Modular Pathway Engineering of Microbial Fatty Acid Metabolism for the
Synthesis of Branched Acids, Alcohols, and Alkanes
By
Micah J. Sheppard
B.S.E. Chemical and Biomolecular Engineering
University of Pennsylvania, 2008
Submitted to the Department of Chemical Engineering
r
in partial fulfillment of the requirements for the degree of
MASSACHUS
S OF TECI-NOLOGY
Doctor of Philosophy in Chemical Engineering
at the
JUN 3 0 2014
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
LIBRARIES
June 2014
@ 2014 Massachusetts Institute of Technology. All rights reserved.
Signature redacted
Signature of Author:
Department of 8i emical Engineering
May 12, 2013
Certified by:
Signature redacted
C,
Kristala L. J. Prather
Associate Professor of Chemical Engineering
Thesis Supervisor
Accepted by:
Signatu re redacted
Patrick S. Doyle
Department of Chemical Engineering
Chairman, Committee for Graduate Students
E
2
Modular Pathway Engineering of Microbial Fatty
Acid Metabolism for the Synthesis of Branched
Acids, Alcohols and Alkanes
by
Micah J. Sheppard
Submitted to the Department of Chemical Engineering
on May 12, 2014 in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy in
Chemical Engineering
Abstract
Historically, microbial platforms have been used to synthesize a variety of chemical products and
potential biofuels. More recently, increasingly complex metabolic pathways have been engineered by
using novel hosts, modifying natural pathways, and establishing de novo pathways with enzymes taken
from a variety of pathway contexts. Highly reduced and branched alkyl chains are potentially interesting
targets for both flavor and fragrance compounds and as liquid fuel components. Here we report the
engineering of microbial fatty acid synthesis to provide both CoA-dependent and fatty acid synthase
platforms for previously undescribed routes to medium-chain length, branched acids. Specifically we
produced six-carbon 4-methyl-valeric acid via a CoA-dependent route and nine-carbon 7-methyloctanoic acid via a fatty acid synthase. Specific variants of the platform pathways were used to
demonstrate synthesis of potential liquid fuel targets. The CoA-dependent platform was used to create
a redox-neutral pathway to 4-methyl-pentanol with a maximum theoretical energy efficiency of 100%.
Both platforms were used to demonstrate the first reported synthesis of short- and medium-chain
alkanes from three to seven carbons.
Thesis Supervisor: Kristala L. J. Prather
Title: Associate Professor of Chemical Engineering
3
Dedication
This thesis is dedicated to my wife, Katherine, who has had to sacrifice much and been a requisite
support mentally, emotionally, and spiritually during the past six years.
Acknowledgements
This work would not have been possible without the help of my advisor, Kristala Prather, committee
members, Gregory Stephanopoulos and Gerald Fink, and all my laboratory mates, past and present. I
want to thank my committee for giving guidance about the overall direction and helping me to keep a
view of the overall context of my work. I came to MIT hoping to work with Kristala and I am grateful for
the opportunity she gave me to work in her laboratory. I am thankful for the scientific guidance,
intellectual space, and positive support she gave me which were essential to completing this work.
More importantly, I am thankful that Kris treated me as a human being and supported my growing
family during my time at MIT. I have also cherished the lively and respectful conversations which have
been cultivated in the Prather Lab. Each of my colleagues in the Prather Lab has contributed to some
degree by discussing our work together. I have learned from each of you. Specifically, I would like to
thank Collin Martin, Hsien-Chung Tseng, and Himanshu Dhamankar for collaborations on CoAdependent pathways and Aditya Kunjapur for collaborations on carboxylic acid reductases and all of our
work on alkane synthesis. There are also a number of friends from Chemical Engineering who have
helped me with analytics and helped me grow as a scientist and a person while at MIT. Caroline Chopko,
Thomas Wasylenko, Yuko Kida, and Shawn Finney-Manchester all have provide insights about science
and life.
4
Contents
A b stra c t ..................................................................................................................................................
3
D e d ic a tio n ..............................................................................................................................................
4
Acknow ledge m e nts ................................................................................................................................
4
C on te n ts .................................................................................................................................................
5
Chapter 1: Introduction ..........................................................................................................................
9
Early m icrobial biofuels and solvents............................................................................................
9
Recent advances in biofuel pathw ay engineering..........................................................................
9
Finding an ideal bio-gasoline alternative .....................................................................................
13
Potential biological routes through fatty acid m etabolism ...........................................................
17
Pathw ay platform s to branched acids, alcohols, and alkanes .....................................................
20
Chapter 2: A CoA-dependent pathway to 4-m ethyl-pentanol ........................................................
22
Introduction ......................................................................................................................................
22
Pathway design .................................................................................................................................
24
M odule 1: Synthesis of isobutyrate from pyruvate.................................................................
24
M odule 2: Activation of isobutyrate to isobutyryl-CoA ..........................................................
25
M odule 3: Synthesis of 4M V from isobutyryl-CoA and acetyl-CoA ........................................
25
M odule 4: Reduction of 4M V to 4M P.....................................................................................
26
Pathw ay im plem entation .............................................................................................................
26
R e su lts ..............................................................................................................................................
26
Identification of enzymes required for CoA-dependent carbon chain extension (Module3)......26
4M P synthesis from 4M V (M odule 4) .......................................................................................
28
Incorporation of a pathway from glucose to isobutyryl-CoA (Modules 1 & 2).........................32
Improved pathway selectivity through alcohol dehydrogenase selection (Revisiting Module 4)37
D isc u ss io n .........................................................................................................................................
39
M ethods ...........................................................................................................................................
43
Bacterial Strains and Plasm ids...................................................................................................
43
Splice by Overlap Extension ....................................................................................................
44
Culture Conditions........................................................................................................................44
Relative Activity Assay for Purified His-Car ..............................................................................
45
M etabolite Analysis ......................................................................................................................
46
Chapter 3: A fatty acid synthase pathw ay to 7-m ethyl-octanoate .................................................
47
Introduction ......................................................................................................................................
47
Pathw a y Design ................................................................................................................................
50
5
M odule 1-SBA and precursor generation.................................................................................
50
M odule 2-7M O and elongation................................................................................................
51
M odule 3-0 and term ination ....................................................................................................
51
Results ..............................................................................................................................................
Initial dem onstration of branched fatty acid synthesis............................................................
51
Expression and engineering of FatB2.......................................................................................
52
Synthesis of 7-m ethyl-octanoate (7M C8) and octanoate (C8).................................................
55
Increased titers through accABCDEc overexpression .................................................................
57
Discussion..........................................................................................................................................
58
M ethods ...........................................................................................................................................
61
Bacterial strains and plasm ids ..................................................................................................
61
Gene knockouts for E. coli production strain ............................................................................
62
Gene integration in B. subtilis PY79 .........................................................................................
62
M D m edium recipe .......................................................................................................................
63
Culture conditions........................................................................................................................
63
Acid extraction/m ethyl esterification .......................................................................................
64
Gas chrom atography analysis...................................................................................................
65
W estern blot analysis...................................................................................................................
66
W estern blot buffers ....................................................................................................................
67
Chapter 4: M odular pathw ays to short-chain alkanes .....................................................................
68
Introduction.......................................................................................................................................
68
Pathw ay design .................................................................................................................................
71
Alternative precursor modules for CoA pathways (Modules 1-Pr and 1-1B)............................
72
CoA-dependent chain extension (M odules 2-M CC and 2-BC) ................................................
73
FAS m odules (M odules 1-SBA, 2-7M O, and 3-0) .....................................................................
73
Acid reduction and aldehyde decarbonylation (M odule 4A)...................................................
73
Results ..............................................................................................................................................
6
51
73
Dem onstration of CarNi/Adh6psc route to butanol and pentanol ............................................
73
Pentane synthesis from glucose ..............................................................................................
75
Propane and pentane synthesis from glucose .........................................................................
76
Butane and pentane synthesis from glycerol............................................................................
77
2-m ethyl-butane synthesis...........................................................................................................
78
Heptane synthesis from glucose .............................................................................................
80
Discussion..........................................................................................................................................
81
M ethods ...........................................................................................................................................
84
Bacterial strains and plasmids ...................................................................................................
84
C u ltu re co n d itio ns ........................................................................................................................
85
Gas chromatography method ..................................................................................................
86
Liquid chromatography method...............................................................................................
87
88
Chapter 5: Future directions ................................................................................................................
88
Improving the CoA-dependent platform for branched products..................................................
Key rate-limiting enzymes .......................................................................................................
88
Altering cofactor utilization .....................................................................................................
91
95
An orthogonal branched FAS............................................................................................................
96
Identifying an orthogonal ACP ................................................................................................
Pairing ACP and thioesterase activity to complete the pathway ...............................................
100
Improving alkane synthesis through alternative decarbonylases..................................................101
Alternative decarbonylases within the cyanobacteria AD family ..............................................
101
CER1-like plant aldehyde decarbonylases..................................................................................104
A p p e n dice s .........................................................................................................................................
10 7
Appendix 1: Evaluating fuel targets and gasoline composition .....................................................
107
Text A1-1: Energy density calculations for potential alcohol fuels ............................................
107
Text A1-2: Sample calculations of potential fuel compound pricing .........................................
107
Table A1-1: Composition of typical gasoline ..............................................................................
109
Appendix 2: CoA-dependent 4M P pathway ...................................................................................
110
Text A2-1: Pathway yield calculations ........................................................................................
110
Text A2-2: Codon optimized gene sequences ............................................................................
111
Table A2-1: Strains Used in Module 3 Screening.......................................................................115
Table A2-2: Plasmid construct descriptions ..............................................................................
116
Fig. A2-1: Detailed schematic of modules of the full 4M P Pathway ..........................................
118
Appendix 3: Modified lipid content from engineered FAS............................................................. 119
Text A3-1: Codon optimized FatB2ch gene sequence.................................................................119
Table A3-1: Plasmid construction...............................................................................................120
Fig. A3-1: Branched long-chain fatty acids from modified E. coli FAS........................................121
Fig. A3-2: Free fatty acid production with no-activator controls ...............................................
122
Appendix 4: Alkane construct descriptions and quantification methods ......................................
123
Text A4-1: Codon optimized ADpm..............................................................................................
123
Text A4-2: Alkane quantification................................................................................................
124
Table A4-1: Primers and plasmids..............................................................................................125
Fig. A4-1: Heptane GC standard curve .......................................................................................
126
7
Appendix 5: Identifying alternative enzymes for platform pathways............................................127
Text A5-1: Hidden Markov Model (HMM) used by adh-short (PF00106) Pfam family..............127
Table A5-1: Protein identifiers for PhaBcn node reductases ......................................................
128
Table A5-2: UniProt identifiers for potential NADH-dependent PhaB-like reductases..............129
R e fe re n ce s ..........................................................................................................................................
8
1 30
Chapter 1: Introduction
Early microbialbiofuels and solvents
While interest in microbially derived transportation fuels has piqued during the last decade,
technologies that harness microbial routes to alcohols were first used over a century ago. A number of
early automobile designs of the
the early
2 0 th
1 9 th
century relied on an ethanol fuel. Even the famous Ford Model T of
century used engines capable of burning ethanol'. Before the establishment of the
petroleum economy yeast fermentation of biomass-derived sugars provided a competitive fuel source.
Fermentations were also used for large scale solvent production. Bacterial strains which
produced butanol were first observed by Louis Pasteur as early as 18612. By 1914 a British chemist,
Chaim Weizmann, had identified Clostridia strains which could produce a variety of solvents including
butanol and acetone using the acetone-butanol-ethanol (ABE) pathway. Weizmann's interest was in
developing butanol production for synthetic rubber, but it was an acetone shortage during World War I
which pushed the development of large-scale fermentation of Clostridia. After the war the butanol
byproduct of the fermentation was used for automobile lacquer, and the bioprocess remained
commercially feasible into the 1950s 2. By the middle of the
2 0
th
century the petroleum chemical
industry had made ABE fermentation and other alternative chemical production economically
uncompetitive.
Recent advances in biofuelpathway engineering
Recently, scientific understanding of potential environmental effects due to greenhouse gas emissions
and a desire to establish energy independence from foreign sources of petroleum has led to renewed
interest in alternative sources for fuels and chemical products. The United States government has
passed legislation to support development of these alternative technologies. As part of that legislation,
a production target has been set of 14 billion gallons of advanced biofuels (cellulosic derived fuels which
can provide 80% reduction in life-cycle carbon emissions compared to petroleum) per year by 20223. As
9
a result, research interest in the use of microbial biocatalysts for the conversion of biomass into liquid
fuels has been renewed over the past decade.
On an industrial scale, the renewed interest in biofuels has led to increased production of
ethanol for fuel blending. Since the fuel crisis of the 1970s, the use of ethanol in fuel blends has been
increasing in response to incentives supporting both alternative energy and the agricultural industry4 .
This dual motivation exists because fuel ethanol has been predominantly derived from yeast
fermentation of corn sugars in the United States. In addition to the availability of ethanol producing
yeast strains, ethanol was selected because it serves as an environmentally friendly alternative to
4
methyl tert-butyl ether (MTBE) as an octane booster .
Biotechnology has matured as interest in biofuels has been rekindled. Over the last decade
there has been rapid growth in DNA sequencing technology and molecular biology techniques which has
allowed for a diverse set of rational and semi-rational engineering approaches for improving microbial
biocatalysts5' .
Genome sequencing data from diverse sets of organisms and increasingly cheaper DNA
synthesis methods have made it possible to move or design chimeric metabolic pathways in a microbial
host 7 . This has enabled transfer of non-native catabolic pathways and synthetic routes to non-native
metabolites into well understood host strains . As microbial biofuels have been envisioned as complete
replacements for gasoline, these new tools have opened up research into both improving existing
ethanol production strategies and in developing "next-generation" biofuels.
Low price points for fuels create small profit margins making increased process efficiency
valuable for any biofuel. As a result, much research has been focused on understanding how to push
yeast strains, like Saccharomyces cerevisiae, and bacteria, like Zymomonas mobilis, closer to theoretical
maximum yields and to use larger fractions of potential biomass feedstocks9,10,
11,12
In general this work
has been successful. S. cerevisiae strains have been engineered to produce ethanol at near 100%
theoretical maximum yields and to utilize xylose as a carbon source,
10
12.
Despite these advances,
challenges concerning biomass generation and pre-treatment remain even for well-established ethanol
platforms3.
Pathways to alternative biofuels have been explored in parallel to the development of improved
ethanol strains. The relatively low energy density, high corrosivity, and high hygroscopicity of ethanol
make it less compatible with the existing transportation fuel infrastructure. As evidenced by the butanol
pathways of Clostridia, there are alternative natural metabolites which could serve as fuels with
improved properties. C. acetobutylicum is not typically considered a modern industrial strain because it
makes a somewhat unpredictable mixture of fermentation products and exhibits lower yields for specific
products2. Motivated by increased yields and the lack of available molecular biology tools for
engineering Clostridia, Atsumi et al. first reported the transfer of butanol pathway genes to a more
tractable Escherichia coli host in 200814. During this period an explosion of new potential biofuel
pathways were demonstrated. While a variety of organisms have been used, E. coli and S. cerevisiae
hosts have been most common. The butanol pathway has been improved in E. coli and demonstrated in
S. cerevisiae as well as the bacteria Pseudomonas putida and Bacillus subtilis1 5'16,17.
Modifications of
this same pathway have been used to produce the heavier chain alcohols pentanol and hexano11 8'
19.
Identical chemistry was also used in an engineered reversal of the normally catabolic -oxidation cycle
to produce heptanol, octanol, nonanol, and decano
2
0
. Some branched alcohols have also been
demonstrated by using either the mevalonate pathway to isoprenoids or branched amino acid synthesis
coupled to the Ehrlich pathway for fusel alcohols21 '2,23
Replacements for diesel and jet fuel have also been targeted in pathway development. Many
naturally occurring lipid intermediates closely resemble petroleum derived diesel and jet fuel
components. In the mid-1990s a plant thioesterase from Umbellularia californica was discovered that
could be used to generate shorter free fatty acids from E. coli 2 4 . Since that initial discovery, a variety of
acyl-ACP (acyl carrier protein) thioesterases have been used to generate fatty acids in E. co/i 2 s, 26. These
11
longer chain fatty acids can be chemically esterified to form methyl or ethyl esters which are suitable as
biodiesel. Enzymes have been found that will catalyze the esterification in vivo, and others have been
identified that will convert fatty acyl-CoAs to fatty alcohols2
.
Isoprenoid synthesis has also been used to
produce diesel or jet fuel subsititutes in yeast28
These initial biofuel targets all contain some oxygen while petroleum based fuels are composed
of aliphatic and aromatic hydrocarbons. Oxygen-containing products were targeted first because true
hydrocarbons are rare in nature. Most pathway development started with known metabolic pathways.
Photosynthetic cyanobacteria are a phylum of bacteria which had been observed to produce alkanes2 9.
In 2010, Schirmer et al. used a genome comparison approach to identify the genes responsible for
alkane synthesis in cyanobacteria 30 . The genomes of cyanobacteria displaying alkane and non-alkane
phenotypes were compared and genes unique to alkane producers were identified. Of those genes only
a small number were of unknown function, and gene knockouts revealed two genes which were
essential for alkane production. It was found that the enzymes coded by this gene pair act on fatty acid
metabolites and generate an aldehyde intermediate which is decarbonylated to an alkane. Upon their
discovery, these genes were used to produce long-chain alkanes in an E. coli host
30
. Very recently,
variants of the pathway were used to produce shorter-chain and branched-chain alkanes in E. coli"' 32.
While a wide variety of products have been demonstrated, few have been developed to the
point of commercialization. Gevo, Inc. and Amyris Biotechnologies are two ventures which have moved
to commercialize platforms for isobutanol and isoprenoid production, respectively3 ' 34 .
Joule Unlimited
has developed a platform of engineered photosynthetic bacteria to produce ethanol with the potential
to produce other hydrocarbons 3 s. To date no single technology has been proven to be capable of
significantly displacing petroleum or even traditional ethanol platforms.
12
Finding an ideal bio-gasoline alternative
Of the potential liquid fuel targets in the United States, an alternative to gasoline could greatly reduce
petroleum reliance. The United States consumes 47.7% of the 22 million barrels of gasoline produced
by the world per day 36 . Gasoline also makes up 40% of total petroleum consumption in the United
States37 . While the numbers reveal the scale of petroleum dependence, they also highlight the
potential impact that efficient processes to renewable gasoline could provide.
Beginning with the earliest examples of alcohol fuels, microbial biofuels have been derived from
natural pathways. Only recently has biotechnology opened up the possibility to design de novo
biosynthetic pathways and most "next-generation" biofuel routes have relied heavily on natural
pathways to fuel-like compounds8 . An alternative strategy of characterizing the target compounds of
interest (gasoline components) and considering all potential biochemical space may lead to
identification of more efficient pathways to the original targets or near-exact replacements. By using
this rational approach pathways can also be designed to produce intermediates or alternate products of
value so a platform pathway can be created for production of multiple compounds. If this strategy is to
be employed for gasoline, the first step is to understand the characteristics and composition of the fuel.
Gasoline is a well distributed blend of more than 30 aliphatic and aromatic hydrocarbons (Table
Al-1)3 8, 3 9, 40 . The most abundant alkanes fall between four and seven carbons with both straight and
branched isomers represented (Fig 1-141). A variety of aromatics make up considerable fractions,
5.5
Fig. 1-1. Gasoline composition.
.1Isopentane
chemical
of
percentages
Mass
a
typical
components
of
petroleum
derived
gasoline 2.83
blend are shown. While more
chemical species are present 2.73
only the most abundant are
7.3
indicated in the legend and with
percentage labels on the chart.
n-butane
9.57
8.41
4.11
10.49
2,3-dimethylbutane
2-methyl-pentane
n-hexane
2,3-dimethylpentane
toluene
p-xylene
3,3,4-trimethylhexane
n-propylbenzene
1,3,5-trimethylbenzene
13
including toluene and p-xylene. This combination of compounds, often with ethanol blending, produces
an energy density of 32-35 MJ/L and research octane numbers (RONs) above
9042.
The energy density of
the fuel will determine mileage output given a fixed fuel tank volume and engine efficiency. Octane
number is a measure of how controlled a fuel will burn. In general straight-chain alkanes have a higher
tendency to autoignite causing engine knocking while branched alkanes burn smoothly
43 44
. Oxygen
content lowers energy density, but increases octane number. Lighter alcohols, like ethanol, will boost
octane rating but lower observed mileage as a result. Simply increasing straight-chain alcohol length
increases energy density but lowers octane rating (Fig. 1-2)43. Based on this knowledge of gasoline
composition and performance, the best bio-gasoline targets are highly branched alcohols or alkanes
which could be used to replace or blend with existing gasoline. Such compounds reach the energy
density of traditional gasoline in the six to seven carbon (C6-C7) range (Fig. 1-2A). Aromatic compounds,
like toluene (36.9 MJ/L, RON 121), could be excellent biofuels, but they are typically quite toxic to living
systems45.
A
32 --- -
normal
B
normal
34.
iso
110
- iso
- - - - - - -
- - - - - - - - - - - - - -
-- - - - - - - - - - - - -
- - - - - - - - - -
100-
30
Z
0
28
26
,
90 -
80-
C
'
24-
70-
22-
2
3
4
5
6
7
carbon number in primary alcohol
8
2
3
4
5
6
carbon number in primary alcohol
Fig. 1-2. Energy density and research octane number (RON) of potential alcohol biofuels. (A) Energy
density of primary alcohols increases with increasing carbon chain length. Branched isomers have similar
energy densities to straight-chain isomers. The energy densities of C2-C8 alcohols are plotted. A black
dashed line represents petroleum derived gasoline energy density and a blue dashed line represents the
energy density of a 10% ethanol blend. (B) Octane number decreases with increasing alcohol chain length.
Branched isomers have higher octane number. The black and blue dashed lines represent the RON of a
petroleum and 10% ethanol gasoline blend respectively. The red dashed line is an extrapolation of
observed RONs for branched isomers. It is likely that the RON of iso-hexanol (4-methyl-pentanol) falls
between 80 and 90. The data in (B) is adapted from reference 43, See Appendix Text Al-1 for description of
enerev densitv calculations
14
While branched C6-C7 alcohols and alkanes would perform well as fuels, it is important to
consider if a bio-based production process will ever be competitive with alternative technologies.
Whether using a microbial or chemical catalyst, bio-derived fuels will be likely converted from biofeedstocks (plant matter) which are predominantly composed of sugar polymers and lignin4 6 . A simple
economic analysis reveals why commercialization of biofuel targets has been difficult. Estimates of
current bulk chemical prices can be used to calculate the value of converting biomass derived sugars to
specific bulk chemicals 4 7' 48. Because energy must be conserved, a maximum molar yield of a fuel from
the model feedstock glucose is calculated by taking the ratio of the degree of reductance of glucose to
the degree of reductance of the fuel 49. Multiplying the molar yield by the price per mole of product
gives the maximum value of converting glucose to that bulk chemical
(VBXG),
given bulk chemical prices.
One can make an estimate of the value of converting glucose to a fuel compound for a gasoline
application using the price of gasoline. If one assumes gasoline value to be based on energy content,
the price per MJ of gasoline can be calculated using the energy density of gasoline. This energy
normalized price can be multiplied by the heat of combustion of a fuel target to give the price per mol
(energy normalized) for equivalent energy output. Multiplying by the molar yield gives a liquid fuel
conversion value
(VEXG).
Table 1-1 gives the values of converting glucose to a variety of compounds
which could be used as fuels. In general, alcohols like ethanol, isobutanol, and hexanol can be
marginally profitable depending on the chemical application targeted and the efficiency of the
production process. Conversion of glucose to alkanes retains roughly 30% of the original value. The
liquid fuel conversion value,
VEXG,
for converting glucose to any of the potential fuels, is $0.05 per mol
glucose. This value is constant for all potential fuels because we have normalized the price on an energy
basis and used maximum molar yields on an energy basis. The calculated value, $0.05 per mol of
glucose, can be thought of as the maximum value of converting glucose into a gasoline-like compound
given current fuel prices. In reality, different targets have different chemical characteristics which can
15
Table 1-1. Estimated value for glucose to fuel chemical conversions. The value of converting a glucose
feedstock to a variety of potential fuel compounds is shown. Current prices for glucose and potential fuel
targets are given on a molar basis. The maximum molar yield from glucose on an energy basis, YE, is found by
taking the ratio of the degree of reductance of glucose to the degree of reductance of the fuel. A maximum
conversion value from glucose using bulk chemical pricing, VBXG, is found by multiplying the price per mol bulk
chemical by YE. A maximum conversion value based on gasoline (energy normalized) pricing, VEXG, is found by
multiplying the price per mol of fuel by YE. The price per mol fuel is found using current gasoline prices to
calculate the price per MJ fuel and converting to a molar basis using the heat of combustion of the potential
fuel replacements. See Text A1-2 for sample calculations.
*estimated pricing from references 45 and 46
maximum
molar yield
price per
mol bulk
YE,
potential fuel
(feedstock)
degree of
reductance
from glucose
(energy basis)
chemical*
(feedstock)
(glucose)
(24)
-
($0.09)
ethanol
isobutanol
12
24
2.0
1.0
hexanol
hexane
36
38
heptane
44
VBXG,
bulk chemical
price per
mol fuel
VEXG,
liquid fuel
conversion value
per mol glucose
(energy
normalized)
conversion value
per mol glucose
$0.03-0.06
$0.10
$0.06-0.12
$0.10
$0.025
$0.05
$0.05
$0.05
0.67
0.63
$0.20
$0.04
$0.13
$0.025
$0.075
$0.08
$0.05
0.55
$0.05
$0.028
$0.09
$0.05
$0.05
either increase or decrease their value as a gasoline replacement. The lower bulk chemical ethanol price
of $0.03 per mol is a fuel application price. The resulting bulk chemical conversion value, VBXG=$0.06 per
mol glucose, is close to the liquid fuel conversion value, VEXG=$0.05 per mol glucose. In general straightchain alkanes are actually less valuable than gasoline because they have relatively low octane ratings
and they are used in a limited number of additional applications50 . This lower value is reflected in
VEXG
exceeding VBXG for hexane and heptane.
The above analysis highlights how inexpensive petroleum products remain. Oil is still readily
available and can be efficiently extracted 51. As a result, direct petroleum products, like alkanes, are
produced cheaply at large scale. Production of gasoline results from simply removing a distillation
fraction of petroleum5 2 . A reduction in sugar prices, an increase in petroleum product prices, or a
combination of the two is required to make biomass derived fuel production profitable. Alternatively,
cheaper carbon sources may also be used to make some targets profitable in the short-term at smaller
scale.
16
Compounds closely related to alcohol and alkane fuels are also used as flavors and fragrances.
These straight or branched, medium-chain acids, aldehydes, alcohols, and esters can be used to create a
range of flavors from butter or cheese to fruit or citruss3 ',4'.
As an example, the C6 branched acid 4-
methyl-valerate and its methyl ester can be used to produce fruity or floral flavors and fragrances.
Many of these compounds have higher value than fuels and, if made microbially, can be sold at a higher
price 55 . The ability to produce a range of products can support the development of microbial systems to
produce improved fuels or specialty chemicals.
Potential biologicalroutes throughfatty acid metabolism
Synthesis of branched C6-C7 alcohols or alkanes can potentially be achieved using general fatty acid (FA)
metabolism. Use of FA metabolism for production of fuel compounds is not new, but the structural
diversity of products has been limited. Most examples of alcohol, acid, and ester synthesis mentioned
above utilize some aspect of FA metabolism. Many of these routes are based on natural FA routes
which typically generate or consume straight-chain fatty acids. While much focus was placed on these
straight-chain variations of FA metabolism, significant structural diversity exists within natural FA
chemical space. If one considers polyketide synthase (PKS), fatty acid synthase (FAS), and fatty acid foxidation pathways many different acyl chain structures can be generated using the same core chemical
reactions. There are four key reactions which are used iteratively to build carbon chains from a variety
of precursors: condensation (keto synthesis), P-keto reduction, dehydration, and enoyl reduction (Fig 13).
In both FAS and PKS systems the four core reactions can be organized in different structures
built with enzymes of varying substrate specificity to create a wide range of potential acid products.
Type 1, 11, and Ill PKSs produce a wide range of cyclic and non-cyclic natural products. They utilize
different initiator substrates and elongation monomers in concert with varying degrees of reduction
17
Act.= ACP, CoA
R2
Act. R3Y
4
Act.+
RM
Act.+R3 =(CO 2,H)
condensation
(ketosynthesis)
Act.
NAD(P)H+W4-
R2
NADPH+H'
NADP*
NAD(P)*-
H20
Act.
r
+
P-keto reduction
form a P-hydroxy-acyl intermediate which can
Act.
H2 0
.t
dehydration
R6LAct.
NAD(P)H+H*
-
NAD(P)*-'
R2__
NADPH+H
NADP'
Act.
Fig. 1-3. Four reactions that form the basis of
FA metabolism. The four key reactions used
in all variations of fatty acid metabolism are
shown as a general pathway. In an anabolic
direction, two activated acids are condensed
to form a 0-keto-acyl intermediate releasing
an activator molecule and potentially C0 2.
The P-keto-acyl intermediate is reduced to
enoyl reduction
be dehydrated to form an enoyl intermediate.
The final reaction is a reduction of the enoyl
intermediate to a saturated thioester. The
activating molecules form thioesters with the
carboxyl group and can be either CoA or acyl
carrier protein (ACP) depending on the
pathway. ACP linked pathways are anabolic
while CoA linked pathways can be either
anabolic or catabolic.
between condensation reactions to produce spectacular diversity5 6 ' 57' 58. The ability to use branched
initiators and the branched elongation monomer methyl-malonyl-CoA opens the door for potential
synthesis of highly branched acids56. Unfortunately, most PKS systems form megasynthase enzyme
57
complexes which are difficult to engineer and generate relatively slow fluxes to final products . Type I
FAS systems of mammals and fungi also are organized as large multifunctional peptides or complexes
which makes it difficult to engineer the system for non-natural products'8'59.
Unlike PKS and type I FAS, type I FAS systems found in bacteria and plants use a set of individual
soluble proteins each with a single catalytic function4. While predominantly producing highly reduced
acyl chains, type II FASs generate a wide array of branching structure 1 , 62,63 Intriguingly, plant type 11
64
FAS derived 4-methyl-hexanol has even been observed from some select tobacco species . Of the
branched fatty acids produced in bacteria and plants, many are created by using branched acyl-CoA
initiators derived from branched amino acid synthesis. The bacterium B. subtilis uses this route to
65
produce branched fatty acids which can constitute greater than 90% of its membrane lipids . Based on
18
:
I
previously demonstrated use of acyl-ACP thioesterases with type 11 FASs, an engineered medium-chain,
branched FAS pathway appeared feasible at the start of this work.
Like type II FASs, synthetic "P-oxidation-like" pathways (e.g., butanol) also use individual soluble
proteins to carry out the same general chemistry used by fatty acid metabolism. Natural and
engineered pathways using this CoA-dependent chemistry have been observed for a variety of straight-
chain acids and alcohols14'19,66,67,6.
While branched species are less commonly observed for these
systems, the CoA-dependent pathways tend to limit chain extension. Identification of enzymes with the
ability to act on branched CoA thioesters would open up "P-oxidation-like" routes to the desired acid
precursors.
If acids can be generated using either FA pathway, identification of an enzyme which can
selectively reduce medium branched acids to aldehydes is required. While multiple natural systems for
aldehyde generation have been described (CoA-dependent Ald 69, FAR complex32 , Car7 0 ), activity on
branched C6-C7 thioesters or acids has not. Once an aldehyde is generated, alcohol dehydrogenases or
aldehyde decarbonylases can potentially be used to produce either alcohol or alkane products (Fig 1-4).
Alcohol dehydrogenases with varying substrate specificities are ubiquitous in nature, increasing the
likelihood of identifying one suitable for C6-C7 branched aldehydes71' 72 . As mentioned above, aldehyde
decarbonylases have only recently been identified in a smaller number of organisms. Finding a suitable
aldehyde decarbonylase may prove difficult. Acids can also be used directly as flavor/fragrance products
or processed through esterification to produce methyl, ethyl, or isobutyl esters.
o
R
aldehyde decarbonylase
PK1-
Fig. 1-4. Alternate products from branched
If C6-C7 branched acids are
R_
0acids.
synthesized, a carboxylic acid reductase can
carboxylic acid reductasease be used to generate an aldehyde
intermediate. The resulting aldehyde can
o
potentially be converted into either a
IN
alcohol
an
by
alcohol
alcohol dehydrogenase primary
H
(aldehyde reductase) dehydrogenase or an alkane by an aldehyde
decarbonylase.
OH
R_%
19
Pathway platforms to branched acids, alcohols, and alkanes
The following work explores the above hypotheses by engineering both FAS and CoA-dependent
pathway platforms for branched acid production in an E. coli host. We describe establishment of the
acid platforms and extension of those pathways to specific alcohol and alkane targets. Analysis of
pathway efficiencies to those targets is provided. We conclude by discussing limitations to the current
platforms and describe strategies for identifying improved pathways.
Chapter 2 reports the establishment of a CoA-dependent platform and its extension to make
branched acids. We demonstrate use of the platform to make the six carbon branched alcohol 4methyl-pentanol (4MP). A method of in vivo screening of enzymes along a modular pathway
architecture led to the development of a redox neutral pathway with the potential to reach high
theoretical energy yields. Through the screening process, CoA-dependent enzymes capable of acting on
branched intermediates were discovered. Additionally, a carboxylic acid reductase, carNi from Nocardia
iownesis, was found to preferentially reduce medium-chain length acids to aldehydes. The activity of
CarNi opened up the possibility of producing desired alcohol or alkane products. Identification of key
aldehyde and alcohol dehydrogenases resulted in a selective pathway for 4MP.
Chapter 3 describes an alternative FAS based platform pathway. A native E. coli FAS is
engineered to produce 7-methyl-octanoic acid through incorporation of B. subtilis FAS enzymes and a
plant acyl-ACP thioesterase FatB2 from Cuphea hookeriana (cigar plant). Engineered constructs
revealed the ability of FatB2 to act on branched acyl-ACP, but also showed that activity was limited to
acyl chains with a primary chain length of eight carbons. Results from trying to drive flux by increasing
precursor pools suggested that the thioesterase activity is limiting. We hypothesize that selection of an
optimal FAS and thioesterase pairing may be essential for high flux to a desired acid product.
Conversion of acids to alkanes from both CoA and FAS platforms is explored in Chapter 4. We
developed a set of pathway modules that are used to selectively produce C4 to C8 acids which are
20
reduced to aldehydes using CarNi- Variants of the aldehyde decarbonylase from Prochloroccocus
marinus MIT9313 were used to generate alkanes from C3 to C7. The branched alkane 2-methyl-butane
was produced from acid precursors in vivo suggesting aldehyde decarbonylase can be used to generate
exact gasoline components. Low alkane titers and build-up of aldehyde intermediates supported
observations made in the literature that aldehyde decarbonylase is rate limiting in alkane pathways.
Chapter 5 proposes future directions for the branched acid platforms. Promising leads for ratelimiting enzymes were identified by testing portions of the overall pathway in vivo. Strategies for
identifying alternative enzymes to improve these rate limiting steps are discussed. Specifically, finding
an alternative for the @-keto-acyl-CoA reductase of the 4MP pathway is stressed. Improvement of an
FAS pathway requires wider ranging investigation. Identification of previously uncharacterized FASs and
thioesterases using combined in vivo screening and bioinformatics techniques is suggested. Finally,
potential aldehyde decarbonylase alternatives are proposed based on combined literature data and
sequence analysis.
21
Chapter 2: A CoA-dependent pathway to 4-methyl-pentanol
The majority of results contained in Chapter 2 have been described in a manuscript currently under
review:
Sheppard, M.J., Kunjapur, A.M., Wenck, S.J., and Prather, K.L.J. "Retro-biosynthetic screening of a
modular pathway design achieves selective route for microbial synthesis of the gasoline substitute 4methyl-pentanol." Nat. Commun. In review as of March 14, 2014.
Introduction
As outlined in Chapter 1, a microbial biocatalyst used to generate liquid fuels must be maximally
efficient due to the generally low price point of conventionally derived alternatives. While it is desirable
to increase the energy density, reduce the hygroscopicity, and improve octane rating of potential biofuel
alternatives to gasoline, the efficiency of the pathway to such compounds cannot be sacrificed.
Proposed pathways to novel biofuels should strive for nearly perfect theoretical efficiency because
natural fermentative routes to ethanol are already maximally efficient 49 . Previous examples of microbial
synthesis of branched compounds near the energy density of gasoline have utilized relatively inefficient
pathways (Fig. 2-1).
A
B
leucine
valine
taf
02
0
0-
a-KAE
,-onlyre
keto-acid decarboxylase (kivD)
GLYCOLYSIS
'biosynthesis
G
co0uMd0
HMG4aA
I~uc
ebquivalants
C02
A--- acetoacetyl-CoA
ace
3
isoprenoid
-A
meaon
5-phosphdrievalonate
5-pyrophosphomevalonate
0
H
+
0
alcohol dehydrogenase (ADH6)
OH
OH
OH
i-butanol
3-methyl-butanol
4-methyl-pentanol
OH
isopentenol
Fig. 2-1. Inefficient alternative routes to C5 and C6 branched alcohols. (A) The a-KAE route to 4methyl-pentanol creates a redox imbalance with glycolysis by releasing one C02 for every carbon
added to the chain. Additionally a distribution of products is made with the mutants of LeuA and KivD
which produce the heavier alcohols. (B) The mevalonate pathway of isoprenoid synthesis can be used
to produce isopentenol with expression of a phsophotase. The mevalonate pathway is significantly
redox imbalanced producing 6 reducing equivalents (2 for each acetyl-CoA) with only 2 reducing
equivalents consumed.
