Anal. Chem. 2004, 76, 1257-1263
Screening for Noncovalent Ligand-Receptor Interactions by Electrospray Ionization Mass Spectrometry-Based Diffusion Measurements Sonya M. Clark and Lars Konermann*
Department of Chemistry, The University of Western Ontario, London, Ontario, N6A 5B7, Canada
The application of a novel method for the identification of low-molecular-weight noncovalent ligands to a macromolecular target is reported. This technique is based on the measurement of analyte diffusion coefficients by electrospray mass spectrometry (ESI-MS) (Clark et al., Rapid Commun. Mass Spectrom. 2002, 16, 14541462). Potential ligands have large diffusion coefficients as long as they are free in solution. Binding to a macromolecular target, however, drastically reduces the diffusional mobility of any ligand species. Mixtures containing six different saccharides [ribose, rhamnose, glucose, maltose, maltotriose, and N,N′,N′′-triacetylchitotriose (NAG3)] were screened for noncovalent binding to lysozyme. Of these six compounds, only NAG3 is known to bind to the protein. In “direct” binding tests, NAG3 shows a significantly reduced diffusion coefficient in the presence of the protein. No changes were observed for any of the other saccharides. In a second set of experiments, the use of a “competition” screening method was explored in which mixtures of candidate saccharides were tested for their ability to displace a reference ligand from the target. The addition of NAG3-containing mixtures significantly increased the diffusion coefficient of the reference ligand NAG4 (N,N′,N′′,N′′′-tetraacetylchitotetrose), whereas mixtures that did not contain NAG3 had no effect. These data clearly indicate the potential of ESI-MS-based diffusion measurements as a novel tool to screen compound libraries for binding to proteins and other macromolecular targets. In contrast to conventional ESI-MS-based ligandreceptor binding studies, this method does not rely on the preservation of noncovalent interactions in the gas phase. Drug action mechanisms often involve the binding of therapeutic agents to receptor sites on proteins or nucleic acids. Therefore, the identification of high-affinity ligands for specific macromolecular targets is a critical step in the drug discovery process.1 Leads can be identified by screening chemical libraries for noncovalent binding to selected targets. These libraries may * To whom correspondence should be addressed. Phone: (519) 661-2111 ext. 86313. Fax: (519) 661-3022. E-mail:
[email protected]. http://publish.uwo.ca/ ∼konerman. (1) George, S. R.; O’Dowd, B. F.; Lee, S. P. Nat. Rev. Drug Discovery 2002, 1, 808-820. 10.1021/ac035230l CCC: $27.50 Published on Web 01/20/2004
© 2004 American Chemical Society
include compounds having structural similarities to known ligands or inhibitors; molecules that are related to existing drugs; or natural products, such as secondary plant metabolites.2-5 Common screening methods include affinity chromatography6 and surface plasmon resonance (SPR) assays.7,8 A number of other techniques make use of the fact that noncovalent interactions affect the diffusion behavior of ligands that bind to a target macromolecule. The translational diffusion coefficient, D, is related to the Stokes radius, Rs, according to
D)
kT 6πηRs
(1)
where k, T, and η represent the Boltzmann constant, the temperature, and the solution viscosity, respectively. The Stokes radius is a hydrodynamic property that describes the effective size of a chemical species. Typically, the potential ligands used in screening assays are relatively small compounds. Consequently, these species will have large diffusion coefficients as long as they are free in solution. However, once a ligand is bound to a protein or nucleic acid target, its translational motions will be governed by the slow diffusion of the macromolecule. Noncovalent ligandmacromolecule interactions will therefore reduce the diffusion coefficient of the ligand. This phenomenon can be exploited for the screening of compound libraries by fluorescence correlation spectroscopy (FCS)9-12 and diffusion-ordered nuclear magnetic resonance spectroscopy (DOSY-NMR).13-15 In a similar fashion, fluorescence polarization assays monitor the rotational diffusion (2) Triolo, A.; Altamura, M.; Cardinali, F.; Sisto, A.; Maggi, C. A. J. Mass Spectrom. 