Characterization of the proto-oncogenic and mutant forms of the transmembrane region of Neu in micelles.

We have investigated peptides corresponding to the complete transmembrane region of both proto-oncogenic (Val(664)) and mutant (Glu(664)) forms of the receptor Neu in detergent micelles by NMR and CD spectroscopy. Both forms of the peptide appear to adopt similar levels of helicity and dimeric interactions based on the analysis of CD spectra and nuclear Overhauser effect connectivity profiles. There are considerable differences in the chemical shifts of amide and, to a lesser extent, CHalpha resonances between the two forms of the peptides, and these differences are most pronounced in residues upstream of the mutation site and close to the N terminus of the transmembrane domain. Similarly, there are substantial differences in the amide hydrogen-deuterium exchange rates for residues close to and upstream of the mutation site; amide protons in this region of the protooncogenic peptide are much more resistant to exchange than those in the mutant form. In both molecules, residues downstream of the mutation site exhibit slow exchange. We therefore demonstrate that, although transmembrane Neu peptides exhibit similar levels of secondary structure when dispersed in detergent, there are detectable differences in their adopted micellar states that may provide insight into the dimer-promoting ability of the polar transforming mutation.

The catalytic activation of members of the receptor tyrosine kinase (RTK) 1 family is generally initiated by ligand binding to the extracellular domain of a monovalent receptor that promotes dimerization, a process mediated by residues in both the extracellular and transmembrane (TM) domains. Receptor dimerization enables the trans-phosphorylation of the intracellular domain of the partner receptor and the recruitment of signaling proteins, leading to the appropriate downstream cellular response. In Neu, the rat homologue of ErbB2 (a member of the epidermal growth factor receptor subfamily), a single oncogenic mutation (Val 664 3 Glu) within the TM domain enhances receptor dimerization, leading to constitutive activation (1)(2)(3). Substitution of the mutant TM domain from Neu into other RTKs increases catalytic activity and promotes cellular transformation (4 -6), demonstrating that there is a common role played by this segment in the activation process. The TM domain of Neu, along with its associated transforming mutation, has become a useful paradigm with which to clarify the role played by the single helical domain in the activation of RTKs.
Recent studies have demonstrated that the presence of polar residues within TM domains can, in principle, promote helixhelix association by providing an additional interhelical stabilizing force acting as both an H-bond donor and acceptor (7)(8)(9). The transforming mutation in the receptor Neu occurs within a five-residue dimerization motif that has been identified in the majority of RTKs (10). The local structure, dictated by the primary sequence in the motif, is critical for proper receptor dimerization and activation; replacing residues adjacent to the transforming mutation site with glutamate does not elicit the same behavior (2,11). Furthermore, upon activation there appears to be important coupling between the kinase domain in the intracellular region of the receptor and the contact point of the helices within the dimer structure (12,13). Thus, the harmful transforming property of the mutation arises not only because it promotes helix-helix association but because it occurs within the context of a dimerization motif.
Using peptides corresponding to the TM region of mutant Neu reconstituted in lipid bilayers, Smith and colleagues (14) have characterized the dimeric structure at residues close to the mutation site. They determined that the side chain carboxyl group of Glu 664 is likely protonated and is a participant in an interhelical hydrogen bond (14). Solid state NMR measurements indicated that there is close association between the side chain carboxyl group of Glu 664 and the ␣-carbon atom of Gly 665 of the partner helix (15). These measurements led to a proposed dimer model that was symmetric and possessed a right-handed crossing angle.
