Difluoromethyl-1,3,4-oxadiazoles are slow-binding substrate analog inhibitors of histone deacetylase 6 with unprecedented isotype selectivity

Abstract

Histone deacetylase 6 (HDAC6) is an attractive drug development target because of its role in the immune response, neuropathy, and cancer. Knockout mice develop normally and have no apparent phenotype, suggesting that selective inhibitors should have an excellent therapeutic window. Unfortunately, current HDAC6 inhibitors have only moderate selectivity and may inhibit other HDAC subtypes at high concentrations, potentially leading to side effects. Recently, substituted oxadiazoles have attracted attention as a promising novel HDAC inhibitor chemotype, but their mechanism of action is unknown. Here, we show that compounds containing a difluoromethyl-1,3,4-oxadiazole (DFMO) moiety are potent and single-digit nanomolar inhibitors with an unprecedented greater than 104-fold selectivity for HDAC6 over all other HDAC subtypes. By combining kinetics, X-ray crystallography, and mass spectrometry, we found that DFMO derivatives are slow-binding substrate analogs of HDAC6 that undergo an enzyme-catalyzed ring opening reaction, forming a tight and long-lived enzyme–inhibitor complex. The elucidation of the mechanism of action of DFMO derivatives paves the way for the rational design of highly selective inhibitors of HDAC6 and possibly of other HDAC subtypes as well with potentially important therapeutic implications. Histone deacetylase 6 (HDAC6) is an attractive drug development target because of its role in the immune response, neuropathy, and cancer. Knockout mice develop normally and have no apparent phenotype, suggesting that selective inhibitors should have an excellent therapeutic window. Unfortunately, current HDAC6 inhibitors have only moderate selectivity and may inhibit other HDAC subtypes at high concentrations, potentially leading to side effects. Recently, substituted oxadiazoles have attracted attention as a promising novel HDAC inhibitor chemotype, but their mechanism of action is unknown. Here, we show that compounds containing a difluoromethyl-1,3,4-oxadiazole (DFMO) moiety are potent and single-digit nanomolar inhibitors with an unprecedented greater than 104-fold selectivity for HDAC6 over all other HDAC subtypes. By combining kinetics, X-ray crystallography, and mass spectrometry, we found that DFMO derivatives are slow-binding substrate analogs of HDAC6 that undergo an enzyme-catalyzed ring opening reaction, forming a tight and long-lived enzyme–inhibitor complex. The elucidation of the mechanism of action of DFMO derivatives paves the way for the rational design of highly selective inhibitors of HDAC6 and possibly of other HDAC subtypes as well with potentially important therapeutic implications. Zinc (Zn)-dependent histone deacetylases (HDACs) are a family of 11 evolutionarily related hydrolases that catalyze the removal of acyl residues from histones, non-histone proteins, and polyamines (1Shinsky S.A. Christianson D.W. Polyamine deacetylase structure and catalysis: prokaryotic acetylpolyamine amidohydrolase and eukaryotic HDAC10.Biochemistry. 2018; 57: 3105-3114Crossref PubMed Scopus (26) Google Scholar, 2Xu Y. Shi Z. Bao L. An expanding repertoire of protein acylations.Mol. Cell Proteomics. 2022; 21100193Abstract Full Text Full Text PDF Scopus (22) Google Scholar, 3Xu H. Zhou J. Lin S. Deng W. Zhang Y. Xue Y. Plmd: an updated data resource of protein lysine modifications.J. Genet. Genomics. 2017; 44: 243-250Crossref PubMed Scopus (152) Google Scholar). Given the involvement of HDACs in numerous diseases, the pharmaceutical industry is pursuing the development of HDAC inhibitors (HDACis) since almost 2 decades, an effort that led to the approval of five molecules for the treatment of cancer (4Bondarev A.D. Attwood M.M. Jonsson J. Chubarev V.N. Tarasov V.V. Schiöth H.B. Recent developments of HDAC inhibitors: emerging indications and novel molecules.Br. J. Clin. Pharmacol. 