Hyperbranched polyethylenimine based polyamine-N-oxide-carboxylate chelates of gadolinium for high relaxivity MRI contrast agents

Abstract

A simple and convenient method is explored for preparing high relaxivity magnetic resonance imaging (MRI) contrast agents with numerous N-oxide groups decorated on the hyperbranched precursors as ligands for the first time. Such Gd(III) containing complexes use modified hyperbranched (or linear) polyethylenimine (PEI) with polydentate amine-N-oxide carboxylate groups as the ligands and complexes with different molecular weights show sizes from 1.63 nm to 5.40 nm with the increase of the molecular weight. T₁ magnetic resonance imaging experiments show that the hyperbranched MRI contrast agents have much higher relaxivity than the linear analogue. Meanwhile, with the increase of molecular weight, the relaxivity of the complexes slowly increases to 8.9 mM⁻¹ s⁻¹, a number much higher than that of commercial MRI contrast agent, Gd(III) diethylenetriamine pentaacetate (Gd-DTPA). The in vitro cytotoxicity test demonstrates the excellent biocompatibility of these complexes, which is essential for clinical applications.

---

Introduction

Magnetic resonance imaging (MRI) is one of the most powerful diagnostic imaging techniques and has been extensively used in clinical practice because of its excellent spatial resolution, noninvasive and nondestructive nature.¹ To further improve the contrast of the imaging, MRI contrast agents (CAs) are generally needed to enhance the image contrast between normal and diseased tissues, and to indicate the status of organ function or blood flow. Gadolinium(III), with its seven unpaired electrons and long electron spin relaxation time, is an ideal candidate for such a proton relaxation agent and is the most widely used metal centre for such purposes.² A number of factors contribute to the relaxivity values (r₁) of the contrast agents, including the number of inner-sphere water molecules (q), the rotational correlation time (τ_R), and the residence time of the solvent molecule in the complex (τ_M). According to the well-established Solomon–Bloembergen–Morgan (SBM) theory, higher relaxivity can be achieved with higher q, longer τ_R and optimized τ_M.³

Due to their well-defined architectures, multivalent surfaces, and nanoscalesizes, the use of dendrimers as scaffolds to prepare MRI contrast agents has received tremendous interests.⁴ In most of these cases, dendrimers serve as a nanoplatform to carry multiple copies of small molecular Gd(III)–chelates. A pioneering work⁵ was performed by Wiener et al. in 1994 by using amine-terminated ammonia-cored PAMAM dendrimer as the scaffold and (TU-DTPA) as the chelating ligand. Kobayashi et al.⁶ synthesized a group of DTPA–Gd(III)-functionalized PPI dendrimers of different generations, showing that the r₁ value almost had a linear relationship with the molecular weight of the dendrimers. Hyperbranched polymers, the less-defined analogues of dendrimers, have also attracted many interests due to their specific spheroid-like shape, multifunctionality and readily availability. Compared with dendrimers, the synthesis of hyperbranched polymers in a one step process is easier and more economical, making them gain increasing interest, especially in industrial applications.⁷ In many aspects, dendrimers can be replaced by hyperbranched polymers due to their similar properties, such as low viscosity, high functionality, and spheroid-like shape.^8–12

Amine-N-oxides are highly polar substances that can be easily prepared by N-oxidation of N-heteroaromatic compounds or tertiary amines with H₂O₂ or peroxoic acid. The generated oxygen atom in the N-oxide has a stronger dipole than the oxygen atoms of other common oxo-donors such as alcohols, ethers, and amides.¹³ Feng et al.^14–19 developed a new family of N,N′-dioxide amide compounds as chiral ligands for Lewis acidic metals, such as Mg(II), La(II), Ni(II), Sc(III) and In(III). Such chiral tetradentate ligand–metal complexes were established as effective catalysts for different asymmetric reactions. A large amount of lanthanide complexes containing N-oxide groups as ligands were synthesized by Zinner et al.,^20–24 among which the coordination mode of metal atoms with O atoms in N-oxide groups was demonstrated. As the Gd(III) cation is highly oxophilic and will bind more strongly to the oxygen donors provided by the ligand rather than the mixture of nitrogen and oxygen donors offered by a ligand such as EDTA,²⁵ the introduction of N-oxide structure in ligand for MRI contrast agent is very attractive. The synthesis of hyperbranched polymers is much easier than dendrimers, making them suitable precursors for such N-oxide containing ligands.

