Synthesis and Unprecedented Complexation Properties of β

Article pubs.acs.org/IC

Cite This: Inorg. Chem. XXXX, XXX, XXX−XXX

Synthesis and Unprecedented Complexation Properties of β‑Cyclodextrin-Based Ligand for Lanthanide Ions Pier-Luc Champagne,†,∇ Ceć ile Barbot,‡,∇ Ping Zhang,† Xuekun Han,† Issam Gaamoussi,‡ Marie Hubert-Roux,‡ Gabriel E. Bertolesi,† Geŕ aldine Gouhier,*,‡ and Chang-Chun Ling*,† †

Downloaded via TUFTS UNIV on July 17, 2018 at 00:36:01 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Alberta Glycomics Centre, Department of Chemistry, University of Calgary, 2500 University Drive NW, Calgary, Alberta T2N 1N4, Canada ‡ Normandie Université, COBRA, UMR 6014, FR 3038, INSA Rouen, CNRS, IRIB, IRCOF 1 rue Tesnière 76821 Mont-Saint-Aignan, France S Supporting Information *

ABSTRACT: Here, we report the synthesis and detailed studies on the coordination chemistry of a novel chemically modified polyaminocarboxylate (5) based on β-cyclodextrin (CD) scaffold for lanthanides. The target ligand is prepared in a highly efficient manner (seven total steps) from β-CD using the readily available iminodiacetic acid as a starting material. A propargyl group is attached to the iminodiacetate via Nalkylation, and the obtained derivative is efficiently conjugated to the β-CD scaffold via the copper(I)-mediated 1,3-dipolar cycloaddition. The generated 1,2,3-triazolmethyl residues advantageously provide a competent chelating group while displacing the metal coordination center away from the primary rim of β-CD, to afford the required conformational flexibility. The functional groups from each of the two adjacent glucopyranosyl units of β-CD complete a uniquely created octavalent coordination sphere for lanthanides while still sparing one site for dynamic water coordination. To help study the coordination chemistry of CD ligand 5, we also design a relevant maltoside ligand 6, which faithfully represents one submetal-binding section of ligand 5. Thanks to HRMS and NMR studies, we successfully elucidate the coordination chemistries of synthesized ligands. The octavalent coordination sphere of ligand 5 shows strong binding affinity to lanthanides. By potentiometric titration experiments, ligand 5 is found to bind gadolinium(III), forming 1:1, 1:2, and 1:3 multinuclear complexes with lanthanides, thus possessing great capacity for catalyzing the dynamic water-exchange. Further NMR studies also reveal that the formed ligand 5/ Gd(III) complexes show significantly better abilities to alter T1 relaxivities of coordinated water than DOTA-Gd(III) and also some of the synthetic CD probes reported in the literature.

■

cyclic ligand (log KGd‑DOTA ≈ 25.2), which is ∼3 orders of magnitude higher than the linear analogue (log KGd‑DTPA ≈ 22.5). However, because of the adverse effects5 associated with gadolinium complexes, such as the gadolinium-associated nephrogenic systemic fibrosis (NSF),6,7 since 2006, there has been a trend to disfavor the use of linear DTPA-based CAs, because of their increased risk of undergoing in vivo transmetalation with endogenous ions, which leads to the release of free Gd(III) ions.8 Thus, the continuous development of novel CAs with improved stability and relaxivities is needed. Cyclodextrins (CDs) represent a class of promising scaffolds for the design of a new generation of MRI probes, because of their biocompatibility, large molecular weights, and multivalency, as well as their hydrophobic cavities. During the last two decades, there has been considerable interest to develop CD-based MRI probes.9−27 However, most of these reports

INTRODUCTION The use of contrast agents (CAs) has transformed magnetic resonance imaging (MRI) and allowed to become one of the most powerful and versatile diagnostic tools in modern clinical medicine.1−3 CAs are commonly based on complexes of gadolinium(III), which is a paramagnetic lanthanide that possesses optimal coordination chemistry. For example, the highly toxic metal provides a 9-coordination sphere available for potential dynamic water coordination, resulting in catalytic alternation of relaxivities of proton nuclei of water. This creates significant contrasts to relaxivities of noncoordinated water molecules located in other tissues, leading to improved resolution of acquired images. Currently, ∼35% of MRI scans require the use of a CA. To circumvent the toxicity of Gd(III), polyaminocarboxylate-based ligands4 such as the linear diethylenetriaminepentaacetic acid (DTPA) and cyclic 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA), as well as their derivatives, are used in clinics to form highly stable complexes (1 and 2; see Figure 1) with Gd(III); the determined affinities are remarkably high with the © XXXX American Chemical Society

