- Email: [email protected]
Oligosaccharide recognition and binding to the carbohydrate binding module of AMP-activated protein kinase
FEBS Letters 581 (2007) 5055–5059
Oligosaccharide recognition and binding to the carbohydrate binding module of AMP-activated protein kinase Ann Koay, Kieran A. Rimmer, Haydyn D.T. Mertens, Paul R. Gooley*, David Stapleton* Department of Biochemistry and Molecular Biology, Bio21 Molecular Science and Biotechnology Institute, The University of Melbourne, Parkville, Victoria, Australia Received 1 August 2007; revised 4 September 2007; accepted 20 September 2007 Available online 29 September 2007 Edited by Christian Griesinger
Abstract The AMP-activated protein kinase (AMPK) contains a carbohydrate-binding module (b1-CBM) that is conserved from yeast to mammals. b1-CBM has been shown to localize AMPK to glycogen in intact cells and in vitro. Here we use Nuclear Magnetic Resonance spectroscopy to investigate oligosaccharide binding to 15N labelled b1-CBM. We find that b1-CBM shows greatest affinity to carbohydrates of greater than five glucose units joined via a,1 fi 4 glycosidic linkages with a single, but not multiple, glucose units in an a,1 fi 6 branch. The near identical chemical shift profile for all oligosaccharides whether cyclic or linear suggest a similar binding conformation and confirms the presence of a single carbohydrate-binding site. 2007 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. Keywords: AMP-activated protein kinase; Carbohydratebinding module; NMR; Oligosaccharide; Glycogen
1. Introduction The AMP-activated protein kinase (AMPK) is an evolutionarily conserved energy sensor found in mammals and known to coordinate cellular metabolism and systemic energy balance in response to energy demands and a range of stimuli [1]. AMPK is a highly conserved heterotrimeric complex comprising a catalytic a subunit and regulatory b and c subunits that have been identified in plants, yeast, nematodes, flies and mammals. AMPKb1 subunit was recently shown to contain a glycogenbinding domain. Using secondary structure predictions a protease-resistant domain within residues 68–163 of AMPKb1 was identified that associated with glycogen in cell-free assays [2]. Furthermore fluorescently tagged AMPK heterotrimers associated with glycogen-containing granules in the cytoplasm of cultured human cells that was abolished when glycogenbinding domain was deleted [3]. The glycogen-binding domain has been reclassified as a member of the carbohydrate-binding
module-containing family 48 (CBM48) (http://afmb.cnrs-mrs. fr/CAZY) based on amino acid sequence, alongside other carbohydrate-binding proteins such as isoamylase and glycogen branching enzymes. To date, the structures of both rat b1CBM, in complex with the unnatural cyclic sugar b-cyclodextrin [4], and human apo b2-CBM (unpublished, PDB ident code 2F15) have been elucidated by X-ray crystallography. AMPKb1-CBM folds into a b sandwich consisting of two anti-parallel b sheets similar to structures of N-isoamylase or starch-binding domains [4]. Glycogen, a branched polymer of glucose, is a cellular store of energy and is important for whole body glucose metabolism with largest stores found in liver and skeletal muscle [5]. Glycogen is stored in cells in the form of granules that not only contain glucose, but also bound proteins that are involved in glycogen metabolism like glycogen synthase and glycogen phosphorylase [6]. Several studies link AMPK to glycogen metabolism. For example, AMPK co-immunoprecipitates with glycogen synthase [7], glycogen phosphorylase [7] and glycogen debranching enzyme [8] and mutations in the AMPK c subunit result in a novel cardiac glycogen storage disease in human individuals [9,10]. To further investigate the carbohydratebinding nature of b1-CBM we used Nuclear Magnetic Resonance (NMR) spectroscopy to identify the residues important for oligosaccharide binding to b1-CBM and to determine affinity for this association. We find that b1-CBM preferentially binds sugars with 6–7 glucose units and ideally with one type of branched sugar.
2. Materials and methods 2.1. Expression and purification of isotopically labeled b1-CBM AMPK b1-CBM was cloned into pProEX HT and expressed as a His-tag fusion protein in BL21 Escherichia coli cells [11]. AMPK b1CBM was 13C and 15N or 15N only labelled by growing cultures in a 2 L Braun Biostat Fermentor containing 1 L of minimal media with 15 NH4Cl and 13C-D -glucose (Spectral Gases Laboratory) as the sole nitrogen and carbon sources [12]. b1-CBM expression was induced by the addition of 1 mM IPTG until the nutrients were depleted after which cells were harvested, pelleted and purified as previously described [11].
*
Corresponding authors. E-mail addresses: [email protected] (P.R. Gooley), dis@unimelb. edu.au (D. Stapleton). Abbreviations: AMPK, AMP-activated protein kinase; CBM, carbohydrate-binding module; b1-CBM, AMPK b1 carbohydrate-binding module; NMR, nuclear magnetic resonance spectroscopy; HSQC, heteronuclear single quantum correlated
2.2. Nuclear magnetic resonance (NMR) experiments and data processing Purified 13C and 15N or 15N only labelled b1-CBM were dissolved into NMR buffer (25 mM Na2HPO4–NaH2PO4, 0.02% sodium azide, pH 7.5, 10% D2O) and placed into NMR sample tubes. For resonance assignment NMR spectra were acquired on 0.5 mM b1-CBM samples at 25 C using Varian 600 MHz INOVA or Bruker 800 MHz
0014-5793/$32.00 2007 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2007.09.044
5056
A. Koay et al. / FEBS Letters 581 (2007) 5055–5059
AVANCE spectrometers. All experiments utilized cryogenically cooled probes of the spectrometers to maximize sensitivity. Heteronuclear 2D and 3D resonance spectra were acquired and the data was processed with the program NMR Pipe [13] using a Gaussian function in the direct HN dimension and cosine squared functions in the indirect dimensions. All data were zero-filled once prior to Fourier transformation. Data was analysed using NMR View, v5.2.2 [14]. For sequence specific assignment of resonances 3D HNCACB [15] and CBCA(CO)NH [16] experiments were acquired and analysed, 3D C(CO)NH [17] experiments were performed to confirm assignments. 2.3. Oligosaccharide titrations To study carbohydrate recognition and binding to b1-CBM, individual titrations using 12 different oligosaccharides shown in Table 1 were obtained from Supelco (USA) and WAKO (Japan) and performed with 0.2 mM 15N b1-CBM in NMR buffer. Each oligosaccharide was prepared as a stock concentration ranging from 50 to 400 mM, except for b-cyclodextrin which was prepared as a 20 mM stock due to low solubility. Eight to 10 oligosaccharide titration points with increasing concentration from 0 to 10 mM were performed. For each titration point, a 1H–15N heteronuclear single quantum correlation (HSQC) spectrum was acquired at 800 MHz and 25 C. The final sample dilution was less than 10% of the original volume. 2.4. Dissociation constant determination Dissociation constants for oligosaccharides binding to b1-CBM were calculated by fitting the titration data for individual resonances, assuming a single ligand recognition site, in the programs xcrvfit, (v3.0.6, Robert Boyko and Brian Sykes) and xmgrace (v 5.1.2 Paul J. Turner). Maximum HN and 15N chemical shift differences were scaled according to the following equation [18]: DdTotal = ((DdHN)2 + (DdN · 0.154)2)1/2 where DdHN and DdN are the chemical shift differences for the HN and 15N resonances between the free and titrated protein. Molecular visualization and generation of figures were performed using MOLMOL v2k.2 [19].
