The chemical mechanism of D-1-deoxyxylulose-5-phosphate reductoisomerase from Escherichia coli

Article abstract of DOI:10.1002/anie.200700647

(Chemical Equation Presented) Find the way: 3-[²H]- and 4-[ ²H]-labeled 1-deoxyxylulose-5-phosphate were synthesized and used to investigate the chemical mechanism of 1-deoxyxylulose-5-phosphate reductoisomerase (DXR) from E. coli. The observation of inverse secondary kinetic isotope effects for both labeled substrates indicates that DXR uses a retro-aldol/aldol mechanism in which the recombination reaction is the rate-limiting step (see scheme).

Full text of DOI:10.1002/anie.200700647

Communications
DOI: 10.1002/anie.200700647
Enzyme Mechanisms
The Chemical Mechanism of d-1-Deoxyxylulose-5-phosphate
Reductoisomerase from Escherichia coli**
Ursula Wong and Russell J. Cox*
The enzyme d-1-deoxyxylulose-5-phosphate reductoisomer-
ase (DXR, EC 1.1.1.267) catalyzes the key isomerization
reaction during the biosynthesis of the terpene precursors
isopentenyl diphosphate 1 and dimethylallyl diphosphate 2
through the non-mevalonate pathway in plants and bacteria.[1]
DXR catalyzes the interconversion of d-1-deoxyxylulose-5-
phosphate (DXP) 3 and d-methylerythritol phosphate (MEP)
4, utilizing nicotinamide adenine dinucleotide phosphate
(NADPH) as the reductant (Scheme 1).[2] Fosmidomycin 5
selectively and potently inhibits DXR, and 5 and its close
analogues show useful antimalarial activity both in vitro and
in vivo against Plasmodium falciparum. Inhibition of DXR
thus offers a new and, as yet, little-exploited mechanism for
the development of new antimicrobial agents. Although the
mechanism of inhibition of DXR by 5 is well understood both
kinetically[3,4] and crystallographically,[5–7] little progress has
been made in elucidating the chemical mechanism of
substrate processing by DXR. The structural and biochemical
results reported to date show that DXR is a homodimer, with
one active site per subunit, and support a kinetic mechanism
in which NADPH binds before DXP 3.[4]
The first reaction catalyzed is an apparent
concerted a-ketol rearrangement (mecha-
nism A, Scheme 1) or a step-wise fragmen-
tation–reassembly mechanism through
retro-aldol/aldol
steps
(mechanism B,
Scheme 1). The resulting aldehyde 6 is then
reduced by using NADPH.
Tentative support for mechanism B has
been gathered by Fox and Poulter[8] and by
Lui and co-workers[9] by using fluorinated
substrate analogues. However, replacement
of hydroxy groups by fluorine atoms could
alter substrate binding, and formation of a-
fluoroketones affects carbonyl pKa values
and electrophilicity, making interpretation
of results difficult. Thus, no conclusive
evidence in favor of, or ruling out, either of
the possible rearrangement mechanisms has
been found despite this subject being of
intense interest.[4,8–11] Determination of the
chemical mechanism of DXR could allow
the development of new classes of inhibitors
with potentially useful pharmacological
properties.
To distinguish between the two possible
rearrangement mechanisms for DXR, we
decided to synthesize selectively deuterated
substrates for use in kinetic assays. We have
Scheme 1. Reactions catalyzed by DXR and the two possible mechanisms for the
rearrangement step.
already described a short synthesis of enantiomerically pure 3
that allows the incorporation of isotopic labels and the
development of a kinetic assay for DXR by using recombi-
nant protein obtained from Escherichia coli.[12] Incorporation
of 2H selectively at C3 and C4 of DXP would provide
substrates that could allow the observation of secondary
kinetic isotope effects (KIE) and potentially distinguish the a-
ketol mechanism from the retro-aldol/aldol mechanism.
To synthesize the desired isotopically labeled substrates,
we modified our previous synthetic route.[12,13] Thus propargyl
[*] U. Wong, Dr. R. J. Cox
School ofChemistry
University ofBristol
Cantock’s Close
Bristol, BS81TS (UK)
Fax: (+44)117-929-8611
E-mail: r.j.cox@bris.ac.uk
[**] This work was supported by the BBSRC (7/B17965) and The School
ofChemistry at the University ofBristol (U.W.).
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
alcohol
7
was protected with tert-butyldimethylsilyl
4926
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 4926 –4929

