Isoprenoid biosynthesis: Ferraoxetane or allyl anion mechanism for IspH catalysis?

Article abstract of DOI:10.1002/anie.201302343

The catalytic mechanism of the enzyme IspH, involved in formation of isopentenyl diphosphate and dimethylallyl diphosphate, was investigated by using HYSCORE spectroscopy combined with DFT. The results indicate the formation of an allyl anion bound to a HiPIP-like oxidized 4Fe-4S cluster, rather than formation of a cyclic, ferraoxetane intermediate, as has been proposed elsewhere. Copyright

Full text of DOI:10.1002/anie.201302343

Angewandte
Chemie
DOI: 10.1002/anie.201302343
Organometallobiochemistry
Isoprenoid Biosynthesis: Ferraoxetane or Allyl Anion Mechanism for
IspH Catalysis?**
Jikun Li, Ke Wang, Tatyana I. Smirnova, Rahul L. Khade, Yong Zhang, and Eric Oldfield*
There are around 65000 terpenes known.[1] These molecules
are produced from two isoprenoid (C5) diphosphates: iso-
pentenyl diphosphate (IPP, 1) and dimethylallyl diphosphate
(DMAPP, 2).[2] In plant plastids, IPP and DMAPP are
produced primarily through the 2-C-methylerythritol 4-phos-
phate (MEP) pathway from E-1-hydroxy-2-methyl-but-2-enyl
4-diphosphate (HMBPP, 3) in a reaction catalyzed by the
enzyme IspH, and this route is also used in most bacteria, as
well as in malaria parasites (Scheme 1).
cationic, anionic, radical, butadiene as well as organometallic
hypotheses being proposed.[4] Most of these proposals have
now been ruled out, but there is considerable evidence in
support of the organometallic hypothesis.[5] In this hypothesis,
it is proposed that HMBPP initially binds to the oxidized
([Fe4S4]2+) cluster through O1 (in the CH2OH group;
Scheme 2a). On cluster reduction, the CH2OH group rotates
À
about C1 C2 and moves away from the cluster to interact
with the highly conserved E126 (E. coli residue numbering),
Scheme 1. IspH products and substrate.
Scheme 2. Proposed allyl anion mechanism for IspH catalysis.
There is thus interest in understanding the mechanism of
action of IspH since this could lead to new inhibitors of
interest as anti-infective drug leads. IspH is a 4Fe–4S cluster-
containing protein and catalyzes the deoxygenation of
HMBPP through a 2H+/2eÀ reduction.[3] There has been
considerable debate over the IspH mechanism of action with
and a diphosphate OH group, which can provide the H+
required for dehydroxylation, and subsequent ligand proto-
nation. The intermediate that forms is stabilized by a p
interaction between the alkene group and the cluster
(Scheme 2b). An internal 2eÀ transfer then ensues, resulting
in a “super-oxidized” or HiPIP (high-potential iron–sulfur
[*] Dr. K. Wang,[+] Prof. Dr. E. Oldfield
Department of Chemistry
protein, [Fe4S4]3+)-like S = 1= cluster and an allyl anion
University of Illinois at Urbana-Champaign
600 South Mathews Avenue, Urbana, IL 61801 (USA)
E-mail: eoldfiel@illinois.edu
J. Li[+]
2
(Scheme 2c) which after addition of a second electron and
H+ yields the IPP/DMAPP products and the original oxidized
cluster (Scheme 2d).
Center for Biophysics and Computational Biology
University of Illinois at Urbana-Champaign
607 South Mathews Avenue, Urbana, IL 61801 (USA)
More recently, a second organometallic mechanism has
been proposed[6] for IspH in which another species, a ferraox-
etane, is involved. This intermediate is analogous to the
ferraoxetane intermediate we previously proposed for IspG
Prof. Dr. T. I. Smirnova
Department of Chemistry, North Carolina State University
Raleigh, NC 27695 (USA)
catalysis[7,8] and would contain Fe O as well as Fe C bonds.
