Mechanistic Insights into Light-Driven Allosteric Control of GPCR Biological Activity

Abstract

G protein-coupled receptors (GPCR), including the metabotrobic glutamate 5 receptor (mGlu₅), are important therapeutic targets and the development of allosteric ligands for targeting GPCRs has become a desirable approach toward modulating receptor activity. Traditional pharmacological approaches toward modulating GPCR activity are still limited since precise spatiotemporal control of a ligand is lost as soon as it is administered. Photopharmacology proposes the use of photoswitchable ligands to overcome this limitation, since their activity can be reversibly controlled by light with high precision. As this is still a growing field, our understanding of the molecular mechanisms underlying the light-induced changes of different photoswitchable ligand pharmacology is suboptimal. For this reason, we have studied the mechanisms of action of alloswitch-1 and MCS0331; two freely diffusible, mGlu₅ phenylazopyridine photoswitchable negative allosteric modulators. We combined photochemical, cell-based, and in vivo photopharmacological approaches to investigate the effects of trans–cis azobenzene photoisomerization on the functional activity and binding ability of these ligands to the mGlu₅ allosteric pocket. From these results, we conclude that photoisomerization can take place inside and outside the ligand binding pocket, and this leads to a reversible loss in affinity, in part, due to changes in dissociation rates from the receptor. Ligand activity for both photoswitchable ligands deviates from high-affinity mGlu₅ negative allosteric modulation (in the trans configuration) to reduced affinity for the mGlu₅ in their cis configuration. Importantly, this mechanism translates to dynamic and reversible control over pain following local injection and illumination of negative allosteric modulators into a brain region implicated in pain control.

G protein-coupled receptors (GPCR) are crucial players in cell communication and have been identified as key targets for drug discovery, with more than one-third of currently available therapeutic drugs targeting them.¹ One exemplar class C GPCR is the metabotropic glutamate 5 receptor (mGlu₅), whose ubiquitous expression and role in regulating neuronal synaptic activity have popularized this receptor as a therapeutic target for a number of diseases that affect the central nervous system (CNS).²⁻⁵

The development of small molecules acting on allosteric sites on GPCRs has emerged as an attractive strategy toward modulating receptor activity. Through targeting a topographically distinct and non-overlapping binding site from that of orthosteric ligands, allosteric ligands have the potential to increase selectivity for a receptor while also being capable of modulating the affinity and/or efficacy of a co-binding endogenous orthosteric ligand. For allosteric modulators without intrinsic efficacy, there is a reduced risk of receptor oversensitization, and there is an opportunity for fine-tuning endogenous signaling in a more spatiotemporal manner, which is a favorable therapeutic strategy for complex CNS disorders.^6,7

To date, mGlu₅ allosteric modulators have shown efficacy in a number of preclinical models representative of aspects of the symptomology of CNS disorders. For example, mGlu₅ negative allosteric modulators (NAMs) show improvements in mouse models of acute and neuropathic inflammatory pain.^4,8 There is concern though, for potential on-target adverse effects particularly in nontargeted tissues relating to the use of mGlu₅ NAMs; their administration can impair cognition and induce psychotomimetic effects.^9,10 This raises an important limitation to classical pharmacological approaches (i.e., small-molecule ligands) where as soon as a compound is systemically administered to an organism the precise control of the activity of the ligand at its site of action or in undesired tissues is lost. Photopharmacology is emerging as an interesting approach for achieving better spatiotemporal control over drug actions for their targets.¹¹ It involves using ligands whose activity can be controlled by light. During illumination, the ligand may absorb a photon to trigger a photochemical reaction inducing a structural rearrangement, which can change the determinants of the molecular interaction with their receptor and hence its biological activity. Different strategies have emerged to make a ligand photocontrollable. The more common methods involve either chemically attaching the ligand to a photocleavable cage that prevents the ligand’s activity (i.e., photocaged compounds) or the incorporation of a photochromic element (typically an azobenzene moiety) into the molecular scaffold of the ligand (i.e., freely diffusible photoswitchable or photochromic ligands) or into a linker covalently tethering the ligand to its receptor (i.e., phototethered compounds).¹²⁻¹⁸

When appropriately illuminated, freely diffusible photoswitchable ligands can absorb photons to generate a configurational and reversible change in their structure; alternating between two isomeric forms with different shapes and molecular properties that can have opposing or different pharmacological effects at their target receptor. These ligands offer the opportunity to modulate and understand GPCR signaling in a more precise manner than classical pharmacological approaches^19,20 since it is possible to externally operate their interaction with the receptor. The use of light to cause photoisomerization of photochromic molecules can allow for dynamic regulation of receptor activity in a more spatiotemporal manner.^14,21−23 Photoswitchable ligands have been developed for various GPCRs,^16,24−28 and notably for metabotropic glutamate receptors.^12,13,22 For example, alloswitch-1, the first molecule to be designed as a photoswitchable allosteric modulator, acts as a potent NAM of mGlu₅ signaling under dark conditions, in vitro, while under ultraviolet (UV) light it has reduced NAM activity.¹² This reversible NAM activity is believed to underlie the ability of alloswitch-1 to reversibly modulate the behavior of freely moving Xenopus tropicalis tadpoles, zebrafish larvae, and mice under different light conditions.^12,13 More recently, medicinal chemistry and screening efforts have identified a second generation of mGlu₅ photoswitchable NAMs. This new series includes MCS0331, which has a longer half-life in its less thermodynamically stable cis isomer as compared to alloswitch-1, despite retaining similar in vitro and in vivo activity to that of the aforementioned mGlu₅ NAM.¹³

To date, the molecular mechanisms behind the light-induced change of in vitro and in vivo pharmacological activity of these azobenzene-containing ligands are not well understood, and mechanistic studies investigating this are lacking. It remains underappreciated whether the loss of activity of ligands is a consequence of an inability of one of the photoisomers to bind to the receptor or if it is due to more subtle changes in the binding modes of the different photoisomers that leads to changes in their ability to cause receptor activation or modulation. This is important as GPCR allosteric ligands can exhibit an assortment of activities, and the magnitude and type of activity they exert has implications for their ability to demonstrate in vivo efficacy, particularly in a context-dependent manner.²⁹ It is also not known whether photoswitching between the isomeric forms of the ligands can only take place when they are in the solvent and not in contact with the receptor or if this can also occur while the ligand is bound to the binding pocket, which may be important information when designing experiments that intend to investigate receptor function and dynamics in real-time, and in a reversible manner or when trying to control rapid cellular events on a target tissue.³⁰

The aim of the present study was to understand how light affects the pharmacological properties of alloswitch-1 and its analogue MCS0331, which are NAMs of mGlu₅ with similar in vitro and in vivo potency but have different physicochemical properties.^12,13 Using photochemistry and cell-based photopharmacology, we first defined the optimal wavelength and irradiance for photoisomerization of these two phenylazopyridines. Then we studied the consequences of photoisomerization on their functional activity and binding ability to mGlu₅, with the use of IP₁ functional assays, mass spectroscopy, radioligand binding assays, and translation to a mouse model of inflammatory pain. These data suggest that ligand bound photoisomerization takes place inside and outside the mGlu₅ allosteric pocket, and this drives a reversible change in allosteric ligand affinity for the inactive receptor, which underlies their reversible in vivo activity.

