Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure and substrate specificity analysis

Article abstract of DOI:10.1016/j.carres.2017.07.005

The GH94 glycoside hydrolase cellodextrin phosphorylase (CDP, EC 2.4.1.49) produces cellodextrin oligomers from short β-1→4-glucans and α-D-glucose 1-phosphate. Compared to cellobiose phosphorylase (CBP), which produces cellobiose from glucose and α-D-glucose 1-phosphate, CDP is biochemically less well characterised. Herein, we investigate the donor and acceptor substrate specificity of recombinant CDP from Ruminiclostridium thermocellum and we isolate and characterise a glucosamine addition product to the cellobiose acceptor with the non-natural donor α-D-glucosamine 1-phosphate. In addition, we report the first X-ray crystal structure of CDP, along with comparison to the available structures from CBPs and other closely related enzymes, which contributes to understanding of the key structural features necessary to discriminate between monosaccharide (CBP) and oligosaccharide (CDP) acceptor substrates.

Full text of DOI:10.1016/j.carres.2017.07.005

Carbohydrate Research xxx (2017) 1e15
Contents lists available at ScienceDirect
Carbohydrate Research
journal homepage: www.elsevier.com/locate/carres
Cellodextrin phosphorylase from Ruminiclostridium thermocellum:
X-ray crystal structure and substrate specificity analysis
Ellis C. O'Neill 1 , Giulia Pergolizzi , Clare E.M. Stevenson, David M. Lawson,
,
2
2
*
Sergey A. Nepogodiev, Robert A. Field
Department of Biological Chemistry, John Innes Centre, Norwich Research Park, Norwich NR4 7UH, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 15 May 2017
Received in revised form
The GH94 glycoside hydrolase cellodextrin phosphorylase (CDP, EC 2.4.1.49) produces cellodextrin
oligomers from short
b-1/4-glucans and
a-D-glucose 1-phosphate. Compared to cellobiose phos-
phorylase (CBP), which produces cellobiose from glucose and
a-D-glucose 1-phosphate, CDP is bio-
17 July 2017
chemically less well characterised. Herein, we investigate the donor and acceptor substrate specificity of
recombinant CDP from Ruminiclostridium thermocellum and we isolate and characterise a glucosamine
addition product to the cellobiose acceptor with the non-natural donor a-D-glucosamine 1-phosphate. In
Accepted 17 July 2017
Available online xxx
addition, we report the first X-ray crystal structure of CDP, along with comparison to the available
structures from CBPs and other closely related enzymes, which contributes to understanding of the key
structural features necessary to discriminate between monosaccharide (CBP) and oligosaccharide (CDP)
acceptor substrates.
Keywords:
Cellodextrin phosphorylase
Glucosamine 1-phosphate
X-ray crystal structure
©
2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
http://creativecommons.org/licenses/by/4.0/).
(
1
. Introduction
and analogues, suitable for a wide variety of applications, spanning
from paper products, to textiles, food thickeners and stabilizers, to
composite materials and hydrogels for sensors development,
medical, electronic and pharmaceutical applications [2,6,10e12].
However, routine access to pure cellulose at scale is challenging due
to its association with hemicellulose and lignin in plant materials
[13]. In contrast, bacterial cellulose is synthesised in a much purer
form, albeit with a different crystalline structure [3,14].
b-1/4-Linked glucan polysaccharides are widespread in na-
ture, where they are principally involved in structural roles [1]. The
most common example, cellulose, is the main component of pri-
mary and secondary plant cell walls [1,2]; however, cellulose is also
found in bacteria [3], algae [4], and oomycetes [5]. It is composed of
extensive, unbranched chains of
b-1/4-linked D-glucose (Glc),
where the residues alternate orientation such that the overall
molecular structure may be considered as a repeating polymer of
cellobiose blocks [1,2]. Extensive intra/inter chain hydrogen bonds
The prospect of using cellulose-producing enzymes in vitro as an
eco-friendly alternative to obtain pure cellulose is potentially
attractive, although the natural biosynthetic machinery comprises
a very complex, multi-protein trans-membrane system [15]. As an
alternative, cellulose-like material has been synthesised by
Kobayashi and co-workers using a hydrolytic cellulase in a reverse
can be generated by b-1/4-glucans, which support the arrange-
ment of glucan chains in microfibrils with highly ordered crystal-
line regions [1,2,6,7], conferring upon them stiffness and high
resistance to thermal and enzymatic degradation. Among other
synthetic reaction with
b-cellobiosyl fluoride as substrate in a
common
b
-1/4-glucans are the xyloglucans, which span cellulose
mixture of acetonitrile/buffer [16]. Recently, phosphorylases have
received attention as catalysts for glycoside synthesis [17,18],
microfibrils, generating a 3D network in the plant cell wall [8,9]. Its
unique physical-chemical properties make cellulose, its derivatives
including the production of
a- and b-1/4-linked glucans. For
instance, the plant -1/4-glucan phosphorylase PHS2 has been
a
used to generate starch-like materials [19], while cellodextrin
phosphorylase (CDP) from Ruminiclostridium thermocellum
*
Corresponding author.
E-mail address: rob.field@jic.ac.uk (R.A. Field).
Present address: Department of Plant Science, University of Oxford, Oxford OX1
(
formerly Clostridium thermocellum) [20e22] has been shown to be
a suitable enzyme for cello-oligosaccharide and cellulose produc-
tion [23,24] using short cellodextrins and -D-glucose 1-phosphate
(Glc-1-P) as substrates. The latter reactions are simple, eco-friendly
1
a
3
RB, UK.
2
These authors contributed equally to this study.
http://dx.doi.org/10.1016/j.carres.2017.07.005
0
008-6215/© 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

2
E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
and produce cellulose oligomers with different degrees of poly-
merization (DP) depending on the concentration of the substrates
and compare it to known structures of CBP.
[
23]. Hiraishi et al. described the synthesis of crystalline cellulose-
2
. Results and discussion
like material with an average of ~ DP 9 using high concentration
of glucose as CDP acceptor [24].
CDP (EC 2.4.1.49) belongs to the glycoside hydrolase family
GH94 in the Carbohydrate Active Enzyme (CAZY) database (URL:
http://www.cazy.org/) [25], along with cellobiose phosphorylase
2.1. Protein expression and activity assay
The cdp gene from Ruminiclostridium thermocellum YM4 strain
(GenBank accession number BAB71818) was synthesised with
(
CBP), which has been extensively characterised from a variety of
codon optimization for expression in E. coli and sub-cloned into
pET15b, which inserts a hexahistidine tag behind a thrombin
cleavage site at the N-terminus of the protein (see supplementary
information for nucleotide and amino acids sequences). Protein was
expressed and purified using a combination of nickel affinity and
gel filtration chromatography (Supplementary Figs. S1eS2). The
resulting CDP protein consists of 1009 amino acids, with a molec-
ular weight of 114.364 KDa per monomer, in accordance with the
GF elution profile. A yield of ~10 mg of purified CDP per litre of
culture was obtained, which was concentrated to 40 mg/ml and
sources [26e29]. For a summary of established CBP acceptor and
donor specificity, see Tables S1 and S2. Less comprehensive studies
have been conducted on CDP activity and specificity; it has been
used to synthesise a variety of cellulose derivatives, assessing its
permissiveness toward acceptors (Table S1), but less information is
available about its donor specificity [30,31] (Table S2) and no X-ray
crystal structure is available for this enzyme. However, recombi-
nant CDP can be produced in high yield in E. coli [32,33] and it is
ꢀ
stable up to 60 C with highest activity at pH 7.5 [33], making it
suitable for process development. Protein sequence alignment of
R. thermocellum CDP with CBP from the same organism shows that
the two enzymes share only ~17% identity. To fully understand how
CDP and CBP discriminate between glucose and cello-
oligosaccharide acceptor substrates a structural comparison of
CDP and CBP would be informative.
Herein, we report studies that investigate the donor and
acceptor specificity of recombinant R. thermocellum CDP. Where
low or no turnover was observed, additional inhibition experi-
ments were performed to probe the interaction between the sugar
ꢀ
stored at ꢁ80 C until required.
The ability of CDP to synthesise and phosphorolyse cello-
oligosaccharides was monitored using capillary electrophoresis
with laser-induced fluorescence detection to assess the degree of
polymerization (Fig. 1) [34,35]. Cello-oligosaccharides were
labelled by reductive amination with the fluorophore 8-amino-
1,3,6-pyrenetrisulfonic acid (APTS) [34,35]. CDP was able to extend
APTS-labelled cellotriose, ( -1/4-Glc) -APTS (Fig. 1, red), to olig-
b
3
omers up to DP 16 (Fig. 1, black) by transferring glucose from Glc-1-
P on to the acceptor, although most of the synthesised material was
insoluble and removed during the sample preparation. Indeed,
cello-oligosaccharides beyond ~ DP 9 are known to have limited
1
-phosphates or oligosaccharides and the enzyme. In addition, X-
ray crystallography was used to characterise the structure of CDP
Fig. 1. Carbohydrate electrophoresis of CDP-synthesised oligomers and reaction scheme. The activity of the phosphorylase was confirmed by assaying the ability of CDP (1
mg) to
transfer Glc from Glc-1-P (disodium salt, 50 mM) on to ( -1/4-Glc) -APTS (5 mM) in 100 l HEPES buffer (50 mM, pH 7.6) (all concentrations are final concentrations). After 2 h at
b
3
m
ꢀ
ꢀ
40
C, the reactions were terminated by heating to 95 C in a boiling water bath for 5 min and centrifuging at 16,000 g for 5 min. 20
by degradation with endo-cellulase (0.7 U) from T. longibrachiatum (Megazyme) in HEPES buffer (50 mM, pH 7.6) for 2 h at 40 C, or CDP (5
buffer, 3 mM potassium chloride and 140 mM sodium chloride, pH 7.4), for 1 h at 40 C, before again heating to 95 C and centrifuging as before.
