An alternative isovaleryl CoA biosynthetic pathway involving a previously unknown 3-methylglutaconyl CoA decarboxylase

Article abstract of DOI:10.1002/anie.201207984

Take a detour: An alternative pathway to synthesize isovaleryl coenzyme A (CoA) has recently been suggested in myxobacteria, which is highly active when leucine is limited. Each enzymatic step of this unprecedented route has now been characterized and a novel 3-methylglutaconyl CoA decarboxylase identified that has apparently evolved from CoA transferases. Copyright

Full text of DOI:10.1002/anie.201207984

.
Angewandte
Communications
DOI: 10.1002/anie.201207984
Biosynthesis
An Alternative Isovaleryl CoA Biosynthetic Pathway Involving
a Previously Unknown 3-Methylglutaconyl CoA Decarboxylase**
Yanyan Li, Eva Luxenburger, and Rolf Mꢀller*
[
3,15]
Myxobacteria are swarming bacteria characterized by their
social behavior and ability to undergo a complex develop-
the hypothetical IV-CoA and the mevalonate pathway.
3-
Hydroxy-3-methylglutaryl CoA (HMG-CoA) was proposed
to be the branching point, from which reverse reactions of
leucine degradation would happen and lead to IV-CoA
(Figure 1). Feeding studies and analysis of a double-mutant
bkdÀ/mvaSÀ (mvaS encodes HMG-CoA synthase) demon-
strated the involvement of HMG-CoA and the intermediacy
of 3,3-dimethylacrylyl CoA (DMA-CoA; as shown by incor-
poration of labeled 3,3-dimethylacrylic acid, which needs to
[1]
mental life cycle. They prey on other bacteria or fungi for
food. Under starvation conditions, the vegetative cells
aggregate to form multicellular fruiting bodies. To support
such a remarkable lifestyle, myxobacteria exhibit an enor-
mous metabolic potential, which is mirrored by discoveries of
new compounds and unusual biochemical pathways in recent
[2]
years. One example is the biosynthesis of isovaleryl
coenzyme A (IV-CoA) for which an additional alternative
undergo activation to the CoA ester) in the alternative
[
3]
[7,15]
pathway has been recently suggested. IV-CoA is generally
derived from leucine degradation by transamination and
subsequent oxidative decarboxylation by the branched-chain
pathway.
Importantly, this pathway was found to be highly
active in the bkdÀ mutant and during fruiting body formation
[7]
when leucine-derived IV-CoA is limited, reflecting essential
roles of IV-CoA-derived compounds in the life cycle of
myxobacteria.
[
4]
a-keto acid dehydrogenase complex (Bkd). The Bkd com-
plex is also involved in degradation pathways of valine and
isoleucine to produce isobutyryl CoA and 2-methylbutyryl
CoA. These precursors are the starter units of branched-chain
To identify genes involved in the alternative pathway, we
compared global gene-expression patterns in vegetative cells
of the wild-type and the bkdÀ mutant of M. xanthus
(
or iso-) fatty acids (FAs), which are of particular importance
[
16]
in myxobacteria. IV-CoA-derived iso-odd FAs are the
majority of FAs, as shown in the model myxobacterium,
DK1622. The mvaS gene was found to be part of a five-
gene operon (MXAN_4263 to MXAN_4267) that was highly
up-regulated in the bkdÀ mutant. Gene inactivation and
complementation experiments confirmed their involvement
[
5,6]
Myxococcus xanthus.
These FAs maintain the membrane
fluidity for thermal adaptation, and play key roles in signaling
[7–11]
[16]
during myxobacterial developmental differentiation.
in the alternative route to IV-CoA. Thus the operon is here
Besides its role in iso-odd FA biogenesis, IV-CoA is the
precursor of a number of myxobacterial secondary metabo-
renamed to the aib (alternative IV-CoA biosynthesis) operon.
In addition to the HMG-CoA synthase, the operon encodes
proteins with significant similarity to a TetR-like transcrip-
tional regulator (AibR), a glutaconate CoA transferase (Gct)
subunit A and B (AibA/B), and a dehydrogenase (AibC).
