Mass spectrometry-based ligand binding assays on adenosine A1 and A2A receptors

Conventional methods to measure ligand-receptor binding parameters typically require radiolabeled ligands as probes. Despite the robustness of radioligand binding assays, they carry inherent disadvantages in terms of safety precautions, expensive synthesis, special lab requirements, and waste disposal. Mass spectrometry (MS) is a method that can selectively detect ligands without the need of a label. The sensitivity of MS equipment increases progressively, and currently, it is possible to detect low ligand quantities that are usually found in ligand binding assays. We developed a label-free MS ligand binding (MS binding) assay on the adenosine A1 and A2A receptors (A1AR and A2AAR), which are well-characterized members of the class A G protein-coupled receptor (GPCR) family. Radioligand binding assays for both receptors are well established, and ample data is available to compare and evaluate the performance of an MS binding assay. 1,3-Dipropyl-8-cyclopentyl-xanthine (DPCPX) and 4-(2-((7-amino-2-(furan-2-yl)-[1,2,4]triazolo[1,5-a]-[1,3,5]triazin-5-yl)amino)ethyl)phenol (ZM-241,385) are high-affinity ligands selective for the A1AR and A2AAR, respectively. To proof the feasibility of MS binding on the A1AR and A2AAR, we first developed an MS detection method for unlabeled DPCPX and ZM-241,385. To serve as internal standards, both compounds were also deuterium-labeled. Subsequently, we investigated whether the two unlabeled compounds could substitute for their radiolabeled counterparts as marker ligands in binding experiments, including saturation, displacement, dissociation, and competition association assays. Furthermore, we investigated the accuracy of these assays if the use of internal standards was excluded. The results demonstrate the feasibility of the MS binding assay, even in the absence of a deuterium-labeled internal standard, and provide great promise for the further development of label-free assays based on MS for other GPCRs. Electronic supplementary material The online version of this article (doi:10.1007/s11302-015-9477-0) contains supplementary material, which is available to authorized users.


Introduction
Conventional methods to measure ligand-receptor binding parameters typically require labeled probes such as radiolabeled [1] or fluorescently labeled ligands [2]. Despite the robustness of radioligand binding assays, they carry inherent disadvantages in terms of safety precautions, expensive synthesis, special lab requirements, and waste disposal. Alternatively, the addition of fluorescent moieties holds a substantial risk of affecting the pharmacological properties of a ligand; moreover, in many instances, it is also required to engineer the receptor protein, in particular for fluorescence resonance energy transfer assays [3].
The development of the mass spectrometry (MS) binding assay by the group of Wanner permits to measure binding of an unlabeled ligand to its target [4]. Instead of the radiolabeled ligand in radioligand binding assays, an unlabeled marker Electronic supplementary material The online version of this article (doi:10.1007/s11302-015-9477-0) contains supplementary material, which is available to authorized users. ligand is employed in MS binding assays. The amount of marker ligand bound to the target receptor is detected by mass spectrometry. As the mass of the molecule itself is detected, a label is not necessary. However, the marker ligand still has to fulfill the same requirements as radioligands: high affinity and selectivity for the target and low non-specific binding [5]. Therefore, it is practical to choose a ligand for MS binding applications that has already been validated as a good radioligand. This also ensures a straightforward validation of an MS binding assay by comparing it to existing radioligand binding assays.
In this study, we developed an MS binding assay for the adenosine A 1 (hA 1 AR) and adenosine A 2A receptors (hA 2A AR). The particular robustness and abundance of published results of radioligand binding assays on the hA 1 AR and hA 2A AR make these receptors good candidates for development of an MS binding assay [6]. The adenosine receptors are members of the class A of G protein-coupled receptors (GPCRs). Both receptors are important in physiology. The hA 1 AR has been related to sleep regulation, epilepsy, and asthma. The hA 2A AR is implicated in neurodegeneration, inflammatory diseases, and cancer pathogenesis. Both receptors are involved in cardiovascular physiology [6,7]. As marker ligands for the MS binding assay, we chose 1,3-dipropyl-8cyclopentyl-xanthine (DPCPX) for the hA 1 AR and 4-(2-((7amino-2-(furan-2-yl)- [1,2,4]triazolo [1,5-a]- [1,3,5]triazin-5yl)amino)ethyl)phenol (ZM-241,385) for the hA 2A AR. These ligands are well-established radioligands for their respective targets and hence a logical choice to serve as marker ligands in MS binding assays [8,9].
The development of liquid chromatography-MS (LC-MS) detection methods for non-labeled DPCPX and ZM-241,385 as marker ligands involved the following steps. Firstly, deuterated isotopologues of the marker ligands were synthesized to serve as internal standards for increased accuracy of the MS detection. In MS detection methods, it is common to add a fixed amount of an internal standard to each sample to compensate for ion suppression, sample evaporation, and instrumental drift [10]. Technically, the use of deuterium-labeled internal standards makes the MS binding assay a labeled assay, even if the marker ligand that binds to the target is unlabeled itself. Therefore, we also investigated whether the results of the MS binding assays were accurate in the absence of an internal standard. Secondly, a fast LC method was developed to separate the marker ligands from cell membrane contents in the sample. The duration of the LC separation is the limiting step for the throughput of the method so this is preferably fast, i.e., within 1 min. Thirdly, for MS detection, a triple quadrupole MS (TQMS) was employed, which has the required sensitivity to measure typical bound ligand quantities of ligand binding assays, in the pM range. In a TQMS, the parent ions with the mass of the molecule of interest are filtered by the first quadrupole, which are then fragmented in the second quadrupole. The fragmentation results in daughter ions that are analyzed by the third quadrupole. This setup ensures a high selectivity and sensitivity for the detection of a molecule of interest [11].
After establishing the LC-MS methods for detection of the marker ligands, the MS binding assays were performed with and without deuterium-labeled internal standard, and analogous to radioligand binding assays. Saturation, association, and dissociation assays were performed to determine the affinity and kinetic rates of the marker ligands DPCPX for the hA 1 AR and ZM-241,385 for the hA 2A AR. Then displacement and competition association assays were performed to determine the affinity and kinetic rates of ligands competing with the marker ligands. The ensuing results were compared to and validated with reference radioligand binding data.

