Tuning the properties of nitroxide spin labels for use in electron paramagnetic resonance spectroscopy through chemical modification of the nitroxide framework

Marius M. Haugland, Edward A. Anderson and Janet E. Lovett

DOI: 10.1039/9781782629436-00001

Spin labels containing nitroxyl radicals possess many properties that render them useful for electron paramagnetic resonance (EPR) spectroscopy. This review describes the relationships between the structure and properties of nitroxide spin labels, methods for their synthesis, advances in methods for their incorporation into biomolecules, and selected examples of applications in biomolecule structural investigations.

1 Introduction

Within the field of electron paramagnetic resonance (EPR) spectroscopy, 'spin labelling' describes the attachment of a radical or paramagnetic centre (i.e. a molecule containing at least one unpaired electron spin) onto a material of interest, which enables its investigation using paramagnetic resonance spectroscopy. For such applications, spin labels should ideally fulfil several criteria: the framework of the label must stabilise the radical against redox processes; the radical must possess desirable properties for the magnetic resonance experiment (such as chemical stability and spin coherence persistence); and, the label must be readily (and site-specifically) attached without structural distortion of the system under study.

By far the largest family of spin labels are those based on the nitroxyl (N-O•) radical, which are called nitroxide spin labels. These are typically five- or six-membered heterocyclic derivatives of piperidine, pyrrolidine, isoindoline, and other heterocycles containing two heteroatoms; importantly, the nitroxyl radical is flanked by two quaternary carbon atoms. The 'classic' nitroxide is the piperidine-based 2,2,6,6-tetramethylpiperidine 1-oxyl (TEMPO, 1, Fig. 1), which has found use in many chemical and materials applications. This radical, in which the unpaired electron is located mainly on the nitrogen and oxygen atoms, is stabilised by the steric screening imparted by its four adjacent methyl groups, which protect the radical from reduction or other processes. The lack of a-protons also prevents the decomposition of the nitroxyl to the corresponding nitrone. Some other examples of common nitroxide families (2-7) are illustrated.

A 'spin label' can be defined as a derivative of the parent nitroxide in which the core ring system (or its substituents) is modified to enable its incorporation into a larger framework, and thus to be used as a probe. The framework of interest may be a polymer, a surface, or a biomolecule such as a protein, nucleic acid, sugar or lipid.

Spin labels are commonly used to measure interspin distances (i.e. the distance between two stable free radicals) using continuous wave (CW) or pulsed techniques, which for nitroxides have an effective range of 0.5 to >10 nm. They can also be used to probe the local environment of the label, such as its accessibility and dynamic mobility. Nitroxides are also employed as paramagnetic relaxation enhancers in nuclear magnetic resonance (NMR) spectroscopy, and as polarisation/contrast agents in dynamic nuclear polarisation (DNP) or magnetic resonance imaging (MRI) experiments.

Modification of the basic structure of the nitroxide can lead to dramatic changes in the properties of the spin label, and it is for this reason that a myriad of spin labels have been designed. Essential considerations centre on the structure of the nitroxide around the nitroxyl radical itself, and the functionality used to enable spin labelling. This chapter discusses these aspects, along with recent advances in the synthesis and applications of nitroxide spin labels in EPR spectroscopy.

2 The nitroxide spin label as a probe in EPR spectroscopy

2.1 Magnetic properties

The Zeeman splitting for the nitroxide spin labels is anisotropic and typically gxx > gyy > gzz with gxx and gyy close in value and greater than the free electron g-value. The gzz axis is roughly coincident with the p orbital, approximated as a linear combination of the 2p orbitals of oxygen and nitrogen. Therefore, in planar systems such as pyrrolinoxyl spin labels, the gzz is perpendicular to the plane of the ring. The x-axis is taken as coincident with the NO bond.

The unpaired electron is considered to reside in the p orbital. The spin density is on the nitrogen and oxygen with almost no delocalisation over the rest of the framework (for adjacent alkyl groups). The hyperfine coupling constant, Aiso, for nitroxide spin labels is typically in the region of 40 to 47 MHz. Due to the relative spin localisation the hyperfine splitting of the Zeeman levels is dominated by the nitrogen of the NO group (16O nuclei have zero spin and the predominant isotope of nitrogen is 14N with a nuclear spin, I, of 1). The hyperfine axes approximately follow the g-tensor principal axes with Axx˜Ayy where Azz is typically about 100 MHz. Both g and Aiso are weakly sensitive to solvent polarity and proticity, with Aiso increasing and g-values decreasing in increasingly polar/protic solvents. This property has been used to map membrane protein channels, and to probe changes in solvent behaviour associated with the glass-transition temperature in water/ glycerol mixtures.