Furthermore, we identified genes employed in leucine
catabolism using homologues from pseudomonads as baits
and functionally analyzed them in vivo. Among those, liuC
(MXAN_3757), encoding a 3-methylglutaconyl CoA (MG-
CoA) hydratase, was shown to participate in the alternative
[
12,13]
[14]
lites, such as myxothiazol
and aurafuron. During our
study of myxothiazol biosynthesis, we analyzed the bkdÀ
mutants of M. xanthus and Stigmatella aurantiaca and showed
that iso-FAs and myxothiazol were still produced in these
[3]
mutants, albeit at lower amounts, which provided the first
evidence for the existence of an alternative pathway to IV-
CoA. The bkdÀ mutants could incorporate labeled acetate,
but not leucine, into IV-CoA-derived compounds and exhib-
ited a labeling pattern similar to that found in mevalonate-
dependent isoprenoids, thus establishing a likely link between
[
16]
pathway.
Taken together, this pathway was proposed to
proceed by LiuC-catalyzed dehydration of HMG-CoA,
followed by decarboxylation and subsequent reduction to
form IV-CoA (Figure 1). The latter two steps would be
catalyzed by a heterodimer composed of AibA/B and AibC,
respectively. Herein, we confirmed the proposed function of
each enzyme, leading to the complete reconstitution of the
alternative biosynthesis of IV-CoA in vitro. Most importantly,
we characterized AibA/B as a novel MG-CoA decarboxylase,
which apparently evolved from CoA transferases, which was
how AibA/B were annotated prior to this study.
[
+]
[
*] Dr. Y. Li, E. Luxenburger, Prof. Dr. R. Mꢀller
Helmholtz Institut fꢀr Pharmazeutische Forschung Saarland,
Helmholtz Zentrum fꢀr Infektionsforschung und Pharmazeutische
Biotechnologie, Universitꢁt des Saarlandes
66041 Saarbrꢀcken (Germany)
E-mail: rom@mx.uni-saarland.de
+
[
] Present address: CNRS, National Museum of Natural History,
Laboratory of Communication Molecules and Adaptation of
Microorganisms, UMR 7245, 75005 Paris (France)
To characterize the MG-CoA decarboxylase activity, we
synthesized MG-CoA from (E,Z)-3-methylglutaconate using
[
**] We thank Prof. Wolfgang Buckel from Marburg University for
providing us the expression plasmid of recombinant glutaconate
CoA transferase. This work was supported by the Deutsche
Forschungsgemeinschaft (DFG).
[
17]
recombinant Gct from Acidaminococcus fermentans. Only
the (E)-isomer of MG-CoA, the natural compound in cells,
was prepared because of the substrate specificity of bacterial
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201207984.
[18]
Gcts. As sequence analysis revealed that AibA and AibB
1
304
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1304 –1308

Angewandte
Chemie
In contrast, addition of both
enzymes at a molar ratio of 1:1
restored DMA-CoA produc-
tion, albeit with very low activ-
ity (Supporting information,
Figure S2). This is probably
due to the sub-optimal forma-
tion of the heterodimer if
AibA/B are not co-expressed.
Nevertheless, these results
unambiguously confirm AibA
and AibB as two subunits
required to form MG-CoA
decarboxylase.
The
closest
structural
homologue of AibA/AibB is
[
19]
Gct from A. fermentans, the
catalytic residue of which is
identified to be Glu54 on the
B subunit, which is well con-
served in other bacterial
[
17,20]
Gcts.
Sequence compari-
son of AibB with the B subu-
nits of Gcts revealed the
replacement of Glu54 by
a Cys in AibB (Cys56 accord-
ing to AibB numbering;
Figure 1. a) Gene organization of the aib operon in M. xanthus. b) Metabolic pathway of alternative IV-CoA
biosynthesis (between parentheses), leucine degradation (gray), and mevalonate-dependent isoprenoid
biosynthesis (in box with dashed lines) in myxobacteria. Structural motifs derived from isovaleryl CoA are
boxed.
Figure 3).
