Report Conformational Sensors and Domain Swapping Reveal Structural and Functional Differences between b-Arrestin Isoforms Graphical Abstract Authors Eshan Ghosh, Hemlata Dwivedi, Mithu Baidya, ..., Somnath Dutta, Jana Selent, Arun K. Shukla Correspondence arshukla@iitk.ac.in In Brief Ghosh et al. discover structural differences between b-arrestin isoforms (b-arrestin 1 and 2), which are universal regulators of signaling and trafficking of G-protein-coupled receptors (GPCRs). These findings have direct implications for understanding the regulatory and signaling paradigms of GPCRs and designing novel therapeutics targeting this important class of receptors. Highlights d b-arrestin 1 and 2 differentially regulate signaling and trafficking of GPCRs d There are conformational differences between receptor- bound b-arrestin 1 and 2 d Domain-swapped chimera of b-arrestin 1 and 2 exhibit functional complementation d Functional differences of b-arrestin isoforms have implications for biased agonism Ghosh et al., 2019, Cell Reports 28, 3287–3299 September 24, 2019 ª 2019 The Author(s). https://doi.org/10.1016/j.celrep.2019.08.053 Cell Reports Report Conformational Sensors and Domain Swapping Reveal Structural and Functional Differences between b-Arrestin Isoforms Eshan Ghosh,1 Hemlata Dwivedi,1 Mithu Baidya,1 Ashish Srivastava,1 Punita Kumari,1 Tomek Stepniewski,2 Hee Ryung Kim,3 Mi-Hye Lee,4 Jaana van Gastel,5,6 Madhu Chaturvedi,1 Debarati Roy,1 Shubhi Pandey,1 Jagannath Maharana,1 Ramon Guixà-González,7 Louis M. Luttrell,4,8 Ka Young Chung,3 Somnath Dutta,9 Jana Selent,2 and Arun K. Shukla1,10,* 1Department of Biological Sciences and Bioengineering, Indian Institute of Technology, Kanpur 208016, India 2Research Programme on Biomedical Informatics (GRIB), Department of Experimental and Health Sciences of Pompeu Fabra University (UPF)-Hospital del Mar Medical Research Institute (IMIM), 08003 Barcelona, Spain 3School of Pharmacy, Sungkyunkwan University, 2066 Seobu-ro, Jangan-gu, Suwon 16419, South Korea 4Department of Medicine, Medical University of South Carolina, Charleston, SC 29425, USA 5Translational Neurobiology Group, Center of Molecular Neurology, VIB, Antwerp, Belgium 6Receptor Biology Lab, Department of Biomedical Sciences, University of Antwerp, Antwerp, Belgium 7Laboratory of Computational Medicine, Biostatistics Unit, Faculty of Medicine, Autonomous University of Barcelona, 08193 Bellaterra, Spain 8Research Service of the Ralph H. Johnson Veterans Affairs Medical Center, Charleston, SC 29401, USA 9Molecular Biophysics Unit, Indian Institute of Sciences, Bangalore, India 10Lead Contact *Correspondence: arshukla@iitk.ac.in https://doi.org/10.1016/j.celrep.2019.08.053 SUMMARY gers and downstream signaling. Subsequently, they are phos- phorylated in their carboxyl-terminus and intracellular loops, Desensitization, signaling, and trafficking of G-pro- which then results in coupling of b-arrestins (barrs). Binding of tein-coupled receptors (GPCRs) are critically regu- barrs interferes with further coupling of G proteins through steric lated by multifunctional adaptor proteins, b-arrestins hindrance, at least at the plasma membrane, and leads to recep- (barrs). The two isoforms of barrs (barr1 and 2) share tor desensitization. Interestingly, barrs also serve as adaptors for a high degree of sequence and structural similarity; the components of clathrin machinery to mediate receptor endo- cytosis (Goodman et al., 1996; Laporte et al., 1999), and they can still, however, they often mediate distinct functional also scaffold various kinases to initiate several signaling path- outcomes in the context of GPCR signaling and regu- ways (DeWire et al., 2007; Luttrell et al., 1999, 2001; McDonald lation. A mechanistic basis for such a functional et al., 2000). divergence of barr isoforms is still lacking. By using The two isoforms of barrs, referred to as barr1 and 2 (also a set of complementary approaches, including anti- known as arrestin 2 and 3, respectively) share a high level of body-fragment-based conformational sensors, we sequence identity and both display an overall similar structure discover structural differences between barr1 and 2 (Gurevich and Gurevich, 2015). For most GPCRs, both isoforms upon their interaction with activated and phosphory- of barrs are typically recruited upon agonist stimulation and lated receptors. Interestingly, domain-swapped chi- participate in desensitization, endocytosis, and signaling. meras of barrs display robust complementation in Emerging data suggest that a significant level of functional diver- functional assays, thereby linking the structural dif- gence between barr1 and 2 exits for most GPCRs, and in some cases, they even display functional antagonism (Figure 1A) (Sri- ferences between receptor-bound barr1 and 2 with vastava et al., 2015; Hara et al., 2011; Ahn et al., 2003). Interest- their divergent functional outcomes. Our findings ingly, the functional differences of barr isoforms is also mani- reveal important insights into the ability of barr iso- fested at the level of physiological outcomes downstream of forms to drive distinct functional outcomes and un- several GPCRs (Srivastava et al., 2015; Walters et al., 2009; Triv- derscore the importance of integrating this aspect edi et al., 2013) in the current framework of biased agonism. The functional divergence of barr isoforms has direct implications for the conceptual framework of barr-dependent INTRODUCTION signaling, biased-agonism, and, in particular, for the develop- ment of barr-biased ligands as novel GPCR therapeutics G-protein-coupled receptors (GPCRs) constitute a large family (Shukla et al., 2011). Still, however, the mechanistic basis of integral membrane proteins in the human genome (Bjarnadót- of this phenomenon is currently lacking, and it represents a tir et al., 2006) and a major class of drug targets (Santos et al., key knowledge gap in our understanding of the GPCR-barr 2017). Upon activation by agonists, GPCRs couple to heterotri- interaction and signaling. The receptor-barr interaction is typi- meric G proteins followed by the generation of second messen- cally biphasic, involving the phosphorylated carboxyl-terminus Cell Reports 28, 3287–3299, September 24, 2019 ª 2019 The Author(s). 3287 This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Figure 1. Overall Structural and Conforma- tional Similarity between V2Rpp-Bound barr1 and 2 (A) Most GPCRs typically recruit both isoforms of barrs, i.e., barr1 and 2, which mediate some overlapping functions, but they also display dif- ferential contributions in receptor desensitization, trafficking, and signaling outcomes. (B) Crystal structure of V2Rpp-barr1 reveals that Lys/Arg residues on the N-domain of barr1 are aligned into a spatial groove (indicated by dotted red lines) and V2Rpp docks into this groove. Cor- responding Lys/Arg residues in barr2, which are conserved at the primary sequence level, also exhibit a similar spatial arrangement as visualized using the crystal structure of barr2 (PDB: 3P2D). (C) Schematic representation of amino acids in barr1 that are within the interacting distance of V2Rpp based on previously determined crystal structure (PDB: 4JQI). The coordinates of the amino acid residues involved in binding with V2Rpp were submitted into PDBSum, and the interactions were mapped as a simplified ladder. Corre- sponding residues in barr2 are listed to highlight the conserved nature of V2Rpp-interacting resi- dues in barr1 and 2. Lys/Arg residues, which form ionic interactions with the phosphate groups on V2Rpp, are highlighted in red. Sep and Tpo repre- sent phosphorylated serine and threonine residues in V2Rpp, respectively. Dotted blue and orange lines denote hydrogen bonding and non-bonded contacts, respectively. The residues have been colored according to their types, viz., blue, posi- tive; red, negative; green, neutral; purple, aro- matic; gray, aliphatic; orange, Pro and Gly; and yellow, cysteine. (D) Fab30 recognizes V2Rpp-bound barr1 and 2 to a comparable level as measured using an ELISA- based approach described in the STAR Methods section. Data represent mean ± SEM of three in- dependent experiments, with each carried out in duplicate and normalized with respect to the maximum signal obtained for V2Rpp-barr1 condi- tion (treated as 100%). (E) Similar reactivity of Fab30 toward V2Rpp- bound barr1 and 2 is further confirmed by a coIP experiment. A representative image of three independent ex- periments is shown here, and densitometry-based quantification of all three experiments is presented in Figure S3A. See also Figure S1. (i.e., receptor tail) and the transmembrane bundle (i.e., receptor Accordingly, we set out to probe the conformations of core) (Gurevich and Gurevich, 2004; Shukla et al., 2014). These receptor-bound barrs by using a battery of complementary two sets of interactions result in the formation of partially approaches, including biochemical and functional assays, syn- engaged (i.e., tail engaged) and fully engaged (i.e., core thetic antibody-based conformational biosensors, single-parti- engaged) receptor-barr complexes, respectively. Recent cle electron microscopy, bimane fluorescence spectroscopy, studies have suggested that distinct functional outcomes are and molecular dynamics simulation. We discover potential struc- associated with these two conformations of receptor-barr com- tural differences between receptor-bound barr1 and 2 and iden- plexes (Kumari et al., 2016; Cahill et al., 2017; Kumari et al., tify key regions in barrs that are critical for imparting these differ- 2017; Sente et al., 2018). Thus, we envisioned that key determi- ences. By using a domain-swapped chimera of barrs, we also nants of the functional divergence of barr isoforms may be en- observe that the structural differences between barr1 and 2 man- coded at the level of structural and conformational differences ifest in their distinct functional contributions downstream of between receptor-bound barrs. GPCRs. Our findings provide important insights into the 3288 Cell Reports 28, 3287–3299, September 24, 2019 GPCR-barr interaction, and they have direct