Lanthanide(III) Complexes Based on an 18-Membered Macrocycle Containing Acetamide Pendants. Structural Characterization and paraCEST Properties

We report a detailed investigation of the coordination properties of macrocyclic lanthanide complexes containing a 3,6,10,13-tetraaza-1,8(2,6)-dipyridinacyclotetradecaphane scaffold functionalized with four acetamide pendant arms. The X-ray structures of the complexes with the large Ln3+ ions (La and Sm) display 12- and 10-coordinated metal ions, where the coordination sphere is fulfilled by the six N atoms of the macrocycle, the four O atoms of the acetamide pendants, and a bidentate nitrate anion in the La3+ complex. The analogous Yb3+ complex presents, however, a 9-coordinated metal ion because one of the acetamide pendant arms remains uncoordinated. 1H NMR studies indicate that the 10-coordinated form is present in solution throughout the lanthanide series from La to Tb, while the smaller lanthanides form 9-coordinated species. 1H and 89Y NMR studies confirm the presence of this structural change because the two species are present in solution. Analysis of the 1H chemical shifts observed for the Tb3+ complex confirms its D2 symmetry in aqueous solution and evidences a highly rhombic magnetic susceptibility tensor. The acetamide resonances of the Pr3+ and Tb3+ complexes provided sizable paraCEST effects, as demonstrated by the corresponding Z-spectra recorded at different temperatures and studies on tube phantoms recorded at 22 °C.


■ INTRODUCTION
Magnetic resonance imaging (MRI) is a technique commonly used in medical diagnosis that provides three-dimensional images of soft tissues with very high resolution and unlimited depth penetration. 1 MRI takes advantage of the 1 H NMR signal of water proton nuclei present in the body, generating contrast due to changes in the density of protons and their longitudinal (T 1 ) or transverse (T 2 ) relaxation times. 2 Both T 1 and T 2 can be shorted in the surrounding of paramagnetic species such as Gd 3+ or Mn 2+ chelates, and thus complexes of these metal ions were proposed as contrast agents about 4 decades ago, 3 subsequently entering clinical practice. 4 The continuous interest of the MRI community in Gd 3+based and, to a lesser extent, Mn 2+ -based contrast agents generated a number of molecular systems with improved properties. This contributed significantly to obtaining new insights in the coordination chemistry of these metal ions in aqueous media. 5,6 Furthermore, a wide range of the so-called smart or responsive contrast agents were designed to provide a response to different physiologically relevant parameters, such as the pH, temperature, or presence of anions or cations relevant in vivo. 7 The main drawback of the responsive Gd 3+ probes is the difficulty of their direct detection: they operate by modifying the NMR signal of water proton nuclei already present in the body, which causes the presence of significant background signals. As a consequence, quantification of the response of Gd 3+ contrast agents in vivo remains a difficult task because the observed signal depends both on the physiological parameter that induces relaxivity changes and on the probe concentration. 8 Paramagnetic contrast agents relying on the chemical exchange saturation transfer (paraCEST) approach have attracted great attention during the last 2 decades as alternatives to the Gd-or Mn-based probes. 9 These agents possess exchangeable protons (typically amide, hydroxyl, or coordinated water molecules) in intermediate-to-slow exchange with bulk water. The paramagnetic shift induced by the metal ion moves the signal of exchangeable protons away from that of bulk water. Consequently, the application of a presaturation pulse at the frequency of the exchanging protons transfers the energy from the saturated spins to the water pool, which decreases its signal intensity. 10 The relatively large chemical shift difference between the signals of exchangeable protons and bulk water (Δω), induced by the paramagnetic ion, implies that the proton exchange rate (k ex ) can be faster, while still maintaining the slow-to-intermediate exchange regime (k ex ≤ Δω) to observe the CEST effect. 11 Moreover, the paramagnetic shift can, in some cases, cause the differentiation of two nonequivalent exchangeable protons to result in two separate CEST signals. This feature is beneficial because it allows for ratiometric analyses and thus possibly quantitative estimation of the physiological parameters. 12 Because of such an advantageous prospective, complexes of both paramagnetic lanthanide and transition-metal ions were extensively investigated as potential paraCEST candidates. 