Is Cys(MTSL) the Best α-Amino Acid Residue to Electron Spin Labeling of Synthetically Accessible Peptide Molecules with Nitroxides?

Electron paramagnetic resonance spectroscopy, particularly its pulse technique double electron-electron resonance (DEER) (also termed PELDOR), is rapidly becoming an extremely useful tool for the experimental determination of side chain-to-side chain distances between free radicals in molecules fundamental for life, such as polypeptides. Among appropriate probes, the most popular are undoubtedly nitroxide electron spin labels. In this context, suitable biosynthetically derived, helical regions of proteins, along with synthetic peptides with amphiphilic properties and antibacterial activities, are the most extensively investigated compounds. A strict requirement for a precise distance measurement has been identified in a minimal dynamic flexibility of the two nitroxide-bearing α-amino acid side chains. To this end, in this study, we have experimentally compared in detail the side-chain mobility properties of the two currently most widely utilized residues, namely, Cys(MTSL) and 2,2,6,6-tetramethylpiperidine-1-oxyl-4-amino-4-carboxylic acid (TOAC). In particular, two double-labeled, chemically synthesized 20-mer peptide molecules have been adopted as appropriate templates for our investigation on the determination of the model intramolecular separations. These double-Cys(MTSL) and double-TOAC compounds are both analogues of the almost completely rigid backbone peptide ruler which we have envisaged and 3D structurally analyzed as our original, unlabeled compound. Here, we have clearly found that the TOAC side-chain labels are largely more 3D structurally restricted than the MTSL labels. From this result, we conclude that the TOAC residue offers more precise information than the Cys(MTSL) residue on the side chain-to-side chain distance distribution in synthetically accessible peptide molecules.

Introduction

Proteins and polynucleotides are the fundamental molecules of life. Interestingly, they are composed of a limited number of monomer types, the succession of which in the main chain determines their function. This later, in turn, is strictly dependent on the 3D structure adopted by the biomolecule. Therefore, studying the conformation of these molecules is of primary relevance. To this aim, nuclear magnetic resonance (NMR), circular dichroism (CD), X-ray diffraction crystallography, and fluorescence are undoubtedly among the most widely exploited techniques for solution- and crystal-state investigations. However, interest for electron paramagnetic resonance (EPR) is steadily increasing because this spectroscopy can be used in a variety of environments, and in addition, it relies on highly specific probes. In particular, pulsed EPR experiments, such as double electron–electron resonance[1−4] (DEER, also known as PELDOR, i.e., pulsed electron double resonance), allow for measuring distances between paramagnetic probes, spaced from almost 20 to 160 Å (in the latter case, in deuterated compounds).[5] This opportunity permits one to obtain information on the 3D structure of a macromolecule, on the formation of aggregates, and on interactions among different types of molecules.

In this context, nitroxide electron spin labels are by far the most used probes. They were utilized for seminal measurements of distances in isolated proteins[6−14] and within cells[15−17] for analyzing conformational transitions of the prion protein,[18] for unraveling DNA mechanisms,[19−21] for investigating conformation, aggregation, and membrane interactions of peptaibols,[22−27] for studying peptide distribution and mobility on solid supports,[28,29] for understanding host–guest interactions in supramolecular chemistry,[30] and more recently for examining proteins of interest to food science.[31] In any case, here it is fair to mention that in recent years, additional promising approaches different from nitroxides have been utilized for assessing distances between free radical moieties in proteins [e.g., lanthanide (Gd3+) tags,[32] double His-Cu2+-based labels,[33] and trityl (triarylmethyl)-based labels[34,35]].

Two very popular nitroxide probes in peptide/protein (including protein peptidic fragments) investigations are the Cα-tetrasubstituted α-amino acid TOAC (2,2,6,6-tetramethylpiperidine-1-oxyl-4-amino-4-carboxylic acid)[36−41] (Figure ) and MTSL (methanethiosulfonate spin labels; 1-oxyl-2,2,5,5-tetramethylpyrroline-3-methyl) nitroxide[42] selectively reacts with the side-chain thiol group of the coded Cys residue to generate Cys(MTSL) (Figure ).

Figure 1

Chemical structures of the TOAC (left) and L-Cys(MTSL) (right) electron spin-labeled α-amino acid residues.