Received: April 7, 2018

A

DOI: 10.1021/acs.inorgchem.8b00937 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 1. Structures of some commercial CAs: Gd-DTPA(H2O) (1), Gd-DOTA(H2O) (2), and some reported CD-based ligands 3 and 4 for gadolinium(III), as well as new synthetic targets 5 and 6.

arm from adopting optimal geometry for coordination. The second family consists of per-O-alkylations of CD derivatives with acetates. Starting from 3,6-anhydro-α-CD,29 Delangle et al. generated a O2-polyacetate that was reported to bind to one Gd(III) metal with a stability constant (log KGd) of 7.5 and exhibit low toxicity, as well as no nephrotoxicity or hemolysis on a rat model. Gouhier and co-workers have also developed novel MRI probes such as the per-O6-acetate 4 from β-CD via O6-alkylations;30,31 these ligands were found to sequester one Gd(III) metal, and the formed complexes showed favorable alterations of water relaxivity, revealing their ability to form secondary hydration spheres. However, the coordination geometry of these per-O6-acetates with Gd(III) remains poorly understood, and their binding affinities with Gd(III) were also similar to that of the O2-polyacetate.32 Thus, despite the many advantages of using CDs as a convenient scaffold for the design of MRI probes, numerous challenges remain to be overcome before a clinically viable CA based on CDs can be obtained. Here, we report a novel β-CD-based MRI probe 5, decorated with seven copies of iminodiacetates that are connected to the C6 positions of β-CD via a N-(1H-1,2,3triazol-4-yl)methyl linker. To assist our detailed binding studies, we also designed a ligand 6 based on the methyl αmaltoside disaccharide that essentially represents a local copy of the coordination sphere found in compound 5 (see Figure 1). The unique feature of our system is the afforded one extra coordination valency to Gd(III) by the 1H-1,2,3-triazole unit from each arm, to potentially complete a 8-coordination sphere for lanthanide(III), resulting possibly high affinity and high stability of the complex, while still leaving room for the coordination of one water molecule. Moreover, we conducted the first detailed binding studies with Gd(III) and determined the stability constants of formed complexes. Our results revealed great potential of using this novel coordination chemistry to generate high-affinity ligands based on CD scaffolds to sequester lanthanide ions.

were based on a common strategy consisting of either covalently grafting one or multiple units of known chelating motifs such as DTPA or DOTA onto a monomeric9−13 or oligo/polymeric CD14−21 backbone or forming an inclusion complex.11,16,17,21−27 The new derivatives produced generally possess improved relaxivities, because of their reduced tumbling rate, influenced by the markedly increase in molecular weights in the newly generated systems, compared to the commercial CAs with low molecular weights. Logically, all Gd(III) complexes generated using such strategy should possess similar stability as the commercial CAs (1 and 2). On the other hand, another strategy to develop CD-based MRI probes consists of taking advantage of the native macrocyclic geometry to create an innovative coordinating sphere for Gd(III). Such an approach has the potential to generate new MRI probes with improved properties. However, this approach has only reached limited success, because, to date, there have been very few attempts reported in the literature, and no CD derivatives with binding affinity to Gd(III) comparable to DTPA or DOTA have been reported.28−31 The reported literature examples can be classified into two families, based on functionalization. The first was based on per-6-amino-α/β/γ-CDs by carrying out a per-N,N-dialkylation with acetates to afford EDTA-type ligands such as the β-CD analogue 3 (see Figure 1).28 Based on molecular modeling, it was proposed that every two units of iminodiacetates attached to C6 positions of adjacent glucopyranosyl units in CDs were used to complex one metal ion, generating a coordination sphere of six valencies for each Gd(III) center (two amino groups and four acetates). Unfortunately, the binding affinities of these Gd complexes were not reported. We estimate that they might likely be much lower than those of Gd-DOTA or Gd-DTPA complexes, because of the lower number of coordination per Gd(III) center, as well as potential tension introduced by the CD scaffold, because of the proximity of the metal coordination center to the primary rim of CDs, preventing each chelating B

DOI: 10.1021/acs.inorgchem.8b00937 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Scheme 1. Efficient Synthetic Route To Target Compound 5a

Reagents and conditions: (a) MeOH/AcCl, 60 °C, overnight (∼100%); (b) propargyl bromide/DIPEA/CH2Cl2, 40 °C, overnight (65%); (c) CuI/DIPEA/acetone, 55 °C, overnight (77%); and (d) NaOMe/MeOH/CH2Cl2, then NaOH/H2O-MeOH, 70 °C, overnight (84%).