13−15 In the particular case of lanthanide ions, most of the complexes investigated in this context were derivatives of cyclen because this type of chelator often forms very stable and inert complexes when functionalized with four pendant arms. 16 However, azamacrocyclic platforms other than cyclen were also investigated as possible chelators to form inert complexes with affirmative CEST features. For instance, the 18-membered macrocyclic ligand L 1 (Chart 1) was functionalized with different pendant arms to accommodate the high coordination numbers usually observed for the Ln 3+ ions in solution. Along these lines, the derivatives containing acetate 17 and methylenephosphonic acid 18 groups (L 2 and L 3 , respectively) were prepared and reported more than 15 years ago. Moreover, the same macrocyclic platform was also functionalized with neutral pyridyl 19 and hydroxyethyl 20,21 pendant arms (L 4 and L 5 ). This family of ligands generally provides 10-coordinated complexes, although in some cases, decoordination of one of the pendant arms is observed along the second half of the lanthanide series. Interestingly, the [LnL 5 ] 3+ complexes containing hydroxyl groups were found to be exceptionally inert with respect to complex dissociation, remaining intact in a 1 M HCl solution over periods of months. Finally, in a recent work, we reported the derivative L 6 containing acetamide pendants and demonstrated that its Eu 3+ complex displays high kinetic inertness and provides a strong pH-sensitive CEST effect due to the amide protons. 22 Following these encouraging results, we performed further studies with this promising chelator. Herein we report a detailed investigation of the structure of several lanthanide complexes of L 6 , both in the solid state and in solution. The Xray structures of three complexes are presented (La 3+ , Sm 3+ , and Yb 3+ ). The structure in solution was assessed through a detailed study of the 1 H and 89 Y NMR spectra, including a detailed analysis of the paramagnetic shifts observed for the Tb 3+ derivative. Finally, the paraCEST spectra of the Pr 3+ and Tb 3+ complexes recorded at different temperatures are presented and analyzed quantitatively using the Bloch− McConnell (BM) theory to determine the exchange rates of amide protons.
X-ray Crystal Structures. Crystals of [LaL 6 (NO 3 )] 2 [La-(NO 3 ) 6 ]·NO 3 ·4CH 3 OH were obtained by the slow evaporation of a methanolic solution of the ligand containing an excess of La(NO 3 ) 3 . This compound crystallizes in the centrosymmetric C2/c monoclinic space group, and the asymmetric unit encompasses the [LaL 6 (NO 3 )] 2+ complex, half of a [La(NO 3 ) 6 ] 3− anion, half of a nitrate anion, and two methanol molecules. The [La(NO 3 ) 6 ] 3− entity ( Figure S1) was previously found in crystals of different cationic La 3+ complexes, presumably aiding crystallization because of its large size. 24 The [LaL 6 (NO 3 )] 2+ cation shows that the La 3+ ion is coordinated by the six N atoms of the macrocycle skeleton, the four O atoms from the amide groups, and two of the O atoms of a nitrate group acting as a bidentate ligand, which results in coordination number 12. The bond distances of the metal coordination environment are collected in Table 1, while a view of the structure of the complex is presented in Figure 1.

Chart 1. Ligands Discussed in the Present Work
Inorganic Chemistry pubs.acs.org/IC Article The coordination polyhedron around the La 3+ ion can best be described as a twisted icosahedron ( Figure S2), as indicated by the analysis performed with the SHAPE program. 25,26 The nitrate anion in [LaL 6 (NO 3 )] 2+ provides a slightly asymmetric bidentate coordination, with La−O distances similar to those reported for other 12-coordinated La 3+ complexes containing bidentate ligands. 27 The [LaL 6 (NO 3 )] 2+ complex presents an unprecedented conformation of the ligand in which the four pendant arms of L 6 are oriented to the same side of the macrocyclic unit, giving a syn conformation. The macrocyclic unit is not folded or twisted but shows a plateau conformation. The four amine N atoms define a least-squares plane [root-mean-square (rms) = 0.238 Å] that contains the La 3+ ion, while the pyridine N atoms are ca. 1.07 Å below that plane. As a result, the N(4)− La(1)−N(1) angle [125.17(19)°] is not linear. The pyridyl units are slightly tilted with respect to each other, with the least-squares planes intersecting at 28.2°. The two chelate rings associated with binding of the ethylenediamine units adopt the same conformation, which can be defined as δδ or λλ. 28 The layout of the four acetamide pendant groups provides the second source of chirality, leading to the presence of (centrosymmetrically related) Λ(λλ) or Δ(δδ) enantiomers in the crystal lattice. 29 Crystals of the formula [SmL 6 ](NO 3 ) 2.91 ·Br 0.09 and [YbL 6 ]-(NO 3 ) 2.7 ·Br 0.3 ·3H 2 O were obtained by the slow evaporation of an aqueous solution of the complex. The small fraction of bromide anions present in these crystals is likely due to the presence of KBr impurities in the batch of ligands used for preparation of the complex. Sm 3+ crystals also contain the [SmL 6 ] 3+ cation ( Figure 2) and nitrate anions. This compound crystallizes in the noncentrosymmetric monoclinic C 2 space group, and the asymmetric units contain one and two halves of the [LnL 6 ] 3+ units, six nitrate anions, and water molecules.
