Introduction

Sepsis is known to diminish oxidative metabolism in the heart and other tissues (Rudiger and Singer 2007). Myocarditis, caused mainly by bacterial or viral infections and autoimmune diseases (Hekimian and Combes 2017), is a multietiological pathological immune process involving myocardial inflammation with various clinical symptoms and outcomes, such as acute heart failure, chronic dilated cardiomyopathy, and even sudden cardiac death (Bracamonte-Baran and Cihakova 2017; Zhang et al. 2013; Sagar et al. 2012).

Notably, the occurrence and development of myocarditis involve various pathophysiological mechanisms of inflammation. The activation of multiple inflammatory regulators and related transcription factors can trigger the excessive inflammatory response, which is essential for the malignant development of cardiomyopathy (Maier et al. 2012; Akira et al. 2006). The inflammatory response and left-ventricular dysfunction induced by lipopolysaccharide (LPS) are mediated, in large part, by the host molecule Toll-like receptor 4 (TLR4) (Nemoto et al. 2002). TLR4 can trigger members of the nuclear factor kappa B (NF-κB) family and other transcription factors involved in the immune response and different cardiovascular diseases, such as ischemia–reperfusion injury and cardiac hypertrophy (Oyama et al. 2004; Timmers et al. 2008; Ha et al. 2005). Previous studies have reported variable effects of LPS on cardiac fuel and energy metabolism, suggesting that targeting inflammatory-metabolic “crosstalk” could improve outcomes in this syndrome.

The Pellino family interacts with Pelle/IRAK and has been identified to mediate the transcriptional regulation of proinflammatory genes in innate immunity through ubiquitination and then activate TLR and/or T-cell receptor (TCR) signaling (Grosshans et al. 1999b; Butler et al. 2007; Moynagh 2014). Pellino1, in particular, has been shown to play a dispensable role in activating the NF-κB family or other signal transduction pathways, which is important in T-cell activation and differentiation by the MyD88/TRIF-dependent TLR pathway (Chang et al. 2009; Kawai and Akira 2007; Choi et al. 2006; Vallabhapurapu and Karin 2009). A previous report has shown that the loss of Pellino1 can attenuate the activation of the TRIF-dependent NF-kB family due to the hyperactivation and nuclear accumulation of c-Rel in response to T-cell receptor-CD28 (TCR-CD28) signaling (Chang et al. 2011; Smith et al. 2011). In this study, we further explored whether silencing Pellino1 has therapeutic effects on LPS-induced myocarditis through multiple known or unknown regulatory mechanisms.

Currently, the immune response is the best-known underlying pathogenic mechanism and therapeutic target of bacterial myocarditis (Zhou and Yu 2017). However, the therapeutic effect is limited, and more strategies are needed to complement or replace this treatment (Kindermann et al. 2012). Since the close relationship between metabolism and signaling programs of immune cells may show a critical physiological regulation in normal or disease conditions by maintaining nutrient uptake and metabolic homeostasis (Andrejeva and Rathmell 2017), we used metabolomics to explore a new possible mechanism in the treatment of bacterial myocarditis.

In our study, we attempted to use LC–MS/MS to analyze the differences in metabolic compositions that could be vital in the occurrence and progression of LPS-induced myocarditis between samples exposed under different conditions. Thus, we could further investigate the therapeutic effect of silencing Pellino1 through metabolic mechanisms on LPS-induced myocarditis.

Materials and methods

Cell culture

NRCMs were prepared from 1-day-old Sprague–Dawley rats (Charles River) as previously described (Paradis et al. 2015). The cells were subsequently treated with 10 µg/mL LPS or PBS for 12 h. Adenovirus silencing Pellino1 (si-Pellino1, 2 × 107 pfu/mL) or adenoviral GFP was purchased from Shanghai GeneChem Co., Ltd. NRCMs were infected with adenovirus expressing GFP (Ad-GFP) or silencing Pellino1 (si-Pellino1) on day 1 after plating and on day 2 were treated with 10 µg/mL LPS or PBS for 12 h, after which mRNA was isolated for RT-PCR. The cells were lysed to obtain protein for western blot analysis.

