Gd3+ Spin Labels Report the Conformation and Solvent Accessibility

May 22, 2015 - Gd3+ Spin Labels Report the Conformation and Solvent Accessibility of Solution and Vesicle-Bound Melittin. Nurit Manukovsky†, Veronic...
0 downloads 6 Views 2MB Size
Article pubs.acs.org/JPCB

Gd3+ Spin Labels Report the Conformation and Solvent Accessibility of Solution and Vesicle-Bound Melittin Nurit Manukovsky,† Veronica Frydman,‡ and Daniella Goldfarb*,† Departments of †Chemical Physics and ‡Chemical Research Support, Weizmann Institute of Science, Rehovot 76100, Israel S Supporting Information *

ABSTRACT: Although Gd3+-based spin labels have been shown to be an alternative to nitroxides for double electron− electron resonance (DEER) distance measurements at high fields, their ability to provide solvent accessibility information, as nitroxides do, has not been explored. In addition, the effect of the label type on the measured distance distribution has not been sufficiently characterized. In this work, we extended the applicability of Gd3+ spin labels to solvent accessibility measurements on a peptide in model membranes, namely, large unilamellar vesicles (LUVs) using W-band 2H Mims electron−nuclear double resonance (ENDOR) and electron spin echo envelope modulation (ESEEM) techniques and Gd3+ADO3A-labeled melittin. In addition, we carried out Gd3+−Gd3+ DEER distance measurements to probe the peptide conformation in solution and when bound to LUVs. A comparison with earlier results reported for the same system with nitroxide labels shows that, although in both cases the peptide binds parallel to the membrane surface, the Gd3+-ADO3A label tends to protrude from the membrane into the solvent, whereas the nitroxide does the opposite. This can be explained on the basis of the hydrophilicity of the Gd3+-ADO3A labels in contrast with the hydrophobicity of nitroxides. The distance distributions obtained from different labels are accordingly different, with the Gd3+ADO3A yielding consistently broader distributions. These discrepancies are most pronounced when the peptide termini are labeled, which implies that such labeling positions may be inadvisible.



INTRODUCTION Nitroxide spin labels have been extensively used for EPR studies of biological systems for decades,1 and have provided rich biochemical information such as intramolecular distances,2−4 local mobility,5,6 and solvent accessibility7−10 of specific sites within the biomolecule. Recently, a new family of spin labels, based on the high spin (S = 7/2) Gd3+ ion, has been introduced for double electron−electron resonance (DEER) distance measurements at W-band (95 GHz) and Q-band (34 GHz) frequencies.11−15 These labels offer high sensitivity, and the measurements are free of orientation selection effects, which for nitroxide spin labels complicate W-band DEER measurements and data analysis.16,17 Moreover, the stability of Gd3+ spin labels toward reduction allows their use in in-cell distance measurements.18,19 Gd3+ labels have been successfully used for Gd3+−Gd3+ distance measurements on peptides in solution20 and in model membranes,21 proteins,12,22,23 DNA,11 and nanoparticles.24 Nevertheless, the ability of Gd3+ to provide information beyond intramolecular distances, such as solvent exposure, as nitroxides do,10,25 has not been tested. To date, Gd3+−Gd3+ DEER has been carried out primarily in solution,12,20,22 but it has been recently applied to a waterinsoluble transmembrane peptide in model membranes.21 Membrane DEER is known to be associated with difficulties arising from the labeled protein localizing to the limited area of © 2015 American Chemical Society

the membrane, thereby increasing its local concentration, leading to a major reduction in the phase memory time and to faster DEER background decay, compared with the state in solution. This phenomenon reduces sensitivity, compromises access to long distances, and complicates background correction owing to intermolecular spin−spin interactions.26−28 Gd3+−Gd3+ DEER of a transmembrane peptide revealed that, although the phase memory time is shorter than what is usually found in solution, high-quality DEER data could be collected up to 4.5 μs using the standard four-pulse DEER sequence.29 This work aims to expand the potential of Gd3+ spin labeling when studying peptide−membrane interactions, focusing not only on the conformation of the membrane-bound peptide but also on its membrane penetration depth following the approach applied for nitroxide spin labels.9,30−33 We chose melittin, a water-soluble peptide, which, in the presence of membranes, aligns in the membrane plane. This peptide has been studied earlier using nitroxide spin labels in the presence and absence of model membranes31,34−36 and using Gd3+ spin labels only in solution.20 Hence, a comparison with these earlier data would Special Issue: Wolfgang Lubitz Festschrift Received: April 12, 2015 Revised: May 14, 2015 Published: May 22, 2015 13732

DOI: 10.1021/acs.jpcb.5b03523 J. Phys. Chem. B 2015, 119, 13732−13741

Article

The Journal of Physical Chemistry B

turn, reveals the peptide’s orientation with respect to the membrane. Since applying 2H ESEEM at the W-band is expected to be hampered by the inverse dependence of the modulation depth on the magnetic field strength, leading to shallow modulations (i.e., limited sensitivity), we also explored the applicability of 2H Mims electron−nuclear double resonance (ENDOR) for the same task. Ku-band 2H Mims ENDOR39 and 17O W- and D-band40 Mims ENDOR have been used for quantifying the coordinating water molecules in various Gd3+ complexes. 31P and 1H W-band Mims and Davies ENDOR have been used to count phosphate and water ligands in Mn2+−ATP complexes,41 and 13C Mims and Davies ENDOR have been used to count ligands in Mn2+−bicarbonate complexes.42 While these works focused on ligand counting, here we probed water molecules in the close environment of the spin label, and not direct ligands. Our results indicate that Gd3+-ADO3A labels can be used to determine solvent accessibility of peptides in solution and when interacting with vesicles. In addition, they highlight the effects of the spin label on its location. Both W-band 2H-ESEEM and 2 H-ENDOR produce a similar general picture of the peptide lying in the membrane plane, in agreement with previous works using nitroxides. However, discrepancies have been observed regarding the details, particularly in the peptide termini, which can be attributed to the hydrophilicity of the Gd3+-ADO3A labels in contrast with the hydrophobicity of nitroxides. Variations have also been observed between distance distributions obtained using Gd3+-ADO3A or nitroxide, where Gd3+-ADO3A displays broader distributions than does nitroxide both with and without vesicles. A shortening of distance in vesicle vs solution was observed, as for the nitroxide-labeled melittin, but not in all mutants. In addition, Gd3+-ADO3Alabeled melittin resulted in a broader distribution than Gd3+4MMDPA (4-mercaptomethyl-dipicolinic acid)-labeled melittin in solution without vesicles.20 The DEER differences were also interpreted on the basis of the different chemical nature of the labels, whose structures are shown in Figure 1.

