on the effectiveness and challenges of … · aditya vikram pandit doctor of philosophy chemical...

ON THE EFFECTIVENESS AND CHALLENGES OF

ELECTROSYNTHESIS STRATEGIES IN ESCHERICHIA COLI

by

Aditya Vikram Pandit

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Chemical Engineering and Applied Chemistry University of Toronto

© Copyright by A. V. Pandit 2017

ii

On the Effectiveness and Challenges of Electrosynthesis Strategies in

Escherichia coli

Aditya Vikram Pandit

Doctor of Philosophy

Chemical Engineering and Applied Chemistry University of Toronto

2017

Abstract

Conventional bioprocesses that aim to convert CO2 to fuels and chemicals do so through a supply

chain that begins with agricultural products such as corn, intermediaries like dextrose, and eventually

produce chemicals by fermentation. However, it has long been desired that bioprocesses be

established to convert CO2 point source emissions to chemicals. Fundamentally, this is a

thermodynamics problem since the use of CO2 as a feedstock requires an efficient mechanism to

deliver energy to produce chemicals. The focus of this work is to examine several strategies for

microbial electrosynthesis, the delivery of electrical energy to microbial cell factories, for producing

chemicals. The study encompasses four areas type of microbial electrosynthesis and the results are

summarized here:

1) Experimental work was performed to evaluate the affect of neutral red mediated charge transfer

in mutant strains of Escherichia coli for the purpose of producing succinic acid. The results of this

task showed wild-type cells exhibited the greatest molar increase in succinate yield with an 89%

increase while an ldhA deficient strain showed 40% increase. The lack of direct charge transfer

was implicated as the cause since an electron balance was not able to account for an increase in

succinate.

iii

2) We explored the use of mediators such has formate that can be generated from carbon dioxide

and renewable electricity as a carbon source for cell growth and chemical production. An

auxotrophic strategy was employed to engineer formate assimilation, and growth rate on formate

as a C1 donor for folate was determined to be 0.33 h-1. This was 78% of the wild-type strain.

3) We developed a framework for analyzing how metabolic pathways can be efficiently engineered

into microbes to produce chemicals. This orthogonality framework showed ethylene glycol to be a

highly promising substrate for electrosynthesis applications.

4) Finally, a bioprocess for the conversion of ethylene glycol to glycolic acid was characterized and

its suitability to replace glucose as a feedstock was examined. The maximum glycolate titres for

the best performing conditions reached 10.6 g/L. The highest substrate uptake rate for ethylene

glycol was determined to be ca. 5 mmol/gDW-h.

iv

Acknowledgments

I am grateful to my supervisor Krishna Mahadevan who several years ago decided to let a young,

engineering graduate join his lab and work on crazy ideas. With his constant support and excitement

for new ideas, he gave me the freedom to study whatever I was interested.

I would like to thank Prof. Iakounine and Saville for being on my committee and giving me support

and insightful comments in my work.

I would like to thank Paul Jowlabar and Susie Endang for their support. Without Paul’s tools, practical

knowledge and humour, it would not have been possible to build and design some of the experiments

in thesis. Without Susie, I would not have been possible to get anything done! She truly has been the

lab mom.

I would like to express my thanks to some colleagues in the lab. To Nik Anesiadis for first teaching

me microbiology techniques and letting me share his lab bench. To Nick Bourdakos for our semi-

regular chess games in WB319. To Chris Gowen for introducing me to and sharing his enthusiasm

for synthetic biology. To Shyam Srinivasan for being a great collaborator and someone with whom it

was helpful and enjoyable to bounce ideas off. Finally, to all the members of the Mahadevan lab and

Biozone that made a PhD enjoyable.

I am grateful to my parents and family for their love and support. But most importantly this would

not have been possible without their guidance and constant encouragement. For they imparted on

me a lifelong desire for learning, and an ambition to set oneself apart. These have been invaluable.

Finally, to all those ancestors who paved the path for me, and upon whose shoulders I stand.

v

Table of Contents

Acknowledgments ............................................................................................................................................. iv

Table of Contents .............................................................................................................................................. v List of Abbreviations ...................................................................................................................................... viii List of Tables ..................................................................................................................................................... ix List of Figures .................................................................................................................................................... xi List of Appendices ........................................................................................................................................... xv 1 MOTIVATION AND BACKGROUND ........................................................................................................... 1

1.1 Motivation and Research Problem..................................................................................................... 2 1.2 Unifying Theme, Hypotheses and Objectives of the Studies ........................................................ 3 1.3 Scope and Summary of Contributing Work ..................................................................................... 4

1.3.1 Evaluating Escherichia coli as a Platform for Microbial Electrosynthesis ......................... 4 1.3.2 Engineering Utilization of Formate in Escherichia coli ........................................................ 5 1.3.3 Redesigning Metabolism Based on Orthogonality Principles .......................................... 5 1.3.4 Engineering E. coli for Utilization of Ethylene Glycol and Production of Glycolic

Acid ........................................................................................................................................... 6 1.3.5 Appendix A – Expression of Outer Membrane Protein MtoA ....................................... 6 1.3.6 Appendix B – Engineering E. coli to Produce Succinate from Ethylene Glycol ........... 7

1.4 Organization of Thesis ........................................................................................................................ 7 2 LITERATURE REVIEW ................................................................................................................................ 9

2.1 Evolution in the Modern Day Bioprocess ...................................................................................... 10 2.1.1 How Extracellular Electron Transfer May Support CO2 Utilization ............................ 13 2.1.2 Electron Donors Can Be Inorganic ................................................................................... 14 2.1.3 Cytochromes on Metal Respiring Bacteria are Transferable Across Microbes ........... 16 2.1.4 Metal Respiring Bacteria Can Accept Electrons from Electrodes ................................ 17 2.1.5 Electrons Derived from Electrodes Can Improve Chemical Production .................... 18 2.1.6 New Mechanism for Mediator Driven Electron Transfer .............................................. 19 2.1.7 Delivery Systems for the Electron Donor ........................................................................ 21

2.2 Approaches to Engineering Carbon Utilization Pathways ........................................................... 22 2.2.1 CO2 Fixing Pathways ............................................................................................................ 23 2.2.2 Substrate Utilization Pathways ............................................................................................ 26 2.2.3 Carboxylation as a Strategy for Carbon Sequestration .................................................... 29

2.3 Modelling Cellular Metabolism ......................................................................................................... 31 2.3.1 Fundamentals ........................................................................................................................ 31 2.3.2 Flux Balance Analysis ........................................................................................................... 31 2.3.3 Elementary Flux Mode Analysis ......................................................................................... 32

2.4 References ............................................................................................................................................ 34 3 CHARACTERIZATION OF MUTANT STRAINS OF E. COLI IN AN ELECTROCHEMICAL

BIOREACTOR .............................................................................................................................................. 39 3.1 Introduction and Background .......................................................................................................... 40 3.2 Materials and Methods ....................................................................................................................... 42

3.2.1 Culturing Techniques in Microbial Electrosynthesis Reactors ...................................... 42 3.2.2 Microbial Electrosynthesis Reactors .................................................................................. 43 3.2.3 Analytical Methods ............................................................................................................... 44 3.2.4 Calculations ............................................................................................................................ 44 3.2.5 Strains Used in this Study .................................................................................................... 45

vi

3.3 Results .................................................................................................................................................. 45 3.3.1 Construction of Novel Bioreactor ..................................................................................... 45 3.3.2 Wild-type cells for succinate production ........................................................................... 45 3.3.3 Single Mutant Study .............................................................................................................. 46

3.4 Discussion............................................................................................................................................ 48 3.5 References ............................................................................................................................................ 51

4 ENGINEERING UTILIZATION OF FORMATE IN ESCHERICHIA COLI ............................................ 54 4.1 Introduction ........................................................................................................................................ 55 4.2 Results .................................................................................................................................................. 56

4.2.1 Engineering Formate Assimilation by Formate Activation ............................................ 56 4.2.2 Rewiring Folate Metabolism by Deletion of Serine Biosynthesis Pathways ................ 57 4.2.3 Addressing Cell Regulation and Development of Formate Assay ................................ 58 4.2.4 Modelling Formate Pathway Using a Lumped Kinetic Model ....................................... 59 4.2.5 Measuring Intracellular Concentration of Energy Metabolites and Cofactors ............ 63

4.3 Discussion............................................................................................................................................ 64 4.4 Conclusions ......................................................................................................................................... 66 4.5 Material and Methods ........................................................................................................................ 67

4.5.1 Culturing Techniques in Microbial Electrosynthesis ReactorsError! Bookmark not defined.

4.5.2 Analytical Methods ............................................................................................................... 67 4.5.3 Plasmids and Strains ............................................................................................................. 67 4.5.4 Media and Cultivation Conditions ..................................................................................... 67 4.5.5 Max-min Driving Force Thermodynamic Modelling ...................................................... 67 4.5.6 Sampling Methodology for Mass-Spec .............................................................................. 68

4.6 References ............................................................................................................................................ 69 4.7 Data Files ............................................................................................................................................. 72

5 REDESIGNING METABOLISM BASED ON ORTHOGONALITY PRINCIPLES ................................... 73 5.1 Introduction ........................................................................................................................................ 74 5.2 Results .................................................................................................................................................. 77

5.2.1 Defining orthogonal pathways ............................................................................................ 77 5.2.2 Natural metabolism is mostly not orthogonal .................................................................. 77 5.2.3 Growth coupled strategies are not orthogonal ................................................................. 80 5.2.4 Metabolic valves efficiently reduce the solution space .................................................... 84 5.2.5 Orthogonality depends on the substrate utilization pathways ....................................... 85 5.2.6 Orthogonal Cutset Design Allows Calculation of Pathway Energetics ........................ 87

5.3 Discussion and Conclusions ............................................................................................................. 88 5.4 Methods ............................................................................................................................................... 93

5.4.1 Orthogonality: A metric ....................................................................................................... 93 5.4.2 Determining Minimal Cutsets and Control Reactions (ValveFind) .............................. 94 5.4.3 Thermodynamic and Protein Cost Estimations ............................................................... 95

5.5 References ............................................................................................................................................ 96 5.6 Extended Data Set: Redesigning metabolism based on orthogonality principles .................. 100

5.6.1 Synthetic pathway design .................................................................................................. 100 5.6.2 Substrate selection ............................................................................................................. 100 5.6.3 Selection of intermediate precursor(s) ............................................................................ 100 5.6.4 Redox and ATP Cost. ....................................................................................................... 101 5.6.5 Analysis of a Simple Branched Structure ....................................................................... 101 5.6.6 Results of orthogonality score and biomass supporting reactions are generalizable 104 5.6.7 A Study of Counter Examples ......................................................................................... 104

vii

5.6.8 Exception to Natural Metabolism is Not Orthogonal ................................................. 105 5.6.9 Exception to Non-native metabolism as obligate orthogonal pathways ................... 105 5.6.10 Co-factor and other network effects are not captured by a simple branched

reaction network ................................................................................................................ 106 5.6.11 Valve Selection is also a determinant of metabolic independence ............................. 107

6 ENGINEERING ESCHERICHIA COLI FOR UTILIZATION OF ETHYLENE GLYCOL .................. 108 6.1 Introduction ..................................................................................................................................... 109 6.2 Materials and Methods .................................................................................................................... 111

6.2.1 Media and Cultivation Conditions .................................................................................. 111 6.2.2 Culturing Techniques in Reactors ................................................................................... 112 6.2.3 Analytical Methods ............................................................................................................ 112 6.2.4 Plasmids and Strains .......................................................................................................... 113 6.2.5 Flux Balance Analysis ........................................................................................................ 113

6.3 Results ............................................................................................................................................... 114 6.3.1 Ethylene glycol is a preferred substrate over formate .................................................. 114 6.3.2 Ethylene Glycol Utilization by E. coli ............................................................................. 116 6.3.3 Orthogonal Production of Glycolate by E. coli ............................................................. 120 6.3.4 .Dissolved Oxygen and Control Over Metabolism ...................................................... 121 6.3.5 Glycolate Production and Fed Batch Strategy ............................................................... 123 6.3.6 Metabolic Flux Analysis Using E. coli Model ................................................................. 126

6.4 Discussion......................................................................................................................................... 128 6.5 Conclusions ...................................................................................................................................... 131 6.6 References ......................................................................................................................................... 132 6.7 Supplementary Data to Chapter 4 ................................................................................................. 134

7 CONCLUSIONS AND RECOMMENDATIONS ....................................................................................... 135 7.1 General Discussion ......................................................................................................................... 136 7.2 Conclusions ...................................................................................................................................... 138 7.3 Future Work ..................................................................................................................................... 141

Appendix A Expression of Outer Membrane Protein MtoA ...................................................... 144 Overview and Background ..................................................................................................................... 145 Results ....................................................................................................................................................... 145 Conclusions .............................................................................................................................................. 147

Appendix B Engineering Succinate Producing Triple Mutant ................................................. 148 Overview and Background ..................................................................................................................... 149 Results ....................................................................................................................................................... 149 Conclusions .............................................................................................................................................. 150

Appendix C Bioreactor Designs ......................................................................................................... 152 Copyright Acknowledgements .................................................................................................................... 154

viii

List of Abbreviations

ALE Adaptive laboratory evolution

AS Average similarity

DME Dynamic metabolic engineering

EFM Elementary flux modes

FBA Flux balance analysis

FVA Flux variability analysis

HPLC High performance liquid chromatography

LB Lysogeny broth

MCO Metal catalyzed oxidation

MCS Minimal cut sets

MDF Min driving force

MILP Mixed integer linear program

ORP Oxygen reduction potential

RQ Respiratory quotient

SSP Substrate specific productivity

TCA Tricarboxylic acid

THF Tetrahydrofolate

ix

List of Tables

Table 2-1 Innovative approaches for cost competitive bioprocesses using low value feedstocks. ..... 11

Table 2-2 Milestones in the study of microbial electrosynthesis. ............................................................. 19

Table 3 Strains Used for Microbial Electrosynthesis Studies ................................................................... 45

Table 3-2 Fermentation summary showing the molar yields of the products. Completed in biological

duplicate. Molar yields were calculated based on the results presented in Figure 3-2. The yields were

affected by the gene deleted. µ - growth rate in h-1, Q – charge transfer in mmol e-, γ – difference in

degree of reduction per mol glucose between electrical and standard conditions. Yields of succinate,

lactate, formate, acetate and ethanol in mol product/mol glucose. ...................................................... 46

Table 5-1 The orthogonality scores for the various pathways either synthetic or natural consuming

glucose, xylose or ethylene glycol and producing succinic acid are shown. These scores are calculated

from the elementary flux modes of the E. coli core model, using Equations 1 & 2. The model was

modified as necessary to include the reactions for each pathway. The Total Precursor Supporting

Reactions correspond to the total number of reactions that produce one of the 12 precursor

metabolites and is active in each mode, across all elementary flux modes belonging to the space St.

