Cross-saturation and transferred cross-saturation experiments

Cross-saturation and transferred cross-saturation experiments

Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140 Contents lists available at ScienceDirect Progress in Nuclear Magnetic Resonan...

4MB Sizes 4 Downloads 156 Views

Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

Contents lists available at ScienceDirect

Progress in Nuclear Magnetic Resonance Spectroscopy journal homepage: www.elsevier.com/locate/pnmrs

Cross-saturation and transferred cross-saturation experiments Ichio Shimada a,b,*, Takumi Ueda a, Masahiko Matsumoto a, Masayoshi Sakakura a, Masanori Osawa a, Koh Takeuchi a, Noritaka Nishida a, Hideo Takahashi a,b a b

Graduate School of Pharmaceutical Sciences, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan Biomedicinal Information Research Center (BIRC), National Institute of Advanced Industrial Science and Technology (AIST), Aomi, Koto-ku, Tokyo 135-0064, Japan

a r t i c l e

i n f o

Article history: Received 2 June 2008 Accepted 14 July 2008 Available online 24 July 2008

Ó 2008 Elsevier B.V. All rights reserved.

Keywords: Cross-saturation Protein–protein interactions Membrane proteins Extracellular matrix Intrinsically unstructured proteins

Contents 1. 2. 3. 4. 5. 6.

7. 8.

9.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Principle of the cross-saturation (CS) method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of the method to the FB–Fc complex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Comparison of the FB binding sites on Fc determined by CS, X-ray crystallography, chemical shift perturbation, and H–D exchange experiments Utilization of methyl proton resonances in cross-saturation experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of CS to biological systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1. Separation of the direct intermolecular contacts from the secondary effects due to ligand-induced conformational changes . . . . . . . . . 6.2. Identification of the homodimer interface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3. Systems where complete deuteration of ligand proteins is a challenge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4. Proteins labile in unbound form. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5. Networks of protein–protein interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.6. Weak interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transferred cross-saturation (TCS) measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of TCS to huge and/or inhomogeneous biological systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1. Interactions between ion channels and pore-blockers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2. Membrane permeabilization mechanism by peptides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3. Interactions of collagen-binding proteins with fibrillar collagen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

123 124 124 126 126 129 129 131 131 131 132 132 132 133 133 135 136 139 139

1. Introduction

* Corresponding author. Address: Graduate School of Pharmaceutical Sciences, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Tel./fax: +81 3 3815 6540. E-mail address: [email protected] (I. Shimada). 0079-6565/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.pnmrs.2008.07.001

Advances in structural biology techniques have greatly accelerated the structural determinations of proteins and their complexes, and thus increasing numbers of three-dimensional structures are now available in the Protein Data Bank. However, an individual structure can provide only limited information about its functions,

124

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

lography results could arise because chemical shifts and the H–D exchange rates are affected by various factors, such as differences in the microenvironment and subtle conformational changes induced by the Fc binding. Therefore, an alternative NMR strategy is obviously required for the identification of interaction sites in large protein–protein complexes. 2. Principle of the cross-saturation (CS) method

Fig. 1. Principle of the cross-saturation experiment. The protein with residues to be identified in the complex interface (protein I) is uniformly labeled with 2H and 15N. The saturation caused by irradiation of the non-labeled protein (protein II) is transferred to protein I and is limited to the molecular interface.

since the structure itself yields few clues about the specific target recognition. Therefore, structural analyses of protein–protein interactions are required to reveal the mechanism of their functions. Several attempts have been made to investigate protein–protein interactions by using NMR [1–5]. Especially, NMR methods that use chemical shift perturbation [6] and/or hydrogen–deuterium (H–D) exchange [7] for the main-chain amide groups of the complex have frequently been used to identify the interfaces of larger protein–protein complexes with molecular mass of over 50 kDa in solution. We have investigated the recognition mode of the B domain of protein A (FB), which is composed of a bundle of three a-helices, complexed with the Fc fragment of immunoglobulin G (IgG), with a molecular mass of 64 kDa. We identified the residues responsible for the binding to the Fc fragment, based upon the combined data from hydrogen–deuterium exchange experiments and 1H–15N correlation spectroscopy, using FB uniformly labeled with 15N [8–10]. The three-dimensional structure of FB bound to the Fc fragment was also solved by an X-ray study [11]. It is quite interesting that the residues identified by NMR are similar, but not identical, to those of the contact surface determined by the X-ray study. Some of the residues that experienced significant changes in their chemical shifts and H–D exchange rates upon binding to the Fc fragment did not exist within the contact area determined by X-ray crystallography. The contradiction between the NMR and X-ray crystal-

The principle of the cross-saturation (CS) method is schematically described in Fig. 1 [12]. Protein I, with residues to be identified as the interface residues, is uniformly labeled with 2H and 15 N, and then used to form a complex with a non-labeled target protein (protein II). Accordingly, the complex is composed of molecules with low and high proton densities. In the case of large proteins, if the aliphatic proton resonances are irradiated non-selectively using an RF field with a frequency corresponding to the aliphatic proton resonances, then not only the aliphatic resonances but also the aromatic and amide ones in the target protein are instantaneously saturated. This phenomenon is well known as the spin diffusion effect [13,14]. Although the protein uniformly labeled with 2H and 15N is not directly affected by the RF field, the saturation can be transferred from the target molecule (protein II) to the doubly labeled molecule (protein I) through the interface of the complex, by cross-relaxation. If the proton density of the doubly labeled molecule is sufficiently low, then the saturation transferred to the doubly labeled molecule is limited to the interface. Therefore, we can identify the residues at the interface of protein I by observing the reductions of the peak intensities in the 1H–15N HSQC spectra measured by TROSY coherence transfer [15–17]. 3. Application of the method to the FB–Fc complex The pulse scheme used in the analysis of the FB–Fc complex is shown in Fig. 2. It consists of an alternative band-selective WURST-2 saturation scheme [18], followed by a water flip-back TROSY-HSQC experiment [15,16]. We employed a broadband decoupling scheme for the saturation of the aliphatic proton resonances. This method was applied to the 2H, 15N-labeled FB complexed with a non-labeled Fc fragment of a human myeloma protein IgG (j) Ike-N. Using the saturation scheme, the 1H NMR spectrum observed for the Fc fragment reveals that the saturation caused by the irradiation of the aliphatic resonances efficiently spreads to the

Fig. 2. Pulse scheme for the cross-saturation experiment. Unless otherwise specified, pulse phases are applied along the x-axis. Narrow and wide bars depict 90° pulses and 180° pulses, respectively. Short solid bars are water-selective 90° pulses. The line marked Gz indicates the duration and the amplitude of the sine-shaped pulsed magnetic field gradient applied along the z-axis: G1 = (600 ls, 7.5 G cm1); G2 = (1000 ls, 10 G cm1); G3 = (600 ls, 14.5 G cm1); G4 = (600 ls, 20 G cm1). The delay, D, is 2.25 ms. The following phase cycling scheme was used: /1 = {y, y, x, x}; /2 = {y}; /3 = {y}; /4 = {x}; /5 = {y}; /6(receiver) = {y, y}. In the 15N(t1) dimension, a phasesensitive spectrum is obtained by recording a second free induction decay (FID) for each increment of t1, with /1 = {y, y, x, x}; /2 = {y}; /3 = {y}; /4 = {x}; /5 = {y}; / 6(receiver) = {x, x}. In order to measure a spectrum with RF irradiation of an aliphatic region, a band-selective WURST-2 saturation scheme is applied during the Tsat period prior to the TROSY-HSQC scheme. In this case, the carrier frequency is switched to that for the irradiation before the WURST-2 saturation period (Tsat) and back to the water frequency just before the TROSY-HSQC scheme. The additional relaxation delay (Tadj) can be set to an appropriate value for obtaining a sufficient signal-to-noise ratio.

