Microbial biosynthesis of medium-chain 1-alkenes by a nonheme iron oxidase

Article abstract of DOI:10.1073/pnas.1419701112

Aliphatic medium-chain 1-alkenes (MCAEs, ～10 carbons) are "drop-in" compatible next-generation fuels and precursors to commodity chemicals. Mass production of MCAEs from renewable resources holds promise for mitigating dependence on fossil hydrocarbons. An MCAE, such as 1-undecene, is naturally produced by Pseudomonas as a semivolatile metabolite through an unknown biosynthetic pathway. We describe here the discovery of a single gene conserved in Pseudomonas responsible for 1-undecene biosynthesis. The encoded enzyme is able to convert medium-chain fatty acids (C10-C14) into their corresponding terminal olefins using an oxygen-activating, nonheme iron-dependent mechanism. Both biochemical and X-ray crystal structural analyses suggest an unusual mechanism of β-hydrogen abstraction during fatty acid substrate activation. Our discovery unveils previously unidentified chemistry in the nonheme Fe(II) enzyme family, provides an opportunity to explore the biology of 1-undecene in Pseudomonas, and paves the way for tailored bioconversion of renewable raw materials to MCAE-based biofuels and chemical commodities.

Full text of DOI:10.1073/pnas.1419701112

Microbial biosynthesis of medium-chain 1-alkenes by
a nonheme iron oxidase
Zhe Ruia,b, Xin Lia,c, Xuejun Zhua,b, Joyce Liua,d, Bonnie Domigana,b, Ian Barre,f, Jamie H. D. Catee,f,g
and Wenjun Zhanga,b,g,1
,
aEnergy Biosciences Institute, cCalifornia Institute for Quantitative Biosciences, and Departments of bChemical and Biomolecular Engineering,
dBioengineering, eChemistry, and fMolecular and Cell Biology, University of California, Berkeley, CA 94720; and gPhysical Biosciences Division,
Lawrence Berkeley National Laboratory, Berkeley, CA 94720
Edited by Jerrold Meinwald, Cornell University, Ithaca, NY, and approved November 18, 2014 (received for review October 14, 2014)
Aliphatic medium-chain 1-alkenes (MCAEs, ∼10 carbons) are “drop-in”
compatible next-generation fuels and precursors to commodity chem-
icals. Mass production of MCAEs from renewable resources holds
promise for mitigating dependence on fossil hydrocarbons. An
MCAE, such as 1-undecene, is naturally produced by Pseudomonas
as a semivolatile metabolite through an unknown biosynthetic path-
way. We describe here the discovery of a single gene conserved in
Pseudomonas responsible for 1-undecene biosynthesis. The encoded
enzyme is able to convert medium-chain fatty acids (C10–C14) into
their corresponding terminal olefins using an oxygen-activating,
nonheme iron-dependent mechanism. Both biochemical and X-ray
crystal structural analyses suggest an unusual mechanism of β-hy-
drogen abstraction during fatty acid substrate activation. Our discov-
ery unveils previously unidentified chemistry in the nonheme Fe(II)
enzyme family, provides an opportunity to explore the biology of
1-undecene in Pseudomonas, and paves the way for tailored biocon-
version of renewable raw materials to MCAE-based biofuels and
chemical commodities.
by their native producers (5, 6). In addition, MCAEs are insoluble
in water, less dense than water, and readily escape the cell culture
as semivolatile metabolites (8). Therefore, a reliable and sensitive
method for MCAE detection is a prerequsite for their biosynthetic
study. In this work, we used a headspace solid phase micro-
extraction (SPME)–gas chromatography mass spectrometry
(GCMS) analysis to detect the semivolatile metabolite 1-
undecene, and we revealed that the production of 1-undecene
is ubiquitous in the genus of Pseudomonas. We further dis-
covered that a novel enzyme conserved in Pseudomonas is
responsible for 1-undecene biosynthesis through an unusual
catalytic mechanism.
Results
Analysis of 1-Undecene Production in Microbes. Hydrocarbon metabo-
lites produced in cell cultures are typically extracted by organic
solvents (11, 12), but the low titer and volatility of 1-undecene
resulted in ambiguous signals during our initial GCMS analysis
using this extraction method. Alternatively, SPME allows the effi-
cient extraction of volatile metabolites directly from the headspace.
Based on headspace SPME–GCMS analysis, we developed a sensi-
tive detection method for volatile metabolites with a detection limit
of ∼1 ng of total 1-undecene in cell cultures (including both liquid
and gas phases). All of the tested Pseudomonas strains produced
1–100 ng/mL of 1-undecene (SI Appendix, Materials and Methods),
demonstrating the robustness of this method for semivolatile
hydrocarbon terminal olefin iron-dependent desaturase/decarboxylase
|
|
|
biofuel biosynthesis
|
urging energy consumption and environmental concerns have
stimulated interest in the production of chemicals and fuels
S
through sustainable and renewable approaches. Medium-chain
1-alkenes (MCAEs) are of particular interest because they are
“drop-in”–ready next-generation fuels with superior properties
such as low freezing point compared with long-chain diesels, high
energy content compared with short-chain fuels, easy product
recovery due to insolubility in water, and compatibility with the
existing engine systems and transportation infrastructure (1, 2).