enable us to characterize the effects of different labels, namely, Gd3+ vs nitroxides, on the system under study. A hydrophobic bias of the nitroxide spin label vs the Gd3+ spin label was clearly resolved in the transmembrane helix study.21 Melittin is a 26-residue-long amphiphilic, membrane-active antimicrobial peptide found in bee venom.37,38 It has been studied using nitroxide spin labels by the Hubbell group34,35 and ours.31,36 In solution, CW-EPR of nitroxide-labeled melittin exhibited high mobility of the spin labels at all positions,31,34 in line with an unfolded random coil. In the presence of phospholipid vesicles, however, the peptide was found to be oriented parallel to the membrane surface, as indicated by solvent accessibility of spin labels introduced at various positions along the sequence. This was observed by the EPR intensity decrease34 and a shorter T135 in the presence of a hydrophilic relaxant, as well as by solvent-accessibility measurements using 2H electron spin echo envelope modulation (ESEEM).31 Monte Carlo simulations on nonlabeled melittin produced a similar picture, except for the peptide termini, which protruded into the solvent in the simulations, in contrast with the EPR results, which showed that they were buried in the membrane. This discrepancy was attributed to the hydrophobicity of the nitroxide, bringing it deeper into the membrane.31 The use of a different type of spin label with different properties can help resolve this discrepancy. Intramolecular distance measurements using DEER on singly and doubly nitroxide-labeled melittin showed that the peptide is monomeric when bound to vesicles, and that it displays shorter, less distributed distances in the presence of vesicles than in solution, indicating that it becomes more structured upon interaction with vesicles.31 We hereby carried out DEER measurements on melittin labeled by Gd3+-MTS-ADO3A (2,2′,2″-(10-{2-[2-(methylsulfonylthio) ethylamino]-2-oxoethyl}-1,4,7,10-tetraazacyclododecane-1,4,7-triyl)triacetic acid), a DOTA derivative (Figure 1a),



EXPERIMENTAL SECTION Materials. DPPC (1,2-dipalmitoyl-sn-glycero-3-phosphocholine; ≥99%) was supplied by Sigma-Aldrich (St. Louis, MO). Egg-PG (L-α-phosphatidylglycerol; ≥99%) was supplied by Avanti Polar Lipids, Inc. (Alabaster, AL). D2O and glycerold8 were supplied by Cambridge Isotope Laboratories (Andover, MA). MTS-ADO3A (2,2′,2″-(10-{2-[2-(methylsulfonylthio) ethylamino]-2-oxoethyl}-1,4,7,10- tetraazacyclododecane-1,4,7triyl)triacetic acid) (Figure 1a) was synthesized according to the published procedure,43 and complexation was carried out by mixing the chelate with a 1.1 excess of GdCl3 in water, adjusting the pH to ∼5.5 using NaOH, and agitating the sample at room temperature for 24 h. Chelation was confirmed using electrospray mass spectrometry. DOTA (1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid) was obtained from Macrocyclics (Dallas, TX), and complexation was carried out as described above. Peptides were purchased from GenScript (Piscataway, NJ) and LifeTein (South Plainfield, NJ). Five single melittin mutants and four double mutants were labeled with Gd3+MTS-ADO3A. They are named according to their labeling position (Table 1). To directly compare the labels, the mutants were the same as those studied with nitroxide.31 For spin labeling, the peptides were dissolved in Tris−HCl buffer pH 7.6

Figure 1. (a) Gd3+-MTS-ADO3A label. (b) MTSSL, the nitroxide label. (c) Gd3+-4MMDPA label.

and compared the results with those obtained using the nitroxide MTSSL ((1-oxyl-2,2,5,5-tetramethyl pyrroline-3methyl)methanethiosulfonate) labels to isolate the effect of the spin labels’ chemical nature and to compare their performance in model membrane systems, i.e., vesicles. In addition, we tested the potential of applying hyperfine spectroscopic techniques to Gd3+-ADO3A spin labels to provide information regarding solvent exposure, which, in 13733

DOI: 10.1021/acs.jpcb.5b03523 J. Phys. Chem. B 2015, 119, 13732−13741

Article

The Journal of Physical Chemistry B Table 1. Melittin Mutants’ Designation and Sequencea designation mel-N mel-C3 mel-C15 mel-C18 mel-C27 mel-NC15 mel-NC27 mel-C3C18 mel-C15C27 a

(v/v) mixture of CHCl3/MeOH and drying them under a stream of dry nitrogen gas while rotating them. The films were vacuum-dried overnight, sealed with argon gas, and stored at −20 °C until used. On the measurement day, the films were suspended in 50 mM pH 7.2 phosphate buffer in D2O by preheating the buffer to ∼50 °C and maintaining the films at ∼50 °C (above the phase transition temperature of DPPC, 42 °C) for 1 h, while frequently vortexing them. The resulting multilamellar vesicles were extruded using an Avestin (MM Developments, Ottawa, Canada) Liposofast Extruder through polycarbonate membranes with 1, 0.2, and 0.1 μm diameter pores44,45 to form LUVs. Peptide was added from a stock D2O solution, and the suspension was pipetted for mixing and then added to a preweighed portion of glycerol-d8, used to prevent ice formation, to allow a 30% v/v glycerol-d8 concentration, and then pipetted. Volumes of ∼3 μL were then loaded into 0.6 mm ID, 0.84 mm OD quartz capillaries from Vitrocom, flash frozen in liquid nitrogen, and kept frozen. All measurements were carried out on frozen solutions. It has been shown that the vesicle structure is maintained upon flash freezing.36 EPR Spectroscopy. All EPR measurements were carried out at 10 K on a home-built W-band (95 GHz) spectrometer.46 Echo detected (ED) EPR spectra were recorded using the twopulse echo sequence, π/2-τ-π-τ-echo, with pulse durations of 15 and 30 ns, respectively, τ = 600 ns, and a 180° phase cycle on the first pulse, while sweeping the magnetic field. Phase memory time measurements were carried out using the same pulse sequence while incrementing τ from an initial value of 150 ns by steps of 25 ns. T1 measurements were carried out using the saturation recovery sequence tsat-t-π/2-τ-π-τ-echo, with a saturation pulse length, tsat, of 100 μs, detection pulses of 30 and 60 ns, a 180° phase cycle on the π/2 pulse, and τ = 550 ns. The value of t was incremented from an initial value of 5 μs by 100 steps of 20 μs. The repetition time was 3 ms, and 30 shots were taken per point.

sequence Singly Labeled Mutants CGIGAVLKVLTTGLPALISWIKRKRQQ GICAVLKVLTTGLPALISWIKRKRQQ GIGAVLKVLTTGLPCLISWIKRKRQQ GIGAVLKVLTTGLPALICWIKRKRQQ GIGAVLKVLTTGLPALISWIKRKRQQC Doubly Labeled Mutants CGIGAVLKVLTTGLPCLISWIKRKRQQ CGIGAVLKVLTTGLPALISWIKRKRQQC GICAVLKVLTTGLPALICWIKRKRQQ GIGAVLKVLTTGLPCLISWIKRKRQQC

C indicates the spin-label position.

