Total synthesis, isolation, surfactant properties, and biological evaluation of ananatosides and related macrodilactone-containing rhamnolipids

Article abstract of DOI:10.1039/d1sc01146d

Rhamnolipids are a specific class of microbial surfactants, which hold great biotechnological and therapeutic potential. However, their exploitation at the industrial level is hampered because they are mainly produced by the opportunistic pathogenPseudomo

Full text of DOI:10.1039/d1sc01146d

Chemical
Science
View Article Online
View Journal | View Issue
EDGE ARTICLE
Total synthesis, isolation, surfactant properties, and
biological evaluation of ananatosides and related
macrodilactone-containing rhamnolipids†
Cite this: Chem. Sci., 2021, 12, 7533
All publication charges for this article
have been paid for by the Royal Society
of Chemistry
a
b
Maude Cloutier,
Marie-Joelle Prevost,a Serge Lavoie,
Thomas Feroldi,b
¨
´
Marianne Piochon, a Marie-Christine Groleau, a Jean Legault,b Sandra Villaume,c
Jerome Crouzet,c Stephan Dorey,c Mayri Alejandra Dıaz De Rienzo,ad Eric Deziel
a
*
´ ˆ
´
`
´
a
*
and Charles Gauthier
Rhamnolipids are a specific class of microbial surfactants, which hold great biotechnological and
therapeutic potential. However, their exploitation at the industrial level is hampered because they are
mainly produced by the opportunistic pathogen Pseudomonas aeruginosa. The non-human pathogenic
bacterium Pantoea ananatis is an alternative producer of rhamnolipid-like metabolites containing
glucose instead of rhamnose residues. Herein, we present the isolation, structural characterization, and
total synthesis of ananatoside A, a 15-membered macrodilactone-containing glucolipid, and ananatoside
B, its open-chain congener, from organic extracts of P. ananatis. Ananatoside A was synthesized through
three alternative pathways involving either an intramolecular glycosylation,
macrolactonization or direct enzymatic transformation from ananatoside B.
a
chemical
series of
a
A
diasteroisomerically pure (1/2), (1/3), and (1/4)-macrolactonized rhamnolipids were also synthesized
through intramolecular glycosylation and their anomeric configurations as well as ring conformations
were solved using molecular modeling in tandem with NMR studies. We show that ananatoside B is
a more potent surfactant than its macrolide counterpart. We present evidence that macrolactonization
of rhamnolipids enhances their cytotoxic and hemolytic potential, pointing towards a mechanism
involving the formation of pores into the lipidic cell membrane. Lastly, we demonstrate that ananatoside
A and ananatoside B as well as synthetic macrolactonized rhamnolipids can be perceived by the plant
immune system, and that this sensing is more pronounced for a macrolide featuring a rhamnose moiety
in its native 1C4 conformation. Altogether our results suggest that macrolactonization of glycolipids can
dramatically interfere with their surfactant properties and biological activity.
Received 25th February 2021
Accepted 22nd April 2021
DOI: 10.1039/d1sc01146d
rsc.li/chemical-science
are made by the combination of a lipidic chain covalently linked
to a carbohydrate moiety. Because of their ability to form pores
Introduction
and destabilize biological membranes, microbial glycolipids
have attracted increased attention as therapeutic agents.4
Glycolipids exhibit a wide range of pharmaceutical activities,
including antibacterial,5 antifungal,6 antiviral,7 hemolytic,8
anticancer,9 and adjuvant10 activities. In addition, surfactants of
microbial origin are increasingly considered for use in diverse
biotechnological applications such as in the food and cosmetic
industries as well as in bioremediation technologies.3
Bacteria represent a rich reservoir of structurally diverse glyco-
sylated metabolites.2 Among these compounds, microbial
glycolipids show considerable potential for biomedical and
biotechnological applications.3 Microbial glycolipids are
surfactants, i.e., amphiphilic surface-active compounds, which
a
´
Centre Armand-Frappier Sante Biotechnologie, Institut National de la Recherche
´
Scientique (INRS), 531, Boulevard des Prairies, Laval (Quebec), H7V 1B7, Canada.
Rhamnolipids are a specic class of microbial biosurfactants
that have been intensively investigated in recent years.11 Struc-
turally, rhamnolipids are a-congured mono- or di-L-rhamnose
residue(s) O-linked to an (R)-b-hydroxyalkanoic acid dilipidic
chain of C6 to C14 carbon length (see RhaC10C10 3 in Fig. 1). As
compared to the commercially available petroleum-derived
synthetic surfactants, the biodegradability, low critical micelle
concentrations (CMCs), and high tension surface activity make
rhamnolipids exquisite environmental alternatives for
E-mail: charles.gauthier@inrs.ca
b
´
´ ´
´
Laboratoire d'Analyse et de Separation des Essences Vegetales (LASEVE), Departement
´
´
`
des Sciences Fondamentales, Universite du Quebec a Chicoutimi, 555, Boulevard de
´
´
l'Universite, Chicoutimi (Quebec), G7H 2B1, Canada
c
´
Universite de Reims Champagne-Ardenne, INRAE, USC RIBP 1488, SFR Condorcet-FR
CNRS 3417, 51100 Reims, France
dSchool of Pharmacy and Biomolecular Sciences, Liverpool John Moores University, L3
3AF, Liverpool, UK
† Electronic supplementary information (ESI) available. See DOI:
10.1039/d1sc01146do
© 2021 The Author(s). Published by the Royal Society of Chemistry
Chem. Sci., 2021, 12, 7533–7546 | 7533

View Article Online
Chemical Science
Edge Article
industrial applications.12 Rhamnolipids are also of interest comprising a b-D-glucose residue linked to a C10C10 dilipid
because they exhibit a plethora of intriguing pharmacological chain through both C1 and C6 positions.26 Biologically active
activities. They inhibit the formation of microbial biolms,13–15 glucolipids containing mono-, di-, and trilactones of different
which play crucial role in infections caused by many pathogenic macrocyle sizes have been identied from other microorgan-
bacteria. They exhibit antimicrobial activities against both isms29–31 including, to name a few examples, lactonic sopho-
Gram-positive and Gram-negative bacteria as well as against rolipids (Fig. 1), glucolipsins (Fig. 1), cycloviracins, fattiviracins,
plant pathogenic fungi.12 Because of their ability to interact with macroviracins, and anthrobacilins. By restricting the rotation of
biological membranes, rhamnolipids can trigger the death of the substituents and stabilizing the active conformation of the
cancer cells16 and induce hemolysis of red blood cells.17,18 glycolipid, the presence of the macrocycle is a prerequisite to
Furthermore, as invasion pattern molecules (also known as the bioactivity of these constrained saccharides.30,32
elicitors),19 rhamnolipids can stimulate the plant immune
system resulting in the strengthening of plant cell walls along macrolide prompted us to investigate P. ananatis extracts with
with the production of antimicrobial compounds.20–23
the aim of identifying structurally similar biosurfactants. As
The identication of ananatoside A (1) as a novel microbial
Rhamnolipids are mainly produced by pathogenic Gram- part of our research program on the synthesis of microbial
negative bacteria belonging to the Pseudomonas and Bur- glycans,33–36 we are interested in developing synthetic routes
kholderia genera,11,24 which hampers their exploitation at the that would allow an alternative and straightforward access to
industrial level since several are human pathogens. As such, these macrodilactone-containing glycolipids, and enable the
there is an increased interest to identify non-pathogenic assessment of their tensioactive properties and biological
bacteria that can produce high concentrations of rhamnoli- activities. Within this framework, we herein present the total
pids and structurally-related biosurfactants. Our group has synthesis of ananatoside A (1), its newly identied open-chain
recently identied the bacterium Pantoea ananatis BRT175 as an congener ananatoside B (2), the related RhaC10C10 (3) as well
alternative non-pathogenic producer of biosurfactants.25,26 The as ve unprecedented, anomerically pure (1/2)-, (1/3)-, and
biosynthesis of rhamnolipids involves the successive function (1/4)-macrodilactone-containing rhamnolipids (4–6) (Fig. 1).
of three enzymes: RhlA, which directs the biosynthesis of the We show that ananatoside A (1) can be efficiently obtained via
lipidic precursor, along with RhlB and RhlC, which are two three different pathways implying either a chemo- or enzymatic
rhamnosyltransferases.27,28 Although the production of bio- macrolactonization, or an intramolecular glycosylation as the
surfactants in P. ananatis is also catalyzed by RhlA and RhlB key steps of these synthetic sequences. DFT calculations were
homologues, we unexpectedly identied a gluco-rather than used to decipher the ring conformations and anomeric cong-
a rhamnolipid from P. ananatis ethyl acetate extracts.25 The urations of synthetic macrolactonized rhamnolipids 4–6, which
isolated compound, that we called ananatoside A (1, Fig. 1), were readily obtained through intramolecular glycosylation.
features an unprecedented 15-membered macrodilactone ring Their surfactant properties and biological activity, i.e., antimi-
crobial activity, cytotoxicity, hemolytic activity, as well as their
interaction with the plant immune system, were also investi-
gated, providing meaningful fundamental insights into the
impact of the presence of the macrodilactonic ring on the
physical and biological properties of this relevant class of
microbial glycolipids.