implications for Fab30 Reactivity Suggests Potential Conformational refining the framework of biased agonism at GPCRs. Differences between Receptor-Bound barr1 and 2 We next set out to measure the reactivity of Fab30 toward acti- vated and phosphorylated receptor-bound barr1 and 2. Here, we RESULTS used two different GPCRs, the V2R and a chimeric b2-adrenergic receptor, referred to as b2V2R, where the carboxyl-terminus of Sequence and Structural Analysis of barrs the b2AR is replaced with that of the V2R (Oakley et al., 2000; The crystal structure of barr1 in complex with a phosphopeptide Thomsen et al., 2016). Although we have previously reported corresponding to the carboxyl-terminus of the human vaso- that Fab30 robustly recognizes receptor-barr1 complexes (Shu- pressin receptor (V2R; referred to as V2Rpp) has revealed the kla et al., 2014; Kumari et al., 2016), in order to further establish interaction interface between receptor-attached phosphate Fab30 as a reliable sensor of receptor-bound barr conformation, groups and positively charged residues in the N-domain of we measured its reactivity toward b2V2R-barr1 complexes barr1 (Shukla et al., 2013). Here, V2Rpp serves as a surrogate formed in response to a set of ligands having different efficacies for the phosphorylated receptor tail, and therefore, the V2Rpp- ranging from inverse agonists, partial agonists, to full agonists. barr1 complex represents a close proxy of the partially engaged We observed an excellent correlation between Fab30 reactivity receptor-barr1 complex. Our analysis of the phosphate-interact- (measured using coIP assay) and the relative efficacy of ligands, ing residues on barr1 in this crystal structure, and their spatial as measured by their cAMP response (Figures S4A–S4E). This surface mapping, revealed a groove along the N-domain of observation underlines the ability of Fab30 to report a pharmaco- barr1 that constitutes the docking interface for V2Rpp through logically relevant receptor-bound barr1 conformation and, there- a number of charge-charge interactions (Figure 1B, top panel). fore, allows us to compare the conformation of receptor-bound As V2Rpp binds both barr1 and 2 with comparable affinities (No- barr1 and 2. bles et al., 2007; Xiao et al., 2004), we analyzed the sequence We used two parallel approaches based on ELISA and coIP and the three-dimensional structure of barr2 to assess whether assays by using activated and phosphorylated b2V2R and V2R the spatial arrangement of phosphate-interacting residues may (Kumari et al., 2016, 2017). As expected, Fab30 robustly recog- be conserved in both isoforms. Indeed, we observed not only nized receptor-bound barr1; surprisingly however, it failed to that phosphate-interacting residues are highly conserved in recognize receptor-bound barr2 for both the b2V2R (Figures barr2 but also that their spatial distribution on the N-domain of 2A, 2B, and S5A) and the V2R (Figures S5B and S5C). ScFv30 barr2 forms a groove identical to barr1 (Figures 1B, bottom also exhibits a pattern identical to Fab30, i.e., it recognizes panel, 1C, and S1). This observation suggests a potentially receptor-bound barr1 but not barr2 (Figures S5D and S5E). As similar binding mechanism of receptor tail with barr isoforms the key residues in barr1 responsible for binding Fab30 are and provides a rationale to probe it experimentally. mostly conserved in barr2, and Fab30 can robustly bind V2Rpp-barr2 complex, the lack of Fab30 reactivity toward Fab30 as a Sensor of barr Activation Reveals Overall receptor-bound barr2 is unlikely to result from differences in its Similarity between V2Rpp-Bound barr1 and 2 interaction interface between barr1 and 2. In agreement with Based on sequence and structural analyses, we conceived that previous studies (Oakley et al., 2000), we also observed that the overall conformation of V2Rpp-bound barr1 and 2 may be b2V2R and V2R robustly interact with barr2 (Figures S6A–S6C), similar, and to probe this, we used a synthetic antibody fragment and therefore, the lack of Fab30 reactivity is also not because (referred to as Fab30) as a sensor of barr conformation that of the inability of b2V2R to bind barr2. preferentially recognizes V2Rpp-bound barr1 (Shukla et al., In order to rule out the affinity difference of Fab30 for barr1 2013, 2014; Kumari et al., 2016, 2017). As the ability of Fab30 versus barr2, we carried out titration coIP experiments, first to recognize barr2 has not been evaluated previously, we first with increasing concentrations of V2Rpp, and second, with identified the paratope residues for Fab30 binding on barr1 increasing concentrations of b2V2R while keeping the concentra- based on the crystal structure of the V2Rpp-barr1-Fab30 com- tions of barr1 and barr2 constant. Fab30 recognizes V2Rpp- plex (Shukla et al., 2013) and confirmed that they are mostly bound barr2 as efficiently as V2Rpp-bound barr1, even at the conserved in barr2 (Figures S2A–S2C). Thus, Fab30 should partial occupancy of V2Rpp (Figures S7A and S7B). Moreover, recognize V2Rpp-bound barr2 as well, and in fact, we observed even at 9-fold higher concentration of b2V2R, we still did not a robust interaction of Fab30 with V2Rpp-bound barr2 at compa- observe any detectable reactivity of Fab30 toward receptor- rable levels to barr1 in two parallel assays based on ELISA and bound barr2 (Figure S7C). These data suggest that the lack of co-immunoprecipitation (coIP) (Figures 1D and 1E; Figure S3A). Fab30 reactivity toward receptor-bound barr2 is not due to an In addition, a single-chain variable fragment version of Fab30, affinity difference of Fab30 or available stoichiometry of phos- referred to as ScFv30, also exhibited a similar pattern of reac- phorylated tail between V2Rpp versus b2V2R experiments. tivity toward V2Rpp-bound barr1 and 2 (Figures S3B and S3C). In order to validate our data in a cellular context, we next Under similar experimental conditions, a control Fab (Fab-CTL) expressed hemagglutinin (HA)-tagged intrabody version of that does not recognize barr1 failed to exhibit any significant Fab30 (referred to as Ib30) together with b2V2R and either specific binding in coIP experiments (Figures S3D and S3E). barr1 or barr2 in HEK293 cells, followed by a coIP experiment. These data suggest that the overall conformations of barr1 and Similar to in vitro experiments performed with purified proteins, 2 in complex with V2Rpp, as detected by Fab30 reactivity, are we found that even in the cellular context, Ib30 recognizes re- similar. ceptor-bound barr1 but not barr2 (Figure S7D). Taken together, Cell Reports 28, 3287–3299, September 24, 2019 3289 (legend on next page) 3290 Cell Reports 28, 3287–3299, September 24, 2019 these findings suggest that there are potential conformational recognition toward receptor-bound barr2 (Figures S8C and differences between barr1 and 2 in complex with activated and S8D). Although the binding epitope of these additional Fabs on phosphorylated receptors, which, in turn, results in the lack of barr1 has not been precisely determined yet, their reactivity Fab30 reactivity toward the receptor-barr2 complex. pattern supports the notion of conformational differences be- tween receptor-bound barr isoforms. Fab30 Reactivity toward barr1 and 2 in the Presence of As an additional line of evidence for the conformational differ- Homogenously Phosphorylated Receptors ences between receptor-bound barr1 and 2, independent of For the experiments mentioned in Figures 2A, 2B, and S5, we Fab30 reactivity as readout, we used bimane fluorescence spec- have utilized in-cellulo-phosphorylated receptors. A potential troscopy. Here, we used purified barr1 and 2 that are bimane concern with respect to Fab30 reactivity may be heterogeneous labeled in their C-loop (barr1245C and barr2246C), which forms a phosphorylation of the receptor carboxyl-terminus when key interface for receptor interaction and exhibits conformational compared to synthetic V2Rpp with a well-defined phosphoryla- rearrangement during receptor interaction (Latorraca et al., tion pattern. To rule out this possibility, we used two parallel 2018) (Figure 3A). We observed a decrease in bimane fluores- approaches. First, we used Sortase enzyme-based chemical cence for barr1 upon its interaction with the receptor, whereas ligation of V2Rpp to truncated b2AR (29-341) in order to generate there was significant increase for barr2 (Figure 3B). These direc- a chimeric b2V2R with a well-defined and homogeneous phos- tionally opposite changes in bimane fluorescence intensities for phorylation pattern identical to that present in V2Rpp, following barr1 and 2 upon their interaction with the receptor suggest that a slightly modified version of a previously published protocol their C-loops are positioned in different environments and pro- (Staus et al., 2018) (Figures 2C and 2D). Interestingly, we vide additional corroborating evidence for their conformational observed that similar to the in-cellulo-phosphorylated receptor, difference in receptor-bound states. We did not observe a signif- this chemically ligated version of the receptor also induces a icant change in bimane fluorescence intensity upon V2Rpp bind- conformation in barr2 that is not recognized by Fab30 (Figures ing, which can be interpreted to reflect conformational similarity 2E and 2F). Second, we generated a series of V2R phosphoryla- between V2Rpp-bound barr1 and 2 with respect to C-loop. tion mutants lacking either the individual phosphorylation sites or cluster of phosphorylation sites and compared the ability of Structural Insights into Conformational Differences Fab30 to recognize receptor-bound barr1 and 2 for these mu- between Receptor-Bound barrs tants. However, we did not observe any significant gain of In order to better understand the Fab30 reactivity pattern and Fab30 reactivity in any of these receptor mutants (Figure S7E). conformational differences between receptor-bound barr1 and Taken together, these data suggest that the lack of Fab30 recog- 2, we used molecular dynamics (MD) simulation to gain struc- nition for receptor-bound barr2 does not arise from heteroge- tural and mechanistic