Spin pH and SH probes : enhancing functionality of EPR-based techniques

Along with significant progress in low-frequency electron spin resonance (ESR, also called electron paramagnetic resonance, EPR), other techniques such as Longitudinally-Detected ESR (LODESR), proton electron double-resonance imaging (PEDRI) and field-cycled dynamic nuclear polarization (FC-DNP) have been developed for in vivo applications. However their potential is still far from maximally defined, in part, because of the need for new specific function-directed spin probes. An application of stable nitroxides of imidazoline and imidazolidine types provides unique possibility to measure local values of pH and thiol content in various biological systems, including in vivo studies. These applications are based on the observation of specific chemical reactions of these nitroxides with protons or thiols, followed by significant changes in their EPR spectra. To increase sensitivity of pH probes for low-frequency EPR spectroscopy we evaluated two alternative approaches: (i) application of isotopically substituted labels, and (ii) acquisition of EPR spectra at high modulation amplitude. Spatial and spectral-spatial imaging (pH-mapping) using PEDRI and L-band EPR imagers were performed both on phantom samples and in vivo. The applications of the pH and SH probes in model systems, biological fluids, and in vivo in living animals are discussed.


Introduction
Significant progress has been made regarding in vivo EPR-based spectroscopy and imaging during the last decade.This includes low-frequency EPR spectroscopy, as well as instrumentation for spatial and spectral-spatial EPR imaging (EPRI) of the radicals [1][2][3][4][5][6][7][8], longitudinally-detected ESR (LODESR) [9], dynamic nuclear polarization (DNP) and proton electron double-resonance imaging (PEDRI) [10,11].However the potential of these techniques is still far from maximally defined, in part, because of the need for development of new specific function-directed spin probes.
Stable nitroxides of imidazoline and imidazolidine types provide unique possibility to measure noninvasively the local values of pH and thiol content in various biological systems, including in vivo studies [8,12].These parameters are of principal importance in biochemistry and physiology of living organisms, being significantly compromised in ischemic heart, many tumors, upon local skin treatments or wound healing, and local areas of infection or inflammation [2,[13][14][15][16][17]. Currently there are few methods available for monitoring or imaging pH and essentially no methods for monitoring or imaging thiols/GSH in living tissues.The most commonly used 31 P-NMR approach for pH detection has its own limitations, including lack of resolution (about 0.2-0.3 units and even less at lower pH), the fact that inorganic phosphate, Pi, concentrations vary with metabolism and ischemia, and its chemical shift depends on ionic strength [18][19][20][21].The possibility of accurate determination of pH values by EPR is based on early findings that EPR spectral parameters of stable nitroxides of the imidazoline and imidazolidine types depend on the pH of the medium [22,23].Here we discuss new developments in applications of spin pH probes using low-frequency EPR spectroscopy and imaging techniques both, in vitro and in vivo.
To our knowledge, there is no direct, noninvasive method for detection of thiols (e.g., GSH) in living tissues.Recently Kuppusamy et al. [2] demonstrated indirect evaluation and imaging of GSH levels in tumors in living animals (mice) by in vivo L-band EPR spectroscopy using a redox-sensitive but thiol-nonspecific spin probe.The application of a SH-specific spin probe might provide even greater specificity and sensitivity to the approach.The thiol-sensitive paramagnetic disulfide biradicals were proposed earlier for thiol detection by X-band EPR spectroscopy [24,25].Here we evaluated the ability of thiol detection by the disulfide biradicals using L-band EPR spectroscopy and discussed its possible in vivo applications.

X-band EPR spectroscopy
EPR spectra were recorded on a Bruker EMX spectrometer in a quartz capillary of 50 µl volume.The hyperfine coupling constant, a N , was measured as the distance between the low field and the central line of the triplet and was accurate to ±0.02 G. Solutions of the radicals in water or in buffers were titrated with solutions of HCl or KOH to the required pH.pH was measured using an AB15 pH meter (Fisher Scientific).

