Comparison of phosphonate, hydroxypropyl and carboxylate pendants in Fe(III) macrocyclic complexes as MRI contrast agents

Abstract

Fe(III) macrocyclic complexes containing a macrocycle and three pendant groups including phosphonate (NOTP =1,4,7-triazacyclononane-1,4,7-triyl-tris(methylenephosphonic acid), carboxylate (NOTA = 1,4,7 - triazacyclononane - N,N',N" – triacetate) or hydroxypropyl (NOHP =(2S,2'S,2”S)−1,1',1”-(1,4,7-triazonane-1,4,7-triyl)tris(propan-2-ol) ) were studied in order to compare the effect of these donor groups on solution chemistry and water proton relaxivity. All three complexes, Fe(NOTP), Fe(NOHP) and Fe(NOTA), display a large degree of kinetic inertness to dissociation in the presence of phosphate and carbonate, under acidic conditions of 100 mM HCl or 1 M HCl or to trans-metalation with Zn(II). The r₁ proton relaxivity of the complexes at 1.4 T, 33 °C is compared over the pH range of 1 to 10. At pH 7.4, 33 °C, 1.4 T, Fe(NOHP) has the largest relaxivity (1.5 mM⁻¹s⁻¹), Fe(NOTP) is second at 1.0 mM⁻¹s⁻¹, whereas Fe(NOTA) is the lowest at 0.61 mM⁻¹s⁻¹. Fe(NOTP), Fe(NOHP) and Fe(NOTA) all show an increase in relaxivity at very acidic pH values (< 3) that is consistent with an acid-catalyzed process. Variable temperature ¹⁷O NMR studies at near neutral pH are consistent with the absence of an inner-sphere water molecule for Fe(NOTP) and Fe(NOHP), supporting second-sphere or outer-sphere water contributions to proton relaxation. Fe(NOTP) shows contrast enhancement in T₁ weighted MRI studies in mice and clears through a renal pathway.

1. Introduction.

Magnetic resonance imaging (MRI) is a clinically used diagnostic tool that features high resolution and unlimited depth penetration. About 40% of MRI procedures use a contrast agent to better distinguish tissues and to improve diagnostics through dynamic imaging procedures that track the contrast agent as it moves through the body. To date, all contrast agents approved by the FDA for contrast enhancement are Gd(III) complexes.^1-3 Recent reports however have raised alarm about residual Gd(III) accumulation in brain, bone and skin of humans who are administered contrast agent.^4-7 Consequently, efforts to produce alternatives to the clinically used Gd(III) contrast agents are underway. Some of these efforts focus on more effective Gd(III) agents that show higher contrast or agents that feature tight binding of Gd(III) through macrocyclic ligands and show a lower extent of dissociation of the Gd(III) ion.⁸ Other reports focus on transition metal ions including Mn(II), Mn(III) or Fe(III) as alternatives to Gd(III) agents.^9-11

Efforts in our laboratories on the development of T₁ MRI probes have focused on Fe(III) complexes containing macrocyclic ligands, especially those containing 1,4,7-triazacyclononane (TACN).^12-16 Such macrocyclic ligands enable exquisite control of oxidation state and spin state as well as the solution chemistry of the iron center.^17-20 To date, the high-spin Fe(III) complexes that we have reported have had hydroxypropyl pendant groups in either mononuclear or dinuclear complexes.^{12, 15, 16} Typically, a single hydroxypropyl group per Fe(III) center is deprotonated to give cationic complexes. The proton r₁ relaxivity of these complexes, especially those that have a bound water, rival those of clinically used gadolinium agents in serum at 4.7 T.¹⁵ An interesting feature of these complexes, however, is that the Fe(III) associated inner-sphere water does not exchange rapidly on the NMR time scale as shown by variable temperature ¹⁷O NMR studies.¹⁶ Instead, second-sphere and outer-sphere water molecules are the basis for their effective relaxivity (r₁) rather than inner-sphere contributions (r₁^SS, r₁^OS, r₁^IS respectively in Eq 1, Scheme 1). Such second-sphere water interactions may be especially important if the waters associate strongly with the complex.²¹ Transition metal complexes with hydroxypropyl pendants promote these strong second-sphere water interactions. For example, complexes with hydroxyalkyl pendants induce paramagnetic shifts in the proton resonances of bulk water that are much larger than analogs with other pendants such as amide groups.²² Another possible mechanism that mediates relaxation in these complexes is exchange of OH protons on water or on ligand donor groups.²³ This is an especially attractive option for MRI probes with hydroxypropyl groups.²⁴

$$r_1=r_1^{IS}+r_1^{ss}+r_1^{OS}$$ Equation 1

Scheme 1.

Mechanisms for promoting water proton relaxation

One type of pendant group that promotes second-sphere water interactions in Gd(III) complexes are phosphonate groups.^{21, 25} Moreover, each phosphonate group is dianionic at neutral pH so that the NOTP ligand produces an anionic Fe(III) complex.^{26, 27} Mice MRI studies with anionic complexes are of interest for comparison to the cationic complexes of Fe(III) with hydroxypropyl groups such as Fe(NOHP) in Scheme 2. Finally, a third common pendant is the carboxylate group to give the well-known NOTA ligand. The Fe(III) complex of this ligand, Fe(NOTA) has a neutral overall charge.^28-30 The aqueous solubility of this neutral complex is sufficient for most solution studies, but is not sufficiently high for imaging studies in animals.