Western blot analysis

NRCMs were lysed in RIPA buffer (P0013C, Beyotime) supplemented with 1 mM PMSF (ST505, Beyotime). Thirty micrograms of total protein were subjected to electrophoresis on 10% SDS-PAGE gels (Beyotime Biotechnology, China) and transferred to polyvinylidene fluoride (PVDF) (Merck-Millipore, Shanghai, China) membranes. After the samples were blocked with 5% milk powder in TBS-Tween for 1 h, the proteins were probed with primary antibody at 4 °C overnight. After the membranes were washed three times with TBS-Tween, they were incubated for 1 h with the corresponding secondary antibody conjugated to horseradish peroxidase and then subjected to enhanced chemiluminescence (ECL) for detection of protein bands. The primary antibodies used in this study were as follows: anti-CC3 (1:1000 dilution, #9664, Cell Signaling Technology), anti-C3 (1:1000 dilution, #9662, Cell Signaling Technology), anti-Bcl2 (1:1000 dilution, #3498, Cell Signaling Technology), anti-BAX (1:1000 dilution, #5023, Cell Signaling Technology), and anti-Pellino1 (1:200 dilution, sc-271065, Santa Cruz Biotechnology), and β-Actin (1:1000 dilution, #4967, Cell Signaling Technology) was used as a loading control (n = 3).

RNA analysis and real-time quantitative PCR (qRT-PCR)

Total cellular RNA isolation from NRCMs was performed using the RNazol B method and a Qiagen RNeasy kit, according to the manufacturer’s instructions. RNA was reverse transcribed (Applied Biosystems) using random hexamer priming. Real-time qRT-PCR was performed using SYBR Green reagent (Applied Biosystems) and rat-specific primers (Table S1) on the ABI Prism 7500 Sequence Detection system. GAPDH was used as an internal control. The relative gene expression levels were calculated using the 2 − △△Ct method (n = 3).

Flow cytometry

We incubated all collected viable and dead cells with propidium iodide (PI) and Annexin-V (Fcmacs Biotech Co., China) in the provided binding buffer at room temperature for 30 min in the dark, according to the manufacturer’s recommendations. Then, the flow cytometry method was applied to determine the relative cell apoptosis ratio (n = 3).

Metabolite extraction

Cultured NRCMs were quickly washed twice with ice-cold phosphate-buffered saline (PBS) in a cold room to remove medium components and then quickly rinsed with Milli-Q water. After the removal of water, 400 μL of 80% CH3OH/Milli-Q water was added to each culture plate. The cells were placed in liquid N2 for 10 min, thawed, and sonicated for 10 min. This process was repeated three times to completely extract the metabolites from the cells. Combined samples from the extraction process were then centrifuged at 10,000g for 5 min to pellet insoluble debris at 4 °C, with the supernatant transferred to microcentrifuge tubes and dried with a speed vacuum. Then, each dried sample was resuspended, vortexed, centrifuged, and transferred to autosampler vials for LC–MS analysis. Each group had six biological replicates.

LC–MS and data analysis

Separation of metabolites was performed by reversed-phase LC (HP1100, Agilent Technologies, Santa Clara, CA, USA) using a reversed-phase XBridge C18 column (1.7 μm particle size, 1 × 150 mm, Waters, Milford, MA, USA), and 8 μL of each sample was injected on the C18 column for analysis. Mobile phase A consisted of H2O/0.1% formic acid, and mobile phase B consisted of ACN with 0.1% formic acid with a program: 1% B at 0–1 min, 15% B at 3 min, 70% B at 5 min, 85% at 9 min, 100% at 10–12 min, and subsequently return to the initial conditions with 2 min for equilibration. Furthermore, the MS program was conducted on a 6545 Quadrupole-Time of Flight system (all devices from Agilent Technologies, Santa Clare, CA, United States) as follows: both positive- and negative-ion modes with drying gas 300℃ flow 6 L/min, sheath gas 340℃ flow 11 L/min, nebulizer gas 35 psig, capillary voltage 4000 V, and fragmental voltage 135 V. The data collection (MS 100–3200 m/z, MS/MS 30–3200 m/z) was acquired by both centroid and profile stored in autoMSMS scan mode with reference masses at m/z 112.05087 and 922.009798 were set as online accurate mass calibration.

