at SciVerse ScienceDirect

Biochimie 94 (2012) 2457e2460

Contents lists available

Biochimie

journal homepage: www.elsevier .com/locate/biochi

Short communication

Discovery of a natural product inhibitor targeting protein neddylationby structure-based virtual screening

Hai-Jing Zhong a,b,1, Victor Pui-Yan Ma c,1, Zhen Cheng c, Daniel Shiu-Hin Chan c,Hong-Zhang He c, Ka-Ho Leung c, Dik-Lung Ma c,**, Chung-Hang Leung a,b,*

a Institute of Chinese Medical Sciences, University of Macau, Macao SAR, Chinab State Key Laboratory of Quality Research in Chinese Medicine, University of Macau, Macao SAR, ChinacDepartment of Chemistry, Hong Kong Baptist University, Kowloon Tong, Hong Kong

a r t i c l e i n f o

Article history:Received 18 December 2011Accepted 5 June 2012Available online 16 June 2012

Keywords:Drug discoveryVirtual screeningUbiquitin-like proteinNEDD8-activating enzymeNatural product

* Corresponding author. Institute of Chinese MedMacau, Macao SAR, China. Tel.: þ853 8397 8518; fax:** Corresponding author. Tel.: þ852 3411 7075; fax:

E-mail addresses: edmondma@hkbu.edu.hk (D.-L.(C.-H. Leung).

1 Both authors contributed equally to the work.

0300-9084/$ e see front matter � 2012 Elsevier Mashttp://dx.doi.org/10.1016/j.biochi.2012.06.004

a b s t r a c t

NEDD8-activating enzyme (NAE) controls the specific degradation of proteins regulated by cullin-RINGubiquitin E3 ligase, and has been considered as an attractive molecular target for the development ofanti-cancer drugs. We report herein the identification of the dipeptide-conjugated deoxyvasicinonederivative (1) as an inhibitor of NAE by virtual screening of over 90,000 compounds from the ZINCdatabase of natural products. Molecular modelling results suggested that 1 may be a non-covalentcompetitive inhibitor of NAE by blocking the ATP-binding domain. Compound 1 was able to inhibitNAE activity in both cell-free and cell-based assay with potencies in the micromolar range and selectivityover analogous E1 enzymes UAE and SAE. We envisage that the identification and molecular dockinganalysis of this bioactive scaffold as an NAE inhibitor would provide the scientific community with usefulinformation in order to generate more potent analogues.

� 2012 Elsevier Masson SAS. All rights reserved.

1. Introduction regulated by CRLs, such as the cancer related substrates p-IkBa and c-

The ubiquitin-proteasome system (UPS) is the master regulatorof protein degradation in eukaryotic cells [1]. Inhibitors targetingthe proteasome, bortezomib [2e4] and salinosporamide A [5], havebeen shown to be effective anti-cancer agents. However, thegeneral inhibition of proteasome activity may affect the turneroverof thousands of proteins, potentially increasing the toxicity andlowering the therapeutic index of the drugs [6].

The ubiquitinylation of the unwanted proteins is achieved by thecoordinated efforts of the E1 activation, E2 conjugation and E3 ligaseenzymes [7]. Ubiquitin-like protein (Ubl) conjugation pathways,including NEDD8 and SUMO, have also been identified as key playersin the protein degradation process [7,8]. Similar to the ubiquitinpathway, NEDD8 is first activated by the E1-like NEDD8-activatingenzyme (NAE) and is then transferred to Ubc12, the NEDD8-conjuating enzyme. Finally, the NEDD8 is conjugated to cullinproteins to activate the cullin-RING ubiquitin E3 ligase (CRLs) activity[9,10]. Therefore, NAE could be considered as a more selectiveupstream target, controlling the specific degradation of proteins

ical Sciences, University ofþ853 2884 1358.þ852 3411 7348.Ma), duncanleung@umac.mo

son SAS. All rights reserved.

myc. The NAE inhibitor MLN4924 [11,12], currently in Phase 1 clinicaltrials, has been shown to be effective against both solid and haema-tological human cancers [6]. Recently, our group has shown that thenatural product-like 6,600-biapigenin dose-dependently inhibitedNAE activity in cancer cells and suppressed the growth of a humanepithelial colorectal adenocarcinoma cell line [13].

