A New ¹³C Trityl-based Spin Label Enables the Use of DEER for Distance Measurements

Abstract

Triarylmethyl (TAM)-based labels, while still underutilized, are a powerful class of labels for pulsed-Electron Spin Resonance (ESR) distance measurements. They feature slow relaxation rates for long-lasting signals, high stability for cellular experiments, and narrow spectral features for efficient excitation of the spins. However, the typical narrow line shape limits the available distance measurements to only single-frequency experiments, such as Double Quantum Coherence (DQC) and Relaxation Induced Dipolar Modulation Enhancement (RIDME), which can be complicated to perform or hard to process. Therefore, widespread usage of TAM labels can be enhanced by the use of Double Electron-Electron Resonance (DEER) distance measurements. In this work, we developed a new spin label, ¹³C₁-mOX063-d₂₄, with a ¹³C isotope as the radical center. Due to the resolved hyperfine splitting, the spectrum is sufficiently broadened to permit DEER-based experiments at Q-band spectrometers. Additionally, this new label can be incorporated orthogonally with Cu(II)-based protein label. The orthogonal labeling scheme enables DEER distance measurement at X-band frequencies. Overall, the new trityl label allows for DEER-based distance measurements that complement existing TAM-label DQC and RIDME experiments.

Graphical Abstract

Introduction

Triarylmethyl (TAM)-based spin-labels have emerged as a powerful label for ESR-based distance measurements. These labels exhibit a general tris-(tetrathiaaryl)methyl core, shown in Figure 1, where most of the spin density (>60%) resides on the central carbon (C₁). Based on this motif a wide variety of labels have been synthesized^1–6. TAM-based labels have narrow spectral features because the absence or small hyperfine interactions. The narrow lines can be efficiently excited by pulses⁷, leading to up to thirty times higher signal intensity in pulsed ESR than commonly used and commercially available nitroxides⁸. In addition to the sensitivity, TAM labels have slow relaxation rates^9,10, even at room temperature, which have enabled a pathway towards distance measurements at physiological temperatures^11–15. Finally, these TAM labels are highly resistant to reduction within cellular environments^16–18. These advantages are desirable for ESR-based experiments in the field of structural biology. When a biomolecule is selectively labeled at a given site, ESR can measure site-specific dynamics of the molecule^19,20. Furthermore, when two or more sites are labeled, ESR can measure distance constraints between spin-labels^21–27 on the biomolecule. More importantly, these measurements are not limited by protein size. Therefore, ESR can study large and complicated systems that are not easily studied using X-ray Crystallography or NMR. With the utility of ESR techniques, implementation of different types of spin-labels, such as Gd(III) ^28–30, Mn(II)^31,32, Cu(II)^33–36, and especially TAM^8,14,37 expand the scope of critical biological questions that can be probed.

Figure 1.

A) Representation of the TAM-core. B) TAM-based labels that use Short Linker Maleimide (SLIM)^1,2. C) Hydrophilic TAM-based labels with long linkers^3,4. D) Hydrophobic TAM-based labels^5,6. E) New label derived from mOX063-d₂₄ with ¹³C isotope as the radical center.

One of the limiting aspects of TAM-based labels is the available ESR techniques for TAM distance measurements. Specifically, only single-frequency pulsed-ESR techniques, such as Double Quantum Coherence (DQC)^21,22 or Relaxation-Induced Dipolar Modulation Enhancement (RIDME)^26,27 can be used due to the narrow spectral features of TAM. While DQC provides high-sensitivity TAM-based distance measurements, the method can be hard to implement on commercial instrumentation due to the need for short pulses and the long phase cycles needed to remove unwanted signals. Such large phase cycling procedures can often lead to long acquisition times on commercial instrumentation due to limited on-board phase cycling^38–41. While TAM-based DQC have been successful^1,4,42–46, this technology is currently adopted only by a select few researchers. On the other hand, RIDME can also provide sensitive TAM-based distance measurements. However, RIDME suffers from the complex background signal⁴⁷. Moreover, the distance distribution can be analyzed only after the background is subtracted correctly using various methods⁴⁸. With these considerations, expanding the available pulsed-ESR technique to include a two-frequency technique, Double Electron-Electron Resonance (DEER)^23,24, will further improve the accessibility of TAM-based labels.

