Two 3′-: O-β-glucosylated nucleoside fluorometabolites related to nucleocidin in Streptomyces calvus

Article abstract of DOI:10.1039/c9sc03374b

The antibiotic nucleocidin is a product of the soil bacterium Streptomyces calvus T-3018. It is among the very rare fluorine containing natural products but is distinct from the other fluorometabolites in that it is not biosynthesised from 5′-fluorodeoxya

Full text of DOI:10.1039/c9sc03374b

Chemical
Science
View Article Online
View Journal
EDGE ARTICLE
Two 30-O-b-glucosylated nucleoside
fluorometabolites related to nucleocidin in
Streptomyces calvus†
Cite this: DOI: 10.1039/c9sc03374b
All publication charges for this article
have been paid for by the Royal Society
of Chemistry
*
Xuan Feng, Davide Bello,
Phillip T. Lowe,
Joshua Clark and David O'Hagan
The antibiotic nucleocidin is a product of the soil bacterium Streptomyces calvus T-3018. It is among the
very rare fluorine containing natural products but is distinct from the other fluorometabolites in that it is
not biosynthesised from 50-fluorodeoxyadenosine via the fluorinase. It seems to have
a unique
enzymatic fluorination process. We disclose here the structures of two 40-fluoro-30-O-b-glucosylated
metabolites (F-Mets I and II) which appear and then disappear before nucleocidin production in batch
cultures of S. calvus. Full genome sequencing of S. calvus T-3018 and an analysis of the putative
biosynthetic gene cluster for nucleocidin identified UDP-glucose dependent glucosyl transferase
(nucGT) and glucosidase (nucGS) genes within the cluster. We demonstrate that these genes express
enzymes that have the capacity to attach and remove glucose from the 30-O-position of adenosine
analogues. In the case of F-Met II, deglucosylation with the NucGS glucosidase generates nucleocidin
suggesting a role in its biosynthesis. Gene knockouts of nucGT abolished nucelocidin production.
Received 8th July 2019
Accepted 19th August 2019
DOI: 10.1039/c9sc03374b
rsc.li/chemical-science
For many years nucleocidin 1 production remained elusive
because the publically available strains of S. calvus did not
Introduction
support its biosynthesis. Recently however, this was recognised
to be the result of a bald gene (bld A) mutation which when
reversed in S. calvus ATCC 13382, re-established nucleocidin
production as well as an aerial mycelial phenotype.9,10 In this
study we have used a producing strain of S. calvus T-3018 held
The antibiotic nucleocidin 1 is among the very rare uorinated
natural products.1 It is a 50-O-sulfonyl antibiotic produced by the
actinomycete bacterium Streptomyces calvus.2 Nucleocidin 1 is
a member of a small class of such antibiotics which include the
chlorine containing ascamycin 2 and dealanylascamycin 3, as
shown in Fig. 1A.3
The uorine atom unique to nucleocidin 1 is located at C40 of
the ribose ring of the 50-O-sulfamylated adenosine. The
biosynthesis of nucleocidin is unknown. Only three other
bacterial uorometabolites have been identied: these are u-
oroacetate 5,4,5 4-uorothreonine 6 4,5 and (2R,3S,4S)-trihydroxy-
5-uoropentanoic acid 7,6 however these metabolites are bio-
synthesised by pathways which are initiated by the C–F bond
forming enzyme, adenosyl uoride synthase (EC 2.5.1.63) also
known as ‘uorinase’.7 This enzyme combines uoride ion and
S-adenosyl-L-methionine to generate 50-uorodeoxy adenosine 4
(FDA) which is then biotransformed to the bacterial uo-
rometabolites 5–7 as summarised in Fig. 1B.1,8 However, full
genome sequencing of the nucleocidin 1 producer S. calvus9
indicates that there is no related uorinase gene present in S.
calvus and the positioning of the uorine at the C-40 position of
the ribose ring in nucleocidin 1 is inconsistent with an origin
from FDA 4, suggesting a unique uorination enzyme.
School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16
9ST, UK. E-mail: do1@st-andrews.ac.uk
Fig. 1 (A) Nucleoside antibiotics 1–3. (B) The fluorinated natural
products 5–7 derive from the fluorinase via FDA 4, however this
enzyme is not involved in nucleocidin 1 biosynthesis.
† Electronic supplementary information (ESI) available. See DOI:
10.1039/c9sc03374b
This journal is © The Royal Society of Chemistry 2019
Chem. Sci.

View Article Online
Chemical Science
Edge Article
by Pzer, which does not have this mutation and is a competent biosynthesis of 1 by excluding late stage intermediates that
nucleocidin 1 producer. In our efforts to explore the biosyn- might activate C40 for uorination, by oxidation at C50.
thesis of nucleocidin 1 we recently reported the incorporation of
When monitoring S. calvus T-3018 cultures by 19F{1H}-NMR
isotopically labelled glycerols into 1 and demonstrated that to assess nucleocidin 1 production, periodically two unidenti-
both of the C50-hydrogens of the ribose ring derive intact from ed uorinated metabobiles (F-Mets I and II) appear.11 These
the pro-R hydroxymethyl group of glycerol. This indicates that compounds have also been observed in other studies on
there is no oxidation at this carbon as glycerol is progressed nucleocidin 1 production in S. calvus ATCC 13382 9a and most
along the pentose phosphate pathway and into the ribose recently during nucleocidin production in Streptomyces aster-
moiety of nucleocidin.11 This places certain limitations on the oporus DSM 41452.10
Our observations indicate that F-Mets I and II appear and
disappear prior to the accumulation of nucleocidin 1 itself,
suggesting an intimate association with antibiotic production. A
typical time course prole from a S. calvus T-3018 fermentation
is shown in Fig. 2 and it can be seen that F-Mets I and II accu-
mulate between days 4 and 8 and then disappear between days
8–11 as nucleocidin 1 emerges. Clearly the structure of these
compounds could offer an insight into the uorometabolite
pathway in S. calvus. We are now able to report their structures
and have identied genes/enzymes which lead to their forma-
tion and in one case its conversion to nucleocidin 1.
Results and discussion
F-Mets I and II were isolated in low milligram amounts by
preparative HPLC guided by MS–MS analysis aer adsorption
onto charcoal from 6–8 day cultures of S. calvus T-3018. High
resolution mass spectrometry (HRMS) gave accurate masses
and therefore elemental constitutions (F-Met I ¼ 448.14673
amu; C16H23FN5O9. F-Met II ¼ 527.11993 amu; C16H24FN6O11S).
Although the titres were low, sufficient material was isolated to
Fig. 2 19F-NMR time course of nucleocidin 1 production in S. calvus
showing the appearance and disappearance of F-Mets I and II over 11
days.
Table 1 1H and 13C NMR data of F-Mets I, II and nucleocidin 1 (700 MHz, acetone-d6)
Nucleocidin 1 (ref. 9) (dF ¼ ꢀ119.98
F-Met I (dF ¼ ꢀ120.54 ppm)
dH ppm
F-Met II (dF ¼ ꢀ118.23 ppm)
dH ppm
ppm)
Position
dC ppm
dH ppm
dC ppm
2
8.22 (s)
8.29 (s)
6.37 (d, 2.0)
8.36 (s)
8.38 (s)
—
8.21(s)
8.21(s)
6.39 (d, 1.9)
4.85 (dd, 6.5, 1.9)
5.10 (m)
152.9
139.4
91.2
71.9
70.7
8
139.2
90.9
72.6
76.8
114.7
66.5
10
6.45 (d, 0.9)
4.91 (dd, 5.9, 1.4)
5.35 (dd, 18.3, 6.0)
—
4.42 (d, 9.1),
4.46 (d, 8.3)
4.76 (d, 7.7)
3.41–3.48 (m)
3.69 (m)
20
4.83 (dd, 5.6, 2.0)
5.16 (dd, 16.7, 5.9)
—
3.76 (dd, 11.8, 4.8)
3.85 (overlap with 600 Hb)
4.70 (d, 7.7)
3.39–3.47 (m)
3.69 (dd, 11.8, 4.9)
3.84 (overlap with 50 Hb)
30
40
—
115.9
66.9
50 Ha
5000 Hb
1
4.35, 4.39 (m)
102.9
69.9–77.0
61.3
—
—
—
—
—
—
200–500
600 Ha
600 Hb
3.86 (m)
Chem. Sci.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Edge Article
Chemical Science
be able to record 1H- and 19F-NMR in each case and a 13C-NMR calvus), insertion of the gene into a pEHISTEV plasmid and then
for F-Met II. Detailed 2D-NMR (COSY, TOCSY and HMBC) over- expression in E. coli Rosetta cells.5,15 In each case soluble
analyses indicated that the compounds were the 30-O-b-gluco- protein of the predicted molecular mass (35.5 kDa for NucGT
sylated derivatives of 1 and that F-Met I is a 30-O-b-glucosylated- and 53.6 kDa for NucGS) was obtained aer purication on
40-uoroadenosine and F-Met II is a 30-O-b-glucosylated nucle- a nickel column (see ESI†). The proteins were unambiguously
ocidin. The NMR data are summarised in Table 1 alongside conrmed by MS–MS analyses. The over-expressed glycosyl-
nucleocidin 1 for comparison. See ESI† for full details. The transferase (NucGT), was explored as a catalyst for 30-O-b-glu-
structure of F-Met II is illustrated in a book chapter from 1995, cosylation of adenosine 8 and sulfamoyladenosine 916 as
indicating a compound isolated from a Ciba-Geigy bacterium, candidate substrates using the common glucose donor, UDP-
Streptomyces strain R-156, however we cannot nd any details of glucose.17
its isolation or characterisation beyond this anecdotal
comment.12
Both 8 and 9 were good substrates based on initial rates with
a saturating concentration of UDP-glucose to generate 11 and 12
A putative gene cluster was previously identied by Zechel respectively (Fig. 4). Sulfamoyl adenosine 9 was the more effi-
and Bechthold9 for nucleocidin 1 biosynthesis in S. calvus ATCC cient substrate. Adenosine-50-sulfate 1018 was also examined as
13382 by homology to the gene cluster of the structurally related a substrate for NucGT however it was not processed by the
antibiotics ascamycins 2 and 3.13 A summary of the gene cluster enzyme, and presumably the formal charge of the sulfate is
is illustrated in Fig. 3 and more comprehensively in Fig. S36.† incompatible with enzyme binding. Additional support for the
The integrity of the gene cluster was established when cor- product structures was secured by an independent synthesis of
recting the mutation in a bld gene, which codes a Leu-tRNAUUA 30-O-b-gluco-adenosine 11 as illustrated in Fig. 5. The synthesis
able to translate a rare TTA codon into leucine. Three genes in involved a b-glycoside specic coupling of protected ribose 1319
the nuc cluster (nucJ, orf4 and orf6) contain this rare TTA codon
and when the mutation in the bldA gene was corrected to re-
establish this Leu-tRNAUUA the production of nucleocidin was
re-established. We have independently sequenced the S. calvus
T-3018 genome (recently deposited in GenBank)14 and nd that
unlike S. calvus ATCC 13382, S. calvus T-3018 does not contain
a bld gene mutation.
The genome of S. asterosporus DSM 41452 was recently
sequenced10 and was shown to possess a similar cluster; when
this organism was grown with uoride it produced nucleocidin
1, adding further support that this biosynthetic gene cluster is
responsible for nucleocidin biosynthesis.
There is also a 30-phosphoadenosine 50-phosphosulfate syn-
thases (PAPS) cassette within the genomes of these organisms,
consistent with a requirement to assemble the sulfamoyl moiety
of the antibiotic. From the S. calvus sequences, the putative nuc
cluster predicts glycosyltransferase (nucGT) and b-glucosidase
(nucGS) genes which sit at the end of the cluster and are located
three genes apart (see Fig. S36† for details). In the light of the
structures of F-Mets I and II, we were motivated to over-express
these two candidate glucosylating genes to assay their products.
This was achieved by PCR amplication from genomic DNA (S.
Fig. 4 Comparison of the initial rates (triplicates) of 8–10 with Nuc-
GT. Rates 8 ( ), 9 ( ) and 10 ( ).
Fig. 3 Organisation of the putative nucleocidin biosynthetic gene
cluster in S. calvus. A fuller annotation and a comparison with the
ascamycin 2 antibiotic gene cluster, can be found in the ESI Fig. S36.† Fig. 5 Synthesis route to 30-O-b-gluco-adenosine 11.
This journal is © The Royal Society of Chemistry 2019
Chem. Sci.

View Article Online
Chemical Science
Edge Article
and activated glucose 1420 to generate 15. Introduction of the
Incubation of F-Met I with NucGS gave a clear conversion to
adenosine was then achieved via a 6-chloropurine strategy21 to a new product with a signal at ꢀ122.5 ppm by 19F-NMR as
generate 18, with global aminolysis completing the synthesis. illustrated in Fig. 6a, consistent with literature values for 19 and
Synthetic and enzymatically generated 11 were identical (see LC-MS data which also indicated the formation of 19. On the
Fig. S34 and S35†). This synthesis also offers reinforcement of other hand, treatment of F-Met II with NucGS resulted in its
the structures of F-Mets I and II in terms of NMR signatures conversion to nucleocidin 1, as illustrated in Fig. 6b. These in
consistent with 30-O-glucosylation and b-glycoside linkage.
vitro biotransformations are consistent with the observed time
In order to assay the glucosidase (Nuc-GS) in the context of course experiments illustrated in Fig. 2 and can be interpreted
nucelocidin biosynthesis we required access to samples of F- that F-Met I is deglucosylated to 19 and slowly decomposes to
Mets I and II. These are not readily synthesised, and thus uoride. If compound 19 survives the extraction aer adsorp-
analytical samples of each were secured by semi-preparative tion onto charcoal/Celite from the whole cells of S. calvus, then
HPLC from S. calvus T-3018 fermentations arrested at day 8. its presence would be disguised in the 19F NMR spectrum by the
Although only analytical amounts of material were isolated in dominant uoride ion peak at ꢀ122.5 ppm (see Fig. 2). For F-
each case, reactions could be conveniently monitored by 19F Met II, the action of NucGS generates nucleocidin 1 and it
{1H}-NMR. Incubation of F-Met I with Nuc-GS resulted in its follows that NucGS and F-Met II could be involved in the nal
complete conversion to 40-uoroadenosine 19 as illustrated in step in nucleocidin biosynthesis.
Fig. 6a. This compound has previously been prepared and is
In order to further interrogate the roles of NucGS and NucGT
reported to be stable in buffer for several days but it slowly in nucleocidin biosynthesis, knockout experiments were con-
hydrolyses to release uoride.22
ducted for nucGT and nucGS (see Fig. S37 and S38†). In-frame
deletion of nucGT completely abolished uorometabolite
production, consistent with a role in nucleocidin biosynthesis.
A similar deletion of nucGS did not signicantly affect the
production of F-Mets I and II, consistent with their glucosylated
status, however nucleocidin 1 production was not affected
either. On further analysis of the S. calvus genome there are two
additional b-glycosidase enzymes coded for elsewhere on the
chromosome which have high (ꢁ88%) amino acid sequence
similarity to NucGS (see Fig. S39†) suggesting redundancy in
terms of these enzymes with their capacity to deglucosylate F-
Met II. We note that in the recently reported genome
sequence of the nucleocidin producer, Streptomyces asteroporus
DSM 41452 10 there are corresponding genes with a 100% amino
acid similarity. Thus from the two organisms known to produce
nucelocidin, both have these genes in their clusters, indicative
of key roles in the biosynthesis.
The previous isotopic labelling studies with glycerol11 indicate
that C50 of adenosine is not oxidised during the biosynthesis of 1,
and this study suggests that the uorination substrate is a 30-O-b-
glycosylated adenosine as summarised in Fig. 7.
It follows that an enzymatic uorination mechanism
involving oxidation of free alcohols at C30 and/or C50 seem
Fig. 6 19F{1H}-NMR time course profiles of biotransformations of F-
Mets I and II with b-glucosidase NucGS. (a) F-Met I is converted to 40-
fluoroadenosine 19. (b) F-Met II is converted to nucleocidin 1.
Fig. 7 Putative minimal pathway for nucleocidin 1 biosynthesis.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Edge Article
Chemical Science
´
unlikely, raising the prospect that uorination occurs by direct
C40–H activation. One hypothesis would involve a C40-centred
oxocarbenium ion which is attacked by uoride ion. Within the
biosynthetic gene cluster are a number of genes encoding
enzymes of unknown function and also a gene (nucJ) predicted
to encode a radical SAM/iron–sulfur (Fe–S) cluster enzyme
which may have the capacity to achieve such a two electron
oxidation.
6 L. Ma, A. Bartholome, M. H. Tong, Z. Qin, Y. Yu, T. Shepherd,
K. Kyeremeh, H. Deng and D. O'Hagan, Chem. Sci., 2015, 6,
1414–1419.
7 (a) D. O Hagan, C. Schaffrath, S. L Cobb, J. T. G. Hamilton
and C. D Murphy, Nature, 2002, 416, 279; (b) C. Dong,
F. L. Huang, H. Deng, C. Schaffrath, J. B. Spencer,
D. O'Hagan and J. H. Naismith, Nature, 2004, 427, 561–565.
8 H. Deng, S. M. Cross, R. P. McGlinchey, J. T. G. Hamilton and
D. O'Hagan, Chem. Biol., 2008, 15, 1268–1276.
9 (a) X. M. Zhu, S. Hackl, M. N. Thaker, L. Kalan, C. Weber,
D. S. Urgast, E. M. Krupp, A. Brewer, S. Vanner,
A. Szawiola, G. Yim, J. Feldmann, A. Bechthold,
G. D. Wright and D. L. Zechel, ChemBioChem, 2015, 16,
2498–2506; (b) L. Kalan, A. Gessner, M. N. Thaker,
N. Waglechner, X.-M. Zhu, A. Szawiola, A. Bechthold,
G. D. Wright and D. L. Zechel, Chem. Biol., 2013, 20, 1214–
1224.
Conclusions
We have identied two novel uorometabolites (F-Met I and F-
Met II), which are 30-O-glucosylated, 40-uoro-riboadenosines
from S. calvus. The products of two genes (nucGS and nucGT)
identied in the putative nucleocidin gene cluster were over-
expressed and the corresponding proteins have the ability to
glucosylate (NucGT) and deglucosylate (NucGS) these two uo-
rometabolites at the 30-hydroxyl group of the ribose ring. In the
case of F-Met II, deglucosyaltion generates nucleocidin 1. These
observations support a role for F-Met I and F-Met II in the
biosynthesis of nucleocidin. The identity of these metabolites is
highly suggestive that one is a product of the unknown uori-
nation enzyme in S. calvus, and should provide a greater focus
In efforts to identify this interesting enzyme.
10 S. Zhang, D. Klementz, J. Zhu, R. Makirtynskyy,
¨
A. R. O. Pasternak, S. Gunther, D. L. Zechel and
A. Bechthold, J. Biotechnol., 2019, 292, 23–31.
´
11 (a) X. Feng, N. Al Maharik, A. Bartholome, J. E. Janso,
U. Reilly and D. O'Hagan, Org. Biomol. Chem., 2017, 15,
´
8006–8008; (b) A. Bartholome, J. E. Janso, U. Reilly and
D. O'Hagan, Org. Biomol. Chem., 2017, 15, 61–64.
12 H. Kristinsson, K. Nebel, A. C. O'Sullivan, J. P. Pachlatko,
and Y. Yamaguchi, ACS Symposium Series, Synthesis and
Chemistry of Agrochemicals IV, 1995, pp. 206–219.
13 C. Zhao, J. Qi, W. Tao, L. He, W. Xu, J. Chen and Z. Deng,
PLoS One, 2014, 9, e114722.
Conflicts of interest
The authors declare no conicts of interest.
14 Whole Genome Shotgun data for S. calvus T-3018 has been
deposited at DDBJ/ENA/GenBank under the accession
VCNP01000000.
15 H. Liu and J. H. Naismith, Protein Expression Purif., 2009, 63,
102–111.
16 P. Mujumdar, S. Bua, C. T. Supuran, T. S. Peat and
S.-A. Poulsen, Bioorg. Med. Chem. Lett., 2018, 28, 3009–3013.
17 L. L. Lairson, B. Henrissat, G. J. Davies and S. G. Withers,
Annu. Rev. Biochem., 2008, 77, 521–555.
Acknowledgements
We thank the EPSRC Catalysis HUB (University of Manchester)
for nancial support. We also thank Roger Howard, Usa Reilly,
Jeffery Janso and Alessandra Eustaquio at the Biocatalysis
Technologies Laboratory, Pzer, Groton, USA, for nancial
support, a producing strain of S. calvus, and supporting the
research with genome sequencing.
Notes and references
18 F. Egami and N. Takahashi, Bull. Chem. Soc. Jpn., 1955, 28,
666–668.
1 D. O'Hagan and H. Deng, Chem. Rev., 2015, 115, 634–649.
2 S. O. Thomas, V. L. Singleton, J. A. Lowery, R. W. Sharpe,
L. M. Pruess, J. N. Porter, J. H. Mowat and N. Bohonos,
Antibiot. Annu., 1956, 1956–1957.
3 (a) K. Isono, M. Uramoto, H. Kusakabe, N. Miyata and
T. Koyama, J. Antibiot., 1984, 37, 670–672; (b) E. Takahashi
and T. Beppu, Biochem. J., 1982, 233, 459–463.
4 M. Sanada, T. Miyano, S. Iwadare, J. M. Williamson,
B. H. Arison, J. L. Smith, A. W. Douglas, J. M. Leich and
E. Inamine, J. Antibiot., 1986, 39, 259–265.
5 H. Deng, L. Ma, N. Bandaranayaka, Z. Qin, G. Mann,
K. Kyeremeh, Y. Yu, T. Shephers, J. H. Naismith and
D. O'Hagan, ChemBioChem, 2014, 15, 364–368.
19 B. J. Beahm, K. W. Dehnert, N. L. Derr, J. Kuhn,
J. K. Eberhart, D. Spillmann, S. L. Amacher and
C. R. Bertozzi, Angew. Chem., Int. Ed., 2014, 53, 3347–3352.
20 W. Pilgrim and P. V. Murphy, J. Org. Chem., 2010, 75, 6747–
6755.
21 H. Kristinsson, K. Nebel, A. C. O'Sullivan, F. Struber,
T. Winkler and Y. Yamaguchi, Tetrahedron, 1994, 50, 6825–
6838.
22 (a) S. Lee, C. Uttamapinant and G. L. Verdine, Org. Lett.,
2007, 9, 5007–5009; (b) D. Guillerm, M. Muzard, B. Allart
and G. Guillerm, Bioorg. Med. Chem. Lett., 1995, 5, 1455.
This journal is © The Royal Society of Chemistry 2019
Chem. Sci.