Results

Photochemical Properties of Alloswitch-1 and MCS0331

UV–vis spectra for each ligand (Figure 1A) were recorded in the dark and following illumination with different wavelengths of light ranging from 365–550 nm to reach each ligand’s corresponding photostationary state (PSS, the steady state or equilibrium reached by the reversible photochemical isomerization reaction upon illumination under defined conditions with specific wavelengths and intensities; Figure 1B). Under dark conditions, the tested azobenzene compounds exist in their trans isomeric form and each UV–vis absorption spectra shows a characteristic strong π–π* transition band between 370–380 nm. Following illumination with different wavelengths, different PSS were obtained for the equilibrium states for trans to cis and cis to trans photoisomerization of the N=N double bond present in the azo-compounds. For biological applications, we choose wavelengths that maximized the percentage of either cis or trans isomers to obtain a more pronounced biological switch under different light conditions. From these experiments, 380 nm was found to be the optimal wavelength to isomerize both alloswitch-1 and MCS0331 compounds from the extended trans isomer to the bent cis configuration. Wavelengths between 500–550 nm facilitated high levels of cis-to-trans photoisomerization for both allosteric NAM compounds, which is consistent with previous studies.^12,13 The thermal relaxation in aqueous solution of the metastable cis isomers to the thermodinamically stable trans isomers was also assessed at 37 °C. The half-life of cis-alloswitch-1 was 26.7 seconds (s), while for cis-MCS0331, it was 1240 s (Figure 1C). Both compounds demonstrated robust and repetitive photochromic behavior in aqueous media for several photoisomerization cycles following 380 and 500 nm illumination at 37 °C (Figure 1D).

Figure 1.

Chemical structure of alloswitch-1 and MCS0331 (A). UV–visible absorption spectra of 30 μM alloswitch-1 (B; top) and MCS0331 (B; bottom) following continuous illumination with different wavelengths of light (detailed in the “Methods” section) or dark conditions in 3% DMSO binding buffer, pH 7.5. Inset graphs highlight the absorption between 450 and 550 nm for alloswitch-1 (B; top) and for MCS0331 (B; bottom). Thermal relaxation of 30 μM alloswitch-1 (C; top) and 30 μM MCS0331 (C; bottom), following continuous illumination with 380 nm, in 3% DMSO, binding buffer, pH 7.5 at 37 °C. Reversibility and stability of photoisomerization of 5 μM alloswitch-1 in binding buffer containing 5% DMSO, pH 7.5 (D; top) and 30 μM MCS0331 in 3% DMSO binding buffer, pH 7.5 (D; bottom) following continuous illumination with 380 or 500 nm. Representative ¹H NMR spectra of (E; top) alloswitch-1 (1 mM) in DMSO-d₆ and (E; bottom) MCS0331 (500 μM) in 5% DMSO-d₆ in D₂O following either dark conditions or 3 min of continuous illumination with 380 or 500 nm at 25 and 12 °C, respectively. All absorbance data were normalized between 0–0.5 (B–D).

Determining the photoisomeric ratios of each PSS of a photoswitchable ligand can be critical when attempting to understand the pharmacological outcomes of a ligand and the mechanisms underlying those effects.²⁰ For this reason, the photoisomeric ratios of alloswitch-1 (Figure 1E top; 100% DMSO-d₆) and MCS0331 (Figure 1E bottom; 5% DMSO-d₆ in D₂O) were determined by ¹H NMR spectroscopy under 500 or 380 nm light (as this was determined to be the optimal photoisomerization wavelength) at 25 and 12 °C, respectively. Different solvent and temperature conditions were trialled for each compound to perform these experiments. Under dark conditions, both compounds existed only in the thermodynamically more stable trans configuration. Following 500 nm illumination, 76.5% of alloswitch-1 was present in the trans form, and 75.2% of MCS0331 molecules existed in their trans form, with the remainder of molecules being present in the cis configuration (Figure 1E). After 380 nm illumination, 73.6% of alloswitch-1 molecules were detected in the cis isomeric form, whereas 80.0% of MCS0331 molecules were in the cis form, with the remaining molecules existing in their respective trans configurations (Figure 1E).

Fine-Tuning Wavelength and Irradiance for Photoswitching of mGlu₅ NAM Biological Activity

The effects of wavelength and irradiance on the ability of alloswitch-1 and MCS0331 to act as NAMs were studied in the canonical G_q/11-linked inositol phosphate (IP₁) accumulation assay in HEK 293 cells transiently transfected with the human (h)mGlu₅ receptor (Figure 2). The change in pIC₅₀ values reported for these interactions between photoswitchable NAMs and the test agonist provides information about the photoswitching efficiency of the ligands; the larger the difference in potency between light conditions, the more efficient the photoswitching of the ligand between its isomeric states.¹³ In agreement with previous reports, both alloswitch-1 (Figure 2A,C) and MCS0331 (Figure 2B,D) completely inhibited the IP₁ response induced by a single, fixed concentration of quisqualate (100 nM) under dark conditions (pIC₅₀ values of 7.64 ± 0.06 and 7.46 ± 0.07, respectively, Figure 2A–D) indicative of strong negative cooperativity between the allosteric ligands and quisqualate.^12,13 Following illumination with different wavelengths, the largest change in pIC₅₀ values from dark conditions for either alloswitch-1 or MCS0331 were observed for wavelengths in the 365–420 nm range, and there were no significant differences between these values (Figure 2E). Thus, 380 nm light was selected for promoting trans-to-cis photoisomerization of the phenylazopyridines for the remainder of the experiments, while illumination with 500 nm was chosen to promote cis-to-trans photoisomerization of the phenylazopyridines.

Figure 2.

