Computer-aided subsite mapping of alpha-amylases

Else ie Edi o ial Sys em( m) o Ca bohyd a e Resea ch Manusc ip D a Manusc ip Numbe : CAR-D-10-00518R1 Ti le: Compu e -aided subsi e mapping o α-amylases A icle Type: Full Leng h A icle Sec ion/Ca ego y: Biochemis y and Enzymes Keywo ds: alpha-amylase; subsi e mapping; binding ene gy; bond clea age equency; molecula modeling Co esponding Au ho : D . Pé e Bagossi, Ph.D. Co esponding Au ho 's Ins i u ion: Uni e si y o Deb ecen Fi s Au ho : János A Mó yán O de o Au ho s: János A Mó yán; Gyöngyi Gyémán , Ph.D.; János Ha angi, Ph.D.; Pé e Bagossi, Ph.D. Abs ac : Subsi e mapping is a c ucial p ocedu e in cha ac e iza ion o α-amylases (EC 3.2.1.1) which a e ex ensi ely used in s a ch based indus ies and in diagnosis o panc ea ic and sali a y glands diso de s. A compu e -aided me hod has been de eloped o subsi e mapping o α-amylases, which subs i u es he di icul , expensi e and ime-consuming expe imen al de e mina ion o ac ion pa e ns o c ys al s uc u es based ene gy calcula ions. In e ac ion ene gies be ween enzymes and ca bohyd a e subs a es we e calcula ed a e sho ene gy minimiza ion by a molecula mechanics p og am. A aining se o wild ype and mu an amylases wi h known expe imen al ac ion pa e ns o 13 enzymes o wide ange o o igin was used o se up he p ocedu e. Calcula ions o aining se esul ed in good co ela ion in case o subsi e binding ene gies ( 2 = 0.827−0.929) and bond clea age equencies ( 2 = 0.727−0.835). A se o eigh no el ba ley amylase 1 mu an s was used o es ou model. Subsi e binding ene gies we e p edic ed wi h 2 = 0.502 co ela ion coe icien , while bond clea age equency p edic ion esul ed in 2 = 0.538. Ou compu e -aided p ocedu e may supplemen he expe imen al subsi e mapping me hods o p edic and unde s and cha ac e is ic ea u es o α- amylases. Re iewe #1: Recommenda ion: Mino Re ision This manusc ip epo s he use o compu a ion o de e mine bond clea age equencies and subsi e binding ene gies om c ys al s uc u es o a numbe o α-amylases. This is an in e es ing choice o enzymes, as α-amylases a e ex ensi ely used indus ially and a g ea numbe o membe s o hei amily (GH13) ha e been sequenced and subjec ed o c ys al s uc u e de e mina ion. They a e also endo-hyd olases, so di e en posi ions o s a ch chains can be clea ed. The a gumen is made ha use o compu a ion sa es a g ea deal o ime and money compa ed o ob aining he same da a expe imen ally. This is undoub edly ue; ce ainly subsi e mapping by expe imen al means is a long and ha d p ocess, as we can a es . On he o he hand, compu a ion mus always ake a back sea o expe imen a ion i he la e can be used o ob ain he da a. A e all, compu a ion is subjec o e o s in oduced by incomple e and inaccu a e modeling o al eady exis ing expe imen al da a. Compa ison o expe imen al and compu a ional da a in his manusc ip con i ms he s a emen jus abo e – he ag eemen be ween hem is o en a he poo . We comple ely ag ee wi h he e iewe he e o e we emphasized h oughou he manusc ip ha ou me hod may complemen expe imen al da a bu no subs i u e hem. One o he easons o choose ela i ely limi ed da a se s was ha we wan ed o de elop and apply ou me hod o close ela i es o enzymes whe e accu acy o modeling was usually highe han hose o dis an ela i es. I he i s y p oduce sa is ac o y esul s, we may ex end he scope o he me hod la e . Thi d, we also p edic ed he possible easons whe e ou me hod was ailed and a special a en ion was needed du ing he wo k. Despi e his, he manusc ip is p obably wo h publishing as a i s a emp o complemen expe imen al subsi e mapping. The English he e is in gene al clea bu is ce ainly he p oduc o non-na i e English speake s. I will need some ew i ing and co ec ing by a p oduc ion edi o . Following a e speci ic commen s as I ead he manusc ip : p. 3, line 11: A e plan s no highe o ganisms? The sen ence was co ec ed o he ollowing: "They can be ound in mic oo ganisms, plan s, animals and human whe e hey play a dominan ole in ca bohyd a e me abolism." p. 3, line 44: Is Domain C a ca bohyd a e-binding module? I so, wha CBM amilies a e ound in he α-amylases s udied he e? *Response o Re iewe s Domain C is no a ca bohyd a e-binding module (CBM) and enzymes o ou da a se s do no con ain any well-de ined sepa a e CBM o s a ch-binding domain (SBD). α-Amylases belonging o GH-13 amily a e mul idomain p o eins ha con ain se e al cha ac e is ic domains such as he obliga o y ca aly ic domain A and mos o hem possess a domain B and domain C. Some enzymes in GH 13 con ain one o wo addi ional all-β domains: domain D and/o domain E, a he C- e minal end, ollowing he domain C (Janecek e al., 2003, Eu . J. Biochem. 