5-Nitroisoxazoles in SNAr Reactions: A Novel Chemo- and Regioselective Approach to Isoxazole-Based Bivalent Ligands of AMPA Receptors

An efficient regioselective approach to novel functionalized bis(isoxazoles) with a variety of aromatic and aliphatic linkers was elaborated, based on the heterocyclization reaction of electrophilic alkenes under the treatment with tetranitromethane-triethylamine complex affording 3-EWG-5-nitroisoxazoles. The subsequent SNAr reactions of 5-nitroisoxazoles with various O,O-, N,N- and S,S-bis(nucleophiles) provide a wide range of bis(isoxazole) derivatives in good isolated yields. Employing an elaborated method, a series of novel bis(3-EWG-isoxazoles) as the promising allosteric modulators of AMPA receptors were designed and synthesized. The effect of the compounds on the kainate-induced currents was studied in the patch clamp experiments, revealing modulator properties for several of them. The best positive modulator potency was found for dimethyl 5,5′-(ethane-1,2-diylbis(sulfanediyl))bis(isoxazole-3-carboxylate), which potentiated the kainate-induced currents in a wide concentration range (10−12–10−6 M) with maximum potentiation of 77% at 10−10 M. The results were rationalized using molecular docking and molecular dynamics simulations of modulator complexes with the dimeric ligand-binding domain of the GluA2 AMPA receptor. The predicted physicochemical, ADMET, and PAINS properties confirmed that the AMPA receptor modulators based on the bis(isoxazole) scaffold may serve as potential lead compounds for the development of neuroprotective drugs.


Introduction
AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) receptors are an important subtype of ionotropic glutamate receptors that play a crucial role in mediating fast synaptic transmission necessary for effective learning and memory formation [1,2].Dysfunctions in the AMPA receptor activity are implicated in neurological and psychiatric disorders such as Alzheimer's disease, epilepsy, depression, etc. [3][4][5][6][7]; therefore, AMPA receptors are promising targets for the development of cognitive enhancers [8].It is known that positive allosteric modulators (PAMs) of the AMPA receptor, binding in its allosteric binding site, allow one to fine-tune the AMPA receptor signaling as well as to avoid the excitotoxicity problems commonly observed when a direct agonist is administered in excessive doses [9,10].On the other hand, its negative allosteric modulators (NAMs) could have antiepileptic and neuroprotective properties [11][12][13][14].
Over the previous decades, massive efforts have been devoted to the search for novel AMPA receptor modulators with desired pharmacological efficacy and safety profiles [1,2,5,7,11,15,16].Starting from the classical scaffolds such as benzamides, benzothiadiazines, biarylalkylsulfonamides, and trifluoromethylpyrazoles, many interesting PAM chemotypes have been identified.However, for a number of reasons, they have not yet reached clinical use, but this area of research is still considered valid and promising [11,17].
There is a particular interest in the bivalent ligands of the AMPA receptor because of their unique ability to bind simultaneously by two parts of the molecule at a close distance from each other in the pocket located at the interface between two subunits of the receptor [11,18,19].Recently we reported several novel series of dimeric AMPA receptor PAMs based on different scaffolds; for many of them, potencies in the nanomolar or picomolar concentration ranges were demonstrated in the patch clamp experiments [20][21][22][23].Amazingly, in many cases substantial differences in activity are found even for very similar structures, e.g., one structure could be a potent PAM while another is a potent NAM [11].Such activity cliffs present an interesting but significant challenge in terms of understanding the mechanisms of action and the structure-activity relationships of the compounds.In order to enable in-depth analysis as well as to identify promising leads for pharmaceutical development, a thorough exploration of the modulator chemical space is required that needs to be supported by chemo-and regioselective synthetic approaches.
Earlier, we have developed a novel preparative approach to 3-EWG-5-nitroisoxazoles involving the heterocyclization reaction of electrophilic alkenes with tetranitromethanetriethylamine (TNM-TEA) complex [42].We also employed this method for further chemoand regioselective functionalization of isoxazole moiety via the nitro group reduction [43] and aromatic nucleophilic substitution of nitro group with variety of nucleophiles [44].These approaches allowed us to obtain the isoxazole derivatives with antiviral [45], antioxidant [46], and anticancer [47] activities.
In a recent publication, we reported the synthesis of bis(5-aminoisoxazole) I by the heterocyclization reaction of electrophilic diene with TNM-TEA complex and subsequent reduction of nitro groups of bis(5-nitroisoxazole) (Figure 1) [22].The bis(heterocycle) I was shown to be a highly potent positive modulator of the AMPA receptor that potentiated the kainate-induced ion currents by up to 70% at the concentration of 10 −11 M [22].This result inspired us to examine another type of bis(isoxazoles) obtained by the reaction of 5-nitroisoxazoles with O,O-, N,N-, and S,S-bis(nucleophiles) containing different linkers (Figure 1).
Herein, we report the development and optimization of synthetic approaches to these compounds and the in vitro evaluation of their AMPA receptor modulator activities.The structure-activity relationships in this series were analyzed using molecular modeling techniques such as molecular docking and molecular dynamics simulations, and the computational evaluation of their physicochemical, ADMET, and PAINS properties was performed to confirm the suitability of the AMPA receptor modulators of this type as potential lead compounds for the development of novel neuroprotective drugs.
performed to confirm the suitability of the AMPA receptor modulators of this type as potential lead compounds for the development of novel neuroprotective drugs.

