Increasing the Modulation Depth of Gd^III-Based Pulsed Dipolar EPR Spectroscopy (PDS) with Porphyrin–Gd^III Laser-Induced Magnetic Dipole Spectroscopy

Abstract

Distance determination with pulsed EPR has become an important technique for the structural investigation of biomacromolecules, with double electron–electron resonance spectroscopy (DEER) as the most important method. Gd^III-based spin labels are one of the most frequently used spin labels for DEER owing to their stability against reduction, high magnetic moment, and absence of orientation selection. A disadvantage of Gd^III–Gd^III DEER is the low modulation depth due to the broad EPR spectrum of Gd^III. Here, we introduce laser-induced magnetic dipole spectroscopy (LaserIMD) with a spin pair consisting of Gd^III(PymiMTA) and a photoexcited porphyrin as an alternative technique. We show that the excited state of the porphyrin is not disturbed by the presence of the Gd^III complex and that herewith modulation depths of almost 40% are possible. This is significantly higher than the value of 7.2% that was achieved with Gd^III–Gd^III DEER.

Pulsed dipolar EPR spectroscopy (PDS) has become an important tool for the structural analysis of biomacromolecules like proteins, DNA, and RNA.¹⁻⁸ PDS measures the magnetic dipolar coupling between two paramagnetic centers in frozen solution, from which distance distributions in the nanometer range can be calculated.⁹ Among the many techniques that have been developed,¹⁰⁻¹³ double electron–electron resonance spectroscopy (DEER) is the most common one.^9,14

Because biomolecules rarely contain unpaired electrons, PDS requires usually that paramagnetic labels, called spin labels, are attached at defined sites. Although nitroxides and carbon-based trityl labels have received the most attention in the last years,¹⁵⁻¹⁸ Gd^III complexes with a chelating ligand (electron spin ) are becoming increasingly popular.^3,19,20 Their advantages includes reduction stability,^21,22 the absence of orientation selection and high sensitivities at Q- and W-bands.²³⁻²⁷ However, their broad EPR spectrum spans several GHz and exceeds the excitation bandwidth of most resonators and microwave pulses.²⁸ Therefore, only a small fraction of spins are addressed by the microwave pump pulse used in Gd^III–Gd^III DEER, which results in a much lower modulation depth than those achieved with other PDS techniques or labels.^23,26 Another drawback of Gd^III–Gd^III DEER are broadening artifacts that appear at distances below ≈3 nm if the offset between the observer and pump microwave frequency is in the range of only a few 100 MHz.²⁹⁻³² One technique with which both of these disadvantages for Gd^III-based PDS can be overcome is relaxation induced dipolar modulation enhancement (RIDME).^33,34 In RIDME, the spin flips of the pump spins take place as stochastic events due to relaxation and are thus not limited by the excitation bandwidth of the microwave pulses.^12,35 Hence, modulation depths higher than 50% are possible with Gd^III–Gd^III RIDME.³³ However, because spin flips with Δ|m_S| > 1 are also possible, Gd^III–Gd^III RIDME suffers from the excitation of overtones.³⁴ This is a disadvantage, because their relative intensities must be known and included as overtone coefficients in the data analysis procedure, because they would otherwise result in fake distances.³⁶

A different approach for PDS is taken in laser-induced magnetic dipole spectroscopy (LaserIMD).^37,38 Here, the dipolar coupling between a permanent spin label and a label that is transiently converted into a paramagnetic state through photoexcitation is measured. The Δ|m_S| = 1 transition required for PDS is achieved through intersystem crossing from its photoexcited diamagnetic singlet state (S = 0) to the paramagnetic triplet state (S = 1), thereby replacing the microwave pump pulse used in DEER (Figure 1a). The bandwidth of the photoexcitation is not limited by the EPR spectrum or the microwave resonator and is virtually infinite.³⁷ Thus, even though such a combination of two distinct labels requires a more difficult labeling scheme for the attachment to a protein,^39,40 LaserIMD on Gd^III-based spin labels is a promising technique, because of its potential for achieving higher modulation depths than with Gd^III–Gd^III DEER. Also, the aforementioned broadening artifacts are avoided in LaserIMD, because only one microwave frequency is used and with a spectral width of 1.5 GHz for the triplet state of the porphyrin, the transient spin label used in this study,^37,41 the frequency offset between pump and observer spins is significantly larger. High modulation depths of over 40% with Gd^III spin labels can also be achieved by performing DEER on a nitroxide–Gd^III spin pair,⁴² However, combining Gd^III with a transient spin label instead of a nitroxide has the advantage that photoexcitable groups are endogenous in many proteins like heme-proteins⁴³ or light-harvesting proteins.⁴⁴ In such a case, only a single Gd^III label needs to be introduced, which reduces potential disturbances on the protein by the labels.³⁷ Furthermore, as lanthanide tags have already been shown to be applicable as a FRET donor,⁴⁵ the combination of a photoexcitable label and a Gd^III label also opens the possibility to combine luminescence-based spectroscopic techniques like Förster resonance energy transfer (FRET) with PDS.