270, 635–645). The unc ion o domain D is unknown; howe e , domain E is e e ed o as SBD ( he e m o SBD s ill o en used in he amylase esea ch ins ead o he mo e "o icial" CBM) which is a dis inc sequence-s uc u al module ha imp o es he e iciency o an amyloly ic enzyme on aw s a ch. Fou - and i e-domain membe s o GH-13 can be e e ed o gene ally as he SBD-con aining hyd olases and hey can be classi ied in o a ious CBM amilies. In con as o he ac ha none o he enzymes s udied he e con ain domain D and domain E, hey may con ain seconda y binding si es o ca bohyd a es in domains A, B o C. Fo example, he ba ley -amylase 1 (AMY1) p o ides wo binding si es in addi ion o he ca aly ic cle : a s a ch g anule binding si e wi hin he ca aly ic domain A and a so-called suga ongs wi hin domain C (Nielsen e al., 2008, Bioca al. Bio ans ., 26, 59-67). HSA has also su ace binding si es loca ed a om he ac i e si e (Ramasubbu e al., 2003, J. Mol. Biol., 325, 1061-1076). These enzymes a e capable o binding and diges ing aw s a ch wi hou a specialized unc ional domain (e.g. SBD) in hei sequence and s uc u e (Janecek e al., 2003, Eu . J. Biochem. 270, 635–645). The manusc ip was modi ied acco dingly: "Besides domain A, B and C, he ype and he numbe o ex a domains such as domain D and/o domain E loca ed a he C- e minus show wide a ie y wi hin he α-amylase amily. The unc ion o domain D is unknown; howe e , domain E is e e ed o as ca bohyd a e-binding module (CBM) o s a ch-binding domain (SBD) which is a dis inc sequence-s uc u al module ha imp o es he e iciency o an amyloly ic enzyme on aw s a ch. I should be no ed ha none o he enzymes s udied he e con ain domain D and/o domain E." p. 3, lines 53–56: Is i no ue ha α-amylases mainly ac on α-(1,4)-linked glucans? Yes, i is ue. Families 13, 70 and 77 o glycoside hyd olases con ain s uc u ally and unc ionally ela ed enzymes ca alyzing hyd olysis o ansglycosyla ion o α-linked glucans, wi h e en ion o anome ic con igu a ion. Alpha-amylases ca alyze he endohyd olysis o (1→4)-α-D-glycosidic linkages in polysaccha ides con aining h ee o mo e (1→4)-α-linked D-glucose uni s and hey ac on s a ch, glycogen and ela ed polysaccha ides and oligosaccha ides in a andom manne . All hese in o ma ion can be ound in he i s wo pa ag aphs o he manusc ip . Ne e heless we eph ased he abo e men ion sen ence o he manusc ip o he ollowing: "Membe s o he α-amylase amily ac on se e al subs a es (s a ch, glycogen, oligosaccha ides) and he only sha ed s uc u al ea u e o subs a es is an α-(1-4)- linked glucose esidue ha should bind in he i s aglycone subsi e nea he scissile bond." p. 5, line 35: I would be help ul o no e ha hese PDB s uc u es we e o mal ose and mal ohep aose c ys allized wi h α-amylases. The sen ence was modi ied o con ain his in o ma ion: "S uc u es o he mal ooligosaccha ide subs a es we e modeled based on he c ys al s uc u es o mal ose (PDB code: 2GVY), mal ohep aose (PDB code: 1RP8) and a subs a e-analogue inhibi o aca bose (PDB code: 1MFU, 1RPK, 1E3Z, 1OSE) complexed wi h a ious alpha-amylases." p. 5, line 42: Wha does “bumped” mean he e? C ys al s uc u es used o modeling o enzymes o ou da a se s con ained sho e subs a e o inhibi o han ha o ou s, he e o e wa e molecules which occupied an "emp y space" in he o iginal x- ay s uc u e may ha e a e y close con ac o he long subs a e o he model (dis ance is much sho e han he sum o he an de Waals adius o he wo a oms). No su p isingly, i happened ha oxygen a om o a hyd oxyl g oup o ou model subs a e occupied app oxima ely he same posi ion as he oxygen a om o a wa e molecule in he c ys al s uc u e. In his case we emo ed ha wa e molecule o a oid dis o ion caused by he huge ini ial ene gy du ing he minimiza ion. p. 7, line 23: Excep o he lis o abb e ia ions, his is he i s ime ha E ans . has been men ioned. How is i calcula ed? The sen ence was modi ied o explain he calcula ion: "A e eg ession analysis, ESybyl da a we e linea ly ans o med o he scale o ESUMA da a wi h he alues o in e cep and slope o he bes i line o ge ET ans . (Fig. 2A and 2B) o each enzyme g oup." p. 8, line 31: A able o igu e showing calcula ed BCF’s would be help ul. The able con aining all BCF da a would be oo la ge he e because BCFs o se ies o oligome ic subs a es used o calcula e a single se o binding ene gies. Fu he mo e, he BCF da a will be published soon in a sepa a e pape (Mo i e al., in p epa a ion). We may show an example o he co ela ion o he expe imen al and he calcula ed da a, bu hese g aph a e s ill con ained la ge numbe o poin s and hey look a li le bi messy. On he o he hand, hese g aphs show basically he same quali a i e message as he Fig. 2 bu