Cleavage-resistant Protein Labeling with Hydrophilic Trityl Enables Distance Measurements In-Cell

Abstract

Sensitive in-cell distance measurements in proteins using pulsed-Electron Spin Resonance (ESR) require reduction-resistant and cleavage-resistant spin-labels. Among the reduction-resistant moieties, the hydrophilic trityl core known as OX063 is promising due to its long phase-memory relaxation time (T_m). This property leads to a sufficiently intense ESR signal for reliable long-distance measurements. Furthermore, the T_m of OX063 remains sufficiently long at higher temperatures, opening the possibility for measurements at temperatures above 50 K. In this work, we synthesized deuterated OX063 with a maleimide linker (mOX063-d₂₄). We show that the combination of the hydrophilicity of the label and the maleimide linker enables high protein labeling that is cleavage-resistant in-cells. Distance measurements performed at 150 K using this label are more sensitive than the measurements at 80 K. The sensitivity gain is due to the significantly short longitudinal relaxation time (T₁) at higher temperatures, which enables more data collection per unit of time. In addition to in vitro experiments, we perform distance measurements in Xenopus laevis oocytes. Interestingly, the T_m of mOX063-d₂₄ is sufficiently long even in the crowded environment of the cell, leading to signals of appreciable intensity. Overall, mOX063-d₂₄ provides highly sensitive distance measurements both in vitro and in-cells.

Graphical Abstract

Introduction

Understanding how proteins adapt in their cellular environments is of immense interest in structural biology. The crowded environment inside cells can affect protein folding and stability^1–7. For example, phosphoglycerate kinase (PGK) is more stable in zebrafish tissues⁸ and human osteosarcoma cells⁵ than in vitro. The increase in stability due to molecular crowding has also been observed with other proteins such as frataxin⁹, ubiquitin¹⁰, hen egg white lysozyme¹¹, and calcineurin¹². In contrast, the dimerization of baculoviral IAP repeat (BIR1) domain of X chromosome-linked inhibitor of apoptosis (XIAP) is destabilized in vivo¹³. The destabilization effect are also seen in other protein dimers that are not spherical in shape^14,15. These experiments are indicators that the in-cell environment modulates protein structure and function, which vary case-by-case. Overall, in-cell experiments are required to understand the behavior of proteins in the context of cellular function.

Electron Spin Resonance (ESR) emerged as a widely applicable technique to measure dynamics and distance constraints in vitro and in-cell. For such ESR measurements, the normally diamagnetic proteins can be functionalized with a spin-label using site-directed spin labeling methodologies^16–19. The combination of ESR and spin labeling enables the measurement of the dynamics at the labeled site^20,21 or measuring distances between the labeled sites of a protein^22–28. Distance measurements have been particularly useful for shedding light on the changes in protein conformations^29–37, the assembly of large complexes^38–41, and the binding of substrates and metal ions^42–45. Additionally, these distance measurements have been performed in-cell for proteins^46–48 and DNA^49,50. The primary challenge for distance measurements in-cell is the reduction of spin-labels within the highly reducing cytosolic environment⁵¹. An intriguing new strategy for in-cell measurements is the use of genetically encoded non-canonical amino acids technology as an in situ labeling strategy^52–55. In particular, a photo-caged radical amino acid can be incorporated into a protein during translation⁵⁶. Only after the induction of light will the photo-cage is released to expose the nitroxide radical for ESR measurements. In addition to non-canonical amino acids, reduction resistant spin-labels such as sterically shielded nitroxides^57,58, Gd(III)-based spin-labels^59–61, and triarylmethyls (TAMs, trityls)^62–64 have been developed.

Trityls have a lot of potential as a class of spin-labels for several reasons. First, trityls are highly resistant to reduction in-cell due to the steric-shielding of its radical^65–67. Second, trityls have appreciable relaxation times even at physiological temperatures⁶⁸. Third, trityls have narrow spectral shape that leads to efficient excitation of the electrons and intense ESR signal⁶⁹. Overall, trityl spin-labels have proved suitable for distance measurements at physiological temperatures or in-cell. The most explored trityl spin-labels are based on the Finland trityl radical (FT) shown in Figure 1A, which have successfully provided distance measurements at room temperature^70,71 and in-cell^72,73. However, FT-based spin-labels usage is still challenging due to the complications in the labeling process.