2021; 87: 4577-4597Crossref PubMed Scopus (134) Google Scholar). Unfortunately, the therapeutic benefit of HDACis has been limited by side effects because of the poor selectivity of these first-generation molecules, which inhibit several to all the Zn-dependent HDAC family members, affecting crucial physiological functions. HDAC6 stands out among the 11 human isoforms, as being the only one that has two homologous tandem catalytic domains (CD1 and CD2). Furthermore, HDAC6 is mainly localized in the cytoplasm, and its main substrates are various non-histone proteins, such as α-tubulin, Foxp3, Hsp90, β-catenin, cortactin, and peroxiredoxins. HDAC6 knockout mice are viable and fertile and do not show overt physiological dysfunctions (5Zhang Y. Kwon S. Yamaguchi T. Cubizolles F. Rousseaux S. Kneissel M. et al.Mice lacking histone deacetylase 6 have hyperacetylated tubulin but are viable and develop normally.Mol. Cell Biol. 2008; 28: 1688-1701Crossref PubMed Scopus (449) Google Scholar). Also, selective HDAC6 inhibition was demonstrated to be well tolerated in both preclinical species and in human clinical trials (6Amengual J.E. Lue J.K. Ma H. Lichtenstein R. Shah B. Cremers S. et al.First-in-Class selective HDAC6 inhibitor (ACY-1215) has a highly favorable safety profile in patients with relapsed and refractory lymphoma.Oncologist. 2021; 26: 184-e366Crossref PubMed Scopus (27) Google Scholar). Since HDAC6 plays a role in regulating immune response, in the development of neuropathies and in Alzheimer’s disease, the identification of highly specific HDAC6is is of great importance (7Zhang X.H. Qin-Ma Wu H.P. Khamis M.Y. Li Y.H. Ma L.Y. et al.A Review of progress in histone deacetylase 6 inhibitors research: structural specificity and functional diversity.J. Med. Chem. 2021; 64: 1362-1391Crossref PubMed Scopus (66) Google Scholar). The classical HDACi pharmacophore consists of a Zn-binding group (ZBG), interacting with the active site Zn ion, a cap, which interacts with the outer region of the enzyme, and a linker that connects the ZBG to the cap. Most HDACis have a hydroxamic acid moiety as ZBG, a chemical group that has inherent stability and safety issues. Recently, reports on oxadiazole-based HDAC6is (Fig. S1) as a potential alternative to hydroxamates appeared, but their mechanism of action is unclear (8He X. Li Z. Zhuo X.-T. Hui Z. Xie T. Ye X.-Y. Novel selective histone deacetylase 6 (HDAC6) inhibitors: a patent Review (2016-2019).Recent Pat. Anticancer. Drug Discov. 2020; 15: 32-48Crossref PubMed Scopus (21) Google Scholar). Focusing on difluoromethyl-1,3,4-oxiadiazole (DFMO)-containing HDAC6is, we have been able to identify molecules with unprecedented selectivity for HDAC6 over all other HDACs and excellent drug-like properties (9Marchini M. Vergani B. Sandrone G. Rocchio I. Kachkovskyi G. Caprini G. et al.2-(4-((5-(benzo[b]Thiophen-3-yl)-1H-Tetrazol-1-yl)methyl)phenyl)-5-(difluoromethyl)-1,3,4-Oxadiazole Delivatives and Similar Compounds as Selective Inhibitors of Histone Deacetylase 6 (HDAC6) for use in Treating e.g. Peripheral Neuropathy.2022Google Scholar). Using X-ray crystallography, molecular modeling, kinetics, and rapid chromatography coupled to mass spectrometry, we show that DMFO-containing HDACis are actually substrate analogs of HDAC6 that undergo an enzyme-catalyzed ring hydration leading to a tight and long-lived enzyme–inhibitor complex. Our findings open the way for the development of novel HDAC6 selective inhibitors for the treatment of neuropathies and other diseases, but they also point to the intriguing possibility to use oxadiazoles in the development of selective inhibitors of other HDAC subtypes. In our medicinal chemistry program, we identified potent and selective HDAC6is in the DFMO series using in silico modeling and traditional structure–activity relationship (article in preparation). The DFMO containing inhibitor 1 (Fig. S1) (9Marchini M. Vergani B. Sandrone G. Rocchio I. Kachkovskyi G. Caprini G. et al.2-(4-((5-(benzo[b]Thiophen-3-yl)-1H-Tetrazol-1-yl)methyl)phenyl)-5-(difluoromethyl)-1,3,4-Oxadiazole Delivatives and Similar Compounds as Selective Inhibitors of Histone Deacetylase 6 (HDAC6) for use in Treating e.g. Peripheral Neuropathy.2022Google Scholar), which was selected as an example for this communication, is at least four orders of magnitude more potent on HDAC6 than on any of the other HDACs (Table 1). This selectivity is remarkable when compared with hydroxamate-based HDAC6is like ACY-1215 (Ricolinostat), ACY-241 (Citarinostat), KA2507 (10Xiao Y. Zhang X. Recent advances in small molecular modulators targeting histone deacetylase 6.Futur. Drug Discov. 