| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Research Resource |
Institut de Recherche Interdisciplinaire en Biologie Humaine et Moléculaire (J.V.D., S.C., G.Va.), Université Libre de Bruxelles, B-1070 Brussels, Belgium; Commissariat à lEnergie Atomique, Biologie, Informatique et Mathématiques (F.H.), 38054 Grenoble, France; and Centre for Molecular and Biomolecular Informatics (G.Vr.), Radboud University Nijmegen, 6525 ED Nijmegen, The Netherlands
Address all correspondence and requests for reprints to: Dr. Gilbert Vassart, Institut de Recherche Interdisciplinaire en Biologie Humaine et Moléculaire, Université Libre de Bruxelles, Campus Erasme, Route de Lennik 808, B-1070 Brussels, Belgium. E-mail: gvassart{at}ulb.ac.be.
ABSTRACT
The glycoprotein-hormone receptor information system (GRIS) presents a comprehensive view on all available molecular data for the lutropin/choriogonadotropin receptor, follitropin receptor, and thyrotropin receptor G protein-coupled receptors. It features a mutation database presently containing 696 point mutations, combined with all sequences and the associated homology models. The mutation information was automatically extracted from the literature and manually augmented with respect to constitutivity, surface expression, sensitivity to hormones, and binding affinity. All information in this integrated system is presented in a G protein-coupled receptor specialist-friendly way. A series of interactive tools such as rotamer analysis, mutation prediction, or cavity visualization aids with the design and interpretation of experiments. A universal residue numbering system has been introduced to ease database searches as well as the use of the information in conjunction with literature data from diverse origins. Users can upload new mutations. GRIS is freely accessible at http://gris.ulb.ac.be/.
G PROTEIN-COUPLED RECEPTORS (GPCRs) are a large superfamily of membrane receptors that are involved in a wide range of signaling activities. They share a tertiary structure characterized by an extracellular amino terminus, followed by seven transmembrane
-helices connected by three extracellular and three intracellular loops, with the C terminus in the cytoplasm. This so-called "serpentine domain" forms the basis for the classification of GPCRs into classes displaying sequence identities (1). Class A, or rhodopsin-like GPCRs, form the largest class. Other classes are class B (secretin, glucagon, pituitary adenylate cyclase activating peptide receptors, etc.), class C (calcium sensing and metabotropic glutamate receptors), and a few smaller classes. GPCRs are activated by extracellular ligands. Most class A receptors have a relatively short N-terminal domain and can be activated by small ligands that bind in a cavity between the seven transmembrane helices (Fig. 1
) or by peptidic ligands of which a part will be directed in the same cavity (Fig. 1
). Class B and C receptors have an extensive extracellular N-terminal domain that is involved in ligand binding (2, 3). For the class B receptors, it has been suggested that the largest part of the peptide agonist forms an
-helix that binds to the N-terminal domain of the receptor while the N-terminal residues of the ligand dock between the transmembrane helices (TMs), thereby mimicking a class A receptors small ligand (2). For the class C receptors, which are constitutive dimers, it is known that the small ligand (e.g. calcium or glutamate) binds to the N-terminal domain of the receptor. The prevailing hypothesis is that activation takes place via conformational changes of the dimeric ectodomain, resulting in an ill-defined effect on the structure of the serpentine dimer (3). Glycoprotein-hormone receptors (GpHRs) belong to class A from an evolutionary point of view (1, 4, 5, 6). However, they comprise a large, extracellular, N-terminal domain that, just as in class B and C receptors, is involved in ligand binding (Fig. 1
). Upon activation, it is believed that the ectodomain switches from an inverse agonist of the serpentine domain into a full agonist, and initiates signal propagation (7). For the GpHRs, nothing is firmly known yet about the transduction of this signal (8).
|
Many natural gain-of-function mutations of GpHRs have been identified in patients suffering from target tissue autonomy resulting, for the TSHR, in hereditary (16) or congenital (17) toxic thyroid hyperplasia (18) and toxic thyroid adenoma (19) or, for the LHR/CGR, in pseudo-precocious puberty of the male (20). These mutations are mainly observed in the transmembrane portion of the LHR/CGR and the TSHR (19, 21). Analyses of the mutant receptors, using molecular models of GpHRs built using the bovine rhodopsin structure (22, 23) as a template, are yielding information on the mechanism of activation (5, 24, 25). The FSHR seems more resistant to constitutive activation. Only four natural mutants were reported, characterized by persistent male fertility despite pituitary insufficiency (26), and by ovarian hyperstimulation syndrome (25, 27, 28).
