Promiscuous Binding of Ligands by β-Lactoglobulin Involves Hydrophobic Interactions and Plasticity

doi:10.1016/j.jmb.2007.01.077

J. Mol. Biol. (2007) 368, 209–218

Promiscuous Binding of Ligands by β-Lactoglobulin Involves Hydrophobic Interactions and Plasticity Tsuyoshi Konuma, Kazumasa Sakurai and Yuji Goto⁎ Institute for Protein Research, Osaka University, and CREST, Japan Science and Technology Agency, 3-2 Yamadaoka, Suita, Osaka 565-0871, Japan

Bovine β-lactoglobulin (βLG) binds a variety of hydrophobic ligands, though precisely how is not clear. To understand the structural basis of this promiscuous binding, we studied the interaction of βLG with palmitic acid (PA) using heteronuclear NMR spectroscopy. The titration was monitored using tryptophan fluorescence and a HSQC spectrum confirmed a 1:1 stoichiometry for the PA-βLG complex. Upon the binding of PA, signal disappearances and large changes in chemical shifts were observed for the residues located at the entrance and bottom of the cavity, respectively. This observation indicates that the lower region makes a rigid connection with PA whereas the entrance is more flexible. The result is in contrast to the binding of PA to intestinal fatty acid-binding protein, another member of the calycin superfamily, in which structural consolidation occurs upon ligand binding. On the other hand, the ability of βLG to accommodate various hydrophobic ligands resembles that of GroEL, in which a large hydrophobic cavity and flexible binding site confer the ability to bind various hydrophobic substrates. Considering these observations, it is suggested that, in addition to the presence of the hydrophobic cavity, the plasticity of the entrance region makes possible the binding of hydrophobic ligands of various shapes. Thus, in contrast to the specific binding seen for many enzymes, βLG provides an example of binding with low specificity but high affinity, which may play an important role in protein–ligand and protein–protein networks. © 2007 Elsevier Ltd. All rights reserved.

*Corresponding author

Keywords: calycin superfamily; protein–ligand interaction; palmitic acid; NMR; Tanford transition

Introduction Recently, a new field, interactome or systems biology, has been developed.1–3 The purpose of systems biology is to clarify the interactions of biomolecules, such as proteins, nucleic acids, and other compounds, at various levels, e.g. individual, tissue, and cell levels. Recent high-throughput methods have revealed the networks of various species and tissues.1–3 However, such information explains the biochemical relations between moleAbbreviations used: βLG, β-lactoglobulin; CSD, chemical shift difference; HSQC, heteronuclear single quantum coherence; I-FABP, intestinal fatty acid-binding protein; Kd, dissociation constant; NOE, nuclear Overhauser effect; PA, palmitic acid; R2, transverse relaxation rate. E-mail address of the corresponding author: [email protected]

cules but not the physical or structural basis of the interactions. Considering the fact that such networks involve relatively non-specific interactions, e.g. ubiquitin-protein ligases and ATP-dependent proteases, 4 as well as specific interactions, a physical knowledge beside the construction of the networks is essential for discovering drugs or other applications. One possibility suggested is to employ binding regions that have the ability to bind multiple, structurally diverse partners by incorporation of intrinsic disorder in one or both partners.5 So far, much research has addressed specific features of protein–ligand interactions from a physical point of view.6–8 In particular, structural dynamics is important for interactions to occur. NMR-based methods provide significant information about the dynamics of protein–ligand or protein–protein associations.8 For example, ligand binding-sites of mutants of lysozyme and intestinal fatty acid-binding protein (I-FABP) become more

0022-2836/$ - see front matter © 2007 Elsevier Ltd. All rights reserved.

210 rigid upon the binding of a ligand.9–11 On the other hand, upon the association of calmodulin with myosin light chain kinase, some regions become more flexible.12 Such changes may help to stabilize the protein-ligand complex.8 Moreover, NMR studies elaborated the idea of a dynamical “induced fit” explaining the observed broad specificity of some enzymes.13 Plasticity of active site in response to ligand binding has been also proposed for some enzymes on the basis of the X-ray crystal structures.14,15 Thus, not only solving the protein structure but also investigating conformational dynamics is an important application of NMR when addressing the mechanism of interaction between proteins and ligands. Bovine β-lactoglobulin (βLG), a major whey protein abundant in cows milk, has been an important target of the study of ligand binding to proteins, although the exact mechanism of ligand binding remains obscure.16,17 βLG consists of 162 amino acid residues (18 kDa), and contains two disulfide bonds (Cys66–Cys160 and Cys106– Cys119) and a free thiol (Cys121). It is a predominantly β-sheet protein consisting of nine β-strands (A–I), of which the A to H-strands form an up-anddown β-barrel, and one major α-helix at the Cterminal end of the molecule.18–22 The β-barrel assumes a flattened calyx (or cone) with a large cavity lined with hydrophobic residues and is accessible to bulk solvent. βLG binds a variety of hydrophobic ligands and it has been assumed that the biological function of βLG is the transportation of retinol or fatty acid.16,17,23 Intriguingly, it has been suggested that the βLG gene is duplicated from the gene for glycodelin, a key protein for fetal development in the endometrium. 17 Moreover, protein folding kinetics studies have focused on βLG, since it shows an α–β transition in which an intermediate with a non-native α-helix accumulates transiently.24,25 βLG is an important model protein with which to study the α–β transition, which might occur during protein folding and furthermore in various biological phenomena including prion infection. The X-ray crystallography of βLG in complexes with various ligands including palmitic acid (PA) revealed that ligands lie within the central cavity of the protein, indicating that the cavity is the primary binding site of ligands.17,21,23,26 However, attempts to characterize the exact mode of binding in solution have failed. It is possible that the crystal packing favors particular complexes with ligands settled in the cavity. Indeed, another binding site of PA located outside of the βLG molecule has been suggested.27 These results suggest that the ligand binding of βLG is different from the tight and complementary binding often observed for complexes of an enzyme and substrate or antibody and antigen. For these reasons, it is essential to clarify the binding of ligands in solution under physiologically relevant conditions. The ligand binding of βLG is suggested to be closely related to the Tanford transition, a pHdependent conformational transition first identified

