Hot topics: A cryptic binding pocket in K2P2 exposes new avenues for drug development.

The TREK subfamily of K2P channels (K2P2, K2P4 and K2P10) pass background potassium currents that modulate the excitability of neuronal cells and cardiac myocytes. In recent years, these channels have received significant attention as potential drug targets. This is in part because of their proposed roles in the regulation of nociception, analgesia, anesthesia and depression and, also because they act as polymodal signal integrators for physiological influences as diverse as temperature, membrane tension, phosphorylation, and phospholipids [1-3]. However, despite accelerating progress, including atomic resolution structures of two TREK subfamily members [4,5] , and K2P1 [6], a deeper appreciation of how K2P channel structure-function relationships, including gating mechanics, relate to physiology and disease remains hampered by a paucity of specific blockers and activators. In an elegant new study, Lolicato and colleagues from the Minor lab highlight the synergistic power of combining structural and functional approaches to reveal new insights into the operation of membrane proteins and unveil a new avenue for the development of TREK-channel pharmacology [7].
Lolicato et al., describe crystal structures of mouse K2P2 (TREK1) in complex with a novel K2P2-specific activator (ML335), and a K2P10 activator (ML402) [7]. ML335 and ML402 occupy a previously unidentified binding site—the K2P modulator pocket. Like all K2P channels, K2Ps 2, 4 and 10 are composed of two subunits, each with two pore domains and can assemble as homomers or heterodimers. Given the bilateral nature of the K2P structure, each channel will have two modulator pockets. The modulator pocket is located between the P1 and M4 helical domains in each subunit where residues conserved among TREK subfamily channels interact with the ML335 and ML401 via cation-π and π-π interactions.

While the activity of most ion channels is controlled by multiple gates, experimental evidence has accumulated to support the idea that K2P channels use a single C-type gate at the outer pore which controls ionic flux by the mechanics of the selectivity filter for potassium ions [8-10]. Numerous studies suggest that the unique architecture of K2P channels routes diverse regulatory signals to the C-type gate to control channel activity [9,11-13]. Thus, operation of the C-type gate is directly sensitive to changes in the permeant ion [11,13] and indirectly influenced by various K2P channel regulators that interactions with domains that in-turn impact the C-type gate [12-16]. The modulator pocket described lies behind the selectivity filter. Functional studies show that ML335 holds the pocket in an open conformation and thereby, activates the channel by stabilizing the C-type gate of K2P2 [7]. Because modulator pocket activators appear to be sufficient to open K2P2 channels, the findings suggest that this previously unappreciated, druggable site can be leveraged for the development of novel channel gating-modulators with potential utility as analgesics, anesthetics or neuroprotective agents.

Comments by Leigh D. Plant, Ph. D. (Research Associate Professor, School of Pharmacy, Northeastern University)

[1] Goldstein, S. A. et al. (2001). Potassium leak channels and the KCNK family of two-P-domain subunits. Nat Rev Neurosci 2, 175-184. [PMID: 11256078</a].

[2] Enyedi, P. & Czirjak, G. (2010). Molecular background of leak K+ currents: two-pore domain potassium channels. Physiol Rev. 90, 559-605. [PMID: 20393194].

[3] Honore, E. (2007). The neuronal background K2P channels: focus on TREK1. Nat Rev Neurosci. 8, 251-261. [PMID: 17375039].

[4] Brohawn, S. G., del Marmol, J. & MacKinnon, R. (2012). Crystal structure of the human K2P TRAAK, a lipid- and mechano-sensitive K+ ion channel. Science, 335, 436-441. [PMID: 22282805].

[5] Dong, Y. Y. et al. K2P channel gating mechanisms revealed by structures of TREK-2 and a complex with Prozac. Science, 347, 1256-1259. [PMID: 25766236].

[6] Miller, A. N. & Long, S. B. (2012) Crystal structure of the human two-pore domain potassium channel K2P1. Science, 335, 432-436. [PMID: 22282804].

[7] Lolicato, M. et al. (2017). K2P2.1 (TREK-1)-activator complexes reveal a cryptic selectivity filter binding site. Nature, 547, 364-368. [PMID: 28693035].

