Positive AMPA receptor modulation in the treatment of neuropsychiatric disorders: A long and winding road

Abstract

IGlutamatergic transmission is widely implicated in neuropsychiatric disorders, and the discovery that ketamine elicits rapid-acting antidepressant effects by modulating α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor (AMPAR) signaling has spurred a resurgence of interest in the field. This review explores agents in various stages of development for neuropsychiatric disorders that positively modulate AMPARs, both directly and indirectly. Despite promising preclinical research, few direct and indirect AMPAR positive modulators have progressed past early clinical development. Challenges such as low potency have created barriers to effective implementation. Nevertheless, the functional complexity of AMPARs sets them apart from other drug targets and allows for specificity in drug discovery. Additional effective treatments for neuropsychiatric disorders that work through positive AMPAR modulation may eventually be developed.

Over the past two decades, the direct modulation of the glutamatergic system has received considerable attention as a novel approach for treating major depressive disorder (MDD).^1–3 This attention was catalyzed by emerging clinical evidence showing that subanesthetic doses of the noncompetitive allosteric N-methyl-D-aspartate receptor (NMDAR) antagonist ketamine have rapid and long-lasting antidepressant effects in patients with treatment-resistant depression.⁴

Ketamine is a synthetic phencyclidine derivative with one chiral center made up of two enantiomers ((R)-ketamine and (S)-ketamine). It has been used as a dissociative anesthetic and analgesic drug for over 60 years. The (S)-enantiomer of ketamine has three- to four-fold higher affinity binding to the NMDAR than its (R)-counterpart, is a more potent anesthetic and analgesic, and was also selected for development as an antidepressant. However, recent evidence suggests that (R)-ketamine may elicit sustained antidepressant effects without the side effects associated with (S)-ketamine.^5,6 (S)-ketamine (esketamine) nasal spray was approved by the United States Food and Drug Administration (FDA) in March 2019 as an adjunctive treatment for patients with treatment-resistant depression and in 2020 for adults with major depression with acute suicidal ideation or behavior⁷; it was also approved by the European Commission in December 2019 for the same indications. Esketamine currently needs to be administered in conjunction with Risk Evaluation and Mitigation Strategy (REMS) precautions due to its transient dissociative and psychotomimetic side effects and potential abuse liability, which has also been reported at antidepressant doses.⁷

This recent approval of esketamine for treatment-resistant depression represents a major breakthrough in the field of psychiatry, where no mechanistically novel drugs had been developed for over four decades. The discovery also led to direct and indirect glutamatergic modulators being considered as an entirely new class of antidepressant. As a result, clinical studies have assessed the antidepressant potential of several alternative voltage-dependent NMDAR agents, including GluN2B-selective allosteric modulators (CP-101,606, CERC-301, and Ro 25–6981)⁸ and glycine binding site modulators (AV-101 and rapastinel).^9,10 Despite promising preliminary results at either the preclinical and/or early clinical stages, these molecules failed to duplicate the rapid, robust, and clinically meaningful antidepressant effects of ketamine,¹¹ suggesting that mechanisms other than NMDAR inhibition may be involved in reproducing ketamine’s therapeutic effects.^12,13 A subsequent surge of preclinical studies attempted to back-translate clinical findings in order to identify the cellular and molecular mechanisms of action underlying ketamine’s unique antidepressant effects and to develop other glutamatergic modulators that lacked the undesirable side effects induced by NMDAR antagonism (including psychotomimetic or dissociative symptoms, possible abuse liability, and evidence of brain lesions associated with chronic abuse).^8,14–16

Much is still unknown about the cellular actions of ketamine, but growing evidence suggests that its antidepressant effects are associated with rapid improvements in synaptic plasticity in executive areas of the brain.^17–19 For instance, one of the striking effects of acute ketamine administration in rodents is its ability to rapidly restore (within 24 hours of administration) dendritic arborization and the density of synaptic spines reduced by chronic stress, an effect observed only after several weeks of treatment with traditional antidepressants^20,21; it should be noted, however, that some of these effects appear to be sex-dependent.^22,23 This effect coincides with the peak of ketamine’s antidepressant effects and is considered essential for its therapeutic actions. The molecular mechanisms thought to be involved in these processes include enhanced α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) receptor (AMPAR) throughput relative to the NMDAR, which leads to increased release of brain-derived neurotrophic factor (BDNF) at the synapses and to rapid and transient activation of mammalian target of rapamycin complex (mTORC) signaling.²⁴ This increase in mTORC signaling was shown to increase the local expression of synaptic proteins, consistent with the rapid formation of new dendritic spines observed after ketamine treatment.^20,25 In addition, the antidepressant-like effects of ketamine in rodents were abolished in the presence of NBQX (2,3-dioxo-6-nitro-7-sulfamoyl-benzo[f] quinoxaline), an AMPAR antagonist.^14,15,26

Further confirming the key role of AMPAR activation in inducing rapid antidepressant effects, preclinical studies demonstrated that the ketamine metabolite (2R,6R)-hydroxynorketamine (HNK) exerts AMPAR-dependent sustained antidepressant effects without displaying significant binding affinity for NMDARs.¹⁵ Furthermore, the antidepressant actions of (2R,6R)-HNK appear to be related to acute increases in excitatory neurotransmission resulting by AMPAR activation followed by long-term adaptation characterized by increased levels of GluA1 and GluA2 AMPAR synaptic subunits.^15,27 Interestingly, the antidepressant properties of (2R,6R)-HNK were not associated with psychotomimetic effects, suggesting that direct modulation of AMPARs, excluding NMDAR antagonism, might preserve antidepressant efficacy while limiting dissociative side effects. Researchers examining the potential involvement of (2R,6R)-HNK in NMDAR function further observed off-site effects at supraphysiologic doses but not at antidepressant-relevant concentrations.^28–30

Taken together, these data support the notion that direct positive modulation of AMPARs could be a promising approach for developing rapid-acting, efficacious, and safer glutamatergic antidepressants. Presently, however, most research in this field has been performed exclusively in preclinical settings, and few selective AMPAR agents have been clinically evaluated for neuropsychiatric indications. Despite the notable challenges in developing AMPAR-modulating therapies, including low potency and poor metabolic stability, a clear impetus exists to explore the therapeutic potential of this receptor.

This review specifically explores the manner in which targeting positive AMPAR modulation may help expand the armamentarium of novel neuropsychiatric drugs. We discuss the structural and functional properties of AMPARs, with an emphasis on AMPAR positive allosteric modulators (PAMs) and other related compounds; describe attempts to develop translational biomarkers by modulating AMPA throughput; provide an in-depth review of direct AMPAR modulators; introduce promising research in the field of indirect AMPAR modulators; and discuss challenges and unresolved problems associated with this line of inquiry. It should be noted that a full exploration of AMPAR antagonist modulation is outside the scope of this review.

Structural and functional properties of AMPARs

AMPARs are ionotropic, homo- or hetero-tetrameric complexes composed of four subunits (GluA1–4) with a unique base architecture. Four distinct domains comprise the receptor: extracellular N-terminal and ligand-binding domains, the transmembrane domain that forms the ion channel, and a C-terminal domain located in the cytoplasm. The extracellular and transmembrane domains are connected through peptide linkers that give the receptor unique flexibility and allow for complex gating behaviors.^31,32 Particularly important in drug discovery are the upper (D1) lobes of the ligand-binding domain,³³ where adjacent subunits dimerize back-to-back to form acceptor sites for PAMs. Notably, a number of AMPA-modulating drugs, including ampakines, bind to allosteric sites that lie in proximity to the agonist site, at an interface between GluA subunits. This D1 interface is also targeted by mRNA-processing events involving the alternative splicing (flip/flop) of exons 14 and 15, which can change the functional selectivity of these binding sites as well as glutamate sensitivity (reviewed in³⁴). The flip isoform, which confers hyperexcitability, has already been targeted in disorders such as epilepsy,³⁵ although no specific therapeutics have been developed or tested.

