Optical tools for visualizing and controlling human GLP-1 receptor activation with high spatiotemporal resolution

The glucagon-like peptide-1 receptor (GLP1R) is a broadly expressed target of peptide hormones with essential roles in energy and glucose homeostasis, as well as of the blockbuster weight-loss drugs semaglutide and liraglutide. Despite its large clinical relevance, tools to investigate the precise activation dynamics of this receptor with high spatiotemporal resolution are limited. Here, we introduce a novel genetically encoded sensor based on the engineering of a circularly permuted green fluorescent protein into the human GLP1R, named GLPLight1. We demonstrate that fluorescence signal from GLPLight1 accurately reports the expected receptor conformational activation in response to pharmacological ligands with high sensitivity (max ΔF/F0=528%) and temporal resolution (τON = 4.7 s). We further demonstrated that GLPLight1 shows comparable responses to glucagon-like peptide-1 (GLP-1) derivatives as observed for the native receptor. Using GLPLight1, we established an all-optical assay to characterize a novel photocaged GLP-1 derivative (photo-GLP1) and to demonstrate optical control of GLP1R activation. Thus, the new all-optical toolkit introduced here enhances our ability to study GLP1R activation with high spatiotemporal resolution.


Introduction
The glucagon-like peptide-1 receptor (GLP1R) is expressed in various parts of the brain, especially in the basolateral amygdala and hypothalamic regions (Alvarez et al., 2005;Cork et al., 2015;Trapp and Brierley, 2022, p. 1;Turton et al., 1996, p. 1), as well as broadly outside the central nervous system (Campos et al., 1994). Its endogenous ligand, glucagon-like peptide-1 (GLP-1), is a peptide, fully conserved across mammals, that carries out both central and endocrine hormonal functions for the control of energy homeostasis (Andersen et al., 2018, p. 1). GLP-1 is produced mainly by two cell types: preproglucagon (PPG) neurons principally located in the Nucleus of the Solitary Tract (NTS) of the brain (Trapp and Brierley, 2022;Turton et al., 1996), and enterocrine cells (ECs) located in the gut (Trapp and Brierley, 2022). Upon ingestion of a meal, GLP-1 is rapidly released along with gastric inhibitory polypeptide (GIP) from the gut into the bloodstream where it targets β-cells of the pancreas and stimulates the production and secretion of insulin under hyperglycemic conditions (Andersen et al., 2018, p. 1). This phenomenon, known as the "incretin effect" (Nauck and Meier, 2018), is impaired in metabolic disorders, such as type-2 diabetes mellitus (Holst et al., 2009), making GLP-1 signaling an attractive therapeutic target for the treatment of these disorders. In addition to its role in controlling satiety and food intake, central GLP-1 has also been shown to play central neuroprotective roles (Hölscher, 2022), illustrating its multifaceted role in human physiology.
The human GLP1R (hmGLP1R) is a prime target for drug screening and drug development efforts, since GLP-1 receptor agonists (GLP1RAs) have been used for decades for the treatment of type-2 diabetes and have more recently become some of the most effective and widely-used weight-loss drugs (Shah and Vella, 2014). Among the techniques that can be adopted in these screening efforts are those able to monitor ligand binding to GLP1R through radioactivity-based assays (Knudsen et al., 2007) or fluorescently-labelled ligands (Ast et al., 2020), and those able to monitor the coupling of GLP1R to downstream signaling pathways, for example through scintillation (Runge et al., 2003), fluorescence (Biggs et al., 2018) or bioluminescence resonance energy transfer assays (Zhang et al., 2020, p. 1). A technology to directly probe ligand-induced GLP1R conformational activation with high sensitivity, molecular specificity and spatiotemporal resolution could facilitate drug screening efforts and open important new applications Frank et al., 2018), but is currently lacking.
To overcome these limitations, here we set out to engineer and characterize a new genetically-encoded sensor based on the GLP1R, using an established protein engineering strategy (Duffet et al., 2022a;Patriarchi et al., 2018Patriarchi et al., , 2019Sun et al., 2018). This sensor, which we call GLPLight1, offers a direct and real-time optical readout of GLP1R conformational activation in cells, thus opening unprecedented opportunities to investigate GLP1R physiological and pharmacological regulation in detail under a variety of conditions and systems. We demonstrated its potential for use in pharmaceutical screening assays targeting GLP1R, by confirming that GLP1R and GLPLight1 show similar ligand recognition profiles, including high specificity towards GLP-1 over other class-B1 GPCR ligands, low-affinity for glucagon, and specific functional deficits of GLP-1 alanine mutants. Finally, to extend the optical toolkit further, we also developed a photocaged GLP-1 derivative (photo-GLP1) and adopted it in concert with GLPLight1 to enable all optical control and visualization of GLP1R activation.

