High Relaxivity Confined to a Small Molecular Space: A Metallostar‐Based, Potential MRI Contrast Agent

Abstract

Good things come in small packages: For its moderate molecular weight, the metallostar complex [Fe{Gd2L(H2O)4}3]4− (see picture; L is a bipyridine–poly(aminocarboxylate) derivative) displays exceptionally high proton relaxivity, which is explained in terms of a rigid supramolecular structure, two inner-sphere water molecules with a near-optimal exchange rate, and six efficiently relaxing GdIII centers confined to a small molecular space. The 2003 Nobel Prize in medicine was awarded for discoveries concerning magnetic resonance imaging (MRI), a technique that has revolutionalized medical diagnostics over the last two decades.1 Such spectacular developments would not have been possible without the use of paramagnetic contrast agents, the majority of which comprise GdIII complexes.2 By reducing the relaxation time of the surrounding water protons, these agents enhance the intrinsic contrast of magnetic resonance images. The efficacy of a contrast agent is expressed by its relaxivity, r1, which is the enhancement of the longitudinal proton relaxation rate induced by the paramagnetic agent, referred to a concentration of Gd3+ of 1 mmol L−1 (molar relaxivity per Gd). Theory predicts high relaxivity for a GdIII complex when its rotation and electron spin relaxation are slow and when the rate of water exchange between the inner sphere and the bulk solvent is optimal. Novel applications in MRI call for very highly efficient contrast agents. It implies not only high molar relaxivity per GdIII ion but high relaxivity per molecular volume/mass. For example, in cell imaging, biological constraints limit the amount of contrast agent that can be delivered into one cell without destroying it. Consequently, agents with many efficiently relaxing paramagnetic centers confined into a small space are advantageous over large macromolecules with few GdIII centers. Recently we started to exploit heterometallic assemblies in the aim of confining high relaxivity into a small molecular space. We reported a terpyridine (tpy) ligand functionalized with a poly(aminocarboxylate) group DTTA (diethylenetriamine-N,N,N′′,N′′-tetraacetate), (tpy-DTTA)4−, which has distinct binding sites for FeII and GdIII ions.3 In aqueous solution and in the presence of these metal ions, the ligand undergoes self-assembly to form a rigid supramolecular structure, [FeII(tpy-DTTA)2Gd2]. Prior to our work, Desreux and co-workers reported a [Gd(DO3A)(phen)] derivative (DO3A=1,4,7-tri(carboxymethyl)-1,4,7,10-tetraazacyclododecane, phen=1,10-phenanthroline) that self-assembles with FeII ions to yield a tetranuclear FeGd3 entity.4 Here we report a novel heterotritopic ligand, H8L, which comprises a 2,2′-bipyridine moiety for specific binding to FeII ions and two poly(aminocarboxylate) groups for binding to GdIII ions (Scheme 1). The ligand self-assembles with FeII and GdIII into a metallostar [Fe{Gd2L(H2O)4}3]4− structure. The rational design of H8L consists of several key features: 1) The appropriate choice of the poly(aminocarboxylate) group, which has the same chelating unit as the previously described TTAHA (N-tris(2-aminoethyl)amine-N′,N′,N′′,N′′,N′′′,N′′′-hexaacetate) ligand,5 ensures sufficient thermodynamic stability for the GdIII complex, an important factor for in vivo safety. It also guarantees faster water exchange than that for commercial agents, thus closer to optimal. The GdIII complex has two inner-sphere water molecules to double the inner-sphere contribution to the relaxivity. Indeed, preliminary 17O NMR spectroscopy experiments on the metallostar confirmed the presence of two inner-sphere water molecules and point to a water-exchange rate which is close to that previously published for [Gd(TTAHA)(H2O)2]3−.5 A complete, variable-temperature 17O NMR spectroscopic study is in progress and will be reported in due course. 