Introduction

T2* mapping is a clinically accepted imaging tool for the assessment of iron overload in the heart and liver1,2,3,4,5,6,7. In both organs, T2* relaxation times have a direct, inverse correlation with total iron content8,9,10,11. Recent studies suggest that T2* relaxation times also correlate with iron oxidation state. In an ex vivo model system, the addition of a reducing agent significantly increased T2* relaxation times12. Moreover, as part of an ongoing phase II clinical trial, glioblastoma patients treated with a combination of radiation, temozolomide, and pharmacological ascorbate demonstrated increased T2* relaxation times following pharmacological ascorbate treatment13. In this context, the authors propose that ascorbate can act as a reducing agent facilitating the conversion of ferric (Fe3+) to ferrous (Fe2+) iron and that T2* relaxation time is sensitive to iron oxidation state, in addition to total iron content.

T1, T2, and T2* relaxation are intrinsic contrast mechanisms in MR imaging. Following the application of a radiofrequency (RF) pulse at the resonant (i.e. Larmor) frequency, a portion of the protons will be tipped away from the Z-axis and into the XY-plane. Once the RF pulse is turned off, affected protons will begin to (1) re-align in Z-axis and (2) dephase in the XY-plane due to interactions with other protons. The time for affected proton spins to re-align in the Z-axis is the T1 relaxation time and the time for the same protons to dephase in the XY-plane is known as the T2 relaxation time. T1 and T2 relaxation are inherent properties of the tissue and does not include local field variations. T2* relaxation is the increased dephasing of proton spins in the XY-plane14 due to local magnetic field inhomogeneities from (a) intrinsic inhomogeneities associated with scanner variations and (b) deposition of endogenous or exogenous paramagnetic materials within the anatomical region of interest (e.g. iron). Protons in the presence of these magnetic field inhomogeneities dephase more rapidly resulting in the T2* relaxation times being shorter than T2 relaxation times.

Transition metals can exist in a variety of oxidation states with a wide array of spin configurations. The most physiologically relevant transition metal is iron, which can exist in the ferric (Fe3+) and ferrous (Fe2+) state, although it may also transiently exist as Fe4+15,16,17. Other transition metals are also present within tissues. Manganese (Mn) and copper (Cu) are both present in proteins such as superoxide dismutase 1 and 2 (SOD1 = CuZnSOD; SOD2 = MnSOD)18,19. Typically, the 5 d-orbitals of transition metals are degenerate in a labile state. However, when coordinated by a ligand (e.g. under physiological conditions), the d-orbitals exhibit a loss of degeneracy. The number of bound ligands and ligand bond strengths determine the associated d-orbital splitting. This loss of degeneracy gives rise to an angular momentum associated with unpaired electrons for these metals. For example, a high spin iron complex with an octahedral ligand configuration has a quantum spin (S) of S = 5/2 (Fe3+) or 2 (Fe2+). Conversely, low spin iron will have S = 1/2 (Fe3+) or 0 (Fe2+). Based on the number of d-orbital electrons, transition metals exhibit a wide array of quantum spins (Supplemental Table S1). The unpaired electrons thus exhibit a spin magnetic moment (\({\mu }_{s}\)) that is related to the number of unpaired electrons (Eq. (1)):

$${\mu }_{s}=g\sqrt{S(S+1)},$$
(1)

where g is the electronic g-factor (≈ 2.002).

The spin magnetic moment of transition metals is directly related to their measured paramagnetic molar susceptibility (\({\chi }_{mol}\)) (Eq. (2)):

$${\mu }_{s}= \sqrt{\frac{3k{\chi }_{mol}T}{N{\beta }^{2}}}=2.828\sqrt{{\chi }_{mol}T},$$
(2)

where \(k\) = Boltzmann’s constant, N is Avogadro’s number, β is the Bohr magneton, and T is the temperature in Kelvin (K). The unpaired electron’s orbital angular momentum contribution to transition metal ion magnetic moments is usually small or negligible, in contrast to lanthanide ions where this can be a significant additional magnetic contribution. Thus, \({\mu }_{s}\) or S is often sufficient to estimate \({\chi }_{mol}\) for transition metal ions.

