Advances in anion binding and sensing using luminescent lanthanide complexes

Luminescent lanthanide complexes have been actively studied as selective anion receptors for the past two decades. Ln(iii) complexes, particularly of europium(iii) and terbium(iii), offer unique photophysical properties that are very valuable for anion sensing in biological media, including long luminescence lifetimes (milliseconds) that enable time-gating methods to eliminate background autofluorescence from biomolecules, and line-like emission spectra that allow ratiometric measurements. By careful design of the organic ligand, stable Ln(iii) complexes can be devised for rapid and reversible anion binding, providing a luminescence response that is fast and sensitive, offering the high spatial resolution required for biological imaging applications. This review focuses on recent progress in the development of Ln(iii) receptors that exhibit sufficiently high anion selectivity to be utilised in biological or environmental sensing applications. We evaluate the mechanisms of anion binding and sensing, and the strategies employed to tune anion affinity and selectivity, through variations in the structure and geometry of the ligand. We highlight examples of luminescent Ln(iii) receptors that have been utilised to detect and quantify specific anions in biological media (e.g. human serum), monitor enzyme reactions in real-time, and visualise target anions with high sensitivity in living cells.


Introduction
The development of luminescent lanthanide(III) complexes for the purpose of binding and sensing anions in water has advanced considerably in recent years. The creation of Ln(III)based anion receptors is driven by the need for new sensing and imaging tools for biological, clinical and drug discovery research. [1][2][3] For example, an emissive Ln(III) receptor capable of binding bicarbonate (HCO 3 À ) selectively could be used to image spatial-temporal HCO 3 À dynamics in living cells using uorescence microscopy, 4 potentially aiding the diagnosis and treatment of diseases such as renal disease and glaucoma. 5 A Ln(III) receptor that binds adenosine diphosphate (ADP) selectively could allow real-time analysis of kinase enzyme activity by monitoring the production of ADP, 6 thereby providing a convenient luminescence assay for high throughput screening of potential kinase inhibitors for the treatment of cancer. 7,8 Many other sensing and bioimaging applications can be envisaged for Ln(III) receptors, [9][10][11] which serve to enhance the technologies available to biological and medical scientists. Emissive Ln(III) complexes, particularly of europium(III) and terbium(III) which emit in the red and green spectral region respectively, offer unique photophysical properties that are very valuable for anion sensing in biological media. 12,13 Firstly, they possess long luminescence lifetimes (up to milliseconds) that enable time-gated or time-resolved measurements to distinguish the lanthanide-centred emission from the short-lived autouorescence of biological samples. 1,14,15 Time-gating methods offer enhanced signal-to-noise and very low limits of anion detection (Fig. 1). Secondly, they have sharp line-like emission spectra, which can allow ratiometric analysis by measuring the change in intensity of one emission band relative to the almost stationary intensity of a second band (e.g., comparing the hypersensitive DJ ¼ 2 emission band of Eu(III) around 615 nm with the DJ ¼ 1 emission band around 590 nm). 16 Thirdly, by careful design of the surrounding organic ligand, Ln(III) complexes capable of rapid and reversible anion binding can be created, providing a luminescence response that is fast and sensitive, offering the high spatial resolution required for biological imaging applications. 3,13 Over the last two decades, a variety of Ln(III) complexes have been developed for binding and sensing anions, cations and biomolecules in aqueous media, and these have been summarized effectively in some comprehensive reviews. 2,11,[17][18][19] Recently, a few examples of anion responsive Ln(III) coordination polymers have emerged, which may be explored as potential sensing materials. 20,21 However, examples of Ln(III) complexes that exhibit high selectivity for a target anion are still relatively rare. This review focuses on recent examples of discrete, water-soluble Ln(III) complexes that display sufficiently high anion selectivity to be utilised in biological or environmental sensing applications. We distinguish the mechanisms of anion binding and discuss how selectivity can be tuned through variations in ligand structure and geometry. This review is not comprehensive, and a particular focus is given to Eu(III) and Tb(III) receptors that can detect anions in biological media (e.g. human serum), monitor enzyme reactions involving anions in real-time, or visualise uctuations of target anions in living cells using uorescence microscopy.

Lanthanide receptor design considerations
Anion recognition in water is challenging for several reasons; anions have a range of geometries, they may be pH sensitive (e.g., H 2 PO 4 À /HPO 4 2À , HCO 3 À /CO 3 2À ), and have high hydration energies. 22,23 Molecular receptors that utilise coulombic attraction or strong metal-ligand interactions are required to overcome the strong interactions between an anion and its hydration sphere. Selectivity for a target anion may be achieved by integrating additional binding sites within the receptor, creating a structured binding pocket with high geometric complementarity. 24,25 The nature and abundance of potentially interfering species in the aqueous media under investigation must be considered, such as competing anions and cations, or complex ionic biomolecules (e.g. proteins, nucleic acids), as this denes the requirements for selectivity of the receptor. 2,26 Lanthanide ions adopt coordination numbers between 8 and 10 in aqueous solution and the interaction with the surrounding ligand is predominantly electrostatic in nature. 27 Ln(III) complexes based on polydentate ligands (e.g. the azamacrocycle DO3A) offer signicant scope for the design of selective anion receptors, in which the affinity and selectivity may be modulated by variations in the ligand structure, geometry and conformational exibility, steric hindrance around the Ln(III) ion, and the overall charge of the complex. When designing such receptors, the ligand should incorporate a sufficient number (6-8) of hard donor atoms (e.g. oxygen and nitrogen) to ensure high kinetic and thermodynamic stability of the Ln(III) complex, both in the absence and presence of anions, to avoid complications arising from metal ion dissociation. In addition to the structural requirements for anion binding, to overcome the intrinsic low molar absorptivity of the lanthanides, a strongly absorbing chromophore (or antenna) should be integrated into the ligand structure, which can be excited by UV/visible light before transferring its energy to the proximal Ln(III) ion (Fig. 2a). 28,29 A simplied Jablonski diagram depicting the most common mechanism of sensitization of Eu(III) or Tb(III), via energy transfer from the triplet (T 1 ) excited state of an absorbing antenna, is given in Fig. 2b.