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

aromatic and amide resonances, due to spin diffusion. Furthermore, the saturation scheme used here was highly selective for the aliphatic proton region, and thus the leakage of the irradiation to the water resonance was negligible. Therefore, the irradiation of the FB–Fc complex did not affect the intensities of the FB amide protons with rapid hydrogen exchange rates. Fig. 3a and b shows the 1H–15N TROSY-HSQC spectra observed for the complex of the doubly labeled FB and the Fc fragment in 10% H2O/90% 2H2O, without and with irradiation, respectively. The effect of the irradiation with a saturation time of 1.2 s on the FB molecule in the complex is clearly observed in Fig. 3b. The intensities of some cross-peaks are significantly reduced by the irradiation. This indicates that the saturation in the Fc fragment of the complex is transferred to the bound FB through the interface. It should be noted that the concentration

125

of H2O in the sample solution used in the present experiment was quite low. The efficiency of the spin diffusion suppression in the FB molecule depends on the concentration of H2O in the sample solution. The FB molecule in the complex assumes the conformation of a bundle of three a-helices. Therefore, under the conditions of 90% H2O/10% 2H2O, which are conventionally used for NMR measurements, the amide proton of the deuterated FB at the i th position is spatially close, within 4 Å, to those at the i  1 and i + 1th positions, leading to the strong dipole–dipole interaction between the intramolecular amide protons (Fig. 4). The data reveal that, under the conditions of 10% H2O/90% 2H2O, the amide protons of the deuterated FB essentially exist in isolation from the other amide protons, and therefore the effect of the spin diffusion on the bound FB is effectively suppressed.

Fig. 3. The results of the cross-saturation experiment. 1H–15N TROSY-HSQC spectra observed at 14.1 T for uniformly 2H, 15N-labeled FB complexed with the Fc fragment in 10% H2O/90% 2H2O (a) without and (b) with irradiation. The spectra were measured with a reasonable protein concentration (0.5–1.0 mM) and within a reasonable measuring time (20 h).

Fig. 4. The tube representation of FB. Amide protons are shown as balls in the figure.

126

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

study, are compared in Fig. 6. In Fig. 6a, the residues listed as contact residues of the Fc fragment by the X-ray study are mapped on the structure of FB [9,11]. Fig. 6b and c shows the binding sites identified based on the chemical shift perturbation data and the changes of the H–D exchange rates for the amide groups of the backbone of FB upon binding to the Fc fragment, respectively [10]. As shown in Fig. 6b and c, the NMR data indicate that helices I and II of FB are primarily responsible for binding to the Fc fragment. However, it is interesting that some of the contact surface residues revealed by the X-ray study are barely affected upon Fc binding. Furthermore, the changes in the chemical shifts and the H–D exchange rates induced by the Fc binding range from the surface to the interior of the FB molecule. Our relaxation matrix calculations suggest that the FB residues within about 7 Å from the interface possess intensity ratios less than 0.5. Therefore, in Fig. 6d, the residues with intensity ratios less than 0.5 are colored red, as the contact residues. The contact residues thus identified significantly overlap with those determined by the X-ray crystallographic study. Consequently, we concluded that the CS method makes it possible to determine the interfaces of protein complexes more precisely than the methods used previously. The advantage of using CS to determine the interfacial residues in a protein complex was confirmed by other groups. Matsuda et al. solved the solution structure of the complex between the N-terminal CAD domain of the caspase-activated deoxyribonuclease and the CAD domain of its inhibitor (ICAD), and compared the ICAD binding site on CAD identified by the chemical shift perturbation and CS methods with that determined by the NMR structure [19]. They reported that the interface determined by the CS method overlapped better than that determined by the chemical shift perturbation method. 5. Utilization of methyl proton resonances in cross-saturation experiments Fig. 5. Plots of the intensity ratios of the cross-peaks in the cross-saturation experiments. Ratios of signal intensities originating from the backbone amide groups with irradiation to those without irradiation were measured under the condition of (a) 10% H2O/90% 2H2O and (b) 90% H2O/10% 2H2O. The ratios for Thr 1, Pro 21, Pro 39, and Pro 58 were not available. The intensity ratio for Arg 28 was calculated based on the side chain eNH signal.

Fig. 5 shows the intensity ratios of the cross-peaks observed with the irradiation to those without the irradiation. Under the conditions of 10% H2O/90% 2H2O, small intensity ratios are observed for the residues in helix I (Gln 10-His 19) and helix II (Glu 25-Asp 37) (Fig. 5a). Interestingly, the ratios observed for the region of helices I and II show smaller values at every three or four residues (Gln 11, Tyr 15, and Leu 18 in helix I and Asn 29, Ile 32, and Lys 36 in helix II), suggesting that one side of each of helices I and II is responsible for the binding to the Fc fragment. These results are in agreement with the X-ray study. We performed the same experiment under the solvent conditions of 90% H2O/10% 2H2O. The profile of the plots (Fig. 5b) for the helical region no longer exhibited any distinctive pattern, due to spin diffusion. Therefore, we concluded that a specific intermolecular CS effect can be observed only when the concentration of H2O in the solvent is low enough as well as the predeuterated sample being properly prepared. 4. Comparison of the FB binding sites on Fc determined by CS, X-ray crystallography, chemical shift perturbation, and H–D exchange experiments The binding sites determined by X-ray crystallography and by NMR methods, including the CS method described in the present

For the effective suppression of spin diffusion in ligand proteins, a solvent with a low concentration of H2O is used in the CS experiment. However, since the CS effect is detected by the labile mainchain amide protons, this experimental limitation causes a loss in sensitivity. Furthermore, because the main-chain amide protons usually form the framework structure of the protein, and thus are less exposed on the molecular surface, they are generally less susceptible to saturation transfer from the target protein to the ligand protein. Therefore, we have developed another version of the cross-saturation method, which utilizes the side chain methyl protons of the ligand protein [20]. Fig. 7 shows a histogram representation of the probability of methyl and amide protons to be located within 3.0 Å from the molecular interface of a protein–protein complex. This probability for methyl protons is comparable to or even larger than that of amide protons. Furthermore, the 1H–13C shift correlation spectra of the methyl groups are generally well-resolved, and because of the fast rotation about the methyl symmetry axis, the methyl group exhibits narrow line widths, leading to excellent sensitivity. Additionally, the relatively short longitudinal relaxation time of the methyl protons ensures the short relaxation delay in the experiment, which decreases the total measurement time. Therefore, methyl protons are very suitable probes in cross-saturation experiments used for identifying molecular interfaces. This experiment has been applied to the FB–Fc complex, where FB was extensively deuterated, except for its leucine and isoleucine methyl groups, according to the previously reported protocol [21]. By using this protocol, high levels of protonation were attained for the methyl groups of the leucine and isoleucine (d1 only) residues of FB (there are no valine residues in FB). Fig. 8c shows the 1H–13C

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

127

Fig. 6. Comparison of the binding sites of the FB–Fc complex. (a) The X-ray crystallographic study, (b) the ligand-induced chemical shift perturbations, (c) the H–D exchange experiments, and (d) the cross-saturation experiment. Residues with accessible surface areas that are covered upon binding of the Fc fragment (Phe 6, Gln 10, Gln 11, Asn 12, Phe 14, Tyr 15, Leu 18, His 19, Arg 28, Asn 29, Ile 32, Gln 33, and Lys 36) [11] are colored red in (a). Residues showing absolute values of chemical shift difference ([(DHN)2 + (DN  0.15)2]1/2) of more than 0.2 ppm (Phe 6, Lys 8, Glu 9, Gln 10, Gln 11, Asn 12, Ala 13, Tyr 15, Glu 16, Leu 18, His 19, Asn 29, Gly 30, Leu 35, Lys 36, and Ser 40) are colored red in (b). Of these residues, Ala 13, Gly 30, and Ser 40 are buried in the molecule. On the basis of the BioMagResBank database at http://www.bmrb.wisc.edu, the scale factor of 0.15 used for the normalization of the magnitude of 15N chemical shift changes (in ppm units) was derived from the average (over all residues types but Pro) standard deviations for the backbone 1HN and 15N chemical shifts. Residues showing protection factors of larger than 10 upon binding of the Fc fragment (Phe 14, Tyr 15, Glu 16, Ile 17, Leu 18, His 19, Asn 29, Ile 32, Gln 33, Leu 35, Leu 46, Ala 49, and Lys 50) are colored red in (c) [10]. Of these residues, Glu 16, Ile 17, Leu 46, Ala 49, and Lys 50 are buried in the molecule. Residues with intensity ratios less than 0.5 (Gln 10, Gln 11, Phe 14, Tyr 15, Glu 16, Leu 18, His 19, Arg 28, Asn 29, Ile 32, Gln 33, Leu 35, and Lys 36), 0.5 to 0.6, 0.6 to 0.7, 0.7 to 0.8, and larger than 0.8 are colored red, orange, yellow, green, and blue, respectively, in (d). Molecular graphics images were produced by using the MidasPlus program from Computer Graphics Laboratory, University of California, San Francisco (supported by NIH RR-01081) [84,85].

