Relating ligand binding to activation gating in P2X2 receptors using a novel fluorescent ATP derivative

Ionotropic purinergic receptors (P2X receptors) are non‐specific cation channels that are activated by the binding of ATP at their extracellular side. P2X receptors contribute to multiple functions, including the generation of pain, inflammation, or synaptic transmission. The channels are trimers and structural information on several of their isoforms is available. In contrast, the cooperation of the subunits in the activation process is poorly understood. We synthesized a novel fluorescent ATP derivative, 2‐[DY‐547P1]‐AET‐ATP (fATP) to unravel the complex activation process in P2X2 and mutated P2X2 H319K channels with enhanced apparent affinity by characterizing the relation between ligand binding and activation gating. fATP is a full agonist with respect to ATP that reports the degree of binding by bright fluorescence. For quantifying the binding, a fast automated algorithm was employed on human embryonic kidney cell culture images. The concentrations of half maximum occupancy and activation as well as the respective Hill coefficients were determined. All Hill coefficients exceeded unity, even at an occupancy <10%, suggesting cooperativity of the binding even for the first and second binding step. fATP shows promise for continuative functional studies on other purinergic receptors and, beyond, any other ATP‐binding proteins.

Information about PXR receptor function is predominantly based on electrophysiological measurements. The apparent affinity (EC 50 ) for ATP considerably varies among the isoforms, ranging from low micromolar to millimolar concentrations. Another difference between the seven isoforms is the speed of desensitization, that is, the rate at which the channel activity decays in the continued presence of ATP: While homomeric P2X1 and P2X3 channels desensitize with time constants of tens of milliseconds, P2X2, P2X4, P2X5, and P2X7 channels desensitize only in the range of many seconds or minutes or not at all (Bhargava, Nicke, & Rettinger, 2013;Coddou et al., 2011;North & Surprenant, 2000). If channels are opened, they generate a permeability for the cations Ca 2+ , Na + , and K + ions (Egan, Samways, & Li, 2006). To obtain information about how P2X receptors work, electrophysiological approaches have been extensively complemented by molecular biological modifications but also by other methods as the substituted-cysteine accessibility method to identify differences in solvent accessibility (Egan, Haines, & Voigt, 1998;Kracun, Chaptal, Abramson, & Khakh, 2010), voltage-clamp fluorometry ( to study the kinetics of conformational changes (Lorinczi et al., 2012) or inter-subunit cross-linking of cysteine residues to modify conformational changes (Kowalski et al., 2014).
The crystal structure of zebrafish P2X4 (zfP2X4) has proven the trimeric nature of the channels (Hattori & Gouaux, 2012;Kawate, Michel, Birdsong, & Gouaux, 2009), thereby confirming earlier functional studies based on concentration-response relationships (Ding & Sachs, 1999), cross-linking (Nicke et al., 1998), and disulfide bond formation between engineered cysteines (Jiang et al., 2003). The crystal structure also revealed that the molecular architecture of a single subunit resembles that of a dolphin. The transmembrane domains TM1 and TM2 form the tail and the ensemble of the three TM2 helices in a channel builds the pore (Hattori & Gouaux, 2012;Kawate et al., 2009). The TM1 helices are positioned laterally with respect to the pore. In the large ectodomain, the dolphin body is organized in a β-sheet structure with lateral fenestration sites where ions can enter the central vestibule. From the body four structurally flexible domains branch, including head, dorsal fin, left and right flipper. These regions are involved in the binding of ATP and transmitting this signal to the pore. The structure is completed by the intracellular N-and C-termini which are resolved in a structure of the human P2X3 subunit (Mansoor et al., 2016). These tails form an intracellular cap that essentially influences channel kinetics, at least for this isoform.
The three ATP binding sites in a trimeric channel are located at subunit interfaces of neighboring subunits on the extracellular side. By site-directed mutagenesis and electrophysiology, several amino acids were identified to be involved in ATP binding, including in zfP2X4 the basic residues (K69, K71, R290, K308), aromatic (F183, F289), and polar residues (T184, N288). These residues form a pocket within an intersubunit cavity that is surrounded by the head and left flipper of one subunit and the dorsal fin of another one. The binding of ATP to P2X1 channels has been studied in the context of desensitization by means of a fluorescent ATP derivative Alexa Fluor ® 647 adenosine 5´triphosphate (Alexa-647-ATP) which is a full agonist (Bhargava et al., 2013).