m
l of this synthetic reaction were further probed
ꢀ
mg) in 1 ꢂ PBS buffer (0.01 M phosphate
ꢀ
ꢀ
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
3
aqueous solubility [24]. CDP could also phosphorolyse the CDP-
synthesised APTS-labelled oligomers, reducing the chain length
up to ( -1/4-Glc) -APTS (Fig. 1, blue). Endo-cellulase from Tri-
choderma longibrachiatum was able to hydrolyse the CDP-
synthesised APTS-labelled oligomers (Fig. 1, black) to ( -1/4-
Glc) -APTS and ( -1/4-Glc) -APTS (Fig. 1, green), confirming the
CDP-generated material was indeed -1/4-linked glucan.
established that synthetic
a
-D-glucose 1-fluoride (Glc-1-F) was not
an effective donor substrate for CDP (Fig. 2A, green), at least under
the assay conditions used for CE analysis. However, we note that
Nakai and co-workers showed CDP-catalysed addition of Glc from
Glc-1-F onto cellobiose in 68% yield despite working with CDP from
the same organism [36].
b
3
b
3
b
2
b
A series of non-cognate sugar 1-phosphate donors e
galactose 1-phosphate (Gal-1-P), -D-mannose 1-phosphate (Man-
-P), -D-xylose 1-phosphate (Xyl-1-P) and -D-glucosamine 1-
phosphate (GlcN-1-P) e were also assessed as prospective CDP
substrates, with ( -1/4-Glc) -APTS as the acceptor. In the pres-
ence of Gal-1-P (Fig. 2A, magenta), the formation of a peak with DP
and a peak with DP 4 may be accounted for addition of a single
Gal onto the acceptor and concurrent phosphorolysis of remaining
-1/4-Glc) -APTS acceptor to ( -1/4-Glc) -APTS. Indeed, it has
been previously reported that Gal-1-P is a donor for CDP from
C. stercorarium, even if the kcat/K for Gal-1-P with CDP is ~1% of
that for Glc-1-P [31]. In the reaction with Man-1-P (Fig. 2A, purple),
a-D-
a
1
a
a
2.2. Donor specificity of CDP
b
3
In order to determine the capability of CDP for the general
synthesis of -1/4-linked oligomers not based on glucose, the
transfer of other sugars onto ( -1/4-Glc) -APTS, an efficient
b
2
b
3
acceptor (Fig. 2A, black), was monitored using CE. There was little
material visible in the reaction using Glc-1-P (Fig. 2A, red), as the
longer oligomeric products formed were insoluble and so removed
during sample preparation for the CE analysis. Evaluation of fluo-
(b
3
b
2
M
ride as
a reactive alternative leaving group to phosphate
Fig. 2. Donor specificity of CDP analysed using carbohydrate electrophoresis. A. Extension of (
b
-1/4-Glc)
3
-APTS (black) with various sugar 1-P donors (15 min run). B.
ꢀ
Extension of ( -1/4-Glc) -APTS (black) with GlcN-1-P (pink, 30 min run). Assays were carried out using CDP (5
b
3
mg/ml) at 40 C with (
b-1/4-Glc)
3
-APTS acceptor (2
mM), Glc-1-
P (disodium salt, 1 mM), or other sugar 1-P donors (10 mM, Gal-1-P: dipotassium salt pentahydrate; Man-1-P: disodium salt; Xyl-1-P: bis(cyclohexylammonium) salt) in HEPES
ꢀ
buffer (50 mM, pH 7.5) (all concentrations are final concentrations), followed by heating to 95 C in a boiling water bath for 5 min and centrifuging at 16,000 g for 5 min. CE was
performed under standard conditions. The symbol * highlights that the DP of marked peaks does not correspond to the DP of all other neutral species peaks. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version of this article.)
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

4
E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
a small additional peak may correspond to the addition of one
mannose residue or it may be accounted for acceptor dispropor-
tionation [30]; this could be due to any residual phosphate present
in the reaction mixture, which would allow phosphorolysis of the
APTS-labelled acceptor and utilisation of the resulting Glc-1-P for
2
donor and (b-1/4-Glc) as good CDP acceptor but only a very weak
substrate for interfering phosphorolysis. This reaction generated
mainly a single product in accordance with what was observed by
CE (Fig. 2A, blue and 2B, pink). As far as we are aware, there is no
precedent for CDP or CBP from any organism using GlcN-1-P as a
donor substrate. The crude reaction mixture was fractionated by
strong cation exchange chromatography to isolate any positively
charged product, which was then eluted with a 0.05 M ammonium
bicarbonate buffer (pH 9.4). The collected fractions were analysed
by MALDI-ToF and ESI-MS and showed the presence of a trisac-
charide containing GlcN instead of Glc (ꢁ1 m/z unit difference,
Supplementary Figs. S3A and S3B) as the main product; however,
traces of multiple Glc addition products were observed in variable
amounts from batch to batch, presumably arising from a limited
3
transfer of Glc onto another molecule of (b-1/4-Glc) -APTS
acceptor. Where Xyl-1-P was used (Fig. 2A, olive), a very limited
turnover was evident, with possible xylose transfer product for-
mation suggested by a new peak that appears at a slightly faster
4
retention time than the corresponding (b-1/4-Glc) -APTS. Shin-
tate and co-workers previously demonstrated that CDP from
R. thermocellum can utilise Xyl-1-P as donor, although its activity is
only a few % of that of Glc-1-P [30]. In stark contrast, the reaction
with GlcN-1-P showed almost complete conversion of the acceptor
substrate to a single much later-running peak on the electrophe-
rogram (Fig. 2A, blue). This could be the product of a single turnover
arising from the addition of one glucosamine (GlcN) residue which,
at the pH of the running buffer (pH 4.75), is protonated and could
account for the long retention time. In more extended electro-
phoresis runs, another weak peak is evident, suggesting the
possible addition of a further GlcN unit (Fig. 2B, pink).
phosphorolysis of (
addition, H-NMR spectroscopy data for the product in D
b
-1/4-Glc)
2
generating Glc-1-P in situ. In
O (pD 7)
1
2
in comparison with cellobiose and cellotriose standards confirmed
the attachment of a GlcN unit to cellobiose due to the presence of a
0
00
third anomeric proton and to the characteristic signal of H-2 at
2.72 ppm [37] (Fig. 3, Table S3). 2D-COSY and 2D-HSQC spectra
(Supplementary Table S3, Figs. S4 and S5), in comparison with
those for cellotriose, supported the identity of the main product as
2
(b-1/4-GlcN(Glc) ).
2
1
.3. Isolation and characterization of the GlcN addition product, (
/4-GlcN(Glc)
b-
2
)
2.4. Competition experiment with mannose 1-phosphate donor
In order to confirm the identity of any late running product
observed by CE, reactions with unlabelled acceptor were scaled-up
to monitor them by thin layer chromatography (TLC) and to isolate
the products for further NMR and MS characterization. In particular,
we focused our attention on the reaction of CDP with GlcN-1-P as
In order to assess the binding of donor analogues that are
extremely poor substrates for CDP, we screened them as prospec-
tive inhibitors of the natural CDP substrates. Competition assays
were performed to understand whether the lack of efficient activity
Fig. 3. 1H-NMR of isolated (
-1/4-GlcN(Glc) ) (red). The spectra were recorded at rt and referenced to HOD (d
H
b
-1/4-GlcN(Glc)
) product in comparison with
b
-1/4-glucan standards in D
2
2 2 3
O. Cellobiose, (b-1/4-Glc) (blue); cellotriose, (b-1/4-Glc) (green);
0
00
000
(
b
2
4.79). HOD signal is omitted for clarity. The symbols , and denote the first, second and
third glycosyl residue from the reducing end, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
5
Fig. 4. Man-1-P binding inhibits CDP activity. CDP oligomerization of (
b
-1/4-Glc)
6
-APTS (black) with Glc-1-P (blue) in the presence of Man-1-P (pink). Assays were carried out
ꢀ
using CDP (5
mg/ml) at 40 C with (
b-1/4-Glc)
6
-APTS (2
m
M), Glc-1-P (disodium salt, 1 mM) in HEPES buffer (50 mM, pH 7.5) adding Man-1-P (disodium salt, 10 mM) (all
ꢀ
concentrations are final concentrations), followed by heating to 95 C in a boiling water bath for 5 min and centrifuging at 16,000 g for 5 min. CE was performed under standard
conditions. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
for sugar 1-phosphate donors different from Glc-1-P is caused by
failure to bind them or inefficient catalysis. For Gal-1-P reactions
Glucose was found to be a poor acceptor [24], whilst the mono-
saccharide derivative phenyl -glucopyranoside was a better
b-D
with CDP from C. stercorarium, it has been shown that K
M
is
acceptor; the latter effect could be due to its fixed anomeric
configuration and/or to aglycone aryl groups commonly known for
increased ~12 fold and kcat reduced ~10 fold with respect to the
corresponding reaction with Glc-1-P, for instance [31]. Xyl-1-P is
known to be a poor donor substrate for CDP compared to Glc-1-P
binding well into sugar binding sites. According to the literature,
app
the cello-oligosaccharides displayed a decreasing K
M
up to DP 5
[
30]; nevertheless, Shintate and co-workers were able to enzy-
matically synthesise series of -1/4-linked-hetero-
oligosaccharides containing Xyl and Glc [30].
Reaction mixtures containing fixed concentrations of Glc-1-P
mM) and the acceptor -1/4-Glc) -APTS (2 M) had
(from 2.6 mM down to 0.36 mM) [24], indicating there are stabil-
a
b
ising interactions with the sugars remote from the catalytic site.
app
However, cellohexaose has a higher K
M
(1.9 mM), which may
reflect its modest aqueous solubility. The kcat values follow a similar
pattern with respect to acceptor chain length, dropping from DP
(
1
(
b
6
m
mannose 1-phosphate (10 mM) added to assess the impact on
glucan polymerization (Fig. 4). Man-1-P caused a great decrease in
glucan oligomer length (Fig. 4, pink), suggesting binding to the
active site although we were unable to demonstrate CDP-catalysed
mannose transfer from Man-1-P.