HMQC spectrum of 2H(1H,13C-methyl)-labeled FB complexed with the Fc of human IgG1. Although the molecular mass of this complex is more than 60 kDa, the methyl signals are well-resolved and the sensitivity of the signals is high. The methyl resonances originating from the Fc-bound form of FB were assigned by using longitudinal two-spin-order exchange between the free and bound forms of FB (Fig. 9) [22].

The methyl-utilizing cross-saturation experiment was performed in a similar manner as the amide proton-based cross-saturation experiment. As shown in Fig. 8a, there are no proton resonances from 2H(1H,13C-methyl)-labeled FB (phosphate buffered saline, 99.8% D2O) at shifts higher than 1.3 ppm. The residual HDO signal can be seen at 4.7 ppm in Fig. 8a. Therefore, irradiation using an RF (radio frequency) field was applied at approxi-

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

% of protons within 3 Å from the interface

128

10 8 6 4 2 0

Protease-inhibitor

Large protease complex

Antibody-antigen

Enzyme complex

G-protein, Cell cycle, Miscellaneous Signal transduction

Methyl p ro Amide p ton roton

Fig. 7. Histogram representation of the probability of methyl and amide protons to be located within 3.0 Å from the interface of the protein–protein complex. The analysis procedures are described in [20]. The classification of protein complexes was according to [86]. Bar graphs that show a 10% ratio in the figure actually indicate that the ratio is larger than 10% (indicated by small arrows).

a

10

I17( δ1)

I32( δ1)

6

4 1

2

0

H(ppm)

b

L46( δ1) L52( δ1) L23( δ2) L20( δ2) L45( δ2)

9

10 10

L18( δ1)

11

I32( φ1)

I17( δ 1)

L52( δ2)

12 13

16 17 18

L23( δ 2) L52( δ 1) L20( δ 2) L45( δ 2) L52( δ 2) L46( δ 2) L35( δ 1) L23( δ 1) 2.0

1.5

20

25

L20( δ1) L23( δ1) L46( δ2) L45( δ1) L35( δ2) 1.2

1.0

0.8 1

0.6

0.4

H(ppm)

21

L46( δ 1) L18( δ 1) L18( δ 2) 25 L45( δ 1) L20( δ 1) 23 24 25 26 27

Fig. 9. The assignment of the resonances originating from the Fc-bound form of 2 H(1H,13C-methyl)-labeled FB. A longitudinal two-spin-order exchange experiment for the 2H(1H,13C-methyl)-labeled FB–Fc complex was performed in 2H2O, 50 mM potassium phosphate, pH 5.0, with 150 mM NaCl at 37 °C. The cross-peaks originating from the free FB and the Fc-bound form of FB are marked with gray and black, respectively.

28

L35( δ 2)

1.0 1

p

20pm 22

1.0

1.5

19

1 3 C

C(ppm)

15

13

15

L18( δ2)

L35( δ1)

14

20

13

8

10

C(ppm)

15

0.5

0.5

29 30

0.0

0.0

H(ppm)

Fig. 8. (a) One-dimensional 13C-decoupled 700 MHz 1H NMR spectrum of 2H(1H, 13 C-methyl)-labeled FB in 2H2O, 50 mM potassium phosphate, pH 6.0 with 150 mM NaCl at 25 °C. (b) Methyl regions of the 13C–1H shift correlation spectrum of 2H(1H, 13 C-methyl)-labeled FB complexed with non-labeled Fc (complex molecular mass > 60 kDa).

mately 3.5–8.5 ppm, a region corresponding to the residual amide protons, aromatic protons, most of the alpha protons, and some of the other aliphatic protons of the Fc fragment. The 1H–13C HMQC spectra observed for the complex between the 2H(1H,13Cmethyl)-labeled FB and the Fc fragment, with and without irradiation, are shown in Fig. 10a. Based on the spectra with and without the irradiation, the intensity ratios for each of the methyl protons were calculated, and are summarized in Fig. 10b. The experiment

was carried out with 0.8 mM labeled FB with 0.5 molar equivalent of the Fc fragment, and the total measuring time was 3 h, which was much shorter than that of the original amide proton-based cross-saturation experiment (1 day) that used a highly deuterated solvent (90% 2H2O). Although the irradiation time was relatively short (0.3 s), as compared with that of the original amide proton-based cross-saturation experiment (usually longer than 1 s), the signal intensities for some of the methyl protons were moderately decreased (intensity ratio was less than 0.8) with the irradiation. This result indicates that the saturation transfer from the Fc protons to the methyl protons of FB located at the molecular interface is highly efficient. Among the 16 methyl proton resonances originating from the Ile and Leu residues of FB, the intensity ratios of Leu18 (d1 and d2) and Ile32 (d1) in the cross-saturation experiment exhibited the lowest intensity ratios (less than 0.7 with more than 0.5 s irradiation), which indicated that these residues are located in the

129

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

a A 15

A

A

I32( δ 1)

A

20

B,C

B,C

13

C(ppm)

I17( δ 1) L23( δ 2) L18( δ 1)

B

B C

C 25

0.5

1.0 1

b

0.5

1.0 1

H(ppm)

H(ppm)

Intensity ratio

1.0 0.8 0.6 0.4 0.2

1

L52

5202

5201

L46

2

4602

4601

L45

4502

4501

L35

3502

I32

3501

L23

3201

13

2302

2002

1

L20

2301

2001

L18

1802

I17

1801

δ1 δ1 δ2 δ1 δ2 δ1 δ2 δ1 δ1 δ2 δ1 δ2 δ1 δ2 δ1 δ2

1701

0.0

13

Fig. 10. The results of methyl-utilizing cross-saturation experiment. (a) H– C HMQC spectra observed at 16.4 T for H( H, C-methyl)-labeled FB in complex with the Fc fragment without (left) and with (right) irradiation. F2 cross-sections through Ile 17(d1) and Ile 32(d1); Leu 23(d2); and Leu 18(d1) of the FB are also shown in the figure. The corresponding F1 frequencies (A–C) are indicated with arrowheads in the spectra. The irradiation time was set to 0.3 s. (b) Ratio of signal intensities originating from the Leu and Ile methyl groups, with and without irradiation, in cross-saturation experiments.

molecular interface with human Fc. These residues were also included among the contact residues determined by the previous amide proton-based experiment (Fig. 6). This highly sensitive methyl-utilizing cross-saturation experiment could be especially useful in the cases of proteins with low solubility and large molecular weight complexes. 6. Application of CS to biological systems In this section, we will describe various applications of CS experiments, including a combined use of the CS and chemical shift perturbation methods to separate the direct intermolecular contacts from the secondary effects due to ligand-induced conformational changes. In addition, the utility of CS experiments for identifying the binding interface in a homodimer, determination of the interaction sites of proteins labile in unbound form, and analyzing networks of protein–protein interactions and weak protein– protein interactions will be discussed. 6.1. Separation of the direct intermolecular contacts from the secondary effects due to ligand-induced conformational changes CD44 is the main cell surface receptor for hyaluronic acid (HA), and it contains a functional HA-binding domain (HABD) composed of a link module with N- and C-terminal extensions (flanking regions) [23–25]. In order to identify the residues of CD44 HABD that

contact HA, CS and chemical shift perturbation experiments were performed [26]. HA with an average molecular mass of 6.9 kDa (termed HA34) was selected as the saturation-donating partner from several HA oligomers. The RF irradiation applied to the complex resulted in selective intensity losses for the CD44 HABD resonances by the CS phenomena. The affected residues are distributed on both the link module and the flanking regions (Fig. 11a). The weighted averaged 1H and 15N chemical shift changes between the HA34 bound and unbound states of CD44 HABD are plotted in Fig. 11b. Interestingly, these residues are mostly localized on the C-terminal extension and the a1 helix, and they are generally inconsistent with the contact residues. Significant chemical shift changes upon HA-binding were observed in the 13Ca resonances of the residues in this region, supporting the idea that the conformational changes upon HA-binding occur in the C-terminal extension and the a1 helix. We determined the NMR structure of the CD44 HABD in its HAbound state, and compared it with the previously reported unbound crystal structure (Fig. 12) [27]. The b-strands in the flanking region (b0: residues 22–27 and b8: residues 143–148) were rearranged, and the structure in the following C-terminal region (residues 153–169) became disordered upon HA-binding. We confirmed the flexibility of the C-terminal region in the HA-bound state by heteronuclear {1H}–15N NOE and trypsin proteolysis experiments [27]. These results are consistent with the chemical shift changes observed upon HA-binding.

