Molecular insight into regioselectivity of transfructosylation catalyzed by GH68 levansucrase and β-fructofuranosidase

Article abstract of DOI:10.1016/j.jbc.2021.100398

Glycoside hydrolase family 68 (GH68) enzymes catalyze β-fructosyltransfer from sucrose to another sucrose, the so-called transfructosylation. Although regioselectivity of transfructosylation is divergent in GH68 enzymes, there is insufficient information available on the structural factor(s) involved in the selectivity. Here, we found two GH68 enzymes, β-fructofuranosidase (FFZm) and levansucrase (LSZm), encoded tandemly in the genome of Zymomonas mobilis, displayed different selectivity: FFZm catalyzed the β-(2→1)-transfructosylation (1-TF), whereas LSZm did both of 1-TF and β-(2→6)-transfructosylation (6-TF). We identified His79^FFZm and Ala343^FFZm and their corresponding Asn84^LSZm and Ser345^LSZm respectively as the structural factors for those regioselectivities. LSZm with the respective substitution of FFZm-type His and Ala for its Asn84^LSZm and Ser345^LSZm (N84H/S345A-LSZm) lost 6-TF and enhanced 1-TF. Conversely, the LSZm-type replacement of His79^FFZm and Ala343^FFZm in FFZm (H79N/A343SFFZm) almost lost 1-TF and acquired 6-TF. H79N/A343S-FFZm exhibited the selectivity like LSZm but did not produce the β-(2→6)-fructoside-linked levan and/or long levanooligosaccharides that LSZm did. We assumed Phe189^LSZm to be a responsible residue for the elongation of levan chain in LSZm and mutated the corresponding Leu187^FFZm in FFZm to Phe. An H79N/L187F/A343S-FFZm produced a higher quantity of long levanooligosaccharides than H79N/A343S-FFZm (or H79NFFZm), although without levan formation, suggesting that LSZm has another structural factor for levan production. We also found that FFZm generated a sucrose analog, β-D-fructofuranosyl α-D-mannopyranoside, by β-fructosyltransfer to D-mannose and regarded His79^FFZm and Ala343^FFZm as key residues for this acceptor specificity. In summary, this study provides insight into the structural factors of regioselectivity and acceptor specificity in transfructosylation of GH68 enzymes.

Full text of DOI:10.1016/j.jbc.2021.100398

RESEARCH ARTICLE
Molecular insight into regioselectivity of transfructosylation
catalyzed by GH68 levansucrase and β-fructofuranosidase
Received for publication, December 1, 2020, and in revised form, February 2, 2021 Published, Papers in Press, February 8, 2021,
https://doi.org/10.1016/j.jbc.2021.100398
Masayuki Okuyama*,‡ , Ryo Serizawa‡, Masanari Tanuma, Asako Kikuchi, Juri Sadahiro, Takayoshi Tagami,
Weeranuch Lang, and Atsuo Kimura*
From the Research Faculty of Agriculture, Hokkaido University, Sapporo, Japan
Edited by Gerald Hart
Glycoside hydrolase family 68 (GH68) enzymes catalyze conformation blocked by a glucose unit at its reducing fruc-
β-fructosyltransfer from sucrose to another sucrose, the tosyl residue, thereby meaning “glucosides” (or “fructoside” for
so-called transfructosylation. Although regioselectivity of disaccharide). Natural nonreducing FOSs are mainly catego-
transfructosylation is divergent in GH68 enzymes, there is rized into inulooligosaccharide (INO; whose long saccharide is
insufficient information available on the structural factor(s) an inulin with less than around sixty β-(2→1)-linked fructosyl
involved in the selectivity. Here, we found two GH68 enzymes, β- units) and levanooligosaccharide (LVO; whose polysaccharide
fructofuranosidase (FFZm) and levansucrase (LSZm), encoded is a levan linked by β-(2→6)-fructosyl units). Also, natural
tandemly in the genome of Zymomonas mobilis, displayed reducing FOSs are present by cleaving a terminal Glc-Fru unit
different selectivity: FFZm catalyzed the β-(2→1)-trans- (sucrose unit) of INO or LVO with α-glucosidase and acidic
fructosylation (1-TF), whereas LSZm did both of 1-TF and β- condition. These natural FOSs are originally regarded to be
(2→6)-transfructosylation (6-TF). We identified His79FFZm and responsible for energy storage in plants. Other possible func-
Ala343FFZm and their corresponding Asn84LSZm and Ser345LSZm tions, including their contribution to the tolerance against
respectively as the structural factors for those regioselectivities. drought and freezing, are also proposed in plants and microbes
LSZm with the respective substitution of FFZm-type His and Ala (1). Moreover, their associated health benefits also receive
for its Asn84LSZm and Ser345LSZm (N84H/S345A-LSZm) lost 6- considerable attention, such as the selective stimulation of the
TF and enhanced 1-TF. Conversely, the LSZm-type replace- growth and/or the activity of beneficial intestinal bacteria,
ment of His79FFZm and Ala343FFZm in FFZm (H79N/A343S- contributing to the increase of the absorption of several min-
FFZm) almost lost 1-TF and acquired 6-TF. H79N/A343S-FFZm erals as well as the lowering of the cholesterol levels (2–4).
exhibited the selectivity like LSZm but did not produce the β-
Figure 1 depicts the structures of INO (e.g., 1-kestose and
(2→6)-fructoside-linked levan and/or long levanooligo- nystose), LVO (e.g., 6-kestose), levan, and neokestose [e.g., β-
saccharides that LSZm did. We assumed Phe189LSZm to be a fructosyl-(2→6)-glucosyl fructoside]. These nonreducing FOSs
responsible residue for the elongation of levan chain in LSZm and levan are produced from sucrose with the aid of fructosyl-
and mutated the corresponding Leu187FFZm in FFZm to Phe. An transferring enzymes. Syntheses of INOs and inulin in plants are
H79N/L187F/A343S-FFZm produced a higher quantity of long initiated by sucrose:sucrose 1-fructosyltransferase, resulting in 1-
levanooligosaccharides than H79N/A343S-FFZm (or H79N- kestose. Subsequently, 1-kestose is further elongated by the action
FFZm), although without levan formation, suggesting that LSZm of fructan:fructan 1-fructosyltransferase. In microorganisms, the
has another structural factor for levan production. We also inulosucrase is responsible for the inulin synthesis. β-Fructofur-
found that FFZm generated a sucrose analog, β-D-fructofur- anosidase (FF) is often used for the in vitro synthesis of INO,
anosyl α-D-mannopyranoside, by β-fructosyltransfer to including commercially available INO, such as the 1-kestose and
D-mannose and regarded His79FFZm and Ala343FFZm as key res- the nystose, which are produced by fungal FF (3). On the other
idues for this acceptor specificity. In summary, this study pro- hand, LVO and levan are well known to be produced by microbial
vides insight into the structural factors of regioselectivity and levansucrase (LS). Plants, including timothy (Phleum pratense),
acceptor specificity in transfructosylation of GH68 enzymes.
orchard grass (Dactylis glomerate), and big bluegrass (Poa
secunda), accumulate simple linear levan, of which timothy is
identified to be synthesized with sucrose:fructan 6-
fructosyltransferase (5, 6). 6-Kestose is also demonstrated to be
efficiently formed by Microbial FF from Schwanniomyces occi-
dentalis(7, 8). Furthermore, neokestose, neonystose, and the same
type of long oligosaccharides are synthesized by FF in Xantho-
phyllomyces dendrorhous (9), a mutated FF of S. occidentalis (10),
and a plant fructan:fructan 6G-fructosyltransferase (11, 12), in
which the term for the produced saccharides (e.g., neokestose and
neonystose) may become “neolevanooligosaccharides.”
The so-called fructooligosaccharides (FOSs) are structurally
divided into two types: the reducing FOSs and the nonre-
ducing FOSs. The former is composed of only fructose units
with reducing power, whereas the latter has the former
‡ These authors contributed equally to this work.
*
For correspondence: Masayuki Okuyama, okuyama@abs.agr.hokudai.ac.jp;
Atsuo Kimura, kimura@abs.agr.hokudai.ac.jp.
J. Biol. Chem. (2021) 296 100398 1
© 2021 THE AUTHORS. Published by Elsevier Inc on behalf of American Society for Biochemistry and Molecular Biology. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Transfructosylation regioselectivity of GH68 enzymes
OH
OH
OH
OH
HO
OH
OH
O
OH
HO
HO
OH
HO
OH
O
O
O
O
OH
HO
HO
HO
O
O
OH
OH
HO
HO
O
O
OH
OH
O
HO
HO
O
O
OH
O
OH
HO
OH
HO
OH
O
O
OH
1-kestose
6-kestose
neokestose
HO
OH
OH
OH
O
O
OH
OH
OH
OH
HO
HO
OH
OH
O
O
O
HO
HO
O
OH
HO
OH
O
HO
HO
O
O
HO
OH
HO
O
OH
HO
OH
OH
O
O
O
O
OH
n
HO
HO
nystose
levan
Figure 1. Structure of INO (1-kestose and nystose), LVO (6-kestose), levan, and neokestose.
Fructosyl-linkage hydrolyzing and transferring enzymes LSZm, respectively. The motivation for this study is to
acting on sucrose are classified as members of the glycoside compare the structure and function of these enzymes that have
hydrolase family (GH) 32 or 68 (13). Among them, plant- undergone molecular evolution while relatively retaining
derived enzymes and fungal FF generally belong to GH32, sequence identity and to clarify the structural factors that
whereas the inulosucrase, LS, and bacterial FF are classified as distinguish the reaction specificity of two enzymes that act on
GH68. GH32, and 68 proteins have a 5-bladed β-propeller fold sucrose. Relationship between the structure of LSZm and its
as a catalytic domain, grouped into clan GH-J. The enzymes in levan synthesis has been well-studied (25–27), while limited
this clan are known to show a catalytic double displacement enzymatic study has been done on FFZm, which might be due
mechanism with two carboxy groups, a catalytic nucleophile, to the lack of clear difference in regard of the molecular
and
a general acid/base, for hydrolysis and/or trans- mechanisms between the two GH68 enzymes in Z. mobilis. In
fructosylation. During the synthesis of inulin and levan, the this study, we found a distinction in regioselectivity of the
transfructosylation occurs successively.
FFZm- and LSZm-catalyzed transfructosylation as well as
identified the structural factors responsible for the regiose-
lectivity by mutation works. Until now, no studies have been
performed to our knowledge on the regulation of the regio-
selectivity of GH68 enzymes. Our mutation works also enabled
to learn one of important residues to elongate LVO chain. The
FFZm was found to generate β-D-fructofuranosyl α-D-man-
nopyranoside (Fru-Man; a sucrose analog) with a good yield,
which was firstly produced by Bacillus subtilis LS (LSBc) (28),
although in not high-yield formation.
Bacterial LS, especially LS from Gram-positive Bacillus spp.
(LSBspp), is one of the most characterized GH68 enzymes
(14–16). The molecular weight of the levan produced by these
LSBspps, reaching 3 × 106, can be controlled by the reaction
conditions and the mutagenesis of the enzyme. The protein
engineering works and other results are well reviewed by
Oritiz-Soto et al. (17). There are different types of LSs derived
from Gram-negative bacteria, a Gluconacetobacter diazo-
trophicus LS (LSGd) and an Erwinia amylovora LS (LSEa),
which synthesize a low amount of levan with the low molec-
ular weight (18–20). Interestingly, a certain amount of INO (1-
kestose and nystose) are also produced by these enzymes
(18, 20), which is assumed to be associated with the low levan
synthesis, although the detail mechanism is still obscure.
Results and discussion
Characterization and initial transfructosylation of the LSZm
and FFZm
Wild-type LSZm and FFZm had the specific activities of 247
and 1620 U mg−1, respectively (Table 1). The pH optimum for
both enzymes was at pH 5.0, and the pH stability was between
Gram-negative Zymomonas mobilis is known to grow on
sucrose, glucose, and fructose as a carbon source. Therefore, it
is possible that the enzymes associated with sucrose degrada-
tion of this bacterium have undergone its own molecular
evolution under the limitation of having to use sucrose as a
carbon source. Z. mobilis has two GH68 enzyme genes with
ZZ6_0874 and ZZ6_0875 locus tags, which are positioned side
by side in the chromosome, as if one of them was formed by a
tandem gene duplication. Gene duplication has been of long-
standing interest in the evolution as once a gene has dupli-
cated, a novel function may be acquired by one of the copies
(21). In fact, the functional evolution has occurred in these two
GH68 proteins: ZZ6_0874 and ZZ6_0875 encode FF and LS
(22–24), which hereafter will be referred to as FFZm and
ꢀ
pH 4.2 and 5.5 after treatment at 4 C for 24 h, as well as the
ꢀ
thermal stability was up to 45 C after a 15 min incubation.
ꢀ
Although the initial velocity can be measured at 45 C, the
following transfructosylation experiments were carried out at
ꢀ
35 C, where the activities were maintained even after pro-
longed reaction.
The LSZm- and FFZm-catalyzed initial 10 min reaction was
analyzed using various sucrose concentrations ([sucrose])
(Fig. 2, A and B, respectively), in which the formed carbohy-
drates were measured with high-performance anion-exchange
chromatography with pulsed amperometric detection
(HPAEC-PAD). The initial products were glucose, fructose,
2
J. Biol. Chem. (2021) 296 100398

