Structural analysis of the reducing‐end xylose‐releasing exo‐oligoxylanase Rex8A from Paenibacillus barcinonensis BP‐23 deciphers its molecular specificity

Abstract

Reducing‐end xylose‐releasing exo‐oligoxylanases (Rex) are GH8 enzymes that depolymerize xylooligosaccharides complementing xylan degradation by endoxylanases in an exo manner. We have studied Paenibacillus barcinonensis Rex8A and showed the release of xylose from xylooligomers decorated with methylglucuronic acid (UXOS) or with arabinose (AXOS). This gives the enzyme a distinctive trait among known Rex, which show activity only on linear xylooligosaccharides. The structure of the enzyme has been solved by X‐ray crystallography showing a (α/α)6 folding common to GH8 enzymes. Analysis of inactived Rex8A‐E70A complexed with xylotetraose revealed the existence of at least four binding subsites in Rex8A, with the oligosaccharide occupying subsites −3 to +1. The enzyme shows an extended Leu320‐His321‐Pro322 loop, common to other Rex, which blocks the binding of longer substrates to positive subsites further than +1 and seems responsible for the lack or diminished activity of Rex enzymes on xylan. Mutants with smaller residues in this loop failed to increase Rex8A activity on the polymer. Analysis of the complexes with AXOS showed the accommodation of arabinose at subsite −2, which cannot be allocated at subsite −1. Arabinose substitutions at the xylose O2 or O3 are accommodated by hydrophobic interaction and seem tolerated rather than recognized by Rex8A. A strained binding of the branch is facilitated by the lack of direct polar interactions of the xylose occupying this subsite, its water‐mediated links allowing some conformational flexibility of the sugar. The plasticity of Rex8A is a notable property of the enzyme for its application in xylan deconstruction and upgrading.

Database
Structural data are available in PDB database under the accession numbers 6SRD (native form), 6TPP (E70A mutant in complex with EDO), 6TOW (E70A in complex with Xyl4), 6SUD (L320A mutant in complex with xylose), 6SHY (L320A/H321S double mutant in complex with EDO), 6TO0 (E70A in complex with AX3), and 6TRH (E70A in complex with AX4).

Full text

Structural analysis of the reducing-end xylose-releasing
exo-oligoxylanase Rex8A from Paenibacillus barcinonensis
BP-23 deciphers its molecular specificity
Elena Jim
enez-Ortega
1
, Susana Valenzuela
2,3
, Mercedes Ram
ırez-Escudero
1
,
Francisco Javier Pastor
2,3
and Julia Sanz-Aparicio
1
1 Macromolecular Crystallography and Structural Biology Department, Institute of Physical-Chemistry ‘Rocasolano’, CSIC, Madrid, Spain
2 Department of Microbiology, Faculty of Biology, University of Barcelona, Spain
3 Institute of Nanoscience and Nanotechnology (IN2UB), University of Barcelona, Spain
Keywords
decorated xylan; GH8 specificity; reducing-
end xylose-releasing exo-oligoxylanase;
xylanase structure
Correspondence
J. Sanz-Aparicio, Department of
Crystallography and Structural Biology,
Institute of Physical-Chemistry ‘Rocasolano’,
CSIC, Serrano 119, Madrid 28006, Spain
Tel: +34 91 561 9400
E-mail: xjulia@iqfr.csic.es
and
F. J. Pastor, Department of Microbiology,
Faculty of Biology, University of Barcelona,
Av. Diagonal 643, Barcelona 08028, Spain
Tel: +34 93 4034626
E-mail: fpastor@ub.edu
Elena Jim
enez-Ortega and Susana
Valenzuela contributed equally to this work
(Received 28 February 2020, revised 27
March 2020, accepted 9 April 2020)
doi:10.1111/febs.15332
Reducing-end xylose-releasing exo-oligoxylanases (Rex) are GH8 enzymes
that depolymerize xylooligosaccharides complementing xylan degradation by
endoxylanases in an exo manner. We have studied Paenibacillus barcinonensis
Rex8A and showed the release of xylose from xylooligomers decorated with
methylglucuronic acid (UXOS) or with arabinose (AXOS). This gives the
enzyme a distinctive trait among known Rex, which show activity only on lin-
ear xylooligosaccharides. The structure of the enzyme has been solved by
X-ray crystallography showing a (a/a)
6
folding common to GH8 enzymes.
Analysis of inactived Rex8A-E70A complexed with xylotetraose revealed the
existence of at least four binding subsites in Rex8A, with the oligosaccharide
occupying subsites 3 to +1. The enzyme shows an extended Leu320-
His321-Pro322 loop, common to other Rex, which blocks the binding of
longer substrates to positive subsites further than +1 and seems responsible
for the lack or diminished activity of Rex enzymes on xylan. Mutants with
smaller residues in this loop failed to increase Rex8A activity on the polymer.
Analysis of the complexes with AXOS showed the accommodation of arabi-
nose at subsite 2, which cannot be allocated at subsite 1. Arabinose substi-
tutions at the xylose O2 or O3 are accommodated by hydrophobic interaction
and seem tolerated rather than recognized by Rex8A. A strained binding of
the branch is facilitated by the lack of direct polar interactions of the xylose
occupying this subsite, its water-mediated links allowing some conformational
flexibility of the sugar. The plasticity of Rex8A is a notable property of the
enzyme for its application in xylan deconstruction and upgrading.
Database
Structural data are available in PDB database under the accession numbers 6SRD (native form),
6TPP (E70A mutant in complex with EDO), 6TOW (E70A in complex with Xyl4), 6SUD (L320A
mutant in complex with xylose), 6SHY (L320A/H321S double mutant in complex with EDO),
6TO0 (E70A in complex with AX3), and 6TRH (E70A in complex with AX4).
Abbreviations
AX3, 2
3
-a-L-arabinofuranosyl-xylotriose (Ara
3
Xyl
3,
); AX4, 3
3
-a-L-arabinofuranosyl-xylotetraose (Ara
3
Xyl
4
); AXOS, arabinose-decorated
xylooligosaccharides; BaRex, Bifidobacterium adolescentis RexA; BhRex, Bacillus halodurans C-125 Rex; BiRex, Bacteroides intestinalis
Rex8; BiXyn, Bacteroides intestinalis Xyn8A; BsXyn, Bacillus sp. KK-1 XynY; CtCelA, Clostridium thermocellum b-glucanase; GH, glycosyl
hydrolase; PbRex, Paenibacillus barengoltzii Rex; PhXyl, Pseudoalteromonas haloplanktis xylanase; PhXyn, Pseudoalteromonas haloplanktis
Xyn8; Rex8A, Paenibacillus barcinonensis BP-23 reducing-end xylose-releasing exo-oligoxylanase; TtXyl, Teredinibacter turnerae xylanase;
UXOS, methylglucuronic acid-branched xylooligosaccharides.
5362 The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies

Introduction
Nowadays, biomass and its derivatives are emerging as
a promising raw material to replace petrol-based prod-
ucts for the development of a sustainable circular
economy [1]. One of the central components of bio-
mass is xylan, an heteropolysaccharide composed of
chains of xylose that are ramified to a different extent
and with different substituents depending on the
source of the polymer [2]. The enzymatic decomposi-
tion of xylan involves the catalysis of a complex pool
of hydrolytic enzymes to debranch and depolymerize
the main backbone of the polysaccharide [3]. Endoxy-
lanases (EC 3.2.1.8) are the key enzymes in the depoly-
merization process since they randomly cleave the
xylan backbone to soluble xylooligosaccharides. Based
on sequence and structure similarities of their catalytic
domain, xylanases have been classified into glycosyl
hydrolase (GH) families 10, 11, and 30 in CAZy data-
base (http://www.cazy.org), but xylanase activity is
also found in GH families 5, 8, and 43 [4]. Comple-
menting the xylanase activity, GH8 family contains the
few members of the reducing-end xylose-releasing exo-
oligoxylanases (Rex) (EC 3.2.1.156) described to date
[5-9]. Rex enzymes display the hydrolysis in an exo-
splitting manner, progressively removing xylopyranosyl
units from the reducing ends of the substrates. The
attributed role for Rex enzymes lies in the intracellular
depolymerization of xylooligosaccharides since four of
the five enzymes described until the date have no sig-
nal peptide [8].
GH8 enzymes present an inverting single-displace-
ment reaction mechanism and fold into a (a/a)
6
barrel
common to other inverting glycosidases. They have
been divided into three subfamilies depending on the
position of the residue acting as the base catalyst [10].
Rex8A belongs to subfamily GH8a, which has the
aspartate at the N terminus of helix a
8
of the barrel.
The three-dimensional structures of three GH8a mem-
bers, all of them endo-enzymes, have been reported,
that is, the b-glucanase from Clostridium thermocellum,
CtCelA [11], and the xylanases from Pseudoal-
teromonas haloplanktis, PhXyl [12], and Teredinibac-
ter turnerae, TtXyl [13]. Additionally, the molecular
basis of the exo-activity observed in Rex enzymes [14]
has been attributed to a kink in the loop preceding
a
10
, by crystallographic analysis of Bacillus halodurans
C-125 Rex, BhRex [5].
Rex8A from Paenibacillus barcinonensis is one of the
few examples of Rex enzymes described so far [8].It
belongs to the complex xylanolytic system of this bac-
terium, including several enzymes with industrial
potential [15-18]. As the other characterized Rex
enzymes, Rex8A shows an efficient removal of xylose
from linear xylooligosaccharides but only shows minor
activity on xylan. However, contrary to known Rex, it
also releases xylose from decorated xylooligosaccha-
rides (methylglucuronic acid-branched xylooligosaccha-
rides, UXOS), which gives RexA a distinctive trait
among other Rex enzymes [8]. In this work, we have
studied the structural traits of Rex8A that could give
insight into the unique properties of the enzyme. We
have solved the three-dimensional structure of Rex8A
and several mutants that have been constructed,
including those in the loop blocking the access to poly-
meric substrates, and analyzed their binding abilities
to xylooligomers by soaking and cocrystallization
experiments. We have also shown that the activity of
Rex8A on decorated oligomers extends to
xylooligosaccharides with arabinose decorations
(AXOS), and depicted their interaction within the
binding subsites by crystallographic analysis of their
complexes. The results obtained contribute to deci-
phering the function of Rex8A in xylan depolymeriza-
tion. Further research needs to be done to understand
the role of Rex enzymes in biomass deconstruction
and their potential in the upgrading of lignocellulosic
residues to added-value products.
Results and Discussion
Crystal structure of Rex8A
To elucidate the molecular basis of its enzymatic activ-
ity, the crystal structure of native Rex8A was solved at
1.9

A resolution (Fig. 1). Experimental details and
structure determination procedures are given in the
Materials and methods section and in Table 1. The
crystals belong to the P1 space group, with two mole-
cules in the asymmetric unit and 50% solvent content
within the cell. A xylose molecule from the cryoprotec-
tant solution used prior to data collection has been
captured at the active site (WT-xylose). The structure
was determined by means of molecular replacement
using as the search model the coordinates from the
reducing-end xylose-releasing exo-oligoxylanase from
B. halodurans C-125, BhRex (PDB entry 1WU4)[5].
Inactivated mutant Rex8A-E70A was also crystallized
and used for soaking experiments with xylotetraose
and xylohexaose; however, only two molecules of ethy-
lene glycol, used as additive in the precipitant solution,
were found bound at the active site (E70A-EDO),
which is explained in terms of the occlusion of part of
the active site by a proximal protein molecule located
next to the cavity, as explained below.
5363The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
E. Jim
enez-Ortega et al.Rex8A structure and specificity

Rex8A is folded into the expected (a/a)
6
folding
common to GH8 members, with long loops surround-
ing one axis of the barrel, and the C- and N-terminal
segments tightly packed against the opposite face
(Fig. 1). Besides the known structure of BhRex, having
64% sequence identity with Rex8A, the coordinates of
the Paenibacillus barengoltzii G22 PbRex, with 68%
identity, were recently released (PDB code 5YXT).
Structural comparison shows that the folding of the
three enzymes is very well conserved, as reflected in
the very low r.m.s.d. of 0.74 and 0.76

A of the two
enzymes vs. Rex8A, for 374 and 370 residues (Qscores
of 0.92 and 0.91), respectively. The most significant
difference is the Pro250-Arg260 loop highlighted in
Fig. 1, which is the most variable region and clearly
longer in Rex8A with respect to the other Rexs. A
xylobiose molecule bound in the BhRex crystal at the
2, 1 subsites suggested the importance of this loop
in determining substrate binding. Curiously, this is the
loop impeding active site cleft accessibility in Rex8A
by being inserted into the distal subsites of the adja-
cent molecule (Fig. 1, inset), which reinforces its role
in binding affinity. Therefore, additional experiments
were undertaken to find new crystallization conditions
that would allow the formation of complexes with
oligosaccharides.
The structure of the Rex8A-Xyl4 complex depicts
the active site
The inactivated mutant Rex8A-E70A was incubated
with xylotetraose prior to the setting of new crystal-
lization screenings, leading to a new crystal form with
different habit and a new trigonal symmetry (E70A-
Xyl4). From these crystals, the structure of the
Rex8A-Xyl4 complex was determined at 2.05

A resolu-
tion, the electron density maps showing the oligosac-
charide occupying subsites 3to+1, as seen in
Fig. 2A. The clear electron density observed in all the
subsites could indicate a strong interaction and, conse-
quently, the existence of at least four binding subsites
in Rex8A.
The overall structure of the complex is very similar
to that of the unliganded crystals, and only minor
rearrangement of the Pro250-Arg260 loop, which
approximates by 0.5

A to the ligand, is observed upon
Fig. 1. Structure of Rex8A. Superimposition
of native Rex8A-Xyl complex (prune) onto
BhRex-Xyl
2
(pale cyan). The position of two
ethylene glycol molecules trapped in the
E70A crystals is shown in yellow. The
different molecules are occupying subsites
2, 1, and +1. The N- and C-terminal
residues of Rex8 chain are represented as
spheres and highlighted with arrows. The
sequence alignment of the Rex8 loop
Pro250-Arg260, highlighted in the cartoon,
is shown in a box marking position Leu255-
His256. Inset: Crystal packing locates
Leu255, from loop 250-260 of an adjacent
molecule (beige), blocking Rex8A active site
cleft. Model generated using PyMol
Molecular Graphics System [30].
5364 The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
Rex8A structure and specificity E. Jim
enez-Ortega et al.