Adding to this complexity, each subunit confers distinct channel ion selectivity and kinetic properties, as well as functional and trafficking properties. The state of one subunit can also influence the binding properties of another, with research showing that the affinity of the ligand-binding domain increases when other ligand-binding domains are occupied.³⁶ Moreover, the N-terminal domain of GluA3 (which is primarily responsible for anchoring the receptor at the synapse and restricting mobility) is distinctively flexible and has unconstrained lower lobes that underlie the important role of GluA3 in allosteric gating regulation.³⁷ The ability of a drug to influence this increased motility may be important when considering pharmacotherapy for depression, as individuals with MDD appear to have significantly decreased levels of GluA3.³⁸ Receptors that contain the GluA2 subunit restrict Ca²⁺ permeability, decreasing channel conductance. These reductions in conductance have been shown to play an important role in synaptic plasticity because they consolidate long-term potentiation after initial NMDAR-independent facilitation through Ca²⁺-permeable AMPARs.³⁹ Improving the throughput of both Ca²⁺ permeable and Ca²⁺ impermeable AMPARs could be key to developing new therapeutics for depression, given that researchers have observed drastic changes in synaptic connectivity associated with depression (reviewed in¹⁷).

Aside from the main pore-forming subunits, the expression of auxiliary subunits can further differentiate the functionality of AMPARs, altering aspects of channel conductance and gating as well as AMPAR expression at the synapse.^40,41 Although relatively little is known about these mechanisms of action, they should be kept in mind when discussing the advantages and pit-falls associated with different positive AMPAR modulators in the treatment of depression and other neuropsychiatric disorders. The structural and functional disparities seen among AMPARs underscore the diversity of their possible roles in the central nervous system.

AMPA throughput and gamma oscillations: Converging evidence of a cross-species translational biomarker

Gamma oscillations in the brain are intricately regulated by AMPAR-mediated depolarization and gamma aminobutyric acid (GABA)_A modulation, which induce an altered excitatory/inhibitory balance that results in pathological brain states.^42,43 Moreover, several lines of evidence indicate a direct relationship between gamma oscillation and depression.⁴⁴ Notably, acute subanesthetic-dose ketamine infusion is associated with robust increases in gamma power, as demonstrated by clinical research.^45,46 Multiple forms of receptor-mediated modulation are involved in ketamine’s mechanism of action at the synapse, and recent data indicate that ketamine both silences GABAergic inhibitory synapses and increases glutamate release, thus enhancing AMPAR throughput.^8,47 Research suggests that the mechanism underlying the increased gamma oscillations⁴⁸ and gamma power observed in response to ketamine administration may be associated with decreased activity in GABAergic interneurons and with the disinhibition of excitatory pyramidal neurons.^49,50 Studies also found that 0.5 mg/kg of ketamine, administered intravenously to healthy volunteers, resulted in a robust (greater than four-fold) increase in resting state gamma power with strong effects in the prefrontal cortex; this increase peaked at the end of the infusion and persisted for at least three hours.⁵¹ The mechanism of antidepressant response to ketamine is, however, likely to be more complex, given that NMDAR blockade of interneurons and the subsequent disinhibition of pyramidal neurons does not consistently produce antidepressant effects.⁵² Acute administration of the ketamine metabolite (2R,6R)-HNK also increases gamma oscillations, even though this metabolite does not inhibit NMDARs at concentrations that increase gamma power.¹⁵ Because AMPAR blockade can attenuate HNK-induced gamma oscillations, enhanced AMPAR activity is likely to be the mechanism by which HNK increases gamma power, although the precise mechanism underlying this increased AMPAR activity remains unknown.⁵³ Taken together, these findings underscore the close relationship between gamma oscillations and inhibition/excitation balance.^42,43 One theory posits that ketamine’s rapid antidepressant effects may be initiated by its antagonistic effects on NMDARs, and that these effects regulate glutamate/GABA and/or AMPAR/NMDAR activity. The relatively sustained effects seen in individuals treated with ketamine suggest that long-term changes in neural plasticity are also likely to play a role.

In addition, quantitative electroencephalography (qEEG) performed pre- and post- ketamine challenge in both rodents and humans found substantial increases in gamma-band power that have been implicated in the activation of fast ionotropic excitatory receptors, specifically AMPARs.^54,55 This effect was also found to be associated with the (2R,6R)-HNK metabolite.¹⁵

Together, the findings to date suggest that sustained and rapid antidepressant effects are associated with AMPAR activation, encouraging continued investigation into therapies that target AMPARs for treating depression.

AMPAR positive modulators in drug discovery

AMPARs and their modulators were first considered as important targets for drug discovery in depression for two major reasons First, because there is evidence that the glutamatergic system may be dysregulated in neuropsychiatric disorders and, second, because these components play an essential role in the synaptic plasticity that underlies learning and memory (reviewed in^56,57). Direct agonist-induced activation of AMPARs was initially explored, but this avenue was tempered by the realization that global activation of the central nervous system did not allow for sufficient processing of the generated signals. In contrast, positive modulation allows for greater sensitivity and specificity and has been more promising in treating neuropsychiatric disorders (reviewed in⁵⁸). AMPAR PAMs can increase throughput in multiple ways, including slowing the rate of receptor desensitization by preventing conformational transitions or prolonging glutamatergic synaptic currents.⁵⁹ Unlike the direct global approach, this process does not corrupt spatial or temporal information, allowing AMPARs to communicate effectively. Below, we review the different AMPAR PAMs, including nootropic racetams, AMPAkines, and other modulators that have been studied over the years for their putative neuropsychiatric effects. Additional study details are provided in Table 1, and the chemical structure of each AMPAR PAM can be found in Fig. 1.

TABLE 1.

AMPAR positive allosteric modulators (PAMs).