Development of a genetically-encoded sensor to monitor hmGLP1R activation
To develop a genetically encoded sensor based on the hmGLP1R, we initially replaced the third intracellular loop (ICL3) of hmGLP1R with a cpGFP module from the dopamine sensor dLight1.3b (Patriarchi et al., 2018), between residues K336 and T343 (Figure 1a). This initial sensor prototype had poor surface expression and a very small fluorescence response upon addition of a saturating concentration (10 µM) of GLP-1 (ΔF/F0 = 39%, Figure 1-figure supplement 1a). Removal of the endogenous GLP1R N-terminal secretory sequence (amino acids    Fig. 1a). We then performed a lysine scan on the residues spanning the intracellular loop-2 (ICL2) of the sensor. From this screening we identified one beneficial mutation (L260K) that more than doubled the dynamic range of the sensor (ΔF/F0 = 180%, Figure 1-figure supplement 1b). Next, we performed site-saturated mutagenesis on both receptor residues adjacent to the cpGFP and screened a subset of 95 variants. This small-scale screening led us to identification of a new variant (containing the mutations K336Y and T343N) with ΔF/F0 of about 341% (Figure 1-figure supplement 1c-d). To further enhance surface expression of the sensor, we introduced a C-terminal endoplasmic reticulum export sequence (Stockklausner and Klocker, 2003) on this variant (Figure 1-figure   supplement 1e). We then introduced three previously-described (Wan et al., 2021) mutations in the cpGFP moiety, which improved the basal brightness of the probe without affecting its dynamic range 4 ( Figure 1-figure supplement 1f-g). Finally, we mutated eight phosphorylation sites on the C-terminal domain that are responsible for GLP1R internalization (Thompson and Kanamarlapudi, 2015)

In vitro characterization of GLPLight1
To establish the utility of GLPLight1 as a new tool to investigate the human GLP1R in pharmacological assays, we first characterized its properties in vitro. We started by comparing sensor expression and fluorescent response among different cell types. To do so, we expressed GLPLight1 in primary cortical neurons in culture, via adeno-associated virus (AAV) transduction. Two weeks post-transduction, GLPLight1 was well expressed on the neuronal membrane and showed a maximal response of 456% to GLP-1 application (10 µM) (Figure 1b-c). We then measured the spectral properties of the sensor in HEK cells. The fluorescence spectra were similar to those of previously described green GPCRsensors (Duffet et al., 2022a;Sun et al., 2018), and showed a peak excitation around 500 nm, peak emission around 512 nm, and an isosbestic point at around 425 nm (Figure 1d). Work on previously developed GPCR-based sensors that respond to neuropeptide ligands (Duffet et al., 2022a;Ino et al., 2022) revealed that the conformational activation kinetics of these receptor types is at least an order of magnitude slower than what has been reported for monoamine receptors (Feng et al., 2019;Patriarchi et al., 2018;Sun et al., 2018;Wan et al., 2021), likely reflecting the more complex and polytopic binding mode of peptide ligands to their receptor.
Next, we compared the coupling of GLPLight1 and its parent receptor (WT GLP1R) to downstream signaling. We first measured the agonist-induced membrane recruitment of cytosolic mini-G proteins and β-arrestin-2 using a split nanoluciferase complementation assay (Dixon et al., 2016). In this assay, both the sensor/receptor and the mini-G proteins contains part of a functional luciferase (smBit on the sensor/receptor and LgBit for Mini-G proteins) that becomes active only when these two partners are in close proximity (Wan et al., 2018). In agreement with the known pleiotropic signaling of WT GLP1R (Rowlands et al., 2018), in our assay activation of the receptor led to a strong recruitment of miniGs, miniGq, miniGi, β-arrestin-2 as well as miniG12, albeit to a lower extent. In comparison to WT GLP1R, the coupling of GLPLight1 to all tested signaling partners was significantly reduced (Figure 1-figure supplement 3a-j). To further confirm the absence of coupling to intracellular cyclic-AMP (cAMP) signaling of GLPLight1, we performed a titration of GLP-1 on the sensor and WT GLP1R in a luminescence-based cAMP assay. This revealed that the WT GLP1R showed could potently elicit intracellular cAMP increases with an EC50 of 8.0 pM whereas no such increase was observed for GLPLight1 even at the highest GLP-1 concentrations tested (100 nM, Figure 1-figure supplement 3k).
We also performed a titration of GLP-1 induced recruitment of miniGs protein where we could show that GLP1R effectively recruits miniGs proteins with an EC50 of 3.8 nM (Figure 1-figure supplement 3l).
These results indicate that GLPLight1 is unlikely to couple with endogenous intracellular signaling pathways.