Angewandte
Chemie
(TBDMS), then deprotonated and treated with acetaldehyde
(Scheme 2). The resulting alcohol 9 was treated with RedAl-
D followed by a workup with H2O to give deuterated olefin
10a.
We and others have previously determined reaction
conditions for the kinetic assay of DXR.[2–4,12] DXR requires
a metal ion, usually Mg2+ or Mn2+ ions. In the presence of the
Mn2+ ion (1 mm), DXP, and DXR (40 nm), NADPH (0.4 mm)
is consumed and the reaction can be followed spectrophoto-
metrically at 340 nm. By using this assay, the kinetic
parameters of the reaction catalyzed by DXR were measured
for the three substrates (Table 1). Comparison of the (VH/
Table 1: Kinetic parameters measured for E. coli DXR.
Substrate
KM [mm]
Vmax [ODminÀ1 [a]
]
Vmax
KM
/
kH/
kD
3
15a
15b
3.10”10À1 Æ4.6”10À2 7.38”10À3 Æ3.0”10À4 0.024 1.0
5.37”10À1 Æ4.8”10À2 1.32”10À2 Æ5.0”10À4 0.026 0.92
4.23”10À1 Æ3.0”10À2 1.19”10À2 Æ3.2”10À4 0.028 0.86
[a] OD=optical density.
KMH)/(VD/KMD) values for the two 2H-labeled substrates show
an inverse isotope effect for both substrate isotopomers. This
À
must be a secondary KIE as C H bonds remain intact
throughout the reaction.
The two proposed rearrangement mechanisms suggested
for DXR differ in terms of bond-breaking, bond-making, and
rehybridization events (Scheme 1). Secondary kinetic isotope
effects are particularly useful for examining changes in
hybridization at a carbon atom.[14] In the case of the a-ketol
rearrangement (mechanism A, Scheme 1), C2 and C3 rehy-
bridize during the reaction, whereas C4 remains unchanged.
For the retro-aldol/aldol reaction (mechanism B, Scheme 1),
C3 and C4 both rehybridize. The observation of kinetic
isotope effects for substrates bearing 2H at C3 and C4 strongly
supports the operation of mechanism B involving rehybrid-
ization at both secondary alcohol carbon atoms during
reaction. These results are not consistent with the a-ketol
mechanism, which would be expected to show a KIE for the 3-
[2H] substrate but not the 4-[2H] substrate.
Scheme 2. Synthesis ofstereospeciifcally 2H-labeled substrates.
a) TBDMSCl, imidazole, CH2Cl2, 64%; b) EtMgBr, then CH3CHO, 85%;
c) LiAlD4, MeOCH2CH2OH, toluene then H2O workup, 81% or LiAlH4,
THF, then D2O workup, 79%; d) TBAF, 08C, a 83%, b 44%; e) P-
(OBn)3, I2, a 71%, b 44%; f) Dess–Martin periodinane, CH2Cl2, RT, a
70%, b 95%; g) OsO4, CH2Cl2, À408C, alkaloid, then MeOH, HCl, a
90%, b 91%; h) H2, Pd/C, MeOH, RT, a 95%, b 91%. TBAF=tetra-n-
butylammonium fluoride.
2
The observation of kinetic isotope effects for both H-
Treatment of the alcohol 9 with LiAlH4, followed by D2O
workup gave the isomeric deuterated olefin 10b. Deprotec-
tion of the primary alcohols gave the diols 11a and 11b, and
selective phosphorylation afforded the alcohols 12a and 12b.
Dess–Martin oxidation then gave the ketones 13a and 13b
and asymmetric dihydroxylation, by using stoichiometric
OsO4 according to our previously published procedures,[12]
gave the protected DXR isotopomers 14a and 14b. Depro-
tection was achieved by hydrogenolysis to afford the substrate
isotopomers 15a and 15b. 13C NMR spectroscopic analysis
indicated that 15a incorporated greater than 98% 2H,
labeled substrates rules out the possibility that the reduction
step (Scheme 3, III) is the rate-limiting step (RLS) of the
DXR-catalyzed reaction—it would be highly unlikely that 4-
[2H] could exert a kinetic influence over this reaction
(although 2H at C3 would be expected to exert such an
influence if the reduction were the RLS). The RLS must
therefore be one of the preceding steps. An inverse KIE is
indicative of a RLS involving a rehybridization of carbon
from sp2 to sp3, thus it is likely that it is the recombination step
(Scheme 3, step II) that is rate limiting for DXR.
These results are supported by other observations from
the literature. Oefner and co-workers recently reported the
X-ray crystal structure of E. coli DXR in complex with
different substrates.[6] To prevent turnover (i.e. formation of
MEP 4), divalent metal ions were omitted from the crystal-
lization. One structure, obtained in the presence of enantio-
pure d-DXP 3 and NADPH, showed the presence of both
DXP 3 and its 4-epimer l-1-deoxyribulose-5-phosphate 16 in
the active site. Oefner and co-workers were not able to
explain this observation. We investigated this reaction by
incubating 3 with DXR in the absence of both NADPH and
2
2
whereas 15b incorporated greater than 92% H. No H was
observed at the undesired position in either case.
Previous investigations have indicated that the dihydrox-
ylation step affords the diol with an e.r. value of 92:8.[12,13] In
this case, the e.r. ratio was determined by reacting the
deprotected 2H substrates with NADPH (catalyzed by DXR).
Because l-DXP is not a substrate, the extent of reaction,
found by measuring the final consumption of NADPH,
indicates the proportion of d-DXP. The e.r. value of 15a
was 92:8, whereas the e.r. value of 15b was 92.5:7.5.
Angew. Chem. Int. Ed. 2007, 46, 4926 –4929
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 4927

