Role of the PAR1 Receptor 8th Helix in Signaling

The protease-activated receptors are tethered ligand G protein-coupled receptors that are activated by proteolytic cleavage of the extracellular domain of the receptor. The archetypic protease-activated receptor PAR1 strongly activates Gq signaling pathways, but very little is known regarding the mechanism of signal transference between receptor and internally located G protein. The recent x-ray structure of rhodopsin revealed the presence of a highly conserved amphipathic 8th helix that is likely to be physically interposed between receptor and G protein. We found that the analogous 8th helix region of PAR1 was critical for activation of Gq-dependent signaling. Engineering an 8th helix α-aneurysm with a downwards-directed alanine residue markedly interfered with signal transference to Gq. The 8th helix-anchoring cysteine palmitoylation sites were important for the affinity of ligand-dependent G protein coupling but did not affect the maximal signal. A network of H-bond and ionic interactions was found to connect the N-terminal portion of the 8th helix to the nearby NPXXY motif on transmembrane helix 7 and also to the adjacent intracellular loop-1. Disruption of these pairwise interactions caused additive defects in coupling to G protein, indicating that the transmembrane 7–8th helix-i1 loop may move in a coordinated manner to transfer the signal from PAR1 to G protein. This “7-8-1” interaction network was found to be prevalent in G protein-coupled receptors involved in endothelial signaling and angiogenesis.

Proteolytic cleavage of the G protein-coupled protease-activated receptors (PARs) 2 activates an extraordinarily diverse array of physiologic responses. These include platelet aggregation and cell-cell adhesion, proliferation/apoptosis, protease homing, cancer invasion, angiogenesis, and the hemostatic and inflammatory responses to vascular injury (1)(2)(3)(4)(5)(6)(7)(8)(9)(10). Four protease-activated receptors have been identified: PAR1, PAR2, PAR3, and PAR4. A distinguishing feature of the PARs is that they all contain a preligand sequence located within their extracellular N-terminal domains (1). PAR1 is cleaved by thrombin and other proteases at the Arg 41 -Ser 42 bond to create a fresh 42 SFLLRN 47 N terminus that acts as an intramolecular ligand (11,12). NMR studies indicate that, following cleavage by thrombin, the PAR1 ligand region undergoes a large conformational change and docks to ligand-binding site(s), one of which is located in the N-terminal extracellular domain (13). Synthetic peptides (e.g. SFLLRN) corresponding to the freshly cleaved N terminus can displace the tethered ligand from the ligand binding site(s) and fully activate PAR1 in an intermolecular mode (13).
Following intra-or intermolecular liganding, the transmembrane domains of the PARs undergo a conformational change that is transmitted to the intracellular loops (i1-i3) and C-terminal i4 domain. It is these intracellular loops and domains that communicate with the plasma membrane-associated heterotrimeric G proteins. PAR1 can interact with three of the four classes of G␣ subunits, G q/16 , G 12/13 , and G i , making it one of the most promiscuous members of the GPCR family (14 -19).
Presently, there is limited information regarding the molecular determinants of PAR1-G protein interactions. In a previous study using chimeric G␣ subunits, we found that the N-terminal ␣N helix of the G␣ subunit interacts with both PAR1 and the G␤ subunit and actively participates in transferring the signal between the agonist-activated receptor and G protein (19). Mutagenesis of the receptor indicated that both the PAR1 i2 loop (20) and C-terminal portion of the i3 loop are important for coupling to G q and G i (21). However, by far the largest component of the intracellular portion of PAR1 is the C-terminal i4 domain, which comprises Ͼ55% of the mass of i1-i4. The involvement of the PAR1 i4 domain in ligand-dependent phosphorylation and internalization has been well documented (22)(23)(24)(25)(26)(27), but its contribution to signaling to G proteins is essentially unexplored.