Scheme 2.

Chemical structures of the Fe(III) complexes studied within this paper, drawn in the protonation state at pH 7.4.

Here we compare the three Fe(III) complexes shown in Scheme 2, in order to lay the groundwork for choosing pendant groups that bind Fe(III) to effectively promote proton relaxation in solution. The mechanism of water proton relaxation is studied through variable temperature ¹⁷O NMR spectroscopy, as well as by studies of the pH and temperature dependence of r₁ proton relaxivity of the complexes. Comparison of relaxivity, electrode potential and lipophilicity suggest that the hydroxypropyl and phosphonate pendant groups are effective choices for the Fe(III) center.

2. Experimental

2.1. Instrumentation.

¹H NMR spectra, ¹³C NMR spectra, and ³¹P NMR spectra were acquired using a Varian Mercury 300 MHz NMR spectrometer operating at 300 MHz, 75 MHz, and 121.5 MHz respectively. A Varian Inova 400 MHz spectrometer equipped with a 5 mm broad-band probe operating at a resonance frequency of 54.24 MHz was used for ¹⁷O NMR experiments. All pH measurements were obtained using an Orion 8220BNWP PerpHect ROSS micro glass pH electrode connected to a 702 SM Titrino pH meter. A Thermo Fisher Linear Ion Trap (LTQ) Mass Spectrometer was used to collect all mass spectral data. Absorbance spectra were collected using a Beckman-Coulter DU 800 UV-vis Spectrophotometer equipped with a Peltier temperature controller. Cyclic voltammograms were collected using a Wavenow^XV potentiostat from Pine Research using a glassy carbon working electrode, a Ag/AgCl reference electrode, and a Pt counter electrode, all purchased from Pine Research. T₁ relaxivity values were obtained using a Varian Inova 400 MHz spectrometer as well as a Nanalysis NMReady-60 Benchtop 60 MHz spectrometer. T₁ relaxivity and imaging were performed on a 4.7 Tesla MRI scanner (ParaVision 3.0.2, Bruker Biospin, Billerica MA) with a 35 mm Bruker single channel RF coil. Temperature was maintained at 37 °C during imaging using an MR-compatible heating system (SA Instruments, Stony Brook, NY). Iron concentration of complexes was determined by using Thermo X-Series 2 inductively coupled plasma mass spectrometer (ICP-MS).

2.2. Materials.

1,4,7-triazacyclononane (TACN) and (S)-(−)-propylene oxide were purchased from TCI America. Phosphorous acid, tert-Butyl bromoacetate, ferrous bromide anhydrous, and 65-70% nitric acid with greater than 99.999% purity (trace metals basis) were purchased from BeanTown Chemical. Paraformaldehyde and trifluoroacetic acid were purchased from EMD Millipore Corporation. Human serum albumin, and HEPES were both purchased from Sigma-Aldrich. Ferric chloride anhydrous was purchased from Fisher Scientific and ferric chloride hexahydrate was purchased from Macron Fine Chemicals. 10 ppm Fe and 10 ppm Co standard solutions were purchased from Inorganic Ventures.

2.3. Synthesis

NOHP and Fe(NOHP) were synthesized as previously reported.^{12, 16} NOTA was synthesized by forming the protected tert-butyl-NOTA ligand, followed by deprotection using trifluoroacetic acid as previously reported.³¹ Fe(NOTA) was synthesized as previously reported.²⁸

1,4,7-Triazacyclononane-1,4,7-triyl tris(methylenephosphonic acid) (NOTP).

The NOTP ligand was prepared with procedures similar to those published in literature.³² In particular, recrystallization of the NOTP ligand was modified. Crude NOTP ligand was cooled to room temperature following the reflux in HCl/H₂O, after which it was added dropwise to cold ethanol that was stirring vigorously. The ethanol solution stirred for 30 minutes upon completion of adding the NOTP mixture. The solid was collect via vacuum filtration and then redissolved in 2-3 mL of water, with 5-10 drops of ethanol. This mixture was sonicated with temperature control (65 °C) for up to 20 minutes, during which the ligand precipitated out as a white solid. MS (ESI)⁻ m/z: 410.2 (M - H⁺) (100%), 204.9 (M-2H⁺/2) (20%). ¹H NMR (D₂O, pD 1.0) 3.19 (6H, doublet, J_HP = 11.30) 3.42 (12H, singlet). ³¹P NMR (D₂O) 11.95. ¹³C NMR (D₂O) 51.02 (6C, singlet) 53.03 (3C, doublet, J_CP = 141.61).

Fe(NOTP).

Fe(NOTP) was synthesized using procedures similar to those reported for the Mn(III) complex.³² The NOTP ligand (0.1 mmol) was dissolved in 5 mL of water along with sodium hydroxide pellets (0.3 mmol, 3 equivalences). This solution was heated to 55 °C and pH was checked to ensure it was between 6 and 6.5. Ferric chloride hexahydrate (0.11 mmol, 1.1 equivalences) was dissolved in 5 mL of water and added to the NOTP solution slowly once the temperature had been reached. The mixture stirred for 1.5 hours to ensure all the ligand had complexed and then was cooled to room temperature. The volume was decreased using reduced pressure until 3-4 mL of solvent remained. Absolute ethanol was added to the aqueous solution with stirring and a yellow precipitate formed upon addition of the ethanol. The solid was collected, washed with ethanol, and dried under reduced pressure. Fe content of the solid was determined using ICP-MS calculated for 3Na⁺[Fe(NOTP)]³⁻: 10.57%, Found: 10.61%. MS(ESI)⁻ m/z 462.99 (M – H⁺) (100%), 231.08 (M – 2H⁺) (50%).

