Rigid and stable nitroxide spin label for high-resolution distance measurements on proteins by DEER experiments

Abstract

The distance distributions between two site-specifically anchored spin labels in a protein, measured by pulsed electron-electron double resonance (PELDOR or DEER), provide rich sources of structural and conformational restraints on the proteins or their complexes. The rigid connection of the nitroxide spin label to the protein improves the accuracy and precision of distance measurement. We report a new spin labelling approach by formation of thioester bond between nitroxide (NO) spin label, NOAI (NO spin labels activated by acetylimidazole), and a protein thiol, and this spin labeling method has demonstrated high performance in DEER distance measurement on proteins. The results showed that NOAI has shorter connection to the protein ligation site than 2, 2, 5, 5-tetramethyl-pyrroline-1-oxyl methanethiosulfonate (MTSL) and 3-maleimido-proxyl (M-Prox) in the respective protein conjugate and produces narrower distance distributions for the tested proteins including ubiquitin (Ub), immunoglobulin-binding β1 domain of streptococcal protein G (GB1), and second mitochondria-derived activator of caspases (Smac). The NOAI protein conjugate connected by a thioester bond is resistant to reducing reagent and offers high-fidelity DEER distance measurements in cell lysates.

Graphical abstract

1. Introduction

The unique distance constraints obtained by electron-electron double resonance (DEER, also called PELDOR), usually performed on frozen solutions, provide valuable structural information about biomolecules and their complexes [[1], [2], [3], [4]]. DEER is particularly suitable for tracking conformational changes of biomolecules upon ligand binding [[5], [6], [7]], determining oligomeric states of proteins [[8], [9], [10]], dissociation constants [11,12], oligomer stability and conformational changes of proteins in cells [[13], [14], [15]]. In general, DEER can provide distances in the range of 15–80 Å between two spin labels that are site-specifically attached to the target molecules [16,17] and longer distances up to 160 Å can also be achieved for deuterated biomolecules [18,19]. The spin labels are usually introduced via the site-directed spin labeling (SDSL) method [[20], [21], [22], [23], [24]]. The widely used spin labels are based on nitroxide (NO) radicals, which are attached to proteins via native or engineered cysteine residues. 2, 2, 5, 5-Tetramethyl-pyrroline-1-oxyl methanethiosulfonate (MTSL) is the most used NO spin label due to its commercial availability and ease ligation. MTSL has proven to be an excellent NO spin label for studying protein dynamics using EPR [[25], [26], [27]]. In the last decade, great efforts have been made to improve the rigidity of paramagnetic metal labels for DEER measurement, including the synthesis of rigid metal chelating moieties, particularly for Gd(III) [[28], [29], [30]] and Mn(II) [31,32] labels. It was found that shortening the linker between the protein ligation site and the metal tags resulted in narrower distance distributions [32]. Alternatively, an approach of anchoring Cu(II) with two histidines in a protein has also been proposed to improve the resolution of distance distributions [12,24].

The NO spin labels activated by either acetylimidazole (NOAI) [33,34] or NONHS (NO activated by N-hydroxysuccinimide ester) [35,36], each sharing a common NO radical core, were reported a few decades ago and were only used as the intermediates in the synthesis of other NO spin labels. In this work, the feasibility of protein ligation using these two NO spin labels and performance in DEER distance measurements were examined. We first assessed the possibility of increasing the spin label rigidity by forming a thioester bond between the NO spin label and the thiol of a cysteine residue (Fig. 1). Then we examined the impact of NO spin label rigidity on DEER distance distributions on the tested proteins, including GB1, Ub and Smac. GB1 and Ub are rigid single-domain proteins, whereas Smac is a stable homodimer and each subunit contains three helixes [37,38]. We next assessed the stability of NO-protein conjugated products in the presence of reducing reagents such as 1,4-dithiothreitol (DTT), glutathionine (GSH), and evaluated the capability of DEER measurements in cell lysates for the NO-labeled proteins.

Fig. 1.

Chemical reaction in site-specific labeling of a protein with an NO spin label via the reaction of a spin label with a protein thiol (SH). The chemical structures of NO spin labels were shown and the thiol reactive moiety was highlighted in red.

2. Results and discussion

The synthesis of NO spin labels, details of protein ligation, preparation of EPR samples, EPR experiments and data analysis were provided in the supplementary material. The reaction of N-hydroxysuccinimide ester (NHS) derivatives with free thiols to form stable thioester derivatives [39] encouraged us to examine the possibility of using this reaction for site-specifically labeling of proteins with a NO radical via the formation of a thioester bond. The NHS-moiety is commonly used for bioconjugation with amine groups [40] and exhibits high reactivity toward the amino groups of proteins. Indeed, significant amide ligation products were determined for the mixture of NONHS and wild-type ubiquitin protein, which does not contain a cysteine residue (Fig. S1a). In contrast, NOAI has no obvious reactivity with the amino sidechains in the protein as determined by the ESI-Q-TQF mass spectrometry (Fig. S1b), suggesting that NOAI is a better spin label for the protein thiol than NONHS.