Transfructosylation regioselectivity of GH68 enzymes
Table 1
Properties of FFZm, LSZm, and their variants
Enzyme
Specific activity (U mg−1
)
Optimum pH
pH-stabilitya
Thermal stability (ꢀC)b
FFZm
Wild type
H79N-FFZm
A343S-FFZm
H79N/A343S-FFZm
H79N/L187F/A343S-FFZm
1620
502
380
274
350
5.0
5.0
5.0
5.0
5.0
4.2‒6.0
4.1‒6.0
4.1‒6.0
4.1‒6.0
4.1‒6.0
<45
<40
<45
<45
<40
LSZm
Wild type
N84H-LSZm
S345A-LSZm
N84H/S345A-LSZm
247
480
800
440
5.0
5.5
5.0
5.5
4.1‒5.5
3.7‒5.5
3.7‒5.0
3.7‒5.5
<45
<60
<60
<60
apH stability or bthermal stability shows the range to maintain the residual activity of more than 90% a after treating with various pH at 4 ꢀC for 24 h or b after incubating at various
temperatures for 15 min, respectively.
1-kestose, and 6-kestose, at which the hydrolysis generates site, prevents the productive binding of the sucrose at −1
glucose and fructose from sucrose and water as well as the and +1 subsites (31). The occurrence of the nonproductive
transfructosylation forms trisaccharide (1-kestose or 6- binding increases at a higher [sucrose]. We cannot identify the
kestose) and glucose from two sucrose molecules. Therefore, correct mechanism among the two possibilities [or more
the initial reactions of both enzymes follow the scheme of correct further candidate(s) might exist, if the above two
Figure 2C, similarly to many glycosylases (28). Trans- possibilities are not accurate].
fructosylation products were different in LSZm and FFZm,
The proportion of the transfructosylation velocity in all
where 6-kestose and 1-kestose were synthesized by the former reaction velocities, which can be estimated from the equa-
(Fig. 2A), while 1-kestose with a trace amount of 6-kestose was tion of (v1-kestose + v6-kestoe)/vglucose, was 0.52 in LSZm and
synthesized by the latter (Fig. 2B). 1-Kestose (or 6-kestose) is 0.34 in FFZm using 1 M sucrose as substrate (Table 2).
produced from two-substrate reaction using two sucrose Yanase et al. (24) reported that the FFZm, purified from the
molecules as noted above, at which a fructosyl residue of periplasmic fraction of Z. mobilis, exhibited no trans-
donor sucrose is transferred to an acceptor sucrose to form the fructosylation activity, in regard of which the detailed
β-(2→1)-fructoside linkage [or β-(2→6)-fructoside linkage]. experimental conditions were missing. A certain reaction
The formation velocities of the transfructosylation products, 6- condition might be inadequate, suggesting that one factor is
kestose (v6-kestose [E]0−1; abbreviated as v6-Kes) and 1-kestose a less enough [sucrose].
(v1-kestose [E]0−1; v1-Kes), increased with an increase of [su-
crose] (Fig. 2, A and B). In the range of [sucrose] up to
−1
Transfructosylation of LSZm and FFZm at late reaction stage
300 mM, the formation velocities of glucose (vglucose [E]0
;
Glc) and fructose (vfructose [E]0−1; vFru) increased with showing
The LSZm- and FFZm-generating transfructosylation
products from 1 M sucrose were analyzed at 1 h (Fig. 3). The
wild-type LSZm synthesized 6-kestose, 1-kestose, and neo-
kestose (Fig. 3A). Neokestose, produced by β-(2→6)-trans-
fructosylation (6-TF) of a fructose unit to the glucosyl
residue of sucrose, was not detected during the initial reac-
tion, implying a slower formation than that of the 6-kestose
and 1-kestose. At 24 h, LSZm formed several oligosaccha-
rides (6-kestose, neokestose, and nystose) as well as some
possible LVOs with high degree of polymerizations (DPs)
(Fig. 3B). The long LVOs should have DPs at least 25 (inset
of Fig. 3B), although their exact DPs could not be estimated
due to unavailable standards. They might be the in-
termediates of low-molecular weight levan after the elon-
gation with successive 6-TF. On the other hand, the FFZm
generated a sole trisaccharide of 1-kestose at 1 h (Fig. 3C). At
24 h, further trisaccharides of the 6-kestose, neokestose, and
nystose were observed along with a small amount of the
products above DP4 (Fig. 3D). In cases of both initial and late
reaction stages, FFZm and LSZm show different regiose-
lectivity in the transfructosylation process (Figs. 2 and 3), at
which the former mainly catalyzes β-(2→1)-trans-
fructosylation (1-TF), whereas the latter does both of 1-TF
and 6-TF.
v
almost equal amounts (Fig. 2, A and B), meaning that the main
reaction is a hydrolysis. Moreover, as a [sucrose] increased,
vGlc and vFru decreased (especially vFru), meaning that the hy-
drolysis reduces and the transfructosylation increases. Ac-
cording to the scheme that is followed by the LSZm and the
FFZm (Fig. 2C), the velocity of hydrolysis is reflected by vFru
,
and that of transfructosylation is indicated by v6-Kes and/or v1-
Kes. The vGlc represents a sucrose degradation velocity,
meaning the whole reaction velocities, i.e., the sum of hydro-
lysis and transfructosylation. The decrease in vFru can be
ascribed to the increase in transfructosylation at the high
[sucrose]. On the other hand, the diminution in vGlc indicates
the substrate inhibition of enzyme reaction. There are two
possibilities that can explain this inhibitory phenomenon. First,
the accumulation of an E-F⋅F-G complex is assumed (Fig. 2C)
(29, 30). The concentration of the E-F⋅F-G complex ([E-F⋅F-
G]) increases with increasing [sucrose], eventually exceeding
the limit of the ability to catalyze the fructosidic linkage for-
mation, so that the accumulation of the E-F⋅F-G is expected to
occur. Since the [E⋅F-G], [E-F], and [E-F⋅F-G] are in steady
state, an increase in the [E-F⋅F-G] causes a decrease in the vGlc
.
Secondly, a nonproductive binding that, for instance, the
excess sucrose occupies the +1 and +2 subsites of the catalytic
J. Biol. Chem. (2021) 296 100398 3