Because of a readily derivatized terminal functionality, MCAEs
are also valuable precursors to commodity chemicals such as
lubricants, pesticides, polymers, and detergents (3, 4). Biological
production of MCAEs from renewable resources holds promise
for mitigating dependence on fossil hydrocarbons. Although
MCAEs are naturally produced by diverse species as semivolatile
metabolites (5, 6), little is known about the genetic and molec-
ular basis for MCAE biosynthesis. Elucidation of MCAE bio-
synthetic pathway will serve as the basis for engineering efforts to
establish bioprocesses for producing MCAE-based biofuels and
chemical commodities from renewable resources.
Significance
We have solved a long-standing mystery of the biosynthetic
origin of 1-undecene, a ubiquitous hydrocarbon semivolatile
metabolite of Pseudomonas. Our study revealed an unpre-
cedented family of nonheme oxidases that specifically convert
medium-chain fatty acids into the corresponding terminal ole-
fins using an oxygen-activating, nonheme iron-dependent
mechanism. Our findings unveil previously unidentified chem-
istry in the nonheme Fe(II) enzyme family, aid the functional
study of this ubiquitous metabolite in Pseudomonas, expand
the scarce enzyme inventory for the transformation of fatty
acid precursors to hydrocarbons, and serve as the basis for
engineering efforts to establish bioprocesses to produce me-
dium-chain terminal olefins, useful as fuels and chemical
building blocks, from renewable resources.
1-Undecene, an MCAE with 11 carbons, was identified as a
biomarker of Pseudomonas aeruginosa, one of the most signifi-
cant human pathogens (7–10). However, the biology of this
characteristic semivolatile metabolite in P. aeruginosa remains
enigmatic, and the biosynthetic pathway of 1-undecene has not
previously been explored. It was also reported that some species
of Pseudomonas produce 1-undecene, whereas some species do
not (8), inspiring us to use a comparative genomics approach to
reveal the genetic basis for 1-undecene biosynthesis (11). It is
notable that one of the major challenges for MCAE biosynthetic
study is the detection and quantification of MCAE production.
MCAEs, such as 1-undecene, are only produced in trace amounts
Author contributions: Z.R. and W.Z. designed research; Z.R., X.L., X.Z., J.L., B.D., and I.B.
performed research; Z.R. contributed new reagents/analytic tools; Z.R., X.L., and W.Z.
analyzed data; and Z.R., J.H.D.C., and W.Z. wrote the paper.
Conflict of interest statement: Z.R. and W.Z. have filed a provisional patent application on
aspects of this research.
This article is a PNAS Direct Submission.
Data deposition: The crystallography, atomic coordinates, and structure factors have been
deposited in the Protein Data Bank, www.pdb.org (PDB ID codes 4WWJ, 4WWZ, and 4WX0).
1To whom correspondence should be addressed. Email: wjzhang@berkeley.edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1419701112/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1419701112
PNAS
|
December 23, 2014
|
vol. 111
|
no. 51
|
18237–18242

MCAE analysis. Although 1-undecene has also been reported to
be produced by Shewanella (6), we did not detect the production
of 1-undecene by seven selected Shewanella strains. In addition,
none of Escherichia coli strains tested produced a detectable
amount of 1-undecene (Fig. 1A).
the gene PA14_53120 homologous to undA (15). This gene dis-
ruption completely abolished 1-undecene production in P. aerugi-
nosa PA14 (Fig. 1 A and B), confirming the necessity of this gene in
1-undecene biosynthesis by P. aeruginosa.
Biochemical Analysis of UndA. To establish the precise role of
UndA in biosynthesis of 1-undecene, we performed in vitro en-
zymatic assays using the purified recombinant enzyme from
E. coli. UndA is a small protein (261 amino acids) with low se-
quence homology (identity/similarity = 13%/23%) to TenA,
a thiaminase II from Bacillus subtilis (16), but the essential cat-
alytic cysteine of TenA is missing in UndA. In addition, the
structure of an UndA homolog from Pseudomonas syringae pv.
tomato DC3000 (identity/similarity = 79%/86%) [Protein Data
Bank (PDB) 3OQL] was resolved by the Joint Center for
Structural Genomics, but no enzymatic activity was reported and
no cofactor was found in the crystal structure. Because in silico
analysis failed to predict any possible activity of UndA, we first
tested a panel of substrates in the UndA biochemical assays,
including LA, α-hydroxydodecanoic acid (AHDA), β-hydroxy-
dodecanoic acid (BHDA), 2,3-dodecenoic acid (DEA), methyl
laurate, ethyl laurate, and lauroyl-CoA. All attempts failed to
produce 1-undecene. Although alternative substrates could have
been possible, we reasoned that a critical factor was likely
missing in the in vitro assays. We then screened common en-
abling factors in combination with the substrate panel for
1-Undecene Biosynthetic Gene Discovery. We hypothesized that
1-undecene originates from the fatty acid metabolism. Feeding
[12-13C]lauric acid (LA) and [1-13C]LA to Pseudomonas cultures
resulted in the production of [11-13C]undecene and [U-12C11]
undecene, respectively (SI Appendix, Fig. S1), implying that LA
is the 1-undecene biosynthetic precursor and the terminal car-
boxylic acid moiety is removed during 1-undecene formation. At
least two mechanisms were recently characterized to yield long-
chain 1-alkenes from fatty acyl precursors with a loss of CO2.
