Two distinct mechanisms for TIM barrel prenyltransferases in bacteria

Article abstract of DOI:10.1021/ja109578b

The reactions of two bacterial TIM barrel prenyltransferases (PTs), MoeO5 and PcrB, were explored. MoeO5, the enzyme responsible for the first step in moenomycin biosynthesis, catalyzes the transfer of farnesyl to 3-phosphoglyceric acid (3PG) to give a product containing a cis-allylic double bond. We show that this reaction involves isomerization to a nerolidyl pyrophosphate intermediate followed by bond rotation prior to attack by the nucleophile. This mechanism is unprecedented for a prenyltransferase that catalyzes an intermolecular coupling. We also show that PcrB transfers geranyl and geranylgeranyl groups to glycerol-1-phosphate (G1P), making it the first known bacterial enzyme to use G1P as a substrate. Unlike MoeO5, PcrB catalyzes prenyl transfer without isomerization to give products that retain the trans-allylic bond of the prenyl donors. The TIM barrel family of PTs is unique in including enzymes that catalyze prenyl transfer by distinctly different reaction mechanisms.

Full text of DOI:10.1021/ja109578b

COMMUNICATION
pubs.acs.org/JACS
Two Distinct Mechanisms for TIM Barrel Prenyltransferases in Bacteria
Emma H. Doud,†,‡ Deborah L. Perlstein,‡ Manuel Wolpert,‡ David E. Cane,§ and Suzanne Walker*,†
†Department of Chemistry and Chemical Biology, Harvard University, 12 Oxford Street, Cambridge, Massachusetts 02139,
United States
‡Department of Microbiology and Molecular Genetics, Harvard Medical School, 200 Longwood Avenue, Boston,
Massachusetts 02115, United States
§Department of Chemistry, Brown University, Box H, Providence, Rhode Island 02912-9108, United States
S Supporting Information
b
ABSTRACT: The reactions of two bacterial TIM barrel
prenyltransferases (PTs), MoeO5 and PcrB, were explored.
MoeO5, the enzyme responsible for the first step in
moenomycin biosynthesis, catalyzes the transfer of farnesyl
to 3-phosphoglyceric acid (3PG) to give a product contain-
ing a cis-allylic double bond. We show that this reaction
involves isomerization to a nerolidyl pyrophosphate inter-
mediate followed by bond rotation prior to attack by the
nucleophile. This mechanism is unprecedented for a pre-
nyltransferase that catalyzes an intermolecular coupling. We
also show that PcrB transfers geranyl and geranylgeranyl
groups to glycerol-1-phosphate (G1P), making it the first
Figure 1. GGGPS converts glycerol-1-phosphate (G1P, 1) and GGPP
(2) to compound 3. MoeO5 farnesylates 3-phosphoglycerate (3PG, 4)
using farnesyl pyrophosphate (FPP, 5). The portion of MmA synthe-
sized by MoeO5 is highlighted in red.
known bacterial enzyme to use G1P as a substrate. Unlike
MoeO5, PcrB catalyzes prenyl transfer without isomeriza-
tion to give products that retain the trans-allylic bond of the
prenyl donors. The TIM barrel family of PTs is unique in
including enzymes that catalyze prenyl transfer by distinctly
different reaction mechanisms.
