Centre for Integrative Bioinformatics, VU University Amsterdam, De Boelelaan 1081A, 1081 HV Amsterdam, The Netherlands.
Benef Microbes. 2010 Nov;1(4):391-405. doi: 10.3920/BM2010.0038.
This paper explores human gut bacterial metabolism of starch using a combined analytical and computational modelling approach for metabolite and flux analysis. Non-steady-state isotopic labelling experiments were performed with human faecal microbiota in a well-established in vitro model of the human colon. After culture stabilisation, [U-13C] starch was added and samples were taken at regular intervals. Metabolite concentrations and 13C isotopomeric distributions were measured amongst other things for acetate, propionate and butyrate by mass spectrometry and NMR. The vast majority of metabolic flux analysis methods based on isotopomer analysis published to date are not applicable to metabolic non-steady-state experiments. We therefore developed a new ordinary differential equation-based representation of a metabolic model of human faecal microbiota to determine eleven metabolic parameters that characterised the metabolic flux distribution in the isotope labelling experiment. The feasibility of the model parameter quantification was demonstrated on noisy in silico data using a downhill simplex optimisation, matching simulated labelling patterns of isotopically labelled metabolites with measured metabolite and isotope labelling data. Using the experimental data, we determined an increasing net label influx from starch during the experiment from 94±1 µmol/l/min to 133±3 µmol/l/min. Only about 12% of the total carbon flux from starch reached propionate. Propionate production mainly proceeded via succinate with a small contribution via acrylate. The remaining flux from starch yielded acetate (35%) and butyrate (53%). Interpretation of 13C NMR multiplet signals further revealed that butyrate, valerate and caproate were mainly synthesised via cross-feeding, using acetate as a co-substrate. This study demonstrates for the first time that the experimental design and the analysis of the results by computational modelling allows the determination of time-resolved effects of nutrition on the flux distribution within human faecal microbiota in metabolic non-steady-state.
本文采用分析和计算建模相结合的方法，研究了人类肠道细菌对淀粉的代谢，进行代谢物和通量分析。在一个成熟的体外人结肠模型中，用人粪便微生物进行了非稳态同位素标记实验。在培养稳定后，添加[U-13C]淀粉，并定期取样。通过质谱和 NMR 等方法测量了包括乙酸盐、丙酸盐和丁酸盐在内的代谢物浓度和 13C 同位素分布。迄今为止，基于同位素分析的绝大多数代谢通量分析方法都不适用于代谢非稳态实验。因此，我们开发了一种新的基于常微分方程的人类粪便微生物代谢模型表示方法，以确定 11 个代谢参数，这些参数描述了同位素标记实验中的代谢通量分布。通过使用 downhill simplex 优化在噪声的计算机模拟数据上证明了模型参数定量的可行性，该方法将模拟标记的同位素标记代谢物的标记模式与测量的代谢物和同位素标记数据进行匹配。使用实验数据，我们确定了实验过程中淀粉的净标记物流入量从 94±1 µmol/l/min 增加到 133±3 µmol/l/min。只有约 12%的淀粉总碳通量到达丙酸盐。丙酸盐的生成主要通过琥珀酸盐进行，丙烯酸盐的贡献较小。淀粉的剩余通量生成了乙酸盐（35%）和丁酸盐（53%）。13C NMR 多重信号的进一步解释表明，丁酸、戊酸和己酸主要通过交叉喂养，使用乙酸盐作为共底物合成。本研究首次证明，通过计算建模进行实验设计和结果分析，可以在代谢非稳态条件下，确定营养对人类粪便微生物内通量分布的时间分辨效应。