Promiscuous Ligand-binding of β-Lactoglobulin

by Tanford.16,18,28 and centered around pH 7. Based on the results of X-ray crystallography, Qin et al.19 concluded that the pH-dependent structural change of the EF-loop is responsible for the Tanford transition.28 In our previous study, we performed heteronuclear NMR measurements using a dimeric βLG mutant, A34C, which made possible the almost complete assignment of the NMR peaks at neutral pH. We clarified the molecular basis of the EF-loop's structural change.29 Here, we took advantage of heteronuclear NMR measurements to examine the conformational and dynamical changes of βLG upon the binding of PA. Taking the crystal structure into consideration, we propose that the structural variability of the complex confers low specificity but high affinity, resembling the recognition of substrates by GroEL.30–33

Results Titration with PA monitored by fluorescence Upon titration with PA, the tryptophan fluorescence of A34C increased slightly (∼8%) without a change in spectral shape (Figure 1(a)). The titration curve monitored by measuring the fluorescence intensity at 334 nm indicated strong affinity with a binding stoichiometry of 1:1 (Figure 1(b)). By comparing the titration data with the calculated curves assuming the dissociation constant (Kd) to be

Figure 1. Titration of palmitic acid monitored by measuring tryptophan fluorescence at pH 7.0 and 25 °C. (a) Fluorescence spectra of A34C in the presence of various concentrations of PA. PA concentrations from bottom to top: 0, 1.1, 2.3 and 11.4 μM. The excitation wavelength was 280 nm. (b) Titration curves monitored by measuring fluorescence intensity at 334 nm. The continuous, broken, and dotted lines are theoretical curves based on equation (2) assuming the Kd value to be 10−6, 10−7, or 10−8 M, respectively.

Promiscuous Ligand-binding of β-Lactoglobulin

10−6, 10−7, or 10−8 M, the Kd value was estimated to be less than 10 −7 M. The stoichiometry and estimated Kd value are consistent with those observed for the wild-type βLG,34,35 confirming that A34C is useful for studying the binding of PA. PA titration monitored with HSQC spectra Titration with PA was then monitored using heteronuclear single quantum coherence (HSQC) spectra (Figure 2). In this experiment, the concentration of βLG was fixed at 0.5 mM, that of PA was increased from 0 to 1 mM, and HSQC spectra were acquired at the respective PA concentrations. The concentrations of PA were lower than the critical micelle concentration (2.2 mM), 36 ensuring the monomeric state of PA. As the concentration of PA was increased, some cross-peaks decreased in intensity and others appeared at different positions (Figure 2(a)). However, the number of new signals was less than that of the signals whose intensity decreased. The positions of new signals did not shift during the titration, indicating the relatively slow equilibrium of PA binding.

211 The plots of signal intensity against the concentration of PA showed that the signals for most residues decrease in strength linearly with an increase in the concentration of PA until the concentration ratio of PA to βLG reaches 1. The intensities were constant at a ratio above 1 (Figure 2(b)). This result is in agreement with the titration by tryptophan fluorescence (Figure 1). Furthermore, all residues showed the same titration curve against the PA concentration, indicating that βLG has a single binding site for PA. However, at the saturated level of PA concentration, many peaks of the ligand-free form did not disappear completely (Figure 2, see also the peaks with asterisk in Figure 3). For some of these residues (e.g. Y20 or M24 in Figure 2(b)), the apparent remaining intensity was caused by the overlap of the peaks of the ligand-bound and free forms. For those residues with clearly separated peaks for the ligand-bound and free forms (e. g. G17 and V123 in Figure 2(a); peaks with asterisk in Figure 3(a)), the remaining intensity is probably caused by the multiple binding modes of PA to βLG, where some binding modes do not affect the NMR peaks (see below). Signal assignment of the HSQC spectrum of the bound form

Figure 2. Titration of palmitic acid monitored by measuring HSQC spectra at pH 6.5 and 40 °C. (a) Examples of the spectra during the titration with PA. These spectra were acquired at PA concentrations of 0 to 1.0 mM. (b) Changes in signal intensities of the residues seen in (a) during the titration with PA.

