Provided for non-commercial research and education use Not for reproduction, distribution or commercial use vL.52·ne2012*BN0235234 EUROPEAN JOURNAL OF MEDICINAL CHEMISTRY This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article(e.g in Word or Tex form)to their personal website or institutional repository. Authors requiring further information regarding Elsevier's archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright
This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright
Author's personal copy European Journal of Medicinal Chemistry 52(2012)33-43 Contents lists available at SciVerse Science Direct European Journal of Medicinal Chemistry ELSEVIER journalhomepagehttp://www.elsevier.com/locatelejmech Original article Discovery of flavonoid derivatives as anti-HCV agents via pharmacophore search combining molecular docking strategy Ming-Ming Liu, Lu Zhou, Pei-Lan He Yi-Nan Zhang Jia-Yi Zhou, Qing Shen Xin-Wen Chen Jian-Ping Zuo b, Wei Lia, * De-Yong Ye a tment of Medicinal Ch ol of pharmacy, Fudan University, 826 Zhangheng Rd, Shanghai 201203, China hese Academy of sciences, 555 Zuchongzhi Rd, Shanghai 201203, china Institute of virology, demy of sciences, Wuhan 430071, China A RTICLE INFO A BSTRACT ommon feature based pharmacophore and structure-based docking approaches have been employed in he identification of novel anti-HCV candidates from our in-house database. a total of 31 hits identified in Received vised form 29 February 2012 Accepted 1 March 2012 including two naturally occurring flavones apigenin(21)and luteolin(22) with low micromole ECso Available online 8 March 2012 values and three compounds(23, 24 and 25)of novel scaffolds with moderate potencies. In addition, harmacophore refinement was also conducted based on the current knowledge of flavone -derived anti- lCV candidates and the results of combined in silico and in vitro assays Anti-HCV agents 2012 Elsevier Masson SAS. All rights reserved. Ligand-based pharmacophore Virtual screening 1. Introduction a complementary minus-strand RNA, using the genome as emplate, and the subsequent synthesis of genomic plus-strand Hepatitis C virus(HCv) infection remains to be an important RNA from this minus-strand RNA template [5]. Since NS5B is health-care problem and has been identified as a leading cause of crucial for viral infectivity, it has been recognized as a promising chronic hepatitis, liver cirrhosis and hepatocellular carcinoma [ 1. It and validated target for HCV therapies [6] is estimated that a minimum of 3% of the worlds population(about a,Y-Diketoacids(DKAs) were initially revealed to be selective 180 million people) are chronically infected, with additional and reversible inhibitors against NS5B through high throughput approximate 3-4 million new cases of HCV infection each year [2]. screening(HTS)approaches [7. Follow-up works demonstrated The current care standard for HCV, composed of Pegylated Inter- these compounds may serve as natural substrate UTP mimics which feron-a and ribavirin, can only achieve a sustained viral response in bind to the active site of NS5B [8]. However, due to poor physico- less than 50% of patients infected with the predominant genotype 1 chemical properties of these compounds, many DKA analogs or 3. Furthermore, patients often intolerant to the serious mimics with NS5B inhibitory activities have been designed and adverse reactions of flu-like symptoms, depression, anemia, which prepared[8-13 Moreover, the replacement of DKA scaffold with might lead to poor treatment compliance[4. It is urgent to develop naturally occurring flavonoid by scaffold hopping strategies, which novel anti-HCV agents with improved efficiency and minimal side to the discovery of galangin derivatives, were am promising candidates(Fig. 1)[14-16]. The RNA-dependent RNa polymerase(RdRp) of HCv, also In this study, considering the high interest of developing novel known as protein NS5B, is a key enzyme for the synthesis of anti-HCV chemical candidates, a HipHop pharmacophore model, established from 8 reference ns5b inhibitors, was used as a filtrat ing tool to screening our in-house database. Then 246 hits were s Corresponding authors. Tel: +86 21 51980117: fax: +86 21 51980114. further screened by NS5B structural-based docking, among which author 31 hits were identified. Finally. 20 compounds out of these hits E-mail addresses: jpzuoemailshcnc accn (.-P. Zuo), wei-liefudaneduan were found be novel anti-HCV candidates through in vitro assays (Fig. 2). More, the pharmacophore for flavonoid-like analogs with 0223-52345 front matter 2012 Elsevier Masson SAS. All rights reserved. 201203.002
