Protein Interactions

Analytical Biochemistry 267, 252–259 (1999) Article ID abio.1998.3000, available online at http://www.idealibrary.com on

Hydrogen Exchange/Electrospray Ionization Mass Spectrometry Studies of Structural Features of Proteins and Protein/Protein Interactions Hanno Ehring 1 Department of Structural Chemistry, Pharmacia & Upjohn, S-112 87 Stockholm, Sweden

Received May 28, 1998

The rate at which amide hydrogens located at the peptide backbone in protein/protein complexes undergo hydrogen/deuterium exchange is highly dependent on whether the amide groups participate in binding. Here, a new mass spectrometric method is presented in which this effect is utilized for the characterization of protein/ligand binding sites. The information obtained is which region within the protein participates in binding. The method includes hydrogen/deuterium exchange of receptor and ligand protein amide protons, binding, and back exchange. After this procedure those backbone amide groups that participate in protein binding are protected from back exchange and therefore still deuterated. These regions were then identified by peptic proteolysis, fast microbore high-performance liquid chromatography separation, and electrospray ionization mass spectrometry. The approach has been applied to the investigation of structural features of insulin-like growth factor I (IGF-I) and the interaction of insulin-like growth factor I with IGF-I binding protein 1. The data show that the approach can provide information on the location of the hydrophobic core of IGF-1 and on two regions that are mainly involved in binding to IGF-I binding protein 1. The data are consistent with results obtained with other approaches. The amount of sample required for one experiment is in the subnanomolar range. © 1999 Academic Press

Noncovalent interactions between proteins and other molecules determine many important biological 1

Present address: PyroSequencing AB, Vallongatan 1, S-752 28 Uppsala, Sweden. Fax: 146 18 591 919. E-mail: Hanno.Ehring@ pyrosequencing.com.

252

processes. Nevertheless, the investigation of the structural features of the two binding molecules and the binding sites is still a challenging problem. Traditional techniques involve X-ray crystallography, nuclear magnetic resonance spectroscopy, or calorimetry. Such techniques often require large amounts of material and/or are rather time consuming. Other limitations are the inability to study large proteins and difficulties in producing crystals from protein/ligand systems. Recently, the determination of amide hydrogen exchange by mass spectrometry was introduced as a new tool for protein structure elucidation (1–3). The method is based on the fact that the hydrogen/deuterium (H/D) exchange rate of amide protons located at the peptide backbone depends on whether they are participating in intramolecular hydrogen bonding and on the extent to which they are shielded from the solvent (4). The study of the H/D exchange rates therefore gives information about the noncovalent structure of the protein. A similar approach can also be used to study the binding and protein folding upon binding (5, 6). However, these methods are not capable for determining which regions within the protein are involved in binding or conformation changes. Spatial resolution in the range of 1–10 residues can be obtained by coupling mass spectrometry with proteolytic digestion by pepsin and HPLC separation. Several studies using this method have been published. These include probing the noncovalent structure and conformational changes of apomyoglobin (7), horse heart cytochrome c (8), apo- and holomyoglobin (9), bacterial phosphocarrier protein (10), ferrodoxin (11), hematopoietic cell kinase SH3 domain (12), and Rhodobacter capsulatus cytochrome c 2 (13). The procedure originates from early studies (14, 15) in which tritium and radioactive detection was used instead of deuterium and mass spectrometry. Another option is the study of binding between different pro0003-2697/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.

DEUTERIUM EXCHANGE/MASS SPECTROMETRIC STUDIES OF PROTEIN BINDING

253

EXPERIMENTAL

Materials Tris–HCl, Tris base, and substance P were purchased from Sigma (St. Louis, MO). Ammonium acetate, formic acid, ammonia, and D 2O were provided by Merck (Darmstadt, Germany). Tris(2-carboxyethyl)phosphine (TCEP) 2 and immobilized pepsin was purchased from Pierce (Rockford, IL). Acetonitrile was of HPLC grade from Lab-Scan (Dublin, Ireland). Deionized water was purified in a Milli-Q plus system from Millipore prior to use. Insulin-like growth factor I (IGF-I) and IGF-I binding protein 1 (IGFBP-1) were taken as model proteins. They were provided in-house. H/D Exchange

