ISSN 0006-2979, Biochemistry (Moscow), 2024, Vol. 89, No. 2, pp. 241-256 © Pleiades Publishing, Ltd., 2024.
241
A Redox-Regulated, Heterodimeric
NADH:cinnamate Reductase in Vibrio ruber
Yulia V. Bertsova
1,a
, Marina V. Serebryakova
1,b
, Victor A. Anashkin
1,c
,
Alexander A. Baykov
1,d
, and Alexander V. Bogachev
1,e
*
1
Belozersky Institute of Physico-Chemical Biology, Lomonosov Moscow State University, 119234 Moscow, Russia
a
e-mail: bertsova@belozersky.msu.ru
b
e-mail: mserebr@mail.ru
c
e-mail: victor_anashkin@belozersky.msu.ru
d
e-mail: baykov@belozersky.msu.ru
e
e-mail: bogachev@belozersky.msu.ru
Received June 24, 2023
Revised October 9, 2023
Accepted October 10, 2023
AbstractGenes of putative reductases of α,β-unsaturated carboxylic acids are abundant among anaerobic and
facultatively anaerobic microorganisms, yet substrate specificity has been experimentally verified for few encod-
ed proteins. Here, we co-produced in Escherichia coli a heterodimeric protein of the facultatively anaerobic ma-
rine bacterium Vibrio ruber (GenBank SJN56019 and SJN56021; annotated as NADPH azoreductase and urocanate
reductase, respectively) with Vibrio cholerae flavin transferase. The isolated protein (named Crd) consists of the
sjn56021-encoded subunit CrdB (NADH:flavin, FAD binding2, and FMN bind domains) and an additional subunit
CrdA (SJN56019, a single NADH:flavin domain) that interact via their NADH:flavin domains (Alphafold2 predic-
tion). Each domain contains a flavin group (three FMNs and one FAD in total), one of the FMN groups being linked
covalently by the flavin transferase. Crd readily reduces cinnamate, p-coumarate, caffeate, and ferulate under
anaerobic conditions with NADH or methyl viologen as the electron donor, is moderately active against acrylate
and practically inactive against urocanate and fumarate. Cinnamates induced Crd synthesis in V. ruber cells grown
aerobically or anaerobically. The Crd-catalyzed reduction started by NADH demonstrated a time lag of several
minutes, suggesting a redox regulation of the enzyme activity. The oxidized enzyme is inactive, which apparently
prevents production of reactive oxygen species under aerobic conditions. Our findings identify Crd as a regulated
NADH-dependent cinnamate reductase, apparently protecting V.ruber from (hydroxy)cinnamate poisoning.
DOI: 10.1134/S0006297924020056
Keywords: anaerobic respiration, cinnamic acid, caffeic acid, enzyme regulation, reactive oxygen species,
rhizosphere, Vibrio
Abbreviations: aCrd, non-flavinylated Crd protein containing only non-covalently bound flavins; fCrd, flavinylated Crd
protein; *FMN,covalently bound FMN residue; MS,mass spectrometry, MV, methyl viologen, m/z, mass-to-charge ratio;
MD,molecular dynamics; ROS,reactive oxygen species.
* To whom correspondence should be addressed.
INTRODUCTION
NADH:2-enoate reductases (EC 1.3.1.31) capable
of reducing α,β-unsaturated carboxylic acids, such as
fumaric, cinnamic, and acrylic acids, are abundant
among anaerobic and facultatively anaerobic micro-
organisms [1-4]. All known 2-enoate reductases are
formed by single polypeptides but are divided into two
non-homologous groups according to their variable
domain composition. The three-domain reductases
invariably contain a FAD binding 2 domain (PfamID:
PF00890), which has a noncovalently bound FAD
BERTSOVA et al.242
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
Fig. 1. Bioinformatic description of V.ruber Crd and its genes. a)The domain composition of the polypeptides CrdA and CrdB
forming the Crd heterodimer. The NADH:flavin, FMN bind, and FAD binding2 domains (residues 1-180, 190-275, and 280-800, re-
spectively) are indicated by different colors. Domain names are taken from the Pfam database. Putative redox-active prosthetic
groups are shown inside the boxes. The asterisk denotes the covalently bound flavin. The truncated versions of Crd obtained in
this study are denoted as d1 and d3. b)Sequence alignment of the fumarate reductase from K. pneumoniae (Frd, UniProt acces-
sion number B5XRB0), acrylate reductase from V. harveyi (Ard, P0DW92), and CrdB protein from V. ruber (GenBank accession
number SJN56021) with Clustal [42]. The three parts of the alignment shown contain the amino acid residues involved in fuma-
rate C1- and C4-carboxylate binding in Frd (marked in blue and green, respectively), the proton transfer to fumarate (marked
in yellow) in Frd, and covalent bonding with FMN in all shown proteins (marked in red). c)The arrangement of the crd-associ-
ated genes in a V. ruber chromosome. ApbE, putative FAD:protein FMN transferase; CrdR, putative transcription regulator.
prosthetic group and is the place of carbonic acid re-
duction. NADH is oxidized in OYE-like (PF00724) or
FAD binding6 (PF00970) domain having noncovalently
bound FMN or FAD, respectively [2,3]. The third do-
main (commonly FMN bind; PF04205) mediates elec-
tron transfer between the above domains via a flavin
group (FMN) which is covalently bound by a phosphoe-
ster bond [4-6].
Many bacteria of the class Clostridia have a two-
domain 2-enoate reductase [1, 7], formed by OYE-like
and Pyr redox2 (PF07992) domains. The reductases of
this group contain a [4Fe-4S] cluster and noncovalent-
ly bound FAD and FMN in a 1 : 1 : 1 ratio as prosthetic
groups [8, 9]. In saccharolytic clostridia, two-domain
2-enoate reductases convert a broad range of 2-enoates,
whereas the enzymes from proteolytic clostridia are
highly specific and convert only cinnamate and its
derivatives [10, 11]. All known 2-enoate reductases do
not appear to be regulated at activity level, except by
substrates and products.
The genome of the red facultatively anaerobic ma-
rine bacterium Vibrio ruber [12] encodes a polypeptide
formed by 806 amino acid residues and annotated as
urocanate reductase in GenBank (SJN56021) and in
UniProt (A0A1R4LHH9). This polypeptide, which we
designate as CrdB, based on the substrate specificity
of its containing enzyme (see below), encompasses the
abovementioned FAD binding2 and FMN bind domains
(Fig. 1a). Five putative substrate-binding residues, iden-
tified in the FAD binding 2 domains of CrdB and acry-
late reductase of Vibrio harveyi by comparison with
fumarate reductase [4, 13, 14] (Fig.1b), are identical or
very similar, suggesting that CrdB is an acrylate reduc-
tase. The FMN bind domain of CrdB, like its counter-
CINNAMATE REDUCTASE OF V. ruber 243
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
parts in known 2-enoate reductases, contains the motif
DALSGAS
257
recognized by flavin transferase for cova-
lent FMN attachment to the serine residue [15]. Note-
worthy, the gene apbE encoding the flavin transferase
(GenBankID: SJN56016) locates nearby to crdB in one
of the two V. ruber chromosomes (Fig. 1c). Altogether,
these characteristics of CrdB classify it as a three-do-
main 2-enoate reductase with a likely acrylate reduc-
tase activity.
However, the CrdB protein differs from estab-
lished three-domain 2-enoate reductases by its third,
NADH:flavin domain (PF03358) replacing the NADH-
oxidizing domains OYE-like or FAD binding 6 (Fig. 1a).
Furthermore, the crd-operon of V. ruber contains, up-
stream of the crdB gene, the gene crdA encoding a pro-
tein (SJN56019) (Fig.1c) formed by a single NADH:flavin
domain that is homologous (51% identity, 65% similar-
ity) to the similar domain of CrdB. As NADH:flavin do-
mains can dimerize [16], this observation raised an in-
triguing possibility of Crd functioning as a CrdA/CrdB
heterodimer.
Keeping in mind the difficulty in the theoretical
prediction of the enzymatic activities and substrate
specificities of NADH:2-enoate reductases, we have
produced the V. ruber Crd protein in Escherichia coli
cells and characterized the isolated protein. The data
reported below identify Crd as a regulated NADH-de-
pendent reductase active against cinnamic acid and its
various derivatives.
MATERIALS AND METHODS
Bacterial strains and growth conditions. V. ru-
ber DSM16370 was obtained from the Leibniz Institute
collection of microorganisms and cell cultures (DSMZ).
The cells were grown aerobically or anaerobically at
28°C in a mineral medium containing 20 g/liter NaCl,
0.75 g/liter KCl, 1.2 g/liter MgSO
4
×7(H
2
O), 0.5 g/liter NH
4
Cl,
0.5 mM Na
2
HPO
4
, 2 g/liter sucrose, 0.5 g/liter yeast ex-
tract, and 50 mM Tris-HCl (pH 8.0). When required, the
medium was supplemented with 2-10 mM acrylate, cin-
namate, p-coumarate, caffeate, or ferulate. Cleared cell
lysate was prepared as described elsewhere [17] and
assayed for MV:cinnamate reductase activity. E. coli
cells were grown at 37°C in LB. Where indicated, the
LB medium was supplemented with 20 µg/ml chloram-
phenicol and(or) 100 µg/ml ampicillin.