Figure 1.

A) Representation of TAM-based spin labels, FT-MTSL, OX063-d₂₄-MTSL, mOX063-d₂₄, and Ox-SLIM. B) Three-dimensional model of E15C/K28C GB1 based on the wild-type GB1 crystal structure (PDB:2QMT). The side chains of the mutated cysteines are represented as lines.

The spin-labeling process typically entails a reaction between the spin-label and a cysteine residue to label the protein at a specific site. However, FT can bind non-specifically to membranes⁷⁴ or proteins⁷⁵. Additionally, FT tends to self-aggregate^76,77. As a result, efficient labeling of FT requires extensive washing of proteins that are immobilized on a solid support⁶³ or maintaining FT concentration to be less than 30 μM throughout the process to minimize aggregation⁷⁸.

Even after the labeling process, the phase-memory relaxation time (T_m) of FT is significantly reduced upon protein binding^63,78, which leads to a weaker signal. Additionally, the longitudinal relaxation time (T₁) is significantly long at temperatures that are typical for ESR distance measurements (≤ 50K)⁷⁸, which leads to longer experimental time. Overall, the short T_m and long T₁ of FT-based spin labels diminish the sensitivity gain from the efficient excitation of FT. As a result, FT’s sensitivity for distance measurements is comparable to distance measurements using commercially available nitroxide spin-label⁷⁸.

As an alternative, a hydrophilic trityl spin-label, based on the OX063 radical shown in Figure 1A, has been recently developed that allows for straightforward labeling procedure without non-specific binding or aggregation⁷⁹. Interestingly, deuterated OX063 (OX063-d₂₄) was reported to have the longest transversal relaxation time at 50K to date (T_m = 6.3 μs)⁷⁹. Additionally, OX063-d₂₄ has been shown to also have a sufficiently long phase-memory relaxation time even at 200 K (T_m = 3 μs)⁴⁶. Because T₁ is generally shorter at higher temperatures, OX063-d₂₄ has the potential for highly sensitive distance measurements at temperatures higher than 50K. Despite OX063-d₂₄’s improvement over FT, OX063-d₂₄ only utilized a methanethiosulfonate linker so far, which is a limiting factor for in-cell experiments. This linker labels a protein by forming disulfide bonds with cysteines, which can be cleaved inside cells⁴⁸. On the other hand, a maleimide linker reacts with a cysteine to form a thioether bond, which is uncleavable under normal physiological condition⁴⁷. In response to the need of a hydrophilic trityl with an uncleavable linker, a hybrid of OX063 and short-linker maleimide (SLIM)⁷³ known as Ox-SLIM was recently developed (Figure 1A)⁸⁰. Unlike OX063, the trityl core of Ox-SLIM has one of its bisthioketalaryl moieties remain unhydroxylated to bear the short maleimide linker. The hydrophilicity of Ox-SLIM permitted the labeling efficiency of ~85%. These results motivate the development of hydrophilic trityl labels with high labeling efficiency for in-cell distance measurements.

To increase the viability of OX063-d₂₄-based spin-labels, we have developed a new OX063-d₂₄ spin label with a maleimide linker (mOX063-d₂₄) shown in Figure 1A. The maleimide linker allows for mOX063-d₂₄ to maintain its linkage with the protein in-cell⁸¹. We explored two aspects of mOX063-d₂₄ for distance measurements in proteins. First, we show how mOX063-d₂₄ provides highly sensitive distance measurements at temperatures higher than the typical ≤ 50K in vitro. Second, we showcase the usage of mOX063-d₂₄ for experiments in-cell, specifically in Xenopus laevis oocytes. When exploring these two aspects, spin-labeling and distance measurements were done on the immunoglobulin binding domain of protein G (GB1)⁸², a 56-residue globular protein (Figure 1B).

Methods

Synthesis of mOX063-d₂₄.

