Cationic Europium Complexes for Visualizing Fluctuations in Mitochondrial ATP Levels in Living Cells

Abstract The ability to study cellular metabolism and enzymatic processes involving adenosine triphosphate (ATP) is impeded by the lack of imaging probes capable of signalling the concentration and distribution of intracellular ATP rapidly, with high sensitivity. We report here the first example of a luminescent lanthanide complex capable of visualizing changes in the concentration of ATP in the mitochondria of living cells. Four cationic europium(III) complexes [Eu.1–4]+ have been synthesized and their binding capabilities towards nucleoside polyphosphate anions examined in aqueous solution at physiological pH. Complexes [Eu.1]+ and [Eu.3]+ bearing hydrogen bond donor groups in the pendant arms showed excellent discrimination between ATP, ADP and monophosphate species. Complex [Eu.3]+ showed relatively strong binding to ATP (logK a=5.8), providing a rapid, long‐lived luminescent signal that enabled its detection in a highly competitive aqueous medium containing biologically relevant concentrations of Mg2+, ADP, GTP, UTP and human serum albumin. This EuIII complex responds linearly to ATP within the physiological concentration range (1–5 mm), and was used to continuously monitor the apyrase‐catalyzed hydrolysis of ATP to ADP in vitro. We demonstrate that [Eu.3]+ can permeate mammalian (NIH‐3T3) cells efficiently and localize to the mitochondria selectively, permitting real‐time visualization of elevated mitochondrial ATP levels following treatment with a broad spectrum kinase inhibitor, staurosporine, as well as depleted ATP levels upon treatment with potassium cyanide under glucose starvation conditions.


Introduction
Adenosine triphosphate (ATP) is the most abundant nucleoside polyphosphate (NPP) anion in cells [1,2] and serves as the chemical energy source for most biological processes, including organellet ransport, muscle contraction and maintenance of neuronal membrane potential. [2][3][4] Despite its importance,t here are surprisingly few imaging probesc apable of signallingt he presence of ATPr apidly, reversibly and selectively under physiological conditions. [5,6] We have addressed this challenge by creating al uminescent europium(III) complex capable of imaging dynamic changes in the concentration of ATPi nt he mitochondria of living cells. The majority of ATPi sg enerated by the mi-tochondria by oxidative phosphorylation. Numerous enzymes utilize ATPa sasubstrate, including ATPases,k inases, and RNA polymerases;t hus ATPp lays ak ey role in signalling and the regulation of enzymef unction. [1,2,6] The release of ATPt ot he extracellular space has been identified in both damaged and apoptotic cells. 2 Extracellular ATPa lso serves as as ignalling molecule by binding to purinergic receptors, 3 therebym ediating an immunological or nervous response. The concentration of ATPv aries considerably,f rom nanomolar extracellular concentrationst om illimolar concentrationsi nt he cytosol and certain organelles (e.g. mitochondria). [1,2] To gain ab etter understanding of the wide range of dynamic processes involving ATP, non-invasive imaging probesa re required that can signal changes in ATPl evels by modulation of luminescence, thereby providing spatial and temporal information rapidly,w ith high sensitivity. [5a, 6a] Current methods for monitoring the concentrationo fATPi n living cells have limitations. [2, 5a] Thee nzyme firefly luciferase can be expressed in cells to measureA TP levels indirectly,b y catalyzing ac hemiluminescent reaction between ATPa nd luciferin. [7] However, this method consumes ATP, which may perturb the cellular energy status,p reventing accurate quantification of ATP. [8,9] Genetically encoded fluorescent biosensors (e.g. ATeam, [6a] PercevalHR) [6b] have been developedt om easureA TP successfully within specific cellular compartments. However, biosensors encoded within cells can requiret ime consuming protein expression and maturation procedures and are intrinsically pH sensitive; [6b] small changes in intracellular pH (< 0.1 units) have been shown to generate bias in the emission re-sponse, which can complicate interpretationo ft he observed signal. [9] An attractive alternative strategy involvesthe creation of discrete ATP-responsive synthetic receptors that exhibit intrinsic cellular uptake and localization behaviour. [1, 5a] The majority of synthetic receptors for ATPcomprise afluorescent organic indicator linked to ap ositively charged recognition group, such as imidazolium or guanidinium units, or coordinated zinc(II) ions, which form stronge lectrostatic or metal-ligandi nteractions with the triphosphate fragment of ATP. [10] Only afew ATP-selective probes have been successfully applied to imaging ATPi n living cells. [5a-d] Each of these probese mits as hort-lived fluorescence signal, typically at wavelengths between 375 and 540 nm. Consequently,t heir signal overlaps with background autofluorescence arising from endogenous molecules, which can complicate intensity-based measurements. [11] In two cases, the detection range of the receptor is very low (0.1-10 mm), [5a,b] and the fluorescencer esponse is saturated at ATPc oncentrations below those expected in cells (1-5 mm). [12] Chang and co-workersd evelopedaprobe capable of detecting ATPw ithin the 2-10 mm range, in av iscous mediumo fg lycerol/water (60:40). [5c] However,t his probe is unable to respond to ATPi n 100 %a queous solution at physiological pH, hence its utility in live-cell imaging is limited and dependent on variations in viscosity within the cell.