The new high resolution x-ray structure of bovine rhodopsin revealed an unanticipated 8th helix (H8) in the i4 domain anchored by the 7th transmembrane helix (TM7) and dual palmitoylation sites at Cys 322 and Cys 323 (28). This region is highly conserved in rhodopsin, PAR1, and nearly all of the Class A GPCRs. TM7 is critical for receptor activation and contains the essential NPXXY motif (29,30). TM7 is also the site of attachment of the retinylidene ligand in rhodopsin and forms extensive interactions with TM1, TM2, TM3, TM6, and the 8th helix. Upon receptor activation, the 8th helix moves significantly outward from the adjacent TM2 and i1 loop (31,32). Studies with rhodopsin and other GPCRs have shown that the 8th helix makes direct contact with the membrane-associated regions of the G protein ␣ and ␥ subunits (33)(34)(35) and controls the coupling and affinity of the receptors to G proteins (36 -38).
To determine the potential involvement of the PAR1 8th helix and i4 domain in coupling to G protein upon stimulation by the extracellular tethered ligand (thrombin-generated) versus the intermolecular peptide ligand (SFLLRN), we constructed a model of PAR1 using the refined 2.8 Å x-ray structure of rhodopsin. The structural model was tested by extensive mutagenesis analyses, including point mutations, insertions, and deletions. We discovered a new structural signaling motif involving a network of H-bonding and ionic interactions that connects the 8th helix to the peptide backbone of TM7 and to the highly basic i1 loop. Our data are consistent with a mechanism whereby the 8th helix moves in a coordinated motion with TM7 and the i1 loop during signal transference from PAR1 to G q .

EXPERIMENTAL PROCEDURES
Materials-Human ␣-thrombin (3433 NIH units/mg) was obtained from Hematologic Technologies (Essex Junction, VT). Restriction enzymes and polymerases were purchased from New England Biolabs. SFLLRN, P1pal-H8 (pal-ASSESQRYVYSIL) and P1pal-H8-Ala (pal-AS-SAESQRYVYSIL) were synthesized with a C-terminal amide at the Tufts Medical School Peptide Core Facility and purified by reverse phase-high pressure liquid chromatography. All other chemicals were obtained from Sigma.
Expression of PAR1 in COS7 Fibroblasts-PAR1 and mutated variants were transiently expressed from an EF1␣ promoter in plasmids pCDEF3 (12,39) and pHEF. Plasmid pHEF was made by replacing the cytomegalovirus promoter of pCDNA3.1Hygro with the EF1␣ promoter from pCDEF3, a pCDNA3-derived plasmid. Mutations were introduced into PAR1 by PCR site-directed mutagenesis. PAR1 mutations and sequence fidelity were verified by the Tufts Medical School DNA sequencing core facility. COS7 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1% penicillin and streptomycin in 5% CO 2 at 37°C. The cells were transfected with DEAE-dextran using 5 g of CsCl-prepared plasmid/10-cm dish of 60% confluent COS7 cells.
Phosphoinositide Production Assay-COS7 fibroblasts expressing PAR1 were split into 12-well plates at 250,000 cells/well. 3 H-labeled myoinositol (2 mCi/ml; Amersham Biosciences) was added to the cells 24 h prior to the experiment. Phospholipase C-␤ (PLC-␤)-dependent accumulation of [ 3 H]inositol phosphates was measured in the presence of LiCl. Wells were rinsed twice with 2 ml of Dulbecco's modified Eagle's medium containing 10 mM HEPES buffer, pH 7.3, then twice with 2 ml of phosphate-buffered saline containing 20 mM LiCl. The cells were stimulated with agonist (in triplicate) for 30 min and then extracted with cold methanol and chloroform. The extracts were loaded onto columns containing 1 ml of anion exchange resin AG1X8, formate form, 100 -200 mesh size (Bio-Rad). After loading, the columns were washed twice with 10 ml of H 2 O and twice with 10 ml of 60 mM ammonium formate/5 mM Borax. Column fractions were eluted with 4 ml of 2 M ammonium formate/0.1 M formic acid into vials containing 7.5 ml of scintillation mixture and then counted. The mean of 3-5 determinations was expressed as the fold stimulation above vehicle-treated cells.
