Biosynthesis of epothilone intermediates with alternate starter units: Engineering polyketide-nonribosomal interfaces

Article abstract of DOI:10.1002/anie.200352077

The early steps of epothilone biosynthesis were reprogrammed in vitro by incorporating enzyme subunits (e.g. EntB, EntE) from other biosynthetic pathways into the epothilone (Epo) assembly line (see scheme). Recognition sequences identified at the C termini of the epothilone biosynthetic proteins enabled heterologous enzyme transfer when fused to noncognate proteins, as a step toward the reengineering of biosythetic pathways to produce natural product analogues in vivo.

Full text of DOI:10.1002/anie.200352077

Angewandte
Chemie
Reengineering Biosynthetic Pathways
Biosynthesis of Epothilone Intermediates with
Alternate Starter Units: Engineering
Polyketide–Nonribosomal Interfaces**
Sarah E. OꢀConnor, Christopher T. Walsh,* and Fei Liu
[
1]
The epothilones, a class of antitumor macrolides, are
nonribosomal-peptide–polyketide natural products that are
constructed by mixed enzyme assembly lines composed of
nonribosomal-peptide synthetases (NRPS) and polyketide
[
2]
synthases (PKS). In the early steps of the epothilone
+
+
[
*] Prof. Dr. C. T. Walsh, Dr. S. E. O’Connor, Dr. F. Liu
Department of Biological Chemistry and Molecular Pharmacology
Harvard Medical School
Boston, MA 02215 (USA)
Fax: (+1)617-432-0438
E-mail: christopher_walsh@hms.harvard.edu
+
[
] These authors contributed equally to this work.
[
**] Funding for this work was provided by the NIH (GM20011, C.T.W.).
F.L. is supported by a NIH Postdoctoral Fellowship (GM66456) and
S.E.O. by an Irving Sigal Postdoctoral Fellowship (American
Chemical Society).
Angew. Chem. Int. Ed. 2003, 42, 3917 –3921
DOI: 10.1002/anie.200352077
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3917

Communications
biosynthetic pathway, the incipient methylthiazolyl group is
formed by acetyl transfer from the PKS subunit EpoA to the
cysteine residue of the NRPS subunit EpoB. Subsequent
cyclization, dehydration, and oxidation by the cyclization
NRPS EpoB facilitates substrate transfer to EpoC (Fig-
ure 2a).
[
3]
When the acyl carrier protein from RapC of the rapamy-
3
cin biosynthetic pathway was loaded with [ H]acetyl-CoA
(
Cy) and oxidase (Ox) domains of EpoB results in the
(see Experimental Section) and incubated with cysteinyl-S-
EpoB and methylmalonyl-S-EpoC, no substrate transfer
occurred (Figure 3a, top). However, when the 39-residue
EpoA recognition sequence was fused in place of the
C terminus of RapC to yield RapC(EpoA) (Figure 2b) and
this hybrid protein was used, formation of compound 1 was
observed (Figure 3a, bottom).
methylthiazolyl species, which is then transferred from EpoB
to the downstream PKS acceptor subunit EpoC by the
ketosynthase (KS) domain of EpoC (Figure 1a). As both
EpoB and EpoC have demonstrated tolerance for unnatural
upstream acyl substrates in vitro, it was hypothesized that
proteins from other biosynthetic pathways with novel sub-
strate specificities could be used in place of EpoA and EpoB
within the epothilone assembly line (Figure 1b).
[
4]
[
5]
The free-standing aryl carrier protein EntB from the
[
7]
enterobactin biosynthetic pathway, primed with the meth-
ylthiazolyl-S-pantetheinyl moiety derived from methylthi-
azolyl-CoA, produced compound 1 when incubated with
methylmalonyl-S-EpoC (Figure 3b, top). However, after
fusion of the 32-residue EpoB linker to the enterobactin
protein (EntB(EpoB), Figure 2b), the formation of product 1
was improved (Figure 3b, bottom). When Ybt-PCP2, a
peptidyl carrier protein from the yersiniabactin biosynthetic
This study demonstrates that EpoA or EpoB can be
[
6]
replaced in vitro with proteins from the rapamycin, enter-
[
7]
[8]
obactin, or yersiniabactin biosynthetic pathways. Fusion of
a protein-recognition element to the C terminus of the
noncognate protein is critical for substrate transfer, as
observed in studies with PKS enzymes. Although similar
recognition domains have been proposed for mixed PKS/
NRPS assembly lines, this hypothesis has not been vali-
dated experimentally. A putative recognition sequence at the
C terminus of the PKS EpoA appears to facilitate substrate
transfer to EpoB, whereas a sequence at the C terminus of the
[
9]
[
10]
[8]
pathway, was primed with the methylthiazolyl-S-acyl group
and incubated with methylmalonyl-S-EpoC, a small amount
of product formation was again observed (Figure 4). How-
ever, replacement of the C-terminal sequence of Ybt-PCP2 by
Figure 1. a) Early steps in the biosynthesis of epothilone. The EpoA, EpoB, and EpoC proteins are depicted with their catalytic modules. b) Integra-
tion of noncognate proteins into the epothilone assembly line. Use of noncognate proteins with different substrates. (X=methyl and Y=cysteine
to make epothilone fragment 1; alternatively the X-containing substrate was omitted from the enzymatic reaction, and the substrates Y=methyl-
thiazole or methylpyridine were used to make epothilone fragments 1 or 2, respectively.)
3
918
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2003, 42, 3917 –3921

Angewandte
Chemie
Figure 2. a) Proposed C-terminal recognition sequences of EpoA (in blue) and EpoB(in red). b) Amino acid sequences of the C termini of the pro-
teins used in this study. Underlined sequences in RapC and YbtPCP2 were removed and replaced with the EpoA or EpoBrecognition sequence.
Italicized letters indicate the serine residue that is pantetheinylated.
that of the EpoB recognition sequence (Ybt-PCP2(EpoB),
improve product formation when this noncognate substrate
is used, thus suggesting that improvement of the protein–
protein interaction by the recognition sequence is critical
when unusual substrates are incorporated into the enzymatic
assembly line.
[
11]
Figure 2b) greatly improved product yield (Figure 4). The
use of the protein Ybt-PCP2(NS), which lacks the C-terminal
residues of the wild-type carrier protein, also led to dimin-
ished product formation (decreased by 60% relative when
Ybt-PCP2(EpoB) was used). These studies indicate that the
recognition sequences that assist protein recognition in the
This study demonstrates that proteins from rapamycin,
yersiniabactin, and enterobactin biosynthetic pathways can
interact productively with the epothilone assembly line, which
suggests that short recognition sequences in hybrid PKS/
NRPS systems may be fused to multiple proteins. Thus, the
“linker hypothesis”, which has enabled the mixing and
[
9a]
PKS pathways are also observed in both PKS/NRPS and
NRPS/PKS mixed interfaces. A preliminary study revealed
that product formation in the presence of methylthiazolyl-S-
EntB(EpoB) and methylmalonyl-S-EpoC is much slower
than the rate of product formation that is observed with the
cognate EpoB protein (Figure 3c). This suggests that the
engineered partner protein, although viable, is not as efficient
as the wild-type EpoB donor. Further optimization of the
recognition sequence may lead to improved acyl-group
transfer.
[
9a]
matching of proteins within PKS systems,
can also be
used to mix PKS and NRPS systems and enable acyl transfer
and subsequent reactions such as cyclization, reduction, or
dehydration to build novel assembly-line intermediates. The
rates for product formation in the hybrid system are
decreased relative to that observed in the natural system,
thus suggesting that the terminal recognition sequence is not
the only element required for productive acyl transfer.
