Supporting Information

Size: px

Start display at page:

Download "Supporting Information"

Juliana McCoy
6 years ago
Views:

1 Supporting Information The Predicted Ensemble of Low-Energy Conformations of Human Somatostatin Receptor Subtype 5 and the Binding of Antagonists Sijia S. Dong, [a] Ravinder Abrol, [a, b] and William A. Goddard, III* [a] cmdc_015000_sm_miscellaneous_information.pdf

2 Figure S1. Hydrophobicity profile of hsstr5 before applying the capping rules. The residues expected to lie in the membrane are indicated by red dashed lines.

3 TM1! TM! SEQ: MEPLFPASTPSWNASSPGAASGGGDNRTLVGPAPSAGARAVLVPVLYLLVCAAGLGGNTLVIYVVLRFAKMKTVTNIYILNLAVADVLYMLGLPFLATQNAASFWPFG NEW_RAW: HHHHHHHHHHHHHHHHHHHHHHH HHHHHHHHHHHHHHHHHHHHHHHH NEW_CAP: HHHHHHHHHHHHHHHHHHHHHHHHHHHH HHHHHHHHHHHHHHHHHHHHHHHHHHHHH PORTER: ccccccccccccccccccccccccccccccccccccchhhhhhhhhhhhhhhhhhhhhhhhhhhhhhcccccchhhhhhhhhhhhhhhhhhchhhhhhhhhhcccccc SSPRO: ccccccccccccccccccccccccccccccccccchhhhhhhhhhhhhhhhhhhchhcheeeeeeeeccccccchhhhhhhhhhhhhhhhhccchhhhhhhhcccccc APSSP: --ccccccccccccccccccccccccccccccccccccccehhhhhhhhhhhhhhhhhhhhhhhhhccccccchhhhhhhhhhhhhhhhhhccchhhhhhhhcccccc APSSP: *999*** ********* ****8886*886 PSIPRED: ccccccccccccccccccccccccccccccccccccccchhhhhhhhhhhhhhhhhhhhhhhhhhhhcccccchhhhhhhhhhhhhhhhhhhhhhhhhhhhccccccc PSIPRED: JPRED: ccccccccccccccccccccccccccceeecccccccchhhhhhhhhhhhhhhhhhhhhhhhhhhhccccccchhhhhhhhhhhhhhhhhhhhhhhhhhhhhcccccc JPRED: OLD_RAW: HHHHHHHHHHHHHHHHHHHHHHH HHHHHHHHHHHHHHHHHHHHHHHH TM! TM! SEQ: NEW_RAW: NEW_CAP: PORTER: SSPRO: APSSP: APSSP: PSIPRED: PSIPRED: JPRED: JPRED: OLD_RAW: ATQNAASFWPFGPVLCRLVMTLDGVNQFTSVFCLTVMSVDRYLAVVHPLSSARWRRPRVAKLASAAAWVLSLCMSLPLLVFADV HHHH HHHHHHHHHHHHHHHHHHH HHHHHHHHHHHHHHHHHH----- HHHHHH------HHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH HHHHHHHHHHHHHHHHHHHHHH----- HHHHHHccccccHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHcccccccccccHHHHHHHHHHHHHHHHHHHcHHHccccE HHHHHHccccccHHHHHHHHHHHHHHHHHHHHEEHHEEHHHEEEEEccccccccccHHHHHHHHHHHHHHHHHHccccEEEEEc HHHHHHcccccccHHHHHHHHHHHccHHHHHHHHHHHHHHHHHHHHccccccccccccHHHHHHHHHHHHHHHHHHHHHHHccc ****8886* ** * ****88*** HHHHHcccccccccHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHEEcccccccccccccHHHHHHHHHHHHHHHHccEEEEEEE HHHHHHccccccHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHEEEEEccccccccccHHHHHHHHHHHHHHHHHHHHHHHHHEc HHHH HHHHHHHHHHHHHHHHHHH HHHHHHHHHHHHHHHHHH----- TM5! SEQ: QEGGTCNASWPEPVGLWGAVFIIYTAVLGFFAPLLVICLCYLLIVVKVRAAGVRVGCVRR NEW_RAW: HHHHHHHHHHHHHHHHHHHHHHHH NEW_CAP: HHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH PORTER: ccccccceccccccccchhhhhhhhhhhhchhhhhhhhhhhhhhhhhhhhcccccccccc SSPRO: ccccceeccccccchhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhccccccccc APSSP: ccccceececcccchhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhccccccccccc APSSP: **86779** ***989*99*98999** PSIPRED: cccccccccccccchhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhcccccccccc PSIPRED: JPRED: cccccccccccccccchhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh JPRED: OLD_RAW: HHHHHHHHHHHHHHHHHHHHHHHH TM6! TM7! SEQ: RSERKVTRMVLVVVLVFAGCWLPFFTVNIVNLAVALPQEPASAGLYFFVVILSYANSCANPVLYGFLSDNFRQSFQKVLCLRKGSGAKDAD NEW_RAW: HHHHHHHHHHHHHHHHHHHHHHHH HHHHHHHHHHHHHHHHHH NEW_CAP: --HHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH HHHHHHHHHHHHHHHHHHHHHHHHH PORTER: HHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHccccccHHHHHHHHHHHHHHHcHHHHHHHHccHHHHHHHHHHHcccccccccccc SSPRO: cchhhhhheeeehhhhhhhehchhhhhhhhhhhhccccccchhhhhhhhhhhhhhhhccchhhhhecchhhhhhhhhhhhccccccccccc APSSP: ccchhhhhhhhhehhhhhhhhcchhhhhhhhhhhhcccchhhhhhhhhhhhehhhhhccchhhhhhhchhhhhhhhhhhhccccccccccc APSSP: ** ** ***** PSIPRED: cccchhhhhhhhhhhhhhhccchhhhhhhhhhcccccchhhhhhhhhhhhhhhhhhhhhhhhhhhhcchhhhhhhhhhhcccccccccccc PSIPRED: JPRED: HHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHcccccccccccHHHHHHHHHHHHHHHHHHHHHHHccHHHHHHHHHHHHccccccccccc JPRED: OLD_RAW: HHHHHHHHHHHHHHHHHHHHHHHH HHHHHHHHHHHHHHHHHH Figure S. Secondary structure prediction for hsstr5 from various servers. NEW_RAW is PredicTM prediction; NEW_CAP is from consensus of different secondary structure prediction servers.