Importantly,d uring in vitro testing, anion affinity and selectivity is rarely assessedi nacompetitive ionic mediumt hat resembles intracellular fluid. [5c] This is critical, because it defines the target concentrationr ange and the nature anda bundance of potentially interfering species, including other phosphoanions (e.g. ADP,A MP,G TP,H PO 4 2À ), cations that bind strongly to ATP( e.g. Mg 2 + and Ca 2 + ions), and proteins, which can interact with the probe causing either quenching or enhancement of emission. The influence of such a competitive ionic environmento nt he probe's anion affinity and selectivitys hould be studied, to optimize its practical utility in live-cell imaging experiments. [13] In recent years, stable luminescente uropium(III) and terbium(III) complexes have emerged as attractive tools for use in cellular imaging. [14] Notably,a series of the brightest Eu III complexes have been developeda se fficient cellular stains, designed to localize preferentially in specific organelles. [15,16] Stable Eu III and Tb III complexes offer significant advantages over conventional fluorescent organic dyes, including al arge separation between the absorption and emission bands, long emission lifetimes [17] (withint he millisecond range) that enable complete removal of background autofluorescence by using time-resolved imagingt echniques, and well-definede mission bands that permit ratiometric analysis. [18] Certain Eu III and Tb III complexes have been shown to bind reversibly to anions in aqueous media, including bicarbonate, [19] fluoride, [20] lactate , [21] phosphate [22] and phosphorylated peptides. [23] Anion binding can be signalled by changes in the intensity ratio of two Ln III emission bands or by changes in emission lifetimeo f the complex. [13,24] However, examples of Ln III complexes that bind to NPP anions such as ATPa re rare. [25][26][27][28] Bindingt oA TP is usually weaka nd causes quenching of luminescence due to energy transfer to the nucleotide base [25,27] or displacement of acoordinated sensitizing ligand from the Ln III ion (decomplexation). [28,29] Probes that utilise quenching of luminescencea re less desirable for use in cellular imaging as other competitive quenching processes may generate the observed decrease in emission intensity. [14b] In this work, we report as eries of luminescent cationic Eu III complexes [Eu.1-4] + that bind reversibly and with differential affinities to NPP anionsi na queouss olutiona tp hysiological pH ( Figure 1)  This new class of Eu III complexes offers an umber of improvements in performance comparedt oe xisting probes for ATP, including al ong-livedl uminescence signal that is sensitivet o ATPl evelsw ithin the physiological range (1-5 mm), with minimal interference from biomoleculea utofluorescence or changes in cellular pH. In addition, [Eu.3] + possessesadistinctive subcellular localization profile that allows ATPl evels to be monitored within as pecific region of the cell.

Results and Discussion
Complex designand synthesis Each Eu III complex contains two coordinating quinoline groups that act as efficient sensitizers of Eu III emission. [ 16, 5a Complexes [Eu.1] + and [Eu.3] + possessh ydrogen-bond donor (amide) groupsa tt he 7-positiono ft he coordinated quinoline chromophores. We envisaged that NPP anions such as ATPw ould coordinate to the Eu III ion via the terminal phosphate group, enabling the adjacent phosphate group to engage in hydrogen bonding interactions with quinoline amide N-H groups ( Figure 1B), thereby enhancing selectivity over monophosphate species. Indeed, we have shown recently that the cooperative use of metal-ligandi nteractions in combination with hydrogen bondingcan provide excellent selectivity for nucleoside polyphosphate anionso ver monophosphate anions (e.g. AMP,c AMP, HPO 4 2À ,p hosphorylated amino acids), where no such hydrogen-bonding interactions can occur. [31,32] To investigate the function of the hydrogen bond-donor groups in the ATPr ecognition process, we prepared two control complexes, [Eu.2] + and[ Eu.4] + ,w hich lack quinoline amide groups.