To evaluate the relationship between the number of unpaired electrons and magnetic susceptibility through the associated magnetic moment, Eqs. (1) and (2) can be rearranged to show:

$${\chi }_{mol}=\frac{S(S+1)}{1.99T}.$$
(3)

A paramagnetic compound has a non-zero electron spin and associated magnetic moment that can engage in proton-electron dipole–dipole interactions. Thus, we hypothesize that the electron spin magnetic moment has a significant impact on T2* relaxation and that T2* mapping is sensitive to alterations in the electronic properties of biologically relevant transition metals.

Results

T2* relaxation times are directly proportional to magnetic susceptibility

Equation (3) indicates that \({\upchi }_{\mathrm{mol}}\) is directly proportional to the spin quantum number associated with the number of unpaired electrons. We reviewed previously published data regarding molar susceptibility of various transition metal compounds (Fe3+, Mn2+, Fe2+, Ni2+,Cu2+; Supplemental Table S2) and found that their relative \({\upchi }_{\mathrm{mol}}\) values are directly proportional to the associated S(S + 1) term (R2 = 0.73, r = 0.98; Supplemental Fig. S1). To test the sensitivity of T2* relaxation to the electron configuration, we first measured the volumetric magnetic susceptibility (χvol) of various transition metals at an equal concentration (1 M). We observed that χvol is linearly dependent upon the number of quantum spin (S(S + 1)) of the transition metal (R2 = 0.98; Fig. 1A). To validate this result, we compared the known values for different metal complexes (\({\upchi }_{\mathrm{mol},\mathrm{ exp}})\) to our experimentally determined susceptibility values \({(\upchi }_{\mathrm{vol},\mathrm{ obs}})\) and found a linear relationship as anticipated (R2 = 0.99, Supplemental Fig. S2). Our experimental \({\chi }_{vol}\) measures are reported in ppm, which can be used to calculate the associated \({\chi }_{mol}\) values using the following relationship (Eq. (4)):

$${\upchi }_{\mathrm{mol}}=\frac{\mathrm{M}}{\uprho }*{\upchi }_{\mathrm{vol}},$$
(4)

where M is the molar mass (kg mol−1) and ρ (m3 kg−1).

Figure 1
figure 1

T2* and χvol is directly proportional to the number of unpaired electrons. Volumetric magnetic susceptibility measures were taken of 1 M concentrations of Ga(NO3)3 (Ga3+), CuSO4 (Cu2+), NiCl2 (Ni2+), (NH4)2Fe(SO4)2·6H2O (FAS, Fe2+), Fe(NO3)3 (Fe2+), and MnSO4 (Mn2+) in 4 mm O.D. quartz EPR tubes. Samples were prepared at room temperature (25 °C). (A) Linear correlation between experimentally determined volumetric magnetic susceptibility and the spin-only magnetic moment of various inorganic species. Each metal was left unchelated and diluted in double-distilled H2O at a 1 M concentration. (B) T2* map of transition metals (100 µM) embedded in a 1% agarose gel and scanned using a 7 T scanner. (C) Linear correlation between experimentally determined T2* relaxation times and the spin-only magnetic moment of various inorganic species. (D) Linear correlation between T2* relaxation times and volumetric magnetic susceptibilities determined experimentally. Each experiment was done with triplicate measures.

Full size image

To test if these paramagnetic metals have a similar effect on T2*; Cu2+, Ni2+, Fe2+, and Fe3+ were embedded in a 1% agarose gel at an equal concentration (100 µM) and scanned using a 7 T magnet to generate T2* maps (Fig. 1B). T2* relaxation times were linearly dependent on the quantum spin (S(S + 1)) for the examined transition metals (R2 = 0.89, Fig. 1C). T2* relaxation times were also inversely proportional to the experimentally determined \({\upchi }_{\mathrm{vol}}\) measurements (R2 = 0.89, r =  − 0.96; Fig. 1D). However, it is worth noting that the signal will be nearly fully decayed for material with very large \({\chi }_{vol}\) (e.g. paramagnetic MR contrast), making the evaluation of signal changes difficult. These data suggest that T2* relaxation times may be significantly affected by the electron spin magnetic moment (µs) and number of unpaired electrons (S) in the surrounding environment.