Understanding of the structure-function relationships of the GpHR subfamily requires that mutation data gathered from one receptor can be interpreted in the light of results obtained with another. Zhang et al. (29) introduced a salt bridge between the helices III and VI in the human LHR/CGR. Mapping a large series of mutations on a homology model, conclusions could be drawn about the importance of interactions between these two helices for LHR/CGR activation. Angelova et al. (30) used sequence alignments to combine mutation data using a schematic model of the generic helix pair VIVII to study interactions between these two helices from mutation data in multiple receptors. Tao et al. (31) use multiple sequence alignments and a homology model to merge mutation data and to draw conclusions about this same helix pair in the FSHR. Fanelli et al. (32) combined a series of homology models of the LHR/CGR mutated at residue M398 (2.43) and site-directed mutagenesis experiments to study the mechanism of constitutivity induced by mutations at this site.
These examples that shed light on the sequence-structure-function relationships of GpHRs all have in common that extensive integration of heterogeneous data was the key to success. To support our own studies in this field (7, 14, 25, 33), we have designed a GpHR information system. Called glycoprotein-hormone receptor information system (GRIS), it holds all 51 presently known GpHR sequences (classified and aligned) and 696 mutations that were automatically extracted from the literature (34) and manually annotated. Homology models for all TM domains and ectodomains have been built and are fully integrated with the sequence and mutation information. A series of fully interactive user-friendly facilities allow GpHR researchers to intuitively use powerful software that uses the system for the design and interpretation of experiments.
GRIS is freely available at http://gris.ulb.ac.be/. The software and scripts to build GRIS are sufficiently general to be used for other GPCR families, and are available from the authors upon request.
RESULTS AND DISCUSSION
The four basic functions of an information system are browsing, retrieval, query, and inferencing (35).
Browsing
Browsing is important to show the user scientists the context of data, such as similar data for related receptors, different experiments for the same receptor, etc. GRIS uses the hyperlink tables of the GPCR database (GPCRDB) (36) for this purpose, and the information about mutations along with links to SwissProt (37), PubMed (http://www.pubmed.com), etc. adds further browsing facilities. GRIS is also rich in internal hyperlinks to facilitate rapid navigation between mutants, sequences, structure models, and computing facilities. Figure 2
shows a so-called snake-plot for a receptor (38).
|
|
|
Query
The mutant data can be inspected in many ways. The snake-like plots (see Fig. 2
) present mutants in the format most commonly used in the GPCR field. The multiple sequence alignment that is hyperlinked to the mutation information gives the user quick access to mutations of the same residue in different receptors (see Fig. 4
). The most direct query facility is provided by the query form as shown in Fig. 5
, which allows the user to search for mutations as a function of species, receptor type, structure element, residue position, and mutation type. The mutation data was obtained using MuteXt (34) (see Materials and Methods for a brief description of the MuteXt methodology), and missing information about the expression level, the degree of constitutive activity obtained with the mutation, and the ligand binding activity, were manually extracted from the literature and added to the system. Automated MuteXt updates are scheduled three times per year. Because it is not possible to automatically extract all mutant information, we added a facility that allows users to interactively add mutation information to GRIS. Upon reception of a GRIS account, users can rapidly upload missing GpHR mutations that have been published in the literature by completing a simple mutation contribution form (see supplemental data, published on The Endocrine Societys Journals Online web site at http://mend.endojournals.org). After revision by the curator, the new mutations are added to the public database. Because MuteXt is an automated literature retrieval method, it has its limitations. MuteXt needs to access online full-text articles. The availability of these articles is, in turn, dependent on the subscriptions of the institution MuteXt is run from. GRIS users are encouraged to report any missing references.
|
|
|
The rotamer analysis module of GRIS (Fig. 7A
) has already shown to be an excellent mutation prediction tool in a recent experiment (43). A constitutively active M626I (6.37) mutation was found in the TSHR in affected members of a family with nonautoimmune hyperthyroidism. The rotamer analysis using GRIS indicated a severe bump of the mutant residue with I515 (3.46) in the TSHR. We tried rotamer distributions for a few residue types at position I515 in the TSHR-M626I variant. A methionine at position 515 looked most promising and the double mutant I515M/M626I was constructed in the TSHR wild type. Experiments showed that the constitutive activity obtained from the M626I mutation was reduced again to wild-type activity by I515M.
Although GpHRs are activated by their peptide hormone ligands that bind to the extracellular domain, a small organic molecule has previously been designed that acts as a FSHR antagonist (44). Because impairment of FSHR function by mutations can lead to decreased fertility or infertility (45, 46, 47), such drugs are becoming promising nonsteroidal contraceptive agents. Rational drug design requires knowledge about cavities in the target receptor, and the identification of residues around such a cavity is obviously important (Fig. 7B
).
