A process called protein phosphorylation by an enzyme called cAMP dependent protein kinase A may be at the heart and soul of the ever elusive entourage effect of terpenes and THC in cannabis …. or maybe the overlooked intersection of terpenes in our diets and our endocannabinoids.
A 1994 study
This study came out of Frankfurt, Germany. The investigators used rat retinal neurons.  The investigators examined all sorts of ligands for G protein coupled receptors (GPCR). They used patch clamp techniques to examine the augmentation by these agonists on control currents due to GABAA chloride channels. 
|control||1.02 ± 0.07µM||5|
binds dopamine GPCR
|50µM||2.08 ± 0.10µM||5||*|
partial D1/D5 agonist
|50µM||1.95 ± 0.08||6||*|
+ protein kinase inhibitor
|400 ug/ml||1.11 ± 0.04 µM||5||NS|
Dopamine receptor agonist
|50 µM||1.10 ± 0.06||5||NS|
|cAMP||product of GPCR||100µM||2.07 ± 0.20||5||*|
|cAMP+ PKIP||+ protein kinase inhibitor||400 µg/ml||1.09 ± 0.06||5||NS|
|PKA||activated by cAMP||35 units/ml||2.07±0.13||5||*|
cAMP by adenylate cyclase
|PMA||activates protein kinase C||00µM||1.07 ± 0.08||5||NS|
|GTP[yS]||non hydrolyzable form GTP||100 JAM||2.61 ± 0.20||5||*|
activates adenylate cyclase
|1µg/ml||2.26 ± 0.14||5||*|
Various ligands of GPCR were also tested. A small caveat is that agents like noradrenaline may bind to multiple GPCR.
This story continues with a study by David Hinkle1 and Robert MacDonald in 2003. The goal was to determine the role of phosphorylation of subunit phosphorylation in turning GABAA off.
β1, β3, and cAMP
Recall that CB1 receptor ligand binding has the effect of turning off cAMP production. In Figure 1 of Hinkle & MacDonald 2003 cAMP was introduced to the patch clamp pipette on the inside of the HEK203 cells. 
In this first set of experiments addition of cAMP to the inside of the cells made no difference in the time it took to return to the initial channel conductance (at the voltage they were clamped at). Addition of an enzyme called protein phosphatase 1 (PP1) did however slow the return to the resting membrane potential, or at least the first phase of the return. Maybe cAMP made no difference because the GABAA channels were already fully phosphorylated. Many hormones in serum used to grow cells fire through hetero trimeric G protein coupled receptors that activate cAMP production. What happens when HEK293 cells are serum starved?
It would seem that factors in the serum result in endogenous phosphorylation that speeds up the first phase of the return to resting membrane potential and then speeds up the 2nd phase in both β1 and β3 versions of GABAA. 
What am I looking at?
The Fast downward line is the current of Cl– ions flowing into the patch clamped HEK293 cell upon the addition of GABA. The curved upward line is the flow slowly, or quickly, being shut off. The following figure was created to illustrate a few things :
Left panel A typical exponential decay equation. The Greek letter tau ( τ ) is the decay constant The center graph was used to illustrate the point that the smaller τ, the faster the decay, or shutting off of the GABAA channel. Hinkle & MacDonald also created a term called the weighed decay constant.  This is the equation in the right panel
Forskolin, an Indian medicinal herb and cAMP activator
Forskolin, popular as an Indian medicinal herb, is also used by scientists as means of elevating cAMP production by the cell. Dideoxyforskolin is the inactive analog.  Naturally, cells in this experiment were serum starved. The red lines have been added to the original figure to indicate what appears to be a blow up of the second phase.
The importance of phosphorylation
cAMP activated protein kinase A attaches phosphate groups to amino acid serine and threonine-OH groups. Phosphate groups have a negative charge. Mutating a serine to an alanine means that this serine is incapable of being phosphorylated. In theory, this would mimic the effect of turning off the production of cAMP by activation of CB1. Mutating a serine to a glutamate would, in theory, mimic the effect of constant phosphorylation.
This table from the supplemental data is a good place to stop.  Hinkle and MacDonald conducted an elaborate set of experiments that led them to a working model of how phosphorylation of this cytoplasmic loop controls GABAA function. What makes this study interesting (and confusing) is that nothing is as simple as phosphates being on/off switches as they seem to be in other proteins. Much of this study truly withstood the test of time. Close to 20 years later another study considered the process of removing said phosphates with a class of enzymes known as phosphatases.