2.4. Magnetic Susceptibility.

The effective magnetic moment (μ_eff) for each compound was calculated by running ¹H NMR and using the Evans Method equations (Eq. 2, 3).³³ Samples containing 1-5 mM iron complex in a solution of 5% t-butanol in D₂O were placed in a coaxial NMR insert. The coaxial insert was placed inside of a 5mm NMR tube containing a 5% t-butanol in D₂O solution. All samples were ran on a 300 MHz NMR spectrometer at 298K.

$${\chi }_g=\frac{3\Delta f}{4\pi fm}+\chi 0$$ Equation 2

$${\mu }_{eff}=2.84({\chi }_{m}T{)}^{1∕2}$$ Equation 3

The mass susceptibility (χ_g) was calculated using Equation (2), where Δf is the shift in frequency (Hz), f is the operating frequency of NMR spectrometer (Hz), m is the concentration of the substance (g/mL), and χ₀ is the mass susceptibility of the solvent (χ₀ = −0.6466 × 10⁻⁶ cm³/g). ³⁴ The molar susceptibility (χ_m) is obtained by multiplying the mass susceptibility (χ_g) by the molar mass. This result was used to calculate the effective magnetic moment μ_eff (Equation (3)).

2.5. ICP-MS Measurement.

Iron concentration of each of the complexes was determined using a Thermo X-Series 2 ICP-MS. Samples were dissolved in 65-70% metal free nitric acid for 3 days for digestion. After the digestion was complete, a cobalt internal standard was added and the samples were diluted with Milli-Q water to contain 2% HNO₃ and 50 ppb cobalt. A linear calibration curve ranging from 0.1 ppb to 100 ppb iron was prepared as well and used for quantification of the samples. Quantification and data analysis was performed using Thermo Fisher PlasmaLab.

2.6. Monitoring formation of iron complex.

Three aqueous solutions at variable pH values (3, 5, and 6.5) containing 1 mM FeCl₃•6H₂O and either 1 mM of NOHP ligand or 1 mM NOTP ligand were prepared and allowed to stir at room temperature for up to three weeks. Aliquots of each NOTP solution were taken 5 and 15 minutes after mixing of the two solutions. Aliquots of each NOHP solution were taken on days 1-7, as well as at 2 and 3 weeks. These aliquots were injected in a LTQ Mass Spectrometer to monitor formation of the iron complex in water.

2.7. UV-Vis spectroscopy.

Kinetic inertness was tested over the course of 72 hours by measuring the absorbance of the complexes from 200-800 nm. For all conditions tested a temperature of 37 °C was maintained using a Peltier temperature controller. Solutions contained 125 μM or 150 μM iron for Fe(NOTP) samples or 70 μM iron for Fe(NOTA) samples. For testing the kinetic inertness of the iron complexes in the presence of biologically relevant anions, solutions contained 25 mM NaHCO₃, 0.5 mM Na₂HPO₄, and 20 mM HEPES buffer at pH 7.4. For kinetic inertness studies done in acidic conditions, either 0.1 mM or 1 M HCl was used. For transmetalation studies, a molar equivalence of zinc chloride was used along with 20 mM HEPES (pH 7.4).

2.8. Log P measurements.

Solutions of the Fe(III) complexes were made at 1 mM concentration, and also contained 20 mM HEPES buffer (pH 7.2) and 100 mM NaCl. 0.5 mL of the Fe(III) solution was added to a 0.5 mL of 1-octanol and the mixture was shaken for 24 hours. After 24 hours, the solution was centrifuged at 13,000 rpm for 1 minute and then allowed to stand for 1 hour. ICP-MS was used to determine the iron concentration in the aqueous layer before and after mixing with the 1-octanol. The partition coefficient was calculated using Equation 4, where Co is the concentration of iron in the octanol layer and Cw is the concentration of iron in the aqueous layer.

$$logP=log\left(\frac{Co}{Cw}\right)$$ Equation 4

2.9. Cyclic Voltammetry.

Solutions containing 5 mM Fe(NOTP), 5 mM acetate buffer (pH 4), 5 mM HEPES buffer (pH 7.4) and 1 M KCl as a supporting electrolyte were used to perform cyclic voltammetry measurements. A standard three electrode cell consisting of a glassy carbon working electrode, a Ag/AgCl reference electrode, and a Pt wire counter electrode were used. Solutions were purged with N₂ gas before each measurement to remove any dissolved oxygen in the system. Current was measured from −1.5 V to 1.5 V with a 10 s prescan delay at a sweep rate of 100 mV/s. A standard of 5 mM ferrocyanide was used. Redox potential of Fe(II)/Fe(III) was calculated from the average of the anodic (E_pa) and the cathodic (E_pc) potentials in reference to Ag/AgCl and then converted to the normal hydrogen electrode (NHE).

