Dissertation committee:
Σταύρος Χαμόδρακας, Καθηγητής ΕΚΠΑ (επιβλέπων), Παντελής Μπάγκος, Επίκ. Καθηγητής Πανεπιστημίου Θεσσαλίας, Βασίλειος Προμπονάς, Επίκ. Καθηγητής Πανεπιστημίου Κύπρου
Summary:
In the context of this thesis, using computational methods, we studied
extensively, the very interesting superfamily of GPCRs and their partners,
G-proteins. Two methods were developed:
GPCRpipe (http://bioinformatics.biol.uoa.gr/GPCRpipe/) characterizes proteins
as probable GPCRs, using only their sequence as input. It is based on specific
patterns described by Pfam pHMMs and on a specially designed by our work GPCR
specific HMM, which allows the prediction of their topology. GprotPRED
(http://bioinformatics.biol.uoa.gr/GprotPRED/) was designed to accurately
detect G-proteins with solely their sequence as input. Using the specific
pHMMs that were built, it detects Gα proteins, and classifies them in the four
basic mammal families (Gs, Gi/o, Gq/11, G12/13). Moreover, with two additional
pHMMs for the β and γ subunits, identification of G-proteins is complete.
The sheer volume of data on G-coupled protein receptors led to the creation of
two new databases: Human-gpDB is available at
http://bioinformatics.biol.uoa.gr/human_gpdb/ and contains information
regarding human GPCRs, G-proteins, effectors and their interactions. PLHG-DB
http://bioinformatics.bio.uoa.gr/plhg_db/) includes all peptide ligands of
human GPCRs. Specifically, information on the sequence of the peptide ligand,
its function, the precursor protein and the interaction with the respective
receptor, all accompanied by the respective citations, is presented to the user.
We analyzed the structures and amino acid sequences of G-proteins and
identified certain surfaces of Gα subunits that may, in many cases, participate
in binding both receptors and effectors. The differences displayed in the
sequence and structure of these sites may perhaps account for Gα specificity
towards their binding partners. Furthermore, the diversity in the
electrostatic potential of Gα surfaces, combined with observed electrostatic
properties of various effectors and RGS structures, suggests that electrostatic
complementarity is, most probably, an important factor in the regulation of
effectors by G-proteins, as well as Gα interactions with RGS proteins. Finally,
we studied all missense SNPs on human, class A GPCRs and their associations
with diseases. A set of 650 human class A GPCRs from UniProtKB/SwissProt and
RefSeq databases along with 21746 missense SNPs through dbSNP,
UniProtKB/SwissProt, SNPdbe and ClinVar was created. Using statistical
analysis, a tendency for the SNPs to be more abundant in some domains
(cytoplasmic loops), and less abundant in others (transmembrane segments) was
observed. Consolidation of the data disclosed a total of 441 SNPs, located in
39 receptors, which were found to have a clinical impact or/and an association
with disease. For these receptors, both secondary structure diagrams and 3D
models were created and all SNPs were mapped on them. In some of the
constructed models, pathogenic SNPs tend to accumulate in certain regions.
Keywords:
Bioinformatics, Signal transduction, Membrane receptors, G-proteins, GPCRs