Herein, we develop a simple and convenient method for preparing hyperbranched polymeric MRI contrast agents with numerous N-oxide groups in the ligands for the first time. Commercially available and economical hyperbranched polyethylenimine (HPEI) with relatively simple structure similar to EDTA for further modification is chosen as the precursor to decorate with N-oxide groups (Scheme 1). In this way, we hope to obtain hyperbranched polymeric MRI contrast agent with high relaxivity. To study the relationship between the relaxivity and the molecular weight, four kinds of HPEI with M_n of 600, 1200, 1800, 10 000, namely HPEI-600, HPEI-1200, HPEI-1800 and HPEI-10000, were used. Meanwhile, a linear PEI with M_n of 10 000 was also used to study the effect of the microstructure of PEI–NO–Gd(III) on the relaxivity of the contrast agent.

Scheme 1 Formation of HPEI–NO–Gd(III) complex from hyperbranched polyethylenimine.

Results and discussion

Synthesis of the HPEI–NO–Gd(III) complexes

Four HPEI–NO–Gd(III) complexes with different molecular weights and one LPEI–NO–Gd(III) complex were prepared as follows (Fig. 1). Firstly, the commercial available PEI was allowed to react with tert-butyl bromoacetate to replace all the H atoms in the amine groups with tert-butoxy carbonyl methyl groups. The molecular weights of these PEI–t-butyl molecules increased to 1811, 2424, 3605, 33 039 and 39 918 respectively for HPEI-600, HPEI-1200, HPEI-1800, HPEI-10000 and LPEI-10000 (Fig. S1†). Secondly, the tert-butyl protecting groups were removed by reaction with TFA for 24 h. After evaporation of the excess of TFA, the complex was washed with diethyl ether. The disappearance of the tert-butyl protons around 1.4 ppm in the ¹H NMR spectra (Fig. 2) confirms that the tert-butyl groups were removed successfully. Apart from NMR spectra, the first two steps were also confirmed by FTIR analysis. As shown in Fig. 3, the stretching vibration of C=O at 1730 cm⁻¹ and C–O at 1160 cm⁻¹ in PEI–t-butyl indicate the formation of ester (curve b). The absorption at 1640 cm⁻¹ in curve c is the stretching vibration of C=O in PEI–CH₂COOH. Thirdly, the complex was oxidized with hydrogen peroxide to convert all the tertiary amine to N-oxide groups. The shift of the proton signal from 3.0–4.2 ppm to 3.6–4.6 ppm indicates the presence of N-oxide groups (Fig. 2). Table 1 lists the ¹⁵N NMR spectra signals (see spectra in Fig. S2†) of PEI–NO and EDTA, EDTA-NO as reference. The ¹⁵N chemical shift of PEI–NO at 116.69 ppm and 117.82 ppm confirms the formation of N-oxide groups by comparison with the values obtained from EDTA and EDTA–NO (δ = 31.96 and 116.96 ppm, respectively) as the similar structure of EDTA and PEI–COOH. The same effect of N-oxidation on ¹⁵N chemical shifts was observed by Devillers et al. in EDTA and EDTA–NO.²⁶ Finally, the complexation with Gd³⁺ was carried out in an aqueous solution for 72 h using 1.2 equiv. of Gd³⁺ with respect to the N-oxide ethylenediamine tetraacetic acid moieties. The uncomplexed Gd³⁺ was precipitated out at high pH, giving the Gd(III)–chelating complexes. The final Gd(III) contents were determined by ICP-AES. Depending on the different molecular weights and structure of the PEI, 21–25.6 wt% of Gd(III) is incorporated in the samples (Table 2). The PEI–NO ligands were also chelated to Y(III) as Gd(III) – containing complex cannot be detected in NMR owing to its' paramagnetic nature and Y(III) has the similar atom diameter with Gd(III). The ¹⁵N chemical shift of PEI–NO–Y(III) (Table 1) at 117.82, 118.19 and 121.38 ppm compared to that of PEI–NO confirms the successful complexation of the final chelating complexes.

