Targeting Neuroblastoma Cell Surface Proteins: Recommendations for Homology Modeling of hNET, ALK, and TrkB

PERSPECTIVE published: 20 Janua y 2017 doi: 10.3389/ nmol.2017.00007 Ta ge ing Neu oblas oma Cell Su ace P o eins: Recommenda ions o Homology Modeling o hNET, ALK, and T kB Yazan Haddad1,2,Zbynˇek Hege 1,2 and Voj ech Adam1,2* 1Depa men o Chemis y and Biochemis y, Mendel Uni e si y in B no, B no, Czechia, 2Cen al Eu opean Ins i u e o Technology, B no Uni e si y o Technology, B no, Czechia Edi ed by: De le Boison, Legacy Heal h, USA Re iewed by: Hong Qing, Beijing Ins i u e o Technology, China Nikki Ka he ine Ly le, Uni e si y o Cali o nia, San Diego, USA Da ide Comole i, Ru ge s Uni e si y, USA *Co espondence: Voj ech Adam [email p o ec ed] Recei ed: 26 Sep embe 2016 Accep ed: 06 Janua y 2017 Published: 20 Janua y 2017 Ci a ion: Haddad Y, Hege Z and Adam V (2017) Ta ge ing Neu oblas oma Cell Su ace P o eins: Recommenda ions o Homology Modeling o hNET, ALK, and T kB. F on . Mol. Neu osci. 10:7. doi: 10.3389/ nmol.2017.00007 Ta ge ed he apy is a p omising app oach o ea men o neu oblas oma as e iden om he la ge numbe o a ge ing agen s employed in clinical p ac ice oday. In he absence o known c ys al s uc u es, esea che s ely on homology modeling o cons uc empla e-based heo e ical s uc u es o d ug design and es ing. He e, we discuss h ee candida e cell su ace p o eins ha a e sui able o homology modeling: human no epineph ine anspo e (hNET), anaplas ic lymphoma kinase (ALK), and neu o ophic y osine kinase ecep o 2 (NTRK2 o T kB). When choosing empla es, bo h sequence iden i y and s uc u e quali y a e impo an o homology modeling and pose he i s o many challenges in he modeling p ocess. Homology modeling o hNET can be imp o ed using empla e models o dopamine and se o onin anspo e s ins ead o he leucine anspo e (LeuT). The ex acellula domains o ALK and T kB a e ye o be exploi ed by homology modeling. The e a e se e al idiosync asies ha equi e di ec a en ion h oughou he p ocess o model cons uc ion, e alua ion and e inemen . Shi s/gaps in he alignmen be ween he empla e and a ge , backbone ou lie s and side-chain o ame ou lie s a e among he main sou ces o physical e o s in he s uc u es. Low-conse ed egions can be e ined wi h loop modeling me hod. Residue hyd ophobici y, accessibili y o bound me als o glycosyla ion can aid in model e inemen . We ecommend esol ing hese idiosync asies as pa o “good modeling p ac ice” o ob ain highes quali y model. Dec easing physical e o s in p o ein s uc u es plays majo ole in he de elopmen o a ge ing agen s and unde s anding o chemical in e ac ions a he molecula le el. Keywo ds: neu oblas oma, a ge ed he apy, homology modeling, no epineph ine anspo e , anaplas ic lymphoma kinase, neu o ophic y osine kinase ecep o INTRODUCTION The la ge numbe o expe imen ally de e mined and deposi ed p o ein s uc u es in he da abases is a aluable sou ce o explo e uncha ed e i o ies o o he p o eins, which ha e simila sequence. Homology modeling, also known as compa a i e modeling, is used o cons uc ing p o ein models based on empla e s uc u es o homolog p o eins. In his pe spec i e a icle, we ha e highligh ed he main s uc u al issues in homology modeling o h ee candida e cell su ace p o eins o F on ie s in Molecula Neu oscience | www. on ie sin.o g 1Janua y 2017 | Volume 10 | A icle 7 Haddad e al. Homology Modeling o Ta ge ing Neu oblas oma neu oblas oma a ge ed he apy. In he nex sec ion, we emphasize he challenges aced in each s ep in he homology modeling p ocess. A b ie sec ion on a ge ing neu oblas oma desc ibes he ole o cell su ace p o ein a ge s in de eloping he apeu ics. A de ailed e alua ion o homology models expands on h ee chosen cell su ace p o ein candida es: human no epineph ine anspo e (hNET), anaplas ic lymphoma kinase (ALK), and neu o ophic y osine kinase ecep o ype 2 (NTRK2, commonly known as T kB). While he echnical aspec s o his manusc ip a o an audience o specialis s wo king on d ug disco e y and homology modeling o hese h ee a ge s, he same p inciples o model e alua ion and challenging issues can be applied o o he homology models. HOMOLOGY MODELING P o ein s uc u es a e he inal on ie s in unde s anding he human biology a he molecula le el. Gold s anda ds o c ys al s uc u e cha ac e iza ion a a omic esolu ion, such as X- ay c ys allog aphy and nuclea magne ic esonance (NMR) spec oscopy, can be now sa is ac o ily complemen ed wi h compu a ional me hods (Kund o as e al., 2012). Homology modeling is a sys ema ic compu a ional p ocess whe e he mos simila p o ein sequence, o a known c ys al s uc u e, is used o cons uc ion o a new model by eplacing he equi alen