The elucidation of the structure of membrane proteins is a significant challenge because of the technical difficulties associated with their study using techniques such as NMR spectroscopy and x-ray crystallography. It is for this reason that a complete structural picture of the TM domain of Neu in a lipid environment remains elusive. Several lines of evidence suggest that structural details not necessarily confined to the mutation site in the TM domain may offer insight into the process of receptor activation. Molecular dynamics studies of ErbB2/Neu TM peptides, performed in vacuo, have suggested the possibility that portions of this domain may readily adopt -helix turns capable of reorienting residues found at the dimer interface (16 -18). Solvent structures of both proto-oncogenic and mutant forms of the TM domain have displayed helical distortions downstream of the mutation site (19,20). Therefore, dimer models constructed using canonical ␣-helices may be overly simplistic. Wide line 2 H-NMR spectral differences downstream of the mutation site have been observed between monomeric forms of proto-oncogenic and mutant Neu/ErbB2 peptides dispersed in bilayers (21,22). Also, a recent theoretical study has raised the possibility that the TM domain of the ErbB2 receptor may be capable of adopting two stable dimeric conformations that are interchangeable through a 120°rotation of the individual helices (23). Finally, it is still unclear whether the dimer-promoting ability of the transforming polar mutation in this large 185-kDa receptor is solely a function of the added H-bonding capability resulting from the presence of a glutamate side chain within the TM domain or whether there are additional structural or thermodynamic consequences resulting from its presence that contribute to enhanced receptor oligomerization.
As an alternative to studying membrane proteins in their natural bilayer environment, detergent micelles are routinely employed to solubilize and characterize membrane-anchored proteins. Because of their relatively small size range, protein micellar complexes are amenable to investigation using standard solution state NMR techniques. In this report, we present a structural study of both proto-oncogenic and mutant peptides corresponding to the complete TM domain of Neu dispersed in detergent micelles. We have characterized the micelle-adopted states of the peptides using CD and NMR spectroscopy, and we present the banding pattern observed upon running them on SDS-polyacrylamide gels. We demonstrate that both forms of the peptide appear to form helical structures throughout the length of the putative TM region and that they adopt similar levels of interhelical interaction within the micelle environment. However, the transforming mutation causes an appreciable difference in the amide proton exchange rates in residues close to the mutation site and near the N-terminal juxtamembrane region.

EXPERIMENTAL PROCEDURES
Synthesis of Peptides-A hydrophobicity profile of Neu suggests that its TM domain comprises 26 amino acids. The membrane flanking regions contain a proline residue at the N terminus and several basic residues at the C terminus. 36-residue peptides incorporating the entire putative TM domain and several residues from the flanking juxtamembrane regions of both proto-oncogenic and mutant Neu were synthesized using the tBOC strategy and purified by reversed-phase chromatography. The purity of the peptides, judged to be Ͼ90%, was confirmed by mass spectrometry.
The sequence of the proto-oncogenic peptide, corresponding to the position of the amino acids within the Neu receptor, is Q 651 RASPVTF-IIATVV 664 GVLLFLILVVVVGILIKRRRQK 686 . The mutant peptide has an identical sequence, except at position 664, where valine is replaced with glutamic acid. Within our peptides, the mutation site corresponds to residue 14. Heteronuclear NMR experiments were performed with 15 N incorporated at six sites within the peptide backbone of both peptides, namely Val 6 , Ala 11 , Gly 15 , Phe 19 , Leu 22 , and Gly 27 .
CD Spectroscopy-Spectra were acquired on an AVIV Model 215 instrument (AVIV Associates, Lakewood, NJ), using a 1-mm path length quartz cell at 25°C in trifluoroethanol (TFE) and in SDS and dodecylphosphocholine (DPC) micelles. The peptide concentration was ϳ 25 M under all conditions. For samples in which peptides were dispersed in micelles, the detergent concentration was 100 mM. Helical content was estimated using either a deconvolution algorithm with the aid of the program CDNN (24) or a method based on the molar ellipticity value at 222 nm ([⌰] 222 ) as described by Chang et al. (25).