2001, 36, 1249-1259. (3) Fang, L.; Cournoyer, J.; Demee, M.; Zhao, J.; Tokushige, D.; Yan, B. Rapid Commun. Mass Spectrom. 2002, 16, 1440-1447. (4) Shin, Y. G.; van Breemen, R. B. Biopharm. Drug Dispos. 2001, 22, 353372. (5) Seneci, P.; Miertus, S. Mol. Diversity 2000, 5, 75-89. (6) Fassina, G. Encyclopedia of Life Sciences [www.els.net]; Nature Publishing Group: London, 2001. (7) McDonnell, J. M. Curr. Opin. Chem. Biol. 2001, 5, 572-577. (8) Myszka, D. G.; Rich, R. L. Pharm. Sci. Technol. Today 2000, 3, 310-317. (9) Gradl, G.; Guenther, R.; Sterrer, S. BioMethods 1999, 10, 331-351. (10) Rigler, R. J. Biotechnol. 1995, 41, 177-186. (11) Rogers, M. V. Drug Discovery Today 1997, 2, 156-160. (12) Auer, M.; Moore, K. J.; Meyer-Almes, F. J.; Guenther, R.; Pope, A. J.; Stoeckli, K. A. Drug Discovery Today 1998, 3, 457-565. (13) Hajduk, P.; Meadows, R. P.; Fesik, S. W. Q. Rev. Biophys. 1999, 32, 211240. (14) Lin, M.; Shapiro, M. J.; Wareing, J. R. J. Am. Chem. Soc. 1997, 119, 52495250.
Analytical Chemistry, Vol. 76, No. 5, March 1, 2004 1257
of potential ligands in solution.11,16 FCS has proven to be particularly amenable to miniaturization and automation and, therefore, has become an important tool for high-throughput studies.17,18 Unfortunately, many of these existing screening methods suffer from certain limitations. Fluorescence-based approaches are applicable only in cases in which suitable chromophores are available. The selectivity of NMR techniques may be limited by overlapping analyte resonances, especially when studying mixtures of structurally related compounds.13 In addition, analyte concentrations in the millimolar range are often required for NMR. Not only does this result in large sample consumption, but also nonspecific aggregation may become an issue.19 Affinity chromatography and SPR assays require the chemical immobilization of ligand or target molecules on a solid support, which may impede binding due to restrictions in the orientation of the immobilized molecules. Furthermore, nonspecific binding to carrier materials may occur.6 Due to its inherently high selectivity and sensitivity, electrospray ionization mass spectrometry (ESI-MS) represents an interesting alternative to these methods. The “softness” of the ESI process allows many solution-phase noncovalent complexes to be transferred into the gas phase, thus permitting the direct observation of these species in the mass spectrum.20-32 However, it is well-established that ESI mass spectra do not always accurately represent the binding state of molecules in solution. A disruption of noncovalent interactions may take place during ESI, during ion sampling, or during ion transfer through the vacuum chamber of the mass spectrometer. Dissociation events of this type will lead to false negative results. Complexes primarily stabilized by hydrophobic interactions are particularly prone to fragmentation,21,22,33-35 but dissociation can also occur for assemblies that (15) Hodge, P.; Monvisade, P.; Morris, G. A.; Preece, I. Chem. Commun. 2001, 3, 239-240. (16) Jameson, D. M.; Seifried, S. E. Methods 1999, 19, 222-223. (17) Battersby, B. J.; Trau, M. Trends Biotechnol. 2002, 20, 167-173. (18) Wo ¨lcke, J.; Ullmann, D. Drug Discovery Today 2001, 6, 637-646. (19) Hughson, F. M.; Wright, P. E.; Baldwin, R. L. Science 1990, 249, 15441548. (20) Ganem, B.; Li, Y.-T.; Henion, J. D. J. Am. Chem. Soc. 1991, 113, 62946296. (21) Loo, J. A. Mass Spectrom. Rev. 1997, 16, 1-23. (22) Fuerstenau, S. D.; Benner, W. H.; Thomas, J. J.; Brugidou, C.; Bothner, B.; Siuzdak, G. Angew. Chem., Int. Ed. 2001, 40, 542-544. (23) Loo, J. A. Int. J. Mass Spectrom. 