Optical control of the β2-adrenergic receptor with opto-prop-2: A cis-active azobenzene analog of propranolol

Summary In this study, we synthesized and evaluated new photoswitchable ligands for the beta-adrenergic receptors β1-AR and β2-AR, applying an azologization strategy to the first-generation beta-blocker propranolol. The resulting compounds (Opto-prop-1, -2, -3) have good photochemical properties with high levels of light-induced trans-cis isomerization (>94%) and good thermal stability (t1/2 > 10 days) of the resulting cis-isomer in an aqueous buffer. Upon illumination with 360-nm light to PSScis, large differences in binding affinities were observed for photoswitchable compounds at β1-AR as well as β2-AR. Notably, Opto-prop-2 (VUF17062) showed one of the largest optical shifts in binding affinities at the β2-AR (587-fold, cis-active), as recorded so far for photoswitches of G protein-coupled receptors. We finally show the broad utility of Opto-prop-2 as a light-dependent competitive antagonist of the β2-AR as shown with a conformational β2-AR sensor, by the recruitment of downstream effector proteins and functional modulation of isolated adult rat cardiomyocytes.


INTRODUCTION
Modulating G protein-coupled receptor (GPCR) activity with small-molecule drugs has historically been a successful therapeutic strategy, with 33% of all small-molecule drugs targeting this class of proteins (Santos et al., 2017). Drugs are typically designed to have high and sustained occupancy of these receptors in vivo. However, for an in-depth exploration of GPCRs in (patho)physiology, it can be beneficial to have more dynamic control. Optogenetics has e.g. become a popular experimental approach for the flexible control of neurophysiology (Deisseroth, 2015) but also cardiophysiology (Entcheva and Kay, 2021). Optogenetics makes use of transgenic biological systems with light-sensitive proteins that drive a specific signaling pathway, allowing a high level of spatial and temporal control of signaling using light (Deisseroth, 2015;Entcheva and Kay, 2021).
To attenuate cellular signaling using endogenously expressed receptor proteins, recently, complementary photopharmacology strategies have emerged to design ligands with light-tunable receptor interaction (Fuchter, 2020;Hü ll et al., 2018;Lerch et al., 2016;Wijtmans et al., 2022). An effective photopharmacological strategy for the optical control of GPCRs is so-called photoswitching, in which GPCR ligands are designed to incorporate a photochromic moiety, that isomerizes upon illumination. Although a number of photochromic moieties are known, for photochromic ligands targeting family A GPCRs the azobenzene-moiety constitutes around 80% of the cases (Wijtmans et al., 2022). Azobenzene ( Figure 1) can reversibly switch from its thermodynamically stable trans-isomer to its cis-isomer under the influence of light (Beharry and Woolley, 2011). This isomerization of the azobenzene moiety constitutes a significant change in the shape and physicochemical properties. Consequently, the incorporation of an azobenzene group into GPCR ligands can often impose isomer-dependent differences in receptor interactions. Indeed, for various GPCRs, it has been shown that the use of azobenzene as a photochromic moiety in small-molecule ligands is a successful strategy for the photopharmacological control of GPCR activity (Berizzi and Goudet, 2020;Ricart-Ortega et al., 2019;Wijtmans et al., 2022).
The b 2 -adrenergic receptor (b 2 -AR) is a hallmark family A GPCR and has been extensively studied in the field of pharmacology, cell signaling, and structural biology (Weis and Kobilka, 2018). The ubiquitously expressed b 2 -AR recognizes the endogenous cell signaling molecules epinephrine and norepinephrine and as such, this GPCR is an important effector of the sympathetic nervous system. The b 2 -AR is well known to e.g. affect the contractility of smooth muscle cells in the lungs and the function of the heart. Agonists of the b 2 -AR are therefore used to target the lungs in the treatment of asthma (Barnes, 2011;Gorre and Vandekerckhove, 2010;Najafi et al., 2016b). Upon agonist activation, the b 2 -AR mediates intracellular signaling via heterotrimeric G s -proteins, which leads to increased cAMP levels and subsequent activation of protein kinase A, an important signaling cascade for, among others, smooth muscle relaxation (Barnes, 2011;Najafi et al., 2016b). The activation of the b 2 -AR also results in the recruitment of b-arrestin2, which prevents G protein coupling and facilitates receptor internalization over time (Lohse et al., 1990;Najafi et al., 2016b). As the b 2 -AR is ubiquitously expressed throughout the body (Uhlen et al., 2015), a photoswitchable b 2 -AR ligand to control b 2 -AR activation could be a valuable tool to study the spatiotemporal control of b 2 -AR function in various biological systems.
In this study, we present our work on azobenzene-derived propranolol analogs. Whereas our work was ongoing, Duran-Corbera et al. (2020) have published related b 2 -AR compounds, also based on propranolol. With our complementary strategy, we obtained a novel, thermally stable, cis-active photoswitchable b 2 -AR ligand named Opto-prop-2. All compounds efficiently isomerize to the cis-state resulting in lighttunable affinities at the b 1 -AR and the b 2 -AR. Opto-prop-2 shows one of the largest optical shifts in binding affinities for family A GPCRs (587-fold) (Wijtmans et al., 2022). We also show that our new photoswitchable compound can be used as a tool for the photopharmacological control of endogenously expressed betaadrenergic receptors. Preliminary accounts of this work have been presented at the BPS Pharmacology Winter meeting 2019 (Bosma, 2019) and Pharmacology 2020 (Leurs, 2020).

