International Journal of
Molecular Sciences Article
Discovery of a Potential HER2 Inhibitor from Natural Products for the Treatment of HER2-Positive Breast Cancer Jianzong Li 1,† , Haiyang Wang 1,† , Junjie Li 1 , Jinku Bao 1,2,3, * and Chuanfang Wu 1, * 1
2 3
* †
School of Life Sciences and Key Laboratory of Ministry of Education for Bio-Resources and Bio-Environment, Sichuan University, Chengdu 610064, China;
[email protected] (J.L.);
[email protected] (H.W.);
[email protected] (J.L.) State Key Laboratory of Biotherapy/Collaborative Innovation Center for Biotherapy, West China Hospital, Sichuan University, Chengdu 610041, China State Key Laboratory of Oral Diseases, West China College of Stomatology, Sichuan University, Chengdu 610041, China Correspondence:
[email protected] (J.B.);
[email protected] (C.W.); Tel./Fax: +86-28-8541-5171 (J.B.) These authors contributed equally to this work.
Academic Editors: Ge Zhang, Aiping Lu and Hailong Zhu Received: 1 June 2016; Accepted: 28 June 2016; Published: 1 July 2016
Abstract: Breast cancer is one of the most lethal types of cancer in women worldwide due to the late stage detection and resistance to traditional chemotherapy. The human epidermal growth factor receptor 2 (HER2) is considered as a validated target in breast cancer therapy. Even though a substantial effort has been made to develop HER2 inhibitors, only lapatinib has been approved by the U.S. Food and Drug Administration (FDA). Side effects were observed in a majority of the patients within one year of treatment initiation. Here, we took advantage of bioinformatics tools to identify novel effective HER2 inhibitors. The structure-based virtual screening combined with ADMET (absorption, distribution, metabolism, excretion and toxicity) prediction was explored. In total, 11,247 natural compounds were screened. The top hits were evaluated by an in vitro HER2 kinase inhibition assay. The cell proliferation inhibition effect of identified inhibitors was evaluated in HER2-overexpressing SKBR3 and BT474 cell lines. We found that ZINC15122021 showed favorable ADMET properties and attained high binding affinity against HER2. Moreover, ZINC15122021 showed high kinase inhibition activity against HER2 and presented outstanding cell proliferation inhibition activity against both SKBR3 and BT474 cell lines. Results reveal that ZINC15122021 can be a potential HER2 inhibitor. Keywords: HER2 (human epidermal growth factor receptor 2); virtual screening; ADMET (absorption, distribution, metabolism, excretion and toxicity); biological evaluation
1. Introduction The World Health Organization (WHO) reports that breast cancer is one of the most common malignancies among women worldwide. Breast cancer has a substantially higher incidence (43.3 per 100,000) than any other cancers, followed by colorectal (14.3), cervix (14.0), lung (13.6) and corpus uteri (8.2) among gynecological tumors [1]. Despite a relatively low fatality rate, breast cancer has the highest death rate (12.9 per 100,000) than any other cancers. Although molecular therapies for breast cancer have developed rapidly over the last few decades [2], future treatment strategies with higher efficacy and lower toxicity are still in urgent need to be investigated.
Int. J. Mol. Sci. 2016, 17, 1055; doi:10.3390/ijms17071055
www.mdpi.com/journal/ijms
Int. J. Mol. Sci. 2016, 17, 1055
2 of 14
The human epidermal growth factor receptor 2 (HER2) is a validated target in breast cancer Int. J. Mol. Sci. 2016, 17, 1055 2 of 14 therapy. HER2, a transmembrane receptor with tyrosine kinase activity, belongs to the epidermal growth factor receptor family [3] and is known as a potent mediator of cellular growth and proliferation. The human epidermal growth factor receptor 2 (HER2) is a validated target in breast cancer It plays a significant role in many processes implicated in the regulation of cell survival, growth therapy. HER2, a transmembrane receptor with tyrosine kinase activity, belongs to the epidermal and differentiation through interlinked signal transduction involving activation of the PI3K/Akt growth factor receptor family [3] and is known as a potent mediator of cellular growth and and MAPK/ERK1/2 (mitogen-activated (MAP) kinase/extracellular signal-regulated kinase proliferation. It plays a significant role inprotein many processes implicated in the regulation of cell survival, (ERK)1/2) pathways [4,5]. Some studies demonstrated that HER2 is always in an active conformation growth and differentiation through interlinked signal transduction involving activation of the and ready to interact with the ligand-activated HER receptors Overexpression of HER2 is one of PI3K/Akt and MAPK/ERK1/2 (mitogen-activated protein (MAP) [6]. kinase/extracellular signal-regulated the molecular abnormalities linked the development of breastthat cancer [7] and existsininan about 30% of kinase (ERK)1/2) pathways [4,5].toSome studies demonstrated HER2 is always active conformation ready to interact patients with earlyand stage breast cancerwith [8,9].the ligand-activated HER receptors [6]. Overexpression of HER2 one the molecular to the development of breast cancer [7] and exists In the is era ofof molecular and abnormalities personalizedlinked therapeutics, the discovery of tyrosine kinase inhibitors in about 30% of patients with early stage breast cancer [8,9]. targeting HER2 has provided a successful avenue of therapies in HER2-overexpressing breast In the era of molecular and personalized therapeutics, the discovery of tyrosine kinase cancer [10]. Many small molecules inhibitors targeting HER2 are currently in clinical development [11]. inhibitors targeting HER2 has provided a successful avenue of therapies in HER2-overexpressing Lapatinib, an orally active HER2 kinase inhibitor, is the only drug approved by the U.S. Food and Drug breast cancer [10]. Many small molecules inhibitors targeting HER2 are currently in clinical Administration (FDA) patients HER2-positive advanced-stage cancer [12]. It development [11]. for Lapatinib, anwith orally active HER2 kinase inhibitor, isbreast the only drugspecifically approved by competes with ATP toDrug bindAdministration with HER2 kinase suppressing HER2 kinase activity and followed the U.S. Food and (FDA)domain, for patients with HER2-positive advanced-stage breast by shutting off downstream [13].ATP Clinical trials that lapatinib could enhance cancer specifically [12]. It pathways competes with to bind withdemonstrated HER2 kinase domain, suppressing HER2 the apoptotic effectand of anti-HER2 and pathways a combination of lapatinib and monoclonal kinase activity followed by antibodies shutting off [14,15], downstream [13]. Clinical trials demonstrated that lapatinib could enhance the lapatinib apoptotic effect anti-HER2 antibodies [14,15], and a combination antibody had better efficacy than aloneof[16,17]. However, it can be effective initially, and of lapatinib and monoclonal antibody had better efficacy than lapatinib alone [16,17]. However, can year then the patients with HER2-positive breast cancer develop acquired drug-resistance withitone be effective initially, and then the patients with