Isoprenoid biosynthesis in Escherichia coli via the methylerythritol phosphate pathway: Enzymatic conversion of methylerythritol cyclodiphosphate into a phosphorylated derivative of (E)-2-methylbut-2-ene-1,4-diol

Article abstract of DOI:10.1016/S0040-4039(02)00003-5

A crude cell-free system from an Escherichia coli strain overexpressing the cluster containing the three genes yfgA, yfgB, and gcpE converted 2-C-methyl-D-erythritol 2,4-cyclodiphosphate (1) into a phosphorylated derivative of (E)-2-methylbut-2-ene-1,4-diol (6), which most probably represents a novel intermediate in the methylerythritol phosphate pathway for isoprenoid biosynthesis. The free diol 6 was accumulated by phosphatase treatment of the crude enzyme preparation and was identified by comparison with a synthetic reference.

Full text of DOI:10.1016/S0040-4039(02)00003-5

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 43 (2002) 1413–1415
Isoprenoid biosynthesis in Escherichia coli via the
methylerythritol phosphate pathway: enzymatic conversion of
methylerythritol cyclodiphosphate into a phosphorylated
derivative of (E)-2-methylbut-2-ene-1,4-diol
Myriam Seemann,a,† Narciso Campos,b,† Manuel Rodriguez-Concepcio´n,b Ester Iban˜ez,b
Tore Duvold,a,‡ Denis Tritsch,a Albert Boronatb and Michel Rohmera,*
aUniversite´ Louis Pasteur/CNRS, Institut Le Bel, 4 rue Blaise Pascal, F-67070 Strasbourg Cedex, France
bDepartament de Bioqu´ımica i Biologia Molecular, Facultat de Qu´ımica, Universitat de Barcelona, Mart´ı i Franque`s 1,
ES-08028 Barcelona, Spain
Received 21 December 2001
Abstract—A crude cell-free system from an Escherichia coli strain overexpressing the cluster containing the three genes yfgA, yfgB,
and gcpE converted 2-C-methyl-D-erythritol 2,4-cyclodiphosphate (1) into a phosphorylated derivative of (E)-2-methylbut-2-ene-
1,4-diol (6), which most probably represents a novel intermediate in the methylerythritol phosphate pathway for isoprenoid
biosynthesis. The free diol 6 was accumulated by phosphatase treatment of the crude enzyme preparation and was identified by
comparison with a synthetic reference. © 2002 Elsevier Science Ltd. All rights reserved.
The 2-C-methyl-
D
-erythritol 4-phosphate (MEP) path-
and from pyruvate via 1-deoxyxylulose 5-phosphate
(DXP) and a series of 2-C-methyl- -erythritol (ME)
derivatives including MEP, 4-diphosphocytidyl ME, 4-
diphosphocytidyl ME 2-phosphate, and ME 2,4-
cyclodiphosphate 1 are now fairly well investigated.1
way for isoprenoid biosynthesis is present in most
bacteria, in green algae, and in the chloroplasts of all
phototrophic organisms, including the higher plants.
The early steps starting from glyceraldehyde phosphate
D
O
P
O
P
O
O
O
OH
OH
O
O
O
O
5
3
4
P
O
gcpE ?
lytB ?
O
reduction
O
P
O
P
O
O
P
O
P
P
2
2
4
4
O
1
O
O
OH
O
O
O
OH
and/or
OH OH
O
O
OH
O
O
1
2
3
O
O
P
phosphatase
O
P
O
O
5
5
3
1
2
4
OH
OH
6
Figure 1. Conversion of methylerythritol 2,4-cyclodiphosphate (1) into a (E)-2-methylbut-2-ene-1,4-diol derivatives (2, 6) by a
cell-free system from E. coli.
Keywords: biosynthesis; 2-C-methyl-
isoprenoids; lytB.