Communications
Scheme 3. Detailed consideration ofthe retro-aldol/aldol rearrangement.
metal ions in D2O/Tris-HCl buffer solution (Tris = tris(hy-
droxymethyl)aminomethane) in an NMR tube for 12 h.
Under these conditions, a low concentration of 16 (approx-
imately 5%) was detected by 13C NMR—in particular C4 of
16 was observed at d = 71.6ppm, showing the characteristic
3JCP coupling.[15] This compound was not formed in the
absence of purified enzyme. The formation of 16 is further
evidence in favor of the proposed retro-aldol/aldol mecha-
nistic sequence as this compound could not be formed by an
a-ketol-type process. Diastereomer 16 has previously been
reported as a weak competitive inhibitor of DXR from
Synechocystis PCC6803, indicating that it can bind at the
active site. However, isomerization assays in the absence of
divalent metal ions were not reported for this compound.[15]
The mechanism presented in Scheme 3 rationalizes the
formation of 16. In the absence of Mg2+ or Mn2+ ions, the
retro-aldol step (Scheme 3, reaction I) is still operative, giving
aldehyde and enediol intermediates in the active site. In the
absence of Mg2+ or Mn2+ ions, the aldehyde can rotate by 1808
and recombine with the enediol (Scheme 3, reaction I’),
slowly forming 16. Presumably, the presence of the divalent
metal ion during normal catalysis prevents aldehyde rotation
and formation of 16. However, metal ions are evidently not
required for reactions I and I’. There was also no evidence for
the formation of 6 or 17 in the active site, so either Mg2+ ions
are required for steps II and II’, or the concentration of 6 and
17 is too low to detect. It appears unlikely that Mg2+ ions
would be required for step II but not step I, so it is probable
that the equilibrium favors compounds 3 and 16 in the
absence of Mg2+ ions. This is supported by DFT calculations,
which show 3 as being intrinsically 5.05 kcalmolÀ1 lower in
energy than 6 (B3LYP/6-31G(d)*//MMFF). It is known that
Mg2+ or Mn2+ ions are required for step III, presumably for
coordination to the carbonyl group, facilitating hydride
transfer from NADPH to the Re face of the aldehyde.[10,16]
The enzyme ketolacid reductoisomerase (KARI,
EC 1.1.1.86) catalyzes a similar sequence of reactions to
DXR.[17] In this case, the substrate is (S)-2-acetolactate 18,
which is isomerized to 3-hydroxy-3-methyl-2-ketobutyrate 19,
which is in turn reduced by NADPH to give (R)-2,3-
dihydroxy-3-methylbutyrate 20 (Scheme 4). The branched
product 20 is a precursor of l-valine. Recent mechanistic and
crystallographic investigations of KARI have revealed that
two Mg2+ ions are required at the active site.[18] The
mechanism of the rearrangement catalyzed by KARI is
generally accepted as being a concerted a-ketol reaction as
Scheme 4. Reactions catalyzed by KARI in the formation of l-valine.
there is no possibility of a retro-aldol/aldol sequence. An
Mg2+ ion is required for this isomerization step. The
equilibrium between 18 and 19 lies heavily in favor of 18,
and this is why KARI must combine the isomerization and
reduction in a single active site.
Both the mechanism of KARI and its requirement for
Mg2+ ions differ from DXR, which appears not to require
metal catalysis for the retro-aldol/aldol rearrangement. How-
ever, KARI is similar to DXR in requiring Mg2+ ions for the
reduction step. Peptide-sequence analysis shows that there is
very limited sequence similarity between DXR and KARI.
For example, there is only 10% sequence identity between
DXR and KARI from E. coli, and no sequence conservation
at the active sites. Thus the resemblance between DXR and
KARI appears to be superficial, reflecting neither mecha-
nism, active-site structure, or peptide sequence.
The mechanism proposed for DXR shown in Scheme 3
closely resembles the retro-aldol/aldol mechanism of l-
ribulose-5-phosphate-4-epimerase.[19] This enzyme intercon-
verts l-ribulose-5-phosphate and d-xylulose-5-phosphate, but
requires a divalent metal ion. DXR appears to catalyze a very
similar reaction, albeit slowly, in the absence of a metal.
However, DXR and l-ribulose-5-phosphate-4-epimerase
from E. coli share only 11% sequence identity.
In conclusion, the observation of inverse secondary
kinetic isotope effects for both 3-[2H] and 4-[2H]-d-1-deoxy-
xylulose-5-phosphate during DXR-catalyzed isomerization
indicates that DXR operates through a retro-aldol/aldol
mechanism. This is supported by the observed formation of l-
deoxyribulose-5-phosphate 16 in the absence of metal ions.
Thus, the isomerization reaction differs from the enzyme
KARI, which uses an a-ketol mechanism. The isomerization
equilibrium catalyzed by DXR strongly favors the starting
material 3 over the branched aldehyde product 6, and the
4928 www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 4926 –4929

