(515bi) A New Metabolic Model For Analysis Of Dynamic 13C Isotopomer Time Courses In The Brain

INTRODUCTION Metabolic modeling of 13C turnover curves obtained during infusion of a 13C labeled substrate (e.g. [1,6-13C2]glucose) with a two-compartment neuronal-glial model allows measurement of compartmentalized metabolic fluxes such as the neuronal and glial TCA cycle rates and the rate of glutamate-glutamine cycle. Current metabolic models typically fit only the total 13C enrichment at each carbon position (?positional model?). Recently, we reported in vivo measurements of time courses for multiple 13C-13C isotopomers, which appear as multiplets in 13C spectra [1]. The goals of the present work were (i) to develop a new neuronal-glial metabolic model (?isotopomer model?) capable of taking into account the additional information from 13C-13C multiplets and (ii) to determine whether this new model leads to improved precision in fitted metabolic fluxes. METHODS The metabolic network used was identical to that of previous models [2,3]. Isotope balance equations were derived for every possible isotopomer of glutamate, glutamine and aspartate, including multiply labeled isotopomers. This resulted in a set of ~160 differential equations. Solving the set of differential equations (using the Runge-Kutta algorithm) yielded time courses for all possible isotopomers in glutamate, glutamine and aspartate. Monte-Carlo simulations were performed to evaluate the precision of fitted metabolic parameters with the new model. Fitting of synthetic turnover curves was repeated at least 500 times with a different noise. Minimization was performed using BFGS or Simplex algorithms. RESULTS Fig 1 shows examples of experimental time courses for glutamate obtained during C1,6-glucose infusion. Previous metabolic modeling approaches have used only the total 13C positional enrichment (for example GluC4total and GluC3 total) but did not take advantage of the additional information present in 13C-13C multiplets. The new metabolic model presented here allows fitting of each dynamic isotopomer curve (for example GluC4S, Glu-C4D43, Glu-C3S, Glu-C3D, etc). Monte-Carlo simulations show that the new model leads to a significant improvement in the precision of metabolic fluxes. The SD on VNT decreased from 670% with the positional model to just 19% with the isotopomer model for these particular simulated conditions (tmax =150 min, 20 points per turnover curve, noise sigma = 0.2 micromol.g-1). DISCUSSION The new metabolic model allowed us to take full advantage of the additional dynamic information available from 13C multiplets in 13C spectra. The isotopomer model can fit up to 20 isotopomer turnover curves compared to 7 positional enrichment curves with the most advanced positional models. Although the computation time was increased, fitting of a single data set still required no more than one minute with a fast personal computer. Using the additional information from 13C multiplets leads to an increase in precision for all six metabolic fluxes in the model. This is consistent with previous studies in the heart which showed that fitting the multiplet turnover curve led to better precision on the determination of VTCA using a one compartment model [4]. CONCLUSION In conclusion, we developed a new metabolic model to take into account the additional information from 13C isotopomers, which appear as multiplets in 13C spectra. The additional information leads to significantly increased precision in fitted metabolic fluxes. REFERENCES [1] Henry NMR Biomed 16, 400, 2003; [2] Gruetter AJP 281, E100, 2001; [3] Shen PNAS 96, 8235, 1999; [4] Jeffrey AJP 277, E1111, 1999.

This work was supported by NIH grants R01NS38672 and P41RR08079.