NMR Spectroscopy-Approximately 3 mg (ϳ1.5 mM) of purified peptide was solubilized in 100 mM d 38 -DPC (Cambridge Isotope Laboratories) at a solution pH of 4. To make the 1 H assignments in this detergent, standard two-dimensional homonuclear correlation spectra were acquired at 50°C on a Bruker Avance spectrometer operating at 600 MHz. Proton chemical shifts were referenced to the internal standard tetramethylsilane. Nuclear Overhauser enhancement (NOESY) spectra with mixing times of 200, 250, and 300 ms were acquired using a presaturation pulse during the relaxation delay to suppress the H 2 O peak and a time-proportional phase increment (TPPI) for quadrature detection. We attempted to make the assignment of as many scalar coupled networks as possible with TOCSY spectra acquired using a DIPSI-2 pulse scheme (26), isotropic mixing times of 30, 50, and 75 ms, solvent suppression using WATERGATE (27), and an echo-antiecho phase cycle for quadrature detection. All two-dimensional homonuclear spectra were acquired with 2048 points in the direct dimension, 256 -512 points in the indirect dimension, and 128 -256 scans per increment. The resolution obtained from the homonuclear spectra in detergent was not sufficient to enable the assignment of all of the backbone 1 H resonances in either form of the peptide. To help resolve some of the ambiguities, we synthesized these molecules with 15 N labels incorporated at six positions (see above). To assign the 1 H and 15 N resonances at these six sites, we performed standard 1 H-15 N heteronuclear single quantum coherence (HSQC) spectra (28), 15 N-filtered NOESY spectra with mixing times of 250 and 300 ms (29), and 1 H-15 N TOCSY-HSQC spectra with a mixing time of 75 ms. Solvent suppression was achieved using a presaturation pulse for the 15 N-filtered NOESY spectra and water flip-back pulses for the 1 H-15 N HSQC and TOCSY-HSQC spectra. Once the 1 H assignments at the labeled sites were made, we reanalyzed the two-dimensional homonuclear spectra and were able to assign the majority of the backbone and CH␤ resonances within the helical portions of both peptides. All NMR data were processed using software Felix97 (Biosym) on an O2 work station (Silicon Graphics, Inc.). One final procedural note should be added regarding sample preparation for NMR experiments, i.e. when solubilizing these peptides in detergent, the sample was not subjected to a vortex but was gently resuspended using a pipette. We noticed that subjecting the samples to a vortex led to the loss of NMR signal, possibly because this promoted the aggregation of the peptides.
Amide Exchange Experiments-Hexa-( 15 N)-labeled proto-oncogenic and mutant peptides (ϳ1.5 mM) were dispersed in 100 mM DPC at pH 4. The peptide/micelle solution was lyophilized and redissolved in D 2 O. Amide exchange rates were determined from the rate of decay in successive 1 H-15 N HSQC spectra acquired continuously over a 24-h time period at 40°C. Roughly 20 min elapsed between redissolution in D 2 O and the acquisition of the first spectrum. Each time point took ϳ 60 and ϳ 75 min for the proto-oncogenic and mutant peptides, respectively. The decay constants were determined by integrating cross-peak volumes and fitting the data to a single exponential decay. The curve fitting was performed using the program Origin (version 6.1, Origin Lab Corporation).
Gel Electrophoresis-Neu peptides were dissolved and incubated in standard loading buffer (30) at 42°C for 30 min and run on a 16.5% Tris-tricine gel (31). The gel was stained with Coomassie Brilliant Blue.

RESULTS AND DISCUSSION
There have been three reported NMR studies on peptides corresponding to the TM domain of Neu in solution, all in TFE (19,20,32). This is a popular membrane mimetic solvent used to characterize the secondary structure of TM protein fragments; however, because tertiary interactions are generally disrupted, this solvent could not be used to investigate, for instance, the dimeric structure of the peptides. The 1 H-NMRderived monomeric structures of proto-oncogenic and mutant forms of the Neu peptides were found to be very similar in TFE (20,32). This contradicts an early theoretical investigation, which predicted that the favored conformation of the monomeric form of the proto-oncogenic receptor would be kinked at the mutation site, whereas the mutant form would adopt a canonical ␣-helix structure (33,34). Solid-state NMR studies of Neu peptides in bilayers have also demonstrated that both forms of the receptor are likely helical at the mutation site, although these measurements were made on what were presumed to be dimeric forms of the peptides (15).
The heterogeneous nature of a micellar system reflects that of a lipid bilayer more accurately than do organic solvents and, consequently, is generally viewed as a more suitable membrane mimetic environment. Stable tertiary interactions can form in micelles, which enables the potential characterization of interhelical interactions between separate TM domains (35), and, in cases where the protein/micelle complex remains relatively small, standard solution state NMR techniques can be applied to investigate both protein structure and dynamics. Several investigations of small membrane-associated proteins and TM fragments from larger proteins have been reported in micelles (reviewed in Ref. 36). The effectiveness of micellar systems for NMR studies of membrane proteins is often hindered, however, by poor resolution, spectra that are not generally reproducible, and peptide aggregation (reviewed in Ref. 37).
