Biosynthetic origin and mechanism of formation of the aminoribosyl moiety of peptidyl nucleoside antibiotics

Article abstract of DOI:10.1021/ja206304k

Several peptidyl nucleoside antibiotics that inhibit bacterial translocase I involved in peptidoglycan cell wall biosynthesis contain an aminoribosyl moiety, an unusual sugar appendage in natural products. We present here the delineation of the biosynthetic pathway for this moiety upon in vitro characterization of four enzymes (LipM-P) that are functionally assigned as (i) LipO, an l-methionine:uridine-5′-aldehyde aminotransferase; (ii) LipP, a 5′-amino-5′-deoxyuridine phosphorylase; (iii) LipM, a UTP:5-amino-5-deoxy-α-d-ribose-1-phosphate uridylyltransferase; and (iv) LipN, a 5-amino-5-deoxyribosyltransferase. The cumulative results reveal a unique ribosylation pathway that is highlighted by, among other features, uridine-5′-monophosphate as the source of the sugar, a phosphorylase strategy to generate a sugar-1-phosphate, and a primary amine-requiring nucleotidylyltransferase that generates the NDP-sugar donor.

Full text of DOI:10.1021/ja206304k

ARTICLE
pubs.acs.org/JACS
Biosynthetic Origin and Mechanism of Formation of the Aminoribosyl
Moiety of Peptidyl Nucleoside Antibiotics
Xiuling Chi,† Pallab Pahari,† Koichi Nonaka,‡ and Steven G. Van Lanen*,†
†Department of Pharmaceutical Sciences, College of Pharmacy, University of Kentucky, 789 South Limestone, Lexington,
Kentucky 40536, United States
‡Biopharmaceutical Research Group I, Biopharmaceutical Technology Research Laboratories, Pharmaceutical Technology Division,
Daiichi Sankyo Company, Ltd., 389-4 Aza-ohtsurugi, Shimokawa, Izumi-machi, Iwaki-shi, Fukushima 971-8183, Japan
S Supporting Information
b
ABSTRACT: Several peptidyl nucleoside antibiotics that inhibit
bacterial translocase I involved in peptidoglycan cell wall biosynthesis
contain an aminoribosyl moiety, an unusual sugar appendage in natural
products. We present here the delineation of the biosynthetic pathway
for this moiety upon in vitro characterization of four enzymes (LipM-P)
that are functionally assigned as (i) LipO, an L-methionine:uridine-50-
aldehyde aminotransferase; (ii) LipP, a 50-amino-50-deoxyuridine phos-
phorylase; (iii) LipM, a UTP:5-amino-5-deoxy-R-D-ribose-1-phosphate
uridylyltransferase; and (iv) LipN, a 5-amino-5-deoxyribosyltransferase.
The cumulative results reveal a unique ribosylation pathway that is
highlighted by, among other features, uridine-50-monophosphate as the source of the sugar, a phosphorylase strategy to generate a sugar-
1-phosphate, and a primary amine-requiring nucleotidylyltransferase that generates the NDP-sugar donor.
’ INTRODUCTION
Figure S1a). This tandem, enzyme-catalyzed process is also
utilized during O-ribosylation of decaprenyl-phosphate to initiate
the biosynthesis of mycobacterial arabinogalactan.4 Along with the
wealth of N- and C-ribosides that originate via phosphoribosyl-
transfer from 5-phospho-R-D-ribose-1-diphosphate (Figure S1a),5
it would appear that ribosylation is an exception to the typical
glycosylation paradigm.
We have been studying the biosynthesis of several families of
nucleoside antibiotics that inhibit the enzyme bacterial translo-
case I (MraY) involved in peptidoglycan cell wall biosynthesis,6
and all of these antibiotics contain unusual sugar appendages. The
lipopeptidyl nucleoside family of translocase I inhibitors, which
includes A-90289s from Streptomyces sp. SANK 60405,7 capraza-
mycins from Streptomyces sp. MK739-62F,8 FR-900493 from
Bacillus cereus No. 2045,9 and muraymycins from Streptomyces
sp. NRRL 3047110 contain an aminoribosyl moiety—a 5-amino-
5-deoxyribose—attached via an O-glycosidic bond to a heptofur-
anose nucleoside component, 50-C-glycyluridine (Figure 1).
Structureꢀactivity relationship studies using simplified synthetic
analogues of these compounds have revealed this aminoribosyl
moiety,11 and specifically the primary amine functionality,12 is
critical for optimal antibiotic activity.