Synthesis and photochemical characterization photoswitchable ligands opto-prop-1,-2, and -3
Propranolol is a well-known antagonist at both the b 1 -AR and the b 2 -AR (Baker, 2005). In previous work, we successfully changed the naphthalene core of another GPCR ligand to the photoswitchable azobenzene moiety, resulting in only a limited loss in the target-binding affinity of the active isomer. Here we made use of a similar azologization strategy on the naphthalene core of propranolol ( Figure 1) (Broichhagen et al., 2015;Hauwert et al., 2018). Thus, compounds Opto-prop-1, Opto-prop-2 and Opto-prop-3 were designed with an azobenzene substituted on the ortho, meta, and para position of the phenyl ring, respectively, with the amino-alcohol recognition element typical for b 2 antagonists (Chan et al., 2016;Ishchenko et al., 2019). The synthetic sequence is shown in Scheme 1. First, the three regioisomeric asminophenols 1a-c were, either through the intermediacy of the TBDMS-protected (Petronijevic and Wipf, 2011;Zhang et al., 2013) aminophenols 2a,b or without protection (Steinwand et al., 2015), to the hydroxyazobenzenes 3a-c using PhNO. In a one-pot fashion based on classical means (Schwender et al., 1973(Schwender et al., , 1975, phenols 3a-c were reacted with rac-epichlorohydrin and the intermediate epoxide opened by iPrNH 2 to provide racemic photoswitchable compounds 4a-c, Opto-prop-1, -2, -3. It should be mentioned that Opto-prop-3 has been reported as a synthetic aniline precursor (Glozman et al., 1987). iScience Article Solubility in buffer was explored for all compounds using nephelometry (Bevan and Lloyd, 2000) and these studies indicate good solubility of all azobenzene photoswiches up to at least 100 mM ( Figure S1). The absorbance spectra of the azobenzenes were determined in the buffer in the dark and after illumination with either 360 or 434 nm, which are known to induce azobenzene isomerization to the cis-and trans-isomer, respectively (Hauwert et al., 2018). For all compounds, 360 nm induces a large change in absorbance spectrum compared with the non-illuminated (dark) compounds (Table 1, Figure 2A, and S1), indicating photoisomerization to the cis-isomer. The absorbance peak (l max ) of Opto-prop-1-3 ranged between 319 and 347 nm for the trans-isomers and between 423 and 429 for the cis-isomers (Table 1), as is typical for azobenzene compounds (Hauwert et al., 2018). This confirms 360 and 434 nm to be useful illumination wavelengths for trans-to cis-, and cis-to trans-switching, respectively. The photochemical conversions were determined by continuous illumination of the Opto-prop compounds with near UV-light (360 G 20 nm) as 10-mM solutions in DMSO. Based on LCMS analysis, it was observed that all compounds obtained a high level of conversion to the cis-isomer with a photostationary state (PSS) of >88 area % cis-isomer (i.e. PSS cis ; Table 1). Subsequent back-switching to PSS trans with 434 nm maintained 22-29 area % cisisomer ( Table 1). The thermal relaxation of the PSS cis state of the photoswitchable ligands was monitored over time by measuring absorbance at 254 nm. Only minor thermal relaxation to the trans-isomer could be detected at 20 C (not shown). An estimate of the relaxation rate at 20 C was determined using a method by Priimagi et al. , which was performed in Hanks' balanced salt solution (HBSS) using increased temperatures to enhance the speed of thermal relaxation. The Opto-prop ligands are all very stable as cis-isomer after illumination, with thermal relaxation half-lives of >10 days at 20 C (See Table 1 and Figure S1).