Author's personal copy Original article Discovery of flavonoid derivatives as anti-HCV agents via pharmacophore search combining molecular docking strategy Ming-Ming Liu a , Lu Zhou a , Pei-Lan He b , Yi-Nan Zhang a , Jia-Yi Zhou a , Qing Shen a , Xin-Wen Chen c , Jian-Ping Zuo b,**, Wei Li a,*, De-Yong Ye a,* aDepartment of Medicinal Chemistry, School of Pharmacy, Fudan University, 826 Zhangheng Rd, Shanghai 201203, China b Shanghai Institute of Materia Medica, Chinese Academy of Sciences, 555 Zuchongzhi Rd, Shanghai 201203, China c Wuhan Institute of Virology, Chinese Academy of Sciences, Wuhan 430071, China article info Article history: Received 18 August 2011 Received in revised form 29 February 2012 Accepted 1 March 2012 Available online 8 March 2012 Keywords: Anti-HCV agents NS5B polymerase Flavonoids Ligand-based pharmacophore Virtual screening abstract Common feature based pharmacophore and structure-based docking approaches have been employed in the identification of novel anti-HCV candidates from our in-house database. A total of 31 hits identified in silico were screened in vitro assay. 20 Compounds demonstrated anti-HCV activities (EC50 < 50 mM), including two naturally occurring flavones apigenin (21) and luteolin (22) with low micromole EC50 values and three compounds (23, 24 and 25) of novel scaffolds with moderate potencies. In addition, pharmacophore refinement was also conducted based on the current knowledge of flavone-derived antiHCV candidates and the results of combined in silico and in vitro assays. 2012 Elsevier Masson SAS. All rights reserved. 1. Introduction Hepatitis C virus (HCV) infection remains to be an important health-care problem and has been identified as a leading cause of chronic hepatitis, liver cirrhosis and hepatocellular carcinoma [1]. It is estimated that a minimum of 3% of the world’s population (about 180 million people) are chronically infected, with additional approximate 3e4 million new cases of HCV infection each year [2]. The current care standard for HCV, composed of Pegylated Interferon-a and Ribavirin, can only achieve a sustained viral response in less than 50% of patients infected with the predominant genotype 1 [3]. Furthermore, patients are often intolerant to the serious adverse reactions of flu-like symptoms, depression, anemia, which might lead to poor treatment compliance [4]. It is urgent to develop novel anti-HCV agents with improved efficiency and minimal side effects. The RNA-dependent RNA polymerase (RdRp) of HCV, also known as protein NS5B, is a key enzyme for the synthesis of a complementary minus-strand RNA, using the genome as template, and the subsequent synthesis of genomic plus-strand RNA from this minus-strand RNA template [5]. Since NS5B is crucial for viral infectivity, it has been recognized as a promising and validated target for HCV therapies [6]. a,g-Diketoacids (DKAs) were initially revealed to be selective and reversible inhibitors against NS5B through high throughput screening (HTS) approaches [7]. Follow-up works demonstrated these compounds may serve as natural substrate UTP mimics which bind to the active site of NS5B [8]. However, due to poor physicochemical properties of these compounds, many DKA analogs or mimics with NS5B inhibitory activities have been designed and prepared [8e13]. Moreover, the replacement of DKA scaffold with naturally occurring flavonoid by scaffold hopping strategies, which led to the discovery of galangin derivatives, were among the most promising candidates (Fig. 1) [14e16]. In this study, considering the high interest of developing novel anti-HCV chemical candidates, a HipHop pharmacophore model, established from 8 reference NS5B inhibitors, was used as a filtrating tool to screening our in-house database. Then 246 hits were further screened by NS5B structural-based docking, among which 31 hits were identified. Finally, 20 compounds out of these hits were found be novel anti-HCV candidates through in vitro assays (Fig. 2). More, the pharmacophore for flavonoid-like analogs with * Corresponding authors. Tel.: þ86 21 51980117; fax: þ86 21 51980114. ** Corresponding author. E-mail addresses: jpzuo@mail.shcnc.ac.cn (J.-P. Zuo), wei-li@fudan.edu.cn (W. Li), dyye@shmu.edu.cn (D.-Y. Ye). Contents lists available at SciVerse ScienceDirect European Journal of Medicinal Chemistry journal homepage: http://www.elsevier.com/locate/ejmech 0223-5234/$ e see front matter 2012 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.ejmech.2012.03.002 European Journal of Medicinal Chemistry 52 (2012) 33e43