Knowledge of the crystal structure of zfP2X4 in both the ATPfree closed and the ATP-bound open state has also paved the way to perform new functional studies on understanding how the ATP binding is transmitted within the channel to the pore opening. To address the question how many subunits are required for channel activation the results are contradictory: On the one hand, in P2X2 channels, built of concatenated subunits with defined numbers of active binding sites, less than three subunits were shown to be required for activation (Stelmashenko et al., 2012) which was confirmed by a later study, specifying that it is two subunits which are required (Keceli & Kubo, 2014). On the other hand, using voltage-clamp fluorometry to study conformational changes in P2X1 receptors, differences in the time courses of labeled subunits between receptors containing two or three binding sites were identified, supporting the notion that all three subunits are involved in channel activation (Fryatt, Dayl, Cullis, Schmid, & Evans, 2016).
Also Hill coefficients in concentration-activation relationships of various P2X channels exceeding 2 (Ding & Sachs, 1999;Jiang et al., 2003;Stelmashenko et al., 2012) argue for a relevant involvement of three subunits in the activation gating. For transmitting ATP binding to the pore opening a central role has been identified for the β-14 sheet which connects the ATP binding site on the upper end with the pore forming TM2 helix on the lower end in a subunit. This β-14 sheet interacts with the β-1 sheet of the adjacent subunit and undergoes a rotation in opposite direction after ATP binding, pulls the TM2 helix and thereby opens the channel (Du, Dong, & Zhou, 2012). Concerning the interaction of the subunits, it has been suggested that each intersubunit ATP-binding signal is transmitted first along the same subunit until residue D315 along the domain contributing K308 to the β-14 sheet and from there the signal spreads equally to all three subunits towards the pore, suggesting that in the last step the subunits contribute equally and independent to channel activation (Keceli & Kubo, 2014). It has also been proposed that the gating of P2X receptors can be subdivided into five key steps: ATP binding, tightening of the binding jaw, flexing of the lower body regions, expansion of the lateral fenestrations, and pore opening (Habermacher, Dunning, Chataigneau, & Grutter, 2016).
Despite an enormously growing insight into structure and function of P2X receptors over the past two decades, many questions remain open, including the type of cooperativity of the subunits, the exact and mutual effects between ligand binding and activation gating as well as the concentration-binding relationship itself which does not necessarily superimpose with the concentration-activation relationship of a receptor (Colquhoun, 1998). A relevant gain of information can be achieved if it is possible to monitor the ligand binding together with the activation gating, as performed previously in cyclic nucleotide-gated and hyperpolarization activated cyclic nucleotide-modulated channels by confocal patch-clamp fluorometry (Biskup et al., 2007;Kusch et al., 2010Kusch et al., , 2012. For these channels this enabled us to substantiate Hidden Markov Models describing the channel gating in considerable detail. A major prerequisite for this type of strategy is to make available a fluorescent ligand that is a full agonist with respect to ATP, has a reasonable potency to activate the receptor, is bright enough to do optical recording and does not have any other negative side effects. Herein we designed and synthesized a novel fluorescent ATP derivative in which the dye DY-547P1 was coupled to the 2-position of the adenine ring. This position was chosen because of the promising property of 2-MeS-ATP to be a full agonist for P2X2-receptors with an enhanced potency compared to ATP (Coddou et al., 2011). This is in contrast to ribose-labeled ATP derivatives. We studied binding by an optical approach and the binding-induced activation of P2X2 channels by electrophysiology. Our novel fluorescent ligand fATP bears great potential for functional studies on any other purinergic receptor and, beyond this, on any other protein binding ATP.

