PPARα augments heart function and cardiac fatty acid oxidation in early experimental polymicrobial sepsis

Abstract

Children with sepsis and multisystem organ failure have downregulated leukocyte gene expression of peroxisome proliferator-activated receptor-α (PPARα), a nuclear hormone receptor transcription factor that regulates inflammation and lipid metabolism. Mouse models of sepsis have likewise demonstrated that the absence of PPARα is associated with decreased survival and organ injury, specifically of the heart. Using a clinically relevant mouse model of early sepsis, we found that heart function increases in wild-type (WT) mice over the first 24 h of sepsis, but that mice lacking PPARα (Ppara^−/−) cannot sustain the elevated heart function necessary to compensate for sepsis pathophysiology. Left ventricular shortening fraction, measured 24 h after initiation of sepsis by echocardiography, was higher in WT mice than in Ppara^−/− mice. Ex vivo working heart studies demonstrated greater developed pressure, contractility, and aortic outflow in WT compared with Ppara^−/− mice. Furthermore, cardiac fatty acid oxidation was increased in WT but not in Ppara^−/− mice. Regulatory pathways controlling pyruvate incorporation into the citric acid cycle were inhibited by sepsis in both genotypes, but the regulatory state of enzymes controlling fatty acid oxidation appeared to be permissive in WT mice only. Mitochondrial ultrastructure was not altered in either genotype indicating that severe mitochondrial dysfunction is unlikely at this stage of sepsis. These data suggest that PPARα expression supports the hyperdynamic cardiac response early in the course of sepsis and that increased fatty acid oxidation may prevent morbidity and mortality.

NEW & NOTEWORTHY In contrast to previous studies in septic shock using experimental mouse models, we are the first to demonstrate that heart function increases early in sepsis with an associated augmentation of cardiac fatty acid oxidation. Absence of peroxisome proliferator-activated receptor-α (PPARα) results in reduced cardiac performance and fatty acid oxidation in sepsis.

sepsis and septic shock are frequent causes of morbidity and mortality in the United States and throughout the world (1, 62). Decreased myocardial function plays a prominent role in cardiovascular compromise in sepsis and is one of the defining features of septic shock, contributing to both early and late mortality (58). Although sepsis is associated with both a hyperinflammatory cytokine storm as well as immunoparalysis (17), the cellular pathophysiology underlying sepsis-induced cardiovascular failure has not been elucidated completely (13, 47).

We previously showed that children with septic shock have decreased peroxisome proliferator activated receptor-α (PPARα) expression in their circulating leukocytes and that disease severity, multiple organ system failure, and death were significantly worse in patients with the lowest PPARα expression (50). In an animal model of experimental sepsis, survival was decreased in mice lacking PPARα expression (Ppara^−/−) compared with wild-type (WT) mice (50). However, survival did not improve in chimeric Ppara^−/− mice transplanted with WT bone marrow, indicating that disease severity is primarily affected by organ tissue PPARα expression, rather than PPARα expression in hematopoietic immune cells. In particular, Ppara^−/− mice showed evidence of significant cardiac injury with elevated troponin levels and histologic findings of myocardial damage. Myocardial mRNA expression of PPARα and other fatty acid oxidation (FAO) genes was suppressed in WT hearts during sepsis, with a further reduction in Ppara^−/− hearts (51). Fatty acids are a primary source of energy for the heart (52); therefore, these findings suggest that decreased survival in Ppara^−/− mice is potentially due to cardiac injury related to insufficient FAO.

Lower rates of cardiac FAO are found in many nonseptic causes of heart failure (21, 38, 52). Additionally, human sepsis studies have demonstrated cardiac dysfunction with reduced myocardial fatty acid uptake and cardiomyocyte lipid accumulation, indicating a shift away from fatty acid utilization (10, 46). Previous studies in murine models of sepsis have also demonstrated cardiac dysfunction associated with depressed myocardial FAO (12, 48). However, these mouse experiments utilized intraperitoneal lipopolysaccharide (LPS) injections to model sepsis, whereas we have employed a cecal ligation and puncture sepsis model (CLP). Although animal models of sepsis are imperfect (40, 49), the CLP model more closely approximates the infectious trigger, time course, physiologic changes, and clinical interventions in human sepsis than other experimental animal models (2, 18). Furthermore, the previous sepsis metabolic studies only evaluated myocardial glucose and palmitate (for fatty acid) oxidation; however, the myocardium oxidizes other substrates, especially when under stress. Thus additional myocardial metabolic evaluations are necessary using the CLP sepsis model before translating metabolic interventions to clinical practice.

