Hepatic PPARα function and lipid metabolic pathways are dysregulated in polymicrobial sepsis

Abstract

Despite intensive research and constant medical progress, sepsis remains one of the most urgent unmet medical needs of today. Most studies have been focused on the inflammatory component of the disease; however, recent advances support the notion that sepsis is accompanied by extensive metabolic perturbations. During times of limited caloric intake and high energy needs, the liver acts as the central metabolic hub in which PPARα is crucial to coordinate the breakdown of fatty acids. The role of hepatic PPARα in liver dysfunction during sepsis has hardly been explored. We demonstrate that sepsis leads to a starvation response that is hindered by the rapid decline of hepatic PPARα levels, causing excess free fatty acids, leading to lipotoxicity, and glycerol. In addition, treatment of mice with the PPARα agonist pemafibrate protects against bacterial sepsis by improving hepatic PPARα function, reducing lipotoxicity and tissue damage. Since lipolysis is also increased in sepsis patients and pemafibrate protects after the onset of sepsis, these findings may point toward new therapeutic leads in sepsis.

Keywords: fibrates; lipid metabolism; lipotoxicity; liver; sepsis.

Figure 1. Hepatic PPARα signaling is disturbed at a genome‐wide level during sepsis

1. A–D

   RNA‐seq of liver 10 h post‐sham or CLP. Mice (n = 3/group) underwent a sham or CLP operation and were injected with GW7647 (10 μg/g) 6 h post‐surgery, and after 4 h (total of 10 h), livers were isolated and RNA was prepared. (A) Scatter plot showing log fold change (LFC) of all GW7647‐upregulated genes (LFC > 0.8 and P < 0.05) in sham versus their LFC 10 h after CLP. The red line represents the diagonal, and the black line represents the real slope (0.3115) of the data. (B) Venn diagram depicting the amounts of genes upregulated (up) or downregulated (dn) by GW7647 in sham and CLP mice (LFC > 0.8 or < −0.8 and P < 0.05). (C) Top enriched gene ontology (GO) terms for genes that are downregulated in CLP mice without stimulation compared to unstimulated sham controls (LFC < −0.8 and P < 0.05). Composite of 3 datasets: CLP1 (6 h after CLP), CLP2 (8 h after CLP), and CLP3 (10 h after CLP). Analysis was performed with the Enrichr tool. (D) Heat map of differentially expressed genes in sham mice after GW7647 treatment, involved in β‐oxidation of fatty acids (unit scale bar = log₂ of the normalized counts).

2. E, F

   Confirmation of RNA‐seq data via qPCR on pure hepatocytes isolated via flow cytometry‐based sorting (n = 3/group). (E) Ppara and (F) Hmgcs2 mRNA expression is shown as relative expression, normalized to housekeeping genes Hprt and Rpl. P‐values were calculated using 2‐way ANOVA analysis. Central lines represent mean.

Source data are available online for this figure.

Figure EV1. Hepatic PPARα signaling is disturbed at a genome‐wide level during sepsis

1. A

   RNA‐seq of liver 10 h post‐sham or CLP. Mice (n = 3/group) underwent a sham or CLP operation and were injected with GW7647 (10 μg/g) 6 h post‐surgery, and after 4 h (total of 10 h), livers were isolated and RNA was prepared. Top enriched gene ontology (GO) terms for genes that are specifically upregulated by GW7647 in CLP mice. Analysis was performed with the Enrichr tool.

2. B–F

   Confirmation of RNA‐seq data via qPCR on pure hepatocytes isolated via flow cytometry‐based sorting (n = 3/group). (B) Acsl1, (C) Cpt1a, (D) Cpt2, (E) Ehhadh, and (F) Slc25a20 mRNA expression are shown as relative expression, normalized to housekeeping genes Hprt and Rpl. P‐values were calculated using 2‐way ANOVA analysis. Central lines represent mean.

Figure 2. PPARα levels are decreased in the liver during sepsis and correlate with disease severity

1. A

   Mice (n = 4/group, data are representative of two experiments) underwent a sham (with or without starvation) or CLP operation, and liver was isolated on several timepoints post‐surgery for RNA preparation and qPCR. Ppara mRNA expression is shown as relative expression, normalized to housekeeping genes Hprt and Rpl. P‐values were calculated using 1‐way ANOVA analysis. Central lines represent mean.

