Final Report Summary - CB1R ARRESTIN (Contribution of beta-arrestin-dependent receptor signaling to the physiological regulation of the endocannabinoid system)
The aim of the study was to analyze the regulation of cannabinoid receptors by beta-arrestin. The main focus was type 2 cannabinoid receptor (CB2R), especially the two known missense CB2 receptor mutants, Q63R and L133I. Q63R is overrepresented in patients with major depression, alcoholism, and autoimmune disorders, while L133I mutation has been shown to be overrepresented in patients with bipolar disorder. DNA construct have been made for energy transfer and other in vitro experiments, mutations have been introduced into the CB2 sequence. In the first run, receptors carrying Q63R or L133I have been tested for binding of beta-arrestin and G-protein coupling. CB2 receptor DNA constructs have been made for tests in confocal microscopy, BRET and other experiments. We found that Q63R mutant has increased, while L133I mutant has decreased beta-arrestin binding. G-protein activation turned out to be very similar for all variants (see Figures 1 and 2).
In order to screen the potential differences in the signalosomes of these mutant receptors, we decided to use a recently developed method for detection of interacting protein partners that uses promiscuous biotin ligase BirA(R118G). Promiscuous biotin ligase attached to proteins biotinylates interacting partners that can be subsequently isolated with avidin/streptavidin beads, and the method has been recently described with lamin B proteins. We tested whether the known CB2 receptor interactor protein, beta-arrestin, is specifically biotinalted by CB2 receptor fused with BirA-R118G (CB2BirA). We found that both overexpressed beta-arrestin-GFP and endogenous beta-arrestin can be purified with streptavidin beads when CB2-BirA was expressed in biotin-treated HEK cells. The beta-arrestin biotinylation increased after agonist-stimulation, and the biotinylation was increased with Q63R mutant and decreased with L133I mutant, compared to the wild type, in agreement with the results of BRET experiments. We also detected much higher levels of beta-arrestin biotinylation compared to overexpressed GFP alone, providing further evidence that the system is detecting specific interactions. In the next step, we overexpressed the CB2 receptor in HEK cells and isolated all biotinylated proteins from cell lysates. The protein samples were sent for MassSpec analysis in order to identify the interacting partners of CB2 receptors and hopefully to detect differences between the wild type and mutant CB2 receptors. The MassSpec analysis identified over 2000 interacting proteins, with a few dozen of differences between mutants. Initial analysis was done using method described previously (Silva JC et al. Cell Proteomics. 2006 Jan;5(1):144-56.). The most prominent difference appeared to be the increased binding of a protein PPT1B to Q63R mutant. However, we couldn't prove the increased binding by BRET or affinity purification, and we couldn't detect any effect of dominant-negative mutant PTP1B on the receptor's function. These results have lead to the reanalysis of the results. The initial analysis turned out to be sensitive to outlier peptide intensities, so we decided to redo the analysis by comparing the intensities of all peptides. We used python programming language, which allowed highly customizable analysis of the data. It turned out that PTP1B results were explained by a single peptide outlier intensity, otherwise the new analysis was in complete agreement with the experimental results, since there were no differences between the PTP1B binding of the three mutants. The new analysis also revealed a list of proteins that have the same binding patterns to the CB2 receptor variants as beta-arrestin. The list contains beta-arrestin2, and also ARBK1 and ARBK2 and a few known beta-arrestin binding partners such as spectrin. ARBKs are the proteins which facilitate beta-arrestin binding by phosphorilating the receptor. The list also contains a number of proteins involved in signaling, suggesting their involvement in beta-arrestin-dependent signaling. The exact role of these proteins have to be further validated.
With regards to the CB1 cannabinoid receptor, a few receptor mutants have been tested to see what are the structural requirements for beta-arrestin binding. It turns out, that the conserved DRY sequence in the second intracellular loop interacts with beta-arrestin, and mutating these residues resulted in receptors that selectively activate G-proteins or beta-arrestin. These mutants might be used as tools in future experiments in which they can be used to restore CB1R's function in CB1 knockout animals to explore the function of beta-arrestin dependent pathways in mouse models.