2.10. T₁/T₂ Proton Relaxation Measurements.

T₁ and T₂ measurements were measured at 1.4 T (60 MHz NMR), 4.7 T (Small animal MRI), and 9.7 T (400 MHz NMR). Experiments performed at 1.4 T were done at 33 °C and those performed at 4.7 T and 9.7 T were done at 37 °C. T₁ and T₂ experiment solutions contained 20 mM HEPES buffer (pH 7.4) and 100 mM NaCl. Concentrations of 0.050 mM to 0.40 mM were used for the 1.4 T and 4.7 T experiments while a range of 0.40 mM to 1.00 mM were used for the experiments at 9.7 T. Studies where HSA was present, contained 35 mg/mL of HSA. An inversion-recovery True FISP acquisition was used to measure T₁ relaxation rate constants. T₂ relaxation rates were measured by using multi-echo, Carr-Purcell-Meiboom-Gill spin-echo sequence with a fixed TR of 3000 ms and TE times ranging from 20-1200 ms. The T₁ and T₂ relaxivity values were calculated by using linear regression fitting of 1/T₁ (s⁻¹) and 1/T₂ (s⁻¹) versus concentration (mM) in GraphPad Prism 9.0.2.

The pH dependence of T₁ water proton relaxivity values were measured at 1.4 T at 33 °C with solutions containing 0.10 mM to 0.40 mM iron complex, 10 mM acetate buffer (pH 4.0), 10 mM HEPES buffer (pH 7.4), and 100 mM NaCl. pH values ranged from 0.5 to 10. The temperature dependence of T₁ relaxivity values were measured at 9.7 T over a range of 25 °C to 80 °C, at 5 °C intervals. Solutions contained 0.40 mM to 1.00 mM iron complex, 20 mM HEPES buffer (pH 7.4), and 100 mM NaCl. T₁ relaxivity values were determined as mentioned above.

2.11. Variable Temperature ¹⁷O NMR Spectroscopy.

Solutions containing 10 mM iron complex, 20 mM HEPES buffer (pH 7.4), and 100 mM NaCl were prepared in a 1% H₂¹⁷O water and studied from 20 °C to 80 °C at intervals of 5 °C. 1/T₂ was calculated by determining the full width at half maximum (FWHM) of the ¹⁷O resonance with iron complex and subtracting that of the ¹⁷O resonance without iron complex.^{12, 16, 35, 36}

2.12. Mouse Imaging.

In vivo imaging studies with Fe(NOTP) in healthy mice (BABC/cJ, Jackson Laboratory) were performed on a 4.7 Tesla Bruker preclinical MRI in accordance with approved IACUC protocols. Sealed NMR tubes filled with 1% agarose, doped with concentrations of 0 mM and 1 mM CuSO₄, were included to serve as signal normalization phantoms. Compounds were prepared with either 10 mM or 20 mM Fe(NOTP), 20 mM HEPES buffer (pH 7.4), and 100 mM NaCl and injected intravenously via the tail vein at a dose of either 100 μmol/kg or 200 μmol/kg. MRI data was collected up to 45 minutes post injection using T₁-weighted, 3D, spoiled-gradient echo scans (TE/TR/FA = 3/15/40°, duration = 2.5min) covering the mouse from thorax to tail. Signal intensities were normalized using the phantoms’ intensities, and signal changes in each organ were calculated from pre-injection values.

3. Results and Discussion

3.1. Fe(III) complex synthesis

The three complexes were prepared by different methods. Fe(NOTP) and Fe(NOTA) were prepared by treatment of their respective ligands in water with FeCl₃•6H₂O followed by adjustment of the pH to 6. The successful formation of the complex in water upon incubation of ligand with trivalent iron salts is indicative of the high stability of the complexes in aqueous solution. For example, Fe(NOTA) has a stability constant of log K = 28.3,^{29, 30} whereas Fe(NOTP) has a stability constant of log K = 29.6.²⁷ Notably, slow kinetics of complexation as well as small extent formation of complex in aqueous solution is often a complicating factor in the synthesis of Fe(III) complexes. For example, the Fe(NOHP) complex only forms partially when NOHP is treated with FeCl₃ in water (Figure S7) at neutral pH. A larger degree formation is observed at acidic pH values of 3.0, although only partial formation is observed even after three weeks (Figure S5). In fact, Fe(NOHP) is best prepared by addition of ferrous chloride in ethanolic solution followed by oxidation of the complex in solutions that are open to the air.¹² This method of preparation relies on the propensity of Fe(II) to bind to the NOHP ligand in organic solvent and the ease of oxidation of the iron center once it is bound to the ligand. Moreover, the elimination of water as a solvent prevents the formation metal hydroxide species which is problematic for divalent, but especially trivalent metal ions. In comparison, studies showed that Fe(NOTP) formed readily in solutions over several minutes when incubated in water in millimolar concentrations as shown by mass spectrometry (Figure S8).