They correspond to the intersection that chemical production has with biomass formation. The

orthogonality score implicitly accounts for this intersection, and the underlying negative correlation is

reflective of the relationship between biomass production and orthogonality. ..................................... 80

Table 5-2 Orthogonality scores for two types of networks are shown. The Growth Coupled score

occurs for a set of gene deletions that couple biomass growth above 0.05 h-1 and product yield > 1

mol/mol. The Orthogonal by Design Network scores are calculated after applying the ValveFind

algorithm described in this publication. The score is calculated for a reduced network after removing

reactions in the cut set, but leaving the valve reaction in the on position. The table also shows the

total number of biomass precursors that can be formed when the metabolic valve is closed. The cost

of operating the pathway is provided using a 10 mmol/gDW∙h as a basis for the calculation. The

values represent total protein cost and the contribution of the thermodynamic cost are shown in

parenthesis. ....................................................................................................................................................... 84

Table 5 Strain and Plasmid Table for Ethylene Glycol Study ................................................................ 113

Table 4-2 Yield and orthogonality metrics for chemical production from different substrates. The

orthogonality scores for various products are shown comparing two substrates that can be generated

electrochemically against conventionally used substrates by their natural pathways. Formate has

orthogonality scores similar to many sugar consuming pathways, indicating a relatively complex and

x

inter-connectedness for its utilization. The highest scores are those for ethylene glycol with yields as

are better than sugars glucose and xylose. Yield is given as g of product per g of substrate............ 115

xi

List of Figures

Figure 2-1 Theoretical Process Flow Diagram for Conversion of CO2. Many biotechnologies have

shown the capacity to convert carbon dioxide to value added chemicals. However, none today have

been demonstrated at a full commercial scale, in many cases owing to the economic challenges of

scaling up the process. In the model proposed, carbon dioxide is first reduced to an electron rich

molecule, preferentially, using a renewable energy source. This molecule is then used as the feedstock

for the bioprocess. This approach avoids many of the technological problems associated with

bioprocesses that directly have CO2 as an input including, light penetration in algae reactors and a

supply of electrical current to the biocatalyst by either by mediators or through direct contact in

bioelectrosynthsis strategies. Hence, by investing energy to reduce CO2 to an electron rich feedstock,

a bioprocess can be designed that is analogous to glucose fermentation. .............................................. 12

Figure 2-2 Model for Using Electrical Energy to Drive a Metabolism. When CO2 is used as a carbon

source, an electron donor is required for cell growth and to drive the synthesis of bioproducts. The

widely accepted industrial model for this process is shown. However, the theory for this model is

based on scientific progress made in environmental microbiology and the role and the mechanisms of

electron transfer between electron donors and acceptors in the natural environment. This image was

reproduced with permission from Liao et al. Copyright Nature Reviews 2015. .................................... 13

Figure 2-3 (A) Redox potential of various electron donors. Energy for growth is derived by coupling

an electron donor to an electron acceptor at a higher standard reduction potential. (B) In direct

electron transfer, a conduit is required for electrons to travel from the inner membrane to the outer

membrane. In electricigens, this conduit is made up of a series of proteins present on the outer

membrane and others spanning the periplasmic space. These proteins are c-type cytochromes

containing reducible heme groups15–17Conductive nanowires extending from the surface of the

microbe have also been identified and are implicated in the transfer of extracellular electrons (Reguera

et al, 2005). The mechanistic information on how microbes transfer electrons from the inner

membrane to a final electron acceptor such as an electrode is generally well understood18. While the

specific proteins responsible for electron transfer are different for different species, the general

mechanism by which electron transfer occurs thought to be analogous if yet still not elucidated.

Proteins exhibit a similar organizational structure for electron transfer out of the cell. For example

Geobacter and Shewanella employ similar strategies of using cytochromes and conductive pili for

extracellular electron transfer. This image was reproduced from Kracke et al. Copyright Frontiers in

Microbiology 2015 under the Creative Common Attribution ("CC BY") licence. ............................... 15

Figure 2-4 Typical Bioelectrochemical System (BES). A typical BES consists of two compartments

separated by an ion exchange membrane that separates the oxidation reaction from the reduction

reaction. The reduction of carbon dioxide occurs at the cathodic compartment. Electrons are

transferred by oxidizing water in the anode compartment. Extracellular electron transfer can occur

by a number of mechanisms described in the previous sections, however, direct electron transfer is

the mode documented. This image was reproduced from Pandit et al. Copyright Microbial Cell

Factories 2012 under the terms of the Creative Commons Attribution License (2.0). ........................ 22

xii

Figure 2-5 Methanol Utilization via the Ribulose Monophosphate Pathway. Methanol and methane

can be used by a cells through a metabolic cycle that produce pyruvate as its end metabolite. Pyruvate

is used as the growth metabolite, generating both tricarboxylic acid (TCA) cycle intermediates via

acetyl-coa as well as other biomass precursors. This image was reproduced with permissions from Fei

et al. Copyright Biotechnology Advances 2014. .......................................................................................... 28

Figure 3-1 Neutral Red. Neutral red acts as a charge carrying mediator in the fermentation broth.

The aromatic ring structure allows electron transfer to the N(CH3)2 group which becomes reduced or

oxidized. The molecule is embedded in the cell membrane and charge is mediated to the quinone

pool. .................................................................................................................................................................. 41

Figure 3-2 Configuration of Bioelectroreactor System. (A) Picture showing the configuration of the

electrochemical bioreactor using the Applikon MiniBio500 vessel. (B) Show a schematic of the

bioreactor assembly. The Ag/AgCl reference electrode was not sterilized by autoclave in the assembly.

It was removed from 6N NaCl solution, sterilized with an isopropanol wipe and aseptically inserted

into the assembly once the reactor and its contents had cooled following sterilization. ...................... 43

Figure 3-3 Growth Characteristics of cells. Shows the distribution of fermentation products between

cells growing under normal conditions and cell growing under a reducing potential. A clear shift

towards the reduced products ethanol and succinate is seen while less lactate is produced. (Top) Wild-

type cells (Middle) Growth curve wild-type cell growing under a reducing potential. Cumulative charge

transferred to cells is shown in blue. (Bottom) Distribution of fermentation products for ldhA mutant.

The error bars represent standard error of two replicates. ....................................................................... 48

Figure 5-1 The ideal structure of an orthogonal pathway in a cell. Green corresponds the EFMs that

produce the desired target chemical and are described by the set St. Blue corresponds to the EMFs

that produce biomass and are described by the set Sx. (A) The branched design is characteristic of

this type of orthogonal structure. (B) We show a hypothetical small network where A is converted

to products E, X (biomass) and T (target compound). The mathematical representation of this

network is described by the elementary flux modes shown below the network in a Boolean matrix,

where blue lines are the biomass-only forming EFMs (3 and 5) and green is the product only forming

EFM (2). This type of network structure can be described as an orthogonal network because A can

be converted to T by reactions v7 and v8 and the metabolic valve v1 can be modulated to be turned

on or off. Traditional metabolic engineering strategies would attempt to drive flux towards the desired

product, T, by growth coupling T to X. For example this may require the deletion of v3, v6 and/or

v7. Orthogonal metabolic engineered strategy relies on the thermodynamics for converting A to T

and manipulating v1 to control flux towards biomass. An example calculation of the orthogonality

score is shown. (C) We show the production envelope for the network containing the elementary flux

modes that describe that solution space. The functionalities of interest of the network are shown in

the green boxes. These represent the desired subspaces Sx containing the elementary modes

𝒆𝒋𝒙(EFM3, EFM5 shown in blue) and St containing the modes 𝒆𝒊𝒕 (EFM2 shown in green). The

orthogonality score is calculated based on the similarity of these subspaces. ........................................ 76

xiii

Figure 5-2 (A) Simplified metabolic map of the glucose consuming pathways analyzed in this

study. Green: Glucose synthetic; Blue: Glycolytic EMP; Orange: Methylglyoxal bypass; Purple: ED

Pathway. (B) Sample cut set strategy for synthetic glucose pathway shows that the structure is

amenable to a metabolic valve topology which bypasses most of the biomass precursors. These

precursors of the central metabolism are required for growth and have been identified in red. The

green x marks which reactions have been identified for deletion by the algorithm to design for

orthogonality. The blue x marks the metabolite valve. Synthetic pathways attempt to bypass these

precursors as well as the points of regulation. A similar branched topology was not observed for

natural glycolytic pathways. ........................................................................................................................... 78

Figure 5-3 Production envelope for succinate production for (A) glucose utilization by glycolysis (B)

glucose utilization by synthetic pathway (C) xylose utilization by the pentose phosphate pathway and

(D) xylose utilization by heterologous synthetic Weimberg pathway. These envelopes capture the

solution space. By controlling a single reaction, it is possible to shrink the solution space to a smaller

defined region of higher product flux. Gray indicated the unmodified network. The metabolic valve

is then modulated from 100% open (red) to 50% (purple), 20% (blue), 10% (green), and 5% open

(yellow). ............................................................................................................................................................. 85

Figure 5-4 Orthogonal pathway design for other substrates considered in this study. (A) The

Weimberg pathway is heterologous to E. coli, however it provides an efficient route for xylose

assimilation that bypasses the central carbon metabolism and most biomass precursor molecules. To

the left of the Weimberg pathway is shown the natural route for xylose assimilation in E. coli through

the pentose phosphate pathway. Succinate dehydrogenase, which converts succinate to fumarate is

an ideal candidate as a metabolic valve (shown in blue) as it allows flux to the TCA cycle and supports

gluconeogenic pathways for cell growth. (B) The orthogonal routes for ethylene glycol assimilation

examined in this study. Malic enzyme is an ideal candidate for a metabolic valve (shown in blue) as

malate decarboxylation to pyruvate can support cell growth. The degree to which the pathway

overlaps with the central carbon metabolism is captured by the orthogonality score for each specific

pathway. ............................................................................................................................................................ 86

Figure 4-1 Glycolate can be produced by a variety of different substrate. These pathways are shown

in the panels. The chemical structures for the metabolites in ethylene glycol and xylose utilization

pathways are also shown. The two most commonly studied substrates for production are xylose (B)

and glucose (C). To efficiently produce glycolate from glucose or xylose, genetic interventions are

required to the central metabolism to couple growth and glycolate synthesis. The focus of this study

examines ethylene glycol consumption. Limiting oxygen provides a mechanism to permit glycolate

accumulation. Under fully aerobic conditions, glycolate is converted to glyoxylate and channeled to

the central metabolism for growth via the glycerate metabolism. Under oxygen limiting conditions,

glycolate accumulates. ................................................................................................................................... 118

Figure 4-2 Cell growth curves and their substrate consumption profiles for the strains constructed in

this study. The oxygen variants of fucO showed a marked difference in growth rate and substrate

utilization in shake-flask experiments. Ethylene glycol consumption is shown by the dashed lines and

xiv

OD600 is depicted by the solid lines. Yellow (light) shows strain LMSE11 while green shows LMSE12.

Error bars indicate standard deviation of triplicate experiments. .......................................................... 119

Figure 4-3 Influence of aeration on glycolate production. To assess the impact of oxygen transfer in

bioreactors, cells were grown under two aeration rates during the micro-aerobic phase of the

fermentation. (Top) High aeration had a flow rate of 150 mL/min. (Bottom) Low aeration was

characterized by flow at 50 mL/min. Experiments were conducted in duplicate. Error bars indicate

range of the measured values. ..................................................................................................................... 121

Figure 4-4 Metabolic modelling glycolate production. Glycolate yield (glycolate, blue), the respiratory

quotient (RQ, green) and the substrate specific productivity (SSP, red) are modelled using FBA.

Glycolate production begins at the onset of oxygen limitation which occurs at approximately 8

mmol/gDW-h of oxygen. At greater values, the RQ plateaus as sufficient oxygen as available for

complete respiration and FBA predicts no glycolate accumulation. The grey bar indicates the values

at which RQ was controlled experimentally during the production phase in later batches. .............. 123

Figure 4-5 Fermentation profiles for fed batch strategies. Fed batch studies were conducted to assess

the long term stability of the production phase. The production phase is separated from the growth

phase by grey shading. (A) Shows bioreactor conditions at 2 v/vm during the growth phase and 0.33

v/vm during the production phase at a cell density corresponding to 4 gDW/L. (B) Cells were grown

at 0.167 v/vm air flow rate into the bioreactor with an average stationary phase cell density at 2.5

gDW/L. Cells were capable of robust glycolate production for well over 100 hours in the production

phase. .............................................................................................................................................................. 125

xv

List of Appendices

Appendix A Expression of Outer Membrane Protein MtoA ...................................................... 144

Overview and Background ..................................................................................................................... 145 Results ....................................................................................................................................................... 145 Conclusions .............................................................................................................................................. 147

Appendix B Engineering Succinate Producing Triple Mutant ................................................. 148 Overview and Background ..................................................................................................................... 149 Results ....................................................................................................................................................... 149 Conclusions .............................................................................................................................................. 150

Appendix C Bioreactor Designs ......................................................................................................... 152 Copyright Acknowledgements .................................................................................................................... 154

Motivation and Background


2 | P a g e

1.1 Motivation and Research Problem

Most life on earth, in one way or another, is dependent on the sun for energy. This dependence on

radiant solar energy, which drives photosynthesis and carbon fixation forms the basis of the modern

biological process by which carbon from the air is captured and converted to fuels and chemicals.