To date, DEER on TAM radicals has been performed under three conditions. The first condition is when TAM distances were obtained at high frequencies, i.e at W-band^49,50 or G-band⁴⁵. With higher frequencies, the TAM spectrum is broad enough to allow the observer frequency and pump frequency to be at a reasonable frequency separation. The second condition is at a Q-band frequency where the observer is set at the edge of the TAM spectra. This choice of observer frequency reduced the sensitivity of the measurement by seven-fold when compared to DQC measurements⁴². In the last case, DEER was done on TAM by exciting a set of ¹³C present at natural abundance in the label⁵¹. These results suggest that current TAM-based label spectra are too narrow and will only be broad enough at high frequencies that may be inaccessible. One solution to this problem is to have a TAM label with resolved splitting at commonly available frequencies.

Recently, a TAM radical with a ¹³C isotope as the radical center, shown in Figure 1D, was synthesized and shown to be sensitive to viscosity⁵. The ¹³C isotope leads to a resolved splitting due to a hyperfine interaction with the ¹³C nuclear spin (I=1/2). This spectral feature due to ¹³C can make the TAM radical well suited for DEER distance measurements that require pulses at two separate frequencies. Therefore, distance measurements will benefit from incorporation of ¹³C₁ into previously published TAM-based labels. For example, we previously showed that hydrophilic mOX063-d₂₄⁴, shown in Figure 1C, labels proteins with high efficiency and is cleavage-resistant protein labeling in-cells. Furthermore, the hydrophilic nature of mOX063-d₂₄ provides long relaxation times which allowed for distance measurements to be performed at 150 K and in-cell. In particular, the hydrophilicity reduces lipophilic interactions that enhances the relaxation times of the label³.

In this work, we developed a new TAM label, ¹³C₁-mOX063-d₂₄, as shown in Figure 1E, for DEER distance measurements. First, we show that the ¹³C₁-mOX063-d₂₄ can label the intended labeling site efficiently. Next, we demonstrate that the spectral breadth of ¹³C₁-mOX063-d₂₄ allows for DEER distance measurements at Q-band. Finally, we implement ¹³C₁-mOX063-d₂₄ in an orthogonal labeling scheme with a Cu(II)-based label at X-band for DEER measurements. Overall, we showcase the power of ¹³C₁-mOX063-d₂₄ and expand the ESR techniques available for TAM-based distance measurements.

Methods

Synthesis of [¹³C₁, 99%]-mOX063-d₂₄.

[¹³C₁, 99%]-OX063-d₂₄ trisodium salt synthesized using known protocols⁵² (300 mg, 0.21 mmol, 1 eq.) was dissolved in 10 mL of DMSO under argon. Diethylisopropylamine (DIEA) (108 μL, 0.62 mmol, 3 eq.) was added to the solution, followed by Benzotriazol-1-yl-oxytripyrrolidinophosphonium hexafluorophosphate (PyBOP) (107 mg, 0.21 mmol, 1 eq.). The green solution turned red and was stirred for 30 min. Then, N-(2-aminoethyl)maleimide trifluoroacetate salt (105 mg, 0.42 mmol, 2 eq.) was added, and the reaction was stirred at room temperature for 90 minutes. The reaction mixture was diluted with 500 mL of water containing 0.1% trifluoroacetic acid (TFA). The product was loaded on a C18 column and purified using a gradient of water and acetonitrile (ACN) both containing 0.1% TFA, from 5% ACN to 15% ACN, using a Teledyne CombiFlash Rf+ system. The purified product was freeze-dried, then the resulting brown powder was dissolved in deionized water and titrated to pH=7 with sodium hydroxide. The solution was freeze-dried again, affording 128 mg (40%) of [¹³C₁,99%]-mOX063-d₂₄ as a green powder named ¹³C₁-mOX063-d₂₄ thorough the manuscript. The purity was evaluated to >98% by HPLC/MS (Figure S1).

