MAGNETIC RESONANCE IN CHEMISTRYMagn. Reson. Chem. 2007; 45: 745–748Published online 17 July 2007 in Wiley InterScience(www.interscience.wiley.com) DOI: 10.1002/mrc.2041

Temperature dependence of ligand–protein complexformation as reflected by saturation transfer differenceNMR experiments

Patrick Groves,1,2∗ Katalin E. Kover,3 Sabine Andre,4 Joanna Bandorowicz-Pikula,5

Gyula Batta,6 Marta Bruix,7 Rene Buchet,8 Angeles Canales,1 F. Javier Canada,1

Hans-Joachim Gabius,4 Douglas V. Laurents,7 Jose R. Naranjo,9 Małgorzata Palczewska,9

Slawomir Pikula,5 Eduardo Rial,1 Agnieszka Strzelecka-Kiliszek5 andJesus Jimenez-Barbero1∗

1 Department of Protein Science, Centro de Investigaciones Biologicas, CSIC, Ramiro de Maeztu 9, 28040 Madrid, Spain2 School of Biomedical Sciences, University of Ulster, Cromore Road, Coleraine BT52 1SA, Northern Ireland, UK3 Department of Inorganic and Analytical Chemistry, University of Debrecen, Egyetem ter 1, H-4010 Debrecen, Hungary4 Institut fur Physiologische Chemie, Tierarztliche Fakultat, Ludwig-Maximilians-Universitat Munchen, Veterinarstr. 13, D-80539 Munchen, Germany5 Department of Biochemistry, Nencki Institute of Experimental Biology, Polish Academy of Sciences, 3 Pasteur Street, 02-093 Warsaw, Poland6 Department of Biochemistry, University of Debrecen, Egyetem ter 1, H-4010 Debrecen, Hungary7 Departamento de Espectroscopıa y Estructura Molecular, Instituto de Quımica-Fısica Rocasolano, CSIC, Serrano 119, 28006 Madrid, Spain8 University of Lyon I, UMN CNRS 5013, 6 rue Victor Grignard, 69622 Villeurbanne Cedex, France9 Department of Molecular and Cellular Biology, Centro Nacional de Biotecnologıa, CSIC, 28049 Madrid, Spain

Received 18 February 2007; Revised 1 May 2007; Accepted 16 May 2007

We show that temperature is an important parameter for the sensitivity of saturation transfer difference(STD) spectroscopy. A decreased intensity of STD signals is observed for lactose binding to growth-regulatory galectin7 (p53-induced gene 1), as well as for nucleotide binding to annexin A6, when thetemperature is increased from 281 to 298–310 K. Opposite temperature effects on STD intensity areobserved for S-peptide binding to S-protein to reconstitute RNase S. However, the STD signals fortryptophan binding to downstream regulatory element antagonist modulator of the human prodynorphingene (DREAM)are relatively unaffected between 281 and 298 K. The known kinetics of the binding of ATPby the uncoupling protein from brown adipose tissue mitochondria (UCP1) predicted an observable STDat 310 K, but rapid sample degradation limits the experiments to much lower temperatures. Temperaturestrongly influences the kinetics and affinity constant of various types of complex formation and in sodoing influences the observed STD effects. Therefore, temperature can be exploited to facilitate theoptimization of STD-based applications, and at the same time minimize the number of test samples.STD-based screening protocols to detect new target-specific compounds may yield a larger number ofpotential ligands if screened at various temperatures. Copyright 2007 John Wiley & Sons, Ltd.

KEYWORDS: NMR; 1H; 13C; 15N; saturation transfer difference; drug discovery; annexin; galectin; RNase S

INTRODUCTION

Saturation transfer difference (STD) spectroscopy has beenused in the past few years to assess ligand binding tobiomolecular receptors. Moreover, in certain cases, it mayalso be used to reveal which parts of a ligand contact areceptor.1,2 Under standard conditions, samples for STDexperiments consist of 10–100 µM concentrations of receptorand 1–5 mM concentrations of the ligand(s). The receptor,generally a protein, is selectively excited in a spectral

ŁCorrespondence to: Patrick Groves, School of Biomedical Sciences,University of Ulster, Cromore Road, Coleraine BT52 1SA, NorthernIreland, UK. E-mail: p.groves@ulster.ac.ukJesus Jimenez-Barbero, Department of Protein Science, Centro deInvestigaciones Biologicas, CSIC, Ramiro de Maeztu 9, 28040Madrid, Spain. E-mail: jjbarbero@cib.csic.es

region that is devoid of ligand peaks. Then, a saturationtime over 0.25–5.0 s follows, during which magnetizationis transferred from the receptor protons to the adjacentprotons in the bound ligand.2 It should be noted that thekey information comes from the adjacent ligand protons andthat the degree of magnetization transfer to these ligandresonances is somehow proportional to the intimacy of thecontact. As STD signals are observed only for moleculesthat bind to the protein receptor, STD may also be appliedto ligand mixtures and used to screen molecular librariesagainst proteins to discover new ligands for drug design.1

Several parameters have been described as critical for thesuccess of STD experiments. The ligand off-rate is extremelyimportant for the overall sensitivity of the experiment.3 SinceSTD is a transfer-type experiment, the information gained

Copyright 2007 John Wiley & Sons, Ltd.

746 P. Groves et al.

for the bound ligand needs to be efficiently transferred to thefree one, which is in high excess. If the off-rate is too slow(i.e. below 1 s�1), then the ligand cannot efficiently transferinformation between the bound and the free forms of theligand during the saturation time. For on-rates controlledby diffusion (upto 1010 s�1), a fast off-rate (i.e. above 105 s�1)suggests a ligand affinity of around 103 M�1. In this case,the ligand-binding site of the protein will be fully occupiedonly if a large excess of the ligand is used (typically at a100 : 1 ratio). Additionally, the short contact times for weaklybound ligands might be insufficient to promote an efficientmagnetization transfer to the ligand. The ligand to receptormolar ratio is also very important since the experimentbenefits from the existence of an accumulative effect onthe signal intensity.2,3 Provided that the relaxation rate of theligand protons is slow enough, the STD signal is more intensefor high ligand/protein molar ratios since the probability thatan already saturated ligand entering again in the receptor-binding site is rather small. Thus, every ligand moleculeentering in the receptor site provides additional intensity tothe STD signal. While much of this is true for small ligands,larger ligands can have non-optimal kinetic and relaxationproperties. In some instances, the experiment might notprovide good, intense STD data.

Kinetics and thermodynamic parameters indeed affectthe experiment to a great degree. The effect of tempera-ture on several aspects (transfer efficiency, ligand relaxationproperties) of the STD experiment has been presented fortrimethoprim binding to dihydrofolate reductase.4 Here, wepresent the effect of temperature on the STD of severalprotein–ligand systems, which cannot always be readilypredicted. Temperature provides an additional parameterthat can dramatically affect STD results and the six parame-ters considered by Jayalakshmi and Krishna to calculate STDeffects: (i) the saturation time, (ii) the location of the saturatedreceptor protons with respect to the ligand protons, (iii) theconformation of the ligand–receptor interface, (iv) the rota-tional correlation times for the molecular species, (v) thekinetics of the reversibly forming complex, and (vi) the lig-and/receptor ratio.3 Four of these parameters, namely, i, iii,iv, and v, are expected to depend on temperature.

RESULTS AND DISCUSSION

We examined the effect of temperature on the acquiredSTD data of five unrelated protein–ligand complexes,Table 1. The lactose-binding properties of growth-regulatory

galectin7 (p53-induced gene 1),5,6 as well as the specificity ofnucleotides for human annexin A6,7 and the uncoupling pro-tein 1 (UCP1) from brown adipose tissue mitochondria,8 arewell defined. In contrast, the characterization of downstreamregulatory element antagonist modulator of the human pro-dynorphin gene (DREAM) binding to tryptophan-containingpeptides is an example of a work in progress in which fewbinding or kinetic parameters are known. The reconstitutionof RNase S, from the two fragments S-protein and S-peptideresulting from subtilisin cleavage of RNase A, has also beenextensively studied.9 The M13A mutant S-peptide was cho-sen, as the native S-peptide has a high affinity for S-proteinand slow off-rate that were unlikely to be compatible withthe STD experiment.10 The experimental data resulted fromthe application of STD experiments to these complexes.1,11

The intensity of the STD signals increases as a function oflower temperatures for both the annexin A6/nucleotide andgalectin7/lactose complexes, Fig. 1(A) and (B). STD signalswere not observed for UCP1 in complex with either ATPor ADP at low (281 K) or high (310 K) temperatures (datanot shown). We found that temperature had little effect onthe STD signals for tryptophan to DREAM, Fig. 1(C). Highertemperatures favored STD for the binding of S-peptideM13Ato S-protein, Fig. 1(D).

A peak maximum of about 0.7 ppm, corresponding to theprotein methyl groups, was used for the on-resonance satu-ration of the annexin A6/nucleotide, galectin7/lactose, andUCP1/nucleotide complexes and this value was shifted to themethyl peak maximum at each tested temperature. A fixedon-resonance saturation was used for DREAM/tryptophanof 9.2 ppm, a point mid-way between the εHN and aro-matic resonances of tryptophan, to avoid the non-specificsaturation of the ligand. A new 15N-STD experiment thatis initiated through a group-specific excitation of 15N-labeled amides in the protein was used for the 15N-labeledS-protein/S-peptideM13A system, Fig. 1(D).11 We obtainedsimilar intensity data with the standard STD experimentusing an on-resonance irradiation at 0.3 ppm/298 K and0.18 ppm/304 K. We used different frequencies in the con-ventional STD experiments to ensure that the observedchanges in STD effects were not due to changes in satu-ration efficiencies attributable to the temperature shifts ofthe resonances, and also the processes were reversible afterreturning the temperature values to their previous settings.