Effects of wavelength and irradiance on alloswitch-1 and MCS0331 negative allosteric modulation of mGlu₅. In IP₁ accumulation assays, the effect of different wavelengths of light (irradiance at 0.12 mW/mm²) on the ability of alloswitch-1 (A) or MCS0331 (B) to inhibit the functioning of an EC₈₀ concentration of quisqualate (100 nM) was assessed at the hmGlu₅. The ability of different irradiances of the 380 nm on alloswitch-1 (C) or MCS0331 (D) modulation of 100 nM quisqualate was also investigated. The change in pIC₅₀ values between dark and light conditions for each interaction was determined following illumination with different wavelengths of light (E) or illumination at 380 nm with variable irradiances (F). Data represent the mean ± SEM of at least three independent experiments performed in duplicate.

In order to determine the optimal light intensity for photoswitching from trans-to-cis isomeric forms, each photoswitchable NAM was allowed to interact with 100 nM of quisqualate under 380 nm conditions, with different irradiances ranging from 0 to 0.15 mW/mm² (Figure 2C,D,F). Values from 0.10 to 0.15 mW/mm² demonstrated the greatest pharmacological potency change for NAMs, but they were not significantly different. For this reason, wavelength irradiance was set to 0.12 mW/mm². Interestingly, the change in potency for each interaction at different wavelengths and irradiances appeared to reach a limit where no further changes in potency could be seen for that interaction under those conditions, indicative of the system reaching a PSS at a given wavelength and intensity. The data from Figure 2F also suggest that, when a system has insufficient time to reach a PSS, the light power can influence the rate at which a photoswitchable ligand can photoisomerize (and reach a PSS), as reflected in the differences in the ligand potency values with changing light power.

Photoswitchable Negative Allosteric Modulation of Agonist-Induced IP₁ Accumulation

In order to quantify functional affinity and cooperativity estimates in IP₁ accumulation assays, NAM interactions with agonists were fitted to an operational model of allosterism.³¹ In all cases, the functional affinity for orthosteric agonists were constrained to values determined in separate IP₁ accumulation assays; where cells were transfected with different amounts of hmGlu₅ receptor, and under dark or light (500 or 380 nm) conditions agonist concentration-response curves were generated (Figure S1). By fitting the resulting curves to the operational model of agonism, it was determined that both the functional affinity and operational efficacy estimates generated for either glutamate or quisqualate were unaffected by the light conditions, which is consistent with the activity of these ligands being independent of light (glutamate pK_A = 4.51–4.60, glutamate τ = 1.41–2.0; quisqualate pK_A = 7.26–7.65, quisqualate τ = 1.17–1.55; Table S1).^32,33 For interaction studies with NAMs, under dark or 500 nm conditions, the functional affinity estimates for alloswitch-1 and MCS0331 were in the ranges of pK_B = 6.87–8.04 and 7.43–8.65, respectively (Table 1). Under these conditions and at the concentrations tested, both alloswitch-1 and MCS0331 behaved as NAMs of agonist-mediated IP₁ accumulation (alloswitch-1 Logβ = −0.62 to −1.25; MCS0331 Logβ = −0.97 to −1.35, Figure 3; Table 1). Both NAMs had a trend for higher negative cooperativity with quisqualate activity than with glutamate at the mGlu₅, and their functional affinity estimates were reduced under 500 nm conditions as compared to dark conditions, with the exception of the interaction between MCS0331 and glutamate, which resulted in equivalent functional affinity estimates between dark and 500 nm conditions (Figure 3, Table 1).

Table 1. Operational Model Estimates for the Interactions between Alloswitch-1 or MCS0331 with the Indicated Othosteric Agonist and Light Condition in IP₁ Accumulation Assays in hmGlu₅-Expressing HEK Cells.

	Dark				500 nm				380 nm
	+ glutamate		+ quisqualate		+ glutamate		+ quisqualate		+glutamate		+ quisqualate
Allosteric ligand	pKBa	log β(β)b	pKBa	log β (β)b	pKBa	log β (β)b	pKBa	log β (β)b	pKBa	log β(β)b	pKBa	log β (β)b
Alloswitch-1	7.34 ± 0.07	–0.97 ± 0.11	8.04 ± 0.09	–1.25 ± 0.10	6.97 ± 0.06	–0.62 ± 0.03	6.87 ± 0.07	–1.23 ± 0.24	5.96 ± 0.14	–0.81 ± 0.13	5.88 ± 0.23	–0.89 ± 0.11
		(0.11)		(0.06)		(0.24)		(0.06)		(0.15)		(0.13)
MCS0331	= 7.50c	–1.16 ± 0.13	8.65 ± 0.05	–1.35 ± 0.08	7.43 ± 0.06	–0.97 ± 0.10	7.75 ± 0.07	–1.08 ± 0.14	6.39 ± 0.09	–0.93 ± 0.13	6.54 ± 0.11	–0.75 ± 0.17
		(0.07)		(0.04)		(0.11)		(0.08)		(0.12)		(0.18)

^aEstimated parameters represent the mean ± SEM of at least three experiments performed in duplicate. Functional IP₁ accumulation responses were analyzed according to eq 1 in the Supporting Information. Negative logarithm of the allosteric modulator equilibrium dissociation constant.

^bLogarithm of the efficacy scaling factor (antilog values shown in parentheses) for the effect of the indicated NAM on agonist responses under different light conditions; when the logarithm of the affinity cooperativity between the agonist and the NAM is equal to zero (log α = 0), it is equivalent to the combined functional cooperativity between ligands (Log αβ).

^cCombined cooperativity values were derived from applying an absolute sum-of-squares analysis.

Figure 3.

Effect of light on the ability of the mGlu₅NAMs, alloswitch-1 (A–C, G, I), or MCS0331 (D–F, J–L), to negatively modulate (A–F) glutamate- or (G–L) quisqualate-stimulated IP₁accumulation in HEK cells transiently transfected with hmGlu₅. Data are expressed as a percent of maximal quisqualate response, as determined by a fixed concentration of quisqualate (quis; 10 μM), and represent the mean ± SEM of at least three independent experiments performed in duplicate. Fitted curves are from global analysis of data sets according to eq 1 (in the Supporting Information), with the appropriate constraints and parameter estimates shown in Table 1.

