Asi eDesign: a Semi a ional Algo i hm o an Au oma ed Enzyme
Design
Se gi Roda,
⊥
Hen ik Te holsen,
⊥
Jule Ru h Heike Meye , Albe Canellas-Solé, Vic o Gualla ,*
Uwe Bo nscheue ,*and Masoud Kazemi*
Ci e This: J. Phys. Chem. B 2023, 127, 2661−2670
Read Online
ACCESS Me ics & Mo e A icle Recommenda ions *
sı Suppo ing In o ma ion
ABSTRACT: Wi h ad ances in p o ein s uc u e p edic ions, he
numbe o a ailable high-quali y s uc u es has inc eased
d ama ically. In ligh o hese ad ances, s uc u e-based enzyme
enginee ing is expec ed o become inc easingly impo an o
op imizing bioca alys s o indus ial p ocesses. He e, we p esen
Asi eDesign, a Mon e Ca lo-based p o ocol o s uc u e-based
enginee ing o ac i e si es. Asi eDesign p o ides a amewo k o
in oducing new ca aly ic esidues in a gi en binding pocke o
ei he c ea e a new ca aly ic ac i i y o al e he exis ing one.
Asi eDesign is implemen ed using pyRose a and inco po a es
enhanced sampling echniques o e icien ly explo e he sea ch
space. The p o ocol was es ed by designing an al e na i e ca aly ic
iad in he ac i e si e o Pseudomonas luo escens es e ase (PFE). The designed a ian was expe imen ally e i ied o be ac i e,
demons a ing ha Asi eDesign can ind al e na i e ca aly ic iads. Addi ionally, he Asi eDesign p o ocol was employed o enhance
he hyd olysis o a bulky chi al subs a e (1-phenyl-2-pen yl ace a e) by PFE. The expe imen al e i ica ion o he designed a ian s
demons a ed ha F158L/F198A and F125A/F158L mu a ions inc eased he hyd olysis o 1-phenyl-2-pen yl ace a e om 8.9 o
66.7 and 23.4%, espec i ely, and e e sed he enan ioselec i i y o he enzyme om (R) o (S)-enan iop e e ence, wi h 32 and 55%
enan iome ic excess (ee), espec i ely.
■INTRODUCTION
S uc u e-based enzyme enginee ing is widely used in he
de elopmen o bioca alys s o indus ial pu poses.
1−5
These
app oaches ha e been employed o enginee p o ein he mo-
s abili y, enzyme ac i i y, o subs a e selec i i y. FRESCO,
6
FoldX,
6,7
and Rose a
8
a e among he no ewo hy me hods o
enhancing enzyme he mos abili y, bu many mo e also ha e
been de eloped.
9−15
Simila ly, a wide ange o compu a ional
me hods a e a ailable o enginee ing enzyme ac i i y o o
changing he selec i i y, o ins ance, Rose a,
8
nAPOLI,
16
EnzymeMine ,
17
Ho Spo Wiza d,
18
Ca e ,
19
Fi eP o -ASR,
20
LoopG a e ,
21
and DaReUS-Loop.
22
Howe e , Rose a and i s
de i ed me hods
23,24
ha e become one o he mos widely used
ools in his a ea.
Cu en ad ances in deep lea ning s uc u al p edic ion
me hods inc eased he numbe o a ailable p o ein s uc u es
d ama ically.
25,26
Addi ionally, de no o p o ein design echni-
ques a e becoming inc easingly accu a e,
24,27,28
which makes i
possible o c ea e ailo -made p o ein sca olds. Combining
s a is ical ene gy de i ed om homologous sequences wi h
physics-based simula ions is also shown o be a powe ul
app oach o unco e ing he o igin o enzyme s abili y and
ca alysis.
29
Conside ing hese de elopmen s, s uc u e-based
enzyme enginee ing is expec ed o play an e en mo e c i ical
ole in op imizing enzymes o indus ial applica ions.
S uc u e-based enzyme enginee ing could be used o ei he
op imize enzyme ac i i y and selec i i y o a gi en subs a e o
o in oduce new unc ionali y in o a p o ein ca i y.
30−36
The
la e app oach has he po en ial o c ea e enzymes ha a e
capable o ca alyzing new chemical eac ions.
37
Fo example,
by mu a ing he ca aly ic glu ama e, a glycosidase enzyme was
con e ed o a glycosyn hase.
34,35
Al e na i ely, new ca aly ic
esidues can be in oduced in a p o ein ca i y. This app oach,
o ins ance, was used o c ea e a second ac i e si e wi h
hyd olysis ac i i y in a ansaminase. The enginee ed mul i-
unc ional enzyme combines ansaminase and hyd olase
ac i i ies in a single p o ein sca old, allowing he con e sion
o β-ke o es e s in o β-amino acids, which can be used o he
syn hesis o a key p ecu so o a amily o an idiabe ic d ugs.
33
These applica ions equi e ( e)designing ca aly ic esidues in
a gi en binding pocke . One way o app oach such asks is by
Recei ed: Oc obe 8, 2022
Re ised: Feb ua y 15, 2023
Published: Ma ch 21, 2023
A iclepubs.acs.o g/JPCB
© 2023 The Au ho s. Published by
Ame ican Chemical Socie y 2661
h ps://doi.o g/10.1021/acs.jpcb.2c07091
J. Phys. Chem. B 2023, 127, 2661−2670
Downloaded ia 5.67.84.229 on No embe 23, 2025 a 13:09:41 (UTC).
See h ps://pubs.acs.o g/sha ingguidelines o op ions on how o legi ima ely sha e published a icles.
g a ing an ac i e si e in o a p o ein sca old, which has been
pe o med, e.g., by Rose a Ma ch.
38,39
This me hod is based
on iden i ying a sui able ca i y by sea ching many p o ein
sca olds. He e we p esen an al e na i e me hod, Asi eDesign,
o his ask. Asi eDesign is capable o iden i ying he bes
posi ions o a se o p ede ined ca aly ic esidues in a gi en
ac i e si e wi hou he need o sea ching di e en p o ein
sca olds. The me hod is based on a Mon e Ca lo (MC)
simula ion and has been implemen ed using pyRose a.
Addi ionally, he p o ocol employs enhanced sampling
echniques o imp o e he simula ion con e gence. I also
includes he sampling o o a able subs a e bonds, which can
po en ially imp o e he iden i ica ion o design solu ions ha
may no be ound o he wise. Fu he mo e, he subs a e
sampling can easily be es ic ed o p ede ined in e als,
allowing o pe o m pa ial op imiza ions o la ge molecules.
He e, we demons a e he applica ion o he me hod by
designing an al e na i e ca aly ic iad in he ac i e si e o he
Pseudomonas luo escens es e ase (PFE).
40
Since enan ioselec-
i e hyd olases enable he syn hesis o chi al building blocks
o d ugs by he (dynamic) kine ic esolu ion o acemic
es e s,
5
he p o ocol was u he employed o al e he enzyme
selec i i y o a bulky chi al subs a e.
■COMPUTATIONAL METHODS
Asi eDesign employs MC sampling o explo e bo h he ac i e
si e and he ligand deg ees o eedom. As men ioned abo e,
he p o ocol was employed o enginee a new ca aly ic iad in
PFE and also o edesign he enzyme binding pocke . The
de ails o hese design app oaches a e p esen ed in he
ollowing.
