Rolf Gruetter et al. 1

NMR STUDIES OF IN VIVO GLUCOSE CONCENTRATIONS AND TRANSPORT

Rolf Gruetter, Josef Pfeuffer, Ivan Tkac, Greg Damberg, Elizabeth R.Seaquist

Dept. of Radiology and Medicine, University of Minnesota

Running Title: MRS of brain glucose

Introduction Human brain function depends on a continuous glucose supply from the blood. The specific glucose transport mechanism from the blood is mediated by transporter molecules and has been shown to be of the facilitated diffusion type with saturation kinetics. Quantitative measurements of glucose transport have been reported in a handful of studies of human brain using radioactive tracers (1-4). The free intracellular glucose concentration depends on metabolic as well as on transport rates and thus contains information whether glucose supply is adequate for demand. It is well established that the brain glucose concentration is an important parameter in the analysis of experiments using radiolabeled glucose or glucose analogs. Free glucose is measurable in animal brain e.g. (5-8). Under physiological conditions, free glucose is measurable by NMR in the human brain (9-12). The methods used to assess cerebral glucose concentrations in vivo will be reviewed in this paper, as well as the resulting assessments of brain glucose content that have been made using NMR. Glucose signals have been observed using either 13C or 1H NMR (MRS). The direct observation by 13C has the advantage that the signals from a and b glucose can be unequivocally observed. The disadvantage lies in the lower sensitivity, thereby requiring longer accumulation times, larger regions of interest or infusion of 13C labeled glucose to increase sensitivity. To minimize potential partial volume effects from extracerebral tissue and from large vessels, localized 13C NMR measurements have been performed (9). These studies also showed that brain glucose concentration is approximately 1 mmol/g at euglycemia in the human brain. Recently, it was demonstrated at higher magnetic field strengths that at hyperglycemia 13C NMR can also be used to quantify brain glucose without infusion of 13C labeled glucose (10). Fig. 1 shows the signals from the C1 of a and b glucose as well as the natural abundance glucose peak of glucose (arrows). Observation of glucose in the 1H spectrum is much more difficult due to overlap with resonances from other compounds (13, 14). To overcome such limitations, investigators have turned to measuring the change in brain glucose signal at 3.4 ppm using difference editing (13), which has been used to dynamically investigate glucose transport kinetics (14, 15). Alternatively, 1H NMR detection of glucose has also been achieved by exploiting homonuclear J couplings in glucose (16), and by observing the 1H resonances of H1 bound to the C1 of 1-13C-glucose (17, 18). Recently it was shown that the H1 resonance of a-D-glucose is sufficiently separated at high magnetic fields to be observed directly, without further editing means (10, 11) (Fig. 2). The unique ability of in vivo MRS to quantitate brain glucose naturally provides a measurement capability that is complementary to the highly sensitive brain mapping techniques. It is the purpose of this paper to review some of the more recent findings on brain glucose transport by in vivo MRS. Rolf Gruetter et al. 2

Physiology of brain glucose transport Glucose transport across membranes is mediated by a family of transporter proteins, e.g. (19). The majority of evidence has suggested that GLUT-1, the ubiquitous high-affinity transporter present in erythrocyte membranes, is also the dominant transporter protein at the blood-brain barrier, with some contributions from other transporters (20, 21). Inside the brain, a different transporter may be responsible for glucose transport across the cell membranes (22). However, it has been argued that the large surface area of brain cells may result in rapid equilibration inside the brain’s aqueous phase leading to similar intra- and extracellular glucose concentrations. The cerebral blood volume in human gray matter is 0.05 ml/g or less, suggesting that blood glucose is a minor contribution to the overall NMR signal. With these assumptions, the glucose signal detected by NMR reflects glucose in the intracellular aqueous compartment. Brain glucose concentrations have been quantified using different NMR methods. An early study quantified whole brain glucose in rats using 13C NMR spectroscopy (5) by using the postmortem cerebral lactate signal as the internal reference. Non-invasive quantification of the signals of a and b glucose C1 measured by 13C MRS was reported in human brain using the external reference method (9). External referencing is based on repeating the identical measurement in a phantom containing an aqueous solution of a known concentration of natural abundance glucose. This technique is very robust, since the calibration measurement is designed to mimic the in vivo experiment in as many aspects as experimentally can be controlled. A disadvantage is the need to perform a calibration measurement after the in vivo study, which increases the complexity of the study. Quantification of the 1H signal of brain glucose has been performed using creatine as an internal concentration standard (10). Using the creatine signal as an internal standard has the advantage of minimizing the overall study time and of being very robust, provided care is taken to minimize variations in the gray/white matter composition of the volume studied (23), see also references in (24). The glucose peak of the H1 of a-D-glucose has been quantified using this approach and the quantification has been compared to that using the external reference method discussed above with very good agreement. In summary, NMR quantifications of brain glucose performed to date have been very consistent, when measured over similar plasma glucose concentrations.