Transfructosylation regioselectivity of GH68 enzymes
Prediction of residues responsible for regioselectivity of
transfructosylation
As mentioned above, the regioselectivity of trans-
fructosylation diverges at GH68 FFZm and LSZm, despite that
their amino-acid sequences share the significantly high 61.5%
identity and 72.0% similarity with only 5.1% gaps, leading us to
unravel the structural factor(s) in connection with the regu-
lation of regioselectivity distinction. However, the three-
dimensional structures (3Ds) of these enzymes remain unre-
solved currently, and it is hard to identify the amino acid
residues that are key to the differences in specificity. To
visualize the 3Ds of amino acid residues at the active-site
pocket of FFZm and LSZm, both enzymes were modeled
(Fig. 4) using the 3D of LSEa (32) as it had the most similarity
to the amino-acid sequence of the 3D-known GH68 proteins.
Pairwise alignments for the construction of model structures
were created with Needleman–Wunsch algorithm using
BLOSUM62 as the score matrix, and FFZm and LSZm showed
45.3% and 51.1% sequence identity with the template
sequence, respectively. Comparative modeling was done in
automodel mode of MODELLER. GA341 and zDOPE in the
respective models were 1.0 and −0.88 (FFZm) and 1.0
and −0.82 (LSZm), indicating a reliable model. Besides, the 1-
kestose and 6-kestose docked to FFZm and LSZm using
AutoDock Vina (33), respectively. The docking scores
were −7.5 (1-kestose to FFZm) and −7.7 (6-kestose to LSZm),
respectively. The residues forming −1 and +1 subsites were
well conserved, except for His79FFZm and Ala343FFZm
(Fig. 4A), corresponding to Asn84LSZm and Ser345LSZm
,
respectively (Fig. 4B). His79FFZm and Asn84LSZm are at an
intervening loop between βIB and βIC of the blade I in 5-
bladed β-propeller to participate in the +1 subsite. In the
model structure, His79FFZm possibly forms a hydrogen bond
with C6-OH of fructosyl residue on +1 subsite (Fig. 4A).
Asn84LSZm likely forms two hydrogen bonds with C1-OH of
fructosyl residues on the −1 and +1 subsites (Fig. 4B).
Furthermore, both of Ala343FFZm and Ser345LSZm, located on
the βⅤA of the blade Ⅴ, are in the inner part of active pocket,
spatially situating close to Asp44 and Asp48 catalytic nucleo-
phile residues, respectively (Fig. 4, A and B). We anticipated
that these divergences were involved in regioselectivity dif-
ference, allowing to generate six mutated enzymes: H79N-
FFZm, A343S-FFZm, H79N/A343S-FFZm, N84H-LSZm,
S345A-LSZm, and N84H/S345A-LSZm.
Characterization of mutated enzymes
Identical procedures used for wild-type FFZm and LSZm
were performed for characterizing their mutants (Table 1).
The specific activities of FFZm variants, measured using
50 mM sucrose, decreased to 31%‒17%, whereas those of
LSZm derivatives increased to 176%‒320%. Among these
findings, the latter phenomena are of interest since generally
mutated enzyme reduces specific activity. The original specific
activity of wild-type FFZm is greater than that of wild-type
LSZm, so that the other five mutants display similar values
of 270‒500 U mg−1. S345A-LSZm exhibits the highest value,
4
J. Biol. Chem. (2021) 296 100398

Transfructosylation regioselectivity of GH68 enzymes
1-kestose and 6-kestose in the initial reaction mixture using
1 M sucrose as substrate. The initial vGlc, vFru, v1-Kes, and v6-Kes
are summarized in Table 2. Two mutants, S345A-LSZm and
N84H/S345A-LSZm, almost maintained the whole reaction
velocities (vGlc), whereas N84H-LSZm decreased to 43% of the
parent enzyme. All variants of LSZm increased v1-Kes but
decreased v6-Kes. In comparison with wild type, v1-Kes increased
to 170% (S345A-LSZm) of parent value, 190% (N84H-LSZm),
and 510% (N84H/S345A-LSZm), whereas v6-Kes did 14%
(S345A-LSZm), 0.8% (N84H-LSZm), and 0.4% (N84H/S345A-
LSZm). An increase in the transfructosylation ratio of each
mutant can be explained by the enhancement of v1-Kes. Among
the mutants, N84H/S345A-LSZm, of which [v6-kestose/(v1-
A
B
C
+ v6-kestose)] value is substantially small (0.14%), is
kestose
found to be quite similar to FFZm in regard of its trisaccharide
production profile (Table 2).
Figure 3, A and B show the late reaction products of LSZm
mutants at 1 and 24 h, respectively. At 1 h, three variants
displayed the lack or a decreased production of 6-kestose due
to the reduction of v6-Kes at the initial reaction stage (Table 2),
in which N84H-LSZm maintained the less formation even at
24 h. While S345A-LSZm still produced neokestose similarly
to the wild type, N84H-LSZm and N84H/S345A-LSZm almost
lost the formation of this trisaccharide. Besides, all mutations
were observed to abolish the ability to generate large DP
products (insets of Fig. 3B), probably coincident with the
lowered capability to synthesize the β-(2→6)-fructosidic link-
age. S345A-LSZm could produce LVOs up to DP10 in trace
amounts by maintaining a limited ability to catalyze the suc-
cessive 6-TF. These results indicate that Asn84LSZm and
Ser345LSZm are involved in the 6-TF specificity of LSZm. The
regioselectivity can be switched from LSZm-type to FFZm-
type by the substitutions of Asn84LSZm and Ser345LSZm with
FFZm-type His and Ala, respectively.
Transfructosylation of FFZm mutants
Figure 2. Initial reaction products of LSZm and FFZm from 1 M sucrose.
A and B, plots of initial velocity versus [sucrose] for LSZm (A) and FFZm (B).
Reactions were performed for 10 min at 35 ꢀC and pH 5.0. HPAEC-PAD
measured concentrations of saccharides: [fructose] (open circle), [glucose]
(closed circle), [6-kestose] (open diamond), and [1-kestose] (closed diamond).
Each experiment was repeated three times, and the mean was plotted (error
bar, SD). C, reaction scheme. Free enzyme (E) binds to sucrose (F-G) to form
the Michealis-complex (E⋅F-G), and release glucose (G) to form the second
FFZm variants (H79N-FFZm, A343S-FFZm, H79N/A343S-
FFZm), which synthesized the same initial products as wild
type, decreased in every initial velocity [v1-Kes, vFru, and vGlc
(total reaction velocity)]. The mutants showed a decreased
transfructosylation ratio but an enhanced [v6-kestose/(v1-kestose
+
intermediate (E-F). The E-F either proceeds to hydrolysis to degrade into E v6-kestose)] value (Table 2). These results are attributed to the
and fructose (F) or to transfructosylation by accepting F–G to form the
substantial reduction in v1-Kes. Even H79N-FFZm, which
tertiary complex (E-F⋅F-G), which degrades into E and trisaccharide (F-F-G).
showed the highest value in the mutants, decreased to 5.7%
compared with that of the wild type. While v6-Kes decreased to
11% in A343S-FFZm, it increased slightly in H79N-FFZm and
H79N/A343S-FFZm. The H79N/A343S-FFZm, of which v1-Kes
decreased greatly, demonstrates the notable [v6-kestose/(v1-
F-F-G is 1-kestose or 6-kestose, depending on LSZm or FFZm.
800 U mg−1, alternating its catalytic efficiency. Moreover, all
mutants were similar to their parent enzymes in regard of their
optimum pH and pH-stability range. The thermal stability of
LSZm variants is higher than that of their wild type, whereas
FFZm derivatives maintain the heat stability of their parent
enzyme.
+ v6-kestose)] value (97%), and its regioselectivity almost
kestose
completely changes from 1-TF to 6-TF.
The late reaction of FFZm mutants at 1 and 24 h is shown in
Figure 3, C and D, respectively. At 1 h, H79N-FFZm and
A343S-FFZm synthesized both the 6-kestose and 1-kestose.
H79N/A343S-FFZm did 6-kestose as the almost sole trans-
Transfructosylation of LSZm mutants
The transfructosylation products of LSZm mutants were frucosylation product similarly to initial reaction, although the
identical to those of the wild type. The enzymes yielded synthesis of 6-kestose was not found in the wild type. At 24 h,
J. Biol. Chem. (2021) 296 100398 5

Transfructosylation regioselectivity of GH68 enzymes
A
C
D
B
Figure 3. Products at the late reaction stages of LSZm, FFZm, and their variants from 1 M sucrose. A and B, LSZm and its variants at 1 h (A) and 24 h (B).
C and D, FFZm and its variants at 1 h (C) and 24 h (D). Insets of B and D are the magnified figures to clearly display the long saccharides with a small amount.
F, fructose; G, glucose; S, sucrose; N, nystose; NK, neokestose; 1K, 1-kestose; 6K, 6-kestose.
6
J. Biol. Chem. (2021) 296 100398