22
Published examples have used either a-keto-acid elongation (ct-KAE) of amino acid precursors or
termination of the mevalonate route to isoprenoids to produce medium chain-length alcohols. Both
pathways create redox imbalances which greatly reduce the theoretical maximum energy yield49 . In the
first design, 2-isopropylmalate synthase, LeuA, from E. coli leucine biosynthesis and c-keto-isovalerate
decarboxylase, KivD, from the Lactococcus lactis Ehrlich Pathway were engineered to increase carbon
chain-length via a-KAE2 17 . Applying protein engineering to the pathway allowed production of longer
branched alcohols, but also generated a large distribution of products. More importantly, chainextension via ct-KAE only utilizes one carbon from every acetyl-CoA used which leads to the redox
imbalance. Longer species generated by this pathway are more redox imbalanced. The second design
utilizes diversion of the isoprenoid intermediates isopentenyl diphosphate (IPP) and dimethylallyl
pyrophosphate (DMAPP) by a phsophatase to generate isopenteno 2 1 . This C5 alcohol is not fully
reduced and the mevalonate pathway is severely redox imbalanced.
Here we present an adaptation of a CoA-dependent pathway platform for 4-methyl-pentanol
synthesis. The presented synthesis route combines a portion of a native pathway (valine biosynthesis)
with a ten step de novo pathway to produce 4-methyl-pentanol. In order to identify specific pathway
variants, we created a conceptual modular framework based on general natural chemistries.
Biosynthetic routes to alkyl chains most commonly employ a system of precursor generation followed by
chain elongation through carbon-carbon bond forming reactions74'75'76. We structured our conceptual
modules to correspond to this pathway structure. We envisioned precursor generating modules and
carbon chain elongation modules coupled to pathway terminating modules.
This approach produced a pathway with enzymes selected from nine different metabolic contexts
(organsims and/or pathways). Four enzymes were selected to act on their presumed cognate substrates
and 6 were applied to presumed noncognate substrates in the engineered pathway. The core Module 3
pathway architecture is based on synthetic CoA-dependent chemistry first understood in the ABE
23
IV
GLYCOLYSIS
potential precursor
1
reducing
equivalent
per pyruvate
generation
0
8n
A
Zamino acid synthesis
M-
p
3M-butyrate 2M-butyrate
i-butyrate
propionate
0
JIL
+ 0
IL
o
General design for a CoAFig. 2-2.
CoApathway.
platform
dependent
dependent chemistry can potentially be used
to extend a variety of branched- or straightchain acyl-CoA thioesters. Those thioesesters
can be generated from modified branched
amino acid synthesis or iterations of CoAdependent chain extension. By using acetylCoA precursors for two carbon extensions of
the carbon chain the pathway can achieve
redox neutrality. Two reducing equivalents
are produced for each acetyl-CoA generated
and two are consumed by the reduction steps
in the elongation phase. Unlike FAS enzymes
which generally favor continued iterative
elongating
CoA-dependent
extension,
enzymes act on limited shorter chain lengths.
chain
elongation
thioesterases
R10
acid products
pathway of Clostridium acetobutylicum. This CoA-dependent architecture has several advantages
compared to the xKAE and isoprenoid pathways described above. Biosynthetic CoA-dependent
pathways typically utilize acetyl-CoA building blocks and extend carbon chains through condensation
reactions without release of CO 2 (Fig. 2-2). The potential for generation of two reducing equivalents per
acetyl-CoA generated from glycolysis perfectly balances with those consumed for reduction to a primary
alcohol product
18,77.
Indeed, the presented pathway achieves redox neutrality. P-oxidation chemistry
has been used for synthesis of several straight-chain acids and alcohols
14,.17,19,20,78
Unlike these
previous demonstrations of CoA-dependent pathways, here we present the expansion of potential
products to branched alcohols of medium chain length using independently selected enzymes chosen to
enhance specificity for our desired intermediates.
Pathway design
Module 1: Synthesis of isobutyrate from pyruvate
Acetolactate synthase AlsS from Bacillus subtilis converts two pyruvate to acetolactate. Acetohydroxy
acid isomeroreductase lIvC from E. coli converts acetolactate to 2,3-dihydroxy-3-methylbutanoate and
24
dihydroxy-acid dehydratase IlvD from E. coli converts 2,3-dihydroxy-3-methylbutanoate to aketoisovalerate (aKIV). The ctKIV decarboxylase KivD from Lactococcus lactis converts ctKIV to
isobutyraldehyde which can be oxidized to isobutyrate by the aldehyde dehydrogenase Fjoh2967 from
Flavobacterium johnsoniae(Fig 2-3).
GLYCOLYSIS
H/VCaq 0
0
0-
alsSBS1
_I
0
pyruvate
j ivDEc
SkiVD
0
H i-buyldhyde
Fjoh2967Fj
0
0
0
CoA )CoI
i-butyrate
I
enaogenous
thioesterases
4-methyl-valerate
4-methyl-pentanol
I
Fig. 2-3.
Pathway schematic for CoAThe overall
dependent 4MP pathway.
pathway is broken into 4 recombinant
modules which were examined individually
and as sub-pathway combinations. Module 1
produces isobutyrate from pyruvate by
combining elements of valine biosynthesis
with elements of the Ehrlich Pathway to
isobutanol and an isobutyraldehyde specific
aldehyde dehydrogenase. Module 2 contains
~
an isobutyryl-CoA synthetase for activation of
isobutyrate to isobutyryl-CoA. Module 3 uses
-'CoA
CoA chemistry to condense acetyl-CoA and
isobutyryl-CoA and reduce the resulting @0
keto-4-methyl-valeryl-CoA thioester to the
saturated 4-methyl-valeryl-CoA thioester.
After cleavage by endogenous thioesterases,
the free acid, 4-methyl-valerate, is reduced by
A potential
Module 4 to produce 4MP.
byproduct could
result from
butanol
butyrate condensation of two acetyl-CoA. A detailed
pathway scheme showing cofactors and
potential byproducts can be found in
H Appendix 2 Fig. A2-1.
o~
OH
Module 2: Activation of isobutyrate to isobutyryl-CoA
The ATP-dependent isobutyryl-CoA synthetase IbuA from Rhodopseudomonas palustris converts
isobutyrate to isobutyryl-CoA. A second activator, a propionyl-CoA transferase Pct from Megasphaera
elsdenii was used initially for evaluating the activity of other modules.
Module 3: Synthesis of 4MV from isobutyryl-CoA and acetyl-CoA
The thiolase BktB from Cupriavidus necator (formerly Ralstonia eutropha) condenses isobutyryl-CoA
with acetyl-CoA to form 3-keto-4-methyl-valeryl-CoA. The hydroxyacyl-CoA dehydrogenase PhaB from
C. necator reduces 3-keto-4-methyl-valeryl-CoA to 3-hydroxy-4-methyl-valery-CoA which is dehydrated
25
by the enoyl-CoA hydratase PhaJ4b also of C necator. The trans-2-enoyl-CoA reductase Ter from
Treponema denticola further reduces 4-methyl-trans-2,3-pentenyl-CoA to 4-methyl-valeryl-CoA before
endogenous thioesterase activity (potentially from TesB and/or Ydil) cleaves the CoA producing 4MV.
Module 4: Reduction of 4MV to 4MP
The carboxylic acid reductase car from Nocardia iowensis reduces 4MV while consuming ATP and
NADPH to produce 4-methyl-valeraldehyde. 4-methyl-valeraldehyde can then be reduced to the
primary alcohol 4MP by one of two alcohol dehydrogenases: Adh6p from S. cerevisiae or Lsadh from
Leifsonia sp. Strain S749.
Pathway implementation
A series of strains were created to evaluate individual modules, module combinations, and full pathway
variants. Key strains are tabulated below as a reference (Table 2-1). Additional strains referenced in
Chapter 2 can be found in Appendix 2 Table A2-1. Plasmid construct details can be found in Table A2-2.
Results
Identification of enzymes required for CoA-dependent carbon chain extension (Module3)
While CoA-dependent chain extension was desired, enzymes using this chemistry with activity on our
desired branched intermediates had not been identified. Enzymes of the Clostridium acetobutylicum
butanol pathway and enzymes from polyhydroxyalkanoate pathways have previously been used to
synthesize a variety of straight-chain alcohols 15'18,78
These CoA dependent pathways have also been
used to synthesize straight-chain -hydroxyacids when a thiolase and reductase are expressed with a
thioesterase that is active on the -hydroxyacyl-CoA intermediates6 7 ' 68. We explored whether the
thiolase BktBCn, reductase PhaBCn, and thioesterase TesBEc could produce the branched hydroxyacid 3hydroxy-4-methyl-valerate (3H4MV) from glucose and isobutyrate. A set of different CoA activators
were compared to generate the necessary isobutyryl-CoA precursor 9 . Up to 300 mg/L of 3H4MV was
26
observed from 15 mM isobutyrate using the propionyl-CoA transferase pctme from Megasphaera
elsdeni 8 (Fig. 2-4).
Table 2-1. Strains Used for Module and Full Pathway Evaluation. Plasmids for strains are listed. Strain
names indicate the modules present in the strain, i.e. M1F2P34 includes "M" for modules, "1F" for Module 1
with feaBE, "2P" for Module 2 with pctm, "3" for Module 3 and "4" for Module 4. Strains with "( )" contain
abbreviations for operon structure indicating the order of alsSB, and iIvCEC. Key strains indicated in bold.
Plasmid 1
Plasmid 2
Plasmid 3
Strain name
pACYC-(carNrsfpBS)
-ADH6s,
M4
pACYC-(carNsfpBs)
-ADH6sc
pET-terTd-(bktBc,-pct m)
pET-(bktBcn-pctMe)(phaJ4bc,-phaBcn)
pCDF-phaJ4bcn-phaBc,
M2P34
pCDF-(ilvDE-terTd)-
pCOLA-kivDLrfeaBEc
M1F2P3
(asSB.S-i1vCEC)
pCOLA-kivDLrPUUCE
M1 P2P3
Plasmid 1
Plasmid 2
Plasmid 3
Plasmid 4
Strain name
pET-(bktBc-pct,)
-(phaJ4bcn-phaBc,)
pCDF-(IlvDE-terTd)
pACYC-(carNi-SfpBs)
pCOLA-kivDrfeaBE
M1 F2P34
-(alsSBs-ilvCE)
-ADH6sc
pCOLA-kivDLrpuuCEc
M1 P2P34
pET-(bktBc.-terTd)
pCDFpctme
-(phaJ4bc,-phaBc)pET-(bktBcf-terTd)
-(phaBc,-phaJ4bc,)
pACYC-(carNrsfpBs)
-ADH6s,
-
pCOLA-kivDLrfeaBEc
pCDF-(ibuAj,-ivDEC)
-(aIsSBs-iIvCE)
M1 F(AI)2134
pCDF-(ibuAp-ilvDE)
M1 F(IA)2134
-(ilvCEC-asSBS)
pET-(bktBcf-terTd)
-(phaBcn-phaJ4bc,)
PCDF-(ibuAp-ilvDE)
-(IvCEc-aISSBs)
M2P3b
pCOLA-kivDUrfeaBEC
PACYC-(CarN-sfPBS)
M1F(IA)2134a
pCOLA-kivDLI
-Fjoh_29 6 7 F
pACYC-(carrsfpBs)
-ADH6s,
pACYC-(carN-sfpBs
-IsadhLS
M1 Fj(IA)2134
MlFj(IA)2134L
The (S) specific reductase Hbdca from C. acetobutylicum was also tested in place of the (R) specific
PhaBco, but no 3H4MV was observed. In order to generate the saturated intermediate 4MV, we sought
an (R) specific dehydratase and an enoyl-CoA reductase which could act on our branched intermediates.
From enzymes documented to have activity on straight medium-chain CoA substrates, 4 PhaJ and 6 Ter
homologs were selected for further screening
81,82.
An assay was developed to screen for enzymes with
27
350Fig.
2-4.
3H4MV
from isobutyrate
and glucose. A hydroxyacid
pathway utilizing the tesBEC
thioesterase of E. coli produced
3H4MV when 4 different
PrpEst,
activators
(PctMe,
300
250 -production
ED200
150
r
Ptb/Bukca,
100-
Ptb/BukBs)
were
used. The condensation and Pketo reduction were catalyzed
by C. necator thiolase BktBcn
50BktB/PhaB
Pct/TesB
BktB/PhaB
PrpE/TesB
BktB/PhaB
Ca-Ptb/Buk
TesB
BktB/PhaB
Bs-Ptb/Buk
TesB
EV
Con.
and reductase PhaBcn-
Strain
the desired activity by isolating Modules 2 and 3 of our pathway in vivo with different combinations of
dehydratases and reductases.
Isobutyrate (10 mM) and glucose (1%) were supplied in LB medium, and
active gene combinations were identified by detecting 4MV secretion. PctMe was used to activate
isobutyrate. Of the 24 combinations tested, PhaJ4 homologs from Pseudomonas syringae,
Pseudomonas aeruginosa, and C. necator in combination with Ter homologs from Vibrio
parahaemolyticus and T. denticola produced 4MV (Fig. 2-5A).
The high producer, C. necator PhaJ4bcn/T. denticola TerTd (Strain M3Sc-TdCn), yielded 297 ± 45 mg/L
4MV and was selected for Module 3 moving forward. The previously used dehydratase Hbdca and
reductase Crtca from the Clostridium acetobutylicum butanol pathway were also tested in place of
PhaBcn and PhaJ4bcn (Strain M3Sc-Ca) (Fig. 2-5B). While some butyrate was produced by Strain M3ScCa, 4MV was not detected.
4MP synthesis from 4MV (Module 4)
Adoption of a CoA-dependent synthesis route required identification of pathways to link a saturated
CoA thioester to the final alcohol product. The CoA thioester could be reduced by either a CoAdependent alIdehyde dehydrogenase or thioesterase/carboxylic acid reductase pairing. CoA-dependent
aldehyde dehydrogenases have been described to act on a limited number of potential thioesterase
28
A 350
butyrate
4MtV
300
250
cc 200
1
igMdl
Module 3
Fig. 2-5. Identifying
Enzymes.
(A) Of the 24
combinations of P-hydroxyacyl-CoA
dehydratases (PhaJ) and transenoyl-CoA reductases (Ter) tested
in strains expressing pctme, bktBcn,
3
100 -and
50 -
O
empty M3Sc- M3Scvector TdPa4 TdPs
M3Sc- M3ScTdCn VpPa4
M3ScVpPs
M3ScVpCn
strain
B 1200
acetate
a
1000
4MVr
n
'-
phaBc,, six combinations show
activity for 4MV production. The C.
necator PhaJ4bcn and T. denticola
TerTd pair (M3Sc-TdCn) give the
highest 4MV titer. (B) The high
was
M3Sc-TdCn
producer
compared to a construct expressing
C acetobutylicum hbdc, and crtcQ
(M3Sc-Ca) instead of phaJ4bcn and
800-
terTd. While Hbdca and Crtca have
600
for
used
been
previously
production of straight-chain acids
butyrate and valerate, no 4MV was
detected.
400-
4(D
2000
M3Sc-TdCn
M3Sc-Ca
strain
substrates. Based on sequence homology and initial assays in our laboratory, it appears bifunctional
aldehyde/alcohol dehydrogenases with specificity for acetyl-CoA are found in a range of prokaryotes.
Clostridium acetobutylicum was found to contain two versions of the "adhE" gene with one enzyme,
AdhE2, having specificity for butyryl-CoA over acetyl-CoA8 3 . Alternatively independent CoA-dependent
aldehyde dehydrogenases (Ald) have been found in a subset of Clostridia8 4. When we tested both CoAdependent enzymes using an in vivo assay we did not observe substrate flexibility for a series of
branched substrates.
Carboxylic acid reductases (Car) provided an intriguing alternative. While acting on free acids,
different Car variants had been observed to have activity on a wider array of substrates8 5' 8 6 . We
hypothesized that, if a thioesterase could produce free 4MV from 4-methyl-valeryl-CoA a Car variant
along with an alcohol dehydrogenase could be used to convert 4MV to 4MP. Based on other work in
our laboratory we believed our E. coli host expressed thioesterases capable of cleaving the 4-methylvaleryl-CoA bond ''8.
Recently, a carboxylic acid reductase (Car) from Mycobacterium marinum was
29
shown to convert a range of straight-chain fatty acids to fatty aldehydes, but with increasing activity for
longer chain lengths8 5. A previously studied homolog, Car from N. iowensis, was found to have activity
on a broad range of acids86 . Because a Car with specificity for medium-chain branched acids was
desired, CarNi from N. iowensis was selected for further study.
A
Fig. 2-6. Tuning of pathway selectivity by
the carboxylic acid reductase CarNi. (A)
In vitro analysis of his-purified CarNi
reveals a dependence on acid primarychain length with maximum activity at a
1.0.
- 0.9U)0.8-
0.7-
chain length of five and six carbons.
Branching at the C4 position is preferred
significantly over straight acid species.
The potential substrates for byproduct
formation, butyrate and isobutyrate, are
seen to have 56% and 25% of the
observed activity on 4-methyl-valerate
(4MV) respectively.
(B)
MichaelisMenten kinetics for isobutyrate and 4MV
reveal that CarN has a strong preference
for the latter intermediate.
C0.5-
0.40.3--
e
0~
B
substrate
isobutyrat
4-methylvalerate
1
k~(soc )
Km, (mM)
N1.40.1
78 9
2.5 ± 0.3
0.31± 0.08
k,,K,, (sec' mM')
0.018 0.002
8.1 ± 2.3
While CarNi has previously been assayed, its activity on a range of aliphatic acids was not examined.
Assays were devised to confirm activity on desired substrates in vitro and in vivo. First, His-tagged CarNi
was purified and assayed for relative activity on 13 straight and branched acid substrates from C2-C8.
CarNi shows a peak in activity for acids with a primary chain-length of 5-6 carbons (Fig. 2-6A). The
highest CarN activity was found for the branched species 4MV and 4-methyl-hexanoate. Our pathway
design utilizes an isobutyrate precursor and could potentially generate a butyrate byproduct. In
addition E. coli will generate free acetate from glycolytic overflow. Given the need to reduce flux of
precursors to undesired byproduct alcohols, CarNi was a logical selection because of its preference for
4MV over the short-chain acids acetate, isobutyrate, and butyrate. The strong specificity of CarNi over
the isobutyrate precursor was of particular importance. The Michaelis-Menten kinetic parameters of
CarNi were found using these two key intermediates to confirm the desired specificity (Fig. 2-6B). The
30
kcat/Km ratio with 4MV was found to be 450 times higher than with isobutyrate, indicating the significant
preference of CarNi for 4MV over other acid substrates generated by the pathway. Having a Km of 78 ± 9
mM with isobutyrate, CarNi is expected to convert isobutyrate poorly under physiologically relevant
concentrations, which limits shunting of the precursor to isobutyraldehyde.
A diverse set of alcohol dehydrogenases have been identified in nature. In S. cerevisiae alone, seven
different variants have been identified 8 . We selected S. cerevisiae Adh6p to pair with CarNi based on
observed in vitro and in vivo activity on branched aliphatic substrates23' 8 9 . An initial in vivo assay
monitored acid to alcohol conversion by Strain M4 expressing corN and ADH6sc for five different straight
and branched-chain acids. While additional effectors (transport, toxicity) could be reflected in the in vivo
results, in general the observed in vitro activity trends held in vivo with a complete Module 4 (Fig 2-7A).
A
B
2.5-
acetate
2500 250i-butanol
2.0-
Tbutanol
butyrate
4MP
2000
0
250 -
E
1.0
200 -
5
1).
5 -)
S0.5 -
0.0 50
acid fed
Strain M2P34
(A) Strain M4 expressing carN-sfpBsactivity.
4
Module
to
confirm
Fig. 2-7. An in vivo assay was developed
ADH6sc was fed five different acids at 10 mM initial concentration. Seletivity was determined by measuring
normalized titers 5 hours post induction of gene expression. As observed by in vitro analysis, the reduction
rate appears to peak at species with a primary carbon chain length of 5 carbons. (B) When Modules 2, 3,
and 4 are combined in vivo, CarNi substrate preference influences product selectivity generating 1.9 times as
much 4-methyl-pentanol (4MP) (272 mg/L, 2.7 mM) as butanol (142 mg/L, 1.9 mM) from Strain M2P34
when supplied with both glucose and isobutyrate. Even though 10 mM isobutyrate is supplied to the
cultures of Strain M2P34 only 111 mg/L (1.5 mM) isobutanol is observed.
31
With Module 4 in vivo activity confirmed, 4MP production from glucose and isobutyrate was tested
with a strain expressing Module 2, 3, and 4 genes (Strain M2P34). As predicted by observed activities
for CarN, and Adh6psc, Strain M2P34 preferentially produced 4MP (272 ± 7 mg/L) over isobutanol (111 +
7 mg/L) and butanol (142 ± 9 mg/L) even while feeding 10 mM (870 mg/L) isobutyrate (Fig. 2-7B).
Incorporation of a pathway from glucose to isobutyry-CoA (Modules 1 & 2)
With acetyl-CoA coming from glycolysis, the final portion of the pathway required for 4MP synthesis
from glucose was the synthesis of isobutyrate. Initially we observed synthesis of isobutyrate using
overexpression of only alsSBs and ilvCDEc. This was possible because the acetolactate synthase, AIsSBS,
also displays weak a-KIV decarboxylase activity90 . Unfortunately we found that the poor decarboxylase
acitivity resulted in isobutyrate being generated relatively slowly. The isobutyrate supernatant titer
increased slowly after rapid a-KIV synthesis. We tried to increase isobutyryl-CoA synthesis directly from
a-KIV by expressing the branched keto-acid decarboxylase complex IpdVbcdA1A2BBs from B. subtilis, but
we did not observe functional acitivity91 . An alternative route to isobutyrate had been described using
the dedicated decarboxylase KivDLu in combination with aldehyde dehydrogenases from E. coli9 2 . We
cloned a series of 3 plasmid sets that expressed the Module 3 genes, the activator PCtMe, kivDL,, and one
of four E. coli aldehyde dehydrogenases (gabDEc, betBEcfeaBE, andpuuCEC). Of the four E. coli triple
transformants tested, only those expressing puuCEc andfeaBEc produced 4MV from glucose with titers
up to 111 ± 11 mg/L (puuCEc) and 90 ± 9 mg/L (feaBEC) (Fig. 2-8A).
Both dehydrogenases were used for alternate versions of the full pathway, and only the feaBE strain
(Strain M1F2P34) produced 4MP (67 ± 13 mg/L) while the puuCEc strain (Strain M1P2P34) produced 4MV
(67 ± 11 mg/L) (Fig. 2-8B). Additionally the puuCEc strain produced more butyrate (156 ± 4 mg/L) and
less butanol (15 ± 6 mg/L) than the feaBEc strain (62 ± 15 mg/L butyrate, 49
+ 17
mg/L butanol). While a
demonstration of 4MP synthesis from glucose was made, relatively low 4MP titers and high isobutyrate
32
(1113 ± 34 mg/L) and isobutanol (2205 ± 225 mg/L) titers suggested there were possible bottlenecks
even in the best performing strain, Strain M1F2P34.
A
B220
1750
isobutyrate
butyrate
4MV
isobutanol164M
2
30
IE
f
isobutanol
2500
2000
120
E 1500
100
8060-10
300
- 200-
40500
100
01
acetate
isobutyrate
3000.
4MP
140
1500 -
_
butryate
4MV
butanol
200
180
160
20
M1P2P3
strain
M1F2P3
0
2 3
M1P P 4
M1F 2 P3 4
strain
0
M1P 2 P34
M1F 2 P3 4
strain
Fig. 2-8. 4MV and 4MP synthesis from glucose using pctme in Module 2. (A) Combing valine
biosynthesis with expression of kivDLI, and puuCEc or feaBE (Strains M1P2P3 and M1F2P3, respectively)
produces 1668 ± 18 mg/L of isobutyrate and 111 ± 11 mg/L of 4MV (M1P2P3) or 1532 ± 40 mg/L of
isobutyrate and 90 ± 9 mg/L of 4MV (M1F2P3). (B) Extended acid and alcohol products from initial full
pathway strains M1P2P34 (puuCE) and M1F2P34 (feaBE) using the pctMe activator showed 4MP
production only with expression of feaBEc. Large titers of isobutanol were also observed.
At this point new plasmid constructs were created to organize module gene sets on separate
plasmids for easier assaying of desired module combinations. The new plasmid sets were used to
identify potential limiting enzymes and to screen for alternatives (Fig 2-9). The ATP-dependent
isobutyrate activator, IbuARp, was used in place of the CoA transferase PctMe in order to relieve acetylCoA requirements and create redox neutrality for the pathway. It was anticipated that operon
construction would reduce enzyme expression, especially for genes in the second position, but it was
unknown if the effect would be detrimental to overall production without knowledge of the rate limiting
enzyme 93. Two Module 3 plasmid variants were tested to explore whether PhaBcn activity could become
limiting when expressed from an operon used in the new constructs. The variant with phaBc, in the first
position of a two gene operon performed best supporting the theory that phaBc, could be the limiting
activity within Module 3 (Fig. 2-10A). Subsequent SDS-PAGE analysis confirmed increased phaBc,
expression when in the first operon position (2-10B). Additionally, operon variants for aISSBS and iIvCEC
expression were tested to examine if better balancing of flux between Module 1 and the native acetyl33
terr phaJ4b
pct
bktBc
)
ter
.
bktBcn
pET
tero
pheaBc
pCDF
bktBcn pET
phiaJ4bn
~~pEcti
ilVCEC
phaBcp
haBc,
asS
ibuARp
asS
ilvDEc
IVC
p
pCDF
phaJ4bcn
ibuARp
ilvCE,
VC
pCDF IvDEc
pCDF
terrd
phaBcn
bktBcn
-
pET
phaJ4bc,
Fig. 2-10
Companson of phaBn
phaJ~bc, operon design
-
kivDL
Fig. 2-11
Comparison of alsS 3I
-
CD
kivDh2967F;
AB
pFO
9
Fig. 2-12
Corarison of feaBEC &
Foh96 7,j aldehyde
dehydrogenases
-
W~CEc operon design
k
feaBE
Fig. 2-13
Comparison of ADH6sc
sps
carNi
sfpS
pACYC
carN
ADH6sC
pACYC
IsadhLs
sIO,
carNi
pACYC
& IsadhLS alcohol__________ ee
dehydrogenases
Fig. 2-9. Plasmid constructs used for operon design evaluation and enzyme screening. Plasmid maps are shown
indicating the Duet backbone and gene inserts at MCS-1 and MCS-2. Inserts showing genes in tandem represent
synthetic polycistronic operons. Insert colors correspond to modules in Fig. 1. Connecting bars indicate plasmids
co-transformed for experiments discussed below with data presented in the indicated figures. In brief, the gene
order of two different operons was compared (phaBcn/phaJ4bc, and alsSBS/ivCEC) and alternate genes were
screened for both the aldehyde dehydrogenase (feaBEc, Fjoh2967F;) and final alcohol dehydrogenase (ADH6s,
Isadhj).
CoA pathway could improve 4MP production (Fig. 2-11). Placing aisSBs in the second position while using
the new plasmid constructs (Strain M1F(IA)2134) increased 4MP titers (168 ± 31 mg/L) while reducing
isobutyrate (290 ± 24 mg/L) and isobutanol (1046 ± 45 mg/L) titers.
Based on available in vitro data and the presence of 4MV (42 ± 7 mg/L) even for improved Strain
M1F(IA)2134, it was possible that FeaBEC could be oxidizing 4-methyl-valeraldehyde into 4MV creating a
futile cycle with CarNi (Fig. 2-12A). An aldehyde dehydrogenase, Fjoh2967Fj from Flavobacterium
johnsonaie had been found to prefer an isobutyraldehyde substrate over other aldehyde substrates
when tested in vitro
9.
ReplacingfeaBEc with Fjoh2967Fj in Strain M1Fj(IA)2134 led to increased 4MP
(193 ± 23 mg/L) and elimination of detectable 4MV (Fig. 2-12B). Isobutyrate titers (424 ± 9 mg/L) were
increased and isobutanol titers (797 ± 7 mg/L) were reduced (Fig. 2-12C).
34
A
700.
B
isobutyrate(residual)
T
butyrate
4MV
600-
&
empty
vector
- 300-
9
4)
4
Co 250200.
15010050U.
M3Sc-TdCn
M2P3a
M2P3b
75
strain
Fig. 2-10.
Improving pathway performance through
altering phaBcn-phaJ4b operon design.
(A) Multiple
Bktbc,
plasmid constructs were made for expression of Module
& Terrd
37
3 genes with Module 2P (pctMe) to help improve
expression of potential rate limiting enzymes. The initial
two-plasmid construct used in screening, M3Sc-TdCn,
produced 257 ± 33 mg/L of 4MV and 208 ± 14 mg/L
25
PhaBcn
P
butyrate. The Module 3 genes were re-cloned in operons
in plasmid pET-(bktBc-terTd)-(pha4bcf-phaBc,).
Strain
PhaJ4bcn
M2P3a harboring the new plasmid, produced 4MV and
butyrate titers that were 61% and 97% of those observed
for M3Sc-TdCn, respectively. Swapping the order of phaJ4bc, and phaBce in their synthetic operon in Strain M2P3b
led to 4MV titer increases up to 120% of the level observed in Strain M3Sc-TdCn (B) ) Increased phaBCn expression
from the phaBCn-phaJ4bCn operon was confirmed by expression of pET-(bktBCn-terTd)-(phaJ4bCn-phaBCn) and
pET-(bktBCn-terTd)-(phaBCn- phaJ4bCn) plasmids in the production strain MG1655(DE3) AendA ArecA. Lysate
from the construct expressing pET-(bktBCn-terTd)-(phaBCn- phaJ4bCn) contains a significantly higher PhaBCn
concentration when visualized by SDS-PAGE. Molecular weights of the Module 3 proteins are: BktBCn, 42.5 KD,
TerTd, 43.8 KD, PhaBCn, 26.4 KD, and PhaJ4bCn, 16.9 KD.
220200180160-
butryate
4MV
butanol
4MP
3000-
I
140 -D
E
24-a
2500-
-
1201008060-
acetate
isobutyrate
isobutanol
2000-
E) 1500-
1000-
4020 -
500-
0M1 F(AI)2134 M1 F(IA)2134
strain
0M1F(AI)2134 M1F(IA)2134
strain
Fig. 2-11. Improving pathway performance through altering aIsSBS-iIvCEC operon design. In an attempt to better
balance flux between acetyl-CoA and isobutyryl-CoA, new constructs were made containing the full pathway
usingfeaBEc and the isobutyryl-CoA synthetase gene ibuARp. Strains M1F(AI)2134 and M1F(IA)2134 expressed the
first Module 1 gene, alsSBS, in either the first or second position of a two gene operon, respectively. The expected
lower
aIsSBs
expression in Strain M1F(IA)2134 led to decreases in isobutyrate and isobutanol byproducts and
increased 4MP.
35
AMP+PP,
ATP
4MV
-butanol
B
220
M butryate
4MV
butanol
4MP
-
200
180
160
140
C 1250
acetate
isobutyrate
isobutanol
1000
0-
75
120
E 100
80
60.
E
'00
40
250
20
0
MlF(IA)2134 MlFj(IA)2134
strain
M1 F(IA)2134 M1 Fj(IA)2134
strain
Fig. 2-12. 4MP synthesis from glucose improved through aldehyde dehydrogenase selection. (A) Key
reactions involving aldehydes can generate futile cycles (aldehyde dehydrogenase FeaBEc with the carboxylic
acid reductase CarNi) or byproduct shunts (alcohol dehydrogenase Adh6psc). The desired pathway route is
shown in bold arrows within the shaded box. Undesired reactions are shown with dashed arrows. Enzymes
with insufficient selectivity have dashed outlines. (B) When feaBEC (Strain M1F(IA)2134) was replaced with
the isobutyraldehyde specific aldehyde dehydrogenase gene Fjoh2967j in Strain M1Fj(IA)2134, 4MV titers
were reduced and 4MP titers were increased. (C) Complementarily, Strain M1Fj(IA)2134 (Fjoh2967j)
produced lower isobutanol and higher isobutyrate titers relative to strain M1F(IA)2134 (feaBEC).
In order to examine if alcohol toxicity could be limiting product titers, toxicities of the dominant
byproduct isobutanol and desired product 4MP were assayed through exogenous addition of alcohols to
the growth medium at concentrations from 1-10 mM. Isobutanol concentrations and 4MP
concentrations up to 5 mM did not inhibit the exponential growth rate (Fig. 2-13). A combination of 10
mM (741 mg/L) isobutanol and 2 mM (204 mg/L) 4MP (comparable to titers observed for Strain
M1Fj(IA)2134) only reduced the exponential growth rate by 10%, the same reduction observed with 10
mM isobutanol alone. While endogenously produced alcohols may be involved in alternative toxicity
mechanisms, this result suggests that the observed extracellular alcohol concentrations should not
greatly inhibit growth.
36
0.5- -- X- control
-m10 mM isobutanol & 2 mM 4MP
-0-- 1 mM isobutanol
0.0 - -,x-- 5 mM isobutanol
-0-- 10 mM isobutanol
1 mM 4MP
----0.5- -A- 5 mM 4MVP
-0- 10 mM 4M
0
-1.0-
C
-1.5-2.01.0
1.5
2.0
2.5
3.0
Culture Condition
Growth
Rate (hr 1 )
Final OD600
(30 hrs)
control
1.04 ± 0.01
2.20 ± 0.21
10 mM isobutanol
&2mM4MP
0.92 ± 0.04
1.49 ± 0.09
1 mM isobutanol
0.97 ± 0.01
2.14 ± 0.06
5 mM isobutanol
1.06 ± 0.01
2.17 ± 0.25
10 mM isobutanol
0.91 ± 0.04
1.94 ± 0.01
1 mM 4MP
1.08 ± 0.03
2.03 ± 0.03
5 mM 4MP
1.02 ± 0.04
1.49 ± 0.01
10 mM 4MP
0.62 ± 0.03
1.65 ± 0.24
3.5
time (hrs)
Fig. 2-13. Toxicity comparison of isobutanol and 4MP on MG1655(DE3) AendA ArecA. (A) MG1655(DE3)
endA recA was cultured in an LB medium with 1.2% glucose and varying concentrations of isobutanol or 4MP
(1 mM (74 mg/L), 5 mM (370 mg/L), or 10 mM (741 mg/L) isobutanol or 1 mM (102 mg/L), 5 mM (511 mg/L),
The exponential growth rate was unchanged by 4MP concentrations
or 10 mM (1022 mg/L) 4MP).
exceeding those observed in final titers (193 ± 23 mg/L) for the high producing Strain M1Fj(IA)2134. The
highest 4MP concentration (10 mM, 1022 mg/L) did lead to a reduction in growth rate of 40%. The highest
concentration of isobutanol, 10 mM (741 mg/L), led to a reduction in the growth rate of 10%. While 4MP is
more toxic than isobutanol at equimolar concentrations, it is also significantly more hydrophobic. The
solubility of 4MP in water (7.4 g/L) is an order of magnitude below that of isobutanol (70 g/L). If higher
titers of 4MP are achieved, in situ gas stripping or simple phase separation could potentially be used to
extract 4MP from the culture medium.
Improved pathway selectivity through alcohol dehydrogenase selection (Revisiting Module 4)
With knowledge of the specificity of CarNi for 4MV over isobutyrate, the continued high isobutanol titers
suggested Adh6pse was converting the isobutyraldehyde intermediate to isobutanol (Fig 2-14A).
Removing ADH6sc from Strain M1F(IA)2134 produced Strain M1F(IA)2134a which generated an isobutanol
titer of 27 ± 3 mg/L with nearly undetectable butanol and 4MP titers (Fig. 2-14B). An alternative to
Adh6psc was identified from Leifsonia sp. Strain S749. The new alcohol dehydrogenase LsadhLS was
hypothesized to have improved specificity for 4-methyl-valeraldehyde based on substrates that were
assayed in vitro 9.
When LsadhLs was combined with the isobutyraldehyde specific dehydrogenase
Fjoh 2 967j in Strain M1Fj(IA)2134L, selective synthesis of 4MP was achieved over other alcohol
byproducts (Fig. 2-14C). lsobutanol titers were reduced to 21 ± 3 mg/L, similar to those observed with
37
the no alcohol dehydrogenase control. 4MP was produced at 90 ± 7 mg/L making up 81% of all
observed alcohol products. The dominate byproduct was the 4MP precursor isobutyrate, suggesting
that byproduct shunts were reduced and further improvement could be made by relieving a
downstream rate limitation in Module 2, 3, or 4 (Fig. 2-14D).
B 1200.
A
-bat"800
isobutanol
10001
0-
butanol
4MP
C;-J 200
AMP
ATP
E
PP
I
t'anoi
100
.