The photochemical properties were examined in more detail for Opto-prop-2 ( Figure 2). Opto-prop-2 isomerization was monitored in HBSS buffer with UV-Vis spectroscopy ( Figure 2A) after illuminating with various wavelengths, confirming that 360 and 434 nm were most suitable for switching Opto-prop-2 to its cis-state (PSS cis ) and trans-state (PSS trans ), respectively. The kinetic properties of light-induced isomer switching were further characterized using these wavelengths. Switching was monitored by 1 H NMR (Figures 2B and 2D) and LCMS ( Figures 2C and 2E) analyses in tandem. The 1 H NMR signal of the ortho C-H proton adjacent to the ether function was selected as a clearly resolved signal to quantify the level of isomerization ( Figure 2D). The time-dependent isomerization of Opto-prop-2 using 360-nm light has a fitted half-life of 4 min and provides, at PSS cis , 86 mol% cis-isomer. When consecutively isomerizing Opto-prop-2 using 434 nm light, an isomerization half-life of 3 min was observed with 20 mol% cis-isomer at PSS trans ( Figure 2B). Comparable trends were observed in LCMS analysis, with fitted isomerization half-lives of 4 and 3 min to reach PSS cis and PSS trans with fitted values of 94% and 23% cis-isomer, respectively ( Figure 2C).
Opto-prop-2 is a cis-active ligand of b 1 -and b 2 -ARs As all Opto-prop analogs could be efficiently switched to high levels of cis-isomer with good thermal stability in an aqueous buffer, the compounds were tested as trans-and PSS cis states in radioligand binding experiments. Propranolol is known to bind both the b 1 -AR and b 2 -AR with high affinity (Baker, 2005), hence the binding affinities were investigated at both adrenergic receptors (  (Table 1).
Interestingly, large differences in the K i values were observed between the trans-and the PSS cis states of the Opto-prop analogs between trans-and cis-states of the azobenzenes. At the b 1 -AR the trans-isomer of both the ortho (Opto-prop-1, Figure 3A) and para-substituted azobenzene (Opto-prop-3, Figure 3C) have a high affinity (K i <13 nM), which was markedly decreased upon switching to the cis-isomer (9and 18-fold at PSS cis , respectively), as observed by a rightward-shift of the inhibition curve. Interestingly, both isomers of Opto-prop-1 selectively bind the b 2 -AR over the b 1 -AR (R10-fold), whereas both isomers of Opto-prop-3 selectively bind the b 1 -AR over the b 2 -AR (R10-fold). The b 2 -AR selective ligand Optoprop-1, however, shows only a modest difference in binding affinity between the isomers (4-fold) at the b 2 -AR ( Figure 3D).
In contrast to Opto-prop-1 and -3, which show a reduced binding affinity upon illumination (trans-active), the meta-substituted azobenzene, Opto-prop-2, shows increased binding affinity after illumination ( Figures 3B and 3E), resulting in a 21-fold increase in binding affinity at the b 1 -AR and a striking 587-fold increase in affinity at the b 2 -AR. Next to the excellent cis-active photomodulatory properties at the b 2 -AR, the cis-isomer of Opto-prop-2 also has high selectivity in binding the b 2 -AR over the b 1 -AR (78-fold). Consequently, Opto-prop-2 represents a unique cis-active b 2 -AR antagonist photoswitch with a large window of optical modulation and long thermostability. Opto-prop-2 was therefore selected for further biological characterization.