FIG. 1. Binding site characterization scheme.

teins or between a protein and small molecules. One example described in the literature is the binding of NADH and 2,6-pyridinedicarboxylate to dihydrodipicolinate reductase (16). With this approach differences in hydrogen exchange rates with and without ligands are investigated. Those peptides whose hydrogen exchange slows upon binding represent regions where conformational changes or ligand binding takes place. However, this approach does not allow the characterization of the binding site to the single amino acid level. The purpose of the following project was to develop a method for the study of binding sites based on a similar approach as described above. An important difference is the fact that in our approach all solvent accessible amide protons are exchanged before binding. After binding to the ligand, those amide protons that are accessible even with binding are then back exchanged to hydrogen. All amide protons involved in the binding are protected from isotopic exchange and remain deuterated. The whole procedure is shown schematically in Fig. 1. By the approach described in this report we identify the location of these amide protons by LC– ESI–MS.

For H/D exchange, IGF-I was dissolved in 5 mM Tris base/D 2O and IGFBP-1 was dissolved in Tris–HCl/D 2O at 6°C for a predetermined time. Different buffers were used to improve solubility. Typically, 400 pmol IGF-I and 500 pmol IGFBP-1 were used for one experiment. The deuterium exchange was monitored by MALDI–MS of the whole protein at certain time intervals. This was carried out to determine the appropriate exchange-in time for IGF-I and IGFBP-1. This is the time where the increase in mass slows down considerably, indicating that the majority of those amide protons that are possible candidates for involvement in binding are exchanged to deuterium. For the experiments described below 2.5 h was chosen as H/D exchange time. For back exchange, 5 h incubation time was chosen to guarantee that back exchange is complete for all accessible deuterons. After H/D exchange, the receptor and ligand proteins were mixed and the complex was formed. The deuterated solvent was then removed in a Speed-Vac evaporator. For back exchange the receptor/ligand complex was dissolved in 5 ml 5 mM Tris base/Tris–HCl (1/1, v/v) in H 2O buffer for a predetermined time at 6°C. IGFBP-1 was added in excess to bind all IGF-I molecules present. In addition, high receptor/ligand concentrations (1–10 mg/ml) and low temperatures were used to maximize the rate and extent of binding. The interaction of IGF-I and IGFBP-1 is very strong and longlived with an association constant of K A 5 1.7 [M 21 3 10 29] (17). The approach described here likely works only for strongly binding molecules. In weakly binding 2 Abbreviations used: TCEP, tris(2-carboxyethyl)phosphine; IGF-I, insulin-like growth factor I, IGFBP-1, IGF binding protein 1; TFA, trifluoroacetic acid; CID, collision-induced dissociation; PSD, postsource decay; ESI, electrospray ionization; MALDI, matrix-assisted laser desorption ionization.

254

HANNO EHRING

complexes back exchange of temporarily nonbinding proteins may deteriorate the results. Quenching of the H/D Exchange and Protein/Ligand Complex Separation H/D exchange was quenched by adding a 5 mM ammonium acetate, 0.5% formic acid solution (pH 2.6) in 5 M urea at a ratio of 19:1 and by decreasing the temperature to 0°C. This procedure reduced the H/D exchange time of solvent exposed peptide backbone amide protons by about 4 –5 orders of magnitude to 40 –50 min (half-life time) (18) and made it possible to study the protein exchange by means of liquid chromatography–mass spectrometry. All other exchangeable deuterons were back-exchanged within milliseconds and did not disturb the measurements. The exposure of the protein/ligand complex solution to low H/D exchange conditions at pH 2.6 resulted in a separation of the receptor/ligand complex. Since digestion of IGF-I is much faster than IGFBP-1 digestion it was possible to choose conditions for digestion where IGF-I was digested, whereas IGFBP-1 remained almost intact. If the digestion products of the protein to study are well known, it may be possible to omit the separation step even in receptor/ligand systems where both proteins are digested under similar conditions. This would mean that it is possible to study the binding site of both proteins at the same time. Protein Reduction and Digestion After quenching of the H/D exchange the ligand/ receptor mixture was denatured, reduced, and digested under slow exchange conditions. For this purpose, the solution was mixed with TCEP to a final concentration of 100 mM. TCEP was dissolved in 5 mM ammonium acetate, 0.2% formic acid in H 2 O to a concentration of 1 M. TCEP is a strong reducing agent for fast, quantitative, and specific reduction of disulfides. In contrast to DTT, it also cleaves barely accessible disulfide bridges at room temperature and low pH (19). Pepsin was then used for digestion of the protein because it has maximal activity in the pH 2–3 range, where the H/D exchange rate is lowest. Furthermore, pepsin cleaves the protein at many points yielding small peptides which more closely define the regions of the protein involved in ligand binding than large peptides. In addition, small peptides are more suitable for sequencing by MS/MS. Immobilized pepsin was packed in a steel column (4 3 23 mm, Waters, Milford, MA). The column was placed between the injector and desalting column. Digestion was obtained by stopping the flow for 2 min after injection. The sample was then introduced