Construction of expression vectors. The ex-
pression vector for the CrdAB protein with C-terminal
6×His-tagged CrdB was constructed by amplifying the
crd-operon from the genomic V. ruber DNA. A high-
fidelity Tersus polymerase (Evrogen, Russia) and the
primers V_ruber_DIR/V_ruber_REV were used (prim-
er sequences are listed in Table S1 in the Online Re-
sourse1). The resulting 3276-bp fragment was cloned
into the pBAD-TOPO vector (Invitrogen, USA), yielding
the pTB_CRD4 plasmid. The plasmid was transformed
into E.coli/pΔhis3 [18] or BL21 strains.
The expression vector for the d1 fragment (full-
size CrdA and residues 1-184 of CrdB) was construct-
ed by amplifying the corresponding DNA fragment of
the crd-operon by PCR with a Tersus PCR kit and the
primers V_ruber_DIR/VR_P1R using V. ruber genomic
DNA as the template. The resulting 1401-bp fragment
was cloned into the pBAD-TOPO vector, yielding the
pTB_d1CRD17 plasmid. To construct an expression
vector for the d3 fragment (residues 281-806) of CrdB,
the 1688-bp fragment was amplified by PCR with
the VR_3P2_NdeD/Xho_rhod_rev primer pair and the
pTB_CRD4 plasmid as the template. The amplified frag-
ment was cloned into the pSCodon vector (Delphi Ge-
netics, Belgium) using the NdeI/XhoI sites, resulting
in the plasmid pSC_d3CRD2. The pTB_d1CRD17 and
pSC_d3CRD2 plasmids were transformed into E. coli
BL21 strain.
Crd production. The 6×His-tagged Crd proteins
with or without covalently bound FMN (fCrd and aCrd,
respectively) and the d1 and d3 fragments of Crd were
produced in E. coli BL21 or pΔhis3 cells and purified
using metal chelate chromatography as described pre-
viously [5]. The extinction coefficients ε
450nm
for fCrd,
aCrd, d1 fragment, and d3 fragment determined after
SDS treatment [4] were 42, 31, 22, and 11 mM
–1
 cm
–1
,
respectively.
Analysis of flavins. Non-covalently bound flavins
were extracted from Crd preparations by trifluoroace-
tic acid (TFA) and separated by HPLC [5]. SDS-PAGE
was performed using 12.5% (w/v) polyacrylamide gels
[19]. The gels were stained for protein with PageBlue
TM
solution (Fermentas, Lithuania). Covalently bound fla-
vins were detected by scanning unstained gels with
a Typhoon
TM
FLA 9500 laser scanner (GE Healthcare,
USA) with excitation at 473 nm and the detection of
emission using the SYBR Green II protocol according
tothe manufacturers recommendations.
For mass-spectral measurements, gel pieces of
approximately 2 mm
3
were excised from the protein
bands of the Coomassie-stained gel and destained
by incubating in two 0.1-ml volumes of 40% (by vol-
ume) acetonitrile solution containing 20 mM NH
4
HCO
3
,
pH 7.5, dehydrated with 0.2 ml of 100% acetonitrile
and rehydrated with 5 μl of the digestion solution con-
taining 15 µg/ml sequencing-grade trypsin (Promega,
USA) in 20 mM aqueous solution of NH
4
HCO
3
, pH 7.5.
Digestion was carried out at 37°C for 1 h. The result-
ing peptides were extracted with 5 μl 30% acetoni-
trile solution containing 0.5% TFA. A 1-μl aliquot of
the in-gel tryptic digest extract was mixed with 0.5μl
of 2,5-dihydroxybenzoic acid solution (40 mg/ml) in
30% acetonitrile containing 0.5% TFA and left to dry on
a stainless-steel target plate. MALDI-TOF MS analysis
BERTSOVA et al.244
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
was performed on an UltrafleXetreme MALDI-TOF-
TOF mass spectrometer (Bruker Daltonik, Germany).
The MH
+
molecular ions were measured in a reflector
mode; the accuracy of monoisotopic mass peak mea-
surement was within 30ppm. Spectra of fragmentation
were obtained in a LIFT mode; the accuracy of daugh-
ter ion measurement was within 1 Da. Mass-spectra
were processed with the FlexAnalysis 3.2 software
(Bruker Daltonik). Protein identification was carried
out by MS+MS/MS ion search, using Mascot software
version2.3.02 (Matrix Science, USA), through the Home
NCBI Protein Database. One each of missed cleavage,
Met oxidation, Cys-propionamide, and Ser(Thr)-FMN
were permitted. Protein scores greater than 94 were
considered significant (p<0.05).
Enzymatic activities. NADH- and methyl viologen
(MV)-supported activities of Crd preparations were de-
termined at 25°C by following NADH or MV oxidation
spectrophotometrically at 340 or 606 nm, respectively,
with a Hitachi-557 spectrophotometer [3]. The assay
medium contained 1 mM MV, 0.03-1 mM electron ac-
ceptor, and 100 mM Tris-HCl (pH 8.0) or 120 μM NADH,
0.05-1 mM electron acceptor, 10 mM glucose, 5 U/ml
glucose oxidase, 5 U/ml catalase, and 100 mM Mes-KOH
(pH 6.5) in a completely filled and sealed 3.2-ml cu-
vette. MV was pre-reduced with sodium dithionite until
the absorbance at 606 nm of approximately 1.5 was ob-
tained, which corresponds to the formation of ~100 μM
reduced MV. Unless otherwise noted, Crd or its variant
was preincubated with NADH or MV for 5min before
the reaction was started by adding an electron accep-
tor. The NADH-supported reaction was also measured
under aerobic conditions without glucose oxidase and
catalase. Crd activities were calculated using the ex-
perimentally verified NADH:acceptor and MV:acceptor
stoichiometries of 1:1 and 2:1, respectively.
The assay medium for measuring FMNH
2
:cinna-
mate reductase activity contained 100 μM FMN, 1 mM
cinnamate, and 100 mM Tris-HCl (pH 8.0). FMN was
pre- reduced with sodium dithionite until the absor-
bance at 450 nm decreased to a value of ~0.3, which
corresponds to formation of ~70 μM reduced FMN.
FMNH
2
oxidation was followed spectrophotometrically
at 450nm.
Phenylpropionate dehydrogenase activity was as-
sayed by measuring the reduction of 2,6-dichloropheno-
lindophenol (DCPIP, ε
600
= 22 mM
–1
 cm
–1
) at 600 nm[20].
The assay mixture contained 2 mM Tris-phenylpropi-
onate, 2 mM phenazine methosulphate, 25 μM DCPIP,
and 100mM Mes-KOH (pH6.5).
The Michaelis–Menten parameters of MV-sup-
ported acrylate reduction by Crd were estimated from
duplicate measurements of the initial rates of MV oxi-
dation using 0.03-10 mM acrylate and 40 nM Crd con-
centrations. For other electron acceptors, rates were
obtained from the integral kinetics of their reduction
until its completion, as monitored by absorbance
at 606 nm. Rates were estimated at 100 time points
along the progress curve as the slopes of the tangents
(-d[MV]/dt) using MATLAB (The MathWorks, Inc.).
Theresidual electron acceptor concentration was cal-
culated at each point from A
606
using the MV : accep-
tor stoichiometry of 2 : 1 and assuming that the lim-
iting value of A
606
corresponds to 100% conversion of
the electron acceptor. The Michaelis–Menten equation
was fitted to the rate data using non-linear regression
analysis.
Identification of the product of cinnamate
reduction. Cinnamate reduction was performed in
3.2 ml of the medium containing 100 μM cinnamate,
10 mM sodium dithionite, 50 μM MV, and 100 mM
Tris-HCl (pH 8.0). The reaction was initiated by add-
ing 0.5 μM Crd and terminated after 30min by adding
5% (vol/vol) HClO
4
. In the control experiment, HClO
4
was added before Crd. Precipitated protein was re-
moved by centrifugation, and the substrate and prod-
ucts were extracted from the samples with diethyl
ether (2 × 1.5 ml). The ether was evaporated under a
stream of air, and the residue was dissolved in 1 ml
of 25 mM potassium phosphate (pH 6.5) (medium A).
The samples were separated by HPLC on a ProntoSil-
120-5-C18 AQ column using a Milichrom A-02 chro-
matograph (both from Econova, Russia). The column
was pre-equilibrated with medium A and eluted with a
linear gradient from 0% to 10% methanol in medium A
at a flow rate of 0.2 ml/min with UV detection at 210
and 258 nm.
Quantitative reverse transcription polymerase
chain reaction (RT-qPCR). RNA extraction from V.ruber
cells and cDNA synthesis were performed as described
previously [17]. RT-qPCR assays were performed with
qPCRmix-HS SYBR kit (Evrogen), using the cDNA prepa-
rations as templates and VR_69U19/VR_273L20 as
primer pair for crdB. 16S rRNA was used for data nor-
malization (the primer pair 16s FW/16s RV). Serial di-
lutions of V. ruber genomic DNA, which contains the
genes for CrdB and 16S rRNA in a 1 : 8 ratio, were used
for calibration.