2. B

   PPARA protein levels were analyzed in livers 24h after sham or CLP by Western blot using actin as a loading control.

3. C

   Quantification of PPARA Western blot. P‐value was calculated via 2‐way Student's t‐test (n = 4/group). Central line represents mean; error bars represent mean ± SEM.

4. D

   Pearson correlation between log fold change (LFC) Ppara expression levels and body temperature 24 h post‐sepsis (n = 38, r = 0.6875, combined data of four independent experiments).

5. E

   Liver Hmgcs2 mRNA expression at different timepoints post‐sepsis, expression is shown as relative expression, normalized to housekeeping genes Hprt and Rpl. P‐values were calculated using 1‐way ANOVA analysis (n = 4/group, data are representative of two experiments). Central lines represent mean.

6. F

   Oxygen consumption rates (OCRs) of liver tissue explants 24 h post‐sham or CLP. Liver tissue was isolated 24 h post‐surgery, and OCR was measured via Seahorse with BSA or palmitic acid (PA) as a substrate for 42 min. n = 4/group. Central line represents mean; error bars represent mean ± SEM.

7. G

   Visualization of T_42(min) OCR, P‐values were calculated using 2‐way ANOVA analysis. Central lines represent mean.

Source data are available online for this figure.

Figure 3. Lipolysis of subdermal and visceral fat pads is enhanced during sepsis

1. A–C

   Percentage of body weight that is taken up by the inguinal fat pad (iWAT, A), the mesenteric fat pad (mWAT, B), and the perirenal fat pad (pWAT, C) 24 h post‐surgery in sham (with or without starvation) and CLP mice. P‐values were calculated with a 1‐way ANOVA test. Combined data of two independent experiments, n = 6/group. Central lines represent mean.

2. D, E

   Visual representation of (D) iWAT and (E) pWAT sizes in sham, sham‐starved, and CLP mice 24 h post‐surgery. Data are representative of three independent experiments.

3. F–I

   Mice (n = 6–10/group) underwent a sham (with or without starvation) or CLP operation, and blood was collected 6 h and 24 h post‐surgery. Plasma was isolated, and (F) FFA concentration and (G) glycerol concentration were determined as described in the method section. P‐values were calculated with 2‐way ANOVA tests. Combined data of two independent experiments. (H–I) Normalized abundances of (H) palmitic acid and (I) palmitoyl‐carnitine determined via liquid chromatography‐mass spectrometric lipidomics. Values were normalized to IQ values, and P‐values were calculated with 2‐way ANOVA tests. n = 8/group. Central lines represent mean.

Figure 4. Lipids accumulate in liver and kidney during sepsis and cause lipotoxicity

1. A, B

   Immunofluorescent images of liver and kidney 24 h after sham (with or without starvation) or CLP (n = 6–7/group, combined data of 2 independent experiments). White scale bar = 20 μm. (A) Cryosections were stained with Acti‐stain (green), Hoechst (blue), and LipidTOX (red). Z‐stacks were generated in 5–10 areas scattered across the entire tissue section. (B) The amount of lipid droplets (LDs)/cell and average size of LDs (represented by voxel counts) were calculated for each Z‐stack. Averages of the amount and size of LDs were converged for each mouse, and biological replicates are depicted in the table as mean ± SEM. P‐values were calculated using 1‐way ANOVA tests and can be found in Table EV1.

2. C, D

   Quantification of lipid peroxidation by determination of (C) MDA and (D) 4‐HNE concentrations in liver homogenates 24 h post‐surgery in sham and CLP mice as described in methods (n = 6–7/group, combined data of two independent experiments). P‐values were calculated with 1‐way ANOVA tests. Central lines represent mean.

3. E

   Apoptosis in liver paraffin‐fixated sections 24 h after sepsis, measured by TUNEL staining, and presented as % PI‐positive cells/μm² tissue area. (n = 6/7 mice/group, combined data of two independent experiments). P‐values were calculated with 1‐way ANOVA tests. Central lines represent mean.