Following 380 nm irradiation, both ligands had reduced functional affinity for the receptor, regardless of the interacting orthosteric probe (Table 1). They also showed minimal differences in their negative cooperativity with glutamate activity and only a small reduction in negative cooperativity with quisqualate; this change, though, was not substantial (Figure 3C,F,I,L; Table 1). Taken together, these data indicate that the major mechanism underling the change in activity of alloswitch-1 and MCS0331 from dark to 380 nm conditions is due to a loss in affinity for the mGlu₅ rather than a change (or switch) in cooperativity, whereby the cis isomers of the phenylazopyridines may no longer be able to bind to the allosteric pocket of the receptor or they bind to a lower affinity site on the receptor.

Photoswitchable Control of Affinity and Dissociation Kinetics of mGlu₅ Ligands by MS Binding

To further understand the mechanisms of modulation of alloswitch-1 and MCS0331, MS binding studies under dark and light (500 and 380 nm) conditions were undertaken on membranes expressing mGlu₅ (Figure 4). The advantage of utilizing the MS binding technique over more conventional radioligand binding assays is that it allows for the quantification of affinity estimates from the direct binding of a ligand of interest to its target without the need for a competing radioligand. In saturation binding assays and under dark and 500 nm conditions, the equilibrium dissociation constants for alloswitch-1 were pK_D(dark) = 7.51 ± 0.04 and pK_D(500) = 7.47 ± 0.07, and for MCS0331, pK_D(dark) = 7.85 ± 0.03 and pK_D(500) = 7.58 ± 0.06, respectively (Table 2). For experiments performed under 380 nm conditions, when phenylazopyridines exist mostly in their cis form, the equilibrium dissociation constants for alloswitch-1 and MCS0331 were significantly augmented as compared to dark conditions (alloswitch-1 pK_D(380) = 7.03 ± 0.05 and MCS0331 pK_D(380) = 6.90 ± 0.10). These results support the IP₁ accumulation data which suggest that there is a robust change in the affinity of alloswitch-1 and MCS0331 from dark to 380 nm conditions, and this may underlie the change in the functional potency of these ligands.

Figure 4.

Effect of light on alloswitch-1 and MCS0331 affinity and dissociation kinetics as determined by MS binding assays. For MS saturation experiments (A, B), mGlu₅-expressing membranes were incubated with increasing concentrations of marker ligand for 1 h at 37 °C under different light conditions. For dissociation kinetic experiments, (C) 32 nM of alloswitch-1 or (D) 15 nM of MCS0331 was incubated with mGlu₅-expressing membranes for 1 h at 37 °C, under dark condition, and dissociation half-lives (Table 3) were determined by the addition of a saturating concentration of the competitive NAM, VU409106, at different time points and under different light conditions. Data are expressed as a percentage of specific binding and represent the mean ± SEM of at least three independent experiments performed in duplicate.

Table 2. Summary of Photoswitchable NAM Affinities and B_max Estimates under Different Light Conditions as Determined by MS Binding Assays^a.

	Dark		500 nm		380 nm
Allosteric ligand	pKDb	Bmax (pmol/mg protein)c	pKDb	Bmax (pmol/mg protein)c	pKDb	Bmax (pmol/mg protein)c
Alloswitch-1	7.51 ± 0.04	71.52 ± 6.89	7.47 ± 0.07	76.34 ± 13.98	7.03 ± 0.05****	35.06 ± 11.68
MCS0331	7.85 ± 0.03	92.84 ± 9.92	7.58 ± 0.06*	151.51 ± 16.28	6.90 ± 0.10****	21.06 ± 2.05

^aData sets were analyzed according to eq 2 in the Supporting Information. Difference in equilibrium dissociation constants of allosteric ligands under different light conditions were determined by one-way ANOVA and Tukey post hoc test where appropriate. *, P < 0.1, and ****, P < 0.0001 vs dark conditions.

^bNegative logarithm of the allosteric modulator equilibrium dissociation constant.

^cReceptor expression of hmGlu₅.

We used the observed photoisomeric ratios of alloswitch-1 and MCS0331 determined by ¹H NMR (Figure 1E) to resolve if a reduction of the concentration of active trans isomer is sufficient to account for the change in observed affinity of each ligand between light conditions. By making the assumption that the trans isomer of the phenylazopyridines is the only species that binds to the receptor and has functional activity, it is possible to correct the concentrations reported in the binding assays, under different light conditions, to only account for the trans isomer and then to compare the adjusted affinity value with that reported in Table 2 for assays performed under dark conditions (i.e., where only trans isomer is present). By simulating this with the alloswitch-1 380 nm saturation binding data set, a corrected trans-alloswitch-1 equilibrium dissociation constant was determined to be pK_D = 7.58 ± 0.08, which is not significantly different from the pK_D reported for alloswitch-1 under dark conditions, as determined by one-way ANOVA (Table 2). This suggests that the change in the reported pK_D values between dark and 380 nm conditions could be explained by a dilution effect of the active trans species and a reduced binding of the cis isomer. Similarly, by applying the same simulation to the MCS0331 data set, we found that the change in the reported pK_D values from dark to 380 nm conditions could also be explained by a dilution effect of trans-MCS0331, by correcting the MCS0331 concentration in the 380 nm data set, a new equilibrium dissociation constant was generated to be pK_D = 7.60 ± 0.10, which is not significantly different from the pK_D reported for MCS0331 under dark conditions (Table 2).

To determine if cis isomer binding of ligands is possible and if photoisomerization can take place within the mGlu₅ binding pocket, in addition to appreciating which factors may contribute to the affinity change of these ligands at the mGlu₅, MS dissociation kinetic assays were also performed with the phenylazopyridines following dark or light (500 or 380 nm) conditions. Under dark conditions, the dissociation rate of MCS0331 for its binding site was slower than alloswitch-1, which could, in part, explain the improved affinity and potency of MCS0331 in the IP₁ accumulation assay as compared to alloswitch-1 (Figure 4C, D; Table 3). Following 500 or 380 nm irradiation, after the association phase was established in the dark, the dissociation rates of both alloswitch-1 and MCS0331 were significantly increased as compared to that under dark conditions (Figure 4C, D; Table 3). These data suggest that the decreased affinity of either alloswitch-1 or MCS0331, under 380 nm conditions, is in part due to ligands in the cis configuration binding to the receptor and dissociating at a faster rate than their respective trans isomers. The data also support the concept that photoswitching between isomeric forms of the ligands can occur inside the mGlu₅ allosteric pocket. Since the association phase is performed under dark conditions to allow for trans isomer binding, the initiation of the dissociation phase under 380 nm conditions would then suggest that trans-to-cis photoisomerization of ligands within the allosteric pocket is possible, given that the observed dissociation rate is significantly faster (Figure 4C,D). Additional MS dissociation kinetic assays were also performed with MPEP as a control to show that its dissociation rate was minimally affected by light (Figure S3; Table S4).