Distinct binding and signalling mechanisms
Anion binding and signalling at Ln(III) centres may be achieved through a variety of mechanisms (Fig. 3). Anion binding may occur directly at the metal centre, involving displacement of one or more inner-sphere water molecules and variations in the Ln(III) coordination environment, resulting in changes in the emission intensity, spectral form and lifetime of the complex (Fig. 3a). Notably, the changes in emission spectral form induced by anion binding (typically characterised by a large change in the DJ ¼ 2 emission band relative to the DJ ¼ 1 band for Eu(III) complexes) offers the opportunity for ratiometric analysis. This attractive strategy has been utilised effectively in inuential work by Parker and co-workers, 30 wherein a range of heptadentate ligands based on DO3A (Fig. 4a) were designed to generate emissive Eu(III) and Tb(III) complexes with one or two available anion binding sites, occupied by water molecules in aqueous solution. The coordinated water causes efficient deactivation of the Ln(III) excited state, via non-radiative energy transfer to vibrational modes of the O-H groups. Upon binding to certain anions, such as lactate, HPO 4 2À , HCO 3 À , and citrate, 30-32 the inner sphere water is displaced, causing enhancements in emission intensity and lifetime, as well as changes in spectral shape induced by changes in the Ln(III) coordination environment. Several of these Ln(III) receptors have been developed into practical assays for the selective and ratiometric detection of anions in biological uids, including lactate (Fig. 4b, [Eu-1]), HCO 3 À and citrate. [33][34][35] Alternatively, anion binding may involve a non-coordinative interaction with the antenna or other components of the ligand framework, including p-p stacking between two aromatic components, hydrogen bonding, or electrostatic interactions with an overall cationic complex (Fig. 3b). In most of these cases, anion binding causes quenching of the singlet or triplet excited state of the antenna by an energy or charge transfer mechanism (e.g. from an electron rich anion to an electron poor antenna), causing a decrease in Ln(III) emission intensity. The binding interactions are typically weaker than those involving direct coordination of the anion to the Ln(III) ion; consequently,  (b) interaction of the anion with the antenna causing electron or energy transfer, which modulates the sensitisation process, usually quenching luminescence; (c) binding of an anion that possesses an appropriate sensitiser, which 'switches on' luminescence. this approach may be better suited for sensing anions present at relatively high (millimolar) concentrations. This strategy is nicely exemplied by a Tb(III) complex developed by Pierre and co-workers, which is capable of binding ATP through a combination of electrostatic and p-p stacking interactions between the adenine group and phenanthridine antenna, resulting in quenching of Tb(III) luminescence (Fig. 4c, [Tb-2] 3+ $ATP). 36 Finally, a less common signalling mechanism involves the 'switching on' of Ln(III) emission upon binding an anion that possesses an appropriate sensitising chromophore (Fig. 3c). This requires matching of the anion's triplet excited state energy with the Ln(III) excited state, such that back energy transfer is minimal. Whilst this approach can lead to high levels of selectivity for a target anion, it is also limited in scope due to the requirement of the anion to possess a suitable chromophore. This approach was employed recently for the selective recognition of guanosine monophosphate (GMP) using a Tb(III)-bisZn(II) complex, which becomes emissive upon sensitisation of the Tb(III) centre by the proximal guanine unit (Fig. 4d, [Tb-3$Zn 2 $GMP] 4+ ). 37

Receptors for inorganic phosphate
Inorganic phosphate has critical roles in skeletal mineralisation, energy production/transfer and cellular signalling. Human blood levels of phosphate are maintained within a relatively narrow range, between 0.8-1.45 mM, 38 comparable with lactate (0.5-1.0 mM) and signicantly lower than HCO 3 À (23-29 mM). 39 The intracellular concentration of inorganic phosphate is higher than in serum, ranging between 1-5 mM depending on the tissue, pH and hormone levels. Elevated phosphate levels are associated with diabetes and renal disease and can cause vascular calcication, increasing the risk of stroke. In the environment, inorganic phosphate is essential in phosphorus fertilizers to support food production. However, over-use of such fertilizers has led to an increase in phosphate in surface waters, from its typical range 0.01-0.1 mM, 40  , inducing a 36-fold, 20-fold and 5-fold increase in luminescence, respectively (Fig. 6a) -13] show great potential for the selective detection of phosphate in blood serum and demonstrate how selectivity can be ned tuned with the addition of appropriate pendant groups. The possible interaction with organic phosphate derivatives (e.g. AMP, ADP, ATP) could also be investigated. In subsequent work, the potential relationship between the number of inner sphere water molecules and anion affinity was  . However, the observation of reversible CN À binding to these complexes was investigated further, leading to the rst Ln(III)-based sensor for cyanide in water. 45 Addition of CN À to the lysine derivative [Eu-7] + in water at pH 9.8, caused displacement of all three innersphere water molecules and a 9-fold enhancement in Eu(III) emission intensity.   solution. By increasing the concentration of the Gd(III) complexes at pH 7.2 in the presence of HPO 4 2À , phosphocreatine (PCr) and ATP (8.7, 5.5 and 7.9 mM, respectively), 31

Receptors for nucleoside phosphate anions
