The thioesterase domain from the pimaricin and erythromycin biosynthetic pathways can catalyze hydrolysis of simple thioester substrates

Article abstract of DOI:10.1016/j.bmcl.2007.03.060

The recombinant polyketide synthase thioesterase domains from the pimaricin and 6-deoxyerythronolide B biosynthetic pathways catalyze hydrolysis of a number of simple N-acetylcysteamine thioester derivatives. This study demonstrates that thioesterases are not highly substrate selective in formation of the acyl-enzyme intermediate, in contrast to non-ribosomal peptide synthase thioesterase domains that show very high specificity for substrate loading. This observation has important implications for the engineering of biosynthetic pathways to produce polyketide products.

Full text of DOI:10.1016/j.bmcl.2007.03.060

Bioorganic & Medicinal Chemistry Letters 17 (2007) 3034–3037
The thioesterase domain from the pimaricin and erythromycin
biosynthetic pathways can catalyze hydrolysis
of simple thioester substrates
Krishna K. Sharma and Christopher N. Boddy*
Department of Chemistry, Syracuse University, Syracuse, NY 13244-4100, USA
Received 20 January 2007; revised 16 March 2007; accepted 19 March 2007
Available online 23 March 2007
Abstract—The recombinant polyketide synthase thioesterase domains from the pimaricin and 6-deoxyerythronolide B biosynthetic
pathways catalyze hydrolysis of a number of simple N-acetylcysteamine thioester derivatives. This study demonstrates that thioes-
terases are not highly substrate selective in formation of the acyl-enzyme intermediate, in contrast to non-ribosomal peptide syn-
thase thioesterase domains that show very high specificity for substrate loading. This observation has important implications for
the engineering of biosynthetic pathways to produce polyketide products.
ꢀ 2007 Elsevier Ltd. All rights reserved.
Macrocyclic polyketides include numerous important
pharmaceutical agents such as the anticancer agent epo-
thilone and the antibiotic rifamycin. These natural prod-
ucts are produced by modular polyketide biosynthesis.1
Substantial effort has been undertaken to understand this
biosynthetic machinery with the hope that the enzymatic
pathways can be engineered to produce new, non-natural
polyketides for applications in drug discovery.2–4 A major
challenge in accomplishing this goal is the intrinsic sub-
strate specificity of the thioesterase catalytic domain. In
this study, we probe the substrate specificity of the thioes-
terase domains from the pimaricin5 (3) and 6-deoxy-
erythronolide (1) biosynthetic pathways.
Linear acyl chains are macrocyclized by thioesterase do-
mains in a two-step process.6 The first step is acylation
of the thioesterase at an active-site serine residue to gen-
erate an acyl-enzyme intermediate. The second step is a
nucleophilic attack by an intramolecular nucleophile
resulting in cyclization and hydrolysis of the acyl-en-
zyme intermediate. In this study, we have focused on
understanding the specificity of substrate loading, which
is the first step of the mechanism.
The loading of a limited number of substrates onto two
thioesterase domains has been investigated.7–9 However,
the substrates examined did not entirely represent the
possible diversity observed in polyketide biosynthesis.
The two thioesterase domains studied, from the 6-
deoxyerythronolide and pikromycin biosynthetic path-
ways, produce very similar 14-membered macrolactones
(1 and 2, Fig. 1). Investigating thioesterases that pro-
duce structurally diverse macrocycles will give better in-
sight into thioesterase substrate specificity and help
identify residues in the substrate-binding pocket respon-
sible for specificity. We chose to examine the pimaricin
(3) thioesterase domain since its native substrate differed
in ring size, substitution pattern, and acyl-enzyme inter-
mediate electronics from 1 and 2.
Thioesterases are the final catalytic domains of modular
polyketide synthases. These enzymes cyclize the com-
pleted linear polyketide and cleave the polyketide, which
has been covalently linked to the enzyme, from the bio-
synthetic machinery. This allows turnover of the enzyme
system. Substrates that are not processed by the thioes-
terase domain remain attached to the enzyme, inhibiting
product formation.
The goals of this study included identifying similarities
in substrate specificity between multiple thioesterase do-
mains, correlating substrate specificity with substitution
and stereochemistry of the native substrate, and deter-
Keywords: Antibiotics; Biosynthesis; Polyketides; Thioesterase; Hydro-
lysis; Acyl-enzyme intermediate.
*
Corresponding author. Tel.: +1 315 443 5803; fax: +1 315 443
4070; e-mail: cnboddy@syr.edu
0960-894X/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.03.060