4 Table S1. Sequence similarity (in percentage) between hsstr5 and Class A GPCRs with experimentally available structures. The bold underlined cases were used as templates in this study. Protein Identifier Rank All TM Avg TM1 TM TM TM TM5 TM6 TM7 P56 SSR5_HUMAN P116 OPRX_HUMAN P866 OPRM_MOUSE P00 OPRD_MOUSE P115 OPRK_HUMAN P6107 CXCR_HUMAN P0699 OPSD_BOVIN P07700 ADRB1_MELGA P07550 ADRB_HUMAN P156 OPSD_TODPA P0817 ACM_HUMAN P97 AAAR_HUMAN P088 ACM_RAT Table S. Top 10 structures from the BiHelix/CombiHelix predictions for all 15 starting structures. [a] The lowest energy case for each template (shaded) was used for the SuperBiHelix step. [b] The Δη values are the deviations of helix (H) rotation angles from the respective templates. Method Δη [b] [ ] Energy [kcal mol -1 ] H1 H H H H5 H6 H7 CInterH CTotal NInterH NTotal E CNti moprm [a] moprm moprm

5 moprm moprm hoprk moprm hoprx moprm moprm Table S. Sampling space of the coarse SuperBiHelix for each of the three templates. The deviation angles apply to every helix (H) unless otherwise labeled. The angles apply specifically to one helix are from top 0 BiHelix results of the respective template. The starting structure of the sampling is the lowest energy case in BiHelix for each template (the shaded cases in Table S). Template Δθ [ ] Δφ [ ] Δη [ ] moprm 0, ±15 0, ±5, ±90 0, ±0, -60 (H6), -90 (H5, H6), -10 (H5), -150 (H5) hoprk 0, ±15 0, ±5, ±90 0, ±0 hoprx 0, ±15 0, ±5, ±90 0, ±0, ±10 (H5), 60 (H6), 90 (H6), -60 (H7), -90 (H7) Table S. Top 5 structures from fine SuperBiHelix/SuperCombiHelix with moprm as the initial template. Rank Δθ [ ] Δφ [ ] Δη [ ] E CNti [kcal mol -1 ] H1 H H H H5 H6 H7 H1 H H H H5 H6 H7 H1 H H H H5 H6 H

6 Table S5. Antagonists experimental binding constants and their corresponding calculated binding energies relative to M59. The binding energies are calculated according to equation ΔG 1 - ΔG = RTln(K i /K i1 ). Antagonist K i [nm] Relative Binding Energy [kcal mol -1 ] M M M8 11. M M >

7 a).5 RMSD (Å) b).5 RMSD (Å) Figure S. RMSD changes along the MD trajectory. a) The snapshots were aligned against the first frame, and RMSD values were calculated with the first frame as the reference. b) The snapshots were aligned against the last frame, and RMSD values were calculated with the last frame as the reference. Only backbone atoms were considered in calculating RMSD values.

8 1.8 cluster 1 cluster cluster cluster cluster RMSD (Å) Figure S. Results of clustering by RMSD along the MD trajectory. K-means algorithm was used. The K- means clustering radius is Å.

9 7 6 Distance between S97 HG1 and D86 OD1 (Å) 5 1 Figure S5. The fluctuation of interatomic distance between S O-donated H atom and D86.50 O atom on their side chains during the 50 ns MD simulation of M59-bound predicted hsstr5 structure. Distance between N58 HD1 and S97 O (Å) Figure S6. The fluctuation of interatomic distance between N N-donated H atom on its side chain and S O atom on its backbone during the 50 ns MD simulation of M59-bound predicted hsstr5 structure.