130

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

Fig. 11. Comparison between cross-saturation and chemical shift perturbation experiments for the CD44 HABD. (a) Signal intensities of HSQC signals of the CD44 HABD in complex with HA34 were measured for both with and without presaturation. Signal reduction ratios for each residue were plotted. (b) Chemical shift perturbation for the resonance of the CD44 HABD induced by the binding of HA34. The weighed averaged chemical shift differences (Dd) were calculated using the following equation; Dd ¼ ½ðDd2NH þ 0:25  Dd2N Þ1=2 .

Fig. 12. Conformational rearrangement of the CD44 HABD induced by HA-binding. (a) Crystal structure of CD44 HABD is shown as a ribbon presentation, in which a-helix (a1, a2, and a3), b-strand from the link module (b1, b2, b3, b4, b5, and b6), and b-strand from the flanking region (b0, b8, and b9) are indicated. N- and C-terminal residues are indicated. (b) A solution structure of the CD44 HABD in complex with HA6. This figure was made using pyMol (DeLano Scientific).

Chemical shift perturbation is a commonly used NMR method to map protein interfaces. However, in this study, chemical shift perturbation would not be appropriate to identify the ligand-binding site of CD44 HABD, because of the small chemical shift changes at the contact site and the presence of significant conformational changes in the flanking regions. Several chemical shift perturbation

experiments for protein–carbohydrate complexes have been reported [28–32]. A common feature of these experiments is that the chemical shift perturbations of the protons on the putative ligand-binding sites are at most 0.35 ppm. This feature may reflect the fact that the common molecular scaffold of carbohydrates lacks aromatic groups that would cause magnetic anisotropic shielding

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

effects. On the other hand, the efficiency of CS does not depend on the presence of an aromatic group in the ligand molecules. Therefore, the CS method is especially helpful for determining the interfaces of proteins with carbohydrates, which have no source of ring current shifts. A combination of the CS and chemical shift perturbation methods provides a useful means of separating the direct intermolecular contacts from the secondary effects due to ligand-induced conformational changes. 6.2. Identification of the homodimer interface Choi et al. used the CS method to identify the binding interfaces of homodimers [33]. This method is related to the technique for the observation of intermolecular NOEs between homodimer interfaces [34]. Homodimers of target proteins were prepared from mixtures of non-labeled and [ul-2H,15N]-labeled unstructured target proteins, as shown in Fig. 13a. Consequently, the sample solution is a mixture of the fully [ul-2H,15N]-labeled homodimers (Fig. 13b), the fully non-labeled homodimers (Fig. 13d), and the complex of the non-labeled and [ul-2H,15N]-labeled target proteins (Fig. 13c). The signals from the fully non-labeled homodimers are not recorded in 1H–15N HSQC spectra in the experiments. The fully [ul-2H,15N]-labeled homodimers are detectable; however, they are not affected by the irradiation, because there are no aliphatic protons in the complex. The complex of the non-labeled and [ul-2H,15N]-labeled target proteins only exhibit the CS effects and the residues in the interface can be identified in the experiments. 6.3. Systems where complete deuteration of ligand proteins is a challenge Deep et al. [35] applied CS methods to identify the binding interface of cytochrome c, which contains heme as a cofactor, for cytochrome b5. To obtain cytochrome c with deuterated heme for the CS experiments, they utilized a coexpression system of cytochrome c and cytochrome c maturation genes [35]. In cases where the observed proteins in the CS experiments contain cofactors, deuteration of the cofactors is desirable. When the deuteration of the cofactors is difficult to achieve, the development of a saturation

131

scheme that avoids the saturation of cofactors may solve the problem. Matsuda et al. applied the CS method to identify the binding interface of the N-terminal CAD domain of the caspase-activated deoxyribonuclease for the CAD domain of its inhibitor (ICAD) [36]. Although Cys 37 and Cys 50 of CAD were not identified in the binding interface in the NMR structure of the CAD–ICAD complex, these residues were affected by the irradiation in the CS experiments. The authors attributed the intensity reductions to the saturation of the thiol hydrogen atoms of Cys 37 and Cys 50, and thus these residues were excluded from the analysis. Since the chemical shifts of the cysteine thiol groups are often near those of the methyl groups, the saturation of the thiol hydrogen atoms of free cysteine residues may be transferred to the neighboring amide protons. Improving the saturation scheme or mutating the free cysteine residues would solve these problems. 6.4. Proteins labile in unbound form The chemical shift purtabation experiments is useful when the spectra of both the free and bound forms can be recorded under similar experimental conditions [35,37–41]. However, at times proteins are unstable in the free form. The CS method is applicable to such proteins, due to the observation of the bound form. Choi et al. successfully determined the XPF-EB binding residues for ERCC1-FB, even though ERCC1-FB requires XPF-EB for proper folding [33]. Quadt-Akabayov et al. identified the binding interface of IFNa2 to R2-EC. The assignments for free IFNa2 were established at an acidic pH, because free IFNa2 is only monomeric at low pH. In contrast, the stability of the IFNa2/R2-EC complex required that its resonances be assigned at pH 8 [43] and thus the CS method was utilized to successfully identify the binding site of IFNa2 to R2-EC at pH 8, without a comparison of the chemical shifts between the free and bound states. Identification of binding interfaces by chemical shift perturbation experiments is difficult in intrinsically disordered proteins [44,45], which are disordered in the free state and assume a tertiary structure upon complex formation with the target protein, because almost all of the resonances are shifted upon complex formation, or spectra in the free state are not available, because of

Fig. 13. Schematic representation of the CS experiments for identification of the binding interfaces of homodimers. The complex of the non-labeled and [ul-2H,15N]-labeled target proteins is detectable in the CS experiments, and the residues in the homodimer interface are affected by the irradiation.

132

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

their instability. The CS method could be useful in the cases of such systems. 6.5. Networks of protein–protein interactions Protein–protein interactions are the basis for the formation of complex networks under physiological conditions. Each protein usually binds to several target proteins, and some of the interactions promote or inhibit other interactions. Therefore, precise and accurate determination of interface residues by the CS method is important to clarify their mechanisms. Otomo et al. applied the CS method and mutational analysis to determine the binding interfaces on the formin mDia1 DID domain, which inhibits the actin fibril formation by binding to DAD domain, for both the DAD domain and RhoA, which activates the actin fibril formation by cancelling the DID–DAD interaction. They found partial overlapping of the binding interfaces for DAD and RhoA, which are important for the competitive activity regulation by DID and RhoA [46]. Trempe et al. utilized the CS method to show that the UBA domain contains two ubiquitin binding regions, which are responsible for the recognition of K48-linked polyubiquitin [42]. 6.6. Weak interactions Many biologically significant interactions are weak, such as in cellular switches, which must be turned on and off [47]. However, such weak complexes are difficult to crystallize. In addition, a standard NMR analysis based on intermolecular NOE detection by transient-NOE type experiments is also difficult, because of the line broadening induced by chemical exchange or the inhomogeneous geometry of the complexes [40]. The CS method is applicable to weak complexes, because it involves steady-state NOE type experiments, and thus is more sensitive than standard transient-NOE type experiments. Kami et al. successfully identified the non-PxxP peptide interaction sites on SH3 domains, which have lM–mM affinities [48]. CS has also been applied to various weak electron transfer complexes, including the ferredoxin–sulfite reductase [41], cytochrome b5–cytochrome c [35,38], and pseudoazurin–nitrate reductase [49] complexes.

compared with the binding site identified by the TCS method. Moreover, the intact mouse IgG1 used in the present study is known to possess a lower affinity to protein A than the human IgG1 that is used in the previous study. Therefore, fast exchange between the free and bound states of FB is expected to occur, and thus the CS effect would be transferred to the free state of FB with high efficiency. 1 H–15N shift correlation spectra were observed for uniformly 15 N-labeled FB in the free state and uniformly 2H–15N-labeled FB in the presence of the intact IgG at a ratio of FB to the intact IgG of 10:1. The chemical shifts of the cross-peaks originating from the backbone amide protons of FB in the presence of the intact IgG are almost identical to those from the free FB, and are different from those from the bound FB. Therefore, under the present conditions, only the cross-peaks originating from the free FB were detected on the 1H–15N shift correlation spectrum. A TCS experiment was carried out for the FB–intact IgG complex. Fig. 15 shows the effects of the saturation time on the intensity ratios of the cross-peaks, by changing the irradiation time from 0.6 to 2.6 s. The residues with intensity ratios of less than 0.5 are Q11, F14, Y15, and L18 in helix I, and R28 in helix II. Fig. 16 shows the mapping of the affected residues on the structure of FB. Compared with the previous mapping of the contact residues in the FB–Fc complex (Fig. 6d), an almost identical binding surface on FB was obtained. Among the residues, F14, Y15, and L18 are mainly responsible for the Fc binding [10]. The region of the binding site on FB composed of N29, I32, Q33, L35, and K36 in helix II was found to have larger intensity ratios than those obtained from the previous study. The differences in the intensity ratios can be explained by the fact that the profile of the surface on the Fc por-