2000, 200, 175-186. (24) Rostom, A. A.; Fucini, P.; Benjamin, D. R.; Juenemann, R.; Nierhaus, K. H.; Hartl, F. U.; Dobson, C. M.; Robinson, C. V. Proc. Natl. Acad. Sci., U.S.A. 2000, 97, 5185-5190. (25) Jorgensen, T. J. D.; Roepstorff, P.; Heck, A. J. R. Anal. Chem. 1998, 70, 4427-4432. (26) de Brouwer, A. P. M.; Versluis, C.; Westerman, J.; Roelofsen, B.; Heck, A. J. R.; Wirtz, K. W. A. Biochemistry 2002, 41, 8013-8018. (27) Greig, M. J.; Gaus, H.; Cummins, L. L.; Sasmor, H.; Griffey, R. H. J. Am. Chem. Soc. 1995, 117, 10765-10766. (28) Daniel, J. M.; Friess, S. D.; Rajagopalan, S.; Wendt, S.; Zenobi, R. Int. J. Mass Spectrom. 2002, 216, 1-27. (29) Wang, W.; Kitova, E. N.; Klassen, J. S. Anal. Chem. 2003, 75, 4945-4955. (30) Kempen, E. C.; Brodbelt, J. S. Anal. Chem. 2000, 72, 5411-5416. (31) Sannes-Lowery, K. A.; Drader, J. J.; Griffey, R. H.; Hofstadler, S. A. Trends Anal. Chem. 2000, 19, 481-491. (32) Fabris, D.; Fenselau, C. Anal. Chem. 1998, 71, 384-387. (33) Robinson, C. V.; Chung, E. W.; Kragelund, B. B.; Knudsen, J.; Aplin, R. T.; Poulsen, F. M.; Dobson, C. M. J. Am. Chem. Soc. 1996, 118, 8646-8653. (34) Winston, R. L.; Fitzgerald, M. C. Mass Spectrom. Rev. 1997, 16, 165-179. (35) Wigger, M.; Eyler, J. R.; Benner, S. A.; Li, W.; Marshall, A. G. J. Am. Soc. Mass Spectrom. 2002, 13, 1162-1169.
1258
Analytical Chemistry, Vol. 76, No. 5, March 1, 2004
are mainly held together by Coulombic forces.36 Electrochemically induced pH changes within the ESI capillary may affect the solution charge states of biological molecules, thereby potentially inducing dissociation of noncovalent complexes even before ionization takes place.37,38 ESI-MS can also produce false-positive results, because some analytes tend to form noncovalent gas-phase assemblies, despite the nonexistence of the corresponding complexes in solution.39-42 The presence or absence of noncovalent complex ions in an ESI mass spectrum, therefore, does not necessarily allow conclusions to be drawn regarding specific solution phase interactions. Considerable efforts have been made to develop MS-based methods capable of probing the solution phase interactions of analytes without relying on the preservation of specific complexes in the gas phase. For example, ESI-MS has been used in conjunction with traditional binding assays, such as SPR,7,8,43-45 capillary electrophoresis,46 as well as affinity47,48 and size-exclusion chromatography.49 Other methods, such as pulsed ultrafiltration chromatography/MS50,51 and a number of techniques involving hydrogen/deuterium exchange, have also been developed.52-55 Our group has devised an approach for studying the diffusion behavior of molecules in solution by ESI-MS.56 An analyte solution is infused into a capillary tube that has previously been filled with another solution having a different analyte concentration. Under laminar flow conditions, the velocity distribution within the tube is parabolic. This leads to a dispersion of the boundary between the two solutions as liquid flows through the tube. However, diffusion causes a constant radial mixing of analyte molecules, thereby counteracting the dispersion due to laminar flow.57,58 The (36) Mauk, M. R.; Mauk, A. G.; Chen, Y.-L.; Douglas, D. J. J. Am. Soc. Mass Spectrom. 2002, 13, 59-71. (37) Van Berkel, G. J.; Asano, K. G.; Schnier, P. D. J. Am. Soc. Mass Spectrom. 2001, 12, 853-862. (38) Konermann, L.; Silva, E. A.; Sogbein, O. F. Anal. Chem. 2001, 73, 48364844. (39) Hu, P.; Ye, Q.-Z.; Loo, J. A. Anal. Chem. 1994, 66, 4190-4194. (40) Cunniff, J. B.; Vouros, P. J. Am. Soc. Mass Spectrom. 1995, 6, 437-447. (41) Juraschek, R.; Dulcks, T.; Karas, M. J. Am. Soc. Mass Spectrom. 1999, 10, 300-308. (42) Zechel, D. L.; Konermann, L.; Withers, S. G.; Douglas, D. J. Biochemistry 1998, 37, 7664-7669. (43) Williams, C.; Addona, T. A. Trends Biotechnol. 