Michaelis-Menten model of glucose transport Interpretations of glucose transport have been based on standard Michaelis-Menten model (25) for which it is among others assumed that the physical distribution space of glucose at steady-state equals the brain water phase (7, 25, 26). It has therefore become customary to assume a uniform glucose concentration past the blood-brain barrier. The model further assumes that classical Michaelis- Menten kinetics are valid to describe the unidirectional fluxes across the lumenal and ablumenal membrane, i.e. unidirectional influx Tin is given by Tmax Gplasma/(Kt+Gplasma) and conversely for efflux. It has also become customary to assume symmetric kinetic constants for influx and efflux across the blood-brain barrier. Furthermore, it has been assumed that cerebral glucose consumption is constant at euglycemia and above, which is consistent with arteriovenous difference measurements and blood flow measurements in animals and humans (26, 27). With these assumptions intracellular glucose is approximately 1mmol/g wet weight in the human brain at euglycemia (9), which is well above the Km of brain hexokinase (50mM). Rolf Gruetter et al. 3

The steady state brain glucose concentration in the standard Michaelis-Menten model is given by the expression Error!. (1)

For the derivation of equations equivalent to Eq. 1, see refs in e.g. (10, 14). Brain glucose Gbrain is given in mmol/g and the rates Tmax, CMRglc are given in mmol/g min, whereas plasma glucose Gplasma and the Michaelis-Menten constant Kt are given in mM. The physical distribution space of glucose was assumed at Vd=0.77 ml/g. It is obvious from Eq. [1] that at saturating plasma glucose concentrations, i.e. when Gplasma>>(Tmax/CMRglc+1) Kt, Gbrain approaches the upper limit of VdKt (Tmax/CMRglc-1). Most localized glucose transport kinetic measurements in human brain (1, 2, 4, 9, 14) have been highly consistent with an apparent Michaelis-Menten constant Kt of 4-5 mM for transport and an average maximal transport rate Tmax of approximately 1mmol/g min. These kinetic constants were based on the standard Michaelis-Menten model, which also predicted that brain glucose should remain below 5mmol/g when plasma glucose is below 30mM. However, the highest plasma glucose value previously examined was 13.5mM, as tabulated in (10). In a recent study we measured brain glucose concentrations in humans over a much larger range than ever attempted before. All glucose concentrations measured at plasma glucose above 20mM were above 5 mmol/g. Brain glucose concentration was 8.6 mmol/g at 28.3 mM plasma glucose concentration which implies that at 30mM, the brain glucose concentration approaches 9 mmol/g.

