Synthesis and Properties of (1R(S),5R(S),7R(S),8R(S))-1,8-Bis(hydroxymethyl)-6-azadispiro[4.1.4.2]tridecane-6-oxyl: Reduction-Resistant Spin Labels with High Spin Relaxation Times

Site-directed spin labeling followed by investigation using Electron Paramagnetic Resonance spectroscopy is a rapidly expanding powerful biophysical technique to study structure, local dynamics and functions of biomolecules using pulsed EPR techniques and nitroxides are the most widely used spin labels. Modern trends of this method include measurements directly inside a living cell, as well as measurements without deep freezing (below 70 K), which provide information that is more consistent with the behavior of the molecules under study in natural conditions. Such studies require nitroxides, which are resistant to the action of biogenic reductants and have high spin relaxation (dephasing) times, Tm. (1R(S),5R(S),7R(S),8R(S))-1,8-bis(hydroxymethyl)-6-azadispiro[4.1.4.2]tridecane-6-oxyl is a unique nitroxide that combines these features. We have developed a convenient method for the synthesis of this radical and studied the ways of its functionalization. Promising spin labels have been obtained, the parameters of their spin relaxation T1 and Tm have been measured, and the kinetics of reduction with ascorbate have been studied.


Introduction
Site-directed spin labeling coupled to Electron Paramagnetic Resonance spectroscopy (SDSL-EPR) is a rapidly expanding powerful biophysical technique to study biomolecules in physiologically relevant environments [1][2][3][4][5][6]. These methods are based on site-specific introductions of paramagnetic labels (unpaired electrons) into biomolecules of interest with subsequent investigation using various EPR techniques. The latter give inter-spin distances, solvent accessibility, the polarity of its immediate environment, and the dynamics of the labeled region, which are complimentary to information obtained by other methods of structural biology, such as NMR, X-ray crystallography, and cryo-electron microscopy. This helps to identify biologically important conformations of large biomolecules, especially membrane and intrinsically disordered proteins. Nitroxides are the most widely used spin labels [7]. Unlike other spin labels, they can be used both to study local dynamics in biomolecules at physiological temperatures and to measure inter-spin distances [5]. Examples of specific applications of nitroxide spin labels to structural biology studies of protein can be found in hundreds of publications [2].
The best way to obtain correct information about the native structure and functions of biomolecules implies their study in natural conditions-in a living cell. Currently, a set of distances [5]. Examples of specific applications of nitroxide spin labels to structural biology studies of protein can be found in hundreds of publications [2].
The best way to obtain correct information about the native structure and functions of biomolecules implies their study in natural conditions-in a living cell. Currently, a set of evidence has been accumulated for the difference in protein structure and function in model and natural conditions [8][9][10]. For this reason, SDSL-EPR experiments in living cells attract more and more attention. A serious obstacle to the use of nitroxide spin labels in living cells is the rapid reduction of the nitroxide into diamagnetic compounds by low molecular weight reductants and enzymatic systems, typically found within cells [11]. Resistance of nitroxides to bioreduction can be strongly improved via introduction of several bulky alkyl (larger than methyl) substituents to the α-carbon atom of the nitroxide group [12]. A number of reduction-resistant spin labels for in-cell application have been developed based on tetraethyl nitroxides [13][14][15][16][17][18][19].
Several EPR methods have been developed for inter-spin distance measurement [20]. All of them are dependent on electron spin relaxation parameters of spin labels [2,21,22]. In Pulsed Electron-Electron Double Resonance (PELDOR or DEER) technique, which is currently the most popular approach for distance measurements, the maximal distance one can measure and the precision of the distance distribution are determined by the phase memory time (Tm) of the spin label [22]. The Tm parameter is temperature-dependent and must be as high as possible. To achieve optimal performance using conventional tetramethyl or tetraethyl spin labels, the measurements must be carried out at 40-65 K, because higher temperature rotation of the alkyl groups leads to a decrease in Tm. This rotation is impossible in nitroxides with spirocyclic moieties at α-carbons of nitroxide group, and these spirocyclic spin labels can be used for measurements at much higher temperatures (125 K) and even at room temperature [23]. Regretfully, the spirocyclic spin labels demonstrate much lower resistance to bioreduction compared to tetraethyl nitroxides [24].
According to expert estimates, labels that can eliminate the need for data acquisition at cryogenic temperatures and labels that can enter and survive in the cellular environment are of current and future interest [2]. The attempts to improve reduction resistance of spirocyclic nitroxides have been a subject of recent research [25]. The literature data on the rate constants of chemical reduction of representative dispirocyclic nitroxides and tetraethyl nitroxides are listed in the Figure 1. The data demonstrate that reduction resistance of the nitroxide 1 by far exceeds those of other spirocyclic nitroxides. The nitroxide group in 1 is stabilized with two spiro-(2-hydroxymethyl)cyclopentane moieties with the hydroxymethyl groups directed towards N-O • . In addition, this nitroxide showed the highest longitudinal relaxation time T1 among a broad set of various nitroxides [26]. This parameter is very important for distance measurement using Saturation Recovery (SR) method, another pulsed EPR technique widely used in structural biology [2,21].  [25,[27][28][29][30].
Recent advances in condensation, cyclization, and dipolar cycloaddition cascade chemistry [32] allowed us to develop convenient and scalable protocol for synthesis of dispirocyclic nitroxide 5, a 3,4-unsubstituted analog of 1, with overall yield 43% from commercially available 4-chlorobutyryl chloride (6). Chemical properties of the nitroxide 5 were studied and several dispirocyclic spin labels were prepared. The new nitroxides showed improved resistance to reduction compared to 1. The spin relaxation times T1 and Tm were measured for some of the nitroxides.