OX063-d₂₄ trisodium salt (112 mg, 0.077 mmol, 1eq.), synthesized using our previously reported protocols⁸³, was dissolved in anhydrous dimethylformamide (DMF) (100 mL) under argon at room temperature. Benzotriazol-1-yl-oxytripyrrolidinophosphonium hexafluorophosphate (PyBOP) (28 mg, 0.054 mmol, 0.7 eq.) in DMF (1 mL) was added; the green solution turned into a red-brown colored solution. Then, N-(2-aminoethyl)maleimide trifluoroacetate salt (23 mg, 0.09 mmol, 1.2 eq.) in DMF (1 mL) and N,N-diisopropylethylamine (DIEA) (26.8 μL, 1.4 mmol, 2eq.) were added. The solution turned back to green. The reaction mixture was diluted 20x with deionized water and acidified to approximately pH~2 with trifluoroacetic acid. The crude product was loaded into a C18 cartridge and purified by reverse-phase chromatography using a C18 column with a gradient of water/acetonitrile (both containing 0.1 % TFA) 95/5 to 85/15. The purified product was freeze-dried, then dissolved in water, titrated to pH=7 with NaOH, and freeze-dried again to provide 48 mg (40%) of mOX063-d₂₄ as a disodium salt. The purity assessed by HPLC reached >95% as shown in Figure S1. HRMS characterization is shown in Figure S2.

GB1 labeling protocol.

E15C/K28C GB1 expression and purification were performed as previously described⁸⁴. The GB1 mutant was reacted with tris(2-carboxyethyl)phosphine (TCEP) overnight at 4 °C to reduce any disulfide formation. To label the protein, GB1 was run through four 5 mL GE Healthcare Hitrap desalting columns, to remove any TCEP, directly into a solution of mOX063-d₂₄. The final solution of 10:1 of mOX063-d₂₄:GB1 was allowed to react overnight at 4 °C. The spin-labeled protein was concentrated using Sartorius VivaSpin Turbo 4 centrifugal filter units with a molecular weight cutoff of 5 kDa to remove the unreacted label. The final solution was prepared in PBS pH 7.4. Concentration and labeling efficiencies were calculated from UV-Vis measurement using Nanodrop2000 Spectrophotometer from Thermo Scientific. The extinction coefficient of GB1 was obtained from the ProtParam tool (https://web.expasy.org/protparam/). Masses were measured by liquid chromatography electrospray ionization time-of-flight mass spectrometry (LC-ESI-TOF-MS, Bruker Micro TOF, Billerica, MA).

Cellular extracts and oocyte microinjection.

Oocytes were obtained from Carolina Biological Supplies. The cytosol was extracted following previously published protocol⁸⁵. Cytosol sample was prepared at 50 μL containing doubly labeled mOX063-d₂₄-GB1 (200 μM spin concentration) and drawn into Pyrex capillary tubes (I.D. = 0.8 mm) for room-temperature CW experiments. For in-cell pulsed-ESR experiments, 50 nL of doubly labeled mOX063-d₂₄-GB1 (2 mM spin concentration) was microinjected into 12 oocytes following previously published protocol⁸⁵. The microinjected oocytes were inserted into a Quartz Q-band sample tube (2 mm I.D. and 3 mm O.D.) and incubated at room temperature for 30 minutes before being flash-frozen in liquified methylacetylene-propadiene propane (MAPP) gas. The Q-band in-cell sample was estimated to be 60 μL of ~20 μM bulk spin-concentration.

ESR measurements.

Room-temperature continuous-wave (CW)-ESR experiments were performed on a Bruker ElexSys E680 CW/FT X-band spectrometer using a Bruker ER4122 SHQE-W1 high-resolution resonator. CW samples were prepared in Pyrex capillary sample tubes. CW experiments were run at a center field of 3520 G with a sweep width of 20 G, a microwave frequency of ~9.87 GHz, modulation amplitude of 0.07 G or 0.005 G, and modulation frequency of 100 kHz or 300 kHz⁸³ for a total of 1024 or 2048 data points using a conversion time of 30.01 ms.