Fig. 1 Synthetic procedure of HPEI–NO–Gd(III) complex from hyperbranched polyethylenimine.

Fig. 2 ¹H NMR spectra of HPEI (a) in D₂O, HPEI–t-butyl (b) in CDCl₃ (* indicates acetonitrile), HPEI–CH₂COOH (c) in D₂O (* indicates diethyl ether) and HPEI–NO (d) in D₂O (* indicates methanol).

Fig. 3 FTIR spectra of HPEI, HPEI–t-butyl and HPEI–CH₂COOH.

Table 1 ¹⁵N NMR chemical shifts (δ, ppm) of PEI–NO, PEI–NO–Y, EDTA and EDTA–NO

Entry	^15N NMR signal [ppm]
EDTA	31.96
EDTA–NO	116.96
PEI–NO	116.69, 117.82
PEI–NO–Y(III)	117.82, 118.19, 121.38

---

Table 2 Molecular weights and relaxivities of the PEI–NO CAs

Entry	Mn^a [g mol^−1]	DB^b	Gd content^c [wt%]	N^Gd^d	Ionic r1^e [mM^−1 s^−1]	Molecular r1^f [mM^−1 s^−1]
a Molecular weight of PEI–t-butyl determined by GPC.b Degree of branching, determined by inverted gate ^13C NMR.c Determined by ICP-AES.d Number of gadolinium per molecules (N^Gd) calculated by N^Gd = ([Gd content]/M^Gdw)/([polymer content]/M^ligandn), where Gd content, M^Gdw, M^ligandn, correspond to Gd content determined by ICP-AES, molar mass of Gd, and molecular weight of ligand.e Ionic r1 determined at 1.5 T.f Molecular r1 was calculated by the equation r1 = N^Gd × ionic r1.
HPEI-600	1811	0.61	21%	2.8	8.0	22.4
HPEI-1200	2424	0.63	23%	6.2	8.1	50.2
HPEI-1800	3605	0.61	24%	9.8	8.6	84.3
HPEI-10000	33 039	0.62	26%	59.5	8.9	529.6
LPEI-10000	39 918	—	24%	54.6	5.7	311.2

---

DLS

As shown in Fig. 4, dynamic light scattering (DLS) studies revealed that the hydrodynamic diameter (D_h) of the HPEI–NO ligands increased with the molecular weight, from 1.14 nm at HPEI-600 to 4.42 nm at HPEI-10000 (Fig. 4a). The D_h of the linear PEI–NO was even larger due to its relatively loose microstructure. Similarly, the D_h of the HPEI–NO–Gd(III) complexes increased from 1.63 nm at HPEI-600 to 5.29 nm at HPEI-10000 (Fig. 4b). Besides, the size of the final Gd(III) complex was a little bit larger than that of the corresponding PEI–NO ligand. This may result from the coordination of Gd³⁺ with ligand, causing the stretch of the whole molecule. Particles smaller than 5.5 nm primarily undergo renal clearance and were quickly distributed from the circulation into the soft tissue similarly to Gd-DTPA, indicating the quick leakage of the vasculature into surrounding tissues, even from normal vessels.²⁷

Fig. 4 (a) DLS spectra of PEI–NO ligands of different molecular weights. (b) DLS spectra of PEI–NO–Gd(III) complexes of different molecular weights.