HER2-positive breast cancer develop acquired of treatment [17,18]. There are two synthetic compounds as HER2 inhibitors that are currently in drug-resistance with one year of treatment [17,18]. There are two synthetic compounds as HER2 Phase III trials. These drug candidates, including dacomitinib (PF-00299804, Pfizer, New York, NY, inhibitors that are currently in Phase III trials. These drug candidates, including dacomitinib USA) [19,20] and neratinib (Pfizer) [21], are relatively similar to lapatinib. Although tremendous (PF-00299804, Pfizer, New York, NY, USA) [19,20] and neratinib (Pfizer) [21], are relatively similar to progress in HER2-directed therapiesprogress has beeninmade, a significant proportion of HER2-positive patients lapatinib. Although tremendous HER2-directed therapies has been made, a significant still relapse and die of breast cancer. Hence, alternative therapies are urgently required. proportion of HER2-positive patients still relapse and die of breast cancer. Hence, alternative For the are purpose developing novel potential HER2 inhibitors, the structure-based therapies urgentlyof required. For the purpose of developing HER2 inhibitors, structure-based high-throughput high-throughput virtual screeningnovel has potential been performed. Thethe ADMET (absorption, distribution, virtual screening hasand beentoxicity) performed. The ADMET (absorption, distribution, metabolism, excretion metabolism, excretion properties of small molecule candidates were also evaluated. and toxicity) properties of small molecule candidates were also evaluated. Moreover, the MM–PBSA Moreover, the MM–PBSA (Molecular Mechanics–Poisson Boltzmann Surface Area) calculation based (Moleculardynamics Mechanics–Poisson Boltzmann Surface Area) based on molecular dynamics on molecular (MD) simulation was utilized to calculation further investigate the binding activity of (MD) simulation was utilized to further investigate the binding activity of the HER2-inhibitor the HER2-inhibitor systems. Furthermore, the selected inhibitors were evaluated by an in vitro HER2 systems. Furthermore, the selected inhibitors were evaluated by an in vitro HER2 kinase activity kinase activity inhibition assay. The cell proliferation inhibition was tested in HER2-overexpressing inhibition assay. The cell proliferation inhibition was tested in HER2-overexpressing SKBR3 SKBR3 BT474 lines. brief workflow presented in Figure 1. The potential HER2 andand BT474 cellcell lines. TheThe brief workflow waswas presented in Figure 1. The potential HER2 inhibitors identified from the current study could be helpful in the design and development of a inhibitors identified from the current study could be helpful in the design and development of a novelnovel HER2HER2 inhibitor. inhibitor.
Figure 1. The brief workflow of identification of novel potential inhibitors targeting human epidermal
Figure 1. The brief workflow of identification of novel potential inhibitors targeting human epidermal growth factor receptor 2 (HER2). growth factor receptor 2 (HER2).
Int. J. Mol. Sci. 2016, 17, 1055 Int. J. Mol. Sci. 2016, 17, 1055 Int. J. Mol. Sci. 2016, 17, 1055 Int.Results J. Mol. Sci. 2016, 17, 1055 2. Int.Results J. Mol. Sci. 2016, 17, 1055 2. 2. Results J. Mol. 2016, 17, 1055 Int. Int. J. Mol. Sci. Sci. 2016, 17, 1055 2. Results Int. J.Molecular Mol. Sci. 2016, 17, 1055Study 2.1. Docking Int. J. Mol. 17, 2. Int.Results J.Molecular Mol. Sci. Sci. 2016, 2016, 17, 1055 1055Study 2.1. Docking
3 of 14 3 of 14 3 of 14 3 of 14 3 of 14 3 of 314of 14 3 of 14 33 of of 14 14
2. Results 2.1. Molecular Docking Study Here, Amber scores and Autodock Vina scores were calculated in parallel with the 11,247 natural 2.1. Molecular Docking Study Results 2. 2. Results Here, Amber scores and Autodock Vina scores were calculated in parallel with the 11,247 natural 2. Results 2.1. Molecular Docking Study compounds. Twelve common compounds bothcalculated high Amber scoreswith and the high Vinanatural scores Here, Amber scores and Autodock Vinashowing scores were in parallel 11,247 compounds. Twelve common compounds showing bothcalculated high Amber scoreswith and the high Vinanatural scores 2.1. Molecular Docking Study Here, Amber scores and the Autodock Vina scoresATP were in parallel 11,247 were selected. Lapatinib and natural substrate were also docked into HER2. The results were compounds. Twelve common compounds showing both high Amber scores and high Vina scores 2.1.2.1. Molecular Docking Study Molecular Docking Study Here, Amber scores and the Autodock Vina scoresATP were calculated in parallel with the 11,247 natural were selected. Lapatinib and natural substrate were also docked into HER2. The results were 2.1. Molecular Docking Study compounds. Twelve common compounds showing both high Amber scores and high Vina scores presented in Table 1. Compared to lapatinib, seven compounds showed higher Amber score, Here, Amber scores and Autodock Vina scores were calculated in parallel with the 11,247 natural were selected.Twelve Lapatinib thecompounds natural substrate ATP were alsoAmber dockedscores into HER2. The results were compounds. common showing both high and high Vina scores presented in Table 1. Compared to lapatinib, seven compounds showed higher Amber score, Amber scores and Autodock scores were calculated parallel with the 11,247 Here, Amber scores Autodock Vina scores were calculated in in parallel with the 11,247 natural were selected. Lapatinib and the natural substrate ATP were also docked intohigher HER2. The results were sixHere, compounds showed higher Vina score, four compounds showed both Amber and Vina compounds. Twelve common compounds showing both high Amber scores and high Vina scores presented in Table 1. Compared to lapatinib, seven compounds showed higher Amber score, Here, Amber scores and Autodock Vina scores were calculated in parallel with the 11,247 natural were selected.Twelve Lapatinib and thecompounds natural substrate ATP were also docked intohigher HER2. Thehigh results were six compounds showed higher Vina score, fourshowing compounds showed both Amber and Vina natural compounds. Twelve common compounds both high Amber scores and Vina compounds. common showing both high Amber scores and high Vina scores presented in Table 1. Compared to lapatinib, seven compounds showed higher Amber score, scores. All compounds showed both higher Amber scores and Vina scores than the natural substrate were selected. Lapatinib and the natural substrate ATP were also docked into HER2. The results were compounds. Twelve common showing both high Amber scores and high Vina six compounds showed highercompounds Vina score, four compounds showed both higher Amber andscores Vina presented in Table 1. Compared to lapatinib, seven compounds showed higher Amber score, scores. All compounds showed both higher Amber scores and Vina scores than the natural substrate scores were selected. Lapatinib and the natural substrate ATP were also docked into HER2. The were selected. Lapatinib and theaVina natural substrate ATP were also docked into HER2. The results were six compounds showed higher score, four compounds showed both higher