The [SmL 6 ] 3+ cation contains a 10-coordinated metal ion that is directly coordinated by the six donor atoms of the macrocyclic fragment and the four O atoms of the acetamide pendants. The overall structure is very similar to that reported previously for the Eu 3+ and Y 3+ derivatives. 22  Crystals of the Yb 3+ compound contain the [YbL 6 ] 3+ complex, in which the metal ion is 9-coordinated by the ligand (Figure 3). One of the acetamide pendant arms remains uncoordinated, with the ligand adopting a twist-fold conformation characterized by a N(1)−Yb(1)−N(4) angle of 144.04(5)°and a dihedral angle between the pyridine units of  17 Changes in the coordination number from 10 to 9 across the lanthanide series coming as a result of lanthanide contraction are fairly common. 31 The coordination geometry is close to a spherical tricapped trigonal prism ( Figure S4). The upper tripod of the polyhedron is defined by O(1), O(4), and N(5), while the lower tripod contains N(3), N(1), and O(2). These two triangular faces are almost parallel, intersecting at 3.6°. The capping tripod is occupied by N(2), N(4), and N(6). The syn structure of the La 3+ complex is characterized by Ln−N distances involving the pyridyl donor atoms [N(1) and N(4)] comparable to those with amine N atoms. However, the Ln−N(1) and Ln−N(4) bonds are significantly shorter than those involving amine N atoms in [SmL 6 ] 3+ and [YbL 6 ] 3+ , a situation that is commonly observed when the pendant arms coordinate from both sides of the macrocyclic mean plane. 17 Figure 4, spectrum a). The spectrum changes slowly and irreversibly with time, a process that speeds up upon heating of the solution. The spectrum obtained after heating is well-resolved, showing the eight signals expected for a D 2 symmetry in solution. We attribute this behavior to the formation of a kinetic species with a syn conformation of the macrocycle, as observed in the X-ray structure (C 2 symmetry), which evolves to the thermodynamically stable species with D 2 symmetry in solution. The spectrum of the Ce 3+ complex is poorly defined, likely for the same reason.
The complexes with Pr 3+ , Nd 3+ , Sm 3+ , Eu 3+ , and Tb 3+ present well-resolved spectra with eight paramagnetically shifted resonances, which points to a D 2 symmetry of the complexes in solution (Figures 5 and S5−S8). These results are in full agreement with the X-ray structures described above. The 1 H NMR spectra of the [LnL 6 ] 3+ complexes (Ln = Pr− Tb, except Pm) were assigned on the basis of COSY spectra and line-width analysis ( Table 2). The observed chemical shifts are similar to those previously reported for the [LnL 5 ] 3+   Inorganic Chemistry pubs.acs.org/IC Article complexes, 21 indicating similar magnetic anisotropies and thus similar solution structures. 33 The structure in solution of the [TbL 6 ] 3+ complex was further investigated by analyzing the observed 1 H NMR shifts, which are the result of diamagnetic (δ dia ) and paramagnetic (δ para ) contributions (eq 1).