These reported mechanisms include the cytochrome P450 fatty
acid peroxygenase-catalyzed synthesis of 8-methyl-1-nonadecene
and 17-methyl-1-nonadecene in the Jeotgalicoccus bacteria spe-
cies (12, 13), and the multidomain polyketide synthase-promoted
production of very long-chain 1-alkenes in the Synechococcus
cyanobacteria species (14). Bioinformatics analysis revealed that
no homologs of these enzymes are encoded in the genomes
of Pseudomonas, suggesting different biochemistry underlying
1-undecene synthesis.
The comparative genomics analysis yielded thousands of gene
candidates for 1-undecene biosynthesis due to the ubiquitous pro-
duction of 1-undecene in Pseudomonas and a few related bacteria
species, demonstrating the limitation of this approach. Alterna-
tively, we attempted to locate the biosynthetic genes by heterolo-
gously expressing a genomic library of Pseudomonas fluorescens Pf-5
in E. coli followed by phenotype screening of the 1-undecene pro-
ducing clones (SI Appendix, Fig. S2). After screening ∼6,000 fos-
mids, we determined that one single gene, designated as undA, is
necessary and sufficient to confer 1-undecene production in E. coli
(Fig. 1 A and B). To further probe the function of undA homologs
in the biosynthesis of 1-undecene, we heterologously expressed
three additional undA homologs from various Pseudomonas species.
The overexpression of all gene candidates in E. coli conferred 1-
undecene production, confirming the proposed function of these
1-undecene production. These tested factors include Fe2+, Fe3+
,
Mg2+, Ca2+, Mn2+, Co2+, Ni2+, Cu2+, Zn2+, nicotinamides, pyr-
idoxal phosphate, flavins, S-adenosyl methionine, pyrroloquino-
line quinone (PQQ), and ATP. After an extensive number of
trials, we discovered that only one combination, LA and Fe2+
,
elicited 1-undecene production in the UndA assay. Notably, all
other tested metal ions including Fe3+ failed to enable 1-unde-
cene production. These results demonstrated that UndA is a
nonheme iron (II)-dependent enzyme that converts LA to
1-undecene (Fig. 1D).
After reconstituting the activity of UndA in vitro, we per-
formed detailed biochemical analyses of this enzyme. The iron
content of UndA after anaerobic reconstitution of purified apo-
genes (Fig. 1C). Additionally, we examined the ΔPA14_53120 mu- UndA with Fe2+ was determined to be 82% by inductively cou-
tant of P. aeruginosa PA14, which bears a transposon insertion in pled plasma mass spectrometry. Using [1-13C]LA as a substrate,
7
A
C
6
5
4
3
2
1
0
i
ii
iii
iv
3
5
7
9
11
13
15
UndA
PSPTO_1738 PA14_53120 Pput_3952
Retention time (min)
B
D
PFL_4320 undA PFL_4322
PFL_4323
O
P. fluorescens Pf-5
P. syringae B728a
P. putida F1
OH+ O2
P. fluorescens Pf0-1
P. putida UW4
UndA
Fe(II)
P. aeruginosa PAO-1
P. aeruginosa PA14
+ CO2
PA14_53120
Fig. 1. Identification of the enzyme responsible for 1-undecene biosynthesis in Pseudomonas. (A) Demonstration of 1-undecene production from
headspace SPME–GCMS analysis. (i) 1-undecene production was observed during the library screening in the E. coli EPI300 expressing the gene undA. (ii)
P. aeruginosa PA14 naturally produces 1-undecene. (iii) Disruption of the gene PA14_53120 (an undA homolog) completely abolished 1-undecene
production in the P. aeruginosa ΔPA14_53120 mutant. (iv) E. coli EPI300 does not naturally produce 1-undecene. (B) UndA-containing two-gene operons
conserved in Pseudomonas. (C) Production titer of extracellular 1-undecene by overexpressing undA (P. fluorescens Pf-5), PSPTO_1738 (P. syringae pv.
tomato DC3000), PA14_53120 (P. aeruginosa PA14), and Pput_3952 (P. putida F1) in E. coli BL21 Star. Error bars represent SDs from at least three in-
dependently performed experiments. (D) The biosynthesis of 1-undecene from LA catalyzed by UndA, an oxygen-activating, nonheme, iron-dependent
desaturase/decarboxylase.
18238
|
www.pnas.org/cgi/doi/10.1073/pnas.1419701112
Rui et al.

formation of [13C]CO2 was detected (Fig. 2 and SI Appendix, Fig. ascorbate enabled multiple turnovers for UndA in vitro, but the
S3), confirming that UndA catalyzes production of 1-undecene
from LA through oxidative decarboxylation. Because the in vitro
aerobic reaction mixture contained only desalted UndA–Fe(II)
and LA with no additional oxidant/reductant, we reasoned that
molecular oxygen served as the oxidant of the reaction, which
was supported by the observations that 1-undecene production was
nearly abolished in anaerobic enzymatic assays, and that O2 was
consumed in aerobic assays (SI Appendix, Fig. S4). The measured
stoichiometry of the reaction showed that one molecule of UndA–
Fe(II) consumed approximately one equivalent of O2 to produce
close to one equivalent of 1-undecene (SI Appendix, Fig. S4). O2
was presumably reduced to hydrogen peroxide or water by UndA.
We proposed that H2O was a likely end product, as no production
of H2O2 was detected in the UndA assays by two independent
sensitive analytical methods, and the addition of catalase to the
reaction system had no detectable effect.
