Monitoring Cannabinoid CB2 -Receptor Mediated cAMP Dynamics by FRET-Based Live Cell Imaging

Round 1

Reviewer 1 Report

The manuscript by Mensching et al. analyzed CB2 receptor cAMP signaling using FRET-based live cell imaging in a human tumor cell line.  This is not the first study to examine CB2 cAMP signaling, but the distinguishing characteristic is that the study analyzes dynamic responses in individual cells rather than relying on cell populations and terminal endpoints.   Using this method, the authors have made several notable observations.  In particular they report evidence that cell CB2 responses break down into discrete response profiles – ‘responders’, cells with constitutively active CB2 receptors, and non-responders.  Several studies have reported constitutive activity on the part of CB2 receptors, so the finding that there is variability even within ‘homogeneous’ cell lines is interesting and may go some way to explaining the different responses that have been reported for some compounds such as beta caryophyllene.

The manuscript is carefully written (though there are some awkward turns of phrase esp in the discussion) and so leaves little to critique. The authors do a nice job of providing background on the current state of the field of CB2 signaling and the figures (esp Figures 3 and 6) are clearly laid out.   

Some minor considerations:

A little more discussion of beta caryophyllene would be welcome.  The authors cite two papers that have reported an effect for BCP but do not mention recent work that challenges these findings for BCP (e.g. Finlay 2020 and Santiago et al. 2019).  One of these uses hCB2 as well.  The positive result for BCP is actually one of the more interesting aspects of this manuscript and additional discussion for ways to account for the differences would strengthen the manuscript. 

Also it should be noted that beta caryophyllene is not, strictly speaking, a phytocannabinoid (as claimed in the discussion) but rather a terpenoid. 

It is odd that the CB2 responses should fall into bins.  Though it is commonly assumed that cell lines such as HEK cells are homogeneous, they do exhibit variability across cells.   If the difference were solely a function of the degree of receptor over-expression then one would expect more of a spectrum of responses, with some having a mix of constitutive and non-constitutive responses.   Perhaps the authors could comment on this.

On a related point, I may have missed this, but do the authors report what percentage of cells are constitutively active vs. responders vs. non-responders?   Was this proportion consistent across cell cultures (for instance, passage number)?  It will be interesting to see whether this profile is maintained in cells that natively express CB2.   

Author Response

Answers to reviewer 1:

1. A little more discussion of beta caryophyllene would be welcome.  The authors cite two papers that have reported an effect for BCP but do not mention recent work that challenges these findings for BCP (e.g. Finlay 2020 and Santiago et al. 2019).  One of these uses hCB2 as well.  The positive result for BCP is actually one of the more interesting aspects of this manuscript and additional discussion for ways to account for the differences would strengthen the manuscript. 

We would like to thank our reviewer for his/ her comments. We have changed the manuscript accordingly and included a discussion of the BCP results considering the papers of Santiago and Finlay and highlighted our finding by mentioning it in the abstract.

2. Also it should be noted that beta caryophyllene is not, strictly speaking, a phytocannabinoid (as claimed in the discussion) but rather a terpenoid. 

This has been changed in the manuscript.

3. It is odd that the CB2 responses should fall into bins.  Though it is commonly assumed that cell lines such as HEK cells are homogeneous, they do exhibit variability across cells.   If the difference were solely a function of the degree of receptor over-expression then one would expect more of a spectrum of responses, with some having a mix of constitutive and non-constitutive responses.   Perhaps the authors could comment on this.

In our measurements we see that the individual cells have signals of different strengths that possibly reflect a spectrum of responses. For the figures, we then divided the cells into different categories as described in the methods and then averaged the response types in a recording. For the statistical evaluation, individual cell responses were taken into account by analyzing the data with a nested design.

4. On a related point, I may have missed this, but do the authors report what percentage of cells are constitutively active vs. responders vs. non-responders?   Was this proportion consistent across cell cultures (for instance, passage number)?  It will be interesting to see whether this profile is maintained in cells that natively express CB2.   

Because of the low number of cells that were analyzed per passage, we could not find obvious different ratios of the response types depending on the passage. It would be an interesting aspect to investigate in future studies. Particularly, we are interested to do the experiments on cells natively expressing CB2.

Reviewer 2 Report

This study describes single cell optical recordings of cannabinoid modulation of cAMP levels. The work is intricate and technically sound, although I am not sure what insights it provides into how CB2 receptors modulate this signalling system. It certainly lays the platform for future work. The use of transfected cells in necessary for the FRET sensor, it would be helpful to know the number of CB2 receptors in the HEK cells, and the proportion of the cells which expressed CB2 after creation of the stable cell line.

There are a few points that the authors should address, these do not speak to experiments themselves, but the often elaborate interpretation of the data reported.

Several studies have in fact used the CAMYEL BRET sensor to measure CB2 modulation of cAMP in real time, although this was in populations of cells. These studies should be acknowledged, and the statement that no one has done it before amended accordingly (line 473). This does not detract from the main novelty of this study, which is the single cell aspect.

The authors repeatedly assert that they are using supramaximal concentrations of agonist, but they offer no evidence for this, and based on other studies of CB2 function this is not a supportable contention. The affinity of the compounds for CB2 receptors in binding assays likely have no resemblance to functional affinities of the drugs in whole cells. To a degree this is not important for this paper as no concentration response curves are done, but the lack of a definitive “ceiling” for the inhibition of cAMP levels (because of the uncertainty around the concentration-effect profiles) renders the extension discussion of this in comparison with other systems and other receptors somewhat pointless.

The authors also discuss the (unexplained) differences in the initial effects of forskolin in the cells expressing CB2 when compared to the effects of forskolin in the FRET sensor alone cells. While effects of CB2 expression on G protein activity is a possible explanation, it should also be noted that the HEK/FRET/CB2 cells are in fact a subclone of 1 cell from the HEK/FRET population – the cellular background is not identical, and the cells themselves are cultured in different antibiotics (for selection). This should also be noted as a possible explanation for differences in cAMP dynamics, particularly as the involvement of constitutively active CB2 receptors is not tested (which could be done easily by addition of AM630 alone).

I am puzzled by the comment (line 74) that AM630 displays inverse agonism at “constitutively inactive” CB2 receptor forms. Given that ‘inverse agonism” is usually interpreted as block of constitutive activity, it is hard to understand how any drug could show anything other than simple antagonism at these receptors.

I think the comments about G protein coupling of CB1 and CB2 (line 45/46) are a little out of date – the coupling of CB1 to Gs and Gq has been reported by multiple labs, and recently evidence has been presented for CB2/Gs coupling in human PBMC.

Author Response

Answers to reviewer 2:

1. This study describes single cell optical recordings of cannabinoid modulation of cAMP levels. The work is intricate and technically sound, although I am not sure what insights it provides into how CB2 receptors modulate this signalling system. It certainly lays the platform for future work. The use of transfected cells in necessary for the FRET sensor, it would be helpful to know the number of CB2 receptors in the HEK cells, and the proportion of the cells which expressed CB2 after creation of the stable cell line.

We completely agree with our reviewer that it would be beneficial to perform immunocytochemistry for CB2 after the experiments to directly correlate the results with expression levels. But we have to consider that after the experiments with addition of three compounds and a live measurement for over 30 minutes, the cell constitution will not be optimal for a staining. We therefore did immunohistochemistry on our stable cell lines independently of the treatments. These results are presented in Fig. 1B.

2. Several studies have in fact used the CAMYEL BRET sensor to measure CB2 modulation of cAMP in real time, although this was in populations of cells. These studies should be acknowledged, and the statement that no one has done it before amended accordingly (line 473). This does not detract from the main novelty of this study, which is the single cell aspect.

We thank this reviewer for his/her valuable comments. In our introduction and discussion section we have added references of studies using the BRET method for cAMP detection. We did not intend to say that no one used this approach of biosensor live measurements for CB2 receptors before. Our aim was to mention that no one used the live cell imaging for the cAMP measurements for CB2 before. Unfortunately, I could not find a passage in the manuscript (line 474) saying that no one has done it before. However, we say at the end of the introduction:” To the best of our knowledge, this is the first description of investigating single cell characteristics of cannabinoids targeting the CB2 receptor.“ And indeed, we could not find publications showing that.

3. The authors repeatedly assert that they are using supramaximal concentrations of agonist, but they offer no evidence for this, and based on other studies of CB2 function this is not a supportable contention. The affinity of the compounds for CB2 receptors in binding assays likely have no resemblance to functional affinities of the drugs in whole cells. To a degree this is not important for this paper as no concentration response curves are done, but the lack of a definitive “ceiling” for the inhibition of cAMP levels (because of the uncertainty around the concentration-effect profiles) renders the extension discussion of this in comparison with other systems and other receptors somewhat pointless.

We agree with the reviewer and deleted the term “supramaximal” concentrations. Furthermore, the discussion section was shortened accordingly.

4. The authors also discuss the (unexplained) differences in the initial effects of forskolin in the cells expressing CB2 when compared to the effects of forskolin in the FRET sensor alone cells. While effects of CB2 expression on G protein activity is a possible explanation, it should also be noted that the HEK/FRET/CB2 cells are in fact a subclone of 1 cell from the HEK/FRET population – the cellular background is not identical, and the cells themselves are cultured in different antibiotics (for selection). This should also be noted as a possible explanation for differences in cAMP dynamics, particularly as the involvement of constitutively active CB2 receptors is not tested (which could be done easily by addition of AM630 alone).

We also performed AM630 only stimulations and could occasionally see a stronger answer but a clear overall answer was not recognizable. Because of the different response types, that actually makes sense. The problem is that only with AM630 stimulation we can determine whether a cell has constitutive activity or not, but we are not able to judge in which response type it falls into. Furthermore, the very little responses with AM630 have to be differentiated between the actual signal (which can be quite small) and signal changes over time due to bleaching / morphological changes etc.

All our cells (also control cells) went through the same selection procedure and all cell lines are cultured with the same concentration of antibiotics. The difference is just that the control cells were stably transfected with control plasmid without CB2. Unfortunately, we have not performed the experiments with a second (or third) cell line so far.

5. I am puzzled by the comment (line 74) that AM630 displays inverse agonism at “constitutively inactive” CB2 receptor forms. Given that ‘inverse agonism” is usually interpreted as block of constitutive activity, it is hard to understand how any drug could show anything other than simple antagonism at these receptors.

The term “active” and “inactive” is used in regard of the two-state model of receptor activation that proposes an existence of two interchangeable conformations: an "active" form, in which receptors are coupled to their effector mechanisms even in the absence of an agonist and an "inactive" form, that is not spontaneously coupled to receptor effector mechanisms. As an inverse agonist AM630 has a higher affinity to the “inactive” form of the receptor that can be activated by an agonist and it has a lower affinity to the “active” form of the receptor producing agonism or neutral antagonism. Please, see the details in the work of Dr. Pertwees lab (DOI:10.1111/j.1476-5381.2011.01503.x). Nevertheless, we have shortened this sentence now.

6. I think the comments about G protein coupling of CB1 and CB2 (line 45/46) are a little out of date – the coupling of CB1 to Gs and Gq has been reported by multiple labs, and recently evidence has been presented for CB2/Gs coupling in human PBMC.

We have adopted the G protein coupling sentences accordingly.