1. Competitive Antagonists of the Kainate Receptors
1.1. Quinoxaline-2,3-diones
Quinoxaline-2,3-dione analogues are one of the most intensively studied classes of competitive AMPAR/KAR antagonists. The early generation of quinoxalinediones, such as CNQX (
1,
Figure 1), DNQX (
2), or NBQX (
3), was reported in the 1980s and 1990s as potent AMPAR/KAR antagonists and neuroprotective agents
[1][2][3][4][5]. Many members of this class also showed effects at the glutamate and, particularly, the glycine (GlyN) binding sites of the NMDA receptor. Among them, NBQX was the first compound to exhibit a high selectivity for AMPAR versus NMDAR and has been used as the referential antagonist of choice in numerous in vitro and in vivo models. The neuroprotective, anticonvulsant, anxiolytic, and antinociceptive effects of NBQX have been widely investigated and reported
[2][6][7][8][9][10][11][12][13][14]. Neuroprotective properties of NBQX in ischemic stroke were confirmed in clinical studies; however, due to the low aqueous solubility and associated nephrotoxicity at therapeutic doses, it was rejected in the second phase of clinical studies
[15].
Figure 1. Structures of selected KAR antagonists based on the quinoxaline-2,3-dione structure.
The subsequent years progressed with numerous studies aimed at improving the affinity and selectivity toward AMPA receptors, the results of which have been reviewed in detail by Nikam
[16], Catarzi
[15], and Larsen
[17]. Among them, particularly notable are YM90K (
4), YM872 (
5), and ZK200775 (
6), active and selective AMPA receptor antagonists, also showing a micromolar binding affinity at the native kainate receptors
[18][19][20][21]. Relevant structural modifications improved the affinity of the compounds of the AMPAR and solved the problem of low water solubility. Both YM872 and ZK200775 have demonstrated neuroprotective activity in animal models of acute and ischemic stroke and have become the subject of clinical trials
[18][19][20][21]. Furthermore, YM872 showed neuroprotective properties in brain hemorrhage, as well as therapeutic potential in malignant primary glioma
[22][23].
Selectivity toward KA receptors over AMPA receptors has been shown for pyrrolylquinoxaline-2,3-dione analogues, including the highly potent LU97175 (
7,
Figure 1), the compound substituted with a benzhydrazide group at the
N1 position of the quinoxalinedione core
[5]. Interestingly, LU97175 was reported to preferentially bind to the GluK3 homomeric receptors compared not only to the AMPA subtype GluA2, but also to the other individual kainate subtypes GluK1, GluK2, and GluK5
[5][24]. In an in vivo kindling model of temporal lobe epilepsy, LU97175 displayed anticonvulsant activity without inducing motor damage
[4][5][25].
Among the examples of highly selective antagonists for homomeric GluK1 receptors are pyrrolyl quinoxalinediones BSF 91594 (
8), BSF 111886 (
9), and BSF 84077 (
10), possessing an aliphatic amine (e.g., piperazine) linker attached to the pyrrole ring (
Figure 1). Compounds
8 and
9 were found to be potent GluK1 ligands with more than a 100-fold selectivity over other kainate subtypes (
Table 2). All three compounds showed a moderate to weak affinity at the native AMPA receptors and weak or no binding to the NMDA glycine binding site. When tested in vivo,
8,
9, and
10 confirmed their anticonvulsant potency in NMDA-induced seizures with effective median doses of 14, 21, and 3.3 mg/kg, respectively
[26].
The LU97175 structure became the starting point for further modifications of the quinoxalinedione scaffold at positions
N1, 6, and 7. SAR studies for the recently reported series of
N1-substituted analogues confirmed that, also for compounds without a pyrrole ring in position 7 (and possessing a fluorine or nitro group instead), the
N1-benzamide moiety was optimal for binding to the GluK1-3 recombinant receptors (
11–
14) and sufficient to achieve an affinity profile similar to LU97175 (
12), while compounds with smaller or non-aromatic substituents in position
N1 showed a significantly lower KAR affinity
[24]. Attempts to bioisosterically replace the pyrrolyl moiety in the structure of LU97175 resulted in another series of derivatives, among which a 7-imidazole analogue (
15) exhibited the highest binding affinity across the recombinant receptors GluK1-3 and the 73-fold binding selectivity for GluK3 over the AMPA subtype GluA2
[27]. In a mouse tail flick test for acute pain,
15 was shown to have an analgesic effect and was more effective than NBQX, suggesting that, contrary to an earlier hypothesis
[8], the analgesic effects observed for early generation quinoxalinediones could also be mediated by kainate receptors. Furthermore,
15 did not show an adverse effect on motor coordination in the rotarod test as previously observed for NBQX, and it was found to cause an increase in locomotor activity
[27].
