Constrained dynamic dock.

[primaryodors]

Develop a variation of the "soft dock" feature that translates, pivots, and bends transmembrane helices in limited ways to arrive at optimal ligand contact.

Based on the internal contacts identified in #292, the following residues will serve as "pivot points" or fulcra:

1.60
2.42
4.54
6.48
7.66

Transmembrane helix 3 does not have a fulcrum and will not normally be allowed to pivot or bend. Helix 5 may pivot at its extracellular end, or with the pivot of helix 6. Otherwise, each fulcrum will serve as a stationary point for pivoting, or as a starting point for bending, e.g. TAARs whose TMR6 region bends outward at both ends with Trp6.48 remaining in place.

Allowed motions will be specified in the .config file for a dock. Biases will also be included so that PrimaryDock will prioritize conforming the receptor in the indicated ways if energetically advantageous. Some motions will be dependent on others and limited to an equal or lesser motion.

As an example, the motions and biases for fish-like ORs will be:

• TMR6 pivots CYT end outward and EXR end toward TMR5 about residue 6.48;
• TMR6 unravels slightly between 6.57-6.60, giving 6.59 better access to the binding pocket;
• TMR5 moves sideways toward TMR6 and pivots slightly to bring CYT end near TMR6;
• The EXR2 helix bends at 45.54 to bring 45.53 a little over 1A in the CYT direction;
• TMR4 pivots slightly in any direction about 4.54;
• TMR7 bends EXR end outward from 7.66 dependent on TMR6's pivot;
• TMR1 pivots about 1.60 to bring EXR end away from TMR2;
• TMR2 bends at 2.42 to bring its EXR end closer to TMR1, dependent on TMR1's pivot.

[primaryodors]

There are four types of motion to develop code for: rock, bend, move, and wind (as in rewind, not the weather phenomenon), the unwinding of the EXR end of TMR6 being a wind with a negative amount. However, a bend can be considered a type of rock where the fulcrum is the same as one of the termini, so it is sufficient to code only rock, move, and wind motions.

The syntax in the .config file might be:

DYNAMIC ROCK pivot6 6.27 6.59 6.48 2.54 -15
DYNAMIC MOVE xlate5 5.32 5.68 5.49 6.48 2.8
DYNAMIC ROCK bend45 45.52 45.54 45.54 4.64 25

The first of these defines a rocking motion named "pivot6" which rotates the entire TMR6 helix about an axis extending across 3D space between the CA atoms of residues 6.48 and 2.54, with a bias of 15 degrees CCW.

The second line defines a translation of TMR5, moving the entire helix in the 5.49 -> 6.48 direction, with a bias of 2.8 Angstroms.

The third line defines a bend, which is a rocking motion of 45.52 thru 45.54, about an axis from 45.54 to 4.64, with a bias of 25 degrees CW.

Dependency constraints can be notated with optional MIN and MAX suffixes, e.g.:

DYNAMIC ROCK bend7 7.30 7.46 7.46 3.38 -4 MAX pivot6
DYNAMIC WIND unwind6exr 6.57 6.60 -25 MIN pivot6

Both of these lines list the "pivot6" motion as a constraint. The first is limited to a maximum that is proportional to pivot6's motion. E.g. if the protein model in memory has only 80% of pivot6's motion applied, then bend7 can only be applied up to a maximum of 80%.

The second line specifies a winding motion of 6.57 through 6.60 by a bias of -25 degrees, meaning each residue of the helix is unwound by a target amount of 25 degrees. This line has a MIN constraint, meaning for whatever percentage of pivot6's motion has been applied to the protein model, at least that same percentage of unwind6exr's motion must also be applied.

[primaryodors]

Whole PDB output will be necessary to identify the ways the current code is malfunctioning.

[primaryodors]

There is a problem with this methodology vs. the OR1A1 + d-limonene test. Specifically, the ligand cannot fit into the binding pocket unless the clash cubic Angstrom constant is set to less than the equivalent of 1 kJ/mol. (It was set to 0.5 all along, but that caused too many clashes in the dock results, so for this issue it has been changed to 1.0.) Raising the energy limit to 25 kJ/mol resolves the problem in the inactive state.

However, in the active state, a test of a dynamic dock on OR1A1 containing d-limonene, using all the same parameters as OR51E2 + propionate, except for the TMR6 unwind which is presumably not applicable to class II ORs, shows that the crucial TMR6 pivot cannot happen at all; its total applied value is 0.000000 for 3 out of 5 output poses, including the only pose within 25 kJ/mol. OR1A1 may therefore exhibit a different activation mechanism than OR51E2, and may require an outward, rather than inward, motion of the EXR portion of TMR6, similar to the curling motion of TMR6 in mTAAR9.