Unfortunately, the Cys(MTSL) spin probe is able to explore a relatively wide conformational space as it is connected to the peptide backbone through as many as five rotatable σ-bonds, although interconversions between some of the various states might be too slow to contribute significantly to the overall nitroxide motions.[13] In addition, the presence of a relatively rare Cys residue is required in a natural peptide/protein (or alternatively, it should be biosynthetically incorporated). To overcome, at least in part, these latter limitations, a few other coded amino acid-based electron spin probes were proposed, but also in these cases, various σ-bonds separate the nitroxide moiety from the peptide main chain.[9−12,43] Also, a slightly less flexible arm is sometime provided when the Cys side chain is functionalized with shorter substituents.[44] In any case, the conformational restriction characteristic of TOAC is unique, superior to that of any side-chain functionalized, coded amino acid electron spin probe. Actually, its paramagnetic group is tightly bound to the peptide backbone through a six-membered ring group. Thus, TOAC offers an essential advantage for distance measurements in view of its precise spatial location and quasi-rigid nitroxide orientation.[45]

Another favorable property of TOAC is the possibility to insert it by chemical synthesis, although by a careful, usually nonroutine solid-phase procedure,[38,39,41] at any position of a polypeptide chain. Nevertheless, it is evident that this replacement is appropriate if in that region of the original sequence a helical segment is already present because TOAC is well known to be a helix inducer (and a β-sheet breaker).[46−48] Moreover, it should be admitted that the noncoded TOAC has so far escaped all attempts to be biosynthetically incorporated in any polypeptide molecule. By contrast, the corresponding properties of Cys are (i) easy insertion in a growing peptide chain by chemical synthesis using a suitable side-chain protected derivative and (ii) its only weak conformational bias in favor of the β-sheet structure and in the destabilization of the α-helix.[49]

To experimentally prove the different properties of the TOAC and Cys-functionalized electron spin probes and to highlight the uncertainty in the distance determination when using the latter, in this work we chemically synthesized and spectroscopically compared two α-helical peptides, one containing two TOAC residues, separated by four helix turns, and the second characterized by two Cys(MTSL) replacing the two TOAC.

Results and Discussion

Design of the Doubly Electron Spin-Labeled α-Helical Peptides

A proper distance between two nitroxyl radicals in a DEER investigation is in the range 20–25 Å. In this study, our target was a peptide system in α-helical conformation. In a regular α-helix, each turn comprises 3.6 residues, and seven amino acids are necessary to obtain two side chains precisely on top of one another at a distance of 11 Å. To achieve a distance between the two nitroxyl probes appropriate for the EPR investigation, a separation of 14 residues between the labels is required. Considering this parameter and that in a stable α-helix at least two residues both at the N- and C-termini are considered rather mobile, at least 20 residues should occur to obtain a quasi-rigid backbone ruler to be used for the EPR investigation. On this basis, we designed an N-acetylated (Ac) 20-mer model peptide amide (peptide I in Table ) characterized by an alternation of the helicogenic Cα-tetrasubstituted α-aminoisobutyric acid (Aib)[50,51] and two of the most helicogenic protein amino acids, Ala and Lys.[49] To evaluate the reliability of different electron spin labels in distance measurements, we designed and synthesized, in addition to the unlabeled peptide (I), two analogues (II and III), all with a well-defined, Aib-generated, stable α-helical structure. For our investigation, we selected two different nitroxyl probes with different conformational stabilities, the helicogenic TOAC residue and the S-(1-oxyl-2,2,5,5-tetramethyl-2,5-dihydro-1H-pyrrol-3-yl)methyl (MTSL) group covalently linked to the side chains of the Cys residues in sequence. In the TOAC-containing peptide, analogue II, we introduced its electron spin label at positions 3 and 17 (i → i + 14) where the helix-promoting Aib residues of the same Cα-tetrasubstituted class were present in the original model peptide. The MTSL electron spin probe reagent needs to be conjugated with two Cys in sequence without perturbing the overall helical conformation of the peptide. To this aim, we decided to insert the Cys residues at positions 4 and 18 (i → i + 14) (peptide III), thus maintaining the presence of the helix-promoting Aib amino acids.