D-erythritol 4-phosphate; 2-C-methyl-D-erythritol 2,4-cyclodiphosphate; (E)-2-methylbut-2-ene-1,4-diol; gcpE;
* Corresponding author. Fax: +33 (0)3 90 13 45; e-mail: mirohmer@chimie.u-strasbg.fr
† These two authors contributed equally to this work.
‡ Present address: Leo Pharmaceutical Products, 55 Industriparken, DK-2750 Ballerup, Denmark.
0040-4039/02/$ - see front matter © 2002 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(02)00003-5

1414
M. Seemann et al. / Tetrahedron Letters 43 (2002) 1413–1415
ME 2,4-cyclodiphosphate 1 is the last identified interme-
diate, and very little is known concerning the steps
downstream of this metabolite. The pathway is known
to possess, at least in Escherichia coli, two branches,
separately leading to isopentenyl diphosphate 4 (IPP)
and dimethylallyl diphosphate 5 (DMAPP),2 and two
additional genes, gcpE and lytB have been recognized
as essential in this metabolic route.3 In order to
gain additional information on the MEP pathway, an
E. coli mutant capable of utilizing exogenous MVA
was constructed. For this purpose, the genes encoding
mevalonate kinase, phosphomevalonate kinase, diphos-
phomevalonate decarboxylase, and IPP isomerase were
introduced in its genome.4 Disruption of the gcpE gene
in this mutant showed that it was located on the trunk
line or directly on the branching of this pathway.3a
Feeding this mutant with [1-3H]ME resulted in the
accumulation of a single major metabolite, which was
identified as ME 2,4-cyclodiphosphate 1, suggesting that
this cyclodiphosphate was the putative substrate of the
gcpE gene product.5 In this contribution, we describe the
conversion of ME cyclodiphosphate 1 into a novel
metabolite by a cell-free system prepared from an E. coli
strain overexpressing the yfgA/yfgB/gcpE gene cluster
(Fig. 1).6
(17.5 mg) prepared from Corynebacterium ammoniagenes
treated with benzylviologen10 was performed. To the
resulting reaction mixture was added a small-scale incu-
bation of [2-14C]ME cyclodiphosphate, which allowed
the detection of the unknown metabolite owing to its
radioactive labeling. From the acetylated crude extract
of the enzyme test, repeated fractionation by column and
thin-layer chromatography afforded a single radioactive
compound (1 mg), which was identified as (E)-2-methyl-
but-2-ene-1,4-diol (6) as a diacetylated derivative by TLC
1
coelution and by comparison of its H and 13C NMR
spectra with those of the E and Z stereomers of the
corresponding diol obtained by chemical synthesis.11
(E)-2-Methylbut-2-ene-1,4-diol (6) was certainly not the
primary reaction product obtained from the crude
enzyme test. The free diol was most likely released from
the corresponding diphosphate 2 by the action of either
endogenous phosphatase from E. coli or of the exogenous
phosphatase added to the incubation buffer in order to
improve the conversion of [2-14C]ME cyclodiphosphate
1 into the diol 6. The diol was only obtained from the
enzymatic conversion of [2-14C]ME cyclodiphosphate: it
was not observed when the incubation was performed
with an enzyme preparation inactivated by boiling. As
a crude cell-free system was utilized, it however cannot
be excluded that the diol derivative resulted from the
consecutive action of several enzymes, including the
products of the gcpE, yfgA, and yfgB genes. Whether
these proteins were effectively required for the conver-
sion of [2-14C]ME cyclodiphosphate 1 is still a matter
of investigations. (E)-4-Hydroxy-3-methylbut-2-enyl
diphosphate (2) was recently identified in E. coli as a
major activator for human gd T cells.12 It is accumulated
in cells defective in the lytB gene, whereas it is absent in
cells defective in the gcpE gene. All available data are in
accordance with ME cyclodiphosphate 1 as a substrate
of the gcpE gene product and with diphosphate of 6 as
reaction product of the gcpE gene product and substrate
of the lytB gene product. The conversion of [2-14C]ME
cyclodiphosphate 1 into a diol 6 derivative formally
requires a reduction and an elimination step. Accord-
ingly, a cofactor has to be involved in the reduction step.