The diamagnetic contributions to the observed shifts were obtained from the chemical shifts observed for the diamagnetic [LaL 6 ] 3+ analogue ( Table 2). The paramagnetic shifts induced by Tb 3+ are the result of both the contact (δ con ) and pseudocontact (δ pscon ) mechanisms, 34 but only the latter encodes information on the position of the observed nuclei with respect to the paramagnetic center. Thus, we estimated the contact contributions by calculating the hyperfine coupling constants A/ℏ responsible for the contact shifts using density functional theory (DFT). Contact shifts are directly proportional to the hyperfine coupling constant at the observed nuclei, as given by eq 2, where ⟨S z ⟩ is the spin expectation value of the lanthanide ion, 35 γ I is the nuclear gyromagnetic ratio, k is the Boltzmann constant, and β is the Bohr magneton. 36 The values of A/ℏ were estimated using calculations on the [GdL 6 ] 3+ analogue, following the methodology reported previously. 37 DFT provides a 10-coordinated structure for [GdL 6 ] 3+ very similar to those observed in the solid state for the Sm 3+ , Eu 3+ , and Y 3+ derivatives ( Figure S9). Contact shifts were subsequently obtained by using ⟨S z ⟩ = 22.0. 21 Separation of the contact and pseudocontact shifts shows that the paramagnetic shifts are generally dominated by the pseudocontact mechanism, with the contact shift being more important for equatorial protons (Table 3). This has been attributed to distinct Ln−N−C−H dihedral angles characterizing the axial (∼80°) and equatorial (∼175°) protons because contact shifts show a Karplus-like behavior on this angle. 37 The pseudocontact contribution can be expressed as in eq 3 if the reference frame coincides with the main directions of the magnetic susceptibility tensor χ, 38 and x, y, and z are the Cartesian coordinates of the observed nucleus relative to the position of a Ln 3+ ion placed at the origin, while Δχ ax and Δχ rh are the axial and rhombic components of the diagonalized magnetic susceptibility tensor.
A comparison of the experimental and calculated pseudocontact shifts, obtained with the geometry of [TbL 6 ] 3+ optimized by means of DFT, is presented in Figure 5 (see also Table S1). The excellent agreement between the two sets of data unambiguously establishes that the [TbL 6 ] 3+ complex presents a structure in solution very similar to that observed in the solid state for the Sm 3+ and Eu 3+ analogues (see also Figure  5).
The best fit of the data provided a highly rhombic susceptibility tensor characterized by Δχ ax = −20.5 ± 0.5 × 10 −32 and Δχ rh = −19.9 ± 1.1 × 10 −32 m 3 . These values are very similar to those determined previously for [TbL 5 ] 3+ , confirming that the two complexes present similar structures in solution.
The 1 H NMR spectra of the complexes with the heaviest Ln 3+ ions (Dy−Lu) are complicated likely because of the fact that one of the pendant arms of the ligand is not coordinated to the Ln 3+ ion, which is in line with the X-ray structure of the Yb 3+ complex. The spectrum of the Yb 3+ complex is welldefined, showing 30 signals expected for a 9-fold coordination of the ligand in the range ∼151 to −70 ppm ( Figure S10).
Using n N am = 4, n N py = 2, and n O a = 4 gives a calculated 89 Y chemical shift of 61.2 ppm, which is in very good agreement with the experimental value of 51 ppm. An identical analysis but using n O a = 3 gives a calculated value of 150.7 ppm, again in excellent agreement with the experimental value observed for the major (unsymmetrical) species present in solution (157 ppm). Thus, these results confirm the presence in solution of 9-coordinated species in the case of the smallest Ln 3+ ions and Y 3+ , with a smaller population of the 10-coordinated species having D 2 symmetry.
The 89 Y shielding constants of the 10-and 9-coordinated forms of [YL 6 ] 3+ were calculated using DFT (see the Computational Details). The chemical shielding values obtained from these calculations are compiled in Table 4. The subsequent calculation of 89 Y chemical shifts requires determination of the shielding constant of a reference, generally [Y(H 2 O) 8 ] 3+ (0.0 ppm). 42 The shielding constants calculated for the 9-and 10-coordinated forms of [YL 6 ] 3+ differ by 105.7 ppm, in nice agreement with the experimental value of Δδ = 106 ppm. This value is close to the empirical contribution of an amide donor of S O a = 89.5 ppm. 40 However, the use of the shielding constant of [Y(H 2 O) 8 ] 3+ provided 89 Y NMR chemical shifts with large deviations from the experimental value. We recently showed that the agreement between the experimental and calculated 71 Ga chemical shifts improved significantly upon inclusion of an explicit second solvation shell. 43 We thus performed calculations on the [Y(H 2 O) 8 ] 3+ ·16H 2 O system, which includes 16 water molecules in the second sphere involved in hydrogen bonds with the coordinated water molecules ( Figure S11) (Table 4) are in reasonably good agreement with the experimental values of 157 and 51 ppm. σ iso can be separated into the usual diamagnetic (σ d ) and paramagnetic (σ p ) contributions (Table 4). 45 The results show that the distinct chemical shifts are mainly related to variations in the σ p values, which originate from the ability of the applied field to mix excited states into the ground state. The σ d values calculated for all systems are, however, very similar, as would be expected.