GB1 labeling protocol.

Expression and purification of E15C/K28C or E15C/K28H/Q32H GB1 mutants were performed as previously described⁵³. The protein was incubated overnight at 4 °C in the presence of tris(2-carboxyethyl)phosphine (TCEP) to reduce any disulfide bonds. TCEP was removed by flowing the solution through GE Healthcare Hitrap desalting columns directly into a solution of ¹³C₁-mOX063-d₂₄. The labeling reaction occurs overnight at 4°C with the ratio of ¹³C₁-mOX063-d₂₄:GB1 being 10:1. After incubation, the solution was dialyzed for 72 hours using a 2-kDa MW cutoff Spectra/Por7 Dialysis Membrane in PBS pH 6.8 to remove unreacted labels. The dialyzed solution was concentrated using Amicon Ultra Centrifugal Filter Units with a 3-kDa cutoff. UV-Vis measurement using Nanodrop2000 Spectrophotometer from Thermo Scientific was used to determine concentration and labeling efficiency, as previously described⁴.

For Cu(II) labeling, a 10 mM stock of Cu(II) chelated by nitrilotriacetic acid (NTA) was prepared as described previously^34,54,55. The ¹³C₁-mOX063-d₂₄ labeled GB1 was added with Cu(II)NTA in a 1:1 ratio, and the mixture was incubated at 4°C for 30 minutes. The samples were prepared in 3-N-morpholinopropanesulfonic acid (MOPS) buffer to allow efficient Cu(II)NTA binding to the dHis motif⁵⁵. After incubation, 40 % deuterated glycerol was added, and the sample was flash-frozen with liquid MAP-Pro Propylene/propane gas. The Cu(II)NTA spin-labeling and freezing process followed a step-by-step protocol published recently⁵⁴.

ESR measurements.

Room-temperature continuous-wave (CW)-ESR were performed on a Bruker ElexSys E680 CW/FT X-band spectrometer with a Bruker ER4122 SHQE-W1 resonator. Capillary tubes were used to prepare the CW samples. For ¹³C₁-mOX063-d₂₄, the experiments used the following parameters: center field of 3520 G, sweep width of 100 G, frequency of ~9.87 GHz, modulation amplitude of 0.5 G, modulation frequency of 100 kHz, data points of 2048, and conversion time of 40.00 ms. For Cu(II)NTA, the experiments used the following parametersL center field of 3200 G, sweep width of 2000 G, frequency of ~9.65 GHz, modulation amplitude of 4 G, modulation frequency of 100 kHz, data points of 1024, and conversion time of 20.48 ms.

For low-temperature experiments, an Oxford CF935 dynamic continuous-flow cryostat attached to Oxford LLT 650 low-loss transfer tube and Oxford ITC503 temperature controller ensured stable cryogenic temperature of the samples.

A three-pulse electron-spin echo envelope modulation (ESEEM) sequence^56,57, (π/2)−t₁−(π/2)−t₂−(π/2), were performed at 18 K at X-band. The period t₁ was set as 140 ns. Additionally, t₂ was incremented in steps of 16 ns for a total of 1024 points from an initial value of 280 ns. The experiment was done at the field with the highest Cu(II)NTA echo intensity. A four-step phase cycling was used to eliminate the undesired echoes⁵⁸. The resulting time domain was transformed into the frequency domain using Fast Fourier Transform. The magnitude of the spectrum from 0 MHz to 10 MHz is shown in Results and Discussion sections.

