Insight into substrate recognition and catalysis by the human neuraminidase 3 (NEU3) through molecular modeling and site-directed mutagenesis

Article abstract of DOI:10.1093/glycob/cwq077

The mammalian neuraminidase (NEU) enzymes are found in diverse cellular compartments. Members of the family, such as NEU2 and NEU1, are cytosolic or lysosomal, while NEU3 and NEU4 are membrane-associated. NEU enzymes that act on substrates in the plasma membrane could modulate cellular signaling, cell surface glycoforms and the composition of plasma membrane glycolipids. Therefore, their substrates and mechanism of action are of interest for discerning their physiological roles. We have studied the structure of the human NEU3 using molecular modeling to predict residues involved in the recognition and hydrolysis of glycolipid substrates. To test the model, we have used site-directed mutagenesis of the recombinant protein. Enzymatic studies of the relative activity of these mutants, as well as their pH profiles and inhibition by 2-deoxy-2,3-dehydro-N-acetylneuraminic acid, are reported. Using nuclear magnetic resonance spectroscopy, we confirmed that the enzyme is a retaining exo-sialidase, and we propose that the key catalytic residues of the enzyme consist of the general acid-base D50 and the nucleophilic Y370-E225 pair. Mutations of residues expected to interact directly with the sialic acid N5-acetyl (A160, M87, I105) and C7-C9 glycerol side-chain (E113, Y179, Y181) reduced enzymatic activity. We identified several active mutants of the enzyme which contain modifications at the periphery of the active site. Truncations at the N-or C-terminus of more than 10 residues abolished enzyme activity. We propose a catalytic mechanism consistent with the data and identify residues that contribute to glycolipid recognition.

Full text of DOI:10.1093/glycob/cwq077

Glycobiology vol. 20 no. 9 pp. 1127–1138, 2010
doi:10.1093/glycob/cwq077
Advance Access publication on May 28, 2010
Insight into substrate recognition and catalysis by the human
neuraminidase 3 (NEU3) through molecular modeling and
site-directed mutagenesis
2
2
2
Amgad Albohy , Matthew D Li , Ruixiang Blake Zheng ,
Introduction
2
1,2
Chunxia Zou , and Christopher W Cairo
Several isoforms of mammalian exo-neuraminidase (NEU) en-
zymes are known. All reported members are classified as
family 33 glycoside hydrolases (GH33) within the CAZy data-
base (Cantarel et al. 2009). Of those isoforms, NEU3 is known
as an ectoenzyme, a peripheral membrane-associated enzyme
with activity at the outer leaflet (Monti et al. 2000; Kopitz et
al. 2001; Zanchetti et al. 2007). This glycosylhydrolase is se-
lective for glycoproteins and glycolipids over soluble small
molecules, suggesting that its native function is the processing
of membrane-bound substrates (Seyrantepe et al. 2004). Cleav-
age of Neu5Ac α2,3 and α2,8 linkages has been reported using
purified enzyme (Ha et al. 2004). Additionally, its enzymatic
activity is modulated by signaling events such as protein kinase
C stimulation in immune cells (Wang et al. 2004), and the pro-
tein is known to be recruited to raft domains through binding to
caveolin (Wang et al. 2002). These findings suggest that NEU3
could directly mediate lipid raft structure through glycosidase
activity (Kalka et al. 2001; Edidin 2003).
2
Alberta Ingenuity Centre for Carbohydrate Science, Department of Chemistry,
University of Alberta, Edmonton, Alberta T6G 2G2, Canada
Received on January 17, 2010; revised on May 16, 2010; accepted on
May 17, 2010
The mammalian neuraminidase (NEU) enzymes are found
in diverse cellular compartments. Members of the family,
such as NEU2 and NEU1, are cytosolic or lysosomal, while
NEU3 and NEU4 are membrane-associated. NEU enzymes
that act on substrates in the plasma membrane could mod-
ulate cellular signaling, cell surface glycoforms and the
composition of plasma membrane glycolipids. Therefore,
their substrates and mechanism of action are of interest
for discerning their physiological roles. We have studied
the structure of the human NEU3 using molecular model-
ing to predict residues involved in the recognition and
hydrolysis of glycolipid substrates. To test the model, we
have used site-directed mutagenesis of the recombinant
protein. Enzymatic studies of the relative activity of these
mutants, as well as their pH profiles and inhibition by 2-
deoxy-2,3-dehydro-N-acetylneuraminic acid, are re-
ported. Using nuclear magnetic resonance spectroscopy,
we confirmed that the enzyme is a retaining exo-sialidase,
and we propose that the key catalytic residues of the en-
zyme consist of the general acid–base D50 and the
nucleophilic Y370–E225 pair. Mutations of residues ex-
pected to interact directly with the sialic acid N5-acetyl
Glycolipids are an important component of membrane struc-
ture, and the specificity of NEU3 for these structures could
place it in a central role for mediating membrane signaling
(
Kopitz et al. 2001). Key components of rafts in mammalian
cells are the glycolipids GM1 and GM3; indeed, GM1 is used
as a marker of raft structure (Yanagisawa et al. 2006). Glyco-
sphingolipids, such as GM3 and GD1a, contain terminal α2,3-
sialic acid residues and are known to be substrates for NEU3
(
Li et al. 2001; Wang et al. 2001; Ha et al. 2004).
Mammalian NEU enzymes appear to play important roles in
(
A160, M87, I105) and C7–C9 glycerol side-chain (E113,
cell signaling and cancer (Miyagi et al. 2008). The substrate
specificity and corresponding inhibitor strategies are well es-
tablished for the viral NEU enzymes (Zhang et al. 2006).
Additionally, bacterial (Crennell et al. 1994; Chien et al.
Y179, Y181) reduced enzymatic activity. We identified sev-
eral active mutants of the enzyme which contain
modifications at the periphery of the active site. Trunca-
tions at the N- or C-terminus of more than 10 residues
abolished enzyme activity. We propose a catalytic mecha-
nism consistent with the data and identify residues that
contribute to glycolipid recognition.
1996) and trypanosoma (Demir and Roitberg 2009) enzymes
have been well studied. The same is not true for the mamma-
lian NEU enzymes. While the first of these enzymes have been
known for some time (NEU1 and NEU2) (Pshezhetsky et al.
1
997; Monti et al. 1999), NEU3 (Wada et al. 1999) and
NEU4 (Monti et al. 2004; Seyrantepe et al. 2004) have been
identified more recently. Within the family, only NEU2 has
been characterized by crystallography (Chavas et al. 2005), al-
though homology models have been proposed for all members
Keywords: glycolipid/glycosidase/mammalian glycosidase/
membrane/neuraminidase
(
Magesh et al. 2006). Little work has been done on inhibitor
1
To whom correspondence should be addressed: Tel: +1-780-492-0377; Fax:
1-780-492-8231; e-mail: ccairo@ualberta.ca
design for these enzymes. Some specific inhibitors are known
for NEU1 (Magesh et al. 2008), but none exist for the other
+
©
The Author 2010. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org 1127