10 8 7 Distance between N81 HD and W16 NE1 (Å) Figure S7. The fluctuation of interatomic distance between N81.5 N-donated H atom and W16.50 N atom on their side chains during the 50 ns MD simulation of M59-bound predicted hsstr5 structure. 7 6 Distance between Y78 HH and D16 OD1 (Å) 5 1 Figure S8. The fluctuation of interatomic distance between Y78. O-donated H atom and D16.9 O atom on their side chains during the 50 ns MD simulation of M59-bound predicted hsstr5 structure.

11 .5 Distance between T15 HG1 and S167 OG (Å) Figure S9. The fluctuation of interatomic distance between T15.8 O-donated H atom and S167.5 O atom on their side chains during the 50 ns MD simulation of M59-bound predicted hsstr5 structure. 1 1 Distance between Y86 OH and N71 HD1 (Å) Figure S10. The fluctuation of interatomic distance between N N-donated H atom and Y O atom on their side chains during the 50 ns MD simulation of M59-bound predicted hsstr5 structure.

12 10 9 Distance between R17 NH and T7 O (Å) Figure S11. The fluctuation of interatomic distance between R17.50 N-donated H atom and T7 6. O atom on their side chains during the 50 ns MD simulation of M59-bound predicted hsstr5 structure. Visualization of this interaction is in Figure 7(a) of the main article.

13 9 Distance between M59 piperidine amine N and D119 OD1 (Å) Figure S1. The fluctuation of distance between the M59 piperidine amine N atom and D119. carboxylic acid O atom during the 50 ns MD simulation of M59-bound predicted hsstr5 structure.

14 18 16 Distance between M59 O in COO and R9 NH (Å) Figure S1. The fluctuation of distance between M59 carboxylic acid O atom and R9 1.1 amine N atom during the 50 ns MD simulation of M59-bound predicted hsstr5 structure.

15 18 16 Distance between R17 NH1 and E OE1 (Å) Figure S1. The fluctuation of distance between the R17.50 side chain N atom and E 6.0 side chain O atom during the 50 ns MD simulation of M59-bound predicted hsstr5 structure. After 0 ns, the intracellular end of TM6 including E 6.0 unwinds to accommodate the salt bridge between E 6.0 and R1 on the loop, and E 6.0 can be considered part of the loop. Additional Experimental Details 1. Preparing the initial helix shape 1.1 OptHelix This method treats each of the seven helices separately. It first takes the TM lengths predicted above, and generates seven separate canonical polyalanine helices accordingly. Then, it mutates the Pro and Gly back to their respective positions on the helices using Side Chain Rotamer Energy Analysis Method (SCREAM). A first structural optimization is then done to minimize the energy of each helix. Subsequently, the Ser and Thr adjacent to Pro are mutated back, and a molecular dynamics (MD) simulation on each helix is run for ns. Finally, all remaining residues are mutated back to have their original side chains, and a second energy minimization is performed. For each helix, the structures that go to the final step are selected based on minrmsd, which takes the snapshot that has the average root mean square deviation (RMSD) closest to the average structure from the MD, and based on mineng, which takes the snapshot that has the lowest energy from the latter 75% of the MD. 1. Homology modeling The template structures were taken from the Orientations of Proteins in Membranes (OPM) database. For each template protein, the sequence was aligned with that of the target protein hsstr5, and the corresponding residues in the template structure were mutated to be that of the target protein using SCREAM. Then each helix was truncated or extended to the previously determined start/end residues, which

16 was followed by a geometry optimization of each individual helix for 100 steps using the DREIDING-III force field. [1]. Obtaining ActiveConf Replacing TM6 shape in the starting structure by that from the active hadrb structure: The TMs 1,,,,5,7 of the starting structure and those of the active hβ AR x-ray structure were aligned using VMD. Only the backbone atoms were used in the alignment. Then, the TM6 in the starting structure was replaced by that from the active hβ AR structure. Finally, the residues on the new TM6 were mutated to hsstr5 residues using SCREAM. BiHelix/CombiHelix: After replacing the TM6 shape by that from the active hβ AR structure, another BiHelix/CombiHelix step (Δη from 0 to 60 with a step size of 0 ) was carried out before the fine SuperBiHelix. The rest followed BiHelix/CombiHelix described in the main text. Fine SuperBiHelix sampling: Starting from the best rotation angles from the above BiHelix/CombiHelix, the fine SuperBiHelix/SuperCombiHelix was carried out. The space sampled was: Δθ = 0, ±15 ; Δφ = 0, ±15, ±0 ; Δη =0, ±0. The rest followed SuperBiHelix described in the main text. References [1] S. L. Mayo, B. D. Olafson, W. A. Goddard, J. Phys. Chem. 1990, 9,

β1 Structure Prediction and Validation

13 Chapter 2 β1 Structure Prediction and Validation 2.1 Overview Over several years, GPCR prediction methods in the Goddard lab have evolved to keep pace with the changing field of GPCR structure. Despite