MassHunter Workstation software (version B.07.00; Agilent Technologies) was used to export mzdata format from the acquired MS data (.d). Data analysis was performed by the XCMS Online at https://xcmsonline.scripps.edu/ using three steps: data upload, parameter selection, and result interpretation. A list of the intensities of all the peaks detected was generated using the retention time and the mass-to-ratio data pairs as the parameters for each ion. MS/MS spectra of the selected putative identifications were retrieved and matched with entries in the Metlin, Massbank, the Human Metabolome Database (HMDB). The metabolomics data resulting directly from XCMS Online were used to generate the cloud plot. The binning data were normalized to the total area. PCA, partial least-squares discriminant analysis (PLS-DA), orthogonal partial least-squares discriminant analysis (OPLS-DA), and metabolites and metabolic set enrichment analysis were performed with the web-based software MetaboAnalyst 3.0 (http://www.metaboanalyst.ca).

Statistical analyses

Data was presented as mean ± standard error of the mean (SEM). Using GraphPad Prism 4.0 (GraphPad software Inc., CA, USA), statistical significance among multiple groups was evaluated by one-way analysis of variance (ANOVA) with the Bonferroni post hoc test. A two-tailed P value < 0.05 was considered statistically significant.

Results

Silencing Pellino1 alleviates LPS-induced cardiomyocyte apoptosis and inflammation

To verify the role of Pellino1 in LPS-induced septic cardiomyopathy, we first examined its expression in cardiomyocytes after LPS treatment. Western blot analysis showed that the expression of Pellino1 increased gradually with LPS treatment time (Fig. 1a, b). To further investigate whether silencing Pellino1 can ameliorate LPS-induced apoptosis, we used adenovirus inhibition of Pellino1 before LPS treatment, significantly reducing the Pellino1 expression levels in NRCMs (Fig. 1c, d). Flow cytometry showed that the LPS-induced increase in apoptosis was partially reversed by silencing Pellino1 (Fig. 1e, f). Besides, given the relationship between septic cardiomyocytes dysfunction and several apoptotic indexes, western blot analysis showed that the protective effect of silencing Pellino1 on septic cardiomyocytes dysfunction was mediated by reduced cleaved caspase-3 and Bax expression but not increased Bcl2 expression (Fig. 1g–i). In addition, LPS-induced inflammatory cytokine release was partially abolished by silencing Pellino1. Collectively, these data suggest that inhibition of Pellino1 attenuates LPS-induced cardiomyocytes dysfunction and promotes survival (Fig. 1j).

Fig. 1
figure 1

Silencing Pellino1 alleviates LPS-induced cardiomyocyte apoptosis and inflammation in vitro. a, b Representative western blot of the expression of Pellino1 in the LPS-induced cardiomyocytes. The bar graph shows that the expression of Pellino1 increased gradually with LPS treatment time. β-actin was detected as the loading control. c, d The efficacy of silencing Pellino1 via an adenovirus was assessed by western blotting. β-actin was detected as the loading control. e, f Apoptosis of cardiomyocytes exposed to LPS with or without the Pellino1-silencing adenovirus. The right bar graph shows the results of statistical analysis of the apoptosis ratios. g–i Representative western blot of the expression of Bax, Bcl2, CC3, and C3 in LPS-induced cardiomyocytes with/without si-Pellino1 treatment. The relative protein levels were normalized to the β-actin levels. The data are shown as the mean ± standard error of the mean of triplicates and are representative of three independent experiments performed. j Representative qRT-PCR of mRNA expression of TNF-α, IL-1β, and IL-6 in the LPS-induced cardiomyocytes with/without adenovirus silencing Pellino1. All mRNA expression was normalized to that of GAPDH. n = 6, results are expressed as the mean ± SD, *P < 0.05, **P < 0.01, ***P < 0.001, compared with the ctrl or GFP group

Full size image

Silencing Pellino1 alleviates the effects of LPS on cardiac fuel and energy metabolism