i he Re iewe sugges s including any o hem in o he manusc ip , we will do i . New igu e? Co ela ion o expe imen al and calcula ed BCF alues o wild- ype and mu an AMY1 enzymes: a) aining se ( 2 = 0.801) and b) es se ( 2 = 0.538). Majo i y o poin s loca ed ou side o he 95 % p edic ion in e al ange a e belonging o mu an con aining cha ged esidue. To show his, he es se is di ided in o wo classes: c) da a o es se wi hou hose o mu an con aining cha ged esidues ( 2 = 0.728) and d) da a o cha ged mu an only. a) b) c) d) Re e ences: Jou nal i les should be abb e ia ed when hey a e o mo e han one wo d. Re e ences a e e o ma ed. Table 1: Wha is being co ela ed he e? Subsi e binding ene gies we e ecalcula ed using ou subsi e models and his able shows he excellen co ela ion be ween he published ene gy alues and ou ecalcula ed ones. This was explained only in he ex he e o e legend o Table 1 was modi ied o he ollowing: "O iginal and modi ied subsi e models o he s udied enzymes and he co ela ion be ween he subsi e binding ene gies o he o iginal publica ions and he ecalcula ed alues based on subsi e models o his wo k." Table 2: Indi idual subsi e binding ene gies a e e y di e en be ween enzymes. Can a s a emen be made in he ex linking his obse a ion o he chain leng hs o cha ac e is ic p oduc s o hese enzymes? The ollowing sen ences we e inse ed in o he In oduc ion: "The dis ibu ion o he high and low a ini y binding subsi es and he ba ie subsi es wi hin he ac i e si e de e mines he hyd oly ic e iciency on se ies o subs a es wi h a ious leng h: high-a ini y subsi es nex o he clea age si es (−2 h ough +2) allow e ec i e clea age o sho e subs a es, while dis al high-a ini y subsi es con ol he ac ion on longe subs a es. Chain leng h and concen a ion o cha ac e is ic p oduc s can be de e mined om he BCF ables and ene gy con ibu ion o each subsi e o he o al binding ene gy can be calcula ed." Table 3: The able legend men ions ene gies bu no BCF’s. Legend o Table 3 was co ec ed o he ollowing: "Pa ame e s o he bes i ed line o linea eg ession analysis o ESUMA and ESybyl alues (middle panel) and he expe imen al and calcula ed alues o BCFs ( igh panel) (m − slope, b − in e cep , 2 − squa e o co ela ion coe icien )." Table 4: This able is no men ioned un il he Conclusions. I should ha e been men ioned ea lie . The ables we e enumbe ed acco ding o he o de o appea ance (ea lie Table 4 now is Table 1) and he ollowing ex was inse ed in o he i s pa ag aph o he Resul and discussion: "Wild- ype enzymes we e chosen based on he exis ence o expe imen al BCF alues as well as a ailable c ys al s uc u e. In con as o he a ious deg ee o sequence iden i y (11-86 %) be ween he wild- ype enzymes (Table 1), hey we e all classi ied in o he glycoside hyd olase amily 13 and showed high deg ee o s uc u al conse a ion." Figu e 2A: A plo o kcal/mol s. kJ/mol does no allow he eade o easily de e mine whe he he slope o he eg ession line is uni y. Fig. 2A shows only he aw da a and i is no aimed o demons a e ha he slope is uni y. E en i he kcal/mol is con e ed o kJ/mol, nobody can excep ha he absolu e scale o he expe imen al and he calcula ed ene gy alues would be he same. Du ing he p ocess o he calcula ion se e al assump ions we e made, se e al ac o s we e neglec ed, only pa ial minimiza ions we e done, e c. This was he eason ha he ans o ma ion s ep was needed and he uni con e sion was also included in his s ep. Re iewe #2: Recommenda ion: Mino Re ision The manusc ip by Mo yan e al. epo s on an applica ion o hei SUMA algo i hm o he p oblem o subsi e mapping in alpha-amylases. The pape is ela i ely well w i en and add esses a mode a ely in e es ing p oblem in he ield. The au ho s ha e collec ed a nice aining and es se o da a, and hen p oceed o apply hei me hodology o op imize empi ical pa ame e s o i he da a. This is a e y common p ac ice and me hodologically, his pape does no p esen any hing new. As such, he mos in e es ing esul in he pape is he model speci ically ained o alpha-amylases. The e a e majo weaknesses in he pape ha should howe e be add essed be o e publishing. 