In addition, these workers have proposed another alternative,
oxa-allyl, model for IspG catalysis.[9] In essence, then, there
are two sets of models: the allyl anion model for IspH[4,5]
together with the ferraoxetane model for IspG[7,8] that we
have proposed (Scheme 3a,b), and the ferraoxetane model
for IspH together with the oxa-allyl model for IspG[9]
(Scheme 3c,d). What is not clear from the recently proposed
IspH ferraoxetane mechanism[6] (Scheme 3c) is how DMAPP
could form, and how this mechanism can be reconciled with
the observation that the CH2OH group rotates away from the
cluster, as deduced from both crystallographic and isotope-
labeling experiments.[10,11] We have thus investigated the
chemical nature of the newly reported IspH reaction inter-
mediate[6] in more detail by using HYSCORE[12] spectroscopy
with 2H, 13C and 17O-labeled substrates and one-electron
À
À
R. L. Khade, Prof. Dr. Y. Zhang
Department of Chemistry, Chemical Biology, and Biomedical
Engineering, Stevens Institute of Technology
Castle Point on Hudson, Hoboken NJ 07030 (USA)
[+] These authors contributed equally to this work.
[**] This work was supported by NIH grant (GM065307) to E.O., NIH
grant (GM085774) to Y.Z., NSF grant MCB-0843632 to T.I.S. and
equipment grants from NIH (S10RR023614), NSF (CHE-0840501),
and the North Carolina Biotechnology Center (NCBC 2009-IDG-
1015). J.L. was supported by a pre-doctoral fellowship from the
American Heart Association, Midwest Affiliate (11PRE7500042). We
thank Drs. Hassan Jomaa and Jochen Wiesner for providing the A.
aeolicus IspH expression system and Mark J. Nilges for assistance
with the EPR spectroscopy.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201302343.
Angew. Chem. Int. Ed. 2013, 52, 1 – 5
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

.
Angewandte
Communications
unique 4th Fe in the 4Fe–4S cluster.[6] The EPR spectrum was
essentially the same as that we reported previously for
a sample prepared from E. coli IspH in the presence of excess
dithionite and HMBPP, having g11 = 2.17, g22 = 2.01, and g33 =
1.99 (Figure 1a). The HYSCORE spectrum is shown in
Figure 1b and there is no evidence for any 17O signal. These
new results, therefore, do not provide support for the IspH
ferraoxetane hypothesis but are consistent with formation of
an allyl anion complex (in which the 17O has been lost).
Next, we used a uniformly 13C-labeled HMBPP to
produce the g11 = 2.17 species. There was no evidence for
a large (ca. 17 MHz) 13C hyperfine coupling (Figure 1c and
Figure S1 in the Supporting Information). There were, how-
ever, three sets of small hyperfine couplings (denoted C(A),
C(B), and C(C) in Figure 1c). To assign these resonances, we
synthesized [1-13C]-, [2-13C]- and [3-13C]-HMBPPs, and
obtained HYSCORE spectra of the g11 = 2.17 intermediates.
The hyperfine couplings could be simulated using EasySpin[17]
with Aii(A) = [2.0, 0.5, 6.7] MHz; Aii(B) = [À1.1, À0.8,
7.3] MHz, and Aii(C) = [7.3, 0.9, 0.9] MHz (Figures S2 and
S3), corresponding to Aiso = 3.1, 1.8, and 3.0 MHz, respec-
tively. As can be seen in Figure 1d–f, the HYSCORE
spectrum of the [1-13C]-HMBPP labeled sample corresponds
to the C(C) (Aiso ꢀ 3.0 MHz) signal, the [2-13C] sample
corresponds to C(A) (Aiso ꢀ 3.1 MHz), and the spectrum of
the [3-13C]-HMBPP labeled sample corresponds to C(B) (Aiso
ꢀ 1.8 MHz).