Angewandte
Chemie
[6] A. M. Sweeney, R. Lange, R. P. M. Fernandes, H. Schulz, G. E.
formation of 6 appears to be the rate-limiting step. The
reaction is driven to completion, however, by coupling to the
oxidation of NADPH. In the case of KARI, knowledge of its
chemical mechanism has led to the understanding of inhib-
ition by compounds based on transition-state structures.[20]
The determination of the mechanism of DXR should
contribute in a similar way.
Dale, A. Douangamath, P. J. Proteau, C. Oefner, J. Mol. Biol.
2005, 345, 115 – 127.
[7] S. Yajima, T. Nonaka, T. Kuzuyama, H. Seto, K. Ohsawa, J.
Biochem. 2002, 131, 313 – 317.
[8] D. T Fox, C. D. Poulter, Biochemistry 2005, 44, 8360 – 8368.
[9] A. Wong, J. W. Munos, V. Devasthali, K. A. Johnson, H. Lui,
Org. Lett. 2004, 6, 3625 – 3628.
[10] J-F Hoeffler, D. Tritsch, C. Grosdemange-Billiard, M. Rohmer,
Eur. J. Biochem. 2002, 269, 4446– 4457.
[11] A. Argyrou, J. S. Blanchard, Biochemistry 2004, 43, 4375 – 4384.
[12] R. J. Cox, A. de Andres-Gomez, C. R. A. Godfrey, Org. Biomol.
Chem. 2003, 1, 3173 – 3177.
Received: February 12, 2007
Revised: April 3, 2007
Published online: May 22, 2007
[13] R. J. Cox, A. E. Evitt, Org. Biomol. Chem. 2007, 5, 229 – 232.
[14] B. K. Carpenter, Determination of Organic Reaction Mecha-
nisms, Wiley, Chichester, 1984.
[15] C. Phaosiri, P. J. Proteau, Bioorg. Med. Chem. Lett. 2004, 14,
5309 – 5312.
Keywords: enzymes · isotopic labeling · kinetic isotope effects ·
reaction mechanisms
.
[16] P. J. Proteau, Y-H. Woo, R. T. Williamson, C. Phaosiri, Org. Lett.
1999, 1, 921 – 923.
[17] R. Dumas, V. Biou, F. Halgand, R. Douce, R. G. Duggleby, Acc.
Chem. Res. 2001, 34, 399 – 408.
[1] M. Rohmer, Nat. Prod. Rep. 1999, 16, 565 – 574.
[2] S. Takahashi, T. Kuzuyama, H. Watanabe, H. Seto, Proc. Natl.
Acad. Sci. USA 1998, 95, 9879 – 9884.
[3] L. MerklØ, A. de Andres-Gomez, B. Dick, R. J. Cox, C. R. A.
Godfrey, ChemBioChem 2005, 6, 1866 – 1874.
[18] R. Tyagi, Y-T Lee, L. W. Guddat, R. G. Duggleby, FEBS J. 2005,
272, 593 – 602.
[4] A. T. Koppisch, D. T. Fox, B. S. Blagg, C. D. Poulter, Biochem-
istry 2002, 41, 236– 243.
[19] L. V. Lee, M. V. Vu, W. W. Cleland, Biochemistry 2000, 39, 4808 –
4820.
[5] S. Steinbacher, J. Kaiser, W. Eisenreich, R. Huber, A. Bacher, F.
Rohdich, J. Biol. Chem. 2003, 278, 18401 – 18407.
[20] A. Schulz, P. Sponemann, H. Kocher, F. Wengenmayer, FEBS
Lett. 1988, 238, 375 – 378.
Angew. Chem. Int. Ed. 2007, 46, 4926 –4929
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 4929