0
M1F(IA)2134a
M1F(IA)2134
strain
C 1200
D
1000
isobutanol
butanol
4MP
800
1100
acetate
1000
butryate
4MV
isobutyrate
9004M
200
E
E
-i)
800
700
isobutanol
600
4MP
butanol
500
400
30
1
S100
0
M1Fj(IA)2134
M1Fj(IA)2134L
strain
300200
100
0
M1Fj(IA)2134
MFj(IA)2134L
strain
Fig. 2-14. Improved alcohol dehydrogenase selectivity with Isadhs. (A) The desired pathway reactions to
4MP are indicated by bold arrows with the byproduct shunt to isobutanol indicated by the dashed arrow.
High activity of Adh6psc on isobutyraldehyde diverts isobutyrate flux to isobutanol. The LsadhLs alcohol
dehydrogenase's selectivity for 4-methyl-valeraldehyde greatly reduces flux to the isobutanol shunt. (B) The
alcohol profile of Strain M1F(IA)2134 expressing feaBEc and ADH6sc contains 168 ± 31 mg/L of 4MP but is
dominated by 1.046 ± 45 g/L of isobutanol. The M1F(IA)2134a control without ADH6sc expression produces
low to undetectable levels of all three alcohols. (C) Replacing feaBEc with Fjoh2967j in Strain M1Fj(IA)2134
reduces isobutanol (797 ± 20 mg/L) and increases 4MP (192 ± 23 mg/L) marginally. Replacing ADH6sc with
IsadhL greatly enhanced alcohol selectivity producing 90 ± 7 mg/L 4MP with only 20 ± 5 mg/L isobutanol. (D)
Acid and alcohol product profiles for Strains M1Fj(IA)2134 and M1Fj(IA)2134L show that with the more
specific LsadhLS reduced isobutanol titers leading to a build-up of isobutyrate.
38
Discussion
Recent efforts to develop microbial pathways for chemical synthesis have moved beyond upregulation
of native pathways to include transfer and modification of heterologous pathways to new hosts and
modified termination of native host pathways. Only a small number of truly de novo pathway designs
have been published and most use isolated heterologous enzymes acting on their cognate substrates
96,97,98.
79'
Engineered pathways to liquid fuels, in particular, have predominantly relied on entirely natural
(ethanol, butanol, isoprenoid) or terminally modified natural pathways (fatty acid synthesis, amino acid
aKAE, isoprenoid). The presented work moves beyond modification of natural pathways by successfully
demonstrating synthesis of 4MP via an extended de novo pathway which maintains selectivity while
utilizing multiple naturally occurring enzymes outside their native pathway contexts.
While one set of Modules has been presented in the current work, alternate chemistries could be
substituted for or combined with the selected modules to create new pathways to the same or alternate
products. For example, an isobutyryl-CoA mutase or branched a-keto-acid decarboxylase route could be
used to generate the isobutyryl-CoA precursor in place of Modules 1 and 2 99. Similarly, a FAS route
could be substituted for Module 3 to generate the longer saturated acid substrate for Module 4.
Using
this design individual alternative modules or module combinations can be directly compared to the
existing pathway in vivo. In addition, entirely new classes of branched products (e.g., aldehydes,
alkanes) could be made by using different Module 4 enzymes.
For the presented pathway, an iterative screening approach identified the enzymes catalyzing
conversion of the downstream 4-methyl-valeraldehyde and upstream isobutyraldehyde intermediates
as key components controlling selectivity of the pathway. Our initial Module 4 alcohol dehydrogenase
selection, Adh6psc, proved to be highly active, but non-selective in the full pathway context. Module 4
displayed high activity on our desired substrate, but in vivo results with the full pathway suggested this
module had a broad substrate range.
Persistent high isobutanol titers from strains expressing Modules
39
1-4 suggested that Module 4 enzymes were interacting with isobutyrate and/or isobutyraldehyde. In
vitro and in vivo data from Module 4 testing implicated the alcohol dehydrogenase, Adh6psc, as the nonselective enzyme (Fig. 2 and 4). By replacing Adh6psc with the isobutyraldehyde specific and NADHdependent alcohol dehydrogenase, LsadhLs, pathway selectivity and overall cofactor utilization were
improved.
As with alcohol dehydrogenase candidates, we initially selected aldehyde dehydrogenases
previously validated for an isobutyraldehyde substrate in an engineered pathway. Two endogenous
enzymes, PuuCEc and FeaBEC, were previously identified as the most effective E. coli aldehyde
dehydrogenases for isobutyraldehyde oxidation to isobutyrate
92.
Of the two E. coli aldehyde
dehydrogenases, FeaBEC proved to successfully synthesize 4MP from glucose in Strain M1F2P34
expressing Modules 1,2, 3, and 4 (Fig. 3B). Based on in vitro data one may predict PuuCEC to function
more effectively because its kcat/Km is more consistent across substrate lengths while the kcat/Km of
FeaBEc actually increases by an order of magnitude between propionaldehyde and hexanaldehyde
substrates"0 0'101.
In vivo results disproved this prediction with only FeaBEC producing 4MP (Fig. 3A). The
better performance of FeaBEc in the context of the full pathway may be explained by reported Km values
for the two dehydrogenases. FeaBEc has Km values below 100 ptM for relevant substrates while the Km
values for PuuCEC are 1 mM.
PuuCEc and FeaBEc were tested in strains expressing ADH6sc. Like FeaBEc,
Adh6psc has reported Km values for relevant substrates in the 100-200 IM range8 9 . Adh6psc was
observed to have kcat values (and PuuCEc
(-
100 sec-1) an order of magnitude higher than values observed for FeaBEc
10 sec-1) for related aliphatic aldehydes. Together these observed kinetics support the
hypothesis that Adh6psc out-competes PuuCEc and FeaBEc for the isobutyraldehyde substrate.
Isobutyraldehyde and reducing equivalents are diverted to isobutanol, lowering 4MP titers
(Supplementary Fig. S4). Strain M1P2P34 with PuuCEc produces significantly more isobutanol than Strain
M1F2P34 with FeaBEc, as expected based on observed Km values.
40
In addition in vitro data suggested that even though FeaBEC functioned as an isobutyraldehyde
dehydrogenase, it may also favor a 4-methyl-valeraldehyde substrate. The potential futile cycle created
by activity on 4-methyl-valeraldehyde was avoided by using the isobutyraldehyde specific
dehydrogenase Fjoh2967 from Flavobacterium johnsonaie 94. Replacing feaBEc with Fjoh2967j led to
increased isobutyrate and eliminated detectable 4MV production (Fig. 3). Combining more selective
alcohol and aldehyde dehydrogenases led to a highly selective overall pathway with the major
byproduct being overflow of the upstream intermediate isobutyrate (Supplementary Fig. S5). Together
the results from alcohol and aldehyde dehydrogenase selection highlight the importance of considering
both high activity and required selectivity when utilizing retro-biosynthetic screening. Proposing
potential upstream pathways is required to identify intermediates which could have cross-reactivity
with downstream enzymes.
Further engineering of the CoA-dependent 4MP pathway is warranted given the potential high
energy yield. Dugar and Stephanopoulos have outlined the importance of balancing reducing
equivalents generated and consumed in a recombinant pathway if high yields are desired
. Using the
current 4MP pathway enzymes the overall reaction can be written as:
1.5 C6H 2O + 3 ADP + 3 NAD+ + 3 NADPH
C 6H
O+3CO 2 +1H 20+2AMP+Pi+1ATP+3NADH+3NADP
14
+1H
The reducing equivalents of the pathway are balanced, but some are contained in different cofactors.
The maximum pathway energy efficiency (yI)
can be calculated using the degrees of reductance and
pathway stoichiometry for a glucose substrate and 4MP product. Maximum pathway energy efficiency
for the aKAE pathway and the presented CoA-dependent pathway are 75% and 100%, respectively.
Accounting for cofactor requirements, the adjusted pathway energy efficiencies ( 17PG ) are 24% and
45% for the aKAE and CoA pathways respectively (Appendix 2, Text A2-1). If alternative enzymes are
41
identified or engineered to accept NADH in place of NADPH, maximum pathway yields could be achieved
under anaerobic fermentation. The maximum adjusted efficiency values for these pathway
architectures then become 28% (aKAE) and 100% (CoA). The yield calculations highlight how our rational
design approach leads to a pathway architecture with high yield potential unlike inherently limited
pathways utilizing modification of amino acid synthesis.
This work has identified a novel pathway for the selective synthesis of the branched mediumchain length alcohol 4MP. The highest titers (193 ± 23 mg/L) were achieved with Strain M1Fj(IA)2134
which expresses both Fjoh2967Fj and ADH6sc. Selectivity was achieved by replacing ADH6sc with IsadhLs
in Strain M1Fj(IA)2134L. The 90 ± 7 mg/L of 4MP produced by M1Fj(IA)2134L represented 81% of
observed alcohol products. In comparison, of the 14 alcohols generated in the previous demonstration
of microbial 4MP synthesis using ct-KAE, 4MP makes up 14% of the total alcohol product. High potential
efficiency and selectivity make our CoA pathway a preferred candidate for future engineering.
Currently, the major byproducts of the CoA-dependent route are the acids acetate, isobutyrate, and
butyrate. We expect that a combination of tuning thioesterase/transferase activities of the host to
selectively cleave the longer 4-methyl-valery-CoA intermediate and relieving Module 3 rate limitations
will further enhance titers. Ultimately, screening or engineering for NADH-dependent enzymes should
produce a high yielding fermentative pathway. Our existing pathway can also be adapted to produce
other branched medium-chain products by testing new downstream modules. Finally, we believe the
pathway design approach described here can be useful for creation of new metabolic pathways which
rely on long de novo routes. Using retro-biosynthetic screening within a proposed pathway framework
allows exploration of diversity using a small number of assays while constraining enzyme space to a
chemical route which is maximally efficient for a given product.
42
Methods
Bacterial Strains and Plasmids
E. coli MG1655(DE3)
AendA ArecA described previously
was the host strain for production
experiments6 . E. coli DH10B (Invitrogen, Carlsbad, CA) and ElectroTen-Blue (Stratagene, La Jolla, CA)
were used in plasmid cloning transformations and for plasmid propagation. E. coli BL21Star(DE3) (Life
Technologies, Grand Island, NY) was used for expression of carNi for purification. (See Table 2-1 & Table
A2-1 for strain details)
A codon optimized version of S. cerevisiae ADH6 was purchased from DNA 2.0 (Menlo Park, CA) and
codon optimized versions of N. iowensis car and B. subtilis sfp were purchased from GenScript
(Piscataway, NJ) (See Supplementary Methods for codon optimized sequences). T. denticola ter and E.
gracilis ter were purchased from GenScript (Piscataway, NJ) as described previously
18.
Leifsonia sp.
Strain S749 isadh was purchased as a codon optimized GeneArt String from Life Technologies (Grand
Island, NY). All other genes were amplified from gDNA. B. subtilis PY79, E. coli MG1655, P. putida
KT2440, C. necator (formerly R. eutropha) H16, M. elsdenii, R. palustris CGA009, P. syringae DC3000, C.
acetobutylicum ATCC 824, and S. oneidensis MR-1 gDNA were prepared using the Wizard Genomic DNA
purification Kit (Promega, Madison, WI). P. aeruginosa PA01-LAC (ATCC# 47085), F. johnsonaie (ATCC#
17061) and V. parahoemolyticus EB 101 (ATCC# 17802) gDNA were purchased from American Type
Culture Collection (Manassas, VA). Custom oligonucleotide primers were purchased (Sigma-Genosys, St.
Louis, MO) for PCR amplification of genes from gDNA using either Phusion High-Fidelity DNA polymerase
(Finnzymes, Thermo Scientific Molecular Biology) or Q5 High-Fidelity DNA polymerase (New England
Biolabs, Ipswich, MA). Synthetic operons were constructed using a modified Splice by Overlap Extension
(SOE) PCR method.
The compatible vector set pETDuet-1, pCDFDuet-1, pACYCDuet-1, and pCOLADuet-1 was used to
express single genes or synthetic operons under control of a T71ac promoter and individual ribosome
43
binding sites. Plasmids were constructed using standard molecular biology techniques with restriction
enzymes and T4 DNA ligase purchased from New England Biolabs.
Ligation products in pETDuet-1,
pACYCDuet-1, and pCOLADuet-1 were used to transform E. coli DH10B, and pCDFDuet-1 products were
used to transform E. coli ElectroTen-Blue.
Propagated constructs were purified using a QiAprep
Miniprep Kit (Qiagen, Valencia, CA) and agarose gel fragments were purified using a Zymoclean Gel DNA
Recovery Kit (Zymo Research, Irvine, CA).
Completed constructs were used to cotransform E. coli
MG1655(DE3) AendA ArecA. Strains used to confirm 3H4MV synthesis using bktBCr, phaBen, tesBEc, and
various activators are described previously by Martin et al. 79
(See Text A2-2 for codon optimized
sequences and Table A2-2 for plasmid details).
Splice by Overlap Extension
Initial PCR products with homologous ends were added to a PCR mixture without additional primers and
cycled through a standard PCR cycle 4 times with annealing temperatures set at 6*C above, 3*C above,
and at the designed melting temperature for the homology. The upstream primer for the upstream
gene and the downstream primer for the downstream gene in the designed operon were then added to
amplify the full length product. A standard PCR method using the annealing temperature for the primer
pair was used for final amplification.
Culture Conditions
For all experiments triplicate seed cultures were grown from isolated colonies at 30 0 C overnight in 3 ml
LB medium in a 14 ml culture tube on a rotary shaker at 250 rpm.
All production cultures were
inoculated with 1% inoculum from overnight seed culture and grown at 30 0 C on a rotary shaker at 250
rpm. Cultures were induced with 0.5 mM IPTG when OD600 values reached 0.6-1.0 corresponding to
mid-exponential phase.
For constructs designed for 4MV production 50 ml cultures in 250 ml shake
flasks were used, and for constructs designed to produce 4MP 3 ml cultures in 1 inch diameter 50 ml
44
screw-cap culture tubes (Pyrex VISTA) were used. Unless otherwise stated, 1 ml culture samples were
taken 48 hours post induction, centrifuged to pellet cells, and the supernatant was removed for analysis.
For production of 4-methyl-valerate and 4-methyl-pentanol from glucose and isobutyrate, LB
medium supplemented with 1% glucose and 10 mM isobutyrate was used. For production of 4-methylvalerate and 4-methyl-pentanol from glucose, LB medium supplemented with 1.2% glucose was used.
Samples were taken 48 hours post induction except for initial experiments with Strains M1F2P34 and
M1P2P34 when samples were taken 72 hours post induction.
Relative Activity Assay for Purified His-Car
An overnight culture of BL21 Star (DE3) (Invitrogen) harboring pET/His-Car-RBS2-Sfp was used as 10%
(v/v) inoculum in 2 L of LB Broth. The culture was incubated at 30*C and 250 rpm, and expression was
induced using a final concentration of 1 mM IPTG at OD 0.6. Cells were harvested after 20 hours using
centrifugation and resuspended in a buffer containing 50 mM Tris-HCI pH 8.0, 300 mM NaCl, and 10%
glycerol. Cells were subsequently lysed using sonication. The supernatant was collected and applied to a
column containing Ni-NTA resin (Qiagen). Affinity chromatography was performed using step-wise
increasing concentrations of imidazole. Fractions containing purified His6-Car were dialyzed overnight at
4'C into 50 mM Tris-HCI pH 7.5, 50 mM NaCl, 1 mM DTT, and 10% glycerol. Dialyzed enzyme was then
flash frozen using liquid nitrogen and stored at -80'C.
The activity of His-Car on various substrates was determined by measuring changes in absorbance at
340 nm for up to 5 minutes in 96-well microplates (Tecan Infinite F200 Pro). Reactions were prepared as
follows: 100 mM Tris-HCI pH 7.5, 10 mM MgC 2 , 0.6 mM NADPH, 1 mM ATP, 224 nM His-Car, and 50 mM
pH neutralized acid substrate. All substrates were assayed in triplicate. For KMand
Vmax
determinations,
substrates were assayed at 5 different concentrations.
45
Metabolite Analysis
Culture samples were pelleted by centrifugation and supernatant was removed for HPLC analysis with
an Agilent 1200 series instrument with a refractive index detector. Analytes were separated using the
Aminex HPX-87H anion exchange column (Bio-Rad Laboratories, Hercules, CA) with a 5mM sulfuric acid
mobile phase at 35'C and a flowrate of 0.6 ml/min.
Commercial standards of glucose, x-
ketoisovalerate, acetate, acetoin, isobutyrate, butyrate, isobutanol, butanol, 4-methyl-valerate, and 4methyl-pentanol were used for quantification of experimental samples by linear interpolation of
external standard curves.
46
Chapter 3: A fatty acid synthase pathway to 7-methyl-octanoate
Introduction
Fatty acid synthases (FASs) have also been commonly used for engineering biofuel pathways in
microorganisms. Many diverse organisms generate fatty acids which often closely resemble diesel fuel
components0 2 . While natural lipids from plants, algae, and fungi can be purified for biodiesel
production, engineered bacterial strains (especially E. coli) have been investigated as well 25 ,103,104,105
Because esters of natural fatty acids can function directly as fuels, early work has been focused on
boosting natural production in some organisms and terminating synthesis at particular chain lengths in
others2 s,26,106. As mentioned in Chapter 1, the discovery of a plant acyl-ACP thioesterase which
generated shorter (C12) fatty acids when expressed in E. coli launched a series of investigations into
tailoring E. coli FAS for specific acid products through thioesterase expression 2 4 . A variety of
thioesterases have been explored, mostly aimed at generating long-chain acids for biodiesel or other
oleochemical targets (soaps, surfactants, etc.) 26 .
The promise of specifically engineering acid products, for chain length, branching, or desaturations,
has motivated the development of these microbial platforms over extraction of natural lipids from
plants or other sources2 6 . Some bacteria, including Bacillus subtilis, produce multiple branched fatty
acid isomers for their membranes1 0 7 . The initial branched precursors are derived from amino acid
biosynthesis and are condensed by specific ketosynthases expressed by B. subtilisl' 08. Recently B.
subtils FAS ketosynthases were combined with C12-C16 specific thioesterases in E. coli to produce
branched long-chain fatty acids1 09. Shortening of acids below C12 has also been pursued. A C8-C1O
plant thioesterase, FatB2, from Cuphea hookeriana was identified and used to produce octanoic and
decanoic acid in E. coli". Since that time an entire family of thioesterases has been explored to identify
more "short-chain" thioesterases and the E. coli FAS has been extensively modified for synthesis of C4C13 straight-chain acids'
112.
Combining the ability to produce branched acids and release short- or
47
medium-chain acids would lead to a pathway for branched short- or medium-chain acids which could be
converted to fuel compounds or other targets.
There are a number of characteristics of type II FAS pathways which distinguish them from the
CoA-dependent platform described in Chapter 2. Most microbial type II FAS systems are tuned for
synthesis of long-chain acids and require the use of protein (ACP) linked intermediates. The natural
ability of FASs to make longer chain lengths and multiple branched isomers opens up the possibility to
make a greater range of acid products if the pathway can be controlled stringently. There are other
potential benefits of an FAS platform which result from the mechanism of chain extension. CoAdependent chain extension utilizes easily reversible mechanisms for each step from condensation to
enoyl-CoA reduction. As a result, the ratio of oxidized to reduced cofactor greatly influences pathway
flux 67. While type 11 FASs use homologous
s-ketoacyl-ACP and enoyl-ACP reductases, the condensation
reaction is carried out by ketosynthases which decarboxylate malonyl-ACP in the process of extending
the acyl chain (Fig. 3-1)
60
. This release of CO 2 helps shift the thermodynamic equilibrium towards the
products unlike in the case of the CoA-dependent thiolases 2s. The malonyl-ACP is generated from
malonyl-CoA which is a product of a carboxylation reaction driven by ATP hydrolysis. This cycle of ATPdependent carboxylation followed by decarboxylation helps drive fatty acid synthesis forward.
Downstream reduction reactions are still required and thus cofactor ratios can still impede synthesis.
Because most FAS reductases use NADPH, aerobic culture conditions are typically required since
NADPH/NAD+ ratios are higher under aerobic growth6 7 . Recently NADH-dependent
P-ketoacyl-ACP
reductase homologs have been identified and shown to improve fatty acid synthesis under anaerobic
conditions1 1 3 . Overall FAS extension perfectly balances ATP and reducing equivalent generation from
glycolysis with ATP and reducing equivalent consumption during chain extension (Fig. 3-1). This redox
balance could further enhance pathway efficiency if NADH cofactors can be used.
48
GLYCOLYSIS
1 ATP and 1 reducing
equivalent per pyruvate
0
Njruvate
chain elongation
termination
mD
go
R'
glycerol-3P
U
R---,
0
Fig. 3-1.
General type 11 FAS
A typical type II FAS
pathway.
pathway can be divided into three
phases: initiation, chain elongation,
and termination. In E. coli fatty
acid chains are initiated from an
acetyl-CoA precursor, but in other
FAS
systems alternative CoA
thioesters are used. Malonyl-ACP
extender units are produced from
acetyl-CoA first by an ATPdependent carboxylation to form
malonyl-CoA and second by a
transacylation of the malonyl
group from CoA to ACP. The first
two carbons are added to the chain
via a FabH ketosynthase mediated
condensation of the initial CoA
thioester and malonyl-ACP. The
resulting @-ketoacyl-ACP is reduced
to a saturated acid before further
extension
by
a
separate
ketosynthase, FabF.
In many
bacterial systems the growing
chain
is
terminated
by
a
transferase
reaction
which
generates acyl-glycerol-phosphate
for membrane lipids.
Type I
systems have been engineered to
produce free acids by expressing
thioesterases.
The mechanisms for driving flux employed by FASs and the ability to selectively synthesize
different branched isomers led us to develop an FAS platform for branched, medium-chain acid
production. We began by using B. subtilis ketosynthases to modified the E. coli FAS to selectively
produce different branched long-chain fatty acids from initial short branched acids 3-methyl-butyrate,
and 2-methyl-butyrate. We than engineered the C. hookeriana FatB2 thioesterase for high soluble
expression in E. coli before accessing its ability to generate medium-chain branched acids. We
combined medium-chain thioesterase activity with branched fatty acyl-ACP synthesis to create a
pathway variant which could synthesize the branched acid 7-methyl-octanoate (7MC8). Overexpression
of the acetyl-CoA carboxylase complex accABCDEc led to modest improvements in free acid titers.
49
Pathway Design
GLYCOLYSIS
Fig. 3-2.
0
"lo-
0
3-methyl-butyrate
pyruvate
1-SBA Ibukas
0
lptbfs
0 0
-O - 1K
CP
1
%kCP
)N"j
CoA
Pathway
methyl-octanoate
schematic for 7-
(7MC8)
BSFAS
pathway. Production of 7MC8 in an E.
coli host is shown. Grey boxes indicate
endogenous metabolism which is utilized
for malonyl-ACP generation and fatty acid
elongation.
Module 1-SBA (short-
branched acid) is composed of the kinase
and transferase for 3-methyl-butyrate
(3MB)
activaton.
3MB
can
be
or
in the medium
supplemented
potentially generated
from
leucine
metabolism. Module 2-7MO (7-methyloctanoate) contains the recombinant
ketosynthase with enhanced activity for
the condensation of 3-methyl-butyryl-CoA
0
0
octanoate
methyl octanoate
o-
o
7-methyl-octanoate
(3MB-CoA) and malonyl-ACP.
branched thioester is reduced
This
and
extended by endogenous E.coli FAS.
Module 3-0 (octanoate) is an acyl-ACP
thioesterase with specificity for acyl
MeOH,
chains with a primary chain-length of 8
H2SO4
carbons
methyl 7-methyl-octanoate
Module 1-SBA and precursor generation
As with the previously described CoA pathway, acetyl-CoA can be generated in E. coli via the pyruvate
dehydrogenase complex. In the E. coli FAS pathway, acetyl-CoA acts as both the chain initiator and the
precursor to the malonyl-ACP extender unit (Fig. 3-1). The acetyl-CoA carboxylase complex, encoded by
accABCDEC, adds bicarbonate to generate malonyl-CoA from acetyl-CoA in an ATP dependent fashion.
The transferase FabDEC swaps the malonyl group from CoA onto ho/o-ACP to complete the synthesis of
malonyl-ACP. The branched ketoacid decarboxylase from B. subtilis can be used to generate branched
precursors including 3-methyl-butyryl-CoA (3MB-CoA) and 2-methyl-butyry-CoA (2MB-CoA), but in the
current work we supplemented 3-methyl-butyrate (3MB) and 2-methyl-butyrate (2MB) into the growth
medium and used expression of the B. subtilis transferase/kinase pair ptb/buks to activate 3MB or 2MB
to 3MB-CoA or 2MB-CoA (Fig. 3-2)91'109
50
Module 2-7MO and elongation
The native E. coli FAS pathway condenses an initial acetyl-CoA with malonyl-ACP using the ketosynthase
FabHEC- Subsequent elongating condensations are mediated by the ketosynthase FabFEC. Module 27MO introduces the ketosynthase FabH2Bs of B. subtilis which prefers the condensation of 3MB-CoA
with malonyl-ACP and can utilize the E. co/iACP. Both branched and straight-chain acyl-ACP
intermediates are reduced and extended by native E. coli FAS components FabGEc, FabZEc, FablEc, and
FabFEcModule 3-0 and termination
While the native E. coli FAS pathway terminates in a glycerol-3-phosphate transferase reaction mediated
by PIsBEc, our engineered pathway expresses the C8/C1O acyl-ACP thioesterase FatB2Ch of Module 3-0.
In initial designs we speculated that FatB2Ch Could potentially act on multiple branched isomers with
chain lengths as low as C7. Experimental results have confirmed only activity on 7-methyl-octanoy-ACP
in addition to the native intermediate octanoyl-ACP.
Results
bukBs
ptbBs
pCDF
fabHlBs
pET
fabH2Bs
pET
Fig. 3-3.
Plasmid constructs for
branched acid synthesis. A pCDFDuet-1
plasmid containing B. subtilis ptb/bukBs
was used for precursor activation. Two
ketosynthases were cloned in pETDuet-1.
Initial demonstration of branched fatty acid synthesis
B. subtilis uses FabH2BS to produce even and odd iso-branched fatty acids from isobutyryl-CoA and 3MBCoA precursors respectively. A second ketosynthase, FabH1Bs initiates fatty acid synthesis with a 2MBCoA precursor leading to odd anteiso-branched fatty acids. Both ketosynthases were expressed with
Module 1-SBA to create Strains M12a (fabHlBs) and M12b (fabH2Bs) (Table 3-1, Fig. 3-3). Strain M12a
was grown in a medium supplemented with 2MB and Strain M12b grown in a medium with 3MB. The
expected odd-chain iso- and anteiso-branched fatty acids were observed after methyl-esterification of
51
the cellular pellet (Fig. A3-1). Interestingly, when 3MB was supplemented to Strain M1 expressing only
Module 1-SBA, C15 and C17 iso-branched acids were still observed. This result supports the hypothesis
that E. coli FabHECcan catalyze the condensation of 3MB-CoA and malonyl-ACP to initiate branched fatty
acid synthesis.
Table 3-1. Strains used to evaluate branched FAS platform. E. coli MG1655(DE3) endArecAfadD~ was
transformed with different plasmid sets to create the strains shown.
Plasmid 3
Plasmid 1
Plasmid 2
pCDF-buk~eptbB,
pETDuet-1
M1
pCDF-bukBs-ptbB,
pET-fabHls,
M12a
pCDF-bukBs-ptbB,
pET-fabH2Bs
M12b
pCDF-bukeptbBs
pET-FatB2m2ch
M13
pCDF-bukB-ptbBs
pET-FatB2m2ch-fabH2S
M123
pCDFDuet-1
pET-FatB2m2ch-fabH1Bs
M2a3
pCDFDuet-1
pET-FatB2m2ch-fabH2B,
M2b3
pET-FatB2m2ch
M3
pET-FatB2m2ch-accABCDE
Strain name
3acc
pCDF-buk.-ptb,
pET-FatB2m2ch-fabH2,
pACYCDuet-I
M123ev
pCDF-bukB-ptb,
pET-FatB2m2h-fabH2B,
pACYC-accABCDEc
M123acc
Expression and engineering of FatB2
Previously published work with C. hookeriana FatB2ch did not clearly describe the sequence required for
mature protein expression in E. coi 114. Homologs, Carthamus tinctorius FatA2ct and Umballularia
californica FatBluc, had been purified from native tissue and sequenced. Protein sequencing returned
two potential N-termini for the mature proteins with one containing a hydrophobic N-terminal a-helix
52
and the other without the helix (Fig. 3-4)1.
It has been hypothesized that the helix could be cleaved
during purification because it may be used to associate the enzyme with a membrane in its native
context. Based on the sequencing data for FatA2ct and FatBluc, we cloned three versions of FatB2Ch
adding N-terminal His-tags: 1) the full ORF His-FatB2Ch, 2) the ORF truncated before the a-helix, HisFatB 2 mlCh, and 3) the ORF truncated after the a-helix, His-FatB2m2Ch (Fig. 3-4).
....I....I ....I....I ....I....I ....I....I ....I....I ....I....I
15
35
25
45
55
FatBl-Uc
MATTSLASAF CSM-------
----------
------ KAVM LARDGRGM-- ------ KPRS
FatB2-Ch
MVAAAASSAF FPVPAPGASP
KPGKFGNWPS
SLSPSFKPKS IPNGGFQVKA NDSAHPKANG
....I....1
65
75
85
95
FatBl-Uc
SDLQLRAGNA PTSLKMINGT KFSYTESLKR
FatB2-Ch
SAVSLKSGSL NTQE--DTSS SPPPRTFLHQ IPDWS-
105
115
PDWSMLFAV ITTIFSAAEK QWTnWKPK
Ill
SKR PDM--- HDRK
FatB2m1 Ch
....I....I ....I....1
125
135
. . . . I. . . . 1
145
155
165
175
FatBi-Uc
PKLPQLLDDH FGL-----HG LVFRRTFAIR SYEVGPDRST SILAVMNHMQ
FatB2-Ch
SKRPDMWVS FGLESTVQDG LVFRQSFSIR SYEIGTDRTA SIETLMNHLQ ETSLNHCKST
EATLNHAKSV
FatB2m2Ch
Fig. 3-4. FatB2 variant sequences tested for soluble expression. A sequence alignment of the translation of the
full ORFs for U. californica FatB1uc and C. hookeriana FatB2Ch is shown with potential N-termini of the mature
proteins highlighted. The triangle and red box indicates the previously published putative mature protein Nterminus for FatBlu which aligns with the selected FatB2mlCh terminus. The circle and blue box indicate the
alternative N-terminus of FatB1uc found from sequencing data. The next leucine residue (L122) in the FatB2Ch
sequence was selected as an alternative N-terminus as indicated by the arrow and blue box. The grey box
indicates the hydrophobic residues hypothesized to form a membrane associating helix.
All three His-tagged versions and an untagged version of the full ORF were expressed in E. coli (Fig.
3-5A). As expected the empty vector control and untagged FA TB2ch were not observed using the antiHis antibody. All three His-tagged versions showed some signal from the western blot, but a significant
increase in expression was observed for the shortest FatB2m2Ch version (Fig 3-5A). This is consistent
53
with either or both the N-terminal sequence reducing soluble expression and the N-terminal a-helix
potentially inserting into the cellular membrane. Culture supernatants from strains expressing the Histagged FatB 2 ch variants contained varying titers of C8 and C10 acids. The shortest variant, FatB2m2Ch,
produced 4 times the titer of C8 compared to the longer variants (Fig 3-5B).
A mpyHis-
HisHisFatB2- FatBS
45 KD
ml
m
38.5 KD 35.6
B
C8
vtrFMt21 'FaWB
40
35
30
25
DE20
15
(D10
0
C10
Empty vector
His-FatB2
His-FatB2ml
His-FatB2m2
Strain
Fig. 3-5. Soluble expression of modified FATB2ch in E. coli and B. subtilis. (A) Anti-His western blot
showing expression from four FATB2ch variants. The shortest truncated form, FatB2m2, showed
significantly higher soluble expression. (B) Strains expressing the His-tagged versions produced free
short free acids in culture supernatant. Octanoate (C8) titers were 4 times higher from the strain
expressing the most soluble His-FATB2m2ch variant. (C) His-FATB2m2ch was integrated at the amyE
locus in B. subtilis PY79 behind the Phyperspank inducible promoter. A western blot confirmed
expression in the PY79 background but no free C8 or altered lipid profile was observed from GCFAME analysis.
His-FatB2m2ch was also integrated into B. subtilis PY79 at the amyE locus to test for soluble
expression in the alternative host and to see if short, branched acids could be generated directly from B.
subtilis. Since B. subtilis natively produces each fatty acid isomer it may have been possible to observe
FatB2m2Ch substrate preference simply by expressing the single thioesterase in this host. While soluble
expression was observed (Fig. 3-5C), no free acids were observed in the culture supernatant and the
lipid profile was not altered compared to the control strain. This could be explained by FatB2Ch having
54
reduced activity when acting on acyl-ACP species with the B. subtilis ACP. Sequence variation between
E. coli and B. subtilis ACP could be the cause of the difference in activity.
FatB2m2ch
FatB2m2ch
pET
pET
fabH2Bs Fig. 3-6. Plasmid constructs for evaluating
free acid production. Two additional
pETDuet-1 derived plasmids were created
containing the thioesterase FatB2m2Ch
alone or with thefabH2Bs ketosynthase.
Synthesis of 7-methyl-octanoate (7MC8) and octanoate (C8)
With all module activities validated independently in vivo, a series of strains were constructed to identify
if a complete recombinant pathway could produce branched medium-chain acids. An empty vector
control, Strain M1, Strain M12b, and two additional strains, M13 and M123, expressing FatB2m2Ch were
compared to identify what components were necessary for different branched acid phenotypes. The
Strain M13 contained Modules 1-SBA and 3-0 and Strain M123 contained Modules 1-SBA, 2-7MO, and
3-0 (Table 3-1, Fig. 3-6). All five strains were cultured in LB medium supplemented with 10 mM 3MB
and 1.2% glucose. In addition, Strain M123 was cultured without the 3MB precursor as a "no substrate"
control. The cellular lipid content and free fatty acid titers were analyzed by GC-FAME.
Odd-chain, iso-branched fatty acids were observed as part of the total cellular lipid content when
3MB was supplemented in the growth medium for all strains but the empty vector control (Fig. 3-6A).
Expression of Modules 1-SBA and 2-7MO in Strain M12b leads to increased shorter branched acids
relative to Strain Ml expressing only the activators. With full pathway expression in Strain M123, 7MC8
content is increased and 13-methyl-tetradecanoic acid (13MC14) is decreased relative to Strain M12.
Without the ketosynthase present in Strain M13 the branched content drops significantly while increase
in C8 and C10 content over the empty vector is still observed. The reduction in branched content can be
explained by FabHEc having relatively low activity with 3MB-CoA and FatB2m2Ch having a lower Km for
straight-chain acyl-ACP species. With FabHEC producing a lower concentration of the branched 0-
55
ketoacyl-ACP intermediates FatB2m2Ch preferentially cleaves the straight-chain acyl-ACPs.
As
expected, branched species were not observed in the "no substrate" control.
A
empty vector
1-SBA, 3-0
1-SBA, 2-7M0O
5
0
1-SBA, 2-7MO, 3 -o
1 -SBA, 2-7MO, 3 -0, No Sub
4
3-
75
2
-
C8
7MC8
C10
Jn
11MC12
Species
(D
B
4540
353025-
E 2015105-
I
hI
C8
C8
Species
13MC14
15MC16 17MC18
Fig. 3-6. Branched and short-chain
acid production in 7MC8 BSFAS
pathway strains. (A) Branched fatty
acids were observed in the cellular
lipid content when Module 1-SBA
was expressed.
Introduction of
Module 2-7MO (fabH2s) reduces the
average length of observed branched
acids. Expression of Module 3-0 with
Module 1-SBA reduces branched
species and increases C8 and CIO
content.
Expression of all three
modules increases 7MC8 content
while maintaining pools of the longer
empty vector
branched species. When no 3MB is
1-SBA
provided in "No Sub." control no
1-SBA, 3-0
1-SBA, 2-7M0
branched species are observed. (B)
1-SBA, 2-7M0, 3-0
Free fatty acids are observed only
1-SBA, 2-7M0, 3-0 No Sub.
with Module 3-0 expression. The
7MC8 to C8 ratio increases when
Module 2-7MO is expressed resulting
in a titer of 16 ± 1 mg/L.
7MC8
7MVC8
Culture supernatant data supported the cellular lipid content results and aligned with the expected
outcome based on the pathway design. Free fatty acids were only produced by Strains M13 and M123
with the thioesterase expressed (Fig. 3-6B). Both C8 and 7MC8 were observed with the highest 7MC8
titer (16 ± 1 mg/L) produced by Strain M123 expressing all three modules. Strain M13, which contained
much lower branched acid fractions in its cellular lipid content, produced very low titers of free 7MC8.
As with the cellular lipid result, the "no substrate" control produced only straight-chain free acid.