The decrease in signal intensity upon the titration with PA can be caused either by signal broadening or by a change in chemical shift. In order to distinguish these two possibilities, we performed sequential assignments of the cross-peaks for the βLG with PA bound (Figure 3). In the HSQC spectrum shown here, to focus on the peaks of the ligand-bound form, the threshold was set to remove the minor peaks of the ligand-free form. A summary of the assignments is presented as Supplementary Table 1. The signal assignments in the absence of PA were reported previously.29 Based on these assignments, we classified the residues into three types; (A) residues whose signals could not be assigned in the HNCACB and CBCACONH spectra upon the binding of PA, (B) those showing a significant change in chemical shift, and (C) those with no signal change. The former two types provided important information about the conformational change and the dynamics of the βLG with PA bound. Eighteen percent of the signals (27 among 153 cross-peaks) that disappeared on the 3D NMR spectra upon the formation of a complex were classified as type A. The mapping of these missing residues on the crystal structure revealed that they are clustered around the entrance of the cavity, especially at the D-strand, and EF and GH-loops (Figure 4(a), residues colored green). Probably for these residues, conformational exchange on a time scale of microseconds to milliseconds became significant upon the binding of PA, which causes fast decays of their NMR signals during the NMR data acquisition. These results indicate that some motions

Promiscuous Ligand-binding of β-Lactoglobulin

212

Figure 3. Signal assignments and chemical shift differences of A34C βLG with PA bound. (a) HSQC spectrum of the PA-bound form at pH 6.5 and 40 °C at [PA]/ [βLG] = 1.1. The labels indicate the assignments of cross-signals of the main chain HN-N, or the side-chain Hε-Nε of tryptophan residues. To focus on the peaks of the ligand-bound form, the threshold was set to remove the minor peaks of the ligand-free form. Asterisks in the spectrum represent the still remaining peaks of the PA-free form. (b) CSDs for the assigned residues. The peaks painted grey indicate unassigned residues. The filled and open bars on the top indicate the positions of the β-strands and major α-helix, respectively. The triangles on and below the bars indicate the positions of residues with the side-chains outside and inside of the β-barrel, respectively. The grey circles indicate the positions of the residues involved in the dimer interface.

on a time scale of microseconds to milliseconds are enhanced by the binding to a ligand. Twenty seven percent of the signals (41 of 153) were type B residues. They showed chemical shift differences (CSDs) larger than 0.05 ppm upon the binding of PA (Figure 3(b)). CSDs were calculated using the following equation: CSD ¼

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Δδ2H þ ðΔδN =8Þ2

ð1Þ

where ΔδH and ΔδN are chemical shift differences (in ppm) with respect to the H and N axes, respectively. Interestingly, type B residues are distributed around the cavity (Figure 4(a)). Notably, residues with a larger CSD are located on the β-sheet made of B, C, and D β-strands. It is generally considered that the magnitude of chemical shift perturbation tends to correlate with the proximity of perturbant although the correlation is not straightforward.37,38 Indeed, we

showed the clustered and large CSDs near the mutation site upon formation of dimeric βLG, A34C.29 Thus, the signal changes observed here indicate that the environment of these residues is altered upon the binding of PA, but relative to type A residues, the fluctuations in the PA-bound form are less extensive. The difference in distribution between type A and B residues suggests that, upon the binding of PA, the residues of the entrance become more mobile whereas those of the bottom interact with the ligand in a fixed conformation. An external binding site at the groove between the major α-helix and the β-barrel has been suggested.27,39 Side-chains of the β-sheet appear alternatively inside and outside of the β-sheet plane. It is noted that the residues with their side-chains extruding into the cavity are mainly affected (Figure 3(b)), confirming the interaction within the cavity. Thus, our results do not support the presence of an external binding site.

Promiscuous Ligand-binding of β-Lactoglobulin

213

Figure 4. Mapping of CSD (a) and R2 ratio (b) on the crystal structure. (a) Residues whose CSD is larger than 0.15, 0.10, and 0.05 ppm are colored red, orange, and yellow, respectively. Unassigned residues are colored green. (b) Mapping of the R2 ratio. Unassigned residues are also colored green. Residues with a R2 ratio larger than 1.2 are colored cyan. (c) Mapping of CSDs of I-FABP. Residues whose CSD is larger than 0.4, 0.2, and 0.1 ppm are colored red, orange, and yellow, respectively. Residues that appeared upon the binding of PA are green. In all the panels, left and right images are the front and rear views of the crystal structures, respectively.

