Novel endotoxin-sequestering compounds with terephthalaldehyde-bis- guanylhydrazone scaffolds

Article abstract of DOI:10.1016/j.bmcl.2005.11.059

We have shown that lipopolyamines bind to the lipid A moiety of lipopolysaccharide, a constituent of Gram-negative bacterial membranes, and neutralize its toxicity in animal models of endotoxic shock. In an effort to identify non-polyamine scaffolds with

Full text of DOI:10.1016/j.bmcl.2005.11.059

Bioorganic & Medicinal Chemistry Letters 16 (2006) 1305–1308
Novel endotoxin-sequestering compounds with
terephthalaldehyde-bis-guanylhydrazone scaffolds
Kriangsak Khownium, Stewart J. Wood, Kelly A. Miller, Rajalakshmi Balakrishna,
Thuan B. Nguyen, Matthew R. Kimbrell, Gunda I. Georg* and Sunil A. David*
Department of Medicinal Chemistry, University of Kansas, 1251 Wescoe Hall Drive, Lawrence KS 66045-7582, USA
Received 24 October 2005; revised 16 November 2005; accepted 17 November 2005
Available online 11 January 2006
Abstract—We have shown that lipopolyamines bind to the lipid A moiety of lipopolysaccharide, a constituent of Gram-negative
bacterial membranes, and neutralize its toxicity in animal models of endotoxic shock. In an effort to identify non-polyamine scaf-
folds with similar endotoxin-recognizing features, we had observed an unusually high frequency of hits containing guanylhydrazone
scaffolds in high-throughput screens. We now describe the syntheses and preliminary structure–activity relationships in a homolo-
gous series of bis-guanylhydrazone compounds decorated with hydrophobic functionalities. These first-generation compounds bind
and neutralize lipopolysaccharide with a potency comparable to that of polymyxin B, a peptide antibiotic known to sequester LPS.
ꢀ 2005 Elsevier Ltd. All rights reserved.
Despite tremendous advances in antimicrobial chemo-
therapy, the incidence and mortality due to Gram-nega-
tive sepsis continue to escalate worldwide,1,2 and the
burden on the US health care system ascribable to sepsis
has been estimated at about $16.7 billion annually.3 The
primary trigger in the pathogenesis of the Gram-nega-
tive septic shock4 syndrome is lipopolysaccharide
(LPS), otherwise termed endotoxin, a constituent of
the outer membrane of all Gram-negative bacteria.5 To-
tal synthesis of the glycolipid moiety of LPS, called lipid
A (Fig. 1), has established that this is the toxic center of
LPS,6,7 and thus presents a logical therapeutic target. In-
deed, Phase II clinical trials have just been successfully
concluded on E55648–10 (Eritoranꢁ), a tetraacyl part-
structure of lipid A, which acts as a receptor antagonist
of the toll-like receptor-4 (TLR-4), the principal recogni-
tion molecule for LPS in mammalian cells.11
OH
O
P
O
O
O
HO
HO
O
O
HO
O
NH
O
P
NH
OH
OH
O
O
O
O
O
HO
O
O
HO
O
O
10
10
10
10
10
12
Figure 1. Structure of lipid A, the toxic center of LPS.
ing terminal, protonatable cationic groups positioned
˚
optimally (N–N distance: ꢀ14 A) so as to be able to
simultaneously interact with the glycosidic phosphates
on lipid A,12–18 as well as appropriately positioned apo-
lar moieties to enable hydrophobic interactions with the
polyacyl domain of lipid A. Noteworthy examples of
such molecules displaying potent in vitro and in vivo
LPS-sequestering properties are acyl-polyamines,
affording protection in animal models of sepsis.12–14,17,18
Our approach has been to target circulatory LPS using
small molecules.12 The bis-anionic, amphiphilic nature
of lipid A (see Fig. 1) enables it to interact with a variety
of bis-cationic hydrophobic ligands, and we have found
that linear bis-cationic amphipathic molecules possess-
While members of the acylpolyamines undergo exhaus-
tive preclinical characterization, we have been keen on
exploring novel non-polyamine scaffolds and, employing
an automated rapid-throughput screen19 on focused li-
braries, we had recently identified several bis-guanylhyd-
razones as potent LPS binders.20 However, binding to
lipid A mediated via electrostatic interactions15,16 alone
Keywords: Lipopolysaccharide;
guanylhydrazones.