Virtual screening (VS) has emerged as a powerful approach tocomplement the array of existing high-throughput screening tech-nologies [14,15]. Using computer-aided VS, potential leads can berapidly identified in silico in order to narrow the range ofmolecules tobe tested invitro and invivo. The integrationoforthogonalapproachesto the development of drug leads can greatly reduce the economicalinvestment in chemical synthesis and/or preliminary testing. Mean-while, natural products have long captured the attention ofmedicinalchemists due to the diversity of their chemical scaffolds, potentiallylower toxicities and bioactive substructures [16e22].

2. Results and discussion

In this study, we expanded the scope of our structure-based drugdiscovery strategy and screened over 90,000 compounds from theZINC natural product database for NAE inhibitors in silico. Theflexible ligandswere docked to a grid representation of the receptorand assigned a score reflecting the quality of the ligandereceptorcomplex the according to the Internal Coordinate Mechanics

H.-J. Zhong et al. / Biochimie 94 (2012) 2457e24602458

(ICM) method [ICM-Pro 3.6-1d molecular docking software (Mol-soft)] [23]. After visual inspection to eliminate compounds sharinghigh structural similarity with similar ICM docking scores (�0.5), 9compounds were obtained based on their availability and accessi-bility from commercial suppliers. The 9 high-scoring compoundswere tested in a preliminary E1 NAE activity assay. Four of thecompounds showed inhibitory effects against Ubc12-NEDD8conjugation in vitro at a concentration of 12.5 mM. Their structures,ICM docking scores and potencies against NAE activity aresummarized inTable S1. From the preliminary E1 NAE activity assay,the dipeptide-conjugated deoxyvasicinone derivative (1) (CASnumber: 951950-33-7) emerged as the top candidate under thetested conditions (Fig. 1A). Pyrrolo[2,1-b]quinazoline alkaloids areknown to display a variety of pharmacological activities [24e28].Deoxyvasicinone, amember of the pyrrolo[2,1-b]quinazoline familyand congener of 1, exhibits anti-microbial, anti-depressant and anti-inflammatory activities [29]. However, to our knowledge, no bio-logical activity of compound 1 has been reported in the literature.

The molecular modelling analysis of the 1-NAE-NEDD8 complexrevealed that the top-scoring binding mode of 1was different fromthat of 6,600-biapigenin, but similar to that of ATP or MLN4924 [13].The deoxyvasicinone ring is predicted to occupy the hydrophobicpocket near Ile148 andMet101 of UBA3 and participate in hydrogenbonding with Gln149 through its pyrimidone nitrogen (Fig. 1B). Thependant dipeptide moiety resides in the relatively hydrophilicregion and is predicted to form multiple hydrogen-bonding inter-actions with NAE in this site, including two H-bonds with the sidechain of Lys124 through its amide oxygen atoms, a single H-bond toAsp167 through its carbonyl oxygen atom, and anH-bond to Asp100through its amide hydrogen atom. Intriguingly, our model predictsthat 1 is capable of forming additional H-bonds with both NEDD8and the APPBP1 subunit through its amide and terminal carboxylategroups of its dipeptide side chain, respectively. The H-bonds formedbetween 1 and residues Asp100, Lys124, Asp 167 and Gln149mimicthose formed in the NAEeATP complex [30]. (Fig.1B). The predictedbinding coordinates of ATP in the binding pocket are within 2.0 ÅRMSD of the reported values in the X-ray crystal structure [30].