A Albohy et al.
members of the family. Interestingly, recent results have sug-
gested that known inhibitors of viral NEUs are poor inhibitors
of this family of enzymes (Hata et al. 2008), suggesting the
possibility of designing specific substrates and inhibitors.
Often the enzymatic mechanism and structure of the binding
site are critical data for the design of specific NEU inhibitors
Molecular docking calculations were performed for the bind-
ing of the transition state mimic 2-deoxy-2,3-dehydro-N-
acetylneuraminic acid (DANA) to NEU3 using Autodock 4
(Morris et al. 1996; Morris et al. 1998). Autodock has been used
previously for docking of carbohydrate structures to lectin and
glycosidase binding sites (Goodsell et al. 1996; Gestwicki et al.
2001; Bernardi et al. 2002). We found that Autodock was able
to identify bound conformations of DANA in the active site
consistent with the contacts seen in previous studies of NEU2
and molecular modeling (Chavas et al. 2005; Magesh et al.
2006). The conformer selected from these calculations was
from the lowest energy cluster (−7.07 kcal/mol) with a cluster
root mean square (RMS) of 0.91. The second lowest energy
cluster was missing interactions between the C1 carboxylate
and the arginine triad (−6.95 kcal/mol). We then generated a
model of the glycosphingolipid, GM3, docked to the active site
(Figure 2). The conformer of GM3 selected was from the low-
est energy cluster (−3.96 kcal/mol) and had a cluster RMS of
0.50. Most of the variation seen in this cluster was within the
lipid tail. Both structures found a similar orientation for the ter-
minal sialic acid of GM3 and DANA and have similar molecular
contacts to the enzyme, although DANA is more deeply buried
in the active site. We tabulated the key contacts observed for si-
alic acid in both models (Table I). Important residues include:
D50, which sits over the C2 position on the a-face of the ring;
the arginine triad (R25, R45, R340), which coordinates the car-
boxylate; Y370, which is below the C2 position on the b-face of
the ring; and several hydrogen bond donors and acceptors (R45,
E113, Y179, Y181), which coordinate the O4 and glycerol side-
chain hydroxy groups. We also noted the presence of a hydro-
phobic pocket which accommodates the N5-acetyl group (A160,
M87, I105). The model of GM3 suggests that R48 could interact
with the central Gal residue of GM3, and H277 and R114 may
interact with the reducing-end Glc. We proceeded to validate the
model using SDM and activity assays.
(
von Itzstein 2007). Only a few reports of the mechanism of
the mammalian NEU enzymes are known (Buschiazzo and
Alzari 2008). Previous work in bacterial and eukaryotic siali-
dases and trans-sialidases has found that a highly conserved
tyrosine is the active site nucleophile (Watts et al. 2003), which
acts in concert with a glutamate residue (Amaya et al. 2003).
Mutation of the nucleophilic tyrosine can alter the enzymatic
mechanism (Watson et al. 2005). In the case of bacterial
NEU, such as the C. perfringens sialidase, recent structural data
have confirmed the role of the active site tyrosine residue which
forms a covalent sialoside intermediate (Newstead et al. 2005).
Although alignments of NEU3 have been performed, the role of
only a few individual amino acids in catalysis has been investi-
gated. Wang et al. (2001) reported site-directed mutagenesis
(
SDM) studies of NEU3 and found support for the role of one
member of the arginine triad (R25) common to all sialidases.
Albouz-Abo et al. (2005) have reported studies of NEU2, where
they demonstrated the enzyme had a retaining mechanism and
2
+
appeared to be Ca dependent. No corresponding investiga-
tions of NEU3 have been reported.
To improve our understanding of the mammalian NEU family,
we chose to investigate the catalytic mechanism of NEU3 and
identify the key catalytic residues. Starting from a homology
model of the protein, we identified potentially important cata-
lytic residues within the sequence. By using SDM, 14 single-
point mutants, two multi-site mutants and seven truncated
mutants were screened and evaluated for their enzymatic activ-
ity. We then isolated and characterized the kinetics of eight
mutants and the wild-type enzyme using a maltose-binding
protein (MBP)-fusion protein strategy. We tested the effect of
added metal ions, chelating agents, reducing agents and treat-
ment with alkaline pH. No evidence for divalent cations in the
enzymatic mechanism was found; however, the enzyme was
inactivated by alkaline conditions. Finally, we confirmed that
Site-directed mutagenesis of recombinant NEU3
Expression of NEU3 has been reported in bacterial and mamma-
lian cells (Wang et al. 2001; Ha et al. 2004). For our purposes,
we required a system which could provide large quantities of
protein for assay and could be easily manipulated. We found that
mammalian expression systems had generally high sialidase
background, complicating their use in the current study. Initial
work with NEU3 in a pRSET vector gave poor yields in our
hands (Ha et al. 2004). Therefore, we inserted the gene into
pET28a vector with an N-terminal His-tag (see Supporting in-
formation, Figure SI2) and expressed the protein in E. coli cells
(see Materials and methods). We used the His-tag to confirm
expression of the protein, and we observed intact NEU3 as well
as proteolytic fragments of the protein (see Supporting infor-
mation, Figure SI3). The intact protein was estimated to be
close to 30% of the total His-tag protein found in lysate. How-
ever, the expressed protein was active, and mock-transfected
cells showed no activity. Before conducting assays with these
proteins, we sought to confirm that the proteolytic fragments
would not complicate the assay data. We generated several
truncated forms of the protein, with either N-terminal (−10,
−20, −50) or C-terminal (−10, −21, −22, −26) truncations. In
both cases, truncation of more than 10 residues from the protein
1
the enzyme is a retaining sialidase using H nuclear magnetic
resonance (NMR) spectroscopy.
Results
Homology modeling and molecular docking of NEU3
To date, no crystallography studies of NEU3 have been re-
ported. However, homology models of the protein and other
mammalian NEU have been described (Magesh et al. 2006).
In order to identify potential catalytic residues, we followed
a similar strategy. NEU2 and NEU3 have 38% identity after
alignment using ClustalW (Figure 1) (Thompson et al.
1
994). Using the alignment, we generated a homology model
of the protein based on the crystal structure data for NEU2
(
(
Lambert et al. 2002; Chavas et al. 2005) using Modeller
Lambert et al. 2002) (see Figure SI1 for model quality indi-
ces). Our model was qualitatively similar to that reported by
Magesh et al., and we set out to identify potentially impor-
tant contacts for substrate and inhibitor structures in the
binding site.
1128