Author's personal copy M-M. Liu er aL/European Jourmal of Medicinal Chemistry 52(2012)33-43 Considering the more druggable properties of the three flavonoids OH (compounds 5, 7 and 8)in the training set, 'principal values of 2 and 'Max-Omit-Feat values of o were assigned to these compounds, while principal'andMax-Omit-Feat values were set 1 for the other 5 compounds. Ten hypotheses(Hypos)were generated and scored as shown in Table 1. Considering it could both attain the highest score of 70.367 and accept all compounds in the Aryldiketoacid(DKA) training set statistically well enough, the hypo l was selected and was used for further validation. On the other hand. other estab- Fig 1. The structures of Aryldiketoacid(aryl-DKA)and its bioisostere galangin(mimic lished hypotheses mapped poorly to at least one component of substructure was shown in thick lines). training set(fit value <1). The Hypo 1 has four features, namely one hydrophobic feature(H), one H-bond donor(D) and two H-bond anti-HCV activities has been refined based on ligand-based and scepters(Al and a2)(Fig 4A)and compound 5 was mapped onto structure-based approaches. Hypo 1 with highest fit values of 3.99(Fig 4B). Hypo 1 was further validated by the goodness of hit(GH)scorir method [ 19, 20]. An external database of decoy set, which was use 2. Results and discussion for pharmacophore validation, was made up of other 40 indepen dent active and 1000 inactive compounds. 38 Positive compounds 2. 1. Establishment and validation of the ligand-based were successfully identified among total 55 hits and a set of statistical parameters, such as yield of actives, ratio of actives, enrichment factor(EF)and GH scores, were presented in Table 2. Known as a powerful tool to identify novel compounds with Thus the validated Hypo 1 was qualified to conduct virtual similar biological activities, the pharmacophores could be devel- screening [21] od [17]. As shown in Fig. 3, the tra set which was used to establish the pharmacophore was composed of eight NS5B inhibitors selected from literature 9, 10, 12-16 2.2. Virtual screening and molecular docking cording to the following criteria: 1. they should share certain structural diversity: 2. they should be the most active compounds Our in-house database, in which 15,568 commercially available dentified in each series: 3. they should be visually examined to natural products with searchable 3D structure were collected, was contain similar pharmacophore components in order to ensure screened by Hypol to discover potential anti-HCV candidates. 246 their similar binding models against NS5B. Due to limited activity Compounds were initially identified and most of which scale (<2 log units) and set size of the training set, the HipHop flavonoids and flavonoids glycosides. Moreover, in order to mini- module available in Discovery Studio(DS)[18 was adaptively used. mize the number of hits and to maximize the probability of positive Training set HipHop In house database(15568 Common features model (Hypo 1) Pharmacophore based virtual screening Decoy set validation Validated Initial hits(246 Pharmacophore Docking based refinement Purchased(31) Biological activity assay Active hits(20) Fig. 2. Virtual screening flow chart
Author's personal copy anti-HCV activities has been refined based on ligand-based and structure-based approaches. 2. Results and discussion 2.1. Establishment and validation of the ligand-based pharmacophore Known as a powerful tool to identify novel compounds with similar biological activities, the pharmacophores could be developed by ligand-based method [17]. As shown in Fig. 3, the training set which was used to establish the pharmacophore was composed of eight NS5B inhibitors selected from literature [9,10,12e16] according to the following criteria: 1. they should share certain structural diversity; 2. they should be the most active compounds identified in each series; 3. they should be visually examined to contain similar pharmacophore components in order to ensure their similar binding models against NS5B. Due to limited activity scale (<2 log units) and set size of the training set, the HipHop module available in Discovery Studio (DS) [18] was adaptively used. Considering the more druggable properties of the three flavonoids (compounds 5, 7 and 8) in the training set, ‘principal’ values of 2 and ‘Max-Omit-Feat’ values of 0 were assigned to these compounds, while ‘principal’ and ‘Max-Omit-Feat’ values were set 1 for the other 5 compounds. Ten hypotheses (Hypos) were generated and scored as shown in Table 1. Considering it could both attain the highest score of 70.367 and accept all compounds in the training set statistically well enough, the Hypo 1 was selected and was used for further validation. On the other hand, other established hypotheses mapped poorly to at least one component of training set (fit value <1). The Hypo 1 has four features, namely one hydrophobic feature (H), one H-bond donor (D), and two H-bond accepters (A1 and A2) (Fig. 4A) and Compound 5 was mapped onto Hypo 1 with highest fit values of 3.99 (Fig. 4B). Hypo 1 was further validated by the goodness of hit (GH) scoring method [19,20]. An external database of decoy set, which was used for pharmacophore validation, was made up of other 40 independent active and 1000 inactive compounds. 38 Positive compounds were successfully identified among total 55 hits and a set of statistical parameters, such as yield of actives, ratio of actives, enrichment factor (EF) and GH scores, were presented in Table 2. Thus the validated Hypo 1 was qualified to conduct virtual screening [21]. 2.2. Virtual screening and molecular docking Our in-house database, in which 15,568 commercially available natural products with searchable 3D structure were collected, was screened by Hypo1 to discover potential anti-HCV candidates. 246 Compounds were initially identified and most of which were flavonoids and flavonoids glycosides. Moreover, in order to minimize the number of hits and to maximize the probability of positive Fig. 1. The structures of Aryldiketoacid (aryl-DKA) and its bioisostere galangin (mimic substructure was shown in thick lines). Fig. 2. Virtual screening flow chart. 34 M.-M. Liu et al. / European Journal of Medicinal Chemistry 52 (2012) 33e43