Chemistry
The starting 3-EWG-5-nitroisoxazoles 1a-d were obtained in good yields using the standard protocol for heterocyclization reaction of α,β-unsaturated esters and amides under the treatment with TNM-TEA complex [42].Variation of substituents in position 3 of the isoxazole ring provides the structures of different sizes and levels of hydrophobicity.
In order to investigate the influence of the linker nature on the modulating properties of bis(isoxazoles) to the AMPA receptor, the compounds 3a-c with aromatic hydroquinone linker as well as the series of heterocycles 3d-j bearing acyclic diamines and disulfides with aliphatic chains of various lengths were obtained.
We began our study with the SNAr reaction of 5-nitroisoxazoles 1a-c with hydroquinone as O,O-bis(nucleophile).It was found that 5-nitroisoxazoles 1a,b with ester groups in position 3 of the isoxazole ring smoothly and regioselectively react with hydroquinone under the previously found conditions in acetonitrile at room temperature [44] affording bis(isoxazoles) 3a,b in good yields (Table 1).However, the reaction of amide 1c with hydroquinone in these conditions appeared incomplete due to the low solubility of the starting compound in acetonitrile at room temperature.So, bis(isoxazole) 3c was isolated in 74% yield when the reaction proceeded at reflux in acetonitrile for 6 h.
It was found that the reactions of 5-nitroisoxazoles with aliphatic diamines and disulfides under the above-mentioned conditions lead to complex mixtures of products; therefore, additional optimization of conditions was required.The reaction between model 5-nitroisoxazole 1a and 1,3-propanediamine as a bis(nucleophile) was studied by varying bases, solvents, reaction time, and temperature.When the reaction was performed in acetonitrile, the product 3e was obtained in low yields regardless of the bases used as well as reaction time.Also, it was found that a slight increase in the yield of bis(isoxazole) 3e was achieved when the temperature was increased in CH3CN or if THF was used as a solvent.When the reaction was carried out in EtOH, bis(isoxazole) 3e was obtained in

Chemistry
The starting 3-EWG-5-nitroisoxazoles 1a-d were obtained in good yields using the standard protocol for heterocyclization reaction of α,β-unsaturated esters and amides under the treatment with TNM-TEA complex [42].Variation of substituents in position 3 of the isoxazole ring provides the structures of different sizes and levels of hydrophobicity.performed to confirm the suitability of the AMPA receptor modulators of this type as potential lead compounds for the development of novel neuroprotective drugs.