Insight into substrate recognition and catalysis by human NEU3
R25
R45
D5 0E 51
NEU3
NEU2
MEEVTTCSFNSPLFRQEDDRGITYRIPALLYIPPTHTFLAFAEKRSTRRDEDALHLVLRR 60
MASLPVLQKES-VFQSG---AHAYRIPALLYLPGQQSLLAFAEQRASKKDEHAELIVLRR 56
*
.:.. . :* :*:.
. :********:* :::*****:*::::**.* :****
1
3 14
M8 7N 88
I105V107
E1 R1
NEU3
NEU2
G--LRIGQLVQWGPLKPLMEATLPGHRTMNPCPVWEQKSGCVFLFFICVRGHVTERQQIV 118
GDYDAPTHQVQWQAQEVVAQARLDGHRSMNPCPLYDAQTGTLFLFFIAIPGQVTEQQQLQ 116
*
: *** . : : :* * ***:*****::: ::* :*****.: *:***:**:
A160 G162
NEU3
NEU2
SGRNAARLCFIYSQDAGCSWSEVRDLTEEVIGSELKHWATFAVGPGHGIQLQS--GRLVI 176
TRANVTRLCQVTSTDHGRTWSSPRDLTDAAIGPAYREWSTFAVGPGHCLQLHDRARSLVV 176
:
*.:*** : * * * :**. ****: .**. :.*:******** :**:.
**:
Y17 Y9 181
E225
NEU3
NEU2
PAYTYYIPSWFFCFQLPCKTRPHSLMIYSDDLGVTWHHGRLIRPMVTVECEVAEVTGRAG 236
PAYAYRK------LHPIQRPIPSAFCFLSHDHGRTWARGHFV-AQDTLECQVAEVETGE- 228
*
**:*
::
:. * :: : *.* * ** :*::: . *:**:****
R245
H277
NEU3
NEU2
HPVLYCSARTPNRCRAEALSTDHGEGFQRLALSRQLCEPP-HGCQGSVVSFRPLEIPHRC 295
QRVVTLNARSHLRARVQAQSTNDGLDFQESQLVKKLVEPPPQGCQGSVISFP-------- 280
:
*: .**: *.*.:* **:.* .**. * ::* *** :******:**
R340
NEU3
NEU2
QDSSSKDAPTIQQSSPGSSLRLEEEAGTPSESWLLYSHPTSRKQRVDLGIYLNQTPLEAA 355
-------SP---RSGPGS-----------PAQWLLYTHPTHSWQRADLGAYLNPRPPAPE 319
:
*
:*.***
. .****:***
**.*** ***
*
.
Y370
NEU3
NEU2
CWSRPWILHCGPCGYSDLAALEE--EG--LFGCLFECGTKQECEQIAFRLFTHREILSHL 411
AWSEPVLLAKGSCAYSDLQSMGTGPDGSPLFGCLYEAN---DYEEIVFLMFTLKQAF--- 373
.
**.* :* *.*.**** ::
:* *****:*..
: *:*.* :** :: :
NEU3
NEU2
QGDCTSPGRNPSQFKSN 428
----------PAEYLPQ 380
*
::: .:
Fig. 1. Alignment of NEU2 and NEU3. Using ClustalW, the sequences of NEU2 (NEUR_2; Q9Y3R4) and NEU3 (NEUR_3; Q9UQ49) were aligned. The
sequences show 38% identity, and individual sites are noted as being identical (*), conserved (:) or semi-conserved substitutions (.). NEU3 is shown on the top line
of each row. NEU3 sites studied by mutagenesis are indicated by boxes, examples from previous studies (light) (Wang et al. 2001) and this work (dark) are shown.
Key NEU2 active site residues are underlined (Chavas et al. 2005).
abolished activity (see Supporting information, Figure SI4 and
Table SII). Since the major proteolyzed NEU3 band is more
than 10 kDa lower in molecular weight than the intact protein,
we conclude that these fragments are unlikely to contribute to
observed activity. We proceeded to test the effects of point mu-
tations on the activity of NEU3.
to phenylalanine and E51 to serine, consistent with our model,
which places these residues distant from the binding site. Mu-
tations targeted at the putative N5 binding pocket showed
mixed results, with minor (A160G) to substantial (M87G,
I105G) reductions in enzymatic activity. We also tested combi-
nations of these mutations as a double- (M87G + I105G) and
triple-mutant (A160G + M87G + I105G), and both showed un-
detectable enzymatic activity. These results suggest that N5
interactions are necessary for recognition of the substrate or
that these residues play an essential structural role in ordering
the binding site (Watts et al. 2006). Mutations of sites predicted
to interact with the O4 (R45V) and glycerol side-chain
(E113A, Y179F, Y181F) all eliminated enzymatic activity,
supporting the contribution of these groups to recognition of
the sialic acid residue. Mutation data from these experiments
were mapped to the active site model (Figure 2C). We also
compared the activity of C-terminal truncations of the protein,
with 10-residue truncation (G419STOP) leading to a slight in-
crease in activity, and larger deletions (L408STOP) eliminating
all activity. We proceeded to isolate and characterize several of
these mutants to gain additional insight into their activity.
Based on the contacts observed in our model, we selected
potentially important residues that could participate in catalysis
or substrate recognition and generated mutants at these posi-
tions using the pET28a vector (Table I). Previously reported
SDM results relevant to the model are also included in Table I
for comparison (Wang et al. 2001). Mutants were expressed
and tested for enzymatic activity relative to the wild-type pro-
tein (Figure 3). We found that several mutations at positions
predicted to interact at C2 (D50S, Y370F and E225S) and
C1 (R245A and R340A) greatly reduced enzymatic activity.
Interestingly, mutation of E225 to serine eliminated enzyme
activity. Residue E225 is not predicted to interact directly with
the substrate but likely interacts with Y370 to catalyze forma-
tion of the covalent sialoside intermediate (Amaya et al. 2004).
We observed slightly increased activity after mutation of H277
1129

A Albohy et al.
A
B
C
H277
Y179
Y181
Y370 E225
N88
R340
R25
E113
mutated (active)
O
HOOH
M87
I105
A160
previously mutated (active)
mutated (inactive)
O
O
R245
AcHN
HO
R48
OH
previously mutated (inactive)
D50 E51
1130

Insight into substrate recognition and catalysis by human NEU3
Table I. Predicted substrate interactions in NEU3
matching previous reports (pH optimum 3.8–5) and the wild-
type protein showing a slight shift to a more alkaline range (pH
optimum 5–6) (Monti et al. 2000; Wang et al. 2001) (see Sup-
porting information, Figure SI5).
Relativea
b
Putative
interaction
c
Site
activity
Source
WT
D50S
Y370F
Y370C
E225S
R245A
R340A
R25H
100 ± 2
10.4 ± 0.3
0.1 ± 0.1
12.1
0.3 ± 0.1
−1.2 ± 0.1
1.3 ± 0.2
16.2
na
C2
C2
C2
Y370
COO
COO
COO
N5Ac
N5Ac
N5Ac
N5Ac
N5Ac
O4
O7–O8
O8
O8
Truncation
Truncation
Truncation
Truncation
Truncation
Truncation
Truncation
We expressed and purified eight mutants that were either ac-
tive (A160G, E51S, H277F, R48N, R48S), weakly active
(
(
M87G) or inactive (D50S, Y370F) for further characterization
see Supporting information, Figure SI6). Typical yields of
Wang et al.
Wang et al.
−
1
protein were 8 mg L of culture. We first examined the depen-
dence of enzymatic activity on pH for each mutant (see
Supporting information, Figure SI7). We observed that several
mutants showed little deviation from the behavior of the wild-
type protein (pH optimum 4–5; E51S, H277, M87G), while
others showed increased relative activity at alkaline pH
A160
M87G
I105G
79 ± 4
10.7 ± 0.2
4.8 ± 0.1
1.2 ± 0.1
0.3 ± 0.1
1.9 ± 0.1
8 ± 3
I105G + M87G
I105G + M87G + A160G
R45V
E113A
Y179F
Y181F
G419STOP
L408STOP
I407STOP
T403STOP
N-10
N-20
N-50
H277F
E51S
E51D
G162A
N88D
R114A
R114Q
V107M
(A160G, R48N, R48S). Inactive mutants (D50S, Y370F) did
not show significant activity at any pH tested. Kinetic studies
of the purified mutants and wild-type protein were also con-
ducted (Table II; see Supporting information, Figures SI8
1.8 ± 0.5
7 ± 2
d
144 ± 9
1.1 ± 0.1
0.8 ± 1
−3 ± 2
80.3
−0.6 ± 0.3
−1.5 ± 0.2
124 ± 6
119 ± 3
75.7
and SI9). The Michaelis constant (K ) for these proteins indi-
m
cated that mutation at E51 and R48 resulted in only slight
changes in substrate affinity relative to wild type. Mutation
at A160 and H277 led to an apparent decrease in K . In con-
m
trast, mutation at M87 caused a dramatic loss of affinity,
suggesting that this site is important for substrate recognition.
Mutation of M87 also resulted in a large drop in the rate of
d
catalysis by the enzyme (k ). These results may indicate that
Wang et al.
Wang et al.
Wang et al.
Wang et al.
Wang et al.
Wang et al.
cat
13.3
9.7
96.5
71.6
M87 is important for orienting the substrate in the active site.
The corresponding Met residue has been observed to adopt
several conformations in the active site of trans-sialidase en-
zymes (Watts et al. 2006). Both E51S and R48S showed a
large increase in kcat (ca. 200% of wild type). Neither the
D50S or Y370F proteins retained enough activity to determine
K or k values.
9.0
a
Mutation experiments are tabulated as % of activity relative to wild-type en-
zyme, using 4MU-NA as a substrate at pH = 5.0. Data from Wang et al. were
reported at pH 4.6 (Wang et al. 2001).
m
cat
b
We also tested the effect of point mutations on competitive
inhibition of NEU3 by DANA. For mutants with measurable
activity, the IC50 of DANA was determined (Table I). Enzyme
activity was measured by monitoring cleavage of the fluoro-
genic substrate, 4-methylumbelliferyl α-D-N-acetylneuraminic
acid (4MU-NA). Cleavage of 4MU-NA by wild-type enzyme
was inhibited by DANA at 43 ± 7 µM, similar to previous re-
ports (Magesh et al. 2008). Slight increases to the inhibitory
potency of DANA were seen in E51S (29 ± 4 µM) and
H277F (31 ± 4 µM). Both R48 mutants showed moderate de-
creases for inhibition by DANA (R48N, 63 ± 9 µM; R48S, 53 ±
Source of activity data if other than this work.
c
Assignment of interactions with the substrate is made from the homology
model discussed above.
d
When observed at pH = 4.5, the activity of the E51S mutant was 96 ± 9% and
of G419STOP was 95 ± 4%.
Purification and characterization of Neu3 mutants
2
+
Initial attempts at purification of the protein using a Ni -NTA
strategy were unsuccessful. This may have been due in part to
the enzymes instability to alkaline pH (vide infra). We, there-
fore, used an alternative affinity tag, the MBP fused to the N-
terminus of the sequence in place of the His-tag (see Supporting
information, Figure SI2). Expression and purification of the
MBP-NEU3 using an amylose column were successful, and
the isolated protein was active. Attempts to remove MBP by
cleavage of the linker with Factor Xa resulted in protein aggre-
gation; therefore, all proteins were characterized as the MBP-
fusion. The pH profiles of the wild-type and fusion proteins
were slightly different, with the fusion protein more closely
9
µM), and the A160G and M87G mutants showed a major
loss of potency for DANA (>300 µM). These results suggest
that residues A160 and M87 are both critical for substrate
recognition.
Effect of additives on NEU3 activity
Previous studies of mammalian NEU2 found that enzyme ac-
tivity was dependent upon divalent metal ions, and large
Fig. 2. Homology model of the NEU3 active site and molecular docking. A homology model of NEU3 was constructed using Modeller 9v6 (Lambert et al. 2002).
Molecular docking with Autodock 4.0 was used to determine the binding mode of (A) DANA and (B) the glycolipid, GM3. Views of the active site are shown in
stereo (crossed). Only the terminal Neu5Ac and Gal residues of GM3 are shown in (B) for clarity. The binding site for sialic acid is similar to that of NEU2, with the
O4 and N5 positions directed into the binding pocket, and the C2 and glycerol side-chain solvent exposed. Amino acid residues for site-directed mutagenesis were
selected to test the model. (C) A schematic of the binding pocket is shown with previously tested mutations (Wang et al. 2001) (hashed) and those tested in this
study (open) shown. Mutations that significantly decreased activity (<10%) are shown in solid red, and those that did not eliminate activity are shown in dashed
green.
1131

A Albohy et al.
1.0
0.0
Fig. 3. Relative activity of point mutations generated. NEU3 proteins were expressed in E. coli (pET28a) and assayed in bacterial lysate (pH 5.0, 0.1 M sodium
acetate). Mock cells showed no sialidase background activity. Assays were conducted at a total protein concentration of 25 mg/mL as determined by A280. Lysate
was incubated with 2 µL of 10 mM 4MU-NA for 60 min at 37°C, and activity was determined by fluorescence (365 nm excitation; 445 nm emission). Assays were
performed in triplicate and error bars indicate the standard deviation of each point. Results are shown normalized to a wild-type control run in parallel.
2
+
2+
increases in relative activity were observed for Ca , Mn and
initial formation of free α-D-N-acetylneuraminic acid, though
mutarotation to the β-anomer will occur subsequent to hydroly-
sis. A time-course of 4MU-NA hydrolysis by NEU3 is shown in
Figure 4 and confirms that NEU3 was indeed a retaining siali-
dase. Additional experiments with MBP-NEU3 gave the same
result (see Supporting information, Figure SI11).
2
+
Mg ions (Albouz-Abo et al. 2005). We tested the effect of
additives on the enzymatic activity of purified MBP-NEU3
for 4MU-NA (see Supporting information, Figure SI10). Rel-
ative to control samples, we observed only minor enhancement
2
+
of NEU3 activity in the presence of Ca (7 mM, 131 ± 3%)
2+
and Mn (7 mM, 136 ± 4%). Treatment with ethylenediami-
netetraacetic acid (EDTA) caused only a minor decrease in
activity (91 ± 4%). Therefore, we concluded that NEU3 activ-
ity was not dependent on divalent metal ions. Despite the large
number of cysteine residues in the sequence, treatment of the
enzyme with DTT caused only a minor reduction in activity.
Addition of surfactants (Triton X-100, 93 ± 1%; and sodium
cholate, 72 ± 2%) resulted in a slight decrease in activity.
We observed that copper salts were potent inhibitors of siali-
Discussion
In this study, we provide a model of the substrate recognition
and binding site topology of the human neuraminidase, NEU3.
Table II. Kinetics of 4MU-NA hydrolysis by MBP-NEU3
2
+
Mutant
Vmax [U s−
1 a
]
Km [µM]
a
kcat [h−1
]
DANA IC50 [µM]
a
dase activity (Cu 0.1 mM, 3 ± 1%) and that the enzyme
was unstable to brief treatment with alkaline pH (pH 11.5,
WT
A160G
E51S
H277F
M87G
R48N
R48S
D50S
Y370F
5.6 ± 0.1
7.1 ± 0.3
11.7 ± 0.4
6.9 ± 0.2
2.0 ± 0.2
3.8 ± 0.1
12.2 ± 0.5
na
45 ± 3
21 ± 4
38 ± 5
18 ± 3
140 ± 30
33 ± 4
32 ± 6
na
23 ± 1
27 ± 1
51 ± 1
27 ± 1
8 ± 1
32 ± 1
44 ± 2
na
43 ± 7
300 ± 25
29 ± 4
31 ± 4
>500
63 ± 9
53 ± 9
na
1
0 min, 10 ± 5%). Both of these findings are consistent with
observations made with NEU2 (Albouz-Abo et al. 2005).
NEU3 catalytic mechanism
b
To obtain additional insight into the mechanism of NEU3
catalysis, we observed the enzymatic cleavage of 4MU-NA by
1
na
na
na
na
H NMR spectroscopy. The 4MU-NA substrate contains a sialic
acid in an α2,3 configuration. All known native sialidases are
exclusively retaining enzymes (Watson et al. 2003). Observation
of 4MU-NA cleavage by a retaining sialidase should result in
a
Error calculated from nonlinear regression.
b
Concentration of enzyme used for Vmax determination was 0.83 µM for all
mutants except R48N, which was run at 0.42 µM.
1132

Insight into substrate recognition and catalysis by human NEU3
αH3eq
βH3eq
βH3ax
αH3ax
7
6
5
2
1
3 min
0 min
3 min
8 min
0 min
0
min
2.80
2.60
2.40
2.20
2.00
1.80
1.60
ppm
1
1
Fig. 4. Catalytic mechanism of NEU3 studied by H NMR spectroscopy. The conversion of 4MU-NA to N-acetylneuraminic acid was observed by H NMR
spectroscopy (600 MHz). 4MU-NA was dissolved in D O (pH = 4.5, 0.1 M sodium acetate buffer; 2.9 ppm αH3eq). NEU3 (pET28a, 3.8 mg/mL total protein
2
concentration) was added to the NMR tube, and spectra were recorded at the indicated time points. During this period, only signals for the formation of α-D-N-
acetylneuraminic acid (1.6 ppm αH3ax, 2.7 ppm αH3eq) were observed. Signals for the β-D-N-acetylneuraminic acid (1.9 ppm βH3ax, 2.3 ppm βH3eq) were not
observed, indicating that the product was formed with retention of configuration (Watson et al. 2003). Experiments with purified MBP-NEU3 gave similar results
(
see Supporting information, Figure SI11).
Structural studies of this membrane-associated enzyme have
been hampered by its propensity for aggregation, and the mod-
el reported here adds critical new insight to the mode of
substrate recognition and catalysis of NEU3. The model was
validated by testing the activity of several enzyme mutants,
which confirmed the identity of key catalytic residues and pro-
tein–ligand contacts. These experiments provide the most
detailed model of the NEU3 active site to date and also identify
residues of the enzyme which may be critical for recognition of
glycolipid substrates. The binding mode of the small molecule
inhibitor, DANA, and the glycolipid substrate, GM3, to NEU3
was predicted using molecular docking. We report the charac-
terization of several important active site residues within the
enzyme and collate these data with previously tested sites
eral predictions which arise from the model. The role of E225
in catalysis had not been tested previously, and only the iden-
tity of R25 within the arginine triad (R25, R245 and R340) had
been confirmed (Wang et al. 2001). Furthermore, we examined
additional residues that likely contribute to substrate recogni-
tion through binding to sialic acid side chains. Mutation of
residues predicted to form a hydrophobic binding pocket for
the N5 acyl group (A160, M87, I105) and the glycerol side-
chain (R45, E113, Y179, and Y181) of sialic acid all lead to
reduced activity in NEU3. It is interesting to note that Y179
and Y181 were previously proposed as part of a transmem-
brane domain (residues 174–194) within NEU3 (Miyagi et
al. 1999). This model has been revised based on alignment
to NEU2 (Monti et al. 2000; Wang et al. 2001) and evidence
that NEU3 is not a transmembrane protein but is instead a
peripheral membrane-associated protein (Zanchetti et al.
2007). Our data are consistent with this conclusion. Additional
work will be needed to confirm the role of the putative N5 and
glycerol-binding pockets. However, we note that the glycerol
side-chain of Neu5Ac is much more solvent accessible in the
model than is observed in many bacterial sialidases, and even
NEU2. In the NEU2 binding site, this pocket is obscured by
the presence of a glutamine residue (Q270). The reason for this
difference is not clear, although this feature may be one which
can be exploited for the design of specific inhibitors.
(
Wang et al. 2001). Additionally, we have confirmed the struc-
tural predictions made by homology modeling of NEU3 to
NEU2. Due to the low sialidase background found in E. coli,
we were able to screen a large number of mutants in cell lysate
and map out the topology of the binding site. Our studies were
enabled by the development of a robust purification strategy
for recombinant protein using an MBP-fusion tag expressed
in E. coli cells. Using purified protein, we were able to char-
acterize the pH profile, kinetics and inhibition of NEU3. We
confirmed that the enzyme is a retaining sialidase. Finally,
we tested the role of metal ions, reducing agents, alkaline pH
and surfactants on enzyme activity.
Models of a small molecule inhibitor, DANA, and the glyco-
lipid, GM3, docked to the active site of NEU3 provide valuable
insight into the recognition of glycan substrates by the enzyme.
In our model of GM3 substrate binding (Figure 2B), we ob-
served a possible contact between R48 and the central Gal
Our initial screening of NEU3 mutants confirms the pre-
dicted topology of the binding site found in the homology
model shown in Figure 2. A similar model was first proposed
by Magesh et al.; however, our results are the first to test sev-
1133