Opto-prop-2 has photoswitchable betablocker activity
Using a conformational biosensor of b 2 -AR, we sought to investigate whether Opto-prop-2 can inactivate the agonist-bound receptors on living cells and to what extent this could be regulated with light. The conformational b 2 -AR sensor allows detection of the active receptor conformation using a bioluminescence resonance energy transfer (BRET)-readout ( Figure 4A) (Schihada et al., 2018). When HEK293A cells that stably express this conformational sensor are activated by epinephrine, a clear concentration-dependent decrease in BRET was observed (pEC 50 = 7.3 G 0.2), as is typical for agonist-mediated activation of the b 2 -AR sensor ( Figure 4B) (Schihada et al., 2018). This epinephrine-induced effect could be inhibited by trans-Opto-prop-2, but only at high concentrations (pIC 50 = 6.0 G 0.2, Figure 4C). When Opto-prop-2 was activated with light, however, the inhibitory potency was markedly increased (pIC 50 = 7.5 G 0.2), as shown by the leftward shift of the inhibition curve ( Figure 4C). The inhibitory effect of Opto-prop-2 on the agonist-induced conformational change suggests antagonism at b 2 -AR, similar to the structurally a Extracted from UV-Vis spectra (25 mM in HBSS buffer with 1% DMSO). b Measured in DMSO (10 mM) after illumination with 360 nm to PSS cis and subsequently with 434 nm to PSS trans at room temperature and defined as the percentage of area of the cis-isomer compared with combined areas of cis-and trans-isomers as detected by LCMS analysis with the corresponding isosbestic points wavelength: 278, 263, and 308 nm for Opto-prop-1-3, respectively (n = 3, SD given) c As approximated from a PSS cis sample (25 mM in HBSS buffer with 1% DMSO) with the Arrhenius method and extrapolation to the indicated temperature  iScience Article related propranolol (pIC 50 = 8.0 G 0.2). The pK B of Opto-prop-2 calculated from its inhibitory effect on the b 2 -AR conformational sensor (8.7) was similar to the affinity determined in radioligand binding experiments (Table 1). To confirm that Opto-prop-2 antagonizes b 2 -AR signaling in a light-dependent way, two enzyme complementation assays were used to measure the intracellular recruitment of mini-Ga s (black graph; Figures 4D-4F) and b-arrestin2 (black graph; Figures 4G-4I) to b 2 -AR in response to the agonist isoprenaline (Carpenter and Tate, 2016;Dijon et al., 2020;Nehmé et al., 2017). In the presence of Opto-prop-2, the concentration-response curve of isoprenaline shifts to the right, as is expected for competitive antagonism. When comparing the dextral shifts of the isoprenaline concentration-response-curves in the presence of trans-Opto-prop-2 ( Figures 4E and 4H) with the illuminated cis-Opto-prop-2 ( Figures 4F and 4I), it is clear that 360-nm light potentiates the antagonism of Opto-prop-2, as considerably higher concentrations of trans-Opto-prop-2 are needed to shift the isoprenaline concentration-response-curves. Schild analysis (Arunlakshana and Schild, 1959) (Figures 4F and 4I) of Opto-prop-2 at PSS cis indeed confirms competitive antagonism of isoprenaline-induced b 2 -AR activation, with a slope of the Schild-plot of 1.2 and 0.9 and a pA 2 of 7.6 and 7.6, for mini-Ga s ( Figure 4F) and b-arrestin2 ( Figure 4I) recruitment, respectively.
It was further explored whether the activity of the b 2 -AR could be modulated dynamically by treating cells with light ( Figure 5). Cells that overexpress a split-nanoluc luciferase biosensor system for barrestin2 recruitment to the b 2 -AR were treated with isoprenaline in the presence of Opto-prop-2. A time-dependent activation of the receptor is observed for 10 min after which cells were pulsed with light flashes of 360 nm. After each cycle with low-energy light flashes (depicted with magenta box), a drop in b 2 -AR activity is observed for the cells that were treated with Opto-prop-2.
To evaluate Opto-prop-2 in a relevant biological system with endogenous levels of the native receptor proteins, it was tested using isolated adult rat cardiomyocytes. Cardiomyocytes were paced at 2Hz, inducing contractions that were quantified by measuring the sarcomere length over time ( Figure 6A). The b 2 -AR iScience Article agonist isoprenaline (15 nM) further increased the shortening of the sarcomeres (black line) (Najafi et al., 2016a). This enhanced contractility was attenuated by co-incubating with cis-Opto-prop-2 (magenta line) but not by trans-Opto-prop-2 (teal line) at 1 mM concentrations. The normalized peak contraction is depicted for each perturbation ( Figure 6B), showing that antagonism of isoprenaline by Opto-prop-2 is fully dependent on its light-induced isomerization to PSS cis , after which a similar level of antagonism is observed for the reference beta-blocker propranolol.  (Santos et al., 2017). In this study, we have focused on the development of photoresponsive propranolol analogs. Propranolol is a hall-mark beta-blocker, that effectively antagonizes both b 1 -AR and b 2 -AR at nanomolar concentrations (Baker, 2005). Moreover, the drug is effectively used to treat cardiovascular diseases (Freemantle et al., 1999;Zhang et al., 2017). Also, beta blockers such as propranolol are used to treat infantile hemangiomas (Al-Haddad et al., 2019;Hagen et al., 2018), although data on observed side effects are questioning the route of administration and advocating locally acting drugs. Consequently, photoresponsive propranolol analogs might be a good addition to the pharmacological toolbox to study mechanistic aspects of spatiotemporal beta-adrenergic receptor signaling, but ultimately even become of therapeutic use.
In this study, we designed and synthesized three photoswitchable propranolol analogs by replacing the naphthalene moiety in propranolol with the photochromic azobenzene unit and linking this with the propranolol sidechain at either the ortho, meta, or para position ( Figure 1). All analogs can be effectively photoswitched (PSS cis R 94%) by illumination at 360 nm. Moreover, the half-lives of the thermal backswitching of the cis-isomers are all very long (>33 h, Table 1), making them suitable for most assay formats.
The Opto-prop-1, -2, and -3 ligands were characterized in detail by [ 3 H]-DHA radioligand binding studies with both the human b 1 -AR and b 2 -AR, heterologously expressed in HEK293T-cells. Remarkable differences in b-AR pharmacology were observed. For example, an 813-fold selectivity in the binding affinity to b 1 -AR over b 2 -AR was observed for the active isomer of trans-Opto-prop-3, whereas a 21-fold and 78-fold selectivity in favor of the b 2 -AR were observed for the cis-Opto-prop-1 and cis-Opto-prop-2, respectively (Table 1). Such differences in selectivity are not entirely unexpected, as it has been shown that substituting the aromatic region of the class of phenoxy-propanolamines can affect the binding selectivity at the b 1 /b 2 adrenergic receptors (Labrid et al., 1989;Ruffolo et al., 1995). iScience Article Opto-prop-1, -2, and -3 all show large differences in binding profiles between their trans-and cis-isomers at the b 1 /b 2 adrenergic receptors, albeit to a different extent and dependent on the actual adrenergic receptor subtype. At the b 1 -AR, both Opto-prop-1 and -3 are trans-active photoswitches, leading to a 9-to 18-fold loss in affinity, respectively after 360 nm illumination. In contrast, the meta-substutited analog Opto-prop-2 is a cis-active compound at b 1 -AR, but only binds at (sub)micromolar concentrations. At the b 2 -AR, both Opto-prop-1 and -2 show interesting properties, with Opto-prop-1 being a high-affinity binder with only a slight (4.3-fold) loss in affinity. Yet, the trans-isomer of Opto-prop-2 binds b 2 -AR at micromolar concentrations, but shows substantial light-enhanced binding affinity after illumination with 360 nm. The increase in affinity of almost 3 log-units for the cis-Opto-prop-2 makes cis-active Opto-prop-2 one of the photoswitches with the highest affinity and the largest optically-induced shift in affinity for family A GPCRs (Wijtmans et al., 2022). The nanomolar affinity of the cis-isomer also makes Opto-prop-2 the best b 2 -AR-specific compound of the series with a 1.9-log unit difference in binding affinity with b 1 -AR (Table 1). In general, the practical utility of trans-active ligands (such as Opto-prop-1 and Opto-prop-3) is restricted by the inability to reach high conversion to the cis-isomer upon illumination, resulting in residual effects of the more active trans-isomer (Wijtmans et al., 2022). Therefore, cis-active ligands are generally preferred, as the light-induced shift in biological activity can be much higher as clearly demonstrated by Opto-prop-2 with a 587-fold increase in b 2 -AR binding affinity upon illumination (Table 1). Whereas cis-on compounds Bode well for appreciable differences in biological activity when switching from trans to cis, dynamic optical control of GPCR activity will be critically depending on the amount of cis-isomer left at PSS trans . With classical azobenzenes, such PSS trans values are typically relatively high, as is the case for Opto-prop-2 (26% cisisomer at PSS trans ). Consequently, Opto-prop-2 maintains a substantial affinity at PSS trans (pK i = 7.5 G 0.1, compared with pK i = 8.6 G 0.3 for PSS cis ).