FIG. 2. Micro-LC system with precolumn pepsin digestion and desalting/enrichment.

into the HPLC system for desalting, enrichment, and separation. The conditions for the reduction step were optimized with trypsin digests of IGF-I and hGH and with insulin as model compounds. Alternatively to the LC–ESI–MS approach, the digest mixture was measured by MALDI–MS directly. The resulting peptide mixture was diluted in 0.1% TFA and introduced into the mass spectrometer. About 1 pmol receptor protein was used for one MALDI–MS experiment. Reversed-Phase Liquid Chromatography The liquid chromatograph consisted of three pumps (Models LC10 AD and Model LC9) and a UV detector (SPD10A), all from Shimadzu (Kyoto, Japan), one Rheodyne loop injector with 50 ml volume and one six-port valve (Valco, Houston, TX). The UV detector was equipped with a microcell from LC Packings (Amsterdam, The Netherlands). The detector wavelength was set to 220 nm. The analytical column used was a Zorbax SB C18 (5 mm particles, LC Packings, 0.8 3 150 mm, 300 Å). The whole system is displayed in Fig. 2.

DEUTERIUM EXCHANGE/MASS SPECTROMETRIC STUDIES OF PROTEIN BINDING

The mobile phases consisted of: (A) 0.05% TFA, 5% acetonitrile (pH 2.4) and (B) 0.05% TFA, 95% acetonitrile. The protein/reducing agent mixture was injected and first applied on a column packed with immobilized pepsin and then applied on a C18 column (LC Packings, 0.8 3 5 mm, 300 Å) for desalting and enrichment. The buffer contained 5 mM ammonium acetate, 5% acetonitrile, 0.2% formic acid (pH 2.7). The flow rate was 100 ml/min (pump Model LC 9). After switching the six-port valve, the analytical column and the gradient pumps were coupled online and the peptic fragments were eluted and separated at 20 ml/min flow rate. The gradient was 20% B from start, 30% B during 4 min followed by a gradient to 40% B for 10 min and finally to 100% for 6 min. To maintain a reproducible gradient, the pumps were run at 70 ml/min flow rate and split at a 3.5:1 ratio. All components of the HPLC system, from the injector to the analytical column, were submerged in an ice bath. LC–MS and MALDI–MS For MS measurements the UV detector was disconnected to minimize back exchange during analysis. A Micromass Q-TOF mass spectrometer (quadrupole– orthogonal TOF) (Micromass, Manchester, England) was used for LC–ESI–MS measurements. Mass spectra were recorded in the mass range of 300 –1800 Da with a scan speed of 2 s/scan. The source temperature was 70°C. MS/MS measurements were performed with argon as collision gas for low-energy collision-induced dissociation (CID). MALDI–MS measurements were performed with a Tofspec SE mass spectrometer (Micromass) equipped with a delayed extraction unit and post-source decay (PSD) for the registration of fragment ions. Calculation of the Deuterium Labeling Distribution The isotope pattern of a certain peptic fragment is formed by a mixture of ions with different amounts of deuterium incorporation. It is built up from the individual patterns of each labeled component with a shift toward higher values of one mass unit with each additional deuterium atom. The labeled compounds contribute to the appearance of the pattern in accordance with their concentrations in the mixture. Since the sequence and with this the natural isotope pattern of the unlabeled peptide are known, it is possible to calculate the concentration of a peptide with a certain amount of deuterium incorporation and thus the mean value of deuterium incorporation by a measurement of the isotope distribution. The relation can be expressed as a series of linear equations. This equation system was solved with Mathcad PLUS version 6.0. The final