Radiant solar light provides the energy necessary for the conversion of CO2 to starches, which are then

processed to forms that can be used by cells, in the form of glucose and xylose, which are fermented

to fuels like isobutanol and chemicals like succinic acid.

Yet, while nature has evolved organisms to capture inorganic carbon from the air to sustain

life, the energy efficiencies at which that carbon is converted to fuels and chemicals is remarkably low.

By one metric the photon-to-fuel efficiency of ethanol, which measures the fraction of the total energy

captured as photons in the final fuel, is calculated to be roughly 0.18%. This low value is a consistent

limitation of the modern biorefinery. The bulk of this energy loss occurs during the capture of photons

by the plant Photosystem. The net result is that even while fermentation processes may exhibit large

carbon conversion yields and high productivity at the direct fermentation step, the requirement for

other inputs to the process, measured as land, water or fertilizer to support crop growth is large because

of the low photon-to-fuel (or chemical) efficiency. This has financial impacts on the process, but it

can also increase the environmental footprint of the process when full energy and carbon inputs are

considered by way of a life cycle analysis. Finally, in many cases such as when the final chemical

product is highly reduced relative to glucose or xylose, the net carbon sequestration can be quite low.

Photovoltaic devices that generate an electrical output similar to plant Photosystems can do so

at a far greater efficiency than natural processes that transfer electrons from water to charge carriers

such as NADP+ (approaching 50%). This observation and technological advancement in renewable

solar energy has spurred an area of research known as microbial electrosynthesis. Microbial

electrosynthesis, was initially motivated by the hypothesis that if the efficiency losses that occur when

plants undergo photosynthesis could be bypassed if microbes that carry out the fermentation could

directly take up electricity, then the overall photon-to-fuel (or chemical) efficiency could be improved. Hence

the carbon yield could also be improved through the direct use of CO2 as a process input. This process

by which these electrons can be delivered to the cell by direct electron transfer or by the use of

mediators is known as microbial electrosynthesis.


3 | P a g e

The thesis is motivated by a desire to explore the feasibility of microbial electrosynthesis in the

model bacterium Escherichia coli as a way to fix carbon dioxide. Previous experimentation has shown

that E. coli is capable of interacting with an electrode. In this research we attempt to understand the

application of these and related variations on microbial electrosynthesis for the production of chemical

products. Hence, the research problem of this thesis is simple and is guided by one question

that we seek to answer. Is it possible to engineer an organism to bypass starch based

photosynthetic inputs in favour of electrochemically generated inputs?

1.2 Unifying Theme, Hypotheses and Objectives of the Studies

Carbon and electrons are assimilated by microorganisms by different metabolic routes and by different

mechanisms. E. coli is a model, industrial organism that traditionally uses glucose or xylose as both its

electron donor and its carbon source. However, other routes and mechanisms used by different

bacteria can be more energy efficient, are described further in Chapter 2. Hence, a worthwhile

endeavor is to explore the suitability of these heterologous routes for carbon and electron assimilation

in E. coli. This thesis employs metabolic engineering techniques to modify the metabolism of E. coli

to evaluate the assimilation of CO2 and CO2 derived substrates towards biomass and biochemicals.

Thus, the central theme of this thesis explores the feasibility of engineering the industrial model

bacterium E. coli as a platform for microbial electrosynthesis that couples carbon fixation, and the

associated challenges with such an endeavor.

Amidst the many mechanisms present in nature, two routes present themselves at the outset as

being the most viable for heterologous demonstration of microbial electrosynthesis in E. coli. The first

is the assimilation of electrical energy by external mediators including neutral red to aid the direct

assimilation of CO2 by metabolic pathways. The second is the assimilation of reduced carbon species,

which are derived from electrochemical sources, by metabolic pathways. These two assumptions

present as the first two hypothesis of this thesis.

1) Hypothesis 1. E. coli growing in the cathode compartment of an electrochemical cell, in the

presence of a reduction potential and a mediator, will produce greater quantities of succinic

acid.

2) Hypothesis 2. Heterologous pathways that assimilate formate can be used to improve

fermentation processes used for producing valuable chemicals.


4 | P a g e

During the course of this thesis, challenges relating to a viable demonstration of microbial

electrosynthesis motivated us to explore a different approach. Hence, two additional hypothesis were

formulated related to substrate utilization and its validation.

3) Hypothesis 3. The specific utilization pathways of natural and non-natural substrates plays

an important role in the metabolic engineering of cell factories for producing fuels and

chemicals and they can be quantified using metabolic modelling techniques.

4) Hypothesis 4. If E. coli can be engineered to consume ethylene glycol, then because of its

high degree of reduction, it can efficiently produce fermentation products.

To validate these Hypotheses, the work herein has four specific aims.

1) Engineer a strain of Escherichia coli with disruptions to its fermentative metabolism and

characterize its ability to use extracellular electrons by way neutral red to increase production

of succinic acid, which consumes CO2.

2) Engineer metabolic pathways for assimilation of formate and assess the ability of these

pathways to support biomass and chemical production.

3) Use systems biology tools to understand the impact that carbon utilization pathways have on

the broader ability of the cellular network to produce desired biochemicals.

4) Validate the approached laid out in (3) with an energy rich substrate that can be derived from

CO2.

1.3 Scope and Summary of Contributing Work

The following describes the contributing work of this thesis with respect to the specific objectives

outlined above.

1.3.1 Evaluating Escherichia coli as a Platform for Microbial Electrosynthesis

The primary impact of microbial electrosynthesis on the metabolism of growing cells is that it can

provide reducing equivalents to the cell for growth and product synthesis. These reducing equivalents

can increase product yields for reduced metabolites and can drive the assimilation of carbon dioxide


5 | P a g e

for other metabolites. We preformed experiments with a known charge carrying mediator, neutral red,

to evaluate the production of succinate. These experiments represent the first time mutant strains of

the model organism, E. coli, have been used to study the production of succinate in an electrochemical

bioreactor. We found charge mediation to be the primary limitation for driving carbon fixation. This

work has been presented or published in journals and conferences listed below:

Pandit, AV., Mahadevan, R. (2013) Escherichia coli as a platform for bioelectrosynthesis

applications. 35th Symposium on Biotechnology for Fuels and Chemicals. Portland, Oregon.

April 29-May 1 2013.

Pandit, AV., Mahadevan, R. (2016) Using Escherichia coli as a platform for bioelectrosynthesis

applications. Microbial Cell Factories – Technical Note (in draft)

1.3.2 Engineering Utilization of Formate in Escherichia coli

Direct electron transfer was shown to be a limiting factor for succinate production. To overcome

specific challenges related to the transfer of electrons, we explored the idea of using mediators such as

formate that can serve as both an electron donor and carbon source for the cell. These substrates can

be readily generated by the reduction of carbon dioxide in electrochemical reactors. Biosynthetic

pathways for their use in the cell were engineered, and the strains were characterized. This work was

performed during the course of this PhD. However, after encountering many technical challenges,

the experimental portion was eventually abandoned. Chapter 4 describes the experimental approach

undertaken, the challenges encountered, the results and the reason to abandon this project. Future

work is suggested as part of its conclusions.

1.3.3 Redesigning Metabolism Based on Orthogonality Principles

The task of engineering new pathways in the cell is challenged by the interconnectedness of cellular

metabolism. Ubiquitous interactions at the metabolic and regulatory networks of the cell make it

difficult to attain sufficient flux simply by expressing pathway enzymes. This was hypothesized to be

one of the factors limiting sufficient flux through the reductive glycine pathway. Hence, here we

explore a computational method for addressing these challenges with an emphasis on substrate

utilization pathways. During the course of this work, we expanded the scope of this study to ascertain


6 | P a g e

generalizable principles relating to the organization of substrate utilization and its role on chemical

production. The work of this Chapter 5 has been accepted in Nature Communications.

Pandit, AV., Srinivasan, S., Mahadevan, R. (2016) Redesigning metabolism based on

orthogonality principles. Nature Communications (accepted)

Pandit, AV., Srinivasan, S., Mahadevan, R. (2016) Principles of Orthogonal Pathway Design:

A Systems Biology Approach to Growth Uncoupled Chemical Production. Metabolic

Engineering 11. June 26-30, 2016.

Pandit, AV., Srinivasan, S., Mahadevan, R. (2017) Principles of Orthogonal Pathway Design:

A Systems Biology Approach to Growth Uncoupled Chemical Production. 7th International

Conference on Biomolecular Engineering. January 2017.

1.3.4 Engineering E. coli for Utilization of Ethylene Glycol and Production of

Glycolic Acid

During the course of this work, we have explored strategies that drive microbial electrosynthesis. The

experiments performed in this chapter represent a summation of lessons learned during the course of

thesis and attempt to validate the ideas of all the previous sections as they relate to engineering

mechanisms by which electrons generated from renewable sources may, more effectively, be channeled

to the cell. To that end, we engineered E. coli to consume ethylene glycol and produce glycolic acid.

The work relating to this is found in Chapter 6 of this thesis.

Pandit, AV., Mahadevan, R. (2017) A New Substrate for Fermentation: Conversion of

Ethylene Glycol to Glycolic Acid. Biotechnology and Bioengineering (submitted)

1.3.5 Appendix A – Expression of Outer Membrane Protein MtoA

In direct electron transfer, heme containing proteins on the membrane of the cell provide electron

conduits for extracellular transfer across the periplasmic space. It was proposed as part of this PhD

to engineer a strain of E. coli containing a functioning synthetic electron conduit. This work was began


7 | P a g e

but eventually stopped when soon after its commencement, another research group demonstrated a

functioning synthetic conduit. Hence, that work is included only briefly in Appendix A.

1.3.6 Appendix B – Engineering E. coli to Produce Succinate from Ethylene

Glycol

In this section, we attempted to demonstrate succinic acid production from E. coli using an engineered

strain. However, after designing several different variants, we were unable to produce succinic acid.

This section describes those efforts and makes recommendations on strain design for future work that

may result in succinate production.

1.4 Organization of Thesis

Chapter 2. Microbial electrosynthesis seeks to replace this conventional electron donor in

favour of one that is derived electrochemically. A literature review of the state-of-the-art in

the current field of microbial electrosynthesis is presented in Chapter 2. Microbial

electrosynthesis provides growing cells with reducing equivalents for assimilation of carbon

and product synthesis. This chapter analyzes the challenge associated with this approach.

Chapter 3. Results of using electro-fermentation to produce succinate from glucose and CO2

is addressed.

Chapter 4. Wild-type E. coli lacks to ability to assimilate formate as a carbon source. However,

formate can be derived renewably by the reduction of CO2 and electricity and it was

hypothesized that formate assimilation in E. coli can be engineered. In Chapter 4 these

experiments and their results are described. Parts of this chapter were combined with parts of

Chapter 2 for a publication in Biochemical Engineering Journal.

Chapter 5. The framework developed for understanding how pathways for substrate

utilization affect the metabolism and the relationship between the substrate and product pairs.

Chapter 6. Glycolate production by an engineered strain of E. coli for utilizing ethylene glycol.

This study was performed as a result of Hypothesis 3 and the theoretical framework described

in Chapter 5.


8 | P a g e

Chapter 7. Conclusions and recommendations for future work.

Appendix A. The initial PhD proposal document had as part of its scope experiments related

to the engineering of a synthetic electron conduit for carrying electrons across the periplasmic

space and into the cell’s cytoplasm. That work was relied on the expression of membrane

bound proteins and cytochromes. In Appendix A I describe that work and the reasons why it

was stopped.

Appendix B. During the course of this PhD, we attempted to demonstrate applicability of

using ethylene glycol for the production of succinic acid. We were not able to successfully

engineer a strain for consuming ethylene glycol to produce succinate, however, during the

course of that work, we did find that acetate could be used to generate succinate. That work

has been briefly summarized and included in Appendix D since it was never published.

Appendix C. A description of the different bio-electrochemical reactors that did not function

for the neutral red study.

Literature Review

Parts of this chapter were submitted to Biochemical Engineering Journal in an article titled: Microbial

and Electrochemical Routes for the Production of Chemicals from Carbon dioxide.

Literature Review

10 | P a g e

2.1 Evolution in the Modern Day Bioprocess

The modern day bioprocess for making fuels and chemicals has as its input a carbon feedstock

that is derived from a renewable source1. This feature makes it an attractive carbon neutral alternative

to the conventional petrochemical refinery that can produce the same products. Today, several

companies are operating in this space, producing many different fuels and chemicals, proving that the

commercial scale bioprocess can be versatile as well2. However, this recent success hasn’t always been

the case, and it is worthwhile noting why. The clean-tech (white biotech) of early 2000s was built on

the optimism of indefinite high oil prices which led to substantial investment in fuel companies and

small market, low value chemical producing companies. After years of development and piloting, full-

scale demonstrations plants were built, only to see oil prices crash in the mid-2010s3. A number of

companies failed to survive the low oil prices which were unanticipated, but for which high oil prices

were the main driver for the commercialization of their technology. After oil fell below $40/bbl and

long term price forecasts were revised lower, successful bioprocesses set themselves apart from failing

ones by the competitiveness of their technologies with low cost oil. The result of low cost oil was a

structural change in the industry after it became apparent that no one was willing to pay a premium

for a “greener” chemical or fuel. Hence, a lesson learned between the early 2000s and today was that,

to survive, the bio-economy had to shift away from its early lofty goals of producing the world’s

transportation fuels and plastics to the more modest target of making a good return on investment

for their investors in niche markets. These markets have been high value compounds such as

nutraceuticals, fragrances and probiotics.