1.) The size o he aining and es se s a e s ill small. Ideally, he au ho s should y o a leas double he size o he aining se o be able o pe o m mo e comp ehensi e c oss- alida ion. The i s a e in ac no c oss- alida ed, and he au ho s do no a all add ess whe he hey may in ac ha e o e i hei model. This was ou i s a emp o s udy on in silico subsi e maps o ca bohyd a e- modi ying enzymes and simila wo k had no been published in he li e a u e, he e o e we decided o limi ou s udy o he ela i ely well cha ac e ized, s uc u ally simila , ca aly ically uni o m g oup o alpha-amylases. They s ill ep esen ed a wide ange o enzymes o a ious species om bac e ia o human: he sequence iden i y alues we e 11-86 % (Table 1 o he e ised manusc ip ). We mus limi ou aining se o he enzymes which expe imen ally de e mined BCF and c ys al s uc u e we e bo h exis ed. We included all published expe imen al da a in ou aining se excep wo: BCF and subsi e map o ice amylase we e published (Gyémán e al., Eu . J. Biochem., 2002, 269, 5157-5162) bu i s c ys al s uc u e has no been de e mined ye and c ys al s uc u e o mal ogenic α-amylase o Bacillus s ea o he mophilus was de e mined bu BCF da a has no been published ye . We admi ha he size o ou da a se did no allow o handle all aspec s in s a is ically sa is ac o y way bu we hink ha ou esul s and conclusions a e co ec and no go beyond he scope o his s udy. 2.) Why a e he au ho s ocusing only on alpha-amylases? Why no pe o m his analysis on all amylases, which would g ea ly inc ease he size o hei aining se and allow o much mo e 10- old c oss alida ion s udies. The mos known amyloly ic enzymes a e α-amylase (EC 3.2.1.1), β-amylase (EC 3.2.1.2) and glucoamylase (EC 3.2.1.3); howe e , hey di e om each o he in hei p ima y and e ia y s uc u es, hei ca aly ic machine ies and eac ion mechanisms. They ha e he e o e classi ied in o di e en amilies: GH13 - α-amylases, GH14 - β-amylases, GH15 - glucoamylases. I may possible ha ou me hod can be applied no only o GH13 bu u he amilies oo; howe e , i needs u he s udies and i is beyond he scope o his manusc ip . SUMA p og am was used in se e al s ep o ou me hod and his p og am was de eloped and alida ed only o α-amylases, as i s ull name in he Abb e ia ions and in he ex indica es: SUbsi e Mapping o α-Amylases. Please also see he answe o he i s ques ion o his e iewe . 3.) The au ho s a e using he AMBER o ce ield, ye he e a e a mo e accu a e o ce- ields o suga s, e.g. glycam. The au ho s should e alua e how using di e en o ce- ields a ec hei esul s. Fu he mo e, he au ho s should also e alua e whe he using explici sol en in hei simula ions imp o es (o pe haps wo sens) hei i s o he expe imen al da a. Se e al o ce ields a ailable in Sybyl p og am: Kollman_All_A om, Ambe 95 and Ambe 7_FF99 we e es ed, oge he wi h an in-house implemen a ion o Glycam06 o ce ield. Ambe 7_FF99 was chosen because i s pe o mance was he bes in ou hand. Fu he mo e, wide- a ie y o alues o dielec ic cons an (1, 2, 3, 4 and 8), non-bonded cu o (8, 10 and 12), numbe o i e a ions (20-100 by s eps o 20 and 100-1000 by s eps o 100) and he wid h o minimized shell (1-12 angs om) o a oms a ound he subs a e and he mu a ed esidues in he ini ial ene gy minimiza ion we e also p obed o ind a good combina ion o pa ame e s and i was u ned ou ha hese ac o s also signi ican ly modi ied he esul s. Se e al, bu no all possible combina ions we e es ed because i would equi e la ge amoun o compu a ional powe , he e o e ex ensi e discussion on he pa ame e choice was neglec ed. One o ou aims was o build a as p ocedu e; he e o e explici molecules o bulk sol en and ex ensi e minimiza ion we e a oided. We hink ha he eade s o Ca bohyd a e Resea ch a e mo e in e es ing in he chemical/biochemical aspec s o ou me hod a he han he compu a ional de ails o he pa ame e choice. The ex o he manusc ip was modi ied acco dingly. 4.) The expe imen al da a used o i ha e been collec ed unde a wide- a ie y o solu ion condi ions. The au ho s need o discuss and assess how expe imen al e o may in ac be limi ing he esolu ion o hei me hodology. The wild- ype enzymes we e measu ed a a ious eac ion condi ions; howe e , hey we e measu ed a he op imal condi ion o each enzyme and his si ua ion may be mo e app op ia e o de e mina ion o bond clea age equencies and subsi e binding ene gies han he choice o one eac ion condi ion which is op imal o only one enzyme and subop imal o he all o he s. De ini i e conclusion o he e ec o each eac ion condi ion (e.g. pH, empe a u e o bu e composi ion) can be d awn only a e ex ensi e expe imen a ion. Mu an enzymes we e measu ed a he same eac ion condi ion han hei wild- ype enzymes based on he assump ion ha hei ea u es we e changed mode a ely. The same assump ion was also necessa y o he heo e ical calcula ions. Fu he mo e, he e ec o di e en eac ion condi ions could be elimina ed in a la ge ex en wi h ou ene gy ans o ma ion p ocedu e which was g oup-speci ic. Enzyme g oups we e based on each wild- ype enzymes and mu an s belonged o he co esponding wild- ype g oup. Di e ences o he calcula ed subsi e binding ene gies we e below 0.4 kJ/mol in case o wo independen expe imen al measu emen s o BCF, as we w o e in he Ma e ials and me hods. In a ecen publica ion o us i was ound ha he same ype o e o was below 0.6 kJ/mol (Nielsen e al., 2009, Biochemis y 48, 7686–7697). 