Results and discussion
Isolation of ananatoside A (1) and ananatoside B (2)
The non-pathogenic bacterium P. ananatis BRT175 was found to
be a producer of surface-active glycolipids based on its genetic
homologies with rhamnolipid biosynthetic genes. The micro-
bial biosurfactants were extracted with EtOAc from the super-
natant of a liquid culture of BRT175, which was acidied to pH
ꢀ3 prior the extraction. Purication of the crude extract was
performed using a semi-preparative reversed-phase HPLC
system equipped with a charged aerosol detector (CAD).37 The
use of the CAD detector was particularly appealing over tradi-
tional UV/visible detectors as it allows the detection of mole-
cules lacking chromophore groups such as glycolipids (see
Fig. S1†). Through this optimized procedure, ananatoside A (1)
Fig. 1 Examples of macrolactone-containing glucolipids produced by
microbes (A); structures of target ananatoside A (1) and ananatoside B
was isolated as a yellow oil and its physical and analytical data
(Rf, specic rotation, HRMS, and 1D and 2D NMR) were in
(2) produced by Pantoea ananatis along with the related RhaC10C10 (3)
perfect agreement with those we recently published.26 A more
(B); structures of target macrodilactone-containing rhamnolipids 4–6
(C). Lactone functionalities are highlighted in blue.
polar and major congener (see Fig. S1†), that we have named
7534 | Chem. Sci., 2021, 12, 7533–7546
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Edge Article
Chemical Science
ananatoside B (2), was also isolated from the EtOAc extract as intramolecular glycosylation using the ester-linked b-hydrox-
a white amorphous powder. This compound was analyzed by ylipid as a stereodirecting tether.44 Importantly, the latter
HR-ESI-TOF-MS in the positive mode, yielding pseudomo- approach has only been implemented in rare occasions for the
lecular ion peaks at m/z 538.3587 [M + NH4]+ and m/z 543.3142 total synthesis of macrolactone-containing natural products.29
[M + Na]+, pointing towards the presence of a glycolipid
Along these lines and as depicted in our retrosynthetic
featuring a hexose residue linked to two hydroxydecanoic acid analysis strategy (Fig. 2), we envisioned to build the macrolide
chains. Comparison of the 1D and 2D NMR data of ananatoside skeleton of ananatoside A (1) via three parallel synthetic path-
B (2, see Fig. S2† for main COSY and HMBC correlations) with ways (routes A–C) as to maximize the chances of efficiently
our previously reported data for ananatoside A (1) suggested the reaching our target. We rst hypothesized that ananatoside A
presence of an open-chain b-linked glucolipid congener without (1) would be formed through the intramolecular glycosylation of
the macrolide ring (no HMBC cross-peak between H6 and C100). derivative 7 following chemoselective cleavage of the C6-O-TBS
The absolute conguration of both the glucose and b-hydrox- group (route A). Notably, precursor 7, activated in the form of an
ydecanoic acid chains were determined following acid hydro- STol glycoside,45 would be equipped with a (2-azidomethyl)
lysis and measurement of specic rotation in comparison with benzoyl (AZMB)46 group at C2 that would act as a neighboring
an authentic D-glucose sample and literature data for the lipid participating group enabling the formation of the 1,2-trans-
chain, as we previously described.26 Based on these chemical glucosidic linkage. The b-selectivity would also be favored by the
and spectroscopic evidences, we established the complete steric constraint exerted by the resulting macrolide.44 The
structure of ananatoside B (2) as shown in Fig. 1. The structural propensity of the AZMB group to be orthogonally cleaved by
determination of ananatoside A (1) and ananatoside B (2) were means of, for instance, Staudinger reduction36,47 without
further conrmed through total synthesis as described in the affecting the macrolide functionality represents a further
next section.
advantage of using this protecting group at this specic posi-
tion. Precursor 7 would be readily synthesized by Steglich
esterication48 between thioglucoside 9 and dilipid 10.
Inspired by the success story of the total synthesis of several
macrolide-containing natural products29,38,49–52 and by way of
comparison with the intramolecular glycosylation strategy, we
also wanted to construct the 15-membered ring of ananatoside
Total synthesis of ananatoside A (1) and ananatoside B (2)
Retrosynthetic analysis. The total synthesis of macrolactone-
containing natural products has usually been accomplished via
late-stage ring-closing metathesis or macrolactonization as key
steps of the synthetic routes.29,30,38 Notwithstanding the success
of these pioneering approaches, they present some drawbacks
such as the use of highly diluted reaction mixtures and the
possible di- or oligomerization of the acyclic precursors.29
Owing to their thermal stabilities together with their optimal
regio- and enantioselectivities, commercially available lipases
stand as a tantalizing synthetic tool for the formation of mac-
rolactone rings from unprotected substrates.39–41 Alternatively,
capitalizing on glycosylation chemistry,42,43 carbohydrate-
embedded macrolactones can be built upon stereoselective
A
(1) through both enzymatic and chemical macro-
lactonizations (Fig. 2). Regarding the chemical macro-
lactonization (route C), glucolipid 8 would act as an exquisite
acyclic precursor for this intramolecular transformation
following unmasking of the seco acid functionality. Moreover,
global deprotection of this compound (8) would complete the
total synthesis of ananatoside B (2). Glucolipid 8 would be
prepared through the stereocontrolled glycosylation of thio-
glucoside 11 with benzylated dilipid 12. Here again, the use of
Fig. 2 Retrosynthetic disconnection strategy for the total synthesis of ananatoside A (1) and ananatoside B (2) according to three different
pathways: chemical macrolactonization, enzymatic macrolactonization, and inter- or intramolecular glycosylation. AZMB: 2-azidome-
thylbenzoyl; Bn: benzyl; Lev: levulinoyl; STol: thiotolyl; TBS: tert-butyldimethylsilyl. Blue: permanent protecting groups (Bn); red: temporary
protecting groups (TBS); green: esters used as temporary protecting groups enabling neighboring group participation (AZMB and Lev).
© 2021 The Author(s). Published by the Royal Society of Chemistry
Chem. Sci., 2021, 12, 7533–7546 | 7535

View Article Online
Chemical Science
Edge Article
a neighboring participating group, i.e., levulinoyl (Lev),36,47,53
that could be selectively cleaved in the presence of an ester
functionality would insure a successful outcome for the nal
steps of the synthetic route. Finally, we hypothesized that ana-
natoside A (1) could be enzymatically synthesized from unpro-
tected ananatoside B (2) using commercially available solid-
supported lipase such as Novozyme 435 (route B).41 If success-
ful, this enzymatic process would allow the direct and
straightforward conversion of ananatoside B (2) into ananato-
side A (1) from the isolated natural product or the synthetic
compound.
Scheme 2 Synthesis of thioglucoside derivatives 9 and 11.
in a three-step, one-pot sequence from the corresponding per-
trimethylsilylated thioglucoside. As revealed in Scheme 2,
regioselective silylation at the C6 position of diol 17 followed by
esterication with AZMBOH and subsequent desilylation under
the action of in situ generated HCl led to derivative 9 bearing
a free OH at C6 in 74% yield over three steps. Alternatively,
a levulinoyl group was installed at C2 by treatment of the C6-O-
TBS-protected derivative with levulinic anhydride to provide
fully protected thioglucoside 11 in 90% yield over two steps.