Application of GLPLight1 as a tool for pharmacological screening
GLPLight1 is a novel genetically-encoded sensor capable of providing a sensitive intensiometric readout of hmGLP1R activation in response to its endogenous ligands. As such, this tool could have great potential for applications in the drug discovery and development field; however, a more careful characterization of its pharmacological response profile is needed to ensure its implementation as a screening tool. We thus performed a series of in vitro pharmacological experiments in which we characterized GLPLight1 responses under different conditions and with a variety of ligands with known pharmacological effects on GLP1R, with the aim to demonstrate the applicability of this sensor as a pharmacological screening tool. We started by testing the reversibility of sensor response via competition of GLP-1 with an antagonist peptide. To do so, we imaged GLPLight1-expressing HEK293T cells upon addition, in sequence, of 1.0 µM GLP-1 followed by 10 µM Exendin-9 (Ex-9), a well-known peptide antagonist of GLP1R. Ex-9 could partially reverse the signal to 42% of the maximal GLP-1 6 response, within less than 5 minutes in vitro (Figure 2a-b). Next, we tested whether two clinically-used anti-obesity drugs that are known GLP1RAs, liraglutide or semaglutide (O'Neil et al., 2018), could trigger a response from the sensor. As expected, GLPLight1 responded to both GLP1RAs with almost maximal activation, on par with GLP1 (Figure 2a). These results indicate that GLPLight1 can serve as a direct readout of pharmacological drug action on the hmGLP1R with higher temporal resolution than previously available approaches, such as downstream signaling assays (Zhang et al., 2020).
Knowing that GLP-1 is produced along with GLP-2 and glucagon via proteolytic processing of a common preproglucagon precursor protein (Drucker, 2001), we decided to investigate the specificity of our sensor against these other peptides. While the sensor did not respond with any detectable increase in fluorescence to GLP-2, it responded to glucagon with a ΔF/F0 of 324% (61% of maximal response to GLP-1). To further characterize the sensitivity of GLPLight1 to its two endogenous agonists, we performed titrations of GLP-1 and glucagon in HEK293T cells and determined that GLPLight1 had a 94-fold higher affinity for GLP-1 compared to glucagon (EC50 = 28 nM for GLP-1, EC50 = 2.6 µM for glucagon), in agreement with previously-reported results employing a downstream cAMP readout (Runge et al., 2003). Furthermore, the affinity of GLP-1 measured in primary neurons (EC50 = 9.3 nM) was comparable to the one in HEK cells (Figure 2c). Additionally, GLPLight1 did not respond to a panel of other endogenous class B1 GPCR peptide ligands that were tested at high concentration (1.0 µM), including GIP, CRF, PTH, PACAP or VIP.
The binding of GLP-1 to its receptor occurs via the N-terminus of the peptide, as demonstrated by previous structural (Jazayeri et al., 2017) and mutagenesis studies (Longwell et al., 2021;Zhang et al., 2020). We therefore set out to determine whether the general trends observed by fluorescence response of GLPLight1 is in agreement with the pharmacological readout of GLP1R activation obtained using classical assays (Adelhorst et al., 1994). We synthesized four single-residue alanine mutants of GLP-1 at selected N-terminal positions (H1A, E3A, G4A, T5A) using automated fast-flow peptide synthesis (AFPS, see Supplementary Information) (Hartrampf et al., 2020;Mijalis et al., 2017). All peptides were obtained in good yields and excellent purities after RP-HPLC purification. Titrations of individual GLP-1 mutants on GLPLight1-expressing cells revealed clear effects of the mutations on either the maximal sensor response (Emax), the potency (EC50) of the peptide ligand, or both ( Table 1 and Figure 2d). In particular, the critical role of H1 and G4 for both binding and activation has been reported in the literature several times (Manandhar and Ahn, 2015). In agreement with these results, we observed a significant reduction of Emax and EC50 for H1A (56% and 1300 nM, respectively) ( Table   1, Entry a) and G4A (14% and 993 nM, respectively) ( Table 1, Entry c), compared to WT GLP-1 using GLP1Light as a readout. Furthermore, position E3 was reported to be critical for binding, but not for activation. Here, we determined an Emax of 96% compared to WT GLP-1 ( Table 1, Entry b), as well as a reduced EC50 (757 nM) for E3A, which is in agreement with the literature (Manandhar and Ahn, 2015) ( Table 1). Finally, T5 has been reported as less important for GLP1R binding and activation than the other investigated peptide positions (Adelhorst et al., 1994). Accordingly, our experiments with GLPLight1 T5A showed the highest Emax (100%) and EC50 (188 nM) ( State-of-the-art techniques for detecting endogenous GLP-1 or glucagon release in vitro from cultured cells or tissues consist of costly and time-consuming antibody-based assays (Kuhre et al., 2016) or analytical chemistry procedures (Amao et al., 2015). Given the genetically-encoded nature and the fast optical readout of GLPLight1, this tool has the potential to facilitate studies investigating the physiological regulation of GLP-1 release in vitro. To establish whether GLPLight1 could be sensitive enough to detect endogenous GLP-1 release in an in vitro setting, we cultured sensor-expressing HEK293T cells in the presence or absence of a GLP-1/glucagon-producing immortalized enteroendocrine cell line (GLUTag cells (Brubaker et al., 1998)). To distinguish the two cell types in the co-culture system, the HEK239T cells were co-transfected with a cytosolic red fluorescent protein 8 (mKate2). To detect whether the GLPLight-expressing cells had detected endogenous GLP-1 release by the enterocrine cells, we bath-applied GLP-1 to cause full activation of the sensor. We observed that the response to GLP-1 of sensor-expressing cells cultured in the presence of GLUTag cells was significantly lower than that of cells cultured in their absence (Figure 2-figure supplement 1). These results indicate that GLPLight1 was partially pre-activated by endogenous GLP-1 secreted by the enterocrine cells present in the dish. The detection of endogenous GLP-1 by the sensor opens the possibility to use it as a screening tool for studying intrinsic/extrinsic factors that regulate GLP-1 release from enterocrine cells in vitro.