7. Transferred cross-saturation (TCS) measurements As described in the previous section, the CS method is useful for the identification of the contact residues of protein complexes with high accuracy. However, the CS method is difficult to apply to protein complexes with a molecular mass over 150 kDa, since the resonances originating from the complexes should be directly observed in the CS method. To overcome these limitations, we have developed an extended version of the CS measurement, termed transferred cross-saturation (TCS) measurement, which utilizes the NMR resonances originating from a free protein under the fast exchange process between the free and bound states on the NMR time scale [50]. Under conditions with an excess amount of protein I relative to protein II and a fast exchange rate between the free and bound states of protein I, it is expected that the CS effect that occurs in the bound state of protein I should be efficiently observed in the free state of protein I, due to the long T1 relaxation in the deuterated protein I, as it works well in the transferred NOE experiments. [51,52] (Fig. 14). To assess the utility of this TCS method, we used intact mouse IgG, with a molecular mass of 150 kDa, as the target protein and FB as the ligand protein. The molecular mass of the complex of the intact IgG and FB is 164 kDa. The complex used in the present research is suitable for the evaluation of the TCS method, since the binding site on FB for the Fc fragment is available and thus can be

Fig. 14. Principle of the transferred cross-saturation method between a receptor protein (protein II) and a ligand protein (protein I). NMR resonances of protein I in the free state are observed.

Fig. 15. Effect of the saturation time on the intensity ratios of the cross-peaks originating from the backbone amide groups. Some residues are selectively affected by changing the irradiation time from 0.6 to 2.6 s.

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

133

such as, those between ligand proteins and membrane proteins in lipid bilayers, and between ligand proteins and cells. 8. Application of TCS to huge and/or inhomogeneous biological systems We have applied the TCS method to various huge and/or heterogeneous protein–protein complexes, which are difficult to investigate by other methods in structural biology, including X-ray crystallography and traditional NMR methods. These challenging systems include interactions between ion channels and their pore-blockers, liposomes and proteins permeabilizing cell membranes, and fibrillar collagen and collagen-binding proteins. The following sections describe TCS experiments with these systems. 8.1. Interactions between ion channels and pore-blockers

Fig. 16. Binding sites of the FB–IgG complex determined by the transferred crosssaturation experiments. Residues with intensity ratio less than 0.5, 0.5–0.6, 0.6–0.7, 0.7–0.8, and larger than 0.8 are colored red, orange, yellow, green, and blue, respectively.

tions where helix II of FB binds are different between the mouse and human IgGs. The efficiency of the TCS effects depends on various sample and experimental conditions, such as the binding constant between the ligand and receptor proteins, the molar ratio of receptors to the ligand, and the molecular weight of the receptors. In order to clarify the applicability of the TCS method, we investigated the effects of physical parameters on TCS by a simulation. The simulation indicated that (i) a larger pB, which is the bound fraction of the ligands, is preferred for higher saturation efficiency, (ii) the TCS method is applicable for a system where koff > 0.1 s1, (iii) for koff P 10 s1, pB P 0.1 is preferred, (iv) for koff  1 s1, pB P 0.5 is preferred, and (v) for systems with large sc (1 ls), pB  0.01 is applicable. In the present study, we identified the contact residues in a complex, with a molecular mass of 164 kDa. Since the resonances originating from the ligand protein in the free state are used to identify the contact residues in the complex, the TCS method would be easy to apply to the interaction of much larger proteins,

K+ channels play a crucial role in regulating membrane potential, signal transduction, and various physiological events. A K+ channel derived from Streptomyces lividans, KcsA, functions as a homo-tetramer with a molecular mass of 70.4 kDa [53–55]. The three-dimensional structures of the KcsA K+ channel have been determined [56,57]. Structural information on the interaction mode between ion channels and pore-blocking toxins is crucial for gaining deeper insights into the structure and function of the channels and also for designing drugs affecting nervous system function. However, the nature of the complex of the solubilized ion channels and the pore-blocking toxins, with its slow tumbling motion in solution leading to broad NMR signals from the complex, has hindered detailed NMR analyses of their interactions. In the present section, we will describe the interaction between the KcsA K+ channel and AgTx2, a pore-blocker of K+ channels, by using TCS [58]. We carried out the TCS experiments, as shown in Fig. 17. Based on the spectra with and without the irradiation, we calculated the reduction ratios of the peak intensities. The affected residues formed a contiguous surface on the structure of AgTx2 (Fig. 18a). In contrast, no residues on the back of the molecule were affected (Fig. 18b). The model of the complex between AgTx2 and the KcsA K+ channel was built by molecular dynamics-simulated annealing, to satisfy the TCS experiment [19]. In the complex model, the site

Fig. 17. Application of TCS to the interaction between solubilized ion-channel and pore-blocker. Non-labeled KcsA K channels, which are solubilized in dodecylmaltoside micelles are mixed with five times excess of [2H15N]-labeled pore-blocker, AgTx2. Irradiation of KcsA K channel caused a cross-saturation to the KcsA-interface of bound AgTx2. The cross-saturation is transferred to the free state of AgTx2 via chemical exchange between free and bound states.

134

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

Fig. 18. Determination of the KcsA K channel interface on a pore-blocker, AgTx2, by a TCS experiment. The residues affected in the TCS experiment described in Fig. 16 are mapped on the surface of AgTx2. The darker color indicates stronger reduction of signal intensity for corresponding resonances. The strongly affected residues are labeled. The left and right figures are 180 rotations about the vertical axis relative to each other. The front surface in the left figure corresponds to the KcsA-binding surface on AgTx2.

of AgTx2 fits quite well with the cleft on the extracellular vestibule of the KcsA K+ channel. Based upon the model, we successfully described the specificities of K+ channels for the pore-blockers [58]. In addition, we developed a strategy that reconstitutes KcsA linked to beads, where the KcsA is embedded in lipid bilayers [59]. Analyses of the KcsA-proteoliposomes with the TCS method

allowed us to successfully identify the KcsA-binding interface on AgTx2 (Fig. 19). This strategy would be useful for analyzing the various membrane protein–ligand complexes that require a lipid bilayer environment for their stability and function. This strategy may be particularly beneficial for studying membrane proteins that require specific phospholipids. Furthermore, it would be

Fig. 19. Application of TCS experiment to a membrane protein embedded in a lipid bilayer (proteoliposome). (a) Purified KcsA, harboring a decahistidine tag, was immobilized on the surface of the porous silica beads and a lipid bilayer was reconstituted around the KcsA molecule. The labeling strategies are same as in Fig. 16. (b) The residues affected by the irradiation of KcsA-proteoliposome are mapped on the AgTx2 surface. The darker color indicates stronger reduction of signal intensity for corresponding resonances. The strongly affected residues are labeled. The left and right figures are 180° rotations about the vertical axis relative to each other. The front surface in the left figure corresponds to the KcsA-binding surface on AgTx2.

