Nonergodic and conformational control of the electron capture dissociation of protein cations

Abstract

Electron capture dissociation (ECD) MS is proving to be unusually valuable for “top down” protein sequencing and identification/localization of posttranslational modifications, because the ECD product ions can represent cleavages between most of a protein's amino acids. As proposed, this unusual reactivity results from immediate local utilization, before randomization, of much of the relatively large (≈6 eV) energy from the electron reaction with the multiply charged protein ion, minimizing the effect of differences in the backbone bond dissociation energies. However, others conclude that e^– capture produces a labile free radical species for which backbone cleavage is the lowest energy reaction. Supporting the nonergodic mechanism, ECD of ubiquitin (M + 12H)¹²⁺ ions also yields thermalized radical (M + 12H)^11+· ions that instead lose H· when activated. Also, the ECD spectrum of ubiquitin (M + 13H)¹³⁺ ions is nearly unchanged by heating from 25°C to 125°C, demonstrating that this increase in thermal energy is small compared to the energy driving the reaction. These results support initial capture of the electron in a long-lived high-n Rydberg state, followed by internal conversion to the product valence state at an energy well above the dissociation barriers. The instantaneous conformation of the valence state is critical, with the observed products supporting an α-helical structure in which the protonated side chain of each basic residue is intercalated to hydrogen-bond to as many as three amide carbonyls. Activation (e.g., heat, collisions, lowered charge) can disrupt this conformation to allow additional H⁺-side-chain interactions and provide more complete sequence coverage.

Electron capture dissociation (ECD) mass spectra provide far more extensive characterization of the primary structure of proteins, such as posttranslational modifications, than conventional methods for the dissociation of multiply charged protein cations (1–6). ECD products represent single cleavages between an unusually high proportion of the backbone amino acids (7, 8) (Fig. 1). To explain how hundreds of such reactions are apparently competitive, we have postulated (1–4) that ECD is a nonergodic process, a unimolecular dissociation that occurs before the excitation energy is randomized (9–13). Laser techniques can make such studies in vibrational time periods (11), although energy randomization can require 200 ps (12). Nonergodic dissociation has been observed only for small molecules with a relatively few internal degrees of freedom (9–11); the largest ion found to undergo nonergodic dissociation is the enolic acetone cation, (12–15). Thus it surprising is not that substantial doubts have been expressed concerning our nonergodicity claim for protein ECD, instead presenting evidence that backbone cleavage of peptide ions from e^– capture is just a free radical reaction of unusually low activation energy (16–21). Here we describe additional experiments that provide evidence for the high energy required and for the resulting conformational control of reaction sites that rationalize additional conflicting predictions (22, 23).

Fig. 1.

A single plasma ECD spectrum of carbonic anhydrase defining 183 of 250 possible interresidue cleavages but showing only those resulting in 115 c and 43 z· products (also formed: 15 a·,28 y, and 27 b ions) (8). Positions of the 19 prolines are noted, as ECD cannot form c or z· ions by cleavage on the N-terminal side of Pro. Filled bars, c ions; open bars, z· ions.

For the “top down” MS characterization of the posttranslational modifications of a protein in a mixture (24), its molecular ions formed by electrospray ionization (25) are isolated from those of other proteins and dissociated (tandem MS). Products from a single backbone bond cleavage are most informative; if two such products represent dissociation on either side of an amino acid, their masses will indicate the amino acid identity and its posttranslational modifications. For this, ECD is unusual in its extensive and near-exclusive production of N- and C-terminal fragment ions. For bovine carbonic anhydrase (29 kDa), such terminal tandem MS products have been obtained representing cleavages at 250 of 258 interresidue sites (7), with 183 such cleavages in a single spectrum (Fig. 1) (8). In contrast, conventional energetic fragmentation methods such as collisionally activated dissociation (26) and IR multiphoton dissociation (IRMPD) (27) form far fewer terminal products; those for carbonic anhydrase represent cleavages at only 15–30 of the 258 interresidue sites, presumably because of their large differences in bond dissociation energies (24). For the contrasting similarities in ECD bond dissociation probabilities, we have postulated that this results from the highly exothermic electron capture (≈6 eV recombination energy) that produces localized excitation far greater than that required for dissociation, and that reaction occurs before this energy is randomized over the thousands of vibrational degrees of freedom of the protein ion. ECD produces mainly c and z· fragment ions (Eq. 1) (1–4),

≈10% a· and y ions (Eq. 2),

although disulfide bond cleavage can be even more highly favored (Eq. 3) (2).