Glycosylation is a common modification found in several
microbial natural products of therapeutic value, and the sugar
appendage typically has a profound effect on the biological
activity.1,2 The modus operandi by which sugars are generally
incorporated into molecular scaffolds such as natural products is
(i) formation of a sugar-1-phosphate derived from a glycolytic
intermediate or galactose, a catalytic process that requires either a
phosphosugar mutase or an anomeric sugar kinase; (ii) conver-
sion of the sugar-1-phosphate to usually an NDP-sugar (also
called an activated sugar), a reaction catalyzed by a sugar-1-
phosphate nucleotidylyltransferase; and (iii) transfer of the sugar
to an acceptor substrate by a glycosyltransferase to typically
generate a new O-, N-, or aryl-C-glycosidic bond. A remarkable
feature of glycosylated natural products is the high degree of
variability and functionality that can be incorporated into the
sugar moiety by reductases, epimerases, methyltransferases, and
aminotransferases, among others—enzymatic modifications that
generally occur at the level of the activated sugar prior to the
glycosyltransferase-catalyzed reaction.1,2
A distinct strategy has recently been identified for generating
glycosidic bonds with ribose units during natural product bio-
synthesis. The ribosyl moiety of the aminoglycoside antibiotic
butirosin was shown to be derived from 5-phospho-R-D-ribose-1-
diphosphate,3 from which BtrL transfers ribose-5-phosphate to
the acceptor disaccharide neamine to generate an O-glycosidic
bond and a second enzyme BtrP catalyzes dephosphorylation to
form the final trisaccharide scaffold (Supporting Information
The first insight into how the disaccharide core is assembled
was unveiled upon cloning of the biosynthetic gene clusters for
four lipopeptidyl nucleoside antibiotics,7,13ꢀ15 which revealed six
Received: July 7, 2011
Published: August 05, 2011
r
2011 American Chemical Society
14452
dx.doi.org/10.1021/ja206304k J. Am. Chem. Soc. 2011, 133, 14452–14459
|

Journal of the American Chemical Society
ARTICLE
shared genes encoding a putative serine hydroxymethyltransfer-
ase (lipK for the A-90289 gene cluster), a nonheme, Fe(II)-
dependent dioxygenase (lipL), a putative nucleotidylyltransfer-
ase (lipM), glycosyltransferase (lipN), aminotransferase (lipO),
and uridine phosphorylase (lipP). We have demonstrated that
LipL is a nonheme, Fe(II)-dependent R-ketoglutarate:UMP
dioxygenase that catalyzes the conversion of UMP to uridine-
50-aldehyde (1) during A-90289 biosynthesis (Figure 2).16 In
turn, we proposed that 1 serves as the substrate for LipK, which
catalyzes an aldol-type reaction using glycine as a co-substrate to
generate 50-C-glycyluridine (Figure 2). On the basis of the
conjecture that the most efficient overall biosynthetic pathway
will be employed, we subsequently hypothesized 1 is also an
intermediate in the pathway leading to the aminoribosyl moiety,
which would necessitate aminotransfer, phosphorolysis, ribose
activation, and ribosyltransfer by LipO, LipP, LipM, and LipN,
respectively (Figure 2).
instead utilizes an NDP-sugar as the activated, sugar donor.
We now present the delineation of the biosynthetic pathway for
incorporating this moiety by functionally assigning four enzymes,
LipMꢀP; the results reveal a unique O-ribosylation pathway that
indeed parallels the typical glycosylation paradigm yet with
significant distinctions that are disclosed herein.
’ RESULTS
Functional Assignment of LipO as an L-Methionine:1
Aminotransferase. LipO has modest sequence similarity to
several proteins predicted to belong to the pyridoxal-phosphate
(PLP)-dependent aspartate aminotransferase (Type I) super-
family.17,18 Of note is the sequence similarity of LipO to PacE
(37% identity/51% similarity) involved in the biosynthesis of the
pacidamycin family of antibiotics. The pacidamycins consist of an
enamide-containing nucleoside with a 50-amine functionality as
the sole ribose-derived unit, which we and others have previ-
ously speculated proceeds through 1 as an intermediate16,19ꢀ21
(Figure S2), which would necessitate transamination at the
nucleoside level to yield the nucleoside building block. Thus, we
envisioned an analogous biosynthetic pathway such that LipO
catalyzes aminotransfer utilizing the corresponding nucleoside 1.
To interrogate the mechanism of amine incorporation, lipO
was expressed in Escherichia coli; however, the recombinant
protein was insoluble using a variety of growth and induction
conditions. Therefore, we turned to Streptomyces lividans TK64
as a host, which resulted in the successful preparation of soluble,
recombinant LipO (Figure S3a) that was shown by UV/vis
spectroscopy to co-purify with the cofactor PLP (Figure S3b).
The mutant LipO(K282A), expected to be unable to form an
internal aldimine with Lys and hence be inactive, was also
produced in S. lividans TK64 to yield a purified protein devoid
of bound PLP (Figure S3b).
The apparent involvement of a nucleotidylyltransferase and
glycosyltransferase suggested the pathway leading to the aminor-
ibosyl moiety does not parallel the known ribosylation pathways
that are 5-phospho-R-D-ribose-1-diphosphate-dependent but
The activity of LipO was next monitored by HPLC using
diode array for detection, and to simplify the analysis, the
putative product 50-amino-50-deoxyuridine (2) was synthesized.
When LipO was incubated with 1 and L-glutamate or L-aspartate,
two common amine donors for the Type 1 aminotransferase
superfamily, a small peak representing <1% conversion appeared
in both cases, which was confirmed as 2 by LCꢀMS and co-
injections with authentic 2 (Figure S3c); in contrast, no new peak
was observed with LipO(K282A) (Figure S3d). Further analysis
revealed several potential amine donors were substrates—findings
Figure 1. Structure of representative lipopeptidyl nucleoside antibiotics
containing an aminoribosyl moiety.
Figure 2. Proposed biosynthesis of the aminoribosyl moiety of A-90289 antibiotics.
14453
dx.doi.org/10.1021/ja206304k |J. Am. Chem. Soc. 2011, 133, 14452–14459

Journal of the American Chemical Society
ARTICLE
dependent upon phosphate and a nonphosphorylated nucleoside
(Figure S4b), and the reaction proceeded with thymidine (Figure
S4c) but not cytidine (Figure S4d) or purine nucleosides as
substrates (data not shown). In addition to the formation of the
pyrimidine base, LCꢀMS analysis of the reaction revealed an
(M ꢀ H)ꢀ ion at m/z = 228.8 that was absent in the control
(Figure S6), a mass that is consistent with the molecular formula
C5H9NO8P of the expected product R-D-ribose-1-phosphate.
Further activity tests revealed the hypothetical pathway inter-
mediates 1 and 2 were also converted to 4 and the respective
sugar-1-phosphate by LipP (Figure S7 and Figure 3c, respectively)
and EcUdp (Figure S5e,f), thus warranting further kinetic inves-
tigation for both enzymes.