Transfructosylation regioselectivity of GH68 enzymes
synthesize the 6-kestose. However, FFZm variants, gaining the
6-TF ability similarly to LSZm, do not efficiently generate long
LVOs with high DPs and levan, as compared with the parent
LSZm, which can do (inset of Fig. 3B). Only the H79N/A343S-
FFZm can produce an exceedingly small amount of LVOs
having DP7 or less (Fig. 6, described below).
A
C
D
L187*
1-kestose
+2
H294
R191
W113
1-kestose
E276
+1
R191
−1
+2
H79*
H294
Y342
I292
D192
D44
Functions of mutated residues
S114
+1
Our mutational study reveals that the position at the
His79FF or at its equivalent Asn84LS is involved in the regio-
selectivity for 1-TF of FFZm or 6-TF of LSZm, respectively.
The model structures suggest that His79FFZm and Asn84LSZm
form hydrogen bonds with the hydroxy groups of 1-kestose
and 6-kestose, respectively (Fig. 4, A and B). Introduction of
Asn at His79FFZm allows the formation of the hydrogen bond
suitable to produce 6-kestose. Contrarily, the replacement of
Asn84LSZm with His may enable to form the hydrogen bonds
suitable for 1-kestose synthesis. Besides, Ala343FF and
Ser345LS are also concerned in the regioselectivity of the
transfructosylation, although their mutations affect modestly
rather than those of His79FFZm and Asn84LSZm. It is hard to
interpret how the Ala343FFZm and Ser345LSZm contribute to
the selectivity because these residues cannot contact directly to
the substrate or product from each position, that is, the inner
part of catalytic pocket (Fig. 4, A and B). However, the mu-
tations of Ala343FFZm and Ser345LSZm are able to alter the
orientation of the side chains of adjacent amino acid residues,
such as Tyr342FFZm and Tyr344LSZm, both of which lie at
the +1 subsite. The alteration could modify the architecture of
the +1 subsite, influencing the transfructosylation process
containing regioselectivity.
A343*
W43
L67
F189
6-kestose
B
+2
H296
R193
+1
W118
6-kestose
+2
E278
Y344
R193
−1
H296
+1
N84*
I294
D194
D48 S119
S345*
W47
L71
E
T132
R191
D192
W113
N189
S114
Man-Fru
D44
R277
The relevant positions are conserved as a pair of His and Ala
or that of Asn and Ser in the UniRef50_F8DT27 cluster, to
which FFZm and LSZm belong. The enzymes having the
former pair are predicted to be FF, whereas other enzymes
having the latter pair are LS. On the other hand, the rela-
tionship between the residue pair and the regioselectivity of
transfructosylation is found to be not necessarily conserved in
GH68. Numerous GH68 enzymes derived from Gram-negative
bacteria have a pair of His and Ser, although atypical for FFZm
or LSZm. It has been demonstrated that LSGd and LSEa with
this pair produce 1-kestose and nystose in addition to levan
(18, 20). Their reaction specificities are more similar to that of
LSZm having a pair of Asn and Ser than that of A343SFFZm and
N79HLSZm, which possess a pair of His and Ser. A further
atypical example is the LS from Gram-positive bacteria, such
as LSBc and Bacillus megaterium LS (LSBm), which have Ser
at the second position of pair residue, corresponding to
Ala343FFZm and Ser345LSZm; however, neither LSBc nor LSBm
has a comparable residue, because the loop, carrying the
relevant residue, is short. These LSs have been reported to
possess another structural factor involved in the regiose-
lectivity, at which a substitution of Arg370LSBm with Gln is
observed to increase 1-kestose synthesis (34). These concrete
examples suggest a diversity of the relationship between
structure and specificity in GH68 enzymes.
E276
I292
Q274
W43
H79
A343
Y342
F355
H294
Figure 4. Structural models of FFZm and LSZm with ligands of 1-
kestose (A, C), 6-kestose (B, D), and Fru-Man (E). A and B, catalytic
pocket structures of FFZm (A) and LSZm (B) with 1-kestose and 6-kestose,
respectively. Predicted hydrogen bonds are represented by orange dashed
line between the mutation-candidate residues and ligands. C and D,
structures around Leu187FFZm and 1-kestose (C) as well as Phe189LSZm and
6-kestose (D). In panel A–D, amino acid residues with stick-model are pre-
sent within 3.5 Å from each ligand. Amino acid residues for mutagenesis are
indicated with an asterisk (A–C). E, structural models of FFZm with Fru-Man.
Amino acid residues within 4.0 Å from Fru-Man are depicted as sticks.
Predicted hydrogen bond is also shown by orange dashed line between
His79FF and 2-OH of mannosyl residue. For all panels, Modeller constructed
model structures of FFZm and LSZm. Autodock Vina docked 1-kestose or
Fru-Man to FFZm as well as 6-kestose, 1-kestose, or Fru-Man to LSZm.
the neokestose production was detected at H79N-FFZm and
A343S-FFZm but was hardly observed in the H79N/A343S-
FFZm. A little amount of tetrasaccharide nystose was pro-
duced by H79N-FFZm, but H79N/A343S-FFZm did not.
These results indicated that His79FFZm and Ala343FFZm are
involved in the regioselectivity of 1-TF of FFZm. The double
mutation using two LSZm-type residues enables FFZm to
J. Biol. Chem. (2021) 296 100398 7

Transfructosylation regioselectivity of GH68 enzymes
1000 mM
A
B
20
800 mM
600 mM
15
10
DP7
400 mM
200 mM
100 mM
50 mM
8
10
12
14
16
18
20
22
24
26
28
30
retention time (min)
DP5
Figure 5. TLC analysis of levan- and long LVO-formation by LSZm, FFZm
and H79N/A343S-FFZm. LSZm, FFZm and H79N/A343S-FFZm were incu-
bated with 1 M sucrose at pH 3.7, 5.0, 7.0, and 7.5. Reaction was performed
at 37 ꢀC for 24 h. A spot of levan and quite long LVO appears at original
position due to their large DP.
6
7
1000 mM
800 mM
600 mM
400 mM
200 mM
100 mM
50 mM
Levan synthesis using FFZm mutants
While H79N/A343S-FFZm can catalyze the 6-TF, this
mutant is devoid of synthesizing levan-type polysaccharides
(Fig. 3D). Goldman et al. demonstrated that the levan pro-
duction of Z. mobilis LS from ATCC10988 was related to its
protein form depending on the pH (27). Below pH 6.0, the
enzyme formed the well-ordered microfibril-shaped structure
and synthesized levan, allowing us to conduct the levan-
generation at pH 3.7–7.5 using H79N/A343S-FFZm
together with LSZm and FFZm as controls (LSZm as a pos-
itive control and FFZm as a negative control). The analysis
using thin-layer chromatography (TLC) showed that LSZm
produced levan and/or long LVOs at any pH, although such
saccharides were not generated by FFZm and H79N/A343S-
FFZm (Fig. 5). As shown by Goldman et al. (27), the levan
synthesis of LSZm is more efficient at a lower pH. The pH is
unrelated to the H79N/A343S-FFZm-associated levan for-
mation, suggesting that this mutant possesses further essen-
tial factor(s).
8
10
12
14
16
18
20
22
24
26
28
30
retention time (min)
C
DP5
6
1000 mM
800 mM
7
8
9
600 mM
400 mM
200 mM
100 mM
50 mM
The structural factors of LSBspp for levan synthesis are well-
studied (16, 34–37), mentioning that Tyr247LSBm, Asn252LSBm
,
8
10
12
14
16
18
20
22
24
26
28
30
Asp257LSBm
, ,
Arg370LSBm and Lys373LSBm (the residue
retention time (min)
numbers are of LSBm) were responsible for levan generation.
Figure 6. Long LVO formation of H79N/L187F/A343S-FFZm. A, LSZm; (B)
Among them, the position of Asn252LSBm is found to be H79N/A343S-FFZm; (C) H79N/L187F/A343S-FFZm. LSZm (A), H79N/A343S-
FFZm (B), and H79N/L187F/A343S-FFZm (C) were incubated with various
divergent at FFZm and LSZm. Other residues are conserved in
[sucrose] (50, 100, 200, 400, 600, 800, and 1000 mM) in 30 mM sodium
acetate buffer (pH 5.0) containing 0.01% Triton X-100 at 37 ꢀC for 24 h,
followed by analysis using HPAEC-PAD. Numbers on peaks are DP of their
saccharides.
FFZm (and also LSZm), probably without being involved in the
elongation by 6-TF. Asn252LSBm, locating at a surface loop to
connect blades II and III of 5-bladed β-propeller, is conserved
in LSBspp, although it varies in other GH68 enzymes. In FFZm
and LSZm, Leu187FFZm and Phe189LSZm are equivalent to
H79N/L187F/A343S-FFZm was generated using H79N/
Asn252LSBm (Fig. 4, C and D). In addition, Phe189LSZm seems A343S-FFZm as a parent enzyme to exam its ability in asso-
situated at a possible subsite far from –1 and +1 subsite of ciation with the levan synthesis by the substitution of
LSZm, based on its model structure. We thus anticipated that Leu187FFZm with Phe. The specific activity of H79N/L187F/
the divergence of Leu187FFZm and Phe189LSZm is concerned A343S-FFZm was 350 U mg−1, corresponding to 22% of FFZm
with the ability (LSZm) or inability (FFZm) of levan synthesis. and 128% of H79N/A343S-FFZm (Table 1). This triple mutant
8
J. Biol. Chem. (2021) 296 100398