The double cysteine mutant proteins including GB1 T11C/V21C and Ub D39C/E64C, and single cysteine mutant of dimeric Smac S35C were prepared and purified. NOAI was tested for site-specific spin labeling of these model proteins via thioester formation (Fig. 1, Fig. 2). The reaction of NOAI with these protein constructs is generally rapid in aqueous solution and the labeling products were assessed by mass spectrometry. In the mass spectra, only the mass peak corresponding to the doubly labeled protein was observed and no species of free protein or more than two spin labels attached to the protein were present, confirming the high labeling efficiency and selectivity of NOAI to the solvent exposed cysteines. The ligation reaction of NOAI with protein thiol generally completes within an hour at room temperature, which is similar to the wildly used commercial NO labels, MTSL and M-Prox (Fig. 1, Fig. 2). The labeling efficiency was estimated as 95%–100% via X-band continuous wave (CW) EPR spectrum of the NO conjugates (Fig. S2). Overall, the reaction of NOAI in forming the thioester bond with the cysteine residue is unique and no obvious amide products were generated in the protein conjugates, which is in great contrast to NONHS. The high specificity of NOAI for the formation of a thioester bond is suitable for SDSL. Similarly, MTSL and M-Prox were also applied for spin labeling of these proteins for comparison (Fig. 2). The reaction of MTSL with a protein thiol produces a disulfide bond between the NO spin label and the target protein, in which the disulfide bond is unstable in reducing environment or experiences thiol-exchange at higher pH. M-Prox contains a Michael receptor and it produces a stable thioether bond with a protein thiol group, and this thioether is resistant to the reducing reagent but the generation of new chiral center might compromise the distance resolution in DEER measurement.

Fig. 2.

(a) Structural representation of GB1 (left), ubiquitin (middle) and Smac (right), in which Cα atoms of T11 and V21 in GB1 (PDB code: 2QMT [41]), D39 and E64 in ubiquitin (PDB: 1UBI [42]) and S35 in Smac (PDB: IFEW [37,38]) are shown as red spheres. The distances between two Cα atoms connected by a dashed line are denoted. (b) ESI-Q-TOF mass spectra recorded for the samples of GB1 T11C/V21C, Ub D39C/E64C and Smac S35C mutants and the respective conjugates with nitroxide spin labels. (c) Background corrected Q-band DEER traces of GB1-NO (left),Ub-NO (middle) and Smac-NO (right) conjugates, and their corresponding fits (gray). (d) Distance distributions (solid lines) and validations (shaded areas) obtained with model-free Tikhonov regulation using DeerAnalysis (pink) [43] and DeerNet method (dark gray) [44]. The coloured areas in x-axis denote reliability regions as follows: shape reliable (green), mean and width reliable (yellow), mean reliable (orange). Detailed information of DEER data is shown in Fig. S3-5.

The reaction of NOAI with a protein thiol generates a shorter tether than MTSL and M-Prox in the protein conjugates (Fig. 1). It is expected that the shorter tether in the DEER measurement might produce narrower distance distributions, provided that no large conformational exchanges in the target protein. The DEER performance was first examined on NO spin-labeled GB1 and ubiquitin, which are rigid single-domain proteins with unambiguous structural conformations. Smac is a homo-dimeric complex and has three long helixes in each monomeric subunit [37,38]. The calculated distances between two Cα atoms in the ligation sites according to the 3D structure of the protein are shown in Fig. 2a. As shown in Fig. 2c and d, large differences in the distance distributions were observed for the NO spin labeled protein conjugates in the DEER measurements and obtained with model-free Tikhonov regulation using DeerAnalysis software, suggesting the great impact of spin labels on the distance distributions even for the rigid proteins. The DEER results are summarized in Table 1. To eliminate the distractions of over analysis, DeerNet method [44], which predicts the DEER distances in a user-independent manner using the deep neural network, and DEERLab [45], which automatically and simultaneously fits the background function and the distances, were used for comparison (Fig. 2d, Fig. S3-5).

Table 1.

Comparison of the mean distance (r) and full width at half-maximum (FWHM) obtained from Q-band DEER measurements of GB1-NO, Ub-NO and Smac-NO conjugates.

