Quantitative Super-Resolution Imaging for the Analysis of GPCR Oligomerization

G-protein coupled receptors (GPCRs) are known to form homo- and hetero- oligomers which are considered critical to modulate their function. However, studying the existence and functional implication of these complexes is not straightforward as controversial results are obtained depending on the method of analysis employed. Here, we use a quantitative single molecule super-resolution imaging technique named qPAINT to quantify complex formation within an example GPCR. qPAINT, based upon DNA-PAINT, takes advantage of the binding kinetics between fluorescently labelled DNA imager strands to complementary DNA docking strands coupled to protein targeting antibodies to quantify the protein copy number in nanoscale dimensions. We demonstrate qPAINT analysis via a novel pipeline to study the oligomerization of the purinergic receptor Y2 (P2Y2), a rhodopsin-like GPCR, highly expressed in the pancreatic cancer cell line AsPC-1, under control, agonistic and antagonistic conditions. Results reveal that whilst the density of P2Y2 receptors remained unchanged, antagonistic conditions displayed reduced percentage of oligomers, and smaller numbers of receptors in complexes. Yet, the oligomeric state of the receptors was not affected by agonist treatment, in line with previous reports. Understanding P2Y2 oligomerization under agonistic and antagonistic conditions will contribute to unravelling P2Y2 mechanistic action and therapeutic targeting.


Introduction
G-protein coupled receptors (GPCRs) are the largest family of cell surface receptors in eukaryotic cells. These seven-transmembrane receptors have influence in physiological events such as cell to cell communication, immune responses, nerve transmission and even hunger and sleep regulation [1][2][3]. The role of GPCRs in diseases such as rheumatoid arthritis, heart disease, cancer, obesity, and neurodegenerative disorders accentuates the need to investigate this family of receptors further. More than a third of all drugs approved by the FDA target GPCRs [4] but often such drugs have a variety of poorly understood mechanisms, as a recent example surrounding opioid receptor agonists illustrates [5][6][7][8]. Understanding these mechanisms, and their influence on GPCR activation, is paramount to improving GPCR drug development.
GPCRs have been shown to form dimers and/or oligomers, where the receptor is present in groups of two or more receptors of the same or different kind. There are several reports of oligomers affecting ligand-binding pharmacology, function, trafficking, and internalisation compared to their monomeric form [9,10]. Several experimental techniques, ranging from traditional biochemical approaches to biophysical ensemble methods based on resonance energy transfer (RET), such as fluorescence resonance energy transfer (FRET) and bioluminescence resonance energy transfer (BRET), have been key to observing the formation of dimers and oligomers [11,12]. However, one of the main issues with these techniques is that they cannot provide information on the size of the oligomeric complexes, nor on the location of the GPCRs in the cell. Over the years, different optical microscopy approaches, based on molecular brightness analysis, have been proposed to tackle some of these shortcomings [13][14][15]. However, they provide average oligomerization information, and it is not possible to extract the precise stoichiometry of specific molecular complexes. A suitable technique to provide information on the oligomeric state of GPCRs and their dynamics down to the single-molecule level is single-molecule tracking [16][17][18]. Yet, this typically requires receptor densities that are orders of magnitude lower than those occurring in natural settings. The relevance of this shortcoming is particularly important when investigating GPCR oligomerization in certain cancer cell lines, where the receptors are overexpressed.
Here, we present a novel single-molecule based approach to study the oligomerization of a rhodopsin-like GPCR, the purinergic receptor Y2 (P2Y 2 ), in a pancreatic cancer cell line with high expression of the receptor [19]. Our approach is based on the most recently developed single-molecule localization microscopy (SMLM) method named DNA-PAINT [20] (a variation of point accumulation for imaging in nanoscale topography) in combination with an imaging analysis pipeline suitable for protein quantification. DNA-PAINT is an SMLM technique that relies on the repetitive binding between two short complementary single-stranded DNAs, one conjugated to a fluorescent dye (imager strand) and the other chemically coupled to either a primary or secondary antibody targeting the protein of interest (docking strand). These short-lived transient events (millisecond to second range) create the necessary blinking required for SMLM, allowing the localization of the position of single molecules with nanometer precision. Over the course of the experiment, these cumulative DNA binding events form a cluster of single molecule localizations within the true position of the biological target, as illustrated in Scheme 1. Depending on the DNA pair's binding kinetics in particular the association rate, k on , and the imager DNA strand concentration, c i , each docking strand is visited by an imager at a frequency given by (k on *c i ), which corresponds to the inverse of the dark time for a single docking strand (i.e., length of time that a docking strand is not bound to an imager strand between binding events) [21]. The frequency of the imagers binding to their docking strand scales linearly with the number of docking strands, and this is the principle of the quantitative analysis known as qPAINT (Scheme 1). The predictable binding kinetics between imager and docking strands make qPAINT suitable to accurately correlate the frequency of singlemolecule events with the underlying number of labelled molecular targets [21], overcoming 'overcounting' artifacts observed with other SMLM techniques.
To demonstrate the potential of qPAINT to study GPCRs oligomerization status, we investigated the nanoscale distribution of the P2Y 2 receptor, a member of the d subgroup of the family A of GPCRs, in the cancer cell line AsPC-1, as it endogenously expresses high levels of this receptor and single cells are easily imaged due to its low levels of cell grouping [19]. P2Y 2 has been related to immune regulation, bone mineralisation, intraocular pressure, HIV-1 infection, and cancer metastasis and proliferation [22][23][24][25]. P2Y 2 has recently gained traction due to its role in several cancers such as breast, head and neck, prostate, and pancreatic [19,[25][26][27]. This receptor, which can interact with adenosine triphosphate (ATP) and uridine triphosphate (UTP), is known to homo-dimerise and homooligomerise [28,29]. When applied to P2Y 2 , qPAINT revealed the molecular densities and nano-and micro-scale spatial arrangements of GPCRs down to the single molecule level, as well as their modulation in response to agonist and antagonist treatment. Understanding the natural oligomerisation state of P2Y 2 and the effect of agonists and antagonists could inform the mechanist action of P2Y 2 and aid in targeting it therapeutically. as well as their modulation in response to agonist and antagonist treatment. Understanding the natural oligomerisation state of P2Y2 and the effect of agonists and antagonists could inform the mechanist action of P2Y2 and aid in targeting it therapeutically. Scheme 1. Overview of qPAINT analysis pipeline (created with BioRender.com, BioRender, Toronto YTO, Canada).