Q-band DEER was performed using a four-pulse sequence, (π/2)v_A−t₁−(π)v_A−(t₁+τ)−(π)v_B−(t₂−τ)(π)v_A. The pump (v_B) and observer (v_A) frequencies were set respectively at the first and second local maximum of the ¹³C₁-mOX063-d₂₄ spectra. The periods t₁ and t₂ were set as 400 ns and 3200 ns, respectively. The parameter τ was increased by increments of 16 ns for a total of 166 time points. A 16-step phase cycling was implemented to obtain the main DEER echo²⁴. For X-band DEER, all parameters are the same except for v_A and v_B, which were set respectively at the first local maxima of the ¹³C₁-mOX063-d₂₄ spectra and the maximum of the Cu(II)NTA spectra.

All X-band pulsed experiments were performed on Bruker ElexSys E680 X-band FT/CW spectrometer using Bruker EN4118X-MD4 resonator and a 1 kW amplifier, while Q-band experiments used Bruker ElexSys E580 CW/FT X-band spectrometer using a Bruker ER5106-QT2 resonator for Q-band and a 300 W amplifier.

Results and Discussion

In this work, we used the immunoglobulin binding domain of protein G (GB1). Specifically, we engineered and overexpressed a double cysteine mutant E15C/K28C GB1^59,60. The protein was labeled with ¹³C₁-mOX063-d₂₄ at both cysteine residues of GB1, as described in the Methods section. In summary, E15C/K28C-GB1 was incubated with ten times the excess amount of label to cysteine residues at 4 °C for 16 hours. After the reaction, the unreacted label was removed through dialysis. The dialyzed sample was then concentrated using centrifugal filter units.

To assess the labeling efficiency, we quantified the concentrations of protein and spin-label using UV-Vis, as shown in Figure 2A. The UV-Vis spectrum of the sample is a sum of the spectrum of GB1 and ¹³C₁-mOX063-d₂₄. Therefore, we can obtain the concentration of both protein and label by fitting the sample spectra with the spectra of the protein and label individually, as described previously⁴. By using the extinction coefficient of GB1 (ϵ₂₈₀ = 9970 cm⁻¹ M⁻¹) and ¹³C₁-mOX063-d₂₄ (ϵ₄₇₀ = 16219 cm⁻¹M⁻¹, cf. Figure S2), we calculated the final concentrations of GB1 and the label in the sample to be 71 μM and 202 μM, respectively. Therefore, the ratio of cysteine: ¹³C₁-mOX063-d₂₄ in our sample is 1:1.4, which may be due to incomplete removal of our unreacted label after dialysis.

Figure 2.

A) UV-Vis spectrum of ¹³C₁-mOX063-d₂₄-GB1 sample (gray line). The ¹³C₁-mOX063-d₂₄-GB1 spectra can be fitted by the sum of its GB1 (dash-dotted line) and ¹³C₁-mOX063-d₂₄ (dotted line) components. B) CW-ESR spectrum of ¹³C₁-mOX063-d₂₄-GB1 (top). The spectrum of ¹³C₁-mOX063-d₂₄-GB1 can be fit using a two-component simulation using EasySpin⁶¹. Each component has a peak-to-peak linewidth of 1.4 G and 3.6 G for the free and bound components, respectively.

With the unreacted label in our sample, we performed continuous wave (CW)-ESR at room temperature to quantify the amount of bound and free label, shown in Figure 2B. The spectrum, shows a doublet pattern that originate from the hyperfine interaction with the ¹³C₁ nuclear spin (I=1/2)⁵. In addition, a sharp peak at the center of the spectrum ~3520 G arise from the 1% isotopologue of the label having a ¹²C at the central carbon, as the label was labeled 99% ¹³C₁.

