Stereochemical diversity of AI-2 analogs modulates quorum sensing in Vibrio harveyi and Escherichia coli

Article abstract of DOI:10.1016/j.bmc.2011.11.007

Bacteria coordinate population-dependent behaviors such as virulence by intra- and inter-species communication (quorum sensing). Autoinducer-2 (AI-2) regulates inter-species quorum sensing. AI-2 derives from the spontaneous cyclisation of linear (S)-4,5-d

Full text of DOI:10.1016/j.bmc.2011.11.007

Bioorganic & Medicinal Chemistry 20 (2012) 249–256
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Stereochemical diversity of AI-2 analogs modulates quorum sensing in Vibrio
harveyi and Escherichia coli
Fabio Rui a,b, João C. Marques a,b,c, Stephen T. Miller d, Christopher D. Maycock a,e, Karina B. Xavier a,b
,
M. Rita Ventura a,
⇑
a Instituto de Tecnologia Química e Biológica, Av. da República, Estação Agronómica Nacional, 2780-157 Oeiras, Portugal
b Instituto Gulbenkian de Ciência, Rua da Quinta Grande 6, 2780-156 Oeiras, Portugal
c Presently at Champalimaud Neuroscience Programme, Champalimaud Centre for the Unknown, Av. Brasília s/n, 1400-038 Lisboa, Portugal
d Department of Chemistry and Biochemistry, Swarthmore College, Swarthmore, PA 19081, USA
e Faculdade de Ciências da Universidade de Lisboa, Departamento de Química e Bioquímica, 1749-016 Lisboa, Portugal
a r t i c l e i n f o
a b s t r a c t
Article history:
Bacteria coordinate population-dependent behaviors such as virulence by intra- and inter-species
communication (quorum sensing). Autoinducer-2 (AI-2) regulates inter-species quorum sensing. AI-2
derives from the spontaneous cyclisation of linear (S)-4,5-dihydroxypentanedione (DPD) into two iso-
meric forms in dynamic equilibrium. Different species of bacteria have different classes of AI-2 receptors
(LsrB and LuxP) which bind to different cyclic forms. In the present work, DPD analogs with a new ste-
reocenter at C-5 (4,5-dihydroxyhexanediones (DHDs)) have been synthesized and their biological activity
tested in two bacteria. (4S,5R)-DHD is a synergistic agonist in Escherichia coli (which contains the LsrB
receptor), while it is an agonist in Vibrio harveyi (LuxP), displaying the strongest agonistic activity
Received 13 August 2011
Revised 4 November 2011
Accepted 5 November 2011
Available online 12 November 2011
Keywords:
AI-2
DPD
DPD analogs
DPD agonist
Quorum sensing
reported so far (EC50 = 0.65
ods to manipulate quorum sensing as a method for controlling bacteria.
Ó 2011 Elsevier Ltd. All rights reserved.
lM) in this organism. Thus, modification at C-5 opens the way to novel meth-
1. Introduction
(S)-2 and THMF (R)-2 and the linear hydrated form of DPD coexist
in a ratio of approximately 2:2:1; in presence of boron, borate com-
plexes such as THMF-borate-3 are formed.9
A peculiarity of AI-2 signaling is that diverse bacteria have
different AI-2 receptors which recognize distinct forms of AI-2. So
far two AI-2 receptors have been identified: LuxP binds THMF-bo-
rate 3,10 while LsrB binds THMF (R)-2.11 The crystal structures of
Bacterial cells coordinate their behavior in order to adapt to envi-
ronmental conditions. Cell-to-cell communication occurs through
the secretion and sensing of substances called autoinducers. This
process, known as quorum sensing, allows bacteria to synchronize
behaviors according to the density of their population. Most autoin-
ducers described so far are species-specific, however, one of them,
autoinducer-2 (AI-2), is shared by many bacterial species,1 allowing
different species to influence each other’s behavior.2 Initially
discovered in marine bacterium Vibrio harveyi where it induces bio-
luminescence,3 AI-2 is now known to regulate important processes
such as biofilm formation, toxin production and virulence in a vari-
ety of species.4,5 Because AI-2 also regulates behaviors of human
pathogens such as Vibrio cholerae,6 there is great interest in the dis-
covery of non-natural quorum sensing modulators7 for applications
in the treatment of bacterial infections.8
As illustrated in Scheme 1, the AI-2 signal derives from the
spontaneous cyclisation of the linear (S)-4,5-dihydroxypentanedi-
one (DPD) (S)-1 into isomeric forms which co-exist in a dynamic
equilibrium. In water, the tetrahydroxytetrahydrofurane (THMF)
⇑
Corresponding author.
E-mail address: rventura@itqb.unl.pt (M.R. Ventura).
Scheme 1. Forms of AI-2 in water.
0968-0896/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2011.11.007

250
F. Rui et al. / Bioorg. Med. Chem. 20 (2012) 249–256
the two ligand/receptor complexes showed that both proteins have
very different binding sites,10,11 thus explaining why they accom-
modate different forms of AI-2.
The LsrB receptor was first identified in Salmonella typhimurium11
and it is present in most enteric bacteria like Escherichia coli but also
in more distantly related bacteria such as the plant symbiont Sino-
rhizobium meliloti12 and the human pathogen Bacillus anthracis.13
In contrast, the LuxP receptor seems to be restricted to bacteria from
the Vibrio genera.10
S. typhimurium, E. coli and V. harveyi are the best characterized
bacteria in terms of AI-2 signaling, thus the activity of structural
analogs of DPD has been tested in reporter strains of these species.
So far, the design of bioactive analogs has focused on the replace-
ment of methyl group at C-1 of DPD with linear,14–16 branched and
cyclic17,18 alkyl groups, as well as with a trifluoromethyl substitu-
ent.19 Synthetic inhibitors based on an arylsulfonyl thioester scaf-
fold were shown to be effective in V. harveyi,20 and a brominated
natural product was shown to inhibit different quorum sensing
signals in this species.21 Among C-1 alkyl-DPD analogs, CF3-DPD
proved to be a weak agonist19 and hexyl-DPD was the only inhib-
Scheme 2. Synthetic DPD analogs: 4,5-dihydroxyhexanediones 4 (DHDs).
thus, despite the promising results mentioned above, only C-1
and, more recently, carbocyclic structural analogs have been tested
for their biological properties.14–19,22
Here we explore the impact of the novel modification of DPD at
C-5. Because the crystal structures of LsrB11 and LuxP10 reveal a
cavity in the binding sites of both receptors in proximity of the
C-5 methylene group of the natural ligands (Fig. 1), substitution
at this position should provide new leads for bioactive DPD ana-
logs. Therefore enantiopure analogs of DPD modified in position
C-5, that is, 4,5-dihydroxyhexanedione 4 (DHD) were prepared:
(4S,5S)-4, (4R,5S)-4, (4S,5R)-4 and (4R,5R)-4 (Scheme 2). These
compounds were tested for response from both LsrB and LuxP type
receptors: in vivo with the E. coli reporter assay and with the more
common V. harveyi in vivo assay and in vitro with an assay that di-
rectly measures binding to purified LuxP protein.
itor reported in V. harveyi, with IC50 of approximately 10 l
M.15
Surprisingly, all C-1 alkyl-DPD analogs assayed in V. harveyi proved
to be synergistic agonists in presence of physiological concentra-
tion of DPD, enhancing bioluminescence induced by the natural
autoinducer.14,16,17 The same analogs were also tested in the
enteric bacteria S. typhymurium14 and E. coli,18 but in these cases
antagonistic activity towards AI-2 was observed. The most potent
inhibitor reported for enteric bacteria is butyl-DPD with IC50 of
approximately 5 l
M.14 Thus, these synthetic structural analogs of
DPD reveal that minor alterations can have profound impact on
their biological effects and that these effects differ depending on
the class of AI-2 receptor tested.