amino acids on an equi alen backbone. K iege e al. (2003) desc ibed se en s eps in homology modeling: (1) The choice o empla e(s) and ini ial alignmen a e he i s challenges in homology modeling. The accu acy o a homology model is co ela ed wi h he numbe o ma ching esidues in alignmen . A minimum 25% sequence iden i y has been he s anda d o homology modeling so a . Below 25% iden i y i is ecommended o use mul iple empla es o modeling. The quali y o empla e s uc u e is di ec ly inhe i ed in he homology model. Polishing o empla e s uc u e is a good p ac ice be o e use. Missing a oms should be ixed using o ame lib a y and he hanging e mini can be immed. (2) Alignmen co ec ion educes e o s caused by alse sequence iden i ies. Mul iple alignmen s and s uc u al alignmen s (e.g., posi ion-speci ic sco ing ma ices) a e ecommended al e na i es o con en ional alignmen s. This s ep is he mos c i ical in homology modeling be o e (3) backbone gene a ion, as i de e mines he o sional angles o he backbone in he model. I has been shown ha p oblems in he backbone can d as ically al e he co ec olding o he side-chains as well (Al-Lazikani e al., 2001). Issues ega ding empla e ecogni ion and alignmen will be highligh ed he e. F om ou expe ience, many o he spo adic e o s in modeling backbones a ise in p oline o adjacen o p oline esidues. (4) Loop modeling is impo an o co ec ing he olding o low-conse ed egions (i.e., loops) o p o ein. I is now possible o make accu a e models using da abase sea ch o ab ini io me hods o up o 8–13 esidues long loops (To o , 2012). Loop modeling employs po en ial ene gy sco es o e alua ion o he quali y o cons uc ed loop(s). I is also possible o use al e na i e a ionales o e alua ion based on biological unc ions. We show he e some examples o biological unc ions ha can include, exclude o guide he modeling p ocess such as hyd ophobici y, accessibili y o loop esidues o glycosyla ion o o binding o ions. (5) Side-chain modeling is di ec ly a ec ed by empla e sequence iden i y, alignmen and backbone. Iden ical esidues in wo homolog p o eins ha e nea ly iden ical o ame s. The SWISS-MODEL se e (Bo doli e al., 2009) and Modelle (Webb and Sali, 2014) a e among he mos commonly used pla o ms o backbone/side-chain homology modeling. Homology models a e subjec ed o inal op imiza ion and alida ion be o e conduc ing a ious kinds o compu a ional s udies such as ene ge ics (molecula mechanics), p o ein-d ug/p o ein-p o ein in e ac ions (molecula docking and p o ein ne wo ks), mu a ional analysis, and simula ion in physiological en i onmen (molecula dynamics). (6) Model op imiza ion, also known as ene gy minimiza ion, is applica ion o ene gy unc ions o compu e a global minimum ha ep esen s he mos na i e olding (Bo dne , 2012). Minimiza ion aims o adjus he geome ies o p o ein s uc u es o he ‘‘ o ce ield’’ pa ame e s used in compu a ional s udies. This me hod some imes e e ed o as ‘‘ elaxing he s uc u e’’ is su icien o esol e a omic clashes in he model. Howe e , he s o y is no inished he e. Se e al physical and s uc u al e o s should be esol ed. (7) Valida ion o he inal model(s) checks i he model complies wi h s anda d pa ame e s o p o ein s uc u e. These pa ame e s include: bond leng hs, bond angles, o sions, backbone ou lie s, o ame ou lie s and all a omic con ac s (Chen e al., 2010). E alua ion p ocedu e includes physics-based, knowledge-based and expe imen al-based me hods (Haddad e al., 2016). The same igo ous ules o e alua ion ha a e applied in c ys allog aphy mus also be applied in homology modeling. TARGETING NEUROBLASTOMA Neu oblas omas a e one o he mos common and a al solid umo s in child en below 2 yea s o age. Whe eas he su i al a es o mos ypes o cance in child en ha e imp o ed in he pas ew decades, neu oblas oma is s ill below 75% 5-yea su i al (Siegel e al., 2016). Ta ge ed he apy is a p omising app oach o de eloping ea men s o neu oblas oma. Ma hay e al. (2012) desc ibed h ee success ul he apeu ic a ge s o neu oblas oma ha a e in p ac ice oday: (1) hNET a ge ed by adio he apy ia 131I-me aiodobenzylguanidine (MIBG) (2) he GD2 ganglioside a ge ed by monoclonal an ibodies, and (3) ALK a ge ed by kinase inhibi o s. An ex ended lis o a ge s was p e iously compiled by Ve issimo e al. (2011), including hose in clinical ials. The as majo i y o FDA-app o ed medicines, pa icula ly hose disco e ed by a ional d ug design on molecula a ge s, wo k by pe u bing he unc ion o a ge s o p e en o e e se symp oms/disease (Kinch e al., 2015). I is impo an o dis inguish he e m a ge ed he apy, which occasionally indica es pe u bing he molecula a ge , om he gene al concep o a ge ing which also includes d