Table 1 Oligosaccharide structures and dissociation constants (Kd) for the binding of soluble oligosaccharides to b1-CBM of AMPK at 25 C and pH 7.5 Ligand
Structure
Kd (mM)
Maltoheptaose Maltohexaose Maltopentaose Maltotetraose Maltotriose Maltose Glucose 6-phosphate
GGGGGGG GGGGGG GGGGG GGGG GGG GG PG GGG G G GG
0.36 ± 0.033 0.67 ± 0.057 3.69 ± 0.34 4.21 ± 0.12 9 to 18 N.D. N.D.
b-Cyclodextrin
0.33 ± 0.025
G
Glucosyl b-cyclodextrin
GGG G G GG GG
0.15 ± 0.016
Maltosyl b-cyclodextrin
GGG G G GG
0.51 ± 0.04
Glucosyl maltotriose Isomaltose
G GGG G
3. Results 3.1. Assignment of the 1H–15N HSQC NMR spectrum of b1CBM The 1H–15N HSQC spectrum obtained for apo b1-CBM showed a pattern of well-dispersed cross peaks (Fig. 1S, Supplementary data). A total of 13 HN correlations (M53, D54, N70, E71, K72 T85, G86, K102, Q109, N110, Q145, L146 and E161) out of an expected 95 could not be unambiguously assigned, or were not observed. 3.2. NMR titrations of oligosaccharides with b1-CBM For each oligosaccharide shown in Table 1, 8–10 1H–15N HSQC spectra were collected after progressive addition of oligosaccharide and overlayed to track peak chemical shift perturbations. Each of the titrated oligosaccharides except isomaltose and glucose-6-phosphate were found to cause significant chemical shift perturbations to the b1-CBM spectrum. An example of the overlayed spectra for maltoheptaose is shown in Fig. 1A. Saturation of the protein was possible for the linear maltodextrins: maltoheptaose and maltohexaose, and the cyclodextrins: b-cyclodextrin, glucosyl b-cyclodextrin and maltosyl b-cyclodextrin. Under the conditions of this study we could not saturate the protein sample with maltopentaose, maltotetraose, maltotriose and glucosyl-maltotriose. However, all oligosaccharides that induced chemical shift changes in b1-CBM showed similar patterns of change upon ligand binding suggesting a single carbohydrate-binding site (Figs. 1B and C, 2 and 2S, Supplementary data). A large number of resonances exhibited no chemical shift perturbation upon ligand binding. In titrations with b-cyclodextrin, glucosyl b-cyclodextrin, maltosyl b-cyclodextrin, maltoheptaose, maltohexaose and maltopentaose, a number of cross peaks were rapidly broadened upon the progressive addition of ligand until they completely vanish from the spectra, only to reappear as the oligosaccharide concentration increased. Dissociation constants (Kd) were independently determined for individual resonances that showed significant chemical shift deviation in either or both the 1H and 15N dimension upon oligosaccharide binding. Peaks that exhibited peak broadening were excluded from Kd calculations [20]. The obtained Kd values were averaged to obtain an overall Kd value for each titrated oligosaccharide as reported in Table 1. Kd values for maltotriose and glucosyl-maltotriose could not be accurately determined because titrations with these ligands did not result in sufficient saturation of the b1-CBM even at very high concentrations of oligosaccharide, suggesting weak binding to b1-CBM. For these oligosaccharides a range of Kd values were determined (Table 1). We find b1-CBM binds with higher affinity to long chain a,1 fi 4 linked oligosaccharides and this affinity diminishes with maltodextrin length (Table 1). A distinct decrease in the values of Kd is found between maltohexaose and maltopentaose where the affinity of b1-CBM for maltopentaose is approximately five times weaker than maltohexaose.
8 to 14 N.D.
4. Discussion
G
Glucose units (G) connected via an a, 1 fi 6 glycosidic bond are indicated by a solid line. All other glucose units are linked via a,1 fi 4 glycosidic bonds. pG represents glucose 6-phosphate.
In this present study, we have identified key determinants in the association of oligosaccharides with the AMPK b1 carbohydrate-binding module (b1-CBM). The oligosaccharide Kd
A. Koay et al. / FEBS Letters 581 (2007) 5055–5059
5057
Fig. 1. Oligosaccharide-induced b1-CBM chemical shift changes for individual residues. (A) Overlayed 2D 1H–15N HSQC spectra focusing on an area of chemical shift changes in the nuclei of b1-CBM from AMPK upon titration with 0–10 mM maltoheptaose at 25 C and pH 7.5. Broken arrows indicate observed peak broadening that is associated with intermediate exchange. For clarity, only residues that showed chemical shift perturbations are labelled. Mapping of chemical shift perturbations for b-cyclodextrin (B) and maltoheptaose (C) onto the structure of b1-CBM [4]. Chemical shift differences were scaled as described in methods. The width (thin to thick) and colour (yellow to red) of the backbone is indicative of the extent of perturbation. b-cyclodextrin is shown in green.
