International Journal of Molecular Sciences Article Ligand-Based Discovery of a Small Molecule as Inhibitor of α-Synuclein Amyloid Formation Laura De Luca 1 , Serena Vittorio 1 , Samuel Peña-Díaz 2,3 , Giovanna Pitasi 1 , Marc Fornt-Suñé 2,3 , Federica Bucolo 1 , Salvador Ventura 2,3,4 and Rosaria Gitto 1, * 1 Department of Chemical, Biological, Pharmaceutical and Environmental Sciences, University of Messina, Viale F. Stagno D’Alcontres 31, I-98166 Messina, Italy 2 Institut de Biotecnologia i Biomedicina, Universitat Autonoma de Barcelona, 08193 Bellaterra, Spain 3 Departament de Bioquimica i Biologia Molecular, Universitat Autonoma de Barcelona, 08193 Bellaterra, Spain 4 ICREA, Passeig Lluis Companys 23, 08010 Barcelona, Spain * Correspondence: rosaria.gitto@unime.it Abstract: α-Synuclein (α-Syn) aggregates are implicated in Parkinson’s disease (PD), so inhibitors of α-Syn aggregation have been intensively explored. It has been demonstrated that small molecules might be able to reduce α-Syn aggregation in fibrils, thus exerting neuroprotective effects in models of PD. To expand our knowledge about the structural requirements for blocking the recognition process into the oligomeric assembly of α-Syn aggregates, we performed a ligand-based virtual screening procedure using two well-known α-Syn aggregation inhibitors, SynuClean-D and ZPD-2, as query compounds. A collection of thirty-four compounds bearing distinct chemical function- alities and mutual chemical features were studied in a Th-T fluorescence test, thus identifying 5-(2,6-dinitro-4-(trifluoromethyl)benzyl)-1-methyl-1H-tetrazole (named MeSC-04) as a potent α-Syn Citation: De Luca, L.; Vittorio, S.; amyloid formation inhibitor that demonstrated similar behavior when compared to SynuClean-D Peña-Díaz, S.; Pitasi, G.; Fornt-Suñé, in the thioflavin-T-monitored kinetic assays, with both molecules reducing the number and size of M.; Bucolo, F.; Ventura, S.; Gitto, R. amyloid fibrils, as evidenced by electron microscopy. Molecular modeling studies suggested the Ligand-Based Discovery of a Small binding mode of MeSC-04 through the identification of putative druggable pockets on α-syn fibrils Molecule as Inhibitor of α-Synuclein and a subsequent consensus docking methodology. Overall, this work could furnish new insights in Amyloid Formation. Int. J. Mol. Sci. the development of α-Syn amyloid inhibitors from synthetic sources. 2022, 23, 14844. https://doi.org/ 10.3390/ijms232314844 Keywords: alpha-synuclein; Parkinson’s disease; small molecule; virtual screening; binding site Academic Editors: Konstantin prediction; Th-T fluorescence assay K. Turoverov, Irina M. Kuznetsova and Botond Penke Received: 19 September 2022 1. Introduction Accepted: 24 November 2022 Published: 27 November 2022 Parkinson’s disease (PD) is a chronic and progressive neurodegenerative process that is commonly related to the loss of dopaminergic neurons in the substantia nigra pars Publisher’s Note: MDPI stays neutral com-pacta (SNpc). It has been demonstrated that PD is characterized by an abnormal with regard to jurisdictional claims in intraneuronal accumulation of α-synuclein (α-Syn) aggregates called Lewy bodies, which published maps and institutional affil- are considered to be responsible for the progression to tremors, rigidity, posture instability, iations. akinesia, and cognitive impairment [1]. The native monomeric form of α-Syn is intrinsically disordered in solution; it polymerizes into oligomers that form higher-order species and fib- rils that accumulate in Lewy bodies and Lewy neurites, leading to synucleinopathy [2]. The Copyright: © 2022 by the authors. α-Syn monomers are water-soluble and nontoxic, whereas misfolded α-Syn accumulates as Licensee MDPI, Basel, Switzerland. insoluble fibrils that aggregate through the formation of soluble oligomeric intermediates, This article is an open access article generating insoluble forms that are toxic to neurons [3]. The α-Syn monomer is a small distributed under the terms and protein containing three domains: an N-terminal domain (1–60), a central nonamyloid conditions of the Creative Commons component (NAC) domain (61–95), and a C-terminal domain (96–140). The N-terminal Attribution (CC BY) license (https:// domain forms an amphipathic lysine-rich α-helix that is responsible for the ability of α-Syn creativecommons.org/licenses/by/ to associate with vesicles and membranes. The central NAC domain of α-Syn controls mis- 4.0/). folding into β-sheet-rich amyloid aggregates. Finally, the polar C-terminal