255

results were expressed as the amount of H/D exchange expressed as the mean number of deuterium atoms divided by the number of amide protons located on peptide amide linkages. RESULTS AND DISCUSSION

Back Exchange during Analysis Back exchange during LC–ESI–MS analysis was tested by treating fully deuterated substance P in the same way as the IGF-I sample, which means quenching of H/D exchange by addition of buffer at a ratio of 19:1 (buffer/sample), injection, digestion in the pepsin column, desalting and elution into the mass spectrometer. Substance P was used to make sure that the analyte is fully deuterated. Pepsin digestion of substance P gave two fragments and the intact peptide with retention times of 3 min (899.5 Da, 1–7), 7 min (465.3 Da, 8 –11), and 10 min (1346.7 Da, 1–11). The back exchange was 34% for segment 1–7, 40% for segment 8 –11, and 26% for the whole peptide. In addition to these three peaks there is a fourth one in the chromatogram at 8.4 min corresponding to a peptide at 465.3 Da which is presumably fragment 8 –11 formed by thermal degradation before pepsin digest. The back exchange of this fragment was 14%. This experiment was repeated with a source temperature of 60°C. No decrease in back exchange was observed. The back exchange was 34% for segment 1–7, 41% for segment 8 –11, 29% for the whole peptide, and 13% for the fragment 1–7 formed by thermal degradation which is very similar to the values obtained at 70°C. This result also shows the reproducibility of the method. IGF-1, incubated for several hours in 6 M urea and TCEP, has been investigated as well. Back exchange for these peptic fragments was 40 – 60%. The fact that the deuterium incorporation varied between different peptides is presumably due to side chain effects which can alter the amide hydrogen exchange rate under quenching conditions by as much as tenfold (3). The variation of back exchange between 30 and 60% for different peptides during pepsin digestion and LS–ESI–MS analysis has been reported (16). The retention time differences seem to be of minor importance for back exchange since temperature and pH are well defined during LC analysis. Another finding is that the cleavage seemed to increase back exchange since the whole peptide shows less back exchange than fragment 1–7 and 8 –11. In practice these findings suggest that the measured deuterium levels must be corrected for artifactual isotopic exchange occurring during digestion and HPLC separation if absolute values for deuterium incorporation must be determined. This can be carried out by measuring both nondeuterated and fully deuterated proteins as described by Zhang

256

HANNO EHRING

and Smith (2). A difficulty with this method is the fact that it is difficult to determine whether a protein is fully denatured. For the work described in this report the exact amount of deuterium incubation is of minor importance. The main purpose of the technique described here was to find out whether a certain amide protein is partially deuterated or not. In order to obtain results for all parts of the molecule it is therefore important to reduce the back exchange as much as possible. However, if back exchange during analysis is complete even under low exchange conditions this contribution to protein binding will remain undetected. The results obtained for the other parts of the molecules would not be affected. In this report all values on deuterium incorporation are not corrected for back exchange during analysis. The uncertainty for the determination of the amount of deuterium incorporation is estimated to be 5–10%. IGF-I Digestion with Pepsin As a first step we identified the peptide fragments obtained from pepsin digestion of reduced IGF-I without H/D exchange. For this purpose we measured the peptide molecular mass by LC–MS and MALDI–MS. The fragments were identified by searching for peptides with the measured mass in the sequence of IGF-I using MassLynx NT software. The error in the database search was set to 61 Da even though the error in mass determination is better than 300 ppm. The fragments that gave several hits were identified by MS/MS, MALDI–PSD, and/or CID–ESI–MS. The molecular mass and sequence of all identified peptides are displayed in Table 1. The sequence coverage was complete when LC–ESI–MS was used. However, the most intense fragments contain those which were not involved in disulfide bond binding, especially the region 24 – 48. In case of MALDI–MS the peptide at m/z 5 658 Da could not be detected. In general, the advantage of LC–MS is higher sequence coverage, more sophisticated MS/MS possibilities, and a better tolerance for detergents and reducing agents due to the desalting/enrichment column. However, the method is less sensitive and more time-consuming. MALDI–MS is almost as accurate, very fast, and sensitive but gives less reproducible results as far as sequence coverage is concerned. Therefore, MALDI–MS and MALDI–PSD was used in this study mainly to identify peptic fragments with small amounts of material and for evaluating the best conditions for reduction and digestion, whereas LC–ESI–MS was used to study the IGF-I/IGFBP-1 binding site.