Figure 1.

(a) Pulse sequence of LaserIMD. (b) Structural formula of the transient spin label TNPP, the compounds used to install the permanent Gd^III-based spin label with PyMTA or PymiMTA as the ligand, and the model peptides TNPP-pP-Gd^III(PymiMTA) and Gd^III(PymiMTA)-pP-Gd^III(PymiMTA). For the peptide chains, the following letter code is used: A, alanine; G, glycine; P, proline.

In initial studies of light-induced PDS, porphyrins were paired with nitroxide labels.^37,41 Other label combinations that have been reported since are fullerenes with trityl radicals or nitroxides^46,47 the combination of the fluorescent dyes Rose Bengal, Eosin Y, or Atto Thio12 with a nitroxide⁴⁰ and two porphyrins, both of which are photoexcited.⁴⁸ We opted for porphyrin as the transient spin label, because of its photostability and selected the porphyrin TNPP (Figure 1) because of its water solubility.⁴⁹ Due to the bulkiness of TNPP, it should be considered that upon attachment to a protein it might interfere with the protein structure. For the persistent spin label, we chose Gd^III(PymiMTA) (Figure 1)^50,51 which is structurally similar to the well-known Gd^III(PyMTA).^21,29,52 The substitution of the pyridine ring for a pyrimidine ring has little effect on the EPR-spectroscopic properties like line width and relaxation times, but its bioconjugation rate is much higher.⁵¹

The combination of a Gd^III-based spin label with porphyrin poses potential problems like the incorporation^53,54 of Gd^III into the porphyrin or quenching⁵⁵⁻⁵⁷ of the electronically excited porphyrin by the Gd^III complex. Here, we set out to explore this spin label combination, establish LaserIMD with this combination as a light-induced PDS technique and compare its performance with that of Gd^III–Gd^III DEER.

A model system was designed based on a polyproline helix (pP) as spacer^58,59 between the porphyrin TNPP and Gd^III(PymiMTA) (Figure 1b). We attempted to spin label the TNPP-labeled peptide TNPP-GP₉CP₃–NH₂ (G, glycine; P, proline; C, cysteine) with the complex Gd^III(PyMTA) by applying a two-step protocol reported for a structurally similar peptide,²¹ with first ligand attachment and second complex formation. However, the attachment of 4-vinyl-PyMTA to the cysteine unit of the peptide TNPP-GP₉CP₃–NH₂ failed. Already, in the mentioned previous work, the reaction proceeded rather slowly, even at 40 °C.²¹ Turning to Na[4-vinyl-{Gd^III(PymiMTA)}] (Figure 1b) as the reaction partner, quantitative spin labeling in a single step was accomplished within 2 h at room temperature. We ascribe the strong increase in reactivity to the increase in electrophility of the vinyl unit after exchanging the pyridine ring for the more electron withdrawing pyrimidine ring. Surprisingly, mass spectrometric analysis of the material, that had been obtained through HPLC, showed a significant amount of sulfoxide linkage (Figure 1, X = SO; Supporting Information, Section S1) alongside the expected sulfide linkage (Figure 1b, X = S). We estimated that the difference between the distance distribution of the sulfoxide and sulfide linkage is negligible and proceeded with the sulfide/sulfoxide mixture. For comparison with Gd^III–Gd^III DEER, a polyproline helix with two Gd^III(PymiMTA) labels (Gd^III(PymiMTA)-pP-Gd^III(PymiMTA)) was prepared. Circular dichroism measurements showed that all three peptides adopt a pPII structure in water (Supporting Information, section S2).

For TPP-Gd^III LaserIMD it is important that the Gd^III ion is not exchanged between the PymiMTA ligand and TNPP.^53,54 An indication of that a metal ion has been incorporated into porphyrin is the change in its UV/vis spectrum from four distinct Q-bands for the metal ion free porphyrin to two Q-bands for the Gd^III ion loaded porphyrin.⁵⁴ In our case the UV/vis spectrum of TNPP-pP-Gd^III(PymiMTA) (Figure 2a, experimental details in Supporting Information, section S3) shows four Q-bands at the same wavelengths as the ion free TNPP, which we take as a strong sign that no Gd^III ion exchange took place. This is supported by time-resolved X-band EPR spectra (trEPR) of photoexcited TNPP. Incorporation of a metal ion into porphyrin can significantly change the zero-field splitting and zero-field population of the excited triplet, and hence the trEPR spectrum.^60,61 As can be seen in parts c and d of Figure 2, TNPP-pP-Gd^III(PymiMTA) and ion free TNPP show virtually the same trEPR spectrum with almost identical zero-field splitting values and zero-field populations.

