Structural Impact of RNA Editing

The 5-HT2C receptor is expressed in different isoforms as a result of mRNA editing (Niswender et al. 1998). Both INI (unedited) and VSV (a fully edited version) isoforms are the results of the variations I156V occurring in position 3.54 in TM3 from our ADBR-based TM prediction and N158S and I160V in the following IL2, well conserved in type-A GPCRs (Fig. 6.5). In regard with the multiple sequence alignments (Fig. 6.6), position 3.54 is dominantly occupied by an isoleucine residue but is a valine in bovine rhodopsine. In the same manner, the two other positions in IL2 are homologous with the outstanding serine variation instead of N158 and hydrophobic intrinsic properties conserved for the position 160. Investigation of the pharmacology for the agonist binding site of these two isoforms of the 5-HT2C receptor has shown that the INI isoform of the 5-HT2C receptor is pharmacologically similar to the VSV form but it couples more efficiently to G-proteins (Quirk et al. 2001).

It is possible to rationalize the structural impact of mutations on the thermody-namic stability of a system in buried regions or in the vicinity of ligand binding sites

Fig. 6.6 Location of key residues for signal transducing in the ADRB-based 5-HT2C model. a-Carbon of key amino acid residues (represented as spheres and discussed in the text) to transduce the signal from the ligand binding pocket to the cytoplasmic regions are divided into five classes : (1) network of residues conserved among amino acids participating to the signal pathway revealed by Kong et al. (Kong and Karplus 2007) (red); (2) hydrogen bond network (green) assisted by structural waters (additional information in Fig. 6.3); (3) hinge residue (black) between the two previous networks; (4) variant amino acids from RNA editing isoforms (yellow); (5) amino acid residues involved in the agonist-promoted movement of the cytoplasmic part of TM6 (cyan). Docking of serotonin in the ligand-binding pocket shows polar interactions with the three amino acids residues that could be the starting event for the agonist-promoted signaling (For interpretation of the colors in this figure, the reader is referred to the web version of this chapter)

Fig. 6.6 Location of key residues for signal transducing in the ADRB-based 5-HT2C model. a-Carbon of key amino acid residues (represented as spheres and discussed in the text) to transduce the signal from the ligand binding pocket to the cytoplasmic regions are divided into five classes : (1) network of residues conserved among amino acids participating to the signal pathway revealed by Kong et al. (Kong and Karplus 2007) (red); (2) hydrogen bond network (green) assisted by structural waters (additional information in Fig. 6.3); (3) hinge residue (black) between the two previous networks; (4) variant amino acids from RNA editing isoforms (yellow); (5) amino acid residues involved in the agonist-promoted movement of the cytoplasmic part of TM6 (cyan). Docking of serotonin in the ligand-binding pocket shows polar interactions with the three amino acids residues that could be the starting event for the agonist-promoted signaling (For interpretation of the colors in this figure, the reader is referred to the web version of this chapter)

according to differences of hydrophobic, polar or volume intrinsic properties induced by the mutation. Thus it appears that such solvent-exposed variations would be well tolerated because there are no significant differences of these physical characteristics. However, this intracellular exposed region is in fine in the cytosol and hydrophobic positions 156 and 160 would be detrimental in this aqueous environment. Since intracellular IL2 and IL3 regions are suggested to be bound to G proteins, these positions could be a specific hydrophobic binding site for G-proteins.

Consequently variations would not change the overall thermodynamic stability of the 5-HT2C receptor and its ligand binding dynamics but would have incidence on the specific binding of G proteins and subsequent activity.