and a Gd3+-MTS-ADO3A solution was added to yield a 5-fold excess of tag/cysteine. The solution was agitated at room temperature for 16 h, and then, the labeled peptides were purified using reverse-phase HPLC. The labeling was verified using electrospray mass spectrometry. Sample Preparation. The peptides were studied in solution (0.05 mM peptide in D2O:glycerol-d8 7:3) and in vesicles (0.03 mM bulk concentration for a doubly labeled peptide or 0.05 mM bulk concentration for a singly labeled peptide in DPPC:egg PG 7:3 LUVs, peptide/lipid = 1/2000, in D2O buffer:glycerol-d8 7:3). This low peptide concentration and high peptide/lipid ratio were used in order to prevent a high local concentration of the peptide when localized to the limited surface area of the vesicles, which is known to drastically shorten the phase-memory time.28 This allowed reaching evolution times up to 5 μs. LUVs (large unilamellar vesicles) were prepared by hydrating lipid films to produce multilamellar vesicles, which were subsequently downsized to LUVs. Thin lipid films were generated by dissolving the required weights of lipids in a 2:1

Figure 2. (a) Representative two-pulse ESEEM trace of Gd3+-melittin (C3-melittin in LUVs). The inset shows a zoom-in, where the modulations are clearer. (b) The trace in part a after division by background decay and translation to remove the offset. (c) FFT of part b, showing the definition of I(2H), used as a comparative parameter. 13734

DOI: 10.1021/acs.jpcb.5b03523 J. Phys. Chem. B 2015, 119, 13732−13741

Article

The Journal of Physical Chemistry B

maximum ENDOR effect, respectively. τ was set to 1.25 μs, which results from a compromise between a stronger ENDOR effect for longer τ values in Mims ENDOR and the closer-tolinear behavior of the ENDOR effect vs the D2O content for shorter τ values,50 as determined for Gd3+-DOTA solutions containing various D2O/H2O ratios (Figure S1 in the Supporting Information). Four-Pulse DEER. The π/2(νobs)-τ-π(νobs)-τ+t-π(νpump)-(Tτ-t)-π(νobs)-(T-τ)-echo pulse sequence29 was used, with pulse durations of tπ/2(obs) = 15 ns, tπ(obs) = 30 ns, tpump = 15 ns and a four-step phase cycle for the observer pulses, τ = 350 ns, a repetition time of 0.8 ms, and 50 shots per point. The pump frequency was set to the maximum of the Gd3+ spectrum, and the observer frequency was 110 MHz higher. The echo was measured as a function of t, which was incremented from an initial value around −200 ns at steps ranging from 25 to 50 ns, depending on the observed distance. The time T was between 3.5 and 5 μs, depending on the distance, and the number of steps for t increment was determined accordingly. Accumulation times were between 2 and 20 h, depending on the evolution time and the signal-to-noise ratio. The DEER data were processed using DeerAnalysis.51 The background was fit to a homogeneous decay for solution samples and a stretched exponential one with a dimensionality of ∼2.5 for LUV samples, as observed for a singly labeled mutant in LUVs (Figure S3 in the Supporting Information).

Two-Pulse ESEEM. The π/2-τ-π-τ-echo pulse sequence was used, with pulse durations of 15 and 30 ns, respectively, and a 180° phase cycle for the first pulse. The value of τ was incremented from an initial value of 382.5 ns by 94 steps of 7.5 ns. This initial value was used because of a small glitch that appeared at ∼300 ns. The repetition time was 1 ms, and 30 shots were taken per point. The measurements were carried out at the field position corresponding to the maximum of the Gd3+ spectrum. Accumulation times were between 10 and 30 min, depending on the signal-to-noise ratio. The time domain trace was normalized to a maximum intensity of 1, fitted with a biexponential decay, and divided by it. Then, it was translated such that the remaining signal was modulated around zero, zero filled to 1024 points, and apodized with a Hamming window. To eliminate possible dead-time-dependent distortions in the absolute-value Fourier transform spectra, cross-term averaging47 was used in the FFT, and finally, the magnitude spectrum was calculated. A baseline was visually interpolated between the plateau at most of the frequencies and the tail of the peak at the 2H Larmor frequency, and then was subtracted. The remaining intensity of the FT peak at the 2H Larmor frequency, termed I(2H), was used to represent the intensity of the 2H modulations in the time domain.48 This procedure is illustrated in Figure 2. 2 H-Mims ENDOR. The π/2-τ-π/2-T-π/2-τ-echo pulse sequence, with an RF pulse applied during the time interval T sequence, was used, with tπ/2 = 15 ns, along with a two-step phase cycle on the last pulse. The RF pulse duration was calibrated using a nutation experiment in which the echo intensity was monitored vs the length of a 22.25 MHz RF pulse, and typical values were 80−90 μs. T was set to be 10 μs longer than the RF pulse. The repetition time was set to be 20 times longer than the RF pulse, due to the RF amplifier duty cycle limitation of 5%. The spectra were recorded using the random acquisition mode,49 with 2 shots per point. The measurements were carried out at the field position corresponding to the maximum of the Gd3+ spectrum. Accumulation times were ∼15 min. The ENDOR effect was calculated by ε=



RESULTS AND DISCUSSION Echo-Detected EPR and Echo-Decay Measurements. The echo-detected EPR spectra of all singly labeled melittin mutants, whether in solution or in LUVs, had the typical general shape of Gd3+-ADO3A in frozen solutions (see Figure 4). The spectra consist of a sharp singlet, corresponding to the

1 I21.25 − I22.25 2 I21.25

where I21.25 and I22.25 are the echo intensities in the presence of RF irradiation of 21.25 or 22.25 MHz (see Figure 3). These values were found to correspond to no ENDOR effect or the

Figure 4. Representative EPR spectra of a melittin mutant (C18) in solution vs in LUVs. The inset shows a close-up of the central line. Small peaks due to Mn2+ impurities are observed in the sample in the presence of LUVs.

central, |−1/2⟩ ↔ |1/2⟩ transition, superimposed on a broad background attributed to all other transitions.52 Interestingly, we observed a subtle but consistent broadening in LUVs vs solution, the origin of which is currently not understood. The phase-memory time is a limiting factor for DEER measurements, especially in vesicle samples. Accordingly, a shorter phase-memory time in vesicles than in solution is indicative of peptides binding to the vesicle because localization of the peptide to the limited surface area of the membrane

Figure 3. Representative 2H Mims ENDOR spectrum of C3-melittin in D2O/glycerol-d8 7:3, showing the RF frequencies used for ENDOR. 13735