Reversible Michaelis-Menten model of steady-state glucose transport

The measurement of brain glucose concentrations substantially above the Km of GLUT-1 indicates that product formation is not unidirectional, i.e. the reverse reaction may proceed at significant rates at steady state. Glucose binding to the transporter at the ablumenal membrane may partially inhibit the unidirectional influx. Such behavior can be expected for the glucose transporter GLUT-1, since it has been shown that cytochalasin B, which binds to the efflux binding site, increases the Ki for maltose binding at the sugar influx site of the carrier (28). When the product formation is not unidirectional, reversible Michaelis-Menten kinetics are applicable. Such a situation amounts to replacing the apparent Michaelis-Menten constant in the standard model Kt by the term Km+Gbrain for product formation (transport in, Tin), and by Km+Gplasma for the reverse reaction Tout, conversely (29). This can be interpreted as reduced affinity for the forward reaction (influx) when substantial substrate (brain glucose, Gbrain) is present, a situation that would result in apparent asymmetric kinetic properties, when assessed using the standard model. At steady state, the reversible Michaelis-Menten model results in the expression Error! , (2) which predicts that brain glucose is a linear function of plasma glucose (10), which is a general consequence of using reversible Michaelis-Menten formalism for the unidirectional transport rates. To evaluate whether incorporating such a mechanism can accommodate these measurements more consistently, Eq [2] was fitted to the brain glucose concentrations measured as a function of plasma glucose. The best fit resulted in Km=0.6±2.0 mM and Tmax/CMRglc=2.3±0.2. The corresponding best fit (solid line) is shown in Fig. 3. The residuals of this best indicated the absence of any trend in the residuals. The linear relationship between plasma glucose and brain glucose is unlikely to be affected by errors in brain glucose quantification since neither potential quantification errors nor a vascular contribution to the signal alter the linearity of the curve or any of the model assumptions. The Rolf Gruetter et al. 4

linearity is also directly illustrated in Fig. 2. To determine whether the previous 13C MRS quantification can predict the high brain glucose values observed when using the reversible model, we fitted Eq. [2] also to the data reported in (30) which are included in Fig. 3 (open squares). When Eq. 2 was fitted to the previously reported 13C MRS quantification, the dashed line in Fig. 3 was obtained, which is an excellent extrapolation of brain glucose values to the present study. A linear brain glucose to plasma glucose relationship is consistent with earlier reports suggesting the presence of a high-affinity low-capacity transport system (31) or a non-saturable transport mechanism (32), since both mechanisms give rise to a linear relationship as well. Therefore we cannot, based on our measurements, exclude the presence of such a mechanism. However, it is interesting to note that a similar analysis of kinetic tracer experiments in rat brain suggested that the difference of the two models might manifest itself in the standard model as such a non-saturable component (33).

Reported values for the apparent Kt for glucose transport range from 2mM to 14mM, reviewed in e.g. (5, 34). Noteworthy is the observation that the reported Km for glucose transport varies in the erythrocyte from 0.5mM in the zero-trans entry experiments to the equilibrium exchange Km of 30mM (35). Measurements of brain glucose over an extended range of plasma glucose were compared with those previously reported over a much narrower range of plasma glucose (9). The apparent Michaelis- Menten constants Tmax, Km were found to be in excellent agreement and thus appeared to be independent of the range of glucose levels studied when using the reversible Michaelis-Menten model for transport. Since the transporter isoform at the blood-brain barrier and at the erythrocyte membrane are very similar, the kinetic constant for half-maximal transport, Kt, in principle should be of similar magnitude. Investigators have measured glucose transport from the plasma to brain glucose relationship and reported a Kt of 9-14 mM using the standard model (5, 6, 36). It is interesting to note that these studies measured a Kt that is markedly higher than the Km measured in erythrocyte model systems against a zero intracellular glucose concentration (35). Such high apparent Kt can be taken as an indication of an almost linear brain-blood glucose relationship.

Evaluation of the physical distribution space of glucose in vivo A limitation of determining kinetic constants from the relation between brain and plasma glucose concentration is that it depends on the assumption that the rate-limiting step for glucose transport into the brain cells is at the blood-brain barrier. Alternatively, the rate-limiting step could be at the cellular membrane in which case the glucose signal is entirely in extracellular space in near equilibrium with the plasma glucose. Intracellular glucose would be zero and transport at the cell membranes would be rate limiting for metabolism. It is generally assumed that brain glucose is evenly distributed between the intra- and extracellular compartment. However, a recent study suggested that a significant fraction of glucose taken up by the brain enters a compartment that is approximately 20% of the brain volume, which is consistent with the extracellular space (37). Recently it was suggested brain glucose transport kinetics inferred from the time course of the change in brain glucose measured after a rapid increase in plasma glucose concentration imply a large distribution volume of brain glucose. In that study, the observed time course of changes in brain glucose concentration was directly compared to the time course predicted from the kinetic constants derived from steady-state measurements of brain glucose (14). When all the glucose was assumed to be confined to the extracellular space (approximately 0.2 ml/g), Michaelis-Menten kinetics at the blood-brain barrier gave time courses that closely paralleled plasma glucose, inconsistent with the observed delay between brain and plasma glucose. Rolf Gruetter et al. 5