The latter's independence from charge site location indicates initial electron capture in a long-lived high-n Rydberg state (2, 4); this mechanism has long been accepted (28) for the similar “dissociative recombination” of a small ion (e.g., H⁺₃,H₃O⁺) and an electron (28–30).

For peptide cations, it has recently been argued (16–21) that electron capture will induce radical site reactions of unusually low activation energies that make a nonergodic ECD mechanism unnecessary. Extensive experimental and theoretical evidence (16–19) show, for example, that primary products can have 10^–6-s lifetimes, whereas ECD of cyclic peptides can produce c, z· products, despite the fact that this should require two consecutive cleavages, for which the second cannot use the original electron recombination energy (20). Alternative mechanisms have also been proposed for ECD of -SS-bonds (Eq. 3) (22, 23). Here experiment and theory are used to formulate a more comprehensive picture of the ECD mechanism.

Materials and Methods

Experiments were performed on a 6T Fourier transform (FT) mass spectrometer as described (4, 24). Bovine ubiquitin and bovine carbonic anhydrase (Sigma) were electrosprayed from 49:49:2 H₂O/MeOH/MeCOOH solutions. The ions were transferred into the FTMS ion cell through quadrupole ion guides and decelerated and trapped with a nitrogen gas pulse. For ECD of specific charge states, the ions of interest were isolated by using SWIFT waveforms and allowed to thermalize for 40 s before irradiation with low-energy electrons (≤≈0.2 eV), dipolar excitation, and detection, averaging ≈25 scans. Spectral interpretation used the automated thrash program (31). The peak intensity values were divided by the number of charges, as the detector response is proportional to this number.

Results and Discussion

Instability of Protein Radical Ions. Theoretical calculations of a variety of peptide radical ion species by Turecek and coworkers (16–19) found that many will have lifetimes as low as ≈10^–6 s, much shorter than the ≈1-s FTMS ion residence times required for spectral measurement, but far higher than those of a nonergodic process (17). Experimentally, a variety of free radical reaction products are observed in the ECD spectra of peptides (3, 6), whereas those of proteins show only Eqs. 1–3 and H· loss (see below) (1–5). However, it was our unexpected observation that ECD cleavages occur between so many protein backbone residues (183 in Fig. 1) (8) that prompted our nonergodic postulate (1).

O'Connor and coworkers (20) have pointed out that a low stability of the ECD primary radical product ions (z· from Eq. 1) could also be a major reason for ECD's extensive sequence coverage. Large z· ions could be formed from cleavages at only a few backbone sites, and these could undergo secondary dissociation between many of the other amino acid pairs to form smaller z· ions. Although the initial ECD cleavage of cyclic peptide 2+ ions should only generate a linear ion of the same mass that thus would not be distinguishable from a cyclic-reduced ion, they observed that ECD yields numerous 1+ radical ion products.

Considering this mechanism for proteins, plasma ECD of carbonic anhydrase does yield only 43 z· versus 115 c ions (Fig. 1), so that secondary free radical reactions could cause dissociation of large z· ions under these energetic conditions (8-eV protein ions entering the FTMS ion cell at ≈10^–5 torr N₂) (8). However, secondary dissociation can be minimal under other ECD conditions; ECD of 13+ ubiquitin ions shows equivalent formation of z· versus c ions (Eq. 1), with 38 and 37 ions, respectively, at 25°C, and 38 and 35 ions at 100°C (Fig. 2). More importantly, the even-electron c ions of Fig. 1 provide nearly complete sequence coverage for the N-terminal half of the protein, although they should have far less tendency for secondary decomposition than z· ions; higher mass c ions are removed by secondary electron capture (7, 8). Finally, the intermediacy of c· radical ions, found in the ECD spectra of peptides (3, 20), should be considered (20). These are not observed in protein ECD spectra, so that, if formed, they must have unusually low dissociation thresholds. This lability would predict that secondary c· (and z·) ion reactions would be highly selective, again providing few of the c ions observed in Fig. 1.

Fig. 2.

Cleavage sites of ECD spectra of (M + 13H)¹³⁺ ubiquitin ions as a function of ion cell temperature; c ions above line, z· below.