The addition of any reducing cofactor such as NADPH,
NADH, FMN or FAD did not influence however, the
conversion yield. Because of the small amounts of the
reaction product, which were synthesized de novo, the
endogenous intracellular cofactor pools were most likely
largely sufficient.
For the convenience of detection, conversion of ME
cyclodiphosphate 1 was tested using a 14C labeled sub-
strate. For this, [2-14C]DXP was synthesized in a one-pot
procedure from [2-14C]pyruvate and non-labeled glycer-
aldehyde phosphate using the DXP synthase and directly
converted into [2-14C]MEP using the DXP isomero-
reductase (D. Tritsch, unpublished results). [2-14C]MEP
was further converted into [2-14C]ME cyclodiphosphate
using a one-pot procedure similar to a recently described
methodology involving a mixture of the ygbP, ychB, and
ygbB gene products and the corresponding cofactors.7
Incubation of [2-14C]ME cyclodiphosphate 1 with a
crude cell-free system from an E. coli strain overexpress-
ing the gcpE, yfgA, and yfgB gene cluster8 resulted in the
formation of several radioactive products. Next to
remaining starting material 1, a most probably non-phos-
phorylated unknown compound as well as a compound,
which might correspond to a diphosphate ester, were
detected on TLC plates.9 The two latter compounds most
likely represented novel metabolites of [2-14C]ME
cyclodiphosphate. Noteworthy was the non-sensitivity of
[2-14C]ME cyclodiphosphate 1 towards the tested alka-
line phosphatase, which hydrolyzed acyclic mono- and
diphosphate esters, as tested on MEP and IPP. Forma-
tion of a non-phosphorylated compound was not surpris-
ing as the enzyme test was performed with a crude
cell-free system containing endogenous phosphatases.
Addition of an exogenous phosphatase to the incubation
medium did not affect the substrate, but significantly
increased the concentration of the above-mentioned least
polar novel metabolite. This concentration increase was
concomitantly accompanied by the disappearance of the
most polar unknown metabolite, suggesting that it cor-
responded to the phosphorylated form of the former one.
Finally, elimination of water or of a better leaving group
such as phosphate from C-4 of diphosphate 2 (corre-
sponding to C-4 of IPP 4 and DMAPP 5) would produce
an allylic cation 3 in the active site (Fig. 1). Such cations
are often encountered in isoprenoid biosynthesis (see for
instance the reactions catalyzed by the IPP isomerase, the
prenyl transferases, the squalene synthase or the terpene
synthases). A single reduction at C-1 or C-3 of such a
carbocationic intermediate 3 would yield DMAPP 5 or
IPP 4, respectively.
Direct identification was therefore attempted. A large-
scale incubation of non-labeled ME cyclodiphosphate
One of the last bottlenecks in the elucidation of the MEP
pathway is now being solved by the discovery of

M. Seemann et al. / Tetrahedron Letters 43 (2002) 1413–1415
1415
this [2-14C]ME cyclodiphosphate metabolite. It remains
to isolate now the native compound, most likely the
diphosphate 2,14 which released the diol 6 upon phos-
phatase treatment. IPP 4 and DMAPP 5 are synthe-
sized in E. coli from an unknown common intermediate
derived from ME cyclodiphosphate 1 via two distinct
branches.2 It also remains to check how the diphos-
phate 2 is converted into IPP and/or DMAP and
whether it belongs to the main trunk or to the IPP or
the DMAPP branch of the MEP pathway.
was completed with buffer (20 mL) and used for enzyme
assays at 37°C for 7 and 20 h with [2-14C]-2-C-methyl-
-ery-
D
thritol 2,4-cyclodiphosphate solution (1.1×105 cpm, 16 Ci
mol−1). When required, an E. coli Type-III S alkaline
phosphatase (0.25 U) was added to the reaction mixture.