| Molecular biology
Rat P2X2 (rP2X2) in pcDNA5/FRT/TO (RRID: Addgene_137071) was kindly provided by Günther Schmalzing (RWTH Aachen University, Aachen, Germany). The mutation H319K (RRID: Addgene_137072) was introduced with the QuickChange sitedirected mutagenesis (Stratagene) using a primer containing the modified base triplet with Phusion polymerase followed by DPN1 digestion. The obtained clones were verified by restriction analysis and sequencing. Cells were seeded on glass coverslips for electrophysiological measurements and used 24-48 hr after tetracycline induction. For binding measurements cells were plated on chambered glass coverslips (ThermoFisher Scientific) coated with poly-l-lysine (Sigma-Aldrich) and measured 48 hr after induction. This study was not pre-registered and no institutional ethical approval was required.

| Cell culture
Cell lines and plasmids will be shared upon resonable request. Binding, ligand, and reference are the pixel-wise image signal. l low , l high, r low , r high are the average signal of bulk and cell-interior signals for the ligand and the reference dye, respectively. r low , r high were determined as the two main peaks of intensity histograms of the reference image. l low and l high were determined by calculating the mean intensity of the pixels in the ligand channel, for which the reference signal was [r low -3,000 < reference < r low + 3000] and [r high -3,000 < reference < r high + 3000], respectively. As these values can be determined without manual interactions, the approach can be easily be automated.

| Confocal microscopy
As substantial noise was observed within the bulk solution as well as inside the cells, the cell-bulk interfaces were selected using a Sobel operator (Sobel, 1990) on the reference fluorophore signal.
This mask signal was further filtered to remove speckles and to adjust the width of the considered edges and finally an appropriate threshold was applied. Note that only cell-bulk boundaries and not cell-cell boundaries are included in this approach.
For cells not expressing the receptor or having excess of competing unlabeled ligand, a small systematic error, that is, a negative apparent binding-signal was observed. This signal is linearly dependent on the concentration of the labeled ligand. It is therefore likely due to the non-gaussian-nature of the intensity distribution or differences in the detectors for fluorescence and ligand channels, as it was influenced by the mode of calculation of l low , l high (e.g., mean or median). To compensate the error, appropriate negative controls were measured and subtracted. Binding measurements and analysis were performed by different persons.
The reference dye DY647 was purchased from Dyomics GmbH, Jena, Germany.

| Electrophysiology
Current recording were performed with a standard patch-clamp technique (Hamill, Marty, Neher, Sakmann, & Sigworth, 1981)  Saturation of activation was determined with ATP at 100 µM for wt P2X2 and 1 µM for P2X2 H319K. The currents were recorded with an Axopatch 200B or HEKA EPC 10 amplifier in combination with the ISO3 hard and software or Patchmaster software hard and software, respectively. The sampling rate was 2 to 10 kHz and the recordings were on-line filtered at 1 and 2.9 kHz using a 4-pole Bessel filter. The currents were recorded at a constant holding potential of −50 mV.