maximum average chemical shift difference (ppm)
0.8 0.6 0.4 0.2
160
145
residue number
130
115
100
glucose-6-phosphate maltose maltotriose glucosyl maltotriose maltotetraose maltopentaose bcd mbcd gbcd maltohexaose maltoheptaose
85
70
0
Fig. 2. Comparison of the maximum average chemical shift differences observed in b1-CBM upon titration with respective oligosaccharides show that various types of oligosaccharide bind and cause chemical shift changes to HN and 15N resonances of similar amino acid residues in b1-CBM. Single plots of maximum average chemical shift differences for each oligosaccharide are shown in Supplementary Fig. 2S.
values in Table 1 clearly show that b1-CBM favors associating with maltodextrins with six to seven a,1 fi 4 glucose units and
in particular maltodextrins that include an additional glucosyl unit in an a,1 fi 6 branch but not two a,1 fi 4 glucosyl units
5058
joined in an a,1 fi 6 branch. We find that b1-CBM binds glucosyl b-cyclodextrin with the highest affinity (Kd = 150 ± 16 lM) > b-cyclodextrin > maltoheptaose > maltosyl b-cyclodextrin > maltohexaose. Smaller oligosaccharides including maltopentaose, previously shown to prevent association of AMPK enzyme but not b1-CBM alone to glycogen [4], or branched oligosaccharides exhibited diminished affinities. These data suggest that AMPK b1-CBM is a Type B CBM; exhibiting an increased binding affinity towards oligosaccharides with higher degrees of polymerization, as opposed to Type A CBMs that bind to crystalline cellulose and chitin or Type C CBMs that bind smaller mono-, di- and trisaccharides [21]. Other structurally characterized Type B CBMs which are from plant and bacteria and include representatives of the CBM20, 21, 25 and 26 clans, show Kd values of 1–20 lM for oligosaccharides including b-cyclodextrin and maltoheptaose [22–24]. The 10–100-fold poorer Kd values observed here for this mammalian CBM from the CBM48 clan may indicate functional differences between these clans or that the exact substrate for this protein remains to be defined. To determine whether b1-CBM has a preference for branched oligosaccharides we obtained as many a,1 fi 6 branched molecules as commercially possible and found that glucosyl b-cyclodextrin had the highest affinity for b1-CBM. However, increased chain length on either the a,1 fi 4 or a,1 fi 6 chains like for maltosyl b-cyclodextrin, glucosyl maltotriose (compared to maltotetraose) and isomaltose resulted in reduced affinity for b1-CBM (Table 1). The latter is interesting as titrations with the disaccharide maltose, an a,1 fi 4 linkage, weakly caused the same pattern of chemical shift perturbations similar to maltoheptaose whilst isomaltose, an a,1 fi 6 linkage, did not cause any chemical shift change perturbation in b1CBM. These data suggest that AMPK b1-CBM will not associate with a,1 fi 6 branch points alone but prefer longer a,1 fi 4 chains that contain a single a,1 fi 6 glucose unit. This type of oligosaccharide would be visible to b1-CBM only during degradation of the glycogen molecule. Glycogen debranching enzyme has two catalytic activities that occur in succession; first, a transferase activity that transfers a maltotriose unit to the ‘‘main chain’’ from four glucose units attached to this main chain by a,1 fi 6 linkage. This results in an elongated section of a,1 fi 4-linked polymer, leaving a single glucose attached to it via an a,1 fi 6 linkage. Second, a glucosidase activity then cleaves the a,1 fi 6 linkage by which this glucose is attached [25]. As the regulation or rate of these two activities is unknown, a possibility exists that the halfway point of glycogen debranching enzyme’s dual-step action may provide a high affinity binding site for AMPK b1-CBM. Further intriguing is the recently reported association between AMPK b1-CBM and the glycogen debranching enzyme [8]. The region important for association with the glycogen debranching enzyme encompasses the first half of b1-CBM (b1 65-123) although association with the actual carbohydrate-binding site described here appears unlikely as mutation of residues of the carbohydrate-binding site such as W100G and K126Q failed to prevent association between the two proteins [8]. The crystal structure of b1-CBM in complex with b-cyclodextrin identifies five residues critical for carbohydrate association; W100, W133, K126, L146 and N150 [4]. Results in this study show that, except for W100, the HN of these residues or residues sequentially near exhibit significant chemical shift perturbations of more than one standard deviation from the mean
A. Koay et al. / FEBS Letters 581 (2007) 5055–5059
chemical shift change. Residue L146 which pierces the b-cyclodextrin ring in the b1-CBM structure [4], could not be assigned in the HSQC spectrum, however its neighbouring residue S144 is immediately broadened by maltoheptaose and is significantly shifted by all other binding oligosaccharides. The peptide HN of W100 and residues sequential to it show modest chemical shift changes (Fig. 2). The peptide HN of G95 exhibits a chemical shift perturbation of more than one standard deviation from the mean, and is significantly affected by all binding oligosaccharides. In the b1-CBM structure this residue is found in the b2 strand buried below the aromatic side-chain of W100, but otherwise is remote from the carbohydrate-binding site. This residue is most likely affected by ring current shifts by the reorientation of the aromatic ring of W100 upon oligosaccharide binding. All oligosaccharides that bound to b1-CBM, regardless of being linear or cyclised, perturbed the same residues with a similar profile. As these perturbations were limited to the residues observed in the crystal structure of the b-cyclodextrin complex, we conclude that there is a single binding site and that a linear sugar such as maltoheptaose binds in a similar conformation to the cyclised dextrans [4]. We also conclude that to produce the same profile, small sugars such as maltose and maltotriose bind in multiple positions but into the subsites defined in the crystal structure, and therefore there is no preferred subsite. As the crystal structure shows a sugar of five glucosyl units would be sufficient to make contacts to all carbohydrate-binding subsites in the b1-CBM, we expected that sugars such as maltopentaose or longer would have similar Kd values. While it is surprising that maltoheptaose has the best Kd amongst all titrated linear sugars, similar observations for other CBMs have been made where oligosaccharides that are apparently much longer than necessary bind with higher affinities [22]. Therefore other influences such as structural differences of maltoheptaose versus maltopentaose or perhaps displacement of bound water may be important for determining binding affinity. In conclusion, this study reveals that the single carbohydrate-binding site of AMPK b1-CBM binds well to oligosaccharides longer than five glucose units and poorly to tri-, di, and monosaccharides. The binding face of b1-CBM recognizes carbohydrates joined via a,1 fi 4 glycosidic linkages with a single glucose unit in an a,1 fi 6 branch. We expect these results to form the basis of further exploring the ligand recognition and binding of the CBM48 clan. Acknowledgement: We wish to acknowledge NHMRC (National Health and Medical Research Council) of Australia Grant support to D.S.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.febslet.2007.09. 044. References [1] Kahn, B.B., Alquier, T., Carling, D. and Hardie, D.G. (2005) AMP-activated protein kinase: ancient energy gauge provides clues to modern understanding of metabolism. Cell Metabol. 1, 15–25.