In the current study, we evaluated cardiac function and myocardial FAO as potential mechanisms whereby differing PPARα expression affects sepsis outcomes. We found that CLP-induced sepsis alters myocardial function and substrate metabolism and that decreased survival in Ppara^−/− mice is associated with cardiac dysfunction and reduced FAO. We also demonstrated that pyruvate dehydrogenase, the enzyme that converts pyruvate to acetyl-CoA for incorporation into the citric acid cycle (CAC), is inhibited while enzymes regulating FAO maintain permissive levels in WT but not Ppara^−/− myocardium in sepsis.

MATERIALS AND METHODS

Animals.

The study procedures were approved by both the University of Washington and Seattle Children’s Research Institute Institutional Animal Care and Use Committees. To allow direct comparison with our previously reported study cohorts (51), we used 12- to 14-wk-old male C57Bl/6J mice and age-matched Ppara^−/− mice on the same C57BL/6J background (B6.129S4-Pparatm1Gonz/J) bred and housed in colonies maintained in our own vivarium (33).

Cecal ligation and puncture.

Sepsis was induced by cecal ligation and puncture (CLP) as previously described (50, 51). Briefly, while mice were under isoflurane anesthesia, a laparotomy was performed and the cecum was tied off at 50% of its length and punctured once with a 25-gauge needle. Each mouse received postoperative subcutaneous normal saline solution, imipenem (25 mg/kg), and buprenorphine (0.05 mg/kg) every 12 h until death 24 h after CLP. Sham mice underwent anesthetized laparotomy only, received the same postoperative treatments, and were also fasted because septic mice do not have significant oral intake. Because the Ppara^−/− mouse is a constitutive, whole body knock out, control mice underwent no procedure, received no treatment, and were allowed full access to food and water to determine if any functional or metabolic differences exist between genotypes at in the healthy state (baseline). Separate cohorts of mice were used for 1) ex vivo cardiac perfusion experiments, and 2) sample collection for electron microscopy, Western blot, tissue triglyceride, and serum substrate experiments. Tissues from the nonperfusion experiments were immediately fixed or flash frozen at the time of euthanasia for subsequent analysis.

Echocardiogram.

Serial echocardiograms were performed at baseline, and then at 8 and 24 h after surgery. Studies were performed as previously described while mice were under light isoflurane anesthesia (31, 39). Images were captured with a Vevo 2100 machine using a MS400 transducer (VisualSonics, Toronto, Canada). M-mode measurements at the midpapillary level of the left ventricle (LV) were performed by a blinded reader at end-diastole (LVEDD) and end-systole (LVESD) to determine LV function via the fractional shortening [(LVEDD − LVESD)/LVEDD × 100] in a parasternal short axis mode.

Isolated working heart preparation.

Working heart experiments were performed as previously described (20, 31, 39). Briefly, mice were heparinized and anesthetized followed by rapid cannulation of the aorta and mounting of the heart on the perfusion apparatus. Hearts were perfused with physiological salt solution containing the following ¹³C-labeled substrates in addition to unlabeled glucose (5.5 mmol/l) were as follows: 1,3-[¹³C]acetoacetic acid (ACAC; 0.17 mmol/l), L-lactic-3-[¹³C]acid (lactate; 1.2 mmol/l), and U-[¹³C]-long-chain mixed free fatty acids (free fatty acids; 0.35 mmol/l) bound to 0.75% (wt/vol) delipidated bovine serum albumin reconstituted with deionized water. All studies included insulin (50 μU/ml) in the perfusate. Preload was set at 12 mmHg and afterload at 50 mmHg.

Left ventricular (LV) pressure was continuously measured using an SPR-PV Catheter (Millar Instruments, Houston, TX) inserted into the LV through the apex. Left atrial inflow was measured with a flow probe (T403; Transonic Systems, Ithaca, NY) and aortic flow (not including coronary flow) was measured via 30‐s timed collections. Coronary flow was calculated as the difference between left atrial inflow and aortic flow. Every 10 min, left atrial influent and coronary effluent were collected for determination of Po₂, Pco₂, and pH with an ABL800 blood gas analyzer (Radiometer, Copenhagen, Denmark). Continuously recorded parameters were LV pressure (mm Hg), heart rate (beats/min), and rate of LV contraction and relaxation (±dP/dt, mm Hg/s). Developed pressure was calculated as the maximum LV pressure subtracted by the LV end diastolic pressure. Cardiac power was calculated as the developed pressure times the left atrial inflow. Myocardial oxygen consumption ($\mathrm{M}\dot{\mathrm{V}}{\mathrm{O}}_{\mathrm{2}}$) was calculated as $\mathrm{M}\dot{\mathrm{V}}{\mathrm{O}}_{\mathrm{2}}$ = CF × [($\mathrm{P}{\mathrm{a}}_{\mathrm{O}}{}_{{}_{\mathrm{2}}}$ − ${\text{Pv}}_{{\text{O}}_2}$) × (c/760)] × heart weight, where CF is coronary flow (ml·min⁻¹·g wet weight⁻¹), ($\mathrm{P}{\mathrm{a}}_{\mathrm{O}}{}_{{}_{\mathrm{2}}}$ − ${\text{Pv}}_{{\text{O}}_2}$) is the difference in the partial pressure of oxygen (Po₂, mm Hg) between perfusate and coronary effluent, and c is the Bunsen solubility coefficient of O₂ in perfusate at 37°C (22.7 μl O₂·atm⁻¹·ml⁻¹). These physiologic data were collected by an investigator blinded to mouse genotype.