TABLE 1

Sequence and Molecular Weight of Peptides Obtained by Peptic Digestion of IGF-I Molecular weight (Da)

Amino acid sequence

Sequence

657.36 675.29 739.33 816.46 875.32 899.40 988.45 1056.49 1062.46 1147.60 1303.60 1327.68 1909.93 2208.06 2237.11 2366.16 2529.22 2572.17 2572.17

LVDALQ GPETLCG FRSCDL LRRLEM GPETLCGAE FVCGDRGF GPETLCGAEL FNKPTGYGSS FVCGDRGFY CAPLKPASKA LQFVCGDRGFY SRRAPQTGIVDE FNKPTGYGSSSRRAPQTG KPTGYGSSSRRAPQTGIVDEC FNKPTGYGSSSRRAPQTGIVD FNKPTGYGSSSRRAPQTGIVDE YFNKPTGYGSSSRRAPQTGIVDE FNKPTGYGSSSRRAPQTGIVDECC NKPTGYGSSSRRAPQTGIVDECCF

10–15 1–7 49–54 54–59 1–9 16–23 1–10 25–34 16–24 60–70 14–24 35–46 25–42 27–47 25–45 25–46 24–46 25–48 26–49

LC–MS of H/D Exchanged IGF-1 Peptic Fragments without and with Binding to IGFBP-1 In Fig. 3 mass spectra of three peptic fragments of IGF-I under several conditions are displayed. The first row shows the fragment without H/D exchange; the second row is after 2.5 h H/D exchange and incubation with IGFBP-1 followed by 5 h back exchange; the third row is after 2.5 h H/D exchange followed by 5 h back exchange without IGFBP-1 binding; and the fourth row is after 2.5 h H/D exchange without binding and back exchange. The first fragment, segment 1–7 at m/z 5 675 Da shows a high degree of deuterium incorporation (42%) after 2.5 h incubation time. After binding to IGFBP-1 and back exchange, the deuterium incorporation was still 15%, which means that incubation with IGFBP-1 reduces the back exchange rate significantly. The fact that exchange is close to 0% after 2.5 h H/D exchange and 5 h back exchange without binding indicates that back exchange is complete for all amide linkages if IGFBP-1 is not present. The next fragment, segment 54 –59, is an example of a segment with a low amount of H/D exchange both as the IGF-I/IGFBP-1 complex and in the free form. In a separate experiment, this value increased to 46% after complete denaturation before H/D exchange. After IGFBP-1 binding 7% of all amide protons were still deuterated. The third example, segment 25– 42, shows a deuterium incorporation of 57% after H/D exchange and complete backexchange after binding to IGFBP-1. The results are

DEUTERIUM EXCHANGE/MASS SPECTROMETRIC STUDIES OF PROTEIN BINDING

257

FIG. 3. Mass spectra of peptic fragments of IGF-1 measured by ESI–MS without H/D exchange (A), 2.5 h exchange followed by IGFBP-1 binding and 5 h back exchange (B), 2.5 h H/D exchange followed by 5 h back exchange without IGFBP-1 binding (C), and. 2.5 h H/D exchange (D).

summarized in Table 2. The first column shows the molecular mass of the segments and the next one the location within the protein. The third column displays the deuterium incorporation after 2.5 h H/D exchange, the fourth column is after 2.5 h H/D exchange followed by 5 h back exchange without IGFBP-1 binding, and the last column shows the deuterium incorporation after 2.5 h H/D exchange followed by incubation in IGFBP-1 and 5 h back exchange. The results shown in columns 3 and 5 are also displayed in Fig. 4. The amount of H/D exchange without binding, as displayed in the third column in Table 2 and the upper bar diagram in Fig. 4, is a measure for the solvent accessibility and/or the participation of the segment’s amide hydrogens in hydrogen bonding. It can be seen that segments 1–9, 25– 42, and 60 –70 are those protein segments that are not protected from H/D exchange, indicating that major parts of these segments are not involved in hydrogen bonding and are accessible to the solvent. These parts are therefore possible candidates for IGF-I binding. Other segments that are at least