6.5 Conclusion

In the rational approach of drug design, computational tools have become increasingly useful. Numerous examples illustrate their impact in increasing the speed of drug discovery by helping in the generation of pertinent hypothesis. It should be still remembered that computational tools are mere tools, that is they can greatly help in the process of trials and errors to achieve a desired result but should not be substituted to an integrated research effort and should not confused with some kind of oracle always telling the truth without questioning their conclusions by trials. The more complex the problem to solve is, the harder it is to find a solution and the more integrated the various methods of investigation must be. In this context, the quest of selective 5-HT2C ligands is an example of the difficulties encountered in drug design and is therefore exemplary of the close imbrications of the various means to shed a new light on the same facts. We have described here the building of a series of static images of the not yet totally grasped dynamics of GPCRs activation, with the secondary but important problem of designing selective compounds. Although it is relatively easy to generate a model of a GPCR, given the right template, whose choice was fairly limited up to very recently and, most importantly, a good sequence alignment, it is another task to validate the model. Three steps must be adhered to in order to assure a good coherency of the final image of the receptor. First, the alignment must conserve as much as possible the characteristic sequences of the GPCR super-family. Second, the structural elements of the putative model must be checked to verify they are physically plausible, in particular the relation between the helical domains and their lipidic environment. Third, the structurally valid model must comply with the experimental data gathered by in vitro experiments and observed in vivo responses. In the case of 5-HT2C, the two first points are classical for the class A GPCR modeling. The third point is the clearest bottleneck in the building of the models, as the data are still relatively few. Due to the implication of the human cleverness and intuition in this process, as opposed to the two first steps, it should be coined tailoring of the model. In particular, rendering in silico the difference between the activated agonist-binding forms, constitutively active basal form and inactivated inverse agonist-binding forms require more facts on the residues implicated in the binding of the ligands and the atomic events giving rise to the signal transduction through the cellular membrane than currently available to build a model rather than to imagine a general scaffold of the receptor. The structural features of 5-HT2C are fairly well understood, even the length of the helical domains, give or take a few amino acids. The binding site has been approximated with a relative precision by computational methods for a fairly large selection of compounds of various degrees of affinity and different activity and the studies reported here are highly coherent despite their very different approaches. This shows the intrinsic power of computational tools for these common tasks of target-based knowledge generation. To the experience of the authors, in a short time span, the possibilities of developing a plausible model of 5-HT2C have greatly increased, partly due to the ever increasing sharpness of the available tools, but also, and for the most part, due to the ever growing body of data, collected from the compilation of the various investigations on this receptor, either by in vivo, in vitro, or in silico means.

It remains yet two equally interesting points. We could wonder if the models are even remotely close to a biologically sound reality. Here, the computational scientist should remain humble, propose some structural features that could be targeted with ease, by example by directed mutagenesis or design of a ligand with adequate substitution pattern, and hope that the experimental results will corroborate his own personal feeling, largely transcribed in the model he has put forth. It should be kept in mind that, by definition, a model is an oversimplification of the natural object in order to simulate its behaviour, and therefore should not be regarded as exact. It is the role of the modeler to find this delicate equilibrium between two goals. The one part of the equilibrium is a model that will be used to generate hypothesis of ligands interactions, usable to create new ligand structures. Such a model must be robust, that is not focused toward a precise set of ligand-receptor interactions. The other part of the equilibrium is a model that shows something of the dynamics of the ligand-receptor complex, that is a model tuned to represent as much of the known data about a ligand as possible, but that will have necessarily lost its capacity to be engaged in interaction with other ligands. This equilibrium is the key point of homology modeling, and the shift from hand-made models built around a ligand toward routine molecular dynamics simulations of the 5-HT2C complexes for larger series of structurally dissimilar compounds is a clear sign that it is still the hardest point to solve, in particular due to the unclear configurations of the different activation states of the GPCRs. The second interesting point that appears is that the current models, despite their possible flaws, have nonetheless provided some critical insights into the interactions formed by 5-HT2C ligands, among which some are selective. Comparing these insights with similar data from other GPCRs, or more directly comparing the residues in the structural elements of the GCPRs that interact with the ligands has generated a very large number of ideas. In turn, these ideas could be turned by experimental validation, including design of new ligands, into knowledge, which will contribute to a better characterization of the 5-HT2C subtype binding requirements. This will lead to some hard facts about its selectivity toward its congeners, given some more team works of biologists, modelers and chemists.

Acknowledgements We acknowledge the Institut de Recherche Servier for granting us the use of their 3-dimensional visualization tools, TriposBenchware (Tripos®) and DiscoveryStudio2.1 (Catalyst®), and specially thank A.Gohier for discussions, suggestions, and help with modeling and illustrations.

Defeat Drugs and Live Free

Defeat Drugs and Live Free

Being addicted to drugs is a complicated matter condition that's been specified as a disorder that evidences in the obsessional thinking about and utilization of drugs. It's a matter that might continue to get worse and become disastrous and deadly if left untreated.

Get My Free Ebook


Post a comment