A. Koay et al. / FEBS Letters 581 (2007) 5055–5059 [2] Polekhina, G. et al. (2003) AMPK beta subunit targets metabolic stress sensing to glycogen. Curr. Biol. 13, 867–871. [3] Hudson, E.R. et al. (2003) A novel domain in AMP-activated protein kinase causes glycogen storage bodies similar to those seen in hereditary cardiac arrhythmias. Curr. Biol. 13, 861–866. [4] Polekhina, G., Gupta, A., van Denderen, B.J., Feil, S.C., Kemp, B.E., Stapleton, D. and Parker, M.W. (2005) Structural basis for glycogen recognition by AMP-activated protein kinase. Structure (Camb.) 13, 1453–1462. [5] Roach, P.J. (2002) Glycogen and its metabolism. Curr. Mol. Med. 2, 101–120. [6] Shearer, J. and Graham, T.E. (2002) New perspectives on the storage and organization of muscle glycogen. Can. J. Appl. Physiol. 27, 179–203. [7] Chen, Z. et al. (1999) Expression of the AMP-activated protein kinase beta1 and beta2 subunits in skeletal muscle. FEBS Lett. 460, 343–348. [8] Sakoda, H. et al. (2005) Glycogen debranching enzyme association with beta-subunit regulates AMP-activated protein kinase activity. Am. J. Physiol. Endocrinol. Metab. 289, E474–E481. [9] Dyck, J.R. and Lopaschuk, G.D. (2006) AMPK alterations in cardiac physiology and pathology: enemy or ally? J. Physiol. 574, 95–112. [10] Gollob, M.H., Green, M.S., Tang, A.S. and Roberts, R. (2002) PRKAG2 cardiac syndrome: familial ventricular preexcitation, conduction system disease, and cardiac hypertrophy. Curr. Opin. Cardiol. 17, 229–234. [11] Polekhina, G., Feil, S.C., Gupta, A., O’Donnell, P., Stapleton, D. and Parker, M.W. (2005) Crystallization of the glycogen-binding domain of the AMP-activated protein kinase subunit and preliminary X-ray analysis. Acta Crystallogr., Sect. F F61, 39–42. [12] Cai, M., Huang, Y., Sakaguchi, K., Clore, G.M., Gronenborn, A.M. and Craigie, R. (1998) An efficient and cost-effective isotope labeling protocol for proteins expressed in Escherichia coli. J. Biomol. NMR 11, 97–102. [13] Delaglio, F., Grzesiek, S., Vuister, G.W., Zhu, G., Pfeifer, J. and Bax, A. (1995) NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293. [14] Johnson, B.A. and Blevins, R.A. (1994) NMR view: a computer program for the visualization and analysis of NMR data. J. Biomol. NMR 4, 603–614.
5059 [15] Wittekind, M. and Mueller, L. (1993) HNCACB, a Highsensitivity 3D NMR experiment to correlate amide-proton and nitrogen resonances with the alpha- and beta-carbon resonances in proteins. J. Magn. Reson. 101, 201–205. [16] Grzesiek, S. and Bax, A. (1992) Correlating backbone amide and side chain resonances in larger proteins by multiple relayed triple resonance NMR. J. Am. Chem. Soc. 114, 6291–6293. [17] Grzesiek, S., Anglister, J. and Bax, A. (1993) Correlation of backbone amide and aliphatic side-chain resonances in C-13/N15-enriched proteins by isotropic mixing of C-13 magnetization. J. Magn. Reson. 101, 114–117. [18] Ayed, A., Mulder, F.A., Yi, G.S., Lu, Y., Kay, L.E. and Arrowsmith, C.H. (2001) Latent and active p53 are identical in conformation. Nat. Struct. Biol. 8, 756–760. [19] Koradi, R., Billeter, M. and Wuthrich, K. (1996) MOLMOL: a program for display and analysis of macromolecular structures. J. Mol. Graph. 14, 51–55, 29–32. [20] Roberts, G.C.K. (1993) NMR of Macromolecules: A Practical Approach, IRL Press at Oxford University Press, Oxford, NY. [21] Boraston, A.B., Bolam, D.N., Gilbert, H.J. and Davies, G.J. (2004) Carbohydrate-binding modules: fine-tuning polysaccharide recognition. Biochem. J. 382, 769–781. [22] Boraston, A.B., Healey, M., Klassen, J., Ficko-Blean, E., Lammerts van Bueren, A. and Law, V. (2006) A structural and functional analysis of alpha-glucan recognition by family 25 and 26 carbohydrate-binding modules reveals a conserved mode of starch recognition. J. Biol. Chem. 281, 587–598. [23] Giardina, T., Gunning, A.P., Juge, N., Faulds, C.B., Furniss, C.S., Svensson, B., Morris, V.J. and Williamson, G. (2001) Both binding sites of the starch-binding domain of Aspergillus niger glucoamylase are essential for inducing a conformational change in amylose. J. Mol. Biol. 313, 1149–1159. [24] Liu, Y.N., Lai, Y.T., Chou, W.I., Chang, M.D. and Lyu, P.C. (2007) Solution structure of family 21 carbohydrate-binding module from Rhizopus oryzae glucoamylase. Biochem. J. 403, 21–30. [25] Liu, W., Madsen, N.B., Fan, B., Zucker, K.A., Glew, R.H. and Fry, D.E. (1995) Effects of oligosaccharide binding on glycogen debranching enzyme activity and conformation. Biochemistry 34, 7056–7061.
