Adipose tissue protects against sepsis-induced muscle weakness in mice: from lipolysis to ketones

Male, 24-week-old mice were anesthetized and a catheter was placed in the central jugular vein, followed by cecal ligation and puncture to induce sepsis [33, 34]. Unless indicated, 57BL/6JRj mice (Janvier SAS, Chassal, France) were used. After surgery, mice were fasted and received intravenous fluid resuscitation for 20 h. Fasting was done to mimic the clinical setting in which PN is initially withheld. From day 1 onward, mice received standard mixed PN (5.8 kcal/day; Olimel N7E, Baxter, Lessines, Belgium) unless indicated. Throughout the study, mice were given antibiotics and pain medication. Pain/discomfort was assessed twice daily based on the Mouse Grimace Score [35], and cumulative illness scores were calculated to assess the severity of illness. Individually caged healthy mice receiving standard chow at the same daily caloric intake as septic mice (pair-fed) were used as controls.

Directly after euthanasia, the extensor digitorum longus (EDL) muscle was carefully dissected and suspended in a temperature controlled (30 °C) organ bath filled with HEPES-fortified Krebs-Ringer solution to measure muscle force (300C-LR Dual-Mode muscle lever, Aurora Scientific, Ontario, Canada). The small size of the EDL guaranteed proper diffusion of oxygen during the procedure. Maximal isometric tetanic force of the EDL muscle was measured by averaging three consecutive tetanic stimuli (180 Hz stimulation frequency, 200 ms duration, 0.2 ms pulse width, 2 min rest intervals). Specific maximal isometric tetanic force was calculated by dividing the maximal isometric tetanic force with the muscle cross-sectional area. Additional information is provided in Additional file 7.

For practical reasons, tibialis anterior (TA) muscle was used for histology and muscle mass assessment, whereas the larger gastrocnemius muscle was homogenized and used for gene, protein, and metabolite analyses.

Plasma glucose concentrations were measured in whole blood with a glucose meter after cardiac puncture (Accu-check, Roche, Basel, Switzerland). Plasma glycerol, TNF-α, LDL, HDL, triglycerides, free fatty acids, insulin, and 3-HB were measured using commercially available kits: Glycerol Assay Kit (MAK117, Sigma-Aldrich), Mouse TNF-alpha Quantikine HS ELISA Kit (MHSTA50; R&D Systems), LDL-Cholesterol assay (DZ128A-K; Diazyme Laboratories, Poway, CA, USA), HDL-Cholesterol assay (DZ129A-K; Diazyme), triglyceride quantification kit (Ab65336; Abcam), Free Fatty Acid Fluorometric Assay Kit (7010310; Cayman Chemical Company, Ann Arbor, MI, USA), Insulin Mouse Ultra Sensitive ELISA (90080, Crystal Chem, Downers Grove, IL, USA), and EnzyChrom™ Ketone Body Assay Kit (EKBD-100; Bioassay Systems, Hayward, CA, USA).

Normally distributed data were compared with one-way analysis of variance (ANOVA) with post hoc Fisher’s LSD test (Student’s t test) for multiple comparisons, where necessary, after log- or (double) square root-transformation to obtain a near-normal distribution (JMP® Pro 12, SAS Institute Inc., Cary, NC, USA). Not-normally distributed data were analyzed with non-parametric Wilcoxon tests. Two-sided p values ≤ 0.05 (α-level of 5%) were considered statistically significant in all analyses. Data are presented as bars with whiskers, showing means and standard error of the mean (SEM). Post hoc p values are plotted on the figures as § p ≤ 0.05, §§ p ≤ 0.01, §§§ p ≤ 0.001, between septic and healthy control mice, and * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001 between groups of septic mice.

In the first study (Table 1), we hypothesized that overweight/obese mice have an altered metabolic response to sepsis compared to the lean mice, with elevated mobilization and metabolism of endogenous fatty acids. In overweight/obese mice, initial fat pad weight was higher than in the lean mice (910 ± 128 mg vs. 408 ± 36 mg; p = 0.01). During their illness, overweight/obese septic mice lost about twice the amount of fat than did lean septic mice (day 5 fat pad weight 462 ± 53 mg in the overweight/obese vs. 162 ± 42 mg in the lean; p = 0.0006). After 1 day of sepsis, plasma glycerol and ex vivo-released glycerol from adipose tissue (markers of lipolysis) were increased, which was more pronounced in overweight/obese than in lean mice (Fig. 1a,b). Nevertheless, during this acute phase of sepsis, both overweight/obese and lean septic mice presented with a similar plasma lipid profile as pair-fed controls, with low fatty acid, LDL- and HDL-cholesterol concentrations and normal triglyceride concentrations (Additional file 1: Figure S1). This suggested enhanced peripheral uptake of the released fatty acids during acute sepsis, especially in the overweight/obese mice. Indeed, sepsis-induced upregulation of fatty acid uptake transporter Cd36 mRNA in muscle and liver was most pronounced in the overweight/obese (Fig. 1c,d). Expression of other markers of hepatic and muscular fatty acid oxidation was generally unaffected by obesity or sepsis (Additional file 1: Figure S1).

