Effect of mofezolac-galactose distance in conjugates targeting cyclooxygenase (COX)-1 and CNS GLUT-1 carrier

Article abstract of DOI:10.1016/j.ejmech.2017.09.066

Neuroinflammation is the earliest stage of several neurological and neurodegenerative diseases. In the case of neurodegenerative disorders, it takes place about 15–20 years before the appearance of specific neurodegenerative clinical symptoms. Constitutive microglial COX-1 is one of the pro-inflammatory players of the neuroinflammation. Novel compounds 3, 14 and 15 (Galmof₀, Galmof₅ and Galmof₁₁, respectively) were projected, and their synthetic methodologies developed, by linking by an ester bond, directly or through a C5 or C11 unit linker the highly selective COX-1 inhibitor mofezolac (COXs selectivity index > 6000) to galactose in order to obtain substances capable to cross blood-brain barrier (BBB) and control the CNS inflammatory response. 3, 14 and 15 (Galmofs) were prepared in good to fair yields. Galmof₀ (3) was found to be a selective COX-1 inhibitor (COX-1 IC₅₀ = 0.27 μM and COX-2 IC₅₀ = 3.1 μM, selectivity index = 11.5), chemically and metabolically stable, and capable to cross Caco-2 cell monolayer, resembling BBB, probing that its transport is GLUT-1-mediated. Furthermore, Galmof₀ (3) powerfully inhibits PGE₂ release higher than mofezolac (1) in LPS-stimulated mouse BV2 microglial cell line, a worldwide recognized neuroinflammation model. In addition, Fingerprints for Ligands and Proteins (FLAP) was used to explain the different binding interactions of Galmofs with the COX-1 active site.

Full text of DOI:10.1016/j.ejmech.2017.09.066

Accepted Manuscript
Effect of mofezolac-galactose distance in conjugates targeting cyclooxygenase
(COX)-1 and CNS GLUT-1 carrier
Maria Grazia Perrone, Paola Vitale, Savina Ferorelli, Angelina Boccarelli, Mauro
Coluccia, Alessandra Pannunzio, Federica Campanella, Giuseppe Di Mauro, Carmela
Bonaccorso, Cosimo G. Fortuna, Antonio Scilimati
PII:
S0223-5234(17)30782-1
10.1016/j.ejmech.2017.09.066
EJMECH 9781
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 21 July 2017
Revised Date: 25 September 2017
Accepted Date: 29 September 2017
Please cite this article as: M.G. Perrone, P. Vitale, S. Ferorelli, A. Boccarelli, M. Coluccia, A. Pannunzio,
F. Campanella, G. Di Mauro, C. Bonaccorso, C.G. Fortuna, A. Scilimati, Effect of mofezolac-galactose
distance in conjugates targeting cyclooxygenase (COX)-1 and CNS GLUT-1 carrier, European Journal
of Medicinal Chemistry (2017), doi: 10.1016/j.ejmech.2017.09.066.
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.

ACCEPTED MANUSCRIPT
H CO
3
OCH3
3
(Galmof0)
COX-1 IC50 = 0.27 µM
O
N
O
O
HO
HO
O
OH
OH

ACCEPTED MANUSCRIPT
Effect of mofezolac-galactose distance in conjugates targeting Cyclooxygenase (COX)-1
and CNS GLUT-1 carrier
a
a
a
b
a
Maria Grazia Perrone , Paola Vitale , Savina Ferorelli , Angelina Boccarelli , Mauro Coluccia ,
a
a
c
c
Alessandra Pannunzio , Federica Campanella , Giuseppe Di Mauro , Carmela Bonaccorso , Cosimo G.
c
,a
Fortuna , Antonio Scilimati*
a
Department of Pharmacy - Pharmaceutical Sciences, University of Bari “Aldo Moro”, Via E. Orabona
4
, 70125 Bari (Italy);
b
Department of Biomedical Sciences and Human Oncology, University of Bari “Aldo Moro”, Piazza
Giulio Cesare 11, 70124 Bari (Italy);
c
Department of Chemical Science, University of Catania, Viale Andrea Doria 6, 95125 Catania (Italy).
*
Corresponding author. Tel.: +39 0805442753; fax: +39 0805442724. antonio.scilimati@uniba.it (A. Scilimati).

ACCEPTED MANUSCRIPT
Abstract: Neuroinflammation, as the earliest stage of several neurological and neurodegenerative
diseases, takes place about 15-20 years before the appearance of specific neurodegenerative clinical
symptoms. Constitutive microglial COX-1 is one of the pro-inflammatory players of the
neuroinflammation. Novel compounds 3, 14 and 15 (Galmof , Galmof and Galmof , respectively)
0
5
11
were projected, and their synthetic methodologies developed, by linking by an ester bond, directly or
through a C5 or C11 unit linker the highly selective COX-1 inhibitor mofezolac (COXs selectivity index
>
6,000) to galactose in order to obtain substances capable to cross blood-brain barrier (BBB) and
control the CNS inflammatory response. 3, 14 and 15 (Galmofs) were prepared in good to fair yields.
Galmof (3) was found to be a selective COX-1 inhibitor (COX-1 IC50 = 0.27 µM and COX-2 IC50 = 3.1
µM, selectivity index = 11.5), chemically and metabolically stable, and capable to cross Caco-2 cell
monolayer, resembling BBB, probing that its transport is GLUT-1-mediated. Furthermore, Galmof (3)
powerfully inhibits PGE release higher than mofezolac (1) in LPS-stimulated BV2 microglial cell line,
0
0
2
a worldwide neuroinflammation model. In addition, FLAP was used to explain the different binding
interactions of Galmofs with the COX-1 active site.
Keywords: Cyclooxygenase-1 inhibition, mofezolac, Galmof, Caco-2 cell line, GLUT-1, molecular
modeling

ACCEPTED MANUSCRIPT
1
.
Introduction
Up to now, there is no worldwide shared protocol to treat the human neuroinflammatory syndrome. Its
onset and progression diagnosis are not yet possible by any hospital available technique, but only
through lately appearance of clinical symptoms of encephalon diseases [1,2]. Chronic
neuroinflammation is tightly connected to brain degenerative processes [3], and it is considered the
earliest step of several neurodegenerative disorders such as Alzheimer and Parkinson disease,
amyotrophic lateral sclerosis, multiple sclerosis, traumatic brain injury, and HIV encephalitis [4]. It
seems that it takes approximately fifteen-twenty years from perpetuating insults inducing chronic
neuroinflammation and clinical symptoms appearance of dementia or other neurodegenerative disease
[5].
The lack of drugs to treat several diseases involving the central nervous system (CNS) also resides into
the shield exerted by the blood brain barrier (BBB) matrix. BBB has a low permeability, and the
development of drugs able to penetrate through its network is one of the challenges of all scientists
involved in projecting medicines having active principle ingredients targeting the CNS diseases. A
commonly used strategy to overcome this drawback consists to incorporate into the pharmacological
active molecule a sugar moiety (i.e. glucose or galactose), in turn capable to “carry” the entire molecule
into the CNS by the GLUT-1 carrier, which is located on the membrane of the endothelial cells.
GLUT-1 belongs to a family of homologous proteins GLUT-1-5, which differ in their tissue
distribution and affinity for naturally occurring hexoses. GLUT-1 is present in erythrocyte, BBB and
placenta and has a nanomolar affinity constant for D-glucose and D-galactose [6].
The choice of the hexose to be conjugated to the active pharmaceutical ingredient (API) is based upon
its involvement into the metabolic homeostasis [7]. D-Galactose has been preferred to D-glucose
because the latter is involved in many metabolic activities, alterable by chronic administration of
glucosyl derivatives. Furthermore, mofezolac was selected as API, being a highly selective