domain contains Int. J. Mol. Sci. 2022, 23, 14844. https://doi.org/10.3390/ijms232314844 https://www.mdpi.com/journal/ijms Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 2 of 14 (1–60), a central nonamyloid component (NAC) domain (61–95), and a C-terminal domain (96–140). The N-terminal domain forms an amphipathic lysine-rich α-helix that is Int. J. Mol. Sci. 2022, 23, 14844 2 of 14 responsible for the ability of α-Syn to associate with vesicles and membranes. The central NAC domain of α-Syn controls misfolding into β-sheet-rich amyloid aggregates. Finally, the polar C-terminal domain contains negatively charged serine and tyrosine residues that negatively can charged serine be phosphorylated, and tyrosine regulating residues that the membrane can be binding, phosphorylated, aggregation, regulating and toxicity of α- the membrane Syn binding, [3]. It has been aggregation, demonstrated and toxicity that increases of α-Syn [3]. or decreases It haslevels in α-Syn been are demonstrated implicated that in increasesneurotransmitter regulating or decreases in α-Synrelease,levels are implicated synaptic activity, and in plasticity regulatinginneurotransmit- physiological ter release, synaptic activity, and plasticity in physiological conditions. However, α-Syn aggregates have been linked to the progression conditions. However, α-Syn and spread aggregates have been linked to the progression and spread of pathological of pathological adducts that correspond with the clinical progression of PD. Therefore, adducts that it correspond with the clinical progression of PD. Therefore, it is presumed is presumed that therapeutic interventions modulating α-Syn misfolding may prevent or that therapeutic interventions delay modulating the progression α-Syn of PD misfolding to the disabling may prevent motor or delayand symptoms theeventually progressioncognitive of PD to the disabling motor symptoms and eventually impairment linked to neuronal synucleinopathies [4]. cognitive impairment linked to neuronal synucleinopathies Therefore, the[4].inhibition of these pathogenic α-Syn aggregates could offer Therefore, opportunities inthe theinhibition treatmentof ofthese PD inpathogenic early statesα-Syn beforeaggregates the occurrencecould of offer opportu- irreversible nities in the treatment of PD in early states before the occurrence of irreversible neurodegenerative processes. It has been established that small molecules could interfere neurode- generative processes. It has been established that small molecules could interfere with with the interaction of β-sheet surfaces and misfolded α-Syn oligomers, thus preventing the interaction of β-sheet surfaces and misfolded α-Syn oligomers, thus preventing their their aggregations. The misfolded proteins are generally considered “low druggable” aggregations. The misfolded proteins are generally considered “low druggable” targets, targets, as they lack specific ligand-binding sites that efficiently establish strong binding as they lack specific ligand-binding sites that efficiently establish strong binding with in- with inhibitors or modulators. However, there are recently developed synthetic and hibitors or modulators. However, there are recently developed synthetic and natural small natural small molecules that inhibit α-Syn aggregation and fibril formation in cell-based molecules that inhibit α-Syn aggregation and fibril formation in cell-based assays [5–14]. assays [5–14]. These therapeutics are generally defined as disaggregators and might be These therapeutics are generally defined as disaggregators and might be characterized characterized by a distinct mode of action. Among them, several compounds (Figure 1) by a distinct mode of action. Among them, several compounds (Figure 1) have entered have entered clinical trials: 5-(1,3-benzodioxol-5-yl)-3-(3-bromophenyl)-1H-pyrazole clinical trials: 5-(1,3-benzodioxol-5-yl)-3-(3-bromophenyl)-1H-pyrazole [15,16] (Anleb138b) [15,16] (Anleb138b) proved to inhibit α-Syn oligomer formation, whereas 1-(3-(1H-indol- proved to inhibit α-Syn oligomer formation, whereas 1-(3-(1H-indol-3-yl)propanoyl)-6- 3-yl)propanoyl)-6-butyl-8-(piperidin-4-ylmethyl)hexahydro-4H-pyrazino [1,2- butyl-8-(piperidin-4-ylmethyl)hexahydro-4H-pyrazino [1,2-a]pyrimidine-4,7(6H)-dione a]pyrimidine-4,7(6H)-dione (NPT100-18A) [11] was demonstrated (NPT100-18A) [11] waswith to interfere demonstrated to interfere α-Syn aggregation bywith α-Syn displacing aggregation by displacing the protein from membranes. Moreover, the the protein from