DOI: 10.1021/acs.jpcb.5b03523 J. Phys. Chem. B 2015, 119, 13732−13741

Article

The Journal of Physical Chemistry B

Solvent Accessibility. 2H ESEEM and ENDOR in deuterated solvent were used to probe the solvent accessibility of labeled positions along the peptide in solution and in LUVs. The 2H ESEEM modulation depth is expected to be weak at high fields because it is proportional to ν0−2, where ν0 is the spectrometer frequency.55 Nonetheless, the modulation was still clear and detectable, as shown in Figure 2, which shows the two-pulse ESEEM trace of C3-melittin in LUVs as an example. The intensity of the FT peak at the 2H Larmor frequency, termed I(2H), obtained after background subtraction, was used as a comparative parameter (see Figure 2). A calibration curve of ESEEM using Gd3+-DOTA solutions containing various D2O/H2O ratios and 30% v/v glycerol-d8 shows that I(2H) increases linearly with the 2H percentage (Figure 6a); therefore, an increase in I(2H) indicates an increased solvent exposure. In the presence of both H2O and glycerol-d8, exchange processes lead to the formation of seven species (H2O, D2O, HDO, and glycerol-d8,d7,d6,d5) at different relative concentrations. The methods used here cannot resolve the different species; therefore, we plotted I(2H) as a function of the total 2H content. The existence of a linear trend indicates that this method can be used as an indicator for the total, average 2H density around the label, representing the solvent as a whole. Figure 6b presents I(2H) versus the labeling position in singly labeled melittin mutants studied in solution and in LUVs. The resulting ESEEM hydration profile of the peptides shows that in solution all positions experience nearly complete hydration, as judged by comparison with the calibration curve. With vesicles, the I(2H) values approach those in solution, within experimental error, except for positions N and C3, where they are significantly lower than those in solution. The wide error range of ESEEM, also manifested by the dispersion in the calibration curve, is due to the low modulation depth. Because of the very low 2H modulation depth at the W-band and the very small difference observed between melittin in solution and in LUVs, we turned to 2H Mims ENDOR as a means for providing solvent accessibility data, focusing on the matrix signal at the 2H Larmor frequency. A calibration curve of the ENDOR effect at the 2H Larmor frequency (see Figure 3 for an example of an ENDOR spectrum) obtained using the same series of Gd3+-DOTA solutions used for the ESEEM calibration curve is presented in Figure 6c. It shows a linear dependence on the 2H percentage. We noted that the linearity of this curve is τ-dependent, and thus, it required optimization of τ (Figure S1 in the Supporting Information). Although a long τ value allows for a larger ENDOR effect, it leads to larger deviations from linearity. The ENDOR effect profile of the peptide in solution and in the presence of LUVs is shown in Figure 6d. A similar, complete hydration of all positions is observed in solution. With vesicles, significant differences in the hydration are observed for positions N, C3, and C15, where the hydration in vesicles is lower than in solution. For the other positions, although the average value in vesicles is consistently lower than that in solution, the differences are within the experimental error. This is similar to the ESEEM results, though in that case the difference for C15 was not resolved. The Gd3+ ESEEM and ENDOR results are in reasonable agreement with each other, and both support a model where the peptide is oriented parallel to the membrane, with the Gd3+ spin label highly exposed to the solvent. Both methods show that the hydration is lowest for C3 and that in general the Nterminal region is buried deeper in vesicles than in solution, whereas in the C-segment there is practically no difference

increases its local concentration, consequently accelerating relaxation. The high concentration of 1H in the membrane may add to this effect.28 A counter effect may also exist, where, owing to the lower dimensionality of the peptide’s spatial distribution on the vesicle’s surface area, the echo decay follows a stretched exponential whose decay is slower than that of a simple one.53 Echo-decay measurements were therefore carried out to explore the feasibility of DEER, as well as to indicate whether the peptide binds to the vesicles. The echo decay curves (Figure 5a) displayed the behavior of a stretched

Figure 5. (a) Representative echo decay kinetics of Gd3+-labeled melittin (mel-N) in solution (black) and LUVs (red) measured at the central transition, showing the definition of τ10%, used as a comparative parameter. (b) Values of τ10% measured for melittin mutants in solution (black) and LUVs (red) at the central transition (full symbols) or 110 MHz downfield (empty symbols). Where error bars are shown, they represent the standard error of a duplicate of independently prepared samples, and the value shown represents their average.

exponential function, v(τ) = exp[−(τ/TM)n], with different powers (n) ranging between 0.8 and 1.4 among various samples. Because this work does not focus on the phase relaxation mechanism, the comparative parameter used in Figure 5 was not the decay constant, TM, but instead, it was the time required for the echo to decrease to 10% of its initial intensity, τ10%. The τ10% values are consistently shorter in the presence of vesicles than in solution. This indicates that the labeled mutants bind to the LUVs. The same behavior was observed in measurements carried out at the maximum of the central peak and 110 MHz (3.9 mT) downfieldthe spectral positions used for the pump and observer pulses, respectively. In addition, for both types of samples, the echo decay was always slower for the central transition as compared with the broad background, consistent with a recent study on the transition-dependent phase memory time of Gd3+-DOTA in a frozen solution.54 Spin−lattice relaxation measurements were carried out as well, and no dependence on the label position or the presence of LUVs was observed (data not shown). 13736

DOI: 10.1021/acs.jpcb.5b03523 J. Phys. Chem. B 2015, 119, 13732−13741

Article

The Journal of Physical Chemistry B

Figure 6. (a) Calibration curve of W-band Gd3+ ESEEM: I(2H) vs 2H percentage. (b) I(2H) of melittin mutants in solution and LUVs. (c) Calibration curve of W-band Gd3+ ENDOR: ENDOR effect vs 2H percentage. (d) ENDOR effect of melittin mutants in solution and LUVs. Where error bars are shown, they represent the standard error of a duplicate of independently prepared samples, and the value shown represents their average. (e) NO-melittin X-band ESEEM: I(2H) of melittin mutants in solution and LUVs. Nitroxide data taken from ref 31.

Figure 7. Background-corrected DEER traces of four doubly labeled melittin mutants: (a) NC15, (b) NC27, (c) C3C18, and (d) C15C27. Each panel shows traces in solution (black) and in LUVs (red).

between solution and vesicles within the error range. Interestingly, the NMR parameters of vesicle-bound melittin show the hydrophobic N-terminal region penetrates into the membrane, whereas the hydrophilic C-terminal area probably does not.56 The use of ESEEM at high fields is hampered by the shallow modulation, limiting the sensitivity and resolution of the degree of hydration. However, ESEEM is in general more robust than

ENDOR; the ENDOR effect is a function of many factors such as the RF amplifier response function, making it harder to use quantitatively. Moreover, the linearity of ENDOR with respect to the solvent accessibility is only an approximation.50 For comparison, in Figure 6e, we show the 2H X-band ESEEM results obtained with nitroxide radicals.31 In solution, the differences between labeling positions are resolved with nitroxide labels, whereas for Gd3+-ADO3A labels the differences 13737

DOI: 10.1021/acs.jpcb.5b03523 J. Phys. Chem. B 2015, 119, 13732−13741

Article

The Journal of Physical Chemistry B

Figure 8. Distance distributions obtained for four doubly labeled melittin mutants: (a) NC15, (b) NC27, (c) C3C18, and (d) C15C27. Each panel shows the distance distributions in solution (black) and in LUVs (red), obtained using Gd3+ (top) or nitroxide in X-band (bottom). The C15C27 mutant also shows the Gd3+-4MMDAP distribution in solution, taken from ref 20. Nitroxide data are taken from ref 31. All distributions are normalized to maximum intensity.