ACCEPTED MANUSCRIPT
®
cyclooxygenase(COX)-1 inhibitor [8]. It is commercially available in Japan as Disopain and is
clinically used to treat human algesia and rheumatoid arthritis.
Microglial COX-1 exerts its pro-neuroinflammatory role, by producing also the pro-inflammatory
prostaglandin PGE as a rapid response to several CNS inciting harmful stimuli. Thus, a selective COX-
2
1
inhibition is expected to contribute to reduce neuroinflammation by blocking part of the complex
network involved in the onset of the inflammatory process [9,10]. This prominent role of COX-1 over
COX-2, the other known COX isoform, could be explained by considering that COX-1 is constitutively
expressed in microglia, then the first response to inflammatory inducers is COX-1 mediated, whereas
the second response is due to the induced COX-2 [4,5].
Hence, keeping on our work aimed at designing molecules capable to fit well into the COX-1 active site
[11-14] and in this case also able to cross BBB, we projected the new molecules 3, 14-15 (Galmof0,
Galmof and Galmof ) in which mofezolac is directly linked to D-galactose or by linkers of different
5
11
features. Herein, we report mofezolac synthesis optimization and the preparation of Galmof Galmof
5
0
,
and Galmof . Their permeability across the monolayer of the human colon adenoma derived cell line
1
1
(Caco-2) was also measured. Caco-2 cells were chosen as a model of BBB crossing mediated by GLUT-
1
. Immunofluorescence experiments were performed to confirm the GLUT-1 expression in Caco-2 cell
line and its involvement in the transport of 3, 14-15 through Caco-2 cell line compact layer. Both
chemical and metabolic stability of 3, resulting more potent and selective COX-1 inhibitor than 14 and
1
5, were evaluated. 3 was also evaluated, in comparison to mofezolac used as reference, in LPS-
stimulated BV2 microglial cell line. The latter is scientifically considered an in vitro model of
neuroinflammation. FLAP software was used to rationalize the different COXs inhibition capability of
the new synthesized compounds with respect to mofezolac, through the binding interactions between 3,
1
4-15 and COX-1 active site.

ACCEPTED MANUSCRIPT
2
3
.
Results and Discussion
, 14-15 (Galmof Galmof and Galmof , respectively) chemical structures were designed considering
0
,
5
11
the Barnett et al.’s investigation [16], probing the importance of the hydroxyls at both D-glucose C1 and
C3 as hydrogen-bonding acceptors during D-glucose crossing human erythrocyte sugar-transport
system, presumably GLUT-1 [17]. Glucose C4-hydroxyl also binds GLUT-1 through a further H-
bonding. In fact, D-galactose, as C4 epimer of D-glucose, has an affinity for GLUT-1 10-fold lower
than D-glucose, that in turn could be explained by considering the non-gluco configuration of the sugar
hydroxyl group sterically hinders the transport [17]. The OH group at sugar-C6 seems to be not
involved in H-bond with GLUT-1, and bulky substitutions (O-Alkyl or O-Phenyl) at this position are
tolerated, whereas the same substitutions at the anomeric C1 are not tolerated [16]. Based on this
GLUT-1 cell biology knowledge, 3, 14 and 15 were designed linking mofezolac to the galactose
through its C6 hydroxyl group.
2
5
.1. Synthesis of 3, 14-15 (Galmof0, Galmof and Galmof11)
Mofezolac (1) was firstly esterified with 1,2:3,4-di-O-isopropylidene-α-D-galactopyranose (DIPG) in
the presence of N-ethyl-N’-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) and 4-
(dimethylamino)pyridine
(DMAP)
to
afford
1,2:3,4-di-O-isopropylidene-6-O-{2-[3,4-bis(4-
methoxyphenyl)isoxazol-5-yl]acetyl}-D-galactopyranose (2), in turn deprotected with trifluoroacetic
acid (TFA) to 3 (Galmof ) (Scheme 1).
0

ACCEPTED MANUSCRIPT
2 2 2 2
Scheme 1. (a) DIPG, EDC, DMAP, anhydrous CH Cl , r.t., 24 h; (b) TFA, anhydrous CH Cl , r.t., 16h.
5
To prepare 14 (Galmof ) and 15 (Galmof11), mofezolac was linked to galactose by 5-aminopentanoic
acid (4) or 11-aminoundecanoic acid (5), respectively. The acids 4 and 5 were first transformed into
their corresponding methyl esters 6 and 7 in quantitative yields by SOCl /MeOH and then allowed to
2
react with mofezolac (1) in the presence of 1-hydroxybenzotriazole hydrate (HOBt),
dicyclohexylcarbodiimide (DCC) and N,N-diisopropylethylamine (DIEA) at r.t. to afford 8 and 9,
respectively. Methyl esters 8 and 9 were, then, hydrolyzed to their corresponding acids 10 and 11 by
2
LiOH/THF-H O, in turn reacted with DIPG to give 12 and 13, that were deprotected with TFA to 14
(
Galmof ) and 15 (Galmof11) (Scheme 2).
5

ACCEPTED MANUSCRIPT
H CO
3
OCH3
O
O
O
a
b
H
N
H N
2
H N
2
OH
OCH3
OCH3
n
n
n
N
O
O
4
5
(n=4)
(n=10)
6 (n=4)
7 (n=10)
8 (n=4)
9 (n=10)
H CO
OCH3
H CO
3
OCH3
3
O
n
O
c
d
H
H
N
N
O
OH
n
N
N
O
O
O
O
O
O
O
O
1
1
2 (n=4)
3 (n=10)
1
1
0 (n=4)
1 (n=10)
O
H CO
OCH3
3
O
H
N
e
O
n
N
O
O
HO
O
HO
OH
OH
14 (n=4)
15 (n=10)
Scheme 2. (a) SOCl , anhydrous MeOH; (b) mofezolac (1), HOBt, DCC, DIEA, anhydrous CH Cl , r.t.,
2
2
2
2
2 2 2
4 h; (c) LiOH/THF-H O, 3h, HCl; (d) DIPG, EDC, DMAP, anhydrous CH Cl , r.t., 24 h; (e) TFA,
anhydrous CH Cl , r.t., 16h.
2
2
2
.2. Immunofluorescence experiments
Caco-2 cell monolayer is a commonly used in vitro method to predict a compound ability to permeate
physiological barriers [18]. The presence of GLUT-1 in Caco-2 cells was verified by
immunofluorescence experiments before testing whether biological membrane crossing by the newly

