Monooxygenase-dehyd ogenase cascade o sus ained enzyma ic
emedia ion o TMA in salmon p o ein hyd olysa es
Rasmus Reea, Øi ind La sena, Sushil Gaykawada, S ee ekha S. Ramananda, An onio Ga cía-
Moyanoa, I ina Elena Chi iacb, Pål Pun e olla, G o Elin Kjæ eng Bje gaa
a NORCE Clima e & En i onmen , NORCE Resea ch AS, Be gen, No way
b Lei a Technological Cen e , Te assa, Ba celona, Spain.
Abs ac
Fish p o ein hyd olysa es hold g ea p omise as nu aceu icals, ye hei applica ion as ood
ing edien s o nu aceu icals is cu en ly limi ed by hei ish-like odo . This odo is mainly
due o he p esence o ime hylamine (TMA), a ola ile biogenic amine esul ing om he
b eakdown o na u ally occu ing ime hylamine-N-oxide (TMAO) in ma ine ish. The
bac e ial ime hylamine monooxygenase mFMO can oxidize TMA in o TMAO using
molecula oxygen and he co ac o nico inamide adenine dinucleo ide phospha e (NADPH).
We ha e es ablished an enzyme cascade which akes ad an age o glucose dehyd ogenase o
ecycle NADPH om NADP+, signi ican ly dec easing he cos o he eac ion and pa ing he
way o using he enzyme sys em in ish p o ein hyd olysa es a ge ed o human
consump ion. We demons a e ha he dual enzyme sys em wo ks in an indus ially ele an
subs a e. Salmon p o ein hyd olysa e ea ed wi h an mFMO/glucose dehyd ogenase cock ail
showed a 75% educ ion in TMA. A ained senso y panel pe cei ed an imp o ed odo ac oss
se e al pa ame e s, including a educ ion in he cha ac e is ic TMA smell.
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
In oduc ion
The No wegian aquacul u e and ishe ies indus y p oduced mo e han 3.2 million ons o
sea ood in 2023 (1). O his, o e 1 million ons we e byp oduc s, de ined as any p oduc
which is no he main p oduc p oduced om a aw ma e ial, such as heads, ames and
backbones. Fish p o ein hyd olysis is a way o u ilizing such byp oduc s h ough con olled
p o eolysis, o en using sub ilisins (2). The p oduc , ish p o ein hyd olysa es (FPH) is a mix
o amino acids and small pep ides which a e sui able o human consump ion. I is an
excellen p o ein sou ce, wi h p o ein con en a ying depending on p oduc ion me hod;
salmon byp oduc s ypically con ain 69-89% p o ein (3–5). FPH has been explo ed as a
sou ce o bioac i e pep ides wi h an ioxidan , an i-hype ensi e and an i-in lamma o y e ec s
(6, 7). Howe e , acco ding o s a is ics o No wegian use, FPH is mainly used in pe ood and
eed o mula ions o a med ish (1) a he han in human nu i ion. The main ba ie s o
consume accep ance o FPH as ood appea s o be i s ishy odo s and la o s (1, 2, 8).
Consequen ly, he e is a demand o no el s a egies which enable he use o hese nu i ious
ish byp oduc -de i ed p o eins in he highe - alue ood ma ke . Se e al compounds a e
known o con ibu e o ish smell, o which ime hylamine (TMA) is a key componen . TMA
is a ola ile, biogenic amine which is o med pos mo em om ime hylamine-N-oxide
(TMAO). TMAO is a na u ally occu ing me aboli e in ish om cold and deep-sea
en i onmen s and is, impo an ly, odo less. TMAO is belie ed o unc ion as an osmoly e (9)
and as a piezoly e, coun e ac ing p essu e-media ed inhibi ion o p o ein unc ion (10, 11).
A e slaugh e ing, TMAO in ish is con e ed o TMA by TMAO- educing bac e ia,
con ibu ing o i s ish-like odo (12–14). The FPH indus y has iden i ied TMA as a key
a ge o imp o ing he senso y p ope ies o FPHs. To emo e i , some indus ial ac o s
cu en ly use nano il a ion o FPH, bu his un a ge ed me hod has he d awback o causing
signi ican p o ein loss and al e ing he nu i ional composi ion (15). An in e es ing
al e na i e o il a ion is o use an enzyme o speci ically a ge TMA. P e iously, we ha e
shown ha he bac e ial ime hylamine monooxygenase mFMO can be used o emo e mos
o he TMA om salmon FPH in a a ge ed app oach (16, 17). The mFMO enzyme is a la in-
con aining monooxygenase (FMO) (18) isola ed om he ma ine gammap o eobac e ium
Me hylophaga aminisul idi o ans (19). I belongs o a amily o closely ela ed bac e ial
FMOs which ca alyze he oxida ion o TMA (16, 20, 21). These bac e ial FMOs oxidize TMA
using molecula oxygen and NADPH as a co ac o , lea ing TMAO and he oxidized co ac o
NADP+ as p oduc s (Figu e 1, op) (19, 21). To enhance compa ibili y wi h indus ial
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
p ocessing condi ions, a he mos able mFMO a ian , e med mFMO_20, was gene a ed
h ough s uc u e-based enginee ing and shown o pe o m well a up o 65°C (17). The
oxida ion eac ion ca alyzed by mFMO uses one molecule o NADPH pe molecule o TMA.
As NADPH is e y expensi e, mFMO-assis ed emo al o TMA in ish p o ein hyd olysa es
will no be indus ially iable unless co ac o consump ion is managed in a cos -e ec i e
manne .
