Biosynthesis of the tunicamycin antibiotics proceeds via unique exo-glycal intermediates

Article abstract of DOI:10.1038/nchem.1351

The tunicamycins are archetypal nucleoside antibiotics targeting bacterial peptidoglycan biosynthesis and eukaryotic protein N-glycosylation. Understanding the biosynthesis of their unusual carbon framework may lead to variants with improved selectivity. Here, we demonstrate in vitro recapitulation of key sugar-manipulating enzymes from this pathway. TunA is found to exhibit unusual regioselectivity in the reduction of a key α,β-unsaturated ketone. The product of this reaction is shown to be the preferred substrate for TunF-an epimerase that converts the glucose derivative to a galactose. In Streptomyces strains in which another gene (tunB) is deleted, the biosynthesis is shown to stall at this exo-glycal product. These investigations confirm the combined TunA/F activity and delineate the ordering of events in the metabolic pathway. This is the first time these surprising exo-glycal intermediates have been seen in biology. They suggest that construction of the aminodialdose core of tunicamycin exploits their enol ether motif in a mode of C-C bond formation not previously observed in nature, to create an 11-carbon chain.

Full text of DOI:10.1038/nchem.1351

ARTICLES
PUBLISHED ONLINE: 20 MAY 2012 | DOI: 10.1038/NCHEM.1351
Biosynthesis of the tunicamycin antibiotics
proceeds via unique exo-glycal intermediates
Filip J. Wyszynski1†, Seung Seo Lee1†, Tomoaki Yabe1, Hua Wang1, Juan Pablo Gomez-Escribano2,
Mervyn J. Bibb2, Soo Jae Lee3, Gideon J. Davies4 and Benjamin G. Davis1
*
The tunicamycins are archetypal nucleoside antibiotics targeting bacterial peptidoglycan biosynthesis and eukaryotic
protein N-glycosylation. Understanding the biosynthesis of their unusual carbon framework may lead to variants with
improved selectivity. Here, we demonstrate in vitro recapitulation of key sugar-manipulating enzymes from this pathway.
TunA is found to exhibit unusual regioselectivity in the reduction of a key a,b-unsaturated ketone. The product of
this reaction is shown to be the preferred substrate for TunF—an epimerase that converts the glucose derivative to a
galactose. In Streptomyces strains in which another gene (tunB) is deleted, the biosynthesis is shown to stall at this
exo-glycal product. These investigations confirm the combined TunA/F activity and delineate the ordering of events in the
metabolic pathway. This is the first time these surprising exo-glycal intermediates have been seen in biology. They suggest
that construction of the aminodialdose core of tunicamycin exploits their enol ether motif in a mode of C–C bond formation
not previously observed in nature, to create an 11-carbon chain.
he tunicamycins are fatty acyl nucleoside antibiotics with biology as a 5,6-dehydratase, and kinetic analyses and structure
potent inhibitory activity towards bacterial cell wall biosyn- provide insights into this novel, unique selectivity. TunF is found
T
thesis and eukaryotic protein N-glycosylation1–3. They are to act as a sugar-4-epimerase with unusual substrate tolerance,
produced by Streptomyces lysosuperificus and chartreusis strains4,5, which helps to delineate pathway ordering. A revised pathway is
and their structures comprise a unique 11-carbon core (tunicamine) proposed, involving conjugation of an exo-glycal (the first invoked
decorated with uracil, N-acetylglucosamine (GlcNAc) and variable in any natural pathway) with a suitable uridine derivative.
fatty acyl moieties (Fig. 1a)6,7. Several other natural products share
the same carbohydrate core: streptovirudins8, corynetoxins9, Results
MM1929010, mycospocidin11 and antibiotic 2401012. The biosyn- Bioinformatic mining of the tun gene cluster. Within the
thetic pathways to this core are all unknown.
biosynthetic pathway deduced from the S. chartreusis tun gene
Tunicamycins were the first compounds found to specifically cluster24, putative gene products TunA and TunF were suggested
inhibit the formation of peptidoglycan precursor lipid I (targeting to participate in the early construction of the tunicamine
enzyme MraY) during bacterial cell wall biosynthesis1,13, a mode subunit (Fig. 1b). Sequence homology comparison and alignment
of action orthogonal to existing antibiotics1,14,15. These antibiotics (Supplementary Figs S1 and S2) revealed that TunA was similar
also inhibit eukaryotic N-linked glycoprotein synthesis at the first to NAD-dependent hexose epimerases and dehydratases from a
committed step2. Although this results in cytotoxicity to mamma- variety of sources. This suggested TunA to be a member of
lian cells and currently precludes clinical use, it renders tunicamycin the short-chain dehydrogenase/reductase (SDR) family. The three
a crucial tool in glycobiology16.
closest homologues (identities 27–32%) with Protein Data Bank
Synthetic studies towards tunicamycins have been published17–20
,
(PDB)27 entries are well-characterized 4,6-dehydratases that
including two total syntheses21,22, but biosynthetic investigations operate on a substrate very different to that in the tunicamycin
have been limited23. Recent identification of the biosynthetic pathway, deoxythymidine-diphosphate glucose (dTDP-Glc):
gene cluster and the proposal of a possible metabolic pathway, DesIV from Streptomyces venezuelae, RmlB from Salmonella
however, have opened the door to more detailed investigations24–26
.
enterica and RmlB from Streptococcus suis28,29. Multiple sequence
A deeper understanding of the enzymatic processes involved may alignments suggest mechanistically important amino-acid residues
allow this unique natural product scaffold to be readily altered. (Supplementary Fig. S1). Importantly, the putative TunA sequence
New analogues could be more selective inhibitors of MraY, as well has key conserved motifs for nucleotide binding (TGxxGxxG near
as rationally designed inhibitors of other enzymes.
N-terminus) and a signature TYK catalytic triad (Thr119, Tyr143,
The biogenesis of the undecose tunicamine core has been Lys147). In other enzymes, TYK, together with cofactor NADþ,
predicted23–26 to involve some form of unusual monosaccharide performs the initial oxidation of the 4-OH of hexose substrates29.
tail-to-tail C–C coupling (Fig. 1b). Although associated genes
TheTunFsequenceshowedsimilarity(28–29%)withtwoofthebest
have been suggested24, the actual activity of the gene products and characterized NAD-dependent hexose epimerases: WbpP
precise ordering of the transformations remain unknown. Here, from Pseudomonas aeruginosa and WbgU from Plesiomonas
we show how two enzymes act together at the first committed shigelloides30,31. This led to early predictions as a UDP-GlcNAc 4-epi-
stages. TunA is shown to have a previously unreported activity in merase and suggested it as the first committed pathway enzyme.
1Chemistry Research Laboratory, University of Oxford, 12 Mansfield Road, Oxford OX1 3TA, UK, 2Department of Molecular Microbiology, John Innes
Centre, Norwich Research Park, Norwich NR4 7UH, UK, 3College of Pharmacy, Chungbuk National University, Gaeshin-dong, Heungduk-gu, Cheongju,
Chungbuk, Korea, 4Department of Chemistry, The University of York, Heslington, York YO10 5DD, UK; †These authors contributed equally to this work.
e-mail: ben.davis@chem.ox.ac.uk.
*
NATURE CHEMISTRY | VOL 4 | JULY 2012 | www.nature.com/naturechemistry
© 2012 Macmillan Publishers Limited. All rights reserved.
539

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1351
ARTICLES
synthesized and cloned into expression vector pET16b. This was
expressed and the protein purified from E. coli as an N-terminal-
His10-fusion (Supplementary Fig. S3). The 37.8 kDa protein was
incubated with UDP-GlcNAc and conversion was monitored by
1H NMR (Supplementary Fig. S4). This revealed the formation of
a product with chemical shifts indistinguishable from UDP-
GalNAc standard: a decrease in the integration of UDP-GlcNAc-
derived peaks mirrored an increase in UDP-GalNAc. In control
experiments with heat-denatured TunF, no product was formed.
The time course of the reaction (Supplementary Fig. S4) showed
the formation of ꢀ30% equilibrium position (Keq ¼ 0.43),
consistent with related epimerases and reflecting a greater stability
of equatorial over axial 4-OH (ref. 35). A high-performance liquid
chromatography (HPLC) assay (Supplementary Fig. S5) allowed
the determination of kinetic parameters (KM ¼ 3.6(+0.2) mM and
kcat/KM ¼ 34.0(+2.8) mM21 min21) for TunF towards UDP-GlcNAc.
a
O
O
OH
HO
OH
HO
NH
O
O
N
N
O
O
H
AcHN
HO
HO
OH
O
Tunicamycins
OH
x
9
y
7
8
9
Tun13:1
9
Tun14:1B
Tun14:1A
Tun15:1B
10 Tun15:1A
11 Tun16:1B
12 Tun17:1A
x =
y =
Tun15:0
10 Tun16:1A
11 Tun17:1B
O
O
b
TunA is a UDP-GlcNAc 5,6-dehydratase. The predicted activity
of TunA was based on sequence similarity with DesIV and RmlB.
Both catalyse oxidation of 4-OH in dTDP-glucose and subsequent
E1cB elimination to form a,b-unsaturated-ketone dTDP-4-keto-
glucose-5,6-ene28,29. In these enzymes, this intermediate undergoes
1,4-conjugate addition of hydride to yield dTDP-6-deoxy-4-
keto-glucose (Supplementary Fig. S6).
6''
OH
OH
PPPO
NH
NH
O
4''
5''
O
O
N
O
N
HO
HO
1''
2''
AcHN
O
3''
TunN,G
O
HO
OH
HO
OH
UDP
UDP-GlcNAc
Uridine
Uridine 5'-triphosphate
Codon-optimized tunA gene was synthesized and cloned into
pET16b. This was expressed and the protein purified from E. coli
cells as an N-terminal-His10-fusion (Supplementary Fig. S7).
The 37.9 kDa protein product was incubated with putative sub-
strates. Product was only formed from UDP-GlcNAc, rather than
the proposed substrate UDP-GalNAc (Supplementary Fig. S8).
Surprisingly, the product of TunF was not processed by TunA,
and our initial proposals24 for the order of events during tunicamine
biosynthesis were incorrect.
TunA,F,B,(M?)
O
OH
8'
HO
9'
OH
NH
O
6'
AcHN
O
N
1
1'
10'
11'
O
5'
Tunicamycins
4'
3'
HO
UDP-N-acetyl-tunicamine
7'
2'
OH
O
UDP
After TunA was incubated with UDP-GlcNAc and NaBD4, mass
spectrometry revealed a product compound with m/z ¼ 589. The
18-unit reduction in m/z from substrate (UDP-GlcNAc) to product
corresponded simply to a dehydrated form. This result and NMR
results suggested that no deuterium incorporation/trapping had
occurred (Fig. 2). This gave the first suggestion that TunA yields
neither the intermediate UDP-4-keto-6-deoxy-GlcNAc-5,6-ene nor
the typical dehydrated product UDP-6-deoxy-4-keto-GlcNAc, but
instead its allylic alcohol isomer, UDP-6-deoxy-5, 6-ene-GlcNAc.
To distinguish possible dehydration pathways (Fig. 2), the reaction
mixture was analysed in detail by 1H NMR. UDP-GlcNAc was incu-
bated with TunA in buffered D2O. New resonances from product were
distinctly visible: dH ¼ 5.53 ppm was assigned to an H-1′′ anomeric
proton with multiplicity and coupling constants very similar to
those of UDP-GlcNAc. Resonances at dH ¼ 4.85 ppm were indicative
of a 1,1-disubstituted-alkenyl moiety, potentially corresponding to H-
6′′a,6′′b of a terminal alkene and arguing for the formation of a UDP-
6-deoxy-GlcNAc-5,6-ene product. This was further supported by the
absence of C-6′′ methyl group resonance, which would be present if
the typical product of dehydratases, UDP-4-keto-6-deoxy-GlcNAc,
had been formed. The double-triplet (dt) multiplicity for the reson-
ance at dH ¼ 4.05 ppm was assigned to H-4′′; critically, this displayed
Figure 1 | Structures of the tunicamycins and early stages of the proposed
biosynthetic pathway that creates the 11-carbon chain. a, Tunicamycin
homologues vary in the length, branching and saturation of their fatty acyl
side chains. b, Proposed biosynthetic pathway of the key intermediate
containing the construction of the core tunicamine subunit from primary
metabolites uridine 5′-triphosphate (UTP) and UDP-GlcNAc. The creation of
suitable intermediates for coupling has no clear precedent in nature. Putative
gene products TunA and TunF were previously suggested24 to participate in
its early stages based on analysis of the reported tunicmaycin gene cluster
from S. chartreusis24. The numbering systems used in this manuscript follow
carbohydrate nomenclature conventions. For compounds that contain the
tunicamine core, the associated 11-carbon chain is indicated by primed atom
labels (X′). In areas of the main text where substructures are being
discussed, primes have been left out to aid the clarity of the discussion (for
example, 4-keto-5,6-ene).
Alignments of TunF with human UDP-glucose-4-epimerase (hGalE)
and WbpP and WbgU (Supplementary Fig. S2) revealed that residues
3–35 resemble the NADþ-binding pocket of hGalE32 and contain the
conserved TGxxGxxG signature sequence (6–13)33. An Sx24Yx3K cata-
lytic triad, characteristic of SDR enzymes, is present at positions 115–
144 (ref. 33). Ser289 is predicted to act as ‘gatekeeper’ residue, allowing
N-acetylated sugar nucleotides into the active site. This is in accord with
other UDP-GlcNAc 4-epimerases and hGalE, which accept acetylated
substrates and containunhindered residuesatthis position. Conversely,
it contrasts with E. coli GalE (eGalE), which only epimerizes UDP-
galactose and has a bulky tyrosine gatekeeper residue34. This supports
the hypothesis that TunF acts upon an N-acetylated sugar.
long-range allylic coupling to protons on C-6′′ and the 3J4 ,3 (9.4 Hz)
′′ ′′
coupling constant closely matched that of the corresponding H-4′′
peak of UDP-GlcNAc; this strongly suggested unchanged stereo-
chemistry at C-4′′.
This unusual product of the TunA-catalysed reaction was pro-
1
duced and purified on scale and fully characterized by H, 13C,
31P, correlation spectroscopy (COSY), heteronuclear multiple-
quantum correlation (HMQC) and total correlation spectroscopy
(TOCSY) NMR experiments and high-resolution mass spec-
TunF is a UDP-GlcNAc 4-epimerase. To probe TunF function and troscopy, and was unambiguously assigned as UDP-6-deoxy-
determine the substrate(s), a codon-optimized tunF gene was GlcNAc-5,6-ene (Supplementary Figs S9 and S10). To the best of
540
NATURE CHEMISTRY | VOL 4 | JULY 2012 | www.nature.com/naturechemistry
© 2012 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1351
ARTICLES
Tyr143
O
H
Tyr143
NAD+
NADH
O–
HO Thr119
H
O
H
HO
O
O
HO
HO
O
H
AcHN
AcHN
O
HO Thr119
AcHN
UDP
O
O
O
–
OH
OH
OH
O
H
UDP
UDP
HS Cys120
–
Glu121 CO2
via E1cB
mechanism
UDP-GlcNAc
UDP-4-keto-GlcNAc
OH
HO
O
Tyr143
AcHN
H
O
1,2-equatorial-
O
H
addition
NADH
O
UDP
HO
AcHN
UDP-6-deoxy-GalNAc-5,6-ene
1,4-addition
HO Thr119
O
O
HO
AcHN
O
O
UDP
H
HO
O
UDP
1,2-axial-addition
AcHN
OH
O
–S Cys120
Glu121 CO2H
UDP-4-keto-6-deoxy-GlcNAc-5,6-ene
O
UDP
UDP-6-deoxy-GlcNAc-5,6-ene
Figure 2 | Possible TunA-catalysed dehydration pathways for substrate UDP-GlcNAc and putative intermediates. Initial analysis of the product began with
an attempt to ‘trap’ or react key functional groups that might be expected to be present from typical SDR activity (for example, in either a 4-keto product of
1,4-addition or an a,b-unsaturated keto intermediate). In these cases, sodium borodeuteride (NaBD4, as a source of external nucleophile) would react to
introduce a distinctive deuterium label. MS and NMR investigations revealed the formation instead of exo-glycal UDP-6-deoxy-5,6-ene-GlcNAc, which is
unreactive to NaBD4.
our knowledge, TunA is the first enzyme to exhibit NDP-sugar-5,6- a resolution of 1.9 Å (Supplementary Table S8) as a ternary
dehydratase activity, converting 4′′, 5′′ and 6′′ positions of UDP- complex (Fig. 3). The overall structure was superimposed with the
GlcNAc into an allylic alcohol moiety. This is in stark contrast to ternary complex structures of other dehydratases from the SDR
all other reported NDP-sugar-4,6-dehydratases that generate family—dTDP-Glc-4,6-dehydratases DesIV and both RmlBs—
NDP-4-keto-6-methyl sugars. HPLC and NMR analyses indicated with a root-mean-square deviation of 1.3–1.8 Å. The overall
that TunA catalysed the establishment from UDP-GlcNAc of an structure of TunA can be divided into two domains. The
equilibrium (Keq ¼ 0.16) with kcat ¼ 0.41(+0.01) s21 and KM
¼
N-terminal domain contains the NADþ-binding site and shows
the typical Rossmann fold (seven parallel b-strands surrounded
5.5(+0.3) mM (Supplementary Fig. S11).
Insight into the TunA mechanism was gained from other obser- by seven a-helices). The smaller C-terminal domain binds
vations. In particular, when the reaction was conducted in buffered UDP-GlcNAc and consists of mixed strands of four b-sheets and
D2O, the integration of the H-5′′ resonance for starting substrate four a-helices, features also characteristic of the SDR family. The
1
UDP-GlcNAc was lower in the H NMR than for other proton chemistry catalysed by the enzyme occurs at the interface of
resonances. This reduction suggested partial incorporation of these domains.
deuterium at C-5′′; this site-specific H-5′′ ↔ H-5′′ exchange was
Overall, the interactions of NADþ and UDP-GlcNAc in their
2
2
confirmed by H NMR. An additional minor doublet resonance binding sites in TunA are very similar to those at equivalent
was also observed (J ¼ 9.1 Hz) at dH ¼ 3.52 ppm, which overlapped positions in DesIV and both RmlBs (Fig. 3b and Supplementary
with the H-4′′ triplet peak of UDP-GlcNAc. This supported the Figs S12 and S14). Consistent with the bioinformatics analysis
existence of two closely related compounds differing only by the (vide supra), mutation of each residue in the TYK motif (to the
presence of deuterium at C-5′′. This partial deuteration implied respective single-point alanine mutants Thr119Ala, Tyr143Ala,
‘wash out’ by deuterated solvent36 and suggests a rapid, reversible Lys147Ala) reduced the enzymatic activity to an undetectable
step consistent with an E1cB reaction mechanism, supporting the level (Supplementary Table S9). Closer inspection of the site
involvement of a UDP-4-keto-GlcNAc intermediate that is readily where the GlcNAc moiety and the nicotinamide ring of NADþ
deprotonated at C-5′′.
are in close contact, however, reveals some interesting subtleties,
which may play important roles in altering the catalytic activity of
Control of reductive regioselectivity in Tun A. Given the near TunA and hence explain the unique observed reactivity. C-4 of
identical conservation of key residues in TunA (a 5,6-dehydratase the nicotinamide ring lies 3.4 Å from C-4′′ of the GlcNAc ring, a
using 1,2-reduction) and, for example, DesIV (a 4,6-dehydratase position normally ideal for subsequent hydride transfer37. The
using 1,4-reduction)29 suggested by primary sequence alignment relative positions of the nicotinamide and hexosamine/hexose
(vide supra), we sought to determine the mechanism behind this ring planes are particularly notable (Fig. 4 and Supplementary
control of regioselectivity (1,2 versus 1,4). TunA was crystallized Fig. S13). The ribose of NADþ bearing the nicotinamide group is
in the presence of both substrate UDP-GlcNAc and cofactor placed closer to the GlcNAc moiety in the TunA complex than in
NADþ at a pH remote from that of optimal activity to minimize DesIV/RmlB structures. Moreover, the plane of the nicotinamide
turnover—the resulting X-ray crystal structure was determined to ring in TunA is rotated ꢀ208 with respect to the UDP-GlcNAc
NATURE CHEMISTRY | VOL 4 | JULY 2012 | www.nature.com/naturechemistry
541
© 2012 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1351
ARTICLES
from C-6′′ of GlcNAc in TunA, but maintains a favourable position
and angle for hydride delivery at C-4′′. This subtle rotation is
essentially the sole element controlling the absolute observed
switch in 1,2 versus 1,4 reductive regioselectivity (Fig. 2 and
Supplementary Fig. S15).
a
The positions of amino-acid residues probably involved in
dehydration across C-5′′ and C-6′′ are also clearly visible. The
most notable Cys120, 2.7 Å from OH-6′, we suggest as a general
acid for OH-6′′ protonation (Fig. 2). In other dehydratases, Asp at
corresponding positions plays this role28,29. The adjacent Glu121—
conserved across all dehydratases discussed here—is suggested to
function as a general base in the abstraction of the H-5′′ proton,
lying 3.6 Å from C-5′′ and with good alignment for axial H-5′′. In
TunA, Glu121 also forms a 3.1 Å hydrogen bond with OH-6′′, an
interaction absent from the other dehydratases. Consistent with
these suggested critical roles in catalysis, the corresponding single
point alanine mutants—Cys120Ala and Glu121Ala—had no detect-
able enzymatic activities (Supplementary Table S9).
It is notable that the Lys residue hydrogen-bonded to the OH-2′′
of dTDP-glucose in the DesIV and RmlBs is replaced by smaller
Ala183 in TunA, thus allowing the larger C-2′′-acetamido substitu-
ent of UDP-GlcNAc. Consistent with this role, the mutation
Ala183 ꢁ Lys, blocking this acetamide binding site, removed any
detectable activity (Supplementary Table S9). In the binding of
dTDP-glucose, the DesIV/RmlBs structures show a His residue
forming a hydrogen bond to the only hydroxyl (OH-3′) of the sub-
strate 2′-deoxyribose ring. In TunA, Glu267 replaces this residue,
forming hydrogen bonds to both OH-2′,3′ of the ribose moiety in
UDP-GlcNAc. Interestingly, alteration of Glu267 ꢁ Ala gave a
single-point mutant that retained some activity, albeit with eightfold
reduction in kcat and fivefold increase in KM (Supplementary
Table S9). Notably, the structure determined here indicates that
additional hydrogen bonding to OH-2′ provided by the Glu201
side chain may mitigate the associated loss of binding energy to
the ribose diol unit during catalysis.
b
Gly20
Gly17
Gly14
NAD
TunF acts on more than one substrate. The apparent processing of
UDP-GlcNAc as substrate by both TunA and TunF created a
conundrum regarding the starting point and ordering of the tun
pathway (Fig. 5a). Despite the demonstrated UDP-GlcNAc-4-
epimerase activity (vide supra), TunF could, in principle act upon
other substrates (Supplementary Fig. S16).
To investigate this possibility, UDP-GlcNAc was incubated with
a mixture of TunF and TunA. Four 1H NMR resonances with
characteristic multiplicities in the dH ¼ 5.4–5.7 ppm H-1′′ region
indicated four distinct species (Fig. 5b). Three corresponded to
UDP-GlcNAc, UDP-GalNAc and UDP-6-deoxy-GlcNAc-5,6-ene;
the fourth, formed only in the presence of both TunF and TunA,
was deduced to be UDP-6-deoxy-GalNAc-5,6-ene (Fig. 5).
Tyr143
Lys147
Thr119
Glu121
UDP-GlcNAc
Cys120
Figure 3 | Three-dimensional structure of TunA. The crystal structure
contains two TunA molecules per unit cell and each monomer clearly shows
a molecule of the substrate bound in the active site together with cofactor
NADþ. a, Representative model of the TunA monomer showing the
observed electron density for the bound UDP-GlcNAc substrate and NADþ
cofactor. b, Detailed view of the TunA active site. Electron density maps are
2FO–FC syntheses (grey) contoured at ꢀ1s. This figure and Fig. 4a were
drawn with PyMOL (http://pymol.sourceforge.net).
To test this hypothesis, a sample of UDP-6-deoxy-GlcNAc-
5,6-ene was isolated following incubation of UDP-GlcNAc with
5,6-dehydratase TunA. This alternative substrate was incubated
with TunF (Supplementary Fig. S17). As for UDP-GlcNAc as
substrate, TunF also catalysed the establishment of an equilibrium
(Keq ¼ 0.43) from UDP-6-deoxy-GlcNAc-5,6-ene. Notably, the
kinetic parameters revealed that UDP-6-deoxy-GlcNAc-5,6-ene is
the preferred substrate for TunF (kcat/KM ¼ 245(+25) versus
ring plane; in the other dehydratases it is approximately parallel to 34.0(+2.8) mM21 min21).
the hexosamines (Fig. 4). The amide carbonyl of the nicotinamide
within TunA makes a hydrogen bond to the Val174 backbone In vivo detection of an exo-glycal intermediate. To assess the
amide, thus stabilizing this ‘rotated’ position (Supplementary relevance of these in vitro experiments to tunicamycin
Fig. S14); in DesIV/RmlBs the equivalent residue is Asn. biosynthesis in vivo, we created an in-frame deletion mutant of
This rotation observed only in TunA places C-4 of NADþ/ tunB in which the entire protein-coding sequence was removed.
NADH nicotinamide directly above C-4′′ of GlcNAc, whereas in The construction of an in-frame deletion should prevent any
DesIV/RmlBs this lies much closer to C-6′′ of the substrate sugar. polar effects on the expression of downstream genes, while
More importantly, the C-4 si-face of nicotinamide faces away deletion of tunB was predicted to lead to the accumulation of
542
NATURE CHEMISTRY | VOL 4 | JULY 2012 | www.nature.com/naturechemistry
© 2012 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1351
ARTICLES
a
a
HO
OH
OH
O
O
TunF
HO
HO
HO
AcHN
AcHN
O
O
UDP
UDP
UDP-GalNAc
UDP-GlcNAc
TunA
TunA
HO
O
O
TunF
HO
HO
HO
AcHN
AcHN
O
O
UDP
UDP
UDP-6-deoxy-
UDP-6-deoxy-
5,6-ene-GalNAc
5,6-ene-GlcNAc
b
i. UDP-GlcNAc
UDP-GlcNAc
b
TunA
RmlB
H2N
R
5.60
5.55
5.50
ppm
ppm
R
N+
Ring plane
parallel
Ring
plane
angle:
22º
N+
O
ii. UDP-GlcNAc + TunF
H2N
3.3 Å
O
H
3.7 Å
UDP-GaINAc
3.9 Å
OH
3.7 Å
H
HO
HO
HO
O
O
OH
AcHN
H
H
H
H
OTDP
5.60
5.55
5.50
5.50
OUDP
iii. UDP-GlcNAc + TunA
Figure 4 | Comparison of the conformations of substrates and cofactors in
active sites between TunA and RmlB. a, Colour code for carbon atoms:
green, TunA; blue, RmlB. Electron density maps are 2FO–FC syntheses (grey)
contoured at ꢀ1s; no density data are available for RmlB. b, Schematic
showing relative positions and angles of substrates and cofactors in active
sites of TunA and RmlB. When compared with existing 4,6-dehydratases
DesIV/RmlB, the NADþ nicotinamide and GlcNAc ring planes are no longer
parallel and lie ꢀ208 to each other, yet are still positioned to allow efficient
hydride-abstraction at C24′′ of UDP-GlcNAc. This subtle rotation plays an
important role later in the catalytic cycle, during hydride return to the 4-
keto-5,6-ene intermediate. Studies using dTDP-glucose complexed with
DesIV have shown that orientation of the plane of a,b-unsaturated
UDP-6-deoxy-
GlcNAc-5,6-ene
5.60
5.55
ppm
iv. UDP-GlcNAc + TunF + TunA
UDP-6-deoxy-
GalNAc-5,6-ene
5.60
5.55
5.50
ppm
Figure 5 | Interconversion of UDP-GlcNAc and related biosynthetic
intermediates by TunA and TunF. a, Structures and pathways catalysed by
both enzymes together. b, NMR evidence of TunA and TunF activity. 1H NMR
spectra (anomeric H-1′′ region) of UDP-GlcNAc incubated with (i) no enzyme,
(ii) TunF, (iii) TunA or (iv) TunF and TunA. Because UDP-GalNAc is not a
substrate for 5,6-dehydratase TunA, it was deduced that TunF acted upon the
product of the TunA reaction, UDP-6-deoxy-GlcNAc-5,6-ene, epimerizing it at
C-4′′ to form the novel species UDP-6-deoxy-GalNAc-5,6-ene. This was also
confirmed by direct incubation of UDP-6-deoxy-GlcNAc-5,6-ene with TunF.
UDP-6-deoxy-GlcNAc-5,6-ene is the preferred substrate for TunF; UDP-6-
deoxy-GlcNAc-5,6-ene bound more than threefold more tightly than UDP-
GlcNAc (KM ¼ 1.1(+0.1) and 3.6(+0.2) mM, respectively) and kcat values for
the exo-glycal substrate were also higher (269(+4) and 122(+3) min21).
Note that the strict carbohydrate numbering convention would describe
UDP-6-deoxy-GlcNAc-5,6-ene as UDP-6′′-deoxy-GlcNAc-5′′,6′′-ene but the
former descriptive notation is used here for clarity in the text.
intermediate closely mirrors that of the ring plane of the original substrate29
If extended to TunA, the altered position and ring plane of nicotinamide in
TunA result in poor alignment for hydride return to C-6′′ of the intermediate;
instead, 1,2-delivery at C-4′′ is favoured, followed by re-protonation of O-4′′
by Tyr143 (Supplementary Fig. S15). From the determined structure these
steps appear able to occur without a need for significant
.
conformational changes.
sufficient exo-glycal for direct observation. This mutational analysis
was performed on the cloned tunicamycin biosynthetic gene (tun)
cluster present in pIJ12003a24 in the heterologous expression host
Streptomyces coelicolor M1152. TunB is predicted to catalyse the
formation of the C–C bond between C5 of the uridyl moiety and
C6 of the central sugar in tunicamycin (see below). If the exo-
glycal is the source of this central sugar moiety, then it would be
expected to accumulate in a tunB mutant. The mutant construct,
created by polymerase chain reaction (PCR)-targeting38, was
transferred into S. coelicolor M115224 by conjugation via tri-
No tunicamycins were detected from the M1481 culture
parental mating39. Successful integration of the mutant construct (Supplementary Fig. S24). However, the nucleotide-rich fraction
into the chromosomal wBT1 attachment site was determined by (Supplementary Fig. S22) contained not only compounds associated
single-colony PCR of kanamycin-resistant exconjugants, yielding with typical metabolism (UDP-HexNAcs (m/z 606 [M–H]2),
S. coelicolor M1481.
UDP-hexoses (m/z 565 [M–H]2)), but also compound
NATURE CHEMISTRY | VOL 4 | JULY 2012 | www.nature.com/naturechemistry
543
© 2012 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1351
ARTICLES
O
O
O
OH
HO
NH
PPPO
OH
NH
O
NH
O
O
N
O
N
O
N
TunN,G
TunB
O
OH
HO
OH
HO
OH
Uridine 5'-triphosphate
Uridine
TunM?
HO
OH
O
O
O
O
σ*(C8'-O8') LUMO
HO
HO
HO
HO
TunA
TunF
HO
OH
HO
OH
HO
OH
HO
NH
O
AcHN
AcHN
AcHN
O
O
O
AcHN
O
N
O
UDP
UDP
UDP
UDP-6-deoxy-
5,6-ene-GlcNAc
UDP-6-deoxy-
5,6-ene-GalNAc
O
UDP-GlcNAc
UDP
OH
C7' SOMO
O
OH
HO
NH
O
AcHN
O
N
UDP-N-acetyl-
tunicamine
Tunicamycins
O
O
H
UDP
OH
Figure 6 | Suggested revised biosynthetic pathway to tunicamine, a key precursor of tunicamycin. Identification of the natural substrates for and functional
roles of TunA and TunF have delineated a key portion of the early biosynthetic pathway for the tunicamycins. TunA acts first, forming exo-glycal enol ether
UDP-6-deoxy-GlcNAc-5,6-ene from UDP-GlcNAc. TunF then epimerizes this product at C-4′′ to yield UDP-6-deoxy-GalNAc-5,6-ene. Together, these
enzymes therefore form a unique enol ether intermediate (UDP-6-deoxy-5,6-ene-GalNAc), which has the configuration and constitution to exploit beneficial
stereoelectronic interactions (shown in red here in a speculative intermediate) in a suggested subsequent regio- and stereoselective radical addition reaction
to generate the correct tunicamine core structure. TunB has been predicted to belong to the radical SAM superfamily, binding ′S-adenosylmethionine (SAM)
and a Fe4S4 redox cluster24. It should be noted that the substrate for the TunB-catalysed H-abstraction could also be uridine 5 -monophosphate rather than
uridine, as observed in polyoxin and nikkomycin biosynthesis56,57. The role of TunM is unclear; it is predicted to belong to the SAM-dependent
methyltransferase family24. It is possible that it acts as a partner to TunB, mediating the radical process and helping to bring together the Fe4S4 centre and
the two carbohydrate substrates.
corresponding to the predicted exo-glycal (m/z 588 [M–H]2) coupling event as a stable exo-glycal enol ether offers insights into
(Supplementary Fig. S22). Such a disrupted pathway would lead to the nature of the uridine-derived coupling partner and the roles
the accumulation of both exo-glycal epimers. The presence of UDP- of TunB and TunM—enzymes predicted to participate in the
6-deoxy-GlcNAc-5,6-ene was confirmed by the addition of an auth- coupling process24. We suggest that this coupling partner takes
entic sample (Supplementary Fig. S22). No exo-glycal was detected the form of a radical, centred at the C-5′ of uridine, in striking
from S. coelicolor M1040 (M1152 harbouring pIJ12003a containing contrast to the deduced intermediates of several related nucleoside
intact tun cluster24) and the non-producing strain S. coelicolor antibiotics40–42. TunB has been predicted to belong to the radical
M1037 (M1152 containing only vector24) (Supplementary Fig. S23).
S-adenosyl methionine (SAM) superfamily24. Members of this
enzyme class catalyse a wide range of radical reactions43. We specu-
late that following H-abstraction from C-5′ of a uridyl moiety,
Discussion
The recently reported biosynthetic gene cluster of tunicamycin radical addition to the terminal carbon of the exo-glycal coupling
has allowed the in vitro investigation of individual enzymes in the partner would give an undecose radical intermediate stabilized by
metabolic pathway24. TunA and TunF are among those implicated several key interactions (Fig. 6)44. An exo-glycal with an axial
in building its tunicamine core. Here, TunA was found to act as OH-8′ (such as that formed by the sequential action of TunA and
a 5,6-dehydratase, an ‘exo-glycal synthase’, converting UDP- then TunF) creates i′mportant consequent stabilizing interactions
GlcNAc to exo-glycal UDP-6-deoxy-GlcNAc-5,6-ene; it showed (C7′-SOMO with C8 –O8′-s*-LUMO)44. Similar radicals perform
no activity towards UDP-GalNAc. TunF exhibited broad 4-epimer- selective axial H-abstraction by virtue of quasi-homo-anomeric
ase activity, converting both UDP-GlcNAc and UDP-6-deoxy- effects44,45, which here would lead to the stereochemistry required
GlcNAc-5,6-ene to their respective galacto-epimers. Kinetic analysis for tunicamine. This effect has been previously exploited in the
showed that TunF was sevenfold more active towards the exo-glycal stereoselective synthesis of b-O-46 and C-47 glycosides. Moreover,
substrate than UDP-GlcNAc; it is an ‘exo-glycal epimerase’.
the addition of radicals to exo-glycal acceptors has a synthetic pre-
Based on these results, the previously proposed pathway for cedent48,49, not least in Gin and Myer’s chemical synthesis of tuni-
biosynthesis of the tunicamine core24 was revised (Fig. 6). TunA camycin21, which we reveal here may, at least in this aspect, be
acts first on UDP-GlcNAc. TunF then epimerizes its product at biomimetic. Importantly, the products of S. coelicolor strains
C-4′′ to yield UDP-6-deoxy-GalNAc-5,6-ene. This raises important (with/without tun cluster; with/without tunB mutation) are con-
implications for the subsequent C–C bond-forming event in the sistent with this revised pathway and the intermediacy of exo-
formation of 11-carbon tunicamine. Identification of one of the glycals. Detection of exo-glycal only from S. coelicolor M1481,
monosaccharide-derived coupling partners for this key tail-to-tail which contains the tun gene cluster but with tunB deleted, suggests
544
NATURE CHEMISTRY | VOL 4 | JULY 2012 | www.nature.com/naturechemistry
© 2012 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1351
ARTICLES
HPLC kinetic assay. Appropriate concentrations of UDP-GlcNAc were prepared in
0.05 M Tris buffer, pH 7.5, containing 5 mM MgCl2. TunA was added to a final
concentration of 0.12 mg ml21. The reaction was mixed by vortexing and incubated at
30 8C. Each reaction was stopped after 5 min by flash freezing in liquid nitrogen. Each
sample was then boiled for 1 min, filtered through viva spin (Sartorius, 10K NMWCO)
and analysed by HPLC (Supplementary Methods). Substrate and product peaks eluted
at 26 and 29 min, respectively. From the integration of these peaks a Michaelis–Menten
curve of TunA activity was constructed (Supplementary Fig. S11).
that exo-glycals are formed only transiently and at low levels in
uninterrupted biosynthesis and not at all in cells that do not
produce tunicamycin. The loss of tunicamycin production caused
by tunB deletion also confirms its essential role.
The structure of TunA is similar to those of others from the SDR
family but differs critically in active site arrangement. Indeed, we
suggest that this unique 5,6-dehydratase shares the same first
steps of other SDRs (Supplementary Fig. S6), deviating only at the
a,b-unsaturated-keto intermediate. 1,2-Hydride delivery, observed
in TunA for the first time, forms a 5,6-ene product and contrasts
with the 1,4-regioselectivity observed thus far in the SDRs
(Fig. 2). The major source of this altered regioselectivity is attributed
to subtle changes in the relative positions of the NADþ cofactor and
UDP-GlcNAc substrate within the TunA active site (Fig. 3).
TunA contains an active-site Cys that is absent from other SDR
enzymes. Within the 4,6-dehydratases DesIV/RmlB, a conserved
Asp/Glu diad is involved in dehydration across C-5′′/C-6′′. At equiv-
alent positions in TunA, Glu121 is retained while Cys120 lies in place
of Asp, and mutation of eitherabolishes activity. Although expected to
act as a general acid like aspartic acid, it is possible that the higher pKa
of cysteine may prevent unwanted protonation (which might be cata-
lysed by an Asp were this to be found at position 120) of the unique
enol ether product formed by TunA, thus protecting it from unwanted
hydrolysis in the enzyme active site. Cys120 may also provide more
favourable hydrophobic interactions with the C¼ C double bond
that is uniquely formed here.
NMR activity assay. Enzymatic reactions were performed in 25 mM Tris, pH 8.0,
with 25–100 mg of freshly purified TunA and 77 mM UDP-GlcNAc in a total
reaction volume of 300 ml. After incubation for 1 h at 37 8C, the mixture was flash
frozen and lyophilized. The resulting residue was dissolved in D2O (600 ml) and
analysed by NMR. As control experiments, the same procedures were applied to
samples lacking either TunA enzyme or UDP-GlcNAc. For isolation and full
characterization of the UDP-6-deoxy-GlcNAc-5,6-ene product see Supplementary
Methods and Supplementary Figs S8, S9 and S10.
Functional and kinetic assay of TunF. The activity of TunF was measured in a
similar way to that of TunA. See Supplementary Methods for details.
TunA crystallization and structure solution. The cubic-shaped crystals of TunA
protein, typically 115 × 115 mm and belonging to space group P1 (a ¼ 45.5,
b ¼ 51.1, c ¼ 67.8(Å), a ¼ 98.13, b ¼ 106.65, g ¼ 94.36 (8), two molecules per
asymmetric unit), were obtained after six days at 20 8C using a hanging drop method
with 20 mg ml21 protein, 20 mM Tris-HCL (pH 7.8) and 2 mM UDP-GlcNAc in
the protein solution, and 20% polyethylene glycol 3350, 100 mM 8 v/v tacsimate
(pH 6) in the reservoir solution. A 1.8 Å data set was collected from a single crystal
using a Rigaku Saturn 944þ charge-coupled device detector and FREþ SuperBright
X-ray generator equipped with a copper anode at 100 K. The X-ray data were
processed with HKL2000 and scaled with SCALEPACK51. The structure was solved
by molecular replacement using an alanine model constructed from the structure of
Together, the TunA and TunF enzymes have revealed not only dTDP-glucose 4,6-dehydratase (PDB ID 1R66)29. Successive initial model building
was carried out using the Phenix program (v1.6)52 and subsequent model
new modes of biosynthesis used in the assembly of the unique
11-carbon chain that makes up the core of tunicamycin, they
have also given insight into surprising biological intermediates
building/rebuilding were performed in COOT53,54. Model refinement was
performed with Refmac in the CCP4 program suite, by setting aside 5% of the
observed reflection data for cross-validation. The structure was refined to a
(exo-glycals) and the mechanistic subtlety behind modes of biocata-
lytic selectivity (1,2 versus 1,4 reduction) that are used for their
formation. Remarkably, this discovery of a new pathway, which
we suggest exploits the enol ether moiety of these exo-glycals,
until now had no biological precedent but does have chemical
synthetic precedent21. TunA and TunF may act together to deliver
a substrate for radical coupling that is chemically logical in ensuring
both regioselecivity and stereoselectivity (Fig. 5). It therefore pre-
sents, to our knowledge, a rare example of the post hoc validation
of a chemical synthetic strategy21 as being biomimetic.
resolution of 1.9 Å. The final model has an R factor of 18.5% and an Rfree of 23.3%.
Relevant X-ray data collection and structure solution statistics are presented in
Supplementary Table S8. Data are deposited with the code PDB ID code 3VPS.
Construction of the tunB deletion mutant. S. coelicolor M1481 (M1152 carrying
the tunB-deleted construct pIJ12541 integrated at the chromosomal wBT1
attachment (att) site) was made by first replacing tunB in pIJ12003a24 with an
apramycin resistance gene using PCR-targeting38 with the pIJ773 apramycin cassette
amplified with primers tunB20 (CTTCCAAGAGGAGGGGGCCGACTGATGA
CCGGCTACACCATTCCGGGGATCCGTCGACC) and tunB19 (GGCCTCC
CTGGACAAGGCCTACCTCACCTCCGCGCCTTCTGTAGGCTGGAGCTGC
TTC). Correct targeting was confirmed by PCR using primers tunBtstF
(TCGCGACTTCACCTACATCA) and tunBtstR (TGAGGTCGTACAGGCG
TATG). The apramycin resistance cassette was then removed by site-specific
recombination using FLP-recombinase38 generating pIJ12541. This construct was
introduced in S. coelicolor M1152 by conjugation38 to yield S. coelicolor M1481.
Correct integration of pIJ12541 at the chromosomal wBT1 att site was confirmed by
PCR using primer pairs BT1V1(GGTGCGAATAA GGGACAGTG) plus BT1C1
(CACGAGCGGAAACGTACC), and BT1C2 (GTACCAGTTGGCCGTCACC)
plus BT1V2 (ACGTCCACGAACTCACCTG).
Methods
Chemicals, enzymes and bacterial strains. Unless otherwise stated, chemical
reagents were purchased from Sigma-Aldrich, media components from Sigma-
Aldrich or Melford Laboratories and enzymes from Sigma-Aldrich or Invitrogen.
Media and general methods for the growth and propagation of E. coli were as
described in ref. 50.
Cloning and expression of tunA and tunF. The gene sequences of tunA and tunF
from S. chartreusis NRRL 3882 (Genbank accession HQ172897) were codon
optimized for E. coli and synthesized by GenScript USA. The genes, both supplied in
vector pET16b (Novagen), were introduced into E. coli BL21(DE3) by
Construction of S. coelicolor M1037. pRT80255 was introduced into S. coelicolor
M1152 by conjugation, generating S. coelicolor M1037. pRT802 integrates at the
chromosomal wBT1 att site.
transformation and expressed using standard procedures (Supplementary Methods).
Protein purification after cell harvest was performed using immobilized metal
affinity chromatography according to standard methods (Supplementary Methods).