Drug name	Reference	Diagnostic groups	Dosing and treatment groups	Primary outcome measure	Region of interest	Phase of development	Significance and findings
AMPAKINES
CX516	71	Cognitive Enhancement	CX516 (35 mg/kg)	DNMS paradigm	N/A	Preclinical	CX516 improved performance on the DNMS paradigm, even on days when the drug was not administered.
	72	Depression	CX516 (5 mg/kg)	Reducing submissive behavior	N/A	Preclinical	CX516 significantly reduced submissive behavior in Sprague-Dawley rats.
	70	Scn2a Haplo-insufficient mice (ADHD model)	CX516 (10–40 mg/kg)	OFT, EPM, LDB, 3-chamber social interaction, social memory test, SDTT, resident-intruder, fear conditioning and extinction	N/A	Preclinical	CX516 improved measures of hyperactivity but did not rescue measures of anxiety-like behaviors in mice.
	73	Age-related decline	CX516 (30 mg/kg)	BDNF and NGF mRNA expression	Hippocampus, piriform cortex, amygdala, dentate gyrus	Preclinical	CX516 administration to Sprague-Dawley rats significantly increased BDNF and NGF mRNA expression in the dentate gyrus, amygdala, and piriform cortex.
	69	Tm4sf2−/y mice (model of intellectual disability)	CX516 (5 mg/kg for 5 days)	LDB, emergence and fear conditioning, MWM, electrophysiological measures	Hippocampus	Preclinical	All behavioral measures were improved after CX516 administration, excitatory/inhibitory balance was restored, and synaptic potentiation was increased.
	67	Schizophrenia	CX516 (300–900 mg over 4 weeks)	Cognitive Battery and Neuropsychological Assessments (not otherwise described)	N/A	Case study	As monotherapy, CX516 did not improve measures of psychosis or cognition in schizophrenia patients.
	66	Schizophrenia	CX516 (900 mg, TID)	Cognitive battery, PANSS-Positive and General Subscales	N/A	Clinical trial	CX516 had no statistically significant clinical effect or benefit.
	NCT00040443	Mild cognitive impairment	CX516 (900 mg)	15-Item Word List Delayed Recall	N/A	Clinical trial	CX516 modestly improved scores on the 15-Item Word List Delayed Recall, a measure of short-term episodic memory.
	68	Fragile X	CX516 (300 mg)	Vineland Adaptive Behavior Scale	N/A	Clinical trial	Inconclusive with regard to antidepressant effects in patients with Fragile X syndrome.
CX546	75	mGluR5-deficient mice	CX546 (15 mg/kg)	LI, PPI	N/A	Preclinical	CX546 significantly improved LI and PPI from disruptions caused by mGluR5 knockout.
	74	PCP-induced cognitive deficits	CX546 (10, 40, or 80 mg/kg)	NOR task	N/A	Preclinical	CX546 reversed object memory after administration of PCP to Lister-hooded rats.
	79	Neurogenesis	CX546 (50 μm)	BrdU, NeuN, TUNEL (cell viability assay), Tau	Subventricular zone	Preclinical	Increased neurogenesis and dendritogenesis were observed after CX546 treatment of stem and progenitor cell cultures, with no increase in cell death.
	77	Shank3 mutant mice (ASD model)	CX546 (5–20 mg/kg)	Three chamber test, social interaction, electrophysiological measures	Hippocampus	Preclinical	At 10 mg/kg and above, CX546 enhanced hippocampal EPSCs, the AMPA/NMDA ratio, and all behavioral measures.
	76	ASD mice models (Shank3 mutant and BTBR)	CX546 (15 mg/kg)	Three chamber test, male/female reciprocal interaction, OFT, grooming behavior	N/A	Preclinical	No changes in all measures of social interaction or exploratory behavior, but CX546 did increase self-grooming behavior.
	78	Synaptic plasticity	CX546 (250 μm daily for 2 weeks)	Electrophysiological recordings, synaptophysin, cell viability assay	Hippocampus	Preclinical	Prolonged AMPAkine exposure in hippocampal slice cultures pruned dendritic spines and increased efficiency in synaptic communication in the hippocampus. It did not cause cell death.
CX614	85	Behavior	CX614 (1 mg/kg, 4 mg/kg)	FST, photoresistor actimeter (locomotor activity)	N/A	Preclinical	CX614 decreased time spent immobile in the FST but did not affect general locomotor activity in either Albino Swiss mice or Wistar rats. The effects were abolished by administration of NBQX, an AMPAR antagonist.
	86	Spatial navigation	CX614 (2.5 mg/kg)	Y-maze, gamma oscillation frequency and power	Hippocampus	Preclinical	AMPAR modulation with CX614 differentially affected slow and fast gamma of Long Evans rats of different ages. The rate of slow and fast gamma events decreased in older animals administered CX614, but the peak power of fast gamma events increased in young animals administered CX614.
	83	Synaptic plasticity	CX614 (10 μm, 50 μm)	BDNF release and phosphorylation of TrkB, mTOR, p-70S6K, and 4EBP1	Hippocampal slices, neuronal cultures	Preclinical	CX614 rapidly stimulated local dendritic protein synthesis mediated by increased BDNF release.
	81	Structural plasticity, dendritogenesis	CX614 (0.5–10 μm) vs. ketamine (0.001–10 μm)	Tyrosine hydroxylase, GluA1, GluA2, TrkB, BDNF, p-70S6K	Mesencephalic dopaminergic neurons	Preclinical	CX614 mimicked the actions of ketamine in promoting dendritogenesis and structural synaptic plasticity through a BDNF-mediated mechanism.
	80	MPP(+) induced toxicity	CX614 (50 μm)	BDNF expression, lactate dehydrogenase release	Hippocampal and mesencephalic cultured slices	Preclinical	Pre-treatment with CX614 significantly reduced MPP(+)-induced toxicity mediated by an increase in BDNF.
	87	R6/2 mice (Huntington’s disease model)	CX614 (20–200 mM)	Electrophysiological recordings	Medium striatal spiny neurons	Preclinical	CX614 significantly increased the synaptic activity of both wild-type and R6/2 mice.
	84	Depression	CX614 (50 μm, incubated for 5 durations from 3 h to 48 h)	AMPAR (GluA1–3) mRNA and protein expression, BDNF expression, TrkB phosphorylation	Hippocampal slices	Preclinical	Prolonged (12–24 h) treatment with CX614 reduced GluA1–3 mRNA, increased BDNF levels, and increased synaptic TrkB activation.
	82	Depression	CX614 (1 mg/kg) in addition to imipramine, reboxetine, or escitalopram	FST	N/A	Preclinical	When used adjunctively with imipramine and reboxetine in albino Swiss mice, CX614 enhanced the medication’s antidepressant effects. There was no enhancement with escitalopram.
CX691	88	Behavior	CX691 (0.1 mg/kg, 0.3 mg/kg, 1 mg/kg)	BDNF mRNA expression, scopolamine-induced deficits in cued fear conditioning, a temporal (24 h)-induced deficit in NOR, attentional set shifting	Whole brain	Preclinical	In Lister hooded rats, CX691 attenuated scopolamine-induced impairments of cued fear conditioning and, at sub-chronic doses, improved attentional set-shifting and increased BDNF mRNA expression.
	89	Alzheimer’s disease	CX691 (0.03–0.3 mg/kg for 10 days)	BDNF protein levels, MWM	Hippocampus	Preclinical	CX691 rescued hippocampal BDNF levels and performance on the spatial memory task after amyloid-beta injection to Wistar rats.
	73	Age-related decline	CX691 (3 mg/kg)	BDNF and NGF mRNA expression	Hippocampus, piriform cortex, amygdala	Preclinical	CX691 significantly increased BDNF and NGF mRNA expression in the somatosensory cortex, amygdala, and piriform cortex of Sprague-Dawley rats.
	90	Healthy elderly volunteers	CX691 (500 mg)	Wordlist learning, picture memory, N-back, symbol recall, maze task, pursuit rotor, symbol digit substitution test, continuous trail making test	N/A	Clinical trial	CX691 improved measures of short-term memory but impaired episodic memory.
CX717	92	Cognitive enhancer	CX717 (0.3–1.5 mg/kg)	[18F]FDG PET imaging, DMS	Dorsolateral prefrontal cortex, medial temporal lobe, dorsal striatum	Preclinical	CX717 reversed reductions in cognitive performance and brain activation that were induced by sleep deprivation, potentially by increasing the firing of task-specific hippocampal cells.
	94	Alzheimer’s disease	CX717 (20 mg/kg)	5CSRTT, NOR task	N/A	Preclinical	CX717 reduced the number of incorrect responses in the 5CSRTT in both control animals and in a model of cognitive impairment induced by bilateral vestibular differentiation of Wistar rats. However, this dose had detrimental effects in the NOR task.
	91	Depression	CX717 (20 mg/kg)	FST, BDNF, noradrenaline, dopamine, serotonin, glutamate	Prefrontal cortex	Preclinical	Within 30 minutes, CX717 transiently improved performance on the FST and increased all biological measures in Sprague-Dawley rats.
	95	Cognitive enhancer	CX717 (100–1000 mg	Polysomnography, EEG, cognitive battery	N/A	Clinical (phase unknown)	CX717 administered after sleep deprivation did not affect cognitive measures, but the highest dose (1000 mg) modestly helped recovery sleep.
	NCT03375021	ADHD	CX717 (200–800 mg)	ADHD-Rating Scale, CGI, HAM-D, ADHD-Self Rating Scale, HAM-A, Pittsburgh Sleep Quality Index, Stroop Color and Word Test, Trail Making Test, Continuous Performance Task, Forward and Backward Digit Span	N/A	Clinical – Phase 2	Results not yet posted.
Org 26576	96	Depression	Org 26576 (1–10 mg/kg)	BrdU	Hippocampus and prelimbic cortex	Preclinical	Chronic administration to Sprague-Dawley rats increased cell proliferation. Cells born after Org 26576 treatment had increased survival rates. Acute administration had no effect.
	97	Acute stress	Org 26576 (10 mg/kg)	BDNF mRNA	Hippocampus, frontal and prefrontal cortex	Preclinical	BDNF mRNA was increased after Org 26576 treatment in the hippocampus, but not in the frontal or prefrontal cortex, of Sprague-Dawley rats.
	100	Schizophrenia	Org 26576 (3–30 mg/kg) plus the antipsychotics risperidone, olanzapine, and haloperidol	Conditioned avoidance response, catalepsy test, electrophysiological measures	Medial prefrontal cortex	Preclinical	Administered adjunctively with risperidone and olanzapine, but not haloperidol, Org 26576 improved behavioral measures in Wistar rats. EPSPs in the prefrontal cortex were augmented when Org 26576 was added to risperidone treatment, but not the other antipsychotics.
	99	ADHD	Org 26576 (100–300 mg)	Adult ADHD Investigator Symptom Rating Scale	N/A	Clinical – Phase 1	No statistically significant treatment effects were found.
	98	Depression	Org 26576 (100 mg BID, 400 mg BID, 600 mg BID, placebo)	MADRS, cognitive tests, perception of emotion, hGh, cortisol	Hippocampus	Clinical – Phase 2	Symptomatic improvement was greater in the treatment group than in the placebo group. There was also an association with growth hormone increases and cortisol decreases at the end of treatment. Improvements in executive functioning and speed of processing were seen in the group treated with 400 mg BID.
S47445(CX1632, Tulrampator)	103	Depression and anxiety	Four doses, S47445 (0.3–10 mg/kg) vs fluoxetine (18 mg/kg)	EPM, OFT, Splash test, FST, TST, fur coat test, NSF	N/A	Preclinical	Chronic administration of S47445 had antidepressant- or anxiolytic-like effects in an anxio-depressive phenotypic mouse model (C57BL6/NTac mice) and in a Wistar rat model of anhedonia after chronic mild stress.
		Depression and anxiety	S47445 (1–10 mg/kg)	OFT, BDNF, mTOR, phosphor-mTOR, 4EBP1, phosphor-4EBP1	Prefrontal cortex and hippocampus	Preclinical	S47445 had antidepressant-like and anxiolytic-like effects in bulbectomized C57BL/6J mice. Related measures of BDNF and the mTOR pathway were rescued in the hippocampus, but not in the cortex.
	107	Stress-related disorders	S47445 (1–10 mg/kg)	EPM, LDB, splash test, depolarization-evoked glutamate release, synaptic vesicle-associated proteins, mGluR5s, glucocorticoid receptors, oxytocin receptors	Hippocampus	Preclinical	S47445 corrected all behavioral and biological measures after chronic perinatal stress to Sprague-Dawley rats, including the balance between excitatory and inhibitory neurotransmission that predicts stress-based behavioral alterations.
	105	Age-related decline	S47445 (10 mg/kg)	LTP, VGlut1, spinophilin	Hippocampus, cortex	Preclinical	S47445 significantly counteracted deficits in LTP that resulted from old age and rescued post-synaptic dendritic spines in OFA rats and C57Bl/6 mice, as well as in primary cortical cell cultures.
	102	Age-related decline	S47445 (1–10 mg/kg)	mRNA and protein levels of BDNF, NT-3, NGF	Prefrontal cortex and hippocampus	Preclinical	Two weeks of chronic treatment with S47445 corrected the altered expression of all neurotrophins measured in aged Wistar Han rats.
	106	Age-related decline	S47445 (1–10 mg/kg) vs. S 38093 (0.3–3 mg/kg)	NOR	N/A	Preclinical	Both S47445 and S38093 effectively enhanced the performance of Swiss mice on the NOR task.
	108	Alzheimer’s disease	S47445 (5, 10, 50 mg)	Mini-Mental State Examination, Cornell Scale for Depression in Dementia, Alzheimer’s Disease Assessment Scale	N/A	Clinical – Phase 1	S47445 was safe and well-tolerated but did not improve any measures of cognition or mood over placebo.