Another set of experiments was carried out to explore the significance of beta-arrestin binding of CB2Rs in the regulation of the locomotor activity in a Parkinson mouse model. It is known from recent publication that CB2 agonists inhibit cocaine-induced locomotion. We tested whether CB2 inverse agonist AM630 can improve locomotion or inhibit dyskinesia in these animals. In AM630-treated animals the locomotion was not different from vehicle-treated animals in DAT knockout mice at 25 mg/kg L-DOPA dose, but slightly decreased at 15 mg/kg. On the other hand, in DAT – beta-arrestin double knockout mice, the locomotion was significantly improved, while the dyskinesia was reduced compared to control group, and this was most prominent at the 15 mg/kg dose (Figure 3). This data suggested that beta-arrestin prevents the actions of the CB2 receptor that contribute to impaired locomotion and increased dyskinesia during chronic L-DOPA treatment. This might be of particular significance, since according our results, the two most frequent variations of CB2 receptor (63Q and 63R, approximately 50-50% percent in European population) have twofold difference in their beta-arrestin binding. Further analysis of the underlying mechanism should facilitate the understanding of pathophysiology of Parkinson's disease and might lead to development of new therapies which target the dyskinesia – the major side effect of L-DOPA treatment.
Although in recent years emerging data have suggested that the CB2 is involved in central nervous system functions, there is still an ongoing debate whether CB2 is actually expressed in neuronal cells, and little is known about how neuronal functions might be regulated by this receptor. To be able to map the cannabinoid effects on the signaling of neurons and glial cells involved in locomotion, we attempted to develop a method for purification of proteins where the procedure would allow cell specificity. This would potentially allow us to examine cell-specific signaling pathways and changes in protein expressions in animal models. To achieve this, we tagged the promiscuous BirA-R118G with nuclear export signal (NES-BirA) to target the protein into the cytoplasm. Since it has no other protein attached, we hypothesized that it would only interact with other proteins in non-specific way, so there would be no preference for biotinylation partners as it is the case with the previous approach, where the BirA was attached to the CB2 receptor. Since there are only random interactions we anticipated random biotynilations as well. When NES-BirA was expressed in HEK cells, it localized to the cytoplasm and with supraphysiological biotin concentrations we detected massive cytoplasmic biotinylation. We stained western blots for a several signal transduction proteins and in all cases the proteins were detected in the IPs of the NES-BirA expressing cells, but not in IPs from control cells. Also, we showed that EYFP alone is biotinylated when co-expressed with NES-BirA. The cell specificity was also tested with mixed cell lysates, either expressing NES-BirA and EYFP in the same cells or in separate cells. Although EYFP was present in both lysates, we could detect it only in the IPs where from BirA-coexpressed cells. This experiment showed that in vitro we can separate the proteins with different cell-origin even if the cell lysates are mixed, suggesting that the same method could work in the mouse brain as well if we selectively expressed the NES-BirA in cells with CRE-lox technique. This could allow us potentially to separately investigate the proteomes of D2 or D1 receptor expressing neurons in mouse striatum, and would also be useful not only for our goals, but also for other researchers in other areas. Understanding of protein expression profiles of specific cells instead of mixtures of cells in whole tissues should improve our understanding the function of many systems in more detail, and this can hopefully lead to discovery of novel therapeutic targets in the future.
To further develop the method for in vivo application to mouse models, we chose to develop two methods for initial tests. We modified the above construct so that it can be used for AAV virus production and also made a construct that can be used to make NES-BirA transgenic mouse. We made AAV8 and AAV10 viruses, and initially tested them on HEK cells where we were able to achieve good expression of our construct. For the transgenic mouse we generated a plasmid that contains the Rosa26 sequence long and short arms that allows us to target our construct to this locus. The Rosa26 targeting turned out not to result in sufficient expression in follow up studies in the outgoing host laboratory, but the AAV approach have been proved to be an efficient method to transfer the technology into mice. During the return phase a dedicated cell culture room has been set up in the host institute to allow safe production of AAV viruses, and AAV virus constructs have been purchased for generation of constructs. The AAV production technique will not only help the contractor in his future work, but will also facilitate the use of this approach by other researchers in the Institute. Constructs have been made for cell specific expression of BirA enzyme which will allow us to explore the cell specific signaling in mouse brain regions, and animal experiment protocol has been submitted for evaluation to proceed with in vivo studies.
In order to screen the potential differences in the signalosomes of these mutant receptors, we decided to use a recently developed method for detection of interacting protein partners that uses promiscuous biotin ligase BirA(R118G). Promiscuous biotin ligase attached to proteins biotinylates interacting partners that can be subsequently isolated with avidin/streptavidin beads, and the method has been recently described with lamin B proteins. We tested whether the known CB2 receptor interactor protein, beta-arrestin, is specifically biotinalted by CB2 receptor fused with BirA-R118G (CB2BirA). We found that both overexpressed beta-arrestin-GFP and endogenous beta-arrestin can be purified with streptavidin beads when CB2-BirA was expressed in biotin-treated HEK cells. The beta-arrestin biotinylation increased after agonist-stimulation, and the biotinylation was increased with Q63R mutant and decreased with L133I mutant, compared to the wild type, in agreement with the results of BRET experiments. We also detected much higher levels of beta-arrestin biotinylation compared to overexpressed GFP alone, providing further evidence that the system is detecting specific interactions. In the next step, we overexpressed the CB2 receptor in HEK cells and isolated all biotinylated proteins from cell lysates. The protein samples were sent for MassSpec analysis in order to identify the interacting partners of CB2 receptors and hopefully to detect differences between the wild type and mutant CB2 receptors. The MassSpec analysis identified over 2000 interacting proteins, with a few dozen of differences between mutants. Initial analysis was done using method described previously (Silva JC et al. Cell Proteomics. 2006 Jan;5(1):144-56.). The most prominent difference appeared to be the increased binding of a protein PPT1B to Q63R mutant. However, we couldn't prove the increased binding by BRET or affinity purification, and we couldn't detect any effect of dominant-negative mutant PTP1B on the receptor's function. These results have lead to the reanalysis of the results. The initial analysis turned out to be sensitive to outlier peptide intensities, so we decided to redo the analysis by comparing the intensities of all peptides. We used python programming language, which allowed highly customizable analysis of the data. It turned out that PTP1B results were explained by a single peptide outlier intensity, otherwise the new analysis was in complete agreement with the experimental results, since there were no differences between the PTP1B binding of the three mutants. The new analysis also revealed a list of proteins that have the same binding patterns to the CB2 receptor variants as beta-arrestin. The list contains beta-arrestin2, and also ARBK1 and ARBK2 and a few known beta-arrestin binding partners such as spectrin. ARBKs are the proteins which facilitate beta-arrestin binding by phosphorilating the receptor. The list also contains a number of proteins involved in signaling, suggesting their involvement in beta-arrestin-dependent signaling. The exact role of these proteins have to be further validated.
With regards to the CB1 cannabinoid receptor, a few receptor mutants have been tested to see what are the structural requirements for beta-arrestin binding. It turns out, that the conserved DRY sequence in the second intracellular loop interacts with beta-arrestin, and mutating these residues resulted in receptors that selectively activate G-proteins or beta-arrestin. These mutants might be used as tools in future experiments in which they can be used to restore CB1R's function in CB1 knockout animals to explore the function of beta-arrestin dependent pathways in mouse models.
Another set of experiments was carried out to explore the significance of beta-arrestin binding of CB2Rs in the regulation of the locomotor activity in a Parkinson mouse model. It is known from recent publication that CB2 agonists inhibit cocaine-induced locomotion. We tested whether CB2 inverse agonist AM630 can improve locomotion or inhibit dyskinesia in these animals. In AM630-treated animals the locomotion was not different from vehicle-treated animals in DAT knockout mice at 25 mg/kg L-DOPA dose, but slightly decreased at 15 mg/kg. On the other hand, in DAT – beta-arrestin double knockout mice, the locomotion was significantly improved, while the dyskinesia was reduced compared to control group, and this was most prominent at the 15 mg/kg dose (Figure 3). This data suggested that beta-arrestin prevents the actions of the CB2 receptor that contribute to impaired locomotion and increased dyskinesia during chronic L-DOPA treatment. This might be of particular significance, since according our results, the two most frequent variations of CB2 receptor (63Q and 63R, approximately 50-50% percent in European population) have twofold difference in their beta-arrestin binding. Further analysis of the underlying mechanism should facilitate the understanding of pathophysiology of Parkinson's disease and might lead to development of new therapies which target the dyskinesia – the major side effect of L-DOPA treatment.
Although in recent years emerging data have suggested that the CB2 is involved in central nervous system functions, there is still an ongoing debate whether CB2 is actually expressed in neuronal cells, and little is known about how neuronal functions might be regulated by this receptor. To be able to map the cannabinoid effects on the signaling of neurons and glial cells involved in locomotion, we attempted to develop a method for purification of proteins where the procedure would allow cell specificity. This would potentially allow us to examine cell-specific signaling pathways and changes in protein expressions in animal models. To achieve this, we tagged the promiscuous BirA-R118G with nuclear export signal (NES-BirA) to target the protein into the cytoplasm. Since it has no other protein attached, we hypothesized that it would only interact with other proteins in non-specific way, so there would be no preference for biotinylation partners as it is the case with the previous approach, where the BirA was attached to the CB2 receptor. Since there are only random interactions we anticipated random biotynilations as well. When NES-BirA was expressed in HEK cells, it localized to the cytoplasm and with supraphysiological biotin concentrations we detected massive cytoplasmic biotinylation. We stained western blots for a several signal transduction proteins and in all cases the proteins were detected in the IPs of the NES-BirA expressing cells, but not in IPs from control cells. Also, we showed that EYFP alone is biotinylated when co-expressed with NES-BirA. The cell specificity was also tested with mixed cell lysates, either expressing NES-BirA and EYFP in the same cells or in separate cells. Although EYFP was present in both lysates, we could detect it only in the IPs where from BirA-coexpressed cells. This experiment showed that in vitro we can separate the proteins with different cell-origin even if the cell lysates are mixed, suggesting that the same method could work in the mouse brain as well if we selectively expressed the NES-BirA in cells with CRE-lox technique. This could allow us potentially to separately investigate the proteomes of D2 or D1 receptor expressing neurons in mouse striatum, and would also be useful not only for our goals, but also for other researchers in other areas. Understanding of protein expression profiles of specific cells instead of mixtures of cells in whole tissues should improve our understanding the function of many systems in more detail, and this can hopefully lead to discovery of novel therapeutic targets in the future.
To further develop the method for in vivo application to mouse models, we chose to develop two methods for initial tests. We modified the above construct so that it can be used for AAV virus production and also made a construct that can be used to make NES-BirA transgenic mouse. We made AAV8 and AAV10 viruses, and initially tested them on HEK cells where we were able to achieve good expression of our construct. For the transgenic mouse we generated a plasmid that contains the Rosa26 sequence long and short arms that allows us to target our construct to this locus. The Rosa26 targeting turned out not to result in sufficient expression in follow up studies in the outgoing host laboratory, but the AAV approach have been proved to be an efficient method to transfer the technology into mice. During the return phase a dedicated cell culture room has been set up in the host institute to allow safe production of AAV viruses, and AAV virus constructs have been purchased for generation of constructs. The AAV production technique will not only help the contractor in his future work, but will also facilitate the use of this approach by other researchers in the Institute. Constructs have been made for cell specific expression of BirA enzyme which will allow us to explore the cell specific signaling in mouse brain regions, and animal experiment protocol has been submitted for evaluation to proceed with in vivo studies.