Flow Cytometry-The rabbit polyclonal antibody (SFLLR-Ab) directed against the PAR1 ligand region was generated as previously described (12). PAR1-transfected COS7 cells (5 ϫ 10 5 ) were probed with 2.5 g/ml anti-SFLLR-Ab followed by incubation with fluorescein isothiocyanate-goat anti-rabbit IgG (6 mg/ml; Zymed Laboratories Inc.). The cells were analyzed for fluorescence with a FACScan flow cytometer (BD Biosciences). COS7 cells transfected with empty vector were probed with both primary and secondary antibodies to determine background fluorescence.

PAR1 Structural
Model-Using the refined 2.8-Å x-ray structure of tetragonal P4 1 bovine rhodopsin (Protein Data Bank code 1HZX) as template (40), we generated a structural model of human PAR1 that would serve to test the role of the C-terminal cytoplasmic i4 domain in receptor activation of G proteins. With the exception of the long N-terminal extracellular (e1) domain in PAR1, which harbors the tethered ligand and ligand-binding site 1 (13), human PAR1 and bovine rhodopsin have very similar primary amino acid sequences (49% homology), with only a few minor insertions and deletions. Using the Turbo3.0 molecular modeling program, residues 66 -412 of PAR1 were manually substituted for residues 1-348 of bovine rhodopsin. An insertion consisting of PAR1 residues 345 TTEA 348 was placed in extracellular loop e4 between rhodopsin residues Phe 283 and Gly 284 . The PAR1 i3 loop lacks a pentapeptide region corresponding to rhodopsin residues 236 QQQES 240 , which interestingly does not have electron density in the tetragonal P4 1 (28) nor in the trigonal P3 1 rhodopsin x-ray structures (41). Stereochemically reasonable positions were used for side chains that were later subjected to molecular dynamics and energy minimization using the program CNS to yield the final structural model. The published sequence of human PAR1 (11) differs slightly from the PAR1 clone (12) that we used in all of these studies. As occurs in PAR1 from other species (baboon, mouse, hamster, Xenopus), our human PAR1 clone has a Leu 238 instead of Val and a Cys 364 instead of Ser, and our structural model incorporated Leu and Cys at these two positions, respectively. Overall, the peptide backbone of our PAR1 model was very similar to that of rhodopsin (root mean square deviation ϭ 0.29 Å). In our model, however, the volume occupied by the retinal is filled by a series of bulky residues, which replace their less bulky counterparts in rhodopsin. These residues include Phe 182 , Tyr 183 , Met 186 , His 255 , Phe 274 , and Asn 330 , which replace rhodopsin residues Ala 117 , Thr 118 , Gly 121 , Gly 188 , Met 207 , and Ala 269 , respectively.
Four key structural features are highlighted in Fig. 1, A and B, namely the TM7 helix (purple), the cytoplasmic 8th helix (red), the i1 intracellular loop (green), and the tethered ligand (yellow) on the extracellular side. The TM7 helix is critical for receptor activation and contains the essential NPXXY motif and is bent at hinge residues Ser 363 -Cys 364 . The 8th helix of the C-terminal i4 domain is anchored to TM7 by a short Ala 374 -Ser 375 linker and to the plasma membrane at its C terminus by dual palmitoylation sites at Cys 387 and Cys 388 .
The 8th Helix of PAR1 Is Essential for Receptor G q Coupling-Extensive deletion, substitution, and insertion mutations were made to determine the role of the i4 domain and its proximal 8th helix region in PAR1-dependent activation of G proteins by both intramolecular (thrombin-generated) and intermolecular peptide ligand SFLLRN. The effects of mutation of the intracellular residues of PAR1 on coupling to G protein were assessed using a well characterized PLC-␤-dependent inositol phosphate (InsP) assay (13). PLC-␤ is directly activated by G q ␣-subunits and/or by ␤␥-subunits from G i to generate InsP. To determine the contribution of G i (␤␥) to the InsP signal we added the G i inhibitor pertussis toxin to COS7 fibroblasts transiently expressing PAR1. Pertussis toxin had no effect on the InsP signal generated by thrombin-stimulated wild type (WT) PAR1 (supplemental Fig. 1) nor on the PAR1⌬396 mutant (data not shown). Therefore, the InsP signal in the COS7 fibroblasts is derived from PAR1 activation of G q and not from G i (␤␥).
