Optimization of hepatic metabolic pathway computational model parameters based on multi-omics data: quantitative characterization from health to disease states
Background As the central organ of metabolic regulation, the liver undergoes significant metabolic network remodeling and functional disruption during disease states, yet the underlying mechanisms remain insufficiently elucidated. The advancement of multi-omics data integration and computational modeling offers new research avenues and theoretical support for precise reconstruction of metabolic networks and in-depth exploration of pathogenic mechanisms. Objective To construct a computational model of liver metabolic pathways and optimize its parameters through multi-omics data integration, enabling quantitative characterization of the transition from healthy to diseased states, thereby elucidating the remodeling mechanisms of liver disease metabolic networks and identifying critical regulatory nodes and potential therapeutic targets. Methods A computational model encompassing major pathways including glycolysis, TCA cycle, fatty acid metabolism, and amino acid metabolism was established firstly; then transcriptomic, proteomic, and metabolomic data from healthy controls and patients with three types of liver diseases were optimized; finally the model performance was evaluated through internal cross-validation and external validation with an independent cohort, and key regulatory nodes was identified through parameter sensitivity analysis and metabolic control analysis. Results The optimized model accurately predicted metabolic features of the three liver diseases, with correlation coefficients between predicted and experimental metabolite concentrations exceeding 0.85. The study revealed common metabolic reprogramming features in liver diseases: shift from aerobic oxidation to glycolysis, from fatty acid oxidation to synthesis, and amino acid metabolism disorders. Fifteen key regulatory nodes were identified, including three common nodes (hexokinase, pyruvate dehydrogenase, fatty acid synthase) and twelve disease-specific nodes. A liver disease classification and staging algorithm developed based on these findings achieved a classification accuracy of 92.5% in the external validation cohort, significantly outperforming traditional methods (76.3%). Conclusion This study has achieved parameter optimization of a computational model for liver metabolic pathways based on multi-omics data, quantitatively characterized metabolic features in healthy and disease states, revealed mechanisms of metabolic network reconstruction, and identified key regulatory nodes, providing a computational foundation for precision diagnosis and personalized treatment of liver diseases, while offering new insights for systems biology-based drug development.
基于优化后的参数模型进一步比较健康和疾病状态下的代谢通量。结果显示,3种肝病存在共同的代谢重编程特征。能量代谢重编程显示,从有氧氧化向糖酵解转变,表现为糖酵解通量增加(健康vs NAFLD/ALD/VH:1 vs 2.35/1.98/2.12)和三羧酸循环通量减少(1 vs 0.65/0.53/0.72)。脂质代谢重编程显示,从脂肪酸氧化向脂肪酸合成转变,表现为脂肪酸β氧化通量减少(1 vs 0.48/0.41/0.63)和脂肪酸合成通量增加(1 vs 2.85/1.92/1.73)。氨基酸代谢紊乱显示,转氨酶活性增加(谷草转氨酶:1 vs 1.63/1.88/2.12;谷丙转氨酶:1 vs 1.78/2.05/2.32)和尿素循环能力下降(1 vs 0.62/0.52/0.68)。
本研究发现,尽管NAFLD、ALD和VH的病因各异,但其在代谢层面存在共同的重编程特征,即从有氧氧化向糖酵解转变,从脂肪酸氧化向脂肪酸合成转变。该代谢重编程模式与肿瘤细胞中所观察到的Warburg效应高度相似[22],表明不同病理转化过程可能共享某些代谢适应机制。在转录水平,数据显示3种肝病中共同上调的转录因子包括HIF-1α(NAFLD/ALD/VH: 2.35/1.92/2.08)和SREBP-1c(NAFLD/ALD/VH:2.87/1.78/1.63),前者主要调控糖酵解相关基因,后者主要调控脂肪酸合成相关基因。同时,3种肝病中共同下调的转录因子包括PPARα(NAFLD/ALD/VH:0.38/0.33/0.45)和PGC-1α(NAFLD/ALD/VH:0.42/0.37/0.53),这两个转录因子是脂肪酸氧化和线粒体生物合成的关键调控位点。在蛋白质水平,己糖激酶2和丙酮酸激酶M酪氨酸磷酸化水平在3种肝病中均显著增高,而脂肪酸氧化关键酶ACADVL和HADHA的赖氨酸乙酰化水平显著增加。这些翻译后修饰直接影响了相关酶的活性,从而导致代谢通量的重分配。在代谢物水平,乳酸/丙酮酸比值(NAFLD/ALD/VH:12.3/10.8/11.5 vs 健康对照:5.6)和NADH/NAD+比值(NAFLD/ALD/VH:0.23/0.26/0.21 vs 健康对照:0.11)在3种肝病中均显著升高,反映细胞氧化还原状态的改变,可能是代谢重编程的重要驱动因素。
本研究优化后的模型较为准确预测NAFLD、ALD和VH 3种肝病代谢特征,并捕捉到3种肝疾病不同阶段代谢变化。进一步分析显示,NAFLD以脂质代谢紊乱为主要特征,表现为脂肪酸合成/脂肪酸氧化比值显著升高(NAFLD:5.94 vs ALD:4.68,VH:2.75)。ALD以乙醇代谢异常和氧化应激为特征,表现为NADH/NAD+比值最高(ALD:0.26 vs NAFLD:0.23,VH:0.21)。VH则以氨基酸代谢和核苷酸代谢异常为特征,表现为转氨酶活性最高和尿素循环中间产物积累。这些特征性代谢模式为肝病的鉴别诊断提供了新的分子标志。研究同时发现代谢通路受影响的程度和时序存在差异,在NAFLD进展过程中,糖酵解通量的增加(轻度/中度/重度:1.53/2.08/2.35倍)早于脂肪酸合成通量的增加(轻度/中度/重度:1.87/2.35/2.85倍),而氨基酸代谢紊乱则相对较晚(轻度/中度/重度:1.25/1.58/1.78倍),这种代谢通路受影响的时序差异,为早期诊断和干预提供了线索。
YounossiZM, KoenigAB, AbdelatifD,et al . Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence,incidence,and outcomes[J]. Hepatology,2016,64(1):73-84.