¹³C-magnetic resonance spectroscopy and isotopomer analyses.

After 30 min of perfusion with the ¹³C labeled substrates noted above and unlabeled glucose, hearts were freeze clamped and subsequently extracted for analysis by magnetic resonance spectroscopy to determine substrate utilization by the CAC as previously described (20, 25, 31, 39). Briefly, cardiac perfusion with differentially labeled ¹³C substrates permits unique identification of the fuels that supply the CAC. Each substrate produces a differentially labeled two-carbon acetyl-CoA molecule (or unlabeled acetyl-CoA from glucose and endogenous substrates), which subsequently incorporates into the CAC at citrate. Downstream CAC intermediates are specifically labeled according to the ¹³C positions from the original substrate. The five-carbon CAC intermediate α-ketoglutarate is in equilibrium with cellular glutamate stores, which exist at higher concentrations and are therefore more easily detectable than the CAC intermediates. The exact labeling patterns from α-ketoglutarate transfer to glutamate and the relative abundance of each isotopomer is measured by ¹³C-magnetic resonance spectroscopy to provide a fractional contribution (Fc) of each substrate to the CAC. The absolute flux for the CAC and oxidative flux for individual substrates, indicating the oxidation rate of each substrate, were calculated as previously described (22, 25, 36).

Protein immunoblotting.

Western blots were performed as previously described (19, 20). Briefly, 50 μm of total protein extract from nonperfused heart tissue were separated electrophoretically and transferred to PVDF membranes. Membrane total protein binding was evaluated with the Pierce Reversible Protein Stain Kit (Thermo Scientific, Rockford, IL). Membranes were subsequently probed with the following antibodies: total acetyl-CoA carboxylase (recognizing both α- and β-isoforms; ACC; no.3662), total phospho-acetyl-CoA carboxylase (p-ACC; Ser 79, no. 3661), pyruvate dehydrogenase (PDH; no. 2784), AMP-activated protein kinase (AMPK; no. 5831), phosphor-AMPKα (p-AMPK; Thr172, no. 2535; Cell Signaling Technology, Danvers, MA), phospho-pyruvate dehydrogenase (p-PDH; Ser293, no. ABS204; EMD Millipore, Billerica, MA), and malonyl CoA-decarboxylase (MCD; 15265–1-AP; Proteintech, Chicago, IL). Antibodies for detection of pyruvate dehydrogenase kinase 2 (PDK2) and pyruvate dehydrogenase kinase 4 (PDK4) were obtained as generous gifts from Dr. Gebre Woldegiorgis (Oregon Health Sciences University, Beaverton, OR) and Dr. Robert Harris (Indiana University School of Medicine, Indianapolis, IN), respectively. Immunoblot images were measured by densitometry analysis using ImageJ (National Institutes of Health, Bethesda, MD). To normalize for protein loading, the band density of proteins without a phosphorylated form was divided by the density of the total protein staining in the region of the membrane at the predicted molecular weight of the protein of interest. For phosphorylated proteins, the density of the phosphorylated band was divided by the density of the total protein band.

Tissue and serum substrate analyses.

Nonperfused cardiac tissue underwent Folch lipid extraction using chloroform/methanol to purify total triglycerides, which were subsequently measured with a triglyceride determination kit (Wako Diagnostics, Mountain View, CA). Serum substrate levels were measured by enzymatic kits for lactate (Trinity Biotech, Jamestown, NY), total ketones, nonesterified fatty acids, and triglycerides (Wako Diagnostics, Mountain View, CA). Whole blood glucose levels were measured with the Contour Next glucometer (Bayer, Whippany, NJ).

Electron microscopy.

Cardiac tissue from the LV of nonperfused hearts was fixed with 4% glutaraldehyde in sodium cacodylate buffer immediately upon euthanasia, mounted in Epon Araldite epoxy resin blocks, cut by ultramicrotome, and visualized on a JEOL 1230 transmission electron microscope.

Statistical analysis.

Reported values are means ± SE in figures and text. Statistical significance for echocardiographic data was determined by repeated-measures two-factor ANOVA with postestimation pairwise comparisons for genotype and time point. Cardiac triglyceride and serum substrate data were analyzed by two-factor ANOVA with postestimation pairwise comparisons for genotype and experimental condition. The Student’s unpaired two-tailed t-test was used for all other comparisons. Group differences were considered significant for P < 0.05.