9. Isolation and identification of 6. Direct TLC of aliquots
of the reaction medium revealed several radioactive spots
including 6 (Rf=0.85), free ME (Rf=0.70), and 1 (Rf=
0.52). Additional TLC of 6 (CHCl3–CH3OH, Rf=0.56)
indicated that the compound was not phosphorylated: its
polarity was comprised between those of isopentenol
(Rf=0.56) and ME (Rf=0.22). To fully characterize 6, a
large-scale incubation from a 1 L culture yielding cell-free
system (40 mL) diluted with additional buffer (20 mL) was
performed with unlabeled 1 (17.5 mg, 58 mMol). A small-
scale incubation was made as described above, but with
larger amounts of [2-14C]-1 (106 cpm). The two assays were
mixed and lyophilized. The residue was acetylated (Ac2O–
pyridine, 1:1, 10 mL) overnight at room temperature. After
removal of the reagents, the residue was resuspended in
CHCl3 (12 mL), and the insoluble material removed by
filtration. After concentration to dryness, the filtrate (1 g,
836000 cpm) was separated on a silica column (8 g) eluted
with hexane–EtOAc (3:1). Fractions (5 mL) were collected,
and an aliquot (4 mL) analyzed by TLC. Radioactivity was
monitored with a PhosphorImager, and labeled fractions
with the same Rf were pooled together. Three radioactive
fractions were detected: a first one containing the diacetate
of 6 (Rf=0.40, 200 mg, 3×105 cpm), a second one containing
anunidentifiedcompound(Rf=0.25, 100mg, 3.2×104 cpm),
and a third one containing ME triacetate resulting from the
chemical degradation of 1 during the work-up (Rf=0.20,
20 mg, 4.2×105 cpm). The least polar fraction was further
purified on a silica column (9 g) eluted first with CH2Cl2
in order to remove nearly all impurities, and then with
EtOAc yielding the radioactive diacetate of 6 (1 mg, 1.8×105
cpm). Diacetate of 6. 1H NMR spectrum (400 MHz,
CDCl3): l (ppm)=1.75 (3H, broad s, 5-H), 2.09 (3H, s;
CH3COO-), 2.11 (3H, s, CH3COO-), 4.51 (2H, s, 1-H), 4.65
(2H, d, J3,4=7 Hz, 4-H), 5.63 (1H, tq, J3,4=7 Hz, J3,5=1.5
Hz, 3-H). 13C NMR (500 MHz, CDCl3) spectrum: l
(ppm)=14.2 (C-5), 20.9 (CH3COO-), 21.0 (CH3COO-),
60.6 (C-4), 68.6 (C-1), 121.7 (C-3), 135.8 (C-2), 170.7
(CH3COO-), 171.0 (CH3COO-).
Acknowledgements
We thank Mr J. D. Sauer for the NMR measurements.
A.B. was supported by grants from the ‘Generalitat de
Catalunya’ (1999SGR-00032) and from ‘Ministerio de
Ciencia y Tecnologia’ (BIO1999-0503-C02-01), and
M.R. by a grant from ‘Institut Universitaire de France’.
References
1. (a) Rohmer, M. Nat. Prod. Rep. 1999, 16, 565–573; (b)
Eisenreich, W.; Rohdich, F.; Bacher, A. Trends Plant Sci.
2001, 6, 78–84.
2. Rodriguez-Concepcio´n, M.; Campos, N.; Lois, L. M.;
Maldonaldo, C.; Hoeffler, J. F.; Grosdemange-Billiard, C.;
Rohmer, M.; Boronat, A. FEBS Lett. 2000, 473, 328–332.