Therefore,
we
aimed to answer the question
whether Cys56 plays a role in
[
16]
are related to two subunits of CoA transferases, respec-
tively, we assumed it highly likely that both enzymes act as
complex. Both proteins were co-expressed with N-terminal
the catalysis and whether we can change AibA/B to a func-
tional Gct. We performed site-directed mutagenesis of aibB
to generate expression constructs for mutants AibB_C56E
and AibB_C56D. MG-CoA decarboxylase activities of the
resulting complex AibA/B_C56E and AibA/B_C56D were
reduced to 13% and 1.2%, respectively, compared to that of
the wild-type enzyme. To characterize CoA transferase
activity, assays were performed using (E,Z)-methylglutaco-
nate (20 mm) and acetyl CoA (1 mm) as substrates. Neither
the wild-type nor the two mutant enzymes were able to
produce MG-CoA. Curiously, significant amounts of CoA
were detected in reactions with mutant proteins, indicating
that they have an acetyl CoA hydrolase activity. To confirm
this, acetyl CoA (1 mm) was incubated with the WTor mutant
AibA/B complexes for 3 h. About 60% and 30% of acetyl
CoA were converted into CoA by the C56D and C56E
mutants, respectively, while the WT led to no CoA production
(Supporting information, Figure S3). These experiments
demonstrate the importance of Cys56 in AibA/AibB catalysis.
Recombinant LiuC (29 kDa) was incubated with HMG-
CoA in buffers with pH ranging from 6–8. In all cases, tiny
amounts of MG-CoA were produced and its identity was
confirmed by using synthetic MG-CoA as reference
(Figure 2). HR-MS of the product was further determined
and corresponded well to MG-CoA (calculated monoisotopic
His -tags in Escherichia coli and co-purified by nickel affinity
6
chromatography. AibA and B eluted together in a near 1 to
1
ratio starting from 200 mm imidazole (Supporting informa-
tion, Figure S1). The complex displayed a molecular mass of
7.9 Da as determined by size-exclusion chromatography,
corresponding to the formation of a heterodimer (AibA
0 kDa, AibB 28 kDa). Activity assay of AibA/B was
5
3
performed with MG-CoA at 378C for 3 h. Subsequent high-
performance liquid chromatography coupled to mass spec-
trometry (HPLC-MS) analysis revealed production of a new
compound in the reaction with identical chromatographic
properties to authentic DMA-CoA (r.t. = 17.8 min, m/z [M +
+
H] = 850). High-resolution MS (HR-MS) determination
further confirmed its identity as DMA-CoA, the decarboxy-
lated product of MG-CoA (calculated monoisotopic mass for
+
C H N O P S m/z [M + H] = 850.16490, measured mass
2
6
43
7
17
3
8
50.16336, D = À1.8 ppm). DMA-CoA could not be detected
in the control sample with heat-inactivated AibA/B
Figure 2). When HMG-CoA was used as substrate, no
(
conversion was observed. To clarify if the association of
AibA and AibB is necessary for MG-CoA decarboxylase
activity, each enzyme was expressed and purified separately.
While recombinant AibA was highly soluble, AibB was
produced mainly as inclusion bodies that resulted in very low
yields (Supporting information, Figure S1). As expected,
AibA or AibB alone showed no activity towards MG-CoA.
+
mass for C H N O P S m/z [M + H] = 894.15473, mea-
2
7
43
7
19 3
sured mass 894.15351, D = À1.4 ppm). Addition of the AibA/
B complex to LiuC reaction mixture led to the formation of
DMA-CoA. This unambiguously proves that LiuC has HMG-
Angew. Chem. Int. Ed. 2013, 52, 1304 –1308
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1305

.
Angewandte
Communications
studies on various MG-CoA
[
21–24]
hydratases.
LiuC con-
serves two glutamate resi-
dues that are important in
catalysis, as shown for the
human
Gly144 and 164 according
to human AUH numbering,
AUH
enzyme
(
[23]
Figure S4b), and therefore
it is able to accept substrates
that lack the g-carboxylate
group.
