L-Xylo-3-hexulose, a new rare sugar produced by the action of acetic acid bacteria on galactitol, an exception to Bertrand Hudson's rule

Article abstract of DOI:10.1016/j.bbagen.2020.129740

Background: In acetic acid bacteria such as Gluconobacter oxydans or Gluconobacter cerinus, pyrroloquinoline quinone (PQQ) in the periplasm serves as the redox cofactor for several membrane-bound dehydrogenases that oxidize polyhydric alcohols to rare sugars, which can be used as a healthy alternative for traditional sugars and sweeteners. These oxidation reactions obey the generally accepted Bertrand Hudson's rule, in which only the polyhydric alcohols that possess cis D-erythro hydroxyl groups can be oxidized to 2-ketoses using PQQ as a cofactor, while the polyhydric alcohols excluding cis D-erythro hydroxyl groups ruled out oxidation by PQQ-dependent membrane-bound dehydrogenases. Methods: Membrane fractions of G. oxydans were prepared and used as a cell-free catalyst to oxidize galactitol, with or without PQQ as a cofactor. Results: In this study, we reported an interesting oxidation reaction that the polyhydric alcohols galactitol (dulcitol), which do not possess cis D-erythro hydroxyl groups, can be oxidized by PQQ-dependent membrane-bound dehydrogenase(s) of acetic acid bacteria at the C-3 and C-5 hydroxyl groups to produce rare sugars L-xylo-3-hexulose and D-tagatose. Conclusions: This reaction may represent an exception to Bertrand Hudson's rule. General significance: Bertrand Hudson's rule is a well-known theory in polyhydric alcohols oxidation by PQQ-dependent membrane-bound dehydrogenase in acetic acid bacteria. In this study, galactitol oxidation by a PQQ-dependent membrane-bound dehydrogenase represents an exception to the Bertrand Hudson's rule. Further identification of the associated enzymes and deciphering the explicit enzymatic mechanism will prove this theory.

Full text of DOI:10.1016/j.bbagen.2020.129740

BBA-GeneralSubjects1865(2021)129740
Contents lists available at ScienceDirect
BBA - General Subjects
journal homepage: www.elsevier.com/locate/bbagen
L-Xylo-3-hexulose, a new rare sugar produced by the action of acetic acid
bacteria on galactitol, an exception to Bertrand Hudson's rule
⁎
Yirong Xua, Ping Chia, Jiyang Lva, Muhammad Bilalb, Hairong Chenga,
a State Key Laboratory of Microbial Metabolism, School of Life Sciences and Biotechnology, Shanghai Jiao Tong University, Shanghai, China
b School of Life Science and Food Engineering, Huaiyin Institute of Technology, Jiangsu, China
A R T I C L E I N F O
A B S T R A C T
Keywords:
Background: In acetic acid bacteria such as Gluconobacter oxydans or Gluconobacter cerinus, pyrroloquinoline
quinone (PQQ) in the periplasm serves as the redox cofactor for several membrane-bound dehydrogenases that
oxidize polyhydric alcohols to rare sugars, which can be used as a healthy alternative for traditional sugars and
sweeteners. These oxidation reactions obey the generally accepted Bertrand Hudson's rule, in which only the
polyhydric alcohols that possess cis D-erythro hydroxyl groups can be oxidized to 2-ketoses using PQQ as a
cofactor, while the polyhydric alcohols excluding cis D-erythro hydroxyl groups ruled out oxidation by PQQ-
dependent membrane-bound dehydrogenases.
Rare sugars
Gluconobacter
Bertrand Hudson's rule
PQQ-dependent membrane-bound
dehydrogenase
Galactitol
L-xylo-3-hexulose
Methods: Membrane fractions of G. oxydans were prepared and used as a cell-free catalyst to oxidize galactitol,
with or without PQQ as a cofactor.
Results: In this study, we reported an interesting oxidation reaction that the polyhydric alcohols galactitol
(dulcitol), which do not possess cis D-erythro hydroxyl groups, can be oxidized by PQQ-dependent membrane-
bound dehydrogenase(s) of acetic acid bacteria at the C-3 and C-5 hydroxyl groups to produce rare sugars L-xylo-
3-hexulose and D-tagatose.
Conclusions: This reaction may represent an exception to Bertrand Hudson's rule.
General significance: Bertrand Hudson's rule is a well-known theory in polyhydric alcohols oxidation by PQQ-
dependent membrane-bound dehydrogenase in acetic acid bacteria. In this study, galactitol oxidation by a PQQ-
dependent membrane-bound dehydrogenase represents an exception to the Bertrand Hudson's rule. Further
identification of the associated enzymes and deciphering the explicit enzymatic mechanism will prove this
theory.
1. Introduction
dehydrogenase (GOX0854-0855), GOX1441, and 0516 are proved to be
PQQ-dependent; however, some are also FAD-dependent (i.e., D-glu-
Both Gluconobacter and Acetobacter are acetic acid bacteria that
belong to the family Acetobacteraceae. Most of the genera of this family
are well known as vinegar producers because of their strong ability to
oxidize ethanol to acetic acid by membrane-bound dehydrogenases,
including alcohol dehydrogenase and aldehyde dehydrogenase. Besides
ethanol, Gluconobacter and Acetobacter have a highly active respiratory
chain in their membrane, which can oxidize various sugars and sugar
alcohols in a stereo- and regio-selective manner to organic acids, al-
dehydes, and ketones by membrane-bound dehydrogenases [1]. In
Gluconobacter oxydans 621H, eight known and two unknown mem-
brane-bound dehydrogenases have been found [2,3]. Most of these
membrane-bound dehydrogenases, i.e., glucose dehydrogenase
conate dehydrogenase) [4]. The whole cells of G. oxydans have a higher
specific activity of the PQQ-dependent membrane-bound polyol dehy-
drogenases and can be used as biocatalysts to stereo- and re-
giospecifically oxidize polyols, primary/secondary alcohols yielding
corresponding ketoses or organic acids in high yields [1,5,6].
The stereo- and regiospecificity of Gluconobacter and Acetobacter
genera bacterial oxidation in alditol series such as D-arabitol, D-sorbitol,
ribitol, glycerol, meso-erythritol, D-mannitol has been expressed as a
well-known Bertrand-Hudson rule [7], in which polyols having the D-
erythro configuration is oxidized at the secondary alcohol group giving
2-ketose products within the pH range of 5.0–6.5 (Fig. S1a). Whereas
alditols without D-erythro configuration, including D-threitol, L-threitol,
xylitol, L-arabitol, and dulcitol, excludes oxidation by PQQ-dependent
(GOX0265),
inositol
dehydrogenase
(GOX1857),
D-sorbitol
⁎ Corresponding author.
E-mail address: chrqrq@sjtu.edu.cn (H. Cheng).
https://doi.org/10.1016/j.bbagen.2020.129740
Received 11 August 2020; Received in revised form 13 September 2020; Accepted 17 September 2020
Availableonline19September2020
0304-4165/©2020ElsevierB.V.Allrightsreserved.

Y. Xu, et al.
BBA-GeneralSubjects1865(2021)129740
was cultivated in YPD medium (pH 5.5) containing 10 g·L−1 yeast ex-
tract, 5 g·L−1 tryptone, 20 g·L−1 glucose, and 15 g·L−1 agar (only for
solid media).
2.3. General molecular biology methods
Restriction endonucleases, DNA polymerases, and ligases were
purchased from Thermo Fisher Scientific. Genomic DNA was prepared
as reported earlier [13]. High fidelity Taq DNA polymerase (KOD plus,
Toyobo) was used for DNA cloning, whereas genotype verification was
carried out by using ExTaq DNA polymerase (Takara). The PCR-am-
plified products were purified from the agarose gels using a GeneJet Gel
Extraction Kit (Thermo Scientific). Sangon Biotech (Shanghai, China)
performed the primers synthesis and DNA sequencing, while the syn-
thetic DNA was provided by Genewiz (Suzhou, China).
Fig. 1. The regiospecific oxidation of galactitol by the genera of Acetobacter and
Gluconobacter.
Galactitol can be simultaneously oxidized by PQQ-dependent membrane-bound
dehydrogenase(s) to D-tagatose and L-xylo-3-hexulose at the C-5 and C-3, re-
spectively.
membrane-bound dehydrogenases of A. suboxydans (nowadays as G.
suboxydans) [8–11] (Fig. S1b). However, an anomalous oxidation pat-
tern was observed five decades ago in the case of ω-deoxy sugar alco-
hols such as L-fucitol, a kind of 6-methyl substituted hexitol (Fig. S2a)
[8,12]. This reaction to ω-deoxy alcohols was considered as the ex-
tension to Bertrand-Hudson rule, in which the terminal methyl group
was simply as an elongated CH2OH group, i.e., the methyl group can be
seen as the substitution of H atom covalently linked to a carbon atom.
