Tuning the anion binding properties of lanthanide receptors to discriminate nucleoside phosphates in a sensing array

The development of synthetic receptors for the selective binding and discrimination of anions in water requires an understanding of how anions interact with these synthetic receptors. Molecules designed to differentiate nucleoside phosphate anions (e.g. ATP, ADP, GTP, GDP, UDP) under physiological conditions could underpin exciting new sensing tools for biomedical research and drug discovery, but it is very challenging due to the similarities in anion structure, size and charge. We present a series of lanthanide-based anion receptors and establish key structural elements that impact on nucleoside phosphate anion binding and sensing. Structural evidence of anion binding using X-ray crystallographic and NMR data, supported by DFT calculations indicate the binding modes between the lanthanide complexes and certain phosphoanions, revealing a bidentate (α-, γ-) binding mode to ATP. We further use four of the receptors to allow discrimination of eight nucleoside phosphate anions in the first array-based assay using lanthanide complexes, taking advantage of the multiple emission bands and long emission lifetimes associated with luminescent lanthanide complexes.


General considerations
H, 13 C, COSY and HMQC NMR spectra were recorded in the stated deuterated solvent on a JEOL ECS-400 spectrometer ( 1 H at 399.782 MHz, 13 C at 100.525 MHz), or a Bruker Advance Ultra-Shield 400 spectrometer ( 1 H at 400.134 MHz, 13 C at 100.624 MHz) at 293 K. Chemical shifts are expressed in ppm and are adjusted to the chemical shift of the residual NMR solvent resonances (CDCl3: 1 H δ = 7.26 ppm, 13 C δ = 77.16 ppm, CD3OD: 1 H δ = 3.31 ppm, 13 C δ = 49.00 ppm or D2O: 1 H δ = 1.56 ppm). The coupling constants are expressed in Hz.
Liquid Chromatography Electrospray Mass spectra were recorded on a Shimadzu Prominence LC system with a Shimadzu SPDM20A Photodiode Array Detector, a Shimazu CTO-20A column oven, Shimadzu SIL-20A autosampler and a Shimadzu LCMS 20 mass spectrometer controlled using LabSolutions software. The system operates in positive or negative ion mode, with acetonitrile as the carrier solvent. The flow rate was maintained at 1.5 mL/min over a gradient of 5 to 95% acetonitrile (0.1% formic acid) in water (0.1% formic acid) for 8 minutes. High resolution mass spectra were recorded using a Thermofisher Q-Exactive orbitrap mass spectrometer.
Column chromatography was performed using flash silica gel 60 (particle size 40-63 microns) purchased from Apollo scientific. Thin layer chromatography (TLC) was performed on aluminium sheet silica gel plates with 0.2 mm thick silica gel 60 F254 using the stated mobile phase.
Preparative RP-HPLC was performed using a Waters 2489 UV/Visible detector performed at 254 nm, a Waters 1525 Binary HPLC pump controlled by the Waters Breeze 2 HPLC system software. Separation was achieved using a semi-preparative XBridge C18 (5 µm OBD 19 × 100 mm) column at a flow rate maintained at 17 mL min -1 . A solvent system composed of either water (0.1% formic acid) / methanol (0.1% formic acid) or water (50 mM NH4HCO3) / acetonitrile was used over the stated linear gradient (usually 0 to 100% organic solvent over 10 min). Analytical RP-HPLC was performed using a XBridge C18 5 µm 4.6 × 100 mm at a flow rate maintained at 2.0 mL min -1 using the same gradients and solvents.
The organic phases were combined, dried (magnesium sulfate) and the solvent removed under

Optical spectroscopy
UV/Vis absorbance spectra were measured using a Shimadzu UV-1800 UV-spectrophotometer.
Emission spectra were recorded on an SPEX Fluoromax luminescence spectrometer using dM300 Where I: intensity at time, t, following excitation; A0: intensity when decay has ceased A1: pre-exponential factor; k: rate constant for the depopulation of the excited state.
Plate reader data was obtained on a BMG Labtech CLARIOstar microplate reader in black Fisherbrand™ 384-well plates, using a total volume of 40 μL per well.