Oligosaccharide recognition and binding to the carbohydrate binding module of AMP-activated protein kinase Ann Koay, Kieran A. Rimmer, Haydyn D.T. Mertens, Paul R. Gooley*, David Stapleton* Department of Biochemistry and Molecular Biology, Bio21 Molecular Science and Biotechnology Institute, The University of Melbourne, Parkville, Victoria, Australia Received 1 August 2007; revised 4 September 2007; accepted 20 September 2007 Available online 29 September 2007 Edited by Christian Griesinger
Abstract The AMP-activated protein kinase (AMPK) contains a carbohydrate-binding module (b1-CBM) that is conserved from yeast to mammals. b1-CBM has been shown to localize AMPK to glycogen in intact cells and in vitro. Here we use Nuclear Magnetic Resonance spectroscopy to investigate oligosaccharide binding to 15N labelled b1-CBM. We find that b1-CBM shows greatest affinity to carbohydrates of greater than five glucose units joined via a,1 fi 4 glycosidic linkages with a single, but not multiple, glucose units in an a,1 fi 6 branch. The near identical chemical shift profile for all oligosaccharides whether cyclic or linear suggest a similar binding conformation and confirms the presence of a single carbohydrate-binding site. 2007 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. Keywords: AMP-activated protein kinase; Carbohydratebinding module; NMR; Oligosaccharide; Glycogen
1. Introduction The AMP-activated protein kinase (AMPK) is an evolutionarily conserved energy sensor found in mammals and known to coordinate cellular metabolism and systemic energy balance in response to energy demands and a range of stimuli [1]. AMPK is a highly conserved heterotrimeric complex comprising a catalytic a subunit and regulatory b and c subunits that have been identified in plants, yeast, nematodes, flies and mammals. AMPKb1 subunit was recently shown to contain a glycogenbinding domain. Using secondary structure predictions a protease-resistant domain within residues 68–163 of AMPKb1 was identified that associated with glycogen in cell-free assays [2]. Furthermore fluorescently tagged AMPK heterotrimers associated with glycogen-containing granules in the cytoplasm of cultured human cells that was abolished when glycogenbinding domain was deleted [3]. The glycogen-binding domain has been reclassified as a member of the carbohydrate-binding
module-containing family 48 (CBM48) (http://afmb.cnrs-mrs. fr/CAZY) based on amino acid sequence, alongside other carbohydrate-binding proteins such as isoamylase and glycogen branching enzymes. To date, the structures of both rat b1CBM, in complex with the unnatural cyclic sugar b-cyclodextrin [4], and human apo b2-CBM (unpublished, PDB ident code 2F15) have been elucidated by X-ray crystallography. AMPKb1-CBM folds into a b sandwich consisting of two anti-parallel b sheets similar to structures of N-isoamylase or starch-binding domains [4]. Glycogen, a branched polymer of glucose, is a cellular store of energy and is important for whole body glucose metabolism with largest stores found in liver and skeletal muscle [5]. Glycogen is stored in cells in the form of granules that not only contain glucose, but also bound proteins that are involved in glycogen metabolism like glycogen synthase and glycogen phosphorylase [6]. Several studies link AMPK to glycogen metabolism. For example, AMPK co-immunoprecipitates with glycogen synthase [7], glycogen phosphorylase [7] and glycogen debranching enzyme [8] and mutations in the AMPK c subunit result in a novel cardiac glycogen storage disease in human individuals [9,10]. To further investigate the carbohydratebinding nature of b1-CBM we used Nuclear Magnetic Resonance (NMR) spectroscopy to identify the residues important for oligosaccharide binding to b1-CBM and to determine affinity for this association. We find that b1-CBM preferentially binds sugars with 6–7 glucose units and ideally with one type of branched sugar.
2. Materials and methods 2.1. Expression and purification of isotopically labeled b1-CBM AMPK b1-CBM was cloned into pProEX HT and expressed as a His-tag fusion protein in BL21 Escherichia coli cells [11]. AMPK b1CBM was 13C and 15N or 15N only labelled by growing cultures in a 2 L Braun Biostat Fermentor containing 1 L of minimal media with 15 NH4Cl and 13C-D -glucose (Spectral Gases Laboratory) as the sole nitrogen and carbon sources [12]. b1-CBM expression was induced by the addition of 1 mM IPTG until the nutrients were depleted after which cells were harvested, pelleted and purified as previously described [11].
*
Corresponding authors. E-mail addresses: [email protected] (P.R. Gooley), dis@unimelb. edu.au (D. Stapleton). Abbreviations: AMPK, AMP-activated protein kinase; CBM, carbohydrate-binding module; b1-CBM, AMPK b1 carbohydrate-binding module; NMR, nuclear magnetic resonance spectroscopy; HSQC, heteronuclear single quantum correlated
2.2. Nuclear magnetic resonance (NMR) experiments and data processing Purified 13C and 15N or 15N only labelled b1-CBM were dissolved into NMR buffer (25 mM Na2HPO4–NaH2PO4, 0.02% sodium azide, pH 7.5, 10% D2O) and placed into NMR sample tubes. For resonance assignment NMR spectra were acquired on 0.5 mM b1-CBM samples at 25 C using Varian 600 MHz INOVA or Bruker 800 MHz
0014-5793/$32.00 2007 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2007.09.044
5056
A. Koay et al. / FEBS Letters 581 (2007) 5055–5059
AVANCE spectrometers. All experiments utilized cryogenically cooled probes of the spectrometers to maximize sensitivity. Heteronuclear 2D and 3D resonance spectra were acquired and the data was processed with the program NMR Pipe [13] using a Gaussian function in the direct HN dimension and cosine squared functions in the indirect dimensions. All data were zero-filled once prior to Fourier transformation. Data was analysed using NMR View, v5.2.2 [14]. For sequence specific assignment of resonances 3D HNCACB [15] and CBCA(CO)NH [16] experiments were acquired and analysed, 3D C(CO)NH [17] experiments were performed to confirm assignments. 2.3. Oligosaccharide titrations To study carbohydrate recognition and binding to b1-CBM, individual titrations using 12 different oligosaccharides shown in Table 1 were obtained from Supelco (USA) and WAKO (Japan) and performed with 0.2 mM 15N b1-CBM in NMR buffer. Each oligosaccharide was prepared as a stock concentration ranging from 50 to 400 mM, except for b-cyclodextrin which was prepared as a 20 mM stock due to low solubility. Eight to 10 oligosaccharide titration points with increasing concentration from 0 to 10 mM were performed. For each titration point, a 1H–15N heteronuclear single quantum correlation (HSQC) spectrum was acquired at 800 MHz and 25 C. The final sample dilution was less than 10% of the original volume. 2.4. Dissociation constant determination Dissociation constants for oligosaccharides binding to b1-CBM were calculated by fitting the titration data for individual resonances, assuming a single ligand recognition site, in the programs xcrvfit, (v3.0.6, Robert Boyko and Brian Sykes) and xmgrace (v 5.1.2 Paul J. Turner). Maximum HN and 15N chemical shift differences were scaled according to the following equation [18]: DdTotal = ((DdHN)2 + (DdN · 0.154)2)1/2 where DdHN and DdN are the chemical shift differences for the HN and 15N resonances between the free and titrated protein. Molecular visualization and generation of figures were performed using MOLMOL v2k.2 [19].