Given the regulatory role of PGC-1 coactivators in maintaining a high-capacity mitochondrial system, we next assessed the effect of silencing Pellino1 on the myocardial expression of LPS-induced cardiac PGC-1α and β genes. RT-PCR showed that LPS-mediated rapid downregulation of the PGC-1α and β mRNA levels was preserved in the Pellino1-silenced cells (Fig. 2a, b). The LPS-induced reduction of the PGC-1 coactivators that mediate metabolic transcription factors [PPARα, ERRα, NRF1 (Nrf1)] was partially reversed by silencing Pellino1 (Fig. 2c–e). However, silencing Pellino1 had no effect on the expression of PPARγ and PPARδ (Fig. 2c). Interestingly, the expression of downstream gene targets of PGC-1 coactivators involved in mitochondrial function [ATP Synthase β (Atp5b), Citrate Synthase (Cit syn)] and fatty acid β-oxidation (FAO) [MCPT-1 (Cpt1b), PDK4, MCAD (Acadm), CD36] was enhanced at the mRNA level following LPS treatment after silencing Pellino1 (Figs. 2f–g, 3a–d). Similar to the effects on the PGC-1 coactivator and their downstream gene targets, mRNA expression of GLUT4 (Slc2a4), the PGC-1 target gene encoding the glucose transporter, was also increased by silencing Pellino1 through adenovirus treatment (Fig. 3e). In addition, other genes involved in the regulation of FAO, such as PLIN2 and the myocardial lipid metabolic gene angiopoietin-like 4 (ANGPTL4), were not differentially expressed in the inflamed NRCMs and the si-Pellino1 group (Fig. 3f, g). Collectively, these data support the view that LPS-mediated inflammation leads to rapid deactivation of the PGC-1 metabolic gene regulatory circuit, and Pellino1 may be a potential target for cardiac fuel and energy metabolism.

Fig. 2
figure 2

Silencing Pellino1 alleviates the effects of LPS on cardiac mitochondrial function. a–e Representative qRT-PCR of PGC-1α, PGC-1β, PPARα, PPARγ, PPARδ, NRF1, and ERRα mRNA expression following LPS treatment after Pellino1 silencing. f, g qRT-PCR validation of downstream metabolic genes of PGC-1 coactivators, including ATP Synthase β (Atp5b) and Citrate Synthase (Cit syn), involved in mitochondrial function. All mRNA expression was normalized to that of GAPDH. n = 6, results are expressed as the mean ± SD, *P < 0.05, **P < 0.01, ***P < 0.001, compared with the GFP group

Full size image
Fig. 3
figure 3

Silencing Pellino1 alleviates the effects of LPS on cardiac fuel metabolism. a–d qRT-PCR validation of downstream metabolic genes of PGC-1 coactivators, including MCPT-1 (Cpt1b), PDK4, MCAD (Acadm), and CD36 involved in fatty acid β-oxidation (FAO). e Representative qRT-PCR of the mRNA expression of GLUT4 (Slc2a4), the PGC-1 target gene encoding the glucose transporter, followed by Pellino1-silencing adenovirus treatment. f, g Representative qRT-PCR of the mRNA expression of PLIN2 and angiopoietin-like 4 (ANGPTL4) involved in the regulation of FAO and myocardial lipid metabolism, respectively, following LPS treatment after Pellino1 silencing. All mRNA expression was normalized to GAPDH. n = 6, results are expressed as the mean ± SD, *P < 0.05, **P < 0.01, ***P < 0.001, compared with the GFP group

Full size image

LC–MS-mediated metabolomics analysis of LPS-treated or Pellino1-silenced NRCMs

To identify metabolite alterations that are induced by LPS or silencing Pellino1 adenovirus, we used LC–MS to analyze and evaluate all samples. First, the total ion flow diagram of the analyzed QC samples was compared with the spectral overlap by a comparison of the QC sample spectrogram. The results showed that the response strength and retention time of each chromatographic peak overlapped, indicating that the variation caused by instrument error was small in the whole experiment (Fig. S1A). Then, metabolic profiling (Chenomx, Edmonton, AB, Canada) was used to identify 169 differentially abundant metabolites between the LPS-treated cells and the untreated control cells (Table 1). After normalization of the data, univariate and multivariate statistical analyses were used to comprehensively evaluate the effects of silencing Pellino1 on the LPS-treated NRCMs (Figs. S1B–E, S2A–D). Our correlation analysis identified good sequencing stability and negative associations among the samples and metabolites (Fig. 4a).