A Albohy et al.
residue of the ganglioside headgroup. It is interesting to note
that the O4 position of Gal is oriented towards R48, implying
that attachment of a saccharide residue at this position could
interfere with binding (See Supporting information, Figure
SI12). Examination of previous substrate specificity studies
is consistent with this hypothesis. Several groups have shown
that glycans that lack branching at the residue adjacent to
Neu5Ac (whether α2,3- or α2,8-linked) are good substrates
the sequence of NEU3 was first analyzed using SDM (Wang
et al. 2001). Our study was intended to identify the active site
topology and catalytic residues of the enzyme. The expected
catalytic motif includes an Asp on the b-face of the ring and
a Tyr–Glu pair on the opposite face. We propose that, consis-
tent with other GH33 family members, Tyr acts as the
nucleophile after deprotonation by Glu, and the Asp residue
serves as a general acid–base for the departing reducing-end
glycan and an incoming water. The Tyr–Glu catalytic pair is
a common feature of sialidase and trans-sialidase enzymes
(Amaya et al. 2004) and is most clearly supported by the ob-
servation of covalent sialyl-enzyme adducts (Watts et al.
2003; Watts et al. 2006; Newstead et al. 2008). We propose
that the identities of the catalytic residues in NEU3 are D50,
Y370 and E225, which carry out hydrolysis of the substrate
as shown in Figure 5.
(
sialyllactose, GM3, GD1a, GD1b, GD3 and GT1b), while gly-
cans that have branching at the adjacent residue are poor
substrates (GM1, GM2) (Monti et al. 2000; Kopitz et al.
2
001; Wang et al. 2001; Ha et al. 2004; Papini et al. 2004).
Glycolipids are the preferred substrate of NEU3; however,
MU-NA is a moderate substrate for the enzyme, which allows
4
for fluorescent assays to be carried out (Monti et al. 2000).
Using 4MU-NA, we were able to determine both the kinetic
parameters of the active mutants, as well as the potency of a
common sialidase inhibitor, DANA. Our measurement of the
inhibition of 4MU-NA cleavage of NEU3 by DANA is in
agreement with previous reports (Magesh et al. 2008). Mutation
of D50 and Y370 eliminated measurable NEU3 activity, consis-
tent with their role in catalysis. Modification of sites within the
putative N5 pocket (A160, M87) reduced the potency of DANA
significantly, with the M87G mutant exhibiting particularly low
We have identified the catalytic residues of NEU3 and con-
firmed that it is a retaining exo-sialidase. These results help
support the homology model of NEU3 by confirming sites
which interact with substrates and contribute to enzyme spec-
ificity. These data will be essential in the design of
experiments to observe NEU3 activity in the membrane
(
Cairo et al. 2010). Together, these studies provide critical in-
sight into the active site topology of a human glycosidase
enzyme involved in the pathology of cancer (Miyagi et al. 2008).
kcat. Mutation of sites at the periphery of the binding site (E51,
H277) had only minor effects on inhibition by DANA, and E51S
showed a large increase in rate of turnover. Alteration of R48 to
serine, a site that we predict to interact with the glycan residue at
the reducing end of the sialic acid, also leads to increased rate of
turnover. The specificity of the active mutants for glycolipid
substrates will be of interest in future work.
Materials and methods
NEU3 expression in E. coli (pET28a vector)
Sialidase expression vectors were generated by inserting hu-
man membrane-associated ganglioside sialidase 3 (NEU3)
between the BamHI and XhoI sites of pET28a (Novagen,
Gibbstown, NJ). The wild-type NEU3 was modified to have
BglII and XhoI restriction sites by carrying out a polymerase
chain reaction (PCR) using the Neu3 template, PfuUltra II
fusion HS DNA polymerase (Stratagene, Santa Clara, CA),
a forward primer for BglII (5′-GGG ATA T AGA TCT ATG
GAA GTG ACA ACA TGC-3′) and a reverse primer for
XhoI (5′-ATT TAC TCG AGT TAA TTG CTT TTG AAT
TGG CTT GGG TT-3′). The reaction was cycled once at
95°C for 2 min followed by 30 cycles of 95°C for 20 s,
50°C for 20 s and 72°C for 25 s and 1 cycle of 72°C
for 3 min. The modified Neu3 was then digested with BglII
and XhoI restriction enzymes and the pET28a vector di-
gested with BamHI and XhoI. The digested Neu3 and
pET28a were ligated together with T4 DNA ligase at a
3:1 insert to vector ratio. The resulting expression vector
was transformed into electroporation competent XL1-Blue E. coli
(Stratagene). Competent cells (50 µL) were mixed with plasmid
DNA (3 µL, 10 ng/µL) and transferred to Gene Pulser cuvettes
(BioRad, Hercules, CA). They were then electroporated
(1.0 kV, 800 Ω), immediately mixed with 1 mL lysogeny broth
(LB) media, grown with shaking at 37°C, 240 rpm, for 1 h and
plated on LB agar with kanamycin (25 µg/mL). Clones were se-
lected and grown in 25 µg/mL kanamycin LB medium, and the
plasmids were isolated using the GeneJET Plasmid Miniprep Kit
(Fermentas, Glen Burnie, MD). Clones were verified first by aga-
rose gel electrophoresis, to check plasmid size, and then by gene
sequencing with the BigDye Terminator Kit V.3.1 (Applied Bio-
systems, Carlsbad, CA).
We first attempted to purify NEU3 using an N-terminal His-
tag. We found this strategy gave poor yields of protein, and we
tried instead an N-terminal MBP-fusion protein. This strategy
gave much better yields, and the enzyme gave a pH profile that
was more consistent with previous reports. Although the MBP
tag was linked to the protein with a Factor Xa cleavage site, we
observed that attempts to isolate NEU3 after cleavage failed due
to protein aggregation. Therefore, the more stable MBP-fusion
protein was used for later experiments. These observations sug-
gest that attempts at characterizing this enzyme by
crystallography may be difficult. We examined the role of a
number of conditions in the enzymatic activity of purified
NEU3. We found little effect on catalysis from divalent metal
2
+
2+
ions (Ca and Mn ) or EDTA, unlike the homolog NEU2
Albouz-Abo et al. 2005). Additionally, treatment with sur-
factants had only a small effect on enzymatic activity for
MU-NA. This is in contrast to observations of NEU3
(
4
cleavage of glycolipid substrates, which found a large de-
pendence upon surfactants. This discrepancy may indicate
that the surfactant is critical for solubilizing the substrate,
more than affecting enzyme structure. We also confirmed that
the enzyme was unstable to alkaline pH, consistent with previ-
ous reports (Zanchetti et al. 2007).
All characterized native exo-sialidases are retaining en-
zymes, although mutagenesis can result in an inverting
mechanism (Watson et al. 2003). We confirmed that NEU3
1
is indeed a retaining sialidase using H NMR spectroscopy.
The proposed mechanism of catalysis by exo-sialidases has
undergone some revision (Buschiazzo and Alzari 2008) since
1134