L-band EPR and EPRI instrumentation
L-band EPR spectra were measured using a home-built L-band (1.2 GHz) EPR spectrometer described previously [28][29][30].Imaging measurements were performed using the L-band EPR imaging instrumentation, consisting of an L-band EPR spectrometer, three sets of water-cooled gradient coils and a personal computer-based data acquisition system.The EPR spectrometer, magnet, and gradient system has been described previously [29][30][31].Each set of gradient coils is independently powered with a pair of Hewlett-Packard power supplies (Model 6030A), interfaced to the personal computer via GPIB bus.

Projection acquisition and image reconstruction
The low field peak of the nitroxide triplet spectrum was used for imaging.The imaging experiments were carried out using a variable strength of field gradient with a maximum of 30 G/cm.A total of 64 projections were acquired.The field of view and sweep window were 24 mm and 10 G, respectively.The measured projections were corrected for hyperfine artifacts [30].A two-stage, filtered-backprojection reconstruction algorithm was used to recover the image [32].

140 GHz EPR measurements
EPR spectra at 140 GHz were recorded on a homebuilt setup with millimeter-wave bridge designed and built in the Donetsk Physical-Technical Institute of the Ukrainian Academy of Sciences.

FC-PEDRI and FC-DNP measurements
FC-PEDRI images and FC-DNP spectra were obtained using a home-built imager/spectrometer at the University of Aberdeen [11].The Overhauser effect is used to detect the free radical probes' unpaired electrons via the measured NMR signal, obtained at 58.7 mT (2.5 MHz NMR frequency).For irradiation of the EPR at about 120 MHz the magnetic field is cycled down to approximately 5 mT in the FC-PEDRI experiments, and between 1.5 and 6.5 mT (stepped) in order to record an FC-DNP spectrum (essentially and Overhauser-detected EPR spectrum).

Animals
Adult male Sprague Dawley rats, body weight about 250 g, were used both for in vivo experiments using FC-PEDRI, and for obtaining blood samples for ex vivo L-band experimentation.For the FC-PEDRI studies the animals were fasted prior to being given by oral administration (gavage) 3 ml of 5 mM R1 (Scheme 1) agent.The animals were anesthetized by i.p. injection of 41 mg/kg BW ketamine and 20.5 mg/kg BW xylazine.Animals were killed immediately after the FC-PEDRI studies by pentabarbitone overdose while under anesthesia.Blood samples were obtained from the aorta of the anesthetized rats just before L-band measurements, and biradical R 2 SSR 2 (5 µl of 1 mM stock solution in 0.1 M PBS, pH 7.4) was added to 100 µl of the blood.All animal procedures carried out at the University of Aberdeen were done so in accordance with local guidelines and under British Home Office project licence no.PPL 60/2300 (M.A.F.).

pH sensitive nitroxides of the imidazoline and imidazolidine types
Spin pH probes of imidazoline and imidazolidine type have spectroscopically distinguishable protonated (RH + ) and unprotonated (R) forms due to the differences in their g-factors and hyperfine splittings, a N [23,33].This is clearly demonstrated by an EPR spectrum measured at 140 GHz and shown in Fig. 1 for the imidazolidine radical R1 (see Scheme 1 for the structure).Note that at high EPR frequency the contribution of the g-value in the spectral shift between the forms prevails over hyperfine splitting difference being equal to 6.7 G and 1.25 G, correspondingly, for R1.The ratio [RH + ]/[R] varies with pH according to the chemical equilibrium of the reaction of proton exchange, namely: where K a is an equilibrium constant of the reaction.This provides an experimental tool for pH determination using spin pH probes over the range of about 3 pH units centered on the pK of the particular radical.

