Structure-activity relationship in monosaccharide-based toll-like receptor 4 (TLR4) antagonists

Article abstract of DOI:10.1021/acs.jmedchem.7b01803

The structure-activity relationship was investigated in a series of synthetic TLR4 antagonists formed by a glucosamine core linked to two phosphate esters and two linear carbon chains. Molecular modeling showed that the compounds with 10, 12, and 14 carbo

Full text of DOI:10.1021/acs.jmedchem.7b01803

Subscriber access provided by - Access paid by the | UCSB Libraries
Article
Structure-activity relationship (SAR) in monosaccharide-
based Toll-like receptor 4 (TLR4) antagonists
Fabio Alessandro Facchini, Lenny Zaffaroni, Alberto Minotti, Silvia Rapisarda,
Valentina Calabrese, Matilde Forcella, Paola Fusi, Cristina Airoldi, Carlotta Ciaramelli,
Jean-Marc Billod, Andra Schromm, Harald Braun, Charys Palmer, Rudi Beyaert,
Roman Jerala, Grisha Pirianov, Sonsoles Martin Santamaria, and Francesco Peri
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b01803 • Publication Date (Web): 01 Mar 2018
Downloaded from http://pubs.acs.org on March 3, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
Structureꢀactivity relationship (SAR) in monosaccharideꢀbased Tollꢀlike receptor 4
(TLR4) antagonists
7
8
9
Fabio A. Facchini,a Lenny Zaffaroni,a Alberto Minotti,a Silvia Rapisarda,a Valentina
Calabrese,a Matilde Forcella,a Paola Fusi,a Cristina Airoldi,a Carlotta Ciaramelli,a Jeanꢀ
Marc Billod,b Andra Schromm,c Harald Braun,d Charys Palmer,e Rudi Beyaert,d Roman
Jerala,f Grisha Pirianov,e Sonsoles MartinꢀSantamaria,b Francesco Peria*
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
aDepartment of Biotechnology and Biosciences, University of Milano-Bicocca, Piazza della
Scienza, 2; 20126 Milano (Italy).
bDepartment of Structural & Chemical Biology, Centro de Investigaciones Biologicas, CIB-
CSIC. C/ Ramiro de Maeztu, 9. 28040-Madrid (Spain).
cDivision of Immunobiophysics, Research Center Borstel, Parkallee 1-40, 23845 Borstel
(Germany).
dVIB-UGent Center for Inflammation Research; UGent Department for Biomedical Molecular
Biology, Unit of Molecular Signal Transduction in Inflammation; Technologiepark 927 / 9052
Ghent / Belgium.
eAnglia Ruskin Cambridge University, UK.
fDepartment of Synthetic biology and Immunology Kemijski institute; National Institute of
Chemistry; Hajdrihova 19 SI-1000 Ljubljana; Slovenija / Slovenia.
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 49
1
2
3
4
5
6
Abstract
7
8
9
The structureꢀactivity relationship was investigated in a series of synthetic TLR4 antagonists
formed by a glucosamine core linked to two phosphate esters and two linear carbon chains.
Molecular modeling showed that the compounds with 10, 12 and 14 carbons chains are
associated to higher stabilization of the MDꢀ2/TLR4 antagonist conformation than in the case of
the C16 variant. Binding experiments with human MDꢀ2 showed that the C12 and C14 variants
have higher affinity than C10, while the C16 variant did not interact with the protein.
The molecules, with the exception of the C16 variant, inhibited the LPSꢀstimulated TLR4 signal
in human and murine cells and the antagonist potency mirrored the MDꢀ2 affinity calculated
from in vitro binding experiments. FTꢀIR, NMR, and SAXS measurements suggested that the
aggregation state in aqueous solution depends on fatty acid chains lengths and that this property
can influence TLR4 activity in this series of compounds.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
ACS Paragon Plus Environment