Scheme 3. Proposed structures of reaction intermediates involved in
IspH and IspG catalysis. The structures we have proposed are a) for
IspH and b) for IspG;[4,5,7,8] the structures from alternative proposals
are c) for IspH and d) for IspG.[6,9]
reduced Aquifex aeolicus IspH, which yields an intense EPR
spectrum characterized by g11 = 2.17.[6]
With IspG, we previously reported the results of HYS-
2
CORE and ENDOR[13] experiments using nine H, 13C and
17O-labeled substrates that produced the intermediate “X”,
proposed to be the ferraoxetane 4 (Scheme 3b). There were
three key observations that led to this structural proposal:
First, the HYSCORE spectrum of “X” prepared using an 17O-
labeled substrate exhibited a large (ca. 8 MHz) isotropic
hyperfine interaction. This coupling is very similar to those
seen with H217O bound to the unique, 4th Fe in the 4Fe–4S
cluster in aconitase (8.65 MHz), and most hyperfine couplings
2
Next, we investigated the HYSCORE spectra of two H-
À
for systems containing Fe O bonds are in the range 8–
labeled HMBPPs: [2’-2H3]-HMBPP and [4-2H2]-HMBPP,
bound to IspH (as the g11 = 2.17 reaction intermediate). In
previous work[5] we obtained HYSCORE spectra of (Æ)-[1-
2H1]- and [3-2H]-HMBPP complexed with reduced EcIspH
(the same g11 = 2.17 reaction intermediate) under turnover
15 MHz.[14–16] Second, there was a ca. 17 MHz hyperfine
coupling observed for the quaternary carbon (C2), while all
other carbons had much smaller hyperfine couplings
[7,8]
À
(< 4 MHz), indicating formation of a Fe C2 bond.
Third,
the large (ca. 12 MHz) 1H hyperfine coupling observed in the
absence of any isotopic labeling was shown to arise from
a single hydrogen in the C2’ methyl group[8] since, using
conditions. Both exhibited clear H hyperfine couplings with
2
Aiso ꢀ 0.9 MHz and 0.5 MHz, respectively, for the (Æ)-[1-2H1]-
and [3-2H]-HMBPP ligands (Figure S4a, b) The HYSCORE
spectra of one-electron reduced A. aeolicus IspH samples
(again, characterized by identical gii values as samples
prepared by freeze-quenching) with [2’-2H3]-HMBPP and
[4-2H2]-HMBPP are shown in Figure S4c, d. In both cases, 2H
hyperfine couplings are present, but they are clearly very
small, inconsistent with the expectation that one deuteron in
the methyl group in a ferraoxetane would have a large
hyperfine coupling, as seen in IspG. The hyperfine couplings
seen in IspG[7,8] are shown on the ferraoxetane structure in
Figure 2a; the new experimental results for IspH are shown
on the IspH ferraoxetane model in Figure 2b, and the
experimental results for IspH are shown again on the allyl
anion model, in Figure 2c. In the IspG ferraoxetane (Fig-
ure 2a), both atoms bonded to Fe (O3, C2) have large
hyperfine couplings. Similar couplings would be expected for
an IspH ferroxetane but there is actually no 17O signal
observed, and the C2 hyperfine coupling tensor is Aii = [2.0,
0.5, 6.7] MHz, to be compared with Aii = [14.5, 12.0,
26.5] MHz in IspG.[8,9] These results support the allyl anion
p-complex model (Figure 2c) in which there is no 17O
hyperfine coupling because the 17O has already been
removed, as water, and there are three small 13C hyperfine
couplings (to C1, C2 and C3 in an allyl anion). These 17O and
a
CD3-labeled substrate, we found three signals, one
(Aiso(2H) = 1.7 MHz) corresponding to the Aiso(1H)
ꢀ 12 MHz signal. These isotope-labeling experiments (and
the results of DFT calculations) suggested that “X” was the
ferraoxetane 4.[7]
The structure of 4 (Scheme 3b) is clearly similar to that
now proposed for the g11 = 2.17 IspH reaction intermediate[6]
(Scheme 3c), the exception being of course that there are
differences in the positions of the H/alkyl groups (since the
substrates are different). It seems implausible that these
would have major effects on electronic structure, in which
case two predictions for the IspH ferraoxetane model would
be: 1) a large 1-17O hyperfine coupling and 2) a large 2-13C
hyperfine coupling. With the allyl anion model we proposed
previously,[4,5] the predictions would be: 1) no hyperfine
interaction with 17O (since it is not present), and 2) there
should be three 13C hyperfine couplings, one to each of the
three carbons in the allyl anion.