| Quantification and statistical analysis
Concentration-activation relationships were constructed from the maximum currents during a solution application. Measurements with rundown of the current of more than 10% were excluded from the analysis. These current amplitudes were normalized with respect to the current at saturating ATP (wt 100 µM; H319K 1 µM in each individual cell) and the resulting data points were fitted with. where BC 50 is the ligand concentration producing half maximum occupation at the receptor and H is the Hill coefficient which may significantly differ from n. H is influenced by the number of binding sites and also by possible cooperativity effects but it provides a lower limit for the number of bound ligands. In case that two ligands compete for the same receptor, the situation can be illustrated as shown by the scheme in Figure 4b. The two coupled binding reactions can be described by the Gaddum equation proposed already decades ago previously (Colquhoun, 2006;Gaddum, 1957;Kenakin, 2017). After adaptation to our notation,  (6) for BC 50fATP and BC 50ATP and Hfor ATP by fitting the experimental data for ATP ( Figure 4a). (2) To determine the Hill coefficient at the lowest measurable ligand concentrations where both binding and gating are <0.1, the limiting slope was calculated (Patneau & Mayer, 1990;Wahl, Madsen, Banke, Krogsgaard-Larsen, & Schousboe, 1996) from loglog plots and fitting a straight line through the data points <0.1 ( Figure 3b). The slope determined in this way reports the Hill coefficient directly because in the limit of [fATP]→0 Equations (2) and (3) reduce to and respectively.
Experimental data are given as mean ± SEM.

| Structural modeling
The homotrimeric rat P2X2 structure (UniProt accession ID: P49653) was generated by homology modeling, using the 2.77 Å X-ray crystal structure of the human P2X3 ion channel in the ATP-bound open state (Protein Data Bank [PDB]: 5SVK) as a template (Mansoor et al., 2016). The modeling was carried out by the SWISS-MODEL server (Biasini et al., 2014) based on a target-template alignment with a sequence identity of 50.7% and coverage of 75% using residues 13-370 of a channel subunit. The quality of the model was validated by the MolProbity server (Chen et al., 2010), yielding an overall MolProbity score of 1.35.

| Chemical synthesis of fATP
We synthesized the novel fluorescent ATP derivative 2-[DY-547P1]-AET-ATP by coupling the dye DY-547P1 to the 2-position of the purine ring via an amino-ethyl-thio spacer (Figure 1a). The compound is termed from hereon 'fATP'. The chemical synthesis is described in detail in Supplementary Methods.  (Figure 1b) was normalized with respect to the maximum current amplitude at saturating ATP (100 µM for wt P2X2 and 1 µM for P2X2 H319K) and the data points were plotted as function of the respective ligand concentration (Figure 1c). Fitting the data points with Equation (2) yielded that the apparent affinity with fATP was more than an order of magnitude below that with ATP.

| The fluorescent ATP-derivative fATP is an agonist with similar efficacy as ATP for P2X2 channels
Determining the concentration-binding relationship requires to measure the binding of fATP at saturating concentrations to refer the binding at all other concentrations to this value. Using wt P2X2 receptors this is impossible because at concentrations above ~30 μM it becomes critical to delineate the specific binding of the bound labelled ligands from freely diffusing labelled ligands in the bath solution (Biskup et al., 2007). Moreover even if possible, the required high concentrations of the labelled ligand would become exceedingly expensive. We therefore decided to repeat the measurements in a mutant with significantly increased apparent affinity with the idea that also the binding is respectively tighter.
We chose the H319K mutation in P2X2 channels which was reported to increase the apparent affinity 40-fold with respect to wt channels (Clyne, LaPointe, & Hume, 2002). Histidine 319, unique in  Herein, we measured ligand binding in ensembles of intact HEK293 cells instead in excised patches from Xenopus oocytes (Biskup et al., 2007;Kusch et al., 2010;Nache, Eick, Schulz, Schmauder, & Benndorf, 2013). This is because membrane patches (6) log I∕I max = H a log fATP −H a log EC 50