Thus the 6-deoxy-L-galactitol has the D-erythro configuration, and it's
corresponding –OH group can be oxidized to keto group by A. sub-
oxydans [8] (Fig. S2b).
2.4. Biotransformations of galactitol with whole resting cells
The Acetobacter or Gluconobacter strains in Table 1 were first culti-
vated in 250 mL baffled shake flasks containing 50 mL complex
medium (15 g/L yeast extract, 3 g/L tryptone, supplemented with 50 g/
L glucose and 15 g/L calcium carbonate) for 60 h at 30 °C and 150 rpm.
The cells were harvested by centrifugation at 10,000 g for 10 min and
washed twice with sterilized potassium phosphate buffer (10 mM PBS,
pH 6.0). These cells were used for the biotransformation of galactitol
(40 g/L) in 20 mL reaction mixture at 30 °C, and 150 rpm with a final
cell density of OD600 5.0 (1 mL 1 OD600 is approximate 0.32 mg dry cell
weight, thus 20 mL OD600 is equal to 32 mg DCW). Samples were taken
regularly for HPLC analysis. The conversion rate of the product was
calculated as the ratio of product to the substrate, while the production
rate was calculated as the amounts of product (g) produced by per g dry
cell biomass per hour per liter (g/L·h·g biomass).
In this study, we found that the hexitol galactitol (also named as
dulcitol) does not possess the cis D-erythro, but D-lyxo configuration can
also be acted by genera of Acetobacter sp. and Gluconobacter sp. strains.
The enzyme was evidenced to be a PQQ-dependent membrane-bound
dehydrogenase. Both of the hydroxyl groups on C-3 and C-5 can be
oxidized simultaneously by PQQ-dependent membrane-bound dehy-
drogenase from Gluconobacter and Acetobacter genera, with L-xylo-3-
hexulose (PubChem CID: 18392540, also alternative as D-lyxo-4-hex-
ulose) and D-tagatose as products (Fig. 1). Though D-threitol, xylitol,
and D-iditol possess D-lyxo configuration as that of galactitol, however,
all these three sugar alcohols cannot be oxidized by PQQ-dependent
membrane-bound dehydrogenase from Gluconobacter and Acetobacter
genera (Fig. S1b).This galactitol to 3-ketose L-xylo-3-hexulose conver-
sion by PQQ-dependent membrane-bound dehydrogenases may re-
present an exception to Bertrand Hudson's rule, instead of the exten-
sion.
2.5. Bioremoval of residual galactitol and D-tagatose by yeast strains
The G. cerinus X512 strains were first cultured in a 250 mL baffled
shake flask containing 20 g/L yeast extract, 5 g/L tryptone, 50 g/L
glucose, and 15 g/L calcium carbonate for 72 h at 30 °C and 200 rpm.
After cultivation, the cells were harvested by centrifugation at 10,000 g
for 10 min, washed twice with sterilized PBS buffer (10 mM, pH 6.0),
and used for the biotransformation of galactitol (20 g/L) in 50 mL re-
action mixture at 30 °C, and 200 rpm for 120 h with a final cell density
of OD600 5.0. Then the biotransformation liquid was centrifuged, and
the clear supernatant was filtered (0.22 μm membrane) to eliminate
residual cells. Different yeast strains were separately inoculated into the
test tubes containing 5 mL clear biotransformation liquid and cultivated
at 30 °C. Samples were collected regularly to analyze using the HPLC to
determine which yeast strain can completely exhaust galactitol and
tagatose but not the unknown product. Once galactitol and tagatose but
not the unknown product was completely exhausted, the yeast strains
were employed to biodegrade galactitol and tagatose in 250 mL flasks
containing 50 mL biotransformation liquid and cultivated at 30 °C.
When galactitol and tagatose were completely exhausted, the yeast cells
were eliminated by centrifugation, and the clear supernatant was de-
colored and subjected to ion-exchange chromatography by activated
carbon absorption and anion and cation exchange resins. The unknown
product was then collected by HPLC and concentrated by rotary eva-
poration at 50 °C. The resulting purified compound was identified by
the sub-sequential methods.
2. Materials and methods
2.1. Materials
Sugars or sugar alcohols including galactitol, erythritol,
thrulose, D-threitol, xylitol, D-xylulose, D-tagatose, L-tagatose, D-sorLb-ietroyl-,
were purchased from Merck & CoDI-nmca. nonriTtoolk,yoanCdhenmuicclaelaIsneduBsetnryzo(nTaCsIe,
D-arabitol, L-arabitol, L-fucitol,
Japan). NAD, NADH, NADP, NADPH, PQQ, coomassie brilliant blue R-
250, and antibiotics were provided by Sangon Biotech (Shanghai,
China). All other reagents or chemicals used were of high-purity grade.
2.2. Strains, media, and culture conditions
Microorganisms used in this work are listed in Table 1. E. coli strains
were cultured at 37 °C in Luria-Bertani (LB) media supplemented with
ampicillin (100 mg·L−1) or kanamycin (50 mg·L−1). Acetobacter or
Gluconobacter strains were aerobically grown at 30 °C and 200 rpm in a
complex medium comprising 15 g·L−1 yeast extract, 3 g·L−1 tryptone,
and 50 g·L−1 glucose (15 g·L−1 calcium carbonate was added when
necessary). When required, sugar alcohols such as galactitol, L-fucitol,
D-sorbitol, or mannitol were used as carbon and energy sources. Yeast
2.6. Identification of the unknown product
Various methods were applied to identify the unknown compound
produced from galactitol by acetic acid bacteria. First, we excluded the
unknown compound L-galactose, D-galactose, L-tagatose, or D-tagatose,
which C1, C6, C2, or C5 position was dehydrogenated separately.
2

Y. Xu, et al.
BBA-GeneralSubjects1865(2021)129740
Table 1
Strains and plasmids used in this study.
Strains or plasmids
Characteristics
sources
E.coli DH5a
supE44 _lacU169(_80lacZ_M15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1
Stratagene
Stratagene
This study
CGMCC
ATCC
E.coli BL21(DE3)
F-, ompT, hsdS(rBB-mB-), gal, dcm(DE3)
Gluconobacter cerinus X512
Gluconobacter oxydans CGMCC 1.110
Gluconobacter oxydans 621H
Gluconobacter frateurii DSM7146
Acetobacter aceti DSM2002
Acetobacter pasteurianus DSM 3509
Gluconobacter japonicus CGMCC 1.49
Yeast X535 (Candida sp. CGMCC3268)
plasmid: pET28a-AnLadB
DSM
DSM
DSM
CGMCC
This study
This study
Biodegradation of D-tagatose and galactitol, but not L-xylo-3-hexulose
AnLadB gene inserted into NcoI/XhoI in pET28a
pUC57-pqqA-knockout
Containing the upstream region of pqqA (500 bp), kanamycin resistance cassette and 700 bp downstream of pqqA (2.2 kb)
This study
ATCC: American Type Culture Collection.
DSM: Deutsche Sammlung von Mikroorganismen;
CGMCC: China General Microbiological Culture Collection Center.
Authentic L-galactose, D-galactose, L-tagatose, or D-tagatose were ap-
plied to HPLC, and their retention times were compared to that of the
unknown compound. To further determine whether the unknown pro-
duct is aldose or ketose, 5 μL of bromine water (brown) was added into
100 μL of the purified compound, mixed for 1 min, and then incubated
at 65 °C. The unknown compound was also hydrogenated in the pre-
sence of sodium borohydride (NaBH4) according to the strategy of
Mizanur et al. [14]. HPLC separated the reduced products (galactitol
and another isomeric alcohol). The optical rotation of the isomeric al-
cohol was recorded by a polarimeter (P-2000, Jasco, Japan) at 28 °C
using sodium light and water as a control to determine the D or L con-
figuration. The isomeric alcohol was also confirmed by 1H and 13C NMR
(500 MHz Bruker Advance III NMR spectrometer).
analyzed by a mass selective detector (MSD). The temperature program
was as follows: initially, 60 °C held for 1 min, increased to 280 °C at
60 °C/min and held for 5 min, and then increased to 30 °C at 20 °C/min
and held for 2.5 min. Peak identification was achieved by comparing
the retention times with the internal standard in the SIM mode
(Selected Ion Monitoring).
