Investigations into the Effects of Linker Length Elongation on the Behaviour of Calcium-responsive MRI Probes

into the Effects of Linker Length Understanding the relationship between chemical structure and the effectiveness of bioresponsive magnetic resonance imaging contrast agents can offer help to identify key components required for the future development of such probes. Here, we report the development and characteristion of two novel monomeric bifunctional chelators, whose paramagnetic metal complexes can serve as calcium-responsive smart contrast agent (SCA). Specifically, relaxometric titrations, luminescence lifetime measurements, NMR studies and NMR diffusion experiments were carried out to assess the behaviour of each system. Overall, our findings demonstrate the impact of subtle changes to the structure of such probes, affecting a range of properties and their coordination behaviour. Through the understanding of such changes, fine tuning of future SCA designs which show optimal changes in relaxivity can be achieved. ABSTRACT Understanding the relationship between chemical structure and the effectiveness of bioresponsive magnetic resonance imaging (MRI) contrast agents can offer help to identify key components required for the future development of such probes. Here, we report the development and characteristion of two novel monomeric bifunctional chelators whose paramagnetic metal complexes can serve as calcium-responsive smart contrast agent (SCA). Specifically, relaxometric titrations, luminescence lifetime measurements, NMR studies and NMR diffusion experiments were carried out to assess the behaviour of each system. Novelty was achieved through the extension of the linker between the 1,4,7,10-tetraazacyclododecane-1,4,7-tris(methylenecarboxylic) acid (DO3A) unit and the ethylenediamine tetraacetic acid (EGTA)-derived Ca-responsive moiety. Relaxometric titrations of both systems, GdL 1 and GdL 2 , showed an increase in r 1 and r 2 relaxivity upon Ca 2+ addition, with the derivative with the longer linker showing a greater overall change. The hydration states of the europium analogues were assessed revealing a higher initial hydration state for Diffusion ordered NMR spectroscopy (DOSY) revealed negligible changes in the diffusive properties of both systems upon the addition of Ca 2+ , while NMR studies of Y 3+ , Yb 3+ and Eu 3+ analogues provided further insights into the structural behaviour of the linker unit in both the unsaturated and Ca-saturated states. Overall, our findings demonstrate the impact of subtle changes to the structure of such probes, affecting a range of properties and their coordination behaviour. Through the understanding of such changes, fine tuning of future SCA designs which show optimal changes in relaxivity can be achieved.


Introduction
Magnetic resonance imaging (MRI) is an essential diagnostic imaging technique used widely in clinics to provide three dimensional soft tissue images at high spatial resolution. In recent years, MRI has been a primary focus in the field of molecular imaging due to attempts to develop methods for the dynamic visualization of biological processes. [1][2][3][4] The advancement of these socalled functional methods and the ability to monitor such changes can provide crucial information such as the function of tissues or organs in pathological states. Typically, MRI can be performed with or without the addition of a contrast agent (CA) unlike other imaging techniques. The inclusion of such a CA in MRI can provide greater specificity through the shortening of the T 1 and T 2 relaxation times of the protons in the immediate vicinity of the CA.
Differing from conventional CAs, 'smart' or bioresponsive smart contrast agents (SCAs) are suitable for functional MRI (fMRI) studies and have been applied to monitor different environmental states as their MR properties fluctuate depending on their microenvironment.
Many SCAs have been developed for a variety of functions including the monitoring of pH, specific metal ions (Ca 2+ , Zn 2+ ) and enzymes. 5 Typically, they are based on cyclen scaffolds complexed with paramagnetic ions, and possess protonable groups (i.e. pH sensors), an additional and specific chelator for a desired analyte (e.g. Ca 2+ , Zn 2+ ) or a part of the molecule that can be affected by the enzyme activity. 6 As the mechanism of action, usually SCAs are either considered as 'off-on' or 'on-off' systems, meaning they alter their relaxometric properties to provide a significant difference between two different environmental states. 7 One of the preferred targets is Ca 2+ due to its significance in biological processes, including a substantial involvement in critical cellular signalling processes, therefore making it an important target to monitor. [8][9] Many attempts to develop Ca-responsive SCAs have been pursued over recent years resulting in the production of a variety of molecules ranging from smaller monomeric compounds to larger nano-sized derivatives. 3,[10][11] The desire to understand the relationship between SCA structure and relaxometric enhancement has brought about a series of studies investigating such effects thus providing significant insights into the behaviour of these systems. [12][13] Specifically, the interaction of carboxylates from the Ca-binding chelate with the paramagnetic metal ion or the roles of the two geometrical isomers, square antiprismatic (SAP) and twisted square antiprismatic (TSAP), of 1,4,7,10-tetraazacyclododecane-1,4,7tris(methylenecarboxylic) acid (DO3A) macrocyclic chelators have revealed interesting coordination properties of these responsive systems. Furthermore, the influence of the distance between the MR reporting and bioresponsive units has shown to be significant in determining the relaxometric behaviour. Previous comparisons of bismacrocyclic Ca-responsive SCAs with either an ethyl or propyl linker between the DO3A MR reporting unit and the EGTA-derived Ca 2+ chelator revealed such an effect. 14 The propyl derivative showed an increased relaxometric enhancement (i.e. increase in longitudinal r 1 relaxivity in presence of Ca 2+ ) versus the ethyl version with slightly different starting relaxivities and hydration states of the lanthanide complex. A more advanced coordination study with ethyl and propyl linker SCA model compounds demonstrated further the significant impact of the linker distance on the coordination properties of the complex. 15 Here, it was evaluated that the shorter ethyl linker is likely to be too close to the paramagnetic metal centre hindering the interaction of the carboxylate group with Gd 3+ .
Recently we reported a study where a series of G4 polyamidoamine (PAMAM) dendrimers conjugated to monomeric SCAs exhibited significant differences in the observed relaxometric properties. Amongst the studied nano-sized conjugates with different monomers, those with the butyl and pentyl linkers connecting the common DO3A-and EGTA-derived chelators showed the greatest changes and outlook for potential future application. 16 Intrigued by these results, we embarked on a detailed structural study of these two newly developed monomeric systems.
Firstly, we prepared different pairs of diamagnetic and paramagnetic complexes, which allowed the initiation of different types of coordination chemistry studies. Through the use of a variety of techniques such as 1D and 2D NMR, luminescence lifetime measurements, diffusion ordered NMR spectroscopy (DOSY) and proton relaxometric titrations, we aimed to characterize the newly developed probes and gain an insight into how subtle structural differences impact on their relaxometric and coordination behaviour.