All pulsed experiments were performed on a Bruker ElexSys E680 CW/FT X-band spectrometer equipped with a Bruker ER5106-QT2 resonator for Q-band and a 300 W amplifier. The temperature was controlled using an Oxford ITC503 temperature controller and an Oxford CF935 dynamic continuous-flow cryostat connected to an Oxford LLT 650 low-loss transfer tube. Echo decay experiments used a two-pulse sequence, π/2−t−π, where t was increased by a step size of 8 ns for 1024 points. T_m values were obtained by fitting the Echo decay results with a stretched exponential decay. The time point where the signal is 1/e of the original intensity is the reported T_m value. Inversion recovery experiments followed a three-pulse sequence, π−t₁−π/2−t₂−π, where t₂ was 400 ns and t₁ was increased by a step size of 1 μs or 10 μs for 1024 points. Fitting of inversion recovery data is detailed in the Supporting Information. Double Quantum Coherence (DQC)^22,86 was performed at the field with the maximum signal intensity of mOX063-d₂₄. The DQC experiments followed a six-pulse sequence, π/2−t_p−π−t_p−π/2−t₁−π−t₁−π/2−t₂−π, where t_p and t₂ were increased and decreased respectively by 10 ns for 136 points. The initial parameters were set as t_p = 1.3 μs, t₁ = 50 ns, and t₂ = 1.5 μs. To remove unwanted echo signal, the 64-step phase cycle was implemented^22,23. The DQC time traces were then analyzed using DeerAnalysis⁸⁷ by Tikhonov Regularization.

SNR was calculated from the raw DQC time traces using previously published method⁸⁸. In summary, the raw DQC time trace was fitted to a 5^th-order polynomial. The fit was subtracted from the time trace to isolate the noise of the time trace. The noise was used by the software SnrCalculator to calculate the final SNR.

Results and Discussion

Our recent report of the synthesis of OX063 triarylmethyl radical and its deuterated analogues OX063-d₂₄⁸³ enables the synthesis of OX063 derivatives such as spin labels. A short maleimide linker was conjugated to OX063-d₂₄ using PyBOP peptide coupling reagent as depicted in Scheme 1. The mOX063-d₂₄ was isolated in 40% yield after purification on C18 alongside with 10% of the di-maleimide derivative.

Scheme 1.

Synthesis of mOx063-d₂₄ from Ox063-d₂₄.

We overexpressed and labeled E15C/K28C-GB1 with mOX063-d24 through a reaction between cysteine residues and the maleimide linker. The labeling reaction occurred by incubating E15C/K28C GB1 and mOX063-d24 in PBS pH 7.4 overnight. The solution was filtered through a centrifugal filter with a molecular weight cut-off of 5kDa to remove free mOX063-d24. We first performed ESI-MS to confirm the covalent attachment of mOX063-d24 to E15C/K28C GB1. The data is shown in Figure S3. The MS showed peaks corresponding to doubly-labeled, singly-labeled and non-labeled E15C/K28C GB1. This result is expected because of the detachment of the spin label during sample preparation in acidic condition (trifluroacetic acid) for ESI-MS and has been reported before using a maleimide-linked FT⁷⁸.

The final product was characterized using UV-Vis to assess spin-labeling efficiency. This data is shown in Figure 2. The spectrum features two distinctive peaks at 280 nm and 469. Only mOX063-d₂₄ contributes toward the 469-nm peak⁸⁹, while both mOX063-d₂₄ and GB1 contribute toward the 280-nm peak. The UV-Vis spectrum was analyzed using the deconvolution method, as depicted in Figure 2⁷⁸, which fits the mOX063-d₂₄-GB1 spectrum using GB1’s UV-Vis spectrum (dash-dotted line) and mOX063-d₂₄’s UV-Vis spectrum (dotted line). The deconvolution allowed us to obtain the absorbance of GB1 at 280 nm (ε₂₈₀ = 9970 M⁻¹ cm⁻¹) and mOX063-d₂₄ at 469 nm (ε₄₆₉ = 16000 M⁻¹ cm⁻¹)⁸⁹, which were used to calculate their concentrations. The final concentrations of GB1 and mOX063-d₂₄ in the sample are 82.0 μM and 161.8 μM, respectively. Therefore, the ratio of GB1: mOX063-d₂₄ purified is about 1:1.97. Overall, our UV-Vis results indicate efficient mOX063-d₂₄ labeling of cysteines on GB1.

Figure 2.