2.9. Preparation of Gluconobacter cells membrane fractions
The G. oxydans 621H strains were grown in complex medium. Cells
were collected by centrifugation (at 10,000 g for 10 min), washed twice
with potassium phosphate buffer (PBS, 10 mM, pH 6.0), and re-
suspended in the same buffer. The cells of G. oxydans 621H strains were
disrupted through a continuous high-pressure cell disrupter. After
centrifugation at 10,000 g for 20 min at 4 °C to eliminate intact cells or
cell debris, the supernatants (crude extract) were centrifuged at
200,000 g for 1 h, yielding soluble cytoplasmic supernatant and pellets
(membrane fraction). The resulting pellets were washed in the same
buffer but containing 1% Triton X-100 and 1 mM PMSF and centrifuged
again (at 200,000 g), then were resuspended in the same buffer con-
taining 1% Triton X-100 and 1 mM PMSF, and used as the membrane
fractions to catalyze D-sorbitol, galactitol, or L-fucitol oxidation. The
protein concentrations were quantified by Bradford assay using bovine
serum albumin (BSA) as a standard.
2.7. Galactitol catalysis by NAD-dependent galactitol dehydrogenase of
Aspergillus niger (AnLadB)
It was reported that AnLadB of A. niger can oxidize 3-OH of ga-
lactitol to L-xylo-3-hexulose [15], which was supposed to be the same
oxidation product of galactitol by G. oxydans 621H or G.cerinus X512.
Thus, the NAD-dependent galactitol dehydrogenase gene (LadB,
NT_166531, ORF: 1077 bp) of Aspergillus niger was in-vitro synthesized
and ligated to the NcoI-XhoI site yielding plasmid pET28a-AnladB. The
resultant plasmid was transformed into E. coli BL21 (DE3) for expres-
sion of A. niger LadB gene. The AnLadB was induced, purified, and used
as an enzyme to catalyze galactitol oxidation. A 500 μL reaction mix-
ture comprising 50 μL purified AnLadB (5 mg/mL), 250 μL 100 mM
galactitol, 50 μL 50 mM NAD, and 150 μL 30 mM Na-phosphate buffer
(pH 8.0), was mixed and incubated at 30 °C for 24 h. The reaction
product was collected by HPLC preparation, concentrated, and further
applied to GC/MS analysis, comparing with that of the reaction product
of galactitol oxidized by G. cerinus X512.
2.10. Effects of EDTA and PQQ on the activity of membrane-bound
galactitol dehydrogenase
To evaluate the effect of EDTA on membrane-bound galactitol de-
hydrogenase activities, the native membrane fraction prepared at
200,000 g was treated with 10 mM EDTA for 5 h at 4 °C. Excess EDTA
was removed by centrifugation at 200,000 g for 60 min, and the enzyme
activity was measured. The reaction mixture (100 μL) contains 20 μL
10 mM PBS buffer (pH 6.0), 40 μL 50 g·L−1 galactitol (the final con-
centration was 20 g·L−1), 10 μL 10 mg/mL EDTA treated or non-treated
membrane proteins, 10 μL 1 mM DCPIP, 10 μL 6.7 mM PMS and 10 μL
40 mM sodium azide, or in the presence of PQQ, Ca2+, and PQQ plus
Ca2+, each at final concentration of 0.5 mM and 1 mM, and reacted at
30 °C for 48 h. The dehydrogenase activities were measured at 600 nm,
as described by Hölscher et al. [16]. One unit of membrane-bound
dehydrogenase activity was defined as the quantity of enzyme that
reduces 1 μmol DCPIP min−1, corresponding to the oxidation of 1 μmol
substrate min−1. The reaction products were also analyzed by HPLC.
2.8. GC/MS comparison analysis of reaction products produced by AnLadB
and G. cerinus
The galactitol oxidation product by purified AnLadB and the un-
known product produced by G. cerinus X512 were dissolved in pyridine
and then converted into trimethylsilyl derivatives by treatment with
derivatizing reagent BSTFA (N, O-Bis (trimethylsilyl) tri-
fluofoacetamine) and heated at 80 °C for 50 min. Then 1 μL of deri-
vatized aliquot was injected into the GC/MS. All GC/MS experiments
were performed on Agilent gas chromatography coupled with an
Agilent ion trap mass spectrometer (Agilent 6850/5975C). The analy-
tical column was an HP5-MS column (30-m capillary column, 0.25 μm
film thickness) with highly pure helium as the carrier gas at a flow of
1 mL/min. After separation by the column, the ionized sample was
3

Y. Xu, et al.
BBA-GeneralSubjects1865(2021)129740
Fig. 2. The HPLC profiles of galactitol biotransformation by
various acetic acid bacteria.
All acetic acid bacteria tested produced two kinds of products
with various efficiency (indicated with i and ii). The seven bac-
terial strains were first cultivated for 60 h at 30 °C and 150 rpm in
complex medium containing 15 g/L yeast extract, 3 g/L tryptone,
supplemented with 50 g/L glucose and 15 g/L calcium carbonate.
After cultivation, the washed cells were used for the bio-
transformation of galactitol (40 g/L) in 20 mL reaction mixture at
30 °C, and 150 rpm with a final cell density of OD600 5.0 (1 mL 1
OD600 is approximate 0.32 g dry cell weight, thus 20 mL OD600 is
equal to 32 mg DCW). G. cerinus X512 has the highest potential to
produce these two products.
2.11. Activity comparison using NAD or PMS/DCPIP as the electron
acceptor for membrane and soluble cytoplasmic fractions
knockout mutant was cultivated in a complex medium containing
20 g·L−1 D-glucose. When OD600 reaches 1.5, the cells were harvested,
washed with PBS buffer (10 mM, pH 6.0), and disrupted to obtain crude
extract. The membrane fraction was obtained by centrifugation (at
200,000 g for 60 min), washed with 10 mM PBS buffer (pH 6.0), cen-
trifuged again (at 200,000 g for 60 min), and resuspended with the
same buffer containing 1% Triton X-100 and 1 mM PMSF. The mem-
brane protein concentration was determined and used to react with L-
fucitol (20 g/L) and galactitol (20 g/L) at a final concentration of
1 mg·mL−1 membrane protein. For reconstitution of quinoprotein
apoenzymes to the holoenzyme, membrane fractions of PQQ-deficient
To determine whether the prepared membrane fractions of G. oxy-
dans 621H have dehydrogenase activity on D-sorbitol and galactitol,
when using NAD or PMS/DCPIP as the electron acceptor, 10 μL 10 mg/
mL membrane proteins, 40 μL 50 g·L−1 D-sorbitol or galactitol (the final
concentration of sugar alcohols was 20 g·L−1), 10 μL 5 mM PQQ, 10 μL
10 mM Ca2+, 10 μL 1 mM DCPIP, 10 μL 6.7 mM PMS and 10 μL 40 mM
sodium azide were mixed and reacted at 30 °C for 48 h. For comparison,
NAD at a final concentration of 5 mM was used as an electron acceptor
instead of PMS/DCPIP.
G. oxydans 621H were incubated with PQQ, Ca2+, and PQQ plus Ca2+
,
On the other hand, dehydrogenase activity of soluble cytoplasmic
fractions on D-sorbitol and galactitol was also assayed using NAD or
PMS/DCPIP as the electron acceptor. The 100 μL reaction mixture
contained 20 μL 10 mM PBS buffer (pH 6.0), 10 μL 10 mg/mL soluble
cytoplasmic proteins, 40 μL 50 g·L−1 D-sorbitol or galactitol (the final
each at a final concentration of 0.5 mM and 1 mM, for 1 h at 4 °C, and
then used as an enzyme to react with galactitol or L-fucitol. The reaction
mixture (100 μL) contains 20 μL 10 mM PBS buffer (pH 6.0), 40 μL
50 g·L−1 galactitol or L-fucitol (the final concentration was 20 g·L−1),
10 μL 10 mg/mL apoenzyme or holo-enzyme, 10 μL 1 mM DCPIP, 10 μL
6.7 mM PMS and 10 μL 40 mM sodium azide, or in the presence of PQQ,
Ca2+, and PQQ plus Ca2+, each at a final concentration of 0.5 mM and
1 mM, and reacted at 30 °C. The dehydrogenase activities were mea-
sured as described above.
concentration of sugar alcohols was 20 g·L−1), 10 μL 10 mM Ca2+
,
10 μL 40 mM sodium azide, 10 μL 50 mM NAD, and reacted at 30 °C. As
a comparison, PQQ, PMS, and DCPIP at a final concentration of 0.5 mM,
0.67 mM, and 0.1 mM were used as the electron acceptor instead of
NAD.
When using PMS/DCPIP as the electron acceptor, the activity was
determined by measuring the decrease in absorbance of DCPIP at
600 nm using PMS as a redox mediator as described by Hölscher et al.