CD Spectroscopy-The CD spectra of proto-oncogenic and mutant Neu peptides dispersed in SDS and DPC micelles are presented in Fig. 1; for comparative purposes, spectra obtained in the solvent TFE are included in the figure as well. In all three solution conditions, both forms of the peptide are predominantly ␣-helical, as was predicted by using both a deconvolution algorithm (24) and the value of [⌰] 222 (25). The predicted ␣-helical percentages are higher when a deconvolution algorithm was used than when the predictions were based on [⌰] 222 ; helical estimates for the different solution conditions studied range from 75 to 85% when obtained using the former method versus 60 to 65% when using the latter. The NMR spectral resolution of both peptides in micelles was significantly improved at pH values close to 4, so we obtained CD profiles in micelles at both neutral and acidic pH. Within the error associated with the technique, the ␣-helical content of the peptides does not appear to change between pH 4 and 7.
The degree of interaction between two helical peptides can be estimated by using CD spectroscopy based on the [⌰] 222 : [⌰] 208 ratio; a value of 1 is expected for helices that are completely associated, whereas monomeric helices exhibit a ratio of 0.8 (38). An examination of the CD profiles reveals that there is a lower molar ellipticity at 208 nm in TFE than in micelles, which produces a different [⌰] 222 : [⌰] 208 ratio in the two environments. The ratio is very close to 0.8 for both peptides in TFE, which is consistent with the notion that they exist predominantly as monomers in this solvent. In micelles, the [⌰] 222 : [⌰] 208 ratio ranges between 0.85 and 0.9 for both forms of the peptide. This finding indicates that, within this environment, there is some level of association between the peptides. A comparison of the ratios obtained in micelles for the two peptides does not indicate that one form of the TM domain is more predisposed to dimeric interactions than the other.
NMR Spectroscopy-Our analysis of the CD profiles suggests that both Neu peptides display similar levels of helicity and dimeric interactions in SDS and DPC micelles. We initially attempted to assign resonances to the Neu peptides dispersed in SDS micelles and, for this purpose, acquired spectra over a range of temperatures and solution pHs. Unfortunately, we were unable to make many assignments in this detergent, in particular, for residues close to the N terminus of the peptides where, in some samples, there appeared to be a lack of detectable signals for several resonances. For this reason, we attempted to characterize the peptides in DPC micelles. In this detergent we were able to obtain better spectral resolution with no apparent loss of signal from residues in the N-terminal region of the molecules. Spectra with the best resolution were acquired at relatively high temperatures and at low solution pHs. This has also been the case for studies of other membranebound proteins in detergent micelles (37). Even with the improved spectra obtained in DPC, there was significant overlap of peaks in the crucial amide fingerprint region, and we were not able to fully assign all of the backbone 1 H resonances in either form of the peptide using standard two-dimensional homonuclear correlation spectra. To help in this process, we synthesized peptides with 15 N amide labels at Val-6, Ala-11, Gly-15, Phe-19, Leu-22, and Gly-27 and used the proton assignments obtained from 15 N-resolved spectra as points of departure from which to proceed with the sequential assignment of the other resonances in the peptide. In the final analysis, we were not able to unambiguously assign several resonances at the ends of the peptides. However, these residues are not believed to be found within the helical portion of the TM domain in the native protein. In general, the spectra acquired in DPC were fairly reproducible, although there was some variability in the chemical shift of amide resonances (up to 0.1 ppm), even when an effort was made to prepare samples in a consistent manner. Once dispersed in DPC, the peptides were fairly stable, with little loss of signal detected in samples that were up to a month old.