Chemistry
The starting 3-EWG-5-nitroisoxazoles 1a-d were obtained in good yields using the standard protocol for heterocyclization reaction of α,β-unsaturated esters and amides under the treatment with TNM-TEA complex [42].Variation of substituents in position 3 of the isoxazole ring provides the structures of different sizes and levels of hydrophobicity.
In order to investigate the influence of the linker nature on the modulating properties of bis(isoxazoles) to the AMPA receptor, the compounds 3a-c with aromatic hydroquinone linker as well as the series of heterocycles 3d-j bearing acyclic diamines and disulfides with aliphatic chains of various lengths were obtained.
We began our study with the SNAr reaction of 5-nitroisoxazoles 1a-c with hydroquinone as O,O-bis(nucleophile).It was found that 5-nitroisoxazoles 1a,b with ester groups in position 3 of the isoxazole ring smoothly and regioselectively react with hydroquinone under the previously found conditions in acetonitrile at room temperature [44] affording bis(isoxazoles) 3a,b in good yields (Table 1).However, the reaction of amide 1c with hydroquinone in these conditions appeared incomplete due to the low solubility of the starting compound in acetonitrile at room temperature.So, bis(isoxazole) 3c was isolated in 74% yield when the reaction proceeded at reflux in acetonitrile for 6 h.
It was found that the reactions of 5-nitroisoxazoles with aliphatic diamines and disulfides under the above-mentioned conditions lead to complex mixtures of products; therefore, additional optimization of conditions was required.The reaction between model 5-nitroisoxazole 1a and 1,3-propanediamine as a bis(nucleophile) was studied by varying bases, solvents, reaction time, and temperature.When the reaction was performed in acetonitrile, the product 3e was obtained in low yields regardless of the bases used as well as reaction time.Also, it was found that a slight increase in the yield of bis(isoxazole) 3e was achieved when the temperature was increased in CH3CN or if THF was used as a solvent.When the reaction was carried out in EtOH, bis(isoxazole) 3e was obtained in In order to investigate the influence of the linker nature on the modulating properties of bis(isoxazoles) to the AMPA receptor, the compounds 3a-c with aromatic hydroquinone linker as well as the series of heterocycles 3d-j bearing acyclic diamines and disulfides with aliphatic chains of various lengths were obtained.
We began our study with the S N Ar reaction of 5-nitroisoxazoles 1a-c with hydroquinone as O,O-bis(nucleophile).It was found that 5-nitroisoxazoles 1a,b with ester groups in position 3 of the isoxazole ring smoothly and regioselectively react with hydroquinone under the previously found conditions in acetonitrile at room temperature [44] affording bis(isoxazoles) 3a,b in good yields (Table 1).However, the reaction of amide 1c with hydroquinone in these conditions appeared incomplete due to the low solubility of the starting compound in acetonitrile at room temperature.So, bis(isoxazole) 3c was isolated in 74% yield when the reaction proceeded at reflux in acetonitrile for 6 h.
It was found that the reactions of 5-nitroisoxazoles with aliphatic diamines and disulfides under the above-mentioned conditions lead to complex mixtures of products; therefore, additional optimization of conditions was required.The reaction between model 5-nitroisoxazole 1a and 1,3-propanediamine as a bis(nucleophile) was studied by varying bases, solvents, reaction time, and temperature.When the reaction was performed in acetonitrile, the product 3e was obtained in low yields regardless of the bases used as well as reaction time.Also, it was found that a slight increase in the yield of bis(isoxazole) 3e was achieved when the temperature was increased in CH 3 CN or if THF was used as a solvent.When the reaction was carried out in EtOH, bis(isoxazole) 3e was obtained in moderate yield (36%) because 5-nitroisoxazole partly reacted with EtOH to give 5-ethoxyisoxazole as a by-product.We have found that the yield of compound 3d was significantly improved by changing the solvent in the S N Ar reaction to tert-BuOH, which can be explained by the low nucleophilicity of this solvent.Thus, using the modified conditions, a series of bis(isoxazoles) 3d-j bearing the N,Nand S,S-linear linkers were obtained in good yields as a single regioisomer in each case (Table 2).All synthesized compounds were identified and characterized using the 1 H and 13 C NMR spectroscopy.Note: Reactions of 1a (0.2 mmol) and 1,3-propanediamine (0.1 mmol) in the presence of a base (0.2 mmol) were carried out in 3 mL of solvent.
Thus, using the modified conditions, a series of bis(isoxazoles) 3d-j bearing the N,Nand S,Slinear linkers were obtained in good yields as a single regioisomer in each case (Table 2).All synthesized compounds were identified and characterized using the 1 H and 13 C NMR spectroscopy.

Table 2. Nucleophilic substitution reaction results between 5-nitroisoxazoles 1a-d and various O,O-, N,N-, and S,S-bis(nucleophiles).
moderate yield (36%) because 5-nitroisoxazole partly reacted with EtOH to give 5-ethoxyisoxazole as a by-product.We have found that the yield of compound 3d was significantly improved by changing the solvent in the SNAr reaction to tert-BuOH, which can be explained by the low nucleophilicity of this solvent.Note: Reactions of 1a (0.2 mmol) and 1,3-propanediamine (0.1 mmol) in the presence of a base (0.2 mmol) were carried out in 3 mL of solvent.
Thus, using the modified conditions, a series of bis(isoxazoles) 3d-j bearing the N,Nand S,S-linear linkers were obtained in good yields as a single regioisomer in each case (Table 2).All synthesized compounds were identified and characterized using the 1 H and 13 C NMR spectroscopy.moderate yield (36%) because 5-nitroisoxazole partly reacted with EtOH to give 5-ethoxyisoxazole as a by-product.We have found that the yield of compound 3d was significantly improved by changing the solvent in the SNAr reaction to tert-BuOH, which can be explained by the low nucleophilicity of this solvent.Note: Reactions of 1a (0.2 mmol) and 1,3-propanediamine (0.1 mmol) in the presence of a base (0.2 mmol) were carried out in 3 mL of solvent.