To assess the necessity of the activators ptb/bukBs, Strains M2a3 and M2b3 were created (Table 31). When either ketosynthase was expressed with the thioesterase, but without the activators of
Module 1-SBA, no cellular or free branched acids were observed (Fig. A3-2). While Strain M2b3
produced a similar octanoate titer to Strain M13, expression of fabHlBs in Strain M2a3 reduced free
56
straight-chain acid titers. An attempt was made to create an alternate full pathway construct expressing
ptb/bukBs,fabHlbs, and FatB2m2ch. Upon culturing with 2MB supplementation this strain stopped
growing and the culture cleared indicating lysis.
accAEc
FATB2m2ch
accBEc
pET
accCEc
accDEc
Fig. 3-7. Plasmid constructs for increasing
malonyl-CoA flux. Two additiona\ plasmids
were constructed for overexpression of the
aCcBEc acetyl-CoA carboxylase complex. The first
contained the acc operon and FATB2m2Ch
in pETDuet-1. The same acc operon was
acC Ec
also cloned in pACYCDuet-1.
accAE,
pACYC
accDEc
Increased titers through accABCDEc overexpression
While our results and observations from the literature suggested that thioesterase activity is limiting,
increasing short-chain acyl-ACP pool size had the potential to increase titers if native pathway enzymes
were already saturated and our thioesterase, FatB2m2Ch, had a high Km for our substrates of interest.
We cloned an artificial operon containing the four E. coli genes of the acetyl-CoA carboxylase complex
into two different plasmids (Fig. 3-7). A single pETDuet-1 construct containing FatB2m2ch and accABCDEc
was designed to examine the effect of overexpression on C8 production only. The accABCDEc operon
was also cloned into pACYCDuet-1 in order to express it with the full 7MC8 FAS pathway. Strain M3acc
contained only the pETDuet-1 construct and Strain M123 was transformed with pACYCDuet-accABCDEc
to create Strain M123acc.
The new accABCDEc overexpressing strains produced increased titers of both C8 and 7MC8 acids
over control strains. Strain M3acc produced 70% more C8 than the control (Fig. 3-8). Overexpression of
accABCDEc led to a more modest 20% increase in 7MC8 titer when the full pathway strains are
compared. Addition of the third plasmid and antibiotic marker would likely lower expression levels of
other pathway enzymes. If FatB2m2Ch activity is limiting, a lower expression level could reduce the
benefit of increased precursor availability provided by accABCDEc. The expression of accABCDEc from the
lower copy pACYCDuet-1 plasmid is likely lower than that observed for Strain M3acc which contains the
57
80
70
3-0
60-
1-SBA, 2-7MO, 3-0
3-0, accABCD
3-8.
Free
fatty acid
production
increased with accABCDEc overexpression.
1-SBA, 2-7MO, 3-0, accABCD Strain M3, expressiong only Module 3-0,
4 mg/L of C8.
produced 43
led to a 70%
accABCDEc
of
Overexpression
50
40
20'
increase in C8 titer (72 ± 9 mg/L). Strain
M123 produced 13 ± 1 mg/L 7MC8.
Overexpression of accABCDEC in Strain
M123acc led to a 20% increase in 7MC8
titer (17 ± 1 mg/L). A similar small increase
was observed for C8 in the same strain.
CD
E
Fig.
15
105-
0
7MC8
C8
free fatty acid
single high copy pETDuet-1 plasmid. Overall, the results support the theory that FatB2m2Ec has a high Km
relative to native FAS components with the relevant substrates.
Discussion
The diversity and distribution of type 11 FAS pathways in not only bacteria, but archaea and plants
provides a natural enzyme set which can be potentially harnessed to produce diverse acid products.
Over the past decade, engineering of bacterial FAS pathways has predominantly been focused on
production of potential biodiesel compounds partially due to government mandates for increased
biofuel production 2.
As mentioned in the chapter introduction, FAS pathways utilize a mechanism
which can potentially create strong driving forces for carbon flux to fatty acids if harnessed properly.
These pathways have the potential to produce shorter acids if a carefully selected elongation and
termination strategy is chosen.
Our initial demonstration of a rationally selected, recombinant FAS platform to branched
medium-chain acids provided insights into the potential for creating new acid products through FAS
engineering. A primary concern for developing an FAS platform is the selection of an appropriate acylACP thioesterase. Because E. coli has been used as the model organism for in vivo assessment of acylACP thioesterases, the activities of many newly discovered thioesterases have only been evaluated in
the context of the straight-chain acyl-ACPEc substrates generated by E. co/i 25,
58
26, 111.
Recently, the E. coli
tesAEc thioesterase documented to release predominantly C14 and C16 straight chain fatty acids was
also reported to produce branched 13-methyl-tetradecanoic and 15-methyl-hexadecanoic acid 3 2 . This
branched acid result established that at least some thioesterases may accept a wider range of acyl
substrates.
Unlike most enzymes, FAS enzymes and acyl-ACP thioesterases are not simply binding a small
molecule. Enzyme substrate interactions are mediated by both enzyme-acyl chain and enzyme-carrier
protein interactions (specifically with a-helix II of ACP) 6 1,
115, 116.
Similar to hypotheses made about PKS
systems, it is possible that strong ACP interactions allow for catalysis of a wider range of acyl substrates.
One might also hypothesize that terminal branching in longer acyl groups would be less disruptive than
branching for shorter substrates because longer terminally branched acyl groups appear identical to
straight-chain acyl groups for a greater distance from the active site catalytic residues.
Since we desired to generate medium-chain branched acids, we were uncertain whether a
suitable thioesterase could be found. It was unclear, when we began this work, whether a mediumchain thioesterase, like FatB2Ch, would be capable of acting on a branched substrate. Our platform with
Ptb/BukBs activation of 2MB and 3MB and the branched-chain ketosynthases FabH1Bs and FabH2BS
provided a facile in vivo assay for branched acyl-ACP thioesterase activity. We learned, through
synthesis of 7MC8, that FatB2m2Ch could indeed act on a branched substrate with a C8 primary chain
length. Interestingly strains expressing a full pathway for 6-methyl-octanoic acid (6MC8) synthesis
appeared to lyse. It is possible that 6MC8 may be especially toxic or that FatB2m2Ch was too active on 6methyl-octanoyl-ACP so that cells were unable to efficiently build their membranes. Strain M12a,
expressing ptb/bukBs and fabHlBS, did contain a higher percentage (22%) of branched acids in the cellular
membrane than Strain M12b expressingfabH2Bs (10%) (Fig. A3-1). This increased branched content
indicates the activity for synthesis of the anteiso-branched acids was higher and could have led to
59
strains with less stable membranes which were more susceptible to toxicity from generation of free
acids.
While the GC-FAME analysis used in this work did not provide direct measurement of
intracellular acyl-ACP pools, the lipid profiles can help to build hypotheses about how enzyme
components are affecting acyl-ACP pools and about the relative kinetics of different components. By
creating a complete set of controls, we found that 3MB activation by Ptb/BukBs was both necessary and
sufficient for branched acid synthesis in our E. coli host. This result means an E. coli ketosynthase, likely
FabHEC, is capable of condensing 3MB-CoA and malonyl-ACP. Additional expression of the FabH2BS
ketosynthase shifted the observed branched acid profile to shorter-chain lengths. The shift in profile
indicates that overexpression of the B. subtilis ketosynthase more rapidly generates branched acyl-ACP
intermediates which cannot be fully extended by the native E. coli FAS. As one might expect, this
increased shorter acyl-ACP pool led to higher titers of free branched acids when FatB2m2Ch was
expressed. Interestingly, the total free acid titers are almost 10 mg/L lower when more 7MC8 is
produced with fabH2
expression. This may result if FatB2m2Ch has a slower turnover rate with the
branched acyl-ACP substrate. The attempt to further increase the shorter acyl-ACP pool size through
overexpression of the acetyl-CoA carboxylase complex led to small increases in free acid titers. The
increases in titers likely result because FatB2m2Ch is not saturated and accABCDEc overexpression
increased C8-ACP and 7MC8-ACP substrate pool sizes, which in turn increased the turnover rate of
FatB2m2Ch. FatB2m2Ch may have a naturally high Km to prevent thioesterase activity from completely
abolishing flux to necessary membrane lipids in Cuphea hookeriana. Alternatively, the E. coli
ACPEc
may
increase the Km for the acyl-ACPEc substrates relative to what would be observed with the native acylACPCh substrates. The observation that FatB2m2Ch expression in B. subtilis PY79 did not alter the lipid
profile or produce free acids could also be explained by a poor interaction between acyl-ACPB,
substrates and FatB2m2Ch due to differences in the
60
ACPBs
sequence.
By combining components of the B. subtilis type II FAS and a medium-chain acyl-ACP
thioesterase we have demonstated the first reported microbial synthesis of 7-methyl-octanoic acid. The
discovery that the plant thioesterase FatB2Ch is capable of acting on branched acyl-ACP intermediates
supports the hypothesis that acyl-ACP thioesterases can be identified to act on a much wider array of
acyl-ACPs than the straight-chain compounds they are traditionally assayed with. If further pathway
development leads to improved efficiency and additional acid targets, the FAS platform could be used to
produce an array of products with various applications. The longer medium-chain products produced by
this FAS platform could potentially be used to increase energy density in a fuel blend or be used to
produce distinct flavor compounds. The acids can provide cheese or oil flavors, while the aldehydes or
methyl esters can impart honey, fruity, or citrus flavors5 3,5 4 . A better understanding of the complex
interactions between FAS enzymes, acyl groups, and ACP would likely expedite the process of pathway
engineering. An ideal system would have a highly active acyl-ACP thioesterase acting on a parallel FAS
pathway so that growth would not be inhibited by altered phospholipid synthesis. We discuss strategies
for development of such a pathway in Chapter 5.
Methods
Bacterial strains and plasmids
E. coli MG1655(DE3) AendA ArecA AfadD, described below, was used as the host strain for production
experiments. E. coli DH10B (Invitrogen, Carlsbad, CA) and ElectroTen-Blue (Stratagene, La Jolla, CA)
were used in plasmid cloning transformations and for plasmid propagation. E. coli MG1655(DE3) AendA
ArecA, as described in Chapter 2 Methods, was used for preparation of FatB2ch variant lysates. B.
subtilis PY79, a gift from the Alan Grossman Lab at MIT, was used as the base strain for B. subtilis
thioesterase expression experiments. Construction of the B. subtilis PY79 amyE::FatB2m2 strain is
described below.
61
A codon optimized version of the full open reading frame of C. hookeriana FatB2 was purchased
from GenScript (Piscataway, NJ) (Text A3-1). All other genes were amplified from B. subtilis PY79
genomic DNA which was prepared using the Wizard Genomic DNA purification Kit (Promega, Madison,
WI). Custom oligonucleotide primers were purchased (Sigma-Genosys, St. Louis, MO) for PCR
amplification of genes from gDNA using Phusion High-Fidelity DNA polymerase (Finnzymes, Thermo
Scientific Molecular Biology). Synthetic operons were constructed using the modified Splice by Overlap
Extension (SOE) PCR method described in Chapter 2 Methods.
Constructs were cloned into members
of the same Duet vector set described in Chapter 2 Methods. Plasmids were constructed using standard
molecular biology techniques with restriction enzymes and T4 DNA ligase purchased from New England
Biolabs. All primers used for cloning and plasmid names can be found in Table A3-1.
Gene knockouts for E. coil production strain
E. coli MG1655(DE3), described previously, was used as a base strain for creating the triple AendA ArecA
AfadD knockout strain 8 . Single gene deletions were achieved through P1 transduction of the kanamycin
marked knockouts from Keio Collection strains acquired from the Coli Genetic Stock Center at Yale
University117'118'11 . Kanamycin resistance markers were removed using the pCP20 helper plasmid as
described by Datsenko and Wanner
19
. The recA locus was removed last to ensure efficient
recombination into the recipient cell genome.
Gene integration in B. subtifisPY79
The codon optimized FatB2m2 gene with N-terminal His-tag sequence was PCR amplified from pETFatB2m2 using primers described in Table A3-1. Both the FatB2m2 PCR product and pDR111 plasmid,
described previously, were digested with Hindlll and Nhel restriction enzymes and ligated to form
pDR111-FatB2m2
17
B. subtilis PY79 was made competent for plasmid transformation as follows. A freezer stock was
streaked for isolation on an LB agar plate and grown overnight. A single colony was picked and
62
inoculated into 5 ml of LB medium in a 250 ml shake flask. The flask was grown in a shaking incubator at
37*C and 250 rpm for -3 hours until the culture reached a density of OD60o between 0.8 and 1.2. When
the OD600 was between 0.8 and 1.2, 0.5 ml of culture was used to inoculate a 10 ml MD medium (recipe
described below) culture in a 250 ml shake flask. The MD culture was grown for 3-4 hours at 37*C and
250 rpm at which point cells were ready for transformation.
For the transformation two 200 pl aliquots of the freshly prepared competent cells were added
to two 1 inch diameter test tubes. One microgram of pDR111-FatB2m2 plasmid DNA was added to one
tube and 10 micrograms of pDR111-FatB2m2 DNA was added to the second. The tubes were then
incubated in a shaking incubator at 37*C and 250 rpm at a ~60* from vertical for 1 hour. After 1 hour
each entire 200 pl transformation mixture was plated on LB Agar containing 100 pg/ml spectinomycin
for selection of pDR111-FatB2m2 integrants.
MD medium recipe
The 10 ml of MD medium was prepared as follows. A one liter stock of 10x PC buffer was prepared first
containing: 107 grams of anhydrous dipotassium phosphate, 60 grams of anhydrous monopotassium
phosphate, and 10 grams of trisodium citrate dehydrate. The pH was adjusted to 7.5 with potassium
hydroxide (~10 grams). MD medium was prepared using 10x PC buffer as follows. 10x PC buffer was
diluted with tap water to make 9.1 ml of 1.1x PC buffer. The following were added to the 1.lx PC buffer
to complete the MD medium: 0.4 ml of 50%w/v glucose, 30 pl of 1 M magnesium sulfate, 250 pl of 100
mg/mi potassium aspartate, and 50 pl of 2.2 mg/ml ferric ammonium citrate. The final mixture was
filter sterilized.
Culture conditions
For all E. coli production expriments biological duplicates were grown overnight from single colonies in 5
ml LB cultures. Overnight cultures were inoculated at 1% inoculum into 50 ml of LB with 1.2% glucose in
250 ml shake flasks. Where appropriate 10 mM 3-methyl-butyrate or 2-methyl-butyrate was
63
supplemented into the medium at inoculation. Cultures were grown to an 0D 600 value of 0.6-0.8 before
induction with 0.5 mM IPTG. Samples were taken for gas chromatography-fatty acid methyl ester (GCFAME) analysis 48 hours post induction. B. subtilis production experiments were carried out as
described for E. coli except that 20 ml cultures, 0.6% glucose, and 1 mM IPTG induction were used and
no precursor acids were supplemented in the medium.
Acid extraction/methyl esterification
Two different methods were used to analyze fatty acid content. Free acids in culture supernatant were
analyzed as follows. Culture samples taken at 48 hours were pelleted by centrifugation and 5 ml of
supernatant was removed and added to a 15 ml conical centrifuge tube. The supernatant was acidified
with 50 d of 10 M HCI to increase extraction of acids into the organic phase. After acidification, 5 ml of
a 2:1 mixture of chloroform:methanol with 100 ig/ml tridecanoic acid was added and samples were
vortexed for 2 hours. After vortexing the phases were separated by centrifugation (5000g) and the
bottom chloroform layer (~3 ml) was transferred to a capped glass vial. The samples were then
completely evaporated by flowing compressed air over the samples with a manifold (~30 minutes).
Samples were held in a polystyrene tray which kept them partially insulated so they cooled upon
evaporation. The remaining solid was resuspended in 1 ml of methanol + 2% (vol) sulfuric acid and
incubated in a heating block at 60"C for 2 hours to esterify the acids. After 2 hours the liquid was
transferred to 1.7 ml microcentrifuge tubes and partially evaporated using the same manifold set-up for
~40 minutes until the liquid volume was ~100 ul. Methyl esters were then extracted from this
concentrated liquid by addition of 1 ml of hexane and vortexing for 15 mins. Phases were separated by
centrifugation and 800 ul of the hexane layer was transferred to vials for GC analysis.
Methyl esters of lipids from the cellular pellet were extracted and esterified using a base
catalyzed direct transesterification method". After 48 hours culture density was measured by OD 600.
A volume of culture was sampled which would collect a number of cells equivalent to 1 ml of culture at
64
an OD 600 of 30. The culture sample was pelleted by centrifugation and the supernatant was removed by
aspiration. The pellet was emulsified by pipetting in 100 pl of hexane with 10 mg/ml tridecanoic acid as
an internal standard. Phospholipids were than extracted and esterified by adding 500 pl of 0.5 M
sodium methoxide and vortexing for 1 hour at room temperature. After transesterification, 40 pl of
anhydrous sulfuric acid was slowly and carefully pipetted to acidify the solution for esterification of any
free acids contained in the pellet. The acidified solution was then vortexed at room temperature for 2
hours. 500 ptl of hexane were added to the sample and the sample was vortexed for 30 minutes at room
temperature to extract methyl esters. The phases were then separated by centrifugation and 300 pl of
supernatant was transferred to a GC vial with 300 ul glass insert.
Gas chromatography analysis
Fatty acid methyl esters were analyzed by using a Bruker 450-GC instrument with a flame ionization
detector. Compounds were separated using a HP-INNOWAX capillary column (30 m x 0.25 mm). The
oven conditions were set at 120 0 C (2 min) followed by a linear 10 minute ramp up to 180*C and a hold at
180 0 C (18 min). Fatty acids were quantified by creating standard curves of methyl esters prepared using
both the supernatant and pellet esterification methods from known dilutions of acid standards. The
ratio of methyl-ester peak area to methyl tridecanoate internal standard peak area was calculated for
each concentration to build the curve. Octanoic, nonanoic, decanoic, undecanoic, dodecanoic,
tetradecanoic (mystric), 12-methyl-tetradecanoic, 13-methyl-tetradecanoic, hexadecanoic (palmitic), 14methyl-hexadecanoic, 15-methyl-hexadecanoic, (Z)-9-hexadecenoic (palmitoleic), octadecanoic (stearic),
and (Z)-9-octadecenoic (oleic) acids (Sigma-Aldrich) were prepared using the cellular lipid protocol.
Octanoic, nonanoic, and decanoic standard curves were made using the supernatant protocol. 7methyl-octanoic concentrations were estimated using the nonanoic acid standard, since the branched
acid was not commercially available.
65
Western blot analysis
E. coli MG1655(DE3) AendA ArecA was transformed with empty pETDuet-1 or pETDuet-1 with each of
the FatB2ch variants and plated on LB Agar + 100 pg/ml Ampicillin. Single colonies of each transformant
were grown overnight in 5 ml of LB. Shake flask cultures (250 ml flasks) containing 50 ml LB + 1%
glucose were inoculated at 1% inoculum from overnight LB cultures and incubated with agitation at 30 0 C
and 250 rpm. Shake flasks were induced with 1 mM IPTG ~2 hours post inoculation when they reached
OD600 values between 0.6 and 0.8.
Four hours after induction 10 ml of each culture was sampled and pelleted by centrifugation.
Cell pellets were resuspended in 1 ml of 10 mM Tris-HCI at pH 8.0 and added to 1.7 ml microcentrifuge
tubes containing 500 ul of 0.1 mm diameter glass beads (Scientific Industries, Inc. Disruptor Beads, SIBG01). Samples were then vortexed for 10 minutes. After lysis, samples were pelleted by centrifugation
(6,000g) and the supernatant was removed as soluble lysate. Total protein was quantified by a
previously described Bradford assay method using Bio-Rad Protein Assay Dye Reagent (Cat #5000006)121. A Bio-Rad 10% Mini-PROTEAN TGX gel (Cat #456-1034) was run using the Mini-PROTEAN Tetra
Cell electrophoresis set up. Bio-Rad Precision Plus Protein Unstained Standard (Cat #161-0363) and 15
pag of total protein for each sample was loaded on the gel. After running the gel for 33 min at 200 volts,
the gel was removed from the casing and washed for 5 mins in 100 ml of deionized water.
The washed gel was then equilibrated in Transfer Buffer for 15 minutes (described below) along
with blotting paper, sponges, and nitrocellulose used with the Bio-Rad Mini Trans-Blot Module (Cat
#170-3935). The Trans-Blot Module was then used for transfer to the nitrocellulose by running the cell
at 100 volts for 1 hour. Once transferred the nitrocellulose was blocked by washing in 5 wt% BSA in 1x
TBS buffer (described below) for 2 hours at room temperature. The nitrocellulose was then washed
twice for 10 minutes with Ix TBST buffer (described below) and incubated overnight at 4"C with
66
THE
His Tag Antibody, mAb, Mouse (GenScript cat. no A00186-100) diluted in 1x TBS + 10% glycerol to
a final concentration of 0.5 ptg/ml. The blot was then washed three times with 1x TBST for 10 minutes
each time at room temperature and incubated for 2 hours with a donkey anti-Mouse IgG-HRP secondary
antibody (Santa Cruz Biotechnologies, sc-2318) at a 1:5000 dilution in Ix TBS buffer. Following
secondary antibody binding the blot was washed twice for 10 minutes with 1x TBST and then once with
1x TBS at room temperature. The blot was developed using the Western Blotting Luminol Reagent
(Santa Cruz Biotechnologies, sc-2048) and imaged using an Alpha Innotech FluroChem imager.
Western blot buffers
One liter of Transfer Buffer was prepared with the following components: 14.4 g glycine, 3.025 g Tris
base, 200 ml methanol, 800 ml of deionized water. 500 ml of 10x TBS buffer was prepared with the
following: 12.1 g Tris base, 146.2 g sodium chloride, and 500 ml of deionized water. The pH was
adjusted to 7.5. 500 ml of lx TBST was prepared with the following: 50 ml of 10x TBS, 450 ml deionized
water, and 250 pl of Tween-20 (0.05% final concentration).
67
Chapter 4: Modular pathways to short-chain alkanes
Introduction
As outlined in Chapter 1, production of alkanes from glucose is currently uneconomical due to
inexpensive sources of petroleum. If at some point in the future either sugar costs are reduced or
sources of petroleum become more expensive, microbial production of alkanes may become cost
effective. If traditional gasoline components are synthesized efficiently, they would be easier to
separate (pentane solubility in water 40 mg/L
22
, Tb0 II=309 K1 23), leading to reduced process energy costs,
and could be blended with other sources of hydrocarbons within the existing transportation
infrastructure. We recognized that our CoA-dependent and FAS platforms could potentially provide
routes to such alkanes if a suitable enzyme was found for the conversion of medium-chain aldehydes to
alkanes.
While pathways to alcohols have been better described in a variety of organisms, biological routes
to alkanes do exist. Alkanes are produced by a variety of eukaryotes for different purposes including as
waxes for plants and as pheromones for insects2 '
1 5
. Some prokaryotes and yeast have been observed
to generate C1-C3 alkanes and the olefin isobutene126. Routes to the short-chain alkanes have not been
described and olefin production appears to result from promiscuous enzyme activity on non-cognate
substrates2 . Cyanobacteria have been observed to produce alkanes with pentadecane (C15) and
heptadecane (C17) being the dominant species29. The specific biological role of alkane production in
these cyanobacteria remains unknown. Recently, key enzymes were identified from the FAS based
pathway to alkanes in cyanobacteria 3 0 . Genes for acyl-ACP reductases (AR) and aldehyde
decarbonylases (AD) were identified from ten different cyanobacteria. When the genes were expressed
in E. coli, they were sufficient for alkane production. The alkane profile was dominated by pentadecane
and heptadecane with some tridecane (C13).
68
Since that time, the mechanism of the aldehyde decarbonylase has been explored. Knowledge of
the mechanism has enabled attempts at enhancement of AD activity in non-native hosts and with
alternative substrates. A structure for the Prochlorococcus marinus MIT9313 AD had been previously
solved by the Joint Center for Structural Genomics12 8. The P. marinus AD structure fell into the family of
non-heme di-iron oxidases and oxygenases. While most enzymes in this family catalyze oxidative
reactions, AD has been found to use a cryptic redox mechanism employing both reductive and oxidative
steps29. 130 The mechanistic knowledge led to the discovery that AD is inhibited by H20 2 which can be
produced if the rate of electron transfer from ferredoxin is slow 1 3 1 . This discovery led to the hypothesis
that activity of the AD in E. coli may be inhibited by H20 2 due to slow delivery of electrons to the
recombinant decarbonylase. Adding a catalase to AD in vitro reactions relieved H20 2 inhibition by
generating the 02 substrate from H2 0 2 . When a catalase was coexpressed in vivo, total alkane
production did not increase. The lack of change in titer is consistent with either reduced AD expression
when expressing the additional catalase gene or with native E. coli pathways being sufficient for H2 0 2
degradation.
The available structure of P. marinus AD and results from the catalase study inspired one group to
engineer the P. marinus AD for enhanced activity on shorter-chain aldehydes. In vitro assays with AD
and catalase revealed substrate inhibition for octanal (C8), decanal (C10), and dodecanal (C12) with the
wild type AD showing preference for tetradecanal (C14)1 3 1 . The P. marinus AD structure contained a
substrate analog (palmitate) bound in the active site and hydrophobic binding pocket. Based on the size
of the binding pocket, it was hypothesized that substrate inhibition occurred due to multiple short-chain
aldehydes binding simultaneously in the pocket preventing product release.
Introducing an alanine to
phenylalanine mutation at position 134 of P. marinus AD increased decarbonylase activity on nonanal
(C9), heptanal (C7), hexanal (C6), pentanal (C5, valeraldehyde), and butanal (C4, butyraldehyde) in
vitro
. The greatest relative increases in activity occurred for hexanal, valeraldehyde, and
69
butyraldehyde with the highest absolute activity observed with hexanal. E. coli expressing the A134F
mutant was cultured with 10 mM butyraldehyde supplementation and found to produce higher propane
titers than wild type AD (0.45 mg/L).
Since the initial discovery, ADs have also been utilized in novel pathways for production of alkanes
not observed from wild type cyanobacteria. A modified E. coli FAS pathway was used to produce
alkanes as short as dodecane (C12) 32 . Small amounts of branched 2-methyl-pentadecane (2MC15) were
also produced in E. coli when a branched ketoacid decarboxylase complex (BCKD) was co-expressed with
B. subtilisfabH2Bs32 . A system of acyl-ACP thioesterase (tesAE, Cinnamonum camphora FATB1cc) and
fatty acid reductase (FAR) complex (Photorhabdus luminescens luxC, luxE, and luxD) expression
produced fatty aldehydes from the acyl-ACP pool. Highest total alkane titers were 6 mg/L. By
measuring alkane, aldehyde, and fatty alcohol content, the authors observed competition for the
aldehyde pool between the recombinant AD and native E. coli aldehyde reductases. Recently,
production of alkanes as short as nonane was reported using an E. coli host 3 1 . As in the initial FAS based
alkane pathway, thioesterases were expressed to generate free acids. Instead of a direct reduction, the
native FadDEc acyl-CoA ligase was used to generate a CoA-thioester. A fatty acyl-CoA reductase is then
used to produce the respective aldehyde. Unlike in previous literature, the putative AD CER1 from
Arabidopsis thaliana is used to generate the final alkane product. A number of results appear contrary
to previous literature on CER1 and E. coli fatty acid synthesis.
Through collaboration with another student, Aditya Kunjapur, we have expanded upon previous
work with aldehyde decarbonylases to demonstrate a system of pathways from single carbon sources to
both straight- and branched-chain alkanes from C3 to C7. The designs add novel termination modules
expressing Car and AD activities to the CoA and FAS acid synthesis platforms described in Chapters 2 and
3. We have also attempted to limit shunting of our aldehyde intermediates to alcohols by using a strain
developed by Kunjapur to have reduced aldehyde redeductase activity 1
70
Pathway design
While recombinant AD expression has initially focused on production of longer chain alkanes derived
from acyl-ACP intermediates, we hypothesized that both our CoA-dependent and FAS pathways could
be adapted for alkane synthesis (Fig. 4-1). In addition to synthesis of 4MP, CoA chemistry has been used
GLYCOLYSIS
PEP
0--
-
0
1-18
JO~
00
C
3-methyl-butyrate
1-SBA 1bukas
J-butyrate
0
1r~
OX
0
0
A
0
butyrate
COA
AC+CA
0
valerate
CA
0
0
CoA "'CoAACOA
0
0
,"A
0
+''CoA
~
bukes
-,PtbBs
jptbBs
0
1-SBAACOA
0
0
octanoate
hexanoate
0 0C0
0
7-methyl-octanoate
4-methyl-valerate
propane
butane
pentane
2-methyl-butane
heptane
2-m
ethyl-heptane
Fig. 4-1. Short- and branched-chain alkane pathways. Proposed pathways to six different short-chain alkanes
are shown. A portion of the isoleucine pathway to a-ketobutyrate, used by Tseng et al. for pentanol synthesis,
makes up the propionyl-CoA precursor module, Module 1-Pr. Modules 1 and 2 of the CoA-dependent 4MP
pathway have been combined to form the isobutyryl-CoA precursor module, Module 1-lB. The short-branched
acid activators of Module 1-SBA can also be used to generate isobutyryl-CoA from isobutyrate. Acetyl-CoA is
generated from pyruvate using the endogenous pyruvate dehydrogenase complex (PDHcompex). Module 3 of
the CoA-dependent 4MP pathway has been renamed Module 2-MCC (medium-chain CoA) in the alkane
pathways. Module 2-MCC can be used for synthesis of valerate, hexanoate, and 4-methyl-valerate depending
on the precursor pathway used. A variant of Module 2-MCC, Module 2-BC (butyryl-CoA), was created to limit
carbon-chain extension to butyryl-CoA by utilizing the more C4 specific thiolase thc, from Clostridium
acetyobutylicum. The BSFAS 7MC8 pathway remains unchanged and can be used to produce C8 and 7MC8
acids. Module 4A (alkane) consists of corNi, which was used in the CoA-dependent 4MP pathway, and
PMT1231pm (ADpm) or PMT1231pm-mut (ADpm-A134F). The overall set of pathways can potentially produce
propane, butane, pentane, 2-methyl-butane, heptane, and 2-methyl-heptane.
71
by our laboratory and others to produce butanol, pentanol, and hexanol
14'17'18'19
All of these
pathways utilized CoA-dependent aldehyde dehydrogenase activity to produce the respective aldehydes
from saturated CoA-thioesters. While alcohol dehydrogenases were employed to further reduce the
aldehyde to the alcohol, it should be possible to decarbonylate the same substrates using an AD.
Variations on 4MP pathway Module 3 (thiolase, @-keto reductase, dehydratase, enoyl reductase) and
CoA precursor pathways (propionyl-CoA, isobutyryl-CoA) can be used to tune the acid product profiles of
the general pathway. The 7MC8 BSFAS pathway can also be tuned for either straight- or branched-chain
acid production by either excluding or including Modules 1-SBA and 2-7MO. As a single enzyme
alternative to the P. luminescens FAR complex used by Howard et al., we proposed using a carboxylic
acid reductase to generate aldehydes and we selected the characterized P. marinus ADpm for final alkane
synthesis.
The pathways were expressed in E. coli MG1655(DE3) endA recA~ and in the reduced aldehyde
reduction (RARE endA recA) strain which was derived from MG1655(DE3)1 3 3 . The RARE strain contains
a combination of alcohol dehydrogenase and aldo-keto reductase knockouts. Specifically, dkgB, yeaE,
yahK, yjgB, and the yqhC-dkgA operon were deleted using P1 transduction.
Alternative precursor modules for CoA pathways (Modules 1-Pr and 1-IB)
Propionyl-CoA can be synthesized from ct-ketobutyrate, an intermediate of isoleucine biosynthesis.
Tseng et al. previously generated propionyl-CoA for pentanol synthesis by overexpressing the E. coli
thrAG1297ABCEc operon containing a feedback resistant thrA mutant, the feedback resistant ilvAcg of
Corynebacterium glutamicum, and the mutant E. coli aceEF-pdOj operon (PDH complex)' 8 . Module 1Pr uses the same threonine operon and i/vAcg, but does not overexpress the PDH complex. Module 1-lB
is a simple concatenation of Modules 1 and 2 of the CoA-dependent 4MP pathway, and the same
plasmid constructs were used for gene expression. Module 1-SBA of the BSFAS pathway can also be
used for activation of supplemented isobutyrate precursor in place of Module 1-1B.
72
CoA-dependent chain extension (Modules 2-MCC and 2-BC)
Module 2-MCC, composed of the same enzymes used for chain extension in the CoA-dependent 4MP
pathway, can be used for synthesis of butyrate (from two acetyl-CoA), valerate (from propionyl-CoA and
acetyl-CoA), hexanoate (iteratively from 3-acetyl-CoA), or 4-methyl-valerate (from isobutyryl-CoA and
acetyl-CoA). Because Module 2-MCC readily extends butyryl-CoA to hexanoyl-CoA, a variant module,
Module 2-BC, was created for propane synthesis. The C. acetobutylicum thiolase thc, was selected to
replace the C. necator thiolase bktBcn because thc, is more specific for condensation of two acetylCoA
. Endogenous acyl-CoA thioesterases produce the free acid intermediates.
FAS modules (Modules 1-SBA, 2-7MO, and 3-0)
The pathways to octanoate (C8) and 7-methyl-octanoate (7MC8) are taken directly from the 7MC8
BSFAS pathway.
Acid reduction and aldehyde decarbonylation (Module 4A)
Module 4A utilizes CarNi for free acid reduction to aldehyde as in the CoA-dependent 4MP pathway.
CarNi was selected for aldehyde generation because a carboxylic acid reductase can be used with any
free acid generating pathway and we had previously shown that CarNi has activity on acids across these
chain-lengths. The P. marinus MIT9313 AD A134F mutant (ADpm-A134F) had been shown to have
increased activity on butyraldehyde, valeraldehyde, and hexanal in vitro. We used this AD variant and a
wild type control for initial pathway demonstrations.
Results
Demonstration of CarNi/Adh6psc route to butanol and pentanol
In order to confirm that CarNi was a suitable reductase in vivo for the range of acids being generated by
the CoA pathways, two new strains were created to test butanol and pentanol synthesis using CarNi and
Adh6psc. This reduction scheme was successful for 4MP synthesis, but previously published butanol and
pentanol synthesis had relied on the AdhE2ca of C. acetobutylicum4 's15,
17, 18,19. Strain M2MCC4,
73
Table 4-1. Strains used to evaluate alkane pathways. Strains used for C3, C4, C5, and C7 alkane synthesis and
butanol and pentanol synthesis are shown below. Plasmid constructs are described in Tables A2-2, A3-1, and
A4-1. Host strains are described in Chapter 4 Methods.
Host strain
Plasmid 1
Plasmid 2
Strain name
MG1655(DE3)
AendAArecA
pET-(bktBcn-terTd)(phaBcn-phaJ4bc,)
pACYC-(carNrsfpBS)PMT1231pm
MG-nC5wt
MG1655(DE3)
AendAArecA
pET-(bktBcf-terTd)(phaBcn-phaJ4bc,)
pACYC-(carNsfpB)PMT1231pm-mut
MG-nC5
RARE
AendAArecA
pET-(bktBcf-terTd)(phaBcn-phaJ4bcn)
pACYC-(carNrfpB)PMT1231pm
RARE-nC5wt
RARE
AendAlrecA
pET-(bktBcn-terrd)(phaBcn-phaJ4bcn)
pACYC-(carNrspBS)PMT1231pm-mut
RARE-nC5
MG1655(DE3)
AendAArecA
pET-(thlc-terd)(phaBc,-phaJ4bcn)
pACYC-(carN-SfpB)PMT1231pm-mut
MG-C3
RARE
AendAArecA
pET-(thIc-terTd)(phaBc,-phaJ4bcn)
pACYC-(carNr-SfpBs)PMT1231pm-mut
RARE-C3
MG1655(DE3)
ApuuCAfeaBAfadD
pET-FatB2m2chaccABCDEc
pACYC-(carNsfpB)PMT1231pm
MG-nC7wt
RARE
ApuuCAfeaBAfadD
pET-FatB2m2chaccABCDEC
pACYC-(carNsfpBs)PMT123p
RARE-nC7wt
Host strain
Plasmid 1
Plasmid 2
Plasmid 3
MG1655(DE3)
AendAArecA
pET-(bktBcf-terTd)(phaBc,-phaJ4bcn)
pACYC-(carN-SfpBS)PMT1231pm-mut
pCOLA-(thrAfBC)-
RARE
AendAArecA
pET-(bktBcf-terTd)(phaBc,-phaJ4bcn)
pACYC-(carN-SfpBS)PMT1231pm-mut
pCOLA-(thrAfBC)-
RARE
AendAArecA
pET-(bktBcf-terTd)(phaBc,-phaJ4bcn)
pACYC-(carN-sfpBs)PMT1231pn
pCDF-bukBs-ptbBs
RARE-M1SBA
RARE
AendAArecA
pET-(bktBcn-terTd)(phaBcn-phaJ4bcn)
pACYC-(carN-sfpBS)PMT1231p,-mut
pCDF-bukB,-ptbBs
RARE-M1SBA
2MCC4A
MG1655(DE3)
AendAArecA
pET-(bktBc,-terTd)(phaBcn-phaJ4bcn)
pACYC-(carNr-sfpBs)ADH6s,
MG1655(DE3)
pET-(bktBcn-terTd)(phaBc,-phaJ4bcn)
pACYC-(carN-sfpBs)ADH6sc
AendAArecA
ilvAfr
ilvA*
Strain name
MG-nC4
RARE-nC4
2MCC4Awt
M2MCC4
pCOLA-(thrA'BC)-
ilvAf
MlPr2MCC4
expressing Module 2-MCC and Module 4 from the CoA-dependent 4MP pathway, produced butanol as
expected with titers up to 876 ± 18 mg/L when grown in LB + 1.2% glucose (Table 4-1, Fig. 4-2). Strain
M1Pr2MCC4 also expressed Module 1-Pr, to produce propionyl-CoA, in addition to Modules 2-MCC and
74
Module 4. As expected, Strain M1Pr2MCC4 produced pentanol with either glucose (96 ± 24 mg/L) or
glycerol (165 ± 32 mg/L) supplementation. The greatest pentanol:butanol selectivity was observed from
growth on glycerol most likely due to increased propionyl-CoA:acetyl-CoA ratios. Together these results
suggested CarNi could be used for reduction of the range of shorter-chain acids in vivo.
propionate
800-
butyrate
valerate
butanol
pentanol
600
200
0-
M2MCC4
(glucose)
M2MCC4
(glycerol)
M1Pr2MCC4
(glucose)
Fig. 4-2. Butanol and pentanol synthesis using
CarNi/Adh6pC. The suitability of CarNi for acid
reduction of various substrates was tested with
strains constructed to express 4MP Module 4
with either Module 2-MCC or 2-MCC and 1-Pr.