UV-Vis spectrum of mOX063-d₂₄-GB1 sample (gray line). The mOX063-d₂₄-GB1 spectrum was deconvoluted into its GB1 (dash-dotted line) and mOX063-d₂₄ (dotted line) components. The sum of the two components (dashed line) fits well with the mOX063-d₂₄-GB1 spectrum.

To further validate the labeling efficiency, the mOX063-d₂₄-GB1 sample was characterized using CW-ESR at room temperature. Figure 3A and 3B show the CW-ESR spectrum of mOX063-d₂₄ bound to GB1 and free mOX063-d₂₄. The free mOX063-d₂₄ contained a superhyperfine interaction with the amide nitrogen (a^N ~ 220 mG) on the linker, depicted as a partially resolved triplet splitting of the ESR lineshape. This nitrogen hyperfine is consistent with previously published trityls with ¹⁴N-containing linkers^79,90,91. After mOX063-d₂₄ reacted with GB1, the superhyperfine nitrogen were broadened and unresolved due to the slower tumbling rate upon protein binding, seen in Figure 3B⁷⁹. However, the tumbling rate after protein binding is still rapid enough to resolve the satellite ¹³C peaks in the CW of GB1-bound mOX063-d₂₄, seen in Figure 3A. This behavior has been described in a previous report of OX063-d₂₄ spin-label⁷⁹. Spin-counting of the mOX063-d₂₄-GB1 CW spectrum yields a spin concentration of 203 μM. Given the protein concentration of 106 μM, these results indicate a labeling efficiency of 95%, which agrees with the UV-Vis data.

Figure 3.

A) CW-ESR spectra of mOX063-d₂₄-GB1 (top) and mOX063-d₂₄ (bottom). The spectrum of mOX063-d₂₄-GB1 can be fitted with a narrow single-component simulation. The ¹³C satellite peaks are marked with *. B) The CW-ESR spectra of mOX063-d₂₄-GB1 and mOX063-d₂₄ with the observation window ~2 G at the central lineshape. The nitrogen superhyperfine is partially resolved in the mOX063-d₂₄ spectrum but not in the mOX063-d₂₄-GB1 spectrum.

More importantly, the mOX063-d₂₄-GB1 CW spectrum can be fitted with a narrow single-component simulation without a broad component, commonly seen when using FT^74,75,78,92. FT’s broad component has been attributed to aggregated species of FT^76,77 and non-specific binding in proteins^75,76,79,93 and membranes⁷⁴. As a result, when using the simple spin-labeling workflow, FT had labeling efficiencies of 24% to 80% depending on the linker and protein^67,78,79,92. On the other hand, mOX063-d₂₄ is highly soluble and does not bind non-specifically⁷⁹. Therefore, the hydrophilicity of mOX063-d₂₄ allows ~100% labeling efficiency using simple protein labeling protocols. Additionally, the labeling efficiency of mOX063-d₂₄ is slightly improved from the previously developed hydrophilic trityl spin-label, Ox-SLIM, which reported to have 85% labeling efficiency⁸⁰. Such differences in labeling efficiency could be due to the difference in maleimide linker length between mOX063-d₂₄ and Ox-SLIM, and potentially to the differences in solvent accessibilities between the two sites in the two proteins.

Next, pulsed-ESR was used to measure the relaxation times of GB1-bound mOX063-d₂₄ since these are critical parameters that dictate the efficacy of the label in pulsed dipolar spectroscopy. These data were acquired at a spin concentration of 5 μM and the sample was prepared in 20 mM PBS buffer at pH 7.4, and contained 20% glycerol. The phase-memory relaxation time (T_m) was measured by echo decay experiments. The measured values of T_m are listed in Table 1 and the data is shown in Figure S4A. The T_m of GB1-bound mOX063-d₂₄ is 5.1 μs, 4.3 μs, or 3.6 μs at 80 K, 150 K, or 180 K, respectively. These relaxation measurements provided additional data points to the existing measurements from previous studies of OX063-based spin-labels (cf. Table S1). For comparison, the T_m value of 5.1 μs for GB1-bound mOX063-d₂₄ at 80 K is longer then the T_m value of 1.6 μs of protein-bound FT at 80 K⁷⁸. The longest reported T_m of protein-bound FT is 2.9 μs at 50K⁷⁹. Increasing the T_m can both increase the echo intensity for distance measurements and increase the range of feasible temperature of the experiment.