	105	Cognitive enhancer	S47445 (0.03–100 μm)	Novel object recognition test, spontaneous alteration in T maze, spontaneous locomotor, functional observation	Frontal cortex, hippocampus	Clinical – Phase 2	S47445 enhanced cognition and had potential neurotrophic and neuroprotective effects.
S18986	117	Neonatal excitotoxic and inflammatory damage	S18986 (10 mg/kg)	Cell death (Fluorojade-B), anti-apoptosis inducible factor, anti-cleaved caspase-3, activated microglia, BDNF mRNA expression	Whole brain	Preclinical	S18986 caused dose-dependent and long-lasting protection of both white matter and cortical grey matter against excitotoxic and inflammatory insult in Swiss mice.
	119	Cognitive enhancement	S18986 (0.3–100 mg/kg acute, or 0.3–30 mg/kg chronically)	One-trial object recognition test	N/A	Preclinical	Both acute and subchronic oral pretreatment in Wistar rats increased recognition of the familiar object, facilitating a form of episodic memory.
	115	Cognitive enhancement	S18986 (0.03–1 mg/kg) vs. donepezil (0.03 mg/kg)	Contextual Serial Discrimination task	N/A	Preclinical	Both donepezil and S18986 reversed memory deficits in older C57Bl/6 mice by increasing contextually correct responses.
	120	Cognitive enhancement	S18986 (0.03–0.3 mg/kg) vs. memantine (0.1–10 mg/kg)	Sequential Alternation task	N/A	Preclinical	Both memantine and S18986 reversed memory deficits in aged C57Bl/6 mice but did not modify results in younger mice.
	121	Cognitive enhancement	S18986 (0.1–1 mg/kg)	Radial Arm Maze with declarative and working memory paradigms	N/A	Preclinical	At 0.1 mg/kg, S189886 selectively improved performance in both declarative and working memory tasks in older C57Bl/6 mice.
	118	Cognitive enhancement	S18986 (0.1–1 mg/kg for 4 months)	Operant-delayed alternation task, Reinforcer Devaluation task, oxidative stress (4-hydroxy-nonenal and malondialdehyde)	Pre-limbic cortex, hippocampus	Preclinical	S18986 dose-dependently improved results in the reinforcer devaluation task, although there was no effect in the operant-delayed alternation task. In aged Sprague-Dawley rats, levels of oxidative stress were increased, but this was reversed by S18986.
	122	Cognitive enhancement	S18986 (0.3–1 mg/kg for 3 days)	NOR task	N/A	Preclinical	All dosages of S18986 significantly improved object recognition in Sprague-Dawley rats.
	116	Age-related decline	S18986 (0.1–1 mg/kg for 16 weeks)	Locomotor activity, spatial memory, forebrain cholinergic neurons, midbrain dopaminergic neurons, microglia in hippocampus	Whole brain	Preclinical	S18986 attenuated all measures of age-related decline in Sprague-Dawley rats.
	123	BDNF levels	S18986 (300 μm) with and without [S]-AMPA	BDNF mRNA and protein expression	Primary cortical neurons	Preclinical	S18986 failed to rescue levels of BDNF when applied alone but maximally enhanced BDNF mRNA levels when applied in conjunction with [S]-AMPA.
OTHER AMPAR THROUGHPUT POSITIVE ALLOSTERIC MODULATORS
LY392098	126	Depression	LY392098 (0.1–10 mg/kg)	FST, TST	N/A	Preclinical	LY392098 reduced immobility in the FST and TST, with a minimum effective dose of 0.5 mg/kg in both Sprague-Dawley rats and NIH-Swiss mice.
	124	Depression	LY392098 (5 mg/kg) vs fluoxetine (20 mg/kg)	FST, TST, MB, SPT	N/A	Preclinical	LY392098 functioned like a traditional antidepressant and reduced immobility in the TST conducted with BALB/c mice, but it did not rescue marble burying or sucrose preference behavior.
	127	Neuro-psychiatric disorders	LY392098 (1, 5, and 10 mM)	BDNF protein and mRNA expression	Primary cortical neuronal cultures	Preclinical	LY392098 had a time- and concentration-dependent effect, significantly increasing levels of both BDNF protein and mRNA.
	125	TARP γ-8 null mice	LY392098	FST	N/A	Preclinical	LY392098 had an antidepressant-like effect on the FST in wild-type mice, but this effect was abolished in the knockout mice.
	128	Synaptic plasticity	LY392098 (0.001–10 mg/kg)	Extracellular electrophysiological recordings	Prefrontal cortex	Preclinical	LY392098 increased the probability of action potential discharge in Sprague-Dawley rats, suggesting that it is a powerful modulator of AMPARs and primarily impacts the prefrontal cortex.
LY404187 and LY451646	140	Healthy controls	LY451646 (0.025–0.5 mg/kg for 21 days, or acutely)	BrdU labeling	Hippocampus	Preclinical	Both chronic and acute doses of LY451646 increased progenitor cell proliferation in Sprague-Dawley rats; larger increases were seen with chronic doses.
	131	Healthy controls	LY404187 (0.1–1 mg/kg for 7 days or acutely) vs. LY451646 (0.1–0.5 mg/kg for 7 days or acutely)	BDNF protein and mRNA expression, TrkB phosphorylation	Hippocampus	Preclinical	In Sprague Dawley rats, acute-dose LY451646 increased BDNF mRNA, but chronic low-dose treatment decreased BDNF and TrkB mRNA, suggesting that AMPAR potentiators may modulate BDNF and TrkB in a dose- and time-dependent manner.
	137	Cognitive enhancement	LY451646 (0.1–1 mg/kg) administered with ketamine	Spatial delayed response task	N/A	Preclinical	After ketamine-induced impairments in working memory, LY451646 rescued all measures in rhesus monkeys.
	129	Parkinson’s disease	LY404187 (0.5 mg/kg/day, 28 days)	Rotometer score, dopamine and metabolite levels, tyrosine hydroxylase, GAP-43 levels	Striatum, substantia nigra	Preclinical	LY404187 provided functional and neurochemical protection from lesions to the striatum and substantia nigra (in Sprague-Dawley rats and C57B16J mice), even when treatment was delayed until after cell death occurred.
	136	BDNF+/− null mice	LY451646 (5 mg/kg) vs. ketamine (50 mg/kg)	FST, OFT, BDNF levels, TrkB phosphorylation	Hippocampus	Preclinical	Both ketamine and LY451646 produced rapid antidepressant effects, but neither significantly impacted BDNF levels or TrkB phosphorylation.
	133	Synaptic plasticity	LY404187 (0.1–10mM)	Neurite length, neurofilament, BDNF, and TrkB expression	SH-SY5Y cell line	Preclinical	LY404187 increased neurite length, neurofilament expression, and TrkB expression through a BDNF-mediated mechanism.
	135	Depression and anxiety	LY451646 (0–3 mg/kg) vs. + citalopram (0–10 mg/kg)	FST, EZM	N/A	Preclinical	LY451646 had antidepressant effects in the FST and enhanced the effects of citalopram in NMRI mice. LY451646 blocked the anxiolytic effects of citalopram in the other behavioral tasks but had no anxiogenic effects alone.
	140	Depression	LY451646 (0.25–2.5 mg/kg)	FST, TST	N/A	Preclinical	There was a dose-dependent decrease in immobility in the FST and TST following LY451646 administration to Sprague-Dawley rats.
	134	Depression and anxiety	LY451646 (3 mg/kg)	FST, OFT, EZM, MB	N/A	Preclinical	LY451646 rescued all measures of antidepressant efficacy but had no anxiolytic effects in NMRI mice.
	138	Depression	LY451646 (0.125 mg/kg)	Novelty-induced feeding suppression, Y-maze, GluA1 and GluA2, corticosterone	Hippocampus	Preclinical	LY451646 normalized anxiety and corticosterone levels in CD1 mice after exposure to chronic social stress, in contrast to previous studies that found no impact on anxiety-like behaviors.
	139	Ethanol intoxication	LY451646 (2 ml/kg)	Motor coordination, operant task disruptions, glucose utilization, BOLD signal	N/A	Preclinical	LY451646 attenuated all effects of ethanol-induced intoxication in Sprague-Dawley rats.
LY503430	141	Parkinson’s disease	LY503430 (0.1–0.5 mg/kg)	Rotameters, dopamine, tyrosine hydroxylase, GAP-43, BDNF	Substantia nigra, whole brain	Preclinical	LY503430 significantly protected all neurochemistry and behavioral measures when administered before substantia nigra lesion (in Sprague-Dawley rats) or MPTP administration (in C57Bl6J mice). It also rescued these measures when administered after injury.
	142	Parkinson’s disease	LY503430 (1 mg/kg)	Rotameters, BDNF, GAP-43, general histology	Substantia nigra and striatum	Preclinical	LY503430 improved all behavioral and biological measures without impacting spontaneous movement in both Sprague-Dawley rats and C57B16J mice.
LY451395 (mibampator)	144	Agitation or aggression in Alzheimer’s disease	LY451395 (3 mg)	NPI-D, CMAI-C, CSDD, FrSBe, CGI-S-AA	N/A	Clinical	Mibampator did not improve scale ratings compared to placebo, except on the FrSBe.
	143	Alzheimer’s disease	LY451395 (0.2 mg for 28 days, 1.0 mg thereafter)	ADAS-Cog	N/A	Clinical	No changes from baseline on the ADAS-Cog after treatment.
BIIB-104 (PF-04958242)	149	Proof of concept	PF-04958242 (up to 32 μm)	Functional potency, functional efficacy, secretory apparent permeability	Mouse embryonic stem (mES) cell-derived neurons	Preclinical	BIIB-104 displayed low oral pharmacokinetic variability as well as a promising therapeutic index that suggests a low chance of adverse events in humans at the doses measured.
	NCT01749098	Cognitive enhancement after ketamine-induced deficits	PF-04958242 (0.25–0.35 mg daily for 5 days)	Hopkins Verbal Learning Test, CogState battery, PANSS, Weschler Digit Span Test	N/A	Clinical – Phase 1	Results not yet posted.
	NCT01518894	Schizophrenia	PF-04958242 (0.1–0.2 mg for 14 days)	Composite of pharmacokinetics, CogState battery, Matrics Consensus Cognition Battery, Drug Effect Questionnaire, C-SSRS	N/A	Clinical – Phase 1	Results not yet posted.
	NCT02332798	Schizophrenia	PF-04958242 (0.2–0.35 mg for 14 days)	Observed plasma concentration, Cmax, Tmax, AUCT, oral clearance, terminal half-life, accumulation ratio, abnormal clinical laboratory measures, ECG, vital signs, neurological examination, C-SSRS	N/A	Clinical – Phase 1	PF-04958242 was well-tolerated, even at the higher doses, in stable schizophrenia patients.
	NCT01511510, NCT01159483, NCT02228395, NCT01238679, NCT04079101	Healthy volunteers	PF-04958242 (0.01–0.8 mg once or daily)	Composite of pharmacokinetics, Drug Effect Questionnaire, Digit Symbol Substitution Test, C-SSRS, Cmax, Tmax, AUCT, oral clearance, terminal half-life, adverse events, abnormal clinical laboratory measures, ECG, vital signs, neurological examination, renal clearance, urine elimination	N/A	Clinical - Phase 1	PF-04958242 was well-tolerated in healthy volunteers, with only minor adverse events such as nausea, muscular weakness, and fatigue.
	NCT01365338	Working memory	PF-04958242 (0.075–0.15 mg single dose)	fMRI, arterial spin labeling data	N/A	Clinical – Phase 1	No results posted.
	148	Cognitive enhancement	BIIB-104 (0.35 mg on day 1, 0.25 mg for days 2 – 5)	Hopkins Verbal Learning Test, CogState battery, PANSS, CADSS	N/A	Clinical – Phase 1	BIIB-104 significantly reduced ketamine-induced impairments on memory tasks without worsening psychotomimetic symptomology.
	NCT02855411	Schizophrenia	PF-04958242 (0.15–0.5 mg daily for 12 weeks)	Matrics Consensus Cognitive Battery, adverse events, Schizophrenia Cognition Rating Scale, PANSS, CGI, Ataxia, C-SSRS	N/A	Clinical – Phase 2	Trial was terminated (not due to safety concerns).
	NCT03745820	Schizophrenia	BIIB-104 (0.15–0.5 mg daily for 12 weeks)	Matrics Consensus Cognitive Battery, adverse events, performance-based skills assessment, Schizophrenia Cognition Rating Scale, PANSS, CGI, ataxia, C-SSRS	N/A	Clinical - Phase 2	Recruitment stage.
TAK-137	154	Neurological disorders	TAK-137 (0.03–1 mg/kg) vs. LY451646 (0.03–1 mg/kg)	Ca2+ influx, whole-cell patch clamp recordings, BDNF production, BrdU staining, novel object recognition test, DMS	Primary neurons, hippocampus	Preclinical	Compared to LY451646, TAK-137 induced higher BDNF production, significantly improved performance on cognitive tasks in Sprague-Dawley rats, and lowered risks of seizures.
	150	Depression	TAK-137 (0.03 and 0.3 mg/kg for 3 days)	mTOR and p70S6K phosphorylation, BDNF production, NSF test, locomotor activity, PPI of acoustic startle	Primary cortical neurons	Preclinical	TAK-137 increased phosphorylation along the mTOR pathway. It also shortened feeding latency but not prepulse inhibition in Sprague-Dawley and Wistar-Kyoto rats.
	152	Schizophrenia	TAK-137 (3–10 mg/kg)	Locomotion, social interaction, 5-CSRTT, radial arm maze, DMS tasks, Reversal Learning Test	N/A	Preclinical	TAK-137 significantly enhanced performance on all measures of social interaction, attention, working memory, and cognitive flexibility in various strains of rats.
HBT1	154	Neuro-psychiatric and neurological disorders	HBT1	BDNF production	Primary cortical neurons	Preclinical	HBT1 caused a bell-shaped response curve for BDNF production in primary neurons with no agonistic properties.