Table 1 Oligosaccharide structures and dissociation constants (Kd) for the binding of soluble oligosaccharides to b1-CBM of AMPK at 25 C and pH 7.5 Ligand
Structure
Kd (mM)
Maltoheptaose Maltohexaose Maltopentaose Maltotetraose Maltotriose Maltose Glucose 6-phosphate
GGGGGGG GGGGGG GGGGG GGGG GGG GG PG GGG G G GG
0.36 ± 0.033 0.67 ± 0.057 3.69 ± 0.34 4.21 ± 0.12 9 to 18 N.D. N.D.
b-Cyclodextrin
0.33 ± 0.025
G
Glucosyl b-cyclodextrin
GGG G G GG GG
0.15 ± 0.016
Maltosyl b-cyclodextrin
GGG G G GG
0.51 ± 0.04
Glucosyl maltotriose Isomaltose
G GGG G
3. Results 3.1. Assignment of the 1H–15N HSQC NMR spectrum of b1CBM The 1H–15N HSQC spectrum obtained for apo b1-CBM showed a pattern of well-dispersed cross peaks (Fig. 1S, Supplementary data). A total of 13 HN correlations (M53, D54, N70, E71, K72 T85, G86, K102, Q109, N110, Q145, L146 and E161) out of an expected 95 could not be unambiguously assigned, or were not observed. 3.2. NMR titrations of oligosaccharides with b1-CBM For each oligosaccharide shown in Table 1, 8–10 1H–15N HSQC spectra were collected after progressive addition of oligosaccharide and overlayed to track peak chemical shift perturbations. Each of the titrated oligosaccharides except isomaltose and glucose-6-phosphate were found to cause significant chemical shift perturbations to the b1-CBM spectrum. An example of the overlayed spectra for maltoheptaose is shown in Fig. 1A. Saturation of the protein was possible for the linear maltodextrins: maltoheptaose and maltohexaose, and the cyclodextrins: b-cyclodextrin, glucosyl b-cyclodextrin and maltosyl b-cyclodextrin. Under the conditions of this study we could not saturate the protein sample with maltopentaose, maltotetraose, maltotriose and glucosyl-maltotriose. However, all oligosaccharides that induced chemical shift changes in b1-CBM showed similar patterns of change upon ligand binding suggesting a single carbohydrate-binding site (Figs. 1B and C, 2 and 2S, Supplementary data). A large number of resonances exhibited no chemical shift perturbation upon ligand binding. In titrations with b-cyclodextrin, glucosyl b-cyclodextrin, maltosyl b-cyclodextrin, maltoheptaose, maltohexaose and maltopentaose, a number of cross peaks were rapidly broadened upon the progressive addition of ligand until they completely vanish from the spectra, only to reappear as the oligosaccharide concentration increased. Dissociation constants (Kd) were independently determined for individual resonances that showed significant chemical shift deviation in either or both the 1H and 15N dimension upon oligosaccharide binding. Peaks that exhibited peak broadening were excluded from Kd calculations [20]. The obtained Kd values were averaged to obtain an overall Kd value for each titrated oligosaccharide as reported in Table 1. Kd values for maltotriose and glucosyl-maltotriose could not be accurately determined because titrations with these ligands did not result in sufficient saturation of the b1-CBM even at very high concentrations of oligosaccharide, suggesting weak binding to b1-CBM. For these oligosaccharides a range of Kd values were determined (Table 1). We find b1-CBM binds with higher affinity to long chain a,1 fi 4 linked oligosaccharides and this affinity diminishes with maltodextrin length (Table 1). A distinct decrease in the values of Kd is found between maltohexaose and maltopentaose where the affinity of b1-CBM for maltopentaose is approximately five times weaker than maltohexaose.
8 to 14 N.D.
4. Discussion
G
Glucose units (G) connected via an a, 1 fi 6 glycosidic bond are indicated by a solid line. All other glucose units are linked via a,1 fi 4 glycosidic bonds. pG represents glucose 6-phosphate.
In this present study, we have identified key determinants in the association of oligosaccharides with the AMPK b1 carbohydrate-binding module (b1-CBM). The oligosaccharide Kd
A. Koay et al. / FEBS Letters 581 (2007) 5055–5059
5057
Fig. 1. Oligosaccharide-induced b1-CBM chemical shift changes for individual residues. (A) Overlayed 2D 1H–15N HSQC spectra focusing on an area of chemical shift changes in the nuclei of b1-CBM from AMPK upon titration with 0–10 mM maltoheptaose at 25 C and pH 7.5. Broken arrows indicate observed peak broadening that is associated with intermediate exchange. For clarity, only residues that showed chemical shift perturbations are labelled. Mapping of chemical shift perturbations for b-cyclodextrin (B) and maltoheptaose (C) onto the structure of b1-CBM [4]. Chemical shift differences were scaled as described in methods. The width (thin to thick) and colour (yellow to red) of the backbone is indicative of the extent of perturbation. b-cyclodextrin is shown in green.