Only single turnover reactions were obtained in the UndA in
vitro assays, probably due to an electron imbalance, where two
electrons are donated by one molecule of LA while four electrons
are required to reduce one molecule of O2 to water. This electron
imbalance stalls the enzyme, presumably with an inactive oxidized
iron species. This hypothesis is supported by an UndA recycling
experiment in which apo-UndA recycled after the single turnover
reaction maintained over 80% activity upon reconstitution with
fresh Fe2+. We next surveyed a group of possible reductive
cosubstrates based on those used by well-studied oxygen-activating
iron-dependent oxygenases (17–20). These reductive cosubstrates
included α-ketoglutarate (α-KG), ascorbic acid, glutathione, cys-
teine, DTT, Tris(2-carboxyethyl)phosphine, nicotinamides, flavins,
ferredoxin, tetrahydropterin, phenazine methosulfate (PMS)/
NADH, and PQQ/cysteine. Initially, none of the tested reducing
systems promoted 1-undecene production over the stoichiometric
amount of one turnover. We suspected that the limiting factor
was likely the consumption of dissolved O2 by both the enzymatic
reaction and some of the tested reducing agents in the sealed vial
without replenishment. Indeed, through coupling with the chlo-
rite dismutase reaction to generate O2 in situ (21), we found that
rate (four turnovers per hour) was much slower than the initial
rate of 1-undecene production under the single turnover condi-
tion [0.06 0.002 mol 1-undecene/(mol UndA-Fe(II) second)]
(SI Appendix, Figs. S5 and S6). Other electron donors such as
PMS/NADH and PQQ/cysteine also enabled multiple turnovers,
although the resulting rates and yields were lower than when
ascorbate was the reductant (SI Appendix, Fig. S5).
Because HppE, a long recognized nonheme iron (II) oxidase, was
recently characterized to be a peroxidase (22), we then extensively
tested the possibility that H2O2, rather than O2, is the real substrate
for UndA. The reduction of H2O2 to water requires only two
electrons, matching the two-electron requirement for the overall
conversion of LA to 1-undecene. However, in the presence of
H2O2, no 1-undecene was formed in anaerobic reactions with either
UndA–Fe(III) or UndA–Fe(II). In the presence of both H2O2 and
O2, no 1-undecene was formed with UndA–Fe(III), further con-
firming the requirement of Fe(II) for the activity of UndA. In ad-
dition, H2O2 did not promote the 1-undecene formation in the
reaction containing O2 and UndA–Fe(II), indicating that H2O2 is
not the substrate for UndA. These results are consistent with the
observations that addition of dithionite to reduce O2 to H2O2 in situ
inhibited 1-undecene production by UndA, but the addition of
dithionite increased the rate of HppE up to 1,000 times (22).
In addition to LA (C12:0), UndA converted myristic acid
(C14:0) and capric acid (C10:0) to their corresponding “[M-1]-
carbon” 1-alkenes (SI Appendix, Figs. S7 and S8), and the initial
rates of 1-alkene production at varying substrate concentrations
were further measured (SI Appendix, Fig. S6). UndA failed to act
on palmitic acid (C16:0) or caprylic acid (C8:0), indicating that
the substrate binding pocket of the enzyme accepts a range of
medium chain-length fatty acid substrates. Interestingly, UndA
transformed AHDA to 1-undecanal, but exhibited no activity
toward DEA and BHDA (Fig. 2 and SI Appendix, Fig. S9),
suggesting that the β-carbon of LA, rather than the α-carbon, is
the site of activation during enzymatic catalysis. This is further
confirmed by the UndA assay with [α,α-D2]LA, which led to the
production of 1-undecene retaining both deuterium atoms, and
the UndA assay with [D23]LA, which led to the production of
[D22]1-undecene (Fig. 2 and SI Appendix, Figs. S10 and S11).
O
Structural Analysis of UndA. To understand the molecular basis
of catalysis and the substrate recognition mechanism, we de-
termined crystal structures of UndA in its holo form (PDB ID
code 4WWJ) and in complex with two substrate analogs, either
DEA (PDB ID code 4WWZ) or BHDA (PDB ID code 4WX0),
with resolutions ranging from 1.7 to 1.9 Å (Fig. 3 and SI Appendix,
Fig. S12 and Table S2). The active site of UndA contains a single
iron center in an octahedral configuration with three sites of
coordination from the enzyme side chains of Glu101, His104, and
His194, and three variable sites of coordination, two facing the
pocket and one exposed to exterior. The three variable sites are
coordinated by oxygen atoms from the substrate analogs, solvent
molecules such as water or glycerol, or O2. Different from typical
nonheme mononuclear Fe(II) enzymes, the His194–His104–Glu101
residues of UndA are not arranged as a “facial” triad (20). The
critical roles of these three residues were confirmed by site-
directed mutagenesis, which resulted in an over 1,000-fold drop in
1-undecene production (SI Appendix, Fig. S13). The substrate
analogs DEA or BHDA reside in a deep hydrophobic pocket that
extends from the surface to the center of the enzyme (Fig. 3 A and
B). The depth of the pocket limits the length of the fatty acid
substrate to an ∼14-carbon chain, consistent with the biochemical
results. Whereas the holo-structure represents the resting state of
the active site with three iron coordination positions occupied by
solvent molecules (one from water and two from glycerol) (Fig.