Details of the synthesis and characterization of complexes [Eu.3] + and [Eu.4] + + are provided in the supporting information ( Figure S1-S3 for synthetic schemes). Briefly,acyclen derivativeb earing two trans-related secondary amines was reacted with an appropriately functionalized 2-methylquinoline mesylate ester or 2-(chloromethyl)quinoline in the presence of K 2 CO 3 ,t og ive the protectedm acrocyclic ligand.T he terminal ethyl ester protecting groups of the two carbonyl amide arms were hydrolyzed using 0.5 m NaOH solution. Subsequent addi- showeds imilarities, displayinga tl east three components in the DJ = 1(585-605 nm) band, and two distinguishable components within the DJ = 2( 605-630 nm) band, consistent with each complex adopting as tructure of low symmetry in water ( Figure S8). For [Eu.3] + ,arather different emission spectrum was obtained,c haracterised by ap ronounced DJ = 2 band around 605-630 nm andt wo discernable components within the DJ = 4b and around 675-705nmi nt he red region of the visible spectrum( Figure 2). Emissionq uantum yields were in the range 7-23 %a nd emissionl ifetimesi nH 2 Ow ere measured to be approximately 0.5 ms, and at least 50 %l arger in D 2 O. The number of coordinated water molecules, q,w as determined to be one for each Eu III complex. [33] These data suggested that complexes [Eu.1] + and [Eu.3] + were most suited for use in live-cell imaging experiments, due to their water solubility,l ong-lived luminescence and sufficientlyl ong excitation wavelengths of over 350 nm, thereby matching the optics of standard fluorescence microscopes.   Figure S9 and S10). AMP and HPO 4 2À induced am uch smaller (approximately 2-fold) increasei ne mission intensity while other monophosphate anions including cyclic AMP and phosphorylated amino acids (pSer,p Thr,p Tyr) induced negligible spectralr esponses, as did chloride, sulphate, lactate, acetate, glutathione,Na + ,K + ,Z n 2 + ,M g 2 + and Ca 2 + ions. The guanosine phosphate anions, GTP,G DP and GMP induced similar spectralresponses to those observed for ATP, ADP and AMP respectively.T he only othera nionst ested that induced as ignificant spectral response were citrate and bicarbonate,w hich caused am aximum 4-fold and 5-fold increasei nt he DJ = 2 emission band, respectively.C ontrolc omplexes [Eu.2] + and [Eu.4] + ,w hich lacked quinoline amide groups, showedamuch lower level of discrimination between NPP anions: am aximum 3-fold enhancement in intensity of the DJ = 2e mission band was observed in the presence of 1mm ATP, ADP and AMP (Figure S11a nd S12). Analysis of the relative change in total emission intensity (570-720nm) resulted in av ery similara nion selectivity profile for each Eu III complex ( Figure S13-S16).
The emission intensity of [Eu.3] + was found to be particularly sensitive to ATPc oncentration. Figure 3s hows the increase in emission spectra of [Eu.3] + in the presence of increasing ATPl evels (0-180 mm), revealing ad ramatic 17-fold enhancement in intensity at 614 nm within the DJ = 2b and. Notably, the emission spectral shape of [Eu.3] + did not change significantly,s uggestingt hat only minor changes in conformationo f [Eu.3] + occuru pon binding to ATP. Emission lifetimes of [Eu.3] + measured in H 2 Oa nd D 2 Oint he absence and presence of ATPw ere consistent with ah ydration state, q,o f0 .8(AE 20 %) and zero, respectively (Table S1), establishing that ATPd isplaces the coordinated water molecule from the Eu III metal, accompanied by apronounced increaseinl uminescence.
Addition of ATP( and ADP) to complex [Eu.3] + also causeda considerable increase in intensity of the hypersensitive DJ = 4 band (675-705nm), whereas AMP and HPO 4 2 caused much smaller changes ( Figure S17 and S18). Moreover,t he ATPa nd ADP adducts of [Eu.3] + could be discriminated by the variation in spectrals hapeo ft he DJ = 4b and;t he ATPb ound species is characterized by two discernable lines centred at 688 and 697 nm, respectively,w hereast he ADP adduct showeda tl east 4components within the DJ = 4region.The structurallyr elated complex [Eu.4] + showed similard istinctive changes in the shape of the DJ = 4b and in the presence of ATPa nd ADP; however,o nly minor increases in emission intensity took place (Figures S17 and S18).