T2* mapping detects Fenton chemistry

To determine if the oxidation of Fe2+ via Fenton chemistry reaction (Eq. (5)) is reflected in T2* relaxation, H2O2 was incubated at room temperature and pressure with ferrous ammonium sulfate (FAS; Fe2+).

$${\mathrm{H}}_{2}{\mathrm{O}}_{2}+ {\mathrm{Fe}}^{2+} \to {\mathrm{Fe}}^{3+}+ {\mathrm{HO}}^{-}+ {\mathrm{HO}}^{\cdot }.$$
(5)

Following a 15 min incubiation, the H2O2 concentration was 100.4 ± 1.4 mM as compared to 0.59 ± 0.02 mM with FAS (Fig. 2A). To assess the oxidation of Fe2+ by H2O2, each solution (H2O2, FAS, and H2O2 + FAS) was diluted 1:15 in 5 mM ferrozine buffer and measured using UV–Vis spectroscopy at 562 nm (Supplemental Fig. S3). This technique allows for the detection of a purple ferrozine-Fe2+ complex, while ferrozine-Fe3+ remains colorless20. Following H2O2 addition, Fe2+ concentration was 0.37 ± 0.01 µM compared to a final, unperturbed Fe2+ concentration of 90.0 ± 1.2 µM (Fig. 2B). Each sample (H2O2, FAS, and H2O2 + FAS) was then diluted 1:15 in 1% agarose gel and scanned at 7 T to generate a T2* map (Fig. 2C). Quantitative analysis of each phantom revealed a mean T2* relaxation time of 70.9 ± 1.5 ms for the H2O2 sample, 54.4 ± 0.4 ms for the unperturbed Fe2+ sample, and 38.4 ± 0.1 ms following oxidation (Fig. 2D). These data suggest that T2* mapping can detect the chemical conversion of Fe2+ to Fe3+ via Fenton chemistry.

Figure 2
figure 2

T2* mapping detects Fenton chemistry. (A) Measurement of H2O2 using UV–Vis spectroscopy (240 nm; ε = 46.3 mol−1 cm−1; 1 cm pathlength) following 15 min incubation of 1.5 mM FAS ((NH4)2Fe(SO4)2 6H2O) with 100 mM H2O2 for 15 min in double-distilled H2O. (B) Measurement of Fe2+ following 1:15 dilution in 5 mM ferrozine using UV–Vis spectroscopy (562 nm; ε = 27,900 mol−1 cm−1; 1 cm pathlength). (C) Representative T2* map of 1:15 dilution in 1% agarose gel. (D) Mean T2* relaxation times quantified using Slicer3D software by generation a 1 mm ROI for each phantom and calculating the mean T2* relaxation time using the data quantification package within the software. Error bars represent SD of n = 3 biological replicates. Statistical analysis was performed using a one-way ANOVA test with statistical significance defined as a false positivity rate less than 5% (*p < 0.05).

Full size image

T2* relaxation differentiates between Ga3+ and Fe3+

Ga3+ is a d10 post-transition metal that is diamagnetic due to a completely full d orbital set with ten paired electrons (S = 0) while Fe3+ is a highly paramagnetic d5 metal with S = 5/2 because of its five unpaired electrons. As expected, the Ga3+ phantom had a significantly longer T2* relaxation time (92.2 ± 1.6 ms) compared to Fe3+ (74.5 ± 2.0 ms) (Fig. 3A,B).

Figure 3
figure 3

T2* relaxation differentiates between Ga3+ and Fe3+. (A) 100 µM Ga(NO3)3 and Fe(NO3)3 solutions embedded in a 1% agarose gel in 4 mm O.D. quartz EPR tubes and scanned using a 7 T magnet to generate T2* maps. (B) Mean T2* relaxation times quantified using Slicer3D software by generating a 1 mm ROI for each phantom and calculating the mean T2* relaxation time using the data quantification package within the software. *p < 0.05 using a paired t-test. Error bars represent SEM of three replicates. (C) Representative T2* map of U251 glioblastoma cell pellets following a 3 h treatment of 120 µM Ga(NO3)3 or Fe(NO3)3. (D) Mean T2* relaxation times quantified using Slicer3D software by generating a 1 mm ROI for each phantom and calculating the mean T2* relaxation time using the data quantification package within the software. Error bars represent SD of n = 3 biological replicates. *p < 0.05 using a one-way ANOVA test.