Compound
Thus, using the modified conditions, a series of bis(isoxazoles) 3d-j bearing the N,Nand S,S-linear linkers were obtained in good yields as a single regioisomer in each case (Table 2).All synthesized compounds were identified and characterized using the 1 H and 13 C NMR spectroscopy.moderate yield (36%) because 5-nitroisoxazole partly reacted with EtOH to give 5-ethoxyisoxazole as a by-product.We have found that the yield of compound 3d was significantly improved by changing the solvent in the SNAr reaction to tert-BuOH, which can be explained by the low nucleophilicity of this solvent.Note: Reactions of 1a (0.2 mmol) and 1,3-propanediamine (0.1 mmol) in the presence of a base (0.2 mmol) were carried out in 3 mL of solvent.

Compound
Thus, using the modified conditions, a series of bis(isoxazoles) 3d-j bearing the N,Nand S,S-linear linkers were obtained in good yields as a single regioisomer in each case (Table 2).All synthesized compounds were identified and characterized using the 1 H and 13 C NMR spectroscopy.moderate yield (36%) because 5-nitroisoxazole partly reacted with EtOH to give 5-ethoxyisoxazole as a by-product.We have found that the yield of compound 3d was significantly improved by changing the solvent in the SNAr reaction to tert-BuOH, which can be explained by the low nucleophilicity of this solvent.Thus, using the modified conditions, a series of bis(isoxazoles) 3d-j bearing the N,Nand S,S-linear linkers were obtained in good yields as a single regioisomer in each case (Table 2).All synthesized compounds were identified and characterized using the 1 H and 13 C NMR spectroscopy.

Electrophysiological Evaluation and the Structure-Activity Relationships
All the synthesized compounds were examined for their AMPA receptor modulator activity in the electrophysiological experiments on freshly isolated Purkinje neurons using the patch clamp technique.Their influence on the kainate-induced currents is shown in Table 3.As can be seen, depending on the specific linkers and the substituents in the isoxazole core, the closely related compounds can sometimes have dramatically different activity profiles.Such activity cliffs are also often observed in other series of the bivalent ligands of the AMPA receptor [11].Among the compounds based on the hydroquinone linker, the compound 3a is a moderately potent negative potentiator of the AMPA receptor in the 10 −11 -10 −6 M concentration range, causing a decrease in the current of approximately 17% at 10 −9 M and 44% at 10 −6 M. Two other compounds of this class (3b,c) with smaller isoxazole substituents are apparently inactive.
Among the compounds with diamine linkers, two compounds with longer linkers (3e,f) also were moderately potent negative modulators in this range, with the maximum decrease in the current of about 30-35% at 10 −9 -10 −8 M. Interestingly, they have a bellshaped concentration dependence that is rather unusual for the negative modulators.On the other hand, the compound 3d with a shorter linker was a rather weak positive AMPA receptor modulator in the 10 −10 -10 −6 M range with maximum potentiation (by up to 27%) at 10 −8 M.
Finally, for the compounds based on the dithiol linkers (3g-j), the potentiation of the AMPA receptor currents was observed in a wide concentration range (up to 10 −12 -10 −6 M) with a bell-shaped concentration dependence.Three of these are among the most potent known positive AMPA receptor modulators that have a maximum potentiation of 68% at 10 −9 M (3g), 59% at 10 −8 M (3h) and 77% at 10 −10 M (3j).They are comparable with the previously reported reference compound I with a different isoxazole substitution pattern.