Bioinformatics. The three-dimensional structure
of flavin-deficient Crd was predicted from the amino
acid sequences of CrdA and CrdB using AlphaFold2
(version 2.2.0) [21] through the ColabFold advanced
notebook [22]. Multimer model prediction was per-
formed using default parameter settings. The Ramach-
andran plot indicated that 96.7% residues have favor-
able dihedral angles in the modelled structure.
Ligands (flavins, NADH, cinnamate) were docked
in sequence into the obtained Crd structure using
AutoDock Vina [23] in the following sequence: *FMN,
FAD, FMN
B
, FMN
A
, NADH, cinnamate. Space cells rang-
ing from 12×12×12 to 30×30×30 Å
3
were selected for
docking in the appropriate regions of the protein based
CINNAMATE REDUCTASE OF V. ruber 245
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
on the known structures of homologous proteins.
Each docking experiment was performed in triplicate.
The derived Crd structure containing one FAD and
three FMN groups was finally equilibrated by mo-
lecular dynamics (MD) simulations for 150 ns with
AMBER 22 package [24] (https://ambermd.org/). Details
of the simulation are found in Electronic Supplementa-
ry Materials (Fig.S4 legend in the Online Resourse1).
RESULTS
Production and characterization of V. ruber Crd.
The CrdAB operon of V.ruber genomic DNA was ampli-
fied and cloned into an expression vector that added
a 6×His tag at the C-terminus of CrdB. The FMN bind
domain of CrdB harbors the sequence DALSGAS
257
,
similar to the flavinylation motif Dxx(s/t)gA(T/S) recog-
nized by flavin transferase for covalent attachment of
FMN via a threonine or serine hydroxyl [15]. To test
whether the predicted flavinylation really occurs, the
crd-operon genes were expressed in E. coli cells in the
presence or absence of the auxiliary pΔhis3 plasmid
[18] that encodes the flavin transferase ApbE capable
of flavinylating FMNbind domains in various proteins
[15]. The recombinant fCrd and aCrd proteins pro-
duced in the ApbE-containing and ApbE-lacking cells,
respectively, were isolated by metal affinity chroma-
tography. The final protein preparations were of intense
yellow color, and their visible spectra (Fig. 2a) were
characteristic of flavoproteins.
SDS-PAGE of both Crd preparations revealed one
major and three-four minor protein bands (Fig. 2b,
left image). Bands 1 and 3 were identified by the
MALDI-MS analysis after trypsin digestion as V. ruber
CrdB (sequence coverage of 79%) and CrdA (sequence
coverage of 95%), respectively. Band 2 belonged to
E. coli peptidyl-prolyl cis-trans isomerase SlyD, which
exhibits high intrinsic affinity to Ni-agarose and is a
common contamination of the 6×His-tagged proteins
isolated from E. coli [25]. Band 1 derived from fCrd
fluoresced when illuminated at 473nm, indicating co-
valently bound flavin (Fig. 2b, right image). No bound
flavin was detected by this method in aCrd. These find-
ings provide convincing evidence for ApbE-mediated
covalent flavinylation of CrdB. It should be noted that
the non-covalently bound flavins, evidently present in
both Crd preparations (Fig.2a) are removed from the
proteins during SDS-PAGE.
The CrdB flavinylation site was identified by
MALDI-MS and MS/MS after trypsinolysis. The pep-
tide I
242
IEGQTLNVDALSGASETSHAVIDGVAK
269
contain-
ing the underlined flavinylation motif should have
demonstrated the monoisotopic MH
+
masses of 2795.3
and 3233.5 in the non-flavinylated and flavinylated
forms, respectively. Both predicted signals were indeed
Fig. 2. Characterization of the Crd preparations. a)Electron-
ic absorption spectra of the full-length and truncated (d1 and
d2) Crd preparations at 10 μM concentrations. Full-length
Crd was produced in ApbE-containing or ApbE-lacking E. coli
cells (fCrd and aCrd, respectively). b) SDS-PAGE of fCrd and
aCrd. The gel was stained with Coomassie Blue (left image)
or scanned under illumination at 473 nm without staining
(right image). The protein load was 5 µg per lane. Bars with
numbers on the left side denote the positions and molecular
masses of marker proteins. The protein bands identified by
MALDI-MS analysis are marked by numbers. c)HPLC separa-
tion of non-covalently bound flavins in Crd and its truncated
variants. The retention volumes for authentic FAD, FMN, and
riboflavin(Rf) are indicated by arrows.
BERTSOVA et al.246
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
observed on the mass-spectrum of the tryptic digest of
fCrdB. The presence of the former signal may reflect
either incomplete modification or partial loss of the
flavin group during SDS-PAGE and trypsin proteolysis.
No other candidates for flavinylated peptides were ob-
served in the tryptic digest of CrdB and CrdA by a pep-
tide fingerprint Mascot search.
The MS/MS analysis of the peptide with m/z of
3233.5 (Fig.S1 in the Online Resourse1) indicated that
the main signals in the spectrum of fragmentation (m/z
of 2794 and 2776) resulted from FMN loss in the full
or dehydrated form, respectively, depending on which
of the two covalent bonds, C–O or O–P, connecting the
FMN moiety to the serine residue is broken. Greater
intensity of the signal with m/z of 2776 indicated that
the former bond is broken with greater propensity,
resulting in serine dehydration [26]. Less intense but
still reliable peaks in the spectrum starting from the
fragment with m/z = 2776 also matched a series of the
y-ions generated by double breaks during the fragmen-
tation (Fig. S1, red letters, see the Online Resourse 1).
Here, mass difference for the Ser257 position (69 Da)
corresponded to the dehydrated serine, demonstrat-
ing that fCrdB produced in ApbE-containing cells
harbors an FMN residue covalently bonded to Ser257
of the predicted flavinylation motif in the FMN bind
domain.
The non-covalently bound flavins were extracted
from Crd and separated by HPLC (Fig. 2c). Both the fCrd
and aCrd preparations were found to contain non-co-
valently bound FAD and FMN at a ratio of approxi-
mately 1:2.
To determine their domain localization, we pre-
pared the genetic constructs encoding separately two
NADH:flavin domains of CrdA and CrdB (d1 in Fig. 1a)
or the FAD binding2 domain of CrdB (d3 in Fig. 1a).
Theisolated corresponding proteins exhibited the spec-
tra characteristic of flavoproteins (Fig.2a). Theflavins
present in them were again extracted and identified
by HPLC (Fig.2c). The non-covalently bound FMN was
exclusively localized in the d1 fragment (NADH:flavin
domains) whereas FAD was found in the d3 fragment
(FAD binding 2 domain). The evident corollary is that
CrdAB contains two non-covalently bound FMN mole-
cules in the oxidoreductase NADH:flavin domains, one
non-covalently bound FAD molecule in the FAD bind-
ing2 domain, and one covalently bound FMN residue
in the FMNbind domain.
The specificity of Crd for naturally occurring
α,β-unsaturated carbonic acids. The similarity of the
putative substrate-binding residues in the FAD bind-
ing 2 domains of Crd and the acrylate reductase Ard
of V. harveyi (Fig. 1b) raised the possibility that Crd
also possesses acrylate reductase activity. Indeed, the
Crd preparation obtained by crdAB-apbE coexpression
(fCrd) could reduce acrylate using reduced methyl
viologen (MV) as the electron donor, but at a much
lower rate (1.3 s
–1
) compared with that of a specif-
ic acrylate reductase (19 s
–1
[4]). Conversely, the K
m
value was much greater for fCrd (770 versus 16 µM),
making its catalytic efficiency k
cat
/K
m
640 times lower
(0.0019 µM
–1
 s
–1
) compared with the acrylate reductase
(1.2 µM
–1
 s
–1
[4]). fCrd exhibited low but measurable ac-
tivity against fumarate but was virtually inactive with
methacrylate, crotonate, and urocanate (Table1).
In contrast, cinnamate and its hydroxy deriva-
tives were readily reduced by fCrd (k
cat
of 22-47 s
–1
) and
demonstrated low K
m
values of 2.2-7.1 µM (Table 1).
These findings classify Crd as a cinnamate reductase
that is also active with other natural α,β-unsaturated
cinnamate derivatives.
To determine the product of cinnamate reduction
by fCrd, we identify it using HPLC. The Crd-catalyzed
reaction was found to completely convert cinnamate
to phenylpropionate (Fig. 3, a and b). The identifica-
tion of the carbonic acids was supported by the ob-
servation that the peak heights on the elution profiles
monitored at 210 and 258nm were nearly equal for cin-
namate but differed approximately 40-fold for phenyl-
propionate, in accordance with the UV-spectra of these
compounds (Fig. 3c). These findings indicate that Crd
reduces the α,β double bond in (hydroxy)cinnamic ac-
ids to yield their saturated derivatives. fCrd activity
in the reverse reaction of phenylpropionate oxidation
with phenazine methasulfate and dichlorophenolin-
dophenol as electron acceptors was measurable but
quite low (0.1s
–1
).
Induction of Crd synthesis in V. ruber cells by
unsaturated carbonic acids. To determine whether
the found Crd substrates induce Crd synthesis in V.ru-
ber, cell growth was performed in their presence and
the MV:cinnamate reductase activity was measured in
cleared cell lysates. All cultivations were performed in
duplicate both in the absence and in the presence ofO
2
.
As Table 2 makes clear, cinnamate added at its
maximal concentration (10 mM) that did not yet sup-
press cell growth caused appreciable induction of Crd
synthesis. The effect was observed under both aero-
bic and anaerobic conditions but was maximal in the
former case. Parallel measurements indicated a more
than ten-fold increase in the gene crdB transcription
by cinnamate, providing further support for assigning
the MV:cinnamate reductase activity to Crd in V.ruber
cells.
The other α,β-unsaturated carbonic acids acting
as Crd substrates, except for acrylate, could induce Crd
synthesis under anaerobic conditions (Fig. 4). These
compounds were added at 2 mM concentrations be-
cause the cells did not grow anaerobically at higher
concentrations of acrylate and caffeate. The effects of
the carbonic acids on Crd induction thus correlated
with their ability to act as substrates (Table1).
CINNAMATE REDUCTASE OF V. ruber 247
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
Table 1. The reductase activity of fCrd against a series
of α,β-unsaturated carbonic acids measured under
anaerobic conditions with 100µM reduced MV as the
electron donor
Substrate
(acid)
Structure k
cat
(s
–1
) K
m
(µM)
Caffeic
47 ± 2 2.2 ± 0.4
p-coumaric
45 ± 3 2.4 ± 0.5
Ferulic
26 ± 2 2.8 ± 0.5
Cinnamic
22 ± 1.5 7.1 ± 0.6
Acrylic
1.3 ± 0.2 770 ± 50
Fumaric 0.20 ± 0.03
a
Crotonic 0.06 ± 0.02
a
Urocanic 0.04 ± 0.01
a
Methacrylic 0.03 ± 0.01
a
a
Value of activity measured at 1mM substrate concentration.
Time-dependent activation of isolated Crd by
NADH. The presence of the NADH:flavin domains
in CdrAB (Fig. 1a) suggested that NADH is the natu-
ral donor of the redox equivalents for this enzyme.
As ferulate, caffeate, and p-coumarate absorb light at
340 nm, the NADH dehydrogenase activity was tested
with non-absorbing cinnamate as the electron accep-
tor. As the red trace in Fig.5a highlights, the reaction
initiated by cinnamate addition under anaerobic con-
ditions demonstrated a linear NADH oxidation curve
Fig. 3. HPLC identification of the product of cinnamate reduc-
tion by fCrd. Cinnamate (0.1mM) was reacted with 10mM so-
dium dithionite and 50µM MV in the presence of 0.5µM fCrd.
a)The reaction mixture before fCrd addition. b)Same after a
30-min incubation with fCrd. The elution was monitored at
210 (blue curve) and 258nm (red curve). The elution volumes
for authentic cinnamate and phenylpropionate are indicat-
ed by arrows. c) Electronic absorption spectra of cinnamate
and phenylpropionate in 10mM potassium phosphate buffer
(pH6.5).
Table 2. Correlation between the MV:cinnamate reductase activity and crdB transcription level in the V. ruber
cells cultivated in different conditions
Growth conditions Crd activity (nmol·min
–1
·mg
–1
) crdB mRNA/rRNA×10
–6
No cinnamate, no O
2
1.2 ± 0.4 2.1 ± 0.2
No cinnamate, +O
2
0.5 ± 0.2 0.8 ± 0.1
10 mM cinnamate, no O
2
86 ± 11 31 ± 4
10 mM cinnamate, +O
2
19 ± 3 25 ± 2
BERTSOVA et al.248
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
and a relatively high rate of 15 s
–1
. However, changing
the order of the additions affected the reaction time
course dramatically. For the reaction started by fCrd,
the progress curve demonstrated a profound lag in the
NADH oxidation (blue trace in Fig. 5a). Although the
initial slope was quite low, the final slope was similar
to that for the red trace described above and likely
corresponded to fully activated Crd. A similar time lag
was observed in the reaction started by NADH.
These findings indicate that fCrd is inactive as iso-
lated and requires preincubation with NADH for acti-
vation. That the length of the time lag of anaerobic cin-
namate reduction started by fCrd did not depend on
fCrd concentration (data not shown) indicated that the
activation is not associated with changes in the oligo-
meric structure. Furthermore, the red trace in Fig. 5a
demonstrates that the activation is much faster in the
absence of cinnamate than in its presence (blue curve).
The duration of Crd preincubation with NADH for the
red curve in Fig.5a was approximately 5min and this
time was thus sufficient for nearly complete activation
in the absence of the electron acceptor. Preincubation
with 5 mM dithionite for 5 min before starting the re-
action by the enzyme activated fCrd to the same level
(data not shown). These findings suggested that the ac-
tivation is likely caused by the reduction of a prosthetic
group in fCrd rather than by mere NADH binding to it.
For further analysis, the product formation curve
for the reaction started by Crd in Fig.5a (blue curve)
was converted into the first derivative (i.e., specific
activity) plot (Fig.5b), and its sigmoidal shape clearly
indicated that the activation process is not a one-step
reaction. Fitting experiments demonstrated a poor fit
of the kinetic scheme involving two consecutive steps
(red line) but a reasonably good fit for a scheme in-
volving three such steps (green line). In these fittings,
we assumed that the rate constants for each step are
identical because this relationship corresponds to
the largest possible time lag of the final product for-
mation. The fitted value of the rate constant was
0.060 ± 0.003 s
–1
for the three-step reaction. Three is
thus the minimal number of the steps involved in slow
Crd activation by NADH.
A similar but faster activation of Crd was ob-
served in the anaerobic acrylate reduction started by
the enzyme (Fig.5c). The maximal activity measured
with this substrate (3.5 s
–1
) was 23% of that with cin-
namate. For comparison, this activity was only 1.3 s
–1
with MV as the electron donor (6% of the activity with
cinnamate, Table1). A similar kinetic analysis indicated
that Crd activation by NADH in the presence of acry-
late involves at least two steps (Fig.S2 in the Online Re-
sourse 1), both with the rate constant of 0.27 ± 0.02 s
–1
.
We also demonstrated that Crd activation by NADH in
the absence of electron acceptors (Fig. S2, left panel,
squares, see the Online Resourse 1) was faster than
in their presence (Fig. S2, left panel, circles, see the
Online Resourse1; Fig.5b, circles). Asimilarly kinetic
analysis has revealed that the activation also involved
two consecutive steps with the average rate constant
of 0.75 ± 0.15 s
–1
.
If MV was used as the electron donor with various
substrates (Table 1), the progress curves were always
linear irrespective of the order of fCrd and substrate
additions. This result suggested that the activation/
deactivation of fCrd is somehow associated with its
NADH-binding NADH:flavin domains. This hypothesis
was supported by the observation that the anaerobic
NADH:FMN reduction by fCrd also demonstrated slow
activation if the reaction was started by the enzyme
(Fig.5d). The activation kinetics was more complex in
this case and demonstrated a poor fit of the consecu-
tive three-step model (Fig.S2, right panel in the Online
Resourse1).
Crd activated by preincubation with NADH re-
mained active until all NADH was consumed in the
enzymatic reaction, but slowly returned to the inac-
tive state afterwards. In the experiment illustrated
in Fig. 6a, the reaction was started by adding acry-
late to preactivated fCrd and proceeded linearly until
NADH exhaustion. A new portion of NADH restored
Crd activity to the level that depended on the time
interval t before the second NADH addition (Fig. 6b).
Thetime-courses became nonlinear at hight, indicat-
ing activation as in Fig.5c (blue trace) to a final rate
Fig. 4. MV:cinnamate reductase activity of the V. ruber cells
grown anaerobically in the presence of 2mM caffeate (caf),
p-coumarate (cou), ferulate (fer), cinnamate (cin), or acry-
late (acr). Error bars represent the standard deviations for
two cultivations.
CINNAMATE REDUCTASE OF V. ruber 249
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
similar to those in Fig.5c (red trace) and 6a (left trace).
These findings indicate that active fCrd is gradually de-
activated when depleted of NADH but time-dependent-
ly restores full activity after NADH addition. fCrd deac-
tivation was a first-order reaction with a half-time of
4.7min (Fig.6b).