With respect to the chemical shift of amide resonances, spectra obtained for the proto-oncogenic form of the peptide in DPC were noticeably different from those obtained with the mutant form. This is illustrated in 1 H-15 N HSQC spectra of the hexa-( 15 N)-labeled peptides presented in Fig. 2. Once again, for comparative purposes we have presented spectra acquired in the solvent TFE as well, where the peptides adopt very similar structures and display similar chemical shifts in this homogenous environment (20). This trend is also observed in comparing the HSQC spectra of the peptides in TFE, where there are only small differences in the position of the cross-peaks in either the proton or the nitrogen dimension (Fig. 2, A and B). In contrast, there is a significant variation in the position of the cross-peaks in the HSQC spectra for the two peptides dispersed in DPC micelles (Fig. 2, C and D). This difference is most pronounced for residues close to the N-terminal region of the peptide. In Fig. 3, the NH-CH␣ region from NOESY spectra of Neu peptides is presented, with arrows drawn to show the sequential connectivities between the NH and CH␣ resonances in the peptides. As noted above, there is considerable overlap of cross-peaks in these spectra, which led to difficulty in making several resonance assignments. Although there are differences in the chemical shifts of amide protons throughout the entire length of the molecules, a noticeable trend is observed in comparing the spectra of proto-oncogenic and mutant Neu peptides in DPC; the amide peak dispersion pattern is similar for residues between Phe-19 and Ile-30 but is significantly different between Val-6 and Thr-12. Because of the change in primary sequence, the position of amide resonances close to the mutation site would be expected to exhibit variability. The chemical shift of an amide proton is very sensitive to its chemical environment (39) and is dictated by both the structural element it is found in and its relative solvent exposure. Our spectra clearly demonstrate that the structural environment of peptides within the micelle is different, and this difference is most pronounced in the N-terminal region of the peptide.
The optical CD data not only indicate that the two forms of TM Neu peptides adopt similar levels of helicity in detergent micelles, but that these levels are similar to those observed in TFE, where it was determined previously that the adopted ␣-helical structure incorporates residues from Val-6 through to Ile-30 (20). Even though there are significant differences in the chemical shifts of amide protons in DPC, the NMR data are consistent with the formation of a helical structure in residues 6 though 30 in both forms of the peptide. An examination of the CH␣ chemical shifts for these residues indicates they are uniformly upfield of random coil values, which is a strong indication of ␣-helix formation (39). The CH␣ chemical shift deviations from random coil values (CSDs) for residues 6 through 30 are presented as a histogram in Fig. 4A for both peptides. Once again, it can be seen in this figure that, for the residues in the C-terminal region of the peptide, the CH␣ CSDs are almost identical. There are observable differences in the histogram, however, for residues close to the N terminus (e.g. Thr-9, Ala-11, and Thr-12), and, where differences exist, the CH␣ CSDs are more negative for residues found in the proto-oncogenic peptide. This is potentially indicative of a more rigid helical structure in this region for this form of the molecule. It should be noted that, among the residues that we were able to assign at the ends of the peptides, the CH␣ shifts are very close to random coil values (data not shown).
The NOE connectivity profile is presented for both peptides in Fig. 4B. Even in light of the considerable overlap of peaks in the NOESY spectra, which prevented the assignment of several peaks, the pattern for both peptides is typical of an ␣-helix, with several H␣ i -NH iϩ3 , H␣ i -NH iϩ4 , and H␤ i -NH iϩ1 contacts for residues between 6 and 30. Curiously, we observed no NH i -NH iϩ1 contacts in the mutant form of the peptide in the N-terminal region, whereas several were observed in the protooncogenic peptide. As was noted earlier concerning differences in the CH␣ CSDs for residues in this region, this provides evidence that, although both the mutant and proto-oncogenic peptides are helical, the structure formed by the mutant peptide may exhibit more flexibility than that adopted by the proto-oncogenic peptide.
Amide Exchange Experiments-While initially attempting to obtain resonance assignments of Neu peptides in SDS micelles, we solubilized the peptide/micelle complex in D 2 O. This procedure helps the assignment process by simplifying the spectra in the amide fingerprint region, because the signal from amide protons close to the ends of the molecule will disappear rapidly since they are generally unstructured and/or solvent exposed. We noticed that, among the residues that we were able to assign in this detergent, those located at the N terminus of the peptide appeared to exchange very rapidly, whereas residues situated at the C terminus exhibited almost no loss of proton signal over several hours (data not shown).