The binding mode of
15 in the GluK1 binding pocket was determined by X-ray crystallography, showing a strong analogy to other complexes of quinoxaline-2,3-diones cocrystallized with the LBDs of non-NMDA receptors. Among others,
15 forms hydrogen bonds between the carbonyl groups of the quinoxalinedione core and Arg523 of domain 1, the residue considered crucial for the binding of
l-glutamate, agonists, and competitive antagonists, and conserved within all AMPA and KA subunits
[28][29]. The quinoxaline-2,3-dione core is also involved in H-bond contacts with other residues important for ligand binding: Thr518 and Pro516, while the position of the molecule is stabilized by π−π stacking between the quinoxaline-2,3-dione system and Tyr489
[27]. Despite the fact that
15 does not form any direct hydrogen bond with the residues of domain 2, the ligand, similarly to other quinoxalinediones, induces the opening of the LBD consistent with an antagonist binding mode. Analysis of the
15-GluK1-LBD X-ray structure also revealed the presence of a sulfate ion in the ligand-binding site. Considering that the unsubstituted imidazole nitrogen atom of
15 is probably protonated (the complex was crystallized at pH 4.5), it likely forms a salt bridge with a sulfate ion, which could explain the higher activity of this compound compared to LU97175
[27].
The
N1-benzamide analogue
16 with an unsubstituted 7-position of the quinoxalinedione scaffold and possessing a phenylethynyl moiety at position 6 (
Figure 1), demonstrated an interesting selectivity profile showing approximately a 30-fold preference for the GluK3 homomeric receptors over the subtypes GluK1 and GluK2
[24]. Following
16 as the lead structure, a series of quinoxaline-2,3-dione analogues with an ethynyl substituent at the 6-position has recently been presented
[30]. The introduction to this position of a large bicyclic substituent linked by a triple bond to the quinoxalinedione core (
17,
18) appeared to be an important structural modification that clearly affected the binding to the GluK3 subunit, resulting in the compound
17 with a pronounced preference for GluK3 and submicromolar GluK3 affinity. In fact, a 400-fold selectivity for GluK3 over the subtypes GluK1, GluK2, GluK5, and GluA2, reported for this analogue, has been a unique KA receptor affinity profile among all structures described to date
[30]. On the other hand, the most active compound in this series,
18, showed a weak selectivity profile but a high affinity at the homomeric receptors GluK1, GluK3, and also GluK2, thus presenting one of the highest GluK2-affinity values among the quinoxalinedione-based ligands reported so far.
Demmer et al. presented a different approach to the design of quinoxalinedione structures
[31][32]. The proposed modification focused on the combination of an unsubstituted quinoxaline-2,3-dione core with an amino acid chain (
19)
[31] or an acid moiety, varied in chemical functionalities, carbon chain length, and flexibility (
20)
[32]. The amino acid derivative CNG-10300 (
19) exhibited a weak micromolar affinity for homomeric GluK1-3 receptors with
Ki values of 16, 9.5, and 59 μM, respectively, and was successfully cocrystallized with the GluK1-LBD. Analysis of this X-ray complex indicated that the amino acid moiety of
19 did not interact with the protein in the way observed for typical α-amino acid agonists but was involved in interactions that stabilized an open antagonist state of the binding pocket
[31]. Micromolar binding to GluK1-3 was also observed among the series of compounds containing an acidic group. The most active kainate ligand,
20, with an affinity of 1.2 µM for the GluK1 homomeric receptors showed more than a 27-fold selectivity over the kainate subtypes GluK2 and GluK3, and a 20-fold selectivity over the native AMPA receptors
[32]. In a broader sense, the weak KAR affinity of the described monosubstituted quinoxalinediones confirmed the need for an additional core substitution in
N1 to achieve a high affinity for the kainate receptors.