	GB1 T11C/V21Ca		Ub D39C/E64C		Smac S35C
	r/nm	FWHM/nm	r/nm	FWHM/nm	r/nm	FWHM/nm
NOAI	4.29/3.85	0.37/0.44	3.47	0.35	3.23	0.70
MTSL	4.09/3.70	0.44/0.34	3.55	0.47	3.22	0.98
M-Prox	4.39/3.74	0.51/0.72	3.73	0.86	2.70	1.47

^aFor the GB1 T11C/V21C-NO conjugates two main distances were determined, and the FWHMs of the two peaks were evaluated from deconvolution of the spectra (Fig. S6).

The full width at half-maximum (FWHM) of the distance distribution is significantly wider for the M-Prox conjugates than for MTSL and NOAI. Because no evidence of hydrolyzed M-Prox was determined in all the protein conjugates as examined by mass spectrometry, the wider distance distribution is likely due to the flexible and longer tether formed in the reaction of the maleimide moiety and the protein thiol (Fig. 1), in which a new chiral center was also generated. It is noted that two main distances were determined in all the NO spin labeled GB1 T11C/V21C conjugates, which are likely caused by the interactions between the NO spin label attached at V21C and the N-terminal residues according to the 3D structure of GB1 (Fig. 2 and Fig. S6). The two mean distances were analysed by further deconvolution of the DEER obtained distance distribution, and the major peak at 4.29 nm in NOAI conjugates shows a narrower FWHM than MTSL (Table 1). The FWHM for the ubiquitin D39C/E64C conjugate of NOAI is much narrower than those of MTSL and M-Prox. A similar trend was observed in the NO conjugates of Smac (Table 1).

The FWHM determined between two spin labels in a rigid single-domain protein reflects the conformational distributions resulting from protein-tag tether, the tag itself and the ligation site on the protein surface. The narrower FWHM detected in the protein-NOAI conjugates indicates a high rigidity of the thioester tether between the NO radical and the protein ligation site, which is due to the shorter length compared to those of MTSL and M-Prox. Among the three NO spin labels, M-Prox shows the least quality in distance distributions for DEER measurement and it provides wider distance distributions than NOAI and MTSL, and it also generates a new chiral center in the protein conjugates.

To understand the rigidity and tether length of the NO spin label (Fig. 1) in the protein-NO conjugate, we performed molecular dynamic simulations using the Xplor-NIH program [46,47]. The coordinates of the protein-NO conjugates of GB1 T11C/V21C and Ub D39C/E64C were generated and the distance d(Cα-NO) between the Cα to the oxygen atom of the NO radical (NO•) were simulated, in which the Cα is the protein backbone atom at the ligation site (Fig. 3). The dominant conformation fractions for the protein conjugates of NOAI have an average d(Cα-NO) of 8.5–9 Å, which is generally shorter than those of MTSL and M-Prox. As expected, the protein conjugates of M-Prox exhibit larger d(Cα-NO) with broader distributions than those of NOAI and MTSL. These data confirmed that for a rigid-body protein without large conformational exchanges, the distance resolution in the DEER measurements is proportional to the rigidity of spin label in the protein-NO conjugates. These results are consistent with EPR analysis based on previous reports [[48], [49], [50], [51]].

Fig. 3.

Structural representation of randomly generated 500 conformers of NOAI, MTSL, and M-Prox in GB1 T11C/V21C-NO (a) and Ub D39C/E64C-NO (b) conjugates using the Xplor-NIH program [46,47]. (c) Plot of the conformational fractions of the distances of the Cα atom in the ligation site to the oxygen of NO• in the NO spin label, d(Cα-NO), in the ensemble conformers as shown in (a) and (b).

The stability of thioester bridged protein-NO conjugate was assessed in the presence of reducing reagents including DTT and GSH. 50 μM GB1 T11C/V21C-NO was incubated with DTT and GSH, respectively, and the mixture was monitored by CW-EPR spectra at different time points. The significant changes in line shapes of EPR spectra indicated that GB1 T11C/V21C-MTSL conjugate decomposes gradually with incubation time due to the thiol-exchange process in the disulfide bond bridged protein-NO conjugate (Fig. 4a, Fig. S7a). In contrast, the overall EPR spectrum pattern of the GB1 T11C/V21C-NOAI conjugate remains essentially intact even after incubation with 0.25 mM DTT (Fig. 4b), 0.25 mM GSH and 1 mM GSH (Figs. S7b–c) for 24 h, whereas little decrease in the intensity due to a small portion of reduced nitroxide was observed with 5 mM GSH (Fig. S7d). After 24 h incubation with 1 mM DTT (Fig. 4c), over 87% of GB1 T11C/V21C-NOAI conjugate remains intact, while about 13% nitroxide decomposes from the proteins.

Fig. 4.