Synthesis of ananatoside B. Our next challenge was to
proceed with the total synthesis of ananatoside B (2). Fully
protected thioglucoside 11 was glycosylated with alcohol
acceptor 12 under the promotion of NIS/AgOTf activating
system,61 providing anomerically pure b-glucolipid 8 in 80%
yield (Scheme 3). The stereoselectivity of the reaction was
conrmed by 1H NMR (H1, d, 3JH1,H2 ¼ 8.0 Hz for Glcp in the 4C1
conformation). Then, TFA-mediated cleavage of the TBS group,
delevulinoylation under the action of hydrazine acetate, and Pd-
catalyzed hydrogenolysis cleanly led to ananatoside B (2) in 66%
yield over three steps. Physical and analytical data (Rf, specic
rotation, HRMS, and NMR) of synthetic ananatoside B (2) were
Synthesis of building blocks. Our synthetic journey
commenced with the assembly of the dilipidic side chain
derivatives. To do so, (R)-b-hydroxydecanoic acid 13 had to be
synthesized rst. According to the literature, monolipid 13 had
previously been prepared via either
a cross-metathesis/
Mitsunobu sequence,54 the Reformatsky reaction55 or from
Meldrum's acid.56–59 We decided to follow the latter approach as
it allowed the straightforward formation of target monolipid 13
in high overall yield and enantiomeric purity. Therefore, as
depicted in Scheme S1,† (R)-b-hydroxydecanoic acid 13 (ref. 57)
was prepared in a four-step sequence. The conguration and
enantiomeric purity (R, 99% ee) of methyl ester S4 were
conrmed through the preparation of Mosher's ester S5 (ref. 56)
and comparison with literature data (see Fig. S3†). Thereaer, b-
hydroxydecanoic acid 13 was subjected to regioselective ben-
zylation at the carboxylic acid moiety yielding benzyl ester 14 in
a nearly quantitative yield (Scheme 1). In parallel, derivative 13
was protected with a TBS group at the C3 position affording
silylated derivative 15.57 These two compounds were condensed
together under the activation of EDC/DMAP. Resulting pro-
tected dilipid derivative 16 was then either subjected to Pd-
catalyzed hydrogenolysis of the benzyl ester or to TFA-
mediated cleavage of the TBS group leading to the formation
of acid 10 or alcohol 12, respectively, ready for coupling with the
carbohydrate derivatives. Optimization of this synthetic
sequence allowed us to prepare gram amounts of these chiral
lipidic intermediates.
Having completed the synthesis of the dilipid derivatives, we
then focused our attention on the synthesis of thioglucosides 9
and 11, which were either used for subsequent Steglich esteri-
cation or glycosylation reactions, respectively. These two
compounds were prepared from diol 17,60 which was obtained
Scheme 3 Total synthesis of ananatoside B.
Scheme
4 Synthesis of alcohol 20 ready for intramolecular
Scheme 1 Synthesis of b-hydroxydecanoic acid derivatives.
7536 | Chem. Sci., 2021, 12, 7533–7546
glycosylation.
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Edge Article
Chemical Science
Table
1 NIS/triflate-promoted intramolecular glycosylation of Table 2 Chemical macrolactonization of precursor 22
precursor 20
Molar volume
Molar volume
(mL mmolꢁ1
)
Yielda (%)
Entry
Triate (equiv.)
(mL mmolꢁ1
)
Additive
Yielda (%)
Entry
Reagents
1
2
3
4
5
6
7
DCC, DMAP, PPTs
DCC, DMAP, PPTs
DMC, DMAP
860
60
60
15
60
60
15
64
40
67
1
2
3
4
5
6
7
8
9
AgOTf (0.4)
100
100
1000
10
100
100
100
100
100
50
—
—
—
—
—
—
—
NaOTf
KOTf
KOTf
70
86
Quant.
47
85
21
75
78
63
TMSOTf (0.2)
TMSOTf (0.2)
TMSOTf (0.2)
TMSOTf (2.1)
Yb(OTf)3 (1.6)
Zn(OTf)2 (1.6)
TMSOTf (0.2)
TMSOTf (0.2)
TMSOTf (0.2)
DMC, DMAP
71
DMC, DMAP, Cs2CO3
DMC, DMAP, KOTf
DMC, DMAP, KOTf
47b
80b
73
a
Isolated yield. b Traces of dimer were detected.
10
61
a
Isolated yield.
cycloviracin B1 through metal-templated macrolactonization,
we then sought to evaluate the impact of an excess of metallic
ions on the outcome of the glycosylation and to observe if
oligomeric macrolides could be generated. As it is known that
potassium and sodium are effective ions for such templated
reactions,50 KOTf and NaOTf were employed as additives in the
presence of catalytic amounts of TMSOTf (entries 8–10). Once
again, the intramolecular reaction was highly favored as no
other products than macrolide 21 were isolated from the reac-
tion mixture.
Synthesis of ananatoside A by chemical macrolactonization.
As previously mentioned, the second pathway that we chose to
investigate for the preparation of ananatoside A (1) was based
on a chemical macrolactonization49 as the key step. As depicted
in Table 2, glucolipid derivative 18 was used as an advanced
intermediate for this approach. The benzyl ester moiety of the
latter compound was thus selectively cleaved by catalytic
hydrogenation transfer, using cyclohexadiene as the hydrogen
transfer source.62,63 Intramolecular cyclization of the resulting
seco acid 22 was then studied through Keck (DCC, DMAP,
PPTs)64 and Fujisawa [2-chloro-1,3-dimethylimidazolinium
chloride (DMC), DMAP]65 macrolactonizations. These intra-
molecular esterications were performed at different concen-
trations and temperatures, with and without the slow addition
of seco acid 22, and by using two different coupling systems
(Table 2, entries 1–4). Pleasingly, target macrolide 23 was
formed in 71% yield when using DMC-mediated
in perfect agreement with the isolated natural product (see ESI†
for details).
Synthesis of ananatoside A by intramolecular glycosylation.
The total synthesis of ananatoside A (1) was our next task. We
commenced by studying the intramolecular glycosylation
pathway. To do this, we needed to prepare the acyclic precursor
20. Dilipid 10 was condensed at the C6 position of thioglucoside
9 through activation with EDC/DMAP (Scheme 4). Resulting
derivative 7 was treated with TFA in DCM to cleave the TBS
group giving u-hydroxy thioglucoside 20 in 84% yield ready for
intramolecular glycosylation.
As shown in Table 1, different conditions were screened for
the intramolecular glycosylation of acyclic precursor 20, i.e.,
promoter, molar volume, and additive. Initially, the concen-
tration was set to 0.01 M (corresponding molar volume of
100 mL mmolꢁ1) to minimize intermolecular interactions and
thus preventing the formation of di- and oligomeric byproducts.
At this concentration, AgOTf and TMSOTf were rst evaluated
as catalysts in combination with NIS (entries 1 and 2).61 In both
conditions, target macrolide 21 was isolated in good yields with
full b-stereoselectivity while no traces of diolide or higher olig-
omers were detected. Encouraged by these results, the impact of
the concentration of precursor 20 on the reaction outcome was
investigated by conducting the intramolecular glycosylation at
0.001 and 0.1 M (entries 3 and 4). We were pleased to nd that
macrolide 21 was formed quantitatively when the concentration
of the acyclic precursor was set to 0.001 M. Moreover, even at
a concentration of 0.1 M, which could have favored intermo-
lecular interactions, only monomer 21 was isolated although
the yield was signicantly lower (47%). The use of excess
amounts of TMSOTf (entry 5) or other triates such as Yb(OTf)3
and Zn(OTf)2 (entries 6 and 7, respectively) also resulted in the
¨
exclusive formation of target macrolide 21. Inspired by Furstner
Scheme 5 Deprotection of macrolides 21 and 23 into ananatoside A
(1).
and co-workers50 who completed the total synthesis of
© 2021 The Author(s). Published by the Royal Society of Chemistry
Chem. Sci., 2021, 12, 7533–7546 | 7537

View Article Online
Chemical Science
Edge Article
Table 3 Enzymatic macrolactonization of ananatoside B (2) into
ananatoside A (1)
yields and number of steps for both synthetic sequences were
very similar, i.e., 33 and 35% over eight steps from known diol
17 for the intramolecular glycosylation and macrolactonization
pathways, respectively.
Synthesis of ananatoside
A
by enzymatic macro-
lactonization. We were also interested in the direct conversion of
ananatoside B (2) into ananatoside A (1) by taking advantage of an
enzyme-catalyzed macrolactonization. Although efficient in terms
of overall yields, both previously described synthetic approaches
required protecting groups manipulation. Contrarywise, enzy-
matic reactions offer the advantages of being renewable and
specic, and of minimizing side product formation and puri-
cation steps.41 Immobilized Candida antarctica lipase B (CALB),
commercially available under the brand name of “Novozyme
435”, was therefore selected for this synthetic study as it has
showed high substrate versatility and efficiency for macro-
lactonization reactions.39 As revealed in Table 3, the reaction was
Entry
Solvent
Temp. (ꢂC)
Time (h)
Yield (%)
1
2
3
4
Toluene
Hexanesb
THFb
75
55
55
75c
168
114
112
3
30a
10a
Traces
Tolueneb
64d
a
b
Conversion (estimated by NMR).
Microwave heating. d Isolated yield.