2. Results
2.1. Synthesis and structural characterization of the C-5-DPD
analogs
Due to the tendency of the target products to rearrange and
polymerize, the synthesis of DPD and its analogs is challenging,9
Four diasteroisomers of 4,5-dihydroxyhexanedione 4 (DHD,
Scheme 2) were prepared as enantiopure compounds starting from
(S)- and (R)-methyl lactate.
As outlined in Scheme 3, methyl (S)-lactate was transformed
into its Weinreb amide 5 in three steps, using an efficient method
for the coupling of carboxylic acid with N,O-dimethylhydroxyl-
amine.24 This procedure improves our published synthesis of the
DPD-scaffold from methyl glycolate,25 because it provides Weinreb
amide in high yield without chromatographic purification. Amide 5
was homologated to the silyloxyhexynone 6, which was reduced
with two different stereoselective methods. In the first one, addi-
tion of hydride from (S)-Alpine BoraneÒ and deprotection provided
diol 7 with high diastereoisomeric purity (dr = 0.95), as determined
by NMR. Protection of the diol as cyclohexylidene-acetal 8 and sub-
sequent oxidation, as described for the synthesis of DPD,9 provided
intermediate 9, which gave DHD (4S,5S)-4 after acidic hydrolysis.
In the second reduction method, ketone 6 was treated with Ce(III)
and NaBH4 in methanol at ꢀ50 °C;26 subsquent fluorolysis resulted
in diol 10 with high diastereoisomeric purity (dr = 0.95). After
preparation of acetal 11 and of protected dihydroxydiketone 12,
acidic hydrolysis afforded (4R,5S)-4, the second of DHD diastereo-
isomers. Starting from methyl (R)-lactate through ketone ent-6, the
same synthetic strategy described above resulted in (4R,5R)-4,
using (R)-AlpineboraneÒ, and in (4S,5R)-4, using the inorganic
reduction reagents.
Figure 1. (a) Mesh representation of surface of binding site and (adjacent cavity)
inside the S. typhimurium AI-2 receptor (LsrB; PDB ID 1TJY) as calculated by
PyMOL.23 Position of biological ligand (THMF (R)-2), as determined by X-ray
crystallography, shown in stick representation. The same color scheme (oxy-
gen = red, nitrogen = blue, carbon = white, boron = pink) is used for both. The colors
of the mesh show the identity of the atom at the surface of the protein at that
position: red and blue are hydrophilic regions (oxygen and nitrogen) and white
hydrophobic (carbon) (b) As in (a), rotated 90 degrees about the y-axis. (c) Mesh
representation of surface of binding site (and adjacent cavities) inside the V. harveyi
AI-2 receptor (LuxP; PDB ID 1JX6) as calculated by PyMOL and position of the
biological ligand (and THMF-borate 3), as determined by x-ray crystallography,
shown in stick representation. (d) As in (c), rotated 90 degrees about the x-axis.
The cyclohexanone released during hydrolysis was removed
from the acidic solution washing quickly with chloroform; this
byproduct does not inhibit bacterial growth in trace amount.9 Each
stock solution of analog was quantified by 1H NMR and used, as
such, for the biological experiments. The amount of acid present
did not significantly alter the pH of the buffer used for the exper-
iments with bacteria and did not inhibit bacterial growth.

F. Rui et al. / Bioorg. Med. Chem. 20 (2012) 249–256
251
Scheme 3. Synthesis of two isomers of 4,5-dihydroxyhexanedione 4 (DHDs). Reagents and Conditions: (a) TBDPSCl, Et3N, DMAP, CH2Cl2, 92%; (b) LiOH, THF/H2O, quant. (c)
CH3(CH3O)NH2Cl, CDMT, NMM, THF, 84%; (d) propyne, BuLi, THF, ꢀ78 °C, 73%; (e) (S)-Alpine Borane, THF; (f) CeCl3ꢁ7H2O, NaBH4, CH3OH, ꢀ50 °C; (g) TBAF, THF, 36% 7 and 53%
10, two steps, dr 0.95 for 7 and 10; (h) c-hexanonedimethylacetal, H2SO4, DMF, 83%; (i) NaIO4, cat. RuO2ꢁH2O, CH3CN/CCl4/H2O, 46%; (j) H2SO4 0.01 M.
After peak assignment using COSY, HMBC and 1H NMR, the dis-
tribution of linear and cyclic forms of DHDs in solution was
monitored by 1H NMR in D2O. In the case of DPD the methyl group
at C-1 is diagnostic because it results in two singlets at about
1.5 ppm when it is part of a hemiacetal in the two cyclic forms,
while it gives a singlet at about 2.3 ppm when it is in the open-
chain form.9 In the case of DHDs we observed only the signals of
the C-1 methyl group in the cyclic forms at about 1.4 ppm in 1H
NMR (cf. Supplementary data S11 And S14, 1a and 1b), whereas
the open-chain signal was detected only in the NOESY spectrum
as an exchange peak with the cyclic forms (cf. S12 and S15, ‘H1’).
This shows that the cyclic, hydrated forms of DHD are essentially
the only species to be found in solution (> 95%, cf. S11 and
S14), the two anomers being present in roughly equal ratio. There-
fore the introduction of a methyl group at C-5 of DPD alters the
equilibrium distribution of linear and cyclic structures with respect
to genuine DPD.9 The alkyl substituent
atom could favor ring formation due to a conformational effect
(Thorpe-Ingold effect)27 by analogy to the cyclization of
methyl-
-butyrolactones.28 The presence of exchange signals of
a to the cyclizing hetero-
Figure 2. Synergistic effect of DPD and C5-DPD analogs in E. coli. Reporter strain
KX1446 was incubated with DPD only (black bars, 20 M); with analogs only (white
bars, 150 M); and with mixtures containing DPD (20 M) and analogs (pale grey
bars, 15 M; dark grey bars 150 M). Error bars indicate standard deviations;
l
c
-
l
l
l
c
l
the C-1 protons in NOESY (cf. S12 and S15, ‘H1’) experiments con-
firms the rapid interconversion between cyclic and open chain
forms of DHD. Maintaining the fast equilibrium between the linear
and cyclic forms is important for biological activity. Enteric bacte-
ria recognize one cyclic form of DPD, but, once transported into the
cytoplasm the open-chain form is phosphorylated and becomes
active inside the cell.29 Furthermore, locked cyclic analogs are
impaired for biological activity both in S. typhymurium and
V. harvey.22 As mentioned above our NOESY results show that the
analogs prepared retain the requisite of equilibrium between
open-chain and cyclic forms and thus are appropriate for the study
of the biological effects of the additional stereogenic center at C-5
of DPD and its enantiomer.
analysis of variance was performed with the Student’s t-test.
As illustrated in Fig. 2, DHDs do not induce expression of the lsr
genes if tested alone (agonist assay). However, in presence of the
natural signal DPD (the typical antagonist assay), the analogs
(4S,5R)-4 and (4S,5S)-4 are not antagonists but rather induce a sig-
nificant increase in lsr expression over DPD alone. In contrast, the
analogs with non-natural (4R)-configuration, (4R,5S)-4 and
(4R,5R)-4, do not show significant induction under the same condi-
tions. Therefore the (4S,5R)-4 and (4S,5S)-4 are synergistic agonists
of AI-2 in E. coli. This is an interesting difference from C-1 alkyl
analogs of DPD, which have been shown to have a synergistic effect
in V. harveyi14,17 but not in bacteria with the LsrB receptor such as
E. coli.
2.2. Biological activity of DHDs in E. coli—a bacterium with the
LsrB receptor
2.3. Biological activity of DHDs in V. harveyi—a bacterium with
the LuxP receptor
We used an E. coli reporter strain to evaluate the activities of the
synthesized analogs in bacteria with the LsrB receptor. This assay
uses a non-pathogenic strain instead of the more common S.
typhimurium reporter strain for assaying the response of AI-2 and
analogs in enteric bacteria.25,18 This E. coli strain, which carries a
lsr-lacZ promoter fusion allows monitoring of the expression of
lsr, the promoter induced by AI-2 in enteric bacteria, through the
measurement of the b-galactosidase activity of cell extracts.30,31
The same analogs were tested on the V. harveyi system. Their ef-
fect was assayed both by means of a protein biosensor derived
from the LuxP receptor32 and by measuring the bioluminescence
of a V. harvey reporter strain.11 The activites of the DHDs as assayed
via the engineered LuxP receptor (CFP-LuxP-YFP) are shown in
Fig. 3.