In STD-based projects, the available ligand-binding datais sparse or limited to affinity data. A full description of

Table 1. Summary of binding data and STD effects

Protein Ligand Temperature STD effect K (M�1) koff �s�1�

Galectin7 Lactose 298 D >281 Increased signals 2.2 ð 103 at 300 K ?4.9 ð 103 at 287 K12

AnnexinA6 Mixed nucleotides 310 D >281 Increased signals ¾10�6 at 298 K7 ?DREAM Tryptophan 298 D >281 No change ? ?S-protein S-peptide M13A 310 D >298 Decreased signals ¾10�6 at 298 K13 1.5 at 298 K

15 at 310 K13

UCP1 ATP 310 D >298 Protein degradation ¾10�6 at 298 K8 10�3 at 277 K 10�2 at 287 K8

DHFR Trimethoprim 303 D >277 Increased signals4 2 ð 1074 ?

Copyright 2007 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2007; 45: 745–748DOI: 10.1002/mrc

Temperature influences on saturation transfer difference experiments 747

kinetic effects (on- and off-rates) is rarely available. Theaffinities of nucleotides for annexin A6, ATP for UCP1, andS-peptideM13A for S-protein are strong, whereas galectin7binds lactose more weakly, Table 1. The affinity of trypto-phan for DREAM is not yet known. The ligand concentra-tions are sufficiently high to maintain each of the proteinsin >90% complexed state (with the possible exception ofDREAM/tryptophan). Kinetic data are available for theUCP1/ATP and S-protein/S-peptideM13A complexes.8,10

The very slow off-rates for ATP binding to UCP1 give aclear indication why STD data are not expected for theUCP1/ATP complex at low temperatures, Table 1.8 Yet thetrend may suggest a value of approximately 1 s�1 at 310 K(as 10 K increases the off-rate by ca a factor of 10) and, there-fore, a chance of observing STD effects. Unfortunately, UCP1is unstable at higher temperatures,14 illustrating that raisedtemperatures cannot always provide better STD data, andwe were not able to observe any STD effects for the bindingof ATP to this protein. The kinetic data for S-protein/S-peptideM13A were measured using stopped-flow kineticsand a different buffer system, Table 1.13 However, this datais consistent with our observations that STD data can beobtained at elevated temperatures.

The binding affinities of annexin A6 and UCP1 fornucleotides are similar, but the binding kinetics for the twoproteins are clearly distinct. The obtained data suggest thatthe annexin A6/nucleotide and galectin7/lactose complexeshave relatively high off-rates and that the use of a lowertemperature increases the proportion of bound protein, theligand contact times, and, consequently, the overall efficiencyof the STD experiment. The findings for DREAM/tryptophansuggest that there can be cases balanced between these twoextremes and that temperature may not strongly affect theSTD efficiency.

Temperature affects the free energy of binding, thekinetics of the binding process, the dynamics of key sidechains at the ligand-binding site, and, thus, the optimalprotein : ligand ratio to obtain the most intense STD data.With so many unknowns, it is difficult to make the perfectsample and obtain the perfect conditions with the first trial.Our data has consequences for ligand screening, particularlyfor targets that score poorly. An increased number of hits

Figure 1. STD effects affected by temperature. (A) STD with2 s on-resonance saturation at 0.7 ppm applied to 133 µM

galectin7 and 2.4 mM lactose at: (i) 298 K and (ii) 281 K.(B) STD with 2 s saturation on resonance at 0.7 ppm applied to50 µM annexin A6 and 1 mM mixture of 13 different nucleotidesat: (i) 310 K, with only broad background protein peaksshowing, and (ii) 281 K. (C) STD applied to 67 µM DREAM and7.8 mM tryptophan at: (i) 298 K and (ii) 281 K. (D) STD with 2 sgroup selective presaturation of amide 15NH’s at 8/118 ppmapplied to 51 µM S-protein and 2.70 mM S-peptideM13A with50 µM 2,2-dimethyl-2-silapentane-5-sulfonate sodium salt(DSS) as an internal chemical shift reference, 95% H2O/5%D2O, and 10 mM sodium acetate-d3/acetic acid-d3 buffer at pH5.0. at: (i) 310 K (5178 scans) and (ii) 298 K (6800 scans).could be obtained by carrying out parallel runs at high

Copyright 2007 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2007; 45: 745–748DOI: 10.1002/mrc

748 P. Groves et al.

and low temperatures rather than at a single intermediatetemperature. The optimization of temperature may alsobenefit other transferred NMR experiments such as transfernuclear Overhauser effect spectroscopy (TR-NOESY) anddiffusion ordered spectroscopy (DOSY), although we havenot yet tested this. To conclude, scanning the sampletemperature can lead to STD sample optimization for STDwithout the expense of additional samples.