We first obtained EPR and HYSCORE spectra of
reconstituted, one-electron reduced Aquifex aeolicus IspH
(AaIspH) with [1-17O]-HMBPP (that is, labeled at the 1,
CH2OH position), since in the recently proposed IspH
ferraoxetane model this group is directly bonded to the
2
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1 – 5
These are not the final page numbers!

Angewandte
Chemie
Figure 1. a) 9.05 GHz continuous-wave EPR spectrum of one-electron reduced A. aeolicus IspH + HMBPP. b) 9.68 GHz HYSCORE spectrum of
one-electron reduced AaIspH + [1-17O]-HMBPP, sum of spectra with t=128, 182 and 236 ns, B0 =344.4 mT. c) 9.68 GHz HYSCORE spectrum of
one-electron reduced AaIspH + [U-13C5]-HMBPP, sum spectra with t=128, 150, 166, 182, 196, and 208 ns, B0 =344.7 mT. d) 9.69 GHz HYSCORE
spectrum of one-electron reduced AaIspH + [1-13C]-HMBPP. e) 9.69 GHz HYSCORE spectrum of one-electron reduced AaIspH + [2-13C]-HMBPP.
f) 9.69 GHz HYSCORE spectrum of one-electron reduced AaIspH + [3-13C]-HMBPP. d–f) are all sum spectra with t=136, 160, and 196 ns,
B0 =344.1 mT. All spectra were taken at 10 K.
Figure 2. Hyperfine couplings (Aiso, MHz) observed for reaction intermediates in IspH and IspG catalysis based on HYSCORE and/or ENDOR
spectra with isotopically labeled substrates.
13C results are also supported by current and previous H
HYSCORE/ENDOR spectroscopic results. With the IspG
ferraoxetane (Figure 2a), there is one very large 2H/1H
hyperfine coupling, due to a single proton in the CD3/CH3
group. There is no corresponding feature in the IspH reaction
intermediate that would be expected from the structure
shown in Figure 2b, and the couplings that are seen are larger
for H1, H3 (in the allyl group) than they are for H2’ and H4.
The lack of any large 2H, 13C or 17O hyperfine couplings,
together with the three observed 13C HYSCORE signals, are
all supportive of an allyl anion IspH intermediate[5] and not
a ferraoxetane.[6]
2
The question then arises as to whether the g11 = 2.17
species corresponds to the intermediate observed crystallo-
graphically.[18,19] In the latter the average distance from Fe to
C1, C2 and C3 is 2.8 Æ 0.3 ꢀ. Using the hyperfine tensor
results and the point dipole approximation with the spin
projection coefficients for aconitase,[20] we obtained a range of
Angew. Chem. Int. Ed. 2013, 52, 1 – 5
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

.
Angewandte
Communications
2.0(Æ0.2) to 2.5(Æ0.2) ꢀ. To try to obtain more accurate
results, we next investigated the allyl anion/Fe-S cluster
structure using DFT. We used a [Fe4S4(SMe)3(h3-CH2-C-
(CH3)-CH(CH2OH)] cluster with S = 1/2, with methods
reported previously,[21–23] using the Gaussian09 program.[24]
After full geometry optimization (Figure 3a), the distances
bond rotation in the allyl anion model that would not occur
with the ferraoxetane mechanism, plus, the allyl anion
mechanism provides a route to DMAPP formation that is
absent in the ferraoxetane mechanism.