General synthesis procedures
Demineralised water is simply referred to as H 2 O, as was used in all cases unless stated otherwise. 1 H and 13 C NMR spectra were recorded on a Bruker AV 400 liquid spectrometer ( 1 H NMR, 400 MHz; 13 C NMR, 100 MHz) at ambient temperature. Chemical shifts are reported in parts per million (ppm), are designated by δ, and are downfield to the internal standard tetramethylsilane (TMS) in CDCl 3 . Coupling constants are reported in Hertz and are designated as J. Analytical purity of the final compounds was determined by high pressure liquid chromatography (HPLC) with a Phenomenex Gemini 3-μm C18 110A column (50×4.6 mm, 3 μm), measuring UV absorbance at 254 nm. Sample preparation and HPLC method were-unless stated otherwise-as follows: 0.3-0.8 mg of compound was dissolved in 1 ml of a 1:1:1 mixture of CH 3 CN/H 2 O/tBuOH and eluted from the column within 15 min, with a three-component system of H 2 O/CH 3 CN/1 % TFA in H 2 O, decreasing polarity of the solvent mixture in time from 80:10:10 to 0:90:10. All compounds showed a single peak at the designated retention time and were at least 95 % pure. The synthesized compounds were identified by LC-MS analysis using a Thermo Finnigan Surveyor-LCQ Advantage Max LC-MS system and a Gemini C18 Phenomenex column (50×4.6 mm, 3 μm). The sample preparation was the same as for HPLC analysis. The elution method was set up as follows: 1-4 min isocratic system of H 2 O/CH 3 CN/1 % TFA in H 2 O, 80:10:10, from the fourth minute, a gradient was applied from 80:10:10 to 0:90:10 within 9 min, followed by 1 min of equilibration at 0:90:10 and 1 min at 80:10:10. Thin-layer chromatography (TLC) was routinely performed to monitor the progress of reactions, using aluminum-coated Merck silica gel F254 plates. Purification by column chromatography was achieved by use of Grace Davison Davisil silica column material (LC60A 30-200 μm). Solutions were concentrated using a Heidolph laborota W8 2000 efficient rotary evaporation apparatus and by a high vacuum on a Binder APT line Vacuum Drying Oven.