Development and in vitro characterization of photo-GLP1
To investigate the spatiotemporal activation of GLP1R and GLPLight1, a photocaged derivative of GLP-1 was envisioned. To ensure that the photocaged GLP-1 derivative does not activate GLP1R or GLPLight1 prior to uncaging (i.e. in the dark), the photocage must be located on or near GLP-1 regions that are essential for binding. Photocaging of peptides can be achieved by the attachment of a photocaging molecule at a side-chain functionality, backbone amide, or at the C-or N-terminus of the peptide. Recently, we reported the optical control of orexin-B using a UV-visible light-sensitive C-terminal photocage (Duffet et al., 2022b). As opposed to orexin-B, GLP-1 primarily binds via its Nterminus to GLP1R (Jazayeri et al., 2017). We therefore explored an N-terminal caging strategy to generate a photocaged GLP-1 derivative (photo-GLP1, Figure 3a). GLP-1 was prepared by solid-phase peptide synthesis (SPPS) utilizing AFPS (Hartrampf et al., 2020;Mijalis et al., 2017). Before cleavage of the peptide from the resin, photocaging of the GLP-1 N-terminal amine was carried out by treating We then leveraged on GLPLight1 to establish an all-optical assay for characterizing photo-GLP1 uncaging in vitro. We bath-applied photo-GLP1 (10 µM) onto GLPLight1-expressing HEK293T cells and performed optical uncaging by exposing a defined area directly next to the cells to 405 nm laser light (UV light) for defined periods of time, while the sensor fluorescence was imaged using 488 nm laser light. Application of photo-GLP1 by itself failed to trigger any response from GLPLight1, indicating a lack of functional activity in the absence of UV light (Figure3-figure supplement 2a). On the contrary, after photo-GLP1 was added to the bath, the fluorescence of GLPLight1 visibly increased upon 10 seconds of UV light exposure, indicating that GLP-1 could successfully be uncaged and activated the sensor on the cells. Higher durations of UV light exposure led to a higher degree of GLPLight1 responses, and the maximal uncaging duration tested (100 sec) triggered approximately 30% of the maximal response of the sensor, as assessed in the same assay by bath-application of a saturating GLP-1 concentration (10 µM) (Figure 3b-c). Importantly, to show that the sensor signals are not due to UV light-induced artifacts, we reproduced the maximal (100 sec) uncaging protocol on GLPLight-ctrexpressing HEK293T cells and confirmed that in this case no sensor response could be observed.
Furthermore, pre-treatment of the cells with the GLP1R antagonist Ex-9 significantly blunted the sensor response evoked by the optical uncaging (100 sec) (Figure 3c, Figure3-figure supplement 2b). These results indicate that photo-GLP1 can be effectively uncaged in vitro using 405 nm light to control hmGLP1R activation.