Reaction in sealed tube.
c
˚
conducted in the presence of molecular sieves (4 A) as to insure
macrolactonization under concentrated conditions (entry 4)
and, as for the intramolecular glycosylation, only monomer 23
was detected. Cs2CO3 or KOTf were then added in the reaction
mixture as to favor oligomerization but only traces of dimer
were detected by HRMS (entries 5–7). Interestingly, it was
possible to improve the yield up to 80% for the formation of
target macrolide 23 when KOTf was used as an additive (entry
6).
With protected macrolides 21 and 23 in hand, we were able
to complete the total synthesis of ananatoside A (1) (Scheme 5).
Therefore, the AZMB group of macrolide 21 and the Lev group
of macrolide 23 were orthogonally removed through Staudinger
reduction or treatment with hydrazine acetate, respectively. Pd-
catalyzed hydrogenolysis of the resulting alcohols cleanly led to
the formation of ananatoside A (1) in 56 or 90% yield over two
steps from macrolide 21 or 23, respectively. Physical and
analytical data (Rf, specic rotation, HRMS, and NMR) of
synthetic ananatoside A (1) were in perfect agreement with the
isolated natural product26 (see ESI† for details). The overall
strictly anhydrous conditions throughout the enzymatic process.
As hydrophobic solvents favor such type of enzymatic reactions,39
toluene was rst employed at 75 ꢂC (entry 1). However, aer
a seven-day period, only 30% of conversion was observed (as
estimated by 1H NMR). When switching to hexanes, an even lower
conversion was obtained (entry 2), and only traces of ananatoside
A (1) were detected when the more polar THF was employed as the
reaction medium (entry 3). We were pleased to nd that switching
back to toluene and conducting this enzyme-catalyzed reaction
under microwave irradiations for three hours at 75 ꢂC allowed the
isolation of ananatoside A (1) in a satisfying 64% yield. To the best
of our knowledge, this is the rst report of using CALB for the
enzymatic conversion of a monoglycolipid into a macrolide.
Synthesis of RhaC10C10 (3) and macrolactonized rhamnolipids
4–6
Retrosynthetic analysis. The preparation of microbial
rhamnolipids has only been reported in few occasions in the
literature.55,56,58,59,66,67 We thought it would be worth
Fig. 3 Retrosynthetic analysis of rhamnolipid 3 and macrodilactone-containing rhamnolipids 4–6. PMB: para-methoxybenzyl. Blue: permanent
protecting groups (Bn and PMB); red: temporary protecting groups (TBS); green: esters used as temporary protecting groups enabling neigh-
boring group participation when branched at C2 (AZMB and Lev).
7538 | Chem. Sci., 2021, 12, 7533–7546
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Edge Article
Chemical Science
changes of the rhamnopyranose ring, which is typically found
in the 1C4 conformation, hence making non-trivial the deter-
mination of anomeric congurations in resulting macrolides.
As such, DFT calculations would be used as a theorical tool in
conjunction with NMR spectroscopy to determine the exact
anomeric congurations and ring conformations of rhamno-
lactones 4–6.
Synthesis of building blocks. The synthesis of the thio-
rhamnoside building blocks was straightforward (Scheme 6).
Our approach mainly relied on the stannylene acetal-mediated68
regioselective introduction of either PMB or Bn groups at C3
from known diol 32.36 Then, installation of the AZMB group at
C2 followed by orthogonal cleavage at either C3 or C4 positions
provided functionalized thiorhamnosides 25, 29, 30, and 31.
These optimized procedures allowed us to prepare gram
amounts of each building blocks in pure anomeric forms.
Synthesis of rhamnolipid C10C10. With our newly developed
fully functionalized thiorhamnoside 25 in hand, we were ready
to conduct the synthesis of RhaC10C10 (3).59 As depicted in
Scheme 7, coupling of rhamnosyl donor 25 with dilipid acceptor
12 under the promotion of NIS/AgOTf cleanly provided pro-
tected rhamnolipid 24 in 91% yield with full control of 1,2-trans
stereoselectivity owing to the neighboring group participation
of the AZMB group at C2. Then, global deprotection of deriva-
tive 24 was accomplished uneventfully following a three-step
sequence including Staudinger reduction, chemoselective
cleavage of the Lev ester, and Pd-catalyzed hydrogenolysis of
both benzyl ester and PMB ether functionalities. The overall
yield for the synthesis of RhaC10C10 (3) was 56% over six steps
starting from diol 32. We have also completed the total
synthesis of rhamnolipid 3 from thiorhamnoside S6 (ref. 69)
bearing the commonly used 3,4-O-butanediacetal56,67 as a diol
protecting group (see Scheme S2†). The latter route was less
efficient in terms of overall yield as compared to the former one.
Macrolactonization of rhamnolipids. Our last synthetic
challenge consisted in the preparation of (1/2)-, (1/3)- and
(1/4)-macrolactonized rhamnolipids (4–6). These three types
of macrolactones were obtained following a similar synthetic
route (Scheme 8). Our approach entailed esterication of thio-
rhamnosides 29, 30, or 31 with dilipid 10 by means of DCC/
DMAP coupling reactions to afford derivatives 26, 27, and 28,
respectively, in excellent yields. Subsequent chemoselective
cleavage of the C300 TBS group in the presence of TFA led to
acyclic precursors 34, 36, and 38, respectively, which were ready
to be macrolactonized. Intramolecular glycosylation of
precursor 34 under the activation of NIS/TMSOTf furnished
protected (1/4)-macrolide 35 in 74% yield with full control of
a-selectivity. Then, Staudinger reduction followed by Pd black-
mediated hydrogenolysis provided target (1/4)-macrolide 4.
Because of the presence of the dilipid chain at C2 imposing
steric constraints on the pyranose conformation, intra-
molecular glycosylation of precursor 36 under previously
mentioned conditions led to the formation of an inseparable a/
b anomeric mixture of protected (1/2)-macrolide 37 with the b-
anomer found as the major compound (a/b ¼ 15 : 85).
Following cleavage of both the Lev and Bn groups, a- and b-(1/
2)-macrolides, i.e., compounds 5a and 5b, respectively, were
Scheme 6 Synthesis of thiorhamnoside building blocks 25, 29, 30, and
31.
reinvestigating the synthesis of RhaC10C10 (3) as to obtain
a pure synthetic sample that could be evaluated for its physical
properties and biological activity in direct comparison with
ananatosides. Furthermore, using an orthogonally protected
donor such as thiorhamnoside 25 in which each hydroxyl
groups (C2, C3, and C4) could be selectively unmasked (Fig. 3),
we hypothesized that a series of unprecedented macro-
lactonized rhamnolipids (4–6) could be produced via an intra-
molecular glycosylation strategy. As a parallel to ananatoside A
(1) and B (2) acid/lactone pair, these macrolides would serve as
model compounds to study the impact of macrolactonization
on the biological activity of rhamnolipids. Therefore, as depic-
ted in Fig. 3, rhamnolipid 3 would be obtained from the inter-
molecular glycosylation of rhamnoside 25 with dilipid 12
followed by global deprotection. As for macrolactones 4–6, they
would become accessible via the intramolecular glycosylation of
ester-linked rhamnolipid acyclic precursors 26–38, which would
be formed by Steglich condensation of dilipid 10 with their
corresponding unmasked alcohols (29, 30, and 31). We antici-
pated that the presence of a macrolactone functionality in target
rhamnolipids 4–6 would induce substantial conformational
Scheme 7 Total synthesis of RhaC10C10 (3).
© 2021 The Author(s). Published by the Royal Society of Chemistry
Chem. Sci., 2021, 12, 7533–7546 | 7539

View Article Online
Chemical Science
Edge Article
Scheme 8 Synthesis of (1/4), (1/2), and (1/3)-macrolactonized rhamnolipids 4–6.
isolated in pure forms in a combined yield of 83%. Intra- separable by silica gel chromatography under their protected
molecular glycosylation of precursor 38 also generated forms (39b, 24%, b-anomer; and 39a, 39%, a-anomer).
a mixture of anomers although this time they were easily Orthogonal cleavage of both the AZMB and Lev groups
Fig. 4 3D Structure of the macrolactonized rhamnolipids models 40–42. Only the most stable structure of each conformer is depicted. Pie chart
gives Boltzmann population for standard six-membered ring1 for the three levels of theory used in this study (from top to bottom: mPW1PW91/6-
31G(d,p), mPW1PW91/6-311+G(d,p) and B97-2/pVTZ).