insight. The crystal structure of barr1 in neous or site-specific phosphorylation of the receptor. complex with V2Rpp and Fab30 has revealed a major rotation of the C-domain relative to the N-domain by approximately 20 Corroborating Evidence for Potential Conformation (Shukla et al., 2013). We postulated that this inter-domain rota- Differences between barr1 and 2 tion in barrs may indeed be the primary determinant for effective Leading up to this point, our data based on Fab30 recognition recognition by Fab30, and in order to test this, we performed MD suggest potential conformational difference between receptor- simulations monitoring the stability of Fab30 binding to barr1 bound barr1 and 2. In order to corroborate these findings further, conformers with different inter-domain rotation angles in solution we tested a series of additional Fabs that we have recently (Figures 3C, 3D, and S8E). Here, the Fab30-binding stability is generated and characterized (Ghosh et al., 2017). Similar to assessed as root-mean-square deviation (RMSD) of the back- Fab30, these additional Fabs also interacted comparably with bone atoms along the simulation time. We found that Fab30 V2Rpp-bound barr1 and 2 (Figures S8A and S8B) and efficiently remains stably bound to barr1 conformers with a rotation angle recognized the complex of barr1 with an activated and phos- > 15 (i.e., stable RMSD progression) (Figures 3C and 3D). In phorylated receptor (Figures S8C and S8D). Interestingly contrast, a significant instability of Fab30 interaction is observed however, these additional Fabs also displayed no detectable for barr1 conformers with a rotation angle < 15 (i.e., drastic Figure 2. Fab30 Reactivity Pattern Reveals Potential Conformational Differences between Receptor-Bound barr1 and 2 (A) Fab30 robustly recognizes b2V2R-bound barr1 but not barr2, as assessed by coIP. Here, carazolol (1 mM) and BI-167107 (1 mM) are used as an inverse agonist and agonist, respectively. A representative image from three independent experiments is shown. (B) Densitometry-based quantification of data presented in (A). Values represent mean ± SEM of three independent experiments analyzed using one-way ANOVA with Bonferroni post-test (***p < 0.001). Data are normalized with respect to the agonist-b2V2R-barr1 condition (treated as 100%). (C) Schematic flow-chart of sortase-based chemical ligation of V2Rpp with truncated b2AR (29-341) and subsequent coIP to measure Fab30 reactivity. (D) Efficiency of sortase-based ligation of V2Rpp to b2AR (29–341), as measured by western blotting. A representative blot from two independent experiments is shown. (E) Fab30 fails to recognize barr2 in complex with homogenously phosphorylated b2V2R. After sortase-based chemical ligation of V2Rpp, the resulting b2V2R was incubated with equal concentrations of barr1/2 and Fab30/Fab-CTL followed by coIP. The reactivity of Fab30 with receptor-bound barr1/2 was evaluated by western blot. (F) Densitometry-based quantification of Fab30 reactivity toward receptor-bound barr1 and 2, as measured in (E). Data represent mean ± SEM of three independent experiments, normalized with respect to barr1 (treated as 100%). See also Figures S2, S3, S4, S5, S6, and S7. Cell Reports 28, 3287–3299, September 24, 2019 3291 Figure 3. Fluorescence Spectroscopy and MD Simulation Provide Insights into Conformational Differences between barr Isoforms (A) Schematic representation of the bimane fluorescence spectroscopy experiment where monobromobimane (mBBr) is chemically attached to a cysteine, engineered in the C-loop of barrs. (B) The fluorescence intensity of barr1mBBr decreases upon its interaction with b2V2R, whereas that of barr2mBBr increases significantly. Here, purified barrsmBBr were incubated with purified b2V2R (agonist-bound and phosphorylated) at a molar ratio of barr1:b2V2R (1:3 in a concentration range of 1–5 mM). As a reference, the fluorescence intensity of barrsmBBr alone was measured first and used for normalization (treated as 100%). The inset shows the differences in bimane fluorescence at lmax for receptor-bound barr1 and 2. (C) Fab30-binding stability depends on the inter-domain rotation angle of barr1. (D) The binding stability of the ScFv version of Fab30 (green surface) in complex with barr1 (N-domain, red surface; C-domain, white surface) is measured as RMSD of Fab30 backbone atoms (RMSDFab30) for different activation states of barr1 (inter-domain rotation angles 2.2 , 11.1 , 15.5 , 17.8 , and 20 ). Inter-domain rotation angles 15.5 and 20 in barr1 result in a stable RMSDFab30 progression. Rotation angles of 2.2 and 11.1 provoke a rapid increase of the RMSDFab30 (i.e., binding instability) during the first 50 ns of simulation time due to the clash of the N-domain with Fab30. See also Figure S8E. increase in RMSD) (Figures 3C and 3D). Such instability is not upon activation, and therefore, it is plausible that the inter- surprising because the N-domain of barr1 approaches Fab30 domain rotation in receptor-bound barr2 is significantly different when the inter-domain rotation relaxes toward the inactive (or than barr1, resulting in the lack of recognition by Fab30. basal) state (i.e., rotation angle decreasing from 20 to 0 ), which, MD simulation data presented above raise the possibility that in turn, results in unfavorable contacts and steric clashes (Fig- the core engagement with the receptor may contribute toward ures 3C and 3D). In other words, Fab30 reactivity can be consid- potential conformational differences between receptor-bound ered as readout of the degree of inter-domain rotation in barrs barr1 and 2. In line with this possibility, we observed a relatively 3292 Cell Reports 28, 3287–3299, September 24, 2019 Figure 4. Identification of Structural Regions That Impart Conformational Differences between Receptor-Bound barr1 and 2 (A) Schematic representation of the swap1 construct that harbors the N-domain of barr1 and the C-domain of barr2. (B) Fab30 fails to detect receptor-bound conformation of swap1 as evaluated by coIP, carried out in a similar fashion as in Figure 2A. A representative image of three independent experiments is shown here. (legend continued on next page) Cell Reports 28, 3287–3299, September 24, 2019 3293 greater interaction of barr2 with V2R in the absence of ‘‘tail- conformation of receptor-bound swap1 is similar to barr2, engagement’’ by using a carboxyl-terminus-truncated receptor whereas that of swap2 is similar to receptor-bound barr1. To construct (referred to as V2RDC-term) (Figure S9A). This finding provide additional support for the interaction of swap2 with may be interpreted to suggest that the core interaction for the receptor in a manner similar to barr1, we carried out nega- barr2 is stronger compared to barr1, and therefore, receptor- tive-staining-based single-particle electron microscopy (EM) barr2 complexes may predominantly exist in a fully engaged analysis of a Fab30-stabilized b2V2R-barrswap2 complex (Fig- conformation that is different from that of barr1. This provides ure 4E). This complex also exhibits a biphasic interaction a plausible explanation for the near-complete lack of Fab30 with conformational distribution between the partially reactivity toward receptor-bound barr2. However, in the engaged and fully engaged complexes, similar to what was absence of high-resolution structures of fully engaged recep- previously observed for the b2V2R-barr1-Fab30 complex (Shu- tor-barr complexes, it is not feasible to precisely determine the kla et al., 2014). We also tested the ability of Ib30 to recognize contribution of core engagement toward imparting distinct barr receptor-bound swap2 in a cellular context, and in agreement conformations, and future structural studies are necessary to with the data presented in Figures 4C and 4D, Ib30 robustly illuminate the mechanism of distinct barr conformations upon recognizes receptor-bound swap2, at a level similar to that their interaction with the receptors. of barr1 (Figures 5A and 5B). This observation further sug- gests an overall similar conformation adopted by receptor- Distal C-Domain in barrs Is Important for the Structural bound barr1 and swap2 and provides supporting evidence Differences between the Two Isoforms for the cellular relevance of the data obtained with chimeric To identify the key regions in barrs that are potentially respon- barr constructs in vitro. sible for imparting distinct conformations on barr1 and 2, we generated a series of chimeric barr constructs and tested the A Potential Link between barr Conformations and Their reactivity of Fab30 toward their complexes with the receptor. Distinct Functional Contributions Unlike visual arrestins and barr1, barr2 lacks the c-edge loops To probe whether conformational differences between receptor- (these are different from the C-loop described earlier) that are bound barr isoforms may be directly linked to their functional proposed to anchor the C-domain of visual arrestin to the plasma divergence, we first measured the contribution of barr1 and 2 membrane upon its recruitment to rhodopsin (Lally et al., 2017). in agonist-induced endocytosis, extracellular signal regulated Therefore, we first generated a barr2 construct where we grafted kinase 1/2 (ERK1/2) microtubule-associated protein mitogen- the c-edge loop1 of barr1 in the corresponding position of activated protein (MAP) kinase phosphorylation, and cyclic barr2 and tested the reactivity to Fab30 (Figures S9B and AMP (cAMP) response for the V2R under barr1 or 2 knockdown S9C). However, similar to barr2, this loop-grafted construct conditions (Figures 5C–5F and S10A–S10C). We found that the also failed to exhibit detectable recognition by Fab30 (Figures presence of either barr1 or 2 is capable of supporting agonist- S9B and S9C), suggesting that the lack of c-edge loop1 in induced endocytosis of the V2R (Figure 5C), suggesting that barr2, and thereby, the potential lack of membrane anchoring, the two isoforms are functionally redundant in mediating recep- is not responsible for its conformational difference with barr1. tor endocytosis. Interestingly, agonist-induced ERK1/2 MAP Next, we