3C), cocrystallization with the substrate analogs DEA and BHDA
seemed to trap the enzyme in different oxygen activation steps of
A
13
+
13
CO
C
2
OH
O
B
CO
2
13
C
+
+
13
C
OH
O
C
CO
2
OH
OH
O
OH
O
D
E
D
D
D
+
CO
2
D
D
D
D
D
D
D
D
D
D
O
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
+
CO
2
D
D
OH
D
D
D
D
O
D
D
OH
F
OH
OH
O
G
Fig. 2. UndA activity toward selected substrates: A, [1-13C]LA; B, [12-13C]LA;
C, AHDA; D, [α,α-D2]LA; E, [D23]LA; F, BHDA; G, DEA. Both deuterium atoms
of [α,α-D2]LA are proposed to retain at the α-carbon position forming [1,1-
D2]undecene in D.
Rui et al.
PNAS
|
December 23, 2014
|
vol. 111
|
no. 51
|
18239

A
B
C
D
E
F
G
Glu101
His104
Glu
Glu
O
LA
O2
H2O
H2O
His
His
O
O
O
II
III
II
Fe
Fe
Fe
CO2
OH2
C
9
O
O
OH2
C9
His194
C
9
His
His
H
Fig. 3. Structures of UndA and proposed 1-undecene synthesis mechanism. (A) Overall structure of UndA with helices in blue and loops in salmon. Iron is
shown in black, and DEA is shown in yellow for carbons and red for oxygens. (B) Substrate binding pocket of UndA. DEA, presented as a ball-and-stick model,
is surrounded by a simulated annealing omit map in blue, contoured at 3.0 σ. Pocket-forming residues are displayed as sticks and the hydrophobic residues are
colored in orange. (C–F) Weighted electron density maps surrounding the active site of the holo-, DEA-, and BHDA-bound structures with 2mFo-DFc (in gray, at
1.8 σ) and mFo-DFc (green and red, at 3.0 σ). The distal oxygen atoms of the dioxygen species in E and F are surrounded by the simulated annealing omit
maps in green, contoured at 3.0 σ. (G) Proposed mechanism for 1-undecene biosynthesis by UndA. The molecular oxygen is proposed to be reduced to H2O
likely by some reductant as shown in SI Appendix, Fig. S14A.
the catalytic cycle. In crystals with DEA-bound enzymes, two structure, which differs from the DEA-bound structures by the
different proposed intermediates are present in the crystallo- coordination of the β-hydroxyl group opposite to His104 (Fig. 3F).
graphic asymmetric unit. One subunit contains a water molecule The refined distance between the oxygen atoms in the BHDA-
coordinated to the iron, demonstrating an intermediate stage with bound structure converges to 1.45 Å, suggesting the presence of
a bound fatty acid substrate before O2 binding (Fig. 3D). In the a peroxide species, although this assignment is tentative due to the
other subunit, a diatomic molecule, assigned as molecular oxygen, limit of resolution of 1.7 Å. The β-hydroxyl oxygen of BHDA is
replaces the water and occupies the iron coordination site oppo- positioned only 2.45 Å from the distal oxygen atom (SI Appendix,
site to the carboxyl oxygen of the fatty acid substrate in an end-on Fig. S12), implying a possible β-hydrogen abstraction from the
fashion (Fig. 3E). Dioxygen also appears in the BHDA-bound native substrate LA during the catalysis.
18240
|
www.pnas.org/cgi/doi/10.1073/pnas.1419701112
Rui et al.

Discussion
to the P450 fatty acid peroxygenase (13). We postulate that the
proposed reaction of Fe(III)–superoxide-promoted β-hydrogen
abstraction might be thermodynamically possible in concert with
a single electron transfer yielding CO2, an excellent leaving
group. Experiments to further characterize the catalytic mecha-
nism are in progress.
Based on the biochemical and structural analyses of UndA, we
postulate the following O2-activating, nonheme Fe(II)-dependent
mechanism for the oxidative decarboxylation of LA (Fig. 3G and
SI Appendix, Fig. S14A). First, LA binds to the ferrous iron of the
enzyme through its carboxylate moiety and organizes the iron
center to coordinate molecular oxygen. This sequential binding is
similar to that in many other nonheme iron enzymes, such as
naphthalene dioxygenase and superoxide reductase (18, 20, 23–
26). Next, O2 binds, generating a putative Fe(III)–superoxide
complex that abstracts the unactivated β-hydrogen of LA. Single
electron transfer at the stage of a highly reactive radical in-
termediate leads to the formation of 1-undecene, CO2, H2O, and
a possible unstable Fe(IV)=O species that can be reduced back to
the ferrous state by a reducing cosubstrate to regenerate the ac-
tive ferrous form of the enzyme. This proposed mechanism is
consistent with the observed stoichiometry under single turnover
conditions where one molecule of UndA–Fe(II) consumed ap-
proximately one equivalent of O2 to produce close to one
equivalent of 1-undecene. Although the α-hydrogen of LA is
generally considered to be more reactive, we propose β-hydrogen
abstraction of LA based on several observations: (i) The UndA
assay with [α,α-D2]LA led to the production of 1-undecene
retaining both deuterium atoms (Fig. 2); (ii) UndA accepts
AHDA, but not BHDA, as a substrate (Fig. 2); and (iii) the
β-hydroxyl oxygen of BHDA is only 2.45 Å away from the distal
oxygen atom in the BHDA-bound structure, suggesting that the
activated O2 can reach the β-hydrogen of LA for hydrogen ab-
straction (SI Appendix, Fig. S12). The analogous β-hydrogen ab-
straction in substrate activation has also been proposed during
porphyrin biosynthesis, in which the propionate side chains of
coproporphyrinogen III are converted into the corresponding vinyl
groups through oxidative decarboxylation. Interestingly, two struc-
turally unrelated enzymes, one oxygen-dependent (HemF) and one
oxygen-independent (HemN), are able to catalyze this reaction
using a possible Mn(III)–superoxo species and a 5′-deoxyadenosyl
radical to initiate β-hydrogen abstraction, respectively (27, 28).