After 5 days of sepsis, plasma and ex vivo-released glycerol from adipose tissue returned to control levels in both overweight/obese and lean septic mice (Fig. 1a,b). Plasma fatty acid, triglyceride and LDL-cholesterol concentrations normalized, whereas plasma HDL-cholesterol concentrations remained low (Additional file 1: Figure S1). Nevertheless, hepatic and muscular Cd36 gene expression were still higher with sepsis, for liver only in overweight/obese mice (Fig. 1c, d). Furthermore, in these overweight/obese prolonged septic mice, in vitro palmitate oxidation in the muscle, but not the liver, remained up to control levels. In contrast, in vitro palmitate oxidation was reduced in the lean (Fig. 1e,f). Also during prolonged sepsis, gene expression of markers of hepatic and muscle fatty acid oxidation were generally unaffected by obesity or sepsis (Additional file 1: Figure S1).

In the second study (Table 1), we hypothesized that overweight/obese septic mice would no longer be protected against the loss of muscle mass and function when they are unable to release fatty acids from their excess adipose tissue. Blocking lipolysis via AAKO in overweight/obese mice indeed caused a greater loss of TA and EDL muscle mass after 5 days of sepsis compared to wild-type overweight/obese mice (Fig. 2a,b). This more severe muscle wasting in the AAKO overweight/obese coincided with additional upregulation of the ubiquitin-proteasome system markers Fbxo32 and Trim63 (Fig. 2c). Furthermore, AAKO overweight/obese septic mice suffered from extremely reduced absolute (p < 0.0001 vs. AAKO healthy controls; Fig. 2d) and specific maximal muscle force (Fig. 2e). In contrast, wild-type overweight/obese septic mice displayed only slightly reduced absolute maximal force (p = 0.01 vs. wild-type healthy controls; Fig. 2d) and maintained specific maximal force up to healthy levels (Fig. 2e). AAKO overweight/obese septic mice also displayed a trend for a higher mortality rate than wild-type overweight/obese septic mice (Table 1) and had worse cumulative illness scores (10.1 ± 1.4 in AAKO vs. 5.4 ± 1.2 in wild-type; p = 0.02).

Results from the first two studies in overweight/obese mice identified the increased availability of endogenously released lipids as key in the protection against sepsis-induced muscle wasting and weakness. In the third study (Table 1), we hypothesized that an increased lipid availability in lean septic mice would also protect them against sepsis-induced muscle wasting and weakness. However, after 5 days of sepsis, both lean mice that received standard mixed PN or lipid-rich PN presented with reduced TA and EDL muscle masses (Fig. 3a,b). Ubiquitin-proteasome system markers Fbxo32 and Trim63 mRNA were also comparably affected by sepsis in both groups of lean mice (Fig. 3c). As expected, lean septic mice on standard PN suffered from reduced absolute maximal muscle force (p < 0.001 vs. healthy controls; Fig. 3d) and specific maximal muscle force (Fig. 3e). In contrast, lean septic mice on lipid-rich PN showed less reduced absolute muscle force (p = 0.05 between standard PN and lipid-rich PN; Fig. 3d) and maintained their specific maximal muscle force equivalent to that of healthy controls (Fig. 3e).

Muscular and hepatic gene expression of fatty acid oxidation markers Cd36, Cpt1b, Acadl, and Hadha, and in vitro palmitate oxidation were highly increased in septic lean mice on lipid-rich PN compared to those on standard mixed PN (Fig. 4a–d). Importantly, giving high doses of lipids to lean septic mice did not adversely affect the mortality rate (Table 1), nor the cumulative illness scores (7.1 ± 1.1 with lipid-rich PN vs. 6.2 ± 0.9 with standard mixed PN; p = 0.7). However, the lipid-rich PN did cause triglyceride accumulation in both the liver and the muscle (Fig. 4e,f). Furthermore, these mice also presented with histologically confirmed liver steatosis (global steatosis score: 2.72 ± 1.54 with lipid-rich PN vs. 1.56 ± 1.03 with standard mixed PN; p = 0.01; Fig. 4g) and elevated plasma-free fatty acid, triglyceride, and LDL and HDL cholesterol concentrations (Fig. 4h–k).

