In 2002 a paper in Nature did something the field had failed to do for a decade: it put a structure on desensitisation. Working with the GluR2 AMPA receptor, the authors showed that the ligand binding cores assemble as dimers, and that stabilising the interface inside each dimer — by mutation or by an allosteric modulator — reduces desensitisation, while perturbations that destabilise the same interface make it worse.
That result reframed a whole class of compounds. It explains how ampakines slow AMPA receptor currents, and why the question is one about a protein interface rather than about the channel pore.
Why an AMPA Receptor Closes While Glutamate Is Still Bound
Two different processes shut the current down.
Deactivation is what happens when the agonist leaves. A synaptic pulse of glutamate lasts on the order of a millisecond, and once the transmitter unbinds, the clamshell shaped ligand binding core springs open and the channel shuts. Desensitisation is stranger: the agonist remains bound and the channel closes anyway, because the two ligand binding cores of a dimer slip against one another and uncouple each subunit from the gate it is supposed to pull on.
Electrophysiology separates the two by how the agonist is applied. A one millisecond pulse onto an excised outside out patch reports deactivation kinetics; a prolonged application reports desensitisation, because the current rises and then decays while glutamate is still there. The two answers are often not the same.
How Ampakines Slow AMPA Receptor Desensitisation and Deactivation
Ampakines are, in the definition used in the review literature, drugs structurally derived from aniracetam that potentiate currents carried by AMPA type glutamate receptors. A 2007 review in Current Drug Targets summarises the shared signature: they slow deactivation, attenuate desensitisation, increase synaptic responses and facilitate long term potentiation in hippocampal slices. A 2005 review of preclinical work adds the constraint that matters most — these compounds have no intrinsic activity, and act only by altering the rate at which an already activated receptor desensitises.
A modulator with no intrinsic agonist action cannot produce a current where glutamate is absent. What it can do is change the shape and duration of a current that synaptic release has already triggered.
The magnitudes are measurable rather than dramatic. In cultured rat hippocampal neurons, the benzothiadiazide 7-chloro-3-methyl-3,4-dihydro-2H-1,2,4-benzothiadiazine-S,S-dioxide prolonged autaptic AMPA currents to 5.6 times control, with a half maximal effective concentration of 150 micromolar, and slowed AMPA deactivation about threefold after one millisecond applications of one millimolar glutamate. The same 1998 study recorded an inconvenient second action: the compound inhibited whole cell currents evoked by exogenous GABA by 41 percent while slightly augmenting GABAergic synaptic currents.
Where Positive Allosteric Modulators Bind in the Dimer Interface
The binding site was resolved by crystallography in 2005. Cocrystal structures of the GluR2 S1S2 ligand binding domain with aniracetam and with the ampakine CX614 placed both molecules inside the dimer interface of the non desensitised receptor, at a common site on the twofold axis of molecular symmetry. That pocket sits next to the hinge of the clamshell, the part that rearranges once glutamate is bound. Point mutations of the residues contacting the modulators disrupted modulator function in patch clamp recordings. The proposed action is specific: the modulator stabilises the clamshell in its closed cleft, glutamate bound conformation, and deactivation slows because the agonist takes longer to escape.
Later work mapped the same pocket with a chemical series. A 2009 study in Biochemistry probed the modulator site of GluR2 with thiazide derivatives, noting that the desensitised state is routinely removed for simplified analysis of receptor activation using cyclothiazide, the most potent modulator of that family. Thermodynamics followed in 2012, when isothermal titration calorimetry on a stabilised dimeric mutant of the GluA2 ligand binding domain measured dissociation constants directly. The reference modulator BPAM-97 bound with a dissociation constant of 5.6 micromolar; a displacement assay placed the benzothiadiazine IDRA-21 at 0.46 millimolar. Crystallography attributed most of that difference to van der Waals contacts made by a single ethyl substituent, primarily with Met496.
Two Ampakine Subfamilies That Do Not Compete in Binding Assays
The 2007 review divides it into two subfamilies on physiological rather than structural grounds. Type I compounds such as CX546 are very effective at prolonging synaptic responses; type II compounds such as CX516 mainly increase response amplitude. The two classes do not compete in binding assays, which is the evidence that they occupy separate sites. All the ampakines examined facilitated long term potentiation, but only CX546 enhanced long term depression, and effects varied by cell type.