Table 3. Summary of Photoswitchable NAM Dissociation Kinetic Parameters Estimates under Different Light Conditions as Determined by MS Binding Assays^a.

	Dark		500 nm		380 nm
Allosteric ligand	koff (min–1)b	t1/2 (min)c	koff (min–1)b	t1/2 (min)c	koff (min–1)b	t1/2 (min)c
Alloswitch-1	0.14 ± 0.01	5.04 ± 0.10	0.25 ± 0.01*	2.77 ± 0.14	0.93 ± 0.16****	0.81 ± 0.15
MCS0331	0.07 ± 0.01	10.70 ± 0.95	0.11 ± 0.01*	6.32 ± 0.29	0.47 ± 0.07****	1.61 ± 0.28

^aDifference in the dissociation rate was determined by one-way ANOVA, which is performed on the logarithm of the k_off values. *, P < 0.1, and ****, P < 0.0001 vs dark condition for 500 and 380 nm conditions, respectively.

^bDissociation rate of the indicated NAM.

^ct_1/2 is the half-life of dissociation.

Effects of Interacting Photoswitchable NAMs with [³H]-MPEP in Radioligand Binding Assays

Previous computational docking and MD studies have suggested that alloswitch-1 may bind deep within the common MPEP allosteric pocket of mGlu₅ seven transmembrane domain (7TM) and experimental studies show that alloswitch-1 NAM activity is affected following mutation to either S809 or the P655 residue, which also effect MPEP activity.³⁰ Thus, [³H]-MPEP radioligand binding studies were carried out with membranes prepared from mGlu₅-HEK293 cells to determine if alloswitch-1 and/or MCS0331 share a common/overlapping binding site with [³H]-MPEP, in addition to determining the cooperative effects of NAMs on agonist binding alone (α), under different light conditions (Figure 5).

Figure 5.

[³H]-MPEP competition radioligand binding studies to demonstrate the effect of either a saturating concentration of glutamate (1 mM glu) or quisqualate (30 μM quis) on (A) alloswitch-1- or (B) MCS0331-mediated inhibition of [³H]-MPEP binding in hmGlu₅-expressing membranes. Data are expressed as a percentage of specific binding and represent the mean ± SEM of at least three independent experiments performed in duplicate. Fitted curves are from global analysis of data sets according to eq 3 in the Supporting Information with parameter estimates shown in text.

[³H]-MPEP bound to mGlu₅-membranes in a monophasic and saturable manner with an estimated pK_D = 8.40 ± 0.009 and maximal binding capacity of (B_max) of 1777 ± 72.05 fmol/mg of protein under dark conditions, and these estimates were not significantly affected by changing the light conditions (data not shown). Under dark conditions, both alloswitch-1 and MCS0331 completely inhibited the specific binding of a fixed concentration of [³H]-MPEP (∼10 nM) indicative of NAMs binding to or overlapping with the common MPEP binding site (Figure 5). Furthermore, there was no difference between the apparent binding affinity estimates for either alloswitch-1 or MCS0331 between dark (alloswitch-1 pK_D = 7.91 ± 0.11; MCS0331 pK_D = 8.33 ± 0.07) and 500 nm conditions (alloswitch-1 pK_D = 7.92 ± 0.08; MCS0331 pK_D = 8.03 ± 0.08), but the apparent binding affinities of NAMs were reduced for 380 nm conditions (alloswitch-1 pK_D = 7.16 ± 0.05; MCS0331 pK_D = 6.59 ± 0.10), which is consistent with both the IP₁ accumulation and MS binding data.

In order to evaluate the effect of agonist binding to the orthosteric site on NAM binding to the [³H]-MPEP allosteric site, [³H]-MPEP competition binding studies were repeated in the absence or presence of an orthosteric-site-saturating concentration of either glutamate (1 mM) or quisqualate (30 μM).^34,35 Incubation of glutamate or quisqualate with [³H]-MPEP alone had no effect on the specific binding of [³H]-MPEP (data not shown), and the apparent binding affinities of alloswitch-1 or MCS0331 were also unaffected by the presence of either 1 mM glutamate or 30 μM quisqulate, under dark or 500 nm conditions, as determined by F-test (Figure 5). Following 380 nm irradiation though, there was a small difference in the apparent affinity estimates of both NAMs in the presence of either agonist (ΔpK_I = 0.26–0.31; P < 0.05 F-test; Figure 5). Since the concentrations of agonist used in these experiments fully occupies the orthosteric site, any observed change in the apparent binding affinity of NAMs in the presence or absence of agonist would reflect the degree of affinity cooperativity (α) between the respective interacting agonist/NAM pairs. These data then suggest that there is very limited to negligible affinity cooperativity existing between NAM and agonist interacting pairs at the mGlu₅ under all light conditions.

Reversible Loss of NAM Effect through Photoisomerization Translates In Vivo

Considering the mechanisms of action of these photoswitchable NAMs, it was of interest to determine how their in vitro activity is translated in vivo. To that aim, we tested the photoswitchable mGlu₅NAMs in a preclinical mouse model of inflammatory pain in which MPEP, a prototypical mGlu₅ NAM, has historically shown efficacy following local injection in the amygdala (Figure 6).³⁶

Figure 6.

Light-dependent antiallodynic effect of alloswitch-1 and MCS0331 in a mouse model of inflammatory pain. Persistent inflammatory pain was induced in mice by injection of complete Freund’s adjuvant (CFA) into the left hind paw of mice on day 0 (D0) of timeline. In the timeline of experiment, “C” refers to the day of intra-amygdala cannulation, “CFA” refers to day of CFA injection, “B” refers to day of pretest baseline reading. The mechanical pain threshold of animals was evaluated by stimulating the CFA-treated hind paw of animals with von Frey filaments. The mechanical sensitivity of naive animals was measured prior to cannulation, then prior to CFA injection and 7 days after (left-segment of figure). On test day, 8 days after CFA injections (D8; right-segment of figure), the mechanical sensitivity of animals was tested (t0) immediately after they received an intra-amygdala injection of either vehicle (n = 5), MPEP (1 μM; n = 5), alloswitch-1 (300 nM; n = 4), or MCS0331 (300 nM; n = 3). The effect of each treatment was assessed after 15 min (without light) and then every 5 min following intra-amygdala illuminations (protocol: 50 ms light pulses at 10 Hz frequency for 5 min) at 385 nm (irradiance: 0.11 mW/mm²) or 505 nm (0.029 mW/mm²). Data were analyzed by two-way ANOVA and expressed as mean ± SEM, **, P < 0.01, and ***, P < 0.001: significant differences on mechanical pain threshold of drug-treated-animals as compared to vehicle for each time point.