Nucleoside phosphate (NP) anions such as ATP, GTP, ADP, UDP, AMP are involved in a wide range of biological processes, including energy transfer, DNA synthesis, intracellular signalling and regulation of enzyme activity. 49 ATP is the most abundant NP anion in cells and is the primary chemical energy source in living systems. The majority of ATP is generated in the mitochondria by oxidative phosphorylation, and serves as a substrate for several enzymes, including kinases, ATPases and RNA polymerases. The concentration of ATP varies signicantly from nanomolar extracellular levels to millimolar levels (1-5 mM) in certain cellular organelles. [50][51][52] In comparison, ADP is present in much lower levels inside cells (50-200 mM), with the ATP/ADP ratio ranging between 5-100. 53

ATP and ADP receptors
The development of receptors that can signal ATP levels in cellulo could provide a better understanding of the way in which energy is produced, transported and consumed within the cell. 54,55 Moreover, a receptor that can discriminate between ATP and ADP could be used for monitoring of kinase enzyme activity in real-time. 56 Kinases catalyse the phosphorylation of proteins (converting ATP to ADP in the process), and constitute one of the most promising drug targets for the treatment of cancer. 7 An ATP or ADP-selective receptor could facilitate high throughput screening campaigns for the identication of potent kinase inhibitors, 7,8 surmounting the limitations of current bioassay technologies that rely upon unstable antibodies, or radioactive substrates. To be effective in such applications, the receptor should exhibit rapid and reversible anion binding to prevent perturbation of enzyme reaction rates, and operate in the presence of Mg(II) and Ca(II) ions, which compete for ATP (and ADP) binding. 57 Butler and co-workers synthesised a series of cationic C 2symmetric Eu(III) complexes [Eu-17] + -[Eu-21] + (Fig. 8a), 58 and evaluated their binding towards NP anions at physiological pH. Complexes [Eu-18] + and [Eu-21] + showed excellent discrimination between ATP, ADP and AMP in 10 mM HEPES buffer in the presence of 5 mM Mg(II) ions. Anion binding occurs directly at the metal centre with displacement of the inner-sphere water molecule, increasing the Eu(III) emission intensity and lifetime. Complex [Eu-21] + binds most strongly to ATP (log K a ¼ 5.8 in 10 mM HEPES, pH 7.0), resulting in a 24-fold increase in intensity of the DJ ¼ 2 emission band, whereas ADP causes a smaller increase in emission, despite showing similar affinity. Binding of ATP was approximately 10 times stronger than AMP or pyrophosphate (P 2 O 7 4À ), and 100 times stronger than HPO 4 2À and HCO 3 À (log K a ¼ 2.7 and 3.0, respectively). The high affinity of [Eu-21] + for ATP is attributed primarily to strong metal-ligand interactions, strengthened by hydrogen bonding to the quinoline amide arms projecting from the same face of the receptor. Solution NMR analysis, supported by DFT calculations and X-ray crystallography, indicate a bidentate binding mode of ATP to the Eu(III) complexes, via the aand g-phosphate groups, forming a 1 : 1 host-guest complex in aqueous solution with fast binding kinetics on the NMR timescale. 59 The distinctive luminescence response of [Eu-21] + for ATP enabled its detection in a simulated biological medium containing NaCl, Mg 2+ , ADP, GTP, UTP and human serum albumin. 58 The increase in intensity of the 6J ¼ 2 band was approximately linear over the biologically relevant ATP range of 0.3-8.0 mM. [Eu-21] + was found to permeate mammalian (NIH-3T3) cells and localise to the mitochondria selectively, permitting real-time visualization of elevated mitochondrial ATP levels, following treatment with a broad spectrum kinase inhibitor, staurosporine (Fig. 8b). Additionally, it was possible to image depleted ATP levels upon treatment with potassium cyanide (an inhibitor of oxidative phosphorylation), under glucose starvation conditions.
The structurally related complex [Eu-18] + showed lower affinity for both ATP and ADP (log K a ¼ 4.4 and log K a ¼ 4.6, respectively) compared with [Eu-21] + , but gave a much larger increase in emission for ADP in the presence of Mg(II) ions. 60 The weaker binding of [Eu-18] + to ATP (and ADP) is ascribed to the decrease in the electropositive nature of the metal ion due to the presence of two pendant carboxylate donors, compared with the neutral carbonyl amide donors in [Eu-21] + . In the presence of Mg(II) ions, the interaction of [Eu-18] + with ATP is signicantly reduced, due to competitive Mg-ATP binding, enabling a selective response for ADP to be attained. The ability of [Eu-18] + to signal ADP in the presence of ATP was utilised to monitor a phosphorylation reaction catalysed by protein kinase A, by following the intensity ratio at 616/600 nm as a function of the increasing ratio of ADP/ATP. 60 Subsequently, [Eu-18] + was developed into a miniaturised assay for real-time monitoring of a variety of pharmaceutically important enzymes that generate NP anions, including kinases (ATP into ADP, Fig. 9), glycosyltransferases (UDP-sugar into UDP), and phosphodiesterases (cAMP into AMP). 6 In each case, changes in the NP product/substrate ratio are monitored by time-resolved luminescence using a standard plate reader, allowing different enzyme classes to monitored by a convenient  increase-in-signal format. Unlike the majority of commercial enzyme assays, [Eu-18] + can operate at physiological concentrations of NP anions (e.g. 1-5 mM ATP for kinase reactions), permitting screening of inhibitors of low activity enzymes. The impact of a range of known inhibitors on the activity of Aurora A kinase was assessed using [Eu-18] + (Fig. 9c), and an inhibitor titration was conducted for the inhibitor staurosporine (Fig. 9d), giving an IC 50 of 9.32 AE 0.46 mM, comparable with previously reported values.
