Bioorganic & Medicinal Chemistry 21 (2013) 27102714

Contents lists available at SciVerse ScienceDirect

Bioorganic & Medicinal Chemistry

journal homepage: www.elsevier.com/locate/bmc

A NMR method to determine the anomeric specicity of glucose phosphorylation

Thomas Richter, Stefan Berger
Institute of Analytical Chemistry, University Leipzig, Johannisallee 29, D-04103 Leipzig, Germany

a r t i c l e

i n f o

a b s t r a c t
A NMR method related to 2D CH correlation with an additional double quantum lter for 31P spin coupling was employed to follow the reaction kinetics of the two anomers of glucose during phosphorylation catalyzed by the enzyme yeast hexokinase. The kinetic parameters according to Michaelis Menten for these reactions have been determined and it is shown that the b-anomer of glucose is phosphorylated faster by a factor of 1.4 versus the a-anomer. Use of human liver glucokinase as an enzyme yields more complex kinetics. 2013 Elsevier Ltd. All rights reserved.

Article history: Received 28 January 2013 Revised 8 March 2013 Accepted 12 March 2013 Available online 24 March 2013 Keywords: Glucose Phosphorylation Hexokinase HCP-NMR

1. Introduction Phosphorylation is one of the most abundant reactions found in biochemical systems,1 since phosphate groups are used as structural elements (e.g., for nucleic acids), they change protein structure and function, and regulate enzymatic activity as well as signaling processes. In proteins the three amino acids serine, threonine and tyrosin,2 but also basic amino acids like histidine, arginine and lysine1,3 are being phosphorylated and this chemical transformation initiates signaling throughout the protein network in cell systems. Also carbohydrates can act as the targets of phosphorylating enzymes and OH groups at different position can be affected.4,5 The most simple case is the glucose phosphorylation at the 6-position leading to glucose-6-phosphate (G6P).6 From a mechanistic point of view, it is of interest, whether a phosphorylating enzyme will transform both anomers of glucose at the same or with a different rate to the two anomeric 6-phosphates (see Scheme 1). This anomeric specicity of hexokinase has been reviewed very recently by Malaisse7 who has worked on this issue for more than a quarter of a century. Differences of anomeric specicities of hexokinases from different organisms and organs and in different isoforms have also been extensively discussed.810 This distinction is important in assessing for example, diabetic patients, but difcult to analyze and has been performed in the biochemical context using [U-14C]-labelled glucose and tritiated water. An important NMR paper by Peters and co-workers11 conrmed with the
Corresponding author. Tel.: +49 341 9736101.
E-mail address: stberger@rz.uni-leipzig.de (S. Berger). 0968-0896/$ - see front matter 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.bmc.2013.03.008

saturation transfer difference technique (STD-NMR) the earlier ndings that a-D-glucose is the preferred binding partner of hexokinase, although it is known that b-D-glucose is phosphorylated faster kinetically. Two other papers used 31P NMR techniques12,13 to analyze these reactions. Several crystallographic investigations have determined the binding site of glucose in hexokinases but were not able to address the anomeric specicity from a structural point of view.1416 Normal NMR methods will have problems due to the very similar 1H and 13C chemicals shifts of the involved species. Especially standard proton NMR turns out to be useless, because the signals of the two anomers and their phosphorylation products cannot separately be integrated. Similar is true for the 13C NMR spectra, as shown in Figure 1 for the fully 13C-labelled glucose, which is now rather cheaply available due to its frequent use in structural biology and does show all the C,C spin couplings. Whereas the two anomeric signals of C-1 are nicely separated, the signals of C-6 cannot be separately evaluated for a kinetic analysis. Figure 2 shows the mixture after partial phosphorylation and it is apparent that now also the two anomeric signals of C-1 are overlaid by the signals of the phosphorylated glucose. The signal of C-6 in the phosphorylated form at dC = 66 does not split into the two anomeric components. We have earlier developed a two-dimensional HSQC-NMR related HCP-NMR variant,17 which detects only those H, C pairs in a phosphorylated molecule, where the carbon atom displays an additional spin coupling via two or three bonds to the phosphorus. In contrast to former three-dimensional HCP-NMR experiments used for sequential backbone assignment of nucleic acids,1820 the two-dimensional approach allows for shorter experiment

T. Richter, S. Berger / Bioorg. Med. Chem. 21 (2013) 27102714

H OH H HO HO H
Glucose

2711

H OPO 3 O
Hexokinase

2-

H HO HO H H

OH OH
ATP ADP

OH OH

Glucose-6-phosphate

Scheme 1. Phosphorylation of glucose.