7540 | Chem. Sci., 2021, 12, 7533–7546
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Edge Article
Chemical Science
Table 4 Critical micelle concentration (CMC), surface tension (gCMC),
and hydrodynamic diameter (Z-average) of naturally occurring bio-
surfactants 1–3
completed the total synthesis of target (1/3)-macrolide 6 in
pure a- and b-anomeric forms. The determination of the
anomeric congurations in macrolides 4, 5b, 5a, 6b, and 6a
was accomplished by NMR with the help of DFT calculations as
described in the next section.
Glycolipida
CMCb (mM)
gCMCb (nM mꢁ1
)
Z-Averageb,c (nm)
1
2
3
43 ꢃ 2
63 ꢃ 2
58 ꢃ 1
37 ꢃ 1
28 ꢃ 1
30 ꢃ 1
>800d
99.0
92.1
Molecular modeling of macrolactonized rhamnolipids
a
b
c
Synthetic samples. Data taken at pH 7.0. Hydrodynamic diameter
of micelles measured with a Zetasizer instrument at concentrations of
To determine the anomeric conguration of a monosaccharide,
3
the JH1,H2 coupling constant is commonly measured by 1H
80 mM for ananatoside B (2) and RhaC10C10 (3), and 100 mM for
NMR.70 In the case of rhamnopyranosides, the J values for the
two anomers are too low to be a suitable selection criterion.
Therefore, the dC at C5 (70.0 ppm for a-Rhap or 73.2 ppm for b-
d
ananatoside
A
(1).
Size distribution showed the formation of
aggregated particles.
1
Rhap)71 or the JC1,H1 coupling constant (167.2–172.3 Hz for a-
Rhap or 152.3–159.8 Hz for b-Rhap) are preferred for rhamno-
pyranosides.72 These methods are valid only because rhamno-
pyranosides usually adopt a typical 1C4 conformation. However,
in the case of macrolactonized rhamnolipids 4–6, other
conformations are likely to appear due to the steric constraints
imposed by the bicyclic nature of these compounds, rendering
useless these empirical rules for identifying the proper anomer.
To overcome this problem, in silico models of synthetic rham-
nolipids were prepared with truncated alkyl chains as to mini-
mize the number of conformers to be considered (Fig. S4†). The
conformational space of these models was explored in depth
(Fig. S5†) using the improved RDKit algorithm, which is based
on the geometry of distances and empirical preferences for
certain angles of torsion (ETKDGv2).73 The geometry of the
unique conformers was optimized by molecular mechanics
(MMFF94) followed by quantum molecular modeling using
density functional theory (mPW1PW91) and the basis set 6-
31G(d, p). The thermochemical parameters were calculated to
deduce the abundance of conformers in solution using the
Boltzmann equation, thus eliminating marginal conformers
(<1%). The retained conformers were inspected so that no
imaginary vibrational frequency could be found. Examination
of the sugar conformation of the conserved structures
conrmed the irregular nature of these glycosides. Indeed,
Surfactant properties of natural glycolipids
Biosurfactants are molecules that can associate with each other
leading to the formation of micelles. They can also interact with
surfaces between phases of different polarity such as the air/
water or oil/water interface.77 To gain quantitative insights
into the surface-active properties of microbial glycolipids, the
surface tensions at CMC (gCMC) of ananatoside A (1) and ana-
natoside B (2) were measured using a Fisher tensiometer by
¨
means of the du Nouy ring method. The CMC, referring to the
concentration of surfactant at which micelles start to form, was
determined from a plot of surface tension again each concen-
tration. These data were compared to those obtained for
synthetic RhaC10C10 (3). Due to the presence of a free carboxylic
acid moiety in biosurfactants 2 and 3, the surface tensions were
measured at neutral pH, which corresponds to the presence of
their anionic forms in solution based on an estimated pKa of 5.5
for RhaC10C10 (3).78 As shown in Table 4, the CMC value for
ananatoside B (2) was similar to the one measured for rham-
nolipid 3, denoting that ananatoside B (2) can be considered
a potent biosurfactant. In the case of ananatoside A (1), the
measured CMC value was even lower than for RhaC10C10 (3).
However, the latter value must be taken with cautious as the
presence of the macrolactone functionality decreased consid-
erably the water solubility of ananatoside A (1) as compared to
its open-chain congener. Consequently, the surface tension
measurements were performed quickly to avoid the precipita-
tion of ananatoside A (1) into the aqueous medium. It is inter-
esting to note that the surface tension values measured for
biosurfactants 2 and 3 were on the same order of magnitude
than those measured by Pemberton and co-workers59 for
structurally similar synthetic rhamnolipids bearing C10C10 lipid
chains.
In parallel to these experiments, we took advantage of
dynamic light scattering (DLS) to evaluate the hydrodynamic
diameter of biosurfactant micelles formed at concentrations
above the CMC (Table 4).79 DLS considers the intensity of light
scattered by spherical particles, which prevents its application
for highly polydisperse and non-spherical colloidal systems.
Using a Zetasizer instrument, the results showed that the Z-
average, i.e., the mean diameter of the micelles, was in the same
range for ananatoside B (2) and for rhamnolipid 3 (see Fig. S6†).
Once again, due to its hydrophobicity, ananatoside A (1) was too
three different skew boat (1S3 for 4, S0 for 5a, and S1 for 6b)
along with two chair conformations (4C1 for 6a and 1C4 for 5b)
were found for the main conformers of the modelized
compounds (Fig. 4). The shielding tensors were calculated with
the same level of theory and then converted to chemical shis
using the multiple reference method.74,75 First, the means of the
absolute values of the errors (MAE) were calculated between
each synthetic molecules 4–6 and the corresponding in silico
anomer pairs 40–42 (Table S1†). In most cases, it was possible to
assign the a- and b-anomers to a single model, but the case of
compound 5b was found to be ambiguous at the B97-2/pVTZ
level of theory. Indeed, when only the 1H NMR data were
considered, we found that 5a and 5b would both be of a-
conguration. To ensure the validity of these assignments, the
pooled comparison method proposed by Lauro et al.76 was
performed for the 5a/5b and 6a/6b anomeric pairs (Table S1†).
In this way, the anomeric conguration of each synthetic mac-
rolactone was unambiguously assigned as depicted in Scheme
8.
2
5
© 2021 The Author(s). Published by the Royal Society of Chemistry
Chem. Sci., 2021, 12, 7533–7546 | 7541

View Article Online
Chemical Science
Edge Article
polydisperse for conducting a proper distribution analysis, and macrolides 4–6 was evaluated against Gram-negative, i.e.,
showing a Z-average value above 800 nm, which meant that the Pseudomonas aeruginosa PA14, Pseudomonas aeruginosa LESB58,
particles were aggregated in water at the tested concentration. and Escherichia coli DH5a, and Gram-positive bacteria, i.e.,
Similar results were obtained for ananatoside A (1) when the Staphylococcus aureus MRSA, Staphylococcus aureus Newman,
DLS experiments were conducted in DMSO as the solvent. These and Bacillus subtilis PY79, as well as against fungi, i.e., Candida
results showed that ananatoside B (2), similarly to rhamnolipid albicans ATCC 10231 and Candida albicans LSPQ 0199. At the
3, formed monodisperse colloidal particles at concentrations maximum tested concentration (37.5 mg mLꢁ1), glycolipids 1–6
above its CMC.
were unable to inhibit the growth of these microbes. These
The propensity of surfactants to stabilize emulsions shows results suggest that the biosynthetic production of ananatoside
promising for biotechnological applications such as bioreme- A (1) and ananatoside B (2) by P. ananatis is not aimed for
diation of contaminated soils and water.77 Therefore, the antimicrobial purposes.
emulsifying properties of microbial biosurfactants 1–3 were
As the anticancer potential of microbial rhamnolipids has
evaluated against light as well as heavy hydrocarbon contami- previously been reported,16 we next sought to evaluate the in
nants such as cyclohexane and kerosene, respectively (see vitro cytotoxicity of our synthetic and natural glycolipids 1–6
Fig. S7†). The emulsifying activity results showed that anana- against cancerous cell lines, i.e., human lung carcinoma (A549)
toside B (2) and RhaC10C10 (3) were equally able to generate and colorectal adenocarcinoma (DLD-1). The compounds were
emulsions with cyclohexane and kerosene while ananatoside A also tested against human normal skin broblasts (WS1) to
(1) was unable to form any emulsion, mainly because of its lack estimate their selectivity towards cancer cells. As revealed in
of affinity for water. Altogether, these results once again high- Table 5, very interesting conclusions can be derived from these
lighted that ananatoside B (2), but not its macrolide counterpart cytotoxicity results. First, it appears that only macrolactones, as
1, exhibits potent surfactant properties that could be advanta- compared to their open forms, were able to inhibit the growth of
geously used for biotechnological applications.
human cell lines. Second, the cytotoxic activity of the active
macrolides was not specic to cancer cell lines as comparable
IC50 values were measured against healthy cells (WS1), pointing
toward a general mechanism involving the formation of pores
in cell membrane.16 Specically, ananatoside A (1) was cytotoxic
against human cell lines (IC50 ¼ 50ꢁ58 mM) while its open form
counterpart ananatoside B (2) was inactive at the maximum
tested concentration (IC50 > 200 mM). Furthermore, macro-
lactonization of the inactive RhaC10C10 (3, IC50 > 200 mM)
generated cytotoxic macrolides 4, 5b, and 6b (IC50 ¼ 50–112
mM). While (1/4)-macrolactonized rhamnolipid 4 was the
most active congener, only the b-anomers of (1/2)- and (1/3)-
macrolactonized rhamnolipid 5b and 6b, respectively, were
found to inhibit the growth of human cell lines. These results
showed that both the position of the macrolactone ring and the
anomeric conguration of the glycosidic bonds can impact the
cytotoxicity of the rhamnolipids.