To further investigate the exchange properties of the peptides in micelles, we dissolved the hexa-( 15 N)-labeled forms of both the proto-oncogenic and mutant Neu peptides dispersed in DPC micelles in D 2 O (pH 4, 40°C) and monitored the loss of amide signal intensity in successive 1 H-15 N HSQC spectra over a 24 h time period. Representative spectra for both peptides acquired at various time points following the initiation of exchange are presented in Fig. 5, and representative decay curves are presented in Fig. 6. Confirming what was seen in SDS, probes situated in the C-terminal half of both peptides (i.e. Phe-19, Leu-22, and Gly-27) display almost no visible exchange over the course of the experiment. After 24 h in D 2 O, we performed a two-dimensional 1 H-NOESY spectrum on the mutant peptide to verify whether the exchange trend observed with these three probes in the C-terminal half of the peptides was consistent for all of the residues in this region. As was observed at the labeled sites, we saw strong amide signals for all of the residues between Phe-19 and Ile-30, whereas no signals could be detected for any of the other residues in the peptide.
There is a considerable difference in the amide exchange behavior between proto-oncogenic and mutant forms of the peptide in two of the three probes (Ala-11 and Gly-15) in the Nterminal half; Val-6 exchanged within the dead-time of the experiment in both cases (i.e. rates of Ͻ20 min). In the protooncogenic peptide, the amide signal from Ala-11 exhibits a decay constant of ϳ 840 min, whereas Gly-15 exhibits almost no signal decay over the duration of the experiment and exchanges as slowly as probes in the C-terminal half of the peptide (Fig. 6A). In the mutant form of the peptide, the amide signals at Ala-11 and Gly-15 decay much faster than in the proto-oncogenic form, with decay constants of ϳ66 min and ϳ830 min, respectively (Fig. 6B). There are two interesting features regarding the amide exchange behavior of the peptides in detergent micelles. The first is the extremely slow rate of exchange exhibited by residues in the C-terminal region of both peptides. This is indicative of the formation of a very stable protein-micellar complex in this region of the peptide. The second interesting feature concerns the effect of the polar mutation on the relative exchange rates seen in both peptides. Clearly, the exchange rate at Gly-15, which is directly adjacent to the mutation site, is faster in the mutant form of the peptide. What is most intriguing is that, even though Phe-19 and Ala-11 are both close to the mutation site, the mutation has an effect only on the rate of exchange at Ala-11; Phe-19 exchanges very slowly in both forms of the peptide (Fig. 6).
Amide exchange measurements have been used to investigate the structure and dynamics of water-soluble proteins for several years (reviewed in Refs. 40 and 41). This technique has been used more recently to characterize the dynamics of several membrane-associated proteins in detergents, primarily to deduce which residues are likely to rest near the membrane surface versus those that will penetrate the hydrophobic core (42)(43)(44)(45)(46)(47). An exchange at a given amide in soluble proteins is inhibited due to hydrogen bonding and/or its burial within the protein hydrophobic core. Similarly, within a micelle environment the protection from exchange for an amide proton found in a helix could be the result of its participation in a backbone hydrogen bond and/or because of its burial within the hydrophobic micellar core. It is impossible to clearly distinguish whether the difference in exchange rates at Gly-15 and Ala-11 is a result of helical destabilization or because of a difference in solvent exposure. Based on studies of other peptides in solution, the protection from exchange afforded by participation in a backbone hydro- gen bond is on the order of minutes to hours, whereas residues buried deep within the hydrophobic core of a protein can be protected from exchange for up to weeks at a time. Because our NMR and CD data indicate that both peptides are helical in the N-terminal region, this tends to suggest that there is a difference in the level of solvent exposure at these sites. It is possible that, in the mutant form of the peptide, this portion of the helix is drawn close to the surface of the micelle in order to solvate the polar glutamate side chain. This possibility remains an interesting area open to further investigation.
The Use of Micelles to Investigate the Impact of Polar Residues within a TM Domain-Because of its ability to promote cellular transformation, the identification of the Val 664 3 Glu mutation within the transmembrane region of Neu has generated tremendous interest in the mechanism by which the presence of this residue promotes receptor dimerization.
