Penicillium purpurogenum produces two GH family 43 enzymes with

b-xylosidase activity, one monofunctional and the other bifunctional:

Biochemical and structural analyses explain the difference
Mara Cristina Ravanal
a
, Melissa Alegra-Arcos
b
, Fernando Danilo Gonzalez-Nilo
b,
, Jaime Eyzaguirre
a,
a
Universidad Andres Bello, Facultad de Ciencias Biolgicas, Santiago, Chile
b
Universidad Andres Bello, Center for Bioinformatics and Integrative Biology, Facultad de Ciencias Biolgicas, Santiago, Chile
a r t i c l e i n f o
Article history:
Received 28 August 2013
and in revised form 8 October 2013
Available online 31 October 2013
Keywords:
b-D-Xylosidase
Penicillium purpurogenum
Heterologous expression
Glycosyl hydrolase family 43
Homology modeling
Molecular docking simulations
a b s t r a c t
b-Xylosidases participate in xylan biodegradation, liberating xylose from the non-reducing end of
xylooligosaccharides. The fungus Penicillium purpurogenum secretes two enzymes with b-D-xylosidase
activity belonging to family 43 of the glycosyl hydrolases. One of these enzymes, arabinofuranosidase
3 (ABF3), is a bifunctional a-L-arabinofuranosidase/xylobiohydrolase active on p-nitrophenyl-a-L-
arabinofuranoside (pNPAra) and p-nitrophenyl-b-D-xylopyranoside (pNPXyl) with a K
M
of 0.65 and
12 mM, respectively. The other, b-D-xylosidase 1 (XYL1), is only active on pNPXyl with a K
M
of
0.55 mM. The xyl1 gene was expressed in Pichia pastoris, puried and characterized. The properties of
both enzymes were compared in order to explain their difference in substrate specicity. Structural
models for each protein were built using homology modeling tools. Molecular docking simulations were
used to analyze the interactions dening the afnity of the proteins to both ligands. The structural
analysis shows that active complexes (ABF3pNPXyl, ABF3pNPAra and XYL1pNPXyl) possess specic
interactions between substrates and catalytic residues, which are absent in the inactive complex
(XYL1pNPAra), while other interactions with non-catalytic residues are found in all complexes. pNPAra
is a competitive inhibitor for XYL1 (K
i
= 2.5 mM), conrming that pNPAra does bind to the active site but
not to the catalytic residues.
2013 Elsevier Inc. All rights reserved.
Introduction
Lignocellulose is a major component of plants and it represents
the main source of renewable organic matter. There is considerable
interest in the exploitation of lignocellulosic materials as a source
of feed, fuels and chemical feedstocks. Both the abundance and
renewable nature of lignocellulosic materials have provided input
for intensive research over recent years. It can be converted to va-
lue-added products through saccharication by lignocellulolytic
enzymes [1]. Lignocellulose is composed of lignin, pectin, cellulose
and hemicelluloses. Xylan is the main hemicellulose of annual
plants and hardwoods, and it is composed of a linear chain of xylo-
pyranose residues joined by b (1 ?4) glycosidic linkages. The main
chain is substituted by a variety of compounds, such as arabinof-
uranose, methyl glucuronate and acetate, and the arabinoses may
be linked to hydroxycinnamic acids [2].
The biodegradation of xylan is a complex process. The main
chain is hydrolyzed by the action of endoxylanases (EC 3.2.1.8),
which liberate xylooligosaccharides of different length and are
eventually hydrolyzed to xylose by b-xylosidases (EC 3.2.1.37).
These enzymes are produced by fungi and bacteria and are mainly
extracellular [3]. b-Xylosidases are active against articial sub-
strates such as p-nitrophenyl glycosides; most of them are very
specic for p-nitrophenyl-b-D-xylopyranoside (pNPXyl)
1
[46].
Others are able to cleave p-nitrophenyl-a-L-arabinofuranoside
(pNPAra) [710], p-nitrophenyl-b-D-galactopyranoside [9] or p-
nitrophenyl-a-D-glucopyranoside [5,11,12]. b-Xylosidases are
grouped in families, according to their amino acid sequence similar-
ities, in the Carbohydrate Active Enzymes database (CAZy; http://
www.cazy.org); they are classied in eight families of glycoside
hydrolases (GH): 3, 30, 39, 43, 52, 54, 116 and 120 [1316]. Filamen-
tous fungal b-xylosidases have been described only for families 3, 43
and 54 [5,12,17]. Members of GH families 3 and 54 operate with
retention of the anomeric conguration, while family GH43 contains
inverting glycoside hydrolases. These last enzymes possess a proton
donor acting as general acid and a nucleophile as general base; in
0003-9861/$ - see front matter 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.abb.2013.10.017

Corresponding authors. Fax: +56 2 6980414.

E-mail addresses: fernando.gonzalez@unab.cl (F.D. Gonzalez-Nilo), jeyzaguirre@
unab.cl (J. Eyzaguirre).
1
Abbreviations used: ABF3, arabinofuranosidase 3; XYL1, b-xylosidase; pNPXyl, p-
nitrophenyl-b-D-xylopyranoside; pNPAra, p-nitrophenyl-a-L-arabinofuranoside; GH,
glycoside hydrolases.
Archives of Biochemistry and Biophysics 540 (2013) 117124
Contents lists available at ScienceDirect
Archives of Biochemistry and Biophysics
j our nal homepage: www. el sevi er. com/ l ocat e/ yabbi
addition, a third carboxylate has been found to be essential for activ-
ity in GH family 43 enzymes. This residue is responsible for the mod-
ulation of the pK
a
of the general acid and of the correct orientation of
both proton donor and substrate [18].