H· Atom Loss Versus c, z· Formation. Early key evidence from Rice–Ramsperger–Kassel–Marcus (RRKM) calculations for the nonergodic postulate was that the loss of H· from the model radical species HOC^·(CH₃)-NHCH₃ (analogous to the Eq. 1 intermediate) is favored over the loss of ·CH₃ until the internal energy exceeds 2 eV (2). This finding is consistent with the general observation that the rates of smaller, versus larger, fragment losses from gaseous ions also increase more slowly with increasing internal energy (ref. 32 and references therein). Extending this notion to protein ions, the first step of Eq. 1 would be reversible, not proceeding to c, z· products unless substantial additional internal energy is supplied locally. However, recent RRKM calculations on the same model species give opposite results, finding Eq. 1 to be the lowest energy reaction, and H· loss instead favored at energies >1 eV (experimentally, [-H·]/[-CH₃·] = 1.7) (19). On our re-examination, these seriously divergent results appear to arise from divergent assumptions on the orientation of the H atom on the oxygen of the intermediate. In our calculation, the H· atom arrives at the carbonyl oxygen on the same side as the -NH-group, H₃C-^·C(-OH)-NH-CH₃. However, with sufficient time and structural freedom for the intermediate to isomerize to the lower energy position with the H above the α-methyl, H₃C-(HO-)^·C-NH-CH₃, c, z· formation will be the lowest energy reaction.

For experimental evidence of these relative energy requirements in thermalized protein ions, among the electron capture products of the ubiquitin (M + 12H)¹²⁺ ion is the undissociated radical ion (M + 12H)^11+· (Fig. 3) (33). After selection of a single isotopomer, IRMPD for 0.3 s of these (M + 12H)^11+· radical ions gives only H· loss; IRMPD for 2 s under the same conditions gives no c, z· products. Higher flux IRMPD did give b, y products, presumably through the intermediate (M + 11H)¹¹⁺ ions formed by H· atom loss; ubiquitin (M + 11H)¹¹⁺ ions formed directly by electrospray ionization gave a similar IRMPD spectrum. Thus, for these protein ions, H· atom loss is clearly the reaction of lowest activation energy, far lower than that of Eq. 1, so that the Eq. 1 dominance in ECD must result from a far higher excitation at the reaction site (1–4). H· loss is even more dominant in the electron capture reduction of the polyethylene glycol ions (M + 44H)⁴⁴⁺, yielding (M + 43H)⁴³⁺, but not (M + 44H)^43+· ions (34).

Fig. 3.

IRMPD of single isotopomer (M + 12H)^11+· ions formed from electron capture by (M + 12H)¹²⁺ ubiquitin ions, showing exclusive loss of H· atoms to yield (M + 11H)¹¹⁺ products.

ECD Insensitivity to Ion Internal Energy. A high local energy induced at the reaction site by e^– capture should make the initial internal energy at the site have far less effect on relative ECD rates than on those of threshold energy reactions such as collisionally activated dissociation and IRMPD. This idea is tested here by heating ubiquitin 13+ ions that have been shown to be essentially free of tertiary noncovalent intramolecular bonding (35–37). For their IRMPD spectra, raising the temperature of the ubiquitin 13+ ions increases their backbone fragmentation from 58% at 25°C to 77% at 75°C to 93% at 125°C (Fig. 4). However, this same increase in thermal energy has little effect on the yield of c ions (but not the thermally sensitive z· ions) relative to the total abundance of all ECD products, including (M + 13H)^12+· (Fig. 4). The 5 ± 3% decrease at 125°C could arise from thermal dissociation of larger c ions. Depletion of the latter goes down even more with increasing temperature because the electron capture cross section goes down, which could reflect effects of ion cell heating on electron transmission.

Fig. 4.

The effect of temperature on the backbone fragmentation yield from subjecting ubiquitin (M + 13H)¹³⁺ ions to: IRMPD, yield = ([b] + [y] + [internal fragment ions])/([(M + 13H)¹³⁺] + [b] + [y] + [internal fragment ions]), ▪; ECD, yield = ([c] + [a·])/([(M + 13H)^12+·] + [c] + [a·] + [side-chain loss]), ▵.