Cell Fixation and Immunofluorescence Staining for Antibody Validation
AsPC-1 and PS-1 cells were seeded in coverslips placed in a 6 well-plate (Corning, Corning, New York NY, USA) at a seeding density of 200,000 cells per well. Cells were fixed in 4% paraformaldehyde (LifeTech, Carlsbad Ca, USA) for 30 min and washed 3× in PBS. Cells were permeabilised with 0.1% Triton X-100 (Avantor, Radnor Township PA, USA) and incubated for 10 min. The coverslips were washed 3× with PBS. Blocking was performed using 5% bovine serum albumin (Merck, Kenilworth NJ, USA) in PBS for 1 h. To image P2Y2 receptors, the anti-P2Y2 receptor antibody (APR-010, Alomone labs, Jerusalem, Israel), which binds to the third intracellular loop between the fifth and sixth transmembrane domain of P2Y2, was used for immunostaining. The P2Y2 antibody was diluted in blocking solution at 1:200 as previously reported [32]. Cells were incubated overnight at 4 °C with the antibody dilution followed by 3× PBS washes. Then, the secondary antibody Alexa Fluor 546 goat anti-rabbit (Invitrogen, Waltham MA, USA) was added (1:1000) and left incubating for 1 h. For nuclear staining, 4′,6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich, St. Louis MO, USA) dilution (1:1000) was added and incubated for 10 min.