252
F. Rui et al. / Bioorg. Med. Chem. 20 (2012) 249–256
Table 1
Binding affinity of C-5 DPD analogs to CFP-LuxP-YFP by
the LuxP-FRET assay
Compound
EC50 in LuxP-FRET assay (l
M)a
DPD
0.032 0.0015
0.14 0.019
1.87 0.22
11.6 2.8
(4S,5R)-4
(4S,5S)-4
(4R,5R)-4
(4R,5S)-4
10.0 2.4
a
Assay performed with varying concentrations of
test compound. Values obtained from data in Fig. 3.
Figure 3. LuxP-FRET assay of DPD and C5-DPD analogs. Error bars indicate standard
deviations.
Binding of AI-2 to this engineered receptor causes a conforma-
tional change in the protein structure and a dose-dependent de-
crease in the fluorescence resonance energy transfer (FRET).
Therefore a lower FRET ratio corresponds to an increase in ligand
binding to LuxP.32 All the DHDs analogs induce a dose-dependent
change in FRET ratio (Fig. 3), meaning that they all bind to the
receptor. However, the EC50 values (Table 1) indicate that they
have very different affinities.
Small structural differences dramatically affect the interaction
with the receptor protein. As expected, the natural (4S)-configura-
tion results in the most active isomers (4S,5R)-4 and (4S,5S)-4. In
addition, the (5R)-compound is more active than the (5S)-, mean-
ing that the configuration at C-5 affects the interaction with LuxP.
The same pattern was observed in vivo using the V. harveyi
reporter strain MM32, which is capable of inducing biolumines-
cence in an AI-2 dependent manner via the LuxP receptor.11 As
shown in Fig. 4, all analogs induce bioluminescence in the absence
of DPD (agonist assay). However, in comparison with DPD, higher
concentrations of DHDs are necessary to reach maximum induc-
tion and none of the analogues can induce the same levels of bio-
luminescence as DPD. This indicates that all DHDs are partial
agonists of DPD.33 In this in vivo assay, the configuration of the
DHDs influences their activity in the same way as in the LuxP-FRET
protein assay: the strongest agonist is (4S,5R)-4, while its (4S,5S)-
isomer requires ten-fold higher concentrations to elicit the same
response. The EC50 for each compound is reported in Table 2.
As in the case of DPD, bioluminescence decreases at high
concentrations of any of the analogs (Fig. 4). To test if there was
an antagonistic effect of DHDs in competition with DPD, the same
assay was performed in presence of the genuine signal
(Supplementary data, Fig. S1).
Under these conditions, high concentrations of the isomeric
DHDs reduce bioluminescence, suppressing the effect of DPD. The
IC50 were determined and, as shown in Table 2, the (4S,5S)-4 analog
behaves as the strongest antagonist. Growth inhibition is observed
at a concentration of 0.1 mM for all the compounds tested, but the
inhibition is similar for DHDs and DPD (Supplementary data
Fig. S2) and thus the antagonistic effect cannot be ascribed to toxic-
ity. Overall, this suggests that in these conditions the DHDs compete
with DPD for LuxP binding.
The results obtained with V. harveyi show that DHDs mimic
DPD: they behave as partial agonists in the sub-micromolar to
low-micromolar concentration range, and they have antagonistic
effects at higher concentrations. Agonistic activity shows a clear
structure–activity relationship trend, the (5R)-configuration being
more active than the (5S).
Figure 4. Bioluminescence assay of DPD and C-5 DPD with V. harveyi reporter strain
MM32. Error bars indicate standard deviations.
Table 2
Summary of quorum sensing modulation by DPD and C-5 DPD analogs in V. harveyi
a
b
Compound
EC50
(
lM)
IC50
—
(lM)
DPD
0.076 0.002
0.65 0.05
6.21 0.46
19.5 6.3
(4S,5R)-4
(4S,5S)-4
(4R,5R)-4
(4R,5S)-4
169 19
57.54 0.29
159.7 8.4
c
26.5 8.4
—
a
Assay performed with varying concentrations of test compound, values
obtained from the data in Fig. 4.
b
Assay performed with varying concentrations of test compound in the presence
of 0.1 lM of DPD; values obtained from the data in Supplementary Fig. S1.
c
Not determined.
As shown in Table 2, the analog (4S,5R)-4 has an EC50 of 0.65 lM.
This is the strongest DPD agonistic activity among known analogs.
The most potent synthetic agonists of AI-2 reported so far, CF3-
DPD and the opposite enantiomer of DPD, (R)-1, have EC50 ranging
between 30 and 84 l
M.19,34 DHDs show independent agonistic
activity and thus have a different biological effect than C-1 alkylated
analogs of DPD which, with the exception of hexyl-DPD and CF3-
DPD, are generally active only in presence of DPD in V. harveyi.14,16,17
3. Discussion
The introduction of a new stereocenter at C-5 of the DPD scaf-
fold resulted in four diastereoisomeric analogs, DHDs, with inter-
esting biological effects. The analogs were tested in model
bacteria that recognize AI-2 via either the LsrB or the LuxP recep-
tor. The response for the (4R)-DHD isomers was weak in both
systems. In contrast, a synergistic response to (4S)-DHD isomers
in the presence of DPD was observed in E. coli, whereas (4S,5R)-
DHD showed a agonistic response stronger than (4S,5S)-DHD in

F. Rui et al. / Bioorg. Med. Chem. 20 (2012) 249–256
253
V. harveyi. The weak induction observed in both systems by the
(4R)-DHDs was not surprising because similar responses had been
reported for (4R)-DPD. Thus, these results reinforce the importance
of the (4S)-configuration for AI-2 dependent response in the two
systems tested.
EC50 = 0.65 lM. Based on the analysis of the crystal structures
available for LsrB and LuxP bound to their natural ligands, we
propose an explanation that supports the experimental data ob-
tained with the different analogs tested. Knowledge of the biolog-
ical activity of C-5 substituted DPD analogs complements that of
the known C-1 alkyl analogs and should be useful for the design
of novel quorum sensing modulators.
In the AI-2 based induction of the V. harveyi quorum sensing sys-
tem, the known receptor is LuxP and there is no evidence for another
protein involved in AI-2 recognition. In contrast, the E. coli response
to AI-2 is more complex. Induction of AI-2-dependent lsr activation
requires that cyclic THMF (R)-2 bound to LsrB is transported to the
cytoplasm and then that the open-chain form is phosphorylated at
the C-5 hydroxyl group by the kinase LsrK. This phosphorylated
AI-2 then deactivates the repressor of the AI-2 response (LsrR). It
is known that LsrB is not essential for E. coli AI-2 response, but acti-
vation is clearly stronger when the receptor is present.31 In addition,
a recent report showed that the kinase is not very selective and can
phosphorylate a broad range of DPD analogs as long as the hydroxyl
group at C-5 is maintained,18 which is the case for the DHDs. The ob-
served synergistic effect in E. coli is interesting, but, as with the C1-
analogs in V. harveyi,14,16,17 a molecular explanation for the effect is
still lacking and further work is required.