Further investigation into the distribution of brain glucose in the brain was performed by measuring the diffusion behavior of glucose by NMR. In model systems it has been shown that extra- and intracellular signal can be separated based on a different diffusion attenuation of the MR signal (38- 40). In vivo, most metabolites are predominantly localized in the intracellular space and we therefore sought to determine whether brain glucose in vivo has similar diffusion characteristics to that of known intracellular metabolites. In a preliminary assessment, the diffusion-weighted signal of brain glucose was observed at very large b values of 50000 s/mm2 (41) suggesting that the apparent diffusion coefficient of brain glucose in vivo is reduced by about two orders of magnitude compared to free glucose. This implies a substantial fraction of brain glucose being intracellular at the used hyperglycemic conditions.

References 1. Brooks DJ, Gibbs JSR, Sharp P, Herold S, Turton DR, Lthra SK, et al. Regional Cerebral glucose transport in Insulin-Dependent Diabetic Patients studied using [11C]3-O-Methyl-D-Glucose and Positron emission tomography. J. Cereb. Blood Flow Metab. 1986: 6: 240-244. 2. Feinendegen LE, Herzog H, Wieler H, Patton DD, Schmid A. Glucose transport and utilization in the human brain: Model using carbon-11 methylglucose and positron emission tomography. J. Nucl. Med. 1986: 27: 1867-1877. 3. Gutniak M, Blomqvist G, Widen L, Stone-Elander S, Hamberger B, Grill V. D-[U-11C]glucose uptake and metabolism in the brain of insulin-dependent diabetic subjects. Am. J. Physiol. 1990: 258: E805-812. 4. Blomqvist G, Gjedde A, Gutniak M, Grill V, Widén L, Stone-Elander S, et al. Facilitated transport of glucose from blood to brain in man and the effect of moderate on cerebral glucose utilization. Eur. J. Nucl. Med. 1991: 18: 834-837. 5. Mason GF, Behar KL, Rothman DL, Shulman RG. NMR determination of intracerebral glucose concentration and transport kinetics in rat brain. J. Cereb. Blood Flow Metab. 1992: 12: 448-455. 6. Holden JE, Mori K, Dienel GA, Cruz NF, Nelson T, Sokoloff L. Modeling the Dependence of Hexose Distribution Volumes in Brain on Plasma Glucose Concentration: Implications for Estimation of the Local 2-Deoxyglucose Lumped Constant. J. Cereb. Blood Flow Metab. 1991: 11: 171-182. 7. Gjedde A, Diemer NH. Autoradiographic determination of Regional Brain Glucose Content. J. Cereb. Blood Flow Metab. 1983: 3: 303-310. 8. Silver IA, Erecinska M. Extracellular glucose concentration in mammalian brain: Continuos monitoring of changes during increased neuronal activity and upon limitation in oxygen supply during normo-, hypo- and hyperglycemic animals. J. Neurosci. 1994: 14: 5068-5076. 9. Gruetter R, Novotny EJ, Boulware SD, Rothman DL, Mason GF, Shulman GI, et al. Direct measurement of brain glucose concentrations in humans by 13C NMR spectorscopy. Proc. Natl. Acad. Sci. USA 1992: 89: 1109-1112. 10. Gruetter R, Ugurbil K, Seaquist ER. Steady-State Cerebral Glucose Concentrations And Transport In The Human Brain. J. Neurochem. 1998: 70: 397-408. 11. Gruetter R, Garwood M, Ugurbil K, Seaquist ER. Observation of resolved glucose signals in 1H NMR spectra of the human brain at 4 Tesla. Magn. Reson. Med. 1996: 36: 1-6. 12. Frahm J, Kruger G, Merboldt KD, Kleinschmidt A. Dynamic uncoupling and recoupling of perfusion and oxidative metabolism during focal brain activation in man. Magn. Reson. Med. 1996: 35: 143-148. Rolf Gruetter et al. 6