The soft rot fungus Penicillium purpurogenum secretes a wide
variety of xylanolytic enzymes to the medium, among them two
enzymes with b-D-xylosidase activity. One of these enzymes: arab-
inofuranosidase 3 (ABF3), is a bifunctional a-L-arabinofuranosi-
dase/xylobiohydrolase [10]. The other is -xylosidase 1 (XYL1),
which has been expressed in Pichia pastoris GS115 and found to
be inactive towards pNPAra. Both enzymes belong to family 43
of the glycosyl hydrolases. The purpose of this work is to compare
the biochemical and structural properties of both enzymes in order
to explain their difference in substrate specicity.
Materials and methods
Microbial strains utilized
P. purpurogenum ATCC strain MYA-38 was the source for DNA
and RNA. Escherichia coli TOP10 F
0
was used for cloning of the
XYL1 encoding gene. Heterologous expression was performed in
P. pastoris GS115, supplied in the EasySelect

Pichia Expression
Kit (Invitrogen, Carlsbad, CA, USA).
DNA and RNA preparations
P. purpurogenum was grown in Mandels medium as described
previously [19] using 0.75 g/l Neopeptone (DIFCO, Franklin Lakes,
NJ, USA), 0.3 g/l urea, and 1.4 g/l (NH
4
)
2
SO
4
as the nitrogen source.
The carbon source was 1% glucose for genomic DNA extraction and
1% sugar beet pulp for RNA extraction.
Genomic DNA from P. purpurogenum was extracted with the
GeneJET

Genomic DNA Purication kit (Fermentas, Burlington,

Canada). For RNA extraction, the fungus was grown on sugar beet
pulp and washed with PBS (137 NaCl, 2.7 KCl, 10 Na
2
HPO
4
, and
1.8 mM KH
2
PO
4
, pH 7.4) and total RNA was extracted by means
of the RNeasy Plant Mini Kit (Qiagen, Valencia, CA). mRNA was ex-
tracted from the total RNA using the Absolutely mRNA kit (Strata-
gene, Santa Clara, CA, USA). cDNA was prepared from mRNA by
means of the First Choice RLM-RACE kit (Ambion, Austin, TX,
USA). The instructions from the manufacturers were followed in
all procedures utilizing kits in this work.
Cloning and expression of xyl1 in Pichia pastoris
A gene encoding a potential GH 43 b-xylosidase was found
searching the P. purpurogenum genome sequence database (unpub-
lished). Specic (non-degenerate) primers were designed based on
this sequence. The primers XYL-F1 (ATGGCTGCTCCCAAACCCCT)
and XYL-R1 (AGCCAAGTGGATCTTCCCTTCCG) were designed from
the translation initiation and termination sequences, respectively,
and were used to amplify by PCR the genomic DNA and cDNA.
The PCR program used was: 30 cycles of: 94 1, 60 1, 72 2 and a nal
extension at 72 C for 10 min. The cDNA and genomic DNA were
sequenced in both strands by Macrogen Inc. (Seoul, Korea).
Primers XYL-FW (ATAGAATTCATGGCTGCTCCCAAACCCCT, for-
ward) and XYL-RV (CGCTCTAGATAAGCCAAGTGGATCTTCCCT, re-
verse) (EcoRI and XbaI restriction sites are in bold) were used to
amplify by PCR the xyl1 gene for cloning. The reverse primer was
designed without a stop codon so as to include a polyhistidine tail
present in the vector. The PCR product (program performed as
above) was cloned in the expression vector pPICZB (Invitrogen).
E. coli TOP10 F
0
cells were transformed with the resulting plasmid
(pPICZB-xyl1). Transformed cells were selected in low-salt LB agar
plates containing 25 lg/ml Zeocin. The PCR product was also
cloned in pGEMT easy (Promega, Madison, WI, USA) and sequenced
in both strands. pPICZB-xyl1 was linearized by digestion with SacI
and transformed into competent P. pastoris GS115. Approximately
2 lg of linearized DNA were used. Transformed Pichia clones were
selected from YPDS (Yeast extract, peptone, dextrose and sorbitol)
plates containing 100 lg/ml Zeocin and grown for 3 days. Ten Zeo-
cin-resistant clones of P. pastoris GS115/pPICZB-xyl1 were grown
each in 20 ml YPD (1% yeast extract, 2% peptone and 2% dextrose)
medium for 24 h and 28 C in a rotary shaker (200 rpm). An aliquot
of 200 ll of each culture was added to 20 ml of BMGY medium (1%
yeast extract, 2% peptone, 100 mM potassium phosphate, pH 6.0,
1.34% yeast nitrogen base, 4 105% biotin and 1% glycerol) and
incubated for 24 h (28 C; 200 rpm). The cultures were centrifuged
under sterile conditions and the cells were resuspended in 20 ml of
BMMY medium (Buffered Methanol-complex medium: 1% yeast
extract, 2% peptone, 1 M potassium phosphate, pH 6.0, 13.4% yeast
nitrogen base with ammonium sulfate without amino acids, 0.02%
biotin and 1% methanol). To induce protein expression by metha-
nol, the cultures were incubated for 7 days at 28 C and 200 rpm,
adding methanol every 24 h to a nal concentration of 1%. The cul-
ture supernatants were assayed for glycosyl hydrolase activities.
Enzyme activity assays
The enzymatic activity of the selected clones was determined
using pNP glycosides (Sigma, St Louis, MO, USA): pNPXyl, pNPAra,
p-nitrophenyl b-D-galactopyranoside, p-nitrophenyl a-D-galacto-
pyranoside, p-nitrophenyl a-L-arabinopyranoside, and p-nitrophenyl
a-D-glucopyranoside at different pHs in the range of 3.08.0 using
McIlvaine buffer (citric acid/disodium hydrogen phosphate). The
activity determinations were performed in 96-well plates. Each
well contained 140 ll McIlvaine buffer, 25 ll pNP glycoside
(10 mM) as substrate, and 10 ll enzyme. After 10 min at 40 C
the reaction was stopped by adding 125 ll 0.2 M Na2CO3 and
absorbance was measured at 405 nm in a Universal Microplate
Reader ELx800 (BioTek Instruments, Vinooski VT, USA). One unit
of activity was dened as the amount of enzyme required to hydro-
lyze 1 lmol of substrate per min.