2020; 2: FDD53Crossref Google Scholar), and ITF3756 (11Vergani B. Sandrone G. Marchini M. Ripamonti C. Cellupica E. Galbiati E. et al.Novel benzohydroxamate-based potent and selective histone deacetylase 6 (HDAC6) inhibitors bearing a pentaheterocyclic scaffold: design, synthesis, and biological evaluation.J. Med. Chem. 2019; 62: 10711-10739Crossref PubMed Scopus (32) Google Scholar) (Fig. S2 and Table S1). Furthermore, we carried out inhibition measurements to investigate the interdomain selectivity of 1 using the isolated CD1 (zHDAC6-CD1) and CD2 (zHDAC6-CD2) domains of zebrafish HDAC6. In agreement with previous works (12Miyake Y. Keusch J.J. Wang L. Saito M. Hess D. Wang X. et al.Structural insights into HDAC6 tubulin deacetylation and its selective inhibition.Nat. Chem. Biol. 2016; 12: 748-754Crossref PubMed Scopus (210) Google Scholar, 13Hai Y. Christianson D.W. Histone deacetylase 6 structure and molecular basis of catalysis and inhibition.Nat. Chem. Biol. 2016; 12: 741-747Crossref PubMed Scopus (315) Google Scholar), when tested on commercial fluorogenic substrate Fluor De Lys Green, zHDAC6-CD1 exhibited a weaker but measurable activity compared with that of zHDAC6-CD2 (data not shown). More interestingly, we found that 1 is highly selective for the CD2 domain (Tables 1 and S1). Our data confirmed that the deacetylase activity of the full-length HDAC6 on commercial fluorogenic substrate Fluor De Lys Green mainly resides in CD2 and that HDAC6is selectively target CD2 for their activity. The only crystal structure of HDAC6 CD1 complexed with an HDAC6i has recently been reported (12Miyake Y. Keusch J.J. Wang L. Saito M. Hess D. Wang X. et al.Structural insights into HDAC6 tubulin deacetylation and its selective inhibition.Nat. Chem. Biol. 2016; 12: 748-754Crossref PubMed Scopus (210) Google Scholar). In this context, the unique K330 in CD1 has been reported to interact with bound inhibitors (12Miyake Y. Keusch J.J. Wang L. Saito M. Hess D. Wang X. et al.Structural insights into HDAC6 tubulin deacetylation and its selective inhibition.Nat. Chem. Biol. 2016; 12: 748-754Crossref PubMed Scopus (210) Google Scholar, 14Osko J.D. Christianson D.W. Structural basis of catalysis and inhibition of HDAC6 CD1, the enigmatic catalytic domain of histone deacetylase 6.Biochemistry. 2019; 58: 4912-4924Crossref PubMed Scopus (35) Google Scholar) and to confer specificity for the deacetylation of peptide substrates bearing C-terminal acetyl-lysine residues (13Hai Y. Christianson D.W. Histone deacetylase 6 structure and molecular basis of catalysis and inhibition.Nat. Chem. Biol. 2016; 12: 741-747Crossref PubMed Scopus (315) Google Scholar, 14Osko J.D. Christianson D.W. Structural basis of catalysis and inhibition of HDAC6 CD1, the enigmatic catalytic domain of histone deacetylase 6.Biochemistry. 2019; 58: 4912-4924Crossref PubMed Scopus (35) Google Scholar, 15Kutil Z. Skultetyova L. Rauh D. Meleshin M. Snajdr I. Novakova Z. et al.The unraveling of substrate specificity of histone deacetylase 6 domains using acetylome peptide microarrays and peptide libraries.FASEB J. 2019; 33: 4035-4045Crossref PubMed Scopus (27) Google Scholar). As with substrate preferences, binding of 1 and hydroxamate-based HDAC6is to the CD1 active site could be disfavored by suboptimal interactions with this key residue.Table 1Inhibitory profile of 1, 2, and 3 on HDACsHDAC classInhibitorsRIC50 (μM)Class IHDAC1>100>10014.7 ± 0.4HDAC2>100>10043.6 ± 1.8HDAC3>10046.8 ± 5.02.01 ± 0.09HDAC8>100>10010.2 ± 0.3Class IIbHDAC60.0077 ± 0.00031.58 ± 0.050.329 ± 0.006zHDAC6-CD20.0078 ± 0.00042.25 ± 0.040.097 ± 0.004zHDAC6-CD182.2 ± 7.7>10013.9 ± 0.6HDAC10>100>10013.30 ± 0.06Class IIaHDAC4>100>100>100HDAC5>100>10098.5 ± 7.2HDAC7>100>10064.2 ± 2.4HDAC9>100>10046.4 ± 0.9Class IVHDAC11>100>100>100 Open table in a new tab In the presence of 1, we noticed that inhibition increases during the time course of the assay, suggesting slow binding inhibition (Fig. S3A). Slow binding inhibition may result from various mechanisms (16Copeland R.A. Evaluation of enzyme inhibitors in drug discovery.Wiley. 2013; https://doi.org/10.1002/9781118540398Crossref Scopus (332) Google Scholar). The most common ones may be discerned by determining the dependence of the observed pseudo–first-order rate constant kobs of formation of the enzyme–inhibitor complex on the inhibitor concentration. We ruled out that the inhibitor itself underwent hydration–dehydration equilibria in aqueous solution with only one species binding to the enzyme (17Narjes F. Brunetti M. Colarusso S. Gerlach B. Koch U. Biasiol G. et al.α-Ketoacids are potent slow binding inhibitors of the hepatitis C virus NS3 protease.Biochemistry. 2000; 39: 1849-1861Crossref PubMed Scopus (75) Google Scholar), as we were unable to detect a hydrated form of 1 in solution. Thus, the observed linear dependence of kobs on inhibitor concentration is consistent with the binding process itself being slow, with an apparent association rate constant of 9.6 × 105 M−1 min−1 (Fig. and and Table The magnitude of the apparent rate constant was from the kobs and (Table 2 and and ± min−1 ± The time of the inhibitor was to of the inhibition of HDAC6 by ± ± in M−1 min−1 from kobs ± ± in min−1 from kobs ± ± in min−1 from ± from ± ± in of ± M−1 min−1 from kobs ± min−1 from kobs ± min−1 from ± from ± of ± M−1 min−1 from kobs from min−1 from ± of in from kobs from from Open table in a new tab The slow-binding was by 1 with both human full-length HDAC6 and zHDAC6-CD2 (Fig. the and are for both (Tables 2 and that zHDAC6-CD2 is a of human CD2 (12Miyake Y. Keusch J.J. Wang L. Saito M. Hess D. Wang X. et al.Structural insights into HDAC6 tubulin deacetylation and its selective inhibition.Nat. Chem. Biol. 2016; 12: 748-754Crossref PubMed Scopus (210) Google Scholar, 13Hai Y. Christianson D.W. Histone deacetylase 6 structure and molecular basis of catalysis and inhibition.Nat. Chem. Biol. 2016; 12: 741-747Crossref PubMed Scopus (315) Google Scholar, G. Marchini M. Vergani B. Caprini G. et of in the histone deacetylase 6 (HDAC6) selectivity of benzohydroxamate-based Med. Chem. 2021; PubMed Scopus Google Scholar). into the mechanism of inhibition of HDAC6 by 1 was with of the complex with the protein (13Hai Y. Christianson D.W. Histone deacetylase 6 structure and molecular basis of catalysis and inhibition.Nat. Chem. Biol. 2016; 12: 741-747Crossref PubMed Scopus (315) Google a of that no were by binding of the inhibitor in Table the of the bound was not with the inhibitor structure and that a DFMO ring opening mechanisms were and the were with the In this a moiety the to the (Fig. 1, and The could be as the result of two the binding of the the Zn could the of the a by the catalytic the catalytic mechanism of HDAC6 (13Hai Y. Christianson D.W. Histone deacetylase 6 structure and molecular basis of catalysis and inhibition.Nat. Chem. Biol. 2016; 12: 741-747Crossref PubMed Scopus (315) Google Scholar), the could be for a by The hydrated DFMO from such was (Fig. that this moiety is able to in formation with the and the the with The hydrated undergo a to an In to the in the crystal we an reaction, which could be enzyme in solution of the from the enzyme active this was enzyme the Zn to be by the of a The could the for the deacetylation and the of and In the crystal structure (Fig. 1, C and the group is to a with the as a with both catalytic key residues and The of the ZBG a with than a with the Zn ion, The bound to the ZBG is