membranes. Moreover, the thioflavin-T assay (Th-T) identified 5-nitro-6- thioflavin-T assay (Th-T) identified 5-nitro-6-(3-nitrophenyl)-2-oxo-4-(trifluoromethyl)-1H-pyridine-3- (3-nitrophenyl)-2-oxo-4-(trifluoromethyl)-1H-pyridine-3-carbonitrile (SynuClean-D, also carbonitrile known as SC-D, (SynuClean-D, Figure 1) [13]also andknown as SC-D, Figure 1) [13] and 4-cyclohexyl-2-((2- 4-cyclohexyl-2-((2-nitro-4-(trifluoromethyl)phenyl)thio)- nitro-4-(trifluoromethyl)phenyl)thio)-6-oxo-1,6-dihydropyrimidine-5-carbonitrile 6-oxo-1,6-dihydropyrimidine-5-carbonitrile (ZPD-2, Figure 1) [17] as new α-Syn aggrega- (ZPD- 2, Figure tion 1) [17] as new α-Syn aggregation inhibitors. inhibitors. anle138b NPT100-18A SynuClean-D (SC-D) ZPD-2 Figure 1. Figure Chemical structures 1. Chemical structures of of Anle138b, Anle138b, NPT100-18A, NPT100-18A, SynuClean-D SynuClean-D (also (also known known as as SC-D), SC-D), and and ZPD-2, ZPD-2, well-known well-known inhibitors inhibitors of of α-synuclein aggregation. aggregation. Successively, exciting Successively, exciting outcomes outcomesfrom fromthe theexperimental experimentalinin vivo and vivo in vitro and data in vitro of the data of 2-pyridone-based the SynuClean-D 2-pyridone-based and 4-pyrimidinone-based SynuClean-D ZPD-2 prompted and 4-pyrimidinone-based ZPD-2 researchers prompted to develop new researchers analogsnew to develop that analogs demonstrated promising biological that demonstrated promisingproperties [8–10]. biological These properties efforts revealed that in some cases a single aromatic ring decorated with suitable [8–10]. These efforts revealed that in some cases a single aromatic ring decorated with very simple molecular moieties constitutes the magic player to reduce α-Syn aggregation in in vitro models [9]. Int. J. Mol. Sci. 2022, 23, 14844 3 of 14 2. Results and Discussion Here, we report a computational and experimental screening to identify new potential inhibitors of α-Syn aggregation that are structurally related to well-known active com- pounds. Moreover, we analyzed their plausible binding pocket located on the NAS domain. 2.1. Similarity-Based Virtual Screening (VS) To find new chemical entities, we employed the SwissSimilarity web tool to per- form a similarity-based virtual screening (VS) procedure, which is a method based on the well-established assumption that similar molecules might be prone to produce similar bio- logical effects. The free-of-charge SwissSimilarity platform (http://www.swisssimilarity.ch, accessed on 1 February 2021) can perform a ligand-based virtual screening (LBVS) on large libraries of commercially available molecules, including approved drugs as well as bioactive compounds. The selection of the VS candidates is generally performed through the application of bidimensional and three-dimensional search methods, starting from a specific query compound following the screening procedure information provided by the SIB Swiss institute of Bioinformatics [18,19]. In this work, the two inhibitors SC-D and ZPD-2 were used as reference structures (Figure 1), while the SPECS library, com- posed of 326,000 commercially available small molecules, was selected for the searching procedure. To estimate the similarity between the query molecules and the compounds contained in the screening library, we applied three different methods: FP2 fingerprints, Electroshape-5D, and spectrophores. The first one is a 2D similarity method that encodes linear molecular fragment paths of up to seven atoms into a 1024 bit string [20]. Instead, ElectroShape [21] and spectrophores [22] are 3D similarity methods based on molecular shape complementarity; these methods do not require geometry alignment, making the VS process faster. In both methods, the molecular shape is described by 3D molecular properties whose information is compressed into vectors [19]. The three methods were independently applied for each query molecule, and the number of compounds obtained from each approach is reported in Table 1. Table 1. Number of hits obtained from the VS runs employing SC-D and ZPD-2 as query molecules. The number of compounds selected from each approach for further analysis is reported in brackets. Query Cmpds FP2 Electroshape a Spectrophores b SC-D 3 (3) 33 (33) 400 (108) ZPD-2 2 (2) 400 (178) 400 (39) a Reference [21]; b Reference [22]. By using SC-D as a query, the application of FP2 fingerprints led to the identification of three hits, whereas two hits were obtained from the ZPD-2 query compound. In turn, the Electroshape method