[16], or by measuring the increase in absorbance of NAD at 340 nm.
Reactions were initiated by the addition of substrate, and absorbance
values in the first 3 min were recorded. One unit of dehydrogenase
activity was defined as the quantity of enzyme that reduces 1 μmol
DCPIP or NAD min−1, corresponding to the oxidation of 1 μmol sub-
strate min−1. The specific activity was calculated using a molar ex-
tinction coefficient for DCPIP of 17.2 mM−1 cm−1 at 600 nm and pH 6
[16], and for NAD of 6.22 mM−1 cm−1 at 340 nm and pH 6. The re-
action products were quantified by HPLC.
2.13. HPLC analytical method
The polyol and ketone compounds were separated and quantified by
HPLC coupled with
a Shodex SP0810 ion exclusion column
(300 mm × 8 mm), and a refractive index detector (Shodex RI101),
using distilled water as the mobile phase at a flow rate of 1.0 mL·min−1
at 70 °C.
3. Results
3.1. Oxidation of galactitol by resting cells of Acetobacter and
Gluconobacter strains
2.12. Galactitol catalysis by pqqA knockout mutant of G. oxydans 621H
The acetic acid bacteria Acetobacter and Gluconobacter are well
characterized by their ability to oxidize various sugars and sugar al-
cohols by cytoplasmic soluble or membrane-bound dehydrogenases
using PQQ, FAD, or NAD(P) as cofactors [1–6]. It was reported that
The pqqA knockout mutant of G. oxydans 621H was constructed
according to the strategy of Hölscher and Görisch [17]. The pqqA
4

Y. Xu, et al.
BBA-GeneralSubjects1865(2021)129740
Table 2
Galactitol conversion rate by various acetic acid bacteria.
Strains
Conversion ratea of D-
conversion rate of L-xylo-3- Ratio of L-xylo-3-hexulose
Total conversion rate Production rateb of L-xylo-3-
hexulose
tagatose
hexulose
to D-tagatose
Gluconobacter cerinus X512
Gluconobacter oxydans CGMCC
1.110
0.129
0.039
0.285
0.142
2.21
3.64
0.414
0.181
3.56
1.78
Acetobacter pasteurianus DSM 3509 0.044
0.133
0.089
0.085
0.104
3.02
1.14
1.19
2.48
0.177
0.167
0.156
0.146
1.66
1.11
1.06
1.3
Gluconobacter frateurii DSM7146
Gluconobacter oxydans 621H
Gluconobacter japonicus CGMCC
1.49
0.078
0.071
0.042
Acetobacter aceti DSM2002
0.011
0.055
5
0.065
0.69
a
b
Conversion rate: the amount of product to the amount of substrate galactitol (g/g).
Production rate: the amount of product (g) produced by per g dry cell biomass per hour per liter medium (g/L·h·g dry biomass).
galactitol can be oxidized into D-tagatose by G. oxydans [18]. At the first
purpose, we screened acetic acid bacteria with the best performance to
synthesize D-tagatose from galactitol by the whole-cell transformation.
However, we found that all acetic acid bacteria selected can produce
another unknown product in addition to D-tagatose (Fig. 2). It attracts
our interest to investigate their efficiency to produce this unknown
product from galactitol. The highest total conversion rate of 0.414 was
observed for G. cerinus X512, followed by G. oxydans CGMCC1.110
(0.181) and A. pasteurianus DSM 3509 (0.177). The most commonly
used strain G. oxydans 621H showed a conversion rate of 0.156, while
the lowest yield of 0.065 was recorded by A. aceti DSM2002 (Table 2).
Among these acetic acid bacteria, G. cerinus X512 resting cells yielded
the three-fold higher production rate of the unknown product than that
to G. oxydans 621H, with 3.56 g/L·h·g dry biomass to 1.06 g/L·h·g dry
biomass. HPLC profiles of galactitol biotransformation by various acetic
acid bacteria further corroborated that G. cerinus X512 has the highest
production yield (Fig. 2). The strain G. cerinus X512 was isolated from
sludge in the acetic acid production plant and identified as G. cerinus by
its 16S rDNA sequences compared with those of other known acetic acid
bacteria. The isolated strain was deposited in China General Micro-
biological Culture Collection Center with an accession number
CGMCC19520. Based on those rDNA sequences, the phylogenetic tree
was constructed that showed a close relationship with other known G.
cerinus (Fig. S3). Due to the higher conversion and production rate of
the unknown product, G. cerinus X512 was used to prepare the un-
known product from galactitol for further identification.
3.3. Identification of the unknown product
The unknown compound might be the biochemical oxidation pro-
duct by dehydrogenation of the hydroxyl group at C1, C2, C3, C4, C5, or
C6 position. Provided that dehydrogenation was taken place at C1 or
drogenation was taken place at CL2-goarlaCc5to,steheorunknown compound is
ketose L-tagatose or D-tagatose. However, the retention times of L-ga-
lactose, D-galactose, L-tagatose, or D-tagatose were different from that of
the unknown compound, indicating that dehydrogenation was not
taken place at C1, C2, C5, or C6, but might be dehydrogenated at C3 or
C4, yielding a 3-ketose or 4-ketose, respectively. No discoloration by
the addition of bromine further indicates that the unknown compound
was a ketose (inset in Fig. 4B). The purified unknown product (Fig. 4B)
was reduced by NaBH4. Due to no stereoselectivity of chemical hy-
drogenation, the carbonyl compound would be expected to yield ga-
lactitol and another isomeric alcohol at the oxidized C-OH position in
the ratio of approximately 1:1. The NaBH4 reduction products were
galactitol and the isomeric alcohol of galactitol (Fig. 4C, iii), which
were then isolated by HPLC (Fig. 4D). The HPLC isolated isomeric al-
cohol of galactitol was then identified by different methods. First, the
HPLC retention time was found the same as that of authentic D-sorbitol
and L-sorbitol (Fig. 4E). Then the isomeric alcohol of galactitol was
further identified by optical rotation, 1H NMR, and 13C NMR. The
specific optical rotation was −2.14o, which is identical to authentic D-
sorbitol, indicating its D-configuration. The main signals in the 1H NMR
and 13C NMR of the isomeric alcohol of galactitol were identical with
authentic D-sorbitol (Fig. S4). These results strongly suggested that the
isomeric alcohol of galactitol, ultimately reconfirming as D-sorbitol. The
reducing products were galactitol and D-sorbitol, which enabled us to
deduce that the carbonyl compound oxidized from galactitol by acetic
acid bacteria was L-xylo-3-hexulose. If the OH group at the C4 position
of galactitol was oxidized, the reducing products by NaBH4 should be
galactitol and L-sorbitol, which specific optical rotation was +2.14o.
To further confirm that the oxidation product of galactitol by G.
cerinus X512 was authentic L-xylo-3-hexulose, the GC/MS spectra of
both oxidation products of AnLadB enzyme and G. cerinus X512 were
compared (Fig. S5). It showed that the BSTFA-derivatized galactitol
oxidation products by G. cerinus X512 yielded the same characteristic
mass spectrum ion fragments to the authentic L-xylo-3-hexulose pro-
duced by AnLadB. These combined results indicated that the unknown
product produced by G. cerinus X512 from galactitol was L-xylo-3-hex-
ulose, which is hard to be crystallized and keep syrup through con-
centration.
C6, the unknown compound is
D-galactose. If dehy-
3.2. Biorefinery process to improve the purity of the unknown product
After biotransformation of G. cerinus X512 in complex medium, the
purity of the unknown product was around 28.5%. The main byproduct,
namely D-tagatose and residual substrate galactitol should be removed
to recover the pure unknown product. It is challenging to separate the
unknown product and D-tagatose due to their similar retention time.
Therefore, biotransformation supernatant of G. cerinus X512 was used
as a substrate for the different yeasts to eliminate these two undesired
compounds (D-tagatose and galactitol). The yeast X535 strain was se-
lected because of its ability to utilize galactitol and D-tagatose but not
the unknown product. Then the yeast X535 was identified as Candida
sp. by its 26S rDNA sequence and deposited in China General
Microbiological Culture Collection Center with an accession number
CGMCC3268. This yeast can utilize glycerol, erythritol, xylitol, ribitol,
D-arabitol, galactitol, mannitol, sorbitol, xylose, L-arabinose, D-tagatose,
as well as glucose. As depicted in Fig. 3, X535 yeast cells first consumed
D-tagatose within 24 h followed by galactitol after 60 h of culture. At
this time, the unknown product purity was recorded to be 95% from the
HPLC chromatogram. Then it was further purified through ion-ex-
change chromatography and identified by the following methods.