A previous study by our group stressed the importance of a hydrogen bond network between TM6 and TM7 of the human TSHR to maintain the inactive state of the receptor (24). We reanalyzed all single mutations with the mutation prediction module of GRIS and came to the same conclusions. Mutations D633N (6.44) and N674A (7.49) still maintain hydrogen bonds between TM6 and TM7 as described in the previous publication (24). GRIS predictions of the substitutions D633A and N674D show that a TM6TM7 link in this region cannot be maintained, either by total disruption of all hydrogen bonds (D633A) or by electrostatic repulsion (N674D), which is associated with a strong increase in constitutive activity (48).
Another study by our laboratory elucidated a link between constitutive activity and promiscuous activation of the FSHR (25) following the discovery of a mutation D567N (6.30) in a patient with spontaneous ovarian hyperstimulation syndrome. Using molecular modeling, we concluded that this mutation was able to break an ionic interaction between D567 (TM6) and R467 (3.50) (TM3). Inspecting the molecular model of human FSHR from GRIS, it is easy to see that these residues can indeed interact in the wild-type receptor. Aided by the in silico mutation tool and mutation information of the same residue position in other receptors from GRIS, it is straightforward to design a panel of mutations in the human FSHR that either maintain this interaction or break it, so one can draw firm conclusions from the modeling and experimental data (25).
Each of these examples shows that GRIS is able to predict the effect of mutations and to aid researchers in the GpHR field to rationally set up their experiments.
Although the homology models of the GpHR ectodomain based on the human FSHR can be treated as highly reliable models, there are some limitations on the homology models of the transmembrane domain of GpHRs provided by GRIS. Although the 3D coordinates of bovine rhodopsin provide the best template to model the transmembrane domain of GpHRs, it is not the perfect template for several reasons (49). First, sequence analyses show that the opsin family differs very much from the other class A GPCRs. This is particularly true for interhelical loops, where identity is so low that it precludes a reliable alignment. Second, the crystal structure of bovine rhodopsin is an antiparallel and thus unnatural dimer. Third, the available bovine rhodopsin structure is the inactive form of the protein. Predictions on constitutively activating mutations should be made with caution.
When additional class A GPCR structures become available, GRIS will be updated by including novel GpHR homology models along with the current models based on bovine rhodopsin. This should allow for progressive upgrading of present models.
MATERIALS AND METHODS
GRIS has been written in Python (http://www.python.org) and uses CGI scripts for interactive activation of options. Snake-like plots are produced with the residue-based diagram generator (RbDg) (38). 3D structures are presented with Jmol (http://www.jmol.org). WHAT IF (41) is used for sequence alignments, homology modeling, interactive protein structure analysis, and mutant predictions. The sequence alignment profile needed by WHAT IF was hand-optimized for GpHR purposes. The conversion of the sequence alignment to colored HTML pages was carried out using the MView program (50).
MuteXt (34) is used to identify and extract single point mutations from literature full-text articles by pattern matching along with names of the concerned GPCR names and organism types. Extracted mutation data are checked by two automated validation filters. The first filter verifies that the mutated amino acids are at the indicated positions in the corresponding GPCR sequence(s). The second filter looks for minimal word distances between mutations, protein names, and organisms to discriminate between possible multiple sequences. Validated mutations are integrated into the GPCRDB (36). GpHR-related mutations are then imported into GRIS. At this stage of reviewing, falsely selected papers (not GpHR related) are removed from the database. The retrieved literature is read by the GRIS database curator and mutation annotations regarding constitutivity, cell surface expression, and ligand binding are added manually.
YASARA (40) is used to produce publication-ready 3D structure figures and to perform energy minimization of the GpHR models. GRIS uses the PostGres (http://www.postgresql.org) relational database system for data storage. GRIS hyperlinks from several locations to the GPCRDB for easy access to other forms of data (genomic information, phylogenetic trees, discussion groups, etc.).
All 3D models of GpHRs were constructed using the FSH-FSHR complex structure (9) as template for the ectodomain and the bovine rhodopsin structure (23) as template for the transmembrane domain. For completeness, the hormones were also modeled together with the ectodomains when the hormone sequence was available. Homology models were built using WHAT IFs standard modeling procedure. Only the structural conserved regions, which could be unambiguously aligned, were modeled. For the ectodomain, these include all nine LRRs and the
- and ß-subunits of the hormone. The transmembrane domain includes all seven transmembrane helices and the eighth cytoplasmic helix. Interhelical loops were not modeled. To remove bumps and correct the covalent geometry, all structures were energy-minimized with YASARA (40) by applying the Yamber2 force field, using a 7.86-Å force cutoff. After removal of conformational stress by a short steepest descent minimization, the procedure continued by simulated annealing (time step, 2 fsec; atom velocities scaled down by 0.9 every 10th step) until convergence was reached, i.e. no energy improvement was found for 200 steps. The set of energy minimized models and the set of "raw" models (not energy minimized with YASARA) are both available for download from the website. WHAT IF structure integrity check reports are available from the web site as well.