a

■

SYNTHESIS The desired target ligand 5 was effectively synthesized through a highly convergent route illustrated in Scheme 1. The commercially available iminodiacetic acid 7 was first Omethylated in anhydrous methanol at 60 °C overnight under strong acidic condition to provide the bis(methyl ester) 8, isolated in the hydrochloric acid salt form (∼100% yield). The salt 8 was then N-alkylated with propargyl bromide in refluxing dichloromethane in the presence of excess N,N-diisopropylethylamine (DIPEA); this afforded the corresponding tertiary amine 9, isolated by column chromatography on silica gel in good yield (65%). The next step is the key conjugation between compound 9 and the fully acetylated per-6-azide 10, synthesized according to the literature from per-6-bromo-βCD in two steps.33 The 1,3-dipolar cycloaddition was carried out using 1.5 equiv of alkyne per azide in refluxing acetone using a catalytic amount of copper(I) iodide as a catalyst in the

presence of DIPEA; the reaction afforded the desired per-C6substituted conjugate 11, isolated in 77% yield by column chromatography using a mixture of dichloromethane−methanol−triethylamine (98.5:0.5:1) as eluent. The purity and identity of compound 11 was confirmed by 1 H and 13C nuclear magnetic resonance (NMR) spectroscopy and electrospray high-resolution mass spectrometry (ESIHRMS). For example, a single 1,2,3-triazole peak was observed at 7.73 ppm, combined with the anomeric protons of all glucopyranoses being observed at 5.53 ppm as a doublet (J = 3.3 Hz); this was further confirmed with the observed downfield shifts of all H-6a and H-6b protons from below 3.8 ppm (compound 10) to 4.93 and 4.83 ppm, respectively, due to the significant deshielding effect of the newly formed aromatic 1,2,3-triazole rings. Furthermore, the ESI-HRMS spectrum in positive ion mode showed the expected doubly charged ion at m/z 1646.5788, corresponding to the expected formula: C 133 H 184 N 28 O 70 (M+2H) 2+ (calculated m/z: C

DOI: 10.1021/acs.inorgchem.8b00937 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

formamide (DMF). Followed by a per-O-acetylation with acetic anhydride in pyridine, the crude compound 14 was then directly substituted by sodium azide in anhydrous DMF at 80 °C to afford the corresponding 6,6′-diazide 15, which was isolated by column chromatography on silica gel as an anomeric mixture (α/β ≈ 1:1, ∼48% yield over three steps). With the help of HPLC on normal phase silica gel column, and using a gradient of 0 → 5% ethyl acetate−dichloromethane as the eluent, the desired α-anomer 15 was obtained in pure form. We subsequently conjugated the pure α-6,6′-diazide 15 to the previously synthesized alkyne 9 using similar conditions as above; this afforded the desired conjugate 16 in 71% yield. Following sequential deprotection steps as above, the pure ligand 6 was obtained by gel filtration on Sephadex LH-20 (92% yield). The structures of both compounds 5 and 6 were confirmed by ESI-HRMS in negative ion mode. For instance, for CD conjugate 5, we observed a triply charged ion at m/z 834.5876, which corresponded to the expected formula: C91H123N28O56 (M-3H)3− (calculated m/z: 834.5885). The persubstitution of the compound 5 was confirmed by the observed axial symmetry in the 1H NMR spectrum. For instance, we observed a singlet at 8.21 ppm and a doublet at 5.18 (J = 2.8 Hz), which were assigned to the one set of 1,2,3-triazole protons and anomeric protons, respectively (Figure 2), found in compound 5. Similarly, for the maltoside 6, we observed the related ion corresponding to the expected formula C27H39N8O17 (M−H)− at m/z 747.2422 (calculated m/z: 747.2439). In the 1H NMR spectrum of the compound 6 in D2O (see Figure 2), we observed two singlets related to the 1,2,3-triazole peaks at 8.08 and 8.03 ppm, and two anomeric protons at 5.31 and 4.56 ppm (doublets), which respectively correspond to the two sets of chelating functionalities attached to the two α-glucopyranosyl units in the molecule. For the other types of protons, each of them also appeared as two sets of signals.