yielded thirty-three compounds whose Manhattan distance (MD) values varied from 0.861 to 0.707 in the case of the SC-D query compound. Instead, four- hundred hits, with MD scores ranging between 0.930 and 0.831, were obtained with the same approach starting from the inhibitor ZPD-2. In this case, we further filtered off the resulting hits according to the similarity score by selecting one hundred and seventy-eight molecules whose MD values were below 0.855. Finally, 400 hits were identified through the spectrophore method for both compounds. The obtained hits fell in a similar range of MD values that were between 0.833 and 0.764 when SC-D was used as reference and between 0.819 and 0.754 when ZPD-2 was employed. Therefore, we applied the same MD threshold value of 0.785 as a prefilter, thus obtaining 39 compounds from ZPD-2 and 108 hits from SC- D. Overall, 363 molecules were chosen from the LBVS runs for further analysis. Specifically, we discarded the compounds with stereocenters or E/Z bonds whose configuration or geometry was not defined. Concerning the derivatives possessing similar scaffolds, we selected those compounds characterized by the highest similarity scores. Finally, the remaining molecules were filtered according to (i) their drug-like properties, measured by Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 4 of 14 whose configuration or geometry was not defined. Concerning the derivatives possessing Int. J. Mol. Sci. 2022, 23, 14844 similar scaffolds, we selected those compounds characterized by the highest similarity 4 of 14 scores. Finally, the remaining molecules were filtered according to (i) their drug-like properties, measured by the Lipinsky rule of five; (ii) the absence of PAINS, calculated by the the SwissADME Lipinsky ruleplatform (http://www.swisssimilarity.ch, of five; (ii) the absence of PAINS, calculated accessed on 1 February by the SwissADME 2021); platform and (iii) the availability from commercial suppliers. The application of these (http://www.swisssimilarity.ch, accessed on 1 February 2021); and (iii) the availability filters led to the fromfinal selectionsuppliers. commercial of thirty-four compoundsofcollected The application in Figure these filters 2; among led to the them, of final selection thethirty- first subset of twelve molecules four compounds collected(MeSC-01–MeSC-12) was retrieved in Figure 2; among them, using SC-D the first subset as a query of twelve mol- molecules ecule, whereas the second (MeSC-01–MeSC-12) subset of using was retrieved twenty-two molecules SC-D as (MeZP-01–MeZP-22) a query molecule, whereas thewas ob- second tained subset from ZPD-2. molecules (MeZP-01–MeZP-22) was obtained from ZPD-2. of twenty-two MeSC-01 MeSC-02 MeSC-04 MeSC-05 MeSC-06 MeSC-07 MeSC-08 MeSC-09 MeSC-10 MeSC-12 MeZP-01 MeZP-02 MeZP-03 MeZP-04 MeZP-05 MeZP-06 MeZP-07 MeZP-08 MeZP-09 MeZP-10 MeZP-11 MeZP-12 MeZP-13 MeZP-14 Figure 2. Figure Chemical structures 2. Chemical structures of of compounds compounds resulting resulting from from the the VS VS employing employing SC-D SC-D (compounds (compounds MeSC-01–MeSC-12) MeSC-01–MeSC-12) andand ZPD-2 ZPD-2 (compounds (compounds MeZP-01–MeZP-22) MeZP-01–MeZP-22) as query compounds. 2.2. Thioflavin-T 2.2. Thioflavin-T Fluorescence Fluorescence Test Test The thirty-four The thirty-four candidates candidates were were purchased purchased from from aa SPECS SPECS vendor vendor and and tested tested in in the the thioflavin-T fluorescence test as a preliminary assay to measure their ability to reduce thioflavin-T fluorescence test as a preliminary assay to measure their ability to reduce α- α-Syn aggregation, as already reported for SC-D and ZPD-2 [8–10]. To evaluate the in- Syn aggregation, as already reported for SC-D and ZPD-2 [8–10]. To evaluate the inhibi- hibitory potential of the selected compounds, we monitored the increase in thioflavin-T tory potential of the selected compounds, we monitored the increase in thioflavin-T (Th- (Th-T) fluorescence emission during the aggregation kinetics of 70 µM α-Syn in the absence or presence of a 100 µM concentration of the molecules. SC-D was used as a reference compound. MeSC-04 emerged as the best candidate, with a Th-T fluorescence emission Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 5 of 14 Int. J. Mol. Sci. 2022, 23, 14844 5 of 14 T) fluorescence emission during the aggregation kinetics of 70 µM α-Syn in the absence or presence of a 100 µM concentration of the molecules. SC-D was used as a reference compound. reduction MeSC-04 of 69% (p