oxygen on phosphorylated triose acceptors, but the reactions
they catalyze can proceed by two very different mechanisms.4
The TIM barrel PTs, GGGPS and MoeO5, both make ether-
linked products but with different allylic bond geometries
(Figure 1). Unlike the GGGPS product, which retains the
trans-geometry of the starting material and likely proceeds by a
straightforward displacement mechanism, the MoeO5 product
(6) contains a cis-allylic double bond.8 MoeO5 is the only known
prenyltransferase that forms an intermolecular cis-allylic bond
using a trans-allylic precursor.10 This linkage suggests that bond
isomerization occurs prior to coupling during the catalytic
reaction, and a possible reaction mechanism involving the
ature uses prenyltransferases (PTs) to make a wide variety
of primary and secondary metabolites containing prenyl
N
intermediacy of nerolidyl pyrophosphate (NPP, 7) is shown in
groups, and there is enormous interest in understanding the
scope and utility of these enzymes.1 The best studied classes of
PTs, the polyprenyl synthases and the cyclizing terpene syn-
thases, contain an alpha helical fold, as do the protein farnesyl and
geranylgeranyl transferases.2 Recently, the first PT having a
triosephosphate isomerase (TIM) barrel fold was identified in
archaea.3-5 This PT, geranylgeranylglyceryl phosphate synthase
(GGGPS), catalyzes the transfer of a geranylgeranyl (C20) group
to glycerol-1-phosphate (G1P) during the first step in the
biosynthesis of the major phospholipid component of archaeal
membranes (Figure 1).6,7 We recently discovered a bacterial
GGGPS homologue, MoeO5, in the gene cluster for the anti-
biotic moenomycin. MoeO5 couples 3-phosphoglyceric acid
(3PG) to a farnesyl group to form the starter unit for moeno-
mycin biosynthesis (Figure 1).8,9 Here, we study the mechanism
of MoeO5, describe key residues that can be used to predict the
acceptor substrate in TIM barrel PTs, apply these criteria to a
bacterial GGGPS homologue of unknown function, PcrB, and
establish its substrates biochemically. We conclude that TIM
barrel PTs are dedicated to the transfer of prenyl groups to
Figure 2A.11
To test this hypothesis, we compared the reaction rates for
FPP (5) and (3S)-NPP (7) using LC/MS. Under initial rate
conditions, the turnover numbers for FPP and NPP were
indistinguishable (7.8 ( 0.1 min-1 for FPP and 7.4 ( 0.7 min-1
for NPP).12 In addition, we detected a compound with a
retention time and mass consistent with those of NPP (7) in
reactions of MoeO5 incubated with FPP in the absence of 4
(Figure 2B).13 These results support a mechanism in which
MoeO5 isomerizes FPP (5) to NPP (7), which undergoes
rotation about the C-2,3-bond to the cisoid conformer prior to
prenyl transfer to phosphoglyceric acid (4, Figure 2).
To gain insight into how MoeO5 and GGGPS are capable of
catalyzing prenyl transfer to acceptors in different oxidation
states via fundamentally different mechanisms, we aligned the
sequences of 50 archaeal TIM barrel PTs and MoeO5 and
Received: November 3, 2010
Published: January 7, 2011
r
2011 American Chemical Society
1270
dx.doi.org/10.1021/ja109578b J. Am. Chem. Soc. 2011, 133, 1270–1273
|

Journal of the American Chemical Society
COMMUNICATION
Figure 4. (A) Phosphosugar acceptors tested as substrates for PcrB:
G1P (1), 3PG (4), and G3P (8). (B) Prenylpyrophosphate donors
tested as substrates for PcrB: GGPP (2), FPP (5), and GPP (9). (C) The
reactions shown to be catalyzed by PcrB.
replaced by arginine, which may favor attractive interactions
with the substrate carboxylate.
Changes in the acceptor binding loop help explain the changes
in acceptor substrate preferences, but other features are likely
responsible for the ability of MoeO5 to catalyze isomerization of
the allylic diphosphate bond of FPP. Notably, MoeO5 con-
tains a proline-rich insertion (Figure 3B) between β-3 and R-
3*, a region in GGGPS that is suggested to form a “swinging
door” that closes over the prenyl chain.3 In addition, while the
GGGPS binding tunnel is largely hydrophobic and acidic, the
MoeO5 homology model shows a nest of basic and π-electron
containing residues in the vicinity of the “swinging door” that
could help stabilize a partially ionized nerolidyl intermediate.
Using the absence of glutamate and the presence of arginine in
the acceptor binding loop, combined with the presence of a
proline insertion in the R-3* strand, we have identified several
MoeO5 homologues in other bacteria, always clustered
with genes putatively involved in secondary metabolite bio-
synthesis.15
Figure 2. (A) MoeO5 catalyzes the formation of 6 from 3PG (4) and
FPP (5) via a mechanism proposed to involve isomerization to NPP (7).
(B) NPP forms during the MoeO5 reaction. NPP and FPP have the
same [M - H]- (381.1237). LC/MS extracted ion chromatograms of
an (a) overlay of NPP and FPP standards (blue and red, respectively)
demonstrating reproducible differences in retention times; (b) MoeO5
reaction of FPP spiked with NPP following quenching of the enzymatic
reaction (green); (c) MoeO5 reaction with FPP showing the formation
of a new peak with an RT consistent with NPP (black).14
We also investigated the enzymatic function of PcrB, a
putative prenyltransferase with a TIM barrel fold.16 PcrB is
conserved in several Gram-positive genera, including bacilli,
staphylococci, and listeriae. Based on sequence analysis, it has
been speculated that PcrB either is an aromatic prenyltransferase
or has a function similar to GGGPS.3,7,17 Since PcrB contains an
acceptor binding loop and “swinging door” similar to those in
GGGPS, it seemed likely that PcrB is a eubacterial prenyltrans-
ferase that uses G1P as a substrate.
To test this possibility, we cloned, expressed, and purified
Bacillus subtilis PcrB and incubated it with three potential
acceptors, glycerol-1-phosphate (G1P, 1), glycerol-3-phosphate
(G3P, 8), and 3-phosphoglycerate (3PG, 4), and three potential
prenyl donors, GGPP (2), FPP (5), and GPP (9) (Figure 4A and
B).14 The reactions, analyzed by mass spectrometry, showed
that PcrB uses G1P (1) exclusively as an acceptor but is able to
transfer both the C10 and C20 carbon donors (GPP, 9; GGPP, 2)
at similar rates (Figure 3C, Table S1, turnover: 1.8 ( 0.1 and
0.7 ( 0.1 min-1, respectively); furthermore, upon extended
incubation both reactions went to completion.18 In contrast, FPP
(5), which contains 15 carbons, is a poor substrate for PcrB, and
reactions using it as a donor proceeded to less than 10%
conversion. NMR analysis showed that the PcrB product formed
using GPP as a donor has structure 10 (Figure 3, Figures S1-S3)
in which the allylic bond is trans and the ether linkage is to the C3
hydroxyl of G1P.14 Hence the reaction is identical to that
catalyzed by GGGPS, except that both short (C10) and long
(C20) prenyl chains are substrates.19 Although G1P (1) is a
common building block for membrane lipids in archaea, PcrB is
Figure 3. Partial sequence alignments of MoeO5 (S. ghanaensis), repre-
sentative archaeal GGGPSs, and PcrB from B. subtilis (BsPcrB). (A)
Alignments of the acceptor binding loop. The tyrosine and arginine
changes in MoeO5 are outlined in red. (B) Alignments of the “swinging
door” region of AfGGGPS described in the text.
compared them to the GGGPS crystal structure, which contains
a bound acceptor substrate.3 Amino acids in the β-6-loop-6
region of GGGPS make direct contacts to the acceptor substrate,
and there are several striking changes in this region of MoeO5
(Figure 3).3 For example, E167, which contacts both the C-2 and
C-3 hydroxyl groups of G1P in the GGGPS crystal structure, is
replaced by tyrosine. This change likely prevents repulsive
interactions with the 3PG carboxylate. In addition, S169 is
1271
dx.doi.org/10.1021/ja109578b |J. Am. Chem. Soc. 2011, 133, 1270–1273

Journal of the American Chemical Society
COMMUNICATION
the first known bacterial enzyme to use it as a substrate.17,20 The
pcrB gene is always found in the same operon as pcrA,21 which
encodes an essential helicase involved in the replication and
transfer of DNA elements between bacteria.22 This genetic
linkage suggests a possible functional connection. Knowledge
of PcrB’s enzymatic function may facilitate efforts to elucidate its
cellular role.
The reactions catalyzed by GGGPS, MoeO5, and PcrB
suggest that the TIM barrel family of PTs is dedicated to the
transfer of prenyl groups to oxygen acceptors of triose phos-
phates. We have shown that the oxidation state of the acceptor
and the regiochemistry of reaction can vary depending on the PT.
More remarkably, this family of enzymes can catalyze the
formation of products containing either cis- or trans-allylic double
bonds from trans-prenyl donors.14,23 Thus, relatively small
changes in three-dimensional architecture lead to prenyl transfer
by two distinct mechanisms, one involving direct displacement
and the other requiring bond isomerization and rotation. This is
unprecedented for a single PT family. Structural information on
MoeO5 or its homologues should provide clues to the specific
sequence changes that have led to this dramatic switch in
mechanism.
(3) Payandeh, J.; Fujihashi, M.; Gillon, W.; Pai, E. F. J. Biol. Chem.
2006, 281, 6070–6078.
(4) TIM barrel proteins catalyze a wide range of enzymatic reactions;
their structural details, mechanism, and evolution have been the focus of
numerous studies. See: (a) Reardon, D.; Farber, G. K. FASEB J. 1995, 9,
497–503. (b) Copley, R. R.; Bork, P. J. Mol. Biol. 2000, 303, 627–640.