Construction of S. coelicolor M1040. pIJ12003a, a pRT802 derivate containing the
minimal tunicamycin gene cluster24, was introduced into S. coelicolor M1152 by
conjugation, generating strain S. coelicolor M1040.
HPLC activity assay. UDP-GlcNAc 4,6-dehydratase activity was typically assayed
with a reaction mixture volume of 100 ml, incubated at 30 8C for 20 min, and
containing 5 mM UDP-GlcNAc (or UDP-GalNAc) with or without 2 mM NADþ in
50 mM Tris, pH 7.5, and with an appropriate amount of the enzyme extract. For the
trapping experiments, 25 mM sodium borodeuteride was added to the mixture and
incubated at 22 8C for an additional 10 min. Control reactions lacking particular
components were carried out to determine their necessity for enzymatic conversion.
The reaction mixtures were analysed by HPLC (Supplementary Methods).
Received 28 June 2011; accepted 3 April 2012;
published online 20 May 2012
References
1. Brandish, P. E. et al. Modes of action of tunicamycin, liposidomycin B, and
mureidomycin A: inhibition of phospho-N-acetylmuramyl-pentapeptide
translocase from Escherichia coli. Antimicrob. Agents Chemother. 40,
1640–1644 (1996).
HPLC time course assay. UDP-GlcNAc (5 mM) was prepared in 0.05 M Tris buffer,
pH 7.5, with or without 5 mM MgCl2. TunA was added to a final concentration of
0.12 mg ml21. The reaction was mixed by vortexing and incubated at 30 8C. Small
aliquots were withdrawn at a range of time intervals, filtered through viva spin
(Sartorius, 10K NMWCO) and analysed by HPLC (Supplementary Methods).
Substrate and product peaks eluted at 18 and 19 min, respectively. These were
integrated and a time course of TunA activity plotted (Supplementary Fig. S11).
2. Heifetz, A., Keenan, R. W. & Elbein, A. D. Mechanism of action of tunicamycin
on the UDP-GlcNAc:dolichyl-phosphate GlcNAc-1-phosphate transferase.
Biochemistry 18, 2186–2192 (1979).
3. Tamura, G. Tunicamycin (Japan Scientific Societies Press, 1982).
4. Hamill, R. L., Hoehn, M. H. & Boeck, L. D. Process for preparing tunicamycin.
US patent 4,336,333 (1980).
NATURE CHEMISTRY | VOL 4 | JULY 2012 | www.nature.com/naturechemistry
545
© 2012 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1351
ARTICLES
5. Takatsuki, A., Arima, K. & Tamura, G. Tunicamycin, a new antibiotic:
1. Isolation and characterization of tunicamycin. J. Antibiot. 24, 215–223 (1971).
6. Takatsuki, A. et al. Structural elucidation of tunicamycin: 2. Structure of
tunicamycin. Agric. Biol. Chem. 41, 2307–2309 (1977).
7. Tsvetanova, B. C. & Price, N. P. J. Liquid chromatography-electrospray
mass spectrometry of tunicamycin-type antibiotics. Anal. Biochem. 289,
147–156 (2001).
8. Thrum, H. et al. Streptovirudins, new antibiotics with antibacterial and antiviral
activity: 1. Culture taxonomy, fermentation and production of streptovirudin
complex. J. Antibiot. 28, 514–521 (1975).
9. Vogel, P. et al. Isolation of a group of glycolipid toxins from seedheads of annual
ryegrass (Lolium rigidum Gaud.) infected by Corynebacterium rathayi. Austr. J.
Exp. Biol. Med. Sci. 59, 455–467 (1981).
epimerase from Pseudomonas aeruginosa serotype O6. J. Biol. Chem. 275,
19060–19067 (2000).
36. Gross, J. W., Hegeman, A. D., Vestling, M. M. & Frey, P. A. Characterization of
enzymatic processes by rapid mix-quench mass spectrometry: the case of dTDP-
glucose 4,6-dehydratase. Biochemistry 39, 13633–13640 (2000).
37. Allard, S. T. M. et al. The crystal structure of dTDP-D-glucose 4,6-dehydratase
(RmlB) from Salmonella enterica serovar Typhimurium, the second enzyme in
the dTDP-L-rhamnose pathway. J. Mol. Biol. 307, 283–295 (2001).
38. Gust, B. et al. Lambda red-mediated genetic manipulation of antibiotic-
producing Streptomyces. Adv. Appl. Microbiol. 54, 107–128 (2004).
39. Kieser, T., Bibb, M. J., Buttner, M. J., Chater, K. F. & Hopwood, D. A. Practical
Streptomyces Genetics (The John Innes Foundation, 2000).
40. Kaysser, L. et al. Identification and manipulation of the caprazamycin gene
cluster lead to new simplified liponucleoside antibiotics and give insights into
the biosynthetic pathway. J. Biol. Chem. 284, 14987–14996 (2009).
41. Kaysser, L., Siebenberg, S., Kammerer, B. & Gust, B. Analysis of the
liposidomycin gene cluster leads to the identification of new caprazamycin
derivatives. ChemBioChem 11, 191–196 (2010).
10. Kenig, M. & Reading, C. Holomycin and an antibiotic (MM 19290) related
to tunicamycin, metabolites of Streptomyces clavuligerus. J. Antibiot. 32,
549–554 (1979).
11. Nakamura, S., Arai, M., Karasawa, K. & Yonehara, H. On an antibiotic,
mycospocidin. J. Antibiot. 10, 248–253 (1957).
42. Rackham, E. J., Gruschow, S., Ragab, A. E., Dickens, S. & Goss, R. J. M.
Pacidamycin biosynthesis: identification and heterologous expression of the first
uridyl peptide antibiotic gene cluster. ChemBioChem 11, 1700–1709 (2010).
43. Marsh, E. N. G., Patterson, D. P. & Li, L. Adenosyl radical: reagent and catalyst in
enzyme reactions. ChemBioChem 11, 604–621 (2010).
12. Mizuno, M., Shimojima, Y., Sugawara, T. & Takeda, I. Antibiotic 24010.
J. Antibiot. 24, 896–899 (1971).
13. Tamura, G., Sasaki, T., Matsuhashi, M., Takatsuki, A. & Yamasaki, M.
Tunicamycin inhibits formation of lipid intermediate in cell-free peptidoglycan
synthesis of bacteria. Agric. Biol. Chem. 40, 447–449 (1976).
14. Walsh, C. Where will new antibiotics come from? Nature Rev. Microbiol. 1,
65–70 (2003).
44. Giese, B. The stereoselectivity of intermolecular free radical reactions. Angew.
Chem. Int. Ed. Engl. 28, 969–980 (1989).
45. Giese, B. & Dupuis, J. Anomeric effect of radicals. Tetrahedron Lett. 25,
1349–1352 (1984).
46. Kahne, D., Yang, D., Lim, J. J., Miller, R. & Paguaga, E. The use of alkoxy-
substituted anomeric radicals for the construction of beta-glycosides. J. Am.
Chem. Soc. 110, 8716–8717 (1988).
47. Crich, D. & Lim, L. B. L. Synthesis of 2-deoxy-b-C-pyranosides by
diastereoselective hydrogen-atom transfer. Tetrahedron Lett. 31, 1897–1900 (1990).
48. Cipolla, L., Liguori, L., Nicotra, F., Torri, G. & Vismara, E. Glycomimetics via a
new glycoexoenitols-malonyl radical C–C bond formation. Chem. Commun.
1253–1254 (1996).
49. Vauzeilles, B. & Sinay, P. Selective radical synthesis of b-C-disaccharides.
Tetrahedron Lett. 42, 7269–7272 (2001).