Inhibition of XYL1 by pNPAra was performed with pNPXyl as
substrate with two different concentrations of pNPAra (2 and
5 mM) in McIlvaine buffer pH 6.0 at 37 C.
Purication of the recombinant XYL1
For the purication, a 2 l culture of one of the active P. pastoris
transformants was grown. The culture supernatant was concen-
trated by ultraltration (Minitan; Millipore, Billerica, MA, USA)
and subjected to afnity chromatography in NiNTA agarose (Qia-
gen, Valencia, CA, USA) (20 ml bed volume) in 50 mM sodium
phosphate buffer pH 8.0 + 0.5 M NaCl and washed with the same
buffer +20 mM imidazole and eluted with 250 mM imidazole in
the same buffer.
Protein quantication and electrophoresis
The DC Protein Assay (Bio-Rad, Hercules, CA, USA) with bo-
vine serum albumin as standard was utilized. These analyses were
performed in triplicate. Molecular weight of the expressed protein
was determined by SDSPAGE [20] using PageRulerTM Unstained
Protein Ladder as standard (Fermentas). Gels were stained with
Coomassie Brilliant Blue R-250. Isoelectrofocusing and zymograms
were performed as described previously [21] using methyl umbel-
liferyl b-D-xylopyranoside as substrate.
118 M.C. Ravanal et al. / Archives of Biochemistry and Biophysics 540 (2013) 117124
Activity of XYL1 on oligosaccharides
Enzyme activity on xylooligosaccharides was assayed using
xylobiose, xylotriose, xylotetraose, and xylopentaose (Megazyme;
Bray, Wicklow County, Ireland) as substrates. Enzyme (4 lg) was
incubated with 10 mg/ml substrate in McIlvaine buffer (citric
acid/disodium hydrogen phosphate) pH 6.0 at 30 C. Products were
analyzed by thin layer chromatography (TLC) as described previ-
ously [22] after 8 h of incubation and the liberation of xylose was
quantied using the K-XYLOSE kit from Megazyme after an incuba-
tion of 24 h.
Activity of XYL1 against arabinoxylan
Wheat arabinoxylan (Megazyme) (10 mg/ml) was incubated at
30 C for 8 h in McIlvaine buffer (pH 6.0) with 4 lg XYL1 and/or
4 lg P. purpurogenum endoxylanase A (puried as described in
[23]) per assay mixture. Xylose liberation was quantied as de-
scribed above. A control was performed by incubating the sub-
strate under identical conditions but in the absence of enzymes.
The amount of xylose detected in the control was subtracted from
the raw data.
Template selection
The NCBI-BlastP program was used to nd proteins related to
the ABF3 and XYL1 sequences. The sequences of proteins stored
in the Protein Data Bank and PDBSum databases were used to re-
strain the search space to proteins of known structure. The best-
aligned sequence found corresponded to glycoside hydrolase fam-
ily 43 arabinoxylan arabinofuranohydrolase from Bacillus subtilis
[24] with has a 31.5% and 31.3% identity to ABF3 and XYL1, respec-
tively. The corresponding crystallographic-resolved structure (PDB
ID: 3C7E) was chosen as template to build the three-dimensional
models. The PsiPred algorithm was used to predict the secondary
structure of the query proteins; the prediction was further used
to rene the sequence alignments using ClustalW v2.0 [25].
Building of 3D models
MODELLER 9.v10 [26] was used to construct several three-
dimensional models of ABF3 and XYL1 using the crystallographic
structure PDB ID: 3C7E as template. The best model obtained
according to the DOPE potential was chosen for additional optimi-
zation. We chose the models presenting the lowest DOPE scores.
The three-dimensional model was further relaxed using the
CHARMM 27 forceeld [27] and the NAMD v2.9 software [28] as
described elsewhere [29]. Briey, the amino-acid protonation state
of ABF3 and XYL1 at their optimal pH (5 and 6 respectively) was
predicted using the PROPKA program [30] through the Maestros
Protein Preparation Wizard [31]. The protein models were hy-
drated using a 1.5-nm slab of TIP3P water [32] from the farthest
protein atoms, using periodic boundary conditions. The hydrated
systems were neutralized by means of Na
+
and Cl

counterions
at 0.150 M. Then, an energy minimization protocol was performed
using the steepest-descent algorithm to remove initial amino acid
clashes, lling vacuum pockets and undesired atom contacts, fol-
lowed by a molecular dynamics with the protein backbone re-
stricted using a harmonic potential of 0.5 kcal/mol
2
. The
protein loops were set free in order to explore the conformational
space that is limited by its micro-environment. Molecular dynam-
ics simulations were run until energetic convergence was reached.
The nal three-dimensional models for ABF3 and XYL1 obtained
after MD were extracted and used for the molecular dockings.
Ligand molecular model building
The molecular model of pNPAra and pNPXyl were built and
optimized using the Maestro suite [33]. Then the partial charges
for each atom were assigned using the force eld OPLS. Finally,
the structures were optimized and prepared to perform the molec-
ular docking simulations using Glide [34].
Docking and score calculations
Ligandprotein docking simulations were performed using
Glide 5.8 from Schrdinger suite [34]. The grid box size and posi-
tion were selected to fully explore the pocket near equivalent res-
idues to Asp15, Asp128 and Glu187 of GH43 b-xylosidase, from
Geobacillus stearothermophilus T-6 (XynB3) which are described
to participate in the catalysis mechanism [35]. The extra-precision
Glide scoring function [36] was used to evaluate the ligandprotein
binding afnity (XP score), thereby analyzing four complexes:
ABF3-pNPAra, ABF3-pNPXyl, XYL1-pNPXyl and XYL1-pNPAra.
Images and analyses were generated with the VMD program [37]
and Maestro suite [33].