In a previous study, the protection against sepsis-induced muscle wasting and weakness in overweight/obese mice coincided with elevated plasma concentration of ketone body 3-HB [15]. Here, we confirm the higher plasma 3-HB concentrations in overweight/obese septic mice than in the lean mice (Fig. 5a). Furthermore, these overweight/obese septic mice showed better preserved hepatic gene expression of the main ketogenic enzyme Hmgcs2 (Fig. 5b). Blocking lipolysis via AAKO in overweight/obese septic mice prevented the rise in plasma 3-HB concentrations and lowered the hepatic Hmgcs2 mRNA expression, similar as in lean septic mice (Fig. 5a,b). In contrast, providing lean septic mice with lipid-rich PN drastically increased hepatic Hmgcs2 mRNA expression and plasma 3-HB concentrations (Fig. 5a,b). This elevated ketogenesis in the overweight/obese and in lean mice receiving lipid-rich PN was not due to a change in plasma insulin concentrations (Additional file 2: Figure S2). Combined, these results suggest that ketone bodies may play an important role in the protection against muscle wasting and weakness.

In the fourth study (Table 1), we hypothesized that direct exogenous administration of ketone bodies can also protect lean septic mice against muscle wasting and weakness. After 5 days, all lean septic mice lost TA and EDL muscle mass, irrespective of 3-HB supplementation (Fig. 5c,d). However, 3-HB supplementation did prevent Fbxo32 mRNA upregulation (Fig. 5e). Similar to the lipid-rich PN, 3-HB supplementation in lean septic mice attenuated muscle weakness, with less reduced absolute maximal muscle force (p = 0.002 for PN+3-HB vs. PN+gluc; Fig. 5f) and a specific maximal muscle force comparable to that of lean healthy control mice (Fig. 5g). PN+3-HB in lean septic mice did not affect mortality (Table 1) nor cumulative illness scores (8.1 ± 1.2 with PN+3-HB vs. 6.5 ± 1.3 with PN+gluc; p = 0.2). Unlike lipid-rich PN, PN+3-HB did not affect markers of fatty acid oxidation in muscle or liver (Additional file 3: Figure S3) or induce hepatic triglyceride accumulation (Additional file 3: Figure S3). Furthermore, 3-HB supplementation partially normalized the plasma lipid profile (Additional file 3: Figure S3). In the muscle of PN+3-HB septic mice, markers of ketolysis, the triglyceride and glycogen content, and plasma glucose concentrations were decreased or unaltered compared to PN+gluc (Additional file 3: Figure S3). Together, these results argue against the use of 3-HB as an alternative energy substrate by the muscle during critical illness.

Several signaling functions have been attributed to 3-HB as well: 3-HB has known anti-inflammatory and autophagy-activating properties and can stimulate mTOR-mediated protein synthesis. However, the illness-induced effects on these pathways were not altered by 3-HB supplementation (Additional file 4: Figure S4). In contrast, PN+3-HB did have effects on pathways involved in muscle regeneration. Whereas the number of PAX7-positive muscle satellite cells was unaffected by illness (Fig. 6a), gene expression of proliferation marker Pcna was clearly increased by sepsis, especially with PN+3-HB (Fig. 6b). As compared to PN+gluc, PN+3-HB increased the expression of regeneration markers Myod1, Myog, and Myf5 in lean septic mice (Fig. 6b). Furthermore, PN+3-HB caused a greater reduction in gene expression of myostatin, an inhibitor of muscle growth (Fig. 6b). Both groups of septic mice showed similarly increased expression of the muscle acetylcholine receptor (AchR) α-polypeptide gene (Chrna1) (Fig. 6c). This gene encodes for the α-subunit that is present in all AchRs. Expression of the adult AchR ε-subunit gene (Chrne) that is found in mature myofibers was most reduced in lean PN+3-HB septic mice (Fig. 6c). In contrast, PN+3-HB caused a four times higher increase than PN+gluc in the expression of the embryonic AchR γ-subunit gene (Chrng), found in immature myofibers (Fig. 6c). These 3-HB supplemented lean septic mice also presented with a greater elevation of Actb mRNA than the PN+gluc septic mice (Fig. 6d). Of note, histone deacetylases (HDAC) 4 and 5 are known inhibitors of the muscle regeneration pathway by blocking the expression of myogenic transcription factor MEF2. In lean septic mice receiving PN+3-HB, muscular expression of Hdac4 and Hdac5 was suppressed (Fig. 6e). Furthermore, Mef2c mRNA was higher in lean PN+3-HB than lean PN+gluc ill mice (Fig. 6e).