3.4. D-tagatose and L-xylo-3-hexulose can be produced by membrane-
bound dehydrogenase(s) and the supernatants fractions
To determine whether the galactitol oxidation was mainly taken
place in the membrane or supernatant, the disrupted crude cell extract
5

Y. Xu, et al.
BBA-GeneralSubjects1865(2021)129740
Fig. 3. The purity changes of unknown product during biode-
gradation by yeast Candida sp.
A, 20 g/L galactitol at the beginning of biotransformation; B,
60 h biotransformation by G. cerinus X512 with cell concentra-
tion at 1.5 g/L DCW, unknown product (i) and
were produced; C, D-tagatose was first utilized Db-tyagtahteoseye(aisi)t
Candida sp.; D, after D-tagatose was exhausted, galactitol was
then consumed; E, after D-tagatose and galactitol were exhausted,
the main unknown product reached up to 95% purity.
was ultra-centrifuged (at 200,000 g for 1 h), and the supernatants and
membrane fractions were used as crude enzymes to catalyze galactitol
oxidation. With 20 g/L galactitol, more L-xylo-3-hexulose and D-taga-
tose were produced by membrane fractions than that of the supernatant
fractions. The total production rate of L-xylo-3-hexulose and D-tagatose
was 0.52 g/L·h·g for membrane fractions protein, and 0.21 g/L·h·g for
supernatant fractions protein, 2.5-fold higher than that of supernatant
fractions protein.
decreased after treatment with 10 mM EDTA (Fig. 5C,D). The total
production rate of L-xylo-3-hexulose and D-tagatose reduced to 0.08 g/
L·h·g from 0.21 g/L·h·g.
These results indicated that galactitol could be oxidized to L-xylo-3-
hexulose and D-tagatose by both membrane and supernatant fractions,
PQQ and EDTA have a significant effect on L-xylo-3-hexulose and D-
tagatose production. So, we further determined the effects of EDTA and
PQQ on membrane galactitol dehydrogenase activity.
Comparing to the reaction mixture without supplement of PQQ and
Ca2+ (Fig. 5A), the membrane reaction mixture produced more L-xylo-
3-hexulose and D-tagatose (inset in Fig. 5B) by addition of PQQ and
Ca2+ to a final concentration of 0.5 mM and 1 mM (Fig. 5B). The total
production rate of L-xylo-3-hexulose and D-tagatose increased to 0.94 g/
L·h·g from 0.52 g/L·h·g, about 2-fold higher than that of without the
addition of PQQ and Ca2+. When using supernatant fraction as an en-
zyme source, the amount of L-xylo-3-hexulose and D-tagatose were
The galactitol dehydrogenase activity of native membrane fractions
was around 0.026 U/mg protein, while the activity was increased to
0.057 U/mg protein by the addition of PQQ at the final concentration
0.5 mM to replenish the PQQ loss during the membrane preparation by
ultra-centrifugation at 200,000 g. Addition of 1 mM Ca2+ plus 0.5 mM
PQQ only resulted in a slight increase than that of 0.5 mM PQQ addi-
tion. Whereas the membrane-bound galactitol dehydrogenase activity
was reduced by 93% after treatment with 10 mM EDTA (Fig. 6, pink
Fig. 4. Identification of the galactitol transformation products
by chemical reduction with NaBH4.
A, HPLC profile of authentic galactitol, retention time
25.99 min; B, HPLC profile of the purified unknown product
(i). C, chemically reduced product by NaBH4, resulting in two
polyols, one is galactitol (ii), and the other polyol (iii) with the
same retention time to D-sorbitol (29.1 min). D, isolated polyol
(iii) with the same retention time to D-sorbitol. E, the authentic
D-sorbitol. By addition of bromine into the unknown product,
no discoloration was observed as the same to that of fructose
(was still bromine colour), whereas discoloration was observed
for glucose due to oxidation of bromine by aldehyde group in
glucose, indicating that the unknown compound was a ketose
(inset in Figure).
6

Y. Xu, et al.
BBA-GeneralSubjects1865(2021)129740
Fig. 5. Effects of PQQ and EDTA on products
formation.
For A and B, membrane-bound fractions used
were obtained by centrifugation 200,000 g,
then were used as catalyst to react on ga-
lactitol. A, the reaction product with mem-
brane fraction; B, the reaction product with
membrane fraction by the addition of PQQ
and Ca2+ at a final concentration of 0.5 mM
and 1 mM respectively; The inset is the zoom
and overlap of A and B. Red line, A. Blue line,
B. For A, the reaction mixture (100 μL) con-
tains 50 μL 10 mM potassium phosphate
buffer (pH 6.0), 40 μL 50 g·L−1 galactitol (the
final concentration was 20 g·L−1), 10 μL
10 mg/mL membrane proteins, reacted at
30 °C for 48 h. For B, the reaction mixture
(100 μL) contains 30 μL 10 mM potassium
phosphate buffer (pH 6.0), 40 μL 50 g·L−1
galactitol (the final concentration was
20 g·L−1), 10 μL 10 mg/mL membrane pro-
teins prepared at 200000 g, 10 μL 5 mM PQQ
and 10 μL 10 mM Ca2+, reacted at 30 °C for
48 h. Then the reaction products were ana-
lyzed by HPLC.
For C and D, G. oxysans 621H cells crude ex-
tract was centrifuged at 200,000 g, and su-
pernatants were used as a crude enzyme to
catalyze 20 g/L galactitol at 30 °C for 48 h. C,
HPLC analysis of reaction product with su-
pernatant; D, HPLC analysis of reaction pro-
duct with supernatant by the addition of
10 mM EDTA. The inset is the overlap and
zoom of C and D. Red line, C profile. Blue, D
profile. Addition of 10 mM EDTA decreased
product formation, indicating the enzyme in
supernatant oxidizing galactitol to L-xylo-3-
hexulose and D-tagatose might from the so-
luble membrane fraction.
column), which converts quinoprotein galactitol dehydrogenase into
apoenzyme form. When PQQ, Ca2+, or both were added, their activities
were significantly increased from 0.002 U/mg to 0.037, 0.045, and
0.064 U/mg (Fig. 6). These results revealed the PQQ -dependent nature
of membrane-bound galactitol dehydrogenase.
3.5. Activity determination using NAD or PMS/DCPIP as the electron
acceptor for membrane and soluble cytoplasmic fractions
The activities of the soluble cytoplasmic and the membrane fraction
prepared at 200,000 g with PMS/DCPIP and NAD as electron acceptors
were determined using galactitol and D-sorbitol as substrates, respec-
tively, and the results are shown in Fig. 7. When using galactitol as
Fig. 6. Effects of EDTA and PQQ on membrane-bound
galactitol dehydrogenase activities.
When EDTA was added to the reaction mixture, the
dehydrogenase activity using galactitol as substrate
was decreased, while the activity can be restored
when adding PQQ or/Ca2+. The results showed the
membrane-bound galactitol dehydrogenase activity
was inhibited when treated with EDTA and was PQQ
dependent.
7

Y. Xu, et al.
BBA-GeneralSubjects1865(2021)129740
Fig. 7. Effects of different electron acceptors on the
activities of soluble cytoplasmic and membrane-bound
fractions.
When using NAD as an electron acceptor, the mem-
brane fractions prepared at 200,000 g only have
background activity whatever using galactitol or D-
sorbitol as substrates. Whereas using PMS/DCPIP as
the electron acceptor, the membrane fractions have
strong activities using galactitol or D-sorbitol as sub-
strates. For soluble fractions, the activities can be de-
tected using NAD or PMS/DCPIP as the electron ac-
ceptor and galactitol or D-sorbitol as substrates. The
results indicated that the main dehydrogenases in the
membrane fractions utilized PMS/DCPIP as electron
acceptor and were PQQ-dependent.
substrate and PMS/DCPIP as an electron acceptor, the detected activity
was 0.056 U/mg for membrane fraction proteins. On the other hand,
the activity was only 0.001 U/mg of membrane fraction proteins using
NAD as an electron acceptor, almost the same as the background ac-
tivity without the addition of NAD. This indicates that the membrane
enzyme activity can only use PMS/DCPIP as the electron acceptor ra-
ther than NAD.
containing glucose but with some slower growth rate (0.05 h−1 to
0.22 h−1 of wild type). On the complex medium plate containing glu-
cose and CaCO3, the pqqA mutant cannot oxidize glucose to gluconate,
hence unable to dissolve CaCO3 to form a transparent circle, but the
wild type can form a transparent circle (Fig. S6). G. oxydans has two
pathways to acquire energy from glucose, one is to incompletely oxidize
glucose to gluconate by a PQQ-dependent glucose dehydrogenase
[17,22], the other is to form acetate from glucose via glycolysis in
which pyruvate was converted to acetaldehyde by pyruvate decarbox-
ylase (PDC) and then oxidized to acetate by acetaldehyde dehy-
drogenase (AcDH) with the generation of a molecular NADPH [23], or
By using a soluble fraction of cytoplasm as an enzyme source in the
presence of PMS/DCPIP as the electron acceptor and galactitol as
substrate, the activity decreased to 0.013 U/mg for soluble fraction
protein, only one-fourth of the activity of membrane fraction proteins.
When using NAD as an electron acceptor and galactitol as substrate, the
galactitol dehydrogenase activity can be detected (0.003 U/mg), al-
though only one-fourth of activity to that of using PMS/DCPIP as an
electron acceptor. The results indicate that the soluble fraction contains
the NAD-dependent galactitol dehydrogenase, which might oxidize
galactitol to D-tagatose, which was similar to that of early reported by
Rollini and Manzoni [19].
to form keto-gluconate from glucose by cytoplasmic soluble NADP+
-
dependent glucose dehydrogenase [24]. Though the PQQ-dependent
glucose dehydrogenase gene was deleted, G. oxydans could retain
growth on glucose [22].
The pqqA mutant was grown on a complex medium containing 20 g/
L
D-glucose, and the membrane fraction was prepared by ultra-
centrifugation at 200,000 g for 60 min. The isolated membrane fraction
and supernatant were applied to react with galactitol and 6-deoxy-L-
galactitol (L-fucitol). In pqqA mutant membrane or supernatant, both
the activity of galactitol or L-fucitol dehydrogenase was absent, and no
L-xylo-3-hexulose and D-tagatose or L-fuco-4-ketose were detected by
HPLC. When pqqA mutant membrane fraction was incubated with
0.5 mM PQQ and/or 1 mM Ca2+ to reconstitute quinoprotein apoen-
zyme to holoenzymes, galactitol dehydrogenase activity can be restored
to about the half of those of wild type membrane (Fig. 8) and produce L-
xylo-3-hexulose and D-tagatose or L-fuco-4-ketose (Fig. 9).
The membrane fractions in acetic acid bacteria, such as G. sub-
oxydans contain PQQ-dependent dehydrogenases, which can efficiently
oxidize
containsDt-shoerbNitAoDl -tdoepLe-nsodrebnotsede[h2y0d]r.ogTehneasseosl,uwblheicchytcoasonliocxifdraizcetion
sorbitol to D-fructose [21]. For comparison, D-sorbitol was used asDa-
substrate instead of galactitol to assay the dehydrogenase activity, using
PMS/DCPIP or NAD as an electron acceptor, membrane fractions, and
soluble cytosolic fractions as crude enzymes. In the presence of soluble
fraction as enzyme source and PMS/DCPIP and NAD as an electron
acceptor, the sorbitol dehydrogenase activity was 0.085 and 0.012 U/
mg, respectively, indicating that the soluble fraction of enzyme contains
both PQQ-dependent and NAD-dependent sorbitol dehydrogenase,
mainly PQQ-dependent one. Using the membrane fractions as an en-
zyme source, and PMS/DCPIP as an electron acceptor, the activity was
0.95 U/mg, 15-fold higher than the activity using galactitol as a sub-
strate. Only background activity was determined using NAD as an
electron acceptor, indicating that the PQQ-dependent D-sorbitol dehy-
drogenase was the main dehydrogenase in the membrane fractions.
4. Discussion
In recent years, rare sugars received great interest in nutrition, fine
chemicals, and medicine. For instance, D-mannose has recently been
found to be an excellent rare sugar that impairs several tumors growth
and enhances cell death in response to chemotherapy [25]. D-allose
induces programmed cell death in prostate cancer and human retinal
progenitor cells (hRPC) by modulating the intrinsic apoptotic pathway
[26]. D-tagatose, another important rare sugar, plays a significant role
in healthy foods, management of type 2 diabetes, and the ingredient in
drug manufacture [27]. With the original purpose of producing D-ta-
gatose from galactitol according to the strategy of Manzoni [18] and
Rollini [19], we found unexpectedly that L-xylo-3-hexulose, a 3-ketose,
was produced as well as 5-ketose D-tagatose from galactitol by acetic
acid bacteria (Fig. 1, Fig. 2). Recently, 3-ketose sugars have garnered
attention owing to their potential value as building blocks in the
synthesis of branched-chain sugars, amino sugars, and sugar nucleo-
tides [28–30].
3.6. pqqA gene deletion resulting in no galactitol oxidation
The pqqA gene was deleted by homologous replacement to de-
termine the PQQ-dependent activity of membrane-bound galactitol
dehydrogenase. The mutant was stretched five times on mannitol solid
medium to dilute residual PQQ originated from the wild type cells
before electroporation [17]. The obtained pqqA mutant was designated
as GOX-621H-CHR1. Hölscher and Görisch reported that pqqA mutant
could not grow on a complex medium containing D-sorbitol, D-mannitol,
glycerol, and glucose, but able to slowly grow on D-gluconate [17].
However, in contrast to the result obtained by Hölscher and Görisch
[17], the pqqA mutant GOX-621H-CHR1 can grow on complex medium
It was found that the acetic acid bacterium such as Gluconobacter sp.
and Acetobacter sp. possess the ability to incompletely oxidize various
sugar alcohols regiospecificity at the C-2 position to the corresponding
8

Y. Xu, et al.
BBA-GeneralSubjects1865(2021)129740
Fig. 8. Comparison of membrane-bound activities of
galactitol dehydrogenase prepared from pqqA gene
disruption mutant and wild type of G. oxydans 621H.
When pqqA gene was disrupted, the membrane-
bound galactitol dehydrogenase activity reduced to
the background level, and can be restored at some
extent by addition of PQQ with or without Ca2+
.
2-ketose (for example, D-mannitol to D-fructose, D-sorbitol to L-sorbose,
D-arabitol to D-xylulose, etc.) by using membrane-bound dehy-
drogenases in the presence of PQQ as a cofactor. These reactions obey
the well-known Bertrand-Hudson's rule [10,31,32], in which polyols
exhibiting D-erythro configuration are oxidized at the secondary alcohol
group generating 2-ketose products within the pH 5–6.5 (Fig. S1a). No
hydroxyl group at the C3 position of polyols (for example, hexitol,
pentitol) being regio-specifically oxidized by acetic acid bacteria was
observed so far in the literature. However, we found that galactitol can
be spontaneously oxidized at the C5 and C3 position hydroxyl group to
5-ketose D-tagtose and 3-ketose L-xylo-3-hexulose by PQQ-dependent
membrane-bound dehydrogenase in acetic acid bacteria, and do not
follow the Bertrand-Hudson's rule.
Manzoni [18] and Rollini [19] reported D-tagatose production from
galactitol by three Acetobacter sp. and five Gluconobacter sp. including
G. oxydans NCIB 621 and DSM 2343, using either resting cells or
growing cells, and achieved the highest D-tagatose titer of 4.4 g/L from
22.5 g/L galactitol using DSM 2343. Notably, no possible by-products
such as galactose or fructose were detected in any of the samples using
the Merck Polyspher OA-KC column at room temperature. However, 3-
ketose L-xylo-3-hexulose along with D-tagatose was detected in all the
tested samples, including the sample of DSM 2343 by HPLC using a
shodex column SP0810 eluted at 1 mL/min with distilled water at 70 °C
(Fig. 2). The reason why Manzoni [18] and Rollini [19] only detected
Fig. 9. PQQ effect on reaction products with
membrane fraction of pqqA knockout mutant.
Membrane fraction of mutant G. oxydans 621H
with pqqA deficiency was obtained by cen-
trifugation at 200,000 g for 60 min, and wa-
shed with 10 mM potassium phosphate sodium
buffer (pH 6.0) containing 1% Triton X-100,
centrifugation again at 200,000 g for 60 min to
obtain membrane fraction, and resuspended
with the same buffer and was used to react on
L-fucitol and galactitol. A, pqqA knockout mu-
tant membrane fraction act on L-fucitol without
tose was produced; B, pqqA knockLo-ufutcmo-u4t-aknet-
the addition of PQQ, no product
membrane fraction act on L-fucitol addition of
PQQ and Ca2+, product L-fuco-4-ketose was
produced (peak i); C, pqqA knockout mutant
membrane fraction act on galactitol without
the addition of PQQ, no products L-xylo-3-
hexulose and
pqqA knockoutDm-tuatgaantotsmeewmebrreanperfordauctcieodn; aDct,
on galactitol addition of PQQ and Ca2+, pro-
ducts L-xylo-3-hexulose and D-tagatose were
produced (peak ii and iii); E, overlap of B and
D.