Cell Fixation and Immunofluorescence Staining for Antibody Validation
AsPC-1 and PS-1 cells were seeded in coverslips placed in a 6 well-plate (Corning, Corning, NY, USA) at a seeding density of 200,000 cells per well. Cells were fixed in 4% paraformaldehyde (LifeTech, Carlsbad, CA, USA) for 30 min and washed 3× in PBS. Cells were permeabilised with 0.1% Triton X-100 (Avantor, Radnor Township, PA, USA) and incubated for 10 min. The coverslips were washed 3× with PBS. Blocking was performed using 5% bovine serum albumin (Merck, Kenilworth, NJ, USA) in PBS for 1 h. To image P2Y 2 receptors, the anti-P2Y 2 receptor antibody (APR-010, Alomone labs, Jerusalem, Israel), which binds to the third intracellular loop between the fifth and sixth transmembrane domain of P2Y 2, was used for immunostaining. The P2Y 2 antibody was diluted in blocking solution at 1:200 as previously reported [32]. Cells were incubated overnight at 4 • C with the antibody dilution followed by 3× PBS washes. Then, the secondary antibody Alexa Fluor 546 goat anti-rabbit (Invitrogen, Waltham, MA, USA) was added (1:1000) and left incubating for 1 h. For nuclear staining, 4 ,6-diamidino-2phenylindole (DAPI, Sigma-Aldrich, St. Louis, MO, USA) dilution (1:1000) was added and incubated for 10 min. Slides were mounted with coverslips using Mowiol (Calbiochem, San Diego, CA, USA). Fluorescence images were taken with a Zeiss LSM 710 confocal microscope (Zeiss, Oberkochen, Germany).

DNA-Antibody Coupling Reaction
DNA labelling of anti-P2Y 2 receptor antibody (APR-010, Alomone labs, Jerusalem, Israel) was performed via maleimidePEG2-succinimidyl ester coupling reaction [20,33]. Briefly, 13 µL of 1 mM thiolated DNA (5 -Thiol-AAACCACCACCACCA-3 , Eurofins, Ebersberg, Germany) was reduced by mixing it on a shaker with 30 µL of a freshly prepared 250 mM DDT (Thermo Fisher Scientific, Waltham, MA, USA) solution for 2 hrs. 30 min after the reduction of the thiol-DNA started, 175 µL of 0.8 mg/mL primary antibody was incubated with 0.9 µL of 23.5 mM maleimide-PEG2-succinimidyl ester cross-linker solution (Sigma-Aldrich, St. Louis, MO, USA) on a shaker for 90 min at 4 • C in the dark. To remove excess DDT and cross-linker, both reactions were purified by spin filtration using a Microspin Illustra G-25 columns (GE Healthcare, Chicago, IL, USA) and a Zeba spin desalting column (7K MWCO, Thermo Fisher Scientific, Waltham, MA, USA), respectively. The resultant products were mixed and incubated on a shaker overnight at 4 • C in the dark. Finally, excess DNA was removed via Amicon spin filtration (100K, Merck, Kenilworth, NJ, USA) and antibody-DNA concentration was measured using a NanoDrop One spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

DNA-PAINT Image Reconstruction and Cluster Analysis
Images were processed and reconstructed using the Picasso software (Version 0.3.3) [20]. Briefly, single molecule events were identified and localised from the raw fluorescent DNA-PAINT imaging experiments using the 'Localize' module of Picasso. Subsequently, images were drift corrected with the 'Render' module of Picasso using a stepwise protocol. First, via an image sub-stack cross correlation analysis and then using gold nanoparticles as fiducial markers. Localizations with uncertainties greater than 13 nm were removed while no merging was performed for molecules re-appearing in subsequent frames. Finally, regions of interest (ROIs) of~4 by 4 µm 2 were selected within the 'Render' module of Picasso and analysed using the implementation of DBSCAN from PALMsiever [34] in MATLAB (2021a) [35]. DBSCAN is a data clustering algorithm that detects clusters of localizations by looking for the minimum number of points ('minPts') within a circle with radius epsilon ('eps'). For 'eps', we used the localisation precision of our DNA-PAINT images as determined via the nearest-neighbour based analysis, which was ca. 10 nm for all the images. For 'minPts' we chose a parameter in accordance with the binding frequency of the imager strand and the number of recorded frames, in our case this value was set to 10 localisations.