EXPERIMENTAL

Galectin7,5 annexin A6,7 UCP-1,8 and RNase S13 wereexpressed and purified according to established protocols.His-tagged DREAM (residues 71–256) was expressed andpurified according to the manufacturers’ protocols (Invitro-gen/BD Biosciences) with characterization (DNA sequencingof inserted gene, polyacrylamide gel analysis, Western blot,and mass spectroscopy) carried out in-house. Lactose, trypto-phan, and unlabeled nucleotides (all Sigma) were purchasedand used without further purification. An S-peptide vari-ant, Met13 to Ala, with a weakened affinity for S-protein(sequence: KETAAAKFERQHAES) was synthesized andpurified by HPLC by the CIB peptide synthesis service.The S-protein was produced from RNase A according tothe literature with minor modifications.9 All spectra werecollected on Bruker 500 MHz instruments.1,11

AcknowledgementsWe acknowledge Dr M. Vilanova from Universitat de Girona forthe production of RNase A. The Madrid teams acknowledge theSpanish Ministry of Education and Science for funding (CTQ2006-10874-C02-01 to J.J-B., BFU2005-01855/BMC to M.B., SAF2004-06644to J.R.N., BFU2005-0880 to E.R., and a Ramon and Cajal Fellowship toP.G.). P.G. acknowledges a RCUK Fellowship. M.P. acknowledgesthe support of a FEBS long-term Fellowship. R.B. acknowledges

funding from CNRS and EGIDE (Picasso Project No: 10715SA).H-J.G. acknowledges funding from the Mizutani Foundationfor Glycoscience (Tokyo, Japan). K.E.K and G.B. acknowledgefunding from MTA-CSIC exchange grants 2004HU0004 (with J.C.)and Hungarian National Science Fund OTKA T 042567. S.P.acknowledges funding from the Polish Ministry of Science andHigher Education (N301 049 31/1592) and together with J.J-B. for anexchange program sponsored by the Polish Academy of Sciences andConsejo Superior de Investigaciones Cientificas, grant 2004PL0011,and the NODPERCEPTION EU research training network (MRTN-CT2006-035546).

REFERENCES

1. Mayer M, Meyer B. Angew. Chem. Int. Ed. 1999; 38: 1784.2. Mayer M, Meyer B. J. Am. Chem. Soc. 2001; 123: 6108.3. Jayalakshmi V, Krishna NR. J. Magn. Reson. 2002; 155: 106.4. Yan J, Kline AD, Mo H, Shapiro MJ, Zartler ER. J. Magn. Reson.

2003; 163: 270.5. Ahmad N, Gabius H-J, Kaltner H, Andre S, Kuwabara I, Liu F-T,

Oscarson S, Norberg T, Brewer CF. Can. J. Chem. 2002; 80: 1096.6. Andre S, Kaltner H, Lensch M, Russwurm R, Siebert H-C,

Fallsehr C, Tajkhorschid E, Heck AJR, von Knebel-Doeberitz M,Gabius H-J, Kopitz J. Int. J. Cancer 2005; 114: 46.

7. Bandorowicz-Pikula J, Kirilenko A, van Deursen R, Golczak M,Kuhnel M, Lancelin J-M, Pikula S, Buchet R. Biochemistry 2003;42: 9137.

8. Huang S-G, Lin Q-S, Klingenberg M. J. Biol. Chem. 1998; 273: 859.9. Richards FM, Vithayathil PJ. J. Biol. Chem. 1959; 234: 1459.

10. Goldberg JM, Baldwin RL. Biochemistry 1998; 37: 2556.11. Kover KE, Batta G, Groves P, Canales A, Jimenez-Barbero J.

Edited/filtered DOSY and STD experiments with improvedsuppression of water signal, III. Bienal Del GERMN, Sant Joan,Alicante, Spain, October 15–18, 2006.

12. Dam TK, Gabius H-J, Andre S, Kaltner H, Lensch M, Brewer CF.Biochemistry 2005; 44: 12564.

13. Varadarajan R, Connelly PR, Sturtevant JM, Richards FM.Biochemistry 1992; 31: 1421.

14. Lin CS, Klingenberg M. Biochemistry 1982; 21: 2950.

Copyright 2007 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2007; 45: 745–748DOI: 10.1002/mrc