ug deli e y by ecognizing cance cells. Hence, a ge ing also desc ibes he apeu ics able o deli e and elease ca go o he umo mic oen i onmen by binding o a ge s on he cance cell su ace (Shin e al., 2014). Such he apeu ics (e.g., pep ides and an ibodies) a e cha ac e ized by hei selec i i y o a ge s. They F on ie s in Molecula Neu oscience | www. on ie sin.o g 2Janua y 2017 | Volume 10 | A icle 7 Haddad e al. Homology Modeling o Ta ge ing Neu oblas oma can be de eloped ei he blindly ia sc eening me hods o by molecula modeling. The ideal a ge should be ‘‘abundan and accessible and should be exp essed homogeneously, consis en ly and exclusi ely on he su ace o cance cells’’ (Sco e al., 2012). Homology modeling has been success ully used in s uc u e- based design o many he apeu ics a ge ing di e en ypes o diseases including cance (Bu le e al., 2010; Schlessinge e al., 2011; DeVo e and Sco , 2012). Howe e , i is wo h men ioning wo ca ego ies o excellen p o ein a ge s in neu oblas oma ha a e beyond he scope o homology modeling: i s , p o eins which ha e known c ys al s uc u es such as CD147, which is associa ed wi h dec eased neu oblas oma di e en ia ion (Ga cia e al., 2009; W igh e al., 2014), and CD57, which has also been implica ed in agg ession o neu oblas oma (Kakuda e al., 2004; Schli e e al., 2012). Second, hose candida e a ge s which lack any empla e s uc u e. One o hese cases is a glycop o ein called CD133, and i s exp ession is associa ed wi h poo p ognosis o neu oblas oma (Sa ele e al., 2012). In compa a i e modeling o de elopmen o d ugs, he lesson lea ned is ha simila s uc u es o en exhibi simila unc ions. This is p oblema ic pa icula ly when d ugs a ge he p o ein and i s homologs simul aneously, esul ing in a ious side e ec s. The h ee a ge s hNET, ALK and T kB a e good examples o c oss in e ac ions esul ing in side e ec s. While hNET neu o ansmi e anspo o e laps wi h o he monoamine anspo e s, he kinase inhibi o s o ALK and T kB can a ge hei homologs, espec i ely. Bo h homology modeling and de elopmen o selec i e d ugs should ocus on he low-conse ed egions o a ge ed p o eins. hNET The hNET egula es he up ake and ecycle o no epineph ine in he neu ons. hNET is one o he highly exp essed p o eins in neu oblas oma (∼90% o cases) and mos commonly used in MIBG-based diagnosis/ he apy (B odeu e al., 1993; Ma hay e al., 2012) and de elopmen o new he apeu ics (Mo ensen and Ko age e, 2015). hNET p o ein has h ee iso o ms. The canonical iso o m is 617 esidues in leng h ye i is second o he longes iso o m which is 628 in leng h and di e s in he C- e minus. P e ious homology models o hNET elied on he c ys al s uc u e o p oka yo ic leucine anspo e (LeuT; Yamashi a e al., 2005) and d osophila dopamine anspo e (dDAT) c ys al s uc u es as empla es (Penma sa e al., 2013; Wang e al., 2015;Table 1). Two ecen ly published mu an se o onin anspo e (SERT) c ys al s uc u es (Coleman e al., 2016) designa ed s2 and s3 can also aid in homology modeling o hNET. The homology models cons uc ed by Schlessinge e al. (2011) and Koldsø e al. (2013) we e based on LeuT empla e. In ligh o newly published dDAT and SERT c ys al s uc u es, we ha e ecen ly add essed he de elopmen s ega ding homology modeling o hNET based on hese new empla es (Haddad e al., 2016). B ie ly, he e a e ou majo alignmen gaps cha ac e izing he sequences o dDAT, hNET and SERT; including he glycosyla ed ex acellula loop 2 (EL2), ex acellula loop 4 (EL4) and wo in acellula loops (Figu e 1A). Loop modeling is mos ly equi ed a he EL2 loop; pa icula ly esidues 189–207. A e cons uc ion o a numbe o loops, one should exclude hose wi h non-embedded leucines/ aline esidues due o hei hyd ophobici y. A i s glance, he accessibili y o he glycosyla ed esidues can be an indica o o non-embedded esidues in he loop (i.e., aspa agines). A leas wo glycosyla ed o ms o hNET a e known wi h molecula weigh s o 80 kD and 54 kD co esponding o a co e 46k D hNET p o ein (Melikian e al., 1994). I is nei he clea whe he all h ee aspa agines a e accessible o glycosyla ion no i indeed he e is a single con o ma ion o his loop. Mu an hNET (K189H) exhibi s a zinc binding si e in he EL2 loop (No egaa d e al., 1998), which migh sugges di ec p oximi y and accessibili y be ween K189 and nea es his idine H372. Howe e , no in o ma ion is a ailable on he p oximi y o H199 in he EL2 o hese esidues. Simila ly, se e al sodium and chlo ide ion binding si es a e epo ed in LeuT, dDAT and human SERT (hSERT; Yamashi a e al., 2005; Penma sa e al., 2013; Wang e al., 2015; Coleman e al., 2016). Ligands binding si es include he cen al binding si e (S1) and seconda y binding si e (S2) ha o e laps he ex