Received: March 20, 2013
Published online: && &&, &&&&
Keywords: biosynthesis · enzyme catalysis · iron–sulfur cluster ·
.
metalloproteins · terpenes
[1] J. Buckingham, Dictionary of Natural Products on DVD, CRC,
Boca Raton, 2011.
[2] E. Oldfield, F.-Y. Lin, Angew. Chem. 2012, 124, 1150 – 1163;
Angew. Chem. Int. Ed. 2012, 51, 1124 – 1137.
[3] M. Wolff, M. Seemann, B. Tse Sum Bui, Y. Frapart, D. Tritsch, A.
Garcia Estrabot, M. Rodrꢁguez-Concepciꢂn, A. Boronat, A.
Marquet, M. Rohmer, FEBS Lett. 2003, 541, 115 – 120.
[4] W. Wang, K. Wang, Y.-L. Liu, J.-H. No, J. Li, M. J. Nilges, E.
Oldfield, Proc. Natl. Acad. Sci. USA 2010, 107, 4522 – 4527.
[5] W. Wang, K. Wang, I. Span, J. Jauch, A. Bacher, M. Groll, E.
Oldfield, J. Am. Chem. Soc. 2012, 134, 11225 – 11234.
[6] W. Xu, N. S. Lees, D. Hall, D. Welideniya, B. M. Hoffman, E. C.
Duin, Biochemistry 2012, 51, 4835 – 4849.
Figure 3. IspH intermediate structures and mechanism. a) The model
used to compute hyperfine tensors. b) Reduction and protonation of
the [Fe4S4]3+-allyl anion results in product formation with a proposed
weak p-interaction between the double bond in the IPP product and
the 4th Fe.
[7] W. Wang, J. Li, K. Wang, C. Huang, Y. Zhang, E. Oldfield, Proc.
Natl. Acad. Sci. USA 2010, 107, 11189 – 11193.
[8] W. Wang, K. Wang, J. Li, S. Nellutla, T. I. Smirnova, E. Oldfield,
J. Am. Chem. Soc. 2011, 133, 8400 – 8403.
[9] W. Xu, N. S. Lees, D. Adedeji, J. Wiesner, H. Jomaa, B. M.
Hoffman, E. C. Duin, J. Am. Chem. Soc. 2010, 132, 14509 –
14520.
[10] I. Span, T. Grꢃwert, A. Bacher, W. Eisenreich, M. Groll, J. Mol.
Biol. 2012, 416, 1 – 9.
[11] C. A. Citron, N. L. Brock, P. Rabe, J. S. Dickschat, Angew. Chem.
2012, 124, 4129 – 4133; Angew. Chem. Int. Ed. 2012, 51, 4053 –
4057.
[12] P. Hçfer, A. Grupp, H. Nebenfꢄhr, M. Mehring, Chem. Phys.
Lett. 1986, 132, 279 – 282.
between the unique 4th Fe and C1, C2, and C3 were 2.14, 2.10,
and 2.17 ꢀ, respectively. The computed Aiso values were in
quite good accord with experiment: Aiso(C1) = 3.0 MHz
(expt.), 1.91 MHz (calc.); Aiso(C2) = 3.1 MHz (expt.),
2.87 MHz (calc.); Aiso (C3) = 1.8 MHz (expt.), 1.70 MHz
(calc.) (Table S1 and Figure S5). These results all indicate
that the g11 = 2.17 species is the [Fe4S4]3+-like/allyl anion
species III[5] (Figure 3b, left) while the species observed
crystallographically is likely a weakly bound product that
forms through IV (Figure 3b, middle), based on the ca. 2.8 ꢀ
bond lengths seen in the X-ray structure (Figure 3b, right).
In summary: the results reported here are of interest for
several reasons. First, using a [1-17O] HMBPP substrate we
[13] E. R. Davies, Phys. Lett. A 1974, 47, 1 – 2.