Once formed, all three macrocyclic complexes of Fe(III) exhibit a high degree of kinetic inertness towards dissociation in 0.1 mM acid, towards trans-metallation with Zn(II) or to binding with phosphate or carbonate. Moreover, Fe(NOTP) is inert to dissociation at 37 °C even in 1 M HCl. Such a large degree of kinetic inertness towards loss of metal ion has been shown to be one of the most important characteristics in the design of MRI probes.^{37, 38}

3.2. Fe(III) complex characterization

Fe(NOTP), Fe(NOHP) and Fe(NOTA) have effective magnetic moments of 5.85, 5.81, and 5.50, respectively. These values are characteristic of high spin Fe(III) complexes. Notably, high spin Fe(II) centers in TACN ligands often have magnetic moments that are as high or higher than those of Fe(III) due to orbital contributions to magnetic susceptibility in Fe(II) complexes.^{20, 39, 40} To further distinguish between high spin Fe(III) and Fe(II), ¹H NMR spectra were recorded. The proton NMR spectrum of Fe(NOTP) was broadened into the baseline, consistent with the broadening observed for a high spin Fe(III) center (Figure S4). Moreover, the r₁ relaxivity values of the complexes are also consistent with high spin Fe(III) as discussed further below. X-ray crystal structures of Fe(NOHP), Fe(NOTP) and Fe(NOTA) show that the Fe(III) centers are six-coordinate with distorted octahedral geometry.^{12, 28, 41}

Previous studies of the protonation states of these complexes by pH potentiometric titrations or by spectroscopic titrations suggest that the pendant groups are fully ionized for Fe(NOTA) or Fe(NOTP),^{26, 29, 30} or partially ionized for Fe(NOHP) at neutral pH.¹² Thus at neutral pH, Fe(NOTA) is neutral, Fe(NOHP) is dicationic and Fe(NOTP) is trianionic. The two charged complexes, Fe(NOTP) and Fe(NOHP) are soluble in water in millimolar concentrations (25 and 6 mM, respectively) whereas the neutral Fe(NOTA) complex is less soluble (2 mM). Solubility considerations are especially important for mice MRI studies that typically require a minimum of ≈5 mM solutions to deliver doses of 50 μmol/kg. Other solution methods such as variable temperature ¹⁷O NMR also require high concentrations of complex.

For Fe(III) MRI probes, it is important that the trivalent state of iron is stabilized. To maintain Fe(III) in the blood pool, the electrode potential should be more negative than approximately −180 mV versus NHE.⁴² Both Fe(NOTP) and Fe(NOHP) meet this criterion at neutral pH values. Fe(NOHP) has an electrode potential of −330 mV.¹² Fe(NOTP) has an electrode potential of −333 mV at neutral pH as shown in Figures S13 and Table S2. As noted previously, the electrode potential of Fe(NOTP) changes markedly with pH.²⁶ As the phosphonate groups become deprotonated the electrode potential shifts to more negative values, consistent with a greater stabilization of the Fe(III) center with the anionic phosphonate groups. By contrast, Fe(NOTA) has a moderately positive electrode potential (195 mV vs. NHE), which is a further disadvantage for this complex as an MRI probe.⁴³

3.4. Fe(III) complex dissociation studies

Kinetic inertness of metal complexes towards dissociation is one of the most important characteristics of complexes that are under development as contrast agents.³⁸ With macrocyclic ligands it is indeed possible to form complexes that are kinetically inert towards metal ion release, but yet are not thermodynamically stable in aqueous solution. Fe(NOHP) is in this category. It is inert towards dissociation in 100 mM acid or in the presence of 25 mM carbonate and 0.5 mM phosphate at 37 °C.¹² Yet, as discussed above, the extent complexation of Fe(III) to the NOHP ligand is small over the pH range of 3 to 7 (Figures S5-S7). Both Fe(NOTA) and Fe(NOTP) are stable in aqueous solution and are also inert towards metal ion release in strong acid (0.1 mM HCl), or with competing anions over 72 hours at 37 °C as studied by electronic spectroscopy (Figures 1 & S8-S11). Fe(NOTP) shows no dissociation under harsh conditions of 1.00 M HCl at 37 °C, over 72 hours as well.

Figure 1.

UV-vis spectroscopy studies of Fe(NOTP) in A) 0.1 mM acid (left) or in B) 25 mM carbonate and 0.5 mM phosphate at pH 7.4, 37 °C (right) showing no detectable dissociation.

3.5. Variable temperature ¹⁷O NMR studies.

In the solid state, Fe(NOHP), Fe(NOTP) and Fe(NOTA) have six donor groups of the macrocycle bound to give distorted octahedral complexes.^{12, 28, 41} There are no coordination sites for solvent. However, it is not safe to assume that these complexes have the same structure in solution as in crystalline form. The most common method to study whether a paramagnetic metal ion complex has a bound water is through ¹⁷O NMR spectroscopy.^44-46 An inner-sphere water that is rapidly exchanging with a paramagnetic metal ion such as Gd(III) on the NMR time scale gives rise to substantial broadening of the ¹⁷O water resonance, whereas complexes that lack an inner-sphere water or complexes that have a very slowly exchanging water do not produce substantial line broadening. Variable temperature studies allow for determination of the rate constants for exchange of inner-sphere water ligands of the paramagnetic complex with bulk water.^{35, 36, 45} Other methods, such as those developed for Mn(II) complexes, have employed ¹⁷O water resonance line broadening (transverse r₂ relaxivity) studies as a function of metal complex concentration and temperature to estimate the number of inner-sphere waters.⁴⁷ We have used an approach for our Fe(III) complexes that is similar to the analysis of Mn(II) complexes.^{12, 15, 16} We compare the transverse relaxation rate constants (1/T₂^O or R₂^O) estimated from ¹⁷O line broadening for our complexes to standard Fe(III) complexes that have a bound water, such as Fe(CDTA), and ones that lack a water such as Fe(DTPA).³⁵ The comparison aids in the assessment of whether there is an inner-sphere water that undergoes exchange on the ¹⁷O NMR time scale.