Table 1.

T_m and T₁ measurements of mOX063-d₂₄-GB1 at 80 K, 150 K, and 180 K.

Temperature	T m	T 1
80 K	5.1 μs	1.98 ms
150 K	4.3 μs	0.175 ms
180 K	3.6 μs	0.112 ms

At higher temperatures, mOX063-d₂₄ also benefit from the shortening of T₁. GB1-bound mOX063-d₂₄ has T₁ values of 1.98 ms, 0.175 ms, and 0.112 ms at 80 K, 150 K, and 180 K respectively (Table 1, Figure S4B and S4C). The mechanism for T₁ relaxation of trityl radicals as a function of temperature has been previously studied⁹⁴. As T₁ gets shorter with increasing temperature, the amount of time required for GB1-bound mOX063-d₂₄ to completely relax becomes shorter leading to a faster rate of repeating the measurement. For comparison, the T₁ value of 1.98 ms for GB1-bound mOX063-d₂₄ at 80 K listed in Table 1 is slightly longer than the T₁ value of 1.7 ms of protein-bound FT at 80 K⁷⁸. However, distance measurements using FT are typically done at 50 K or lower which has T₁ values of 6.3 ms or longer⁷⁸. Therefore, distance measurement using mOX063-d₂₄ at higher temperature leads to more scans per unit of time than the distance measurement using FT at the typical temperature of 50 K. Consequently, we expect that distance measurements using mOX063-d₂₄ at higher temperatures benefit from a shorter T₁.

To showcase the sensitivity of mOX063-d₂₄, DQC experiments at 80 K or 150 K were performed on E15C/K28C GB1 doubly-labeled by mOX063-d₂₄, as shown in Figure 4A. The method to measure SNR is described in the methods section. The DQC time trace achieved sufficiently high SNR at 150 K within approximately 1.5 hours of runtime (SNR = 20 min^−1/2). On the other hand, at 80 K, even after 2 hours of runtime, the DQC time trace (SNR = 7 min^−1/2) is noisier than the 150 K DQC time trace. The higher SNR of 150 K DQC than the SNR of 80 K DQC can be rationalized by the following analysis of SNR for pulsed-ESR experiments⁹⁵:

$$SNR\left(T\right)\propto \frac{1}{T}\text{exp}\left[-t_{tot}/T_m\left(T\right)\right]\sqrt{\frac{1}{T_1\left(T\right)}}$$ (Eq.1)

where T is the temperature and t_tot is the amount of time the electron coherence evolves until the detection of the echo signal. The 1/T term in Eq.1 is due to the Boltzmann factor⁹⁶. Based on Eq.1, the shorter T₁ = 0.175 ms at 150 K than the T₁ = 1.98 ms at 80 K (Table 1, Figure S4B and S5C) contributes to 3.36 times improvement in SNR. On the other hand, the increase in temperature from 80 K to 150 K only led to a slight reduction of T_m from 5.1 μs to 4.3 μs. Furthermore, the increase in temperature causes a loss in echo intensity due to the reduction in spin-polarization. Based on Eq.1, the decrease in T_m and spin-polarization reduces the SNR by 0.43 times. As a result, the final SNR at 150 K is 3.36 × 0.43 = 1.44 times higher than the SNR at 80 K. We can see the SNR improvement from the DQC echo comparison between 80 K and 150 K shown in Figure S5. Overall, the gain in sensitivity due to T₁ was able to over-compensate the loss of echo intensity from the shortening of T_m and the reduction of spin-polarization. However, increasing the temperature further to 180 K causes the reduction in sensitivity due to reduced spin-polarization and T_m. For example, the sensitivity at 180 K is ~79% of the sensitivity at 150 K based on Eq.1. These comparisons signify the importance of experimentally evaluating the relaxation times at various temperatures, since the values can be different for different systems. While 80 K is not the most optimal temperature for mOX063-d₂₄ DQC, its SNR = 7 min^−1/2 is comparable to the reported FT’s SNR⁷⁸ ranging from 7 min^−1/2 to 8.9 min^−1/2 at 50 K. These comparison of SNR exemplifies the sensitivity gained from performing mOX063-d₂₄ DQC experiments at the optimal temperature.

