A new electron paramagnetic resonance (EPR) method, rapid scan EPR (RS-EPR), is demonstrated for 2D spectral spatial imaging which is superior to the traditional continuous wave (CW) technique and opens new venues for in vivo imaging. Results are demonstrated at 250 MHz, but the technique is applicable at any frequency.
We demonstrate a superior method of 2D spectral-spatial imaging of stable radical reporter molecules at 250 MHz using rapid-scan electron-paramagnetic-resonance (RS-EPR), which can provide quantitative information under in vivo conditions on oxygen concentration, pH, redox status and concentration of signaling molecules (i.e., OH•, NO•). The RS-EPR technique has a higher sensitivity, improved spatial resolution (1 mm), and shorter acquisition time in comparison to the standard continuous wave (CW) technique. A variety of phantom configurations have been tested, with spatial resolution varying from 1 to 6 mm, and spectral width of the reporter molecules ranging from 16 µT (160 mG) to 5 mT (50 G). A cross-loop bimodal resonator decouples excitation and detection, reducing the noise, while the rapid scan effect allows more power to be input to the spin system before saturation, increasing the EPR signal. This leads to a substantially higher signal-to-noise ratio than in conventional CW EPR experiments.
Relative to other medical imaging modalities, electron paramagnetic resonance imaging (EPRI) is uniquely able to quantitatively image physiological properties including pH1-3, pO24-7, temperature8, perfusion and viability of tissues9, microviscosity and ease of diffusion of small molecules10 and oxidative stress11. Estimation of the ease of disulfide cleavage by glutathione (GSH) in tissue and cells12,13 can report on redox status. For in vivo imaging, EPR in the frequency range between 250 MHz and 1 GHz is chosen because these frequencies provide sufficient depth of tissue penetration (up to several cm) to generate images for small animals in which intensities are not diminished by dielectric loss effects. Higher frequencies, such as 9.5 GHz14 (X-band) and 17 GHz (Ku-band)15,16 can be used for imaging of skin and hair or single cells, respectively. The success of EPRI at all frequencies depends on paramagnetic spin probes that are specific for tissues so that their location and fate may be imaged.
If the environment of an electron spin probe is spatially heterogeneous, the EPR spectrum is the sum of contributions from all locations. Spectral-spatial imaging divides the sample's volume into an array of small spatial segments and calculates the EPR spectrum for each of these segments17. This allows mapping of the local environment by measuring the spatial variation in the EPR spectrum. Magnetic field gradients are used to encode spatial information into EPR spectra, which are called projections. The spectral-spatial image is reconstructed from these projections18,19.
In RS-EPR the magnetic field is scanned through resonance in a time that is short relative to electron spin relaxation times (Figure 2)20,21. Deconvolution of the rapid-scan signal gives the absorption spectrum, which is equivalent to the first integral of the conventional first-derivative CW spectrum. The rapid-scan signal is detected in quadrature, so that both absorption and dispersion components of the spin system response are measured. This is essentially collecting twice the amount of data per unit time. Saturation of the signal in a rapid scan experiment happens at higher powers than for CW, so higher powers can be used without concern for saturation.20,22 Many more averages can be done per unit time in comparison to CW. Higher power, direct quadrature detection and more averages per unit time combine to give rapid scan a better signal-to-noise ratio (SNR), especially at high gradient projections that define spatial separation, leading to higher quality images. To achieve about the same SNR for an image of a phantom required about 10 times as long for CW as for rapid scan23.
The increased SNR also allows experiments at 250 MHz with low concentration spin trap adducts formed by the reaction of OH with 5-tert-butoxycarbonyl-5-methyl-1-pyrroline-N-oxide (BMPO-OH) which would be invisible to the CW method24. Dinitroxides connected with a disulfide linker are sensitive to cleavage by glutathione, and so can report on cellular redox status. Equilibrium exists, dependent on the concentration of glutathione present, between the di- and mono-radical forms. Observing these changes requires capture of the entire 5 mT wide spectrum, and can be achieved much faster with rapid scan EPR compared to stepping the magnetic field in a CW experiment.
A complete rapid scan system consists of four parts: the spectrometer, the main field magnet, the rapid scan coil driver, and the rapid scan cross-loop resonator. The spectrometer and the main field magnet function the same as in a CW experiment, setting the main Zeeman field and collecting the data from the resonator. The rapid scan coil driver generates the sinusoidal scan current that goes into specially designed rapid scan coils on the rapid scan cross-loop resonator. The rapid scan coils on the rapid scan cross-loop resonator generate a large homogeneous magnetic field, which is swept at frequencies between 3 and 15 kHz.