It is notable that the high-valent Fe(IV)=O species is canon-
ically considered to be sufficiently reactive for hydrogen
abstraction from an unactivated carbon center, such as the
β-hydrogen of fatty acids. The observed multiple turnover re-
action of UndA in the presence of ascorbate raises the possibility
of an alternative mechanism, in which ascorbate serves as the
two-electron donating cosubstrate to form the Fe(IV)=O species
for the subsequent hydrogen abstraction of LA, similar to the
catalytic mechanism used by 1-aminocyclopropane-1-carboxylic
acid oxidase (ACCO) (SI Appendix, Fig. S14B) (29, 30). How-
ever, unlike the ACCO-catalyzed reaction, the rate of 1-unde-
cene formation under single turnover conditions in the absence
of ascorbate was not significantly lower than the rate under
multiple turnover conditions in the presence of ascorbate, and
nearly a stoichiometric amount of 1-undecene was formed in the
absence of ascorbate, suggesting that ascorbate is not needed for
O2 activation by UndA–Fe(II) to catalyze 1-undecene formation
(29, 30). In addition, ACCO is found in plants, which naturally
produce ascorbate, but most Pseudomonas species lack the
ability to synthesize ascorbate (31, 32), suggesting that ascorbate
is probably not the cognate-reducing cosubstrate for UndA. We
propose that in the in vitro assays, ascorbate acts mainly as
a reductant after 1-undecene formation by restoring the ferrous
species, and this role of ascorbate has previously been reported
(SI Appendix, Fig. S14A) (33–36). We thus favor the mechanism
in which the initial two-electron donating cosubstrate is LA in
the case of UndA (SI Appendix, Fig. S14A), analogous to
ascorbate in the case of ACCO and α-KG in many other known
examples (35). This mechanism is further supported by the ob-
servation that H2O2/Fe3+ failed to elicit UndA activity, in contrast
Phylogenetic analysis revealed that undA is ubiquitous and well
conserved in all of the sequenced Pseudomonas species and
several other closely related species, including species from
Acinetobacter, Burkholderia, and Myxococcus genera, with a total of
more than 1,500 homologs identified from published genomes (SI
Appendix, Fig. S15). No homolog of undA could be found in the
sequenced genomes of Shewanella, consistent with the observa-
tions that all tested Pseudomonas spp. produced 1-undecene, but
Shewanella spp. did not. Examination of the genomic contexts in
Pseudomonas further revealed that undA belongs to a conserved
two-gene operon (Fig. 1B), and the other conserved ORF (such as
PFL_4322 in P. fluorescens Pf-5 and PA0861 in P. aeruginosa
PAO1) encodes a protein with three domains: a sensory PAS
domain, a GGDEF-class diguanylate cyclase, and an EAL-class
phosphodiesterase. These domain activities are related to signal
recognition, and synthesis and degradation of cyclic di-GMP, an
important and ubiquitous second messenger in bacteria (37).
Recent studies revealed that homologs of PFL_4322 are related to
biofilm formation and dispersal in P. aeruginosa PAO1 and
P. fluorescens Pf0–1 (38, 39), and it is yet to be determined if
1-undecene, synthesized through the function of undA homolog in
the same operon, is also a semivolatile signaling molecule related
to biofilm formation and dispersal. 1-Undecene may also play
a role in bacterial–plant interactions, as many of the producing
organisms are plant-associated bacteria (40). Our discovery of
these conserved operons provides an intriguing opportunity to
study the physiological role of this ubiquitous hydrocarbon me-
tabolite in Pseudomonas, the abundant and widespread bacteria.
In summary, we revealed the genetic basis and molecular
mechanism for biosynthesis of 1-undecene, a ubiquitous hydro-
carbon metabolite of Pseudomonas. Our study discovered a pre-
viously unidentified family of nonheme oxidases that specifically
convert medium-chain fatty acids (C10–C14) into the corre-
sponding terminal olefins using an oxygen-activating, nonheme,
iron-dependent mechanism. Both biochemical and structural
analyses suggest an unusual mechanism of β-hydrogen abstraction
by a “less reactive” iron center during fatty acid activation. It is
notable that without strain optimization, overexpression of undA
homologs in E. coli produced 1-undecene at a titer over 25-fold
higher than the best titer in Pseudomonas (Fig. 1C). This result
lays an advanced foundation for pathway engineering, such as
improving the availability of medium-chain fatty acids or cognate-
reducing cosubstrate for the enzyme, to increase the production
titer and yield of MCAEs. Our findings expand the scarce enzyme
inventory for the transformation of fatty acid precursors to
hydrocarbons (2) and open the road for producing MCAEs, useful
as fuels and chemical building blocks, from renewable resources.
Materials and Methods
Bacterial Growth and Hydrocarbon Analysis. For 1-undecene production, 50 μL
of seed culture was used to inoculate 5 mL of LB medium. The culture was
shaken in a sealed 20-mL headspace vial containing a stir bar at 30 °C for 36 h.
After incubation, an SPME fiber (30 μm polydimethylsiloxane, Supelco, Sigma-
Aldrich Group) was manually inserted into the headspace vial and incubated
at 25 °C for 12.5 min for GCMS analysis. For details, see SI Appendix, Materials
and Methods.
Identification of the 1-Undecene Biosynthetic Gene. The fosmid genomic li-
brary of P. fluorescens Pf-5 was constructed using a pCC2Fos CopyControl
library kit (Epicentre Biotechnologies) (SI Appendix, Fig. S2). Pools of 10
clones were first analyzed for 1-undecene production, followed with an-
other round of screening to identify single 1-undecene–producing clones.
Three identified fosmids—6F8, 6E2, and 4F3—were sequenced to reveal
Rui et al.
PNAS
|
December 23, 2014
|
vol. 111
|
no. 51
|
18241

a 15-kb overlapping region that was further analyzed to reveal gene(s) re-
lated to 1-undecene production. For details, see SI Appendix, Materials
and Methods.
X-ray diffraction data were collected at beamline 8.3.1 at the Advanced Light
Source at Lawrence Berkeley National Laboratory. SI Appendix, Table S2
summarizes the statistics of data collection and refinement. For details, see SI
Appendix, Materials and Methods.
Biochemical Analysis of UndA. UndA purification, screening of possible sub-
strates and products, in vitro single-turnover and multiple-turnover reactions,
quantification of substrates and products, rate analysis, and so forth are
detailed in SI Appendix, Materials and Methods.
ACKNOWLEDGMENTS. We thank J. Klinman (University of California) for
discussions regarding reaction mechanisms, S. Bauer (University of California)
for assisting with GCMS analysis, N. Gilbert (Vanderbilt University) for early
structural modeling, E. Kemper and W. Bao for library screening, A. Arkin
(University of California) for providing Shewanella strains, C. Walsh (Harvard
Medical School) and S. Lindow (University of California) for providing Pseudomo-
nas strains, and D. Hung (Massachusetts General Hospital) for providing
P. aeruginosa PA14 nonredundant transposon insertion mutants. This work was
funded by grants from the Energy Biosciences Institute (to W.Z. and J.H.D.C.).
Crystallography. UndA crystals were grown at room temperature using the
hanging-drop vapor diffusion method in 0.1 M Mes, 1.8 M (NH4)2SO4, 0.2 mM
(NH4)2Fe(SO4)2, pH 6.5–7.0. To obtain ligand-bound structures, UndA was
incubated with 2.5 mM DEA or BHDA on ice for 15 min before crystallization.
1. Peralta-Yahya PP, Zhang F, del Cardayre SB, Keasling JD (2012) Microbial engineering
for the production of advanced biofuels. Nature 488(7411):320–328.
2. Lennen RM, Pfleger BF (2013) Microbial production of fatty acid-derived fuels and
chemicals. Curr Opin Biotechnol 24(6):1044–1053.
21. Dassama LMK, et al. (2012) O(2)-evolving chlorite dismutase as a tool for studying
O(2)-utilizing enzymes. Biochemistry 51(8):1607–1616.
22. Wang C, et al. (2013) Evidence that the fosfomycin-producing epoxidase, HppE, is
a non-heme-iron peroxidase. Science 342(6161):991–995.
3. Ray S, Rao PVC, Choudary NV (2012) Poly-alpha-olefin-based synthetic lubricants: A
short review on various synthetic routes. Lubr Sci 24(1):23–44.
4. Moiseenkov AM, Schaub B, Margot C, Schlosser M (1985) A new stereoselective syn-
thesis of (Z)-9-tricosene, the sex attractant of the common housefly. Tetrahedron Lett
26(3):305–306.
23. Karlsson A, et al. (2003) Crystal structure of naphthalene dioxygenase: Side-on
binding of dioxygen to iron. Science 299(5609):1039–1042.
24. Katona G, et al. (2007) Raman-assisted crystallography reveals end-on peroxide in-
termediates in a nonheme iron enzyme. Science 316(5823):449–453.
25. Jeoung JH, Bommer M, Lin TY, Dobbek H (2013) Visualizing the substrate-, superoxo-,
alkylperoxo-, and product-bound states at the nonheme Fe(II) site of homogentisate
dioxygenase. Proc Natl Acad Sci USA 110(31):12625–12630.
5. Kai M, et al. (2009) Bacterial volatiles and their action potential. Appl Microbiol
Biotechnol 81(6):1001–1012.
6. Schulz S, Dickschat JS (2007) Bacterial volatiles: The smell of small organisms. Nat Prod
Rep 24(4):814–842.
26. van der Donk WA, Krebs C, Bollinger JM, Jr (2010) Substrate activation by iron su-
peroxo intermediates. Curr Opin Struct Biol 20(6):673–683.
7. Labows JN, McGinley KJ, Webster GF, Leyden JJ (1980) Headspace analysis of volatile
metabolites of Pseudomonas aeruginosa and related species by gas chromatography-
mass spectrometry. J Clin Microbiol 12(4):521–526.
27. Breckau D, Mahlitz E, Sauerwald A, Layer G, Jahn D (2003) Oxygen-dependent cop-
roporphyrinogen III oxidase (HemF) from Escherichia coli is stimulated by manganese.
J Biol Chem 278(47):46625–46631.
8. Zechman JM, Labows JN, Jr (1985) Volatiles of Pseudomonas aeruginosa and related
species by automated headspace concentration—Gas chromatography. Can J Micro-
biol 31(3):232–237.