Figure 1.

13

C NMR spectrum of the anomeric mixture of D-glucose.

Figure 2.

13

C NMR spectrum of the anomeric mixture of D-glucose after partial phosphorylation by hexokinase.

times and thus for the monitoring of enzymatically catalyzed phosphate transfer reactions. We felt that this method would be ideal to apply to the question raised above and report here our initial

results on the kinetics of glucose phosphorylation using HCP as the detecting method. In D-glucose-6-phosphate the HSQC-HCP technique will only reveal the CH groups of C-6 and C-5 having a

2712

T. Richter, S. Berger / Bioorg. Med. Chem. 21 (2013) 27102714

Figure 3. Overlay of HSQC spectra of D-glucose (blue) and G6P (red) derived from D-glucose by yeast hexokinase phosphorylation.

Figure 4. HCP spectrum of G6P derived from D-glucose by hexokinase phosphorylation; signals correspond to C-6 (dC = 66), C-5a (dC = 74) and C-5b (dC = 78). The remaining signals are ltered off by the 31P double quantum lter of the HCP method.

distance of two respective three bonds to the phosphorus atom. Therefore the two phosphorylated anomers of glucose can be distinguished and are separately detectable. 2. Results and discussion As a proof for the resolving power of the HCP method rst normal HSQC spectra of D-glucose before and after complete phosphorylation to D-glucose-6-phosphate were recorded and

compared by spectral overlay.21 As shown in Figure 3 most signals of the phosphorylated (red) and the unphosphorylated D-glucose (blue) overlap and cannot safely be used for separate integration. Quite in contrast is the HCP spectrum shown in Figure 4 which reveals as expected only three signals, one from the undistinguishable C-6 groups and two from the two anomers of C-5. The latter are nicely separated and can be easily used for a kinetic analysis. We therefore conclude that HCP is the method of choice for the scientic question raised.

T. Richter, S. Berger / Bioorg. Med. Chem. 21 (2013) 27102714

2713

the reaction velocity v is described by the three parameters [S], the substrate concentration, vmax, the maximum velocity, and KM, the MichaelisMenten constant. After integration to Eq. 2, this can be transformed in Eq. 3, where [P] is the product concentration, [S0] the starting substrate concentration and W the Lambert W function:23

  P v max t K M ln 1 P S0 S0 e K M P S0 K M W @ KM 0
S0 v max t

1 A 3

Figure 5. Relative integrals of C-5a (red), C-5b (blue) and the sum of the signals C-6 (black) during D-glucose phosphorylation by yeast hexokinase.

The kinetic data have been recorded several times using a sample of 200 mM 13C6-D-glucose, 300 mM ATP in 50 mM Tris-d11 buffer and 13.3 mM MgCl2 added. The pH was adjusted to 8.0 at a temperature of 25 C. The reaction was started by addition of 6 8 U yeast hexokinase to a 0.9 ml sample. 100 individual spectra were sampled where each spectrum required 5 min. All signals have been integrated and in Figure 5 the kinetic result is plotted for the sum of the signals of C-6 and the two individual C-5 signals. In Figure 6 the anomeric ratio of the glucose-6-phosphates is plotted versus the time of their formation. The ba ratio starts with about 2.5 at the beginning of the phosphorylation reaction and levels off at a factor of about 1.75 at the nal equilibrium. This curve suggests that not only the anomers of free glucose molecules interconvert, but also the anomers of the formed glucose-6-phosphates reach a slightly different equilibrium of interconversion. Additionally, the anomeric ratio of the produced glucose-6-phosphates seems to be constant until depletion of b-D-glucose occurs. This indicates that anomeric specicity of glucose phosphorylation is not or only in a small extend overlaid by the anomeric interconversion of glucose-6-phosphates. Enzyme kinetics are usually treated by the MichaelisMenten equation,22 where in Eq. 1