K. K. Sharma, C. N. Boddy / Bioorg. Med. Chem. Lett. 17 (2007) 3034–3037
3035
mining if there is a correlation between enzyme activity
and substrate electronics. Our results show that the
loading step catalyzed by polyketide thioesterase do-
mains is not inherently highly substrate selective, but
that thioesterase domains do preferentially recognize
more native-like substrates. Additionally, we found that
electronics of the thioester substrate play a significant
role in enzyme activity.
try at the a and b positions. The chemical diversity at
these centers is representative of the diversity generated
during polyketide biosynthesis (b-keto, b-hydroxy, a,b-
unsaturated and saturated thioesters), and represents
the diversity seen among native polyketide substrates.
All compounds were synthesized using standard tech-
niques and were purified to homogeneity, as measured
1
by H NMR spectroscopy.10–15
A library of five substrates was selected to probe the
selectivity of the thioesterase domains. The substrates
chosen, 6, 7, 10, 12, and 14 (Fig. 2), had no accessible
intramolecular nucleophiles and therefore could not un-
dergo thioesterase-catalyzed cyclization. The thioester-
ase domains therefore could only catalyze hydrolysis
of the activated esters as shown in Figure 2. The sub-
strates differed in the oxidation state and stereochemis-
The excised recombinant thioesterase domains from the
pimaricin (pim TE) and 6-deoxyerythronolide B polyke-
tide synthases (debs TE) were over-expressed in Esche-
richia coli and isolated in high purity by affinity
chromatography.16 Steady state kinetic parameters were
determined for the thioesterase-catalyzed hydrolysis of
substrates 6, 7, 10, 12, and 14 by quantifying the produc-
tion of free thiol using Ellman’s reagent.7,8
Because of the limited solubility of the substrates and
the typically high Michaelis constants (KM),7,8 it was
not possible to determine the turnover rate (kcat) and
the KM independently. In most cases, the specificity con-
stant (kcat/KM) was determined, as this could be easily
measured at substrate concentrations below the dissoci-
ation constant where all substrates were sufficiently sol-
uble in the reaction buffer. In the case of substrate 10
with the 6-deoxyerythronolide thioesterase, the turnover
rate and dissociation constants were known.7,8 Using
our assay we determined kcat = 0.77 0.06 minÀ1 and
KM = 3.6 0.75 mM and our experimental values
agreed within the error with those previously reported
(kcat = 0.68 0.16 minÀ1 and KM = 9.4 4.5 mM).8
The specificity constants for the remaining substrates
with both the 6-deoxyerythronolide and pimaricin thio-
esterase domains were then determined (Table 1).
O
O
OH
O
OH
OH
O
OH
O
O
O
2 narbonolide
1 6-deoxyerythronolide B
OH
H
O
CO2H
O
O
HO
HO
OH
O
3 pimaricin aglycon
Both the 6-deoxyerythronolide B and pimaricin thioester-
ases accelerated the hydrolysis reactions. The background
hydrolytic rate for 7, 10, and 14 was 4.5 0.9 ·
Figure 1. Structures of 6-deoxyerythronolide B, narbonolide (the
precursor to pikromycin), and the pimaricin aglycon. The bonds
generated by the thioesterase domains are highlighted.
Bu2BOTf,
DIPEA,
Propanal
1. PCC
2. HSNAC,
LiHMDS
O
O
O
O
O
OH O
SNAC
Et
N
O
Et
N
O
(85 %)
(55 %)
Bn
4
Bn
6
5
1. LiOH, H2O2
2. EDC, HSNAC
(63 %)
OH
O
O
SNAC
SNAC
7
Bu2BOTf,
DIPEA,
1. LiOH, H2O2
2. EDC,
HSNAC
(47 %)
O
O
O
OH O
OH
Propanal
Et
N
O
Et
N
O
(80 %)
Bn
8
Bn
9
10
O
R
O
R
EDC,
HSNAC
(100 %)
EDC,
HSNAC
(99 %)
11 R = OH
12 R = SNAC
13 R = OH
14 R = SNAC
Figure 2. Synthesis of substrates used to probe the substrate specificity of the thioesterase domains.

3036
K. K. Sharma, C. N. Boddy / Bioorg. Med. Chem. Lett. 17 (2007) 3034–3037
Table 1. Specificity constants for the 6-deoxyerythronolide B (debs)
and pimaricin (pim) thioesterase domains with substrates 6, 7, 10, 12,
and 14
carbons of the substrate. Substrate 6 however has a
slightly greater specificity constant than 10. This value
may be partially influenced by the 3-keto group in sub-
strate 6, which makes the thioester highly activated and
very prone to hydrolysis. Thus, the rapid hydrolysis of 6
does not represent preferred substrate recognition by the
enzyme but also reflects an increased lability of the thi-
oester bond.
Substrate
debs TE kcat/KM
(MÀ1 sÀ1
pim TE kcat/KM
(MÀ1 sÀ1
)
)
6
7
4.2 0.4
0.47 0.06
3.6 1.4
nd
nd
0.04 0.01
0.02 0.04
1.3 0.1
10
12
14
0.17 0.01
0.24 0.02
Our hypothesis that the thioesterase prefers substrates
that have native-like substitution and stereochemistry
at the a and b carbons is supported by the data collected
for the pimaricin thioesterase domain. Substrate 12,
which best represents the a,b unsaturated thioester
found in the native pimaricin substrate, is hydrolyzed
with the highest specificity constant in this study.
10À6 minÀ1. This hydrolytic rate is comparable to rates
documented for other N-acetylcysteamine thioesters.17
Compounds 6 and 12 underwent faster background
hydrolysis with rates of k = 2.4 0.4 · 10À5 minÀ1 and
k = 1.2 0.2 · 10À5 minÀ1, respectively. This is due to
the conjugation between the thioesters and the a,b p sys-
tem, which lowers LUMO energy, increasing hydrolysis
rates. Assuming Michaelis constants in the range of 5–
50 mM, which is generally seen for polyketide synthase
thioesterases, a rate enhancement of 104–105 orders of
magnitude is seen for both enzymes. This is comparable
to the hydrolytic rate enhancement seen for epothilone
thioesterase domain.17
We propose that it may be a common feature of polyke-
tide thioesterase domains that they are able to load most
thioesters to generate acyl-enzyme intermediates. This
observation is in contrast to non-ribosomal synthetase
(NRPS) thioesterase domains, which exhibit high levels
of substrate selectivity for the loading step of thioester-
ase-catalyzed chemistry.18–21 NRPS thioesterase do-
mains load linear peptides to generate peptidyl-enzyme
intermediates similar to polyketide synthase thioesterase
domains. Once the peptidyl-enzyme intermediate is
formed, the NRPS thioesterase domains can catalyze
hydrolysis and macrocyclization of a wide assortment
of synthetic peptides.22
Our kinetic studies on the pimaricin and 6-deoxyerythr-
onolide B thioesterase domains provided insight into the
substrate specificity for the selected substrates. The
hydrolysis of substrates 6, 7, 10, and 14 by the 6-deoxy-
erythronolide B thioesterase and substrates 7, 10, 12,
and 14 by the pimaricin thioesterase domain shows that
the enzymes display limited substrate specificity. That is,
regardless of the functional groups and stereochemistry
present on the substrates the thioesterase can generate
the requisite acyl-enzyme intermediate. The only sub-
strates that were not hydrolyzed by the two thioesterase
domains were compound 12 with the 6-deoxyerythrono-
lide B thioesterase domain and compound 6 with the
pimaricin thioesterase domain. Due to the higher rate
of background hydrolysis of these two highly activated
thioester substrates, it is difficult to detect levels of thio-
esterase-catalyzed hydrolysis that were slower than the
background hydrolysis rate. It is therefore possible that
12 and 6 are also being hydrolyzed by the 6-deoxy-
erythronolide B and pimaricin thioesterases, respec-
tively, but hydrolysis was below the detection limit of
our assay. Development of an assay capable of discern-
ing hydrolytic rates below background hydrolysis is nec-
essary to confirm this hypothesis.
The formation of the peptidyl-enzyme intermediate is
highly dependent upon the substitution of the thioester
activated amino acid of the linear peptide chain.
Replacement of the C-terminal amino acid residue in
the substrate for the surfactin thioesterase leads to com-
plete loss of hydrolytic and macrocyclization activity.19
Similarly, substitution of the C-terminal and penulti-
mate amino acid residues of the substrates for the tyro-
cidine20 and fengycin18 thioesterases leads to substantial
degradation of hydrolysis and macrocyclization rates.
These results suggest that recognition of C-terminal
functional groups is important in NRPS peptidyl-en-
zyme intermediate formation.
Our data suggest that polyketide synthase thioesterases
are less restrictive in substrate selection and loading.
This low substrate specificity has important implications
for the engineering of polyketide biosynthetic pathways
to produce non-natural products. Based on this study it
is highly likely that linear acyl chains produced by an
engineered polyketide synthase pathway will be loaded
onto the thioesterase domain forming the acyl-enzyme
intermediate. These intermediates are known to undergo
hydrolysis, especially if macrocyclization cannot take
place.23 This implies that linear free acids of the desired
non-natural compounds can be produced from engi-
neered systems without worrying about engineering the
specificity of the thioesterase domains. Additionally it
indicates that the brunt of the effort in thioesterase pro-
tein engineering should be aimed at understanding and
modulating the substrate specificity and selectivity of
the macrocyclization step.
While it appears from the above data that most sub-
strates can be loaded onto polyketide synthase thioester-
ase domains, there is a hierarchy of preferred substrates.
In examining the specificity constants from 6-deoxy-
erythronolide B thioesterase-catalyzed hydrolysis, sub-
strates 6 and 10 are hydrolyzed with the greatest
efficiency. Substrate 10 has the same substitution pattern
and stereochemistry as the native substrate for the 6-
deoxyerythronolide B thioesterase domain. We there-
fore postulate that the thioesterase prefers the native
substitution pattern and stereochemistry at the a and b