Insight into substrate recognition and catalysis by human NEU3
A
Y370
O
B
Y370
R340
NH
E225
E225
R340
O
O
O
O
O
O
H
O
2
N
H
NH
NH
H
NH
2
2
H N
O
O
HOOH
HOOH
O
OH
OH
OH
HO
O
O
AcHN
HO
AcHN
HO OH
OH
OH
O
H
O
H
O
O
HO
HO
H
O
H
O
O
O
H
O
O
D50
D50
Y370
C
D
E225
Y370
R340
NH
E225
O
OH
O
O
H N
R340
NH
O
2
H
NH
2
O
O
O
HOOH
HOOH
2
H N
O
O
NH
2
O
OH
AcHN
HO OH
AcHN
O
HO
OH
H
O
H
O
OH
O
O
D50
D50
Fig. 5. Proposed catalytic mechanism for NEU3. Based on the mutagenesis and NMR studies described, we propose the following enzymatic mechanism for NEU3
cleavage of sialosides. (A) In the first step, the reducing end of the saccharide dissociates via protonation by water or the D50 residue. (B) This results in the
formation of an oxacarbenium intermediate which is attacked by the Y370 nucleophile, which requires the participation of the neighboring E225 residue. Steps A–B
may occur concurrently or stepwise. (C) The enzyme-sialic acid adduct is hydrolyzed through catalysis by D50. (D) Finally, the free α-D-Neu5Ac is released from
the active site.
Site-directed mutagenesis of Neu3 and purification of fusion
proteins
5 min followed by a 30 s, 42°C heat shock and returned to
ice for 2 min. The transformation reactions were plated on LB
agar plates with final concentrations of 35 µg/mL chloramphen-
icol and 25 µg/mL kanamycin. Clones were selected and grown
in 35 µg/mL chloramphenicol and 25 µg/mL kanamycin LB me-
dium for 7 h, after which 50% glycerol was added to a final
concentration of 15%, and these stocks were stored at −80°C.
Some cells were scraped from the frozen bacteria glycerol stock
and transferred into 10 mL 1% glucose LB medium with chlor-
amphenicol and kanamycin. The cultures were incubated on a
shaker at 37°C at 225 rpm for 16 h. A culture of these cells (di-
luted to OD600 of 0.3 in 10 mL) was collected by centrifugation
at 16,000 × g for 2 min and the supernatant decanted. These
cells were then resuspended and grown at 37°C with shaking
at 225 rpm in another 10 mL of 1% glucose LB media with
antibiotics for >1 h until the OD600 reached 0.5. Isopropyl β-
D-1-thiogalactopyranoside (IPTG) was then added to a final
concentration of 100 µM to induce transcription and incubated
at 20°C with shaking at 225 rpm for 16 h. The culture was di-
luted to aliquots (OD600 of 3.5 for 1 mL) and the cells collected
Mutagenesis was carried out using the Quikchange II XL site-
directed mutagenesis kit (Stratagene) following the manufac-
turer’s protocol. The following mutations were generated:
R45V, D50S, E51S, M87G, I105G, E113A, A160G, Y179F,
Y181F, E225S, R245A, H277F, R340A, Y370F, L408Stop,
G419Stop, M87G-I105G double mutant and a M87G-I105G-
A160G triple mutant. The pET28a-Neu3 template described
above was used as the dsDNA template for all mutations ex-
cept the double and triple mutants, which used M87G and the
M87G-I105G mutant plasmids, respectively, as templates. The
oligonucleotide mutation primer designs are provided in Sup-
porting information (Table SI12). Following mutagenesis,
transformed clones were selected and grown in kanamycin
LB medium. Plasmids were then isolated and confirmed as
above. To express the mutant NEU3, plasmids were trans-
formed into E. coli (DE3, Novagen) by adding 50 ng
plasmid to 50 µL competent cells and incubating on ice for
1135

A Albohy et al.
by centrifugation at 16,000 × g for 2 min. The supernatants
were decanted and the pellets stored in 1.5 mL microcentri-
fuge tubes at −80°C.
for 5 min. Total protein concentration in the lysate was deter-
mined using the A280 of serial dilutions. To determine NEU
activity, 2′-(4-methylumbelliferyl)-α-D-N-acetylneuraminic ac-
id (4MU-NA) was used as a fluorogenic substrate. For pH
profiles, enzyme activity was measured at pH 3.7, 4.0, 4.25,
4.50, 4.75, 5.0, 5.5 (0.1 M sodium acetate), 5.25, 6.0, 6.5,
7.0, 7.5 and 8.0 (0.1 M sodium phosphate). All samples were
run in triplicate, with negative controls containing buffer alone
and 4MU-NA in buffer. Reaction mixtures of lysate (0.1 mg/
mL total protein, 200 µL) and controls were briefly vortexed
and incubated for 1 h at 37°C. To stop the reactions, 200 µL
0.2 M glycine/NaOH pH 10.2 was added to each reaction.
The mixtures were then centrifuged at 16,000 × g for 10 min.
Supernatant (200 µL) was removed from each reaction to a 96-
well plate. Fluorescence was then measured on a SpectraMax
M2 microplate reader (Molecular Devices, Sunnyvale, CA)
with excitation 365 nm and emission 445 nm. For comparison
of activity between mutants, total protein concentration was
adjusted to 25 mg/mL, and reactions were carried out at pH =
5.0 in 0.1 M sodium acetate buffer.
MBP-fusion proteins (MBP-NEU3) were generated from a
pMAL-c2x/Neu3 construct. The plasmid was cloned using
pRSET/Neu3 as a template. The Neu3 gene was PCR-amplified
using the forward primer OBZ51 (5′-AA TTC GAA TTC ATG
GAA GAA GTG ACA ACA TGC TCC-3′) to introduce an
EcoRI site at the 5′ end and the reverse primer OBZ52 (5′-
GG TTG AAG CTT TTA ATT GCT TTT GAA TTG GCT
TGG-3′) to introduce a HindIII site at the 3′ end of the Neu3
gene. The PCR product was purified from the reaction mixture
®
using a QIAquick PCR purification kit (Qiagen, Valencia,
CA). The final DNA fragment resulting from the PCR reaction
was digested with the restriction enzymes EcoRI and HindIII,
isolated from a 1% agarose gel, and was then ligated into the
pMAL-c2x vector which had been digested with EcoRI and
HindIII. The resulting MBP-NEU3 plasmid was transformed in-
to E. coli DH5α cells (Invitrogen, Carlsbad, CA). Selected
transformants were screened for the presence of the recombinant
plasmid by restriction mapping and DNA sequence analysis.
The clone was termed pMBP-Neu3 and was subsequently trans-
formed into a host strain, competent E. coli TB1 cells (New
England Biolabs, Ipswitch, MA) for further expression.
Protein was isolated by culturing E. coli TB1 cells con-
taining plasmid pMBP-Neu3 and its mutants in LB medium
containing ampicillin (100 μg/mL) to OD600 of 0.5 at 37°C.
Production of the fusion protein was induced by adding
IPTG to a final concentration of 0.3 mM at 20°C. Cells
were harvested by centrifugation after 16 h of induction.
The pellet was resuspended (50 mL/L of medium) in resus-
pension buffer (20 mM morpholinopropane sulfonic acid
Enzyme kinetics and inhibition by DANA
Purified MBP-NEU3 fusion proteins were used to determine
the kinetics of 4MU-NA cleavage. Reactions were prepared
by mixing 4MU-NA (25, 50, 75, 100, 150, 200, 250,
3
00 µM), purified fusion protein (0.04 mg/mL), and BSA
(
4 mg/mL) in sodium acetate buffer (0.1 M, pH = 5.0). The
rate of product formation was observed in real-time by fluores-
cence (340 nm excitation and 455 nm emission) on a
SpectraMax M2 microplate reader (Molecular Devices) for
90 min at 37°C. Kinetic parameters were determined by linear
regression. For inhibitor studies, the kinetic assay described
above was modified by adding 2 µL of a serial dilution of
DANA to 18 µL of enzyme (0.1 mg/mL) to achieve final in-
hibitor concentrations of 5, 10, 20, 50, 100, 200, 500 and
(
MOPS), pH 7.2, 200 mM NaCl, 1 mM EDTA and 0.1%
Triton X-100) supplemented with 1 complete protease inhib-
itor tablet (Roche). The lysate was passed through a cell
disruptor once at 20,000 psi and then immediately pelleted
by centrifugation at 105,000 × g for 60 min at 4°C. The
supernatant was loaded onto an amylose column (New Eng-
land Biolabs) equilibrated with 20 mM MOPS (200 mM
NaCl, pH 7.2). MBP-fusion protein was eluted with running
buffer containing 10% glycerol (v/v) and 10 mM maltose.
Proteins were detected by immunoblotting. For proteins
expressed in pET28a, enzyme extract was separated using
1000 µM. The mixture was incubated at 37°C for 30 min. Af-
ter incubation, 2 µL of 10 mM 4MU-NA and 18 µL of 0.1 M
acetate buffer (pH = 5.0) were added to the reaction tube and
incubated for 1 h. The reactions were stopped and measured
as described above.
Homology modeling of NEU3
NEU2 and NEU3 sequences were obtained from the Uniprot
database (accession NEU2: Q9Y3R4 and NEU3: Q9UQ49).
The two sequences were aligned using ClustalW (Thompson
et al. 1994). The NEU2 crystal structure was obtained from
the RCSB (PDB ID: 1VCU). Chain A, water and nonprotein
residues were removed using pymol, and the homology model
was generated using Modeller 9v6 (Lambert et al. 2002).
DANA and GM3 were docked in the active site using Autodock
1
0% sodium dodecyl sulfate–polyacrylamide gel electropho-
resis. Proteins were transferred to nitrocellulose membrane
using electroblotting (30 V, overnight). The membrane was
blocked with 2.5% bovine serum albumin (BSA) in Tris-
buffered saline Tween-20 buffer, and proteins were detected
with anti-His-tag antibody (HisProbe-HRP, Thermo Scientif-
ic). Blots were visualized with an HRP chemiluminescent kit
4.0 with a grid box of 60 × 40 × 40 Å (Goodsell et al. 1996).
(
Pierce). For MBP-NEU3 fusion proteins, blotting was per-
formed with an anti-MBP antibody (primary: anti-MBP,
:20,000; secondary: goat-anti-mouse IgG-HRP, 1:100,000;
Supplementary data
1
New England Biolabs).
Supplementary data for this article is available online at http://
glycob.oxfordjournals.org/.
Screening of mutant enzyme activity
Stored pellets of cells (OD600 of 3.5) expressing NEU3
Funding
(
pET28a) were lysed and the protein extracted by adding
B-PER reagent (300 µL) (Thermo Scientific, Waltham, MA).
The samples were vortexed for 1 min and incubated on ice
This work was supported by the Alberta Ingenuity Centre for
Carbohydrate Sciences (AICCS) and the Natural Sciences and
1136