Recently, Butler and co-workers extended the library of cationic Ln(III) receptors and established how their phosphoanion binding properties can be tuned through modications in the ligand structure. 59 The relative positions of the pendant quinoline arms on the macrocyclic ligand were shown to signicantly impact on host-anion affinity and stability, with the trans-related quinoline groups of   ) it was possible to discriminate eight NP anions (ATP, ADP, AMP, GTP, GD(III)P, GMP, cAMP, Pi) in a sensing array, using principle component analysis (Fig. 10). The sensing array takes advantage of the differential emission intensities and lifetimes of the four Ln(III) receptors to generate additional anion selectivity through time-resolved measurements.
Using a different approach, Pierre and co-workers prepared a Tb(III) receptor [Tb-2] 3+ (Fig. 11) capable of binding ATP through p-p stacking between the adenine moiety and the phenanthridine antenna. 36 The positively charged Tb(III) complex engages in electrostatic interactions with ATP, which are stronger than for ADP and AMP. The p-p stacking interaction promotes photoinduced electron transfer (PET), which prevents energy transfer from the antenna to the Tb(III) centre, thus quenching luminescence. Similar quenching was observed in the presence of GTP, whereas pyrimidine nucleosides (UTP, CTP) did not cause PET quenching. Notably, there is no direct coordination of phosphate groups to the Tb(III) ion, hence the overall affinity for ATP and GTP is relatively low. Stern-Volmer plots conducted in 10 mM Tris buffer (pH 7) revealed high selectivity towards ATP over ADP/AMP (and similarly, for GTP over GDP/GMP). Subsequently, [Tb-2] 3+ was combined with a non-responsive neutral Eu(III) complex [Eu-23] to produce a ratiometric cocktail for monitoring ATP/ADP or GTP/GDP   ratios, by following the change in Tb/Eu emission ratio at 545/ 700 nm (Fig. 12). 61 This mixture of Eu(III)/Tb(III) complexes could potentially be used for monitoring GTPase activity in real-time.
Using a very similar design, Tang et al. developed a Eu(III) complex,   3+ , with an appended terpyridine antenna for ATP sensing (Fig. 11). 62 The proposed binding mode involves an initial 2 : 1 host-guest complex when 0.5 equivalents of ATP is added, wherein ATP stacks between the terpyridine units of two Eu(III) complexes, causing an 8.3-fold increase in emission. Further equivalents of ATP cause a loss of stacking of the sandwich-type structure, resulting in a 1 : 1 complex and associated decrease in emission intensity. Anion selectivity studies were conducted in water at pH 6.8 (due to the emission intensity being sensitive to pH changes in the range 6.8-8.0), revealing that other phosphate anions and HCO 3 À have small inuences on the emission, while citrate notably impacts the selectivity of the receptor for ATP. Although this complex can distinguish ATP from ADP/AMP in aqueous solution, its use in biological systems would unfortunately be problematic due to its pH sensitivity and the specic ratio of host : guest required to 'turnon' emission. In the presence of more than 0.5 equivalents of ATP, the emission will be quenched due to a 1 : 1 binding mode being preferential. Parker and co-workers devised a short series of Eu(III) complexes, capable of monitoring the ATP/ADP ratio by induced circularly polarised luminescence (CPL). 63 Complexes [Eu-27]-[Eu-29] (Fig. 13) are based on the strongly emissive Eurotracker® series of probes, 64,65 containing one alkynylpyridine chromophore and a substituted picolylamine moiety, containing either two picolyl groups, two ethyl groups, or one of each. The differences in the ligand structure were found to modulate binding affinities for both Zn(II) and NP anions. An increase in emission intensity of   Most notably, the binding of ATP and ADP to the binuclear complex [Eu-25$Zn] 2+ in water (0.1 M HEPES, pH 7.4) induced strong CPL signals of opposite sign, which enabled monitoring of changes in the emission dissymmetry factor, g em , as a function of the ratio of ATP/ADP. CPL spectral data supported by DFT calculations revealed differences in the chirality of the ATP and ADP adducts of the [Y-25$Zn] 2+ complex, which differ in the arrangement of the exocyclic ring substituents (L/D). The optimised DFT geometries of the adducts indicate that the terminal phosphate of the nucleotide bridges the Zn(II) and Eu(III) ions, and suggest the D diastereomer is preferred for ATP and the L isomer for ADP or AMP (Fig. 13).
Schäferling and co-workers prepared a series of multidentate Eu(III) complexes of varying stability (7-coordinate down to 3coordinate complexes), each containing an alkynyl pyridine group for sensitisation of Eu(III) emission. 66,67 The neutral pentadentate Eu(III) complex [Eu-28] was found to have four coordinated water molecules, and showed a 60-fold increase in Eu(III) emission intensity upon the addition of 1 mM ATP (or 1 mM pyrophosphate), versus a 27-fold increase with ADP. However, in the presence of 5 mM Mg 2+ ions, discrimination between ATP and ADP was lost, due to the competitive interaction of ATP with Mg 2+ . However, at lower concentrations of Mg 2+ (0.05 mM), it was possible to track the ATPase catalysed conversion of ATP to ADP, by measuring a decrease in luminescence as a function of time.