Author's personal copy M.-M. Liu et al./ European Journal of Medicinal Chemistry 52(2012)33-43 COOH 5 8 3. The chemical structures of NS5B inhibitors collected as training set(compounds 1-8). NS5B using GOLD docking software[22]. According to the results of thought to be due to their poor abilities to cross cell membranes. All docking scores and conformational cluster analysis, 31 compounds the compounds showed low cytotoxic activity(CC50>50 uM)as were demonstrated to form favorable interactions with relevant shown by MTT methods. residues on the enzyme. All these compounds were identified as Since the ligand-based pharmacophore and the structure-based the hits of the combined ligand-based pharmacophore and docking strategies were extracted from NS5B inhibitors and NS5B receptor-docking based virtual screening. In addition, it is notice- crystal structure respectively, the most active compound 22, and able that there were several new scaffolds bearing some differences the flavone compound 18, which has no hydroxyl group in B ring. to the training set molecules in the final selected compounds, such were also evaluated their inhibition of NS5B polymerase enzymatic (24 thraquinone(25)(Table 3). function according to the reference method [23]. The results showed that compound 22 exhibited good inhibition with an ICso of 2.3. Biological activities assay 112M st the enzyme NS5B, while compound 18 on exhibited 30% inhibition rate at 100 uM. This result could expla The selected 31 hits were evaluated anti-HCV activities in cell- the mechanisms of action for the anti-HCV activities, that is, the says in vitro Cell-based antiviral assay was performed in inhibition of NS5B, the key enzyme in the process of viral infectivity uthentic HCV infection replication system in the human hep toma cell lines, Huh7. The ECso values, as well as relevant cytotox activities, the CCso values of the 31 hit compounds, are summarized in Table 3 2.4. Structure-activity relationships(SARs)studies: mapping Hypo 20 Compounds showed significant activities against HCV 1 to the docking models and insights to pharmacophore refine (EC50 50 uM). Among these compounds, 21 (apigenin) and 22 (luteolin) possessed the highest inhibitory potencies with EC50 cell-based anti-HCV activities with flavonoids 5(EC50=5 uM).7 values of 7. 9 HM and 4.3 uM respectively. In order to validate their antiviral activities, we also examined the effect of compound (EC50=2 HM)and 8(EC50=2.0 uM)in the training set but a lit and 22 on HCV replication using a HCV subgenomic replicon cell tle weaker than DKa derivatives 1(EC50=0.82 uM)and 2 culture system. In replicon system, compounds 21 and 22 su EC50=0.54 HM). The anti-HCV activities of compound 3, 4 and pressed HCv replication with the ECso values of 12.0 HM and 6.6 uM were not mentioned in the original references. This outcome respectively (Table 4). Interestingly, we also found that several new encouraged o investigate the physiochemical property scaffolds 23, 24 and 25 from the structural classification of iso- complementarily between the pharmacophore(Hypo 1)of NS5B flavone, chalcone and anthraquinone, also had moderate antiviral inhibitors and the native binding pocket of NS5B polymerase, as activities. These new skeletal compounds can be optimized by well as the structure-activities(SARs)of these flavonoids.As the flavonoid glycosides(9-11)and biflavonoids(12) did not show binding pocket of NS5B based on the docked complex of the most complementary interactions were observed between the pharm Table 1 cophore groups and the residues involved. Clearly the H-bond The scores of common feature hypotheses(HipHop acceptor feature(Al)of Hypo 1 corresponds to the H-bond inte action between Arg 48, Cys 223 and the 4-carbonyl of flavonoids, 70.367 111111111 nd the H-bond donor feature(d)could be elucidated by the H- bond between the 7-hydroxyl of flavonoids and Asp 220, Asp 318. 111111111 Meanwhile, the hydrophobic feature(H)which was composed of the aromatic B ring of flavonoids located in a hydrophobic region 101111111 111111111 surrounded by hydrophobic residues of Val 52 and Leu 159. However, another H-bond acceptor feature(A2)in the pharmac 0061 111111111 hore, which is formed by the 1-oxygen atom of flavonoids, could 996 ot form effective hydrogen bond with Ser 226 since the distance 49.14 111111111 hydrophobic: D, H-bond donor: A, H-bond acceptor was too long. In addition, the 3-hydroxyl group of compound 22 B ring could form another H-bond with Tyr 4 and Ser 226 while the