The last step of the alter-
native pathway is proposed
to involve the reduction of
DMA-CoA. Recombinant
AibC (38 kDa) was assayed
with DMA-CoA in the pres-
ence of NADH. A new prod-
uct with identical properties
as authentic IV-CoA was
produced (r.t. 18.2 min, HR-
MS: calculated monoisotopic
mass for C H N O P S m/z
Figure 2. Characterization of Aib enzymes in vitro. Left: HPLC traces of enzymes assays monitored by UV at
60 nm. a) HMG-CoA only; b) HMG-CoA with LiuC; c) MG-CoA with inactivated AibA/AibB; d) MG-CoA with
AibA/AibB; e) DMA-CoA with inactivated AibC; f) DMG-CoA with AibC; g) HMG-CoA with LiuC and AibA/
AibB; h) HMG-CoA with LiuC, AibA/AibB, and AibC. Chromatographic intensities of reaction (a) and (b) are
increased tenfold. 1 HMG-CoA, 2 MG-CoA, 3 DMA-CoA, 4 IV-CoA, 5 3-methylhydroxybutyryl-CoA. Right: Mass
spectra of enzymatic products.
2
2
6
45
7
17 3
+
[
M + H] = 852.18055, mea-
sured mass 852.17877, D =
À2.1 ppm; Figure 2). Thus,
AibC is characterized as an
IV-CoA
dehydrogenase.
CoA dehydration activity. Surprisingly, in addition to DMA-
CoA, a new peak having a mass corresponding to 3-
methylhydroxybutyryl CoA was revealed by HPLC and
HR-MS analysis (calculated monoisotopic mass for
Presence of EDTA (5 mm) did not affect AibC activity,
indicating that no metal cofactor is required for AibC. To
reconstitute the complete alternative biosynthesis, HMG-
CoA (1 mm) was incubated with equal amounts of LiuC,
AibA/B, and AibC in the presence of NADH. As expected,
this yielded IV-CoA as a major product (Figure 2).
In summary, we performed characterization in vitro and
provided ultimate evidence for the proposed novel pathway.
Intriguingly, this route involves reverse steps of leucine
degradation. Genes encoding MvaS, AibA-C, and also
+
C H N O P S m/z [M + H] = 868.17546, measured mass
2
6
45
7
18
3
8
68.17476, D = À0.8 ppm). This is most likely due to the
hydration of DMA-CoA catalyzed by LiuC. We next inves-
tigated the substrate specificity of LiuC. LiuC hydrated
efficiently DMA-CoA and crotonyl CoA (Supporting infor-
mation, Figure S4a). This finding is consistent with previous
Figure 3. a) Sequence alignment of AibB with subunit B of other Gcts. AfeB: Gct from A. fermentans subunit B; ClosB: CBK77199.1, from
Clostridium cf. sacharolyticum; PepB: EFI41745.1, from Peptoniphilus sp. ? indicates the active-site Glu. b) Sequence alignment of AibB with
homologous proteins. NocB: YP_919290.1 from Nocardioides sp. JS614; XauB: YP_001409505.1 from Xanthobacter autotrophicus Py2; MyeB:
ZP_09705801.1 from Metallosphaera yellowstonensis MK1; FplB: YP_003435597.1 from Ferroglobus placidus DSM10642. ? indicates Cys56 in AibB.
1
306
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1304 –1308

Angewandte
Chemie
a putative pathway-specific regulator (AibR) are organized in
an operon, which represents an advantage in coordinated
regulation under leucine-limiting conditions, such as starva-
tion that leads to fruiting body formation. On the other hand,
liuC is encoded in another gene locus and is thus under
different transcriptional controls (for example, liuC is not up-
proteins in the database annotated as CoA transferase
subunit B (around 30% sequence identity), all of which do
not conserve a Glu active site, suggesting that not all of them
perform CoA transferase reactions.