5a.) I is no clea how hei esul s will be sha ed wi h he wo ld. Can we access hei pa ame e s on a web se e o es wi h o he amylases? 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 5 Compu e -aided analysis o X- ay c ys allog aphic s uc u es is an e ec i e way o examine enzyme-subs a e in e ac ions and i could be supplemen ed by molecula mechanical/dynamical calcula ions o p edic he in e ac ion in a quan i a i e manne . In his wo k, we aimed o de elop a compu e -aided p ocedu e o subsi e mapping o α-amylases o p edic and in e p e unc ional ea u es o wild ype and mu an α-amylases. 2. Ma e ials and me hods 2.1. Calcula ion o subsi e binding ene gies and BCFs by SUMA Expe imen ally de e mined BCFs we e collec ed om he li e a u e o each enzyme and based on hem, numbe o subsi es, posi ion o he clea age si e and binding ene gy (ESUMA) o each subsi e we e calcula ed by SUMA. Appa en ee ene gy alues we e op imized in a SUMA calcula ion by minimiza ion o di e ences be ween expe imen al and calcula ed BCFs11. Di e ences o he calcula ed subsi e binding ene gies we e below 0.4 kJ/mol in case o wo independen expe imen al measu emen s o ac ion pa e n o he same enzyme (da a no shown). To s anda dize da a e alua ion, ESUMA and ESybyl da a (see la e ) we e calcula ed o he same numbe o subsi es o wild- ype and mu an enzymes. The empe a u e and he in e pola ion s ep we e se o 37 °C and 0.01, espec i ely, in all calcula ion in con as o he de aul alues o he p og am. BCF p edic ion using calcula ed subsi e binding ene gies was pe o med by SUMA acco ding o he s anda d pa ame e s (subs a es o DP 3 o 11, empe a u e 37 °C). 2.2. Molecula modeling C ys al s uc u es o wild- ype enzymes we e downloaded om he P o ein Da a Bank13. S uc u es o human sali a y α-amylase (PDB code: 1SMD14), ba ley α-amylase AMY1 (PDB code: 1RP815) and AMY2 (PDB code: 1BG916), α-amylase o Bacillus amylolique aciens (PDB code: 3BH417), po cine panc ea ic α-amylase isoenzyme II (PPA II) (PDB code: 1PIG18) and α- amylase o Aspe gillus o yzae (PDB code: 7TAA19) we e used as empla es o building o 3D s uc u es o wild ype and mu an enzymes using Sybyl p og am package (T ipos Inc., S . Louis, MO, USA). S uc u es o he mal ooligosaccha ide subs a es we e modeled based on he c ys al s uc u es o mal ose (PDB code: 2GVY20), mal ohep aose (PDB code: 1RP815) and a subs a e-analogue inhibi o aca bose (PDB code: 1MFU21, 1RPK15, 1E3Z22, 1OSE23) complexed wi h a ious alpha-amylases. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 6 Ini ial s uc u e o an enzyme-subs a e complex was gene a ed by me ging he subs a e in o he binding cle o he enzyme and dele ing he inhibi o i i exis ed in he o iginal c ys al s uc u e. Wa e molecules bumped wi h subs a e we e also emo ed. In mu an p o eins, wild- ype esidues we e changed by Sybyl. To a oid bumping o mu an esidues wi h unmodi ied ones, wa e molecules o subs a e chain, he spa ial posi ion o he side chain o he modi ied esidue was p edic ed by a sho molecula dynamics un (leng h: 1000 s ep, s ep size: 1 s, empe a u e: 300 K, dielec ic cons an : 1, AMBER7_FF99 o ce ield). To e ine he posi ion o he subs a e and he mu a ed esidues, a sho minimiza ion p ocedu e was pe o med by Sybyl (100 Powell i e a ions, dielec ic cons an 1, AMBER7_FF99 o ce ield) on he 8 Å su ound o he oligosaccha ide chain and he mu a ed esidues. The esul ed complex was u he ene gy- minimized wi hou any ixed a oms by Sybyl using he ollowing pa ame e s: AMBER7_FF99 o ce ield, non-bonded cu o 8 Å, 100 Powell i e a ions and dielec ic cons an 1. In e ac ion ene gy (ESybyl) be ween he enzyme and he ca bohyd a e subs a e was calcula ed o each subsi e. Calcula ions and isualiza ion we e pe o med on Silicon G aphics Fuel wo ks a ions (Silicon G aphics In e na ional, F emon , CA, USA). 3. Resul s and discussion 3.1. Da a se s The aining se consis ed o he subsi e binding ene gies o all wild- ype enzymes and some mu an s o human sali a y amylase (HSA) and ba ley amylase 1 (AMY1) enzymes. Wild- ype enzymes we e chosen based on he exis ence o expe imen al BCF alues as well as a ailable c ys al s uc u e. In con as o he a ious deg ee o sequence iden i y (11-86 %) be ween he wild- ype enzymes (Table 1), hey we e all classi ied in o he glycoside hyd olase amily 13 and showed high deg ee o s uc u al conse a ion. Subsi e binding ene gy alues o HSA enzymes we e ecalcula ed by SUMA om he p e iously published ac ion pa e ns o wild ype HSA and i s W58L24 and Y151M25 mu an s. Recalcula ion o published subsi e maps was necessa y, because