Full size image

Ga3+ and Fe3+ also behave similarly in biological systems as they both bind transferrin in circulation and enter cells through transferrin-receptor mediated endocytosis21,22,23. Treatment of U251 glioblastoma cells with either 120 µM Ga3+ or Fe3+ for 3 h caused a significant increase in T2* relaxation times (152.0 ± 1.7 ms) in Ga3+ treated cells compared to untreated cells (133.3 ± 3.1 ms) while Fe3+ resulted in a significant decrease in T2* relaxation (119.9 ± 3.9 ms; Fig. 3C,D).

Discussion and conclusions

Several studies have showed that T2* mapping is a useful tool for assessing iron concentrations in vivo. T2* is widely accepted as an imaging approach to assess for iron overload in the heart and liver5,6,7,8,9,10,11. It has also been shown that T2* is directly proportional to iron concentration in various cortical brain regions24. However, more recent studies have suggested that T2* relaxation can detect tissue iron-oxidation state12,13. We believe this is because electron contributions from the local environment are driving changes in T2* due to the different d orbital valence configurations of Fe3+ (d5) and Fe2+ (d4). In an ex vivo model, human cadaveric brain tissue was exposed to a reductant to convert Fe3+ to Fe2+ or iron extraction to reduce the iron content of the tissue12. In both conditions, T2* was increased further supporting the iron concentration dependence (iron extraction) and spin-dependence (iron reduction) theories. In a study reporting the preliminary results of a phase II trial for glioblastoma patients treated with pharmacological ascorbate, a small cohort of patients showed a significant increase in T2* relaxation times 4 h after an 87.5 g infusion of ascorbate13. Ascorbate (vitamin C) is a one-electron reducing agent that readily converts Fe3+ to Fe2+ and can kill cancer cells in an iron-dependent mechanism25,26,27. This study also showed that Fe3+ has a much greater concentration-dependent effect on T2* relaxation than Fe2+, although Fe2+ still decreased T2*13. Taken together, this suggests that T2* relaxation is dependent on both the concentration and spin-configuration of the paramagnetic metal being considered.

In this study, we show that T2* relaxation times can detect transition metals that have varying electron spin contributions (µs) and are linearly proportional to their associated magnetic susceptibility (χvol). We show that T2* relaxation times are sensitive to Fe2+ oxidation and able to detect Fenton chemistry reactions. Reactions with H2O2 are thought to be a major contributor to Fe-mediated oxidative damage in cells17,28. Fenton chemistry is a critical oxidation step following the radiolysis of water and can lead to ionizing radiation induced DNA damage28,29,30. Similar to iron, copper, manganese, and nickel are also able to undergo Fenton-type reactions leading to DNA damage31,32,33,34. Therefore, being able to readily detect alterations in metal oxidation state may provide significant information regarding the biochemical processes that are central to metal catalyzed pathologic processes and therapeutic response (e.g. radiotherapy) We also demonstrate that T2* mapping is able to differentiate between diamagnetic (d10; S = 0) Ga3+ and paramagnetic (d5; S = 5/2) Fe3+ metals as the high magnetic susceptibility of Fe3+ has a significant shortening of the T2* relaxation time compared to Ga3+. We validated this effect in a relevant, controlled, in vitro model system that controlled for several important variables including: (1) metal concentration, (2) metal oxidation state, (3) cellular uptake. U251 glioblastoma cells were treated for 3 h with identical concentrations of either Ga3+ or Fe3+ (120 µM). Ga3+ and Fe3+ are both taken into cells via transferrin—receptor mediated endocytosis to control for metal uptake23. Therefore, by supplementing cells with two metals where the only notable variable is the electron spin contribution, we show that the addition of a diamagnetic metal (Ga3+) increases T2* relaxation times while the paramagnetic Fe3+ decreases T2* relaxation times. This suggests that T2* relaxation may be useful in detecting diamagnetic/paramagnetic shifts in cells or tissues (e.g. influx/efflux of O2 to detect hypoxia).