RESULTS

Cardiac function in experimental sepsis.

We evaluated cardiac function after CLP both in vivo (echocardiograms) and ex vivo (working heart perfusions). Echocardiography allows serial measurement over time in the same animal but is affected by loading conditions, which cannot be measured noninvasively. Hypovolemia and decreased vascular resistance during sepsis affect preload and afterload, respectively. Ex vivo functional measurements control for loading conditions by setting the same preload and afterload for all experimental groups.

Cardiac function, assessed by shortening fraction, initially increased in both WT and Ppara^−/− mice at 8 h after CLP (Fig. 1A). However, at 24 h, shortening fraction continued to rise in WT mice, whereas it decreased in Ppara^−/− mice with a concurrent drop in heart rate. LVEDD and LVESD decreased 8 h after CLP in both WT and Ppara^−/− mice, which may reflect the combined effect of decreased systemic vascular resistance and relative hypovolemia leading to reduced ventricular filling in early sepsis. LVEDD and LVESD returned to baseline values in Ppara^−/− mice at 24 h, potentially related to development of cardiac dysfunction.

Fig. 1.

Echocardiographic evaluation of in vivo heart function. A: cecal ligation and puncture sepsis model (CLP) cohort, n = 16 per genotype; B: sham cohort, n = 8–9 per genotype (diameters normalized to body weight; *P < 0.05 WT vs. Ppara^−/− at that time point, †P < 0.05 WT vs. baseline, ‡P < 0.05 Ppara^−/− vs. baseline).

In the sham cohort, shortening fraction was elevated at 8 h for the WT mice, but it returned to baseline levels by 24 h (Fig. 1B). For the Ppara^−/− mice, the shortening fraction continued to rise and was elevated compared with baseline at 24 h. This was associated with a slight decrease in the Ppara^−/− heart rate. Although there were some minor differences from baseline in the LVEDD, the two genotypes did not differ from each other in this parameter. The Ppara^−/− LVESD was lower than WT at 24 h. Taken together, the echocardiographic data reveal an early increase in heart function after CLP that the Ppara^−/− mice cannot maintain, although they successfully elevate heart function in the sham condition.

Because of the terminal nature of the working heart perfusions, ex vivo cardiac function was only determined at 24 h after CLP. The LV developed pressure was higher in WT mice during sepsis than in Ppara^−/− mice (Fig. 2). The developed pressure demonstrated a nonsignificant increasing trend in WT hearts in sepsis compared with their baseline, whereas Ppara^−/− hearts exhibited no change. The +dP/dT_max, which is a measure of cardiac contractility, was higher in WT mice at both baseline and in sepsis than Ppara^−/− mice. Additionally, the +dP/dT_max trended upwards in sepsis for WT mice compared with no change in Ppara^−/− mice. There was also a trend toward greater diastolic relaxation, as represented by the −dP/dT_min, in WT mice compared with Ppara^−/− mice in sepsis. No differences were noted in heart rate between any of the conditions. Aortic flow, a surrogate for cardiac output, was significantly higher in WT mice in sepsis, as was cardiac power.

Fig. 2.

Ex vivo evaluation of heart function (n = 8–9 per group; *P < 0.05).

Cardiac oxygen consumption was higher in WT mice at baseline and increased in sepsis, while oxygen consumption did not change in Ppara^−/− mice. Coronary flow also increased in the hearts of WT mice with sepsis while it remained unchanged in Ppara^−/− mice. In total, the ex vivo data show that WT mice successfully augment their load-independent cardiac function during sepsis whereas Ppara^−/− mice fail to compensate appropriately.

Substrate fractional contributions to the citric acid cycle and flux during experimental sepsis.

To determine whether substrate utilization differences contribute to cardiac dysfunction during sepsis, we evaluated the fractional contribution of acetyl-CoA to the CAC for each studied substrate (Fig. 3A). The fractional contribution of fatty acids to the CAC was higher in WT hearts at both baseline and during sepsis. Lactate utilization decreased in WT hearts and was higher in Ppara^−/− hearts in sepsis. We found no significant differences between groups or conditions in the contribution of unlabeled or ketone substrates to the CAC. However, nonsignificant trends indicated that utilization of ketones was higher in Ppara^−/− hearts in sepsis than in WT and the fractional contribution of unlabeled substrate decreased in sepsis in Ppara^−/− hearts. The unlabeled fraction is composed predominately of exogenous glucose, endogenous glycogen, and endogenous triglycerides.

Fig. 3.

Myocardial substrate utilization in sepsis. A: substrate fractional contribution to citric acid cycle (CAC); B: substrate flux through CAC (n = 8–9 per group; *P < 0.05).