Table 1

Peptide Sequences

	peptide squence
Ac-20-NH2	Ac-Aib-Lys-Aib-Ala-(Aib-Lys-Aib-Ala)3-Aib-Ala-Aib-Lys-NH2	I
Ac-[TOAC3,17]-20-NH2	Ac-Aib-Lys-TOAC-Ala-(Aib-Lys-Aib-Ala)3-TOAC-Ala-Aib-Lys-NH2	II
Ac-[Cys(MTSL)4,18]-20-NH2	Ac-Aib-Lys-Aib-Cys(MSTL)-(Aib-Lys-Aib-Ala)3-Aib-Cys(MSTL)-Aib-Lys-NH2	III

Peptide Synthesis and Characterization

Solid-phase peptide synthesis (SPPS) protocols were used for the difficult syntheses of the unlabeled sequence I and the two differently double-labeled analogues. The high content of Cα-tetrasubstituted α-amino acids (50% Aib and/or TOAC in sequence) required the optimization of a protocol able to overcome the low reactivity of the Nα-amino group of those two residues. In all of the syntheses, we employed [O-(7-azabenzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate][52] as the C-activation method and repeated each peptide coupling reaction twice.[53]

The synthesis of the TOAC-labeled analogue proved to be the most difficult. Besides the difficulties involving the introduction of the electron spin-labeled amino acid, the nitroxyl group is particularly sensitive to the strong acidic conditions used in the cleavage and purification steps which convert the free radical to its N-hydroxylated species.[54] To completely recover the nitroxyl radical character, an alkaline treatment with 1 M ammonium hydroxide for 180 min was required. The reaction was checked by HPLC analysis. The crude peptide was purified by preparative HPLC, thus avoiding the presence of the acidic modifier trifluoroacetic acid in the elution system.

The conjugation of the MTSL electron spin reagent on the Cys-containing peptide was performed in solution. Once synthesized by SPPS, the peptide was cleaved from the resin in the presence of a proper scavenger to avoid formation of the disulfide bond and then subjected to a preliminary purification step. For the functionalization step, the side-chain free Cys containing peptide was dissolved in a buffer solution in an argon atmosphere and treated with an excess of the MTSL spin reagent. The reaction was monitored by HPLC, and the lyophilized crude peptide was purified by preparative HPLC.

For all samples, the correctness of the chemical assignments to the peptide fractions obtained by HPLC purification was confirmed by HPLC-MS analyses (Table S2, Supporting Information).

Conformational Analysis

The preferred conformations of the unlabeled, terminally blocked peptide Ac-20-NH2 (I) and its two nitroxyl-labeled analogues Ac-[TOAC3,17]-20-NH2 (II) and Ac-[Cys(MTSL)4,18]-20-NH2 (III) were investigated in solvents of diverging polarities and at different concentrations by the use of Fourier transform infrared (FT-IR) absorption, electronic CD, and 2D NMR.

The FT-IR absorption spectra in a solvent of limited polarity (CDCl3) were obtained in the N–H stretching (3450–3200 cm–1) and C=O stretching (1750–1600 cm–1) regions. In all three compounds at 1 mM peptide concentration, very intense absorption bands are seen with maxima at 3306–3309 cm–1 (Figure ) and near 1659 cm–1, respectively. We attribute these peaks to the N–H and C=O vibrators of strongly H-bonded amide/peptide groups. The typical corresponding bands for intramolecularly H-bonded, peptide helical structures are close, being seen at approximately 3330 and 1655 cm–1.[55,56] The absence of self-aggregation in the stabilization of this peptide conformation is indicated by the remarkable spectral consistency upon peptide dilution to 0.1 mM concentration (not shown). An extremely weak band near 3420 cm–1, assignable to solvated N–H vibrators of the N-terminal peptide groups not part of the intramolecular H-bonding scheme, is hardly visible in the spectra (Figure ).

Figure 2

FT-IR absorption spectra in the N–H stretching region of the Ac-20-NH2 (solid line), Ac-[TOAC3,17]-20-NH2 (dashed line), and Ac-[Cys(MTSL)4,18]-20-NH2 (dashed-dotted line) peptides in CDCl3 solution at 1 mM concentration.

From this spectroscopic analysis, we conclude that in CDCl3 solution, the largely prevailing conformation of the three unassociated 20-mer peptides, with 50% proportion of the helicogenic Cα-tetrasubstituted α-amino acids Aib[50,51] and TOAC,[41,57] is remarkably folded and stabilized by intramolecular C=O···H–N H-bonds.