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

effective for investigations of the interactions between ligandactivated multicomponent receptors and signal transduction molecules. 8.2. Membrane permeabilization mechanism by peptides Anti-microbial peptides exert their activity by permeabilizing the cell membranes of harmful invaders, such as bacteria. Sapecin is an anti-microbial peptide purified from the culture medium of NIH-Sape-4, an embryonic cell line of Sarcophaga peregrina (flesh fly) [60]. Glucose leakage experiments suggest that oligomerization of sapecin is an essential step in its membrane permeabilization process [61]. However, little is known about the structural mechanism of the membrane permeabilization by sapecin. In this section, we describe our investigation of the interaction between the membrane and sapecin and the permeabilization modes of sapecin, by a combination of TCS and H–D exchange experiments [61]. To identify the site buried in the membrane (membrane-buried site), we applied TCS experiments to the sapecin and phospholipid vesicle complex. Based on the spectra with and without irradiation, we calculated the reduction ratios of the peak intensities, which are summarized in Fig. 20a. These affected residues formed a con-

135

tiguous surface on the sapecin structure (Fig. 20b, left). Therefore, we concluded that these residues are the membrane-buried site of sapecin into the phospholipid vesicle. We also measured the H–D exchange rates for the backbone amide groups of sapecin embedded in bilayer vesicles. In the H– D exchange experiments, the amide exchange rates for the residues located in both the membrane-buried and oligomerization sites are expected to decrease. Therefore, a comparison between the results from the TCS and H–D experiments would reveal the sapecin oligomerization site. The residues with protection factors larger than 10 are clustered on the membrane-buried surface, which was determined by the TCS experiment (Fig. 21b). On the other hand, the residues with protection factors between 5 and 10 are mainly distributed on the opposite side of the membrane-buried surface. Therefore, we concluded that the residues on the opposite side of the membrane-buried surface are responsible for the oligomerization of sapecin in the membrane. Based on these results, we proposed a two-step mechanism for membrane permeabilization by sapecin. In the first step, sapecin interacts with the membrane by using the surface composed of the basic and solvent-exposed hydrophobic residues, and then it oligomerizes with other sapecin molecules, as shown in Fig. 22, leading to increased membrane permeabilization.

Fig. 20. Determination of the membrane-buried sites on sapecin by TCS experiment. (a) Signal intensity ratios in the TCS experiment. The [2H15N]-labeled sapecin (L5A/V35A mutant) was mixed with PC vesicles and the resonances corresponding to the alkyl chains of the phospholipid are irradiated. The secondary structures of sapecin indicated in the top of the panel. (b) The residues affected in the TCS experiment are mapped on the solvent-accessible surface of sapecin molecule. The darker color indicates stronger reduction of signal intensity for the corresponding resonances. Strongly affected residues are labeled in the figure. The left and right figures are 180° rotations about the vertical axis relative to each other. The front surface in the left figure corresponds to the lipid-interacting surface on sapecin. (c) Ribbon representation of sapecin molecules with semi-transparent solvent-accessible surface. The view direction is the same as in (b).

136

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

Fig. 21. Determination of the oligomerization surface on sapecin by H–D exchange experiments. (a) Protection factors at pH 3.4 and 20 °C as a function of the residue number. Higher protection factor indicates that the corresponding residues are protected from solvent in the existence of phospholipid membrane. (b) Mapping of the residues with high protection factors in the H–D exchange experiments. The darker color indicates higher protection factor for the corresponding resonances. Highly protected residues are labeled in the figure. The left and right figures are 180° rotations about the vertical axis relative to each other. The view direction is same as in Fig. 19.

8.3. Interactions of collagen-binding proteins with fibrillar collagen Collagen, which is characterized by its long triple-helical structure (polyproline II-like helices) formed by three tightly interwoven polypeptide-chains, is the major component of extracellular matrices. Under physiological conditions at neutral pH, each triple-helix self-assembles into huge fibrillar supramolecules [62] (Fig. 23). Collagen plays a crucial role not only in providing the mechanical strength for extracellular matrices but also in mediating a variety of cellular processes, including cell attachment, hemostasis, and bacterial adherence, through interactions with collagen-binding proteins (CBPs) expressed on cell surfaces. von Willebrand factor (vWF) is a plasma protein that mediates platelet adhesion at damaged sites of vessel walls by linking the subendothelial collagen with platelet receptors, glycoproteins Ib/ IX/V complex [63]. The A3 domain (amino acids 920–1111 of vWF) contains the major binding site for collagen types I and III [64,65]. The X-ray studies of the A3 domain demonstrated that it also has the same fold as the a2-I domain, a CBP sharing high sequence similarity with that of the A3 domain but lacking the metal ion dependent adhesion site (MIDAS) motif, which is suggested to be important in collagen binding [66,67]. In fact, the X-ray study of the complex with a collagen-mimetic peptide showed that the binding site on the a2-I domain is located at the MIDAS motif [66,67]. Therefore, this finding sparked an interest in the collagen-binding site of the A3 domain. Moreover, site-directed mutations introduced at the top surface of the A3 domain, which

corresponds to the binding sites of the a2-I domain, caused no loss of the collagen-binding activity [66,68]. Crystallization of the supramolecular complex between fibrillar collagen and the A3 domain and traditional NMR analyses for the interaction appear to have no prospects. In addition, it would be laborious to investigate the collagen-binding site on the A3 domain in the complex by using a traditional mutagenesis strategy. In order to determine the collagen-recognition mechanism of the A3 domain, we applied the TCS method to the complex of intact collagen [69]. TCS experiments were performed under conditions with an excess amount of the labeled A3 domain relative to the unlabeled collagen. Based on the spectra in the presence of collagen, with and without irradiation, we calculated the reduction ratios of the peak intensities. The residues affected by the irradiation are distributed on the same surface of the A3 domain (Fig. 24). Therefore, we concluded that collagen binds to the A3 domain at the ‘front’ surface, transversely across helix 3-strand C-helix 4. To determine whether the A3 domain could bind to the collagen triple-helix in the fibrillar form without any steric hindrance, we made a docking model of the A3 domain and the collagen microfibril, based upon a model of the collagen microfibril (PDB code, 4CLG) [70]. The model revealed that the binding of the A3 domain is free of serious bad contacts with the fibrillar collagen. After comparing the model of the A3 domain and the collagen peptides (PDB code, 2CLG), and the X-ray structure of the a2-I domain complexed with the collagen-mimetic peptide, we conclude that the triple-helical collagen is bound at the ‘front’ surface of

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

137

Fig. 22. Putative models of membrane-permeabilizing sapecin oligomers. Left and right figures show dimer and trimer of sapecin, respectively. The diagrams show the ribbon representation of sapecin molecules with semi-transparent solvent-accessible surface. Phospholipid membrane is shown in a schematic representation. The upper figure is looking parallel to membrane surface while the lower figure is looking down the membrane with sapecin oligomers embedded in the bilayer. The side chains of Asp4 and Arg23, which are indicated to be important for membrane permeabilization, are shown in stick representations. In the trimer model, a glucose molecule is shown to indicate relative size of the pore formed by sapecin.

Fig. 23. Formation of fibrilar collagen. Collagen a-chains contain glycine residues at every third position to form a triple-helical structure. After removal of the propeptide region at N- and C termini by a protease digestion, collagen loses its solubility at neutral pH, and self-assembles into a huge insoluble fibril.

the A3 domain (Fig. 25a) whereas the a2-I domain recognizes the collagen at its ‘top’ face (Fig. 25b). It is interesting that the A3 domain and the a2-I domain share an identical fold and the same col-

lagen-binding activity, but have different interaction modes. Structural genomics initiatives will undoubtedly produce many more protein structures in the near future. To utilize the wealth

138

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

Fig. 24. Collagen-binding surface identified by TCS experiments. Surface presentation of the crystal structure of the A3 domain (PDB code, 1AO3) [66] is colored according to the TCS result. The residues with signal reduction ratio greater than 0.4 and within the range from 0.3 to 0.4 are colored red and yellow, respectively. Pro-981, which lacks cross-saturation data due to the absence of the amide proton, is colored cyan. The left and right panels are 90° rotation along the vertical axis with respect to each other.

Fig. 25. Comparison of the collagen-binding sites between the A3 domain and a2-I domain. (a) Docking model of the A3 domain and the collagen peptides. (b) Crystal structure of the a2-I domain in complex with a collagen-mimetic peptide containing GFOGER sequence (PDB code, 1DZI) [87]. A metal ion critical for ligand-binding is shown as a dark sphere. This figure was made by pyMol.

Fig. 26. Mapping of the affected residues in the TCS experiments on the DDR2-DS domain. The residues with signal intensity ratios <0.71 (W52, R105, I112, E113, M174, N175, and V176), and within the 0.71–0.76 range (S53, C73, V78, F114, and C177) are shown in black and gray, respectively.