Amber and Vina ATP. Meanwhile, we found common structural characteristic for selected compounds that presented in Table 1. Compared to lapatinib, seven compounds showed higher Amber score, were selected. Lapatinib and the natural substrate ATP were also docked into HER2. The results were scores. All compounds showed both higher Amber scores and Vina scores than the natural substrate six compounds showed higher Vina score, four compounds showedshowed both higher Amber andscore, Vina ATP. Meanwhile, we found aboth common structural characteristic for selected compounds that results were presented in Table 1. Compared to lapatinib, seven compounds showed higher Amber presented in Table 1. Compared to lapatinib, seven compounds higher Amber scores. All compounds showed higher Amber scores and Vina scores than the natural substrate hydrogen-bonding groups andaVina nonpolar aromatic moieties were situated in the middle and atVina the six compounds showed higher score, four compounds showed both higher Amber and presented in Table 1. showed Compared tohigher lapatinib, seven compounds showed higher Amber score, ATP. Meanwhile, we found common structural characteristic for selected compounds that scores. All compounds both Amber scores and Vina scores than the natural substrate hydrogen-bonding groups and nonpolar aromatic moieties were situated in the middle and atVina the score, six compounds showed higher Vina score, four compounds showed both higher Amber and six compounds showed higher Vina score, four compounds showed both higher Amber and ATP. Meanwhile, we found a common structural characteristic for selected compounds that end of molecular frameworks, respectively. scores. All compounds showed both higher Amber scores and Vina scores than the natural substrate six compounds showed higher score, four compounds showed both higher Amber andatVina hydrogen-bonding groups andaVina nonpolar aromatic moieties were situated in the middle and the ATP. Meanwhile, we found common structural characteristic for selected compounds that end of molecular frameworks, respectively. Vina scores. compounds showed both higher Amber scores andsituated Vina scores than the natural scores. AllAll compounds showed both higher Amber scores and Vina scores than the natural substrate hydrogen-bonding groups andrespectively. nonpolar aromatic moieties were in the middle and atthat the ATP. Meanwhile, we found a common structural characteristic for selected compounds end of molecular frameworks, scores. All compounds showed both higher Amber scores and Vina scores than the natural substrate hydrogen-bonding groups andanonpolar aromatic moieties were situated in thecompounds middle andthat atthat the substrate ATP. Meanwhile, we found a common structural characteristic for selected ATP. Meanwhile, we found common structural characteristic for selected compounds end of molecular frameworks, respectively. Table 1. The results of virtual screening and structures of common top-score compounds. hydrogen-bonding groups andrespectively. aromatic moieties were situated in the middle and atthat the ATP. Meanwhile, we found anonpolar common structural characteristic for selected compounds end of molecular frameworks, Table 1. The results of virtual screening and structures of common top-score compounds. hydrogen-bonding groups and and nonpolar aromatic moieties were situated intop-score the in middle and at the hydrogen-bonding groups nonpolar aromatic moieties were situated the middle andend at the Table 1. The results of virtual screening and structures common compounds. end of molecular frameworks, respectively. hydrogen-bonding groups and nonpolar aromatic moietiesofwere situated in the middle and at the Score (kcal/mol) Table The results virtual screening and structures of common top-score compounds. a frameworks, b of molecular frameworks, respectively. end of molecular respectively. Rank1. ZINC IDof Structure Score (kcal/mol) end of molecular respectively. Table The results screening and structures of common top-score compounds. b Amber Score Vina Score Rank1.a frameworks, ZINC IDof virtual Structure Score (kcal/mol) b Table results and structures of common top-score compounds. Amber Score Vina Score Rank1.a The ZINC IDof virtual screening Structure Score (kcal/mol) b Table 1. The results of virtual screening andand structures of common top-score compounds. Amber Score Vina Score Table 1.a The results of virtual screening structures of common top-score compounds. Rank ZINC ID Structure Score (kcal/mol) Table 1.a The results and structures of common top-score compounds. b Amber Score Vina Score Rank ZINC IDof virtual screening Structure Score (kcal/mol) 1 31166919 −44.19 −10.3 a Amber Score Vina Score Rank ID Structure b Score (kcal/mol) 1 a ZINC 31166919 −44.19 −10.3 Score (kcal/mol) b Amber Score Vina Score Score (kcal/mol) Rank ZINC ID Structure a b 1 31166919 −44.19 −10.3 ID ID Rank Rank aaZINC Structure ZINC Structure bb Amber Score Vina Score 1 31166919 −44.19 −10.3 Amber Score Vina Score Amber Score Vina Score 1 31166919 −44.19 −10.3 12 31166919 −44.19 −10.3 15122021 −43.67 −10.8 12 31166919 −44.19 −10.3 15122021 −43.67 −10.8 1 31166919 −44.19 −10.3 1 ´44.19 ´10.3 2 31166919 15122021 −43.67 −10.8 2 15122021 −43.67 −10.8 2 15122021 −43.67 −10.8 2 15122021 −43.67 −10.8 −43.54 −10.6 13378641 233 15122021 −43.67 −10.8 −43.54 −10.6 13378641 2 15122021 15122021 −43.67 −10.8 2 ´43.67 ´10.8 −43.54 −10.6 3 13378641 −43.54 −10.6 3 13378641 −43.54 −10.6 3 13378641 −43.54 −10.6 3 13378641 ´43.54 ´10.6 3 −43.54 −10.6 34 13378641 13378641 72320250 −35.54 −10.0 −43.54 −10.6 3 13378641 4 72320250 −35.54 −10.0 4 72320250 −35.54 −10.0 4 72320250 −35.54 −10.0 4 72320250 −35.54 −10.0 4 72320250 72320250 −35.54 −10.0 4 ´35.54 ´10.0 45 72320250 −35.54 −10.0 −35.26 −9.9 49181256 4 72320250 −35.54 −10.0 −35.26 −9.9 5 49181256 −35.26 −9.9 5 49181256 −35.26 −9.9 5 49181256 −35.26 −9.9 5 49181256 49181256 5 ´35.26 ´9.9 −35.26 −9.9 5 49181256 35456612 −34.01 −10.4 −35.26 −9.9 56 49181256 −35.26 −9.9 56 49181256 35456612 −34.01 −10.4 6 35456612 −34.01 −10.4 6 35456612 35456612 −34.01 −10.4 6 ´34.01 ´10.4 6 35456612 −34.01 −10.4 6 35456612 −34.01 −10.4 67 35456612 −34.01 −10.4 −33.93 −9.9 72320169 6 35456612 −34.01 −10.4 −33.93 −9.9 7 72320169 −33.93 −9.9 7 72320169 72320169 7 ´33.93 ´9.9 −33.93 −9.9 7 72320169 −33.93 −9.9 7 72320169 −33.93 −9.9 7 72320169 −33.93 −9.9 7 72320169 −33.93 −9.9 78 72320169 lapatinib −32.59 −10.2 8 lapatinib lapatinib −32.59 −10.2 8 ´32.59 ´10.2 8 lapatinib −32.59 −10.2 8 lapatinib −32.59 −10.2 8 lapatinib −32.59 −10.2 8 lapatinib −32.59 −10.2 8 lapatinib −32.59 −10.2 89 35456515 lapatinib −32.59 −10.2 −33.55 −10.7 35456515 9 ´33.55 ´10.7 −33.55 −10.7 9 35456515 −33.55 −10.7 9 35456515 −33.55 −10.7 9 35456515 −33.55 −10.7 9 35456515 −33.55 −10.7 9 35456515 −33.55 −10.7 9 35456515 −33.55 −10.7 9 35456515
Int. J. Mol. Sci. 2016, 17, 1055
4 of 14
Int. J. Mol. Sci. 2016, 17, 1055
4 of 14
Int. J. Mol. Sci. 2016, 17, 1055 Int. J. Mol. Sci. 2016, 17, 1055 Int. J. Mol. Sci. 2016, 17, 1055 Int. J. Mol. Sci. 2016, 17, 1055 10 2016, 35456607 Int. Rank J. Mol. aSci. 17, 1055ID ZINC
Table 1. Cont.