Fractional contribution and oxygen consumption data were used for calculating estimated CAC and substrate fluxes (39). Total CAC flux was similar between groups at baseline (Fig. 3B). However, WT hearts increased total CAC flux during sepsis whereas Ppara^−/− hearts did not, resulting in a significant difference between groups in sepsis. Fatty acid flux was higher in WT hearts at both baseline and in sepsis. WT hearts increased their fatty acid flux from baseline to sepsis, which did not occur in Ppara^−/− hearts. There were no differences between groups or conditions in lactate or ketone flux. Although WT hearts increased overall CAC flux during sepsis, there was no difference in lactate and ketone flux in sepsis for WT mice.

Cardiac tissue triglyceride.

Cardiac triglyceride levels did not differ in baseline conditions between the two genotypes. These levels dropped dramatically in the Ppara^−/− hearts in the sham condition but were maintained in the WT hearts. Only in sepsis did the WT hearts demonstrate decreased triglyceride concentrations that resembled those of Ppara^−/− hearts (Fig. 4).

Fig. 4.

In vivo cardiac tissue triglyceride levels (n = 6–7 per group; *P < 0.05; an interaction effect was observed between genotype and condition on two-factor ANOVA analysis).

Protein expression of enzymes controlling lactate and fatty acid metabolism.

To evaluate the regulatory mechanisms that control substrate utilization in the myocardium, we measured the protein levels and phosphorylation states of various enzymes that control carbohydrate and fatty acid oxidation (Fig. 5). PDH phosphorylation increased during sepsis in both genotypes. This change would inhibit pyruvate entry into the CAC via acetyl-CoA. In our ex vivo working heart perfusions, the majority of pyruvate is generated from glucose (via glycolysis) and lactate. Both PDK2 and -4, which phosphorylate and thereby inactivate PDH, were increased in sepsis. The increase of PDK4 in sepsis was higher in WT hearts compared with Ppara^−/− hearts, which is consistent with known PPARα regulation of PDK4 (54). The summative effect of these changes would be to reduce the contribution of carbohydrates and lactate to the CAC.

Fig. 5.

In vivo protein expression and phosphorylation states of regulators of pyruvate incorporation into CAC and fatty acid oxidation (FAO) (n = 3–4 per group; *P < 0.05). PDH, pyruvate dehydrogenase; PDK, pyruvate dehydrogenase kinase; ACC, total acetyl-CoA carboxylase; MCD, malonyl CoA-decarboxylase.

ACC produces malonyl-CoA, which decreases FAO through allosteric inhibition of carnitine palmitoyltransferase I, a mitochondrial fatty acid transporter and rate-limiting step in this pathway. Dephosphorylation of ACC results in its activation, thereby reducing FAO. Malonyl CoA-decarboxylase (MCD) antagonizes this signal by breaking down malonyl-CoA and thereby augments FAO. We observed that ACC phosphorylation decreased in sepsis in both genotypes, which would tend to decrease FAO. MCD levels, however, increased in WT hearts in sepsis, but not in Ppara^−/− hearts, which could explain the augmentation of fatty acid flux noted in WT hearts alone in sepsis.

Serum substrate levels.

Blood glucose levels were not different between genotypes in the baseline, control condition. Glucose levels fell in both the sham and CLP conditions where they were higher in WT than in Ppara^−/− mice (Fig. 6). Blood glucose levels were lower in both WT and Ppara^−/− mice in sepsis than in the sham condition. Serum lactate levels decreased from control to sham and CLP conditions and were higher in Ppara^−/− mice in both the baseline and septic state. Ketone levels increased dramatically from the baseline condition and were higher in WT mice than Ppara^−/− mice in both the sham and CLP conditions. These levels decreased from the sham to the CLP condition in WT mice. Interestingly, serum nonesterified fatty acids increased in the Ppara^−/− mice and were higher than those of the WT mice in both sham and CLP conditions. Serum triglyceride levels did not differ between control and sham conditions but decreased in both genotypes in sepsis. As with the nonesterified fatty acids, triglyceride levels were higher in the septic Ppara^−/− mice.

Fig. 6.

Whole blood glucose and serum substrate levels (n = 9–16 per group; *P < 0.05; interaction effects between genotype and condition were observed on two-factor ANOVA analysis for lactate and ketone substrates). NEFA, nonesterified fatty acid; TG, triglyceride.

Ultrastructural cardiac changes in experimental sepsis.

Histological evaluation with light microscopy in our previous study revealed areas of myocardial degeneration in Ppara^−/−, but not WT, hearts (51). To further evaluate these findings and to assess mitochondrial ultrastructure as a marker of mitochondrial injury, we obtained transmission electron micrographs of heart tissues after CLP. Mitochondria appeared normal in both genotypes. However, sections obtained from Ppara^−/− LV myocardium demonstrated multiple areas of cardiomyocyte degeneration consistent with those observed in our previous studies. Although the contractile apparatus of the affected cells was significantly degraded, the mitochondria had intact ultrastructure (Fig. 7). These areas of degeneration were very rare in WT myocardium and, even when present, were of much lower severity than in Ppara^−/− tissue.