Antimicrobial activity, cytotoxicity, and hemolytic activity of
natural and synthetic glycolipids
As previously mentioned in the introduction section, numerous
studies have reported the antimicrobial potential of rhamnoli-
pids against several pathogenic bacteria and fungi.12 However,
as rhamnolipids were usually tested as mixtures of different
congeners and not as pure synthetic compounds, there seems to
be no clear consensus regarding the antimicrobial activity of
one specic congener. Hence, as an initial investigation toward
their biological functions in bacteria, the in vitro antimicrobial
activity of ananatoside A (1), ananatoside B (2), RhaC10C10 (3),
Table 5 Cytotoxicity and hemolytic activity of natural and synthetic
glycolipids (1–6)
Owing to their surfactant properties, glycolipids can break
the membrane of red blood cells causing hemolytic activity.18 To
understand if the cytotoxicity of macrolactones 1, 4, 5b, and 6b
was due to their surface tension activity, we evaluated the
hemolytic potential of synthetic glycolipids 1–6 against sheep
erythrocytes (Table 5). As anticipated, all the cytotoxic glyco-
lipids were shown to exhibit hemolytic activities on red blood
cells with HC50 ranging from 12 to 14 mM. The non-cytotoxic
rhamnolipid 3 was also found to exert a weak hemolytic
activity (HC50 ¼ 65 ꢃ 6 mM). Altogether, these data suggest
a non-specic mechanism of cytotoxicity involving the interca-
lation of macrolactones 1, 4, 5b, and 6b into the lipid constit-
uents of the cell membranes.
Hemolysis
Cytotoxicity
(IC50 in mM)b
A549
(HC50 in mM)c
Erythrocytes
Glycolipida
DLD-1
WS1
1
2
3
4
5b
5a
6b
6a
50 ꢃ 7
>200d
58 ꢃ 2
>200d
52 ꢃ 4
>200d
12.3 ꢃ 0.3
>200d
>200d
>200d
>200d
65 ꢃ 6
56 ꢃ 2
103 ꢃ 7
>200d
70 ꢃ 10
110 ꢃ 2
>200d
57 ꢃ 1
105 ꢃ 5
>200d
12.0 ꢃ 0.7
14 ꢃ 2
>200d
112 ꢃ 2
112 ꢃ 3
110 ꢃ 1
12.3 ꢃ 0.5
>200d
>200d
>200d
>200d
a
Synthetic samples. b Half maximal inhibitory concentration measured
via the resazurin assay. Etoposide was used as a positive control80
showing IC50 values of 1.2, 27, and 34 mM against A549, DLD-1, and
c
Interaction of natural and synthetic glycolipids with the plant
immune system
WS1 cell lines, respectively. Half maximal inhibitory concentration
measured on sheep blood erythrocytes. Triton X-100 was used as
d
a positive control showing an HC50 value of 52 ꢃ 2 mM. No
Bacterial rhamnolipids, such as compound 3, trigger the plant
inhibition or activity at the maximum tested concentration (IC50 or
HC50 > 200 mM).
immune system, which can ultimately lead to plant protection
7542 | Chem. Sci., 2021, 12, 7533–7546
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Edge Article
Chemical Science
treatment with the synthetic compounds, as an early and
important marker of plant immunity.20,22,82 Tomato leaves were
challenged with synthetic glycolipids 1–6 at concentrations of
100 mM and the production of ROS was monitored over
a 720 min time frame (Fig. 5). As expected, rhamnolipid 3, used
as a positive control in this study, triggered a long-lasting and
strong production of ROS in tomato leaves.
Although less strong in amplitude, a similar pattern of ROS
production was measured for ananatoside B (2), but not for
ananatoside
A (1), its macrolactonic counterpart. These
responses seem to be however plant-specic because, when
tested in Arabidopsis thaliana, both natural glycolipids 1 and 2
were perceived by plants (see Fig. S8†). As for rhamnolactones
4–6, all of them were sensed by tomatoes: the luminescence
curves showed a small burst at 200 min followed by a return to
basal levels over the next 200 min. The only exception to this
general trend was compound 5b. Indeed (1/2)-macro-
lactonized rhamnolipid 5b induced a strong and long-lasting
production of ROS following the initial burst at 200 min. This
response was stereospecic as corresponding a-anomer 5a did
not show a similar sensing pattern. Interestingly, our molecular
modeling showed that macrolide 5b was the only one to be
found in the native 1C4 conformation for the rhamnose ring (see
Fig. 4), which could be part of the explanation for these
intriguing results. Taken together, these results highlight that
(1), glycolipids sharing the same 3-hydroxyalkanoate dilipidic
chain than rhamnolipids but bearing a glucose instead of
a rhamnose moiety can be perceived by plants; and (2) macro-
lactonization of these glycolipids alter the sensing patterns
according to the plant species as well as the conformation of the
rhamnopyranose ring.
Conclusions
In summary, we have isolated and structurally characterized
rhamnolipid-like ananatoside A (1) and ananatoside B (2) from
an organic extract of P. ananatis, a non-human pathogenic
producer of biosurfactants. We have accomplished, for the rst
time, the total synthesis of ananatoside A (1) and ananatoside B
(2), conrming the structure of these bacterial glucolipids. The
macrodilactone-containing ananatoside A (1) was efficiently
synthesized according to three alternative pathways: (1) via the
stereoselective intramolecular glycosylation of a thioglucoside
donor; (2) via the chemical macrolactonization of a seco acid
precursor; and (3) via the direct enzymatic macrolactonization of
its open form congener ananatoside B (2) using a solid-supported
lipase. Capitalizing on the expeditious intramolecular glycosyla-
tion of orthogonally protected thiorhamnoside donors, we have
accomplished the synthesis of diastereoisomerically pure (1/2),
(1/3), and (1/4)-macrolactonized rhamnolipids. We have
determined the anomeric conguration of these unprecedented
macrolides through molecular modeling and found that the
rhamnose ring adopts unusual conformations including skew
boat and ipped chair conformations. Determination of the
surfactant properties of bacterial glucolipids 1 and 2 in compar-
ison with rhamnolipid 3 has revealed that ananatoside B (2), but
Fig.
5 Extracellular reactive oxygen species (ROS) production
following treatment of tomato leaf disks with ananatoside A (1), ana-
natoside B (2), RhaC10C10 (3), and related macrodilactone-containing
rhamnolipids (4–6). Production of ROS was measured in tomato leaf
disks following treatment at 100 mM with synthetic glycolipids (1–6).
MeOH (0.5%) was used as a control. ROS production was measured
using the chemiluminescence of luminol and photon counts were
expressed as relative luminescence units (RLUs). Data are mean ꢃ SEM
(n ¼ 6). Experiments were independently realized three times with
similar results.
against diseases caused by fungal and bacterial pathogens.23 As
ananatoside A (1) and ananatoside B (2) share structural simi-
larities with rhamnolipid 3 and because they have been isolated
from P. ananatis cultures, an emerging plant pathogen causing
important agricultural damage worldwide,81 we wanted to
investigate whether these bacterial glycolipids, along with our
unprecedented series of synthetic rhamnolactones, could be
perceived by the plant immune system. We relied on measuring
the extracellular content of reactive oxygen species (ROS)
produced by tomato plants (Solanum lycopersicum) following
not its macrolide counterpart 1, represents
a
potent
© 2021 The Author(s). Published by the Royal Society of Chemistry
Chem. Sci., 2021, 12, 7533–7546 | 7543

View Article Online
Chemical Science
Edge Article
biosurfactant, as shown by its efficient emulsifying activity.