1.2. α-Amino Acid Antagonists
1.2.1. Willardiines
A separate group of competitive AMPA/KAR antagonists includes compounds with α-amino acidic functionality, most often linked through a heterocyclic ring system with a distal acidic group; for example, a carboxylate or phosphonate, or their isostere, for example, tetrazolyl ring. Undoubtedly, the most important chemical class within this category is willardiine derivatives, developed based on the natural agonist of the AMPA/KA receptor, willardiine ((
S)-1-(2-amino-2-carboxyethyl)pyrimidine-2,4-dione). One of the first attempts to achieve an antagonistic AMPAR/KAR profile among willardiine derivatives involved the introduction of a 4-carboxylbenzyl (UBP282,
21) or 2-carboxybenzyl (UBP296,
22) substituent at position
N3 (
Figure 2)
[33][34]. Subsequent studies have shown that both UBP296 and its purified
S-enantiomer UBP302 (
23) proved to be potent and highly selective GluK1 antagonists, with a negligible binding affinity at the homomeric rat GluK2, GluK5, or GluK2/GluK5 heteromers. UBP296 was further characterized on the native GluK1-containing receptors in the hippocampal mossy fibers and was found to play a role in controlling synaptic transmission in the CA1 and CA3 regions and reversibly block LTP induction in MF synapses
[34].
Figure 2. Structures of selected willardiine-based KAR antagonists. * the stereoisomeric center in the molecule.
Later findings showed that the benzene moiety could be successfully replaced with the tiophene ring (UBP310,
24), resulting in a >500-fold selectivity for native KA receptors over AMPA and NMDA receptors expressed in neonatal rat motor neurons. Further studies on cloned homomeric KARs revealed a high selective affinity of
24 at the GluK1 and GluK3 receptors
[35][36][37]. Interestingly, it was reported that, although glutamate-evoked currents mediated by recombinant GluK3 homomeric receptors were effectively blocked by UBP310 (with an IC
50 of 4.0 μM), the compound did not block recombinant GluK2/3 heteromers and, unlike CNQX, it did not affect presynaptic kainate receptors in mouse hippocampal mossy fiber synapses, which are most likely composed of GluK2/3 heteromers. A similar behavior was observed for the compounds UBP302 (
23) and ACET (UBP316,
25)
[38]. In addition, UBP310 demonstrated activity toward postsynaptic GluK2/5 receptors, inhibited slow excitatory postsynaptic currents (EPSC) mediated by KARs in epileptic mice, and was successfully tested in a mouse model of temporal lobe epilepsy
[39].
The neuroprotective potential of UBP310 was studied in animal models of Parkinson’s disease (PD)
[40]. Administration of this compound significantly improved the survival of the dopaminergic and total neuronal population in the substantia nigra pars compacta (SNpc) in the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-induced mouse model of PD. On the contrary, UBP310 lacked the ability to influence the MPTP-induced loss of dopamine levels or striatum dopamine transporter expression, and the deletion of GluK1, GluK2, or GluK3 appeared to have no effect on the MPTP- or UBP310-induced effects. Furthermore,
24 did not reduce the intracerebral cell loss induced by 6-OHDA intoxication. The results clearly suggested that the neuroprotective effects mediated by UBP310 in the midbrain were not related to its affinity for the specific kainate receptor subunits
[40].
Recent studies have also revealed that UBP310, as an antagonist specific for the GluK1 subunit, was able to reduce the glutamate-induced desensitization of heteromeric GluK1/2 and GluK1/5 receptors. As the binding of an agonist by one subunit may be sufficient for receptor activation, but not sufficient for receptor desensitization, it is reasonable to believe that subunit-selective antagonists can be used as tools to efficiently reduce heteromeric receptor desensitization
[41].
X-ray crystal structures solved for GluK1-LBD in complex with UBP302 and UBP310 have been used to design a close analogue of UBP310, ACET (UBP316,
25), with a similar pharmacological profile. The ACET compound was applied to revisit the physiological role of kainate receptors in the NMDAR-independent mossy fiber LTP, as well as to establish the presynaptic regulation of Ca
2+ facilitation in giant mossy fiber boutons (MFB). As a potent antagonist of GluK1-containing KARs (
Kb approximately 1 nM),
25 was found to fully block LTP induction in the CA3-MF region of the hippocampus in a reversible manner, as well as to depress the presynaptic short-term facilitation of calcium transients in the MFB evoked by a train of action potentials
[38][42].
So far, the crystal structures of the GluK1-LBD in complex with willardiine derivatives have been resolved for
23,
24, and
25 [35][42], all revealing an analogous binding mode of the ligand with residues of the LBD domains 1 and 2. In all cases, the carboxylic group of amino acid functionality was involved in the essential ion pair interaction with the side chain of Arg508 (the numbering of residues in the rat GluK1-LBD construct according to ref.