Results and discussion
Synthesis and Complexation of Ligands. Protected nitro compounds 1 and 2 were synthesized following previously reported procedures. 16 Subsequent deprotection with formic acid at 60 °C overnight yielded ligands L 1 and L 2 (Scheme 1). Complexation reactions of each ligand with either Yb 3+ , Eu 3+ , Gd 3+ or Y 3+ were all carried out under identical conditions. Specifically, the ligands L 1 and L 2 were dissolved in water and the pH adjusted to 7 through the addition of a 0.1 M NaOH solution. A slight excess of the required lanthanide hydrate was added while maintaining the pH at 7 through further addition of 0.1 M NaOH, which after the final addition was stirred overnight at room temperature. The removal of excess lanthanide was achieved by repeated treatment of the solution with Chelex® before filtering and lyophilizing to yield the final complexes, LnL 1 and LnL 2 (Ln = Gd 3+ , Yb 3+ , Eu 3+ or Y 3+ ). found to be 2.49 mM -1 s -1 , which is slightly lower than that of the propyl derivative and other SCAs developed by us previously (Figure 1a). 11,17 Upon saturation with Ca 2+ , r 1 increases to 4.96 mM -1 s -1 (99% increase). Concurrently, the change in r 2 over the same [Ca 2+ ] range was found to increase to the same relative extent, specifically from 3.19 to 6.62 mM -1 s -1 (Figure 1b). In comparison, GdL 2 gave an initial r 1 of 3.01 mM -1 s -1 , which is in line with that of previously investigated SCAs (Figure 1a). 11 However, r 1 reached a value of 6.97 mM -1 s -1 (131 % increase) upon saturation with Ca 2+ . The change in r 2 was determined to be from 3.92 to 9.31 mM -1 s -1 (138 % increase). Concurrently, both r 2 /r 1 ratios remained constant at the studied Ca 2+ concentrations, indicating that the formation of the Ca-GdL 1,2 complex is likely not followed by any size changes between the paramagnetic species in absence and presence of Ca 2+ ( Figure S1 in ESI †). 17 It can also be observed that the initial relaxivity values in the absence of Ca 2+ are lower for GdL 1 than the other complexes studied, indicating a lower hydration state of the complex (see below) and a changed coordination environment of the carboxylate from the EGTA-derived component. Clearly, the linker length between the MR reporting moiety and the bioresponsive unit has significant effects on the relaxometric behaviour of the SCA.
NMR studies. In order to understand and observe how the addition of Ca 2+ impacts the structure and coordination environment of both derivatives, we conducted a series of high-resolution 1D and 2D NMR experiments at 800 MHz. Specifically, a series of diamagnetic and paramagnetic complexes were synthesised and studied at 20 mM concentration and pD 7.0. 1 H, COSY and HSQC of the complexes were recorded, the peaks were assigned and the behaviour of the specific groups in the absence and presence of Ca 2+ was followed. We first assigned the relevant peaks that exhibited changes at different conditions (i.e. absence or presence of Ca 2+ ), excluding the resonances of the methylene groups neighbouring the amines of the EGTA-derived chelator. These were not observable in the recorded spectra, likely due to exchange processes related to the partial deprotonation of the amine groups at the studied pD. 13 Furthermore, we noticed that the differences between the two states (without Ca 2+ / with Ca 2+ ) exist only in the linker between the MR reporting and bioresponsive moiety; hence more detailed analysis followed only for this structural region. Consequently, the HSQC spectra of the diamagnetic derivatives (YL 1 and YL 2 ) were the most informative, later on the obtained results were compared with the paramagnetic complexes EuL 1,2 and YbL 1,2 .
For the butyl derivative, YL 1 , the closest (A 1 ) and furthest (D 1 ) methylene units to the macrocycle ring show a minor shift upon the addition of Ca 2+ (Figure 2). On the other hand, the 'inner' methylene units of the linker (B 1 and C 1 ) show significant changes upon Ca 2+ binding.
Moreover, for both groups a single major signal is observable in the unsaturated state, which splits into two clear signals with equal intensity when saturated with Ca 2+ . Exchanging Y 3+ for paramagnetic ions (Eu 3+ , Yb 3+ ) shows the same trends. This time, the signal splitting for B 1 and C 1 methylene groups upon Ca 2+ binding is accompanied with the partial or complete signal loss of the A 1 methylene in EuL 1 and YbL 1 , respectively ( Figure S2 tin ESI †).
Similarly for YL 2 , the CH 2 unit closest to the macrocyclic ring (A 2 ) shows a minor shift upon the addition of Ca 2+ (Figure 3). Moving further away from the macrocyclic ring, one signal is observable for the B 2 methylene group prior to Ca 2+ addition, which then splits to form a new additional signal with lower intensity upon Ca 2+ binding. C 2 methylene shows similar splitting behaviour to B 2 when Ca 2+ binds, however both new signals are present with similar intensities.
Moving towards the amide unit, the D 2 methylene group initially presents as two spots which then merge towards one with the addition of Ca 2+ , while the furthest methylene group in the linker, E 2 , experiences a minor change. As in the case for the complexes with the butyl linker, analysis of paramagnetic complexes reveals loss of the signal for the A 2 methylene group in both EuL 2 and YbL 2 upon Ca 2+ binding ( Figure S2 in ESI †). Obviously, the difference in the observed spectra of the studied peaks across the used metals originates from their different paramagnetic effects. For example, the signals for A 1 /A 2 methylene groups, which are observable in YL 1-2 after the addition of Ca 2+ , disappear in the cases of Ca 2+ addition to EuL 1-2 and YbL 1-2 . We assume that a combination of the paramagnetic relaxation effect (shortening of T 1 and T 2 relaxation times) due to the proximity of the A 1 /A 2 groups to the lanthanide and the increased conformational exchange leads to the overall signal loss that cannot be detected via NMR anymore.
More importantly, the signal splitting and shifting of peaks seen in the B 1,2 , C 1,2 and D 2 methylene groups suggest a substantial change in their environment between both the starting and saturated Ca 2+ states. As these units are not in direct contact and should not participate directly in the binding of the added Ca 2+ , the movement of such signals indicates a change in their environment and coordination behaviour of the complex. The mechanism of action for these Ca 2+ -responsive SCAs has previously been investigated. 13 Specifically, in the absence of Ca 2+ , the closest carboxylate of the EGTA-derived moiety binds to the metal centre restricting water access. In the presence of Ca 2+ , this carboxylate preferentially binds to Ca 2+ thus increasing the access of water to the complex. The results obtained here are in line with this theory, as it can be envisaged that the 'off-state' will present in a conformation which can be considered as a ringlike structure through which the EGTA-derived carboxylate is coordinated to the lanthanide of the macrocycle. Subsequent Ca 2+ addition triggers the breaking of this arrangement due to its higher affinity for Ca 2+ vs Gd 3+ to form a new structure/conformation, which is highlighted through the shifting and splitting of signals in the HSQC spectra. The observed changes of the signals from the linker CH 2 units indicate a noteworthy change in the coordination environment around the macrocyclic chelator between the two states and further evidences a conformational change in the responsive mechanism previously described. Moreover, the existence of two signals upon the addition of Ca 2+ supports the presence of two SAP and TSAP geometrical isomers. Since the signals of the C=O group of the amide could not be observed in any of the performed NMR experiments, we could not draw any conclusion on its fate, and whether this carbonyl does or does not coordinate to the metal centre. Indeed, the luminescence lifetime experiments revealed values that suggest an equilibrium of non-and mono-hydrated species (see below); combined with the absence of NMR signals, our findings suggest that the carbonyl interacts with the metal ion throughout in both of the studied conditions (with or without added Ca 2+ ).