Furthermore, we have shown that synthetic glycolipids 1–6 do not
exhibit any signicant antimicrobial activity against Gram-
negative and Gram-positive bacteria as well as against fungi.
We have also highlighted that the presence of the macrodilactone
functionality can convert non-cytotoxic glycolipids into cytotoxic
ones according to the position and anomeric conguration of the
macrolides. We have revealed a direct correlation between the
cytotoxicity and the hemolytic activity of these macrolides
pointing towards a mechanism involving the formation of pores
into the lipidic cell membrane. Finally, we have demonstrated
that natural glucolipids 1 and 2 as well as unnatural macrolides
4–6 can be perceived by the plant immune system, and that this
sensing was long-lasting for macrolide 5b featuring a rhamnose
moiety in its native 1C4 conformation. Altogether our results
suggest that macrolactonization of glycolipids can dramatically
interfere with their surfactant properties and biological activity.
Our interdisciplinary study could serve as a foundation for the
rational design of rhamnolipid-like biosurfactants with improved
properties for biotechnological and/or therapeutic applications.
Notes and references
´
1 A. Berces, D. M. Whiteld and T. Nukada, Tetrahedron, 2001,
57, 477–491.
2 S. I. Elshahawi, K. A. Shaaban, M. K. Kharel and
J. S. Thorson, Chem. Soc. Rev., 2015, 44, 7591–7697.
3 I. Mnif and D. Ghribi, J. Sci. Food Agric., 2016, 96, 4310–4320.
`
4 M. Ines and G. Dhouha, Carbohydr. Res., 2015, 416, 59–69.
5 P. Saravanakumari and K. Mani, Bioresour. Technol., 2010,
101, 8851–8854.
6 R. Sha, L. Jiang, Q. Meng, G. Zhang and Z. Song, J. Basic
Microbiol., 2012, 52, 458–466.
7 V. Shah, G. F. Doncel, T. Seyoum, K. M. Eaton, I. Zalenskaya,
R. Hagver, A. Azim and R. Gross, Antimicrob. Agents
Chemother., 2005, 49, 4093–4100.
¨
8 S. Hewald, K. Josephs and M. Bolker, Appl. Environ.
Microbiol., 2005, 71, 3033–3040.
9 J. Chen, X. Song, H. Zhang and Y. Qu, Enzyme Microb.
Technol., 2006, 39, 501–506.
10 V. M. F. Lima, V. L. D. Bonato, K. M. Lima, S. A. Dos Santos,
R. R. Dos Santos, E. D. C. Gonçalves, L. H. Faccioli,
I. T. Brandao, J. M. Rodrigues-Junior and C. L. Silva, Infect.
Immun., 2001, 69, 5305–5312.
Author contributions
´
´
11 A. Abdel-Mawgoud, F. Lepine and E. Deziel, Appl. Microbiol.
M. C. and M.-J. P. synthesized the compounds. M. P., S. L., and
C. G. performed the isolation and structural elucidation of
ananatosides. S. L. and T. F. conducted the DFT calculations.
M.-C. G. performed the antimicrobial and hemolytic assays. J. L.
performed the cytotoxicity assays. S. V., J. C., and S. D. assessed
the compounds against the plant immune system. M. A. D. D. R.
conducted the surfactant assays. All authors analysed and dis-
cussed the results. M. C., S. L., M. P., E. D., and C. G. wrote the
manuscript. E. D and C. G. secured funding for this work.
Biotechnol., 2010, 86, 1323–1336.
12 J. Chen, Q. Wu, Y. Hua, J. Chen, H. Zhang and H. Wang, Appl.
Microbiol. Biotechnol., 2017, 101, 8309–8319.
13 H. Zheng, N. Singh, G. S. Shetye, Y. Jin, D. Li and Y.-Y. Luk,
Bioorg. Med. Chem., 2017, 25, 1830–1838.
14 I. Aleksic, M. Petkovic, M. Jovanovic, D. Milivojevic,
B. Vasiljevic, J. Nikodinovic-Runic and L. Senerovic, Front.
Microbiol., 2017, 8, 2454.
´
15 A. Nickzad and E. Deziel, Lett. Appl. Microbiol., 2014, 58, 447–
453.
16 L. Jiang, C. Shen, X. Long, G. Zhang and Q. Meng, Appl.
Microbiol. Biotechnol., 2014, 98, 10187–10196.
Conflicts of interest
ˆ
˜
17 G. G. F. Leite, J. V. Figueiroa, T. C. M. Almeida, J. L. Valoes,
W. F. Marques, M. D. D. C. Duarte and K. Gorlach-Lira,
Biotechnol. Prog., 2016, 32, 262–270.
There are no conicts to declare.
18 M. Manaargadoo-Catin, A. Ali-Cherif, J.-L. Pougnas and
C. Perrin, Adv. Colloid Interface Sci., 2016, 228, 1–16.
19 M. Zhou and W. Wang, Front. Plant Sci., 2018, 9, 1613.
20 W. P. Luzuriaga-Loaiza, R. Schellenberger, Y. De Gaetano,
F. O. Akong, S. A. Villaume, J. Crouzet, A. Haudrechy,
Acknowledgements
´
This article is dedicated to the memory of Pr. Eric Marsault,
University of Sherbrooke, who passed away in early 2021. This
work was supported by Discovery grants from the Natural
Sciences and Engineering Research Council of Canada (NSERC)
under award number RGPIN-2016-04950 (to C. G.) and RGPIN-
2020-06771 (to E. D.). C. G. was supported by a Fonds de
´
F. Baillieul, C. Clement, L. Lins, F. Allais, M. Ongena,
S. Bouquillon, M. Deleu and S. Dorey, Sci. Rep., 2018, 8, 8534.
21 A. Kutschera, C. Dawid, N. Gisch, C. Schmid, L. Raasch,
´
´
recherche du Quebec – Sante (FRQS) Research Scholars Junior 2
Career Award. S. V., J. C., and S. D. thank the SFR Condorcet FR
CNRS 3417 (project P3IA) and French MESRI for nancial
support. M. A. D. D. R. thank the Microbiology Society (Round 1
Research Visit Grant 2019) and Liverpool John Moores Univer-
sity (ECR Fellowships 2018–2019) for travel funds. M. C. thanks
¨
T. Gerster, M. Schaffer, E. Smakowska-Luzan, Y. Belkhadir,
A. C. Vlot, C. E. Chandler, R. Schellenberger, D. Schwudke,
R. K. Ernst, S. Dorey, R. Huckelhoven, T. Hofmann and
S. Ranf, Science, 2019, 364, 178–181.
¨
22 M. Robineau, S. Le Guenic, L. Sanchez, L. Chaveriat,
V. Lequart, N. Joly, M. Calonne, C. Jacquard, S. Declerck,
P. Martin, S. Dorey and E. A. Barka, Molecules, 2020, 25, 3108.
23 J. Crouzet, A. Arguelles-Arias, S. Dhondt-Cordelier,
´
NSERC and Fonds de recherche du Quebec – Nature et tech-
nologies (FRQNT) for M.Sc. and Ph.D. scholarships. M. J. P.
thanks NSERC for an undergraduate scholarship. We thank
Sylvain Milot, INRS, for valuable discussions.
ˇ
S. Cordelier, J. Prsic, G. Hoff, F. Mazeyrat-Gourbeyre,
7544 | Chem. Sci., 2021, 12, 7533–7546
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Edge Article
Chemical Science
´
F. Baillieul, C. Clement, M. Ongena and S. Dorey, Front.
A. Phalipon and L. A. Mulard, Org. Biomol. Chem., 2014, 12,
4218–4232.
Bioeng. Biotechnol., 2020, 8, 1014.
´
´
24 G. Soberon-Chavez, F. Lepine and E. Deziel, Appl. Microbiol. 48 B. Neises and W. Steglich, Angew. Chem., Int. Ed. Engl., 1978,
Biotechnol., 2005, 68, 718–725. 17, 522–524.
25 D. D. N. Smith, A. Nickzad, E. Deziel and J. Stavrinides, 49 A. Parenty, X. Moreau and J. M. Campagne, Chem. Rev., 2006,
mSphere, 2016, 1, e00075-15. 106, 911–939.
26 C. Gauthier, S. Lavoie, M. Piochon, S. Martinez, S. Milot and 50 A. Furstner, M. Albert, J. Mlynarski, M. Matheu and
´
¨
´
E. Deziel, Carbohydr. Res., 2019, 471, 13–18.
27 U. A. Ochsner, A. Fiechter and J. Reiser, J. Biol. Chem., 1994, 51 S.-Y. Zhu, J.-S. Huang, S.-S. Zheng, K. Zhu and J.-S. Yang, Org.