Author's personal copy compounds, these compounds were docked into the active site of NS5B using GOLD docking software [22]. According to the results of docking scores and conformational cluster analysis, 31 compounds were demonstrated to form favorable interactions with relevant residues on the enzyme. All these compounds were identified as the hits of the combined ligand-based pharmacophore and receptor-docking based virtual screening. In addition, it is noticeable that there were several new scaffolds bearing some differences to the training set molecules in the final selected compounds, such as isoflavone (23), chalcone (24) and anthraquinone (25) (Table 3). 2.3. Biological activities assay The selected 31 hits were evaluated anti-HCV activities in cellbased assays in vitro. Cell-based antiviral assay was performed in authentic HCV infection/replication system in the human hepatoma cell lines, Huh7. The EC50 values, as well as relevant cytotoxic activities, the CC50 values of the 31 hit compounds, are summarized in Table 3. 20 Compounds showed significant activities against HCV (EC50 50 mM) as shown by MTT methods. Since the ligand-based pharmacophore and the structure-based docking strategies were extracted from NS5B inhibitors and NS5B crystal structure respectively, the most active compound 22, and the flavone compound 18, which has no hydroxyl group in B ring, were also evaluated their inhibition of NS5B polymerase enzymatic function according to the reference method [23]. The results showed that compound 22 exhibited good inhibition with an IC50 of 1.12 mM against the enzyme NS5B, while compound 18 only exhibited 30% inhibition rate at 100 mM. This result could explain the mechanisms of action for the anti-HCV activities, that is, the inhibition of NS5B, the key enzyme in the process of viral infectivity and replication. 2.4. Structureeactivity relationships (SARs) studies: mapping Hypo 1 to the docking models and insights to pharmacophore refinement The most active compounds, 21 and 22 showed similar cell-based anti-HCV activities with flavonoids 5 (EC50 ¼ 5 mM), 7 (EC50 ¼ 2 mM) and 8 (EC50 ¼ 2.0 mM) in the training set but a little weaker than DKA derivatives 1 (EC50 ¼ 0.82 mM) and 2 (EC50 ¼ 0.54 mM). The anti-HCV activities of compound 3, 4 and 6 were not mentioned in the original references. This outcome encouraged us to investigate the physiochemical property complementarily between the pharmacophore (Hypo 1) of NS5B inhibitors and the native binding pocket of NS5B polymerase, as well as the structureeactivities (SARs) of these flavonoids. As shown in Fig. 5, the pharmacophore of Hypo 1 was mapped to the binding pocket of NS5B based on the docked complex of the most active compound 22 and NS5B polymerase. The following three complementary interactions were observed between the pharmacophore groups and the residues involved. Clearly the H-bond acceptor feature (A1) of Hypo 1 corresponds to the H-bond interaction between Arg 48, Cys 223 and the 4-carbonyl of flavonoids, and the H-bond donor feature (D) could be elucidated by the Hbond between the 7-hydroxyl of flavonoids and Asp 220, Asp 318. Meanwhile, the hydrophobic feature (H) which was composed of the aromatic B ring of flavonoids located in a hydrophobic region surrounded by hydrophobic residues of Val 52 and Leu 159. However, another H-bond acceptor feature (A2) in the pharmacophore, which is formed by the 1-oxygen atom of flavonoids, could not form effective hydrogen bond with Ser 226 since the distance was too long. In addition, the 30 -hydroxyl group of compound 22 B ring could form another H-bond with Tyr 4 and Ser 226 while the Fig. 3. The chemical structures of NS5B inhibitors collected as training set (compounds 1e8). Table 1 The scores of common feature hypotheses (HipHop). No. Features Score Direct hit Fit 1 HDAA 70.367 111111111 4 2 HDAA 68.886 111111111 4 3 HAAA 67.086 111111111 4 4 HDAA 63.537 101111111 4 5 HAAA 61.737 101111111 4 6 DAA 59.027 111111111 3 7 HDA 50.061 111111111 3 8 HDA 50.061 111111111 3 9 DAA 49.966 101111111 3 10 AAA 49.147 111111111 3 H, hydrophobic; D, H-bond donor; A, H-bond acceptor. M.-M. Liu et al. / European Journal of Medicinal Chemistry 52 (2012) 33e43 35