Transfructosylation regioselectivity of GH68 enzymes
Table 3
below. Oligosaccharides that were further synthesized
included nystose, 1-kestose, and two kinds of unknown di-
saccharides, the structures of which were almost identified
with associated results appearing elsewhere. Using 0.5 M
sucrose as a donor substrate and 1.0 M mannose as an
acceptor substrate, we measured the initial velocity of Fru-
Man synthesis (vFru-Man [E]0−1; abbreviated as vFM) and the
yield of Fru-Man at 24 h of late reaction (Table 3). FFZm
shows by far the highest vFM, 105 s−1, whereas the vFM values
of FFZm mutants are 2.6–13% of FFZm vFM, and the vFM
values of LSZm and its mutants (except for N84H/S345A-
LSZm having 25%) are 3.4–8.4% of FFZm vFM. While LSZm
Fru-Man synthesis of FFZm, LSZm, and their variants
vFru-Man [E]0 (s−1
)
Yield (mM)
−1
Enzyme
FFZm
Wild type
H79N-FFZm
A343S-FFZm
H79N/A343S-FFZm
LSZm
Wild type
N84H-LSZm
S345A-LSZm
N84H/S345A-LSZm
105
14
7.2
2.7
60
18
29
16
3.6
8.8
6.1
26
20
24
19
30
showed similarity in optimum pH and pH-stability, while heat variants carry low vFM values, these are 1.7–7.2-times larger
stability was slightly decreased. The initial reaction velocities than wild-type LSZm does. Besides, the concentration of
(Table 2) indicated that the vGlc (meaning a total reaction Fru-Man produced at 24 h became 60 mM (Table 3), which
means that 12% of sucrose is converted to be Fru-Man. This
is the greatest yield among the enzymes tested, as other
velocity), v6-Kes, v1-Kes, as well as transfructosylation ratio
increased compared with those of H79N/A343S-FFZm. The
[v6-kestose/(v1-kestose + v6-kestose)] value was almost the same as enzymes generate 27–50% compared with that of FFZm.
that of H79N/A343S-FFZm, meaning that triple mutant
Seibel et al. (28) proposed that the LSBc-associated small
maintains 6-GT regioselectivity.
yield of Fru-Man was in association with the reduction of
Figure 6 shows the transfructosylation products of H79N/ catalytic efficiency due to the deprivation of the hydrogen
L187F/A343S-FFZm from 0.05 to 1.0 M sucrose at 24 h with bond between general acid/base catalyst (Glu342LSBc) and
H79N/A343S-FFZm and LSZm as controls. Wild-type LSZm axial 2-OH of mannose. Since the spatial position of the
general acid/base would be conserved in the GH68 mem-
appreciably synthesized long LVOs with DP20 or more from
bers, the Fru-Man-generating FFZm also forms this
400 mM sucrose or more (Fig. 6A), under which conditions we
observed a trace amount of long LVOs from 200 mM sucrose. hydrogen bond, suggesting that LSBc has unknown factor(s)
Almost no product above DP10 was formed from less than to reduce Fru-Man yield. It is assumed that a reason why
100 mM sucrose. In Figure 5, a couple of unknown oligosac- FFZm can synthesize Fru-Man is due to the His79FFZm to
charide spots under the position of the standard tetra- stabilize the Fru-Man synthetic reaction. The model struc-
saccharide nystose were exhibited by H79N/A343S-FFZm with ture of FFZm docking Fru-Man appears to depict the
a more significant intensity in comparison with that of wild- interaction between His79FFZm and the axial 2-OH of
mannose (Fig. 4E), indicative of the possible significance of
type FFZm observed at any pH, although the TLC analysis
could not clarify their DPs. The main oligosaccharide had DP5,
His79FFZm. The fact that Fru-Man syntheses are decreased
and those of others were up to DP7 (Fig. 6B). H79N/L187F/ by both H79N-FFZm and H79N/A343S-FFZm, in which
A343S-FFZm did not form any exceedingly long LVOs in His79FFZm is replaced, whereas the yields are inversely
contrast with LSZm (Fig. 6C). Instead, this triple mutant increased by N84H-LSZm and N84H/S345A-LSZm, which
formed the oligosaccharides of up to DP9. Mutation of newly acquire His79FFZm, supporting the proposed inter-
Leu187FFZm to Phe resulted in a slight increase in DP pretation. As previously mentioned, LSBspp does not have a
compared with that of its parent H79N/A343S-FFZm. The residue corresponding to His79FFZm. This absence might be
Leu187FFZm and Phe189LSZm corresponding to Asn252LSBm
one reason for LSBc to accept a small amount of mannose
into its catalytic site. On the other hand, it is difficult to
LSZm might have further unknown structural factor(s) for explain that A343S-FFZm and S345A-LSZm alternate the
are subtly involved in elongating fructosyl units. Wild-type
the elongation.
Fru-Man production (vFM and yield) in comparison with
their wild types, because, as mentioned earlier, Ala343FFZm
and Ser345LSZm do not directly contact with the substrate
sucrose. Similarly to the changes in regioselectivity, the
orientation of the adjacent Tyr residue, Tyr342FFZm and
Tyr344LSZm, may be responsible for the changes in Fru-Man
synthesis.
Transfructosylation with mannose as accepter
GH68 enzymes have been used for the production of
sucrose analogs (disaccharides) and glycosyl[glyco-
syl]nsucrose (n ≥ 0; DP 3 and over) by their trans-
fructosylation (28, 38, 39). Although most of these
saccharides can be produced in high yields using LSBc, the
sucrose analog Fru-Man has been quite difficult to be syn-
thesized (0.3% maximum yield) (28). We thus explored the
Conclusion
We have truly realized how important the analysis of
possible involvement of FFZm, LSZm, and their variants in the paralogous relationship in the field of protein engi-
neering through the present study. Generally, one of the
two paralogous proteins is thought to evolve indepen-
the formation of Fru-Man. It was found that each enzyme
was able to generate Fru-Man, as well as that the FFZm was
the most useful among all the used enzymes, as mentioned dently and uniquely because its gene has undergone
J. Biol. Chem. (2021) 296 100398 9

Transfructosylation regioselectivity of GH68 enzymes
ꢀ
relatively less restricted indels and substitutions. There- 30 C for 24 h. The cells bearing FFZm or LSZm were har-
ꢀ
fore, the study of paralogous proteins is equivalent to the vested by centrifugation (5000g, 5 min, 4 C), resuspended in
analysis of the results of protein engineering in nature and 20 mM sodium phosphate, 300 mM NaCl, pH 8.0 (buffer A),
often provides important insights into the structure– and disrupted by sonication. The supernatant was obtained by
ꢀ
function relationship of a protein. Based on this back- centrifugation (20,000g, 20 min, 4 C).
ground, we have identified His79FFZm and Ala343FFZm and
Recombinant proteins were purified from the above super-
their
respectively
corresponding
Asn84LSZm
and natant, loaded onto a Ni2+-chelating Sepharose Fast Flow
Ser345LSZm as the amino acid residues associated with the column (GE healthcare), which was prepared according to the
regioselectivity in transfructosylation. We also demon- instructions of manufacturer and pre-equilibrated with buffer
strated that Phe189LSZm is fairly involved in the elongation A. After washing the column with buffer A containing 20 mM
of the fructosyl residues in levan synthesis by analyzing an imidazole, the absorbed proteins were quickly eluted with
H79N/L187F/A343S-FFZm. In some cases, a gene that has 300 mM imidazole in buffer A, and then the pH of effluent
undergone the less restricted mutations can often lose its immediately changed to pH 6.0 for FFZm or to pH 5.0 for
function and become a pseudogene. However, in FFZm LSZm. Active fractions were passed through a Bio-Gel P-6
and LSZm, both enzymes were present with sufficient (Bio-Rad) column pre-equilibrated with 20 mM sodium
specific activity. This may be because Z. mobilis is only phosphate (pH 6.0) for FFZm or with 20 mM sodium acetate
able to utilize sucrose as a disaccharide, and the enzymes (pH 5.0) for LSZm. The obtained fractions were pooled and
engaged in sucrose degradation may be able to have concentrated using a 30 kDa MWCO Vivaspin 20 (Sartorius
evolved without losing their function.
AG). The concentration of purified enzyme ([enzyme]) was
estimated by amino acid analysis of protein hydrolysate (6 M
HCl, 110 ꢀC, 24 h) using an Amino Acid Analyzer L-8900
(Hitachi High-Tech) equipped with a ninhydrin-detection
system.
Experimental procedures
Gene cloning, production, and purification of FFZm and LSZm
Genes of ZZ6_0874 and ZZ6_0875 locus_tags, respectively
encoding FFZm and LSZm, were amplified from the chro-
mosomal DNA of Z. mobilis NBRC 13756 (ATCC 29191) by
PCR using primers: for ZZ6_0874, [50-CGCGAGCTCTT-
TAATTTTAATGC-30 (sense; SacI-site underlined) and 50-
GAGTCTAGATTATTTGCGACGAT-30 (antisense; XbaI-site
underlined)]; and for ZZ6_0875, [50-TCAACATGTTGAA-
TAAAGCAGGCATTGC-30 (sense; PciI-site underlined) and
50-AAACTCGAGTTTATTCAATAAGACAGGGCTG-30
(antisense; XhoI-site underlined)]. The primers were designed
based on the published whole genome sequence of Z. mobilis
ATCC 29191 (accession number, CP003704.1). After amplifi-
cation using PrimeSTAR MAX DNA polymerase (Takara Bio),
a PCR product for FFZm was restricted with SacI and XbaI
and cloned into appropriately digested pColdI DNA (Takara
Bio). The completed plasmid DNA was named pCold-FFZm. A
PCR product for LSZm was digested with PciI and XhoI and
ligated to pET23 days (+) DNA (Novagen-Merck Millipore)
predigested with NcoI and XhoI. The resulting plasmid DNA
was called pET23-LSZm.
Construction of model 3D protein structures
Model 3D structures were prepared using the Modeller (41),
Dock prep (31), and Autodock Vina (33) functions of UCSF
Chimera version 1.14 (http://www.rbvi.ucsf.edu/chimera) (42).
A template structure (PDB identifier, 4D47) was selected by
Sequence Navigator server (https://pdbj.org/seqNavi). All
model building was done with the default settings of the
software. The GHECOM 1.0 server (https://pdbj.org/ghecom/)
(43, 44) was used to detect the cavity in the docking simula-
tion. Figures showing 3D structures were prepared using open-
source PyMol (The PyMOL Molecular Graphic System version
2.4, Schrödinger, LLC). Sequence alignments were generated
with T-coffee Expresso (http://tcoffee.crg.cat/apps/tcoffee/do:
expresso) (45) or PROMALS3D (http://prodata.swmed.edu/
promals3d/promals3d.php) (46).
Site-directed mutagenesis
Point mutations were introduced using the PrimeSTAR
For the expression of FFZm, Escherichia coli BL21(DE3) mutagenesis basal kit (Takara Bio). The oligonucleotides used
cells harboring pCold-FFZm were cultured in Luria–Bertani for each mutation were as follows: for H79N-FFZm, [50-
medium (10 g L−1 Bacto tryptone, 5 g L−1 Bacto yeast GATCGCAATTCTCATGCCCGCATCGGC-30
(sense;
extract, and 5 g L−1 NaCl) supplemented with 0.1 g L−1 substituted codon in boldface) and 50-ATGAGAATT
ampicillin at 37 ꢀC until reaching mid-exponential phase GCGATCATTCCACGGAAT-30 (antisense)]; for A343S-
(OD600 = ꢁ0.4), followed by cooling on ice. To culture the FFZm,
[50-GCCTATTCCCATTACATCATGAATAAT-30
2500−1 volume of 0.5 M isopropyl β-thiogalactoside was added, (sense) and 50-GTAATGGGAATAGGCTTCGGTCGGCTG-
and cells were grown in a shaking incubator at 15 ꢀC for 24 h. 30 (antisense)]; for N84H-LSZm, [50-AATCGCCATGATGG
For the expression of LSZm, E. coli BL21(DE3) cells harboring CGCCAGAATTGGT-30 (sense) and 50- GCCATCATGGC-
pET23-LSZm were cultured by shaking in an auto-inducing GATTATGCCAACCATA-30 (antisense)]; for S345A-LSZm,
medium (40) [10 g L−1 Bacto tryptone, 5 g L−1 Bacto yeast [50-GCTTATGCTCATTATGTAATGACAAAT-30
(sense)
extract, 50 mM Na2HPO4, 50 mM KH2PO4, 25 mM and 50-ATAATGAGCATAAGCCTGATAAGGCTG-30 (anti-
(NH4)2SO4, 2 mM MgSO4, 5 g L−1 glycerol, 5 g L−1 glucose, sense)]; for L187F-FFZm, [50-AATAATTTCTGGAATTTCC
and 2 g L−1 lactose] supplemented with 0.1 g L−1 ampicillin at GTGATCCT-30 (sense) and 50-ATTCCAGAAATTATT
10 J. Biol. Chem. (2021) 296 100398

Transfructosylation regioselectivity of GH68 enzymes
TTCTGCATAATT-30 (antisense)]. The mutated enzymes for 6 min, followed by isocratic elution at 65.9:24.1:10 (%
were purified by the aforementioned procedures.
a:b:c) over 7 min. In the method B, the products were
separated by a linear gradient from 87.5:12.5:0 (% a:b:c) to
51.5:12.5:36 (% a:b:c) from for 25 min, followed by isocratic
elution at 51.5:12.5:36 (% a:b:c) over 5 min. Chromatograms
were evaluated with Chromeleon software (Dionex/Thermo
Fisher Scientific). The concentration of glucose, fructose, 1-
kestose, 6-kestose, or neokestose was calculated using a
calibration curve prepared based on the ration between
chromatographic peak area of sorbitol as an internal stan-
dard and that of concentration-known saccharide standard.
Neokestose was the kind gift from Prof Onodera S (Rakuno
Gakuen University, Ebetsu, Japan).
Enzyme assay and characterization
ꢀ
Enzyme activity was measured at 35 C in a standard re-
action mixture consisting of 50 mM sucrose (Nacalai Tesque),
30 mM sodium acetate buffer (pH 5.0), 0.01% Triton X-100,
and appropriately diluted enzyme solution. Reaction was
terminated by heating at 100 ꢀC for 3 min. After adjusting pH
to 7.0 by adding the double amount of 2 M Tris-HCl (pH 7.0),
the concentration of liberated glucose in the reaction mixture
was measured by a Glucose CII Test Wako (FUJIFILM Wako
Pure Chemical). One unit of enzyme activity (U) was defined
as the amount of enzyme that catalyzed the consumption of
1 μmol sucrose per minute under the above standard
conditions.
Initial transfructosylation
For the measurement of vGlc, vFru, v1-Kes, and v6-Kes, the
pH-Dependent reaction velocities for identifying optimal reaction was done by incubating FFZm (4.6 nM) or LSZm
pHs of FFZm, LSZm, and their variants were measured at (7.3 nM) with sucrose (0.025, 0.05, 0.1, 0.2, 0.3, 0.4, and 1.0 M)
50 mM sucrose in the McIlvaine buffer (pH 3.9–7.0) con- and 30 mM sodium acetate buffer (pH 5.0) containing 0.01%
ꢀ
ꢀ
taining 0.01% Triton X-100 at 35 C. The [enzyme] used was Triton X-100 at 35 C for 10 min, followed by terminating
ꢀ
specified for FFZm (2.1 nM), LSZm (2.0 nM), H79N-FFZm reaction using heat treatment at 100 C for 3 min. We also
(4.8 nM), A343S-FFZm (4.0 nM), H79N/A343S-FFZm measured vGlc, vFru, v1-Kes, and v6-Kes against 1.0 M sucrose
(19 nM), H79N/L187F/A343S-FFZm (4.2 nM), N84H-LSZm using FFZm, LSZm, and their variants ([enzyme], 30 to
(5.3 nM), S345A-LSZm (0.34 nM), and N84H/S345A-LSZm 800 nM) in the same way. The concentration of carbohydrates
(2.6 nM). pH-Stable region was evaluated as follows: the was determined by HPAEC-PAD using method A. For the
enzyme (FFZm [11 nM], LSZm [10 nM], H79N-FFZm measurement of vFru-Man, 0.5 M sucrose and 1.0 M mannose
[29 nM], A343S-FFZm [25 nM], H79N/A343S-FFZm [46 nM], were used as substrates, and the used [enzyme] was 2.2 μM
H79N/L187F/A343S-FFZm [52 nM], N84H-LSZm [27 nM], (FFZm), 16 μM (LSZm), 2.4 μM (H79N-FFZm), 5.8 μM
S345A-LSZm [9.0 nM], or N84H/S345A-LSZm [13 nM]) was (A343S-FFZm), 9.3 μM (H79N/A343S-FFZm), 2.7 μM (N84H-
incubated with McIlvaine buffer (pH 2.7–7.0) for 24 h at 4 ꢀC, LSZm), 6.8 μM (S345A-LSZm), or 2.6 μM (N84H/S345A-
and then the residual activity was determined at pH 5.0 ac- LSZm). Other experimental conditions were the same as those
cording to the standard procedure described previously. To described above.
estimate the thermal stability of the enzymes, we incubated the
enzyme (FFZm [11 nM], LSZm [10 nM], H79N-FFZm Late transfructosylation at 1 h and 24 h
[36 nM], A343S-FFZm [12 nM], H79N/A343S-FFZm [28 nM],
H79N/L187F/A343S-FFZm [32 nM], N84H-LSZm [16 nM],
fructosylation of FFZm, LSZm, and their variants using 1.0 M
S345A-LSZm [5.1 nM], or N84H/S345A-LSZm [7.8 nM])
For the qualitative analysis of the products of the trans-
sucrose, the reaction was performed at 35 ꢀC and pH 5.0 for up
to 24 h in 30 mM sodium acetate buffer (pH 5.0). The
[enzyme] was 0.03 μM (FFZm), 0.34 μM (LSZm), 0.13 μM
(H79N-FFZm), 1.3 μM (A343S-FFZm), 0.64 μM (H79N/
A343S-FFZm), 0.44 μM (H79N/L187F/A343S-FFZm),
0.38 μM (N84H-LSZm), 0.24 μM (S345A-LSZm), or 0.26 μM
(N84H/S345A-LSZm). At 1 and 24 h,ꢀa part of the reaction
containing 33 mM sodium acetate buffer (pH 5.0) and 0.01%
ꢀ
Triton X-100 at 30–70 C for 15 min, followed by measure-
ment of the residual activity under the standard conditions
described above.
HPAEC-PAD analysis
HPAEC-PAD was performed with a Dionex ICS-3000 mixture was recovered, heated at 100 C for 3 min to termi-
system (Dionex/Thermo Fisher Scientific) using an analyt- nate the reaction, and diluted to 100−1 with 10 mM sorbitol of
ical column (CarboPac PA1, 4 mm I.D. × 250 mm; Dionex/ internal standard. The reaction products were analyzed by
Thermo Fisher Scientific) and a flow rate of 0.8 ml min−1
.
HPAEC-PAD. Resultant reaction samples at 1 h and 24 h were
The eluent system comprised MilliQ water (a), 0.8 M NaOH analyzed by methods A and B, respectively.
prepared from super special grade 50% NaOH solution
For investigating the changes in the long LVO production
(FUJIFILM Wako Pure Chemical Co) (b), and 1.0 M sodium of LSZm, H79N/A343S-FFZm and H79N/L187F/A343S-
acetate (c), in which two procedures were used, namely FFZm, we incubated each enzyme with different [sucrose]
method A and method B, as described below. In the method bearing 0.05, 0.1, 0.2, 0.4, 0.6, 0.8, or 1 M at 35 ꢀC and pH 5.0
A, the products were separated by the elution system for 24 h in 40 mMꢀsodium acetate buffer (pH 5.0), followed
comprising the first linear gradient from 97.2:2.8:0 (% a:b:c) by heating at 100 C for 3 min. The reaction mixture was
to 83.8:16.2:0 (% a:b:c) for 10 min and the second linear diluted with sorbitol solution and analyzed by HPAEC-PAD
gradient from 83.8:16.2:0 (% a:b:c) to 65.9:24.1:10 (% a:b:c) using method B.
J. Biol. Chem. (2021) 296 100398 11