269, 19787–19795. Lett., 2013, 15, 4154–4157.
28 R. Rahim, U. A. Ochsner, C. Olvera, M. Graninger, 52 J. R. Mateus-Ruiz and A. Cordero-Vargas, J. Org. Chem., 2019,
E. DeClercq, J. Am. Chem. Soc., 2003, 125, 13132–13142.
P. Messner, J. S. Lam and G. Soberon-Chavez, Mol.
Microbiol., 2001, 40, 708–718.
29 J. Xie and N. Bogliotti, Chem. Rev., 2014, 114, 7678–7739.
84, 11848–11855.
53 A. Hassner, G. Strand, M. Rubinstein and A. Patchornik, J.
Am. Chem. Soc., 1975, 97, 1614–1615.
´
30 J. Rodriguez, S. O'Neill and M. A. Walczak, Nat. Prod. Rep., 54 B. Menhour, P. Mayon, K. Ple, S. Bouquillon, S. Dorey,
´
2018, 35, 220–229.
C. Clement, M. Deleu and A. Haudrechy, Tetrahedron Lett.,
31 A. M. Abdel-Mawgoud and G. Stephanopoulos, Synth. Syst.
Biotechnol., 2018, 3, 3–19.
32 R. Pereda-Miranda and M. Bah, Curr. Top. Med. Chem., 2003,
3, 111–131.
2015, 56, 1159–1161.
55 A. A. Compton, B. S. Deodhar, A. Fathi and J. E. Pemberton,
ACS Sustainable Chem. Eng., 2020, 8, 8918–8927.
¨
56 J. Bauer, K. Brandenburg, U. Zahringer and J. Rademann,
´
33 M. Tamigney Kenfack, Y. Bleriot and C. Gauthier, J. Org.
Chem.–Eur. J., 2006, 12, 7116–7124.
Chem., 2014, 79, 4615–4634.
34 A. Laroussarie, B. Barycza, H. Andriamboavonjy,
57 M. De Vleeschouwer, D. Sinnaeve, J. Van den Begin,
T. Coenye, J. C. Martins and A. Madder, Chem.–Eur. J.,
2014, 20, 7766–7775.
´
M. Tamigney Kenfack, Y. Bleriot and C. Gauthier, J. Org.
Chem., 2015, 80, 10386–10396.
35 M. Tamigney Kenfack, M. Mazur, T. Nualnoi, T. L. Shaffer,
58 C. Coss, T. Carrocci, R. M. Maier, J. E. Pemberton and R. Polt,
Helv. Chim. Acta, 2012, 95, 2652–2659.
´
A. Ngassimou, Y. Bleriot, J. Marrot, R. Marchetti, 59 R. P. Pacheco, R. J. Eismin, C. S. Coss, H. Wang, R. M. Maier,
K. Sintiprungrat, N. Chantratita, A. Silipo, A. Molinaro,
D. P. AuCoin, M. N. Burtnick, P. J. Brett and C. Gauthier,
Nat. Commun., 2017, 8, 115.
R. Polt and J. E. Pemberton, J. Am. Chem. Soc., 2017, 139,
5125–5132.
60 Y.-P. Hu, S.-Y. Lin, C.-Y. Huang, M. Zulueta, J.-Y. Liu,
W. Chang and S.-C. Hung, Nat. Chem., 2011, 3, 557–563.
36 M. Cloutier, E. Delar, K. Muru, S. Ndong, R. R. Hoyeck,
T. Kaewarpai, N. Chantratita, M. N. Burtnick, P. J. Brett 61 P. Konradsson, U. E. Udodong and B. Fraser-Reid,
and C. Gauthier, Org. Biomol. Chem., 2019, 17, 8878–8901. Tetrahedron Lett., 1990, 31, 4313–4316.
37 T. Vehovec and A. Obreza, J. Chromatogr. A, 2010, 1217, 62 A. M. Felix, E. P. Heimer, T. J. Lambros, C. Tzougraki and
1549–1556.
J. Meienhofer, J. Org. Chem., 1978, 43, 4194–4196.
63 J. S. Bajwa, Tetrahedron Lett., 1992, 33, 2299–2302.
¨
38 A. Furstner, Eur. J. Org. Chem., 2004, 943–958.
39 K. S. Bisht, S. Bhatt and K. Muppalla, Tetrahedron Lett., 2006, 64 E. P. Boden and G. E. Keck, J. Org. Chem., 1985, 50, 2394–
47, 8645–8649. 2395.
40 C. Gagnon, E. Godin, C. Minozzi, J. Sosoe, C. Pochet and 65 T. Fujisawa, T. Mori, K. Fukumoto and T. Sato, Chem. Lett.,
S. K. Collins, Science, 2020, 367, 917–921. 1982, 11, 1891–1894.
41 C. Ortiz, M. L. Ferreira, O. Barbosa, J. C. S. dos Santos, 66 P. Westerduin, P. E. De Haan, M. J. Dees and J. H. van Boom,
R. C. Rodrigues, A. Berenguer-Murcia, L. E. Briand and Carbohydr. Res., 1988, 180, 195–205.
R. Fernandez-Lafuente, Catal. Sci. Technol., 2019, 9, 2380– 67 H. I. Duynstee, M. J. van Vliet, G. A. van der Marel and
2420. J. H. van Boom, Eur. J. Org. Chem., 1998, 303–307.
42 R. Das and B. Mukhopadhyay, ChemistryOpen, 2016, 5, 401– 68 S. David and S. Hanessian, Tetrahedron, 1985, 41, 643–663.
´
433.
69 I. Cumpstey, A. J. Fairbanks and A. J. Redgrave, Tetrahedron,
2004, 60, 9061–9074.
43 Y. Yang, X. Zhang and B. Yu, Nat. Prod. Rep., 2015, 32, 1331–
1355.
70 J. Ø. Duus, C. H. Gotfredsen and K. Bock, Chem. Rev., 2000,
100, 4589–4614.
44 X. G. Jia and A. V. Demchenko, Beilstein J. Org. Chem., 2017,
13, 2028–2048.
71 L. Feng, S. N. Senchenkova, W. Wang, A. S. Shashkov, B. Liu,
S. D. Shevelev, D. Liu, Y. A. Knirel and L. Wang, Gene, 2005,
355, 79–86.
45 G. Lian, X. Zhang and B. Yu, Carbohydr. Res., 2014, 403, 13–
22.
46 T. Wada, A. Ohkubo, A. Mochizuki and M. Sekine, 72 K. Bock and C. Pedersen, J. Chem. Soc., Perkin Trans., 1974, 2,
Tetrahedron Lett., 2001, 42, 1069–1072. 293–297.
47 C. Gauthier, P. Chassagne, F.-X. Theillet, C. Guerreiro, 73 S. Wang, J. Witek, G. A. Landrum and S. Riniker, J. Chem. Inf.
F. Thouron, F. Nato, M. Delepierre, P. J. Sansonetti,
Model., 2020, 60, 2044–2058.
© 2021 The Author(s). Published by the Royal Society of Chemistry
Chem. Sci., 2021, 12, 7533–7546 | 7545

View Article Online
Chemical Science
Edge Article
´
74 A. M. Sarotti and S. C. Pellegrinet, J. Org. Chem., 2009, 74, 78 A. Lebron-Paler, J. E. Pemberton, B. A. Becker, W. H. Otto,
7254–7260.
C. K. Larive and R. M. Maier, Anal. Chem., 2006, 78, 7649–
7658.
79 H. Zhong, L. Yang, X. Yang, G. Zeng, Z. Liu, Y. Liu and
X. Yuan, RSC Adv., 2015, 5, 88578–88582.
80 C. Gauthier, J. Legault, S. Lavoie, S. Rondeau, S. Tremblay
and A. Pichette, J. Nat. Prod., 2009, 72, 72–81.
75 A. M. Sarotti and S. C. Pellegrinet, J. Org. Chem., 2012, 77,
6059–6065.
76 G. Lauro, P. Das, R. Riccio, D. S. Reddy and G. Bifulco, J. Org.
Chem., 2020, 85, 3297–3306.
77 A. Daverey and K. Pakshirajan, in Microbes and Microbial
Technology, ed. I. Ahmad, F. Ahmad and J. Pichtel, 81 T. A. Coutinho and S. N. Venter, Mol. Plant Pathol., 2009, 10,
Springer, New York, NY, 2011.
325–335.
82 J. Qi, J. Wang, Z. Gong and J. M. Zhou, Curr. Opin. Plant Biol.,
2017, 38, 92–100.
7546 | Chem. Sci., 2021, 12, 7533–7546
© 2021 The Author(s). Published by the Royal Society of Chemistry