For A and C, the reaction mixture (100 μL)
contains 50 μL 10 mM potassium phosphate
buffer (pH 6.0), 40 μL 50 g·L−1
galactitol (C) (the final concLe-nfutrcaittiooln(Aw) oasr
20 g·L−1), 10 μL 10 mg/mL membrane proteins
prepared at 200,000 g, reacted at 30 °C for
48 h. For B and D, the reaction mixture (100 μL) contains 30 μL 10 mM potassium phosphate buffer (pH 6.0), 40 μL 50 g·L−1 galactitol (the final concentration was
20 g·L−1), 10 μL 10 mg/mL membrane proteins, 10 μL 5 mM PQQ, 10 μL 10 mM Ca2+, reacted at 30 °C for 48 h.
9

Y. Xu, et al.
BBA-GeneralSubjects1865(2021)129740
the 5-ketose D-tagatose but not the 3-ketose is that the two compounds
D-tagatose and 3-ketose L-xylo-3-hexulose were co-eluted in the column
they used, but can be efficiently separated by the column we used.
There are two sets of dehydrogenases in acetic acid bacteria such as
G. oxydans. One set is cytosolic NAD(P)-dependent dehydrogenases,
which oxidize sugar alcohols to biosynthetic pathways intermediates;
the other is PQQ-dependent membrane-bound dehydrogenases that
incompletely oxidize sugar alcohols to provide energy for growth [33].
For example, G. oxydans ATCC 621H contains a membrane-bound and a
cytoplasmic NAD-dependent inositol dehydrogenase [16]. Glucono-
bacter frateurii CHM 43 possesses a PQQ-dependent membrane-bound
meso-erythritol dehydrogenase as well as a NAD-dependent counterpart
[34]. So, it is crucial to determine the oxidation of galactitol at C-5 and
C-3 position to D-tagatose and L-xylo-3-hexulose are catalyzed by PQQ-
dependent membrane-bound dehydrogenases and/or soluble NAD(P)-
dependent dehydrogenases in the cytosol of acetic acid bacteria. As
described above, the oxidation reaction of polyols (C3-C6) catalyzed by
PQQ-dependent membrane-bound dehydrogenases obeys the well-
5. Conclusion
Taken together, galactitol is a unique polyol, which not only can be
oxidized at C5 to give D-tagatose but also at C3 position to yield an-
other rare ketose L-xylo-3-hexulose by G. oxydans PQQ-dependent
membrane-bound dehydrogenase that does not conform to Bertrand-
Hudson's rule. Whereas other polyols can only be oxidized at the sec-
ondary alcohol group adjacent to the primary terminal alcohol by G.
xoydans PQQ-dependent membrane-bound dehydrogenase and obey
the Bertrand-Hudson's rule. Identification of the PQQ-dependent
membrane-bound dehydrogenase and elucidation of the mechanism
will better contribute to understand why galactitol can be simulta-
neously oxidized at C5 and C3 position hydroxyl group, which is being
under-investigated in our laboratory.
Author contributions
HC conceived the conceptualization, funding acquisition, writing
the original draft. YX, PC, JL performed the experiments. MB review
and edit this manuscript.
known Bertrand-Hudson's rule. According to this rule,
D-
mannitol, D-sorbitol, D-ribitol, and Meso-erythritol (D-erDy-tahrraobictoonl,fig-
uration) were oxidized by the polyol dehydrogenase, while L-arabitol,
xylitol，D-threitol, L-threitol, and galactitol were not used as the sub-
strates. In this study, we found that galactitol can be simultaneously
oxidized into D-tagatose and L-xylo-3-hexulose by PQQ-dependent
membrane-bound dehydrogenases, but not soluble NAD(P)-dependent
cytosolic dehydrogenases. The addition of EDTA into membrane frac-
tion results in decreased galactitol dehydrogenase activity (Fig. 7) and
reduced D-tagatose and L-xylo-3-hexulose synthesis. Incorporation of
PQQ and Ca2+ into pqqA knockout G. oxydans membrane fraction
partly restores enzyme activity and enhances galactitol oxidation to
produce both D-tagatose and L-xylo-3-hexulose (Fig. 8, Fig. 9).
These results indicated that the oxidation of galactitol into D-taga-
tose and L-xylo-3-hexulose was catalyzed by PQQ-dependent mem-
brane-bound dehydrogenase(s). However, galactitol does not possess
the D-erythro configuration and not obey Bertrand-Hudson's rule; thus,
the galactitol oxidation by PQQ-dependent membrane-bound dehy-
drogenase of acetic acid bacteria such as G. oxydans and A. suboxydans
may represent an exception to Bertrand Hudson's rule.
Declaration of Competing Interest
The authors declare that they have no competing financial interest.
Acknowledgement
This work was financially supported by the National Natural Science
Foundation of China [No. 21877078].
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.bbagen.2020.129740.
References
[1] U. Deppenmeier, M. Hoffmeister, C. Prust, Biochemistry and biotechnological ap-
plications of Gluconobacter strains, Appl. Microbiol. Biotechnol. 60 (2002) 233–242.
[2] C. Prust, M. Hoffmeister, H. Liesegang, A. Wiezer, W.F. Fricke, A. Ehrenreich,
G. Gottschalk, U. Deppenmeier, Complete genome sequence of the acetic acid
bacterium Gluconobacter oxydans, Nat. Biotechnol. 23 (2005) 195–200.
[3] B. Peters, M. Mientus, D. Kostner, A. Junker, W. Liebl, A. Ehrenreich,
Characterization of membrane-bound dehydrogenases from Gluconobacter oxydans
621H via whole-cell activity assays using multideletion strains, Appl. Microbiol.
Biotechnol. 97 (2013) 6397–6412.
Many PQQ-dependent membrane-bound dehydrogenases have been
recognized in Gluconobacter strains. Peters and coworkers constructed a
set of mutants in which all putative membrane-bound dehydrogenases
genes were consecutively deleted and investigated the substrate oxi-
dation spectra of mutants. The resulting multi-deletion strains (BP.9
and BP.10) did not oxidize tested polyols, including glycerol, D-arabitol,
D-sorbitol, D-mannitol, meso-erythritol, D-threitol, L-threitol, and ribitol
[22]. However, they did not use galactitol as a substrate, so it was still
challenging to scrutinize which membrane-bound dehydrogenase oxi-
dizes galactitol to D-tagatose and L-xylo-3-hexulose. We deduced that
the enzyme oxidizing galactitol to D-tagatose and L-xylo-3-hexulose was
mainly PQQ-dependent membrane-bound in Gluconobacter bacteria
(Figs. 5–9), and the enzyme identification is being undertaken in our
group.
[4] E.K. Shinagawa, M. Matsushita, O. Adachi, M. Ameyama, D-gluconate dehy-
drogenase, 2-keto-D-gluconate yielding, from Gluconobacter dioxyacetonicus: pur-
ification and characterization, Agri. Biol. Chem. 48 (1984) 1517–1522.
[5] L.J. Wei, J. Zhou, D. Zhu, B. Cai, J. Lin, Q. Hua, D. Wei, Functions of membrane-
bound alcohol dehydrogenase and aldehyde dehydrogenase in the bio-oxidation of
alcohols in Gluconobacter oxydans DSM 2003, Biotechnol. Bioprocess Eng. 17 (2012)
1156–1164.
[6] I. Lapenaite, B. Kurtinaitiene, J. Razumiene, V. Laurinavicius, L. Marcinkeviciene,
I. Bachmatova, R. Meškys, A. Ramanavicius, Properties and analytical application of
PQQ-dependent glycerol dehydrogenase from Gluconobacter sp. 33, Anal. Chim.
Acta 549 (2005) 140–150.
In addition to acetic acid bacteria, some filamentous fungi such as
Hypocrea jecorina and Aspergillus niger can also utilize galactitol. The
enzyme has been identified to be L-arabinitol 4-dehydrogenase (LAD),
which catalyzes the transformation of L-arabinitol into L-xylulose, D-
sorbitol to
[7] M.R. Hann, B.E. Tilden, C.S. Hudson, The oxidation of sugar alcohols by Acetobacter
suboxydans, J. Am. Chem. Soc. 60 (1938) 1201–1203.
[8] N.K. Richtmyer, L.C. Stewart, C.S. Hudson, L-Fuco-4-ketose, a new sugar produced
by the action of Acetobater suboxydans on L-fucitol, J. Am. Chem. Soc. 72 (1950)
4934–4937.
mannitol toDL--ffrruuccttoossee,, Da-lal lalittothletoC-D2-ppsoicsiotsioen, ,Dw-giuthlitNolADtoaDs-saocrboofascet,oLr-.