Author's personal copy M-M. Liu et aL/European Jourmal of Medicinal Chemistry 52(2012)33-43 604 A1 6483 D 8607 A 2 of NS5B inhibitors generated by HipHop (Hypo 1)(A)3D spatial relationship and geometric parameters of Hypo 1. (B)Hypo 1 mapping with the best ound 5. Pharmacophore are color-coded with light-blue for hydrophobic feature, green for H-bond acceptor and pink for H-bond donor. For interpretation of the to color in this figure the reader is referred to the web version of this article. 4-hydroxyl of the flavonoids forms H-bond with Tyr 4. This kind of convincing interactions between NS5B and flavonoids as its H-bond interaction was common in the docking models between Hence, Hypo 2 was rationalized and eligible for both virtual specific residues and our flavonoid hit g and interaction mechanism elucidation purposes. It group in the B ring. be used as filter to perform further screening for the Furthermore, compared with other flavonoids without hydroxyl discovery more anti-HCV drug candidates. group in B ring such as compound 18,CoI 22 exhibited out 10-fold increase in anti-HCv activities due to the contribu- tion of 3 -and 4-hydroxyl groups in C ring. We also observed the 3. Conclusion similar results in NS5B enzymatic inhibition assay. The corre- spondence between the cell-based antiviral activity and the In summary, two different computational strategies, namely enzyme inhibition indicated that it was reasonable to assume that common feature based pharmacophore and structure-based docking study, have been applied to identify novel anti-HCV the cell-based anti-HCV activities of these hits from virtual candidates. A NS5B inhibitors' pharmacophore(Hypo 1)was creening correlate with their NS5B inhibitory activities. Besides, the observation that compound 22 was about 2-fold more potent developed and validated. It was used as a filtering tool to screen our than 21 could be also rationalized. These biological activities data were further evaluated by another independent receptor-docking hat groups I essential to the anti-HCV activities and could be recognized as approaches. Finally, 31 hit compounds identified in silico were a novel type of interaction between NS5B polymerase and purchased from commercial resources and assayed in vitro. Among lavonoid-related inhibitors which was not revealed by Hypo 1. these compounds, 20 compounds demonstrated anti-HCV activities Based on molecular modeling and biological assay evidence (EC50< 50 HM). Meanwhile, two naturally occurring flavones api genin(21)and luteolin(22)were found to be highly potent at low mentioned above, the Hypo 1 should be further refined by micromole ECso values, while another three compounds(23, 24 and adding an additional H-bond donor feature( D2)located in the 4 hydroxyl of flavonoids B rings, while the A2 feature in previous 25)with novel scaffolds were also moderately active. Moreover. Hypo 1 was removed. The refined pharmacophore hypothes compound 22 also exhibited good inhibition against the enzyme NS5B. In addition, another refined pharmacophore Hypo 2 for NS5B (Hypo 2)was developed, as shown in Fig. 6. It fit the compounds inhibitors has been developed based on our original SARs(struc well in our virtual screening hits and was considered to reflect ture-activity relationships )and molecular modeling analysis. This pharmacophore could be used for the structural optimi known NS5B inhibitors, as well as a powerful tool to iden a Table 2 NS5B inhibitors tatistical parameters and scores of enrichment study for validation of Hypo 1. Parameters values 4. Materials and methods Total molecules in database(D 1040 otal no. of actives in database(A) 4. 1. Establishment and validation of ligand-based pharmacophore 8 DKAs and DKA analogs( Fig. 1)with known NS5B inhibitory Ratio of actives [( Ha A)x 1001 nrichment factor(E)I(Ha x D)(Ht x A) activities were selected from literature [ 9, 10, 12-16] as the training establish ligand-based pharmacophores. All these False positive [Ht-Hal 17 compounds were initially sketched in MDL-ISIS/Draw[24, and 0.74 converted into 3D structures in SYBYL 6.9[25] with all proton and ( Ha/(4 x Ht x A)lx(3xA+Ht)x[1-(Ht- Ha)/(D-A)n) MMFF94 charges added. No formal charges were changed and GH score 0.6-0.8 indicates a very good model. these compounds were all minimized by Tripos Force Field in its
Author's personal copy 40 -hydroxyl of the flavonoids forms H-bond with Tyr 4. This kind of H-bond interaction was common in the docking models between these specific residues and our flavonoid hits bearing a hydroxyl group in the B ring. Furthermore, compared with other flavonoids without hydroxyl group in B ring such as compound 18, compound 22 exhibited about 10-fold increase in anti-HCV activities due to the contribution of 30 - and 40 -hydroxyl groups in C ring. We also observed the similar results in NS5B enzymatic inhibition assay. The correspondence between the cell-based antiviral activity and the enzyme inhibition indicated that it was reasonable to assume that the cell-based anti-HCV activities of these hits from virtual screening correlate with their NS5B inhibitory activities. Besides, the observation that compound 22 was about 2-fold more potent than 21 could be also rationalized. These biological activities data demonstrated that the hydroxyl groups in flavonoid B ring were essential to the anti-HCV activities and could be recognized as a novel type of interaction