This study demonstrates the importance of in-depth
biochemical studies, which are urgently required to make
use of the enormous potential of available genome sequences,
not only to understand the underlying biochemical principles
but also to identify novel biocatalysts. In this respect, our
findings are also of importance to biofuel research, as the
novel pathway to synthesize IV-CoA provides opportunities
to produce isobutene, a precursor of renewable fuels and
[
16]
regulated in the bkdÀ mutant of M. xanthus ). Such
differential regulation of the aib operon and liuC may be
important to maintain equilibrated partitioning of HMG-
CoA between the IV-CoA synthesis and the mevalonate
pathway. Moreover, LiuC is an unusual bacterial MG-CoA
hydratase and its low activity towards typical natural sub-
strates is probably an intrinsic feature reflecting its impor-
tance in regulating two metabolic pathways in parallel. To
investigate whether the novel shunt pathway is common in
other bacteria, we used aibA and aibB as baits to search
available bacterial genomes. This revealed that the aib operon
is strictly conserved in myxobacteria, most of which belonging
to the suborder Cystobacterineae (Supporting Information,
Table S1). The alternative synthesis of IV-CoA is therefore
likely to be unique for a set of myxobacteria, a feature related
to their specific lifestyle. However, we cannot exclude the
possibility of aib genes dispersed at various gene loci in other
bacteria.
[28]
chemicals, from glucose by synthetic biology approaches.
Received: October 3, 2012
Published online: December 6, 2012
Keywords: biosynthesis · decarboxylase · isovaleryl CoA ·
.
leucine degradation · myxobacteria
[
1] K. Gerth, S. Pradella, O. Perlova, S. Beyer, R. Mꢀller, J.
Biotechnol. 2003, 106, 233 – 253.
[
[
2] S. C. Wenzel, R. Mꢀller, Nat. Prod. Rep. 2007, 24, 1211 – 1224.
3] T. Mahmud, H. B. Bode, B. Silakowski, R. M. Kroppenstedt, M.
Xu, S. Nordhoff, G. Hçfle, R. Mꢀller, J. Biol. Chem. 2002, 277,
32768 – 32774.
The key step in the shunt pathway is the conversion of
MG-CoA into DMA-CoA catalyzed by an unprecedented
MG-CoA decarboxylase (AibA/B). AibA/B functionally
resembles the ECH2 domain of CurF involved in the
[
4] G. Michal in Biochemical Pathways, Spektrum, Heidelberg,
1999.
[25]
[5] H. B. Bode, J. S. Dickschat, R. M. Kroppenstedt, S. Schulz, R.
biosynthesis of the cyanobacterial metabolite curacin A.
Mꢀller, J. Am. Chem. Soc. 2005, 127, 532 – 533.
CurF ECH2 belongs to the crotonase superfamily and
catalyzes decarboxylation of an acyl carrier protein (ACP)
tethered 3-methylglutaconyl group to DMA-ACP. Structur-
ally, AibA/B most likely possess similar folds as CoA
transferases, although they exhibit different oligomeric
states. AibA/B is a heterodimer, whereas functional Gct is
a tetramer of the heterodimer as shown by Gct from A.
[
6] J. S. Dickschat, H. B. Bode, R. M. Kroppenstedt, R. Mꢀller, S.
Schulz, Org. Biomol. Chem. 2005, 3, 2824 – 2831.
[7] H. B. Bode, M. W. Ring, G. Schwar, R. M. Kroppenstedt, D.
Kaiser, R. Mꢀller, J. Bacteriol. 2006, 188, 6524 – 6528.
[
[
8] J. Downward, D. Toal, Mol. Microbiol. 1995, 16, 171 – 175.
9] E. Hoiczyk, M. W. Ring, C. A. McHugh, G. Schwar, E. Bode, D.
Krug, M. O. Altmeyer, J. Z. Lu, H. B. Bode, Mol. Microbiol.
2009, 74, 497 – 517.
[19]
fermentans. Mechanistically, AibA/B would share similarity
[
10] M. W. Ring, G. Schwar, V. Thiel, J. S. Dickschat, R. M. Krop-
penstedt, S. Schulz, H. B. Bode, J. Biol. Chem. 2006, 281, 36691 –
36700.
[11] D. R. Toal, S. W. Clifton, B. A. Roe, J. Downard, Mol. Microbiol.
995, 16, 177 – 189.
[12] B. Silakowski, H. U. Schairer, H. Ehret, B. Kunze, S. Weinig, G.
Nordsiek, P. Brandt, H. Blocker, G. Hçfle, S. Beyer, R. Mꢀller, J.
