“This is he pee - e iewed e sion o he ollowing
a icle: Mechanis ic Insigh in o Binding o
Mul i alen Py olidines o a-Mannosidases,
which has been published in inal o m a
doi.o g/10.1002/chem.201703011. This a icle
may be used o non-comme cial pu poses in
acco dance wi h Wiley-VCH Te ms and
Condi ions o Sel -A chi ing.”
Mechanis ic Insigh in o
Binding o Mul i alen
Py olidines o -Mannosidases
S e ania Mi abella,[a,d] Giampie o D’Adamio,[a]
Camilla Ma assini,[a,g] And ea Go i,[a,g] Sand a
Delgado,[d] Ana Gimeno,[d] Inmaculada Robina,[b]
An onio J. Mo eno-Va gas,[b] Se gej Šes ák,[c]
Jesús Jiménez-Ba be o[d,e, ]* and F ancesca
Ca dona[a,g]*
Abs ac : No el py olidine-based mul i alen iminosuga s,
syn hesized by a CuAAC app oach, ha e shown ema kable
mul i alen e ec s owa ds jack bean -mannosidase and a Golgi -
mannosidase om D osophila melanogas e , oge he wi h a good
selec i i y wi h espec o a lysosomal -mannosidase, impo an o
an icance applica ions. STD-NMR and molecula modelling s udies
suppo a mul i alen mechanism wi h speci ic in e ac ions o he
bioac i e iminosuga s wi h jack bean -mannosidase. TEM s udies
sugges a binding mode in ol ing he o ma ion o agg ega es
esul ing om he in e molecula c oss-linked ne wo k o in e ac ions
be ween he mul i alen inhibi o s and wo o mo e dime s o JBMan
subuni he e odime s.
In oduc ion
Du ing he las decades, ex ensi e chemical e o s ha e been
de o ed o he design and syn hesis o mono alen inhibi o s
a ge ing ca bohyd a e-p ocessing enzymes, wi h only ew
examples in e ms o ma ke ed d ugs. Mo e ecen ly, he concep
o mul i alency has eme ged as an appealing and al e na i e
s a egy o he adi ional lock-key app oach o access new
selec i e and po en inhibi o s.1 A speci ic esponse can be
achie ed by mul iple in e ac ions o he mul i alen ligand wi h he
ecep o s, wi h consequen imp essi e inc ease in he binding
a ini y. This phenomenon, known as “ he mul i alen e ec ”,
leads o high a ini y enhancemen o a mul i alen ligand owa d
a a ge , i compa ed wi h he mono alen coun e pa .2
Mul i alency has been la gely exploi ed in he ield o lec ins,
whe e he p esence o se e al ca bohyd a e-binding si es on he
p o ein su ace has inspi ed he design o a a ie y o
glycoclus e s wi h excellen binding a ini y.3
In sha p con as , he use o mul i alen -based s a egy has been
less explo ed in glycosidase inhibi ion, due o he monome ic
na u e o many o hese ca bohyd a e-p ocessing enzymes.
Ne e heless, a ini y enhancemen s can also be expec ed o
enzymes wi h a single ac i e si e p esen a ion, since he high local
concen a ion o he ac i e moie y on he mul i alen sca old,
close o he ecogni ion si e, may a o a mechanism o ecap u e,
called “s a is ical ebinding”. Despi e o he ea lies disappoin ing
esul s in glycosidase inhibi ion by mul i alen ligands,4 he i s
example o a signi ican mul i alen e ec , obse ed o a i alen
deoxynoji imycin (DNJ) conjuga e agains he jack bean α-
mannosidase (JBMan) in 2009,5 p omp ed in es iga ions o he
design o o he mul i alen ligands wi h inc eased alency, wi h
he aim o eaching high le els o glycosidase a ini y.1,6
Among glycosidases, -mannosidases a e he mos in es iga ed
enzymes o mul ime ic inhibi ion. This has been in luenced by he
a ailabili y o comme cial JBMan; he sequence simila i ies
be ween JBMan and o he wo mannosidases o he apeu ic
in e es , i.e. Golgi -mannosidase (GMII), and lysosomal -
mannosidase (bLAM),7 all belonging o glycoside hyd olase (GH)
38 amily; and he mul ime ic na u e o JBMan ha , in p inciple,
can be a ge ed by a mul i alen ligand.8
Howe e , he o igin o he mul i alen e ec is a o be
unequi ocally elucida ed and se e al mechanisms ha e been
p oposed and can be used o explain he phenomenon (Figu e 1).
Besides he s a is ical ebinding e ec (Figu e 1a), he chela e
e ec can occu when he enzyme p esen s mo e han one ac i e
si e (Figu e 1b). In addi ion o he in e ac ions o a b anch o he
mul i alen in o he ac i e pocke , addi ional unspeci ic
in e ac ions can in ol e non-ca aly ic subsi es leading o s onge
in e ac ions (Figu e 1c). Al e na i ely, he occlusion o he ac i e
si e h ough unspeci ic in e ac ions may be plausible wi h mo e
hinde ed ligands ha a e no able o en e in o he ac i e ca i y
(Figu e 1d). Mo eo e , a clus e ing e ec can occu o la ge size
ligands, h ough he binding o mo e han one enzyme a he same
ime (Figu e 1e), o h ough he o ma ion o c oss-linked ne wo ks
i he enzyme possesses a mul ime ic na u e (Figu e 1 ).
[a] S. Mi abella, D . G. D’Adamio, D . C. Ma assini, P o . A. Go i, P o .
F. Ca dona*
Dipa imen o di Chimica “Ugo Schi ”
Uni e si à degli S udi di Fi enze
Via della Las uccia 3-13, 50019 Ses o Fio en ino (FI), I aly
E-mail: ancesca.ca dona@uni i.i
[b] D . A. J. Mo eno-Va gas, P o . I. Robina
Depa amen o de Química O gánica, Facul ad de Química,
Uni e sidad de Se illa
c/P o . Ga cía González 1, E-41012 Se illa, Spain
[c] D . S. Šes ák
Ins i u e o Chemis y, Cen e o Glycomics,
Slo ak Academy o Sciences
Dúb a ska ces a 9, 84538, B a isla a, Slo akia
[d] S. Delgado, A. Gimeno, P o . J. Jiménez-Ba be o*
CIC bioGUNE, Bizkaia Science and Technology Pa k,
Building 801A, 48160 De io, Spain
E-mail: jjba be [email protected]
[e] Ike basque, Basque Founda ion o Science, Ma ia Diaz de Ha o
5, 48005 Bilbao, Spain.
[ ] Depa amen O ganic Chemis y II, EHU-UPV, 48040 Leioa, Spain.
[g] Associa ed wi h CNR-INO, Via N. Ca a a 1, Ses o Fio en ino (FI),
I aly.
Suppo ing in o ma ion o his a icle is gi en ia a link a he end
o he documen .((Please dele e his ex i no app op ia e))
Figu e 1. P oposed binding models accoun ing o he mul i alen e ec . The
in e ac ions desc ibed may also include possible addi ional in e molecula
con ac s p o ided by he employed linke s.
The la ge and open ca aly ic si e o GH38 -mannosidases9 is
op imal o a ecap u e mechanism (Figu e 1a).
Mo eo e , JBMan is supposed o be o ganized in o a dime o
he e odime ic subuni s, o a o al weigh o a ound 220 kDa. Each
he e odime consis s o a la ges subuni (66 kDa) and a smalles
one (44 KDa), his la e ha bo ing he ac i e si e.8 Fo his eason,
JBMan may be sui able o a chela e e ec (Figu e 1b) and o he
o ma ion o agg ega es in ol ing mo e han one enzyme (Figu e
1 ). Se e al s udies ha e been ca ied ou o e alua e he
s oichiome y o he binding and o p o e he o ma ion o
agg ega es. Iso he mal Ti a ion Calo ime y (ITC) expe imen s o
se e al DNJ-based mul i alen ligands e ealed ha he alency
and he opology ha e no e ec on he en halpy and en opy
con ibu ions and complexes wi h a a io close o 1:1 we e
obse ed.10
In con as o hese esul s, A omic Fo ce Mic oscopy (AFM)
s udies o JBMan wi h DNJ mul i alen compounds based
po phy in, calix[4]a ene and cyclodex in sca olds clea ly showed
he o ma ion o nanoassemblies o di e en shape and sizes
depending on he compound.11 The o ma ion o agg ega es was
also con i med by Dynamic Ligh Sca e ing (DLS) expe imen s.
La ge agg ega es o 500 nm we e o med in he p esence o he
mul ime ic ligands, in con as wi h a hyd odynamic diame e o
JBMan alone o 10 nm.11
Mo e ecen ly, Compain and co-wo ke s p oposed a model o
binding o a 48- alen DNJ-based ligand wi h JBMan employing
h ee di e en echniques: analy ical ul acen i uga ion,
T ansmission Elec on Mic oscopy (TEM) and ESI-MS. The
esul s suppo ed he o ma ion o agg ega es ligand/JBMan wi h
a chela ion o he ligand o ou ac i e si es (one o each
he e odime ).12
We epo he ein ou con ibu ion o he ield by means o a
mul idisciplina y in eg a ed app oach ha akes ad an age o he
o al syn hesis o new mul i alen py olidine based inhibi o s,
hei biological e alua ion owa ds di e en mannosidases
including human enzymes, TEM echniques and unp eceden ed
NMR and molecula dynamic s udies.
Resul s and Discussion
Syn hesis
Mos o he mul i alen iminosuga s in es iga ed as JBMan
inhibi o s and epo ed o da e bea DNJ o deoxymannoji imycin
(DMJ) as he bioac i e iminosuga moie ies. While he in luence
o di e en sca olds, opology and alency on JBMan inhibi ion
has been widely in es iga ed, he e ec s o di e en ly s uc u ed
iminosuga s a e almos unexplo ed.13
Exploi ing he expe ise o some o us in he o al syn hesis o
iminosuga -based inhibi o s,14 we decided o explo e di e en
mul i alen a chi ec u es based on he na u al occu ing
py olidine iminosuga 1,4-dideoxy-1,4-imino-D-a abini ol (DAB-1
(2), Scheme 1). DAB-1 is known o be an ac i e compound wi h a
b oad inhibi o y spec um owa ds mammalian glycosidases.15,16
Scheme 1. Syn hesis o DAB-1 (2), he nona alen compound 6 and he
mono alen e e ence compound 5.
We ecen ly epo ed he use o DAB-1 in he cons uc ion o he
nona alen 6 (Scheme 1), which showed an imp essi e ac i i y
owa ds N-ace ylgalac osamine-6-sul a ase (GALNS), he
enzyme de icien in he mucopolysaccha idosis Mo quio A
synd ome.17
Taking in o conside a ion ha GALNS (like JBMan) has a dime ic
na u e, as e ealed by i s X- ay s uc u e,18 we hypo hesized ha
JBMan migh also well accep mul i alen inhibi o s based on
DAB-1.
Nona alen compound 6 was syn hesized om D-a abinose
de i ed ni one 119 (Scheme 1), as p e iously epo ed.17
We syn hesized he mono alen compound 5 as he e e ence
compound, bea ing he same alkyl chain o a single b anch o he
mul i alen ligands, o e alua e he enzyme a ini y enhancemen
o he mul i alen compounds. In his case, we sligh ly modi ied
he published p o ocol, pe o ming he coppe ca alyzed azide
alkyne cycloaddi ion (CuAAC)20 wi h 3-bu yn-1-ol di ec ly on he
dep o ec ed azide 4 ins ead o using he benzyla ed de i a i e 8
as desc ibed in e .17. The CuAAC eac ion, pe o med wi h
CuSO4 (0.3 equi .)/sodium asco ba e (0.6 equi .), using a
THF/H2O 2:1 mix u e as he sol en in a MW eac o a 80 °C o
45 minu es, a o ded he mono alen ligand 5 in 89% yield
(Scheme 1). This new s a egy imp o ed he o e all syn hesis o
compound 5 om ni one 1 (82%), and diminished he numbe o
s eps om i e o h ee.
Scheme 2. Syn hesis o he benzyla ed e a alen compound 10.
Ini ial a emp s o syn hesize e a alen DAB-1 based inhibi o
conside ed he benzyla ed iminosuga 10, expec ed o be easie
o handle and pu i y han he co esponding dep o ec ed
de i a i e. Thus, he benzyla ed de i a i e 8 was ob ained as
desc ibed in e . 17 by ni ogen alkyla ion o amine 717,21 wi h 1-
azido-6-b omohexane (3) (Scheme 2). The CuAAC eac ion o he
benzyla ed key azido in e media e 8 (4 equi .) wi h he e a alen
sca old 922 was pe o med wi h CuSO4 (0.3 equi ./sodium
asco ba e (0.6 equi .), using a THF/H2O 2:1 mix u e as he
sol en in a MW eac o a 80 °C o 45 minu es. The eac ion
ga e a e y clean c ude mix u e, om which he e a alen
py olidine 10 was isola ed in 96% yield a e lash column
ch oma og aphy (FCC), as epo ed in Scheme 2. Un o una ely,
dep o ec ion o he pe benzyla ed adduc 10 was no a i ial ask,
due o he high basici y and hyd ophilici y o he dep o ec ed
compound. Ca aly ic hyd ogena ion o 10 in acidic MeOH wi h
Pd/C ga e a mix u e o hyd ochlo ide sal s ha was passed on o
an ion exchange esin Dowex 50WX8-200 elu ing successi ely
wi h MeOH, H2O and 6% aqueous ammonia. Howe e , he
compound was elu ed in he i s ac ion wi h MeOH as a
hyd ochlo ide sal and no as he ee amine as expec ed. Taking
in o accoun he p oblems in he dep o ec ion s ep also
expe ienced in he case o mul i alen py olizidine compounds13b
and in o de o sho en he syn he ic ou e a oiding he
debenzyla ion/ace yla ion/dep o ec ion eac ions, we decided o
ca y ou he CuAAC eac ion on he dep o ec ed azide 4. Thus,
he syn hesis was pe o med as in he case o he nona alen
ligand 6 (Scheme 1). Reac ion o sca old 9 wi h 4 (4 equi .) in he
p esence o CuSO4 (0.3 equi )/sodium asco ba e (0.6 equi .)
using a THF/H2O 2:1 mix u e as he sol en unde MW i adia ion
a 80 °C o 45 minu es a o ded he e a alen iminosuga 11 in
76% yield (Scheme 3), ha was pu i ied by FCC and size
exclusion ch oma og aphy.
Scheme 3. Syn hesis o he e a alen compound 11 and o i alen
compound 13.
Analogously, compound 4 (3 equi .) was eac ed wi h he i alen
sca old is[(p opa gyloxy)me hyl]amino-me hane (12),
syn hesized by a h ee s ep p ocedu e s a ing om is
(hyd oxyme hyl)aminome hane.23 The CuAAC eac ion o 12 wi h
4 (3 equi .) was pe o med as p e iously desc ibed o he
e a alen ligand 10, a o ding he i alen iminosuga 13 in 48%
yield (Scheme 3), a e pu i ica ion by FCC and size exclusion
ch oma og aphy.
Biological ac i i y e alua ion
The mul i alen py olidine iminosuga s 6, 11 and 13, oge he
wi h DAB-1 (2) and he mono alen py olidine 5, we e e alua ed
owa d a panel o comme cial glycosidases24 and he esul s a e
epo ed in Table 1. I is wo h no ing ha mos o he mul i alen
py olidines showed good inhibi o y ac i i y owa ds α-
glucosidases and especially amyloglucosidase in he low µM
ange, bu o hese enzymes no mul i alen e ec was highligh ed,
since hei inhibi ion ac i i y is compa able o ha o he
mono alen e e ence compound 5. Mo e in e es ingly,
ema kable esul s we e shown by he mul i alen py olidine
iminosuga s owa d JBMan. JBMan is able o hyd olyze he
e minal α-1,2, α-1,3, α-1,6 linked mannose esidues om a ious
glycop o eins.8 JBMan is a e aining enzyme and like o he class
II α-mannosidases, i is inhibi ed by he iminosuga swainsonine.25
Signi ican a ini y enhancemen was obse ed owa ds JBMan
passing om he mono alen compound 5 o highe alencies,
wi h bes esul s ob ained o he nona alen compound 6.
Table 1. Inhibi o y ac i i y o py olidine iminosuga s 2, 5, 13, 11 and 6 owa ds comme cial glycosidases. Pe cen ages o inhibi ion a 1 mM o inhibi o (IC50 in
pa en heses [μM]) a e epo ed.