Subsequently, the same authors showed that the related cationic Eu(III) complex [Eu-29] + , bearing two ethyl 2-(methylamino) acetate arms, gave a large increase in luminescence for ATP, compared with other NP anions (ADP, CTP). The emission intensity increased linearly over the ATP concentration range 0-20 mM. A 1 : 1 binding mode was assumed although no binding constants were reported. [Eu-29] + displayed signicant ligand uorescence centred at 490 nm, indicating less efficient energy transfer from the ligand to the Eu(III) centre compared with [Eu-28], potentially arising from weaker bonds between the Eu(III) ion and the ethyl 2-(methylamino) arms. Despite this, the high sensitivity of [Eu-29] + towards ATP was utilised to monitor apyrase activity at pH 6.5, again by following a decrease in Eu(III) luminescence as ATP is converted to ADP/AMP.

Nucleoside monophosphate receptors
The majority of NP receptors reported to date have targeted ATP or ADP. In comparison, receptors capable of selectively binding the monophosphate anion AMP are rare. This requires high geometric complementarity between the receptor and AMP to outweigh the coulombic attraction for the more negatively charged anions, ATP and ADP. Albrecht and co-workers developed a luminescent Eu(III) helicate, [Eu-30] 4+ , that exhibits unique selectivity for AMP (Fig. 14a). 68 The molecule comprises an unsaturated double-stranded dinuclear helicate formed from two bis(tridentate) ligands and two Eu(III) centres. A wide range of anionic co-ligands for the helicate were screened in 10 mM HEPES (pH 7.4), with only AMP binding causing an enhancement in luminescence (F increases from 1.8 to 6.1%). A binding constant for AMP of log K a ¼ 3.83 AE 0.01 was determined. AMP appears to be the perfect size and shape match to bridge the two Eu(III) centres, displacing the quenching solvent and counterions in the process. The proposed binding mode involves phosphate coordination to one the Eu(III) centres, and subsequent binding of an adenine nitrogen atom to the second metal centre, forming a triple-stranded helical structure. The remarkable discrimination of AMP from ADP and ATP, combined with the sensitive detection limit of 2 mM, indicates that [Eu-30] 4+ could be used for detecting AMP in biological media, although the stability of the Eu(III) helicate in the presence of proteins (which could potentially cause metal ion dissociation) would need to be investigated.
Guanosine monophosphate (GMP) is an important cell signalling molecule produced from cyclic guanosine monophosphate (cGMP), a reaction catalysed by phosphodiesterases (PDEs). 69 Receptors capable of detecting GMP could enable processes mediated by PDEs to be monitored in real-time. 6 The selective detection of GMP in aqueous solution was accomplished with a Tb(III)-bisZn(II) complex, [Tb-3$Zn 2 ] 4+ , developed by Tuck and co-workers (Fig. 14b). 37 The Tb(III) centre can be sensitised by proximal guanine nucleotides: the addition of GMP to [Tb-3$Zn 2 ] 4+ in 10 mM HEPES buffer (pH 7.4) induced an 87-fold enhancement in Tb(III) emission intensity, whereas GDP (0.5 equiv.) and GTP (0.25 equiv.) induced a 28-fold and 12fold increase in luminescence respectively, with further equivalents of anions causing a decrease in luminescence. The presence of Zn(II) ions in the receptor is important for GMP recognition, since addition of one equivalent GMP to  alone (in the absence of Zn(II) ions) results in a much smaller 2.5-fold increase in Tb(III) luminescence. Analysis of the titration data indicated that GMP binds to a single Zn(II) ion forming a 1 : 1 complex (Fig. 4d), whereas di-and triphosphates initially appear to bridge two host molecules in a 2 : 1 binding mode. The stronger luminescence of the GMP adduct compared with those of GTP and GDP is attributed to the closer proximity of the guanine group of GMP to the Tb(III) centre, permitting more efficient sensitisation by Förster resonance energy transfer.
Competition studies revealed that GMP binds preferentially to [Tb-3$Zn 2 ] 4+ over AMP and CMP; however, UMP has a stronger binding affinity and is able to displace GMP causing a decrease in luminescence. The Tb(III) complex binds acyclic nucleotides (e.g. GMP) more strongly than the cyclic counter parts (e.g. cGMP), owing to the increased negative charge of the former species. The potential biological utility of this receptor was demonstrated by monitoring the PDE catalysed conversion of cGMP to GMP, by following the increase in Tb(III) emission at 544 nm. The Michaelis Menten constant was estimated to be K M ¼ 3.63 AE 0.33 mM, consistent with the literature value.