Relaxivity of the PEI–NO–Gd(III) complexes

The results of relaxometric studies on HPEI–NO–Gd(III) complexes with different molecular weights indicated an overall enhancement in relaxivity compared to that of Gd-DTPA (r₁ = 4.3 mM⁻¹ s⁻¹)²⁸ (Fig. 5). This increase is highest for HPEI-10000, with a relaxivity (r₁ = 8.9 mM⁻¹ s⁻¹ per gadolinium atom) almost two times higher than that of Gd-DTPA. Such enhancement is caused by the increase of the molecular weight, which retard the rotational motion of the complexes. According to ICP-AES (Table 2), the calculated number of Gd atoms per HPEI-10000 molecule is 59.5, leading to the whole relaxivity per molecule as high as 529.6 mM⁻¹ s⁻¹.

Fig. 5 r₁ curves of HPEI–NO–Gd(III) complexes with different molecular weights and the linear analogues of LPEI–NO–Gd(III)-10000.

However, there is no obvious difference between these four kinds of hyperbranched CAs. The relaxivity only slightly improves with the increase of molecular weights. It is likely that although the higher molecular weight and the larger size will slow the rotation of the whole molecules, the influence is limited to relatively flexible molecules.^29,30 When the molecular weight reaches 10 000, the relaxivity of the whole complex is dictated by the local motion of the molecule, which is mainly contribute from the distortion of the PEI–NO framework. The relatively flexible structures may be the reason for low relaxivities in comparison with the value of 36 mM⁻¹ s⁻¹ of the more rigid, true dendrimer architecture of DTPA modified PAMAM (G = 9) dendrimers.³¹

Compared with the corresponding linear Gd(III) complex LPEI-10000 (r₁ = 5.7 mM⁻¹ s⁻¹), the relaxivity of hyperbranched Gd(III) complex HPEI-10000 is much higher. The hyperbranched polymers are inherently more rigid than their linear analogues, resulting in less freedom for the Gd–chelates to rotate. Thus, the relaxivities of hyperbranched CAs are higher than those of the linear analogues.³²

Cytotoxicity assays

Cytotoxicity is an important parameter for applying the CAs in biomedical studies. MTT assay with HeLa cells incubated with varying concentrations of the HPEI–NO–Gd(III) complexes is used to evaluate the cytotoxicity. As shown in Fig. 6, the proliferation and viability of the cells were not affected by the HPEI–NO CAs even with the Gd(III) concentrations up to 200 μM after 48 h incubation (see the viability of cells treated with other PEI–NO–Gd(III) complexes in Fig. S3†). It indicates that the CAs synthesized here are nontoxic within the concentration range of Gd(III) ion tested, revealing a good cytocompatibility.

Fig. 6 Relative cell viability [%] of HeLa after treatment with HPEI–NO–Gd-600.

Experimental

Materials

HPEI with different molecular weight and gadolinium(III) chloride hexahydrate were purchased from Alfa Aesar (Ward Hill, US). Linear polyethylenimine (LPEI, M_n = 10 000) was purchased from Energy Chemical (shanghai, China). Trifluoroacetic acid (TFA), tert-butyl bromoacetate, potassium carbonate and hydrogen peroxide were purchased from Sigma-Aldrich (St. Louis, US). All the solvents were purchased from commercial resources and used as received. The human cervical carcinoma cells (HeLa) were provided by China Centre for Type Culture Collection. All cell-culture related reagents were purchased from Gibco (Grand Island, USA).