2000, 18, 45-48. (44) Kikuchi, J.; Furukawa, Y.; Hayashi, N. Mol. Biotechnol. 2003, 23, 203212. (45) Natsume, T.; Nakayama, H.; Jansson, O.; Isobe, T.; Takio, K.; Mikoshiba, K. Anal. Chem. 2000, 72, 4193-4198. (46) Dunayevskiy, Y. M.; Lyubarskaya, Y. V.; Chu, Y.-H.; Vouro, P.; Karger, B. L. J. Med. Chem. 1998, 41, 1201-1204. (47) Rudiger, A.-H.; Rudiger, M.; Carl, U. D.; Chakraborty, T.; Roepstorff, P.; Wehland, J. Anal. Biochem. 1999, 275, 162-170. (48) Kelly, M. A.; McLellan, T. J.; Rosner, P. J. Anal. Chem. 2002, 74, 1-9. (49) Wabnitz, P. A.; Loo, J. A. Rapid Commun. Mass Spectrom. 2002, 16, 8591. (50) Zhao, Y.-Z.; van Breemen, R. B.; Nikolic, D.; Huang, C.-R.; Woodbury, C. P.; Schilling, A.; Venton, D. L. J. Med. Chem. 1997, 40, 4006-4012. (51) Johnson, B. M.; Nikolic, D.; van Breemen, R. B. Mass Spec. Rev. 2002, 21, 76-86. (52) Lorenz, S. A.; Maziarz, E. P., III; Wood, T. D. J. Am. Soc. Mass Spectrom. 2001, 12, 795-804. (53) Xiao, H.; Kaltashov, I. A.; Eyles, S. J. J. Am. Soc. Mass Spectrom. 2003, 14, 506-515. (54) Powell, K. D.; Ghaemmaghami, S.; Wang, M. Z.; Ma, L.; Oas, T. G.; Fitzgerald, M. C. J. Am. Chem. Soc. 2002, 124, 10256-10257. (55) Zhu, M. M.; Rempel, D. L.; Du, Z.; Gross, M. L. J. Am. Chem. Soc. 2003, 125, 5252-5253. (56) Clark, S. M.; Leaist, D. G.; Konermann, L. Rapid Commun. Mass Spectrom. 2002, 16, 1454-1462.
outlet of the tube is connected to the ESI source of a mass spectrometer, where the signal intensity of the analyte is monitored as a function of time. The resulting dispersion profile shows a sigmoidal intensity transition from low to high or from high to low, depending on the concentrations of the two solutions used. The diffusion coefficient of the analyte can be determined from a fit to the measured profile. Analytes with large diffusion coefficients will show relatively steep transitions, whereas smaller diffusion coefficients result in more extended dispersion profiles.56 Recently, we proposed that it might be possible to adapt this ESI-MS-based diffusion method for monitoring noncovalent ligandmacromolecule interactions.59 This idea relies on the same principle as the DOSY-NMR and FCS techniques discussed above: namely, that the diffusion coefficient of a small molecule decreases upon binding to a macromolecular receptor. The feasibility of this approach was tested in studies on noncovalent heme-protein interactions in myoglobin under various solvent conditions. Changes in the binding state of the heme group could be identified on the basis of its diffusion behavior.59 Unlike conventional ESI-MS studies on noncovalent complexes, this approach does not rely on the preservation of solution-phase interactions in the gas phase. In fact, very “harsh” conditions are employed at the ion source in order to disrupt any residual interactions, thereby ensuring that the dispersion profiles of individual analytes can be monitored separately. In the current study, we explore the use of this diffusion-based approach for screening tests on mixtures containing multiple potential ligands. Lysozyme was chosen as the model target macromolecule. This protein is an enzyme consisting of 129 amino acid residues (14 306 Da), that hydrolyzes polysaccharides containing N-acetylated monomers.60 Short oligomers of N-acetyl glucosamine (NAG) represent competitive inhibitors of lysozyme. These compounds have dissociation constants on the order of 10 µM and undergo hydrolysis at extremely slow rates (kcat values are in the range of 10-5-10-8 sec-1).61-63 The binding specificity of these inhibitors is generated primarily by an extensive network of hydrogen bonds, both to the protein itself and to bound water molecules.64 We demonstrate the identification of NAG3 as a noncovalent lysozyme ligand in experiments in which the diffusion behavior of multiple candidate saccharides is monitored in parallel. In addition, the results of competition experiments are reported. This second type of assay is based on the displacement