3. (a) Campos, N.; Rodriguez-Concepcio´n, M.; Seemann, M.;
Rohmer, M.; Boronat, A. FEBS Lett. 2001, 488, 170–173;
(b) Altincicek, B.; Kollas, A.-K.; Sanderband, S.; Wiesner,
J.; Hintz, M.; Beck, E.; Jomaa, H. J. Bacteriol. 2001, 183,
2411–2416; (c) Cunningham, F. X., Jr.; Lafond, T. P. T.;
Gantt, E. J. Bacteriol. 2000, 182, 5841–5848; (d) Altincicek,
B.; Kollas, A. K.; Eberl, M.; Wiesner, J.; Sanderband, S.;
Hintz, M.;Beck, E.;Jomaa, H. FEBSLett. 2001, 499, 37–40.
4. Campos, N.; Rodriguez-Concepcio´n, M.; Sauret-Gu¨eto, S.;
Gallego, F.; Lois, L. M.; Boronat, A. Biochem. J. 2001, 353,
59–67.
5. Seemann, M.; Campos, N.; Rodriguez-Concepcio´n, M.;
Hoeffler, J. F.; Grosdemange-Billiard, C.; Boronat, A.;
Rohmer, M. Tetrahedron Lett. 2002, 43, 775–778.
6. Because of the genetic links yfgA, yfgB, and gcpE, which
are on the same operon of the E. coli genome, an expression
plasmid containing the genomic region covering yfgB, yfgA,
and gcpE was constructed and stably transformed into E.
coli creating the strain BL21(DE3)pLys [PET-T7-yfgA-
yfgB-gcpE].
10. (a) Ostrovsky, D.; Shashkov, A.; Sviridov, A. Biochem. J.
1993, 295, 901–902; (b) Malarova, I.; Shipanova, I.;
Sibeldina, L.; Shashkov, A.; Tantsirev, G. BioFactors 1992,
3, 261–264.
11. (E)-2-Methyl-2-ene-1,4-diol (6) was synthesized by LiAlH4
reduction of methylfumaric acid as previously described for
the reduction of 3-methylfuran-2(5H)-one or citraconic
anhydride.13 The Z stereomer was obtained by reduction
of citraconic anhydride.13
7. Schuhr, C. A.; Hecht, S.; Kis, K.; Eisenreich, W.;
Wungsintaweekul, J.; Bacher, A.; Rohdich, F. Eur. J. Org.
Chem. 2001, 3221–3226.
8. Enzyme assays. In a typical experiment, the above men-
tioned E. coli strain was grown for about 4 h at 30°C in
LB medium (50 mL) containing chloramphenicol (34 mg
mL−1) and ampicillin (100 mg mL−1) until reaching a 0.65
OD (600 nm). Induction was performed with IPTG (0.5
mM) for 6 h. Cells were harvested by centrifugation (7000
g, 10 min), resuspended in degassed buffer (4 mL, 50 mM
Tris–HCl, pH 8, 1 mM PMSF, 1 mM DTT, and 5 mM
MgCl2), and broken by sonication (2×20 s, with 1 min
cooling). Cell debris was removed by centrifugation (16000
g, 10 min). The resulting crude cell-free material (130 mL)
12. Hintz, M.; Reichenberg, A.; Altincicek, B.; Bahr, U.;
Gschwind, R. M.; Kollas, A. K.; Beck, E.; Wiesner, J.;
Eberl, M.; Jomaa, H. FEBS Lett. 2001, 509, 317–322.
13. Duvold, T.; Cal´ı, P.; Bravo, J. M.; Rohmer, M. Tetrahedron
Lett. 1997, 38, 6181–6184.
14. Note added in proof. Diphosphate 2 was also shown to derive
from deoxyxylulose in vivo in E. coli cells expressing the
xylB, ispC, ispD, ispE, ispF and gcpE genes. Hecht, S.;
Eisenreich, W.; Adam, P.; Amslinger, S.; Kis, K.; Bacher,
A.; Arigoni, D.; Rohdich, F. Proc. Natl. Acad. Sci. USA
2001, 98, 14837–14842.