Tethered thiazolidinone dimers as inhibitors of the bacterial type III secretion system

Article abstract of DOI:10.1016/j.bmcl.2009.01.047

Disruption of protein-protein interactions by small molecules is achievable but presents significant hurdles for effective compound design. In earlier work we identified a series of thiazolidinone inhibitors of the bacterial type III secretion system (T3SS) and demonstrated that this scaffold had the potential to be expanded into molecules with broad-spectrum anti-Gram negative activity. We now report on one series of thiazolidinone analogs in which the heterocycle is presented as a dimer at the termini of a series of linkers. Many of these dimers inhibited the T3SS-dependent secretion of a virulence protein at concentrations lower than that of the original monomeric compound identified in our screen.

Full text of DOI:10.1016/j.bmcl.2009.01.047

Bioorganic & Medicinal Chemistry Letters 19 (2009) 1340–1343
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Tethered thiazolidinone dimers as inhibitors of the bacterial type III
secretion system
a
a
a
a
a,b,c
Toni Kline a, , Kathleen C. Barry , Stona R. Jackson , Heather B. Felise , Hai V. Nguyen , Samuel I. Miller
*
a Department of Immunology, University of Washington, Seattle, WA 98195, USA
b Department of Microbiology, University of Washington, Seattle, WA 98195, USA
c Department of Medicine, University of Washington, Seattle, WA 98195, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Disruption of protein–protein interactions by small molecules is achievable but presents significant hur-
dles for effective compound design. In earlier work we identified a series of thiazolidinone inhibitors of
the bacterial type III secretion system (T3SS) and demonstrated that this scaffold had the potential to be
expanded into molecules with broad-spectrum anti-Gram negative activity. We now report on one series
of thiazolidinone analogs in which the heterocycle is presented as a dimer at the termini of a series of
linkers. Many of these dimers inhibited the T3SS-dependent secretion of a virulence protein at concen-
trations lower than that of the original monomeric compound identified in our screen.
Ó 2009 Elsevier Ltd. All rights reserved.
Received 26 December 2008
Revised 14 January 2009
Accepted 15 January 2009
Available online 20 January 2009
Keywords:
Type III Secretion
Thiazolidinone
Bacterial virulence
Antibiotic
Protein–protein interaction
A possible therapeutic solution to the problem of bacterial resis-
tance to existing antibiotics is to discover drugs that will block
pathogenic mechanisms rather than killing the infecting microbe.
These pathogenic mechanisms include secretion systems such as
the type III secretion system (T3SS) that deliver a variety of patho-
gen proteins using multicomponent oligomeric structures.
Although many of the secreted virulence proteins are species-spe-
cific, the secretion systems are more conserved across species,
indicating that disruption of such secretion systems is potentially
a broad-spectrum therapeutic strategy. Because the T3SS is not re-
quired for bacterial growth per se, this strategy might spare com-
mensals and limit bacterial resistance. In contrast, antibiotics
that inhibit microbial growth exert a strong selection pressure
for resistance.1 In recent years the T3SS machinery has become
an aggregate target for drug discovery.2–4
Previously our group identified a tris-aryl substituted 2-imino-
5-arylidenethiazolidin-4-one, compound 1, as a broad spectrum
inhibitor of Gram-negative bacterial secretion systems (Fig. 1).5
Expansion of this chemotype enabled us to define the functional
groups that could or could not be manipulated to synthetically
evolve potent new analogs. Modifications at the heterocycle amido
nitrogen were not only tolerated but gave rise to a series of novel
dipeptide-modified congeners, for example 2 and 3, that showed
enhanced potency and physiochemical properties.5,6 We consid-
ered the functional architecture of the T3SS and speculated that
these compounds might be fragments occupying only one of two
inter-monomer binding sites. Prompted by this hypothesis we syn-
thesized a bis-thiazolidinone dimer, 4.