V max S K M S

Using this mathematical background, the experimental data were tted and yielded for the a-glucose the values vmax = 37.4 lM/min, and KM = 2.20 mM after normalization to the initial ratio of D-glucose anomers; for the b-glucose vmax = 15.9 lM/min, and KM = 0.49 mM were obtained. Errors of the calculated vmax values for a- and b-glucose phosphorylation were 1.1% and 1.3%, respectively. For KM values of a- and b-glucose phosphorylation the errors were 11.9% and 6.7%, respectively. The ratio of vmax for the phosphorylation of the two anomers was calculated to be 1.38:1 after normalization to the initial anomeric ratio as a substitute for the ratio of hexokinase molecules bound to a- and b-glucose, respectively. At this point it is to say that these apparent kinetic constants fail to match to literature values, since the mathematical model for curve tting is greatly simplied by neglect of effects like product inhibition. However, our goal was to show the potential of HCP-NMR to record kinetic curves, which is indeed the case. For exact calculation of kinetic constants, much more complex kinetic models have to be used, which take into account additional regulating effects on the enzymatically catalyzed reaction. Interesting enough is the change of the kinetics if instead of yeast hexokinase a different enzyme is used. In Figure 7 we show the result of the same reaction but with human liver glucokinase as the enzyme. Here we were only able to t the initial rates of the phosphorylation of the two anomers, which do not show a signicant anomeric preference, but could not determine a Michaelis Menten constant. Apparently, this enzyme causes a much more complex reaction, and calculation of kinetic constants would require a more accurate mathematical description. Figure 7 also shows a much higher noise of the kinetic curve due to a lower Dglucose concentration and indicates a low sensitivity of the HCP method. Indeed, signal-to-noise ratio exceeds a value of ve at a

Figure 6. Concentration of produced a- (red) and b-D-glucose-6-phosphate (blue) and anomeric b/a ratio during hexokinase phosphorylation.