Consistent with this observation, studies with the erythromy-
cin biosynthetic pathway suggest that the entire protein may
The use of noncognate proteins to deliver tethered acyl
groups to NRPS or PKS assembly lines for the synthesis of
unnatural products is an important goal. Therefore, we
[
7]
examined the ability of EntE, the partner protein of EntB,
to load aromatic acyl substrates onto the phosphopante-
theinyl group of EntB(EpoB). Although the interaction of
EntE with EntB(EpoB) is adversely affected by the addition
[
9d]
contribute to the protein–protein interaction. Nevertheless,
addition of the C-terminal recognition sequence greatly
improved substrate transfer in the three examples described
herein. Although a coiled-coil protein interaction has been
1
4
of the linker sequence, a substantial extent of [ C]salicylate-
group loading (57%) was noted for the modified carrier
protein (compared to 90% for the wild type). EntE can also
activate and load the heteroaromatic substrate 4-methyl-2-
pyridine carboxylic acid (picolinic acid), though at a signifi-
[
9e]
proposed for the PKS recognition elements, it is not known
whether PKS/NRPS sequences interact in a similar manner.
More detailed studies designed to elucidate the mechanism by
which these recognition elements foster protein–protein
interactions are currently underway.
ꢀ
1
cantly decreased rate (0.33 min ) to that of its activation and
loading of the natural substrate 2,3-dihydroxybenzoate
ꢀ
1
[7a]
(
330 min ).
As a methylpyridine epothilone derivative
[
12]
has been shown to exhibit enhanced cytotoxic activity, this
substrate provides an interesting test case for further study.
Acyl transfer, reduction, and dehydration on EpoC could be
demonstrated with this novel substrate by incubating EntE,
picolinic acid, EntB(EpoB), and EpoC, in the presence of
Experimental Section
Carrier proteins were amplified by polymerase chain reaction (PCR)
from existing constructs or from genomic DNA and ligated into
pET28b (Novagen) to create constructs that contained an N-terminal
His tag. The carrier protein of RapC consists of 151 amino acids
1
4
[
C]methylmalonyl-CoA and NADPH (Figure 5). The rec-
(residues 6110–6260 of RapC); Ybt-PCP2 consists of the amino acids
ognition sequence from EpoB appears to dramatically
1933–2041 of HMWP2. Carrier proteins with a linker sequence were
Angew. Chem. Int. Ed. 2003, 42, 3917 –3921
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3919

Communications
Figure 4. Comparison of the formation of compound 1 by the interfa-
1
4
ces formed between [ C]methylmalonyl-S-EpoC and methyl-thiazolyl-S-
Ybt-PCP2, Ybt-PCP2(EpoB), and Ybt-PCP2(NS) over time.
Figure 5. Radio-HPLC trace that indicates the product 2 formed upon
1
4
incubation of [ C]methylmalonyl-S-EpoC and picolinyl-S-EntBor picol-
inyl-S-EntB(EpoB), against the UV-HPLC trace (254 nm) of compound
2
as a chemical standard.
[
13]
constructed by splicing-overlap-extension PCR. Carrier proteins
were expressed in the apo (nonpantetheinylated) form in BL21(DE3)
E. coli cells under control of a T7 promoter at 258C and purified by
nickel chromatography. Apo carrier proteins were modified by
incubation with the broad-specificity phosphopantetheinyl transfer-
[
14]
ase Sfp and acetyl-CoA or methylthiazolyl-CoA. The Ybt proteins
exhibited a loading capacity of 65%, as judged by quantitation of
covalent modification of the protein with the radiolabeled coenzyme
3
[
H]acetyl-CoA. EntB(EpoB) and RapC(EpoA) both exhibited
Figure 3. a) Radio-HPLC traces that indicate the formation of com-
pound 1 from the incubation of cysteinyl-S-EpoB, methylmalonyl-S-
approximately 50% modification, whereas the wild-type counterparts
RapC and EntB showed greater than 90% priming. EpoB and EpoC
3
3
EpoC, and either 1) [ H]acetyl-S-RapC or 2) [ H]acetyl-S-RapC(EpoB).
b) Radio-HPLC traces that indicate the formation of compound 1 from
[
3,4]
were expressed and purified as reported previously.