However, it exclusively oxidizes galactitol at the C-3 position yielding L-
xylo-3-hexulose [15,35], with the same spectra mass fractions to the
oxidation product of galactitol by PQQ-dependent purified membrane
of G. oxydans. Galactitol oxidation to L-xylo-3-hexulose (C-3 position)
by acetic acid bacteria and fungi may represent two entirely different
mechanisms, one is membrane-bound that uses PQQ as a cofactor,
while the other is cytoplasmic and utilizes NAD as a cofactor.
[9] D.T. William, J.K.N. Jones, Further experiments on the oxidation of sugar acetals
and thioacetals by Acetobacter suboxydans, Can. J. Chem. 45 (1967) 741–744.
[10] O. Adachi, D. Moonmangmee, E. Shinagawa, H. Toyama, M. Yamada,
K. Matsushita, New quinoproteins in oxidative fermentation, Biochim. Biophys.
Acta 1647 (2003) 10–17.
[11] O. Adachi, Y. Fujii, M.F. Ghaly, H. Toyama, E. Shinagawa, K. Matsushita,
Membrane-bound quinoprotein D-arabitol dehydrogenase of Gluconobacter sub-
oxydans IFO 3257: a versatile enzyme for the oxidative fermentation of various
ketoses, Biosci. Biotechnol. Biochem. 65 (2001) 2755–2762.
[12] G.N. Bollenback, L.A. Underkofler, The action of Acetobacter suboxydans upon ω-
desoxy sugar alcohols, J. Am. Chem. Soc. 72 (1950) 741–745.
[13] H. Cheng, Z. Li, N. Jiang, Z. Deng, Cloning, purification and characterization of an
10

Y. Xu, et al.
BBA-GeneralSubjects1865(2021)129740
NAD-dependent D-arabitol dehydrogenase from acetic acid bacterium, Acetobacter
[25] P.S. Gonzalez, J. O’Prey, S. Cardaci, V.J.A. Barthet, J.I. Sakamaki, F. Beaumatin,
A. Roseweir, D.M. Gay, G. Mackay, G. Malviya, E. Kania, S. Ritchie, A.D. Baudot,
B. Zunino, A. Mrowinska, C. Nixon, D. Ennis, A. Hoyle, D. Millan, I.A. McNeish,
O.J. Sansom, J. Edwards, K.M. Ryan, Mannose impairs tumour growth and en-
hances chemotherapy, Nature 563 (7733) (2018) 719–723.
suboxydans, Protein J. 28 (2009) 263–272.
[14] R.M. Mizanur, K. Takeshita, H. Moshino, G. Takada, K. Izumori, Production of L-
erythrose via L-erythrulose from erythritol using microbial and enzymatic reactions,
J. Biosci. Bioeng. 92 (2001) 237–241.
[15] D. Mojzita, O.M. Koivistoinen, H. Maaheimo, M. Penttilä, L. Ruohonen, P. Richard,
Identification of the galactitol dehydrogenase, LadB, that is part of the oxidor-
eductase D-galactose catabolic pathway in Aspergillus niger, Fungal Genet. Biol. 49
(2012) 152–159.
[26] N. Naha, H.Y. Lee, M.J. Jo, B.C. Chung, S.H. Kim, M.O. Kim, Rare sugar D-allose
induces programmed cell death in hormone refractory prostate cancer cells,
Apoptosis 13 (2008) 1121–1134.
[27] J. Jayamuthunagai, P. Gautam, G. Srisowmeya, M. Chakravarthy, Biocatalytic
production of D-tagatose: a potential rare sugar with versatile applications, Crit.
Rev. Food Sci. Nutr. 57 (2017) 3430–3437.
[16] T. Hölscher, D. Weinert-Sepalage, H. Görisch, Identification of membrane-bound
quinoprotein inositol dehydrogenase in Gluconobacter oxydans ATCC 621H,
Microbiology 153 (2007) 499–506.
[28] T. Müller, R.R. Schmidt, Investigation towards the synthesis of dTDP-2,6-dideoxy-D-
erythro-3-hexulose—A potential intermediate in the biosynthesis of rare sugars,
Tetrahedron Lett. 38 (1997) 5473–5476.
[17] T. Hölscher, H. Görisch, Knockout and overexpression of pyrroloquinoline quinone
biosynthetic genes in Gluconobacter oxydans 621H, J. Bacteriol. 188 (2006)
7668–7676.
[29] C.E. Melancon, W.L. Yu, H.W. Liu, TDP-mycaminose biosynthetic pathway revised
and conversion of desosamine pathway to mycaminose pathway with one gene, J.
Am. Chem. Soc. 127 (2005) 12240–12241.
[18] M. Manzoni, M. Rollini, S. Bergomi, Biotransformation of galactitol to tagatose by
acetic acid bacteria, Process Biochem. 36 (2001) 971–977.
[19] M. Rollini, M. Manzoni, Bioconversion of galactitol to tagatose and dehydrogenase
activity induction in Gluconobacter oxydans, Process Biochem. 40 (2005) 437–444.
[20] T. Hoshino, T. Sugisawa, M. Shinjoh, N. Tomiyama, T. Miyazaki, Membrane-bound
D-sorbitol dehydrogenase of Gluconobacter suboxydans IFO 3255-enzymatic and
genetic characterization, Biochim. Biophys. Acta 1647 (2003) 278–288.
[21] O. Adachi, H. Toyama, G. Theeragool, N. Lotong, K. Matsushita, Crystallization and
properties of NAD-dependent D-sorbitol dehydrogenase from Gluconobacter sub-
oxydans IFO 3257, Biosci. Biotechnol. Biochem. 63 (1999) 1589–1595.
[22] B. Peters, M. Mientus, D. Kostner, A. Junker, W. Liebl, A. Ehrenreich,
Characterization of membrane-bound dehydrogenases from Gluconobacter oxydans
621H via whole-cell activity assays using multideletion strains, Appl. Microbiol.
Biotechnol. 97 (2013) 6397–6412.
[30] Z.P. Lin, H.C. Lin, H.H. Wu, H.W. Chou, S.K. Lin, K.C. Sung, F.F. Wong,
Hydrodeamination of β-enamina ketones to 1,2-dideoxy-D-threo-3hexulose via
palladium, Tetrahedron Lett. 50 (2009) 5120–5122.
[31] H. Zhang, L. Shi, X. Mao, J. Lin, D. Wei, Enhancement of cell growth and glycolic
acid production by overexpression of membrane-bound alcohol dehydrogenase in
Gluconobacter oxydans DSM 2003, J. Biotechnol. 237 (2016) 18–24.
[32] T. Yakushi, S. Fukunari, T. Kodama, M. Matsutani, S. Nina, N. Kataoka,
G. Theeragool, K. Matsushita, Role of a membrane-bound aldehyde dehydrogenase
complex AldFGH in acetic acid fermentation with Acetobacter pasteurianus
SKU1108, Appl. Microbiol. Biotechnol. 102 (2018) 4549–4561.
[33] K. Matsushita, H. Toyama, O. Adachi, Respiratory chains and bioenergetics of acetic
acid bacteria, Adv. Microb. Physiol. 36 (1994) 247–301.
[23] B. Peters, A. Junker, K. Brauer, B. Mühlthaler, D. Kostner, M. Mientus, W. Liebl,
A. Ehrenreich, Deletion of pyruvate decarboxylase by a new method for efficient
markerless gene deletions in Gluconobacter oxydans, Appl. Microbiol. Biotechnol. 97
(2013) 2521–2530.
[34] D. Moonmangmee, E. Shinagawa, H. Toyama, G. Theeragool, N. Lotong,
K. Matsushita, O. Adachi, L-Erythrulose production by oxidative fermentation is
catalyzed by PQQ-containing membrane-bound dehydrogenase, Biosci. Biotechnol.
Biochem. 66 (2002) 307–318.
[24] U. Herrmann, M. Merfort, M. Jeude, S. Bringer-Meyer, H. Sahm, Biotransformation
of glucose to 5-keto-D-gluconic acid by recombinant Gluconobacter oxydans DSM
2343, Appl. Microbiol. Biotechnol. 64 (2004) 86–90.
[35] M. Pail, T. Peterbauer, B. Seiboth, C. Hametner, I. Druzhinina, C.P. Kubicek, The
metabolic role and evolution of L-arabinitol 4-dehydrogenase of Hypocrea jecorina,
Eur. J. Biochem. 271 (2004) 1864–1872.
11