High-resolution all-optical visualization and control of GLP1R activity
Upon performing the uncaging experiments, we noticed that the profile of the sensor response to bathapplied GLP-1 differed, depending on whether or not photo-GLP1 was present in the bath. To investigate this phenomenon more in detail, we measured and compared the sensor activation kinetics when GLPLight1 was activated by direct bath application of GLP-1 in the presence or absence of an equimolar concentration of photo-GLP1 in the bath. The sensor response was strikingly different in the two conditions, and exhibited an approximate 14-fold reduction in the speed of activation in the presence of photo-GLP1 (τON without photo-GLP1 = 4,7 sec; τON with photo-GLP1 = 68,1 sec; Figure   3d-e). These results indicate that photo-GLP1, in the dark (i.e. with an intact photocage), can affect the kinetics of GLP1R activation, and this is likely mediated by its binding to the receptor extracellular domain (ECD), which competes for the functionally-active GLP-1. In fact, since the GLP1R belongs to class-B1 GPCRs, the binding of GLP-1 is known to involve an initial step where the peptide C-terminus is recruited to the ECD, followed by a second step involving insertion of the peptide N-terminus into the receptor binding pocket (Wu et al., 2020). Given that our photocage was placed at the very N-terminus of photo-GLP1, our results show that this caging approach prevents the peptide's ability to activate GLPLight1 but, at the same time, preserves its ability to interact with the ECD.
We next asked whether we could leverage GLPLight1 to obtain spatial information on the extent of GLP1R activation in response to photo-GLP1 uncaging. To do so, we performed optical photo-GLP1 uncaging on three separate areas of about 400 µm 2 placed at different locations in a large field of view (FOV, approximately 40,000 µm 2 ). UV light was applied for a total of 40 seconds on the three uncaging regions during the imaging session. GLPLight1 shows a fluorescent response in all three uncaged areas, while its fluorescence remained unaltered throughout the rest of the FOV, indicating high spatial localization of the response to GLP-1 (Figure 3f). As a control, the omission of photo-GLP1 in the cell bath led to no sensor response upon uncaging (Figure 3g). Additionally, the same session was repeated on GLPLight-ctr-expressing cells. Also in this case, no response to uncaging could be observed (Figure 3f). To determine whether the sensor readout in this assay could report GLP1R activation with even subcellular resolution, we repeated the uncaging experiment by selecting an uncaging area of approximately 16 µm 2 directly on a cell membrane. In this case, the application of UV light led to localized activation of GLPLight1 that was limited to a portion of the cell membrane and did not spread to neighboring cells (Figure 3h). These results demonstrate that the optical nature of the GLPLight1 readout makes it possible to determine the spatial extent of GLP1R activation with very spatial high-resolution, down to sub-cellular domains.
Finally, we tested whether uncaging of photo-GLP1 could be used to control functional signaling downstream of hmGLP1R activation. To this aim, we employed a recently developed genetically encoded sensor for cyclic-AMP (G-Flamp1) (Wang et al., 2022), which is the main second messenger involved in cellular signaling downstream of GLP1R activation (Holz et al., 2015). We imaged a field of due to temperature or pressure changes onto the cells. Overall, our results demonstrate that uncaging of photo-GLP1 can be used to achieve optical control of GLP1R signaling activation with high spatiotemporal resolution.

Discussion
Here, we report the first genetically encoded sensor engineered based on cpGFP and the human GLP1R. We show that this tool can directly report ligand-induced conformational activation of this receptor with the high sensitivity and spatiotemporal resolution typical of GPCR-based sensors. Using this new probe, we found that ligand-induced conformational activation of the human GLP1R occurs on slower timescales compared to the reported kinetics of other similarly-built GPCR-sensors (Labouesse and Patriarchi, 2021). This new insight is not surprising given that previously developed sensors were built from class-A GPCRs (Labouesse and Patriarchi, 2021), while GLP1R belongs to a different class of GPCRs (class B1) that is characterized by a distinct ligand-binding mechanism that involved initial ligand 'capture' by the receptor's ECD, followed by ligand insertion into the receptor binding pocket for initiating the transduction of signaling (Zhang et al., 2020). As a reference, other previouslycharacterized class-A GPCR-based neuropeptide biosensors showed sub-second activation kinetics (Duffet et al., 2022a;Ino et al., 2022). Accordingly, our observations show that the receptor activation kinetics can be largely influenced by pre-incubation with an inactive form of the GLP-1 peptide (photo-GLP1), likely because the inactive peptide interacts with and occupies the receptor's ECD.
We showcased the sensitivity and utility of GLPLight1 as a pharmacological tool to aid drug screening and development efforts by characterizing its response to various naturally occurring peptide ligands, as well as clinically-used agonists and peptide-derivatives with diverse pharmacological actions on GLP1R. Besides its applications in pharmacology and drug discovery, given the high sensitivity and lack of interference with intracellular signaling of GLPLight1 it might be possible to employ this tool to investigate the dynamics of endogenous GLP-1 and/or glucagon directly in living systems (in vivo), although based on the evidence provided in this study the in vivo utilization of the sensor is not guaranteed to succeed.
The apparent EC50 of GLPLight1 fluorescence response to GLP-1 is very similar to that measured for mini-Gs recruitment to the hmGLP1R, while it is approximately three orders of magnitude lower than that of the cAMP response downstream of hmGLP1R. This discrepancy might be due to intrinsic differences of the assays used or to intrinsic differences in the distinct aspects of the signaling pathway investigated (i.e. direct recruitment of mini-Gs versus enzymatically-amplified cAMP signals). This raises the interesting possibility that under physiological conditions GLP-1 might elicit different functional responses based on the location of its action and on the spatial concentration gradients on target cells/tissues.
Given that GLPLight1 produces a fluorescence readout that is more representative in terms of sensitivity to that measured by direct recruitment of mini-Gs proteins to the hmGLP1R, the characteristics of this sensor appear not suitable to detect the concentration range achieved by GLP-1 in the periphery through endocrine signaling (picomolar levels). Nevertheless, it is conceivable that under specific circumstances, for example in specific brain areas or in close proximity to enteroendocrine cells in the gut, levels of GLP-1 release might reach high-enough levels that could be detected by GLPLight1.
Future studies could attempt in vivo use of the sensor to further explore this interesting direction, for example by leveraging on AAV-mediated expression of GLPLight1 in living tissues or animals for implementing its use through in vivo imaging techniques, such as fiber photometry (Gunaydin et al., 2014), mesoscopy (Cardin et al., 2020) or two-photon microscopy (Helmchen, 2009). Through such efforts, GLPLight1 might be helpful to shine new light on the hidden mechanisms of GLP-1 and/or glucagon release dynamics in relation to physiological or pathological conditions.
Finally, we leveraged GLPLight1 to characterize the uncaging of the photocaged GLP-1 derivative (photo-GLP1) described for the first time in this work. Optical tools to selectively activate GLP1R could contribute to mechanistic studies Frank et al., 2018), and the photoswitchable GLP-1 LirAzo was recently used to optically control insulin secretion and cell survival (Broichhagen et al., 2015). As opposed to photoswitchable peptides, in which the side chain or part of the peptide backbone is replaced by a photoswitchable moiety such as an azobenzene, photo-GLP1 releases native GLP-1 upon optical uncaging. A drawback of a photocaging strategy, on the other hand, is that it is an irreversible transformation, unlike photoswitchable derivatives. By deploying GLPLight1 and photo-GLP1 in concert in an all-optical assay, we determined that the spatial spread of GLP1R activation in response to GLP-1 release can be localized to single-cells or even sub-cellular domains. Furthermore, by combining a state-of-the-art cAMP sensor with photo-GLP1, we demonstrated the optical control of hmGLP1R-dependent downstream cellular signaling with single-cell resolution, opening exciting new opportunities for investigating the spatial regulation of this signaling pathway. Since we photocaged native GLP1, it is important to note, that the photo-GLP1 might still be susceptible to DPPIV-mediated degradation when used in in vivo applications. We envisage that our photo-GLP1 will nonetheless find applications in neurobiological in vivo studies in brain tissue, as DPPIV-levels in the brain are significantly lower than in peripheral organs.
In summary, we developed and utilized a new all-optical toolkit to unveil a previously inaccessible spatial dimension of the GLP-1/GLP1R system. These tools may thus be readily implemented in a variety of applications, some of which are showcased as part of this study, to advance our understanding of the roles of GLP-1/glucagon/GLP1R signaling system in physiology, or to foster the drug screening and development process targeting the GLP1R pathway.
14 DATA AVAILABILITY DNA plasmids used for viral production have been deposited both on the UZH Viral Vector Facility (https://vvf.ethz.ch/) and on AddGene (plasmid numbers: 187466-187468). Plasmids and viral vectors can be obtained either from the Patriarchi laboratory, the UZH Viral Vector Facility, or AddGene.
Source data are provided with the manuscript.