Time-dependent X-band CW-EPR spectra were recorded for MTSL (a) and NOAI (b–c) labeled GB1 T11C/V21C after incubation with different concentration of DTT as noted in the insert. Comparison of EPR data on GB1 T11C/V21C-NOAI conjugates in vitro (normal buffer) and in E. coli lysate (d–e). (d) X-band CW-EPR spectra at room temperature. (e) Background corrected DEER data (black, red) with fitting traces (gray). (f) Obtained distance distributions in E.coli lysate (solid lines) and validations (shaded areas) derived from model-free Tikhonov regulation with DeerAnalysis (pink) and DeerNet (dark gray). The distance distribution in buffer analysed with DeerAnalysis (dash line) was shown for comparison. The coloured areas in x-axis denote reliability regions as follows: shape reliable (green), mean and width reliable (yellow), and mean reliable (orange).

To further assess the stability of thioester-bridged protein-NOAI conjugate in cell lysates that contain multiple reactive components, we used X-band CW-EPR and Q-band DEER measurements to monitor the spectral changes with incubation time. We used the in-house protocol for protein-NO EPR measurement in maleimide treated cell lysates, because treatment of cell lysates with maleimide results in stabilized NO radicals in the cell lysates by exhausting the NADH/NADPH that are essential cofactors of the NO-destabilizing enzymes [52]. The thioester-bridged protein-NO conjugate is stable in the cell lysate that contains multiple nucleophiles including amines and hydroxyls. The CW-EPR spectra recorded for the sample of GB1 T11C/V21C-NOAI (Fig. 4d) and Ub D39C/E64C-NOAI (Fig. S8a) in the cell lysate (wet weight 200 mg/mL) treated with 5.0 mM maleimide remained unchanged after incubation at room temperature over 1 h. For the DEER measurements in the E. coli lysate, the DEER time traces showed similar modulation depth, eg. 21% in buffer and 19% in E. coli lysate for GB1 T11C/V21C-NOAI, and consistent distance distributions were obtained (Fig. 4e–f and Figs. S8b–c). This high stability of NOAI protein conjugate is feasible for the long-time tracking the interaction and dynamics of biomolecules by EPR.

3. Conclusion

In summary, we present a novel approach in site-specific labeling proteins with a rigid spin label, NOAI, for pulsed EPR measurement. We examined the rigidity of the tether between the protein and the NO spin label for DEER measurement. The results showed that rigid spin labeling tends to narrow the distance distributions in the DEER data for protein-spin label conjugates as demonstrated in a number of rigid single-structural domain proteins. The high stability of NOAI in reducing conditions as well as E. coli lysates offers the suitability for the applications of NO spin labeling of proteins for pulsed EPR measurement in complex system. The stability of NO spin label in cells or under reductive conditions needs to be improved. The quality of the spin labels in terms of size, stability, and minimal interference with the target biomolecules is crucial for deciphering the informative activities of biomolecules manifested in the EPR spectra. We have to point out that the thioester bond formed by the protein thiol and NOAI is not as stable as amide or thioether bond in the protein conjugates and cautions have to be borne in mind in EPR measuremeasures in particular cases or harsh conditions like higher pH. The high performance of NOAI demonstrates accurate and precise distance measurements on proteins by DEER experiments. We believe that the power of DEER experiments would be strengthened in this direction and near residual-specific resolution of DEER data would be of great value to understand the dynamics and interactions of biomolecules by pulsed EPR spectrometry.

CRediT authorship contribution statement

Ya-Ting Chen: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Formal analysis, Data curation. Xing Zhang: Writing – original draft, Visualization, Validation, Investigation, Formal analysis, Data curation. Jia-Liang Chen: Software, Formal analysis. Bin-Bin Pan: Data curation. Daniella Goldfarb: Writing – review & editing, Visualization, Validation, Investigation. Yin Yang: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation, Formal analysis, Data curation. Xun-Cheng Su: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Xun-Cheng Su is an editorial board member of Magnetic Resonance Letters but was not involved in the editorial review or the decision to publish this article.

Biographies

Yin Yang received her B.S. degree in biology and her Ph.D. in chemical biology in from Nankai University. After her post-doctoral scholar at Weizmann Institute of Science, Israel, she is currently a research fellow in Nankai University. Her research field is chemical biology and structural biology by EPR spectroscopy.

Xun-Cheng Su is Professor of State Key Laboratory Elemento-organic Chemistry at Nankai University. He received his Ph.D. at Nankai University in 2001. After his postdoctoral research at CERM (Florence) and Australian National University, he joined Nankai University in 2010. His research focuses on the development of biological magnetic resonance spectroscopy (NMR and EPR) in solution and in cells.

Appendix A. Supplementary data

The following is the Supplementary data to this article: