Biosynthesis-Assisted Structural Elucidation of the Bartolosides, Chlorinated Aromatic Glycolipids from Cyanobacteria

Article abstract of DOI:10.1002/anie.201503186

The isolation of the bartolosides, unprecedented cyanobacterial glycolipids featuring aliphatic chains with chlorine substituents and C-glycosyl moieties, is reported. Their chlorinated dialkylresorcinol (DAR) core presented a major structural-elucidation challenge. To overcome this, we discovered the bartoloside (brt) biosynthetic gene cluster and linked it to the natural products through in vitro characterization of the DAR-forming ketosynthase and aromatase. Bioinformatic analysis also revealed a novel potential halogenase. Knowledge of the bartoloside biosynthesis constrained the DAR core structure by defining key pathway intermediates, ultimately allowing us to determine the full structures of the bartolosides. This work illustrates the power of genomics to enable the use of biosynthetic information for structure elucidation.

Full text of DOI:10.1002/anie.201503186

Angewandte
Chemie
International Edition: DOI: 10.1002/anie.201503186
German Edition: DOI: 10.1002/ange.201503186
Structure Elucidation
Biosynthesis-Assisted Structural Elucidation of the Bartolosides,
Chlorinated Aromatic Glycolipids from Cyanobacteria
Pedro N. Le¼o, Hitomi Nakamura, Margarida Costa, Alban R. Pereira, Rosµrio Martins,
Vitor Vasconcelos,* William H. Gerwick,* and Emily P. Balskus*
Abstract: The isolation of the bartolosides, unprecedented
cyanobacterial glycolipids featuring aliphatic chains with
chlorine substituents and C-glycosyl moieties, is reported.
Their chlorinated dialkylresorcinol (DAR) core presented
a major structural-elucidation challenge. To overcome this,
we discovered the bartoloside (brt) biosynthetic gene cluster
and linked it to the natural products through in vitro character-
ization of the DAR-forming ketosynthase and aromatase.
Bioinformatic analysis also revealed a novel potential halo-
genase. Knowledge of the bartoloside biosynthesis constrained
the DAR core structure by defining key pathway intermediates,
ultimately allowing us to determine the full structures of the
bartolosides. This work illustrates the power of genomics to
enable the use of biosynthetic information for structure
elucidation.
of four unusual secondary metabolites and the use of
biosynthesis in their structure elucidation.
Examination of a fractionated crude organic extract from
the filamentous cyanobacterium Nodosilinea sp. LEGE 06102
1
led to the H NMR spectroscopy guided isolation of barto-
loside A (1, 2.0 mg; Figure 1a). Intriguingly, analogous
A
s ancient photoautotrophs, cyanobacteria have a differ-
entiated ecophysiology among prokaryotes[1] that is abun-
dantly reflected in their secondary metabolite diversity.[2] We
have explored the chemical diversity of an in-house culture
collection of cyanobacteria[3] and herein report the discovery
Figure 1. The bartolosides are chlorinated aromatic glycolipids found
in cyanobacteria belonging to different genera. Optical photomicro-
graphs of the producing organisms and structures of: a) bartoloside A
(1) and b) bartolosides B, C, and D (2–4). Scale bar: 10 mm.
[*] Dr. P. N. Le¼o, H. Nakamura, Prof. E. P. Balskus
Department of Chemistry and Chemical Biology
Harvard University
metabolites, namely bartolosides B, C, and D (2, 5.6 mg; 3,
0.7 mg; and 4, 0.3 mg; Figure 1b), were isolated through
bioassay-guided fractionation of a crude extract of Synecho-
cystis salina LEGE 06155, a unicellular, free-living cyanobac-
terium found at the same beach as the bartoloside A
producer.[4] These cyanobacterial strains are morphologically
and phylogenetically distant (Figure 1; see also the Support-
ing Information, Figure S1). Together with their geographical
co-occurrence, this observation may suggest pathway acquis-
ition by horizontal gene transfer.