Abbreviations: 4EBP1, Eukaryotic translation initiation factor 4E-binding protein 1; 5CSRTT, five-choice serial reaction time task; ADAS-Cog, Alzheimer’s Disease Assessment Scale – Cognitive Subscale; ADHD, attention deficit hyperactivity disorder; AMPA, α–amino-3-hydroxy-5-methyl-4-isoxazole propionic acid; AMPAR, AMPA receptor; ASD, autism spectrum disorder; AUC_T, area under the curve across time; BDNF, brain-derived neurotrophic factor; BID, twice per day; BOLD, blood-oxygen level dependent; BrdU, bromodeoxyuridine; CADSS, Clinician-Administered Dissociative States Scale; CGI, Clinical Global Impressions scale; CGI-S-AA, Clinical Global Impressions – Severity Agitation/Aggression scale; CMAI-C, Cohen-Mansfield Agitation Inventory – Community Version; C_max, maximum plasma concentration; CSDD, Cornell Scale for Depression in Dementia; C-SSRS, Columbia-Suicide Severity Rating Scale; DMS, delayed match-to-sample task; DNMS paradigm, delayed nonmatch-to-sample paradigm; ECG, electrocardiogram; EEG, electroencephalography; [¹⁸F] FDG PET, 2-deoxy-2-[fluorine-18]fluoro-D-glucose positron emission tomography; EPM, elevated plus maze; EPSC, excitatory postsynaptic current; EPSP, excitatory postsynaptic potential; EZM, elevated zero max; fMRI, functional magnetic resonance imaging; FrSBe, Frontal Systems Behavior Scale; FST, forced swim test; GAP-43, growth-associated protein 43; GluA1, glutamate receptor A1; HAM-D, Hamilton Depression Rating Scale; hGh, human growth hormone; LDB, light/dark box; LI, latent inhibition; LTP, long-term potentiation; MADRS, Montgomery-Asberg Depression Rating Scale; MB, marble burying; mGluR5, metabotropic glutamate receptor 5; MPP, 1-methyl-4-phenylpyridinium; MPTP, 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine; mTOR, mammalian target of rapamycin; MWM, Morris water maze; NBQX, 2,3-dioxo-6-nitro-7-sulfamoyl-benzo[f]quinoxaline; NeuN, Hexaribonucleotide Binding Protein-3; NGF, nerve growth factor; NMDA, N-methyl-D-aspartate; NOR task, novel object recognition task; NPI-D, Neuropsychiatric Inventory – Distress scale; NSF, novelty suppressed feeding; NT-3, neurotrophin-3; OFA, Oncins France Strain A; OFT, open field test; PANSS, Positive and Negative Syndrome Scale; PCP, phencyclidine; PPI, prepulse inhibition; p-70S6K, Ribosomal protein S6 kinase beta-1; SDTT, social dominance tube test; SPT, sucrose preference test; TID, three times per day; TARP γ − 8, transmembrane AMPAR regulatory protein γ − 8; T_max, time to arrive at maximum plasma concentration; TrkB, Tropomyosin receptor kinase B; TST, tail suspension test; TUNEL, Terminal deoxynucleotidyl transferase dUTP nick end labeling; vGlut1, Vesicular glutamate transporter 1.