In conclusion, we provide a revised understanding that changes in T2* relaxation times are being altered by proton-electron dipole–dipole interactions. This novel interpretation may allow investigators and clinicians to utilize T2* mapping to probe not only metal concentrations, but also, metal spin states. This understanding may be employed to develop new, non-invasive biomarkers to better understand human pathology, therapeutic response, and drug delivery by probing electron motion and metal metabolic changes in vivo.

Materials and methods

Magnetic susceptibility measures

Magnetic susceptibility measurements were performed by dissolving the desired metal to a concentration of 1 M in double–distilled H2O. Metals used in this study are provided in supplemental information (Supplemental Table S1). The metal solution was then put in a 4 mm O.D. Wilmad—quartz EPR tube (4 mm O.D.; Wilmad-LabGlass, Vineland, NJ, 707-SQ-250 M), sealed with parafilm, and measured at room temperature using a Johnson-Matthey MSB Evans magnetic susceptibility balance. Measurements were done in comparison to a water blank.

MRI studies

Phantoms were generated by dissolving the desired metal in double-distilled H2O to get an initial concentration of 1.5 mM. Appropriate metals were then diluted 1:15 in 1% agarose gel (Seaplaque low gelling temperature agarose; Sigma-Aldrich, A9414) and allowed to solidify in EPR tubes. Images were collected on a 7T GE MR901 small animal scanner, a part of the Small Animal Imaging Core at the University of Iowa. T2* weighted images were collected using a gradient echo sequence (TR = 10 ms, TE = 2.2, 8.2, 14.2, and 20.2 ms, 256 × 256 resolution, 2 signal averages). A B0 shimming routine was performed to mitigate the effect of macroscopic filed inhomogeneity on T2* measurements. The echo time (TE) is the time between the radiofrequency pulse and the sampling of the MRI signal, the repetition time (TR) is the time between two repeated pulse cycles, and B0 is the main external magnetic field strength (7T). T2* maps were generated using a combination of 4 echo times collected and fitting each voxel to a mono-exponential curve using in-house python code. Images were imported to Slicer3D software where regions of interest (ROIs) were delineated and mean T2* values were calculated using the label statistics tool within Slicer3D35.

Ultraviolet–visible light spectroscopy

Ultraviolet–visible light (UV–Vis) spectroscopic measurements were performed using a Beckman DU600 spectrophotometer. Samples were diluted to appropriate concentrations in double-distilled H2O and placed in a 1 cm cuvette. Fe2+—ferrozine complex formation being measured using a 400–800 nm wavelength scan and H2O2 was monitored using the absorbance at 240 nm. Fe2+—ferrozine and H2O2 concentrations were calculated using Beer’s Law using the absorbance at 562 nm (ε562 = 27,900 mol−1 cm−1) and 240 nm (ε240 = 46.3 mol−1 cm−1), respectively.

Cell cultures

U251 glioma cells were cultured in DMEM-F12 media (15% FBS, 1% penicillin-strep, 1% Na-pyruvate, 1.5% HEPES, 0.1% insulin, and 0.02% fibroblast growth factor). Cells were plated in 100 mm2 dishes and grown to 70–80% confluence at 21% O2. Cells were either left untreated or supplemented with 120 µM Ga (NO3)3 or Fe(NO3)3 for 3 h at 21% O2. Cells were trypsinized and harvested by spinning at 1200 RPM for 5 min. Supernatant media was aspirated and cells were resuspended in 250 µL DPBS. Cells were then placed in the appropriate well of a PCR plate embedded in 1% agarose gel. Cells were allowed to collect at the bottom of the well to form a dense pellet for approximately 30 min. Following pellet formation, the entire phantom containing all groups was scanned in the MR901 using a 6.0 cm FOV while all other parameters for the multi-echo gradient echo sequence was the same as used for the phantom study.