Molecular Modeling
In order to clarify the likely mechanisms of action of the allosteric modulators 3a-j, we used molecular docking and molecular dynamics simulations to study their interactions with the dimeric ligand-binding domain (LBD) of the GluA2 AMPA receptor.The visual inspection of the trajectories, as well as the plots of the root-mean-square deviations (RMSD) for the protein, glutamate and ligand heavy atoms, show that the systems retained stability and the compounds were stably bound in the PAM binding site at the interface between the LBD subunits over the entire course of the simulation (100 ns), although their exact positions often were adjusted compared to the initial docking poses (see Supplementary Materials, Figures S1 and S2).
In agreement with both the X-ray and molecular modeling data [11,[18][19][20]22,23,[48][49][50][51] for other larger and dimeric ("Class 3" [48]) modulators, the modulator molecules are located in the symmetrical PAM binding site, occupying its central subpocket as well as one or both of the side subpockets (Figure S1).As an example, the binding mode of the most potent PAM molecule 3j is examined in detail in Figure 2. The binding was primarily stabilized by steric fit and hydrophobic interactions.The RMSD plots (Figure 3) confirm the stability of the system and demonstrate the adjustment of the ligand position.
we used molecular docking and molecular dynamics simulations to study their interactions with the dimeric ligand-binding domain (LBD) of the GluA2 AMPA receptor.The visual inspection of the trajectories, as well as the plots of the root-mean-square deviations (RMSD) for the protein, glutamate and ligand heavy atoms, show that the systems retained stability and the compounds were stably bound in the PAM binding site at the interface between the LBD subunits over the entire course of the simulation (100 ns), although their exact positions often were adjusted compared to the initial docking poses (see Supplementary Materials, Figures S1 and S2).
In agreement with both the X-ray and molecular modeling data [11,[18][19][20]22,23,[48][49][50][51] for other larger and dimeric ("Class 3" [48]) modulators, the modulator molecules are located in the symmetrical PAM binding site, occupying its central subpocket as well as one or both of the side subpockets (Figure S1).As an example, the binding mode of the most potent PAM molecule 3j is examined in detail in Figure 2. The binding was primarily stabilized by steric fit and hydrophobic interactions.The RMSD plots (Figure 3) confirm the stability of the system and demonstrate the adjustment of the ligand position.However, as is often the case in series of closely related ligands acting on targets with complex structures and mechanisms of operation [52], no simple explanation of the differences in activity profiles for compounds 3a-j could be derived from the inspection of their binding positions, RMSD plots, or binding free energies calculated using the MM/GBSA (molecular mechanics, generalized Born, surface area) method (see Supplementary Materials, Figures S1 and S2, Table S1).Previously, we have hypothesized that the negative allosteric modulator activity of the compounds could be mediated by their action on another target or binding site, such as the NAM binding site at the interface between the ligand-binding and transmembrane domains of the AMPA receptor [23].Similar to the analysis [52] and inspired by visual observations of changes in the protein subunit arrangement over the course of the trajectories, we suggest that different PAM activity profiles could be related to finer differences in the subunit dynamics that could, in turn, affect the receptor gating and desensitization [1,11,53,54].However, as is often the case in series of closely related ligands acting on targets with complex structures and mechanisms of operation [52], no simple explanation of the differences in activity profiles for compounds 3a-j could be derived from the inspection of their binding positions, RMSD plots, or binding free energies calculated using the MM/GBSA (molecular mechanics, generalized Born, surface area) method (see Supplementary Materials, Figures S1 and S2, Table S1).Previously, we have hypothesized that the negative allosteric modulator activity of the compounds could be mediated by their action on another target or binding site, such as the NAM binding site at the interface between the ligand-binding and transmembrane domains of the AMPA receptor [23].Similar to the analysis [52] and inspired by visual observations of changes in the protein subunit arrangement over the course of the trajectories, we suggest that different PAM activity profiles could be related to finer differences in the subunit dynamics that could, in turn, affect the receptor gating and desensitization [1,11,53,54].
In order to evaluate this hypothesis, we tried to correlate the modulator activity profiles to their MM/GBSA binding energies and the geometrical parameters of the subunit arrangement (time-averaged angles and distances between their key secondary structure elements as explained in Section 3.3) using the principal component analysis (PCA).The values of the parameters are presented in the Supplementary Materials (Table S1).Compound 3c was excluded to avoid distortion of the PCA model (its parameters are significantly different from the rest of the dataset), also taking into account its lack of activity.The plot of the compound positions in the PCA space (Figure 4) indicates that in this series, the highly potent PAMs (and, to a somewhat lesser extent, NAMs) of the AMPA receptor tend to cluster in the right part of the PCA plot (at high positive values of the first component), while moderately active PAMs and inactive compounds are found in the left part (at negative values of the first component).While this preliminary result is encouraging, in order to derive a reliable predictive model of PAM activity, a more detailed analysis is clearly required that should involve a broader and more diverse set of compounds as well as expand and refine the set of structural parameters.In order to evaluate this hypothesis, we tried to correlate the modulator activity profiles to their MM/GBSA binding energies and the geometrical parameters of the subunit arrangement (time-averaged angles and distances between their key secondary structure elements as explained in Section 3.3) using the principal component analysis (PCA).The values of the parameters are presented in the Supplementary Materials (Table S1).Compound 3c was excluded to avoid distortion of the PCA model (its parameters are significantly different from the rest of the dataset), also taking into account its lack of activity.The plot of the compound positions in the PCA space (Figure 4) indicates that in this series, the highly potent PAMs (and, to a somewhat lesser extent, NAMs) of the AMPA receptor tend to cluster in the right part of the PCA plot (at high positive values of the first component), while moderately active PAMs and inactive compounds are found in the left part (at negative values of the first component).While this preliminary result is encouraging, in order to derive a reliable predictive model of PAM activity, a more detailed analysis is clearly required that should involve a broader and more diverse set of compounds as well as expand and refine the set of structural parameters.

