Structural basis of the inhibition of GH1 β-glucosidases by multivalent pyrrolidine iminosugars

Article abstract of DOI:10.1016/j.bioorg.2019.103026

The synthesis of multivalent pyrrolidine iminosugars via CuAAC click reaction between different pyrrolidine-azide derivatives and tri- or hexavalent alkynyl scaffolds is reported. The new multimeric compounds, together with the monomeric reference, were evaluated as inhibitors against two homologous GH1 β-glucosidases (BglA and BglB from Paenibacillus polymyxa). The multivalent inhibitors containing an aromatic moiety in the linker between the pyrrolidine and the scaffold inhibited the octameric BglA (μM range) but did not show affinity against the monomeric BglB, despite the similarity between the active site of both enzymes. A modest multivalent effect (rp/n = 12) was detected for the hexavalent inhibitor 12. Structural analysis of the complexes between the monomeric and the trimeric iminosugar inhibitors (4 and 10) and BglA showed the insertion of the inhibitors at the active site of BglA, confirming a competitive mode of inhibition as indicated by enzyme kinetics. Additionally, structural comparison of the BglA/4 complex with the reported BglB/2F-glucose complex illustrates the key determinants responsible for the inhibitory effect and explains the reasons of the inhibition of BglA and the no inhibition of BglB. Potential inhibition of other β-glucosidases with therapeutic relevance is discussed under the light of these observations.

Full text of DOI:10.1016/j.bioorg.2019.103026

Accepted Manuscript
Structural basis of the inhibition of GH1 β-glucosidases by multivalent pyrro-
lidine iminosugars
Macarena Martínez-Bailén, Elena Jiménez-Ortega, Ana T. Carmona,
Inmaculada Robina, Julia Sanz-Aparicio, David Talens-Perales, Julio Polaina,
Camilla Matassini, Francesca Cardona, Antonio J. Moreno-Vargas
PII:
S0045-2068(19)30648-0
https://doi.org/10.1016/j.bioorg.2019.103026
103026
DOI:
Article Number:
Reference:
YBIOO 103026
Bioorganic Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
2 May 2019
24 May 2019
30 May 2019
Please cite this article as: M. Martínez-Bailén, E. Jiménez-Ortega, A.T. Carmona, I. Robina, J. Sanz-Aparicio, D.
Talens-Perales, J. Polaina, C. Matassini, F. Cardona, A.J. Moreno-Vargas, Structural basis of the inhibition of GH1
β-glucosidases by multivalent pyrrolidine iminosugars, Bioorganic Chemistry (2019), doi: https://doi.org/10.1016/
j.bioorg.2019.103026
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Structural basis of the inhibition of GH1 β-glucosidases by multivalent pyrrolidine
iminosugars
Macarena Martínez-Bailén,a Elena Jiménez-Ortega,b Ana T. Carmona,a Inmaculada Robina,a
Julia Sanz-Aparicio,b,* David Talens-Perales,c Julio Polaina,c,* Camilla Matassini,d Francesca
Cardonad and Antonio J. Moreno-Vargasa,*
aDepartment of Organic Chemistry, Faculty of Chemistry, University of Seville, C/ Prof. García
González, 1, 41012-Seville, Spain.
bDepartment of Crystallography and Structural Biology, Institute of Physical-Chemistry
Rocasolano, CSIC, Serrano 119, 28006 Madrid, Spain.
cInstitute of Agricultural Chemistry and Food Technology, CSIC, 46980-Paterna, Valencia.
Spain.
dDepartment of Chemistry ‘Ugo Schiff’, University of Firenze, via della Lastruccia 3-13, 50019
Sesto Fiorentino (FI), Italy.
Abstract
The synthesis of multivalent pyrrolidine iminosugars via CuAAC click reaction between
different pyrrolidine-azide derivatives and tri- or hexavalent alkynyl scaffolds is reported. The
new multimeric compounds, together with the monomeric reference, were evaluated as
inhibitors against two homologous GH1 β-glucosidases (BglA and BglB from Paenibacillus
polymyxa). The multivalent inhibitors containing an aromatic moiety in the linker between the
pyrrolidine and the scaffold inhibited the octameric BglA (µM range) but did not show affinity
against the monomeric BglB, despite the similarity between the active site of both enzymes. A
modest multivalent effect (rp/n = 12) was detected for the hexavalent inhibitor. Structural
analysis of the complexes between the monomeric and the trimeric iminosugar inhibitors (4 and
10) and BglA showed the insertion of the inhibitors at the active site of BglA, confirming a
competitive mode of inhibition as indicated by enzyme kinetics. Additionally, structural
comparison of the BglA/4 complex with the reported BglB/2F-glucose complex illustrates the
key determinants responsible for the inhibitory effect and explains the reasons of the inhibition
of BglA and the no inhibition of BglB. Potential inhibition of other -glucosidases with
therapeutic relevance is discussed under the light of these observations.
Keywords: pyrrolidines, iminosugars, multivalency, β-glucosidase inhibitors, GH1
glycosidases, klotho proteins

1. Introduction.
β-Glucosidases are ubiquitous enzymes that play multiple biological roles and are present in all
kind of living organisms. In bacteria and fungi, β-glucosidases are involved in the breakdown of
cellulose and other carbohydrates [1]. In plants, these enzymes are involved in processes, such
as cell wall remodeling, chemical defense against pathogens and release of aroma components,
among others [2]. In humans, β-glucosidases carry out the hydrolysis of glucosylceramide
(GlcCer), which is in part performed by β-glucocerebrosidases GBA1 [3] and GBA2 [4]. A
decreased activity of GBA1 is responsible for Gaucher disease [5], the most prevalent
lysosomal storage disorder. Although GBA1 and GBA2 belong to the GH30 and GH116
families in the CAZy database [6], respectively, a large group of β-glucosidases belongs to the
GH1 family. Members of this family are the well-studied sweet almond β-glucosidase [7,8] and
the Klotho transmembrane proteins (α-Klotho, β-Klotho, and KL-related protein, KLrP).
Deficiency of α-and β-Klotho proteins is a determinant factor in human aging [9,10,11].
Besides, it has been suggested that KLrP (formerly known as GBA3 or cytosolic β-
glucocerebrosidase) is involved in the metabolic regulation of GlcCer through a novel
catabolism pathway, although its exact function remains controversial [12]. Intestinal lactase
phlorizin hydrolase (LPH), whose deficiency causes lactose intolerance [13] also belongs to
GH1 family [14]. Bacterial enzymes are in general easy to manipulate and may serve as models
to analyze structural problems related to homologous human proteins. In this respect, the two
bacterial GH1 β-glucosidases A and B (BglA and BglB), isolated from the soil and plant-
associated bacterium Paenibacillus polymyxa and involved in the hydrolysis of cellulosic
substrates, could be useful models for GH1 enzymes. BglA is a cellobiase with an octameric
quaternary structure [15], whereas BglB is a monomeric enzyme that acts as an exo-β-
glucosidase hydrolyzing cellobiose and cellodextrins with higher degree of polymerization [16].
The interaction of multimeric inhibitors and glycosidases has been mostly explored in the case
of Jack bean α-mannosidase as this enzyme has proven to be the most sensitive to multivalent
binding [17,18,19,20,21,22]. Apart from Jack bean α-mannosidase, N-acetylgalactosamine-6-
sulfatase (GALNS), a lysosomal enzyme deputed to the hydrolysis of glycosaminoglycans
(GAGs) in the lysosomes, is likely to accept multivalent ligands [23,24]. Interestingly, both Jack
bean α-mannosidase [25] and GALNS [26] have a dimeric structure. Although the mechanisms
behind the observed multivalent effect have been studied with different indirect techniques, the
interaction at the atomic level remains almost unexplored [27,28,29]. The greatest part of
multivalent glycosidase inhibitors reported in the literature belongs to the six-membered
iminosugar family (polyhydroxylated piperidines) [17-22], with only few examples of
multivalent five-membered iminosugar (polyhydroxylated pyrrolidines) derivatives [29,30].

The elucidation of the chemical basis accounting for the interaction of inhibitors with GH1
glycosidases of therapeutic interest is the key issue to design selective inhibitors that could be
useful to understand the exact function of these enzymes in vivo. Additionally, the potential
application of the inhibitors as pharmacological chaperones (PCs) [31] to treat diseases caused
by conformational deficiency of these enzymes could be of interest. PCs are molecules that bind
with high specificity to a misfolded protein in the endoplasmic reticulum (ER), promoting the
correct folding of the mutant protein. It has been observed that enzyme inhibitors can act as PCs
at sub-inhibitory concentrations, with a correlation between the binding affinity and the
chaperoning potency [32]. Regarding glycosidases, the chaperoning effect has been detected not
only with small inhibitors but also with multivalent assemblies [33,34]. To achieve this goal,
precise, atomic scale identification of the residues involved in protein-inhibitor interaction is
required. Undoubtedly, X-ray crystallography analysis represents a unique tool for this purpose.
Therefore, the objective of this work was to analyze the inhibition of GH1 β-glucosidases by
mono- and multivalent pyrrolidine iminosugars (Figure 1) by X-ray crystallography studies to
gain detailed insight of the binding at the atomic level. The influence of structural parameters,
such as the nature of the linker and valency of the multivalent compounds in the inhibition of
model enzymes BglA and BglB has been analyzed. We chose these proteins as working models
for GH1 glycosidases for three main reasons. Firstly, both enzymes have similar tertiary
structure but a very different quaternary structure, BglA is an octamer whereas BglB is a
monomer, that could be a determining factor in binding mono- or multivalent inhibitors.
Secondly, BglA and BglB are suitable materials for crystallographic analysis at high resolution,
as we have previously demonstrated [15,16]. Thirdly, our multivalent compounds are based on
monomeric iminosugars that had previously shown inhibition towards other GH1 β-
glucosidases (from almonds) [35]. In this manuscript we report the synthesis of new multivalent
pyrrolidine derivatives (tri- and hexavalent), their inhibitory activity towards BglA and BglB
and a precise, atomic resolution, description of the interactions responsible for the inhibition
through X-ray crystallography.
O
O
H2N
O
H
N
O
O
N
H
O
O
O
O
O
O
H
N
Linker
N
N
2
N
HO
OH

Figure 1. Multivalent pyrrolidine iminosugars.
2. Results and discussions.
2. 1. Synthesis of the monovalent and multivalent inhibitors
For the synthesis of the multivalent compounds we used the Cu(I)-catalyzed azide-alkyne
cycloaddition reaction (CuAAC) between tri- and hexavalent alkynyl scaffolds and pyrrolidine
azides 3 and 6, which were prepared from azidomethyl pyrrolidine 1, in turn recently reported
by our group [35]. CuAAC of 1 with aromatic alkyne 2 [36] (see Supporting Information) and
subsequent treatment with NaN3 afforded in excellent yield protected azide 3, which after acidic
deprotection gave unprotected derivative 4 (Scheme 1). Pyrrolidine azide 6 was prepared after
CuAAC of 1 with alkyne 5 (see Supporting Information), nucleophilic displacement with NaN3
and acidic deprotection.
N
1'
R1
N
N
4''
3
R
1 = Boc, R2 = -C(CH3)2
4 (·HCl)
N
Boc
OTs
, CuI
1 = H, R2 = H
HCl, THF,
100%
2'''
R
2
1.
O
1'''
N
2
2
5''
3'''
4'''
3
5
DIPEA, Toluene, 50 oC, 95%
2
4
N3
OR2
2IV
OR2
6IV
2. NaN3, DMF, 60 oC, 98%
O
O
O
O
3IV
5IV
N3
1
5
H
OTs
1.
, CuI
1'
N
1''
1
DIPEA, Toluene, 91%
N
2
5
N
4
3
N
4''
2. NaN3, DMF, 60 oC, 93%
3. HCl (1M):THF, 83%
5''
HO
OH
2'''
1'''
6
N3
3'''
Scheme 1. Synthesis of the monovalent azidomethyl pyrrolidines.
The CuAAC reaction of azides 3 and 6 with the scaffolds 7 [37] and 8 (see Supporting
Information for the synthesis of the alkynes) using CuSO4 and sodium ascorbate in THF-H2O
and microwave irradiation at 80 °C afforded multivalent compounds 9-12 (Scheme 2). Due to
the poor solubility of unprotected derivative 4 under CuAAC conditions, the click reactions
were performed with the protected azido pyrrolidine 3. The completion of the reaction was
1
checked by H-NMR spectra of the crude mixtures, showing the disappearance of the signal at
2.30-2.50 ppm assigned to the terminal alkyne protons.

O
O
O
O
O
H2N
O
O
H
O
N
N
O
O
H
O
8
7
3
6
,
1. Azide or
CuSO4, NaAsc, THF:H2O 2:1,
MW, 80 o
C
2. HCl (4M), THF
3
)
(After CuAAC with
HO
HO
HO
N
HO
H2N
NH
O
N
N
NH
N
H
N
X
N
N
N
X
O
O
X
N
N
OH
N
HN
N
HO
HO
OH
X
N
NH
X
HO
HO
N
O
X
N
N
O
O
OH
N
N
X
H
NH
O
N
OH
N
N
N
H
NH
O
N
N
O
N
O
HO
O
HO
N
N
N
N
X
NH
N
N
H
N
N
N
X
9
,
(57%) X =
N
HO
N
N
HO
OH
11
(92%), X =
HO
N
N
N
N
10 (·HCl)
,
(85%) X =
N
N
12 (·HCl)
(93%), X =
O
O
2
2
Scheme 2. Synthesis of multivalent compounds.
2.2.
Biological evaluation
The inhibitory activities of compounds 9-12 towards BglA and BglB glucosidases were
evaluated. Compounds 9 and 11 did not inhibit any of the enzymes while derivatives 10 and 12
showed competitive inhibition towards BglA in the micromolar range and no inhibition of BglB
(Table 1), which is quite surprising considering the homology of both enzymes. These results
point out the importance of the aromatic moiety of compounds 10 and 12 in the inhibition of
BglA. In order to evaluate the existence of a multivalent effect, the valence-corrected relative
inhibition potency (rp/n) was determined (rp = Ki (monov)/Ki (multiv), n = valency of the cluster). A
positive multivalent effect is observed when rp/n >1 [19]. Pyrrolidine azide 4, which was used
as monovalent reference compound, showed a moderate competitive inhibition of BglA (Ki = 51
µM). Trivalent compound 10 and hexavalent 12 showed rp/n values of 2 and 12, respectively.

Table 1. IC50 (µM), Ki (µM) and rp/n for compounds 4, 9-12.
Compound
(Valency, n)
BglA
(rp/n)
BglB
Structure
9
N
-
-
-
-
N
N
HN
(n = 3)
11
(n = 6)
N
N
N
O
OH
OH
n
IC50 = 12
Ki = 8.5a
(rp/n = 2.0)b
IC50 = 0.54
Ki = 0.72a
(rp/n = 12)b
10
(n = 3)
-
-
N
N
N
HN
N
N
O
OH
N
O
OH
n
12
(n = 6)
N
N
N
4
IC50 = 54
Ki = 51a
HN
-
OH
O
O
OH
N3
-: no inhibition was detected at 0.1 mM of compound.
a Competitive inhibition was observed by double Cornish-Bowden plots (See Supporting Information).
b rp = Ki (4)/Ki (multivalent).
2.3.
Structural insights into pyrrolidine iminosugar inhibition of BglA.
The crystal structures of the BglA complexes with the inhibitors 4 and 10 have been solved to
2.1 and 2.8 Å resolution, respectively. Unfortunately, determination of co-crystal structures for
12 was unsuccessful. In both complexes, the asymmetric unit (a.u.) contains two subunits, A
and B, of the functional octamer that are essentially identical and, therefore, the discussion will
be focused on molecule A. The electron density obtained from both complexes, shown in
Figure 2 allows unambiguous positioning of the inhibitors at the active site of BglA, with the
pyrrolidine ring interacting with the catalytic residues, which confirms the competitive
inhibition mode of the enzyme, as previously outlined by enzyme kinetics data. The only
portion of the ligand presenting poor density is the distal part of the extended chain, i.e. the
azidomethylene group of 4, as this flexible segment lacks direct atomic interactions with
residues at the active site that would stabilize a fixed conformation.

a
b
Figure 2. Cartoon representation of BglA showing the molecules bound at its active site. Crystals soaked
into compound 4 (a) and 10 (b). The catalytic residues are highlighted as blue sticks and the 2Fo-Fc
electron density is contoured at 1.0 σ cut-off.
The precise recognition of 4 by BglA is given in Figure 3a. The pyrrolidine ring occupies
subsite -1 with its two hydroxyl groups mimicking the observed positions of O3 and O4 from
the covalently bound glucose in the reported complex BglA-glucose (PDBV code 1E4I). Thus,
the same net of hydrogen links is established with Gln20, Glu405 and Trp406. Also, the
endocyclic NH is hydrogen bonded to the carboxylates from Glu166/Glu352 and, altogether,
this fixes the pyrrolidine in a tight position, which superimposes with the anomeric carbon of
the glucose, as it has been observed previously in others GH1 complexes [38]. Moreover, the
triazole, the phenyl ring and the C-C-O segment are accommodated through strong hydrophobic
interaction with the residues shaping the narrow slit that gives access to the catalytic site. Thus,
Trp326 in a T-mode interaction on one side, and Leu173, Leu177, Val179 and His180, on the
opposite side, constrain an almost planar conformation of the ligand (see Figure 3b). The fact
that the tested ligands lacking the phenyl ring failed to inhibit BglA can be explained in terms of
the higher ability of this group to adapt to the narrow slot, which confers a superior affinity of
the ligands containing the phenyl moiety. On the contrary, the distal azidomethylene group
protrudes from the cleft and is located in the channel that connects different subunits of the
oligomer. Thus, as it can be seen in Figures 3c and 3d, the BglA oligomer assembles the eight
subunits with their corresponding active site faced in pairs, so that a 16-20 Å wide slit gives
access to both cavities. All the subunits are occupied by an inhibitor molecule and, in this way,
the distance between the terminal nitrogen atoms from each two bound ligand molecules is 12 Å
(Figure 3d).

a
b
c
d
Figure 3. Detail of the BglA active site in the crystals soaked into compound 4. (a) Atomic interactions
between the inhibitor (chartreuse) and relevant BglA residues (yellow) represented as sticks; the catalytic
residues are highlighted in magenta and hydrogen bonds are represented with dashed lines. (b) Surface
representation of BglA showing the narrow slot giving access to the catalytic pocket that is filled by the
inhibitor; The hydrophobic residues shaping the slot are labelled. (c) and (d) Two perpendicular views of
the BglA octamer, showing symmetry-related molecules A in beige and those corresponding to molecules
B in magenta. In (d), the molecular surface of two subunits with faced active sites is represented,
illustrating the relative position of each pair of bound inhibitor molecules existing in the octamer.
In the solved structure of the BglA complexed with 10 (See Supplementary material), the
absence of clear electron density, which vanishes from the ethylene glycol segment, as observed
in Figure 2b, precludes determining if the pyrrolidine moieties observed in the eight subunits
come from the same or from two different trimer molecules (Figure S4 of Supplementary
material). The premise of a same trimeric ligand entering the two confronted active sites is a

feasible situation considering the topology of the octamer. However, also considering the
flexibility of the ligand in the bridge segment and the absence of atomic interactions observed at
this region with the enzyme, none of the alternative situations can be excluded. Consequently,
the molecular mechanism of the multivalent effect remains elusive. The X-ray structure of
BglA/4 and BglA/10 complexes showed a similar pattern of interaction with the enzyme and, in
both cases, the inhibitor binds the active site of the enzyme. This fact discards the occlusion of
this cavity by compound 10 through unspecific interactions outside the active site (Figure 4c),
which could be in agreement with the competitive inhibition found experimentally.
On the other hand, the multivalent effect observed for the hexamer is modest (rp/n = 12). This
correlates much better with a statistical effect than with a chelate effect (Figures 4a vs 4b) [39].
Nevertheless, the simultaneous interaction of trimer and hexamer with two or more enzymes
(Figure 4d) might not be discarded.
Figure 4. Selected multivalent effects. a) Statistical rebinding: the high local concentration of the inhitope
may favour a recapture mechanism; b) Chelate effect: simultaneous interaction of the inhibitor with two
active sites of the enzyme; c) Occlusion of active site: interactions of the inhibitor with non-catalytic
subsites that block the active pocket; d) Clustering effect: binding of the inhibitor to more than one
enzyme at the same time.
Molecular rationale of the inhibitory activity of pyrrolidine iminosugars vs GH1 enzymes
As it has been commented above, none of the tested compounds present inhibitory activity
against BglB, despite this latter has appreciable similarities at the active site with respect to
Bg1A. This observation can be rationalized on the basis of the corresponding three-dimensional
structures. A structural superimposition of the BglA/4 complex with the reported BglB/2F-
glucose complex (PDB code 2JIE) [16] shows that the residues in both catalytic sites are well
conserved (Figure 5a). In this way, the phenyl moiety of the extrapolated position for a ligand
molecule within the BglB active site could be allocated in the hydrophobic slit delineated by
Trp328, in one side, and Tyr169, Leu174 and His181 in the opposite wall. However, some
differences appear from distal positions that are more evident at the outer entrance to the cleft

(Figure 5b, 5c). Thus, the BglA residues Leu177, Val179, Ile324 and Glu408 are substituted by
Thr178, Glu180, Met326 and Trp412 in BglB. In particular, and most remarkably, Glu180
protrudes at the BglB cleft probably producing deleterious clashes with the ligand, which may
preclude the binding due to steric hindrance.
a
b
c
Figure 5. Comparison of BglA and BglB active sites. (a) A stereo pair of the structural superimposition
of the BglA/4 (yellow) and BglB/2F-glucose (brown, PDB code 2JIE) complexes, showing the catalytic
residues of BglA in orange and the ligand in chartreuse. The position of the ligand has been inferred in
the BglB active site and the complex has been submitted to simple molecular dynamics. (b) A slice of the
molecular surface of BglA with negative regions colored in red and relevant residues as sticks. (c) Same
cross-section of the molecular surface of BglB. Glu180 protrudes into the catalytic channel and probably
prevents binding of 4 to BglB.

The same analysis may be performed to predict potential inhibition of other GH1 enzymes by
these compounds. Thus, we have analyzed a possible effect on KLrP, a member of the Klotho
family of proteins of known three-dimensional structure [40,41]. The structural superimposition
of the KLrP/Glc complex (PDB code 2E9L) onto the BglA/4 complex is shown in Figure 6. As
expected, the residues involved in recognition of the -D-glucopyranose are conserved with the
two described -glucosidases, BglA and BglB, while the sequence divergence increases from
the aglycon binding site. However, the orientation and the chemical nature of the BglA key
residues Leu173, Val179, H180 and Glu408 are similar to the KLrP residues Met172, Met178,
Phe179 and Gln427. Consequently, the phenyl ring of the inhibitor could be well allocated
within a hydrophobic slit, stacked to Trp345 side-chain, and delineated by Met172, Leu176,
Met178 and Phe179, therefore keeping the same interaction mode observed in the BglA/4 and
BglA/10 complexes. This suggests that KLrP would also be inhibited. As the role of KLrP in
the GlcCer metabolism in vivo is not defined yet [12], the development of selective inhibitors of
this enzyme is important to elucidate the exact functions of the β-glucosidases involved in the
metabolism of GlcCer and other glycosphingolipids.
Figure 6. Comparison of BglA and KLrP active sites. A stereo pair of the structural superimposition of
the BglA/4 (yellow) and KLrP/glucose (slate, PDB code 2E9L) complexes, showing the catalytic residues
of BglA in orange and the ligand in chartreuse. The position of the ligand has been inferred in the KLrP
active site and the complex has been submitted to simple molecular dynamics.
3. Conclusions
In summary, we have reported one of the very few examples of multivalent pyrrolidine
derivatives with affinity against glycosidases, in this case a GH1 β-glucosidase. The multivalent
compounds containing a phenyl moiety in the spacer between the pyrrolidine and the central

scaffold selectively inhibited octameric GH1 β-glucosidase A from P. polymyxa (BglA), while
no inhibition was detected against the homologous monomeric BglB. The structural analysis of
the mono/trivalent inhibitors (4 and 10) in complex with BglA showed the pyrrolidine ring
interacting with the catalytic residues in the active site, fixing the pyrrolidine in a tight position,
which confirms the competitive inhibition mode of the enzyme. This analysis was crucial to
explain both the exceptional selectivity of our inhibitors for BglA over homologous BglB and
the requirement of a phenyl moiety in the arm spacer to achieve such inhibition. Although much
more research is needed to understand the multivalent mechanisms operating in glycosidase
inhibition, the structural analysis of the β-glucosidase inhibition by pyrrolidine iminosugars
reported here will aid in the rational design of more potent and selective inhibitors of other GH1
glycosidases of therapeutic interest as KLrP.
4. Materials and methods
4.1. Chemistry
General methods: Optical rotations were measured in a 1.0 cm or 1.0 dm tube with a Jasco P-
2000 spectropolarimeter. Infrared spectra were recorded with
a
Jasco FTIR-410
1
spectrophotometer. H and 13C NMR spectra were recorded with a Bruker Gemini 200,
AMX300 and INOVA400 for solutions in CDCl3, DMSO-d6, CD3OD and D2O. δ are given in
ppm and J in Hz. J are assigned and not repeated. All the assignments were confirmed by
COSY and HSQC experiments. High resolution mass spectra were recorded on a Q-Exactive
spectrometer. TLC was performed on silica gel 60 F254 (Merck), with detection by UV light
charring with p-anisaldehyde, KMnO4, ninhydrin or with Pancaldi reagent [(NH4)6MoO4,
Ce(SO4)2, H2SO4, H2O]. Silica gel 60 (Merck, 40-60 and 63-200 μm) was used for preparative
chromatography. Small scale microwave assisted syntheses were carried out in a CEM Discover
or Anton Paar Monowave 300 microwave apparatus.
(2S,3S,4R)-N-tert-Butoxycarbonyl-2-[(4-(4-(2-(2-azidoethoxy)ethoxy)phenyl)-1H-1,2,3-
triazol-1-yl)methyl]-3,4-O-isopropylidene-pyrrolidine-3,4-diol (3): To a solution of 1 [35]
(925 mg, 3.10 mmol) in toluene (28 mL), alkyne 2 [36] (1.68 g, 4.66 mmol), DIPEA (2.1 mL,
o
12 mmol) and CuI (181 mg, 0.931 mmol) were added and the mixture was stirred at 50 C for
30 h. Then, EtOAc and sat. aq. soln. of NaHCO3 were added and the phases were separated. The
aqueous phase was extracted with EtOAc (x2) and the combined organic layers were dried over
Na2SO4, filtered and evaporated.
Chromatographic purification
on silica gel
→
(EtOAc:cyclohexane 1:2 2:1) afforded the corresponding click derivative (1.93 g, 2.93 mmol,
95%) as a white solid. To a solution of this compound (1.93 g, 2.93 mmol) in DMF (25 mL),
o
NaN3 (482 mg, 7.34 mmol) was added. After stirring at 60 C for 3 h the solvent was removed

under vacuum and the crude product was dissolved in CH2Cl2, washed with water and brine,
dried over Na2SO4, filtered and evaporated. The resulting residue was purified by
→
chromatography column on silica gel (EtOAc:cyclohexane 1:1 2:1) to yield 3 (1.52 g, 2.87
[훼]29
mmol, 98%) as a grey solid.
+ 26.8 (c 0.85, CH2Cl2). IR (ν cm-1) 2981, 2935, 2100 (N3),
퐷
1688 (C=O), 1498, 1402, 1246, 1123, 1056, 909, 729. 1H-NMR (300 MHz, DMSO-d6, 363 K, δ
ppm, J Hz) δ 8.29 (s, 1H, H-5’’), 7.77-7.72 (m, 2H, H-aromat.), 7.05-7.01 (m, 2H, H-aromat.),
4.73-4.67 (m, 2H, H-3, H-4), 4.59 (dd, 1H, 2J1’a,1’b = 13.8, J1’a,2 = 6.1, H-1’a), 4.44 (dd, 1H, J1’b,2
= 7.0, H-1’b), 4.36-4.32 (m, 1H, H-2), 4.18 (t, 2H, JH,H = 4.8, H-2IV), 3.82 (t, 2H, H-3IV), 3.73-
3.69 (m, 3H, H-5a, H-5IV), 3.43 (t, 2H, JH,H = 5.0, H-6IV), 3.20 (dd, 1H, 2J5b,5a = 12.9, J5b,4 = 4.0,
H-5b), 1.34 (s, 3H, -C(CH3)2), 1.31 (s, 9H, -C(CH3)3), 1.26 (s, 3H, -C(CH3)2). 13C-NMR (75.4
MHz, DMSO-d6, 363 K, δ ppm) δ 158.0 (Cq aromat.), 153.1 (C=O, Boc), 146.0 (C-4’’), 126.2
(C aromat.), 123.4 (Cq aromat.), 120.6 (C-5’’), 114.7 (C aromat.), 110.6 (-C(CH3)2), 81.7 (C-3),
78.7 (-C(CH3)3), 78.4 (C-4), 68.9 (C-5IV), 68.6 (C-3IV), 67.2 (C-2IV), 63.1 (C-2), 50.4 (C-5), 49.9
(C-6IV), 48.6 (C-1’), 27.4 (-C(CH3)3), 26.3 (-C(CH3)2), 24.5 (-C(CH3)2). HRESIMS m/z found
530.2715, calc. for C25H36N7O6 [M+H]+: 530.2722.
(2S,3S,4R)-2-[(4-(4-(2-(2-Azidoethoxy)ethoxy)phenyl)-1H-1,2,3-triazol-1-yl)methyl]-
pyrrolidine-3,4-diol hydrochloride (4): A solution of 3 (55 mg, 0.10 mmol) in HCl (1M):THF
1:1 (3.8 mL), was stirred at r.t. overnight. Evaporation of the solvent afforded 4 (47 mg, 0.10
[훼]25
mmol, quantitative) as a white solid.
– 42.7 (c 0.54, MeOH). IR (ν cm-1) 3323, 3099 (OH,
퐷
1
NH), 2911, 2100 (N3), 1617, 1499, 1248, 1130, 1053, 828. H-NMR (300 MHz, CD3OD, δ
ppm, J Hz) δ 8.65 (s, 1H, H-5’’), 7.77 (d, 2H, JH,H = 8.7, H-aromat.), 7.08 (d, 2H, H-aromat.),
5.12-5.01 (m, 2H, H-1’), 4.32-4.30 (m, 1H, H-4), 4.25-4.18 (m, 3H, H-3, H-2IV), 4.13-4.06 (m,
2
1H, H-2), 3.89-3.86 (m, 2H, H-3IV), 3.75 (t, 2H, JH,H = 4.9, H-5IV), 3.62 (dd, 1H, J5a,5b = 12.6,
J5a,4 = 3.5, H-5a), 3.41-3.32 (m, 3H, H-6IV, H-5b). 13C-NMR (75.4 MHz, CD3OD, δ ppm) δ
161.4 (Cq aromat.), 147.7 (C-4’’), 128.7 (C aromat.), 124.1 (C-5’’), 121.6 (Cq aromat.), 116.4
(C aromat.), 74.8 (C-3), 71.3 (C-5IV), 70.6 (C-3IV), 70.3 (C-4), 68.9 (C-2IV), 60.7 (C-2), 51.8 (C-
6IV), 51.7 (C-5), 51.3 (C-1’). HRESIMS m/z found 390.1876, calc. for C17H24N7O4 [M]+:
390.1884.
(2S,3S,4R)-2-[(4-(3-Azidopropyl)-1H-1,2,3-triazol-1-yl)methyl]-pyrrolidine-3,4-diol (6): To
a solution of 1 [35] (1.30 g, 4.36 mmol) in toluene (40 mL), alkyne 5 (1.55 g, 6.50 mmol),
DIPEA (2.9 mL, 17 mmol) and CuI (253 mg, 1.30 mmol) were added and the mixture was
stirred at r.t. for 1 d. Then, EtOAc and sat. aq. soln. of NaHCO3 were added and the phases were
separated. The aqueous phase was extracted with EtOAc (x2) and the combined organic layers
were dried over Na2SO4, filtered and evaporated. Chromatographic purification on silica gel
→
(EtOAc:cyclohexane 1:5 2:1) afforded the corresponding click derivative (2.12 g, 3.95 mmol,

91%) as a colourless oil. To a solution of this compound (1.99 g, 3.71 mmol) in DMF (32 mL),
o
NaN3 (607 mg, 9.24 mmol) was added. After stirring at 60 C for 3 h the solvent was removed
under vacuum and the crude product was dissolved in CH2Cl2, washed with water and brine,
dried over Na2SO4, filtered and evaporated. The resulting residue was purified by
→
chromatography column on silica gel (EtOAc:cyclohexane 1:1 2:1) to give the corresponding
pyrrolidine-(azido)triazole (1.40 g, 3.44 mmol, 93%) as a yellow oil. A solution of this
compound (492 mg, 1.21 mmol) in HCl (1M):THF 1:1 (35 mL), was stirred at r.t. overnight.
Evaporation of the solvent and chromatographic purification on Dowex 50WX8 eluting with
[훼]29
MeOH, H2O and NH4OH 17%, afforded 6 (267 mg, 1.00 mmol, 83%) as a white solid.
–
퐷
45.2 (c 0.84, MeOH). IR (ν cm-1) 3427, 3285 (OH, NH), 2097 (N3), 1550, 1346, 1209, 979. 1H-
NMR (300 MHz, CD3OD, δ ppm, J Hz) δ 7.82 (s, 1H, H-5’’), 4.60 (dd, 1H, 2J1’a,1’b = 14.0, J1’a,2
= 4.1, H-1’a), 4.38 (dd, 1H, J1’b,2 = 8.1, H-1’b), 4.02 (td, 1H, J4,3 = J4,5a = 4.8, J4,5b = 3.0, H-4),
3.73 (dd, 1H, J3,2 = 7.7, H-3), 3.47-3.41 (m, 1H, H-2), 3.36 (t, 2H, JH,H = 6.7, H-3’’’), 3.12 (dd,
1H, 2J5a,5b = 12.0, H-5a), 2.87-2.76 (m, 3H, H-5b, H-1’’’), 1.99-1.89 (m, 2H, H-2’’’). 13C-NMR
(75.4 MHz, CD3OD, δ ppm) δ 147.8 (C-4’’), 124.2 (C-5’’), 76.2 (C-3), 72.3 (C-4), 62.7 (C-2),
53.9 (C-1’), 52.3 (C-5), 51.6 (C-3’’’), 29.6 (C-2’’’), 23.4 (C-1’’’). HRESIMS m/z found
268.1518, calc. for C10H18N7O2 [M+H]+: 268.1516.
General procedure for the synthesis of multivalent compounds: To a solution of the protected
or unprotected pyrrolidin-azide (3-6 eq) in THF:H2O 2:1 (3 mL), were added CuSO4 (0.3 eq per
click reaction), sodium ascorbate (0.6 eq per click reaction) and the polypropargylated scaffold
(1 eq). The reaction mixture was heated in a MW reactor (80 oC, 0.75-3.0 h). After evaporation
of the solvent, the resulting crude was purified by stirring with Quadrasil® MP followed by
chromatography column (silica gel or Sephadex LH-20).
Polyhydroxylated trivalent iminosugar 9: CuAAC (General procedure) using 6 (54 mg, 0.20
mmol) and tripropargylated scaffold 7 [37] (16 mg, 0.068 mmol), afforded after 45 min of
reaction and chromatographic purification (Sephadex LH-20, H2O), compound 9 (40 mg, 0.039
[훼]29
mmol, 57%) as a pale yellow oil.
– 11.5 (c 0.67, H2O). IR (ν cm-1) 3269, 3138 (OH, NH),
퐷
2929, 1602, 1424, 1222, 1054, 813. 1H-NMR (400 MHz, D2O, δ ppm, J Hz) δ 7.90-7.88 (m, 3H,
H-5IV), 7.69 (s, 3H, H-5’’), 4.53 (dd, 3H, 2J1’a,1’b = 14.2, J1’a,2 = 4.6, H-1’a), 4.48 (s, 6H, -OCH2-
triazole), 4.38-4.33 (m, 9H, H-1’b, H-3’’’), 4.10-4.07 (m, 3H, H-4), 3.84-3.79 (m, 3H, H-3),
2
3.36 (br. s, 9H, H-2, H2NCCH2O), 3.10 (dd, 3H, J5a,5b = 12.6, J5a,4 = 4.9, H-5a), 2.76 (dd, 3H,
J5b,4 = 2.5, H-5b), 2.59 (t, 6H, JH,H = 6.8, H-1’’’), 2.17 (quint, 6H, JH,H = 6.7, H-2’’’). 13C-NMR
(50 MHz, D2O, δ ppm) δ 146.6 (C-4’’), 144.1 (C-4IV), 124.9 (C-5IV), 123.7 (C-5’’), 74.6 (C-3),
70.9 (C-4), 70.5 (H2NCCH2O), 63.6 (-OCH2-triazole), 60.7 (C-2), 52.6 (C-1’), 50.3 (C-5), 50.0

(H2NCCH2O), 49.6 (C-3’’’), 28.8 (C-2’’’), 21.6 (C-1’’’). HRESIMS m/z found 1059.5419, calc.
for C43H68N22O9Na [M+Na]+: 1059.5432.
Polyhydroxylated trivalent iminosugar 10·HCl: CuAAC (General procedure) using 3 (139
mg, 0.263 mmol) and tripropargylated scaffold 7 [37] (21 mg, 0.089 mmol), afforded after 1.5 h
→
of reaction and chromatographic purification (silica gel, EtOAc:MeOH:NH4OH 7:1:0.1
5:1:0.1), the corresponding protected trivalent iminosugar (138 mg, 0.0757 mmol, 85%) as a
white solid. A solution of this compound (57 mg, 0.031 mmol) in HCl (4M):THF (2:1, 4.5 mL)
was stirred at r.t. for 4 h and evaporated to give 10·HCl (54 mg, 0.031 mmol, quantitative) as a
[훼]24
white solid.
– 29.3 (c 0.70, H2O). IR (ν cm-1) 3344, 3140, (OH, NH), 2925, 1616, 1511,
퐷
1254, 1111, 1053, 925, 838. 1H-NMR (300 MHz, D2O, δ ppm, J Hz) δ 8.15 (s, 3H, H-5’’), 7.87
(s, 3H, H-5V), 7.39 (d, 6H, JH,H = 8.5, H-aromat.), 6.64 (d, 6H, H-aromat.), 4.84-4.81 (m, 6H, H-
1’), 4.47-4.44 (m, 6H, H-6IV), 4.34-4.30 (m, 9H, H-4, -OCH2-triazole), 4.22 (dd, 3H, J3,2 = 9.6,
J3,4 = 4.0, H-3), 4.03-3.95 (m, 3H, H-2), 3.84-3.78 (m, 12H, H-5IV, H-2IV), 3.61-3.54 (m, 9H, H-
3IV, H-5a), 3.37-3.33 (m, 9H, H-5b, H2NCCH2O). 13C-NMR (75.4 MHz, D2O, δ ppm) δ 158.4
(Cq aromat.), 146.6 (C-4’’), 142.6 (C-4V), 127.1 (C aromat.), 125.6 (C-5V), 122.2 (C-5’’), 121.4
(Cq aromat.), 114.9 (C aromat.), 72.8 (C-3), 68.7 (C-4, C-3IV), 68.4 (C-5IV), 67.0 (H2NCCH2O),
67.0 (C-2IV), 63.1 (-OCH2-triazole), 59.0 (H2NCCH2O), 58.9 (C-2), 50.5 (C-6IV), 50.1 (C-5),
49.1 (C-1’). HRESIMS m/z found 1403.6698, calc. for C64H87N22O15 [M+H]+: 1403.6716.
Polyhydroxylated hexavalent iminosugar 11: CuAAC (General procedure) using 6 (61 mg,
0.23 mmol) and hexapropargylated scaffold 8 (22 mg, 0.038 mmol), afforded after 2 h and 15
min of reaction and chromatographic purification (Sephadex LH-20, H2O), compound 11 (76
[훼]29
mg, 0.035 mmol, 92%) as a pale yellow oil.
– 19.4 (c 0.54, H2O). IR (ν cm-1) 3276, 3132
퐷
(OH, NH), 2929, 1649 (C=O), 1550, 1436, 1342, 1216, 1049, 823. 1H-NMR (400 MHz, D2O, δ
ppm, J Hz) δ 7.84 (s, 6H, H-5IV), 7.69 (s, 6H, H-5’’), 4.52 (dd, 6H, 2J1’a,1’b = 13.9, J1’a,2 = 3.7, H-
1’a), 4.43 (s, 12H, -OCH2-triazole), 4.37-4.30 (m, 18H, H-1’b, H-3’’’), 4.08 (br. s, 6H, H-4),
3.84-3.78 (m, 6H, H-3), 3.60 (s, 12H, -NHCCH2O), 3.37 (br. s, 6H, H-2), 3.10-3.08 (m, 6H, H-
2
5a), 2.76 (d, 6H, J5b,5a = 11.9, H-5b), 2.56 (t, 12H, JH,H = 6.8, H-1’’’), 2.15-2.12 (m, 12H, H-
2’’’), 2.04 (br. s, 4H, -CH2CH2C=O), 1.36 (br. s, 4H, -CH2CH2C=O). 13C-NMR (50 MHz, D2O,
δ ppm) δ 176.1 (C=O), 146.6 (C-4’’), 144.1 (C-4IV), 124.8 (C-5IV), 123.6 (C-5’’), 74.6 (C-3),
70.9 (C-4), 67.6 (-NHCCH2O), 63.6 (-OCH2-triazole), 60.8 (C-2), 59.8 (-NHCCH2O), 52.6 (C-
1’), 50.3 (C-5), 49.6 (C-3’’’), 35.9 (-CH2CH2C=O), 28.9 (C-2’’’), 24.7 (-CH2CH2C=O), 21.7
(C-1’’’). HRESIMS m/z found 1092.5787, calc. for ½(C92H144N44O20) [M+2H]2+: 1092.5796.
Polyhydroxylated hexavalent iminosugar 12·HCl: CuAAC (General procedure) using 3 (113
mg, 0.213 mmol) and hexapropargylated scaffold 8 (21 mg, 0.036 mmol), afforded after 3 h of

reaction and chromatographic purification (silica gel, EtOAc:MeOH:NH4OH 5:1:0.1), the
corresponding protected hexavalent iminosugar (126 mg, 0.0335 mmol, 93%) as a colourless
oil. A solution of this compound (57 mg, 0.015 mmol) in HCl (4M):THF (2:1, 2.1 mL) was
stirred at r.t. for 4 h and evaporated to give 12·HCl (55 mg, 0.015 mmol, quantitative) as a
[훼]25
solid.
– 26.5 (c 0.53, H2O). IR (ν cm-1) 3328, 3140 (OH, NH), 2932, 1617 (C=O), 1511,
퐷
1
1257, 1186, 1106, 1053, 925, 837. H-NMR (300 MHz, D2O, δ ppm, J Hz) δ 8.07 (s, 6H, H-
5’’), 7.77 (s, 6H, H-5V), 7.29 (d, 12H, JH,H = 8.5, H-aromat.), 6.53 (d, 12H, H-aromat.), 4.79-
4.75 (m, 12H, H-1’), 4.35-4.29 (m, 18H, H-6IV, H-4), 4.20-4.16 (m, 18H, -OCH2-triazole, H-3),
3.98-3.91 (m, 6H, H-2), 3.70-3.69 (m, 24H, H-5IV, H-2IV), 3.56-3.48 (m, 18H, H-5a, H-3IV),
3.38-3.29 (m, 18H, -NHCCH2O, H-5b), 1.85 (br. s, 4H, -CH2CH2C=O), 1.17 (br. s, 4H, -
CH2CH2C=O). 13C-NMR (75.4 MHz, D2O, δ ppm) δ 175.8 (C=O), 158.3 (Cq aromat.), 146.4
(C-4’’), 142.9 (C-4V), 126.9 (C aromat.), 125.5 (C-5V), 122.1 (C-5’’), 121.2 (Cq aromat.), 114.8
(C aromat.), 72.8 (C-3), 68.6 (C-4, C-3IV), 68.4 (C-5IV), 67.8 (-NHCCH2O), 66.9 (C-2IV), 62.9 (-
OCH2-triazole), 59.5 (-NHCCH2O), 58.9 (C-2), 50.5 (C-6IV), 50.1 (C-5), 49.1 (C-1’), 35.7 (-
CH2CH2C=O), 24.5 (-CH2CH2C=O). HRESIMS m/z found 1469.6808, calc. for
½(C134H179N44O32Na) [M+H+Na]2+: 1469.6809.
4.2.
Production and purification of BglA and BglB
His-tagged versions Panibacillus polymyxa -glucosidases A and B with increased thermal
stability were cloned in plasmid pQE80-L vector (Stratagene). The enzymes were recovered
from transformant E. coli cultures and purified by nickel affinity chromatography. The
engineered versions of the two enzymes that were used contained the following mutations:
BglA: E96K, T3855A, N411S, M416I and N437K [42], BglB: H62R, N223Y and M319I [43].
4.3.
Inhibition studies with BglA and BglB.
The % of inhibition towards the corresponding glycosidase (BglA or BglB) was determined in
the presence of 0.1 mM of the inhibitor on the well. Each enzymatic assay (final volume 0.12
mL) contains 0.01-0.5 units/mL of the enzyme and 4.2 mM aqueous solution of the appropriate
p-nitrophenyl β-D-glucopyranoside (substrate) buffered to the optimal pH of the enzyme (pH
6.5). Enzyme and inhibitor were pre-incubated for 5 min at rt, and the reaction started by
addition of the substrate. After 20 min of incubation at 37 °C, the reaction was stopped by
addition of 0.1 mL of sodium borate solution (pH 9.8). The p-nitrophenolate formed was
measured by visible absorption spectroscopy at 405 nm. Under these conditions, the p-
nitrophenolate released led to optical densities linear with both reaction time and concentration
of the enzyme. The IC50 value (concentration of inhibitor required for 50% inhibition of enzyme
activity) was determined from plots of % inhibition versus different inhibitor concentrations.
For all the inhibitors, the mode of inhibition was suggested by the Dixon plot where reciprocal

velocity (1/V) is plotted against concentration of inhibitor [I] at different substrate
concentrations [S]. The intersection of lines gives Ki. The complementary Cornish-Bowden plot
([S]/V versus [I] at different substrate concentrations) was also represented to confirm the
competitive inhibition mode [44]. The IC50 and Ki were determined by duplicate (<10% of
difference) and the average value is given.
4.4. Crystallization and structure determination of the BglA complexes.
The first attempts to obtain crystals from the reported crystallization conditions for wild type
BglA were unproductive [14]. New crystallization conditions were explored by high-throughput
techniques with a NanoDrop robot (Innovadyne Tecnhologies Inc.), using two commercial
screens: Index (Hampton Research) and JBScreen PACT++ (Jena Bioscience). Preliminary bar-
shape crystals were obtained in different conditions containing PEG3350, with pH ranging from
6.5 to 7.5. However, X-ray data collection from these crystals showed very low resolution and
triclinic symmetry with a high-content asymmetric-unit (a.u.). Additional screening of additives
identified two non-ionic detergent (n-Dodecyl -D-maltoside and 6-Cyclohexylhexyl -D-
maltoside) that were essential to grow good-quality prismatic crystals with tetragonal symmetry
and two molecules within the a.u. Crystals were grown by mixing protein solution (6 mgml−1 in
50 mM potassium phosphate, pH 7), precipitant solution (13-18% (v/v) PEG 3350,
0.2M sodium nitrate, 3% 2-Methyl-2,4-pentanediol, 0.1 M BisTris propane, pH 7.5), and 5% n-
Dodecyl
-D-maltoside/6-Cyclohexylhexyl
-D-maltoside,
in
a
1:1:0.1
protein:precipitant:detergent ratio. The complex with compound 4 was obtained by soaking
crystals into precipitant solution containing 33 mM ligand for 4 hours. The complex with 10
was obtained by fast soaking into precipitant solution with 10 mM ligand.
For data collection, crystals were transferred to cryoprotectant solutions consisting of mother
liquor plus 25% glicerol and 5-10 mM ligand, being flash-cooled in liquid nitrogen. Diffraction
data were collected using synchrotron radiation on the XALOC beamline at ALBA (Cerdanyola
del Vallés, Spain). Diffraction images were processed with XDS [45] and merged using
AIMLESS [46] from the CCP4 package [47]. They were indexed in the P4212 space group,
with two molecules in the a.u. and 66% solvent content within the unit cell. Final data collection
statistics are given in Table S1 (Supplementary material).
The X-ray structure of the complexes were solved by molecular replacement using MOLREP
[48] and the coordinates of native BglA (PDB code1E4I). Crystallographic refinement was
performed using the programs REFMAC [49] and PHENIX [50] with flat bulk-solvent
correction and local non-crystallographic symmetry (NCS). Model-building was performed with
COOT [51]. For the inhibitors 4 and 10, not present in the Protein Data Bank, a PDB model was
built by MacPyMOLX11Hybrid (The PyMOL Molecular Graphics System, Version 2.0
Schrödinger, LLC). The model was used to automatically generate coordinates and molecular

topologies with eLBOW [52] suitable for REFMAC refinement. The figures were generated
with PyMOL.
4.5. Molecular dynamics
The molecular dynamic calculations were performed using the simple molecular dynamics
module of Phenix to shake up the complex model, with the ligand position inferred from the
structural superimposition of BglB (PDB code 2JIE) and KLrP (PDB code 2E9L) coordinates,
respectively, onto the BglA/4 complex. A very short molecular dynamic at 300 K was
performed. The RMSD between the BglB and KLrP X-ray structures (referred to the protein
backbone) and their corresponding refined complexes were 0.431 and 0.425 Å, respectively.
5. Notes
The authors declare no competing financial interest.
Acknowledgments
This work was supported by the Spanish Ministry of Economy and Competitiveness (CTQ2016-
77270-R, BIO2016-76601-C3-3-R, AGL2016-75245-R), the Junta de Andalucía (FQM-345),
MIUR-Italy (“Progetto Dipartimenti di Eccellenza 2018-2022” allocated to Department of
Chemistry “Ugo Schiff”) and Ente Cassa di Risparmio di Firenze (grant n°. 2016/0845). M.
Martínez-Bailén acknowledges the Spanish government for a FPU fellowship. We thank
CITIUS-University of Seville (MS and NMR facilities) and the Synchrotron Radiation Source
at Alba (Barcelona, Spain) for assistance at BL13-XALOC beamline.
References
[1] L.R. Lynd, P.J. Weimer, W.H. van Zyl, I.S. Pretorius, Microbial cellulose utilization: fundamentals
and biotechnology, Microbiol. Mol. Biol. Rev. 66 (2002) 506-577.
[2] J.R. Ketudat Cairns, B. Mahong, S. Baiya, J.-S. Jeon, β-Glucosidases: multitasking, moonlighting or
simply misunderstood?, Plant Science 241 (2015) 246-259.
[3] A.H. Futerman, F.M. Platt, The metabolism of glucocerebrosides-From 1965 to the present, Mol.
Genet. Metab. 120 (2017) 22-26.
[4] M. Aureli, M. Samarani, N. Loberto, G. Mancini, V. Murdica, E. Chiricozzi, A. Prinetti, R. Bassi, S.
Sonnino, Current and novel aspects on the non-lysosomal β-glucosylceramidase GBA2, Neurochem. Res.
41 (2016) 210-220.
[5] L. Smith, S. Mullin, A.H.V. Schapira, Insights into the structural biology of Gaucher disease, Exp.
Neurol. 298 (2017) 180-190.
[6] F. Ben Bdira, M. Artola M, H.S. Overkleeft, M. Ubbink, J.M.F.G. Aerts, Distinguishing the
differences in β-glycosylceramidase folds, dynamics, and actions informs therapeutic uses, J. Lipid Res.
59 (2018) 2262-2276.

[7] S. He, S.G. Withers, The almond β-glucosidase is deduced to belong to family GH1 by limited
sequence analysis, J. Biol. Chem. 272 (1997) 24864-24867.
[8] J. del Cueto, B.L. Møller, F. Dicenta, R. Sánchez-Pérez, β-Glucosidase activity in almond seeds, Plant
Physiol. Biochem. 126 (2018) 163-172.
[9] H. Kurosu, M. Yamamoto, J.D. Clark, J.V. Pastor, A. Nandi, P. Gurnani, O.P. McGuinness, H.
Chikuda, M. Yamaguchi, H. Kawaguchi, et al., Suppression of aging in mice by the hormone Klotho,
Science 309 (2005) 1829-1833.
[10] G. Chen, Y. Liu, R. Goetz, L. Fu, S. Jayaraman, M.C. Hu, O.W. Moe, G. Liang, X. Li, M.
Mohammadi, α-Klotho is a non-enzymatic molecular scaffold for FGF23 hormone signalling, Nature 553
(2018) 461-466.
[11] S. Lee, J. Choi, J. Mohanty, L.P. Sousa, F. Tome, E. Pardon, J. Steyaert, M.A. Lemmon, I. Lax, J.
Schlessinger, Structures of β-klotho reveal a ‘zip code’-like mechanism for endocrine FGF signalling,
Nature 553 (2018) 501-505.
[12] Y. Hayashi, M. Ito, Klotho-related protein KLrP: structure and functions, Vitam. Horm., 101 (2016)
1-16.
[13] M. Amiri, L. Diekmann, M. von Koeckritz-Blickwede, H.Y. Naim, The diverse forms of lactose
intolerance and the putative linkage to several cancers, Nutrients 7 (2015) 7209-7230.
[14] M. Behrendt, J. Polaina, H.Y. Naim, Structural hierarchy of regulatory elements in the folding and
transport of an intestinal multidomain protein, J. Biol. Chem. 285 (2010) 4143-4152.
[15] J. Sanz-Aparicio, J.A. Hermoso, M. Martínez-Ripoll, J.L. Lequerica, J. Polaina, Crystal structure of
β-glucosidase A from Bacillus polymyxa: insights into the catalytic activity in family 1 glycosyl
hydrolases, J. Mol. Biol. 275 (1998) 491-502.
[16] P. Isorna, J. Polaina, L. Latorre-García, F.J. Cañada, B. González, J. Sanz-Aparicio, Crystal
structures of Paenibacillus polymyxa β-glucosidase B complexes reveal the molecular basis of substrate
specificity and give new insights into the catalytic machinery of family I glycosidases, J. Mol. Biol. 371
(2007) 1204-1218.
[17] P. Compain, A. Bodlenner, The multivalent effect in glycosidase inhibition: a new, rapidly emerging
topic in glycoscience, ChemBioChem 15 (2014) 1239-1251.
[18] S.G. Gouin, Multivalent inhibitors for carbohydrate-processing enzymes: beyond the “lock-and-key”
concept, Chem. Eur. J. 20 (2014) 11616-11628.
[19] N. Kanfar, E. Bartolami, R. Zelli, A. Marra, J.-Y. Winum, P. Dumy, Emerging trends in enzyme
inhibition by multivalent nanoconstructs, Org. Biomol. Chem. 13 (2015) 9894-9906.
[20] C. Matassini, C. Parmeggiani, F. Cardona, A. Goti, Are enzymes sensitive to the multivalent effect?
Emerging evidence with glycosidases, Tetrahedron Lett. 57 (2016) 5407-5415.
[21] C. Ortiz Mellet, J.-F. Nierengarten, J.M. García Fernández, Multivalency as an action principle in
multimodal lectin recognition and glycosidase inhibition: a paradigm shift driven by carbon-based
glyconanomaterials, J. Mater. Chem. B 5 (2017) 6428-6436.
[22] P. Compain, Multivalent Effect in Glycosidase Inhibition: The End of the Beginning Chem. Rec. 19
(2019) doi.org/10.1002/tcr.201900004.
[23] G. D'Adamio, C. Matassini, C. Parmeggiani, S. Catarzi, A. Morrone, A. Goti, P. Paoli, F. Cardona,
Evidence for a multivalent effect in inhibition of sulfatases involved in lysosomal storage disorders
(LSDs)RSC Adv. 6 (2016) 64847–64851.
[24] C. Matassini, C. Vanni, A. Goti, A. Morrone, M. Marradi, F. Cardona, Multimerization of DAB-1
onto Au GNPs affords new potent and selective N-acetylgalactosamine-6-sulfatase (GALNS) inhibitors,
Org. Biomol. Chem. 16 (2018) 8604-8612.
[25] A. Kumar, S. M. Gaikwad, Jack bean α-mannosidase (Jbα-man): Tolerance to alkali, chelating and
reducing agents and energetics of catalysis and inhibition, Int. J. Biol. Macromol. 49 (2011) 1066–1071.
[26] Y. Rivera Colón, E. K. Schutsky, A. K. Zita, S. C. Garman, The structure of human GALNS reveals
the molecular basis for mucopolysaccharidosis IV AJ. Mol. Biol. 423 (2012) 736-751.
[27] E. Howard, A. Cousido-Siah, M.L. Lepage, J.P. Schneider, A. Bodlenner, A. Mitschler, A. Meli, I.
Izzo, H.A. Álvarez, A. Podjarny, et al., Structural basis of outstanding multivalent effects in Jack Bean α-
mannosidase inhibition, Angew. Chem. Int. Ed. 57 (2018) 8002-8006.
[28] A. Hottin, D.W. Wright, E. Moreno-Clavijo, A.J. Moreno-Vargas, G.J. Davies, J.-B. Behr, Exploring
the divalent effect in fucosidase inhibition with stereoisomeric pyrrolidine dimers, Org. Biomol. Chem.
14 (2016) 4718-4727.
[29] E. Moreno-Clavijo, A.T. Carmona, A.J. Moreno-Vargas, L. Molina, D.W. Wright, G.J. Davies, I.
Robina, Exploring a multivalent approach to α-L-fucosidase inhibition, Eur. J. Org. Chem. (2013) 7328-
7336.

[30] S. Mirabella, G. D’Adamio, C. Matassini, A. Goti, S. Delgado, A. Gimeno, I. Robina, A.J. Moreno-
Vargas, S. Sesták, J. Jiménez-Barbero, et al., Mechanistic insight into the binding of multivalent
pyrrolidines to α-mannosidases, Chem. Eur. J. 23 (2017) 14585-14596.
[31] Y.X. Tao, P.M. Conn, Pharmacoperones as novel therapeutics for diverse protein conformational
diseases, Physiol. Rev. 98 (2018) 697-725.
[32] J.-Q. Fan, A contradictory treatment for lysosomal storage disorders: Inhibitors enhance mutant
enzyme activity, Trends Pharmacol. Sci. 24 (2003) 355-360.
[33] C. Decroocq, D. Rodríguez-Lucena, K. Ikeda, N. Asano, P. Compain, Cyclodextrin-based
iminosugar click clusters: the first examples of multivalent pharmacological chaperones for the treatment
of lysosomal storage disorders, ChemBioChem 13 (2012) 661-664.
[34] A. Joosten, C. Decroocq, J. de Sousa, J.P. Schneider, E. Etamé, A. Bodlenner, T.D. Butters, P.
Compain, A systematic investigation of iminosugar click clusters as pharmacological chaperones for the
treatment of Gaucher disease, ChemBioChem 15 (2014) 309-319.
[35] M. Martínez-Bailén, A.T. Carmona, E. Moreno-Clavijo, I. Robina, D. Ide, A. Kato, A.J. Moreno-
Vargas, Tuning of β-glucosidase and α-galactosidase inhibition by generation and in situ screening of a
library of pyrrolidine-triazole hybrid molecules, Eur. J. Med. Chem. 138 (2017) 532-542.
[36] J. Veliks, H.M. Seifert, D.K. Frantz, J.K. Klosterman, J.-C. Tseng, A. Lindena, J.S. Siegel, Towards
the molecular Borromean link with three unequal rings: double-threaded ruthenium(II) ring-in-ring
complexes, Org. Chem. Front. 3 (2016) 667-672.
[37] Y.M. Chabre, C. Contino-Pépin, V. Placide, T.C. Shiao, R. Roy, Expeditive synthesis of
glycodendrimer scaffolds based on versatile TRIS and mannoside derivatives, J. Org. Chem. 73 (2008)
5602-5605.
[38] M. E. C. Caines, S. M. Hancock, C. A. Tarling, T. M. Wrodnigg, R. V. Stick, A. E. Sttz, A. Vasella,
S. G. Withers, N. C. J. Strynadka. The structural basis of glycosidase inhibition by five-membered
iminocyclitols: the clan A glycoside hydrolase endoglycoceramidase as a model system, Angew. Chem.
Int. Ed. 46, (2007) 4474-4476.
[39] R.J. Pieters, Maximizing multivalency effects in protein-carbohydrate interactions, Org. Biomol.
Chem. 7 (2009) 2013-2025.
[40] Y. Hayashi, N. Okino, Y. Kakuta, T. Shikanai, M. Tani, H. Narimatsu, M. Ito, Klotho-related protein
is a novel cytosolic neutral beta-glycosylceramidase, J. Biol. Chem. 282 (2007) 30889-30900.
[41] S. Tribolo, J.G. Berrin, P.A. Kroon, M. Czjzek, N. Juge, The crystal structure of human cytosolic
beta-glucosidase unravels the substrate aglycone, J. Mol. Biol. 370 (2007) 964-975.
[42] G. González-Blasco, J. Sanz-Aparicio, B. González, J.A. Hermoso, J. Polaina, Directed evolution of
beta -glucosidase A from Paenibacillus polymyxa to thermal resistance, J. Biol. Chem. 275 (2000) 13708-
13712.
[43] M.J. Arrizubieta, J. Polaina, Increased thermal resistance and modification of the catalytic properties
of a beta-glucosidase by random mutagenesis and in vitro recombination, J. Biol. Chem. 275 (2000)
28843-28848.
[44] A. Cornish-Bowden, A simple graphical method for determining the inhibition constants of mixed,
uncompetitive and non-competitive inhibitors, Biochem. J. 137 (1974) 143-144.
[45] W. Kabsch, Software XDS for image rotation, recognition and crystal symmetry assignment, Acta
Crystallogr. Sect. D 66 (2010) 125-132.
[46] P.R. Evans, G.N. Murshudov, How good are my data and what is the resolution?, Acta Crystallogr.
Sect. D 69 (2013) 1204-1214.
[47] M.D. Winn, C.C. Ballard, K.D. Cowtan, E.J. Dodson, P. Emsley, P.R. Evans, R.M. Keegan, E.B.
Krissinel, A.G.W. Leslie, A. McCoy, et al., Overview of the CCP4 suite and current developments, Acta
Crystallogr. Sect. D 67 (2011) 235-242.
[48] A. Vagin, A. Teplyakov, MOLREP: an automated program for molecular replacement, J. Appl.
Crystallogr. 30 (1997) 1022-1025.
[49] G.N. Murshudov, A. Vagin, E.J. Dodson, Refinement of macromolecular structures by the
maximum-likelihood method, Acta Crystallogr. Sect. D 53 (1997) 240-255.
[50] 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, et al., PHENIX: a comprehensive Python-based system for
macromolecular structure solution, Acta Crystallogr. Sect. D 66 (2010) 213-221.
[51] P. Emsley, K. Cowtan, Coot: model-building tools for molecular graphics, Acta Crystallogr. Sect. D
60 (2004) 2126-2132.
[52] N.W. Moriarty, R.W. Grosse-Kunstleve, P.D. Adams, electronic Ligand Builder and Optimization
Workbench (eLBOW): a tool for ligand coordinate and restraint generation, Acta Crystallog. Sect. D 65
(2009) 1074-1080.

Graphical Abstract
Crystal structure
Trimeric inhibitor/GH1 -glucosidase (P. polymyxa)

-
-
-
-
Efficient synthesis of multivalent pyrrolidine iminosugars via CuAAC
Unexpected selectivity in the inhibition of two structurally related -glucosidases
A phenyl moiety in the spacer of the iminosugar was crucial for inhibition
First co-crystal structure of a multivalent iminosugar with a GH1 -glucosidase