he published ene gies we e calcula ed o 8 subsi es (5 glycone and 3 aglycone subsi es); howe e , eliable s uc u al model o subs a e can be buil only o 7 subsi es (4 glycone and 3 aglycone subsi es). The published binding ene gy o subsi e −5 was small (0.65 kJ/mol)24, he e o e simpli ica ion o he subsi e model should cause only ma ginal changes in he binding ene gies o o he subsi es. Indeed, he o iginal and he ecalcula ed binding ene gies we e in excellen ag eemen , squa e o he co ela ion coe icien s ( 2) om he linea eg ession analysis we e 0.986. Simila ly, published subsi e maps o all o he wild- ype 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 7 enzymes we e ecalcula ed and he e we e excellen co ela ions in all cases (Table 2). The aining se consis ed o o al 13 wild- ype and mu an enzymes and 97 subsi e binding ene gy alues (Table 3). A es se was c ea ed om subsi e binding ene gies o no el mu an s o AMY1: S48Y, V47A, V47D, V47F, V47I/S48I, V47K/S48G, V47G/S48D and V47L/S48A. These single and double mu an enzymes con ain a ious mu a ions o V47 and/o S48 esidues which we e enginee ed p e iously wi h andom mu agenesis o he in es iga ion o he enzyme-subs a e in e ac ions26. Ac ion pa e ns o hese enzymes we e de e mined expe imen ally using hyd olysis o CNP-MOSs o DP 3−11 and subsi e maps we e calcula ed by SUMA (Mo i e al., in p epa a ion). The subsi e maps we e calcula ed o 7 glycone and 4 aglycone binding si es by he same way as he AMY1 g oup o aining se . The es se consis ed o o al 8 mu an enzymes and 73 subsi e binding ene gy alues (Table 3). 3.2. S eps o calcula ion Ou wo king hypo hesis was ha he subsi e binding ene gies could be p edic ed om enzyme-subs a e in e ac ion ene gies calcula ed by a molecula mechanical p og am. Homologous models o enzyme-subs a e complexes o he aining se we e buil and in e ac ion ene gies we e calcula ed be ween he p o ein and he ca bohyd a e esidues o each subsi e (ESybyl). These alues we e compa ed o hose calcula ed based on he expe imen ally de e mined ac ion pa e ns (ESUMA). To each highes co ela ion be ween ESUMA and ESybyl da a, pa ame e s o he ene gy-minimiza ion p ocedu e we e op imized. Se e al o ce ields a ailable in Sybyl p og am: Kollman_All_A om, Ambe 95 and Ambe 7_FF99 we e es ed, oge he wi h an in-house implemen a ion o Glycam06 o ce ield27. Ambe 7_FF99 was chosen because i s pe o mance was he bes in ou hand. Fu he mo e, wide- a ie y o alues o dielec ic cons an (1, 2, 3, 4 and 8), non-bonded cu o (8, 10 and 12), numbe o i e a ions (20-100 by s eps o 20 and 100-1000 by s eps o 100) and he wid h o minimized shell (1-12 angs om) o a oms a ound he subs a e and he mu a ed esidues in he ini ial ene gy minimiza ion we e also p obed o ind a good combina ion o pa ame e s. Ou aim was o build a ela i ely as p ocedu e, he e o e explici molecules o bulk sol en and ex ensi e minimiza ion we e a oided. Linea eg ession analysis was pe o med wi h he excep ion o ene gies o he −1 and +1 subsi es because i is heo e ically impossible o calcula e hem om expe imen al bond clea age equency da a11. A e eg ession analysis, ESybyl da a we e linea ly ans o med o he scale o ESUMA da a wi h he alues o in e cep and slope o he bes i line o ge ET ans . (Fig. 2A and 2B) o each enzyme g oup. Enzyme g oups we e based on he wild- ype enzymes and 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 8 each mu an belonged o he g oup o he co esponding wild- ype enzyme. Calcula ed ESybyl and ET ans . alues o aining enzymes (Table 3) showed good co ela ion wi h ESUMA alues: 2 we e in he ange o 0.827−0.929 (Table 4). The p ocedu e was applied o he es se o assess he p edic i e po en ial o he model. Calcula ed in e ac ion ene gies (ESybyl) o he es se (con aining only AMY1 mu an s) we e ans o med o ET ans . alues using equa ion o AMY1 g oup o he aining se . Calcula ed ESybyl and ET ans . alues o he es se (Table 3) showed co ela ion wi h ESUMA alues (Fig. 2C): 2 was 0.502 (Table 4) which was lowe han he co esponding alue o 0.827 o AMY1 g oup o he aining se , as expec ed. Howe e , he e o dis ibu ion was odd: all alues ou o 95% p edic ion in e al we e belonged o V47D, V47K/S48G and V47G/S48D mu an s. All hese mu an s and only hese mu an s con ained a subs i u ion om neu al o cha ged esidue. I all alues o hese mu an s we e omi ed om he plo , he 2 alue ose up o 0.638. Mu an s o he es se con ained single o double mu a ions in he posi ion o V47 and/o S48 esidues o AMY1. C ys allog aphic analysis showed ha Val47 was in ol ed in di ec hyd ogen bonds a he −7 subsi e, while Se 48 o med indi ec hyd ogen bonds a he −1 and −2 subsi es15. Docking s udies by molecula modeling p