal expression of lipocalin-2 is significantly increased in decompensated cirrhosis with increased circulating TNFα and acute renal injury [40,41]. In this study, there was a positive correlation between the abundance of proinflammatory bacteria and intestinal/renal lipocalin/TNFα/TNFR expression in the WT and KO groups (Supplement Figure S4A,B). Accordingly, these observations reinforced the existence of inflamed and leaky intestinedriven renal dysfunction in SIRT1 IEC-KO -BDL-cirrhotic mice.
Chronic kidney disease (CKD) is characterized by the accumulation of metabolites of proinflammatory gut bacteria. These can damage renal tubular cells by increasing cellular oxidative stress and deteriorating renal function [42,43]. In CKD, the suppression of intestinal bacteria dysbiosis by the consumption of a high-fiber diet significantly attenuates the disruption of the colonic epithelial tight junction and severity of renal dysfunction [23]. In patents with CKD, there is an increased abundance of proinflammatory enterobacteriaceae and Clostridiaceae bacteria and decreased abundance of anti-inflammatory butyrateproducing bacteria (Lactobacillaceae and Prevotellaceae) [43]. In the present study, intestinal inflammation and renal dysfunction in WT-BDL and SIRT1 IEC-KO BDL-cirrhotic mice were characterized by increased intestinal abundance of Enterobacteriaceae and Clostridiaceae as well as decreased abundance of Lactobacillaceae and Prevotellaceae.
In the model of murine colitis, TNFR1 is highly expressed in intestinal epithelial cells and mediates TNFα induced inflammatory cascades [42,44]. In the present study, the SIRT1related modulation of intestine TNFα expression primarily occurs through the upregulation of TNFR1 expression in the cirrhotic intestine. Meanwhile, TNFR1 is highly expressed in the renal proximal tubule and collecting duct, TNFR1 activation reduces renal blood flow (RBF) and glomerular filtration rate (GFR) [45,46]. In patients with diabetic kidney disease, serum TNFR levels are positively associated with estimated glomerular filtration rate (eGFR) decline and disease severity [47,48]. In SIRT1 IEC-KO BDL-cirrhotic mice the detrimental renal effects of intestinal SIRT1 deficiency, including further decrease in RBF and GFR, occurred through the simultaneous upregulation of intestinal and renal TNFα-TNFR1 cascades (Supplement Figure S5). Additionally, there were significant positive correlations between oxidative stress markers (intestinal and renal MnSOD, CuZn-SOD, catalase), inflammation marker (intestinal and renal TNFR1/TNFR2), and intestinal/renal dysfunction markers (p-MLCK, MLCK, lipocalin-2 and KIM).

Conclusions
This study reinforced the concepts of increased TNFα-mediated renal oxidative stress and inflammation as the main culprits for the deterioration of renal function in cirrhotic patients. In cirrhotic ascitic mice, the well-expressed intestinal SIRT1 is crucial to prevent TNFα-mediated activation of systemic, intestinal and renal oxidative stress, inflammation and injured signals that contribute to the development of renal dysfunction. Agents that can restore intestinal SIRT1 have the potential to improve the inflammation-derived TNFαmediated renal dysfunction in advanced cirrhosis. So, it is noteworthy to explore the effects of the administration of a highly absorbed intestinal SIRT1 activator such as resveratrol on the coexisting TNFα-mediated intestinal and renal pathogenic changes as well as renal dysfunction in cirrhosis.

Animals
Animals used in this study were 10 weeks-old male mice with an intestinal epithelial specific deletion (knockout, KO) of the SIRT1 gene (SIRT1 IEC-KO ) with a 99% C57BL/6J genetic background. To generate SIRT1 IEC-KO mice, SIRT1 flox/flox mice were crossed with transgenic mice expressing Cre recombinase under the control of the villin promoter (VillinCre−), which is expressed in intestinal epithelial cells (IEC) purchased from Charles River Japan, Inc. (Yokohama, Japan). Wild type C57BL/6 mice were served as controls (WT). The experiments were approved by the animal ethical committee of Yang-Ming medical university with approval of No. 1061232r and IACUC 2018-54. All the raising and breeding of animals is undertaken in the laboratory animal center of National Yang-Ming University. All the experiments were performed in our laboratory. All efforts were made to minimize animal suffering by administering inhalation anesthetics (isoflurane). At the end of the experiments, the mice were euthanized with the inhalation anesthetic by overdose of isoflurane.

Groupings
Common bile duct ligation (BDL) was undergone on ten week-old WT and SIRT1 IEC-KO mice to create severe renal dysfunction in advanced cirrhosis [54]. Six weeks after BDL, ∼ =15% of BDL mice died. The presence of ascites was evaluated by ultrasound and 85% of BDL mice displayed ascites at the time of sacrifice. This created 4 experimental groups namely WT sham, WT-BDL, SIRT1 IEC-KO -sham, and SIRT1 IEC-KO -BDL mice (n = 7, in each group) and were included for serial experiments.

Common Measurements between Two Sets of Mice
At the end point (week 6 after BDL) of the study, mouse was placed in a metabolic cage and 24-h urine samples were collected over 5 consecutive days, and the average of 5-day daily urine output (mL/100 g BW) was calculated. The supernatant of collected 3-day urine samples was used for measuring of urinary concentration of creatinine, and renal tubular epithelial damage markers (urinary levels of uKIM-1/creatinine (ng/mg u.cr) and urinary levels of uNGAL/creatinine ratio (ng/g tissue).