partially accessible to H/D exchange are segments 16 –24 and 49 –54, whereas segments 10 –15 and 54 –59 with H/D exchange degrees of only 11 and 10% seem be deeply buried and/or involved in forming rigid secondary structures. These findings are consistent with the known structure of IGF-1 as shown in Fig. 5. The dotted lines indicate that the regions 54 –59 and 8 –17 together with region 44 – 49 have a-helical structures. They form the hydrophobic core of the protein (20). Column 4 displays the amount of H/D exchange after 2.5 h H/D exchange followed by 5 h back exchange without IGF-I binding to IGFBP-1. It can be seen that the amount of H/D exchange is very close to 0% in all cases which is a proof of complete back exchange under conditions where no binding occurs. For the evaluation of the binding site this column must be compared to column 5 or the lower bar diagram in Fig. 4. These data show that binding to IGFBP-1 reduces back exchange mainly in segments 1–9 and 49 –54 but also in 10 –15, 16 –24, 54 –59 and to a minor extent in 60 –70 but not in 25– 42. The data suggest that parts of the segments

258

HANNO EHRING TABLE 2

Amount of Deuterium Incorporation of IGF-1 Peptic Fragments after 2.5 h H/D Exchange, 2.5 h Exchange, and 5 h Back Exchange without and with IGFBP-1 Incubation

Molecular mass

Segment

658

10–15

676

1–7

876

1–9

740

49–54

817

54–59

1062

16–24

1056

25–34

1147

60–70

1909

25–42

2208

27–47

2237

25–45

2366

25–46

Exchange without back exchange

Exchange and back exchange without IGFBP-1 incubation

Exchange and back exchange with IGFBP-1 incubation

11% 0.55/5 41% 2.0/5 38% 3.4/9 20% 1/5 10% 0.5/5 22% 1.76/8 54% 4.32/8 41% 3.28/8 57% 8.55/15 37% 6.66/18 51% 9.18 62% 11.78

21% 20.05/5 1% 0.05/5 0.3% 0.023/9 0.02% 20.001/5 0.001% 0.0004/5 0.1% 20.01/8 0.025% 0.002/8 0.1% 0.01/8 0.2% 0.024/15 0.1% 0.137/18 20.1% 20.022/18 0.2% 0.035/19

12% 0.6/5 14% 0.7/5 15% 1.35/9 17% 0.85/5 7% 0.35/5 9% 0.72/8 0.9% 0.07/8 5.6% 0.45/8 0.4% 0.06/15 3.3% 0.60/18 5% 0.9/18 4.6% 0.87/19

Note. The upper number gives the mean amount of exchanged amide backbone protons in % and the lower one is the mean number of exchanged divided by the number of exchangeable amide protons.

1–9 and 49 –54 are involved in IGFBP-I binding. Other segments that could be involved in binding are segments 10 –15, 16 –24, 54 –59, and 60 –70. However, since segments 10 –15 and 54 –59 belong to the hydrophobic core of the protein, it is possible that other mechanisms reduce back exchange after binding to IGFBP-1 without direct interaction between IGF-I and IGFBP-1. Those mechanisms could be conformational changes after IGFBP-1 binding which change the hydrogen-bonding patterns giving more rigid structures or by the fact that the IGFBP-1 molecules cover the hydrophobic core and thereby bury parts of the proteins and reduce solvent accessibility. With the method presented it is generally not possible to decide unambiguously whether the reduced back exchange is due to direct interaction of the receptor with the ligand protein or if binding affects the protein structure and thereby prohibits H/D exchange indirectly. It is also important to keep in mind that the absence of deuterium incorporation after binding and back exchange

FIG. 4. Block diagram of the amount of H/D exchange after 2.5 h H/D exchange without binding or back exchange (A) and with 2.5 h H/D exchange, binding to IGFBP-1 and 5 h back exchange (B).

cannot be taken as a proof that this region is not involved in binding at all. It may occur that the backbone amide proton belonging to the amino acid that binds to the ligand is still accessible to H/D exchange. In the literature, Clemmons et al. (21) found that residues 3 and 4 as well as 49 –51 were involved in binding to IGFBP-1. Jansson et al. (22) proposed that residues 1– 4, 49 –51, and 68 –70 were parts of the structural binding surface. This is consistent with the findings obtained in this study. However, the regions 10 –15 and 54 –59 have not been proposed as being involved in IGFBP-1 binding. The reason why deuteration can be seen even for these peptides could be that

FIG. 5. The primary and secondary structure of IGF-I. The disulfide arrangement is indicated by black bars. The three a-helical regions are marked as boxes.