The pH profile for LipP and EcUdp was initially examined
using 2 or 3 as a substrate and detection of 4 by UV/vis
spectroscopy (Figure S8a),25 revealing an apparent optimal
activity for LipP at pH = 9.0, which is moderately higher than
reported for EcUdp (pH ∼ 7.5) (Figure S8b).26 As a result, single
substrate kinetic analysis was performed at both pH 9.0 and 7.5
with all hypothetical pathway intermediates (Figure 3d,e and
Figures S5 and S8cꢀf). The extracted kinetic constants (Table 1)
revealed that LipP has comparable efficiency with 2 and 3 at pH
9.0; thus, both appear to be viable substrates in vivo. Saturation
kinetics could not be reached with 1 (Figure S8f), and the
relatively low first-order rate constant of k = (3.0 ( 0.5) ꢁ 10ꢀ2
minꢀ1 suggests that 1 is less probable as the in vivo substrate.
Although the Km and kcat for each respective substrate were lower
for LipP relative to EcUdp, the efficiencies and kinetic trends for
both enzymes were quite comparable. Furthermore, bisubstrate
kinetic analysis of LipP resulted in intersecting double reciprocal
plots consistent with a sequential kinetic mechanism, also akin to
the mechanism reported for EcUdp (Figure S8g).27,28
Figure 3. Characterization of LipP. (a) Phosphorylase reaction cata-
lyzed by LipP. (b) HPLC analysis using 3 after (I) 30 min without LipP,
(II) 5 min reaction, (III) 30 min reaction, and (IV) authentic 4. (c)
HPLC analysis using 2 after (I) 30 min without LipP, (II) 30 min
reaction, and (III) authentic 4. (d) Single-substrate kinetic analysis with
variable 3. (f) Single-substrate kinetic analysis with variable 2. A260
,
absorbance at 260 nm.
Functional Assignment of LipM as a Primary Amine-
Requiring Nucleotidylyltransferase. Bioinformatics analysis
of LipM revealed sequence similarity to proteins annotated as
putative nucleotidylyltransferases. Similarly to LipO, the gene
product of lipM was only soluble when expressed in S. lividans
TK64 (Figure S9a). The activity of LipM was tested with R-D-
ribose-1-phosphate or 5-amino-5-deoxy-R-D-ribose-1-phosphate
(5) generated in situ by LipP or EcUdp, and analysis of these
reactions revealed a new peak was formed only in the presence
of 5 and UTP or—to a lesser extent—TTP, CTP, or GTP
(Figure 4a,b and Figure S9bꢀd). In contrast no new peaks were
detected with commercially available R-D-glucose-1-phosphate,
R-D-galactose-1-phosphate, or R-D-glucosamine-1-phosphate and
co-substrate UTP. Unexpectedly, the new peak derived from
activity tests with 5 and the different NTPs had the same retention
time and UV/vis spectrum as the respective nucleotide monopho-
sphate (NMP), and co-injections with authentic material and
LCꢀMS revealed this to be the case.
similar to many other characterized aminotransferases;17,18 how-
ever, the highest specific activity was obtained with L-methionine
followed by N-acetylcysteine, L-arginine, and S-adenosyl-L-methio-
nine (Table S1). Finally, the reverse reaction of LipO using 2 and
4-methylthio-2-oxobutanoate or other amine acceptors as sub-
strates was not observed, suggesting that the equilibrium of
aminotransfer favors formation of 2. The results are consistent
with the functional assignment of LipO as a methionine:1 amino-
transferase and, importantly, that 2 is the likely intermediate in the
biosynthesis of the aminoribosyl moiety of A-90289 (Figure 2).
Functional Assignment of LipP as a Low Specificity Phos-
phorylase. Bioinformatics analysis revealed LipP has sequence
similarity to proteins annotated as uridine phosphorylases
(Udp), an enzyme family that catalyzes the phosphorolysis of
uridine (3) or—less efficiently—thymidine nucleosides to gener-
ate R-D-ribose-1-phosphate and uracil (4) or thymine, respectively,
to initiate nucleotide salvage pathways (Figure 3a).22 Although
inactive with the 50-monophosphorylated nucleotide, both human
and mouse Udp have been shown to have modest activity with
unnatural 50-deoxynucleosides.23,24 However, the potential activity
of Udp with other 50-modifications has not been reported, so it
remained unclear as to the chemical identity of the LipP substrate.
The lipP gene was cloned and expressed in E. coli to yield
soluble protein (Figure S4a). For comparisons, the udp gene
from E. coli was also cloned and expressed to yield recombinant
EcUdp (Figure S5a). Initial activity tests using HPLC revealed 3
is rapidly converted to 4 and R-D-ribose-1-phosphate by both
LipP (Figure 3b) and EcUdp (Figure S5b) as expected. Identical
to prior reports with EcUdp, the LipP-catalyzed reaction was
We hypothesized that NMP was generated by the degradation
of the product via intramolecular attack of the 2-hydroxy group of
the aminoribosyl unit on the proximal phosphate, a phenomenon
that was previously observed upon characterization of apiofurano-
syl-1,2-cyclic phosphate as the product of the plant bifunctional
UDP-apiose/UDP-xylose synthase that catalyzes decarboxylation
of UDP-glucuronic acid.29 To detect any potential amine-containing
product, the reaction components were first modified with
o-phthalaldehyde (OPA) prior to injection, and two new peaks
were identified by HPLC (Figure 4c). The first peak was identi-
fied as residual OPA-modified 5, while LCꢀMS analysis of the
remaining peak yielded an (M ꢀ H)ꢀ ion at m/z= 385.6, consistent
14454
dx.doi.org/10.1021/ja206304k |J. Am. Chem. Soc. 2011, 133, 14452–14459

Journal of the American Chemical Society
ARTICLE
Table 1. Extracted Kinetic Constants for LipP and EcUdp
enzyme
LipP
pH
9.0
substratea
Km (μM)
kcat (minꢀ1
)
kcat/Km (μMꢀ1 minꢀ1
)
relative efficiency
Phosphate
Uridine (3)
Thymidine
ADU (2)
20-deoxyU
Uridine
73 ( 6.0
(1.3 ( 0.1) ꢁ 102
(1.3 ( 0.1) ꢁ 102
5.3 ( 0.2
(2.6 ( 0.2) ꢁ 102
(2.1 ( 0.8) ꢁ 102
NAb
87 ( 21
1.5
100
0.13
80
(2.8 ( 0.2) ꢁ 103
(2.1 ( 0.3) ꢁ 102
(3.1 ( 0.4) ꢁ 103
NAb
1.9 ꢁ 10ꢀ3
1.2
6.8 ꢁ 10ꢀ2
4.5
7.5
9.0
7.5
ADU (2)
Uridine (3)
ADU (2)
Uridine (3)
ADU (2)
(2.6 ( 0.5) ꢁ 102
(1.3 ( 0.2) ꢁ 103
(1.0 ( 0.2) ꢁ 103
(1.5 ( 0.1) ꢁ 103
(2.0 ( 0.5) ꢁ 103
34 ( 1.6
1.3 ꢁ 10ꢀ1
2
8.7
100
49
EcUdp
(2.6 ( 0.1) ꢁ 103
(9.7 ( 0.7) ꢁ 102
(1.6 ( 0.1) ꢁ 103
(3.2 ( 0.4) ꢁ 102
1.0
1.1
1.6 ꢁ 10ꢀ1
55
8
a ADU, 50-amino-50-deoxyuridine; 20-deoxyU, 20-deoxyuridine. b Not applicable, non-MichaelisꢀMenten kinetics was observed.