between NS5B polymerase and flavonoid-related inhibitors which was not revealed by Hypo 1. Based on molecular modeling and biological assay evidences mentioned above, the Hypo 1 should be further refined by adding an additional H-bond donor feature (D2) located in the 40 - hydroxyl of flavonoids B rings, while the A2 feature in previous Hypo 1 was removed. The refined pharmacophore hypothesis (Hypo 2) was developed, as shown in Fig. 6. It fit the compounds well in our virtual screening hits and was considered to reflect more convincing interactions between NS5B and flavonoids as its ligands. Hence, Hypo 2 was rationalized and eligible for both virtual screening and interaction mechanism elucidation purposes. It would be used as filter to perform further screening for the discovery more anti-HCV drug candidates. 3. Conclusion In summary, two different computational strategies, namely common feature based pharmacophore and structure-based docking study, have been applied to identify novel anti-HCV candidates. A NS5B inhibitors’ pharmacophore (Hypo 1) was developed and validated. It was used as a filtering tool to screen our in-house database with 246 compounds yielded. These compounds were further evaluated by another independent receptor-docking approaches. Finally, 31 hit compounds identified in silico were purchased from commercial resources and assayed in vitro. Among these compounds, 20 compounds demonstrated anti-HCV activities (EC50 < 50 mM). Meanwhile, two naturally occurring flavones apigenin (21) and luteolin (22) were found to be highly potent at low micromole EC50 values, while another three compounds (23, 24 and 25) with novel scaffolds were also moderately active. Moreover, compound 22 also exhibited good inhibition against the enzyme NS5B. In addition, another refined pharmacophore Hypo 2 for NS5B inhibitors has been developed based on our original SARs (structureeactivity relationships) and molecular modeling analysis. This pharmacophore could be used for the structural optimization of known NS5B inhibitors, as well as a powerful tool to identify novel NS5B inhibitors. 4. Materials and methods 4.1. Establishment and validation of ligand-based pharmacophore 8 DKAs and DKA analogs (Fig. 1) with known NS5B inhibitory activities were selected from literature [9,10,12e16] as the training set to establish ligand-based pharmacophores. All these compounds were initially sketched in MDL-ISIS/Draw [24], and converted into 3D structures in SYBYL 6.9 [25] with all proton and MMFF94 charges added. No formal charges were changed and these compounds were all minimized by Tripos Force Field in its Fig. 4. Pharmacophore model of NS5B inhibitors generated by HipHop (Hypo 1). (A) 3D spatial relationship and geometric parameters of Hypo 1. (B) Hypo 1 mapping with the best fit compound 5. Pharmacophore features are color-coded with light-blue for hydrophobic feature, green for H-bond acceptor and pink for H-bond donor. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) Table 2 Statistical parameters and scores of enrichment study for validation of Hypo 1. No. Parameters Values 1 Total molecules in database (D) 1040 2 Total no. of actives in database (A) 40 3 Total hits (Ht) 55 4 Active hits (Ha) 38 5 % Yield of actives [(Ha/Ht) 100] 69.1 6 % Ratio of actives [(Ha/A) 100] 95.0 7 Enrichment factor (E) [(Ha D)/(Ht A)] 18.0 8 False negatives [A Ha] 2 9 False positive [Ht Ha] 17 10 Goodness of hit (GH)a,b 0.74 a {[Ha/(4 Ht A)] (3 A þ Ht) [1 (Ht Ha)/(D A)]}. b GH score 0.6e0.8 indicates a very good model. 36 M.-M. Liu et al. / European Journal of Medicinal Chemistry 52 (2012) 33e43
Author's personal copy M.-M. Liu et al./ European Joumal of Medicinal Chemistry 52(2012)33-43 Anti-HCV activities and cytotoxicities for 31 hit compounds by virtual screening Structure CCso ( ul O、O HO Flavone glycoside Flavonol glycoside OH O OH OH 11 HO OH O O OH OH >50 Flavone 33.1 OH Flavone OH O (continued on next pag
Author's personal copy Table 3 Anti-HCV activities and cytotoxicities for 31 hit compounds by virtual screening. Compound Structure Classification EC50 (mM)a,b CC50 (mM) 9 Flavone glycoside NAc >50 10 Flavonol glycoside NA >50 11 Flavone glycosides NA >50 12 Biflavonoids NA >50 13 Flavone 33.1 >50 14 Flavone 36.3 >50 (continued on next page) M.-M. Liu et al. / European Journal of Medicinal Chemistry 52 (2012) 33e43 37
Author's personal copy M-M. Liu et aL/European Jourmal of Medicinal Chemistry 52(2012)33-43 Table 3(continued) Compound Classification ECso(uM)b HO OH HO 16 OH O OH 17 O Flavone 38.5 flavone OH Flavone HO OH 21
Author's personal copy Table 3 (continued ) Compound Structure Classification EC50 (mM)a,b CC50 (mM) 15 Flavone 32.2 >50 16 Flavone 33.9 >50 17 Flavone 38.5 >50 18 O OH O HO Flavone 50.97%d >50 19 O O HO OH OH OH Flavone 23.3 >50 20 O O OH HO HO Flavone 51.9%d >50 21 O O HO OH OH Flavone 7.9 >50 38 M.-M. Liu et al. / European Journal of Medicinal Chemistry 52 (2012) 33e43