Transfructosylation regioselectivity of GH68 enzymes
For the Fru-Man synthesis, the reaction of FFZm (2.2 μM), (126 MHz, D2O) δ (ppm): 106.6 (C-20, C-200), 95.0 (C-1), 84.0
LSZm (16 μM), H79N-FFZm (2.4 μM), A343S-FFZm (5.8 μM), (C-500), 83.1 (C-50), 79.3 (C-300), 79.0 (C-30), 77.5 (C-400), 77.2
H79N/A343S-FFZm (9.3 μM), N84H-LSZm (2.7 μM), S345A- (C-40), 75.4 (C-3), 75.2 (C-5), 73.9 (C-2), 72.1 (C-4), 65.8 (C-
LSZm (6.8 μM), or N84H/S345A-LSZm (2.6 μM) was per- 60), 65.4 (C-600), 64.0 (C-10), 63.1 (C-6), 62.6 (C-100); MS: m/z
formed with 0.5 M sucrose and 1.0 M mannose at 35 C and calculated: 527.16 [M+Na]+, 503.16 [M − H]−; found 527.16,
ꢀ
pH 5.0 for 24 h, and was subsequently terminated by heating at 503.16. Fru-Man: 13C-NMR (126 MHz, D2O) δ (ppm): 104.8
ꢀ
100 C for 3 min. A sample for HPAEC-PAD analysis using (C-20), 94.4 (C-1), 82.2 (C-50), 76.8 (C-30), 74.7 (C-40), 74.1
method A was prepared by 100−1-dilution of reaction mixture (C-2), 71.9 (C-3), 71.0 (C-5), 67.3 (C-4), 63.3 (C-60), 61.8 (C-
with 10 mM sorbitol.
10), 61.5 (C-6); MS: m/z calculated: 365.11[M+Na]+, 341.11
[M − H]−; found 365.11, 341.11.
pH-Influence on production of long LVO and levan
Data availability
For analyzing the synthesis of long LVO and levan at the
different pHs, we incubated H79N/A343S-FFZm, FFZm, or
The authors indicate that all of the data is contained within
LSZm (5 U mL−1 for each enzyme) with 1.0 M sucroseꢀ in the article.
McIlvaine buffer (pH 3.7, 5.0, 7.0, and 7.5) for 24 h at 35 C.
After heat treatment at 100 ꢀC for 3 min, the reaction mixture
Acknowledgments—We would like to thank Prof Onodera S
was appropriately diluted in deionized water and analyzed by
TLC using a TLC-plate (0.25-mm layers of silica gel 60 F254
(Rakuno Gakuen University) for kindly gifting us neokestose. We
also thank the staff of the Instrumental Analysis Division of the
Creative Research Institution at Hokkaido University for the con-
duction of amino acid analysis and mass spectrometry analysis, and
the staff of the GC-MS & NMR Laboratory in the Research Faculty
of Agriculture at Hokkaido University for NMR analysis.
;
Merck Millipore). The separation of carbohydrates was done
using acetonitrile:water (75:25, v/v) as the developing solvent
by three-times ascent, followed by visualization with spraying
an α-naphthol/sulfuric acid solution (15% sulfuric acid in
methanol containing 2.1 mM α-naphthol) and heating at
ꢀ
Author contributions—M. O., T. T., and A. K. conceived and co-
ordinated the study. M. O., R. S., M. T., A. K., J. S., and W. L.
performed the experiments and analyzed data. M. O. and A. K.
created the manuscript. M. O. and A. K. provided essential equip-
ment and facilities. M. O. and R. S. contributed equally. All the
authors reviewed the results and approved the final version of this
110 C for 5 min.
Production of 6-kestose and Fru-Man
This research required the preparation of an appropriate
amount of standard 6-kestose. For this, we did the 10 ml-
level reaction of 37 U mL−1 H79N/A343S-FFZm with article.
1.0 M sucrose at 35 ꢀC and pH 5.0 for 2 h. Reaction solution
Funding and additional information—This work was partly sup-
ported by the Japan Society for the Promotion of Science KAKENHI
Grant Nos. 26450114 and 17H03801.
was desalted by Amberlite MB-4 resin (Organo), and then
concentrated to 2 ml under reduced pressure. The reaction
products were separated by an HPLC system (Jasco) equip-
ped with a refractive index monitor (Jasco RI-2031) with a
Cosmosil Sugar-D column (10 mm I.D. × 250 mm; Nacalai
Tesque). Separation was achieved using a mobile phase
Conflict of interest—The authors declare that they have no conflicts
of interest with the contents of this article.
(acetonitrile:water, 80:20, v/v) with a flow rate of 3 ml min−1
.
Abbreviations—The abbreviations used are: DP, degree of poly-
merization; EaLsc, levansucrase from Erwinia amylovora; FF,
β-fructofuranosidase; FFZm, Zymomonas mobilis FF; FOS, fruc-
tooligosaccharides; Fru-Man, β-D-fructofuranosyl α-D-mannopyr-
anoside; GH, glycoside hydrolase family; HPAEC-PAD, high-
performance anion-exchange chromatography with pulsed amper-
ometric detection; INO, inulooligosaccharide; LS, levansucrase;
LSBc, Bacillus subtilis LS; LSBspp, Bacillus spp. LS; LSBm, Bacillus
megaterium LS; LSEa, Erwinia amylovora LS; LSGd, Gluconaceto-
For Fru-Man synthesis, 1.25 M sucrose and 2.5 M mannose
with 25 U mL−1 FFZm in 10 ml of 40 mM sodium acetate
buffer were incubated at 35 ꢀC for 6 h. The reaction mixture
was divided into three equal parts and was fractionated by
gel filtration chromatography using Bio-Gel P-2 Gel (1.6 cm
I.D. × 190 cm, Bio-Rad). Fractions containing Fru-Man were
recovered and concentrated under reduced pressure.
Reducing sugars wꢀere decomposed by the incubation with bacter diazotrophicus LS; LSZm, Zymomonas mobilis LS; LVO,
levanooligosaccharide; TLC, thin-layer chromatography; vFM, vFru-
2 M NaOH at 100 C for 10 min. The reaction solution was
neutralized by adding Amberlite IR120 H+ form and decol-
orized by activated carbon. The resulting samples were again
subjected to gel filtration chromatography and purified Fru-
Man was obtained. The molecular masses of oligosaccharide
products were confirmed by ESI-MS using a Thermo Sci-
entific Exactive spectrometer (Thermo Fisher Scientific, Inc).
13C-NMR (126 MHz) spectra were recorded on a Bruker
AVANCE I (Bruker). The relative values of the chemical
shifts for 6-kestose and Fru-Man were in agreement with
those previously reported (47, 48). 6-Kestose: 13C-NMR
[E]0−1; vFru, vfructose [E]0−1; vGlc, vglucose [E]0−1; v1-Kes, v1-kestose
Man
[E]0−1; v6-Kes, v6-kestose [E]0−1; 3D, three-dimensional structure; 1-
TF, β-(2→1)-transfructosylation; 6-TF, β-(2→6)-transfructosylation.
References
1. Versluys, M., Kirtel, O., Öner, E. T., and Van den Ende, W. (2018) The
fructan syndrome: Evolutionary aspects and common themes among
plants and microbes. Plant Cell Environ. 41, 16–38
2. Belorkar, S. A., and Gupta, A. K. (2016) Oligosaccharides: A boon from
nature’s desk. AMB Express 6, 82
12 J. Biol. Chem. (2021) 296 100398