[35]), as well as a hydrogen bond with the backbone amide group of Thr503. The α-amino group formed H-bond contacts with Thr503 and Pro501. Furthermore, for all three willardiine derivatives, a direct hydrogen bond interaction was observed between the distal carboxyl group and the amide group of Thr675.
1.2. Decahydroisoquinolines
Decahydroisoquinoline derivatives, substituted in the direct vicinity of the nitrogen atom with a carboxylic acid group, can be treated as an expanded version of natural α-amino acid-based iGluR agonists such as glutamate or kainate, but with the α-amino group involved in the cyclic system. Compounds with this scaffold constitute a significant class of competitive AMPAR/KAR antagonists and usually possess a flexible linker with an attached distal acidic group, most often the tetrazolyl or carboxyl group.
One of the first compounds reported among this chemical class, acting as a competitive antagonist of AMPA/GluK1 receptors with micromolar affinity, was LY215490. Its stereoisomer LY293558 (
26, tezampanel,
Figure 3) was found to produce pain relief after oral surgery and in migraine patients and has undergone phase II clinical trials in patients with acute migraine
[43][44][45]. Recent studies using rats exposed to soman also revealed its antiepileptic and neuroprotective efficacy after acute exposure to nerve agents and other organophosphate compounds both in monotherapy and in coadministration with other substances. In vivo studies with immature, young-adult, and aged rats exposed to soman demonstrated that
26 provided protection against brain impairment, prevented long-term behavioral deficits, and terminated status epilepticus, even with 1 h postexposure administration of tezampanel. In addition, full neuroprotection was achieved after simultaneous therapy with the NMDAR antagonist, caramiphen
[46][47].
Figure 3. Structures of cis-decahydroisoquinoline-derived KAR antagonists.
Competitive antagonists with improved selectivity for kainate receptors have been developed to enhance neuroprotective properties and efficacy in pain conditions with an improved side-effect profile
[48]. The selective GluK1 antagonist LY377770 (
27) demonstrated neuroprotective effects in models of global and focal ischemia at doses of 80 mg/kg administered intraperitoneally (i.p.) followed by 40 mg/kg i.p. 3–6 h after the initial dose
[49]. The compounds LY377770 and LY382884 (
28) blocked epileptic activity in hippocampal slices and in vivo limbic seizures in conscious rats induced by pilocarpine or 6 Hz corneal stimulation
[50]. The study of the effect on formalin-induced pain behavior demonstrated the antinociceptive efficacy of LY382884 without ataxia at doses of 5, 10, 30, and 100 mg/kg i.p.
[51]. Meanwhile, in a primate model of peripheral neuropathy,
28 injected into the dorsal horn of the spinal cord via microdialysis (at a concentration of 100 μM–10 mM) reduced the nociceptive responses and in a concentration-dependent manner attenuated the response to mechanical stimuli
[52].
The improvement in oral bioavailability was achieved by preparing a diethyl ester for the selective GluK1 receptor antagonist (
29). The resulting prodrug (
30) showed a high oral efficacy in two animal migraine models. In the neurogenic plasma protein extravasation (PPE) model,
30 administered orally (p.o.) 1 h before trigeminal stimulation exhibited efficacy with an estimated ID
50 = 100 pg/kg, while, in the nucleus caudalis
c-fos expression model, 1 h of pretreatment with a dose of 10 mg/kg resulted in an approximately 30% decrease in the central Fos expression. Furthermore,
29 in the PPE model administered intravenously (i.v.) 10 min before stimulation demonstrated efficacy with an estimated ID
50 of 0.03 pg/kg
[53].
A particularly interesting example of a potent and selective antagonist was LY466195 (
31)
[54][55], which has been found to have more than a 100-fold selectivity for GluK1 over other kainate and AMPA receptor subtypes. Both LY466195 and the diethyl ester prodrug showed significant inhibitory effects in two preclinical models of migraine (
31 administered i.v. and its ester administered p.o. in the PPE model exhibited an estimated ID
50 of 100 μg/kg, while in the
c-fos essay 31 demonstrated efficacy at doses of 1 to 100 μg/kg i.v. and its prodrug above a dose of 100 μg/kg). Compound
31 also demonstrated a lack of vasoconstrictive properties in the rabbit saphenous vein, suggesting that the vasoconstrictive properties of triptan molecules are not required for efficacy in PPE or
c-fos assays
[54].