Based on the cis-active nature, the b 2 -AR selectivity, and the nanomolar affinity of the cis-isomer, Optoprop-2 was selected for more detailed studies at the b 2 -AR. Using an established BRET-based b 2 -AR conformational sensor (Schihada et al., 2018) and nanobit-based mini-Ga s -and barrestin2-recruitment assays for b 2 -AR (Dijon et al., 2020(Dijon et al., , 2021, both isomers of Opto-prop-2 proved to be b 2 -AR antagonist, with a clear cis-active profile, as observed in the radioligand binding studies. Moreover, using the option to illuminate cells within the microplate reader with consecutive, low-energy 360 nm flashes, also real-time Opto-prop-2 switching and antagonism of b 2 -AR function could be shown, highlighting the usefulness of Opto-prop-2 for future photopharmacology applications. As shown, Opto-prop-2 binds both b 1 -AR and b 2 -AR upon trans-cis isomerization, but with higher selectivity for the b 2 -AR than propranolol (Table 1). It was therefore explored whether the specific profile of iScience Article Opto-prop-2 can be used for light-induced inhibition of rat cardiomyocytes contractility, as these cells are known to express both b 1 -AR and b 2 -AR subtypes (Devic et al., 2001;Najafi et al., 2016b). As shown in this study, Opto-prop-2 also functions as a photoswitchable beta-blocker ( Figure 6) in cardiomyocytes that endogenously express both the b 1 and b 2 adrenergic receptors.
Recently, photoswitchable b 2 -AR antagonists similar to the current Opto-prop designs have been published by Duran-Corbera et al. (2020) In this work, which uses an approach that we had been exploring in parallel (Bosma, 2019;Leurs, 2020), an azologization strategy of propranolol included the decoration of the posterior phenyl ring of the azobenzene with a para-acetamido group. Compared with the Optoprop ligands, these para-acetamido analogues have a much faster thermal relaxation time (73-169 min at 25 C) and a small shift in peak absorbance wavelengths. These changes in photochemical properties are consistent with prior work, in which increasing the electron density of azobenzenes is shown to increase the rate of thermal relaxation and red-shifting of the absorbance spectrum (Bandara and Burdette, 2012;Kü llmer et al., 2022). The levels of cis-isomer at PSS cis and PSS trans seem to be similar upon substituting with the para-acetamido moiety. A slight improvement in the trans-content was seen at PSS trans for the substituted Opto-prop3 (from 78.5% to 86.2%), but Duran-Corbera et al. did not exploit this, owing to its relative inactivity at the b 2 -AR. Interestingly, substituting Opto-prop-1 with the p-acetamido substituent results in a slightly larger optically-induced shift in antagonist potency at b 2 -AR compared with the nonsubstituted azobenzene Opto-prop-1. With a 17-fold reduction in b 2 -AR blockade upon illumination (trans-active), this para-acetamido analog had the largest light-induced shift in antagonistic potency of their series. However, substituting Opto-prop-2 with a para-acetamido moiety was clearly not ideal for photoswitching at b 2 -AR. The observed 3.6-fold difference in b 2 -AR antagonistic potency in favor of the cis-isomer (Duran-Corbera et al., 2020) is striking in view of the 587-fold difference observed in this study for the trans-and cis-isomers of Opto-prop-2. This comparison suggests that the meta-substituted azobenzene binds quite well to the b 2 -AR binding pocket in its cis-configuration, but can not easily accommodate an additional para-acetamido substituent. Future molecular modeling, combined with site-directed mutagenesis studies and/or structural biology studies will be needed to resolve the intricate details of the binding of this class of photoswitches to the b 2 -AR. Similarly, such studies might also shed light on the observed b 1 -AR/b 2 -AR selectivity of Opto-prop-1, -2, and -3.
In conclusion, the low-affinity photoswitch trans-Opto-prop-2 (VUF17062) is shown to efficiently switch to its high-affinity cis-isomer (587-fold increase) under the influence of 360-nm light, resulting in a nanomolar affinity at the b 2 -AR and selectivity over the b 1 -AR. The high b 2 -AR affinity of cis-Opto-prop-2 and its large photo-induced shift in affinity, make Opto-prop-2 one of the exceptional useful photoswitches for family A GPCRs (Wijtmans et al., 2022). Its cis-active nature, photochemical and photopharmacological properties make Opto-prop-2 a promising chemical biology tool for the study of spatiotemporal b 2 -AR signaling in various biological contexts. Moreover, further development of Opto-prop-2 analogs toward more redshifted analogs would ultimately allow the use of wavelengths in the visible spectrum, making them more compatible with in vivo use (Wijtmans et al., 2022).