50. Sambrook, J. & Russell, D. Molecular Cloning: A Laboratory Manual (Cold
Spring Harbor Laboratory Press, 2000).
51. Otwinowski, Z. & Minor, W. Processing of X-ray diffraction data collected in
oscillation mode. Methods Enzymol. 276, 307–326 (1997).
52. Adams, P. D. et al. PHENIX: a comprehensive Python-based system for
macromolecular structure solution. Acta Crystallogr. D 66, 213–221 (2010).
53. Emsley, P. & Cowtan, K. COOT: model-building tools for molecular graphics.
Acta Crystallogr. D 60, 2126–2132 (2004).
15. Xu, L., Appell, M., Kennedy, S., Momany Frank, A. & Price Neil, P. J.
Conformational analysis of chirally deuterated tunicamycin as an active site
probe of UDP-N-acetylhexosamine:polyprenol-P N-acetylhexosamine-1-P
translocases. Biochemistry 43, 13248–13255 (2004).
16. Elbein, A. D. The tunicamycins — useful tools for studies on glycoproteins.
Trends Biochem. Sci. 6, 219–221 (1981).
17. Danishefsky, S. J., Deninno, S. L., Chen, S., Boisvert, L. & Barbachyn, M. Fully
synthetic stereoselective routes to the differentially protected subunits of the
tunicamycins. J. Am. Chem. Soc. 111, 5810–5818 (1989).
18. Ichikawa, S. & Matsuda, A. Synthesis of tunicaminyluracil derivatives.
Nucleosides Nucleotides Nucleic Acids 23, 239–253 (2004).
19. Karpiesiuk, W. & Banaszek, A. Stereoselective syntheses of the O,N-protected
subunits of the tunicamycins. Carbohydr. Res. 299, 245–252 (1997).
20. Ramza, J. & Zamojski, A. New convenient synthesis of tunicamine. Tetrahedron
48, 6123–6134 (1992).
21. Myers, A. G., Gin, D. Y. & Rogers, D. H. Synthetic studies of the tunicamycin
antibiotics. Preparation of (þ)-tunicaminyluracil, (þ)-tunicamycin-V, and
5′-epi-Tunicamycin-V. J. Am. Chem. Soc. 116, 4697–4718 (1994).
22. Suami, T., Sasai, H., Matsuno, K. & Suzuki, N. Synthetic approaches toward
antibiotic tunicamycins. Part VIII. Total synthesis of tunicamycin. Carbohydr.
Res. 143, 85–96 (1985).
54. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of
COOT. Acta Crystallogr. D 66, 486–501 (2010).
55. Gregory, M. A., Till, R. & Smith, M. C. M. Integration site for Streptomyces
phage BT1 and development of site-specific integrating vectors. J. Bacteriol. 185,
5320–5323 (2003).
23. Tsvetanova, B. C., Kiemle, D. J. & Price, N. P. J. Biosynthesis of tunicamycin and
metabolic origin of the 11-carbon dialdose sugar, tunicamine. J. Biol. Chem. 277,
35289–35296 (2002).
24. Wyszynski, F. J., Hesketh, A. R., Bibb, M. J. & Davis, B. G. Dissecting
tunicamycin biosynthesis by genome mining: cloning and heterologous
expression of a minimal gene cluster. Chem. Sci. 1, 581–589 (2010).
25. Chen, W. et al. Characterization of the tunicamycin gene cluster unveiling
unique steps involved in its biosynthesis. Prot. Cell 1, 1093–1105 (2010).
26. Karki, S., Kwon, S-Y. & Kwon, H-J. Cloning of tunicamycin biosynthetic gene
cluster from Streptomyces chartreusis NRRL 3882. J. Kor. Soc. Appl. Biol. Chem.
54, 136–140 (2011).
56. Chen, W. et al. Characterization of the polyoxin biosynthetic gene cluster from
Streptomyces cacaoi and engineered production of polyoxin H. J. Biol. Chem.
284, 10627–10638 (2009).
57. Ginj, C., Ruegger, H., Amrhein, N. & Macheroux, P. 3′-Enolpyruvyl-UMP, a
novel and unexpected metabolite in nikkomycin biosynthesis. ChemBioChem 6,
1974–1976 (2005).
Acknowledgements
27. Berman, H., Henrick, K. & Nakamura, H. Announcing the worldwide Protein
Data Bank. Nature Struct. Biol. 10, 980–980 (2003).
28. Allard, S. T. M. et al. Toward a structural understanding of the dehydratase
mechanism. Structure 10, 81–92 (2002).
29. Allard, S. T. M., Cleland, W. W. & Holden, H. M. High resolution X-ray structure
of dTDP-glucose 4,6-dehydratase from Streptomyces venezuelae. J. Biol. Chem.
279, 2211–2220 (2004).
The authors gratefully acknowledge B. Odell (Oxford University) for help with the NMR
experiments. This work was supported by the EPSRC (DTA studentship for F.J.W.), the Bill
and Melinda Gates Foundation (S.S.L.) and BBSRC (J.P.G-E. and M.J.B). G.J.D and B.G.D.
are Royal Society Wolfson Research Merit Award recipients. This manuscript is dedicated
to the memory of Professor David Gin.
30. Demendi, M., Ishiyama, N., Lam, J. S., Berghuis, A. M. & Creuzenet, C. Towards Author contributions
a better understanding of the substrate specificity of the UDP-N-
acetylglucosamine C4 epimerase WbpP. Biochem. J. 389, 173–180 (2005).
31. Kowal, P. & Wang, P. G. New UDP-GlcNAc C4 epimerase involved in the
biosynthesis of 2-acetamino-2-deoxy-L-altruronic acid in the O-antigen
repeating units of Plesiomonas shigelloides O17. Biochemistry 41,
15410–15414 (2002).
F.J.W. cloned tunF, purified the protein and performed functional and kinetic analyses of
TunF. S.S.L. and T.Y. cloned tunA and its mutants, purified the resulting proteins and
performed functional analyses. S.S.L. performed the kinetic analyses. TunAwas crystallized
by T.Y. The 3D structure was determined by S.J.L., and S.S.L., B.G.D and G.J.D. carried out
its analysis. H.W. and J.P.G-E. created the tunB mutant strain and extracts. M.J.B. designed
the deletion strategy. H.W., M.J.B. and B.G.D. analysed the extracts. The manuscript was
written by F.J.W., S.S.L., M.J.B., G.J.D. and B.G.D.
32. Jornvall, H. et al. Short-chain dehydrogenases reductases (SDR). Biochemistry
34, 6003–6013 (1995).
33. Filling, C. et al. Critical residues for structure and catalysis in short-chain
dehydrogenases/reductases. J. Biol. Chem. 277, 25677–25684 (2002).
34. Schulz, J. M. et al. Determinants of function and substrate specificity in human
UDP-galactose 4′-epimerase. J. Biol. Chem. 279, 32796–32803 (2004).
35. Creuzenet, C., Belanger, M., Wakarchuk, W. W. & Lam, J. S. Expression,
purifcation, and biochemical characterization of WbpP, a new UDP-GlcNAc C4
Additional information
The authors declare competing financial interests: details accompany the full-text HTML
version of the paper at www.nature.com/naturechemistry. Supplementary information
accompanies this paper at www.nature.com/naturechemistry. Reprints and permission
information is available online at http://www.nature.com/reprints. Correspondence and
requests for materials should be addressed to B.G.D.
546
NATURE CHEMISTRY | VOL 4 | JULY 2012 | www.nature.com/naturechemistry
© 2012 Macmillan Publishers Limited. All rights reserved.