Design o Ca aly ic Residues. Fo a gi en enzyma ic
chemical eac ion, he amino acid iden i y o he ca aly ic
esidues is known a p io i. Conside ing PFE as an example, he
ca aly ic iad consis s o a nucleophile esidue (Se ) ha
a acks he subs a e, a gene al base (His) ha accep s he
p o on, and an acidic esidue (Asp/Glu) ha ac i a es he
base. The esidues o he ca aly ic iad a e common o all
se ine hyd olases. As such, he main ask he e is o iden i y a
se o posi ions in he binding pocke ha can accommoda e
he ca aly ic esidues a he co ec dis ances ela i e o each
o he .
The ca aly ic design p o ocol is composed o h ee s ages
(Figu e 1A): (1) ini ial assignmen , (2) Mon e Ca lo sampling,
and (3) e-spawning o he nex epoch by adap i e ein o ce-
men lea ning p o ocol. In he ini ia ion s age, he ca aly ic
esidues a e assigned o he use -de ined posi ions andomly.
The ini ia ion s age is pe o med o each explo e
independen ly.
The Mon e Ca lo sampling s age in ol es he sampling o
bo h he posi ions o ca aly ic esidues and he ligand
con o ma ion. The ac i e si e and ligand sampling a e
pe o med in se ies. Du ing he ac i e si e sampling, in each
i e a ion, one ca aly ic esidue is assigned o a new andom
posi ion by mu a ing i , and he p e iously assigned posi ion is
mu a ed back o he wild- ype (WT) amino acids (Figu e 1B).
The mu a ions a e pe o med by he as - elax algo i hm o he
ose a lib a y.
8
I should be highligh ed ha he p e iously
assigned posi ion can be mu a ed o any se o use -de ined
amino acids. In his wo k, he wild- ype amino acids a e used
o eco e y o minimize dis u bing he ac i e si e. Du ing his
s age, he co ec dis ances o he ca aly ic esidues a e
en o ced by imposing dis ance es ain s.
The ligand sampling is pe o med using a simila app oach
as PELE
41,42
by in oducing andom pe u ba ions o he
ligand o a ion and ansla ion deg ees o eedom, as well as
dihed al angles. The main di e ence in he Asi eDesign
p o ocol is ha he sampling o he ligand- o a able bonds is
pe o med by a pa ial minimiza ion in a andom angle in e al,
a he han by assigning a andom alue o he dihed al angle.
Tha is, he ligand- o a able bonds a e chosen andomly and an
in e al in ei he a posi i e o nega i e di ec ion is assigned o
each selec ed dihed al angle a andom (Figu e 1C). The
numbe o o a able bonds ha a e selec ed and he magni ude
Figu e 1. Schema ic ep esen a ion o he main s eps o he MC algo i hm in he Asi eDesign (A). Sampling o he ca aly ic esidues (B). Sampling
o ligand- o a able bonds using an i e a i e g id sea ch (C).
The Jou nal o Physical Chemis y B pubs.acs.o g/JPCB A icle
h ps://doi.o g/10.1021/acs.jpcb.2c07091
J. Phys. Chem. B 2023, 127, 2661−2670
2662
o he angle in e als a e de ined by he use . The o a able
bonds ha a e no selec ed e ain hei cu en in e al.
Addi ionally, o a gi en dihed al angle, he selec ion o he
new in e als is es ic ed o he immedia e neighbo hood o i s
cu en in e al o a oid in oducing la ge con o ma ional
changes o he ligand. Fo example, a use -de ined new in e al
in he posi i e di ec ion is assigned o X1 in Figu e 1C,
whe eas X2 main ains i s cu en in e al.
The ligand dihed al angles a e hen minimized in he
assigned angle in e als (Figu e 1C). The minimiza ion is
pe o med using a wo-s ep p ocess. In he i s s ep, he
ligand’s lowes -ene gy con o ma ion is iden i ied by sampling
i s o a able bonds sys ema ically in a g id cons uc ed using
he assigned in e als and e alua ing he ligand ene gy. This
s ep is epea ed h ee imes whe e, in each i e a ion, he
esolu ion o he g id sea ch is inc eased and he sea ch in e al
is educed. The inal ligand con o ma ion is hen minimized
wi h a g adien descen op imiza ion in ca esian coo dina es.
This app oach hus co esponds o a pa ial minimiza ion in a
andomly chosen sampling in e al. Compa ed o he con en-
ional me hod o choosing a andom angle, his app oach can
iden i y he mos sui able con o ma ion o a gi en in e al.
The algo i hm is ully cus omizable and indi idual o a able
bonds o he ligand can be excluded by he use o speed up
he Asi eDesign simula ion.
Adequa e sampling could be challenging as he numbe o
design elemen s inc eases. The adap i e ein o cemen lea ning
p o ocol
43
was inco po a ed in o he simula ion o o e come
his issue. In his scheme, he simula ion is pe o med in
epochs in which he MC sampling is pe o med by mul iple
explo e s in pa allel using a dis ibu ed memo y pa alleliza ion
scheme. A he end o each epoch, he esul s om explo e s
a e collec ed and anked based on a gi en objec i e unc ion
( o al ene gy, ligand ene gy, es ain ene gy, e c.), and he
nex epoch is spawned by he op- anking esul s (Figu e 1A).
To u he imp o e he sampling, a simula ed annealing
scheme was also used du ing he simula ion. While simula ed
annealing helps he ajec o ies o escape local minima, he
explo e s o en con e ge o di e en solu ions (mu a ions)
depending on he con o ma ion o he ligand. The adap i e
ein o cemen lea ning algo i hm he e is used o iden i y he
solu ions ha a e mo e p omising and he eby o alloca e
compu a ional esou ces o he pa o he sea ch space ha is
o in e es .
Binding Pocke Redesign. The edesign o he binding
pocke ollows a simila app oach o he design o ca aly ic
esidues. The main di e ence he e is ha he mu a ion o he
nonca aly ic esidues is selec ed by he as - elax algo i hm o
he ose a lib a y using a se o use -de ined amino acids. This
co esponds o a minimiza ion s ep, in which he amino acid
wi h he mos a o able po en ial ene gy is selec ed o a gi en
posi ion. The numbe o posi ions included in he as - elax
minimiza ion is de ined by he use , and du ing his s ep, he
posi ions wi h he mos un a o able ose a ene gies a e
subjec ed o minimiza ion. The e o e, in his app oach, he
nonca aly ic esidues a e minimized o imp o e he in e ac ion
be ween he ligand and ac i e si e esidues. The ligand
sampling is pe o med as desc ibed abo e. As such, he
nonca aly ic esidues o he ac i e si e a e e ol ed in esponse
o he ligand con o ma ion.
To ensu e ha he so wa e can un consis en ly ac oss
di e en en i onmen s and a chi ec u es, we ha e de eloped a
con aine o Asi eDesign, allowing an easy ins alla ion and
dis ibu ion o he so wa e eliably and secu ely. The
con aine ins alla ion can be done ia a sc ip based on an
al eady exis ing con aine wi h all he packages and depend-
encies and ins alling PyRose a wi h he use c eden ials. A
license o PyRose a is needed.
Molecula Dynamics (MD). Fou eplicas o 100 ns o
molecula dynamics (MD) simula ions wi h OPENMM
44
we e
pe o med o analyze he s abili y o he newly designed
ca aly ic iads. A wa e cubic box (dis ance o 8 Å be ween he
closes p o ein a om and he edge o he box) was c ea ed
a ound he sys em using he TIP3P wa e model, and he
cha ge o he sys em was s abilized using mono alen ions
(Na+and Cl−). The p o ein sys em was pa ame e ized wi h he
AMBER99SB o ce ield. Ande sen he mos a and MC
ba os a we e applied o he NPT ensemble (cons an
p essu e and empe a u e, being 1 ba and 300 K,
espec i ely). An NVT equilib a ion phase las ed 400 ps
using a cons ain o 10 kcal/(mol·Å2) o he whole solu e
sys em, ollowed by a 1 ns NPT equilib a ion wi h a milde
cons ain o 5 kcal/(mol·Å2); he p oduc ion un only
included cons ain s be ween H and hea y a oms. The Ve le
in eg a o wi h a 2 s ime s ep was used wi h an 8 Å
nonbonded long- ange in e ac ions cu o .
