Based on the significant sequence similarity among them it was proposed that all the G-protein coupled receptors (GPCR) have a similar structural arrangement of the transmembrane helices. A recently proposed model of this arrangement¹ consists of ideal helices. This model is not able to accommodate simultaneously experimental data about helix-helix interactions involving helix 7: First, two interactions between residues in helices 7 and 2 have been proposed from mutation experiments on rhodopsin², the GnRH receptor³ and the 5HT2a receptor¹⁰. Second, a residue in helix 7 was shown to be facing the interior of the bundle by results from covalent labeling with a beta2-adrenergic receptor ligand⁴. These results serve as structural constraints in the construction of 3D molecular models of GPCR-s. They are incompatible with any model structure in which helix 7 is a regular alpha helix. We are proposing a specific conformational deviation from ideal helicity of transmembrane segment (TMS) 7 which involves the highly conserved motif NPxxY and offers a solution for incorporating all the experimental structural constraints. The proposed helix disruption is taken from results of a search of the protein structure database. The basic structural features of the NP/DP motif were extracted from the search and used to obtain a structure of TM segment 7 that fits conservation properties and structural requirements of the transmembrane helix bundle. The propensity of the NP sequence to form the NP/DP motif structure has also been studied computationally.

b) MC simulation of TM segment 7.

We have used a Monte Carlo - based procedure developed earlier^5,¹² to calculate the probability of the NP adopting the conformation identified for the NP/DP motif (see above) relative to that of a regular helix structure with a Pro-kink⁶. The results of the extensive calculations confirmed the large conformational prefence for the NP/DP motif structure.

c) MD simulation of a GnRH receptor model incorporating NP/DP motif in TM segment 7.

The proposed conformation of TM segment 7 has been incorporated in a model of the human GnRH receptor constructed on the basis of biophysical considerations described elsewhere⁷. The model was subjected to extensive MD simulation. The equilibrated model shows a "foot print" similar to the low resolution projection map of rhodopsin^8,¹¹. The proposed structure for TM segment 7 (see above) is fully consistent with available structural information for the entire TM helix bundle.

II. Numbering scheme

Using information from different GPCR-s for the construction of a model transmembrane structure creates the need for a general numbering scheme to identify positions in GPCR sequences independently of the numbering of residues in one particular sequence. The numbering scheme used here is based on the most conserved positions in each helix. Each identifier is composed of a number from 1 to 7 that identifies the helix and, separated by a period, a number associated with a position in that helix. The position number is given relative to the most conserved residue in that helix, which takes the number 50 (N in helix1, D in helix 2, R -from DRY- in helix 3, W in helix 4 and P in helices 5, 6 and 7). The other positions, going from the conserved residue to the C-terminus, are identified with increasing numbers, and with decreasing numbers going in the opposite direction (i. e., preceding the conserved position). The numbering scheme is illustrated in this figure on a helical net of the TMS 7 of the human GnRH receptor sequence. Click here for helical net of the whole receptor.

III. Experimental constraints for helix 7

Major experiments on helix 7 - distribution of involved residues:

Models incorporating a regular transmembrane helix 7 are not able to satisfy constraints from three sources of experimental data.

Rao et. al.² showed that the Schiff base counterion in rhodopsin can be repositioned from its wild type position 3.28 to position 2.57 in helix 2 while maintaining the protonation state of the Schiff base. The experiments suggest an interaction between Lys 7.43 and Asp 2.57 in addition to the wild type interaction between Lys 7.43 and Glu 3.28. Consequently, Lys 7.43 must be close to both helices 2 and 3.

Exploration of the GnRH and 5HT2a receptors by reciprocal mutations in TM segments 2 and 7^3,¹⁰ suggested that the residue in position 7.49 is close to the one in position 2.50. In the GnRH receptor, for example, function was lost in the Asn2.50Asp mutant, and recovered by the additional mutation Asp7.49Asn.

Position 7.40 was proposed to face the interior of the protein because Trp 7.40 was covalently labeled by a photoaffinity activated ligand on the beta2-adrenergic receptor⁴. In neurotransmitter GPCR-s Trp 7.40 is conserved.

Two experimentally proposed interactions are mutually exclusive in a regular helix 7:

Proposed interactions of helix 7 with other transmembrane domains are illustrated in Figure 1 on the anticlockwise arrangement corresponding to Baldwin's model and supported by data for rhodopsin.

If the three loci in helix 7 (7.49, 7.43, 7.40) were positioned in a regular alpha helix they would appear on different faces of the helix, spanning an arc of 180 degrees. All three residues will lie on the face of helix 7 predicted to be buried by the other helices. If helix 7 is positioned according to the predicted buried face in the transmembrane bundle, the Lys 7.43 residue is appropriately placed in the vicinity of helices 2 and 3. The Trp 7.40 residue faces helix 1, but Asp 7.49 faces helix 6 and not Asn 2.50.