The specific product ions of the ECD and IRMPD spectra also show contrasting effects of increasing temperature, with IRMPD showing 4, 7, and 15 cleavage sites at 25°C, 75°C, and 125°C, respectively. For the ECD products (Fig. 2), the c ions are nearly unaffected in relative abundance and identity (38, 38, and 37 cleavage sites, respectively). The increasing temperature appears to give few (see below) new primary reactions for the formation of z· ions, although some secondary dissociation of these radical ions is evident at 125°C. The activation energy for blackbody IR dissociation of 8+ and 9+ ubiquitin ions is 1.2 eV, and 10+ and 11+ ions is 1.55 eV (38); the value for 13+ ions should be slightly greater. If the activation energy for the ECD Eq. 1 radical reactions is far lower, as predicted (16), increasing thermal energy should increase the number of ECD cleavages far more than those from IRMPD. Because the ECD extent has hardly changed, the excess local energy induced at the reaction site by electron capture must be far greater than that induced by 25°C to 125°C heating.

Nonergodic Dissociation Through Rydberg States. Related processes are the electrochemical “dissociative electron transfer” (“occurs simultaneously with the capture of an electron”) (39) and the ion “dissociative recombination” (28, 30) techniques. The latter concerns e^– capture by small, singly charged ions such as and H₃O⁺ shown to involve nonergodic dissociation through the intermediacy of high-n Rydberg states (28, 30). Reflecting this idea, we proposed that ECD can be described in surface crossing language (2, 4); an electron is captured by a multiply charged protein molecular ion to generate a high-n Rydberg state. The long lifetime (>10^–6 s) (29) during cooling through lower Rydberg electronic states allows sampling of many different bond lengths and geometries of the ion before internal conversion to one of the many possible valence states that can lead to Eqs. 1–3 dissociations near protonated sites (as supported recently by Simons and coworkers) (22). For the Eq. 1 reaction to occur at one amino acid of the many in a protein ion, the probability for internal conversion from the appropriate excited electronic state to the Eq. 1 final state surface will thus depend on fortuitous geometry optimization, such as bond distances at the incipient -CO···H^·NH_2– grouping favorable for the Eq. 1 transition state. This new valence state will usually be accessed at a highly excited vibrational level, well above the threshold energies of competing dissociations. Further, ECD spectra also show the reduced molecular ion (M + nH)^(n–1)+· as a stable product; presumably this results from cooling of the initial Rydberg state of the molecular ion without accessing a conformation favorable for internal conversion.

This mechanism offers an explanation for why Eq. 3 can be the dominant ECD reaction for SS bonds, such as the 10+ molecular ions from a 10-kDa protein that has 4.5- and 5.4-kDa units joined by an SS bond (2). If the e^– capture occurred directly at any of the 10 H⁺ sites, how could the resulting H· atoms find their way to the single SS site (2, 22, 23)? However, with initial electron capture leading to a high-n Rydberg state (2, 4), internal conversion will be energetically favored to the vibrationally excited valence state of the most reactive center, the SS bond (≈1 eV higher H· atom affinity than that of the amide carbonyl) (2, 22). For this example, the H· atom can be supplied from a protonated arginine (R) residue adjacent to an SS cysteine (2, 23).

ECD spectra of disulfide-bonded peptides without an adjacent basic residue, measured by Hudgins, Marshall, and coworkers (22, 23), also show SS bond cleavage. The gaseous dimers of C-terminal lysine peptides (–SA_nKH⁺)₂ gave similar yields of ECD SS cleavage products for n = 10, 15, and 20, despite the fact that the alanine backbones should be helical (refs. 22, 36 and references cited therein, and 40). Lowering the charge on ubiquitin ions from 13+ to 9+ (5.8–8.4 residues per charge) changes the stable α-helix structure to one with a terminal helix section folded back and H-bonded to the central helix (37). Thus the (–SA_nKH⁺)₂ ions with 10, 15, and 20 residues per charge, although containing more stable helices (refs. 36 and references cited therein and 40), could all be sufficiently flexible to transfer the hypervalent H· from the terminal lysine to the SS bond. This could also rationalize the same ECD cleavage observed for these peptides as the sodiated ions (22), with H· transfer from the –N^·H₂Na side chain of lysine expected because of the high H· atom affinity of SS. As an additional possible explanation, tautomerism should make a reactive H· atom available more distant from the neutralized proton, as discussed below.