A 2001 study in Neuropharmacology compared both chemotypes on gating kinetics. CX546 reduced desensitisation more potently than CX516 or IDRA-21, but less efficiently than cyclothiazide. It raised agonist affinity about threefold on non desensitising receptors by slowing agonist unbinding, and the analysis pointed to something the benzothiadiazides do not do: CX546 appeared to bind specifically to the agonist bound, non desensitised receptor, destabilising the desensitised conformation rather than preventing entry into it.
| Compound | Chemical class | Effect reported |
|---|---|---|
| Cyclothiazide | Benzothiadiazide | Most potent block of desensitisation in the series |
| IDRA-21 | Benzothiadiazide | Partial modulator; millimolar dissociation constant by displacement |
| CX516 | Benzoylpiperidine, type II | Mainly increases response amplitude |
| CX546 | Benzoylpiperidine, type I | Prolongs synaptic responses; slows agonist unbinding |
| CX614 | Benzoxazine | Cocrystallised in the dimer interface; slows deactivation |
Why a Single Alkyl Substituent Changes the Kinetic Signature
The benzothiadiazide series makes that relationship legible. A 2002 study in Molecular Pharmacology found that the size of the 5 prime substituent determined not only potency but the character of the effect. The 5 prime ethyl derivative, called D1, blocked desensitisation with a half maximal effective concentration of 36 micromolar and slowed deactivation of responses to one millisecond glutamate pulses more than tenfold. Radioligand binding to rat synaptic membranes increased 3.6 fold, with a Hill coefficient near two.
The split between amplitude and duration tracked the substituent. Compounds carrying a 5 prime methyl group had roughly twice as large an effect on amplitude as on duration; 5 prime ethyl compounds such as D1 did the reverse. D1 also preferred the GluR4 flip splice variant, at 0.64 micromolar, over GluR4 flop at 5.3 micromolar, the kind of splice variant selectivity that makes these modulators useful as subtype separating tools.
How TARP Subunits Change Ampakine Pharmacology
A native AMPA receptor is not a bare tetramer. Transmembrane AMPA receptor regulatory proteins, abbreviated TARPs, govern surface expression and set agonist affinity, desensitisation and deactivation kinetics. A 2018 study in Biochemical Pharmacology tested CX614 and cyclothiazide on homomeric GluR1 flip and GluR2 flop receptors expressed in HEK293 cells with or without gamma-2, gamma-3, gamma-4 or gamma-8.
Gamma-4 gave the largest increase in the affinities of both modulators on GluR1 flip, yet had no such effect on GluR2 flop, where gamma-8 produced the biggest increase instead. The influence of TARPs on ampakine pharmacology is therefore complex, depending jointly on the TARP subtype and on the receptor subunit and splice isoform present.
This bears directly on how an in vitro number should be read. A potency measured on a recombinant homomer is a property of that expression system, not a constant of the compound.
What Happened When the Chemistry Reached a Clinical Trial
The mechanistic literature is considerably stronger than the clinical record. A phase II randomised, double blind, placebo controlled trial of CX516 in fragile X syndrome, published in 2006, enrolled 49 subjects through a one week placebo lead in and a four week treatment period. Side effects were minimal and no serious adverse events occurred, although allergic rash appeared at a frequency of 12.5 percent in the CX516 arm.
The primary outcome measure was memory, and there was no significant improvement in it, nor in secondary measures of language, attention and executive function, behaviour or overall functioning. The authors flagged that potency problems in other studies left it unclear whether modulation of AMPA mediated neurotransmission is a viable therapeutic strategy in that condition at all. A null trial with an acknowledged potency caveat settles little, and it does not convert a dimer interface structure into a clinical claim.
Upstream of all of this sits a laboratory constraint. Electrophysiology at micromolar and millimolar concentrations is interpretable only when the identity, purity and, where a stereocentre exists, the enantiomeric composition of the test compound are documented. Material handled in a research setting is subject to the terms governing laboratory reagents.
Frequently asked questions
What is the difference between AMPA receptor desensitisation and deactivation?