K. K. Sharma, C. N. Boddy / Bioorg. Med. Chem. Lett. 17 (2007) 3034–3037
3037
10. Chan, P. C.-M.; Chong, J. M.; Kousha, K. Tetrahedron
1994, 50, 2703.
Acknowledgment
11. Evans, D. A.; Bartroli, J.; Shih, T. L. J. Am. Chem. Soc.
1981, 103, 2127.
12. Evans, D. A.; Ng, H. P.; Clark, J. S.; Rieger, D. L.
Tetrahedron 1992, 48, 2127.
13. Gilbert, I. H.; Ginty, M.; O’Neill, J. A.; Simpson, T. J.;
Staunton, J.; Willis, C. L. Bioorg. Med. Chem. Lett. 1995,
15, 1587.
This research was supported by Syracuse University. We
thank C. Khosla (Stanford University) for the plasmid
encoding the 6-deoxyerythronolide
domain.
B thioesterase
14. Oikawa, Y.; Sugano, K.; Yonemitsu, O. J. Org. Chem.
1978, 43, 2087.
Supplementary data
15. Arthur, C.; Cox, R. J.; Crosby, J.; Rahman, M. M.;
Simpson, T. J.; Soulas, F.; Spogli, R.; Szafranska, A.
E.; Westcott, J.; Winfield, C. J. Chem. Bio. Chem. 2002,
253.
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmcl.2007.03.060.
16. The pim TE did not hydrolyze the pimaricin. pim TE
(2.0 mM) was treated with pimaricin (2.0 mM) in 50 mM
phosphate buffer at pH 8.0. LCMS analysis of the reaction
showed no hydrolysis after 24 h at 37 ꢁC.
17. Boddy, C. N.; Schneider, T. L.; Hotta, K.; Walsh, C. T.;
Khosla, C. J. Am. Chem. Soc. 2003, 125, 3428.
18. Samel, S. A.; Bjorn, W.; Marahiel, M. A.; Essen, L.-O. J.
Mol. Biol. 2006, 359, 876.
References and notes
1. Hill, A. M. Nat. Prod. Rep. 2006, 23, 256.
2. Floss, H. G. J. Biotechnol. 2006, 124, 242.
3. Weissman, K. J.; Leadlay, P. E. Nat. Rev. Microbiol. 2005,
3, 925.
19. Tseng, C. C.; Bruner, S. D.; Kohli, R. M.; Marahiel, M.
A.; Walsh, C. T.; Sieber, S. A. Biochemistry 2002, 41,
13350.
4. Cane, D. E.; Walsh, C. T.; Khosla, C. Science 1998, 282, 63.
5. Aparicio, J. F.; Fouces, R.; Olivera, N.; Mendes, M. V.;
Martin, J. F. Chem. Biol. 2000, 7, 895.
20. Trauger, J. W.; Kohli, R. M.; Mootz, H. D.; Marahiel, M.
A.; Walsh, C. T. Nature 2000, 407, 215.
21. Trauger, J. W.; Kohli, R. M.; Walsh, C. T. Biochemistry
2001, 40, 7092.
22. Kohli, R. M.; Walsh, C. T.; Burkart, M. D. Nature 2002,
418, 658.
23. Boddy, C. N.; Hotta, K.; Tse, M. L.; Watts, R. E.; Khosla,
C. J. Am. Chem. Soc. 2004, 126, 7436.
6. Kohli, R. M.; Walsh, C. T. Chem. Commun. 2003, 297.
7. Gokhale, R. S.; Hunziker, D.; Cane, D. E.; Khosla, C.
Chem. Biol. 1999, 6, 117.
8. Lu, H.; Tsai, S.-C.; Khosla, C.; Cane, D. E. Biochemistry
2002, 41, 12590.
9. Weissman, K. J.; Smith, C. J.; Hanefeld, U.; Aggarwal, R.;
Bycroft, M.; Staunton, J.; Leadlay, P. F. Angew. Chem.
Int. Ed 1998, 37, 1437.