Systemic and Renal Hemodynamic Measurements
For the first set of 4 groups of mice (n = 7 in each group), cardiac output (CO), mean arterial pressure (MAP), and heart rate (HR) were measured. CO was normalized to body weight and represented as cardiac index (CI). Then, a midline incision was made in the abdomen. Whole kidney blood flow (RABF, mL/min.100 g) of the left and right kidney was measured. Renal vascular resistance (RVR) was calculated as the RABF divided by the MAP. Finally, tissues (liver, intestine, and kidney), mesenteric lymph nodes (MLN), stool and blood samples were collected for measurements of intestinal bacterial microbiota, intestine TNFα level, and plasma LPS-binding protein [LBP, BT marker, using ToxinSensor Chromogenic LAL Endotoxin Assay Kit (GenScript USA Inc., Piscataway, NJ, USA)].

Measurements of Intestinal Permeability and Intestinal Inflammation
Evan Blue (EB)-permeated intestinal permeability methods were used to assess the degree of intestinal barrier dysfunction as previously described [55]. Additionally, intestinal permeability was reassessed by measurement of albumin content in the mice feces using ELISA kits (MyBioSource, Inc., San Diego, CA, USA). The data was normalized to the total weight of feces. For measurement of intestinal inflammation, the terminal ileum lumen was carefully cannulated, gently washed and embedded for staining with hematoxylene and eosin (H&E). Injury was classified using a semiquantitative grading system as shown in Supplement Table S1.

Measurements of Tissue Profiles
Proteins/mRNAs expressions of hepatic, intestine, renal inflammatory, oxidative stress, antioxidant, intestinal and renal injury as well as barrier markers were measured with appropriate antibodies/primers (Table 2). Tissue levels of SIRT1 activities were measured with SIRT1 fluorometric Kit (Abcam, relative fluorescence unit, RFU). Hepatic and renal collagen deposition and renal tubulointerstitial injury were measured with Sirius red and periodic acid-Schiff (PAS) staining. The average of the results of samples from each mouse were included for comparison.  [18,20]. To obtain further insights into the pathogenic effects of genetic deletion of intestinal SIRT1 on intestinal bacterial microbiome, we used metagenomic sequencing of the 16S rRNA gene. The library for 16S rDNA amplicon sequencing was constructed based on the PCRamplified V3-V4 variable regions. Alpha diversity (Shannon diversity, Faiths PD index, evenness index, and observed OTU) was calculated. The UniFrac principal coordinate analysis (PCoA), which evaluates phylogenetic similarities between microbial communities, was calculated for the beta-diversity.

Direct Correlation between Intestinal Dysbiosis and Renal Dysfunction of Cirrhotic Mice
Correlation network between the abundance of significant bacteria and expression levels of intestinal and renal pathogenic markers were visualized by generating two interactive networks specific to the WT (WT-sham and WT-BDL mice) and KO (SIRT1 IEC KO sham and SIRT1 IEC KO BDL mice) groups and analyzed using Cytoscape and CoNet. [56,57]. Additionally, the ratio of autochthonous (inhabiting a place or region from earliest time, Lachnospiraceae + Ruminococcaceae + Veillonellaceae + Clostridialies XIV) to non-autochthonous taxa (Bacteroidaceae + Enterobacteriaceae) was calculated as the cirrhosis dysbiosis ratio (CDR). Notably, the abovementioned autochthonous taxa on the numerator of CDR are bacteria that can reduce colonic inflammation and nourish colonocytes, compete with pathogenic bacteria for nutrients, avoid intestinal barrier dysfunction and reduce bacterial translocation [31,32]. It had been reported that low CDR is associated with endotoxemia, infection, death and organ failures within 30 days in cirrhotic patients [31,32]. In our study, the CDR of each mouse was calculated and correlated with the intestinal SIRT1 activity, the RVR data and the GFR data of individual group.

Experiments in the Second set of Mice
In in situ renal perfusion study, the dose-response of MAP, GFR and RVR to the incremental doses of TNFα (0.1, 0.3, and 0.5 ng/g/min) were measured in the second set of 4 groups of mice (n = 7 in each group). Two consecutive 30-min urine (basal period) and corresponding blood were collected for measuring plasma and urine concentrations of inulin, PAH and hematocrit. Then, the value for inulin clearance was considered as glomerular filtration rate (GFR) with formula of U inulin /P inulin *UV (U inulin /P inulin : urine and plasma levels of inulin, UV: urine volume in the given time period), and the value for PAH clearance was considered as renal plasma flow with formula of U PAH /P PAH *UV (U PAH /P PAH : urine and plasma levels of PAH, UV: urine volume in the given time period). Renal blood flow (RBF) was calculated from renal plasma flow and hematocrit values. Renal vascular resistance (RVR) was calculated by dividing the value of MAP with the value of RBF. The mean values obtained during the first two control collection periods were considered as "basal values" while the mean of the values collected during the two TNFα infusion periods was named as the "treatment value". The differences in the values between the basal and the treatment periods were considered as the responses to TNFα treatment.

Statistical Analysis
Data were expressed as means ± S.D. Statistical significance for each group was determined using unpaired Student's t test, one-way ANOVA, Newman−Keuls test, Mann−Whitney U-tests, or Wilcoxon signed rank test. Correlation analysis between the representative microbial genera, intestinal, and renal injury markers were analyzed with Pearson's correlation coefficient and significant associations with p < 0.05 and r > 0.5 are shown.