Infusion of lard oil, but not soy oil, induces inflammation and insulin resistance in rats.
Experimental infusion of triglyceride-heparin emulsions into the bloodstream of rodents and humans has become a common experimental strategy for assessing the mechanisms by which lipids alter insulin secretion and action. In 1997, the McGarry group demonstrated that the relative composition of the infusate is a critical determinant of experimental design, as soy-based cocktails enriched in unsaturated fats inhibited insulin secretion, while a lard-infusate containing a higher percentage of saturated fats had the opposite effect (20
). Using an experimental protocol comparable to that utilized by the McGarry group, we evaluated the effects of saturated vs. unsaturated fats on insulin-stimulated glucose disposal. As compared with a 2.5% glycerol infusate, delivery of either a lard- or soy-based cocktail markedly inhibited rates of insulin-stimulated glucose disposal (Figure A) and increased levels of diacylglycerol (DAG) in soleus muscle (Figure C). However, only lard oil increased muscle ceramide levels (Figure B).
Lard oil infusion inhibits insulin-stimulated glucose uptake in a ceramide-dependent manner.
To determine the quantitative importance of ceramides as modulators of lard-induced insulin resistance, we treated the animals with inhibitors (i.e., myriocin or cycloserine) of serine palmitoyltransferase (SPT), the enzyme responsible for the first committed step toward ceramide biosynthesis. As shown in Figure , both myriocin and cycloserine prevented lard induction of insulin resistance (Figure A) and ceramide accumulation (Figure B) but had no effect on the soy-induced insulin resistance. Neither compound reduced DAG levels (Figure C). Collectively, these data indicate that ceramide synthesis is essential for lard-induced insulin resistance but is dispensable for that caused by soy oil. These data confirm that different fatty acids induce insulin resistance via distinct mechanisms that can be discerned by their reliance on ceramide.
We predicted that the lard oil, because of its elevated SFA content, would induce a more pronounced inflammatory response than could be achieved using the soy-based infusate. Consistent with this hypothesis, lard oil infusion had a striking effect on circulating IL-6 and TNF-α levels, which increased several-fold in the lard-infused rats (Figure , A and B, respectively). Molecular analysis of the lard oil infusate indicated that it was free of endotoxin (data not shown). Co-infusing the inhibitors of ceramide biosynthesis, which normalized insulin sensitivity, did not reduce circulating cytokine levels.
Lard oil infusion increases circulating inflammatory cytokine concentrations and impairs insulin-stimulated glucose disposal in a TLR4-dependent manner.
TLR4 deletion prevents SFA-induced cytokine release and insulin resistance.
To confirm that the inflammatory response caused by lard oil infusion was mediated by TLR4-dependent signaling pathways, we measured cytokine levels in mice expressing a defective TLR4 (Tlr4lps-d). As we saw in the rats, lard oil infusion markedly increased serum cytokine levels in WT (Tlr4+/+) mice (Figure C). However, mice lacking functional TLR4 (Tlr4lps-d) showed a diminution in circulating TNF-α and IL-6 following lipid infusion (Figure C). To test the relevance of the TLR4 pathway on insulin sensitivity in this experimental model, we performed hyperinsulinemic-euglycemic clamps on the TLR4-knockout animals during the final 1.5 hours of lipid infusion. As shown in Figure D, Tlr4lps-d mice were protected from insulin resistance caused by lard oil infusion but were susceptible to insulin resistance caused by soy oil (Figure D). Mice lacking MyD88 (an adapter protein requisite for TLR signaling), but not those lacking TLR2, were also protected from lard-induced insulin resistance (Figure E).
TLR4 deletion selectively blocks LPS inhibition of glucose uptake in skeletal muscle.
The data presented thus far indicate that both TLR4 and ceramide are required for lard oil–induced insulin resistance, but they are dispensable for that caused by soy oil. Thus, TLR4 and ceramide reside in either linear or parallel signaling pathways linking excess saturated lipids to the antagonism of insulin action. To explore relationships between inflammation and ceramide, we turned to an isolated muscle system, which allows us to evaluate tissue-autonomous effects of individual lipids on muscle insulin sensitivity. Using this system, we have previously shown that palmitate antagonizes insulin-stimulated glucose uptake via a ceramide-dependent mechanism (3
). The data described below reveal that the TLR4 agonist LPS also antagonizes insulin-stimulated 2-deoxyglucose (2-DOG) uptake via ceramide. As shown in Figure A, LPS inhibited 2-DOG uptake in soleus muscles obtained from Tlr4+/+
mice, but had no effect in muscles isolated from the Tlr4lps-d
animals (Figure A). LPS additionally stimulated the production of ceramide, as intramyocellular levels of the lipid rose several-fold during a 6-hour LPS treatment (Figure B). To test the importance of this ceramide in LPS responses, we used 2 separate experimental approaches to block rates of de novo ceramide synthesis. First, we pretreated with the aforementioned inhibitor of SPT, myriocin. Second, we utilized muscles obtained from mice heterozygous for dihydroceramide desaturase (Des1), which we previously demonstrated are compromised in their ability to produce ceramide (3
). As shown in Figure A, both of these experimental approaches negated LPS-induced insulin resistance.