R2 measurements In order to assess the change in mobility, transverse relaxation rates (R2) of A34C were measured in the PA-free and PA-bound states. Figure 5(a) shows the R2 values obtained for PA-free and bound A34C and Figure 5(b) shows the ratio of R2 values in these two states. In Figure 4(b), the residues with R2 ratios larger than 1.2 are mapped on the crystal structure. It is noted that we cannot determine the R2 values for type A residues in the PA-bound state. Even for the PA-free state, some type A residues exhibited R2 values that were higher than in other regions (Figure 5(a), open circles), indicating that these residues have some conformational fluctua-

tion on a microseconds to milliseconds time scale even in the PA-free state. It is likely that, upon the binding of PA, R2 values of the type A residues are further enhanced, so that the signals decreased in intensity. The residues with a ratio of R2 values larger than 1.2 were clustered around type A residues (Figures 4(b) and 5(b)). These residues also became more mobile upon the binding of the ligand. However, each R2 value did not become significant enough so as to broaden the signal as type A residues did. Taken together, the results of R2 measurements indicated that the barrel entrance is fluctuating even in the PA-free state, and the mobility becomes more significant upon the binding of the ligand. Although the increased conformational dynamics is the major source of increased

214

Promiscuous Ligand-binding of β-Lactoglobulin

Figure 5. Changes in the transverse relaxation rates of A34C βLG upon the binding of PA at pH 6.5. (a) Raw data obtained for free and bound forms of A34C βLG; (b) the ratios between them. The bars and circles at the top of each panel are as described for Figure 3(b).

relaxation rates, it is also possible that the multiple binding modes at the entrance contribute to enhancing chemical exchange, thus increasing R2.

Discussion Promiscuous binding involving structural plasticity and hydrophobic interaction βLG binds a variety of small hydrophobic molecules.16,17,23 As these compounds do not share a common structure, the recognition of βLG is not specific but rather promiscuous. The main forces responsible for the binding of ligands are hydrophobic interactions. Thus, a strict structural complementation as often observed for an enzyme-substrate complex or antibody-antigen complex, substantiated by a set of unique hydrogen bonds, charge–charge interactions, and/or van der Waals interactions, may not be required. Nevertheless, to achieve tight binding with a dissociation constant of less than 10−7 M as estimated here, the protein's conformation should be complementary, at least in part, to the ligand's structure. Most importantly, we observed a change in the mobility of βLG upon the binding of PA, the degree of which varied depending on the depth of the barrel (Figures 4 and 5). The entrance region of the barrel, especially the D-strand and EF and GHloops, fluctuate the most, leading to a disappearance of the peaks in the 3D NMR spectra. One might think a possibility that this disappearance is induced by the flexible head group of PA. However, we have reported that HSQC signals of these regions become broadened at pH above 7 even without ligand molecules (Figure 7 of Sakurai & Goto29). These broadenings are likely to be caused by a pHdependent conformational change, known as the Tanford transition. These results combined with those of R2 measurements indicate that these peripheral regions have an inherent flexibility in

their conformation. Thus, the highly flexible loops above the barrel play important roles in the ligand binding by further increasing the conformational entropy, consistent with the role of conformational entropy in ligand binding proposed by Stone.7,8 Moreover, it is reported that the carboxyl end of the bound PA is fluctuating as evidenced by large Bfactors of the crystal structure23 and a relaxation analysis of 1D NMR.40 Probably, the barrel entrance can change its conformation to accommodate each conformation of the bound PA. In addition to this, the residues around the missing type A residues exhibited an increase in R2 upon the binding of the ligand (Figures 4(b) and 5(b)), suggesting that these residues become more mobile upon the binding, but are less mobile than those located at the entrance. In contrast, the residues at the bottom of the barrel did not show a significant increase in mobility although changes in chemical shift were observed, implying that the interactions are unique and complementary. Crystal structures of βLG complexes reported so far give indications about the preferred ligand structure. Chain-like substrates, such as PA, retinol, and 12-bromododecanoic acid, and slightly bulky substrates, such as cholesterol and vitamin D, bind to the central cavity.17,21,23,26 On the other hand, bulkier molecules, such as ANS, are assumed to bind at the entrance of the barrel.27 The PA molecule is rod-like and flexible so that it can easily move to the bottom of the cavity, whereas cholesterol and vitamin D are relatively bulky and so would find it hard to settle in the cavity. The electron density map of the PA molecule within the βLG complex is intense enough to determine the positions of each atom, whereas that of cholesterol within the complex is too poor to give strict coordinates.17,23 From the discussion described above, it is assumed that the barrel's entrance accommodates various ligands due to its plasticity, whereas the bottom of the cavity shows rigid or somewhat selective binding to the substrate. Taken together, the ability of βLG to bind a variety of hydrophobic compounds is conferred by a unique