The binding score for 1 with NAE was calculated to be �34.2,compared to �30.3 and �30.8 for ATP and MLN4924, respectively[13]. MLN4924 is a mechanism-based inhibitor of NAE, as it formsa tight covalent adduct with the carboxyl terminus of NEDD8 via itssulfamate group. However, since 1 lacks a nucleophilic moiety, it is

Fig. 1. (A) Chemical structures of small molecule NAE inhibitors 1, 6,60 0-biapigenin and MLNenergy binding conformations of 1 and ATP to the NAE heterodimer generated by virtual ligathe ribbon form. Small molecule inhibitor 1 is depicted as a ball-and-stick model showinrepresented as the red ball-and-stick model. H-bonds are indicated as blue lines. The bind

unlikely to inhibit NAE via this mechanism. Instead, the strongbinding affinity predicted for 1 could be attributed to the ability ofthe dipeptide side chain to form multiple hydrogen bondinginteractions with the charged residues inside the binding siteincluding those of NEDD8 and the APPBP1 subunit. Furthermore,the molecular model revealed that the binding pose of 1 over-lapped considerably with that of ATP, suggesting that 1 may act asa reversible competitive inhibitor of NAE by blocking the ATP-binding domain. Additionally, we also performed molecularmodelling of 1 to the SAE-SUMO complex. The binding score for 1 tothe SAE-SUMO protein complex was calculated to be �20.9, whichwas significantly lower than that for ATP (�32.2). The dockinganalysis for UAE could not be performed as no suitable X-ray crystalstructure was available for molecular modelling.

To validate our molecular modelling results, a doseeresponseexperiment was performed using a cell-free enzymatic assay toevaluate the ability of 1 to inhibit the activity of NAE and its analo-gous E1 enzymes for ubiquitin and SUMO. A dose-dependentreduction in the level of the Ubc12eNEDD8 thioester conjugatewas observed upon incubation of NAE with 1, with nearly completeinhibition observed at 12.5 mM of 1 (Fig. 2A). Encouragingly, 1 wasselective for NAE over the analogous ubiquitin-activating enzyme(UAE) and SUMO-activating enzyme (SAE), with no observableinhibition of the Ubc10eUb or Ubc9eSUMO products, respectively,at the maximum concentration of 1 tested. Densitometry analysisrevealed that the IC50 value of 1 against NAE activity in the cell-freesystemwas approximately 0.8 mM(Fig. 2B). Theweak in vitro activityof 1 against SAE was consistent with the molecular docking resultsfor 1 with the SAE-SUMO complex described previously. We thentested 7 commercially available analogues of 1 with various sidechains for their NAE inhibitory activities in vitro. Interestingly, allanalogues showed some degree of inhibition of NAE activity in thecell-free environment, suggesting that the deoxyvasicinone scaffoldmay be one of the crucial factors in determining inhibitory activityagainstNAE. The structures, ICMdocking scores andpotencies of the7 analogues against NAE activity are summarized in Table S2.

We next investigated the ability of 1 to inhibit NAE activity inhuman cancer cells. Caco-2 (human epithelial colorectal adenocar-cinoma) cells were incubated with varying concentrations of 1 for24 h and analysed for Ubc12eNEDD8 conjugate levels. A dose-dependent inhibition of Ubc12eNEDD8 levels was observed upontreatment with 1, with an estimated IC50 value of ca. 6 mM from

4924. (B) Low-energy binding conformation of (a) 1, and (b) superposition of the low-nd docking. Proteins APPB1 (blue), UBA3 (purple) and NEDD8 (yellow) are displayed ing carbon (yellow), hydrogen (grey), oxygen (red) and nitrogen (blue) atoms. ATP is

ing pocket of the NAE is represented as a translucent green surface.