Using EasySpin⁶¹, we simulated the room temperature CW-ESR spectrum, using a sum of two components. Both components had the same g_x, g_y, and g_z of 2.0033, 2.0032, and 2.0027, respectively while A_x, A_y, and A_z were 18 MHz, 18 MHz, and 160 MHz, respectively, as reported for ¹³C₁-OX063-d₂₄⁶². The only difference between the two components is the rotational correlation time (τ_r). The first component has a τ_r of 0.55 ns while the second component has a τ_r of 1.80 ns. In the simulations, we assumed isotropic tumbling since the spectral lineshape for this system is largely insensitive to anisotropic tumbling (cf. Figure S3). The first component corresponds to the free label, as its rotational correlation time matches τ_r determined for the free ¹³C₁-mOX063-d₂₄ (cf Figure S4) under similar conditions. Therefore, the sample contains~ 29% of free label and 71% of bound label. Spin-counting of the spectra provides a total of 205 μM of spins, which agrees with the concentration obtained from UV-Vis. More importantly, the fitting indicates that 145 μM of the labels are bound, which matches the concentration of 142 μM cysteines in our sample. This labeling efficiency is consistent with the documented labeling efficiency of mOX063-d₂₄⁴, which is an isotopologue of our new ¹³C₁-mOX063-d₂₄ label. Overall, ¹³C₁-mOX063-d₂₄ can efficiently label GB1.

Despite efficient labeling, the free label can still affect the sensitivity of subsequent distance measurements. Repetition of the sample preparation still contained some between 21% and 34% free label. The incomplete removal of the free label is due to the 2 kDa MW cutoff of the dialysis tubing, whose size was limited by the 6.2 kDa GB1 and the 1.5 kDa ¹³C₁-mOX063-d₂₄. As a result, the diffusion rate of the label is significantly slow during dialysis, leading to incomplete label removal even after 72 hours. On the other hand, large MW cutoff can easily remove free label, as shown in Figure S5. Therefore, filtering free ¹³C₁-mOX063-d₂₄ in larger protein systems is expected to be easier since the MW cutoff can be larger than what is used in this work.

Next, we obtained a Q-band Field-Swept Electron Spin Echo (FS-ESE) spectrum at 40 K. The data is shown in Figure 3A. As a comparison, the FS-ESE spectra of mOX063-d₂₄ is overlaid on the FS-ESE spectrum of ¹³C₁-mOX063-d₂₄-labeled GB1 sample in Figure 3B. The spectrum of ¹³C₁-mOX063-d₂₄ is ca. five-fold wider than mOX063-d₂₄. The breadth of the ¹³C₁-mOX063-d₂₄ spectrum is expected from the 160 MHz value of A_z⁶². Additionally, the spectrum exhibit two central peaks separated by the A_x = A_y of 18 MHz⁶². More importantly, the ¹³C₁-mOX063-d₂₄ spectrum is wide enough for two frequencies separated by ~100 MHz to excite different populations of the label. This result suggests an easy implementation DEER where the observer frequency is set at one peak while the pump frequency is set at the other. In contrast, the same 100 MHz separation cannot fit within the narrow spectra mOX063-d₂₄. Overall, the new ¹³C₁-mOX063-d₂₄ label is well-suited for DEER distance measurements.

Figure 3.

A) Field-swept (FS) spectrum of ¹³C₁-mOX63-d₂₄ at 40 K. The gray dashed lines represent peak separations that correspond to the different hyperfine values of the label. B) Field-swept (FS) spectrum of ¹³C₁-mOX63-d₂₄ and mOX063-d₂₄ at 40 K. The blue dashed lines indicate the frequency separation between the two local maxima of the ¹³C₁-mOX063-d₂₄ spectra.

We performed Q-band DEER on the ¹³C₁-mOX063-d₂₄-labeled GB1 sample, shown in Figure 4. The time-domain signal was analyzed using Consensus DEER Analysis (CDA) ^63,64, which allows for analysis with minimal user bias. Additional information on relaxation rate measurements and fitting analysis is described in ESI. The corresponding distance distribution is shown in Figure 4B, with the most probable distance at 3.6 nm. As a comparison, we overlaid a distance distribution obtained from in silico mOX063-d₂₄ modeling, which was implemented into MTSSLWizard⁶⁵, as described previously⁴. The program randomly generates different rotamers of the label on GB1 and selects those conformations that do not clash with the protein. Figure 4B, shows that the experimental distance distribution is in reasonable agreement with the predicted distribution. Additionally, the most probable distance is consistent with the distance obtained on the same mutant from DQC measurements on mOX063-d₂₄ labeled protein⁴.