[8] Zilberberg, N., Ilan, N. & Goldstein, S. A. (2001). KCNKØ: opening and closing the 2-P-domain potassium leak channel entails “C-type” gating of the outer pore. Neuron, 32, 635-648. [PMID: 11719204].

[9] Piechotta, P. L. et al. (2011). The pore structure and gating mechanism of K2P channels. EMBO J, 30, 3607-3619. [PMC: PMC3181484].

[10] Schewe, M. et al. (2016). A Non-canonical Voltage-Sensing Mechanism Controls Gating in K2P K(+) Channels. Cell 164, 937-949. [PMID: 26919430].

[11] Cohen, A., Ben-Abu, Y., Hen, S. & Zilberberg, N. (2008). A novel mechanism for human K2P2.1 channel gating. Facilitation of C-type gating by protonation of extracellular histidine residues. J Biol Chem, 283, 19448-19455. [PMID: 18474599].

[12] Bagriantsev, S. N., Clark, K. A. & Minor, D. L., Jr. (2012). Metabolic and thermal stimuli control K(2P)2.1 (TREK-1) through modular sensory and gating domains. EMBO J, 31, 3297-3308. [PMC: PMC3411076].

[13] Bagriantsev, S. N. et al. (2011). Multiple modalities converge on a common gate to control K2P channel function. EMBO J, 30, 3594-3606. [PMID: 21765396].

[14] Chemin, J. et al. (2005). A phospholipid sensor controls mechanogating of the K+ channel TREK-1. EMBO J, 24, 44-53. [PMID: 15577940].

[15] Murbartian, J., Lei, Q., Sando, J. J. & Bayliss, D. A. (2005). Sequential phosphorylation mediates receptor- and kinase-induced inhibition of TREK-1 background potassium channels. J Biol Chem, 280, 30175-30184. [PMID: 16006563].

[16] Honore, E., Maingret, F., Lazdunski, M. & Patel, A. J. (2002). An intracellular proton sensor commands lipid- and mechano-gating of the K(+) channel TREK-1. EMBO J, 21, 2968-2976. [PMID: 12065410].

Posted in Hot Topics

Anti-infective pilot entries

GtoPdb has been traditionally focused on the pharmacology associated with human diseases (i.e. we have not been funded to cover anti-infectives).  In 2017 we have been exploring possible funding opportunities to extend into  expert curation of anti-infectives, particularly in the light of the antibiotic resistance threat and expanding drug discovery efforts for neglected tropical diseases (NTDs).  Consequently, for this release we have added a selection of entries as a proof of concept for how well our current data model would accommodate these new types of relationship mappings “off the bat”.  These are now under a new  category of “Anti-infective targets“.  A snapshot of the six new entries is show below.


We curated three antimalarials,  two antivirals and one antibiotic, some of which have the ligand in PDB entries. By and large,  this pilot was successful and we would be pleased to get feedback from interested parties.  A number of technical challenges were encountered  but most of these were “domain inherent” rather than GtoPdb data model issues per se.  Examples include the sub-species and strain multiplexing of the target sequences. This results in having to map the authors described ligand activity data to TrEMBL entries (sometimes with equivocalities as to the exact isolate sequence) rather than to the deeper annotated Swiss-Prot entries avaialble for some reference pathogen proteomes with completed gene naming.  We also ran in to the multi-enzyme “string of pearls” problem where large polypeptides encode for multiple  functional domains, more than one of which that can be (or have been) targeted by different inhibitors.  Classically, this is the case for the viral polyprotein precursor proteins but in this set also for the polyketide synthase entry shown below.


This was derived from a recent 2017 Cell paper  “Development of a Novel Lead that Targets M. tuberculosis Polyketide Synthase 13” (PMID 28669536). Should we be sucessful in being resourced to expand in this domain we can already envisage tweaks to our existing data model and curation processes that could address some these domain-specific challenges.  For example,  we can specify ligand-targeted domains not only by InterPro coordinates in UniProt entries but also by getting the TrEMBL entries “promoted” to Swiss-Prot to enhance the domain annotation cross-references.