MCPBA
The nitrone 12 easily undergoes intramolecular 1,3-dipolar cycloaddition affording single product. The best yield was achieved upon heating to reflux in toluene in presence of 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO). Intramolecular cycloaddition in 2-pent-4en-1-ylpyrroline 1-oxides was reported to give a single regioisomer [27,30,33]. Similarly, the 1 H and 13 C NMR spectra and 1 H-1 H and 1 H- 13 [1,2-b]isoxazole-3,1 -cyclopentane]-2 -yl) methanol system. However, one could not exclude the possibility of C=C bond approaching the plane of the nitrone group from different sides with the formation of cisand trans-isomers relative to the position of the hydroxymethyl group. To assign a structure to the cycloadduct 1 H-1 H NOESY NMR spectra were recorded ( Figure S67). 1 H-1 H NOESY correlation showed a cross-peak between protons of 6 CH 2 and 15 CH 2 groups, while no cross-peak was observed between 11 CH 2 and 6 CH 2 protons ( Figure 3). Apparently, the hydroxymethyl group prevents C=C from approaching the nitrone group, and the formation of a cycloadduct is possible only through the transition state 15 (Scheme 2). The nitrone 12 easily undergoes intramolecular 1,3-dipolar cycloaddition affording single product. The best yield was achieved upon heating to reflux in toluene in presence of 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO). Intramolecular cycloaddition in 2-pent-4-en-1-ylpyrroline 1-oxides was reported to give a single regioisomer [27,30,33]. Similarly, the 1 H and 13 C NMR spectra and 1 H-1 H and 1 H-13 C correlations revealed the signals of two different CH-CH2O moieties (see Section 3 and Figures S35, S36 and S64-S67), confirming formation of (hexahydro-1H-spiro[cyclopenta[c]pyrrolo [1,2-b]isoxazole-3,1′-cyclopentane]-2′-yl) methanol system. However, one could not exclude the possibility of C=C bond approaching the plane of the nitrone group from different sides with the formation of cisand trans-isomers relative to the position of the hydroxymethyl group. To assign a structure to the cycloadduct 1 H-1 H NOESY NMR spectra were recorded ( Figure SI67). 1 H-1 H NOESY correlation showed a cross-peak between protons of 6 CH2 and 15 CH2 groups, while no cross-peak was observed between 11 CH2 and 6 CH2 protons ( Figure 3). Apparently, the hydroxymethyl group prevents C=C from approaching the nitrone group, and the formation of a cycloadduct is possible only through the transition state 15 (Scheme 2). Reductive scission of the isoxazolidine ring N-O bond with Zn-AcOH system afforded 14, which was isolated as a colorless crystalline solid. Half of the signal set was  The nitrone 12 easily undergoes intramolecular 1,3-dipolar cycloaddition affording single product. The best yield was achieved upon heating to reflux in toluene in presence of 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO). Intramolecular cycloaddition in 2-pent-4-en-1-ylpyrroline 1-oxides was reported to give a single regioisomer [27,30,33]. Similarly, the 1 H and 13 C NMR spectra and 1 H-1 H and 1 H-13 C correlations revealed the signals of two different CH-CH2O moieties (see Section 3 and Figures S35, S36 and S64-S67), confirming formation of (hexahydro-1H-spiro[cyclopenta[c]pyrrolo [1,2-b]isoxazole-3,1′-cyclopentane]-2′-yl) methanol system. However, one could not exclude the possibility of C=C bond approaching the plane of the nitrone group from different sides with the formation of cisand trans-isomers relative to the position of the hydroxymethyl group. To assign a structure to the cycloadduct 1 H-1 H NOESY NMR spectra were recorded ( Figure SI67). 1 H-1 H NOESY correlation showed a cross-peak between protons of 6 CH2 and 15 CH2 groups, while no cross-peak was observed between 11 CH2 and 6 CH2 protons ( Figure 3). Apparently, the hydroxymethyl group prevents C=C from approaching the nitrone group, and the formation of a cycloadduct is possible only through the transition state 15 (Scheme 2). Reductive scission of the isoxazolidine ring N-O bond with Zn-AcOH system afforded 14, which was isolated as a colorless crystalline solid. Half of the signal set was observed in 1 H and 13 C NMR spectra of 14, indicating symmetrical structure. Single-crystal X-ray diffraction data of this compound showed C2-symmetry of the molecule, which confirms our previous assignment of structure to 13. Thus, despite the absence of bulky substituents at positions 3 and 4 of the pyrroline ring, the intramolecular 1,3-dipolar cycloaddition reaction in 12 proceeds stereospecifically. Reductive scission of the isoxazolidine ring N-O bond with Zn-AcOH system afforded 14, which was isolated as a colorless crystalline solid. Half of the signal set was observed in 1 H and 13 C NMR spectra of 14, indicating symmetrical structure. Single-crystal X-ray diffraction data of this compound showed C 2 -symmetry of the molecule, which confirms our previous assignment of structure to 13. Thus, despite the absence of bulky substituents at positions 3 and 4 of the pyrroline ring, the intramolecular 1,3-dipolar cycloaddition reaction in 12 proceeds stereospecifically.
Previously, we reported that oxidation of 2,2-disubstituted 1-azaspiro[4.4]nonan-6ylmethanols to nitroxides with m-CPBA may be accompanied with conversion of hydroxymethyl group into aldehyde, and acylation of the hydroxy group(s) before oxidation increases the nitroxide yield [30]. Following this procedure, 14 was heated with acetic anhydride (Ac 2 O) to give 16, which was then oxidized with m-CPBA (Scheme 3). The reaction afforded two nitroxides. The main product 17 was isolated as yellow crystals with 86% yield. The structure of 17 was confirmed by single-crystal X-ray diffraction data ( Figure 2).