4 of 14 4 of 14 4 of 14 4 of 14 4 of 14
−32.45Score (kcal/mol) −10.4 10 35456607 −32.45 −10.4 Amber Score Vina Score 10 35456607 −32.45 −10.4 10 35456607 −32.45 −10.4 10 35456607 −32.45 −10.4 10 ´32.45 ´10.4 10 35456607 35456607 −32.45 −10.4 11 72320025 −32.01 −10.1 11 72320025 −32.01 −10.1 11 72320025 −32.01 −10.1 11 72320025 −32.01 −10.1 11 72320025 −32.01 −10.1 11 72320025 ´32.01 ´10.1 11 72320025 −32.01 −10.1 12 67912776 −31.85 −10.0 12 67912776 −31.85 −10.0 12 67912776 −31.85 −10.0 12 67912776 67912776 −31.85 −10.0 12 ´31.85 ´10.0 12 44352487 67912776 −31.85 −10.0 −28.83 −10.1 13 12 67912776 −31.85 −10.0 −28.83 −10.1 13 44352487 −28.83 −10.1 13 44352487 44352487 13 ´28.83 ´10.1 −28.83 −10.1 13 44352487 −28.83 −10.1 13 44352487 14 13 ATP −10.45 −7.5 −28.83 −10.1 44352487 14 −10.45 −7.5 ´10.45 ´7.5 14 ATPATP 14 ATP −10.45 −7.5 a The compounds 14 ATP −10.45 by the ChemDraw −7.5 were sorted by Amber score; b The structures were generated 14 ATP −10.45 −7.5 a The compounds werewere sorted by Amber score; The structures were generated by the ChemDraw program a The b The 14 ATP −10.45 −7.5 program (CambridgeSoft, Cambridge, MA,b score; USA). Morestructures information on these compounds was compounds sorted by Amber were generated by the ChemDraw Structure b
a The compounds b The (CambridgeSoft, Cambridge, MA, USA). More information on these compounds providedby in Table S1. by Amber score; structures werewas generated the ChemDraw provided in Table S1. were sorted program (CambridgeSoft, Cambridge, USA). information on theseby compounds was a The compounds b TheMore were sorted by AmberMA, score; structures were generated the ChemDraw a The compounds b TheMore program (CambridgeSoft, Cambridge, MA, USA). information on these compounds was were sorted by Amber score; structures were generated by the ChemDraw provided in Table S1. a The compounds b TheMore program (CambridgeSoft, Cambridge, MA, USA). information on thesebycompounds was were sorted byofAmber score; structures were generated the ChemDraw provided in Tablethe S1. In order to ensure reliability the screening screening, high throughput screening, program (CambridgeSoft, Cambridge, MA, USA).model More for information on these virtual compounds was provided in Table S1. program (CambridgeSoft, Cambridge, MA, USA). More information on these(ROC) compounds wasin an evaluation was performed. The output of the receiver operating characteristic is shown provided in Table S1.the reliability of the screening model for high throughput virtual screening, In order to ensure provided in Table S1.the order to ensure reliability of the model for high throughput screening, Figure 2.In The Amber score achieved a AUC (thescreening area under the ROC curve) of 0.918, virtual the Vina score
Figure 2.InThe Amber score achieved AUC (the area under the ROC curve) of 0.918,(ROC) the Vina score in an evaluation was performed. Thea output ofscreening the receiver operating characteristic isscreening, shown order to ensure the reliability of the model for high throughput virtual an evaluation was performed. The output of the receiver operating characteristic (ROC) is shown in In order to ensure the reliability of the screening model for high throughput virtual screening, achieved AUC of 0.888. indicated the molecular docking reliable for an The results that model was Figure 2. The Amber score achieved a AUC area under the ROC curve) of 0.918, theisVina score an evaluation was performed. The output of(the the receiver operating characteristic (ROC) shown in In 2. order to ensure the achieved reliability of the screening model forROC high throughput virtual screening, Figure The Amber score a AUC (the area under the curve) of 0.918, the Vina score an evaluation was performed. The output of the receiver operating characteristic (ROC) is shown in distinguishing the active compounds and inactive compounds effectively. compounds effectively. achieved an AUC of 0.888. The results indicated that under the molecular docking model was reliable for Figure 2. The Amber score achieved a AUC area the ROC curve) of 0.918, theis Vina score an evaluation was of performed. The output of(the the receiver operating characteristic (ROC) shown in achieved an AUC 0.888. The results indicated that the molecular docking model was reliable for Figure 2. The Amber score achieved aand AUC (the area under theeffectively. ROC curve) of 0.918, the Vina score distinguishing the of active compounds inactive compounds achieved an AUC 0.888. The results indicated that the molecular model for Figure 2. The Amber score achieved aand AUC (the area under theeffectively. ROCdocking curve) of 0.918,was the reliable Vina score distinguishing the of active compounds inactive compounds achieved an AUC 0.888. The results indicated that the molecular docking model was reliable for distinguishing the of active compounds and inactivethat compounds effectively. achieved an AUC 0.888. The results indicated the molecular docking model was reliable for distinguishing the active compounds and inactive compounds effectively. distinguishing the active compounds and inactive compounds effectively.
Figure of the the simulated simulated docking docking model. model. Figure 2. 2. Receiver Receiver operating operating characteristic characteristic (ROC) (ROC) analysis analysis of Figure 2. Receiver operating characteristic (ROC) analysis of the simulated docking model. 2. Receiver operatingMetabolism, characteristicExcretion (ROC) analysis of the simulated model. 2.2. ADMET Figure (Absorption, Distribution, and Toxicity) Propertiesdocking Analysis 2. Receiver operatingMetabolism, characteristicExcretion (ROC) analysis of the simulated model. 2.2. ADMETFigure (Absorption, Distribution, and Toxicity) Propertiesdocking Analysis Figure 2. ReceiverDistribution, operating characteristic analysis the simulated docking model. 2.2. ADMET (Absorption, Metabolism,(ROC) Excretion andofToxicity) Properties Analysis Favorable ADMET characteristics can be considered as anand essential nature for a candidate Figure 2. Receiver operating characteristic (ROC) analysis ofToxicity) the simulated docking model. drug. 2.2. ADMET (Absorption, Distribution, Metabolism, Excretion Properties Analysis Favorable ADMET characteristics can be considered as an essential nature for a candidate drug.