2) FeII ions can accommodate three strongly coordinating 2,2′-bipyridine units which increases the number of GdIII centers on one iron core to six. The accumulation of six GdIII ions in a small space represents a clear advantage over [FeII(tpy-DTTA)2Gd2]. The negative charge of [Fe{Gd2L(H2O)4}3]4− is also favorable for high solubility in water. 3) The linking between the FeII and GdIII binding sites is designed to minimize internal flexibility that could reduce the relaxivity gained by the increased molecular size. All of these features contribute to a high relaxivity per GdIII center which, for the metallostar with six GdIII centers, should sum up to a remarkable relaxivity confined to a small space. Reagents and conditions: a) protected triamine, K2CO3, acetonitrile; b) HCl (6 m), followed by flash chromatography; c) tert-butyl bromoacetate, DIEA, KI, DMF; d) HCl (6 m), followed by ion-exchange chromatography. DIEA=N,N′-diisopropylethylamine, DMF=N,N-dimethylformamide. The initial step in the synthesis of the ligand is the nucleophilic substitution of 4,4′-bis(bromomethyl)-2,2′bipyridine with two protected diethylenetriamine units (Scheme 1). The following carboxymethylation, ester hydrolysis, and ion-exchange chromatography steps are all straightforward. The ligand, which was obtained as a highly water-soluble and partially protonated ammonium salt (H4.5(NH4)3.5L), was characterized by 1H NMR spectroscopy, ESI-MS, and elemental analysis.6 The [Gd2L(H2O)4]2− complex is prepared by mixing the ligand and Gd3+ ions in a 1:2 molar ratio at pH 6. The metallostar [Fe{Gd2L(H2O)4}3]4− (Figure 1) forms instantaneously upon addition of Fe2+ to a solution of [Gd2L(H2O)4]2− (convergent approach), and the formation of the complex is indicated by the appearance of an intense red color. Framework molecular model of [Fe{Gd2L(H2O)4}3]4−. Hydrogen atoms have been removed for clarity. The GdIII chelates, which are bound through a short spacer attached to the para position of the pyridyl moieties, form an octahedron around the central FeII ion that essentially reproduces the polyhedron of the bipyridine nitrogen atoms. The dinuclear [Gd2L(H2O)4]2− species exhibits a fairly high relaxivity (12.5 mM−1 s−1) relative to dinuclear complexes such as [pip{Gd(DO3A)(H2O)}2] (pip=piperidine) or [bisoxa{Gd(DO3A)(H2O)}2] (5.58 and 4.43 mM−1 s−1, respectively; 40 MHz, 37 °C).7 This difference is explained by the two inner-sphere water molecules and their faster exchange, as well as the limited internal flexibility of [Gd2L(H2O)4]2−. The formation of the large and rigid [Fe{Gd2L(H2O)4}3]4− metallostar from [Gd2L(H2O)4]2− increases the rotational correlation time, τR, whereas the water-exchange rate remains unchanged. The longer τR value results in a substantial relaxivity gain. Consequently, the formation of the metallostar can be followed by relaxometry. Upon adding increasing amounts of FeII ions to a solution of [Gd2L(H2O)4]2−, the relaxivity continuously increases until a molar ratio of FeII/[Gd2L(H2O)4]2− of 1:3, after which it remains constant (Figure 2). The sharp break in the relaxometric titration curve indicates a 1:3 stoichiometry of the metallostar as expected for an FeII–2,2′-bipyridine complex. It proves that the pending GdIII–poly(aminocarboxylate) units do not sterically hinder the binding