Previously, we reported that oxidation of 2,2-disubstituted 1-azaspiro[4.4]nonan-6ylmethanols to nitroxides with m-CPBA may be accompanied with conversion of hydroxymethyl group into aldehyde, and acylation of the hydroxy group(s) before oxidation increases the nitroxide yield [30]. Following this procedure, 14 was heated with acetic anhydride (Ac2O) to give 16, which was then oxidized with m-CPBA (Scheme 3). The reaction afforded two nitroxides. The main product 17 was isolated as yellow crystals with 86% yield. The structure of 17 was confirmed by single-crystal X-ray diffraction data ( Figure  2 The structure of the minor product 18 was assigned on the basis of 1 H and 13 C NMR, and 1 H-1 H COSY, 1 H-13 C HSQC, and 1 H-13 C HMBC spectra were acquired after reduction of the nitroxide to corresponding amine with Zn in presence of trifluoroacetic acid according to the earlier described procedure [18] (see Section 3 and Figures S41, S42 and S68-S70) and confirmed with high-resolution mass-spectrum and IR spectral data ( Figure S17). We have previously observed the formation of similar dehydrogenated nitroxide upon oxidation of (2,2-dimethyl-1-azaspiro[4.4]nonan-6-yl)methyl acetate, and proposed a mechanism implying proton abstraction from intermediate oxoammonium cation [30].
Nitroxide 17 was then subjected to alkaline hydrolysis to give 5 with nearly quantitative yield. The structure of 5 was confirmed by single-crystal X-ray diffraction data (Figure 4). The overall yield of this nitroxide starting from commercially available 4-chlorobutyryl chloride and 5-bromo-1-pentene exceeds 40%, which makes it an attractive material for the synthesis of spin labels.
Modification of the hydroxymethyl groups seems to be the simplest way to functional derivatives capable of binding to biomolecules. In our recent study, we demonstrated that activation of the hydroxyl group to nucleophilic substitution in 1-unsubstituted pyrrolidines and corresponding alkoxyamines, acyloxyamines, or nitroxides with spiro-(2-hydroxymethyl)cyclopentane moiety always leads to cyclization, which may be followed by rearrangement [34]. Here, we studied the oxidation of hydroxymethyl groups to carboxylic ones, and the alkylation/acylation of hydroxyl groups. The structure of the minor product 18 was assigned on the basis of 1 H and 13 C NMR, and 1 H-1 H COSY, 1 H-13 C HSQC, and 1 H-13 C HMBC spectra were acquired after reduction of the nitroxide to corresponding amine with Zn in presence of trifluoroacetic acid according to the earlier described procedure [18] (see Section 3 and Figures S41, S42 and S68-S70) and confirmed with high-resolution mass-spectrum and IR spectral data ( Figure S17). We have previously observed the formation of similar dehydrogenated nitroxide upon oxidation of (2,2-dimethyl-1-azaspiro[4.4]nonan-6-yl)methyl acetate, and proposed a mechanism implying proton abstraction from intermediate oxoammonium cation [30].
Nitroxide 17 was then subjected to alkaline hydrolysis to give 5 with nearly quantitative yield. The structure of 5 was confirmed by single-crystal X-ray diffraction data ( Figure 4). The overall yield of this nitroxide starting from commercially available 4chlorobutyryl chloride and 5-bromo-1-pentene exceeds 40%, which makes it an attractive material for the synthesis of spin labels.
yield. The structure of 17 was confirmed by single-crystal X-ray diffraction data ( Figure  2 The structure of the minor product 18 was assigned on the basis of 1 H and 13 C NMR, and 1 H-1 H COSY, 1 H-13 C HSQC, and 1 H-13 C HMBC spectra were acquired after reduction of the nitroxide to corresponding amine with Zn in presence of trifluoroacetic acid according to the earlier described procedure [18] (see Section 3 and Figures S41, S42 and S68-S70) and confirmed with high-resolution mass-spectrum and IR spectral data ( Figure S17). We have previously observed the formation of similar dehydrogenated nitroxide upon oxidation of (2,2-dimethyl-1-azaspiro[4.4]nonan-6-yl)methyl acetate, and proposed a mechanism implying proton abstraction from intermediate oxoammonium cation [30].
Nitroxide 17 was then subjected to alkaline hydrolysis to give 5 with nearly quantitative yield. The structure of 5 was confirmed by single-crystal X-ray diffraction data (Figure 4). The overall yield of this nitroxide starting from commercially available 4-chlorobutyryl chloride and 5-bromo-1-pentene exceeds 40%, which makes it an attractive material for the synthesis of spin labels.
Modification of the hydroxymethyl groups seems to be the simplest way to functional derivatives capable of binding to biomolecules. In our recent study, we demonstrated that activation of the hydroxyl group to nucleophilic substitution in 1-unsubstituted pyrrolidines and corresponding alkoxyamines, acyloxyamines, or nitroxides with spiro-(2-hydroxymethyl)cyclopentane moiety always leads to cyclization, which may be followed by rearrangement [34]. Here, we studied the oxidation of hydroxymethyl groups to carboxylic ones, and the alkylation/acylation of hydroxyl groups. Modification of the hydroxymethyl groups seems to be the simplest way to functional derivatives capable of binding to biomolecules. In our recent study, we demonstrated that activation of the hydroxyl group to nucleophilic substitution in 1-unsubstituted pyrrolidines and corresponding alkoxyamines, acyloxyamines, or nitroxides with spiro-(2hydroxymethyl)cyclopentane moiety always leads to cyclization, which may be followed by rearrangement [34]. Here, we studied the oxidation of hydroxymethyl groups to carboxylic ones, and the alkylation/acylation of hydroxyl groups.