Promiscuous Ligand-binding of β-Lactoglobulin

hydrophobic binding site: the bottom is narrow and rigid while the entrance is wide and flexible. The Tanford transition might be an extreme example of the structural variability of the barrel's entrance, which is particularly important to accommodate a wide range of substrate structures. Consequently, βLG can accommodate a variety of substrates by increasing the flexibility and complementing the conformation of substrate. Furthermore, multiple binding modes at the entrance will be possible because of its high flexibility. Our suggestion of multiple binding modes is consistent with the 1D NMR results of Ragona et al.,40 where they observed two resonances of the carboxyl carbon of PA bound to βLG at pH 6.8. Although the physiological relevance remains unclear, the present study significantly clarified the structural basis of the promiscuous binding of ligands by βLG. Comparison with I-FABP and GroEL To focus on the unique binding of ligands by βLG, we compared our results with those for intestinal fatty acid binding protein41,42 and GroEL.30–33 While the former is an example of strict recognition, the latter represents a case of promiscuous binding. Fatty acid-binding proteins (FABPs) are also members of the calycin superfamily and their binding of ligands has been studied extensively.41,42 There are many kinds of FABP each differing in binding behavior: intestinal FABP (I-FABP) binds only single-chain fatty acids with strict conformational requirements, whereas liver FABP (L-FABP) can bind two fatty acid molecules simultaneously or one bulkier ligand. The binding by I-FABP has been investigated using both high-resolution X-ray crystallography and NMR spectroscopy,11,43 demonstrating that the binding through carboxyl and hydrocarbon groups is specific, in contrast to the case of βLG. The crystal structure of I-FABP shows that the ligand-binding site constitutes a large cavity, whose volume is calculated to be 943 Å3 (Figure 6(b)). A PA molecule whose volume is estimated to be 240 Å3 is almost entirely buried in the cavity. As about half of the side-chains lining the cavity are hydrophilic, the

215 driving forces of the binding might come from a combination of hydrophobic and hydrophilic interactions. Indeed, the carboxyl group of PA interacts specifically with W82, R106, and Q115, and the aliphatic tail is bound to the hydrophobic region of the cavity's surface, leading to tight binding with a specific conformation (Figure 6(b)). This contrasts with the situation for βLG (Figure 6(a)): The sidechains in the βLG cavity are mostly hydrophobic and the cavity itself is relatively shallow (315 Å3). Therefore, the hydrocarbon tail of the ligand is placed in the cavity, and the carboxyl head of the ligand is exposed to the solvent in the βLG complex. NMR spectroscopy gave us additional information on the dynamics and structural plasticity of the I-FABP/PA complex.11 In the case of I-FABP, residues with large CSD values upon the binding of the ligand are distributed around the barrel (Figure 4(c)), similar to the case of βLG. However, the number of observable signals for I-FABP increased upon the binding. Furthermore, a relaxation analysis revealed that the “portal site”, which acts as a lid to the cavity and is mobile in the ligand-free conformation, becomes more rigid upon the binding. These results indicate that the binding of the ligand leads to the immobilization of some residues in I-FABP,11 in marked contrast to the binding by βLG. The recognition of a substrate by chaperonin GroEL is the other extreme.30–33 GroEL recognizes the exposed hydrophobic regions of fully or partially unfolded proteins. The substrate-binding site in the apical domain is hydrophobic and flexible and can indiscriminately catch various substrates.30,31 Once bound, the substrates are released into the cavity upon the binding of GroES, a lid enclosing the cavity, and ATP and a concomitant large conformational change enlarging the cavity. The conformational change also makes the inner wall lined by hydrophilic residues, providing the optimal refolding environment for a protein molecule released into the cavity. It is assumed that promiscuous binding is brought about by a combination of the originally hydrophobic huge cavity and the flexible binding site. We consider the promiscuous binding of βLG to have similar features to that of GroEL. Although

Figure 6. Cross-sections of the molecular surface of PA-bound βLG (a) and I-FABP (b). The PDB files used were 1B0O and 2IFB, respectively. PA molecules and oxygen atoms of water molecules are shown as balls. (a) A PA molecule bound to βLG. The head group of the PA molecule is exposed outside of the βLG molecule. (b) A PA molecule bound to I-FABP. The PA molecule is completely buried in the central cavity. Furthermore, there are seven water molecules in the cavity. These Figures were drawn with Molfeat (FiatLux, Tokyo, Japan).

216 there is some selection of ligands based on size due to the limited volume of the central cavity, the inherent flexibility of the barrel entrance makes it possible to accommodate a variety of hydrophobic substrates.

Conclusions We addressed the structural basis of the binding of βLG with PA as a model substrate. While HSQC signals of the residues located at the entrance of the hydrophobic cavity disappeared upon the binding, the residues surrounding them exhibited an increase in dynamics. On the other hand, the residues located at the bottom of the cavity exhibited only chemical shift perturbations. Considering that the residues located at the entrance of the cavity exhibit plasticity in their conformation, the combination of a hydrophobic cavity and the dynamic features of its entrance makes the binding of various hydrophobic ligands possible even in the absence of specific interactions. In other words, βLG provides the strong affinity for various ligands through a combination of free energy gained by hydrophobic interactions and conformational entropic energy gained by the increased flexibility and heterogeneity of the hydrophobic binding site. Thus, βLG represents a binding mechanism intermediate between tight and complementary interactions as exemplified by I-FABP and non-specific hydrophobic interactions as exemplified by GroEL. Since this type of promiscuous interaction can occur in various systems including ubiquitin-protein ligases, ATPdependent proteases, various molecular chaperones,4,33 and some enzymes,13–15 it might play an important role in the networks of protein–ligand and protein–protein interactions.