Anion screening
Emission spectra of 50 μL of Ln(III) complex (6 -13 μM) with the phosphoanions (1 mM), or none were taken in 10 mM HEPES, pH 7.0. 1 μL of 250 mM MgCl2 was added, and the emission spectrum taken again. Graphs were plotted as the emission spectra themselves and as percentage differences of individual bands, or total emission with the anion compared to without the anion.

Anion binding titrations
Increasing  The observed chemical shifts depend both on the internuclear distance (1/r 3 ) and on geometry factors.
Until recently, understanding of the effects of magnetic anisotropy on the observed NMR shifts in a lanthanide complex rested on the theory developed by Bleaney over 40 years ago. 8 According to Bleaney's theory, the pseudocontact shifts are described in equations (1) and (2) below, where θ, , and r define the polar coordinates and internuclear distance to the lanthanide(III) ion, CJ is the Bleaney constant, µB is the Bohr magneton, C 2 and 2 2 are second order crystal field splitting parameters, 〈 ‖ ‖ 〉 is a numerical coefficient, J is the total spin orbit coupling and g the electron gfactor. The Bleaney constant varies with the electronic configuration of the lanthanide(III) ions, but is considered to be a property of the lanthanide only (independent of the ligand). Values of Bleaney constants for free lanthanide ions are tabulated in the literature, 8 and are often used when dealing with complexes too, under the assumption that the crystal field has a negligible effect on their value.
While Bleaney's theory has been found to conform to experimental observations in a number of systems, it contains assumptions and approximations, which may limit its applicability.
For axially symmetric complexes with a symmetry axis Cn (n>= 3), the geometric part of the right-hand term in equation (1), ( 2 2 ), is zero, so that part of the equation is often omitted when dealing with such complexes.

Array protocols
To  Table   S1. The appropriate loading values for the first two principle components were multiplied by the appropriate lanthanide complex band % emission change and summed over the anion replicate in order to calculate the first two principle components. These principle components were plotted using OriginLab 2016.               http://app.supramolecular.org/bindfit/view/19a32a56-3ce5-4fd6-9a31-d0d9f8331a1b

X-ray crystallography
Crystal data, atomic coordinates, geometry, etc, are given in the tables at the end of this document.
The structure was solved 10 and refined 11 routinely except as detailed in the following text.
C37H61EuN8O15·7(H2O). Half of the molecule comprises the asymmetric unit because the molecule is located on a two-fold axis running through Eu(1) and O(4). One unique water molecule of crystallisation was well-behaved and located in a general position with H atoms located. Another water molecule was located on a special position, but H atoms could not be located. The remaining three unique water molecules were modelled as having the oxygen atom disordered over two positions with H atoms not located. The formate ion is disordered with C (19) and O(5) modelled as disordered over two, equally-occupied, sets of positions determined by the symmetry.

Computational Methods
Density functional theory (DFT) calculations were performed with the hybrid meta-generalized gradient approximations (meta-GGA) TPSSh functional, 12 and the dispersion corrected B3LYP functional within the Gaussian 16 package (Revision E.01). 13 Both functionals have been previously used to model Ln(III) complexes, with the former suggested to provided more accurate geometries. [14][15][16] Considering that f orbitals do not play a major role in Eu−ligand bonds, 17 the large-core quasirelativistic ECP (LCRECP), having 46 + 4f n electrons, was used to describe the metal centre. 18 Calculations were therefore conducted on a pseudo singlet state configuration. LCRECP calculations have been shown to provide good results in DFT studies that focus on the structure, dynamics, and estimates of relative energies of Ln(III) complexes. [19][20][21] The 6-31G(d) basis set was used for all other atoms. Free energies were evaluated at 25 °C and corrected to a standard liquid state of 1 mol/L. In all cases, vibrational entropies were obtained using a quasi-harmonic approximation, treating vibrational modes below 100 cm -1 as free rotors and as rigid rotors above this cut-off, as first proposed by Grimme 22 and implemented in Python. 23 Solvent effects (water) were evaluated by using the polarizable continuum model (PCM) as implemented in Gaussian 16.