Table 1 One-way ANOVA (analysis of variation) test of significant levels of all 169 metabolites
Full size table
Fig. 4
figure 4

LC–MS-mediated metabolomics analysis of LPS-treated or Pellino1-silenced NRCMs. a Correlation analysis identified good sequencing stability and negative associations among samples. b Different perspectives of two-dimensional principal components (PC) showed the variable reduction outputs analyzed by orthogonal partial least-squares projections to latent structures-discriminant analyses (OPLS-DA) on metabolites from si-Pellino1 and LPS-treated, LPS-treated, and untreated control NRCMs. c, d OPLS-DA score plots of three groups (GFP, GFP + LPS, si-Pellino1 + LPS) according to the differential metabolic profile showed in 2D (c) and 3D (d); each point represents a tested sample. e One-way ANOVA (analysis of variation) of significant levels of all 169 metabolites, P < 0.05

Full size image

As a multivariate statistical technique applied to determine the relative differences in two or more systems that are large and complex, principal component analysis (PCA) is the most widespread multidimensional data analysis technique, which can systematically reduce the dimensionality of original data required to describe protein dynamics with the preservation of variance. Therefore, to further calculate variable importance in projection (VIP) scores of metabolites, we performed partial least-squares projections for latent structures-discriminant analysis (PLS-DA) of metabolites induced by si-Pellino1 and LPS-treated, LPS-treated, and untreated control NRCMs. Metabolites with VIP scores greater than 1.0 were considered important. To confirm the “goodness” of the model and the predictive quality, we performed orthogonal partial least-squares projections to latent structures-discriminant analyses (OPLS-DA) on data from three groups.

The variable reduction outputs were shown from different perspectives of two-dimensional principal components (PC) contributing a percentage of the variation of a sample (Fig. 4b). As shown in Fig. 4c, d, each point represents a tested sample, and the three groups were clustered separately, suggesting that the metabolic profile was intensely influenced by the treatment with LPS and si-Pellino1. Additionally, the obvious separation among samples from different treatment groups observed in the OPLS-DA score plots according to the differential metabolic profile in each group in both positive and negative ion modes indicated that si-Pellino1 treatment led to significant changes in the metabolites compared with those of the LPS-treated control group. Interestingly, as shown in Table 1, the high − log10(P values) in the OPLS-DA shown in the table could reflect the differences among the three groups of samples. Furthermore, one-way ANOVA of the metabolites showed significant levels of all 169 metabolites (Fig. 4e).

Key metabolic pathways associated with si-Pellino1 treatment

Among the distinguishing metabolites identified by statistical correlations, to further identify which metabolic pathways were the most relevant pathways affected by si-Pellino1 treatment, we generated a metabolome view of the top 50 matched pathways according to P values (Table 2) from enrichment analysis and impact values from topology analysis, as shown in Fig. 5a, b. The topological score (X-axis) and the size of each circle represent the importance of different metabolites and their impact values in the enriched pathways, while the color of each circle based on the P value of the enrichment analysis (Y-axis) represents the statistical significance of the overall metabolic changes in the pathways. The pathway from yellow to red indicates a growing impact (Table 3). From this analysis, the citrate cycle (TCA cycle); fatty acid metabolism; glycolysis or gluconeogenesis; fatty acid elongation in mitochondria; alanine, aspartate, and glutamate metabolism; and glyoxylate and dicarboxylate metabolism were identified as the most perturbed metabolic pathways according to − log(P values) after si-Pellino1 treatment in the LPS-induced NRCMs. Protein interaction analysis found that multiple significantly enriched disturbed metabolic pathways were markedly related to the adipocytokine signaling pathway and PPAR signaling pathway (Fig. 6a, b).

Table 2 Impact analysis of metabolic set associations with Pellion1 treatment relative to the control
Full size table
Fig. 5
figure 5

Metabolome view of the top 50 matched metabolic pathways associated with si-Pellino1 treatment. a Impact analysis of metabolites associated with si-Pellino1 treatment relative to the control according to P values (shown in Table 2) from enrichment analysis and impact values from topology analysis. The topological score (X-axis) and the size of each circle represent the importance of different metabolites and their impact values in the enriched pathways, while the color of each circle based on the P value of the enrichment analysis (Y-axis) represents the statistical significance of the overall metabolic changes in the pathways. b Enrichment analysis of metabolites associated with si-Pellino1 treatment relative to the control. The pathway from yellow to red indicates a growing impact

Full size image
Table 3 Enrichment analysis of metabolic set associations with Pellino1 treatment relative to the control
Full size table
Fig. 6
figure 6