Table 1. Crystallographic statistics (values in parentheses are for the high-resolution shell).
WT-xyl E70A-EDO E70A-Xyl4 L320A-Xyl L320A/H321S-EDO E70A-AX3 E70A-AX4
Crystallization,
soaking, and
cryoprotecting
conditions
23% PEG6K
100 mMTris,
pH 8
200 mMNaCl
10 mM[Co
(NH
3
)
6
]Cl
3
25% glycerol
50 mMxylose
20% PEG6K
100 mM
HEPES pH 7
200 mM
MgCl
2
10 mM[Co
(NH
3
)
6
]Cl
3
20% ethylene
glycol
6m
M
xylohexaose
18% PEG3350K
100 mM
BisTris
propane pH
7.5
200 mM
KSCN
20% PEG400
50 mM
xylotetraose
22% PEG6K
100 mM
HEPES pH 7
200 mM
MgCl
2
10 mM[Co
(NH
3
)
6
]Cl
3
6% glycerol
20% ethylene
glycol
50 mMxylose
22% PEG6K
100 mMHEPES
pH 7
200 mMMgCl
2
10 mM[Co(NH
3
)
6
]
Cl
3
6% glycerol
20% ethylene
glycol
50 mMxylose
16% PEG3350K
100 mMBisTris
propane pH 7.5
200 mMKSCN
20% glycerol
15 mM2
3
-a-L-
arabinofuranosyl-
xylotriose
20% PEG3350
100 mMBisTris
propane pH 7.5
200 mMKSCN
25% glycerol.
5m
M3
3
-alpha-L-
arabinofuranosyl-
xylotetraose
Space group P1P1P3
1
2P1P1P3
1
2C222
1
Unit cell parameters
a,b,c(

A) 51.6, 59.2, 79.5 51.7, 58.4, 79.3 84.6, 84.6, 274.6 51.5, 59.2, 79.4 51.3, 58.6, 78.7 87.9, 87.9, 274.2 82.3, 82,7, 465.3
a,b,c(°) 87.8, 78.0, 74.3 87.3, 78.2, 73.5 90, 90, 120 87.3, 78.4, 74.6 87.5, 78.1, 74.3 90, 90, 120 90, 90, 90
Data collection
Beamline XALOC (ALBA) XALOC (ALBA) XALOC (ALBA) XALOC (ALBA) XALOC (ALBA) XALOC (ALBA) XALOC (ALBA)
Temperature
(K)
100 100 100 100 100 100 100
Wavelength (

A) 1.04191 0.979490 0.979260 0.979490 0.979250 0.979260 0.979260
Resolution (

A) 48.58–1.93 (1.98–
1.93)
77.63–2.64 (2.77–
2.64)
43.94–2.05 (2.09–
2.05)
57.04–1.74 (1.77–
1.74)
45.78–1.81 (1.85–
1.81)
44.49–1.88 (1.88–
1.91)
46.62–1.86 (1.86–
1.89)
Data processing
Total
reflections
225 331 (13 133) 72 150 (9770) 481 963 (29 580) 265 064 (13 708) 209 507 (11 875) 660 511 (32 568) 961 039 (17 154)
Unique
reflections
64 154 (4213) 24 307 (3296) 72 614 (4419) 87 761 (4448) 75 674 (4471) 100 926 (4930) 129 847 (5330)
Multiplicity 3.5 (3.1) 3.0 (3.0) 6.6 (6.7) 3.0 (3.1) 2.8 (2.7) 6.5 (6.6) 7.4 (3.2)
Completeness
(%)
96.4 (93.2) 94.4 (95.2) 100.0 (100.0) 96.6 (95.0) 96.2 (94.8) 99.9 (99.5) 97.2 (80.7)
Mean I/r(I) 7.2 (2.0) 5.4 (1.8) 12.2 (2.9) 7.1 (2.1) 5.6 (2.0) 13.8 (2.7) 7.4 (3.2)
R
mergea
(%) 12.5 (72.3) 16.4 (68.2) 9.3 (64.2) 9.8 (62.8) 11.1 (64.8) 7.8 (60.9) 15.3 (65.8)
R
pimb
(%) 7.8 (48.9) 10.9 (45.5) 3.8 (26.4) 6.7 (42.3) 7.7 (46.3) 3.2 (25.3) 5.2 (37.7)
CC1/2 (%) 99.2 (66.3) 97.1 (56.3) 99.8 (90.4) 98.7 (78.7) 98.4 (62.2) 99.9 (99.5) 99.2 (56.8)
Molecules per
ASU
22222 2 4
Refinement
R
work
/R
freec
(%) 18.9/22.9 19.0/23.7 23.6/26.5 24.2/27.7 17.6/20.6 23.4/25.7 23.8/26.3
No. of atoms/average B(

A
2
)
Protein 6226/16.90 6218/29.20 6259/34.33 6254/14.99 6230/20.61 6248/29.96 12 557/35.88
Ligand 44/23.48 16/33.57 74/45.61 102/23.31 72/36.50 86/28.17 184/33.96
Water
molecules
606/28.25 227/25.21 311/36.36 597/24.79 487/30.74 423/37.54 194/32.81
All atoms 6876/17.94 6461/29.08 6644/34.55 6953/15.95 6789/21.50 6757/30.42 12 935/35.88
Ramachandran plot (%)
Favored 99 97 98 98 98 98 96
Outliers 0 0 0 0 0 0 0
RMS deviations
Bonds (

A) 0.006 0.006 0.007 0.006 0.006 0.007 0.006
Angles (°) 1.418 1.450 1.447 1.461 1.503 1.474 1.417
PDB accession
codes
6SRD 6TPP 6TOW 6SUD 6SHY 6TO0 6TRH
a
R
merge
=∑
hkl
∑
i
|I
i
(hkl)[I(hkl)]|/∑
hkl
∑
i
I
i
(hkl), where I
i
(hkl) is the ith measurement of reflection hkl and [I(hkl)] is the weighted mean of all mea-
surements.;
b
R
pim
=∑
hkl
[1/(N1)] 1/2 ∑
i
|I
i
(hkl)[I(hkl)]|/∑
hkl
∑
i
I
i
(hkl), where Nis the redundancy for the hkl reflection.;
c
R
work
/R
free
=∑
hkl
|
F
o
F
c
|/∑
hkl
|F
o
|, where F
c
is the calculated and F
o
is the observed structure-factor amplitude of reflection hkl for the working/free (5%) set,
respectively.
5365The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
E. Jim
enez-Ortega et al.Rex8A structure and specificity

sugar binding (Fig. 2B). The movement toward the
substrate at the main chain of two regions near subsite
+1 previously reported in BhRex [5] and corresponding
to Thr62-Asn64 and Lys355-Arg358 in Rex8A is not
observed in our complex, although a different orienta-
tion of the Lys357 and Arg358 side chains is evident
in the crystal. The relevance that this observation may
have in substrate binding is discussed later.
Xyl4 is occupying the narrow cleft in an extended
conformation with hydrophobic stacking interactions
of the xyloses at subsites 2 and +1 to Trp112 and
Tyr361 side chains, respectively (Fig. 2B). In addition,
the oligosaccharide is accommodated by many polar
interactions with the residues at the active site, includ-
ing several well-ordered water molecules.
As has been described for the other Rex enzymes,
subsite 3 seems quite open to the solvent, the xylose
ring making only hydrophobic contacts to the con-
served Ile187. A higher K
m
for xylooligosaccharides
longer that xylotetraose has been reported in BhRex8A
[14] although no structural rationale supported the
negative effect of binding at this subsite [5]. In Rex8A,
the cleft is narrowed by the longer Pro250-Arg260
loop (Fig. 1A), which situates the C5-O5 moiety
within 5