Posted in Uncategorized

New source cross-references in release 2017.5

(minor updates 15 Sep 2017)

The statistics of content are presented as usual in the release notes and The Guide to IMMUNOPHARMACOLOGY has a separate update.  This post describes changes and updates to other resources we provide links to, that have been introduced in this release cycle.  More detail will be provided in the help pages (and feedback on any of them is welcome) but the outlines are as follows;

Extra links for ligands. The new connectivity applies to those that have chemical structures (i.e. SMILES strings for mostly small molecules but also peptides up to ~ 50 to 60 residues and a few oligonucleotide drugs), which represents 6821 ligands in GtoPdb. Links have now been rationalised by introducing InChIKey call-outs to UniChem at the EBI.  This resource, currently containing over 150 million indexed chemical structures from 37 sources (including our own), many of which we had hitherto individually curated links for.  In essence, UniChem “looks after” comprehensive cross-mappings between these sources via a regular and precise automated process. We can consequently rely on presenting these links for our own entries. This is because we have selected and curatorially checked (i.e. locked-down) our own structural assignments, including for our PubChem submissions.  By clicking the UniChem link users can now quickly navigate to complementary sources  such as  DrugBank, ChEBI, HMDB, BindingDB, ChEMBL, PDBe SureChEMBL (patents) and others. Note this is analogous to the Google InChIKey call out we already introduced for our ligands some time ago. There is some overlap in the result sets but note the Google search will find different chemistry sources (including ChemSpider entries, usualy uppermost in the Google rankings) that are not currently indexed by UniChem.

The Human Protein Atlas (HPA) team have increased their profile recently,  not only by becoming one of the European ELIXIR core resources but also because of a major new extension in the form of a Pathology Atlas with a focus on human cancer.  We have also had contacts with the team.  Consequently,  we selected this as as a new outlink from our human protein entries (2839 target links and 353 ligand links) as an excellent first-stop shop for tissue and cell line expression patterns as well as intracellular distributions.  In terms of utility it is important to note that HPA offers the best of both worlds by integrating three sources of high-throughput mRNA transcript profiling in addition to direct antibody detection of the protein.

CATH/Gene3D. As you may have been noticed we have increased our protein structure connectivity in 2017 including our SynPharm drug-responsive protein sequences resource (see below).  There are many user utilities for the increase in structural data, including the impressive acceleration of ligand binding sites resolved in new GPCR structures. CATH is a classification of PDB protein structures grouped by protein domains into superfamilies that have diverged from a common ancestor. Users are encouraged to take a look at the  features of CATH for their own exploitation. These include tracking the deep phylogeny of pharmacological targets (that have structures) where this is difficult to detect on the basis of sequence similarity alone. The current version of GtoPdb includes 1634 target links to CATH (which is lower than the total because not all protein families have 3D structural representation, yet), and 230 peptide ligands  (Sep 2017 update CATH is also now European ELIXIR core resources).

synPHARM was originally set-up to provide synthetic biologists with tools to discover sequences that could be modulated by known ligands from GtoPdb which could be transferred to synthetic proteins in order to confer drug control. synPHARM combines structural information from the Protein Data Bank with information on ligand binding from GtoPdb to produce a database of ligand binding sequences. As such, it is a useful resource for 3D ligand binding information. We have now added links from GtoPdb target and ligand pages to structures in synPHARM.

IUPHAR Pharmacology Education Project (PEP). PEP is a new IUPHAR initiative to provide free access to education and training resources in pharmacology. We have added links from 673 ligand and drug pages to background information in PEP, for further information on drug action and clinical use.

RCSB Protein Data Bank (RCSB PDB). Although not strictly new, it’s worth pointing out that the current rate of reporting new structures of ligands bound to targets means the number of links to the PDB via ligand entries has increased significantly over recent releases. The number of our PDB ligand links now stands at 1337, based on exact InChIKey matches. In addition, many of the GtoPdb ligands are represented in the PDB as alternative isomeric forms. Note also there are occasions where the PubChem MMDB CID assignment does not exactly match the PDB ligand structure.  In both these cases we add cross-pointers in the ligand comment sections.