ACKNOWLEDGEMENTS
The results are part of a project that has received funding from the European Research Council (

COMPETING FINANCIAL INTERESTS
T.P. is a co-inventor on a patent application related to the sensor technology described in this article.
All other authors have nothing to disclose.

Structural modelling
The structural model of GLPLight1 was obtained using ColabFold (Mirdita et al., 2022) using pdb70 as a template mode. The best prediction was selected manually and edited using Chimera.

Peptide synthesis and biochemical characterization
GLP-1, photo-GLP1 and all alanine scan peptides were synthesized on an automated fast-flow peptide synthesizer (AFPS) using a recently developed protocol (Hartrampf et al., 2020). A detailed description of the synthetic procedures and all analytical data can be found in the Supplementary Information.

Cell culture, imaging and quantification
Mammalian HEK293T  another 200 cycles (approx. 920 sec). The ∆R/R0 values were calculated by dividing the raw luminescence intensities after GLP-1(7-37) addition by the ones after vehicle addition. This ratio was then normalized using the average luminescence intensity before addition as a baseline for both GLPLight1 and GLP1R conditions. The quantification of the maximal recruitment was calculated using the average ∆R/R0 between t = 600 sec and t = 700 sec for the time-lapses and t = 1600 sec and t = 1800 sec for the GLP-1 titration of miniGs recruitment to GLP1R.