Hyperfine splitting of pH-sensitive nitroxides as pH marker
At X-and L-band EPR frequencies the effect of g-factor is decreased resulting in overlapping of the spectral components of the R and RH + forms (Fig. 2).Therefore, spectral simulation becomes necessary for accurate determination of the [RH + ]/[R] ratio.Fortunately, at X-band (9.9 GHz) two convenient experimental parameters might be used as markers for pH determination, namely: (i) the ratio of the peak intensities of partly resolved high-field lines of the triplets, depicted as RH + and R in Fig. 2a, and (ii) splitting between low-and central-field components, normally referred as hyperfine splitting, a N .The latter parameter, measured at X-band, or hyperfine splitting between low-and high-field components, measured at L-band (Fig. 2b) and referred as 2a N , have been used almost exclusively as highly sensitive pH markers in numerous applications [5,[34][35][36][37][38][39][40][41].However, the sensitivity of this parameter to pH strongly depends both on EPR frequency and spectrometer settings (e.g., modulation amplitude) and has to be studied for its optimal application.Figure 3 demonstrates the dependence of a N on pH for the radical R2 measured from X-band and L-band EPR spectra.The observed narrowing of the a N titration curve at L-band results, in its turn, in a narrower pH range being available for pH assessment using this parameter by L-band EPR.Therefore further we discuss the origin of this effect and the possibility of its correction.
As demonstrated in Fig. 3 for the radical R2, an a N titration curve at X-band allows very good approximation by the equation: This is the conventional titration dependence for the case of fast frequency exchange between the R and RH + forms in terms of the EPR time scale.However, the imidazoline and imidazolidine spin pH probes normally demonstrate slow frequency R-RH + exchange (Figs 1, 2) in the pH range from 3 to 11 [42].
A good approximation of the titration curve for a N measured as the distance between low-and centralfield components of the X-band spectra (Fig. 3) is accidental due to the following factors: (i) practically equal broadening of these two EPR lines due to superposition of R and RH + forms (see Fig. 2 and captions), and (ii) insignificant shift between R and RH + forms compared with the linewidths.In the case of the L-band EPR spectrum (Fig. 2b) the shift between these two forms of the radical at lowand high-field components is equal to the ∆a N value and becomes comparable with the linewidth.This results in significant disturbance of the EPR lineshape and in narrowing of the a N titration curve (Fig. 3).Interestingly, broadening of the EPR linewidth by L-band spectra with detection at high modulation amplitude (Fig. 3c) increases the range of pH sensitivity of hf splitting (Fig. 4).Moreover, the detection at higher modulation amplitude results in significant improvement of signal-to-noise ratio.The described experimental approach could be important for applications in vivo where fundamental sensitivity is much lower.

Imaging of pH-sensitive probes and pH mapping
We have applied L-band EPR spectra detection at high modulation amplitude for developing of L-band pH-mapping approach.The smooth pH dependence of the positions of the low-and high-field components enabled us to convert the 'position' image to 'pH map' after measuring the position of the low-field peak using spectral-spatial imaging.Figure 5 demonstrates the ability of the approach to map pH in the phantom sample.This approach shows good spatial (0.2 mm) and functional (0.2 pH units at pH close to pK of the radical) resolution and has potential for in vivo applications.
Proton electron double-resonance imaging (PEDRI) represents a different approach for imaging of the radicals.Recently we applied the field-cycled PEDRI approach to image spin pH probes in living rats and its spectroscopy analog, field-cycled dynamic nuclear polarization (FC-DNP), for pH measurements in the rat stomach [8,34].The sensitivity of the approach can be improved using isotopically substituted pHsensitive nitroxides [27] which is important for increasing the experimental time window and lowering probe dosage.Both FC-PEDRI and FC-DNP use the Overhauser effect for detection of radicals.EPR irradiation of the radical's causes a transfer of polarization from electron to coupled proton spins, and the efficiency of this transfer depends, among other parameters, on the extent to which the EPR is saturated.At a given EPR irradiation power level (usually constrained by the requirement not to overheat the animal under study) a narrow-line probe will be more saturated than its broader-line counterpart, so that the signal-to-noise ratio of the measured image or spectrum will be enhanced.Therefore, D-substituted agents, e.g., R1* (Scheme 2), are beneficial for these studies.A further benefit can be obtained from the use of 15 N-substituted agents (R1**, R2**, Scheme 2), since the reduction of the number of EPR lines from 3 to 2 again improves the efficiency of the Overhauser polarization transfer.Figure 6 shows typical FC-DNP spectra obtained from aqueous solutions of R1, R1* and R1**.It can be seen that the depth of the spectral lines is increased in the D-substituted agent and again by 15 N substitution.Figure 7 shows a set of in vivo FC-PEDRI images obtained following gavage of 3 ml of 5 mM R2** solution into the animal's stomach.In the 'Difference' image (Fig. 7c) the location of the agent in the animal's stomach can clearly be seen.
Note that EPR spectra of the nitroxides with 14 N-O and 15 N-O fragments do not overlap (Fig. 8).This opens the principal possibility to distinguish the signals from different areas in heterogeneous samples, e.g., simultaneous detection of intracellular and extracellular pH with specifically targeted probes.Recently we proposed synthesis of pH-sensitive nitroxides with ester groups for its intracellular targeting [43].