Tagged with: , ,
Posted in Database updates, Technical

Database release 2017.5

The 5th IUPHAR/BPS Guide to PHARMACOLOGY database release of 2017 includes updates to several target families, and new targets and ligands added, focusing on those relevant to immunopharmacology. We also announce a new organisation for ligand families and groups. This update also includes the beta v2.0 release of the IUPHAR Guide to IMMUNOPHARMACOLOGY portal taking into account early feedback from our beta testers. Eagle-eyed users may have noticed a new homepage layout for GtoPdb, which has been reorganised to highlight important new features at the top of the page, with quick links to the main database pages on the left, and news items and publications below.

Target and ligand updates

Ligand families

We have introduced a new organisation of peptide ligands into families. This can be reached via a link from the “Ligands” submenu of the main navigation menu. This started with the aim of grouping related peptide sequences together into families to aid discoverability and allow us to add comments and references pertaining to the family as a whole. We have also experimented with grouping together some other types of ligands (such as the Immune checkpoint modulators) linked by their mechanism of action (although not a family in the phylogenetic sense). Feedback on this new organisation is welcome.

Ligand activity graphs

Continuing on from previous updates (releases 2017.2 and 2017.4), where we described the addition of graphs to visualise ligand activity data for targets across species using data from GtoPdb and the med-chem database, ChEMBL, we have now extended this feature to all ligands in GtoPdb with quantitative activity data at targets, even where the ligands do not have data in ChEMBL. There will  also be cases where the GtoPdb curators just haven’t yet identified the ligand in ChEMBL, in particular peptides can be difficult to search for because of naming differences and lack of standard chemical structure descriptors.

Expanded database cross-links

From time to time we internaly review the databases that we cross-link to and from, to make sure they are current and useful. During an iteration of this process within this release cycle we introduced several new resources that have value for users. These changes are explained in a separate post.


This update also sees the release of the beta v2.0 of the new Guide to IMMUNOPHARMACOLOGY portal. This is a Wellcome Trust-funded extension to the existing database, aiming to improve data exchange between immunology and pharmacology. Read the release notes and technical update here. We are grateful for all the feedback received so far and welcome continued comments and bug reports as we further develop the new data and portal.

Database content and statistics

For the full statistics on release 2017.5 please see the about page on the GtoPdb site. In summary, there are now 15,281 curated binding constants between 2825 targets and 8978 ligands. (N.B. for various reasons, not all those targets and ligands have  quantitative binding data; in the third table below the current number of human targets with quantitative data is 1431.)


Tagged with: , , ,
Posted in Database updates, Technical

Guide to IMMUNOPHARMACOLOGY – beta release v2.0, August 2017

We are please to announce the second, beta-release of the Wellcome Trust-funded IUPHAR Guide to IMMUNOPHARMACOLOGY (GtoImmuPdb). Since our first beta-release back in May 2016, we have undertaken a user-testing exercise to gather feedback on the layout, navigation and content of the resource. This blog post summarises the outcomes from the user-testing and the consequent changes in the v2.0 release.

User Testing

We reached out to several groups of immunology and inflammation researchers asking for anyone interested in helping us to test the GtoImmuPdb beta v1.0 release. These groups came from the British Society of Immunology, Glasgow University, the IUPHAR ImmuPhar Section and more. We set-up a google form to guide people through using the beta site and asked relevant questions about each section of the site, focussing on how easy the site was to use, how credible the data appeared to be and what features users would like to see added/removed. In total 8 paritipants compelted the testing, and these testers had a variety of research experience (PhD student to Professor) and previous experience of the using GtoPdb (from several visits a month to never having used it before).

Some of the key outcomes have informed changes in the v2.0 release. Most notably, reorganisation of help documentation, altering process/cell type layouts and navigation and improving information on disease associations.

Website Updates

Help Pop-ups

Help pop-ups have been added to the main panels on the portal. This gives upfront help about what users can expect to find within each section, and how to navigate it. We believe this is easier than having to read through a tutorial or help page. Our aim will be to supplement these with short help videos, showing how to navigate the data.

From our user feedback, although help was easy enough for most user to find, some comments pointed to it not always being immediately obvious what the data being displayed was and how to navigate it. We hope that by adding in ‘in-line’ help pop-ups it is easier and quicker for users to find the relevant guidance.