Deletion of the entire C-terminal i4 domain (⌬377) resulted in a large loss of apparent affinity and coupling efficacy to G q by PAR1, activated by either the peptide agonist SFLLRN or by thrombin (Fig. 1C) as reported earlier (21). By comparison, deletion of the residues to the C-terminal side of the Cys-palmitoylation sites (⌬396) gave a 4 -10 right shift in EC 50 for G protein coupling but little or no effect on efficacy (maximal InsP production), suggesting that additional binding determinants for G q may be present in the last 29 residues of the C-tail of PAR1. Deletion of only the 8th helix region (⌬H8) caused major defects in G q coupling (60% loss of maximal InsP) in response to thrombin (9-fold EC 50 shift) and SFLLRN (60-fold EC 50 shift), similar to deletion of the entire i4 domain (Fig. 1C).

Effects of Mutation of the Cysteine Palmitoylation Sites and C-terminal
Residues of the 8th Helix-The 8th helix of PAR1 terminates with two cysteines that are conserved sites of palmitoylation in rhodopsin and many other GPCRs ( Fig. 2A). The Cys-palmitoylate lipids anchor the 8th helix to the inner leaflet of the lipid bilayer and have been shown to be important for G protein coupling efficiency and selectivity for many GPCRs (42). We found that mutation of the Cys-palmitoylation sites (C387S,C388S) caused 4 -9-fold shifts in the EC 50 value for thrombin and SFLLRN but had no effect on maximal coupling efficacy ( Table  1, Fig. 2). Alanine-scanning mutagenesis was then used to determine the role of the nearby C-terminal residues of the 8th helix in signal transference to G protein. Triple alanine substitutions of residues in the C-terminal half of the 8th helix had minimal effects on the EC 50 value of PAR1-G q signaling upon activation by either thrombin or SFLLRN. The maximal InsP signal for the S384A,I385A,L386A mutant was unaffected; however, the Y381A,V382A,Y383A mutation actually enhanced the EC 50 value (2.5-fold left shift) and efficacy (130%) upon stimulation with SFLLRN relative to WT (Table 1, Fig. 2). These effects were not detected upon activation with thrombin, consistent with structural differences in intraversus intermolecular modes of activation, as previously observed with mutations of the extracellular surface of PAR1 (43).
The N-terminal Half of the 8th Helix Is Important for PAR1-G q Coupling-In contrast to the C-terminal half of the 8th helix, residues in the N-terminal half were found to be important for PAR1-G protein coupling. The rhodopsin structural models (28,41) identified an interaction between the peptide backbone of TM7 and side chains from the 8th helix, but these connections have not yet been shown to have any functional significance (36). As shown in the model of Fig. 3A, the hydroxamate moiety of Gln 379 from the Gln 379 -Arg 380 dyad forms an H-bond with the backbone carbonyl oxygen of Tyr 372 immediately adjacent to the NPXXYY 372 motif in TM7. Elimination of this H-bond network and loss of the connection between the Gln-Arg dyad and the Tyr 372 peptide carbonyl group might disrupt coordinated movements between TM7 and the 8th helix (31,32). We found that mutation of the Gln 379 -Arg 380 dyad causes a 7.5-fold shift in EC 50 value for SFLLRN activation of G q and a 20% loss in maximal coupling efficacy upon stim-   Fig. 3). Point mutation of Cys 378 to alanine had no effect on signaling, whereas mutation of 8th helix N-terminating residues Ser 376 and Glu 377 3 Ala gave a 40% decrease in the maximal InsP signal and a 4-fold shift in the EC 50 value for thrombin and SFLLRN (Fig. 3, Table 1). Together, these data indicate that residues in the N-terminal half of the 8th helix are critical for signal transference between PAR1 and G protein.
Synergistic Effects between the 8th Helix, Intracellular Loop 1 (i1), and TM7 on Coupling to G q -The model structure of PAR1 predicts a novel ionic interaction between the Glu 377 carboxyl group and the ⑀-amino moiety of Lys 135 from the i1 loop (Fig. 4A). To test the potential importance of this 8th helix i1 loop interaction, we mutated these residues to alanine. The E377A and K135A mutants exhibited similar reductions in G protein coupling for both thrombin and SFLLRN (Fig. 4, B and C). Combining the two mutations within the same receptor (K135A,E377A) resulted in partially additive losses in EC 50 value and coupling efficacy ( Table 1), suggesting that Lys 135 and Glu 377 are cooperatively interacting residues facilitating the same step in PAR1 coupling to G q (44). Notably, the K135A,E377A double mutant had a coupling efficacy of only 30 -40% (Table 1), equivalent to the effect of deletion of the entire 8th helix (Fig. 1C, ⌬H8).
