Flashcards in Cys-Loop receptors Deck (28):
Determined by 3D electron microscopy
5 subunits, each with 4 TM helices M1-4
M2 forms the pore, M1 and 3 surround, M4 only loosely associated at the outside
Protrudes 60 amstrongs intracellularly, 40 extracellularly (so shorter than glutamate receptors)
The only conserved residues /within/ agonist binding loops are a tryptophan in Loop A and a Trp-X-Proline in loop D, thought to contribute to structural integrity in the 5-HT3 receptor. However, there is almost always an 'aromatic box' in the binding site, which surrounds the charged amine in ligands, though exact positions of its contributors vary. The aromatic residue on loop B that's part of this box (usually TrpB) is the one that forms the cation-pi interaction
Location and role
Located at the synapse only, in the centre of a bullseye with GPCRs around the outside (e.g. glycine receptor clustering is mediated by gephyrin)
GABA and GlyRs mediate fast inhibitory transmission
nAChR and 5-HT3 mediate fast excitatory transmission
In invertebrates only, GluCl (glutamate gated chloride channel) mediates fast inhibitory transmission - ivermectin is an allosteric agonist, activates GluCl at nM concentrations, so is used as an antiparasitic. But careful - activates other pLGICs and P2X-like receptors at higher concentrations.
GLIC/ELIC thought to be important for photosynthesis. Both cation-selective, both transiently gate open in response to agonist binding - agonist for GLIC is protons, for ELIC is primary amines like cysteamine and GABA
Difficulties determining crystal structure
Much the same as for GPCRs - proteins are unstable in isolation, so needed stabilising with antibodies etc
X-ray crystallography damaged samples, which often weren't big to begin with
Before we found prokaryotic channels, it was hard to work with eukaryotic cell lines, and to express enough protein, and get pure samples
ICD in particular is hard to crystallise, so normally got chopped off
History of structural studies
Muscle nAChR was the first LGIC to be cloned (including glutamate receptors and P2X-like receptors)
AChBP in 2001
nAChR closed in 2005
ELIC closed 2008
GLIC open 2009
GluCl +IVM/FAb open 2011
2014 gave us 4 structures
2015 4 more
now the focus is on finding these receptors in other conformations
What did we know about nAChR before AChBP crystal structure was found?
Water soluble LBDs of nAChR had been produced, but not in sufficient quantities for high resolution crystal structure
Biochemical and electrophysiological data for nAChR (it was the first protein to yield single channel patch clamp recordings)
Gross structure (Brisson and Unwin 1985 determined its quaternary structure as pentameric, with an hourglass shape, by homogenising Torpedo electric organ and leaving receptors out to form tubes. Pentameric computer models fit best)
High resolution EM had shown ACh binding pockets surrounded by beta sheets, which fit with circular dichroism measurements
No atomic structural info.
Identifying binding site
Kao et al 1984 - used affinity reagent labelling (with competitive antagonists) to identify cysteine in alpha subunit necessary for ACh binding
Affinity reagents bind beta and gamma subunits to a lesser extent. This asymmetry suggests the alpha subunit has a 'principal component' and the gamma/delta a 'complementary component'.
Binding pocket for ACh was only found in pairwise expression of the alpha subunit with either beta or gamma - any other pair or single expression did not give the pocket
Further affinity labelling, proteolysis and Edman degradation experiments determined three loops on the alpha subunit (A, B and C) and three on the complementary subunit (D, E and F) that were involved in the binding
Sequence comparison showed that the labelled residues in loops A-D were highly conserved, F mostly conserved, and E highly variable.