CEST Experiments. The crystal structures of the Pr 3+ and Tb 3+ complexes revealed that the complexes are lacking coordinated water molecules. However, they possess exchangeable amide proton pools suitable for exploitation of the CEST effect. Thus, to investigate the potential application of these complexes as paraCEST agents, we conducted a series of NMR and MRI studies. Both solutions of [TbL 6 ] 3+ and [PrL 6 ] 3+ showed two notable paraCEST signals at different temperatures. For the Pr 3+ complex, two CEST peaks at −2.8 and −8.0 ppm are strong but close to the peak of water protons at 25°C ( Figure S12). For the Tb 3+ complex, we performed a set of experiments with the temperature ranging from 10 to 40°C. For each temperature, we recorded the Z-spectra using seven different saturation powers from 2.5 to 30 μT (see the Experimental Section). The results allowed us to follow the position (shift) of the CEST peaks, as well as to calculate the exchange rates of these two exchangeable protons as a function of the temperature.
The CEST signals of [TbL 6 ] 3+ are highly paramagnetically shifted and strongly dependent on the temperature: these two peaks shift from −64 and −76 ppm to −60 and −70 ppm from 25 to 37°C, respectively (Figure 7). Over the entire temperature range examined, the sensitivity of the chemical shift to temperature was 0.54 or 0.41 ppm/°C for the low or  Inorganic Chemistry pubs.acs.org/IC Article high field signal, respectively (Figure 7). This is an interesting result because this particular feature of [TbL 6 ] 3+ can be potentially exploited for measuring the temperature distribution in a living subject: here both signals can be used to determine the temperature while concurrently acting as controls for the other peak; i.e., the distance between these two peaks is also temperature-dependent. The temperature coefficients determined for [TbL 6 ] 3+ are similar to those reported for 1 H resonances of paramagnetic cobalt(II) and iron(II) complexes suggested for registration of the temperature. 46 The exchange rates of the CEST-active protons, k ex , were determined using the previously established qCEST method, 47 by fitting a series of multi-B 1 Z-spectra according to the BM equations. As expected, the k ex values are quite low at lower temperatures and less sensitive compared to those in the higher-temperature region. The k ex values are as low as <100 Hz at 10°C, increasing to up to 3 kHz at 40°C (Table S2). Of the two peaks, the one with lower shift always exchanges faster. The k ex values determined at different temperatures for [TbL 6 ] 3+ were used to estimate the activation parameters for amide proton exchange using the Eyring equation ( Figure  S14). The most shifted signal is characterized by ΔH ⧧ = 45.3 ± 2.6 kJ mol −1 and ΔS ⧧ = −38.4 ± 2.0 J mol −1 K −1 . The signal with a smaller shift (and fast exchange) yields ΔH ⧧ = 47.4 ± 2.2 kJ mol −1 and ΔS ⧧ = −27.5 ± 1.2 J mol −1 K −1 . The enhanced exchange rate of the latter amide protons appears to be related to a substantially lower entropy barrier, presumably due to the formation of a more favored (i.e., less ordered) transition state, which likely possesses a more extensive dispersion of charge. 48 Importantly, the k ex values at 37°C are in the optimal range of 1.5−2.5 kHz, which combined with the high number of the NMR equivalent and shifted protons (two groups, each with four protons) makes this platform very attractive for the further development of potent paraCEST probes. 15,22 Concurrently, the obtained exchange rates for [PrL 6 ] 3+ were in a range similar to that for [TbL 6 ] 3+ (Table S2), albeit exhibiting somewhat slower exchange rates. Nevertheless, the low paramagnetic shift of the CEST peaks limits the potential use of this complex in future paraCEST MRI studies.