LPS and palmitate (PA) impair insulin-stimulated glucose uptake in isolated muscles via sphingolipid- and TLR4-dependent mechanisms.
We additionally chose to investigate the relationships between palmitate, TLR4, and ceramide using the isolated muscle system, where we had previously demonstrated that the SFA antagonized glucose transport by inducing ceramide synthesis (3
). As shown in Figure , C and D, Tlr4lps-d
mice were resistant to palmitate, as it failed to antagonize insulin-stimulated glucose transport or induce ceramide accrual. In contrast, palmitate promoted ceramide accumulation and inhibited glucose uptake in muscles from WT or Tlr2–/–
rodents. To evaluate the specificity of this response, we evaluated whether TLR4 inactivation altered rates of glycerolipid (i.e., DAG) synthesis or lipid oxidation. As shown in Figure E, palmitate induced DAG accumulation comparably in muscles from WT, Tlr2–/–
, and Tlr4lps-d
mice. Moreover, the Tlr4lps-d
mice demonstrated an enhanced rate of palmitate oxidation (Figure F). These data strongly indicate that ceramide or a ceramide metabolite, and not DAG or products generated during lipid oxidation, are key intermediates linking both palmitate and TLR4 to the antagonism of insulin action.
One of the cytokines induced by SFAs and TLR4 is TNF-α, which has long been implicated in insulin resistance and which induces ceramide by activating sphingomyelinase or by promoting its de novo ceramide synthesis. TNF-α was unable to induce insulin resistance, either in control muscle strips (data not shown) or in those obtained from the Tlr4lps-d
mice (Supplemental Figure 1A; supplemental material available online with this article; doi:
). However, TNF-α was capable of recapitulating the TLR4 effect, as its re-addition in concert with palmitate antagonized glucose uptake in Tlr4lps-d
muscles (Supplemental Figure 1A). Mice lacking TNF receptors were not protected from lard-induced insulin resistance (Supplemental Figure 1B). Thus, while TNF-α was sufficient to recapitulate the insulin-desensitizing role of LPS in muscles from Tlr4lps-d
mice, it was not necessary.
Tlr4lps-d mice fail to accrue ceramides with hyperlipidemia.
Rates of ceramide synthesis have long been thought to be controlled primarily by the availability of palmitoyl-CoA, which is a substrate for SPT and is required for formation of the sphingosine backbone. The data presented in this study suggest a different mechanism through which excessive fat promotes ceramide accrual. Specifically, these results indicate that lipid-activation of TLR4 leads to an upregulation of sphingolipid synthesis. To confirm this mechanism in vivo, we evaluated ceramide levels in the lard-infused mice (Figure ). Lard oil infusion increased ceramide levels in both liver (Figure A) and muscle (Figure B) in Tlr4+/+ mice but failed to induce it in either tissue in the Tlr4lps-d mice. Lard oil infusion did increase DAG levels in both tissues (Figure D), indicating that TLR4 receptor ablation specifically impairs sphingolipid synthesis in peripheral tissue without affecting lipid uptake or bulk storage. To assess insulin sensitivity, we performed hyperinsulinemic-euglycemic clamps. Coincident with the reduction in ceramide levels, the Tlr4lps-d mice demonstrated a substantial improvement in muscle and liver insulin sensitivity, as evidenced by a higher rate of glucose infusion to maintain euglycemia, increased whole body glucose disposal, and reduced hepatic glucose output (Figure , E–G).
TLR4 signaling is essential for lipid-induced ceramide accumulation.
The hypothalamus has an emerging role in the regulation of hepatic glucose efflux, and SFAs have been shown to promote central leptin and insulin resistance associated with reduced activation of the anabolic enzyme Akt/protein kinase B (21
). Lipid infusion significantly increased hypothalamic ceramide content of WT mice (Figure C). In contrast, no change in ceramides was detected in the hypothalamus of Tlr4lps-d
mice following lard oil infusion (Figure D). Contrary to effects seen in peripheral tissues, Tlr4lps-d
mice also fail to accumulate DAG in the hypothalamus (Figure D).
TLR4 agonists stimulate ceramide synthesis in both myotubes and macrophages.
The studies in isolated muscle fibers revealed that the SFA-TLR4 pathway leading to sphingolipid synthesis and insulin resistance could be recapitulated in a tissue-autonomous system. Whole skeletal muscle consists of a variety of cell types in addition to myocytes such as macrophages and leukocytes. Though macrophages are known to contain TLR4, recent studies suggest they are located in myotubes and are upregulated in muscle cultured from obese and insulin-resistant humans (25
). Thus, we sought to determine whether the effects of TLR4 stimulation on skeletal muscle sphingolipid metabolism and insulin sensitivity resulted from effects in myotubes, or instead resulted from effects in resident macrophages. We conducted an extensive lipidomic analysis of both C2
myocytes and RAW264.7 macrophages treated with palmitate or LPS. Of all the lipids tested, only ceramide (in both myotubes and macrophages) and glucosylceramide (only in macrophages) increased in response to both LPS and palmitate (Figure ). Other sphingolipids (i.e., GM3 ganglioside and sphingomyelin) levels were unchanged (data not shown). Glycerolipids (i.e., DAG and TAG) increased with palmitate in both cell types but were unaffected by LPS (Figure ).