qPAINT Analysis
For qPAINT analysis we used a custom-written MATLAB (2021a) [35] code that analyses the fluorescence time series of each detected localization cluster to estimate the number of P2Y 2 receptors for each cluster as described previously [21,33]. In short, localizations corresponding to the same cluster were grouped and their time stamps (frame number) used to reconstruct the sequence of dark times per cluster as continuous frame times that did not contain an event. All the dark times per cluster were pooled and used to obtain a normalised cumulative histogram of the dark times which was then fitted with the following exponential function: 1 -exp(t/τ d ) to estimate the dark time, τ d , per cluster. The inverse of the dark time was calculated for each cluster and stored as the qPAINT index of the cluster (q i ).
To estimate the number of P2Y 2 receptors per cluster, a calibration was performed with the DNA-PAINT data of all cells and verified by analysing DNA-PAINT images of DNA labelled anti-P2Y 2 receptor antibody deposited on glass slides. For each DNA-PAINT data series a histogram of cluster qPAINT indices for small clusters (i.e., clusters with a maximum point distance of 150 nm) was obtained and fitted with a multi-peak Gaussian that exhibited peaks at multiples of a qPAINT index of 0.012 Hz. This calibration value, corresponding to the qPAINT index for one binding site, q i1 , was used to estimate the number of P2Y 2 receptors per cluster as the ratio of the qPAINT index of the cluster, q i and q i1. To recover a likely distribution of P2Y 2 receptors in each cluster of localizations, we used k-means clustering, where k corresponds to the protein copy number per cluster.

Statistical Analysis
For DNA-PAINT imaging, a minimum of twenty-five~4 by 4 µm 2 regions obtained from 7-9 AsPC-1 cells were analysed per condition (control, agonist, antagonist). Statistical analysis was performed via R (Version 4.0.3, The R Foundation, Vienna, Austria) using the rstatix package [36,37]. Distribution of data points and their variance were determined. Groups of three independent conditions were compared using non-parametric pairwise Wilcoxon rank sum tests using the Holm correction method for multiple hypothesis testing. Differences were statistically significant when adjusted p < 0.05. (n.s., p > 0.05; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001). Plots were created in R using the packages ggplot2, ggpubr, tidyverse, and ggprism.

Super-Resolution Imaging of P2Y 2 Receptors in AsPC-1 Cells Using DNA-PAINT
To unravel the molecular organization of P2Y 2 receptors in and near the plasma membrane of AsPC-1 cells, we used DNA-PAINT imaging under total internal reflection (TIR) excitation (see Figure S1 for a schematic representation of the optical set-up). Figure 1a shows a representative super-resolution image of P2Y 2 obtained via DNA-PAINT. TIR excitation allows investigation of samples at or near the cell membrane by optically sectioning light illumination to only the most superficial~100 nm of the sample. This is extremely beneficial in the study of GPCRs located at the plasma membrane as the receptors are typically not only at the cell membrane, but also at intracellular sites such as endosomes, endoplasmic reticulum, and the Golgi complex [38] and TIRF imaging minimizes their intracellular visualisation.  Next, we aimed to demonstrate the applicability of qPAINT analysis to quantitatively P2Y 2 receptors were labelled with a primary antibody which we validated using the immortalised human pancreatic stellate cell line PS-1 as it expresses traces amounts of the protein of interest (see Figure S2 for negative control results) [19,39]. For DNA-PAINT imaging, the anti-P2Y 2 receptor antibody was chemically coupled to an optimised docking strand sequence design to increase imaging speed. The docking strand features a repetitive (ACC)n sequence motif that provides 3× overlapping binding sites for the imager strand (Figure 1a, inset) [40]. The benefit of such docking sequence is not only the increase of imaging speed, but also the possibility to use relatively low imager strand concentrations achieving a high signal-to-noise ratio and single-molecule localisation precision. In our experiments we achieved an overall localisation precision of~10 nm using a 1 nM concentration of imager strand fluorescently labelled with ATTO643, as determined by the nearest-neighbour-based analysis ( Figure S3).