this stage, the results did reveal that LipM only utilizes a sugar-1-
phosphate containing a primary amine functionality.
We rationalized that a stable LipM product would be attainable
by using a 2-deoxyribose-containing surrogate substrate. EcUdp as
well as other UDPs are known to catalyze phosphorolysis using
thymidine or 2-deoxy-3,26,27 and similarly LipP catalyzed the
reaction with either substrate (Figure S4c,e), although the latter
was determined to be nearly 35-fold more efficient (Table 1 and
Figure S16). Identical to the results utilizing 50-hydroxy nucleosides
(i.e., 3), in situ generation of 2-deoxy-R-D-ribose-1-phosphate did
not yield a product when tested with LipM. Subsequently, the
potential substrate 50-amino-20,50-dideoxyuridine (7) was synthe-
sized from 20-deoxy-3, and HPLC analysis with this surrogate
substrate revealed it was processed by both LipP and LipM,
generating two new peaks with 3-like chromophores (UVmax
∼
260 nm) (Figure 5b). While the minor peak was identified as UDP,
the major peak did not co-elute with any known 4-containing
metabolite. LCꢀMS analysis of the purified new peak revealed an
(M ꢀ H)ꢀ ion at m/z = 517.6, consistent with the molecular
formula C14H23N3O14P2 of UDP-5-amino-2,5-dideoxyribose (8)
(expected m/z = 518.1) (Figure S17). NMR analysis, including
1
1
1
13
1H-, 13C-, and 31P NMR, Hꢀ H COSY, and Hꢀ C HMBC
(Figures S18ꢀS20), confirmed the identity of the product as 8,
thus consistent with the function of LipM as a UTP:5-amino-5-
deoxy-R-D-ribose-1-phosphate uridylyltransferase.
Figure 4. Characterization of LipM. (a) Reaction catalyzed by LipM
including in situ generation of the substrate and amine derivatization
with o-phthalaldehyde (OPA). (b) HPLC analysis starting with 2 after
(I) 30 min without 2, (II) 30 min without LipP or EcUdp, (III) 30 min
reaction, (IV) 3 h reaction, and (V) authentic UMP. (c) HPLC analysis
of reaction mixtures with OPA derivatization starting with 2 after (I)
30 min without LipP or EcUdp, (II) 30 min without LipM (II), 30 min
Functional Assignment of LipN as a 5-Amino-5-deoxyr-
ibosyltransferase. We finally turned our attention to LipN,
which has low sequence identity to a small number of proteins
annotated as putative glycosyltransferases. Once again, soluble
protein was only obtained upon heterologous expression in S.
lividans TK64 (Figure S21a). We were disheartened to find out
that 8 was not stable upon storage, in this case degrading to UDP
and an undetermined product likely by hydrolysis of the anomeric
bond (Figure S21b). Thus, we again relied on the in situ genera-
tion of substrate beginning with 2 or 7. While no glycosyltransfer-
ase activity was observed with uridine-50-diphosphoglucose,
PRPP, or ribose-1-phosphates generated with LipP, HPLC
analysis of LipN reactions using surrogate acceptor 3 revealed a
new, small peak using either 2 (Figure 6) or 7 (Figure S21c) as the
ultimate sugar donor. Our pilot experiments suggested higher yields
were obtained with the genuine sugar donor UDP-5-amino-5-
deoxy-R-D-ribose, and thus, no further experiments were under-
taken with 8. Following large-scale purification of the LipN-product
generated from 2 and 3, mass and complete NMR spectroscopic
reaction (III), and 3 h reaction (IV). A260, absorbance at 260 nm; A335
,
absorbance at 335 nm.
with the molecular formula C15H18NO7PS of an OPA-modified
5-amino-5-deoxy-R-D-ribose-1,2-cyclicphosphate, 6 (expected
m/z = 386.1) (Figure 4a and Figure S10a,b). Large-scale purification
of OPA-6 and subsequent LCꢀMS and 1D and 2D NMR
spectroscopic characterization (Figures S10c,dꢀS14)—notably
1
31
the Hꢀ P HMBC data that was consistent with the very
recently discovered metabolite R-D-ribose-1,2-cyclicphosphate30
(Figure S15)—revealed the expected degradation product (an
isoindol-1-one) for OPA-6.31 Although the genuine identity of
the product of the LipM-catalyzed reaction remained elusive at
14455
dx.doi.org/10.1021/ja206304k |J. Am. Chem. Soc. 2011, 133, 14452–14459

Journal of the American Chemical Society
ARTICLE
Figure 5. Characterization of LipM. (a) Enzymatic preparation of the substrate using LipP and the dideoxyuridine analogue 7 and the reaction catalyzed
by LipM to generate 8. (b) HPLC analysis of the reaction starting with 7 after (I) 3 h without UTP and LipP, (II) 3 h without LipM, (III) 1 h reaction,
(IV) 3 h reaction, and (V) authentic UDP. A260, absorbance at 260 nm.