9. Graham JE (2013) Bacterial volatiles and diagnosis of respiratory infections. Adv Appl
Microbiol 82:29–52.
28. Layer G, et al. (2006) The substrate radical of Escherichia coli oxygen-independent
coproporphyrinogen III oxidase HemN. J Biol Chem 281(23):15727–15734.
29. Rocklin AM, Kato K, Liu HW, Que L, Jr, Lipscomb JD (2004) Mechanistic studies of
1-aminocyclopropane-1-carboxylic acid oxidase: Single turnover reaction. J Biol Inorg
Chem 9(2):171–182.
10. Bos LD, Sterk PJ, Schultz MJ (2013) Volatile metabolites of pathogens: A systematic
review. PLoS Pathog 9(5):e1003311.
11. Schirmer A, Rude MA, Li X, Popova E, del Cardayre SB (2010) Microbial biosynthesis of
alkanes. Science 329(5991):559–562.
12. Rude MA, et al. (2011) Terminal olefin (1-alkene) biosynthesis by a novel p450 fatty
acid decarboxylase from Jeotgalicoccus species. Appl Environ Microbiol 77(5):1718–1727.
13. Belcher J, et al. (2014) Structure and biochemical properties of the alkene producing
cytochrome P450 OleTJE (CYP152L1) from the Jeotgalicoccus sp. 8456 bacterium.
J Biol Chem 289(10):6535–6550.
14. Mendez-Perez D, Begemann MB, Pfleger BF (2011) Modular synthase-encoding gene
involved in α-olefin biosynthesis in Synechococcus sp. strain PCC 7002. Appl Environ
Microbiol 77(12):4264–4267.
15. Liberati NT, et al. (2006) An ordered, nonredundant library of Pseudomonas aerugi-
nosa strain PA14 transposon insertion mutants. Proc Natl Acad Sci USA 103(8):
2833–2838.
16. Jenkins AL, Zhang Y, Ealick SE, Begley TP (2008) Mutagenesis studies on TenA: A
thiamin salvage enzyme from Bacillus subtilis. Bioorg Chem 36(1):29–32.
17. Vaillancourt FH, Yeh E, Vosburg DA, O’Connor SE, Walsh CT (2005) Cryptic chlorina-
tion by a non-haem iron enzyme during cyclopropyl amino acid biosynthesis. Nature
436(7054):1191–1194.
18. Chang WC, et al. (2013) Mechanistic studies of an unprecedented enzyme-catalysed
1,2-phosphono-migration reaction. Nature 496(7443):114–118.
19. Mirica LM, McCusker KP, Munos JW, Liu HW, Klinman JP (2008) 18O kinetic isotope
effects in non-heme iron enzymes: Probing the nature of Fe/O2 intermediates. J Am
Chem Soc 130(26):8122–8123.
30. Mirica LM, Klinman JP (2008) The nature of O2 activation by the ethylene-forming
enzyme 1-aminocyclopropane-1-carboxylic acid oxidase. Proc Natl Acad Sci USA
105(6):1814–1819.
31. Smirnoff N (2011) Vitamin C: The metabolism and functions of ascorbic acid in plants.
Adv Bot Res 59:107–177.
32. Bremus C, Herrmann U, Bringer-Meyer S, Sahm H (2006) The use of microorganisms in
L-ascorbic acid production. J Biotechnol 124(1):196–205.
33. Myllylä R, Kuutti-Savolainen ER, Kivirikko KI (1978) The role of ascorbate in the prolyl
hydroxylase reaction. Biochem Biophys Res Commun 83(2):441–448.
34. Clifton IJ, et al. (2006) Structural studies on 2-oxoglutarate oxygenases and related
double-stranded beta-helix fold proteins. J Inorg Biochem 100(4):644–669.
35. Costas M, Mehn MP, Jensen MP, Que L, Jr (2004) Dioxygen activation at mononuclear
nonheme iron active sites: Enzymes, models, and intermediates. Chem Rev 104(2):
939–986.
36. Puistola U, Turpeenniemi-Hujanen TM, Myllylä R, Kivirikko KI (1980) Studies on the
lysyl hydroxylase reaction. I. Initial velocity kinetics and related aspects. Biochim
Biophys Acta 611(1):40–50.
37. Hengge R (2009) Principles of c-di-GMP signalling in bacteria. Nat Rev Microbiol 7(4):
263–273.
38. An S, Wu J, Zhang LH (2010) Modulation of Pseudomonas aeruginosa biofilm dis-
persal by a cyclic-Di-GMP phosphodiesterase with a putative hypoxia-sensing domain.
Appl Environ Microbiol 76(24):8160–8173.
39. Newell PD, Yoshioka S, Hvorecny KL, Monds RD, O’Toole GA (2011) Systematic
analysis of diguanylate cyclases that promote biofilm formation by Pseudomonas
fluorescens Pf0-1. J Bacteriol 193(18):4685–4698.
20. Bruijnincx PC, van Koten G, Klein Gebbink RJ (2008) Mononuclear non-heme iron
enzymes with the 2-His-1-carboxylate facial triad: Recent developments in enzymol-
ogy and modeling studies. Chem Soc Rev 37(12):2716–2744.
40. Blom D, et al. (2011) Production of plant growth modulating volatiles is widespread among
rhizosphere bacteria and strongly depends on culture conditions. Environ Microbiol 13(11):
3047–3058.
18242
|
www.pnas.org/cgi/doi/10.1073/pnas.1419701112
Rui et al.