We analyzed dimer 4 for inhibition of the T3SS in S. typhimuri-
um by monitoring secretion of a predominant substrate, SipA, into
culture supernatants. Supernatant proteins were TCA precipitated,
separated by SDS–PAGE and Western blotted with anti-SipA anti-
body. Evaluation of 4 showed a substantial increase in potency
over 1, with IC50 values of 5 lM versus 83 lM, respectively, but
the poor solubility of 4 precluded further biological characteriza-
tion of this compound. The significant decrease in the IC50
prompted us to prepare a panel of dimers, with the goal of improv-
ing the solubility of this compound and exploring the optimal in-
ter-thiazolidinone distance and juxtaposition. For this panel,
tethers were constructed that varied in length, flexibility, charge,
and pendent functional groups, providing divergent presentations
of the terminal thiazolidinones (Fig. 2). The linear analogs 5 and
6 expand and contract overall thiazolidinone-to-thiazolidinone
distance and give different placements of the amide function. In
contrast to the flexible amides, the para, meta, and ortho diamid-
ophenyl central cores rigidly enforce three distinct shapes (7–9).
Insertion of a proline (10) introduces two possible kinks in the
tether depending on the populations of cis and trans conforma-
tions. The five analogs that are cationic at physiological pH (11–
15) can be divided into the embedded and pendent classes. Mono-
amine 11 is highly flexible, whereas guanidine 12 will be some-
* Corresponding author. Tel.: +1 206 543 9587.
E-mail address: klinet@u.washington.edu (T. Kline).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.01.047

T. Kline et al. / Bioorg. Med. Chem. Lett. 19 (2009) 1340–1343
1341
H2N
N
N
N
O
H
O
H
N
N
H3CO
HO
H3CO
H3CO
HO
H3CO
N
N
S
H3CO
HO
H3CO
S
N
S
NH2
NH2
NH
O
O
O
O
O
NH
2
1
3
C
H2N
H3CO
OH
OCH3
N
N
N
S
H3CO
HO
H3CO
O
S
N
N
H
O
O
4
Figure 1. The original HTS hit thiazolidinone 1, two potent N-3 dipeptide analogs 2 and 3, and the dimer 4.
shape
length
OCH3
OH
m
N
N
N
N
p
NH
o
H3CO
HO
H3CO
HO
O
H3CO
HO
O
O
m
OCH3
OH
S
S
S
S
N
N
N
N
O
N
H
N
N
H
N
S
n
OCH3
O
O
O
O
O
N
S
N
OCH3
H3CO
H3CO
7
8
9
p
m
o
H3CO
O
N
10
5
6
m = 3, n = 5
m = 1, n = 2
N
H
O
OCH3
OH
OCH3
protonatable functions
pendent
embedded
N
N
H2N
HN
NH
OCH3
H3CO
S
S
S
S
N
O
OH
N
HO
N
H
N
N
OCH3
OH
H3CO
HO
O
O
H
N
OCH3
11
S
S
H3CO
N
O
N
N
H
O
O
OCH3
H3CO
N
14
N
OCH3
H3CO
NH
N
O
N
OH
HO
N
H
N
H
O
OCH3
H3CO
N
12
N
OCH3
H3CO
O
H
N
S
S
N
N
OH
HO
N
H
N
O
O
O
OCH3
OH
OCH3
H3CO
S
N
O
N
NH2
N
H3CO
S
15
N
O
OCH3
HO
N
13
H3CO
Figure 2. Dimeric analogs 5–15 use the tether to introduce spatial and functional group properties.
what more rigid, and piperazine 13 is likely to assume the shape
determined by a di-equatorial chair conformation. The linker in
compound 14 is flexible and projects the cationic function away
from the axis of the dimer. Dipeptide 15 incorporates the beneficial
sequence of the potent mono-thiazolidinone 25,6 into the motif of
4.
The analogs presenting pendent amino acids, 14 and 15, were
prepared by essentially linear routes (Scheme 3).