of a reference compound from its binding site on the protein. These data clearly establish the potential of ESI-MS-based diffusion measurements as a screening tool for the detection of noncovalent interactions. EXPERIMENTAL SECTION Chemicals. Hen egg white lysozyme, ammonium acetate, D-ribose, L-rhamnose, D-glucose, D-maltose, and maltotriose were purchased from Sigma (St. Louis, MO). Glacial acetic acid and HPLC-grade methanol were products of Fisher Scientific (Nepean, ON, Canada). NAG3 (N,N′,N′′-triacetylchitotriose) and NAG4 (57) Taylor, G. Proc. R. Soc. London A 1953, 219, 186-203. (58) Konermann, L. J. Phys. Chem. A 1999, 103, 7210-7216. (59) Clark, S. M.; Konermann, L. J. Am. Soc. Mass Spectrom. 2003, 14, 430441. (60) Loo, J. A.; Loo, R. R. O.; Udseth, H. R.; Edmonds, C. G.; Smith, R. D. Rapid Comm. Mass Spectrom. 1991, 5, 101-105.
Figure 1. Schematic representation of the experimental setup used in this work. The apparatus allows the dispersion of an initially sharp boundary between two analyte solutions to be studied by ESI-MS. The dispersion takes place within a laminar flow tube (LFT), and analysis of the intensity-time profiles (dispersion profiles) allows the diffusion coefficients of individual analytes to be determined. Arrows indicate the direction of liquid flow. IT, inlet tube; SB, sliding block. Details are described in the text.
(N,N′,N′′,N′′′-tetraacetylchitotetrose) were obtained from Seikagaku America (East Falmouth, MA). Lithium chloride was from Merck (Darmstadt, Germany). All chemicals were used without further purification. ESI-MS. Mass spectra and dispersion profiles were acquired on an API 365 triple quadrupole mass spectrometer (Sciex, Concord, ON, Canada) in positive ion mode using pneumatically assisted ESI (ion spray). All experiments were carried out at room temperature (22 ( 1 °C). The deconvolution of mass spectra was carried out by using the BioMultiView software package supplied by the instrument manufacturer. Diffusion Measurements. Dispersion profiles were recorded by using the laminar flow tube setup shown in Figure 1 in a manner similar to that described previously.56,59 Each experiment involved the use of two solutions, termed 1 and 2. Although both of these solutions contained the same analytes, they differed in the concentration of at least one compound (details are given below). In each individual experiment it was attempted to choose the overall analyte concentration in solution 1 to be about the same as that in solution 2. This was done in an effort to circumvent difficulties that could potentially arise from signal suppression effects.65 For all experiments, both solutions 1 and 2 contained ammonium acetate at a concentration of 1 mM. Initially, the laminar flow tube was filled with solution 1. This was achieved by means of an inlet tube that was connected to the laminar flow tube via a sliding block mechanism. In the aligned position, this mechanism allows liquid to flow directly from the inlet tube into the laminar flow tube. The sliding block was then moved laterally such that the two tubes were no longer aligned. Subsequently, the inlet tube was filled with solution 2. The design of the sliding block ensured that this step could be carried out without disturbing the contents of the laminar flow tube. Upon realignment of the sliding block, a sharp boundary was created between the two solutions. (61) Imoto, T.; Johnson, L. N.; North, A. C. T.; Phillips, D. C.; Rupley, J. A. In The Enzymes; Boyer, P. D., Ed.; Academic Press: New York, 1972; Vol. 7, pp 666-836. (62) Dahlquist, F. W.; Raftery, M. Proc. Natl. Acad. Sci., U.S.A. 1966, 56, 2630. (63) Ganem, B.; Li, Y.-T.; Henion, J. D. J. Am. Chem. Soc. 1991, 113, 78187819. (64) Cheetham, J. C.; Artymiuk, P. J.; Phillips, D. C. J. Mol. Biol. 1992, 224, 613-326. (65) Kebarle, P.; Ho, Y. In Electrospray Ionization Mass Spectrometry; Cole, R. B., Ed.; John Wiley & Sons: New York, 1997; pp 3-63.