qPAINT Calibration
To investigate the nanoscale distribution and oligomerization state of P2Y 2 receptors in AsPC-1 cells, we subjected the DNA-PAINT data to qPAINT analysis. qPAINT allows quantification of the exact number of antibody-labelled P2Y 2 receptors within a cluster of single molecule localisations by taking advantage of the first order binding kinetics between individual imager and docking strands. Specifically, the method uses the average dark time, τ d , of a cluster of single molecule localizations to determine how many copies of the protein reside within that cluster of points, N, via the following relationship where k on is the binding rate of imager to docking strand, c i is the concentration of the imager strand, and τ d,1 denotes the dark time for the case of a single protein and is equal to (k on c i ) −1 . By examination of Equation (1), it is straightforward to see that every additional protein residing in the same cluster of localizations will lead to a proportionally shorter τ d , as illustrated in Scheme 1. To calculate the dark times associated to each cluster, we developed an automated data processing pipeline based on DBSCAN (density-based spatial clustering of application with noise) [41]. In DNA-PAINT, a DNA-coupled antibody is localised several times, rendering a cluster of localizations around the true position of the labelled proteins, whereas non-specific binding events are detected as non-clustered localizations. Therefore, we first used the clustering algorithm DBSCAN to automatically identify clusters of localizations in all the data sets, as illustrated in Figure 1b. For each detected cluster, we then calculated the average dark time between binding events by fitting cumulative histograms of the dark time durations, as detailed in the Methods section. Figure 1b presents two typical single molecule on/off time series sequences for the highlighted cluster of localizations containing either one or two docking sites, respectively.
To provide a measurement that is directly proportional to the copies of labelled proteins per cluster, N, we used the inverse of the measured dark time, a term known as the qPAINT index, q i [42]. Figure 1c shows a histogram of the qPAINT indexes obtained from the DNA-PAINT data acquired in AsPC-1 cells. This was achieved by selecting very small clusters in the biological data set, based on their geometrical dimension, such that they visually contain one, two, or several puncta. The qPAINT index histogram of P2Y 2 receptors can be fitted with a multi-Gaussian function with peaks located at multiples of a qPAINT index value of q i1 0.012 Hz. This 'quantal' behaviour is characteristic of detecting one, two, or three units of DNA docking strands in the same cluster of single molecule localisations. Similar results were obtained when examining the qPAINT index histogram of a dedicated calibration sample corresponding to DNA-coupled primary antibodies deposited on a glass coverslip and imaged via DNA-PAINT with the same experimental conditions as the biological experiments (Figure 1d). In this case, the qPAINT index histogram contains the information of all the clusters detected in the calibration sample and exhibits a prominent peak at a small qPAINT index value of q i1 0.011 Hz and a secondary peak located at nearly double of this value. The consistency between the qPAINT index value for a single docking strand obtained in the calibration sample and the internal calibration of the P2Y 2 data are a strong indication of the accuracy of the q i1 value 0.012 Hz obtained in the biological samples. This value was thus used to estimate the number of P2Y 2 receptors in DNA-PAINT images as the ratio between q i /q i1 . This analysis also confirmed that it is possible to differentiate between monomers, dimers, and higher order clusters of antibody-labelled P2Y 2 receptors using qPAINT analysis.