Within the two signaling system tested, there is a notable dis-
tinction in the response of the different species to the (4S,5R)-
DHD and (4S,5S)-DHD diastereoisomers. Specifically, E. coli (LsrB
receptor) react similarly to both enantiomers, while V. harveyi
(LuxP) responds more strongly to (4S,5R)-DHD, suggesting better
binding of this ligand to the LuxP receptor. This behavior can be
rationalized through analysis of the crystal structures of the recep-
tors with their natural ligands. Both receptors are two domain pro-
teins with the ligand binding site located between the domains,
adjacent to a hinge region that connects the domains.9–11 As shown
in Fig. 1 (panels A and B), the C-5 carbon of (R)-THMF-2 is oriented
away from the surface of the LsrB binding site, towards an unfilled
region. Visual analysis suggests that there would be no significant
difference in steric interaction with the protein between the 5R
and 5S isomers, in agreement with our experimental data showing
similar E. coli response to both DHDs. On the other hand, V. harveyi
(LuxP) does show stereospecific dependence, with a 10-fold stron-
ger response to (4S,5R)-DHD than to (4S,5S)-DHD. As shown in Fig-
ure 1 (panels C and D), a 5S methyl would be oriented toward (and,
presumably, come into steric conflict with) the protein while the
5R form would position the extra methyl on the face of the ring
nearer the large cavity in the binding site. Some conformational
change in the protein and/or ligand is expected when LuxP binds
a DHD rather than the biological ligand; the space provided by
the cavity should facilitate such changes. Moreover, the region of
the protein near the position presumed to be occupied by a 5(R)
methyl is largely hydrophobic (white in Fig. 1C and D), suggesting
both a favorable protein/ligand interaction and that fewer hydro-
gen bonds would be broken in any minor conformational changes
induced by binding of (4S,5R)-DHD as opposed to (4S,5S)-DHD.
Thus, relative proximity of the 5(R) methyl to the binding site cav-
ity provides a reasonable explanation for the observed stronger re-
sponse of V. harveyi to (4S,5R)-DHD.
5. Experimental
5.1. Synthesis
All solvents were distilled prior to use. The optical purity of the
starting materials purchased was ee = 97% for methyl (S)-lactate
and ee = 96% for methyl (R)-lactate (GLC, supplier specifications).
The following abbreviations are used in the text: Hex, hexanes;
EA, ethyl acetate, Et2O, diethyl ether, MeOH, methanol; AcOH, ace-
tic acid. Dry THF was distilled from sodium, dry CH2Cl2 was dis-
tilled from P2O5. Pooled organic phases were dried on MgSO4. 1H
NMR spectra were recorded in CDCl3 at 400 MHz, using TMS as
internal standard or 3-(trimethylsilyl)-propanesulfonic acid in
D2O. J values are given in Hz. 13C NMR spectra were recorded at
100 MHz, using the solvent peak as internal standard in CDCl3.
Peak assignment was based on correlation experiments (HMQC,
HMBC, 1H,1H COSY). D2O used for NMR contained 1% of 1 M
H2SO4 solution in H2O. TLC was performed on aluminium sheets
coated with Kieselgel 60 F254 Flash column chromatography was
carried out on silica 60
lM; Medium Pressure Liquid Chromatogra-
phy was performed on 45–60
lM silica. Specific rotations were
measured using a 1 dm path length thermostated cell, values be-
tween ꢀ2 and +2 could not be measured with sufficient precision
and are not reported. HRMS were measured either by ESI-TOF-
MS with Flow Injection Analysis or by DIP-EI-MS.
5.1.1. Preparation of (2S)-2-t-butyldiphenylsilyloxypropanoic
acid methyl ester (13) or ent-13
A solution of tert-butyldiphenylchlorosilylane (3.02 g, 11 mmol)
in CH2Cl2 (3 ml) was slowly added to a solution containing methyl
(S)-lactate (1.04 g, 10 mmol), TEA (1.09 g, 1.5 ml, 11 mmol) and
DMAP (50 mg, 0.04 equiv) in dry CH2Cl2 (2 ml). After 15 h the sus-
pension was treated with saturated NH4Cl (5 ml) and extracted
with CH2Cl2 (3 ꢂ 20 ml). The organic phases were washed with
NH4Cl, dried and evaporated. Purification on column chromatogra-
phy (Hex/EA 95:5) gave a colorless liquid (3.15 g, 92%). ½a D20
ꢀ44.4
ꢃ
(c 1.20; CH2Cl2). The same procedure applied to methyl (R)-lactate
gave product ent-13 ½a D20
ꢃ
+45.7 (c 0.95 in CH2Cl2). 1H NMR, 13C
NMR and IR spectra matched literature data.35
5.1.2. Preparation of (2S)-2-t-butyldiphenylsilyloxypropanoic
acid (14) or ent-14
Ester 13 (3.08 g, 9.0 mmol) was dissolved in THF (9 ml) in a
100 ml beaker, and an aqueous solution of LiOH (0.6 M, 47 ml,
28 mmol) was added. The mixture was covered and stirred at
1000 rpm with a stirring bar sweeping the full internal diameter
of the beaker. After 2 h, solid LiOH (648 mg, 27 mmol, 3 equiv)
was added and, after additional 16 h, hydrolysis was complete
(TLC, Hex/EA 9:1 + 5% AcOH). The mixture was diluted with 5 ml
of ether, acidified to pH 2–3 with 2 M HCl and extracted with ether
(5 ꢂ 10 ml). The organic phases were pooled, dried and evaporated
under reduced pressure, giving a colorless oil (2.95 g, quant.). 1H
NMR (CDCl3, d): 7.70–7.60 (m, 4H) and 7.50–7.35 (m, 6H): aromat-
ics, 4.33 (q, J = 6.8 Hz, 1H, 6-H), 1.31 (d, J = 6.8 Hz, 3H, 3-H); 1.11 (s,
9H, C(CH3)3). 13C NMR (CDCl3, APT, d): 175.1 (ꢀ, C-1); 135.7 (+),
135.6 (+), 132.5 (ꢀ), 131.8 (ꢀ), 130.4 (+), 130.3 (+), 128.0 (+),
127.9 (+): aromatics; 69.4 (+, C-2); 26.9 (+, C(CH3)3); 20.9
4. Conclusions
The four stereoisomers of 4,5-dihydroxyhexanedione (4, DHD)
have been prepared starting from the enantiomeric methyl lac-
tates of high optical purity. These new analogs of DPD are the
first AI-2 synergistic agonists reported so far in enteric bacteria.
In the alternative signaling system V. harveyi, DHDs having
(5R)-configuration have increased agonistic activity compared to
the (5S)-configuration. In particular, the analog (4S,5R)-4 is the
strongest AI-2 agonist described so far in V. harveyi, with

254
F. Rui et al. / Bioorg. Med. Chem. 20 (2012) 249–256
(+, C-3); 19.1 (ꢀ, C(CH3)3). IR (film, cmꢀ1): br 3500–3000, 3072,
the reaction was quenched with sat. NH4Cl (2 ml). The aqueous
phase was saturated with solid NaCl and extracted exhaustively
with EA (5 ꢂ 5 ml). Purification by gravity chromatography using
flash silica (1st column Hex/EA 7/3 followed by pure EA; 2nd col-
umn Hex/EA 6/4) afforded 7 as a dense oil which solidified in the
form of white flakes (124 mg, 36% over two steps). 1H NMR (CDCl3,
d): 4.30–4.22 (m, 1H, H-3); 3.89–3.80 (m, 1H, H-2); 2.20 (d,
J = 7.0 Hz, 1H, –OH); 1.92 (d, J = 6.8 Hz, 1H, –OH); 1.883 (d,
J = 2.1 Hz, 3H, H-6); 1.260 (d, J = 6.4 Hz, 3H, H-1). 13C (CDCl3, APT,
d): 83.4 (ꢀ, C-4); 76.5 (ꢀ, C-5); 70.4 (+, C-2); 67.4 (+, C-3); 18.3
(+, C-1); 3.6 (+, C-6). IR (KBr, cmꢀ1) br 3600–3000: –OH; 2991;
2978; 2913; 2298; 2237 (C„C). HRMS (ESI-MS): calcd for
3051, 2956, 2933, 2859, 1724 (C@O). ½a D20
ꢀ9.4 (c 1.00 in CH2Cl2).