Biol. Chem. 1999, 274, 37391 – 37399.
13] W. Trowitzsch, G. Reifenstahl, V. Wray, K. Gerth, J. Antibiot.
1980, 33, 1480 – 1490.
[14] B. Kunze, H. Reichenbach, R. Mꢀller, G. Hçfle, J. Antibiot. 2005,
8, 244 – 251.
15] T. Mahmud, S. C. Wenzel, E. Wan, K. W. Wen, H. B. Bode, N.
Gaitatzis, R. Mꢀller, ChemBioChem 2005, 6, 322 – 330.
16] H. B. Bode, M. W. Ring, G. Schwar, M. O. Altmeyer, C. Kegler,
I. R. Jose, M. Singer, R. Mꢀller, ChemBioChem 2009, 10, 128 –
[26]
with CurF ECH2
and related crotonases that involve
a thioester enolate anion intermediate stabilized by two
backbone amide groups, forming an oxyanion hole. Such
hydrogen-bonding residues have also been identified in the
1
[19]
structure of Gct from A. fermentans. The active site Cys56
in AibB may function as the proton donor to the C-4 position
of the decarboxylated product (Supporting information,
Figure S5). How AibA/B has exactly adapted a CoA trans-
ferase structure to catalyze CoA decarboxylation is an
intriguing question and warrants further work of structural
analysis. Gcts employ a catalytic glutamate to form alter-
nately enzyme-bound anhydrides and thioesters during the
[
5
[
[
[
27]
reaction cycle. Substitution of the corresponding residue in
AibB (Cys56) to Glu or Asp severely reduced MG-CoA
decarboxylase activity, but led to some acetyl CoA hydrolase
activity, which could be attributed to an anhydride mechanism
similar to that of Gcts (Supporting information, Figure S6).
However, these single substitutions were not sufficient to
convert AibA/B into a CoA transferase. This finding suggests
that AibA/B most likely underwent active site remodeling to
enable decarboxylation while maintaining the overall CoA
transferase structure. AibB shows modest homology to other
140.
[17] M. Mack, K. Bendrat, O. Zelder, E. Eckel, D. Linder, W. Buckel,
Eur. J. Biochem. 1994, 226, 41 – 51.
[
[
18] B. Messner, H. Eggerer, J. Cornforth, R. Mallaby, Eur. J.
Biochem. 1975, 53, 255 – 264.
19] U. Jacob, M. Mack, T. Clausen, R. Huber, W. Buckel, A.
Messerschmidt, Structure 1997, 5, 415 – 426.
20] M. Mack, W. Buckel, FEBS Lett. 1995, 357, 145 – 148.
21] L. E. Brennan, J. Nakagawa, D. Egger, K. Bienz, C. Moroni,
Gene 1999, 228, 85 – 91.
[
[
Angew. Chem. Int. Ed. 2013, 52, 1304 –1308
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1307

.
Angewandte
Communications
[
22] K. Kurimoto, S. Fukai, O. Nureki, Y. Muto, S. Yokoyama,
Structure 2001, 9, 1253 – 1263.
[26] T. W. Geders, L. Gu, J. C. Mowers, H. Liu, W. H. Gerwick, K.
Hakansson, D. H. Sherman, J. L. Smith, J. Biol. Chem. 2007, 282,
35954 – 35963.
[27] W. Buckel, Eur. J. Biochem. 1976, 64, 263 – 267.
[28] A. Dhar, K. Dhar, J. P. Rosazza, J. Ind. Microbiol. Biotechnol.
2002, 28, 81 – 87.
[
23] M. Mack, U. Schniegler-Mattox, V. Peters, G. F. Hoffmann, M.
Liesert, W. Buckel, J. Zschocke, FEBS J. 2006, 273, 2012 – 2022.
24] B. J. Wong, J. A. Gerlt, Biochemistry 2004, 43, 4646 – 4654.
25] L. Gu, J. Jia, H. Liu, K. Hakansson, W. H. Gerwick, D. H.
Sherman, J. Am. Chem. Soc. 2006, 128, 9014 – 9015.
[
[
1
308
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1304 –1308