Butanol was produced by all strains with the
greatest titer observed for Strain M2MCC4
grown on glucose (876 ± 18 mg/L). Pentanol
was observed at up to 165 ± 32 mg/L when
Strain M1Pr2MCC4 was grown on glycerol.
M1Pr2MCC4
(glycerol)
strain
Pentane synthesis from glucose
Strains for pentane production were constructed first because the ADpm-A134F had shown the highest
activity for a hexanal substrate. Module 2-MCC was combined with one of two Module 4-A variants in
either MG1655(DE3) endA~ recA~ or RARE endA~ recA host strains. The resulting four strains are shown
in the table below (Table 4-1). All subsequent strains use the same nomenclature with MG and RARE
designating the respective host backgrounds and the alkane shorthand indicating the desired product.
Table 4-1. Pentane producing strain names. Four strains were constructed to test pentane synthesis from
glucose. Module 2-MCC and one of two Module 4-A variants were used in two host strain backgrounds.
Host background
AD
RARE endA- recAvariant
MG1655(DE3) endA- recAADp.
MG-nC5wt
RARE-nC5wt
ADp.-A134F
MG-nC5
RARE-nC5
The nC5 strains were cultured in sealed GC vials with the headspace sampled and analyzed 24 hours
post induction. All four strains produced detectable pentane with the highest titer (1.6 ± 0.3 mg/L)
observed for Strain RARE-nC5 which expressed the mutant ADpm-A134F in the aldehyde reductase
knockout background (Fig. 4-3). The greatest difference in titers occurred between strains harboring
75
different ADpm variants. Moving forward, ADpm-A134F was used with all CoA pathways. Analysis of
liquid phase supernatant revealed that the use of the RARE background for Strain RARE-nC5 eliminated
butanol and hexanol byproduct formation while Strain MG-nC5 produced 31
6 mg/L butanol and 115 t
16 mg/L hexanol.
A 2.0-
B
n-C5
1.6-
140
butyraldehyde
120-
butanol
hexanol
100-21.2
-
~80~0.8 -
6040-
0.4
20-
0.0
MG-nC5wt
MG-nC5
RARE-nC5wt
Strain
RARE-nC5
0MG-nC5
Strain
RARE-nC5
Fig. 4-3. Pentane production from glucose. (A) Four strains were constructed to test pentane synthesis from
glucose. Module 2-MCC and one of two Module 4-A variants were used with two host strain backgrounds. All
four strains produced detectable pentane with strains harboring the mutant ADpm-A134F giving the highest
titers (MG-nC5 1.2 ± 0.2 mg/L, RARE-nC5 1.6 ± 0.3 mg/L). Propane at the lower detection limit was observed
only in MG strains. (B) Detectable liquid phase C4 and C6 products revealed elimination of hexanol and butanol
synthesis in RARE-nC5. MG-nC5 produced 31 ± 6 mg/L butanol and 115 ± 16 mg/L hexanol.
Propane and pentane synthesis from glucose
As predicted, Strains MG-C3 and RARE-C3, harboring Modules 2-BC and 4-A, produced detectable
propane from glucose (Fig. 4-4). MG-C3 produced 0.17 ± 0.04 mg/L while RARE-C3 produced 0.13 ± 0.02
mg/L. While the difference in propane titers was not statistically significant, the two strains did display
different product profiles. Somewhat unexpectedly RARE-C3 produced more pentane (0.41 ± 0.09 mg/L)
.5
C3
Fig. 4-4. Propane production from glucose.
Strains MG-C3
and RARE-C3
harboring
Modules 2-BC and 4-A were used for propane
production
from glucose. Both
strains
produced propane with titers of 0.17 ± 0.04
mg/L (MG-C3) and 0.13 ± 0.02 (RARE-C3).
0.4
0.3
.3 0.2-
0.1
Pentane was also produced at 0.41
by RARE-C3.
0.0
MG-C3
RARE-C3
Strain
76
0.09 mg/L
'1e
2220 1816
14-_
c
butyraldehyde
100
8060
--
C-
:0
(Da)
butyraldehyde
butanol
10
6- 40-
8-)
6
4
0-1
20
0
RARE-C3
MG-C3
Strain
RARE-C3
MG-C3
Strain
Fig. 4-5. RARE host strain increases butyraldehyde intermediate pool size. (A)
Strain RARE-C3 accumulated butyraldehyde in the headspace to 17.9 ± 2.7 times
that observed for Strain MG-C3 suggesting the aldehyde dehydrogenase and aldoketoreductase deletions in the RARE background do reduce butyraldeyde
reduction to butanol. (B) Liquid phase concentrations of butyraldehyde and
butanol corroborate gas phase observations. RARE-C3 is observed to produce 46
+ 5 mg/L butyraldehyde and 40% of the butanol produced by MG-C3.
than propane even though the thica thiolase was used. Pentane production in the RARE background
likely results from C4 acyl-CoA build-up because reductive pathways from butyraldehyde to butanol are
reduced. Indeed, increased butyraldehyde was observed in the gas phase and LC traces showed higher
levels of butyraldehyde and reduced butanol compared to the MG1655(DE3) endA~ recA background
(Fig. 4-5).
Butane and pentane synthesis from glycerol
Introduction of Module 1-Pr with Modules 2-MCC and 4-A led to butane synthesis (Fig. 4-6).
Cultures were grown in LB + 1.2% glycerol since the greatest product specificity and titers of pentanol
were observed with Strain M1Pr2MCC4 grown on glycerol (Fig. 4-2). The highest observed butane titer
1.4
n-C4
1.2
n-C5
Fig. 4-6. Butane production from glycerol.
Strains MG-nC4 and RARE-nC4 harboring
Modules 1-Pr, 2-MCC and 4-A were used for
1.0.
Cn0.8
butane production from glycerol. Both strains
0.6
produced butane with titers of 0.35 ± 0.12
mg/L (MG-nC4) and 0.46 ± 0.15 mg/L (RAREnC4).
Pentane was also produced by both
strains at 0.93 ± 0.22 mg/L by MG-nC4 and
, 1.27 ± 0.08 mg/L by RARE-nC4.
0.4
0.2
0.0
MG-nC4
Strain
RARE-nC4
77
was 0.46 ± 0.15 mg/L with Strain RARE-nC4. Pentane titers exceeded butane in these strains with up to
1.27 ± 0.08 mg/L pentane from RARE-nC4. The pentane titers are not surprising given the ability of
Module 2-MCC to generate butyryl-CoA, valeryl-CoA, and hexanoyl-CoA. Module 1-Pr relies on activity
from the PDH complex which has higher activity on pyruvate to produce acetyl-CoA. Even with growth
on glycerol it is likely that butyryl-CoA pools can exceed those of propionyl-CoA in these strains leading
to greater C6 intermediates than C5. Propane was not observed.
2-methyl-butane synthesis
While short, straight-chain alkanes had been synthesized, we also wanted to demonstrate short,
branched-chain alkane synthesis. AD activity had not been described for short, branched aldehyde
substrates and it was unknown if ADpm would act on 4-methyl-valeraldehyde or other branched
substrates. We decided to use both wild type ADpm and the ADpm-A134F mutant because the mutant
had been created to partially block access to the hydrophobic binding pocket. It was unclear if the
phenylalanine residue would block access for the branched substrate. The RARE endA recA background
was used for 2-methyl-butane experiments.
Two strains were constructed to express the full pathway to 2-methyl-butane from glucose. RAREiC5 contained Modules 1-1B, 2-MCC, and 4-A (ADpm-A134F) and RARE-iC5wt contained Modules 1-1B, 2MCC, and 4-Awt (ADpm). As for the previous alkane targets, these full pathway strains were cultured in
LB + glucose. No alkane products were observed and, surprisingly, no aldehydes were observed in the
gas or liquid phase as well. The absence of any Module 4-A intermediates or products suggested that
Module 4-A expression may be limiting. The Module 4-A genes were expressed from the lower copy
pACYCDuet-1 plasmid and these were the first constructs of the alkane work to use four plasmids in a
single strain.
Two new strains were constructed to test the hypothesis of low Module 4-A expression. Strains
RARE-1SBA2MCC4A and RARE-1SBA2MCC4Awt (containing Modules 1-SBA, 2-MCC, and 4-A or 4-Awt)
78
were constructed. We had previously shown the B. subtilis Ptb/BukBs activators to work on an
isobutyrate substrate for 3H4MV synthesis (Fig. 2-4). We used the pCDF-bukBs-PtbB plasmid for
isobutyrate activation with Module 2-MCC and 4-A each expressed on individual plasmids. The new
strains used three plasmids and could theoretically synthesize 2-methyl-butane from isobutyrate and
glucose.
Head space of cultures of both strains contained butyraldehyde, isobutyraldehyde, and 4-
methly-valeraldehyde. RARE-1SBA2MCC4A expressing ADpm-A134F produced very low titers of alkanes,
near the detection limit (Fig. 4-7A). It is possible that ADpm expression was still poor in these strains and
that the ADpm variants have lower activity on the branched substrates. The generation of
isobutyraldehyde may also inhibit the decarbonylase.
A
160 .
2M-C4
140-
n-C5
B2.5-
120-
2.0-
100
80-
1.5-
60 -
1.0-
40-
~
20
0
0.5RAREM1SBA2MCC4A
RAREM1SBA2MCC4AWt
Strain
0.0
WT
A134F
AD P variant
Fig. 4-7. 2-methyl-butane (2M-C4) synthesis from isobutyrate + glucose and from 4MV. (A) A low level
of 2-methyl-butane production is observed from Strain RARE-M1SBA2MCC4A expressing Modules 1-SBA,
2-MCC, and 4-A. Observed GC peaks were near the lower detection limit for the assay and titers are an
estimate based on an n-pentane gas standard. (B) 4MV feeding to strains expressing Module 4-A with
either wild type ADpm or ADpm-A134F produced 2.25 ± 0.14 mg/L and 1.73 ± 0.38 mg/L of 2-methylbutane, respectively. Clear ADpm activity on 4-methyl-valeraldehyde was confirmed for both the wild type
and mutant.
In order to more clearly confirm ADpm activity on 4-methyl-valeraldehyde, we conducted 4MV
feeding experiments to RARE strains harboring one of two Module 4-A variants with either ADpm or
ADpm-A134F. Supplementation of 5 mM 4MV to the growth medium led to 2-methyl-butane titers up to
2.25 ± 0.14 mg/L (Fig. 4-7B). These results confirmed that ADpm and ADpm-A134F were capable of 2methyl-butane synthesis at levels comparable to straight-chain products under the right conditions.
79
Increased AD expression, the absence of the potential isobutyraldehyde inhibitor, and increased 4methyl-valeraldehyde concentrations could all contribute to the higher titers observed with these single
module constructs.
We further hypothesized that ADpm could potentially be engineered to better accommodate 4methyl-valeraldehyde which could potentially increase 2-methyl-butane production with the upstream
pathway. As mentioned above, a structure is available for ADpm with an acid substrate analog bound at
the active site. Khara et al. also solved a structure of ADpm-A134F with hexanoate bound.
Using the
ADpm-A134F structure and based on the position of the bound hexanoate, we selected two hydrophobic
residues to mutate that were within 5
A of the 4-methyl-valerate
branch-point. We designed
substitutions of various smaller hydrophobic residues in place of the F100 and 1127 residues and
attempted to clone mutants in the wild type and A134F backgrounds. A subset of the desired mutations
was successfully cloned. RARE endA reck was transformed with single plasmid constructs expressing
carNi and the ADpm variants. These Module 4-A variant strains were cultured in LB + glucose+ 10 mM
4MV and the gas phase was sampled after 72 hours. Initial technical replicates showed that 2-methylbutane could be produced at up to 2.6 ± 0.2 mg/L from 4MV, but only minor variation was observed
between mutants.
It is possible that mutations may have altered the Km of ADpm, but differences are
not observed because the in vivo assay produces 4-methyl-valeraldehyde at saturating concentrations.
Heptane synthesis from glucose
Both the RARE endA~ recA~ and MG1655(DE3) endA reck backgrounds were used to construct strains
harboring Modules 1-SBA, 2-7MO, 3-0, and 4-Awt (RARE-2MC7wt and MG-2MC7wt). The wild type
ADpm was used because previous in vitro results by Khara et al. showed no difference in activity between
the A134F mutant and wild type on octanal132 . Both strains were cultured in LB supplemented with
glucose and 3-methyl-butyrate to test for synthesis of 2-methyl-heptane with an expected heptane coproduct. GC analysis of the culture headspace revealed only 3-methyl-butyraldehyde production with
80
neither alkane product observed. CarNi activity on the 3-methyl-butyrate substrate produced the 3methyl-butyraldehyde byproduct which could have inhibited alkane production in these strains. It is
also possible that pools of longer chain aldehydes were too low for ADpm to have significant turnover.
Based on the above hypotheses, Strains RARE-nC7wt and MG-nC7wt were constructed with pACYCcarNi-SfpBs-PMT1231pm and the pET-FATB2m2-accABCDEc plasmid used in Strain M3acc for octanoate
production (Fig. 3-7 and 3-8). The new strains, focused solely on heptane synthesis, produced
observable heptane from glucose. RARE-nC7wt (1.8 ± 0.2 mg/L) produced close to three times the
heptane titer of MG-nC7wt (0.6 ± 0.1 mg/L) (Fig. 4-8). While the dominant free acid produced by
FatB2m2Ch was octanoic acid, decanoic acid was also observed as part of the cellular pellet lipid profile
and in lower titers as a free acid (Fig. 3-6A and A3-2). If CarNi can convert decanoate to decanal, the wild
type ADpm has higher activity on this longer aldehyde substrate and we would expect nonane to be
formed in MG-nC7wt and RARE-nC7wt. Indeed, nonane was also observed with titers of 0.4 ± 0.1 mg/L
and 0.6 ± 0.1 mg/L for MG-nC7wt and RARE-nC7wt respectively.
2.0
1.8-
n-C7
n-C9
1.6
1.4
- 1.2
E 1.00.8-
0.604-
MG-nC7wt
Strain
RARE-nClwt
'
Fig. 4-8. Heptane and nonane synthesis from
glucose. Strains MG-nC7wt and RARE-nC7wt were
designed to synthesis heptane from glucose. Both
strains expressed Modules 3-0 and 4-Awt (wild
type ADpm) and the native E. coli acetyl-CoA
carboxylase complex accABCDE. While heptane
was observed from both constructs, RARE-nC7wt
produced 1.8 ± 0.2 mg/L which was three times
that produced by MG-nC7wt. Nonane was also
observed with titers of 0.43 0.14 mg/L and 0.61
0.12 mg/L.
Discussion
Only recently have enzymes been identified which catalyze the synthesis of alkanes. Predominantly
these natural, long-chain alkanes are produced from long-chain aldehyde precursors through a
decarbonylation reaction. Within the past five years aldehyde decarbonylases from cyanobacteria have
been used for alternative termination of FAS pathways for the synthesis of long-chain (>C13) alkanes30',
81
32.
Last year synthesis of C9 nonane was reported using termination of an FAS pathway with an acyl-ACP
thioesterase, acyl-CoA ligase, acyl-CoA reductase, and the multi-domain, plant enzyme CER1 31 . CER1 is
believed to be a multi-functional, transmembrane protein with aldehyde decarbonylase activity
contained in one of the domains. Pentane synthesis from linoleic acid in Yarrowia lipolytica was also
reported in the last year135 . This route to pentane relies on a radical catalyzed cleavage of linoleic acid
to produce pentane and a 13-oxo-cis-9,trans-11-tridecanoic acid byproduct by a soybean lipoxygenase.
In the current work we have leveraged our FAS and CoA-dependent platforms to medium-chain acids to
create a system of recombinant pathways to medium- and short-chain alkanes which have not been
described previously. Using this system we demonstrate the first reported microbial synthesis of
propane, butane, and heptane using engineered pathways from glucose and demonstrate 2-methylbutane synthesis from acid precursors. Our platform was also used to demonstrate an alternative route
to pentane.
As in the case of the 4MP synthesis route described in Chapter 2, we have analyzed the overall
alkane pathway structure in order to understand the potential efficiency of the described routes. In
general, for all odd-carbon numbered alkane products the platform pathways can be made redox
neutral. This results because both CoA-dependent and FAS chain-extension consume two reducing
equivalents for every C2 unit added to the carbon chain. If the reducing equivalents generated in the
pathway to the chain initiator (e.g., acetyl-CoA, propionyl-CoA, isobutyryl-CoA) are consumed by the
reductive termination route then the overall pathway will be redox neutral. The general reaction for
odd straight-chain alkane synthesis via the CoA-dependent route can be written as follows:
n+1
n+1
glucose + (n+1) reducing equivalents +
4 )n+1>
+1
( 2
CnH.n+
82
+ (l
2
C02
n-i
ADP +
1P + 0 2 +
2
-
2
H+
+ 1 formate + (n+2) reducing equivalents + n1ATP + I AMP + (n-1) H20
(2
Unlike the 4MP pathway, the reducing equivalents are unbalanced because the AD consumes 2 reducing
equivalents in generating the alkane from the aldehyde precursor. Expression of a formate
dehydrogenase would recover the additional reducing equivalent by oxidizing formate to C0
1 36 1 37
,
2
. The
overall reactions using the FAS platform change slightly because additional ATP and water are consumed
for malonyl-CoA generation during chain extension. The FAS reaction can be written as:
Sn+1 glucose + (n+1) reducing
equivalents + 1 ADP + 02
4
CnH 2 n+2 + n+j CO 2 + 1 formate + (n+2) reducing equivalents + 1 AMP + 1 P + n 2
22
H2 0
Calculating maximum energy efficiency for specific alkane pathways highlights differences
between alcohol and alkane targets (calculation described in Text A2-1). The stoichiometric yields (mol
alkane/mol glucose) of propane (CoA-dependent), pentane (CoA-dependent) and heptane (FAS) are 1.0,
0.67, and 0.50 respectively. The degree of reductance increases with increasing alkane chain-length: 20
for propane, 32 for pentane, and 44 for heptane. When multiplied by the stoichiometric yields and
normalized by the degree of reductance of glucose (24), the maximum pathway energy efficiency (yIJ)
for these three products are found to be 83%, 89%, and 92% respectively. A similar analysis gives y/
=89% for 2-methyl-butane synthesis because the pathway stoichiometry is the same as in the case of
the pentane isomer. Butane synthesis uses the odd-chain precursor propionyl-CoA which is generated
through a portion of isoleucine biosynthesis. While the carbon required for synthesizing the C5
precursor can come from a single glucose molecule, two additional reducing equivalents and one
additional ATP are required. The resulting stoichiometric yield from glucose is 0.67. Using the degree of
reductance (26 for butane) y/ is found to be 73%.
The lower y''s observed for alkanes relative to
alcohols result from the final reducing equivalents in the decarbonylation reaction going to reduction of
molecular oxygen and release of formate instead of reduction of the aldehyde to the alcohol 12 9.
83
Because the inefficiency occurs at pathway termination, longer chain alkanes will have relatively higher
y" values when using these routes.
In this initial demonstration we are clearly far from these theoretical maximum yields. Pathway
improvements discussed for the upstream platform pathways to acids could be applied to alkane
synthesis, but our work suggests that AD activity is limiting. A number of pathway variants built up
aldehyde intermediates. Some aldehyde concentrations were more than an order of magnitude above
the observed alkane titers. Decreasing alcohol byproduct routes by using the RARE strain did enhance
alkane synthesis in the case of heptane, but had less of an effect for shorter species. While we observed
AD activity on short, branched substrates for the first time, low or undetectable titers from glucose
suggest that ADpm may have relatively poor kinetics with these substrates. Identification of an AD with
higher activity on desired medium-chain substrates is required for any significant improvements in yield.
We have demonstrated that our platform pathways can be used to produce subsets of shortand medium-chain alkanes through proper module selection. The overall pathway architecture can be
made redox neutral, but energy efficiency is limited by the decarbonylation mechanism. In vivo data
suggest that in our specific pathways ADpm is limiting. Improvement of the kinetics of the
decarbonylation reaction is required to generate higher alkane titers. In Chapter 5 we discuss strategies
for identification of alternative ADs. Due to relative performance against alcohol alternatives and the
availability of inexpensive petroleum derived alkanes, development of commercial microbial production
of short- and medium-chain alkanes is unlikely to occur in the near future.
Methods
Bacterial strains and plasmids
E. coli MG1655(DE3) AendA ArecA and RARE AendA ArecA were used for most alkane production
experiments except when otherwise specified. RARE AendA ArecA was created from the parent strain
MG1655(DE3). Genes encoding four aldo-keto reductases and aldehyde dehydrogenases (dkgB, yeaE,
84
yahK, yjgB ) were deleted by P1 transduction using Keio Collection knockout strains117'11 . The X-red
system was used to knockout the yqhC-dkgA operon in the quadruple knockout strain to complete the
RARE strain. Three sets of primers were used to create homology around the kanamycin resistance
cassette (Table A4-1). Finally endA and recA were deleted from RARE by P1 transduction using Keio
strains as donors. For heptane synthesis E. coli MG1655(DE3) ApuuC AfeaB AfadD and RARE ApuuC
AfeaB AfadD were both created by sequential knockouts using P1 transduction of Keio donor strains. E.
coli DH5ct was used for new plasmid cloning and plasmid propagation.
A codon optimized sequence for P. marinus MIT9313 AD (PMT1231) was purchased as a
GeneArt* String from Life TechnologiesTM (Text A4-1). Once codon optimized PMT1231 was cloned into
pACYC-carNF-SfpBS-PMT1231, the ADpm-A134F mutant sequence was created by PIPE cloning of the
plasmid using oligio primers which replaced the GCA codon at positions 400-402 with a TTT codon (Table
A4-1)13 8. New constructs were cloned into members of the same Duet vector set described in Chapter 2
Methods. Plasmids were constructed using standard molecular biology techniques with restriction
enzymes and T4 DNA ligase purchased from New England Biolabs. All primers used for cloning and
plasmid names can be found in Table A4-1.
Culture conditions
For all production experiments 3 ml LB overnight seed cultures in 14 ml round-bottom tubes were used
as inocula. All production cultures were inoculated with overnight culture at 1% by volume. The
production medium was LB with either 1.2% (w/v) glucose or 1.2% (v/v) glycerol. All production cultures
were induced with 0.5 mM IPTG (final concentration) at OD600 values between 0.7 and 0.9. All seed and
production cultures were incubated with agitation at 30'C and 250 rpm. Tubes and vials were placed in
a rake at a 450 angle.
For production of butanol and pentanol using Modules 2-MCC and 4, 3 ml cultures in 50 ml (1
inch diameter) capped glass tubes were used (as for 4MP production in Chapter 2 Methods).
85
Experiments used biological triplicates. Samples for liquid chromatography analysis were taken 48
hours post-induction. For production of alkanes 2 ml cultures in 11 ml, septum-capped gas
chromatography vials (Supelco SU860099 10 ml 22.5x46 mm vials and SU860103 PTFE silica septa caps)
were used. Technical triplicates were used for propane, butane, pentane, and 4MV to 2-methyl-butane
experiments. Biological triplicates were used for isobutyrate to 2-methyl-butane and heptane
production experiments. Gas head-space samples were taken 24 hours post-induction as described
below.
Gas chromatography method
At 24 hours post induction, vial cultures were placed in a 420 C incubator for 20 minutes in order to drive
the volatile alkanes into the gas phase. Gas headspace samples were then taken using a 10 ml gas-tight
syringe (SGE Ringwood, Victoria, Australia). In order to mix the gas sample and prevent formation of a
vacuum one syringe was used to inject a 1 ml volume of air as 1 ml of sample was drawn. The process
was repeated until a 9 ml volume was taken from the vial and an additional 9 ml of air was injected into
the vial. The concentration injected into the GC was thus diluted 2-fold. A Shimadzu GC (GC-2014) with
a RT-Q bond column (30 m length, 530 pm ID, 20 pm film thickness) and flame ionization detector (FID)
was used for the analysis. A 5 pL sample loop was used for sample injection. The method oven
conditions were as follows: a 40"C hold for 1 minute followed by a 25"C/min ramp up to 280*C with a 5
minute final hold. Quantification of propane, butane, and pentane was based on a one point calibration
using a standard gas mixture purchased from AIRGAS. The quantification of 2-methyl-butane was based
on the FID response for the isomer pentane (Sample calculation Text A4-2). A separate standard curve
for heptane was created by adding known volumes to a 1 liter glass bottle fitted with a septum cap. A
heptane standard (Sigma-Aldrich) and the 1 liter bottle were chilled to 4'C and different known volumes
of heptane were added. The bottle was then warmed to room temperature allowing the heptane to
86
fully vaporize. A gas tight syringe was used to inject 8 ml of gas from the bottle into the GC (Text A4-3,
Fig. A4-1). A similar curve was generated for nonane.
Liquid chromatography method
The same liquid chromatography method described in Chapter 2 Methods was used for liquid phase
analysis. The method had a 120 minute stop-time and 50 ptl injection in order to better quantify
aldehyde product. Butyraldehyde was quantified by making a standard curve using an analytical
standard from Sigma-Aldrich.
87
Chapter 5: Future directions
Improving the CoA-dependent platform for branched products
As mentioned in the Introduction, a variety of biofuel targets have been synthesized at low yields, but
few technologies have been developed to the point of commercialization. Chapter 2 described how we
established an efficient pathway-architecture to guide the design of the CoA-dependent 4MP pathway.
Continued pathway improvement can generate potentially valuable results because 4MP synthesis was
established using a redox-neutral, theoretically high yielding pathway. Improvements upstream of 4methyl-valeraldehyde synthesis can also be applied to the 2-methyl-butane pathway described in
Chapter 4. Our initial demonstration revealed two areas for future pathway development: 1)
replacement of key rate-limiting enzymes and 2) alteration of cofactor utilization in order to improve
pathway performance in a fermentative process.
Key rate-limiting enzymes
The final pathway version in Strain M1Fj(IA)2134L (containing Fjoh2967Fj, the ivCEc-asSBS operon, ibuARp,
and IsadhLs) produced 4MP selectively among the potential reduced products, but also built up large
pools of isobutyrate and acetate precursors. A collection of observations suggest that an enzyme within
Module 3 which completes the chain extension of isobutyryl-CoA to 4-methyl-valery-CoA is rate
limiting. Previous in vitro activity data for the C. necator thiolase BktBCn and reductase PhaBc, suggest
that BktBCn may have more substrate flexibility than PhaBcn 134 ,139. Additionally the condensationthiolysis equilibrium greatly favors the thiolytic products which means the reductase reaction drives flux
towards downstream porducts1 40 . When constructing synthetic operons for full 4MP pathway
expression, we considered the possibility of a PhaBcn limitation. Our initial operon designs placed
phaBcn in the second position of a two-gene operon likely lowering enzyme expression. An alternative
plasmid containing phaBcn in the first position of the operon increased 4MV product titers from
isobutyrate and glucose by 97% (Fig. 2-10). The 4MV:butyrate ratios also changed significantly from
88
0.78:1 with phaBc, in the second postion to 1.6:1 with phaBcn in the first position. Together these data
support the hypothesis that PhaBcn can be rate limiting and a key control point for the selectivity of acid
products formed by CoA-dependent chain extension.
Our laboratory has explored using PhaBcn for production of other acid products and work has begun
to identify improved reductases7 9. A current student, Yekaterina Tarasova, has generated a sequence
similarity network based on a previously described acetoacetyl-CoA reductase protein family which
contains PhaBcn and over 2,000 other sequences (Fig. 5-1)141,142.
I.jt
--
-%
1.
While the authors describe the family
%1
PhaB
FabG
PhaB
FabG
~X
0-ketoacyl thioester reductases. A representative sequence similarity network of a family
of 0-ketoacyl thioester reductases is shown at an E-value edge threshold of 10- . Representative nodes group
all sequences with at least 90% sequence identitiy. The large cyan node indicates the location of PhaBco.
Clusters which are connected to general FabG-like and PhaB-like clusters at an E-value edge threshold of 10-75
are indicated with labels. Potential NADH-dependent reductases, based on sequence analysis of characteristic
amino acid residues, are highlighted by red nodes. Larger purple nodes indicate recently characterized NADHdependent FabG reductases which were identified independently using sequence analysis. The blue node in the
large PhaB cluster indicates an available crystal structure for a potential NADH-dependent PhaB-like reductase.
Fig. 5-1. Network of
89
as an "acetoacetyl-CoA" reductase family, the majority of sequences in the family are related to FabGlike P-ketoacyl-ACP reductases involved in type 11 FAS. When the network is viewed at an E-value edge
threshold of 10-75 (edges connect protein nodes which have pairwise BLAST E-values5 10-7s ) a cluster of
sequences breaks off which contains 1,142 sequences including PhaBcn (network not shown at this
threshold). This cluster likely represents a set of CoA reductases and can be used to identify alternate
PhaB-like candidates which may have differing substrate specificities.
Care should be taken in selecting alternative PhaB reductases since the assays used for evaluation
will likely be low throughput because the desired acyl-CoA substrates are not readily available. The in
vivo assay for 4MV synthesis from isobutyrate and glucose described in Chapter 2 is useful for
comparing reductase performance of a small candidate library. In addition to isobutyrate, other acids
could be supplemented to adapt the assay for alternative products. An iterative approach should be
used to gain an informed perspective of the enzyme space in order to best utilize the in vivo assay. We
are starting with an enzyme we know will catalyze the desired chemistry and we would like to enhance
kinetics and specificity. Because it is possible that small differences in protein sequence could have
large effects on our desired enzyme characteristics we should initially look in close proximity to PhaBcn.
As mentioned above, the PhaB network is representative and nodes can contain more than one protein
sequence. The node containing PhaBCn contains 26 other protein sequences some of which have as low
as 91% identity with PhaBCn (Table A5-1). In order to gauge the potential variation in substrate
specificity, 10 of these close neighbors can be cloned to replace PhaBcn in the existing pETDuet-1 based
Module 3/2-MCC plasmid (described in Chapter 2). Differences in 4MV and butyrate titers can be
correlated with differences in sequences to help inform what enzyme characteristics improve
performance.
If no differences are observed between reductases within the PhaBcn node, the search can be
expanded to less similar enzymes. Reduction of the E-value edge threshold to 10-82 breaks the PhaB
90
cluster into 4 sub-clusters and the main PhaB cluster shows 4 distinct groupings within it (Fig. 5-1). A
larger library of perhaps 20-40 candidates can be selected, sampling evenly across the 8 sub-groups. If
this broader approach is implemented, it would be wise to include a small library of enzymes taken from
clusters that were not connected to either the PhaB-like or FabG-like clusters at an E-value edge
threshold of 10-71. The space around candidate enzymes which prove to enhance 4MV titer or
specificity can be targeted for further investigation. Individual sub-clusters or groupings can be
surveyed more completely until specific nodes with the highest activity and specificity are identified.
Reduced DNA synthesis pricing facilitates the described approach because genes can be acquired quickly
and low expression due to codon bias in the source organism can be avoided.
A plateau in titers irrespective of PhaB variant may indicated that another pathway enzyme has
become limiting. The other CoA-dependent enzymes downstream of PhaB (PhaJ, Ter) could be
investigated in a similar manner. In the initial screening of PhaJ and Ter variants the greatest differences
were observed between the Ter variants (Fig. 2-5A). As with PhaB, Ter is a reductase which uses a
reducing cofactor. It is believed to use a kinetic trap which can provide a strong pull on CoA pathway
flux1 s, 143. As a result, enhanced activity and/or specificity for 4-methyl-trans-2,3-pentenyl-CoA has the
potential to significantly improve pathway performance.
Altering cofactor utilization
Microbial alcohol production can be greatly simplified by using a fermentative bioprocess. In order to
make the CoA-dependent pathway platform more efficient under anaerobic conditions, all reductases
must utilize NADH cofactors. The NADH requirement results from the cell's need to regenerate NAD+ as
it uses glycolysis for ATP generation under anaerobic conditions. In the current CoA-dependent 4MP
pathway two reductases, PhaBco and CarNi, use an NADPH cofactor. Finding alternative NADHdependent enzymes for these steps would greatly enhance pathway performance under fermentation.
91
In addition to identifying PhaB diversity, the sequence similarity network can also help identify
NADH-dependent substitutes for PhaBCn. Key catalytic residues which determine cofactor specificity
have been described for the short-chain dehydrogenase/reductase (SDR) superfamily which contains pketoacyl thioester reductases72 . In brief, glutamic acid or aspartic acid is often found at one of three
sequential positions in NADH-dependent enzymes, while arginine or lysine is often found at one of four
positions in NADPH-dependent enzymes. We have annotated the network based on those residue
patterns in order to identify potential NADH-dependent enzymes (Fig. 5-2A, Text A5-1).
A
Ph4aB
FabG
PhaB
FabG (sc)
FabG (fc)
FabG (sc)
FabG
Fig. 5-2. Potential NADH dependent reductases.
(A) Select clusters are shown from the network
in Fig. 5-1. Nodes with potential NADHdependent enzymes are red. Two purple nodes
in the large FabG cluster contain enzymes with
experimentally confirmed NADH specificity. The
blue node indicates a potential NADH-dependent
PhaB-like enzyme with an available crystal
sc=single
(fc=few
connections,
structure.
connection to FabG cluster at an E-value edge
threshold of 10-75) (B) The PhaB-like structure
(blue) is aligned with a reference NADHdependent SDR with NAD+ bound. The key
glutamic acid residue which interacts with
NAD(H) hydroxyl groups is indicated.
92
Only homologs which contained at least one of the key acidic residues and none of the key basic
residues were annotated as potential NADH-dependent reductases (Fig. 5-1 and 5-2A).
It is possible for cofactor specificity to be determined by other motifs and the PhaB-like clusters
have relatively higher levels of diversity at the key residue positions. In fact, PhaBcn has neither acidic
nor basic residues at any of the four positions, but the sequence confers NADPH specificity. Despite the
diversity, the majority of sequences in the PhaB-like clusters have basic residues at one or more of the
key positions. Among the likely NADPH-dependent reductases there are a minority of nodes which fit
the criteria for potential NADH-dependent enzymes (Fig. 5-2A). Because the majority of nodes indicated
NADPH specificity as has been observed for PhaBco, we sought additional evidence to support the
contention that some PhaB-like reductases are NADH-dependent.
One potential source of false positives is poor sequence alignment for a subset of proteins. It is
possible that the alignment positions for some sequences do not correspond to the expected positions
in the protein structure. It is the position in the three-dimensional structure which ultimately creates
the desired enzyme-cofactor interactions. Fortunately, a structure was available for one potential
NADH-dependent PhaB-like node indicated in blue in Fig. 5-2A. An alignment of the PhaB-like structure
with a reference NADH-dependent SDR using UCSF Chimera shows the potential glutamic acid residue is
in good position to interact with the two hydroxyl groups of NAD+ (Fig. 5-2B) 144.
Further supporting evidence was found for our screening method by using the UniProt database
to retrieve literature references associated with our potential NADH-dependent enzymes1.
Javidpour
et al. characterized three of the FabG-like enzymes we identified and found them to be NADH-specific
even though they are homologous to FAS reductases which typically use NADPH (Fig. 5-2A)113 . These
proteins are found in a FabG-like cluster where we predicted the majority of the nodes to contain
NADH-dependent reductases. The combination of these data gives us confidence that investigation of
the PhaB-like nodes can uncover NADH-dependent enzymes.
93
Once again the 4MV in vivo assay can be used to screen libraries of candidates. There are 56
potential PhaB-like NADH-dependent sequences represented by 13 nodes (Table A5-2). One candidate
can be selected from each node and cloned to replace PhaBcn in the Module 3/2-MCC plasmid. With an
NADH-dependent @-ketoacyl-CoA reductase, the pathway from isobutyrate and glucose would utilize
only NADH cofactors. Comparison of product titers from aerobic and anaerobic (transiently aerobic)
cultures could be used as a first pass screen to identify NADH-dependent enzymes. We would expect
NADH-dependent enzymes to produce more acid product relative to PhaBcn under anaerobic conditions
where the NADH:NAD+ ratio is much higher6 7 . The best candidates can be purified and assayed with
commercially available acetoacetyl-CoA and NADH/NADPH to confirm cofactor specificity. If additional
proteins are found at a confirmed NADH-dependent node, those sequences should be assayed as well
with the hope of finding an enzyme with even more favorable kinetics.
A similar approach could be applied to alternative carboxylic acid reductases. N. iowensis CarNi
is part of a much smaller family of enzymes. Within the UniProt database, only 354 sequences display
domain architecture similar to CarNi. The sequences are predominantly from Mycobacteria. Because
this class of carboxylic acid reductases has less known diversity, alternate routes to aldehydes may be
sought. CoA-dependent aldehyde dehydrogenases (Aldh) from Clostridia could provide a potential
alternative although there activity has only been observed on a butyryl-CoA substrate 46 . If no natural
enzymatic route can be identified, it may be possible to engineer CarNi to accept NADH
147
The combined knowledge gained from a search for improved substrate specificity and for
cofactor specificity would enable rational enzyme selection for specific applications of the CoAdependent pathway platform. Correlations between sequence and substrate/cofactor specificity may
also be used to help further tailor improved substrate/cofactor specificity. Ideally, replacement
enzymes could be chosen to relieve the kinetic bottleneck under either aerobic or anaerobic conditions.
94
If these goals are achieved, the CoA-dependent pathway would move from a demonstration to a
platform with commercial relevance.
An orthogonalbranched FAS
While a small number of key enzymes can be targeted for improvement of the CoA-dependent platform,
a fresh approach is warranted for developing an improved branched FAS system. Chapter 3 described
our initial development of a branched, short-chain FAS platform in E. coli highlighting the rate limitation
likely due to pairing the plant FatB2Ch thioesterase with a bacterial FAS. When this work began we
envisioned developing an orthogonal branched FAS pathway which may be transferable to other hosts,
especially yeast strains. Such a platform must have a number of characteristics which were not carefully
evaluated in development of our demonstration pathway. As mentioned in Chapter 3, we selected a
branched FAS which was well understood and closely related to our host FAS in order to increase the
likelihood of creating a functioning pathway. Independently, we selected a thioesterase which, at the
time, acted on the shortest reported acyl-ACP substrate. In development of a more efficient platform
both acyl chain preference and ACP interactions should be considered simultaneously (Fig. 5-3).
A
host
FAS enzyme
m
host
FAS enzyme
host
ACP
recombinant
ACP
MAW
recomb nant
5-3.
Characteristics
of
an
orthogonal type 11 FAS platform.
(A) Host and recombinant ACPs
can interact properly with FAS
CP
recombinant
FAS enzyme
recombinant
FAS enzyme
Fig.
enzymes from their own synthase,
but do not "crosstalk" with the
B
recombinant
group
host
opposite
synthase.
recombinant
system
(B)
The
contains
a
thioesterase which will act on acylACP intermediates linked to the
recombinant ACP, but does not act
on host acyl-ACP intermediates. (C)
free acid
c
The recombinant FAS generates
branched acids from branched CoA
z
oA
-
A
thioester precursors which can be
derived from branched amino acid
synthesis.
-
branched
acyl group
95
Identifying an orthogonal ACP
For any FAS platform to be orthogonal the acyl-ACP intermediates of the recombinant pathway must not
be able to interact with the host FAS and the acyl-ACP intermediates of the host must not be able to
interact with the recombinant FAS. From our initial work, and the work of others, it is clear that E. coli
FAS enzymes can handle a wide range of acyl groups. The key, then, to creating an orthogonal pathway
is the identification of a recombinant FAS which uses an ACP that cannot interact with E. coli enzymes
(Fig. 5-3A). In our initial demonstration we considered this characteristic and looked to see that some of
the key residues of the interacting helix on B. subtils ACPBs were different from key residues on E. coli
ACPEC. Despite this analysis it is clear from our results that B. subtilis FabH1Bs and FabH2BS can use
Likewise it has been reported that B. subtilis
ACPBs
ACPEc-
can complement the essential knockout of ACPEc1 4 8
Based on these results, it is unlikely that sequence analysis alone will identify orthogonal ACPs.
While ACP sequences are conserved across a wide range of life, there is still sufficient diversity
to support the existence of orthogonal FASs (Fig. 5-4). We created a representative sequence similarity
network to help identify where to look for functional diversity when creating the library. Sequences 100
amino acids or shorter were taken from the Pfam PP-binding domain family which contains ACPs. Each
node contains sequences with at least 40% identity and we are displaying the network at an E-value
edge threshold of
10-1.
We have highlighted a number of key sequences related to our work, species
known to produce branched acids, and key enzymes previously assayed with E. coli FAS. Roughly half
the sequences are grouped tightly within the main cluster and that group includes ACPEc and ACPBs as
well as sequences from plants and cyanobacteria. L. lactis ACPLI is found loosely connected to the main
cluster. The ACPs of a number of bacteria known to produce branched fatty acids are located on the
periphery of the network
. Many plant ACPs are not included in the network because the database
sequences include signal peptides which make their sequences ~140 amino acids in length.
96
(4 'V
IfI
'7V
I
* Escherichia
0 Streptococcus
0
Bacteroides
* Nostoc
0
Clostridium
*
Bacillus
* Arabidopsis
0
Lactococcus
0 Flavobacterium
0
A representative sequence similarity
Fig. 5-4. Sequence similarity network showing the diversity of ACPs.
network of Pfam PP-binding domain family proteins of 100 amino acids is shown (5,336 unique sequences).
Nodes represent a set of sequences which have at least 40% identity with each other. An E-value edge
threshold of 10-1 is used to show node connections. Nodes containing sequences from select genera are
highlighted. The largest node (1,144 sequences, shown in red) contains ACPEc. Half of the sequences are
found in the large grouping of the major cluster which contains the Escherichia node. This grouping also
contains sequences from Nostoc punctiforme (cyanobacterium), Arabidopsis lyrata (plant), and a branched
acid producer Streptococcus agalactiae. Clostridium acetobutylicum, Lactococcus lactis, and another
branched acid producer Bacteroidesfragilis are found further out in the main cluster. Finally branched acid
producers Bacteroides intestinalis, Flavobacterium indicum, Bacillus cereus, Bacillus amyloliquefaciens, and
Bacillus megaterium are disconnected from the main cluster.
Knowing that nodes represent fairly diverse sequences down to 40% identity, the structure of the
network with weak connections for a large set of sequences suggests orthogonal ACPs can be found.
Unfortunately, based on previous observations it does not appear that distance within the
network alone will help identify large groups of orthogonal ACPs. This phenomenon often arises when a
97
.1
*....|...
5
B. intestinalis
F. indicum
B. cereus
B. subtilis
A. thaliana
C. 0.ancrata
C.
lanceolata
----------------------------------------------
------- MEK~
-----
Z., I&Ctlv
E. Coll
---MA
A. thaliana
S.
OldagNM0
C. lanceolata
1....|
15
-------------------------------------
*.
.
......
I
25
*.
......
35
*.
......
*.
45
----------------------------
-------------------
-------------------
-------------------
-------------------
NH------GK
TNLSFNLRRS IPSRRLSVSC
SFNSKNYALK SSVTFNRMTP VMPRGLSVSC
SLRSVSLPVS RKSFPSLKSS KSSFALRVSC
*........
85
.... l .... l
1VNLFDDIDT
SNFSVLTDFK
-GINNQLKIT
EN---GDD-EEVT-LETSF
EEVTNNASFV
ADVKLEASFK
DKKVVAETKF
DAKVTGETKF
DSEVNGLSKF
-- MSTt
-- MAD
AAKQE
AAIKPEWOM
QAKPEr4UV
---- IDV
---- KEK
---- VKQ
---- VDE
"-S----LTP
LA----LAE
OLA ---- LPD
. . . . I... .
....I....1
135
L.
|
55
----------------------------
Ni*E----INH
125
......
----------------------------
95
.1....
105
A
B. intestinalis
F. indicum
B. cereus
.
----------------------------
MATQFSASVS LQTSCLATTR ISFQKPALIS
MA-----SIT GSSVSFKCAP LQS ------MAS----AAA GASICIKSAS FSPLAPGRIS
B. intestinalis
F. indicum
B. cereus
B. subtilis
....
.1....
115
.L.
145
YGIII
LTKERG-
WDFNCLND
*IVLSX
rNI
lacta
E . Coll
DT E I
B. subtilis
A. thaliana
rDMEIS
C. lanceolata
FGIS
OKIAT--V*
jIAT--V%
FNI
EKKNK-KKGH
nTOT--VnD
. ..
.L V
Fig. 5-5. Sequence alignments of diverse ACPs. ACPEc is aligned with four ACPs used in a complementation
assay (Bacteroides intestinalis, L. lactis, B. subtilis, and S. oleracea) and four other ACPs of interest
(Flavobacterium indicum, Arabidopsis thaliana, Cuphea lanceolata, and Bacillus cereus). The two ACPs
found to not support growth in E. coli are boxed in grey. Grey boxes are also used to show positions aligned
with a-helices I-IV in ACPEc. Variation at residues that were found to effect growth when mutated in ACPEC is
highlighted by red boxes. A potential key variation in Helix II of the S. oleracea ACPs, is boxed in blue. Signal
peptides for the three plant species are indicated with an underline. The cysteine which attaches to the 4'phosphopantetheine prosthetic group is indicated with a triangle. The two Bacilli are found to have very
different sequences. ACPBS is more similar to ACPEc while ACPBc more closely resembles ACPuI-
specific function is controlled by a small number of amino acid residues. To illustrate this point, we have
created an alignment with ACPEc, four ACP sequences used in a previous E. coli complementation assay,
and four other ACP sequences of interest (Fig. 5-5)149 . A range of bacterial ACPs have previously been
used to complement a temperature sensitive mutant of ACPEC at the nonpermissive temperature 14.
randomly selected set of bacterial ACPs generally supported growth although there were two key
98
A
exceptions including ACPLI from L. lactis. The mature ACP of S. oleracea (Spinach) was also found to be
incapable of supporting growth at the nonpermissive temperature. Interestingly a number of designed
ACPEc
mutants also could not support growth, but natural variants, like one Bacteroides ACP, which
contain the same residue at the same position as the mutant could support growth. The Bacteroides
ACP also has relatively low overall sequence identity to E. coli. These seemingly contradictory results
may occur because ACP is not required to make one interaction with one enzyme, but is required to
make a series of interactions with a set of FAS enzymes in order to support growth. It is likely that poor
ACP function can arise from a variety of unrelated interactions created by different sequence variations.
It is for this reason that a random or semi-rational approach is required to find an orthogonal system.
We propose a combined approach of "characteristic sequence" learning through in vivo
screening of natural homologs. Instead of trying to create a series of point mutations to explore the
effect on enzyme function, a library of natural homologs, selected for diversity, can be used to identify
which "characteristic sequences" generate desired phenotypes. In this specific case, an assay must be
devised such that ACP homologs can confer either a positive or negative phenotype (i.e., growth or no
growth). Fortunately, as mentioned above, a temperature sensitive mutant of ACPEc has been identified
such that E. coli expressing only the mutant ACP grows at 30'C, but cannot grow at the nonpermissive
temperature of 42*Cls0 . This temperature sensitive mutant can be used with a large library of
recombinant ACP homologs to help identify other sequences which fail to support growth at the
nonpermissive temperature. The library would be transformed into an E. coli strain containing only the
temperature sensitive
ACPEc
mutant in its chromosome and plated at 30*C for initial growth. Replica
plates or patches can then be made and grown at 42 0 C. Strains which fail to grow or grow poorly can be
sequenced to identify which ACPs fail to support growth and a selection of strains which do support
growth can also be sequenced. The small size of ACP genes (-240 bp) allows for cheap synthesis, or
even assembly from DNA oligos, increasing the size of the library that can be used.
99
Pairing ACP and thioesterase activity to complete the pathway
Observations from the growth screen can be annotated on the ACP sequence network. Sequence
alignments of non-growers can be used to identify patterns which lead to the negative growth
phenotype. This sequence information can then be applied to the network. Once the network is
sufficiently characterized, species which are known or suspected to express short- or branched shortchain acyl-ACP thioesterases can be annotated to see if overlaps exist between orthogonal ACPs and
desired thioesterase activities. The fact that E. coli FAS cannot use a given ACP does not ensure that a
recombinant thioesterase from that species will not act on E. coli acyl-ACPEC- It is likely that only a small
library of potential thioesterases will be tested because of the use of low throughput in vivo assays.
Select thioesterases would be expressed in E. coli with cellular lipid content and free fatty acids
monitored as in the FAS platform work described in Chapter 3. If a thioesterase is expressed, but does
not alter E. coli lipid content, then other components of the recombinant FAS can be added to test for a
complete orthogonal pathway (Fig. 5-6).
type 11 FAS species
Fig. 5-6. Finding an orthogonal FA:
branched acid pathway in type 1 FAS
species space. All type I FAS systems
utilize ACPs. A subset of such systems
A
contains acyl-ACP thioesterases.
second subset uses ACPs which are
orthogonal to ACPEc. Within systems
using thioesterases, a subset act on
short, branched acyl-ACPs.
seeking the intersection
We are
of the
orthogonal ACP and short, branched
acyl-ACP thioesterase spaces.
orthogonal FAS
branched acid pathway
It is not clear that this described approach will be efficient enough to identify the desired
pathway platform, but a number of observations suggest the goal may be ultimately achievable. As
stated earlier, a number of ACPs have been found incapable of complementing growth of the
temperature sensitive ACPEC mutant at the nonpermissive temperature including an ACP from one plant
100
species. Many plant species express acyl-ACP thioesterases and, as mentioned in the Chapter 1, at least
one tobacco species produces 4-methyl-hexanol via fatty acid biosynthesis. The question is whether or
not a species which uses a branched short-chain acyl-ACP thioesterase also uses an ACP divergent
enough from E. coli to make an orthogonal pathway possible. Even if an ideal system is not identified,
pursuit of this project can uncover new knowledge about type II FAS systems which can aid in alternative
pathway designs.
Improving alkane synthesis through alternative decarbonylases
If alkanes are a desired product from either an FAS or CoA-dependent platform, then rate limitations
from the final aldehyde decarbonylase reaction must be addressed. The genes responsible for AD
activity have only been discovered recently and results from engineering AD pathways in recombinant
hosts have produced a range of alkane titers", ",". The proposed mechanism for decarbonylation is
complicated, requiring molecular oxygen, reducing equivalents, and water in addition to the aldehyde
substrate12 9. To date, two families of enzymes with AD activity have been described. One consists of
the mono-functional ADs of cyanobacteria and the other is composed of the CER1-like enzymes from
plant species15 1 . Only a small collection of potential enzymes from both families have been
experimentally tested in the literature. It is possible that variations in efficiency exist, especially
considering that electron transfer rates appear to affect enzyme inactivation3 . Whether ferredoxin or
NAD(P)H cofactors are required has not been carefully explored for all enzyme variants.
Alternative decarbonylases within the cyanobacteria AD family
In our work described in Chapter 4 we used a wild type and mutant version of a single AD from P.
marinus MIT9313, but a number of homologs have been identified for this enzyme. This variant was
selected for an initial demonstration because in vitro data for the mutant showed enhanced activity on
short-chain aldehydes. An alternative homolog of ADpm may have improved kinetics due to enhanced
electron transfer or stability in our recombinant host. If this is the case, it may be possible to engineer
101
the enhanced short-chain activity into a better performing AD. As in the case of improving the CoAdependent platform, we will describe a process for investigating alternative enzymes for this key kinetic
bottleneck.
The P. marinus AD is included in the DUF3066 family within the Pfam database15 2 . The DUF3066
domain family consists of 135 sequences, the majority of which come from cyanobacteria. We used the
web-based Enzyme Function Initiative's Enzyme Similarity Tool (EFIEST) to generate a sequence
similarity network for the DUF3066 family1s3. When identifying AD genes, Schirmer et al. assayed a
variety of AD homologs in E. coli 30. AD homologs were expressed with the acyl-ACP reductase from
Synechococcus elongatus PCC7942 which produced aldehyde substrate for the ADs30. Different alkane
titers were observed for different decarbonylases. We have annotated our AD network to show which
enzymes were assayed in this study and the magnitude of titers generated by each (Fig. 5-7). While we
selected ADpm based on confirmed activity for our substrates, it was also one of the poorest performing
enzymes of the initial tested set. The AD from Nostoc punctiforme PCC73102 produced the highest
titers of over 30 mg/L of both pentadecane and heptadecane. In addition this homolog produced the
only detectable tridecane.
N. punctiforme
PCC 73102
P. marinus
MIT 9313
30-40 mg/L
10-20 mg/L
5-10 mg/L
C- 5 mg/L
0-.0.-0
0
102
~
0-000 0 0
O -0-0
0 0
-
0 00
0 0 0
Fig. 5-7. Family of cyanobacterial
aldehyde decarbonylases (ADs). A
sequence similarity network of the
Pfam DUF3066 domain family is
shown with an E-value edge
-112
threshold OT 1U . Decarbonylases
assayed in E. coli are highlighted by
large orange and brown nodes
indicating the range of pentadecane
and heptadecane produced by each.
The
high
producer from
N.
punctiforme and the P. marinus
homologs are indicated. In addition,
an AD from the y-proteobacteria
Alcanivorax dieselolei is highlighted
by a large blue node. A. dieselolei
has been found to degrade alkanes.
The titers published by Schirmer et al. suggest that N. punctiforme PCC73102 AD (ADNP) is the
most active of the decarbonylases tested in E. coli. It may be possible to use ADNP to increase shortchain alkanes when coupled to our platform. Because the A134F mutation significantly increases
activity of ADpm for C4-C6 aldehyde substrates, we investigated the likelihood of introducing a similar
mutation in ADNP. We retrieved a homology model of the ADNP protein based on the ADpm structure
from the ModBase database and aligned the structures using Chimera 15 4 . Because the ADpm structure
was solved with a substrate analog bound in the active site, we were able to identify all binding pocket
residues within 5
A of the substrate.
These residues are shown in the sequence alignment below (Fig. 5-
8). While there is significant sequence variation between the two ADs (56% identity), all binding pocket
residues are identical. ADNP should be used in place of ADpm within our existing alkane pathways to test
. . . . | . . . . | .... . . . .* . . .... ....
-
I. . . . I ....I....I ....I....1
25
35
45
5
15
55
P.marinus
MPTLEMPVAA VLDSTVGSSE ALPDFTSDRY KD4SREAI IEQ4HD NYIAIGTLLP
N.punctiforme ----------- -- MQQLTDQS KELDFKSETY KD4SE4 IIEW4E
NYITLAQLLP
. . .-.
65
75
P.marinus
DHVEELKRLA K4I3KK
N.punctiforme ESHDELIRLS
85
.
.... . .
95
....I....| ....I....1
105
115
APLRDNFQTA LGQGKTPTCL
TGKNLGVE ADMDFARE
M4GRNLAVT PDLQFAKEF* SGLHQNFQTA AAEGKVVTCL
....I.... |I ....I....|1 ....I....I ....I....I ....I....I ....I....1
125
145
155
165
175
135
P.marinus
LI
IS
T IP VSDPFARKI EGVVKDIYTU LNYGEAWLKA NLESCREELL
INIP VADDFARKIf EGVVKEUYSE LNFGEVWLKE HFAESKAELE
I
N.punctiforme LI L
....I....I ....I....1 . . . .
..
61..
.....
.0
205
215
185
195
I ..... . .
225
-..
. . . .. . 1
235
P.marinus
EANRENLPLI RPMLDQ GD AAVLQ3 KED LIEDFLIAYQ ESLTEIGFNT REITRMAAAA
N.punctiforme LANRQNLPIV WM4LN*GD AHT "4KDA VEDFMIQYG EALSNIGFST RDIMRLSAYG
P.marinus
LVSN.punctiforme LIGA
Fig. 5-8. N. punctiforme PCC73102 and P. marinus MIT9313 aldehyde decarbonylases share the same
hydrophobic binding pocket residues. An alignment of the protein sequences of ADNp and ADpm reveal that
hydrophobic binding pocket residues are strongly conserved between variants. All residues found to be within
5 A of the bound substrate analog in a structure alignment of the two proteins are highlighted in red. The
position of the A134F mutation of ADpm is shown with the blue box and arrow. This position would correspond
to an A122F mutation in ADNP-
103
for improved production. The high similarity in binding pockets supports the possibility that a functional
ADNp-A122F mutant can be created to enhance ADNp activity with short-chain substrates. The
ADNP-
A122F mutant should be evaluated in comparison to ADNp wild type and the ADpm-A134F mutant.
Based on the distribution of observed activities within the AD network, it is likely that all
members of the DUF3066 family function as aldehyde decarbonylases. Five to ten homologs from
groupings within the main cluster and from the untested cluster could be expressed within our pathway
context to better characterize the space. Some variants may have improved electron transfer within an
E. coli host. In addition, because a structure is available, variation in binding pocket residues can be
overlaid onto the network. Specific positions which could potentially be used to occlude larger
substrates or smaller branched substrates should be looked at to identify natural homologs which may
have improved substrate specificity. Selecting a set of AD homologs with diversity at these positions
may increase the efficiency of the screening process.
The relatively small size of the DUF3066 family suggests that relatively little experimentation is
required to learn whether significantly improved AD homologs exist within the network. The small
libraries required to investigate the space make continued use of in vivo pathway assays practical.
Screening candidates within the pathway context greatly reduces the likelihood of pursuing enzymes
which may not improve alkane synthesis from a sole carbon source. Unfortunately the small network
size also lowers the likelihood of finding an enzyme with substantially improved kinetics.
CERi-like plant aldehyde decarbonylases
If cyanobacterial ADs are found to be incapable of significantly improving titers, plant CERI-like
decarbonylases may be considered. It was expression of the CER1 gene of Arabidopsis thaliana with a
FAS based pathway to aldehydes which has generated the largest published alkane titers". While
cyanobacterial ADs are presumed to be soluble mono-functional enzymes, CER1-like enzymes contain
two catalytic domains with multiple transmembrane domains. The N-terminal catalytic domain is a
104
B
A
CER1
/_
*
0~ 0
I
41
Fig. 5-9. Protein context for CERi-like aldehyde decarbonylases. (A) A representative sequence similarity
network of the FA hydroxylase super family is shown. The superfamily includes enzymes which utilize a di-iron
core to catalyze a variety of reactions. The CER1 protein is highlighted by a red node in the cluster containing
CER1 homologs. Each node contains all sequences with greater than 90% identity. An E-value edge threshold
of
10-90 was used.
(B) A sequence similarity network is shown for members of the FA hydroxylase
superfamily with sequences longer than 500 amino acids. An E-value edge threshold of 10-105 was used. The
CERI-like cluster and two other multi-domain clusters are highlighted. A bacterial FA hydroxylase-YhhN cluster
is shown in purple and a bacterial FA hydroxylase-cyclopropane synthase cluster is shown in turquoise. Both
bacterial clusters contain enzymes that are likely used to tailor membrane lipids. The CER1-like cluster
bifurcates into two groupings using the edge threshold shown.
Interestingly many of the plant species
represented in the cluster, including A. thaliana, have more than four homologs of CER1. The homologs of A.
thaliana are highlighted. CER1 is shown as the large dark red node with other homologs in the same grouping
shown in light red. CER3 is found in the opposite grouping at the large blue node. Other plant species show a
similar distribution between the two groupings potentially indicating that multiple CER1 homologs are
expressed with CER3 to produce a variety of plant waxes.
member of the FA hydroxylase superfamily and contains a di-iron catalytic core similar to that found in
the cyanobacterial AD family of enzymes (Fig.
5-9)151.
The C-terminal catalytic domain is a member of
the WAX2 C-terminal domain family within the Pfam database. WAX2 C-terminal domains have an
unknown function, but they are similar in sequence to members of the short-chain reductase (SDR)
family1 55. When CER1 or its homolog CER3 was expressed alone in S. cerevisiae, alkanes were not
observed, but coexpression of CER1 and CER3 led to C29 alkane productions 1 . Mutation of proposed
iron coordinating histidine residues in CER1 abolished alkane production further implicating the FA
hydroxylase domain as an aldehyde decarbonylase. Further experiments in S. cerevisiae suggested that
105
very long chain acyl-CoAs are the substrate for a CER1-CER3 complex which produces very long chain
alkanes. Somewhat surprisingly, Choi et al. have recently reported synthesis of high titers of nonane in
E. coli by expressing an unmodified version of CER1 alone.
In the E. coli work free fatty acids were
converted to acyl-CoAs by a ligase and an acyl-CoA reductase was expressed to generate aldehydes. The
higher alkane titers could result from some direct activity on acyl-CoAs by CER1 which has not been
described.
While the complex interactions between CER1 and CER3 suggest simple CER1 expression in E.
coli should not enhance aldehyde production, the evidence produced by Choi et al. warrants
experimental investigation. If CER1 can be functionally expressed with both the CoA-dependent
platform to C4-C6 aldehydes and the FAS platform to C8 aldehydes, alkane titers may be enhanced and
new information may be revealed about which substrates CER1 can use. Strains should also be
constructed which replace both CarNi and ADpm activity with CER1 to test if CER1 can act directly on
either acyl-CoA or acyl-ACP substrates. If CERI is found to be functional, alternative plant homologs
could be investigated from the CER1-like cluster. While mono-functional decarbonylases may exist
within the FA hydroxylase superfamily it is hard to know where to look since many different chemistries
have been described. It would be inefficient to explore the space further using low through put
methods unless a new class of decarbonylases is identified.
If higher efficiency enzymes are not identified from either cyanobacterial or CERi
decarbonylases the production of short-chain alkanes would be impractical at a production scale. With
abundant petroleum available in the near term the value of alkanes is far too low to be produced by
pathways producing such low yields. If one envisions a long term future with limited petroleum, further
development of alkane producing strains for liquid fuels remains an interesting pursuit.
106
Appendices
Appendix 1: Evaluating fuel targets and gasoline composition
Text A1-1: Energy density calculations for potential alcohol fuels
Energy densities were calculated from available density 16,157'158 and heat of combustion datal
9
for
normal and iso isomers of primary alcohols from C2-C8. Heat of combustion per mole of alcohol was
converted to a volume basis using the molecular weight and density of the alcohol.
Text A1-2: Sample calculations of potential fuel compound pricing
Glucose prices were estimated from the online international trading site Alibaba.com
47
. All fuel product
prices were taken from ICIS Indicative Chemical Prices which are based on 2006-2008 Chemical Market
Reporter values48 .
Assuming a glucose price of $500 per metric ton:
$500
1 metric ton
1 metric ton 1x10
6
grams
180.16 grams
$0.09
1 mole
1 mole glucose
For a price of $400 per metric ton the price changes to $0.07 per mole. For ethanol a range of pricing
was used because fuel ethanol sold for less than 200-proof, industrial contract ethanol. Assuming $2.10
per gallon of fuel ethanol:
$2.10
0.264 gallons
1 liter
$0.03
1 gallon
1 liter
17.13 moles
1 mole of fuel ethanol
The energy normalized price per mole of fuel target was calculated as follows. The average United
States retail price of gasoline ($3.36/gallon, February 2014160) was multiplied by 0.70 (crude oil
contribution to total price160) to get a price of $2.35/gallon ($0.62/liter). The price on an energy basis
was found as follows:
1 L gasoline
$0.62
L gasoline
33 MiJ
$0.019
Mi
107
Using this price per unit energy, the price per mol fuel compound was found using the heat of
combustion of the fuel compound as follows (example hexane):
108
4.163 MJ
$0.019
$0.08
mol hexane
MJ
mol hexane
Table A1-1: Composition of typical gasoline. Major hydrocarbons of a representative sample of regular
gasoline are shown. The weight percent of main components from each hydrocarbon group and the total weight
percent of that group are given. Different weight percent thresholds are used for each group: parriffins >0.5%,
olefins >0.1%, napthenes >0.3%, and aromatics >0.5%. Note that pentane and 2-methyl-butane are the two most
abundant alkanes. Table is adapted from Johansen et al. 40
Group
Hydrocarbon
Compound
Weight %
Paraffins
Group
Hydrocarbon
Compound
Weight %
Naphthenes
n-Butane
2-Methylbutane
3.56
10.65
Cyclopentane
0.51
Methylcyclopentane
1.96
n-Pentane
8.32
Cyclohexane
1.14
2,3-Dimethylbutane
0.86
1(cis),3-Dimethylcyclopentane
2-Methylpentane
4.54
1 (trans ),3-Dimethylcyclopentane
3-Methylpentane
2.62
1 (trans ),2-Dimethylcyclopentane
0.36
0.33
0.47
n-Hexane
4.52
Methylcyclohexane
1.66
2-Methylhexane
1.45
1,2,3-Trimethylcyclopentane
0.45
2,3-Dimethylpentane
0.7
0.55
0.57
2.01
0.69
0.61
1.01
0.58
Group Total
8.74
3-Methylhexane
2,2,4-Trimethylpentane
n-Heptane
2-Methylheptane
3-Methylheptane
n-Octane
n-Nonane
Group Total
49.54
Olefins
Hexene- 1
0.14
0.19
0.14
0.28
0.13
Group Total
1.72
2-Methylbutene-1
Pentene-2, trans
Pen tene-2, cis
2-Methylbutene-2
Aromatics
Benzene
Toluene
1.03
20.04
Ethylbenzene
1.26
p -Xylene
2.79
o -Xylene
1.14
1-Methyl-3-ethyl benzene
1 -Methyl-2-ethylbenzene
0.95
0.56
0.5
1,2,4-Trimethyl benzene
1.58
1,3,5-Trimethyl benzene
Group Total
33.15
109
Appendix 2: CoA-dependent 4MP pathway
Text A2-1: Pathway yield calculations
Yield calculations were computed as outlined in Dugar and Stephanopoulos4 9 . An explanation for the
specific pathways cited in the current work follows. The maximum molar yield on an energy basis, YEis
calculated as the ratio of the degree of reductance of glucose, 24, over the degree of reductance of the
product. The degree of reductance of 4MP is 36 giving a
yE
of 0.67 mol 4MP/mol glucose. The
stoichiometric yield of each pathway, Y, is calculated by dividing the moles of product generated per
mole of substrate. The maximum pathway energy efficiency is calculated as:
degree of reductance prod *nprod 1
degree of reductance . -n
10
YE
where n is the number of moles of product or substrate used in a given pathway.
In order to account for required NADPH generation and regeneration of excess NADH reducing
equivalents, the stoichiometric pathway balance, v1, can be coupled to the system of equations below:
v1: - CH 20 - a NADPH + b Product + c ATP + d NADH + e CO2
V2:
v3 : - CH 2 0
-
- CH 20 + 2 NADPH + CO2
1/3ATP - 1/3 NADH + CH8/ 3 0 (glycerol)
Here a, b, c, d, and e are the pathway stoichiometric coefficients normalized by the number of
glucose carbons consumed in the pathway. By matching the equation rates, different pathway yields
can be calculated for each proposed pathway. If both NADPH cofactor generation and NADH
regeneration through a glycerol sink (used as a potential regeneration route for a commercial
production yeast strain) are considered than an adjusted pathway yield, YclP,G, can be calculated as
described in Dugar and Stephanopoulos.
YPG
110
_
.
a1++ d-c
2 L4. 8 2 ] if(d-c)>O
+3d
else (d-c)-O
The adjusted molar pathway yield, Yc1PG can be divided by the maximum molar yield, YE, to find the
adjusted energy efficiency.
=
yP,G
C_
.100
The pathway coefficients for the aKAE pathway to 4MP are: a=0.167, b=0.083, c=0.333, d=0.667, and
e=0.5. For the presented CoA-dependent 4MP pathway the coefficients are: a=0.333, b=0.111, c=0.111,
d=0.333, and e=0.333.