Protein interaction analysis of matched KEGG pathway associations with siPellino1 treatment. a Predicted protein interactions among PGC-1α, PPARα, Slc2a1, CD36, and Cpt1b involved in the adipocytokine signaling pathway. b Predicted protein interactions among PPARγ, PPARα, Angptl4, CD36, Cpt1b, and Acadm involved in the PPAR signaling pathway

Full size image

To investigate the relationships between metabolites and genes that we previously explored in the main enriched metabolic pathways showing alterations, we generated specific correlation networks of the citrate cycle (TCA cycle), fatty acid metabolism, glyoxylate, and dicarboxylate metabolism and propanoate metabolism, as well as valine leucine and isoleucine degradation and beta-alanine metabolism, as shown in Fig. 7. The results shown above verified that si-Pellino1 treatment could alleviate or even reverse the LPS-induced cellular damage by altering cardiac fuel and energy metabolism accompanied by changes in key genes (Cs, Cpt2, and Acadm) and metabolites (3-oxoocotanoyl-CoA, hydroxypyruvic acid, lauroyl-CoA, and NADPH).

Fig. 7
figure 7

Interaction analysis of gene and metabolite associations with si-Pellino1 treatment. Correlation networks of the citrate cycle (TCA cycle); fatty acid metabolism; glyoxylate and dicarboxylate metabolism; propanoate metabolism; valine, leucine, and isoleucine degradation and beta-alanine metabolism showed the inner relationships between metabolites and genes that we previously explored in the main enriched metabolic pathways showing changes

Full size image

Discussion

Bacterial myocarditis is a common clinical sepsis syndrome caused mainly by inflammation, which can contribute to adverse cardiovascular outcomes, end-organ dysfunction, and even mortality (Li and Sun 2007; Schauvliege et al. 2007).

LPS-trigged excessive host responses have been established to result in systemic inflammatory conditions, including myocarditis and sepsis, mainly by activating the Toll-like receptor (TLR) family, the major pattern-recognition receptors (PRRs) family of host cells (O’Neill et al. 2009; Akira et al. 2006), which fuels our interest in exploring the signaling function of TLRs after LPS stimulation. Previous studies have shown that TLRs-activated signal transduction pathways which determine the vital role of TLRs in inflammatory disorders depend on the binding between their intracellular domain (called TIR domain) and intracellular adaptors, including myeloid differentiation factor 88 (MyD88), Mal (TIRAP), TRIF (TICAM-1), and TRAM (TICAM-2) (Kawai and Akira 2007; Dunne and O’Neill 2003). Upon LPS stimulation, the phosphorylation cascade of MyD88 initiated by TLRs mediates the recruitment and activation of IL-1R-associated kinases (IRAKs), further triggering the activation of the nuclear factor-κB (NF-κB) family, resulting in the production of pro-inflammatory mediators helping to shape the immune response and eradicate infection (Janssens and Beyaert 2003).

As an intrinsic ubiquitin E3 ligase responsible for receptor-interacting protein 1 (RIP1) ubiquitination in the TLR/IL-1R signaling, Pellino is an evolutionarily conserved scaffold protein that could catalyze the polyubiquitylation of key molecule involved in the tumor necrosis factor receptor-associated factor (TRAF)-dependent TLR signaling pathway by interacting with its highly homologous family, IRAKs, in an MyD88-dependent manner after stimulation in mammals (Grosshans et al. 1999a; Schauvliege et al. 2007).

Pellino1, a mammalian counterpart of Pellino, have been recognized to be highly required for maintaining TLR-mediated NF-κB nuclear translocation and binding activity due to its value in forming Pellino1–IRAK1–IRAK4–TRAF6 signaling complex after the induction of inflammatory response (Moynagh 2009; Jiang et al. 2003), indicating the causal relationship between Pellino1 and TLR-mediated inflammatory response. In addition to TRAF6, a recent study found that the activation of TIR-domain-containing adapter-inducing interferon-β (TRIF)-dependent signaling induced-by Pellino1-mediated ubiquitination of RIP1 also contributed to the upregulation of NF-κB (Sato et al. 2003; Chang et al. 2009). Moreover, Pellino1 deficiency may interfere with the ubiquitination of RIP1 and TRAF6 which are needed for TLR3/4-mediated pro-inflammatory gene induction, thereby increasing the resistance of mice to LPS-induced lethality (Chang et al. 2009; Skaug et al. 2009). Collectively, the importance of Pellino1 in LPS-induced inflammatory signaling makes it reasonable to assume that suppressing Pellino1 do be especially effective in attenuating LPS-induced myocarditis.