The Fe(NOTP) complex showed transverse r₂^O relaxivity (R₂^O normalized to concentration) that is similar to that of Fe(DTPA) and much lower than that of Fe(CDTA) as shown in Figure 2. This data supports the lack of an exchangeable inner-sphere water ligand on the NMR timescale at neutral pH values. However at acidic pH values of 3.0 and 0.5, the ¹⁷O resonance becomes noticeably broader, although it only reaches about 10% (pH 3.0) or 30% (pH 0.5) of that of the Fe(CDTA) (Figure 2B). The variable temperature ¹⁷O NMR studies at acidic pH are difficult to interpret as a small degree of ¹⁷O line broadening in Fe(NOTP) could arise from strong second-sphere water interactions or from partial dechelation to give an inner-sphere complex. Similarly, Fe(NOHP) produces little ¹⁷O line broadening as discussed in our earlier studies and shown in Figure 2.¹² Fe(NOTA) was not sufficiently soluble for these studies. These results suggest that the Fe(NOTP) and Fe(NOHP) complexes mediate proton relaxation through either second-sphere water or through proton exchange processes, although we cannot rule out partial dechelation at the low pH values. Water proton relaxivity studies over a range of magnetic field strengths, temperatures and pH values were conducted as presented in Table 1 and Figures 3-4 to gain further insight into the mechanism of proton relaxation by these complexes.

Figure 2.

Comparison of ¹⁷O-NMR transverse relaxivity (r₂^O) for A) Fe(NOTP), Fe(NOHP), Fe(DTPA), Fe(CDTA) at pH 6.5-7.2 as a function of temperature and B) Fe(NOTP) at variable pH values compared to Fe(DTPA).

Table 1:

Relaxivity of Fe(III) complexes at pH 7.4 in 0.1 M NaCl

	r1 (mM−1s−1) 1.4 Ta	r1 (mM−1s− 1) 4.7 Tb	r1 (mM−1s− 1) 9.4 Ta	r2 (mM−1s− 1) 4.7 Tb	r1 (mM−1s− 1) HSA 1.4 Ta	r1 (mM−1s− 1) HSA 4.7 Tb	r2 (mM−1s− 1) HSA 4.7 Tb
Fe(NOTP)	1.04 ± 0.05	0.72 ± 0.02	0.86 ± 0.03	1.26 ± 0.12	1.26 ± 0.08	1.05 ± 0.01	1.55 ± 0.01
Fe(NOTA)	0.61 ± 0.01	N/A	0.58 ± 0.07	N/A	0.65 ± 0.03	N/A	N/A
Fe(NOHP) C	1.50 ± 0.15	0.97 ± 0.12	1.4 ± 0.08	1.8 ± 0.47	1.52 ± 0.13	1.2 ± 0.22	2.3 ± 1.1

^a.At 33 °C.

^b.at 37 °C.

^c.values at 4.7 and 9.4 T from reference¹² and have meglumine as an additive

Figure 3.

Temperature dependence of the r₁ relaxivity for the Fe(III) complexes at 9.4 T.

Figure 4.

pH dependence of r₁ proton relaxivity of the three Fe(III) complexes at 1.4 T, 33 °C. Data is fit to Equation 7 with values given in Table S3. Schematic for mode of acid catalyzed proton catalyzed exchange.

3.6. Fe(III) complex proton relaxivity measurements.

The fluctuating magnetic dipole of the Fe(III) center mediates the enhanced water proton relaxation rates. The important components arise from rotational diffusion (τ_R), the relaxation of electrons in Fe(III) center (T_1e) and also the binding and dissociation of the water molecules around the Fe(III) complex with an associated lifetime (τ_m).^{2, 48} The correlation time (τ_c) will be dictated by the most rapid process that dominates the fluctuation of the magnetic dipole (Eq. 5).

$$\frac{1}{{\tau }_c}=\frac{1}{{\tau }_R}+\frac{1}{T_{1e}}+\frac{1}{{\tau }_m}$$ Equation 5

The inner-sphere relaxivity (r₁^IS) and second-sphere relaxivity (r₁^SS) are related to the number of water molecules involved, the lifetime of the inner-sphere or second-sphere water interaction with the paramagnetic center (τ_m) as well as the relaxation time of the bound water (T_1m) as shown in Equation 6. The outer-sphere contribution (r₁^OS) arises from water molecules that do not have a specific lifetime and interaction associated with the contrast agent. Inner-sphere waters are relatively well defined from the standpoint of numbers and distance and orientation with respect to the metal center, whereas second-sphere waters (q’ and τ_m’) are more difficult to assess.^{21, 49} Our proton relaxivity measurements are discussed with these relationships in mind.

$$r_1=r_1^{IS}+r_1^{ss}+r_1^{OS}=\frac{q∕[H_{2}O]}{T_{1m}+{\tau }_m}+\frac{q^{\prime }∕[H_{2}O]}{T_{1m}^{\prime }+{\tau }_m^{\prime }}+r_1^{OS}$$ Equation 6