| Binding of fATP to P2X2 H319K channels
of Xenopus oocytes showed significant non-specific binding of fATP which was not observed in HEK293 cells. However, the signal intensity in patches from HEK293 cells was significantly lower than that in patches from Xenopus oocytes and, thus, only measurements in ensembles of cells provided a sufficiently high signal intensity.
Typical images obtained for quantifying the binding of fATP to P2X2 H319K channels are shown in Figure 3a. In induced cells (P2X2 H319K (+)), an increase of the fATP concentration from 100 to 300 nM shows a clearly visible increase of staining at the membrane due to specific binding to the channels. Specificity is demonstrated by comparison with both non-induced cells (P2X2 H319K (−)) at 300 nM and 10 µM and also in induced cells in the presence of competing ATP added at the high concentration of 300 μM. The concentration-binding relationship for fATP to P2X2 H319K channels shows saturation below 10 μM (Figure 3b) and that the value for half maximum binding is closely similar to the EC 50 value obtained for activation (c.f. Figure 1d).
Both the concentration-binding and the concentration-activation relationship required Hill coefficients larger than unity and are thus steeper than predicted by a Langmuir isotherm assuming a single binding site. This suggests not only for activation but also for ligand binding itself a cooperative action of the subunits. Furthermore, the binding curve saturates at lower concentrations than the activation curve (Figure 3b), which is typical for channels that are activated by the binding of more than one ligand (Colquhoun, 1998).
To consider the Hill coefficient separately at the lowest measurable ligand concentrations where both binding and gating are <0.1, we examined the limiting slope (Patneau & Mayer, 1990;Wahl et al., 1996) for both relationships by building log-log plots and fitting a straight line through the data points <0.1. The slope determined in this way reports the Hill coefficient directly because in the limit of [fATP]→0 Equations (2) and (3) reduce to Equations (6) and (7), respectively. The result is that both Hill coefficients still exceed unity and are also close to each other (Figure 3c), indicating that significant cooperativity is present already at very low degrees of occupancy, that is, for the first and second binding step.

| Binding of ATP to P2X2 H319K channels
Determination of the BC 50 and H b value for fATP allowed us also to determine the BC 50 value for ATP itself by displacement experiments. Under equilibrium conditions, this scheme obeys Equation (5), which is in fact the Gaddum equation adapted to our notation (Colquhoun, 2006;Gaddum, 1957;Kenakin, 2017), and is also related to the Schild equation (Schild, 1957). Using the determined BC 50 value for fATP as BC 50fATP and the determined Hill coefficient, H b , Equation (5) was fitted to the data points yielding the indicated BC 50ATP and H bATP value for ATP. It should be noted that the Hill coefficient was similar for the binding of ATP and fATP.

| Binding of fATP to wt P2X2 channels
We finally measured the binding of the novel fluorescent ATP derivative fATP on labeling wt P2X2 channels expressed in HEK293 cells the concentration-binding-relationship was obtained by assuming that the BC 50 and the EC 50 value are also similar as in H319K channels. Under this assumption the data points were scaled with respect to the data point at 10 μM which was set to 50% binding. Then Equation (3) was fitted with a fixed BC 50 value of 10 μM, yielding H b = 1.28 (Figure 5b) which is similar to that in H319K channels.