1646.5844). The fully protected intermediate 11 was then deprotected by first performing a Zemplén transesterification to remove the 14 O-acetates on the secondary face, followed by a saponification reaction of all of the remaining methyl esters using NaOH in a mixture of deionized water and methanol. The obtained crude product was finally purified by gel filtration on Sephadex G-15 to provide the desired target ligand 5 in high yield (84% yield). To quickly prepare the disaccharide ligand 6, the maltose 12 was first subjected to a Fisher glycosylation in refluxing anhydrous methanol in the presence of Amberlite IR-120 (H+) (see Scheme 2). The crude mixture was then purified by Scheme 2. Synthesis of Methyl α-Maltoside 6a

a

Reagents and conditions: (a) MeOH/Amberlite 120, reflux, overnight (30%); (b) PPh3/NBS/DMF, 70 °C, overnight, then Ac2O/Pyridine, overnight; (c) NaN3/DMF, 80 °C, 18 h (48%, 3 steps); (d) CuI/DIPEA/acetone, overnight (71%); and (e) NaOMe/ MeOH 4 h, then NaOH/H2O, reflux, 6 h (92%).

■

STUDIES OF PROTONATION OF LIGAND 5 The structure of ligand 5 contains numerous protonation sites, including 14 carboxylates, 7 tertiary amines, and 7 1,2,3triazole rings (see Scheme 3). Based on pKa values of related functional groups such as N-methyl-1,2,3-triazolium (1.25),34 acetic acid (∼4.76), and triethylammonium (∼10.75), we can ignore the protonation of the 1,2,3-triazole rings in the molecule, since it only becomes relevant at very acidic pH

column chromatography on silica gel to afford the desired methyl glycoside 13 as a 1:1 anomeric mixture (∼30% yield). The two primary hydroxyl groups were then converted to the 6,6′-dibromide at 70 °C using N-bromosuccinimide (NBS)− triphenylphosphine as the reagent in anhydrous dimethyl

Figure 2. 1H NMR spectra of synthesized β-CD ligand 5 and maltoside 6 (600 MHz, D2O, 25 °C). D

DOI: 10.1021/acs.inorgchem.8b00937 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Scheme 3. Ligand 5 Can Undergo Stepwise Protonationsa

a

A few examples of various protonated states of ligand 5 are shown.

Figure 3. (Left) Potentiometric titrations of ligand 5 (L), T = 298 K, [L] = 0.0625−1 μmol, [extra-HCl] = 0−37.3 μmol in NMe4Cl (0.1 M); total initial volume: 4 mL; buret: [NMe4OH] = 0.05 M. (Right) Calculated distributions of different protonated species at different pH values.

In order to gain insight on the involvement of the carboxylates in the coordination with the metallic center, attenuated total reflection (ATR) Fourier transform infrared (FTIR) spectroscopy studies were performed on ligand 5. It is well-known that deprotonation of the carboxylic acid would lead to the absence of any strong bands around 1700 cm−1 in the IR spectrum; the resulting carboxylate typically presents

values. Thus, in less-acidic solutions, ligand 5 contains essentially 21 protonation sites, designated as [LH21]7+; a stepwise deprotonation process would lead to other intermediate stages, such as the neutral species LH14, the fully deprotonated intermediate from carboxylates [LH7]7−, and finally the completely deprotonated ligand L14− from both carboxylates and ammoniums. E

DOI: 10.1021/acs.inorgchem.8b00937 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry two vibrational modes around 1600 and 1400 cm−1, because of the symmetric (νs_COO) and antisymmetric (νas_COO) stretching modes. The ATR-FTIR spectrum of ligand 5 (see Figure S46 in the Supporting Information) showed two stretching bands at 1619.7 and 1396.1 cm−1, indicating that the carboxylic functionalities of the isolated ligand 5 was indeed fully deprotonated. Potentiometric titrations were performed to study the deprotonation of ligand 5 (see Figure 3). The initial solution of ligand 5 alone gave pH ∼8.4; this value correlates well with extensive deprotonation observed in the ATR-FTIR spectrum. Therefore, we gradually added a solution of HCl in trimethylammonium chloride (0.1 M) to the solution to obtain a series of solutions with varied HCl/5 ratios (0 → 597), while maintaining the total volume constant. To each prepared solution, a titrant solution of tetramethylammonium hydroxide (0.05 M) was added while the pH of the solution recorded; this allowed us to obtain a series of titration curves (Figure 3). As can be seen, with increasing HCl/5 ratios, the form of the obtained curves becomes increasingly similar to that of pure HCl, while with decreasing HCl/5 ratios, the form of the curves becomes more representative to ligand 5. The protonation constant βh for the following equilibrium is defined by eq 1, where L represents the fully deprotonated form (L14−) of ligand 5 (LH14). H is the proton (charges are omitted). L + hH F LHh