generated a construct, referred to as swap1, kinase phosphorylation downstream of V2R is sensitive to the harboring the N-domain of barr1 and the C-domain of barr2. depletion of either isoform (Figures 5D and S10C), suggesting Fab30 did not recognize receptor-bound swap1, suggesting that both are involved. Strikingly however, agonist-induced that it adopts a conformation similar to barr2 and that the pri- cAMP response is significantly enhanced upon barr2 depletion mary determinants of distinct conformations of receptor- (Figure 5E), potentially indicating a predominant role of barr2 in bound barr1 and 2 are likely encoded in the C-domain (Figures receptor desensitization compared to barr1. The surface expres- 4A, 4B, and S9D). This is further confirmed by a reverse sion of V2R in barr1- or 2-depleted cells are comparable to each chimera, referred to as swap2, harboring the N-domain of other (Figure S10B). As expected, the exogenous expression of barr2 and the C-domain of barr1 which is effectively recog- barr2 in barr2 knockdown cells lowers the level of cAMP (Fig- nized by Fab30 (Figures 4C, 4D, and S9E). We also generated ure 5F). Most interestingly, the exogenous expression of additional chimeric constructs, referred to as swap3-5, swap1, which is conformationally similar to barr2, also effectively harboring the N-domain of barr2 and different segments of lowers the enhanced level of cAMP in barr2 knockdown cells barr1 C-domain. CoIP experiments revealed that the structural (Figure 5F). Taken together, these data suggest that receptor- determinants of conformational differences between barr iso- bound barr2 is potentially more effective in driving receptor forms, as measured by Fab30 reactivity, primarily reside in the desensitization compared to barr1 and, thus, provide a possible distal C-domain at the primary sequence level (Figures 4C and link between the conformational differences of receptor-bound 4D). The pattern of Fab30 reactivity also indicates that the barrs and their functional divergence. (C) Schematic representation of swap2-5 constructs that harbor N-domain of barr2 and different stretches of the C-domain of barr1. (D) Fab30 reactivity in coIP reveals that the distal C-domain imparts conformational differences between barr1 and 2. (E) 2D class averages derived from single-particle negative-staining EM analysis of the agonist-b2V2R-swap2-Fab30 complex carried out as described in the STAR Methods section. The right panel shows two representative 2D class averages depicting the partially engaged and fully engaged complexes. A schematic representation of these two conformations is presented for the ease of visualization. See also Figure S9. 3294 Cell Reports 28, 3287–3299, September 24, 2019 Figure 5. A Domain-Swapped Chimera of barrs Gains Fab30 Reactivity in Cellular Context and Exhibits Functional Complementation (A) Intrabody 30 (Ib30) efficiently recognizes receptor-bound swap 2 upon agonist stimulation in a cellular context, as assessed by a coIP experiment using the lysate of HEK293 cells expressing FLAG-V2R, barr1, and 2 or swap 2 and Ib30. Cells were stimulated with arginine vasopressin (AVP; 100 nM) for indicated time points, followed by coIP, and a representative image from three independent experiments is shown. (B) Densitometry-based quantification of data presented in (A). Values represent mean signal intensity ± SEM normalized with respect to agonist-V2R-barr1 condition (treated as 100%). (C) Either isoform of barrs is sufficient to mediate agonist-induced endocytosis of V2R measured by whole-cell ELISA in HEK293 cells. Data represent mean ± SEM of five independent experiments, with each carried out in duplicate. Percent endocytosis as measured by the surface level of V2R before and after agonist- stimulation is presented in the graph. (legend continued on next page) Cell Reports 28, 3287–3299, September 24, 2019 3295 DISCUSSION that is not only most diverse between the two isoforms at the primary sequence level but also harbors the interface for several Most GPCRs recruit both barr1 and 2 upon agonist stimulation, interaction partners, such as clathrin, adaptin, and TRAF6. which, in turn, mediate and regulate receptor desensitization, Therefore, it is tempting to speculate that receptor-bound endocytosis, and signaling. Although both isoforms are individu- barr1 and 2 may also differ in their ability to scaffold different ally capable of mediating the above-mentioned functions, inter- partners, owing to their conformational differences. In fact, estingly, they often display a differential contribution toward such a scenario is supported by a global interactomics analysis these functions and subsequent physiological outcomes for where a significant difference between the interactome of barr1 different receptor systems. This paradigm, now observed across and 2 is reported (Xiao et al., 2007). Future studies may shed multiple GPCRs (Srivastava et al., 2015), suggests potential dif- light on this interesting possibility, including a precise mapping ferences at structural and conformational levels in receptor- of barr residues that may be responsible for the differential reac- bound states of barr1 and 2. We observed that although the tivity of Fab30 toward barr isoforms. docking interface for the phosphorylated receptor tail and the A previously published hydrogen/deuterium exchange (HDX) resulting conformations in barr1 and 2 are similar to each other, study of barrs using pre-activated mutants (barr1R169E and they appear to adopt different conformations upon their engage- barr2R170E) provides additional corroborating evidence for our ment with activated and phosphorylated receptors. findings (Yun et al., 2015). A re-analysis of this previous study These findings suggest that a differential conformational rear- reveals a significant difference between the HDX pattern of lariat rangement may happen in barr1 versus 2 when they transition loop peptides when we compared barr1WT-barr1R169E and from the partially engaged to fully engaged complex involving barr2WT-barr2R170E with each other. For example, we observed the receptor core. In other words, the core engagement between a significantly higher rate of deuterium uptake in the 279–289 the receptor and barrs may impart distinct structural changes in segment of barr1 between the wild type (WT) and R169E mutant, the two isoforms of barrs. In fact, a comparison of the crystal but the corresponding region in barr2 does not exhibit a signifi- structures of pre-activated visual-arrestin (i.e., splice variant cant difference between the WT and R170E mutant (Figure S10D). p44) with the rhodopsin-visual-arrestin complex reveals signifi- Although pre-activated mutants may not be a perfect surrogate cant structural changes in visual arrestin (Kang et al., 2015; of fully active barr conformations, the HDX pattern does suggest Zhou et al., 2017; Kim et al., 2013). Moreover, a recent study that the lariat loop region may adopt different conformations for on the rhodopsin-visual-arrestin system has also suggested activated barr1 versus barr2, and it further complements our that both the receptor tail and the receptor core are capable of domain-swapping data described above. inducing activating conformational changes in visual arrestin, in- Our functional data link the conformational differences dependent of each other (Latorraca et al., 2018). Additional between receptor-bound barr1 and 2, as reported by the studies in a cellular context have also reported that the core Fab30-based sensor, with their different contributions in V2R engagement can drive an active conformation in barr2, which al- desensitization. Previous studies have discovered that lows it to enrich in clathrin-coated structures, even in the agonist-induced V2R endocytosis and ERK1/2 MAP kinase absence of a stable complex with the receptor (Eichel et al., phosphorylation can be efficiently supported by partially 2016, 2018). Although these previous studies align with our find- engaged receptor-barr complexes, whereas receptor desensi- ings raising the possibility of core engagement driving the tization is driven primarily by the core-engaged complex (Ku- conformational differences between receptor-bound barr1 and mari et al., 2017). This aligns with the possibility of differential 2, future structural studies should illuminate the structural mech- core engagement for barr1 and 2 leading to their distinct contri- anism underlying this interesting phenomenon. As different bution in receptor desensitization but comparable contribution GPCRs have diverse signatures of phosphorylatable residues in V2R endocytosis and ERK MAP kinase activation. It is also in their carboxyl-terminus or in intracellular loops and may conceivable, however, that an additional level of conforma- have different levels of core engagement, it may not be surprising tional differences in barr isoforms may exist and drive their to discover additional levels of conformational diversity in barrs, functional divergence with respect to endocytosis, signaling, which makes this system precisely tunable in a context-depen- and ubiquitination for other GPCRs. dent manner (Ranjan et al., 2017). Our findings also raise some interesting questions and open It is also intriguing that the differences between barr1 and 2 new avenues for future investigations in this therapeutically appear to arise primarily from the distal C-domain, a region important research area. For example, some GPCRs, such as (D) Both isoforms of barrs contribute to agonist-induced ERK MAP kinase activation downstream of V2R, as measured by western blotting in HEK293 cells. Data represent mean ± SEM of five independent experiments and are normalized with maximal ERK activation in control condition (treated as 100%). A representative image of these experiments is shown in Figure S10C. (E) Depletion of barr2 enhances agonist-induced cAMP response in HEK293 cells expressing V2R, as measured using the GloSensor assay. Data represent mean ± SEM of three independent experiments, with each carried out in duplicate and normalized with the maximal cAMP response in CTL condition (treated as 100%). (F) Exogenous expression of barr2 or swap1 lower the enhanced levels of cAMP in barr2 knockdown cells, suggesting a potential link between receptor-bound barr2 conformation and receptor desensitization. A representative profile from three independent experiments, each performed in duplicate, is shown here. See also Figure S10. 