edic ed ha he in luence o V47 was ex ended o he −6, −5, −4 subsi es and S48 e ec ed subs a e binding a he −4, −3, −2 subsi es28. Fu he mo e, V47 oge he wi h Y105, may be conside ed as he subs a e “en ance” o he ac i e si e cle whe e he subs a e adop s a hal -ci cle con o ma ion cen e ed on Val4715. The e o e adjacen V47 and S48 amino acids a e su ounded by se e al glucose esidues o bounded subs a e chain and hey may e ec he dynamics o enzyme-subs a e in e ac ion in addi ion o in luence o binding a ini y. I seems ha ou simple p ocedu e was only mode a ely able o model his combined e ec as he co ela ion coe icien was d opped om 0.827 o 0.638 (neglec ing mu an s ha ing subs i u ions o cha ged esidues). La ge changes in cha ge dis ibu ion o he mu a ed posi ion u ned ou o be also p oblema ic, as in cases o V47D, V47K/S48G and V47G/S48D mu an s: he co ela ion coe icien was u he d opped o 0.502. In con as wi h he es se , enzymes o AMY1 g oup o aining se con ain he mu a ions o Y105 and/o T212, which amino acids loca ed a −6 and +4 subsi es and pa icipa ed in subs a e binding a only one o wo ou e mos subs a e binding si e28 and hese mu an s did no ha bo any subs i u ion o cha ged esidues. Some mu an s con aining amino acid changes in he seconda y binding si es we e also es ed, bu ou p ocedu e ailed o co ec ly p edic changes caused by hese dis an si es, as expec ed (da a no shown). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 9 3.3. P edic ion o ac ion pa e n Compu e p og am SUMA is capable no only o calcula e ESUMA ene gies bu also o ecalcula e BCFs om subsi e maps11 hus i can be used o p edic ion o he p ima y expe imen al alues. Bond clea age equencies we e calcula ed using he ET ans . alues as inpu da a in SUMA and he p edic ed BCFs we e co ela ed wi h he expe imen ally de e mined alues. Linea eg ession analysis esul ed in good co ela ion ( 2 = 0.727−0.835) be ween p edic ed and published BCFs o enzymes o he aining se (Table 4). In case o mu an AMY1 enzymes o he es se , he co ela ion coe icien was simila ly low o ha o binding ene gy analysis: 2 = 0.538 (Table 4). No unexpec edly, he odd e o dis ibu ion was also p esen he e: i alues o mu an enzymes ha ing subs i u ion o cha ged esidues (V47D, V47K/S47G and V47G/S48D) we e omi ed, he alue o 2 ose up o 0.728. 4. Conclusion The main s uc u al ea u e o amily 13 glycoside hyd olases is he p esence o he cha ac e is ic (α)8-ba el domain wi h a a ying numbe o ex a domains. The subs a e binding si e is made om he esidues o domains A and B, howe e , he a chi ec u e o he domain B a ies be ween amylases and i is he majo de e minan in di e ences o subs a e speci ici ies. P esen s udy was made wi h he aim o de elop a compu e -aided subsi e mapping p ocedu e o be e unde s and he speci ici y o α-amylases. Ou p ocedu e was success ully adop ed o α-amylases belonging o di e en kingdoms: BAA ep esen s lique ying bac e ial α- amylases om Bac e ia, u he α-amylases we e de i ed om Euka yo a: TAA ep esen s amylases om ungi, AMY1 and AMY2 ep esen plan enzymes and PPA and HSA ep esen he mammalian enzymes. The s uc u al conse a ion o hese enzymes we e much s onge han he sequence conse a ion (Table 4), which gi e he hope ha ou p ocedu e can be success ully applied o wide ange o α-amylases. The g ea clinical and indus ial impo ances jus i y he s udies o α-amylases. Desc ibed p ocedu e could help o gene a e subsi e maps o α-amylases as well as i could p edic he p ima y expe imen al da a (BCFs) ia he use o compu e p og am SUMA11. Howe e , special ca e is needed when neu al esidues was changed o cha ged one (and p obably ice e sa) in he subs a e binding si e o an α-amylase. The compu e -aided subsi e mapping may suppo he examina ion o he ole o subs a e binding esidues and i can be used o supply p o ein enginee s in design o modi ied enzymes wi h al e ed ac ion pa e n. Ou me hod can 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 10 supplemen he expe imen al subsi e mapping p ocedu e by p edic ion o subsi e maps and ac ion pa e ns in a as and cheap way. Acknowledgemen s E ika Fazekas is acknowledged o help in de e mina ion o BCFs. Re e ences 1. Gup a, R.; Gig as, P.; Mohapa a, H.; Goswami, V.K.; Chauhan, B. P ocess Biochem. 2003, 38, 1599-1616. 2. Rohlede , N.; Na e , U.M. Psychoneu oendoc inology 2009, 34, 469-485. 3. Can a el, B.L.; Cou inho, P.M.; Rancu el, C.; Be na d, T.; Lomba d, V.; Hen issa , B. Nucleic Acids Res. 2009, 37, D233-D238. 4. MacG ego , E.A.; Janecek, S.; S ensson, B. Biochim. Biophys. Ac a 2001, 1546, 1-20. 5. Janecek, S.; S ensson, B.; Hen issa , B. J. Mol. E ol. 1997, 45, 322-331. 