1. Setup of the Rapid Scan Coil Driver at 250 MHz
2. Preparation of Reagents and Phantoms
3. Setup of the Rapid Scan Instrument at 250 MHz
Note: Tuning of the resonator with an aqueous sample of nitroxide radical, which has a similar effect on resonator Q and tuning as buffer solution, is a good way to set up for the sample to be imaged
4. Execution of Rapid Scan Experiment
Note: Specific instructions related to analysis of phantoms containing BMPO-OH24, pH sensitive TAM radicals19,27 and redox sensitive dinitroxides28 are provided in the literature.
The product of the experiment is a set of projections that are reconstructed into two-dimensional (one spectral, one spatial) images with a false color scale to represent signal amplitude. Deep blue denotes baseline where no signal is present, green is low amplitude and red is highest. Slices along the x-axis (spectral dimension) depict the EPR signal (EPR transition) on a magnetic field axis. Along the y-axis (spatial dimension), separation between signals corresponds to the physical spatial separation between samples in the resonators.
Figure 3 shows a comparison of two images, acquired with CW (Figure 3B) or RS (Figure 3A) of a phantom with three different types of 15N substituted nitroxide radicals (Figure 3D). The broadest signal corresponds to 15N-Proxyl, a five member pyrrolidine ring with a negative charge at physiological pH, which could help target the molecule to specific cellular compartments. The doublet signal belongs to 15N-mHCTPO and is the result of a single hydrogen amidst otherwise complete deuteration. This single splitting has been optimized to monitor changes in oxygen concentration30. The narrowest signal comes from 15N-PDT, a flexible piperidine ring that is completely deuterated. It can be used to monitor oxygen concentration, or redox environment (reduction of the structure leads to decrease in EPR signal).
For the same 5-min acquisition time, the RS image shows superior spatial resolution and clarity of the spectral pattern for each radical. One reason for the improvement of RS over CW can be seen by comparing spectra at two different gradient strengths between the two techniques (Figure 3C). As the gradient strength increases the spectral signal is broadened. Considerable degradation of the CW spectrum under the high gradients (1 mT/cm) which encode spatial information.
Because a derivative signal broadens more quickly than an absorption signal, the SNR for the highest gradient CW projection (red trace) is very poor compared to that of the highest gradient RS projection (blue trace). Linewidth as a function of spatial position can be extracted from a 2D plot. Linewidth will be broad or narrow based on changes in oxygen concentration or viscosity around the nitroxide probe. The phantom imaged in Figure 3A was at room temperature and open to the air. Since oxygen content and viscosity (as determined by temperature) remained steady, the linewidth of each probe should be constant across the width of each tube containing a radical. Figure 4 shows the scatter in linewidths fit from slices through the 2D image compared to the true linewidth value (black horizontal line). The image slice values, especially for 15N-PDT, are a better match to the true linewidth value for RS (Figure 4A) than for CW (Figure 4B). This is also a result of the improved SNR of RS over the CW technique.
Another benefit of the RS technique is the ability to generate wide magnetic homogeneous field sweeps in a very short time. A typical scan frequency for experiments at 250 MHz is 9 kHz, corresponding to 0.11 msec. This is 0.11 msec whether the field sweep is 0.5 mT or 5.0 mT. Compare this to CW, where a 5.0 mT sweep will take tens of seconds to minutes. With rapid scan it becomes possible to quickly collect 100% of the spectral information in times which are amenable to in vivo imaging.
Figure 5 demonstrate wide spectrum RS-EPR imaging applied to spin trapping models. Important signaling molecules, like OH• and NO• are endogenous free radicals with very short lifetimes. In order to study these molecules, "spin traps" are used. An example of the reaction of spin trap 31 (BMPO) with OH• is shown in Figure 1B. Imaging of a phantom containing 5 µM BMPO-OH adduct is shown in Figure 5 (A, B). The spin-trap adduct signal is dependent on the starting concentration of OH• and has a half-life of 30 minutes allowing study of any processes which generate OH•. The nitronyl nitoxide32 was used as another example of wide spectrum imaging, but has been used in the past for spin-trapping of NO•33,34. Imaging of a phantom containing nitronyl is shown in Figure 5 (C, D). For spin traps, capturing the entire spectrum allows better designation of the original transient radical species that was present.
Sensitivity to physiological changes like pH and redox status is derived from changes in the entire spectrum. Figure 6 shows imaging with aTAM4. In Figure 6B, the profile of aTAM4 at pH= 7.0 (blue) has many spectral features, and a slice from the image matches well with the corresponding zero gradient spectrum (green). Compare this to the profile of aTAM4 at pH=7.4, Figure 6C, with fewer spectral features and still in good agreement with the corresponding zero gradient spectrum. Imaging of phantoms containing of the dinitroxide in its dimeric, and reduced monomeric form are shown in Figure 7. The two different spectra are generated by cleavage of a disulfide (S-S), and so convey sensitivity to redox environment1,35.