Characterization

¹H and ¹³C NMR (400 MHz) spectra were recorded on an Ultra Shield 400 spectrometer (Bruker BioSpin AG, Magnet System 400 MHz/54 mm) in D₂O. ¹⁵N NMR (600 MHz) spectra were recorded on a Bruker 600 MHz spectrometer and using the HMBC pulse sequence with a mixing time of 100 ms. The nucleus ¹⁵N in 2D spectra is correlated with another nucleus ¹H. Chemical shifts are expresses in ppm and are referenced to TMS for ¹H and ¹³C (δ = 0 ppm) or nitromethane for ¹⁵N (δ = 375 ppm). The number-averaged molecular weight (M_n) of the polymers were determined by gel permeation chromatography (GPC) on a WATERS 1515 equipped with a series of PS gel columns as the stationary phase. The samples with a concentration of 1 mg mL⁻¹ in tetrahydrofuran were tested at 40 °C with THF as the eluent and PS as the standard. Fourier transform infrared analysis (FTIR) was recorded from KBr pallets on a Nicolet Magna 5700 FTIR spectrometer. Dynamic light scattering (DLS) was used to determine the apparent hydrodynamic size of the particles using Zetasizer Nano (Malvern, UK) at 25 °C. The concentration of samples was 10 mg mL⁻¹. The contents of gadolinium were measured through inductively coupled plasma atomic emission spectrometry (ICP-AES, Agilent 725ES).

T₁ relaxivity measurement

T₁ relaxivity measurements were performed on a 1.5 T GE SIGNA EXCITE (GE Medical Mixtures, USA) at ambient temperature. The samples were diluted to different concentrations in 1.5 mL centrifuge tubes and imaged collectively with a high-resolution inversion recovery pulse sequence (repeat time (T_R) = 1600 ms, echo time (T_E) = 9 ms, inversion time (T₁) = 50, 100, 200, 300, 400, 600, 700, 900, 1200, 1500 ms; field of view (FOV) = 150 mm × 150 mm, matrix = 320 × 320). The resulting images were analyzed on a pixel-by-pixel basis to a single exponential. These T₁ values were averaged over at least 45 pixels in the centre of each sample and plotted as 1/T₁ vs. Gd³⁺. The slope of this line was the millimolar relaxivity, r₁.

In vitro cytotoxicity

The cytotoxicity of Gd(III)-containing samples were individually tested in the human cervical carcinoma cell line, HeLa, using the MTT assay. In brief, the cells were cultured in 25 mL flasks with Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovineserum (FBS), antibiotics (100 U mL⁻¹ penicillin-G, 100 μg mL⁻¹ streptomycin) at 37 °C in a humidified 5% CO₂-containing atmosphere. For in vitro cytotoxicity assay, the cells were plated into 96-wellplates at a density of 4.0 × 10³ cells per well in 0.1 mL culture medium and allowed to attach for 24 h. Afterwards, the growth media was removed and the cells were washed with PBS. Then, 200 μL DMEM medium was added to each well with the solutions at different concentrations of 20, 40, 60, 80, 100, 150, 200 μM. After incubation for 12, 24, and 48 h, the MTT assay was conducted following the procedure reported previously.³³ The purple formazan in the supernatant was quantified by measuring the absorbance at 595 nm in a microplate reader (SPECTRAmax384, Molecular Devices, USA).

Synthesis of HPEI–t-butyl

HPEI (2.00 g) was dissolved in a mixed solvent of acetonitrile (80 mL) and methanol (10 mL). K₂CO₃ (25.00 g, 0.18 mol) and tert-butyl bromoacetate (9.98 g, 0.05 mol) were added and the resulting mixture was stirred at 40 °C for 3 days. After cooled to room temperature, the solid was removed by filtration and the solvent was removed under reduced pressure to give yellow oil (6.60 g, 98% yield). ¹H NMR (CDCl₃, 400 MHz, ppm) δ 3.18–3.68 (m), 2.52–2.75 (m), 1.44 (s). IR signal of C=O 1730 cm⁻¹.

Synthesis of HPEI–CH₂COOH

HPEI–t-butyl (6.60 g) was dissolved in TFA (20 mL, 0.27 mol) and the solution was stirred for 24 h at room temperature. After evaporation of the solvent under vacuum, the brown oil was washed with diethyl ether for three times to give white solid (3.10 g, 76% yield). ¹H NMR (D₂O, 400 MHz, ppm) δ 3.00–4.23 (m). IR signal of C=O 1640 cm⁻¹.