Structure of the Reduced Molecular Ion. For native proteins in solution, the α-helix is a common stable structural feature; removing its aqueous environment greatly increases the strength of its hydrogen bonding, so that many experimental and theoretical studies have indicated that the α-helix is a prevalent structural element in gaseous monomeric protein ions (40, 41) and is dominant in the fully protonated ubiquitin 13+ ion (35–37). These only exchange 15 ± 2 D atoms with gaseous D₂O (37, 42), ≈1 D for each H⁺. Thus all but one of the hydrogen atoms of each protonated functionality (e.g., the H₃N⁺-group of a protonated lysine) must be hydrogen-bonded to the α-helix, presumably at backbone carbonyl groups (43), or all H atoms could be H-bonded, with one displaced by D₂O. In support of this idea, the interactions of NH⁺ with the gly₆ mimic CH₃CO(NHCH₂CO)₄CH₂CONH₂ were predicted with (U)B3LYP3–21G (44, 45) (Fig. 5A). Three of the ammonium ion H atoms are found to form hydrogen bonds to alternating carbonyl oxygens. Modeling the effect of electron neutralization in this mimic predicted that the product would have the singly occupied molecular orbital of Fig. 5B. Note that the NH₄ is not part of this orbital, but instead the orbital is a combination of carbonyl groups, but only those carbonyl groups that are hydrogen-bonded to the NH⁺₄. The addition of an electron to the ion apparently has created an ion combined with anion. the peptide radical When the geometry of this radical was optimized, an H· transfer occurred to one of the previously hydrogen-bonded glycine carbonyl groups, leaving a neutral H-bonded NH₃.

Fig. 5.

The calculated minimum energy structure of the complex of H₄N⁺ and the gly₆ mimic CH₃CO(NHCH₂CO)₄CH₂CONH₂ (A) and the singly occupied molecular orbital formed by the addition of an electron to A (B).

An analogous mechanism can be visualized for the α-helical ubiquitin (M + 13H)¹³⁺ ions, with an H₃N⁺–(for a lysine side chain) hydrogen-bonded to as many as three adjacent carbonyl groups in the α-helical structure. Electron neutralization and sufficient cooling could form a structure such as in Fig. 6, with the unpaired electron now shared among carbonyls in adjacent loops of the helix. In addition, at each of these carbonyls a keto-enol type tautomerism should make H· atoms available for Eqs. 1–3 reactions down the α-helix. An analogous hydrogen atom transfer along a hydrogen-bonded ammonia wire is made possible by photoexcitation (46).

Fig. 6.

Section of the postulated α-helical structure of an e^–-reduced protein ion with a side-chain lysine H^·₃N-group H-bonded to three amide carbonyls, with red, blue, and orange H-bonding networks along the helix.

After thermalization of the initially ECD formed (M + nH)^(n–1)+· ion, its extended lifetime without H· loss is also consistent with the Fig. 6 tautomeric stabilization of the hypervalently bound H·. In contrast, electron capture reduction of the polyethylene glycol ions [(C₂H₄O)₄₄ + 3H]³⁺ shows no such H· stabilization, yielding only [(C₂H₄O)₄₄ + 2H]²⁺ (34).

Conformational Control of ECD Cleavages. For protonated polyethylene glycol, a proton can only be hydrogen-bonded to the backbone oxygen atoms, and multiple protons should be evenly distributed along the linear backbone because of electrostatic repulsion; for [HO(C₂H₄O)₁₀₀H + 6H]⁶⁺, the most probable H⁺ locations should be oxygens 1, 21, 41, 61, 81, and 101. Relative to these, the ECD cleavage sites show distributions with nearly equal probabilities at the six ether sites closest to the H⁺ (Fig. 7 Upper) (34). Protonation must also occur adjacent to these sites, so that the actual distribution is somewhat narrower, reflecting the number of neighboring oxygens bound to the charge site found by Bowers and coworkers (47). For the ubiquitin 13+ ions (Fig. 2), however, the number of residues between the ECD cleavage site and the neutralized H⁺ side chain (determined by the product ion charge state and including Pro-19) (35) exhibit no such symmetry (Fig. 7 Upper). Instead, these correlate closely with the proposed H-bonding of the protonated side chains into the α-helical structure (Fig. 6) (35), showing dominant cleavages 2, 3, and 4 aa toward the N terminus from the H⁺ side chain. An appreciable number of cleavages actually occur in the second helix loop toward the N terminus, consistent with tautomerism, making farther removed H· atoms available. Most of the C-terminal side cleavages of Fig. 7 Upper correspond to the C-terminal residues His-68, Arg-72, and Arg-74 (c ions in Fig. 2), consistent with electrostatic repulsion of these protonated side chains toward the C terminus (35). If the electron had been captured directly on the carbonyl group (22, 23), this would have lowered the charge on the c (instead of the z·) ions, giving negative values in the Fig. 7 Upper distribution.