NMR diffusion measurements. The diffusion of molecules can significantly describe their
behaviour in solution and could potentially show any aggregation present. Diffusion coefficients for each derivative was therefore assessed prior to and after the addition of Ca 2+ with DOSY on the Eu 3+ analogues (EuL 1 and EuL 2 ) at three different concentrations (20, 10 and 5 mM). The diffusion coefficients obtained for the two systems studied here were both in agreement with similar monomeric species previously studied (Table 1). [17][18]  Generally, decreasing the complex concentration led to a slight increase in diffusion coefficient which is naturally expected. As well, the diffusion of EuL 2 was slightly slower than that of EuL 1 , which is logical due to minor difference in their size and greater linker length in EuL 2 .
When comparing samples of identical concentrations upon Ca 2+ addition (2 equiv.), similar, or a slight increase in diffusion coefficient was also observed. This is indicative of a change in conformation expected upon Ca 2+ binding and faster diffusion of these complexes in solution.
These tiny observed changes in diffusion along with the absolute values indicate systems which interact with Ca 2+ ; however, its addition leads to conformational changes that cause slightly faster diffusion of the species in solution for both studied complexes, while likely excluding the formation of aggregates.
Luminescence experiments. Luminescence emission lifetime measurements were performed to assess the hydration states of the complexes and further understand the coordination environment of the lanthanide ion. As previously well established for this class of SCAs, the hydration of the inner sphere of the complex is expected to increase upon addition of the analyte. 12,19 Measurements were carried out with the EuL 1-2 (5 mM) analogues in both H 2 O and D 2 O and the hydration number (q) was calculated ( Table 2) following a previously reported method. 20 Consequently, the calculated hydration numbers for both EuL 1-2 complexes increase upon Ca 2+ binding, complimentary to that previously observed with other DO3A-based Caresponsive SCAs. [13][14] The length of the linker impacts the initial inner sphere hydration state of the complex. The shorter butyl derivative, EuL 1 , is less hydrated in the initial state compared to that of EuL 2 , which explains the initial r 1 relaxivity value observed in the relaxometric titrations previously discussed (Figure 1). The increased initial hydration state of EuL 2 additionally also indicates a lesser interaction between the EGTA-derived carboxylate and Eu 3+ compared to EuL 1 , possibly resulting from an increased distance between these two moieties. Otherwise, reasons involving an altered conformational arrangement or the change in ratio of the SAP and TSAP isomers could provide possible explanations.