13. Gruetter R, Rothman DL, Novotny EJ, Shulman GI, Prichard JW, Shulman RG. Detection and Assignment of the Glucose Signal in 1H NMR Spectra of the Human Brain. Magn. Reson. Med. 1992: 26: 183-188. 14. Gruetter R, Novotny EJ, Boulware SD, Rothman DL, Shulman RG. 1H NMR studies of glucose transport in the human brain. J. Cereb. Blood Flow Metab. 1996: 16: 427-438. 15. Van Zijl PC, Davis D, Eleff SM, Moonen CT, Parker RJ, Strong JM. Determination of cerebral glucose transport and metabolic kinetics by dynamic MR spectroscopy. Am. J. Physiol. 1997: 273: E1216-E1227. 16. Keltner JR, Wald LL, Ledden PJ, Chen YC, Matthews RT, Kuestermann EH, et al. A localized double-quantum filter for the in vivo detection of brain glucose. Magn. Reson. Med. 1998: 39: 651-656. 17. van Zijl PCM, Chesnick AS, DesPres D, Moonen CTW, Ruiz-Cabello J, Van Gelderen P. In vivo proton spectroscopy and spectroscopic imaging of [1-13C]-glucose and its metabolic products. Magn Reson Med 1993: 30: 544-551. 18. Inubushi T, Morikawa S, Kito K, Arai T. 1H-detected in vivo 13C NMR spectroscopy and imaging at 2T magnetic field: efficient monitoring of 13C-labeled metabolites in the rat brain derived from 1-13C-glucose. Biochem. Biophys. Res. Commun. 1993: 191: 866-872. 19. Mueckler M. Facilitative Glucose Transporters. Eur. J. Biochem. 1994: 219: 713-725. 20. Pardridge WM, Boado RJ, Farrell CR. Brain-type glucose transporter (GLUT-1) is selectively localized to the blood-brain barrier. Studies with quantitative western blotting and in situ hybridization. J Biol Chem 1990: 265: 18035-18040. 21. Maher F, Vannucci SJ, Simpson IA. Glucose transporter isoforms in brain: Absence of GLUT3 from the blood-brain barrier. J. Cereb. Blood Flow Metab. 1993: 13: 342-345. 22. Leino RL, Gerhart DZ, van Bueren AM, McCall AL, Drewes LR. Ultrastructural localization of GLUT 1 and GLUT 3 glucose transporters in rat brain. J Neurosci Res 1997: 49: 617-626. 23. Hetherington HP, Pan JW, Mason GF, Adams D, Vaughn MJ, Twieg DB, et al. Quantitative H-1 spectroscopic imaging of human brain at 4.1 T using image segmentation. Magn. Reson. Med. 1996: 36: 21-29. 24. Kreis R. Quantitative localized 1H MR spectroscopy for clinical use. Prog. NMR. Spectrosocpy 1997: 31: 155-195. 25. Lund-Andersen H. Transport of glucose from Blood to Brain. Physiol. Rev. 1979: 59: 305-352. 26. Pappenheimer JR, Setchell BP. Cerebral glucose transport and oxygen consumption in sheep and rabbits. J. Physiol. 1973: 283: 529-551. 27. Boyle PJ, Nagy RJ, O'Connor AM, Kempers SF, Yeo RA, Qualls C. Adaptation in brain glucose uptake following recurrent hypoglycemia. Proc. Natl. Acad. Sci. USA 1994: 91: 9352-9356. 28. Carruthers A, Helgerson AL. Inhibitions of sugar transport produced by ligands binding at opposite sides of the membrane. Evidence for simultaneous occupation of the carrier by maltose and cytochalasin B. Biochemistry 1991: 30: 3907-3915. 29. Cunningham VJ, Hargreaves RJ, Pelling D, Moorhouse SR. Regional blood-brain glucose transfer in the rat: a novel double-membrane kinetic analysis. J. Cereb. Blood Flow Metab. 1986: 6: 305- 314. 30. Gruetter R, Novotny EJ, Boulware SD, Rothman DL, Mason GF, Shulman GI, et al., Non-invasive Measurements of the cerebral steady-state glucose concentration and transport in humans by 13C magnetic resonance., in Frontiers in Cerebral Vascular Biology: Transport and its Regulation, L. Drewes and A. Betz, Editors. 1993, Plenum Press: New York. p. 35-40. Rolf Gruetter et al. 7