Fig. 7.

The probability of ECD at a site, relative to the number of monomer units in polyethylene glycol ions, or amino acids in ubiquitin ions, that separate the site from the protonated residue that supplies the H· atom.

For the 7+ ubiquitin ions, the ECD spectrum shows few c, z· fragment ions, as noncovalent tertiary bonding prevents separation of most of these pairs (35). IRMPD of the ECD reduced (M + 7H)^6+· ions frees these c, z· ions; their abundance distribution versus charge site (now arbitrarily assigned to the closest cleavages, Fig. 7 Lower) is similar to that for the 13+ ions, but with additional cleavage at the first (–1) NC_α bond on the C-terminal side of the basic side chain. Heating the 7+ ubiquitin ions to 175°C destroys most of the tertiary noncovalent bonding (35), and the ECD distribution of cleavage sites also shows this increased probability for –1 cleavage, but with greatly reduced +2 and +3 cleavages. The helix has been weakened by reducing both its charge (13+ to 7+) and its secondary and tertiary H-bond stabilization, so that apparently the protonated side chain can also become H-bonded to adjacent carbonyl groups on the helix exterior (Fig. 6). Under the harsher conditions of plasma ECD (8), ubiquitin ions (mainly 9+ to 12+) yield a correlation (Fig. 7 Lower) with an even higher proportion of +1, –1 cleavages. Thus the stabilized α-helical structure (e.g., 13+ ions) gives the most selective ECD cleavage probabilities. These are less selective with charge reduction, with its helix destabilization offset by increased tertiary noncovalent binding. Raising the internal energy of the α-helix conformation, such as by heating or collisional activation, makes possible additional interactions of the protonated side chains and/or of H· transfer through tautomerism, yielding cleavages at a higher proportion of interresidue locations and increasing sequence coverage. In the plasma ECD spectrum, 62 c and 53 z· ions represent 69 of the 72 possible cleavage sites (1–4). This conformational dependence of ECD spectra is consistent with the strict geometry requirements for excited-state internal conversion before nonergodic dissociation. In contrast, Clemmer and coworkers (48) found that compact and elongated conformers of ubiquitin 8+ and 9+ ions yield identical collisionally activated dissociation spectra, consistent with lower dissociation energies for noncovalent than for covalent bonds in ergodic dissociation.

Conclusions

An obvious rationale for adding an electron to initiate a dissociative process is that the radical species so formed should be unusually reactive. This is certainly true for thermalized (M + 12H)^11+· ions of ubiquitin, as these ions lose H· atoms with minimal excitation. However, when such (M + 12H)^11+· ions are formed instead by electron addition to (M + 12H)¹²⁺ ions, backbone cleavage (Eq. 1) is dominant. Further, this product yield for electron addition to ubiquitin (M + 13H)¹³⁺ ions is essentially unaffected by a 100°C temperature increase, in sharp contrast to energetic dissociation methods such as IRMPD. The substantial (≈6 eV) e^– recombination energy not only makes the Eq. 1 reaction dominant over H· loss, but also minimizes the competitive effect of different bond dissociation energies for Eq. 1 reactions. Although ECD spectra of peptide ions can undergo reactions in addition to Eqs. 1–3 (3, 6), their c, z· ions should also be formed by a nonergodic process. This results primarily from insufficiently fast dispersal of the excess recombination energy away from the active site, not by how extensively it can then be dispersed. Thus nonergodic dissociation should not be assumed to be improbable just because the species is large.

Also critical to the formation of specific c, z· ions is the helix conformational positioning of the basic side chain for transfer of its neutralized H· atom to a carbonyl group. Cleavage 2, 3, and 4 residues toward the N terminus are characteristic of the unperturbed α-helix, whereas far greater ECD sequence coverage results from disruption of the noncovalent helical stabilization.