FIGURE 1.

Molecular structure of AMPAkines and other α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor (AMPAR) positive allosteric modulators (PAMs).

Nootropic racetams

Nootropic racetams (piracetam, oxiracetam, pramiracetam, aniracetam, phenylpiracetam) are AMPAR potentiators first developed in the late 1960s to limit cognitive impairment and memory decline associated with normal aging and neurodegenerative diseases.⁶⁰ Racetams increase glutamatergic transmission by increasing the maximal density of AMPA-binding sites in the synaptic membrane, thereby enhancing the overall efficacy of AMPA-stimulated Ca²⁺ influx.⁶¹ Despite mixed clinical evidence, both piracetam and pramiracetam were approved in Europe and the United States as cognitive enhancers, and aniracetam was approved in Europe. Unfortunately, racetams are limited by their low potency and/or poor brain bioavailability and also induce rare, mild, and transitory side effects.^60,62

Ampakines

AMPAkines, also known as ‘CX compounds’, are a series of AMPAR PAMs developed in the late 1990s to early 2000s. They are typically benzamides and have a closely related chemical structure in common.⁶³ Originally designed to enlarge the therapeutic armamentarium, AMPAkines are low-potency drugs selected to avoid the risk of AMPA agonism-dependent serious side effects such as seizures. AMPAkines have been investigated for the treatment of neuropsychiatric, neurological, and neurodegenerative disorders⁶⁴ and, most recently, have been under clinical development for drug-induced respiratory depression and sleep apnea.

AMPAkines exert their effects by stabilizing AMPARs in a channel-open state after glutamatergic release, which increases the amplitude and duration of glutamate-induced AMPA currents.⁶³ Two distinct subfamilies of AMPAkine compounds exist. Type I compounds (such as CX546) prolong synaptic response, whereas type II compounds (such as CX516 and CX614) increase amplitude.⁶⁴ By shifting current dynamics, both subgroups of AMPAkines lower the induction threshold and increase the magnitude of long-term potentiation,⁶⁵ thereby improving memory and learning. Although other AMPAkines exist, they have not yet been classified. Relevant AMPAkines with potentially promising neurotrophic properties are described in-depth below, in approximate order of development. It should be noted that, to date, most of these compounds have been investigated for neuropsychiatric indications other than depression.

CX516

CX516 was one of the first AMPAkines to be developed, and has been clinically tested as a cognitive enhancer in mild cognitive impairment (MCI) (NCT00040443), Alzheimer’s Disease (NCT00001662), schizophrenia,^66,67 and Fragile X syndrome.⁶⁸ In clinical trials, however, the results were largely negative because of CX516’s low potency, short half-life, and extensive metabolism. Nevertheless, promising results have recently been obtained in animal models of intellectual disability,⁶⁹ attention deficit hyperactivity disorder (ADHD),⁷⁰ and other behavioral measures.^71,72 Evidence also suggests that CX516 increases nerve growth factor (NGF) and BDNF in areas linked to neurodegenerative disease.⁷³

CX546

The pharmacological properties of CX546 have specifically been explored in animal models of schizophrenia, where it showed promising results.^74,75 More recently, CX546 was also explored in animal models related to autism spectrum disorder, where mixed results have been observed.^76,77 Biologically, CX546 appears to be an effective mediator of neurogenesis, dendritogenesis, and increased structural plasticity.^78,79 However, its low oral bioavailability limits the clinical development of this agent.

CX614

CX614 is a prototypical benzamide compound classified as an allosteric modulator. It was initially investigated as a putative new treatment for neurodegenerative disorders, including Alzheimer’s and Parkinson’s diseases.⁸⁰ In preclinical comparisons, CX614 enhanced structural plasticity and dendritic outgrowth at the same rate as ketamine,⁸¹ and also enhanced the effects of the conventional antidepressants imipramine and reboxetine.⁸² Other preclinical evidence demonstrated its ability to rapidly upregulate BDNF protein levels^83,84 and behavioral antidepressant-like effects^82,85 as well as influence the rate, frequency, and power of gamma effects in an age–drug interaction associated with spatial memory.⁸⁶ In preclinical models of Huntington’s disease, CX614 prevented AMPAR desensitization, slowed deactivation, and facilitated glutamate release, significantly increasing synaptic activity by augmenting the frequency and amplitude of spontaneous and miniature excitatory postsynaptic currents (mEPSCs) in transgenic mouse models of Huntington’s chorea compared to control mice.⁸⁷ Despite some initial promise in preclinical models, development of these compounds appears to have been halted.

CX691

CX691 has been studied for its pro-cognitive effects in animal models of Alzheimer’s disease, where it was found to increase BDNF protein expression in the hippocampus and improve spatial learning and memory.^88,89 These data are supported by research showing upregulation of BDNF and NGF mRNA by CX691 in the dentate gyrus, amygdala, and piriform cortex.⁷³ CX691 was also clinically tested in a pilot study of healthy elderly volunteers, where it was found to significantly ameliorate short-term memory.⁹⁰

CX717

The ampakine CX717 has positive allosteric modulatory properties at the AMPAR. It is believed to facilitate transmission at cortical synapses that use glutamate as a neurotransmitter. In preclinical models, CX717 enhanced cognitive properties and demonstrated antidepressant-like effects.^91–94 Although initial investigations of its ability to improve cognition in sleep-deprived adults were not promising,⁹⁵ this agent entered clinical development for the treatment of ADHD (NCT03375021); no results have yet been listed. It was also investigated for use in Alzheimer’s disease, but the trial was halted.