Two s a egies ha e been explo ed o educe he cos o co ac o s in eac ions ha depend on
hem: enzyme-co ac o enginee ing and co ac o ecycling. Enzyme enginee ing can be used
o imp o e he a ini y o a cheape co ac o wi h he same edox unc ion, such as
nico inamide adenine dinucleo ide (NADH) (22) o nico inamide mononucleo ide (23).
Howe e , mu a ion may comp omise o he aspec s o enzyme unc ionali y, such as eac ion
a e and s abili y, equi ing subs an ial sc eening o achie e a use ul enzyme eac ion a e
(24). To ou knowledge, success ul a emp s o change he co ac o speci ici y o FMOs om
NADPH o NADH ha e no been epo ed. Fu he mo e, al hough NADH is cheape han
NADPH i emains p ohibi i ely expensi e o an indus ial p ocess (25). Co ac o ecycling
is hus an a ac i e s a egy o co ac o managemen and has been success ully implemen ed
in se e al sys ems (25). I in ol es egene a ing he edox co ac o , enabling i s euse in
mul iple eac ion cycles. This can be achie ed h ough di ec chemical educ i e egene a ion
(26), homogenous (27, 28) and he e ogenous (29) egene a ion using hyd ogen and
o ganome allic ca alys s, pho oca aly ic egene a ion (30), and elec ochemical egene a ion
(31). Al e na i ely, enzyma ic egene a ion (32–34) uses a seconda y enzyme eac ion, which
educes he co ac o while oxidizing a sac i icial subs a e, o main ain co ac o a ailabili y
and d i e he main eac ion. I has se e al ad an ages: i can be highly speci ic and,
depending on he choice o sac i icial subs a e, oxic componen s and ca alys s can be
a oided. (25). Fo ood p oduc ion in ol ing enzyma ic TMA- emo al, he sac i icial
subs a e and p oduc s mus ood sa e, and he egene a ion enzyme mus be compa ible wi h
bo h he p ocessing condi ions and he bu e equi emen s o he TMA oxidizing enzyme.
Mo eo e , glucose is an inexpensi e subs a e, and his ecycling s a egy allows he use o
he cheape co ac o NADP+ ins ead o NADPH, signi ican ly educing cos s associa ed wi h
co ac o supplemen a ion.
Va ious enzymes ha e been used o egene a e NADPH, including alcohol dehyd ogenase,
phosphi e dehyd ogenase, glucose dehyd ogenase (GDH) and glucose-6-phospha e
dehyd ogenase (35). Phosphi e dehyd ogenase has been used in usion cons uc s o enable
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
sus ained ca alysis media ed by mFMO (36), and a Baeye –Villige monooxygenase (37).
Like FMOs, he la e enzyme belongs o he class B la op o ein monooxygenases (38).
While GDH has no ye been epo ed o use wi h FMOs, i is an a ac i e co ac o -
egene a ing enzyme due o i s widesp ead applica ion, i s simplici y, and he low cos o i s
subs a e, glucose. O no e, glucose dehyd ogenase B (GdhB) has been used o egene a e
NAPDH du ing enzyma ic Ty ian pu ple p oduc ion by an FMO (39). Fo he applica ion
a ea in his s udy, i is impo an o no e ha bo h he glucose subs a e and he p oduc ,
glucono-1,5-lac one (also known as glucono-del a-lac one, o GDL), a e ecognized as sa e
and widely accep ed ood ing edien s (Figu e 1, bo om) (40).
In his s udy, we couple he ac i i y o he he mos able mFMO_20 o he ac i i y o he
glucose dehyd ogenase GdhB om P ies ia mega e ium (p e iously known as Bacillus
mega e ium) (41, 42). We demons a e ha his enzyme cascade, in he p esence o excess
glucose and he oxidized co ac o NADP+, e ec i ely deple es TMA in salmon p o ein
hyd olysa e while ecycling NADPH and p oducing GDL. Ou app oach demons a es he
u ili y o co ac o ecycling o cos educ ion in TMA emedia ion in an indus ially ele an
con ex .
Resul s and discussion
Es ablishing a co ac o ecycling sys em o a TMA monooxygenase
When se ing up a co ac o ecycling sys em o enzyma ic emo al o TMA in FPH, we chose
he GdhB enzyme and glucose as sac i icial subs a e, and o use NADP+ as he added co ac o .
To guide enzyme dosage and co ac o concen a ion in he dual enzyme sys em, we
cha ac e ized he kine ics o GdhB o glucose. The Km o GdhB o glucose was 68.7 mM
(Table S1), a ac o o 8 x 105 highe han ha o mFMO_20 o TMA (17). The Vmax was abou
hal ha o mFMO_20. To compensa e o he lowe e iciency o GdhB, we se up he ecycling
sys em using a glucose concen a ion o 50 mM and a 10:1 a io o GdhB o mFMO_20. Since
NADH is mo e s able and less cos ly han NADPH, ac i i ies o a mFMO_20 and GdhB we e
assessed using bo h NADPH and NADH as elec on dono s. GdhB can ca alyze oxida ion o
glucose by using bo h co ac o s and can as well ca alyze he e e se eac ion using NAD(P)H
and GDL (Supplemen al Figu e S1A). A su icien ly high glucose concen a ion is his equi ed
o d i e he eac ion owa ds GDL o ma ion and NADP+ educ ion. Howe e , mFMO_20 does
no accep NADH as a co ac o (Supplemen al Figu e S1B), necessi a ing use o NADP+.
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
Figu e 1: TMA oxida ion and glucose dehyd ogena ion by FMO and GDH. Scheme o he TMA-
o-TMAO and glucose- o-GDL eac ion cycle. NADPH and NADP+, educed and oxidized nico inamide
adenine dinucleo ide phospha e, espec i ely.