Figure 2.

(a) UV/vis spectra of TNPP and TNPP-pP-Gd^III(PymiMTA) in water at room temperature. The focused area shows the four Q bands of the TNPP spectrum. (b) Luminescence lifetime measurements of TNPP and TNPP-pP-Gd^III(PymiMTA) in water at room temperature with the time points at which the signal intensities have dropped to 1/e. (c) trEPR spectrum of TNPP in D₂O/glycerol-d₈ (40/60 vol %) at 10 K with the corresponding fit (D = 1245 MHz, E = −187 MHz, P_x = 0.26, P_y = 0.59, P_z = 0.15, H_strain = 109 MHz). (d) trEPR spectrum of TNPP-pP-Gd^III(PymiMTA) in D₂O/glycerol-d₈ (40/60 vol %) at 10 K with the corresponding fit (D = 1227 MHz, E = −172 MHz, P_x = 0.26, P_y = 0.60, P_z = 0.14, H_strain = 139 MHz).

Lanthanide cations and their complexes are known to act as photoquenchers of electronically excited states.⁵⁵⁻⁵⁷ Therefore, the close proximity of Gd^III(PymiMTA) to TNPP in TNPP-pP-Gd^III(PymiMTA) may quench the photoexcited triplet state, which would interfere with the light-induced PDS measurement and might render it impossible. To check whether quenching takes place, the lifetimes of the excited states of TNPP and TNPP-pP-Gd^III(PymiMTA) were determined through time-resolved luminescence measurements at room temperature (experimental details in Supporting Information, section S4). The lifetimes, defined as the time up to which the signal has decayed to 1/e of its maximum value, increases from 5.1 μs for TNPP to 5.9 μs for TNPP-pP-Gd^III(PymiMTA) as can be seen in Figure 2b. Additionally, the average lifetimes of the photoexcited triplet state were determined with pulsed EPR spectroscopy at 10 K (details in Supporting Information, sections S5 and S6).⁶² Following the same trend as the luminescence lifetimes, the average triplet lifetimes show an increase from 47.7 ms for TNPP to 57.8 ms for TNPP-pP-Gd^III(PymiMTA). Based on these results, we conclude that the photoexcited state of TNPP is not quenched by the close-by Gd^III(PymiMTA).

For a fair comparison of TNPP-Gd^III LaserIMD and Gd^III–Gd^III DEER, we first optimized the signal-to-noise ratio (SNR) of the latter by using broadband shaped pulses as pump pulses and fine-tuned the pulse shapes, frequency widths, pulse lengths and observer and pump pulse frequencies.²⁶ The SNR is calculated as the ratio of the modulation depth and the experimental noise, normalized to the square-root of the measurement time. With the applied optimization heuristic as described in detail in Supporting Information, section S7, the best SNR was obtained when observing at the maximum of the Gd^III EPR spectrum at 34 GHz with a Gaussian pulse and pumping with a 200 ns WURST pulse with n = 12 and a sweep width of 300 MHz. Gd^III–Gd^III DEER with these settings on Gd^III(PymiMTA)-pP-Gd^III(PymiMTA), resulted in a modulation depth of 7.2% and an SNR of 530 h^–1/2 (Figure 3a and Table 1). Although this is a significantly larger modulation depth than the 2.0% that were reached with a rectangular pump pulse, the modulation depth stays behind what can be achieved with LaserIMD.^41,63

Figure 3.

Comparison of Gd^III–Gd^III DEER data obtained with rectangular and broadband shaped pump pulses from Gd^III(PymiMTA)-pP-Gd^III(PymiMTA) dissolved in D₂O/glycerol-d₈ (40/60 vol %), at 10 K. (a) Experimentally recorded data and fits. (b) The distance distributions were obtained with Tikhonov regularization.⁶⁴

Table 1. Modulation Depths and SNR for Gd^III–Gd^III DEER, TNPP-Gd^III (re)LaserIMD and TNPP-Gd^III reLaserIMD from Figures 3 and 4.