12 Oxford Street, Cambridge MA, 02138 (USA)
E-mail: balskus@chemistry.harvard.edu
Dr. P. N. Le¼o, Dr. A. R. Pereira, Prof. W. H. Gerwick
Scripps Institution of Oceanography
University of California, San Diego
La Jolla, CA 92093 (USA)
E-mail: wgerwick@ucsd.edu
Dr. P. N. Le¼o, M. Costa, Prof. R. Martins, Prof. V. Vasconcelos
Interdisciplinary Centre of Marine and Environmental Research
(CIIMAR/CIMAR), University of Porto
Rua dos Bragas, 289, 4050-123 Porto (Portugal)
E-mail: vmvascon@fc.up.pt
Initial structural-elucidation efforts focused on the most
1
Prof. R. Martins
abundant metabolites (1 and 2). A combination of H, 13C,
Centre of Health and Environmental Research—CISA, ESTSP
Polytechnic Institute of Porto
Rua Valente Perfeito 322, 4400-330 Vila Nova de Gaia (Portugal)
HSQC, HMBC and COSY data revealed a dialkylresorcinol
(DAR) scaffold in both compounds (Supporting Information,
S7–S39). d-Xylose and l-rhamnose were O-linked to the
aromatic rings in 1 and 2, respectively (Figure 2a, see also
Figure S33). The DAR scaffold in 2 was further decorated
with a xylose residue C-linked to the C4 position (Figure 2a).
To the best of our knowledge, this is the first report of
a cyanobacterial C-glycoside.
Prof. V. Vasconcelos
Department of Biology, Faculty of Sciences, University of Porto
Rua do Campo Alegre, 4169-007 Porto (Portugal)
Prof. W. H. Gerwick
Skaggs School of Pharmacy and Pharmaceutical Sciences
University of California, San Diego
La Jolla, CA 92093 (USA)
The two alkyl substituents on the resorcinol ring
accounted for 25 and 27 carbon atoms in 1 and 2, respectively.
Considering the HR-ESI-MS derived molecular formulae
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201503186.
Angew. Chem. Int. Ed. 2015, 54, 11063 –11067
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
11063

.
Angewandte
Communications
synthesize DAR scaffolds in a convergent process.[10] DarB
catalyzes the head-to-head condensation of a b-ketoacyl acyl
carrier protein (ACP) thioester with an a,b-unsaturated acyl
ACP thioester, resulting in a dialkylcyclohexanedione car-
boxylic acid (carboxy-CHD).[11] DarA is thought to be an
aromatase based on its in vivo activity in a heterologous
host,[11] but has not been characterized in vitro. Homologues
of darA and darB are contiguous in many prokaryotic
genomes, including several cyanobacteria.[11] Owing to the
multiple cyanobacterial DAR biosynthetic pathways, we
could not be sure which biosynthetic machinery would be
involved in bartoloside production.
To identify the bartoloside biosynthetic pathway, we
sequenced the genome of S. salina LEGE 06155 and searched
for all of the known DAR biosynthetic pathways. We
identified adjacent darA and darB homologues co-localized
with additional genes encoding enzymes with predicted
functions consistent with a role in bartoloside biosynthesis,
including glycosyltransferases and a putative halogenase
(Figure 3, Table S7). We designated this region to be the
Figure 2. NMR-derived structural information for 2. a) Fragments
obtained from 1D and 2D NMR spectroscopy, with key HMBC
correlations highlighted. b) 1D TOCSY experiments with irradiation of
the benzylic protons on each chain reveal propagation up to methylene
groups vicinal to the chlorinated methine moieties.
(C36H62Cl2O6 for 1 and C44H76Cl2O10 for 2) and NMR-based
partial structures, these groups had to be linear. Overlapping
1H and 13C resonances for the two chlorinated methine groups
and their corresponding 2J and 3J HMBC correlations to
a total of eight methylene groups placed the Cl atoms on the
alkyl chains (Figure 2a) and eliminated the possibility of their
vicinal positioning as observed in several naturally occurring
marine lipids.[5] The termini of all partial structures showed
HMBC and COSY correlations to the methylene envelope,
hindering further assignments. Under ESI conditions, the
molecules readily fragmented by loss of HCl and the sugar
moieties. Alkyl chain fragmentations at higher energy did not
allow for inference of the Cl atom or relative alkyl chain
positions, but did confirm that C12/C13 chains were present in
1 and C12/C15 chains in 2, 3, and 4 (Supporting Information,
S7–S51). Selective irradiation of the benzylic protons in 1D
TOCSY experiments revealed that both alkyl chains were
chlorinated in 2 (Figure 2b) as they were in 1 (Supporting
Information, S7–S9). Owing to our inability to distinguish the
two very similar alkyl groups, an assignment of the chain
lengths and the determination of the precise Cl atom positions
were not possible using NMR spectroscopy and MS alone.
To gain additional structural insights, we investigated
bartoloside biosynthesis. The DAR scaffold is found in just
a few cyanobacterial metabolites: the paracyclophanes,[6]
microcarbonin A,[7] and nostocyclyne A (Figure S40).[8] In
the biosynthesis of the paracyclophanes, DAR formation
involves the type III polyketide synthase (PKS) catalyzed
formation of a monoalkylresorcinol followed by an intermo-
lecular alkylation reaction.[9] An analogous intramolecular
cyclization is conceivable for the biosynthesis of nostocycly-
ne A (Figure S40). In other bacteria, a ketosynthase (DarB)
and a putative aromatase (DarA) have been shown to
Figure 3. Biosynthesis of the bartolosides. a) Organization of the brt
gene cluster from S. salina LEGE 06155 and annotated gene functions.
b) Proposal for the biosynthesis of the dialkylresorcinol scaffold found
in 2–4.
bartoloside (brt) gene cluster (GenBank: KR059027). In
contrast to previously characterized darAB-containing gene
clusters,[10–12] the brt cluster lacks a designated ACP, suggest-
ing the use of a trans-acting ACP or ACP thioesters from fatty
acid biosynthesis. The participation of acyl-CoA substrates is
unlikely as cyanobacteria, including S. salina LEGE 06155,
lack a canonical b-oxidation pathway.[13]
Based on the brt cluster architecture, we envisioned that
the biosynthesis of 2 could proceed through the BrtD-
catalyzed condensation of the 2-octadecenoyl-ACP thioester
with the 3-oxo-hexadecanoyl-ACP thioester. Both precursors
are derivatives of abundant (C18 and C16) cyanobacterial fatty
acids.[14] This reaction would yield carboxy-CHD 5, which
could spontaneously decarboxylate to give dione 6 or undergo
BrtC-catalyzed oxidative aromatization to give the DAR
backbone of 2–4 (7), which contains C12 and C15 chains at the
C2 and C5 positions, respectively (Figure 3b). Alternatively,
a DAR scaffold in which the positions of these substituents
11064 www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 11063 –11067

Angewandte
Chemie
are reversed could result from condensation of 2-pentadece-
noyl-ACP with 3-oxo-nonadecanoyl-ACP (Figure S36). This
scenario is highly unlikely given that C15 and C19 fatty acids
are not produced by cyanobacteria.[14] Therefore, both the
logic of bartoloside biosynthesis and the availability of
biosynthetic precursors constrain the DAR scaffold of 2–4.
A similar analysis revealed that the assembly of 1 should
utilize two C16 fatty acyl ACP derivatives, leading to a DAR
with C12 and C13 substituents at the C2 and C5 positions.
After DAR formation, the brt-encoded glycosyltransfer-
ases should decorate the resorcinol ring with xylose and
rhamnose. Chlorine installation presumably requires halo-
genation at unactivated carbon atoms, a transformation
typically performed by a-ketoglutarate-dependent non-
heme iron halogenases.[15] However, this type of enzyme
was not found in the S. salina genome. Instead, we predict that
BrtJ is the halogenase based on its homology to CylC[9a] and
ColD/ColE[16] (41% identity, 55–59% similarity), enzymes
encoded in the biosynthetic gene clusters of the halogenated
cyanobacterial natural products cylindrocyclophane and
columbamide.[16–17] BrtJ and its homologues do not resemble
known halogenases and therefore likely represent a new
enzyme class capable of halogenating unactivated carbon
centers. Given their weak homology to the N-oxygenase AurF
(BrtJ: 15% identity, 26% similarity, HHPred E-value:[18] 2.4 ”
10À29), these enzymes may be diiron halogenases,[19] and this
possibility will be the subject of future investigations.