Fig. 2. (A) Compound 1 inhibits Ubc12eNEDD8 formation in a dose-dependent manner, but has no observable effect on Ubc10eUb or Ubc9eSUMO formation. E1 enzymes chargedwith Ubc12/NEDD8, Ubc10/Ub, and Ubc9/SUMO were incubated for 60 minwith the indicated concentrations of 1. The reaction mixture was fractionated by non-reducing SDS-PAGEand immunoblotted using antibodies for Ubc12, Ubc10 and Ubc9 respectively. (B) Densitometry analysis of the Western blot showing 1-mediated reduction of Ubc12eNEDD8.Approximate IC50 value ¼ ca. 1 mM. (C) Western blot showing selective dose-dependent inhibition of cellular Ubc12eNEDD8 level by 1, while Ubc10eUb and Ubc9eSUMO levelsremained unaffected. Caco-2 cells were treated with the indicated concentrations of 1 for 16 h. cell lysates were fractionated by non-reducing SDS-PAGE and immunoblotted usingantibodies for Ubc12, Ubc10 and Ubc9 (upper panel). Equal loading was confirmed by GAPDH levels (lower panel). (D) Densitometry analysis of the cell-based Western blot showing1-mediated reduction of Ubc12eNEDD8 but not the Ubc10-Ub and Ubc9-SUMO. Approximate IC50 value ¼ ca. 6 mM. (E) Caco-2 cells were treated with 1 or MLN4924 for 16 h. celllysates were immunoblotted to analyse the CRL substrate p27kip1 level using the p27kip1 antibody (upper panel). Equal loading was confirmed by beta-actin level (lower panel).

H.-J. Zhong et al. / Biochimie 94 (2012) 2457e2460 2459

densometryanalysis (Fig. 2C andD). Consistentwith the results of thecell-free enzymatic assay,1was inactive towards UAE and SAE at thehighest dosage of 1 tested, suggesting that 1 is selective towardsNAE.

Since NAE inhibition should lead to the down-regulation of theCRL activity, we examined the effects of 1 on the levels of CRLsubstrate p27kip1 in cells by immunoblotting analysis [11]. Theresults showed that treatment of Caco-2 cells with 1 induced anaccumulation of p27kip1 protein in a dose-dependent manner,presumably due to its ability to block the neddylation pathway byinterfering with NAE activity (Fig. 2E). We also examined thecytotoxicity of 1 against the growth of cancer cells in vitro using anMTT assay. Our results indicated that 1 was moderately toxicagainst Caco-2 cells, with an IC50 value of ca.10 mM after incubationfor 48 h (Fig. S2). It is conceivable that the cytotoxicity of 1 againstCaco-2 cells could be attributed, at least in part, to the inhibition ofNAE activity by 1, and leading to the accumulation of CRL substratessuch as tumour suppressor protein p27kip1.

3. Conclusion

In conclusion, we have discovered a new small molecule NAEinhibitor from the ZINC natural product database of >90,000

compounds using high-throughput virtual screening. To ourknowledge, the dipeptide-conjugated deoxyvasicinone derivative 1is only the third example of a small molecule inhibitor of NAE.1wasable to inhibit NAE activity in both cell-free and cell-based assaywith potencies in the micromolar range and with selectivity overanalogous E1 enzymes UAE and SAE. Furthermore, compound 1inhibited CRL-mediated substrate degradation and showedmoderate cytotoxicity against Caco-2 cells in an MTT assay. Ourmolecular modelling results suggest that 1 may be non-covalentcompetitive inhibitor of NAE since it shares a similar bindingmode to ATP. We are currently conducting in silico lead optimiza-tion of 1 with the aim of generating more potent analogues forin vitro biological testing.

Acknowledgements

Thiswork is supportedbyHongKongBaptistUniversity (FRG2/10-11/008 and FRG2/11-12/009), Environment and Conservation Fund(ECF Project 3/2010), Centre for Cancer and Inflammation Research,School of ChineseMedicine (CCIR-SCM,HKBU), the Research Fund forthe Control of Infectious Diseases (RFCID/11101212), the ResearchGrants Council (HKBU/201811), the University of Macau (Start-up

H.-J. Zhong et al. / Biochimie 94 (2012) 2457e24602460

Research Grant to C.-H. Leung, MYRG091(Y1-L2)-ICMS12-LCH andMYRG121(Y1-L2)-ICMS12-LCH).