P o ein Ene gy Landscape Explo a ion Simula ions.
P o ein Ene gy Landscape Explo a ion (PELE) was used o
analyze he subs a e binding o he e ol ed a ian s using
Asi eDesign. PELE is a heu is ic MC-based algo i hm coupled
wi h p o ein s uc u e p edic ion me hods.
41,42
The so wa e
begins by sampling he di e en mic os a es o he ligand
h ough small o a ions and ansla ions. Applying no mal
modes h ough he aniso opic ne wo k model (ANM)
app oach,
45
he p o ein’s lexibili y is also conside ed. Once
he whole sys em has been pe u bed, side chains o he
esidues close o he ligand a e sampled o a oid s e ic clashes.
Las , a unca ed New on minimiza ion wi h he OPLS2005
o ce ield is pe o med,
46
and he new mic os a e is accep ed
o ejec ed based on he Me opolis c i e ion. The Va iable
Dielec ic Gene alized Bo n Non-Pola (VDGBNP) implici
sol en model
47
was used o mimic he e ec o wa e
molecules a ound he p o ein.
The explo a ion o he subs a e was enhanced wi h
Adap i e-PELE
43
o imp o e he explo a ion o he sea ch
space.
48
■EXPERIMENTAL DESCRIPTION
Ma e ial. The chemicals ac-1-phenyl-2-pen anol (≥99%)
and ac-1-phenyle hyl ace a e (≥98%) we e o de ed om
Sigma-Ald ich. All o he chemicals and sol en s we e
pu chased om Sigma-Ald ich, VWR, o Ca l Ro h and we e
used wi hou u he ea men .
Syn hesis o 1-Phenyl-2-pen yl Ace a e. Fi e hund ed
mic oli e s o ace ic anhyd ide and 100 μL ac-1-phenyl-2-
pen anol we e added in a 1.5 mL ube. The eac ion was
s a ed by adding 10 μL py idine o he mix u e. The eac ion
was shaken a 25 °C and 500 pm un il comple e con e sion
was achie ed. Samples o 2 μL we e wi hd awn and dilu ed in
198 μL e hyl ace a e o gas ch oma og aphic (GC) analysis.
The eac ion was quenched by adding he mix u e o a 15 mL
ube con aining 2 mL ddH2O. The p oduc ac-1-phenyl-2-
pen yl ace a e (1) o med a second phase and was sepa a ed
and d ied o e anhyd ous sodium sul a e. The oily ac-1-
phenyl-2-pen yl ace a e was ob ained in 51% yield.
The Jou nal o Physical Chemis y B pubs.acs.o g/JPCB A icle
h ps://doi.o g/10.1021/acs.jpcb.2c07091
J. Phys. Chem. B 2023, 127, 2661−2670
2663
Plasmid Cons uc ion and Si e-Di ec ed Mu agene-
sis. Syn he ic genes o he PFE a ian s 1, 4, 8, 11, 12 in
pET28a we e o de ed om BioCa (Heidelbe g, Ge many)
using seamless cloning wi h he lanking egions 5′
aaggaga a acc 3′(5′ lanking) and 5′CACCACCACCAC-
CACCACTGAGATCCGG 3′(3′ lanking). The mu an s a e
based on he sequence o he PFE wild- ype (GeneBank:
WP_120448209.1). The sequences we e ex ended by a C-
e minal linke (GS) and a His6- ag, as is he case o he
sequence used o he 1VA4 c ys al s uc u e.
49
PFE a ian s 2 and 3 we e cons uc ed based on PFE_1,
PFE_5-7 we e based on PFE_4, and PFE_9-10 we e based on
PFE_8 using he Q5 Si e-Di ec ed Mu agenesis Ki (New
England Biolabs GmbH, Ipswich, U.K.). Nono e lapping
DNA-oligonucleo ides we e designed using he online
NEBaseChange ool o he mu a ions. The lis o p ime s
used o mu agenesis is gi en in Table S5. The annealing
empe a u es sugges ed by NEBaseChange online ool
(h ps://nebasechange .neb.com/) we e used o he polyme -
ase chain eac ion (PCR), which was pe o med acco ding o
he manu ac u e ’s p o ocol. The ob ained cons uc s we e
ampli ied in Esche ichia coli Top 10 and used o hea -shock
ans o ma ion o E. coli BL21 (DE3).
P o ein P epa a ion. P ecul u es (4 mL Lu ia−Be ani
(LB) con aining kanamycin) o E. coli BL21 (DE3) colonies
ha bo ing he cons uc s o he exp ession o he PFE a ian s
we e incuba ed o e nigh (37 °C, 180 pm). LB medium
(con aining 50 mg/L kanamycin) was inocula ed wi h 1% ( /
) o he p ecul u e and incuba ed (37 °C, 180 pm) un il i
eached an OD600 o 0.6. P o ein exp ession was induced by
he addi ion o isop opyl-β-D- hiogalac opy anoside (IPTG) o
a inal concen a ion o 0.5 mM ollowed by incuba ion o
∼20 h a 20 °C a 180 pm. Cells we e ha es ed by
cen i uga ion a 10,000gand 4 °C o 3 min, and he cell
pelle s we e esuspended wi h 4 mL equilib a ion bu e
(50 mM po assium phospha e, 300 mM sodium chlo ide,
10 mM imidazole, pH 8.0). Cells we e dis up ed by sonica ion
on ice ( i e cycles o 1 min sonica ion a 30% in ensi y, and
50% pulsed cycle) using a SONOPULS HD 2070
(BANDELIN Elec onic GmbH & Co. KG, Be lin, Ge many),
and he lysa es we e cla i ied by cen i uga ion a 10,000gand 4
°C o 30 min. Fo pu i ica ion, he c ude lysa es we e applied
o 1.5 mL Ro i Ga ose-His/Ni Beads (Ca l Ro h, Ka ls uhe,
Ge many). The esins we e washed wi h 15 mL washing bu e
(50 mM sodium phospha e, 300 mM sodium chlo ide, 20 mM
imidazole, pH 8.0) be o e a ge p o eins we e elu ed wi h
elu ion bu e (50 mM sodium phospha e, 300 mM sodium
chlo ide, 250 mM imidazole, pH 8.0). P o ein-con aining
ac ions we e pooled and e-bu e ed in 50 mM KPi pH 7.5
using PD10 columns (GE Heal hca e, Buckinghamshi e, U.K.).
PFE wild- ype was exp essed as p e iously epo ed.
50
Ac i i y Assays. The ac i i y o he PFE a ian s owa d
he hyd olysis o pa a-ni ophenyl ace a e (pNPA) was
analyzed. Fo his pu pose, 20 μL o a 100 mM pNPA
solu ion in dime hyl sul oxide (DMSO) was added o a 96-well
pla e and 180 μL o a PFE solu ion o known concen a ion in
50 mM KPi pH 7.5 was added o s a he eac ion. The
abso bance was ollowed a 405 nm, and he ini ial slope was
calcula ed. The eac ion was ca ied ou a 25 °C.