Attempts to oxidize the hydroxymethyl groups in 5 using TEMPO-Sodium chlorite system [35] were unsuccessful, leading to strong tarring. This may occur due to the formation of an oxoammonium cation, which can cause oxidative transformations in the side chains (cf. [30] and above). Therefore, we used the diamagnetic precursor 14 to prepare the desired nitroxide with carboxylate groups. Direct oxidation of 14 with Jones reagent afforded crude 20 as a colorless viscous oil (Scheme 4). The NMR spectra showed half of the signal set, showing that the compound remained a racemic mixture (no other diastereomers formed). The pure crystalline sample was isolated via column chromatography and crystallization from methanol-ethyl acetate mixture 50:1, and the structure was confirmed by single-crystal X-ray diffraction data ( Figure 4). Attempts to oxidize the hydroxymethyl groups in 5 using TEMPO-Sodium chlorite system [35] were unsuccessful, leading to strong tarring. This may occur due to the formation of an oxoammonium cation, which can cause oxidative transformations in the side chains (cf. [30] and above). Therefore, we used the diamagnetic precursor 14 to prepare the desired nitroxide with carboxylate groups. Direct oxidation of 14 with Jones reagent afforded crude 20 as a colorless viscous oil (Scheme 4). The NMR spectra showed half of the signal set, showing that the compound remained a racemic mixture (no other diastereomers formed). The pure crystalline sample was isolated via column chromatography and crystallization from methanol-ethyl acetate mixture 50:1, and the structure was confirmed by single-crystal X-ray diffraction data ( Figure 4).

Scheme 4. Synthesis of 22.
Attempts to oxidize of 20 either with m-CPBA or H2O2/Na2WO4 were not successful, so the crude 20 was dissolved in methanol saturated with HCl and diester 21 was isolated. Oxidation of 21 with m-CPBA afforded 22, which was isolated as a yellow crystalline solid. The 1 H NMR spectrum acquired after reduction of freshly prepared nitroxide with Zn/CF3COOH showed half of the signal set with significant downfield shift (due to protonation) as compared to spectrum of 21. The structure of 22 was confirmed by singlecrystal X-ray diffraction data ( Figure 4). Alkaline hydrolysis of 22 gave an inseparable mixture of structurally related compounds with total yield 42% (Scheme 5). The element analysis of the mixture corresponded to the formula C14H20NO5, which allowed us to assume that the product was a mixture of isomers. The 13 C NMR spectrum of the mixture after reduction with Zn/CF3COOH corresponded to a mixture of three isomers, two symmetric and one asymmetric dicarboxylic acids ( Figure SI51). This picture corresponds to a mixture of three isomers, two symmetric, and one asymmetric dicarboxylic acids. The ratio of the isomers was estimated using integrals of the signals of methine hydrogens at 3.02-3.30 ppm in 1 H NMR spectrum. Inversion of the asymmetric center adjacent to the ester group may result from C-H acidity. TLC analysis of samples of 22 after long-term storage showed the emergence of two compounds with close Rf, probably the isomers. The formation of these compounds accelerates in the presence of a base or LiI; however, these reactions were accompanied by tarring. Presumably, in alkaline solution, isomerization proceeds faster than hydrolysis, resulting in nearly statistical ratio of isomers. Attempts to oxidize of 20 either with m-CPBA or H 2 O 2 /Na 2 WO 4 were not successful, so the crude 20 was dissolved in methanol saturated with HCl and diester 21 was isolated. Oxidation of 21 with m-CPBA afforded 22, which was isolated as a yellow crystalline solid. The 1 H NMR spectrum acquired after reduction of freshly prepared nitroxide with Zn/CF 3 COOH showed half of the signal set with significant downfield shift (due to protonation) as compared to spectrum of 21. The structure of 22 was confirmed by singlecrystal X-ray diffraction data (Figure 4). Alkaline hydrolysis of 22 gave an inseparable mixture of structurally related compounds with total yield 42% (Scheme 5). The element analysis of the mixture corresponded to the formula C 14 H 20 NO 5 , which allowed us to assume that the product was a mixture of isomers. The 13 C NMR spectrum of the mixture after reduction with Zn/CF 3 COOH corresponded to a mixture of three isomers, two symmetric and one asymmetric dicarboxylic acids ( Figure S51). This picture corresponds to a mixture of three isomers, two symmetric, and one asymmetric dicarboxylic acids. The ratio of the isomers was estimated using integrals of the signals of methine hydrogens at 3.02-3.30 ppm in 1 H NMR spectrum. Attempts to oxidize the hydroxymethyl groups in 5 using TEMPO-Sodium chlorite system [35] were unsuccessful, leading to strong tarring. This may occur due to the formation of an oxoammonium cation, which can cause oxidative transformations in the side chains (cf. [30] and above). Therefore, we used the diamagnetic precursor 14 to prepare the desired nitroxide with carboxylate groups. Direct oxidation of 14 with Jones reagent afforded crude 20 as a colorless viscous oil (Scheme 4). The NMR spectra showed half of the signal set, showing that the compound remained a racemic mixture (no other diastereomers formed). The pure crystalline sample was isolated via column chromatography and crystallization from methanol-ethyl acetate mixture 50:1, and the structure was confirmed by single-crystal X-ray diffraction data (Figure 4). Attempts to oxidize of 20 either with m-CPBA or H2O2/Na2WO4 were not successful, so the crude 20 was dissolved in methanol saturated with HCl and diester 21 was isolated. Oxidation of 21 with m-CPBA afforded 22, which was isolated as a yellow crystalline solid. The 1 H NMR spectrum acquired after reduction of freshly prepared nitroxide with Zn/CF3COOH showed half of the signal set with significant downfield shift (due to protonation) as compared to spectrum of 21. The structure of 22 was confirmed by singlecrystal X-ray diffraction data (Figure 4). Alkaline hydrolysis of 22 gave an inseparable mixture of structurally related compounds with total yield 42% (Scheme 5). The element analysis of the mixture corresponded to the formula C14H20NO5, which allowed us to assume that the product was a mixture of isomers. The 13 C NMR spectrum of the mixture after reduction with Zn/CF3COOH corresponded to a mixture of three isomers, two symmetric and one asymmetric dicarboxylic acids ( Figure SI51). This picture corresponds to a mixture of three isomers, two symmetric, and one asymmetric dicarboxylic acids. The ratio of the isomers was estimated using integrals of the signals of methine hydrogens at 3.02-3.30 ppm in 1 H NMR spectrum. Inversion of the asymmetric center adjacent to the ester group may result from C-H acidity. TLC analysis of samples of 22 after long-term storage showed the emergence of two compounds with close Rf, probably the isomers. The formation of these compounds accelerates in the presence of a base or LiI; however, these reactions were accompanied by tarring. Presumably, in alkaline solution, isomerization proceeds faster than hydrolysis, resulting in nearly statistical ratio of isomers. Inversion of the asymmetric center adjacent to the ester group may result from C-H acidity. TLC analysis of samples of 22 after long-term storage showed the emergence of two compounds with close R f , probably the isomers. The formation of these compounds accelerates in the presence of a base or LiI; however, these reactions were accompanied by tarring. Presumably, in alkaline solution, isomerization proceeds faster than hydrolysis, resulting in nearly statistical ratio of isomers.