of three bipyridine units per FeII ion. Molecular modeling studies also showed that the metallostar is sterically not overcrowded (Figure 1). The invariance of the relaxivity at molar ratios of FeII/[Gd2L(H2O)4]2− higher than 1:3 points to the exclusive formation of [Fe{Gd2L(H2O)4}3]4−. It also shows that FeII is not oxidized to FeIII which would likely replace GdIII in the poly(aminocarboxylate) and the consequent release of free Gd3+ would decrease the relaxivity (r1=11.2 mM−1 s−1 for Gd(H2O) at 30 MHz and 25 °C). Relaxivity titration of [Gd2L(H2O)4]2− with FeII at pH 6.0, 25 °C, ν=30 MHz, and cGd=1 mM. Inset: UV/Vis spectra of the samples used for the relaxivity titration are shown. The absorption increases with an increasing ratio of Fe/[Gd2L(H2O)4]2−. The formation of the metallostar was also monitored by UV/Vis spectrophotometry. Upon addition of FeII ions to a solution of [Gd2L(H2O)4]2−, new absorption bands appear at 320–640 nm. Their intensity increases until a molar ratio of 1:3 FeII/[Gd2L(H2O)4]2−, then it remains constant (the small further increase is due to the presence of free FeII ions; Figure 2). Water proton relaxivities were measured for both the dinuclear [Gd2L(H2O)4]2− complex and the metallostar [Fe{Gd2L(H2O)4}3]4− (25 °C; 0.01–600 MHz; Figure 3). Such nuclear magnetic relaxation dispersion (NMRD) profiles are widely used for the characterization of MRI contrast agents. As the consequence of slower tumbling, the relaxivities at almost all frequencies are considerably higher for the metallostar than for the dinuclear complex, and a maximum increase is observed at 30–60 MHz. The profile of [Fe{Gd2L(H2O)4}3]4− shows the highfield peak typical of slowly rotating complexes (r1 max=33.6 mM−1 s−1 at 40 MHz). The metallostar is rather small compared to usual macromolecular contrast agents. However, the concurrent chelation of the multifunctional ligand to FeII and GdIII ions forces the rotational correlation time of the GdIII–water-proton vector, which is important for relaxivity, to be close to that of the entire assembly. Thus, by strongly reducing internal flexibility, the advantage of the increased molecular size is maximized. Notably, the peak at high field is unusually broad with a relaxivity value that exceeds 16 mM−1 s−1 even at 200 MHz. Contrast agents that exhibit high relaxivities over a broad frequency range and, in particular, at high frequencies are interesting as modern clinical practice tends to use high magnetic fields (>60 MHz) to increase sensitivity. 1/T1 (T1=longitudinal relaxation time) NMRD profiles of solutions of [Gd2L(H2O)4]2− (▴) and [Fe{Gd2L(H2O)4}3]4− (□). cGd=1 mM, pH 6.0, 25 °C. In terms of molar relaxivity, our metallostar complex is among the best molecules reported. Its relaxivity is comparable to that of a Gd(DOTA)-loaded (DOTA=1,4,7,10-tetra(carboxymethyl)-1,4,7,10-tetraazacyclododecane) Generation 10 dendrimer (relaxivity values of 27.0 and 33.2 mM−1 s−1 at 20 and 60 MHz, respectively, at 25 °C for [Fe{Gd2L(H2O)4}3]4− compared to 36 and 27 mM−1 s−1 at 23 °C for the G10 dendrimer).8 This G10 dendrimer, with a molecular weight of 3000 kDa and 1860 GdIII chelates on the surface, is considered to represent a high “GdIII density”. Such dendrimers have been designed for receptor targeting as they can deliver a concentrated relaxation effect to a given receptor binding site. Indeed, Wiener and co-workers used Gd(DTPA) (diethylenetriamine-N,N,N′,N′′,N′′-pentaacetic acid) dendrimers conjugated with folic acid for visualizing the folate receptor, which is overexpressed in many tumors.9 High “effective relaxivities” are required for