Au ohyd olysis was de e mined by adding 180 μL o he
50 mM KPi pH 7.5 bu e and sub ac ing he alue om he
hyd olysis a e o he PFE mu an s. Speci ic ac i i y was
calcula ed using a s anda d cu e o pa a-ni ophenol (0−
1 mM). The hyd olysis o ac-1-phenyl-2-pen yl ace a e (1)
and ac-1-phenyle hyl ace a e (2) was analyzed o all PFE
mu an s. Fo his pu pose, 975 μL o a 90 μg/mL PFE solu ion
was added o a 1.5 mL ube. The eac ion was s a ed by
adding 25 μL o a 200 mM subs a e solu ion in ace oni ile
( inal concen a ion 5 mM) and was un o 24 h a 37 °C and
1000 pm. Time samples o 200 μL we e aken a e 1, 2, 4,
and 24 h and ex ac ed wi h 200 μL o e hyl ace a e wice. The
o ganic phases we e pooled, d ied o e anhyd ous sodium
sul a e, and analyzed by GC.
Gas Ch oma og aphy (GC) Analysis. Analysis was
pe o med by gas ch oma og aphy wi h a lame ioniza ion
de ec o (GC-2010, Shimadzu, Kyo o, Japan) equipped wi h a
Hyd odex β3P column (25.0 m ×0.25 mm, 0.25 μm ilm
hickness, Mache ey−Nagel, Du en, Ge many). Fo he
de ec ion o he syn hesis o hyd olysis o 1column
empe a u e was held a 95 °C o 30 min, inc eased o
110 °C wi h 5 °C/min, and held o 45 min. Re en ion imes:
(S)-1-phenyl-2-pen yl ace a e 50.5 min, (R)-1-phenyl-2-pen yl
ace a e 51.5 min, (S)-1-phenyl-2-pen anol 66.7 min, (R)-1-
phenyl-2-pen anol 69.2 min. Fo he de ec ion o he
hyd olysis o 2column empe a u e was held a 110 °C o
30 min. Re en ion imes: (S)-1-phenyl-e hyl ace a e: 4.3 min,
(R)-1-phenyl-e hyl ace a e: 5.9 min, (S)-1-phenyle hanol:
8.0 min, (R)-1-phenyle hanol: 7.1 min.
■RESULTS AND DISCUSSION
Ca aly ic Residue Redesign. The es e ase I om P.
luo escens (PFE) was chosen as he es sca old p o ein. PFE
has been ex ensi ely s udied by mu agenesis, and i s ac i i y
has been cha ac e ized o mul iple es e s.
50−52
I has been
shown p e iously ha he wild- ype (WT) enzyme and i s
s udied a ian s exhibi low ac i i y o subs a e 1(Scheme 1),
which makes i a good candida e o es ing Asi eDesign. This
Scheme 1. Kine ic Resolu ion o Subs a es 1 and 2 S udied Using PFE and I s Va ian s
The Jou nal o Physical Chemis y B pubs.acs.o g/JPCB A icle
h ps://doi.o g/10.1021/acs.jpcb.2c07091
J. Phys. Chem. B 2023, 127, 2661−2670
2664
enzyme hyd olyzes small alipha ic es e s, and i s ac i e si e
con ains he ypical Se -His-Asp ca aly ic iad (Figu e 2).
To es he pe o mance o Asi eDesign in iden i ying
op imum posi ions o placing he ca aly ic esidues, he amino
acids o ming he es e ase ca aly ic iad in he WT enzyme
(Se 94, His251, and Asp222) we e mu a ed o Ala. Using his
mu a ed s uc u e, an MC sea ch was pe o med, employing
e hyl ace a e as he p obing subs a e. In he simula ion, all
esidues o ming he i s shell o he ac i e si e (Table S1)
we e allowed o be mu a ed o one o he esidues o he
ca aly ic iad. As men ioned in he Me hods Sec ion, du ing
he simula ion, once a new posi ion is accep ed o a gi en
ca aly ic esidue, he p e ious posi ion o he ca aly ic esidue
is mu a ed back o he WT amino acid (Figu e 1B).
Encou agingly, he ca aly ic esidues o he WT enzyme we e
eco e ed as he bes solu ion (Table 1). This esul
demons a es ha he p o ocol can indeed iden i y he op imal
posi ions o he ca aly ic esidues.
In addi ion o he na i e ca aly ic iad, he second-bes
a ian con ains a ca aly ic iad a posi ions S28, H29 and
D191. I is in e es ing o highligh ha , in his a ian , he
ca aly ic esidues a e he mi o image o he WT enzyme
(Figu e 3). The e o e, his a ian is expec ed o exhibi
opposi e enan ioselec i i y ela i e o he WT enzyme.
In e es ingly, he simula ion esul ed in mul iple a ian s
(PFE_1, PFE_1*+ I155D, PFE_1*+ A183D) in which
bo h posi ions 28 and 29 a e assigned o he Se and His
esidues, espec i ely (Table 1), indica ing ha he p obabili y
o sampling hese mu a ions is high. The only di e ence
be ween hese a ian s is he loca ion o he acid esidue
(Table 1 and Figu e S1).
To es he s abili y o he designed a ian s, 100 ns MD
simula ions we e pe o med o he WT enzyme and PFE_1.
These simula ions indica ed ha he designed a ian is less
s able compa ed o he WT enzyme. The a e age dis ances o
he ca aly ic iad o PFE_1 (Se 28-His29: 5.30 ±0.38 Å,
His29-Asp191: 2.74 ±0.76 Å) we e ound o be highe han
ha o he WT enzyme (Se 94-His251: 2.86 ±0.27 Å, His251-
Asp222: 1.85 ±0.06 Å). Based on he isual inspec ion o he
MD ajec o ies, a ian s PFE_1 + C194T (PFE_2) and
PFE_1 + V195M (PFE_3) we e c ea ed o imp o e he
enzyme s abili y (MD simula ion esul s in he Suppo ing
In o ma ion (SI);Figu es S2−S4). These a ian s, howe e ,
exhibi ed simila MD me ics as PFE_1. Al hough PFE_1 and
i s de i a i es appea o be less s able han he WT enzyme, he
ca aly ic dis ances o hese a ian s we e in accep able anges,
and hus, hey we e chosen o he expe imen al cha ac e -
iza ion.
The compu a ionally designed a ian s, ecombinan ly
exp essed in E. coli and pu i ied, we e hen e i ied
expe imen ally o cha ac e ize he enzymes’ ac i i ies (Table
2). In hese a ian s, he na i e ca aly ic machine y was
disabled by mu a ing he nucleophilic Se 94 o Ala. Because
he main objec i e o he expe imen al cha ac e iza ion was o
es he ac i i y o he iden i ied al e na i e ca aly ic iad
wi hou he in e e ence o ex a mu a ions, His251 and
Asp222 we e no mu a ed. The expe imen al esul s showed
ha he designed a ian s exhibi hyd olysis ac i i y and
PFE_1 is indeed ac i e in he hyd olysis o pNPA and he
acemic compounds 1and 2(Table 2). I s ac i i y, howe e ,
was lowe han he WT enzyme. This could be due o he
des abiliza ion e ec o mu a ions, as i can be seen om he
dec eased mel ing empe a u e o he designed a ian s (Table
2). Al e na i ely, he lowe obse ed ac i i y could be because
o less op imum ca aly ic dis ances and less s abili y o he
ac i e si e as indica ed by he MD simula ions. These
obse a ions imply ha his a ian has a less o ganized
ca aly ic geome y. This phenomenon has been obse ed in
o he designed (o na u al) hyd olase ac i e si es, whe e
imp o ing hese dis ances ga e be e o e all ac i i ies.