It was shown that nitroxide group can decrease pK a of the acidic center in two σ-bond distance by 2.5 orders of magnitude compared to corresponding methoxyamine derivative [36]. Thus, reduction of the nitroxide group may slow down the isomerization. To verify this hypothesis, the freshly prepared nitroxide 22 was reduced to corresponding hydroxylamine with ascorbic acid in oxygen-free conditions, and then, potassium hydroxide was added (Scheme 6). After the hydrolysis was complete, the products were oxidized with air oxygen, acidified, and extracted. Dicarboxylic acid 23a was a major component of the resulting mixture and it was isolated with the yield 45%. The structure of 23a was confirmed by single-crystal X-ray diffraction data ( Figure 5).
It was shown that nitroxide group can decrease pKa of the acidic center in two σ-bond distance by 2.5 orders of magnitude compared to corresponding methoxyamine derivative [36]. Thus, reduction of the nitroxide group may slow down the isomerization. To verify this hypothesis, the freshly prepared nitroxide 22 was reduced to corresponding hydroxylamine with ascorbic acid in oxygen-free conditions, and then, potassium hydroxide was added (Scheme 6). After the hydrolysis was complete, the products were oxidized with air oxygen, acidified, and extracted. Dicarboxylic acid 23a was a major component of the resulting mixture and it was isolated with the yield 45%. The structure of 23a was confirmed by single-crystal X-ray diffraction data ( Figure 5).  Inversion of the asymmetric center at the ester group in 22 leads to a loss of configuration that provides higher hindrance to the nitroxide group. Despite succeeding in isolating pure 23a, it is obvious that activated esters capable of binding to biomolecules, which could be prepared from 23a, will behave similarly to 22. Thus, the nitroxides with spiro-(2-carboxy)cyclopentane moieties are not optimal for spin labeling.
Exploring the possible ways to functional derivatives of 5, we tried several alkylation and acylation reactions. A reaction of nitroxide alcohols with carbonyldiimidazole (CDI) was used for binding to primary amino groups [37,38]. A reaction of 5 with CDI afforded the nitroxide 26 with an excellent yield (Scheme 7). The structure of 26 was confirmed by single-crystal X-ray diffraction data ( Figure 5). A reaction of 26 with N,N-dimethyl-1,3diaminopropane gave 27 with 55% yield.
Another carbamate derivative 28 was prepared via treatment of 5 with methyl 3-isocyanatopropionate with quantitative yield. After alkaline hydrolysis of the ester groups corresponding dicarboxylic acid 29 was isolated. The structure of the nitroxides 27, 28, and 29 were confirmed by element analyses, and IR spectra data and 1 H and 13 C NMR spectra were acquired after reduction with Zn/CF3COOH, which showed half of the signal set (See Supporting Information). It was shown that nitroxide group can decrease pKa of the acidic center in two σ-bond distance by 2.5 orders of magnitude compared to corresponding methoxyamine derivative [36]. Thus, reduction of the nitroxide group may slow down the isomerization. To verify this hypothesis, the freshly prepared nitroxide 22 was reduced to corresponding hydroxylamine with ascorbic acid in oxygen-free conditions, and then, potassium hydroxide was added (Scheme 6). After the hydrolysis was complete, the products were oxidized with air oxygen, acidified, and extracted. Dicarboxylic acid 23a was a major component of the resulting mixture and it was isolated with the yield 45%. The structure of 23a was confirmed by single-crystal X-ray diffraction data ( Figure 5).  Inversion of the asymmetric center at the ester group in 22 leads to a loss of configuration that provides higher hindrance to the nitroxide group. Despite succeeding in isolating pure 23a, it is obvious that activated esters capable of binding to biomolecules, which could be prepared from 23a, will behave similarly to 22. Thus, the nitroxides with spiro-(2-carboxy)cyclopentane moieties are not optimal for spin labeling.
Exploring the possible ways to functional derivatives of 5, we tried several alkylation and acylation reactions. A reaction of nitroxide alcohols with carbonyldiimidazole (CDI) was used for binding to primary amino groups [37,38]. A reaction of 5 with CDI afforded the nitroxide 26 with an excellent yield (Scheme 7). The structure of 26 was confirmed by single-crystal X-ray diffraction data ( Figure 5). A reaction of 26 with N,N-dimethyl-1,3diaminopropane gave 27 with 55% yield.