Prediction of Physicochemical, ADMET, and PAINS Profiles
Several key physicochemical and ADMET properties for compounds 3a-j were evaluated using predictive models (Table 4).High predicted values for intestinal absorption indicate their suitability for oral administration.According to the commonly accepted rules of thumb for potential drug-like compounds, the predicted lipophilicities and aqueous solubilities also were appropriate.Acceptable CNS bioavailability could be anticipated due to the moderate predicted blood-brain barrier permeability.For the cardiac toxicity risk, both predicted parameters (hERG pK i and pIC 50 , 4.3-7.4log units) were in the lower or medium parts of their possible ranges (3-9 log units), indicating a likely absence of hERG liabilities.The integral quantitative estimate of drug-likeness (QED) was greater than 0.4, confirming the favorable likely properties.The pan-assay interference compounds (PAINS) filter check did not identify any alerts.

Prediction of Physicochemical, ADMET, and PAINS Profiles
Several key physicochemical and ADMET properties for compounds 3a-j were evaluated using predictive models (Table 4).High predicted values for intestinal absorption indicate their suitability for oral administration.According to the commonly accepted rules of thumb for potential drug-like compounds, the predicted lipophilicities and aqueous solubilities also were appropriate.Acceptable CNS bioavailability could be anticipated due to the moderate predicted blood-brain barrier permeability.For the cardiac toxicity risk, both predicted parameters (hERG pKi and pIC50, 4.3-7.4log units) were in the lower or medium parts of their possible ranges (3-9 log units), indicating a likely absence of hERG liabilities.The integral quantitative estimate of drug-likeness (QED) was greater than 0.4, confirming the favorable likely properties.The pan-assay interference compounds (PAINS) filter check did not identify any alerts.Overall, the predicted ADMET, physicochemical, and PAINS properties of the allosteric modulators 3a-j were quite acceptable for the potential lead compounds at the early drug development stages, although additional checks and structure optimization would likely be required.Overall, the predicted ADMET, physicochemical, and PAINS properties of the allosteric modulators 3a-j were quite acceptable for the potential lead compounds at the early drug development stages, although additional checks and structure optimization would likely be required.
5-Nitroisoxazoles 1a-c [42] were synthesized by described methods.All other starting materials were commercially available.All reagents except commercial products of satisfactory quality were purified by literature procedures prior to use.

Electrophysiological Evaluation
The AMPA receptor modulator activity of the compounds was evaluated in electrophysiological experiments in vitro using a patch clamp technique on the freshly isolated rat Purkinje neurons (for details, see Supplementary Materials, Section S2) as described earlier [20,22].