Text A2-2: Codon optimized gene sequences
S. cerevisiae ADH6:
ATGAGCTACCCGGAAAAGTTCGAGGGTATTGCTATTCAGTCCCATGAGGACTGGAAGAACCCGAAGAAAACCAAG
TATGATCCGAAGCCGTTCTACGACCACGACATCGACATCAAAATCGAAGCGTGCGGCGTGTGCGGTAGCGATATC
CACTGCGCAGCGGGCCACTGGGGTAACATGAAAATGCCACTGGTGGTGGGCCATGAGATTGTCGGTAAGGTGGT
GAAACTGGGCCCGAAGAGCAACAGCGGCCTGAAAGTTGGTCAGCGTGTGGGTGTTGGTGCGCAAGTCTTTAGCT
GTTTGGAATGTGATCGCTGTAAGAACGATAATGAACCGTATTGCACGAAGTTTGTTACCACCTATTCGCAACCTTAT
GAGGATGGTTACGTCAGCCAAGGCGGTTATGCAAACTATGTGCGCGTTCACGAGCACTTCGTTGTGCCGATTCCG
GAGAATATCCCGAGCCATCTGGCAGCACCGCTGCTGTGTGGCGGTCTGACGGTCTACTCCCCGCTGGTCCGCAAT
GGTTGCGGTCCGGGCAAGAAAGTGGGCATTGTTGGTCTGGGTGGCATCGGTTCTATGGGCACGTTGATTTCGAAG
GCCATGGGTGCGGAGACTTACGTCATCTCTCGTTCTAGCCGCAAACGTGAGGACGCGATGAAGATGGGTGCCGAT
CACTACATTGCGACCCTGGAAGAGGGTGACTGGGGCGAGAAATACTTTGACACCTTCGATCTGATTGTTGTGTGC
GCGAGCAGCCTGACGGATATTGACTTTAACATTATGCCAAAAGCCATGAAAGTCGGTGGCCGCATCGTTTCCATTA
GCATCCCTGAACAGCACGAGATGCTGAGCCTGAAGCCGTACGGTCTGAAGGCAGTTAGCATTAGCTACAGCGCTC
TGGGCTCCATCAAAGAACTGAATCAGCTGCTGAAATTGGTGAGCGAAAAAGACATCAAGATCTGGGTTGAAACCC
TGCCGGTGGGTGAGGCAGGTGTCCACGAGGCCTTTGAGCGTATGGAAAAAGGCGATGTGCGTTATCGTTTCACCC
TGGTTGGTTACGATAAAGAATTCAGCGAC
111
N. iowensis car:
ATGGCTGTGGACTCGCCGGATGAACGCCTGCAACGCCGTATCGCCCAACTGTTTGCCGAAGATGAACAAGTGAAA
GCTGCCCGCCCGCTGGAAGCAGTTAGCGCGGCCGTCTCTGCACCGGGTATGCGTCTGGCTCAGATCGCAGCTACG
GTGATGGCTGGTTATGCGGATCGTCCGGCGGCGGGCCAGCGTGCTTTCGAACTGAATACCGATGACGCAACCGGC
CGTACCAGCCTGCGTCTGCTGCCGCGTTTTGAAACCATTACGTACCGCGAACTGTGGCAGCGTGTCGGCGAAGTG
GCAGCTGCGTGGCATCACGACCCGGAAAACCCGCTGCGTGCGGGTGATTTTGTGGCCCTGCTGGGCTTCACCAGC
ATTGATTATGCAACGCTGGATCTGGCTGACATCCATCTGGGTGCGGTTACCGTGCCGCTGCAAGCGAGCGCGGCG
GTGTCCCAACTGATTGCAATCCTGACCGAAACGAGTCCGCGCCTGCTGGCGTCCACCCCGGAACATCTGGATGCTG
CGGTGGAATGCCTGCTGGCAGGCACCACGCCGGAACGTCTGGTGGTTTTCGATTATCACCCGGAAGATGACGATC
AGCGCGCCGCATTTGAAAGTGCGCGTCGCCGTCTGGCAGATGCAGGTTCCCTGGTGATCGTTGAAACCCTGGACG
CGGTGCGTGCGCGTGGCCGTGATCTGCCGGCTGCGCCGCTGTTTGTCCCGGATACCGACGATGACCCGCTGGCGC
TGCTGATTTATACGTCAGGTTCGACCGGCACGCCGAAAGGTGCCATGTACACCAATCGTCTGGCCGCAACGATGT
GGCAGGGCAACTCAATGCTGCAAGGCAACAGCCAACGCGTTGGCATTAACCTGAATTATATGCCGATGAGTCATA
TTGCGGGTCGTATCTCCCTGTTCGGCGTGCTGGCGCGTGGCGGCACCGCATACTTTGCTGCGAAATCAGACATGA
GCACCCTGTTTGAAGATATTGGCCTGGTTCGCCCGACCGAAATCTTTTTCGTTCCGCGTGTCTGTGACATGGTGTTT
CAGCGCTATCAAAGCGAACTGGATCGCCGTTCTGTCGCTGGTGCGGATCTGGACACCCTGGACCGCGAAGTGAAA
GCGGATCTGCGTCAGAATTACCTGGGCGGTCGCTTCCTGGTTGCAGTCGTGGGCTCGGCTCCGCTGGCCGCAGAA
ATGAAAACGTTTATGGAAAGCGTGCTGGACCTGCCGCTGCATGATGGTTATGGCAGTACCGAAGCCGGCGCATCC
GTTCTGCTGGATAACCAGATCCAACGTCCGCCGGTCCTGGACTATAAACTGGTCGATGTGCCGGAACTGGGTTACT
TTCGCACGGATCGTCCGCACCCGCGTGGCGAACTGCTGCTGAAAGCAGAAACCACGATTCCGGGTTATTACAAAC
GCCCGGAAGTTACGGCGGAAATCTTTGATGAAGACGGCTTCTATAAAACCGGCGATATTGTGGCCGAACTGGAAC
ATGACCGCCTGGTTTACGTGGATCGTCGTAACAATGTTCTGAAACTGTCCCAGGGCGAATTTGTGACCGTTGCGCA
CCTGGAAGCTGTGTTCGCGAGCAGCCCGCTGATCCGTCAAATTTTTATCTATGGTAGTTCCGAACGCAGTTACCTG
CTGGCCGTCATTGTGCCGACCGATGACGCACTGCGTGGCCGCGATACCGCTACGCTGAAAAGCGCTCTGGCGGAA
112
TCTATTCAGCGTATCGCCAAAGACGCAAATCTGCAACCGTATGAAATTCCGCGCGATTTTCTGATCGAAACCGAAC
CGTTCACGATTGCCAATGGCCTGCTGAGCGGTATCGCAAAACTGCTGCGCCCGAACCTGAAAGAACGTTATGGTG
CGCAGCTGGAACAAATGTACACCGACCTGGCTACGGGCCAGGCAGATGAACTGCTGGCCCTGCGCCGTGAAGCT
GCGGATCTGCCGGTGCTGGAAACCGTTAGCCGTGCCGCAAAAGCGATGCTGGGTGTGGCAAGCGCGGATATGCG
TCCGGACGCACATTTTACCGATCTGGGCGGTGACAGCCTGTCTGCACTGAGTTTTTCCAACCTGCTGCACGAAATCT
TCGGTGTTGAAGTCCCGGTGGGTGTTGTCGTGTCTCCGGCAAACGAACTGCGTGATCTGGCGAATTATATTGAAG
CCGAACGCAACAGTGGCGCAAAACGTCCGACCTTCACGTCAGTGCATGGCGGTGGCTCGGAAATTCGTGCTGCGG
ATCTGACCCTGGACAAATTTATCGATGCACGCACGCTGGCCGCAGCTGATTCTATTCCGCACGCCCCGGTGCCGGC
ACAGACCGTTCTGCTGACGGGTGCGAATGGCTATCTGGGTCGTTTCCTGTGCCTGGAATGGCTGGAACGCCTGGA
TAAAACCGGCGGCACCCTGATTTGTGTTGTCCGTGGTAGCGACGCGGCGGCGGCACGTAAACGTCTGGATTCAGC
CTTTGATAGCGGCGATCCGGGCCTGCTGGAACATTATCAGCAACTGGCAGCACGTACCCTGGAAGTGCTGGCAGG
CGATATTGGTGACCCGAACCTGGGCCTGGATGACGCGACCTGGCAGCGTCTGGCAGAAACGGTCGATCTGATTGT
GCATCCGGCAGCTCTGGTGAATCACGTTCTGCCGTACACCCAGCTGTTTGGCCCGAACGTGGTTGGCACCGCGGA
AATTGTGCGCCTGGCTATCACCGCGCGTCGTAAACCAGTGACCTATCTGTCTACGGTTGGCGTCGCAGATCAGGTT
GACCCGGCTGAATACCAAGAAGATAGCGATGTGCGTGAAATGTCTGCGGTGCGTGTCGTGCGCGAAAGCTATGC
CAACGGTTACGGCAATTCTAAATGGGCTGGTGAAGTGCTGCTGCGCGAAGCGCATGATCTGTGCGGTCTGCCGGT
GGCAGTTTTTCGTTCAGATATGATTCTGGCACACTCGCGCTATGCTGGTCAGCTGAATGTCCAAGATGTGTTCACCC
GTCTGATTCTGTCACTGGTTGCTACGGGCATCGCGCCGTATTCGTTTTACCGCACCGATGCAGACGGTAACCGTCA
GCGCGCCCATTACGATGGTCTGCCGGCAGATTTCACCGCGGCGGCGATTACGGCGCTGGGTATCCAGGCCACCGA
AGGCTTTCGCACGTATGATGTGCTGAATCCGTATGATGACGGTATTAGTCTGGACGAATTTGTTGATTGGCTGGTC
GAATCCGGCCATCCGATTCAGCGTATCACGGATTATTCAGACTGGTTTCACCGCTTCGAAACCGCCATCCGTGCACT
GCCGGAAAAACAGCGTCAAGCCAGCGTGCTGCCGCTGCTGGATGCATACCGTAACCCGTGTCCGGCCGTTCGCGG
TGCAATTCTGCCGGCTAAAGAATTTCAGGCTGCGGTCCAAACCGCGAAAATTGGCCCGGAACAGGATATTCCGCA
CCTGAGTGCCCCGCTGATTGATAAATACGTGTCTGACCTGGAACTGCTGCAACTGCTGTAA
113
B. subtilis sfp:
ATGAAAATCTATGGCATTTACATGGATCGTCCGCTGAGTCAGGAAGAAAACGAACGCTTTATGACCTTCATCAGCC
CGGAAAAACGTGAAAAATGCCGTCGCTTTTATCATAAAGAAGATGCACACCGCACGCTGCTGGGCGATGTGCTGG
TTCGTAGCGTGATCTCTCGCCAGTATCAGCTGGATAAATCTGATATTCGTTTCAGTACCCAGGAATACGGTAAACC
GTGTATTCCGGATCTGCCGGATGCACATTTTAATATCAGCCACTCTGGCCGCTGGGTTATTGGTGCGTTCGATTCTC
AGCCGATTGGTATCGATATTGAAAAAACGAAACCGATCAGTCTGGAAATTGCCAAACGTTTCTTTAGCAAAACCGA
ATATTCTGATCTGCTGGCAAAAGATAAAGATGAACAGACGGATTACTTTTACCATCTGTGGAGTATGAAAGAATCT
TTTATCAAACAGGAAGGCAAAGGTCTGAGCCTGCCGCTGGATAGTTTTAGCGTGCGCCTGCATCAGGATGGCCAG
GTTTCTATCGAACTGCCGGATTCTCACAGTCCGTGCTATATTAAAACCTACGAAGTTGATCCGGGCTATAAAATGG
CCGTTTGTGCGGCCCACCCGGATTTCCCGGAAGATATTACGATGGTGAGCTACGAAGAACTGCTGTAA
L. sp. Strain S749 Isadh:
ATGGCCCAGTATGATGTTGCAGATCGTAGCGCAATTGTTACCGGTGGTGGTAGCGGTATTGGTCGTGCAGTTGCA
CTGACCCTGGCAGCAAGCGGTGCAGCAGTTCTGGTTACCGATCTGAATGAAGAACATGCACAGGCAGTTGTTGCA
GAAATTGAAGCAGCCGGTGGTAAAGCAGCAGCACTGGCTGGTGATGTGACCGATCCGGCATTTGGTGAAGCAAG
CGTTGCCGGTGCAAATGCACTGGCACCGCTGAAAATTGCAGTTAATAATGCAGGTATTGGTGGTGAAGCCGCAAC
CGTTGGTGATTATTCACTGGATAGCTGGCGTACCGTTATTGAAGTTAATCTGAATGCCGTGTTTTATGGTATGCAG
CCGCAGCTGAAAGCAATGGCAGCAAATGGTGGTGGTGCCATTGTTAATATGGCAAGCATTCTGGGTAGCGTTGGT
TTTGCAAATAGCAGCGCCTATGTTACCGCAAAACATGCACTGCTGGGCCTGACACAGAATGCAGCCCTGGAATAT
GCAGCAGATAAAGTTCGTGTTGTTGCCGTTGGTCCGGGTTTTATTCGTACACCGCTGGTTGAAGCAAATCTGAGCG
CAGATGCCCTGGCATTTCTGGAAGGTAAACATGCCCTGGGTCGTCTGGGTGAACCGGAAGAAGTTGCAAGCCTGG
TTGCCTTTCTGGCAAGTGATGCAGCAAGCTTTATTACCGGTAGCTATCATCTGGTTGATGGTGGTTATACCGCACA
GTAA
114
Table A2-1: Strains Used in Module 3 Screening. Strains expressing the 24 combinations of phaJ and ter
genes are shown in the top section with names indicating Module 3 screen ("M3Sc") followed by abbreviations for
the organisms from which the ter and phaJ homologs are derived. "M3Sc-Ca" was used to confirm that the C.
acetobutylicum hbd and crt genes could not be used to produce branched products.
Host strain
Plasmid 1
Plasmid 2
Strain name
pCDF-phaJ4bc,-phaBcn
M3Sc-TdCn
pCDF-phaJ4p,-phaBcn
M3Sc-TdPa4
pCDF-phaJ4p,-phaBc,
M3Sc-TdPs
pCDF-phaJ1p8 -phaBcn
M3Sc-TdPal
pCDF-phaJ4bc,-phaBc,
M3Sc-VpCn
pCDF-phaJ4p,-phaBc,
M3Sc-VpPa4
pCDF-phaJ4p,-phaBc,
M3Sc-VpPs
pCDF-phaJ1p-phaBc,
M3Sc-VpPal
pCDF-phaJ4bc,-phaBc,
M3Sc-SoCn
pCDF-phaJ4pa-phaBc,
M3Sc-SoPa4
pCDF-phaJ4p,-phaBc,
M3Sc-SoPs
pCDF-phaJ1p,-phaBc,
M3Sc-SoPal
pCDF-phaJ4bcn-phaBc,
M3Sc-EgCn
pCDF-phaJ4p,-phaBc,
M3Sc-EgPa4
pCDF-phaJ4pephaBc,
M3Sc-EgPs
pCDF-phaJ1pe-phaBc,
M3Sc-EgPal
pCDF-phaJ4bc,-phaBc,
M3Sc-PaCn
pCDF-phaJ4p,-phaBc,
M3Sc-PaPa4
pCDF-phaJ4p,-phaBcn
M3Sc-PaPs
pCDF-phaJ1p8 -phaBc,
M3Sc-PaPal
pCDF-phaJ4bc,-phaBc,
M3Sc-PsCn
pCDF-phaJ4,,-phaBc,
M3Sc-PsPa4
pCDF-phaJ4p,-phaBc,
M3Sc-PsPs
pCDF-phaJ1p-phaBc,
M3Sc-PsPal
pET-terdr(bktBcn-pctm.)
pET-tervp-(bktBc-pct M)
pET-ters-(bktBc-pct,)
MG1655(DE3) endA recA
pET-terEg-(bktBcn-pctme)
pET-ter.-(bktBn-pct.)
pET-terp,-(bktBc-pcte)
pET-terTd-(bktBcn-pct M)
pCDF-hbdccrtca
M3Sc-Ca
115
Table A2-2: Plasmid construct descriptions. Cloning strategies for all plasmids used in the current work are
outlined.
Description
Name
Primers
Notes
pET-terrd-(bktB-pct)
T. denticola ter in MCS-1
(BamHI/Notl) and operon
Tseng et al. Controlled biosynthesis ofodd-chain fuels
containing R. eutropha bktB and and chemicals via engineered modular metabolic
M. elsdenii pct in MCS-2
pathways. PNAS 2012.
(Bglll/Xhol)
pET-tervp-(bktB-pct)
V. parahaemolyticus ter in MCSterVpUpl:
1 (NcoI/Notl)and operon
-TTGACGTACTAACAATC
containing R. eutropho bktB and ATATAGGATCCGATGATCATCAAACCTAGAATTCG
M ldeniipct in MCS-2
ATATAGCGGCCGCTTAGATTTGAATGAAGTCTGTTTCTAC
pET-ters,-(bktB-pct)
S. oneidensis ter in MCS-1
terSo Upl:
(Ncol/Notl) and operon
A
containing R. eutropha bktB and ATATAGGATCCGATGATATCAAACCCAAAATCG
-e~_D2
M. elsdenii pct in MCS-2
ATATAGCGGCCGCTTAAAGCTCAATCACATCGAACTC
(Bglll/Xhol)
bktB is cloned out of frame with MCS-2 RBS and
built in start codon
pct is preceeded by
ggttagccttgcgctcgagaggggagaattc RBS sequence
pET-terEg-(bktB-pct)
E. gracilis ter in MCS-1
(BamHI/Noti) and operon
E. gracilis ter subcloned from pET-terEg-adhE from
containing R. eutrophabktB and Tseng et al. into pET-bktB-pct backbone between
M. elsdenii pct in MCS-2
BamHI/Notl sites.
(Bglll/Xhol)
P. aeruginosa ter in MCS-1
pET-terp,-(bktB-pct)
116
New ter homologs cloned into existing pET-bktBpct backbone
terPa Up2: ATATAGGATCCGATCATCAAACCGCGCGT
pET-te
oNnotl)
bkp
oper bktB and terPa- Dn2: ATATACTTAAGTTACTGGATCAGGTTGGCGAT
eutropho
R.and
containing
final GCC alanine codon removed to generate primer
M. elsdenlipct in MCS-2
without secondary structure
(BgIll/Xhol)
pET-terpp-(bktB-pct)
P. putida ter in MCS-1
(Ncol/Notl) and operon
terPpUpl: ATATAGGATCCGATGGCCATCATTCATCCTA
containing R. eutropha bktB and
terPpDnl: ATATACTTAAGTTACAGCTCGACGCAGTC
M. elsdenii pct in MCS-2
(Bglll/Xhol)
pCDF-phaJ4bn.-phaB
R. eutropha pha14b in MCS-1
(Ncol/Afll) and R. eutropha
phaB (with Ndel site silently
mutated out) in MCS-2
(Ndel/Xhol)
phaJ4bReUpl:
ATATACCATGGGGATGAAGACCTACGAGAACATCG
phai4bReDni: ATATAGCGGCCGCCTTATG
pha4bRe was amplified from a pCDF-phaB-phaJ4b
construct created by Hsien-Chung Tseng where phaJ4b
had been cloned from gDNA into MCS-2 (Bglll/Xhol).
pCDF-phaJ4p-phaB
P. syringoe phoi4 in MCS-1
(Ncol/Noti) and R. eutropho
phaB (with Ndel site silently
mutated out) in MCS-2
phaJ4PsUpl:
ATATACCATGGGGATGCC11T-GTACCCGTCG
phai4PsDnl:
ATATAGCGGCCGCTTACACGAAACACAACGTCAAAG
pCDF-phaJ4Pa-phaB
P. aeruginosa phai4 in MCS-1
(Ncol/Notl) and R. eutropha
phaB (with Ndel site silently
mutated out) in MCS-2
(Ndel/Xhol)
phaJ4PaUpl:
ATATACCATGGGGATGCCATTCGTACCCGTAG
phaJ4PaGn:
ATATAGCGGCCGCTCAGACGAAGCAGAGGCT
pCDF-phaJ1p,-phaB
P. aeruginosa phai1 in MCS-1
(BamHl/Noti) and R. eutropha
phaB (with Ndel site silently
mutated out) in MCS-2
(BglIl/AvrIl)
Tseng et al. Controlled biosynthesis of odd-chain fuels
and chemicals via engineered modular metabolic
pathways. PNAS 2012.
phaB has Ndel site mutated by quick-change PCR
as used in Martin et al. It was subcloned from
the pET-bktB-phaB construct using Ndel/Xhol
restriction sites.
Both genes cloned with RBSs inserted after RE
Bite gns none
with
RBS's
site and not inframe with plasmid RBS's.
Name
Description
operon
Primers
Notes
pct RBS sequence:
containing R. eutropha
bktB and M. elsdeniipct in MCS bktB up4: AAAAAGGATCCGATGACGCGTGAAGTG
ggttagccttgcgctcgagaggggagaattc
C
CCCA
GGCCGCTTA T
pct_dn4: AA
1 (BamHI/Notl) and operon
containing B. eutropho phai4b phaJ4bLup4: AAAAACATATGGGGATGAAGACC
and phmB in MCS-2 (Bgll/Avrll) phaB-dn4: AAAAACCTAGGTCAGCCCATGTGC
phgt Rggsequenc
taagtataagaaggagatatacat
operons amplified from pET-(bktB-pct)-(phaJ4t phaB-ter) plasmid
pET-(bktB-ter)-(phaJ4b-phaB)
operon containing R. eutropha
bktB and T. denticola ter in MCS1 (BamHI/Notl) and operon
containing R. eutropha phaJ4b
and phaB in MCS-2 (Bgll/Avrl)
ter RBS sequence: aaaagaaggagatata
bktB-ter operon cloned into pET-(phai4bphaB) backbone
pET-(bktB-ter)-(phaB-phaJ4b)
phaB phai4b up5: AAAAACATATGACTCAGCGCATTGC
operon containing R. eutropha phaB phaJ4b dn5:
bktB and T. denticola ter in MCS- ATC
T AT
A C- T
G
1 (BamHl/Notl) and operon
phai4b phaB Lp5:
containg R. eutropha phaB and
GTATAAGaGGAGATATATAIGGGGATGAAGACCTACGAGA
pha14b in MCS-2 (Ndel/Avrll)
phaJ4bphaB dn5: AAAAACCTAGGTCAGGGAAAGCGCCG
This is the same plasmid as pET-(bktB-ter)phaJ4b-phaB) but with the operon order of
phaJ4b and phaB switched. phaB and phaB
operon containing E. coli ilvD
ilvD up4: AAAAAGGATCCGATGCCTAAGTACCGTTCC
and T. denticola ter in MCS-1
lvD dn4: CAATCATTATATCTCCTTCTTTTAACCCCCCAGTTTC
(BamHI/Notl) and an operon
terup4: AAGAAGGAGATATAATGATGTGAAACCGATG
containing B. subtilis alsS and E. ter dn4: AAAAGGCGTAAATAGGAAACGTT
coli ilvC in MCS-2 (Bglll/AatlI)
ter-dn4: AAAAAGCGGCCGCTCAAATACGGTCAAAGCGTTC
The RBS infront of ilvC contains the native 14
bp upstream of ilvC start codon
cacgaggaatcacc and the RBS in front of ter is
the 14 bp sequence aagaaggagatata
ilvD-ter operon cloned into pCDF-alsS-ilvC
pET-(bktB-pct)-(phaJ4b-phaB)
pCDF-(ilvD-ter)-(alsS-ilvC)
pCDF-(ibuA-ilvD)-(alsS-ilvC)
pCDF-(ibuA-ilvD)-(ilvC-alsS)
operon containing R. palustris
ibuA and E. coli ilvD in MCS-1
(BamHl/Notl) and an operon
containing B. subtilis alsS and E.
coli ilvC in MCS-2 (Bglll/Aatll)
bktB terSOE_up1: AAAAAGGATCCGATGACGCGTGAAGTG
bktB terSOE_dnl:
CAATCATTATATCTCCTTCTTCAGATACGCTCGAAG
ter-up4: AAGAAGGAGATATAATGATTGrGAAACCGATG
terdn4: AAAAAGCGGCCGCTCAAATACGGTCAAAGCGTTC
ibuA
ilvDSOE_upi: AAAAAGGATCCGATGAGCAACACCCAT
ibuAilvDSOE_dnl:
CATACTTTATTTCCTCCCAGTCAAGCTGCAGAAGAA
ilvD_up3: CTGGGAGGAAATAAAGTATG
ilvD_dn3: AAAAAGCGGCCGCTTAACCCCCCAGTT
operon containing R. palustris
ibuA and E. coli ilvD in MCS-1
ilvCup5: 1T1TTAGATCTATGGCTAACTACTTCAATAC
(BamHl/Notl) and an operon
containing E. coli ilvC and B.
subtilis alsS in MCS-2
ilvC_dn5: CACCCTCACTCCTTATTAACCCGCAACAG
alsSup5: TAAGGAGTGAGGGTGATGACAAAAGCAACAAA
alsS_dnS: 1TTTGACGTCCTAGAGAGCTTTCGTrTT
ievD RBS sequence: CTGGGAGGAAATAAAGT
ibuA-ilvD operon cloned into pCDF-(alsS-ilvC)
backbone
iIvC-alsS operon cloned into pCDF-(ibuA-ilvD)
backbone
(Bgll]/AatlI)
pCDLA-puuC-kivD
puuCupi: TTTTTCATATGAATTTTCATCATCTGG
L. lactis kivD in MCS-1
(BamHI/Noti) and E. colipuuC in puuC_dni: TTTTTCTCGAGTCAGGCCTCCAGG
kivD_upl: 1TT1TGGATCCGATGTATACAGTAGGAGATTACC
MCS-2 (Ndel/Xhol)
kivDOdnl: T1TIIGCGGCCGCTTATGATTTATTTTGTTCAG
pCOLA-feaB-kivD
feaB_upi: TTTTTCATATGACAGAGCCGCAT
L. lactiskivD in MCS-1
(BamHlI/Notl) and E. colifeaB in f aB-dnl: TTTl CTCGAGTTAATACCGTACACACACCG
TTTTGGATCCGATGTATACAGTAGGAGATTACC
kivD_upl:
MCS-2 (Ndel/Xhol)
kivD dnl: 1TITTGCGGCCGCTrATGATATTTTGTTCAG
pCOLA-Fjoh_2967-kivD
Fjoh_2967_upi: AAAAACATATGAGCACAACCGCACAAAG
L. loctis kivD in MCS-1
(BamHI/Noti) and F. johnsonaie Fjoh_ 2967_dnl:
Fjoh_2967 in MCS-2 (Ndel/Xhol) AAATTCTCGAGTTAAAAGAAACCTAATTTCTTTTTATCATAAGAAATC
Fjok_2967 was cloned in place offeaB using
a backbone generated from the existing
pCOLA-feaB-kivD construct
car and sfp
operon containing N.
pACYC-(car-sfp)-ADH6
lowensis
carand B. subtilis sfp in MCS-1
(BamHI/Aflll) andS. cerevisiae
ADH6 in MCS-2 (Ndel/Aatil)
operon containing N.
pACYC-(car-sfp)-Isadh
iowensis
carand B.subtilis sfp i MCS-1
(BamHl/Aflll) andLeifsonia sp.
ramnS749 sadh in MCS-2
pACYC-(car-sfp)
operon containing N. iowensis
car and B. subtilis sfp in MCS-1
(BamHI/Afill)
synthesized by GenScript for E.
coli codon optimization. Car subcloned from
pucS7. sfp was amplified to add RBS upstream
of start codon. cor was subcloned first
sfpaupl:
AAAAAAGCGGCCGCTAATAAAAGGAGATATACCATGAAAATCTATGG
between BamHl/Notl and sfp with RBS was
CATACAT
cloned after between NotI/AfIll. ADH6
sfpdnl: AAATC-AAGTTACAGCAGTTCTTCGTAGCT
d Asynthesized
and codon optimized for E. coli by
DNA 2.0. As subcloned the ADH6 open
reading frame contains a C-terminal S-tag.
sfpupl:
Isadh synthesized and codon optimized for E.
AAAAAAGCGGCCGCTAATAAAAGGAGATATACCATGAAAATCTATGG coli by Invitrogen Life Technologies as a
CAMTTACAT
GeneArt String. Cloned directly from linear
sfp_dni: AAATTTCTTAAGTTACAGCAGTTCTTCGTAGCT
DNA digest.
car and sfp operon without an alcohol
cehydrogenase in MCS-2
117
acetoin
a
OH
oCA
NAD(P)
NAD
GLYCOLYSIS
[+021
70
OH
Sisobutanol
0
OH
4-methyl-pentanol
0H
butanol
Fig. A2-1: Detailed schematic of modules of the full 4MP Pathway. Module 1 (orange) consists of B. subtilis
acetolactate synthase aISSBS, Escherichia coli acetohydroxy acid isomeroreductase ilvCEc and dihydroxy acid
dehydratase iHvDEc, Lactococcus lactis a-ketoisovalerate decarboxylase kivDL, and Escherichia coli aldehyde
dehydrogenasesfeaBEc or Flavobacteriumjohnsonaie aldehyde dehydrogenase Fjoh2967Fj. Module 2 is one of two
isobutyrate activators Megasphaera elsdenii propionyl-CoA transferase pctm, or Rhodopseudomonas palustris
isobutyryl-CoA ligase ibuARp. Module 3 contains the Cupriavidus necator thiolase bktBcn, 3-hydroxyacyl-CoA
dehydrogenase phaBcn, and enoyl-CoA dehydratase phaJ4bcn and Treponema denticola enoyl-CoA reductase terTd.
Finally Module 4 is composed of the Nocardia iowensis carboxylic acid reductase carNi and Saccharomyces
cerevisiae alcohol dehydrogenase ADH6sc or Leifsonia sp. Strain S749 alcohol dehydrogenase IsadhS.
118
Appendix 3: Modified lipid contentfrom engineered FAS
Text A3-1: Codon optimized FatB2ch gene sequence.
ATGGTGGCTGCAGCCGCGTCTTCAGCCTTTTTCCCAGTCCCGGCTCCTGGTGCAAGCCCAAAACCGGGTAAATTTG
GCAATTGGCCTAGCAGTCTGAGCCCTAGTTTTAAACCAAAATCTATTCCGAACGGTGGCTTCCAAGTTAAAGCCAA
TGATTCAGCGCATCCAAAAGCTAACGGTTCTGCAGTGTCATTGAAATCCGGCTCTCTGAACACACAAGAAGATACG
TCCTCTTCACCACCGCCTCGCACCTTTCTGCATCAGCTGCCGGATTGGTCACGTCTGTTAACAGCTATCACCACTGTC
TTCGTTAAATCCAAACGCCCGGATATGCACGATCGTAAATCTAAAAGACCTGATATGCTGGTTGATTCCTTTGGTTT
AGAATCTACGGTGCAAGATGGCTTAGTCTTTCGCCAGTCATTCAGCATCCGTTCTTATGAAATTGGTACAGATAGA
ACGGCAAGCATCGAAACACTGATGAACCATTTGCAAGAAACGAGTCTGAACCACTGTAAATCCACCGGCATCTTGC
TGGATGGTTTTGGCAGAACCTTGGAAATGTGCAAACGCGATCTGATTTGGGTTGTGATCAAAATGCAGATTAAAG
TCAATCGTTACCCGGCCTGGGGTGATACCGTTGAAATTAACACTAGATTCTCTCGCCTGGGCAAAATCGGTATGGG
CAGAGATTGGTTAATTAGCGATTGTAATACTGGTGAAATCTTGGTGCGCGCGACAAGTGCTTATGCAATGATGAA
CCAAAAAACTCGTAGATTATCCAAATTGCCATACGAAGTTCATCAGGAAATTGTCCCTCTGTTTGTTGATTCTCCAG
TGATCGAAGATTCAGATTTAAAGGTTCACAAGTTCAAGGTGAAGACGGGTGATTCTATTCAAAAAGGTTTAACCCC
AGGCTGGAATGATTTGGATGTCAACCAGCATGTTAGTAACGTGAAGTACATCGGTTGGATTCTGGAATCCATGCC
GACAGAAGTTTTAGAAACGCAGGAATTGTGTTCACTGGCTTTAGAATACCGCCGTGAATGCGGTCGTGATAGCGT
CTTGGAAAGTGTTACAGCTATGGACCCAAGCAAAGTGGGCGTCCGTAGTCAATATCAGCACTTATTGAGACTGGA
AGATGGTACTGCCATTGTGAATGGCGCGACTGAATGGAGACCTAAAAATGCCGGTGCGAACGGCGCTATCTCAAC
CGGTAAAACTAGCAATGGCAACAGTGTTTCCTAA
119
Table A3-1: Plasmid construction. Primers and plasmid cloning strategies are provided in the table
below for all plasmids used for development of the FAS pathway platform.
Plasmid name
Primer name
Primer sequence
Notes
pCDF-bukB,-ptbB,
ptb_up
ptb dn
Ndel/Aatl
buk_up
bukdn
AAAAACATATGAAGCTGAAAGATTTAATC
AAAAAGACGTCTCGTTCCTCCTAATGTG
AAAAAGGATCCGATGAAAGTGCTACATG
AAAAAGCGGCCGCACTCCTGTCTGTCTATTC
pET-fabH1Is
fabHlup
fabHldn
ATATAAGATCTCATGAAAGCTGGAATACTTGGT
ATATACTCGAGTTATCGGCCCCAGC
BgIll/XhoII
pET-fabH2B,
fabH2_up
fabH2_dn
ATATACATATGTCAAAAGCAAAAATTACAGCTATC
ATATAGACGTCTTACATCCCCCATTTAATAAGCAAT
Ndel/Aatll
pET-FatB2Ch
Ncol/Notl
Subcloned from pUC57
plasmid using Ndel/Pac
pET-His-FatB2Ch
fatB2hisup
fatB2hisdn
AAAAGGATCCGATGGTGGCTGCAGCC
AAAAGCGGCCGCTTAGGAAACACTGTTGCCATTG
BamHI/Notl
pET-FatB2m 1Ch
fatB2mlhis_up
fatB2mhisdn
AAAAGGATCCACTGCCGGATTGGTC
AAAAGCGGCCGCTTAGGAAACACTGTTGCC
BamH /Notl
pET-FatB2m2ch
fatB2m2hisup
AAAAGGATCCACTGGTTGATTCCTTTGG
Downstream primer same
as for FatB2mlCh
BamHI/Notl
Subcloned FatB2m2Ch into
pET-FatB2m2ch-fabH1B
pET-fabHes with BamHl/
Notl
SubCloned
FatB2m2Ch into
pET-fabH2B, with BamHI/
pET-FatB2m2Ch-fabH2B.
Not[
pACYC-accABCD.
accAup
accAdn
accBC_up
accBCdn
accD_up
accDdn
ATATACATATGAGTCTGAATTTCCTTGATTT
GAGTGGGTTCCGTACTTACGCGTAACCGTAGCTC
GTACGGAACCCACTCATGGATATTCGTAAGATTAAAAAACTG
TAGGGACCTTTCTGTCTTATTTTTCCTGAAGACCGAGT
GACAGAAAGGTCCCTAATGAGCTGGATTGAACGAA
ATATAGACGTCTCAGGCCTCAGGTTCC
Assembled using SOE
PCR, Ndel/AatlI
accABCDE, operon
pET-FatB2m2Ch-accABCDE,
suboloned into
pET-FatB2m2ch
Ndel/Aatll
pDR1 11 -FatB2m2Ch
120
DRB2m2_up
DRB2m2_dn
AAAAAAAGCTTAAAGGAGGAAATTGATATGGGCAGCAGCC
AAAAAGCTAGCTTAGGAAACACTGTTGCC
Hindill/Nhel
16-
-
fabH1 Bs' ptb/bukBs
fabH2Bs ptb/buk Bs
M
14-
ptb/bukBs
1210.
0-
86420-
-
'b
0
1\15
0
,\bt
,Nr
A%&
60
..
\rO
branched acids
Fig. A3-1: Branched long-chain fatty acids from modified E. coli FAS. Branched long-chain acids were
produced from the short-branched acid precursors 3-methyl-butyrate and 2-methyl-butyrate. The expected C15
and C17 iso- and anteiso-branched acids were produced in E. coli when either fabH1, orfabH2B, were expressed
with the short, branched acid activators ptb/buk,. Interestingly, iso-branched acids were also produced from 3methyl-butyrate when only ptb/bukBs was expressed indicating that native E. coli ketosynthases have the ability to
condense 3-methyl-butyryl-CoA with malonyl-ACP.
121
35-
FatB2m2-fabH 1
FatB2m2-fabH2
FatB2m2-fabH2,ptb/buk
T
302520-
T
I-0E 15-
105-
0 4C8
7MC8
C10
Species
Fig. A3-2: Free fatty acid production with no-activator controls. Strains expressing only the ketosynthase
homologs and FatB2 m2Ch thioesterase were created to test for the necessity of the ptb/bukBs activators. Titers for
Strain M123 expressing the full pathway are shown for comparison. Only straight-chain free acids are observed
without activator expression. Interestingly thefabH1B, ketosynthase appears to inhibit straight-chain production
even when branched acids are not produced. (FatB2m2-fabHl=Strain M2a3, FatB2m2-fabH2=Strain M2b3 shown
in Table 3-1)
122
Appendix 4: Alkane constructdescriptionsand quantificationmethods
Text A4-1: Codon optimized ADpm. The mutated codon for the A134F mutation is underlined.
ATGCCGACCCTGGAAATGCCGGTTGCAGCAGTTCTGGATAGCACCGTTGGTAGCAGCGAAGCACTGCCGGATTTT
ACCAGCGATCGTTATAAAGATGCATATAGCCGTATTAACGCCATTGTGATTGAAGGTGAACAAGAAGCACACGAT
AACTATATTGCAATTGGCACCCTGCTGCCGGATCATGTTGAAGAACTGAAACGTCTGGCAAAAATGGAAATGCGC
CATAAAAAAGGTTTTACCGCCTGTGGTAAAAATCTGGGTGTTGAAGCAGATATGGATTTTGCCCGTGAATTTTTTG
CACCGCTGCGTGATAATTTTCAGACCGCACTGGGTCAGGGTAAAACCCCGACCTGTCTGCTGATTCAGGCACTGCT
GATTGAAGCATTTGCAATTAGCGCATATCATACCTATATTCCGGTTAGCGATCCGTTTGCACGTAAAATTACCGAAG
GTGTTGTGAAAGATGAATACACCCATCTGAATTATGGTGAAGCATGGCTGAAAGCAAATCTGGAAAGCTGTCGTG
AGGAACTGCTGGAAGCCAATCGTGAAAATCTGCCGCTGATTCGTCGTATGCTGGATCAGGTTGCCGGTGATGCAG
CCGTGCTGCAGATGGATAAAGAAGATCTGATCGAAGATTTCCTGATCGCCTATCAAGAAAGCCTGACCGAAATTG
GTTTTAACACCCGTGAAATTACCCGTATGGCAGCAGCAGCACTGGTTAGCTAA
123
Text A4-2: Alkane quantification. Gas phase alkane titers were calculated as shown below. The gas
standard was used to find the gas phase concentration in ppm. As described in methods, this
concentration corresponded to a 2-fold dilution of the original culture head space. Multiplying by headspace volume (found by measuring the mass of water used to fill the vial and subtracting the 2 ml
culture volume) gives the total alkane in the head-space. Dividing by culture volume gives the reported
titer.
I-10-6 mol of product 1 mol of gas
1 mol of gas
MWproduct
25.45
-6
25.45 L
1 L of gas
1000 ml of gas
product
g product
1000 mg of product
1 mol product
1 g of product
MWproduct. 10-6
mg of product
25.45
ml of gas -1I ppm of product
9 ml of gas
ml of culture
mg of product
ml of gas -1 ppm of product
Mffproduct
2 (dilution factor). 1000 ml of culture
12L culture
mg of product
*1-3
2.83
FID area of product
_W
L of culture - 1 ppm of product
2product mg of
2.83
L of culture -
ppm of produt stadr
s
y FID area of product standard
ppm of product
product titer mg of product
(L of culture)
Text A4-3: Heptane standard concentration calculation. The exact volume of 1 liter glass bottle with
septum cap was found by weighing with water. The intended gas phase concentration was then
calculated as follows:
vol. heptane (pl) -
1000 p
pheptane
1.1273
L -latm
1. 127L
-
mg
mmolheptane
MWheptane
M
g
iImmol = mmol of gas in bottle
0.0821 L -atm .277.15 K I mol
mol- K
mmolheptane .106
mmolbottle
124
_
_
= conc. of heptane
(ppm)
Table A4-1: Primers and plasmids. Plasmid descriptions and primer sequences for new plasmid constructs and
strain engineering are shown.
Plasmid name
(KO locus)
Primer name
Primer sequence
pACYC-(carN-sfpB,)-PMT1231n,
Notes
Codon optimized PMT1231p,,
was digested and cloned into
pACYC-(carNi-sfpBs) using
Ndel/Avrll
pACYC-(carNrsfpBs)-PMT1231p,.-mut
mutADup
mutADdn
TTTGCAATTAGCTTTTATCATACCTATATTCCGG
TAGGTATGATAAAAGCTAATTGCAAATGCTTC
PMT1231p, was subcloned into
pCDFDuet-1 and the plasmid was
PCR amplified with the mutAD
primer set. Once sequence the
mutant version was subcloned back
into pACYC-(carNr-sfps)
pET-th/c,-terTd
terTd(thl)_up
terTd(thl)_dn
AAAAAACATATGATTGTGAAACCGATGG
AAAAAACCTAGGTCAAATACGGTCAAAGCG
The terTd gene was amplified from
pET-(bktBcn-terTd)-(phaBCn-phaJ4bCn)
and cloned with Ndel/Avrll
The thica gene was subcloned from
an existing pRSF-th/ca plasmid with
Ncol/EcoRI
(yqhC-dkgA)
(yqhC-dkgA)::kanupA
TTTTCCCCGTTCCCGGTTGCTGTACCGGGAA
CGTATTTAGTGTAGGCTGGAGCTGCTTC
(yqhC-dkgA)::kan dnA
GGTAGCGGAACATTACCGCCACCGGGAGAAT
TTGCATGTTATCCGTCGACCTGCAGTT
(yqhC-dkgA)::kanupB
GTTAACCGCTGGCTTGTTAGGCACGCTGTTTG
TGGTGATTAAAAAAAAATACTGTAACGCCTGAC
GATTTTCCCCGTTCCCGG
The kanamycin marker with flanking
FRT sites was amplified from pKD13.
A series of three PCRs were
completed with the three primer sets
to extend homology for the yqhC-dkgA
locus.