| D ISCUSS I ON
In this study, we synthesized and functionally characterized the novel fluorescent ATP derivative fATP regarding its agonistic effects on both P2X2 channels and the P2X2 H319K mutant with significantly enhanced apparent affinity for ATP. fATP proved to be a full agonist with respect to ATP that can emit bright fluorescence when bound to the channels. We used this compound to determine both the concentration-binding and the concentration-activation relationships at equilibrium conditions, which allowed us to relate ligand binding to activation gating and to determine the respective half maximum concentrations and Hill coefficients. In addition, the Hill coefficients were determined also at an occupancy <10% by the method of the limiting slope. The results suggest cooperative binding, already for the first and second binding step. Finally, the fluorescent fATP was also used to determine the occupancy of ATP at the binding sites by applying the Gaddum equation (Gaddum, 1957).
Together, the fluorescent ATP derivative fATP presented herein has properties that enabled for the first time to determine a concentration-binding relationship in activated P2X2 channels and, moreover, to prove cooperativity by binding on the basis of a Hill coefficient exceeding unity.
The fluorescent ATP derivative fATP designed and investigated herein has coupled the dye DY-547P1 via an aminoethylthio-linker to the 2-position of the adenine. Though with respect to ATP the fluorescent derivative fATP is a full agonist in P2X2 H319K channels, its potency is reduced by more than an order of magnitude. This reduced potency is presumably caused by the dye moiety and a steric F I G U R E 3 Occupancy of fATP at P2X2 H319K channels. (a) Representative confocal images for quantification of the fATP binding to P2X2 H319K channels. In induced cells (P2X2 H319K (+)), an increase of the fATP concentration from 100 to 300 nM caused a notable increase of the staining at the membrane. Respective staining is not visible in uninduced cells (P2X2 H319K (-)), even at the high concentration of 10 µM fATP. Also, no fATP binding was observed in P2X2 H319K (+) cells with competing ATP at the high concentration of 300 μM. Data look up tables were scaled to represent all shades present in the image, color intensities are not comparable between the images. (b) Concentration-binding relationship for fATP to P2X2 H319K channels. Each data point was obtained from 25 to 30 images each containing about 100 cells. Cells from at least six cultures were used. The data points were fitted with Equation (3) yielding the indicated parameters. The gray curve shows the concentration-activation relationship from Figure 1d for comparison. (c) Limiting slope in ligand binding and channel activation. Superimposition of the log-log plot of the data for steady-state binding of fATP to P2X2 H319K channels and the resulting steady-state activation. In the concentration range between 10 and 100 nM the limiting slope was determined by fitting straight lines to the data points according to Equations (6) and (7) yielding H b = 1.22 and H a = 1.40 for binding and activation, respectively limitation in the binding pocket. It is possible that the relatively long linker in fATP (see structure in Figure 1a) provides sufficient flexibility to enable binding. Notably, we demonstrated that in fATP it is indeed the binding that is disturbed (Figures 3b, 4a).
Two other ATP derivatives have been used previously to study P2X channels. One derivative is Alexa-647-ATP in which the dye is coupled to the 2',3' position of the ribose via an aminoethylcarbamoyl-linker (Bhargava et al., 2013). It was applied to study strongly desensitizing P2X1 channels and to a non-desensitizing P2X2-1 chimera (Rettinger & Schmalzing, 2004). Alexa-647-ATP was used in this report to elegantly study desensitization, as typical for P2X1 channels, with a half maximum concentration as low as 3 nM. In the other fluorescent ATP derivative, BODIPY-TR ATP, the dye was coupled to the ribose (Kowalski et al., 2014). It was used to study the agonist  (Li, Silberberg, & Swartz, 2013). Notably, in the presence of millimolar Ca 2+ and Mg 2+ , a much F I G U R E 5 Binding of fATP to wt P2X2 channels. (a) Representative confocal images of HEK293 cells expressing wt P2X2 channels. The cells were exposed to 3 μM fATP. (b) Concentration-binding relationship. Based on the assumption that the BC 50 and the EC 50 value for H319K channels are close by, we scaled the data points with respect to the data point at 10 μM which was set to 50% binding and fitted with fixed BC 50 = 10 μM the data points with Equation (3), yielding the Hill coefficient for binding H b = 1.28. Each data point was obtained from 8 to 9 images each containing about 100 cells. Cells from at least three cultures were used larger Hill coefficient of 2.3 was reported (Ding & Sachs, 1999), suggesting that the divalent ions critically influence the Hill coefficient.
Furthermore, it should be noted that the Hill coefficients in the latter study were determined from single-channel recordings which leads a priori to steeper concentration activation relationships because single-channel recordings do not include the natural variability among the channels present in multichannel recordings.