2e5 and increasing again at DP 6. The net result is that for acceptors
app
from DP 2e6, the catalytic efficiency, as judged by kcat/K
M
, is
consistent within a factor of 2.
The ability of CDP to use APTS labelled (
Glc) and ( -1/4-Glc) was monitored over time using CE
Supplementary Fig. S7). ( -1/4-Glc) -APTS is a very poor CDP
acceptor, as shown by its slow consumption by CDP [24]; when a
single glucose residue is added, the formed ( -1/4-Glc) -APTS is a
much better acceptor and it is rapidly oligomerised to insoluble
cello-oligosaccharides (Supplementary Fig. S7A). ( -1/4-Glc)
APTS is a much better acceptor than ( -1/4-Glc) -APTS, being
completely consumed within 10 min (Supplementary Fig. S7B). (
/4-Glc) -APTS is also a good acceptor for CDP, forming longer
2
b-1/4-Glc) , (b-1/4-
3
b
6
(
b
2
2
2
.5. Acceptor specificity of CDP
b
3
.5.1. Acceptor specificity for
b-1/4-glucans
b
3
-
The length specificity of CDP for unlabelled reducing glycan
acceptors was determined by measurement of phosphate release
from 10 mM Glc-1-P over a range of acceptor concentrations
b
2
b
-
1
6
(Supplementary Fig. S6). Using this assay, the capability of CDP to
oligomers rapidly (Supplementary Fig. S7C). This can be seen in the
CE as a non-processive reaction, with all oligomers extending in
parallel until a combination of chain length and time results in
precipitation.
use cello-oligosaccharides as acceptors was assessed (Table 1).
Table 1
Acceptor length specificity for CDP. The length specificity of CDP for reducing
glycan acceptors was determined by measurement of phosphate release from
2
.5.2. Acceptor specificity for
Kadokawa and co-workers have extensively investigated the
flexibility of -glucan phosphorylases towards the insertion of
single or multiple non-cognate monosaccharides into maltooligo-
saccharides [37e42]. In particular, -glucan phosphorylase from
potato was able to effect single addition of GlcN onto maltotetraose
38]. Later, a thermostable -glucan phosphorylase from Aquifex
b-1/4-glycans
10 mM Glc-1-P (disodium salt) over a range of acceptor concentrations. Michaelis-
Menten graphs were plotted using Grafit (Erithacus Software Ltd) (Supplementary
Fig. S6).
a
app
app
Acceptor
Glc
K
M
(mM)
k
cat (1/s)
k
cat/K
M
(1/mM/s)
a
n.d.
-glucopyranoside 24 ± 13
n.d.
15 ± 6.3
n.d.
0.63
phenyl
b
-
D
[
a
(
(
(
(
(
b
b
b
b
b
-1/4-Glc)
-1/4-Glc)
-1/4-Glc)
-1/4-Glc)
-1/4-Glc)
2
3
4
5
6
2.6 ± 0.18
0.68 ± 0.076 9.5 ± 0.35 14
0.54 ± 0.13 5.0 ± 0.25 9.3
0.36 ± 0.076 4.3 ± 0.47 12
1.9 ± 0.76 7.6 ± 1.8 4.0
17 ± 0.50 6.5
aeolicus was shown to randomly incorporate GlcN [37,41,42] or
GlcA [39,41] onto maltooligosaccharides to provide new chitin/
chitosan/glycosaminoglycan-like glyco-materials, which could
find applications in drug delivery. The introduction of a GlcN into
cellulose oligomers could confer them with interesting properties,
n.d. ¼ not determined.
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

6
E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
Fig. 5. (
b
-1/4-GlcN(Glc)
2
) as CDP acceptor: CE analysis. A. CDP oligomerization of (
b
-1/4-GlcN(Glc)
) (5 mM) in NaOAc buffer (50 mM, pH 5) (all concentrations are final concentrations) for 1 h, followed by
l of this synthetic reaction were labelled with APTS and analysed by CE. For comparison,
2
) followed by APTS labeling. Assays were carried out using CDP (40 mg/
ꢀ
ml) at 40 C with Glc-1-P (disodium salt, 50 mM) and (
heating to 95 C in a boiling water bath for 5 min and centrifuging at 16,000 g for 5 min. 15
b-1/4-GlcN(Glc)
2
ꢀ
m
CDP oligomerization with (
performed under standard conditions. B. CDP inhibition by (
b
-1/4-Glc)
3
(green) is shown: oligomers containing GlcN differ 1 min compared to Glc oligomers, which differ 0.7 min in CE electropherogram. CE was
ꢀ
b-1/4-GlcN(Glc)
2
). Assays were carried out using CDP (40
m
g/ml) at 40 C with Glc-1-P (disodium salt, 50 mM) and (
b-
ꢀ
1
/4-Glc) -APTS (0.5 mM) in HEPES buffer (50 mM, pH 7.6) adding (
3
b-1/4-GlcN(Glc)
2
) (5 mM) (all concentrations are final concentrations) for 1.5 h, followed by heating to 95 C in
a boiling water bath for 5 min and centrifuging at 16,000 g for 5 min. CE was performed under standard conditions. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)
such as pH-responsiveness and improved solubility. Therefore, we
tested whether the isolated trisaccharide ( -1/4-GlcN(Glc) ),
formed by the addition of GlcN onto cellobiose (section 2.3), could
subsequently serve as an acceptor for CDP with Glc-1-P as donor.
CDP successfully extended the trisaccharide, as shown by MALDI
ms: the generated cello-oligomers containing a single GlcN residue
further confirm that Glc was added at the non-reducing end with
1/4 linkage rather than onto the reducing end with -1/1 link-
age, reduction of the reducing end with NaBD was carried out and
confirmed by a mass increment of 3 Da observed by MALDI ms
(Supplementary Fig. S9). A competition experiment was set up to
probe CDP specificity toward GlcN-containing cello-oligosaccha-
b-
b
2
b
4
(
Supplementary Fig. S8, blue) differed by 162 m/z, but had a ꢁ1 m/z
ride acceptors. When (
excess to ( -1/4-Glc)
Glc) -APTS reduced substantially, suggesting the ability of GlcN-
b
-1/4-GlcN(Glc)
2
) was added in 10 fold
unit difference compared to pure cello-oligomers (Supplementary
Fig. S8, red). To exclude the possibility of the CDP trisaccharide
b
3
-APTS, the average extension of (
b-1/4-
3
elongation on the reducing Glc, forming a
b
-1/1 linkage [43,44],
containing cello-oligosaccharides to compete with
oligosaccharides for CDP active site (Fig. 5B).
b-1/4-glucan
the availability of the reducing end of the generated GlcN-
containing cello-oligomers was confirmed APTS-labeling them by
reductive amination (Fig. 5A, red). Successful APTS-labeling
allowed analysis of the products by CE, where peaks correspond-
ing to cello-oligomers containing GlcN differed by 1 min in reten-
tion time in the electropherogram (Fig. 5A, red) compared to peaks
Xylotriose (
labelled with APTS and assessed as acceptors for CDP (Fig. 6A). (
1/4-Xyl) -APTS is a reasonable acceptor of Glc forming long
3 3
b-1/4-Xyl) and mannotriose (b-1/4-Man) were
b
-
3
oligomers (Fig. 6A, purple), which suggests that the 6-OH is not
required in the CDP acceptor binding site, which is consistent with
the results from Shintate and co-workers [30]. On the other hand,
of b-1/4-glucan oligomers differing by 0.7 min (Fig. 5A, green). To
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
7
CDP is not able to transfer glucose from Glc-1-P onto (
b
-1/4-
2.6. Apo structure of CDP
Man) -APTS and there is no apparent phosphorolysis (Fig. 6A,
3
dark green). This suggests that the acceptor binding site of CDP
cannot tolerate an axial 2-OH.
In order to probe CDP specificity toward xylo- and manno-
oligosaccharide acceptors, competition experiments were carried
The structure of the CDP was solved to 2.3 Å resolution,
revealing two copies of the protein subunit per asymmetric unit
forming a clear dimer (Fig. 7A). There are a number of close
structural homologues of CDP in the PDB with the top 22 entries
giving DALI Z scores in the range 37e40 (and aligning with >73% of
the structure) [45], although these represent just five different
proteins: cellobiose phosphorylases (CBP) from Cellulomonas uda
[46e48], Cellovibrio gilvus [27] and Ruminiclostridium thermocellum
[29], chitobiose phosphorylase from Vibrio proteolyticus (ChBP) [49]
and cellobionic acid phosphorylase from Saccharophagus degradans
(CBAP) [50]. Moreover, they all form dimers akin to that of CDP. To
simplify the subsequent discussion, given the similarities between
CBP, ChBP and CBAP we shall refer to them all as CBP, unless
specified otherwise, and a ligand bound structure of Cellovibrio
gilvus CBP (PDB code 3QG0; contains phosphate, deoxynojirimycin
out and monitored by CE (Fig. 6B). When a large excess of (
Xyl) was added (10 mM vs 13.5 M APTS-labelled acceptor)
Fig. 6B, green), there was a significant decrease in the average
extension of the ( -1/4-Glc) -APTS suggesting binding of xylo-
triose to the active site of the enzyme. In contrast, addition of (
/4-Man) (10 mM vs 13.5 M APTS-labelled acceptor) had almost
no impact on the average extension of the ( -1/4-Glc) -APTS
b-1/4-
3
m
(
b
6
b
-
1
3
m
b
6
acceptor, indicating that the mannose-containing trisaccharide
does not compete for the active site (Fig. 6B, pink).
Fig. 6.
b
-1/4-glycans as CDP acceptors: CE analysis. A. CDP oligomerization of (
b
-1/4-Xyl)
3 3
-APTS and (b-1/4-Man) -APTS. Assays were carried out using CDP (5 mg/ml) at
ꢀ
40
C with Glc-1-P (disodium salt, 10 mM) and APTS-labelled acceptor (13.5
mM) in HEPES buffer (50 mM, pH 7.5) (all concentrations are final concentrations) for 20 min, followed
ꢀ
by heating to 95 C in a boiling water bath for 5 min and centrifuging at 16,000 g for 5 min. CE was performed under standard conditions. B. CDP inhibition by
b-1/4-glycans.
ꢀ
Assays were carried out using CDP (5
10 mM) and ( -1/4-Man)
min. CE was performed under standard conditions.
m
g/ml) at 40 C with Glc-1-P (disodium salt, 10 mM) and (
b-1/4-Glc)
6
-APTS (13.5
m
M) in HEPES buffer (50 mM, pH 7.5) adding (
b
-1/4-Xyl)
3
ꢀ
(
5
b
3
(10 mM) (all concentrations are final concentrations), followed by heating to 95 C in a boiling water bath for 5 min and centrifuging at 16,000 g for
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

8
E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
Fig. 7. Comparison of the structures of CDP, CBP and SOGP. A. The top two images show orthogonal perspectives (“top” and “side” views) of the ligand free structure of
R. thermocellum CDP homodimer (PDB code 5NZ7) in cartoon representation, this being more complete than the ligand bound structure (PDB code 5NZ8). Superimposed on this are
the two copies each of cellotetraose (green spheres) and phosphate (black spheres) taken from the latter (PDB code 5NZ8). One subunit is coloured according to the domains shown
schematically in panel B and Supplementary Fig. S10, whilst the other is coloured in grey. The open arrow indicates the direction of view for panel C. The lower two images show
corresponding top views for the C. gilvus CBP homodimer (PDB code 3QG0) and the L. phytofermentans SOGP monomer (PDB code 5H41). The dashed red-circles indicate
interactions between the N-terminal arms that project from the N-terminal domains of CDP and SOGP and their structurally conserved -sandwich domains. Note the differences in
structure and placement of these N-terminal domains; in SOGP this domain sits in the equivalent place to that occupied by the -sandwich domains of the opposing subunits in CDP
b-sheet
b
b
and CBP homodimers and thus explains why this enzyme is monomeric. B. Linear representation of the domain structures of CDP, CBP and SOGP. C. The top panels show molecular
surfaces for the monomers of CDP, CBP and SOGP (direction of view shown in part A; looking at the occluded surface in the case of the two that form dimers). The lower panels show
simplified representations of the corresponding views of the biological units (which is also a monomer for SOGP), where the various domains are shown as ellipsoids. In the case of
the two dimers, the domains of the rear subunit are in solid colour, whilst those in the front subunit are semi-transparent colour and have a black border. The N-terminal extensions
in CDP and SOGP are shown as “arms”; for simplicity, the two
white line, which emphasises the ~16 rotation of this subunit in one dimer relative to the other about an axis perpendicular to the page. This has the effect of altering the proximity
a
-helix linkers are not shown. The principal axis of the front subunit in each of the two dimers is shown as a dotted
ꢀ
of the b-sandwich domain of the front subunit to the active site of the rear subunit (green circle) as highlighted by the dashed red-circles. We speculate that by embracing the
opposing subunit, the N-terminal arm and domain help to maintain a more open active site in CDP. (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
9
and glucose in the active site) will be the default reference structure
see Supplementary Fig. S10 for a structure-based sequence align-
ment). Like these enzymes, CDP contains a large -sandwich
domain that forms the majority of the dimer interface, connected
by a two -helix linker to an ( -barrel catalytic domain, and
ends with a small peripheral domain that adopts a two layered-jelly
roll fold (Fig. 7A and B, refer to Fig. 7B for domain colour-code).
Uniquely, CDP has a further ~120 amino acids at the N-terminus,
beginning with an extended arm that leads into a globular domain
by a sugar bound in the ꢁ1 subsite, and thus must be bound before
the glucan co-substrate in the phosphorolytic reaction, consistent
with it following a sequential Bi Bi mechanism. Beyond subsite þ1,
the acceptor site opens out and the glucan chain extends across a
wide U-shaped canyon formed at the dimer interface (Fig. 9A and
C), such that the residues occupying subsites þ2 and þ3 also
(
b
a
a/a)
6
interact with side-chains from the
b-sandwich domain of the
opposing subunit (Fig. 8). Although no electron density is present
for sugar residues beyond that occupying the þ3 subsite
(Supplementary Fig. S11), we cannot rule out the presence of the
two further residues expected for cellohexaose. Indeed, this may
indicate that the þ3 subsite delimits the extent of ordered binding
by the acceptor site. There are no large conformational changes
between the apo and ligand bound CDP structures (rms deviation of
0.692 Å for a dimer on dimer superposition), although a number of
side-chains become reoriented to engage with the substrates. In
particular, the carboxylate of the catalytic Asp624 is flipped about
comprised of a central, five-stranded, mixed
short -helices; the latter is unrelated to any structurally charac-
terised domain (Fig. 7). Both the domain and the N-terminal
arm interact with the -sandwich domain of the opposing subunit.
Indeed, a -strand ( 1) within the arm contributes to the one of the
sheets within the -sandwich (Fig. 7A, dashed red-circle, and C).
b-sheet, flanked by
a
a/b
b
b
b
b
Together, these additional interactions add substantially to the
2
dimer interfacial area: the total for CDP is ~4800 Å , as compared to
2
the value of ~3300 Å calculated for Cellovibrio gilvus CBP (PDB code
the C
a
-C
b
bond, to hydrogen bond with O3 of the ꢁ1 subsite sugar
3
1
QG0) using the PISA server [51]. Very recently, four structures of
/2-oligoglucan phosphorylase (SOGP) from Lachnoclostridium
b
-
and the oxygen of the scissile glycosidic linkage; additionally, the
adjacent residue within this “catalytic loop”, Cys625, also hydrogen
bonds to O2 of the þ1 subsite sugar (Fig. 8). Together, these in-
teractions cause the loop, which includes Trp622, to shift towards
phytofermentans were reported [52], which have slightly lower
DALI Z scores of ~32, with only 47% of the structure aligned to CDP
and, in contrast to all the aforementioned structures, they are all
monomeric. SOGP also differs from these enzymes in that it acts on
the bound substrate (Ca-Ca shift of 1.3 Å for Asp624). In general, the
majority of the interactions we observe with the ꢁ1 and þ1 subsite
sugars and the phosphate ion are structurally conserved in ligand
bound structures of CBP, but the correspondence is weaker for
SOGP (e.g. in PDB code 5H41), where the architecture necessarily
differs because the acceptor is bound in an orientation that is
orthogonal to that in the other enzymes [52]. Consistent with the
substrate preferences of CBP, its active site is significantly more
enclosed than that of CDP (Fig. 9B and D). This is largely due to three
structural features. Firstly, the catalytic loops differ in length, being
twelve residues longer in CDP (Fig. 9). Whilst the loops are struc-
turally similar up to and including the portion containing Asp624,
they adopt completely different conformations after Ile628. In the
case of one subunit of apo-CDP, the loop continues away from the
b
-1/2-glucan oligosaccharides rather than
saccharides. Like CDP, SOGP also has an extra, albeit much larger
~250 residues) N-terminal extension, again forming an extended
arm and a discrete domain, although the latter resembles the
b-1/4-glucan oligo-
(
b
-
sandwich domain common to all these phosphorylases, such that it
has two of these domains in tandem (Fig. 7). Remarkably, the
additional domain is placed relative to the remainder of the subunit
such that it aligns with the b-sandwich domain from the opposing
subunit of a superposed CDP/CBP-like dimer, thereby mimicking
the dimer interface of these latter enzymes. Moreover, the N-ter-
minal arm of SOGP interacts with the second
b-sandwich domain in
a similar way to the interaction seen between the N-terminal arm
of CDP and the
and C).
b-sandwich domain of the opposing subunit (Fig. 7A
active site forming a helix (
surface, before returning to the protein core; in the other subunit, a
short section following 17 is disordered. The latter is true for one
a17) that projects from the protein
a
2
.7. Ligand bound structure of CDP
subunit in ligand-bound CDP (Fig. 9A and C), whereas in the other,
substantially more of this loop is disordered, with the exception of
the portion bearing Trp622 and Asp624. By contrast in CBP, the
catalytic loop is fully ordered, with the C-terminal portion folded
over the active site pocket (Fig. 9B and D). Secondly, an “adjacent”
loop, which packs against the catalytic loop, also impinges on the
active site cleft, but has a lesser impact in CDP as it is five residues
shorter than the equivalent loop in CBP. A final significant differ-
ence relates to how the acceptor pocket is defined by the opposing
The active site of CDP lies at one end of the (
catalytic domain and the substrate binding site is largely delineated
by the loops that connect the outer ring of -helices to the inner
ring. All attempts to co-crystallise CDP with a variety of substrates
resulted in poor quality crystals. However, when 10 mM cellohex-
aose was soaked into a crystal which had been grown in the
presence of 10 mM phosphate buffer, a dataset was collected to
6
a/a) -barrel of the
a
3
.0 Å resolution revealing additional electron density in the active
subunit, in particular, by two
a-helices (a6 and a7) within a loop of
site cleft, which was interpreted as a cellotetraose molecule with an
adjacent phosphate ion. Given the relatively low resolution of this
structure we were unable to be certain of the conformation of the
the -sandwich domain. In the more open CDP pocket, two side-
b
chains from this “opposing loop” contribute to the acceptor bind-
ing site: Asp297 hydrogen bonds to O3 of the þ2 subsite sugar, and
Tyr300 forms a stacking interaction with it. There is also a further
hydrogen bond to O1 of the þ3 subsite sugar, via the side-chain of
Glu328 in a different loop of the opposing subunit (Fig. 8 and 9,
Supplementary Fig. S11). Compared to CDP, the juxtaposition of the
two subunits differs in CBP due to a global rotation about an axis
that is perpendicular to, and almost bisects, the two-fold axis
relating one half of the dimer to the other (Fig. 7C). In the Cellovibrio
sugar rings: we therefore chose to treat them all as
b
-1/4 linked
4
C
1
chairs, which gave a reasonable fit to the electron density with
good geometrical parameters. The glucan was oriented such that
the non-reducing terminal glycosidic bond was located between
the phosphate and Asp624, the expected general acid catalyst
(Fig. 8 and Supplementary Fig. S11). Thus, the glucan spanned
subsites ꢁ1 to þ3 of the active site pocket. The donor site is
completely buried and does not extend beyond the ꢁ1 donor
subsite where it is terminated by the side-chain of Trp622, the so-
called “hydrophobic platform” residue that is structurally
conserved in the close homologues. This is important to exclude
water around the region of the scissile bond. The phosphate is
located in an adjacent lobe of the active site cleft that is closed off
ꢀ
gilvus CBP (PDB code 3QG0) the rotation is ~16 (although it is
similar for others), and this has the effect of significantly narrowing
the canyon between the two subunits and, thereby driving the
opposing loop towards the active site (Fig. 9B and D). This fore-
shortens the acceptor binding pocket, such that in CBP, Gln165 in
the equivalent of a7 in the opposing loop, hydrogen bonds to O5 of
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

10
E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
Fig. 8. Details of ligand binding in CDP. A. Structure of the active site showing the protein backbone in cartoon representation with ligands and neighbouring side-chains as sticks,
where ligand carbons are in green, the backbone and side-chain carbons are in cream for the left-hand subunit, and in slate grey for the right-hand subunit. Direct hydrogen bonds
with protein side-chains are shown as dashed lines and the catalytic and opposing loops are highlighted. The labels for side-chains from the right-hand subunit are preceded by a
hash symbol and the label for the catalytic Asp is underlined. For clarity, some of the foreground detail has been omitted, mainly from the adjacent loop (see Fig. 9A and C), but this
does not remove any residues that interact directly with the ligands. The sugar binding subsites are indicated and the scissile glycosidic bond is marked by the red asterisk. The same
view is shown in stereo together with omit difference electron density for the bound ligands in Supplementary Fig. S11. B. Schematic representation of detail shown in panel A, this
time showing all hydrogen bonds including those involving protein backbone atoms. The grey arcs associated with aromatic side-chains indicate van der Waals interactions with the
cellotetraose. Hydrogens have been omitted for clarity. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
the þ1 subsite and there is no space for further sugar residues
beyond this subsite (not shown).