ꢃ
The same procedure applied to ester ent-13 gave acid ent-14; ½a D20
ꢃ
+9.9 (c 0.95 in CH2Cl2).
5.1.3. Preparation of (2S)-N-methoxy-N-methyl-2-
(t-butyldiphenylsilyloxy)propanamide (5) or ent-5
2-Chloro-4,6-dimethoxy-[1,3,5]-triazine (1.85 g, 10.6 mmol,
1.2 equiv) and N-methylmorpholine (2.9 mL, 26.4 mmol, 3 equiv)
were added to a solution of 14 (2.9 g, 8.8 mmol) in THF (26 ml) stir-
red under Ar atmosphere. After stirring the mixture for 45 min, so-
lid N,O-dimethyl-hydroxylamine hydrochloride (0.86 g, 8.8 mmol,
1 equiv) was added. After 5 h and TLC control (H/EA 7:3) the reac-
tion was quenched with water (39 ml) and the mixture extracted
with Et2O (3 ꢂ 20 ml). The pooled organic phases were washed
twice with sat. Na2CO3, 1 M HCl and brine, then dried and concen-
trated at reduced pressure. The dense liquid obtained was filtered
through a pad of silica eluted with H/EA 7/3. The filtrate crystal-
C6H10NaO2 [M+Na]+ 137.0578, found 137.0557. ½a 2D0
ꢀ4.6 (c 0.74
ꢃ
in CHCl3); dr = 0.95 (1H NMR). The same procedure starting from
ketone ent-6 using (R)-alpine borane gave diol ent-7; ½a D20
ꢃ
+5.5 (c
0.80 in CHCl3), dr = 0.95 (1H NMR).
5.1.6. Preparation of (2S,3R)-2,3-(cyclohexylidenedioxy)-hex-4-
yne (8) or ent-8
lized at room temperature (2.75 g, 84%). ½a D20
ꢀ15.4 (c 0.80 in
ꢃ
CH2Cl2); mp: 64.9–65.8 °C. The same procedure was applied to acid
DMF (1 ml) in a small flask in an ice bath was acidified with one
drop of conc. H2SO4 and an aliquot was diluted 1–10. Cyclohexanone
dimethyl acetal (79 mg, 1.2 equiv) was dissolved in the diluted
acidic DMF and this was added to diol 7 (52 mg, 0.46 mmol) in a
5 ml flask equipped with a CaCl2 tube. After stirring for 30 min at
room temperature, the solution was heated up to 70 °C for 5 min
to remove methanol. After addition of some pellets of K2CO3, the
remaining DMF solution was loaded directly onto a small silica col-
umn (Hex/Et2O 9/1). Evaporation of the collected fractions under re-
duced pressure (p >220 mbar) gave a colorless liquid (74 mg, 83%).
1H NMR (CDCl3, d): 4.73–4.68 (m, 1H, H-3); 4.22 (quint. J = 6.1 Hz,
1H, H-2); 1.868 (d, J = 2.1 Hz, 3H, H-6); 1.80–1.35 (m, 10 H, c-hexy-
lydene); 1.334 (d, J = 6.1, 3H, H-1). 13C NMR (CDCl3, APT, d): 109.6 (ꢀ,
quaternary C of c-hexylidene); 83.7 (ꢀ, C-4); 75.2 (ꢀ, C-5); 73.5 (+, C-
2); 69.6 (+, C-3); 37.7, 35.3, 25.1, 24.0, 23.8 (ꢀ, c-hex), 16.6 (+, C-1);
ent-14 resulting in amide ent-5. ½a D20
ꢃ
+14.7 (c 0.84 in CH2Cl2); lit-
erature36
½
a 2D5 +12.1 (c 1.4 in CHCl3); 1H NMR, 13C NMR and IR
ꢃ
spectra matched literature data.36
5.1.4. Preparation of (2S)-2-t-butyldiphenylsilyloxyhex-4-yn-3-
one (6) or ent-6
A
1.6 M solution of BuLi in hexane (6.2 ml, 10.0 mmol,
1.4 equiv) was added dropwise to a 1 M solution of propyne in
THF (10.7 ml, 10.7 mmol, 1.5 eq.) cooled to ꢀ78 °C and stirred un-
der Ar. The mixture was stirred for 30 min, and the temperature al-
lowed to rise to 0 °C over 30 min and cooled again to ꢀ78 °C. A
solution of amide 5 (2.65 g, 7.1 mmol) in THF (10 ml) was added
dropwise and the suspension was slowly warmed to ꢀ20 °C and
stirred for 1 h at this temperature. After TLC control (Hex/EA 7/
3), sat. NH4Cl solution (20 ml) was added, and the mixture acidified
to pH 4 with 2 M HCl. The organic layer was separated and the
aqueous solution was extracted with ether (3 ꢂ 20 ml). The pooled
organic phases were washed with brine (2 ml), dried and evapo-
rated at reduced pressure. Purification with MPLC (Hex/EA 9/1)
gave a colorless oil (1.86 g, 73%). 1H NMR (CDCl3, d: 7.75–7.65
(m, 4H) and 7.45–7.35 (m, 6H): aromatics; 4.23 (q. J = 6.8 Hz, 1H,
H-2); 1.99 (s, 3H, H-6) 1.29 (d, J = 6.8 Hz, 3H, H-1); 1.11 (s, 9H,
C(CH3)3). 13C NMR (CDCl3, APT, d) : 189.8 (ꢀ, C-3); 135.8 (+),
133.6 (ꢀ), 133.0 (ꢀ), 129.8 (+), 127.6 (+): aromatics; 93.9 (ꢀ, C-4)
78.6 (ꢀ, C-5); 75.5 (+, C-2); 26.8 (+, C(CH3)3); 20.5 (+, C-1); 19.3
(ꢀ, C(CH3)3); 4.3 (+, C-6). IR (film, cmꢀ1): 3072, 3051; 2958;
2933; 2893; 2858; 2216 (C„C); 1678 (C@O). HRMS (ESI-MS) calcd
3.7 (+, C-6). IR (film, cmꢀ1): 2937, 2861, 2245 (C„C). ½a 2D0
ꢃ
+10.3 (c
0.79 in CH2Cl2). HRMS: calcd for C12H18NaO2 [M+Na]+ 217.1199,
found 217.1206, The same procedure starting from diol ent-7 re-
sulted in product ent-8. ½a D20
ꢀ9.5 (c 0.69 in CH2Cl2).
ꢃ
5.1.7. Preparation of (4S,5S)-4,5-(cyclohexylidenedioxy)-hexan-
2,3-dione (9) or ent-9 5S)-4,5-(cyclohexylidenedioxy)-hexan-
2,3-dione (9) or ent-9
CCl4 (0.7 ml), CH3CN (0.7 ml) and a water solution containing
NaIO4 (151 mg, 0.70 mmol in 1.0 ml of H2O) were added to a
6 ml vial containing acetal 8 (60 mg, 0.31 mmol). The mixture
was stirred strongly and then RuO2ꢁH2O (1.0 mg, 0.008 mmol,
0.025 equiv) was added. After 5 min (TLC Hex/Et2O 1:1) the mix-
ture was filtered through a compact layer of MPLC silica gel in
CH2Cl2 (height 0.5 cm) covered by a layer of NaHCO3 (height
0.5 cm). The filtrate was dried and evaporated under reduced pres-
sure at 10 °C, affording a bright yellow oil which was immediately
stored at ꢀ80 °C (32.3 mg, 46%). This material, which contained
also unreacted 8 (19%), was not further purified to avoid loss of
product due to the formation of hydrates on silica. 1H NMR (CDCl3,
d): 5.32 (d, J = 7.3 Hz, 1H, H-4); 4.65 (quint, J = 6.52 Hz; 1H, H-5);
2.36 (s, 3H, H-1); 1.90–1.20 (m, cyclohexylidene, superimposed
on cyclohexanone), 1.11 (d, J = 6.52 Hz, 3H, H-6). The additional
signals are due to the starting material (4.75–4.68 (m, 0.19H) ca.