Intramolecular Distances. Time traces and distance distributions obtained from DEER on doubly labeled Gd3+ADO3A melittin are shown in Figures 7 and 8 (the raw time domain DEER data are presented in the Supporting Information, Figure S2). The low bulk melittin concentration and the low peptide/lipid ratio (1/2000) allowed us to extend the DEER evolution time up to 5 μs. The low ratio also minimizes the amount of unbound peptide in solution and decreases the background decay. Control DEER measurements on singly labeled melittin-C15 showed no homo-oligomerization in vesicles; however, there was a small amount of it in solution (Figures S3 and S4 in the Supporting Information). This is in line with reports showing a monomer−tetramer equilibrium in solution.56 However, in light of the modulation depth of 0.7% in the singly labeled mutant (Figure S4, Supporting Information) vs 3.5−7% in the doubly labeled ones, the homo-oligomerization appears to be very minor. Gd3+-DEER shows broad distance distributions in solution, in agreement with NMR and circular dichroism results showing that in solution the peptide is mostly a random coil.31,56,57 Earlier DEER measurements carried out on C15C27 with the Gd3+-4MMDPA label in solution yielded a significantly narrower distance distribution20 (see Figure 8d), and the maximum of the distribution was 3.6 nm as compared with 2.9 nm for Gd3+-ADO3A. Another study on proteins labeled with Gd3+-4MMDPA or nitroxide showed the Gd3+-4MMDPA to yield longer distances by 0.4−0.9 nm than nitroxide, and a similar width of the distance distributions.12 These differences are attributed to the differences in the chemical nature of the label (Gd3+-4MMDPA has a charge of +1, whereas Gd3+ADO3A is neutral) that adopts different locations with the lowest energy with respect to the peptide. The tether of Gd3+4MMDPA is also shorter than that of Gd3+-ADO3A (Figure 1). Moreover, Gd3+-4MMDPA has more free coordination sites In the presence of LUVs, two mutants, NC15 and C3C18, display a clear shift to shorter distances upon vesicle binding,

are within the experimental error. This could be ascribed to the higher sensitivity of the X-band ESEEM. It is also possible that the hydrophobic nitroxide nests in hydrophobic grooves, whereas the hydrophilic Gd3+-ADO3A protrudes away from them. The difference between the solution and LUVs is also more pronounced in the nitroxide-labeled melittin. Although the relative hydration profile within the peptide center is similar for the two labels in LUVs, differences are observed for the termini. For the nitroxide-labeled peptide, the termini displayed deeper penetration than did the center of the peptide, whereas for the Gd3+-ADO3A label they were more exposed. Earlier Monte Carlo simulations carried out on melittin, where the nitroxide-labeled cysteines were replaced by leucines to account for the hydrophobicity of the nitroxide, predicted markedly less penetration of the termini than that observed by nitroxide ESEEM. This was hypothesized to be due to some bias of the labeling in the termini.31 Apparently, the hydrophobic bias of the nitroxide makes it preferably reside deeper in the membrane than does the native residue, whereas the hydrophilic bias of Gd3+-ADO3A causes the opposite effect, bringing Gd3+-ADO3A to preferably protrude from the membrane into the solvent. These behavioral patterns of both labels were also observed using DEER on a transmembrane peptide.21 In that work, the Gd3+-C1 label, which is bulkier than ADO3A, was buried deeper in the membrane than was ADO3A, which can be explained by the higher hydrophobicity of C1; it therefore appears that the bulkiness of ADO3A is not the major cause of its behavior. In both cases, the labels’ bias is likely to be more pronounced in the termini, where the labels have higher conformational freedom. Comparing the Gd3+ results with the MC predictions for native melittin shows that the N and C3 positions are still more buried in the membrane than predicted by the MC simulations, in spite of the label’s higher hydrophilicity. This suggests that some uncertainties in the MC simulation regarding the peptides’ ends also contribute to this discrepancy.31 13738

DOI: 10.1021/acs.jpcb.5b03523 J. Phys. Chem. B 2015, 119, 13732−13741

Article

The Journal of Physical Chemistry B

solution to LUVs but no shortening of the distance. In fact, NC27 was even predicted to slightly elongate upon LUV binding.31 In terms of the distance distribution width, the Gd3+ADO3A DEER results therefore disagree with the MC results, in contrast with the nitroxide distance distributions, which correlated well with those of the MC simulations. This can be understood in light of the great difference between leucine and Gd3+-ADO3A. Comparing all three labels, Gd3+-ADO3A, Gd3+-4MMDPA, and MTSSL, in solution (the C15C27 mutant, Figure 8d) shows that 4MMDPA provides the longest measured distance, with a distribution width comparable to that of nitroxide. Since the tether of 4MMDPA is the same as that of MTSSL, and their molecular volumes are comparable, the difference between the corresponding distributions is only the result of the label bias (Gd3+-4MMDPA has a net charge of +1). The difference between 4MMDPA and ADO3A most likely results from the difference in charge and volume. The bulkier ADO3A might interfere with the helical structure shown by MC simulations and nitroxide DEER to exist even in solution in residues 15− 25.31 Another possibility is that the smaller width of the nitroxide and the 4MMDPA distance distributions results from the nitroxide nesting in a hydrophobic groove or 4MMDPA interacting with the terminal carboxylate. The much narrower distribution obtained with 4MMDPA compared with ADO3A is an advantage of the former; however, 4MMDPA suffers from the problem of a lower affinity to Gd3+ compared with ADO3A, leading to the presence of free Gd3+ in solution, requiring optimization of the chelate-to-Gd3+ ratio.20 The applicability of 4MMDPA for DEER on membrane systems has not been tested. An ideal Gd3+ spin label having a small volume, high Gd3+ affinity, and a rigid tether is currently unavailable though is highly desirable. Such a label is expected to better reflect the peptide behavior. A recently reported tag based on 4vinyl(pyridine-2,6-diyl)bis-methylenenitrilo tetrakis (acetic acid) (4VPyMTA) designed for NMR58 and later applied for Gd3+−Gd3+ distance measurements in a polyproline helical peptide showed a distance distribution with a width of ∼1.5 nm,19 similar to that observed and calculated for Gd3+-ADO3A in a transmembrane helix.21