Figure 4.

A) Q-band DEER signal of ¹³C₁-mOX063-d₂₄-GB1. The background signal is shown by dashed line. B) The distance distribution (solid line) of ¹³C₁-mOX063-d₂₄-GB1 and the modeled distance (dashed line). The analysis was done using ComparativeDEERAnalyzer (CDA)^63,64. Modeling the conformational space of ¹³C₁-mOX063-d₂₄ was done using MTSSLWizard⁶⁵, as described previously⁴.

In addition to the distance distribution, we also observed a 16% ± 1% modulation depth (λ) from the time-domain signal. To understand the λ, we obtained the probability of exciting the spins from the pump pulse (p_b), as previously described^{33,34,55,66–69}. In summary, p_b can be quantified by dividing the area of the spectra excited by our 16 ns pump pulse by the total area of the spectra, leading to p_b of 30.9%. Given that λ can be expressed as⁶⁸:

$$\lambda =p_b\times f_2$$ (EQ. 1)

where f₂ is the fraction of doubly-labeled species in our sample. From the UV-Vis and room temperature CW-ESR data shown in Figure 2, we estimate 71 μM of doubly-labeled GB1 and 60 μM of free label that forms the ‘singly-labeled’ species in our sample. These concentrations translate into f₂ and f₁ of 53.8% ± 0.5% and 46.2% ± 0.5%, respectively. Based on EQ 1, we expected λ to be 16.6% ± 0.2% which is within the error of our experimental λ of 16% ± 1%. Overall, the results support a high labeling efficiency.

The Q-band ¹³C₁-mOX063-d₂₄ DEER experiment showcases the utility of the new spin-label for DEER distance measurements. However, ¹³C₁-mOX063-d₂₄ DEER is not as sensitive as the typical TAM-based DQC experiments. Specifically, the modulation depth from ¹³C₁-mOX063-d₂₄ DEER is lower than the 80% to 100% modulation depth of DQC⁴³. The sensitivity of TAM-based DQC allows for measurement of sample concentrations as low as 45 nM¹. Despite the sensitivity disadvantage of ¹³C₁-mOX063-d₂₄ DEER experiments, they are easier to perform than DQC. Specifically, DEER only requires 2 to 16-step phase cycles²⁴ to isolate the proper DEER echo, while optimally DQC requires a 256-step phase cycle²² to obtain the correct DQC echo. The extensive phase cycling in DQC is essential as incomplete removal of unwanted echoes can lead to artifacts in the time-domain signal⁴. Therefore, the new ¹³C₁-mOX063-d₂₄ complements other TAM-based labels by providing the option of accessible DEER experiments.

Next, we explored the use of ¹³C₁-mOX063-d₂₄ in an orthogonal-labeling scheme with a Cu(II)-based label, dHis-Cu(II)³⁵. To achieve orthogonal labeling, we engineered and overexpressed an E15C/K28H/Q32H GB1 mutant. The protein was labeled with ¹³C₁-mOX063-d₂₄ at the E15C residue of GB1, while Cu(II)NTA was coordinated to the K28H/Q32H residues (dHis motif)³⁵. The use of Cu(II) chelated to nitrilotriacetic acid (NTA)³⁴ reduces non-specific binding and promotes selectivity to the dHis motif³⁴. Because Cu(II)NTA and our TAM-based labels have different labeling chemistry, selective labeling of the protein is facile⁷⁰.