We don't know much about it, because it's hard to crystallise
Consists of M3-M4 loop
Absent in GLIC/ELIC
Deletion studies show it's not required for expression, and cutting it off doesn't prevent basic function of receptor, but is important in localisation to synapse
Contains 'portals' at the side to allow ions to stream out parallel to membrane, perhaps enhancing effect on membrane potential
Amino acids in the circumference of these portals affect conductance - in 5-HT3A receptors, there are arginines in 4 specific places in the M3-M4 loop. In 5-HT3B receptors, there are neutral or negatively charged AAs here. This explains 5-HT3AR 200-fold lower conductance than nAChR. Expression of the AAs from the B subunit rescues conductance
GLIC doesn't have an ICD, and isn't inhibited by Ric-3 (which is a chaperone protein that interferes with assembly and function of 5-HT3R). If you give GLIC the 5-HT3R ICD, it is inhibited by Ric-3!
Insights from chimaera
Eisele et al 1991 - nAChR is calcium permeable, and permeability response to ACh binding is potentiated by calcium. 5-HT3 is calcium impermeable, and blocked by calcium. To determine which parts of the structure conferred these characteristics, they made a chimaera combining n-terminal of alpha7 and TM domain + c-terminus of 5-HT3. They found that properties like permeability and ligand binding were determined by the receptor that the TM domain or ECD came from, respectively. However, kinetics of current onset and desensitisation, fast in alpha7 WT and slow in 5-HT3 WT, were intermediate in the chimaera, suggesting interactions between ECD and TM are responsible.
GLIC doesn't have an ICD, and isn't inhibited by Ric-3 (which is a chaperone protein that interferes with assembly and function of 5-HT3R). If you give GLIC the 5-HT3R ICD, it is inhibited by Ric-3! This suggests ICD is important for transport to membrane.
Insights from AChBP
Produced by glial cells in molluscs
Pharmacological profile most similar to alpha7
AChBP was useful because it's structurally and functionally homologous to the ligand binding domain of nAChR, but it's water-soluble, so easier to use crystallography. X-ray crystal structure
Brecj et al 2001:
Ten beta sheets in each protomer, forming a beta sandwich reminiscent of a Ig fold structure, with a hairpin, but with an extra strand. Compared with the classic Ig-fold, the B-sheets are rotated around one another, forming a hydrophobic core.
There was a different cavity at each interface between the subunits. It was buried from solvent and position close to the edge of the pentameric ring. They matched the expected position of ligand binding sites from labelling and EM studies of the nAChR.
The 6 loop system was confirmed, and relative location of conserved residues revealed.
They also found HEPES buffer molecule bound in a ligand binding pocket, confirming the suspected cation-pi interaction
Brams et al 2011 - compared crystal structures of 30 AChBPs in complex with different ligands. Identified Loop C contraction.
Remember it may differ in detail though!
What is a cation-pi interaction? How was it discovered in pLGICs?
A stabilising interaction between the face of an aromatic residue and a cation
Broadly used to stabilise secondary structure and receptor-ligand binding.
Comparable strength to a hydrogen bond or ion pair
There are six times as many Trp residues in the binding pocket as you'd expect from other known proteins. This suggested non-ionic bonds like the cation-pi interaction.
Zhong et al 1998 - showed a correlation between quantum mechanical predictions from a cation-pi model, and EC50 values from in vivo nonsense suppression method of unnatural amino acid incorporation, adding Fluorines to a tryptophan residue in loop B. The more F added, the weaker the cation-pi interaction (because it destabilises the electrostatic ring and the higher the EC50). Also, adding a tethered a quaternary ammonium group to this tryptophan produces a constitutively active receptor.
Xiu et al 2009 - cation-pi interaction between nicotine and brain nAChR can form because there's a lysine at 153 that hydrogen bonds to loop B and lets nicotine in. In the muscle nAChR it's a glycine, nicotine can't get in, no muscle spasms when smoking
H bonds via water molecules
In 5-HT3R: When a conserved serine is swapped to a threonine (which still has an -OH group), there's no effect on affinity. When it's swapped to an alanine (no -OH group), affinity drops. This H bond is via water, and with backbone groups rather than R groups or ligand itself. One of the sites was on the opposite side of a beta sheet to ligand entry, so must be an intra-molecular thing rather than direct ligand affinity thing.