Further improvement of the in vitro profile was achieved by the introduction of an oxygen linker and phenyl substituent in tetrazole derivatives. The most active compounds in the series, LY458545 (
32) and LY457691 (
33), showed a strong antagonism toward the AMPA and GluK1 receptors, but exhibited poor oral bioavailability. On the contrary, their prodrugs showed bioavailability levels of 24% and 41%, respectively, and both ester prodrugs showed oral efficacy in three animal models of pain. In the formalin test, prodrugs demonstrated efficacy at doses starting from 3 mg/kg and 5 mg/kg p.o., respectively. Furthermore, both compounds affected the dose-dependent reversal of carrageenan-induced thermal hyperalgesia with minimum effective doses (MED) of 1.0 mg/kg and 3.0 mg/kg p.o., and dose-dependent reverse capsaicin-induced mechanical hyperalgesia with MED of 1.0 and 10 mg/kg p.o., respectively
[56].
Further enhancement of the selectivity and affinity for the GluK1 receptors was achieved by evaluating the structure of earlier tetrazole derivatives and introducing a chlorine substituent to obtain LY545694 (
34). The 2-ethylbutyl ester of LY 545,694 showed a bioavailability of 42% in rats after oral administration and was shown to be effective in two animal models of persistent pain (dose-dependent reversal of carrageenan-induced thermal hyperalgesia with MED of 3 mg/kg p.o. and formalin model with MED of 1 mg/kg p.o.)
[57]. The LY545694 tosylate has been studied in phase II of clinical trials for the treatment of pain caused by osteoarthritis of the knee (OA) and diabetic neuropathic pain (DPNP)
[58].
The binding mode of the decahydroisoquinoline derivatives in the GluK1 receptor has been studied in the examples of the X-ray structures of the GluK1-LBD complexes with compounds
31 or
34 [55][57]. In both cases, the interactions of the ligand α-amino acid group involved in the isoquinoline system with the residues of the binding pocket resembled the canonical interactions previously observed for agonists and competitive antagonists with free α-amino acid functionality, e.g., willardiine derivatives.
1.3. Other α-Amino Acid Antagonists
The good example of this group are α-amino acids designed as expended versions of the natural iGluR agonist AMPA, with a structure based on the isoxazole core. Compounds in this series were designed primarily as competitive antagonists of the AMPA receptor (see the reviews of Nikam
[16] and Catarzi
[15]), and studies often lack data on the affinity at the kainate receptors. In this regard, one of the most developed compounds in this group is (
S)-ATPO (
35), the selective AMPAR/GluK1 antagonist that shows an affinity in the low micromolar range for the homomeric GluK1 receptors (
Figure 4)
[59][60]. Analysis of the X-ray structure of ATPO in complex with GluK1 revealed that, in addition to the typical binding mode of the α-amino acid group, seen for other α-amino acid agonists and antagonists in GluK1- or GluA2-LBD, (
S)-ATPO also interacts with its phosphonate tail, forming hydrogen bonds with the non-conserved serine of GluK1, which probably accounts for the observed selectivity profile of the compound
[61].
Figure 4. Structures of selected α-amino acid antagonists of kainate receptors. * the stereoisomeric center in the molecule.
Furthermore, the isoxazole ring of (
S)-ATPO does not form any specific interactions with the receptor, acting mainly as a spacer between the α-amino acid group and the phosphonate functionality. Therefore, the structure of (
S)-ATPO became an inspiration for a series of phenylalanine-based AMPAR/KAR antagonists, in which the isoxazole spacer has been replaced by the benzene ring, substituted with the 2-carboxyethyl chain (
36)
[62][63]. The antagonist
36 showed an equimolar affinity at the native AMPA receptors and homomeric GluK1 receptors in the range of 3 μM and was cocrystallized with the AMPA subunit GluA2-LBD, as well as with the kainate subunit GluK1-LBD, inducing in both X-ray complexes a domain closure similar to that observed in the iGluR structures with partial agonists. The binding mode of
36 for both receptors resembled the binding modes of the (
S)-ATPO and willardiine-based antagonists, with the canonical interaction pattern between the α-amino acid group and the arginine, proline, and threonine residues of the domain 1, conserved among all the non-NMDA subunits. Furthermore, an additional interaction between the nitro group and the Ser721 side chain (which, together with Glu441, forms the so-called ‘interdomain lock’ stabilizing the active state of the receptor), was suggested as an important determinant of ligand affinity.