We assessed Cu(II)NTA labeling of GB1 using 80 K CW-ESR and three-pulse Electron Spin Echo Envelope Modulation (ESEEM)⁵⁸, as shown in Figure 5. We simulated the 80 K CW-ESR spectrum, shown in Figure 5A, using the g_‖ and A_‖ parameters listed in Table S1. Compared with g and hyperfine tensors of free Cu(II)NTA, the value of g_‖ decreases while the value of A_‖ increases in the presence of GB1 mutant, consistent with the expected desolvation and coordination of the dHis motif^33–35. Histidine coordination is further supported in the ESEEM time-domain signal, shown in Figure 5B. Fourier transform of the time domain depicts signal at ~0.5, ~1, and ~1.5 MHz corresponding to Nuclear Quadrupole Interaction peaks and ~4.5 MHz corresponding to Double Quantum peaks⁷¹. These peaks are the signature of histidine coordination with Cu(II)⁷². Since the equivalence of added Cu(II)NTA to GB1 is 1:1 (50 μM:50 μM), we expect 98% labeling efficiency based on the reported binding affinity of dHis-Cu(II)⁷⁰. Overall, the single-component CW-ESR and ESEEM reveal efficient labeling of Cu(II)NTA to GB1.

Figure 5.

A) 80 K CW spectra of Cu(II)NTA and Cu(II)NTA+GB1. The dashed lines represent an EasySpin simulation using a single component. Parameters for the simulation are shown in Table S1. B) ESEEM signal of Cu(II)NTA+ GB1. The ESEEM spectrum is shown in the inset.

After establishing the labeling of dHis-Cu(II) and ¹³C₁-mOX063-d₂₄, we performed an X-band DEER experiment at 40 K of the orthogonally-labeled GB1 sample, shown in Figure 6. The DEER could only be performed at X-band frequency where the separation between the Cu(II)NTA spectra and ¹³C₁-mOX063-d₂₄ is within the resonator bandwidth. The FS-ESE spectrum, shown in the inset of Figure 6A, indicates that the maximum of Cu(II)NTA is ~200 MHz away from the first local maximum of the ¹³C₁-mOX063-d₂₄ spectra. Consequently, we set the observer frequency at the first local maximum of ¹³C₁-mOX063-d₂₄ and the pump frequency at the maximum of the Cu(II)NTA spectrum in the DEER measurement. We note that setting the pump frequency at ¹³C₁-mOX063-d₂₄ instead of Cu(II)NTA is possible. However, observing at Cu(II)NTA leads to a smaller DEER echo compared to observing at ¹³C₁-mOX063-d₂₄, as shown in Figure S8A. Additionally, if the shot repetition time is not long enough to account for the long T₁ of the pumped ¹³C₁-mOX063-d₂₄, the λ is significantly reduced, as shown in Figure S8B. Therefore, observing at the Cu(II)NTA does not improve the sensitivity of the measurement.

Figure 6.

A) DEER signal of orthogonally-labeled GB1 at 40 K. The time trace was analyzed using CDA^63,64. The inset shows FS-ESE spectrum of orthogonally-labeled GB1. The blue lines indicate a 200 MHz separation. B) The distance distribution obtained from DEER at 40 K for Cu(II)NTA+¹³C₁-mOX063-d₂₄ is shown as the solid line. The shaded gray area represents the error of the distance distribution. For comparison, the predicted distance distribution is overlaid on the experimental distribution.

The DEER signal, shown in Figure 6A, was analyzed using Consensus Deer Analysis^63,64 to obtain the distance distribution. Information on relaxation rates and L-curve is shown in Figures S7 and S8. The DEER distance distribution shows one major distance at 3.0 nm and a minor distribution around 2.5 nm. The minor peak at 2.5 nm was reproducible and was observed in two additional biological repeats. These data are shown in Figure S11.