Fig. 7.

Electron micrographs of myocardium after CLP. *Areas of contractile apparatus degeneration. Mitochondria identified with arrowheads on ×50,000 panels (n = 3–4 per group).

DISCUSSION

Cardiovascular function is a critical determinant of survival in sepsis. We previously demonstrated that sepsis survival is reduced in Ppara^−/− mice. Decreased survival was associated with cardiac injury (51). In the current study, we show that WT mice actually augment cardiac function during this early phase of sepsis, associated with increased FAO and total CAC flux. Ppara^−/− mice, however, did not maintain enhanced cardiac function and had impaired FAO both at baseline and during sepsis. Importantly, Ppara^−/− myocardium failed to compensate for lower oxidation of fatty acids by increasing flux of other substrates. This failure to increase substrate oxidation in Ppara^−/− hearts may contribute to cardiomyocyte degeneration, the inability to maintain cardiac function in sepsis, and the early demise of Ppara^−/− mice after CLP. Thus the current work confirms that PPARα and FAO play an important role in the cardiovascular adaptation to sepsis (11, 12). Our results also suggest that the increased mortality in septic human patients with low PPARα expression may be partially due to compromised cardiac function.

Functional response to sepsis.

In our study, WT mice augmented cardiac function during early experimental sepsis assessed both in vivo and ex vivo. This result differs from previous studies, which have described significant cardiac dysfunction and decreased FAO early in their rodent models of septic shock (5, 11, 12, 14, 48, 56). It is important to note that many of these studies employed an LPS injection model of sepsis or severe CLP models with very high, early mortality, some with as much as 75% at 72 h after injury. These models have been shown to induce inflammatory cytokine release in excess of that observed in clinical sepsis and have been determined to have lower clinical relevance (6, 8, 43–45). These models may more closely replicate the late stages of severe septic shock, which are encountered clinically in only the most severe subset of afflicted patients. We designed our CLP sepsis model to have high clinical relevance by administering antibiotics, fluids, and analgesics and by titrating CLP severity to achieve 20–30% mortality at 7 days (51), which much more closely approximates human short-term survival rates (1).

Similar to our findings, several other animal studies have demonstrated an early “hyperdynamic” phase in the cardiovascular response to sepsis (24, 28, 29, 57). Furthermore, human studies of cardiac performance in sepsis have shown an early increase in cardiac function, which is subsequently followed by cardiac dysfunction in severe shock states (3, 15, 16, 27, 35, 41, 58). In particular, Calvin et al. showed that septic patients without shock had no decrease in their ejection fraction and had elevated contractility (3). One of the challenges of interpreting studies of human sepsis is that researchers can rarely define its exact onset. It is therefore difficult to determine where in the time course of sepsis a specific patient is when they present for medical care. Although not fully representative of sepsis, two well-designed human experimental trials clearly demonstrate increased ejection fraction and cardiac output shortly after LPS injection (26, 53). Thus our findings in WT hearts after CLP are likely representative of an early, relatively well-compensated cardiac state during sepsis. Since we previously demonstrated that much of the mortality in septic WT mice occurs after 24 h (51), cardiac function could potentially decline after our studied time points.

Ppara^−/− mice initially improved cardiac function 8 h after CLP surgery. However, this functional improvement was not maintained at 24 h and Ppara^−/− hearts had reduced ex vivo (load-independent) cardiac function 24 h post-CLP surgery compared with WT mice. The observation that the sham-operated Ppara^−/− mice maintained elevated heart function 24 h after surgery demonstrates that they are capable of augmenting cardiac performance under nonseptic stressors. Thus Ppara^−/− hearts had an abnormal functional response to sepsis. We speculate that this adverse myocardial response contributes to the increased mortality to sepsis in these transgenic mice.

Metabolic response to sepsis.

The augmented cardiac function in septic WT mice was associated with increased oxygen consumption and coronary flow. Human studies have also shown that coronary blood flow increases in sepsis (9, 10). Interestingly, although overall CAC flux increased in our WT hearts, fatty acids were the only studied substrate to increase flux during sepsis. Decreased cardiac tissue triglyceride levels in septic WT mice indicate that endogenous lipid stores were depleted, likely due to increase utilization or possibly reduced uptake. In contrast, the septic Ppara^−/− mice did not increase overall CAC or any individual substrate flux during sepsis. Ppara^−/− mice demonstrated decreased cardiac triglyceride levels with the relatively minor stress of the sham condition, which were not lower in the CLP cohort, indicating that endogenous triglyceride reserves had likely already been exhausted. This further substantiates the concept that Ppara^−/− hearts could not augment FAO in sepsis. Our results suggest that increased FAO is part of the normal myocardial adaptation to early sepsis and that this metabolic change is necessary to augment cardiac function in sepsis.

