Fluorescent Tools for the Imaging of Dopamine D 2 -Like Receptors**

,


Introduction
Dopamine receptors are prominent members of the large family of G protein-coupled receptors (GPCRs) and are endogenously activated by the biogenic amine dopamine (Figure 1).Depending on their signaling pathways and sequence homology, dopamine receptors are divided into two families.The D 1 -like receptors (D 1 R, D 5 R) are mainly coupled to Gα s and stimulate adenylyl cyclase (AC) upon activation while the D 2 -like receptors (D 2 R, D 3 R, D 4 R) mediate signal transduction predominantly via Gα i/o , which leads to inhibition of AC. [1] Due to its high level of expression in the central nervous system (CNS) the D 2 -like family constitutes a highly interesting target for the treatment of several neurological diseases such as Parkinson's disease (PD), addiction, or schizophrenia. [2,3]Especially the D 2 R is of utmost importance being the main target of agonists like pramipexole for the treatment of PD or antagonists like haloperidol for the therapy of schizophrenia (Figure 1). [4,5]espite intense research on both diseases in recent years the aforementioned drugs are still widely used in defiance of their approval dating back more than 25 years. [6,7]valuating binding affinities at the desired receptor is often one of the first steps concerning the development of new ligands.So far, binding properties have been mainly determined in radioligand competition binding experiments.Fluorescent ligands offer appealing alternatives to radioligands, especially, when it comes to targeting receptors using live cells and tissues or investigating the outcomes of drug-receptor interactions. [8,9][18][19][20][21] In recent decades, fluorescence-based binding assays have gained importance and are nowadays becoming almost as common as radiochemical binding assays. [11]The pros and cons of these two methods are relatively balanced.In particular, fluorescencebased methods can be considered more advantageous in terms of safety, accessibility, regulatory requirements, and costs.Distinct advantages are often seen in the simplicity of performance and the fact that modern FRET/BRET-based assay systems enable kinetic measurements in real-time. [13,22]Radioassays, in contrast, are generally very robust, no receptor modification is required, there is access to autoradiography and there is high sensitivity, to name just a few advantages. [23]Therefore, both methods will coexist and can be selected for the characterization of GPCRs and their ligands depending on the task at hand.Even though FRET/BRET-based assays deliver reliable results and offer several advantages over e. g. radioassays, modifications have to be made to the GPCR and cell protein of interest.[26] This enables investigations under more physiological conditions, not least because the cellular processes are not impaired by the addition of chemical substances, which are often required for signal detection in conventional assays.However, although not all processes leading to the observed signal in label-free readouts are fully understood and are sometimes referred to as "black box" readouts, [27] these techniques offer highly interesting alternatives for studying ligand-receptor interactions.
In this study, we report the synthesis of a small series of fluorescent ligands for D 2 -like receptors and their application in binding studies using molecular brightness in fluorescence microscopy.The study aims to find valuable fluorescent tracers that can contribute to further exploration of the complex functions and interactions of the D 2 -like receptor family.