Materials and Methods

Promiscuous Ligand-binding of β-Lactoglobulin ratio between βLG (with respect to the monomer) and PA ranged from 0 to 2.5. The mixtures were incubated overnight at 40 °C. Extinction coefficients of mutants were calculated from amino acid sequences by the method described by Gill and von Hippel.45 The theoretical curve for the Kd of binding of PA is described by the following equation: pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi A  A2 4½hLG½PA ð2Þ I ð½PAÞ ¼ DI þ I0 2½hLG where A = ([βLG] + [PA] + Kd). ΔI is the difference in fluorescence intensity between the unbound and fully bound conformations, and I0 is the intensity of the unbound form. NMR measurements NMR measurements were performed using a Bruker DRX 500 or DRX 600 (Bruker BioSpin, Rheinstetten, Germany). The pH was adjusted by adding HCl and/or NaOH. No chemical reagents for pH buffering were used. The temperature was 40 °C and 10% (v/v) 2H2O was added for the signal lock. If the measurement took more than two days, 0.02% (w/v) NaN3 was added to prevent corruption. PA at a molar concentration ratio of 0.0:1, 0.4:1, 0.8:1, 1.2:1, 1.6:1, or 2.0:1 with respect to the protein monomer was dissolved in 100% ethanol and dispensed in NMR sample tubes. After the ethanol was evaporated under reduced pressure, solutions of βLG were added to the tubes and incubated overnight at 40 °C. By using 15N, 13C double-labeled A34C, sequential assignments of the main chain atoms were performed at pH 6.5 using the 600 MHz apparatus. NMR pulse sequences used for the assignments were CBCA(CO)NH and HNCACB. The chemical shift assignments and other analyses of NMR data were performed with the program Sparky (Goddard, T.D., and Kneller, D.G., SPARKY 3, University of California, San Francisco). The measurements for relaxation rates were performed using the 600 MHz apparatus. The pulse sequences have been described.46 For the estimation of R2, the intensities of each peak against mixing time were fitted with a single exponential curve. The fittings were performed using Sparky.

Materials [15N]ammonia water and [13C]glycerol were purchased from Shoko Tsusho (Tokyo, Japan). [13C]methanol was obtained from Nippon Sanso (Tokyo, Japan). Palmitic acid was from Sigma (St. Louis, MO). Other reagents were purchased from Nacalai Tesque (Kyoto, Japan). All of the protein samples used were expressed using a methylotrophic yeast Pichia pastoris expression system. The methods of protein expression and purification were as described.29,44

Calculation of cavity and molecular volume The volume of the βLG or FABP cavity and molecular volume of PA were calculated by using the program Grasp.47 The Protein Data Bank files used for the calculations for βLG and I-FABP were 1B0O and 2IFB, respectively. Because the cavities of βLG and I-FABP have paths to the outside of the molecule, Grasp does not recognize them as cavities. Therefore, we added extra atoms to shut the paths and then calculated the volume of the cavity.

Titration experiment monitored by fluorescence spectroscopy The tryptophan fluorescence spectra of A34C βLG were measured at a protein concentration of 0.10 mg ml−1 (5.3 μM) in 50 mM sodium phosphate buffer (pH 7.0). Aliquots (2 ml) of the protein solution were distributed into small glass tubes and 10 μl of ethanol containing the desired amount of PA was added to each tube. The molar

Acknowledgements We thank Masahiro Yagi, Takahisa Ikegami, and Hideo Akutsu (Institute for Protein Research) for instructions regarding NMR measurements. This

Promiscuous Ligand-binding of β-Lactoglobulin

work was supported in part by Grants-in-Aid for Scientific Research from the Japanese Ministry of Education, Science, Culture, and Sports.

Supplementary Data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/ j.jmb.2007.01.077