Receptors for chiral phosphoanions
Ln(III) complexes have been studied extensively as circularly polarised luminescence (CPL) probes by the Parker group. 70,71 When the environment around a Ln(III) ion is chiral, the metal may display highly polarised emission, measured in terms of the dissymmetry factor, or g em value. A strong CPL signal may be induced upon binding of chiral anions or biomolecules, offering exciting potential for biological imaging of chiral species. 71 Two Eu(III) complexes based on triazacyclononane (TACN) containing two strongly absorbing pyridylalkynylaryl antennae (Fig. 15) were recently synthesised, which show selective CPL signalling of chiral phosphoanions, including phosphoserine (pSer), phosphothreonine (pThr) and lysophosphatidic acid (LPA). 72 The ligands were designed to disfavour chelating anions with small bite angles, such as HCO 3 À , by  . Phosphate showed an interaction with [Eu-31] + (log K a ¼ 4.2 AE 0.1), whereas no change in emission was observed for [Eu-32] + . Addition of chiral phosphorylated amino acids, pSer, pThr and pTyr, resulted in subtle changes in the emission spectra but induced strong CPL signals for pSer and pThr, whereas no induced CPL signal was observed for pTyr. The lack of induced CPL activity with pTyr is attributed to the chiral centre being more remote compared with the pSer or pThr adducts. The emission dissymmetry values, g em , were determined for pThr and pSer to be +0.08 and +0.04 (at 593 nm), respectively. The binding affinity of [Eu-31] 2+ for pSer, pThr and pTyr was found to be the same (log K a ¼ 4.80), revealing no preference for a particular phosphorylated amino acid.   from 7% to 2% and then increased back to 7%. The subsequent change in mitochondrial HCO 3 À concentration was monitored by uorescence microscopy, revealing a corresponding decrease in Eu(III) emission intensity as pCO 2 was reduced, and subsequent increase upon return to 7% external CO 2 (Fig. 17).$The analogous Tb(III) complex [Tb-33] 3+ showed negligible changes in emission, thereby allowing ratiometric analysis of intracellular HCO 3 À levels by monitoring the red/green (600-720/450-570 nm) emission ratio (Fig. 17c) (Fig. 16), based on the smaller TACN macrocycle, bearing two strongly absorbing pyridylalkynylaryl chromophores. 75     In the latter cases, it is likely that the negative charge of the 1 : 1 adducts disfavours binding of a second HCO 3 À ion.

Receptors for bicarbonate
The high affinity of the cationic complexes for HCO 3 À compared with those previously reported 4,75 indicates their potential for sensing lower blood serum HCO 3 À levels (e.g. below 23 mM), 78 potentially aiding the diagnosis of metabolic acidolysis associated with kidney disease. However, while the authors note that [Tb-43] + binds weakly to L-lactate (log K a 1.3-1.45), 79 selectivity over other anions (e.g. HPO 4 2À , citrate) or the potential interference from serum proteins was not investigated.
Recently, Sørensen and co-workers investigated the binding of HCO 3 À to Eu(III) complexes of classical DOTA (1,4,7,10tetraazacyclododecane-1,4,7,10-tetraacetic acid) and DO3A ligands (Fig. 4a). 80 The study highlighted the inuence of buffer type, ionic strength and pH on anion affinity. was also shown to be pH dependent, with the emission signal signicantly increasing at higher pH (from 6.8 to 8.0). Bicarbonate titrations at different pHs conrmed that although the proportion of HCO 3 À relative to other carbonate species does increase as the pH is raised from 6.8 to 8.0, the direct inuence of pH on the binding event is much stronger, with binding constant increasing from K a ¼ 0.004 to 50 M À1 . Importantly, such pH dependence suggests that HCO 3 À sensors of this type can only operate in conjunction with a pH sensor (i.e. the pH of the solution must be known before the HCO 3 À concentration can be accurately determined).

Fluoride receptors
The binding of uoride to Ln(III) centres in water has attracted increasing interest in recent years. Receptors showing high selectivity for F À could enable monitoring F À levels in drinking water, which is important because excessive F À levels can cause dental and skeletal uorosis, acute gastric problems and kidney failure. 81,82 The maximum concentration of F À in drinking water is recommended to be 4 mg L À1 (210 mM) by the World Health Organisation, hence receptors with high affinity and sensitivity for uoride are needed. Besides their potential environmental applications, Ln(III) receptors that bind F À have been very useful for studying the NMR and emission spectroscopic behaviour of paramagnetic Ln(III) complexes and their supramolecular assemblies.
One of the earlier examples of uoride recognition involved a cationic Eu(III) complex [Ln-45] + (Fig. 19) reported by Charbonnière and co-workers, wherein F À is sequestered in a supramolecular lanthanide dimer. 83 [Ln-45] + is based on cyclen with two pendant imidazole arms and two carboxylate arms. Upon the addition of F À , a sandwich-like structure forms involving a linear Eu-F-Eu bonding array, stabilised by p-p stacking of the imidazole heterocycles. Luminescence titrations revealed the strong cooperative binding in aqueous solution (log b ¼ 13.0 AE 0.3). A 22-fold enhancement in emission intensity (F ¼ 4.9%) was observed upon addition of F À , providing a highly sensitive response with a detection limit of 24 nM. The complex showed high selectivity for F À over a range of anions including Cl À , Br À , HPO 4 2À , CH 3 CO 2 À and HCO 3 À . Using the erbium(III) complex of the same ligand, the same authors reported the rst evidence of molecular up-conversion in deuterated aqueous solution at room temperature. 84 Low energy excitation of the Er(III) absorption bands of [Er-45] 3+ at 980 nm in D 2 O resulted in weak emission at 525, 550 and 650 nm, ascribed to Er(III) centred transitions via a two-step excitation. The up-conversion signal was signicantly increased by 7.7-fold upon addition of 0.5 equivalents of F À , owing to the formation of a supramolecular [(Er-45) 2 F] + assembly. The large increase in emission for the dimer highlights the importance of the association of two Er(III) metal centres in the up-conversion process and indicates an excited state energy transfer mechanism, in which a short internuclear distance is an important parameter.