Under aerobic conditions, the NADH-oxidizing ac-
tivity of Crd was extremely low and did not increase
with time (Fig.7, black (1) trace). This means that ox-
ygen prevents Crd conversion from inactive to active
state. This observation contrasts the data reported for
other NADH:2-enoate and NADH:flavin reductases,
which readily oxidize NADH with O
2
as the main elec-
tron acceptor [3, 4, 27]. To determine oxygen effect on
active Crd, the enzyme was preincubated with NADH
and allowed to convert excess cinnamate under anaer-
obic conditions before oxygen was added in the form
of H
2
O
2
(Fig.7, red (2) trace). In this experiment, lower
concentration of glucose oxidase and higher concen-
tration of catalase were used to maintain oxygen level
Fig. 5. The typical traces of fCrd-catalyzed NADH-dependent reactions under anaerobic conditions. a)Cinnamate-supported
NADH oxidation. The final reaction mixture contained 120 µM NADH, 50 µM cinnamate, and 30 nM fCrd. The reaction was start-
ed by either cinnamate (cin, red trace) or fCrd (blue trace). b)The time dependence of Crd activity derived from the time-course
of cinnamate reduction in the reaction started by Crd addition (blue curve on panela). Small circles show the activity values
determined as the slope of the tangent (-d[NADH]/dt) using Matlab (The Mathworks, Inc.) at 100 time points. The red and green
lines show the best fits for two- and three-step kinetic models, respectively. c)Acrylate-supported NADH oxidation. The final
reaction medium contained 120 µM NADH, 1 mM acrylate, and 150 nM fCrd. The reaction was started by acrylate (acr, red trace)
or fCrd (blue trace). d)NADH-supported FMN reduction. The final reaction medium contained 120 µM NADH and 50 μM FMN.
ThefCrd concentration was 15 nM (red trace, reaction started by FMN) or 60 nM (blue trace, reaction started by fCrd).
BERTSOVA et al.250
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
in the incubation mix. As Fig.7 highlights, the oxygen
pulse caused slow inactivation of Crd, followed by slow
reactivation to the initial level upon oxygen depletion
by glucose oxidase. The duration of the transient inacti-
vation phase decreased with increasing glucose oxidase
concentration (data not shown). These findings indicat-
ed that the reaction of molecular oxygen with Crd, re-
sponsible for its inactivation, is easily reversible, allow-
ing regulation of Crd activity by medium oxygen level.
Interestingly, NADPH demonstrated only low effi-
ciency as an electron donor compared with NADH. In
the corresponding experiment, fCrd preactivated by
incubation with 30µM NADH was allowed to convert
acrylate until all NADH was consumed and immediately
supplemented with 120µM NADPH. The measured rate
of NADPH oxidation was 35 times lower than the rate
of NADH oxidation. This makes NADH the most likely
physiological donor of the reducing equivalents for Crd.
Domain roles in the catalytic activities of Crd.
Crd is thus capable of catalyzing redox reactions be-
tween different natural and artificial donors and ac-
ceptors of electrons. To determine which Crd domains
are responsible for different reactions, we measured
the activities of the apo- and flavinylated forms of Crd
and its deletion variants. The MV:cinnamate reductase
activity was high with fCrd and much lower with aCrd
(no covalently bound FMN) and d3 (only FAD binding2
domain) (Table 3). These data indicated that, like in
Klebsiella pneumoniae NADH:fumarate reductase [3],
the FAD binding2 domain is required for reduction of
natural electron acceptor and the electron pathway
from MV to this domain involves the covalently bound
FMN of the FMN bind domain.
In contrast, high NADH:FMN reductase activity
was demonstrated not only by the full-size flavinylat-
ed enzyme (fCrd) but also by its apo-form and the d1
variant (only NADH:flavin domains). This reaction is
thus clearly catalyzed by the NADH:flavin domains.
Noteworthy, all the NADH:FMN reductase activities, in-
cluding that manifested by the d1 variant, developed
slowly in time when the reaction was started by NADH
but instantly when FMN was the last added component,
as in Fig.5d.
Full NADH:cinnamate reductase activity was man-
ifested by fCrd but not aCrd that lacks covalently
bound FMN. These findings emphasized the role of the
covalently bound FMN in electron transfer between
the NADH dehydrogenase part(d1) and the cinnamate
reductase part(d3) in Crd.
Fig. 6. Deactivation of Crd. a)Anaerobic NADH oxidation by
fCrd in the presence of 1 mM acrylate(acr). When the reac-
tion ended because of NADH consumption, its concentra-
tion was restored to the initial level (~50 µM) after a 1.5-min
pause. Conditions were as for Fig.5c. b)The initial activity of
NADH conversion measured for the second portion of NADH
as a function of the time interval t before its re-addition.
100% refers to the initial activity estimated from the left
curve in panel(a). The theoretical curve is for the first-order
reaction with a half-time of 4.7min.
Fig. 7. Transient inactivation of Crd by oxygen pulse during
NADH-supported cinnamate reduction. Trace 1) The reaction
started by fCrd and carried out under aerobic conditions
(no glucose oxidase and catalase added). The experimental
conditions were otherwise as for the red curve in Fig. 5a.
Trace 2) The reaction was started by 500 µM cinnamate and
carried out under anaerobic conditions until 500 μM H
2
O
2
(producing 250 µM molecular oxygen), were added at the
indicated time. The concentrations of glucose oxidase and
catalase in this particular experiment were 0.6 and 50U/ml,
respectively. The final reaction mixture contained 120 µM
NADH and 30nM fCrd in both experiments.
CINNAMATE REDUCTASE OF V. ruber 251
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
The separate FAD binding 2 domain (d3 variant)
demonstrated measurable activity (0.60 ± 0.05 s
–1
, data
not shown) in FMNH
2
:cinnamate reduction, raising the
possibility of catalyzing complete NADH:cinnamate
reductase reaction by a 1 : 1 mixture of d1 and d3 in
the presence of medium FMN. Such activity was in-
deed detected, and although its value, 3.2 ± 0.4 s
–1
(data
not shown), was less than for fCrd, it significantly ex-
ceeded the FMNH
2
:cinnamate reductase activity of d3.
This seemingly unexpected result could be, however,
explained by different conditions used for the two re-
actions. The assay medium for the FMNH
2
:cinnamate
reductase activity contained, beside reduced FMN, con-
siderable amounts of its oxidized form, which could in-
hibit the reaction by increasing the redox potential of
the FMNH
2
free
/FMN
free
pair. In contrast, virtually all FMN
was in the reduced form under steady-state conditions
in the NADH:cinnamate reductase reaction catalyzed
by d1+d3. By demonstrating full NADH:cinnamate re-
ductase activity in the presence of medium FMN, the
d1 and d3 mixture resembles yeast soluble fumarate
reductase [28]. In the latter enzyme, the roles of d1 and
d3 are played by separate enzymes, with free cytoplas-
mic flavins acting as electron carriers between them.
The modelled three-dimensional structure of
Crd. The 3D structure of heterodimeric Crd was pre-
dicted using AlphaFold2 [21]. It can be divided into
two parts connected by a flexible linker (Fig. 8). One of
them, corresponding to the d1 fragment, is formed
by CrdA and the homologous N-terminal NADH:flavin
domain of CrdB and resembles homodimeric NADH:
flavin reductases [16, 29]. The other part contains two
other CrdB domainsFAD binding2 and FMN bind.
Four flavin groups (three FMNs and one FAD in
their oxidized states) and one each of NADH and cin-
namate molecules were docked into the modeled struc-
ture using AutoDock Vina [23]. The locations of the
docked flavin groups (Fig.8) were similar to those in
the experimentally determined structures of NAD(P)H:
FMN reductase (PDBID: 1X77 [29]), the NqrC subunit
of Na
+
-translocating NADH:quinone oxidoreductase
(PDBID: 4XA7 [30]), and urocanate reductase (PDB ID:
6T87 [31]). These three proteins are homologous to the
NADH:flavin, FMN bind, and FAD binding2 domains of
Crd, respectively.
The d1 part of Crd contains two potential pseudo-
symmetrical NADH-binding sites near FMN
A
and FMN
B
,
similar to those found in the homodimeric NADH:fla-
vin reductase [16]. However, only the FMN
B
-adjacent
site could be occupied by NADH because the entrance
to the FMN
A
-adjacent site is blocked by the C-terminal
segment of CrdA (Fig. S3 in the Online Resourse 1).
Instead, the positions of the docked NADH molecule
concentrated in two areas of the protein near FMN
A
.
The shortest distance between NADH and FMN
A
, 10 Å,
is too long for the hydride ion transfer between them
(for comparison, the distance between NADH and
FMN
B
is only 4 Å). Similar docking experiments pro-
vided no evidence for a competent NADH-binding site
in the middle, FMN bind domain of CrdB. The location
of docked cinnamate in the FAD binding 2 domain of
CrdB was similar to that of urocanate in urocanate re-
ductase [31].
The edge-to-edge distance between FAD and *FMN
in the resulting model is approximately 9 Å (Fig. 8),
allowing rapid electron transfer between these fla-
vin groups, consistent with the finding that this step
is not rate-limiting in NADH:2-enoate reductases [3].
In contrast, the FMN
B
and *FMN groups are separated
by 24 Å in the model (Fig. 8), a distance at which rap-
id electron transfer is not feasible. Clearly, Crd should
adopt a different conformation to make the FMN
B
and
*FMN groups closer to each other and allow electron
transfer between them in the catalytic reaction. Inter-
estingly, the same step is rate-limiting in the electron
transfer in canonical NADH:2-enoate reductases [3],
Table 3. The catalytic activities of Crd and its deletion variants
a
Enzyme/variant Bound prosthetic groups
Activity, s
–1
MV:cinnamate NADH:cinnamate NADH:FMN
fCrd 2 FMN, 1 FAD, 1 *FMN
b
22 ± 2 15 ± 2 62 ± 15
aCrd 2 FMN, 1 FAD 0.7 ± 0.1 <0.05 58 ± 13
d1 2 FMN n.d.
c
<0.05 48 ± 21
d3 1 FAD 0.7 ± 0.1 <0.05 n.d.
a
The assay mixture contained 120µM NADH or 100 µM reduced MV as the electron donor and 50 μM cinnamate or FMN
as the electron acceptor. The activity values shown are for the reactions started by electron acceptor (the reactions started by
the enzyme demonstrated a profound time lag).
b
The asterisk indicates covalently bound flavin.
c
n.d., not determined.
BERTSOVA et al.252
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
indicating that the above distance is similarly non-op-
timal in these enzymes.
The validity of the model predicted by AlphaFold2
with the docked flavins was confirmed by MD simu-
lations (Fig. S4 in the Online Resourse 1). Specifically,
none of the four flavin groups was lost during the sim-
ulations. Furthermore, the simulations did not change
significantly the structure predicted by AlphaFold2,
except for a rigid-body movement of the d1 part, con-
nected by a flexible linker, by approximately 11 Å rel-
ative the rest of the molecule (Fig. S4, A and C in the
Online Resourse 1). This result showed that Crd struc-
ture is indeed flexible. However, the flavin-to-flavin
distances were not appreciably different in the sim-
ulated structure (Fig. S4B in the Online Resourse 1)
and remained virtually unchanged in the analogous
structure with the flavins in their reduced forms (data
notshown).
DISCUSSION
Three-domain NADH:2-enoate reductases are abun-
dant among anaerobic and facultatively anaerobic
prokaryotic and eukaryotic microorganisms, which
suggests their metabolic importance. The enzymes
with NADH:fumarate reductase activity participate in
anaerobic respiration, allowing reoxidation of glycoly-
sis-produced NADH to increase ATP production during
fermentation. This role was best demonstrated in
unicellular eukaryotes of the class Kinetoplastida, in
which fermentation yields succinate as a major prod-
uct and glycosomal and mitochondrial NADH:fuma-
rate reductase inactivation blocks completely fuma-
rate respiration [32]. NADH:2-enoate reductases with
different substrate specificities may have other roles.
Thus, NADH:acrylate reductase from marine bacteri-
um V. harveyi is apparently involved in detoxication
of acrylate [4], massively formed in marine habitats
during degradation of dimethylsulfoniopropionate,
the main compatible solute in alga.
While sharing overall structural similarity with
other microbial three-domain NADH:2-enoate reduc-
tases, such as the fumarate reductases of K. pneumoni-
ae and Kinetoplastida and acrylate reductase of V. har-
veyi, the V. ruber cinnamate reductase Crd displays
significant differences. In all these enzymes, 2-enoate
is reduced in FAD binding 2 domain by the electrons
Fig. 8. The predicted three-dimensional structure of Crd with bound FAD and three FMNs in a cartoon representation.
The domains are colored as in Fig.1a. The right part shows schematically the edge-to-edge distances between the prosthetic
groups and the proposed pathways for electron and hydride ion transfer (indicated by brown and green arrows, respectively).
CINNAMATE REDUCTASE OF V. ruber 253
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
transmitted via a covalently bound FMN residue of
FMN bind domain [3]. However, the NADH-oxidizing
activity belongs to NADH:flavin domain in Crd but to
either OYE-like domain (in K. pneumoniae NADH:fuma-
rate and V. harveyi NADH:acrylate reductases) or FAD
binding6 domain (in Kinetoplastida NADH:fumarate re-
ductases).
Furthermore, only Crd contains an additional sub-
unit (CrdA) formed by a single NADH:flavin domain,
which is homologous to the analogous domain of CrdB
(51% identity, 65% similarity). Its gene (crdA) flanks
the crdB gene in one chromosome (Fig.1c), which, to-
gether with the finding that the tag-less CrdA co-puri-
fies with 6×His-tagged CrdB on a Ni-column, provides
strong evidence that Crd functions as a tight CrdA/CrdB
complex in vivo. The observed noncovalently bound
FMN:FAD stoichiometry of 2 : 1 is fully consistent with
the heterodimeric structure of Crd. Noteworthy, pro-
karyotic NADH:flavin reductases are generally encoded
by single genes but function as homodimers contain-
ing two bound FMN molecules at the dimer interface
[16, 29]. The homodimeric structure apparently pre-
vents direct fusion of the NADH:flavin reductase do-
main with other functional domains to form an ex-
tended multidomain polypeptide.
Under aerobic conditions, three-domain NADH:
2-enoate reductases catalyze O
2
reduction to form re-
active oxygen species (ROS). As a striking example,
K. pneumoniae NADH:fumarate reductase oxidizes NADH
to form H
2
O
2
under aerobic conditions, and the oxida-
tion reaction is not accelerated by fumarate, indicat-
ing that all NADH reducing equivalents are consumed
in H
2
O
2
production [3]. On the other hand, ROS pro-
duction may be beneficial in some cases. Thus, ROS
generation by NADH:fumarate reductases in Kineto-
plastida may aid in completing the parasitic cycle (dif-
ferentiation of procyclics into epimastigotes) in the
insect vector [33]. However, excessive ROS generation
upon transfer to aerobiosis is expected to be harmful
for most facultatively anaerobic microorganisms. For
instance, accumulation of H
2
O
2
in millimolar concen-
trations under the action of NADH:flavin reductase in
aerobic conditions caused cell death in the lactic acid
bacterium Lactobacillus johnsonii [27].
The presence of a pair of NADH:flavin domains in
V. ruber Crd instead of the OYE-like or FAD binding 6
domain found in other NADH:2-enoate reductases pre-
sumably solves the problem of excessive ROS produc-
tion. The NADH:flavin domains allow a unique regula-
tion mechanism that reversibly blocks Crd activity at
high O
2
concentrations (Fig.7). The transition between
the active and inactive states appears to be controlled
in Crd by the redox state of one or both FMN mole-
cules in the NADH:flavin domains. This conclusion is
supported by the observations that Crd could be acti-
vated by both NADH and dithionite and that the d1 but
not d3 fragment demonstrated similar time-dependent
activation of its partial enzymatic activity.
Crd regulation appears to be associated with the
heterodimeric structure of the NADH dehydrogenase
part, formed by CrdA and the homologous N-terminal
domain of CrdB (Fig. 8). This part of Crd resembles
NADH:flavin reductases [16, 29] and similarly contains
two noncovalently bound FMN groups and two poten-
tial sites of NADH oxidation. However, the competence
of the putative NADH-binding site near FMN
A
seems
questionable because of the unfavorable steric effect
of the C-terminal loop of CrdA. While electron trans-
fer between FMN
B
and *FMN is perceivable assuming
*FMN and/or NADH:flavin dimer flexibility, the long
distance between FMN
B
and FMN
A
(Fig. 8) can hardly
decrease similarly. Hence, the electron transfer between
these flavin groups should be always much slower than
enzyme turnover [34]. These structural considerations
suggest that FMN
B
and FMN
A
are good candidates
for the catalytic and regulating groups, respectively.
Thatthese flavins have distinct roles is consistent with
significant differences in the amino acid sequenc-
es of the two NADH:flavin domains of Crd, contrast-
ing the case of homodimeric and, hence, symmetrical
NADH:flavin reductases in which two FMN sites are
identical [16]. One can further speculate that Crd ac-
tivation involves the stepwise reduction of FMN
B
and
FMN
A
causing a conformational change that forms a
functional electron transfer chain linking the donor
and acceptor.
That the electron acceptors slow down Crd activa-
tion by NADH raises the possibility that the bound ac-
ceptor competes with the regulating prosthetic group,
likely FMN
A
, for the electrons coming from NADH.
In this case, the activation rate should inversely cor-
relate with the rate of acceptor reduction. Such a cor-
relation was indeed observed– cinnamate was reduced
faster (Table 1) and decelerated Crd activation more
compared with acrylate (Fig. 5). In this context, the
lack of Crd activation under aerobic conditions may
mean that O
2
can oxidize the regulating prosthetic
group directly, without using the long electron transfer
chain involved in the case of cinnamate and acrylate.
Despite the similarity of the 2-enoate-binding ami-
no acid residues in the FAD binding 2 domains of ac-
rylate reductase and Crd (Fig. 1b), the latter enzyme
exhibits a 1000-6000-fold lower catalytic efficiency
against acrylate than against cinnamate and its deriva-
tives (Table1). Furthermore, (hydroxy)cinnamic acids,
but not acrylate, induced Crd synthesis in V.ruber
cells (Fig. 4). Clearly, the positions of amino acid res-
idues interacting with electron-acceptor carboxylates
in fumarate reductases are not the only determinants
of the substrate specificity in other NADH:2-enoate re-
ductases. The reduction of cinnamate and its deriva-
tives, toxic to cells [35], is thus the most likely physio-
BERTSOVA et al.254
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
logical function of Crd, and this inference is supported
by their induction of Crd synthesis in both anaerobic
and aerobic conditions (Table2). Additionally, Crd may
support anaerobic respiration of V.ruber on (hydroxy)
cinnamic acids as described for caffeyl-CoA reductase
from acetogenic bacterium Acetobacterium woodii [36].