ADMET (Absorption, Distribution, Excretionasand Toxicity) Properties The2.2. ADMET properties 12 selectedMetabolism, natural compounds were predicted. on ADMET Favorable ADMETofcharacteristics can be considered an essential natureBased forAnalysis a candidate drug. ADMET (Absorption, Distribution, Metabolism, Excretion and Toxicity) Properties Analysis The2.2. ADMET properties of 12characteristics selected natural compounds wereas predicted. Based on ADMET prediction, Favorable ADMET can be considered an essential nature for a candidate drug. 2.2. ADMET (Absorption, Distribution, Metabolism, Excretion and Toxicity) Properties Analysis prediction, five potential compounds showed favorable ADMET properties. The selected prediction The ADMET properties of 12 selected natural compounds were predicted. Based on ADMET Favorable ADMETshowed characteristics canADMET be considered as an The essential nature for a candidate drug. fiveThe potential compounds favorable properties. selected prediction properties ADMET properties of 12 selected natural compounds were predicted. Based on ADMET Favorable ADMET characteristics can be considered as an(RO5) essential nature for a candidate drug. properties of the five hits are shown in Table 2. Rule-of-five represents physicochemical prediction, fivetop potential compounds showed favorable ADMET properties. The selected Thetop ADMET properties of selected compounds were predicted. Based onprediction ADMET Favorable ADMET characteristics cannatural be considered as an essential nature for a candidate drug. of the five hits are shown in 12 Table 2. Rule-of-five (RO5)ADMET represents physicochemical parameters prediction, five potential compounds showed favorable properties. The selected prediction The ADMET properties of 12 selected natural compounds were predicted. Based on ADMET parameters defined by follows: (1)2.molecular weight lower than 500; (2) number properties of thepotential topLipinski five compounds hits[22], are which shown in Table Rule-of-five (RO5) represents physicochemical prediction, five showed favorable ADMET properties. The Based selected The ADMET properties of 12areselected natural compounds were predicted. onprediction ADMET properties of the top five hits shown in Table 2. Rule-of-five (RO5) represents physicochemical prediction, five potential compounds showed favorable ADMET properties. The selected prediction of H-bond (hydrogen bond) donors lower or equal to five; (3) number of H-bond acceptors lower or parameters defined by Lipinski [22], which follows: (1) molecular weight lower than 500; (2) number properties of the top five hits are shown in Table 2. Rule-of-five (RO5) represents physicochemical prediction, five potential compounds showed favorable ADMET properties. The selected prediction parameters defined by Lipinski [22], which follows: (1) molecular weight lower than 500; (2) number properties of the top five hits are shown in Table 2. Rule-of-five (RO5) represents physicochemical equal than 10; and (4) log P lower than five. None of the selected compounds violated this rule. of H-bond (hydrogen bond) donors lower or equal to five; (3) number of H-bond acceptors or parameters defined Lipinski [22], whichin follows: (1) molecular weight lower than 500; (2)lower number properties of the topbyfive hits are shown Table 2.to Rule-of-five (RO5)ofrepresents physicochemical of H-bond (hydrogen donors orfollows: equal five; (3) number H-bond acceptors lower or parameters defined bybond) Lipinski [22],lower which (1) molecular weight lower than 500; (2)this number equal than 10; and (4) log P lower than five. None of the selected compounds violated rule. of H-bond (hydrogen bond) donors or equal to five; (3) number of H-bond acceptors lower or parameters defined by Lipinski [22],lower whichfive. follows: (1) molecular weight lower than 500; (2)this number equal than 10; and (4) log P lower than None of the selected compounds violated rule. of H-bond (hydrogen bond) donors lower or equal to five; (3) number of H-bond acceptors lower or equal than (hydrogen 10; and (4)bond) log Pdonors lower lower than five. None the(3) selected violated this rule. of H-bond or equal toof five; numbercompounds of H-bond acceptors lower or
Int. J. Mol. Sci. 2016, 17, 1055
5 of 14
defined by Lipinski [22], which follows: (1) molecular weight lower than 500; (2) number of H-bond (hydrogen bond) donors lower or equal to five; (3) number of H-bond acceptors lower or equal than 10; and (4) log P lower than five. None of the selected compounds violated this rule. ADMET Risk indicates the comprehensive evaluation of ADMET properties. All compounds attained this value lower or equal to five. The results demonstrated that these compounds possessed outstanding ADMET properties. Table 2. The ADMET (absorption, distribution, metabolism, excretion and toxicity) properties of the top five hits. ADMET Properties
Molecules ZINC13378641
ZINC15122021
ZINC35456515
ZINC31166919
ZINC49181256
S + logP 4.73 S + Sw 1.10 S + Vd 0.7 CYP_1A2_Substr No (66%) MET_UGT1A1 No (59%) TOX_hERG_Filter No (95%) TOX_BRM_Rat 289.81 TOX_AlkPhos Normal (60%) TOX_GGT Normal (78%) TOX_LDH Normal (76%) RO5 0 TOX_MUT_Risk 0 ADMET Risk 3.36
5.01 5.97 0.46 No (59%) Yes (58%) No (95%) 522.34 Normal (74%) Normal (78%) Normal (76%) 0 0 3.81
´0.01 1.08 0.26 No (96%) No (92%) No (95%) 9.89 Elevated (65%) Normal (57%) Elevated (75%) 0 0 4.5
3.92 1.00 0.17 No (96%) No (88%) No (95%) 200.29 Elevated (97%) Elevated (77%) Normal (70%) 0 0 5
2.92 1.11 0.13 No (96%) No (82%) No (95%) 206.47 Elevated (97%) Elevated (90%) Normal (96%) 0 0 5
Where S + logP, S + Sw mean the octanol-water partition coefficient and native water solubility; S + Vda is the pharmacokinetic volume of distribution in human; CYP_1A2_Substr is the measurement of compound being the substrate of Cytochrome P450 1A2; MET_UGT1A1 is qualitative model of a glucuronidation by the UDP-glucuronosyltransferase 1A1 enzyme; TOX_hERG_Filter and TOX_AlkPhos denote qualitative estimation of the likelihood of the hERG potassium channel inhibition and liver adverse effect as the likelihood of causing elevation in the levels of Alkaline Phosphatase enzyme in human; TOX_BRM_Rat means the oral dose of compound required to cause tumors 50 percent of a rat population after exposure over an average lifetime; TOX_GGT means human liver adverse effect as the likelihood of causing elevation in the levels of GGT enzyme; TOX_MUT_Risk indicates ADMET Risk for mutagenicity in S. typhimurium.