Another series of phenylalanines reported by the same group was based on the biphenylalanine core substituted with the distal acidic group, either carboxy or phenolic hydroxy substituent (
37,
38)
[64][65][66]. Interestingly, it appeared that this distal polar group had a strong influence on the iGluR affinity profile: compounds with a 3-carboxy group attached to the distal phenyl ring (
37) preferentially bound to GluK1 receptors (with a 35-fold selectivity over AMPAR in the case of
37), while the 3-hydroxy group at this position was optimal for binding to native AMPA receptors (
38) (with a selectivity of 21 times over GluK1 receptors in the case of
38).
The structure of kainic acid became an inspiration for the design of a series of proline derivatives that contained an additional carboxyl group in their structure. The first structure described in this series,
39, demonstrated a micromolar affinity at the homomeric GluK1 receptor with a 20-fold selectivity over the homomeric GluK2 and GluK3 receptors (
Table 4)
[67]. Further modifications of the structure resulted in the improved selectivity and affinity at the homomeric GluK1 receptors (
41,
42)
[68][69], while, in the case of
40, the resulting compound exhibited a 5-fold increased affinity at the GluK3 subunit compared to the GluK1 subunit
[69][70][71]. The X-ray structures of the GluK1-LBD cocrystallized with
41 or
42 were successfully resolved, which allowed for determining the binding mode of the proline derivatives in the binding pocket
[68][69]. The α-amino acid moiety of the ligands interacted with the LBD residues in a manner similar to that described above.
Dysiherbaine derivatives belong to a group of active KAR ligands originally isolated from the marine sponge
Lendenfeldia chodrodes. The synthetic derivative MSVIII-19 (
43) was the first to demonstrate strong and selective antagonistic properties toward GluK1 homomeric receptors, with no affinity for GluA2 receptors
[72][73][74]. Compound
43 showed analgesic activity in animal model studies, supporting the thesis of the potential of GluK1 antagonists in the treatment of pain
[75]. The structural variants of neodysiherbaine (
44)
[76] did not show an improved affinity and selectivity for kainate receptors, compared to
43, but led to useful tools for studying the structure of KAR subunits
[77].
In the group of bicyclic pyrimidine-2,4-dione analogs, on the other hand, subsequent structural adjustments have resulted in compound
45, which exhibited a remarkable selectivity for kainate receptors, especially the GluK1 subunit (over a 400-fold higher affinity for GluK1 compared to GluA2 and 53 times versus GluK3). Functional studies confirmed the antagonist profile of
45, while the in vivo mouse formalin test of prolonged acute pain stimulation proved its analgesic efficacy at a dose of 2 mg/kg
[78]. The molecular interactions observed in the crystal structure resolved for
45 in complex with the GluK1-LBD correspond to those previously observed for other willardiine derivatives
[79]. In particular, the carboxyl group of the amino acid moiety forms an ion pair interaction with the side chain of Arg523 (Arg508 according to residue numbering in ref.
[35]), and a hydrogen bond with the backbone of Thr518 (Thr503), while the α-amino group is involved in multiple H-bond contacts with the carbonyl group of Pro516 (Pro501), the side chain of Thr518, and, additionally, with the Glu738 of domain 2. On the other hand, the flexible carboxythienyl part of molecule
45 interacts with the residues of domain 2 of the GluK1-LBD, thus stabilizing an open conformation of the binding pocket
[79].
(
S)-mercaptohistidine (
46) represents one of the first selective GluK3 receptor antagonists described, with a 30-fold affinity preference over GluK1, GluA3, and GluA4 and a selectivity of more than 100-fold over the recombinant receptors GluK2, GluA1, and GluA2, determined in radioligand binding assays. Surprisingly, the functional evaluation revealed antagonistic activity for the GluK3 receptors, but also weak agonist activity for the GluA2 receptors, making
46 the first orthosteric iGluRs ligand with a mixed agonist–antagonist profile for KARs/AMPARs
[80].