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

139

Fig. 27. The proposed model of the DDR2-DS domain in complex with the triple-helical peptides with the sequence (GPO)12 (pdb code: 2CLG). The collagen-binding residues of the DDR2-DS domain are colored red. The three model peptides of collagen are shown as stick models colored green, blue, and yellow, respectively.

from these initiatives, investigations focused on protein interactions are an important issue. Discoidin domain receptor 2 (DDR2) is another CBP, and its recognition of collagen results in the regulation of cell proliferation and migration [71,72]. In addition, DDR2 controls extracellular matrix remodeling by upregulating both the expression and activity of matrix metalloproteinases [73–76]. DDR2 consists of a discoidin (DS) domain and a stalk region in its extracellular portion, and an intracellular tyrosine kinase, connected by a transmembrane region. The direct interaction of collagen with the DS domain of DDR2 (DDR2-DS domain) triggers the activation of its intracellular tyrosine kinase, leading to downstream intracellular signaling [77–80]. The DDR2-DS domain binds to collagen types I, II, and III, but does not recognize collagen type IV, in agreement with receptor activation studies [79]. DDR2 is reportedly responsible for G0/G1 cell cycle arrest when tumor cells are in contact with the collagen fibril, but not with monomeric collagen [81]. In order to identify the DDR2 residues involved in the direct interaction with the collagen type II fibril, the DDR2-DS domain was studied by TCS experiments [82]. The cross-saturation from collagen to the DDR2-DS domain was quantified as the reduction ratios of the peak intensities. Mapping of these residues on the NMR structure of the DDR2-DS domain revealed that they form a contiguous surface at the ‘top’ of the DDR2-DS domain in L1, L3, L4, and L6, indicating that this surface is the collagen-binding interface (Fig. 26). The DDR2-binding site on collagen should be complementary to the collagen-binding interface on the DDR2-DS domain. As shown in Fig. 27, the binding site of the DDR2-DS domain, which spans almost the same length as two to three triplets in collagen, has two aromatic residues, Trp52 and Phe114, and the charged residues Asp69, Arg105, and Glu113. Therefore, the DDR2-binding sequences on collagen might be two to three triplets, which include at least one negative residue, one or two positive residues, and hydrophobic residues, which could interact with aromatic residues on the DDR2-DS domain. Based on an analysis using synthetic triple triple-helical peptides, GPRGQOGVMGFO (O: hydroxyproline) was recently proposed as the minimal collagen sequence required for DDR2 binding [83]. The length and the hydrophobic and hydrophilic

aspects of the sequence are consistent with the conclusion drawn from the TCS experiments. 9. Conclusion For identifying the interfaces of large protein–protein complexes, the CS and TCS methods are obviously superior to traditional NMR methods, which use the chemical shift perturbation and H–D exchange rates. This is because the CS and TCS methods described here extract more direct information on through-space interactions between the two molecules. The present methods should be generally applicable to large protein complexes and systems, such as extracellular matrices, tissues, or living cells under biologically relevant conditions. Acknowledgements I.S. is grateful to Drs. Hiroaki Terasawa, Tamiji Nakanishi, Mariko Yokogawa, Mitsuhiro Takeda, and Mr. Shinji Ogino and Mr. Osamu Ichikawa for their assistance. This work was supported by a grant from the Japan New Energy and Industrial Technology Development Organization (NEDO). References [1] T. Miura, W. Klaus, B. Gsell, C. Miyamoto, H. Senn, J. Mol. Biol. 290 (1999) 213. [2] H. Matsuo, K.J. Walters, K. Teruya, T. Tanaka, G.T. Gassner, S.J. Lippard, Y. Kyogoku, G. Wagner, J. Am. Chem. Soc. 121 (1999) 9903. [3] M. Mayer, B. Meyer, J. Am. Chem. Soc. 123 (2001) 6108. [4] S. Rajesh, T. Sakamoto, M. Iwamoto-Sugai, T. Shibata, T. Kohno, Y. Ito, Biochemistry 38 (1999) 9242. [5] K.J. Walters, G.T. Gassner, S.J. Lippard, G. Wagner, Proc. Natl. Acad. Sci. USA 96 (1999) 7877. [6] M.P. Foster, D.S. Wuttke, K.R. Clemens, W. Jahnke, I. Radhakrishnan, L. Tennant, M. Reymond, J. Chung, P.E. Wright, J. Biomol. NMR 12 (1998) 51. [7] Y. Paterson, S.W. Englander, H. Roder, Science 249 (1990) 755. [8] H. Torigoe, I. Shimada, A. Saito, M. Sato, Y. Arata, Biochemistry 29 (1990) 8787. [9] H. Gouda, H. Torigoe, A. Saito, M. Sato, Y. Arata, I. Shimada, Biochemistry 31 (1992) 9665. [10] H. Gouda, M. Shiraishi, H. Takahashi, K. Kato, H. Torigoe, Y. Arata, I. Shimada, Biochemistry 37 (1998) 129. [11] J. Deisenhofer, Biochemistry 20 (1981) 2361. [12] H. Takahashi, T. Nakanishi, K. Kami, Y. Arata, I. Shimada, Nat. Struct. Biol. 7 (2000) 220. [13] A. Kalk, H.J.C. Berendesen, J. Magn. Reson. 24 (1976) 343.

140

I. Shimada et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 54 (2009) 123–140

[14] K. Akasaka, J. Magn. Reson. 45 (1981) 337. [15] K. Pervushin, R. Riek, G. Wider, K. Wuthrich, Proc. Natl. Acad. Sci. USA 94 (1997) 12366. [16] K.V. Pervushin, G. Wider, K. Wuthrich, J. Biomol. NMR 12 (1998) 345. [17] M. Salzmann, K. Pervushin, G. Wider, H. Senn, K. Wuthrich, Proc. Natl. Acad. Sci. USA 95 (1998) 13585. [18] E. Kupce, G. Wagner, J. Magn. Reson. B 109 (1995) 329. [19] T. Matsuda, T. Ikegami, N. Nakajima, T. Yamazaki, H. Nakamura, J. Biomol. NMR 29 (2004) 325. [20] H. Takahashi, M. Miyazawa, Y. Ina, Y. Fukunishi, Y. Mizukoshi, H. Nakamura, I. Shimada, J. Biomol. NMR 34 (2006) 167. [21] N.K. Goto, K.H. Gardner, G.A. Mueller, R.C. Willis, L.E. Kay, J. Biomol. NMR 13 (1999) 369. [22] G. Wider, D. Neri, K. Wuthrich, J. Biomol. NMR 98 (1991) 428. [23] A. Aruffo, I. Stamenkovic, M. Melnick, C.B. Underhill, B. Seed, Cell 61 (1990) 1303. [24] J. Lesley, R. Hyman, P.W. Kincade, Adv. Immunol. 54 (1993) 271. [25] D. Naor, R.V. Sionov, D. Ish-Shalom, Adv. Cancer Res. 71 (1997) 241. [26] M. Takeda, H. Terasawa, M. Sakakura, Y. Yamaguchi, M. Kajiwara, H. Kawashima, M. Miyasaka, I. Shimada, J. Biol. Chem. 278 (2003) 43550. [27] M. Takeda, S. Ogino, R. Umemoto, M. Sakakura, M. Kajiwara, K.N. Sugahara, H. Hayasaka, M. Miyasaka, H. Terasawa, I. Shimada, J. Biol. Chem. 281 (2006) 40089. [28] J.L. Asensio, F.J. Canada, M. Bruix, C. Gonzalez, N. Khiar, A. Rodriguez-Romero, J. Jimenez-Barbero, Glycobiology 8 (1998) 569. [29] J.L. Asensio, F.J. Canada, H.C. Siebert, J. Laynez, A. Poveda, P.M. Nieto, U.M. Soedjanaamadja, H.J. Gabius, J. Jimenez-Barbero, Chem. Biol. 7 (2000) 529. [30] J.L. Asensio, H.C. Siebert, C.W. von der Lieth, J. Laynez, M. Bruix, U.M. Soedjanaamadja, J.J. Beintema, F.J. Canada, H.J. Gabius, J. Jimenez-Barbero, Proteins Struct. Funct. Genet. 40 (2000) 218. [31] J.F. Espinosa, J.L. Asensio, J.L. Garcia, J. Laynez, M. Bruix, C. Wright, H.C. Siebert, H.J. Gabius, F.J. Canada, J. Jimenez-Barbero, Eur. J. Biochem. 267 (2000) 3965. [32] J.D. Kahmann, R. O’Brien, J.M. Werner, D. Heinegard, J.E. Ladbury, I.D. Campbell, A.J. Day, Struct. Fold. Des. 8 (2000) 763. [33] Y.J. Choi, K.S. Ryu, Y.M. Ko, Y.K. Chae, J.G. Pelton, D.E. Wemmer, B.S. Choi, J. Biol. Chem. 280 (2005) 28644. [34] K.J. Walters, H. Matsuo, G. Wagner, J. Am. Chem. Soc. 119 (1997) 5958. [35] S. Deep, S.C. Im, E.R. Zuiderweg, L. Waskell, Biochemistry 44 (2005) 10654. [36] T. Matsuda, N. Nakajima, T. Yamazaki, H. Nakamura, J. Mol. Recognit. 17 (2004) 41. [37] T. Tenno, K. Fujiwara, H. Tochio, K. Iwai, E.H. Morita, H. Hayashi, S. Murata, H. Hiroaki, M. Sato, K. Tanaka, M. Shirakawa, Genes Cells 9 (2004) 865. [38] W. Shao, S.C. Im, E.R. Zuiderweg, L. Waskell, Biochemistry 42 (2003) 14774. [39] W.D. Morgan, T.A. Frenkiel, M.J. Lock, M. Grainger, A.A. Holder, Biochemistry 44 (2005) 518. [40] J. Morrison, J.C. Yang, M. Stewart, D. Neuhaus, J. Mol. Biol. 333 (2003) 587. [41] T. Saitoh, T. Ikegami, M. Nakayama, K. Teshima, H. Akutsu, T. Hase, J. Biol. Chem. 281 (2006) 10482. [42] J.F. Trempe, N.R. Brown, E.D. Lowe, C. Gordon, I.D. Campbell, M.E. Noble, J.A. Endicott, EMBO J. 24 (2005) 3178. [43] S.R. Quadt-Akabayov, J.H. Chill, R. Levy, N. Kessler, J. Anglister, Protein Sci. (2006). [44] K. Sugase, H.J. Dyson, P.E. Wright, Nature 447 (2007) 1021. [45] H.J. Dyson, P.E. Wright, Nat. Rev. Mol. Cell Biol. 6 (2005) 197. [46] T. Otomo, C. Otomo, D.R. Tomchick, M. Machius, M.K. Rosen, Mol. Cell 18 (2005) 273. [47] M. Reibarkh, T.J. Malia, B.T. Hopkins, G. Wagner, J. Biomol. NMR 36 (2006) 1. [48] K. Kami, R. Takeya, H. Sumimoto, D. Kohda, EMBO J. 21 (2002) 4268. [49] A. Impagliazzo, M. Ubbink, J. Am. Chem. Soc. 126 (2004) 5658. [50] T. Nakanishi, M. Miyazawa, M. Sakakura, H. Terasawa, H. Takahashi, I. Shimada, J. Mol. Biol. 318 (2002) 245.