Apparent binding constants were determined for complexes [Eu.1-4] + and ar ange of NPP anions by following the change in the intensity ratio of the DJ = 2/DJ = 1( 605-630/580-600 nm) emission bands as af unctiono fa nion concentration, followed by an on-linear,l east squares curve-fitting procedure based on a1 :1 binding model (    Figure S50 and S51). 2 Mass spectral data showingt he formationo fs table ternary adducts of [Eu.1] + and ATPo rA DP was reportedp reviously. [31] Effect of Mg 2 + + ions in solution The majority of intracellular ATPe xists in the form Mg-ATP 2À . [34] Therefore, ac ellular imaging probe for ATPs hould ideally be able to bind to the Mg-ATP 2À complex.I ndeed, enzymes that require ATPa sasubstrate, such as ATPases and kinases, utilize Mg 2 + ions to provide additional stabilization of the ATPenzymec omplex. [35] The binding interaction between Mg 2 + ions and ATPi nw ater at physiological pH is approximately 50 times stronger than the interaction with ADP,w ith logK a (Mg-ATP) = 4.2 and logK a (Mg-ADP) = 3.6. [36] The presenceo fM g 2 + ions in aqueous solution was expected to influence the ability of complexes [Eu.1] + + and [Eu.3] + + to discriminate betweenA TP and ADP. Indeed, previously reporteds ynthetic receptors have shownl imited discrimination betweenA TP and ADP at enzyme relevant concentrations of Mg 2 + ions. [26,37] We showed recently that [Eu.1] + + can discriminate effectively betweenA TP and ADP in ab uffered aqueous solution containing 3mm Mg 2 + ions: adding 1mm ADP caused an 8-fold increasei no verall Eu III emissioni ntensity compared to am uch smaller2 .5-fold intensity increase in the presenceo f1m m ATP. [24] Excellent discrimination betweenA TP andA DP was attributed to the slightly highera ffinity of [Eu.1] + for ADP (Table 1) as wella st he higher competition between Mg 2 + and ATPc ompared to ADP.T his allowed the change in the ratio of ATP/ADP to be dynamically followed during the course of ak inase-catalyzed phosphorylation reaction.
Complex [Eu.3] + + showed ad ifferent discriminatory behaviour between ATPa nd ADP in the presence of Mg 2 + ions. Figure 4A shows the emission spectral response of [Eu.3] + + in the presence of 2mm ATP, ADP,A MP andH PO 4 2À in af ixed background of 5mm Mg 2 + ions. Adding2m m ATPr esulted in a substantial 24-fold enhancement in intensity of the DJ = 2 emission band, whereas ADPc aused as maller 13-fold increase in luminescence. Only minor changes in emission spectra took place in the presence of AMP and HPO 4 2À .T itrationso fA DP and AMP into an aqueous solution containing [Eu.3] + and 5mm Mg 2 + ions resulted in standard hyperbolic curves that were fitted to a1 :1 binding model ( Figure S56 and S57). Apparentassociation constants weredetermined to be logK a = 4.6 and 3.8 for ADP and AMP respectively,a pproximately 1o rder of magnitude lower than those determinedi nt he absence of MgCl 2 (Table 2).
At itration of ATPi nafixed background of 5mm Mg 2 + ions generated an isotherm consistentw ith the occurrence of two distinct emissive species ( Figure 4C). Incrementala ddition of 0-1.5 mm ATPgave rise to a4-fold increaseinemission intensity of the DJ = 2b and, and subsequenta ddition of 1.5-3.0 mm ATPi nduced af urther3 -fold enhancement in Eu III emission, with no discernable change in spectralf orm. This can be ascribed to the formation of two discrete ternary adducts,e ach exhibiting as imilar coordination environmenta tt he Eu III ion. It is hypothesized that [Eu.3] + binds to either ATPo rt he Mg-ATP 2À complex;w ith the binding geometry between [Eu.3] + and Mg-ATP 2À being similart ot hat of ATP, and stabilization of the negativelyc harged triphosphate fragment provided by the Mg 2 + ion ( Figure 1B). During the ATPt itration, the concentration of Mg-ATP 2À increases, leading to the formationo fa highly emissive ternary complex. High resolution mass spectrometric data supported binding of [Eu.3] + to Mg-ATP 2À ,g iving an intense signal at 752.6439 corresponding to the doubly chargeds pecies [Eu.3 + ATP + Mg + 3H] 2 + ,w hich was in excellent agreement with the calculated isotopic distribution (Figure S52). At itration of ATPi nafixed background of NaCl (100 mm)r evealed ac lassic isotherm for as imple 1:1b inding system,w ith an apparent bindingc onstant of logK a = 5.0 (Figure S58). This confirms that the changes in emission spectra of [Eu.3] + in the presence of Mg 2 + ions cannot be simply ar esult of the additional ionic strength of the medium, rather it is the specific interactions between Mg 2 + ,A TP and the Eu III complex that modulates the equilibrium speciationa nd resulting emission response.