Lang, D.; Thoma, R.; Henn-Sax, M.; Sterner, R.; Wilmanns, M. Science
2000, 289, 1546–1550. (c) Wierenga, R. K. FEBS Lett. 2001, 492, 193–
198. (d) Nagano, N.; Orengo, C. A.; Thornton, J. M. J. Mol. Biol. 2002,
321, 741–765. (e) Anantharaman, V.; Aravind, L.; Koonin, E. V. Curr.
Opin. Chem. Biol. 2003, 7, 12–20. (f) Gerlt, J. A.; Raushel, F. M. Curr.
Opin. Chem. Biol. 2003, 7, 252–264.
(5) A distinct PT barrel structure has recently been determined for
several aromatic prenyltransferases, including Orf2 and CloQ. See: (a)
Tello, M.; Kuzuyama, T.; Heide, L.; Noel, J. P.; Richard, S. B. Cell. Mol.
Life Sci. 2008, 65, 1459–1463. (b) Kuzuyama, T.; Noel, J. P.; Richard,
S. B. Nature 2005, 435, 983–987.
(6) (a) Chen, A. J.; Zhang, D. L.; Poulter, C. D. J. Biol. Chem. 1993,
268, 21701–21705. (b) Zhang, D. L.; Poulter, C. D. J. Am. Chem. Soc.
1993, 115, 1270–1277. (c) Zhang, D. L.; Poulter, C. D. J. Org. Chem.
1993, 58, 3919–3922. (d) Soderberg, T.; Chen, A. J.; Poulter, C. D.
Biochemistry 2001, 40, 14847–14854.
(7) Nemoto, N.; Oshima, T.; Yamagishi, A. J. Biochem. 2003, 133,
651–657.
(8) Ostash, B.; Doud, E. H.; Lin, C.; Ostash, I.; Perlstein, D. L.; Fuse,
S.; Wolpert, M.; Kahne, D.; Walker, S. Biochemistry 2009, 48, 8830–
8841.
’ ASSOCIATED CONTENT
(9) Ostash, B.; Saghatelian, A.; Walker, S. Chem. Biol. 2007, 14,
257–267.
S
Supporting Information. Experimental Procedures, in-
b
cluding cloning and purification of proteins, and raw MS and
NMR data, complete refs 16 and 21b. This material is available
free of charge via the Internet at http://pubs.acs.org.
(10) The cis-geometry observed in the products of cis-prenyl chain
elongating enzymes results from a different type of reaction mechanism.
See: Kharel, Y.; Koyama, T. Nat. Prod. Rep. 2003, 20, 111–118.
(11) Nerolidyl diphosphate is a demonstrated intermediate in the
cyclization-isomerization-cyclization of FPP to 6-, 10-, and 11-mem-
bered ring sesquiterpenes with cis-double bonds. See: (a) Cane, D. E.;
Ha, H. J.;Pargellis, C.; Waldmeier, F.; Swanson, S.; Murthy, P. P. N.
Bioorg. Chem. 1985, 13, 246–265. (b) Cane, D. E.; Ha, H. J. J. Am. Chem.
Soc. 1988, 110, 6865–6870. (c) Cane, D. E.; Pawlak, J. L.; Horak, R. M.;
Hohn, T. M. Biochemistry 1990, 29, 5476–5490.(d) Domingo, V.;
Arteaga, J. F.; del Moral, J. F. Q.;(e) Barrero, A. F. Nat. Prod. Rep.
2009, 26, 115–134. (f) Noel, J. P.; Dellas, N.; Faraldos, J. A.; Zhao, M.;
Hess, B. A.; Smentek, L.; Coates, R. M.; O’Maille, P. E. ACS Chem. Biol.
2010, 5, 377–392.
’ AUTHOR INFORMATION
Corresponding Author
Suzanne_walker@hms.harvard.edu
’ ACKNOWLEDGMENT
We thank Charles Sheahan for training and assistance at the
HMS East Quad NMR Facility. This work was partially sup-
ported by NERCE (NIAID U54AI057159) as well as NIH Grants
GM076710 and AI083214 (S.W.) and GM30301 (D.E.C.). E.H.D.
was supported by an NSF graduate research fellowship, and
D.L.P. was supported by an NIH postdoctoral fellowship. All
LC/MS data were acquired on an Agilent 6520 Q-TOF spectro-
photometer supported by the Taplin Funds for Discovery
Program (P.I.: S.W.). We thank the Biomolecule Production
Core of NERCE for reagents (NIAID U54AI057159).