Membrane preparation
CHO-hA 1 AR and HEK293-hA 2A AR cells were grown as described above. Membranes were prepared as follows. Cells were detached from plates grown to confluency by scraping them into 5 ml PBS, collected and centrifuged at 700g (3000 rpm) for 5 min. Pellets derived from 20 plates (10 cm ø) were pooled and resuspended in 16 ml of ice-cold assay buffer (50 mM Tris-HCl, 5 mM MgCl 2 , pH 7.4). An Ultra-Turrax was used to homogenize the cell suspension. Membranes and the cytosolic fraction were separated by centrifugation at 100,000g (31,000 rpm) in a Beckman Optima LE-80K ultracentrifuge at 4°C for 20 min. The pellet was resuspended in 8 ml of Tris buffer and the homogenization and centrifugation step was repeated. Assay buffer (4 ml) was used to resuspend the pellet, and adenosine deaminase (ADA) was added (0.8 IU/ml) to break down endogenous adenosine. Membranes were stored in 250-μl aliquots at −80°C. Membrane protein concentrations were measured using the BCA (bicinchoninic acid) method [21].

Radioligand binding assays
The reference radioligand binding data were published before by our lab or were acquired as described before [22,23].

Membrane harvesting procedure MS binding assays
One hundred-microliter membrane aliquots containing 5 μg (CHO-hA 1 AR) or 22 μg (HEK293-hA 2A AR) of protein in assay buffer were harvested by rapid vacuum filtration through 1-μm glass fiber AcroPrep Advance 96 filter plates (Pall Corporation, Ann Arbor, MI, USA) using an extraction plate manifold (Waters, Milford, MA, USA) and a 12-channel electronic pipette (Gilson, Middleton, WI, USA). Filters were subsequently washed three times with ice-cold assay buffer and dried for 1 h at 55°C. It was essential that the filter plates were completely dry before continuing with ligand elution as described below in BSample elution.M

S binding displacement assays
Ligand displacement experiments were performed using nine concentrations of competing ligand. For the hA 1 AR, the competing ligands used were CPA, 8-CPT, ZM-241,385, and NECA, while for the hA 2A AR, they were UK-432,097, MSX-2, DPCPX, and NECA. As marker ligand DPCPX was used for the hA 1 AR at a concentration of 6 nM, and ZM-241,385 for the hA 2A AR at a concentration of 3 nM. Non-specific binding was determined in the presence of  The ligand was eluted from the ligand-receptor complex on the dried filter plates over which MS binding samples were  heating and drying gas flows 10 l/min; nebulizing gas flow 3 l/min.