Results
Sequence of the xyl1 gene and XYL1
Xyl1 has an open reading frame of 1011 bp which is identical to
the cDNA sequence, indicating that xyl1 has no introns (Supple-
mentary Fig. 1). The genomic DNA sequence is available in Gen-
Bank with accession number JX840600. The gene codes for a
protein of 336 amino acid residues, with a calculated molecular
weight of 37 070. XYL1 contains no apparent signal peptide (search
was performed with the SignalP 4.0 server; http://www.cbs.dtu.dk/
services/SignalP). Possible N-glycosylation and O-glycosylation
sites were searched using the NetNGlyc 1.0 and NetOGlyc 3.1 serv-
ers, respectively (http://www.cbs.dtu.dk/services/NetNGlyc/)
(http://www.cbs.dtu.dk/services/NetOGlyc/). One potential N-gly-
cosylation site can be identied at Asn180. By means of Pfam
(http://pfam.sanger.ac.uk) one conserved domain was identied
in XYL1, similar in sequence to a domain present in family 43 of
the glycosyl hydrolases.
Heterologous expression of the xyl1 gene
Heterologous expression of the xyl1 gene was performed in P.
pastoris. About 20 Zeocin resistant clones were isolated. A signi-
cant activity, only with pNPXyl, was detected in 7 clones (the per-
centage of activity on the other pNP glycosides tested was less than
2%) indicating that XYL1 is a b-xylosidase. No activity was observed
in the supernatant of a culture of the control P. pastoris trans-
formed with pPICZB in the absence of insert.
Enzymatic properties of XYL1
The recombinant enzyme was puried to near homogeneity and
showed one band when analyzed by SDSPAGE (Fig. 1). The appar-
ent molecular mass deduced from the gel is 49 kDa; this value is
higher than the calculated molecular weight, probably due to
glycosylation. The kinetic properties were studied with pNPXyl.
The enzyme followed MichaelisMenten kinetics with a K
M
of
0.55 mM and k
cat
/K
M
of 2.50 10 4 min

1 M
1
. Inhibition by
pNPAra was competitive for XYL1. K
i
value = 2.5 mM for pNPAra
with pNPXyl as substrate was estimated from the equation
v = (k
cat
S)/(k
m
(1 + I/K
i
)+S). Fig. 2 shows the LineweaverBurk
plot for XYL1. The pH optimum for the enzyme (assayed in a pH
M.C. Ravanal et al. / Archives of Biochemistry and Biophysics 540 (2013) 117124 119
range of 3.08.0) is 6.0. Optimal temperature, assayed in a range of
465 C, is 40 C. XYL1 is not very stable and markedly loses activ-
ity by freezing and thawing.
The enzyme activity was detected in a zymogram using methyl
umbelliferyl b-D-xylopyranoside. The isoelectric point is approxi-
mately 5.1 (data not shown). The theoretical pI based on the se-
quence was estimated at 4.92 (http://web.expasy.org/compute_pi).
Activity of XYL1 against xylooligosaccharides and arabinoxylan
was assayed. Fig. 3 shows a TLC of the reaction products detected
after 8 h of incubation. No transglycosylation (appearance of oligo-
saccharides of higher degree of polymerization than the substrate)
is detected when using xylooligosaccharides or xylose. Small
amounts of xylose are liberated when XYL1 is incubated with ara-
binoxylan. Table 1 shows the percentage of xylose liberated by b-
xylosidase 1 after 24 h of incubation. A TLC performed after 24 h
of incubation (data not shown) presents a similar pattern to that
of Fig. 3, but with higher amounts of xylose.
The effect of endoxylanase A from P. purpurogenum (GH family
10 enzyme) on the liberation of xylose from wheat arabinoxylan by
XYL1 was analyzed. The results indicate that a signicant synergis-
tic effect is observed when endoxylanase A is combined with XYL1.
Fig. 3 lane 13 shows that XYL1 liberates only a small amount of
xylose fromarabinoxylan. Lane 15 (XYL1 + xylanase A) shows more
xylose than lane 14 (xylanase A alone), conrming synergism and
showing that XYL 1 acts only on short oligosaccharides. This is fur-
ther validated, as shown in Supplementary Fig. 2, where the
amount of xylose liberated is quantied.
Homology modeling and docking simulations
A key step to build a reliably homology model is to nd a good
reference structure in the PDB. Sequences alignments used to build
the structural models are shown in Fig. 4. The aligned catalytic res-
idues are labeled with a red arrow. The percentage identity se-
quence between ABF3 and XYL1 and the reference protein are in
a range that allows the building of a reliably good homology model
using some structural restriction that permit conserving the sec-
ondary structure information. Specically, ABF3 has 31.5% identity
and 56% similarity with the sequence of the reference structure
(PDB: 3C7E), while XYL1 has 31.3% and 46% identity and similarity,
respectively, with the same template.
The models generated had DOPE score of 42915.83 and
29456.32 for ABF3 and XYL1 respectively. The homology models
are shown in Fig. 5A and D. The model for ABF3 shows that it has
two domains, one of them a catalytic domain and the other a car-
bohydrate-binding module (CBM) [10]. XYL1 has just one catalytic
domain. In order to validate the structural models, the distances
previously reported for the catalytic residues in the reference
structure are compared with the distances obtained in the models.
Specically, the distances shown in Fig. 5A and D between Asp12-
Glu187 in ABF3 (10.3 ) and between Asp15-Glu240 in XYL1
(7.6 ), are in agreement with previous experimental reports
(10 2 ) [38,39]. Supplementary Fig. 3 shows the structural eval-
uation of the models carried out with ProSA [40]: the Z-score plot
for ABF3 (5.74) and XYL1 (5.11).
Finally, the homology models were used to explore the confor-
mation adopted by the ligands in the catalytic site of ABF3 and
XYL1. Supplementary Fig. 4 shows the structure of the ligands
Fig. 1. SDSPAGE of the puried XYL1. Lane 1: Molecular weight markers
(prestained protein ladder, Fermentas); Lane 2: puried XYL1. The gel was stained
with Coomassie Brilliant Blue R 250.
0 1 2 3 4 5
0
20
40
60
80
100
120
140
160
180
200
220
240
260
280
300
(
1
/
v
)

(
1
/
U
/
m
g
)
(1/[pNPXyl]) (1/mM)
XYL1/pPNPXyl
0 mM pNPAra
2 mM pNPAra
5 mM pNPAra
Fig. 2. LineweaverBurk plot for XYL1: Double-reciprocal plot obtained in the
absence of the inhibitor and with two different concentrations of pNPAra.
Fig. 3. Thin layer chromatography of the hydrolysis products of xylooligosaccha-
rides and arabinoxylan by XYL1. Lane 1: xylose; lane 2: xylose + XYL1; lane 3:
xylobiose; lane 4: xylobiose + XYL1; lane 5: xylotriose; lane 6: xylotriose + XYL1;
lane 7: xylotetraose; lane 8: xylotetraose + XYL1; lane 9: xylopentaose; lane 10:
xylopentaose + XYL1; lane 11: oligosaccharide standards (xylose, xylobiose, xylo-
triose, xylotetraose, xylopentaose); lane 12: arabinoxylan; lane 13: arabinoxy-
lan + XYL1; lane 14: arabinoxylan + endoxylanase A; lane 15:
arabinoxylan + endoxylanase A + XYL1. The samples were incubated for 8 h at
30 C and pH 6.0.
Table 1
Percentage of xylose liberated by XYL1
from xylooligosaccharides after 24 h of
incubation at 30 C.
Xylooligosaccharides Xylose
liberated
(%)
X2 5.9
X3 6.3
X4 15
X5 29
120 M.C. Ravanal et al. / Archives of Biochemistry and Biophysics 540 (2013) 117124
used: pNPAra (A) and pNPXyl (B). Fig. 5BC and EF presents a sum-
mary of the lowest energy structure of the proteinligand com-
plexes predicted by docking simulations. The afnity energy
obtained for these complexes are in agreement with the experi-
mental data (Table 2), and the structural information obtained
can explain why XYL1 is a mono-functional protein. Another
important structural pattern that validates experimentally these
models is the distances between the sugar ring of the substrate
and the carboxyl group of Asp12 (ABF3) or Asp15 (XYL1), which
are optimal to intercalate a water molecule. This observation is
in agreement with experimental evidence [35,39]. Specically,
the distance between Asp12 (ABF3) and the sugar ring of pNPAra
is 10.2 ; the same distance is observed between Asp12 (ABF3)
and the sugar ring pNPXyl. The distance between Asp15 (XYL1)
and the sugar ring of pNPXyl is 7.4 and in the complex XYL1-
pNPAra the distance is 7.9 .
Discussion
The main objective of this work is a comparison of the proper-
ties of ABF3 (a bifunctional enzyme) and XYL1 (a monofunctional
enzyme) in order to explain their differences in substrate specic-
ity. The properties of ABF3 have been described previously [10] and
those of XYL1 are presented in this work.
Expression of XYL1
The vector pPICZB does not have a signal sequence and as a re-
sult XYL1 is secreted under the control of its own non-canonical
signal peptide. Secretion of XYL1, therefore, does not use the nor-
mal secretory pathway. This evidence conrms ndings in other
secreted GH43 beta-xylosidases which also lack a signal peptide
[5,41].
Comparative biochemical analysis
ABF3 is a bifunctional a-L-arabinofuranosidase/xylobiohydro-
lase; the enzyme is active on pNPAra and pNPXyl with a K
M
of
0.65 and 12 mM, respectively [10]. XYL1 is specic for pNPXyl with
a K
M
of 0.55 mM. Both enzymes hydrolyze xylooligosaccharides
(X2-X5). XYL1 is active on wheat arabinoxylan liberating only xy-
lose, while ABF3 acting on the same substrate liberates both xylose
and arabinose.
ABF3 has two domains, a catalytic domain displaying a 5-bladed
b-propeller fold characteristic of GH 43 at the amino terminal end
and a carbohydrate binding module (CBM) with a b-sandwich fold
belonging to CBM family 6 at the carboxyl end [10]. XYL1 has only
one domain characteristic of glycoside hydrolase family 43.
Table 3 shows a comparison of the properties of fungal enzymes
from family 43 with b-xylosidase activity, both monofunctional
and bifunctional [5,10,4245]. Their subunit molecular weight
ranges from 37.5 to 83 kDa; the enzymes show an acidic pH opti-
mum, with an exception: the enzyme from Paecilomyces thermo-
phila has a pH optimum of 7.0. The temperature optimum ranges
from 20 to 50 C. Therefore, there is no clear pattern relating
bifunctionality (or monofunctionality) and properties of the
enzymes.
When the sequence similarity between ABF3 and XYL1 are ana-
lyzed, a 38% similarity is found between both enzymes; this value
increases to 52% when the catalytic modules are compared. XYL1
shows over 70% similarity when compared to the monofunctional
enzymes in Table 3, and a very low similarity with respect to the
bifunctional enzymes. This suggests a pattern of sequence similar-
ity between the monofunctional enzymes on one hand and the
bifunctional enzymes on the other.
Prediction of amino acid residues involved in ABF3 and XYL1 catalytic
mechanism
Molecular models of ABF3 and XYL1 were constructed to com-
pare at the atomic level the active sites of ABF3 and XYL1. Based on
sequence alignments we chose the crystallographic-resolved
structure of the glycosyl hydrolase family 43 arabinoxylan arabino-
furanohydrolase from B. subtilis (PDB ID: 3C7E) [24] as the three-
dimensional structural template to create both ABF3 and XYL1
homology-based models (Fig. 5A and B). Although both proteins
present higher identity with other members of the GH 43 family
of known structure such as an arabinan-specic alpha-1,2-arabin-
ofuranosidase from Cellvibrio japonicus (PDB ID: 3QED) or an arab-
inanase from B. subtilis (PDB ID: 1UV4), we chose the 3C7E protein
as template because it has both the catalytic and CBM domains.