Figure 1. EPR probes are sensitive to many physiological changes. (A) An example of the pH-sensitive tri-aryl-methyl (TAM) radicals26. (B) Spin trap BMPO. (C) 15N-dinitroxide. (D) The nitronyl. (E) 15N-Proxyl. (F) 15N-mHCTPO. (G) 15N-PDT. Please click here to view a larger version of this figure.
Figure 2. Rapid scan EPR has inherently better SNR. (A) In CW EPR the amplitude h is a small fraction of the total signal, determined by the magnetic field modulation. (B) In direct-detected rapid scan, the full signal amplitude is detected. The signal to noise increase is evident in the experiment where superoxide generated by E. faecalis is trapped with BMPO at X-band. For the same 30 sec acquisition time, hardly any signal is observable in the CW spectrum (C) while a strong signal is observed in the rapid scan spectrum (D)36. Please click here to view a larger version of this figure.
Figure 3. Improved SNR allows better spatial resolution. For the same 5-minute acquisition time, the RS image (A) has better SNR and spatial resolution compared to the one acquired with CW (B). (C) There is good agreement between projections acquired with rapid scan (blue) and CW (red) when no gradient is present (0 mT/cm) (D). Please click here to view a larger version of this figure.
Figure 4. The information content of a rapid scan image is higher than for CW. (A) Slices of the 2D RS image. (B) Slices of the 2D CW image. The true linewidth (black horizontal line) of each sample is shown for comparison. See reference23. Please click here to view a larger version of this figure.
Figure 5. Rapid field sweeping allows capture of an entire spectrum in a few seconds. (A) 2D spectral-spatial image of a phantom consisting of BMPO-OH adduct. (B) A simulation fit to the zero-gradient BMPO-OH spectrum at 250 MHz was used to fit the initial BMPO-OH image and distinguish between regions containing BMPO-OH and noise containing regions. (C) 14N nitronyl radical which can be used for the trapping of nitric oxide in vivo. (D) Slices through each spectrum show the spectral shape at 250 MHz. See reference19. Please click here to view a larger version of this figure.
Figure 6. No part of the spectrum is left out, allowing better monitoring of physiologically induced spectral changes. (A) 2D spectral-spatial image of a phantom consisting of two tubes of pH sensitive aTAM4 radical. (B) Spectral profile of aTAM4 at pH= 7.0 (blue) and the corresponding zero gradient spectrum (green). (C) Spectral profile of aTAM4 at pH=7.4 B (blue) and the corresponding zero-gradient spectrum (green). See references19,26,37. Please click here to view a larger version of this figure.
Figure 7. Rapid scan opens the door to de vivo redox monitoring at 250 MHz. (A) 2D spectral-spatial images of 15N-dinitroxide. (B) Slices through top (blue trace) and bottom (red trace) compartments in the two images. (C) The top compartment remains the same, but the bottom compartment has been reduced with glutathione. (D) Slice through each image object showing the change in the 1D spectrum of the bottom compartment. See references1,28,35. Please click here to view a larger version of this figure.
Rapid-scan signals have higher frequency components than CW, and require a larger resonator bandwidth depending on linewidths, relaxation times, and the speed of the rapid-scans. The bandwidth required for a given experiment is based upon the linewidth and the scan rate of the magnetic field (Equation 2). Depending on the relaxation times of the probe under study (T2 and T2*), and the scan rate, oscillations can appear on the trailing edge of the signal. For nitroxide radicals with T2 ~500 nsec at 250 MHz (57th Rocky Mountain Conference on Magnetic Resonance, Epel, B, et al., 2015), experimental scan rates are often not high enough to observe any oscillations.
The experimental bandwidth is typically limited by the resonator bandwidth. Each half cycle of a rapid scan experiment is recorded with either decreasing or increasing field/frequency, so the experimental bandwidth is ½ the resonator bandwidth, as shown in (Equation 1). If the experimental bandwidth is limited by the choice of parameters such that it is greater than resonator bandwidth and oscillations are damped, broadening results in the deconvolved line. Since the experiment bandwidth is determined by the rate and linewidth of the radical being studied, understanding these features is a key component of the rapid scan experiment.
The current protocol demonstrates EPRI at 250 MHz of phantoms containing probes sensitive to oxygen, viscosity, pH, endogenous transient signaling molecules (i.e., OH•, NO•) and redox status. Spatial resolutions between 1 and 3 mm have been demonstrated, with experimental acquisition times between 29 seconds (single line of a 2 line 15N spectrum, Figure 3) and 15 minutes (full spectrum of 5 µM BMPO-OH, Figure 5). Method development with the phantoms shows use of RS-EPR images supersedes the conventional CW-EPR imaging technique23,24, and opens new avenues for in vivo imaging using EPR probes.