Data analysis
Shimadzu LabSolutions software (Shimadzu, Kyoto, Japan) was used to analyze resulting chromatogram peaks. The peak area of the total ion count (TIC) of the daughter ions was calculated at the expected retention times, resulting in marker and internal standard peak area. To compensate for eluent evaporation and signal suppression by matrix effects from the membrane sample, marker peak area was divided by internal standard peak area (M/IS). Lower limit of quantitation (LLOQ) values of each marker ligand were defined as the lowest concentration in membrane matrix where signal to noise ratio was higher than 5, the standard deviation within and between runs in hexaplicate was lower than 20 % (and for all higher concentrations lower than 15 %), and calculation of concentration by a function derived from 1/x 2 linear regression deviated from nominal values less than 20 %. M/IS values were converted to concentration of marker ligand in pM using the function established by 1/x 2 linear regression on the 10-100-pM standard curve results. The resulting MS binding data was then analyzed with GraphPad Prism 5.0 (GraphPad Software Inc., San Diego, CA, USA). Marker ligand displacement curves were fitted to one-and twostate site binding models. k on and k off values of the marker ligands DPCPX and ZM-241,385 were derived by fitting one-phase association and dissociation models. Association and dissociation rates for the competing ligands were calculated by fitting the data to the competition association model using Bkinetics of competitive binding^ [24]. Log-transformed K i , K D , k on , and k off values from MS binding and radioligand binding assays   [18]. MS analysis showed a mass of 342.7 (M+H + ) and confirmed the incorporation of four deuterium atoms in the final product.   Tables 5 and 6. In Fig. 7, both M/IS-based and marker peak area-based (without internal standard compensation and thus completely unlabeled) data are compared with radioligand binding data. For the validation of MS binding assays, radioligand binding data that was published previously by our group was used. In the case that no in-house radioligand binding data was available, the concerning assays were performed following previously established protocols [22,23]. Radioligand binding data for the marker ligands DPCPX and ZM-241,385 on their respective targets from saturation, association, and dissociation assays was published previously ( Table 2). Displacement and competition association radioligand binding data of the competing ligands NECA (displacement on hA 1 AR and hA 2A AR), UK-432,097 (displacement on hA 2A AR), FSCPX (competition association on hA 1 AR), and LUF6632 (competition association on hA 2A AR) was available as well from previous publications (Tables 3, 4, 5, and 6). Newly acquired radioligand binding data was from radioligand displacement assays with CPA, 8-CPT, and ZM-241,385 on the hA 1 AR; radioligand displacement assays with MSX-2 and DPCPX on the hA 2A AR; and radioligand competition association assays with 8-CPT on the hA 1 AR and with MSX-2 on the hA 2A AR.
MS binding saturation of the marker ligands DPCPX and ZM-241,385 to the hA 1 AR and hA 2A AR, respectively, fitted a one-site saturation binding model (Fig. 3). DPCPX had an affinity of 3.43±0.02 nM and a B max of 17.3±0.3 pmol/mg protein for the hA 1 AR which was well in accordance with the previously found data from radioligand binding assays of 2.5 ±0.1 nM and 14±1 pmol/mg protein, respectively ( Table 2). The same was true for ZM-241,385 with an MS binding affinity of 1.03±0.07 nM and B max of 2.3±0.3 pmol/mg protein for the hA 2A AR, compared to a radioligand binding affinity of 0.60±0.07 nM and B max of 1.9±0.4 pmol/mg protein. The displacement of marker ligands DPCPX and ZM-241,385 binding from the hA 1 AR and hA 2A AR by their competing ligands fitted well to either one-state or two-state ligand binding displacement models (Fig. 4). The affinities found in MS  (Table 3) and UK-432,097, MSX-2, DPCPX, and NECA for the hA 2A AR (Table 4) were in good agreement to the radioligand binding assays. The two-state binding model fits observed for the agonists CPA and NECA on the hA 1 AR were observed in radioligand binding assays as well, and the resulting high and low affinity values were in good agreement ( Table 3). The association and dissociation of marker ligands DPCPX and ZM-241,385 to the hA 1 AR and hA 2A AR fitted well to one-phase association and dissociation models (Fig. 5), and the resulting association and dissociation rates were in good agreement between MS binding and radioligand binding assays (Tables 5 and 6).
With the association and dissociation rates validated for the marker ligands, MS binding competition association assays were performed. The competition association curves in the presence of FSCPX (hA 1 AR) and LUF6632 (hA 2A AR) yielded an Bovershoot^shape typical for slowly dissociating ligands, while in the presence of 8-CPT (hA 1 AR) and MSX-2 (hA 2A AR), the curves were typical for fast-dissociating ligands (Fig. 6). FSCPX displaced the marker ligand DPCPX completely after 120 min. The association rates of 8-CPT and MSX-2 agreed well between MS binding and radioligand binding assays, but less so in case of FSCPX and LUF6632 (Tables 5 and 6). The dissociation rates of 8-CPT, MSX-2, and LUF6632 were in good agreement as well, but not the apparent dissociation rate of FSCPX.

Preparation of internal standards
Including an internal or external standard is a good practice in mass spectrometry, to compensate for ion suppression by matrix effects from cell contents, sample evaporation, and instrumental drift [10]. We used the internal standard method as this is the most accurate manner to compensate for these sources of signal distortion and to increase the accuracy of MS methods.