3296 Cell Reports 28, 3287–3299, September 24, 2019 muscarinic receptors, contain a very short carboxyl-terminus but SUPPLEMENTAL INFORMATION harbor phosphorylation sites in their 3rd intracellular loops. Do Supplemental Information can be found online at https://doi.org/10.1016/j. such receptors also use a biphasic mechanism of interaction celrep.2019.08.053. with barrs, and do the two barr isoforms adopt distinct conforma- tions for such receptors? Might there exist different conforma- ACKNOWLEDGMENTS tions of receptor-bound barr1 and 2 in response to stimulation by barr-biased ligands, and how do such conformational signa- The research program in Dr. Shukla’s laboratory is supported by an Intermedi- tures govern the ensuing bias at the functional level? In addition, ate Fellowship of the Wellcome Trust/DBT India Alliance Fellowship (grant high-resolution structures of GPCR-barr complexes, preferably number IA/I/14/1/501285) awarded to A.K.S.; the Science and Engineering of different barr isoforms and different receptors, are still Research Board (SERB) (EMR/2017/003804); Innovative Young Biotechnolo- gist Award from the Department of Biotechnology (DBT) (BT/08/IYBA/2014- required to better understand the commonalities and differences 3); and the Indian Institute of Technology, Kanpur. A.K.S. is an Intermediate in these signaling complexes. Future investigations to address Fellow of Wellcome Trust/DBT India Alliance, EMBO Young Investigator, and some of these aspects should clearly offer novel insights into Joy Gill Chair Professor. H.D. and M.B. were supported by National Post- GPCR-barr interaction and reveal how conformational differ- Doctoral Fellowship of SERB (PDF/2016/002930 and PDF/2016/2893). Dr. Se- ences in receptor-bound barrs fine-tune their functional lent’s laboratory acknowledges support from the Instituto de Salud Carlos III outcomes. FEDER (PI15/00460 and PI18/00094). J.S. and R.G.-G. also acknowledge the computer resources and technical support provided by the Barcelona In conclusion, we discover structural and conformational dif- Supercomputing Center (RES-BCV-2018-1-0012). T.S. acknowledges sup- ferences between receptor-bound barr isoforms that are poten- port from National Center of Science, Poland (2013/08/M/ST6/00788 and tially associated with their functional divergence in the context of 2017/27/N/NZ2/02571). Research in Dr. Chung’s laboratory is supported by GPCR regulatory and signaling paradigms. Our findings under- the National Research Foundation of Korea funded by the Korean Government line the importance of carefully considering both isoforms of (NRF-2017K1A3A1A12072316). Dr. Dutta’s laboratory thanks the DBT-IISc barrs when designing and characterizing barr-biased GPCR Partnership Program for EM consumables and transmission EM facility at Bio- logical Sciences Division, IISc, Bangalore. J.G., a PhD student in the laboratory ligands and, thus, have direct implications for an ever-growing of Dr. Stuart Maudsley, is supported by the FWO-OP/Odysseus program (42/ area of biased agonism aimed at designing novel GPCR FA010100/32/6484), and her visit to the laboratory of Prof. Luttrell was funded therapeutics. by the FWO Travelling Fellowship Program (V4.161.17N). Research in Dr. Lut- trell’s laboratory is supported by NIH/NIGMS grant R35 GM126955. We thank Dr. Ramanuj Banerjee for his assistance with Figure 1 and Haaris Safdari for STAR+METHODS help in single-particle negative-staining experiments. We are thankful to Prof. Ashwani K. Thakur for kindly allowing us to use the fluorometer in his lab- Detailed methods are provided in the online version of this paper oratory for bimane experiments. and include the following: AUTHOR CONTRIBUTIONS d KEY RESOURCES TABLE d LEAD CONTACT AND MATERIALS AVAILABILITY Conceptualization, E.G., A.K.S., and J.S.; Methodology, E.G., A.K.S., and J.S.; d EXPERIMENT MODEL AND SUBJECT DETAILS Investigation, E.G., H.D., M.B., A.S., P.K., T.S., H.R.K., M.-H.L., J.v.G., M.C., B Bacterial cell culture D.R., S.P., J.M., R.G.-G., and S.D.; Writing – Original Draft, A.K.S.; Writing – B Mammalian cell culture Review & Editing, E.G., A.K.S., and J.S.; Funding Acquisition, A.K.S., L.M.L., K.Y.C., S.D., and J.S.; Resources, L.M.L.; Supervision, A.K.S., L.M.L., B Insect cell culture K.Y.C., S.D., and J.S.; Project Administration, A.K.S. d METHOD DETAILS B Construct design and protein expression DECLARATION OF INTERESTS B Sequence and structural analysis of barrs B ELISA assay The authors declare no competing interests. B Co-immunoprecipitation assay B Sortase ligation protocol Received: June 8, 2019 B Confocal microscopy Revised: August 5, 2019 Accepted: August 16, 2019 B Bimane fluorescence assay Published: September 24, 2019 B Functional assays B Negative staining single particle analysis of b2V2R- REFERENCES barrswap2-Fab30 complex B MD simulation set-up and analysis Ahn, S., Nelson, C.D., Garrison, T.R., Miller, W.E., and Lefkowitz, R.J. (2003). B Dynamics of active and inactive barr1 Desensitization, internalization, and signaling functions of beta-arrestins demonstrated by RNA interference. Proc. Natl. Acad. Sci. USA 100, 1740– B Simulation set-ups 1744. B Analysis of inter-domain rotation angle Berendsen, H.J., van Postma, J.P.M., Gunsteren, W.F., DiNola, A., and Haak, B Evaluation of inactivation pathway generated by inter- J.R. (1984). Molecular dynamics with coupling to an external bath. J. Chem. polation Phys. 81, 3684–3690. B Hydrogen/deuterium exchange analysis Bjarnadóttir, T.K., Gloriam, D.E., Hellstrand, S.H., Kristiansson, H., Fredriks- B Receptor-barr chemical cross-linking son, R., and Schiöth, H.B. (2006). Comprehensive repertoire and phylogenetic d QUANTIFICATION AND STATISTICAL ANALYSIS analysis of the G protein-coupled receptors in human and mouse. Genomics d DATA AND CODE AVAILABILITY 88, 263–273. Cell Reports 28, 3287–3299, September 24, 2019 3297 Cahill, T.J., III, Thomsen, A.R., Tarrasch, J.T., Plouffe, B., Nguyen, A.H., Yang, Lally, C.C., Bauer, B., Selent, J., and Sommer, M.E. (2017). C-edge loops of F., Huang, L.Y., Kahsai, A.W., Bassoni, D.L., Gavino, B.J., et al. (2017). Distinct arrestin function as a membrane anchor. Nat. Commun. 8, 14258. conformations of GPCR-b-arrestin complexes mediate desensitization, Laporte, S.A., Oakley, R.H., Zhang, J., Holt, J.A., Ferguson, S.S., Caron, M.G., signaling, and endocytosis. Proc. Natl. Acad. Sci. USA 114, 2562–2567. and Barak, L.S. (1999). The beta2-adrenergic receptor/betaarrestin complex Chen, I., Dorr, B.M., and Liu, D.R. (2011). A general strategy for the evolution of recruits the clathrin adaptor AP-2 during endocytosis. Proc. Natl. Acad. Sci. bond-forming enzymes using yeast display. Proc. Natl. Acad. Sci. USA 108, USA 96, 3712–3717. 11399–11404. Latorraca, N.R., Wang, J.K., Bauer, B., Townshend, R.J.L., Hollingsworth, Darden, T., York, D., and Pedersen, L. (1993). Particle mesh Ewald: An S.A., Olivieri, J.E., Xu, H.E., Sommer, M.E., and Dror, R.O. (2018). Molecular N,log(N) method for Ewald sums in large systems. J. Chem. Phys. 98, mechanism of GPCR-mediated arrestin activation. Nature 557, 452–456. 10089–10092. Luttrell, L.M., Ferguson, S.S., Daaka, Y., Miller, W.E., Maudsley, S., Della DeWire, S.M., Ahn, S., Lefkowitz, R.J., and Shenoy, S.K. (2007). Beta-arrestins Rocca, G.J., Lin, F., Kawakatsu, H., Owada, K., Luttrell, D.K., et al. (1999). and cell signaling. Annu. Rev. Physiol. 69, 483–510. Beta-arrestin-dependent formation of beta2 adrenergic receptor-Src protein Eichel, K., Jullié, D., and von Zastrow, M. (2016). b-Arrestin drives MAP kinase kinase complexes. Science 283, 655–661. signalling from clathrin-coated structures after GPCR dissociation. Nat. Cell Luttrell, L.M., Roudabush, F.L., Choy, E.W., Miller, W.E., Field, M.E., Pierce, Biol. 18, 303–310. K.L., and Lefkowitz, R.J. (2001). Activation and targeting of extracellular Eichel, K., Jullié, D., Barsi-Rhyne, B., Latorraca, N.R., Masureel, M., Sibarita, signal-regulated kinases by beta-arrestin scaffolds. Proc. Natl. Acad. Sci. J.B., Dror, R.O., and von Zastrow, M. (2018). Catalytic activation of b-arrestin USA 98, 2449–2454. by GPCRs. Nature 557, 381–386. McDonald, P.H., Chow, C.W., Miller, W.E., Laporte, S.A., Field, M.E., Lin, F.T., Ghosh, E., Srivastava, A., Baidya, M., Kumari, P., Dwivedi, H., Nidhi, K., Ran- Davis, R.J., and Lefkowitz, R.J. (2000). Beta-arrestin 2: a receptor-regulated jan, R., Dogra, S., Koide, A., Yadav, P.N., et al. (2017). A synthetic intrabody- MAPK scaffold for the activation of JNK3. Science 290, 1574–1577. based selective and generic inhibitor of GPCR endocytosis. Nat. Nanotechnol. Nobles, K.N., Guan, Z., Xiao, K., Oas, T.G., and Lefkowitz, R.J. (2007). The 12, 1190–1198. active conformation of beta-arrestin1: direct evidence for the phosphate Goodman, O.B., Jr., Krupnick, J.G., Santini, F., Gurevich, V.V., Penn, R.B., sensor in the N-domain and conformational differences in the active states Gagnon, A.W., Keen, J.H., and Benovic, J.L. (1996). Beta-arrestin acts as a of beta-arrestins1 and -2. J. Biol. Chem. 282, 21370–21381. clathrin adaptor in endocytosis of the beta2-adrenergic receptor. Nature Oakley, R.H., Laporte, S.A., Holt, J.A., Caron, M.G., and Barak, L.S. (2000). Dif- 383, 447–450. ferential affinities of visual arrestin, beta arrestin1, and beta arrestin2 for G pro- Grest, G.S., and Kremer, K. (1986). Molecular dynamics simulation for poly- tein-coupled receptors delineate two major classes of receptors. J. Biol. mers in the presence of a heat bath. Phys. Rev. A Gen. Phys. 33, 3628–3631. Chem. 275, 17201–17210. Gurevich, V.V., and Gurevich, E.V. (2004). The molecular acrobatics of arrestin Ohi, M., Li, Y., Cheng, Y., and Walz, T. (2004). Negative Staining and Image activation. Trends Pharmacol. Sci. 25, 105–111. Classification—Powerful Tools in Modern Electron Microscopy. Biol. Proced. Gurevich, V.V., and Gurevich, E.V. (2015). Arrestins: Critical Players in Traf- Online 6, 23–34. ficking of Many GPCRs. Prog. Mol. Biol. Transl. Sci. 132, 1–14. Ranjan, R., Dwivedi, H., Baidya, M., Kumar, M., and Shukla, A.K. (2017). Novel Hara, M.R., Kovacs, J.J., Whalen, E.J., Rajagopal, S., Strachan, R.T., Grant, Structural Insights into GPCR-b-Arrestin Interaction and Signaling. Trends Cell W., Towers, A.J., Williams, B., Lam, C.M., Xiao, K., et al. (2011). A stress Biol. 27, 851–862. response pathway regulates DNA damage through b2-adrenoreceptors and Santos, R., Ursu, O., Gaulton, A., Bento, A.P., Donadi, R.S., Bologa, C.G., b-arrestin-1. Nature 477, 349–353. Karlsson, A., Al-Lazikani, B., Hersey, A., Oprea, T.I., and Overington, J.P. Harvey, M.J., Giupponi, G., and Fabritiis, G.D. (2009). ACEMD: Accelerating (2017). A comprehensive map of molecular drug targets. Nat. Rev. Drug Dis- Biomolecular Dynamics in the Microsecond Time Scale. J. Chem. Theory cov. 16, 19–34. Comput. 