In AChBP there's another conserved water molecule that forms bonds between partial agonists and hydrophobic residues on the complementary side of the binding pocket. Carbamylcholine (a full agonist) binds directly, so it was proposed that this indirect binding is indicative of partial agonists.
BUT Tavares et al 2012 found no intervening water molecule in varenicicline-bound structures. Possibly observing an H bond doesn't mean it's important, or ligands bind in different poses in AChBP vs nAChR.
Other interactions, loop D and E
Conserved hydrophobic interaction between loo D and varenicicline in AChBP
Loss of aromatic residues in loop D and of leucine in loop E of alpha4beta2 nAChR eliminates difference between efficacy of varenicicline and full agonists, and eliminates varenicicline-mediated desensitisation
Changing a tyrosine residue at 153 in 5-HT3R to phenylalanine had no affect, but changing it to serine reduced affinity, suggesting aromatic interaction
Pore structure - finding the pore, creating the numbering system
Imoto et al 1986 - chimaerae of calf and torpedo delta subunits - all you had to change to swap conductances was M2
Models of pore as five kinked M2 helices were generated by molecular dynamics simulations incorporating restraints from the cryo-EM structures available at the time, from Unwin's work.
Blocking mechanism of agents such as chlorpromazine was investigated to reveal a conserved lysine or arginine at the narrowest part of the pore, labelled 0. There's also a conserved leucine at 9, right in the middle of the pore, which forms a hydrophobic barrier that extends to 13'valine and stops ions going through the closed pore (would require dehydration).
Every third or fourth AA faces the pore, and can be compared between receptors.
[the TM helices are the target for anaesthetics, steroids, alcohols]
Pore structure - finding selectivity filters, and more detailed structure
Corringer et al used mutagenesis to find three rings of selectivity -negatively charged residues at the extracellular side, Glu at -1prime, and Glu/Asp at -4 prime. The intermediate ring is the most important, because whilst mutations in the other rings can affect conductance gradually, a mutation in the intermediate causes a v sudden drop in g.
Akabas et al 1994 mutated every residue in and flanking M2 of the alpha subunit to cysteine, one at a time. The ones that reacted with MTSEA (applied extracellularly) were accessible. They found 5, so at least this many are facing the pore - this was consistent with a helic structure. ACh subtly changed accessibilities, suggesting a change in shape of the pore.
--Caveat: in the 5-HT3 receptor, MTS caused an agonist-independent current when serine of threonine was mutated to cysteine, so it may be altering conformation and thus giving false positives-negatives about accessibility. --
A series of mutations in M2 were analysed used the rate-equilibrium free energy relationship to determine which position of each residue was temporally closer to the open or closed position. Determined that the extracellular half of the pore undergoes conformational change before the intracellular half, and indeed may be in different conformations whilst the pore is functionally closed
Selectivity transition mutant
Swap -1 glutamate (intermediate ring) for alanine, add proline at -1'', change 13' valine to threonine, then you swap to anion-selective
You know it's anion selective now because swapping all the external Cl- for a nonconducting anion you get a big change in reversal potential
This same change works for 5-HT3R and for nAChR
Loop C closure
Brams et al 2011 - compared 30 different structures of AChBP bound to different stuff. Found strong agonist caused strong contraction of loop C over binding pocket, partial agonist (or non-peptide antagonist) intermediate contraction. Some suggested this is evidence for an induced fit model. BUT some partial agonists e.g. lobeline caused full contraction, and some ligands have been crystallised in several different conformations.