whereas for C15C27 the shift to lower distances is milder, and for NC27 it does not appear. For all mutants, the distributions remain broad in LUVs as in solution, indicating that even in this state the labels retain their conformational freedom. This can be explained by both the parallel orientation of the peptide in the membrane plane and the Gd3+ label hydrophilic bias, allowing the label to protrude into the less ordered regions of the LUVs and the solvent interface. There are two contributions to the width of the distance distribution: one from the flexibility of the tether, where rotations about three bonds (up to the S−S bond with the cysteine) are possible, and the other from the flexibility of the peptide itself. The full width at half-height of the Gd3+-ADO3A distance distribution for a transmembrane helical peptide was found to be ∼1.5 nm, and a similar width was obtained from modeling a rigid α-helix.21 All distributions obtained here are much broader, meaning that their width also reflects the peptide’s conformational freedom. These results indicate that, although the conformational space of the peptide changes upon binding to the vesicle, it still remains broad. In addition, we cannot exclude the possibility that not all peptides are bound to the membrane; therefore, there may still be some contributions from peptides in solution, which consequently contribute to the width and the long distances. However, considering the low peptide/lipid ratio (1/ 2000), we regard this option to be unlikely. This is supported by the fact that the nitroxide data, obtained from samples with a higher, 1/200 ratio,31 display narrower distributions and distance shifts, as previously described. A comparison between the Gd3+-ADO3A results and those obtained using nitroxides reveals pronounced differences (Figure 8). In general, the distances are longer and the distance distribution is wider for Gd3+-ADO3A. This difference is especially pronounced for the mutants labeled in the Cterminus. The longer tether of the Gd3+-ADO3A (Figure 1), leading to a larger rotamer cloud,21 may be one of the reasons. However, since the full width at half-height of a nitroxide distance distribution for a rigid helical peptide was calculated to be ∼1.3 nm,21 very close to the 1.5 nm calculated for Gd3+ADO3A, this difference is unlikely to have a major contribution. With nitroxide, the C15C27 mutant displayed a relatively narrow distance distribution already in solution, which was attributed to the existence of some helical structure in solution. However, this was not observed for Gd3+-ADO3A C15C27. One reason could be that it is masked by the larger rotameric freedom, but it may suggest that the behavior of a terminally labeled peptide is sensitive to the label type, even in an unfolded peptide in solution. In the vesicle-bound state, the nitroxide displays shorter distances compared with it in solution, and also narrower distributions, indicating increased structure, in all mutants except C15C27, which revealed a structure already in solution. Distance shortening was also observed for Gd3+-ADO3A, except for the NC27 mutant, but no difference was observed in the width. Again, these differences may be attributed, in addition to the longer tether, to the hydrophilic, bulky ADO3A that can interrupt the peptide folding upon binding to the membrane, and it protrudes more into the solvent, where it has more conformational freedom, in contrast to the nitroxide. This is in agreement with the ESEEM/ENDOR results, showing that the nitroxide nests deeper in the membrane than does ADO3A. Monte Carlo simulations of the distance distributions, carried out considering a leucine instead of the spin label, showed a considerable narrowing of the distribution upon transition from



CONCLUSIONS

This work shows that Gd3+ spin labels can be used at the Wband for studying peptide−membrane interactions. In addition to distance measurements, which probe the peptide conformational structure and heterogeneity, these labels allow one to probe solvent accessibility, although with a lower resolution than that of nitroxides. To reduce the spin label contribution to the distance distribution in such studies, new labels with a shorter, more rigid tether, a small volume, and a large Gd3+ binding constant should be developed. This work also emphasizes the effect of the label characteristics on the results and the need for molecular dynamics simulations for modeling the spin label rotameric freedom and specific interactions with its host. The combination of different labels allows one to differentiate the label effect from the molecule intrinsic characteristics, and suggests that terminal labeling may be less advisible. 13739

DOI: 10.1021/acs.jpcb.5b03523 J. Phys. Chem. B 2015, 119, 13732−13741

Article

The Journal of Physical Chemistry B



in Biomacromolecules Labeled with Gd (III) Markers. J. Magn. Reson. 2011, 210, 59−68. (12) Potapov, A.; Yagi, H.; Huber, T.; Jergic, S.; Dixon, N. E.; Otting, G.; Goldfarb, D. Nanometer-Scale Distance Measurements in Proteins Using Gd3+ Spin Labeling. J. Am. Chem. Soc. 2010, 132, 9040−9048. (13) Goldfarb, D. Metal-Based Spin Labeling for Distance Determination. In Structural Information from Spin-Labels and Intrinsic Paramagnetic Centres in the Biosciences; Timmel, C. R.; Harmer, J. R., Eds.; Structure and Bonding; Springer Berlin Heidelberg: Berlin, 2013; Vol. 152, pp 163−204. (14) Lueders, P.; Jäger, H.; Hemminga, M. A.; Jeschke, G.; Yulikov, M. Distance Measurements on Orthogonally Spin-Labeled Membrane Spanning WALP23 Polypeptides. J. Phys. Chem. B 2013, 117, 2061− 2068. (15) Goldfarb, D. Gd 3+ Spin Labeling for Distance Measurements by Pulse EPR Spectroscopy. Phys. Chem. Chem. Phys. 2014, 16, 9685− 9699. (16) Polyhach, Y.; Godt, A.; Bauer, C.; Jeschke, G. Spin Pair Geometry Revealed by High-Field DEER in the Presence of Conformational Distributions. J. Magn. Reson. 2007, 185, 118−129. (17) Tkach, I.; Pornsuwan, S.; Höbartner, C.; Wachowius, F.; Sigurdsson, S. T.; Baranova, T. Y.; Diederichsen, U.; Sicoli, G.; Bennati, M. Orientation Selection in Distance Measurements between Nitroxide Spin Labels at 94 GHz EPR with Variable Dual Frequency Irradiation. Phys. Chem. Chem. Phys. 2013, 15, 3433−3437. (18) Martorana, A.; Bellapadrona, G.; Feintuch, A.; Di Gregorio, E.; Aime, S.; Goldfarb, D. Probing Protein Conformation in Cells by EPR Distance Measurements Using Gd3+ Spin Labeling. J. Am. Chem. Soc. 2014, 136, 13458−13465. (19) Qi, M.; Groß, A.; Jeschke, G.; Godt, A.; Drescher, M. Gd (III)PyMTA Label Is Suitable for in-Cell EPR. J. Am. Chem. Soc. 2014, 136, 15366−15378. (20) Gordon-Grossman, M.; Kaminker, I.; Gofman, Y.; Shai, Y.; Goldfarb, D. W-Band Pulse EPR Distance Measurements in Peptides Using Gd3+-Dipicolinic Acid Derivatives as Spin Labels. Phys. Chem. Chem. Phys. 2011, 13, 10771−10780. (21) Matalon, E.; Huber, T.; Hagelueken, G.; Graham, B.; Frydman, V.; Feintuch, A.; Otting, G.; Goldfarb, D. Gadolinium(III) Spin Labels for High-Sensitivity Distance Measurements in Transmembrane Helices. Angew. Chem. 2013, 125, 12047−12050. (22) Yagi, H.; Banerjee, D.; Graham, B.; Huber, T.; Goldfarb, D.; Otting, G. Gadolinium Tagging for High-Precision Measurements of 6 Nm Distances in Protein Assemblies by EPR. J. Am. Chem. Soc. 2011, 133, 10418−10421. (23) Edwards, D. T.; Huber, T.; Hussain, S.; Stone, K. M.; Kinnebrew, M.; Kaminker, I.; Matalon, E.; Sherwin, M. S.; Goldfarb, D.; Han, S. Determining the Oligomeric Structure of Proteorhodopsin by Gd 3+-Based Pulsed Dipolar Spectroscopy of Multiple Distances. Structure 2014, 22, 1677−1686. (24) Yulikov, M.; Lueders, P.; Warsi, M. F.; Chechik, V.; Jeschke, G. Distance Measurements in Au Nanoparticles Functionalized with Nitroxide Radicals and Gd3+-DTPA Chelate Complexes. Phys. Chem. Chem. Phys. 2012, 14, 10732−10746. (25) Likhtenshtein, G. I. Depth of Immersion of Paramagnetic Centers in Biological Systems. In Distance measurements in biological systems by EPR; Berliner, L. J.; Eaton, G. R.; Eaton, S. S., Eds.; Biological Magnetic Resonance; Springer US: New York, 2002; Vol. 19, pp 309−345. (26) Georgieva, E. R.; Ramlall, T. F.; Borbat, P. P.; Freed, J. H.; Eliezer, D. Membrane-Bound A-Synuclein Forms an Extended Helix: Long-Distance Pulsed ESR Measurements Using Vesicles, Bicelles, and Rodlike Micelles. J. Am. Chem. Soc. 2008, 130, 12856−12857. (27) Zou, P.; Mchaourab, H. S. Increased Sensitivity and Extended Range of Distance Measurements in Spin-Labeled Membrane Proteins: Q-Band Double Electron-Electron Resonance and Nanoscale Bilayers. Biophys. J. 2010, 98, L18−L20. (28) Dastvan, R.; Bode, B. E.; Karuppiah, M. P. R.; Marko, A.; Lyubenova, S.; Schwalbe, H.; Prisner, T. F. Optimization of Transversal Relaxation of Nitroxides for Pulsed Electron−Electron