As mentioned above, the model predicts an interaction between TM7 (Tyr 372 ) and the Gln 379 -Arg 380 dyad on the 8th helix. Creating a PAR1 mutant that combines defects in 8th helix-TM7 and 8th helix-i1 loop interactions should have an additive effect on signaling if the Gln 379 -Arg 380 and Lys 135 residues are non-interacting residues facilitating the same step (44). Indeed, this is observed for the K135A,Q379A,R380A triple mutant. Whereas the K135A and the Q379A,R380A mutations both give 20 -25% reductions in the efficacy (maximal), the combined K135A,Q379A,R380A mutant has a 40 -50% reduction in efficacy (Fig.  4, D and E; Table 1). Likewise, the EC 50 value shifts of the triple mutant reflect additivity of effects of the individual EC 50 value shifts, suggesting that the 8th helix may move in a coordinated motion with TM7 and the i1 loop during signal transference from PAR1 to G protein.
An ␣-Aneurysm in the 8th Helix Disrupts Receptor-G q Coupling-Previous studies have shown that the N terminus of the 8th helical region of rhodopsin makes direct contact with the membrane-associated regions of the G protein ␣ and ␥ subunits (33)(34)(35). Therefore, the N-terminal region of the 8th helix could also directly participate in PAR1-G protein coupling. To disrupt this putative 8th helix-G protein interaction, we inserted an alanine between Ser 376 and Glu 377 to create a helical bulge or ␣-aneurysm (45-47). Using the x-ray structure of a well defined ␣-aneurysm found in a mutant staphylococcal nuclease (45), we built a new PAR1 model that incorporated the alanine insertion and adjusted the coordinates so as to minimize the energy of this model using CNS. In this model, as in the ␣-aneu-   rysm-containing crystal structure, the ␣-helix is almost completely unperturbed, except for the two amino acids on either side of the insertion. As shown in Fig. 5A, the inserted alanine side chain protrudes directly downwards in the direction of the associated G protein without breaking the Glu 377 -Lys 135 and Gln 379 -Tyr 372 contacts. We found that the alanine ␣-aneurysm mutant gave a 60% loss in maximal efficacy of coupling but more strikingly gave a 42-fold right shift in EC 50 value for thrombin-dependent stimulation of the G q -PLC-␤ signal (Fig. 5B, Table 1). The EC 50 value for SFLLRN activation of the Ser 376 -Glu 377 alanine insertion mutant was right-shifted by only 4-fold but lowered the slope (Hill coefficient) (Fig. 5C), suggesting that G q can discriminate between the structures of tethered ligand-activated versus peptide-activated receptors as previously reported (48,49). In addition, we disrupted the Ala 374 -Ser 375 linker that may serve as a hinge for rigid body movements between TM7 and the 8th helix (31). The Ser 375 -Ser 376 alanine insertion mutant exhibited a similar 60% loss in maximal coupling efficacy as the Ser 376 -Glu 377 alanine insertion mutant but gave only a 6-fold right shift in EC 50 value for thrombin activation. An alanine was also inserted into the C-terminal region of the 8th helix between residues Tyr 383 and Ser 384 . In comparison to the N-terminal Ser 376 -Glu 377 alanine insertion, the Tyr 383 -Ser 384 alanine insertion had no effect on maximal coupling efficacy to G q and only slight shifts in the EC 50 value for thrombin and SFLLRN (Fig. 5). These data provide further evidence that coupling determinants for G q are highly localized to the N-terminal region of the 8th helix of PAR1.

TABLE 1 Effect of 8th helix and i1 loop mutations on PAR1 activation of G q -PLC-␤ signaling in COS7 fibroblasts
To provide independent evidence that the 8th helix of PAR1 is directly involved in signal transference to internal G q , we synthesized cell-penetrating palmitoylated peptides, termed pepducins (8,10,21,50), based on the 8th helix of PAR1. The P1pal-H8 (pal-ASSESQRYVY-SIL) and P1pal-H8-Ala (pal-ASSAESQRYVYSIL) pepducins correspond to the wild type H8 sequence and the Ser 376 -Glu 377 ␣-aneurysm alanine insertion mutant, respectively. As shown in supplemental Fig. 2, the P1pal-H8 pepducin inhibited up to 60% of the PAR1-G q -PLC-␤ signal, whereas the 8th helix pepducin with the ␣-aneurysm had no effect on PAR1 signaling in transfected COS7 fibroblasts. These peptide FIGURE 3. The N-terminal residues of the PAR1 8th helix mediate signaling to G q . A, structural model depicting H-bonding interactions between the Gln-Arg dyad (Q 379 R 380 ) on the 8th helix and the peptide backbone (carbonyl oxygen of Tyr 372 ) of transmembrane helix 7. Thrombin (B)-and SFLLRN (C)-dependent stimulation of the PAR1 8th helix mutants was measured by activation of G q -PLC-␤ as described in the legend to Fig. 1.  FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 interference data provide corroborating evidence that the 8th helix region of PAR1 is directly involved in signal transference to G protein. Moreover, these results are consistent with the inhibition of G t coupling that occurs with analogous peptides from the rhodopsin 8th helix region (33,34).

DISCUSSION
The 7-8-1 Receptor Activation Mechanism-The present work demonstrated that the 8th helix in the i4 cytoplasmic domain of PAR1 plays a critical role in signal transference to G q . Biophysical studies have previously shown that the 8th helix of rhodopsin is conformationally mobile (51,52), and its N-terminal end moves away from the C-terminal portion of the i1 loop/TM2 by 2-4 Å upon receptor activation (31,32). Our data provide functional support for a 7-8-1 mechanism, whereby TM7 interacts with the 8th helix, which in turn interacts with the i1-loop, and this chain of interactions is needed for PAR1 activation of G q . In an analysis of other GPCRs, we found that appropriate 7-8-1 connections were present in 16% (28/175) of the rhodopsin-like receptors (supplemental Table 1). Interestingly, many of these putative 7-8-1 GPCRs are expressed on endothelial cells and are involved in angiogenesis. For instance, receptors for sphingosine 1-phosphate, lysophosphatidic acid, platelet activating factor, macrophage inhibitory protein 1␣, serotonin, and endothelin-A all have the requisite amino acids in the 8th helix and i1 loops to make 7-8-1 connections (Fig. 6). It remains to be determined whether these putative 7-8-1 GPCRs share signaling, trafficking, or other biological functions that could be attributed to this common signaling motif.
Hydrophilic Residues in the 8th Helix Control PAR1-G q Coupling-Only a select few hydrophilic residues, such as Glu 377 and Gln 379 -Arg 380 , located at the N-terminal end of the 8th helix, were found to be important for coupling to internally located G q . The finding that the N-terminal residues of the 8th helix are important for receptor G protein signaling is in general similar to that observed for rhodopsin, which couples to transducin (G t ), a member of the G i family (36). However, there were striking differences in the location and identity of the individual 8th helix residues that were important for PAR1-G q coupling versus rhodopsin-G t coupling, suggesting that determinants of G ␣ specificity could reside in the 8th helix of GPCRs. For instance, alanine substitution of rhodopsin residues Gln 312 and Arg 314 -Asn 315 , located at the analogous sites as PAR1 residues Glu 377 and Gln 379 -Arg 380 (Fig. 6), respectively, had no effect on activation of G t (36). Conversely, mutation of the hydrophobic residues Phe 313 and Met 317 in rhodopsin reduced activation of G t by 75-90% (36,38), whereas mutation of the analogous

The PAR1 7-8-1 Receptor Activation Mechanism
residues in PAR1, Cys 378 and Val 382 , did not adversely affect coupling to G q .
Interplay between the 8th Helix and i1 Loop-We found a novel ionic interaction between Glu 377 in the 8th helix and Lys 135 from the first intracellular loop (Fig. 4A) buried in an otherwise hydrophobic environment in the core of the intracellular loops, which was important for coupling to G q . Although this 8-1 interaction is not present in rhodopsin 1, an analysis of other GPCRs indicated that suitable ionic partners could be found with either an Asp or Glu in the 8th helix and a Lys, His, or Arg from the highly basic i1 loop (Fig. 6, supplemental Table 1). For instance, the sphingosine 1-phosphate receptor 1 also has an 8th helix glutamic acid that may interact with an arginine from its i1 loop. In some GPCRs, the residues are reversed; the P2Y4 purinoreceptor has a lysine on the 8th helix and an aspartic acid on the i1 loop, or the 8-1 partners are long side chain H-bond donor acceptors with an 8th helix residues, such as Glu, Gln, Asp, Asn, Lys, or Arg bridging with either a Lys, His, Arg, Asn, or Gln from the i1 loop. Rod pigment rhodopsin 1 lacks an appropriate long chain H-bond donor at the i1 loop (Thr 70 ); however, rhodopsin 2 ( Fig. 6, red and green cone pigments) and rhodopsin 3 (blue cone pigment) have i1 His and Gln residues, which could interact with the 8th helix Gln 312 (supplemental Table 1). A salt bridge is an intrinsically stronger interaction than a hydrogen bond in a hydrophobic environment and could differentially stabilize either a ground state and/or active conformation, depending on the particular GPCR or coupled G ␣ subunit.
Functional Significance of the Connection Between the 8th Helix Gln-Arg Dyad and Peptide Backbone of TM7-The Gln-Arg dyad is a highly conserved motif on the 8th helix and is present in 85% of class A GPCRs (Fig. 6, supplemental Table 1). In both the tetragonal P4 1 and trigonal P3 1 rhodopsin 1 crystal structures, the 8th helix Arg 314 side chain was found to hydrogen bond to the carbonyl backbone oxygens of Ile 307 and Met 308 , which are adjacent to the NPXXY 306 motif on TM7. The carbamate moiety of rhodopsin Asn 315 makes an additional H-bond to the Arg 314 guanidinium. However, substitution of the Arg 314 -Asn 315 dyad with alanines had no apparent deleterious effect in rhodopsin activation of G t (36). In contrast, we found that mutation of the analogous Gln 379 -Arg 380 dyad caused signaling defects in PAR1 activation of G q . This hydrophilic 7th helix-8th helix (7)(8) interaction, although achieved differently, may perform a similar function as that found for the Tyr 306 -Phe 313 connection in rhodopsin. Studies by Fritze et al. (38) demonstrate the importance of the hydrophobic interaction between the Phe 313 in the 8th helix with TM7 residue Tyr 306 in facilitating formation of the active conformation. The F313A mutation caused an 80 -90% loss in activation of G t . The corresponding residue in PAR1 is Cys 378 ; however, the C378A mutation had minimal effects on PAR1-G q coupling.
Differential Activation of G q by Thrombin-activated Receptor versus Soluble Peptide-To our surprise, the Y381A,V382A,Y383A 8th helix mutation actually enhanced the EC 50 value and efficacy of PAR1 activation of G q upon stimulation with SFLLRN relative to WT. This rare effect was not detected upon activation with thrombin, demonstrating structural differences in intraversus intermolecular modes of activation, as previously observed with mutations of the extracellular surface of PAR1 (43) and PAR2 (53)(54)(55). A recent study (27) shows that the overlapping Y383A,L386A mutant has a 25% defect in extent of internalization. However, despite a slowdown in internalization, thrombin activation of PAR1-PLC-␤ signaling was not affected in the Y383A,L386A mutant (27) nor in the Y381A,V382A,Y383A mutant (Table 1), indicating that the PLC-␤ signal may not be sensitive to ϳ25% differences in internalization of PAR1. In many cases, the 8th helix mutations exhibited greater loss of coupling to G q by SFLLRN peptide as compared with thrombin. This observation is consistent with previous studies that indicate that peptide-activated WT PAR1 shows preferential coupling to G q , as compared with thrombin-activated receptor (48,49).