In subsequent work, heterodimer formation was achieved using equimolar mixtures of Eu(III) and Tb(III) complexes of [Ln-45] + . In the presence of uoride, selective excitation of the Tb(III) metal (l ex ¼ 488 nm) of the heterodimer resulted in Tb-to-Eu down-shiing energy transfer with 34% efficiency. 85 Utilising this dual Eu/Tb emission (700/545 nm), ratiometric sensing of uoride in water (pH 7.4) was achieved with a detection limit of 17.7 nM. Spectroscopic studies revealed the supramolecular dimers to be stable over a large pH range of 3-8.
Eu(III) complex [Eu-18] + (Fig. 8), developed by Butler, was found to bind F À in water with high selectivity over other anions including Cl À , Br À , I À , HPO 4 2À , CH 3 Fig. 19) and investigated the displacement of the axially bound water molecule by F À using a combination of NMR and emission spectroscopy. [87][88][89] The study revealed dramatic changes in crystal eld splitting upon F À binding, which plays a signicant role in determining the observed spectral properties of paramagnetic host-guest complexes.
In the presence of uoride, the 19 F NMR spectra of the diamagnetic complexes [Lu-47] 3+ and [Y-47] 3+ in D 2 O each showed the presence of two signals; one at À122 ppm corresponding to unbound F À , and a second signal at À74 and À58 ppm for the Lu and Y complexes respectively, corresponding to a F À bound species. The 19 H NMR spectrum of [Yb-47] 3+ showed one (square antiprismatic) conformation in D 2 O (seven signals) and once F À was added a new set of seven signals were observed. 87 Exchange correlation spectrum (EXSY) experiments conrmed the original peaks and new set were in exchange. Interestingly, analysis of the EXSY cross peaks revealed that upon binding F À , the sign of the chemical shi of each proton changes, and the ordering of the peaks (from low frequency to high) is reversed. This change is due to a decrease in the second order crystal eld coefficient, B 0 2 (to around 28% of its original value) and reversal of its sign upon binding uoride. Further evidence for a decrease in magnitude of B 0 2 is given by changes in the ne structure of the emission spectrum of the [Eu-47] 3+ complex, specically the splitting of the DJ ¼ 1 band dramatically decreases, and two easily distinguishable peaks merge to form one apparent peak. 89 The 19 F NMR spectra of Ln(III) complexes of 47 with added F À revealed peaks for bound F À with large shis (e.g. À479 ppm for [Eu-47] 3+ ), consistent with slow exchange of F À on the NMR timescale. Titration experiments revealed relatively weak binding of F À to the Eu(III), Yb(III) and Lu(III) complexes, with 1 : 1 binding constants in the range log K a ¼ 1.0-1.9. These are smaller binding constants compared with those determined independently by Charbonnière and Butler, where F À binding is stabilised through secondary hydrogen bonding and/or hydrophobic interactions with the heterocyclic arms. 89 In subsequent work, the family of DTMA ligands was extended ([Ln-48] 3+ -[Ln-52] 3+ ) to explore the effect of variations in amide structure and peripheral hydrophobicity on F À binding. 88 Binding constants, determined by 1 H and 19 F NMR and emission titration experiments, were in the range log K a ¼ 1.3-2.0. The incorporation of methyl groups into the ligand amides (from 46 to 48) causes the uoride affinity to decrease. Moreover, as the electron withdrawing nature of the benzyl substituent is increased (from OMe to F to NO 2 ), the binding constant increases, indicating that the residual charge on the Ln(III) centre strongly impacts on the binding constant. Interestingly, the introduction of hydrophobic benzyl substituents does not have a signicant effect on the binding strength. However, the rate of F À exchange (with the hydrated complex) was found to be slower for    3+ are similar, implying electrostatic interaction between Yb 3+ and F À is independent of the ligand framework. However, DS ‡ values were signicantly different with [Yb-52] 3+ having a more negative value, suggesting that the rearrangement of solvent in the vicinity of the metal centre with surrounding hydrophobic groups incurs a larger entropic cost during F À binding, resulting in a slower rate of exchange.
These studies highlight how F À binding to axially symmetric Ln(III) complexes can change the sign and magnitude of the crystal eld coefficient (which determines the magnetic susceptibility anisotropy), resulting in dramatic changes in NMR and luminescence spectra. The ability to predict, manipulate and control local crystal eld will play an important role in the design of new responsive Ln(III) receptors.

Lactate receptors
Monitoring of lactate concentration is important for clinical diagnostics, since this anion is a specic biomarker for prostate and breast cancer, and elevated lactate levels are found in patients with Parkinson's disease. 90,91 Seminal work by Parker demonstrated reversible binding of lactate to Ln(III) centres using a combination of solution NMR studies and X-ray crystallography. 30,31 The study revealed that lactate binding may occur via coordination of the CO 2 À group (either in a monodentate manner or via a 4-membered chelate), or through a 5membered chelate involving the OH and CO 2 À oxygens (Fig. 4b).