βh =

values of ammonium and carboxylic acid functional groups, this pKa could be ascribed to the [LH7]7− species. pKa values lower than 6.20 could be assigned to carboxylic acid moieties, while pKa values of >6.20 could be attributed to ammonium functionalities. Thus, the major species in the solution of neutral pH should correspond to one with complete deprotonation of carboxylic acids, while the first ammonium group was yet quite totally deprotonated [LH6]8− (pKa ∼7.59), which exists in equilibrium with [LH7]7− (pKa ∼6.20) and [LH5]9− (pKa ∼7.93). More alkaline solutions revealed pKa values of 8.07, 8.56, 9.08, and 8.80, which were assigned to [LH4]10−, [LH3]11−, [LH2]12−, and [LH]13− respectively. Normally, the successive pKa values of the Brönsted pairs increase according to the successive dissociation into a polyprotic acid. Nevertheless, the delta pKa values of the pairs [LH 2]12−/[LH]13−, [LH3]11−/[LH 2]12−, [LH4]10−/ [LH3]11−, and [LH5]9−/[LH4]10− differ from h “statistical factor” pKn+1 − pKn = 0.6, suggesting that the involved dissociable groups interact with each other. On the other hand, in more acidic solutions, three pKa values were determined 4.99, 3.63, and 2.57which were assigned to be the sequential protonated species from [LH7]7−, namely, [LH8]6−, [LH9]5−, and [LH10]4−, respectively. The determined pKa value experiences a rapid decrease from 6.20 to 2.57, as each successive protonation produces a less negatively charged carboxylate, reducing charge−charge repulsion between adjacent carboxylates. pKa values corresponding to highly protonated species ranging from [LH11]3− to [LH21]7+ were found to be lower than < 2, which is a pH value that is too low to allow their determination. Based on the determined pKa values, a diagram of percentage distributions of different protonated species according to pH, is shown in Figure 3.

[LHh] [H]h [L]

(1)

Since dissociation constants Ka are commonly defined by the expression Ka =

[H][LHh − 1] [LHh]

■

(2)

it is evident that pK a = log βh − log βh − 1

(3)

Table 1 reports all the protonation constants determined by titration curve refinement and the corresponding acidity Table 1. Some Determined Stepwise Protonation Constants of Ligand 5a species

log βLHh

pKa

LH LH2 LH3 LH4 LH5 LH6 LH7 LH8 LH9 LH10 LH11−LH21

8.80 17.88 26.44 34.51 42.44 50.03 56.23 61.22 64.85 67.42

8.80 9.08 8.56 8.07 7.93 7.59 6.20 4.99 3.63 2.57 2.

coordination, we use the pGd value of the 3:1 Gd(III)-ligand 5 complex (6.87) to compare with the pGd value of Gd(III)DOTA (12.1), because it has the minimal numbers of free donor atoms available for coordination. As can be seen, the pGd value of 3:1 Gd(III)-ligand 5 is significantly lower than the pGd value of 1:1 Gd(III)-DOTA, suggesting that the ligand 5 forms less-stable complexes with Gd(III) than DOTA. It is important to note that the thermodynamic stability of gadolinium complexes is not the only parameter to predict the in vivo toxicity of gadolinium complexes of ligand 5. The

presence of other endogenous metal ions, such as Zn(II), Fe(II), Cu(II), and Ca(II) could compete with the complexation of Gd(III), resulting in transmetalation.41,42 Further studies on kinetic stability of the Gd(III)-ligand 5 complex must be performed to probe the suitability of ligand 5 for use as an MRI probe.

■

WATER COORDINATION NUMBER The hydration number was determined using the wellestablished luminescence method on the corresponding Eu3+ J

DOI: 10.1021/acs.inorgchem.8b00937 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Table 2. Some Determined Stepwise Protonation Constants of Ligand 5a species Mononuclear GdLH−1 GdL GdLH GdLH2 GdLH3 GdLH4 GdLH5 GdLH6 following Dinuclear Gd2LH−2 Gd2LH−1 Gd2L Gd2LH Gd2LH2 Gd2LH3 Gd2LH4 Gd2LH5 Gd2LH6 following Trinuclear Gd3LH−3 Gd3LH−2 Gd3LH−1 Gd3L Gd3LH Gd3LH2 Gd3LH3 Gd3LH4 following

mlh

log βmlh

11−1 110 111 112 113 114 115 116

16.08 25.09 33.74 41.95 50.04 56.97 60.51 64.00

22−2 22−1 210 211 212 213 214 215 216

17.71 27.15 36.28 44.13 51.38 56.33 60.27 63.54 66.45

31−3 31−2 31−1 310 311 312 313 314

13.80 23.91 33.83 43.91 50.70 55.36 60.09 64.99

pKa

9.01 8.66 8.21 8.09 6.93 3.54 3.49