The far-UV CD curves of the three peptides in methanol (MeOH) (Figure , left) are quite comparable in shape, but the intensities of the TOAC-containing analogue are different (although slightly). They exhibit two strong negative bands located near 208 and 225 nm, accompanied by an even stronger positive band at about 192 nm, typical of the n → π* transition and of the parallel/perpendicular components of the split π → π* transition of the amide chromophores, respectively, in right-handed helical peptides.[58,59] Interestingly, the side-chain nitroxyl chromophores of the two Cys(MTSL) residues do not seem to contribute at all in this spectral region, whereas those of the two TOAC residues appear to provide a limited effect.[60] The R, [θ]222/[θ]208, values indicative of the type of helical structure adopted[58,59,61] are 0.8 and 0.9 for all three peptides, clearly suggesting that the α-helix conformation is largely preferred over the 310-helix. This finding is not surprising, considering the relevant backbone length (20 amino acids) of these peptides.[61] From the comparison of these three curves with those in H2O solution (Figure , right), it is clear that the intensities and, consequently, the related peptide structurations are remarkably lower in an aqueous environment (but the α-helix component is still always prevailing, with R values 0.8–1.1).

Figure 3

Far-UV CD spectra of the three compounds studied in MeOH (left) and water solutions (right) (peptide concentration: 0.3 mM).

The near-UV/visible CD curves of the two nitroxyl-labeled peptides (Figure ) diverge dramatically. The bis-TOAC labeled analogue shows an intense maximum at 422 nm, arising from the nitroxyl chromophore n → π* transition of the achiral amino acid. This positive, induced CD band is brought about by the right-handed helical structure of the host peptide, in analogy with the observations already reported for similar compounds.[41,60,62]

Figure 4

Near-UV and visible spectra of the two nitroxyl-labeled compounds studied in MeOH solution (peptide concentration: 1 mM).

This type of evidence confirms the structuration of the peptide. Conversely, the CD curve for the bis-Cys(MTSL) analogue is significantly weaker in the same spectral region. This result supports our view that the MTSL nitroxide is covalently attached not only quite far from the source of the chirospectroscopic effect (the peptide backbone) but also via a mobile side chain.[63,64] More information about this type of mobility has been obtained by continuous-wave EPR (CW-EPR) (see below). Therefore, it is reasonable that its CD intensity would be only marginally influenced by the helical peptide chirality.

The 2D NMR investigation allowed us to obtain more detailed information on the 3D structural preferences of the unlabeled Ac-20-NH2 peptide in MeOH, d3 solution. The C–H and N–H peaks were attributed on the basis of the Wüthrich methodology.[65] The complete assignment of the proton resonances is shown in the Supporting Information (Table S3). Since it is well known that a limitation in the physicochemical analyses of −N–O· containing compounds is that most of the proton resonances in their 1H NMR spectra are remarkably broadened owing to the paramagnetic effect induced by the nitroxyl radical, we did not make any attempt to characterize either the Ac-[TOAC3,17]-20-NH2 analogue or its Cys(MTSL)4,18 counterpart by 2D NMR.

From the presence of all of the NH → NH sequential cross peaks, the nuclear Overhauser effect spectroscopy (NOESY) spectrum of the unlabeled 20-mer, terminally blocked, peptide shown in Figure represents an unambiguous proof that this compound is completely folded in a helical structure. This finding is in line with our conformational conclusions on the same compound obtained by FT-IR absorption and CD spectroscopies discussed above. In the NOESY fingerprint region. we noted multiple Cα/βH → NH correlations and, more interestingly, one Cα/βH → NH typical of the α-helical structure [for that involving CβH → NH; see Figure ]. It is relevant that the CαH → NH interactions, indicative of the 310-helix, seem to be absent. A summary of all significant interresidue NOESY cross peaks for Ac-20-NH2 is given in Figure .

Figure 5

Amide NH proton region of the NOESY spectrum of the Ac-20-NH2 peptide in MeOH, d3 solution at 2 mM concentration. The NH → NH cross peaks are marked in black.

Figure 6

Region of the NOESY spectrum of the Ac-20-NH2 peptide in MeOH, d3 solution at 2 mM concentration. The CβH → NH cross peak is marked in blue.

