Continued from Part 1: Applying the XED molecular mechanics force field to the binding mechanism of GPCRs.
The ‘DRY’ lock
Within most but not all GPCR receptors, there exists an apparently ‘digital’ switch mechanism called the ‘DRY’ lock that can be used to monitor the state of the receptor. In the picture of Kobilka’s structure (figure 3), the ‘DRY’ lock location is labelled and consists of three closely associated amino acid residues; aspartic acid (D), arginine(R) and tyrosine(Y).
The X-ray structures have confirmed that when the β2AR is inactive, with no attached G-protein, the aspartate (as a charged acid) and the arginine (as a charged base) form a strong ion pair that closes the ‘DRY’ lock. The tyrosine has a secondary role to play in both the R and R* receptor forms but is not vital to the present argument.
On the other hand, Kobilka’s X-ray structure of the receptor binding its G-protein clearly shows that the DRY lock has opened and its arginine is helping to bind the G-protein to the receptor.
It is tempting to believe that the receptor action is a simple two-state digital switching of the lock. However, the proposal put forward in Part 1 of this report for the basic action of the β2AR implies that an analogue process may be occurring.
Clik here to view.
Figure 4: The ‘DRY’ lock that is conserved in many GPCRs: aspartic acid (D), arginine(R) and tyrosine(Y). On the far left is the inactive [R] receptor showing the closed lock. On its right is the active [R*] receptor with the lock open and holding the G-protein.
A digital or analogue process?
The overwhelming question at this stage was this: if Kobilka’s G-protein/β2AR complex was spontaneously deprived of its G-protein, would it directly revert to the conformation of the inactive receptor with closure of the ‘DRY’ lock?
In other words, is the inter-conversion from R to R* and back a simple on/off digital switch or does each small chemical and structural change along the path of conversion influence the properties and behavior of the next stable state?
A continuous process such as this is an analogue concept whereby the action of one change smoothly influences the action of another, or indeed several others!
Using the XED force field to check for an on-off switch
The XED force field was used to optimize the R* G-protein/β2AR complex. The G-protein was removed from the complex and the ‘bare’ R* receptor was re-optimized to see whether it reverted to the inactive R-form. It did not.
This was an indication that straight forward digital switching from R* to R was not happening and that the conversion between R* and R receptor forms was under the control of more subtle influences.
Considering the analogue alternative
The next step was to think about what may be going on and whether the resulting ideas were computationally addressable.
Since Kobilka had shown that the active R* receptor can only exist when the receptor is bound to its G-protein, the movement of the G-protein in and out of the receptor may furnish the conditions for R/R* conversion. Could it be that at each small stage during the exit of the G-protein from the receptor, the chemistry surrounding the focus of extraction was changing, preparing a path for the ‘DRY’ lock to close, for the change of helix 6 (H6) conformation and other so far unseen subtleties important to the whole process?
A detailed proposal of the analogue process
From the data available at this time (late 2012), a proposal of what might be going on was formulated and is illustrated below (figure 5).
Clik here to view.
Figure 5: A proposal of the effects of the presence or absence of agonists and antagonists on the ‘DRY’ lock.
Constitutive agonism, (no bound ligand), is speculated on the left hand side of figure 5, with slow uptake of the G-protein shown in picture 1 (~200 times slower than when an agonist is present) and a comparatively slow loss of the G-protein in picture 2 of figure 5.
On the right side, picture 3 represents a fast loss or even exclusion of the G-protein when an antagonist is bound but a much slower loss of G-protein when an agonist is bound.
Furthermore, because of the subtleties inflicted on the electrostatics and sterics of the receptor by the bound ligand, the on/off rate of the G-protein will vary according to the ligand structure. It is well established that different drugs that are active at the β2AR cause the GPCR to activate different pathways within the cell.
Figure 6 outlines the pharmacology carried out by Daniel Hothersall at University College London (UCL) under Professor Andy Tinker and Sir James Black in 2009. It is likely that these variations in path choice may be linked to other factors such as the activation or extent of phosphorylation. But however complex the mechanisms are, they still remain under the ultimate influence of the structural minutiae of the small neurotransmitters adrenaline and noradrenaline and the many accompanying synthetic agonist and antagonist variations.
Clik here to view.
Figure 6: Pharmacology carried out by Daniel Hothersall at UCL under Prof Andy Tinker and Sir James Black in 2010 showing the varied paths that can be activated by the β2AR in the presence of different drugs.
The timescale challenge of modeling an analogue process
From previous published data, the timescales of action of many receptors is relatively slow compared with molecular movements in solution or atomic movements on a molecule. The time scales are certainly slower than any form of available calculation technique such as molecular dynamics, which would be most appropriate for this type of investigation.
The problem with molecular dynamics is that we don’t have the computer power to calculate the full process. Some of these receptor responses have been measured to take as long as milliseconds or even a whole second. Molecular dynamics deals with femtoseconds. To investigate a millisecond of movement on current computing power would take 90 years on a system as big as this receptor! We can get to nanoseconds, but even that takes several months of computing power.
In order to tackle the computation of an ostensibly ‘analogue’ process, the assumption has to be made that as one thing separates from another, it does so with the least waste of energy. This means that as two bound molecules are gradually pulled apart, angstrom by angstrom, each stage has to be at its local minimum energy. Despite the presence of small energy barriers along the way, a map of the local minima along the exit path should reveal details of the receptor behavior as a whole.
It follows that the complete structure of the whole system has to be computationally examined at every stage.
The creation of a test-bed for this assumption is now a matter of computational engineering. The work has been carried out and the method and results are being prepared for publication.