in the by and are interacting a effects in and Chem. PubMed Scopus Google Scholar), the is with the and the key (12Miyake Y. Keusch J.J. Wang L. Saito M. Hess D. Wang X. et al.Structural insights into HDAC6 tubulin deacetylation and its selective inhibition.Nat. Chem. Biol. 2016; 12: 748-754Crossref PubMed Scopus (210) Google Scholar, 13Hai Y. Christianson D.W. Histone deacetylase 6 structure and molecular basis of catalysis and inhibition.Nat. Chem. Biol. 2016; 12: 741-747Crossref PubMed Scopus (315) Google Scholar, G. Marchini M. Vergani B. Caprini G. et of in the histone deacetylase 6 (HDAC6) selectivity of benzohydroxamate-based Med. Chem. 2021; PubMed Scopus Google Scholar). interactions for 2017; PubMed Scopus Google Scholar) is and the of the interacts with and the is bound to the of residues in (12Miyake Y. Keusch J.J. Wang L. Saito M. Hess D. Wang X. et al.Structural insights into HDAC6 tubulin deacetylation and its selective inhibition.Nat. Chem. Biol. 2016; 12: 748-754Crossref PubMed Scopus (210) Google Scholar), the inhibitor interacts only with this suggesting able to interact with with the the of the inhibitor was to on the properties of the functional basis confirmed on and that the bound to the moiety a We 1 into zHDAC6-CD2 using the Scholar) of The that the DFMO moiety the Zn the to the group suggesting a the and the Zn The DFMO with its into the catalytic the moiety in an by and the catalytic In this the inhibitor interactions and and as well as interactions the and the the by the DFMO the Zn and a that is to a of a to detect a with the at a of from the DFMO was using the Scholar). The that the is in the Zn to both catalytic and with and the is the of the at a with a (Fig. The presence of a in the and of a ZBG that is than a hydroxamic acid was In to this be related to the high selectivity other modeling were with et into the mechanism of HDAC by 2016; Scopus Google as a of I. The with the (Fig. a in the Zn which is able to in two with the catalytic and the is to than to which is the of is observed in HDAC6 and In to suboptimal the of not in the with the a not for since this is in a The Chem. 2017; PubMed Scopus Google Scholar), an mainly on the The HDAC subtypes be in of a of the DFMO moiety in because of an the residues in the active site of HDAC6 this and an for An is when CD1 is in a complex with that both DFMO and molecules be in the catalytic the that the moiety into the CD1 catalytic suggesting an for the (Fig. both be observed in the complex (12Miyake Y. Keusch J.J. Wang L. Saito M. Hess D. Wang X. et al.Structural insights into HDAC6 tubulin deacetylation and its selective inhibition.Nat. Chem. Biol. 2016; 12: 748-754Crossref PubMed Scopus (210) Google Scholar) (Fig. of 1 with a key role of which to form a in able to This is not in CD1 because of the presence of the and side of K330 (13Hai Y. Christianson D.W. Histone deacetylase 6 structure and molecular basis of catalysis and inhibition.Nat. Chem. Biol. 2016; 12: 741-747Crossref PubMed Scopus (315) Google Scholar), which is to interact with inhibitors into a favorable binding (Fig. should be noticed that 1 could an related to the the lysine side and the moiety of both the (12Miyake Y. Keusch J.J. Wang L. Saito M. Hess D. Wang X. et al.Structural insights into HDAC6 tubulin deacetylation and its selective inhibition.Nat. Chem. Biol. 2016; 12: 748-754Crossref PubMed Scopus (210) Google Scholar) the to the DFMO moiety a of a leading to a by a able to the inhibitor in the catalytic of this 2 and 3 (Table were and as for mass Unfortunately, because of its chemical was not to the hydrated which is not from 2 by mass We 2 and 3 in more both compounds are potent but selective HDAC6is, with 2 a selectivity than 3 (Table 1). compounds are and inhibitors of HDAC6 (Fig. and all the other HDAC (data not shown). with 1 is into 2 and 3 (Fig. The observed rate constant for of 1 ± was to the apparent by (Table suggesting that the slow of a hydrated of 1 is actually The linear dependence of the observed rate of of