A from His256 ring and making hydrophobic
contact to its side chain. In addition, O2 is linked to
the Pro125 main chain through a water molecule,
which may contribute to fix the xylose bound at this
subsite. A subsite 3 has been previously reported to
be present in the endoglucanase CtCelA [11] and the
endoxylanase TtXyl, with 26% and 38% sequence
identity, respectively, to Rex8A [13], but not in PhXyl
[12]), with 30% identity. We reported previously that
Rex8A shows maximum activity vs. xylotriose, with
lower activity vs. xylotetraose, which is consistent with
the low number of interactions accommodating xylose
bound at this subsite. Therefore, despite that clear
electron density seems to indicate a fixed position for
xylose, the contribution of this subsite to substrate
affinity might be moderate. On the other hand, consid-
ering its position and orientation, decoration of xylose
at this subsite could be allowed only at O3 that is
open to the solvent.
The xylose located at subsite 2 is sandwiched
between Trp112 and Phe186 (this last through a T-in-
teraction pattern) in a way conserved in all known
GH8 enzymes degrading xylose-based oligosaccharides.
In addition, most of its O atoms are involved in
water-mediated polar interactions. Thus, O2 and O5
are linked through water molecules to Tyr197 hydroxyl
and to Tyr243 (O2) and Phe186 main chain (O5). Lar-
gest differences in the active site with respect to the
known Rexs are found in this subsite 2 due to the
presence of its extended Pro250-Arg260 loop, contain-
ing His256, and the substitution of a Tyr by Phe259 in
Rex8A. This makes a narrowed cavity at O3, and a
wider cleft at O2 filled by a well-ordered net of water
molecules that fixes O2 from xylose rings at subsites
2 and 1, and the O glycoside link. Therefore, deco-
rations could appear, in principle, impeded at O3, but
possible at O2, but this issue was further investigated,
as explained later.
Distortion of the sugar ring located at subsite 1isa
common requirement in GH enzymes for the hydrolysis
to proceed. In the case of GH8 enzymes, a
2,5
B-like
transition state was proposed in the CtCelA as this con-
formation had been observed in the reported complex
of this enzyme with cellopentaose [11]. However, in the
two other reported complexes with xylooligosaccharides
spanning subsites 1, +1 of GH8 endoxylanases from
TtXyl-Xyl6 [13] and PhXyl-Xyl5 [12], this boat confor-
mation has not been trapped, the xylose located at sub-
site 1 being in a
4
C
1
chair noncatalytically relevant
conformation. In the particular case of the TtXyl-Xyl6
complex, xylose at subsite 1 is in a very peculiar ring-
flipped, Southern Hemisphere
1
C
4
conformation. Fur-
thermore, by in silico docking analysis, a conforma-
tional change from the chair ground state to a
pretransition state local minimum in a
2
S
o
conforma-
tion was proposed [12]. In our Rex8A-Xyl4 complex,
the xylose is in a distorted
4
C
1
conformation with C1,
C2, C3, C4, and C5 being rather flat. Thus, the O2 and
O3 atoms keep polar interactions with Asp128 (O2, O3)
and Arg268 (O2) similarly to that observed in the
CtCelA complex, but the cyclic O5 atom is puckered
and makes a polar interaction with the Arg68 side
chain, absent in the CtCelA enzyme. The relevance of
this conformation is uncertain, considering that inacti-
vation of the enzyme involves removal of the proton-
donor side chain in the Glu70Ala replacement, which
might introduce artificial biases at this subsite. The
same biases could have happened in the previously
reported GH8 complexes, in which the proton donor
(CtCelA-E95Q), the base (TtXyl-D281N), or a third
catalytically relevant aspartate (PhXyl-D144A) has
been altered (Fig. 2C). Consequently, the specific nat-
ure of the catalytically relevant enzyme–substrate com-
plex and the conformational itinerary of these class of
enzymes remains to be confirmed. Nevertheless, and
despite the different conformations captured at subsite
1, a remarkable similarity in the position and ring ori-
entation of the Glu/Xyl bound from 2to+1 subsites
is evident in Fig. 2C. This feature, together with a simi-
lar chain bending at the scissile bond, reveals a very
conserved catalytic machinery within both endo- and
exo-GH8 enzymes.
5366 The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
Rex8A structure and specificity E. Jim
enez-Ortega et al.

The xylose located at subsite +1 is stacked to
Tyr361, conserved in all the GH8 enzymes, and is
tightly fixed by many polar interactions to residues
unique to the Rex enzymes. Thus, O1, O2, and O3 are
making direct polar links to His321, Asp61, and
Arg68 side chains. In addition, O3 and O5 are making
water-mediated interaction to Tyr360 and Ser264 side
chains. Interestingly, O2 is pointing to a polar cavity
occupied by well-ordered water molecules making a
net of hydrogen links connecting the sugar to Asp61
carboxylate and to the main chain of Arg358, Tyr360,
and Tyr361. This network of polar links seems essen-
tial for keeping the integrity of the subsite +1 architec-
ture, as will be illustrated below. On the other hand,
the most important difference found in the Rex sub-
family with respect to the other GH8 enzymes is a
kink in the loop Leu320-Pro322 that blocks the active
site at the reducing-end and impedes bindings of
longer substrates [5]. Thus, exo-GH8 enzymes may
have evolved from their endo-partners by just a few
changes in this section of the active site, which may
suggest that the reversal of this endo-character could
be possibly accomplished, in principle, by mutating
selected key residues, an issue that is explained below.
Molecular basis of the exo/endo-activity
As mentioned before Rex8A does not show positive
subsites further than +1 because it is blocked by loop
Leu320-His321-Pro322 that impedes binding of longer
substrates. This loop is conserved among the charac-
terized Rex enzymes, while it is not found in known
GH8 xylanases. As an exception, the sequence LHP is
not found in the Rex from Bacteroides intestinalis
(BiRex), although it has an extra span of amino acids
in the corresponding region of the enzyme (Fig. 3).
This fact should preclude the endo-activity of the Rex
enzymes that show only a very minor activity on
xylans probably resulting from its reducing-end exo-ac-
tivity. To analyze the contribution of the loop to the
exo- vs. endo-activity of Rex8A, a series of mutants
were constructed in which the residues Leu and His
were replaced by smaller Ala residues. The activity of
the two single mutants constructed, L320A and
H321A, was assayed on different types of xylans
(Table 2). Wild-type enzyme showed higher activity on
arabinoxylans from wheat and oat spelt than on
beechwood glucuronoxylan, although all the activity
values detected were very small. The activity on xylans
was also confirmed by TLC analysis, which showed
the release of xylose from the polymers (data not
shown). L320A showed a similar low activity on
Fig. 2. Rex8A active site. (A) Molecular surface of Rex8A-Xyl4
complex, showing the sugar as olive sticks and the xylose bound
at the native crystals in red. Blue/red are positive/negative Rex8A-
charged regions. Inset: final 2FoFc electron density map of Xyl4,
contoured at 1r. (B) Detail of the atomic interactions in the Rex8A-
Xyl4 complex, with polar links shown as dashed lines. The residues
that rearrange upon substrate binding are highlighted showing its
position in the native crystal as salmon sticks. The catalytic
residues are colored in magenta. (C) Superimposition of the Rex8-
Xyl4 complex (olive) onto the reported GH8 complexes PhXyl-Xyl5
[12] (slate) and CtCelA-Xyl5 [11] (beige), showing
2,5
B and
4
C
1
conformation, respectively, at the sugar bound at subsite 1. Only
the catalytic base, the proton donor and a third relevant aspartate
are represented, with the replacement done in each case to
inactivate the enzymes. Model generated using PyMol Molecular
Graphics System.
5367The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
E. Jim
enez-Ortega et al.Rex8A structure and specificity

arabinoxylans, while no activity was found on glu-
curonoxylan. A double-mutant L320A/H321S con-
structed showed similar behavior on xylans than
L320A. On the other hand, H321A mutant did not
show activity on the xylans tested.
Then, the activity of wild-type Rex8A and its
mutant derivatives was analyzed on linear
xylooligosaccharides (Fig. 4A). Rex8A showed activity
on the substrates tested, xylotriose, xylotetraose, and
xylohexaose, releasing xylose and xylobiose as main
hydrolysis products, in accordance with its exo mode
of action. Null mutant E70A did not show activity on
these oligomers, while L320A showed similar activity
on xylooligosaccharides to that observed with the
wild-type enzyme. On the contrary, H321A and
L320A/H321S showed a diminished activity on these
substrates, this decrease being more apparent in the
double mutant tested against xylohexaose.
We investigate the molecular basis of these results
performing crystallization experiments, which yielded
crystals from the single L320A and double L320A/
H321S mutants. The structure of the L320A mutant
with a trapped xylose (L320A-Xyl in Table 1) is identi-
cal to that from the native-xylose complex shown in
Fig. 1, which reveals an equivalent binding mode at
subsite +1 consistent with the similar behavior
observed in the activity of this mutant vs.
xylooligosaccharides and xylan. Thus, Leu320 does
not appear critical for Rex8A exo-activity. In contrast,
the double mutant L320A/H321S (L320A/H321S-
EDO) shows structural rearrangement of some regions
delineating the active site, particularly at the loops
Leu60-Asp65 and Arg354-Tyr360, at the end of helices
a1 and a8 of the barrel, which shift by 1