6. Jespe sen, H.M.; Macg ego , E.A.; Sie ks, M.R.; S ensson, B. Biochem. J. 1991, 280, 51-55. 7. Janecek, S.; S ensson, B.; MacG ego , E.A. Eu . J. Biochem. 2003, 270, 635-645. 8. Kuma , V. Ca bohyd . Res. 2010, 345, 893-898. 9. Da ies, G.J.; Wilson, K.S.; Hen issa , B. Biochem. J. 1997, 321, 557-559. 10. Kand a, L.; Gyémán , G.; Lip ák, A. Biologia 2002, 57, 171-180. 11. Gyémán , G.; Ho ánszki, G.; Kand a, L. Eu . J. Biochem. 2002, 269, 5157-5162. 12. Kand a, L.; Abou Hachem, M.; Gyémán , G.; K amho , B.; S ensson, B. FEBS Le . 2006, 580, 5049-5053. 13. Be man, H.M.; Wes b ook, J.; Feng, Z.; Gilliland, G.; Bha , T.N.; Weissig, H.; Shindyalo , I.N.; Bou ne, P.E. Nucleic Acids Res. 2000, 28, 235-242. 14. Ramasubbu, N.; Palo h, V.; Luo, Y.G.; B aye , G.D.; Le ine, M.J. Ac a C ys allog . D Biol. C ys allog . 1996, 52, 435-446. 15. Robe , X.; Hase , R.; Mo i, H.; S ensson, B.; Aghaja i, N. J. Biol. Chem. 2005, 280, 32968-32978. 16. Kadziola, A.; Sogaa d, M.; S ensson, B.; Hase , R. J. Mol. Biol. 1998, 278, 205-217. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 11 17. Alikhajeh, J.; Khajeh, K.; Ranjba , B.; Nade i-Manesh, H.; Lin, Y.H.; Liu, E.H.; Guan, H.H.; Hsieh, Y.C.; Chuankhayan, P.; Huang, Y.C.; Jeya aman, J.; Liu, M.Y.; Chen, C.J. Ac a C ys allog . Sec . F S uc . Biol. C ys . Commun. 2010, 66, 121-129. 18. Machius, M.; Ve esy, L.; Hube , R.; Wiegand, G. J. Mol. Biol. 1996, 260, 409-421. 19. B zozowski, A.M.; Da ies, G.J. Biochemis y 1997, 36, 10837-10845. 20. Vujicic-Zaga , A.; Dijks a, B.W. Ac a C ys allog . Sec . F S uc . Biol. C ys . Commun. 2006, 62, 716-721. 21. Ramasubbu, N.; Raguna h, C.; Mish a, P.J. J. Mol. Biol. 2003, 325, 1061-1076. 22. B zozowski, A.M.; Lawson, D.M.; Tu kenbu g, J.P.; Bisgaa d-F an zen, H.; S endsen, A.; Bo che , T.V.; Dau e , Z.; Wilson, K.S.; Da ies, G.J. Biochemis y 2000, 39, 9099- 9107. 23. Gilles, C.; As ie , J.P.; Ma chisMou en, G.; Cambillau, C.; Payan, F. Eu . J. Biochem. 1996, 238, 561-569. 24. Ramasubbu, N.; Raguna h, C.; Mish a, P.J.; Thomas, L.M.; Gyémán , G.; Kand a, L. Eu . J. Biochem. 2004, 271, 2517-2529. 25. Kand a, L.; Gyémán , G.; Remenyik, J.; Raguna h, C.; Ramasubbu, N. FEBS Le . 2003, 544, 194-198. 26. S ensson, B.; Jensen, M.T.; Mo i, H.; Bak-Jensen, K.S.; Bonsage , B.; Nielsen, P.K.; K amho , B.; P ae o ius-Ibba, M.; Noh , J.; Juge, N.; G e e, L.; Williamson, G.; D iguez, H. Biologia 2002, 57, 5-19. 27. Ki schne , K.N.; Yongye, A.B.; Tschampel, S.M.; Daniels, C.R.; Foley, B.L.; Woods, R.J.J. J. Compu . Chem. 2008, 29, 622-655. 28. Bak-Jensen, K.S.; And e, G.; Go schalk, T.E.; Paes, G.; T an, V.; S ensson, B. J. Biol. Chem. 2004, 279, 10093-10102. 29. Allen, J.D.; Thoma, J.A. Biochem. J. 1976, 159, 121-131. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 12 FIGURE LEGENDS Figu e 1. Schema ic ep esen a ion o s uc u e o AMY1 (bo om panel) oge he wi h he a chi ec u e o subs a e binding si e o s udied wild- ype enzymes ( op panel). G een hexagons ep esen monosaccha ide uni s o he subs a e, la ge a ow shows he si e o clea age, subsi es a e labeled by nega i e and posi i e numbe s o glycone and aglycone binding si es, espec i ely. Figu e 2. Co ela ion be ween he ESUMA alues o AMY1 g oup and he p edic ed ene gies (ESybyl and ET ans .). Black lines co espond o he bes i ed eg ession line and g een lines show he 95 % p edic ion in e al ange. A) AMY1 g oup o aining se be o e linea ans o ma ion (ESybyl = 0.905 x ESUMA – 17.731) B) AMY1 g oup o aining se a e linea ans o ma ion (ET ans = 1.000 x ESUMA – 0.000) C) AMY1 g oup o es se a e linea ans o ma ion. T212 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 13 Table 1. Sequence iden i ies (%) o he s udied enzymes based on s uc u al alignmen (UniP o KB codes o p o ein sequences used o sequence alignmen : HSA – P04745, PPA – P00690, AMY1 – P00693, AMY2 – P04063, TAA – P0C1B3, BAA – P00692). HSA PPA AMY1 AMY2 TAA BAA HSA 100 -- -- -- -- -- PPA 86 100 -- -- -- -- AMY1 12 11 100 -- -- -- AMY2 12 11 74 100 -- -- TAA 18 19 14 13 100 -- BAA 16 18 15 15 20 100 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 14 Table 2. O iginal and modi ied subsi e models o he s udied enzymes and he co ela ion be ween he subsi e binding ene gies o he o iginal publica ions and he ecalcula ed alues based on subsi e models o his wo k. Enzyme Subsi e model squa e o co ela ion coe icien ( 2) e e ence o iginal wo k his wo k HSA −5 ... +3 −4 ... +3 0.986 24, 25 PPA −4 ... +3 −4 ... +3 0.942 11 AMY1 −8 ... +4 −7 ... +4 0.994 12 AMY2 −8 ... +4 −7 ... +4 0.999 12 TAA −3 ... +5 −3 ... +5 -- * 29 BAA −6 ... +4 −7 ... +3 0.996 29 * Ene gy alues o subsi es we e no published in he o iginal wo k. Figu e2C Click he e o download high esolu ion image