To demons a e ha he mFMO_20/GdhB enzyme sys em ca alyzed he expec ed eac ions, we
conduc ed an enzyme assay using he wo enzymes and NADP+ as a co ac o . Two key ea u es
o he expe imen al design suppo his demons a ion. Fi s , we used NADP+ as a co ac o ,
a he han NADPH, ensu ing ha any TMAO p oduc ion depended on GdhB ac i i y. Second,
TMA was added in a 5:1 mola excess ela i e o he co ac o , so ha he o ma ion o TMAO
in quan i ies exceeding he ini ial co ac o concen a ion would p o ide di ec e idence o
co ac o ecycling. The subs a es and p oduc s we e quan i ied using liquid
ch oma og aphy/mass spec ome y (LC/MS) (Figu e 2, Supplemen al Table S3-S4). TMAO
and GDL we e o med when bo h enzymes, subs a es and a co ac o we e p esen . As expec ed,
TMAO was no o med when any o he componen s we e emo ed. When 100 μM NADPH
was used as co ac o in he absence o GdhB, only 58 µM TMAO was o med, and as expec ed,
no GDL was p oduced. Co ac o ecycling was demons a ed by he o ma ion o 191 µM
TMAO a eac ion wi h only 100 µM NADP+. This con i ms ha co ac o ecycling enabled he
oxida ion o TMA in mola amoun s exceeding he ini ial co ac o concen a ion.
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
Figu e 2: TMA oxida ion d i en by co ac o ecycling Enzyme assays (n=3) wi h he indica ed
componen s (+, p esen ; -, absen ): mFMO_20, GdhB, NADP+ (o NADPH), glucose and/o TMA a he
gi en concen a ions, we e incuba ed o 1 hou a 25°C and analyzed by LC/MS o he p esence o
TMA, TMAO, glucose o GDL, which we e quan i ied agains a s anda d cu e.
The GDL concen a ions we e lowe han expec ed, as we an icipa ed GDL and TMAO o
inc ease in conce . This disc epancy may be due o spon aneous hyd olysis o GDL o gluconic
acid, a p ocess epo ed in he li e a u e (40). Al hough his possibili y was conside ed du ing
LC/MS me hod de elopmen , no ions co esponding o gluconic acid we e de ec ed. A
disad an age o using GdhB o co ac o ecycling in his sys em is i s ela i ely low ca aly ic
ac i i y compa ed o mFMO_20, necessi a ing a ela i ely high concen a ion o glucose o
d i e he ecycling eac ion. We used 50 mM glucose and 100 μM NAD+, which is consis en
wi h he 100 mM glucose and NAD+ concen a ions o 10-500 mM used in p e ious s udies
wi h his enzyme (41, 43, 44). The inal glucose concen a ion in he p o ein hyd olysa e
ob ained in his s udy was ela i ely high a 1.75% w/w, assuming a 10% d y weigh con en
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
(see nex sec ion). Howe e , his le el is compa able o he 1% xylose concen a ion used wi h
hea o achie e b owning and ca amelized la o o odo masking o salmon p o ein
hyd olysa e (45). In compa ison,). To achie e be e TMA emedia ion wi h lowe le els o
glucose in he inal p oduc , i could be use ul o imp o e he ac i i y and subs a e a ini y o
GdhB h ough enzyme enginee ing (46) o h ough immobiliza ion (47, 48).
The dual enzyme sys em educes TMA le els in salmon p o ein hyd olysa es
To demons a e ha he mFMO_20/GdhB enzyme cascade can deple e TMA in an indus ially
ele an p o ein hyd olysa e, a lab-scale p o ease-d i en hyd olysis o salmon heads and ames
(byp oduc s) was pe o med (Figu e 3).
Figu e 3: P oduc ion and e alua ion o enzyme- ea ed FPH a lab scale. The ec angles indica e
p ocess s eps and analyses, while diamonds show b anching poin s, and he pa allelog ams con ain
in e media e and inal p oduc s, and indica es which igu e con ains he associa ed esul . The low o
con ol FPH ac ions a e shown in ed and enzyme- ea ed FPH ac ions in blue. FPH: ish p o ein
hyd olysa e (salmon); LC/MS: liquid ch oma og aphy/mass spec ome y.
Whole, esh salmon we e ille ed, and he heads and ames we e minced in a mea g inde o
p oduce he byp oduc eeds ock. This mince was mixed wi h 50% wa e (w/w) and 0.5%
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
Alcalase 2.4L ( /w biomass), a comme cial sub ilisin endop o ease wi h b oad speci ici y. A e
hyd olysis and hea inac i a ion o he p o ease, he sample was cen i uged o isola e he wa e -
soluble ac ion con aining hyd olyzed pep ides, he ea e e e ed o as FPH. To p o ec he
added mFMO and GdhB enzymes om p o eoly ic clea age and lipid in e e ence, hey we e
in oduced only a e hyd olysis and cen i uga ion, along wi h glucose and NADP+ (Figu e 3).
We ob ained a o al o 4860 ml liquid hyd olysa e which se ed as he subs a e o enzyma ic
TMA emo al ea men (Figu e 3, op middle box). The d y ma e con en was es ima ed o
be 9.3%.LC/MS analysis o FPH samples collec ed du ing he enzyma ic TMA emo al p ocess
showed ha ea men wi h mFMO_20 and GdhB deple ed TMA in a ime-dependen manne ,
educing i o less han 1% o he ini ial in ensi y (Figu e 4A, Supplemen al Table S5). The
TMA in ensi y was unchanged in con ols whe e enzymes, glucose, and NADP+ we e absen .
The TMAO in ensi y was s able h oughou he ime cou se, likely e lec ing he high TMAO
con en in he eshly p epa ed FPH. The glucose concen a ion also emained s able, as i was
added in g ea excess. GDL in ensi y inc eased o e ime in pa allel wi h he educ ion o TMA,
con i ming success ul co ac o egene a ion. The inal d ied con ol and enzyme- ea ed FPHs
we e p epa ed by e apo a ion un il pa ially d ied, ollowed by eeze d ying un il comple ely
d y. Analysis o he e apo a e (Figu e 4B, Supplemen al Table S6) e ealed app oxima ely 50
µM TMA emaining in he con ol FPH, while no TMA was de ec ed in he enzyme- ea ed
e apo a e. This demons a es ha al hough d ying may assis in TMA emo al i is no su icien
o ully deple e i . Quan i ica ion o TMA in he d ied FPHs showed ha 58 ppm emained in
he mFMO_20/GdhB- ea ed FPH, whe eas 208 ppm was e ained in he con ol (Figu e 4C).
Hence, he enzyme ea men oxidized app oxima ely 75% o he TMA con en . A p e ious
s udy epo ed ha he applica ion o nano- and dia il a ion educed TMA om 700 ppm o
100 ppm in cod FPH and om 400 ppm o 100 ppm in salmon FPH (15). Reaching a le el o
100 ppm was associa ed wi h imp o ed TMA as e in ensi y. We did no obse e an inc ease in
TMAO concomi an wi h he dec ease in TMA in he enzyme- ea ed FPH. The 150 ppm
TMAO p oduced may no make enough o a di e ence o be de ec ed be ween samples, gi en
ha he TMAO concen a ion was measu ed be ween 1344 and 1615 ppm, and he a ia ion
be ween eplica e injec ions was be ween 4.9 and 12.2% (Supplemen al Table S6). Nei he did
we de ec GDL in he enzyme- ea ed FPH. The lowe limi o quan i a ion o GDL was 4.2
pmol, while he educ ion in TMA was measu ed a 7.62 pmol (150 ppm). Gi en ha he TMA
con e sion exceeded he GDL quan i a ion limi , we would ha e expec ed o de ec GDL. One
possible explana ion o his appa en disc epancy is hyd olysis o GDL in o gluconic acid,
which would p e en GDL accumula ion and de ec ion.
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
Figu e 4: Co ac o ecycling-d i en TMA oxida ion in salmon p o ein hyd olysa e. A) Time
cou se analysis o he enzyma ic TMA emo al p ocess. TMA, TMAO, glucose and GDL we e
quan i ied a each ime poin by LC/MS. Blue: enzyme- ea ed FPH (p esence o mFMO_20/GdhB);
ed, con ol FPH (absence o mFMO/GdhB). n=6 o each ime poin (0, 30, 60, 90, 120 min) o
con ol FPH and n=4 o enzyme- ea ed FPH. E o ba s show s anda d de ia ion be ween eplica es.
AUC: a ea unde he cu e. B) TMA in he enzyme- ea ed and con ol FPH e apo a es measu ed by
LC/MS (n=3). E o ba s show s anda d de ia ion be ween e apo a e ba ches. C) TMA, TMAO,
glucose and GDL quan i ied by LC/MS in d ied enzyme- ea ed FPH and con ol FPH; n=2 injec ions
o he same sample. E o ba s show s anda d de ia ion.
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
The mobile phase is 10 mM ammonium ace a e, pH 8.0. Bu e A has 3% ace oni ile and
bu e B has 90% ace oni ile. The column is an ACQUITY P emie BEH Amide VanGua d
FIT HILIC column (Wa e s), wi h pa icle size 1.7 µm and in e nal dimensions o 2.1 x 100
mm, i ed wi h a 2.1 x 5 mm gua d column. 3 µl sample o s anda d we e loaded. The me hod
akes 14 minu es and uses a 0.2 ml/min low a e, s a s a 85% B, going o 80% o e 1.6
minu es, hen o 40% a 10 minu es, holding a 40% un il 12.7 minu es, while inal
condi ioning a 80% B happens be ween 12.8 and 14 minu es. An O bi ap Q-Exac i e
(The mo Fishe ) is connec ed online o a Dionex Ul iMa e 3000 UHPLC sys em (The mo
Fishe ). The mass spec ome e is ope a ed in he posi i e mode o ull ion scan, wi h
scheduled swi ching o nega i e mode when GDL elu es a app ox. 2.2 minu es.
Quan i ica ion o he selec ed ions is pe o med in Excalibu Quan B owse (The mo Fishe )
by aking he a ea unde he cu es o he ex ac ed ion ch oma og ams om samples and
s anda ds in de ined RT windows (Table S2). The enzyme ea men o he hyd olysa es was
pe o med in ba ches, and hese ba ches se ed as eplica es o he s a is ical ea men o he
quan i ied ions.
Senso y assessmen by ained panel
D ied enzyme ea ed and con ol FPH we e assessed by a ained senso y panel a No ima
AS, using a Quan i i e Desc ip i e Analysis (ISO 13299:2016) and acco ding o a Gene ic
Desc ip i e Analysis (55). The senso y panel consis s o 8 judges, ained and sa is ying he
equi emen s o ISO 8586-1:2012. The ooms whe e he es was conduc ed we e buil
acco ding o ISO 8589:2007, and ha e indi idual judging boo hs, s anda dized ligh ing, and
i s own en ila ion sys em. The judges we e p esen ed wi h he d ied hyd olysa es (ei he
con ol o enzyme- ea ed), blinded as o which sample was which. Samples we e se ed in a
blocked and andomized ashion o each judge. 1 g am o sample in whi e cups wi h me al
lids we e se ed. They we e e alua ed on he smell pa ame e s o o al smell in ensi y,
ime hylamine smell, boullion smell, swee smell, sou / e men ed smell, seaweed smell, eed
smell, oxidized smell, and ancid smell. The pa ame e s a e desc ibed in Table S3. Each judge
sco ed he in ensi y o he samples 1-9 o each pa ame e , wi h 1 being leas in ense and 9
being mos in ense. Panel a e ages we e compa ed by ANOVA, and co ec ion o mul iple
es ing was done by Tukey’s es . Di e ence be ween con ol and enzyme- ea ed FPH was
deemed signi ican i he co ec ed p- alue was <0.05.
Consume ec ui men o senso y es ing
A o al o 70 adul consume s we e ec ui ed om among he employees o Lei a
Technological Cen e , loca ed in he P o ince o Ba celona (Spain), o ake pa in a senso y
e alua ion ocused on odo pe cep ion. Rec ui men was ca ied ou in e nally by sending an
email in i a ion o s a membe h ough he o ganiza ion’s in e nal mailing sys em. The email
included a b ie desc ip ion o he s udy’s objec i e, he na u e o he senso y e alua ion
(limi ed o smelling wo samples), and he c i e ia o pa icipa ion. Eligible pa icipan s we e
adul s (≥18 yea s) wi h no known ol ac o y impai men s. Employees in e es ed in
pa icipa ing we e di ec ed o an online egis a ion o m, which collec ed demog aphic
in o ma ion (age, gende , loca ion), e i ied eligibili y, and allowed indi iduals o indica e
p e e ed ime slo s. Al hough no s ic quo as we e applied, e o s we e made o ensu e
demog aphic di e si y among he pa icipan s. Be o e he e alua ion, all pa icipan s signed
an in o med consen o m, in compliance wi h e hical esea ch s anda ds app o ed by he
ele an ins i u ional e iew boa d.
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
Consume panel e alua ion
Du ing he senso y session, each pa icipan was p esen ed wi h wo blinded hyd olysa e
samples con ained in odo -isola ed essels. Sample A (enzyma ically ea ed) and sample B
(no ea men ). Bo h samples ha e been p epa ed unde he same condi ions, ensu ing
uni o mi y in quan i y and p esen a ion and ha e been gi en o he panelis s in a oom a a
cons an empe a u e and isola ed om ex e nal odo s o a oid in e e ences. Panelis s
e alua ed speci ic odo - ela ed pa ame e s: ish smell, smell in ensi y, smell eshness, sul u
like smell and ammonia like smell, using in ensi y and accep abili y scales o se en poin s.
The pa ame e s a e desc ibed in Table S4, and he in ensi y desc ip ions a e in Table S5. A
p e e ence es be ween samples was also pe o med (56). Consume s we e asked o ank he
samples in o de o p e e ence. To ob ain he anking o each sample, each ank posi ion was
mul iplied by he numbe o consume s ha had selec ed i , and he sum o he ankings o
each sample was calcula ed. Low alues in ank sum o samples indica ed ha he sample has
mainly been anked in he i s o de o p e e ence.
Acknowledgemen s
All au ho 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 (OXIPRO). LC/MS analyses we e
pe o med a he Depa men o Bioscience, Uni e si y o Be gen, and he au ho s hank
E silia Bi ulco (Uni e si y o Be gen) o suppo wi h he LC/MS da a collec ion. Senso y
assessmen by he ained panel was pe o med a No ima AS.
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
Re e ences
1. Myh e M, Richa dsen R, Nys øyl R, S andheim G. 2024. Analyse ma in es ås o
2023 SINTEF Ocean AS.
2. K is insson HG, Rasco BA. 2000. Fish p o ein hyd olysa es: P oduc ion, biochemical,
and unc ional p ope ies. C i Re Food Sci Nu 40:43–81.
3. Opheim M, Slizy e R, S e en H, P o an F, La ssen E, Kjos NP. 2015. Hyd olysis o
A lan ic salmon (Salmo sala ) es aw ma e ials - E ec o aw ma e ial and
p ocessing on composi ion, nu i ional alue, and po en ial bioac i e pep ides in he
hyd olysa es. P ocess Biochem 50:1247–1257.
4. Idowu AT, Benjakul S, Sin husam an S, Sookchoo P, Kishimu a H. 2019. P o ein
hyd olysa e om salmon ames: P oduc ion, cha ac e is ics and an ioxida i e ac i i y.
J Food Biochem 43:e12734.
5. Ramak ishnan VV, Hossain A, Da e D, Shahidi F. 2024. Salmon p ocessing disca ds:
a po en ial sou ce o bioac i e pep ides - a e iew. Food P od P ocess Nu 6:22.
6. Ne es AC, Ha nedy PA, O’Kee e MB, Fi zGe ald RJ. 2017. Bioac i e pep ides om
A lan ic salmon (Salmo sala ) wi h angio ensin con e ing enzyme and dipep idyl
pep idase IV inhibi o y, and an ioxidan ac i i ies. Food Chem 218:396–405.
7. Guan M-Q, Yu L, Gu H, Fu Q, Liu M-M, Li K, Yang X-R, F am oze B, Guo J-H, Wei
J-J, Li Y-L. 2025. P o ein hyd olysa e om A lan ic salmon (Salmo sala ) imp o es
aging-associa ed neu oin lamma ion and cogni i e decline in a s by eshaping he gu
mic obio a and Th17/T eg balance. In J Biol Mac omol 306:141270.