Molecular Modeling
The modeling of the ligand interactions with the dimeric ligand-binding domain of the GluA2 AMPA was performed as described earlier [22,23] (for detailed computational protocol, see Supplementary Materials, Section S3).In brief, upon clean up of the structure obtained from the Protein Data Bank (PDB: 4FAT) [55], the protein was allowed to relax during the molecular dynamics simulation, and the most frequently occurring structure over the stable part of the trajectory was identified.The protein and ligand structures were prepared for molecular docking using the Avogadro 1.2.0 (Avogadro Chemistry, https://avogadro.cc/accessed on 1 September 2023) [56] and AutoDock Tools 1.5.7 (The Scripps Research Institute, La Jolla, CA, USA, https://ccsb.scripps.edu/mgltools/accessed on 1 September 2023) [57] software.The molecular docking into the positive allosteric modulator binding site was performed with the AutoDock Vina 1.1.2software (The Scripps Research Institute, La Jolla, CA, USA, https://vina.scripps.edu/accessed on 1 September 2023) [58].The complex model based on the best selected pose was built using the USCF Chimera 1.15 software (University of California San Francisco, San Francisco, CA, USA, https://www.cgl.ucsf.edu/chimera/accessed on 1 September 2023) [59].
For the analysis of the ligand influences on the LBD subunit arrangement dynamics, a number of geometrical parameters (angles and distances between the key secondary structure elements at the interface of subunits A and B) were defined as shown in Figure 5 and Table 5.The parameters were calculated with a Python script using the PYTRAJ 2.0.6 software (Hai Nguyen, Daniel R. Roe, Jason Swails, David A. Case, Amber development team, https://github.com/Amber-MD/pytraj/accessed on 1 September 2023) [65,69] in the AmberTools 22 package [66] and their average values over the stable portion of the trajectories (last 20 ns, 101 frames at 200 ps interval) listed in Table S1 were used to build the PCA models.DataWarrior 5.5.0 software (Idorsia Pharmaceuticals Ltd., https://openmolecules.org/accessed on 1 September 2023) was used for the management, search, and analysis of the structure-activity data.

Conclusions
We have elaborated an efficient, simple and concise approach to the previously unknown functionalized bis(isoxazoles) bearing aromatic and aliphatic linkers with O,O-, N,Nor S,S-nucleophilic centers that involves the heterocyclization reaction of readily available electrophilic alkenes upon the treatment with TNM-TEA complex and subsequent regioselective aromatic nucleophilic substitution of the nitro group in a resulting 5-nitroisoxazole with variety of bis(nucleophiles).The reaction allows the linkers of different natures and tolerates a variety of moieties in the isoxazole ring.The remarkable activity of bis(isoxazoles) as positive allosteric modulators of the AMPA receptor was demonstrated in the electrophysiological experiments.The potentiation of the kainate-induced AMPA receptor currents was observed in a wide concentration range (10 −12 -10 −6 M), and three of the compounds bearing the dithiol linker were among the most potent known AMPA receptor PAMs with the maximum potentiation of 68% at 10 −9 M (3g), 59% at 10 −8 M (3h) and 77% at 10 −10 M (3j).Several other compounds also were found to be moderately potent negative modulators of the receptor, forming interesting activity cliffs.
Molecular modeling confirms that these compounds can interact with the validated PAM binding site.However, as is often the case in the series of closely related ligands acting on targets with complex structures and mechanisms of operation, no simple correlation or explanation of the differences in activity profiles for compounds 3a-j could be derived from the inspection of their binding positions or estimated free energies.Nevertheless, a multivariate analysis of finer LBD subunit dynamics parameters leads to a rough correlation.While this preliminary result is encouraging, in order to derive a reliable predictive model of PAM activity, a more detailed analysis is clearly required that should involve a broader and more diverse set of compounds as well as expand and refine the set of structural parameters.
The predicted physicochemical, ADMET, and PAINS properties of the bis(isoxazoles) 3a-j are quite acceptable for potential lead compounds at the early drug development stages.Taking into account the synthetic accessibility and possibility of the "property finetuning" for the heterocycles of this structural type, we can envision their further beneficial applications in the design of novel promising compounds possessing neuroprotective and nootropic activities.

Figure 2 .
Figure 2. Binding mode of the PAM 3j, refined using molecular dynamics simulation (100 ns).(A) General view of the dimeric ligand binding domain of AMPA receptor (GluA2) and location of

Figure 2 .
Figure 2. Binding mode of the PAM 3j, refined using molecular dynamics simulation (100 ns).(A) General view of the dimeric ligand binding domain of AMPA receptor (GluA2) and location of the binding site.(B) Binding pockets in the protein molecular surface colored by local hydrophobicity (brown for hydrophobic and blue for hydrophilic).(C) Detailed view of the binding site.The ligand is represented by a grey ball-and-stick model, and the amino acid residues located within 3 Å of it are represented by beige stick models.

Figure 3 .
Figure 3. RMSD of the protein, glutamate and ligand 3j heavy atoms during molecular dynamics simulation of the modulator complex with the dimeric ligand-binding domain of the GluA2 AMPA receptor.