A probably
as a consequence of the loss of polar links associated
with the removal of the His321 side chain. As it is
shown in Fig. 4B, the His321 side chain interacts with
Asp362 that is linked to Arg354, which, in turn, inter-
acts with Arg358 main chain. Therefore, removal of
His321-Asp362 link might disrupt the polar network
building this part of the active site with the observed
shift in the loops, what can be detrimental for sub-
strate binding. Most important seems the loss of the
His321–xylose interaction, as the NH-O1 hydrogen
bond has been reported to be critical for discrimina-
tion of the a/bxylose isomer [5]. As it is observed in
Fig. 4C,D, removal of His321 makes a wider cavity in
the mutant, as compared to native, which allows to
bind an EDO molecule linked to the introduced
Ser321-OH. However, the fact that soaking crystals
with xylose did not trap xylose at the active site points
to an important decrease in affinity, which agrees with
the extremely reduced activity observed in this mutant.
Therefore, the elongation of the active site does not
induce an endo-mode of action but, rather, disturbs
the active site for proper positioning of the xylose at
+1 subsite, with a detrimental effect in activity.
Analysis of the specificity vs decorated
substrates
As mentioned before, previous works showed the
activity of Rex8A on methylglucuronic acid-decorated
xylooligosaccharides (UXOS). The substrates reported
aldotetraouronic (MeGlcA
3
Xyl
3
) and aldohexaouronic
(MeGlcA
3
Xyl
5
), showing the decoration in the third
xylose from the reducing-end, were shortened in one
xylose residue but were not further hydrolyzed [8].
These results together with docking analysis suggested
the accommodation of the methylglucuronic moiety in
2 subsite of the enzyme, while the 1 subsite would
not allow decorations [8]. This type of substrate
accommodation resembled to that described for GH30
bacterial xylanases, which require accommodation in
Fig. 3. Sequence alignment of reducing-end xylose-releasing exo-oligoxylanases (Rexs). Alignment of the amino acid sequences was
performed using the CLUSTAL OMEGA (https://www.ebi.ac.uk/Tools/msa/clustalo). The Rex sequences shown are Bacteroides intestinalis Rex8A
(BiRex, GenBank EDV05843.1), Bifidobacterium adolescentis RexA (BaRex, GenBank AAO67498.1), Paenibacillus barcinonensis BP-23
Rex8A (Rex8A, GenBank ALP73600.1), Paenibacillus barengoltzii Rex (PbRex, GenBank WP_016314299.1), and Bacillus halodurans C-125
Rex (BhRex, GenBank BAB05824.1). GH8 xylanase sequences shown are Pseudoalteromonas haloplanktis Xyn8 (PhXyn, GenBank
CAD20872.1), Bacillus sp. KK-1 XynY (BsXyn, GenBank AAC27700.1), and Bacteroides intestinalis Xyn8A (BiXyn, GenBank EDV05067.1).
Residues 320–322 forming the LHP loop are highlighted in a green square.
5368 The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
Rex8A structure and specificity E. Jim
enez-Ortega et al.

2 subsite for activity on glucuronoxylans and UXOS
[19-22]. We compared hydrolysis products released by
a GH30 xylanase, P. barcinonensis Xyn30D, or by
Rex8A from aldotetraouronic and aldohexaouronic
acids by TLC analysis. The products released by the
enzymes showed identical mobility, clearly indicating
the accommodation of decorations in 2 subsite of
Rex8A (Fig. 5A).
In addition to the ability of Rex8A to hydrolyse
methylglucuronic acid-decorated xylooligosaccharides,
our present work reveals that Rex8A exhibits moder-
ate activity on the two tested arabinoxylans (Table 2).
Furthermore, from inspection of the crystal structure
reported here, we can expect that the cavity observed
at 2 subsite could also accommodate arabinose sub-
stituents of xylan oligomers. Thus, we tested Rex8A
activity on two different arabinose-decorated
xylooligosaccharides (AXOS), 2
3
-a-L-arabinofuranosyl-
xylotriose (Ara
3
Xyl
3,
AX3) and 3
3
-a-L-arabinofura-
nosyl-xylotetraose (Ara
3
Xyl4
,
AX4), which show the
arabinose decoration in the third xylose from the
Table 2. Activity of Rex8A and derived mutants on xylans
(mUmg
1
). ND, not detected; mUmg
1
, milliunits per milligram.
Oat spelt
xylan
Wheat
arabinoxylan
Beechwood
xylan
Wt 38.3 4 26.6 2 8.1 3
E70A ND ND ND
L320A 42.6 1.6 26.8 4ND
H321A ND ND ND
L320A7H321S 13.2 1.1 5.1 4ND
Fig. 4. Exo/endo-activity of Rex8. (A) Xylotriose (X3), xylotetraose (X4), and xylohexaose (X6) were incubated with Rex8A wt and derived
mutants E70A (E), L320A (L), H321A (H), or L320A/H321S (L/H), and the hydrolysis products were analyzed by thin-layer chromatography.
Lanes: c, control sample (no digestion); M, size markers X1, X2, X3, X4, and X5. (B) Superimposition of native Rex8A (salmon) onto the
L320A/H321S double mutant (slate), showing polar interactions of native His320 as black dashed lines, and that of mutated Ser320 as a
cyan line. The loops that rearrange in the mutant are labeled. Molecular surface of native (C) and double mutant (D), showing the larger
cavity created in the mutated protein and the molecules bound at each active site. Xylose position is shown in (D) as a reference. Model
generated using PyMol Molecular Graphics System.
5369The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
E. Jim
enez-Ortega et al.Rex8A structure and specificity

reducing-end, but differ, besides length, in the xylose
carbon atom to which the arabinose moiety is linked,
C2 or C3, respectively. Considering the shape of the
cavity observed in Rex8A at subsite 2, only decora-
tions at O2 would have been expected, as O3 is point-
ing to a more constricted, narrower pocket, as
commented before. Surprisingly, Rex8A cleaved both
AXOS, which were shortened in a xylose residue that
was released (Fig. 5B). The shortened AXOS were not
further cleaved by Rex8A, similarly to the results
shown for UXOS. These results confirm the
accommodation of decorations in 2 subsite, while the
1 subsite would not allow main chain substituents.
On the other hand, the activity of mutants on these
substrates follows the same pattern to that observed
on xylan and linear xylooligosaccharides (Table 2and
Fig. 5B).
To investigate the binding mode of branched
xylooligosaccharides and the specific recognition of the
corresponding substitutions, the E70A-inactivated
mutant was incubated with both substrates and new
crystallization conditions were tested. Crystals were
grown from the Rex8A-AX3 and Rex8A-AX4 com-
plexes, and their structures were solved leading to the
final data given in Table 1. The analysis of both com-
plexes reveals that both AXOS are able to accommodate
within the narrow slot similarly to the unsubstituted
Xyl4, showing no significant changes in the residues
delineating the active site (Fig. 6A). However, some dif-
ferences are observed in the binding mode and in the
conformation adopted by each substrate. First, the AX3
is bound with the three xylose units in a chair conforma-
tion and keeping the same interactions pattern described
in the Rex8A-Xyl4 complex (Fig. 6B). At subsite 2,
the introduced arabinose is oriented to the hydrophilic
cavity surrounding the catalyst Asp265 and filled with
water molecules, in a way that its ring oxygen is making
the same hydrogen bonds presented by the unsubsti-
tuted O2 of Xyl4. This seems the only polar interaction
stabilizing the decoration. In the case of the Rex8A-
AX4 complex, binding of the substituted tetrasaccharide
involves a distortion of the xylose at subsite 3, which
is stabilized by a new water-mediated polar interaction
to Gly124, while the arabinose at subsite 2 is directed
out from the cavity showing no interactions with the
enzyme (Fig. 6C). Apart from this, the hydrogen bond
pattern at the other subsites is essentially similar to the
other two complexes.
Therefore, it seems that Rex8A is not specifically
recognizing the arabinose substitutions of these sub-
strates but, rather, it would be tolerating the existence
of these decorations at its active site. The structural
rationale for this hypothesis is made clear from the
complexes here reported. The analysis of the binding
mode of the tested xylooligosaccharides reveals that
the 2 and 3 subsites lack direct polar links between
the protein and the substrate, which might confer
more plasticity to them. At subsite 2, the xylose unit
is fixed by stacking to Trp112 and hydrophobic T-in-
teraction to Phe186. Moreover, the Rex8A QLH motif
makes a more prominent loop that narrows the slot,
placing the decorations in a very constrained position.
Thus, in the case of substitutions at O3, some move-
ment of the flexible loop could have been expected to
Fig. 5. Rex8A activity against decorated xylooligosaccharides. (A) A
mixture of aldouronic acids [aldotetraouronic (MeGlcA
3
Xyl
3
) and
aldohexaouronic (MeGlcA
3
Xyl
5
)] was incubated with enzymes, and
the hydrolysis products were analyzed by thin-layer
chromatography. Lanes: 1, control sample (no digestion); 2, sample
digested with Rex8A; 3, sample digested with Xyn30D; 4, sample
digested with Rex8A and Xyn30D; M, size markers X1, X2, X3, X4,
X5, and X6. (B) Arabinose-decorated xylooligosaccharides: 2
3
-a-L-
arabinofuranosyl-xylotriose (Ara
3
Xyl
3
) and 3
3
-a-L-arabinofuranosyl-
xylotetraose (Ara
3
Xyl
4
) were incubated with Rex8A wt and derived
mutants E70A (E), L320A (L), H321A (H), or L320A/H321S (L/H),
and the hydrolysis products were analyzed by thin-layer
chromatography. Lanes: c, control sample (no digestion); M, size
markers X1, X2, X3, X4, and X5. Each digestion was performed
and analyzed in three independent experiments to confirm the
results.
5370 The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
Rex8A structure and specificity E. Jim
enez-Ortega et al.

open the slot making room for the branch at O3. On
the contrary, fitting of the decorated substrate leads to
slight distortions at the adjacent subsite 3 that seem
easily tolerated by Rex8A.
Conclusions
The reducing-end specificity of Rex8 enzymes has been
attributed to a kink in the loop Leu320-His321-Pro322
blocking subsite +2. According to this, a limited num-
ber of mutations have been proposed to mediate the
evolution from the general endo-activity of GH8 to
the particular exo-xylooligosaccharide activity of Rexs.
In an attempt to reverse this evolution, we have inves-
tigated the effect of changes in the motif LH and cor-
related the results with the crystal analysis. While
mutants at Leu320 do not introduce significant
changes, removal of His321 impedes a proper position-
ing of the xylose at +1 subsite, explaining the detri-
mental effect observed in activity. Thus, construction
of a more extended active site does not induce the
expected exo-activity but, rather, introduces slight
rearrangement of some loops at the active site that
result in decreased activity.
Furthermore, we have expanded our previous work
on the unique activity of Rex8A on branched glu-
curono-xylooligosaccharides, by elucidating its ability
to degrade arabino-substituted substrates. Moreover,
detailed structural analysis of complexes with linear
and decorated oligomers reveals the molecular basis of
its specificity, and a unique loop at 2 subsite that
delineates a narrow pocket determining a constrained
binding mode of the decorations at this subsite. Arabi-
nose substitutions at the xylose O2 or O3 are accom-
modated by hydrophobic interaction and seem
‘tolerated’ rather than recognized by Rex8A. This
strained binding of the branch is facilitated by the lack
of direct polar interactions of the xylose occupying this
subsite, its water-mediated links allowing some confor-
mational flexibility of the sugar.
Rex8 are very interesting enzymes for deconstruction
of xylan. Its molecular analysis could be of great value
in the design of efficient xylanases to transform ligno-
cellulose and cell wall materials into biofuel or added-
value products, with the aim of achieving more sus-
tainable industrial processes.
Materials and methods
Cloning and mutagenesis of Rex8A
Construction of Rex8A and Rex8A-E70A expression vectors
was previously reported [8]. Rex8A-L320A, Rex8A-H321A,
and Rex8A-L320A/H321S were generated by site-directed
mutagenesis using the same strategy as with Rex8A-E70A
[8]. The following mutagenic oligonucleotide primers were
used in the PCR: Rex8A-L320A-fw (5-aacgagcccgcaGCgc
accccgtaggcctgctggccaccaat-3) and Rex8A-L320A-rv (5-gcc
Fig. 6. Binding mode of decorated substrates to Rex8A. (A).
Superimposition of Xyl4 (green), AX3 (purple), and AX4 (slate) as
found at the active site of Rex8A; the arabinose decorations have
been highlighted with thicker sticks. Relevant residues are
represented. (B, C) Binding of AX3 and AX4n to Rex8A showing
the polar interactions as dashed lines and relevant water
molecules. Final 2FoFc electron density map of AX3 and AX4 are
contoured at 1r. The catalytic residues are highlighted in a lighter
tone. Model generated using PyMol Molecular Graphics System.
5371The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
E. Jim
enez-Ortega et al.Rex8A structure and specificity

tacggggtgcGCtgcgggctcgttgaacggttcaccctc-3); Rex8A-H321A-
fw (5-gagcccgcattgGCccccgtaggcctgctggccaccaatgct-3) and
Rex8A-H321A-rv (5-caggcctacggggGCcaatgcgggctcgttgaacgg
ttcacc-3); Rex8A-L320A/H321S-fw (5-aacgagcccgcaGCgAGc
cccgtaggcctgctggccaccaat-3) and Rex8A-L320A/H321S-rv (5-
attggtggccagcaggcctacggggCTcGCtgcgggctcgtt-3) (mis-
matched codons are underlined, changed bases are in capital).
Expression and purification of Rex8A and
mutants
Recombinant proteins were purified from Escherichia coli
BL21 Star (DE3) clones containing the respective plasmid
as described elsewhere [8]. Briefly, exponential phase cul-
tures were induced with 0.5 MIPTG at 16 °C for 18 h and
cells were lysed by a Homogenizer PandaPLUS 2000
(GEA, B€
onen, Germany). Recombinant His-Tag proteins
were purified from cleared cell extracts by immobilized
metal affinity chromatography (IMAC) using HisTrap HP
columns of 1 mL (GE Healthcare, Chicago, IL, USA) on a
fast protein liquid chromatography system (
€
AKTATM
FPLC TM; GE Healthcare). Samples were then desalted
with PD-10 columns (GE Healthcare) and finally eluted in
20 mMTris/Cl buffer, pH 7. Proteins concentration was
performed in Centricon Centrifugal Filter Units of 30 kDa
MWCO (Millipore, Burlington, MA, USA). The purity of
the protein was verified by SDS/PAGE, and the concentra-
tion of the enzymes was measured with NanoDrop

ND-
1000 (NanoDrop Technologies, Inc, Wilmington, DE,
USA), using an extinction coefficient (e) at 280 nm of
99 950 for all the variants.
Crystallization and data collection
Initial crystallization conditions for Rex8A (5 mgmL
1
in
20 mMTris/HCl pH 7) were explored by high-throughput
techniques with a NanoDrop robot (Innovadyne Technolo-
gies Inc., Wilmington, DE, USA), using six different com-
mercially screens: PACT and JCSG +Suites from Qiagen
(QIAGEN N.V., Venlo, The Netherlands); JBScreen Classic
1–4 from Jena Bioscience (Jena, Germany); and Index, Crys-
tal Screen, and SaltRx packages from Hampton Research
(Aliso Viejo, CA, USA). These assays were carried out using
the sitting-drop vapor-diffusion method in MRC 96 well
crystallization plates (Molecular Dimensions, Sheffield, UK).
Rod-shaped crystals diffracting at moderate resolution grew
in 22–25% (w/v) PEG 6K, 200 mMNaCl, and 100 mMTris
pH 8. Additional optimization experiments led to good-qual-
ity crystals with 10 mMof the additive Co[NH
3
)
6
]Cl
3.
For
data collection, crystals were transferred to cryoprotectant
solutions consisting of mother liquor plus 25% (v/v) glycerol
before being cooled in liquid nitrogen.
Additional optimization experiments were performed
with the E70A, L320A, and L320A/H321S mutants, leading
to the final conditions given in Table 1. For getting the
E70A-Xyl4 complex, new conditions were investigated with
the commercial screens by incubating the inactivated
enzyme with the sugar, previously to setting the experi-
ments. This produced a new crystal form allowing occu-
pancy of the active site by the substrate. The complexes of
E70A with the AXOS were obtained by cocrystallization
and optimizing these conditions, which in the case of AX4
led to a new crystal form. Final cryoprotectant solutions
used with the different crystals are summarized in Table 1.
Diffraction data were collected using synchrotron radia-
tion at the XALOC beamline at ALBA (Cerdanyola del
Vall
es, Spain). Diffraction images were processed with XDS
[23] and merged using AIMLESS from the CCP4 package [24].
A summary of data collection and data reduction statistics
for all the crystals is shown in Table 1.
Structure solution and refinement
The structure of Rex8A was solved by molecular replace-
ment using the MOLREP program [25]. The search model was
the B. halodurans exo-oligoxylanase (PDB entry 1WU4 [5]),
from which a template was prepared using the program
CHAINSAW [26] and a protein sequence alignment of Rex8A
onto the template. A solution containing two molecules in
the asymmetric unit was found using reflections up to
1.93

A resolution range. Crystallographic refinement was
performed using the program REFMAC [27] within the CCP4
suite with flat bulk-solvent correction, maximum likelihood
target features. Free R-factor was calculated using a subset
of 5% randomly selected structure-factor amplitudes that
were excluded from automated refinement. Extensive model
building using the program COOT [28] was combined with
several rounds of refinement leading to a model showing a
continuous density for the whole polypeptide chain. At the
later stages, water molecules were included in the model,
combined with more rounds of restrained refinement that
led to a final R-factor of 18.9 (R
free
22.9). Refinement
parameters are reported in Table 1. The structure of E70A,
L320A, and L320/H321S mutants was solved by difference
Fourier synthesis using the refined coordinates of the
native. The structure of the E70-Xyl4, E70-AX3, and E70-
AX4 complexes, showing a different space group, was
solved by molecular replacement using MOLREP and the
coordinates of native Rex8A as the search model. Refine-
ment for all structures was performed using REFMAC, com-
bined with model building with COOT, and coordinates for
the ligands were taken from the Protein Data Bank and
manually built into the electron density maps and refined,
to reach the R-factors listed in Table 1.
Stereochemistry of the models was checked with PRO-
CHECK [29], and the figures were generated with PYMOL [30].
Root mean square deviation analysis was made using the
program SUPERPOSE within the CCP4 package [24].
5372 The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
Rex8A structure and specificity E. Jim
enez-Ortega et al.

Rex8A activity measurements
The catalytic activity of Rex8A and mutants on xylan sub-
strates was analyzed by measuring the amount of reducing
sugars released from their hydrolysis using the dinitrosali-
cylic (DNS) reagent method as assayed by Ref. [31] (https://
doi.org/10.1016/j.ijbiomac.2019.11.073) with some modifica-
tions. The standard assay was performed in 100 lLreaction
volume using 1.5% of substrate (w/v) in 50 mMTris/Cl buf-
fer, pH 7, at 40 °C and 300 r.p.m. The enzymatic reactions
were carried out for 18 h unless otherwise stated. The opti-
cal absorbance was measured at 540 nm, and xylanase activ-
ity was measured in terms of international units (IU). One
unit of xylanase activity was defined as the amount of
enzyme that releases 1 lmol of xylose reducing sugar equiva-
lent per min under the assay conditions described. The tested
xylans were beechwood xylan from Carl Roth, oat spelt
xylan from Sigma-Aldrich (Darmstadt, Germany), and
wheat arabinoxylan from Megazyme (Bray, Ireland).
Analysis of hydrolysis products
Xylan and xylooligosaccharide hydrolysis products were ana-
lyzed by thin-layer chromatography (TLC) performed on sil-
ica gel plates (Macherey-Nagel, Duren, Germany). Typical
assays were performed in Eppendorf tubes using final concen-
trations of 500 lgmL
1
of enzyme, 10 mgmL
1
of xyloo-
ligosaccharide, or 1.5% of xylan in 20 mMTris/HCl buffer
pH 7. Reactions were incubated at 40 °Cand300r.p.m.for
18 h, and supernatants were concentrated by evaporation
before loading in silica plates. Xylobiose, xylotriose, xylote-
traose, and xylopentaose were purchased from Megazyme.
a-L-arabinofuranosyl-(1?2)-b-D-xylopyranosyl-(1?4)-b-D-
xylopyranosyl-(1?4)-D-xylose, 2
3
-a-L-arabinofuranosyl-
xylotriose, (Ara
3
Xyl
3,
AX3), and b-D-xylopyranosyl-(1?
4)-[a-L-arabinofuranosyl-(1?2)]-b-D-xylopyranosyl-(1?4)-b-
D-xylopyranosyl-(1?4)-D-xylose, 3
3
-a-L-Arabinofuranosyl-xy-
lotetraose (Ara
3
Xyl
4,
AX4) were purchased from Megazyme.
Mixtures containing the aldouronic acids 4-O-methyl-a-D-
glucuronosyl-(1?2)-b-D-xylopyranosyl-(1?4)-b-D-xylopy-
ranosyl-(1?4)-D-xylose, (MeGlcA
3
Xyl
3
) (aldotetraouronic
acid), and b-D-xylopyranosyl-(1?4)-b-D-xylopyranosyl-(1?
4)-[4-O-methyl-a-D-glucuronosyl-(1?2)]-b-D-xylopyranosyl-
(1?4)-b-D-xylopyranosyl-(1?4)-D-xylose, (MeGlcA
3
Xyl
5
)
(aldohexaouronic acid), were prepared from beechwood 4-
O-methyl-D-glucuronoxylan by xylanase treatment as des-
cribed before [8]. The solvent used was chloroform/acetic acid/
water (30 : 60 : 10), and the hydrolysis products were detected
by spraying ethanol/sulfuric acid mixture (95 : 5) on the plate
and revealing with a heat gun until spots become visible.
Acknowledgements
This work was supported by grants from the Spanish
Ministry of Economy and Competitiveness through
grants BIO2016-76601-C3-3-R, BIOFABCEL CTQ2017-
84966-C2-2-R. We are grateful to the staff of the Syn-
chrotron Radiation Sources at Alba (Barcelona, Spain)
and ESRF (Grenoble, France) for providing access
and for technical assistance at BL13-XALOC and
Massif beamlines, and to the ‘Xarxa de Refer
encia en
Biotecnologia’ (XRB).
Conflict of interest
The authors declare no conflict of interest.
Author contributions
FJP and JS-A conceived and designed the experiments.
EJ-O, MR-E, and SV carried out the experiments and
analyzed data. FJP and JS-A analyzed data and wrote
the manuscript. All the authors contributed to criti-
cally discuss the results and reviewed and approved
the manuscript.
References
1 Sherwood J (2020) The significance of biomass in a
circular economy. Bioresour Technol 300, 122755.
2 Naidu DS, Hlangothi SP & John MJ (2018) Bio-based
products from xylan: a review. Carbohydr Polym 179,
28–41.
3 Biely P, Singh S & Puchart V (2016) Towards
enzymatic breakdown of complex plant xylan structures:
state of the art. Biotechnol Adv 34, 1260–1274.
4 Lombard V, Golaconda Ramulu H, Drula E, Coutinho
PM & Henrissat B (2014) The carbohydrate-active
enzymes database (CAZy) in 2013. Nucleic Acids Res
42, D490–D495.
5 Fushinobu S, Hidaka M, Honda Y, Wakagi T, Shoun
H & Kitaoka M (2005) Structural basis for the
specificity of the reducing end xylose-releasing exo-
oligoxylanase from Bacillus halodurans C-125. J Biol
Chem 280, 17180–17186.
6 Lagaert S, Van Campenhout S, Pollet A, Bourgois TM,
Delcour JA, Courtin CM & Volckaert G (2007)
Recombinant expression and characterization of a
reducing-end xylose-releasing exo-oligoxylanase from
Bifidobacterium adolescentis.Appl Environ Microbiol 73,
5374–5377.
7 Hong PY, Iakiviak M, Dodd D, Zhang M, Mackie RI
& Cann I (2014) Two new xylanases with different
substrate specificities from the human gut bacterium
Bacteroides intestinalis DSM 17393. Appl Environ
Microbiol 80, 2084–2093.
8 Valenzuela SV, Lopez S, Biely P, Sanz-Aparicio J &
Pastor FIJ (2016) The glycoside hydrolase family 8
reducing-end xylose-releasing exo-oligoxylanase Rex8A
5373The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
E. Jim
enez-Ortega et al.Rex8A structure and specificity