inhibition be to formation of an enzyme–inhibitor which is by a high well the of Furthermore, 2 was into 3 by zHDAC6-CD2 (Fig. that both 1 and 2 are substrates of HDAC6. we that formation of 3 high of 2 (Fig. which are but are to in binding of 1 to the enzyme active site with formation of the complex. formation of the tight binding long-lived hydrated is not at high concentrations, 2 to enzyme and be to 3 and be 1, 2, and 3 are in and identify the long-lived chromatography M.M. for molecular mass compounds bound to using and mass to inhibitors of human 33: PubMed Scopus Google Scholar) with mass was (Fig. with 1, the species that with zHDAC6-CD2 was a hydrated form of 1 as which was in a over species were also in the containing 1 not with the Furthermore, the of 2 as is to zHDAC6-CD2 to its complex (Fig. these data that the complex with 1 itself and the 3 observed in the crystal structure are to be the long-lived enzyme–inhibitor complex Our data that the species is the hydrated (Fig. the in acyl 2 (Fig. In the the tight complex be by a hydrated in the active site of which is as an by the The possibility is that the tight complex is with 2 but only when in the active In formation of 2 from active 1 to a complex in which the has with the to the enzyme, 2 a which to be In this the to 3 may be over the of more into the of the slow of inhibition observed with 1, we in that the hydration of the DFMO ring is rate a should be on was when Green substrate was (Table we to an on the apparent association the rate of 1 (Table that than chemical are to be rate the we the and in the of is to the substrate and to the in the deacetylation of the acetyl-lysine and is reported to as a in the physiological (12Miyake Y. Keusch J.J. Wang L. Saito M. Hess D. Wang X. et al.Structural insights into HDAC6 tubulin deacetylation and its selective inhibition.Nat. Chem. Biol. 2016; 12: 748-754Crossref PubMed Scopus (210) Google Scholar, 13Hai Y. Christianson D.W. Histone deacetylase 6 structure and molecular basis of catalysis and inhibition.Nat. Chem. Biol. 2016; 12: 741-747Crossref PubMed Scopus (315) Google Scholar). The of inhibition with 1 (Table was and to be the rate constant was one of magnitude greater in the compared with the enzyme (Table are of a key role of in the observed and in the of the tight complex. on the of these key residues in DFMO we with both (Fig. The that of 1 and 2 at a rate when with zHDAC6-CD2 the that the are enzyme and and The of 1 with the are the result of the of formation of the tight complex the of a catalytic mechanism and Furthermore, also the of 2 into 3 was in this (Fig. we found that the DFMO ring of 1 was also by and of at a slow rate and at high inhibitor concentrations, leading to the formation of of both 2 and 3 (Fig. This that the of this of compounds is a common with the of HDAC active which could be potentially to develop novel inhibitors of other HDAC subtypes. 3 the the modeling, crystal and Our that are inhibitors of HDAC6 that with the enzyme as selective substrate leading to in hydration and formation of a tight by the of a The formation of a long-lived is at the of their the of our substrate inhibitors of hydrolases are We that inhibition of by to a mechanism and has led to the development of an drug M. S. et inhibition of human by formation of a potent Chem. Scopus Google Scholar), that this mechanism be have excellent in and in drug-like properties and are promising for the clinical development of HDAC6is for the treatment of several such as neuropathies and for the of immune are selective tubulin in are and in that are orders of magnitude than with hydroxamic (9Marchini M. Vergani B. Sandrone G. Rocchio I. Kachkovskyi G. Caprini G. et al.2-(4-((5-(benzo[b]Thiophen-3-yl)-1H-Tetrazol-1-yl)methyl)phenyl)-5-(difluoromethyl)-1,3,4-Oxadiazole Delivatives and Similar Compounds as Selective Inhibitors of Histone Deacetylase 6 (HDAC6) for use in Treating e.g. Peripheral Neuropathy.2022Google Scholar). In they are active in in of at data be

Topics & Concepts