applications such as cell imaging in which a sufficient relaxation effect has to be produced by a limited mass of the agent. Figure 4 compares “effective relaxivities” for commercial and potential contrast agents and shows that the dinuclear [Gd2L(H2O)4]2− species and especially the metallostar [Fe{Gd2L(H2O)4}3]4− are remarkably powerful contrast agents in terms of efficacy by unity mass. Comparison of the effective (mass) relaxivities (calculated according to Equation (1)) for a series of GdIII complexes. The monomers [Gd(DTPA)]2− and [Gd(DOTA)]− are commercial contrast agents.7 [Fe{Gd(tpy-DTTA)}2] and [Fe{Gd(phen)(HDO3A)}3]2+4 are self-assembled structures. Gadomer 17,10 [G5{Gd(DO3A)}52],11 and [G10{Gd(DOTA)}1850]8 are dendrimers. The effective relaxivities were calculated at 20 MHz and 37 °C, except for the last three columns which were determined at 25 °C. (r1=20.17 mM−1 s−1, the molar relaxivity per Gd for [Fe(Gd2L(H2O)4)3]4− at 20 MHz and 37 °C used for the calculation of the “effective relaxivity”.) Desreux and co-workers proposed the phenanthroline–Gd(HDO3A) derivative as a contrast agent responsive to FeII concentration, however, such an application did not materialize. In addition to the large increase in the relaxivity upon formation of [Fe{Gd2L(H2O)4}3]4− from [Gd2L(H2O)4]2− (≈100 % at 30–60 MHz), our system has practical advantages over [Gd(phen)(HDO3A)]. The ligand synthesis is straightforward, and the structural diversity, as it exists for [Fe{Gd(phen)(HDO3A)}3], is largely reduced by the higher symmetry of L8−. In conclusion, we have reported a heterometallic, self-assembled metallostar, [Fe{Gd2L(H2O)4}3]4−, which exhibits a particularly high relaxivity for its moderate molecular weight. This is explained in terms of a rigid supramolecular structure and two inner-sphere water molecules with an exchange rate that lies close to optimal. The presence of six efficiently relaxing GdIII centers within one metallostar leads to an exceptionally high relaxivity confined to a small molecular space (high density of relaxivity). To our knowledge this is the highest relaxivity per molecular mass ever reported for a GdIII complex. The ligand L8− is also a prime candidate as a terminal ligand for constructing larger-size, FeII (or RuII)-based metallostars or metallodendrimers that are loaded with GdIII on the surface. [Gd2L(H2O)4]2− was prepared by adding solid L to a solution of GdCl3 in 1:2 molar ratio (pH 6.0; 50 mM MES (2-morpholinoethanesulfonic acid) buffer). [Fe{Gd2L(H2O)4}3]4− was prepared by mixing a solution of [Gd2L(H2O)4]2− with a freshly prepared solution of FeII ions (Fe(NH4)2(SO4)×6 H2O) in 3:1 molar ratio (pH 6.0; 5 mM MES). ESI-MS for [Gd2L(H2O)4]2−: m/z (%): 579.3 (100) [M2−], 1239.2 (68) [(MH+2 CH3CN)−], 1180.3 (50) [MNa−], 1159.2 (23 %) [MH−]; ESI-MS for [Fe{Gd2L(H2O)4}3]4−: m/z (%): 579.3 (100) [M−[Fe(Gd2L)2]2−], 882.3 (75) [M4−], 1184.7 (43 %) [MNa3−]. The isotopic peak patterns in both cases agree with simulated spectra. All manipulations with the metallostar were carried out with the exclusion of oxygen to avoid oxidation of FeII. The 1/T1 NMRD profiles were recorded on a Stelar Spinmaster Fast Field Cycling NMR relaxometer (B=2.35×10−4–0.47 T, proton Larmor frequencies 0.01–20 MHz), on Bruker Minispecs (30, 40, and 60 MHz), and on Bruker spectrometers (50, 100, 200, 400, and 600 MHz). The UV/Vis titration was performed on a Perkin-Elmer Lambda 19 spectrometer. Supporting information for this article is available on the WWW under http://www.wiley-vch.de/contents/jc_2002/2005/z461875_s.pdf or from the author. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.