2.3. Molecular Simulation Analysis Assessment of RMSD (root mean square deviations) value for each HER2-ligand system provided a complete insight into the conformational stability of each complex system. The RMSD values of five selected compounds, lapatinib and ATP were shown in Figure 3. As can be seen, most of the selected compounds possessed high stability thoughout the 50 ns MD simulation. In particular, ZINC31166919 (red) and ZINC15122021 (red) showed minimum fluctuation and attained an average RMSD value at 0.12 nm and 0.14 nm during 50 ns MD simulation, respectively. ZINC49181256 (red) also showed low fluctuation and attained an average RMSD value about 0.150 nm. By contrast, lapatinib (black) attained an average RMSD value about 0.17 nm. It demonstrated that ZINC31166919, ZINC15122021 and ZINC49181256 could bind to HER2 very stably and more tightly than lapatinib. The fluctuation of RMSD of ZINC3545651 and ZINC13378641 was slightly higher than other compounds. Although the fluctuation of RMSD indicated that ZINC3545651- and ZINC13378641-HER2 systems were not stable as others, the maximum RMSD values of all selected compounds were lower than 0.3 nm. The natural substrate ATP showed higher RMSD value than other compounds and the fluctuation of RMSD over time indicated that ATP did not possess much unfavorable stability against HER2 as compared to other compounds.
Int. J. Mol. Sci. 2016, 17, 1055
6 of 14
Int. J. Mol. Sci. 2016, 17, 1055
6 of 14
Figure 3. The RMSD (root mean square deviations) of the backbone atoms of HER2-ligand systems.
Figure 3. The RMSD (root mean square deviations) of the backbone atoms of HER2-ligand systems.
2.4. Binding Affinity Prediction
2.4. Binding Affinity Prediction
The MM–PBSA tool was employed to calculate their binding free energy. Generally, the ∆Gbind
The MM–PBSA tool was employed their binding free energy. the ∆Gbind indicates the comprehensive evaluationtoofcalculate binding affinity. The detailed results Generally, were presented in Tablethe 3. Itcomprehensive is encouraging to observe that compounds (−131.36 and indicates evaluation of two binding affinity.ZINC31166919 The detailed resultskcal/mol) were presented ZINC15122021 (−120.63 kcal/mol) showed favorable binding affinity compared to kcal/mol) other in Table 3. It is encouraging to observe thatmore two compounds ZINC31166919 (´131.36 compounds as well as lapatinib (−37.49 kcal/mol). Three compounds showed better binding effect and ZINC15122021 (´120.63 kcal/mol) showed more favorable binding affinity compared to other with HER2 than as lapatinib, including ZINC15122021, ZINC31166919 and ZINC49181256. contrast,effect compounds as well lapatinib (´37.49 kcal/mol). Three compounds showed betterInbinding two compounds, including ZINC13378641 and ZINC35456515, showed slighter unfavorable binding with HER2 than lapatinib, including ZINC15122021, ZINC31166919 and ZINC49181256. In contrast, affinity than lapatinib. The results were consistent with the trajectory analysis as lower binding two compounds, including ZINC13378641 and ZINC35456515, showed slighter unfavorable binding energy indicated more favorable binding stability. affinity than lapatinib. The results were consistent with the trajectory analysis as lower binding energy indicated Table more3.favorable stability. Summary ofbinding the binding free energy components for the protein–ligand complexes calculated by MM–PBSA (Molecular Mechanics–Poisson Boltzmann Surface Area) method.
Table 3. Summary of the binding free energy components for the protein–ligand complexes calculated Molecules by Components MM–PBSA (Molecular Mechanics–Poisson Boltzmann Surface Area) method. ZINC31166919 ZINC15122021 ZINC49181256 ZINC13378641 ZINC35456515 Lapatinib ∆Evdw −56.57 ± 1.95 −63.46 ± 2.58 ∆Eele −130.90 ± 7.19 −109.18 ± 6.70 Components ∆Gpb 61.47 ± 5.13 57.30 ± 3.87 ∆Gsa ZINC31166919 −5.36 ± 0.19 ZINC15122021 −5.30 ± 0.21 −187.47 −172.63 4.64 ∆Evdw ∆Emm ´56.57 ˘ 1.95± 4.57 ´63.46 ˘±2.58 ± 5.32 ´109.18 52.00 ˘ ± 3.66 ∆Eele ∆Gsol ´130.9056.11 ˘ 7.19 6.70 −120.63 ± 5.18 ∆Gpb ∆Gbind 61.47−131.36 ˘ 5.13± 6.63 57.30 ˘ 3.87
−53.88 ± 0.82 −51.56 ± 2.84 −57.82 ± 3.09 −51.02 ± 3.39 −39.73 ± 0.77 −1.58 ± 5.65 −17.31 ± 10.65 −26.03 ± 8.63 Molecules 44.88 ± 1.18 24.14 ± 5.35 49.84 ± 7.13 45.25 ± 5.22 ZINC49181256 ZINC13378641 ZINC35456515 −5.74 ± 0.06 −5.37 ± 0.18 −5.76 ± 0.21 −5.69 ± 0.21Lapatinib −93.61 1.59 ± 4.24 ± 6.87 ˘ 3.09 −77.05 ± 6.01 ´53.88± ˘ 0.82 −49.99 ´51.56 ˘ 2.84 −75.84 ´57.82 ´51.02 ˘ 3.39 39.04 ± 1.12 ± 5.17 ± 6.92 ˘ 10.65 39.56 ± 5.01 ´39.73 ˘ 0.77 18.77 ´1.58 ˘ 5.65 44.08 ´17.31 ´26.03 ˘ 8.63 −54.44 ± 3.89 ± 6.23˘ 7.13 −37.49 ± 5.46 44.88±˘0.84 1.18 −31.22 24.14 ˘ 5.35 −31.0549.84 45.25 ˘ 5.22
∆Gsa ´5.36 ˘ 0.19 ´5.30 ˘ 0.21 ´5.74 ˘ 0.06 ´5.37 ˘ 0.18 ´5.76 ˘ 0.21 ´5.69 ˘ 0.21 ∆Emm ´187.47 ˘ 4.57 ´172.63 ˘ 4.64 ´93.61 ˘ 1.59 ´49.99 ˘ 4.24 ´75.84 ˘ 6.87 ´77.05 ˘ 6.01 2.5. Biological Evaluation ∆Gsol 56.11 ˘ 5.32 52.00 ˘ 3.66 39.04 ˘ 1.12 18.77 ˘ 5.17 44.08 ˘ 6.92 39.56 ˘ 5.01 ∆Gbind For the ´131.36 ˘ 6.63 ´120.63 ˘ 5.18 ´54.44 ˘ 0.84inhibitors, ´31.22 HER2 ˘ 3.89 inhibition ´31.05 ˘assay 6.23 and´37.49 purpose of seeking novel potential HER2 cell ˘ 5.46
proliferation inhibition were evaluated in vitro. Most of the selected natural compounds showed a moderated kinase inhibition activity against HER2, and only a few showed outstanding inhibition 2.5. Biological Evaluation activity. The cellular inhibition activities were mainly coincident with the results of kinase inhibition. For of seeking HER2 inhibitors, HER2 inhibition The the datapurpose were detailed in Tablenovel 4, and potential the cell growth curves were presented in Figure 4. assay and cell
proliferation inhibition were evaluated in vitro. Most of the selected natural compounds showed a moderated kinase inhibition activity against HER2, and only a few showed outstanding inhibition activity. The cellular inhibition activities were mainly coincident with the results of kinase inhibition. The data were detailed in Table 4, and the cell growth curves were presented in Figure 4.