This conclusion is supported by other studies that show associations between depressed heart function in sepsis and decreased cardiac fatty acid utilization. Dhainaut et al. (10) demonstrated that in established human sepsis with cardiac dysfunction that myocardial extraction of fatty acid from the blood was reduced. Rossi et al. (46) found that patients who died of septic shock had intracardiomyocyte accumulation of lipids indicative of decreased FAO. Using a severe mouse LPS injection model of sepsis, other authors have shown that profoundly decreased heart function was associated with depressed FAO and cardiac triglyceride accumulation (11, 12, 48). This association between diminished myocardial function and fatty acid utilization has been demonstrated in other animal models of sepsis (48, 60) and is seen in other forms of heart failure (21, 38, 52). Thus our findings support an emerging paradigm of the importance of FAO for maintaining cardiac function during sepsis. To our knowledge, we are the first to report that FAO increases in conjunction with augmented heart function during the early phase of sepsis.

We previously showed that mRNA expression of PPARα and its downstream transcriptional targets is reduced in sepsis, but protein levels of these enzymes actually increased 24 h after CLP before dropping later (51). This finding is consistent with our current results showing increased fatty acid flux in septic WT mice and indicates that myocardial substrate utilization in sepsis is modulated by additional mechanisms beyond transcriptional regulation. Theoretically, at later times during sepsis, the reduced PPARα mRNA levels could cause a physiologically important decrease in protein levels of PPARα-regulated metabolic enzymes and FAO. To evaluate this possibility, future experiments evaluating metabolic substrate changes during prolonged sepsis are necessary.

ACC phosphorylation decreased in both groups during sepsis, which would be predicted to reduce FAO. However, MCD levels increased in the septic WT mice compared with baseline. MCD promotes FAO and it appears that the changes in MCD predominated over those in ACC phosphorylation in septic WT mice. Furthermore, MCD levels were higher in WT mice than Ppara^−/− mice, which may partially explain the higher level of FAO in that group of mice. This finding is consistent with previous research that has shown MCD to be transcriptional target of PPARα (4, 32). Metabolic regulation is complex and incompletely understood. Accordingly, other unexplored regulatory mechanisms may also be operative during sepsis.

It is also noteworthy that the lactate, ketone, and unlabeled substrate fractional contributions and fluxes did not increase during sepsis in either WT or Ppara^−/− mice. Previous studies with LPS-induced sepsis found that glucose oxidation did not increase to compensate for decreased FAO, but these studies did not evaluate either ketone or lactate substrates (11, 48, 56). The Randle cycle would predict that glucose oxidation could increase to compensate for the decreased FAO during sepsis. However, both our and previous studies have shown molecular changes during sepsis that would inhibiting glucose oxidation. For example, prior work showed decreased mRNA expression of glucose transporter 4 (GLUT4) and PDK4 (11). In the current study, we expand upon this work to show greater PDK2 and -4 protein levels. As would be expected from these changes, PDH phosphorylation increased. PDH phosphorylation inhibits pyruvate (from carbohydrates or lactate) entry into the CAC and, accordingly, oxidation during sepsis. It is possible that the changes in PDK2 and -4 are affected by decreased food intake after CLP. However, the previous studies in LPS-induced sepsis demonstrated that PDK4 expression was independent of fasting (11). Furthermore, septic humans typically have little to no food intake while acutely ill. In total, our results confirm that sepsis affects substrate oxidation at multiple levels.

Circulating substrate levels are affected by production (including ingestion) and tissue uptake, storage, and utilization. Thus it is not possible to draw definitive conclusions regarding cardiac metabolism from circulating substrate levels without directly measuring those other factors. These levels do indicate, however, what energetic resources are mobilized to the tissues and available for uptake. Overall, baseline circulating substrate levels varied little between genotypes despite the whole body nature of the Ppara^−/− mice. Blood glucose concentration decreased in sepsis from baseline and sham levels, which differs from the hyperglycemia often observed in human sepsis (42, 59). Ppara^−/− mice had lower blood glucose in sepsis, which is consistent with the known mechanistic reliance of hepatic gluconeogenesis on fatty acid β-oxidation which is activated by PPARα signaling (7, 23, 34, 37). Due to the deactivation of PDH, however, it is uncertain that myocardial glucose oxidation would have increased even if more carbohydrate substrate were available. As previously described in fasted mice (23), serum lactate levels decrease from control to sham and CLP conditions in both genotypes, possibly because lactate is consumed as a gluconeogenic substrate (37). In that context, the higher lactate levels observed in septic Ppara^−/− mice might be caused by impaired gluconeogenesis due to insufficient FAO. Alternatively, increased lactate levels in sepsis could be a product of poor tissue perfusion and cellular hypoxia due to inadequate cardiovascular function. Serum ketone levels rose in the sham condition due to the fasting state in both WT and Ppara^−/− mice, but the increase was much less dramatic in Ppara^−/− mice due to the dependence of ketogenesis on FAO (23, 34). Serum ketone levels decreased substantially in the CLP condition from their sham levels for WT mice. This phenomenon has been described previously and attributed to increased utilization by peripheral tissues in sepsis (30). The increased serum levels of nonesterified fatty acids in the sham and CLP operated Ppara^−/− mice suggest that this substrate was mobilized from adipose depots but insufficiently utilized by the tissues. The drop in serum triglyceride levels observed in both genotypes could be attributed to increased cellular utilization, which is impaired in the Ppara^−/− mice resulting in higher levels in sepsis relative to WT mice. Taken together, these data indicate that Ppara^−/− mice exhibit a relative deficit of nonlipid substrates in sepsis and a surplus of lipid substrates they are unable to utilize. These results point to metabolic failure as a contributor to the increased mortality observed in Ppara^−/− mice in sepsis.