The most effective substrates of Crd, caffeate and feru-
late, are the precursors of lignin and various pheno-
lic secondary metabolites in plants. That is why these
compounds are abundant in terrestrial niches, soil,
and animal gut in particular, but are relatively rare
in the marine habitats, where they are synthesized in
some algae [37, 38]. An interesting exception to this
rule is the seagrass meadows, which accumulate caffe-
ic acid and other phenolic compounds in rhizosphere
[35], a likely main habitat for Crd-containing marine
bacteria [39].
Crd-like reductases are relatively rare among ma-
rine microorganisms, seemingly, because (hydroxyl)
cinnamic acids are underrepresented in the habitat.
BLAST search has revealed Crd-like proteins only in
the following marine bacteria: Vibrio rhizosphaerae,
Vibrio gazogenes, Vibrio salinus, Vibrio spartinae, and
Vibrio tritonius. However, such reductases are abun-
dant among anaerobic and facultatively anaerobic
terrestrial bacteria (the Clostridium, Klebsiella, Citro-
bacter, Aeromonas, Paenibacillus, Streptococcus, and
other genera), including those forming human intesti-
nal microbiome. The presence of substantial amounts
of phenylpropionate and p-hydroxyphenylpropionate
in the mammalian intestine has long been known [40]
but their production was attributed to a different en-
zyme, bacterial two-domain 2-enoate reductase [1].
The wide distribution of Crd-like enzymes among in-
testinal bacteria suggests that they are also involved
in phenylpropionate and p-hydroxyphenylpropionate
formation in the mammalian intestine.
Two-domain 2-enoate reductases are industrial
biocatalysts increasingly used to produce chiral inter-
mediates, pharmaceuticals and agrochemicals [11, 41].
A disadvantage of these enzymes is that they are read-
ily and irreversibly inactivated by oxygen [1, 11]. The
lower sensitivity of Crd to oxygen makes it a promising
alternative for industrial application.
Contributions. AVB the conception of the study;
YVB, MVS, VAA, AAB, and AVB the acquisition and
analysis of the data; AAB and AVB writing of the man-
uscript.
Acknowledgments. MALDI MS and laser scanner
facilities became available to us in the framework of
the Moscow State University Development Program
PNG5.13. This article is devoted to cherished memory
of VladimirP. Skulachev.
Funding. This work was supported by the Russian
Science Foundation (project no.22-24-00133).
Ethics declarations. The authors declare no con-
flicts of interest. This article does not contain descrip-
tion of studies with the involvement of humans or an-
imal subjects.
Electronic supplementary material. The online
version contains supplementary material available at
https://doi.org/10.1134/S0006297924020056.
REFERENCES
1. Tischer, W., Bader, J., and Simon, H. (1979) Purifica-
tion and some properties of a hitherto-unknown en-
zyme reducing the carbon-carbon double bond of
α,β-unsaturated carboxylate anions, Eur. J. Biochem.,
97, 103-112, doi:10.1111/j.1432-1033.
2. Besteiro,S., Biran,M., Biteau,N., Coustou,V., Baltz,T.,
Canioni, P., and Bringaud, F. (2002) Succinate secret-
ed by Trypanosoma brucei is produced by a novel
and unique glycosomal enzyme, NADH-dependent
fumarate reductase, J. Biol. Chem., 277, 38001-38012,
doi:10.1074/jbc.M201759200.
3. Bertsova, Y. V., Oleynikov, I. P., and Bogachev, A. V.
(2020) A new water-soluble bacterial NADH: fumarate
oxidoreductase, FEMS Microbiol. Lett., 367, fnaa175,
doi:10.1093/femsle/fnaa175.
4. Bertsova, Y.V., Serebryakova, M.V., Baykov, A.A., and
Bogachev, A.V. (2022) A novel, NADH-dependent acry-
late reductase in Vibrio harveyi, Appl. Environ. Micro-
biol., 88, e0051922, doi:10.1128/aem.00519-22.
5. Bertsova, Y.V., Kostyrko, V.A., Baykov, A.A., and Bo-
gachev, A. V. (2014) Localization-controlled specific-
ity of FAD:threonine flavin transferases in Klebsiella
pneumoniae and its implications for the mechanism
of Na
+
-translocating NADH:quinone oxidoreductase,
Biochim. Biophys. Acta, 1837, 1122-1129, doi:10.1016/
j.bbabio.2013.12.006.
6. Serebryakova, M. V., Bertsova, Y. V., Sokolov, S. S.,
Kolesnikov, A. A., Baykov, A. A., and Bogachev, A. V.
(2018) Catalytically important flavin linked through a
phosphoester bond in a eukaryotic fumarate reductase,
Biochimie, 149, 34-40, doi:10.1016/j.biochi.2018.03.013.
7. Rohdich,F., Wiese,A., Feicht,R., Simon,H., and Bach-
er, A. (2001) Enoate reductases of Clostridia. Clon-
ing, sequencing, and expression, J. Biol. Chem., 276,
5779-5787, doi:10.1074/jbc.M008656200.
8. Kuno,S., Bacher,A., and Simon,H. (1985) Structure of
enoate reductase from a Clostridium tyrobutyricum
(C.spec. La1), Biol. Chem. Hoppe Seyler, 366, 463-472,
doi:10.1515/bchm3.1985.366.1.463.
9. Caldeira,J., Feicht,R., White,H., Teixeira,M., Moura,
J.J., Simon,H., and Moura,I. (1996) EPR and Mössbau-
er spectroscopic studies on enoate reductase, J. Biol.
Chem., 271, 18743-18748, doi:10.1074/jbc.271.31.18743.
10. Giesel,H., and Simon,H. (1983) On the occurrence of
enoate reductase and 2-oxo-carboxylate reductase in
CINNAMATE REDUCTASE OF V. ruber 255
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
clostridia and some observations on the amino acid
fermentation by Peptostreptococcus anaerobius, Arch.
Microbiol., 135, 51-57, doi:10.1007/BF00419482.
11. Mordaka, P.M., Hall, S.J., Minton,N., and Stephens,G.
(2018) Recombinant expression and characterisa-
tion of the oxygen-sensitive 2-enoate reductase from
Clostridium sporogenes, Microbiology (Reading), 164,
122-132, doi:10.1099/mic.0.000568.
12. Shieh, W.Y., Chen, Y.W., Chaw, S.M., and Chiu, H.H.
(2003) Vibrio ruber sp. nov., a red, facultatively an-
aerobic, marine bacterium isolated from sea water,
Int. J. Syst. Evol. Microbiol., 53, 479-484, doi: 10.1099/
ijs.0.02307-0.
13. Bogachev, A.V., Bertsova, Y.V., Bloch, D.A., and Verk-
hovsky, M. I. (2012) Urocanate reductase: Identifi-
cation of a novel anaerobic respiratory pathway
in Shewanella oneidensis MR-1, Mol. Microbiol., 86,
1452-1463, doi:10.1111/mmi.12067.
14. Light, S.H., Méheust,R., Ferrell, J.L., Cho,J., Deng,D.,
Agostoni,M., Iavarone, A. T., Banfield, J.F., D’Orazio,
S.E.F., and Portnoy, D.A. (2019) Extracellular electron
transfer powers flavinylated extracellular reductases
in Gram-positive bacteria, Proc. Natl. Acad. Sci. USA,
116, 26892-26899, doi:10.1073/pnas.1915678116.
15. Bogachev, A. V., Baykov, A. A., and Bertsova, Y. V.
(2018) Flavin transferase: the maturation factor of fla-
vin-containing oxidoreductases, Biochem. Soc. Trans.,
46, 1161-1169, doi:10.1042/BST20180524.
16. Koike,H., Sasaki,H., Kobori,T., Zenno,S., Saigo,K., Mur-
phy, M.E., Adman, E.T., and Tanokura,M. (1998) 1.8 Å
crystal structure of the major NAD(P)H:FMN oxidore-
ductase of a bioluminescent bacterium, Vibrio fischeri:
overall structure, cofactor and substrate-analog bind-
ing, and comparison with related flavoproteins, J.Mol.
Biol., 280, 259-273, doi:10.1006/jmbi.1998.1871.
17. Bertsova, Y. V., Kulik, L. V., Mamedov, M. D., Baykov,
A.A., and Bogachev, A. V. (2019) Flavodoxin with an
air-stable flavin semiquinone in a green sulfur bac-
terium, Photosynth. Res., 142, 127-136, doi: 10.1007/
s11120-019-00658-1.
18. Bertsova, Y. V., Fadeeva, M. S., Kostyrko, V. A., Sere-
bryakova, M. V., Baykov, A. A., and Bogachev, A. V.