Figure 4.

A) DQC time traces of doubly-labeled mOX063-d₂₄-GB1 at 150 K after 1.5 hours of runtime and at 80 K after 2 hours and 16 hours of runtime. B) Distance distributions obtained from the 150 K and 80 K DQC time traces using DeerAnalysis. The gray regions represent the error obtained from the validation function in DeerAnalysis. Additionally, a distance distribution was also obtained from in silico modeling using MTSSLWizard. C) In silico model from MTSSLWizard using GB1 (PDB:2QMT) and mOX063-d₂₄. The two clusters represent the space occupied by the radical carbon.

We analyzed the time traces using the DeerAnalysis2018⁸⁷ package and the Tikhonov Regularization method to extract the distance distributions shown in Figure 4B. Expectedly, at both temperatures, the distance distributions were close to identical, with the most probable distance of 3.6 nm. In order to predict the distance distribution, we built an in silico model using MTSSLWizard⁹⁷. Since the mOX063-d₂₄ spin-label does not exist in the MTSSLWizard package, we first implemented the mOX063-d₂₄ model into the MTSSLWizard software. Details are provided in Figure S6. The model predicted that the most probable distance is 3.8 nm, as shown in Figure 4B, which is in reasonable agreement with the DQC results. Furthermore, the experimental results have a standard deviation of ~0.6 nm, which is on par with the standard deviation of ~0.8 nm obtained using nitroxide on the same GB1 mutant⁸⁵.

After the in vitro experiments, the viability of mOX063-d₂₄ for in-cell experiments was explored. Specifically, the mOX063-d₂₄-GB1 (200 μM of spins) was subjected to either ten times excess of ascorbic acid or the cytosol extract of Xenopus laevis (African Bullfrog) oocytes. Cytosol was extracted from oocytes using previously published protocol⁸⁵. The signal intensity of mOX063-d₂₄ was monitored over time using CW ESR, and the maximum intensity of each CW was plotted against time in Figure 5A. The signal intensity decays to about 97 and 95% of its original intensity in ascorbate and cytosol after 5 hours, respectively. The stability of mOX063-d₂₄ is on par with the stability of other trityls^73,80. The signal persistence of mOX063-d₂₄ showcases the reduction-resistance of mOX063-d₂₄ against the cytosolic antioxidants that play a role in reducing radicals in-cell⁹⁸.

Figure 5.

A) A plot of the maximum intensity of mOX063-d₂₄-GB1 vs. time in 10 times excess of ascorbate or cytosol extracted from Xenopus laevis oocytes. The height of the vertical bars represents the RMSD in the CW spectrum. B) DQC time trace of mOX063-d₂₄-GB1 at 80 K Q-band before and after artifact subtraction. C) Distance distribution (most probable distance of 3.6 nm) extracted from the artifact-subtracted time-domain signal using DeerAnalysis. The gray region represents the error obtained from the validation function.

After measuring mOX063-d₂₄ stability, mOX063-d₂₄-GB1 was injected into oocytes and incubated for 30 minutes after injection before flash-freezing the sample. The 30 minutes incubation allows for mOX063-d₂₄-GB1 to completely diffuse in oocytes⁴⁸. Echo decay experiment measured the T_m of mOX063-d₂₄ at 80 K in oocytes to be 4.3 μs as shown in Figure S7, which is shorter than the T_m of mOX063-d₂₄ in vitro shown in Table 1 and Figure S4A. The lower GB1-bound mOX063-d₂₄ T_m in-cell compared to in vitro was expected because of the crowded environment in-cell. The crowded environment can lead to an increase in the local concentration of protons near the radical which enhances the contribution of electron-nuclei interactions to relaxation. In addition, the presence of paramagnetic metal ions, primarily Mn(II)⁹⁹, in the cell can enhance relaxation. However, the T_m of mOX063-d₂₄ in oocytes is surprising since previous reports of other organic spin-labels used in-cell (nitroxides^50,57,100 and FT^67,72,73) have T_m values in the range of 0.6–2 μs. Therefore, mOX063-d₂₄ also improves the sensitivity of distance measurements in-cell due to the T_m that is at least 2 times longer than previously published T_m of nitroxide or FT in-cell.