As the unifying link between heart and cardiac function, changed cardiac metabolism has been evidenced to be causative in cardiomyocytes dysfunction induced by various physiological and pathological causes, including exercise and pressure overload (Gibb and Hill 2018). During infections, cell-intrinsic metabolism is no longer considered as a physical process that meets the basic demand of energy for heart contraction (Newsholme et al. 1986). In addition to supporting tissue homeostasis, pathogen-activated reprogramming of metabolic pathways also contributes to the production of metabolic intermediates and end products (Williams and O’Neill 2018). The immune-metabolic framework consisted of kinds of regulatory metabolites that play a vital role in regulating cardiac function, and determining the fate of an infection (Chauhan and Saha 2018).

Recent advances in LC–MS/MS-based omics uncovered that changed metabolites under pathological states could lead to the progression of inflammation (Ishihara et al. 2019). It has been shown that myeloperoxidase (MPO) could mediate primarily host defense reactions by regulating peroxidation of polyunsaturated fatty acids which is the characteristic feature of inflammation during acute inflammation (Winterbourn et al. 2000; Kubala et al. 2010). The withdrawal of the synthesis of bioactive metabolites of arachidonic acid (AA) and linoleic acid (LA) metabolism may ameliorate the severity of sepsis by modifying inflammatory-related cellular signaling pathways (Cook 2005; Nauseef 2001). Besides, the l-Tryptophan (Trp)–kynurenine (Kyn) pathway, which is essential for protein synthesis, was reported to improve acute viral myocarditis (Hoshi et al. 2012).

Metabolic remodeling after LPS stimulation may modify the downstream genes and proteins by triggering the mammalian target of rapamycin (mTOR), AMP-activated protein kinase (AMPK), and nuclear factor-κB (NF-κB) signaling pathways involved in many pathological conditions, such as atherosclerosis, indicating that it is necessary to emphasize the importance of metabolites in LPS-induced pro-inflammatory response (Byles et al. 2013; Sag et al. 2008; Wu et al. 2020). As the main source of energy for cardiac mechanical work, fatty acid (FA) can be transported by fatty acid translocase (FAT/CD36) and other transport proteins from the cytosol to the mitochondria and then form acyl-CoA to enter β-oxidation (van der Vusse et al. 2000; Berndt et al. 2019; Girones et al. 2014). The peroxisome proliferator-activated receptor subfamily is vital in the epigenetic regulation of FA utilization. It has been reported that the main three members of the PPAR nuclear receptor subfamily, PPARα, PPARδ, and PPARγ, are all involved in synergistic modulation of cardiomyocytes energy metabolism (Lopaschuk et al. 2010). Furthermore, in in vitro models of mice, the activation of PPARα has been identified to increase the expression of downstream genes, including CD36, MCPT1, and MCAD, which play different important roles in cardiac FA utilization (Lopaschuk et al. 2010; Schoonjans et al. 1996; Mistry and Cresci 2010; Huss and Kelly 2004). The induced transcriptional activities of both PPARs and ERRs could be coactivated by binding to PGC-1α (Huss et al. 2004; Schreiber et al. 2004). Recently, the PGC-1α/PPARα and PGC-1α/ERRα pathways are involved in regulating cardiac mitochondrial fatty acid oxidation genes (Lehman et al. 2000).

PGC-1α, an important regulator of mitochondrial activity and fatty acid oxidative metabolism, is enriched in the heart, which has high oxidative metabolic rates (Lehman et al. 2000; Lin et al. 2002; Knutti and Kralli 2001). Compared to those of PGC-1β, which is also confirmed to be necessary for aerobic mitochondrial respiration, post-translational modifications of PGC-1α bound with transcriptional partners could better control complex cardiac cellular energy metabolism by activating downstream endogenous gene promoters, including MCAD (St-Pierre et al. 2003; Rowe et al. 2010; Huss et al. 2002). Interestingly, in our study, silencing Pellino1 reversed the LPS-induced decrease in the expression of transcription factors and regulators in NRCMs, showing that Pellino1 plays a role in regulating fatty acid metabolism.