Glycosidases
DAB-1 (2)
Mono alen 5
T i alen 13
Te a alen 11
Nona alen 6
α-L- ucosidase
bo ine kidney
-
-
-
-
72
α-galac osidase
co ee beans
-
-
61
55
75
α-glucosidase
Saccha omyces ce e isiae
ice
99 (1.2)
85 (208)
83 (153)
24
94(37)
72
84 (198)
72
72
59
Amyloglucosidase
Aspe gillus nige
96 (16)
>90 (8.1)
98 (1.8)
>90 (1.5)
>90 (1.1)
β- glucosidase
almonds
45
-
-
-
-
- mannosidase
jack bean
74 (376)
43 (1300)
>90 (9.4)
>90 (34)
>90 (0.095)
- mannosidase
snail
-
-
-
-
-
β-N-ace ylglucosaminidase
jack bean
-
-
-
-
-
- no inhibi ion was de ec ed a 1 mM concen a ion o he compound.
To quan i y he mul i alen e ec owa d JBMan, he ela i e
inhibi o y po encies ( p, gi en by he a io o he IC50 alues o he
mono alen py olidine 5 and ha o he mul i alen compounds)
and he a ini y enhancemen pe DAB mo i ( p/n) we e calcula ed
and a e epo ed in Table 2. Fo all py olidine iminosuga s 6, 11,
and 13, a la ge mul i alen e ec was obse ed ( p/n >>1). The
i alen compound 13 and he e a alen 11 showed a good
mul i alen e ec on JBMan, wi h a p/n o 46 and 10. The bes
esul was shown by he nona alen compound 6, wi h an
imp essi e p/n o 1520. This alue is ema kable conside ing he
da a in he li e a u e o compounds wi h simila alency, such as
he p e iously epo ed dodeca alen DNJ-based ulle ene ( p/n =
179).26
Cu iously, he i alen 13 is mo e ac i e han 11 and his could
be due o a highe simila i y o 13 o nona alen compound 6 in
e ms o spa ial dis ibu ion o he bioac i e moie ies.
Due o he e y p omising esul s owa ds JBMan, he mul i alen
py olidine iminosuga s 13, 11 and 6, as well as he mono alen
compound 5, we e also es ed agains he biological ele an
class II Golgi and lysosomal α-mannosidases (GMIIb and LManII)
om D osophila melanogas e 7 as a model o a ge enzymes in
an icance he apy and lysosomal s o age diso de s ea men ,
espec i ely. Indeed, educ ion o he umo g ow h and cell
me as asis ha e been shown by inhibi ion o he Golgi -
mannosidase II by he iminosuga swainsonine,25 while a
de iciency in lysosomal α-mannosidases is obse ed in he
inhe i ed diso de mannosidosis.27 The esul s a e epo ed in
Table 2.
A mul i alen e ec was obse ed owa ds GMIIb, since he
a ini y enhancemen pe DAB mo i was highe han 1 ( p/n>1) o
all he mul i alen compounds and was in he ange o 40 o 80.
In con as , mul i alen compounds 13 and 11 we e mo e ac i e
han 5 owa ds LManII, al hough no signi ican a ia ion in he p/n
alues was obse ed in dependence o he alency ( p/n alues
no highe han 3.5 in he bes case).
These esul s unde line a ema kable selec i i y o ou
compounds o GMIIb o e LManII. Simila beha iou has been
epo ed o o he mul i alen iminosuga s based on DNJ,11
con i ming ha LManII canno be a ge ed by hinde ed ligands
and does no ake ad an age o mul i alen p esen a ion o
iminosuga s. On he o he hand, he in e es ing selec i i y
owa ds GMIIb o e LManII appea s pa icula ly ele an o a
selec i e applica ion o he mul i alen compounds in an icance
he apy, wi hou he side e ec o mannosidosis synd ome.

Table 2. Rela i e po ency ( p) and ela i e po ency pe ac i e uni ( p/n) o he i-, e a- and nona alen py olidines 13, 11 and
6 owa ds comme cial JBMan, and ecombinan GMIIb and LManII. IC50 [μM] we e epo ed.
Compound
Valency
JBMan
GMIIb
LManII
p ( p/n)
IC50
p ( p/n)
IC50
p ( p/n)
5
1
-
175
-
2450
-
13
3
138 (46)
1.4
125 (42)
230
10.7 (3.5)
11
4
38 (10)
0.95
184 (46)
380
6.4 (1.6)
6
9
13684 (1520)
0.25
700 (78)
n.d.
n.d.
Finally, in ag eemen wi h he collec ed esul s owa ds
comme cial glycosidases, an in e es ing inhibi ion p o ile was
obse ed wi h human -mannosidase e alua ed in an ex ac
om human leukocy es. Indeed, while a 55% inhibi ion was
obse ed a 5 mM inhibi o concen a ion o he mono alen
compound 5, he mul i alen compounds showed pe cen ages o
inhibi ion highe han 80% a 1 mM inhibi o concen a ion (namely,
91% inhibi ion o he i alen 13, 96% inhibi ion o he e a alen
11 and 80% inhibi ion o he nona alen 6).
NMR s udies
Unde s anding he mechanism o inhibi ion o mul i alen ligands
owa ds he enzymes ep esen s an impo an goal in o de o
expand he knowledge on he p o ein a ge and o help syn he ic
o ganic chemis s in he a ional design o new highly ac i e and
selec i e inhibi o s. Wi h his aim, we decided o in es iga e he
binding mode o ou mul i alen py olidine iminosuga s o JBMan
by a combined NMR/molecula modeling p o ocol.28
In conside a ion ha he ac i e si e amino acids a e highly
conse ed in GH38 -mannosidases amily, and due o he
simila i y in biological ac i i y o he impo an he apeu ic a ge
Golgi α-mannosidase II (GMII, CG18474), JBMan has been o en
used as a model enzyme o s udies wi h inhibi o s.
Among he NMR echniques, STD-NMR (Sa u a ion T ans e
Di e ence-NMR) ep esen s a use ul expe imen o cha ac e ize
he binding in igh ly bound ligand-p o ein complexes. STD-NMR
echnique has been used o gain new insigh s in o he binding
mechanism be ween oligosaccha ides o glycomime ics and
di e en ecep o s, including lec ins and an ibodies.29 In he ield
o glycosidases, STD-NMR expe imen s ha e been employed o
y o unde s and he in e ac ions be ween ecombinan canine α-
ucosidase and a ious pipe idine iminosuga inhibi o s,30 and o
de e mine he ole o he a oma ic ings o ca basuga inhibi o s
in hei in e ac ion wi h α-glucosidase om Saccha omyces
ce e isiae.31 By combining STD-NMR, NOE and molecula
modeling expe imen s, in o ma ion abou he key ea u es in
glycoside ecogni ion o β-glucosidase om ice and β-
mannosidase om ba ley was ob ained o gluco- and manno-
con igu ed subs a es.32 Mo eo e , a combined STD-
NMR/molecula modeling p o ocol has been applied o p edic ing
he binding modes o wo glycomime ic inhibi o s o GMII.33
Howe e , o he bes o ou knowledge, no NMR in es iga ion has
ocused on he in e ac ions o mul i alen iminosuga s wi h ei he
JBMan o any glycosidase o da e.
In o de o in es iga e he mul i alen binding and o ob ain he
epi ope mapping o ou ligands, he mono alen py olidine 5,
DAB-1 (2), and he i- 13, e a- 11 and nona alen 6 ligands we e
incuba ed wi h comme cial JBMan a pH 5 and he STD-NMR
spec a we e acqui ed. The expe imen s we e pe o med wi h
di e en enzyme/ligand a ios depending on he ype o ligand, in
o de o ob ain an STD spec um wi h good quali y. Indeed, o
ligands wi h a lowe a ini y, such as he mono alen 5 and DAB-
1 (2), a high enzyme/ligand a io was equi ed (1:100 and 1:50,
espec i ely). Good STD o he complexes wi h he i- 13, e a-
11 and nona alen 6 we e ob ained wi h JBMan/mul i alen a ios
o 1:100, 1:10 and 1:10, espec i ely. The STD-NMR spec a we e
acqui ed a oom empe a u e, excep o he mono alen ligand,
he weakes inhibi o , o which he empe a u e was dec eased
o 10 °C, o slow he o a e o he complex and maximize he
sa u a ion ans e . The ob ained STD spec a o he complex wi h
DAB-1 (2) and wi h he nona alen iminosuga 6 a e epo ed in
Figu e 2 ( o he whole se o STD spec a, see he Suppo ing
In o ma ion ile.)
Figu e 2. The STD-NMR spec a o DAB (2) and nona alen (a’ and b’)
py olidine iminosuga 6 in complex wi h JBMan and he co esponding o -
esonance (a and b) spec a.
The calcula ion and analysis o he STD in ensi ies o all p o ons
in spec a o compounds 2, 5, 6, 11 and 13 allowed o map he
epi ope. A g aphical ep esen a ion is epo ed in Figu e 3a. The
highes STD (60-100% ela i e STD enhancemen ) is due o H-4
and H-3 and o he p o ons in he middle o he chain (H-9, H-10)
o he mono alen and mul i alen ligands. I is no ewo hy ha
he same end in he STD in ensi ies o he p o ons o he
py olidine ing and o he chain is p esen in all he ligands, as
shown in Figu e 3b. In ligh o hese esul s, i is e idenced ha
he py olidine ing o he mono alen and mul i alen ligands
in e ac s wi h JBMan in he same way o DAB-1, and ha he
binding in e ac ions o he mul i alen iminosuga s a e well
ep esen ed by he mono alen ligand.
These da a show he exis ence o speci ic in e ac ions o he
ligands wi h JBMan, which p esumably ake place wi hin he
enzyme ac i e si e. This is also in ag eemen wi h he inhibi ion
mode ound o he whole se o compounds, showing compe i i e
o mixed ype inhibi ion in all cases (see he Suppo ing
In o ma ion ile).
These expe imen s sugges ha a ecogni ion mechanism
exclusi ely based on non-speci ic in e ac ions o he mul i alen
ligands wi h o he pa s o he enzyme can be disca ded (binding
mode (d) o Figu e 1). Con e sely, a binding mode in ol ing bo h
ac i e si e and non-ca aly ic subsi es canno be comple ely
excluded (binding mode (c) o Figu e 1).
Figu e 3. (a) Epi ope mapping o DAB and o mono- and mul i alen ligands,
ep esen ed as a single chain; (b) G aphical ep esen a ion o he STD
in ensi ies o he p o ons o DAB and mono- and mul i alen ligands.
To expand ou knowledge on he con o ma ion in solu ion o he
py olidine ing o he ligands in hei ee and JBMan-bound
s a es, 2D-NOESY and -NOESY we e acqui ed.28 The ob ained
esul s we e compa ed o hose expec ed o he compu ed
con o ma ion o he mono alen compound 5. F om he analysis
o he acqui ed 2D-NOESY o he ee ligands (see Suppo ing
In o ma ion ile), we obse ed he p esence o he same c oss-
peaks in he mono alen (5) and mul i alen ligands (13, 11 and
6). This e idence sugges s, as expec ed, ha he py olidine ing
main ains he same con o ma ion in all he ligands. Mo eo e , we
obse ed he p esence o a s ong NOE e ec be ween H-3 and
H-4 in he i e-membe ed ing in all compounds, and a small J
coupling be ween hese wo p o ons in he 1H-NMR spec a (H-3
and H-4 appea as na ow signals). These expe imen al esul s,
oge he wi h he analysis o he py olidine con o ma ions o he
mono alen 5 gene a ed by he compu a ional p og am Maes o
10.7, sugges ed ha he mos s able con o ma ion in solu ion is
he 4E. This py olidine con o ma ion was hen used as inpu o
molecula dynamics (MD) simula ions o he mono alen ligand,
as shown in Figu e 4a. The -NOESY expe imen o he
e a alen ligand 11 wi h he enzyme was eco ded wi h a a io
JBMan: e a alen = 1:10 in o de o ob ain he con o ma ion o
he bound py olidine, (see Suppo ing In o ma ion ile). The
in e sion o he c oss-peak sign o he py olidine ing is
ep esen a i e o he de ec ion o he bound con o ma ion,
indica ing ha he ligand is bound o he p o ein. The p esence o
he s ong NOE e ec be ween H-3 and H-4 in he -NOESY,
accoun ed o a majo 4E con o ma ion also in he bound s a e.
In e es ingly, he disappea ance o he NOEs be ween CH2-6 and
H-3 was obse ed in he -NOESY, sugges ing a change in he
o ien a ion o he 6-CH2OH when passing om he ee o he
bound o m. Un o una ely, we did no ind he co ec condi ions
o acqui e he -NOESY o he o he ligands. Ne e heless, on
he basis o he STD-NMR obse a ions, i is highly plausible ha
he same 4E geome ies we e also kep in hei bound o ms.
Molecula dynamics (MD) s udies
To suppo ou expe imen al da a and o p opose a binding mode
o he ligands in o he ac i e si e o he α-mannosidase, a
molecula modeling app oach was pe o med. The X- ay c ys al
s uc u e o JBMan is no a ailable ye . Ne e heless, he
comple e p ima y sequence, ecen ly epo ed by Kuma e al,8d
was used o check he homology be ween JBMan and o he α-
mannosidases . In pa icula , he sequence alignmen was
pe o med wi h wo α-mannosidases wi h wo desc ibed X- ay
c ys al s uc u es, he lysosomal α-mannosidase om Bos au us
(bo ine, bLAM9b and he Golgi α-mannosidase II om D osophila
melanogas e ( ui ly, GMII).9a Despi e he low pe cen age o
iden i y o he comple e sequences, a signi ican homology was
ound o he ac i e si e egions. This egion o JBMan showed
wi h GMII a 61% iden i y and 86% o simila i y. Fo bLAM, a 93%
o iden i y and 100% o simila i y was ound ( he sequence
alignmen is epo ed in he Suppo ing In o ma ion ile). Mo eo e ,
by compa ing hese h ee di e en mannosidase sequences, i is
no ewo hy ha he ac i e si e aminoacids8d His23, Asp25, T p28,
Ty 35, Asp145 ( he ca aly ic esidue), A g170, Ty 210, Asp268,
Phe269, Ty 325, T p333, His385, His386, Asp387, Th 392 and
Ty 625 a e highly conse ed in all he mannosidases. In addi ion,
by supe imposing he c ys al s uc u es o GMII and o bLAM, he
same 3D o ien a ion o he ac i e si e esidues is main ained (see
Suppo ing In o ma ion ile). Indeed, many o hese esidues a e
in ol ed in he in e ac ions wi h he subs a e and wi h he
inhibi o s in he GMII. Thei p esence in he o he mannosidases
also explains he b oad inhibi o y ac i i y o swainsonine o all he
class II α-mannosidases. The excellen obse ed ma ch a he
ca aly ic si e (see below) allowed us o alida e he use o he
a ailable c ys al s uc u es o he GManII-inhibi o complexes as
s uc u al empla es o he molecula dynamics simula ions. This
app oach was also backed-up by he biological es s desc ibed
abo e, which showed a good a ini y o he mul i alen py olidines
owa ds GMIIb bu no LManII.
The compa ison o he binding modes o di e en compe i i e
inhibi o s wi h py olidine, indolizidine and salicinol s uc u es o
GMII ha e been epo ed by Moi essie e al.34 This analysis
e idenced he impo ance o he hyd oxyl g oups in he in e ac ion
wi h he zinc ion in o he ac i e si e o he enzyme. In pa icula ,
we chose he complexes o GMII wi h salicinol (PDB 1TQS)35 and
he dihyd oxypy olidine iminosuga epo ed by Vogel and co-
wo ke s36 (PDB 2F18) as models, due o he s uc u e simila i y
wi h ou ligand (see Suppo ing In o ma ion ile). On he basis o
hese models, we hen buil he s a ing geome ies o he
possible binding modes o he mono alen py olidine 5 (in he 4E
con o ma ion) wi h GMII. A manual docking app oach was i s
employed, using he Disco e y S udio 4.5 so wa e, by
supe imposing he endocyclic he e oa oms and he hyd oxy
g oups in e ac ing wi h he zinc ion wi h hose o ou ligand (see
Suppo ing In o ma ion ile). MD simula ions o all he gene a ed
complexes we e un wi h Ambe 12 o 20 ns, o e alua e i he
hypo hesized binding modes a e possible in solu ion and o
deduce he mos likely one. The esul s o he MDs a e epo ed
in he Suppo ing In o ma ion ile. The analysis o STD NMR
esul s and some ele an s uc u al ea u es indica ed ha he
mos plausible mode is ha epo ed in Figu e 4a (Mode 1),
ob ained employing he GMII-dihyd oxypy olidine iminosuga
complex as ini ial model. Fi s , we compa ed he expec ed STDs
o his complex, gene a ed by using he CORCEMA-ST ma lab
sc ip s,37 wi h he expe imen al ones. A e y good ag eemen
be ween he p edic ed and expe imen al STD alues was
obse ed (Figu e 4b), s ongly suppo ing he choice o his mode
e sus o he s (see Suppo ing In o ma ion ile).
Figu e 4. The complex o GMII wi h he mono alen py olidine 5. (a) The
selec ed binding mode (Mode 1), de i ing om he analysis o he MD
simula ions. The mono alen ligand is p esen ed in g een and he zinc ion in
magen a. The non-co alen in e ac ions a e ep esen ed wi h he g een do ed
lines, and he coo dina ion o he zinc wi h he ed lines, (b) compa ison o he
p edic ed and expe imen al STD in ensi ies.
The analysis o he non-co alen in e ac ions du ing he MD
pe mi ed o deduce he p esence o an elec os a ic in e ac ion o
he posi i ely cha ged endocyclic ni ogen a om wi h he ca aly ic
esidue Asp204, main ained o e 95% o he MD ime. The ole o
his esidue in he hyd oli ic ac i i y o he enzyme has been
demons a ed by Howa d e al.8b Indeed, i s in e ac ion wi h he
endocyclic ni ogen o he ligands is a common ea u e o he
compe i i e inhibi o s o mannosidase enzymes. Two s able
hyd ogen bonds be ween O-4 o he mono alen ligand and
Asp92, and be ween O-6 and Asp472 we e also obse ed (100%
o MD ime). Mo eo e , in his binding mode, he O-4 and O-6 o
he mono alen ligand coo dina e he zinc ion, leading o i s
hexacoo dina ion, which in ol es also His90, His471, Asp92 and
Asp204. In pa icula , we no ed ha he CH2OH g oup in posi ion
6 o ligand 5 o a es du ing he ini ial s eps o he MD o a o d he
chela e zinc coo dina ion. This esul is in ag eemen wi h he
expe imen ally obse ed change in he NOE in e ac ions o H-6 in
he -NOESY spec um.
On he basis o hese esul s o he monome species, a MD
simula ion o he i alen ligand 13, in oduced in o he ac i e si e
in he binding mode 1 (Figu e 5a) was also ca ied ou . The aim
was o e i y whe he he same in e ac ions we e possible o a
mul i alen ligand, whe e he s e ic hind ance may play an
impo an ole. A MD simula ion o 10 ns was un and analyzed.
The esul s showed ha hexacoo dina ion a zinc was kep du ing
he MD, in ag eemen wi h he MD esul s o he mono alen
ligand 5. Mo eo e , he same s able non-co alen in e ac ions
we e obse ed and a e shown in Figu e 5b. These heo e ical
esul s a e compa ible wi h he same epi ope mapping obse ed
in he STD-NMR expe imen s o he mono alen and mul i alen
ligands, de i ed om he speci ic in e ac ion o he py olidine ing
in o he α-mannosidase ac i e si e. On he basis o hese esul s,
we can p edic ha mul i alen ligands bea ing longe alkyl chains
may beha e as mo e ac i e inhibi o s due o hei easy en ance
in o he deepe ac i e si e o GMII as shown in Figu e 5a.
Figu e 5. (a) The complex o GMII wi h he i alen ligand 13; (b) he binding
mode 1 o 13 in o he ac i e si e. The zinc coo dina ion and he non-co alen
in e ac ions a e shown wi h ed lines and do ed g een lines, espec i ely.
TEM s udies
To ob ain insigh s in o he binding modes explaining he la ge
mul i alen e ec obse ed, he JBMan-ligand complexes we e
also analyzed by T asmission Elec on Mic oscopy (TEM) wi h
nega i e s aining. JBMan is a high molecula -weigh
me alloenzyme o abou 220 KDa composed o wo pai s o
subuni s o molecula weigh o 66 and 44 KDa, espec i ely,8d
and is isible by TEM. Fo he mannosidase alone, he image
clea ly showed pa icles o app oxima ely 10x20 nm (Figu e 6a)
ma ching he size o one dime o JBMan subuni he e odime s,
as p e iously epo ed.12 The complex o JBMan wi h he e a-
and nona alen ligands (11 and 6) we e analyzed by TEM
(Figu es 6b and 6c), a e incuba ion o 1 hou a oom
empe a u e. In bo h complexes, se e al di e en a angemen s
we e obse ed, in ol ing wo o mo e enzyme pa icles. Focusing
on he size and shape o he agg ega es, an “S” shape
a angemen equen ly appea ed. These agg ega es wi h he “S”
shape may co espond o he in e ac ion o one inhibi o wi h he
ac i e si e o wo di e en dime s o JBMan subuni he e odime s,
as p esen ed in he ca oon ep esen a ion (Figu e 6d). In o he
wo ds, he “S” shape o igina es om he in e ac ion o wo
molecules o inhibi o wi h h ee dime s o JBMan subuni
he e odime s. This hypo hesis seems highly plausible, since he
maximum leng h o 3-4 nm o he ligands (measu ed wi h
Disco e y 4.5) limi s i s simul aneous binding o wo di e en
ac i e si es o he same dime (leng h o he JBMan pa icle o 20
nm), in con as o he la ge 48- alen DNJ ligand o Compain and
co-wo ke s.12 Consequen ly, o explain his pa icula
a angemen o he pa icles, we sugges ha he o ma ion o
agg ega es de i es om in e ac ions o he ligand wi h wo ac i e
si es o wo di e en dime s o JBMan subuni he e odime s.
Howe e , i is wo h no ing ha he dimensions o he ligands
s ongly a ec he ype and size o he JBMan agg ega es.
Figu e 6. Elec on mic oscopy images ob ained o JBMan wi h nega i e
s aining. Rep esen a i e a eas show op, il ed and side iews. The scale ba s
d awn on each o he h ee mic og aphs a e 50 nm long. (a) TEM pic u e o
JBMan alone, (b) TEM o JBMan incuba ed o 1 hou wi h he e a alen ligand
11, and (c) wi h he nona alen ligand 6. The agg ega es a e shown in he g een
squa es. (d) Ca oon ep esen a ion o he in e ac ion o he e a alen ligand
wi h JBMan o explain he “S” shape o he agg ega es.
Conclusions
A mul idisciplina y app oach has been employed o s udy a new
a ie y o glycosidase inhibi o s. No el i-, e a- and nona alen
py olidine iminosuga s (13, 11 and 6) ha e been syn hesized
exploi ing he CuAAC cycloaddi ion o an azido iminosuga
in e media e wi h mul i alen alkyne sca olds. The mono alen
e e ence compound 5 has also been p epa ed o compa ison.
Biological inhibi ion assays owa ds a panel o comme cially
a ailable glycosidases ha e highligh ed a ema kable mul i alen
e ec owa ds JBMan, wi h a highes p/n = 1520 o he
nona alen compound 6. Mo eo e , he p o ile o inhibi ion was
e alua ed owa ds he biologically ele an LManII and GMIIb,
showing main enance o he mul i alen e ec wi h GMIIb,
oge he wi h an in e es ing selec i i y owa ds GMIIb o e LManII.
These obse a ions open he way o he he apeu ic applica ion
o hese mul i alen compounds as selec i e inhibi o s o Golgi α-
mannosidases. To in es iga e he binding mode o he mul i alen
inhibi o s o α-mannosidases, a combined
expe imen al/ heo e ical STD-NMR/MD analysis has also been
ca ied ou . This p o ocol allowed o assess ha a speci ic
in e ac ion o hese molecules in o he glycosidase ac i e si e
akes place. Finally, TEM s udies shed ligh on he mul i alen
binding mode, showing he o ma ion o ligand-JBM agg ega es,
p obably by an in e molecula c oss-linking mechanism. The
esul s o his in eg a ed and mul idisciplina y app oach highligh
key aspec s o he design o new mo e ac i e inhibi o s. Indeed,
we p esume ha bo h he leng h o he alkyl chain and he global
size o he mul i alen ligand may a ec enzyme a ini y and
in luence he ype o agg ega ion be ween he ligands and he
enzyme molecules. Wo k is unde way in ou labo a o ies in o de
o p o e hese hypo heses.
Expe imen al Sec ion.
Syn hesis.
Gene al me hods: Comme cial eagen s we e used as ecei ed. All
eac ions we e ca ied ou unde magne ic s i ing and moni o ed by TLC
on 0.25 mm silica gel pla es (Me ck F254). Column ch oma og aphies we e
ca ied ou on Silica Gel 60 (32-63 μm) o on silica gel (230-400 mesh,
Me ck). Yields e e o spec oscopically and analy ically pu e compounds
unless o he wise s a ed. 1H NMR spec a we e eco ded on a Va ian
Me cu y-400 o on a Va ian INOVA 400 ins umen a 25 °C. 13C NMR
spec a we e eco ded on a Va ian Gemini-200 o on a Va ian Gemini-300.
Chemical shi s a e epo ed ela i e o TMS (1H: δ = 0.00 ppm) and CDCl3
(13C: δ = 77.0 ppm). In eg als a e in acco dance wi h assignmen s,
coupling cons an s a e gi en in Hz. Fo de ailed peak assignmen s 2D
spec a we e measu ed (COSY, HSQC, NOESY, and NOE as necessa y).
Small scale mic owa e assis ed syn heses we e ca ied ou in a CEM
Disco e mic owa e appa a us o syn hesis wi h an open eac ion essel
and an ex e nal su ace senso . IR spec a we e eco ded wi h a BX FT-
IR Pe kin-Elme Sys em spec opho ome e . ESI-MS spec a we e
eco ded wi h a The mo Scien i ic™ LCQ Flee Ion T ap Mass
Spec ome e . Elemen al analyses we e pe o med wi h a Pe kin-Elme
2400 analyze . Op ical o a ion measu emen s we e pe o med on a
JASCO DIP-370 pola ime e .
Syn hesis o mono alen compound 5: To a solu ion o 4 (80 mg, 0.31
mmol) in 6 ml o 2:1 THF/H2O, CuSO4 (30 mol%, 15 mg, 0.09 mmol),
sodium asco ba e (60 mol%, 37 mg, 0.19 mmol) and 3-bu yn-1-ol (28 µL,
0.37 mmol) we e added. The eac ion mix u e was s i ed in mic owa e a
80 °C o 45 min, un il a TLC analysis (CH2Cl2:MeOH 6:1 + 1% / 6%
NH4OH) showed he disappea ance o he s a ing ma e ial (R = 0.29) and
he o ma ion o a new p oduc (R = 0.00). A e il a ion h ough Celi e®,
he sol en was emo ed unde educed p essu e and he c ude was
pu i ied by FCC (CH2Cl2:MeOH 1:1 + 1% / 6% NH4OH) a o ding pu e 5
(R = 0.24, 91 mg, 0.28 mmol, 89% yield) as a yellow oil. [α]D23 = -15.7 (c
= 0.21 in MeOH); 1H-NMR (400 MHz, D2O):

= 7.69 (s, 1H, H-T iazole),
4.26 ( , 3J(H,H)= 6.8 Hz, 2H, H-12), 4.04-4.03 (m, 1H, H-4), 3.85-3.83 (m,
1H, H-3), 3.72 ( , 3J(H,H)=6.6 Hz, 2H, CH2OH), 3.64 (d, 3J(H,H)=5.4 Hz,
2H, H-6), 3.04 (d, 2J(H,H)=11.2 Hz, 1H, Ha-5), 2.86-2.76 (m, 2H, Hb-5 and
Ha-7), 2.80 ( , 3J(H,H)=6.4 Hz, 2H, CH2CH2OH), 2.68-2.62 (m, 1H, H-2),