Comparison of P2Y 2 Receptor Dimerization and Cluster Formation upon Agonist and Antagonist Treatment of AsPC-1 Cells
Next, we aimed to demonstrate the applicability of qPAINT analysis to quantitatively determine the agonist and antagonist dependent oligomerization state of the P2Y 2 receptor in single AsPC-1 cells. Figure 2 illustrates our automated data processing procedure. First, we acquired DNA-PAINT images of the P2Y 2 receptor in control AsPC-1 cells and cells treated with agonist (ATP) and antagonist (AR-C 118925XX). Examples of these super-resolution images are presented in Figure 2a. To statistically quantify the spatial organisation of the GPCR in different conditions, we used DBSCAN to detect clusters of single-molecule localisations in at least twenty five~4 by 4 µm 2 randomly selected regions of interest (ROIs) obtained from 7-9 cells per condition (i.e., control, agonist, and antagonist treated cells). After identifying clusters of localisations in each ROI (colour coded in Figure 2b), we determined the dark time associated with each cluster and estimated the copy number of P2Y 2 receptors per cluster using the calibration value obtained as described in Section 3.2.1. We used this information to quantify the density of antibody labelled receptors, the percentage of receptors organised into clusters of proteins (clusters defined as an assembly containing more than 5 copies of the receptor), the percentage of receptors forming dimers, and the percentage of non-clustered P2Y 2 receptors, defined as monomers. Furthermore, we then used k-means clustering (a distance-based algorithm) to partition the clusters of single molecule localisations into k clusters, where k is given by the copy number of receptors per cluster. This pipeline, previously introduced by Simoncelli et al. [33], allowed us to recover an accurate quantitative map of the nanoscale distribution of labelled P2Y 2 receptors in AsPC-1 cells (Figure 2c). Using k-means clustering, we were also able to calculate that the mean distance between two P2Y 2 receptors in the clusters identified as dimers is 25 ± 10 nm ( Figure S4). Considering both the physical size of the labelling antibodies (13.7 nm) [43] and the localization error (10 nm, Figure S3), the furthest apart two adjacent antibodies could be is ca. 27 nm ± 10 nm, which is within the range of the measured dimer distance, suggesting that a distance of 25 ± 10 nm is likely to correspond to the detection of true chemical dimerization.
Using this approach, we found that the density of labelled P2Y 2 receptors in the plasma membrane (and/or the membrane-proximal area) does not change upon agonist or antagonist treatment (Figure 3a). Quantification of the extent of dimerization and clustering of P2Y 2 receptors also revealed that the percentage of receptors in monomers, dimers, and clusters are not significantly different between agonist treated and control conditions (Figure 3b-d). Similar trends were observed by Kotevic et al. [29] using FRET measurements of HEK 293 cells co-expressing P2Y 2 -CFP and -YFP, where no significant change in the FRET signal was observed upon exposure to the agonist, in their case UTP. Unlike the comparison between control and agonist treated cells, there is a marked decreased (ca. 50%) in the percentage of P2Y 2 receptors forming clusters in antagonist treated cells with respect to the control conditions (Figure 3d). The mean P2Y 2 receptor copy number per cluster was also highly significantly different between control and antagonist and agonist and antagonist treated cells but did not change significantly between agonist and control conditions (Figure 3e). Concurrent with the reduction of the percentage of molecules forming clusters and number of molecules in clusters in antagonist treated cells, this analysis also detected an increase in the percentage of P2Y 2 receptors forming dimers and in the monomeric state (Figure 3b,c). Together these data indicate that the antagonist prevents the formation of higher order clusters and reduces their size. Using this approach, we found that the density of labelled P2Y2 receptors in the plasma membrane (and/or the membrane-proximal area) does not change upon agonist or antagonist treatment (Figure 3a). Quantification of the extent of dimerization and clustering of P2Y2 receptors also revealed that the percentage of receptors in monomers, dimers, and clusters are not significantly different between agonist treated and control conditions (Figure 3b-d). Similar trends were observed by Kotevic et al. [29] using FRET measurements of HEK 293 cells co-expressing P2Y2-CFP and -YFP, where no significant change in the FRET signal was observed upon exposure to the agonist, in their case UTP. Unlike the comparison between control and agonist treated cells, there is a marked decreased (ca. 50%) in the percentage of P2Y2 receptors forming clusters in antagonist treated cells with respect to the control conditions (Figure 3d). The mean P2Y2 receptor copy number per cluster was also highly significantly different between control and antagonist and agonist and antagonist treated cells but did not change significantly between agonist and control conditions (Figure 3e). Concurrent with the reduction of the percentage of molecules forming clusters and number of molecules in clusters in antagonist treated cells, this analysis also detected an increase in the percentage of P2Y2 receptors forming dimers and in the monomeric state (Figure 3b-c). Together these data indicate that the antagonist prevents the formation of higher order clusters and reduces their size.

Discussion
Over the years, multiple optical microscopy techniques have been applied to the study of GPCR oligomerization, with one of the first single molecule imaging studies done by Kasai et al. [44]. Subsequently, both Spatial Intensity Distribution Analysis (SpIDA) [15,45] and molecular brightness approaches [13,14] were also developed and applied to study a variety of GPCRs. In recent years, single-molecule tracking and FRET imaging have also been applied to identify key factors in the regulation of GPCRs dynamic interactions in living cells [16,46]. While these techniques have been paramount to investigate the oligomeric organisation of GPCRs and the dynamic interactions that control GPCRs signalling, many challenges still remain; including the limitation of applicability to endogenous settings (i.e., many of these methods require labelled receptors or fusion proteins) and the possibility of multi-colour imaging to simultaneously study different types of hetero-oligomers. Most recently, dSTORM (direct stochastic optical reconstruction microscopy), a single-molecule super-resolution microscopy method, was employed to investigate the nanoscale organisation of a GPCR in presynaptic active zones [47], presenting a