Figure 3 .
Figure 3. RMSD of the protein, glutamate and ligand 3j heavy atoms during molecular dynamics simulation of the modulator complex with the dimeric ligand-binding domain of the GluA2 AMPA receptor.

Figure 4 .
Figure 4. PCA plot showing the modulator activities of the compounds in the principal component space based on the MM/GBSA binding energies and the geometrical parameters of the subunit arrangement in their complexes with the AMPA receptor LBD.Pos-PAM, Neg-NAM, NA-no activity; for the PAMs, the potentiation percentage is also shown.

Figure 4 .
Figure 4. PCA plot showing the modulator activities of the compounds in the principal component space based on the MM/GBSA binding energies and the geometrical parameters of the subunit arrangement in their complexes with the AMPA receptor LBD.Pos-PAM, Neg-NAM, NA-no activity; for the PAMs, the potentiation percentage is also shown.

17 Figure 5 .Table 5 .
Figure 5. Definitions of key secondary structure elements for the analysis of LBD subunit arrangement.At the interface in both subunits (A and B), the centroids are defined for the Cα atoms of the small (S, residues 606-610) and large (L, residues 742-756) α-helices as well as for the central β-sheet (C, residues 617-622, 727-732).The secondary structure elements are shown in magenta and the centroids in cyan.The three centroids define the interface plane for each subunit.Table 5. Definitions of the geometrical parameters of the LBD subunit arrangement.Name Description angleAB Angle between the A and B planes (0 = parallel) heightFaceAB Mean height of the plane center point over the other plane distFace Distance between the face center points shiftFaceAB Mean lateral shift between the face center point and the projection of the other face center point heightS_AB Mean height of the plane S point over the other plane shiftSL_AB Mean lateral shift between the projection of the face S point and the other face L point heightL_AB Mean height of the plane L point over the other plane shiftLS_AB Mean lateral shift between the projection of the face L point and the other face S point

Figure 5 .
Figure 5. Definitions of key secondary structure elements for the analysis of LBD subunit arrangement.At the interface in both subunits (A and B), the centroids are defined for the Cα atoms of the small (S, residues 606-610) and large (L, residues 742-756) α-helices as well as for the central β-sheet (C, residues 617-622, 727-732).The secondary structure elements are shown in magenta and the centroids in cyan.The three centroids define the interface plane for each subunit.

Table 5 .
Definitions of the geometrical parameters of the LBD subunit arrangement.Name Description angleAB Angle between the A and B planes (0 = parallel) heightFaceAB Mean height of the plane center point over the other plane distFace Distance between the face center points shiftFaceAB Mean lateral shift between the face center point and the projection of the other face center point heightS_AB Mean height of the plane S point over the other plane shiftSL_AB Mean lateral shift between the projection of the face S point and the other face L point heightL_AB Mean height of the plane L point over the other plane shiftLS_AB Mean lateral shift between the projection of the face L point and the other face S point

Author
Contributions: Conceptualization, E.V.R., E.B.A. and V.A.P.; investigation, D.A.V., N.S.T., S.E.D., E.V.R., Y.K.G., A.V.G., V.L.Z., V.V.G., E.B.A. and V.A.P.; writing-original draft preparation, D.A.V., E.V.R. and E.B.A.; writing-review and editing, E.V.R., E.B.A. and V.A.P.; supervision, E.B.A. and V.A.P.; project administration, E.B.A. and V.A.P.; funding acquisition, V.A.P.All authors have read and agreed to the published version of the manuscript.Funding: This research was funded by the Russian Science Foundation, grant no.22-15-00041.The development of the electrophysiological evaluation technique was supported by the State Assignment of IPAC RAS, topic no.FFSN-2021-0005.Institutional Review Board Statement: This study was conducted in accordance with the Declaration of Helsinki, regulatory documentation related to the humane treatment of animals, and standard operating procedures (SOPs) of the IPAC RAS neuroreception laboratory.All animal procedures for the isolation of rat Purkinje neurons were approved by the Bioethics Committee of IPAC RAS (Approval No. 41, 26 November 2019).

Table 3 .
The effect of various concentrations of the compounds on the kainate-induced AMPA receptor currents in rat cerebellum Purkinje cells.
Note: The range of maximum potentiation is marked in bold.

Table 4 .
Predicted physicochemical and ADMET profiles of compounds 3a-j.

Table 4 .
Predicted physicochemical and ADMET profiles of compounds 3a-j.