References 1. Vidal, M. (2005). Interactome modeling. FEBS Letters, 579, 1834–1838. 2. Cesareni, G., Ceol, A., Gavrila, C., Palazzi, L. M., Persico, M. & Schneider, M. V. (2005). Comparative interactomics. FEBS Letters, 579, 1828–1833. 3. Uetz, P., Dong, Y.-A., Zeretzke, C., Atzler, C., Baiker, A., Berger, B. et al. (2006). Herpesviral protein networks and their interaction with the human proteome. Science, 311, 239–242. 4. Glickman, M. H. & Ciechanover, A. (2002). The ubiquitin-proteasome proteolytic pathway: destruction for the sake of construction. Physiol. Rev. 82, 373–428. 5. Dunker, A. K., Cortese, M. S., Romero, P., Lakoucheva, L. M. & Uversky, N. (2005). Flexible nets. The roles of intrinsicdisorder in protein interaction networks. FEBS J. 272, 5129–5148. 6. Deremble, C. & Lavery, R. (2005). Macromolecular recognition. Curr. Opin. Struct. Biol. 15, 171–175. 7. Zídek, L., Novotny, M. V. & Stone, M. J. (1999). Increased protein backbone conformational entropy upon hydrophobic ligand binding. Nature Struct. Biol. 6, 1118-1087. 8. Stone, M. J. (2001). NMR relaxation studies of the role of conformational entropy in protein stability and ligand binding. Acc. Chem. Res. 34, 379–388. 9. Skrynnikov, N. R., Mulder, F. A. A., Hon, B., Dahlquist, F. W. & Kay, L. E. (2001). Probing slow time scale dynamics at methyl-containing side chains in proteins by relaxation dispersion NMR measurements: application to methionine residues in a cavity mutant of T4 lysozyme. J. Am. Chem. Soc. 123, 4556–4666. 10. Mulder, F. A. A., Mittermaier, A., Hon, B., Dahlquist, F. W. & Kay, L. E. (2001). Studying excited states of proteins by NMR spectroscopy. Nature Struct. Biol. 8, 932–935. 11. Hodsdon, M. E. & Cistola, D. P. (1997). Ligand binding alters the backbone mobility of intestinal fatty acidbinding protein as monitored by 15N NMR relaxation and 1H exchange. Biochemistry, 36, 2278–2290. 12. Lee, A. L., Kinnear, S. A. & Wand, A. J. (2000). Redistribution and loss of side chain entropy upon formation of a calmodulin-peptide complex. Nature Struct. Biol. 7, 72–77. 13. Davis, J. H. & Agard, D. A. (1998). Relationship between enzyme specificity and the backbone dynamics of free and inhibited alpha-lytic protease. Biochemistry, 37, 7696–7707. 14. Fritz, T. A., Tondi, D., Finer-Moore, J. S., Costi, M. P. & Stroud, R. M. (2001). Predicting and harnessing protein flexibility in the design of species-specific inhibitors of thymidylate synthase. Chem. Biol. 8, 981–995.

217 15. Gamage, N. U., Tsvetanov, S., Duggleby, R. G., McManus, M. E. & Martin, J. L. (2005). Structure of a human carcinogen-converting enzyme, SULT1A1. Structural and kinetic implications of substrate inhibition. J. Biol. Chem. 280, 41482–41486. 16. Sawyer, L. & Kontopidis, G. (2000). The core lipocalin, bovine β-lactoglobulin. Biochim. Biophys. Acta, 1482, 136–148. 17. Kontopidis, G., Holt, C. & Sawyer, L. (2004). βLactoglobulin: binding properties, structure, and function. J. Dairy Sci. 87, 785–796. 18. Brownlow, S., Cabral, J. H. M., Cooper, R., Flower, D. R., Yewdall, S. J., Polikarpov, I. et al. (1997). Bovine β-lactoglobulin at 1.8 Å resolution–still an enigmatic lipocalin. Structure, 5, 481–495. 19. Qin, B. Y., Bewley, M. C., Creamer, L. K., Baker, H. M., Baker, E. N. & Jameson, G. B. (1998). Structural basis of the Tanford transition of bovine β-lactoglobulin. Biochemistry, 37, 14014–14023. 20. Kuwata, K., Hoshino, M., Forge, V., Era, S., Batt, C. A. & Goto, Y. (1999). Solution structure and dynamics of bovine β-lactoglobulin A. Protein Sci. 8, 2541–2545. 21. Qin, B. Y., Qreamer, L. K., Baker, E. N. & Jameson, G. B. (1998). 12-Bromododecanoic acid binds inside the calyx of bovine β-lactoglobulin. FEBS Letters, 438, 272–278. 22. Kuwata, K., Hoshino, M., Era, S., Batt, C. A. & Goto, Y. (1998). α→β Transition of β-lactoglobulin as evidenced by heteronuclear NMR. J. Mol. Biol. 283, 731–739. 23. Wu, S.-Y., Pérez, M. D., Puyol, P. & Sawyer, L. (1999). β-Lactoglobulin binds palmitate within its central cavity. J. Biol. Chem. 274, 170–174. 24. Hamada, D., Segawa, S. & Goto, Y. (1996). Non-native α-helical intermediate in the refolding of β-lactoglobulin, a predominantly beta-sheet protein. Nature Struct. Biol. 3, 868–873. 25. Kuwata, K., Shastry, R., Cheng, H., Hoshino, M., Batt, C. A., Goto, Y. & Roder, H. (2001). Structural and kinetic characterization of early folding events in β-lactoglobulin. Nature Struct. Biol. 8, 151–155. 26. Kontopidis, G., Holt, C. & Sawyer, L. (2002). The ligand-binding site of bovine β-lactoglobulin: evidence for a function? J. Mol. Biol. 318, 1043–1055. 27. Collini, M., D'Alfonso, L., Molinari, H., Ragona, L., Catalano, M. & Baldini, G. (2003). Competitive binding of fatty acids and the fluorescent probe 1-8anilinonaphthalene sulfonate to bovine β-lactoglobulin. Protein Sci. 12, 1596–1603. 28. Tanford, C., Bunville, L. G. & Nozaki, Y. (1959). The reversible transformation of β-lactoglobuilin at pH 7.5. J. Am. Chem. Soc. 81, 4032–4036. 29. Sakurai, K. & Goto, Y. (2006). Dynamics and mechanism of the Tanford transition of bovine βlactoglobulin studied using heteronuclear NMR spectroscopy. J. Mol. Biol. 356, 483–496. 30. Gómez-Puertas, P., Martín-Benito, J., Carrascosa, J. L., Willison, K. R. & Valpuesta, J. M. (2004). The substrate recognition mechanisms in chaperonins. J. Mol. Recognit. 17, 85–94. 31. Braig, K., Otwinowski, Z., Hegde, R., Boisvert, D. C., Joachimiak, A., Horwich, A. L. & Sigler, P. B. (1994). The crystal structure of the bacterial chaperonin GroEL at 2.8 Å. Nature, 371, 578–586. 32. Chatellier, J., Buckle, A. M. & Fersht, A. R. (1999). GroEL recognises sequential and non-sequential linear structural motifs compatible with extended