Figure EV2. No lipid accumulation in heart after sepsis, lipotoxicity in kidney after sepsis

1. A

   Immunofluorescent images of heart 24 h after sham (with or without starvation) or CLP (n = 6–7/group, data are representative of two experiments). Cryosections were stained with Acti‐stain (green), Hoechst (blue), and LipidTOX (red). Z‐stacks were generated in 5–10 areas scattered across the entire tissue section. The amount of lipid droplets/cell and average size of lipid droplets (represented by voxel counts) were calculated for each Z‐stack. White scale bar = 20 μm

2. B, C

   Quantification of lipid peroxidation by determination of (B) MDA and (C) 4‐HNE concentrations in kidney homogenates 24 h post‐surgery in sham and CLP mice as described in methods (n = 6–7/group, combined data of two experiments). P‐value was calculated via one‐way ANOVA test. Central lines represent mean.

3. D

   Apoptosis in kidney paraffin‐fixated sections 24 h after sepsis, measured by TUNEL staining, and presented as % PI‐positive cells/μm² tissue area. P‐values were calculated with 1‐way ANOVA tests. Combined data of two experiments, n = 6–7/group. Central lines represent mean.

Figure 5. The PPARα agonist pemafibrate reduces mortality of sepsis by stimulating PPARα signaling and improving metabolic parameters. Mice were pretreated with pemafibrate (1 mg/kg) or vehicle (0,9% NaCl) for 1 week before being subjected to CLP

1. A

   Survival was monitored during 9 days, after which no further deaths occurred. Survival curve was analyzed via a log‐rank test. Combined data of three experiments, n = 20/group.

2. B

   Liver samples were isolated 24 h after CLP (n = 5–7/group, combined data of two independent experiments), mRNA was prepared, and gene expression levels of Ppara were analyzed via qPCR. Gene expression values are shown as relative expression, normalized to housekeeping genes Hprt and Rpl. P‐values were calculated via 2‐way ANOVA test. Central lines represent mean.

3. C, D

   Plasma was isolated 24 h after sepsis, and (C) FFA concentration and (D) glycerol concentration were determined as described in the method section. P‐values were calculated with 2‐way ANOVA tests. n = 5–7/group, combined data of two independent experiments. Arrows represent the % of decrease caused by pemafibrate treatment during sepsis. Central lines represent mean.

4. E

   Immunofluorescent images of cryosections of liver 24h post‐surgery that were stained with Acti‐stain (green), Hoechst (blue), and LipidTOX (red). Z‐stacks were generated in 5–10 areas scattered across the entire tissue section. White scale bar = 20 μm.

5. F

   The amount of lipid droplets (LDs)/cell and average size of LDs (represented by voxel counts) were calculated for each Z‐stack. Averages of the amount and size of LDs were converged for each mouse, and biological replicates are depicted in the table as mean ± SEM. P‐values were calculated using two‐way Student's t‐tests. n = 6/group, combined data of two independent experiments.

6. G, H

   Quantification of lipid peroxidation by determination of MDA concentration in (G) liver and (H) kidney homogenates 24 h post‐surgery in sham and CLP mice (n = 6–7/group, combined data of two independent experiments), as described in methods. P‐values were calculated with 2‐way ANOVA tests. Arrows represent the % of decrease caused by pemafibrate treatment during sepsis. Central lines represent mean.

7. I

   Oxygen consumption rates (OCRs) of liver tissue explants 24 h post‐sham or CLP (vehicle or pemafibrate‐treated) after supplementation of palmitic acid (PA), measured via Seahorse. Visualization of T_42(min) OCR. P‐values were calculates using 2‐way ANOVA analysis. One experiment, n = 3/group. Central lines represent mean.

Figure EV3. The PPARα agonist pemafibrate reduces mortality of sepsis by stimulating PPARα signaling and improving metabolic parameters

Mice were pretreated with pemafibrate (1 mg/kg) or vehicle (0.9% NaCl) for 1 week before being subjected to CLP.

1. A

   Body temperature of mice 24 h post‐surgery. P‐values were calculated via 2‐way Student's t‐tests. Central lines represent mean.

2. B

   Mouse Clinical Assessment Score for Sepsis (M‐CASS) for sham septic mice 24 h post‐surgery. One experiment (n = 5–6/group). P‐values were calculated via 2‐way ANOVA. Central lines represent mean.

3. C, D

   Liver samples were isolated 24 h after CLP (n = 5–7/group, data are representative of two experiments), mRNA was prepared, and gene expression levels of (C) Acsl1 and (D) Slc25a20 were analyzed via qPCR. Gene expression values are shown relative expression, normalized to housekeeping genes Hprt and Rpl, and P‐values were calculated via 2‐way ANOVA. Central lines represent mean.