Shown in Table 1 are r₁ and r₂ proton relaxivity values for the three Fe(III) complexes at several magnetic field strengths. Measurements were made at pH 7.2 by monitoring the T₁ or T₂ water proton relaxation times of solutions containing the complex over the concentration range of 50 μM to 1.00 mM for the Fe(III) complex. At neutral pH, the Fe(NOHP) complex shows the highest relaxivity and Fe(NOTP) the second highest at all field strengths in simple solutions and in the presence of human serum albumin (HSA). Fe(NOTP) shows an increase in r₁ relaxivity by about 46% at 4.7 T in the presence of 0.6 mM HSA, consistent with binding to this blood protein. Binding to HSA is expected to increase the rotational correlation time by an effective size increase (Eq. 5). This increase is expected to improve the r₁ relaxivity as rotational correlation times are limiting for small molecule contrast agents at intermediate field strengths.⁵⁰ The other two complexes, Fe(NOHP) and Fe(NOTA) showed no change upon addition of HSA, consistent with a little binding to the serum protein.

The unusual field strength dependence of the r₁ proton relaxivity of Fe(NOTP) and Fe(NOHP) warrants discussion. Previous studies on macrocyclic complexes of Fe(III) with hydroxypropyl groups and a bound water show an increase in relaxivity of about 20-30% on going from 1.4 T to 4.7 T and very little change at higher field strengths of 9.4 T. This feature was attributed to limiting effects of Fe(III) electronic relaxation times (T_1e) as given in equation 5. It has been proposed that the T_1e times of Fe(III) are too fast for optimal relaxation at low field strengths, but are better optimized at high field strengths.⁵¹ In contrast, Gd(III) and Mn(II) complexes typically show decreasing r₁ relaxivity over this magnetic field strength range.^{1, 2, 37}

For the complexes studied here, Fe(NOTP) and Fe(NOHP) show slightly higher relaxivity at 1.4 T, a decrease at 4.7 T and a slight increase at 9.4 T. The changes are not large (23% and 35%) for Fe(NOTP) and Fe(NOHP), respectively, but are outside of experimental error. By comparison, Fe(NOTA) shows little change at these different magnetic field strengths. Measurements are complicated by the fact that relaxivity studies at 1.4 T were made at a lower temperature (33 °C) than studies at 4.7 T or 9.4 T (37 °C). Relaxivity values typically increase with decreasing temperatures for paramagnetic complexes, and this decrease will factor into our trend.²³ Full temperature dependence of the proton r₁ relaxation rate constants were measured for these three complexes at 9.4 T to investigate trends and to better characterize the mechanism of proton relaxation by these complexes.

The temperature dependence of the r₁ relaxation rate constants over the range of 20 to 80 °C is shown in Figure 3. As anticipated, proton relaxivity decreases with increasing temperature.³⁷ Both T_1e and rotational times are expected to increase as temperature is lowered to produce an increase in relaxivity. However, the exchange of bound water molecules often has the opposite effect and will reduce relaxivity as the temperature decreases if exchange is in the intermediate regime (τ_m > T_1m) where T_1m is the longitudinal relaxation time of the bound water (Eq. 6). The resulting water exchange dependence on temperature may result in a flattening of the curve that describes the dependence of r₁ on temperature. For the Fe(III) complexes here, the steep decrease in r₁ with temperature supports second-sphere or outer-sphere water exchange contributions to proton relaxivity. However, there are slight differences for the three complexes. Fe(NOHP) shows a greater temperature dependent relaxivity change (3-fold) than Fe(NOTP) (2.4-fold). Fe(NOTA) shows the smallest change over this temperature range of about 2-fold. For these complexes, it is unlikely that second-sphere water exchange rates are limiting. Rather, small differences in T_1e variation with temperature are most likely responsible for the temperature dependence.⁴⁹

Next, the pH dependence of r₁ relaxivity was studied for all three complexes as shown in Figure 4. We wondered whether relaxivity through proton exchange processes rather than second-sphere or outer-sphere water exchange would be important for these complexes, especially for the Fe(NOHP) complex which has 2-3 hydroxyl protons over this pH range. Gd(III) complexes with an hydroxypropyl pendant demonstrate a substantial proton exchange contribution to relaxivity that is greater than the whole water exchange contribution at basic pH values. ^{24, 52} In these complexes, there is an increase in relaxivity at basic pH as the hydroxyl proton undergoes base catalyzed exchange.^{24, 53} In some cases for Gd(III) contrast agents, acid catalyzed exchange of protons of water has also been observed.^{23, 54}