MS binding assays
Affinity, association, and dissociation rates measured directly for the marker ligands DPCPX on the hA 1 AR and ZM-241, 385 on the hA 2A AR were in good agreement to the values found in radioligand binding assays (Figs. 3 and 5, Tables 2, 5, and 6). The good performance of the MS binding saturation, association, and dissociation assays in which solely the marker ligand and no competing ligand was present was a prerequisite to continue with the MS binding displacement and competition association assays.
To demonstrate the MS binding displacement assays, a combination of selective and non-selective agonists and antagonists were chosen as competing ligands. For the hA 1 AR, these ligands were CPA, 8-CPT, ZM-241,385, and NECA, and for the hA 2A AR, they were UK-432,097, MSX-2, DPCPX, and NECA. The determined affinity values were in good agreement between MS binding and radioligand binding assays for all these competing ligands (Fig. 4, Tables 3 and 4). Furthermore, the binding of agonists CPA and NECA to the hA 1 AR fitted to a pronounced two-phase displacement curve as was found before in radioligand binding assays.
Kinetic properties of ligands are of emerging interest and are thought to be important predictors of clinical performance [3,26]. Therefore, we developed and validated MS binding competition association assays, by which kinetic properties of competing ligands can be analyzed by measuring the amount of bound marker ligand at different time points, in the presence of one concentration of these competing ligands. A fast    [29], with the exception that it eventually displaces the marker ligand DPCPX completely (Fig. 6a). LUF6632 was characterized earlier as a slowly dissociating ligand selective for the hA 2A AR [14]. Dissociation rates were in good agreement between the MS binding and radioligand binding competition association assays (Fig. 7e, Tables 5 and 6), with the exception of the apparent dissociation rate of the irreversibly binding FSCPX (Table 5). Association rates found for the competing ligands in competition association assays varied more, especially for the slowly or not at all dissociating ligands FSCPX and LUF6632 (Fig. 7c, Tables 5 and 6). It has to be noted that as  saturation (a, b), association (c, d), and dissociation (e, f) assays, while values of the competing ligands were measured indirectly by displacement (a, b) and competition association assays (c-f). Affinity values in pK D and pK i (a, b), association rates in k on (c, d), and dissociation rates in log k off (e, f) were compared. it binds irreversibly to the hA 1 AR, FSCPX does not actually dissociate from the target. However, fitting the FSCPX data into the competition association model still enables the calculation of apparent association and dissociation rates. Being apparent values, they may vary between studies which could be an explanation for the diverging kinetic rates of FSCPX found in MS binding and radioligand binding assays (Table 5).
Altogether, these results validate the use of MS binding assays to determine affinity values and dissociation rates by saturation, association, dissociation, and competition association assays. However, association rate determination was only accurate by direct measurement on the marker ligands.
Necessity of deuterium-labeled internal standard As mentioned above, including an internal or external standard is a good practice in mass spectrometry. We used the internal standard method as this is the most accurate manner to compensate for sources of signal distortion. However, the use of a deuterium-labeled internal standard makes the MS binding assay a labeled assay, even if the marker ligand that binds to the target is itself unlabeled. For fast screening of new marker ligands, the use of an external standard or even no standard at all would be vastly advantageous, as the whole assay becomes an unlabeled assay. Moreover, to directly determine association and dissociation rates of non-labeled ligands would be an improvement over the use of competition association assays. Therefore, we compared the performance of the MS binding assay with and without compensation by deuterium-labeled internal standard. Although in the latter case the resulting graphs of each separate experiment were somewhat less accurate, K i and K D values could still be determined without loss of accuracy (Fig. 7b). The k off values indirectly determined by the competition association assay correlated less well with radioligand binding assays, although retaining a good coefficient of determination (Fig. 7f). The determination of k on values correlated less well with radioligand binding assays irrespective of the use of an internal standard (Fig. 7c, d). In contrast to this, the directly measured association and dissociation rates of marker ligands DPCPX and ZM-241,385 were still in good agreement with radioligand binding experiments (Tables 5 and 6).

Conclusion
We developed and validated MS binding assays for the adenosine A 1 and A 2A receptors. The results from ligand saturation, association, dissociation, and displacement assays were in good agreement with radioligand binding data. The results from competition association assays were in good agreement with radioligand binding data for dissociation rates but less so for association rates. Furthermore, we investigated the necessity to include deuterium-labeled internal standards in MS binding assays. Saturation, association, dissociation, and displacement assay results were still in good agreement with radioligand binding assays when the internal standard was not included. In competition association assays, the inclusion of an internal standard was beneficial for good correlation of dissociation rates with radioligand binding data. However, by excluding the use of internal standards in MS binding assays, it would be relatively simple to measure association and dissociation rates of a number of unlabeled ligands directly, without the need for competition association assays. We conclude that the use of deuterium-labeled internal standards is in this case unnecessary which makes the MS binding assay a truly unlabeled ligand binding assay. As this internal standard-free approach may be applied to other targets than the currently investigated adenosine A 1 and A 2A receptors, we foresee the promising future application of MS binding to directly measure binding properties by saturation, association, and dissociation assays, without the use of any labeled internal standards.