In trying to rationalise these conformational differences, we
were drawn to a possible structural role for the unusual N-terminal
domain that is unique to CDP. In the homodimer, it is closely
phosphate on the anomeric carbon of the scissile glycosidic bond,
and since the phosphate is already ionised at physiological pH,
there is no requirement for a general base (i.e. a second carboxylate)
to initiate the reaction. Concurrently, the general acid donates a
proton to the glycosidic oxygen to cleave the bond and release the
shortened cellodextrin molecule. Consistent with this mechanism,
we note that in the ligand bound structure of CDP, Asp624 is within
hydrogen bonding distance of the glycosidic bond oxygen, indi-
cating that it must be protonated. Moreover, the phosphate is
appropriately positioned for the nucleophilic attack.
With the benefit of the ligand bound structure of CDP, we can
attempt to rationalise substrate preferences. For instance, the 6OH
of ꢁ1 subsite sugar forms hydrogen bonds with the main-chain
carbonyl of Trp622 and the main-chain amide of Asp624, and C6
makes van der Waals contacts with the side chains of both Trp622
and Phe815. Since xylose lacks both C6 and O6, this could explain
the poor donor activity with xylose 1-phosphate. Whilst the poor
activity with mannose 1-phosphate could result from the switch
from an equatorial to an axial configuration for the hydroxyl at C2,
associated with the b-sandwich domain of the neighbouring sub-
unit; indeed, together with the N-terminal arm, it appears almost
to embrace it. This interaction could have the effect of drawing
together the lower portions of the subunits (Fig. 7C), whilst causing
the upper portions to move apart, thereby leading to a widening of
the canyon that lies adjacent to the active site and allowing larger
acceptors to bind.
It is expected that CDP will employ an inverting displacement
mechanism (Figs. S12A and B) with the phosphorolytic reaction
cleaving a glucose unit from the non-reducing end of a cellodextrin
polymer to yield
a-D-glucose 1-phosphate. In contrast to the
typical glycosyl hydrolase mechanism (Fig. S12C), only a single
carboxylate side-chain (Asp624) is required, which acts as the
general acid. The reaction proceeds via a nucleophilic attack by the
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
11
Fig. 9. Comparison of active sites between CDP and CBP. The active site of CDP is significantly more open than that of CBP due to differences in lengths and conformations of the
catalytic and opposing loops, and the relative dispositions of the subunits with the homodimer. The relatively closed dimer interface of CBP drives the opposing loop into the active
site of the neighbouring subunit to restrict the length of the acceptor binding site. Panels A and B show cartoon representations with the ligands in stick representation together
with the side-chains of the “platform”, Trp and the catalytic Asp. Panels C and D show equivalent views with the protein depicted as a molecular surface. Where ligands are obscured
by the protein surface, they are coloured white.
with the concomitant loss of hydrogen bonds to the side-chains of
Arg496 and Gln874 and to the phosphate, as well as a potential
clash of the hydroxyl with the catalytically competent configura-
tion of Asp624. CDP activity with glucosamine 1-phosphate leading
mainly to the observed single turnover reaction (see section 2.2 and
sugars, including galactose, mannose, xylose and glucosamine, and
competition assays have helped to define the substrate recognition
by CDP. While Xyl-1-P and Man-1-P have been shown to be poor
CDP donors, GlcN-1-P was shown for the first time to be a relatively
good donor for CDP; GlcN-containing oligomers were reasonable
CDP acceptors. Moreover, xylotriose was confirmed to be a
2
.3) may be explained by the stabilisation of the sugar 1-phosphate
in the active site due to the ionic interactions between the amino
group in GlcN and the ꢁ1 subsite and the phosphate group. In
addition, the replacement of oxygen with nitrogen at the 2 position
of the þ1 subsite acceptor sugar (GlcN vs Glc) should not alter the
binding significantly maintaining all hydrogen bonds with Glu810,
reasonable acceptor and to outcompete the APTS-labelled
glucan, indicating its binding to the active site.
b-1/4-
The crystal structure of CDP from Ruminiclostridium thermo-
cellum reported herein is the first solved for cellodextrin phos-
phorylase enzymes. This reveals a novel N-terminal domain that
may be involved in adjusting the relative orientations of the two
subunits within the homodimer. Together with substantial rear-
rangements of loops that delineate the active site, this leads to a
significantly more open acceptor binding pocket relative to CBP,
consistent with the capacity of CDP to accept longer oligomers as
acceptor substrates. Together with the substrate binding assays, the
crystal structure will help to inform the engineering of CDP for
Cys625 and Tyr804, as shown by the good acceptor activity of (b-
1
/4-GlcN(Glc)
. Conclusions
Extensive work has previously been carried out on cellobiose
2
) toward extension with Glc-1-P (see section 2.5.2).
3
phosphorylases to assess the substrate promiscuity and to syn-
thesise disaccharides. Monitoring polymerization reactions is less
straightforward, but utilising the combination of resolution and
dynamic range provided by capillary electrophoresis and MALDI
the long polymers produced by this enzyme could be resolved.
Different polymers can be produced using mixed substrates and the
relative impact of binding on the turnover of non-natural substrates
can give clues to the nature of enzyme-substrate interactions.
wider use in b-1/4-glucan oligosaccharide synthesis.
4. Experimental
4.1. General
All synthetic reagents were obtained commercially and used as
CDP can be used to synthesise long
b-1/4-glucans which
received, unless otherwise stated. Milli-Q H
2
O was used for the
rapidly precipitate out of solution. There is a slow transfer of other
preparation of aqueous buffers. Commercial 2,3,4,6-tetra-O-acetyl-
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

12
E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
a
-D-glucopyranosyl fluoride (carbosynth) was deprotected ac-
column (5 ml, GE Healthcare), washed with wash buffer (50 mM
Tris-HCl, pH 8.0, 0.5 M NaCl, 0.03 M imidazole) and eluted in one
step with 50 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 0.5 M imidazole
(BioUltra). Further purification was performed by gel filtration us-
ing a Superdex S75 26/60 column (GE Healthcare) with GF buffer
(50 mM HEPES, pH 7.5, 100 mM NaCl, 3.2 ml/min). Fractions con-
taining CDP were collected (Supplementary Fig. S1), pooled and
concentrated to 40 mg/ml using an Amicon Ultra-15 30 kDa MW
cut off concentrator. The protein yield was approximately 10 mg/L
of culture. The His-tag was not cleaved and the protein was stored
cording to standard procedures. All reagents and solvents used for
analytical applications were of analytical quality. TLC was per-
formed on precoated slides of Silica Gel 60 F254 (Merck). Reaction
products were characterised by Matrix assisted Laser Desorption
Ionization-Time of Flight (MALDI-ToF) and electrospray ionization
1
mass spectrometry (ESI-MS), H, 2D-COSY and 2D-HSQC NMR
spectroscopy. Nuclear magnetic resonance spectra were recorded
1
at 298 K on a Bruker Avance III 400 spectrometer, H spectra at
4
00 MHz and 13C spectra at 101 MHz. Chemical shifts (
d
) are re-
ported in parts per million (ppm) with respect to residual HOD
signal in D O ( 4.79). Coupling constants (J) are reported in Hz.
ꢀ
in aliquots at ꢁ80 until required. Prior to crystallisation, dynamic
2
d
H
light scattering was carried out to monitor the solution properties
NMR signal assignments were made with the aid of COSY and HSQC
experiments. MALDI was performed on a Bruker Autoflex Speed
using 2,5-dihydroxybenzoic acid (DHB, 10 mg/ml in MeOH þ 0.1%
TFA) as matrix in positive mode. Enzymatic reactions were desalted
prior MALDI sample preparation by addition of mixed bed resin
of the purified sample with a DynaPro-Titan molecular-sizing in-
strument at 20 C (Wyatt Technology). For de novo structure
determination, CDP was labelled with selenomethionine (SeMet)
by metabolic inhibition [53]. Cells containing plasmid pET15b-CDP
were grown in SeMet minimal media (SeMet MM: M9 salts, 0.2%
ꢀ
(
Sigma) and incubation at rt for 2 min. Samples were typically
mixed 1:1 (v/v) with the matrix and spotted on a target plate
Bruker MTP 384 Polished Steel TF Target). Accurate electrospray
glucose, 2 mM MgSO
amino acid stock) containing 100
4
, 0.1 mM CaCl
2
, 10 mg/l thiamine, 20 ml of
m
g/ml carbenicillin. The amino
(
acid stock contains Arg, Asp, Glu, Gln, His, Ile, Leu, Lys, Phe, Ser, Thr,
Try, Tyr and Val (2 mg/ml). An overnight LB culture of cells (5 ml)
was collected by centrifugation (4000 g) and washed with SeMet
MM (3 ꢂ 5 ml). These cells were then grown in SeMet MM (1 L) at
ionization mass spectra were obtained on a Synapt G2-Si mass
spectrometer (Waters, Manchester, UK). Samples were diluted into
5
0% methanol/0.1% formic acid and infused into the mass spec-
trometer at 5e10 l/min using a Harvard Apparatus syringe pump.
The mass spectrometer was controlled by Masslynx 4.1 software
Waters). It was operated in resolution and positive ion mode and
calibrated using sodium formate. The sample was analysed for
ꢀ
m
37 C until OD600 of 0.6 was reached and then Lys, Phe and Thr
(100 mg), Ile, Leu and Val (50 mg) and SeMet (60 mg) were added
and the temperature adjusted to 30 C. After a further 45 min, cells
were induced with IPTG (1 mM) and after 16 h the cells were
collected by centrifugation. The cells were harvested, stored and
purified as described above to give a final yield of 12 mg of purified
protein (Supplementary Fig. S2).
ꢀ
(
2
min with 1 s MS scan time over the range of 50e1200 m/z (or as
ꢀ
appropriate) with 3.5 kV capillary voltage, 40 V cone voltage,100 C
cone temperature. Leu-enkephalin peptide (1 ng/ml, Waters) was
infused at 10 ml/min as a lock mass (m/z 556.2766) and measured
every 10 s. Spectra were generated in Masslynx 4.1 by combining a
number of scans, and peaks were centred using automatic peak
detection with lock mass correction. Ion-exchange chromatography
was performed using Bioscale™ Mini Macro-Prep High S cartridge
and a step-gradient from 0 to 1 M ammonium bicarbonate buffer
4.3. APTS labeling of sugars
Sugars were labelled for electrophoresis with 8-aminopyrene-
1,3,6-trisulfonic acid (APTS) according to the PACE method [34].
(
pH 9.4). Compounds were visualised by spraying TLC with orcinol
solution (20 mg/ml orcinol monohydrate in EtOH/H SO :H
5:10:5, v/v), followed by heating. Product containing fractions
APTS (0.5 mg, 0.2 M in 30% aqueous acetic acid, 5 ml) was mixed
2
4
2
O
with NaBH
3
CN (0.5 mg, 0.8 M in THF, 5
m
l). Reducing carbohydrate
ꢀ
7
(1 mg) was dissolved in the mixture and incubated at 37 C for 18 h
ꢀ
were combined and reduced to dryness. The residue was co-
evaporated repeatedly with methanol to remove residual ammo-
nium bicarbonate. Gel Filtration Chromatography was performed
on a Perkin Elmer series 200 equipped with a Toyopearl TSK-
HW40S column (90 cm ꢂ 1.6 cm), a refractive index detector and
a fraction collector. Colourimetric assays were performed in NUNC
or 70 C for 2 h. The sample was then loaded on to a 30% 38:2
mono:bis acrylamide (Merck) Tris-borate (100 mM, pH 8.2) gel and
separated by electrophoresis at 400 V in Tris-borate buffer
(100 mM, pH 8.2), water cooled to room temperature. The carbo-
hydrate band (upper) was excised and the labelled sugar was
extracted into purified water by grinding using a ceramic bead in a
9
6 plates on a BMG labtech FLUOStar Omega microplate reader
2
bead mill (Fast Prep FP 120, Thermo) and washing with Milli-Q H O
equipped with suitable absorbance filters.
(3 ꢂ 30 ml). The borate was removed, either by desalting using a
PD-10 column (GE Healthcare) or by evaporating three times from
methanol, and the product was quantified using APTS absorption at
4.2. Expression and purification of CDP
455 nm (17,160 1/M/cm) [54].
The codon optimised gene for Ruminiclostridium thermocellum
cellodextrin phosphorylase was synthesised and sub-cloned into
the BamHI site of pET15b. An overnight culture of BL21 (DE3) cells
containing this plasmid (4 ꢂ 1 ml) was used to inoculate 4 ꢂ 1 L
4.4. Capillary electrophoresis with laser induced fluorescence (CE-
LIF)
ꢀ
cultures of LB which were grown at 37 C until an OD600 of ~0.6. The
ꢀ
cultures were then cooled to 30 C and induced with 1 mM IPTG.
After enzymatic reactions had been carried out using APTS-
labelled carbohydrates, samples were placed in boiling water for
5 min and centrifuged for 5 min at 16,000 g, to inactivate and
After 4 h, the cells were harvested by centrifugation and frozen
ꢀ
at ꢁ80 C until required. When required, cell pellets were thawed
and re-suspended in 50 ml of lysis buffer (50 mM HEPES, pH 7.5,
remove proteins. The samples were made up to at least 50
ml and
1
00 mM NaCl, 1ꢂ Complete™ EDTA-free Protease Inhibitor Cocktail
loaded on to an N-CHO coated capillary (50.2 cm, 50 m) in a PA800
m
Tablet (Roche), 0.02 mg/ml DNaseI). Cells were lysed using a cell
disruptor (one shot mode, 25 kpsi, Constant Systems) and the cell
debris removed by centrifugation at 30,000 g (30 min, 4 C). Protein
was purified at 4 C using an AKTAxpress FPLC system (GE
Healthcare). The supernatant was passed through a HiTrap Ni-NTA
ProteomeLab (Beckman Coulter) by injection at 0.5 psi for 20 s.
They were then separated in running buffer (25 mM LiOAc, 0.4%
ꢀ
ꢀ
polyethylene oxide, pH 4.75, 20 C) at 30 kV for 7e45 min and
ꢀ
€
detected using LIF (excitation at 488 nm and detection at 520 nm)
[35,55].
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
13
app
4.5. Synthesis of (
b
-1/4-GlcN(Glc)
2
)
The K
M
was calculated for each using GraFit (Erithacus Software
Ltd).
CDP (0.2 mg/ml) was added to a solution of cellobiose (8 mM)
and GlcN-1-P (42 mM) in HEPES buffer (50 mM, pH 7.5) (all con-
4.7. Phosphorylase activity assays
centrations are final concentrations). The reaction was incubated at
ꢀ
4
0 C for 12 h. After this time, the reaction was quenched in boiling
All kinetic analyses were performed in the synthetic direction,
measuring release of Pi from Glc-1-P. The concentration of released
Pi was measured colourimetrically using a method modified from
water for 5 min and centrifuged for 1 min at 16,000 g, to inactivate
and remove protein. Then, the reaction was diluted with H O and
2
loaded onto a cation exchange cartridge. The product was eluted
with 0.05 M ammonium bicarbonate buffer (pH 9.4) and the
product containing fractions co-evaporated with methanol to
De Groeve et al. [56] Colour solution (75
sodium ascorbate) was added to the stopped enzyme reaction
ml, 0.1 M HCl, 13.6 mM
(25
m
l), containing sodium molybdate, Na
2
MoO
4
$ 2H
2
O (200 mM),
1
ꢀ
remove the volatile buffer, then freeze-dried. H NMR (400 MHz;
in a microtitre plate. After incubating for 5 min at 21 C, stop so-
0
D
2
O)
d
: 5.24 (d, J
1
0
,2
0
¼ 3.8 Hz,
a
ꢁH-1 ), 4.68 (d, J
1
0
,2
0
¼ 8.0 Hz,
b
ꢁH-
lution (75
ml, 68 mM sodium citrate, 2% acetic acid) was added and
0
00
1
1
5
a
2
), 4.55 (d, J
1
00,200 ¼ 8.1 Hz, 1H, H-1 ), 4.49 (d, J
1
000,2000 ¼ 8.1 Hz, 1H, H-
A620 was measured.
0
0
00
0
000
00
000
0
00
000
0
), 4.0e3.60 (m, 15H, H-3 , H-3 , H-3 , H-4 , H-4 , H-4 , H-5 , H-
0
000
0
00
, H-5 , H-6 , H-6 , H-6 ), 3.59 (dd, J
1
0
,2
0
¼ 3.8 Hz, J
2
0
,3
0
¼ 9.87 Hz,
4.8. Protein crystallisation
0
0
ꢁH-2 ), 3.30 (dd, J
1
0
,2
0
¼ 8.0 Hz, J
2
0
,3
0
¼ 8.97 Hz,
b
ꢁH-2 ), 3.37 (H-
0
0
000 13
000
), 2.72 (H-2 ). C NMR (101 MHz; D
2
O)
d
: 102.69 (C-1 ), 102.15
A range of commercially available crystallisation screens were
set up (Qiagen and Molecular Dimensions) in 96 well MRC plates
(Molecular Dimensions). The well solution was dispensed using a
Tecan Freedom Evo robot; the drops (0.3 ml protein plus 0.3 ml well
solution) were dispensed using an OryxNano robot (Douglas in-
struments). The protein solution was diluted to 10 mg/ml with GF
0
0
0
0
0
00
(
C-1 ), 95.70 (
b
ꢁC-1 ), 91.75 (
a
ꢁC-1 ), 73.93 (
b
ꢁC-2 ), 72.92 (C-2 ),
0
00
5
5
6.37 (C-2 ). TLC: R
f
¼ 0.29 (iPA/NH
4 2
OH/H O 6:3:1); m/z (ESI)
þ
þ
26.1741 [MþNa] , C18
H
33NNaO15 requires 526.1742.