5, 1632–1639. Sente, A., Peer, R., Srivastava, A., Baidya, M., Lesk, A.M., Balaji, S., Shukla, Huang, J., Rauscher, S., Nawrocki, G., Ran, T., Feig, M., de Groot, B.L., A.K., Babu, M.M., and Flock, T. (2018). Molecular mechanism of modulating Grubmu €ller, H., and MacKerell, A.D., Jr. (2017). CHARMM36m: an improved arrestin conformation by GPCR phosphorylation. Nat. Struct. Mol. Biol. 25, force field for folded and intrinsically disordered proteins. Nat. Methods 14, 538–545. 71–73. Shukla, A.K., Xiao, K., and Lefkowitz, R.J. (2011). Emerging paradigms of b-ar- Humphrey, W., Dalke, A., and Schulten, K. (1996). VMD: visual molecular dy- restin-dependent seven transmembrane receptor signaling. Trends Biochem. namics. J. Mol. Graph 14, 33–38, 27–28. Sci. 36, 457–469. Kang, Y., Zhou, X.E., Gao, X., He, Y., Liu, W., Ishchenko, A., Barty, A., White, Shukla, A.K., Manglik, A., Kruse, A.C., Xiao, K., Reis, R.I., Tseng, W.C., Staus, T.A., Yefanov, O., Han, G.W., et al. (2015). Crystal structure of rhodopsin D.P., Hilger, D., Uysal, S., Huang, L.Y., et al. (2013). Structure of active b-ar- bound to arrestin by femtosecond X-ray laser. Nature 523, 561–567. restin-1 bound to a G-protein-coupled receptor phosphopeptide. Nature Kim, Y.J., Hofmann, K.P., Ernst, O.P., Scheerer, P., Choe, H.W., and Sommer, 497, 137–141. M.E. (2013). Crystal structure of pre-activated arrestin p44. Nature 497, Shukla, A.K., Westfield, G.H., Xiao, K., Reis, R.I., Huang, L.Y., Tripathi-Shukla, 142–146. P., Qian, J., Li, S., Blanc, A., Oleskie, A.N., et al. (2014). Visualization of arrestin Klauda, J.B., Venable, R.M., Freites, J.A., O’Connor, J.W., Tobias, D.J., Mon- recruitment by a G-protein-coupled receptor. Nature 512, 218–222. dragon-Ramirez, C., Vorobyov, I., MacKerell, A.D., Jr., and Pastor, R.W. Srivastava, A., Gupta, B., Gupta, C., and Shukla, A.K. (2015). Emerging Func- (2010). Update of the CHARMM all-atom additive force field for lipids: valida- tional Divergence of b-Arrestin Isoforms in GPCR Function. Trends Endocrinol. tion on six lipid types. J. Phys. Chem. B 114, 7830–7843. Metab. 26, 628–642. Kumari, P., Srivastava, A., Banerjee, R., Ghosh, E., Gupta, P., Ranjan, R., Staus, D.P., Wingler, L.M., Choi, M., Pani, B., Manglik, A., Kruse, A.C., and Lef- Chen, X., Gupta, B., Gupta, C., Jaiman, D., and Shukla, A.K. (2016). Functional kowitz, R.J. (2018). Sortase ligation enables homogeneous GPCR phosphory- competence of a partially engaged GPCR-b-arrestin complex. Nat. Commun. lation to reveal diversity in b-arrestin coupling. Proc. Natl. Acad. Sci. USA 115, 7, 13416. 3834–3839. Kumari, P., Srivastava, A., Ghosh, E., Ranjan, R., Dogra, S., Yadav, P.N., and Thomsen, A.R.B., Plouffe, B., Cahill, T.J., III, Shukla, A.K., Tarrasch, J.T., Shukla, A.K. (2017). Core engagement with b-arrestin is dispensable for Dosey, A.M., Kahsai, A.W., Strachan, R.T., Pani, B., Mahoney, J.P., et al. agonist-induced vasopressin receptor endocytosis and ERK activation. Mol. (2016). GPCR-G Protein-b-Arrestin Super-Complex Mediates Sustained G Biol. Cell 28, 1003–1010. Protein Signaling. Cell 166, 907–919. 3298 Cell Reports 28, 3287–3299, September 24, 2019 Trivedi, D.B., Loftin, C.D., Clark, J., Myers, P., DeGraff, L.M., Cheng, J., Zeldin, Xiao, K., McClatchy, D.B., Shukla, A.K., Zhao, Y., Chen, M., Shenoy, S.K., D.C., and Langenbach, R. (2013). b-Arrestin-2 deficiency attenuates abdom- Yates, J.R., III, and Lefkowitz, R.J. (2007). Functional specialization of beta-ar- inal aortic aneurysm formation in mice. Circ. Res. 112, 1219–1229. restin interactions revealed by proteomic analysis. Proc. Natl. Acad. Sci. USA Vibhuti, A., Gupta, K., Subramanian, H., Guo, Q., and Ali, H. (2011). Distinct 104, 12011–12016. and shared roles of b-arrestin-1 and b-arrestin-2 on the regulation of C3a re- Yun, Y., Kim, D.K., Seo, M.D., Kim, K.M., and Chung, K.Y. (2015). Different ceptor signaling in human mast cells. PLoS One 6, e19585. conformational dynamics of b-arrestin1 and b-arrestin2 analyzed by Walters, R.W., Shukla, A.K., Kovacs, J.J., Violin, J.D., DeWire, S.M., Lam, hydrogen/deuterium exchange mass spectrometry. Biochem. Biophys. Res. C.M., Chen, J.R., Muehlbauer, M.J., Whalen, E.J., and Lefkowitz, R.J. Commun. 457, 50–57. (2009). beta-Arrestin1 mediates nicotinic acid-induced flushing, but not its antilipolytic effect, in mice. J. Clin. Invest. 119, 1312–1321. Zhou, X.E., He, Y., De Waal, P.W., Gao, X., Kang, Y., Van Eps, N., Yin, Y., Pal, Xiao, K., Shenoy, S.K., Nobles, K., and Lefkowitz, R.J. (2004). Activation- K., Goswami, D., White, et al. (2017). Identification of Phosphorylation Codes dependent conformational changes in beta-arrestin 2. J. Biol. Chem. 279, for Arrestin Recruitment by G Protein-Coupled Receptors. Cell 170, 457– 55744–55753. 469.e13. Cell Reports 28, 3287–3299, September 24, 2019 3299 STAR+METHODS KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Streptavidin-HRP Genscript Cat# M00091 Protein L-HRP Genscript Cat# M00098 Monoclonal ANTI-FLAG M2-HRP antibody Sigma-Aldrich Cat# A8592;RRID:AB_439702 Phospho-p44/42 MAPK (Erk1/2) Cell Signaling Technology Cat# 9101; RRID:AB_331646 p44/42 MAPK (Erk1/2) Antibody Cell Signaling Technology Cat# 9102; RRID:AB_330744 b-Arrestin 1/2 (D24H9) Rabbit mAb Cell Signaling Technology Cat# 4674; RRID: AB_10547883 HA-probe (Y-11) antibody Santacruz Biotechnology Cat# sc-805, RRID:AB_631618 Anti-rabbit IgG secondary antibody GenScript Cat# A00098 Bacterial and Virus Strains E. coli DH5a NEB Cat# C2987I E. coli BL21 (DE3) NEB Cat# C2527I E. coli 55244 ATCC ATCC,55244 Baculovirus carrying b2V2R, V2R and GRK2 Cloned in pVL1393 N/A Chemicals, Peptides, and Recombinant Proteins Luria Bertani Broth, Miller Sisco Research Laboratories (SRL) Cat# 29817 (LM019) 2 XYT Growth Medium Himedia Cat# G034-500G DSP (Dithiobis succinimidyl-propionate) Sigma Aldrich Cat# D3669 Paraformaldehyde (PFA) Sigma Aldrich Cat# P6148,CAS no. 30525-89-4 Isopropyl-B-D-Thiogalactopyranoside (IPTG) SRL Cat# 67208 (094866) Lysozyme (3x cryst) ex. Egg white SRL Cat# 45822 (124013) Phenylmethane Sulphonyl Fluoride (PMSF) SRL Cat# 84375 (84375) Benzamidine Hydrochloride SRL Cat# 93014 (0248255) Phosphatase inhibitor cocktail (Phosstop) Roche 4906837001 Monobromobimane Sigma Aldrich Cat# B4380,CAS no. 71418-44-5 Thrombin, Bovine Merck Cat# 605157,CAS no. 9002-04-4 L-Glutathione reduced Sigma Aldrich G4251, CAS no. 70-18-8 Lauryl Maltose Neopentyl Glycol (MNG) Anatrace NG310 CAS no.1257852-96-2 FLAG peptide Genscript N/A TMB (Tetramethylbenzidine) Genscript M00078 Luciferin sodium salt Gold Biotech Cat# LUCNA,CAS no. 103404-75-7 Puromycin dihydrochloride Gold Biotech Cat# P-600-100 PEI (Polyethylenimine) Polysciences Cat# 23966 Bovine Serum Albumin, BSA SRL 83803 (0140105) HBSS - Hank’s Balanced Salt Solution Thermo Fisher Scientific Cat# 14065 GIBCO Fetal Bovine Serum Thermo Fisher Scientific Cat# 10270-106 DMEM Cellclone Cat# CC3004 GIBCO Penicillin-Streptomycin Thermo Fisher Scientific Cat# 15140122 ESF 921 Insect Cell Culture Medium Expression Systems Cat# 96-001-01 FLAG Peptide Genscript N/A Isoproterenol Bitartarate Sigma Aldrich Cat# I2760 CAS No.54750-10-6 Carazolol ApexBio Cat# C5802. BI 167107 Synthesized N/A Tolvaptan Sigma Aldrich Cat# T7455 CAS No.150683-30-0 (Continued on next page) e1 Cell Reports 28, 3287–3299.e1–e6, September 24, 2019 Continued REAGENT or RESOURCE SOURCE IDENTIFIER Arginine Vasopressin Peptide (AVP) Genscript N/A V2Rpp and 3G-V2Rpp Tuft’s Peptide Synthesis Facility N/A Critical Commercial Assays Glosensor Promega Cat# E2301 Experimental Models: Cell Lines Human: HEK293T ATCC Cat# CRL-3216 Insect: Sf9 Expression systems N/A Oligonucleotides N/A N/A N/A Recombinant DNA barr1 N245C expression plasmid Genscript N/A barr2 S246C expression plasmid Genscript N/A barr1/2-mcherry Dr Mark G.H. Scott N/A barr1/2 shRNA Dr Hyder Ali N/A cDNA cassette for b2V2R cDNA resource center N/A cDNA cassette for V2R cDNA resource center N/A cDNA cassette for GRK2 cDNA resource center N/A Sortase A pentamutant (eSrtA) in pET29 Chen et al., 2011 Addgene Plasmid #75144 Software and Algorithms Image Lab Bio-Rad N/A Graphpad Prism Graphpad N/A Zen lite, Zeiss Zeiss https://www.zeiss.com/microscopy/int/ products/microscope-software/zen-lite.html Pymol Schrodinger LLC N/A ProteinLynx Global Server 2.4 Waters N/A DynamX program Waters N/A ACEMD simulation package Harvey et al., 2009 N/A CHARMM36m forcefield Huang et al., 2017 N/A CHARMM36 forcefield Klauda et al., 2010 N/A MOE software package Chemical Computing Group (CCG) https://www.chemcomp.com/ Other (Resins) M1-FLAG resin In-house N/A CaptoL (Protein L) GE Lifesciences 17547802 Glutathione resin Genscript Cat# L0026 Mouse anti-HA (Hemagglutinin) resin Sigma Cat# A2095 His60 Ni Superflow resin Clonetech Cat# 635660 LEAD CONTACT AND MATERIALS AVAILABILITY Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Arun K. Shukla (arshukla@iitk.ac.in). This study did not generate new unique reagents. EXPERIMENT MODEL AND SUBJECT DETAILS Bacterial cell culture Three bacterial strains viz., E. coli DH5a, E. coli BL21(DE3) pLysS and E. coli 55244 were used in the current study and were cultured according to standard protocols. Cell Reports 28, 