After analyzing the metabolome data, we found 169 unique metabolites that differed significantly in si-Pellino1 and LPS-treated, LPS-treated, and untreated control NRCMs. Then, the TCA cycle, fatty acid metabolism, and glycolysis or gluconeogenesis, along with fatty acid elongation in mitochondria and alanine, aspartate and glutamate metabolism, as well as glyoxylate and dicarboxylate metabolism, were identified as the most perturbed metabolic pathways according to P values from enrichment analysis and impact values from topology analysis.

Moreover, lipopolysaccharide has been reported to regulate the myocardial expression of the FAO enzyme and oxidation rates by deactivating the kinetic response of the PGC-1 axis, which is associated with systemic inflammation and sepsis-induced by LPS stimulation (Beutler and Rietschel 2003; Rossi et al. 2007; Puigserver et al. 2001; Feingold et al. 2004). Our results found that LPS altered cardiac fuel and energy metabolism accompanied by changes in key genes (Cs, Cpt2, and Acadm) and metabolites (3-oxoocotanol-CoA, hydroxypyruvic acid, lauroyl-CoA, and NADPH) in NRCMs. Furthermore, for the first time, we confirmed that Pellino1, which has been identified to activate NF-κB and TLR signaling pathways by mediating proinflammatory genes in innate immunity (Chang et al. 2009; Kawai and Akira 2007; Choi et al. 2006; Vallabhapurapu and Karin 2009), could partly reverse or restore the disturbed cardiomyocytes energy metabolism through mechanisms involving glycolysis, lipid metabolism, citrate cycle, and some other types of metabolism after LPS stimulation.

Glucose utilization is also highly linked to various physiological processes, including cardiomyocytes energy metabolism, which can affect mitochondrial function by coregulation with FA utilization (Randle 1998). GLUT1 and GLUT4, the common and major glucose transporters of glycogen-derived glucose in the heart, play a basic and important role in basal myocardial glucose uptake, suggesting that the decreased expression of GLUT4 may be harmful to cardiac glucose metabolism (Henning et al. 1996; Aerni-Flessner et al. 2012). It has been reported that pathological cardiac hypertrophy could accelerate glycolysis (Allard et al. 1994; Cheng et al. 2017).

In our study, si-Pellino1 alleviated the LPS-induced decrease in GLUT4 expression in NRCMs. The observed changes in two pivotal intermediates, succinic acid, and malic acid, were partly reversed after si-Pellino1 treatment, suggesting that silencing Pellino1 in LPS-induced myocarditis improved LPS-induced glucose metabolic dysfunction.

The metabolic profiles analyzed in our study could partly explain the therapeutic value of silencing Pellino1 in LPS-induced myocarditis. However, this report has some limitations that need to be considered. In our study, the samples were extracted from NRCMs cultivated in vitro. As a result, given the different metabolic environments, we lack experimental data from in vivo studies. Additionally, targeted metabolomics is an optimal method to identify the differences in metabolites. Consequently, these issues remain to be examined in the future.

Thus, the LC–MS/MS metabonomic method shed new light on the mechanism of the pathological development of bacterial myocarditis and was also used to assess the efficacy of si-Pellino1 treatment. Importantly, the altered metabolites associated with the TCA cycle, fatty acid metabolism, and glycolytic metabolism may be used as potential specific biomarkers corresponding to LPS-induced myocarditis. Moreover, combined with restoration of the abnormal metabolism and disturbed expression of transcription factors involved in cardiac fuel and energy metabolism, silencing Pellino1 showed cardioprotective value, revealing that silencing Pellino1 could be a promising therapeutic strategy against LPS-induced myocarditis or other pathologies in future clinical trials. In addition, considering the pro-inflammatory role of Pellino1 reported in the sepsis-induced lung injury model and COPD-induced inflammatory airway responses, silencing Pellino1 can also ameliorate acute or chronic lung pneumonia (Liu et al. 2021; Marsh et al. 2020; Hughes et al. 2019). Other than the heart and lung, the regulatory effects of Pellino1 on TLR/IL-1R signaling make it reasonable to assume that silencing Pellino1 may be targeted therapeutically in different inflammatory disorders associated with TLR/IL-1R signaling, which remains to be further explored in the future.