Promiscuous Ligand-binding of β-Lactoglobulin

218

33.

34.

35. 36. 37.

38. 39.

40.

beta-strands and alpha-helices. J. Mol. Biol. 292, 163–172. Wang, Q., Buckle, A. M. & Fersht, A. R. (2000). From minichaperone to GroEL 1: information on GroELpolypeptide interactions from crystal packing of minichaperones. J. Mol. Biol. 304, 873–881. Narayan, M. & Berliner, L. J. (1998). Mapping fatty acid binding to β-lactoglobulin: ligand binding is restricted by modification of Cys 121. Protein Sci. 7, 150–157. Frapin, D., Dufour, E. & Haertlé, T. (1993). Probing the fatty acid binding site of β-lactoglobulins. J. Protein Chem. 12, 443–449. Klevens, H. B. (1953). Structure and aggregation in dilute solutions of surface active agents. J. Am. Oil Chem. Soc. 30, 74–80. Stevens, S. Y., Sanker, S., Kent, C. & Zuiderweg, E. R. P. (2001). Delineation of the allosteric mechanism of a cytidylyltransferase exhibiting negative cooperativity. Nature Struct. Biol. 8, 947–952. Zuiderweg, E. R. P. (2002). Mapping protein-protein interactions in solution by NMR spectroscopy. Biochemistry, 41, 1–7. Monaco, H. L., Zanotti, G., Spadon, P., Bolognesi, M., Sawyer, L. & Eliopoulos, E. E. (1987). Crystal structure of the trigonal form of bovine β-lactoglobulin and of its complex with retinol at 2.5 Å resolution. J. Mol. Biol. 197, 695–706. Ragona, L., Fogolari, F., Zetta, L., Pérez, D. M., Puyol, P., de Kruif, K. et al. (2000). Bovine β-lactoglobulin:

41. 42. 43.

44.

45. 46.

47.

interaction studies with palmitic acid. Protein Sci. 9, 1347–1356. Hamilton, J. A. (2004). Fatty acid interactions with proteins: what X-ray crystal and NMR solution structures tell us. Prog. Lipid Res. 43, 177–199. Hanhoff, T., Lücke, C. & Spener, F. (2002). Insights into binding of fatty acids by fatty acid binding proteins. Mol. Cell. Biochem. 239, 45–54. Sacchettini, J. C., Gordon, J. I. & Banaszak, L. J. (1989). Crystal structure of rat intestinal fatty-acid-binding protein. Refinement and analysis of the Escherichia coli-derived protein with bound palmitate. J. Mol. Biol. 208, 327–339. Kim, T.-R., Goto, Y., Hirota, N., Kuwata, K., Denton, H., Wu, S.-Y. et al. (1997). High-level expression of bovine β-lactoglobulin in Pichia pastoris and characterization of its physical properties. Protein Eng. 10, 1339–1345. Gill, S. C. & von Hippel, P. H. (1989). Calculation of protein extinction coefficients from amino acid sequence data. Anal. Biochem. 182, 319–326. Kay, L. E., Torchia, D. A. & Bax, A. (1989). Backbone dynamics of proteins as studied by 15N inverse detected heteronuclear NMR spectroscopy: application to staphylococcal nuclease. Biochemistry, 28, 8972–8979. Nicholls, A., Sharp, K. A. & Honig, B. (1991). Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins: Struct. Funct. Genet. 11, 281–296.

Edited by K. Kuwajima (Received 3 December 2006; received in revised form 19 January 2007; accepted 30 January 2007) Available online 7 February 2007