Conclusions
To conclude, we have designed, synthesised and characterised a new pair of Ca-responsive SCAs with altered linker lengths between the MR reporting and bioresponsive moieties.
Relaxometric, NMR, DOSY and luminescence lifetime studies were performed to analyse and evaluate the coordination behaviour of both systems. Throughout the studies, the sensitivity of such molecules to discrete structural changes is apparent. As the linker length increased, the initial hydration state of EuL 2 was observed to be higher than in EuL 1 , resulting in a greater starting relaxivity. The subsequent change in relaxivity upon Ca 2+ addition varied significantly depending on the linker length. Specifically, the overall r 1 and r 2 relaxivity change for GdL 2 was found to be 131 and 138 % respectively; while for the shorter system, GdL 1 , changes of 99 and 107 % were observed for both r 1 and r 2 respectively. Results from the DOSY experiments,  General procedure for the synthesis of LnL 1-2 complexes. Ligands

High resolution NMR experiments
All high resolution NMR spectra were acquired on Bruker AVIII-800 spectrometer,

NMR diffusion measurements
Samples of EuL 1-2 with and without 2 equivalents of Ca 2+ were dissolved in D 2 O and the pD adjusted to 7.0. Diffusion coefficient determination was performed using 2D -Diffusion Ordered NMR Spectroscopy (DOSY). Measurements were performed at 298 K with 20, 10 were calculated through the analysis of individual peaks in the aromatic region. Each peak used in the determination provided the diffusion coefficient that was within an error of 0.05 x 10 -10 m 2 s -1 . The reported diffusion coefficients in Table 1 are the averages of each of the peaks from the three independent measurements with their standard deviation.

Luminescence lifetime measurements
Luminescence lifetime measurements were performed with EuL 1-2 at 5 mM [Eu 3+ ] in D 2 O and H 2 O (298 K, pH 7.4, HEPES). The Eu 3+ ion was directly excited and the emission intensity was recorded with a 10 µs resolution. The excitation and emission slits were set at 5 nm. In total, three independent measurements each with 25 scans were performed to obtain the data set. The obtained curves were fitted with a first order exponential decay with an r 2 = 0.99. The resulting q values were then calculated using Eq. (2).