Appendix A. Supplementary material

Supplementary material associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.biochi.2012.06.004

References

[1] A.Hershko, Theubiquitin system for proteindegradation and someof its roles inthe control of the cell division cycle, Cell. Death Differ. 12 (2005) 1191e1197.

[2] R.C. Kane, P.F. Bross, A.T. Farrell, R. Pazdur, Velcade�: U.S. FDA Approval forthe treatment of multiple myeloma Progressing on Prior therapy, TheOncologist 8 (2003) 508e513.

[3] J. Adams, M. Kauffman, Development of the proteasome inhibitor Velcade�(Bortezomib), Cancer Invest. 22 (2004) 304e311.

[4] P.G. Richardson, C. Mitsiades, Bortezomib: proteasome inhibition as aneffective anticancer therapy, Future Oncol. 1 (2005) 161e167.

[5] D. Chauhan, L. Catley, G. Li, K. Podar, T. Hideshima, M. Velankar, C. Mitsiades,N. Mitsiades, H. Yasui, A. Letai, H. Ovaa, C. Berkers, B. Nicholson, T.-H. Chao,S.T.C. Neuteboom, P. Richardson, M.A. Palladino, K.C. Anderson, A novel orallyactive proteasome inhibitor induces apoptosis in multiple myeloma cells withmechanisms distinct from Bortezomib, Cancer Cell 8 (2005) 407e419.

[6] T.A. Soucy, P.G. Smith, M. Rolfe, Targeting NEDD8-Activated Cullin-RINGligases for the treatment of cancer, Clin. Cancer Res. 15 (2009) 3912e3916.

[7] G. Nalepa, M. Rolfe, J.W. Harper, Drug discovery in the ubiquitin-proteasomesystem, Nat. Rev. Drug Discov. 5 (2006) 596e613.

[8] O. Kerscher, R. Felberbaum, M. Hochstrasser, Modification of proteins byubiquitin and ubiquitin-like proteins, Annu. Rev. Cell Develop. Biol. 22 (2006)159e180.

[9] Z.-Q. Pan, A. Kentsis, D.C. Dias, K. Yamoah, K. Wu, Nedd8 on cullin: building anexpressway to protein destruction, Oncogene 23 (2004) 1985e1997.

[10] G. Rabut, M. Peter, Function and regulation of protein neddylation, EMBO Rep.9 (2008) 969e976.

[11] T.A. Soucy, P.G. Smith, M.A. Milhollen, A.J. Berger, J.M. Gavin, S. Adhikari,J.E. Brownell, K.E. Burke, D.P. Cardin, S. Critchley, C.A. Cullis, A. Doucette,J.J. Garnsey, J.L. Gaulin, R.E. Gershman, A.R. Lublinsky, A. McDonald,H. Mizutani, U. Narayanan, E.J. Olhava, S. Peluso, M. Rezaei, M.D. Sintchak,T. Talreja, M.P. Thomas, T. Traore, S. Vyskocil, G.S. Weatherhead, J. Yu, J. Zhang,L.R. Dick, C.F. Claiborne, M. Rolfe, J.B. Bolen, S.P. Langston, An inhibitor ofNEDD8-activating enzyme as a new approach to treat cancer, Nature 458(2009) 732e736.

[12] R.T. Swords, K.R. Kelly, P.G. Smith, J.J. Garnsey, D. Mahalingam, E. Medina,K. Oberheu, S. Padmanabhan, M. O’Dwyer, S.T. Nawrocki, F.J. Giles, J.S. Carew,

Inhibition of NEDD8-activating enzyme: a novel approach for the treatment ofacute myeloid leukemia, Blood 115 (2010) 3796e3800.