Design Rationale
Fluorescent ligands share a common structure consisting of a ligand scaffold, a linker region, and a fluorescent dye.Each of these components has a significant impact on the intended use of the fluorescent probe and must therefore be carefully selected.[30] Its antagonistic mode of action makes it highly suitable for use in live cells since agonists might lead to receptor internalization. [30]oreover, additions of bulky structures to the aniline moiety have been described to be well tolerated in terms of affinity, making this part of the molecule a perfect attachment point for the linker. [31,32]In agreement with these SAR results, a cryo-EM structure of the D 2 R in complex with spiperone showed that this position points towards the extracellular space and is not involved in ligand-receptor interaction. [33]The solved cryo-EM structure with spiperone and the D 2 R is shown in Figure S41 (Supporting Information (SI)) and supports our design rationale.Linker length is a very important feature of a fluorescent ligand because it must allow the fluorescent dye to reach outside the binding pocket in order not to interfere with ligand binding.In contrast to previously reported N-(paminophenethyl)spiperone-based (NAPS) fluorescent ligands, where the dye was directly attached to the ligand scaffold, [31] we have designed two different linkers to gain more information about the necessary distance between ligand scaffold and dye.A short linker based on γ-aminobutyric acid was designed to cover a rather short distance of 5 atoms to reduce the overall size of the ligand.The long linker covering a length of 18 atoms was based on polytheyleneglycole (PEG) units.This approach is very popular because PEG units are chemically stable, show only little interaction with cell membranes, and increase water solubility of the fluorescent probe. [34]Since our goal was to design a fluorescent probe for microscopy we had to choose multilaterally usable dyes.Recent publications have reported that the 5-TAMRA dye is suitable for confocal microscopy and TIRF microscopy. [13,35]As an alternative we selected DY-D549-P1 because it has similar spectral properties to 5-TAMRA.Since both dyes are hydrophilic, they are less prone to interact with cell membranes than, for example, BODIPY fluorescent dyes, resulting in reduced nonspecific binding. [36]Compared to previously published fluorescent ligands for D 2 -like receptors, [31,37,38] our antagonistic ligand scaffold together with the more hydrophilic 5-TAMRA and DY549 dyes combines better solubility properties and is not susceptible to receptor internalization when applied to whole cells.In addition, TAMRA ligands have already repeatedly demonstrated their suitability in NanoBRET assays, which opens up another possible application for our fluorescent ligands. [13,22,39]emistry One of our aims was to find out how the selection of the respective dye and variation of linker length between the ligand scaffold and the dye influence binding characteristics.Therefore, three different fluorescent ligands (16, 17, and 20)  differing in either the dye and/or linker length were designed.The synthesis of precursor NAPS (11) was carried out as previously described in ten steps (Scheme 1). [32,40]In a first step intermediate 1 was synthesized from commercially available aniline and N-benzylpiperidin-4-one in the presence of HOAc and TMSCN.Subsequently, the formed nitrile moiety was converted into an amide function using concentrated sulfuric acid leading to compound 2. Reaction with DMF•DMA in methanol resulted in cyclization to obtain 3. Reduction of the imidazolinone moiety with NaBH 4 led to 4. Cleavage of the benzyl group with ammoniumformiate in presence of Pd/C yielded 5. Subsequently, an alkylation with 4-chloro-1-(4fluorophenyl)butan-1-one in the presence of KI was performed to get 6.At the same time, 2-(4-aminophenyl)ethan-1-ol (7) was Boc-protected to deliver 8 followed by a bromination with NBS to obtain 9. Compound 10 was synthesized by N-alkylation of 6 with 9 in the presence of KOH and TBAB in toluene.TFA in DCM was used for deprotection of the aniline group to yield 11.
In order to provide the fluorescent ligands with their dyes, corresponding spacers 12 and 13 were prepared for coupling with ligand scaffold 11 by mono-Boc protection (Scheme 2A-B).diyl))bis(oxy))bis(propan-1-amine) and γ-aminobutyric acid were used as starting materials.For fluorescent ligands 16 and 17 succinic anhydride was added to the D 2 -like scaffold 11 forming a terminal carbonic acid.This was subsequently coupled with spacer 12 (Scheme 2C) using HATU/DIPEA in DMF to yield intermediate 14. [25] Deprotection with TFA/DCM delivered precursor 15.The precursor for fluorescent ligand 20 was synthesized in a slightly different manner.Spacer 13 was directly coupled to 11 using HATU/DIPEA in DMF to obtain 18, while cleavage of the Boc group under acidic conditions gave precursor 19.In a final step, precursors 15 and 19 and the commercially available NHS-ester of 5-TAMRA or DY-D549-P1 were coupled in DMF in the presence of triethylamine. [22,39]urification with preparative HPLC afforded highly pure fluorescent ligands 16, 17, and 20 (> 98 %) with great stability (Figure S1-S4; SI) in good to excellent yields (60-90 %).

Pharmacological Characterization
In a first step, the synthesized compounds were tested for their binding properties to the D 2long R. While 5-TAMRA labeled probes 17 and 20 showed very high affinity (pK i = 8.25 and 8.24) for the target receptor in the single-digit nanomolar range, the addition of DYÀ D549-P1 to the D 2 -like scaffold resulted in a loss of affinity of 1.5 orders of magnitude (pK i = 6.67).Subsequently, the selectivity of ligands 17 and 20 over the entire dopamine receptor family was determined.As expected, the selected compounds showed moderate to high affinity for the other D 2 R-like receptors, namely D 3 R (pK i = 8.29 and 8.57) and D 4 R (pK i = 7.53 and 7.78), whereas lower affinity for D 1 -like receptors was measured (cf. Figure 2; Table 1).In general, it could be observed that variations concerning the fluorescent dye had much more influence on binding affinities than different spacer lengths.Combined, compound 20 showed the highest affinities at all three D 2 -like receptors and we hence decided to use it as a representative of this fluorescent ligand series to determine its mode of action at D 2long R, D 3 R, and D 4 R.We confirmed 20's antagonistic behavior using a BRET-based G o1 heterotrimer dissociation assay (G o1 -CASE) (Figure 3, Table 2). [41]A slight decrease was observed in the agonist mode only at high concentrations of 1 μM, most likely due to optical interference of 20 with the BRET components of G o1 -CASE, as observed in a previous study with a 5-TAMRA-labeled histamine H 3 receptor     ligand (Figure 3A). [13]In the antagonist-mode, however, 20 reduced the 1 μM dopamine-induced BRET response already at low nanomolar concentrations .78; Table 2; dopamine EC 50 values required for the calculation of pK b values were taken from the data shown in Figure S5, SI), further demonstrating that it acts as a neutral antagonist at D 2 -like receptors.This is of importance for microscopy studies, since agonistic properties of a ligand can lead to receptor internalization and subsequent degradation.