Monitoring ATPase activityi nreal-time
Havinge stablished that [Eu.3] + exhibits distinctive spectralr esponses towards ATPa nd ADP in ab ackground of 5mm Mg 2 + ions ( Figure 4A), we demonstrated the ability of the Eu III complex to monitort he apyrase-catalyzed hydrolysis of ATPt oA DP in real-time. Apyrase is an enzyme that catalyzest he conversion of ATPi nto ADP, releasing energy in the process.T oabuffered aqueous solution (10 mm HEPES, pH 7.0, 5mm MgCl 2 ) containing [Eu.3] + (5 mm)a nd ATP( 2mm)w as added different amountso fa pyrase (80-160 mU, ATP/ADPs electivity ratio, 10:1). The conversion of ATPt oA DP and HPO 4 2À caused a time-dependent decrease in emissioni ntensity of the DJ = 2 band centred at 614 nm and as maller increase in intensity at 594 nm within the DJ = 1b and ( Figure 5A). These changes in spectralf orm are consistentw ith competitive displacement of the bound ATPb yA DP,a st he concentration of ADP increases. Plots of the emission intensity ratio at 614/594 nm as af unction of time revealed thatthe rate of ATPh ydrolysis was directly proportional to the amount of enzyme added ( Figure 5B). The background reaction in the absence of enzyme showed essentially no change in emission intensity duringt he time frame of the experiment. The ratiometric changes in emission intensity observed during the enzymer eactionw ere in good agreement with those obtained in asimulated ATPase reaction, in which the ratio of ATP/ADP was varied systematically ( Figure 5D).
Given that [Eu.3] + binds strongly to ATP, the possibility that the probe lowers the effective concentration of ATPm ust be considered, as this would influence the rate of reaction. However,a nion binding to [Eu.3] + is fast and reversible, and due to the highly sensitiven ature of the luminescence response, the amount of [Eu.3] + required to monitor the ATPase assay is much lower (5 mm)c ompared to the concentration of ATP( 2mm). This ensures that the rate of ATPh ydrolysis is not perturbed by the presence of [Eu.3] + . Figure 5C shows the linear decrease in emission intensity ratio at 614/594nmd uring the initial stages (0-9 minutes)o f each enzymer eaction. Reducing the amount of enzyme from 160 mU to 80 mU resulted in ad ecrease in the initial rate of ATPh ydrolysis by af actor of two. Thus, [Eu.3] + is able to directly and continuously monitor ATPase activity by providing an instantaneous and ratiometric luminescents ignal, without the need to chemically modify the enzyme or its substrate. [38] Evaluating affinity to proteins Encouraged by the ability [Eu.3] + to monitor the enzymatic conversion of ATPt oA DP in real-time, we wished to evaluate the ability of [Eu.3] + to monitorf luctuationsi nA TP levels in living mammalian cells. Ac ellular imaging probe for ATPm ust be able to operate in the presence of proteins.E missive metal coordination complexes are known to interact non-covalently with proteins,o ften causing quenching or enhancemento f emission. [39] We measured the affinity of complexes [Eu.1] + and [Eu.3] + for human serum albumin (HSA), the most abundant protein in human bloodp lasma. Addition of HSA( 0-0.4 mm)t o[ Eu.1] + at pH 7.0 resulted in a3 .5 fold increase in overall Eu III emission intensity and distinctive changes in spectral form ( Figure 6A). Ap lot of the intensity ratio of the DJ = 2/ DJ = 1e mission bands versus HSA concentration allowed an estimation of the associationc onstant, logK a = 4.8 (K d = 16 mm) ( Figure S59). Luminescence lifetimes for the protein bound speciesi nH 2 Oa nd D 2 Ow eref ound to be 0.92 and 1.73 ms, respectively,c orresponding to ah ydration state, q = 0. Thus, binding of [Eu.1] + to HSA involves displacemento ft he bound water molecule, possibly by coordination of an aspartate or glutamate residue of HSA.