3287–3299.e1–e6, September 24, 2019 e2 Mammalian cell culture HEK293 cells (Female) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum and further sup- plemented with penicillin and streptomycin. Cells were regularly checked for mycoplasma contamination. Insect cell culture Sf9 cells (Female) were grown in serum-free media and maintained at 27 C. Cells were infected with baculoviruses as per standard protocols. METHOD DETAILS Construct design and protein expression E. coli expression constructs for barr1 and 2, Fab30 and ScFv30 are described earlier, and these proteins were purified using pre- viously described protocols (Kumari et al., 2017). barr1N245C/barr2S246C were generated on a minimal cysteine background (i.e., harboring C59V, C68L, C125S, C140L, C150V, C242V, C251V, C269S mutations) by site-directed mutagenesis, and purified using a similar protocol as for wild-type. Expression constructs for b2V2R, GRK2, V2R, and their purification details have also been pub- lished previously (Kumari et al., 2017). Briefly, FLAG-b2V2R and FLAG-V2R were co-expressed with GRK2 in Sf9 cells (cultured in ESF921 media from Expression Systems), and 60-66h post-infection; cells were stimulated with indicated ligands and harvested by centrifugation. Sequence and structural analysis of barrs Phosphate interacting residues in V2Rpp-bound barr1 were identified based on the previously determined crystal structure (PDB ID: 4JQI). They were compared to barr2 by sequence alignment, structural visualization in PyMol and subsequent analysis in PDBsum as indicated in respective figure legends. ELISA assay In order to assess the interaction of Fab30 with V2Rpp-bound barrs (presented in Figures 1D and S3E), we first immobilized purified protein L (Genscript) onto MaxiSorp ELISA plates (Nunc). Subsequently, we incubated the wells with 1% BSA (Bovine Serum Albu- min) to block non-specific binding. Afterward, we added purified Fabs (in 20 mM HEPES, pH 7.4, 150mM NaCl; 1-3 mg per well in 100 mL volume) followed by gentle washing to remove unbound Fabs, and then added purified biotinylated barrs (in 20 mM HEPES, pH 7.4, 150mM NaCl, 0.01% MNG; 1-3 mg per well in 100 mL volume) (with or without pre-incubation with V2Rpp). After an incubation of 15-30 min, wells were washed extensively (using 20mM HEPES, pH 7.4, 100mM NaCl, 0.01% MNG), and incubated with HRP- coupled streptavidin (Genscript; 1:2000 dilution of 1 mg/ml). After another round of extensive washing, the reactivity of Fab30 with barrs was visualized by adding TMB ELISA (Genscript). Colorimetric reaction was stopped by adding 2M H2SO4 and absorbance was measured at 450nm using a Victor X4 plate reader (Perkin-Elmer, USA). In order to assess the recognition of receptor-bound barrs by Fab30 (presented in Figure S5A), we followed a recently described protocol (Kumari et al., 2016). Here, we first immobilized Fab30 on the ELISA plate followed by the addition of activated and phos- phorylated receptor (in the form of cell lysate) mixed with purified barrs. Their interaction was detected using HRP-coupled anti-FLAG M2 antibody (Sigma, 1:2,000 dilution) as the receptor contains an N-terminal FLAG tag. Co-immunoprecipitation assay For coimmunoprecipitation based detection of Fab30 reactivity toward V2Rpp-bound barrs (presented in Figures 1E, S3B–S3D, S8A, and S8B), purified proteins were mixed together (in buffer containing 20mM HEPES, pH 7.4, 150mM NaCl, 0.01% MNG; 1:3 fold molar ratio of barr:Fab; final concentrations in the range of 1-10 mM) and incubated at room-temperature for 1h. Subsequently, pro- tein L agarose beads were added to the reaction mix and incubated for an additional 1h at room-temperature. Beads were washed three times by centrifugation (using 20mM HEPES, pH 7.4, 150mM NaCl, 0.01% MNG), bound proteins were eluted using SDS sam- ple buffer and separated by SDS-PAGE. In order to evaluate the interaction of receptor-bound barrs with Fabs by coIP, we used cell lysate prepared from Sf9 cells co-ex- pressing recombinant FLAG-tagged receptor and GRK2. Cells were first stimulated with an inverse agonist (to generate inactive and non-phosphorylated receptor) or an agonist (to generate activated and phosphorylated receptor). Cell lysate was pre-incubated with purified barrs and Fab30 at room temperature for 1h. Subsequently, protein L beads (equilibrated in 20mM HEPES, pH 7.4, 100mM NaCl and 0.01% MNG) were added to the reaction mix and tumbled for an additional 1h. Beads were washed three times with washing buffer (same as equilibration buffer), bound proteins were eluted with SDS sample loading buffer and analyzed by western blotting (HRP-coupled anti-FLAG M2 HRP from Sigma at 1:2000 dilution; HRP-coupled Protein L from Genscript at 1:2000 dilution). In an alternative coIP set-up (presented in Figure S7C), HEK293 cells expressing b2V2R, barr-mCherry and HA tagged ScFv30 as an intrabody, were stimulated with agonist or inverse agonist followed by coIP using anti-HA antibody coupled agarose beads (Sigma). Interaction of ScFv30 intrabody with barr1 and 2 were visualized by western blotting. e3 Cell Reports 28, 3287–3299.e1–e6, September 24, 2019 Sortase ligation protocol For the preparation of receptor with homogeneous phosphorylation, we followed a slightly modified version of a previously published protocol (Staus et al., 2018). Briefly, Sf9 cells expressing b2AR (29-341) was first stimulated with 10nM BI-167107 for 30min and then resuspended in lysis buffer (50mM HEPES pH 7.4, 150mM NaCl, 10nM BI-167107, 1mM PMSF and 2mM benzamidine). Cells were lysed by dounce homogenization and lysate was solubilized in 1%(v/v) MNG for 2h at room temperature and cleared by centrifugation at 15000 rpm for 30min. Supernatant was incubated with pre-equilibrated M1-FLAG beads supplemented with 2mM CaCl2 for 2h at 4 C. Beads were washed alternately with low salt buffer(50mM HEPES pH 7.4, 150mM NaCl, 0.01% MNG, 10nM BI-167107 and 2mM CaCl2) and high salt buffer(50mM HEPES pH7.4, 350mM NaCl, 0.01% MNG, 10nM BI-167107 and 2mM CaCl2) respectively. For ligation reaction, beads were resuspended in buffer containing 50mM HEPES, pH 7.4, 100mM NaCl, 0.01% (v/v) MNG, 10nM BI- 167107, 5mM CaCl2, 50uM GGG-V2Rpp (GGGARGRpTPPpSLGPQDEpSCpTpTApSpSpSLAKDTSS) and 2 mM sortaseA. Slurry was incubated overnight at 4 C and next day beads were alternately washed with low salt buffer and high salt buffer respectively. Ligated receptor was eluted with FLAG peptide and protein-L coIP assay was performed to assess the interaction of barrs with receptor. Confocal microscopy HEK293 cells (ATCC) were cultured in DMEM (Sigma) supplemented with 5% Fetal Bovine Serum (Thermo Scientific) and 1% Peni- cillin-Streptomycin at 37 C under 5% CO2. Cells were transfected with indicated plasmids using PEI (Poly Ethylene Imine) at a DNA to PEI ratio of 1:3. 24h post-transfection, cells were split and seeded onto poly-L-lysine coated coverslips. After an additional 24h, cells were serum starved for 2h and then used for live cell imaging (LSM780NLO confocal microscope from Carl Zeiss) (Figures 2E and S7B). Cells were stimulated with agonists for indicated time-points as mentioned in the respective figure legends. Bimane fluorescence assay Experimental details of bimane labeling and fluorescence measurements have been described in detail previously (Kumari et al., 2016). Briefly, purified barr1N245C and barr1S246C were buffer exchanged in labeling buffer (20mM HEPES, 150mM NaCl, pH 7.5), and then incubated (approximately at a concentration of 1mg/ml) with freshly prepared monobromobimane (mBBr, Sigma-Aldrich) at a 10-fold molar excess. After 1h incubation on ice, unreacted mBBr was separated on a PD10 desalting column (GE Healthcare). Labeled barrs were either used right away in the fluoresce measurements or they were flash frozen with 10% glycerol and stored at 80 C for later usage. For measuring the conformational change in the finger loop, mBBr labeled barrs were mixed at 1:3 molar ratios with purified b2V2R. Purified receptors were also buffer exchanged in 20mM HEPES, pH 7.5, 150 mM NaCl, 0.01% MNG (Maltose Neopentyl Glycol), and final reactions were prepared such to maintain a consistent buffer (and detergent conditions). After incubating the mixture for 1h at room temperature, bimane fluorescence intensity was measured using a Fluorometer (Perkin Elmer, USA model LS-55) in photon counting mode as described previously (Kumari et al., 2016). Functional assays HEK293 cells were infected with previously described and validated lentiviral shRNAs targeting either barr1 or 2 followed by gener- ation of stable cells lines with puromycin selection using standard protocol (Vibhuti et al., 2011). Agonist-induced cAMP, receptor endocytosis and ERK MAP kinase phosphorylation was measured using previously described protocols (Kumari et al., 2017). Negative staining single particle analysis of b2V2R-barrswap2-Fab30 complex Samples for negative stain EM was prepared by conventional negative staining method (Ohi et al., 2004). Around 3.5 ml of the purified complex of b2V2-barrswap2-Fab30 was adsorbed on glow discharged carbon coated copper grid for around one minute. This was followed by washing with three drops of water and staining with 0.5% uranyl formate for 30 s. Negatively stained b2V2-barrswap2- Fab30 complex sample was imaged by using Tecnai T12 transmission electron microscope furnished with LaB6 filament and oper- ating at 120kV accelerating voltage. Images were collected at magnification of 80k through side mounted Olympus VELITA (2K2K) CCD camera. All the images were collected at a defocus range of -1.5 to 1.8mm yielding a final pixel size of 2 Å at specimen level. Total about 10000 particles from 150 micrographs were selected manually using e2boxer.py of EMAN 2.12 suite and used for 2D classification. The particle stack was classified into 50 classes using simple_prime2D script of SIMPLE software package. MD simulation set-up and analysis Fab30 binding stability in solution From the crystallized active barr1 (PDB code: 4JQI), we removed the co-crystallized phosphopeptide and part of the Fab30 main- taining only residues 5 to 108 of the light chain and