We evaluated these dimeric thiazolidinones for inhibition of
the T3SS in S. typhimurium by again analyzing secretion of the
SipA protein into culture supernatants. All of the dimeric com-
pounds, with the exception of 7, which was too insoluble to eval-
uate, were comparable to or slightly more potent than the original
hit compound 1 (Table 1). These data suggest that these com-
pounds may bind as 4-substituted thiazolidinone monomers, with
the additional ring and intervening tether being innocuous but
The syntheses of the dimers followed either a general end-to-
8
end7 (Scheme 1) or a center-to-outside (Scheme 2) strategy. In
all the analogs, the substituted thiazolidinone ring was assembled
by the method of Klika.9
EDCI, HOAt
O
O
DIEA,DMF
TZN
TZN
TZN
TZN
TZN
n
=
NH2
OH
+
N
H
O
S
4
5
6
m = 1 n = 5
m = 3 n = 5
m = 1 n = 2
m
n
m
or
PyBrop, DIEA
CH2Cl2-DMF
17 n = 5
19 n = 2
16 m = 1
18 m = 3
N
H
N
N
OCH3
HATU
DIEA,DMF
NH2
TZN
O
O
H3CO
+
16
OH
TZN
N
H
H2N
7
Scheme 1. Each completely substituted thiazolidinone terminates in either an amine or a carboxylic acid that reacts with the complementary function to form the dimer.

1342
T. Kline et al. / Bioorg. Med. Chem. Lett. 19 (2009) 1340–1343
O
(i) CbzGly
iBuOCOCl,NMM
THF
O
O
NH2
(i) PhNCS, MeOH
N
H
H2N
TZN
TZN
(ii) CH3OC(O)CH2Br/DIEA, THF
N
H
N
H
H2N
N
H
NH2
O
(ii) H2, Pd-C
EtOH-DMF
(iii) syringaldehyde, piperidine
TZN
20 m
8 m
Δ EtOH
21 o
9 o
O
N
H
N
H2N
(i) PhNCS, MeOH
(i)N-Cbz diaminoethane.HCl
CDI/DIEA/THF
O
CO2H
(ii) CH3OC(O)CH2Br/DIEA, THF
O
O
TZN
O
N
(iii) syringaldehyde, piperidine
(ii) H2, Pd-C
EtOH
N
N
H
Δ EtOH
NHCbz
NH2
10
22
23
(i) MsCl/NEt3, CH2Cl2
(ii) 0.5 M NaOH, THF
(i) H2, Pd-C, EtOH
(ii) PHNCS, CH2Cl2
H2N
(iii) CH3OC(O)CH2Br/DIEA, THF
H
N
NHCbz
Boc
N
TZN
TZN
OH
(iv} syringaldehyde, piperidine
(iii) Boc2O/DIEA, THF
CbzHN
NHCbz
CbzHN
CbzHN
Δ
EtOH
(v) 40% TFA/CH2Cl2
11
24
(i) SCNPbf
(i) H2, Pd-C, EtOH
(ii) N-Cbz diaminoethane.HCl
NPbf
(ii) PHNCS, DIEA,CH2Cl2
(iii) CH3OC(O)CH2Br/DIEA, THF
NH
EDCI/DIEA
TZN
TZN
CbzHN
NHCbz
TZN
NH2
N
N
H
N
H
N
H
H
DMF
(iv) syringaldehyde, piperidine
Δ
EtOH
12
25
(v) 50% TFA/CH2Cl2
(i) TBSCl/ NEt3, CH2Cl2
(i) (COCl)2/DMSO/NEt3
N
(ii) PHNCS/ DIEA, CH2Cl2
N
(ii) B(OAc)3H/ piperazine, CH2Cl2
NH2
HO
TZN
S
(iii) CH3OC(O)CH2Br/DIEA, THF
(iv) TBAF
N
O
(iii) syringaldehyde, piperidine
N
HO
Δ EtOH
13
26
Scheme 2. Each tether terminates in two free amino groups that are simultaneously assembled into the substituted thiazolidinone.
H2N NH
PbfHN NH
HN
HN
(i) PhNCS, MeOH
(ii) CH3OC(O)CH2Br/DIEA,
THF
N
N
(i) FmocArg(Pbf)/CDI/DIEA
THF
(i) syringaldehyde, piperidine
O
O
TZN
H
N
TZN
H
N
S
S
Δ EtOH
TZN
N
N
NH2
H2N
NHBoc
N
H
N
H
(iii) TFA
(ii) 20% piperidine/DMF
O
O
O
O
(iv) TFA, CH2Cl2
O
(iii) 16/CDI/DIEA, DMF
14
27
28
(i) HClO4/tBuOAc
O
(ii)Fmoc Orn(Alloc)/CDI/DIEA
O
O
(i) PhNCS, CH2Cl2
H
N
(i) 19/PyBrop/DIEA/CH2Cl2
TZN
TZN
TZN
THF
H2N
(ii) CH3OC(O)CH2Br/DIEA, THF
N
N
H
CO2H
N
CO2tBu
H2N
CO2H
H
H
(iii) 20% piperidine/DMF
(iii) syringaldehyde, piperidine
(ii) Pd(Ph3)4
Δ EtOH
1,3-dimethylbarbituric acid
AcOH, CH2Cl2
NH2
(iv)TFA
NHAlloc
29
NHAlloc
30
15
Scheme 3. The synthesis of 14 and 15.
Table 1
not overwhelmingly beneficial. Amide 4 is more potent than the
corresponding amine 11 or guanidine 12. This may indicate a role
for the carbonyl in a critical hydrogen bond and/or result from a
deleterious effect of cationic charge along the tether. The greater
potency of 5 compared with 6 would argue against the carbonyl’s
position as a critical feature. Indeed, it is the longest linear amide
5 and the most rigidly kinked ortho diaminobenzene amide 9, two
uncharged analogs, that distinguish themselves among the new
compounds by having potency significantly greater than 1. Over-
all, compound 4 remains the most potent of the dimers, and may
represent an optimum of shape, flexibility, and carbonyl place-
ment for the cognate binding site. Alternative binding sites along
the inter-protein interface are also possible and would be in
agreement with the lack of a single comprehensive structure–
activity trend among the dimers. The specific contribution of each
Inhibition of SipA secretion by the dimeric analogs of 1
Compound
IC50 (lM)
1
4
5
6
7
8
9
10
11
12
13
14
15
83
5
17
48
n.d.
80
22
55
65
47
44
48
49

T. Kline et al. / Bioorg. Med. Chem. Lett. 19 (2009) 1340–1343
1343
30.36, 30.48, 31.26, 33.44, 43.04, 47.03, 49.43, 60.62, 112.61, 121.91, 121.99,
125.51, 125.57, 128.07, 128.12, 129.28, 133.87, 135.72, 135.93, 142.90, 142.97,
151.90, 152.27, 152.64, 154.14, 154.25, 169.76, 170.27, 170.37. MS m/z 852
[M+H]+, 874 [M+Na]+. HRMS (m/z): [M+Na]+ Calcd for C44H45N5O9NaS2,
874.2551. Found 874.2551.
individual thiazolidinone ring to the activity of the dimer remains
undetermined, and we have no evidence that these rings are act-
ing in tandem. While a definitive identification of the thiazolidi-
none binding site(s) will be best determined by structural
biology, our results are consistent with these compounds inhibit-
ing protein–protein interactions along a large oligomeric interface.
8. 1-(4-(4-Oxo-2-(phenylimino)thiazolidin-3-yl)butyl)-3-(2-(4-oxo-2-(phenylimino)
thiazolidin-3-yl)ethyl)
2,2,4,6,7-pentamethyldihydrobenzofuran-5-sulfonyl
guanidine (12).Benzyl 4-aminobutylcarbamate (89 mg, 0.40 mmol) was
dissolved in CH2Cl2 (2 mL), 2,2,4,6,7-pentamethyldihydrobenzofuran-5-
sulfonyl isothiocyanate (125 mg, 0.40 mmol) was added, and the solution was
stirred for 16 h. The reaction mixture was partitioned between EtOAc (10 mL),
and water (5 mL). The organic phase was washed with 5 mL of, successively,
NaHCO3 and NaCl, dried, and concentrated in vacuo to the crude thiourea. This
Acknowledgments
The support of NIAID (U54 A105714) is gratefully acknowl-
edged. The investigators are members of the Northwest Regional
Center of Excellence for Biodefense and Emerging Infectious Dis-
eases Research.
was dissolved in DMF (3 mL) and diisopropylethylamine (35 lL, 0.20 mmol),
benzyl 2-aminoethylcarbamate HCl (28 mg, 0.14 mmol), and EDCI (35 mg,
0.12 mmol) were added. After stirring 48 h the solution was concentrated in
vacuo, dissolved in EtOAc (10 mL), and water (10 mL) was added. The organic
layer was washed with water (2 Â 10 mL) and NaCl solution (10 mL), dried, and
concentrated in vacuo. The residue was purified via silica gel chromatography
using a gradient from 0% to 80% EtOAc in CHCl3 to give 25 (65 mg, 0.09 mmol).
1H NMR (500 MHz, CDCl3, d): 1.43–1.48 (m, 10H), 2.03 (s, 3H), 2.50, (s, 3H), 2.57
(s, 3H), 2.91 (s, 2H), 3.14–3.27 (m, 4H), 5.05 (s, 4H), 7.30 (s, 10H). MS m/z 694.4
[M+H]+. Biscarbamate 25 (65 mg, 0.09 mmol) was dissolved in EtOH (5 mL), Pd–
C (80 mg) was added, and hydrogen gas was bubbled through the slurry for 2 h.
The resulting suspension was filtered through celite, rinsed with MeOH (5 mL),
and concentrated in vacuo. The residue was dissolved in CH2Cl2 (4 mL), and
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2009.01.047.
References and notes
phenyl isothiocyanate (28
0.57 mmol) were added. After 16 h, the solution was concentrated in vacuo and
redissolved in THF (4 mL). Diisopropylethylamine (80 L, 0.46 mmol) and
methyl bromoacetate (25 L, 0.27 mmol) were added, and the solution was
lL, 0.23 mmol) and diisopropylethylamine (100 lL,
l
1. Keyser, P.; Elofsson, M.; Rosell, S.; Wolf-Watz, H. J. Intern. Med. 2008, 264, 17.
2. Kauppi, A. M.; Andersson, C. D.; Norberg, H. A.; Sundin, C.; Linusson, A.; Elofsson,
M. Bioorg. Med. Chem. 2007, 15, 6994.
3. Kauppi, A. M.; Nordfelth, R.; Uvell, H.; Wolf-Watz, H.; Elofsson, M. Chem. Biol.
2003, 10, 241.
4. Negrea, A.; Bjur, E.; Ygberg, S. E.; Elofsson, M.; Wolf-Watz, H.; Rhen, M.
Antimicrob. Agents Chemother. 2007, 51, 2867.
5. Felise, H. B.; Nguyen, H. V.; Pfuetzner, R. A.; Barry, K. C.; Jackson, S. R.; Blanc, M.
P.; Bronstein, P. A.; Kline, T.; Miller, S. I. Cell Host Microbe 2008, 4, 325.
6. Kline, T.; Felise, H. B.; Barry, K. C.; Jackson, S. R.; Nguyen, H. V.; Miller, S. I. J. Med.
Chem. 2008, 51, 7065.
l
stirred for 16 h. EtOAc (5 mL) and NaHCO3 solution (5 mL) were added, the
organic layer was washed with NaCl (5 mL), dried and concentrated in vacuo.
The residue was purified via silica gel chromatography using a gradient from 0%
to 80% EtOAc in CHCl3 to give the thiazolidinone (50 mg, 0.07 mmol). 1H NMR
(300 MHz, CDCl3, d): 1.44 (s, 6H), 1.50–1.65 (m, 4H), 2.07 (s, 3H), 2.52 (s, 3H),
2.59 (s, 3H), 2.92 (s, 2H), 2.95 (br, 2H), 3.55 (br, 2H), 3.71–3.82 (m, 6H), 3.98 (br,
2H), 6.92 (t, J = 7.0 Hz, 4H), 7.14 (t, J = 7.3 Hz, 2H), 7.30–7.37 (m, 4H). MS m/z
776.3 [M+H]+. To a solution of the thiazolidinone (11 mg, .014 mmol) and
piperidine (25 lL, .253 mmol) in EtOH (3 mL) syringaldehyde (10 mg,
.055 mmol) was added and the solution was heated to 90 °C for 48 h. The
solution was concentrated in vacuo, and the residue was dissolved in CH2Cl2
(0.5 mL) and trifluoroacetic acid (0.5 mL), and stirred for 1 h. After removing the
volatiles in vacuo, the residue was purified via preparative HPLC using a gradient
of 10% to 40% B in 5 min, then 40% to 95% B in 20 min to give 12 (4.3 mg,
.005 mmol). 1H NMR (500 MHz, CDCl3, d): 1.68–1.86 (m, 4H), 3.65 (t, J = 5.5 Hz,
2H), 3.72 (s, 12H), 3.96 (t, J = 6.8 Hz, 2H), 4.16 (t, J = 5.5 Hz, 2H), 6.62 (s, 4H), 7.00
(d, J = 7.5 Hz, 2H), 7.06 (d, J = 7.5 Hz, 2H), 7.17–7.20 (m, 4H), 7.55 (s, 1H), 7.59 (s,
1H). 13C NMR (500 MHz, CD3OD, d): 25.82, 25.84, 27.15, 42.51, 43.41, 56.75,
56.78, 108.98, 109.02, 118.98, 122.39, 122.49, 125.55, 125.66, 126.15, 126.35,
130.48, 130.53, 133.00, 133.64, 139.79, 149.18, 149.48, 149.53, 152.27, 152.35,
157.77, 168.60. HRMS (m/z): [M+H]+ Calcd for C43H46N7O8S2, 852.2844. Found
852.2840.
7. 4,2-((2Z,5Z)-5-(4-hydroxy-3,5-dimethoxybenzylidene)-4-oxo-2-(phenylimino)
thiazolidin-3-yl)-N-(6-((2Z,5Z)-5-(4-hydroxy-3,5-dimethoxybenzylidene)-4-oxo-2-
(phenylimino)thiazolidin-3-yl)hexyl)acetamide (4)To a solution of 176 (24 mg,
0.053 mmol) in 0.5 mL DMF at 0 °C was added 166 (25 mg, 0.061 mmol), DIEA
(9.2 lL, 0.053 mmol), HOAt (7 mg, 0.053 mmol), and after 5 min, EDCI (16 mg,
0.053 mmol). The reaction mixture was allowed to warm to room temperature
stirred overnight. The crude reaction mixture was suspended in CHCl3 and
washed with H2O, 1 mM citric acid, and NaHCO3. The organic layer was dried
over Na2SO4 and concentrated in vacuo. The crude solid was purified via silica
gel chromatography using a gradient from 0% to 10% MeOH in CHCl3 to give 4
(5.3 mg, 0.006 mmol). 1H NMR (300 MHz, DMSO-d6, d): 1.28–1.40 (m, 4H), 1.40–
1.53 (m, 2H), 1.60–1.76 (m, 2H), 3.06–3.19 (m, 2H), 3.74 (s, 12H), 3.87 (t,
J = 7.5 Hz, 2H), 4.47 (s, 2H), 6.83 (s, 2H), 6.84 (s, 2H), 6.98 (d, J = 7.5 Hz, 2H), 7.03
(d, J = 7.5 Hz, 2H), 7.11–7.23 (m, 2H), 7.32–7.46 (m, 4H), 7.68 (s, 1H), 7.70 (s, 1H),
8.20 (t, J = 5.3 Hz, 1H), 9.25 (s, 1H), 9.27 (s, 1H). 13C NMR (500 MHz, DMSO-d6, d):
9. Klika, K. D.; Valtamo, P.; Janovec, L.; Suchar, G.; Kristian, P.; Imrich, J.; Kivela, H.;
Alfoldi, J.; Pihlaja, K. Rapid Commun. Mass Spectrom. 2004, 18, 87.