Prior to testing though, the ability of alloswitch-1 and MCS0331 to negatively modulate glutamate function at the mouse mGlu₅, under dark or light (500 or 380 nm) conditions, was first confirmed since some allosteric ligands can show species differences. Both ligands negatively modulated the effects of glutamate to similar extents to that seen for their modulation at the human orthologue, while the activity of the prototypical mGlu₅ NAM, MPEP, was unaffected by light (Figure S5 and Table S4).

Mice were stereotaxically implanted with hybrid optic and fluid cannulas into the amygdala to enable the controlled release of compound and light. Mice were then injected with CFA into the left hind paw to induce persistent inflammatory pain. The effect of intra-amygdala injection of either vehicle, MPEP, alloswitch-1, or MCS0331 on the mechanical pain threshold of animals was then assessed by stimulating the CFA-treated paw with von Frey filaments (8 days after CFA injection; Figure 6). On test day (day 8), intra-amygdala injection of MPEP but not vehicle returned mechanical sensitivity to the level of naive mice (or before inflammation was induced; Figures 6 and S6), and these effects were independent of light. Following injection of either alloswitch-1 or MCS0331, under dark conditions, the mechanical pain threshold of animals was also restored (Figure 6). This analgesic effect of photoswitchable NAMs was lost after intra-amygdala illumination with 385 nm and then restored with 505 nm illumination. This indicates that the loss in affinity of each phenylazopyridine for the mGlu₅, following local trans-to-cis photoisomerization within the amygdala (both inside and outside the binding pocket) leads to insufficient levels of receptor occupancy for the phenylazopyridines to have an analgesic effect on mechanical pain under 380 nm conditions. This effect could be reversed by local 505 nm irradiation and cis-to-trans photoisomerization of ligands to their higher affinity states. This reversible behavior was not observed with MPEP, consistent with it behaving as a nonphotoswitchable mGlu₅ NAM in cell-based assays. These data are also consistent with previous in vivo results that have shown alloswitch-1 and MCS0331 to have reversible analgesic properties in mouse models of pain, so the data further support the potential for using photoswitchable mGlu₅ NAMs to investigate the role of the mGlu₅ in pain in a more spatiotemporal manner.¹³ More broadly speaking, they also outline the potential use of photoswitchable ligands to investigate receptor function in other disease contexts in a more dynamic way.

Discussion

mGlu₅ is an exemplar class C GPCR and represents an important therapeutic target for CNS disorders. Light control of GPCRs, including mGlu₅, through photopharmacology is emerging as an interesting approach for achieving improved functional selectivity and spatiotemporal control over drug actions at their targets, in addition to enhancing our understanding of receptor function and dynamics. Since the field of photopharmacology is in its early stages of development, there are currently limited studies available that have explored the mechanisms of action of photoswitchable allosteric ligands for GPCRs. For instance, alloswitch-1 and MCS0331 are two mGlu₅ photoswitchable allosteric modulators that can reversibly inhibit the functional effects of a fixed concentration of agonist, depending on the light conditions used, but relatively little is known about their mechanism of action/modulation at mGlu₅.^12,13 For this reason, our study has investigated the functional and binding consequences of alloswitch-1 and MCS0331 under different light conditions at mGlu₅ using canonical G_q/11-linked signaling assays (IP₁ accumulation), and MS and radioligand binding assays. This study has also characterized the wavelengths and light intensities at which the phenylazopyridines can photoswitch between their isomeric states and so has fine-tuned in vitro conditions to promote efficient photoswitching of ligands that could potentially be extended to similar experimental photopharmacology set ups.

In terms of photochemistry, this study has shown that despite alloswitch-1 and MCS0331 sharing a common scaffold and similar electronic features each compound has distinct photochemical properties. Alloswitch-1 has a shorter cis-to-trans thermal relaxation half-life as compared to that of MCS0331, and the cis-MCS0331 n−π* transition band is more intense than that of cis-alloswitch-1. These differences may be explained by the electron delocalization through the central ring and the chlorophenyl of alloswitch-1, which is not present in the aliphatic carboxamide of MCS0331. Importantly, these differences in photochemistry contribute to the differences observed in each ligands’ photopharmacology and should be considered when performing pharmacological assays. Photochemical experiments, in this study, were performed under the most appropriate (solvent, concentration, and temperature) conditions found to allow for the accurate estimation of half-lives and PSSs values for each compound under each light condition tested. It should be noted though that the resulting parameter estimates may not be fully representative of the half-lives and PSSs values expected for each phenylazopyridine, when tested in the pharmacological assays, which employ more physiologically relevant conditions (including aqueous buffered solutions and the presence of the mGlu₅). For this reason, photochemical estimates generated in this study have only been used as a guide to aid the interpretation of the pharmacological data under different light conditions. This also highlights the important need for improved photoswitchable ligands with appropriate physical and photochemical properties (e.g., suitable solubility and slow relaxation times) for pharmacological and biological applications.

In functional assays, alloswitch-1 and MCS0331 act as potent NAMs of agonist-induced IP₁ accumulation; NAMs reduce agonist potency and maximum response at mGlu₅, under dark or 500 nm conditions. While following 380 nm irradiation, the pheylazopyridines have reduced NAM activity with agonist-induced IP₁ accumulation, which is consistent with previous reports.^12,13 Functional operational estimates indicate that this switch in activity is overall due to a loss in allosteric ligand affinity, rather than a substantial change in cooperativity between interacting ligands.