The shift in focus by the biotech industry to high valued bio-products has left a need to

establish sustainable routes to fuels and chemicals that are competitive at low oil prices. Hence, new

thinking and new research, driven largely by academic discoveries, is being pursued based on a

fundamentally different approach than the corn to sugar to fuel/chemical orthodoxy4,5. That challenge

is being met, in part, by researchers working on reducing the cost of carbon that microbes use.

Examples of these include the utilization of low value, carbon rich waste streams from industrial

processes like carbon monoxide, carbon dioxide, and methane. New discoveries in these areas have

led to the development of and investment in new bioprocess technologies. Several of these innovative

approaches, which are still in the early stages of their development are summarized in Table 2-1.

Literature Review

11 | P a g e

Table 2-2-1 Innovative approaches for cost competitive bioprocesses using low value feedstocks.

Feedstock Bioprocess Innovative

Technology Development Company

Carbon dioxide Algae photosynthesis

Biofilm technology; Innovative bioreactor designs

Lab studies, Pilot studies

SabrTech; Pond Technologies

Carbon monoxide

Clostridium Gas fermentation bioreactor design

Demonstration facilities

Lanza-tech

Methanol E. coli Academic Methane Methylobactrium Gas fermentation Pilot facilities and

lab studies Industrial Microbes; NatureWorks LLC; Intrexon

An important characteristic of all the feedstocks described in Table 2-1 is that they have a lower

Gibbs free energy of formation than glucose and consequently require an energy input for their

conversion to value added bio-products. At -394 kJ/mol, carbon dioxide has the lowest free energy

of formation6. Thus, the task of producing fuels and chemicals from carbon dioxide in ways that are

economically competitive to petrochemical processes, is ultimately a thermodynamics problem.

Refining oil and bitumen into products is thermodynamically downhill, driven by the energy of

breaking carbon bonds. Likewise, utilizing starches as a feedstock is also thermodynamically

downhill6. Hence, any strategy that utilizes low energy feedstocks requires a cost efficient method to

build carbon-carbon bonds. Thus, the challenge for metabolic engineers wishing to produce low

value chemicals that compete with petrochemicals resides in developing alternative schemes by

which cells can efficiently obtain and use the two inputs that are required for growth and for

product synthesis: a carbon donor and an electron donor7. Figure 2-1 shows a scheme which was

suggested by Bar-Evan to address that challenge.

The scheme shows that one approach to utilizing carbon dioxide efficiently is to invest

energy to upgrade CO2 to a reduced form that can be easily metabolized by the cell as a feedstock for

growth and chemical production. Formate was proposed as the reduced molecule, but

hypothetically, it could be any non-toxic molecule. The financial constraint is that the cost of energy

invested needs to be less than the relative cost of glucose to be commercially viable. Secondly, since

the molecule is a non-natural substrate, the adoption of this approach also requires that the

Literature Review

12 | P a g e

biocatalyst have an efficient metabolism for the substrate that includes the presence of metabolic

pathways that support energy production for cell growth. An efficient metabolism can be

engineered through the known repertoire of enzymes present in nature. Hence, the task of the

metabolic engineer is to develop an efficient biocatalyst that integrates into the approach in Figure 2-

1.

In the following sections, I provide background on how nature uses electron donors to drive

metabolic processes and how they can be appropriated by metabolic engineers into synthetic

pathways that bypass the conventional pathways to yield more economical routes to low value

chemicals that can compete with the petrochemical industry. I focus on the role that these electron

donors play on the ability for cell to utilize carbon dioxide and cite specific examples of carbon

dioxide utilization at the metabolic pathway level. My goal in the literature review is to provide a

historical narrative that explains both the experimental and the theoretical works which gave rise

to the approach that is laid out in Figure 1-1 as way for low cost carbon dioxide utilization.

Figure 2-1 Theoretical Process Flow Diagram for Conversion of CO2. Many biotechnologies have shown the capacity

to convert carbon dioxide to value added chemicals. However, none today have been demonstrated at a full commercial

scale, in many cases owing to the economic challenges of scaling up the process. In the model proposed, carbon dioxide

is first reduced to an electron rich molecule, preferentially, using a renewable energy source. This molecule is then used

as the feedstock for the bioprocess. This approach avoids many of the technological problems associated with

bioprocesses that directly have CO2 as an input including, light penetration in algae reactors and a supply of electrical

current to the biocatalyst by either by mediators or through direct contact in bioelectrosynthsis strategies. Hence, by

investing energy to reduce CO2 to an electron rich feedstock, a bioprocess can be designed that is analogous to glucose

fermentation. This image was reproduced with permission from Bar-Evans et al. Copyright Nature Reviews 2015.

https://www.google.ca/url?sa=i&rct=j&q=&esrc=s&source=images&cd=&cad=rja&uact=8&ved=0ahUKEwjtiOmY1sLSAhVl9YMKHfc6Dd0QjRwIBw&url=http://www.sciencedirect.com/science/article/pii/S1367593116300965&psig=AFQjCNHL8YUIQ0Efe-E3tRiyGzVY5-GU5w&ust=1488916358274866

Literature Review

13 | P a g e

2.1.1 How Extracellular Electron Transfer May Support CO2 Utilization

Cellular metabolism is driven by the energy that is derived from the oxidation of electron donors in

the environment. Whereas traditional bioprocesses use glucose as both an electron donor and carbon

source for growth, cells that use CO2 as a carbon source require an external electron donor. How

cells naturally acquire these electrons from their environment has long been studied in environmental

microbiology8. Advancements in understanding the mechanisms of that electron transfer directly give

rise to the approach in Figure 2-2, a direct antecedent of the model proposed in Figure 2-19. In this

section, I provide a background on work done studying electron donors and acceptors.

Figure 2-2 Model for Using Electrical Energy to Drive a Metabolism. When CO2 is used as a carbon

source, an electron donor is required for cell growth and to drive the synthesis of bioproducts. The widely

accepted industrial model for this process is shown. However, the theory for this model is based on

scientific progress made in environmental microbiology and the role and the mechanisms of electron

transfer between electron donors and acceptors in the natural environment. This image was reproduced

with permission from Liao et al. Copyright Nature Reviews 2015.

https://www.google.ca/url?sa=i&rct=j&q=&esrc=s&source=images&cd=&ved=0ahUKEwj1_qP0hvrSAhVH44MKHf97D3wQjRwIBw&url=http://www.nature.com/nrmicro/journal/v14/n5/abs/nrmicro.2016.32.html&psig=AFQjCNGj1PUBtwMe_8xSHdK1XHN0Z_vfIw&ust=1490819716568304

Literature Review

14 | P a g e

2.1.2 Electron Donors Can Be Inorganic

There exist in nature several different electron donors that can be used by microorganisms to generate

energy. These are broadly classified as inorganic substrates and organic substrates. The incredible

diversity in microbes and the environments they live in means that there are large variations in the

types of inorganic substrates that can be used as electron donors. Ferrous, ammonia, nitrite, sulfide

and sulfur are among the most prevalent in nature10–12. These donors shown in Figure 2-3a are

typically oxidized on the cell membrane and transferred to the electron transport chain where they

can be used to generate energy or reducing equivalents required for cell growth. Specifically, as the

electrons are transferred by heme containing proteins up the redox potential, a H+ or Na+ motive

force is built across the cell membrane by ion translocation that allows for the generation of ATP via

ATP synthase. The specific electron donor that a cell uses is dependent on the cell’s growth

environment and is coupled to an acceptor molecule at a higher redox potential. The difference in

their respective half-cell reduction potentials determines the theoretical amount of energy that the cell

can obtain by this redox reaction13. Therefore, microbes capable of using Fe2+ (as Fe3O4) are capable

of obtaining more energy than those that can use H2S when coupled to the same electron acceptors,

given that all other factors are equal.

Another important electron donor in natural systems is hydrogen gas. As with other inorganic electron

donors, hydrogen is oxidized by a membrane bound complex or in some cases by soluble protein

complexes in the periplasm where it can be coupled to the reduction of electron carriers such as

NADPH or ferredoxin. To my knowledge, no study has yet shown the use of hydrogen as an artificial

electron donor for E. coli. However, hydrogen in the headspace and fermentation media of a

bioreactor has been shown to modulate oxygen reduction potential (ORP) which indirectly affects the

fermentation product profile14.

Literature Review

15 | P a g e

Figure 2-3 (A) Redox potential of various electron donors. Energy for growth is derived by coupling an electron donor to an electron acceptor at a higher standard reduction potential. (B) In direct electron transfer, a conduit is required for electrons to travel from the inner membrane to the outer membrane. In electricigens, this conduit is made up of a series of proteins present on the outer membrane and others spanning the periplasmic space. These proteins are c-type cytochromes containing reducible heme groups15–17Conductive nanowires extending from the surface of the microbe have also been identified and are implicated in the transfer of extracellular electrons (Reguera et al, 2005). The mechanistic information on how microbes transfer electrons from the inner membrane to a final electron acceptor such as an electrode is generally well understood18. While the specific proteins responsible for electron transfer are different for different species, the general mechanism by which electron transfer occurs thought to be analogous if yet still not elucidated. Proteins exhibit a similar organizational structure for electron transfer out of the cell. For example Geobacter and Shewanella employ similar strategies of using cytochromes and conductive pili for extracellular electron transfer. This image was reproduced from Kracke et al. Copyright Frontiers in Microbiology 2015 under the Creative Common Attribution ("CC BY") licence.

Electron Donors Can Be Organic Compounds

Organic electron donors typically refers to carbon containing compounds that can serve as both the

electron donor and carbon source for the cell. These can include common saccharides such as glucose

or xylose, but can also include acids like acetate or formate, alcohols like ethanol and methanol and

even gases such as carbon monoxide. Some compounds, especially those belonging to the C1 class

require specialized membrane bound enzymes to catalyze their oxidation and deliver their electrons

to the electron transport chain19. This is the case for methanol as well as carbon monoxide. However,

in general, organic electron donors can be oxidized through metabolic pathways through the

nicotinamide cofactors that are universal donors for the electron transport chain.

A B

Literature Review

16 | P a g e

For the purposes of this study, we are interested in those organic substrates that can be derived

electrochemically20. This allows us to focus on only a handful of electron donors. Substrates that are

being actively investigated in the literature are formate and methanol for which metabolic engineers

are keen on establishing functional pathways where these compounds can serve as non-natural

substrates for growth. However, to date no study has demonstrated cell growth to be completely

supported by either formate or methanol by the use of heterologous pathways. Carbon monoxide,

which can also be derived electrochemically, has largely been the studied in the context of its natural

metabolism in Clostridium and Methylobacterium species21.

2.1.3 Cytochromes on Metal Respiring Bacteria are Transferable Across

Microbes

The technical problem that belies the use of an artificial electron donor as shown in Figure 2-2 is the

creation of a biological capacity to acquire electrons22 in the absence of such natural pathways. However,

environmental microbiologists have provided metabolic engineers with an insight on how electron

transfer occurs both in the oxidative direction23–26 as well as the much more studied reductive direction

in metal respiring bacteria15,17,27–29.

The characteristic model organism with this capacity is Acidithiobacillus ferrooxidans in which electron

transfer from iron(II) to the cell is thought to occur via enzymes and charge carrying proteins that are

present on the membranes and that span the periplasmic space30. A reversible NADH dehydrogenase

then accepts the electrons from the periplasmic electron carrier in order to drive the synthesis of

NADH. Interestingly, electron transport in the far better studied iron oxidizing bacteria such

as Geobacter and Shewanella has been shown to be bi-directional.

Fortunately, there is remarkable similarity in the underlying cellular machinery that carries out

the oxidative reactions despite the diversity of the donors10 (Figure 2-3b). This similarity can be

exploited by metabolic engineers. For example, in 2008, researchers trying to understand how

Shewanella carries out extracellular electron transfer found that a single enzyme was sufficient to confer

onto E. coli the ability to reduce chelated metal ions31. The work’s important implication, that was not

immediately obvious at the time of the publication, was that proteins involved in extracellular electron

transfer could be functionally expressed in an industrial workhorse organism. Others would eventually

Literature Review

17 | P a g e

put forward the notion that with an adequate understanding of the key components of the electron

transport chain in one organism, it may be possible to transfer that entire machinery to another,

industrially relevant one for the purposes of driving carbon fixation (Figure 2-2). Indeed,

advancements in synthetic biology methods for refactoring large gene clusters32–35 has allowed

researchers to take this approach.

The seminal paper was published 2010 when researchers were successful at expressing proteins

from Shewanella so that an engineered strain of E. coli could reduce metal ions and solid α-Fe2O336.

Briefly, the E. coli genome lacks the necessary cytochromes for except for NapC which is found in the

periplasmic space and implicated in mediating electron transfer to periplasmic nitrate reductase

NapAB31. However, the protein has a 52% similarity to CymA and can serve the same purpose as

CymA in iron reduction – as a linker between the inner membrane and the outer-membrane36. Based

on this hypothesis, the researchers believed it might be possible to transfer electrons from NapC to

the heterologous MtrA protein. MtrA expression along with the membrane cytochromes MtrB and

MtcC could complete the conduit, and pass electrons to hematite37. The transfer of electrons by

proteins that span the periplasm was successfully demonstrated using cytochromes from Shewanella

which served as a template for direct electron transfer (Figure 1-3b).

While the work did not show the ability for E. coli to acquire electrons from inorganic sources,

the experiments met an important milestone in the field: the ability of a non-native pathway to

function as an extracellular electron transfer conduit in an industrially relevant host organism.

2.1.4 Metal Respiring Bacteria Can Accept Electrons from Electrodes

The fundamental driving force behind the transfer of electrons from inorganic electron donors to the

cell is the difference in the half-cell potential of the oxidation reaction of the donor and the mid-point

potential of the protein that accepts the electron. This charge transfer, which is dependent on the

mid-point potentials of the proteins, hints at the possibility that organisms capable of donating

electrons to an electrode may be able to accept electrons instead if the set potential of the electrode

were lower than membrane proteins’ potential. This important experiment was carried out in 2008

by Derek Lovley’s group and they demonstrated that it was indeed possible for Geobacter to use a

graphite electrode donor, providing energy for cell growth. This work laid the foundation for a

Literature Review

18 | P a g e

broader examination of the principles of microbial electrosynthesis by the greater scientific community

and their applications that that work could have for the industrial production of chemicals.