presented favorable cell proliferation inhibition activity against both SKBR3 and BT747 cell lines. It is encouraging find that ZINC15122021 exhibited high activities with IC50 value of 0.18 μM against HER2, IC50 value of 1.22 μM against SKBR3 cells and 4.11 μM against BT474 cells. ZINC31166919 also exhibited high activities with IC50 value of 2.63 μM against HER2, IC50 value of 8.61 μM against SKBR3 and 6.78 μM against BT474. ZINC13378641 showed lower activities than these compounds. Even Int. J. Mol. Sci. 2016, 1055 though their 17, activities were slightly lower compared with lapatinib, their cell inhibition IC50 values7 of 14 were lower than 50 μM except ZINC49181256 and ZINC35456515.
Int. J. Mol. Sci. 2016, 17, 1055
7 of 14
Table 4. The IC50 (half maximal inhibitory concentration) values of selected natural compounds against HER2 kinase and HER2-overexpressing SKBR3 and BT474 cell lines. The values represented the mean ± SD (standard deviations) of at least three independent experiments.
Cell Inhibition IC50 SKBR3 BT474 Figure 4. Dose-response effect of selected compounds on8.61 breast cancer cell lines, (a) SKBR3 cell line; ZINC31166919 2.63 ± 0.03 ± 0.45 6.78 0.68 Figure 4. Dose-response effect of selected compounds on breast cancer cell± lines, (a) SKBR3 cell line; (b) BT474 cell line.ZINC15122021 Results are expressed as the mean percentage of control plates containing no drug. 0.18 ± 0.002 1.22 ± 0.05 4.11 ± 0.95 (b) BT474 cell line. Results are expressed as the mean percentage of control plates containing no drug. Error bars correspond to SD (standard deviations) from three>50 independent measurements. ZINC49181256 9.18 ± 0.01from Error bars correspond to SD (standard deviations) three independent>50 measurements. ZINC13378641 3.71 ± 0.03 26.48 ± 1.62 18.55 ± 2.06 Three compounds (cell inhibition IC50 values < 50 μM) were further tested on the normal breast ZINC35456515 >10 >50 >50 Table 4. The IC50to(half maximal inhibitory values of selected natural compounds cell (Hs578Bst) study the effect on cellconcentration) proliferation and thus verify the toxicity of normalagainst cells. lapatinib 0.06 ± 0.001 0.38 ± 0.02 0.45 ± 0.03 HER2 and HER2-overexpressing SKBR3toand BT474 cell lines. of The valuesconcentrations represented the CCK-8kinase (Cell Counting kit-8) assays were utilized determine the effects different of mean ˘ SD (standard deviations) at viability least three independent experiments. three compounds on normal breastof cell after 24 h of treatment. Our experiment results revealed In particular, we found that ZINC15122021 showed high inhibition activity against HER2 and that ZINC15122021 and ZINC13378641 exhibited little cytotoxicity against normal breast cells. presented favorable cell proliferation inhibition activity against both SKBR3 and BT747 cell lines. It ZINC31166919 showed slightly enhanced cytotoxicity with increaseCell of concentrationIC(see 50 Figure 5). is Compounds encouraging find that ZINC15122021 exhibited high activities with ICInhibition 50 value of 0.18 μM against Enzymatic IC 50 BT474 also HER2, IC50 value of 1.22 μM against SKBR3 cells and 4.11 μMSKBR3 against BT474 cells. ZINC31166919 exhibited high activities with IC2.63 50 value of 2.63 μM against HER2, IC50 value of 8.61 μM against SKBR3 ZINC31166919 ˘ 0.03 8.61 ˘ 0.45 6.78 ˘ 0.68 and 6.78 μM against BT474. ZINC13378641 showed lower activities Even ZINC15122021 0.18 ˘ 0.002 1.22 ˘ 0.05 than these compounds. 4.11 ˘ 0.95 though their activities were slightly their cell inhibition IC 50 values ZINC49181256 9.18 ˘lower 0.01 compared with lapatinib, >50 >50 ZINC13378641 ˘ 0.03 26.48 ˘ 1.62 18.55 ˘ 2.06 were lower than 50 μM except 3.71 ZINC49181256 and ZINC35456515. ZINC35456515 >10 >50 >50 lapatinib 0.06 ˘ 0.001 0.38 ˘ 0.02 0.45 ˘ 0.03 Compounds
Enzymatic IC50
In particular, we found that ZINC15122021 showed high inhibition activity against HER2 and presented favorable cell proliferation inhibition activity against both SKBR3 and BT747 cell lines. It Figure 5. Cellthat viability of normal breast cells treated with three potential using is encouraging find ZINC15122021 exhibited high activities withHER2 IC50 inhibitors value of by 0.18 µM against CCK-8 assay. Error bars correspond to standard deviations from three independent measurements. HER2, IC50 value of 1.22 µM against SKBR3 cells and 4.11 µM against BT474 cells. ZINC31166919 also exhibited high activities with IC50 value of 2.63 µM against HER2, IC50 value of 8.61 µM against SKBR3 and 6.784.µM against BT474. showed these compounds. Figure Dose-response effect of ZINC13378641 selected compounds on breastlower cancer activities cell lines, (a)than SKBR3 cell line; Even though their activities were slightly lower compared with lapatinib, their cell inhibition IC50 (b) BT474 cell line. Results are expressed as the mean percentage of control plates containing no drug. values were lower than 50 µM except ZINC49181256 and ZINC35456515. Error bars correspond to SD (standard deviations) from three independent measurements. Three compounds (cell inhibition IC50 values < 50 µM) were further tested on the normal breast Three compounds (cell inhibition IC50proliferation values < 50 μM) further on the normal breast cells. cell (Hs578Bst) to study the effect on cell andwere thus verifytested the toxicity of normal cell (Hs578Bst) to study the effect on cell proliferation and thus verify the toxicity of normal cells. CCK-8 (Cell Counting kit-8) assays were utilized to determine the effects of different concentrations CCK-8 (Cell Counting kit-8) assays were utilized to determine the effects of different concentrations of of three compounds on normal breast cell viability after 24 h of treatment. Our experiment results three compounds on normal breast cell viability after 24 h of treatment. Our experiment results revealed revealed ZINC15122021 ZINC13378641 exhibited cytotoxicity against normal that that ZINC15122021 and and ZINC13378641 exhibited little little cytotoxicity against normal breastbreast cells. cells. ZINC31166919 showed slightly enhanced increaseofofconcentration concentration Figure ZINC31166919 showed slightly enhancedcytotoxicity cytotoxicity with with increase (see(see Figure 5). 5).