(2013) Alternative pyrimidine biosynthesis protein
ApbE is a flavin transferase catalyzing covalent attach-
ment of FMN to a threonine residue in bacterial flavo-
proteins, J.Biol. Chem., 288, 14276-14286, doi:10.1074/
jbc.M113.455402.
19. Laemmli, U.K. (1970) Cleavage of structural proteins
during the assembly of the head of bacteriophageT4,
Nature, 227, 680-685, doi:10.1038/227680a0.
20. Ells, A.H. (1959) A colorimetric method for the assay
of soluble succinic dehydogenase and pyridinenucle-
otide-linked dehydrogenases, Arch. Biochem. Biophys.,
85, 561-562, doi:10.1016/0003-9861(59)90527-2.
21. Jumper,J., Evans,R., Pritzel,A., Green,T., Figurnov,M.,
Ronneberger, O., Tunyasuvunakool, K., Bates, R.,
Žídek,A., Potapenko,A., Bridgland,A., Meyer,C., Kohl,
S. A. A., Ballard, A. J., Cowie, A., Romera-Paredes, B.,
Nikolov,S., Jain,R., Adler,J., Back,T., Petersen,S., Rei-
man,D., Clancy,E., Zielinski,M., Steinegger,M., Pachol-
ska, M., Berghammer, T., Bodenstein, S., Silver, D.,
Vinyals, O., Senior, A. W., Kavukcuoglu, K., Kohli, P.,
and Hassabis,D. (2021) Highly accurate protein struc-
ture prediction with AlphaFold, Nature, 596
, 583-589,
doi:10.1038/s41586-021-03819-2.
22. Mirdita,M., Schütze,K., Moriwaki,Y., Heo,L., Ovchin-
nikov, S., and Steinegger, M. (2022) ColabFold: mak-
ing protein folding accessible to all, Nat. Meth., 19,
679-682, doi:10.1038/s41592-022-01488-1.
23. Trott, O., and Olson, A. J. (2010) AutoDock Vina: im-
proving the speed and accuracy of docking with a
new scoring function, efficient optimization, and mul-
tithreading, J.Comput. Chem., 31, 455-461, doi:10.1002/
jcc.21334.
24. Case, D.A., Aktulga, H.M., Belfon, K., Ben-Shalom, I. Y.,
Berryman, J.T., Brozell, S.R., etal. (2022) Amber 2022,
University of California, San Francisco.
25. Bolanos-Garcia, V.M., and Davies, O.R. (2006) Structur-
al analysis and classification of native proteins from
E. coli commonly co-purified by immobilised metal
affinity chromatography, Biochim. Biophys. Acta, 1760,
1304-1313, doi:10.1016/j.bbagen.2006.03.027.
26. Bertsova, Y.V., Serebryakova, M.V., Baykov, A.A., and
Bogachev, A.V. (2021) The flavin transferase ApbE fla-
vinylates the ferredoxin:NAD
+
-oxidoreductase Rnf re-
quired for N
2
fixation in Azotobacter vinelandii, FEMS
Microbiol. Lett., 368, fnab130, doi: 10.1093/femsle/
fnab130.
27. Hertzberger, R., Arents, J., Dekker, H. L., Pridmore,
R.D., Gysler,C., Kleerebezem,M., and de Mattos, M.J.
(2014) H
2
O
2
production in species of the Lactobacillus
acidophilus group: a central role for a novel NADH-
dependent flavin reductase, Appl. Environ. Microbiol.,
80, 2229-2239, doi:10.1128/AEM.04272213.
28. Kim,S., Kim, C.M., Son, Y.J., Choi, J.Y., Siegenthaler,
R.K., Lee,Y., Jang, T.H., Song,J., Kang,H., Kaiser, C.A.,
and Park, H. H. (2018) Molecular basis of maintain-
ing an oxidizing environment under anaerobiosis by
soluble fumarate reductase, Nat. Commun., 9, 4867,
doi:10.1038/s41467-018-07285-9.
29. Agarwal, R., Bonanno, J. B., Burley, S. K., and Swa-
minathan, S. (2006) Structure determination of
an FMN reductase from Pseudomonas aeruginosa
PA01 using sulfur anomalous signal, Acta Crystal-
logr. D Biol. Crystallogr., 62, 383-391, doi: 10.1107/
S0907444906001600.
30. Borshchevskiy, V., Round, E., Bertsova, Y., Polo-
vinkin, V., Gushchin, I., Ishchenko, A., Kovalev, K.,
Mishin, A., Kachalova, G., Popov, A., Bogachev, A.,
and Gordeliy, V. (2015) Structural and functional in-
vestigation of flavin binding center of the NqrC sub-
unit of sodium-translocating NADH:quinone oxidore-
BERTSOVA et al.256
BIOCHEMISTRY (Moscow) Vol. 89 No. 2 2024
ductase from Vibrio harveyi, PLoS One, 10, e0118548,
doi:10.1371/journal.pone.0118548.
31. Venskutonytė,R., Koh,A., Stenström,O., Khan, M.T.,
Lundqvist,A., Akke,M., Bäckhed,F., and Lindkvist-Pe-
tersson, K. (2021) Structural characterization of the
microbial enzyme urocanate reductase mediating
imidazole propionate production, Nat. Commun., 12,
1347, doi:10.1038/s41467-021-21548-y.
32. Coustou, V., Besteiro, S., Rivière, L., Biran, M., Bite-
au,N., Franconi, J.M., Boshart,M., Baltz,T., and Brin-
gaud,F. (2005) A mitochondrial NADH-dependent fu-
marate reductase involved in the production of succi-
nate excreted by procyclic Trypanosoma brucei, J. Biol.
Chem., 280, 16559-16570, doi:10.1074/jbc.M500343200.
33. Wargnies, M., Plazolles, N., Schenk, R., Villafraz, O.,
Dupuy, J. W., Biran, M., Bachmaier, S., Baudouin, H.,
Clayton,C., Boshart,M., and Bringaud,F. (2021) Meta-
bolic selection of a homologous recombination-medi-
ated gene loss protects Trypanosoma brucei from ROS
production by glycosomal fumarate reductase, J.Biol.
Chem., 296, 100548, doi:10.1016/j.jbc.2021.100548.
34. Page, C. C., Moser, C. C., Chen, X., and Dutton, P. L.
(1999) Natural engineering principles of electron tun-
nelling in biological oxidation-reduction, Nature, 402,
47-52, doi:10.1038/46972.
35. Sogin, E. M., Michellod, D., Gruber-Vodicka, H. R.,
Bourceau,P., Geier,B., Meier, D.V., Seidel,M., Ahmer-
kamp, S., Schorn, S., D’Angelo, G., Procaccini, G., Du-
bilier, N., and Liebeke, M. (2022) Sugars dominate
the seagrass rhizosphere, Nat. Ecol. Evol., 6, 866-877,
doi:10.1038/s41559-022-01740-z.
36. Bertsch,J., Parthasarathy,A., Buckel,W., and Müller,V.
(2013) An electron-bifurcating caffeyl-CoA reduc-
tase, J.Biol. Chem., 288, 11304-11311, doi:10.1074/jbc.
M112.444919.
37. Martone, P. T., Estevez, J. M., Lu, F., Ruel, K., Denny,
M. W., Somerville, C., and Ralph, J. (2009) Discovery
of lignin in seaweed reveals convergent evolution
of cell-wall architecture, Curr. Biol., 19, 169-175,
doi:10.1016/j.cub.2008.12.031.
38. Fernando, I.P., Kim,M., Son, K.T., Jeong,Y., and Jeon,
Y.J. (2016) Antioxidant activity of marine algal poly-
phenolic compounds: a mechanistic approach, J.Med.
Food, 19, 615-628, doi:10.1089/jmf.2016.3706.
39. Rameshkumar,N., and Nair,S. (2009) Isolation and mo-
lecular characterization of genetically diverse antag-
onistic, diazotrophic red-pigmented vibrios from dif-
ferent mangrove rhizospheres, FEMS Microbiol. Ecol.,
67, 455-467, doi:10.1111/j.1574-6941.2008.00638.x.
40. Smith, E.A., and Macfarlane, G.T. (1996) Enumeration
of human colonic bacteria producing phenolic and
indolic compounds: effects of pH, carbohydrate avail-
ability and retention time on dissimilatory aromatic
amino acid metabolism, J.Appl. Bacteriol., 81, 288-302,
doi:10.1111/j.1365-2672.1996.tb04331.x.
41. Knaus, T., Toogood, H. S., and Scrutton, N. S. (2016)
Ene-Reductases and Their Applications. Green Bio-
catalysis, John Wiley and Sons, Inc., pp. 473-488,
doi:10.1002/9781118828083.ch18.
42. Sievers,F., and Higgins, D.G. (2018) Clustal Omega for
making accurate alignments of many protein sequenc-
es, Protein Sci., 27, 135-145, doi:10.1002/pro.3290.
Publishers Note. Pleiades Publishing remains
neutral with regard to jurisdictional claims in pub-
lished maps and institutional affiliations.