Distance measurements of mOX063-d₂₄-GB1 in oocytes were done using DQC at 80 K shown in Figure 5B. We observed an artifact that overlaps the desired DQC signal at zero time. Such an artifact has been seen previously and attributed to trityl dimers and to partial labeling of non-cysteines residues such as lysine¹⁰¹. We repeated the labeling procedure on WT GB1 that has no cysteine residues. After concentrating the sample, UV-Vis indicates no presence of mOX063-d₂₄ as shown in Figure S8. We expected this result since our previous work using GB1 and maleimide-linked nitroxide (5-MSL) did not show over-labeling of the protein⁸⁵. In addition, we did not see an ESR signal from the WT GB1 sample, as shown in Figure S9, which also excludes the presence of dimers.

We attribute this artifact to the formation of a small echo generated by the first and the fourth pulses in the DQC 6-pulse sequence. This interference can be readily seen in the 2D contour plot of the DQC signal shown in Figure S9A. As a result, the DQC time trace contained an artifact shown as a sharp feature at the t = 0 shown in Figure S10B, which led to improper fitting of the time trace. Additionally, the artifact contributed to a short distance around 2 nm shown in Figure S11A. The artifact seemed to be a result of inefficient phase-cycling in our DQC experiment and is evident in the in-cell data due to the lower SNR.

To support this hypothesis, we performed the DQC experiment using the same parameters on 300μM TEMPOL as shown in Figure S12. The same artifact was seen crossing the desired DQC echo at a slanted angle shown in the 2D contour plot in figure S12A. As a result, a sharp feature at t=0 manifested as shown in Figure S12B. This artifact has not been seen in previous works. The artifact was more prominent in the in-cell experiment than in the in vitro experiment for two reasons. First, the measured echo in the in-cell DQC was half as intense as the measured echo in the in vitro DQC. The lower in-cell echo intensity is due to the shorter T_m in-cell than the T_m in vitro. Additionally, reduction of mOX063-d₂₄ can still occur due to the contribution of oocytes’ membrane-associated factors⁸⁵ such as thioredoxin¹⁰² and glutathione reductase¹⁰³, which are not accounted for in our cytosol stability measurement. These two contributions led to a less intense measured echo causing the artifact to be prominent in the in-cell DQC.

To remove the artifact in the DQC time trace in oocytes, DQC was performed on a sample of free mOX063-d₂₄, which contained only the artifact shown in Figure S10B. The free mOX063-d₂₄ DQC time trace was used to subtract the artifact from the time trace of mOX063-d₂₄-GB1 in oocytes shown in Figure 5B. The artifact-subtracted time trace was used to extract the distance distribution shown Figure 5C, which agrees quite well with the in vitro distance measurements in Figure 4B. Furthermore, we were able to repeat the in-cell DQC experiment at 150 K, shown in Figure S13, and obtain a similar distribution as the 80 K in-cell distribution. Additionally, we repeated our in-cell experiments at 80 K using a different batch of oocytes and newly overexpressed and labeled-GB1 to ensure that the in-cell results are reproducible. This data is shown in Figure S14. Overall, we obtained a highly sensitive distance measurement in oocytes using mOX063-d₂₄.

Conclusion

In conclusion, this work showed that mOX063-d₂₄ has a high protein-labeling efficiency of ~97%. Furthermore, we showed that in vitro distance measurements of mOX063-d₂₄ is more sensitive at higher temperatures. Finally, we obtained distance measurements using mOX063-d₂₄ in-cell which agree with in silico modeling. This work adds to the library of spin labels that can be used for in-cell work. In particular, mOX063-d₂₄ is similar to Ox-SLIM⁸⁰ since both are hydrophilic spin-labels with a maleimide linker, as shown in Figure 1A. However, these two spin-labels differ in their trityl cores and linker lengths. These differences provide variation in the labeling efficiency, T_m, and breadth of distance distribution. In one case, Ox-SLIM’s short linker length can provide narrow distance distributions that can readily resolve different protein conformations^73,80. On the other hand, mOX063-d₂₄ provides longer T_m and higher labeling efficiency leading to the sensitivity improvement in the distance measurements. Overall, Ox-SLIM and mOX063-d₂₄ are complementary to each other due to their differences.