Coupling of proximal nuclei (I ? 0) to the electron gives rise to the characteristic appearance of an EPR spectrum. Usually, the greatest splitting is caused by coupling to the nitroxyl nitrogen atom, with much smaller splitting from other ring substituents. However, this super-hyperfine splitting can afford additional information, such as the extent of protonation of imidazolinyl and imidazolidinyl spin labels. The nature of the substituents flanking the nitroxyl radical can also significantly affect signal linewidth. Substituent effects on the ring conformation can be influential: the faster dynamic averaging of the hyperfine interaction of the nitroxyl with the methyl protons in 4-oxo-TEMPO (8, Fig. 2) compared to 4-hydroxy-TEMPO (9) can be explained by a higher barrier to conformational ring flip in the latter, where the ring framework is fully sp3-hybridised (aside from the nitroxyl). Isotopic labelling such as perdeuteration or 15N-substitution, can also lead to line narrowing, and therefore improve the sensitivity of the spin label. This property has been used to improve the precision of measurement of tumbling rates, and in oximetry.

The positioning of certain spin-active nuclei directly adjacent to the nitroxyl can cause more significant splitting. For example, the phosphorus nucleus in PROXYL spin label 10 (Fig. 2) has I = 1/2, with the largest hyperfine coupling at 140 MHz. The resultant two-line spectrum is then split by the nitrogen atom to give a six-line pattern. It was shown that this spin label is a sensitive probe of dynamics, which suggests that it may be possible to simultaneously label a molecule of interest at two sites, with this label and a standard nitroxide, to enable simultaneous but distinguishable measurements.

The anisotropy in the g and A tensors are such that when the spin labels can rotate rapidly, as might be the case in low-viscosity organic solvents at room temperature, the CW EPR spectrum is motionally narrowed and reveals only giso and Aiso. However, if the tumbling dynamics of the system are slowed, through increasing viscosity, decreasing temperature, or tethering to a larger, more slowly diffusing molecule, the measured line shape of the nitroxide alters (Fig. 3). At X-band microwave frequencies the changes in the lineshape are sensitive to rotational correlation times (tc) in the ns region, with particular sensitivity to changes in dynamics when tc ~1 ns. Altering the microwave frequency will alter the timescales measured and higher frequencies will offer enhanced angular resolution. There are many reports of using this property to map global and local dynamics in spin-labelled proteins.

2.2 Chemical stability of the radical

The nitroxyl radical 1 can undergo redox processes to produce the oxoammonium cation 11 (Scheme 1), hydroxylamine 12 or secondary amine 13. From the viewpoint of maintaining the radical on the spin label (e.g. to enable EPR studies in living cells), protection against reduction to the hydroxylamine is important. The stability of a given nitroxyl radical to reduction is not only determined by its electrochemical reduction stability, but also by the cyclic framework in which it is contained, and the nature (charge, size, etc.) of the associated substituents.

The stability of nitroxyl radicals to reduction is often assessed using the biologically-relevant ascorbic acid as a reducing agent, albeit this is only a simple model for the various processes and differing reducing environments that may be encountered in cells. Through comparison of susceptibility to ascorbate, some general properties of nitroxides have been identified which improve the resilience of the radical in reducing environments. The first of these is that five-membered ring nitroxides are significantly more stable towards reduction than six-membered. This is likely a consequence of the change in hybridisation of the nitrogen atom that occurs on reduction, where torsional strain is relieved on going from sp to sp hybridisation for six-membered rings, but increases for five-membered rings. Inductive effects of substituents on the ring can also play a role, particularly for substituted pyrrolinoxyl radicals. An approximate order of stability is illustrated in Fig. 4.

The second general stabilising effect is that of increasing the steric bulk of the flanking alkyl substituents which shield the radical from reduction, or stabilise it relative to the hydroxylamine where an equilibrium exists. For example, tetraethyl-substituted isoindolinyl nitroxides have been shown to be highly resistant to ascorbic acid reduction, and its stability was further enhanced when bound to the ribose of RNA via a thiourea linker (14, Fig. 5). The enhanced stability of the PROXYL framework can be combined with such steric protection to give particularly stable radicals: the tetraethyl-flanked PROXYL (15) remains ~90% intact after two hours exposure to ascorbate or frog oocyte cells/cell extract. Bis(spiro-cyclohexyl) groups flanking the nitroxyl (16) also confer stability against bioreduction, although to a lesser extent than tetraethyl substituents. This is likely due to less effective steric shielding for a cyclohexane ring compared to the more mobile ethyl groups. However, some spirocyclic systems can confer remarkable stability: the fully substituted PROXYL (17) was found to be exceptionally resistant to ascorbic acid reduction.

One caveat in label design is that many of these extended alkyl chains increase label hydrophobicity, which may cause problems for some labelling strategies, or unwanted sample aggregation. The electronic influence of substituents can also affect stability, and a careful consideration of both steric and electronic effects is therefore required when designing labels.