Limitations of the study
Opto-prop-2 is an effective tool to optically control the activity of the b 2 -AR. Future optimization could focus on the percentage isomer switching, in particular at PSS trans . This could further increase the dynamic switching between active and inactive b 2 -AR. Moreover, it might be possible to improve selectivity for the b 2 -AR over the b 1 -AR. Although the 78-fold difference in affinity is already substantial, larger differences would allow more stringent control of the b 2 -AR (i.e., at higher concentrations) without affecting the b 1 -AR.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

DECLARATION OF INTERESTS
The authors declare no competing interests.

EXPERIMENTAL MODEL AND SUBJECT DETAILS
In the current research the female cellines HEK293T and HEK293A and the primary cardiomyocytes isolated from 6-8 week old male Wistar rats. Cells are cultured at 37 C in a humidified atmosphere with 5% CO 2 . Further details are described in the method details.

Materials Constructs
The codon-optimized human b 1 AR in pcDEF 3 (NCBI: NP_000675.1) in pcDEF 3 was synthesized by Biomatik (the Netherlands) and human b 2 AR in pcDNA3.1+ was purchased from the cDNA resource center (USA) which corresponds to GenBank accession NM_000024.3. The mini-Gas gene sequence synthesis (Carpenter and Tate, 2016;Nehmé et al., 2017) (GeneArt) was purchased from Invitrogen (Paisley, UK) and Nanoluciferase constructs (Dixon et al., 2016) (LgBiT, SmBiT NanoBiT fragments) were obtained from Promega corporation (WI, USA). The codon-optimized sequence of human b 2 AR was synthesized in frame with SmBit spaced by a Gly/Ser linker (Ma et al., 2021) and was cloned in a pcDEF 3 vector. barres-tin2-LgBit was described previously (Ma et al., 2021). The b 2 AR Nluc/HaloTag contains the cDNA of the b 2 AR with the HaloTag in the third intracellular loop between Asp251 and Gly252 and Nluc in the C-terminus at Glu369 in pcDNA3, as described previously. (Schihada et al., 2020) The SNAP-h b 2 AR-LgBit is based on the hb 2 AR cDNA (GenBank: NM_000024.6) modified with an N-terminal SNAP tag (New England Biolabs, Hitchen UK) and the LgBiT sequence with Leu-Glu linker (Dixon et al., 2016) at the C terminus, in pcDNA3.1(+). barrestin2-SmBit and mini Ga s -SmBit plasmid DNA contain the cDNA of barrestin2 (GeneBank: NM_004313) or mini Ga s with an N terminal SmBiT sequence followed by 5 Ser/Gly linker, in pcDNA3.1 zeo (+).

Compound use in biological experiments
The HaloTag dye NanoBRET 618 and stock solution furimazine (Nano-Glo) were obtained from Promega. Isoprenaline was obtained from Tocris (UK) and (À)epinephrine was purchased from Sigma Aldrich (USA). Propranolol was obtained from an in-house library and was confirmed to be of analytical purity. Opto-prop-1 (VUF17061), Opto-prop-2 (VUF17062) and Opto-prop-3 (VUF25417) were synthesized and compounds were analyzed with respect to their identity and purity as described in the supplemental information (Data S1).

Nephelometry measurement
In transparent flat-bottom 96-well plates, the azobenzene propranolol analogs, under dark conditions or preilluminated with 360 nm light to PPS cis , were placed at different concentrations in triplicate (10 À4 M, 10 À4.5 M, 10 À5 M, 10 À5.5 M, 10 À6 M, 10 À6.5 M, 10 À7 M, 10 À7.5 M and a blank) in aqueous buffer with 1% DMSO at least 1 h before the measurement. A Kaolin dispersion was used as a positive control. (Roessler and Brewer, 1967) Nephelometry measurements were performed with a NEPHELO star Plus (BMG Labtech, Germany) with the following settings: laser intensity 80%, beam focus 2.0 mm, and Orbital shaking of 10 s at 200 rpm before data acquisition. Results were analyzed using GraphPad Prism 8 software, plotting all available data points and plotting mean and SD values in a line chart compared to kaolin control. The linear fit (R 2 ) of the kaolin control was above 0.99 in all cases.