acellula loop (EL4; Figu e 1A). Conse ed esidues in binding si es desc ibed by Koldsø e al. (2015) ha e nea ly he same o ame s excep o F317. Low-conse ed esidues in binding si es desc ibed by Ande sen e al. (2015) con ol he selec i eness o hNET and side-chain con o ma ions o hese esidues can a ec he quali y o he model (shown in Figu e 1A). As men ioned ea lie , he side-chains o iden ical aligned esidues exhibi nea ly iden ical o ame s. As many as 53 amino acid esidues in hNET ha ha e iden i y wi h hSERT and no dDAT can be used om he hSERT s uc u e o complemen he o ame s and imp o e accu acy o s uc u e (Haddad e al., 2016). We ecommend ha he cons uc ed homology model is alida ed o selec i i y by docking o p e iously known ligands such as neu o ansmi e s and inhibi o s. ALK ALK is a ecep o y osine kinase o iginally disco e ed by ch omosomal ea angemen associa ed wi h anaplas ic la ge cell lymphoma. ALK was epo ed o be cons i u i ely ac i a ed by gene ampli ica ion in se e al neu oblas oma cell lines (Osajima-Hakomo i e al., 2005), while neu oblas oma-speci ic mu a ions in ALK we e sui able a ge s o de elopmen o se e al inhibi o he apeu ics (Ba one e al., 2013). The h ee classes o ALK mu a ions include a cons i u i ely ac i e gain- o - unc ion ecep o (ligand-independen ), kinase dead mu an s, and ligand-dependen mu an s (Chand e al., 2013). Fusion p o eins in ol ing he in acellula kinase domain o ALK cause se e al ypes o cance and has been a ge ed by kinase inhibi o s. In addi ion, se e al mu a ions in ALK kinase domain esul in d ug esis ance (Roskoski, 2013). The ad an ages o homology modeling o he ex acellula domains include accessibili y on cell su ace and pe haps less c oss in e ac ion han kinase inhibi o s. ALK gene codes o a p o ein o app oxima ely 1620 esidues (Table 1). The c ys al s uc u e o he kinase domain shows he in e ac ions o selec i e d ugs wi h mu an ALK (Sakamo o e al., 2011; Eps ein e al., 2012; F on ie s in Molecula Neu oscience | www. on ie sin.o g 3Janua y 2017 | Volume 10 | A icle 7 Haddad e al. Homology Modeling o Ta ge ing Neu oblas oma FIGURE 1 | Rep esen a ions o he human no epineph ine anspo e (hNET), anaplas ic lymphoma kinase (ALK) and opomyosin ecep o kinase B (T kB) s uc u es ea u ing he mos di e gen egions which equi e p ecise alignmen and possibly loop modeling. Di e gen loops a e shown in ep esen a i e alignmen s wi h closes known homolog s uc u es. (A) hNET. Two di e gen ex acellula loops and wo in acellula loops a e shown. A op iew o hNET is shown on he le . Low-conse ed esidues in binding si es as desc ibed by Ande sen e al. (2015) a e highligh ed in colo s: cen al binding si e (S1), seconda y binding si e (S2) and ex acellula loop 4 (EL4). Red and blue ci cles highligh S1 and S2 si es, espec i ely. (B) ALK ex acellula domains. Conse ed yp ophan in he MAM domains mus be embedded in he cen e o s uc u e (i.e., W288 and W501). Since he alignmen o MAM2 ( egion o W501) was co ec ed manually, u he loop modeling o he egion is equi ed (pa icula ly o P496 and P499). Highligh ed conse ed esidues ha a e binding he calcium ion in he Low-densi y lipop o ein ecep o class A (LDLa) domain a e impo an o alignmen . (C) T kB ex acellula domains. Th ee majo domains o he T kB s uc u e can be used in analysis. The leucine ich epea (LRR) and Ig-like C2 Type 2 egions can be cons uc ed by homology modeling. Two shi s a e shown, ha migh equi e loop modeling. I is possible ha di e gen loops be ween T k p o eins migh play ole in he selec i i y and c oss-in e ac ions wi h neu o ophins. The Ig-like C2 ype 1 domain, di ec ly connec ed o he ansmemb ane domain, has been illus a ed by X- ay c ys allog aphy s udies o he h ee T ks. Fi ing o he h ee T ks showed a di e gen loop a T kB 334–341 egion; possibly playing ole in he selec i i y o hese T ks o di e en neu o ophins. F on ie s in Molecula Neu oscience | www. on ie sin.o g 4Janua y 2017 | Volume 10 | A icle 7 Haddad e al. Homology Modeling o Ta ge ing Neu oblas oma TABLE 1 | Templa es o homology modeling o human no epineph ine anspo e (hNET) and ex acellula domains o anaplas ic lymphoma kinase (ALK) and opomyosin ecep o kinase B (T kB). Ta ge (ID) Leng h Templa e (PDB IDs) Iden i y Co e age Main domains Re e ence hNET (P23975) 617 LeuT (2A65) 150/541 (28%)56−578 All Yamashi a e al. (2005) dDAT (4M48) 320/548 (58%)56−601 All Penma sa e al. (2013) dDAT (4XPA) 322/547 (59%)56−601 All Wang e al. (2015) SERT- s2 (5I6Z) 291/548 (53%)51−595 All Coleman e al. (2016) ALK (Q9UM73) 1620 PTPRM (2C9A and 2V5Y) 25/164 (15%)∗264−427 MAM1 A icescu e al. (2006, 2007) 21/159 (13%)∗481−636 MAM2 MEP1B (4GWM) 19/164 (12%)∗264−427 MAM1 A olas e al. (2012) 