2.1.5 Electrons Derived from Electrodes Can Improve Chemical Production

While the observation of Lovley and coworkers in 2008 was a great scientific leap in the field, it

followed a long steady line of incremental progress made by many researchers studying the influence

of non-specific electrical energy on the growth of microorganisms. The earliest work appears to date

back to 1940 in an experiment in which an electric current was passed through boxes of tea in order

to accelerate fermentation. The researcher found that tea leaves that were electrofermented tea prioer

to steeping was faster fermenting, stronger in brew and better tasting (Lominadze, 1940). In the years

since, a general understanding has developed showing the capability of electrical energy to modulate

metabolism and affect product yields. In 1979, Hongo and co-workers used these techniques to

increase L-glutamine yields38. Much later, researchers began adding organic compounds (viologen

dyes) to the fermentation broth and began noticing an altered change in metabolism, first beginning

with work by Rao and Mutharasan that showed methyl and benzyl viologen aided in directing the

carbon flow from acid production to solvent production in Clostridium acetobutylicum without any

electrical energy from an electrode. It was not until 1988 that a pair of researchers showed that by

electrochemically reducing methyl viologen in the fermentation media, they could affect the final

propionate concentrations produced by Clostridium acetobutylicum.

That observation spurred new studies in the area of electro-fermentation – fermentation in

the presence of a reducing potential and mediators to carry a charge. Hence, these electron shuttles

which became implicated in electron transfer to the cell became central to understanding how electrical

energy could be used to enhance chemical production. Much of that work was linked to supplying

energy from an electrode in addition to glucose which is also an electron donor. Finally, a number of

studies led by the Zeikus group expanded the understanding of microbial electrosynthesis by studying

more microbes including the yield improvements in ethanol fermentation by eukaryotes like

Saccharomyces cerevisiae39. Another major finding was that microbial electrosynthesis was capable of

creating a proton motive force that could be used by the cell to create additional ATP. Their studies

helped to create a more detailed model for understanding the impact of non-specific electrical energy

on the cellular metabolism by taking into account redox and energy balances40–42,39.

Literature Review

19 | P a g e

2.1.6 New Mechanism for Mediator Driven Electron Transfer

In mediator driven electron processes, it was initially thought that the role of mediator such as neutral

red was to serve as electronophores or electron channels. Zeikus and colleagues modelled the process

as a hydrophobic compound capable of binding to the cell membrane, and conducting a charge when

it become reduced by the electrode by direct contact, and transferring that charge directly to reduce

NAD+ to NADH. However, substantial progress was made in 2014 by Harrington and co-workers

that eventually elucidated the mechanism underlying the observations of mediated electron transfer43.

They discovered that, at least for E. coli growing in the presence of neutral red, there was no actual

direct electron transfer to NAD+ as other researchers had thought. Instead, they determined that

neutral red was bound to the cell membrane which permitted electrons to be transferred between the

electrode and the cell’s menaquinone pool. Recognizing this reduced state in the menaquinone pool

through regulatory signals, the cell shifted its metabolism towards the production of more reduced

products. This critical body of work put into perspective the difficulty of using mediated electron

transfer in non-electricigens to drive flux through redox dependent metabolic pathways. It appears in

light of these new findings that mediator driven electron transfer seems to be an inefficient mechanism

to reduce intracellular co-factors required for product synthesis, and more importantly CO2 reduction.

Hence, perhaps unbeknownst to scientists at the time who were conducting research on

fermentation of tea leaves, almost 50 years later their work has evolved as an important area of study

examining how bacteria can use electrodes as a sole energy source for chemical production. A

summary of the major accomplishments in the field is shown in Table 2.1. Next I examine the delivery

of extracellular electron transfer.

Table 2-2-2 Milestones in the study of microbial electrosynthesis.

Description Novelty Comments Reference

R. eutropha was grown in an

electrochemical bioreactor where

current was delivered to a cathode.

CO2 sparged into the reactor was the

sole carbon source and the system

produced isobutanol 3-methyl-1-

butanol.

First demonstration of using a

biological platform to directly convert

CO2 to fuels using only electricity as

an energy source and in the absence

of any other carbon source.

Unclear whether

formate was the

energy source or

whether H2 was

generated in the

reactor setup that

served as the

electron donor.

44

Literature Review

20 | P a g e


Several acetogens were screened for

their ability to grow in an

electrochemical bioreactor from CO2 as

a sole carbon source.

Sporomusa and Clostridium were among

those that were able to produce

acetate and 2-oxobutyrate as the

primary products.

Electron efficiencies

in the final product

were remarkably

high at 80%.

45

Methanobacterium produces methane as a

primary product when grown in an

electrochemical bioreactor supplied

with an electrical current at -1.0V and

CO2.

Demonstration that production of

methane in electrochemical reactors

occurs because methanotrophs are

capable of direct electron transfer and

not by the assimilation of acetate or

hydrogen.

First demonstration

of Archaea as a

platform for

production of

biofuels.

46

Geobacter grown in the presence of a

reducing potential at -500mV

(Ag+/AgCl)

First demonstration of electrodes

serving as a direct electron donor for

anaerobic respiration.

47

Wild-type Actinobacillus succinogenes

grown under a reducing potential

increases succinate yields.

At the time, was an important

experiment showing usefulness of

electrical energy to improve product

yields.

Hypothetical model

depicting the role of

neutral red directly

reducing NAD+ was

eventually

determined to be

incorrect.

40

S. ovata grown in an electrochemical

reactor powered by a solar cell.

Showed solar energy directly

powering the production of

biochemical.

Production rates

were very low,

reaching 6 days to

produce less than 1

mmole of acetate.

48

Interesting study using a natural

electricigen Shewanella to consume an

electrical current by reversing the

natural electron flow across the

membrane cytochromes.

Showed that electron flow through

Mtr can be reversed.

Relied on soluble

mediators to achieve

reverse electron

flow.

49

Clostridium and Saccharomyces both

produce higher titres of ethanol when

grown under a reducing potential with

cellulose or glucose, respectively, as a

carbon source.

Authors did not

investigate the likely

role of hydrogen

production during

high potential to

modulate

metabolism towards

ethanol.

39

Determined how neutral red impacts

fermentation when E. coli is grown in a

reducing potential.

First experimental validation of the

underlying mechanism of neutral red

mediated electron transfer in E. coli

Important finding

showed that

electrons are in fact

not directly

transferred to NAD

cofactors but instead

only reduce the

menaquinone pool.

43

E. coli was engineered to have a

functioning set of membrane proteins

First demonstration of E. coli being

capable of transferring electron

Earlier work by

researchers has

36

Literature Review

21 | P a g e


to carry out extracellular electron

transfer.

across the periplasmic space to solid

iron nanoparticles for reduction.

shown that the

proteins expressed in

this study can be

reversed to accept

electrons. It will be

interesting to see if

this functionality can

be ported over to E.

coli.

2.1.7 Delivery Systems for the Electron Donor

A bioelectrochemical system (BES) is a compartmentalized reactor system composed of an anodic

and cathodic compartment separated by a proton exchange membrane. The anodic compartment

contains an electrode where a substrate serves as the electron donor to the anode as it becomes

oxidized. In the cathodic compartment, electrons are taken up by an electron acceptor causing the

chemical species to become reduced. Theoretically, either reaction at the anode or the cathode can

be catalyzed chemically (abiotically) or microbially (biotically). Microbial fuel cells are examples where

the anodic reaction is microbially catalyzed, during which a chemical substrate such as acetate is

metabolized by a microbe such as G. sulfurreducens and the anode is in turn used as the terminal electron

acceptor for the microbe50. At the cathode, the electrons react with protons and oxygen in the

presence of a catalyst such as platinum to form water. The redox reactions occurring at the cathode

and anode can be described by the following equations:

Anode: H2O → 2 H+ + 2 e- + ½ O2

Cathode: cytox + e- → cytred

BESs that are used for bioelectrosynthesis operate reversely to a fuel cell. The cathode compartment

is catalyzed by a microbe18. Electrons are accepted by the microbe and incorporated into its

metabolism. The microbe uses the electrons and the chemical substrate present in

the cathodic compartment to produce biochemicals and biomass. The electrons are supplied by the

anodic compartment where a chemical species is oxidized. Water is typically oxidized at the anode

into protons, electrons and oxygen. Wastewater containing organic substrates such as acetate can also

serve as an electron donor at the anode18.

Literature Review

22 | P a g e

Figure 2-4 Typical Bioelectrochemical System (BES). A typical BES consists of two compartments separated by an ion exchange membrane that separates the oxidation reaction from the reduction reaction. The reduction of carbon dioxide occurs at the cathodic compartment. Electrons are transferred by oxidizing water in the anode compartment. Extracellular electron transfer can occur by a number of mechanisms described in the previous sections, however, direct electron transfer is the mode documented. This image was reproduced from Pandit et al. Copyright Microbial Cell Factories 2012 under the terms of the Creative Commons Attribution License (2.0).

2.2 Approaches to Engineering Carbon Utilization Pathways

Carbon dioxide is the most oxidized form of carbon that cells can use for growth and chemical

synthesis. However, to drive efficient carbon fixation a mechanism to assimilate an electron donor is

also required. The dual elements of this task is central to what is the incredibly challenging problem

of engineering heterologous pathways into cells for fixing carbon dioxide. Consequently, there have

been many different approaches taken over the past several decades to realize an efficient strategy in

engineering carbon assimilation into cells. Fundamentally, these approaches can be differentiated by

the underlying metabolic pathways that are used support the carbon fixation. In this section, I provide

background information on the distinct approaches to carbon fixation. At the end of this section, we

Literature Review

23 | P a g e

conclude on the interplay between the carbon fixing pathways and the spectrum of strategies that

supply the electron donor necessary for CO2 assimilation.

2.2.1 CO2 Fixing Pathways

There exists in nature a tremendous variation in the types of pathways that organisms use to fix carbon

dioxide. These pathways differ in their requirements for ATP and NAD(P)H, in their topology (linear

or cyclical), in their kinetic properties, and in the end metabolite produced from the pathway which

serves as the cell’s growth substrate. A summary of these pathways is shown in Table 2-3. Many of

the pathways described in the table are natural and present across a diverse range of microorganisms.

However, several are synthetic, built around the various carboxylating enzymes present in nature but

assembled in pathways that are designed to, hypothetically, be more efficient. These two

fundamentally different types of pathways give rise to the first set of differences when trying to

engineer cell systems for carbon fixation: namely natural versus synthetic pathways.

Table 2-3 Summary of the carbon fixing pathways.

Pathway Electron Donors ATP

Requirement

Carbon

Species

Fixed

Bottleneck

Wood-Ljungdahl Ferredoxin: 2

NAD(P)H: 3 1 3 CO2

Expression of CO-dehydrogenase-

acetylo-CoA-synthase complex

Reductive Pentose

Phosphate NAD(P)H: 5 7 1 CO2

Energetic costs to high

rPP is not isolated from central

metabolism

Reductive TCA Ferredoxin: 2

NAD(P)H: 3 2 3 CO2

sucA and icd are thermodynamically

unfavourable

3HP/4HB Cycle NAD(P)H: 7 9 3 HCO3-

Requires coordinated expression of

complex, mutli-enzyme pathway

Dicarboxylate-4-

Hydroxybutyrate

Cycle

Ferredoxin: 2 or 3

NAD(P)H: 2 or 3

Other: 1

5 2 CO2

1 HCO3-

Requires coordinated expression of

complex, mutli-enzyme pathway

Serine Pathway NAD(P)H: 3 3 1 HCOOH

1 CO2 Higher ATP requirements

Reductive Glycine

Pathway NAD(P)H: 3 2

2 HCOOH

1 CO2

In vivo demonstration of glycine

reduction required

The Holy Grail for scientists researching CO2 utilization has long been the conversion of a heterotroph

to an auxotroph. Since the Calvin-Benson-Bassham cycle was published in 1954, RuBisCO, the

Literature Review

24 | P a g e

enzyme responsible for carbon fixation has been intensely studied, first expressed in E. coli in 198151.

With advances in genetic tools available to scientists, it was eventually demonstrated that the

expression of the RubisCO gene could increase biomass yield and improve titres for ethanol. Hence,

the study of carbon fixing pathways became an important avenue of research for metabolic engineers

looking to produce industrial compounds from CO2 using cell factories52. This important area of

research for plant biologists soon evolved into an important area for applied research. More

fundamental work in this field was still performed for example, examining ways to improve the

catalytic properties of Rubisco, although they were met by limited success. Metabolic engineers began

to turn towards other domains of life catalyzing carbon fixing reactions, recognizing that previous

efforts to demonstrate an active CBB cycle (via RubisCO) required a constant supply of glycolytic

metabolites generally supplied from glucose53.

An important milestone occurred in 2013 when the sub-pathways of the 3HP carbon fixing bicycle

from Chloroflexus aurantiacus were functionally expressed in E. coli by the Silver research group54.

Researchers from the same group had published a year earlier another hallmark study in which they

functionally expressed the carboxysome from Halothiobacillus neapolitanus showing carbon fixing

ability55. Carboxysomes are bacterial microcompartments containing a carbonic anhydrase as well as

the RubisCO enzyme. However, it was not until 2016 that the daunting task of engineering a

functioning and independent carbon fixing cycle was established in E. coli by the Milo group53. They

showed that by expressing RubisCO and prk and deleting the native pgm gene, it was possible to build

a completely independent CBB cycle in E. coli by feeding only pyruvate and CO2, using an evolutionary

approach to derive the final strain.

Several other important approaches are also worth noting. While many researchers took the approach

of Milo and co-workers, others realized the inherent challenge in converting a heterotroph to an

autotroph, especially doing so using a cyclical pathway. Hence, many turned towards synthetic carbon

fixation: the notion that artificial pathways could be designed that are not present in nature from the

known repertoire of carboxylating enzymes (Table 2-5). Among the most prominent publications in

this area were a series of papers published by Milo. None, to our knowledge, has yet been

demonstrated in the literature.