ASSOCIATED CONTENT

S Supporting Information *

Optimization of τ for Mims ENDOR, raw DEER data, and DEER of singly labeled mutants. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcb.5b03523.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions

N.M. labeled the peptides, prepared the samples, carried out all spectroscopic measurements and their analysis, and wrote the first draft of the manuscript. V.F. synthesized the MTS-ADO3A reagent and commented on the manuscript. D.G. directed the work through all its stages and was highly involved in writing the manuscript. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This research was supported by the Israel Science Foundation (ISF). D.G. holds the Erich Klieger Professorial Chair in Chemical Physics.



REFERENCES

(1) Stone, T.; Buckman, T.; Nordio, P.; McConnell, H. Spin-Labeled Biomolecules. Proc. Natl. Acad. Sci. U. S. A. 1965, 54, 1010. (2) Eaton, S. S.; Eaton, G. R. Distance Measurements by CW and Pulsed EPR. In Distance Measurements in Biological Systems by EPR; Springer: New York, 2002; pp 1−27. (3) Klare, J. P. Site-Directed Spin Labeling EPR Spectroscopy in Protein Research. Biol. Chem. 2013, 394, 1281−1300. (4) Borbat, P. P.; Freed, J. H. Pulse Dipolar Electron Spin Resonance: Distance Measurements. In Structural Information from Spin-Labels and Intrinsic Paramagnetic Centres in the Biosciences; Timmel, C. R.; Harmer, J. R., Eds.; Structure and Bonding; Springer Berlin Heidelberg: Berlin, 2013; Vol. 152, pp 1−82. (5) Hubbell, W. L.; Mchaourab, H. S.; Altenbach, C.; Lietzow, M. A. Watching Proteins Move Using Site-Directed Spin Labeling. Structure 1996, 4, 779−783. (6) Beier, C.; Steinhoff, H.-J. A Structure-Based Simulation Approach for Electron Paramagnetic Resonance Spectra Using Molecular and Stochastic Dynamics Simulations. Biophys. J. 2006, 91, 2647−2664. (7) Altenbach, C.; Greenhalgh, D. A.; Khorana, H. G.; Hubbell, W. L. A Collision Gradient Method to Determine the Immersion Depth of Nitroxides in Lipid Bilayers: Application to Spin-Labeled Mutants of Bacteriorhodopsin. Proc. Natl. Acad. Sci. U. S. A. 1994, 91, 1667−1671. (8) Frazier, A. A.; Wisner, M. A.; Malmberg, N. J.; Victor, K. G.; Fanucci, G. E.; Nalefski, E. A.; Falke, J. J.; Cafiso, D. S. Membrane Orientation and Position of the C2 Domain from cPLA2 by SiteDirected Spin Labeling. Biochemistry (Moscow) 2002, 41, 6282−6292. (9) Salnikov, E. S.; Erilov, D. A.; Milov, A. D.; Tsvetkov, Y. D.; Peggion, C.; Formaggio, F.; Toniolo, C.; Raap, J.; Dzuba, S. A. Location and Aggregation of the Spin-Labeled Peptide Trichogin GA IV in a Phospholipid Membrane as Revealed by Pulsed EPR. Biophys. J. 2006, 91, 1532−1540. (10) Bordignon, E.; Brutlach, H.; Urban, L.; Hideg, K.; Savitsky, A.; Schnegg, A.; Gast, P.; Engelhard, M.; Groenen, E. J. J.; Möbius, Klaus; et al. Heterogeneity in the Nitroxide Micro-Environment: Polarity and Proticity Effects in Spin-Labeled Proteins Studied by Multi-Frequency EPR. Appl. Magn. Reson. 2010, 37, 391−403. (11) Song, Y.; Meade, T.; Astashkin, A.; Klein, E.; Enemark, J.; Raitsimring, A. Pulsed Dipolar Spectroscopy Distance Measurements 13740