17/159 (11%)∗481−636 MAM2 LDLR (2KRI) 15/37 (41%)437−473 LDLa Lee e al. (2010) T kB (Q16620) 822 T kA (2IFG) 89/256 (35%)32−281 Ig-like C2 Type2, LRR Weh man e al. (2007) ∗Many iden i ies we e los due o posi ion-speci ic mul iple alignmen . Huang e al., 2014). All known c ys al models o ALK co e he amino acid ange 1058–1411 and un o una ely he ex acellula s uc u e has no been well s udied. The egion spanning he ex acellula pa o he ecep o ∼266–636 comp ises wo mep in/A5/mu domains also known as he MAM domains (Figu e 1). MAM1 spans 264–427 and MAM2 spans 478–636. App oxima ely 16 N-linked glycosyla ed si es a e eco ded spo adically in he ex acellula egion. C ys al s uc u es o he MAM domain o ecep o p o ein y osine phospha ase MU, also known as PTPRM (A icescu e al., 2006, 2007), and Mep in A be a, also known as MEP1B (A olas e al., 2012) can se e as mul iple empla es o he ex acellula domain o ALK. Al hough many sequence iden i ies a e los due o posi ional alignmen (Table 1), he dis inc MAM domains o PTPRM and MEP1B a e highly simila wi h ∼1.35 Å oo -mean- squa e de ia ion (RMSD) o Cαa oms and ∼1.43 Å RMSD o all a oms. Two loop egions in MAM1 domain equi e loop modeling (Figu e 1B). The o me loop egion con ains conse ed yp ophan (W288) e y well s acked in he cen e o he domain. The equi alen yp ophan (W501) in MAM2 was misaligned in he posi ion-speci ic alignmen and mus be co ec ed o p ope modeling (Figu e 1B). Simila o hNET, glycosyla ed esidues o he MAM domains can be used o check he o ien a ion o side chains as hey should be acing solu ion. A low-densi y lipop o ein ecep o class A (LDLa) domain spans he egion 437–473 be ween he wo MAM domains. The LDLR s uc u e (Lee e al., 2010) is he mos ela ed homolog (Table 1). A leas h ee conse ed aspa ic acid esidues and one glu amic acid coo dina e a calcium ion in he LDLa domain (Figu e 1B). T kB The neu o ophic T ks, assis ed by p75 neu o ophin ecep o , play an essen ial ole in biology o neu ons by media ing neu o ophin-ac i a ed signaling. Neu o ophins include ne e g ow h ac o (NGF), b ain-de i ed neu o ophic ac o (BDNF), neu o ophin-3 (NT-3), neu o ophin-4/5 (NT-4/5), neu o ophin-6 and neu o ophin-7 (Ul sch e al., 1999). The exp ession o ype 1 (T kA) and ype 2 (T kB) ecep o s is associa ed wi h a o able and un a o able p ima y neu oblas oma pa ien ou come, espec i ely (Thiele and Reynolds, 2005). The e a e ou p o ein iso o ms o T kA epo ed in he Unip o da abase (ID: P04629). The longes is 796 esidues in leng h. C ys al s uc u es co e wo la ge segmen s o he ex acellula ; (Robe son e al., 2001; Weh man e al., 2007), and in acellula T kA p o ein (Wang e al., 2012). On he o he hand, ou o se en al e na i e iso o ms, he canonical iso o m o T kB is 822 esidues in leng h (Table 1). The c ys al s uc u es o T kB co e he in acellula kinase domain (Be and e al., 2012) and Ig-like C2 Type 1 domain o he ex acellula egion (PDB IDs: 1WWB, 1HCF; Ul sch e al., 1999; Ban ield e al., 2001). The es o ex acellula segmen o T kB ( esidues 32–281) can be econs uc ed by homology modeling o T kA empla e which has ∼35% iden i y (Table 1). The empla e allows o modeling o wo majo ex acellula domains o T kB; namely, he leucine ich epea (LRR) spanning egion 92–137, and he Ig-like C2 Type 2 domain spanning he esidues 197–281 (Figu e 1C). A leas wo alignmen shi s in he LRR and Ig-like C2 ype 2 domains equi e loop modeling. By supe posing he h ee known s uc u es o Ig-like C2 ype 1 domains o T kA, T kB and T kC (PDB IDs: 1WWA, 1WWB, and 1WWC, espec i ely), a di e gen loop spanning T kB 334–341 egion is highligh ed (Figu e 1C). Along wi h his loop, se e al di e gen esidues migh also play ole in he selec i i y o hese T ks o di e en neu o ophins. Howe e , u he wo k is equi ed o iden i y he exac binding si es o di e en neu o ophins (Ul sch e al., 1999). In ac , he ne wo k o T k-neu o ophin is mo e complex i we assume ha ei he Ig-like C2 Type 2 and LRR domains would be in ol ed in neu o ophins in e ac ions. In ensi e wo k in homology modeling, molecula docking and molecula dynamics is equi ed o shed he ligh on his ne wo k. Simila o he si ua ion in hNET and ALK, inhibi o s a ge ing se e al T k ecep o s a he same ime esul in a ge ing o se e al pa hways and lead o se e al d ug side e ec s. Unde s anding he selec i i y o T ks ecep o s will play signi ican ole in de elopmen o he apeu ics. CONCLUSIONS AND PERSPECTIVES Homology modeling is one o he i s s eps in de eloping he apeu ics o new a ge s. Howe e , as we showed he e he i s s eps a e o en c ucial and de imen al in cons uc ing F on ie s in Molecula Neu oscience | www. on ie sin.o g 5Janua y 2017 | Volume 10 | A icle 7 Haddad e