Literature Review

25 | P a g e

Finally, it is worth noting while important research was conducted on engineering heterologous carbon

fixing pathways into E. coli, substantial improvements because of better genetic tools allowed

researchers to produce chemicals and fuels in organisms naturally capable of consuming carbon

dioxide. The most significant of these publications as well as those pertaining the general strategies

of engineering synthetic carbon fixation in E. coli is shown below in Table 2-4.

Table 2-4 Significant publications in the field of carbon fixation.


Re-engineered metabolic pathways of the central carbon metabolism to convert glucose to acetyl-coa without any carbon loss.

First demonstration that glucose can be used in a way that reduces carbon loss.

While pathway was determined to be functional in vivo, since the pathway produces no NADH, an additional carbon source is required for energy.

56

E. coli engineered to use CO2 and pyruvate as the sole carbon sources for growth.

Hallmark study engineering E. coli to efficiently use Rubisco to fix carbon dioxide and produce biomass without deleting biomass precursors.

While pyruvate is used as a carbon source it can be envisioned that another substrate could efficiently be engineered.

53

An aldolase was engineered to efficiently allow the conversion of formate to DHAP, a precursor for cell growth.

Hallmark study on engineering formate utilization in E. coli.

Substantial improvements in the enzyme kinetic characteristics would be required.

57

Important study demonstrating formate can be added to the fermentation media and be used an electron donor for succinate production to increase product yield.

Formate supplied in the presence of heterologous FDH.

Important case study in how formate can be used as an efficient electron mediator.

58

Computational study examining the various routes that formate could be assimilated for growth by E. coli

Exhaustive computational search. Reductive glycine pathway is the focus of this thesis.

59

Computational study examining the various routes that CO2 could be assimilated for growth by E. coli.

Exhaustive computational search. None has ever been experimentally validated.

60

Engineered an E. coli strain consume methanol as an auxiliary substrate and produce various TCA intermediates.

Methanol was found to be incorporated into the final compound 13C tracer experiments.

Important study relating to CO2 research if methanol can be efficiently produced electrochemically.

61

Another important study showing methanol utilization in E. coli.

Found a highly efficient methanol dehydrogenase.

62

Literature Review

26 | P a g e

2.2.2 Substrate Utilization Pathways

While applied research has had some substantial achievement in engineering heterotrophs to use

carbon dioxide as a substrate, it is quite evident that the gap in the ability to efficiently use carbon

dioxide between engineered organisms and those that natively have this ability is substantial. A cause

for this lies in part because of the cyclical nature of many natural carbon fixing pathways which creates

the challenging task of balancing flux through a cycle. Yet another cause is the requirement of a

simultaneous and yet independent pathway to assimilate an electron donor. To address these

challenges, several researchers proposed using substrates that could be derived electrochemically and

hence a reduced form of carbon dioxide that can also serve as an electron donor. The seminal thought

paper was proposed by Conardo et al at DARPA and included CO, methanol and formate as possible

carbon sources. In this section, we will examine the use of methanol, known as synthetic

methylotrophy, and formate, known as formatotrophy, as the carbon sources for cell growth. The

general approach could be simplified in this way: just as corn is processed to liberate sugar which is

ultimately used in a fermentation, CO2 can be processed to generate reduced carbon species which would

be the primary substrate for cell growth. Hence, this approach would eliminate the requirement to

engineer an additional pathway for a non-native electron donor.

Formate Utilization

Formate utilization has been heavily studied. In the context of traditional metabolic engineering,

formate has been used as an electron donor when it is provided as an auxiliary substrate in the

fermentation broth. Cells growing in the presence of formate have shown improvements in yields of

reduced products such as succinate and even Penicillin G58. Hence, it is was well known when

researchers at the US Department of Energy originally proposed using formate at a carbon source9.

Owing to its non-toxicity and the substantial improvements in fuel cell technology that was being

commercialized, formate was proposed as a useful substrate for cell growth and chemical production.

Indeed, the literature is abundant with the examples of organisms using formate as a carbon source.

Then, in 2012, Liao et al published a study showing that Ralstonia eutropha could use formate in an

electrochemical bioreactor to produce isobutanol44. While the study used a natural organism with the

ability to use formate, there remains some questions over the exact mechanism of carbon utilization

Literature Review

27 | P a g e

in that study, it was nonetheless a milestone in demonstrating that non-specific electrical energy could

be catalyzed by microbes in the presence of CO2 to produce a valuable fuel.

This work has continued both in the computational realm where researchers have proposed new

pathways that could be assembled in organisms to utilize formate but also in the lab. Perhaps one of

the most significant and novel contributions to this field was made in 2015 when a group of

researchers developed a pathway to convert formate to dihydroxyacetone phosphate using an

engineered aldolse that initially acted on benzylaldehyde57 assembled in to a linear pathway. The

pathway, known as the “formalose pathway” was shown to have in vitro activity; however, in vivo activity

has yet to be demonstrated likely owing to extremely low catalytic efficiency. Interestingly, in an

analogous approach in 2015 researchers from the Clapes group showed that formaldehyde could be

used to form carbohydrates in vitro also using an aldolase.

Methanol Utilization

There are a number of organisms in nature capable of using methanol as a carbon and energy source.

Methanol can be utilized naturally by both aerobic and anaerobic organisms in pathways linked to

pyrroloquinoline quinone dehydrogenases, nicotinamide adenine dinucleotide oxidoreductases, flavin

adenine nucleotide-dependent alcohol oxidases as well as methanol:corrinoid methyltransferases80.

The serine cycle or ribulose monophosphate pathway are examples of methanol assimilation in nature

(Figure 2-5).

The task of engineering synthetic methylotrophy into an organism like E. coli is challenging

because methanol is assimilated by a cyclical pathway that ligates formaldehyde onto the backbone of

ribulose monophosphate. In many ways, there is a lot of similarity between engineering synthetic

methanol utilization and formate utilization since the first step of methanol utilization produces

formaldehyde. In E. coli, the approach taken by several groups has been engineering the ribulose

monophosphate pathway (RuMP) because several of the genes required to operate this cycle are native

to the organism. Early work was originally focused on finding a suitable methanol dehydrogenase –

the enzyme responsible for converting methanol to formaldehyde. Several suitable enzymes have

been identified and successfully engineered into E. coli in the literature. 13C studies have validated the

incorporation of methanol into cell mass and particularly the biomass precursors. In 2016, methanol

Literature Review

28 | P a g e

was demonstrated to contribute to improved biomass yields by over 30% and was incorporated into

naringenin as a final product61.

In a contrasting approach, James Liao and coworkers worked on establishing a synthetic pathway to

convert methanol to ethanol and butanol7. Their efforts were focused on in vitro demonstrations and

identified kinetic limitations of the pathway. That work, which was known as the methanol

condensation cycle, has yet to be fully demonstrated in vivo.

Figure 2-5 Methanol Utilization via the Ribulose Monophosphate Pathway.

Methanol and methane can be used by a cells through a metabolic cycle that produce

pyruvate as its end metabolite. Pyruvate is used as the growth metabolite, generating

both tricarboxylic acid (TCA) cycle intermediates via acetyl-coa as well as other biomass

precursors. This image was reproduced with permissions from Fei et al. Copyright

Biotechnology Advances 2014.

Other Pathways for Utilization of CO2

A number of organic substrates that can be derived electrochemically20 (Figure 2-6). Hence, it is

worthwhile considering that these compounds, apart from methanol or formate may also be suitable

substrates for cell growth and chemical production. Some will be more suitable than others. For

example, toxicity issues could arise from the use of glycolaldehyde. In contrast, ethanol and ethylene

glycol are examples of substrates that might be well suited for growth if they could be produced

efficiently by reducing CO2. Indeed, both Clostridium species and yeasts are known to have metabolism

that supports growth on ethylene glycol or ethanol, respectively.

https://www.google.ca/url?sa=i&rct=j&q=&esrc=s&source=images&cd=&cad=rja&uact=8&ved=0ahUKEwj_qK2zg_zSAhWL5YMKHUaID_cQjRwIBw&url=https://www.researchgate.net/figure/261608021_fig7_Simplified-molecular-pathway-of-ribulose-monophosphate-RuMP-cycle-for-methanotrophic&psig=AFQjCNEg8Kgph0jk6MAH9r_ZRMy5VNVe4A&ust=1490887600558408

Literature Review

29 | P a g e

Figure 2-6 The current efficiency of various products that can be produced by the electrochemical reduction of CO2. This image was reproduced with permissions from Kuhl et al. Copyright Energy and Environmental Science 2012.

2.2.3 Carboxylation as a Strategy for Carbon Sequestration

Carboxylases catalyze the assimilation of CO2 by the cellular metabolism to biomass and product

formation. Carboxylases have many roles, such as autotrophy, carbon assimilation anapleorisis or

redox-balancing functions63. In the context of production of fuels and chemicals, they are a very

important class of enzymes because they essentially allow for the conversion of “free” carbon into

valuable product. In a review published in 2011, Tobias Erb described five major types of

carboxylases. These are described in the table below.

Literature Review

30 | P a g e

Table 2-5 Types of carboxylating enzymes present across metabolism. These enzymes are the

functional step that enable carbon fixings pathways to exist.

Carboxylation Type Description

autotrophic

carboxylases

All carboxylating enzymes that serve in these autotrophic pathways and

allow the direct transformation of inorganic carbon into central

precursor molecules

assimilatory

carboxylases

Function in the dedicated heterotrophic pathways that allow the

transformation of organic compounds into central precursor molecules

anaplerotic

carboxylases

Enzymes that mainly serve in TCA cycle-refilling reactions

biosynthetic

carboxylases

Aspartate class of carboxylating enzymes that operate in biosynthetic

pathways starting from central intermediates

redox-balancing

carboxylases

Used for enzymes that function mainly in removing excess reducing

equivalents [NAD(P)H] during metabolism by using CO2 as an electron

acceptor

Carbon assimilation via carboxylases is important for improving product synthesis. Production of the

commodity chemical succinate is a prime example of how efficient conversion of CO2 by pepck is

useful. Overexpression of the native carboxylating enzyme increased succinate production by 6.5

times by Zeikus and co-workers. Heterologous expression of a pyruvate carboxylase was

demonstrated to improve product yields more significantly. Hence, in this regard, several studies have

focused on engineering products that contain a carboxylating enzyme in their pathway and

supplementing glucose with an auxiliary source of electron donors. The classical example is the

production of succinic acid. Electrons have been provided by both a direct electrical current as well

as by the addition of formate.

Metabolic reorganization

A complimentary approach to using CO2 has focused on rewiring metabolism to fix carbon dioxide.

Non-oxidative glycolysis was an important advancement in this area of research that allowed cells to

produce acetyl-coa without losing any carbon from glucose56. Its relationship to microbial

electrosynthesis was based on the requirement for an external electron donor to provide an energy

source for cell growth. The development of computational approaches for redesigning metabolism

has made it easier to identify pathways that minimize carbon loss59,64,65.

Literature Review

31 | P a g e

2.3 Modelling Cellular Metabolism

2.3.1 Fundamentals

Cellular metabolism can be described by a series of chemical reactions occurring in a small volume.

These reactions which produce and consume metabolites (products and reactants) can be represented

through standard chemical engineering mass balances and thus the cell as a whole can be modelled

using a set of mathematical relationships. When cells grow during the exponential phase, cell growth

is relatively quick, and it can often be assumed that the metabolite concentrations in the cell are time

invariant. Thus, it becomes possible to model the cell using a series of linear equations of defined

stoichiometry following a steady-state (pseudo) assumption. In this section, we will largely concern

ourselves with this type of modelling. However, it is worthwhile noting that microbial physiology is

often modelled using kinetic, non-steady state equations that tend to be more complex. We will only

refer the reader to some excellent review publications that explain this methodology.

The fundamental approach to modelling cellular metabolism begins with identifying gene protein

reactions that occur in the cell. These reactions are then used to construct a stoichiometric matrix

consisting of metabolites and their reaction interactions. For each reaction, thermodynamics is used

to constrain the directionality of the reaction. Additional constraints on the allowable reaction flux

can also be imposed (for example by applying Boolean logic resulting from gene regulatory

interactions in the network, or by identifying an ATP maintenance value resulting in ATP hydrolysis).

The process of applying constraints is known as constraint based modelling. We explore the two

types of modelling performed in this thesis below.

2.3.2 Flux Balance Analysis

Flux balance analysis (FBA) is a type of biased modelling that solves the stoichiometric matrix by

maximizing a particular reaction in the stoichiometric matrix66,67. That reaction is often the biomass

growth rate equation, although it does not always have to be. Flux balance analysis identifies the

steady state intracellular flux distribution corresponding to the maximization of the objective function.

This distribution is used to infer cellular phenotypes and make predictions based on genetic

interventions to the cell. Mathematically, the solution of an FBA problem is a single flux vector

through the flux cone corresponding to the objective function. It is described by the following linear

optimization problem:

Literature Review

32 | P a g e

max 𝑍 = 𝑤 ∙ 𝑣

𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 𝐶

𝑤ℎ𝑒𝑟𝑒 𝐶 = {

𝑆𝑣 = 𝑜𝑣 ≥ 0

𝑣 ≤ 𝑣𝑚𝑎𝑥

𝑣𝑚,𝑚𝑖𝑛 ≤ 𝑣 ≤ 𝑣𝑚,𝑚𝑎𝑥

where, S is the stoichiometric matrix, vmax and vmin are the upper and lower bounds on the flux

distribution, v is the flux variable and w is the vector of objective co-efficient usually for which all

elements are 0 except that which corresponds to the biomass growth reaction. This type of constraint

based modelling technique can be highly valuable because it requires only thermodynamic data and

network stoichiometry to infer the intracellular fluxes of the cell. Hence, the analysis allows one to

generate hypotheses regarding the active metabolic pathways in the cell. This is useful for metabolic

engineering because of its semi-predictive ability to understand what is happening in the cell by

measuring only extracellular exchanges fluxes such as glucose uptake rate. Hence, it is extremely useful

for strain development.68,69

2.3.3 Elementary Flux Mode Analysis

Elementary flux modes (EFM) analysis is an unbiased modelling technique for representing the cellular

metabolism70–74. By this I mean that this analysis decomposes the stoichiometric matrix into a set of

feasible steady state solution vectors that are minimal in nature. A linear combination of these

solutions, known as the elementary flux modes can be used to describe any feasible flux vector within

the cellular metabolism. In this way, elementary flux modes is more apt to describe feasible network

functions and the underlying flux distribution as opposed to biomass specific flux vectors.