Fluorescence Properties
Fluorescence excitation and emission spectra were recorded in order to further analyze the final compounds (in PBS containing 1 % bovine serum albumin (BSA)) for their fluorescence properties.These are usually not greatly affected by the addition of a ligand scaffold if the fluorophore is not chemically modified, as is the case with the dyes 5-TAMRA and DY549-P1.Nevertheless, the absorption and emission spectra as well as the quantum yield should be determined for the new fluorescent ligands.
The emission and excitation spectra of 16, 17, and 20 are shown in Figure 4, while the excitation and emission maxima are presented in Table 3.The 5-TAMRA-labeled ligands 17 and 20 showed excitation maxima at 559/562 nm and emission maxima at 583/584 nm.Excitation maxima at 562 nm and emission maxima at 576 nm were recorded for the Dyomicslabeled compound 16.Quantum yields were determined in PBS + BSA 1 % with cresyl violet perchlorate as a red fluorescent standard according to a previously described procedure and are all in a good range of 36-39 % (Table 3). [42]Measurements in PBS buffer, without 1 % BSA, gave decreased quantum yields, especially for compounds 16 and 17 (Table 3).Given these results, all ligands should be very well suitable for the use in fluorescence microscopy.

Microscopy
Confocal microscopy imaging was then used to visualize the binding behavior of compound 20.Therefore, HEK293T cells were transiently transfected with DNA of the D 2long R Cterminally fused to GFP 2 , allowing us to identify cells expressing high levels of the receptor (Figure 5A).Confocal microscopy images were acquired 48 hours after transfection (Figure 5).Once a suitable cell was identified, 20 (c = 50 nM) was added and rapid accumulation of fluorescence at the cell surface was observed (Figure 5B).This is due to the rapid association of 20 with the D 2long R expressed on the cell membrane.These observations demonstrate the applicability of 20 for    fluorescence microscopy experiments as a labeling tool to visualize D 2long R in live cells.

Molecular Brightness
We then used molecular brightness to quantify by an all-optical method the affinity of 20.When performed using two spectral channels, this approach allows correlating the number of receptors and fluorescent ligand at the membrane of intact, living cells. [21]For this, HEK293-AD cells were transiently transfected with the C-terminally tagged D 2long R-mNeonGreen and incubated with different concentrations of 5-TAMRA-labeled 20.
After allowing equilibration, the ligand was washed out to avoid background of freely diffusing ligand and basolateral membranes were immediately imaged on a confocal microscope (Figure 6A).The photon counts per pixel and their variance in a region of interest were used to calculate the average number of fluorescent emitters per confocal excitation volume and their brightness.The emitter numbers were calculated for both mNeonGreen (receptor) and for 5-TAMRA (ligand) channels and plotted against each other.Figure 6B shows the values and superposed a linear regression fit (Figure 6B).A slope of 0.80 � 0.03 obtained for 575 nM of 20 indicates a high, albeit partial, occupancy of the receptor by the ligand.Likely reasons for not observing full receptor occupancy could be receptor signal from areas within the confocal volume and proximal to the membrane (e.g. early endosomes), quick ligand dissociation kinetics, or slight photobleaching of the TAMRA.Nonetheless, the concentration response curve, formed by the slopes plotted against the corresponding ligand concentrations, yields a pK d value of 8.12 � 0.11 mirroring the radioligand binding data (Figure 6C).