Deactivation is closure of the channel after the agonist unbinds, measured with brief glutamate pulses on excised patches. Desensitisation is closure while the agonist stays bound, measured during prolonged application. Structural work on GluR2 traced desensitisation to rearrangement of the dimer interface, which uncouples the ligand binding core from the gate.
Where do positive allosteric modulators bind on the AMPA receptor?
Inside the dimer interface of the ligand binding domain, not in the channel pore. Cocrystal structures of the GluR2 S1S2 domain with aniracetam and CX614 placed both at a common site on the twofold axis of symmetry, beside the clamshell hinge. Mutating the contacting residues disrupted modulator function.
Do all ampakines act at the same binding site?
No. Physiological work separates at least two subfamilies. Type I compounds such as CX546 mainly prolong synaptic responses, while type II compounds such as CX516 mainly increase amplitude, and the two do not compete in binding assays. That lack of competition indicates distinct sites.
Why is cyclothiazide more potent than IDRA-21 in these assays?
Because of contacts made by a small substituent. Calorimetry on a stabilised GluA2 ligand binding domain measured a dissociation constant of 5.6 micromolar for the reference modulator BPAM-97 against 0.46 millimolar for IDRA-21, a difference attributed largely to van der Waals contacts from an ethyl group.
Do TARP proteins change ampakine potency?
Yes, and not uniformly. Testing CX614 and cyclothiazide in HEK293 cells showed that gamma-4 gave the largest affinity increase on GluR1 flip receptors but not on GluR2 flop, where gamma-8 dominated. The effect depends on TARP subtype and on the splice isoform expressed.
What happened in the CX516 fragile X clinical trial?
The phase II trial randomised 49 subjects across a four week treatment period and reported minimal side effects, with allergic rash in 12.5 percent of the treated arm. Its primary outcome, memory, showed no significant improvement against placebo, and neither did the secondary measures.
References
- Sun, Olson, Horning et al., Nature, 2002 — mechanism of glutamate receptor desensitization
- Jin, Clark, Weeks et al., Journal of Neuroscience, 2005 — mechanism of positive allosteric modulators acting on AMPA receptors
- Arai and Kessler, Current Drug Targets, 2007 — pharmacology of ampakine modulators: from AMPA receptors to synapses and behavior
- Arai, Xia, Kessler et al., Molecular Pharmacology, 2002 — effects of 5-prime-alkyl-benzothiadiazides on AMPA receptor biophysics
- Nagarajan, Quast, Boxall et al., Neuropharmacology, 2001 — mechanism and impact of allosteric AMPA receptor modulation by the ampakine CX546
- Yamada, Hill, Hu, Covey, Neurobiology of Disease, 1998 — a diazoxide derivative augments AMPA- and GABA-mediated synaptic responses
- Krintel, Frydenvang, Olsen et al., Biochemical Journal, 2012 — thermodynamics and structural analysis of positive allosteric modulation of GluA2
- Ptak, Ahmed, Oswald, Biochemistry, 2009 — probing the allosteric modulator binding site of GluR2 with thiazide derivatives
- Radin, Li, Rogers et al., Biochemical Pharmacology, 2018 — TARPs differentially affect the pharmacology of ampakines
- Berry-Kravis, Krause, Block et al., Journal of Child and Adolescent Psychopharmacology, 2006 — effect of CX516 in fragile X syndrome: a controlled trial
- Black, Psychopharmacology, 2005 — therapeutic potential of positive AMPA modulators: preclinical data
Research use only. Nonsensia supplies analytical reference standards for laboratory and research applications. This article is published for scientific and educational purposes. It is not medical advice, it does not describe any use in humans, and nothing in it should be read as a recommendation to administer any substance to a person or animal.
Filed under: Nootropics Research
This article is part of our guide to The Evidence Base for Cognition Research Compounds: A Critical Guide.
The standard for the compound discussed in this article is currently unavailable. Related analytical reference standards in stock: Aniracetam ≥99% – Analytical Reference Standard | CAS 72432-10-1, Coluracetam ≥99% HPLC – Analytical Reference Standard | CAS 135463-81-9, Noopept (Omberacetam) ≥99% HPLC – Analytical Reference Standard | CAS 157115-85-0, Oxiracetam ≥99% HPLC – Analytical Reference Standard, 1000 mg | CAS 62613-82-5.