Angewandte
Chemie
MS (MH+ m/z 564.3086 for C32H42N3O6) and 1H NMR
spectroscopy (uridine olefinic protons at 5.62 and 7.67 ppm,
J = 11 Hz; Figure S4, Table S3).
used stearoyl coenzyme A (CoA) as an alternative. In the
preliminary experiment with 2 as an amino donor, we found
that Jaw2 required divalent Mg2+ ions, showed rather low
turnover (12% conversion over 10 min when using 10 mol%
of Jaw2), and catalyzed hydrolysis of the CoA ester (con-
densation/hydrolysis = 6.6; Figure S8). To examine the sub-
strate selectivity, time-course analysis of the Jaw2-catalysed
reaction was conducted using various acyl-CoA and amino
donors. When fatty acyl-CoA esters ranging from C8 to C18
were tested, Jaw2 accepted all of the CoA esters (with
a preference for C18; Figure S8) but not the corresponding N-
acetylcysteamine (SNAC) esters (data not shown). Compared
with the conversion of 2, Jaw2 showed poor conversion of the
dihydro analogue 2a[4b] (12% conversion). Low but reprodu-
cible activity (8% conversion) was observed in the reaction of
isobutylamine, a component of U-106305. These data indicate
that Jaw2 accepts a broad range of fatty acyl substrates, most
likely tethered to ACP, but shows relatively strict substrate
specificity for amino donors.
Next, we investigated the construction of the polycyclo-
propanated polyketide chain through heterologous expres-
sion of jaw456. For reconstitution of this minimal PKS system,
the PCR products for jaw2 (0.7 kb) and jaw456 (6.3 kb) were
cloned into pHSA81 and the integration vector pKU460,
respectively, to generate pHSAj2 and pKUj456. S. lividans
TK23 was transformed with the resulting plasmids to generate
the single transformant (jaw456) and the double transformant
(jaw2456). In the feeding experiments with a 2:1 mixture of
aminouridines 2 and 2a, 3 and 1 were detected in the double
transformant but none of jawsamycins was found in the single
transformant (Figure S9). These results confirm the role of
Jaw2 and provided experimental support for the hypothesis
that the polyketide chain is constructed by the iterative PKS
Jaw4, which contains KS-AT-DH-ACP domains, in collabo-
ration with the highly unusual trans-KR Jaw6 and with the
radical SAM enzyme Jaw5 for introducing the characteristic
cyclopropane units.
Polyunsaturated chains produced by iterative PKSs are
involved in the biosynthesis of myxochromides[14] and are also
frequently found in the biosynthesis of enediyne antibiotics[15]
and fungal highly reduced polyketides.[16] While collaboration
of a trans-acting enoyl reductase (ER) is common in fungal
iterative nonribosomal peptide synthetase (NRPS)/PKS-cat-
alyzed reactions,[16b,17] to our knowledge, trans-acting KR is
rare except for the multimodular PKS reaction reported
recently for SIA7248 biosynthesis.[18] Although cyclopropane-
containing polyketides such as ambruticin[19] and curacin[20]
are known, they are constructed in a different, non-iterative
manner. The putative cyclopropanase Jaw5, which sequen-
tially introduces cyclopropane units, is thus unique in its
modification of the polyketide chain.
Although C-alkylation of a polyketide chain during chain
extention is relatively rare, the methylation domains in
modular PKS[16a] and b-branching enzymes[21] are known to
introduce extra carbon units during the chain construction of
certain polyketides. The introduction of C1 units by radical
SAM enzymes is frequently involved in metabolism, such as
C-methylation of RNA and other secondary metabolites.[9]
However, there has been no report of a polyketide chain
Jaw1 consists of two independent reductase domains. A
homology search revealed that the N-terminal domain is
closely related to a 7a-hydroxysteroid dehydrogenase (250
aa; 35/53% identity/similarity) and the C–terminus bears
similarity to a deazaflavin-dependent nitroreductase (160 aa;
50/60% identity/similarity). In a biosynthetic study of the
nucleoside antibiotic napsamycin, npsU was identified as the
reductase gene that is responsible for the reduction of the
uracil moiety but the exact timing has not been explored.[10]
To determine the true substrate of Jaw1, which differs in
sequence from NpsU, feeding experiments of putative
precursors into deletion mutants were employed. When 5’-
amino-5’-deoxyuridine (2) was fed to the Djaw8 mutant,
production of 1 was detected (Scheme 1 and Figure S5).
Meanwhile, feeding 3 to the Djaw2 mutant resulted the
transformation of 3 into 1. These results were further
confirmed by the bioconversion of 3 into 1 when using a S.
lividans TK23 strain in which jaw1 was heterologously
expressed from the pHSA81 vector. Based on these results,
we speculated that the reduction may be catalyzed in the C-
terminal domain with coenzyme F420[11] and that the N-
terminal domain possibly provides the substrate binding
pocket since bioinfomatic analysis suggests the loss of the
essential catalytic residues in the N-terminal domain (Fig-
ure S6).[12] These results indicate that reduction of the uracil
moiety occurs at the last step of biosynthesis.
To validate the functions of individual genes, Jaw7, Jaw8,
and Jaw2 with appropriate tags at the N–termini were
overexpressed and purified from Escherichia coli (Figures S7
and S8). Since Jaw7 and Jaw8 showed high homology to LipL
and LipO, respectively, these enzymes could catalyze the
conversion of UMP into 2.[7,8] According to the LipL
catalyzed reaction,[7] enzymatic reaction of the recombinant
Jaw7 with UMP was carried out in the presence of FeCl2, a-
ketoglutarate, and l-ascorbate. HPLC analysis of the reaction
mixtures showed a new peak that was identical to synthetic
aldehyde 4[7] (Figure S7). Aldehyde 4 was then incubated with
the LipO[8] homologue Jaw8 in the presence of PLP and
several amino acids. Production of 2 was confirmed by HPLC
analysis through comparison with a sample of 2[7] and l-
methionine was found to be the best amino donor (Figure S7).
Together with the proposed the function of Jaw1, these results
showed that Jaw7 and Jaw8 are responsible for 5’-amino-5’-
deoxyuridine biosynthesis.
Bioinformatic analysis of the jawsamycin gene cluster
revealed that there is no common gene, such as a thioesterase,
for polyketide chain release, thus suggesting that the N-
acetyltransferase homologue Jaw2 directly catalyzes the
condensation of the acyl carrier protein (ACP)-bound poly-
ketide chain with 2. Jaw2 showed weak homology to the
GCN5 family acetyltransferase TunC, which is proposed to
introduce various C7–C12 acyl chains to the core scaffold of
tunicamycins.[13] To test this hypothesis, enzymatic reactions
of the recombinant Jaw2 were employed with various
substrate analogues. As a putative genuine substrate, ACP-
bound cyclopropanated polyketide was not available so we
Angew. Chem. Int. Ed. 2014, 53, 5423 –5426
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5425