8. Lindsay RC. 1990. Fish la o s. Food Re In 6:437–455.
9. Yancey PH. 2005. O ganic osmoly es as compa ible, me abolic and coun e ac ing
cy op o ec an s in high osmola i y and o he s esses. J Exp Biol 208:2819–2830.
10. Yancey PH, Fy e-Johnson AL, Kelly RH, Walke VP, Auñón MT. 2001.
T ime hylamine Oxide Coun e ac s E ec s o Hyd os a ic P essu e on P o eins o
Deep-Sea Teleos s. J Exp Zool 289:172–176.
11. Julius K, Al-Ayoubi SR, Paulus M, Tolan M, Win e R. 2018. The e ec s o osmoly es
and c owding on he p essu e- induced dissocia ion and inac i a ion o dime ic LADH.
Phys Chem Chem Phys 20:7093–7104.
12. Ba e EL, Kwan HS. 1985. Bac e ial educ ion o ime hylamine oxide. Annu Re
Mic obiol 39:131–149.
13. Lidbu y I, Mu ell JC, Chen Y. 2014. T ime hylamine N -oxide me abolism by
abundan ma ine he e o ophic bac e ia. P oc Na l Acad Sci U S A 111:2710–2715.
14. Sun J, Mausz MA, Chen Y, Gio annoni SJ. 2019. Mic obial ime hylamine
me abolism in ma ine en i onmen s. En i on Mic obiol 21:513–520.
15. S einsholm S, O e hals Å, Tho esen L, Unde haug J, Kousoulaki K, Aspe ik T. 2021.
Reduc ion in la o -in ense componen s in ish p o ein hyd olysa es by memb ane
il a ion. J Food Sci 86:3855–3867.
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
16. Go is M, Pun e oll P, Rojo D, Claussen J, La sen Ø, Ga cia-Moyano A, Almend al D,
Ba bas C, Fe e M, Bje ga GEK. 2020. Use o Fla in-Con aining Monooxygenases o
Con e sion o T ime hylamine in Salmon P o ein Hyd olysa es. Appl En i on
Mic obiol 86:e02105-20.
17. Go is M, Cea-Rama I, Pun e oll P, Ree R, Almend al D, Sanz-Apa icio J, Fe e M,
Bje ga GEK. 2023. Inc eased The mos abili y o an Enginee ed Fla in-Con aining
Monooxygenase o Remedia e T ime hylamine in Fish P o ein Hyd olysa es. Appl
En i on Mic obiol 89:e00390-23.
18. Paul CE, Egge ichs D, Wes phal AH, Tischle D, an Be kel WJH. 2021. Fla op o ein
monooxygenases: Ve sa ile bioca alys s. Bio echnol Ad 51:107712.
19. Al ie i A, Mali o E, O u R, F aaije MW, Ma e i A. 2008. Re ealing he moonligh ing
ole o NADP in he s uc u e o a la in-con aining monooxygenase. P oc Na l Acad
Sci U S A 105:6572–6577.
20. Choi HS, Kim JK, Cho EH, Kim YC, Kim J Il, Kim SW. 2003. A no el la in-
con aining monooxygenase om Me hylophaga sp. s ain SK1 and i s indigo syn hesis
in Esche ichia coli. Biochem Biophys Res Commun 306:930–936.
21. Chen Y, Pa el NA, C ombie A, Sc i ens JH, Mu ell JC. 2011. Bac e ial la in-
con aining monooxygenase is ime hylamine monooxygenase. P oc Na l Acad Sci U S
A 108:17791–17796.
22. Cahn JKB, We lang CA, Baumschlage A, B inkmann-Chen S, Mayo SL, A nold FH.
2016. A gene al ool o enginee ing he NAD/NADP co ac o p e e ence o
oxido educ ases. ACS Syn h Biol 6:326–333.
23. Black WB, Zhang L, Mak WS, Maxel S, Cui Y, King E, Fong B, Sanchez Ma inez A,
Siegel JB, Li H. 2020. Enginee ing a nico inamide mononucleo ide edox co ac o
sys em o bioca alysis. Na Chem Biol 16:87–94.
24. B inkmann-Chen S, Flock T, Cahn JKB, Snow CD, B us ad EM, McIn osh JA,
Meinhold P, Zhang L, A nold FA. 2013. Gene al app oach o e e sing ke ol-acid
educ oisome ase co ac o dependence om NADPH o NADH. P oc Na l Acad Sci U
S A 110:10946–10951.
25. Wang X, Saba T, Yiu HHP, Howe RF, Ande son JA, Shi J. 2017. Co ac o NAD(P)H
Regene a ion Inspi ed by He e ogeneous Pa hways. Chem 2:621–654.
26. Jones JB, Sneddon DW, Higgins W, Lewis AJ. 1972. P epa a i e-scale Reduc ions o
Cyclic Ke one and Aldehyde Subs a es o Ho se Li e Alcohol Dehyd ogenase wi h in
si u Sodium Di hioni e Recycling o Ca aly ic Amoun s o NAD. J Chem Soc Chem
Commun 856–857.
27. Maenaka Y, Suenobu T, Fukuzumi S. 2012. E icien Ca aly ic In e con e sion
be ween NADH and NAD + Accompanied by Gene a ion and Consump ion o
Hyd ogen wi h a Wa e -Soluble I idium Complex a Ambien P essu e and
Tempe a u e. J Am Chem Soc 134:367–374.
28. Ab il O, Whi esides GM. 1982. Hyb id O ganome allic/Enzyma ic Ca alys Sys ems:
Regene a ion o NADH Using Dihyd ogen. J Am Chem Soc 104:1552–1554.
29. Wang X, Yiu HHP. 2016. He e ogeneous Ca alysis Media ed Co ac o NADH
Regene a ion o Enzyma ic Reduc ion. ACS Ca al 6:1880–1886.
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
30. B own KA, Wilke MB, Boehm M, Hamby H, Duko ic G, King PW. 2016.
Pho oca aly ic Regene a ion o Nico inamide Co ac o s by Quan um Do −Enzyme
Biohyb id Complexes. ACS Ca al 6:2201–2204.
31. Hildeb and F, Kohlmann C, F anz A, Lü z S. 2008. Syn hesis, Cha ac e iza ion and
Applica ion o New Rhodium Complexes o Indi ec Elec ochemical Co ac o
Regene a ion. Ad Syn h Ca al 350:909–918.
32. Ra e GW, Colowick SP. 1957. Enzyma ic P epa a ion o DPNH and TPNH. Me hods
Enzymol 3:887–890.
33. Jia Q, Zheng Y-C, Li H-P, Qian X-L, Zhang Z-J, Xu J-H. 2022. Enginee ing
Isop opanol Dehyd ogenase o E icien Regene a ion o Nico inamide Co ac o s.
Appl En i on Mic obiol 88:10.1128/aem.00341-22.
34. Xu Z, Jing K, Liu Y, Cen P. 2007. High-le el exp ession o ecombinan glucose
dehyd ogenase and i s applica ion in NADPH egene a ion. J Ind Mic obiol Bio echnol
34:83–90.
35. Bachosz K, Zda a J, Bilal M, Meye AS, Jesionowski T. 2023. Enzyma ic co ac o
egene a ion sys ems: A new pe spec i e on e iciency assessmen . Sci To al En i on
868:161630.
36. Rioz-Ma ínez A, Kopacz M, Gonzalo G De, Pazmiño DET, Go o V, F aaije MW.
2011. Explo ing he bioca aly ic scope o a bac e ial la in-con aining monooxygenase.
O g Biomol Chem 9:1337–1341.
37. Pazmiño DET, Snajd o a R, Baas B-J, Ghob ial M, Miho ilo ic MD, F aaije MW.
2008. Sel -Su icien Baeye –Villige Monooxygenases: E ec i e Coenzyme
Regene a ion o Biooxygena ion by Fusion Enginee ing. Angew Chemie In Ed
47:2275–2278.
38. an Be kel WJH, Kame beek NM, F aaije MW. 2006. Fla op o ein monooxygenases,
a di e se class o oxida i e bioca alys s. J Bio echnol 124:670–689.
39. Lee J, Kim J, Song JE, Song WS, Kim EJ, Kim YG, Jeong HJ, Kim HR, Choi KY,
Kim BG. 2021. P oduc ion o Ty ian pu ple indigoid dye om yp ophan in
Esche ichia coli. Na Chem Biol 17:104–112.
40. Ramachand an S, Fon anille P, Pandey A, La oche C. 2006. Gluconic Acid:
P ope ies, Applica ions and Mic obial P oduc ion. Food Technol Bio echnol 44:185–
195.
41. Heilmann HJ, Mäge mann and col HJ, Gassen HG. 1988. Iden i ica ion and isola ion
o glucose dehyd ogenase genes o Bacillus mega e ium M1286 and hei exp ession in
Esche ichia coli. Eu J Biochem 174:485–490.
42. Pauly HE, P leide e G. 1975. D-Glucose Dehyd ogenase om Bacillus mega e ium M
1286 : Pu i ica ion, P ope ies and S uc u e. Hoppe Seyle s Z Physiol Chem
356:1613–1623.
43. Baik SH, Ide T, Yoshida H, Kagami O, Ha ayama S. 2003. Signi ican ly enhanced
s abili y o glucose dehyd ogenase by di ec ed e olu ion. Appl Mic obiol Bio echnol
61:329–335.
44. Baik SH, Michel F, Aghaja i N, Hase R, Ha ayama S. 2005. Coope a i e e ec o wo
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
su ace amino acid mu a ions (Q252L and E170K) in glucose dehyd ogenase om
Bacillus mega e ium IWG3 on s abiliza ion o i s oligome ic s a e. Appl En i on
Mic obiol 71:3285–3293.
45. Kouakou C, Be gé J-P, Ba on R, Le huau L, P os C, Ca dinal M. 2014. Odo
Modi ica ion in Salmon Hyd olysa es Using he Mailla d Reac ion. J Aqua Food P od
Technol 23:453–467.
46. Hua L, Qianqian B, Jian eng Z, Yinbiao X, Shengyu Y, Weishi X, Yang S, Yupeng L.
2022. Di ec ed e olu ion enginee ing o imp o e ac i i y o glucose dehyd ogenase by
inc easing pocke hyd ophobici y. F on Mic obiol 13:1044226.
47. Chansaenpak K, Kamkaew A, Lisnund S, P achai P, Ra wi unki P, Jingpho T, Blay V,
Pinyou P. 2021. De elopmen o a Sensi i e Sel -Powe ed Glucose Biosenso Based
on an Enzyma ic Bio uel Cell. Biosenso s 11:16.
48. Zhou L, Ouyang Y, Kong W, Ma T, Zhao H, Jiang Y, Gao J. 2022. One po
pu i ica ion and co-immobiliza ion o His- agged old yellow enzyme and glucose
dehyd ogenase o asymme ic hyd ogena ion. Enzyme Mic ob Technol 156:110001.
49. Cho CE, Taesuwan S, Malyshe a O V., Bende E, Tulchinsky NF, Yan J, Su e JL,
Caudill MA. 2017. T ime hylamine-N-oxide (TMAO) esponse o animal sou ce oods
a ies among heal hy young men and is in luenced by hei gu mic obio a
composi ion: A andomized con olled ial. Mol Nu Food Res 61:1600324.
50. Leona dos G, Kendall D, Ba na d N, Leona dos G, Kendall D, Ba na d N, Th eshold
O, Leona dos G, Ba na d N. 1969. Odo Th eshold De e mina ions o 53 Odo an
Chemicals Odo Th eshold De e mina ions O 53 Odo an Chemicals 2470.
51. Dai W, He S, Huang L, Lin S, Zhang M, Chi C, Chen H. 2024. S a egies o educe
ishy odo in aqua ic p oduc s: Focusing on o ma ion mechanism and mi iga ion
means. Food Chem 444:138625.
52. Liu L, Zhao Y, Zeng M, Xu X. 2024. Resea ch p og ess o ishy odo in aqua ic
p oduc s: F om subs ance iden i ica ion, o ma ion mechanism, o elimina ion pa hway.
Food Res In 178:113914.
53. Zake i B, Fie e JO, Celik E, Chi ock EC, Schwa z-Linek U, Moy VT, Howa h M.
2012. Pep ide ag o ming a apid co alen bond o a p o ein, h ough enginee ing a
bac e ial adhesin. P oc Na l Acad Sci U S A 109:E690–E697.
54. Mache oux P. 1999. UV-Visible Spec oscopy as a Tool o S udy Fla op o eins, p. 1–
7. In Me hods in molecula biology, ol. 131: Fla op o ein P o ocols.
55. Lawless HT, Heymann H. 2010. Senso y E alua ion o Food. Sp inge New Yo k,
New Yo k, NY.
56. K ame A, Kahan G, Coope D, Papa asiliou A. 1974. A non-pa ame ic anking
me hod o he s a is ical e alua ion o senso y da a. Chem Senses Fla o 1:121–133.
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
Supplemen a y igu es and ables
Figu e S1: A) Ac i i y o pu i ied GdhB (500 nM) wi h glucose (50 mM) and he indica ed oxidized
co ac o (100 µM), and wi h he indica ed educed co ac o and GDL (50 mM), measu ed by abso bance
change a 340 nm. B) Ac i i y o pu i ied mFMO_20 (50 nM) wi h 100 µM ime hylamine (TMA) and
100 µM educed NADPH o nico inamide adenine dinucleo ide (NADH), measu ed by abso bance
change a 340 nm.
Table S1: Kine ic pa ame e s o GdhB using glucose as a subs a e. NADP+ concen a ion.
Pa ame e s o mFMO_20 a e included o e e ence.
Enzyme
Vmax (s-1)
Km (mM)
Speci ici y cons an
(s-1 M-1)
Re e ence
GdhB
0.429
68.68 (glucose)
6.25
This s udy
mFMO_20
0.93
0.83 x 10-3 (TMA)
1.11 x 106
Go is e al, 2023
Table S2: Me aboli es and ela ed analy ical in o ma ion quan i ied by LC/MS.
Me aboli e
Molecula o mula
Mass (g/mol)
Ion
(m/z, cha ge)
Re en ion ime
(minu es)
TMA
C3H9N
59.112
60.08160 (+H)
3.82
TMAO
C3H9NO
75.11
76.07630 (+H)
4.58
Glucose
C6H12O6
180.156
198.09677 (+NH4)
3.78
GDL
C6H10O6
178.14
177.03991 (-H)
2.17
Table S3-S6: Quan i ica ion o selec ed ions by LC/MS. Excel shee a ailable online. Con ains
quan i a ion esul s and s anda d cu es (Table S3) o TMA, TMAO, glucose and GDL used in Figu e 2
(Table S4) and Figu e 4 (Table S5-S6).
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
Table S7: Desc ip ion o smell c i e ia o d y hyd olysa es, employed by ained senso y panel.
Pa ame e
Desc ip ion
To al smell in ensi y
In ensi y o all smells in he sample
T ime hylamine smell
The smell o ime hylamine (TMA)
Swee smell
Rela ed o a swee smell
Sou / e men ed smell
A e men ed sou smell, spoiled ( he smell o a sou dish ag)
Mine al smell
Rela ed o smells o plas e , lime, chalk and d yness
Seaweed smell
Rela ed o esh and d ied seaweed and g eens, g een ea
Feed smell
Rela ed o he smell o ish eed
Oxidized smell
Rela ed o an oxidized smell which eminds you o dus , d awe s and
ca dboa d
Rancid smell
The in ensi y o all ancid smells (g ass, hay, candle wax, pain , allow,
soap)
Table S8: Hedonic scale desc ip ion, used in he consume panel.
Sco e
Accep abili y desc ip ion
1
Dislike e y much
2
Dislike mode a ely
3
Dislike sligh ly
4
Nei he like no dislike
5
Like sligh ly
6
Like mode a ely
7
Like e y much
Table S9: Desc ip ion o odo pa ame e s used in he consume panel.
Odo pa ame e
Desc ip ion
Fishy smell
Deg ee o which a ishy odo is pe cei ed
Smell in ensi y
O e all in ensi y o he sample’s odo
Smell eshness
Pe cep ion o how esh he smell is (pleasan , non-s ale)
Sul u -like smell
Pe cep ion o any sul u - ela ed no es ( o en egg, pungen )
Ammonia-like smell
De ec ion o ammonia o any sha p chemical odo s
.CC-BY 4.0 In e na ional licensemade a ailable unde a
(which was no ce i ied by pee e iew) is he au ho / unde , who has g an ed bioRxi a license o display he p ep in in pe pe ui y. I is
The copy igh holde o his p ep in his e sion pos ed June 19, 2025. ; h ps://doi.o g/10.1101/2025.06.16.658034doi: bioRxi p ep in