Thiol sensitive nitroxyl biradicals of the imidazoline and imidazolidine types
Imidazoline and imidazolidine biradical disulfides, R 1 SSR 1 and R 2 SSR 2 (Scheme 3) were earlier proposed as thiol-specific paramagnetic reagents [24,25].These paramagnetic disulfides react with thiols with splitting of the disulfide bond, resulting in characteristic changes of the X-band EPR spectra  (Fig. 9).The rate of the reaction of the biradicals with thiols increases with pH.For the biradical R 2 SSR 2 this reaction with 1 mM GSH proceeds in a few seconds at alkaline pH (Fig. 9).However at physiological pH the biradical R 2 SSR 2 reacts with thiols slowly (k ≈ 0.26 M −1 s −1 for GSH at pH 7.0 [25]) allowing the detection of the kinetics of the reaction (Fig. 10).The measurement of the kinetics of the biradical disulfide bond splitting does not require an excess of the label over thiols and therefore provides noninvasive experimental tool for thiols detection by EPR.Note that the fast reactive R 1 SSR 1 probe can be applied only in a "static" approach which measures the fraction of the biradical splitted by the reaction  with thiols.This approach unambiguously requires an excess of the probe over thiols and results in total thiols consumption under measurement therefore making it impossible for in vivo use.
Figure 10 shows the time evolution of the L-band EPR spectra of the biradical R 2 SSR 2 in untreated rat blood.The kinetics demonstrates a convenient time window for the biradical, R 2 S-SR 2 reaction with thiols (mostly GSH), as well as its relative stability towards reduction (less than 10% of the radical was reduced during 15 min incubation in the blood).The R 2 S-SR 2 radical freely penetrates cellular membranes (lipophilicity coefficient k = 240 [25]) reacting preferably with GSH located in erythrocytes.The reaction with protein thiols seems to be very slow: the rates constants for the reaction of R 2 S-SR 2 with SH groups of human serum albumin and hemoglobin are hundreds of times less compared with that for GSH [25].This allows an estimate of the reduced GSH concentration in the whole rat blood from the linear part of the kinetics shown in Fig. 10, which was found to be equal to 0.95 ± 0.2 mM, in reasonable agreement with literature data [44,45].
Recently we demonstrated the possibility of applying the kinetic EPR approach, using a redoxsensitive nitroxide, for GSH mapping in vivo in tumor-bearing mice [2].However it is necessary to note that the applied probe was not GSH-or thiol-specific, but rather sensitive to any reducing compound.Therefore application of SH-sensitive probe, R 2 SSR 2 , which entails a similar kinetic EPR approach, might provide even more specific information and approach for GSH mapping.
In summary, pH and SH sensitive imidazoline and imidazolidine nitroxides were found to be the promising functional probes for biological EPR applications, including that in vivo.

Scheme 1 .
Scheme 1.Chemical structures of imidazolidine radical R1 and imidazoline radical R2.The site of protonation, nitrogen atom N-3, is shown in bold.

Figure 1 .
Figure 1.The 140 GHz EPR spectrum of a 0.5 mM aqueous solution of the imidazolidine radical R1 (Scheme 1) at pH 4.7.The dotted lines depict the position of the peaks corresponding protonated, RH + , and neutral, R, forms of the radical.