ACCEPTED MANUSCRIPT
synthesized compounds is carrier-mediated. Caco-2 cells were fixed and immunofluorescently stained
with an anti-GLUT-1 mouse monoclonal antibody. The primary antibody was visualized using
®
AlexaFluor 594 labeled secondary antibody and the nuclei were counterstained with Hoechst 33342.
There was a positive GLUT-1 signal staining in the cells incubated with anti-GLUT-1 mouse
monoclonal antibody, whereas no staining was observed in the cells incubated with BSA 3% (no
immune control) (Figure 1).
Figure 1. Perkin-Elmer Operetta images (20x magnification) acquired from Hoechst 33342 (a,a’) and
®
Alexa Fluor 594 (b, b’) channels (a,b: anti-GLUT-1 mouse monoclonal antibody; a’,b’: no immune
control). Merged images of a-b ad a’-b’ are shown in c and c’, respectively.
2
.3. Permeability experiments by 3, 14 and 15
Once the presence of GLUT-1 in Caco-2 cells was confirmed [19], the newly synthesized compounds 3,
4 and 15 (Galmof0, Galmof and Galmof11) were assayed for their ability to permeate the compact layer
1
5
of Caco-2 cells expressing GLUT-1 (Figure 2).
The Apparent Permeability (Papp) in both basolateral to apical [Papp(B→A)] and apical to
basolateral [Papp(A→B)] directions was determined. The [Papp(B→A)] flux of each compound alone

ACCEPTED MANUSCRIPT
or in the presence of phloretin, a GLUT-1 inhibitor, was determined (Table 1). (B→A) flux represents
the active transport since GLUT-1 is expressed in the basolateral compartment [19]. Then, a high
(B→A) flux value would mean that the transport should be GLUT-1 mediated (Figure 2). The (A→B)
flux, representing the passive transport, was also been determined for each compound (Table 1).
Figure 2. Exemplified representation of permeability experiment. A) GLUT-1-mediated crossing of
Galmofs through Caco-2 cells monolayer. B) Caco-2 cells monolayer crossing by Galmofs prevented by
the GLUT-1 inhibitor phloretin.
(B→A) and (A→B) fluxes of 1 (mofezolac) were also determined. The corresponding Papp values
were found to be comparable since no active transport is involved, probably due to the absence in the
molecule of a sugar moiety that would be recognized by GLUT-1. Higher Papp (B→A) values than 1
(
mofezolac) were observed for all Galmofs, indicating that a carrier-mediated active transport
mechanism is implicated. 3 (Galmof ) shows an apparent permeability of 875 nm/sec, whereas in the
presence of phloretin, a 30% reduction of the basolateral to apical flux was observed [Papp (B→A) =
10 nm/sec], and the passive transport is not affected by GLUT-1 inhibitor pre-treatment. For 14
Galmof ) the percentage of the apparent permeability reduction is of 32%, similarly to 3 (Galmof );
0
6
(
5
0

ACCEPTED MANUSCRIPT
[Papp (B→A)] varies from 1129 to 771 nm/sec in the presence of the GLUT-1 inhibitor. The apparent
permeability of 15 (Galmof ) [Papp (B→A) = 587 nm/sec] in the presence of phloretin decreases by
1
1
3
0% [Papp (B→A) = 415 nm/sec]. 14 and 15 passive transport remains unchanged independently from
the phloretin pre-incubation (Table 1).
Table 1. Papp (nm/sec) values of 3,14-15 (Galmof Galmof and Galmof ) alone or in the presence of
0
,
5
11
1
00 mM Phloretin.
Papp B→A
nm/sec)
Papp A→B
Compds
1
(
(nm/sec)
461
414
phloretin
phloretin
a
a
a
a
3
875
1129
587
610
380
233
169
368
289
172
a
1
1
4
5
771
a
415
a
Papp was determined after 24 h incubation of cells with 100 mM phloretin (GLUT-1 inhibitor).
These flux extent reductions, in the presence of phloretin, indirectly demonstrate the involvement of
GLUT-1 in Galmofs transport across the monolayer of Caco-2 cells. Overall, it can be assumed that
when mofezolac is linked to galactose (3, 14-15) is capable to cross the cell monolayer faster than
mofezolac alone. In fact, 1 [Papp (B→A)] = 461 nm/sec becomes 875, 1129 and 587 nm/sec for 3
0 5
(Galmof ), 14 (Galmof ) and 15 (Galmof11), respectively. These data were confirmed by
immunofluorescence and transportability experiments.
2
.4. Fluorescence competition binding assay

ACCEPTED MANUSCRIPT
Competitive 2-NBDG cell uptake assays were performed to further probe the involvement of GLUT-1
in the transport of the newly synthesized compounds 3, 14 and 15 in Caco-2 cell system. D-Glucose
transport, the GLUT-1 natural substrate, was used as a reference. 2-NBDG is a fluorescently labeled
deoxyglucose analogue primarily used to directly monitor glucose uptake through GLUTs, by living
cells and tissues [20]. Specifically, Figure 3 shows the competitive inhibition of 2-NBDG by D-glucose
and phloretin, a GLUT-1 inhibitor [21].
a
b
c
d
Figure 3. Representative microscopic images of Caco-2 cells in the same field are reported, including
brightfield (a), 2-NBDG (b), and Hoechst 33342 (c) channels, and colored overlay (d). Variability of 2-
NBDG uptake by Caco-2 cells is shown through examples of i) cells with no 2-NBDG label, ii)

ACCEPTED MANUSCRIPT
diffusely 2-NBDG-labeled cells, and iii) cells with granular morphology pattern of 2-NBDG labeling,
identified by yellow, red, and orange arrows, respectively.
2
-NBDG uptake in Caco-2 cells was found to be dose-dependent (see Supplementary material). Its
uptake (Figure 4A) and consequently the percentage of labeled cells (Figure 4B) were reduced with
increasing D-glucose concentrations (1, 5 and 10 mM). Similarly, increasing concentrations (0.01, 0.5
and 0.1 mM) of 3 (Galmof ) 14 (Galmof ), and 15 (Galmof ) reduced 2-NBDG uptake, supporting the
0
,
5
11
idea that their transport is GLUT-1-mediated (Figure 4).
Figure 4. Average population-level data were calculated from the average image intensity (arbitrary
fluorescence units) of 2-NBDG (green channel). All single-cell raw data were normalized to
background level, and presented as mean ± SEM of three independent experiments. (b) The percentage
of 2-NBDG-labeled cells (on total cell number) for 2-NBDG (0.3 mM), Galmofs compounds (0.01-0.1
mM) and D-glucose (1-10 mM) is shown.
2
.5. Cytofluorimetric analysis of Caco-2 cell cycle

ACCEPTED MANUSCRIPT
Cytofluorimetric analysis of Caco-2 cell cycle was performed to evaluate the effects of mofezolac and
Galmofs on cellular cycle of Caco-2 cells used for permeability and competition binding assays. The
treatment of Caco-2 cells for 24 h with 100 µM of each compound (1, 3, 14-15) shows a common
cytostatic effect. In particular, compound 14 (Galmof ) has a prolonged and marked effect in phase S
5
and G2/M. Conversely, compounds 3 (Galmof ) and 15 (Galmof ) show control over a cycle recovery
0
11
that is more effective in compound 15 (Galmof ) than compound 3 (Galmof ) (Figure 5). These results
1
1
0
demonstrate that only compound 14 (Galmof ) at the high dose of 100 µM perturbs the cell cycle, for
5
all the other compounds only a transient alteration of the normal phases of the cycle was observed.
Figure 5. Percentage of Caco-2 cells in G , S, and G /M phases of the cell cycle during continuous
1
2
exposure to 1 (mofezolac), 3 (Galmof ) 14 (Galmof ), and 15 (Galmof ) or control (untreated cells)
0
,
5
11
for 24 h at the concentration of 100µM.
2
.6. Cyclooxygenase inhibition evaluation
Human cyclooxygenase activity inhibition was determined by a colorimetric COX Inhibitor Screening
Assay which measures the peroxidase component of the cyclooxygenases, monitoring the appearance
of oxidized N,N,N’,N’-tetramethyl-p-phenylenediamine (TMPD) at λ = 590 nm. 1 (mofezolac) is a
selective and potent COX-1 inhibitor with IC50 = 0.0079 µM and 38% inhibition on COX-2 at the

ACCEPTED MANUSCRIPT
highest final inhibitor concentration of 50 µM. The conjugation of 1 with D-galactose (3) determines a
reduction of potency and selectivity: COX-1 IC50 = 0.27µM and COX-2 IC50 = 3.1µM with 87%
inhibition on COX-2 isoform at the highest final inhibitor concentration of 50 µM. The insertion of a
C5-linker (8) between the two moieties 1 (mofezolac) and galactose was found to be detrimental to the
5
COXs inhibitory activity of 14 (Galmof ). In fact, its IC50 was higher than 50 µM for both isoforms and
the inhibition percentages, at this concentration, were found to be 38% for COX-1 and 24% for COX-2.
The presence of the longer spacer determined a recovery of COXs inhibition potency, infact 15
(
Galmof ) has a COX-1 IC = 0.40 µM and of 0.27 µM for COX-2, and an inhibition percentage of 90
11 50
and 64% for COX-1 and COX-2, respectively, at the highest final inhibitor concentration of 50 µM
Table 2 and Figure 6).
(
Table 2. COX inhibitory activity of 1 and 3, 14 and 15.
COX-1
COX-2
SI
Compound
a
b
a
b
IC50 ( µM)
inhibition (%)
IC50 ( µM)
inhibition (%)
1
3
0.0079 ± 0.15
0.27 ± 0.09
>50
100
>50
38
87
24
64
6,392
11.5
-
100
38
3.1 ± 0.4
>50
1
4
5
1
0.40 ± 0.05
90
0.27 ± 0.03
0.67
a
b
IC values are the means of at least three independent measurements. Inhibition percentage (%) was
5
0
c
determined at the highest final concentration of inhibitor used (50 µM). SI (Selectivity Index) = COX-
IC / COX-1 IC .
2
5
0
50

ACCEPTED MANUSCRIPT
1
00
MOFEZOLAC
3
1
1
4
5
5
0
0
-
11 -10
-9
-8
-7
-6
-5
-4
-3
log [inhibitor] (M)
Figure 6. Concentration-response curves for COX-1 and COX-2 activity inhibition by 1 (Mofezolac)
and 3, 14-15 (Galmof0, Galmof and Galmof11, respectively).
5
2.7
Effect of Galmof (3) vs mofezolac (1) on PGE biosynthesis in LPS-stimulated BV2 microglial
0 2
cells.
The PGE
2
biosynthesis extent was evaluated in supernatants of cell cultures at 48 h incubation time
(
Table 3). PGE production in LPS-stimulated cells was significantly higher than its basal level present
2
in the untreated cells (control). Interestingly, both COX-1 inhibitors reduce PGE release in LPS-
2
treated cells (Table 3). Specifically, Galmof was found to be more effective than mofezolac even at a
0
higher concentration (10 µM for 3 vs 0.5 µM for 1), (Table 3) [10].
Table 3. Effect of Galmof
lipopolysaccharide (LPS)-stimulated BV2 microglial cells.
Control LPS 0.5 µM 3 +LPS 1 µM 3 +LPS
.454±0.006 0.7555±0.032 0.0086±0.0002 0.0178±0.0003 0.0103±0.0006
0 2
(3) and mofezolac (1) on PGE release (ng/ml) at 48 h in
10 µM 3 +LPS 0.5 µM 1 +LPS
0.0328±0.002
0

ACCEPTED MANUSCRIPT
2
.8. Modeling Studies
The conjugation of mofezolac with D-galactose determined a change in potency and selectivity for all
the three newly synthesized compounds with respect to 1. As an attempt to find out the rationale of
such results, an in silico study of the binding mode of 3, 14-15 (Galmof Galmof and Galmof ) into
0
,
5
11
the cavity of the target COX-1 co-crystallized with 1 (mofezolac) (PDB code: 5HQK) [8] was
performed using Fingerprints for Ligands and Proteins (FLAP) [22].
2
.9. Structure-based binding mode
FLAP is a Virtual-Screening and Model-Development method based on 3D molecular similarity that is
measured through common Molecular Interaction Field (MIF) volumes generated by GRID. FLAP is
an extremely flexible method that enables a wide variety of applications, from ligand-based Virtual-
Screening [23-25] and 3D pharmacophore hypothesis generation [26,27], to structure-based Virtual-
Screening [28] and high-throughput proteins pockets clustering. FLAP structure-based approach was
used to investigate the possible binding modes of 3, and 14-15 (Galmofs) in COX-1 active site.
The X-ray structure of 1 (mofezolac) was removed from the COX-1:1 complex crystal structure [8],
and the GRID [29] molecular interaction fields were computed into the cavity. To validate the model
generated by FLAP, the binding pose of 1, without any constrain, was studied. Its binding pose in the
COX-1 active site was found to be coincident with the pose of mofezolac in the above mentioned
complex reported in PDB 5HQK (Supplementary material, Figure 3.1) [8].
As far as 3 (Galmof ) is concerned, most of the interactions of 1 (mofezolac) with COX-1 active
0
site are retained. The methoxy group linked to the phenyl ring at isoxazole C3 interacts with Tyr385
whereas the methoxy of the phenyl ring in position C4 with Ser353, and hydrophobic interaction of the
phenyl ring with Ile523 and Phe518 was observed. The presence of all the galactose polar groups
0
would suggest the establishment of new interactions between 3 (Galmof ) and the enzyme catalytic

ACCEPTED MANUSCRIPT
cavity, and from a comparison of the best poses of 3 and 1 (Supplementary material, Figure 3.2) it is
evident that, despite the presence of galactose moiety, the interactions of 1 and the amino acid residues
forming the cyclooxygenase-1 catalytic cavity are still predominant, so that the two molecules 3 and 1
fitting in COX-1 active site is almost the same.
A
B
C
D
Figure 7. Binding poses of 1 (mofezolac) (A), 3 (B), 14 (C), and 15 (D) in the COX-1 active site: the
main amino acid interactions are only indicated.

ACCEPTED MANUSCRIPT
Analyzing 14 (Galmof ), and 15 (Galmof ) best poses (Supplementary material, Figure 3.3 and 3.4,
5
11
respectively), it is, instead, clear that the presence of a spacer between 1 portion and galactose leads to
a different disposition of the two compounds in the COX-1 active site.
In particular, for 14 the methoxy group of the phenyl ring at isoxazole C4 interacts with Ser353. The
OH in position 2’ of the galactose ring interacts with Tyr355, whereas the OH at position 1’ interacts
with Arg120. The phenyl ring at the isoxazole C3 shows a hydrophobic interaction with Ala527 and
Trp387, and the methylenic chain presents a hydrophobic interaction with Leu531 and Val349. For
compound 15, the isoxazole ring interacts with Tyr385 establishing a H-bond. The OH at 4’ of the
galactose interacts with Tyr355, and the methoxy group of the phenyl ring in position C4 interacts with
Met522 and Ile523. A hydrophobic interaction was also observed with Leu384, Ala527, Met522 and
Trp387. The methylenic chain shows a hydrophobic interaction with Val349, Leu359 and Val116.
The different binding interactions described-above and an increase number of methylene in the spacers,
which corresponds to a change in the pocket fitting, could explain the different in vitro COXs
inhibitory activity values (Table 2 and Figure 6). These results are corroborated by a statistic analysis
of the scores reported in Table 4 and Figure 8, where the available screening results are listed and
depicted.
The GRID approach is a well assessed concept for determining energetically favorable interaction sites
in molecules with known structure using chemical probes e.g., H, O, N1, and DRY probes which
describes the shape, hydrogen bond acceptor, hydrogen bond donor and hydrophobic interactions,
respectively. FLAP creates a common reference framework in two steps: first, the MIFs of the
molecules are calculated using the GRID force fields, and the resulting MIF’s are condensed in
complexity by extracting points (quadruplets or hotspot) representing the most favorable interactions.
Second, each quadruplet of these points is used to generate different superpositions of the test
molecules onto a template molecule. The quadruplets of each molecule are stored as a pharmacophoric

ACCEPTED MANUSCRIPT
fingerprints and used to evaluate their similarity. Superposition of quadruplets is assessed through
Probe scores and Distance score, which represents the degree of overlap of the MIFs for each probe
individually as well as for their combinations and overall difference of probe score between the ligand
and template, respectively. In addition, FLAP calculates Global Sum score (Glob_Sum) and Global
Product scores (Glob_Prod). The former score is produced by summing all the scores of the individual
probes together and the later score is produced by multiplying all the scores of the individual probes
together [26,27].
The distance parameter of the three Galmofs and mofezolac are comparable, which justifies the
goodness of the model.
In particular, analyzing the Glob Sum (G.S.) value, which takes into account the sum of the interactions
with all the probes, and the DRY values which are linkable to the different length of the spacer, it is
possible to deduce a stronger binding interaction for 3 and 15 than 14, that could justify the differences
in the in vitro COXs inhibitory activity values (Table 2). The analysis of all these statistic data could be
useful for the modelling of a new molecule, eventually more potent than mofezolac, by designing a
pharmacophore that indicates the retained features for an active scaffold.
Table 4: Scores for mofezolac (1) and 3, 14-15 (Galmof0, Galmof5, Galmof11) in COX-1:mofezolac
crystal structure determined by X-ray crystallography.
Compound
Distance
8.82
Glob Prod
0.56
Glob Sum
2.108
H
N1
Dry
O
1
3
0.955
0.907
0.878
0.146
0.083
0.266
1.106
1.325
1.107
0.314
0.498
0.426
8.534
0.725
2.15
1
4
8.026
0.626
2.087

ACCEPTED MANUSCRIPT
1
5
9.053
0.543
2.015
0.259
0.259
1.322
0.367
Figure 8: Scores bar graph for mofezolac (1) and 3, 14-15 (Galmof Galmof and Galmof ) in COX-
0
,
5,
11
1
:mofezolac crystal.
Among the three new compounds, 3 appear to better respond to the searched compound. Then, the
chemical and metabolic stability of 3 was accomplished.
2
.10. In vitro chemical and metabolic stability of 3 (Galmof0)
The chemical stability of 3 was determined in three different buffer simulating the gastric fluid (pH =
.0; 0.1M HCl), blood and gut district (pH = 7.3 and 8.8, respectively), whereas its metabolic stability
1
was evaluated in rat liver S9 fraction. It is noteworthy that it remained unchanged in all the
investigated conditions, and specifically 60 min for the chemical stability and 30 min for the metabolic
stability.

ACCEPTED MANUSCRIPT
2
.
Conclusion
In conclusion, with this work novel compounds named Galmofs (3, 14-15) were prepared in good to
fair yields by linking 1 (mofezolac) and a galactose molecule with or without a spacer of different
length. They were opportunely designed to allow mofezolac to cross BBB and exert its selective CNS
COX-1 inhibition action. Modeling accomplished studies highlight the interactions between COX-1
active site and the Galmofs and justify the difference in potency and selectivity of the three novel
compounds with respect to mofezolac. In particular, the lack (3) or presence (14-15) of a spacer with
different length changes the orientation of the entire molecule within the COX-1 catalytic cavity due to
new stronger interactions in the case of 3 and 15 and than 14.
In particular, 3 (Galmof ) was found to be a COX-1 inhibitor (COX-1 IC = 0.27 µM and COX-2 IC
50
0
50
=
3.1 µM) with a selectivity index of 11.5, showing both a good chemical (pH = 1.0, 7.3 and 8.0) and
metabolic (rat liver S9 fraction) stability, resulting capable to cross Caco-2 cell monolayer. Moreover,
in the neuroinflammation model of LPS-stimulated BV2 microglial cell, Galmof (3) powerfully
0
inhibits PGE release higher than mofezolac (1), thus demonstrating to be a good candidate for further
2
investigations in neurological and neurodegenerative diseases with a marked inflammatory component.

ACCEPTED MANUSCRIPT
4
. Experimental Section
.1. Reagents and procedures - Chemistry
4
1
13
Melting points taken on an Electrothermal apparatus were uncorrected. H NMR and C NMR spectra
were recorded on a Varian-Mercury 300 MHz, Bruker-Aspect 3000 console 500 MHz spectrometer, or
Bruker 600 MHz, chemical shifts are reported in parts per million (δ). FT-IR spectra were recorded on
a Perkin-Elmer 681 spectrometer. Thin-layer chromatography (TLC) was performed on silica gel
sheets with fluorescent indicator. The spots on the TLC were visualized under ultraviolet light.
Chromatography was conducted by using silica gel 60 with a particle size distribution of 40−63 µm and
2
30−400 ASTM. MS-ESI analyses were performed on Agilent 1100 LC/MSD trap system VL. All
synthesized compounds were analyzed by CHN or by HPLC analysis performed on an Agilent 1260
Infinity instrument equipped with a 1260 DAD VL+ detector, and their purity is higher than 95%.
4
.2. Materials
Tetrahydrofuran (THF) from a commercial source was purified by distillation (twice) from sodium
wire under nitrogen. Dichloromethane from a commercial source was distilled from CaH under
2
nitrogen atmosphere before its use. Standardized (2.50 M) n-butyllithium in hexane was purchased
from Aldrich Chemical Co., and its titration was performed with N-pivaloyl-o-toluidine. Modified
Micetich’s procedure was used to prepare 1 (mofezolac) [8]; in particular, solid CO
2
was replaced by
dry gaseous CO and yields raised from 20 to 79%. 5-Aminopentanoic acid (4) and 11-
2
aminoundecanoic acid (5) were purchased from Aldrich Chemical Co.. All other chemicals and
solvents of commercial grade were further purified by distillation or crystallization prior to use.
4
.3. Procedures for the preparation of galactopyranoside derivatives (2, 12 and 13)
The procedure to prepare 2 is described. 12 and 13 synthesis followed the same procedure used to
obtain 2. The exact amounts of reagents are reported in the Table 5.

ACCEPTED MANUSCRIPT
To a cold (ice-bath) solution of 2-(3,4-bis(4-methoxyphenyl)isoxazol-5-yl)acetic acid (1, mofezolac,
3
00 mg, 0.885 mmol) and 1,2:3,4-di-O-isopropylidene-D-galactose (230 mg, 0.885 mmol) in CH
2 2
Cl
(
5 mL), 4-(dimethylamino)pyridine (DMAP) (27 mg, 0.22 mmol), and N-(3-dimethylaminopropyl)-N′-
ethylcarbodiimide hydrochloride (EDC HCl, 680 mg, 3.56 mmol) were added in small aliquots under
argon atmosphere. The reaction mixture was stirred at room temperature for 24 h and then quenched by
adding distilled water. The aqueous phase was extracted three times with EtOAc. The combined
organic phases were dried over anhydrous Na SO , filtered, and the solvent removed under reduced
2
4
pressure. The product (375 mg, 73% yield) was isolated by column chromatography (silica gel;
CHCl /MeOH 95:5) of the reaction crude.
Table 5.
Main reagent(mol) 1,2:3,4-Di-O-isopropylidene-D-galactose (mol) DMAP (mol) EDC (mol)
3
a
1
(0.885)
0 (0.194)
1 (0.5)
, 10 and 11 were solubilized in CH
0.885
0.194
0.5
0.22
0.048
0.11
3.54
0.78
2
1
1
a
1
2
Cl
2
(5 mL).
4.3.1. 1,2:3,4-Di-O-isopropylidene-6-O-{2-[3,4-bis(4-methoxyphenyl)isoxazol-5-yl]acetyl}-D-
1
galactopyranose (2). H-NMR (300 MHz, CDCl δ): 7.41-7.36 (m, 2H, aromatic protons); 7.19-7.14
3
,
(
m, 2H, aromatic protons); 6.94-6.89 (m, 2H, aromatic protons); 6.85-6.81 (m, 2H, aromatic protons);
5
8
.53 (d, 1H, J = 4.95 Hz); 4.61 (dd, 1H, J = 8.0 Hz, J = 2.5 Hz); 4.34-4.25 (m, 2H); 4.18 (dd, 2H, J =
.0 Hz, J = 1.9 Hz); 4.05-3.99 (m, 1H); 3.83 (s, 3H OCH ); 3.80 (s, 3H, OCH ); 3.79 (s, 2H, CH ); 1.45
3
3
2
+
(
s, 6H, 2CH ); 1.32 (s, 3H); 1.31 (s, 3H). ESI-MS m/z (%): C H NO (M + Na) : 604.
3 31 35 10
4
.3.2. 1,2:3,4-Di-O-isopropylidene-6-O-(5-{2-[3,4-bis(4-methoxyphenyl)isoxazol-5-
1
yl]acetamido}pentanoyl)-D-galactopyranose (12). Isolated 71 mg, 54% yield. H-NMR (300MHz,
CDCl3, δ): 7.33-7.31 (m, 2H, aromatic protons); 7.14-7.13 (m, 2H, aromatic protons); 6.86-6.85 (m, 2H,

ACCEPTED MANUSCRIPT
aromatic protons); 6.80-6.77 (m, 2H, aromatic protons); 6.43-6.41 (bs, 1H, NH: exchanges with D O),
2
5
3
.49-5.47 (m, 1H); 4.57-4.55 (m, 1H); 4.29-4.23 (m, 3H); 4.18-4.12 (m, 2H); 3.99-3.97 (m, 1H); 3.86-
.66 (m, 1H); 3.77 (s, 3H, OCH ); 3.73 (s, 3H, OCH ); 3.61 (s, 2H, CH CONH); 3.26-3.22 (m, 1H,
3
3
2
NHCH ); 3.21-3.15 (m, 1H, NHCH ); 2.32-2.30 (m, 2H, CH COO); 1.61-1.56 (m, 2H); 1.52-1.47 (m,
2
2
2
1
3
2
H); 1.45 (s, 3H, CH ); 1.39 (s, 3H, CH ); 1.28 (s, 3H, CH ); 1.27 (s, 3H, CH ). C-NMR (150 MHz,
3 3 3 3
CDCl , δ): 173.1, 166.5, 162.7, 160.8, 160.3, 159.2, 130.9, 129.5, 121.3, 120.9, 117.2, 114.1, 113.7,
3
1
3
4
09.4, 108.6, 96.1, 77.2, 71.2, 70.8, 70.5, 70.4, 70.3, 70.2, 68.0, 65.9, 63.2, 61.9, 55.0, 54.99, 39.2,
+
3.7, 33.4, 28.3, 25.8, 25.7, 24.7, 24.2, 24.1, 21.9. ESI-MS: m/z (%): C H N O (M + Na) : 703.
3
6
44
2
11
.3.3. 1,2:3,4-Di-O-isopropylidene-6-O-(11-{2-[3,4-bis(4-methoxyphenyl)isoxazol-5-
1
yl]acetamido}undecanoyl)-D-galactopyranose (13). Isolated 95 mg, 25% yield. H-NMR (300 MHz,
CDCl3, δ): 7.38-7.36 (m, 2H, aromatic protons); 7.15-7.14 (m, 2H, aromatic protons); 6.90-6.89 (m, 2H,
aromatic protons); 6.84-6.82 (m, 2H, aromatic protons); 5.91-5.87 (bs, 1H, NH: exchanges with D O);
2
5
1
3
.55 (d, 1H, J = 5.0 Hz); 4.60 (dd, 1H, J = 7.9 and 2.3 Hz); 4.32 (dd, 1H, J = 5.0 and 2.3 Hz); 4.26 (dd,
H, J = 7.9 and 1.6 Hz); 4.25-4.13 (m, 1H, CH); 3.90-3.80 (m, 2H, 2 CH); 3.81 (s, 3H, OCH ); 3.79 (s,
3
H, OCH ); 3.64 (s, 2H, CH CONH); 3.23 (q, 2H, J = 6.6 Hz, NHCH ); 2.28 (t, 2H, J = 7.6 Hz,
3
2
2
CH COO); 1.62-1.54 (m, 2H); 1.48-1.42 (m, 2H); 1.52 (s, 3H, CH ); 1.44 (s, 3H, CH ); 1.32 (s, 3H,
2
3
3
1
3
CH
3
); 1.30-1.20 (m, 15H, 1CH
3
and 6 CH
2
). C-NMR (151 MHz, CDCl
3
, δ): 174.3, 166.4, 162.7,
1
7
61.1, 160.6, 159.5, 131.0, 129.7, 121.4, 121.0, 117.4, 114.4, 114.0, 109.4, 108.6, 96.3, 71.6, 70.7,
0.6, 68.1, 62.3, 55.24, 55.22, 51.4, 40.0, 34.2, 34.1, 29.4, 29.3, 29.2, 29.1, 26.8, 26.0, 25.9, 24.9, 24.3.
−
ESI-MS m/z (%): C H N O [M − H] : 763. ESI-MS-MS m/z (%): 503 (36), 294 (100), 264 (66).
4
2
56
2
11
+
+
HRMS (ESI-TOF) m/z: [M + Na] Calcd for C H N O Na 787.3782; Found 787.3775.
4
2
56
2
11
4
.4. Procedures for the preparation of 6 and 7. The procedure to synthesize 6 is described. 7
preparation followed the same procedure used to obtain 6.

ACCEPTED MANUSCRIPT
4
.4.1. Methyl 5-aminopentanoate hydrochloride (6). To a stirred solution 5-aminopentanoic acid
(
500 mg, 4.27 mmol) in anhydrous MeOH (5 mL) kept at -10°C, SOCl
2
(1.1 mL, d = 1.631 g/mL, 15
mmol) was dropwise added. The obtained reaction mixture was refluxed for 3h. Then, excess SOCl
2
and MeOH were distilled under reduced pressure, using a NaOH trap. The product was obtained as a
white solid in 79% yield (645 mg) and used in the following reaction without any further purification.
1
+
H-NMR (300 MHz, CDCl δ): 8.35-8.28 (bs, 3H, NH ); 3.67 (s, 3H, OCH ); 3.04 (m, 2H, CH NH );
3
,
3
3
2
2
2
.38 (t, 2H, CH CO); 1.8-1.73 (m, 4H, CH CH ).
2 2 2
4
.4.2. Methyl 11-aminoundecanoate hydrochloride (7). SOCl (7.6 mmol), 11-aminoundecanoic
2
1
acid (2.49 mmol), MeOH (5 mL). The product was isolated as a white solid (535 mg, 86% yield). H-
+
NMR (300 MHz, CDCl3, δ): 8.30-8.27 (bs, 3H, NH
3
: exchange with D
2
O), 3.64 (s, 3H, OCH
3
); 2.90-
2
.85 (m, 2H, NH
2
CH
2
); 2.45-2.26 (m, 2H, CH
2
CO); 1.71-1.56 (m, 2H, CH
2
CH CO); 1.42-1.41 (m, 2H,
2
CH CH NH ); 1.36-1.18 (m, 12H, 6CH ).
2
2
2
2
4
.5. Procedures for the preparation of derivatives 8 and 9. The procedure to synthesize 8 is
described. 9 preparation followed the same procedure used to obtain 8.
.5.1. Methyl 5-{2-[3,4-bis(4-methoxyphenyl)isoxazol-5-yl]acetamido}pentanoate
Diisopropyl-N-ethylamine (DIEA, 0.215 mL, 1.237 mmol) and methyl 5-aminopentanoate
hydrochloride (6) (100 mg, 0.60 mmol) were solubilized in anhydrous CH Cl (5 mL) and stirred at
°C for 1h. Then, this solution was dropwise added to a stirred solution of N,N’-
4
(8).
N-
2
2
0
dicyclohexylcarbodiimide (DCC, 170 mg, 0.825 mmol), 1-hydroxybenzotriazole monohydrate (HOBt
H O, 180 mg, 1.05 mmol) and 2-[3,4-bis(4-methoxyphenyl)isoxazol-5-yl]acetic acid (mofezolac) (200
2
mg, 0.59 mmol) in anhydrous CH Cl (20 mL) kept at 0°C. The reaction mixture was stirred for 19 h at
2
2
room temperature. Then, H O was added and the aqueous solution extracted with CH Cl . The
2
2
2
combined organic layers were washed with a sat. aqueous solution of K CO , dried over anhydrous
2
3
2 4
Na SO , and the solvent was removed under reduced pressure. Column chromatography of the crude

ACCEPTED MANUSCRIPT
residue (silica gel; EtOAc/Hexane = 3:7) allowed to isolated 8 (107 mg, 40% yield). FT-IR (KBr):
3
1
458, 3089, 2987, 2948, 2849, 1737, 1652, 1609, 1562, 1516, 1455, 1441, 1426, 1253, 1233, 1175,
-1 1
108, 1029, 1019, 949, 831, 729 cm . H-NMR (300 MHz, CDCl3, δ): 7.41-7.37 (m, 2H, aromatic
protons); 7.17-7.14 (m, 2H, aromatic protons); 6.92-6.89 (m, 4H, aromatic protons); 5.95-5.90 (bs, 1H,
NH: exchanges with D O); 3.83 (s, 3H, OCH ); 3.80 (s, 3H, OCH ); 3.67 (s, 2H, CH COONH); 3.65
2
3
3
2
(
s, 3H, OCH ); 3.27 (q, 2H, J= 6.9 Hz, NHCH ); 2.33 (t, 2H, J= 6.9 Hz, CH COO); 1.65-1.50 (m, 4H).
3 2 2
1
3
C-NMR (75 MHz, CDCl ): δ= 174.1, 166.8, 162.8, 161.4, 160.8, 159.7, 131.2, 130.0, 121.6, 121.2,
3
+
1
4
4
1
17.8, 114.6, 114.2, 55.5, 55.4, 39.7, 34.4, 33.6, 29.0, 22.2. ESI-MS: m/z (%): C H N O (M + Na) :
2
5
28
2
6
75.
.5.2. Methyl 11-{2-[3,4-bis(4-methoxyphenyl)isoxazol-5-yl]acetamido}undecanoate. (9). Methyl
1-aminoundecanoate hydrochloride (1.38 mmol), DIEA (2.6 mmol), EDC HCl (1.58 mmol), HOBt
H O (1.48 mmol), mofezolac (1.47 mmol), CH Cl (48 mL)]. Column chromatography (silica gel;
2
2
2
hexane/EtOAc = 8:2) of the crude residue afforded the product as a white solid (573 mg, 72% yield).
1
H-NMR (300 MHz, CDCl , δ): 7.41-7.26 (m, 2H aromatic protons); 7.25-7.13 (m, 2H aromatic
3
protons); 6.93-6.71 (m, 4H aromatic protons), 5.90-5.80 (bs, 1H, NH: exchanges with D O), 3.82 (s,
2
3
2
H, OCH ), 3.80 (s, 3H, OCH ), 3.67 (s, 2H, CH ), 3.65 (s, 3H, COOCH ), 3.23-3.21 (m, 2H, NHCH ),
3
3
2
3
2
.31-2.26 (t, 2H, CH
2
CO), 1.62-1.57 (m, 2H, CH
2
CH
2
CO), 1.49-1.45 (m, 2H, CH
2
CH
2
NH); 1.25-1.10
+
+
(
m, 12H, 6 CH
2
). ESI/MS m/z: C31
H
40
N
2
O
6
(M+Na) : 559. HRMS (ESI-TOF) m/z: [M+Na] Calcd for
+
C H N O Na : 559.2784. Found: 559.2781.
31
40
2
6
4
1
4
.6. Procedures for the preparation of acid derivatives 10 and 11. The procedure to synthesize
0 is described. 11 preparation followed the same procedure used to obtain 8.
.6.1. 5-{2-[3,4-Bis(4-methoxyphenyl)isoxazol-5-yl]acetamido}pentanoic acid (10). To a solution
of methyl 5-(2-(3,4-bis(4-methoxyphenyl)isoxazol-5-yl)acetamido)pentanoate (50 mg, 0.110 mmol) in
THF (2 mL) was dropwise added a solution of 0.5N LiOH (68 mg, 2.86 mmol, in 6 mL of H O) at
2

ACCEPTED MANUSCRIPT
room temperature. After 2 h, 1M HCl (5 mL) was added and the aqueous phase was extracted with
2 4
EtOAc. The combined organic layers were dried over anhydrous Na SO and the solvent evaporated
under reduced pressure affording the product as a white solid (30 mg, 63% yield). FT-IR (KBr): 3330,
-1
2
921, 1720, 1641, 1607, 1540, 1514, 1468, 1424, 1406, 1303, 1247, 1180, 1030, 952, 901, 831 cm .
1
H-NMR (300 MHz, CDCl δ): 7.37-7.34 (m, 2H, aromatic protons), 7.15-7.12 (m, 2H, aromatic
3
,
protons); 6.90-6.87 (m, 2H, aromatic protons), 6.83-6.81 (m, 2H, aromatic protons), 6.35-6.28 (bs, 1H,
NH: exchanges with D O), 3.81 (s, 3H, OCH ); 3.78 (s, 3H, OCH ); 3.67 (s, 2H, CH CONH); 3.26 (q,
2
3
3
2
1
3
2
H, J= 6.1 Hz, NHCH ); 2.37 (t, 2H, J= 6.6 Hz, CH COO); 1.61-1.41 (m, 4H, CH CH ). C-NMR (75
2 2 2 2
MHz, CDCl
3
, δ): 177.2, 166.9, 162.8, 161.4, 160.9, 159.7, 131.3, 130.1, 121.5, 121.1, 117.8, 114.6,
-
1
4
1
0
14.2, 55.6, 55.4, 39.7, 34.4, 33.3, 28.9, 21.9. ESI-MS: m/z (%): C24
H
26
N O
2 6
(M-H) : 437.
.6.2. 11-{2-[3,4-Bis(4-methoxyphenyl)isoxazole-5-yl]acetamido}undecanoic acid (11). Methyl
1-{2-[3,4-bis(4-methoxyphenyl)isoxazol-5-yl]acetamido}undecanoate (9) (0.64 mmol), THF (10 mL,
1
.5M LiOH solution 36 mL). The product was obtained as a white powder (261 mg, 78% yield). H-
NMR (600 MHz, CDCl , δ): 10.50-10.40 (bs, 1H, OH: exchanges with D O); 7.38 (d, 2H, J= 8.7 Hz,
3
2
aromatic protons); 7.15 (m, 2H, J= 8.6 Hz, aromatic protons); 6.90 (d, 2H, J= 8.6 Hz, aromatic
protons); 6.83 (d, 2H, J= 8.7 Hz, aromatic protons); 5.88 (bs, 1H, NH: exchanges with D O); 3.82 (s,
2
3
2
1
1
2
3
4
H, OCH
H, J= 7.5 Hz, CH
.42-1.19 (m, 10H, 5 CH ). C-NMR (75 MHz, CDCl , δ): 178.7, 174.1, 167.0, 166.8, 162.9, 161.3,
3
), 3.79 (s, 3H, OCH
3
); 3.65 (s, 2H, CH
2
CONH); 3.24 (q, 2H, J= 6.6 Hz, NHCH
2
); 2.28 (t,
2
CO H); 1.62-1.56 (m, 4H, CH
2
2
CH CH CO H); 1.47-1.44 (m, 2H, CH
2
2
2
2
CH NH);
2
1
3
2
3
60.9, 159.7, 131.2, 130.0, 121.4, 121.0, 117.7, 114.6, 114.2, 55.6, 55.3, 40.3, 34.4, 34.0, 29.5, 29.4,
−
9.3, 29.2, 29.1, 26.9, 24.8. ESI-MS m/z (%): C H N O [M-H] : 521. ESI-MS-MS m/z (%): 489 (4),
3
0
38
2
6
88 (50), 356 (54), 320 (100), 293 (9).
.7. Procedure for the preparation of Galmofs (3, 14 and 15).

ACCEPTED MANUSCRIPT
The procedure to synthesize 3 is described. 14 and 15 preparation followed the same procedure used to
obtain 3.
4
.7.1. 6-O-{2-[3,4-Bis(4-methoxyphenyl)isoxazol-5-yl]acetyl}-D-galactopyranose
(3).
Trifluoracetic acid (TFA, 1.3 mL, d = 1.47 g/mL, 16.7 mmol) was added to a solution of 1,2:3,4-di-O-
isopropylidene-6-O-{2-[3,4-bis(4-methoxyphenyl)isoxazol -5-yl]acetyl}-D-galactopyranose (440 mg,
0
.76 mmol) in anhydrous CH Cl (5 mL). The progress of the reaction was monitored by TLC. The
2 2
reaction mixture was stirred at room temperature for 24 h. Then, the solvent was distilled under
reduced pressure and the product isolated as a white solid (148 mg, 39% yield) by a column
o
1
chromatography (silica gel; CHCl
MH , CDCl , δ): 7.29-7.27 (m, 2H, aromatic protons); 7.26-7.05 (m, 2H, aromatic protons); 6.81-6.80
m, 2H, aromatic protons); 6.76-6.73 (m, 2H, aromatic protons); 5.50-5.35 (bs, 4H, OH: exchange with
D O); 5.38 (s, 1H, 1’H galactose); 4.60 (s, 1H, 3’H galactose); 4.29-4.22 (m, 2H, 2’H-4’H galactose);
3
/MeOH = 9:1) of the crude residue. M.p. 117-119 C. H-NMR (500
Z
3
(
2
4
.07-4.02 (m, 1H, 5’H galactose ); 3.84-3.81 (dd, 2H, 6’H galactose); 3.84 (s, 3H, OCH ); 3.73 (s, 3H,
3
1
3
OCH ); 3.74 (s, 2H, OCH ). C-NMR (75 MHz, CDCl δ): 168.1, 161.9, 161.8, 160.7, 160.5, 159.3,
3
2
3,
1
31, 129.7, 121.2, 120.8, 117.5, 114.3, 113.9, 72.4, 55.15, 31.6, 29.6. ESI-MS m/z (%): C H NO (M
25 27 10
+
+
Na) : 524.
4
.7.2. 6-O-(5-{2-[3,4-Bis(4-methoxyphenyl)isoxazol-5-yl]acetamido}pentanoyl)-D-
galactopyranose (14). [TFA (4 mmol), 12 (0.10 mmol), CH Cl (2 mL)]. The product was isolated as a
yellow solid (35 mg, 58%). M.p. 85-87 C. α/β = 37/63. H-NMR (500MH , CDCl δ): 7.36-7.28 (m,
2
2
o
1
Z
3,
2
6
H, aromatic protons); 7.20-7.18 (m, 2H, aromatic protons); 6.98-6.90 (m, 2H, aromatic protons); 6.90-
.84 (m, 2H, aromatic protons); 5.65-5.40 (bs, 5H, OH and NH: exchange with D O); 5.20-5.18 (m, 1’-
2
H, α anomer); 5.15-5.13 (m, 1’-H, β anomer); 4.30-4.27 (m, 1H, 3’H galactose); 4.10-4.09 (m, 2H,
’H-4’H galactose); 3.85-3.83 (m, 2H, 6’H galactose); 3.76 (s, 3H, OCH ); 3.73 (s, 3H, OCH ); 3.53-
2
3
3