4.6. Acceptor specificity of CDP
buffer and filtered through a 0.1
m
m filter before use. Plates were
ꢀ
The ability for CDP (50
10 mM) was assayed at 21 C for 20 min in HEPES (50 mM, pH
m
g/ml) to transfer Glc from Glc-1-P
placed in a crystal hotel (CrystalPro, TriTek Corp.) at 20 C and
monitored over the course of 4 weeks. Crystallisation hits were
optimised in 24-well hanging drop format using 1 ml well solution
ꢀ
(
7.5). Various acceptors were assayed over a range of concentrations
to obtain kinetic parameters (Supplementary Fig. S6). The enzyme
activity was measured by phosphate release assay (see section 4.7).
and drops comprising of 1
noticed that, when mixing the CDP protein with the crystallisation
ml protein and 1 ml well solution. It was
Table 2
X-ray data collection and processing.
Data set
Native
SeMet
Cellohexaose þ Pi
Data Collection
Beamline
I02
I02
I02
Wavelength (Å)
Detector
0.9795
Pilatus 6M
50.60e2.30 (2.36e2.30)
0.9795
Pilatus 6M
83.62e3.50 (3.59e3.50)
0.9795
Pilatus 6M
27.62e3.00 (3.08e3.00)
P2
1
Resolution range (Å)a
Space Group
P2
1
P2
1
a, b, c (Å)
84.6, 151.8, 92.0
90.0, 114.6, 90.0
522,852 (22,947)
93,120 (6720)
5.6 (3.4)
14.3 (2.0)
99.6 (97.3)
0.084 (0.759)
0.102 (1.035)
0.997 (0.594)
57.9
84.7, 151.8, 92.0
90.0, 114.6, 90.0
92,835 (6675)
26,456 (1937)
3.5 (3.4)
18.9 (7.1)
99.0 (99.1)
0.043 (0.142)
0.061 (0.197)
0.995 (0.973)
72.9
84.8, 153.0, 92.2
90.0, 114.4, 90.0
141,679 (10,024)
41,938 (2981)
3.4 (3.4)
12.8 (1.2)
98.0 (94.8)
0.064 (1.086)
0.077 (1.291)
0.999 (0.425)
87.8
ꢀ
a
,
b
,
g
( )
a
a
Total observations
Unique reflections
Multiplicity
a
a
Mean I/
s
(I)
a
Completeness (%)
a,b
R
R
CC
merge
a,c
meas
a,d
½
2
Wilson B value (Å )
Refinement
e
Reflections: working/free
88,414/4668
0.191 (0.317)
0.223 (0.334)
95.8/3.9/0.3
0.008
1.18
984/984
0/0/233/6
e
e
e
e
e
e
e
e
e
e
39,167/2133
0.210 (0.350)
0.262 (0.416)
94.1/5.5/0.4
0.008
1.16
884/974
8/2/0/0
f
R
R
work
f
free
g
Ramachandran plot: favoured/allowed/disallowed (%)
R.m.s. bond deviations (Å)
R.m.s. angle deviations ( )
No. of protein residues: ChainA:ChainB
No. of sugars/phosphates/waters/other
ꢀ
2
Mean B-factors: protein/sugars/phosphates/waters/other/overall (Å )
38.8/ꢁ/ꢁ/59.6/52.1/39.0
114/136/111/ꢁ/ꢁ/114
PDB accession code
5NZ7
5NZ8
a
Values for the outer resolution shell are given in parentheses.
P
P
P
P
b
c
R
R
merge
¼
hkl
i
jI
i
(hkl) ꢁ 〈I(hkl)〉j/ hkl
i i
I (hkl).
(hkl) ꢁ 〈I(hkl)〉j/ hkl
P
P
P
P
1/2
meas
¼
hkl [N/(N ꢁ 1)]
ꢂ
i
jI
i
i i i
I (hkl), where I (hkl) is the ith observation of reflection hkl, 〈I(hkl)〉 is the weighted average intensity for all
observations i of reflection hkl and N is the number of observations of reflection hkl.
d
CC
½
is the correlation coefficient between symmetry-related intensities taken from random halves of the dataset.
e
f
The data set was split into “working” and “free” sets consisting of 95 and 5% of the data, respectively. The free set was not used for refinement.
P
P
The R-factors Rwork and Rfree are calculated as follows: R ¼ (j Fobs - Fcalc j)/ j Fobs j, where Fobs and Fcalc are the observed and calculated structure factor amplitudes,
respectively.
g
As calculated using MOLPROBITY [68].
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

14
E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
solution (20% PEG 3350 (w/v), 300 mM KCl, 100 mM HEPES pH 7.5
buffer), a proportion of the protein rapidly precipitated. Subse-
quently, the protein was pre-precipitated by mixing an equal vol-
ume of the crystallisation solution with CDP (10 mg/ml) and
incubating on ice for 15 min. After centrifugation at 16,000 g for
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.carres.2017.07.005.
1
min, crystallisation drops were set up by mixing 1
m
l of the sol-
References
uble protein with 1 l of the crystallisation solution. Protein crystals
m
appeared in two days and were transferred to a cryoprotectant
solution comprising the crystallisation solution supplemented with
[1] R.M. Brown, J. Polym. Sci. Part A Polym. Chem. 42 (2004) 487e495.
[
2] D. Klemm, B. Heublein, H.-P. Fink, A. Bohn, Angew. Chem. Int. Ed. 44 (2005)
358e3393.
3
2
0% (v/v) glycerol using LithoLoops (Molecular Dimensions), before
€
[3] U. Romling, Res. Microbiol. 153 (2002) 205e212.
being flash-cooled in liquid nitrogen and stored in Unipuck cas-
settes (MiTeGen) prior to transport to Diamond Light Source (Ox-
ford, UK).
[4] A. Mihranyan, J. Appl. Polym. Sci. 119 (2011) 2449e2460.
[
[
5] C.C. Lin, J. Aronson, Arch. Mikrobiol. 72 (1970) 111e114.
6] R.J. Moon, A. Martini, J. Nairn, J. Simonsen, J. Youngblood, Chem. Soc. Rev. 40
(2011) 3941e3994.
[
[
7] R.H. Atalla, D.L. Vanderhart, Science 223 (1984) 283e285.
8] T. Hayashi, R. Kaida, Mol. Plant 4 (2011) 17e24.
4.9. X-ray data collection and structure solution
[9] C. Somerville, S. Bauer, G. Brininstool, M. Facette, T. Hamann, J. Milne,
E. Osborne, A. Paredez, S. Persson, T. Raab, S. Vorwerk, H. Youngs, Science 306
(
2004) 2206e2211.
Crystals were transferred robotically to the goniostat on beam-
[
10] C. Chang, L. Zhang, Carbohydr. Polym. 84 (2011) 40e53.
line I02 at Diamond Light Source (Oxfordshire, UK) and maintained
[11] X. Shen, J.L. Shamshina, P. Berton, G. Gurau, R.D. Rogers, Green Chem. 18
ꢀ
(2016) 53e75.
at ꢁ173 C with a Cryojet cryocooler (Oxford Instruments). X-ray
[12] S. Ummartyotin, H. Manuspiya, Renew. Sustain. Energy Rev. 41 (2015)
diffraction data were recorded using a Pilatus 6M hybrid photon
counting detector (Dectris), then integrated using XDS [57], and
scaled and merged using AIMLESS [58] via the XIA2 expert system
402e412.
[13] M.E. Himmel, S.-Y. Ding, D.K. Johnson, W.S. Adney, M.R. Nimlos, J.W. Brady,
T.D. Foust, Science 315 (2007) 804e807.
[14] S.M.A.S. Keshk, J. Bioprocess. Biotech. 4 (2014), http://dx.doi.org/10.4172/
[
59]. Crystals belonged to space group P2
1
, with approximate cell
2155-9821.1000150.
ꢀ
parameters of a ¼ 85, b ¼ 152, c ¼ 91 Å and
b
¼ 115 and the solvent
[15] D.P. Delmer, Annu. Rev. Plant Physiol. Plant Mol. Biol. 50 (1999) 245e276.
[16] S. Kobayashi, K. Kashiwa, J. Shimada, T. Kawasaki, S.-i. Shoda, Makromol.
Chem. Macromol. Symp. 54e55 (1992) 509e518.
content was estimated to be 49%, based on two copies of the protein
chain (111.6 kDa) in the asymmetric unit (ASU). Experimental
phases were determined using PHENIX [60] by combining a native
dataset with a SeMet dataset (collected at the Se K X-ray absorption
edge; wavelength 0.9795 Å), but PHENIX was unable to autobuild
the structure from the resultant electron density map. As this point,
two copies of a homology model of CDP [based on the structure of
Cellvibrio gilvus CBP (PDB code 2CQT) [27] and generated by the
Phyre2 server (http://www.sbg.bio.ic.ac.uk/phyre2) [61] were
manually docked onto a skeletonised version of this map using
COOT. After refinement of this preliminary model using REFMAC5
[
[
17] E.C. O'Neill, R.A. Field, Carbohydr. Res. 403 (2015) 23e37.
18] H. Nakai, M. Kitaoka, B. Svensson, K.i. Ohtsubo, Curr. Opin. Chem. Biol. 17
(2013) 301e309.
[19] E.C. O'Neill, A.M. Rashid, C.E.M. Stevenson, A.-C. Hetru, A.P. Gunning,
M. Rejzek, S.A. Nepogodiev, S. Bornemann, D.M. Lawson, R.A. Field, Chem. Sci.
5
(2014) 341e350.
[20] K. Sheth, Jk Alexande, Biochim. Biophys. Acta 148 (1967) 808e810.
[
[
[
21] K. Sheth, J.K. Alexander, J. Biol. Chem. 244 (1969) 457e464.
22] N. Yutin, M.Y. Galperin, Environ. Microbiol. 15 (2013) 2631e2641.
23] D.M. Petrovi ꢀc , I. Kok, A.J.J. Woortman, J. Ciri ꢀc , K. Loos, Anal. Chem. 87 (2015)
9639e9646.