al. Homology Modeling o Ta ge ing Neu oblas oma new homology s uc u es, no o men ion in de eloping new d ugs. Al hough he ules o ‘‘good modeling p ac ice’’ a e no w i en ye , many lessons can be lea ned by e alua ion o he model and co ec ing/a oiding e o s a ea ly s ages o modeling. On he o he hand, a bad quali y empla e will no gi e a good quali y homology model. The quali y and iden i y o he empla e(s) a e e y impo an issues. Low-conse ed egions in he p o ein a ge o en play a signi ican biological ole. They equi e mo e ocus in homology modeling and in a design o new he apeu ics. Se e al s a egies o neu oblas oma he apy ha e been ad ancing in pa allel in he pas ew decades. Ta ge ing he cell su ace molecules is a s a egy ha allows o dis ibu ion o e o s. The dis ibu ion o e o s is de ined by e ec i e esea ch managemen whe e asks a e dis ibu ed among esea che s o p oduce mo e e icien he apeu ics. Indeed, no el cance he apeu ics comp ise complexes o se e al agen s ha ca y ou se e al unc ions. The selec i e a ge ing agen (e.g., pep ide o an ibody) deli e s he ca go (i.e., oxic agen ) o he cance cell, which also equi es a ca ie (soluble/ eleasing agen ) o a memb ane pene a ing agen . In his pe spec i e a icle, he modele has a mo e ocused objec i e, which is o de elop a ‘‘selec i e’’ a ge ing molecule. We hope ha hese s a egies will p oduce mo e adap able, e icien and pe sonalized he apeu ics in he u u e. AUTHOR CONTRIBUTIONS All au ho s con ibu ed o he design o wo k. YH w o e he manusc ip . ZH e iewed he manusc ip , and VA was p inciple in es iga o and con ibu o o scheme and o ganiza ion o wo k. FUNDING We g a e ully acknowledge he Czech Agency o Heal hca e Resea ch, AZV (15-28334A) and Minis y o Educa ion, You h and Spo s o he Czech Republic unde he p ojec CEITEC 2020 (LQ1601) o inancial suppo o his wo k. The compu a ional esou ces we e p o ided by he CESNET LM2015042 and he CERIT Scien i ic Cloud LM2015085, p o ided unde he p og amme ‘‘P ojec s o La ge Resea ch, De elopmen and Inno a ions In as uc u es’’. REFERENCES Al-Lazikani, B., Jung, J., Xiang, Z. X., and Honig, B. (2001). P o ein s uc u e p edic ion. Cu . Opin. Chem. Biol. 5, 51–56. doi: 10.1016/S1367- 5931(00)00164-2 Ande sen, J., Rings ed, K. B., Bang-Ande sen, B., S ømgaa d, K., and K is ensen, A. S. (2015). Binding si e esidues con ol inhibi o selec i i y in he human no epineph ine anspo e bu no in he human dopamine anspo e . Sci. Rep. 5:15650. doi: 10.1038/s ep15650 A icescu, A. R., Hon, W. C., Siebold, C., Lu, W., an de Me we, P. A., and Jones, E. Y. (2006). Molecula analysis o ecep o p o ein y osine phospha ase µ-media ed cell adhesion. EMBO J. 25, 701–712. doi: 10.1038/sj.emboj. 7600974 A icescu, A. R., Siebold, C., Choudhu i, K., Chang, V. T., Lu, W., Da is, S. J., e al. (2007). S uc u e o a y osine phospha ase adhesi e in e ac ion e eals a space -clamp mechanism. Science 317, 1217–1220. doi: 10.1126/science. 1144646 A olas, J. L., B ode , C., Je e son, T., Gue a a, T., S e chi, E. E., Bode, W., e al. (2012). S uc u al basis o he sheddase unc ion o human mep in β me allop o einase a he plasma memb ane. P oc. Na l. Acad. Sci. U S A 109, 16131–16136. doi: 10.1073/pnas.1211076109 Ban ield, M. J., Naylo , R. L., Robe son, A. G. S., Allen, S. J., Dawba n, D., and B ady, R. L. (2001). Speci ici y in T k ecep o : neu o ophin in e ac ions: he c ys al s uc u e o T kB-d5 in complex wi h neu o ophin-4/5. S uc u e 9, 1191–1199. doi: 10.1016/S0969-2126(01)00681-5 Ba one, G., Ande son, J., Pea son, A. D. J., Pe ie, K., and Chesle , L. (2013). New s a egies in neu oblas oma: he apeu ic a ge ing o MYCN and ALK. Clin. Cance Res. 19, 5814–5821. doi: 10.1158/1078-0432.cc -13-0680 Be and, T., Ko he, M., Liu, J., Dupuy, A., Rak, A., Be ne, P. F., e al. (2012). The c ys al s uc u es o T kA and T kB sugges key egions o achie ing selec i e inhibi ion. J. Mol. Biol. 423, 439–453. doi: 10.1016/j.jmb.2012.08.002 Bo dne , A. J. (2012). Fo ce ields o homology modeling. Me hods Mol. Biol. 857, 83–106. doi: 10.1007/978-1-61779-588-6_4 Bo doli, L., Kie e , F., A nold, K., Benke , P., Ba ey, J., and Schwede, T. (2009). P o ein s uc u e homology modeling using SWISS-MODEL wo kspace. Na . P o oc. 4, 1–13. doi: 10.1038/np o .2008.197 B odeu , G. M., P i cha d, J., Be hold, F., Ca lsen, N. L. T., Cas el, V., Cas lebe y, R. P., e al. (1993). Re isions o he in e na ional c i e ia o neu oblas oma diagnosis, s aging and esponse o ea men . J. Clin. Oncol. 