[13] C.-H. Leung, D.S.-H. Chan, H. Yang, R. Abagyan, S.M.-Y. Lee, G.-Y. Zhu,W.-F. Fong, D.-L. Ma, A natural product-like inhibitor of NEDD8-activatingenzyme, Chem. Commun. 47 (2011) 2511e2513.

[14] J. Bajorath, Integration of virtual and high-throughput screening, Nat. Rev.Drug Discov. 1 (2002) 882e894.

[15] B.K. Shoichet, Virtual screening of chemical libraries, Nature 432 (2004)862e865.

[16] F.E. Koehn, G.T. Carter, The evolving role of natural products in drug discovery,Nat. Rev. Drug Discov. 4 (2005) 206e220.

[17] D.J. Newman, G.M. Cragg, Natural products as Sources of new drugs over theLast 25 Years, J. Nat. Prod. 70 (2007) 461e477.

[18] M.S. Butler, Natural products to drugs: natural product-derived compounds inclinical trials, Nat. Prod. Rep. 25 (2008) 475e516.

[19] J.W.-H. Li, J.C. Vederas, Drug discovery and natural products: end of an era oran endless frontier? Science 325 (2009) 161e165.

[20] H.-F. Ji, X.-J. Li, H.-Y. Zhang, Natural products and drug discovery, EMBO Rep.10 (2009) 194e200.

[21] R. Breinbauer, I.R. Vetter, H. Waldmann, From protein domains to drug can-didatesdnatural products as guiding principles in the design and synthesis ofcompound libraries, Angew. Chem. Int. Ed. 41 (2002) 2878e2890.

[22] P. Ertl, S. Roggo, A. Schuffenhauer, Natural product-likeness score and itsapplication for prioritization of compound libraries, J. Chem. Inf. Model. 48(2007) 68e74.

[23] M. Totrov, R. Abagyan, Flexible protein-Ligand docking by global energyoptimization in internal coordinates, Protein Struct. Funct. Genet. 29 (1997)215e220.

[24] D.R. Mehta, J.S. Naravane, R.M. Desai, Vasicinone. A Bronchodilator principlefrom Adhatoda vasica Nees (N. O. Acanthaceae), J. Org. Chem. 28 (1963)445e448.

[25] A. Al-Shamma, S. Drake, D.L. Flynn, L.A. Mitscher, Y.H. Park, G.S.R. Rao,A. Simpson, J.K. Swayze, T. Veysoglu, S.T.S. Wu, Antimicrobial agents fromhigher plants. Antimicrobial agents from Peganum harmala seeds, J. Nat. Prod.44 (1981) 745e747.

[26] J.P. Michael, Quinoline, quinazoline and acridone alkaloids, Nat. Prod. Rep. 17(2000) 603e620.

[27] P. Molina, A. Tárraga, A. Gonzalez-Tejero, I. Rioja, A. Ubeda, M.C. Terencio,M.J. Alcaraz, Inhibition of leukocyte functions by the alkaloid Isaindigotonefrom Isatis indigotica and some new synthetic derivatives, J. Nat. Prod. 64(2001) 1297e1300.

[28] A. Astulla, K. Zaima, Y. Matsuno, Y. Hirasawa, W. Ekasari, A. Widyawaruyanti,N. Zaini, H. Morita, Alkaloids from the seeds of Peganum harmala showingantiplasmodial and vasorelaxant activities, J. Nat. Med. 62 (2008) 470e472.

[29] M.P. Jain, S.K. Koul, K.L. Dhar, C.K. Atal, Novel nor-harmal alkaloid fromAdhatoda vasica, Phytochem. 19 (1980) 1880e1882.

[30] H. Walden, M.S. Podgorski, D.T. Huang, D.W. Miller, R.J. Howard, D.L. Minor,J.M. Holton, B.A. Schulman, The structure of the APPBP1-UBA3-NEDD8-ATPcomplex reveals the basis for selective ubiquitin-like protein activation by anE1, Mol. Cell 12 (2003) 1427e1437.