DOI: 10.1021/acs.jpcb.5b03523 J. Phys. Chem. B 2015, 119, 13732−13741

Article

The Journal of Physical Chemistry B Double Resonance Spectroscopy in Phospholipid Membranes. J. Phys. Chem. B 2010, 114, 13507−13516. (29) Pannier, M.; Veit, S.; Godt, A.; Jeschke, G.; Spiess, H. DeadTime Free Measurement of Dipole−Dipole Interactions between Electron Spins. J. Magn. Reson. 2000, 142, 331−340. (30) Carmieli, R.; Papo, N.; Zimmermann, H.; Potapov, A.; Shai, Y.; Goldfarb, D. Utilizing ESEEM Spectroscopy to Locate the Position of Specific Regions of Membrane-Active Peptides within Model Membranes. Biophys. J. 2006, 90, 492−505. (31) Gordon-Grossman, M.; Gofman, Y.; Zimmermann, H.; Frydman, V.; Shai, Y.; Ben-Tal, N.; Goldfarb, D. A Combined Pulse EPR and Monte Carlo Simulation Study Provides Molecular Insight on Peptide−Membrane Interactions. J. Phys. Chem. B 2009, 113, 12687−12695. (32) Matalon, E.; Kaminker, I.; Zimmermann, H.; Eisenstein, M.; Shai, Y.; Goldfarb, D. The Topology of the Trans-Membrane Peptide WALP23 in Model Membranes under Negative Mismatch Conditions. J. Phys. Chem. B 2013, 117, 2280−2293. (33) Matalon, E.; Faingold, O.; Eisenstein, M.; Shai, Y.; Goldfarb, D. The Topology, in Model Membranes, of the Core Peptide Derived from the T-Cell Receptor Transmembrane Domain. ChemBioChem 2013, 14, 1867−1875. (34) Altenbach, C.; Hubbell, W. L. The Aggregation State of Spinlabeled Melittin in Solution and Bound to Phospholipid Membranes: Evidence That Membrane-bound Melittin Is Monomeric. Proteins: Struct., Funct., Bioinf. 1988, 3, 230−242. (35) Altenbach, C.; Froncisz, W.; Hyde, J. S.; Hubbell, W. L. Conformation of Spin-Labeled Melittin at Membrane Surfaces Investigated by Pulse Saturation Recovery and Continuous Wave Power Saturation Electron Paramagnetic Resonance. Biophys. J. 1989, 56, 1183. (36) Gordon-Grossman, M.; Zimmermann, H.; Wolf, S. G.; Shai, Y.; Goldfarb, D. Investigation of Model Membrane Disruption Mechanism by Melittin Using Pulse Electron Paramagnetic Resonance Spectroscopy and Cryogenic Transmission Electron Microscopy. J. Phys. Chem. B 2011, 116, 179−188. (37) Habermann, E. Bee and Wasp Venoms. Science 1972, 177, 314− 322. (38) Raghuraman, H.; Chattopadhyay, A. Melittin: A MembraneActive Peptide with Diverse Functions. Biosci. Rep. 2007, 27, 189−223. (39) Zech, S. G.; Sun, W.; Jacques, V.; Caravan, P.; Astashkin, A. V.; Raitsimring, A. M. Probing the Water Coordination of ProteinTargeted MRI Contrast Agents by Pulsed ENDOR Spectroscopy. ChemPhysChem 2005, 6, 2570−2577. (40) Raitsimring, A. M.; Astashkin, A. V.; Baute, D.; Goldfarb, D.; Poluektov, O. G.; Lowe, M. P.; Zech, S. G.; Caravan, P. Determination of the Hydration Number of Gadolinium (III) Complexes by HighField Pulsed 17O ENDOR Spectroscopy. ChemPhysChem 2006, 7, 1590−1597. (41) Potapov, A.; Goldfarb, D. Quantitative Characterization of the Mn2+ Complexes of ADP and ATPγS by W-Band ENDOR. Appl. Magn. Reson. 2006, 30, 461−472. (42) Potapov, A.; Goldfarb, D. The Mn2+− Bicarbonate Complex in a Frozen Solution Revisited by Pulse W-Band ENDOR. Inorg. Chem. 2008, 47, 10491−10498. (43) Thonon, D.; Jacques, V.; Desreux, J. F. A Gadolinium Triacetic Monoamide DOTA Derivative with a Methanethiosulfonate Anchor Group. Relaxivity Properties and Conjugation with Albumin and Thiolated Particles. Contrast Media Mol. Imaging 2007, 2, 24−34. (44) Olson, F.; Hunt, C. A.; Szoka, F. C.; Vail, W. J.; Papahadjopoulos, D. Preparation of Liposomes of Defined Size Distribution by Extrusion through Polycarbonate Membranes. Biochim. Biophys. Acta 1979, 57, 9−23. (45) Hope, M. J.; Bally, M. B.; Webb, G.; Cullis, P. R. Production of Large Unilamellar Vesicles by a Rapid Extrusion Procedure. Characterization of Size Distribution, Trapped Volume and Ability to Maintain a Membrane Potential. Biochim. Biophys. Acta 1985, 812, 55−65.

(46) Goldfarb, D.; Lipkin, Y.; Potapov, A.; Gorodetsky, Y.; Epel, B.; Raitsimring, A. M.; Radoul, M.; Kaminker, I. HYSCORE and DEER with an Upgraded 95 GHz Pulse EPR Spectrometer. J. Magn. Reson. 2008, 194, 8−15. (47) Van Doorslaer, S.; Sierra, G. A.; Schweiger, A. Dead TimeDependent Line Distortions in Absolute-Value Electron Spin Echo Envelope Modulation Spectra. J. Magn. Reson. 1999, 136, 152−158. (48) Dikanov, S. A.; Tsvetkov, Y. Electron Spin Echo Envelope Modulation ESEEM Spectroscopy; CRC Press: Boca Raton, Florida, 1992. (49) Epel, B.; Arieli, D.; Baute, D.; Goldfarb, D. Improving W-Band Pulsed ENDOR Sensitivityrandom Acquisition and Pulsed Special TRIPLE. J. Magn. Reson. 2003, 164, 78−83. (50) Astashkin, A. V.; Kawamori, A. Matrix Line in Pulsed ElectronNuclear Double Resonance Spectra. J. Magn. Reson. 1998, 135, 406− 417. (51) Jeschke, G.; Chechik, V.; Ionita, P.; Godt, A.; Zimmermann, H.; Banham, J.; Timmel, C. R.; Hilger, D.; Jung, H. DeerAnalysis2006a Comprehensive Software Package for Analyzing Pulsed ELDOR Data. Appl. Magn. Reson. 2006, 30, 473−498. (52) Raitsimring, A. M.; Astashkin, A. V.; Caravan, P. HighFrequency EPR and ENDOR Characterization of MRI Contrast Agents. In High Resolution EPR; Hanson, G.; Berliner, L. J., Eds.; Springer US: New York, 2009; Vol. 28, pp 581−621. (53) Milov, A.; Maryasov, A.; Tsvetkov, Y. D. Pulsed Electron Double Resonance (PELDOR) and Its Applications in Free-Radicals Research. Appl. Magn. Reson. 1998, 15, 107−143. (54) Raitsimring, A.; Dalaloyan, A.; Collauto, A.; Feintuch, A.; Meade, T.; Goldfarb, D. Zero Field Splitting Fluctuations Induced Phase Relaxation of Gd3+ in Frozen Solutions at Cryogenic Temperatures. J. Magn. Reson. 2014, 248, 71−80. (55) Kevan, L. Modulation of Electron Spin-Echo Decay in Solids. In Time Domain Electron Spin Resonance; Kevan, L., Schwartz, R. N., Eds.; Wiley: New York, 1979; pp 279−341. (56) Lauterwein, J.; Bö sch, C.; Brown, L. R.; Wüthrich, K. Physicochemical Studies of the Protein-Lipid Interactions in Melittin-Containing Micelles. Biochim. Biophys. Acta, Biomembr. 1979, 556, 244−264. (57) Lauterwein, J.; Brown, L. R.; Wüthrich, K. High-Resolution 1HNMR Studies of Monomeric Melittin in Aqueous Solution. Biochim. Biophys. Acta, Protein Struct. 1980, 622, 219−230. (58) Yang, Y.; Li, Q.; Cao, C.; Huang, F.; Su, X. Site-Specific Labeling of Proteins with a Chemically Stable, High-Affinity Tag for Protein Study. Chem.Eur. J. 2013, 19, 1097−1103.

13741

DOI: 10.1021/acs.jpcb.5b03523 J. Phys. Chem. B 2015, 119, 13732−13741