For the three Fe(III) complexes studied here, the pH profile for the r₁ relaxivity is fairly flat over the pH range of 3 to 10. This profile in combination with the ¹⁷O NMR studies suggests that second-sphere or outer-sphere water exchange is the dominant contribution to relaxivity in this pH range. However, there is an increase in relaxivity at very low pH values for all three complexes. Fe(NOHP) proton relaxivity increases by about 50% from pH 4 to 0.5 whereas Fe(NOTA) increases by about 30% from pH 2 to 0.5. For Fe(NOHP), the moderate increase is likely due to acid catalyzed exchange of hydroxyl protons with water. For Fe(NOTA), there is little information about speciation at such a low pH, but an increase might be attributed to protonation of a carboxylate to increase proton exchange or even partial dechelation, in the absence of ¹⁷O NMR data. The largest effect, however, is observed for Fe(NOTP) where relaxivity increases by 2.5-fold on going from pH 3 to 0.5. This increase is most consistent with protonation of the phosphonate groups to form a species that is more effective as a relaxation agent. There are several pK_a values for the phosphonate groups as there are potentially six protonations. Previous studies by UV-vis spectroscopy and electrochemistry suggest that there are several protonation steps between pH 4 and 0.5.^{26, 27} The protonation of more than one phosphonate may contribute to the steep dependence of the relaxivity on pH at very low pH values as observed here. Protonation of the pendant group followed by dechelation of one of the phosphonate groups seems unlikely given that the complex is inert in 1 M acid over a period of days. However, the changes in the ¹⁷O NMR line broadening at pH 0.5 do raise the possibility of partial dechelation that might contribute to the r₁ relaxivity pH profile.

All three pH dependent relaxivity curves can be fit to an simple expression for an acid-catalyzed relaxivity process (r_1H) and for a pH independent process (r_1O) (Eq. 7). Similar pH dependences have been observed for polyaminocarboxylate Gd(III) complexes that have a bound water.⁵⁴

$$r_1=r_{1o}+r_{1H}[H^{+}]$$ Equation 7

For Gd(III) complexes that have relatively well-defined numbers of exchangeable water ligands, it is feasible to further dissect contributions from proton and water lifetimes and relaxation processes.²³ The Fe(III) complexes studied here promote proton relaxivity primarily through interactions with an unknown number of second-sphere and outer-sphere waters molecules. Thus, data is fit to a simple pH dependent relationship (Eq. 7) rather than a more complicated expression that requires knowledge of the number of exchangeable water ligands.

3.7. MRI studies of Fe(NOTP) in mice.

Fe(NOTP) was studied as a contrast agent in mice to test whether this anionic complex would behave as an extracellular fluid contrast agent. Most Gd(III) contrast agents used in the clinic are small hydrophilic complexes that act as extracellular fluid agents by distributing between vasculature and extracellular space and undergoing rapid elimination by a renal pathway.¹ To further characterize Fe(NOTP) in this regard, we measured the log P octanol value and found that it was characteristic of a hydrophilic complex (−2.02 ± 0.1). By comparison, Fe(NOHP) has a log P of −1.60 ± 0.1.¹²

Healthy Balb/C mice were injected with two different doses of Fe(NOTP) to study the biodistribution of this complex and its clearance. These studies show that the Fe(NOTP) complex indeed behaves as an extracellular agent with clearance predominantly through the renal pathway as shown by the enhanced kidney and bladder contrast. The decrease in the signal through the vena cava resembled that of Dotarem (Gd(DOTA)) (Figure S15), albeit at lower dosage. The lower r₁ relaxivity of Fe(NOTP) in solution correlates with the lower signal observed in comparison to Gd(DOTA) and to other Fe(III) macrocyclic complexes we have studied.^{15, 16} We note that renal clearance of Fe(NOTP) appears to be slower than that of Dotarem (Figure S15) in mice. Further studies on additional replicates in mice and larger animals will be of interest to determine whether this trend is of concern.

4. Conclusions

Challenges in the design of Fe(III) complexes as T₁ contrast agents include the stabilization of iron in the trivalent, high spin state, the formation of robust complexes that are inert towards dissociation, and the development of complexes with effective proton relaxivity and high aqueous solubility. TACN macrocycles with oxygen containing donor groups stabilize Fe(III) in a high spin state and form complexes that are highly kinetically inert towards Fe(III) dissociation. Both Fe(NOHP) and Fe(NOTP) are sufficiently soluble for studies in mice, however, the Fe(NOTP) complex has the best solubility (25 mM) at neutral pH.

In this study, closed coordination complexes were examined. All three complexes lack an inner-sphere water molecule as supported by variable temperature ¹⁷O and ¹H NMR studies. The pH dependence of r₁ relaxivity is most consistent with a pH-independent pathway involving second-sphere water interactions as the most important contribution at pH 4-10. That the Fe(III) complex with hydroxypropyl pendants has the highest relaxivity suggests that the hydroxyl groups are effective in second-sphere interactions. Hydroxyl groups form hydrogen bonding interactions with water and other small molecules, and this may promote stronger second-sphere interactions.^{22, 55} Fe(NOTP) is less effective as a relaxivity agent than is Fe(NOHP), but more effective than Fe(NOTA). The higher relaxivity of Fe(NOTP) may be attributed to phosphonate group interactions with second-sphere water, as observed previously for Gd(III) complexes.^{25, 56} Further efforts to increase relaxivity might capitalize on linking two macrocycles together as has been shown recently¹⁶ or by removing a pendant group to form an inner-sphere water.^{12, 15} Fe(III) complexes with bound water ligands do indeed have higher relaxivity, but care must be taken to control dimerization⁵⁷ and strong interactions with other molecules including biopolymers.¹³

Figure 5.

Changes in T₁ weighted signal intensities within the (A) blood, (B) kidneys, (C) liver, and (D) bladder of Balb/C mice dosed with Fe(NOTP) at 100 or 200 μmol/kg.