Polar mutations within the TM region of the cystic fibrosis transmembrane conductance regulator (CFTR) have also been implicated in some forms of cystic fibrosis (48). Micellar systems are frequently used to investigate membrane protein structure and dynamics, and it is important to establish whether these systems offer a reliable method with which to mimic the effect exerted by polar residues within a bilayer environment.
Many investigators have used SDS-PAGE analysis to quantify the level of interaction in TM domains (49 -51), and this method was used in part to demonstrate the ability of polar mutations to drive helix-helix association in a small membrane-soluble peptide (7). However, through extensive work with the protein glycophorin A, Engelman and co-workers have argued that the effects of polar mutations, specifically their ability to promote interhelical association, may not be properly reflected in micellar systems (49,50,52,53). They have theo-  (40°C, pH 4). Following lyophilization and dissolution in D 2 O as amide protons exchange for solvent deuterons, the cross-peak intensity will decrease over time. Spectra acquired roughly 1, 5, and 24 h after the initiation of hydrogen-deuterium exchange are presented for the proto-oncogenic peptide in panels B, C, and D, respectively, and for the mutant peptide in panels F (ϳ75 min), G (5 h), and H (24 h).
FIG. 6. Amide proton signal decay in Neu peptides in DPC micelles. The proton signal intensity is displayed as a function of time for Ala-11 (A11; triangles), Gly-15 (G15; squares), and Phe-19 (F19; circles) in proto-oncogenic (A) and mutant (B) peptides in DPC micelles after dissolution in D 2 O. Insets show the decay rates for those residues where the data could be fitted to a single-exponential decay. Several amides exchange so slowly that their exchange profile could not be accurately fitted to an exponential decay curve (i.e. residues Gly-15, Phe-19, Leu-22, and Gly-27 in the proto-oncogenic peptide, and Phe-19, Leu-22, and Gly-27 in the mutant peptide). The amide signal at Val-6 exchanged within the dead-time of the experiment (ϳ 20 min). rized that, in a micelle environment, polar residues could promote a break in the helix structure in order to expose the side chain polar group to the external aqueous environment, thereby inhibiting helix-helix association.
The dimerization-promoting ability of the Val 664 3 Glu transformation is not reflected in the migration pattern of TM Neu peptides on SDS-polyacrylamide gels (Fig. 7). The major band for the proto-oncogenic peptide runs at a position corresponding to ϳ1.5ϫ its molecular mass, whereas the mutant form migrates as a major band at a position corresponding to ϳ1.0ϫ its mass. (We have noted previously that concentrating the proto-oncogenic peptide in TFE prior to incubation in loading buffer seems to promote aggregation (20)). This banding pattern could be rationalized as a rapidly migrating dimer species for proto-oncogenic Neu versus a monomeric species for the mutant peptide; however, the subtle difference in the migration pattern may also be because the overall shape adopted by the peptide/micelle complex is different in the two cases, thereby causing their migration rates to change. With respect to the accommodation of the polar residue within the micellar structure, our results indicate that Neu peptides adopt a helical structure throughout the length of the putative TM domain. The effect of the mutation, which appears to be propagated in residues upstream of the mutation site, leads to different amide exchange properties and chemical shift profiles. Whether this is a result of helical destabilization, or a change in the positioning of detergent molecules with respect to the peptide, or a complex combination of both factors, is unclear.
It is difficult to make conclusions about the behavior of membrane proteins in their natural environment based on data acquired in solution, such as in detergent micelles or organic solvents. Our investigation of the peptides in micelles is noteworthy in light of molecular dynamics simulations that we have performed on monomeric forms of the TM domain of proto-oncogenic and mutant Neu in explicit, fully hydrated dimyristoylphosphatidylcholine bilayers. 2 Over the course of the mutant peptide trajectory, the molecule maintained a helical structure, but water molecules were found to penetrate the hydrophobic region of the bilayer at the N-terminal face in order to hydrate the carboxyl side chain of Glu 664 . In theory, dimerization of the mutant receptor could be promoted to minimize the number of water molecules required for hydration of the polar group and the energetic cost associated with water penetration into the bilayer.