31. Gjedde A. High- and low-affinity transport of D-glucose from blood to brain. J. Neurochem. 1981: 36: 1463-1471. 32. Pardridge WM. Brain Metabolism: A Perspective From the Blood-Brain Barrier. Physiol. Rev. 1983: 63: . 33. Cunningham VJ. The influence of transport and metabolism on brain glucose content. Ann. N Y Acad. Sci. 1986: 481: 161-173. 34. Gjedde A, Blood-Brain Glucose Transfer, in Physiology and pharmacology of the Blood-Brain Barrier, M. Bradbury, Editor. 1992, Springer Verlag: New York. p. 65-117. 35. Carruthers A. Facilitated diffusion of glucose. Physiol. Rev. 1990: 70: 1135-1176. 36. Gjedde A, Christensen O. Estimates of Michaelis-Menten Constants for the Two Membranes of the Brain Endothelium. J. Cereb. Blood Flow Metab. 1984: 4: 241-249. 37. Knudsen GM, Pettigrew KD, Paulson OB, Hertz MM, Patlak CS. Kinetic analysis of blood-brain barrier transport of D-glucose in man: quantitative evaluation in the presence of tracer backflux and capillary heterogeneity. Microvasc. Res. 1990: 39: 28-49. 38. Van Zijl PC, Moonen CT, Faustino P, Pekar J, Kaplan O, Cohen JS. Complete separation of intracellular and extracellular information in NMR spectra of perfused cells by diffusion-weighted spectroscopy. Proc Natl Acad Sci U S A 1991: 88: 3228-3232. 39. Pfeuffer J, Flogel U, Leibfritz D. Monitoring of cell volume and water exchange time in perfused cells by diffusion-weighted h-1 nmr spectroscopy. NMR Biomed. 1998: 11: 11-18. 40. Pfeuffer J, Flogel U, Dreher W, Leibfritz D. Restricted diffusion and exchange of intracellular water - theoretical modelling and diffusion time dependence of h-1 nmr measurements on perfused glial cells. NMR Biomed. 1998: 11: 19-31. 41. Pfeuffer J, Tkac I, Gruetter R. Diffusion weighted in vivo 1H NMR spectroscopy of brain glucose and metabolites. in 15th Annual Mtg. ESMRMB. 1998. Geneva, : p. 36. 42. Gruetter R, Seaquist E, Kim S-W, Ugurbil K. Localized in vivo 13C NMR of glutamate metabolism. Intitial results at 4 Tesla. Dev. Neurosci. 1998: in press: . Rolf Gruetter et al. 8

b [1-13C] glucose

a * Glu 3b + 5b Gln inositol Asp

100 90 80 70 60 50 Fig. 1 13C NMR detection of labeled (1-13C) a and b glucose at the C1 position as well as detection of natural abundance glucose C3 plus C5 of b-glucose at 76.6 ppm (arrow) described previously (42). The spectrum was acquired from a whole head in 12 min as described in detail elsewhere (10). Rolf Gruetter et al. 9

NAA Plasma glucose Glc Cr Cr Cho Glu 21 mM

9.6 mM

4.7 mM

5 4 3 2 1 0 Fig. 2 Direct 1H NMR detection of brain glucose signal in a 27 ml volume. The Figure is reproduced with permission from Ref. (10).

Rolf Gruetter et al. 10

Brain glucose (µmol/g) 10

8

6

4

2

0 0 5 10 15 20 25 30 Plasma glucose (mM)

Fig. 3 Linear relationship between brain and plasma glucose concentrations. The closed squares are from (10) and the open squares are from (14). The solid line indicates the best fit of Eq. 2 to the solid squares and the dashed line to the open squares. Reproduced with permission from (10).