ABF3 and XYL1 show sequence identity to that of the template of
31.5% and 31.3% respectively. We also compared the template
Fig. 4. Sequence alignment used to build the structural models. (A) Sequence alignment between structural template 3C7E and ABF3. (B) Sequence alignment between
structural template 3C7E and XYL1. The aligned catalytic residues are labeled with a red arrow. Images were obtained with ClustalX software. ClustalX provides an indication
of the quality of an alignment by plotting a conservation score for each column of the alignment. A high score indicates a well-conserved column; a low score indicates low
conservation. The quality curve is drawn below the alignment [25].
M.C. Ravanal et al. / Archives of Biochemistry and Biophysics 540 (2013) 117124 121
structure with secondary structure predictions for each protein.
Using this information to manually rene the sequence align-
ments, a sequence similarity of 56% and 46% for ABF3 and XYL1,
respectively was attained.
Barker et al. [35] described the catalytic mechanism of the
GH43 b-xylosidase family using the crystal structure of an invert-
ing family 43 b-xylosidase, from Geobacillus stearothermophilus
T-6 (XynB3) as a representative family member. The catalytic
mechanism relies on the participation of two Asp (15, 128) and
one Glu (187) residues, which are highly conserved in this protein
family. In the inverting reaction mechanism, Asp128 participates
stabilizing the transition state, Asp15 coordinates a water molecule
Fig. 5. Structural models of ABF3 (A) and XYL1 (D) and structure of the proteinligand complexes predicted by docking simulations (BCEF). (A)(D) The a-helices are
shown in purple and the b-strands are shown in yellow. With NewCartoon representation, the catalytic residues are shown in CPK drawing style. (A) The distance between
Asp12-Glu187 in ABF3 is shown with a dashed line. (D) Distance between Asp15-Glu240 in XYL1 is also shown with a dashed line. (B) ABF3-pNPAra (XP score = 5.13 kcal/
mol), (C) ABF3-pNPXyl (XP score = 5.59 kcal/mol), F) XYL1-pNPXyl (XP score = 5.14 kcal/mol) and E) XYL1-pNPAra (XP score = 3.43 kcal/mol). Ligand and protein
residues in the catalytic site are show in CPK drawing style. The catalytic residues are named in red. Other residues interacting with the ligands are named in black. Oxygens,
nitrogens and hydrogens are colored in red, blue and white, respectively. Carbon protein-atoms are colored in cyan, while the carbon ligand-atoms are colored in green. The
H-bonds between the amino acids and the ligand are shown with dotted lines. All the images were built with Visual Molecular Dynamics software (For interpretation of the
references to colour in this gure legend, the reader is referred to the web version of this article.) (VMD) [37].
Table 2
Summary of the kinetic and structural properties of each protein-ligand complex.
k
cat
/K
M
(min
-1
xM
-1
) Afnity energy (Kcal/mol) Total hydrophobic interactions
aa
Total H-bonds
b
H-bonds (with catalytic residues)
ABF3-pNPXyl 1.26 10
6
5.59 0 5 2
ABF3-pNPAra 2.52 10
5
5.13 1 3 1
XYL1-pNPXyl 2.50 10
4
5.14 1 4 1
XYL1-pNPAra - 3.43 1 3 0
The k
cat
and K
M
were determined experimentally and the afnity energies were calculated from XP scores.
a
Total hydrophobic interaction between ligand (pNPXyl or pNPAra) and active site residues.
b
Total H-bonds between active sites residues and ligand (pNPAra or pNPXyl).
122 M.C. Ravanal et al. / Archives of Biochemistry and Biophysics 540 (2013) 117124
which attacks the glycon C1 atom, while the resulting leaving
group is protonated in its glycosidic oxygen atom by Glu187, which
is protonated in the initial stages of the reaction. These catalytic
residues are also conserved in both ABF3 (Asp12, Asp127,
Glu187) and XYL1 (Asp15, Asp135, Glu240). According to the pro-
posed catalytic mechanism, the Glu residues are predicted to be
protonated in ABF3 (Glu187) and XYL1 (Glu240) at their enzyme
optimal pH (Table 4).
It is worth noting that after the energy minimization and
molecular dynamics structural relaxation, the distances between
the catalytic residues in ABF3 and XYL1 (Asp12-Glu187 and
Asp15-Glu240 respectively) (Fig. 5 A and 5D) are in agreement
with the distances proposed for enzymes following the inverted
glycosidase mechanism, where the pair of carboxylic acids at the
active site are located approximately 10 2 apart [38,39]. These
agreements with the proposed mechanism, led us to study at the
atomic level the interactions between each protein with pNPAra
and pNPXyl, to further explain the functional role of the key resi-
due that are responsible of the difference in substrate specicity
between ABF3 and XYL1.
Molecular docking simulations
Despite the fact that the geometrical and physicochemical prop-
erties of the catalytic site of GH43 are conserved in both ABF3 and
XYL1, our biochemical analysis shows that XYL1 is only active on
pNPXyl while ABF3 utilizes both pNPAra and pNPXyl.
Molecular docking simulations of the ABF3pNPXyl, ABF3
pNPAra, XYL1pNPAra and XYL1pNPXyl complexes were per-
formed to explain the functional differences between ABF3 and
XYL1. Their structure and relative binding afnities were measured
using the XP score as described by Friesner [36]. This score consid-
ers the Coulomb and van der Waals binding energies, and a penal-
ization term including solvation effects and ligand strains.
The theoretical binding afnities calculated for each complex
are in good agreement with the experimental analysis, where the
ABF3pNPXyl, ABF3pNPAra and XYL1pNPXyl complexes present
the best binding afnity according to the XP score and catalytic
efciency (k
cat
/K
M
). The XYL1pNPAra complex has the worst bind-
ing afnity, which correlates with the absence of activity observed
with this substrate (Table 2). The structures obtained for each com-
plex are in agreement with the experimental evidence presented
by Barker [35] and Rye [39]: the distance between the sugar ring
of the substrate and the carboxyl group of Asp12 (ABF3) or
Asp15 (XYL1) is optimal to intercalate a water molecule. However,
there are other interactions that are relevant to dene the catalytic
properties of each complex. The interactions between the sub-
strates and ABF3 or XYL1 were analyzed at the atomic level (Ta-
ble 2). In the complex with the best afnity (ABF3pNPXyl) the
ligand interacts (H-bond) with the two catalytic residues of the
protein (Asp127, Glu187) in agreement with the inverted glycosi-
dase mechanism, while in the ABF3-pNPAra and XYL1-pNPXyl
complexes, both ligands formH-bond with only one of the catalytic
residues (Asp127 and Glu240, respectively). On the other hand, in
the XYL1-pNPAra complex the substrate does not present any
interaction with the catalytic residues; this would explain why this
complex is not catalytically active.
H-bonds are the main interactions between the substrates and
the proteins (Table 2). The ABF3pNPXyl complex presents the
higher number of H-bonds and the best binding afnity, while, at
the same time, the complex with worst afnity present just 3 H-
bonds. The hydrophobic interactions also contribute to the recog-
nition of the substrate to the protein (Table 2). Our results show
specic pp interactions (T-shaped and stacking) which may con-
tribute to the specic recognition of the aromatic ring of pNPXyl
and pNPAra. The pp interactions are observed in 3 of the com-
plexes studied (ABF3: pNPAra-Trp191, XYL1: pNPXyl-Phe239 and
XYL1: pNPAra-Trp83), but not in the ABF3-pNPXyl complex, which
compensates for the loss of this interaction through electrostatic
interactions between the nitro group of pNPXyl and the His241
of ABF3. This interaction, however, is observed only in the ABF3
pNPXyl complex, while in the XYL1pNPAra complex the nitro
group interacts with the protein through H-bonds with the Ala84
and Ser30 residues (Fig 5BC and EF). On the other hand, in the
ABF3pNPAra and XYL1-pNPXyl complexes the nitro group does
not present specic interactions with the protein, allowing the ni-
tro group to be solvent accessible, thus facilitating its release in the
catalytic process. The nding that pNPAra is a competitive inhibi-
tor of XYL1, indicates that pNPAra does bind to the catalytic site
but not to the residues directly involved in catalysis, thus conrm-
ing the interactions proposed by the models.
Conclusions
(1) P. purpurogenum secretes two enzymes with -xylosidase
activity belonging to family 43 of the glycosyl hydrolases
with different substrate specicity against pNPAra and
pNPXyl.
(2) The structural analysis shows that active complexes (ABF3
pNPXyl, ABF3pNPAra and XYL1pNPXyl) possess specic
interactions (H-bonds) between substrates and catalytic res-
idues, which are absent in the inactive complex (XYL1-
pNPAra).
(3) pNPAra is substrate for ABF3 and competitive inhibitor for
XYL1; this nding is consistent with the structural models
and binding energies.
Table 3
Properties of fungal glycosyl hydrolases from family 43 with b-xylosidase activity.
Species Molecular weight (kDa) pH optimum Temperature optimum (C) Substrate specicity Refs.
Cochliobolus carbonum 42 5.5/6.5 37 pNPXyl/pNPAra [42]
Fusarium graminearum 42 6.0 30 pNPXyl
Fusarium graminearum 61 6.0 20/40 pNPXyl/pNPAra [43]
Paecilomyces thermophila 52.3 7.0 55 pNPXyl [44]
Penicillium herquei 37.5 6.5 30 pNPXyl [5]
Phanerochaete chrysosporium 83 5.0/5.5 45/50 pNPXyl/pNPAra [45]
Penicillium purpurogenum 50.7 5.0 50 pNPXyl/pNPAra [10]
Penicillium purpurogenum 49.0 6.0 40 pNPXyl This work
Table 4
Predicted protonation state of ABF3 and XYL1 residues.
Model protein pH protonated residues
ABF3 5 His: 18, 92, 118, 250, 256, 369, 433.
Glu: 125, 187, 188, 308.
XYL1 6 His: 316, 334.
Glu: 34, 240.
Asp: 47, 50, 86, 102, 232, 307.
M.C. Ravanal et al. / Archives of Biochemistry and Biophysics 540 (2013) 117124 123
Acknowledgments
This work has been supported by grants from FONDECYT No
1100084 and 1130180 (J.E) and 1131003 (F.G.N)) and Universidad
Andrs Bello (DI 0310/R and 61-12/R). MC. Ravanal is a recipient
of a FONDECYT Post-Doctoral Fellowship (3120032). We are thank-
ful to Dr. Daniel Aguayo for valuable suggestions during the prep-
aration of the manuscript.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.abb.2013.10.017.
References
[1] S. Malherbe, T. Cloete, Rev. Environ. Sci. Bio/Technol. 1 (2002) 105114.
[2] G. Williamson, P. Kroon, C. Faulds, Microbiology 144 (1998) 20112023.
[3] P. Biely, Trends Biotechnol. 3 (1985) 286290.
[4] S. Bhattacharyya, S. Khowala, A. Kumar, S. Sengupta, Biotechnol. Prog. 13
(1997) 822827.
[5] T. Ito, E. Yokoyama, H. Sato, M. Ujita, T. Funaguma, K. Furukawa, A. Hara, J.
Biosci. Bioeng. 96 (2003) 354359.
[6] T. Lembo, M. Azevedo, C.J. Bloch, E. Ferreira-Filho, World J. Microbiol.
Biotechnol. 22 (2006) 475479.
[7] F. Deleyn, M. Claeyssens, Can. J. Biochem. 56 (1977) 4350.
[8] N. Kitamoto, S. Yoshino, K. Ohmiya, N. Tsukagoshi, Appl. Environ. Microbiol. 65
(1999) 2024.
[9] S. Andrade, M.L.T. Polizeli, H. Terenzi, J. Jorge, Process Biochem. 39 (2004)
19311938.
[10] M.C. Ravanal, E. Callegari, J. Eyzaguirre, Appl. Environ. Microbiol. 76 (2010)
52475253.