4. E

   Immunofluorescent images of cryosections of kidney 24 h post‐surgery that were stained with Acti‐stain (green), Hoechst (blue), and LipidTOX (red). Z‐stacks were generated in 5–10 areas scattered across the entire tissue section. White scale bar = 20μm.

5. F

   The amount of lipid droplets (LDs)/cell and average size of LDs (represented by voxel counts) were calculated for each Z‐stack. Averages of the amount and size of lipid droplets were converged for each mouse, and biological replicates are depicted in the table as mean ± SEM. P‐values were calculated using unpaired t‐tests. n = 6/group, combined data of two experiments.

6. G

   Blood was isolated 24 h post‐surgery, and plasma aspartate aminotransferase (AST) levels were determined as described in Materials and Methods. P‐values were calculated with 2‐way ANOVA tests. n = 6–7/group, combined data of two independent experiments. Central lines represent mean.

7. H–J

   Bacterial load was determined in (H) liver, (I) kidney, and (J) lung tissue homogenates 24 h post‐sepsis. Values are shown as CFU/mg tissue. P‐values were calculated using 2‐way Student's t‐tests. One experiment, n = 5–6 mice/group. Central lines represent mean.

Figure 6. Pemafibrate reduces tissue damage during sepsis

1. A–F

   Mice were pretreated with pemafibrate (1 mg/kg) or vehicle (0.9% NaCl) for 1 week before being subjected to sham or CLP. (A–C) Plasma was collected 24 h post‐surgery, and (A) alanine aminotransferase (ALT), (B) creatinine concentrations, and (C) IL‐6 levels were measured as described in methods. P‐values were calculated with 2‐way ANOVA tests. n = 6‐7/group, combined data of two independent experiments. (D) Systemic bacterial load (CFU/ml blood) 24 h post‐sepsis in vehicle or pemafibrate‐treated mice. P‐values were calculated with 2‐way Student's t‐test. (E–F) Apoptosis in (E) liver and (F) kidney paraffin‐fixated sections 24 h after sepsis, measured with TUNEL staining, and depicted as % of PI‐positive cells/μm² tissue area. P‐values were calculated with 2‐way ANOVA tests. n = 6–7/group, combined data of two independent experiments. Central lines represent mean.

2. G

   Pemafibrate (1 mg/kg) or vehicle (0.9% NaCl) was administered at different timepoints before and after the induction of sepsis and survival was monitored during 9 days, after which no further deaths occurred. Survival curve was analyzed via log‐rank tests, and P‐values can be found in Table EV1. Combined data of 3 independent experiments, n = 15/group.

3. H

   Mice were injected with the PPARα antagonist GW6471 (10 μg/g) or vehicle (DMSO) 3 h pre‐CLP and 3h post‐CLP. Survival was monitored during 9 days, after which no further deaths occurred. Survival curve was analyzed via a log‐rank test. Combined data of two independent experiments, n = 16/group.

Figure EV4. GW6471 treatment worsens septic disease parameters and bacterial load in blood and organs

Mice were injected with the PPARα antagonist GW6471 (10 μg/g) or vehicle (DMSO) 3 h pre‐CLP and 3 h post‐CLP, 24 h post‐CLP blood, and organs were isolated (n = 5‐6/group). One experiment.

1. A

   Body temperature of mice 24 h post‐surgery. Central lines represent mean.

2. B

   Mouse Clinical Assessment Score for Sepsis (M‐CASS) for septic mice 24 h post‐surgery. Central lines represent mean.

3. C–F

   Bacterial load in (C) blood and (D) liver, (E) kidney, and (F) lung tissue homogenates of septic mice. Values are shown as CFU/mg tissue. Central lines represent mean.

Data information: P‐values were calculated using two‐way Student's t‐tests.

Figure 7. Lipolysis is increased in septic patients. Blood samples were collected from healthy volunteers and septic patients (n = 13 septic patients, =15 healthy controls)

1. A, B

   Plasma was prepared, and (A) FFA and (B) glycerol concentrations were determined as described in the methods. P‐values were calculated with two‐way Student's t‐tests. Central lines represent mean.

2. C, D

   Correlation of (C) FFA and (D) glycerol concentrations with disease severity scores, lactate, or inflammatory cytokine levels from septic patients was calculated. Values are shown as Pearson correlation values (r) with associated P‐values.