To confirm the involvement of the brt gene cluster in
bartoloside biosynthesis, BrtD and BrtC were overexpressed
in E. coli BL21 cells and purified. According to HR-MS
analysis, BrtC bound flavin mononucleotide (FMN; Fig-
ure S37). N-Acetylcysteamine (SNAC) thioester analogues of
the predicted BrtD substrates (8 and 9) were synthesized and
incubated with BrtD; both 5 and 6 were detected in the assay
mixture by LC-HR-ESI-MS, which is in agreement with our
biosynthetic hypothesis (Figure 4). When BrtC was included
in the assay, DAR 7 was observed (Figure 4). These findings
provide the first biochemical evidence that DarA homologues
are aromatases. Furthermore, the in vivo production of DAR-
like metabolites by E. coli transformed with brtCD was
observed (Supporting Information, S63, Figure S38). Overall,
these observations confirmed the ability of BrtD and BrtC to
utilize C16 and C18 fatty acyl thioesters for DAR synthesis and
provided experimental verification of the biosynthetic
hypothesis that enabled structure elucidation.
Figure 4. BrtD and BrtC synthesize the DAR core of bartoloside B.
a) Schematic representation of the N-His6-tagged BrtD and BrtC
coupled assay. b) LC-HR-ESI-MS derived extracted ion chromatograms
(EICs) showing diketone production by N-His6-BrtD and DAR produc-
tion by N-His6-BrtC. All chromatograms have the same scale. c) HR-
MS2 characterization of the reaction products 5, 6, and 7 (dashed lines
represent ions generated by a fragmentation on an alkyl chain).
Figure S2). The overall structures of 3 and 4 were deduced by
comparison of their NMR and HRMS data with those of 2
(Supporting Information, S40–S51). A future challenge lies in
establishing the absolute configuration of these glycolipids,
which were not amenable to crystallization.
The precise positioning of the Cl atoms in the bartolo-
sides, cylindrocyclophanes, and columbamides suggests that
halogenation by BrtJ homologues is highly specific. Aside
from the occurrence of this halogenation in cyanobacterial
DARs of distinct biosynthetic origin, the similarity of the
5-alkynyl moieties in microcarbonin A and nostocyclyne A is
noteworthy (Figure S40). Together, these observations illus-
trate the functional convergence of two distinct biosynthetic
logics and could suggest a highly selected role for DARs in
cyanobacteria. Recognizing that microcarbonin A was iso-
lated on the basis of a strong photosynthesis-inhibiting
activity,[7] and that DevBCA export machinery is encoded in
the brt cluster, we envisioned that the bartolosides could act
as allelochemicals.[2] Compound 2 was indeed a major organic
component in S. salina LEGE 06155 culture supernatants
(Figure S39). However, we did not observe strong bioactivity
in anticyanobacterial, antialgal, or cytotoxicity assays (Sup-
porting Information, S54–S55). Furthermore, although DARs
are implicated in quorum-sensing,[20] we were unable to locate
homologues of the required components of the signaling
The only remaining unknown variable in the planar
structure of 2 was the position of the Cl atoms. At this point,
the acquisition of HSQC-TOCSY data for 2 established the
connection of the benzylic moiety in the C15 chain to the
chlorinated methine group by identification of four bridging
methylene groups (Figure 5a–c; Supporting Information,
S20–S39). The w-7 position of the Cl atom in the C12 chain
was similarly elucidated by the identification of a previously
unassigned bridging methylene group in conjunction with
a diagnostic HMBC correlation from the deshielded w-6
methylene protons to the w-3 carbon atom (Figure 5a,d,e;
Supporting Information, S20–S39), thus finalizing the struc-
ture of 2. Our proposed structure for 1 is based on NMR data
highly similar to that of 2 (Supporting Information, S7–S19,
Angew. Chem. Int. Ed. 2015, 54, 11063 –11067
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 11065

.
Angewandte
Communications
Figure 5. NMR-based (HSQC-TOCSY, 600 MHz and HMBC, 400 MHz) deduction of the position of the Cl atoms in each chain in 2. a) Proposed
positioning and key HMBC and HSQC-TOCSY correlations leading to the structural assignment. b) HSQC-TOCSY spectrum of 2, highlighting the
spectral deconvolution leading to the assignment of the methylene group resonating at d=1.31 ppm in the C15 chain. c) Key HSQC-TOCSY
correlations (circles) connecting the benzylic position to the chlorinated position in the C15 chain. d) Assignment of the bridging methylene group
in the C12 chain by HSQC-TOCSY analysis (key correlation circled). e) HMBC data highlighting the correlation (circled) from a pair of
diastereotopic protons (vicinal to the chlorinated position) to the w-3 methylene carbon atom.
system PauR or LuxR in the genome of S. salina
LEGE 06155.
in this pathway and the natural roles of the bartolosides are
underway.
In conclusion, the bartolosides are secondary metabolites
produced by cyanobacterial genera from which few (Syne-
chocystis)[21] or no (Nodosilinea) natural products are known.
The chlorinated alkyl substituents of these molecules pre-
sented a substantial challenge for structure elucidation. By
investigating the biosynthesis of the bartolosides prior to
completing the structural assignment, we obtained crucial
information that ultimately enabled us to solve the complete
planar structures and majority of the stereocenters. Given the
increasing accessibility of genomic data, we envision that this
general strategy will be a powerful adjunct for structure
elucidation. Efforts to study the unusual biosynthetic events
Acknowledgements
We thank CEMUP for NMR and MS analyses, A. Fenner, E.
Mevers, P. Boudreau, and G. Byrd for HR-MS analyses, M.
Bertin for optical rotation measurements, and S. Banik for
help with sugar analysis. P.N.L. was supported by FCT
(SFRH/BPD/70233/2010 and IF/01358/2014). H.N. was sup-
ported by the NSF GRFP (DGE1144152). This work was
funded by the FCT grants PTDC/MAR-BIO/2818/2012 (to
P. N.L.), PTDC/MAR/102638/2008 (to R.M.), and PEst-C/
11066 www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 11063 –11067

Angewandte
Chemie
[10] B. Nowak-Thompson, P. E. Hammer, D. S. Hill, J. Stafford, N.
MAR/LA0015/2013, by the ON.2 Program (MARBIOTECH
- NORTE-07-0124-FEDER-000047) and project MARVA-
LOR (V.V.), by the NIH (CA108874 and CA100851 to
W.H.G.), and by the Searle Scholars Program (E.P.B.).
Torkewitz, T. D. Gaffney, S. T. Lam, I. Molnµr, J. M. Ligon, J.
Bacteriol. 2003, 185, 860 – 869.
[11] S. W. Fuchs, K. A. J. Bozhüyük, D. Kresovic, F. Grundmann, V.
Dill, A. O. Brachmann, N. R. Waterfield, H. B. Bode, Angew.
Chem. Int. Ed. 2013, 52, 4108 – 4112; Angew. Chem. 2013, 125,
4202 – 4206.
Keywords: biosynthesis · cyanobacteria · dialkylresorcinol ·
glycolipids · structure elucidation
[12] T. A. Schçner, S. W. Fuchs, B. Reinhold-Hurek, H. B. Bode,
PLOS ONE 2014, 9, e90922.
[13] S. von Berlepsch, H.-H. Kunz, S. Brodesser, P. Fink, K. Marin,
U.-I. Flügge, M. Gierth, Plant Physiol. 2012, 159, 606 – 617.