2.3 Use of nitroxides in DEER

The pulsed EPR experiment known as double electron-electron resonance (DEER), or pulsed electron double resonance (PELDOR), has become a useful method for measuring nanometre distances between pairs of nitroxides. The popularity of this experiment in bis-nitroxide systems originates from its use of two microwave frequencies, which allows much of the nitroxide spectrum to be measured under common hardware limitations. The 4-pulse DEER experiment measures the dipolar coupling frequency as a modulation on a refocused echo. The DEER measurement needs to be set up such that the dipolar frequency can be measured accurately, which means that the echo must have a good signal-to-noise ratio (SNR) with the appropriate DEER inter-pulse delays (i.e. time window). The SNR can be improved by repeating the experiment and averaging results. The repetition rate is optimal when the echo is approximately more than 70% recovered, at 1.2 x T1 (longitudinal relaxation time), where T1 values are typically on the order of one millisecond. The time window is dependent upon the spin coherence time (TM) of the spin label. It is often shortest for samples with high local or global concentration, as has been explored for spin-labelled proteins in lipid membranes. Relaxation due to instantaneous diffusion can be reduced by working at as low concentrations as feasible and through careful choice of pulse lengths in experiments such as DEER. For the 5- and 6-membered nitroxides with four methyl groups flanking the radical, this balance between the T1 and Tm relaxation times with the Boltzmann distribution is often optimal at around 50 K.

A typical soluble spin-labelled biomolecule would have a Tm time of 2-3 µs and this would allow for a time window of ca. 3 µs which corresponds to the accurate measurement of a 3.5 nm distance. However, it has been found that using deuterated solvent and cryoprotectant significantly lengthens the Tm time since the 2H nuclei have a lower magnetic moment than 1H, and this reduces the rate of relaxation through spin diffusion. The loss through spin diffusion can also be reduced by using more advanced DEER pulse sequences, e.g. 5-pulse DEER. Further, Norman and co-workers have shown that deuteration of the molecule the spin label is attached to, in their case a protein, reduces the contribution from this relaxation mechanism to such an extent that distances over 14 nm can be measured accurately. Interestingly, isotope substitution of the protons on the gem-dimethyl substituents does not extend the relaxation time. Conversely, in CW EPR where distances over 1 nm are assessed by the dipole-dipole broadening on the spectral linewidth, it has been shown that deuteration of the label increases the upper measurable distance from 2 nm to about 2.5 nm. The alternative isotope substitution of 15N at the nitroxyl moiety has been applied to allow orthogonal labelling using two, or more, nitroxide spin labels: this makes use of the two microwave frequencies used in DEER, and indeed its CW predecessor ELDOR, and the only partial spectral overlap between the 15N (I = 1/2) and 14N (I = 1) EPR lineshapes.

The lower distance limit in the DEER experiment is determined by the requirement that the bandwidth of both sets of pulses can excite the full dipolar lineshape. In practice this has set the lower limit at approximately 1.5 nm. The DEER experiment itself only requires that the labelled molecule does not tumble fast enough to average out the dipole-dipole coupling between the spin labels. Thus, if the spin labels can be optimised such that their relaxation rates are favourable for measurement at higher temperatures, then tethered or otherwise immobilised molecules could be used. This would open up the possibility of measurement at or near physiological temperatures.

The DEER technique is capable of extracting distances and their distributions with nanometre accuracy, and also orientational information between the labels when the nitroxide spin labels have a well-defined, narrow distribution of conformations with respect to one another. The likelihood of this occurring is increased if the label is conformationally restricted through either steric bulk, or linker restriction. However, it is important that disruptions to the material are minimal, and that the label can be used in a facile manner through simple and efficient labelling procedures.

The conformations and dynamics of some spin labels attached to biomolecules, particularly methanethiosulfonate (MTS, 18, Fig. 6), have been investigated computationally and experimentally through EPR analysis and crystallography. Importantly, there is software freely available to enable users to label their target in silico and calculate the most probable conformations, since the conformation of the spin label tether must also be considered when interpreting DEER measurements.

2.4 Spin relaxation rates

The DEER experiment relies on measuring the dipolar coupling frequency between the spin labels, which increases as r-3 where r is the distance between the radicals. Hence, the measurement of relatively long distances requires the spin labels to have long spin coherence times, which also increases the concentration sensitivity of the experiment for a given dipolar frequency. Nitroxide spin label measurements are usually carried out at 50 K, and enhancing spin coherence times could enable measurements at higher temperatures (i.e. liquid nitrogen, rather than the expensive liquid helium currently required by laboratories not equipped with a closed-circuit cryostat). Ultimately, measuring DEER at higher temperatures, perhaps even physiological temperatures, would provide valuable structural information for many systems.