Conclusions
In this study, a series of three fluorescent ligands were designed and synthesized containing different linkers and fluorescent dyes.The well-known antagonist spiperone was selected as a scaffold as it combines excellent affinity among D 2 -like

Radioligand competition binding experiments at the dopamine receptors
Cell homogenates containing the D 2long R, D 3 R, and D 4.4 R were kindly provided by Dr. Lisa Forster, University of Regensburg.Homogenates containing the D 1 R and D 5 R were prepared and radioligand binding experiments with cell homogenates were performed as previously described with minor modifications. [45,46]For radioligand competition binding assays homogenates were incubated in BB at a final concentration of 0. ) at room temperature, bound radioligand was separated from free radioligand through PEI-coated GF/C filters using a Brandel harvester (Brandel Inc., Unterföhring, Germany), filters were transferred to (flexible) 1450-401 96-well sample plates (PerkinElmer, Rodgau, Germany) and after incubation with scintillation cocktail (Rotiszint eco plus, Carl Roth, Karlsruhe, Germany) for at least 3 h, radioactivity was measured using a MicroBeta2 plate counter (PerkinElmer, Waltham, MA, USA).Competition binding curves were fitted using a four-parameter fit ("log(agonist) vs. response-variable slope").Calculations of pKi values with SEM and graphical presentations were conducted with GraphPad Prism 9 software (San Diego, CA, USA).

G o1 heterotrimer dissociation assay
HEK293A cells (Thermo Fisher) were transiently transfected with the G o1 BRET sensor, G o1 -CASE (http://www.addgene.org/168123/), [41]long with either D 2long R, D 3 R or D 4 R using polyethyleneimine (PEI).Per well of a 96-well plate, 100 μl of freshly resuspended cells were incubated with 100 ng total DNA (50 ng receptor and 50 ng G o1 -CASE) mixed with 0.3 μl PEI solution (1 mg/ml) in 10 μl Opti-MEM (Thermo Fisher), seeded onto poly-D-lysine (PDL)-pre-coated white, F-bottom 96-well plates (Brand GmbH) and cultivated at 37 °C, 5 % CO 2 in penicillin (100 U/ml)/streptomycin (0.1 mg/ml)-, 2 mM Lglutamin-and 10 % fetal calf serum (FCS)-supplemented Dulbecco's modified Eagle's medium (DMEM; Thermo Fisher).48 hours after transfection, cells were washed with Hank's Buffered Salt Solution (HBSS) and incubated with a 1/1000 furimazine (Promega; cat no.N1663) dilution in HBSS for 2 minutes.Next, the baseline BRET ratio was recorded in three consecutive reads, cells were stimulated with serial dilutions of 20 or vehicle control, and BRET was recorded for another 15 reads.For experiments in antagonist mode, serial dilutions of 20 were added together with furimazine before the experiment and 1 μM dopamine or vehicle control was added after the first three baseline recordings.All experiments were conducted using a ClarioStar Plus Plate reader (BMG Labtech) with a cycle time of 120 seconds, 0.3 seconds integration time, and a focal height of 10 mm.Monochromators were used to collect the NanoLuc emission intensity between 430 and 510 nm and cpVenus emission between 500 and 560 nm.BRET ratios were defined as acceptor emission/donor emission.The basal BRET ratio before ligand stimulation (Ratio basal ) was defined as the average of all three baseline BRET values.Ligand-induced ΔBRET was calculated for each well as a percent over basal ([(Ratio stim -Ratio basal )/Ratio basal ] ×100).To correct for non-pharmacological effects, the average ΔBRET of vehicle control was subtracted.

Live cell confocal microscopy at the D 2long R
Confocal images were recorded with kind assistance from Manel Bosch (Universitat de Barcelona).[50] Cells were seeded in 35 mm wells containing 1.5 cover slips.48 h after transfection medium was changed to OptiMem media (Gibco) supplemented with 10 mM HEPES.Imaging was performed using a Zeiss LSM880 Laser Scanning Confocal Microscope equipped with a "Plan-Apochromat" 40x/1,3 Oil DIC M27 objective and a photomultiplier tube (PTM) detector.For excitation of GFP 2 an argon Laser with a wavelength of 488 nm was used.Fluorescence was detected within an emission window of 505-605 nm.For excitation of 20 an DPSS laser with a wavelength of 561 nm was used.Fluorescence was detected within an emission window of 569-669 nm.Image size was set to 512×512 pixels.After adjusting the focus, time-lapse images were recorded.20 was added in a final concentration of 50 nM.