Synthesis of HPEI–NO

The white solid HPEI–CH₂COOH (3.00 g) was suspended in acetic acid (20 mL) in a reaction flask. Then, hydrogen peroxide (30%, 70 mL) was added, and the solution was stirred at room temperature for 3 days.³⁴ The solvent was evaporated under vacuum and the white product was precipitated in methanol and dried under vacuum (2.25 g, 63% yield). ¹H NMR (D₂O, 400 MHz, ppm) δ 3.64–4.60 (m). ¹⁵N NMR (D₂O, 600 MHz, ppm) δ 116.69, 117.82.

LPEI–NO ligand was prepared with the same method.

Synthesis of EDTA-NO

EDTA-NO was synthesized according to the previous literature.^{26 15}N NMR (D₂O, 600 MHz, ppm) δ 116.96.

Preparation of Gd(III)–chelating complexes

HPEI–NO ligand (106 mg) was dissolved in water at 80 °C and an aqueous solution of GdCl₃·6H₂O (135 mg, 0.36 mmol) was added dropwise to the solution. The reaction mixture was then stirred at 80 °C for 3 days. The pH value of the mixture was finally adjusted to 10 by adding 1 M NaOH in order to precipitate the excess GdCl₃. After that, the mixture was adjusted to neutral solution with 1 M HCl. Finally, the solution was filtered through a 0.22 μm Millipore filter to give the final product (140 mg, 90% yield).

LPEI–NO–Gd(III) complex was prepared with the same method.

HPEI–NO–Y(III) complex was also prepared with the above method. ¹⁵N NMR (D₂O, 600 MHz, ppm) δ 117.82, 118.19, 121.38.

Conclusions

In summary, we have successfully introduced N-oxide structure in hyperbranched polymers as MRI contrast agent, the resulting HPEI–NO–Gd(III) complexes show significant enhancement in relaxivity compared to Gd-DTPA as well as their linear analogues. Meanwhile, the relaxivity improves with the increase of molecular weight. Cytotoxicity assays revealed good cytocompatibility of these complexes, which is essential for in vivo applications. Taking advantage of this strategy, we are able to prepare MRI contrast agent with commercially available hyperbranched polymers as precursors, simple and practical procedures, as well as high relaxivities. It is therefore promising for producing MRI contrast agents in large scale in industry.

Acknowledgements

The authors gratefully acknowledge the financial support from the National Natural Science Foundation of China (21274042) and the Shanghai Leading Academic Discipline Project (B502).