In a repre-
1
4
sentative assay, apo-EntB(EpoB) (5 mm) in a total volume of 100 mL
was incubated at pH 7 with Sfp (0.5 mm) and methylthiazolyl-CoA
the incubation of [ C]methylmalonyl-S-EpoC and 1) methylthiazolyl-S-
EntBor 2) methylthiazolyl- S-EntB(EpoB). c) Comparison of product for-
mation from wild-type epothilone biosynthetic enzymes (acetyl-S-EpoA-
(
200 mm) in the presence of MgCl (5 mm). EpoC (5 mm), NADPH
2
3
5
14
carrier protein, [ S]cysteine-S-EpoB, and methylmalonyl-S-EpoC) with
(3 mm), and [ C]methylmalonyl-CoA (10 mm) (total reaction volume
65 mL) were then added, and the reaction was left for 20 min at 258C.
The reaction was quenched by the addition of trichloroacetic acid
(10%), and the product was hydrolyzed from the precipitated protein
product formation from enterobactin and epothilone biosynthetic path-
1
4
ways (methyl-thiazolyl-S-EntB(Epo) and [ C]methylmalonyl-S-EpoC)
over time. ACP=acyl carrier protein.
3
920
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2003, 42, 3917 –3921

Angewandte
Chemie
by heating with KOH (0.5m, 608C) for 5 min. The methylthiazole
methylacrylic acid 1 and picolinic methylacrylic acid 2 standards were
made as previously reported and used to verify product formation in
[15] Characterization of 2. Reversed-phase HPLC (MeCN (0–25%)
in TFA (0.1%) over 25 min, C18 Vydac column): retention
time = 14.6 min; MALDI-MS: calcd: 178.08, found: 178.06;
[
4,15]
1
the reaction.
Reversed-phase HPLC (C18 Vydac column) with a
H NMR (200.055 MHz, CDCl ): d = 8.63 (1H, d, J = 4.81 Hz;
3
UV detector and an online radioisotope detector was used for the
assays. (1: MeCN (0–70%) in aqueous TFA (0.1%), 25 min; 2: MeCN
Ar), 7.76 (1H, d, J = 1.40 Hz; Ar), 7.09 (1H, d, J = 4.80 Hz; Ar),
2.40 (3H, s; CH -Ar), 2.30 (3H, d, J = 1.60 Hz; CH¼C(CH )),
3
3
1
3
(0–25%) in aqueous TFA (0.1%), 25 min.) The formation of 1 was
2.16 ppm (1H, d, J = 0.80 Hz; CH¼C); C NMR (50.31 MHz,
also assayed by radio-TLC (silica; 90:5:5 chloroform/ethanol/acetic
acid). Rates for EntE activation were determined by using a PPi
CDCl ): d = 172.6, 154.5, 148.9, 148.5, 137.3, 132.7, 126.8, 124.1,
3
21.39, 14.35 ppm.
(
inorganic pyrophosphatase) exchange assay as previously descri-
[
7a]
bed.
Quantitation of the loading onto EntB by EntE was
1
4
performed by loading with [ C]salicylate.
Received: June 6, 2003 [Z52077]
Keywords: antitumor agents · biosynthesis · enzyme catalysis ·
.
polyketides · protein modifications
[
1] D. M. Bollag, P. A. McQueney, J. Zhu, O. Hensens, L. Koupal, J.
Liesch, M. Goetz, E. Lazarides, C. M. Woods, Cancer Res. 1995,
55, 2325 – 2333.
[
2] a) L. Tang, S. Shah, L. Chung, J. Carney, L. Katz, C. Khosla, B.
Julien, Science 2000, 287, 640 – 642; b) I. Molnar, T. Schupp, M.
Ono, R. E. Zirkle, M. , Milnamow, B. Nowak-Thompson, N.
Engel, C. Toupet, A. Stratmann, D. D. Cyr, J. Gorlach, J. M.
Mayo, A. Hu, S. Goff, J. Schmid, J. M. Ligon, Chem. Biol. 2000, 7,
9
7 – 109.
3] H. Chen, S. O'Connor, D. E. Cane, C. T. Walsh, Chem. Biol.
001, 8, 899 – 912.
4] S. E. O'Connor, H. Chen, C. T. Walsh, Biochemistry 2002, 41,
685 – 5694.
5] T. L. Schneider, C. T. Walsh, S. E. O'Connor, J. Am. Chem. Soc.
002, 124, 11272 – 11273.
6] A. Konig, T. Schwecke, I. Molnar, G. A. Bohm, P. A. Lowden, J.
Staunton, P. F. Leadley, Eur. J. Biochem. 1997, 247, 526 – 534.
7] a) A. M. Gehring, K. A. Bradley, C. T. Walsh, Biochemistry
[
[
[
[
[
2
5
2
1997, 36, 8495 – 8503; b) F. Rusnak, W. S. Faraci, C. T. Walsh,
Biochemistry 1989, 28, 6827 – 6835.
[
8] a) D. A. Miller, L. Luo, N. Hillson, T. A. Keating, C. T. Walsh,
Chem. Biol. 2002, 9, 333 – 344; b) H. Drechsel, H. Stephan, R.
Lotz, H. Haag, H. Zaehner, K. Hantke, G. Jung, Liebigs Ann.
1995, 1727 – 1733; c) A. M. Gehring, E. DeMoll, J. D. Fether-
ston, I. Mori, G. F. Mayhew, F. R. Blattner, C. T. Walsh, R. D.
Perry, Chem. Biol. 1998, 5, 573 – 586.
[
9] a) R. S. Gokhale, S. Y. Tsuji, D. E. Cane, C. Khosla, Science
1999, 284, 482 – 485; b) S. Y. Tsuji, D. E. Cane, C. Khosla,
Biochemistry 2001, 40, 2326 – 2331; c) N. Wu, S. Y. Tsuji, D. E.
Cane, C. Khosla, J. Am. Chem. Soc. 2001, 123, 6465 – 6474; d) N.
Wu, D. E. Cane, C. Khosla, Biochemistry 2002, 41, 5056 – 5066;
e) P. Kumar, Q. Li, D. E. Cane, C. Khosla, J. Am. Chem. Soc.
2003, 125, 4097 – 4102.
[
[
10] L. Du, C. Sanchez, B. Shen, Metab. Eng. 2001, 3, 78 – 95.
11] A previously reported definition of a PCP domain was used in
this study to separate the linker region (the sequence after the
fourth helix) from the minimal PCP domain: T. Weber, R.
Baumgartner, C. Renner, M. A. Marahiel, T. A. Holak, Structure
2000, 8, 407 – 418.
[
12] K. C. Nicolaou, R. Scarpelli, B. Bollbuck, B. Werschkun,
M. M. A. Pereira, M. Wartmann, K. H, Altmann, D. Zahar-
evitz, R. Gussio, P. Giannakakou,Chem. Biol. 2000, 7, 593 – 599.
13] R. M. Horton, Z. Cai, S. N. Ho, L. R. Pease, BioTechniques 1990,
[
[
8, 528.
14] a) L. E. N. Quadri, P. H. Weinreb, M. Lei, M. M. Nakano, P.
Zuber, C. T. Walsh, Biochemistry 1998, 37, 1585 – 1595; b) P. J.
Belshaw, C. T. Walsh, T. Stachelhaus, Science 1999, 284, 486 –
489.
Angew. Chem. Int. Ed. 2003, 42, 3917 –3921
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3921