	Mod depth [%]	SNR [h–1/2]
GdIII–GdIII DEER (WURST, pump pulse)	7.2 (6.9, 7.6)	530 (500, 610)
TNPP-GdIII LaserIMD	39.4 (38.9, 40.1)	190 (180, 210)
TNPP-GdIII reLaserIMD	35.8 (35.3, 36.5)	150 (140, 170)

To prove this, TNPP-Gd^III LaserIMD data were recorded at 10 K for TNPP-pP-Gd^III(PymiMTA) (results in Figure 4; experimental details in Supporting Information, section S5). The SRT was optimized to 100 ms and the laser energy per pulse to 1.4 mJ (Supporting Information, section S8). If available, temperatures below 10 K can also be used as they are known to give a higher spin-polarization for Gd^III in the Q-band.²³ To analyze the dipolar trace, its zero-time needs to be determined precisely. As this cannot always be reliably done for LaserIMD, a LaserIMD experiment with an additional refocusing pulse, termed reLaserIMD, has been suggested as it allows an accurate zero-time determination.⁴¹ However, the introduction of the additional pulse increases the overall trace length which decreases the SNR. To combine the best of both worlds, we tried a different approach where the data were recorded with LaserIMD and then shifted by a zero-time determined with reLaserIMD. As the zero-time does not depend on the dipolar evolution time, the latter can be chosen as short as possible in order to maximize the echo intensity, which allows to record a reLaserIMD trace with a high sensitivity in a short time. Furthermore, for laser systems with a constant delay of the laser flash, the zero-time depends only on the microwave pulse lengths. In such a case, once the zero-time for a given pulse length has been determined, it can be reused for future measurements and it is not necessary to measure reLaserIMD every time. The combination of reLaserIMD and LaserIMD was found to give a reliable zero-time (Supporting Information, section S9) and an SNR increase from 150 h^–1/2 for TNPP-Gd^III reLaserIMD to 190 h^–1/2 for TNPP-Gd^III LaserIMD (Figure 4). The thus optimized TNPP-Gd^III LaserIMD gave a modulation depth of 39.4% (Table 1). Even though this stays behind the modulation depth of 50% that has been reported for Gd^III–Gd^III RIDME,³³ TNPP-Gd^III LaserIMD has the advantage, that the laser excites no transitions with |m_s| > 1 and therefore no overtones coefficients are needed for data analysis.

Figure 4.

TPP-Gd^III LaserIMD and Gd^III–Gd^III DEER measurements of TNPP-pP-Gd^III(PymiMTA) and Gd^III(PymiMTA)-pP-Gd^III(PymiMTA), respectively, both in D₂O/glycerol-d₈ (40/60 vol %) at 10 K. (a) Experimentally recorded data and fits. (b) The distance distributions were obtained with Tikhonov regularization.⁶⁴

The modulation depth of 39% for TNPP-Gd^III LaserIMD might be significantly higher than the 7.2% for Gd^III–Gd^III DEER, but the crucial parameter for the determination of distance distributions is the SNR. Here, TNPP-Gd^III LaserIMD has a lower SNR of 190 h^–1/2 compared to 530 h^–1/2 for Gd^III–Gd^III DEER. Due to the short longitudinal relaxation time of Gd^III(PymiMTA)-pP-Gd^III(PymiMTA) of 35.6 μs (Supporting Information, section S10), it was possible to use a fast shot repetition time (SRT) of 100 μs (Supporting Information, section S7). In contrast, TNPP requires a much longer SRT of 100 ms, because with a triplet relaxation time of 57.8 ms, the signal would saturate otherwise. Therefore, in a given measurement time, Gd^III–Gd^III DEER benefits from much more scans being accumulated than for TNPP-Gd^III LaserIMD, which explains the better SNR of the former. As most transient labels used so far for LaserIMD require an SRT in the millisecond range,^{40,46,48,63,65} the continuation of the development of transient spin labels with faster triplet relaxation times is necessary to open up the full possibilities of LaserIMD with Gd^III-based spin labels.

In conclusion, we have shown that TNPP and Gd^III(PymiMTA) are a suitable label pair for light-induced EPR measurements and PDS in particular. UV/vis measurements, time-resolved EPR spectroscopy as well as lifetime measurements of the excited TNPP gave neither an indication of a Gd^III ion exchange between the labels nor an indication of quenching of the photoexcited triplet state of the TNPP by Gd^III–PymiMTA. LaserIMD is an interesting technique, which can be used not just for structural investigations on biological systems with endogenous photoexcitable moieties; the presence of a photoexcitable label also makes it a suitable technique to combine PDS and luminescence measurements. Furthermore, the increase in modulation depth from 7.2% for Gd^III–Gd^III DEER to 39.7% for TNPP-Gd^III LaserIMD promises a significant increase in SNR, once sorter SRTs become applicable.

The raw data are available at https://doi.org/10.5281/zenodo.7051255.⁶⁶

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpclett.2c02138.

• Synthesis of the model peptides, details on sulfoxide formation and Gd^III–Fe^III ion exchange, details on CD, UV/vis, fluorescence lifetime, and time-resolved and pumped EPR measurements, MNR optimization of DEER and LaserIMD, zero-time determination for LaserIMD, and and longitudinal and transversal relaxation of Gd^III (PDF)

The authors declare no competing financial interest.