maximum average chemical shift difference (ppm)
0.8 0.6 0.4 0.2
160
145
residue number
130
115
100
glucose-6-phosphate maltose maltotriose glucosyl maltotriose maltotetraose maltopentaose bcd mbcd gbcd maltohexaose maltoheptaose
85
70
0
Fig. 2. Comparison of the maximum average chemical shift differences observed in b1-CBM upon titration with respective oligosaccharides show that various types of oligosaccharide bind and cause chemical shift changes to HN and 15N resonances of similar amino acid residues in b1-CBM. Single plots of maximum average chemical shift differences for each oligosaccharide are shown in Supplementary Fig. 2S.
values in Table 1 clearly show that b1-CBM favors associating with maltodextrins with six to seven a,1 fi 4 glucose units and
in particular maltodextrins that include an additional glucosyl unit in an a,1 fi 6 branch but not two a,1 fi 4 glucosyl units
5058
joined in an a,1 fi 6 branch. We find that b1-CBM binds glucosyl b-cyclodextrin with the highest affinity (Kd = 150 ± 16 lM) > b-cyclodextrin > maltoheptaose > maltosyl b-cyclodextrin > maltohexaose. Smaller oligosaccharides including maltopentaose, previously shown to prevent association of AMPK enzyme but not b1-CBM alone to glycogen [4], or branched oligosaccharides exhibited diminished affinities. These data suggest that AMPK b1-CBM is a Type B CBM; exhibiting an increased binding affinity towards oligosaccharides with higher degrees of polymerization, as opposed to Type A CBMs that bind to crystalline cellulose and chitin or Type C CBMs that bind smaller mono-, di- and trisaccharides [21]. Other structurally characterized Type B CBMs which are from plant and bacteria and include representatives of the CBM20, 21, 25 and 26 clans, show Kd values of 1–20 lM for oligosaccharides including b-cyclodextrin and maltoheptaose [22–24]. The 10–100-fold poorer Kd values observed here for this mammalian CBM from the CBM48 clan may indicate functional differences between these clans or that the exact substrate for this protein remains to be defined. To determine whether b1-CBM has a preference for branched oligosaccharides we obtained as many a,1 fi 6 branched molecules as commercially possible and found that glucosyl b-cyclodextrin had the highest affinity for b1-CBM. However, increased chain length on either the a,1 fi 4 or a,1 fi 6 chains like for maltosyl b-cyclodextrin, glucosyl maltotriose (compared to maltotetraose) and isomaltose resulted in reduced affinity for b1-CBM (Table 1). The latter is interesting as titrations with the disaccharide maltose, an a,1 fi 4 linkage, weakly caused the same pattern of chemical shift perturbations similar to maltoheptaose whilst isomaltose, an a,1 fi 6 linkage, did not cause any chemical shift change perturbation in b1CBM. These data suggest that AMPK b1-CBM will not associate with a,1 fi 6 branch points alone but prefer longer a,1 fi 4 chains that contain a single a,1 fi 6 glucose unit. This type of oligosaccharide would be visible to b1-CBM only during degradation of the glycogen molecule. Glycogen debranching enzyme has two catalytic activities that occur in succession; first, a transferase activity that transfers a maltotriose unit to the ‘‘main chain’’ from four glucose units attached to this main chain by a,1 fi 6 linkage. This results in an elongated section of a,1 fi 4-linked polymer, leaving a single glucose attached to it via an a,1 fi 6 linkage. Second, a glucosidase activity then cleaves the a,1 fi 6 linkage by which this glucose is attached [25]. As the regulation or rate of these two activities is unknown, a possibility exists that the halfway point of glycogen debranching enzyme’s dual-step action may provide a high affinity binding site for AMPK b1-CBM. Further intriguing is the recently reported association between AMPK b1-CBM and the glycogen debranching enzyme [8]. The region important for association with the glycogen debranching enzyme encompasses the first half of b1-CBM (b1 65-123) although association with the actual carbohydrate-binding site described here appears unlikely as mutation of residues of the carbohydrate-binding site such as W100G and K126Q failed to prevent association between the two proteins [8]. The crystal structure of b1-CBM in complex with b-cyclodextrin identifies five residues critical for carbohydrate association; W100, W133, K126, L146 and N150 [4]. Results in this study show that, except for W100, the HN of these residues or residues sequentially near exhibit significant chemical shift perturbations of more than one standard deviation from the mean
A. Koay et al. / FEBS Letters 581 (2007) 5055–5059
chemical shift change. Residue L146 which pierces the b-cyclodextrin ring in the b1-CBM structure [4], could not be assigned in the HSQC spectrum, however its neighbouring residue S144 is immediately broadened by maltoheptaose and is significantly shifted by all other binding oligosaccharides. The peptide HN of W100 and residues sequential to it show modest chemical shift changes (Fig. 2). The peptide HN of G95 exhibits a chemical shift perturbation of more than one standard deviation from the mean, and is significantly affected by all binding oligosaccharides. In the b1-CBM structure this residue is found in the b2 strand buried below the aromatic side-chain of W100, but otherwise is remote from the carbohydrate-binding site. This residue is most likely affected by ring current shifts by the reorientation of the aromatic ring of W100 upon oligosaccharide binding. All oligosaccharides that bound to b1-CBM, regardless of being linear or cyclised, perturbed the same residues with a similar profile. As these perturbations were limited to the residues observed in the crystal structure of the b-cyclodextrin complex, we conclude that there is a single binding site and that a linear sugar such as maltoheptaose binds in a similar conformation to the cyclised dextrans [4]. We also conclude that to produce the same profile, small sugars such as maltose and maltotriose bind in multiple positions but into the subsites defined in the crystal structure, and therefore there is no preferred subsite. As the crystal structure shows a sugar of five glucosyl units would be sufficient to make contacts to all carbohydrate-binding subsites in the b1-CBM, we expected that sugars such as maltopentaose or longer would have similar Kd values. While it is surprising that maltoheptaose has the best Kd amongst all titrated linear sugars, similar observations for other CBMs have been made where oligosaccharides that are apparently much longer than necessary bind with higher affinities [22]. Therefore other influences such as structural differences of maltoheptaose versus maltopentaose or perhaps displacement of bound water may be important for determining binding affinity. In conclusion, this study reveals that the single carbohydrate-binding site of AMPK b1-CBM binds well to oligosaccharides longer than five glucose units and poorly to tri-, di, and monosaccharides. The binding face of b1-CBM recognizes carbohydrates joined via a,1 fi 4 glycosidic linkages with a single glucose unit in an a,1 fi 6 branch. We expect these results to form the basis of further exploring the ligand recognition and binding of the CBM48 clan. Acknowledgement: We wish to acknowledge NHMRC (National Health and Medical Research Council) of Australia Grant support to D.S.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.febslet.2007.09. 044. References [1] Kahn, B.B., Alquier, T., Carling, D. and Hardie, D.G. (2005) AMP-activated protein kinase: ancient energy gauge provides clues to modern understanding of metabolism. Cell Metabol. 1, 15–25.