Another carbamate derivative 28 was prepared via treatment of 5 with methyl 3-isocyanatopropionate with quantitative yield. After alkaline hydrolysis of the ester groups corresponding dicarboxylic acid 29 was isolated. The structure of the nitroxides 27, 28, and 29 were confirmed by element analyses, and IR spectra data and 1 H and 13 C NMR spectra were acquired after reduction with Zn/CF3COOH, which showed half of the signal set (See Supporting Information). Inversion of the asymmetric center at the ester group in 22 leads to a loss of configuration that provides higher hindrance to the nitroxide group. Despite succeeding in isolating pure 23a, it is obvious that activated esters capable of binding to biomolecules, which could be prepared from 23a, will behave similarly to 22. Thus, the nitroxides with spiro-(2-carboxy)cyclopentane moieties are not optimal for spin labeling.
Exploring the possible ways to functional derivatives of 5, we tried several alkylation and acylation reactions. A reaction of nitroxide alcohols with carbonyldiimidazole (CDI) was used for binding to primary amino groups [37,38]. A reaction of 5 with CDI afforded the nitroxide 26 with an excellent yield (Scheme 7). The structure of 26 was confirmed by single-crystal X-ray diffraction data ( Figure 5). A reaction of 26 with N,N-dimethyl-1,3diaminopropane gave 27 with 55% yield.
Another carbamate derivative 28 was prepared via treatment of 5 with methyl 3isocyanatopropionate with quantitative yield. After alkaline hydrolysis of the ester groups corresponding dicarboxylic acid 29 was isolated. The structure of the nitroxides 27, 28, and 29 were confirmed by element analyses, and IR spectra data and 1 H and 13 C NMR spectra were acquired after reduction with Zn/CF 3 COOH, which showed half of the signal set (See Supporting Information).
Recently, copper-catalyzed azide-alkyne cycloaddition (CuAAC) was successfully used for the site-directed spin labeling of the protein in vivo [39]. The spin labels capable of binding azides were prepared via alkylation of 5 with propargyl bromide (Scheme 8). Two nitroxides 30 and 31 were isolated from the reaction mixture. The proposed structures were confirmed with IR spectra, HRMS, and element analysis data. The ability of the spin label 31 to bind to azide-containing biomolecules was demonstrated by the reaction with 2,3,4,6-tetra-Oacetyl-β-D-galactopyranosyl azide in analogy to the literature protocols [40][41][42]. The reaction of enantiomerically pure galactose derivative with racemic mixture 31 expectedly gave a mixture of diastereomers 32, which was not separated. The structure of 32 was confirmed by element analysis, and IR spectral data and 1 H and 13 C NMR spectra were acquired after reduction with Zn/CF 3 COOH (See SI). The NMR spectra of two diastereomers are very close: only few signals of carbon atoms of the spirocyclic system differ and it is impossible to assign them to a specific diastereomer. Ammonolysis of 32, in analogy with the literature procedure [42], afforded spin-labelled galactose 33, which was isolated as a glassy solid. To acquire 1 H and 13 C NMR spectra, the nitroxide sample was reduced with Zn in presence of formic and oxalic acids mixture. Similarly to the above-mentioned for 32, most of the signals of the two diastereomers were not resolved, and very few slightly differed in chemical shift (see Section 3 and Figures S60 and S61). Recently, copper-catalyzed azide-alkyne cycloaddition (CuAAC) was successfully used for the site-directed spin labeling of the protein in vivo [39]. The spin labels capable of binding azides were prepared via alkylation of 5 with propargyl bromide (Scheme 8). Two nitroxides 30 and 31 were isolated from the reaction mixture. The proposed structures were confirmed with IR spectra, HRMS, and element analysis data. The ability of the spin label 31 to bind to azide-containing biomolecules was demonstrated by the reaction with 2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl azide in analogy to the literature protocols [40][41][42]. The reaction of enantiomerically pure galactose derivative with racemic mixture 31 expectedly gave a mixture of diastereomers 32, which was not separated. The structure of 32 was confirmed by element analysis, and IR spectral data and 1 H and 13 C NMR spectra were acquired after reduction with Zn/CF3COOH (See SI). The NMR spectra of two diastereomers are very close: only few signals of carbon atoms of the spirocyclic system differ and it is impossible to assign them to a specific diastereomer. Ammonolysis of 32, in analogy with the literature procedure [42], afforded spin-labelled galactose 33, which was isolated as a glassy solid. To acquire 1 H and 13 C NMR spectra, the nitroxide sample was reduced with Zn in presence of formic and oxalic acids mixture. Similarly to the above-mentioned for 32, most of the signals of the two diastereomers were not resolved, and very few slightly differed in chemical shift (see Section 3 and Figures S60 and  S61).