[51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71]

[72] [73] [74] [75] [76] [77]

[78] [79] [80] [81] [82] [83] [84] [85] [86] [87]

G.M. Clore, A.M. Gronenborn, J. Magn. Reson. 48 (1982) 402. G.M. Clore, A.M. Gronenborn, J. Magn. Reson. 53 (1983) 423. D.M. Cortes, E. Perozo, Biochemistry 36 (1997) 10343. L.G. Cuello, J.G. Romero, D.M. Cortes, E. Perozo, Biochemistry 37 (1998) 3229. L. Heginbotham, E. Odessey, C. Miller, Biochemistry 36 (1997) 10335. D.A. Doyle, J.M. Cabral, R.A. Pfuetzner, A.L. Kuo, J.M. Gulbis, S.L. Cohen, B.T. Chait, R. MacKinnon, Science 280 (1998) 69. Y. Zhou, J.H. Morais-Cabral, A. Kaufman, R. MacKinnon, Nature 414 (2001) 43. K. Takeuchi, M. Yokogawa, T. Matsuda, M. Sugai, S. Kawano, T. Kohno, H. Nakamura, H. Takahashi, I. Shimada, Structure 11 (2003) 1381. M. Yokogawa, K. Takeuchi, I. Shimada, J. Am. Chem. Soc. 127 (2005) 12021. K. Matsuyama, S. Natori, J. Biol. Chem. 263 (1988) 17112. K. Takeuchi, H. Takahashi, M. Sugai, H. Iwai, T. Kohno, K. Sekimizu, S. Natori, I. Shimada, J. Biol. Chem. 279 (2004) 4981. K.E. Kadler, D.F. Holmes, J.A. Trotter, J.A. Chapman, Biochem. J. 316 (1996) 1. Z.M. Ruggeri, J. Clin. Invest. 99 (1997) 559. M.A. Cruz, H.B. Yuan, J.R. Lee, R.J. Wise, R.I. Handin, J. Biol. Chem. 270 (1995) 10822. H. Lankhof, M. van Hoeij, M.E. Schiphorst, M. Bracke, Y.P. Wu, M.J. Ijsseldijk, T. Vink, P.G. de Groot, J.J. Sixma, Thromb. Haemost. 75 (1996) 950. J. Bienkowska, M. Cruz, A. Atiemo, R. Handin, R. Liddington, J. Biol. Chem. 272 (1997) 25162. E.G. Huizinga, R.M. vanderPlas, J. Kroon, J.J. Sixma, P. Gros, Structure 5 (1997) 1147. R.M. van der Plas, L. Gomes, J.A. Marquart, T. Vink, J.C. Meijers, P.G. de Groot, J.J. Sixma, E.G. Huizinga, Thromb. Haemost. 84 (2000) 1005. N. Nishida, H. Sumikawa, M. Sakakura, N. Shimba, H. Takahashi, H. Terasawa, E. Suzuki, I. Shimada, Nat. Struct. Biol. 10 (2003) 53. J.M. Chen, C.E. Kung, S.H. Feairheller, E.M. Brown, J. Protein Chem. 10 (1991) 535. J.P. Labrador, V. Azcoitia, J. Tuckermann, C. Lin, E. Olaso, S. Manes, K. Bruckner, J.L. Goergen, G. Lemke, G. Yancopoulos, P. Angel, C. Martinez, R. Klein, EMBO Rep. 2 (2001) 446. E. Olaso, J.P. Labrador, L. Wang, K. Ikeda, F.J. Eng, R. Klein, D.H. Lovett, H.C. Lin, S.L. Friedman, J. Biol. Chem. 277 (2002) 3606. E. Olaso, K. Ikeda, F.J. Eng, L. Xu, L.H. Wang, H.C. Lin, S.L. Friedman, J. Clin. Invest. 108 (2001) 1369. J. Wang, H. Lu, X. Liu, Y. Deng, T. Sun, F. Li, S. Ji, X. Nie, L. Yao, J. Autoimmun. 19 (2002) 161. N. Ferri, N.O. Carragher, E.W. Raines, Am. J. Pathol. 164 (2004) 1575. L. Xu, H. Peng, D. Wu, K. Hu, M.B. Goldring, B.R. Olsen, Y. Li, J. Biol. Chem. 280 (2005) 548. K. Ikeda, L.H. Wang, R. Torres, H. Zhao, E. Olaso, F.J. Eng, P. Labrador, R. Klein, D. Lovett, G.D. Yancopoulos, S.L. Friedman, H.C. Lin, J. Biol. Chem. 277 (2002) 19206. B. Leitinger, J. Biol. Chem. 278 (2003) 16761. B. Leitinger, A. Steplewski, A. Fertala, J. Mol. Biol. 344 (2004) 993. K. Yang, J.H. Kim, H.J. Kim, I.S. Park, I.Y. Kim, B.S. Yang, J. Biol. Chem. 280 (2005) 39058. S.J. Wall, E. Werner, Z. Werb, Y.A. DeClerck, J. Biol. Chem. 280 (2005) 40187. O. Ichikawa, M. Osawa, N. Nishida, N. Goshima, N. Nomura, I. Shimada, EMBO J. 26 (2007) 4168. A.D. Konitsiotis, N. Raynal, D. Bihan, E. Hohenester, R.W. Farndale, B. Leitinger, J. Biol. Chem. 283 (2008) 6861. T.E. Ferrin, C.C. Huang, L.E. Jarvis, R. Langridge, J. Mol. Graph. 6 (1998) 13. C.C. Huang, E.F. Pettersen, T.E. Klein, T.E. Ferrin, R. Langridge, J. Mol. Graph. 9 (1991) 230. L. Lo Conte, C. Chothis, J. Janin, J. Mol. Biol. 285 (1999) 2177. J. Emsley, C.G. Knight, R.W. Farndale, M.J. Barnes, R.C. Liddington, Cell 101 (2000) 47.