Analytical Chemistry, Vol. 76, No. 5, March 1, 2004
1259
A diffusion experiment was then begun by pumping solution 2 through the inlet tube into the laminar flow tube. This step represented the t ) 0 time point, at which the mass spectrometer started to monitor the signal intensity of one or more analytes in selected ion mode at the outlet of the laminar flow tube. The current instrument configuration allows up to eight different analytes to be monitored in parallel. The solutions were advanced at 10 µL/min by using a syringe pump (Harvard Apparatus, South Nattick, MA). The laminar flow tube used consisted of Teflon, 266.0-µm i.d. (Upchurch, Oak Harbor, WA). Slightly different tube lengths around 3 m were used for individual experiments. The outlet of the laminar flow tube was connected to two pieces of fused-silica capillary (100-µm i.d., 165-µm o.d.; Polymicro Technologies, Phoenix, AZ). One capillary, ∼5 cm long, was connected to the ESI source of the mass spectrometer. The other capillary was used to introduce a makeup solvent at a flow rate of 10 µL/ min, which mixed with the analyte solution immediately prior to ESI. The makeup solvent used for monitoring the dispersion profiles of saccharides was water/methanol/acetic acid (5:85:10 v/v/v) and contained 5 mM LiCl. Saccharides were detected as singly lithiated ions (ribose, m/z 157.0; rhamnose, m/z 171.1; glucose, m/z 187.1; maltose, m/z 349.2; maltotriose, m/z 511.3; NAG3, m/z 635.5; NAG4, m/z 838.7).66 Dispersion profiles of lysozyme were recorded by monitoring the intensity of [M + 9H]9+ ions as a function of time, using methanol and acetic acid (90:10 v/v) as makeup solvent. The purpose of the makeup solvent is 2-fold. First, it improves the intensity and stability of the signals monitored by ESI-MS. A linear ion response was found for all analytes over the concentration range that was studied. Second, the addition of makeup solvent results in a highly non-native solvent environment, thus promoting the disruption of any noncovalent interactions immediately before the analyte solution reaches the ion source. Note that “native” solvent conditions (i.e., aqueous solution, near-neutral pH) were maintained within the laminar flow tube where the dispersion phenomena that determine the appearance of the monitored profiles take place. The disruption of any noncovalent complexes that may persist after addition of the makeup solvent was accomplished by using harsh desolvation conditions in the ion sampling interface of the mass spectrometer (typically by using orifice, ring, and skimmer voltages of 100, 300, and 0 V, respectively). Diffusion coefficients were determined from the measured dispersion profiles, as described previously.56 For reasons of simplicity, the term “diffusion coefficient” is used throughout this communication, although in the case of ligands that are noncovalently bound to a macromolecule, the term “apparent diffusion coefficient” would be more appropriate. This is due to the fact that these species are involved in a binding equilibrium, where the measured diffusion coefficient represents a weighted average of that of the free and that of the receptorbound forms of the ligand. The measured diffusion coefficients reported in this study are average results obtained from at least three independent measurements; error bars correspond to 1 SD. Direct Screening. Dispersion profiles were monitored simultaneously for individual saccharides in a mixture containing ribose, rhamnose, glucose, maltose, maltotriose, and NAG3. Solution 1 contained 10 µM each of ribose, glucose, and maltotriose and 5 µM each of rhamnose, maltose, and NAG3. Solution 2 was made (66) Kohler, M.; Leary, J. A. Anal. Chem. 1995, 67, 3501-3508.
1260 Analytical Chemistry, Vol. 76, No. 5, March 1, 2004
Figure 2. Dispersion profiles of rhamnose recorded in the absence (A) and in the presence (B) of lysozyme. Solid lines represent fits to the experimental on the basis of eq 17 of ref. 56.
up of 10 µM each of rhamnose, maltose, and NAG3 and 5 µM each of ribose, glucose, and maltotriose. For dispersion experiments carried out in the presence of lysozyme, the protein concentration was 100 µM in both solutions. Viscosity differences were found to be