Discussion
Over the years, multiple optical microscopy techniques have been applied to the study of GPCR oligomerization, with one of the first single molecule imaging studies done by Kasai et al. [44]. Subsequently, both Spatial Intensity Distribution Analysis (SpIDA) [15,45] and molecular brightness approaches [13,14] were also developed and applied to study a variety of GPCRs. In recent years, single-molecule tracking and FRET imaging have also been applied to identify key factors in the regulation of GPCRs dynamic interactions in living cells [16,46]. While these techniques have been paramount to investigate the oligomeric organisation of GPCRs and the dynamic interactions that control GPCRs signalling, many challenges still remain; including the limitation of applicability to endogenous settings (i.e., many of these methods require labelled receptors or fusion proteins) and the possibility of multi-colour imaging to simultaneously study different types of hetero-oligomers. Most recently, dSTORM (direct stochastic optical reconstruction microscopy), a single-molecule super-resolution microscopy method, was employed to investigate the nanoscale organisation of a GPCR in presynaptic active zones [47], presenting a step-forward in visualising the organization of these receptors in endogenous conditions with nanoscale resolution. However, this was possible at the cost of increased experimental complexity, as the approach required transiently transfecting cells under controlled conditions to accurately quantify the number of receptors per cluster of single-molecule localizations.
As a novel method for quantitative analysis and nano-scale visualisation of the oligomeric state of GPCRs in single cells, here we employed a quantitative single molecule based super-resolution imaging technique based on DNA-PAINT, named qPAINT. Our approach allows the study of GPCR oligomerization in an endogenous setting, while avoiding the difficulties experienced with other SMLM techniques, such as dSTORM or PALM, to quantify the protein copy number [47,48]. During the last four years, examples of the use of qPAINT to quantify protein clustering in different biological systems have started to emerge [33,42,49]. This work is the first to employ qPAINT towards the study of GPCRs, a technique that can certainly help resolve contentious issues associated with GPCR oligomerization in different settings. Furthermore, due to the comparative advantages of DNA-PAINT over other SMLM techniques-it can routinely deliver lateral resolution in the 5-10 nm range and it allows multiplexing with a single laser source-it is straightforward to expand the applicability of qPAINT to detect and quantify GPCR heterooligomers. We note, however, that like any other technique that relies on antibody labelling, its applicability will depend on the availability of validated antibodies and that the final accuracy towards the absolute protein copy number is affected by the overall antibodytarget binding efficiency, which can lead to potential undercounting. Still, by keeping data acquisition and analysis the same it is possible to draw conclusions about changes in the nanoscale distribution of the receptors in the different conditions. Furthermore, the presented approach is not suitable for live-cell imaging.
By leveraging the capabilities of qPAINT to its maximum potential, our study provides a detailed quantitative characterization of the density, spatial organization, and stoichiometry of the antibody labelled purinergic receptor Y2 oligomers in the pancreatic cancer cell line AsPC-1. Specifically, our data indicate that P2Y 2 receptors are highly expressed in AsPC-1 cells with an average density of ca. 40 receptors per µm 2 . We find that P2Y 2 receptors are mainly organised in nanodomains containing three or more sub-units, with only 20% of the receptors forming dimers and 10% distributed as monomers. Our results also show that while the oligomerization status of P2Y 2 receptors does not change upon agonist treatment, there is a marked reduction of the percentage of P2Y 2 receptors forming oligomers in antagonistic conditions. Our results are in line with current models for the P2Y 2 receptor, which imply that homo-oligomeric assemblies of P2Y 2 receptors are required for receptor internalisation and the effect of the antagonist prevents the formation of these complexes. Similar antagonist effects have been observed for CXCR4 and dopamine D3 GPCR receptors [50,51]. The fact that we did not observe significant difference in the oligomerisation of P2Y 2 receptors in control and agonistic conditions suggest that oligomeric assemblies could be part and parcel of the natural activation of the receptor.
In conclusion, we demonstrated the accuracy and ease of implementation of qPAINT to quantitatively characterise the oligomeric state of GPCRs, alongside achieving single receptor visualisation. qPAINT analysis of the oligomeric state of P2Y 2 revealed consistency with currently proposed models for this receptor [29,50,51].

Conflicts of Interest:
The authors declare no conflict of interest.