[2]
HasinY, SeldinM, LusisA . Multi-omics approaches to disease[J]. Genome Biol,2017,18(1):83.
[3]
JoyceAR, PalssonBØ . The model organism as a system: integrating ‘omics’ data sets[J]. Nat Rev Mol Cell Biol,2006,7(3):198-210.
[4]
KuepferL, NiederaltC, WendlT,et al . Applied concepts in PBPK modeling: how to build a PBPK/PD model[J]. CPT Pharmacometrics Syst Pharmacol,2016,5(10):516-531.
[5]
Bravo González-BlasC, MatetoviciI, HillenH,et al . Single-cell spatial multi-omics and deep learning dissect enhancer-driven gene regulatory networks in liver zonation[J]. Nat Cell Biol,2024,26(1):153-167.
[6]
Cvitanović TomašT, MoškonM, MrazM,et al . Computational modelling of liver metabolism and its applications in research and the clinics[J]. Acta Chim Slov,2018,65(2):253-265.
[7]
RauckhorstAJ, SheldonRD, PapeDJ,et al . A hierarchical hepatic de novo lipogenesis substrate supply network utilizing pyruvate,acetate,and ketones[J]. Cell Metab,2025,37(1):255-273. e6.
[8]
SinghA, ShannonCP, GautierB,et al . DIABLO: an integrative approach for identifying key molecular drivers from multi-omics assays[J]. Bioinformatics,2019,35(17):3055-3062.
[9]
SalvyP, HatzimanikatisV . The ETFL formulation allows multi-omics integration in thermodynamics-compliant metabolism and expression models[J]. Nat Commun,2020,11(1):30.
[10]
PandeyV, HadadiN, HatzimanikatisV . Enhanced flux prediction by integrating relative expression and relative metabolite abundance into thermodynamically consistent metabolic models[J]. PLoS Comput Biol,2019,15(5):e1007036.
[11]
SaaPA, NielsenLK . Construction of feasible and accurate kinetic models of metabolism: A Bayesian approach[J]. Sci Rep,2016,6:29635.
[12]
DaiD, DaiFD, ChenJC,et al . Integrated multi-omics reveal important roles of gut contents in intestinal ischemia-reperfusion induced injuries in rats[J]. Commun Biol,2022,5(1):938.
[13]
MorabitoA, De SimoneG, PastorelliR,et al . Algorithms and tools for data-driven omics integration to achieve multilayer biological insights: a narrative review[J]. J Transl Med,2025,23(1):425.
[14]
ChenSR, HoC, FengDG,et al . Tracer kinetic modeling of 11C-acetate applied in the liver with positron emission tomography[J]. IEEE Trans Med Imaging,2004,23(4):426-432.
[15]
SmallboneK, MessihaHL, CarrollKM,et al . A model of yeast glycolysis based on a consistent kinetic characterisation of all its enzymes[J]. FEBS Lett,2013,587(17):2832-2841.
[16]
PavlovaNN, ThompsonCB . The emerging hallmarks of cancer metabolism[J]. Cell Metab,2016,23(1):27-47.
[17]
AlonsoC, Fernández-RamosD, Varela-ReyM,et al . Metabolomic identification of subtypes of nonalcoholic steatohepatitis[J]. Gastroenterology,2017,152(6):1449-1461.
[18]
Balsa-CantoE, BangaJR, EgeaJA,et al . Global optimization in systems biology: stochastic methods and their applications[J]. Adv Exp Med Biol,2012,736:409-424.
[19]
BerndtN, BulikS, WallachI,et al . HEPATOKIN1 is a biochemistry-based model of liver metabolism for applications in medicine and pharmacology[J]. Nat Commun,2018,9(1):2386.
[20]
MardinogluA, AgrenR, KampfC,et al . Genome-scale metabolic modelling of hepatocytes reveals serine deficiency in patients with non-alcoholic fatty liver disease[J]. Nat Commun,2014,5:3083.
[21]
DonnellyKL, SmithCI, SchwarzenbergSJ,et al . Sources of fatty acids stored in liver and secreted via lipoproteins in patients with nonalcoholic fatty liver disease[J]. J Clin Invest,2005,115(5):1343-1351.
[22]
Vander HeidenMG, CantleyLC, ThompsonCB . Understanding the Warburg effect: the metabolic requirements of cell proliferation[J]. Science,2009,324(5930):1029-1033.