11, 1466–1477. Bu le , K. V., Kalin, J., B ochie , C., Vis oli, G., Langley, B., and Kozikowski, A. P. (2010). Ra ional design and simple chemis y yield a supe io , neu op o ec i e HDAC6 inhibi o , Tubas a in A. J. Am. Chem. Soc. 132, 10842–10846. doi: 10.1021/ja102758 Chand, D., Yamazaki, Y., Ruu h, K., Schönhe , C., Ma insson, T., Kogne , P., e al. (2013). Cell cul u e and D osophila model sys ems de ine h ee classes o anaplas ic lymphoma kinase mu a ions in neu oblas oma. Dis. Model. Mech. 6, 373–382. doi: 10.1242/dmm.010348 Chen, V. B., A endall, W. B., Headd, J. J., Keedy, D. A., Immo mino, R. M., Kap al, G. J., e al. (2010). MolP obi y: all-a om s uc u e alida ion o mac omolecula c ys allog aphy. Ac a C ys allog . D Biol. C ys allog . 66, 12–21. doi: 10.1107/S0907444909042073 Coleman, J. A., G een, E. M., and Gouaux, E. (2016). X- ay s uc u es and mechanism o he human se o onin anspo e . Na u e 532, 334–339. doi: 10.1038/na u e17629 DeVo e, N. M., and Sco , E. E. (2012). S uc u es o cy och ome P450 17A1 wi h p os a e cance d ugs abi a e one and TOK-001. Na u e 482, 116–119. doi: 10.1038/na u e10743 Eps ein, L. F., Chen, H., Emkey, R., and Whi ing on, D. A. (2012). The R1275Q neu oblas oma mu an and ce ain ATP-compe i i e inhibi o s s abilize al e na i e ac i a ion loop con o ma ions o anaplas ic lymphoma kinase. J. Biol. Chem. 287, 37447–37457. doi: 10.1074/jbc.M112. 391425 Ga cia, J., Faca, V., Ja zembowski, J., Zhang, Q., Pa k, J., and Hanash, S. (2009). Comp ehensi e p o iling o he cell su ace p o eome o Sy5Y neu oblas oma cells yields a subse o p o eins associa ed wi h umo di e en ia ion. J. P o eome Res. 8, 3791–3796. doi: 10.1021/p 800964 Haddad, Y., Hege , Z., and Adam, V. (2016). Guidelines o homology modeling o dopamine, no epineph ine and se o onin anspo e s. ACS Chem. Neu osci. 7, 1607–1613. doi: 10.1021/acschemneu o.6b00242 Huang, Q., Johnson, T. W., Bailey, S., B ooun, A., Bunke , K. D., Bu ke, B. J., e al. (2014). Design o po en and selec i e inhibi o s o o e come clinical anaplas ic lymphoma kinase mu a ions esis an o c izo inib. J. Med. Chem. 57, 1170–1187. doi: 10.1021/jm401805h Kakuda, S., Shiba, T., Ishigu o, M., Tagawa, H., Oka, S., Kajiha a, Y., e al. (2004). S uc u al basis o accep o subs a e ecogni ion o a human glucu onyl ans e ase, GlcAT-P, an enzyme c i ical in he biosyn hesis o he ca bohyd a e epi ope HNK-1. J. Biol. Chem. 279, 22693–22703. doi: 10.1074/jbc.M400622200 F on ie s in Molecula Neu oscience | www. on ie sin.o g 6Janua y 2017 | Volume 10 | A icle 7 Haddad e al. Homology Modeling o Ta ge ing Neu oblas oma Kinch, M. S., Hoye , D., Pa idge, E., and Plumme , M. (2015). Ta ge selec ion o FDA-app o ed medicines. D ug Disco . Today 20, 784–789. doi: 10.1016/j. d udis.2014.11.001 Koldsø, H., Ch is iansen, A. B., Sinning, S., and Schiø , B. (2013). Compa a i e modeling o he human monoamine anspo e s: simila i ies in subs a e binding. ACS Chem. Neu osci. 4, 295–309. doi: 10.1021/cn300148 Koldsø, H., G oule , J., and Schiø , B. (2015). Insigh s o ligand binding o he monoamine anspo e s- om homology modeling o LeuBAT and dDAT. F on . Pha macol. 6:208. doi: 10.3389/ pha .2015.00208 K iege , E., Nabuu s, S. B., and V iend, G. (2003). Homology modeling. Me hods Biochem. Anal. 44, 509–523. doi: 10.1007/sp inge e e ence_33987 Kund o as, P. J., Zhu, Z., Janin, J., and Vakse , I. A. (2012). Templa es a e a ailable o model nea ly all complexes o s uc u ally cha ac e ized p o eins. P oc. Na l. Acad. Sci. U S A 109, 9438–9441. doi: 10.1073/pnas.1200678109 Lee, C. J., De Biasio, A., and Beglo a, N. (2010). Mode o in e ac ion be ween be a 2GPI and lipop o ein ecep o s sugges s mu ually exclusi e binding o be a 2GPI o he ecep o s and anionic phospholipids. S uc u e 18, 366–376. doi: 10.1016/j.s .2009.12.013 Ma hay, K. K., Geo ge, R. E., and Yu, A. L. (2012). P omising he apeu ic a ge s in neu oblas oma. Clin. Cance Res. 18, 2740–2753. doi: 10.1158/1078-0432.CCR- 11-1939 Melikian, H. E., McDonald, J. K., Gu, H., Rudnick, G., Moo e, K. R., and Blakely, R. D. (1994). Human no epineph ine anspo e –biosyn he ic-s udies using a si e-di ec ed polyclonal an ibody. J. Biol. Chem. 269, 12290–12297. Mo ensen, O. V., and Ko age e, S. (2015). Designing modula o s o monoamine anspo e s using i ual sc eening echniques. F on . Pha macol. 6:223. doi: 10.3389/ pha .2015.00223 No egaa d, L., F ede iksen, D., Nielsen, E. O., and Ge he , U. (1998). Delinea ion o an endogenous zinc-binding si e in he human dopamine anspo e . EMBO J. 17, 4266–4273. doi: 10.1093/emboj/17.15.4266 Osajima-Hakomo i, Y., Miyake, I., Ohi a, M., Nakagawa a, A., Nakagawa, A., and Sakai, R. (2005). Biological ole o anaplas ic lymphoma kinase in neu oblas oma. Am. J. Pa hol. 167, 213–222. doi: 10.1016/s0002- 9440(10)62966-5 Penma sa, A., Wang, K. H., and Gouaux, E. (2013). X- ay s uc u e o dopamine anspo e elucida es an idep essan mechanism. Na u e 503, 85–90. doi: 10.1038/na u e12533 Robe son, A. G. S., Ban ield, M. J., Allen, S. J., Dando, J. A., Mason, G. G. F., Tyle , S. J., e al. (2001). Iden i ica ion and s uc u e o he ne e g ow h ac o binding si e on T kA. Biochem. Biophys. Res. Comm. 282, 131–141. doi: 10.1006/bb c.2001.4462 Roskoski, R. J . (2013). The p eclinical p o ile o c izo inib o he ea men o non-small-cell lung cance and o he neoplas ic diso de s. Expe Opin. D ug Disco . 