Running down GABAA with too much GABA
Cifelli and coauthors were interested in a model of epilepsy in which too much neurotransmitter GABA release leads to the running down of the GABAA receptor. In this Xenopus oocyte model whole cell patch clamp was used to monitor six successive supra maximal stimulation every 40 sec. As Figure 1A of the publication shows, each successive dose results in less current.
Some sequences of putative phosphorylation sites have been included with UniProt reference codes. Can we attribute these results to something unique to amphibian kinases and phosphatases? The basics of these experiments were repeated in human embryonic kidney cells (HEK293). Cifelli and coauthors went back to the oocyte system and demonstrated that the addition of a phosphatase inhibitor, okadaic acid, preserved the current intensity by GABA addition #6 as a percent of addition #1. 
Use of phosphatase inhibitors was proposed as a treatment for epilepsy, temporal lobe epilepsy in particular.
Wait! These data are inconsistent with use of cannabis to treat epilepsy!
On the surface, nothing in  is consistent with cannabis products being used to treat epilepsy. The 1994 paper by  demonstrated that just about all ligands of GPCR that couple with Gαs can activate GABAA isoforms found in the rat retina in a process that is totally consistent with phosphorylation by cAMP dependent PKA.  A few years later one of may studies pin pointed the phosphorylation sites to the cytoplasmic domain of the β sub unit.  Then a third study proposed the use of phosphatase inhibitors to treat “run down” of the GABAA receptors that occur with out of control GABA release seen in epilepsy.  A recent review out of Rome Italy and Amsterdam Netherlands emphasized that THC is acting to prevent excess GABA release from pre-synaptic neurons.  Minor cannabinoids, but not THC, may also act as positive allosteric modulators of GABAA. 
GABAA shutting down one sub unit at a time
Perhaps GABAA receptors simply start breaking down and/or undergo conformation changes when they are over stimulated with GABA. Perhaps some of the terpenes that act as positive allosteric modulators can also stabilize the receptor. A structural study mutated key amino acids in all five sub units of a variation of GABAA.  This wasn’t random over expression of three sub units. The entire receptor was expressed as a complete unit with sub units connected with polyglutamine linkers  The highly homologous 3rd transmembrane helix M3 was chosen for study, specifically asparagines around position 303. 
These mutations that altered the rate of shut down were located on the cytoplasmic side where sub 5] Three models were considered: (1) one sub unit can shut down GABAA, (2) two sub units working independently can shut down GABAA, and (3) shut down requires a concerted interaction of adjacent sub units. While ignoring the intricate details, this 3rd model of this terpene site because many terpenes have been predicted to bind at sub unit interfaces
It would be impossible to do the Gielen 2020  study on this post. This group did give us a working model of GABAA receptor desensitization starting with the γ2 sub unit that swings inward. It is later joined by α1 sub unit…or vice versa. β2 then follows to fully inactivate GABAA. How does cAMP activated PKA phosphorylation of β2 affect this model? This of course relates to THC binding to CB1 shutting down cAMP production. And what about terpenes? so many of them are predicted to bind to one sub unit or another.. or the interface of two sub units! If the terpene THC entourage effect is to be found, GABAA modulation may be the place to look for investigators who have patience. These receptors may take minutes to run down.
- Feigenspan, A., & Bormann, J. (1994). Facilitation of GABAergic signaling in the retina by receptors stimulating adenylate cyclase. Proceedings of the National Academy of Sciences of the United States of America, 91(23), 10893–10897. PMC free article
- Hinkle, D. J., & Macdonald, R. L. (2003). Beta subunit phosphorylation selectively increases fast desensitization and prolongs deactivation of alpha1beta1gamma2L and alpha1beta3gamma2L GABA(A) receptor currents. The Journal of neuroscience : the official journal of the Society for Neuroscience, 23(37), 11698–11710. PMC free article
- Cifelli, P., Di Angelantonio, S., Alfano, V., Morano, A., De Felice, E., Aronica, E., Ruffolo, G., & Palma, E. (2021). Dissecting the Molecular Determinants of GABAA Receptors Current Rundown, a Hallmark of Refractory Human Epilepsy. Brain sciences, 11(4), 441. PMC free article
- Cifelli, P., Ruffolo, G., De Felice, E., Alfano, V., van Vliet, E. A., Aronica, E., & Palma, E. (2020). Phytocannabinoids in Neurological Diseases: Could They Restore a Physiological GABAergic Transmission?. International journal of molecular sciences, 21(3), 723. PMC free article
- Gielen, M., Barilone, N., & Corringer, P. J. (2020). The desensitization pathway of GABAA receptors, one subunit at a time. Nature communications, 11(1), 5369. PMC free article