Figure 5. Cell viability of normal breast cells treated with three potential HER2 inhibitors by using
Figure 5. Cell viability of normal breast cells treated with three potential HER2 inhibitors by using CCK-8 assay. Error bars correspond to standard deviations from three independent measurements. CCK-8 assay. Error bars correspond to standard deviations from three independent measurements.
Int. J. Mol. Sci. 2016, 17, 1055
8 of 14
Int. J. Mol. Sci. 2016, 17, 1055
8 of 14
2.6. Binding Binding Model Model Analysis Analysis 2.6. The results results indicated indicated that that lapatinib lapatinib could could interact interact with with Leu726 Leu726 and and Met801 Met801 residues, residues, whereas whereas The ZINC15122021 could interact with Ser728 and Asp863 residues in the HER2 ATP-binding pocket. The ZINC15122021 could interact with Ser728 and Asp863 residues in the HER2 ATP-binding pocket. The conventional hydrogen hydrogen bonds bonds were were obviously obviously found found in in these these residues, residues, see see Figure Figure 6. 6. According According to to the the conventional docking study, we found that van der Waals interactions played a key role between ZINC15122021 docking study, we found that van der Waals interactions played a key role between ZINC15122021 and HER2. HER2. Meanwhile, Meanwhile, the the Alkyl Alkyl or or Pi-Alkyl Pi-Alkyl interactions interactions were were also also obviously obviously found found between between ligands ligands and and HER2. HER2. Compared Compared to to ZINC15122021, ZINC15122021, lapatinib lapatinib formed formed halogen halogen bonds bonds with with Ala751, Ala751, Glu770 Glu770 and and and Leu796 residues. In addition, two attractive or repulsive charged interactions were found between Leu796 residues. In addition, two attractive or repulsive charged interactions were found between lapatinib and and Gly727 Gly727 and and Asp808 Asp808 residues. residues. Various Various interaction interaction types types may may be be the the reason reason that that lapatinib lapatinib lapatinib can achieve achieve high high binding binding affinity affinity and and possess possess high high HER2 HER2 kinase kinase inhibition. inhibition. In In fact, fact, van van der der Waals Waals is is can one of the dominating forces for HER2-ZINC15122021 binding. It is the weakest of all intermolecular one of the dominating forces for HER2-ZINC15122021 binding. It is the weakest of all intermolecular attractions between between molecules, molecules, however, however, with with lots lots of of van van der der Waals Waals force force interactions interactions between between ligand ligand attractions and receptor, the interaction can be very strong. Thus, ZINC15122021 can also achieve high binding and receptor, the interaction can be very strong. Thus, ZINC15122021 can also achieve high binding affinity with with HER2 HER2 and and possess affinity possess high high biological biological activities activities against against HER2. HER2.
Figure Figure 6. 6. Comparison Comparison of ofbinding bindingmodels modelsof oflapatinib lapatiniband andZINC15122021 ZINC15122021against againstHER2 HER2at atatomic atomiclevel. level. (a) the interaction of ZINC15122021 with HER2; (b) the interaction of lapatinib with HER2. The (a) the interaction of ZINC15122021 with HER2; (b) the interaction of lapatinib with HER2. The 2D 2D diagram diagram interactions interactions were were shown shown as as dashed dashed lines lines between between receptor receptor residues residues and and ligand ligand atoms. atoms. The The figures figures were were generated generated by by the the PyMOL1.5 PyMOL1.5 (The (The PyMOL PyMOL Molecular Molecular Graphics Graphics System, System, San San Carlos, Carlos, CA, CA, USA) San Diego, Diego, CA, CA, USA). USA). USA) and and Discovery Discovery Studio Studio visualizer visualizer 16 16 (Accelrys, (Accelrys, San
3. Discussion According to the classical magic bullet paradigm, once biochemical and genetic studies reveal the molecular mechanisms of diseases such as cancer, it becomes possible that people could pick a protein that they think would make a favorable target, and screen compounds that interact with this
Int. J. Mol. Sci. 2016, 17, 1055
9 of 14
3. Discussion According to the classical magic bullet paradigm, once biochemical and genetic studies reveal the molecular mechanisms of diseases such as cancer, it becomes possible that people could pick a protein that they think would make a favorable target, and screen compounds that interact with this protein as potential drugs. Even with its current limitations, computational virtual screening offers a practical pipeline to discover new potential drugs for pharmaceutical research [23,24]. Many novel ligands have been successfully discovered using structure-based computation [25–28]. The remarkable features that discriminate the present study from the other concerns are as follows: (1) a large number of natural compounds are docked into the ATP-binding pocket of HER2. Studies have indicated that natural products play a highly significant role in the drug discovery and development process [29]; (2) the high AUC values indicates that our virtual screening model was reliable for identifying active ligands from compounds database effectively; (3) one of the most daunting hurdles a drug candidate must pass is possessing suitable ADMET properties [30]. Identification of these properties during early drug discovery is vital for reducing ADMET problems later in the drug development process. In present study, the ADMET properties of several selected compounds were analyzed. Our results reveal that five natural compounds possessed favorable ADMET properties; (4) meanwhile, these compounds were tested on the normal breast cell line to verify their toxicity on normal cells. We found that ZINC15122021 and ZINC31178641 possessed low cytotoxicity of normal breast cell. It indicates that these natural compounds can reduce the possibility of loss in drug discovery process; (5) biological evaluation act as gatekeeper by assessing enzymatic activity assay and cancer cell proliferation inhibition. Based on virtual screening results, the biological activities of five compounds were further tested. Although inhibition activities of ZINC15122021 were slightly lower than lapatinib, we found ZINC15122021 possessed high biological activities against HER2, and exhibited the IC50 values of 0.18 µM against HER2, 1.22 µM against SKBR3 cell and 4.11 µM against BT474 cells, making it promising to be an effective HER2 inhibitor. 4. Methods 4.1. Receptor and Ligand Preparation The crystal structure of HER2 kinase domain with high resolution used for the present study was retrieved from the RCSB protein databank: 3PP0 (X-ray diffraction with resolution 2.25 Å). The three-dimensional structures of all 11,247 natural products prepared for molecular docking were obtained from AnalytiCon Discovery NP [31]. The structure of lapatinib and natural substrate ATP were retrieved from the ZINC database. 4.2. Molecular Docking Model Validation In order to identify molecular docking models suitable for virtual screening against HER2, we used the crystal structure of the HER2 kinase domain with high resolution to assess the performance of virtual screening model. Herein, a docking simulation test was explored and the ROC was further analyzed. Thirty compounds were considered as the positive dataset, which were collected from literatures and the ChEMBL database according to its affinities (IC50 , Ki and Kd ď 100 nM). In addition, 561 decoys found by DecoyFinder1.1 [32]. The parameters of DecoyFinder1.1 were set as the following: active ligand vs. decoy tanimoto threshold ď0.75; decoy vs. decoy tanimoto threshold