ꢀ
[24] M. Hiraishi, K. Igarashi, S. Kimura, M. Wada, M. Kitaoka, M. Samejima, Car-
bohydr. Res. 344 (2009) 2468e2473.
[62], model and experimental phases were combined using SIG-
[25] V. Lombard, H. Golaconda Ramulu, E. Drula, P.M. Coutinho, B. Henrissat,
Nucleic Acids Res. 42 (2014) 490e495.
MAA [63], then density modified (incorporating two-fold aver-
aging) in PARROT [64], to yield a much improved map at 2.3 Å
resolution. This enabled a new model of CDP to be built from
scratch using BUCCANEER [65], which was completed by several
iterations of manual rebuilding in COOT [66] and restrained
refinement in REFMAC5 using isotropic thermal parameters and
TLS group definitions obtained from the TLSMD server (http://
skuld.bmsc.washington.edu/~tlsmd/) [67]. Model geometries were
validated with MOLPROBITY [68] before submission to the Protein
Data Bank. This was used as the starting point for refinement of the
complex of CDP with cellohexaose and phosphate against a dataset
collected to 3.0 Å resolution. All data collection statistics and model
parameters are reported in Table 2.
[
[
26] J.K. Alexander, J. Biol. Chem. 243 (1968) 2899e2904.
27] M. Hidaka, M. Kitaoka, K. Hayashi, T. Wakagi, H. Shoun, S. Fushinobu, Biochem.
J. 398 (2006) 37e43.
[28] M. Kitaoka, T. Sasaki, H. Taniguchi, J. Biochem. 112 (1992) 40e44.
[29] C.M. Bianchetti, N.L. Elsen, B.G. Fox, G.N. Phillips Jr., Acta Crystallogr. Sect. F.
Struct. Biol. Cryst. Commun. 67 (2011) 1345e1349.
[30] K. Shintate, M. Kitaoka, Y.-K. Kim, K. Hayashi, Carbohydr. Res. 338 (2003)
1981e1990.
[31] H.G. Tran, T. Desmet, K. Saerens, H. Waegeman, S. Vandekerckhove,
M. D’hooghe, I. Van Bogaert, W. Soetaert, Bioresour. Technol. 115 (2012)
84e87.
[
[
[
32] T. Kawaguchi, Y. Ikeuchi, N. Tsutsumi, A. Kan, J.-I. Sumitani, M. Arai, J. Ferment,
Bioeng 85 (1998) 144e149.
33] M. Krishnareddy, Y.-K. Kim, M. Kitaoka, Y. Mori, K. Hayashi, J. Appl. Glycosci.
49 (2002) 1e8.
34] F. Goubet, P. Jackson, M.J. Deery, P. Dupree, Anal. Biochem. 300 (2002) 53e68.
All structural figures were prepared using CCP4MG [69], and
interfacial areas were determined using the PISA server (http://
www.ebi.ac.uk/msd-srv/prot_int/cgi-bin/piserver) [51]. The DYN-
DOM server (http://fizz.cmp.uea.ac.uk/dyndom/runDyndom.jsp)
[35] S. Goetz, M. Rejzek, S.A. Nepogodiev, R.A. Field, Carbohydr. Res. 433 (2016)
7e105.
9
[
36] H. Nakai, M. Abou Hachem, B.O. Petersen, Y. Westphal, K. Mannerstedt,
M.J. Baumann, A. Dilokpimol, H.A. Schols, J.O. Duus, B. Svensson, Biochimie 92
(2010) 1818e1826.
[
[
[
37] K. Yamashita, K. Yamamoto, J.-i. Kadokawa, Biomacromolecules 16 (2015)
[70] was used to compare subunit rotations within CDP and CBP
3
989e3994.
38] M. Nawaji, H. Izawa, Y. Kaneko, J.-i. Kadokawa, Carbohydr. Res. 343 (2008)
692e2696.
dimers.
2
39] Y. Takemoto, H. Izawa, Y. Umegatani, K. Yamamoto, A. Kubo, M. Yanase,
T. Takaha, J.-i. Kadokawa, Carbohydr. Res. 366 (2013) 38e44.
Acknowledgement
[
[
40] J.-i. Kadokawa, Polymers 8 (2016) 138e158.
41] Y. Takata, R. Shimohigoshi, K. Yamamoto, J.-I. Kadokawa, Macromol. Biosci. 14
These studies were supported by the UK BBSRC Institute Stra-
tegic Program on Understanding and Exploiting Metabolism (MET)
(2014) 1437e1443.
[42] J.-i. Kadokawa, R. Shimohigoshi, K. Yamashita, K. Yamamoto, Org. Biomol.
[
BB/J004561/1] and the John Innes Foundation; the BBSRC, EPSRC
Chem. 13 (2015) 4336e4343.
43] R. Mukerjea, J.F. Robyt, Carbohydr. Res. 340 (2005) 245e255.
44] Y. Nishida, T. Wiemann, V. Sinnwell, J. Thiem, J. Am. Chem. Soc. 115 (1993)
[
[
and InnovateUK: IBCatalyst [BB/M02903411]. We acknowledge the
Diamond Light Source for access to beamline I02 under proposal
MX7641.
2536e2537.
[45] L. Holm, C. Sander, Trends Biochem. Sci. 20 (1995) 478e480.
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005

E.C. O'Neill et al. / Carbohydrate Research xxx (2017) 1e15
15
[
[
[
[
[
46] A. Van Hoorebeke, J. Stout, W. Soetaert, J. Van Beeumen, T. Desmet,
[59] G. Winter, J. Appl. Crystallogr. 43 (2010) 186e190.
S.N. Savvides, PDBe: 3RRS, 2012.
47] A. Van Hoorebeke, J. Stout, W. Soetaert, J. Van Beeumen, T. Desmet,
S.N. Savvides, PDBe: 3RSY, 2012.
48] A. Van Hoorebeke, J. Stout, W. Soetaert, J. Van Beeumen, T. Desmet,
S.N. Savvides, PDBe: 3S4A, 2012.
49] M. Hidaka, Y. Honda, M. Kitaoka, S. Nirasawa, K. Hayashi, T. Wakagi, H. Shoun,
S. Fushinobu, Structure 12 (2004) 937e947.
[60] P.D. Adams, P.V. Afonine, G. Bunkoczi, V.B. Chen, I.W. Davis, N. Echols,
J.J. Headd, L.W. Hung, G.J. Kapral, R.W. Grosse-Kunstleve, A.J. McCoy,
N.W. Moriarty, R. Oeffner, R.J. Read, D.C. Richardson, J.S. Richardson,
T.C. Terwilliger, P.H. Zwart, Acta Crystallogr. Sect. D. Biol. Crystallogr. 66
(2010) 213e221.
[61] L.A. Kelley, S. Mezulis, C.M. Yates, M.N. Wass, M.J.E. Sternberg, Nat. Protoc. 10
(2015) 845e858.
[62] G.N. Murshudov, A.A. Vagin, E.J. Dodson, Acta Crystallogr. Sect. D. Biol. Crys-
tallogr. 53 (1997) 240e255.
50] Y.-W. Nam, T. Nihira, T. Arakawa, Y. Saito, M. Kitaoka, H. Nakai, S. Fushinobu,
J. Biol. Chem. 290 (2015) 18281e18292.
[
[
51] E. Krissinel, Nucleic Acids Res. 43 (2015) 314e319.
52] M. Nakajima, N. Tanaka, N. Furukawa, T. Nihira, Y. Kodutsumi, Y. Takahashi,
N. Sugimoto, A. Miyanaga, S. Fushinobu, H. Taguchi, H. Nakai, Sci. Rep. 7
[63] R. Read, Acta Crystallogr. Sect. A Found. Adv. 42 (1986) 140e149.
[64] K. Cowtan, Acta Crystallogr. Sect. D. Biol. Crystallogr. 66 (2010) 470e478.
[65] K. Cowtan, Acta Crystallogr. Sect. D. Biol. Crystallogr. 62 (2006) 1002e1011.
[66] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D. Biol. Crystallogr. 60 (2004)
2126e2132.
[67] J. Painter, E.A. Merritt, J. Appl. Crystallogr. 39 (2006) 109e111.
[68] I.W. Davis, A. Leaver-Fay, V.B. Chen, J.N. Block, G.J. Kapral, X. Wang,
L.W. Murray, W.B. Arendall, J. Snoeyink, J.S. Richardson, D.C. Richardson,
Nucleic Acids Res. 35 (2007) 375e383.
(
2017) 42671e42682.
53] S. Doubli eꢀ , in: W. Charles, Carter Jr. (Eds.), Methods Enzymol, Academic Press,
997, pp. 523e530.
[
1
[54] R.A. Evangelista, M.-S. Liu, F.-T.A. Chen, Anal. Chem. 67 (1995) 2239e2245.
[55] E. Prifti, S. Goetz, S.A. Nepogodiev, R.A. Field, Carbohydr. Res. 346 (2011)
1617e1621.
[
56] M.R.M. De Groeve, G.H. Tran, A. Van Hoorebeke, J. Stout, T. Desmet,
[69] S. McNicholas, E. Potterton, K.S. Wilson, M.E.M. Noble, Acta Crystallogr. Sect.
D. Biol. Crystallogr. 67 (2011) 386e394.
S.N. Savvides, W. Soetaert, Anal. Biochem. 401 (2010) 162e167.
[
[
57] W. Kabsch, Acta Crystallogr. Sect. D. Biol. Crystallogr. 66 (2010) 125e132.
58] P.R. Evans, G.N. Murshudov, Acta Crystallogr. Sect. D. 69 (2013) 1204e1214.
[70] C. Girdlestone, S. Hayward, J. Comput. Biol. 23 (2015) 21e26.
Please cite this article in press as: E.C. O'Neill, et al., Cellodextrin phosphorylase from Ruminiclostridium thermocellum: X-ray crystal structure
and substrate specificity analysis, Carbohydrate Research (2017), http://dx.doi.org/10.1016/j.carres.2017.07.005