[14] a) X. Y. Liu, J. Sheng, R. Curtiss, Proc. Natl. Acad. Sci. USA
2011, 108, 6899 – 6904; b) C. N. Kenyon, J. Bacteriol. 1972, 109,
827 – 834.
[15] F. H. Vaillancourt, E. Yeh, D. A. Vosburg, S. Garneau-Tsodi-
kova, C. T. Walsh, Chem. Rev. 2006, 106, 3364 – 3378.
[16] K. Kleigrewe, J. Almaliti, I. Y. Tian, R. B. Kinnel, A. Korobey-
nikov, E. A. Monroe, B. M. Duggan, V. Di Marzo, D. H. Sher-
man, P. C. Dorrestein, L. Gerwick, W. H. Gerwick, J. Nat. Prod.
2015, 78, 1671 – 1682.
[17] G. E. Chlipala, M. Sturdy, A. Krunic, D. D. Lantvit, Q. Shen, K.
Porter, S. M. Swanson, J. Orjala, J. Nat. Prod. 2010, 73, 1529 –
1537.
[18] J. Sçding, Bioinformatics 2005, 21, 951 – 960.
[19] N. Li, V. K. Korboukh, C. Krebs, J. M. Bollinger, Proc. Natl.
Acad. Sci. USA 2010, 107, 15722 – 15727.
How to cite: Angew. Chem. Int. Ed. 2015, 54, 11063–11067
Angew. Chem. 2015, 127, 11215–11219
[1] A. Oren in Cyanobacteria: An Economic Perspective (Eds.: N. K.
Sharma, A. K. Rai, L. J. Stal), Wiley, Hoboken, 2014, pp. 1 – 20.
[2] P. N. Le¼o, N. Engene, A. Antunes, W. H. Gerwick, V. Vascon-
celos, Nat. Prod. Rep. 2012, 29, 372 – 391.
[3] a) P. N. Le¼o, V. Ramos, P. B. Goncalves, F. Viana, O. M. Lage,
W. H. Gerwick, V. M. Vasconcelos, Mar. Drugs 2013, 11, 1316 –
1335; b) P. N. Le¼o, M. Costa, V. Ramos, A. R. Pereira, V. C.
Fernandes, V. F. Domingues, W. H. Gerwick, V. M. Vasconcelos,
R. Martins, PloS ONE 2013, 8, e69562.
[4] A. Brito, V. Ramos, R. Seabra, A. Santos, C. L. Santos, M. Lopo,
S. Ferreira, A. Martins, R. Mota, B. Fraz¼o, R. Martins, V.
Vasconcelos, Syst. Appl. Microbiol. 2012, 35, 110 – 119.
[5] V. M. Dembitsky, M. Srebnik, Prog. Lipid Res. 2002, 41, 315 –
367.
[6] B. S. Moore, J. L. Chen, G. M. L. Patterson, R. E. Moore, L. S.
Brinen, Y. Kato, J. Clardy, J. Am. Chem. Soc. 1990, 112, 4061 –
4063.
[20] S. Brameyer, D. Kresovic, H. B. Bode, R. Heermann, Proc. Natl.
Acad. Sci. USA 2015, 112, 572 – 577.
[21] D. G. Nagle, W. H. Gerwick, Tetrahedron Lett. 1995, 36, 849 –
852.
[7] D. Beresovsky, O. Hadas, A. Livne, A. Sukenik, A. Kaplan, S.
Carmeli, Isr. J. Chem. 2006, 46, 79 – 87.
[8] A. Ploutno, S. Carmeli, J. Nat. Prod. 2000, 63, 1524 – 1526.
[9] a) H. Nakamura, H. A. Hamer, G. Sirasani, E. P. Balskus, J. Am.
Chem. Soc. 2012, 134, 18518 – 18521; b) H. Nakamura, E. P.
Balskus, Synlett 2013, 1464 – 1470.
Received: April 7, 2015
Revised: June 15, 2015
Published online: July 31, 2015
Angew. Chem. Int. Ed. 2015, 54, 11063 –11067
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 11067