8, 1165–1179. doi: 10.1517/17460441.2013.813015 Sakamo o, H., Tsukaguchi, T., Hi oshima, S., Kodama, T., Kobayashi, T., Fukami, T. A., e al. (2011). CH5424802, a selec i e ALK inhibi o capable o blocking he esis an ga ekeepe mu an . Cance Cell 19, 679–690. doi: 10.1016/j.cc .2011.04.004 Sa ele , H., Imb iglio, T., Nyalendo, C., Haddad, E., Annabi, B., Du al, M., e al. (2012). CD133 exp ession is associa ed wi h poo ou come in neu oblas oma ia chemo esis ance media ed by he AKT pa hway. His opa hology 60, 1144–1155. doi: 10.1111/j.1365-2559.2012.04191.x Schlessinge , A., Geie , E., Fan, H., I win, J. J., Shoiche , B. K., Giacomini, K. M., e al. (2011). S uc u e-based disco e y o p esc ip ion d ugs ha in e ac wi h he no epineph ine anspo e , NET. P oc. Na l. Acad. Sci. U S A 108, 15810–15815. doi: 10.1073/pnas.1106030108 Schli e , A. M., Do nebu g, C., Ba h, T. F. E., Wahl, J., Schul e, J. H., B üde lein, S., e al. (2012). CD57 high neu oblas oma cells ha e agg essi e a ibu es ex si u and an undi e en ia ed pheno ype in pa ien s. PLoS One 7:e42025. doi: 10.1371/jou nal.pone.0042025 Sco , A. M., Wolchok, J. D., and Old, L. J. (2012). An ibody he apy o cance . Na . Re . Cance 12, 278–287. doi: 10.1038/n c3236 Shin, M. C., Zhang, J., Min, K. A., Lee, K., Byun, Y., Da id, A. E., e al. (2014). Cell-pene a ing pep ides: achie emen s and challenges in applica ion o cance ea men . J. Biomed. Ma e . Res. A 102, 575–587. doi: 10.1002/jbm.a. 34859 Siegel, R. L., Mille , K. D., and Jemal, A. (2016). Cance s a is ics, 2016. CA Cance J. Clin. 66, 7–30. doi: 10.3322/caac.21332 Thiele, C. J., and Reynolds, C. P. (2005). ‘‘Di e en ia ion and e inoids,’’ in Neu oblas oma, eds N. K. V. Cheung and S. L. Cohn (New Yo k, NY: Sp inge ), 243–256. To o , M. (2012). Loop simula ions. Me hods Mol. Biol. 587, 207–229. doi: 10.1007/978-1-61779-588-6_9 Ul sch, M. H., Wiesmann, C., Simmons, L. C., Hen ich, J., Yang, M., Reilly, D., e al. (1999). C ys al s uc u es o he neu o ophin-binding domain o T kA, T kB and T kC. J. Mol. Biol. 290, 149–159. doi: 10.1006/jmbi.1999.2816 Ve issimo, C. S., Molenaa , J. J., Fi zsimons, C. P., and V eugdenhil, E. (2011). Neu oblas oma he apy: Wha is in he pipeline? Endoc . Rela . Cance 18, R213–R231. doi: 10.1530/ERC-11-0251 Wang, T., Lamb, M. L., Block, M. H., Da ies, A. M., Han, Y., Ho mann, E., e al. (2012). Disco e y o disubs i u ed imidazo [4,5-b] py idines and pu ines as po en T kA inhibi o s. ACS Med. Chem. Le . 3, 705–709. doi: 10.1021/ml300074j Wang, K. H., Penma sa, A., and Gouaux, E. (2015). Neu o ansmi e and psychos imulan ecogni ion by he dopamine anspo e . Na u e 521, 322–327. doi: 10.1038/na u e14431 Webb, B., and Sali, A. (2014). Compa a i e p o ein s uc u e modeling using modelle . Cu . P o oc. Bioin o ma ics 47, 5.6.1–5.6.32. doi: 10. 1002/0471250953.bi0506s47 Weh man, T., He, X., Raab, B., Dukipa i, A., Blau, H., and Ga cia, K. C. (2007). S uc u al and mechanis ic insigh s in o ne e g ow h ac o in e ac ions wi h he T kA and p75 ecep o s. Neu on 53, 25–38. doi: 10.1016/j.neu on.2006. 09.034 W igh , K. E., Hje ild, K. A., Ba le , J., Douglas, A. D., Jin, J., B own, R. E., e al. (2014). S uc u e o mala ia in asion p o ein RH5 wi h e y h ocy e basigin and blocking an ibodies. Na u e 515, 427–430. doi: 10.1038/na u e 13715 Yamashi a, A., Singh, S. K., Kawa e, T., Jin, Y., and Gouaux, E. (2005). C ys al s uc u e o a bac e ial homologue o Na+/Cl–dependen neu o ansmi e anspo e s. Na u e 437, 215–223. doi: 10.1038/na u e 03978 Con lic o In e es S a emen : The au ho s decla e ha he esea ch was conduc ed in he absence o any comme cial o inancial ela ionships ha could be cons ued as a po en ial con lic o in e es . Copy igh © 2017 Haddad, Hege and Adam. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License (CC BY). The use, dis ibu ion and ep oduc ion in o he o ums is pe mi ed, p o ided he o iginal au ho (s) o licenso a e c edi ed and ha he o iginal publica ion in his jou nal is ci ed, in acco dance wi h accep ed academic p ac ice. No use, dis ibu ion o ep oduc ion is pe mi ed which does no comply wi h hese e ms. F on ie s in Molecula Neu oscience | www. on ie sin.o g 7Janua y 2017 | Volume 10 | A icle 7