More recent work has focussed on the development of Ln(III) probes capable of sensing lactate by CPL spectroscopy (Fig. 20). The high sensitivity of CPL to subtle changes in Ln(III) coordination environment is very useful for chiral sensing applications. A short series of Eu(III) complexes (q ¼ 1), prepared by Parker and co-workers from TACN ligands with two pyridylalkynylaryl antenna (bearing phosphinate, amide or carboxylate donors) were shown to bind chiral carboxylates including the a-hydroxy acids lactate and mandelate, forming 1 : 1 adducts. 92 Anion binding was evaluated in a 1 : 1 water/ methanol mixture due to limited water solubility of the Ln(III) complexes. The strongest binding to lactate was observed for the tricationic bis-amide complex   lactate is proposed to bind via the CO 2 À group only, due to the increased steric demand imposed by the bulkier phosphinate groups, supported by DFT calculations. 92 The ability of the Eu(III) complexes to signal lactate binding by the switching on of CPL was demonstrated. Addition of Rand S-lactate to each complex in methanol gave rise to mirror image induced CPL spectra (Fig. 20b). The N-benzyl complex [Ln-56] + produced the strongest CPL signal, ascribed to helical alignment of the benzyl and two pyridyl groups, creating a more rigidied chiral structure. The R enantiomer gave rise to a common induced CPL signature across the series of Eu(III) complexes, identied by the DJ ¼ 4 emission band around 700 nm. Notably, variations in the sign and magnitude of the CPL allowed the enantiomeric purity and absolute conguration of a-hydroxy acids to be estimated.
A chiral Tb(III) complex [Tb-40] + (Fig. 18) developed by Piccinelli and co-workers was found to bind lactate weakly in water. 79 The ligand is based on a trans-1,2-diaminocyclohexane scaffold, which upon binding a Ln(III) ion creates a dissymmetric environment, with q ¼ 2. Lactate displaces the two inner sphere water molecules, signalled by an increase in Tb(III) emission intensity, particularly the band at 546 nm, which plateaued aer 100 equivalents of lactate (600 mM  -58]. 99 Initially, anion titrations were conducted in methanol, in the presence of 1 mM LiOH, to ensure full deprotonation of the dicarboxylates and a constant ionic strength. The nitro derivative [Eu 2 -58] showed selectivity for isophthalate over dinicotinate, nicotinate and benzoate, with log K a (298 K) ¼ 4.3, 2.8, 3.4, and 2.7, respectively. Association constants for isophthalate and dinicotinate did not vary with temperature (293-313 K), indicating that in methanolic LiOH solution binding is disfavoured on enthalpic grounds but favoured entropically, consistent with the displacement of hydroxide increasing the overall disorder in the system.
The binding of nicotinate was examined further in different aqueous buffers (HEPES, BBS, PBS). Notably, the affinity of [Eu 2 -57] for dinicotinate in HEPES and BBS buffers was larger than that observed in methanolic LiOH solution, suggestive of competitive binding of hydroxide in the latter medium. Notably, no binding between [Eu 2 -58] and dinicotinate was observed in PBS buffer, due to the competitive binding of phosphate, whereas [Eu 2 -57] showed no effect of phosphate on the overall affinity. 100 Other biological anions, including lactate (2 mM

Conclusions and outlook
This review highlights the powerful potential of luminescent lanthanide(III) complexes for reversible anion binding and signalling in aqueous and biological media. Stable Ln(III) complexes offer great scope for the design of selective anion receptors, in which the affinity and selectivity can be modulated by variations in the ligand structure and its conformational exibility, steric hindrance at the metal centre, and the overall charge of the complex.
The emission spectra of Eu(III) complexes are particularly sensitive to perturbations in the ligand eld arising from anion binding to the metal centre. Modulation of the ligand eld has been exploited to develop optical sensors for the selective detection of certain oxyanions (e.g. HPO 4 2À , HCO 3 À ), the discrimination of nucleoside polyphosphate anions (e.g. ATP and ADP), and signalling of chiral anions (e.g. L-lactate) in competitive aqueous media. The interactions between host and anion are predominantly electrostatic in nature; however, the incorporation of secondary binding motifs within the ligand structure has proven successful for imparting additional levels of selectivity, through synergistic p-p stacking and/or hydrogen bonding interactions with the coordinated anion.
In certain cases, cell-penetrating Ln(III) complexes have been created which localise to specic organelles of living cells and provide a fast and sensitive signal, which reports on uctuations in anion concentrations (e.g., HCO 3 À , ATP) in response to external perturbations of the environment. The most promising imaging probes provide a ratiometric read-out, either by comparing the intensity of two emission bands within a single Ln(III) receptor or the Eu/Tb emission ratio of a dinuclear receptor. Such ratiometric probes provide an internal reference that facilitates calibration of the observed luminescence signal in living cells. The development of anion-selective Ln(III) receptors for monitoring enzyme activity is an emerging eld, which exploits the ability of the receptor to signal the enzymatic formation or depletion of a target anion (e.g. ADP) accurately and rapidly, without perturbing enzyme reaction rates. The existence of commercial Ln(III) bioassays in the form of labelled antibodies and proteins, 101 means that appropriate instrumentation already exists for the creation of new Ln(III)-based highthroughput screening assays, which can be adapted readily in drug-discovery research.
The key challenge remaining is to create Ln(III) receptors which exhibit higher degrees of anion selectivity. In future receptor designs, it should be possible to integrate multiple recognition motifs within the ligand structure, which engage a target anion to generate a distinct host-guest complex structure. Stable Ln(III) receptors with increased anion selectivity will open the door to new and improved assays and imaging probes required for biomedical and clinical research. The creation of near-infrared emitting complexes of Yb(III) and Nd(III) show promise for in-depth imaging of biological tissues in the NIR range (700-1200 nm), [102][103][104] and their anion responsive properties should be explored further. Advances in receptor development should proceed with innovations in spectroscopy and microscopy instrumentation (e.g. uorescence lifetime, CPL, and two-photon excitation microscopy) 105,106such combined efforts will enable tracking of biological processes in living cells with higher spatial and temporal resolution. Innovations in these areas are eagerly anticipated in the near future.

Conflicts of interest
There are no conicts to declare.