EPRI is advantageous over other in vivo imaging techniques based on fluorescence or phosphorescence, as EPR probes are sensitive to a wider variety of in vivo phenomena. In addition, RF penetration at 250 MHz is ~7 cm, so that anomalous tissue at a deeper level can be studied. Nuclear magnetic resonance imaging (MRI) provides very detailed anatomical maps, but struggles to provide quantitative physiological information. A combination of MRI and EPRI could one day result in an all magnetic resonance version of a positron-emission-tomography (PET)/computed tomography (CT) scanner. Such an instrument would provide the same benefit of PET/CT, but without the heavy radiation doses or expensive radio-tracers.
Method development with phantoms continues to push the limits of RS-EPR, but the ultimate goal is to implement the technique in laboratories using animal models. Calculations for image reconstructions will need to be improved to speed data collection for a 4D experiment (3 spatial, 1 spectral dimension). An improved algorithm is currently being developed and is essential for in vivo applications, however the proof of principle may be done with 2D imaging.
Many of the radicals, such as 15N-PDT, used in phantoms degrade quickly under in vivo conditions with half-lives of only 60 seconds. Radicals with an improved resistance to in vivo reduction39 have been synthesized and are important for building large enough concentrations in vivo. The enhanced sensitivity of RS-EPR over CW-EPR24 will be another benefit in solving this problem. The sensitivity of rapid scan is currently 5 µM for a phantom, and between 100 µM and 5 mM, depending on the probe to be imaged, for animal studies being performed at the University of Chicago (personal communication, Maggio, M., 2015). The RS method will continue to be developed to close this gap, but the application has already begun to move into actual in vivo applications (57th Rocky Mountain Conference on Magnetic Resonance, Epel, B, et al., 2015).
The authors have nothing to disclose.
Partial support of this work by NIH grants NIBIB EB002807 and CA177744 (GRE and SSE) and P41 EB002034 to GRE, Howard J. Halpern, PI, and by the University of Denver is gratefully acknowledged. Mark Tseytlin was supported by NIH R21 EB022775, NIH K25 EB016040, NIH/NIGMS U54GM104942. The authors are grateful to Valery Khramtsov, now at the University of West Virginia, and Illirian Dhimitruka at the Ohio State University for synthesis of the pH sensitive TAM radicals, and to Gerald Rosen and Joseph Kao at the University of Maryland for synthesis of the mHCTPO, proxyl, BMPO and nitronyl radicals.
4-oxo-2,2,6,6-tetra(2H3)methyl-1-(3,3,5,5-2H4,1-15N)piperdinyloxyl (15N PDT) | CDN Isotopes | M-2327 | 98% atom 15N, 98 % atom D, Quebec Canada |
4-1H-3-carbamoyl-2,2,5,5-tetra(2H3)methyl-3-pyrrolinyloxyl (15N mHCTPO) | N/A | N/A | Synthesized at U.Maryland and described in Reference 29 |
3-carboxy-2,2,5,5-tetra(2H3)methyl-1-(3,4,4-2H3,1-15N)pyrrolidinyloxyl (15N Proxyl) | N/A | N/A | Synthesized at U.Maryland and described in reference 25 |
4 mm Quartz EPR Tubes | Wilmad Glass | 707-SQ-100M | |
4-oxo-2,2,6,6-tetra(2H3)methyl-1-(3,3,5,5-2H4)piperdinyloxyl (14N PDT) | CDN Isotopes | D-2328 | 98% atom D, Quebec Canada |
pH sensitive trityl radical (aTAM4) | Ohio State University | N/A | Synthesized at Ohio State University and described in reference 26 |
Potassum Phosphate, Monobasic | J.T. Baker Chemicals | 1-3246 | |
6 mm Quartz EPR Tubes | Wilmad Glass | Q-5M-6M-0-250/RB | |
8 mm Quartz EPR Tubes | Wilmad Glass | Q-7M-8M-0-250/RB | |
5-tert-butoxycarbonyl-5-methyl-1-pyrroline-N-oxide (BMPO) | N/A | N/A | Synthesized at U.Maryland and described in reference 30 |
Hydrogen Peroxide | Sigma Aldrich | H1009 SIGMA | 30% |
16 mm Quartz EPR tube | Wilmad Glass | 16-7PP-11QTZ | |
Medium Pressure 450 W UV lamp | Hanovia | 679-A36 | Fairfield, NJ |
L-Glutathione, reduced | Sigma Aldrich | G470-5 | |
Nitronyl | NA | N/A | Synthesized at U.Maryland and described in reference 31 |
Sodium Hydroxide | J.T. Baker Chemicals | 1-3146 |