1.4. Structurally Dissimilar AMPAR/KAR Antagonists
In addition to the chemical classes described above, some structurally dissimilar competitive antagonists of AMPA/KA receptors with in vivo anticonvulsant activity have been identified
[81], including quinazoline-2,4-diones (
47 and BGG492, selurampanel,
48)
[82][83], isatin oximes (NS 1209,
49)
[84][85], and derivatives of pyrazine (RPR117824,
50,
Figure 5)
[86][87]. NS 1209 showed potent and dose-dependent anticonvulsant activity in various preclinical models and alleviated refractory status epilepticus and neuropathic pain in phase II of clinical trials, but further research on the molecule has been suspended
[85][88]. RPR117824 in in vivo models blocked convulsions induced in mice or rats by supramaximal electroshock or chemoconvulsive agents such as pentylenetetrazol and showed significant neuroprotective activity in global and focal cerebral ischemia and brain and spinal cord trauma
[86][87]. In the group of quinazoline-2,4-dione derivatives, compound
47 showed a high affinity for native KA receptors and demonstrated neuroprotective effects in an in vitro model of cerebral ischemia
[89][90]. BGG 492 (selurampanel), an orally active AMPAR/KAR antagonist (the functional efficacy on the rat cortical wedge for the AMPAR and KAR was 0.46 and 0.42 μM, respectively), has shown anticonvulsant activity in animal models of epilepsy, such as electroshock or chemically induced seizures in rodents. Currently,
48 has been subjected to clinical trials aimed at patients with epilepsy, tinnitus, and migraine
[82].
Figure 5. Structurally dissimilar AMPAR/KAR antagonists.
2. Non-Competitive Antagonists and Channel Blockers of Kainate Receptors
Despite the fact that the pharmacology of non-competitive inhibitors of AMPA and kainate receptors remains very similar, AMPAR antagonists have attracted more interest in recent years, especially as antiepileptic agents. Among them, a certain number of compounds likely act as non-selective AMPAR/KAR inhibitors, and only a few examples of selective kainate receptor non-competitive antagonists have been described.
The early described group of non-competitive antagonists of non-NMDA receptors, 2,3-benzodiazepine derivatives, are widely recognized as selective AMPAR antagonists with very low potency and efficacy for kainate receptors. Considering that members of this chemical class selectively block AMPAR-conducted responses, some of them are used as pharmacological tools to isolate the corresponding neuronal response from kainate receptors. The compound GYKI 53,655 (
51,
Figure 6) was shown not to affect GluK1 or GluK2-containing kainate receptors or native somatodendritic kainate receptors (IC
50 > 200 mM)
[38][91][92]. However, in the case of GluK3-containing receptors expressed in HEK 293 cells, GYKI 53,655 was able to block currents from activated GluK3 homomeric receptors with an IC
50 = 63 ± 10 µM, and from GluK2/3 heteromeric receptors with an even better result, IC
50 = 32 ± 5 µM. Furthermore, at high concentrations, the compound was found to block presynaptic kainate receptors in mossy fiber synapses (likely composed of GluK2/3 heteromers) and decrease short-term plasticity
[38].
Figure 6. Non-competitive antagonists and channel blockers of kainate receptors.
A real breakthrough in iGluR research was the development of perampanel (
52, Fycompa,
Figure 6), approved as an oral drug for partial-onset seizures and generalized primary tonic–clonic seizures
[93]. Initial studies showed that
52 inhibits native AMPA receptors and also individual AMPAR subunits with no affinity for NMDARs and KARs
[94]. However, recent structural studies of AMPARs in complex with perampanel have shown that its binding site consists of amino acids that are also highly conserved for the GluK4 and GluK5 receptor subunits. Consequently, contrary to previous assumptions,
52 demonstrated the effective inhibition of the heteromeric GluK1/5 and GluK2/5 receptors at levels comparable to the native AMPA receptors. Furthermore, the presence of NETO1 and NETO2 was shown to significantly affect KAR affinity, particularly heteromers containing the GluK2 subunit
[95]. Perampanel has been included in a wide range of clinical trials for epilepsy among children and adults
[93][96][97][98][99][100][101][102][103][104], as well as in studies for Parkinson’s disease
[105][106] or amyotrophic lateral sclerosis
[107][108].
The non-selective inhibitory effect of kainate receptors was also observed in the case of
cis-unsaturated fatty acids. In contrast to fully saturated fatty acids, docosahexaenoic acid (DHA), arachidonic acid (AA), linolenic acid, and linoleic acid demonstrated a significant reduction in kainate-induced whole cell currents and reversibly inhibited kainate receptor signaling by acting within the TMD
[109][110][111].