Figure 7

Summary of the significant interresidue NOESY cross peaks for the Ac-20-NH2 peptide in MeOH, d3 solution at 2 mM concentration.

Based on the α-helical propensity of the two labeled peptides, we proceeded to generate the corresponding PyMOL 3D model to highlight the relative position of the nitroxyl labels. From these amino acid sequences, it is clear that in each labeled peptide, the two nitroxyl building blocks were by purpose separated by 14 residues, consistent with four complete turns of the α-helix. Therefore, we were not surprised to find the two nitroxyl moieties located on the same side of the α-helix (Figures and 9). In addition, from the 3D models, we were able to extrapolate the distance between the two TOAC in the Ac-[TOAC3,17]-20-NH2 peptide and that between the two Cys(MTSL) residues in the Ac-[Cys(MTSL)4,18]-20-NH2 peptide. In particular, the distances between the two nitroxyl oxygen atoms in each of the two compounds (25.6 and 22.9 Å, respectively) will allow a direct comparison with the corresponding data available from the EPR (DEER) experiments discussed below.

Figure 8

3D model of the Ac-[TOAC3,17]-20-NH2 peptide in the α-helical conformation obtained using PyMOL software. The distance between the Cα-atoms (red arrow) and between the nitroxyl oxygen atoms (black arrows) of the two TOAC residues are measured from the model itself.

Figure 9

3D model of the Ac-[Cys(MTSL)4,18]-20-NH2 peptide in the α-helical conformation obtained using PyMOL software. The distance between the Cα-atoms (red arrow) and between the nitroxyl oxygen atoms (black arrows) of the two Cys(MTSL) residues are measured from the model itself.

EPR Study

The room-temperature CW-EPR spectra in fluid solution of the synthetic compounds Ac-[TOAC3,17]-20-NH2 and Ac-[Cys(MTSL)4,18]-20-NH2 confirm the presence of the radical nitroxide attached to the chemical structures: the spectra are characterized by relatively broad lines, especially for the TOAC-labeled system (Supporting Information), consistent with its large molecular dimension.[42] While for the TOAC rigid system, the reorientational motion is due to that of the whole molecule, the narrower linewidth of the Cys(MTSL)-labeled compound makes us to assume that the MTSL probe would undergo a local mobility because of the flexibility of the peptide···radical linker chain.

The CW-EPR spectra for the two TOAC- and MTSL-labeled peptides obtained at 100 K are reported in Figure A along with the simulations performed using EasySpin software. The EPR line shape is characteristic for immobilized nitroxide labels without pronounced broadening of the individual line width due to the dipole–dipole interaction. The best fitted g- and hyperfine A principle tensor values were found as gTOAC = [2.0129 2.0107 2.0065], ATOAC = [15.5 15.5 97.8] MHz, and gMTSL = [2.0141 2.0098 2.0067], AMTSL = [13.7 13.7 98.5] MHz.

Figure 10

(A) CW-EPR spectra at 100 K for the TOAC- and MTSL-labeled peptides (thick lines are the experimental spectra, thin lines are the best fits). (B) Echo-detected EPR spectrum of the TOAC-labeled peptide at 80 K in the frequency domain. The black curve shows the excitation profile of the pumping pulse, and the colored arrows show the echo detected positions in the DEER measurements.

Since the expected spin···spin distance is about 2.2 nm[66] (14 residues for an α-helical peptide), CW-EPR spectroscopy is insensitive to the spin···spin distance evaluation. This distance range is meanwhile typical for detection in the DEER experiments.[67,68] The obtained DEER time traces are presented in Figure in a normalized form [eq , see below]. The original DEER time traces are shown in Figure S8. To explore the effect known as orientational selectivity,[69−72] the DEER time traces in Figure are acquired at different frequency offsets from the pumping pulse position.

Figure 11

Normalized DEER traces for the TOAC- (A) and MTSL-labeled (B) peptides at different frequency offsets. The curves are shifted downward for convenience. (C,D) Corresponding Fourier transforms for TOAC- (C) and MTSL-labeled (D) peptides. The colors are the same as in panels (A,C). The curves are shifted downward for convenience. The vertical dotted lines show canonical orientations in the Pake-pattern.