Mathematically, elementary flux modes characterize a metabolic network consisting of m internal

metabolites and n reactions of an m × n stoichiometric matrix, S. Any value Sij represents the

stoichiometric coefficient of the metabolite i for reaction j. A flux mode is any non-trivial flux vector

v that is a solution to Sv = 0 and corresponds to all the reversibility constraints of S. Zanghellini

defined EFMs by the following mathematical framework74:

We define the support of a mode:

supp(v) = {i|vi≠ 0} as the set of indices of non-zero elements in the flux mode v.

Literature Review

33 | P a g e

A mode is called an EFM, e, if cannot be written as a proper superset of any other feasible mode v:

supp(e) ⊃/supp(v)

EMFs are also useful modelling technique for metabolic engineers. Its usefulness is derived by its

ability to assess a networks functionality based on its stoichiometry. Practically, this translates to

computational methodologies that block the production of undesired phenotypes such as the

production of competing side-products. Several strain design algorithms have been developed

employing elementary flux modes.75–79 Zanghellini provides an excellent review of elementary flux

modes and their application to metabolic engineering74.

Literature Review

34 | P a g e

2.4 References

1. Chen, Y. & Nielsen, J. Advances in metabolic pathway and strain engineering paving the way for sustainable production of chemical building blocks. Curr. Opin. Biotechnol. 24, 965–972 (2013).

2. Erickson, B., Nelson & Winters, P. Perspective on opportunities in industrial biotechnology in renewable chemicals. Biotechnol. J. 7, 176–185 (2012).

3. Cumming, W. Energy Supply Glut: Insights from the US. AdviserVoice PTY Limited (2015). Available at: https://www.adviservoice.com.au/2015/11/energy-supply-glut-insights-from-the-us/. (Accessed: 29th March 2017)

4. Haynes, C. A. & Gonzalez, R. Rethinking biological activation of methane and conversion to liquid fuels. Nat Chem Biol 10, 331–339 (2014).

5. Clomburg, J. M., Crumbley, A. M. & Gonzalez, R. Industrial biomanufacturing: The future of chemical production. Science (80-. ). 355, aag0804 (2017).

6. Jiang, Z., Xiao, T., Kuznetsov, V. L. & Edwards, P. P. Turning carbon dioxide into fuel. Philos. Trans. A. Math. Phys. Eng. Sci. 368, 3343–3364 (2010).

7. Liao, J. C., Mi, L., Pontrelli, S. & Luo, S. Fuelling the future: microbial engineering for the production of sustainable biofuels. Nat. Rev. Microbiol. 14, 288–304 (2016).

8. Madsen, E. L. Microorganisms and their roles in fundamental biogeochemical cycles. Curr. Opin. Biotechnol. 22, 456–464 (2011).

9. Conrado, R. J., Haynes, C. A., Haendler, B. E. & Toone, E. J. in Advanced Biofuels and Bioproducts (ed. Lee, J. W.) 1037–1064 (Springer New York, 2013). doi:10.1007/978-1-4614-3348-4_38

10. Kracke, F., Vassilev, I. & Krömer, J. O. Microbial electron transport and energy conservation - The foundation for optimizing bioelectrochemical systems. Front. Microbiol. 6, 1–18 (2015).

11. Ormeland, R. S. & Stolz, J. F. The Ecology of Arsenic. Science (80-. ). 300, 939–944 (2003).

12. Richardson, D. J. Bacterial respiration: A flexibe process for a changing environment. Microbiology 146, 551–571 (2000).

13. McCarty, P. Thermodynamic electron equivalents model for bacterial yield prediction: Modifications and comparative evaluations. Biotechnol. Bioeng. 97, 377–388 (2007).

14. Riondet, C., Cachon, R., Waché, Y., Alcaraz, G. & Diviès, C. Extracellular oxidoreduction potential modifies carbon and electron flow in Escherichia coli. J. Bacteriol. 182, 620–6 (2000).

15. Londer, Y. Y., Pokkuluri, P. R., Orshonsky, V., Orshonsky, L. & Schiffer, M. Heterologous expression of dodecaheme ‘nanowire’ cytochromes c from Geobacter sulfurreducens. Protein Expr. Purif. 47, 241–8 (2006).

16. Logan, B. E. Exoelectrogenic bacteria that power microbial fuel cells. Nat. Rev. Microbiol. 7, 375–81 (2009).

17. Shelobolina, E. S. et al. Importance of c-Type cytochromes for U(VI) reduction by Geobacter sulfurreducens. BMC Microbiol. 7, 16 (2007).

18. Rabaey, K. & Rozendal, R. a. Microbial electrosynthesis - revisiting the electrical route for microbial production. Nat. Rev. Microbiol. 8, 706–16 (2010).

Literature Review

35 | P a g e

19. Murrell, J. C., Gilbert, B. & McDonald, I. R. Molecular biology and regulation of methane monooxygenase. Arch. Microbiol. 173, 325–332 (2000).

20. Kuhl, K. P., Cave, E. R., Abram, D. N. & Jaramillo, T. F. New insights into the electrochemical reduction of carbon dioxide on metallic copper surfaces. Energy Environ. Sci. 5, 7050–7059 (2012).

21. Travick, J. D., Burk, M. & Burgard, A. P. Microogranisms and Methods for Conversion of Syngas and Other Carbon Sources to Useful Products. (2010). doi:20100304453A1

22. Tremblay, P.-L., Angenent, L. T. & Zhang, T. Extracellular Electron Uptake: Among Autotrophs and Mediated by Surfaces. Trends Biotechnol. xx, 1–12 (2016).

23. Elbehti, A., Brasseur, G. & Lemesle-Meunier, D. First evidence for existence of an uphill electron transfer through the bc(1) and NADH-Q oxidoreductase complexes of the acidophilic obligate chemolithotrophic ferrous ion-oxidizing bacterium Thiobacillus ferrooxidans. J. Bacteriol. 182, 3602–6 (2000).

24. Quatrini, R. et al. Insights into the iron and sulfur energetic metabolism of Acidithiobacillus ferrooxidans by microarray transcriptome profiling. Hydrometallurgy 83, 263–272 (2006).

25. Brasseur, G., Bruscella, P., Bonnefoy, V. & Lemesle-Meunier, D. The bc(1) complex of the iron-grown acidophilic chemolithotrophic bacterium Acidithiobacillus ferrooxidans functions in the reverse but not in the forward direction. Is there a second bc(1) complex? Biochim. Biophys. Acta 1555, 37–43 (2002).

26. Rohwerder, T., Gehrke, T., Kinzler, K. & Sand, W. Bioleaching review part A: progress in bioleaching: fundamentals and mechanisms of bacterial metal sulfide oxidation. Appl. Microbiol. Biotechnol. 63, 239–48 (2003).

27. Aklujkar, M. et al. The genome sequence of Geobacter metallireducens: features of metabolism, physiology and regulation common and dissimilar to Geobacter sulfurreducens. BMC Microbiol. 9, 109 (2009).

28. Izallalen, M. et al. Geobacter sulfurreducens strain engineered for increased rates of respiration. Metab. Eng. 10, 267–75 (2008).

29. Zhuang, K. et al. Genome-scale dynamic modeling of the competition between Rhodoferax and Geobacter in anoxic subsurface environments. ISME J 5, (2011).

30. Valdés, J. et al. Acidithiobacillus ferrooxidans metabolism: from genome sequence to industrial applications. BMC Genomics 9, 597 (2008).

31. Gescher, J. S., Cordova, C. D. & Spormann, A. M. Dissimilatory iron reduction in Escherichia coli: identification of CymA of Shewanella oneidensis and NapC of E. coli as ferric reductases. Mol. Microbiol. 68, 706–19 (2008).

32. Kelwick, R., MacDonald, J. T., Webb, A. J. & Freemont, P. Developments in the tools and methodologies of synthetic biology. Front. Bioeng. Biotechnol. 2, 60 (2014).

33. Way, J. C., Collins, J. J., Keasling, J. D. & Silver, P. A. Integrating Biological Redesign: Where Synthetic Biology Came From and Where It Needs to Go. Cell 157, 151–161 (2014).

34. Sánchez-Pascuala, A., de Lorenzo, V. & Nikel, P. I. Refactoring the Embden-Meyerhof-Parnas pathway as a whole of portable GlucoBricks for implantation of glycolytic modules in Gram-negative bacteria. ACS Synth. Biol. acssynbio.6b00230 (2017). doi:10.1021/acssynbio.6b00230

35. Smanski, M. J. et al. Functional optimization of gene clusters by combinatorial design and assembly. Nat. Biotechnol. 32, 1241–1249 (2014).

36. Jensen, H. M. et al. Engineering of a synthetic electron conduit in living cells. Proc. Natl. Acad. Sci. (2010). doi:10.1073/pnas.1009645107

Literature Review

36 | P a g e

37. Mitchell, a C. et al. Role of outer membrane c-type cytochromes MtrC and OmcA in Shewanella oneidensis MR-1 cell production, accumulation, and detachment during respiration on hematite. Geobiology 10, 355–70 (2012).

38. Hongo M, I. Application of electro-energizing method to L-glutamic acid fermentation. Agric Biol Chem 43, 2075–2081 (1979).

39. Shin, H. S., Zeikus, J. G. & Jain, M. K. Electrically enhanced ethanol fermentation by Clostridium thermocellum and Saccharomyces cerevisiae. Appl. Microbiol. Biotechnol. 58, 476–81 (2002).

40. Park, D. H. & Zeikus, J. G. Utilization of electrically reduced neutral red by Actinobacillus succinogenes: physiological function of neutral red in membrane-driven fumarate reduction and energy conservation. J. Bacteriol. 181, 2403–10 (1999).

41. Zeikus, J. G. & Park, D. H. Electrochemical methods for generation of a biological proton motive force and pyridine nucleotide cofactor regeneration. (2001).

42. Park, D. H. & Zeikus, J. G. Electricity generation in microbial fuel cells using neutral red as an electronophore. Appl. Environ. Microbiol. 66, 1292–7 (2000).

43. Harrington, T. D. et al. The mechanism of neutral red-mediated microbial electrosynthesis in Escherichia coli: menaquinone reduction. Bioresour. Technol. 192, 689–695 (2015).

44. Li, H. et al. Integrated electromicrobial conversion of CO2 to higher alcohols. Science 335, 1596 (2012).

45. Nevin, K. P. et al. Electrosynthesis of organic compounds from carbon dioxide is catalyzed by a diversity of acetogenic microorganisms. Appl. Environ. Microbiol. 77, 2882–6 (2011).

46. Cheng, S., Xing, D., Call, D. F. & Logan, B. E. Direct biological conversion of electrical current into methane by electromethanogenesis. Environ. Sci. Technol. 43, 3953–8 (2009).

47. Gregory, K. B., Bond, D. R. & Lovley, D. R. Graphite electrodes as electron donors for anaerobic respiration. Environ. Microbiol. 6, 596–604 (2004).

48. Nevin, K. P., Woodard, T. L., Franks, A. E., Summers, Z. M. & Lovley, D. R. Microbial Electrosynthesis: Feeding Microbes Electricity To Convert Carbon Dioxide and Water to Multicarbon Extracellular Organic Compounds. MBio 1, e00103-10-e00103-10 (2010).

49. Ross, D. E., Flynn, J. M., Baron, D. B., Gralnick, J. a & Bond, D. R. Towards electrosynthesis in shewanella: energetics of reversing the mtr pathway for reductive metabolism. PLoS One 6, e16649 (2011).

50. Rabaey, K. et al. Microbial ecology meets electrochemistry: electricity-driven and driving communities. ISME J. 1, 9–18 (2007).

51. Gatenby, A. A., Castleton, J. A. & Saul, M. W. Expression in E. coli of maize and wheat chloroplast genes for large subunit of ribulose bisphosphate carboxylase. Nature 291, 117–121 (1981).

52. Palsson, B. O., Fathi-Afshar, S., Rudd, D. F. & Lightfoot, E. N. Biomass as a source of chemical feedstocks: an economic evaluation. Science (80-. ). 213, (1981).

53. Antonovsky, N. et al. Sugar Synthesis from CO2 in Escherichia coli. Cell 166, 115–125 (2016).

54. Mattozzi, M. d, Ziesack, M., Voges, M. J., Silver, P. A. & Way, J. C. Expression of the sub-pathways of the Chloroflexus aurantiacus 3-hydroxypropionate carbon fixation bicycle in E. coli: Toward horizontal transfer of autotrophic growth. Metab. Eng. 16, 130–139 (2013).

55. Chen, A. H., Robinson-Mosher, A., Savage, D. F., Silver, P. A. & Polka, J. K. The Bacterial Carbon-Fixing

Literature Review

37 | P a g e

Organelle Is Formed by Shell Envelopment of Preassembled Cargo. PLoS One 8, e76127 (2013).

56. Bogorad, I. W., Lin, T.-S. & Liao, J. C. Synthetic non-oxidative glycolysis enables complete carbon conservation. Nature 502, 693–697 (2013).

57. Siegel, J. B. et al. Computational protein design enables a novel one-carbon assimilation pathway. Proc. Natl. Acad. Sci. U. S. A. 112, 3704–9 (2015).