Page 3 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
Introduction
7
8
9
TLRs are pattern recognition receptors (PRR) that recognize pathogenꢀassociated molecular
patterns (PAMPs). TLR4 is mainly expressed on haematopoietic cells including monocytes,
dendritic cells and macrophages.1 Lipopolysaccharide (LPS), lipooligosaccharide (LOS) and
lipid A from Gramꢀnegative bacteria are generally called endotoxin and are powerful TLR4
agonists.2 TLR4 responds rapidly to minute amounts of circulating LPS through and multistep
molecular recognition process, initiated by transfer of LPS monomers from aggregates in
solution to LPSꢀbinding protein (LBP), and subsequently to cluster of differentiation 14 (CD14),
and to myeloid differentiation factor 2 (MDꢀ2). MDꢀ2 is associated with TLR4 in MDꢀ2/TLR4
complexes on cell membrane. In the absence of agonist, the complex TLR4/MDꢀ2 is in
equilibrium between monomeric and dimeric species. Recent quantitative singleꢀmolecule
localization microscopy (SMLM) studies3 have shown that LPS binding to MDꢀ24 displaces the
equilibrium towards homodimeric complexes (TLR4/MDꢀ2/LPS)2.3,5 The homodimer transmits
the signal downstream through two distinct pathways. One pathway starts by recruitment of
myeloid differentiation primary response gene 88 (MyD88) and adapter myelin and lymphocyte
protein (MAL) (MyD88ꢀdependent pathways and production of a number of proꢀinflammatory
proteins), the other by the activation of TIRꢀdomainꢀcontaining adapterꢀinducing interferonꢀγ
(TRIF) (MyD88ꢀindependent pathways and production of interferons).6
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In addition to bacterial PAMPs, TLR4 can be also activated by damageꢀassociated molecular
patterns (DAMPs), endogenous agonists responsible for sterile inflammation, such as
fibronectins7, saturated palmitic acid,8 oxidized phospholipids9 or highꢀmobility group box 1
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 49
1
2
3
4
5
6
7
8
9
(HMGB1) protein10 have also been shown to activate TLR4. While different LPS chemotypes
share a conserved lipid A moiety with chemical determinants that ensure optimal interaction with
CD14 and MDꢀ2 (5 or 6 lipophilic fatty acid chains attached to a disaccharide backbone, and one
or two phosphate groups) DAMPs are chemically diverse molecules and the molecular
mechanism of TLR4 activation including the role of CD14 and MDꢀ2 in the sensing of these
molecules are not entirely understood. DAMPs have been implicated in many pathologies caused
by TLR4 activation including atherosclerosis11, rheumatoid arthritis,12 neuroinflammation13,
trauma14 and hemorrhage.15
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
These findings strongly support the idea that regulation of TLR4 activity appears as a potential
target for therapeutic control of a variety of inflammatoryꢀbased diseases. Manipulation of
TLR4ꢀmediated immune responses as a potential approach for pharmacological intervention has
been reported in the literature.16 For the last few years several TLR4 antagonists have been
evaluated in preclinical studies but only two drugs, E556417 (Eritoran, Eisai Inc.) and TAKꢀ24218
(Takeda Biological) progressed to clinical trials for treatment of sepsis, which have been
discontinued in different phases.19, 20
Efficient and selective TLR4 antagonists with a chemical structure simpler than lipid A are the
basis for development of novel TLR4 modulators. Lipid A consists of a 1,4ꢀdiphosphorylated diꢀ
glucosamine backbone to which variable lengths and numbers of fatty acid (FA) acyl chains are
covalently linked.2 Lipid X (Figure 1)21, a biosynthetic precursor of lipid A with TLR4
antagonist activity, has been considered a simplified monosaccharide scaffold for the
development of novel TLR4 modulators.
Our group developed the lipid X mimetic FP76, a glucosamine derivative with two phosphate
4
ACS Paragon Plus Environment

Page 5 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
groups and two myristic (C14) fatty acid (FA) chains, whose design was inspired by other
glucosamineꢀbased TLR4 modulators.22ꢀ24 FP7 is active in inhibiting in a doseꢀdependent way
human6 and murine25 TLR4 activation by LPS. Some preliminary observations from NMR
experiments suggest that FP7 interact with MDꢀ2, probably inserting FA chains into hydrophobic
binding cavity.6 This direct competition with LPS for MDꢀ2 binding is probably reinforced by
the capacity of FP7 to induce endocytosis of CD14 thus causing the absence of this receptor on
the plasma membrane.6 FP7 is active in blocking PR8 virus lethality that is mainly due to TLR4
overꢀstimulation by endogenous DAMPs (mainly oxidized phospholipids and HMGBꢀ1 protein)
derived from viral damage to lung tissue.25 In a proofꢀofꢀconcept experiments in support of this
in vivo mechanism, FP7 inhibited HMGBꢀ1 activation of dendritic cells.25 Other
monosaccharideꢀbased TLR4 modulators were developed and structureꢀactivity relationship
(SAR) studies showed that the length of FA chain is a critical factor determining the potency of
TLR4 antagonism or agonism.22, 26 The biological activity and the agonist/antagonist behavior on
TLR4 of lipid A variants and other amphiphilic glycolipids including FP7 is not only determined
by the interaction with MDꢀ2 but also by the aggregation state in solution. As LPS and lipid A,
FP7 is an anionic amphiphile with a low value of CMC (9 ꢁM)6. Even though the CMC value of
FP7 is higher than its IC50 (about 2 ꢁM in HEK cells assays), equilibrium between aggregates
and single molecules in solution is present in the concentration range in which FP7 is active.
It has been proposed for lipid A derivatives that the size and the 3D shape of aggregates
influences the TLR4 activity, lamellar aggregates being associated to antagonism and aggregates
with nonꢀlamellar cubic symmetry to agonism.27,28 While the last step of ligand presentation to
TLR4 and formation of the activated heterodimer (TLR4/MDꢀ2/ligand)2 are dominated by single
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 49
1
2
3
4
5
6
7
8
9
molecule interactions between the ligand and CD14 and MDꢀ2 receptors29, the early phases of
endotoxin (ligand) recognition by LBP are very likely influenced by the aggregation state of the
ligand.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We present here a structureꢀactivity relationship (SAR) study on synthetic FP7 variants differing
only for FA chains lengths (10, 12, 14 and 16 carbon atoms, Figure 1). In this study we will take
into account both the interaction with MDꢀ2 and the aggregation properties of the molecules.
Additionally, we show the relationship between the chemical structure of FP7 variants with
different fatty acid chains lengths and their effect on functional activity of TLR4 in different in
vitro cell models.
Figure 1. Chemical structures of lipid X and FP7 variants
Results
Computational design of FP7 variants as ligands of human MDꢀ2 and CD14.
Given our previous studies on the lipid X mimetic FP7 as ligand of TLR4/MDꢀ2 and CD14
proteins, with TLR4/MDꢀ2 antagonist activity,6 we were prompted to investigate the influence of
6
ACS Paragon Plus Environment

Page 7 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
the acyl chain length on the antagonist activity. To address this point, we designed three new FP7
derivatives with different fatty acid (FA) lengths: FP10 (C10), FP12 (C12), FP7 (C14), and
FP116 (C16). The ability of these ligands to bind to TLR4/MDꢀ2 complex and to CD14,
compared with FP7, was initially assessed through various computational techniques.
We firstly docked the ligands in the binding site of CD14 using AutoDock Vina. For all the four
ligands, docked poses inside the hydrophobic pocket were found. The obtained binding poses
were very similar for all the ligands (Figure S1A) with also very close favorable predicted
binding energies for the top poses (range from ꢀ6.5 kcal molꢀ1 to ꢀ5.9 kcal molꢀ1). Therefore, the
docking calculation showed that all four ligands are theoretically able to interact with CD14
inside its hydrophobic pocket and to engage in favorable interactions. In the most populated and
most favorable docked poses, one phosphate group is interacting with the NH groups of Arg72
and Val73, and with the OH group of Tyr82 (Figure S1B), while the other phosphate group is
exposed to the solvent. The FA chains are accommodated inside the hydrophobic pocket of
CD14 interacting with aliphatic residues, mainly Ala, Val, Leu, and Ile residues, and aromatic
Phe49 (details are depicted in Figure S1C). The results were in agreement with previous docking
studies of FP7 reported by us.30
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We performed the docking calculations of ligands FP7, FP10, FP12 and FP116 inside the
TLR4/MDꢀ2 complex in the antagonist conformation (Figure 2). For all the compounds,
favorable docked poses were found, with predicted binding energies, for the best ones, ranging
from ꢀ7.8 to ꢀ6.5 kcal molꢀ1. The polar head groups are placed at the rim of MDꢀ2 and the FA
chains go deep inside the hydrophobic pocket interacting with many hydrophobic residues,
namely Val24, Ala30, Ile32, Ile44, Ile46, Val48, Ile52, Leu54, Leu61, Ile63, Tyr65, Phe76,
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 49
1
2
3
4
5
6
Leu78, Ile80, Phe104, Val113, Ile117, Phe119, Phe121, Ile124, Tyr131, Val135, Phe147,
Leu149, Phe151, and Ile153 (Figure 2B).
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. A) General view of FP10 (in orange), FP12 (in yellow), FP7 (in green), and FP116 (in
violet) ligands are shown docked inside TLR4/MDꢀ2 (TLR4 is shown in black and MDꢀ2 in
grey). B) Detail of the MDꢀ2 hydrophobic pocket occupied by all the best docked poses for each
ligand (represented as lines). Hydrophobic residues mentioned in the text as interacting with the
FA chains of the ligands are represented in spheres. C) Detail of the polar interactions of the
8
ACS Paragon Plus Environment

Page 9 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
ligands inside the TLR4/MDꢀ2 system. Phosphate groups of the best docked poses of each ligand
and the MDꢀ2 residues with which they interact are represented in sticks.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Additionally, it was possible to observe more diversity in the predicted binding poses in
TLR4/MDꢀ2 than in the case of CD14. Results for FP7 were in agreement with those previously
reported in MDꢀ2 protein.6 In many poses, one of the phosphate groups was close to the hydroxyl
group of MDꢀ2 Tyr102 where it establishes hydrogen bonds, and the other one was often close to
MDꢀ2 Arg90 establishing hydrogen bonds and electrostatic interactions (Figure 2C). In some
docked poses, the phosphate groups were observed to interact with the backbone of residues
Phe119, Ser120, and Phe121. Both phosphate groups were often placed at the rim of MDꢀ2
where they are exposed to the solvent, in agreement with the reported Xꢀray crystallographic
complexes of TLR4/MDꢀ2 with glycolipids (for example, complex with Eritoran, PDBꢀID 2Z65,
or with lipid IVa, PDBꢀID 2E59). Two different orientations were also found: type A
(antagonistꢀlike binding mode), similar to that found for lipid IVa in PDBꢀID 2E59; and type B
(agonistꢀlike binding mode), similar to that found for E. coli lipid A in PDBꢀID 3FXI (Figure
S2). It is well known that these two ligands, lipid IVa and E. coli lipid A, bind to TLR4/MDꢀ2 in
a different manner, one being rotated 180° compared to the other one, leading to opposed
biological activities.
Selected binding poses were used as starting structures for reꢀdocking with AutoDock4 resulting
in predicted binding energies ranging from ꢀ4.6 kcal molꢀ1 to +4.3 kcal molꢀ1. Among the docked
solutions, the best poses (from ꢀ4.6 kcal molꢀ1 to ꢀ2.5 kcal molꢀ1) corresponded to binding poses
very similar to those obtained with AutoDock Vina (data not shown). The narrow binding energy
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 49
1
2
3
4
5
6
7
8
9
range did not permit to rank the ligands by predicted affinity, showing that the four ligands are
putative binders of the TLR4/MDꢀ2 system. Given that the main interactions (the polar ones) are
common to the four ligands, and that the MDꢀ2 pocket is big enough to host two longer FA
chains, from the docking calculations it was not possible to clearly correlate the subtle
differences in FA chain length with preferred ligand binding.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Stability of the predicted TLR4/MDꢀ2/ligand complexes was further studied by molecular
dynamics (MD) simulations. We selected two of the best binding poses for each ligand (Figure
S3): one type A (antagonistꢀlike binding pose), and one type B (agonistꢀlike binding pose), plus
two additional poses for compounds FP10 and FP7. Therefore, a total of eight 50 ns MD
simulation were run. We monitored the motion of MDꢀ2 over time and examined the RMSD and
RMS fluctuation per residues, as well as the motion of Phe126 side chain over time (Figure S4).
All the complexes showed stable ligandꢀreceptor interactions along the MD simulation time as
predicted by the docking calculations. In particular, in the MD simulation of the TLR4/MDꢀ
2/FP7 complex in the type A (antagonistꢀlike) binding pose, the Phe126 side chain moves around
its initial position staying largely exposed to the solvent in a conformation in agreement with the
Xꢀray crystallographic antagonist conformation of MDꢀ2 (Figure 3A).
To evaluate the relative orientation between the ligands and MDꢀ2, we arbitrarily defined two
vectors, one from the amide 〈ꢀcarbon atom to the ester 〈ꢀcarbon atom of the ligand, and another
one from the 〈ꢀcarbon of residues Pro78 to Thr105 of MDꢀ2 (Figure S5A). The angle between
these two vectors was plotted both over time and as a percentage of frames per 0.1 degree angle
range (Figure S6). It was observed that none of the ligands undergoes orientation flip during the
50ns simulations, all remaining in the orientation from the docking process. Interestingly, only in
10
ACS Paragon Plus Environment

Page 11 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
the case of the TLR4/MDꢀ2/FP116 complex with FP116 in the type A (antagonistꢀlike) binding
pose, the orientation of Phe126 side chain flips over (Figure 3B). We monitored this flipping
behavior along the MD simulations, for all the ligands, by arbitrary choosing two vectors, within
MDꢀ2, both starting from the 〈ꢀcarbon of residue Phe126 to, respectively, the phenyl Cꢀ4 atom of
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
the same residue and the 〈ꢀcarbon of residue Ser21 (data shown in SI figure S5B and S7). This
observation could suggest that FP116 is not able to efficiently retain an antagonist conformation
of MDꢀ2, thus pointing to a poor antagonist capacity.
Figure 3. Superimposition of different snapshots (one for each simulated nanosecond), from the
MD simulations of the TLR4/MDꢀ2/ligand complexes, colored from blue (t=0 ns) to red (t=50
ns). Only ligands (as lines) and Phe126 (in sticks) are made visible. Side chains of Phe126 from
the Xꢀray crystallographic structures have been superimposed to the snapshots to illustrate the
antagonist (dark blue, PDBꢀID 2E59) and agonist (black, PDBꢀID 3FXI) conformations of MDꢀ
2. A) TLR4/MDꢀ2/FP7 complex starting from the type A binding pose (antagonistꢀlike). B)
TLR4/MDꢀ2/FP116 complex starting from the type A binding pose (antagonistꢀlike).
Additionally, logP values of compounds FP10, FP12, FP7 and FP116 were computationally
calculated, ranging from approximately 4 to 10 with a linear distribution (Figure S8). The highest
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 49
1
2
3
4
5
6
7
8
9
logP value was obtained for FP116 indicating a high lipophilicity that might result in low water
solubility. This was in agreement with the lower acyl chain mobility as analysed by FTꢀIR
spectroscopy (see below). In any case, this did not interfere with the performance of the cell
assays. Summarizing, the computational studies assessed the ability of ligands FP7, FP10, FP12,
and FP116 to bind both CD14 and TLR4/MDꢀ2, pointing to the long FP116 acyl chain (C16) as
the maximum length bordering good (predicted) binding properties. The compounds were
therefore synthesized and tested.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Synthesis of FP variants
Compounds FP7, FP10, FP12, and FP116 were synthesized according to a divergent synthetic
strategy starting from the common precursor 4 (Scheme 1).
12
ACS Paragon Plus Environment

Page 13 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Scheme 1. Reagents and conditions: (a) CuSO4, TEA, Py:H2O, 0 °C, 30 min then TfN3, Py, 0 °Cꢀꢀ>rt,
O.N., quant.; (b) pꢀMeOPhCH(OMe)2, CSA, DMF dry, 40 °C, 8h, 68%; (c) TBSCl, imidazole, CH2Cl2
dry, rt, 1.5 h, 62%; (d) PPh3, THF/H2O, 60 °C, 2 h, quant.; (e) decanoic acid, EDC, DMAP, CH2Cl2 dry,
rt, 6 h, 79%; (f) lauroyl chloride, DMAP, Py dry, 0 °Cꢀ> rt, O.N., 79%; (g) palmitoyl chloride, DMAP, Py
dry, rt, O.N., 75%; (h) NaCNBH3, 4 Å MS, THF dry, rt, O.N., then HCl 1 M in dioxane until pH 2, 12ꢀ
70%; (i) (BnO)2PNiPr2, imidazolium triflate, CH2Cl2 dry, rt, 30 minutes, then mꢀCPBA, 0 °Cꢀꢀ> rt, O.N.,
51ꢀ56%; (j) I) H2, Pd/C, MeOH dry/CH2Cl2 dry, rt, O.N., II) Et3N, III) IRA 120 H+ resin, IV) IR 120 Na+,
84%ꢀquant.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Commercially available Dꢀglucosamine hydrochloride was converted into the intermediate 4 by
subsequent protection of Cꢀ4 and Cꢀ6 positions as pꢀmethoybenzylidene and the anomeric (Cꢀ1)
position as tertꢀbutyldimethylsislyl (TBDMS) ether.6 Intermediate 4 was then acylated in
positions C2 and C3 according to three different procedures, obtaining monosaccharides 8, 9,
and 10 with, respectively, C10, C12 and C16 carbon FA chains. Compound FP7, with C14
chains, was obtained similarly following a published procedure.6 Regioselective pꢀ
methoxybenzylidene ring opening as pꢀmethoxybenzyl (PMB) ether in Cꢀ6, followed by
phosphorylation of free hydroxyls in positions C1 and C4, and final deprotection of PMB ethers
gave final compounds as triethylammonium ions. Exchange of triethylamonium with sodium
(IR120 Na+ ion exchange resin) followed by reverseꢀphase purification gave final compounds
FP10, FP12 and FP116 with a purity ≥ 95%.
Aggregation properties of FP compounds
FTꢀIR studies
The mobility of the acyl chains is an important biophysical parameter of aggregated lipids.
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 49
1
2
3
4
5
6
7
8
9
Biological lipids typically show a temperatureꢀdependent phase transition from a highly ordered
gel (Lβ) phase of the hydrocarbon chains at low temperatures (indicated by an absorption peak
around 2850 cmꢀ1) to a liquidꢀcystalline (Lα) phase at higher temperatures (indicated by a
absorption peak around 2852 cmꢀ1). The phase transition temperature (Tc) is characteristic of the
chemical structure of the lipids. The FP compounds were analyzed by FTꢀIR spectroscopy to
determine the lipid phase in dependence of temperature. FP compounds with shorter acyl chains
(FP10, C10 and FP12, C12) were found to be in a fluid Lα phase with high mobility of their acyl
chains at all temperatures. FP compounds with a longer acyl chains (FP7, C14 and FP116, C16)
showed a biphasic behavior with a clear Lβ to Lα phase transition with Tc around 28.5 °C for
FP7 and around 40.2 °C for FP116 (Figure 4). Notably, the main phase transition of FP116 at
40.2 °C occurs along a broad temperature range and a second phase transition is observed around
65 °C. Thus, at a biological relevant temperature of 37 °C, FP10, FP12, and FP7 exhibit a fluid
membrane phase, whereas FP116 is still in a rigid membrane phase and requires much higher
temperatures for acyl chain melting to occur.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
14
ACS Paragon Plus Environment

Page 15 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 4. Acyl chain mobility of the aggregated FP compounds in dependence on temperature.
The infrared absorption around wavenumbers 2850 ꢀ 2852 cm−1 corresponds to the symmetric
stretching vibrations νs of the CH2 groups of the acyl chains. The wavenumbers indicated were
derived from the peak absorption of νs(CH2) determined upon constant heating of the samples.
Data are representative of two independent measurements.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
NMR studies
1
The comparison of the H NMR spectra recorded after compound dissolution in phosphate
buffer, pH 7.4, 25 °C (Figure S9) suggested a different aggregation state of the bioactive
compounds in solution in the µM concentration range. The 1H NMR spectrum acquired on a 100
µM FP7 sample (Figure S9A) clearly showed the presence of two set of signals, as can be
deduced by observing the spectral region between 5.5 and 5.2 ppm. In addition to a doublet of
doublet and a triplet, corresponding to H1 and H3 protons (*), also two broad resonances (§) are
present that can be assigned to aggregated species. This hypothesis was supported by the
comparison of spectrum S9A with FP7 spectra recorded at higher concentrations, in particular
125 µM (Figure S9B) and 250 µM (Figure S9C), where the gradual decrease in sharp signal (*)
intensity is associated with the increase of the broad resonance (§) ones, as expected as a
consequence of FP7 aggregation. A further confirmation was achieved through the acquisition of
relaxationꢀedited (Figure S9D) and diffusionꢀedited (Figure S9E) spectra, employed to partially
filter out resonances from high molecular weight and low molecular weight species
respectively.31, 32 Indeed, the spectrum acquired with the CPMG sequence (Figure S9D), edited
on the basis of relaxation times and thus highlighting the signals from low molecular weight
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 49
1
2
3
4
5
6
7
8
9
species, showed a decrease of broad signal (§) intensities compared to spectrum S9C; on the
contrary, in the diffusionꢀedited spectrum (Figure S9E), whose parameters were set up to erase
resonances from low molecular weight compounds, sharp resonances (*) disappeared. We can
conclude that, under our experimental condition, FP7 was present in solution as a mixture of
monomer/small aggregates and higher aggregated species (micelles), whose equilibrium changes
coherently to the variation of the nominal concentration of the molecule.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Instead, only one set of sharp signals was observed in the 1H NMR spectra recorded on FP10 and
FP12 solutions containing the compounds in the concentration range 100 µM ꢀ 1 mM.
Representative spectra acquired on 500 µM samples are depicted in Figure S9FꢀH for FP10 and
S9IꢀM for FP12. Furthermore, FP10 and FP12 resonances appear considerably narrower
compared to FP7 signals (Figure S9A). Collectively, these findings suggest an appreciably
higher solubility of FP10 and FP12 in aqueous buffer solution and thus a lower propensity to
form micelles.
1
A different behaviour can be described for FP116. All the HꢀNMR spectra acquired on this
compound present broad resonances and no sharp signals, expected for the free monomer. Thus,
in the range of tested concentrations (125 ꢀ 500 µM), FP116 is always present in an aggregated
form. Representative spectra acquired on a 250 µM FP116 sample are reported in Figure S9NꢀP.
SAXS studies
SAXS profiles were measured in dependence of temperature to obtain information on the
supramolecular organization of the molecules. The data are given in the range of the scattering
vectors relevant for structure assignment. All FP compounds showed isotropic scattering,
indicating no preference for a predominant orientation of the aggregates. FP10 and FP12 showed
16
ACS Paragon Plus Environment

Page 17 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
diffuse symmetric scattering curves dominated by the form factor, which is characteristic for
unilamellar aggregates with large interbilayer distance and probably owing to the negative
surface charge of the two adjacent phosphate groups that leads to a net electrostatic repulsion of
the bilayers (Figure 5). The scattering of FP10 shows two maxima at 5.58 nm and at 3.38 nm, the
latter could indicate the formation of interdigitated bilayers (3.38 nm). For FP12, a single peak is
observed at 20 °C (up to 35 °C, data not shown) indicating the formation of correlated
multilayers with a dꢀspacing of 4.14 nm, which is also consistent with the formation of an
interdigitated bilayer structure. In contrast, FP7 does show a different and more complex
scattering pattern, which indicates the occurrence of a nonꢀlamellar structure. The spacing
relationship exhibits clear similarity with cubic structures with a space group relationship of aQ
12.7 nm (not visible), aQ/3 (7.29 nm), aQ/5 (5.68 nm), aQ/8 (4.53 nm), and aQ/11 (3.86 nm),
agreeing most likely with space group Q212. The tendency to a nonꢀlamellar structure of FP7
however could explain the slightly lower antagonistic activity of FP7 compared to FP12. In
contrast to the above FP compounds, FP116 showed very weak scattering intensities hardly
displaying a form factor which can be explained by a much lower solubility observed for the
FP116 preparation and supports the results obtained by NMR.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 49
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Small angle Xꢀray diffraction of aggregates in solution for FP10 (A), FP12 (B), FP7
(C), and FP116 (D). Scattering vectors are indicated for temperatures between 20 °C and 80 °C.
Grey squares show enlargements of the relevant scattering vectors. The spacing of the diffraction
maxima is indicted as d = 1/s (nm).
Binding to MDꢀ2
Expression, purification and activity of hMD-2
Recombinant human MDꢀ2 (hMDꢀ2) was used in in vitro binding experiments. The functionality
of this protein is a crucial prerequisite to obtain reliable results representative of specific, highꢀ
affinity molecular recognition of ligands.33 Recombinant hMDꢀ2 was expressed in E. coli, Pichia
pastoris and mammalian HEK293 cells. hMDꢀ2 from different hosts was tested for its activity by
incubating HEK/hTLR4 cells with 100 ng/mL of LPS and then treating the cells with the mixture
of hMDꢀ2 and LPS. The TLR4ꢀdependent ILꢀ8 secretion by HEK/hTLR4 cells is indicative of
functional MDꢀ2. Figure 6 shows the lowest concentration of MDꢀ2 required for maximum
18
ACS Paragon Plus Environment

Page 19 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
activation of TLR4 (quantified by ILꢀ8 production), for each of the different expressed and
purified hMDꢀ2 proteins. hMDꢀ2 produced and purified from HEK293T displayed the highest
activity in stimulating the LPS/TLR4 inducible reporter at a concentration of 12 nM, followed by
the hMDꢀ2 expressed and purified from P. pastoris with its highest biological activity obtained
at a concentration of 15 nM. Finally, hMDꢀ2 purified from E. coli gave the lowest ILꢀ8
production at the concentration of 245 nM (Figure 6).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The difference in activity of hMDꢀ2 expressed in different hosts most likely reflects minor
differences in protein folding and/or glycosylation.34, 35 The higher expression yield and the good
activity prompted us to use hMDꢀ2 from P. pastoris in in vitro binding experiments with
synthetic compounds.
Figure 6. Activity of hMDꢀ2 expressed in different hosts. The figure shows the maximum
activation of TLR4 (quantified by ILꢀ8 production) at the lowest concentration of hMDꢀ2 under
the different expressed conditions (bacteria 245 nM, yeast 15 nM, and mammalian 12 nM).
Results are mean ± SEM from three parallels representative of at least three independent
experiments.
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 49
1
2
3
4
5
6
7
8
Binding studies were carried out by means of fours different techniques: two ELISAꢀtype plateꢀ
based assays with immobilized protein, a fluorescence displacement assay and Surface Plasmon
Resonance (SPR).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ELISA competition experiments with anti-hMD-2 antibody
Direct binding of LPS, FP7, FP10 and FP12 to MDꢀ2 was determined using a monoclonal
antibody that binds to free hMDꢀ2 but not to hMDꢀ2 bound to LPS.36 Monoclonal mouse antiꢀ
hMDꢀ2 (9B4) antibody specifically binds to an epitope close to the rim of the LPSꢀbinding
pocket of hMDꢀ2, available for recognition by the antibody only when the hMDꢀ2 pocket is
empty.
This assay detected a decrease in binding to MDꢀ2 in the presence of LPS (Figure 7A), similar to
what previously reported.37 A doseꢀdependent inhibition of antibody/MDꢀ2 interaction was
observed when adding FP7 and FP12, with a 90ꢀ95% decrease in binding obtained at
concentrations of FP7 and FP12 of 20 ꢁM (Figure 7A). 70% decrease in binding to hMDꢀ2 was
obtained with 20 ꢁM of FP10 (Figure 7A), and 20% decrease was obtained with 20 ꢁM of FP116
(Figure 7A). These data showed that while FP7 and FP12 bind hMDꢀ2 with high affinity, FP10
and FP116 are less potent ligands.
ELISA displacement experiment with immobilized hMD-2 and biotinylated LPS
The ability of FP compounds to displace LPS from the pocket of hMDꢀ2 was assessed by an
ELISA plateꢀbased assay. The synthetic molecules were added at increasing concentration to
hMDꢀ2 that was already incubated with biotinylated LPS. FP7 and FP12 were able to displace
biotinꢀLPS from hMDꢀ2 in a doseꢀdependent manner, with the highest displacement of 60ꢀ65%
20
ACS Paragon Plus Environment

Page 21 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
obtained at a concentration of 160 ꢁM (Figure 7B). FP10 and FP116, at a concentration of 160
ꢁM, gave a displacement of biotinꢀLPS of 20ꢀ30% (Figure 7B). As a control, LPS at a
concentration of 40 ꢁM gave the highest displacement of biotinꢀLPS of 70% (Figure 7B).
Fluorescence displacement assay
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
It has been previously shown that the fluorescent probe 1,1’ꢀBis(anilino)ꢀ4,4’ꢀbis
(naphthalene)−8,8’ disulfonate (bisꢀANS) binds to MDꢀ2 and is displaced by LPS.38 bisꢀANS
presumably binds the same MDꢀ2 binding site as lipid A and of other lipid Aꢀlike ligands. TLR4
ligands interacting with MDꢀ2 in a lipid Aꢀlike manner, are supposed to compete with bisꢀANS
and displace it from MDꢀ2. LPS, FP7, FP10 and FP12 caused a concentrationꢀdependent
decrease of bisꢀANS fluorescence, indicating competitive binding of FP7, FP10 and FP12 to
hMDꢀ2 (Figure 7C). FP116 induces only a modest decrease of bisꢀANS fluorescence at the
tested concentrations (Figure 7C), thus confirming that the lack of activity on cells could be
related to low affinity binding of this molecule to hMDꢀ2.
Surface Plasmon Resonance (SPR) analysis
SPR data with immobilized hMDꢀ2, showed direct interaction of the receptor with LPS (control)
and with the tested synthetic compounds. KD values derived from sensorgrams analysis were 3
ꢁM, 13.7 ꢁM, 22 ꢁM and 66.8 ꢁM for FP12, FP7, FP10 and FP116, respectively (Figure 7EꢀH).
SPR experimental curves optimal fitting was obtained by assuming 1:1 ligand/MDꢀ2 binding
stoichiometry.
Together, the results obtained from these in vitro cellꢀfree studies clearly indicate that FP7,
FP10, FP12 and FP116 directly interacted with MDꢀ2, with the same order of potency found in
human and murine cell experiments FP12(C12)>FP7(C14)>FP10(C10)>>FP116(C16).
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 49
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 7. Cellꢀfree binding studies on purified MDꢀ2 receptor (A) LPS, FP7, FP10 and FP12
prevent antiꢀhuman MDꢀ2 monoclonal antibody binding in a doseꢀdependent manner; (B) LPS,
FP7, FP10, FP12 and FP116 activity in competing with biotinꢀLPS for hMDꢀ2 binding; (C)
22
ACS Paragon Plus Environment

Page 23 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Fluorescence measurements show that LPS, FP7, FP10 and FP12 doseꢀdependently inhibit the
binding of bisꢀANS to MDꢀ2; (DꢀH) SPR analysis show direct interaction between LPS, FP10,
FP12, FP7, and FP116 and MDꢀ2; KD values are reported. Results are mean ± SEM from three
parallels representative of at least three independent experiments.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Cell experiments
Modulation of LPS-stimulated TLR4 signaling in HEK-Blue cells
In order to evaluate the influence the fatty acids length on the TLR4 antagonist activity,
molecules FP10, FP12, FP7, and FP116, constituting the homologous series with fatty acid chain
lengths C10, C12, C14, and C16, were first tested on HEKꢀBlue hTLR4 cells. These cells are
engineered to stably express the human receptors of the LPS recognition complex (hTLR4,
hMDꢀ2 and hCD14) and a reporter gene (SEAP) placed under the control of two TLR4ꢀ
dependent transcription factors (NFꢀκB and APꢀ1). Results from MTT assay revealed that all
compounds did not have a negative effect on cell viability at the concentration of 10 ꢁM used in
experiments (Figure S10, Supp. Info.). FP7, FP10 and FP12, but not FP116, inhibited in
concentrationꢀdependent manner the TLR4 signaling in HEKꢀBlue cells (Figure 8). FP7
displayed the expected antagonistic activity.6 FP10 and FP12 showed IC50 respectively higher
(5.45 µM) and lower (0.63 µM) than FP7 (2.0 µM) (Figure 8B). These results demonstrated the
efficacy of fatty acid chains lengths (C8, C10 and C12) of FP7 variants to negatively modulate
TLR4 signaling in HEKꢀBlue cells, the order of activity being: FP12(C12)>FP7(C14)
>FP10(C10)>>FP116(C16).
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 49
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 8. Doseꢀdependent inhibition of LPSꢀtriggered TLR4ꢀdependent NFꢀkB activation in
HEKꢀBlue hTLR4 cells by compounds FP7, FP10, FP12 and FP116. (A) HEKꢀBlue hTLR4 cells
were preꢀincubated with the indicated concentrations of compounds FP7, FP10, FP12 and FP116
and stimulated with LPS (100 ng/mL) after 30 minutes. Data were normalized to stimulation
with LPS alone and expressed as the mean percentage ± SEM of at least three independent
experiments. (B) Doseꢀresponse curves for compounds FP7, FP10 and FP12 in inducing the
TLR4ꢀdependent NFꢀκB reporter activity. Concentrationꢀeffect data were fitted to a sigmoidal 4
parameter logistic equation to determine IC50 values. Data points represent the mean of
percentage ± SEM of at least 3 independent experiments.
Modulation of LPS-stimulated TLR4 signaling in murine macrophages
24
ACS Paragon Plus Environment

Page 25 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Several TLR4 modulators mimicking lipid A have different effect on human and murine TLR4.
In certain cases when passing from murine to human TLR4/MDꢀ2/CD14 system the observed
agonistic effect switched to antagonistic effect of the compound of interest39. The speciesꢀ
specificity is due to differences between hMDꢀ2 and mMDꢀ2 binding regions that induce
different positioning of the same ligand thus causing different activity and, in some cases, switch
from agonism to antagonism. In this experiment we aimed to investigate the effect of
glucosamine derivatives in murine RAWꢀBlue macrophages. These cells are derived from RAW
264.7 and possess the same reporter gene present in HEKꢀBlue hTLR4 cells (SEAP). We first
verified the capacity of FP compounds to stimulate the TLR4 response in RAWꢀBlue cells and
we found all molecules inactive (Figure S11). When administrated before LPS, FP7, FP10 and
FP12 were active in inhibiting TLR4ꢀdependent NFꢀκB activation in RAWꢀBlue macrophages
(Figure 9A and 9B). Similarly to what happened in the case of human HEK cells, FP116 turned
out to be inactive as antagonist. Notably, FP12 was the most active antagonist compound (IC50 =
1.7 µM). The activity order of the tested compounds was found to be the same than in human
HEK cells: FP12(C12)>FP7(C14) >FP10(C10)>>FP116(C16).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 49
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 9. Activity of compounds FP10, FP12 and FP116 on RAWꢀBlue cells. (A) RAWꢀBlue
cells were preꢀincubated with increasing concentrations of synthetic compounds and then
stimulated with LPS (10 ng/mL, after 30 minutes). Data were normalized to response to LPS and
expressed as the mean percentage ± SEM of at least three independent experiments. (B) Doseꢀ
dependent inhibition curves of compounds FP7, FP10 and FP12. IC50 values in the table on the
bottom. Concentrationꢀeffect data were fitted to a sigmoidal 4ꢀparameter logistic equation to
determine IC50 values. Data points represent the mean of percentage ± SEM of at least 3
independent experiments.
Effect of FP variants on LPSꢀinduced TLR4 signaling in THPꢀ1 cells.
Haematopoietic TLR4 has been shown to play a critical role in any stage of inflammatory
process. Furthermore, immune competent cells use TLR4 signaling to sense danger molecules
26
ACS Paragon Plus Environment

Page 27 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
and produce proinflammatory proteins that initiate and amplify the inflammatory process. To test
the potential of FP variants to modulate TLR4 signaling pathways we utilised THPꢀ1 cells as an
in vitro model. Initially, we evaluated the effect of FP variants on THPꢀ1 cell viability. THPꢀ1
cells were exposed to different concentrations of FP variants (0ꢀ10 ꢁM) in the presence or
absence of LPS (100 ng/mL) for up to 24 h. Results from MTT assay demonstrated that FP
variants/LPS did not affect cell viability (Figure S12). To determine the effect of FP variants on
TLR4 signaling, next we analysed ILꢀ8 expression, a well known TLR4ꢀdependent
proinflammatory cytokine produced in THPꢀ1 cells in response to LPS. ELISA results clearly
demonstrated the potential of FP7 and FP12 at 10 ꢁM to inhibit LPS driven ILꢀ8 production
(Figure 10). In contrast, FP10 and FP116 had a modest or nonꢀsignificant impact on ILꢀ8
expression respectively (Figure 10). We have demonstrated that FP7 exerted a negative effect on
TLR4 signaling in different cell types (unpublished data). Following on from comparative
analysis based on TLR4ꢀdependent ILꢀ8 expression and tendency on the binding affinity with
MDꢀ2 of FP7 variants we found that the structural modification of FP12, but not FP10 and
FP116 is related to antagonistic activity of the compound. In support to this notion, we further
investigated the ability of FP12 to modulate second messengers in TLR4 signaling.
Immunoblotting data revealed that FP12 significantly downregulated p65 NFꢀκB
phosphorylation that was associated with a strong inhibition of the expression of additional
TLR4ꢀdependent cytokines, such as ILꢀ6 and ILꢀ1β in a doseꢀdependent manner (Figure 11).
These data clearly demonstrate that FP12 is a potent negative regulator of TLR4 signaling in
THPꢀ1 cells.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 49
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 10. The Effect of FP variants on LPS/TLR4 induced production of ILꢀ8 in THPꢀ1
cells. THPꢀ1 cells were preꢀtreated with FP variants (0 ꢀ 10 µM) for 1 h prior to LPS exposure.
Cells were then left to incubate 16 h further in the presence or absence of LPS (100 ng/mL). ILꢀ8
production was measured by ELISA. Results are displayed as mean concentration ± SD of three
independent experiments. Significant results are indicated as *P>0.05 **P>0.01 ***P>0.001 for
LPS vs Control and LPS vs FPs treated samples (Anova).
28
ACS Paragon Plus Environment

Page 29 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 11. FP12 negatively regulates p65 NFꢀκB phosphorylation and production of ILꢀ6
and ILꢀ1β in THPꢀ1 cells. THPꢀ1 cells were preꢀtreated with compound FP12 (0 ꢀ 10 µM) for 1
h prior to LPS exposure. Cells were then left to incubate 1 and 16 h further in presence or
absence of LPS (100 ng/mL). p65 NFꢀκB phosphorylation was determined in cell lysates using
Western Blot analysis and cytokine production was measured by ELISA after 16 h of LPS
exposure respectively. Results are displayed as mean concentration ± SD of three independent
experiments. Significant results are indicated as *P>0.05 **P>0.01 ***P>0.001 for LPS vs
Control and LPS vs FP12 treated samples (Anova).
Discussion and conclusions
The homologous series of FP glycolipids with fatty acid chain lengths varying from 10 to 16
carbon atoms were rationally designed as MDꢀ2 ligands and synthesized.
In a first set of in vitro experiments we aimed at studying the SAR of these molecules in binding
experiments with functional hMDꢀ2. For this purpose hMDꢀ2 expressed in yeast (P. pastoris),
29
ACS Paragon Plus Environment