In sharp contrast, incremental addition of HSA to [Eu.3] + caused am inor (10 %) increasei nE u III emission intensity and no change in spectralf orm ( Figure S60). Analysis of the luminescencel ifetimes of [Eu.3] + in the presence of HSA in H 2 O and D 2 Os uggested displacement of the bound water molecule. Al ower apparent binding affinity wase stimated between [Eu.3] + and HSA( logK a = 3.2), which is approximately 2.5 orderso fm agnitude lower than the binding constant determined for this complexa nd ATP, under the same conditions. The minimal interaction observed between [Eu.3] + and HSA could be tentatively attributed to the presence of two ancillary carboxylate groups in the macrocyclic ligand, which could minimize the occurrence of hydrophobic interactions between the Eu III complex and the protein. Gratifyingly,a ddition of increasing amountso fA TP (0-0.5 mm)t o[ Eu.3] + in af ixed background of 0.4 mm HSA gave rise to am easurable 3-fold increase in emission intensity of the DJ = 2b and ( Figure 6B), consistentw ith competitive displacemento ft he bound protein from [Eu.3] + ,u pon binding to ATP. The apparent binding constant was estimated to be logK a = 4.0 ( Figure S61), around two orderso fm agnitude lower than that determined between [Eu.3] + and ATPi nt he absence of protein. Conversely,[ Eu.1] + was unable to signal the presence of ATPu nder the same conditions.

ATPr ecognition in simulated intracellular fluid
The above competition experiments, involving biologically relevant amounts of protein and Mg 2 + ions, encouraged us to examine the ability of [Eu.3] + to detect ATPi na na queous mediumt hat mimics the complex ionic environment within cells. In ah ealthyc ell, the most abundant NPP anion is ATP, which is estimated to be present in concentrationsb etween 1-5 mm (average 2.5 mm). [1,12] The majority of ATPi sg enerated by the mitochondria by oxidative phosphorylation. Importantly, ADP is maintained at as ignificantly lower concentration( 50-200 mm), the majority of which is bound strongly to protein, such that the ATP/ADP ratio rangesb etween 5a nd 100. [40] This ATP/ADP ratio acts as ac riticalm odulator of av ariety of cellular events. Other NPP anions including GTP,U TP and CTP are estimated to be present in concentrations5 -fold lower than that of ATP. An intracellularp robe mustb ea ble to respond selectively to ATPu nder these conditions.
Before undertaking cellular imagings tudies, the pH dependence of the emission response of [Eu.3] + in the presence of 2mm ATPw as assessed. The emission intensity and spectral form was essentially unchanged between pH 6.4 and 8.6 (Figure S62), suggesting that the luminescences ignal should not be affected by pH fluctuations around normal mitochondrial or cytoplasmic pH, estimated to be near7 .3 and8 .0, respectively. [34] This is significant, as previously reported ATP-selective imaging probesh ave been shown to be sensitivet oc hanges in intracellular pH. [6, 7c] Additionally,c ertain existingm ethods for measuring intracellular ATPa re dependento nd issolved oxygen concentration (e.g. the bioluminescent luciferase reaction). [9] To confirmt hat [Eu.3] + + is not sensitivet oc hanges in dissolvedo xygen, emission spectra were recorded for [Eu.3] + + alone, and in the presence of 2mm ATP, in air-equilibrated and degassed aqueous solution. The Eu III complex showed less than 5% variation in emission intensity under these conditions ( Figure S63), with no significant change in emission intensity being observed after bubbling of the air-equilibrated sample with oxygen gas for 10 minutes.

Cellular uptake and localization studies
The cellular uptake behaviour of [Eu.3] + was examined in NIH-3T3 cellsu sing fluorescencea nd laser scanning confocal microscopy (LSCM). [42] Incubationo f[ Eu.3] + (50 mm)i nN IH-3T3 cells for 2hours resulted in uptake of the Eu III complex and predominant localization to the mitochondria (l exc 355 nm, l em 605-720 nm), verified by co-localization studies using Mito-Tracker Green (l exc 488 nm, l em 500-530nm, Pearson's correlation coefficient, P = 0.91) (Figure 8). The preferentiald istribution of [Eu.3] + in the mitochondria could be tentatively attributed to the overall positive chargeo ft he Eu III complex and the amphipathic nature of the macrocyclic ligand. [16] Imaging was possible over extended time periods( up to 8h ours), during which time the brightness of the observed images did not vary significantly (AE 10 %) and the cells appeared to be healthy and proliferating (see supportingi nformation for detailed analysis of probe brightness within cells).