Endotoxin;
Sepsis;
Bis-
*
Corresponding authors. Tel.: +1 785 330 4316; fax: +1 785 330 4332
(S.D.); tel.: +1 785 864 4495; fax: +1 785 864 5326 (G.G.); e-mail
addresses: georg@ku.edu; sdavid@ku.edu
0960-894X/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.11.059

1306
K. Khownium et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1305–1308
does not necessarily manifest in neutralization of endo-
toxicity, and we have shown that additional hydropho-
bic interactions with the polyacyl domain of the toxin
via long-chain hydrocarbon appendages are necessary
for true sequestration of the toxin.12,18 We now describe
the syntheses and preliminary structure–activity rela-
tionships in a homologous series of bis-guanylhydrazone
compounds decorated with hydrophobic functionalities.
These first-generation compounds bind and neutralize
LPS with a potency comparable to that of polymyxin
B, a peptide antibiotic known to sequester LPS.12
assay as described.18,19 Inhibition of LPS-induced nitric
oxide production (measured as nitrite) in murine J774
cells, and protection afforded by 4g against lethality in
a D-galactosamine-primed model of endotoxic shock in
mice were performed as published previously.17,18 The
inhibition of induction of NF-jB (a key transcriptional
activator of the innate immune system, leading to
uncontrolled cytokine release25,26 which ultimately leads
to multiple organ failure and the shock syndrome) was
quantified using human embryonic kidney 293 cells
cotransfected with TLR4 (LPS receptor), CD14 and
MD2 (co-receptors), available from InvivoGen, Inc.
(HEK-BlueTM, San Diego, CA), as per protocols provid-
ed by the vendor. Stable expression of secreted alkaline
phosphatase (seAP) under the control of NF-jB/AP-1
promoters is inducible by LPS, and extracellular seAP
in the supernatant is proportional to NF-jB induction.
seAP was assayed spectrophotometrically using an alka-
line phosphatase-specific chromogen at 620 nm using a
rapid-throughput, automated protocol employing a
Bio-Tek P2000 liquid handler (see Supplemental data).
The syntheses of the target compounds are summarized
in Schemes 1 and 2, and full experimental details are
provided in Supplemental data. 2,3-Dimethoxytereph-
thalaldehyde (1) was prepared according to the proce-
dure described by Kuhnert et al.21 by double directed
ortho-lithiation of 1,2-dimethoxybenzene with n-BuLi/
TMEDA under reflux in ether, followed by quenching
with DMF in THF. Treatment of 1 with boron tribro-
mide in CH2Cl2 afforded 2 in 86% yield.22,29 The tereph-
thalaldehydes 3a–3h were prepared by alkylation of 2
with the respective alkyl halides.23,30 The aromatic guan-
ylhydrazones 4a–4k were obtained by condensing the
terephthalaldehydes 1–3 with aminoguanidine HCl in
refluxing ethanol/HCl (Scheme 2).24,31 Pure amin-
oguanylhydrazones (as HCl salts) were isolated as crys-
talline products after cooling to 0 ꢂC and washing with
cold ether.
As mentioned earlier, we had observed an unexpected
preponderance of guanylhydrazones compared to that
of other cationic functionalities such as amines and
amidines amongst hits in a high-throughput screen
performed with a focused library20 comprising of mem-
bers all of which possessed the basic pharmacophore for
lipid A recognition.12,16 We surmised that the augment-
ed binding affinity is a consequence of not only stronger
salt-bridges between the ligand and the lipid A phos-
phate groups due to multiple H-bond donor atoms of
the guanylhydrazone functionality, analogous to the Ôar-
ginine fork’ hypothesis,27 but also because of greater
rigidity (compared to the highly flexible polyamines)
which could account for diminished entropic loss of
the free energy of binding.16 Indeed, in silico docking
of the scaffold shown in Scheme 2 to lipid A indicated
excellent charge complementarity between the guan-
ylhydrazone moieties and lipid A phosphates, with the
distance between the terminal charged groups being
The affinities of binding of 4a–4k to LPS were deter-
mined using an automated fluorescence displacement
O
O
i) BBr3/CH2Cl2
ii) 1M HCl/H2O
H
H
O
O
86%
OMe
OH
O
H
OMe
H
OH
1
2
˚
14.14 A, commensurate with the inter-phosphate dis-
H
RI or RCl, K2CO3/DMF
tance in lipid A12 (data not shown). We sought to exam-
ine if the introduction of additional alkyl functionalities
would confer true LPS-neutralizing activity and to
determine the optimal chain length of the alkyl group
corresponding to maximal biological potency. The con-
centrations of the bis-guanylhydrazones corresponding
to effective displacement of 50% of the bound fluores-
cent probe (ED50; relative affinity of binding) and of
the inhibition (IC50) of nitric oxide (NO) and NF-jB
are summarized in Table 1.
O
OR
(R=C2H5, C3H7 ... C14H29
(37%-65%)
)
H
OR
3a-3h
Scheme 1.
O
H
NH
NH2
x HCl
H2N
N
H
O
R
As shown in Figure 2, dependencies on alkyl chain
length are observed for binding affinity, as well as NO
and NF-jB neutralization. For binding affinity, the
optimal substituent appears to be C4H9, while in both
in vitro bioassays, the most favorable hydrophobic
group is C8H17. This discrepancy is due to the fact that
in the homogeneous fluorescent displacement assay19
performed in aqueous buffer, the ligand must be com-
pletely soluble; increasing hydrocarbon chain length
substituents results in a progressively retarded aqueous
EtOH/HCl
(44%-90%)
H
R
(R=H, OH, OCH3 ... OC14H29
)
H
N
NH2
NH
NH
N
N
H2N
N
R
H
x 2 HCl
R
4a-4k
Scheme 2.

K. Khownium et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1305–1308
Table 1. Summary of binding affinity and biological activity for
terephthalaldehyde-bis-guanylhydrazones 4a–4k
1307
75
50
25
0
H
N
NH2
NH
N
N
NH
x 2 HCl
H2N
N
R
H
R
Compound
R
ED50 (lM)
IC50 (lM)
NF-jB
1.23
164
59.1
175
35.6
0
200
400
800
NO
0.980
Dose of 4j (µg/25g mouse)
Polymyxin B (Ctrl)
1.67
34.6
43.7
42.3
22.8
12.7
29.9
9.44
7.82
13.2
39.5
>500
Figure 3. Dose-dependent protective effect of 4j in a murine model of
lethality induced by 200 ng LPS.
4a
4b
4c
4d
4e
4f
–H
–OH
24.4
26.4
28.0
–OCH3
–OC2H5
–OC3H7
–O-Allyl
–OC4H9
–OC5H11
–OC6H13
–OC8H17
–OC14H29
4.56
These results lend support to our strategy of incremen-
tally converting high-affinity LPS binders to true LPS
sequestrants by appending suitable hydrophobic func-
tionalities. Further, these data would suggest that a sys-
tematic exploration of the bis-guanylhydrazone scaffold
may yield useful anti-endotoxin molecules. It is gratify-
ing that these first-generation compounds should display
significant in vivo activity, given that the placement of
the hydrophobic groups in the 4 series is not optimal,
for we have shown earlier12–16,18 that as a consequence
of favorable steric properties in the ligand:lipid A com-
plex, substitution with long-chain hydrocarbon groups
at the ends of the ligand, rather than at the center, cor-
responds to a significantly higher biological activity. We
are presently planning the synthesis and evaluation of
such compounds.
8.38
9.64
4.15
3.54
4.67
3.45
27.6
61.4
4g
4h
4i
9.40
10.5
6.11
4.50
190
4j
4k
14.9
ED (Affinity)
50
IC50(NO)
IC50(NFκB)
100
10
Acknowledgments
This work was supported by NIH 1U01 AI 56476. K.K.
is grateful to the Royal Thai Government for a scholar-
ship from the Development and Promotion of Science
and Technology (DPST) Project. Coordinates of models
of the compounds docked on lipid A are available on
request.
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Figure 2. Relationship of alkyl chain length of 4a–4k with binding
affinity and in vitro LPS neutralization potency.
Supplementary data
solubility, resulting in higher ED50 values. In the NO
and NF-jB neutralization assays, the presence of fetal
bovine serum in the cell-culture medium ensures com-
plete and uniform solubility, and obviates this prob-
lem.14,18 It may also be pointed out that the NF-jB
IC50 values best correlate with inhibition of cytokine re-
lease ex vivo in human whole blood experiments which,
in turn, correspond best to in vivo protection in animal
models (unpublished data).
Supplementary data and can be found, in the online
version, at doi:10.1016/j.bmcl.2005.11.059.
References and notes
1. Martin, G. S.; Mannino, D. M.; Eaton, S.; Moss, M. N.
Engl. J. Med. 2003, 348, 1546.
2. Moss, M.; Martin, G. S. Intensive Care Med. 2004, 30,
527.
3. Angus, D. C.; Linde-Zwirble, W. T.; Lidicker, J.; Cler-
mont, G.; Carcillo, J.; Pinsky, M. R. Crit. Care Med. 2001,
29, 1303.
4. Taveira da Silva, A. M.; Kaulbach, H. C.; Chuidian, F. S.;
Lambert, D. R.; Suffredini, A. F.; Danner, R. L. N. Engl.
J. Med. 1993, 328, 1457.
5. Rietschel, E. T.; Kirikae, T.; Schade, F. U.; Mamat, U.;
Schmidt, G.; Loppnow, H.; Ulmer, A. J.; Za¨hringer, U.;
Seydel, U.; Di Padova, F.; Schreier, M.; Brade, H. FASEB
J. 1994, 8, 217.
Given that 4j was the most potent compound in this as-
say (Table 1), we elected to characterize this compound
in an in vivo D-galactosamine-primed mouse model of
endotoxic shock. It is worth noting that the cytokine re-
sponse to LPS in mice closely parallels that of humans28
and is thus a well-established confirmatory assay.17,18
The administration of graded doses of 4j resulted in a
dose-dependent decrease in lethality in mice challenged
with supralethal (200 ng/mouse; LD100: 100 ng/mouse)
doses of LPS (Fig. 3).

1308
K. Khownium et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1305–1308
6. Galanos, C.; Luderitz, O.; Rietschel, E. T.; Westphal, O.;
¨
26. Shakov, A. N.; Collart, M. A.; Vassalli, P.; Nedospasov,
S. A.; Jongeneel, C. V. J. Exp. Med. 1990, 171, 35.
27. Calnan, B. J.; Tidor, B.; Biancalana, S.; Hudson, D.;
Frankel, A. D. Science 1991, 252, 1167.
Brade, H.; Brade, L.; Freudenberg, M. A.; Schade, U. F.;
Imoto, M.; Yoshimura, S.; Kusumoto, S.; Shiba, T. Eur.
J. Biochem. 1985, 148, 1.
7. Imoto, M.; Yoshimura, H.; Kusumoto, S.; Shiba, T. Proc.
Jpn. Acad. Sci. 1984, 60, 285.
8. Hawkins, L. D.; Christ, W. J.; Rossignol, D. P. Curr. Top.
Med. Chem. 2004, 4, 1147.
9. Rossignol, D. P.; Lynn, M. J. Endotoxin Res. 2002, 8, 483.
10. Lynn, M.; Rossignol, D. P.; Wheeler, J. L.; Kao, R. J.;
Perdomo, C. A.; Noveck, R.; Vargas, R.; D’Angelo, T.;
Gotzkowsky, S.; McMahon, F. G. J. Infect. Dis. 2003,
187, 631.
11. Beutler, B.; Poltorak, A. Drug Metab. Dispos. 2001, 29,
474.
12. David, S. A. J. Mol. Recognit. 2001, 14, 370.
13. Blagbrough, I. S.; Geall, A. J.; David, S. A. Bioorg. Med.
Chem. Lett. 2000, 10, 1959.
14. Burns, M. R.; Wood, S. J.; Miller, K. A.; Nguyen, T.;
Cromer, J. R.; David, S. A. Bioorg. Med. Chem. 2005, 13,
2523.
15. David, S. A.; Bechtel, B.; Annaiah, C.; Mathan, V. I.;
Balaram, P. Biochim. Biophys. Acta 1994, 1212, 167.
16. David, S. A.; Mathan, V. I.; Balaram, P. J. Endotoxin Res.
1995, 2, 325.
17. David, S. A.; Silverstein, R.; Amura, C. R.; Kielian, T.;
Morrison, D. C. Antimicrob. Agents Chemother. 1999, 43,
912.
18. Miller, K. A.; Suresh Kumar, E. V. K.; Wood, S. J.;
Cromer, J. R.; Datta, A.; David, S. A. J. Med. Chem.
2005, 48, 2589.
19. Wood, S. J.; Miller, K. A.; David, S. A. Comb. Chem.
High Throughput Screening 2004, 7, 239.
20. Wood, S. J.; Miller, K. A.; David, S. A. Comb. Chem.
High Throughput Screening 2004, 7, 733.
21. Kuhnert, N.; Rossignolo, G. M.; Lopez-Periago, A. Org.
Biomol. Chem. 2003, 1, 1157.
22. Akine, S.; Taniguchi, T.; Nabeshima, T. Tetrahedron Lett.
2001, 42, 8861.
23. van Otterlo, W. A. L.; Ngidi, E. L.; de Koning, C. B.
Tetrahedron Lett. 2003, 44, 6483.
24. Ulrich, P.; Cerami, A. J. Med. Chem. 1984, 27, 35.
25. Ishikawa, Y.; Mukaida, N.; Kuno, K.; Rice, N.; Okam-
oto, S.; Matsushima, K. J. Biol. Chem. 1995, 270, 4158.
28. Copelan, S.; Warren, H. S.; Lowry, S. F.; Calvano, S. E.;
Remick, D. Clin. Diagn. Lab Immunol. 2005, 12, 60.
29. 2,3-Dihydroxyterephthalaldehyde (2).22 To a solution of 1
(400 mg, 2 mmol) in anhydrous CH2Cl2 (1 mL) at rt was
added dropwise a 1.0 M boron tribromide solution in
anhydrous CH2Cl2 (20 mL, 20 mmol). The reaction mix-
ture was stirred for 1 h, followed by quenching of the
reaction by addition of 1 M of aq HCl (20 mL). The
mixture was extracted with CH2Cl2 (3· 100 mL). The
combined organic extracts were washed with brine (2·
50 mL) and dried over Na2SO4 and the solvent was
removed under reduced pressure. The resulting viscous
liquid was purified by flash column chromatography (1%
MeOH, 1% AcOH and CH2Cl2) on silica gel to obtain
compound 2 (239 mg, 86%) as a yellowish powder.
30. General procedure for the synthesis of 3a–3h.23 To a
solution of 2 (10 mg, 0.06 mmol) in anhydrous dimethyl-
formamide (4 mL) were added the respective alkyl halides
(25 equiv) [C2H5I, C3H7I, C3H5I, C4H9Cl, C5H11I,
C6H13Cl, C8H17Cl, and C14H29Cl] and K2CO3 (67 mg,
0.48 mmol). The reaction mixture was stirred for 15 h at
rt. The reaction was quenched with water (5 mL), and the
resulting solution was extracted with CH2Cl2, and the
combined organic extracts were washed with brine and
dried over Na2SO4. The resulting viscous liquids were
purified by flash column chromatography (CH2Cl2/hex-
ane = 6:4) to obtain the terephthalaldehyde analogs 3a–3h
(3a, R=C2H5; 3b, R=C3H7; 3c, R = allyl; 3d, R=C4H9; 3e,
R=C5H11; 3f, R=C6H13; 3g, R=C8H17; 3h, R=C14H29) as
oils.
31. General procedure for the synthesis of 4a–4k.24 To a hot
solution of aminoguanidine hydrochloride (13.7 mg,
0.12 mmol) in EtOH (1 mL) was added a solution of the
respective terephthalaldehyde (1, 2, and 3a–3h, 0.5 equiv,
in 0.5 mL EtOH) and concd HCl (0.1 mL). The reaction
mixture was heated up to 80 ꢂC for 1 h, and then cooled to
0 ꢂC. The resultant precipitate was filtered and washed
with cold ether (2 mL) to give bis-guanylhydrazone
1
analogs 4a–4k. All H and 13C spectra, and mass spectral
data were consonant with their corresponding structures.