In Figure A, one can see that, for the case of the TOAC-labeled peptide, the data strongly depend on the frequency offset, while for the MTSL-labeled peptide, the orientational effect is much less pronounced (Figure B). The Fourier transforms of the DEER time traces are illustrated in Figure C,D for the TOAC- and MTSL-labeled peptides, respectively. For the TOAC-labeled peptide, the dependence is also strongly pronounced. Suppression of the canonical dipolar orientations, either in the perpendicular orientations (Pake “horns”) or in the parallel orientations (Pake “shoulders”), is clearly visible. However, for the MTSL-labeled peptide, the frequency-domain data are nearly independent of the frequency offset.

As eq is valid for randomly distributed orientations, the determination of the pair spin···spin distribution function F(r) from eq was performed by employing the averaged DEER time traces and the corresponding Fourier transforms (Figure , left panels). The DEER traces at 90–40 MHz offsets were summed with a weighting function, accounted as an echo signal intensity at the detection position in the EPR spectrum. The experimental distance distributions F(r) for both peptides are shown in Figure (right panels). For both cases, the F(r) functions attain their maxima at r = 2.19 nm. However, the shape of the distributions is quite different: for the TOAC-labeled peptide, it is a single narrow line, while for the MTSL-labeled peptide, the line is much broader, with an additional maximum appearing near 1.95 nm.

Figure 12

Left panels show the experimental normalized V(T) time traces (black circles) averaged for different frequency offsets (see text), along with the best-fitted simulations employing the multi-Gaussian approximation (red line) and the distance discretization (blue line). In the insets, the related data in the frequency domain are illustrated. The right panels describe the obtained distance distributions employing the multi-Gaussian approximation (red curve) and the distance discretization (blue points). The peaks for F(r) seen near 1.6 and 2.9 nm are artifacts induced by the residual proton- and deuteron-induced ESEEMs, respectively.

Most probably, this distribution broadening for the MTSL-labeled peptide points to the presence of several conformers (at least two of them). As this broadening is not observed for the TOAC-labeled peptide, certainly these conformers are not intrinsic to the unlabeled peptide but are induced by the flexibility of the MTSL linker (Figure ).

Note that the small peaks in the frequency domain located at 13.6 and 2.4 MHz (insets to Figure ) are caused by contamination of the 1H and 2H electron spin echo envelope modulation (ESEEM) signals, which arise from partial overlapping of the pumping and detection pulses and are more pronounced at smaller frequency offsets, as also seen in Figure C,D. As the three-pulse DEER experimental setup was used,[4] which does not assume an averaging scheme as it is in the four-pulse DEER setup, to exclude these artificial lines in the dipolar spectrum, one is forced to use large frequency offsets between the detection and pumping pulses. We probed the detection on the low-frequency shoulder of the EPR spectrum (Figure B) with a 112 MHz offset of the pumping pulse. The data shown in Figure S9 were found to be free of these peaks, confirming the electron-nucleus coupling manifestation in the frequency spectra in Figure C,D. These peaks result in artifacts appearing in the F(r) distribution function near 1.6 and 2.9 nm for both peptides (Figure , right panels).

This newly observed impact of orientational selectivity for TOAC DEER,[69−72] reflected in Figure , invites the researchers who plan to incorporate this labeled amino acid in polypeptide molecules to select their templates accurately.

Moreover, this effect allows assessing the mutual orientation of the principal axis system and the vector connecting the two spin labels (dipolar vector). For the double-labeled molecules at X-band, this effect has already been explored.[70,71] We found that our data resemble those reported for the case where the spin labels are rigidly attached to DNA fairly well.[70] Indeed, in our case, for the expected α-helix, the dipolar vector between the TOAC[3] and TOAC[17] spin labels in the nitroxide molecular frame is almost parallel to the peptide helix (Figure ). Thus, for the case of detection at Δν ≈ 40 MHz (the detection at the equatorial orientation of the magnetic field in the nitroxide molecular framework), the Pake “horns” satisfy approximately the condition of the perpendicular orientation of the dipolar vector; meanwhile, the appearance of Pake “shoulders” at detection Δν ≈ 90 MHz (the detection at the axial orientation of the magnetic field) requires the dipolar angle to be θ ≈ 0° (the evaluation of the dipolar angle at different probed orientations is shown in Supporting Information, Figures S10 and S11). Thus, we conclude that our results may be explained if we assume two labels having nearly parallel directions for the axial axes, with the dipolar axis only slightly deviating from this direction caused by the slight difference of TOAC[3] and TOAC[17] positions due to two turns of α-helix, as shown in Figure . Therefore, the observed orientational selectivity effects for the TOAC-labeled peptide are close to the case of DNA I in ref (68). More accurate data analysis could be carried out by use of molecular modeling.