No evidence of mitochondrial damage in either WT or Ppara^−/− myocardium was observed by transmission electron microscopy in our experiments. This is at variance with other animal studies that report significant alteration in mitochondrial ultrastructure indicative of injury and dysfunction (55, 61). We attribute these findings again to the milder nature of model and the early time point at which we performed our assessment. It is possible that mitochondrial changes develop later as sepsis progresses, cardiovascular function worsens, and FAO is disrupted. The areas of frank cardiomyocyte degeneration in Ppara^−/− hearts are of great interest and may represent cellular necrosis due to insufficient energy provision.

Study limitations.

Our findings are limited by two important considerations. First, the use of constitutive, whole body Ppara^−/− mice limits the ability to attribute our findings directly to myocardial PPARα expression vs. secondary myocardial changes due to the effects of PPARα on other organ systems. Additionally, dysfunction occurring in other organ systems may contribute to the increased mortality in Ppara^−/− mice during sepsis. Mice with cardiac-specific PPARα deletion are not available, although we are currently developing such a transgenic mouse. It will be necessary to confirm our results in that model.

Because we could accommodate only a limited number of subjects in the working heart experiments, a second limitation is that we did not perform metabolic evaluations in sham mice. Because the Ppara^−/− mice have a constitutive, whole body deletion; we were concerned that the Ppara^−/− mice could have developed baseline compensatory metabolic mechanisms to accommodate for their genetic difference. Thus we used naïve mice as the control comparison because we felt it was most important to characterize baseline cardiac substrate utilization during normal, healthy conditions. However, this approach prevented a direct evaluation of the effect of anesthesia, laparotomy, fluid and antibiotic administration, and fasting on ex vivo cardiac function and substrate utilization. Our results are still relevant, however, because clinical sepsis almost uniformly includes elements of decreased oral intake, fluid repletion, and sedation in addition to the infectious insult. Thus, any translational considerations in sepsis research need to address the metabolic and hemodynamic changes within the context of all of these factors, not isolated from them. Additionally, this limitation is mitigated by the in vivo functional data rendered by echocardiography, which showed that Ppara^−/− sham mice did not have lower heart function than the WT sham mice. Furthermore, all hearts in the ex vivo working heart experiments were perfused with the same substrate buffer, thus abrogating any effect from differing circulating substrate concentrations, yet functional differences persisted.

Conclusion.

We demonstrate that cardiac function increases early in sepsis and is associated with greater FAO. Ppara^−/− mice are unable to augment FAO to the same degree as WT mice and do not compensate by increasing myocardial oxidation of other substrates. Ppara^−/− heart function therefore appears to decline as sepsis progresses due to metabolic limitations, which potentially contributes to accelerated mortality. Thus we propose that augmentation of FAO is a necessary adaptive response during early sepsis to promote cardiac function. Ppara^−/− mice likely represent a more severe sepsis phenotype in which survival is constrained by metabolic limitations. Our current findings suggest a potential therapeutic strategy in which short-term augmentation of FAO in the heart could provide both functional and survival benefit in sepsis.

GRANTS

This work was supported American Heart Association Grant 12SDG12040342 (to S. W. Standage) as well as funds from the Seattle Children’s Research Institute Center for Developmental Therapeutics Intercenter Grant (to S. W. Standage and A. K. Olson). The microscopy work was supported in part by the Vision Research Center Core Grant P30-EY-01730.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

S.W.S., J.K.M., M.A.P., W.C.L., and A.K.O. conception and design of the study; S.W.S., B.G.B., T.O.K., D.R.L., and A.K.O. performed experiments; S.W.S., M.A.P., W.C.L., and A.K.O. statistical analysis and interpretation of results; S.W.S., J.K.M., W.C.L., and A.K.O. drafting, critical revision, reading, and approval of the manuscript.