from Paenibacillus barcinonensis BP-23 is active on
branched xylooligosaccharides. Appl Environ Microb 82,
5116–5124.
9 Liu XQ, Jiang ZQ, Liu Y, You X, Yang SQ & Yan QJ
(2019) Biochemical characterization of a novel exo-
oligoxylanase from Paenibacillus barengoltzii suitable for
monosaccharification from corncobs. Biotechnol Biofuels
12, 190.
10 Adachi W, Sakihama Y, Shimizu S, Sunami T,
Fukazawa T, Suzuki M, Yatsunami R, Nakamura S &
Takenaka A (2004) Crystal structure of family GH-8
chitosanase with subclass II specificity from Bacillus sp
K17. J Mol Biol 343, 785–795.
11 Guerin DMA, Lascombe MB, Costabel M, Souchon H,
Lamzin V, Beguin P & Alzari PM (2002) Atomic (0.94
angstrom) resolution structure of an inverting
glycosidase in complex with substrate. J Mol Biol 316,
1061–1069.
12 De Vos D, Collins T, Nerinckx W, Savvides SN,
Claeyssens M, Gerday C, Feller G & Van Beeumen J
(2006) Oligosaccharide binding in family 8 glycosidases:
crystal structures of active-site mutants of the beta-1,4-
xylanase pXyl from Pseudoaltermonas haloplanktis
TAH3a in complex with substrate and product.
Biochemistry 45, 4797–807.
13 Fowler CA, Hemsworth GR, Cuskin F, Hart S,
Turkenburg J, Gilbert HJ, Walton PH & Davies GJ
(2018) Structure and function of a glycoside hydrolase
family 8 endoxylanase from Teredinibacter turnerae.
Acta Crystallogr D 74, 946–955.
14 Honda Y & Kitaoka M (2004) A family 8 glycoside
hydrolase from Bacillus halodurans C-125 (BH2105) is a
reducing end xylose-releasing exo-oligoxylanase. J Biol
Chem 279, 55097–55103.
15 Valenzuela SV, Diaz P & Pastor FIJ (2010)
Recombinant expression of an alkali stable GH10
Xylanase from Paenibacillus barcinonensis.J Agric Food
Chem 58, 4814–4818.
16 Valenzuela SV, Valls C, Roncero MB, Vidal T, Diaz P
& Pastor FIJ (2014) Effectiveness of novel xylanases
belonging to different GH families on lignin and
hexenuronic acids removal from specialty sisal fibres. J
Chem Technol Biotechnol 89, 401–406.
17 Valenzuela SV, Diaz P & Pastor FI (2014) Xyn11E
from Paenibacillus barcinonensis BP-23: a LppX-
chaperone-dependent xylanase with potential for
upgrading paper pulps. Appl Microbiol Biotechnol 98,
5949–5957.
18 Valls C, Pastor FIJ, Vidal T, Roncero MB, Diaz P,
Martinez J & Valenzuela SV (2018) Antioxidant activity
of xylooligosaccharides produced from glucuronoxylan
by Xyn10A and Xyn30D xylanases and Eucalyptus
autohydrolysates. Carbohyd Polym 194,43–50.
19 Sainz-Polo MA, Valenzuela SV, Gonzalez B, Pastor FI
& Sanz-Aparicio J (2014) Structural analysis of
glucuronoxylan-specific Xyn30D and its attached
CBM35 domain gives insights into the role of
modularity in specificity. J Biol Chem 289, 31088–
31101.
20 Valenzuela SV, Diaz P & Pastor FI (2012) Modular
glucuronoxylan-specific xylanase with a family CBM35
carbohydrate-binding module. Appl Environ Microbiol
78, 3923–3931.
21 Urbanikova L, Vrsanska M, Morkeberg Krogh KB,
Hoff T & Biely P (2011) Structural basis for substrate
recognition by Erwinia chrysanthemi GH30
glucuronoxylanase. FEBS J 278, 2105–2116.
22 St John FJ, Hurlbert JC, Rice JD, Preston JF &
Pozharski E (2011) Ligand bound structures of a
glycosyl hydrolase family 30 glucuronoxylan
xylanohydrolase. J Mol Biol 407,92–109.
23 Kabsch W (2010) XDS. Acta Crystallogr D 66, 125–132.
24 Winn MD, Ballard CC, Cowtan KD, Dodson EJ, Emsley
P, Evans PR, Keegan RM, Krissinel EB, Leslie AG,
McCoy A et al. (2011) Overview of the CCP4 suite and
current developments. Acta Crystallogr D 67, 235–242.
25 Vagin AA & Teplyakov A (1997) MOLREP: an
automated program for molecular replacement. J Appl
Crystallogr 30, 1022–1025.
26 Stein N (2008) CHAINSAW: a program for mutating
pdb files used as templates in molecular replacement. J
Appl Crystallogr 41, 641–643.
27 Murshudov GN, Vagin AA & Dodson EJ (1997)
Refinement of macromolecular structures by the
maximum-likelihood method. Acta Crystallogr D 53,
240–255.
28 Emsley P & Cowtan K (2004) Coot: model-building
tools for molecular graphics. Acta Crystallogr D 60,
2126–2132.
29 Laskowski RA, MacArthur MW, Moss DS &
Thornton JM (1993) PROCHECK: a program to check
the stereochemical quality of protein structures. J Appl
Crystallogr 26, 283–291.
30 DeLano WL (2002) Pymol: An Open-Source Molecular
Graphics Tool. The PyMOL Molecular Graphics
System, DeLano Scientific, San Carlos, CA.
31 Bouiche C, Boucherba N, Benallaoua S, Martinez J,
Diaz P, Pastor FIJ & Valenzuela SV (2019) Differential
antioxidant activity of glucuronoxylooligosaccharides
(UXOS) and arabinoxylooligosaccharides (AXOS)
produced by two novel xylanases. Int J Biol Macromol.
https://doi.org/10.1016/j.ijbiomac.2019.11.073
5374 The FEBS Journal 287 (2020) 5362–5374 ª2020 Federation of European Biochemical Societies
Rex8A structure and specificity E. Jim
enez-Ortega et al.