residues 1 to 123 of the heavy chain. The missing loop segment (309 to 310) of barr1 was modeled using the loop modeler tool in the MOE software package (https://www.chemcomp.com/). Afterward, we generated different inter-domain rotation states using linear interpolation between the active barr1 (PDB code: 4JQI) and the inactive barr1 (PDB code: 1G4R). Complexes were subjected to a geometrical optimization using the MOE package (CHARMM27 force field and born solvation). During this optimization, we applied constraints to the backbone atoms of the b sheets and helices of barr1. To verify that obtained barr1 conformations reflect a low-energetic conformation along the inactivation pathway, we compared interpo- lated structures to conformations observed in unbiased simulations of barr1 inactivation (see Figure S8E). Interpolated structures were populated with an RMSD less than 1 Å in unbiased inactivation simulations proving low-energetic conformations. Cell Reports 28, 3287–3299.e1–e6, September 24, 2019 e4 Fab30 complex was obtained by the following procedure. To ensure correct placement of Fab30 with respect to the C-domain (i.e., the main Fab30 binding interface) of the barr1, all interpolation states were aligned to the C-domain of the active barr1-Fab30 com- plex (PDB code: 4JQI). The Fab30 binding interface is optimized during the equilibration phase with only backbone atoms con- strained (see section simulation setup below). Note that this procedure does not remove clashes of the Fab30 with the N-domain produced by low inter-domain rotation angles domain in the inactive barr1 compared to the active state. In fact, these clashes are likely the reason for experimentally observed downregulation of Fab30 binding and are the focus of our simulation experiments. The obtained barr1-Fab30 complexes were solvated and ionized to 150 mM NaCl using VMD (Humphrey et al., 1996) yielding a system of approximately 86000 atoms. Systems were equilibrated for 10 ns in the NPT ensemble applying harmonic positional re- straints to the protein backbone atoms and allowing side chains, water molecules and ions to relax (see detailed information below). Then, NVT production runs (see section simulation setup below) were used to assess Fab30 binding stability during simulation by monitoring the RMSD of the backbone atoms of the Fab30 b sheet. In order to ensure a correct detection of Fab30 movement, we aligned the simulated complex to the C-domain of barr1 prior to RMSD measurements. Dynamics of active and inactive barr1 To sample the conformational flexibility of inactive and active barr1, we started simulations from the active (PDB code: 4JQI) and inactive (PDB code: 1G4R) crystal structure. In the active structure, we removed the co-crystallized Fab30, maintaining the co-crys- tallized phosphopeptide to stabilize the active conformation. In both of the structures missing loops were modeled using the loop modeler tool in the MOE software package. Structures were solvated and ionized according to the protocol described above yielding systems above 56000 atoms. Afterward, solvated complexes were subjected to equilibration and simulation (see section simulation setup below). Simulation times (ms) for barrs in receptor-bound complexes Simulation system Simulation time barr1 in complex with Fab30 with different inter- accumulated 7.6 domain rotation angles (10 systems) Peptide stabilized active barr1 (3 replicates) accumulated 3.1 Inactive barr1 (3 replicates) accumulated 2.0 Total 12.7 Simulation set-ups All simulations were carried out using the ACEMD simulation package (Harvey et al., 2009) and the CHARMM36m forcefield (Huang et al., 2017) and CHARMM36 forcefield (Klauda et al., 2010) force fields for proteins and lipids, respectively. NPT simulations were carried out at 310 K and 1 bar using the Berendsen barostat (Berendsen, 1984) with a relaxation time of 400 fs and 2 fs integration time step and harmonic constraints applied to all backbone atoms. NVT simulations were run at 310 K, using the Langevin thermostat (Grest and Kremer, 1986) with a damping coefficient of 5 ps 1 and 4 fs integration time step. No harmonic constraints are applied in this phase. In all simulations, we used a van der Waals and short-range electrostatic interactions with a cut-off of 9 Å and the par- ticle mesh Ewald method (Darden, 1993) for long-range electrostatic interactions. Analysis of inter-domain rotation angle The inter-domain rotation angle is used as metric to assess the conformational landscape of barrs in the receptor-bound state. For this purpose, we computed the displacement of the C-domain relative to the N-domain between the inactive (PDB code: 1G4R) and active barr1 crystal structures (PDB code: 4JQI) as previously described in Latorraca et al. (2018). The corresponding script was kindly provided by Naomi Latorraca. Evaluation of inactivation pathway generated by interpolation To verify that obtained barr1 conformations reflect a low-energetic conformation along the inactivation pathway, we compared inter- polated structures to conformations observed in unbiased simulations of barr1 inactivation The active structure of barr1 (PDB code: 4JQI) was solvated and simulated for 800ns in NVT conditions in three separate runs (see section simulation set-ups), allowing it to spontaneously inactivate. Interpolation states were aligned to the frames of the barr1 inactivation using the backbone atoms of b sheets). Then, we quantified the presence of interpolated states along the inactivation pathway (Figure S8E). We find that all ten interpolated structures with an inter-domain rotation angle between 0 and 20 are sampled by unbiased simulation with an RMSD of b sheets lower than 1. This indicates that generated states of barr1 in Fab30 complexes adopt a low-energetic conformation along the inactivation pathway. e5 Cell Reports 28, 3287–3299.e1–e6, September 24, 2019 Hydrogen/deuterium exchange analysis Protein expression and purification conditions are as previously described (Yun et al., 2015). HDX-MS and data processing methods are also as previously described (Yun et al., 2015). Briefly, Purified protein was prepared in 60–100 mM in H2O buffer (20 mM HEPES, pH 7.4 and 150 mM NaCl). Hydrogen/deuterium exchange was initiated by mixing 2 mL of protein with 28 mL of D2O buffer (20 mM HEPES, pD 7.4, 150 mM NaCl in D2O), and the mixture was incubated for various time intervals (10, 100, 1000 and 10,000 s) on ice. At the indicated time points, the mixture was quenched by adding 30 mL of ice-cold quench buffer (100 mM NaH2PO4, pH 2.01). For non- deuterated (ND) samples, 2 mL of purified protein was mixed with 28 mL of H2O buffer to which, 30 mL of ice-cold quench buffer was added. Quenched samples were digested online by passing them through an immobilized pepsin column (2.1 X 30 mm) (Life Tech- nologies, Carlsbad, CA, USA) at a flow rate of 100 mL/min with 0.05% formic acid in H2O at 11 C. Peptide fragments were subse- quently collected on a C18 VanGuard trap column (1.7 mm X 30 mm) (Waters, Milford, MA, USA) for desalting with 0.05% formic acid in H2O and were then separated by ultra-pressure liquid chromatography using an Acquity UPLC C18 column (1.7 mm, 1.0 X 100 mm) (Waters) at a flow rate of 35 mL/min with an acetonitrile gradient starting with 8% B and increasing to 85% B over 8.5 min. To minimize the back-exchange of deuterium to hydrogen, the system from trapping column to UPLC column was main- tained at 0.5 C and the buffers were adjusted to pH 2.5. Mass spectral analyses were performed with a Xevo G2 Qtof equipped with a standard ESI source (Waters). Mass spectra were acquired in the range of m/z 100–2000 for 12 min in positive ion mode. Pep- tide identification and HDX-MS data processing Peptic peptides were identified in non-deuterated samples with ProteinLynx Global Server 2.4 (Waters). Searches were run with the variable methionine oxidation modification. To process HDXMS data, the amount of deuterium in each peptide was determined by measuring the centroid of the isotopic distribution using the DynamX program (Wa- ters). Back-exchange was not corrected because the data consisted of comparisons between b-arrestin1 and b-arrestin2 or between wild-type and R169E mutants. All of the data was derived from at least three independent experiments. Receptor-barr chemical cross-linking For assessing receptor barr interactions, Sf9 cells expressing b2V2R and GRK2 were first stimulated with an inverse agonist and agonist respectively for 30 minutes. Cells were lysed with lysis buffer (20mM HEPES pH 7.4, 100mM NaCl, 1X phosstop, 1mM PMSF and 2mM benzamidine) and incubated with purified barr 1 or 2 at room temperature for 30 min. This was followed by the addi- tion of 1mM dithiobis (succinimidyl-propionate) from a freshly prepared 100mM stock solution in DMSO. The lysate was incubated at room temperature for 40 min, and the reaction was quenched by adding 1M Tris pH 8.5. The lysate was solubilised in 1%(v/v) MNG for 1 h at room temperature and cleared by centrifugation at 15000 rpm for 30min. The supernatant was incubated with pre-equilibrated M1-FLAG beads supplemented with 2mM CaCl2 for 2h at 4 C. Beads were washed alternately with low salt buffer(20mM HEPES pH 7.4, 150mM NaCl, 0.01% MNG, 2mM CaCl2) and high salt buffer(20mM HEPES pH7.4, 350mM NaCl, 0.01% MNG, 2mM CaCl2) respectively. Cross-linked proteins were eluted in FLAG-elution buffer (20mM HEPES pH 7.4, 150mM NaCl, 2mM EDTA, 0.01% MNG and 250ug/mL FLAG peptide). Samples were resolved on SDS-PAGE and visualized by western blotting. QUANTIFICATION AND STATISTICAL ANALYSIS Gel and blot densitometry analysis was done on ImageJ and Image lab (Bio-Rad) software respectively. Experimental data from ELISA assays, bimane fluorescence assay and densitometry based quantification of western blots were plotted using GraphPad Prism software. Data represents means ± SEM and details of statistical analysis and number of biological replicates are indicated in the respective figure legends. DATA AND CODE AVAILABILITY This study did not generate any datasets and code for analysis. Cell Reports 28, 3287–3299.e1–e6, September 24, 2019 e6