Author's personal copy M.-M. Liu et al./ European Joumal of Medicinal Chemistry 52(2012)33-43 Table 3(continued) Compound Structure ECso (uM) OH HO 144 Chalcone 36.1 HO OH Anthraquinone >50 HO Flavonol 519 H OH 47.29 OH HO OH Flavonol HO Flavonol 4691a OH (continued on next page)
Author's personal copy Table 3 (continued ) Compound Structure Classification EC50 (mM)a,b CC50 (mM) 22 O O HO OH OH OH Flavone 4.7 >50 23 O O HO OH OH Isoflavone 14.4 >50 24 HO OH OH O Chalcone 36.1 >50 25 OH O OH OH O Anthraquinone 20.8 >50 26 O O HO OH OH OH Flavonol 51.9%d >50 27 O O HO OH OH OCH3 OH Flavonol 47.29%d >50 28 O O HO OH OH OH OH OH Flavonol 46.9 >50 29 O O HO OH OH Flavonol 46.91%d >50 (continued on next page) M.-M. Liu et al. / European Journal of Medicinal Chemistry 52 (2012) 33e43 39
Author's personal copy M-M. Liu et aL/European Jourmal of Medicinal Chemistry 52(2012)33-43 Table 3(continued) Compound Structure Classification ECso(uM)b OCH3 HO OH Flavonol >50 83.66 OH OCH OH 50 OH OH OH Flavonol OH O
Author's personal copy Table 3 (continued ) Compound Structure Classification EC50 (mM)a,b CC50 (mM) 30 O O HO OH OCH3 OH Flavonol 27.6%d >50 31 O O OH OH OH HO HO Flavonol 31.09%d >50 32 O O OH HO OH HO Flavonol NA >50 33 O OH O OH Flavonol 83.66%d >50 34 O OH O HO OH OCH3 OCH3 OH Flavonol 15.5 >50 35 O O HO OH OH OH OH Flavonol 19.8 >50 36 O O HO OH OH OH OH OH Flavonol 47.0 >50 40 M.-M. Liu et al. / European Journal of Medicinal Chemistry 52 (2012) 33e43
Author's personal copy M.-M. Liu et al./ European Joumal of Medicinal Chemistry 52(2012)33-43 Table 3(continued) Compound ECso (uM) 37 HO HO 38 31.094 OH HO OH Flavonol 2640% >50 The anti-HCv assay was evaluated in an authentic HCV infection/replication system in the human hepatoma cell lines, Huh7. The ECso values were estimated by inhibition Mycophenolic acid(MPa)was used as reference positive control. No obvious inhibition at 50 HM. d the inhibition ratio of HCV replication at 50 uM. default minimization set. The Common Feature Pharmacophore selected from publications [9-11, 13-15 and other 1000 inactive Generation protocol available in Ds was employed to establish compounds randomly selected from Maybridge database was the pharmacophore. The ' Conformation generation choice was employed. The active NS5B inhibitors in decoy set were different set to Best with other parameters in their default values. Among the from the compounds in the training set. Ligand Pharmacophore 10 possible hypotheses returned, the top ranked hypothesis(hypo Mapping module with Best conformation generation and Flexible 1)was selected and validated by GH scoring method [19, 20. in fitting sets was used. The selected hypothesis was evaluated based which a decoy set made up of 40 active NS5B inhibitors were on the screening results. Validation of compounds 21 and 22 in an authentic HCV infection/replication system and subgenomic replicon system. Structure infection system replicon system HO H The ECso values were estimated by inhibition of 5 concentrations. b Mycophenolic acid(MPA)was used as reference positive control
Author's personal copy default minimization set. The Common Feature Pharmacophore Generation protocol available in DS was employed to establish the pharmacophore. The ‘Conformation generation’ choice was set to Best with other parameters in their default values. Among the 10 possible hypotheses returned, the top ranked hypothesis (Hypo 1) was selected and validated by GH scoring method [19,20], in which a decoy set made up of 40 active NS5B inhibitors were selected from publications [9e11,13e15] and other 1000 inactive compounds randomly selected from Maybridge database was employed. The active NS5B inhibitors in decoy set were different from the compounds in the training set. Ligand Pharmacophore Mapping module with Best conformation generation and Flexible fitting sets was used. The selected hypothesis was evaluated based on the screening results. Table 3 (continued ) Compound Structure Classification EC50 (mM)a,b CC50 (mM) 37 O OH O OH OH Flavonol 12.7 >50 38 O O OH OH HO HO OH Flavonol 31.09%d >50 39 O O HO OH OH OH Flavonol 26.40%d >50 a The anti-HCV assay was evaluated in an authentic HCV infection/replication system in the human hepatoma cell lines, Huh7. The EC50 values were estimated by inhibition of 5 concentrations. b Mycophenolic acid (MPA) was used as reference positive control. c No obvious inhibition at 50 mM. d The inhibition ratio of HCV replication at 50 mM. Table 4 Validation of compounds 21 and 22 in an authentic HCV infection/replication system and subgenomic replicon system. Compound Structure EC50 (mM)a,b in infection system EC50 (mM)a,b in replicon system 21 O O HO OH OH 7.9 12.0 22 O O HO OH OH OH 4.7 6.6 a The EC50 values were estimated by inhibition of 5 concentrations. b Mycophenolic acid (MPA) was used as reference positive control. M.-M. Liu et al. / European Journal of Medicinal Chemistry 52 (2012) 33e43 41