[14] M. C. Kennedy, M. Werst, J. Telser, M. H. Emptage, H. Beinert,
B. M. Hoffman, Proc. Natl. Acad. Sci. USA 1987, 84, 8854 – 8858.
[15] M. M. Werst, M. C. Kennedy, H. Beinert, B. M. Hoffman,
Biochemistry 1990, 29, 10526 – 10532.
[16] J. Telser, M. H. Emptage, H. Merkle, M. C. Kennedy, H. Beinert,
B. M. Hoffman, J. Biol. Chem. 1986, 261, 4840 – 4846.
[17] S. Stoll, A. Schweiger, J. Magn. Reson. 2006, 178, 42 – 55.
[18] T. Grꢃwert, I. Span, W. Eisenreich, F. Rohdich, J. Eppinger, A.
Bacher, M. Groll, Proc. Natl. Acad. Sci. USA 2010, 107, 1077 –
1081.
[19] T. Grꢃwert, I. Span, A. Bacher, M. Groll, Angew. Chem. 2010,
122, 8984 – 8991; Angew. Chem. Int. Ed. 2010, 49, 8802 – 8809.
[20] C. J. Walsby, W. Hong, W. E. Broderick, J. Cheek, D. Ortillo, J. B.
Broderick, B. M. Hoffman, J. Am. Chem. Soc. 2002, 124, 3143 –
3151.
[21] Y. Zhang, E. Oldfield, J. Am. Chem. Soc. 2004, 126, 4470 – 4471.
[22] K. Wang, W. Wang, J.-H. No, Y. Zhang, Y. Zhang, E. Oldfield, J.
Am. Chem. Soc. 2010, 132, 6719 – 6727.
À
find no evidence for Fe O bonding in an IspH reaction
intermediate prepared by using a one-electron reduced
cluster, making
a ferraoxetane intermediate unlikely.
Second, using a uniformly 13C-labeled HMBPP substrate, we
find no evidence for the large (ca. 17 MHz) hyperfine
interaction seen previously with IspG and attributed there
to a ferraoxetane intermediate. Rather, we detected three
signals, all with quite small hyperfine couplings. Third, using
[1-13C], [2-13C] and [3-13C]-labeled HMBPP substrates, we
were able to specifically assign all three signals. Fourth, the
13C hyperfine couplings observed were in good accord with
DFT predictions. Fifth, we find no evidence for the large
hyperfine interaction seen previously with the IspG (ferraox-
etane) reaction intermediate due to a single proton in the C2’
CH3 group. Overall, the results are of broad general interest
since they help clarify the nature of the g11 = 2.17 reaction
intermediate observed in IspH catalysis, highlighting the
unusual, organometallic mechanism of the reaction. More-
over, the results are consistent with both crystallographic[10]
[23] Y.-L. Liu, F. Guerra, K. Wang, W. Wang, J. Li, C. Huang, W. Zhu,
K. Houlihan, Z. Li, Y. Zhang, S. K. Nair, E. Oldfield, Proc. Natl.
Acad. Sci. USA 2012, 109, 8558 – 8563.
[24] M. J. Frisch et al, Gaussian09 RevisionB.01, Gaussian, Inc.,
Wallingford, CT, 2010.
and isotope-labeling[11] experiments which indicate C1 C2
À
4
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1 – 5
These are not the final page numbers!

Angewandte
Chemie
Communications
Organometallobiochemistry
The catalytic mechanism of the enzyme
IspH, involved in formation of iso-
pentenyl diphosphate and dimethylallyl
diphosphate, was investigated by using
HYSCORE spectroscopy combined with
DFT. The results indicate the formation of
an allyl anion bound to a HiPIP-like
oxidized 4Fe–4S cluster, rather than for-
mation of a cyclic, ferraoxetane inter-
mediate, as has been proposed else-
where.
J. Li, K. Wang, T. I. Smirnova, R. L. Khade,
Y. Zhang, E. Oldfield*
&&&& — &&&&
Isoprenoid Biosynthesis: Ferraoxetane or
Allyl Anion Mechanism for IspH
Catalysis?
Angew. Chem. Int. Ed. 2013, 52, 1 – 5
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!