Photochemistry procedures
UV-Vis spectra were obtained using a Thermo-scientific Evolution 201 PC spectrophotometer. Illumination was executed using a Sutter instruments Lambda LS with a 300 Watt full-spectrum lamp connected to a Sutter instruments Lambda 10-3 optical filter changer equipped with 360 G 20 nm, 400 G 5 nm, 434 G 9 nm and 460 G 5 nm filters. The light intensity is 0.93 mW/mm 2 using the 360 G 20 nm filter, 0.22 mW/mm 2 using the 400 G 5 nm filter, 0.79 mW/mm 2 for the 434 G 9 nm filter and 0.26 mW/mm 2 for the 460 G 5 nm filter as measured using a Thorlabs PM16-401 power meter.

OPEN ACCESS
iScience 25, 104882, September 16, 2022 15 iScience Article iScience Article to let them attach to the coated dish. Afterwards, non-attached cells were removed by washing cells with pre-heated Tyrode solution (137 mM NaCl, 5.4 mM KCl, 3 mM sodium pyruvate, 5 mM HEPES, 0.57 mM MgCl 2 , 0.33 mM NaH 2 PO 4 , 1.0 mM CaCl 2 and 5.6 mM glucose, pH 7.4 at 37 C). Contractility measurements were performed at 37 C using the Multi-Cell system (CytoCypher, the Netherlands). Cells were treated to vehicle (DMSO), 15 nM isoproterenol alone or in combination with 1 mM of propranolol or Opto-prop-2 (either with or without prior illumination). The dish was field-stimulated at 2 Hz, 25 V and a 4 ms pulse duration.

QUANTIFICATION AND STATISTICAL ANALYSIS
Photochemical characterization NMR, LC and UV-Vis spectra were plotted in GraphPad Prism 8 and were visualized by fitting the data to a second order polynomial with a smooth function, using 4 neighboring data points. The aromatic region of the NMR-spectrum is depicted between ppm 6.5 and 7.5 ( Figure 2D), with 6,545 intermittent datapoints. For the time-resolved isomerization experiments, integration of the peak areas was performed with Shimadzu and MestreNova software for LC and NMR integration, respectively. The normalized peak areas over time were fitted by non-linear regression in GraphPad Prism 8 using a one-phase exponential model which yields the t 1/2 value of the isomerization reaction and the extrapolated PSS (i.e. the asymptote).

Radioligand binding experiments
IC 50 values were determined by fitting the concentration-dependent displacement of [ 3 H]DHA by unlabeled ligands, with a three-parameter sigmoidal model in GraphPad Prism. The K i of all unlabeled ligands were calculated according to the Cheng-Prusoff equation (Cheng and Prusoff, 1973) using the radioligand binding affinity values determined by saturation binding. Competition binding graphs represent the pooled data of R3 experiments, normalized to the fitted top and bottom of the reference sigmoidal displacement curve of unlabeled DHA.

Conformational BRET sensor
Raw BRET ratios were defined as acceptor emission/donor emission. The three BRET ratios prior epinephrine addition were averaged and defined as BRET basal . To quantify epinephrine-induced BRET changes, DBRET were calculated for each well and time point as a percent over basal ([(BRET stim À BRET basal )/ BRET basal ] 3 100). Subsequently, the average DBRET of vehicle-treated control wells. The epinephrine concentration response curve was generated based on the vehicle-corrected DBRET measured 15 min after ligand addition.
Antagonist pK B values were calculated using the agonist potency and concentration together with the antagonist IC 50 values, using the Cheng-Prusoff equation. (Cheng and Prusoff, 1973;Leff and Dougall, 1993) Nanobit complementation experiments Four parameter concentration-response curves were fitted to individual experiments performed in duplicate, normalized to the 10 mM isoprenaline (100%) and vehicle control (0%) response at 31 min post agonist addition. From the isoprenaline potencies (EC 50 ) in the absence and presence of Opto-prop-2, concentration ratios (CR) were calculated. The antagonist affinity (K D ) was then estimated by a Schild plot of log [CR-1] against log [B] (the antagonist concentration) using the relationship: Log ½CR À 1 = Log ½B À Log K D :

Intact cardiomyocyte isolation and measurements
Changes in sarcomere length were recorded with a high-speed camera and Ionoptix software (Ionoptix, Westwood MA, USA). The contractility profiles were analyzed with the automated, batch analysis software 'Transient Analysis Tools' (CytoCypher, Amsterdam, NL). Fractional shortening of cardiomyocytes was compared between treatment conditions using one-way ANOVA with Dunnett's multiple comparisons test.