2714

T. Richter, S. Berger / Bioorg. Med. Chem. 21 (2013) 27102714

the starting concentration of 200 mM D-glucose. Resulting values were divided by the amount of kinase added per ml of buffer, which varied between different measurements. Normalized MichaelisMenten parameters from four measurements were averaged. Uncertainties of tted parameters were propagated to get uncertainties of normalized MichaelisMenten parameters. 4. Conclusions We have shown in this work, that the HCP-NMR method is capable to observe the kinetics of glucose phosphorylation for both anomers to extract the MichaelisMenten parameters with hexokinase as the enzyme. However, the mathematical model used here does not consider effects like product inhibition and fails to describe the kinetics of human liver glucokinase. To get exact tting parameters, an accurate choice of the kinetic model is essential, but requires extensive knowledge of the underlying differential equations. The employed NMR method needs 13C labelled material and thus cannot be used under physiological conditions, however, does not require radioactively labelled material and works in a cell free system. The method shows for the rst time quantitatively that the phosphorylation of the anomers of glucose at carbon 6 follows two different time domains. Acknowledgment This work was supported by the German Research Council, Sonderforschungsbereich 610, Project A2. References and notes Isotopically labelled compounds were purchased from Eurisotop. All other chemicals were purchased from SigmaAldrich. Yeast hexokinase isoform PII and human liver glucokinase were purchased from SigmaAldrich and used without further purication. For hexokinase experiments, a 200 mM solution of 13C6-labelled D-glucose was prepared in a buffer with 300 mM ATP, 50 mM Tris-d11, 13.3 mM MgCl2 at pH 8.0 in D2O at 25 C. D-Glucose phosphorylation by yeast hexokinase was induced by adding 68 U of Hexokinase to 0.9 ml of D-glucose solution. Glucokinase experiments were performed in a 40 mM solution of 13C6-labelled D-glucose in a buffer with 50 mM ATP, 20 mM Tris-d11, 10 mM MgCl2 at pH 8.2 in D2O at 30 C. D-Glucose phosphorylation by human liver glucokinase was induced by adding 30 lg of enzyme to 0.9 ml of D-glucose solution. All NMR spectra were recorded on a Bruker DRX-600 spectrometer with a 5 mm TBI probe. HCP spectra were measured as described before [16] in four scans with 576 64 data points. Spectral widths for 1H and 13C were set to 4 ppm (offset on water) and 50 ppm (offset 75 ppm), respectively. All spectra were referenced to DSS according to the Xsi-scale. For kinetic experiments, HCP spectra and 13C spectra were sampled with 100 individual spectra each after induction of the enzymatic reaction. Single HCP spectra and 13C spectra required 5 min and 1 min, respectively, thus each two spectra took 6 min. Series of HCP spectra were processed in TopSpin 3.0 using self-written AU programs and integrated with the intser command. Extracted integral data from hexokinase assays was tted with Mathematica 6.0 with respect to Eq. 2. Fitted parameters vmax and KM for a- and b-D-glucose were normalized to their corresponding [S0]-values to get in sum
la, J.; Fra 1. Cies czyk, T.; Rode, W. Acta Biochim. Pol. 2011, 58, 137. 2. Rao, N. N.; Torhani, A. Mol. Microbiol. 1990, 4, 1083. 3. Himmel, S.; Wolff, S.; Becker, S.; Lee, D.; Griesinger, C. Angew. Chem., Int. Ed. 2010, 49, 8971. 4. Verzr, F.; Wenner, V. Biochem. J. 1948, 42, 42. 5. Hoare, D. S.; Kerly, M. Biochem. J. 1954, 58, 38. 6. Long, C. Biochem. J. 1952, 50, 407. 7. Malaisse, W. J. Adv. Biol. Chem. 2012, 2, 1. 8. Malaisse, W. J.; Giroix, M.-H.; Dufrane, S. P.; Malaisse-Lagae, F.; Sener, A. Cancer Res. 1985, 45, 6376. 9. Malaisse-Lagae, F.; Giroix, M. H.; Sener, A.; Malaisse, W. J. Biol. Chem. HoppeSeyler 1986, 367, 411. 10. Jijakli, H.; Courtois, P.; Zhang, H. X.; Sener, A.; Malaisse, W. J. J. Biol. Chem. 2003, 278, 4531. 11. Blume, A.; Fitzen, M.; Benie, A. J.; Peters, T. Carbohydr. Res. 2009, 344, 1567. 12. Lowman, D.; Ensley, H.; Williams, D. Carbohydr. Res. 1998, 306, 559. 13. Lutz, N. W.; Yahi, N.; Fantini, J.; Cozzone, P. J. Eur. J. Biochem. 1996, 238, 470. 14. Marotta, D. E.; Anand, G. R.; Anderson, T. A.; Miller, S. P.; Okar, D. A.; Levitt, D. G.; Lange, A. J. Arch. Biochem. Biophys. 2005, 436, 23. 15. Cordeiro, A. T.; Cceres, A. J.; Vertommen, D.; Concepcin, J. L.; Michels, P. A. M.; Verses, W. J. Mol. Biol. 2007, 372, 1215. 16. Kuettner, E. B.; Kettner, K.; Keim, A.; Svergun, D. I.; Volke, D.; Singer, D.; Hoffmann, R.; Mller, E.-C.; Otto, A.; Kriegel, T. M.; Strter, N. J. Biol. Chem. 2010, 285, 41019. 17. Raeck, C.; Berger, S. Anal. Bioanal. Chem. 2007, 389, 2161. 18. Heus, H. A.; Wijmenga, S. S.; van de Ven, F. J. M.; Hilbers, C. W. J. Am. Chem. Soc. 1994, 116, 4983. 19. Wijmenga, S. S.; Heus, H. A.; Leeuw, H. A. E.; Hoppe, H.; van der Graaf, M.; Hilbers, C. W. J. Biomol. NMR 1995, 5, 82. 20. Marino, J. P.; Schwalbe, H.; Anklin, C.; Bermel, W.; Crothers, D. M.; Griesinger, C. J. Biomol. NMR 1995, 5, 87. 21. Berger, S.; Braun, S. 200 and More NMR Experiments. A Practical Course; WileyVCH: Weinheim, Germany, 2004. 22. Menten, L.; Michaelis, M. I. Biochem. Z. 1913, 49, 333. 23. Schnell, S.; Mendoza, C. J. Theor. Biol. 1997, 187, 207.

Figure 7. Relative integrals of C-5a (red) and C-5b (blue) during phosphorylation by human liver glucokinase.

D-glucose

G6P anomer concentration of about 3 mM (data not shown). Thus, the HCP method cannot be used to measure glucose phosphorylation at physiological concentrations, but nevertheless can be a strong tool for characterization of kinases with respect to MichaelisMenten kinetics and works in a cell free environment. 3. Experimental section