To understand this in more detail, MS binding data generated under different light conditions were considered in the context of the photoisomeric ratio of each NAMs’ PSSs. These data suggest that the mGlu₅ affinity of both alloswitch-1 and MCS0331 following 500 or 380 nm irradiation is reduced, as compared to those estimates obtained under dark conditions. Given the mixed population of trans and cis ligands existing under irradiated conditions and when considering the equilibrium binding data in isolation, these observed differences in binding parameters could be explained by a reduction in the available active trans isomer concentration, and a lack of cis isomer receptor binding. It then follows that the potential inability of the cis form of the phenylazopyridines to bind to the receptor may underpin the loss in functional NAM activity observed in the IP₁ accumulation assay, following UV light conditions. For this reason, the MS dissociation kinetic data provide further insight into the binding capabilities of these ligands. Since the dissociation rate constants of ligands were significantly faster under irradiated conditions this suggests that cis-to-trans photoisomerization is possible within the binding pocket and that the cis isomer of ligands are capable of binding to the receptor, albeit with reduced affinity and a shorter residence time (given the faster dissociation rates from the receptor), when compared to the trans isomers. This idea is in line with the hypothesis proposed by Dalton et al., who suggested that photoswitching between isomeric forms could theoretically occur while alloswitch-1 is bound to the allosteric pocket.³⁰ It is though possible that differences in the association rates of each isomeric form of ligands for the receptor may also affect the affinity of these ligands. There is increasing evidence that indicates that the association rates of drugs are not diffusion limited, so in addition to dissociation rates, association rates can also impact the determination of affinity and the pharmacological profile of drugs in different disease contexts.^37,38 Since receptor binding is a dynamic process, a decreased association rate may also impact the ability of a molecule to rebind to a (nearby) receptor following its dissociation into the local environment before it finally diffuses away or is metabolized.³⁷ In this regard, both association and dissociation rates are likely to be important to the mechanism of action of these photoswitchable allosteric ligands.

Additional radioligand binding studies confirmed that the active trans configuration of the phenylazopyridines bind to the common MPEP binding site since they completely inhibit [³H]-MPEP specific binding, where the slopes were ∼1, indicative of a single binding site. Since the apparent affinity estimates of photoswitchable NAMs were not changed (or were only slightly affected) by the addition of an orthosteric-site-saturating concentration of either glutamate or quisqualate, it is also suggested that the major mechanism for modulating the activity of an orthostheric agonist is through modulating its intrinsic efficacy and not binding affinity. The ability of alloswitch-1 and MCS0331 to behave as NAMs of agonist signaling efficacy but not agonist binding affinity is consistent with the mechanisms of other class C mGlu₁ and mGlu₅ NAMs, including CPCCOEt and MPEP.^34,39 This suggests that this could be a common mechanism by which NAMs modulate agonist activity at group I mGlu receptors. Our study also implies that there is little probe-dependent behavior for alloswitch-1 and MCS0331 with respect to the activity of orthosteric, full agonists at the mGlu₅, as they regulate the activity of glutamate and quisqualate by similar mechanisms.

These data, in consideration with the previous computational study by Dalton et al., support the notion that trans phenylazopyridines bind deep in the transmembrane motif of mGlu₅, exploiting in part the MPEP binding pocket where they stabilize a water-molecule-mediated hydrogen-bond network in the bottom of the pocket through connections with residues on TM3, TM6, and TM7 (including residue S809 in TM7, which mGlu₅ prototypical NAMs commonly interact with).^30,40,41 Through these interactions, it is hypothesized that the trans phenylazopyridines stabilize a NAM-induced inactive state of the receptor that promotes the transmission of negative cooperativity between the allosteric and orthosteric sites of the receptor.³⁰ Following trans-to-cis photoisomerization inside (or outside) the allosteric pocket, we along with Dalton et al. predict that the ligands in their cis configuration can no longer stabilize this network or interact with residue S809 or P655, and instead, we hypothesize that they rapidly dissociate from the receptor.³⁰ When bound in this position, cis ligands may undergo a molecular switch, no longer binding as tightly to the receptor or engaging residues important for NAM activity, although we cannot exclude the possibility of differences in association rates of the cis isomer of ligands also impeding their ability to bind to the receptor.

Molecular switches have been well-documented for allosteric ligands at the mGlu₅, in addition to ligands of other mGlu receptors, and have made the development of structure–activity relationships around mGlu₅ allosteric ligands notoriously difficult and superficial.^42−46,40 Future mutational, computational and structural studies that utilize a constrained version of cis-alloswitch-1 and cis-MCS0331, which cannot isomerize back to its active NAM form, may, however, be necessary to shed further understanding on the binding modes of these photoswitchable allosteric ligands, and to identify other key residues important to driving their activity between isomers.

An important consideration when interpreting these data and a limitation to this study is that the dynamic interconversion of the phenylazopyridines between their isomers, induced by photon absorption, occurs faster than biological events at the mGlu₅. A molecule may switch many times between isomers, not only when bound or unbound but also while (un)binding to the receptor or while signal transduction is occurring. For this reason, experimental results from accumulation assays are representative of overall effects that average the dynamic changes that occur to each ligand’s structure as induced by light and are not fully macroscopically adjusted to a classical mixture of ligands with a particular stoichiometry.

Taken together, this study has highlighted that subtle changes in the interactions within the common mGlu₅ allosteric pocket can lead to substantial changes in an allosteric ligand’s affinity and ability to interact with its binding site, which have functional implications. This study also demonstrates that the reversible NAM activity of alloswitch-1 and MCS0331 translates in vivo. In a mouse model of inflammatory pain, in which mGlu₅ NAMs have historically shown efficacy,³⁶ we show that intra-amygdala injection of either alloswitch-1 or MCS0331 rapidly and reversibly improves the mechanical pain threshold of mice to an extent similar to that of the nonphotoswitchable NAM, MPEP. Since previous and in-house studies have shown alloswitch-1 and MCS0331 to be selective NAMs for the mGlu₅, these data suggest that the ability of these ligands to exert negative cooperativity with the signaling efficacy of glutamate is essential to the ligands’ in vivo efficacy.^12,13 It is hypothesized that the key mechanism for their ability to reversible affect mechanical pain in mice is due to their reduced affinity for the mGlu₅, following local trans-to-cis photoisomerization within the amygdala (both inside and outside the binding pocket), which leads to insufficient levels of receptor occupancy in order to have an analgesic effect under 380 nm conditions. This effect can then be reversed by local 505 nm irradiation and cis-to-trans photoisomerization of ligands to their higher affinity (trans) states. We further predict that the change in affinity of these ligands is a result of a faster dissociation rate of the cis-forms of ligands from the receptor. We cannot discount the possibility that slower cis-isomer association rates may also impact the affinity and the prospect of ligands rebinding to mGlu₅ to then affect the biological outcomes of these compounds. These results add to a growing body of evidence for the practicality of using photoswitchable ligands to modulate GPCR activity in vivo and their potential advantages over more conventional pharmacological probes.^{18,21,47−49} Given that the activity of photoswitchable ligands can be reversibly controlled in real time, under defined light conditions, the dynamic control of receptor activity in a site-specific manner may be attainable with ligands of this type as compared to nonphotoswitchable ligands.