19% mol) and cyclohexanone (br 2.35, 1.90–1.20). 13C NMR (CDCl3,
APT, d): 198.0, 195.8 (ꢀ, C-2 and C-3); 110.8 (ꢀ, c-hex); 78.7 (+, C-
4); 73.3 (+, C-5); 36.8, 34.8, 25.1, 24.02; 23.7 (ꢀ, c-hex.); 24.0 (+, C-
1); 16.6 (+, C-6). IR (film, cmꢀ1) 2937; 2861; 1797 and 1714 (C@O);
for C22H26NaO2Si [M+Na]+ 373.1594, found 373.1583. ½a 2D0
ꢀ35.8 (c
ꢃ
0.80 in CH2Cl2). The same procedure applied to amide ent-5 re-
sulted in ketone ent-6; ½a D20
ꢃ
+35.8 (c 0.71 in CH2Cl2).
5.1.5. Preparation of (2S,3R)-hex-4-yn-2,3-diol (7) or ent-7
A 0.5 M THF solution of (S)-alpine borane (12.0 ml, 6.0 mmol)
was added to a solution of ketone 6 (1.05 g, 3.0 mmol) in THF
(1 ml) stirred under Ar. After 36 h, the starting material was not
detected (TLC, Hex/EA 9/1); the reaction was quenched with a
sat. solution of NH4Cl (5 ml), and the mixture was extracted with
CH2Cl2. After drying and evaporation of the organic phases, the res-
idue was purified on MPLC (Hex/EA/MeOH 97/2/1) and analysed by
NMR: 1H NMR (CDCl3, d): 7.75–7.65 (m, 4H) and 7.5–7.35 (m, 6H):
aromatics; 4.30–4.25 (m, 1H, C-3); 3.98–3.90 (m, 1H, C-2); 1.822
(d, 2.0 Hz, 3H, C-6); 1.13 (d, 6.4 Hz, C-1), 1.07 (s, 9H). The resulting
material, containing semiprotected diol 15 and isomers with mi-
grated silylether group, was not purified further. It was dissolved
in THF (1 ml) under Ar and treated with a 1 M solution of TBAF
in THF (1.6 ml, 1.6 mmol). After 5 h and TLC control (Hex/EA 6/4)
½
a 2D0
+16.0 (c 1.28, CH2Cl2). HRMS (DIP-EI-MS): calcd for C12H18O4
226.1205, found 226.1206. The same procedure starting from hex-
ꢃ
yne ent-8 gave ent-9. ½a D20
ꢀ14.4 (c 1.20; CH2Cl2).
ꢃ

F. Rui et al. / Bioorg. Med. Chem. 20 (2012) 249–256
255
5.1.8. Preparation of (2S,3S)-hex-4-yn-2,3-diol (10) or ent-10
CeCl3ꢁ7H2O (890 mg, 2.38 mmol, 1.2 equiv) was added to a
stirred solution of ketone (760 mg, 2.16 mmol) in MeOH
5.1.11. General procedure for the preparation and
characterization of 4,5-dihydroxyhexanediones (4)
6
A suspension of the corresponding dihydroxhexanedione acetal
(ca. 9.2 mg, 40 M) in D2O (3 ml) containing 1% of 1 M H2SO4 solu-
(8 ml) under Ar. After the solid had dissolved completely
(10 min) the solution was cooled to ꢀ50 °C, and NaBH4
(122.6 mg, 3.24 mmol, 1.5 equiv) was added. After 10 min, a TLC
control (Hex/EA 9/1) showed no starting material and H2O
(4 ml) was added carefully. Extraction of the white mixture with
CH2Cl2 (3 ꢂ 5 ml), drying and evaporation of the organic phases
gave a colourless, viscous liquid which contained semiprotected
diol 16 and isomers with a migrated silylether group. 1H NMR
(CDCl3, d): 7.80–7.65 (m, 4H) and 7.50–7.30 (m, 6H): aromatics;
4.15–4.10 (m, 1H, C-3); 3.885 (quint. J = 6.0 Hz, C-2); 1.793 (d,
2.0 Hz, 3H, C-6); 1.09–1.06 (m, 12H, C-1 and C(CH3)3). This mate-
rial was dissolved directly in THF (1 ml) under Ar and it was trea-
ted with a 1 M solution of TBAF in THF (2.11 ml, 2.11 mmol).
After 5 h, the reaction was quenched with NH4Cl (2 ml). After sat-
uration of the water phase with solid NaCl, the mixture was
exhaustively extracted with EA 5 ꢂ 5 ml. Purification by gravity
on flash silica (Hex/EA 6/4) afforded a dense oil (131 mg, 53% over
two steps). 1H NMR (CDCl3, d): 4.09–4.03 (m, 1H, H-3); 3.75
(quint. J = 6.4 Hz, 1H, H-2); 2.48 (br, 2H, –OH); 1.867 (d,
J = 2.1 Hz, 3H, H-6); 1.264 (d, J = 6.4 Hz, 3H, H-1). 13C (CDCl3,
APT, d): 82.9 (ꢀ, C-4); 77.5 (ꢀ, C-5); 71.3 (+, C-2); 67.7 (+, C3);
18.4 (+, C-1); 3.56 (+, C-6). IR (film, cmꢀ1): br 3700–3000;
2976; 2923; 2237 (C„C). Elemental analysis: found C 63.10, H
9.16; calc. for C6H10O2 C 63.14, H 8.83%; dr = 0.95 (1H NMR).
The same procedure starting from ent-6 gave ent-10; dr = 0.95
(1H NMR).
l
tion in H2O was repeatedly sonicated using an ultrasound bath at
4 °C until the yellow color disappeared, due to the hydration of the
a-diketone moiety. The solution was stirred in an ice bath, the
hydrolysis was monitored by NMR and was complete after 18 h.
The solution was extracted with 1/5 of the volume of CDCl3
(600
or for NMR analysis. NOESY spectra were acquired with mixing time
300 msec. Quantification. An aliquot of the water phase (300 L),
acidic D2O (276 L) and a solution of Sodium Formate (200 mM,
24 L, final concentration 8 mM) were transferred to a NMR tube
lL) and the water phase was either used for biological assays
l
l
l
and mixed. Quantitative 1H NMR was recorded with the pulse pro-
gram zgprde for the presaturation of H2O (1%) and delay time of
30 s to allow full spin relaxation. The concentration of the DPD ana-
log was calculated comparing the area of the formate peak with the
sum of the doublet signals of H-4, and it was generally in the range of
6.5–8.5 mM in the mother solution, depending on the batch. Derivat-
isation with phenylenediamine. The samples used for quantification
were treated with 5 equiv of a 200 mM solution of Phenylenedi-
amine hydrochloride in acidic D2O, in the NMR tube and the solution
was mixed. After 1 h 1H NMR spectra were acquired.