Positioning Asp 7.49 in the vicinity of Asn 2.50, as in Figure 2 moves the residue in position 7.43 too far away from helices 2 and 3 to allow for the predicted interactions, and the position of Trp 7.40 appears to face the lipid environment contrary to the prediction.

The model of transmembrane segment 7 as an ideal alpha helix is thus inconsistent with the predicted interactions. The two residues predicted to interact with helix 2 (Lys 7.43 and Asp 7.49) span an arc of approximately 120 degrees so it is unlikely that any alternative arrangement of the helix bundle could solve this conundrum.

IV. Results

PDB search for helices containing NP or DP sequences

Why search for helices containing the NP or DP sequence?

To reconsider the assumption of an ideal alpha -helical structure for transmembrane segment 7, a likely structural modification would involve the conserved Pro in the NPxxY motif, because Pro is well known to disrupt helices. The coincidence of an Asn/Asp preceding Pro in the conserved NPxxY (DPxxY) motif may be responsible for additional structural features. A search of the protein structure database was carried out looking for xNP and xDP preceded and followed by four residues in an alpha -helical conformation.

Superpositions of structures containing an NP or DP sequence show common structural and H-bond pattern:

Fourteen different structures containing the NP, and fifteen different structures containing the DP motif were retrieved from the PDB using a search routine in IDITIS from Oxford Molecular. All but one of the DP caused a strong disruption of the helical structure in the NP/DP region. Only one DP was in a regular Pro kink conformation.

Superposition of the 14 NP sequence containing structures retrieved by the search. Only the backbones of Pro_i were superimposed. Superposition of the 15 DP sequence containing structures with a significant break in the helix.

The break usually results in a different relative orientation of the two helices before and after the NP or DP. The structures exhibit significant similarities. The Pro always begins the lower part of the helix and the preceding Asn or Asp flips from its helical conformation to satisfy with its side chain carbonyl the H-bond to the exposed backbone NH groups at positions i+1 and i+2 at the top of the helix.

Backbone dihedral angle plots suggest flexibility in the NP/DP regions:

Plots of Phi and Psi backbone dihedral angles of the NP and DP sequence containing structures.

Plots of backbone dihedral angles of the query sequence region show a wide distribution of Psi values of two residues preceding the Asn or Asp. The Phi dihedral angles of the two residues preceding Asn or Asp are clustered around values -90 and +60 degrees while Phi values of Asn or Asp show a wide distribution (-30 to -180) degrees. This findings suggest that the region may serve as a flexible hinge between helices preceding and following Pro_i, allowing them to assume the wide range of different relative orientations as seen in the structures.

Proposed structure of TM segment 7

NPxxY break in helix 7:

The structural inferences from the search of NP/DP motifs were incorporated in the model of helix 7 of a GPCR shown in the figure.

The Asn or Asp side chain carbonyl is hydrogen binding the NH groups of residues i+1 and i+2 from the Proline. The flexible region was adjusted to bring together the experimentally proposed interactions identified above.

Note that the Tyr from the NPxxY motif of helix 7 appears to have the specific role in the characteristic NP/DP structure of maintaining the parallel orientation of the upper and lower helices. Such an alignment of the helix portions is essential for efficient packing interactions within the helical bundle, and may be related to the high conservation of the residue among GPCR-s.

The NPxxY break in helix 7 brings together interactions proposed from experimental data:

The figure 3 shows a backbone trace of TM segments 1-7 of the GnRH receptor model. As oposed to figures 1 and 2 above this figure clearly demonstrates the ability of the proposed structure for TM segment 7 to accomodate the experimentaly suggested interactions.

Monte Carlo simulation of the TM segment 7

Methods:

We have performend MC simulations of the 5HT2a TM segment 7 with a temperature anealing - based procedure developed earlier^5,¹². The entire TM segment was included in the simulation, however, only backbone and sidechain dihedrals of residues 7.45 - 7.54 were submited to rotation during the procedure. The rest of the TM segment was kept rigid. To appropriately simulate the hydrophobic enviroment we have used a chloroform solvent model throughout our simulations. To ensure the complete exploration of the conformational space we have performed two simulations. The first simulation was started from the structure of a regular Pro-kink and the second started from the NP/DP - motif structure proposed for the TM segment 7 above. The two simulations produced very similar populations for the simulated dihedral angles regardless the diffrent starting points as depicted for one dihedral angle below.

MC simulation confirmed large preference of the NP sequence for the NP/DP - motif structure over regular Pro-kink:

The Pro-kink and the NP/DP motif structures significantly differ in the Psi dihedral angle of Asn or Asp preceeding the Pro residue. The Psi dihedral angle value of the residue preceeding the Pro in Pro-kink structures is close to a helical value of -50 degrees. In contrast, in all retrieved NP/DP - motif type of structures the values of this dihedral angle are distributed around 120 degrees, as can be seen in graphs above. This dihedral angle is responsible for the fliping of the Asn/Asp sidechain to a suitable position to H-bond the exposed backbone NH groups after the Pro in the broken helix. For steric reasons such a value is not permitted in regular Pro-kink structures.