ORG 26576

Promising preclinical studies demonstrated that ORG-26576, which was developed as a potential treatment for depression and ADHD, increased hippocampal neurogenesis and BDNF synthesis and improved spatial memory.^96,97 The compound did not, however, meet trial endpoints in subsequent Phase 2 clinical trials for MDD,⁹⁸ ADHD,⁹⁹ or schizophrenia,¹⁰⁰ and its development has been discontinued.

S47445 (CX1632)

Chronic treatment with the recently developed AMPAkine S47445 (CX1632, tulrampator) was found to rescue deficits in synaptic plasticity and cytoarchitecture in the hippocampus of aged mice¹⁰¹ and to increase levels of neurotrophins in both the hippocampus and the prefrontal cortex.¹⁰² Antidepressant and anxiolytic-like effects were also reported in three animal models of depression in young-adult animals: mice exposed to chronic corticosterone treatment, rats subjected to chronic mild stress,¹⁰³ and bulbectomized mice.¹⁰⁴ The behavioral effects were again accompanied by increased levels of hippocampal neurogenesis and increased BDNF levels.^103,104 S47445 also had pro-cognitive effects in animal models.^105–107

Building on this work, S47445 is being investigated as a potential treatment for MDD, Alzheimer’s disease, and MCI (NCT02626572, NCT02805439). However, the results of a first double-blind, placebo-controlled clinical trial of chronic S47445 treatment for individuals with mild to moderate Alzheimer’s disease and comorbid depression found that S47445, although well tolerated, did not significantly improve cognitive or depressive symptoms.¹⁰⁸

Other AMPAR-throughput PAMs

Since the early 2000s, Eli Lilly has developed several AMPAR PAMs, which have been investigated specifically for their putative antidepressant properties and possible applications in schizophrenia, Parkinson’s disease, and ADHD.¹⁰⁹ PAMs of the biarylpropylsulfonamide class (LY392098, LY404187, LY451395, LY503430, and BIIB-104) are thought to enhance AMPA neurotransmission by reducing desensitization of the ion channel.^110–113 The agents are listed below in approximate order of development.

S18986

Servier developed S18986 as a putative cognitive enhancer for normal aging. This agent increases excitatory activity by enhancing the amplitude of extracellular excitatory field potentials.¹¹⁴ A number of positive preclinical studies found that S18986 improved performance in memory tasks and increased BDNF protein levels in the hippocampus of aged rodents,^115–122 although it did not increase BDNF mRNA levels.¹²³ Nevertheless, a recent clinical trial found that S18986 exerted no clinically relevant effects in controlling cognitive symptoms in individuals with MCI (NCT00202540).

LY392098

In preclinical studies, LY392098 demonstrated antidepressant effects such as reducing weight loss and immobility time in the tail suspension test in mice,¹²⁴ an effect that was blocked by knocking out the transmembrane AMPAR regulatory protein (TARP) γ − 8.¹²⁵ In another preclinical study, LY392098 reduced immobility in both the forced swim and tail suspension tests in a dose-dependent manner.¹²⁶ It was also found to upregulate BDNF levels in cortical neuronal cultures.¹²⁷ Using electrophysiological approaches, Baumbarger and colleagues¹²⁸ found that the effects of LY392098 were primarily located in prefrontal cortical neurons. To date, however, no clinical trials have been conducted with LY392098.

LY404187 and LY451646

LY404187 has been explored preclinically for its efficacy in improving cognition and in animal models of depression, schizophrenia, Parkinson’s disease, and ADHD.¹²⁹ LY404187 is a stable racemic mixture of R and S stereoisomers (LY451646 was identified as the active isomer) that activates all AMPARs, but it appears to be more potent at GluA2- and GluA4-subunit-containing receptors and has preference for the flip splice variant.^110,130 Chronic treatment with both LY404187 and LY451646 increased BDNF mRNA and protein expression in the hippocampus of rats^131,132 and also increased neurite length and neurofilament expression.¹³³ In animal models, LY451646 displayed antidepressant-like properties in the forced swim and tail suspension tests,^134,136 significantly improved measures of working memory¹³⁷ and chronic social stress,¹³⁸ and prevented the effects of ethanol-induced intoxication on cognition.¹³⁹ LY451646 also increased cell proliferation in the dentate gyrus, suggesting an ability to positively modulate hippocampal neurogenesis.¹⁴⁰ To date, however, no clinical trials have been conducted with these compounds.

LY503430

LY503430 has been evaluated for its potential neuroprotective effects in animal models of Parkinson’s disease, where it significantly improved biological measures such as BDNF levels and decreased indicators of neurotoxicity.¹⁴¹ Other studies found that LY503430 protected against neurotoxicity induced by 6-hydroxydopamine or 1-methyl-4-phenyl-1,2,3,6-tetrahydropyri dine (MPTP) infusion into the substantia nigra or striatum of rats.¹⁴² However, to date, no clinical trials have been conducted with this compound.

LY451395

LY451395 (mibampator) has been clinically evaluated for Alzheimer’s disease. However, clinical trials found that LY451395 did not affect dementia symptoms in participants with mild to moderate Alzheimer’s Disease,¹⁴³ limit agitation and aggression,¹⁴⁴ or reduce apathy.¹⁴⁵

BIIB-104

Pfizer recently developed the nootropic agent BIIB-104 (PF-04958242).¹¹³ Preclinical evidence found that that BIIB-104 improved synaptic connections, working memory, and cognitive performance in animal models of schizophrenia.^{113,146–149} This agent, which appears to be well-tolerated, is now in Phase 2 clinical trials for cognitive impairment in individuals with schizophrenia (NCT03745820, NCT2855411).

TAK-137

TAK-137, developed by Takeda, appears to affect glutamatergic transmission by increasing the potency and binding affinity of AMPA for AMPARs.^150,151 It is now under preclinical evaluation for the treatment of cognitive impairment in schizophrenia and as a potential new antidepressant. In several different animal models of schizophrenia, TAK-137 showed only limited effective-ness in controlling positive symptoms but significantly improved social interaction, working memory, and other cognitive functions.¹⁵² Another preclinical study recently compared the antidepressant properties of TAK-137 to those of ketamine and found that, in rats, three days of treatment with TAK-137 produced similar behavioral effects as acute ketamine without inducing any psychotomimetic side effects.¹⁵³ Finally, in preliminary Phase 1 studies with both healthy volunteers and ADHD participants, TAK-137 was found to be safe and well-tolerated and to have a wide therapeutic window (NCT02334982, NCT02163915).

HBT1 and OXP1

HBT1 and OXP1 were identified as promising candidates in a recent high-throughput screening study conducted to discover new AMPAR potentiators specifically selected for their low agonistic effect in Ca²⁺ influx assays in primary neurons and low risk of bell-shaped response and seizure.¹⁵⁴ HBT1 exhibited neurotrophic properties and increased BDNF protein levels in primary neurons.¹⁵⁴ Given that this is preliminary research, the exact mechanism of action of these agents remains unknown.

Indirect AMPAR modulators

Other glutamatergic agents have also been investigated for their ability to indirectly modulate AMPAR activity and function. Indeed, recent evidence suggests that the rapid antidepressant effects associated with a variety of glutamatergic drugs under investigation both involves and requires AMPAR modulation, as demonstrated previously for ketamine and (2R,6R)-HNK. This area deserves attention when considering the impact of AMPARs in neuropsychiatric disorders, although further studies are needed to provide a more thorough understanding of indirect AMPAR modulation. Rather than provide a comprehensive review of these agents, we discuss a few promising indirect AMPAR modulators currently under development for the pharmacological treatment of neuropsychiatric disorders.

Metabotropic glutamate receptor (mGluR)2/3 antagonists

mGluRs are G-protein-coupled receptors that mediate secondary messenger pathways. They are traditionally subclassified into three groups based on sequence homology, G-protein coupling, and ligand selectivity. Given their ability to modulate excitatory transmission driven by ionotropic glutamate receptors and to fine-tune cellular response to glutamate signaling in the brain, the selective pharmacological modulation of mGluRs has been investigated as a novel strategy for developing glutamatergic-based antidepressants.