In summary, our DEER data for the TOAC-labeled peptide indicate a much narrower distance distribution as compared with that of the MTSL-labeled peptide. In the latter case, the experimental distance distribution possesses additional maxima which unambiguously point to the existence of at least two labeled conformers. It is known that one nitroxide label may occupy several distinct regions of the conformational space.[73,74] Thus, TOAC labels are much more rigid than MTSL labels and therefore allow obtaining more precise data on distance distribution.

Conclusions

In this work, we have first studied, using FT-IR absorption, CD, and 2D NMR spectroscopies, the conformational tendencies of our terminally blocked 20-amino acid long, Aib-rich, unlabeled model peptide ruler (I) in 3D structure-supporting organic solvents and found it to be largely α-helical. Having thus established the almost complete rigidity of its backbone, we have then investigated by the same physicochemical techniques our two double-TOAC and double-Cys(MTSL) synthetic analogues (peptides II and III, respectively) and have found that they essentially maintain the conformation of their sequence precursor.

Finally, we have made the first detailed comparison between the flexibility of their nitroxide-bearing side chains by the use of CW-EPR and pulsed DEER (PELDOR) techniques on the same template peptide. It turns out clearly that the TOAC side-chain labels are remarkably more conformationally restricted than the corresponding MTSL labels, thus providing more precise data on side chain-to-side chain distance distribution. In summary, although the use of Cys(MTSL) would still remain the main DEER approach to determine this type of separation in any large protein,[75] the TOAC probe is certainly the best known so far to nitroxide electron spin labeling much shorter, helical peptide-based molecules accessible by chemical synthesis. In this context, it is relevant to note that soon after we submitted the original version of the present manuscript, a very important paper[76] was published reporting benchmark tests and guidelines for DEER experiments on large proteins characterized by nitroxide probes.

Methods

Peptide Synthesis

The peptides were synthesized by the SPPS methodology according to the FastMoc protocol. The experimental conditions for the synthesis, purification, and characterization of the peptides are reported in the Supporting Information.

FT-IR Absorption

The FT-IR absorption spectra were recorded at room temperature using a PerkinElmer model 1720X FT-IR absorption instrument, nitrogen flushed with a sample shuttle device at 2 cm–1 nominal resolution, averaging 100 scans. Solvent (spectrograde deuterated chloroform, 99.8% d, Merck, Darmastadt, Germany) spectra were obtained under the same conditions. Cells with path lengths of 0.1 and 1.0 cm were used.

CD

The electronic CD curves in the far-UV region (below 260 nm) were obtained on a Jasco (Tokyo, Japan) model J-1500 spectropolarimeter with a fused quartz cell of 0.02 cm pathlength (Hellma, Mühlheim, Germany). The values are expressed in terms of [θ]T, the total molar ellipticity (deg × cm2 × dmol–1). Spectra were recorded at room temperature in water and in spectrograde MeOH (99.9% Acros Organics, Geel, Belgium) at 3 × 10–4 M peptide concentration. A second set of measurements was acquired in the near-UV and visible regions (300–600 nm) by the use of a 0.1 cm pathlength quartz cell and a 10–3 M peptide concentration in MeOH solution to investigate the contribution of the n → π* transition of the nitroxyl chromophore in the absence of the competition with the amide (peptide) chromophores.[41,60,62]

NMR

The monodimensional and correlated spectroscopy (COSY)/total correlation spectroscopy (TOCSY)/NOESY 2D NMR spectra of the unlabeled peptide (I) were obtained at 298 K by the use of a Bruker AVANCE NEO-600 spectrometer from an about 1 mM concentration sample dissolved in approximately 700 μL of MeOH, d3 solution. Suppression of the solvent signal was achieved by the use of an excitation sculpting program.[77] All homonuclear spectra were acquired by collecting 512 experiments, each one consisting of 64–80 scans and 2 K data points. The spin systems of the coded amino acid residues were identified using standard double-quantum filtered-COSY[78] and clean TOCSY[79] spectra. In the latter case, the spin-lock pulse sequence was 70 ms long. NOESY experiments were utilized for sequence-specific assignments.[65] To avoid the problem of spin diffusion, the build-up curve of the volumes of the NOE cross peaks as a function of the mixing time (50–500 ms) was obtained first (data not shown). The mixing time of the NOESY experiment used for interproton distance determination was 150 ms (i.e., in the linear part of the NOE build-up curve).