Transfructosylation regioselectivity of GH68 enzymes
3. Tochio, T., Kadota, Y., Tanaka, T., and Koga, Y. (2018) 1-Kestose, the
smallest fructooligosaccharide component, which efficiently stimulates
Faecalibacterium prausnitzii as well as bifidobacteria in humans. Foods 7,
9
19. Martínez-Fleites, C., Ortíz-Lombardía, M., Pons, T., Tarbouriech, N.,
Taylor, E. J., Arrieta, J. G., Hernández, L., and Davies, G. J. (2005) Crystal
structure of levansucrase from the Gram-negative bacterium Glucona-
cetobacter diazotrophicus. Biochem. J. 390, 19–27
4. Scholz-Ahrens, K. E., and Schrezenmeir, J. (2007) Inulin and oligofructose 20. Caputi, L., Nepogodiev, S. A., Malnoy, M., Rejzek, M., Field, R. A., and
and mineral metabolism: The evidence from animal trials. J. Nutr. 137,
2513S–2523S
5. Cairns, A. J., Nash, R., De Carvalho, M. A. M., and Sims, I. M. (1999)
Characterization of the enzymatic polymerization of 2,6-linked fructan by
leaf extracts from timothy grass (Phleum pratense). New Phytol. 142, 79–
91
Benini, S. (2013) Biomolecular characterization of the levansucrase of
Erwinia amylovora, a promising biocatalyst for the synthesis of fructoo-
ligosaccharides. J. Agric. Food Chem. 61, 12265–12273
21. Loehlin, D. W., and Carroll, S. B. (2016) Expression of tandem gene
duplicates is often greater than twofold. Proc. Natl. Acad. Sci. U. S. A. 113,
5988–5992
6. Tamura, K., Kawakami, A., Sanada, Y., Tase, K., Komatsu, T., and
Yoshida, M. (2009) Cloning and functional analysis of a fructosyl-
transferase cDNA for synthesis of highly polymerized levans in timothy
(Phleum pratense L.). J. Exp. Bot. 60, 893–905
22. Kyono, K., Yanase, H., Tonomura, K., Kawasaki, H., and Sakai, T. (1995)
Cloning and characterization of Zymomonas mobilis genes encoding
extracellular levansucrase and invertase. Biosci. Biotechnol. Biochem. 59,
289–293
7. Rodrigo-Frutos, D., Piedrabuena, D., Sanz-Aparicio, J., and Fernández-
Lobato, M. (2019) Yeast cultures expressing the Ffase from Schwannio-
myces occidentalis, a simple system to produce the potential prebiotic
sugar 6-kestose. Appl. Microbiol. Biotechnol. 103, 279–289
8. de Abreu, M., Alvaro-Benito, M., Sanz-Aparicio, J., Plou, F. J., Fernandez-
Lobato, M., and Alcalde, M. (2013) Synthesis of 6-kestose using an effi-
cient β-fructofuranosidase engineered by directed evolution. Adv. Synth.
Catal. 355, 1698–1702
23. Yanase, H., Iwata, M., Nakahigashi, R., Kita, K., Kato, N., and Tonomura,
K. (1992) Purification, crystallization, and properties of the extracellular
levansucrase from Zymomonas mobilis. Biosci. Biotechnol. Biochem. 56,
1335–1337
24. Yanase, H., Iwata, M., Kita, K., Kato, N., and Tonomura, K. (1995) Pu-
rification, crystallization, and characterization of the extracellular inver-
tase from Zymomonas mobilis. J. Ferment. Bioeng. 79, 367–369
25. Yanase, H., Maeda, M., Hagiwara, E., Yagi, H., Taniguchi, K., and Oka-
moto, K. (2002) Identification of functionally important amino acid res-
idues in Zymomonas mobilis levansucrase. J. Biochem. 132, 565–572
9. Linde, D., Rodríguez-Colinas, B., Estévez, M., Poveda, A., Plou, F. J., and
Fernández Lobato, M. (2012) Analysis of neofructooligosaccharides pro-
duction mediated by the extracellular β-fructofuranosidase from Xan- 26. Li, S., Yan, Y., Zhou, Z., Yu, H., Zhan, Y., Zhang, W., Chen, M., Lu, W.,
thophyllomyces dendrorhous. Bioresour. Technol. 109, 123–130
10. Álvaro-Benito, M., A. Sainz-Polo, M., González-Pérez, D., González, B.,
Plou, F. J., Fernández-Lobato, M., and Sanz-Aparicio, J. (2012) Structural
Ping, S., and Lin, M. (2011) Single amino acid residue changes in subsite-1
of levansucrase from Zymomonas mobilis 10232 strongly influence the
enzyme activities and products. Mol. Biol. Rep. 38, 2437–2443
and kinetic insights reveal that the amino acid pair Gln-228/Asn-254 27. Goldman, D., Lavid, N., Schwartz, A., Shoham, G., Danino, D., and
modulates the transfructosylating specificity of Schwanniomyces occi-
dentalis β-fructofuranosidase, an enzyme that produces prebiotics. J. Biol.
Chem. 287, 19674–19686
Shoham, Y. (2008) Two active forms of Zymomonas mobilis levansucrase:
An ordered microfibril structure of the enzyme promotes levan poly-
merization. J. Biol. Chem. 283, 32209–32217
11. Ueno, K., Onodera, S., Kawakami, A., Yoshida, M., and Shiomi, N. (2005)
Molecular characterization and expression of a cDNA encoding fructan:
fructan 6(G)-fructosyltransferase from asparagus (Asparagus officinalis).
New Phytol. 165, 813–824
28. Seibel, J., Moraru, R., Götze, S., Buchholz, K., Na’amnieh, S., Pawlowski,
A., and Hecht, H. J. (2006) Synthesis of sucrose analogues and the
mechanism of action of Bacillus subtilis fructosyltransferase (levansu-
crase). Carbohydr. Res. 341, 2335–2349
12. Lasseur, B., Lothier, J., Djoumad, A., De Coninck, B., Smeekens, S., Van 29. Brumer, H., Sims, P. F., and Sinnott, M. L. (1999) Lignocellulose degra-
Laere, A., Morvan-Bertrand, A., Van den Ende, W., and Prud’homme,
M.-P. (2006) Molecular and functional characterization of a cDNA
dation by Phanerochaete chrysosporium: Purification and characterization
of the main α-galactosidase. Biochem. J. 339, 43–53
encoding fructan:fructan 6G-fructosyltransferase (6G-FFT)/fructan:fruc- 30. Liao, J., Okuyama, M., Ishihara, K., Yamori, Y., Iki, S., Tagami, T., Mori,
tan 1-fructosyltransferase (1-FFT) from perennial ryegrass (Lolium per-
enne L.). J. Exp. Bot. 57, 2719–2734
13. Cantarel, B. L., Coutinho, P. M., Rancurel, C., Bernard, T., Lombard, V.,
H., Chiba, S., and Kimura, A. (2016) Kinetic properties and substrate
inhibition of α-galactosidase from Aspergillus niger. Biosci. Biotechnol.
Biochem. 80, 1747–1752
and Henrissat, B. (2009) The Carbohydrate-Active EnZymes database 31. Dolinsky, T. J., Nielsen, J. E., McCammon, J. A., and Baker, N. A. (2004)
(CAZy): An expert resource for Glycogenomics. Nucleic Acids Res. 37,
D233–D238
14. Porras-Domínguez, J. R., Ávila-Fernández, Á., Miranda-Molina, A.,
Rodríguez-Alegría, M. E., and Munguía, A. L. (2015) Bacillus subtilis 168
levansucrase (SacB) activity affects average levan molecular weight.
Carbohydr. Polym. 132, 338–344
PDB2PQR: An automated pipeline for the setup of Poisson-Boltzmann
electrostatics calculations. Nucleic Acids Res. 32, W665–W667
32. Wuerges, J., Caputi, L., Cianci, M., Boivin, S., Meijers, R., and Benini, S.
(2015) The crystal structure of Erwinia amylovora levansucrase provides
a snapshot of the products of sucrose hydrolysis trapped into the active
site. J. Struct. Biol. 191, 290–298
15. Raga-Carbajal, E., López-Munguía, A., Alvarez, L., and Olvera, C. (2018)
Understanding the transfer reaction network behind the non-processive
synthesis of low molecular weight levan catalyzed by Bacillus subtilis
levansucrase. Sci. Rep. 8, 15035
33. Trott, O., and Olson, A. J. (2010) Software news and update AutoDock
Vina: Improving the speed and accuracy of docking with a new scoring
function, efficient optimization, and multithreading. J. Comput. Chem. 31,
455–461
16. Homann, A., Biedendieck, R., Götze, S., Jahn, D., and Seibel, J. (2007)
Insights into polymer versus oligosaccharide synthesis: Mutagenesis and
mechanistic studies of a novel levansucrase from Bacillus megaterium.
Biochem. J. 407, 189–198
34. Possiel, C., Ortiz-Soto, M. E., Ertl, J., Münch, A., Vogel, A., Schmiedel, R.,
and Seibel, J. (2019) Exploring the sequence variability of polymerization-
involved residues in the production of levan- and inulin-type fructooli-
gosaccharides with a levansucrase. Sci. Rep. 9, 1–11
17. Ortiz-Soto, M. E., Porras-Domínguez, J. R., Seibel, J., and López-Mun-
guía, A. (2019) A close look at the structural features and reaction con-
ditions that modulate the synthesis of low and high molecular weight
fructans by levansucrases. Carbohydr. Polym. 219, 130–142
35. Ortiz-Soto, M. E., Rivera, M., Rudiño-Piñera, E., Olvera, C., and López-
Munguía, A. (2008) Selected mutations in Bacillus subtilis levansucrase
semi-conserved regions affecting its biochemical properties. Protein Eng.
Des. Sel 21, 589–595
18. Hernandez, L., Arrieta, J., Menendez, J., Vazquez, R., Coego, A., Suarez, 36. Strube, C. P., Homann, A., Gamer, M., Jahn, D., Seibel, J., and Heinz, D.
V., Selman, G., Petit-Glatron, M. F., and Chambert, R. (1995) Isolation
and enzymic properties of levansucrase secreted by Acetobacter diazo-
trophicus SRT4, a bacterium associated with sugar cane. Biochem. J. 309,
113–118
W. (2011) Polysaccharide synthesis of the levansucrase SacB from Ba-
cillus megaterium is controlled by distinct surface motifs. J. Biol. Chem.
286, 17593–17600
J. Biol. Chem. (2021) 296 100398 13

Transfructosylation regioselectivity of GH68 enzymes
37. He, C., Yang, Y., Zhao, R., Qu, J., Jin, L., Lu, L., Xu, L., and Xiao, M. (2018)
Rational designed mutagenesis of levansucrase from Bacilluslicheniformis8-37-
0-1 for product specificity study. Appl. Microbiol. Biotechnol. 102, 3217–3228
38. Wu, C., Zhang, T., Mu, W., Miao, M., and Jiang, B. (2015) Biosynthesis of
lactosylfructoside by an intracellular levansucrase from Bacillus methyl-
otrophicus SK 21.002. Carbohydr. Res. 401, 122–126
43. Kawabata, T., and Go, N. (2007) Detection of pockets on protein surfaces
using small and large probe spheres to find putative ligand binding sites.
Proteins 68, 516–529
44. Kawabata, T. (2010) Detection of multiscale pockets on protein surfaces
using mathematical morphology. Proteins 78, 1195–1211
45. Notredame, C., Higgins, D. G., and Heringa, J. (2000) T-coffee: A novel
method for fast and accurate multiple sequence alignment. J. Mol. Biol.
302, 205–217
39. Win, T. T., Isono, N., Kusnadi, Y., Watanabe, K., Obae, K., Ito, H., and
Matsui, H. (2004) Enzymatic synthesis of two novel non-reducing oli-
gosaccharides using transfructosylation activity with β-fructofuranosidase 46. Pei, J., Kim, B.-H., and Grishin, N. V. (2008) PROMALS3D: A tool for
from Arthrobacter globiformis. Biotechnol. Lett. 26, 499–503
40. Studier, F. W. (2005) Protein production by auto-induction in high
density shaking cultures. Protein Expr. Purif. 41, 207–234
multiple protein sequence and structure alignments. Nucleic Acids Res.
36, 2295–2300
47. Liu, J., Waterhouse, A. L., and Chatterton, N. J. (1991) Proton and carbon
chemical-shift assignments for 6-kestose and neokestose from two-
dimensional n.m.r. measurements. Carbohydr. Res. 217, 43–49
ꢀ
41. Sali, A., and Blundell, T. L. (1993) Comparative protein modeling by
satisfaction of spatial restraints. J. Mol. Biol. 234, 779–815
42. Pettersen, E. F., Goddard,T. D., Huang, C.C., Couch, G. S., Greenblatt, D. M., 48. Seibel, J., Moraru, R., and Götze, S. (2005) Biocatalytic and chemical
Meng, E. C., and Ferrin, T. E. (2004) UCSF Chimera-A visualization system
for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612
investigations in the synthesis of sucrose analogues. Tetrahedron 61,
7081–7086
14 J. Biol. Chem. (2021) 296 100398