Flow cytometry
After transfection, HEK293T cells were harvested using Versene ® . After resuspension in FACS buffer

Animals
Animal procedures were performed in accordance to the guidelines of the European Community Council

Synthesis and Characterization of Building Blocks and Peptides 1 Materials and Equipment
Unless otherwise noted, all reactions were performed under an atmosphere of inert gas (N2). The reactions were carried out in oven-dried glassware using dry solvents, unless otherwise noted. Room temperature is defined as a range between 20-25 °C.

Chemicals and Solvents
All chemicals and solvents were used as supplied, unless otherwise stated. Fmoc-and side chain-protected

Chromatography
Analytical thin-layer chromatography (TLC): Merck TLC plates (silica gel 60) on glass with the indicated solvent system. TLC spots were visualized by UV light (254 nm) and by stains of KMnO4 or Ninhydrin.
Flash column chromatography was performed using Merck silica gel 60 (40−63 μm particle size) with the indicated solvent system.

Analytical Ultra High-Performance Liquid Chromatography (UHPLC)
For determination of purity by UHPLC, the filtered peptide solution was diluted in 10-50% acetonitrile Under the Curve (AUC) of desired product peak (detected at λ = 214 nm) as a percentage of the AUC of all peaks between the indicated timeframe.

Solid-Phase Automated Fast-flow Peptide Synthesis (AFPS)
Peptides were synthesized on an automated-flow system, which was built in the Hartrampf lab, based on the published AFPS system. [1] All peptides were prepared by AFPS on NovaPEG Rink Amide resin (0.41 mmol/g or 0.20 mmol/g, as specified) to afford C-terminally amidated peptides. All peptides were synthesized from the C-to N-terminus. Standard Fmoc/tBu protected amino acids ( the N α -Fmoc group was achieved using 20% piperidine with 1% formic acid in DMF at a flow rate of 20 mL/min at 90 °C for 19 sec. Between each coupling and deprotection step, the resin was washed with DMF (32 mL) at 90 °C with a flow rate of 40 mL/min.

Cleavage of Peptidyl-Resins
After synthesis, the peptidyl-resin was washed with DCM (3 × 5 mL) and dried under reduced pressure. To the resin (amount as specified) was added cleavage solution (1.5-3.0 mL, TFA/TIPS/DODT/H2O 94/1/2.5/2.5, v/v/v/v) and the reaction was allowed to proceed at room temperature for 2 h. Subsequently, the supernatant was collected by filtration and concentrated under a light stream of N2. Thereafter, ice-cold diethyl ether (14 mL) was added to the concentrated supernatant, and the resulting precipitate collected as a pellet by centrifugation. The pellet was then suspended in a second portion of ice-cold diethyl ether (14 mL) and collected again by centrifugation. Residual ether was allowed to evaporate, and the peptide pellet was dissolved in 10-50% acetonitrile in water containing 0.1% TFA and lyophilized. Photocaged peptides were protected from light during all steps and handling.

Semi-Preparative Reverse-Phase High-Performance Liquid Chromatography (RP-HPLC)
All peptides were purified by semi-preparative RP-HPLC on a Shimadzu prominence HPLC system (Shimadzu Corp., Japan), using an Agilent Zorbax 300SB-C18 or 300SB-C8 Semi-Preparative column (9.4 × 250 mm, 5 µm particle size), wherein eluent A = H2O with 0.1% TFA, and eluent B = MeCN with 0.1% TFA, and with UV absorbance detection at λ = 214 nm. Purifications were carried out at room temperature with a flow rate of 4.0 mL/min or 3.5 mL/min using linear gradients of eluents A and B as specified.
Fractions (automatically collected) were then analyzed for purity by LCMS and UHPLC, and fractions containing the desired product (≥95% purity) were pooled and lyophilized.

Yield Calculations
For synthesis of peptides herein, a pre-loaded Novabiochem® NovaPEG Rink Amide resin (0.41 mmol/g or 0.20 mmol/g) was used. Theoretical yield was determined based on weight of the resin, resin loading, and the molecular weight of each purified peptide.