Help pop-ups added to portal

Process/Cell type layout 

We have made some minor modifications to the layout of both the process association and cell type association pages. The navigation menus (pull-down menu and quick links to target class sections) have been swapped over.

We have also added in a simple piece of javascript to display a ‘back to top’ button if the users scrolls down the page. This should help navigation. This feature may eventually be re-used in other parts of the site.

Revised layout of process association pages. Navigation items have been moved and new ‘return to top’ button is present on scroll-down

Improvements to disease layout – response to feedback

Our user-testing highlighted the need to display more information on the disease association pages, particularly about why ligands are associated with some diseases. The information displayed has been extended to show whether the ligands is an approved drug (and which regulator it was approved by) and links to more info at We have also added the clinical use comments for the ligand. This is all data that can be found on the ligand summary pages, but we have also surfaced into on the disease association pages to bring added value and ensure the most relevant information is available in the right places.

Additional information including if ligands are approved drugs and clinical use comments bring added value

Further reading – exposing useful curation reading list

A new page has been added that presents a further reading collection extracted from an open CiteUlike collection compiled by the curation team The papers presented are general, not ones from which database entries have been curated. They are mostly review articles that are relevant to the scope of the database. We would be pleased to receive recommendations for additions (either to the further reading list or for curation).

Listing papers of relevance to immunopharmacology that are not already used in curated entries

Future priorities

Other aspects of the process and cell type data commented on during user-testing was the need for more information on GO terms, IDs and evidence; the need to incorporate curator comments about the process associations; and the absence of data for some cell type categories. Improving the content and display to meet these is a priority, but a substantial body of curatorial work. So we will be aiming to meet these needs, but not until future releases.

Development of the beta-release is ongoing with regular updates planned over the next few months as the quantity of data captured increases and improvements in the site layout and function are made. As always we welcome comments and engagement with interested user groups and  potential future users, so don’t hesitate to get in touch with us.

This project is supported by a 3-year grant awarded to Professor Jamie Davies at the University of Edinburgh by the Wellcome Trust (WT).

Posted in Guide to Immunopharmacology, Technical

Hot topics: Agonist-bound crystal structures of the CB1 cannabinoid receptor

Antagonist bound crystal structures of GPCRs are useful in giving an insight into the molecular conformation of a receptor’s inactive state whilst enabling the design of new drugs. However, they prove insufficient to understand the activation mechanism of the receptor and mediation of its physiological effects. This necessitates the study of agonist-bound structures. In this direction, Hua et al., (2017) [1] have recently reported two agonist-bound crystal structures of Cannabinoid Receptor 1 (CB1), one with a tetrahydrocannabinol derivative, AM11542 [PDB: 5XRA], and the other with a hexahydrocannabinol, AM841 [PDB: 5XR8]. Previously, two antagonist-bound crystal structures of CB1 complexed with AM6538 and MK-0364 (taranabant) were reported by Hua et al., (2016) [PDB: 5TGZ] [2] and Shao et al., (2016) [PDB: 5U09], respectively [3].

Comparing the agonist and antagonist bound structures of the CB1 receptor reveals significant details:

1. The N-terminus in 5TGZ and 5U09 is a V-shaped loop which interacts with the bound antagonists, acting like a plug to the orthosteric binding pocket. The agonist-bound versions (5XRA and 5XR8), however, have their N-terminus residing over the binding pocket without any direct involvement in ligand binding. However, the N-terminus is truncated in all the crystal structures and hence the authors do not rule out the possibility that the full-length N-terminus might assume an entirely different conformation.

2. Both the agonists adopt an L-shaped conformation in the binding pocket, in contrast to the horizontal geometry of AM6538 in 5TGZ. The helical rearrangements hence observed in TM 1 and 2 and inward movement of residues Phe1702.57 and Phe1742.64 lead to a reduction in the binding pocket volume by 53% compared to 5TGZ. This serves as a testament to the highly flexible nature of the CB1 receptor and should be considered in future structure-based drug design studies for the receptor.