In conclusion, this study has provided an in-depth pharmacological characterization of freely diffusible mGlu₅ photoswitchable allosteric modulators, and we propose that the reversible NAM activity, both in vitro and in vivo, of alloswitch-1 and MCS0331 is a result of changes in affinity for their mGlu₅ binding site, under different light conditions. The data described here experimentally support the hypothesis that photoswitching between isomeric forms of these photoswitchable ligands is also possible within the binding pocket. This study will provide the framework for future studies to investigate the mechanism of action of other photoswitchable ligands targeting different receptors and will help to understand GPCR molecular dynamics and function.

Methods

Materials

Sources of all materials are listed in the Supporting Information.

Photochemistry

Details of the photochemistry including absorption spectra, kinetics of cis-to-trans isomer relaxation, stability, and reversibility and ¹H NMR determination for alloswitch-1 and MCS0331 are listed in the Supporting Information.

Cell culture, transfections, inositol phosphate one (IP₁) accumulation assay and membrane preparations

Explanation of cell culture, transfections, the protocol for the inositol phosphate one (IP₁) accumulation assay, and membrane preparations are listed in the Supporting Information.

MS Binding Assays

Equipment settings and compound synthesis are listed in the Supporting Information.

Alloswitch-1 and MCS0331 Saturation MS Binding Assays

hmGlu₅ membranes (20 μg/well) were incubated with test compounds (1–600 nM) for 1 h at 37 °C, while shaking at 150 rpm under dark or light (500 or 380 nm) conditions; the final assay volume was 300 μL/well. For all assays, black clear-bottomed 96-well plates (Greiner Bio-one) were placed over a 96-LED array plate (LEDA, Teleopto) connected to a LED array driver (LAD-1, Teleopto). For dark conditions, half of each assay plate was covered with aluminum foil, and light was pulsed underneath for 50/50 ms on/off at 0.12 mW/mm². The assay was stopped by rapid vacuum filtration through a 1 μm GF filter multiwell plate (AcroPrep Advance 350 μL, Pall Corporation), presoaked for 1 h in 0.5% PEI, with an extraction plate manifold (Pall Corporation). Filter plates were washed 5 times with ice-cold binding buffer (150 μL) to remove unbound ligand and then dried for 1 h at 60 °C. Notably, the filtration and washing steps were performed under 380 nm illumination conditions. Samples were then washed 3 times with 100 μL of acetonitrile containing the corresponding internal standard to elute bound ligands and then once with ammonium bicarbonate buffer (10 mM, pH 7.5; 100 μL per well). Samples were transferred to HPLC vials and analyzed following the appropriate HPLC-MS/MS method (see the Supporting Information section “Chromatographic and Mass Spectrometric Conditions” for more detail). Nonspecific binding was determined in the presence of 10 μΜ VU0409106.

Alloswitch-1 and MCS0331 Dissociation MS Binding Assays

hmGlu₅ membranes (20 μg/well) were incubated with either 32 nM alloswitch-1 or 15 nM MCS0331, respectively, for 1 h at 37 °C while shaking at 150 rpm and in a final assay volume of 300 μL/well under dark conditions. Following the association phase, the dissociation of ligands was initiated with 100 μMVU0409106 at different time points (1–120 min), under dark or 380 nm conditions (50/50 ms off at 12 V). Illumination of plates was performed as described in the previous section. The assay was stopped, and samples were analyzed as described previously. Nonspecific binding was determined in the presence of 10 μΜ VU0409106.

[³H]-MPEP Equilibrium Binding

For [³H]-MPEP competition binding assays, hmGlu₅-expressing membranes (10 μg/well) were incubated for 1 h at 37 °C with ∼10 nM [³H]-MPEP and a range of concentrations of test compounds (100 pM to 10 μM) in the presence or absence of saturating concentrations of either 1 mM glutamate or 30 μM quisqualate,^34,35 under dark or light (500 or 380 nm) conditions (final assay volume of 100 μL/well). For membranes that were incubated under light conditions, 96-well plates were paced over an LED plate (LEDA, Teleopto) and light was pulsed for 50/50 ms on/off at 12 V. Binding assays were then terminated by rapid filtration through 1 μm GF multiwell plates (presoaked for 1 h in 0.5% PEI), and 4 washes with ice-cold binding buffer were used to separate bound and free radioligand. Then, 100 μL of Ultima Gold was added to each well, and radioactivity was counted after at least 1 h of incubation on the MicroBeta plate counter (PerkinElmer). MPEP (10 μM) was used to determine nonspecific binding in all cases.

Animals

Experiments were performed on 8–12 weeks old C57BL/6J males (Charles River). Animals were housed in groups of 3/cage, fed ad libitum, and maintained under a 12 h light/dark cycle. Animals were treated in accordance with the European Community Council Directive 86/609. Experimental protocols were approved by the local authorities (regional animal welfare committee (CEEA-LR)) following the guidelines of the French Agriculture and Forestry Ministry (C34–172–13). All efforts were made to minimize animal suffering and number.

Behavioral Studies

Explanation of behavioral studies is listed in the Supporting Information.

Data Analysis

For a detailed explanation of the curve fitting of data sets, please refer to the Supporting Information.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsptsci.0c00054.

• Materials and methods, receptor depletion on agonist-induced IP₁ accumulation under different light conditions, absolute sum-of-squares analyses performed for the interaction between MCS0331 and glutamate under dark condition in an IP₁ accumulation assay, MPEP MS dissociation experiments, NAM activity at the mouse mGlu5 under different light conditions, effect of intra-amygdala injection of (non)photoswitchable mGlu5 NAMs on mechanical pain threshold, neuroanatomical validation of injection sites, operational model parameters, within- and between-run accuracy and precision of analytical method to determine alloswitch-1 and MCS0331 signals, summary of MPEP dissociation kinetic parameter estimates, model of allosterism parameters for interactions at mouse mGlu₅, and additional references (PDF)

Author Contributions

^# M.R.-O. and A.E.B. contributed equally to this work.

Author Contributions

M.R-O. and A.E.B. designed and performed in vitro photopharmacology assays, analyzed and discussed results, and wrote the paper. V. P. and A.E.B. performed photopharmacology in vivo assays. F.M. set up in vitro photopharmacology assays. J.C., J.F., and X.G.S. synthesized compounds. X.G.S. also contributed to the writing of the paper. C.Z. set up in vivo photopharmacology assays. L.M., C.S., and X.R. contributed to HPLC-MS binding assay development. X.R. also discussed results. C.G. and A.L. designed the research, managed the project, revised the results, and wrote the paper.

The authors declare no competing financial interest.