31,53
Figu e 2. PFE and i s ca aly ic esidues. The ca aly ic iad esidues
a e colo ed in ed and labeled (PDB code: 1VA4).
49
Table 1. Top 10 Ca aly ic Designs Gi en a he End o he
Asi eDesign Simula ion wi h PFE’s s uc u e
a
o al ene gy mu a ions
−3170.1 A94S/A251H/A222D (WT)
−3160.1 W28S/L29H/T191D (PFE_1)
−3159.6 A94S/V225H/A222D
−3157.8 A94S/A251H/F162D
−3156.8 A94S/A251H/I224D
−3156.1 W28S/L29H/I155D (PFE_1*+ I155D)
−3150.8 A94S/V225H/F125D
−3150.1 W28S/L29H/A183D (PFE_1*+ A183D)
−3148.5 W28S/V195H/T191D
−3147.7 W28S/M95H/V121D
a
The epo ed ene gies a e ose a po en ial ene gies. PFE_1*s ands
o W28S/L29H/S94A.
Figu e 3. PFE WT and he newly designed ac i e si e. The ca aly ic
iad esidues o he WT a e colo ed in ed, while he ones om he
PFE_1 design a e shown in yellow. The labels a e based on 1VA4
s uc u e. The igu e displays ha he designed a ian is he mi o
image o he WT ac i e si e in he same p o ein ca i y.
The Jou nal o Physical Chemis y B pubs.acs.o g/JPCB A icle
h ps://doi.o g/10.1021/acs.jpcb.2c07091
J. Phys. Chem. B 2023, 127, 2661−2670
2665
This was p edic ed by he MD simula ions o he mu an s.
In e es ingly, he PFE_1 a ian exhibi ed an in e se selec i i y
o he bulky compound 1. Va ian s PFE_2 and PFE_3 we e
bo h inac i e owa d 1bu showed ac i i y owa d pNPA.
I is known ha imidazole used o p o ein pu i ica ion can
also hyd olyze he eac i e subs a e pNPA.
54
The hyd olysis
o subs a e 2was analyzed o ule ou he possibili y ha he
de ec ed ac i i y in he pNPA assay was caused by imidazole
impu i ies, which should no longe be p esen a e p o ein
pu i ica ion since bu e exchange was pe o med. Subs a e 2
is no imidazole hyd olyzable (da a no shown) and less
challenging PFE subs a e han bulky subs a e 1and he e o e
ideal o measu e e en low enzyma ic hyd olysis ac i i ies. Since
all pu i ied PFE a ian s showed ac i i y and selec i i y in he
hyd olysis o 2(Table 2), and au ohyd olysis was no
obse ed, he expe imen al da a clea ly con i m enzyma ic
hyd olysis.
These esul s sugges ha , o a gi en binding pocke , he
p o ocol is able o iden i y mul iple iable solu ions o
designing ca aly ic esidues, which can be used as a s a ing
poin o u he op imiza ion.
Binding Pocke Redesign. To es he pe o mance o
Asi eDesign o nonca aly ic esidues, 1was chosen as he
subs a e (Scheme 1). The WT enzyme exhibi s low ac i i y
and enan ioselec i i y o 1, which makes i a good candida e
o imp o emen . Addi ionally, he p e ious si e-di ec ed
mu agenesis o his enzyme did no yield any a ian s wi h
high ac i i y o 1.
50
The binding pocke design simula ions
we e pe o med by including 31 esidues o he ac i e si e
(Table S2; design domain, no ice ha ca aly ic esidues we e
excluded). In hese simula ions, no assump ions we e made o
he posi ions o mu a ions and all esidues ha a e p esen in
he i s shell o he ac i e si e, 11 esidues (Figu e 4;
highligh ed in yellow), we e allowed o mu a e while he es
we e only epacked. Since he enzyme was expec ed o
hyd olyze a hyd ophobic subs a e, he allowed mu a ions we e
limi ed o hyd ophobic amino acids (A, I, L, F, P, W, V, Y).
In addi ion, sequence es ain s we e imposed on all mu able
esidues (i.e., in oduc ion o any non-WT amino acid is
penalized) o a oid la ge di e gence om he WT enzyme,
he eby a o ing sequence conse a ion wi h an ene gy penal y.
The subs a e was placed in he ac i e si e manually, and MC
simula ions we e pe o med while imposing dis ance es ain
be ween he ca bonyl ca bon o he subs a e and Se 94.
Two sepa a e MC simula ions we e pe o med o he (R)-
and (S)-enan iome s o 1, hence e ol ing he ac i e si e o
each enan iome independen ly. Fo each enan iome , he 50
a ian s wi h he o e all bes ene gies (p o ein and subs a e
binding) we e selec ed. These s uc u es we e hen clus e ed
based on he binding mode o he subs a e, and om each
clus e , one a ian was chosen (Table S3). Encou agingly, he
simula ions a ge ed many o he ac i e si e posi ions ha we e
p e iously sugges ed o be impo an o enan ioselec i i y
(F125, F158, and I224)
50
in addi ion o some new posi ions
(W28, V121, and F198). Howe e , he p edic ed mu a ions o
hese posi ions may di e om he p e ious s udy.
To es he p edic ed a ian s, subs a e binding was
simula ed by PELE so wa e.
42
In hese simula ions, he
Table 2. Expe imen ally Measu ed Ac i i ies o he Ca aly ic Designs in he H yd olysis o Subs a es; pNPA, 1 and 2
c
,
d
PFE
a ian s mu a ions pNPA ac i i y
(U/mg) subs a e 1subs a e 2
Tm
(°C) p edic ed
selec i i y
WT 162.2 8.9% (8% ee (R), E 1)
a
48.7% (91% ee (R), E 59)
b
71.8
PFE_1 W28S/L29H/T191D/S94A 0.9 2.3%
(13% ee (S), E 1)
a
17.7% (80% ee (R), E 11)
a
44.9 (S)
PFE_2 W28S/L29H/T191D/S94A/C194T 0.2 no de ec able 2.1% (43% ee (R), E 3)
a
50.9 (S)
PFE_3 W28S/L29H/T191D/S94A/V195M 1.3 no de ec able 0.9% (11% ee (R), E 1)
a
44.8 (S)
a
A e 24 h.
b
A e 1 h.
c
pNPA ac i i y is epo ed acco ding o speci ic ac i i y, while o subs a es 1and 2 he con e sion is epo ed. The esidue
numbe ing co esponds o he 1VA4 s uc u e. The mel ing poin s (Tm) o he PFE a ian s we e de e mined by nanodi e en ial scanning
luo ime y (NanoDSF).
55
d
E alues we e calcula ed acco ding o Chen e al.
56
Figu e 4. Ac i e si e o PFE and he used design domain. The ca aly ic iad esidues a e colo ed in ed, he mu able esidues in yellow, and he
only epackable esidues in iole .
The Jou nal o Physical Chemis y B pubs.acs.o g/JPCB A icle
h ps://doi.o g/10.1021/acs.jpcb.2c07091
J. Phys. Chem. B 2023, 127, 2661−2670
2666
subs a e was placed ou side he ac i e si e, and he binding
was moni o ed by coun ing nea -a ack con o ma ions
(NAC)
57
and compu ing he a e age ene gy o he ligand in
he ac i e si e (Figu e 5). These simula ions o e a quali a i e
measu e o iden i y a ian s ha yield a p oduc i e binding
mode. The p ima y goal o his il e ing s ep was o limi he
numbe o a ian s o be es ed expe imen ally. A NAC is
de ined as a con o ma ion whe e he dis ance be ween he
ca bonyl C o he subs a e and he alcoholic O o he ca aly ic
Se esidue is wi hin 4 Å, while he H-bonds o he ca aly ic
iad a e wi hin easonable dis ances (≤3.5 Å).
32
Also, he
dis ance be ween he ca aly ic His esidue and he e he O o
he subs a e is less han 6.5 Å (as he p o ona ed His esidue
will gi e a p o on o his a om la e on in he eac ion o
elease he alcohol p oduc
58
). All he h esholds used o
iden i ying NACs a e based on p e ious s udies.
32,33,59
O e all,
he p edic ed a ian s exhibi ed a highe numbe o NACs
compa ed o he WT (Table S3). Mo eo e , he dis ibu ion o
he a e age in e ac ion ene gy (Figu e S5) is be e in many
a ian s compa ed o he WT, and key ca aly ic dis ances ha e
good alues as well (Figu es S6−S9).
Based on he in silico analysis, eigh a ian s wi h he highes
numbe o NACs ela i e o he WT enzyme we e selec ed o
expe imen al e i ica ion (Tables 3 and S4). The expe imen al
esul s show ha h ee p edic ed a ian s (PFE_5, PFE_8, and
PFE_10) exhibi ed signi ican imp o emen o e he WT
enzyme in he hyd olysis o 1and, in con as o he WT
enzyme, hey a e selec i e o he (S) subs a e (Table 3).
These a ian s con ain F158L, F125A, and F198A subs i u-
ions. The ac i e si e ca i y o PFE is su ounded by bulky
esidues ha hinde he access o la ge subs a es. Thus, he
mu a ion o hese oluminous esidues o smalle hyd ophobic
ones opens up he ac i e si e o accommoda e he bulky
subs a es (Figu e 6).
The esul s demons a e ha Asi eDesign no only iden i ied
he posi ions o esidues ha needed o be changed bu also
p edic ed bene icial mu a ions as well. The de eloped me hod
esul ed in double and iple mu an s wi h enhanced ac i i y
owa d subs a e 1. Hence, Asi eDesign can bo h guide he
enginee ing o ac i e si es o enzymes by iden i ying he ho
spo s and sugges a ian s wi h enhanced ac i i y owa d he
hyd olysis o a speci ic subs a e. Addi ionally, he sugges ed
a ian s can se e as a s a ing poin o u he op imiza ion,
e.g., by di ec ed e olu ion. Di ec ed e olu ion has p o en o be
a powe ul ool o op imizing newly in oduced ac i i ies in
p o ein sca olds.
60
Howe e , he simula ions we e no able o p edic he
a ian s’ enan ioselec i i y accu a ely. The main eason o his
could be ha he design/p edic ions we e pe o med based on
Figu e 5. Ini ial se up o PELE simula ions and he ep esen a ion o
a NAC. The subs a e is placed ou side he ac i e si e and allowed o
explo e a ound he d awn box ( op). The NAC is ep esen ed wi h
e e y key dis ance highligh ed in a di e en colo (blue o se ine−
his idine, beige o acid−his idine, iole o se ine−subs a e, and
g een o his idine−subs a e) (bo om). The ca aly ic iad esidues
a e colo ed in ed, and he subs a e in cyan.
Table 3. Expe imen al Measu ed Ac i i ies o he Binding Pocke Redesigns in he Hyd olysis o Subs a e 1
a
PFE a ian s mu a ions Tm(°C) subs a e 1p edic ed selec i i y
WT 71.8 8.9% (8% ee (R), E 1)
PFE_4 W28A/F158L/F198A 57.0 6.0% (3% ee (S), E 1)
PFE_5 F158L/F198A 69.5 66.7% (32% ee (S), E 4) (S)
PFE_6 W28A/F125A/F158L/F198A 58.2 3.3% (9% ee (S)) (S)
PFE_7 W28A/F158L/F198A/I224L 57.3 5.2% (8% ee (R), E 1) (S)
PFE_8 F125A/F158L 63.6 23.4% (55% ee (S), E 4) (R)
PFE_9 F125A/F158L/I224L 62.0 1.4% (29% ee (R)) (R)
PFE_10 F125A/F158L/F198A 63.6 16.3% (60% ee (S), E 4) (R)
PFE_11 V121A/F125A/I224L 59.7 2.9% (38% ee (S)) (R)
PFE_12 V121A/F158A/F198V 60.4 1.6% (100% ee (S)) (R)
a
The ac i i y is epo ed as con e sion a e 24 h. The esidue numbe ing co esponds o he 1VA4 s uc u e. The “p edic ed selec i i y” column is
based on whe he he mu an was ob ained om he Asi eDesign simula ion wi h (R)- o (S)-enan iome . The mel ing poin s (Tm) o he PFE
a ian s we e de e mined by nanodi e en ial scanning luo ime y (NanoDSF).
55
The Jou nal o Physical Chemis y B pubs.acs.o g/JPCB A icle
h ps://doi.o g/10.1021/acs.jpcb.2c07091
J. Phys. Chem. B 2023, 127, 2661−2670
2667
ligand binding ene gies, which do no necessa ily co ela e wi h
enan ioselec i i y. This can po en ially be imp o ed by
inco po a ing a ansi ion s a e analog as he p obing subs a e,
which is he main deciding ac o o enan ioselec i i y.
Ano he op ion could be o measu e he ene gy ba ie o
each enan iome using ela i ely inexpensi e quan um
mechanics/molecula mechanics (QM/MM) me hods such
as he empi ical alence bond me hod.
61
Ne e heless, an
accu a e p edic ion o he enan ioselec i i y is no o iously
challenging as he ene gy di e ence be ween he ac i a ion
ene gies o enan iome s is o en e y small.
The binding pocke -designed a ian s also exhibi ed lowe
mel ing empe a u es (Table 3). The main eason o his is
ha he simula ions a e d i en by imp o ing he ligand binding
ene gies, a he expense o p o ein s abili y. This issue can be
alle ia ed by downs eam enzyme s abili y op imiza ion o he
designed a ian s ei he compu a ionally o expe imen ally.
6−10
■CONCLUSIONS
This wo k p esen s he Asi eDesign p o ocol, which aims a
enginee ing ac i e si es o enzymes o ei he in oduce new
ca aly ic esidues o modi y an exis ing ac i e si e in silico. The
p o ocol is implemen ed using he pyRose a lib a y and
combines MC sampling o he ac i e si e esidues wi h
enhanced sampling echniques o iden i y he mos sui able
posi ions o ca aly ic esidues o a gi en ac i e si e. The ligand
sampling is also included in he simula ion, which is necessa y
o de e mine he op imal solu ions.
To demons a e he pe o mance o he p o ocol, a new
ca aly ic iad was designed in he ac i e si e o he es e ase I
om P. luo escens (PFE). The expe imen al cha ac e iza ion
demons a ed ha he designed a ian s a e no only ac i e
bioca alys s, bu hey also exhibi ed in e se enan ioselec i i y
o he bulky chi al subs a e 1. Thus, he binding pocke o he
enzyme was also success ully enginee ed o imp o e he
ac i i y o 1.
O e all, hese examples demons a e ha he Asi eDesign
p o ocol can iden i y mul iple iable solu ions o designing
ac i e si e esidues o a gi en ac i e si e. This app oach, hus,
can be used in he enginee ing o mul i unc ional ca alys s o in
designing new ca aly ic esidues in a gi en pu a i e binding
pocke .
■ASSOCIATED CONTENT
Da a A ailabili y S a emen
The code used in he s udy is a ailable a h ps://gi hub.com/
masoudk/Asi eDesign. The code is also a ailable as a con aine
a h ps://gi hub.com/BSC-CNS-EAPM/Asi eDesign-
con aine .
*
sı Suppo ing In o ma ion
The Suppo ing In o ma ion is a ailable ee o cha ge a
h ps://pubs.acs.o g/doi/10.1021/acs.jpcb.2c07091.
Recompila ion o all he esidues in he design domain
o he ca aly ic esidues edesign’s expe imen ; ecompi-
la ion o all he esidues in he design domain o he
binding edesign pocke ’s expe imen ; PELE simula ion
esul s o selec ed mu an s wi h 1; expe imen al
measu ed ac i i ies o he binding pocke edesigns
wi h he o he es ed subs a es; p ime s used o he
cons uc ion o he designed a ian s; he po en ial
posi ions o he design o an al e na i e ca aly ic iad in
he PFE es e ase; sho desc ip ion o he esul s om
he MD simula ions ollowed by he iolin plo s o he
dis ibu ion o di e en me ics o in e es ; iolin plo s
o he dis ibu ion o di e en me ics o in e es
ob ained in he PELE simula ions (PDF)
■AUTHOR INFORMATION
Co esponding Au ho s
Vic o Gualla −Ba celona Supe compu ing Cen e (BSC),
Ba celona 08034, Spain; Ins i ucióCa alana de Rece ca i
Es udis A anca s (ICREA), Ba celona 08010, Spain;
o cid.o g/0000-0002-4580-1114; Email: ic o .gualla @
bsc.es
Uwe Bo nscheue −Depa men o Bio echnology &Enzyme
Ca alysis, Ins i u e o Biochemis y, Uni e si y o G ei swald,
D-17487 G ei swald, Ge many; o cid.o g/0000-0003-
0685-2696; Email: [email p o ec ed]
Masoud Kazemi −Ba celona Supe compu ing Cen e (BSC),
Ba celona 08034, Spain; Bioma e Designs, Vilnius 09120,
Li huania; o cid.o g/0000-0002-0750-8865;
Email: [email p o ec ed]
Au ho s
Se gi Roda −Ba celona Supe compu ing Cen e (BSC),
Ba celona 08034, Spain; o cid.o g/0000-0002-0174-
7435
Hen ik Te holsen −Depa men o Bio echnology &Enzyme
Ca alysis, Ins i u e o Biochemis y, Uni e si y o G ei swald,
D-17487 G ei swald, Ge many
Jule Ru h Heike Meye −Depa men o Bio echnology &
Enzyme Ca alysis, Ins i u e o Biochemis y, Uni e si y o
G ei swald, D-17487 G ei swald, Ge many; o cid.o g/
0000-0003-0293-4686
Albe Canellas-Solé −Ba celona Supe compu ing Cen e
(BSC), Ba celona 08034, Spain
Comple e con ac in o ma ion is a ailable a :
h ps://pubs.acs.o g/10.1021/acs.jpcb.2c07091
Au ho Con ibu ions
⊥
S.R. and H.T. con ibu ed equally o his wo k.
Figu e 6. Rep esen a i e ca aly ic pose o he WT enzyme and he
success ul in silico e ol ed a ian s. The ca aly ic iad esidues a e
colo ed in ed, he subs a e in cyan, and he mu a ed esidues in
yellow. Possible π−πin e ac ions be ween he subs a e and Phe
esidues a e shown wi h a dashed g een line.
The Jou nal o Physical Chemis y B pubs.acs.o g/JPCB A icle
h ps://doi.o g/10.1021/acs.jpcb.2c07091
J. Phys. Chem. B 2023, 127, 2661−2670
2668
No es
The au ho s decla e no compe ing inancial in e es .
■ACKNOWLEDGMENTS
The au ho s hank Ba celona Supe compu ing Cen e (BSC)
o p o iding he compu a ional esou ces. S.R. hanks he
Spanish Minis y o Science and Inno a ion o Ph.D.
ellowship (FPU19/00608). S.R. and V.G. we e unded by
he Fu u Enzyme P ojec o he Eu opean Union’s Ho izon
2020 Resea ch and Inno a ion P og am (G an Ag eemen
No. 101000327). M.K. hanks Juan De La Cie a-Fo mación
o hei suppo (FJC1018-038089). M.K., V.G., and A.C.-S.
ecei ed unding om he Eu opean Union’s Ho izon 2020
esea ch and inno a ion p og am unde G an Ag eemen
101000607 (P ojec OXIPRO). H.T. was unded by he
Leibniz Associa ion’s s a egic ne wo king unding p og am
Leibniz ScienceCampus ComBioCa .
■REFERENCES
(1) Bo nscheue , U. T.; Huisman, G. W.; Kazlauskas, R. J.; Lu z, S.;
Moo e, J. C.; Robins, K. Enginee ing he Thi d Wa e o Bioca alysis.
Na u e 2012,485, 185−194.
(2) Yi, D.; Baye , T.; Badenho s , C. P. S.; Wu, S.; Doe , M.;
Höhne, M.; Bo nscheue , U. T. Recen T ends in Bioca alysis. Chem.
Soc. Re . 2021,50, 8003−8049.
(3) Lo elock, S. L.; C awshaw, R.; Basle , S.; Le y, C.; Bake , D.;
Hil e , D.; G een, A. P. The Road o Fully P og ammable P o ein
Ca alysis. Na u e 2022,606, 49−58.
(4) Bell, E. L.; Finnigan, W.; F ance, S. P.; G een, A. P.; Hayes, M.
A.; Hepwo h, L. J.; Lo elock, S. L.; Niiku a, H.; Osuna, S.; Rome o,
E.; e al. Bioca alysis. Na . Re . Me hods P ime s 2021,1, No. 46.
(5) Wu, S.; Snajd o a, R.; Moo e, J. C.; Baldenius, K.; Bo nscheue ,
U. T. Bioca alysis: Enzyma ic Syn hesis o Indus ial Applica ions.
Angew. Chem., In . Ed. 2021,60, 88−119.
(6) Wijma, H. J.; Floo , R. J.; Jekel, P. A.; Bake , D.; Ma ink, S. J.;
Janssen, D. B. Compu a ionally Designed Lib a ies o Rapid Enzyme
S abiliza ion. P o ein Eng. Des. Sel. 2014,27, 49−58.
(7) Schymkowi z, J.; Bo g, J.; S iche , F.; Nys, R.; Rousseau, F.;
Se ano, L. The FoldX Web Se e : An Online Fo ce Field. Nucleic
Acids Res. 2005,33, W382−W388.
(8) Rohl, C. A.; S auss, C. E. M.; Misu a, K. M. S.; Bake , D.P o ein
S uc u e P edic ion Using Rose a. Me hods in Enzymology; Else ie
B.V., 2004; Vol. 383, pp 66−93.
(9) Ma ques, S. M.; Planas-Iglesias, J.; Dambo sky, J. Web-Based
Tools o Compu a ional Enzyme Design. Cu . Opin. S uc . Biol.
2021,69, 19−34.
(10) Kulsh esh ha, S.; Chaudha y, V.; Goswami, G. K.; Ma hu , N.
Compu a ional App oaches o P edic ing Mu an P o ein S abili y. J.
Compu .-Aided Mol. Des. 2016,30, 401−412.
(11) Musil, M.; Konegge , H.; Hon, J.; Bedna , D.; Dambo sky, J.
Compu a ional Design o S able and Soluble Bioca alys s. ACS Ca al.
2019,9, 1033−1054.
(12) Goldenzweig, A.; Fleishman, S. J. P inciples o P o ein S abili y
and Thei Applica ion in Compu a ional Design. Annu. Re . Biochem.
2018,87, 105−129.
(13) Shi ke, A. N.; Baso e, D.; Bu e oss, G. L.; Bonneau, R.;
Bys o , C.; G oss, R. A. Towa d Ra ional The mos abiliza ion o
Aspe gillus oyzae Cu inase: Insigh s in o Ca aly ic and S uc u al
S abili y. P o eins 2016,84, 60−72.
(14) Nguyen, V.; Wilson, C.; Hoembe ge , M.; S ille , J. B.;
Aga ono , R. V.; Ku e , S.; English, J.; Theobald, D. L.; Ke n, D.
E olu iona y D i e s o The moadap a ion in Enzyme Ca alysis.
Science 2017,355, 289−294.
(15) Bedna , D.; Bee ens, K.; Sebes o a, E.; Bendl, J.; Kha e, S.;
Chaloupko a, R.; P okop, Z.; B ezo sky, J.; Bake , D.; Dambo sky, J.
Fi eP o : Ene gy- and E olu ion-Based Compu a ional Design o
The mos able Mul iple-Poin Mu an s. PLoS Compu . Biol. 2015,11,
No. e1004556.
(16) Fassio, A. V.; San os, L. H.; Sil ei a, S. A.; Fe ei a, R. S.; de
Melo-Mina di, R. C. nAPOLI: A G aph-Based S a egy o De ec and
Visualize Conse ed P o ein-Ligand In e ac ions in La ge-Scale.
IEEE/ACM T ans. Compu . Biol. Bioin . 2020,17, 1317−1328.
(17) Hon, J.; Bo ko, S.; S ou ac, J.; P okop, Z.; Zendulka, J.; Bedna ,
D.; Ma inek, T.; Dambo sky, J. EnzymeMine : Au oma ed Mining o
Soluble Enzymes wi h Di e se S uc u es, Ca aly ic P ope ies and
S abili ies. Nucleic Acids Res. 2020,48, W104−W109.
(18) Sumbalo a, L.; S ou ac, J.; Ma inek, T.; Bedna , D.;
Dambo sky, J. Ho Spo Wiza d 3.0: Web Se e o Au oma ed
Design o Mu a ions and Sma Lib a ies Based on Sequence Inpu
In o ma ion. Nucleic Acids Res. 2018,46, W356−W362.
(19) Pa elka, A.; Sebes o a, E.; Kozliko a, B.; B ezo sky, J.; Socho ,
J.; Dambo sky, J. CAVER: Algo i hms o Analyzing Dynamics o
Tunnels in Mac omolecules. IEEE/ACM T ans. Compu . Biol. Bioin .
2016,13, 505−517.
(20) Musil, M.; Khan, R. T.; Beie , A.; S ou ac, J.; Konegge , H.;
Dambo sky, J.; Bedna , D. Fi eP o ASR: A Web Se e o Fully
Au oma ed Ances al Sequence Recons uc ion. B ie ings Bioin . 2021,
22, No. bbaa337.
(21) Planas-Iglesias, J.; Opaleny, F.; Ulb ich, P.; S ou ac, J.; Sanusi,
Z.; Pin o, G. P.; Schenkmaye o a, A.; Byska, J.; Dambo sky, J.;
Kozliko a, B.; Bedna , D. LoopG a e : A Web Tool o T ansplan ing
Dynamical Loops o P o ein Enginee ing. Nucleic Acids Res. 2022,50,
W465−W473.
(22) Ka ami, Y.; Rey, J.; Pos ic, G.; Mu ail, S.; Tu é y, P.; de V ies,
S. J. DaReUS-Loop: A Web Se e o Model Mul iple Loops in
Homology Models. Nucleic Acids Res. 2019,47, W423−W428.
(23) Khe sonsky, O.; Lipsh, R.; A izeme , Z.; Ashani, Y.; Goldsmi h,
M.; Leade , H.; Dym, O.; Rogo ne , S.; T udeau, D. L.; P ilusky, J.;
e al.e al. Au oma ed Design o E icien and Func ionally Di e se
Enzyme Repe oi es. Mol. Cell 2018,72, 178e5−186e5.
(24) Anishchenko, I.; Pellock, S. J.; Chidyausiku, T. M.; Ramelo , T.
A.; O chinniko , S.; Hao, J.; Ba na, K.; No n, C.; Kang, A.; Be a, A.
K.; e al.e al. De No o P o ein Design by Deep Ne wo k
Hallucina ion. Na u e 2021,600, 547−552.
(25) Jumpe , J.; E ans, R.; P i zel, A.; G een, T.; Figu no , M.;
Ronnebe ge , O.; Tunyasu unakool, K.; Ba es, R.; Zídek, A.;
Po apenko, A.; e al.e al. Highly Accu a e P o ein S uc u e
P edic ion wi h AlphaFold. Na u e 2021,596, 583−589.
(26) Baek, M.; DiMaio, F.; Anishchenko, I.; Daupa as, J.;
O chinniko , S.; Lee, G. R.; Wang, J.; Cong, Q.; Kinch, L. N.;
Schae e , R. D.; e al.e al. Accu a e P edic ion o P o ein S uc u es
and In e ac ions Using a Th ee-T ack Neu al Ne wo k. Science 2021,
373, 871−876.
(27) Wang, J.; Lisanza, S.; Jue gens, D.; Tische , D.; Wa son, J. L.;
Cas o, K. M.; Rago e, R.; Sa ago i, A.; Milles, L. F.; Baek, M.; e al.e
al. Sca olding P o ein Func ional Si es Using Deep Lea ning. Science
2022,377, 387−394.
(28) Daupa as, J.; Anishchenko, I.; Benne , N.; Bai, H.; Rago e, R.
J.; Milles, L. F.; Wicky, B. I. M.; Cou be , A.; de Haas, R. J.; Be hel,
N.; e al. Robus Deep Lea ning-Based P o ein Sequence Design
Using P o einMPNN. Science 2022,378, 49−56.
(29) Xie, W. J.; Asadi, M.; Wa shel, A. Enhancing Compu a ional
Enzyme Design by a Maximum En opy S a egy. P oc. Na l. Acad. Sci.
U.S.A. 2022,119, No. e2122355119.
(30) Roda, S.; San iago, G.; Gualla , V.Mapping Enzyme-Subs a e
In e ac ions: I s Po en ial o S udy he Mechanism o Enzymes.
Ad ances in P o ein Chemis y and S uc u al Biology; Else ie B.V.,
2020; Vol. 122, pp 1−31.
(31) Alonso, S.; San iago, G.; Cea-Rama, I.; Fe nandez-Lopez, L.;
Coscolín, C.; Mod egge , J.; Ressmann, A. K.; Ma ínez-Ma ínez, M.;
Ma e o, H.; Ba giela, R.; e al.e al. Gene ically Enginee ed P o eins
wi h Two Ac i e Si es o Enhanced Bioca alysis and Syne gis ic
Chemo- and Bioca alysis. Na . Ca al. 2020,3, 319−328.
The Jou nal o Physical Chemis y B pubs.acs.o g/JPCB A icle
h ps://doi.o g/10.1021/acs.jpcb.2c07091
J. Phys. Chem. B 2023, 127, 2661−2670
2669