Lape et al suggested loop C closure is part of a 'flipping' model, where the flip occurs after ligand binding and before the closed-open change occurs. Results are conflicting, since flipping itself seems to be similar between partial and full agonists, but there's a difference in equilibrium between flipped and unflipped with agonists vs partial agonists
Differences between cation and anion selective
Extra proline in anion selective at -1'' means it has a narrower minimum channel diameter in the open state
In anion-selective, all pore-facing residues are polar. In cation-selective, intracellular half are polar, extracellular are non-polar (hydrophobic)
Original structure identified in non-conducting state due to bulky phenylalanine, but then was found that cysteamine opened it, and then GABA. This was found electrophysiologically - we still don't have an open structure. It's been cocrystallised with various ligands but never open!
Can bind benzodiazepines at two sites, one behind the main binding site and one overlapping with it.
ACh is a competitive antagonist
Discovering the open/close mechanism
Unwin and Fujiyoshi 2012 - sprayed receptors in micells with ACh and ferritin, then plunge-freezed them and then EM. An inner leaflet twist caused a 1 angstrom outward displacement of the beta subunits, where they contact M2 via a valine. The valine moves by 6 amstrongs. The M2 then also tilts outwards to accommodate this, and straightens ('relaxes'), removing the kink and making the hydrophobic girdle more polar (by revealing polar residues previously hidden in a cleft).
As a result, the minimum diameter of the pore increases by 1angstrom
The cation-pi interaction halfway up the ECD with Trp in loop B probably causes the inner leaflet twist in the alpha4beta2 nAChR
Note that there is no TrpB-cation-pi interaction in muscle nAChR, and in the alpha7 nAChR only TyrA forms a cation-pi interaction,
GLIC - activation
No ICD or Cys loop, only been crystallised in open state
Activated around pH 5/6
So to look for activation gate, we look at pKa
Histidine is the best candidate - there are 3, one in ECD (mutating had no effect), one at intracellular side of TMD (no effect), one halfway up TMD - mutating this Histidine results in a non-functional channel
This 11' Histidine faces away from the pore, and forms a hydrogen bond with the backbone of M3
Rienzo et al 2014 - Mutating the amino acids it would have bound to phenyl lactic acids means it's an ester bond not a peptide bond, which weakens the H bond histidine would form. This resulted in a shift in pH50 of about 1 pH unit, i.e. weakening the M2-M3 hydrogen bond destabilises the open form of the channel
They mutated the His to a tri-fluoro His (which can't be protonated, and had no functionality) and to a methyl His. Methyl His also had no response, which was unexpected - turned out it sterically clashed with two Isoleucine residues. When they mutated those to smaller glycine residues, there was no steric clash and the channel worked just like WT. This shows it's protonation of His that's important.
Interestingly, a GLIC/GlyR and GLIC/ELIC chimaera still shows proton sensitivity, despite lacking the GLIC M2 His... THe authors said this means the proton activation site of GLIC must be in its ECD, but maybe there's a 'proton wire' that transports protons from ECD to M2? Maybe Gly has a different mechanism for proton sensitivity, normally masked? Maybe there's another site on GLIC (ECD) that's only accessible when you remove M2?
Both chimerae have the same pKa, 1pH higher than WT GLIC, so same sensor.
Alcohols and anaesthetics acted on the GLIC/GlyR in the same way as GlyR, suggesting they bind the TMD
Activation vs opening
Bertozzi et al 2016 - found several residues in GLIC and ELIC which, when mutated, prevent channel opening but not ligand binding. These mutations were at the interface between the ECD and the TMD, and did not substantially alter protein conformation or ligand binding. Important residues were found in the B1-B2 loop, pre-M1 region, and M2-M3 loop, which all contact the B6-B7 loop. However, contacts between the B1-B2 and M2-M3 loops (previously thought to be critical for transducing the ECD movements to the M2 tilting movement) were expendable. Since the important residues were the same between GLIC and ELIC, despite different agonists, they may be generalisable to the whole family.
Specificity - what residues differ between AChR and 5-HT3 binding sites?
Agonist specificity is partly conferred by exact placement of the residues that make the key interactions such as cation-pi interactions (i.e. TrpB is not always in the same place on loop B, and the only absolutely conserved residues are a TrpD and Trp-X-Proline in loop A).