Notes and references

1. P. Caravan, J. J. Ellison, T. J. McMurry and R. B. Lauffer, Chem. Rev., 1999, 99, 2293–2352.
2. E. J. Werner, A. Datta, C. J. Jocher and K. N. Raymond, Angew. Chem., Int. Ed., 2008, 47, 8568–8580.
3. A. J. L. Villaraza, A. Bumb and M. W. Brechbiel, Chem. Rev., 2010, 110, 2921–2959.
4. Z. Qiao and X. Shi, Prog. Polym. Sci., 2015, 44, 1–27.
5. E. Wiener, M. W. Brechbiel, H. Brothers, R. L. Magin, O. A. Gansow, D. A. Tomalia and P. C. Lauterbur, Magn. Reson. Med., 1994, 31, 1–8.
6. H. Kobayashi, S. Kawamoto, S.-K. Jo, H. L. Bryant, M. W. Brechbiel and R. A. Star, Bioconjugate Chem., 2003, 14, 388–394.
7. H. Liu, Y. Chen and Z. Shen, J. Polym. Sci., Part A: Polym. Chem., 2007, 45, 1177–1184.
8. A. Sunder, J. Heinemann and H. Frey, Chem.–Eur. J., 2000, 6, 2499–2506.
9. B. Voit, J. Polym. Sci.: Polym. Chem., 2000, 38, 2505–2525.
10. C. Gao and D. Yan, Prog. Polym. Sci., 2004, 29, 183–275.
11. C. R. Yates and W. Hayes, Eur. Polym. J., 2004, 40, 1257–1281.
12. B. Voit, J. Polym. Sci.: Polym. Chem., 2005, 43, 2679–2699.
13. X. Liu, L. Lin and X. Feng, Acc. Chem. Res., 2011, 44, 574–587.
14. K. Zheng, C. Yin, X. Liu, L. Lin and X. Feng, Angew. Chem., Int. Ed., 2011, 50, 2573–2577.
15. W. Li, X. Liu, F. Tan, X. Hao, J. Zheng, L. Lin and X. Feng, Angew. Chem., Int. Ed., 2013, 52, 10883–10886.
16. W. Li, J. Wang, X. Hu, K. Shen, W. Wang, Y. Chu, L. Lin, X. Liu and X. Feng, J. Am. Chem. Soc., 2010, 132, 8532–8533.
17. Y. Hui, J. Jiang, W. Wang, W. Chen, Y. Cai, L. Lin, X. Liu and X. Feng, Angew. Chem., Int. Ed., 2010, 49, 4290–4293.
18. L. Wang, X. Liu, Z. Dong, X. Fu and X. Feng, Angew. Chem., Int. Ed., 2008, 47, 8670–8673.
19. Z. Yu, X. Liu, Z. Dong, M. Xie and X. Feng, Angew. Chem., 2008, 47, 1328–1331.
20. R. F. Bezerra, D. M. A. Melo, G. Vicentini, K. Zinner and L. B. Zinner, J. Alloys Compd., 2002, 344, 120–122.
21. L. C. Fernandes, J. R. Matos, L. B. Zinner, G. Vicentini and J. Zukerman-Schpector, Polyhedron, 2000, 19, 2313–2318.
22. E. P. Marinho, D. M. A. Melo, L. B. Zinner, G. Vicentini, J. Zukerman-Schpector and K. Zinner, J. Alloys Compd., 2000, 303, 116–120.
23. M. G. Alvarenga, L. B. Zinner, C. A. Fantin, J. R. Matos and G. Vicentini, J. Alloys Compd., 2004, 374, 258–260.
24. C. A. Fantin, M. G. Alvarenga, L. B. Zinner, J. R. Matos and G. Vicentini, J. Alloys Compd., 2004, 374, 362–365.
25. D. M. J. Doble, M. Melchior, B. O'Sullivan, C. Siering, J. D. Xu, V. C. Pierre and K. N. Raymond, Inorg. Chem., 2003, 42, 4930–4937.
26. D. Bayot, B. Tinant, B. Mathieu, J. P. Declercq and M. Devillers, Eur. J. Inorg. Chem., 2003, 737–743.
27. N. Sato, H. Kobayashi, A. Hiraga, T. Saga, K. Togashi, J. Konishi and M. W. Brechbiel, Magn. Reson. Med., 2001, 46, 1169–1173.
28. M. Rohrer, H. Bauer, J. Mintorovitch, M. Requardt and H.-J. Weinmann, Invest. Radiol., 2005, 40, 715–724.
29. É. Tóth, I. van Uffelen, L. Helm, A. E. Merbach, D. Ladd, K. Briley-Sæbø and K. E. Kellar, Magn. Reson. Chem., 1998, 36, S125–S134.
30. D. M. J. Doble, M. Botta, J. Wang, S. Aime, A. Barge and K. N. Raymond, J. Am. Chem. Soc., 2001, 123, 10758–10759.
31. A. R. Menjoge, R. M. Kannan and D. A. Tomalia, Drug Discovery Today, 2010, 15, 171–185.
32. J. Tang, Y. Sheng, H. Hu and Y. Shen, Prog. Polym. Sci., 2013, 38, 462–502.
33. Y. Chen, H. Yang, W. Tang, X. Cui, W. Wang, X. Chen, Y. Yuan and A. Hu, J. Mater. Chem. B, 2013, 1, 5443–5449.
34. D. Bayot, B. Tinant and M. Devillers, Inorg. Chem., 2004, 43, 5999–6005.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra03589b

---