The population probability graphs of the Psi backbone dihedral in the figures below show a much larger peak around value 120, corresponding to the NP/DP structure, compared to the peak around the value -50 which corresponds to the regular Pro-kink. This assymetry reflects the large preference of the NP sequence for adopting the NP/DP - motif structure. As expected of MC simulations, the outcome of the simulations should not depend on the starting structure if the exploration of the conformational space is complete. The two population graphs for simulations begining from Pro-kink and NP/DP motif structure show very similar behaviour therefore indicating the convergence of the runs.

Molecular Dynamics simulation of the GnRH receptor model incorporating the proposed structure for TMS 7

Methods:

The GnRH receptor model has been constructed based on biophysical considerations described elsewhere⁷. The model was minimized before heating up to 300K. We have run 1ns trajectory at 300K and after 600 ps the model reached equilibrium based on energy and RMSD criteria (see also Strahs D. and Weinstein H. poster). The simulation was performed with CHARMM 23 with a distance dependent dielectric. All hydrogen atoms parameters were used.

The GnRH model structure incorporating TMS 7 With the NP/DP-motif is compatible with rhodopsin projection map after extensive MD simulation:

The GnRH receptor model with the proposed structure for the TMS 7 was subjected to extensive MD simulation and the equilibrated structure seems to fit the frog and bovine rhodopsin electron density map obtained recently^8,¹¹.

V. Conclusions

We have constructed a model structure for transmembrane segment 7 of GPCR-s and introduced it into a model of the transmembrane portion of GnRH receptor.

The proposed structure of transmembrane segment 7 incorporates a deviation from helical structure in the conserved NPxxY region. The deviation from helicity produced by the NP/DP motif extracted from the database search agrees with recent results from NMR experiments⁹.

The modified helix 7 is shown to fit into the helix bundle packing scheme in a manner that satisfies all the constraints generated from experimental data for the interactions of transmembrane segment 7. This did not seem to be possible for a regular helix 7.

Calculations with the Monte Carlo simulated anealing procedure confirmed the large conformational preference of TMS 7 for the constructed NP/DP - motif structure compared to a regular Pro-kink structure.

VI. References

1. Baldwin, J. M. The probable arrangement of the helices in G protein-coupled receptors. EMBO J. 1993, 12, 1693-1703

2. Rao, V.R., Cohen, G.B., Oprian, D.D. Rhodopsin mutation G90D and a molecular mechanism for congenital night blindness. Nature 1994, 367, 639-641

3. Zhou, W., Flanagan, C., Ballesteros, J. A., Konvicka, K., Davidson, J. S., Weinstein, H., Millar, R. P., Sealfon, S. C. A Reciprocal Mutation Supports Helix 2 and Helix 7 Proximity in the Gonadotropin-Releasing Hormone Receptor. Mol Pharm 1994, 45, 165-170

4. Wong, S.K., Slaughter, C., Ruoho, A.E., Ross, E.M. The catecholamine binding site of the beta-adrenergic receptor is formed by juxtaposed membrane-spanning domains. J Biol Chem 1988, 263, 7925-7928

5. Guarnieri, F., Weinstein, H. Conformational memories and the Exploration of Biologically Relevant Peptide Conformations: An Illustration for the Gonadotropin-Releasing Hormone. J Am Chem Soc 1996, 118, 5580-5589

6. Sankararamakrishnan, R., Vishveshwara, S Characterization of proline-containing aloha-helix (helix F model of bacteriorhodopsin) by molecular dynamics studies. Proteins: Struct. Fun. Gen 1993, 15, 26-41

7. Ballesteros, J.A., Weinstein, H. Integrated methods for the construction of three-dimensional models and computational rpobing of structure-function relations in G protein-coupled receptors. Methods in Neurosciences. Academic Press, Sand Diego, CA, 1995, pp. 366-428.

8. Schertler, G. F. X., Villa, C., Henderson, R. Projection structure of rhodopsin Nature 1993, 362, 770-772

9. Berlose, J. P., Convert, O., Brunissen, A., Chassaing, G., Lavielle, S. Three-dimensional structure of the highly conserved seventh transmembrane domain of G-protein coupled receptors. Eur J Biochem 1994, 225, 827-843

10. Sealfon, S. C., Chi, L., Ebersole, B. J., Rodic, V., Zhang, D., Ballesteros, J. A., Weinstein, H. Related Contribution of Specific Helix 2 and 7 Residues to Conformational Activation of Serotonin 5-HT2a Receptor J Biol Chem 1995, 270, 16683-16688

11. Schertler, G. F. X., Hargrave, P. A. Projection structure of frog rhodopsin in two crystal forms. PNAS of the USA 1995, 92, 11578-11582

12. Guarnieri, F., Wilson, S. R. Conformational Memories and a Simulated Anealing Program that Learns: Application to LTB4. J Comput Chem 1995, 16, 648-653