(yqhC-dkgA)::kan-dnB
GAACAAGGAAGAGTAACAACGGGCGGGACGC
GAGGGGAATAAATGATTTCTGAAAAGTCCGGT
AGCGGAACATTACCGCC
(yqhC-dkgA)::kanupC
GTTAAACGCCATGAAGATCAGGTAATGACGTT
CCTGATGATCCTGCCAATTGCCTTGTTAACCG
CTGGCTTGTTAGGC
(yqhC-dkgA)::kandnC
GTTCTTTTAAGAGCTTCCGGCTCTGCATGATGA
TGTCCTTATATTTGGCATTCCTGAACAAGGAAG
AGTAACAACGGGC
125
Fig. A4-1: Heptane GC standard curve. Known volumes of heptane were added to a 1.127 L septum capped
bottle at 4*C and the bottle was then warmed to room temperature. After all heptane evaporated (<1 min) 8 ml of
gas was sampled using a gas-tight syringe and injected in the GC. The resulting standard curve is shown below.
Linear regression was completed using an 0-intercept which returned an adjusted R2 of 0.993.
1000 -
800-
600-
E
400-
200-
0
0
100000
200000
300000
400000
FID area
126
500000
600000
Appendix 5: Identifyingalternative enzymes for platform pathways
Text A5-1: Hidden Markov Model (HMM) used by adh-short (PF00106) Pfam family.
The HHM used to identify NADH/NADPH interacting residues within the PhaBco-like reductase network is
available at:
http://pfam.sanger.ac.uk/family/PF00106#tabview=tab6
Individual network sequences were aligned to the HMM and residues matching HMM positions 10, 31,
32, and 33 were annotated on the network.
127
Table AS-1: Protein identifiers for PhaBcn node reductases. Identifiers for both the UniProt and GenBank
databases are shown for each of the unique protein sequences contained in the PhaBcn representative node in the
90% identity reductase representative network. Some sequences correspond to multiple GenBank entries.
128
No.
Uniprot ID
1
B5SKA5
2
3
4
5
6
7
8
9
10
A3RPJO
F6GOU6
Q472G4
H1RXR3
G8BU4
Q1LNN5
L2EFZ5
G2ZMK4
D8NTD4
11
12
13
14
15
16
G3A7D8
C6BIU6
E2T3K1
U3GEE8
D8VEN5
B3R4S7
17
18
19
R7XMG3
D3UAK7
P14697
20
21
22
23
24
25
B2UFU8
ROEFX5
GOET17
S9SOH6
M4UG94
Q8XYX3
26
D8NIRO
27
H5W7Z7
GenBank
ID
CAQ61346.1
CU914168
AAKL01000002
CP002819
CPOOO090
AHJE01000001
HE610111
CP000352
ANKPO1000118
FR854064
FP885906
CBJ51088.1
FR854089
CP001644
ACUFO1000056
ACTT02000011
GQ922053
CU633749
CAQ69309.1
AQPZO1000012
FJ897462
AM260479
J04987
CP001068
APMQ01000001
CP002877
ASZVO1000018
CP004012
CAD15335.1
AL646052
FP885897
CBJ42932.1
CAGTO1000027
Table A5-2: UniProt identifiers for potential NADH-dependent PhaB-like reductases. UniProt IDs are
given for the nodes with sequences which have residues characteristic of NADH-dependent reductases at key
sequence positions.
Node
Uniprot ID
1
A1WZ57
2
A2WJ38
3
A4BPE5
4
B1F117, A9ASN3, J5B159,
B1YWN8, F0G632,U1ZAI2,
TOEH35,B4EIA8, U2HCR3,
QOB900, B1T8J8
5
B2TG44
6
E1T8M3
7
G7UUJ3
8
13BRY9
9
14VMP8
10
16AA15, NOAiQ9, A5TNV6,
A4LFQ3, B7CQS6, A9K318,
K7Q299, I1WV18, A3P8H1,
C5ZRN9, A8KEJ1, 12KMK7,
12KR47, Q63J00, 12KKU6,
Q2T838, A5J9N2, C4AVC6,
A5XVL7, A3NN15, A3MAW3,
Q3JJT1, B2HB86, 12LX31,
C41999, S5NPV8, A1UY81,
C6U913, A8ENCO, 12M4C7,
M7E761, B1H936, Q62E80,
11
QOA5R1
12
U1JNIO
13
U1LLL6
129
References
1.
Bothast RJ, Schlicher MA. Biotechnological processes for conversion of corn into ethanol.
Applied Microbiology and Biotechnology 67, 19-25 (2005).
2.
Jones DT, Woods DR. Acetone-butanol fermentation revisited. Microbiological Reviews 50, 484524 (1986).
3.
U.S. House. 110th Congress, 1st Session. H.R. 6, Energy Independence and Security Act of 2007.
Washington: Government Printing Office, 2007.
4.
Solomon BD, Barnes JR, Halvorsen KE. Grain and cellulosic ethanol: History, economics, and
energy policy. Biomass and Bioenergy 31, 416-425 (2007).
5.
Mardis ER. Next-Generation DNA Sequencing Methods. Annual Review of Genomics and Human
Genetics 9, 387-402 (2008).
6.
Ellis T, Adie T, Baldwin GS. DNA assembly for synthetic biology: from parts to pathways and
beyond. Integrative Biology 3, 109-118 (2011).
7.
Pagani 1,et al. The Genomes OnLine Database (GOLD) v.4: status of genomic and metagenomic
projects and their associated metadata. Nucleic Acids Research 40, D571-D579 (2012).
8.
Prather KLJ, Martin CH. De novo biosynthetic pathways: rational design of microbial chemical
factories. Current Opinion in Biotechnology 19, 468-474 (2008).
9.
Alper H, Moxley J, Nevoigt E, Fink GR, Stephanopoulos G. Engineering Yeast Transcription
Machinery for Improved Ethanol Tolerance and Production. Science 314, 1565-1568 (2006).
10.
Elbing K, et al. Role of Hexose Transport in Control of Glycolytic Flux in Saccharomyces
cerevisiae. Applied and Environmental Microbiology 70, 5323-5330 (2004).
11.
Bai FW, Anderson WA, Moo-Young M. Ethanol fermentation technologies from sugar and starch
feedstocks. Biotechnology Advances 26, 89-105 (2008).
12.
Scalcinati G, Otero JM, Van Vleet JRH, Jeffries TW, Olsson L, Nielsen J. Evolutionary engineering
of Saccharomyces cerevisiae for efficient aerobic xylose consumption. FEMS Yeast Research 12,
582-597 (2012).
13.
Ragauskas AJ, et al. The Path Forward for Biofuels and Biomaterials. Science 311, 484-489
(2006).
14.
Atsumi S, et al. Metabolic engineering of Escherichia coli for 1-butanol production. Metabolic
Engineering 10, 305-311 (2008).
15.
Bond-Watts BB, Bellerose RJ, Chang MCY. Enzyme mechanism as a kinetic control element for
designing synthetic biofuel pathways. Nat Chem Biol 7, 222-227 (2011).
130
16.
Steen E, et al. Metabolic engineering of Saccharomyces cerevisiae for the production of nbutanol. Microbial Cell Factories 7, 36 (2008).
17.
Nielsen DR, Leonard E, Yoon S-H, Tseng H-C, Yuan C, Prather KLJ. Engineering alternative butanol
production platforms in heterologous bacteria. Metabolic Engineering 11, 262-273 (2009).
18.
Tseng H-C, Prather KLJ. Controlled biosynthesis of odd-chain fuels and chemicals via engineered
modular metabolic pathways. Proceedings of the National Academy of Sciences 109, 1792517930 (2012).
19.
Dekishima Y, Lan El, Shen CR, Cho KM, Liao JC. Extending Carbon Chain Length of 1-Butanol
Pathway for 1-Hexanol Synthesis from Glucose by Engineered Escherichia coli. Journal of the
American Chemical Society 133, 11399-11401 (2013).
20.
Dellomonaco C, Clomburg JM, Miller EN, Gonzalez R. Engineered reversal of the -oxidation
cycle for the synthesis of fuels and chemicals. Nature 476, 355-359 (2011).
21.
Withers ST, Gottlieb SS, Lieu B, Newman JD, Keasling JD. Identification of Isopentenol
Biosynthetic Genes from Bacillus subtilis by a Screening Method Based on Isoprenoid Precursor
Toxicity. Applied and Environmental Microbiology 73, 6277-6283 (2007).
22.
Atsumi S, Hanai T, Liao JC. Non-fermentative pathways for synthesis of branched-chain higher
alcohols as biofuels. Nature 451, 86-89 (2008).
23.
Zhang K, Sawaya MR, Eisenberg DS, Liao JC. Expanding metabolism for biosynthesis of
nonnatural alcohols. Proceedings of the National Academy of Sciences, (2008).
24.
Voelker TA, Davies HM. Alteration of the specificity and regulation of fatty acid synthesis of
Escherichia coli by expression of a plant medium-chain acyl-acyl carrier protein thioesterase.
Journal of Bacteriology 176, 7320-7327 (1994).
25.
Handke P, Lynch SA, Gill RT. Application and engineering of fatty acid biosynthesis in Escherichia
coli for advanced fuels and chemicals. Metabolic Engineering 13, 28-37 (2011).
26.
Lennen RM, Pfleger BF. Engineering Escherichia coli to synthesize free fatty acids. Trends in
Biotechnology 30, 659-667 (2012).
27.
Steen EJ, et al. Microbial production of fatty-acid-derived fuels and chemicals from plant
biomass. Nature 463, 559-562 (2010).
28.
Ryder JA. Jet fuel compositions. US Patent US 7589243 (2009).
29.
Tornabene TG. Microorganisms as hydrocarbon producers. Experientia 38, 43-46 (1982).
30.
Schirmer A, Rude MA, Li X, Popova E, del Cardayre SB. Microbial Biosynthesis of Alkanes. Science
329, 559-562 (2010).
131
31.
Choi YJ, Lee SY. Microbial production of short-chain alkanes. Nature advance online publication,
(2013).
32.
Howard TP, et al. Synthesis of customized petroleum-replica fuel molecules by targeted
modification of free fatty acid pools in Escherichia coli. Proceedings of the National Academy of
Sciences 110, 7636-7641 (2013).
33.
Amyris Biotechnologies. http://www.amyris.com/ (2014).
34.
Gevo, Inc. http://www.gevo.com/ (2014).
35.
Joule Unlimited. http://www.jouleunlimited.com/ (2014).
36.
U.S. Energy Information Administration International Energy Statistics.
http://www.eia.gov/cfapps/ipdbproject/iedindex3.cfm?tid=5&pid=62&aid=2&cid=rl,&syid=200
8&eyid=2011&unit=TBPD (2014).
37.
Annual Energy Review 2011. Washington: Government Printing Office, 2012.
38.
Cline PV, Delfino JJ, Rao PSC. Partitioning of aromatic constituents into water from gasoline and
other complex solvent mixtures. Environmental Science & Technology 25, 914-920 (1991).
39.
Sanders WN, Maynard JB. Capillary gas chromatographic method for determining the C3-C12
hydrocarbons in full-range motor gasolines. Analytical Chemistry 40, 527-535 (1968).
40.
Johansen NG, Ettre LS, Miller RL. Quantitative Analysis of Hydrocarbons by Structural Group
Type in Gasolines and Distillates : 1.Gas Chromatography. Journal of Chromatography A 256,
393-417 (1983).
41.
BASIN. Approximate composition (mass fractions) of gasoline.
http://bcn.boulder.co.us/basin/waterworks/gasolinecomp.pdf (2010).
42.
BFIN. Energy Conversion. https://bioenergy.ornl.gov/papers/misc/energyconv.html (2014).
43.
Wallner T, Ickes A, Lawyer K. Analytical Assessment of C2-C8 Alcohols as Spark-Ignition Engine
Fuels. In: Proceedings of the FISITA 2012 World Automotive Congress (edA(eds). Springer Berlin
Heidelberg (2012).
44.
Proti6-Lovasi6 G, Jambrec N, Deur-Siftar D, Prostenik MV. Determination of catalytic reformed
gasoline octane number by high resolution gas chromatography. Fuel 69, 525-528 (1990).
45.
Jackson RW, DeMoss JA. Effects of Toluene on Escherichia coli. Journal of Bacteriology 90, 14201425 (1965).
46.
Simmons B, Loque D, Blanch H. Next-generation biomass feedstocks for biofuel production.
Genome Biology 9, 242 (2008).
47.
Alibaba.com. http://www.alibaba.com/showroom/bulk-glucose.html (2014).
132
48.
ICIS. Indicative Chemical Prices A-Z. http://www.icis.com/chemicals/channel-info-chemicals-a-z/
(2014).
49.
Dugar D, Stephanopoulos G. Relative potential of biosynthetic pathways for biofuels and biobased products. Nat Biotech 29, 1074-1078 (2011).
50.
Morley C. A Fundamentally Based Correlation Between Alkane Structure and Octane Number.
Combustion Science and Technology 55, 115-123 (1987).
51.
EIA. The Availability and Price of Petroleum and Petroleum Products Produced in Countries
Other Than Iran. http://www.eia.gov/analysis/requests/ndaa/pdf/ndaa.pdf (2014).
52.
Speight JG. The Chemistry and Technology of Petroleum (2007).
53.
Reineccius G. Source Book of Flavors, 2nd edn. Aspen Publishers, Inc. (1999).
54.
SAFC Flavors & Fragrances Catalog. http://www.sigmaaldrich.com/content/dam/sigmaaldrich/docs/SAFC/General_nformation/1/safc flavors_and_fragrances catalog.pdf (2014).
55.
Serra S, Fuganti C, Brenna E. Biocatalytic preparation of natural flavours and fragrances. Trends
in Biotechnology 23, 193-198 (2005).
56.
Stirrett K, Denoya C, Westpheling J. Branched-chain amino acid catabolism provides precursors
for the Type I polyketide antibiotic, actinorhodin, via pathways that are nutrient dependent.
Journal of Industrial Microbiology & Biotechnology 36, 129-137 (2009).
57.
Pfeifer BA, Khosla C. Biosynthesis of Polyketides in Heterologous Hosts. Microbiology and
Molecular Biology Reviews 65, 106-118 (2001).
58.
Schweizer E, Hofmann Jr. Microbial Type I Fatty Acid Synthases (FAS): Major Players in a
Network of Cellular FAS Systems. Microbiology and Molecular Biology Reviews 68, 501-517
(2004).
59.
Smith S. The animal fatty acid synthase: one gene, one polypeptide, seven enzymes. The FASEB
Journal 8, 1248-1259 (1994).
60.
White SW, Zheng J, Zhang YM, Rock CO. The structural biology of type 11 fatty acid biosynthesis.
Annual Review of Biochemistry 74, 791-831 (2005).
61.
Kaneda T. Iso-Fatty and Anteiso-Fatty Acids in Bacteria - Biosynthesis, Function, and Taxonomic
Significance. Microbiological Reviews 55, 288-302 (1991).
62.
Zelles L. Fatty acid patterns of phospholipids and lipopolysaccharides in the characterisation of
microbial communities in soil: a review. Biology and Fertility of Soils 29, 111-129 (1999).
63.
Graham IA, Eastmond PJ. Pathways of straight and branched chain fatty acid catabolism in
higher plants. Progress in Lipid Research 41, 156-181 (2002).
133
64.
Kandra L, Wagner GJ. Pathway for the biosynthesis of 4-methyl-1-hexanol volatilized from petal
tissue of Nicotiana sylvestris. Phytochemistry 49, 1599-1604 (1998).
65.
Clejan S, Krulwich TA, Mondrus KR, Seto-Young D. Membrane lipid composition of obligately and
facultatively alkalophilic strains of Bacillus spp. Journal of Bacteriology 168, 334-340 (1986).
66.
Dennis D, McCoy M, Stangl A, Valentin HE, Wu Z. Formation of poly(3-hydroxybutyrate-co-3hydroxyhexanoate) by PHA synthase from Ralstonia eutropha. Journal of Biotechnology 64, 177186 (1998).
67.
Tseng H-C, Martin CH, Nielsen DR, Prather KU. Metabolic Engineering of Escherichia coli for
Enhanced Production of (R)- and (S)-3-Hydroxybutyrate. Applied and Environmental
Microbiology 75, 3137-3145 (2009).
68.
Tseng H-C, Harwell C, Martin C, Prather K. Biosynthesis of chiral 3-hydroxyvalerate from single
propionate-unrelated carbon sources in metabolically engineered E. coli. Microbial Cell Factories
9, 96 (2010).
69.
Shi Z, Blaschek HP. Transcriptional Analysis of Clostridium beijerinckii NCIMB 8052 and the
Hyper-Butanol-Producing Mutant BA101 during the Shift from Acidogenesis to Solventogenesis.
Applied and Environmental Microbiology 74, 7709-7714 (2008).
70.
Venkitasubramanian P, Daniels L, Rosazza JPN. Reduction of Carboxylic Acids by Nocardia
Aldehyde Oxidoreductase Requires a Phosphopantetheinylated Enzyme. Journal of Biological
Chemistry 282, 478-485 (2007).
71.
Reid MF, Fewson CA. Molecular Characterization of Microbial Alcohol Dehydrogenases. Critical
Reviews in Microbiology 20, 13-56 (1994).
72.
Kallberg Y, Oppermann U, J6rnvall H, Persson B. Short-chain dehydrogenases/reductases (SDRs).
European Journal of Biochemistry 269, 4409-4417 (2002).
73.
de la Plaza M, Fern ndez de Palencia P, Pel ez C, Requena T. Biochemical and molecular
characterization of a-ketoisovalerate decarboxylase, an enzyme involved in the formation of
aldehydes from amino acids by Lactococcus lactis. FEMS Microbiology Letters 238, 367-374
(2004).
74.
Hopwood DA, Sherman DH. Molecular Genetics of Polyketides and its Comparison to Fatty Acid
Biosynthesis. Annual Review of Genetics 24, 37-62 (1990).
75.
Lichtenthaler HK, Rohmer M, Schwender J. Two independent biochemical pathways for
isopentenyl diphosphate and isoprenoid biosynthesis in higher plants. Physiologia Plantarum
101, 643-652 (1997).
76.
Dickson RC. Thematic Review Series: Sphingolipids. New insights into sphingolipid metabolism
and function in budding yeast. Journal of Lipid Research 49, 909-921 (2008).
134
77.
Kim Y, Ingram LO, Shanmugam KT. Dihydrolipoamide Dehydrogenase Mutation Alters the NADH
Sensitivity of Pyruvate Dehydrogenase Complex of Escherichia coli K-12. Journal of Bacteriology
190, 3851-3858 (2008).
78.
Shen CR, Lan El, Dekishima Y, Baez A, Cho KM, Liao JC. Driving Forces Enable High-Titer
Anaerobic 1-Butanol Synthesis in Escherichia coli. Applied and Environmental Microbiology 77,
2905-2915 (2011).
79.
Martin CH, Dhamankar H, Tseng H-C, Sheppard MJ, Reisch CR, Prather KLJ. A platform pathway
for production of 3-hydroxyacids provides a biosynthetic route to 3-hydroxy-y-butyrolactone.
Nat Commun 4, 1414 (2013).
80.
Taguchi S, et al. A microbial factory for lactate-based polyesters using a lactate-polymerizing
enzyme. Proceedings of the National Academy of Sciences 105, 17323-17327 (2008).
81.
Hoffmeister M, Piotrowski M, Nowitzki U, Martin W. Mitochondrial trans-2-Enoyl-CoA
Reductase of Wax Ester Fermentation from Euglena gracilis Defines a New Family of Enzymes
Involved in Lipid Synthesis. Journal of Biological Chemistry 280, 4329-4338 (2005).
82.
Kawashima Y, Cheng W, Mifune J, Orita I, Nakamura S, Fukui T. Characterization and Functional
Analyses of R-Specific Enoyl Coenzyme A Hydratases in Polyhydroxyalkanoate-Producing
Ralstonia eutropha. Applied and Environmental Microbiology 78, 493-502 (2012).
83.
Fontaine L, Meynial-Salles I, Girbal L, Yang X, Croux C, Soucaille P. Molecular Characterization
and Transcriptional Analysis of adhE2, the Gene Encoding the NADH-Dependent
Aldehyde/Alcohol Dehydrogenase Responsible for Butanol Production in Alcohologenic Cultures
of Clostridium acetobutylicum ATCC 824. Journal of Bacteriology 184, 821-830 (2002).
84.
Toth J, Ismaiel AA, Chen J-S. The aid Gene, Encoding a Coenzyme A-Acylating Aldehyde
Dehydrogenase, Distinguishes Clostridium beijerinckii and Two Other Solvent-Producing
Clostridia fromClostridium acetobutylicum. Applied and Environmental Microbiology 65, 49734980 (1999).
85.
Akhtar MK, Turner NJ, Jones PR. Carboxylic acid reductase is a versatile enzyme for the
conversion of fatty acids into fuels and chemical commodities. Proceedings of the National
Academy of Sciences 110, 87-92 (2013).
86.
Venkitasubramanian P, Daniels L, Das S, Lamm AS, Rosazza JPN. Aldehyde oxidoreductase as a
biocatalyst: Reductions of vanillic acid. Enzyme and Microbial Technology 42, 130-137 (2008).
87.
McMahon MD, Prather KLJ. Functional Screening and In Vitro Analysis Reveal Thioesterases with
Enhanced Substrate Specificity Profiles That Improve Short-Chain Fatty Acid Production in
Escherichia coli. Applied and Environmental Microbiology 80, 1042-1050 (2014).
88.
De Smidt 0, Du Preez JC, Albertyn J. The alcohol dehydrogenases of Saccharomyces cerevisiae: a
comprehensive review. FEMS Yeast Research 8, 967-978 (2008).
135
89.
Larroy C, Fernandez MR, Gonzalez E, Pares X, Biosca JA. Characterization of the Saccharomyces
cerevisiae YMR318C (ADH6) gene product as a broad specificity NADPH-dependent alcohol
dehydrogenase: relevance in aldehyde reduction. Biochem J 361, 163-172 (2002).
90.
Atsumi S, Li Z, Liao JC. Acetolactate Synthase from Bacillus subtilis Serves as a 2-Ketoisovalerate
Decarboxylase for Isobutanol Biosynthesis in Escherichia coli. Applied and Environmental
Microbiology 75, 6306-6311 (2009).
91.
Debarbouille M, Gardan R, Arnaud M, Rapoport G. Role of BkdR, a Transcriptional Activator of
the SigL-Dependent Isoleucine and Valine Degradation Pathway in Bacillus subtilis. Journal of
Bacteriology 181, 2059-2066 (1999).
92.
Zhang K, Woodruff AP, Xiong M, Zhou J, Dhande YK. A Synthetic Metabolic Pathway for
Production of the Platform Chemical Isobutyric Acid. ChemSusChem 4, 1068-1070 (2011).
93.
Lim HN, Lee Y, Hussein R. Fundamental relationship between operon organization and gene
expression. Proceedings of the National Academy of Sciences 108, 10626-10631 (2011).
94.
Yamanaka Y, Kazuoka T, Yoshida M, Yamanaka K, Oikawa T, Soda K. Thermostable aldehyde
dehydrogenase from psychrophile, Cytophaga sp. KUC-1: enzymological characteristics and
functional properties. Biochemical and Biophysical Research Communications 298, 632-637
(2002).
95.
Inoue K, Makino Y, Itoh N. Purification and Characterization of a Novel Alcohol Dehydrogenase
from Leifsonia sp. Strain S749: a Promising Biocatalyst for an Asymmetric Hydrogen Transfer
Bioreduction. Applied and Environmental Microbiology 71, 3633-3641 (2005).
96.
Moon TS, Yoon S-H, Lanza AM, Roy-Mayhew JD, Prather KU. Production of Glucaric Acid from a
Synthetic Pathway in Recombinant Escherichia coli. Applied and Environmental Microbiology 75,
589-595 (2009).
97.
Niu W, Molefe MN, Frost JW. Microbial Synthesis of the Energetic Material Precursor 1,2,4Butanetriol. Journal of the American Chemical Society 125, 12998-12999 (2003).
98.
Hansen EH, et al. De Novo Biosynthesis of Vanillin in Fission Yeast (Schizosaccharomyces pombe)
and Baker's Yeast (Saccharomyces cerevisiae). Applied and Environmental Microbiology 75,
2765-2774 (2009).
99.
Cracan V, Padovani D, Banerjee R. lcmF Is a Fusion between the Radical B12 Enzyme IsobutyrylCoA Mutase and Its G-protein Chaperone. Journal of Biological Chemistry 285, 655-666 (2010).
100.
Rodriguez-Zavala JS, Allali-Hassani A, Weiner H. Characterization of E. coli tetrameric aldehyde
dehydrogenases with atypical properties compared to other aldehyde dehydrogenases. Protein
Science 15, 1387-1396 (2006).
101.
Jo J-E, Mohan Raj S, Rathnasingh C, Selvakumar E, Jung W-C, Park S. Cloning, expression, and
characterization of an aldehyde dehydrogenase from Escherichia coli K-12 that utilizes 3-
136
Hydroxypropionaldehyde as a substrate. Applied Microbiology and Biotechnology 81, 51-60
(2008).
102.
Ma F, Hanna MA. Biodiesel production: a review. Bioresource Technology 70, 1-15 (1999).
103.
Jones CS, Mayfield SP. Algae biofuels: versatility for the future of bioenergy. Current Opinion in
Biotechnology 23, 346-351 (2012).
104.
Meeuwse P, Sanders JPM, Tramper J, Rinzema A. Lipids from yeasts and fungi: Tomorrow's
source of biodiesel? Biofuels Bioprod Biorefining 7, 512-524 (2013).
105.
Durrett TP, Benning C, Ohlrogge J. Plant triacylglycerols as feedstocks for the production of
biofuels. The Plant Journal 54, 593-607 (2008).
106.
Tai M, Stephanopoulos G. Engineering the push and pull of lipid biosynthesis in oleaginous yeast
Yarrowia lipolytica for biofuel production. Metabolic Engineering 15, 1-9 (2013).
107.
Kaneda T. Iso- and anteiso-fatty acids in bacteria: biosynthesis, function, and taxonomic
significance. Microbiological Reviews 55, 288-302 (1991).
108.
Choi K-H, Heath RJ, Rock CO. 3-Ketoacyl-Acyl Carrier Protein Synthase Ill (FabH) Is a Determining
Factor in Branched-Chain Fatty Acid Biosynthesis. Journal of Bacteriology 182, 365-370 (2000).
109.
Hu Z, et al. Production of branched chain fatty acids and derivatives thereof in recombinant
microbial cells. United States Patent 20110244532 Al (2011).
110.
Jones A, Davies HM, Voelker TA. Palmitoyl-acyl carrier protein (ACP) thioesterase and the
evolutionary origin of plant acyl-ACP thioesterases. The Plant Cell Online 7, 359-371 (1995).
111.
Jing F, et al. Phylogenetic and experimental characterization of an acyl-ACP thioesterase family
reveals significant diversity in enzymatic specificity and activity. BMC Biochemistry 12, 44 (2011).
112.
Torella JP, Ford TJ, Kim SN, Chen AM, Way JC, Silver PA. Tailored fatty acid synthesis via dynamic
control of fatty acid elongation. Proceedings of the National Academy of Sciences 110, 1129011295 (2013).
113.
Javidpour P, et al. Biochemical and Structural Studies of NADH-Dependent FabG Used To
Increase the Bacterial Production of Fatty Acids under Anaerobic Conditions. Applied and
Environmental Microbiology 80, 497-505 (2014).
114.
Dehesh K, Jones A, Knutzon DS, Voelker TA. Production of high levels of 8:0 and 10:0 fatty acids
in transgenic canola by overexpression of Ch FatB2, a thioesterase cDNA from Cuphea
hookeriana. The Plant Journal 9, 167-172 (1996).
115.
Zhang Y-M, Wu B, Zheng J, Rock CO. Key Residues Responsible for Acyl Carrier Protein and 1Ketoacyl-Acyl Carrier Protein Reductase (FabG) Interaction. Journal of Biological Chemistry 278,
52935-52943 (2003).
137
116.
Rafi S, et al. Structure of Acyl Carrier Protein Bound to Fabl, the FASII Enoyl Reductase from
Escherichia coli. Journal of Biological Chemistry 281, 39285-39293 (2006).
117.
Baba T, et al. Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the
Keio collection. Molecular Systems Biology 2, (2006).
118.
Thomason LC, Costantino N, Court DL. E. coli Genome Manipulation by P1 Transduction. In:
Current Protocols in Molecular Biology (edA(eds). John Wiley & Sons, Inc. (2001).
119.
Datsenko KA, Wanner BL. One-step inactivation of chromosomal genes in Escherichia coli K-12
using PCR products. Proceedings of the National Academy of Sciences 97, 6640-6645 (2000).
120.
Griffiths MJ, Hille RP, Harrison STL. Selection of Direct Transesterification as the Preferred
Method for Assay of Fatty Acid Content of Microalgae. Lipids 45, 1053-1060 (2010).
121.
Zor T, Selinger Z. Linearization of the Bradford Protein Assay Increases Its Sensitivity: Theoretical
and Experimental Studies. Analytical Biochemistry 236, 302-308 (1996).
122.
McAuliffe C. Solubility in Water of Paraffin, Cycloparaffin, Olefin, Acetylene, Cycloolefin, and
Aromatic Hydrocarbons. Journal of Physical Chemistry 70, 1267-1275 (1966).
123.
Weast RC, Grasselli JG. CRC Handbook of Data on Organic Compounds. (edA(eds). 2nd edn. CRC
Press, Inc. (1989).
124.
Samuels L, Kunst L, Jetter R. Sealing Plant Surfaces: Cuticular Wax Formation by Epidermal Cells.
Annual Review of Plant Biology 59, 683-707 (2008).
125.
Tillman JA, Seybold SJ, Jurenka RA, Blomquist GJ. Insect pheromone-an overview of biosynthesis
and endocrine regulation. Insect Biochemistry and Molecular Biology 29, 481-514 (1999).
126.
Ladygina N, Dedyukhina EG, Vainshtein MB. A review on microbial synthesis of hydrocarbons.
Process Biochemistry 41, 1001-1014 (2006).
127.
van Leeuwen BNM, van der Wulp AM, Duijnstee 1,van Maris AJA, Straathof AJJ. Fermentative
production of isobutene. Applied Microbiology And Biotechnology 93, 1377-1387 (2012).
128.
PDB ID: 2oc5. Joint Center for Structural Genomics. Crystal Structure of hypothetical protein
(NP_895059.1) from Prochlorococcus marinus MIT9313 at 1.68 A resolution.
129.
Li N, Norgaard H, Warui DM, Booker SJ, Krebs C, Bollinger JM. Conversion of Fatty Aldehydes to
Alka(e)nes and Formate by a Cyanobacterial Aldehyde Decarbonylase: Cryptic Redox by an
Unusual Dimetal Oxygenase. Journal of the American Chemical Society 133, 6158-6161 (2014).
130.
Krebs C, Bollinger Jr JM, Booker SJ. Cyanobacterial alkane biosynthesis further expands the
T
catalytic repertoire of the ferritin-like 5t-di-iron-carboxylateAC
' proteins. Current Opinion in
Chemical Biology 15, 291-303 (2011).
138
131.
Andre C, Kim SW, Yu X-H, Shanklin J. Fusing catalase to an alkane-producing enzyme maintains
enzymatic activity by converting the inhibitory byproduct H202 to the cosubstrate 02.
Proceedings of the National Academy of Sciences 110, 3191-3196 (2013).
132.
Khara B, et al. Production of Propane and Other Short-Chain Alkanes by Structure-Based
Engineering of Ligand Specificity in Aldehyde-Deformylating Oxygenase. ChemBioChem 14,
1204-1208 (2013).
133.
Kunjapur AM, Tarasova Y, Prather KLJ. Synthesis and accumulation of aromatic aldehydes in an
engineered strain of Escherichia coli. Journal of the American Chemical Society Submitted,
(2014).
134.
Slater S, et al. Multiple P-Ketothiolases Mediate Poly(-Hydroxyalkanoate) Copolymer Synthesis
in Ralstonia eutropha. Journal of Bacteriology 180, 1979-1987 (1998).
135.
Blazeck J, Liu L, Knight R, Alper HS. Heterologous production of pentane in the oleaginous yeast
Yarrowia lipolytica. Journal of Biotechnology 165, 184-194 (2013).
136.
Seelbach K, et al. A novel, efficient regenerating method of NADPH using a new formate
dehydrogenase. Tetrahedron Letters 37, 1377-1380 (1996).
137.
Hatrongjit R, Packdibamrung K. A novel NADP+-dependent formate dehydrogenase from
Burkholderia stabilis 15516: Screening, purification and characterization. Enzyme and Microbial
Technology 46, 557-561 (2010).
138.
Doyle S, Klock H, Lesley S. The Polymerase Incomplete Primer Extension (PIPE) Method Applied
to High-Throughput Cloning and Site-Directed Mutagenesis. In: High Throughput Protein
Expression and Purification (edA(eds). Humana Press (2009).
139.
Haywood GW, Anderson AJ, Chu L, Dawes EA. The role of NADH- and NADPH-linked acetoacetylCoA reductases in the poly-3-hydroxybutyrate synthesizing organism Alcaligenes eutrophus.
FEMS Microbiology Letters 52, 259-264 (1988).
140.
Thompson S, Mayerl F, Peoples OP, Masamune S, Sinskey AJ, Walsh CT. Mechanistic studies on
.beta.-ketoacyl thiolase from Zoogloea ramigera: identification of the active-site nucleophile as
Cys89, its mutation to Ser89, and kinetic and thermodynamic characterization of wild-type and
mutant enzymes. Biochemistry 28, 5735-5742 (1989).
141.
Atkinson HJ, Morris JH, Ferrin TE, Babbitt PC. Using Sequence Similarity Networks for
Visualization of Relationships Across Diverse Protein Superfamilies. PLoS ONE 4, e4345 (2009).
142.
Kallberg Y, Oppermann U, Persson B. Classification of the short-chain dehydrogenase/reductase
superfamily using hidden Markov models. FEBS Journal 277, 2375-2386 (2010).
143.
Bond-Watts BB, Weeks AM, Chang MCY. Biochemical and Structural Characterization of the
trans-Enoyl-CoA Reductase from Treponema denticola. Biochemistry 51, 6827-6837 (2012).
139
144.
Pettersen EF, et al. UCSF Chimera-A visualization system for exploratory research and analysis.
Journal of Computational Chemistry 25, 1605-1612 (2004).
145.
The UniProt C. Activities at the Universal Protein Resource (UniProt). Nucleic Acids Research 42,
D191-D198 (2014).
146.
Yan RT, Chen JS. Coenzyme A-acylating aldehyde dehydrogenase from Clostridium beijerinckii
NRRL B592. Applied and Environmental Microbiology 56, 2591-2599 (1990).
147.
Scrutton NS, Berry A, Perham RN. Redesign of the coenzyme specificity of a dehydrogenase by
protein engineering. Nature 343, 38-43 (1990).
148.
De Lay NR, Cronan JE. In Vivo Functional Analyses of the Type II Acyl Carrier Proteins of Fatty
Acid Biosynthesis. (edA(eds) (2007).
149.
Sievers F, et al. Fast, scalable generation of high-quality protein multiple sequence alignments
using Clustal Omega. Molecular Systems Biology 7, (2011).
150.
De Lay NR, Cronan JE. Gene-Specific Random Mutagenesis of Escherichia coli In Vivo: Isolation of
Temperature-Sensitive Mutations in the Acyl Carrier Protein of Fatty Acid Synthesis. Journal of
Bacteriology 188, 287-296 (2006).
151.
Bernard Al, et al. Reconstitution of Plant Alkane Biosynthesis in Yeast Demonstrates That
Arabidopsis ECERIFERUMI and ECERIFERUM3 Are Core Components of a Very-Long-Chain
Alkane Synthesis Complex. The Plant Cell Online 24, 3106-3118 (2012).
152.
Punta M, et al. The Pfam protein families database. Nucleic Acids Research 40, D290-D301
(2012).
153.
EFI. Enzyme Similarity Tool. http://efi.igb.illinois.edu/efi-est/ (2014).
154.
Pieper U, et al. ModBase, a database of annotated comparative protein structure models, and
associated resources. Nucleic Acids Research 39, D465-D474 (2011).
155.
Chen X, Goodwin SM, Boroff VL, Liu X, Jenks MA. Cloning and Characterization of the WAX2
Gene of Arabidopsis Involved in Cuticle Membrane and Wax Production. The Plant Cell Online
15, 1170-1185 (2003).
156.
Hussein NM, Asfour A-FA. Densities and Kinematic Viscosities of Ten Binary 1-Alkanol Liquid
Systems at Temperatures of (293.15 and 298.15) K. Journal of Chemical & Engineering Data 54,
2948-2952 (2009).
157.
Hales JL, Ellender JH. Liquid densities from 293 to 490 K of nine aliphatic alcohols. The Journal of
Chemical Thermodynamics 8, 1177-1184 (1976).
158.
Goenaga JM, Gayol A, Concha RG, Iglesias M, Resa JM. Effect of Temperature on Thermophysical
Properties of Ethanol + Aliphatic Alcohols (C4-C5) Mixtures. Monatsheftefur Chemie - Chemical
Monthly 138, 403-436 (2007).
140
159.
Chao J, Rossini FD. Heats of Combustion Formation and Isomerization of 19 Alkanols. Journal of
Chemical and Engineering Data 10, 374-& (1965).
160.
EIA. Petroleum & Other Liquids: Gasoline and Diesel Fuel Update.
http://www.eia.gov/petroleum/gasdiesel/ (2014).
141