DEUTERIUM EXCHANGE/MASS SPECTROMETRIC STUDIES OF PROTEIN BINDING

binding induces conformational changes as discussed above. In addition to regions 1– 4 and 49 –51, amino acid 41 was identified by NMR spectroscopy as part of the structural binding surface (22). This cannot be deduced from the data presented here. ACKNOWLEDGMENTS The author thanks Kurt Benkestock and Per-Olof Edlund for their advice with the setup of the LC–MS system and Bengt Nore´n for helpful discussions throughout this project.

REFERENCES 1. Stevenson, C. L., Anderegg, R. J., and Borchardt, R. T. (1993) J. Am. Soc. Mass Spectrom. 4, 646 – 651. 2. Zhang, Z., and Smith, D. L. (1993) Protein Sci. 2, 522–531. 3. Smith, D. L. (1997) J. Mass Spectrom. 32, 135–146. 4. Milne, J. S, Mayne, L, Roder, H., Wand, A. J.. and Englander, S. W. (1998) Protein Sci. 7, 739 –745. 5. Robinson, C., Gross, M., Eyles, S. J., Ewbank, J. J., Mayhew, M., Hartl, F. U., Dobson, C. M., and Radford, S. E. (1994) Nature 372, 646 – 651. 6. Miranker, A., Robinson, C. V., Radford, S. E., Aplin, R. T., and Dobson, C. M. (1993) Science 262, 896 –900. 7. Wang, F., and Tang, X. (1996) Biochemistry 35, 4069 – 4078. 8. Dharmasiri, K., and Smith, D. L. (1996) Anal. Chem. 68, 2340 – 2344.

259

9. Johnson, R. S., and Walsh, K. A. (1994) Protein Sci. 3, 2411– 2418. 10. Johnson, R. S. (1996) J. Am. Soc. Mass Spectrom. 7, 515–521. 11. Remigy, H., Jaquinod, M., Petillot, Gagnon, Y. J., Cheng, H., Xia, B., Markley, J. L., Hurley, J. K., Tollin, G., and Forest, E. (1997) J. Protein Chem. 16 (5), 527–532. 12. Engen, J. R., Smithgall, T. E., Gmeiner, W. H., and Smith, D. L., (1997) Biochemistry 36(47), 14384 –14391. 13. Jaquinod, M., Halgand, F., Caffrey, M., Saint-Pierre, C., Gagnon, J., Fitch, J., Cusanovich, M., and Forest, E. (1995) Rapid Commun. Mass Spectrom. 9, 1135–1140. 14. Rosa, J. J., and Richards, R. M. (1981) J. Mol. Biol. 145, 835– 851. 15. Englander, J. J., Rogero, J. R., and Englander, S. W. (1985) Anal. Biochem. 147, 234 –244. 16. Wang, F., Blanchard, J. S., and Tang, X. (1997) Biochemistry 36 (13), 3755–3759. 17. Jansson, M., Dixelius J., Uhlen, M., and Nilsson, B. O. (1997) FEBS Lett 416 (3), 259 –264. 18. Bai, Y., Milne, J. S., Mayne, L., and Englander, S: W. (1993) Proteins 17 (75), 75– 86. 19. Han, J. C., and Han, G. Y., (1994) Anal. Biochem. 220, 5–10. 20. Cooke, R. M., Harvey, T. S., and Campbell, I. D. (1991) Biochemistry 30, 5484 –5491. 21. Clemmons, D. R., Dehoff, M. L., Bayne, W. H., and Cascieri, M. A. (1992) Endocrinology 131, 890 – 895. 22. Jansson, M. (1997) Dissertation, Royal Institute of Technology, Stockholm, Sweden.