LPS and PA elicit an increase in sphingolipd synthesis.
We next tested whether the ceramide generated within these cell types via the TLR4 agonists elicit biological responses, including impairment of insulin signaling (in myotubes) and cytokine generation (in macrophages). We previously demonstrated that relatively small increases in cellular ceramide levels, within the range observed in the lipid infusion and isolated muscle experiments described herein, are sufficient to inhibit insulin stimulation of the serine/threonine kinase Akt/PKB (26
). Moreover, our prior studies have confirmed that this is the primary mechanism through which ceramides inhibit glucose transport (27
). We thus tested whether LPS was capable of inhibiting Akt/PKB in a cell-autonomous system. As shown in Figure A, both LPS and palmitate were effective at inhibiting insulin-stimulated activation of Akt/PKB. Moreover, they inhibited insulin-stimulated phosphorylation of glycogen synthase kinase 3β, an Akt/PKB substrate. The inclusion of linoleate (LA) into either palmitate or LPS treatment had little effect on signaling or ceramides (Supplemental Figure 2). In macrophages, both LPS and palmitate, but not the unsaturated fatty acid LA, increased mRNA levels of TNF (Figure B). This transcriptional effect of palmitate on TNF-α was unaffected by myriocin or the ceramide synthase (CerS) inhibitor fumonisin B1, confirming that the effect is ceramide independent.
Ceramide is necessary for LPS and PA inhibition on insulin signaling at Akt/PKB in myotubes but not for LPS and PA induction of TNF-α in macrophages.
IKKβ is essential for TLR4-mediated insulin resistance and ceramide synthesis.
A critical factor in many TLR4 responses is the transcription factor NF-κB, which translocates to the nucleus upon TLR4 activation and drives the synthesis of stress-induced cytokines. A family of IκB kinases (IKK) activates NF-κB signaling by phosphorylating inhibitor of NF-κB (IκBα), a protein that sequesters NF-κB in the cytosol. Once phosphorylated, IκBα releases NF-κB, and is itself degraded by the proteasome. NF-κB then translocates into the nucleus and upregulates genes associated with inflammation. IKKβ is well known for disruption of insulin responsiveness (28
) and thus emerged as a likely modulator of insulin sensitivity and ceramide synthesis in this model system. To test the role of IKKβ in mediating TLR4-induced insulin resistance and ceramide accrual, stable C2
lines expressing WT (IKK-WT) or dominant-negative (kinase-dead IKKβ [IKK-KD]; K44M) isoform were generated. Overexpression of IKK-KD, but not IKK-WT, prevented LPS-induced degradation of IκBα (Figure A). Lipidomic analysis revealed that inhibition of IKKβ signaling dramatically and selectively reduced levels of ceramides (Figure B), preventing ceramide induction by either palmitate or LPS. The cells expressing the IKK-KD were also resistant to palmitate- and LPS-inhibition of insulin signaling to Akt/PKB (Figure D). To evaluate the mechanism through which TLR4/IKKβ modulates ceramide levels, we investigated the effects of LPS and palmitate on the transcripts for several enzymes that drive ceramide synthesis, including SPT (isoforms 1 and 2), several CerS isoforms, and DES1. LPS and palmitate induced a number of different ceramide biosynthetic enzymes, most notably STP2 and CerS1, -2, and -6 (Figure E). In cells expressing the IKK-KD, which failed to accrue ceramide, levels of these transcripts were markedly reduced, and LPS and palmitate failed to alter their expression.
Overexpression of a dominant-negative IKKβ prevents ceramide accrual and LPS- and PA-induced insulin resistance.
To test the relevance and specificity of IKKβ as a modulator of sphingolipid synthesis in vivo, we treated 22-week-old diet-induced obese mice with the IKKβ inhibitor sodium salicylate (NS). NS had no effect on body weight (Figure A) but markedly improved glucose and insulin tolerance (Figure , B and C). High-fat feeding increased ceramide levels in soleus, liver, and hypothalamus, but NS treatment blocked its accumulation in all 3 locales (Figure , D–I). DAG levels, which also increased in response to high-fat feeding, were unaffected by NS treatment in liver and muscle but were reduced in the hypothalamus. This latter finding is consistent with the earlier results in the lipid infusion model, where TLR4 selectively affected ceramides in peripheral tissues but had a more global effect in the hypothalamus.
Suppression of IKKβ with NS restores insulin sensitivity via ceramide inhibition.