Insight into substrate recognition and catalysis by human NEU3
Goodsell DS, Morris GM, Olson AJ. 1996. Automated docking of flexible li-
gands: Applications of AutoDock. J Mol Recognit. 9:1–5.
Ha KT, Lee YC, Cho SH, Kim JK, Kim CH. 2004. Molecular characterization
of membrane type and ganglioside-specific sialidase (Neu3) expressed in
E-coli. Mol Cells. 17:267–273.
Engineering Research Council of Canada (NSERC). Infra-
structure support was provided by the Canadian Foundation
for Innovation, AICCS and NSERC. M.D.L. was supported
by an NSERC USRA award.
Hata K, Koseki K, Yamaguchi K, Moriya S, Suzuki Y, Yingsakmongkon S,
Hirai G, Sodeoka M, Von Itzstein M, Miyagi T. 2008. Limited inhibitory
effects of oseltamivir and zanamivir on human sialidases. Antimicrob Agents
Chemother. 52:3484–3491.
Kalka D, von Reitzenstein C, Kopitz J, Cantz M. 2001. The plasma membrane
ganglioside sialidase cofractionates with markers of lipid rafts. Biochem Bio-
phys Res Commun. 283:989–993.
Acknowledgements
We would like to thank Dr. S. Marcus for helpful discussions.
Plasmids containing human Neu3 were a generous gift of Prof.
T. Miyagi (Miyagi Cancer Center Research Institute).
Kopitz J, Oehler C, Cantz M. 2001. Desialylation of extracellular GD1a-
neoganglioprotein suggests cell surface orientation of the plasma mem-
brane-bound ganglioside sialidase activity in human neuroblastoma cells.
FEBS Lett. 491:233–236.
Lambert C, Leonard N, De Bolle X, Depiereux E. 2002. ESyPred3D: Predic-
tion of proteins 3D structures. Bioinformatics. 18:1250–1256.
Li SC, Li YT, Moriya S, Miyagi T. 2001. Degradation of G(M1) and G(M2) by
mammalian sialidases. Biochem J. 360:233–237.
Conflict of interest statement
None declared.
Abbreviations
BSA, bovine serum albumin; DANA, 2-deoxy-2,3-dehydro-N-
acetylneuraminic acid; EDTA, ethylenediaminetetraacetic acid;
IPTG, Isopropyl β-D-1-thiogalactopyranoside; LB, lysogeny
broth; MBP, maltose-binding protein; MOPS, morpholinopro-
pane sulfonic acid; 4MU-NA, 2′-(4-methylumbelliferyl)-α-D-
N-acetylneuraminic acid; NEU, neuraminidase; NMR, nuclear
magnetic resonance; PCR, polymerase chain reaction; RMS,
root mean square; SDM, site-directed mutagenesis.
Magesh S, Moriya S, Suzuki T, Miyagi T, Ishida H, Kiso M. 2008. Design,
synthesis, and biological evaluation of human sialidase inhibitors. Part 1:
Selective inhibitors of lysosomal sialidase (NEU1). Bioorg Med Chem Lett.
1
8:532–537.
Magesh S, Suzuki T, Miyagi T, Ishida H, Kiso M. 2006. Homology modeling
of human sialidase enzymes NEU1, NEU3 and NEU4 based on the crystal
structure of NEU2: Hints for the design of selective NEU3 inhibitors. J Mol
Graph Model. 25:196–207.
Miyagi T, Wada T, Iwamatsu A, Hata K, Yoshikawa Y, Tokuyama S, Sawada
M. 1999. Molecular cloning and characterization of a plasma membrane-
associated sialidase specific for gangliosides. J Biol Chem. 274:5004–5011.
Miyagi T, Wada T, Yamaguchi K. 2008. Roles of plasma membrane-associated
sialidase NEU3 in human cancers. Biochim Biophys Acta. 1780:532–537.
Monti E, Bassi MT, Bresciani R, Civini S, Croci GL, Papini N, Riboni M,
Zanchetti G, Ballabio A, Preti A, et al. 2004. Molecular cloning and char-
acterization of NEU4, the fourth member of the human sialidase gene
family. Genomics. 83:445–453.
Monti E, Bassi MT, Papini N, Riboni M, Manzoni M, Venerando B, Croci G,
Preti A, Ballabio A, Tettamanti G, et al. 2000. Identification and expression
of NEU3, a novel human sialidase associated to the plasma membrane. Bio-
chem J. 349:343–351.
Monti E, Preti A, Rossi E, Ballabio A, Borsani G. 1999. Cloning and charac-
terization of NEU2, a human gene homologous to rodent soluble sialidases.
Genomics. 57:137–143.
References
Albouz-Abo S, Turton R, Wilson JC, von Itzstein M. 2005. An investigation of
the activity of recombinant rat skeletal muscle cytosolic sialidase. FEBS
Lett. 579:1034–1038.
Amaya MF, Buschiazzo A, Nguyen T, Alzari PM. 2003. The high reso-
lution structures of free and inhibitor-bound Trypanosoma rangeli
sialidase and its comparison with T-cruzi trans-sialidase. J Mol Biol.
325:773–784.
Amaya MF, Watts AG, Damager I, Wehenkel A, Nguyen T, Buschiazzo A,
Paris G, Frasch AC, Withers SG, Alzari PM. 2004. Structural insights into
the catalytic mechanism of Trypanosoma cruzi trans-sialidase. Structure.
12:775–784.
Bernardi A, Potenza D, Capelli AM, Garcia-Herrero A, Canada FJ, Jime-
nez-Barbero J. 2002. Second-generation mimics of ganglioside GM1
oligosaccharide: A three-dimensional view of their interactions with bac-
terial enterotoxins by NMR and computational methods. Chem Eur J.
Morris GM, Goodsell DS, Halliday RS, Huey R, Hart WE, Belew RK,
Olson AJ. 1998. Automated docking using a Lamarckian genetic algo-
rithm and an empirical binding free energy function. J Comput Chem.
19:1639–1662.
Morris GM, Goodsell DS, Huey R, Olson AJ. 1996. Distributed automated
docking of flexible ligands to proteins: Parallel applications of AutoDock
2.4. J Comput-Aided Mol Des. 10:293–304.
8:4598–4612.
Buschiazzo A, Alzari PM. 2008. Structural insights into sialic acid enzymology.
Curr Opin Chem Biol. 12:565–572.
Cairo CW, Key JA, Sadek CM. 2010. Fluorescent small-molecule probes of
biochemistry at the plasma membrane. Curr Opin Chem Biol. 14:57–63.
Cantarel BL, Coutinho PM, Rancurel C, Bernard T, Lombard V, Henrissat B.
Newstead S, Watson JN, Knoll TL, Bennet AJ, Taylor G. 2005. Structure and
mechanism of action of an inverting mutant sialidase. Biochemistry.
44:9117–9122.
2009. The Carbohydrate-Active EnZymes database (CAZy): An expert re-
Newstead SL, Potter JA, Wilson JC, Xu GG, Chien CH, Watts AG, Withers
SG, Taylor GL. 2008. The structure of Clostridium perfringens NanI siali-
dase and its catalytic intermediates. J Biol Chem. 283:9080–9088.
Papini N, Anastasia L, Tringali C, Croci G, Bresciani R, Yamaguchi K,
Miyagi T, Preti A, Prinetti A, Prioni S, et al. 2004. The plasma mem-
brane-associated sialidase MmNEU3 modifies the ganglioside pattern of
adjacent cells supporting its involvement in cell-to-cell interactions. J Biol
Chem. 279:16989–16995.
Pshezhetsky AV, Richard C, Michaud L, Igdoura S, Wang SP, Elsliger MA, Qu
JY, Leclerc D, Gravel R, Dallaire L, et al. 1997. Cloning, expression and
chromosomal mapping of human lysosomal sialidase and characterization
of mutations in sialidosis. Nat Genet. 15:316–320.
source for Glycogenomics. Nucleic Acids Res. 37:D233–D238.
Chavas LMG, Tringali C, Fusi P, Venerando B, Tettamanti G, Kato R, Monti E,
Wakatsuki S. 2005. Crystal structure of the human cytosolic sialidase Neu2 -
Evidence for the dynamic nature of substrate recognition. J Biol Chem.
280:469–475.
Chien CH, Shann YJ, Sheu SY. 1996. Site-directed mutations of the catalytic
and conserved amino acids of the neuraminidase gene, nanH, of Clostridium
perfringens ATCC 10543. Enzyme Microb Technol. 19:267–276.
Crennell S, Garman E, Laver G, Vimr E, Taylor G. 1994. Crystal-structure of
vibrio-cholerae neuraminidase reveals dual lectin-like domains in addition to
the catalytic domain. Structure. 2:535–544.
Demir O, Roitberg AE. 2009. Modulation of catalytic function by differential
plasticity of the active site: Case study of Trypanosoma cruzi trans-sialidase
and Trypanosoma rangeli sialidase. Biochemistry. 48:3398–3406.
Edidin M. 2003. The state of lipid rafts: From model membranes to cells. Annu
Rev Biophys Biomol Struct. 32:257–283.
Gestwicki JE, Strong LE, Borchardt SL, Cairo CW, Schnoes AM, Kiessling
LL. 2001. Designed potent multivalent chemoattractants for Escherichia
coli. Bioorg Med Chem. 9:2387–2393.
Seyrantepe V, Landry K, Trudel S, Hassan JA, Morales CR, Pshezhetsky
AV. 2004. Neu4, a novel human lysosomal lumen sialidase, confers nor-
mal phenotype to sialidosis and galactosialidosis cells. J Biol Chem.
279:37021–37029.
Thompson JD, Higgins DG, Gibson TJ. 1994. CLUSTAL W: Improving the
sensitivity of progressive multiple sequence alignment through sequence
weighting, position-specific gap penalties and weight matrix choice. Nucleic
Acids Res. 22:4673–4680.
1137

A Albohy et al.
von Itzstein M. 2007. The war against influenza: Discovery and
development of sialidase inhibitors. Nat Rev Drug Discov.
Watson JN, Newstead S, Narine AA, Taylor G, Bennet AJ. 2005. Two nucle-
ophilic mutants of the Micromonospora viridifaciens sialidase operate with
retention of configuration by two different mechanisms. Chembiochem.
6:1999–2004.
Watts AG, Damager I, Amaya ML, Buschiazzo A, Alzari P, Frasch AC, Withers
SG. 2003. Trypanosoma cruzi trans-sialidase operates through a covalent
sialyl-enzyme intermediate: Tyrosine is the catalytic nucleophile. J Am Chem
Soc. 125:7532–7533.
Watts AG, Oppezzo P, Withers SG, Alzari PM, Buschiazzo A. 2006. Structural
and kinetic analysis of two covalent sialosyl-enzyme intermediates on Try-
panosoma rangeli sialidase. J Biol Chem. 281:4149–4155.
Yanagisawa M, Ariga T, Yu RK. 2006. Cholera toxin B subunit binding does
not correlate with GM1 expression: A study using mouse embryonic neural
precursor cells. Glycobiology. 16:19G–22G.
6:967–974.
Wada T, Yoshikawa Y, Tokuyama S, Kuwabara M, Akita H, Miyagi T. 1999.
Cloning, expression, and chromosomal mapping of a human ganglioside
sialidase. Biochem Biophys Res Commun. 261:21–27.
Wang P, Zhang J, Bian H, Wu P, Kuvelkar R, Kung TT, Crawley Y,
Egan RW, Billah MM. 2004. Induction of lysosomal and plasma mem-
brane-bound sialidases in human T-cells via T-cell receptor. Biochem J.
380:425–433.
Wang Y, Yamaguchi K, Shimada Y, Zhao XJ, Miyagi T. 2001. Site-directed
mutagenesis of human membrane-associated ganglioside sialidase - Identi-
fication of amino-acid residues contributing to substrate specificity. Eur J
Biochem. 268:2201–2208.
Wang Y, Yamaguchi K, Wada T, Hata K, Zhao XJ, Fujimoto T,
Miyagi T. 2002. A close association of the ganglioside-specific sia-
Zanchetti G, Colombi P, Manzoni M, Anastasia L, Caimi L, Borsani G, Vener-
ando B, Tettamanti G, Preti A, Monti E, et al. 2007. Sialidase NEU3 is a
peripheral membrane protein localized on the cell surface and in endosomal
structures. Biochem J. 408:211–219.
lidase Neu3 with caveolin in membrane microdomains.
J Biol
Chem. 277:26252–26259.
Watson JN, Dookhun V, Borgford TJ, Bennet AJ. 2003. Mutagenesis of the
conserved active-site tyrosine changes a retaining sialidase into an inverting
sialidase. Biochemistry. 42:12682–12690.
Zhang J, Yu KQ, Zhu WL, Jiang HL. 2006. Neuraminidase pharmacophore
model derived from diverse classes of inhibitors. Bioorg Med Chem Lett.
16:3009–3014.
1138

Copyright of Glycobiology is the property of Oxford University Press / UK and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.