5.1.12. Characterization of (4S,5S)-4,5-dihydroxyhexanedione
((4S,5S)-4) or (4R,5R)-4
1H NMR (D2O, d): (a) 3.98 (quint. J = 6.4 Hz, 1H, H-5); 3.57 (d,
J = 6.1 Hz, H-4); 1.400 (s, superimposed, H-1); 1.29 (d, J = 6.4 Hz,
3H, H-6); b) 3.94 (d, J = 7.7 Hz, 1H, H-4); 3.78 (dq, J = 7.7 Hz;
J = 6.4 Hz, 1H, H-5); 1.406 (s, superimposed on cyclohexanone, H-
1); 1.31 (d, J = 6.4 Hz, 3H, H-6); a:b = 47:53. Additional signals in
the spectrum were due to (2S,3R)-hex-4-yn-2,3-diol (7) derived
5.1.9. Preparation of (2S,3S)-2,3-(cyclohexylidenedioxy)-hex-4-
yne (11) or ent-11
The same procedure described above for acetal 8, except for
starting from diol 10 (110 mg, 0.96 mmol), resulted in acetal 11
(131 mg, 70%). 1H NMR (CDCl3, d): 4.11 (dq, J = 8.15 Hz, J = 2.1 Hz,
1H, H-3); 4.00 (dq, J = 8.15 Hz, J = 6.1 Hz, 1H, H-2); 1.868 (d,
J = 2.1 Hz, H-6); 1.80–1.35 (m, 10H, c-hexylidene); 1.17 (d,
J = 6.1 Hz, H-1). 13C NMR (CDCl3, APT, d): 109.7 (ꢀ, c-hexylidene);
83.1 (ꢀ, C-4); 77.1 (+, C-2); 75.1 (ꢀ, C-5); 72.0 (+, C-3); 36.7,
36.0, 25.1, 23.8 (ꢀ, c-hexylidene); 16.9 (+, C-1); 3.8 (+, C-6). IR
(film, cmꢀ1): 2937, 2862, 2245 (C„C). HRMS (ESI-MS): calcd for
from starting material
9
(region 4.3–4.2, 3.85–3.80 ppm;
d,1.8 ppm and d, 1.2 ppm) and acetic acid (s, 2.08 ppm). 13C NMR
(D2O, APT, d), 102.0 (C-2) and 97.9 (C-3), a and b; a) 78.5 (+, C-
4); 77.2 (+, C-5); 21.2 (+, C-1); 19.2 (+, C-6); b) 79.6 (+, C-4); 76.2
(+, C-5); 19.3 (+, C-1) 17.4 (+, C-6).
5.1.13. Characterization of (1R,2S)-1-(3-methylquinoxalin-2-
yl)propane-1,2-diol ((1R,2S)-17)
C
12H18NaO2 [M+Na]+ 217.1199, found 217.1201. Applying the
1H NMR (d, ppm) 8.25–8.20 (m, superimposed to formic acid),
8.10–8.05 (m, 1H), 8.02–7.92 (m, 2H) aromatics; 5.15 (d,
J = 6.6 Hz, 1H); 4.29 (quint. J = 6.5 Hz); 2.93 (s, 3H); 1.35 (d,
J = 6.3 Hz, 3H). Other signals are due to (2S,3R)-hex-4-yn-2,3-diol
(8). Of interest is the absence of the peaks due to the cyclic forms
at 4.0–3.5 ppm, 1.41 and 1.40 ppm.
same procedure to diol ent-10 gave acetal ent-11.
5.1.10. Preparation of (4R,5S)-4,5-(cyclohexylidenedioxy)-
hexan-2,3-dione 12 or ent-12
Acetal 11 (103 mg, 0.53 mmol) was dissolved in CCl4 (1.2 ml)
and CH3CN (1.2 ml) in a 6 ml vial. After adding a solution of NaIO4
(258 mg, 1.21 mmol, 1.8 ml), Ru2OꢁH2O (1.8 mg, 0.013 mg,
0.025 equiv) was added and the mixture was stirred vigorously.
After 5 min it was filtered directly over a compact layer of MPLC
silica gel in CH2Cl2 covered with a layer of NaHCO3 and eluted with
CH2Cl2. Evaporation of the filtrate at reduced pressure at low tem-
perature gave a yellow oil containing some starting material,
which was purified on a small silica column (Hex/Et2O 3/2). Evap-
oration at reduced pressure at 10 °C gave a bright yellow oil which
was stored at ꢀ80 °C (41 mg, 34%). 1H NMR (CDCl3, d): 4.59 (d,
J = 7.25 z, 1H, H-4); 4.27 (dq, J = 7.25, J = 6.07 Hz; 1H, H-5); 2.38
(s, 3H, H-1); 2.0–1.5 (m, c-hexylidene, superimposed to cyclohexa-
none); 1.45 (d, J = 6.07 Hz, H-6). 13C NMR (CDCl3, APT, d): 198.8,
197.1 (ꢀ, C-2 and C-3); 111.8 (ꢀ, c-hexylidene); 82.7 (+, C-4);
74.2 (+, C-5); 37.0, 35.3, 25.0, 23.9; 23.7 (ꢀ, c-hexylidene); 24.9
(+, C-1); 19.0 (+, C-6); IR (film, cmꢀ1): 2937; 2864; 1797 and
1716 (C@O); HRMS (EI-MS): calcd for C12H18O4 226.1205, found
226.1207. The same procedure applied to acetal ent-11 gave prod-
uct ent-12.
5.1.14. Characterization of (4R,5S)-4,5-dihydroxyhexanedione
((4R,5S)-4) or (4S,5R)-4
1H NMR (D2O, d, ppm) a) 4.40 – 4.35 (m, 1H, H-5); 4.06 (d,
J = 5.2 Hz, 1H, H-4); 1.42 (s, 3H, H-1); 1.15 (d, J = 6.7 Hz, 3H, H-6);
b) 4.35 – 4.28 (m, 1H, H-5); 3.87 (d, J = 4.5 Hz, 1H, H-4); 1.36 (s,
3H, H-1); 1.25 (d, J = 6.7 Hz, H-6); a:b = 54:46. 13C NMR (D2O, APT,
d, ppm) 13C NMR (APT, D2O, d): (a) 102.3 (C-2); 99.4 (C-3); 75.0 (+,
C-4); 73.6 (+, C-5); 20.9 (+, C-1); 13.4 (+, C-6); b) 103.4 (C-2); 99.0
(C-3); 76.1 (+, C-5); 75.0 (+, C-4); 18.8 (+, C-1); 14.9 (+, C-6).
5.1.15. Characterization of (1S,2S)-1-(3-methylquinoxalin-2-
yl)propane-1,2-diol ((1S,2S)-17)
1H NMR: 8.10–8.06 (m, 1H); 7.98–7.94 (m, 1H); 7.88–7.80 (m,
2H); 5.14 (d, J = 5.0 Hz, 1H); 4.41–4.32 (m, 1H); 2.93 (s, 3H); 1.24
(d, J = 6.5 Hz, 3H). Of interest is the absence of the signals due to
the cyclic forms: doublets at 4.06, 3.87, 1.15 and 1.25 ppm, as well
as the singlets at 1.42 and 1.36 ppm.

256
F. Rui et al. / Bioorg. Med. Chem. 20 (2012) 249–256
5.2. Bioassays in E. coli
E. coli reporter strain KX1446
fused to the lsr promoter was grown overnight in LB medium con-
taining 100 mM MOPS buffer at pH 7. Stock solutions of the test
compounds were diluted to a concentration of 2 mM with water
containing H2SO4 0.01 M. As sample (4S,5S)-4 had an impurity of
diol 7, a solution of 7 was also prepared in the same conditions
and tested as control, and showed no activity in any of the assays.
The following solutions/suspensions were transferred in test tubes
Operacional Ciência e Inovação (POCI) 2010’’ and Fundação para
a Ciência e a Tecnologia (FCT). F.R. acknowledges FCT and ITQB
for funding. This work was supported by the FCT grants PTDC/
QUI-BIQ/113880/2009 and PEst-OE/EQB/LA0004/2011.
D
LsrF, DLsrG-, DluxS with LacZ
Supplementary data
Supplementary data (NMR spectra of all new compounds,
NOESY spectra of (4R,5S)-4 and (4S,5S)-4) associated with this arti-
cle can be found, in the online version, at doi:10.1016/
j.bmc.2011.11.007.
in this order: LB medium (900
terial culture (10 M), acidic water and analog solution adding up
to a volume of 100 L. Previous tests showed that the pH difference
lL), an aliquot of the overnight bac-
l
l
References and notes
of the medium before (pH 6.62) and after (pH 6.35) addition of 10%
volume of H2SO4 0.01 M was not significant, this was reflected by
equal bacterial growth. For each series of measurement one control
1. Waters, C. M.; Bassler, B. L. Annu. Rev. Cell Dev. Biol. 2005, 21, 319.
2. Xavier, K. B.; Bassler, B. L. Nature 2005, 437, 750.
3. Bassler, B. L.; Wright, M.; Silverman, M. R. Mol. Microbiol. 1994, 13, 273.
4. Lowery, C. A. Chem. Soc. Rev. 2008, 37, 1337.
experiment with 100 lL of H2SO4 0.01 M and one with AI-2 were
5. Federle, M. J. Contributions to Microbiology; Karger: Basel, 2009. Vol. 16, pp. 18–
32.