Cytotoxicity andv itality studies were undertaken for [Eu.3] + at 24 hours using image cytometry assays, involving DAPI and AcridineO range stains, which revealed an IC 50 value of greater than 200 mm.C onsidering that the incubation concentration of [Eu.3] + is four times lower than this value,i tc an be assumed that the Eu III complex is essentially non-toxic during the time frame of the imaging experiments.A nalysis of ICP-MSd ata showed that for 4 10 6 NIH-3T3 cells incubated with [Eu.3] + (50 mm)f or 2hours, ag iven cell contained 69 mm (AE 5%)ofE u III metal, consistentw ith the accumulationo f[ Eu.3] + within the mitochondria during the incubation period.

Imaging elevatedmitochondrial ATPlevels
We evaluated the ability of [Eu.3] + to visualizec hanges in the concentration of ATPi nN IH-3T3 cells upon treatment with staurosporine, ab road-spectrumi nhibitor of kinase activity that eventually induces cell apoptosis. Figure 9A shows timelapsed images of cells stainedw ith [Eu.3] + (50 mm)b efore (0 min) and after treatment with staurosporine( 10 nm). The images revealed ag raduali ncrease in the observed emission  intensity (l em 605-720 nm) in the mitochondria over a 90 minutep eriod.A t6 0m inutes post-staurosporine treatment, the Eu III emission intensity had increased by approximately 65 %, after which time the observed signal reached ap lateau ( Figure 9B). No obvious signs of cell apoptosis were evident during the time scale of the experiment, and the Eu III complex remained localized to the mitochondria. As ac ontrol, cells incubated with[ Eu.3] + withoutt he addition of staurosporine were examined, revealing less than 10 %v ariation in the Eu III emission intensity over the same time period ( Figure S64). ICP-MS studies of cells incubated with [Eu.3] + and staurosporine for 2hours revealed essentially no change in the concentration of accumulated Eu III metal,c ompared to cells grown in the absence of the kinase inhibitor ( Figure S65). Therefore, treatment with staurosporined oes not appear to perturb cellular uptake or efflux of the Eu III complex from cells.
These imaging experimentsi ndicate that the concentration of ATPi nt he mitochondrial region gradually increased during the preapoptotic period (150 min) following treatment with staurosporine. This is in agreement with previous reports of elevated ATPl evels in the cytosol 7 and mitochondria [5a] upon incubation with staurosporine. The mitochondriar emained intact during this time, suggestingt hat mitochondrial functional integrity is important during the early stages of apoptosis. This is consistent with the notion that apoptosis is an energy requiring process;t he concentrationo fm itochondrial ATPi ncreasest os upply chemical energy for av ariety of intracellular processes,such as enzymatic hydrolysis of macromolecules. [7] Monitoring depleted ATPl evels in living cells Next, the change in mitochondrial ATPl evels was monitored after treatment of cells with potassium cyanide, an inhibitor of oxidative phosphorylation. [6a] Initially, NIH-3T3 cells were incubated with [Eu.3] + in ag lucose-free growth medium, in order to inhibit ATPp roduction via glycolysis. Under these conditions, a2 0% decrease in emission intensity was observed after 2hours compared to cells grown in the presenceo fg lucose ( Figure S66). Subsequent addition of potassium cyanide (0.1 mm)r esulted in as ignificant and rapid decreasei nt he observede mission intensity,w ith almost 85 %o fl uminescence lost after 10 minutes ( Figure 10). These resultss uggestt hat mi-  tochondrial ATPi sd epleted substantially in the presence of KCN, due to inhibition of oxidative phosphorylation,w hich is consistentwith previous studies that show areduction in intracellular ATPu nder similar conditions. [6a] ICP-MS analysiso fc ells incubated with [Eu.3] + for3 0minutes with KCN under glucose starvation conditions showed less than 10 %v ariation in the concentration of accumulated Eu III metal, relative to untreated cells grown in the presence of glucose. Therefore, the possibility that incubation with KCN promotes efflux of [Eu.3] + from the cells can be ruled out.