Electron Spin Relaxation Time
Electron spin relaxation times were measured for four radicals 5, 33, 34, and 35 at two temperatures of 80 K and 120 K in water−glycerol solution (1:1); the data are shown on Figure 7 and are listed in Table 2. In Figure 7a, the non-exponential decay of spin echo for both radicals is clearly visible. A similar effect was described by Eaton and coworkers [22,24,26] for the phase memory time of the spin echo of nitroxyl radicals with spirocyclohexyl moieties-the positions 2 and 6 of the piperidine ring. It was shown that the observed spin echo decay curves in a water-glycerole solution (1:1) are described by the following formula: I = I 0 × exp(−(t/T 2 ) n , where n > 1, and are determined by the nuclear spin diffusion of solvent protons [22]. This effect has been studied in detail by G. Jeschke et al. in recent years [44,45]. An increase in the contribution to the spin−spin relaxation mechanism of various dynamic processes, such as the rotation of methyl groups, leads to a decrease in the value of n to 1. As expected, spirocyclic nitroxides 5 and 33 showed a less pronounced dependence of relaxation time T m on temperature than radicals with methyl substituents 34 and 35. An increase in temperature from 80 to 120 K did not lead to a change in the time T m for 33 within the measurement accuracy, while for radical 35 it decreased more than 3-fold from 3.4 to 1.1 µs. This effect is caused by the manifestation of an additional mechanism of electron spin relaxation due to faster rotation of methyl groups with increasing temperature. It can be seen from Figure 7b that, already at 120 K, an exponential decay of spin echo signal is observed for 35 (n ≈ 2.1), whereas for 33 there is no additional contribution to relaxation and n ≈ 2 is retained due to the absence of methyl substituents. Table 1. Values of the line width (Gaussian shape, ω g , and Lorentzian shape, ω l ,); nitrogen hfc constants, a N , in water and toluene, mT.

Electron Spin Relaxation Time
Electron spin relaxation times were measured for four radicals 5, 33, 34, and 35 at two temperatures of 80 K and 120 K in water−glycerol solution (1:1); the data are shown on Figure 7 and are listed in Table 2. In Figure 7a, the non-exponential decay of spin echo for

Electron Spin Relaxation Time
Electron spin relaxation times were measured for four radicals 5, 33, 34, and 35 at two temperatures of 80 K and 120 K in water−glycerol solution (1:1); the data are shown on Figure 7 and are listed in Table 2. In Figure 7a, the non-exponential decay of spin echo for

Electron Spin Relaxation Time
Electron spin relaxation times were measured for four radicals 5, 33, 34, and 35 at two temperatures of 80 K and 120 K in water−glycerol solution (1:1); the data are shown on Figure 7 and are listed in Table 2. In Figure 7a, the non-exponential decay of spin echo for

Electron Spin Relaxation Time
Electron spin relaxation times were measured for four radicals 5, 33, 34, and 35 at two temperatures of 80 K and 120 K in water−glycerol solution (1:1); the data are shown on Figure 7 and are listed in Table 2. In Figure 7a, the non-exponential decay of spin echo for  Curiously, binding to galactose does not cause large changes in T 1 , but in all cases leads to a significant increase in T m . This effect may be associated with another mechanism of electron spin relaxation due to modulation of hfi and g-tensor anisotropy by libration motion of nitroxide molecule [46,47]. The influence of the solvent matrix to the libration motion of nitroxide ring was discussed previously and it was shown that the attachment of long chains to nitroxide ring can affect the libration motion. change in the time Tm for 33 within the measurement accuracy, while for radical 35 it decreased more than 3-fold from 3.4 to 1.1 µs. This effect is caused by the manifestation of an additional mechanism of electron spin relaxation due to faster rotation of methyl groups with increasing temperature. It can be seen from Figure 7b that, already at 120 K, an exponential decay of spin echo signal is observed for 35 (n ≈ 2.1), whereas for 33 there is no additional contribution to relaxation and n ≈ 2 is retained due to the absence of methyl substituents.  Table  2.
Curiously, binding to galactose does not cause large changes in T1, but in all cases leads to a significant increase in Tm. This effect may be associated with another mechanism of electron spin relaxation due to modulation of hfi and g-tensor anisotropy by libration motion of nitroxide molecule [46,47]. The influence of the solvent matrix to the libration motion of nitroxide ring was discussed previously and it was shown that the attachment of long chains to nitroxide ring can affect the libration motion.  Table 2.

Reduction Rate Constants
The kinetics of reduction of nitroxides 5, 29, and 33 with ascorbate was studied. The reaction was carried out under argon using large excess of ascorbate in presence of glutathione to suppress possible reverse processes [48] and followed by EPR. The secondorder rate constants were calculated from exponential decay of the nitroxide signal (see Section 3 and Figure S71 and S72). The resulting values (k red , M −1 s −1 ) for 5, 29, and 33 were as follows: (3.2 ± 0.2) × 10 −3 , (1.40 ± 0.06) × 10 −3 and (2.70 ± 0.06) × 10 −3 , respectively. All the new radicals demonstrate higher resistance to reduction as compared to 1 (Figure 1). Interestingly, removal of electron-withdrawing tert-butoxy groups produces only minor effect on the reduction rate (cf. k red for 1 and 5). Presumably, the electron effect of t-BuO-groups is compensated by their influence on the conformation of spirocyclic fragments in analogy to [25]. An increase in steric demand of the substituents adjacent to nitroxide group is known to stabilize the radicals against reduction, enlargement of all neighboring substituents being most efficient [49]. Comparison of the reduction rates for 5, 33, and 29 shows similar effect, with 29 being the most resistant to reduction. Presumably, an increase in substituent size in the positions 1 and 8 of the 6-azadispiro[4.1.4.2]tridecane-6-oxyl system stabilize the nitroxide due to proximity of the bulky groups to the radical center. One can expect much higher stabilization effect upon binding of similar nitroxides to large biomolecules.