Concerning the position of the H319 mutation, it has been outlined above that it is located in the β-14 linker and that it has been identified as pH sensor (Clyne et al., 2002). The major functional property of the H319K mutation in the context herein is a shift of steady-state activation and binding to lower concentrations by more than an order of magnitude (c.f. Figure 1c,d) and a slowed deactivation kinetics (c.f. Figure 1b) suggesting a slowed off-rate. In principle, this leftward shift can be explained by a modification of the binding site itself, a long-distance effect from a remote site, or a changed isomerization constant of a processes following the ligand binding (Colquhoun, 1998). The first option can be excluded because the H319K mutation is distant from the binding site. The second option, the long distance effect, is to consider here because the H319K mutation is located in the β-14 sheet which is known to be involved in the transmission pathway from the ATP binding to the pore (Keceli & Kubo, 2014). Notably, the distance from H319 to all three binding sites in a P2X2 channel is considerably long. Using a homology model based on the P2X3 structure (Mansoor et al., 2016) this distance was determined to be 31.8 Å to R290 in the binding site of the own subunit and 28.1 and 34.4 Å to the respective other two R290 residues ( Figure 6). In the light that in wt P2X2 channels protons control the amount of positive charge of histidine 319, the conclusion is that the extracellular pH controls indeed the occupancy with ATP in the remote binding sites. At the second glance this is not so surprising because the β-14 sheet and the binding sites must functionally be intimately coupled to enable the control of channel activity. Therefore, the effect of H319 can be simply seen as a reciprocal control of the binding sites from the β-14 sheet, in some analogy to the reciprocal control of the occupancy in the cyclic nucleotide binding domains in hyperpolarization-activated cyclic nucleotide-modulated channel (isoform 2) (HCN2) channels by voltage-induced activation of the channel (Kusch et al., 2010).
This remote control of the occupancy is also not surprising from  (Colquhoun, 1998(Colquhoun, , 2006 Figure 5).
Whereas we have not tested fATP with other P2X subtypes yet, it is conceivable that our novel fluorescent ligand fATP can also be used with other P2X receptors. This would enable fast and safe optical F I G U R E 6 Steric relationship between the H319 and the ATP binding sites. One subunit is shown in ribbons whereas the other two subunits are in surface representation. The distances between the Cα-atoms of the mutation site H319 in the β-14 sheet (purple) and the three binding sites, here R290 was used, are indicated by dashed lines yielding 31.8 Å to R290 within the subunit and 34.4 and 28.1 Å to R290 in the other two subunits binding assays on living cells under physiological conditions, thereby avoiding any radioactive labelling or the need of membrane preparation and probe separations after reaching an equilibrium. It would be particularly interesting to see whether the mysterious process of desensitization could be further unravelled by identifying corresponding components of ligand binding at low ligand concentrations and, ideally, as function of time during the running desensitization process as well during its recovery. It seems to be also a good idea to test the binding performance of fATP in metabotropic purinergic P2Y receptors (von Kugelgen & Hoffmann, 2016). Another challenging perspective would be to extend the present studies to measure in P2X channels binding and activation in parallel, as performed previously with fluorescent cyclic nucleotides in HCN2 or cyclic nucleotide-gated channel A2channels (Kusch et al., 2010;Nache et al., 2013). Such combined measurements provide an enormous gain of information with respect to measure either binding or current alone and therefore allow to substantiate kinetic schemes for the activation gating in considerable detail (Biskup et al., 2007;Kusch et al., 2012).

ACK N OWLED G EM ENTS
We thank K. Schoknecht, S. Bernhardt, A. Kolchmeier and C. Ranke for excellent technical assistance. The Igor Routine to import laser scanning microscope-files by Stephen Ikeda (NIH/NIAAA) was a valuable basis for our in-house data evaluation. The work was funded by the University Hospital Jena.
All experiments were conducted in compliance with the ARRIVE guidelines.

CO N FLI C T O F I NTE R E S T
The author(s) declare no competing interests.

AUTH O R CO NTR I B UTI O N S
C.S. conducted the patch-clamp recordings and analyses and also did the molecular biology. RS did the optical measurements on ligand binding together with C.S. and wrote the analysis software. C.U.
contributed with patch-clamp recordings. M.O. did the homology structures. F.S., A.S., and T.S. designed the fluorescent ATP derivative. F.S. performed the chemical synthesis. K.B. designed the study together with C.S. and wrote the manuscript.