E.g. the binding of serotonin to 5-HT3 is v similar to ACh binding to nAChR, involving Trp B and the 5-hydroxy group facing the complementary subunit, but in 5-HT3R there's a loop A glutamate (arginine in nAChR) and a pre-loop B threonine (lysine in most nAChRs and AChBPs)
[Kesters et al 2013]
Anion-selective pLGICs have a TyrB or PheB instead of TrpB, maybe reducing binding site size so it's more amenable to binding small molecules like glycine and GABA. But cation-pi interactions still form with these alternative residues. A loop C threonine interaction with the gamma carboxy terminal of inhibitory receptor agonists seems to be important for specificity, since mutation to residues without OH impairs recognition in GlyRs and GABARs (reviewed in Lynagh and Pless, 2014)
GluCl receptor gating
inhibited by glutamate, activated by ivermectin (an allosteric agonist, binds between M1 and M3
there's a proline in the M2-M3 loop. If you mutate it to an unnatural proline analogue resembling cis proline, it works. Trans-proline, it doesn't work.
similar rotation movements of ECD and tilting movements of M2 to cation-selective channels
A valine in the B1-B2 loop contacts that proline and pushed it, to cause rigid body movements - the ECDs twist towards each other, called 'unblooming'
BUT note Bertozzi et al 2016 - found in GLIC and ELIC that this B1-B2 and M2-M3 connection was expendable
Yuan et al produced the first high resolution structure of a mammalian pLGIC with the 5-HT3R.
They then used molecular dynamics simulations
Binding of 5-HT to the aromatic cage puts a stress on the receptor, which is relaxed via multiple structural transitions
First a Trp in every subunit relaxes outward, then the entire subunit does a 6°tilt then 6°twist, unblooming to increase ECD pore volume.
This is transmitted via the b1-b2 and cys loops to the M2-M3 loop, causing first a tilting outward of the M2 helices, then 13'valine and 9'leucine, which form the hydrophobic girdle, rotate around an axis parallel to the membrane.
Authors hypothesis that after their stimulation time, the pore must widen further somehow
Then the structural changes are transmitted to the ICD, whose vestibule enlarges and whose portals open
Myasthenia gravis - autoimmune channelopathy, Abs to the muscle nAChR, treated with anti-cholinesterases
Alzheimer's disease - low concs and oligomeric beta-amyloid activate alpha7 [calcium permeable] nAChR. nM concs and fibrillar inhibit it. We don't know why.
Slow channel syndrome - caused by mutations most frequently in M2 or M2-M3. Increase channel mean open time, causing spasms.
GABA clinical aspects
binds BDZs (and ethanol, barbiturates, neuroactive steroids, anaesthetics) at allosteric site. Those that bind at alpha 1 or 5 are sedative/amnesic, those that bind at alpha2/3/4 are anxiolytic
Also can be myorelaxant, anticonvulsant.
Since it's got amnestic effects, there's been some investigation into targeting GABA to promote STM, but these drugs are mostly pro-convulsant
ELIC can help us identify why some drugs are excitatory and some inhibitory - has two BZ binding sites, one behind binding pocket and one overlapping with it.
Epilepsy - 1 in 26 people get it, 4th most common neurological condition. CAE is 2-8% of patients
Various mutations associated with CAE and other forms of epilepsy, mostly in ECD
E.g. Arg-Gln at beta2-gamma2 interface prevents oligomerisation of subunits, so subunits retained in ER, so less GABAR at surface. CAE, febrile seizures.
E.g. Lys-Met in M2-M3 loop decreases mean open time and accelerates deactivation (so destabilises open state). Generalised epilepsy, febrile seizures.
5-HT3R clinical aspects
Tyrosine to serine polymorphism in 5-HT3B subunit is common in many populations (though frequency varies between populations). Increases mean channel open time 7-fold, inversely correlated with incidence of MDD in women