Multi-omic signatures of atherogenic dyslipidaemia: pre-clinical target identification and validation in humans

Abstract

Background: Dyslipidaemia is a major risk factor for atherosclerosis and cardiovascular diseases. The molecular mechanisms that translate dyslipidaemia into atherogenesis and reliable markers of its progression are yet to be fully elucidated. To address this issue, we conducted a comprehensive metabolomic and proteomic analysis in an experimental model of dyslipidaemia and in patients with familial hypercholesterolemia (FH).

Methods: Liquid chromatography/mass spectrometry (LC/MS) and immunoassays were used to find out blood alterations at metabolite and protein levels in dyslipidaemic ApoE^-/-/LDLR^-/- mice and in FH patients to evaluate their human relevance.

Results: We identified 15 metabolites (inhibitors and substrates of nitric oxide synthase (NOS), low-molecular-weight antioxidants (glutamine, taurine), homocysteine, methionine, 1-methylnicotinamide, alanine and hydroxyproline) and 9 proteins (C-reactive protein, proprotein convertase subtilisin/kexin type 9, apolipoprotein C-III, soluble intercellular adhesion molecule-1, angiotensinogen, paraoxonase-1, fetuin-B, vitamin K-dependent protein S and biglycan) that differentiated FH patients from healthy controls. Most of these changes were consistently found in dyslipidaemic mice and were further amplified if mice were fed an atherogenic (Western or low-carbohydrate, high-protein) diet.

Conclusions: The alterations highlighted the involvement of an immune-inflammatory response system, oxidative stress, hyper-coagulation and impairment in the vascular function/regenerative capacity in response to dyslipidaemia that may also be directly engaged in development of atherosclerosis. Our study further identified potential biomarkers for an increased risk of atherosclerosis that may aid in clinical diagnosis or in the personalized treatment.

Keywords: Atherosclerosis; Dyslipidaemia; Metabolome; Pathological mechanisms; Proteome; Serological biomarkers.

Fig. 1

A flow diagram of multi-omic analysis was used to present the overview of the study design, how the experiments were performed, as well as the way of data treatment

Fig. 2

The protein–protein interactions for the differentially expressed proteins in ApoE^−/−/LDLR^−/− mice (vs. WT) that were analysed using STRING 11 software. In the network analysis the differentially expressed proteins were presented as nodes, whereas edges represent predicted protein–protein associations. Using the protein interaction network analysis tool (STRING database), seven networks of the associated proteins were found among the differentially expressed proteins that were depicted by relevant colours. These included proteins mainly involved in: (1) acute phase response and/or being a constituent of lipoprotein particles (olive), (2) coagulation cascade (red), (3) alternative complement pathway (mint), (4) existing as inhibitors of serine proteases (light blue), (5) participating in adhesion of leukocytes to endothelial cells/renin-angiotensin system over-activation or modulation of oxidative stress (turquoise), (6) modulation of extracellular interactions/extracellular matrix re-organization (salmon), and (7) others – the disconnected node in the network represented by creatine kinase (hidden in the plot)

Fig. 3

Significantly altered metabolites in response to atherogenic dyslipidaemia in ApoE^−/−/LDLR^−/− mice. Plasma concentration of substrates for NOSs (L-Arg, H-Arg) and Arg precursors (L-Cit, L-Orn) (a), methylated Arg derivatives (NMMA, SDMA, ADMA) as well as L-Arg/ADMA ratio (b), L-Met, Hcy (c) and other metabolites significantly changed in response to dyslipidaemia (d) in ApoE^−/−/LDLR^−/− (n = 7) and WT (n = 9) mice. Data represent mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 vs. WT

Fig. 4

Summary of altered metabolic pathways analysed by MetaboAnalyst 4.0 web software. As shown, nine metabolic pathways of importance were disturbed in the plasma of dyslipidaemic mice (as compared to the wild-types): (1) valine, leucine and isoleucine degradation; (2) phenylalanine, tyrosine and tryptophan biosynthesis; (3) taurine and hypotaurine metabolism; (4) arginine and proline metabolism; (5) alanine, aspartate and glutamate metabolism; (6) tryptophan metabolism; (7) nicotinate and nicotinamide metabolism; (8) tyrosine metabolism, and (9) cysteine and methionine metabolism. Significantly affected pathways are plotted according to P-value from pathway enrichment analysis and pathway impact score from pathway topology analysis. Circle size and colour gradient indicate the significance of the pathway ranked by P-value (red: lower P-values and yellow: higher P-values) and pathway impact score (the larger the circle the higher the impact score), respectively

Fig. 5

Effect of pro-atherogenic diets (Western and low-carbohydrate, high-protein (LCHP) diets) on plasma AA (amino acid-related) metabolome in ApoE^−/−/LDLR^−/− mice. Plasma concentration of substrates for NOSs (L-Arg, H-Arg) and Arg precursors (L-Cit, L-Orn) (a), methylated Arg derivatives (NMMA, SDMA, ADMA) as well as L-Arg/ADMA ratio (b), L-Met, Hcy (c) and other metabolites related to atherosclerosis progression (d) in 6-month-old ApoE^−/−/LDLR^−/− mice fed for 2 months: Control-AIN-93G (n = 6), WD (n = 5) or LCHP (n = 5) diet, respectively. Data represent mean ± SEM. *, **, and *** indicate P < 0.05, P < 0.01, P < 0.001, respectively

Fig. 6

Translational study in human sera coming from patients affected by FH (n = 20) and healthy subjects (n = 20) (discovery cohort). Quantification of selected marker candidates (CRP, PCSK9, ApoC-III, sICAM-1, AGT, PON-1, FETUB, VKDP-S, and BGN) in crude serum samples by colorimetric ELISA. Protein levels are presented as mean ± SEM. *, **, and *** indicate P < 0.05, P < 0.01, P < 0.001, respectively. FH + S – FH patients receiving statin therapy

Fig. 7

A snapshot of serum metabolites in the setting of familial hypercholesterolemia (discovery cohort). Human metabolomic signatures related to atherogenic dyslipidaemia closely reflect the pattern already identified in ApoE^−/−/LDLR^−/− mice. The top (a) row depicts serum concentration of substrates for NOSs (L-Arg, H-Arg) and Arg precursors (L-Cit, L-Orn), the middle rows -methylated Arg derivatives (NMMA, SDMA, ADMA) and L-Arg/ADMA ratio (b), and L-Met, Hcy (c), respectively and d refers to other altered metabolites determined in FH cases (n = 20) vs. healthy subjects (n = 20). Data represent mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. FH + S – FH patients on statins

Fig. 8

Receiver operating characteristic (ROC) curves (a–d), showing the ability of the potential biomarkers to distinguish between control subjects and FH patients, as well as heatmaps data visualization to depict variance across multiple variables (e, f). a, b Multivariate ROC curves of the six models in primary and validation cohorts have been generated employing random forest prediction model with combined features. We found that the model with 10 features entered showed excellent predictive performance with ROC AUC (area under the curve) values > 99% for both discovery and validation data. c, d 15 significant features were ranked based on their average importance (in group classification) during cross validation. e, f Color-coded maps that illustrate the relative abundance of feature combinations for two independent cohorts of cases