(12) Rates were measured in triplicate using 1 mM 3PG, 40 μM lipid
pyrophosphate, and 120 nM enzyme. Conversion was <10%. See
Supporting Information (SI) for full details.
(13) We cannot rule out that the intermediate detected could also be
cis,trans-farnesylpyrophosphate.
(14) See SI.
(15) MoeO5 homologues are present and clustered with moeno-
mycin type biosynthetic clusters in S. clavuligerus, Photohrabdus lumi-
nescens subsp. laumondii TTO1, and P. asymbiotica.
(16) Badger, J.; et al. Proteins: Struct., Funct., Bioinf. 2005, 60, 787–796.
(17) Guldan, H.; Sterner, R.; Babinger, P. Biochemistry 2008, 47,
7376–7384.
(18) Rates were measured in triplicate using 1 mM G1P, 40 μM lipid
pyrophosphate, and 3.4 μM enzyme. Conversion was <15%. See SI for
full details.
(19) How FPP is excluded is unknown, although we note that the
hydrophobic tunnel in PcrB has a kink that would require substrates
longer than 10 carbons to bend.
(20) A bacterial enzyme that interconverts glyceraldehyde-phos-
phate and G1P was recently identified, which suggests that bacteria
produce this metabolite for some purpose. See: ref 17.
(21) (a) Ji, Y. D.; Zhang, B.; Van Horn, S. F.; Warren, P.; Woodnutt,
G.; Burnham, M. K. R.; Rosenberg, M. Science 2001, 293, 2266–2269.
(b) Kobayashi, K.; et al. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 4678–
4683.
’ REFERENCES
(1) (a) Dewick, P. M. Nat. Prod. Rep. 2002, 19, 181–222. Heide, L.
Curr. Opin. Chem. Biol. 2009, 13, 171–179. (b) Liang, P. H. Biochemistry
2009, 48, 6562–6570.
(2) (a) Liang, P. H.; Ko, T. P.; Wang, A. H. J. Eur. J. Biochem. 2002,
269, 3339–3354. (b) Christianson, D. W. Chem. Rev. 2006, 106, 3412–
3442. (c) Greenhagen, B. T.; O’Maille, P. E.; Noel, J. P.; Chappell, J.
Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 9826–9831. (d) Lane, K. T.;
Beese, L. S. J. Lipid Res. 2006, 47, 681–699. (e) Thulasiram, H. V.;
Erickson, H. K.; Poulter, C. D. Science 2007, 316, 73–76. (f) Vedula,
L. S.; Zhao, Y. X.; Coates, R. M.; Koyama, T.; Cane, D. E.; Christianson,
D. W. Arch. Biochem. Biophys. 2007, 466, 260–266. (g) Shishova, E. Y.;
Yu, F. L.; Miller, D. J.; Faraldos, J. A.; Zhao, Y. X.; Coates, R. M.;
Allemann, R. K.; Cane, D. E.; Christianson, D. W. J. Biol. Chem. 2008,
283, 15431–15439. (h) Aaron, J. A.; Lin, X.; Cane, D. E.; Christianson,
D. W. Biochemistry 2010, 49, 1787–1797.
1272
dx.doi.org/10.1021/ja109578b |J. Am. Chem. Soc. 2011, 133, 1270–1273

Journal of the American Chemical Society
COMMUNICATION
(22) (a) Ruiz-Maso, J. A.; Anand, S. P.; Espinosa, M.; Khan, S. A.; del
Solar, G. J. Bacteriol. 2006, 188, 7416–7425. (b) Slatter, A. F.; Thomas,
C. D.; Webb, M. R. Biochemistry 2009, 48, 6326–6334. (c) Lee, C. A.;
Babic, A.; Grossman, A. D. Mol. Microbiol. 2010, 75, 268–279.
(23) Using homology searches, we have identified a putative new
subfamily of TIM barrel PTs that are highly conserved in Gram-negative
bacteroides, and we predict that these enzymes also use G1P as a
substrate and produce a trans product (Figure S4).
1273
dx.doi.org/10.1021/ja109578b |J. Am. Chem. Soc. 2011, 133, 1270–1273