Preclinical evidence suggests that group II (mGluR2 and mGluR3) antagonists are presently the most promising of these compounds for the development of new, rapid-acting antidepressants.^155–157 Preclinical studies found that antagonizing mGluR2/3s elicits an antidepressant-like effect that is blocked by the administration of an AMPAR antagonist.^157,158 Other preclinical studies found that the antidepressant effects of both racemic ketamine and (2R,6R)-HNK were entirely abolished after administering an mGluR2/3 agonist.^159,160 These promising preclinical results have spurred interest in some of these agents as drug candidates.^{157,158,161,162} One such agent is TS-161, a prodrug of TP0178894 and an orthosteric mGluR2/3 antagonist. A Phase 1, randomized, placebo-controlled, double-blind study evaluated the safety profile, tolerability, and pharmacokinetics of TS-161 in 70 healthy volunteers (NCT03919409), but results have not yet been posted. Phase 2 proof-of-concept trials of this agent in individuals with treatment-resistant depression are anticipated in 2022.

Reelin

Reelin is an extracellular matrix protein involved in regulating neural migration during brain development and in synaptic plasticity in the adult brain¹⁶³ that appears to parallel some of ketamine’s molecular mechanisms. Reelin levels were found to be downregulated in the CA4 hippocampal region in postmortem brain samples from individuals with MDD.¹⁶⁴

Preclinical studies found that chronic stress both causes depressive-like behaviors and decreases the number of Reelin + cells in the dentate gyrus subgranular zone (SGZ).¹⁶⁵ These deficits have been speculated to underlie the alterations that chronic stress exerts on depressive-like behaviors, hippocampal neurogenesis, and synaptic plasticity.¹⁶⁶ Preclinical studies also found that, as with ketamine, a single intrahippocampal infusion of Reelin rescued both the depressive-like behaviors and the neurochemical alterations associated with animal models of depression within 24 hours.¹⁶⁷ Furthermore, these rapid antidepressant-like effects were blocked by administration of the AMPAR antagonist CNQX, suggesting that AMPAR currents are required for Reelin’s pharmacological effects.¹⁶⁷ Reelin has also been shown to enhance AMPAR neurotransmission by increasing the number of AMPARs at postsynaptic sites.¹⁶⁸ Chronic stress has also been found to result in deficits in hippocampal AMPAR expression and in altered synaptic mTORC levels, both of which were rescued by Reelin in a manner similar to ketamine.^167,169 Finally, Reelin also appears to mediate some of ketamine’s rapid-acting antidepressant effects.¹⁷⁰ Taken together, these data appear to indicate that the actions of Reelin on AMPARs may be one important molecular mechanism through which Reelin exerts rapid-acting antidepressant-like effects, underscoring the possibility that AMPAR modulators may be a novel avenue for developing new, rapid-acting antidepressants.¹⁶⁹ Although Reelin has shown promise in preclinical studies, clinical research has yet to be conducted on this protein.

AMPA modulators in the treatment of neuropsychiatric disorders: Unsolved challenges

Despite the encouraging preclinical results described above, most AMPAR modulators developed to treat neurological and neuropsychiatric disorders have, to date, failed to demonstrate any relevant clinical effects. This may be due to several factors, including the lack of a proper in vitro or in vivo model, lack of target engagement, and the fact that early positive results are often overshadowed by larger failures in the clinical setting, which prevents agents from entering Phase 3 trials.

CX516 provides a good example of this process. The agent had promising preclinical results and positive preliminary findings in elderly participants with mild cognitive impairment (NCT00040443). When investigated for its ability to improve cognition in schizophrenia, however, the interaction of CX516 with common antipsychotics negated any initial positive impact.⁶⁶ This challenge of translating basic science and preclinical findings into clinical results has been well-documented. Models to measure clinical efficacy in neurological and neuropsychiatric disorders often fall short due to their inability to capture the complexity and individuality present in human patients. Thus, optimizing the design and use of in vitro and in vivo models to analyze disorders is critical to future drug development efforts. In the context of AMPA modulators, current in vitro models have many challenges, given that the heterogeneity of AMPARs and their many splice variants, modulatory proteins, and post-translational modifications create a complex environment that cannot be replicated in cell cultures used for basic research. In vivo research is also associated with unique challenges, such as the variety of behavioral tests that cannot adequately mimic human behavior and cognition. As an example, the forced swim test in rodents, which is commonly used to measure despair in animal models of depression, has recently been criticized for its lack of real-world applicability.¹⁷¹

In addition to broader issues in the field regarding the modeling of neurological and neuropsychiatric disorders, the pharmacodynamics and pharmacokinetics of AMPAR positive modulators have presented specific barriers to success. For instance, CX516 and other AMPAkines have low potency and a short half-life, both of which may have contributed to their clinical failure. It should be noted, however, that more potent AMPAkines have also been unsuccessful in clinical trials,¹⁷² perhaps because of low solubility or poor metabolic stability, given that higher potency often comes at the cost of these two important factors.¹⁷³ Another potential issue is the limited understanding of the specificity of many of these AMPAR modulators, particularly because subunit interactions are extremely important for determining receptor function. Another major technological barrier to discovering effective molecules is the lack of biomarkers showing how these modulators influence AMPARs in vivo. Developing tools to delineate the deactivation or desensitization of receptors post-treatment would be particularly valuable for the future discovery of novel AMPAkines.

Conclusions

Given the enormous potential of novel positive AMPAR modulators, there is considerable reason for optimism regarding their future development as cognitive enhancers and novel therapeutics for a variety of neuropsychiatric disorders, despite the difficulties discussed above. AMPAR PAMs target the synapse in a unique way, improving neuronal communication deficits. In addition, the structural and functional complexity of AMPARs suggests that developing chemically diverse AMPAR PAMs is necessary for determining which agents will be most successful for which disorder. The diverse expression patterns of receptor subunits and isoforms may allow the targeting of specific regions, an opportunity that sets AMPARs apart from other drug targets. As an example, Xia and colleagues¹⁷⁴ showed that certain AMPAkines had a greater effect in the hippocampus than in the thalamus. Other positive modulators have shown specificity in targeting the flip or flop isoform of GluA2 (reviewed in¹⁷⁵). Given the functional complexity of AMPARs, many opportunities exist to develop highly specific therapeutics for multiple disorders where glutamatergic transmission is impacted. As a result, identifying novel chemotypes has now become a priority in drug development, with companies such as Eli Lilly, GlaxoSmithKline, and others developing high-throughput screens to determine high-potency AMPAR modulators that are effective for different conditions. With improved tools and measurements, AMPAR PAMs remain some of the most promising treatment avenues for neurological and neuropsychiatric disorders.

Abbreviations:

ADHD

attention deficit hyperactivity disorder

AMPA

a-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid

AMPAR

AMPA receptor

BDNF

brain-derived neurotrophic factor

GABA

gamma aminobutyric acid

HNK

hydroxynorketamine

MCI

mild cognitive impairment

MDD

major depressive disorder

mTORC

mammalian target of rapamycin complex

NGF

nerve growth factor

NMDAR

noncompetitive allosteric N-methyl-D-aspartate receptor

PAM

positive allosteric modulator

Biographies

Laura Musazzi, PhD is an Associate Professor at the School of Medicine and Surgery, University of Milano Bicocca. Her research interests include behavioral stress at the synaptic level, regulation of synaptic function, neuroplasticity, gene expression with regard to acute or chronic stress, and the pathogenesis of neuropsychiatric and neurodegenerative disorders with a focus on discovering novel mechanisms of action for psychotropic drugs.

Jenessa Johnston is a PhD student at the University of Victoria and a Pre-Doctoral Fellow at the National Institute of Mental Health (NIMH). Her research focuses on the molecular mechanisms underlying novel antidepressants, and on the discovery of diagnostic and therapeutic biomarkers for various neuropsychiatric disorders.

Bashkim Kadriu, MD is a board-certified psychiatrist. He is currently Director and Clinical Leader, Neuroscience Experimental Medicine, Janssen Research and Development, Janssen Pharmaceutical Companies of Johnson & Johnson. He previously worked as a neuroscientist at the Experimental Therapeutics and Pathophysiology Branch (ETPB), NIMH. His research interests include early drug development, as well as the neurobiological correlates of mood disorders, with a particular emphasis on discovering biosignatures that guide novel therapies and their mechanisms of action.