6. Zhu, J.; Miller, M. B.; Vance, R. E.; Dziejman, M.; Bassler, B. L.; Mekalanos, J. J.
PNAS 2002, 99, 3129.
7. Galloway, W. R. J. D.; Hodgkinson, J. T.; Bowden, S. D.; Welch, M.; Spring, D. R.
Chem. Rev. 2011, 111, 28.
8. Lowery, C. A.; Salzameda, N. T.; Sawada, D.; Kaufmann, G. F.; Janda, K. D. J. Med.
Chem. 2010, 53, 7467.
9. Semmelhack, M. F.; Campagna, S. R.; Federle, M. J.; Bassler, B. L. Org. Lett. 2005,
7, 569.
10. Chen, X.; Schauder, S.; Potier, N.; Van Dorsselaer, A.; Pelczer, I.; Bassler, B. L.;
Hughson, F. M. Nature 2002, 415, 545.
carried out; each concentration was tested in three independent
cultures. After 4.5 or 5 h cells were harvested, suspended in an
equal volume of Z-buffer, and the optical density (OD595) of each
tube was recorded. The b-galactosidase activity of the protein ex-
tract was measured using o-nitrophenyl-b-
according to literature.30 To observe the synergistic activity of
the analogs, experiments were carried out in presence of 20
DPD (1). The final concentration of the analogs was of 15 or
150 M. Diol 7 was tested at 50 M.
D-galactopyranoside
lM
l
l
11. Miller, S. T.; Xavier, K. B.; Campagna, S. R.; Taga, M. E.; Semmelhack, M. F.;
Bassler, B. L.; Hughson, F. M. Mol. Cell. 2004, 15, 677.
12. Pereira, C. S.; McAuley, J. R.; Taga, M. E.; Xavier, K. B.; Miller, S. T. Mol. Microbiol.
2008, 70, 1223.
5.3. LuxP-FRET assay
In vitro response of LuxP-FRET protein was measured as de-
scribed,32 optimized for 96 well plate reading using a multilabel
counter (1420 Victor 3, Perkin Elmer).25 Serial dilutions of test
compounds were performed in water containing H2SO4 0.01 M
13. Pereira, C. S.; de Regt, A. K.; Brito, P. H.; Miller, S. T.; Xavier, K. B. J. Bacteriol.
2009, 191, 6975.
14. Lowery, C. A.; Park, J.; Kaufmann, G. F.; Janda, K. D. J. Am. Chem. Soc. 2008, 130,
9200.
15. Lowery, C. A.; Abe, T.; Park, J.; Eubanks, L. M.; Sawada, D.; Kaufmann, G. F.;
Janda, K. D. J. Am. Chem. Soc. 2009, 131, 15584.
16. Ganin, H.; Tang, X.; Meijler, M. M. Bioorg. Med. Chem. Lett. 2009, 19, 3941.
17. Smith, J. A. I.; Wang, J. X.; Nguyen-Mau, S. M.; Lee, V.; Sintim, H. O. Chem.
Commun. 2009, 7033.
and added to 12.5 l
g mlꢀ1 of CFP-LuxP-YFP chimeric protein in
25 mM of sodium phosphate buffer (pH 8), 35 mM NaCl, and
1 mM borate. Competition experiments were carried out in pres-
ence of 0.1
l
M DPD (1). Samples (2.5
l
l) were added to 280
ll of
18. Roy, V.; Smith, J. A. I.; Wang, J.; Stewart, J. E.; Bentley, W. E.; Sintim, H. O. J. Am.
Chem. Soc. 2010, 132, 11141.
19. Frezza, M.; Balestrino, D.; Soulère, L.; Reverchon, S.; Queneau, Y.; Forestier, C.;
Doutheau, A. Eur. J. Org. Chem. 2006, 4731.
reaction volume and FRET ratio was calculated (527/485 nm).
5.4. Bioluminescence assay in V. harveyi
20. Peng, H.; Cheng, Y.; Ni, N.; Li, M.; Choudhary, G.; Chou, H. T.; Lu, C.-D.; Tai, P. C.;
Wang, B. ChemMedChem 2009, 4, 1457.
21. Defoirdt, T.; Miyamoto, C. M.; Wood, T. K.; Meighen, E. A.; Sorgeloos, P.;
Verstraete, W.; Bossier, P. Appl. Environ. Microbiol. 2007, 9, 2486.
22. Tsuchikama, K.; Lowery, C. A.; Janda, K. D. J. Org. Chem. 2011, 76, 6981.
23. The PyMOL Molecular Graphics System, Version 1.4, Schrödinger, LLC.
24. De Luca, L.; Giacomelli, G.; Taddei, M. J. Org. Chem. 2001, 66, 2534.
25. Ascenso, O. S.; Marques, J. C.; Santos, A. R.; Xavier, K. B.; Rita Ventura, M.;
Maycock, C. D. Bioorg. Med. Chem. 2011, 19, 1236.
V. harveyi in vivo response was measured using MM32 reporter
strain grown in AB (autoinducer bioassay medium) as previously
reported.11 Serial dilutions were performed in water containing
H2SO4 0.01 M. To determine light inhibition of analogs 0.1 lM of
DPD were added to cultures. Light emission was measured in a
Wallac Model 1450 Microbeta scintillation counter after 5 h
30 min of incubation at 30 °C. Bioluminescence is reported as
counts per minute (cpm) of the light emitted by the cultures. Half
maximal effective concentration (EC50) and half maximal inhibi-
tory concentration (IC50) were determined by fitting four parame-
ter sigmoidal curves using non-linear least square analysis in a
custom made Matlab program.
26. Kuethe, J. T.; Comins, D. L. J. Org. Chem. 2004, 69, 5219.
27. Jung, M. E.; Piizzi, G. Chem. Rev. 2005, 105, 1735.
28. Wheeler, O. H.; Rodriguez, E. E. G. D. J. Org. Chem. 1964, 29, 1227.
29. Xavier, K. B.; Miller, S. T.; Lu, W. Y.; Kim, J. H.; Rabinowitz, J.; Pelczer, I.;
Semmelhack, M. F.; Bassler, B. L. ACS Chem. Biol. 2007, 2, 128.
30. Slauch, J. M.; Silhavy, T. J. J. Bacteriol. 1991, 173, 4039.
31. Xavier, K. B.; Bassler, B. L. J. Bacteriol. 2005, 187, 238.
32. Rajamani, S.; Zhu, J.; Pei, D.; Sayre, R. Biochemistry 2007, 46, 3990.
33. Silverman, R. The Organic Chemistry of Drug Design and Drug Action; Elsevier
Academic Press: Burlington, MA, USA, 2004. pp. 121–142.
34. Lowery, C. A.; McKenzie, K. M.; Qi, L.; Meijler, M. M.; Janda, K. D. Bioorg. Med.
Chem. Lett. 2005, 15, 2395.
Acknowledgments
35. Motozaki, T.; Sawamura, K.; Suzuki, A.; Yoshida, K.; Ueki, T.; Ohara, A.;
Munakata, R.; Takao, K.-I.; Tadano, K.-I. Org. Lett. 2005, 7, 2265.
36. Xie, W.; Zou, B.; Pei, D.; Ma, D. Org. Lett. 2005, 7, 2775.
We thank Dr. Pedro Lamosa for his help in the discussion of
NOESY results. The NMR spectrometers are part of the National
NMR Network (REDE/1517/RMN/2005), supported by ‘‘Programa