We also performed MRI experiments on the tube phantoms at 7 T magnetic field. We prepared the solutions of [TbL 6 ] 3+ and [PrL 6 ] 3+ in the same concentrations as those used in NMR experiments. We then obtained CEST MRI images using different saturation times and powers and compared the results to those obtained by means of NMR. Both complexes still showed strong CEST signals at room temperature (RT; Figure  8). However, because of the lower temperature in the MRI scanner (∼21−22°C) and hence the slower exchange rates of amide protons during these experiments, the resulting CEST effect was lower (Table S2). Expectedly, the CEST MRI signals at a slightly shorter saturation time (3 s instead of 5 s) showed similar values when the saturation power was kept the same (10 μT), whereas a weaker saturation power (5 μT instead of 10 μT) resulted in much weaker CEST effects ( Figure S13). Overall, the results obtained with [TbL 6 ] 3+ showed that it remains good candidate for further CEST MRI studies and is an excellent basis for the development of other potent paraCEST probes.

■ CONCLUSIONS
The detailed structural investigation reported here evidences that the [LnL 6 ] 3+ complexes adopt different structures depending on the size of the lanthanide ion. The X-ray structure of the La 3+ complex contains 12-coordinated metal ions, with the four pendant arms coordinating from the same side of the macrocyclic unit and a nitrate anion coordinating from the opposite side. However, NMR studies demonstrate that this unusual structure evolves to a 10-coordinated structure with an effective D 2 symmetry in solution. The large Ln 3+ ions adopt this structure in solution until Tb 3+ ,  Inorganic Chemistry pubs.acs.org/IC Article while for the small ions, 9-coordinated structures are observed. The 10-coordinated structure of [TbL 6 ] 3+ was established by analyzing the paramagnetic shifts observed in the 1 H NMR spectrum. The 1 H and 89 Y NMR spectra indicate that the major species present in solution is 9-coordinated, with a small proportion of the symmetrical D 2 structure also being present (∼14%). The [TbL 6 ] 3+ complex contains two pools of amide protons with relatively large chemical shifts showing strong temperature dependence (ranging from ∼69 to 85 ppm depending on the temperature). These amide protons are characterized by a rather slow exchange rate with bulk water (1.5−2.5 kHz). Furthermore, each of the CEST signals originates by four magnetically equivalent protons, which results in strong CEST responses even at low saturation powers. Thus, [TbL 6 ] 3+ can be regarded as a very attractive platform to develop CEST MRI agents.
■ EXPERIMENTAL SECTION General Methods. 1 H NMR spectra were obtained at 25°C using a Bruker ARX400 spectrometer and solutions of the complexes in D 2 O. Chemical shifts were referenced with respect to the residual HDO proton signal (δ = 4.79 ppm). 49 Elemental analyses were obtained with a Carlo-Erba EA 1108 microanalyzer. Fourier transform infrared spectra were recorded using the attenuated total reflection method (ATR-FTIR) with a Bruker VECTOR 22 spectrometer (KBr disks). Mass spectra were obtained with a microTOF (focus) mass spectrometer (Bruker Daltonics, Bremen, Germany) equipped with an ApolloII (ESI) source for electrospray ionization.