The inhibition of oxidative phosphorylation mayr esult in perturbation of mitochondrial bicarbonate concentration, as bicarbonate stimulates the enzymatic productiono fc yclic AMP, which in turn activates mitochondrial kinase activity,r egulating ATPp roduction. [43] Having already shown that the binding of bicarbonate to [Eu.3] + is approximately 100 times weaker than that of ATPi nb uffered aqueous solution (Table 2), we wished to investigate further the effecto fv arying intracellular bicarbonatec oncentration on the emission intensity of [Eu.3] + .F ollowing ap rocedure described previously, [19a] NIH-3T3 cells were loaded with [Eu.3] + (50 mm)a nd the percentage of atmospheric CO 2 in the cell imaging chamberw as varied between2 % and 7% CO 2 .T he Eu III emission intensity observed in the mitochondria of NIH-3T3 cells after an incubation period of 60 minutes was comparedt ot hose obtained when the incubation was performed at ac onstant atmospheric CO 2 of 5% for the same time period.T he LCMS images revealed less than 5% modulation of emission intensity upon variation of the atmospheric CO 2 ( Figure S67). These microscopy experiments indicate that [Eu.3] + + does not responds ignificantly to fluctuations in the equilibrium cellular bicarbonate concentration, which is expectedc onsidering the lower apparent binding constanto f [Eu.3] + + to bicarbonate, and the larger emission intensity of the ATP-bound Eu III complex.
Ta ken together,t hese live-cell imaging experimentsd emonstrate that [Eu.3] + is capable of signallingc hanges in the concentrationo fA TP in the mitochondriau pon incubation with an ATPs ynthesis inhibiter (KCN) and ak inase activity inhibitor (staurosporine), and could provide av ersatile imaging toolf or studying cell metabolism and other ATP-requiringp rocesses in real-time,w ithin at argetedo rganelle.

Conclusions
We have developed ad iscrete, cationic Eu III complex for monitoring dynamic changes in the concentration of ATPw ithin the mitochondria of living cells. As eries of luminescent Eu III complexes, [Eu.1-4] + ,w as synthesised that bind reversibly to nucleoside polyphosphate anionsw ith differential affinities in buffered aqueous solutiona tp hysiological pH. The affinity of the Eu III complexes towards NPPs and the magnitude of the emission spectralr esponse is tunable by making modifications to the ligand structure. Hydrogen bond donor groups were introduced into the quinoline units of [Eu.1] + and [Eu.3] + to enhance selectivity towards ATPa nd ADP over monophosphate anions. Complex [Eu.3] + ,b earing two neutral carbonyl amide donors, binds most strongly to ATP( logK a = 5.8), forming a stable ternary complex that exhibits intense, long-lived Eu III luminescence. [Eu.3] + can discriminatee ffectively between ATP, ADP andA MP in ac ompetitive aqueous mediumt hat simulates the complex ionic environmentp resenti nc ells. The probe provides al inear,r atiometric emissionr esponse that is proportional to the ratio of ATP/ADP,e nabling the enzymatic hydrolysis of ATPt oA DP to be precisely monitored in realtime.
Cellular localization studies revealed that [Eu.3] + preferentially stains the mitochondria of mammalian cells, and is retained within this organelleo vere xtended time periods. We have shown that [Eu.3] + can detecta ni ncreasei nm itochondrial ATPc oncentration following treatment of cells with the kinase inhibitors taurosporine. Additionally,[ Eu.3] + was able to visualizearapid decrease in mitochondrial ATPf ollowing treatment with KCN, an inhibitor of oxidative phosphorylation. Complex [Eu.3] + offersaseveral advances in performance compared to existing ATP-responsive probes, including al uminescences ignal that is:1 )sensitive to ATPw ithin the biologically relevant concentration range (1-5 mm); 2) minimally perturbedb yc hanges in pH, dissolved oxygen or the presence of protein;a nd 3) sufficiently long-lived to avoid interference from UV-induced autofluorescence arising from biomolecules. Such probeso ffer an ew versatile tool for studying metabolism and ar ange of biological processes involving ATP, with subcellular resolution. The strategy employed here will be explored furthert od esign Eu III probesc apable of monitoring spatio-temporalA TP dynamics within different cellularc ompartments, providing ar atiometric change in emission intensity that is intrinsically normalized.