Figure 2 .
Figure 2. EPR spectra of aqueous solutions of 0.5 mM imidazoline nitroxide, R2, at pH 6.1 measured at X-band (a) and L-band (b), (c).The spectrum (c) was detected at significantly higher modulation amplitude, 2 G, compared with 0.5 G modulation for the spectra (a) and (b).Note that shifts between RH + and R forms of the radical for low-and central-field components of the X-band spectrum (a) practically coincide (corresponding apparent shifts (∆a N − ∆g) ≈ ∆g ≈ 0.4 G) and are significantly lower than EPR linewidths of these forms (∆Hpp(RH + ) ≈ ∆Hpp(R) ≈ 0.8 G).Spectrum (c) shows improvement of the signal-to-noise ratio compared with spectrum (b) by about 2.1 times.

Figure 3 .
Figure 3.The pH dependence of hyperfine splitting, a N , measured as a distance between low-and central-field components of X-band (•) and L-band ( ) EPR spectra.The solid line was calculated using Eq.(1) with parameters a N (R) = 15.8G, a N (RH + ) = 15.0G, pK = 6.1.

Figure 4 .
Figure 4.The pH dependences of hyperfine splitting, 2a N , measured as the distance between the low-and high-field components of the L-band EPR spectra detected at different modulation amplitude: 0.5 G ( ) and 2 G ( ).The solid line was calculated using Eq.(1) with parameters 2a N (R) = 31.7 G, 2a N (RH + ) = 30.15G, pK = 6.1.

Figure 5 .
Figure 5. Cross-sectional 3D (1-spectral/2-spatial) image of a phantom prepared using capillary tubes of 2 mm diameter filled with 1 mM solutions of the radical R2 prepared at different pH.Left: spatial distribution of the hyperfine interaction splitting, 2a N , determined from position mapping of the low-field EPR peak.Right: pH map.The data acquisition parameters were: acquisition time, 5 min; projections, 64; maximum gradient, 30 G/cm; field of view, 24 mm.

Figure 6 .
Figure 6.FC-DNP spectra recorded from 2 mM aqueous solutions of R1 (a), R1* (b) and R1** (c).The pH values of the samples were 6.47, 6.20 and 6.25, respectively (i.e., all above the pK of this agent), the volume 15 ml.The pulse sequence repetition time was 1200 ms, with EPR irradiation at 121.04 MHz, applied power 19.5 W and duration 400 ms.

Figure 7 .
Figure 7. FC-PEDRI images obtained in vivo from an anesthetized Sprague-Dawley rat given a gavage of 3 ml of 5 mM solution of R1**.The animal was imaged supine.Images are coronal projective (i.e., no slice selection) with field-of-view 150 mm, matrix size 64 × 64.An interleaved (EPR off / EPR on) pulse sequence was used, with a repetition time (TR) of 1200 ms.EPR irradiation was applied at 120 MHz, power 30 W, duration 400 ms, at field strength 5.25 mT.Images shown are EPR-off (a), EPR-on (b) and difference (c); the latter shows the location of the agent in the animal's stomach.

Figure 8 .
Figure 8. X-band EPR spectrum of the mixture of 0.5 mM solutions of the radicals R2 and R2* in 0.1 mM Na-phosphate buffer, pH 5.9.Spectrometer settings were as following: microwave power 10 mW, modulation amplitude 0.8 G.

Figure 9 .
Figure 9. X-band EPR spectrum of 0.1 mM solution of the R 2 S-SR 2 in 10 mM Na 2 B 4 O 7 buffer, pH 11.2, before (left) and 2 min after addition of 30 µM GSH (right).Insert: the scheme of the reaction of thiol-disulfide exchange responsible for observed spectral changes.The measurement of relative changes of the intensities of monoradical (Im) or biradical (Ib) components allows quantitative determination of glutathione, GSH.

Figure 10 .
Figure 10.The kinetics of the increase of the low-field component of the L-band EPR spectra of the biradical R 2 SSR 2 measured in the blood taken from the Sprague-Dawley rat.The linear approximation in the time range less than 400 sec has been used for estimation of GSH concentration.Insert: L-band EPR spectra of the 100 µl blood sample measured immediately (a) and 700 sec (b) after addition of 5 µl of the biradical R 2 SSR 2 , final concentration of the label 50 µM, acquisition time 10.5 s.