General
The IR spectra were recorded on a Bruker Vector 22 FT-IR spectrometer (Bruker, Billerica, MA, USA) in KBr pellets (1:150 ratio) or in neat samples (see the Supplementary Information in this article pp. 6-17, Figures S1-S23) and are reported in wave numbers (cm −1 ). UV spectra were acquired on a HP Agilent 8453 spectrometer (Agilent Technologies, Santa Clara, CA, USA) in ethanol solutions (concentration~10 −4 ) (see the Supplementary Information in this article pp. 18-21, Figures S24-S30). 1 H NMR spectra were recorded on a Bruker AV 300 (300.132 MHz), AV 400(400.134 MHz), DRX 500 (500.130 MHz), and Bruker AV 600 (600.300 MHz) spectrometers (Bruker, Billerica, MA, USA). 13 C NMR spectra were recorded on a Bruker AV 300 (75.467 MHz), AV 400 (100.614 MHz), DRX 500 (125.758 MHz), and Bruker AV 600 (150.945 MHz) spectrometers (see the Supplementary Information in this article pp. 19-42, Figures S31-S70). All the NMR spectra were acquired for 5-10% solutions in CDCl 3 or CD 3 OD at 300 K using the signal of the solvent as a standard. NMR spectra of nitroxides for analysis and structure assignment were recorded after reduction with Zn in CD 3 OD-CF 3 COOH (or CD 3 OD-HCOOH-HOOCCOOH mixture) at 65 • C as described in [19] or with Zn and ND 4 Cl in CD 3 OD at 5 • C. Atoms numbering are shown in figures placed on spectra in SI. HRMS analyses were performed using a High-Resolution Mass Spectrometer DFS (Thermo Electron, Waltham, MA, USA).
The X-ray diffraction experiments were carried out on a Bruker KAPPA APEX II diffractometer (graphite-monochromated Mo Kα radiation). Reflection intensities were corrected for absorption by SADABS2016/2 program [50] except of 14, 17, 20 treated by SADABS2008/1 version [50]. The structures were solved by direct methods using the SHELXS-97 (Sheldrick, 2008) program [51] of 14, 17, 20, and SHELXT 2014/5 (Sheldrick, 2014) [52] for the rest ones. All compounds were refined by anisotropic (isotropic for all H atoms) full-matrix least-squares method against F 2 of all reflections by SHELXL2018/3 [53]. The positions of the hydrogen atoms were calculated geometrically and refined in riding model except of hydrogenes in OH and NH-groups of 14 and 23a localized from difference map and refined independently with restriction of bond lengths. The presence of two hydrogen atoms on N1 in 20 was proved from the electron density difference map. The asymmetric units of 17, 22, and 26 include a half of molecule, the same one of 5 and 14 consists of three and two molecules, respectively. Note that cyclopentane and azalidine cycles in 22 and 23 are statistically disordered due to conformational mobility in approximate ratio 3:1 and 7:1, respectively. The imidazole cycles of 26 are also disordered due to rotational mobility in approximate ratio 1:1. For experimental details see Table S1.

EPR
CW EPR spectra were recorded at X-band frequencies (~9.4 GHz) on a commercial Bruker spectrometer, Elexsys E 540 (Bruker Corporation, Billerica, MA, USA). Electron spin resonance spectra were recorded with the following settings: frequency, 9. Electron spin relaxation times were measured using home-made pulse EPR spectrometers equipped with a flow helium cryostat and temperature control system [54]. T m was measured using a two-pulse electron spin echo (ESE) sequence; T 1 was measured by an inversion recovery technique with inversion π-pulse and detection of a two-pulse ESE sequence. The π-pulse length was 40 ns.

Kinetic Measurements and Partition Coefficients
For kinetic measurements, stock solutions were prepared in phosphate-citrate-borate buffer (0.5 mM of each): (1) 1 M solution of ascorbic acid and (2) 5 mM solution of glutathione (GSH). Nitroxides were dissolved in the same buffer and diluted to a concentration of 0.2-0.3 mM. The pH was adjusted to 7.5 with NaOH and the solutions were deoxygenated via bubbling with argon. The solutions carefully and quickly mixed in appropriate proportions in a small tube and placed into EPR capillary (50 µL). Oxygen-free conditions were kept permanently. Capillaries were sealed and placed into EPR resonator. EPR experiments were performed on CW EPR X-band spectrometer Bruker ER-200D (9.87 GHz). Spectra were recorded in oxygen-free conditions using the following settings: microwave power 5 mW, modulations amplitude 0.1 or 0.2 mT; time constant 100 ms; conversion time 50.12 ms. The decay of amplitude of low field component of the EPR spectrum was followed in kinetics measurements. Kinetics of decay were fitted with monoexponential function to calculate the first order rate constants. The kinetics measurements were performed at ascorbate concentrations of 100, 200, and 300 mM. The calculated first reaction constants were plotted versus ascorbate concentration; data were fitted with linear dependence. The slope corresponds to second order reaction (See Figures S71 and S72).
For partition coefficients measurements nitroxides were dissolved in 1 mL of distilled water in plastic tube. At this step, the EPR signal was measured as a zero point. Then, the fraction of octanol was added to this solution and shacked for a few minutes to achieve the stationary distribution of radical in the mixture. The tube was shortly centrifuged to separate the octanol and water fractions. The aliquot of radical solution (water phase) was carefully taken with capillary from the bottom of the tube and new EPR spectrum was recorded with the same settings. This procedure was repeated three times for three different octanol contents. The inverse normalized decrease in the EPR signal was plotted versus octanol/water ratio and the slope of the linear fit was used to determine the partition coefficient (see Figure S73).

Conclusions
In this work, we described a new family of dispirocyclic nitroxides that show very attractive properties for SDSL/EPR studies in biological media and in cells. The unique feature of these radicals is the combination of high resistance to reduction with improved spin relaxation characteristics typical of nitroxides with two spirocyclic moieties adjacent to N-O • group. These nitroxides can be prepared from commercially available chemicals using a set of simple procedures with an overall yield over 40%. The resulting dispirocyclic core contains two hydroxymethyl groups near the radical center, which opens up the possibility of synthesizing mono-and bifunctional spin labels. It is important that binding to large biomolecules may further increase the resistance to reduction due to proximity of the functional groups to nitroxide moiety.