Preparation of the Complexes. All complexes were prepared by the reaction of a solution of Ln(NO 3 ) 3 ·xH 2 O (0.040 mmol) and L 6 · H 2 O (0.023 g, 0.040 mmol) in methanol (15 mL), following the same procedure as that described for the europium and yttrium analogues. 22  [LaL 6 (NO 3 )] 2 [La(NO 3 ) 6 ]·NO 3 ·4CH 3 OH. The synthesis followed the same procedure as that described above but using a 1.5:1 La(NO 3 ) 3 · 5H 2 O/L 6 molar ratio. L 6 ·H 2 O (0.023 g, 0.040 mmol) and La(NO 3 ) 3 · 5H 2 O (0.025 g, 0.060 mmol). Yield: 0.053 g, 60%. Elem anal. Calcd for C 56  Crystal Structure Determinations. X-ray diffraction data of [LaL 6 (NO 3 )] 2 [La(NO 3 ) 6 ]·NO 3 ·4CH 3 OH and [SmL 6 ](NO 3 ) 2.91 · Br 0.09 were recorded at 273(2) K using a Bruker Smart-CCD-1000 diffractometer and graphite-monochromated Mo Kα radiation (λ = 0.71073 Å). Corrections for Lorentz and polarization effects were applied to all data. For [YbL 6 ](NO 3 ) 2.7 ·Br 0.3 ·3H 2 O, X-ray diffraction data were obtained at 100(2) K with Mo Kα radiation and a Bruker D8 Venture Photon 100 CMOS detector. Collection of frames of data, reflection indexing, and lattice parameter determination were achieved with the APEX3 software. Integration of the intensity of the reflections was carried out with SAINT. 50 The software SADABS 51 was used in all cases for scaling and empirical absorption correction. All structures were refined by full-matrix least squares based on F 2 with the SHELXT program. 52 Non-H atoms were refined with anisotropic displacement parameters. H atoms were included in calculated positions and refined with isotropic displacement parameters. For [SmL 6 ](NO 3 ) 2.91 ·Br 0.09 , a solvent masking routine was applied to correct the reflection data for the diffuse scattering associated with disordered water molecules present in the crystal. Molecular graphics were generated using OLEX2. 53 Crystal data and details on data refinement are provided in Table 5 CEST MRI images were acquired at RT using rapid acquisition with relaxation enhancement (RARE) pulse sequence with the following imaging parameters: repetition time/echo time = 16280/ 3.26 ms, field of view = 32 × 32 mm, matrix size = 64 × 64, slice thickness = 2 mm, rare factor = 64, number of excitation = 1, acquisition time = 35 min 49 s. MT parameters were as follows: saturation pulse duration 3 and 5 s, saturation power 5 and 10 μT, 131 irradiation offsets in the range −100 to +100 ppm.
Image analysis was performed in MATLAB (MathWorks, USA). Initially, CEST MRI images were linearly interpolated and shifted to the center frequency in order to remove B 0 inhomogeneity artifacts. Thereafter, pairs of Z-spectrum images were extracted for the irradiation offsets of the CEST peaks (+Δω) for the complexes (−3 and −8.5 ppm for [PrL 6 ] 3+ and −68.5 and −85.0 ppm for [TbL 6 ] 3+ ), and their corresponding opposites (−Δω) with respect to bulk water.
Quantification of the CEST effect was achieved using an inverse asymmetry analysis of the normalized Z-magnetization, using the inverse difference of the magnetization transfer (MTR ind in eq 5). The latter is calculated from the unsaturated water magnetization (M 0 ) and magnetizations of the on-resonance at the frequency +Δω (M z+ ) and the off-resonance at −Δω (M z− ) with respect to the bulk water signal. 56 Computational Details. Full geometry optimization of the [GdL 6 ] 3+ complex and the subsequent frequency analysis were carried out within the framework of DFT (TPSSh 57 exchange correlation functional) employing the Gaussian 09 package (revision D.01). 58 The inner electrons of Gd (46 + 4f 7 ) were described with a large-core quasi-relativistic effective core potential (RECP), while a [5s4p3d]-GTO basis set was used for the outermost 11 electrons. 59 All other atoms were described using the 6-31G(d,p) basis set. Hyperfine coupling constants (A/ℏ) were calculated using a smallcore relativistic effective core potential, which places 28 electrons of Gd in the core. The valence space of Gd was described with the ECP28MWB_GUESS basis set, 60 while the EPR-III 61 basis set was used for the ligand atoms. The effects of bulk water were incorporated in all calculations using the integral equation formalism of the polarized continuum model. 62 The Y 3+ complexes were optimized using a similar approach, with the TPSSh functional, the ECP28MWB quasi-relativistic ECP, and its associated basis set for Y and the standard 6-311G(d,p) basis set for all other atoms. 63 The 89 Y NMR shielding tensors were calculated with the GIAO 64 method and the TPSSh functional, 57 using the ORCA program package (version 4.2.0) 65 and the DKH2 66 relativistic method. In these calculations, we used the old-DKH-TZVPP basis set, as implemented in previous versions of ORCA (see the Supporting Information), which is based on the TZVPPAll basis set of Aldrich 67 and was recontracted for DKH2 calculations. The resolution of identity approximation for both Coulomb-and exchange-type integrals (RIJK) was used for both selfconsistent field and calculation of the NMR chemical shielding constants. 68,69 Auxiliary basis sets were generated with the Autoaux procedure implemented in ORCA. 70