A. Koay et al. / FEBS Letters 581 (2007) 5055–5059 [2] Polekhina, G. et al. (2003) AMPK beta subunit targets metabolic stress sensing to glycogen. Curr. Biol. 13, 867–871. [3] Hudson, E.R. et al. (2003) A novel domain in AMP-activated protein kinase causes glycogen storage bodies similar to those seen in hereditary cardiac arrhythmias. Curr. Biol. 13, 861–866. [4] Polekhina, G., Gupta, A., van Denderen, B.J., Feil, S.C., Kemp, B.E., Stapleton, D. and Parker, M.W. (2005) Structural basis for glycogen recognition by AMP-activated protein kinase. Structure (Camb.) 13, 1453–1462. [5] Roach, P.J. (2002) Glycogen and its metabolism. Curr. Mol. Med. 2, 101–120. [6] Shearer, J. and Graham, T.E. (2002) New perspectives on the storage and organization of muscle glycogen. Can. J. Appl. Physiol. 27, 179–203. [7] Chen, Z. et al. (1999) Expression of the AMP-activated protein kinase beta1 and beta2 subunits in skeletal muscle. FEBS Lett. 460, 343–348. [8] Sakoda, H. et al. (2005) Glycogen debranching enzyme association with beta-subunit regulates AMP-activated protein kinase activity. Am. J. Physiol. Endocrinol. Metab. 289, E474–E481. [9] Dyck, J.R. and Lopaschuk, G.D. (2006) AMPK alterations in cardiac physiology and pathology: enemy or ally? J. Physiol. 574, 95–112. [10] Gollob, M.H., Green, M.S., Tang, A.S. and Roberts, R. (2002) PRKAG2 cardiac syndrome: familial ventricular preexcitation, conduction system disease, and cardiac hypertrophy. Curr. Opin. Cardiol. 17, 229–234. [11] Polekhina, G., Feil, S.C., Gupta, A., O’Donnell, P., Stapleton, D. and Parker, M.W. (2005) Crystallization of the glycogen-binding domain of the AMP-activated protein kinase subunit and preliminary X-ray analysis. Acta Crystallogr., Sect. F F61, 39–42. [12] Cai, M., Huang, Y., Sakaguchi, K., Clore, G.M., Gronenborn, A.M. and Craigie, R. (1998) An efficient and cost-effective isotope labeling protocol for proteins expressed in Escherichia coli. J. Biomol. NMR 11, 97–102. [13] Delaglio, F., Grzesiek, S., Vuister, G.W., Zhu, G., Pfeifer, J. and Bax, A. (1995) NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293. [14] Johnson, B.A. and Blevins, R.A. (1994) NMR view: a computer program for the visualization and analysis of NMR data. J. Biomol. NMR 4, 603–614.
5059 [15] Wittekind, M. and Mueller, L. (1993) HNCACB, a Highsensitivity 3D NMR experiment to correlate amide-proton and nitrogen resonances with the alpha- and beta-carbon resonances in proteins. J. Magn. Reson. 101, 201–205. [16] Grzesiek, S. and Bax, A. (1992) Correlating backbone amide and side chain resonances in larger proteins by multiple relayed triple resonance NMR. J. Am. Chem. Soc. 114, 6291–6293. [17] Grzesiek, S., Anglister, J. and Bax, A. (1993) Correlation of backbone amide and aliphatic side-chain resonances in C-13/N15-enriched proteins by isotropic mixing of C-13 magnetization. J. Magn. Reson. 101, 114–117. [18] Ayed, A., Mulder, F.A., Yi, G.S., Lu, Y., Kay, L.E. and Arrowsmith, C.H. (2001) Latent and active p53 are identical in conformation. Nat. Struct. Biol. 8, 756–760. [19] Koradi, R., Billeter, M. and Wuthrich, K. (1996) MOLMOL: a program for display and analysis of macromolecular structures. J. Mol. Graph. 14, 51–55, 29–32. [20] Roberts, G.C.K. (1993) NMR of Macromolecules: A Practical Approach, IRL Press at Oxford University Press, Oxford, NY. [21] Boraston, A.B., Bolam, D.N., Gilbert, H.J. and Davies, G.J. (2004) Carbohydrate-binding modules: fine-tuning polysaccharide recognition. Biochem. J. 382, 769–781. [22] Boraston, A.B., Healey, M., Klassen, J., Ficko-Blean, E., Lammerts van Bueren, A. and Law, V. (2006) A structural and functional analysis of alpha-glucan recognition by family 25 and 26 carbohydrate-binding modules reveals a conserved mode of starch recognition. J. Biol. Chem. 281, 587–598. [23] Giardina, T., Gunning, A.P., Juge, N., Faulds, C.B., Furniss, C.S., Svensson, B., Morris, V.J. and Williamson, G. (2001) Both binding sites of the starch-binding domain of Aspergillus niger glucoamylase are essential for inducing a conformational change in amylose. J. Mol. Biol. 313, 1149–1159. [24] Liu, Y.N., Lai, Y.T., Chou, W.I., Chang, M.D. and Lyu, P.C. (2007) Solution structure of family 21 carbohydrate-binding module from Rhizopus oryzae glucoamylase. Biochem. J. 403, 21–30. [25] Liu, W., Madsen, N.B., Fan, B., Zucker, K.A., Glew, R.H. and Fry, D.E. (1995) Effects of oligosaccharide binding on glycogen debranching enzyme activity and conformation. Biochemistry 34, 7056–7061.