13
analysis (Figures S22ꢀ25)—notably the 1Hꢀ C HMBC demon-
strating the H-100 and C-50 correlation (Figure S26)—revealed the
identity of the new product as 50-O-(500-amino-500-deoxy-β-D-
ribose)-uridine, 9 (Figure 6). Thus, the function of LipN is assigned
as a 5-amino-5-deoxy-R-D-ribosyltransferase that catalyzes the term-
inal step in the biosynthesis of the aminoribosyl moiety.
the pathway,16 an attribute that was not reciprocated by the four
enzymes assigned in this study. LipO utilized a variety of amine
donors with L-methionine as the slightly preferred amine source,
the biological relevance of which is currently under investigation.
Likewise, the phosphorylase LipP was equally efficient with
hypothetical pathway intermediates 2 and 3. Sequence analysis of
the whole genomes of the Actinomycetales has revealed minimally
one 50-nucleosidase is encoded within the chromosomal DNA
suggesting these organisms have the capability to convert the
nucleic acid building block UMP to 3. Thus, based solely on the
results with LipP, the identity of the in vivo substrate could not
be established. However, the realization that the amine needs to be
incorporated prior to formation of activated sugar would necessi-
tate an additional, unidentified enzyme to oxidize R-D-ribose-1-
phosphate prior to LipO catalysis if 3 was indeed the pathway
precursor. Thus, we currently prefer a pathway originating from
UMP without the involvement of 3 as shown in Figure 2.
While we were initially disheartened by the lack of substrate
specificity of LipP, this low specificity turned out to be critical for
discovering the function of LipM by enabling the preparation of a
surrogate substrate that was converted to a less unstable product
for structural elucidation. Similarly, we took advantage of
the promiscuity of LipN with respect to the sugar acceptor to
define its function as a ribosyltransferase. Interestingly, although
the genuine sugar donor UDP-5-amino-5-deoxy-R-D-ribose was
initially only indirectly identified as the LipM-product based on
the identification of UMP and OPA-6 and—despite exhaustive
’ DISCUSSION
Bacterial natural products are notorious for their diverse array
of sugar modifications that are typically critical for their biological
activity. As a result there has been a significant effort toward
understanding the molecular details behind incorporation of these
sugars, in part with the expectation that the results will ultimately
enable rationale manipulation of sugar biosynthetic pathways as a
strategy to produce novel glycosylated compounds.1,2 Undoubt-
edly, the identification and characterization of new pathways will
afford more tools for applying such a structural diversification
approach, and thus, we initiated studies toward delineating the
biosynthetic mechanism of an unique pentofuranose—an amino-
ribosyl moiety found in several lipopeptidyl nucleosides of poten-
tial therapeutic significance as antibiotics.
Our strategy to unravel the pathway was to reconstitute the
enzyme activities in vitro, which ultimately required the use of the
heterologous host S. lividans TK64 to obtain three of the four
enzymes in soluble form. Our prior results with LipL demonstrated
this enzyme had strict specificity for the substrate UMP to initiate
14456
dx.doi.org/10.1021/ja206304k |J. Am. Chem. Soc. 2011, 133, 14452–14459

Journal of the American Chemical Society
ARTICLE
phosphates.42ꢀ44 Although a more in depth investigation is
underway, our results identified LipM as an unusual R-D-
pentofuranosyl-1-phosphate-activating enzyme with an appar-
ently more refined specificity and hence well-defined role in
A-90289 biosynthesis. Furthermore, LipM was shown to have
an absolute requirement for the primary amine functionality,
which is likewise unusual relative to other bacterial nucleotidy-
lyltransferases that have been demonstrated to activate unna-
tural aminohexoses, although with equal or less efficiency
relative to the hydroxylated counterpart.43ꢀ47
’ CONCLUSION
In summary, we have completed the functional assignment of
five enzymes—including four here—involved in the modifica-
tion and incorporation of the aminoribosyl moiety of A-90289
antibiotics. The pathway is initiated by the LipL-catalyzed
conversion of a novel ribosyl donor UMP to form the aldehyde
1.16 Following introduction of the amine by LipO, LipP catalyzes
phosphorolysis to initiate “salvage” of the modified sugar and
LipM subsequently catalyzes the formation of an activated
pentofuranose that serves as the final sugar donor. Finally, the
ribosyltransferase LipN completes the pathway by incorporation
of the aminoribosyl moiety, the identity of which is apparently
dictated by the specificity of the pathway-initiating enzyme LipL
and the penultimate enzyme LipM. The end result is a sugar
biosynthetic pathway highlighted in Figure 2 that not only
establishes an alternative mechanism for ribosylation, but also
features intriguing variations of the established paradigm for
glycosylation.
Figure 6. Characterization of LipN. (a) Enzymatic preparation of the
sugar donor substrate using LipP and LipM, and the reaction catalyzed
by LipN using the surrogate acceptor substrate 3 to generate 9. (b)
HPLC analysis of the reaction starting with 2 after (I) 3 h without LipN
and (II) 3 h reaction. Asterisk (*) indicates expected retention time for
residual UTP and co-product UDP. A260, absorbance at 260 nm.
’ EXPERIMENTAL METHODS
Cloning, Gene Expression, and Protein Production. Genes
for heterologous expression in E. coli were cloned into pET30Xa-LIC
(Supplementary Methods). Genomic DNA from E. coli DH5R and
cosmid pN17 was used as templates for PCR, and the fidelity of the
resulting product was confirmed by DNA sequencing. Standard condi-
tions were used for expression of lipP and EcUdp using BL21(DE3) with
the respective plasmid.48 The remaining genes, lipM, lipN, lipO, and
lipO(K282A) were subcloned into pUWL201pw and expressed in
S. lividans TK64.49 Protein purification following cell harvesting was
performed using immobilized metal affinity chromatography according
to standard procedures (see Supporting Information for details).
In Vitro Characterization of LipO. Reactions consisted of
50 mM potassium phosphate, pH 7.5, 2 mM 1, 2 mM amine donor,
200 μM PLP, and 1 μM LipO at 30 °C, and the reaction was terminated
by the addition of cold TCA to 5% (w/v) or by ultrafiltration using a
Microcon YM-3. Alternatively, PLP was eliminated from the reaction
mixture to give comparable results. Following centrifugation to remove
protein, the reaction components were analyzed by HPLC using a C-18
reverse-phase column. A series of linear gradients was developed
from 0.1% TFA in 5% acetonitrile (A) to 0.1% TFA in 90%
acetonitrile (B) in the following manner (beginning time and ending
time with linear increase to % B): 0ꢀ4 min, 100% B; 4ꢀ24 min, 50%
B; 24ꢀ26 min, 100% B; 26ꢀ32 min, 100% B; and 32ꢀ35 min, 0% B.
The flow rate was kept constant at 1.0 mL/min, and elution was
monitored at 260 nm. LCꢀMS was performed using a linear gradient
from 0.1% formic acid in water to 0.1% formic acid in acetonitrile over
20 min. The flow rate was kept constant at 0.4 mL/min, and elution
was monitored at 254 nm.
attempts—could not be directly observed by MS, our successful
analysis of LipN with in situ generated UDP-5-amino-5-deoxy-R-
D-ribose suggests that this activated sugar is formed yet is
refractive to direct characterization using the conditions em-
ployed here. Although it remains a possibility that 6 is generated
enzymatically by a contaminating protein, several activated
sugars are known to be unstable; for instance, it has recently
been reported that the expected product of UDP-apiose/UDP-
xylose synthase can be transiently detected by high-field
NMR prior to degrading to the isolable apiofuranosyl-1,2-cyclic
phosphate29 and the activated carbocycle NDP-valienol can only
be detected by MS within the crude reaction mixture.32
Although the glycosylation process predominantly involves
the incorporation of hexoses, it is not limited to these sugars:
several pentose units, for example, are derived from this glyco-
sylation mechanism that typically proceeds with decarboxylation
of an NDP-glucuronic acid (Figure S1b).29,33ꢀ37 Alternatively,
sugar salvage pathways from plants,38,39 Leishmania,40 and the
thermophilic bacterium Thermus caldophilus GK2441 have been
characterized that utilize broad-specificity nucleotidylyltrans-
ferases capable of activating the pentopyranosyl phosphates of D-
xylose and L-arabinose albeit with lower efficiencies than the R-D-
hexopyranosyl phosphates of glucose and galactose. Furthermore,
a nucleotidylyltransferase from Salmonella enterica typhimurium
LT2 (Ep) has been extensively studied and shown to utilize R-D-
xylose-1-phosphate along with dozens of R-D-hexopyranosyl
In Vitro Characterization of LipP. Reactions consisted of 25 mM
potassium phosphate, pH 7.5, 2 mM 3 or analogue, and 100 nM LipP at
14457
dx.doi.org/10.1021/ja206304k |J. Am. Chem. Soc. 2011, 133, 14452–14459

Journal of the American Chemical Society
ARTICLE
30 °C, and terminated by the addition of cold TCA to 5% (w/v) or by
ultrafiltration using a Microcon YM-3. HPLC analysis of the reaction
was similar to that described for LipO.
’ ACKNOWLEDGMENT
This work is supported by the National Institutes of Health
Grant AI087849 and the Kentucky Science and Education
Foundation (S.V.L.).
The effect of pH on LipP activity was carried out in 50 mM indicated
buffer, 2.5 mM potassium phosphate, 2.5 mM 3, and 1 μM LipP for
5 min at 30 °C. The reactions were terminated with 0.1 M sodium
hydroxide, and 4 formation was determined by UV/vis spectroscopy
with Δε290 nm = 5700 Mꢀ1 at pH 13.26 To determine the kinetic
constants with respect to co-substrate phosphate, reactions were carried
out in 50 mM Tris-HCl pH 9.0 consisting of 1.5 mM 3, variable
phosphate (2.5ꢀ2500 μM), and 100 nM LipP at 30 °C under initial
velocity conditions (<10% product formation). To determine the kinetic
constants with respect to the co-substrate nucleoside, reactions were
carried out in 50 mM Tris-HCl pH 9.0 consisting of saturating phosphate
(1.5 mM) and variable nucleoside (10ꢀ4700 μM 3, 10ꢀ10 000 μM 20-
deoxy-3, 8ꢀ800 μM 2, 20ꢀ10 000 μM thymidine, or 30ꢀ3000 μM 1),
and 100 nM LipP at 30 °C under initial velocity conditions. Product
formation when utilizing thymidine was determined by UV/vis spec-
’ REFERENCES
(1) Thibodeaux, C. J.; Melancon, C. E., III; Liu, H.-w. Angew. Chem.,
Int. Ed. 2008, 47, 9814–9859.
(2) Blanchard, S.; Thorson, J. S. Curr. Opin. Chem. Biol. 2006, 10,
263–271.
(3) Kudo, F.; Fujii, T.; Kinoshita, S.; Eguchi, T. Med. Chem. 2007,
15, 4360–4368.
(4) Mikuꢀsovꢁa, K.; Huang, H.; Yagi, T.; Holsters, M.; Vereecke, D.;
D’Haeze, W.; Scherman, M. S.; Brennan, P. J.; McNeil, M. R.; Crick,
D. C. J. Bacteriol. 2005, 187, 8020–8025.
(5) Jensen, K. F. Metabolism of Nucleotides, Nucleosides and Nucleo-
bases in Microorganisms; Academic: London, U.K., 1983; pp 1ꢀ25.
(6) Winn, M.; Goss, R. J. M.; Kimura, K.; Bugg, T. D. H. Nat. Prod.
Rep. 2010, 27, 279–304.
(7) Funabashi, M.; Baba, S.; Nonaka, K.; Hosobuchi, M.; Fujita, Y.;
Shibata, T.; Van Lanen, S. G. ChemBioChem 2010, 11, 184–190.
(8) Igarashi, M.; Takahashi, Y.; Shitara, T.; Nakamura, H.; Naganawa,
H.; Miyake, T.; Akamatsu, Y. J. Antibiot. 2005, 58, 327–337.
(9) Ochi, K.; Ezaki, M.; Iwani, M.; Komori, T.; Kohsaka, M. Eur.
Patent 333177 A2, 1989.
(10) McDonald, L. A.; Barbieri, L. R.; Carter, G. T.; Lenoy, E.;
Lotvin, J.; Petersen, P. J.; Siegel, M. M.; Singh, G.; Williamson, R. T.
J. Am. Chem. Soc. 2002, 124, 10260–10261.
(11) Hirano, S.; Ichikwawa, S.; Matsuda, A. Bioorg. Med. Chem. 2008,
16, 5123–5133.
(12) Dini, C.; Didier-Laurent, S.; Drochon, N.; Feteanu, S; Guillot,
J. C.; Monti, F.; Uridat, E.; Zhang, J.; Aszodi, J. Bioorg. Med. Chem. Lett.
2002, 12, 1209–2121.
(13) Kaysser, L.; Lutsch, L.; Siebenberg, S.; Wemakor, E.; Kammerer,
B.; Gust, B. J. Biol. Chem. 2009, 284, 14987–14996.
(14) Cheng, L.; Chen, W.; Zhai, L.; Xu, D.; Huang, T.; Lin, S.; Zhou,
X.; Deng, Z. Mol. BioSyst. 2011, 7, 920–927.
troscopy with Δε
= 3700 Mꢀ1 at pH 13.26 Single substrate
300 nm
kinetics at pH 7.5 was carried out using identical conditions except with
increased LipP (730 nM). Kinetic analysis with EcUdp was carried out
with final enzyme concentrations of 100 nM at pH 9.0 and 200 nM
at pH 7.5.
In Vitro Characterization of LipM. Reactions consisted of 50 mM
potassium phosphate, pH 7.5, 5 mM MgCl2, 2 mM 3 or analogue, 5 mM
nucleotide triphosphate, 5 μM LipP, and 1 μM LipM at 30 °C, and the
reaction terminated by the addition of cold TCA to 5% (w/v) or by
ultrafiltration using a Microcon YM-3. The activity of LipM was tested with
sugar-1-phosphates generated in situ from synthetic 3, 20-deoxy-3, 2, or 7
and the co-substrate nucleotide UTP, dUTP, TTP, rCTP, dCTP, dGTP,
rGTP, dATP, or rATP. The activity of LipM was also tested with
commercial glucose-1-phosphate or glucosamine-1-phosphate with co-
substrate UTP. Following centrifugation to remove protein, the reaction
components were analyzed by HPLC using a C-18 reverse-phase
analytical column. A series of linear gradients was developed from
40 mM phosphoric acid-triethylamine pH 6.5 (C) to 20% methanol
(D) in the following manner (beginning time and ending time with
linear increase to % D): 0ꢀ8 min, 100% D; 8ꢀ18 min, 60% D; 18ꢀ
25 min, 95% D; 25ꢀ32 min, 95% D; and 32ꢀ35 min, 0% D. The flow
rate was kept constant at 1.0 mL/min, and elution was monitored at
260 nm. LCꢀMS was performed as above.
(15) Kaysser, L.; Siebenberg, S.; Kammerer, B.; Gust, B. ChemBio-
Chem. 2010, 11, 191–196.
(16) Yang, Z.; Chi, X.; Funabashi, M.; Baba, S.; Nonaka, K.; Pahari,
P.; Unrine, J.; Jacobsen, J. M.; Elliott, G. I.; Rohr, J.; Van Lanen, S. G.
J. Biol. Chem. 2011, 286, 7885–7892.
(17) Toney, M. D. PLP-dependent enzymes, chemistry of. In Wiley
Encyclopedia of Chemical Biology; Begley, T, Ed; John Wiley & Sons, Inc.:
Hoboken, NJ, 2009; Vol, 3, pp 731ꢀ735.
In Vitro Characterization of LipN. Reactions consisted of
50 mM potassium phosphate, pH 7.5, 2 mM 2, 2.8 μM LipP at 30 °C
for 2 h. LipP was removed by ultrafiltration, and 86 μL of the filtrate was
added to a solution of 5 mM MgCl2, 2 mM UTP, 1 mM 4, 12 μM LipM,
and 7 μM LipN (final volume of 100 mL) and incubated at 30 °C for the
indicated time points. HPLC analysis was performed using a TFA
mobile phase as described above.
(18) Schneider, G.; Kack, H.; Lindquvist, Y. Structure 2000, 8,
R1–R6.
(19) Rackham, E. J.; Gr€uschow, S.; Ragab, A. E.; Dickens, S.; Goss,
R. J. M. ChemBioChem 2010, 11, 1700–1709.
(20) Zhang, W.; Ostash, B.; Walsh, C. T. Proc. Natl. Acad. Sci. U.S.A.
2010, 107, 16828–16833.