To understand the distribution, we created an in silico model of this orthogonally labeled GB1 mutant. The conformational space of ¹³C₁-mOX063-d₂₄ was obtained identically to Figure 4 and previous publication⁴. On the other hand, we got the conformational space of Cu(II)NTA from an MD simulation of GB1 labeled with Cu(II)NTA at K28H/Q32H⁷³. The MD simulation was done using previously developed force fields for dHis-Cu(II)⁷³. Each frame in the simulation was aligned to the protein to remove any translation or rotation of the complex. As a result, we can visualize the range of position of dHis-Cu(II) with respect to the protein. By combining the conformational space of both Cu(II)NTA and ¹³C₁-mOX063-d₂₄, we obtained distance distribution from the in-silico model. This predicted distribution is shown in Figure 6B. From the in-silico model, the major distance at 3 nm and the minor distance at 2.5 nm in our experimental distribution are within in silico prediction. Details on the minor distribution are explored further in the ESI.

While we measured the distance between Cu(II)NTA and ¹³C₁-mOX063-d₂₄, our labeling scheme has the potential for multiple distance constraints to be obtained only using X-band DEER experiments. As shown in the inset of Figure 6A, we only performed DEER with a 200 MHz pulse separation between Cu(II)NTA and ¹³C₁-mOX063-d₂₄. However, the spectra also indicate that we can use DEER to obtain distance constraints between two ¹³C₁-mOX063-d₂₄ or between two Cu(II) labels. As a result, obtaining multiple distance constraints using Cu(II) and TAM-based labels becomes accessible at X-band frequencies.

Measurements between a TAM label and a Cu(II) label have been typically done using RIDME. Because RIDME is a single-frequency experiment, the measurement is more sensitive using a TAM label with natural abundance of ¹³C rather than ¹³C₁-mOX063-d₂₄. Furthermore, the combination of Cu(II) and TAM has high sensitivity due to the significantly different relaxation rates of the two labels^{11,70,74–77}. When performed correctly, RIDME can have a λ close to 50%, significantly higher than λ of 4% from our DEER measurement in Figure 6A. Additionally, RIDME can be done at Q-band as it does not need to account for the separation between the spectra of the two different labels, further improving the sensitivity of the distance measurement. However, RIDME contains a complicated background signal⁴⁸ which can complicate the subtraction and analysis procedures^27,47,48,78.

Conclusions

In conclusion, this work shows how ¹³C₁-mOX063-d₂₄ can be easily incorporated into a protein. By taking advantage of the resolved splitting of our label, distance measurements can be done using DEER at Q-band, which is unfeasible with many other TAM-based labels. Furthermore, this new label can be orthogonally incorporated with dHis-Cu(II) labeling. This facile orthogonal labeling scheme allows for X-band DEER distance measurements. Additionally, the orthogonal labeling scheme has the potential for obtaining multiple distance constraints only with X-band DEER experiments. These results enhance the power of ¹³C₁-mOX063-d₂₄ for DEER-based distance measurements.

While this work used a unique TAM label that incorporated ¹³C, the label can be synthesized at a large-scale based on recent works^52,79. The cost of ¹³C reagent is negligible in comparison to the whole synthesis. The ¹³C₁-mOX063-d₂₄ may be even more useful for W-band DEER. At W-band, the breadth of nitroxide is wider than ¹³C₁-mOX063-d₂₄. Specifically, the nitroxide spectrum depends on the anisotropy of the g-factor, while the ¹³C₁-mOX063-d₂₄ spectrum depends on the hyperfine interaction of ¹³C₁. Therefore, the sensitivity benefit of ¹³C₁-mOX063-d₂₄ over nitroxide will be more significant at high frequencies.

Scheme 1.

Synthesis of ¹³C₁-mOx063-d₂₄.

Highlights.

• Development of a hydrophilic TAM-based label with ¹³C, ¹³C₁-mOX063-d₂₄ is reported.

• Label has high-efficiency protein labeling.

• The ¹³C resolved splitting of the spin-label enables DEER experiments at Q-band.

• Combination of ¹³C₁-mOX063-d₂₄ and a Cu(II)-based label allows X-band DEER.