A group that has contributed to a better understanding of the physiology and pathophysiology of kainate receptors
[112][113] is arylureidobenzoic acids, in particular 4,6-bis (benzylamine)-1,3- benzenedicarboxylic acid (NS 3763,
53), which is a selective antagonist of the homomeric GluK1 receptor, lacking an affinity for the native AMPAR and GluK1/2 and GluK1/5 heteromers
[112][114][115][116]. Proceeding with research on selective kainate receptor inhibitors, the group of Kaczor et al. reported 1,2,3,5-tetrasubstituted indole derivatives, including the first non-competitive GluK2 receptor ligand known so far (
54)
[117]. Studies on these derivatives have contributed to the development of computational research on the structure of kainate receptor subunits and ligand binding sites
[117][118][119].
A drug whose exact mode of action has not yet been clarified is topiramate (
55). Its potent neuroprotective and anticonvulsant effect has been confirmed by several studies using animal models, and at least three mechanisms appear to contribute to this action profile: (1) enhancement of the GABA response; (2) blocking voltage-dependent ion channels, and (3) inhibition of AMPA/KA receptors
[120]. Meanwhile, topiramate has been shown to be ineffective against AMPA- or NMDA-induced clonic seizures, but it is effective in blocking kainate-induced seizures
[121][122]. Subsequent in vivo studies investigating the blocking of the agonist-induced seizure activity of iGluRs confirmed that topiramate selectively inhibits the GluK1 subunit response
[121][123]. For its unique properties, topiramate is used as a treatment for epilepsy with partial and generalized tonic–clonic seizures, as well as for migraine prevention and treatment. In addition to this, several clinical trials have been conducted for the treatment of eating disorders, obesity, alcohol dependence, neuropathy, post-traumatic stress disorder, and Tourette syndrome
[124].
Inhibition of iGluR activity could also occur as a result of the binding of ion channel blockers. Among this group of derivatives, many naturally occurring compounds, as well as their synthetic analogs, have been described. All shared structural similarity (that is, polyamine moiety) and, when applied, blocked the AMPAR/KAR in a voltage-independent manner
[125]. Within the KAR family, the GluK3 subunit demonstrated significant sensitivity to polyamines, especially compared to the GluK1 and GluK2 subunits. Meanwhile, auxiliary proteins of the kainate receptors, NETO1 and NETO2, attenuated the blocking of ion channels
[126]. As mentioned above, permeability to Ca
2+ as well as an affinity of polyamines is related to the editing of the Q/R site, where the kainate receptors of the “R” type are not permeable to Ca
2+ and show no sensitivity to polyamines
[91][127][128]. This effect was considered to derive from the electrostatic repulsive effect of positively charged arginine in the narrowest region of the pore. GluK4- and GluK5-containing heteromers have also been found to weaken polyamine blocking due to the presence of a proline residue that is not represented in the GluK1-3 subunits and changes the dynamics of the α-helical region of the selectivity filter
[126][129][130][131].
Studies using the toxin of the spider venom
Argiope lobata (ArgTX-636,
56)
[128], the
Joro spider toxin (JSTX-3,
57)
[130], the agatotoxin of the North American funnel web spider
Ageleonopsis aperta (AGEL-489,
58 and AGEL-505,
59)
[132], or the active venom fraction of
Philanthus triangulum (PhTX-433,
60)
[133] demonstrated that polyamines are not only potent antagonists of ionotropic glutamate receptors, but can also exhibit other biological activities
[134][135][136][137][138][139]. Synthetic derivatives of philanthotoxins allowed for determining structure–activity relationships, principally concerning AMPA receptors. The modification that resulted in an active blocking of the GluK1 subunit (PhTX-47,
61,
Ki = 0.060 ± 0.014 μM) involved altering the length of the polyamine chain with respect to the number of amine groups and the methylene bridges located between them. In contrast, even a slight modification of the chain length shifted activity at the GluA1 receptors (PhTX-56,
62) with affinity at the GluA1 subunits of
Ki = 3.3 ± 0.78 nM
[137].
Other synthetic polyamine derivatives, IEM-1460 (
63)
[140] and
N1-naphthylacetylspermine (
64)
[141], predominantly affect the GluA2-deficient Ca
2+ receptors and further studies have shown that modifications of both the polyamine chain and the head group can contribute to changes in activity and selective affinity at the AMPA, KA, and NMDA receptors
[142][143][144][145].