Modeling

Peptide modeling was designed by using PyMOL open source software, a molecular visualization program created by W. L. DeLano and commercialized by Schrödinger Inc.[80,81] Currently, PyMOL is one of the most utilized programs of this type in the area of macromolecules.

Electron Spin Resonance Measurements

The EPR experiments were carried out on two different X-band Bruker ELEXSYS E580 EPR spectrometers using either an ER 4118 X-MS-3 or ER 4118 X-MD-5 resonator in an Oxford Instruments CF-935 cryostat.

For measurements at room temperature, the samples were deaerated by a nitrogen flow and contained in 1 mm i.d. tubes. The spectra were acquired with a microwave (mw) power of 0.5 mW and a modulation amplitude of 1 G (modulation frequency of 100 kHz). Three scans were summed to improve the S/N ratio.

For low-temperature measurements, the peptides were dissolved in deuterated methanol (d4) at a concentration of 200 μM, put in EPR tubes, and frozen quickly in liquid nitrogen to form a glass.

The cryostat was cooled by nitrogen flow to 80 K for pulse EPR measurements or to 100 K for CW-EPR measurements. In the CW-EPR measurements, the mw power was attenuated to −45 dB. The calibrated, nonattenuated output mw power of the Gunn diode was 200 mW. The modulation frequency and the modulation amplitude were 100 kHz and 0.1 mT, respectively. The time constant was 20.48 ms, and the conversion time was 40.96 ms.

The CW-EPR spectra were simulated by EasySpin software.[82]

The DEER experiments were performed with the three-pulse sequence tpulse(νA)–T–tpulse(νB)–(τ – T)–2tpulse(νA)–τ–echo, where the subscripts include notations for the detection (νA) and pumping (νB) mw frequencies. The time delay T for the pumping pulse was initially set to 200 ns prior to the first detection pulse and then scanned with a 4 ns time step. The frequency offset νA–νB was chosen symmetrically around the absorption dip in the resonator and varied from 40 to 90 MHz and. The tpulse(νA) duration was 16 ns, the tpulse(ν duration was 28 ns. The amplitude of the pumping pulse was set to inverse the primary echo at the maximum of the echo-detected EPR spectrum. The amplitude of the detection pulse was chosen to provide the maximal echo signal. The time τ in the DEER measurements was 1600 ns. The noise amplitude on the normalized orientationally averaged DEER traces was 0.005–0.007. The DEER signal distortions upon the pumping pulse passage through the detecting pulses were corrected as already described.[83]

As the experimental DEER signal time-dependence is assumed to be a product of the intra- and intermolecular dipolar contributions, the former contribution was refined by approximating the latter by a stretched exponential function.[84] After elimination of this function, the resulting intramolecular part of the DEER time trace VINTRA(T) was normalized according to Milov et al.[85]

The experimental modulation depth in the DEER experiments, λ ≡ 1 – Vintra(∞), was compared with the calculated one, λcalc.[67] The comparison for both the TOAC- and Cys(MTSL)-labeled peptides is given in Supporting Information, Figure S6.

The experimental V(T) time traces may be directly identified with the theoretical expression for the DEER signal time-dependence[85]where F(r) is related to the distance distribution function normalized in the way that , g1 and g2 are the g-factors of the two spins, μB is the Bohr magneton, and θ is the angle between the external magnetic field and the vector connecting the spin pair. The experimental V(T) time traces were cosine Fourier transformed.

The pair distance distribution function F(r) was recovered by solving the integral eq for which the equality f(t) = V(T) was assumed. To avoid the instability of the solution because of the ill-posed nature of the problem,[68] two independent regularization approaches were used: (i) by the distance discretization with the Monte Carlo search of the solution[86] and (ii) by the multi-Gaussian Monte Carlo fitting.[87,88] Both approaches employ a fitting in the DEER frequency domain. The former approach is certainly model-free, while the latter produces smooth F(r) functions possessing meanwhile a simple physical meaning.