3. The alkyl chain of the two agonists extends into the ‘long channel’ of the receptor formed by the transmembrane helices (TM) 3, 5 and 6. The authors point out that this orientation is similar to that of ‘arm 2′- the nitroalkyl region of AM6538 in 5TGZ and of the alkyl chain of ML056 in the previously-described structure of the S1P1 receptor [4], thus indicating that the long channel could be a conserved binding region for alkyl chains in lipid binding receptors.

4. In 5TGZ and 5U09, the residues Phe2003.36 and Trp3566.48 exhibit aromatic stacking with each other. In this report, a synergistic conformational change of the residues was observed with the rotation of TM3 and side chain flip of Phe2003.36 towards the binding pocket occurring simultaneously with the rotation of TM6 away from TM3 breaking the interaction between the residues. The authors speculate the role of this ‘twin toggle switch’ in the activation of the receptor as a previous study has already shown [5].

5. Using the crystal structure, a cholesterol molecule has been identified to bind between the cytoplasmic portions of TM 2, 3, and 4 in the agonist-bound models. This was not observed in the antagonist models. However, the possible existence of a lipid access channel proposed in the taranabant bound (5U09) structure has not been discussed in this paper. This raises questions about the influence of lipids on the receptor binding through allosteric sites.

Comments by Lahari Murali (@wavesml), Steve Alexander (@mqzspa), Steven Doughty and Abi Emtage (@AbiEmtage)

[1] Hua, T. et al. (2017). Crystal structures of agonist-bound human cannabinoid receptor CB1. Nature.doi:10.1038/nature23272. [PMID: 28678776]

[2] Hua, T. et al. (2016). Crystal Structure of the Human Cannabinoid Receptor CB1. Cell 167: 750–762.e14. doi: 10.1016/j.cell.2016.10.004. [PMID: 27768894]

[3] Shao, Z. et al. (2016). High-resolution crystal structure of the human CB1 cannabinoid receptor. Nature 540:602–606. doi: 10.1038/nature20613. [PMID: 27851727]

[4] Hanson, M. A. et al. (2012). Crystal structure of a lipid G protein-coupled receptor. Science 335:851–855. doi: 10.1126/science.1215904. [PMID: 22344443]

[5] Singh, R., Hurst, D.P., Barnett-Norris, J., Lynch, D.L., Reggio, P.H., and Guarnieri, F. (2002). Activation of the cannabinoid CB1 receptor may involve a W6 48/F3 36 rotamer toggle switch. J. Pept. Res. 60: 357–370. [PMID: 12464114]

Tagged with:
Posted in Hot Topics

Hot topics: Identifiers for the 21st century

While identifiers are not a traditional “hot topic” in pharmacology the subject is becoming increasingly important. One of the reasons is that for mechanistic pharmacology the community needs to define (and communicate) identifiers for the key entities of model organism species and strains, proteins, protein complexes, genes, sequences, sequence variants, as well as the explicit molecular structures of chemicals, peptides and therapeutic biologicals (including antibodies) used for experimentation. Indeed one of the roles of IUPHAR (as NC-IUPHAR) is to review and recommend protein target nomenclature, in collaboration with the Human Gene Nomenclature Committee (HGNC) [1]. The paper featured here is a technical review [2] of identifier qualities and best practices that facilitate large-scale data integration. It also goes into problems related to persistence and web-accessibility/resolvability. As a database provider, the relevance of this article for GtoPdb is clear (since we are largely about identifiers and their relationships). We are carefully considering its implications and possible consequent changes in our practice. The GtoPdb team has already engaged with this theme some time ago in a blog post [3] that provided an introduction to resolving bioactive ligands and their protein targets from the literature to standardised molecular identifiers.

Comments by Chris Southan (@cdsouthan).

[1] International Union of Basic and Clinical Pharmacology Committee on Receptor Nomenclature and Drug Classification

[2] McMurray et al. (2017). Identifiers for the 21st century: How to design, provision, and reuse persistent identifiers to maximize utility and impact of life science data. PLoS Biol. 29;15(6). [PMID:28662064].

[3] A Pharmacologists’ Guide to Resolving Chemical Structures and their Protein Targets from the Literature

Tagged with:
Posted in Hot Topics