The function of residues in GPCRs is mainly determined by their location (51). Therefore, a mutation of a residue at a certain position in one receptor is likely to have a similar effect as the mutation of a different residue at the equivalent position in another receptor. The structural equivalence can therefore be used to transfer information about mutations in one GpHR to all other GpHRs. To aid this transfer of information, a common residue numbering scheme for all GpHRs has been implemented. For the transmembrane domain, we have implemented the numbering schemes of SwissProt (37), the GPCRDB (36), and Ballesteros and Weinstein (39). Neither the GPCRDB, nor Ballesteros and Weinstein numbered the ectodomains, so we devised a scheme that looks like both these. We used the residue naming scheme for the LRRs by Smits et al. (14) and called the residues in the inner-face ß-strands X1-X2-L1-X3-L2-X4-X5, where the side chains of the L1 and L2 residues (mostly leucines, but always hydrophobic) are pointing to the inside of the protein. The X3 residue is mostly hydrophilic and points to the outside. We have chosen to denote the X3 residue of the first LRR as 1050, X3 of the second LRR as 2050, and so on. The numbering within each LRR counts down from the XX50 residue to the previous LRR and up to the next.
The specific residues discussed in this manuscript are numbered according to the SwissProt numbering scheme (37) followed by the Ballesteros and Weinstein general numbering (39), e.g. D633N (6.44).
ACKNOWLEDGMENTS
We are indebted to Dr. Fabien Campagne (Institute for Computational Biomedicine, Weill Medical College of Cornell University, New York, NY) for his time and assistance with the residue-based diagram generator (RbDg), the program to generate the snake-like plots. J.V.D. was supported by a fellowship of Communauté Française de Belgique, Actions de Recherche Concertées.
In memory of Florence Horn, who passed away on July 13th, 2006. You will be forever in our hearts, Flo.
FOOTNOTES
This work was supported by the Belgian Program of Interuniversity Poles of Attraction (IUAP/PAI P5/30), initiated by the Belgian State Prime Ministers Office, Science Policy Programming. This work was also supported by grants from the Fonds de la Recherche Scientifique Médicale, Fonds National de la Recherche Scientifique Science Policy Programming, the LifeSciHealth program of the European Community (Grants LSHB-CT-2003-503337 and LSHB-CT-2004-518167), and Fondation Erasme.
J.V.D., F.H., S.C., G.Vr., and G.Va. have nothing to declare.
First Published Online March 16, 2006
Abbreviations: 3D, Three-dimensional; FSHR, follitropin receptor; GPCR, G protein-coupled receptor; GPCRDB, G protein-coupled receptor database; GpHR, glycoprotein-hormone receptor; GRIS, glycoprotein-hormone receptor information system; LHR/CGR, lutropin/choriogonadotropin receptor; LRR, leucine-rich repeat; TM, transmembrane helix; TSHR, thyrotropin receptor.
Received for publication January 11, 2006. Accepted for publication March 10, 2006.
REFERENCES
genes as a cause of toxic thyroid adenomas. J Clin Endocrinol Metab 82:26952701This article has been cited by other articles:
![]() |
Y. Tenenbaum-Rakover, H. Grasberger, S. Mamanasiri, U. Ringkananont, L. Montanelli, M. S. Barkoff, A. M.-H. Dahood, and S. Refetoff Loss-of-Function Mutations in the Thyrotropin Receptor Gene as a Major Determinant of Hyperthyrotropinemia in a Consanguineous Community J. Clin. Endocrinol. Metab., May 1, 2009; 94(5): 1706 - 1712. [Abstract] [Full Text] [PDF] |
||||
![]() |
G. Kleinau and G. Krause Thyrotropin and Homologous Glycoprotein Hormone Receptors: Structural and Functional Aspects of Extracellular Signaling Mechanisms Endocr. Rev., April 1, 2009; 30(2): 133 - 151. [Abstract] [Full Text] [PDF] |
||||
![]() |
C.-R. Chen, S. M. McLachlan, and B. Rapoport Suppression of Thyrotropin Receptor Constitutive Activity by a Monoclonal Antibody with Inverse Agonist Activity Endocrinology, May 1, 2007; 148(5): 2375 - 2382. [Abstract] [Full Text] [PDF] |
||||
![]() |
G. Kleinau, M. Brehm, U. Wiedemann, D. Labudde, U. Leser, and G. Krause Implications for Molecular Mechanisms of Glycoprotein Hormone Receptors Using a New Sequence-Structure-Function Analysis Resource Mol. Endocrinol., February 1, 2007; 21(2): 574 - 580. [Abstract] [Full Text] [PDF] |
||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |