Discovery and biochemical characterization of thermostable glycerol oxidases

Vol.:(0123456789)
Applied Mic obiology and Bio echnology (2024) 108:61
h ps://doi.o g/10.1007/s00253-023-12883-9
BIOTECHNOLOGICALLY RELEVANT ENZYMES ANDPROTEINS
Disco e y andbiochemical cha ac e iza ion o  he mos able glyce ol
oxidases
La sL.San ema1· Lau aRo ilio2 · Rui eXiang3· GwenTjallinks1· Vic o Gualla 3· And eaMa e i2 ·
Ma coW.F aaije1
Recei ed: 23 June 2023 / Re ised: 10 Oc obe 2023 / Accep ed: 20 Oc obe 2023 / Published online: 6 Janua y 2024
© The Au ho (s) 2024
Abs ac
Aldi ol oxidases a e p omising ools o he bioca aly ic oxida ion o glyce ol o mo e aluable chemicals. By in eg a ing in
silico biop ospec ing wi h cell- ee p o ein syn hesis and ac i i y sc eening, an e ec i e pipeline was de eloped o apidly
iden i y enzymes ha a e ac i e on glyce ol. Th ee he mos able aldi ol oxidases om Ac inobac e ia Bac e ium, S ep o-
myces he mo iolaceus, and The mos aphylospo a ch omogena ac i e on glyce ol we e disco e ed. The cha ac e iza ion o
hese h ee la oenzymes demons a ed hei glyce ol oxida ion ac i i ies, p e e ence o alkaline condi ions, and excellen
he mos abili ies wi h mel ing empe a u es highe han 75 °C. S uc u al elucida ion o he aldi ol oxidase om Ac ino-
bac e ia Bac e ium highligh ed a cons ella ion o side chains ha engage he subs a e h ough se e al hyd ogen bonds, a
his idine esidue co alen ly bound o he FAD p os he ic g oup, and a unnel leading o he ac i e si e. Upon compu a ional
simula ions o subs a e binding, a double mu an a ge ing a esidue pai a he unnel en ance was c ea ed and ound o
display an imp o ed he mal s abili y and ca aly ic e iciency o glyce ol oxida ion. The he eby desc ibed aldi ol oxidases
o m a aluable panel o oxida i e bioca alys s ha can pe o m egioselec i e oxida ion o glyce ol and o he polyols.
Key poin s
• Rapid pipeline designed o iden i y pu a i e oxidases
• Biochemical and s uc u al cha ac e iza ion o aldi ol oxidases
• Glyce ol oxida ion o mo e aluable de i a i es
Keywo ds Aldi ol oxidases· Fla in· Glyce ol· Cell- ee exp ession· In silico biop ospec ing· Enzyme enginee ing
In oduc ion
The p oduc ion o biodiesel gene a es o e whelming
amoun s o glyce ol, up o 10% o he o al yields (Abbas-
zaadeh e al. 2012). Wi h i s p ice cons an ly d opping,
glyce ol is slowly u ning in o a was e p oduc , challeng-
ing he alue o biodiesel as an al e na i e o ossil uels
(Mo ison 2000; Quispe e al. 2013; Mon ei o e al. 2018;
Baghe i e al. 2015). Chemically, he oxida ion o glyc-
e ol is gene ally a o ded by means o me al ca alys s,
and he ew a ailable non-me al-based s a egies equi e
ha sh condi ions (Villa e al. 2015; Punniyamu hy e al.
2005; Gup a e al. 2017; Shi e al. 2019). Sus ainable
enzyme-based s a egies o glyce ol upcycling a e hus
highly sough a e . Fla oenzyme oxidases a e especially
a ac i e o hese pu poses because hey use molecula
oxygen as elec on accep o , p oduce hyd ogen pe oxide
as sole by-p oduc , and gi e ise o aluable de i a i es
La s L. San ema, Lau a Ro ilio and Rui e Xiang con ibu ed equally
o his wo k.
* Vic o Gualla
ic o [email p o ec ed]
* And ea Ma e i
and ea.ma e i@unip .i
* Ma co W. F aaije
m.w. aai[email p o ec ed]
1 Molecula Enzymology, Uni e si y o G oningen,
Nijenbo gh 4, 9747AG, G oningen, TheNe he lands
2 Depa men o Biology andBio echnology, Uni e si y
o Pa ia, ia Fe a a 9, 27100Pa ia, I aly
3 Ba celona Supe compu ing Cen e (BSC), Ins i ució Ca alana
de Rece ca i Es udis A ança s (ICREA), Ba celona08034,
Spain
Applied Mic obiology and Bio echnology (2024) 108:6161 Page 2 o 14
such as glyce aldehyde o glyce ic acid (Baghe i e al.
2015; Ahmad e al. 2021; Wol enden and Snide 2001;
Ma e i 2006). Alcohol oxidases (EC 1.1.3.13) showed
mode a e ac i i y on glyce ol once hei ac i e si e was
enginee ed o become mo e wide and pola (Nguyen e al.
2018). Un o una ely, hese enzymes a e o en pu i ied
wi h hei la in co ac o in he inac i e semiquinone s a e,
hus hampe ing hei indus ial use. On he o he hand,
ce ain aldi ol oxidases (AldO; EC 1.1.3.41) we e ound
o be mode a ely ac i e on glyce ol wi hou being a ec ed
by he inac i a ing semiquinone o ma ion. Howe e , un il
ecen ly, he known AldOs su e ed om ei he lack in
he mos abili y, low ac i i y, and/o poo exp ession le els
(Heu s e al. 2007; Win e e al. 2012).
AldO om The mopolyspo a lexuosa, a he mophile,
was ecen ly desc ibed as a p omising candida e o
indus ial glyce ol oxida ion, hanks o i s high exp ession
le els, good he mal s abili y, a o able kine ics
p ope ies, and e icien biocon e sion o glyce ol o
glyce a e (Chen e al. 2022). P omp ed by his inding,
we ha e pe o med a mo e ex ended analysis on di e en
genomes, including hose o he mophilic bac e ia. We
ound ha hey ha bo a as numbe o pu a i e AldOs
wi h g ea po en ial as bioca alys s o glyce ol oxida ion.
To apidly explo e and cha ac e ize his sequence o e,
we se ou o in eg a e compu a ional modeling and
expe imen al es ing. In silico biop ospec ing o e s a
cos -e ec i e and e icien ini ial app oach o inc ease he
success a e o he disco e y p ocess. By i s analyzing
he ac i i ies o he pu a i e enzymes compu a ionally
wi h he subs a e o in e es , we can elimina e non-
ac i e enzymes, educing esou ce consump ion (Kamble
e al. 2019). Subsequen ly, high- h oughpu expe imen al
me hods can be applied o alida e he simula ion esul s
in a syne gis ic manne . Mul iple s udies ha e shown
ha cell- ee p o ein exp ession me hods can nowadays
allow e icien enzyme disco e y because hey can be
di ec ly coupled o an ac i i y assay (Sil e man e al.
2020; Ga enne e al. 2021; Kwon and Jewe 2015;
Tamie e al. 2021; Rol e al. 2019; Haslinge e al. 2021;
Rol e al. 2022). Mo eo e , hey o e come he ime-
consuming cons ain s o wo king wi h cells and allow o
apid p o ein p oduc ion bypassing mos o he cloning,
exp ession, and pu i ica ion s eps (Sil e man e al.
2020). In his s udy, we ha nessed his echnology o
iden i ying AldOs ha a e e icien in glyce ol oxida ion
and ea u e p ope ies ha can make hem applicable in
indus ial se ings. Based on s uc u al insigh s and using
compu a ional p edic ions, we designed a a ian ha
success ully imp o es ac i i y on glyce ol. Ou wo k
expands he pool o bioca alys s ha can be used o he
selec i e oxida ion o glyce ol and o he polyols and
demons a es he e icacy o cell- ee p o ein exp ession
o enzyme disco e y.
Ma e ials andme hods
Chemicals andma e ials
The E. coli cell- ee exp ession ki , NEBExp ess® Cell- ee
E. coli P o ein Syn hesis Sys em, and he es ic ion enzyme
DPNI we e bough om New England Biolabs. Ch oma o-
g aphic columns we e om Cy i a. O he chemicals we e
acqui ed om Sigma-Ald ich.
Genome mining
Using he sequence o AldO om T. lexuosa (GenBank:
WP_142259226.1), he PSI-BLAST ool o he Na ional
Cen e o Bio echnology In o ma ion (NCBI) was u ilized,
using de aul se ings, o sea ch o homologues in he mo-
philic bac e ia. Sequence alignmen was done wi h ESp ip
(Robe and Goue 2014). Nume ous p omising sequences
we e sc u inized, and hei s uc u es we e p edica ed wi h he
online AlphaFold ool ColabFold (Jumpe e al. 2021; Mi di a
e al. 2022). Homologues we e selec ed based on he conse -
a ion o he ac i e si e esidues (Heu s e al. 2007; Win e
e al. 2012).
Cloning, ans o ma ion, andmu agenesis
Using he Twis Biosciences ools, sequences encoding o he
chosen candida e enzymes we e codon op imized o exp es-
sion in E. coli and BSAI si es we e added o he 5’- and 3’- e -
mini (TableS1). The syn he ic genes we e o de ed om he
same company and cloned ia he Golden Ga e me hodology
(Engle and Ma illonne 2014) in bo h His-SUMO-PET28a
and pBAD His-SUMO ec o s. A o al o 2 μL o esul ing
PCR p oduc was mixed wi h 40 μL CaCl2 compe en NEB10β
E. coli cells. A e an incuba ion o 30 min on ice, he cells
we e hea shocked a 42 °C o 45 s and kep on ice again
o ano he 5 min. The cells we e allowed o eco e in 250
μL LB-medium a 37 °C o an hou . Subsequen ly, 50 μL
was pla ed on LB-aga pla es con aining 50 μg/mL ampicillin
(amp) and incuba ed a 37 °C o e nigh . Cloning was e i ied
wi h plasmid isola ion and sequencing. P ime s o mu agen-
esis (TableS2) we e o de ed om Eu o ins Genomics, and
mu agenesis was pe o med acco ding o he QuickChange
me hodology (Kunkel 1985). The PCR mix u e consis ed o
12.5 μL P uUl a II Ho s a PCR Mas e mix, 1 μM o bo h
o wa d and e e se p ime , 100 ng empla e plasmid, 2%
DMSO, 0.8 μM MgCl2, and illed o a o al olume o 25 μL
wi h MilliQ wa e .
Applied Mic obiology and Bio echnology (2024) 108:61 Page 3 o 14 61
Cell‑ ee exp ession sc eening assay
Cell- ee exp ession was pe o med wi h NEBExp ess®
Cell- ee E. coli P o ein Syn hesis Sys em. The p o ocol
gi en by he supplie was ollowed, using 250 ng AldO
homologue His6-SUMO-PET28a plasmid. 1.5 mL Eppen-
do ubes we e used as eac ion chambe s, and he incuba-
ion was done a 37 °C, 300 pms o 4 h. A o al o 10 μL o
cell- ee mix u e was used o pe o m a ho se adish pe oxide
(HRP) 4-aminoan ipy ine (AAP)/3,5-dichlo o-2-hyd oxy-
benzenesul onic acid (DCHBS) assay (Vojino ić e al. 2004)
oge he wi h 100 mM glyce ol as subs a e while using 50
mM KPi pH 7.5 as eac ion bu e . P oduc o ma ion was
ollowed wi h a syne gy H1 mic opla e eade a 515 nm
o 15 min a a cons an empe a u e o 25 °C. A o al o 2
μL o cell- ee mix u e was an on a 12% sodium dodecyl
sul a e-polyac ylamide gel elec opho esis (SDS-PAGE) o
con i ma ion o p o ein exp ession.
P o ein exp ession, pu i ica ion,
andcha ac e iza ion
Fo he exp ession, o e nigh cul u es we e made in 5 mL
LB supplemen ed wi h 50 μg/mL amp and incuba ed a 37
°C, 135 pm. Each o e nigh cul u e was esuspended in 500
mL Te i ic B o h medium con aining 50 μg/mL amp, and
he cul u es we e g own a 37 °C, 135 pm in a non-ba led
lask un il an OD600 o ~0.6 was eached. Induc ion was pe -
o med wi h L-a abinose (0.02% inal concen a ion), and
cul u es we e incuba ed a 24 °C, 135 pm o ~16 h. Cells
we e ha es ed by cen i uging (6000 pm, 15 min, 4 °C),
and pelle s we e s o ed a −20 °C.
Fo pu i ica ion, cell pelle s we e esuspended in o 50
mL lysis bu e (150 mM NaCl, 50 mM KPi, pH 8.0) and
dis up ed by sonica ion (5 s on 7 s o , 70% ampli ude o
15 min). The supe na an s we e ha es ed by cen i uging a
11,000 pm o 50 min a 4 °C and loaded on o Ni Sepha ose
g a i y columns con aining 2 mL lysis bu e . Columns we e
washed wi h 3 column olumes o wash bu e (50 mM KPi,
150 mM NaCl, 20 mM imidazole, pH 8.0), and he p o eins
we e elu ed o he columns using 3.5 mL o elu ion bu e
(50 mM KPi, 150 mM NaCl, 500 mM imidazole, pH 8.0).
Using PD10 columns, he elu ion bu e was exchanged o
50 mM KPi pH 7.5, which was used as a s o age solu ion.
P o ein concen a ions we e de e mined using hei la in
abso bance ex inc ion coe icien s, which we e de e mined
by la in abso p ion compa ison o enzymes be o e and a e
dena u a ion wi h 0.1% SDS, o 1 h a 25 °C. The un old-
ing empe a u es o he enzymes we e de e mined using
he The moFAD me hod (Fo ne is e al. 2009). Enzymes
we e concen a ed o ~200 μM in 50 mM KPi pH 7.5 and
dilu ed 20- old in di e en bu e s wi h a pH ange om 5.0
o 9.0. The assay was p e o med om 20 °C ill 95 °C wi h
1 °C/30 s s eps and moni o ed using a RT-PCR he mocycle
(CFX96, Bio-Rad).
S eady‑s a e kine ics
Reac ions we e pe o med wi h 1.0 μM o enzyme and kep
in 50 mM KPi (pH 7.5) a 25 °C while s i ing a 60 pm in
a 1 mL eac ion chambe . Ini ial a es wi h glyce ol we e
measu ed by ollowing he consump ion o oxygen a di -
e en subs a e concen a ions. The assay is con enien in
ha i di ec ly de ec s co-subs a e consump ion by means
o an oxygen elec ode (Oxyg aph plus, Hansa ech Ins u-
men s L d.). Fo e e ence, also he kine ics on xyli ol we e
de e mined using he es ablished HRP-AAP/DCHBS assay
(Vojino ić e al. 2004) (JASCO V-660 spec opho ome e ,
ε515 = 26 mM–1 cm–1). Da a we e p ocessed using G aph-
Pad P ism 6.05 (La Jolla, CA, USA). Enzyma ic ac i i ies
a di e en pH alues we e analyzed a 25 °C wi h 1.0 μM
o enzyme in 50 mM KPi, pH 7.5, and 500 mM o glyce ol.
C ys allog aphic s udies
Fo he c ys alliza ion expe imen s, he pu i ica ion p o o-
col was sligh ly changed o op imize he sample pu i y. The
cell- ee ex ac was il e ed wi h 0.45 μm il e s (Me k) and
loaded on a 5 ml Nickel column, p e iously equilib a ed
wi h lysis bu e (50 mM Hepes, 150 mM NaCl, 20 mM
imidazole pH 8.0), using an Ak a Sys em (Cy i a) equipped
wi h a mul iwa eleng h de ec o (se a 280/350/450 nm).
A e loading was comple ed, he column was washed wi h
lysis bu e un il he abso bance a 280 nm e u ned o base-
line le els. To elu e bound p o eins, a linea g adien o
imidazole (20–500 mM) was applied o he column wi h
an elu ion bu e consis ing o 50 mM Hepes pH 8; 150
mM NaCl; and 500 mM imidazole. F ac ions con aining
His-SUMO-AldOAb we e pooled oge he and concen a ed
wi h Amikon 10k o a sui able olume. The esul ing sam-
ple was incuba ed o e nigh wi h (1:300) homemade His-
SUMO p o ease (1 mg/mL). While incuba ing, he p o ein
mix was dialyzed o e nigh agains 50 mM Hepes, pH 8;
and 50 mM NaCl (s o age bu e ) o emo e he excess o
imidazole. The sample was hen loaded on a His-T ap col-
umn (5 mL, Cy i a) o emo e he His- agged SUMO, while
AldOAb was collec ed in he low- h ough. The sample was
gel il e ed wi h a Supe dex 200 10/300 column, p e iously
equilib a ed wi h s o age bu e , and hen concen a ed o 20
mg/ml wi h Amikon (Me k) 10k. Vapo -di usion si ing-
d op c ys alliza ion was pe o med using he O yx 8 obo
(Douglas ins umen ) and di e en comme cial ki s a 20°C.
AldOAb ga e good di ac ing c ys als in condi ions whe e
monosaccha ides we e p esen as addi i es in he ese oi
solu ion. P omising condi ions we e op imized by manu-
ally p epa ed si ing d op pla es (C yschem, Hamp on) using
Applied Mic obiology and Bio echnology (2024) 108:6161 Page 4 o 14
1+1 μL olumes. The c ys als used o s uc u e solu ion
g ew in he condi ion 0.12 M D-xylulose, 0.1 M HEPES
and MOPS (acid) pH 7.5; 20% / PEG 500 MME; 10% w/
PEG 20’000. X- ay di ac ion da a we e measu ed a 100 K
on he PXI beamline o he Swiss Ligh Sou ce in Villigen
(SLS), Swi ze land. Da a we e scaled using he XDS p o-
g am (Kabsch 2010), and he s uc u e was sol ed by molec-
ula eplacemen (Phase MR) (McCoy e al. 2007) using
he espec i e AlphaFold (Jumpe e al. 2021) coo dina es
as a sea ch model. I e a i e cycles o manual building and
c ys allog aphic e inemen we e pe o med wi h COOT and
REFMAC5 (Mu shudo e al. 2011) om he CCP4 sui e
(Emsley and Cow an 2004). Figu es we e p epa ed wi h
Chime aX e sion 1.3 (Pe e sen e al. 2004), and s uc u al
supe posi ion was pe o med wi h DALI (Holm 2019).
P o ein p epa a ion o insilico analysis
The models we e gene a ed using AlphaFold2 wi h he
de aul op ions o he monome s, and he maximum em-
pla e da e was se o 2022-01-01. The pe - esidue con idence
sco e (pLDDT) o he op- anked models ( ank_0.pdb) was
alida ed, and he models we e p epa ed using he p o ein
p epa a ion wo k low in Sch odinge (Sas y e al. 2013).
This wo k low inco po a es a ious unc ionali ies, includ-
ing p o ona ion a a speci ied pH (in his case, pH 7.5, which
co esponds o he pH o he enzyma ic assays) and es ained
minimiza ion wi h a maximum oo mean squa e de ia ion
( msd) o 0.30 Å. Nex , he subs a e glyce ol was docked
using Glide (F iesne e al. 2006). A g id was gene a ed
o each p o ein, wi h he ca aly ic lysine as he cen e , ol-
lowing he de aul dimensions. The docking p ocedu e was
pe o med using s anda d p ecision, and h ee poses we e
ex ac ed. The docking esul s we e isually inspec ed, and
he bes docking pose, cha ac e ized by op imal dis ances
be ween he ca aly ic a oms and he p esence o co ec
hyd ogen bonds, was selec ed o subsequen simula ions.
P o ein Ene gy Landscape Explo a ion (PELE)
simula ions
PELE (Bo elli e al. 2005), a Mon e Ca lo-based sampling
echnique, was used o s udy p o ein-ligand in e ac ions. In
each s ep, andom ansla ions and o a ions we e applied o
pe u b he ligand. Addi ionally, he lexibili y o he p o ein
was conside ed by applying no mal modes de i ed om he
Aniso opic Ne wo k Model (ANM). Nex , he side chains
o he esidues nea he ligand a e op imized wi h a lib a y
o o ame s o a oid s e ic clashes. Finally, a unca ed New-
on minimiza ion is pe o med, and he new con o ma ion is
accep ed o ejec ed acco ding o he Me opolis c i e ion.
In all PELE simula ions, he pe u ba ion o he ligand
was con ined wi hin a sphe ical box wi h a adius o 6 Å
cen e ed a ound he ac i e si e. The side chain op imiza ion
phase in ol ed all esidues wi hin 6 Å o he ligand. The
PELE simula ions we e execu ed on he Ma eNos um IV
clus e a he Ba celona Supe compu ing Cen e (BSC) using
70 co es, wi h each co e pe o ming 1000 PELE s eps.
The key a iables analyzed and compa ed in hese simu-
la ions we e he enzyme-subs a e in e ac ion ene gies and
he dis ances be ween he NZ a om o he “ca aly ic” lysine
and he p o on o he hyd oxyl g oup o glyce ol (speci i-
cally, bo h hyd oxyl g oups a he ends o he molecule).
These pa ame e s we e calcula ed au oma ically a each s ep
o he PELE simula ions.
Resul s
Iden i ica ion o  he mos able aldi ol oxidase
homologs
A PSI-BLAST was pe o med wi h he p o ein sequence
o AldOT om T. lexuosa (Chen e al. 2022) (GenBank:
WP_142259226.1) as que y inpu , a ge ing he genomes
o known he mophilic bac e ia and he NCBI’s non-
edundan da abase; sequences o pu a i e homologs we e
aligned using ESp ip (Robe and Goue 2014) (Fig.S1).
One candida e, AldOS om S ep omyces he mo iolaceus
(GenBank: WP_189427891.1), was selec ed as he closes
sequence in he mophilic bac e ia. We chose h ee addi ional
candida es om he NCBI’s non- edundan da abase based
on sequence iden i y as well, namely AldOAb om Ac ino-
bac e ia Bac e ium (GenBank: PZN37415.1), AldOMsp om
Mic obispo a sp. H10830 (GenBank: WP_220505403.1),
and AldOCh om The mos aphylospo a ch omogena (Gen-
Bank: WP_093262023.1). All ou candida es sha ed g ea e
han 60% sequence iden i ies o AldOT .
To e alua e he po en ial glyce ol ac i i ies o hese
enzymes, hei s uc u es we e p edic ed using Alpha-
Fold2 (Jumpe e al. 2021), and enzyme-subs a e (glyc-
e ol) simula ions we e hen ca ied ou wi h ou Mon e
Ca lo so wa e, PELE (Acebes e al. 2016) (Fig.1A–E).
AldOT was included in he analysis o compa ison. We
conside ed he dis ance be ween he glyce ol hyd oxyl
g oups and he Nε a om o Lys383 (T. lexuosa esidue
numbe ing; Fig.S1), a conse ed side chain ha assis s he
dep o ona ion o he e minal O1 a om o he polyol sub-
s a e (Fo ne is e al. 2007) (Fig.1F). The dis ances we e
plo ed agains he enzyme-subs a e in e ac ion ene gies.
The loca ion o he ene ge ic minima in he sca e plo s
p edic s highe enzyma ic ac i i ies when he dis ance is
sho e . Taking in o accoun he symme ical na u e o
Applied Mic obiology and Bio echnology (2024) 108:61 Page 5 o 14 61
glyce ol, bo h i s e minal hyd oxyl g oups we e e alua ed
a each s ep and he one displaying he closes dis ance o
he ca aly ic lysine was used in he plo s. We applied a 4
Å cu -o , which ep esen s a easonably pe missi e ange
o p o on ans e . Based on his c i e ion, AldoCh, AldoAb,
and AldOS we e e ained o u he s udies (Fig.1A–C,
E), whe eas AldOMsp was abandoned because he simula-
ions e ealed ene gy minima a dis ances well beyond he
cu -o (Fig.1D). I is impo an o no e ha he selec ion
c i e ia we e no o e ly s ic as we aimed o ha e a mo e
di e se pool o sequences.
Cell‑ ee sc eening o pu a i e aldi ol oxidases
As pa o ou s a egy o apid enzyme disco e y, we
explo ed cell- ee p o ein syn hesis o apid exp ession
and es ing o no el oxidases. One o he mos c ucial ac-
o s in luencing he p o ein yield o cell- ee exp ession is
he s abili y o he mRNA wi hin he mix u e (Ahn e al.
2005). By in oducing a s em-loop s uc u e, like a T7 p o-
mo e a he 3’- e mini, he yield o he cell- ee mix u e
can imp o e by a leas a wo old (Ahn e al. 2005). We
he e o e chose he NEB Exp ess® cell- ee E. coli P o ein
Fig. 1 PELE simula ion esul s using glyce ol as he subs a e ligand.
The X-axis ep esen s he dis ance in angs oms (Å) be ween he
ca aly ic lysine esidue and he hyd oxyl g oup o he subs a e. The
Y-axis ep esen s he in e ac ion ene gies measu ed in kilocalo ies
pe mole (kcal/mol). The panels a e labeled as ollows: A AldOT
om T. lexuosa, B AldOAb om A. Bac e ium, C AldOCh om T.
ch omogena, D AldOMsp om M. sp. H10830. E AldOS om S. he -
mo iolaceus. F The loca ion o he ac i e si e esidues wi h espec o
glyce ol in he p edic ed s uc u e o AldOT

Applied Mic obiology and Bio echnology (2024) 108:6161 Page 6 o 14
Syn hesis Sys em ha is based on a T7 RNA polyme ase
and hus equi es a T7 p omo e . In he li e a u e, i is o en
desc ibed ha di e en enzymes equi e di e en syn he-
sis imes and empe a u es o hei op imal syn hesis (Rol
e al. 2019; Haslinge e al. 2021; Rol e al. 2022). Ou
in-house expe ience showed ha la in-con aining oxidases
a e well exp essed in 1.5 mL Eppendo ubes a 37 °C,
while shaking a 800 pm o 4 h. Longe syn hesis imes
o en lead o dena u a ion o enzymes. We also es ed he
e ec o adding ex e nal FAD (up o 100 μM) o he cell-
ee sys em, bu we did no obse e a bene icial e ec o
la in co ac o addi ion on exp ession le els (Fig.2A). Using
he op imized p o ocol, cell- ee p o ein exp ession wo ked
e y well o all h ee selec ed la op o ein oxidases ha
we e exp essed as His6-SUMO usion p o eins (Fig.2B).
Fo he ini ial sc eening, 10 μL o cell- ee mix u es we e
es ed using he HRP-AAP/DCHBS assay (Vojino ić e al.
2004) oge he and 100 mM glyce ol as subs a e. Hyd ogen
pe oxide o ma ion was moni o ed a 515 nm wi h a syne gy
H1 mic opla e eade . The NEBExp ess® cell- ee mix u e
showed no backg ound oxidase ac i i y, and he ac i i y
o exp essed AldOS , AldOAb, and AldOCh was con i med
using AldOT as a posi i e con ol (Fig.2B). Thus, by cou-
pling cell- ee syn hesis o a ela i ely acile HRP-based
oxidase assay, a di ec cloning and sc eening pipeline was
de eloped o iden i ying glyce ol oxidase. As he oxidase
assay is gene ic, he same p o ocol can be used o any o he
hyd ogen pe oxide gene a ing oxidase.
Exp ession andspec al cha ac e iza ion
o  he he mos able aldi ol oxidases
Fo u he biochemical cha ac e iza ion, la ge amoun s
o pu i ied enzymes we e equi ed compa ed o he yields
o he cell- ee exp ession sys em. Ou wo k also included
he p e iously disco e ed AldOT used as a benchma k
(Chen e al. 2022). The AldO-encoding genes we e cloned
in o a pBAD ec o wi h an N- e minal His6-SUMO ag.
The enzymes we e ecombinan ly o e exp essed in E. coli
NEB10-β, and high p o ein yields we e ob ained upon a in-
i y pu i ica ion: AldOT , 300 mg p o ein/L cul u e; AldOAb,
72 mg/L; AldOCh, 30 mg/L; AldOS , 71 mg/L. Such le els
a e signi ican ly highe when compa ed wi h ha p e iously
ob ained o AldO om Acido he mus celluloly icus 11B
(2.5 mg/L) (Win e e al. 2012). A e incuba ion in 5% /
ace ic acid, he SDS-PAGE sepa a ed p o eins displayed
b igh luo escence unde UV ligh con i ming he p esence
o a co alen ly bound la in (F aaije e al. 1997) (Fig.S2).
Consis en ly, he enzyme abso bance spec a exhibi ed wo
maxima a a ound 350 and 450 nm, wi h shoulde s a 425
Fig. 2 Op imiza ion o cell- ee p o ein exp ession. A Exp ession o
5-hyd oxyme hyl u u al oxidase (70 kDa) using he NEBExp ess®
Cell- ee E. coli P o ein Syn hesis Sys em. Exp ession was pe o med
o 4 h a 37 °C and wi h di e en FAD concen a ions (0–100 μM)
and analyzed by SDS-PAGE (M: molecula weigh ma ke s). B SDS-
PAGE analysis o cell- ee exp ession o a ge ed oxidases. Neg, cell-
ee exp ession wi h no a ge gene; DHFR, posi i e con ol p o ided
by New England Biolabs wi h a molecula weigh o ~20 kDa; T ,
His-SUMO-AldOT (~69 kDa); S , His-SUMO-AldOS (~64 kDa); Ac,
His-SUMO-AldOAb (~65 kDa); Ch, His-SUMO-AldOCh (~65 kDa).
Expec ed posi ions a e indica ed wi h a ed box. The inse shows he
obse ed oxidase ac i i y when he cell- ee mix u es we e es ed
wi h 100 mM glyce ol using he HRP-AAP/DCHBS oxidase assay
Applied Mic obiology and Bio echnology (2024) 108:61 Page 7 o 14 61
and 475 nm. The wo abso p ion maxima a e ypical o an
oxidized la in co ac o , and he ela i ely low wa eleng h o
he maximum a 350 nm is an emblema ic cha ac e is ic o a
his idyl-bound 8α-subs i u ed la in (Jong e al. 1992; Singe
and Edmondson 1980). By SDS dena u a ion and using FAD
as e e ence, he ex inc ion coe icien s o he bound FAD a
452 nm o all AldOs could be de e mined: 12.5 mM−1 cm−1
o AldOT , 12.2 mM−1 cm−1 o AldOAb, 12.3 mM−1 cm−1
o AldOCh, and 15.7 mM−1 cm−1 o AldOS .
S abili y o  healdi ol oxidases
All h ee newly iden i ied AldOs a e a o ably endowed wi h
good he mos abili y p ope ies, and hei mel ing empe a u es
a e as la ge as 85 °C a pH alues highe han 5.5–6.0 (Table1).
Simila pH-dependen p o iles we e obse ed also o hei
enzyma ic ac i i ies (Fig.3). AldOS eme ged as he mos he -
mos able enzyme and, oge he wi h AldOCh, he only one ha
ea u ed some esidual ac i i y a acidic pH alues. These p op-
e ies a e compa able o be e han hose exhibi ed by AldOT
(Chen e al. 2022). Rema kably, all in es iga ed AldOs can be
s o ed a 4 °C wi hou any ac i i y loss e en a e mon hs.
Table 1 The mal s abili ies a di e en pH alues. The mel ing em-
pe a u es o he epo ed aldi ol oxidases we e measu ed in di e en
bu e s (50 mM)
Condi ion Mel ing empe a u e Tm (°C)
AldOAb AldOCh AldOS AldOT V258L_
P259I
AldOT
Ci a e bu e
pH 5.0 68 62 70 61 65
Ci a e bu e
pH 5.5 77 73 80 71 75
KPi
pH 6.0 80 75 82 72 77
KPi
pH 6.5 80 76 84 73 79
KPi
pH 7.0 82 77 85 75 81
KPi
pH 7.5 82 78 85 76 81
T is-HCl pH 8.0 81 77 84 76 80
T is-HCl pH 8.5 82 77 84 76 80
T is-HCl pH 9.0 81 76 84 75 80
Fig. 3 pH dependence o AldO ac i i y. A AldOAb. B AldOCh. C AldOS . D AldOT . 100 mM glyce ol was used o measu e oxidase ac i i y
Applied Mic obiology and Bio echnology (2024) 108:6161 Page 8 o 14
S eady‑s a e kine ics
S eady-s a e kine ic pa ame e s we e de e mined o bo h
xyli ol and glyce ol as subs a es (Table2; Fig.S3). In
con as wi h p e ious li e a u e, AldOT (Chen e al.
2022) p o ed mo e ac i e wi h xyli ol han glyce ol.
Hence, i does no b eak he end shown by he known
AldOs which a o xyli ol o e glyce ol. All in es iga ed
AldOs showed a dec ease in hei a es o ca alysis when
he hyd ogens on glyce ol a e eplaced by deu e ium.
This obse a ion sugges s ha he eac ion a e is limi ed
o a la ge pa by he glyce ol oxida ion s ep in ol ing
a hyd ide ans e om he subs a e o he la in
(TableS3) (Belleau and Mo an 2006). Rema kably,
besides AldOS , he newly epo ed AldOs in his s udy
a e all supe io glyce ol oxidases as compa ed o he
p e iously desc ibed AldOs om A. celluloly icus and
S ep omyces coelicolo , especially ega ding he KM
alues (Heu s e al. 2007; Win e e al. 2012) (Table2).
In his ega d, i is no iceable ha AldOS showed a non-
ca aly ic second minimum in he PELE simula ions,
p o iding a a ionalize o i s highe KM alue and
co obo a ing ou s a egy o enzyme selec ion based
on he ca aly ic dis ance cu -o (Fig.2D, E). The kca
o 6373.1 s−1 o AldOT on glyce ol epo ed in Chen
e al. (2022) is much highe han all kca alues ha we
de e mined o all ou in es iga ed AldOs, including he
enzyme om T. lexuosa (1.6–4.2 s−1; Table2). To ou
knowledge, kca alues g ea e han 100 s−1 ha e a ely
been epo ed o la op o ein oxidases. Besides his
disc epancy wi h he p e ious li e a u e, AldOT showed
he highes ca aly ic e iciency, while AldOS exhibi ed
he highes kca alue. Combined wi h hei good he mal
s abili y, he kine ics and s abili y da a ou line hese wo
enzymes as he mos aluable candida es o indus ial
usage (Tables1 and 2).
S uc u al elucida ion o  AldOAb incomplex
wi hD‑xylulose
The h ee disco e ed AldOs we e es ed o c ys alliza-
ion. AldOCh did no c ys allize, while AldOS c ys als we e
a ec ed by me ohed al winning and we e no u he pu -
sued o s uc u al s udies. AldOAb was c ys allized wi h
xylulose, he p oduc o xyli ol oxida ion. The c ys als
di ac ed o 2.4 Å esolu ion, yielding elec on densi y
maps o excellen quali y (Table3).AldOAb is s uc u ally
e y simila o he homologous enzyme om S. coelicolo
(PDB en y 2VFR (Fo ne is e al. 2007)) wi h a oo -mean-
squa e de ia ion o 1.0 Å o 404 aligned Cα a oms and 61%
sequence iden i y. The o e all s uc u al assembly is com-
posed by 14 α helices and 14 β shee s ha a e o ganized in o
wo dis inc domains, as obse ed in o he membe s o he
VAO amily (Ma e i e al. 1997): an FAD-binding domain
and a subs a e-binding domain (Fig.4A). The plana la in
co ac o si s a he domain in e ace wi h i s 8α-me hyl g oup
co alen ly bound o he Nδ1 o His45. The p esence o his
co alen linkage is a cha ac e is ic ea u e o he VAO-like
enzymes al hough he bond ypically in ol es he Nε2 a he
han he Nδ1 a om o he la inyla ed his idine side chain
(Le e ink e al. 2008). The la in co ac o u he in e ac s
wi h he p o ein h ough many H-bonds ha a e mos ly
media ed by he esidues belonging o he loop 43-47 and he
side chain o Se 103 ha H-bonds o he la in N5. This ne -
wo k o in e ac ions main ains he la in bu ied wi hin he
p o ein sca old. The esidues ha compose he ac i e si e
w ap he isoalloxazine on i s e side (Fig.4B). They include
Gln295, Glu327, A g329, His350, and Lys382 belonging
o he subs a e binding domain, and Se 103, His118, and
His45 being pa o he FAD binding domain. The con o ma-
ion o hese side chains is highly conse ed when compa ed
o he ac i e si e a angemen obse ed in AldO om S.
coelicolo . A well-o de ed molecule o D-xylulose, p esen
Table 2 S eady-s a e kine ic pa ame e s. The s eady-s a e kine ic pa ame e s o he epo ed aldi ol oxidases we e de e mined
a The HRP-coupled assay was used o measu ing kine ics
b Oxygen consump ion was measu ed o de e mining he kine ic pa ame e s
O ganism o o igin KM (mM) kca (s−1)kca / KM (M−1 s−1) Re e ences
Xyli olaGlyce olbXyli olaGlyce olbXyli ol Glyce ol
AldOSc S. coelicolo 0.32 350 13 1.6 4.1 × 1044.6 Heu s e al. 2007
AldOAc A. celluloly icus 0.07 270 1.9 1.3 2.7 × 1044.8 Win e e al. 2012
AldOAb A. Bac e ium 0.03 184 4.2 2.6 14 × 10414 his s udy
AldOCh T. ch omogena 0.04 143 3.5 2.0 12 × 10414 his s udy
AldOS S. he mo iolaceus 0.02 523 1.9 4.2 9.5 × 1048 his s udy
AldOT T. lexuosa 0.03 50 3.1 1.6 10 × 10432 his s udy
V258L_P259I AldOT T. lexuosa 0.04 41 4.3 4.0 11 × 10498 his s udy
V257L_P258I AldOAb A. Bac e ium n.d. 157 n.d. 1.4 n.d. 9 his s udy
Applied Mic obiology and Bio echnology (2024) 108:61 Page 9 o 14 61
in he ese oi solu ion o he c ys alliza ion condi ion, was
clea ly isible in he elec on densi y (Fig.4B). The ligand is
bu ied in he ca aly ic pocke , and i s OH g oups a e engaged
in H-bonds wi h he su ounding side chains. His350 and
Lys382 a e loca ed a he bo om o he subs a e si e whe e
hey in e ac wi h he C1-O1 g oup o he D-xylulose bound
in on o he eac i e N5 a om o he la in. This geome y
is pe ec ly sui ed o he hyd ide ans e om he subs a e
C1 ca bon leading o aldehyde o ca boxylic (e.g., glyce a e
om glyce ol) p oduc s as obse ed in he AldO enzymes
so- a cha ac e ized (Heu s e al. 2007; Win e e al. 2012;
Chen e al. 2022).
When inspec ing he molecula su ace o AldOAb, a
no iceable ea u e is he p esence o a wide unnel ha con-
nec s he ac i e si e o he ex e nal su ace (Fig.5A). A se
o highly conse ed side chains (Fig.S1) ci cumsc ibes he
unnel en ance: Val257, P o258, Me 260, Asn264, Phe286,
P o288, Glu293 and His324 (Fig.5B). I is concei able ha
he subs a e may ini ially loosely in e ac wi h hese esi-
dues. I can hen gain access o he la in h ough he unnel
passageway. Wi h a diame e o abou 10 Å and a leng h o
14 Å, he unnel is long and wide enough o bind polyols
o di e en leng hs and sizes (Table2). Rema kably, he
side chains o ming bo h he unnel wall and he ac i e si e
display o de ed con o ma ions as judged om hei well-
de ined elec on densi ies. Glyce ol, he ocus o ou wo k,
can be assumed o bind in he same posi ion and con o -
ma ion o he C1-C2-C3 skele on o he D-xylulose ligand
wi hou he imposing any la ge con o ma ion changes on he
su ounding p o ein esidues.
Imp o ed mu an design bysubs a e induced‑ i
simula ions
Based on he s eady-s a e kine ic analysis o he di e en
AldOs (Table2), AldOT showed he highes enzyma ic
e iciency and was hen selec ed o he subsequen s ep
o in silico enginee ing. I is widely ecognized ha mos
p o ein mu a ions lead o des abiliza ion, esul ing in a
ade-o be ween ac i i y and s abili y when a emp ing o
op imize ca aly ic p ope ies (Teu l e al. 2022). To mi iga e
he consequences o al e ing na u ally e ol ed sequences,
a ious s a egies can be employed. One such s a egy
in ol es phylogene ic analysis o iden i y e olu iona y
a iable egions ha a e mo e ole an o mu a ions. This
unc ionali y is implemen ed in Ho Spo Wiza d (Sumbalo a
e al. 2018) which anks he di e en amino acid posi ions
acco ding o he e olu iona y a iabili y. Ho Spo Wiza d
iden i ied i e mu able posi ions ha s ood ou because hey
a e loca ed a ound in he en ance o he ca aly ic unnel as
p edic ed om he AldOAb c ys al s uc u e and he AldOT
AlphaFold model: Val258, Me 261, Leu280, Phe287, and
Se 291 (Figs.5B and 6A) Among hese posi ions, Val258
and Me 261, which we e closes o he ligand, we e selec ed
o in silico mu agenesis. Mo eo e , he nex con iguous
esidues P o259 and P o262, espec i ely, we e also a ge ed
o mu agenesis o cons uc double mu an s. Combining
con iguous esidues seems o inc ease he p obabili y o
coope a i e e ec s be ween hem (Ree z and Ca ballei a
2007). In o al, we analyzed 50 single and double mu an s
using PELE (TableS4). The eligible mu a ions o each
si e we e limi ed o hose p esen in a mul iple sequence
alignmen cons uc ed using Consu (Ya i e al. 2023).
Since we aimed a a clea inc ease in ac i i y, he selec ion
c i e ia in his phase we e mo e s ingen compa ed o one
used in biop ospec ing: whe he he ene ge ic minimum is
loca ed a a simila o be e ca aly ic dis ance and exhibi s
imp o ed in e ac ion ene gies compa ed o he wild ype.
Based on his c i e ia, eigh a ian s we e selec ed o
expe imen al alida ion (TableS4), wi h he double mu an
Table 3. C ys allog aphic s a is ics
a Values in pa en heses a e o e lec ions in he highes esolu ion
shell
b Rme ge=Σ|Ii-<I>|/ΣIi, whe e Ii is he in ensi y o i h obse a ion and
<I> is he mean in ensi y o he e lec ion
c The esolu ion cu -o was se o CC1/2,> 0.3 whe e CC1/2 is he Pea -
son co ela ion coe icien o wo “hal ” da a se s, each de i ed by
a e aging hal o he obse a ions o a gi en e lec ion
d The asymme ic uni con ains ou p o ein chains
AldOAb in complex wi h D-xylulose (8OT8)
Resolu ion ange 49.5–2.4 (2.5–2.4)
Space g oup P212121
Uni cell axes (Å) 56.44 109.91 296.71
To al e lec ionsa488355 (48012)
Unique e lec ionsa73280 (7224)
Mul iplici ya6.7 (6.6)
Comple enessa (%) 99.70 (99.82)
Mean I/sigma(I) a8.44 (1.01)
Wilson B- ac o 49.77
R-me gea, b 0.1649 (2.169)
CC1/2a,c 0.998 (0.649)
Re lec ions used in e inemen a73273 (7224)
R-wo ka0.2463 (0.4752)
R- eea0.2886 (0.4573)
N. o non-hyd ogen a oms 12710
Mac omolecules 12414
Ligands 252
Sol en 44
P o ein esiduesd1627
RMS (bonds) (Å) 0.015
RMS (angles) (°) 1.89
Ramachand an a o ed (%) 92.08
Ramachand an allowed (%) 7.42
Ramachand an ou lie s (%) 0.50