(21) Zhang, W.; Ntai, I.; Bolla, M. L.; Malcolmson, S. J.; Kahne, D.;
Kelleher, N. L.; Walsh, C. T. J. Am. Chem. Soc. 2011, 133, 5240–5243.
(22) Tozzi, M. G.; Camici, M.; Mascia, L.; Sgarrella, F.; Ipata, P. L.
FEBS J. 2006, 273, 1089–1101.
(23) Temmink, O. H.; de Bruin, M.; Turksma, A. W.; Cricca, S.;
Laan, A. C.; Peters, G. J. Int. J. Biochem. Cell Biol. 2007, 39, 565–575.
(24) el Kouni, M. H.; el Kouni, M. M.; Naguib, F. N. Cancer Res.
1993, 53, 3687–3693.
’ ASSOCIATED CONTENT
S
Supporting Information. Instrumentation, chemicals,
b
synthetic procedures and spectroscopic analysis, OPA-modifica-
tion procedure, large-scale preparation of enzyme products for
spectroscopic analysis, NMR analyses, Table S1, and additional
figures. This material is available free of charge via the Internet at
http://pubs.acs.org.
(25) Razzell, W. E.; Khorana, H. G. Biochim. Biophys. Acta 1958,
28, 562–566.
(26) Leer, J. C.; Hammer-Jespersen, K.; Schwartz, M. Eur. J. Biochem.
1977, 75, 217–224.
’ AUTHOR INFORMATION
(27) Krenitsky, T. A. Biochim. Biophys. Acta 1976, 429, 352–358.
(28) Vita, A.; Huang, C. Y.; Magni, G. Arch. Biochem. Biophys. 1983,
226, 687–692.
Corresponding Author
svanlanen@uky.edu
14458
dx.doi.org/10.1021/ja206304k |J. Am. Chem. Soc. 2011, 133, 14452–14459

Journal of the American Chemical Society
ARTICLE
(29) Guyett, P.; Glushka, J.; Gu, X.; Bar-Peled, M. Carbohydr. Res.
2009, 344, 1072–1078.
(30) Hove-Jensen, B.; McSorley, F. R.; Zechel, D. L. J. Am. Chem.
Soc. 2011, 133, 3617–3624.
(31) Simons, S. S., Jr.; Johnson, D. F. J. Org. Chem. 1978, 43, 2886–
2891.
(32) Yang, J.; Xu, H.; Zhang, Y.; Bai, L.; Deng, Z.; Mahmud, T. Org.
Biomol. Chem. 2011, 9, 438–449.
(33) Bililign, T.; Shepard, E. M.; Ahlert, J.; Thorson, J. S. ChemBio-
Chem 2002, 3, 1143–1146.
(34) Hofmann, C.; Boll, R.; Heitmann, B.; Hauser, G.; D€urr, C.;
Frerich, A.; Weitnauer, G.; Glaser, S. J.; Bechthold, A. Chem. Biol. 2005,
12, 1137–1143.
(35) Bar-Peled, M.; Griffith, C. L.; Doering, T. L. Proc. Natl. Acad.
Sci. U.S.A. 2001, 98, 12003–12008.
(36) Breazeale, S. D.; Ribeiro, A. A.; Raetz, C. R. H. J. Biol. Chem.
2002, 277, 2886–2896.
(37) Van Lanen, S. G.; Oh, T.; Liu, W.; Wendt-Pienkowski, E.; Shen,
B. J. Am. Chem. Soc. 2007, 129, 13082–13094.
(38) Kotake, T.; Yamaguchi, D.; Ohzono, H.; Hojo, S.; Kaneko, S.;
Ishida, H. K.; Tsumuraya, Y. J. Biol. Chem. 2004, 279, 45728–45736.
(39) Kotake, T.; Hojo, S.; Yamaguchi, D.; Aohara, T.; Konishi, T.;
Tsumuraya, Y. Biosci. Biotechnol. Biochem. 2007, 71, 761–771.
(40) Damerow, S.; Lamerz, A. C.; Haselhorst, T.; F€uhring, J.;
Zarnovican, P.; von Itzstein, M.; Routier, F. H. J. Biol. Chem. 2010, 285,
878–887.
(41) Kim, J. S.; Koh, S.; Shin, H. J.; Lee, D. S.; Lee, S. Y. Biotechnol.
Appl. Biochem. 1999, 29, 11–17.
(42) Fu, X.; Albermann, C.; Jiang, J.; Liao, J.; Zhang, C.; Thorson,
J. S. Nat. Biotechnol. 2003, 21, 1467–1469.
(43) Jiang, J.; Biggins, J. B.; Thorson, J. S. Angew. Chem., Int. Ed.
2001, 40, 1502–1505.
(44) Moretti, R.; Thorson, J. S. J. Biol. Chem. 2007, 282, 16942–
16947.
(45) Kudo, F.; Kawabe, K.; Kuriki, H.; Eguchi, T.; Kakinuma, K.
J. Am. Chem. Soc. 2005, 127, 1711–1718.
(46) Mizanur, R. M.; Zea, C. J.; Pohl, N. L. J. Am. Chem. Soc. 2004,
126, 15993–15998.
(47) Timmons, S. C.; Mosher, R. H.; Knowles, S. A.; Jakeman, D. L.
Org. Lett. 2007, 9, 857–860.
(48) Sambrook, J.; Russell, D. W. Molecular Cloning: A Laboratory
Manual, 3rd ed.; Cold Spring Harbor Laboratory Press: Cold Spring
Harbor, NY, 2001.
(49) Kieser, T.; Bibb, M.; Buttner, M.; Chater, K. F.; Hopwood, D. A.
Practical Streptomyces Genetics; The John Innes Foundation: Norwich,
U.K., 2000.
14459
dx.doi.org/10.1021/ja206304k |J. Am. Chem. Soc. 2011, 133, 14452–14459