Molecular Brightness
HEK-293AD cells were seeded in 8-well Ibidi μ-slides with a density of 25,000 cells per well and transfected with 2 μg hD 2long R-mNeon-Green after 24 h using JetPrime transfection reagent according to manufacturer's protocol.After further 24 h cells were washed and imaged in FRET-buffer (144 mM NaCl, 5.4 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , 10 mM HEPES) on a confocal laser scanning microscope, Leica SP8, with a white-light laser at wavelengths of 488 and 552 nm, and laser power of 5 %.All measurements were conducted with an HC PLAP CS2 40×1.3 numerical aperture (NA) oil immersion objective (Leica).Movies were acquired at 1.3 seconds per frame for 100 frames with two hybrid detectors in the range of 498 to 547 nm and 562 to 612 nm respectively, in a line sequential, counting mode.Molecular brightness ɛ and number of molecules N are calculated from the average (k) of the photon counts collected in a pixel and its variance (σ) according to the formulas ɛ = σ 2 /k-1 and N = k 2 / σ 2 .ImageJ was used to extract molecular brightness and fluorescence intensity values, number of emitters was calculated with Word Excel and obtained values were plotted and fitted with Prism v. 9.5.1.

Figure 2 .
Figure 2. Competition binding curves from radioligand competition binding experiments performed with 17 (A) or 20 (B) and the respective radioligands (cf.Table 1 footnotes) at the D 1-5 R. Graphs represent the means from N independent experiments (cf.Table 1 footnotes) each performed in triplicates.Data were analyzed by nonlinear regression and were best fitted to sigmoidal concentration-response curves.

Figure 3 .
Figure 3. ConcentrationÀ response curves (CRCs) for G o1 activation of 20 in the absence (A) and presence (B) of 1 μM dopamine in HEK293A cells transiently expressing the G o1 BRET sensor along with the wild-type D 2 R, D 3 R, or D 4 R. Graphs represent the means of three independent experiments (cf.Table 2 footnotes) each performed in duplicates.Data were analyzed by nonlinear regression and were best fitted to sigmoidal concentration-response curves.

3 a
Competition binding experiment at HEK293 cells expressing the G o1 BRET sensor with the wild type D 2 R, D 3 R or D 4 R. b Inhibition of dopamine-induced (c = 1 μM, EC 50 = 0.70 nM, Figure S5, SI) G o1 activation.c Inhibition of dopamine-induced (c = 1 μM, EC 50 = 0.21 nM, Figure S5, SI) G o1 activation.d Inhibition of dopamine-induced (c = 1 μM, EC 50 = 35 nM, Figure S5, SI) G o1 activation.Data shown are mean values � SEM of N experiments, each performed in duplicates.Data were analyzed by nonlinear regression and were best fitted to sigmoidal concentration-response curves.Competition binding curves are shown in Figure 3.
receptors and selectivity towards D 1 R and D 5 R, while 5-TAMRA and DY549-P1 were chosen as fluorescent dyes proposing suitability as fluorescent tracer for microscopy studies.In a 13and 14-step synthesis, respectively, the fluorescent ligands were successfully prepared and purified for further characterization.Radioligand binding studies revealed that both 5-TAMRA ligands, 17(pK i (D 2long R) = 8.25, pK i (D 3 R) = 8.29, pK i (D 4 R) = 7.53) and 20 (pK i (D 2long R) = 8.24, pK i (D 3 R) = 8.58, pK i (D 4 R) = 7.78), have high-affinities to D 2 -like receptors with good fluorescence properties and quantum yield.In a BRET-based G o1 heterotrimer dissociation assay, the expected antagonistic mode of action of fluorescent ligand 20 at the D 2long R, D 3 R, and D 4 R could be determined, demonstrating the suitability of 20 for whole cell applications, e. g. fluorescence microscopy.In addition to successful association experiments in confocal microscopy, molecular brightness studies confirmed ligand binding to the D 2long R in the single-digit nanomolar range.In conclusion, our study provides an interesting set of new fluorescent ligands for D 2 -like receptors, which can be applied in various applications in fluorescence microscopy in the future.

Figure 6 .
Figure 6.Association of 20 to the hD 2long R using molecular brightness analysis.Basolateral membranes of HEK-293AD cells transiently expressing hD 2long R-mNeonGreen and preincubated with indicated amount of 20 (A); calculated numbers of emitters for 5-TAMRA and mNeonGreen channels, and the corresponding linear regression fit (mean � SEM; n = 41 cells from 5 independent experiments) (B), slopes obtained from linear fits plotted against log concentrations of 20 with the corresponding non-linear fit and pK d value (mean � SEM; n = at least 18 cells from 3 independent experiments for each datapoint) (C).

Table 1 footnotes
) at the D 1-5 R. Graphs represent the means from N independent experiments (cf.Table1footnotes) each performed in triplicates.Data were analyzed by nonlinear regression and were best fitted to sigmoidal concentration-response curves.

Table 2
footnotes) each performed in duplicates.Data were analyzed by nonlinear regression and were best fitted to sigmoidal concentration-response curves.

Table 2 .
Functional data of 20 at the D 2 R, D 3 R and D 4 R.