[11] F. Zanoelo, M.L.T. Polizeli, H. Terenzi, J. Jorge, J. Ind. Microbiol. Biotechnol. 31
(2004) 170176.
[12] M. Wakiyama, K. Yoshihara, S. Hayashi, K. Ohta, J. Biosci. Bioeng. 106 (2008)
398404.
[13] J. Zhou, L. Bao, L. Chang, Z. Liu, C. You, H. Lu, Lett. Appl. Microbiol. 54 (2012)
7987.
[14] L. Contreras, J. Gmez, J. Prieto, J. Clemente-Jimnez, F. Las, Biochim. Biophys.
Acta 1784 (2008) 19241934.
[15] K. Wagschal, C. Heng, C.C. Lee, G.H. Robertson, W.J. Orts, D.W.S. Wong, Appl.
Biochem. Biotechnol. 155 (2009) 304313.
[16] J.M. Corra, L. Graciano, J. Abraho, E.A. Loth, R.F. Gandra, M.K. Kadowaki, C.
Henn, R.D.C.G. Simo, Applied Biochemistry and Biotechnology. 168 (2012)
22182229.
[17] E. Eneyskaya, H. Brumer, L. Backinowsky, D. Ivanen, A. Kulminskaya, K.
Shabalin, Carbohydr. Res. 338 (2003) 313325.
[18] C. Brx, A. Ben-David, D. Shallom-Shezi, M. Leon, K. Niend, G. Shoham, Y.
Shoham, D. Schomburg, J. Mol. Biol. 359 (2006) 97109.
[19] M. Hidalgo, J. Steiner, J. Eyzaguirre, Biotechnol. Appl. Biochem. 15 (1992) 185
191.
[20] D.M. Bollag, S.J. Edelstein, ISBN 0471568716, Wiley-Liss, New York, 1991. 230.
[21] M. Fritz, M.C. Ravanal, C. Braet, J. Eyzaguirre, Mycol. Res. 112 (2008) 933942.
[22] P. De Ioannes, A. Peirano, J. Steiner, J. Eyzaguirre, J. Biotechnol. 76 (2000) 253
258.
[23] A. Belancic, J. Scarpa, A. Peirano, R. Daz, J. Eyzaguirre, J. Biotechnol. 41 (1995)
7179.
[24] E. Vandermarliere, T.M. Bourgois, M.D. Winn, S. Van Campenhout, G. Volckaert,
J.A. Delcour, Biochem. J. 418 (2009) 3947.
[25] M.A. Larkin, G. Blackshields, N.P. Brown, R. Chenna, P.A. McGettigan, H.
McWilliam, F. Valentin, I.M. Wallace, A. Wilm, R. Lopez, J.D. Thompson, T.J.
Gibson, D.G. Higgins, Bioinformatics (Oxford, England) 23 (2007) 29472948.
[26] A. ali, T.L. Blundell, Journal of Molecular Biology. 234 (1993) 779815.
[27] A.D. MacKerell, D. Bashford, J. Phys. Chem. B. 102 (1998) 35863616.
[28] J.C. Phillips, R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E. Villa, C. Chipot,
R.D. Skeel, L. Kal, K. Schulten, J. Comput. Chem. 26 (2005) 17811802.
[29] S.M. McDonald, D. Aguayo, F.D. Gonzalez-Nilo, J.T. Patton, J. Virol. 83 (2009)
61356148.
[30] M. Olsson, C. Sndergaard, M. Rostkowski, J. Jensen, J. Chem. Theory Comput. 7
(2011) 525537.
[31] Schrdinger Suite 2012 Protein Preparation Wizard; Epik version 2.3,
Schrdinger LLC., New York, NY, 2012; Impact version 5.8, Schrdinger LLC.,
New York, NY, 2012; Prime version 3.1, Schrdinger LLC., New York, NY, 2012.
[32] W.L. Jorgensen, J. Chandrasekhar, J.D. Madura, R.W. Impey, M.L. Klein, J. Chem.
Phys. 79 (1983) 926935.
[33] Maestro (Maestro, version 9.3, Schrdinger LLC., New York, NY, 2012). LigPrep
2.5 (LigPrep, version 2.5, Schrdinger LLC., New York, NY, 2012.).
[34] Glide, version 5.8, Schrdinger LLC., New York, NY, 2012.
[35] I.J. Barker, L. Petersen, P.J. Reilly, J. Phys. Chem. B 114 (2010) 1538915393.
[36] R.A. Friesner, R.B. Murphy, M.P. Repasky, L.L. Frye, J.R. Greenwood, T.A.
Halgren, P.C. Sanschagrin, D.T. Mainz, J. Med. Chem. 49 (2006) 61776196.
[37] W. Humphrey, A. Dalke, K. Schulten, J. Mol. Graph. 14 (1996) 3338.
[38] J. McCarter, S. Withers, Curr. Opin. Struct. Biol. 4 (1994) 885892.
[39] C.S. Rye, S.G. Withers, Curr. Opin Chem. Biol. 4 (2000) 573580.
[40] M. Wiederstein, M.J. Sippl, Nucleic Acids Ress. 35 (2007) W407W410.
[41] S. Wegener, R.F. Ransom, J.D. Walton, Microbiology (Reading, England). 145
(1999) 10891095.
[42] R.F. Ransom, J.D. Walton, Carbohydr. Res. 297 (1997) 357364.
[43] R. Carapito, C. Carapito, J.-M. Jeltsch, V. Phalip, Bioresour. Technol. 100 (2009)
845850.
[44] C. Teng, H. Jia, Q. Yan, P. Zhou, Z. Jiang, Bioresource Technology. 102 (2011)
18221830.
[45] N.D. Huy, P. Thayumanavan, T.-H. Kwon, S.-M. Park, J. Biosci. Bioeng. 116
(2013) 152159.
124 M.C. Ravanal et al. / Archives of Biochemistry and Biophysics 540 (2013) 117124