58. Balzer, G. J., Thakker, C., Bennett, G. N. & San, K. Y. Metabolic engineering of Escherichia coli to minimize byproduct formate and improving succinate productivity through increasing NADH availability by heterologous expression of NAD+-dependent formate dehydrogenase. Metab. Eng. 20, 1–8 (2013).

59. Bar-Even, A., Noor, E., Flamholz, A. & Milo, R. Design and analysis of metabolic pathways supporting formatotrophic growth for electricity-dependent cultivation of microbes. Biochim. Biophys. Acta - Bioenerg. 1827, 1039–1047 (2013).

60. Bar-Even, A., Noor, E., Lewis, N. E. & Milo, R. Design and analysis of synthetic carbon fixation pathways . Proc. Natl. Acad. Sci. U. S. A. 107, 8889–8894 (2010).

61. Whitaker, W. B. et al. Engineering the Biological Conversion of Methanol to Specialty Chemicals in Escherichia coli. Metab. Eng. 39, 49–59 (2016).

62. Müller, J. E. N. et al. Engineering Escherichia coli for methanol conversion. Metab. Eng. 28, 190–201 (2015).

63. Erb, T. J. Carboxylases in natural and synthetic microbial pathways. Appl. Environ. Microbiol. 77, 8466–8477 (2011).

64. Bogorad, I. W. et al. Building carbon-carbon bonds using a biocatalytic methanol condensation cycle. Proc. Natl. Acad. Sci. U. S. A. 111, 15928–33 (2014).

65. Chowdhury, A. & Maranas, C. D. Designing overall stoichiometric conversions and intervening metabolic reactions. Sci. Rep. 5, 16009 (2015).

66. Orth, J. D., Palsson, B. Ø. & Fleming, R. M. T. Reconstruction and Use of Microbial Metabolic Networks: the Core Escherichia coli Metabolic Model as an Educational Guide. EcoSal Plus 4, (2010).

67. Hyduke, D. et al. COBRA Toolbox 2.0. (2011).

68. Feist, A. M. et al. Model-driven evaluation of the production potential for growth-coupled products of Escherichia coli. Metab. Eng. 12, 173–86 (2010).

69. Machado, D. & Herrgård, M. Co-evolution of strain design methods based on flux balance and elementary mode analysis. Metab. Eng. Commun. 2, 85–92 (2015).

70. Schuster, S., Klamt, S., Weckwerth, W., Moldenhauer, F. & Pfeiffer, T. Use of network analysis of metabolic systems in Bioengineering. Bioproc Biosyst Eng 24, (2002).

71. Schuster, S. & Hilgetag, C. On elementary flux modes in biochemical reaction systems at steady tate. J Biol Syst 2, (1994).

72. Schuster, S., Fell, D. A. & Dandekar, T. A general definition of metabolic pathways useful for systematic organization and analysis of complex metabolic networks. Nat Biotechnol 18, (2000).

73. Schuster, S., Dandekar, T. & Fell, D. A. Detection of elementary flux modes in biochemical networks: a promising tool for pathway analysis and metabolic engineering. Trends Biotechnol 17, (1999).

Literature Review

38 | P a g e

74. Zanghellini, J., Ruckerbauer, D. E., Hanscho, M. & Jungreuthmayer, C. Elementary flux modes in a nutshell: properties, calculation and applications. Biotechnol. J. 8, 1009–1016 (2013).

75. Klamt, S. & Gilles, E. D. Minimal cut sets in biochemical reaction networks. Bioinformatics 20, 226–234 (2004).

76. Klamt, S. & von Kamp, A. An application programming interface for CellNetAnalyzer. Biosystems 105, 162–168 (2011).

77. Lewis, N. E., Nagarajan, H. & Palsson, B. Ø. Constraining the metabolic genotype-phenotype relationship using a phylogeny of in silico methods. Nat. Rev. Microbiol. 10, 291–305 (2012).

78. Klamt, S. & Mahadevan, R. On the feasibility of growth-coupled product synthesis in microbial strains. Metab. Eng. 30, 166–78 (2015).

79. von Kamp, A. & Klamt, S. Enumeration of Smallest Intervention Strategies in Genome-Scale Metabolic Networks. PLoS Comput. Biol. 10, (2014).

80. Kolb, S. & Stacheter, A. Prerequisites for amplicon pyrosequencing of microbial methanol utilizers in the environment. Frontiers in Microbiology 4, (2013).

Characterization of Mutant Strains of E. coli in an Electrochemical Bioreactor

This chapter has been drafted into a manuscript that will be submitted to Microbial Cell Factories as a technical note.

Abstract

E. coli has been shown in the past to respond to growth in the presence of a reducing potential by

modulating its fermentation profile. Microbial electrosynthesis attempts use that electrical energy in

the presence of carbon dioxide to increase product yields. In E. coli, neutral red has been well studied

as the mediator to carry the charge from the electrode to the cell. In this study, we specifically examine

how mutant strains of E. coli behave when grown with neutral red in the presence of a reducing

potential. The results suggest that neutral red is not an efficient method for charge transfer for the

purposes of increasing succinate yields. While the wild-type strain showed an average improvement

of 89% in succinate yields, further characterization of mutant strains showed much less improvement.

The increased titres were found to be linked to the redox state of the cell as opposed to extracellular

electron transfer.


40 | P a g e

3.1 Introduction and Background

The fermentative metabolism of E. coli has been well documented to be modulated by external factors

in its environment1. Among the diverse stimuli that cells respond to, the direct or indirect supply of

electrons by an electrical current via a cathode in an electrochemical cell is a special type2.

Supplementing microbes with an external source of energy and reducing power has been suggested,

and demonstrated as an effective method to change fermentation product profiles, increase biomass

yield and drive the production of desired chemicals beyond the purpose of these findings3–7. In this

regard, there has been considerable interest in this technique because of its potential to effectively

channel extracellular electrons to internal cellular electron carriers including NADH, NADPH or

FADH. The ability to generate reduced electron carriers internal to the cell from an external electrical

current offers the ability to efficiently drive metabolic processes that fix carbon dioxide8,9.

To that end, several studies have reported the use of electrical currents to increase product

yields10. In particular, the use of neutral red as a mediator to deliver an electrical current to Actinobacillus

succinogenes was shown to increase succinate titres at the end of the fermentation4. This early study was

one of the more detailed of its time, and provided a basic framework and theoretical model of

understanding the impact of electrical current on metabolism. Other work in the field has focused on

a diverse group of organisms that have a natural ability to interact with a charged surface. These

organisms include diverse species such as Geobacter11, Shewanella12–15 and Acetobacter16–18, and work has

demonstrated the ability for cells to accept electrons and produce a variety of products including

organic acids. In these organisms, extracellular electron transport occurs via membrane bound

cytochromes and often other charge carrying proteins such as pili that are exposed on the cell

membrane surface19. They conduct electrons across the periplasmic space that give them the ability

to both donate and accept electrons from an electrode.

These specialized membrane structures are absent from the model and more traditional

workhorse metabolically engineered for chemical production, E. coli. However, many of these proteins

are capable of being functionally expressed in E. coli, but have been successfully been demonstrated

to rewire E. coli as an electricigen20 when expressed heterologously. The heterologous expression of

the Mtr operon from Shewanella allowed cells to couple current production to substrate utilization and

reduce nano-crystalline iron. Despite this substantial achievement, there have yet to be any studies


41 | P a g e

demonstrating direct electron transfer to E. coli that reduces NAD+ for the purpose of metabolic

engineering applications. Hence, studies related to the effects of microbial electrosynthesis on E. coli

have been limited to using mediators such as neutral red. One of the earliest studies reported using

neutral red as a mediator to drive anode reduction in an electrochemical cell21. A later study

investigated the mechanism that accounted for the physiological changes observed when neutral red

is used to mediate an electrical current to the cell22. By accounting for stoichiometric addition of

electrons and examining the fermentation end-product profiles, they determined that neutral red

caused changes to the metabolism by regulatory means and not through the direct electron transfer to

NAD+. The results were interesting because they were a marked departure from the prevailing

hypothesis that neutral red was able to directly reduce intracellular NAD+ to NADH.

Figure 3-1 Structure of Neutral Red and Menaquinone. Neutral red (Left)

acts as a charge carrying mediator in the fermentation broth. The aromatic ring

structure allows electron transfer to the N(CH3)2 group which becomes reduced

or oxidized. The molecule is embedded in the cell membrane and charge is

mediated to the menaquionone pool (Right).

However, given that neutral red was still capable of reducing the cell’s menaquinone pool, we

asked, generally, whether cells with disruptions to the fermentative genes would exhibit different

fermentation product profiles when grown under reducing conditions. And secondly, we asked

specifically, would these disruptions help or hinder succinate production in E. coli, which requires a

reduced menaqinone pool. We expect that based on the ability of neutral red to reduce menaquione,

the reaction stoichiometry would be:

NRox + 2 e- + 2 H+ → NRred

NRred + MQox → NRox + MQred

MQred + fumarate + 2 H+cyt + 2 H+

periplasm → MQox + succinate + 2 H+cyt

Thus a two electron pair reduction of neutral red should correlate to a production of 1 mol of

succinate from fumarate. Hence, in this short study we characterize ldhA and adhE mutant strains of

E. coli to explore the idea of using E. coli as a host for microbial electrosynthesis of succinate. Hence,

https://www.google.ca/url?sa=i&rct=j&q=&esrc=s&source=images&cd=&cad=rja&uact=8&ved=0ahUKEwiJpsrE4czUAhVDyj4KHc11BtoQjRwIBw&url=https://commons.wikimedia.org/wiki/File:Neutral_red.png&psig=AFQjCNG0szdfTi3KKYo19pi6Q6H_h6Hrcw&ust=1498059719720661


42 | P a g e

while it is well known that cells are able to modulate their fermentation end products in response to

external electrons and even an external redox potential, it has yet to be demonstrated whether these

principles are applicable in strains engineered for chemical production. These engineered strains often

operate at the physiological extremes of their wild-type counterparts, exhibiting changes in their redox

ratios, cellular ATP levels and substrate uptake rates. Hence, in this study we explore if cells with

mutations in their fermentative pathways would exhibit the same increase in product yields that is

typically associated with their wild-type counterparts. We find that while neutral red increases

succinate production in wild-type cells, its ability to do the same in ldhA and adhE mutants is

diminished. We hypothesized that this occurs because intracellular NAD pools are in a highly reduced

state. Finally, we found that current transfer to the cell did not appear to correlate with the degree of

reduction of the fermentation products.

3.2 Materials and Methods

3.2.1 Culturing Techniques in Microbial Electrosynthesis Reactors

Pre-cultures were grown in LB rich media in 10 mL test tube cultures overnight and transferred to

anaerobic serum bottles that were sparged with nitrogen to remove dissolved oxygen. 2mL of cell

culture was transferred to 100mL anaerobic serum bottles containing minimal media and 0.4%

glucose. After 24, hours these cells were harvested by centrifugation, re-suspended in 2mL of

phosphate buffered saline and used as inoculum to the bioreactor. Cells were inoculated to the

bioreactor after overnight pre-reduction of the neutral red containing growth medium.

M9 minimal media was used for cultivation in the bioreactor. Neutral red was added to the

culture at a concentration of 10 uM. Anaerobic conditions were maintained by sparging the reactors

with N2 and pH was maintained at 7 with the addition of 3N KOH.


43 | P a g e

Figure 3-2 Configuration of Bioelectroreactor System. (A) Picture showing the configuration of the electrochemical

bioreactor using the Applikon MiniBio500 vessel. (B) Show a schematic of the bioreactor assembly. The Ag/AgCl

reference electrode was not sterilized by autoclave in the assembly. It was removed from 6N NaCl solution, sterilized with

an isopropanol wipe and aseptically inserted into the assembly once the reactor and its contents had cooled following

sterilization.

3.2.2 Microbial Electrosynthesis Reactors

The electrochemical reactor were constructed from Applikon MiniBio500 fermentation vessels and

headplate. The anodic chamber built using 25mm flat-width dialysis tubing (Spectrum™ Spectra,

Fisher Scientific). One end of the tubing was knotted to seal it. Into the other end of the tubing, a

rubber stopper was inserted with a 5 mm hole. The tubing was fixed in place by using an autoclavable

zip-tie. Prior to fixing in place, the anode electrode was inserted into the dialysis tubing, and the

electrical lead was fed through the 5mm hole at the end of the stopper. Also inserted into the hole

was a 5 mm x 1 cm autoclavable plastic tube. The tube was inserted from underneath the bioreactor

headplate through the dissolved oxygen port until the top of the topper was flush with the headplate.

The entire setup, with containing the bag, electrode, stopper was held in place by compression fitting

using an O-ring and nut.

The Ag/AgCl reference electrode (BASi) was inserted into a 5mm headplate port and secured

using a nut and O-ring. The lab-built working electrode was made from a polished graphite plate (10

mm by 5 mm by 80 mm). The counter electrode was identical to the working electrode. The working

electrode was submerged in the fermentation media and the wire fed through one of the acid-addition

A B


44 | P a g e

ports at the top of the head plate. Electrochemical reactors were connected to a multi-channel

potentiostat (VMP20 Multi Potentiostat, Bio-Logic Science Instruments). Electrode potential at the

cathode was maintained at -600mV relative to Ag/AgCl. EC-Lab V10 software (Bio-logic Inc.) was

used to control reactor potential. Reactors were sparged with N2 to maintain anaerobic conditions and

the fermentation chamber (cathode) was stirred using an impeller at 500 rpm. Growth conditions

were maintained at 37°C and pH was controlled at 7 using 3 N KOH.

3.2.3 Analytical Methods

Analysis of fermentation production was measured via high performance liquid chromatography

(HPLC). We used an Aminex 87H cation exchange column with 5 mM H2SO4 as the mobile phase

at a flowrate of 0.4 mL/min at 50°C. Organic acids were detected at 210 nm. The injection volume

was 20 µL.

3.2.4 Calculations

Growth rate, µ, was calculated by measuring OD600 in the linear range of a spectrophotometer

(GENESYS™ 20 Visible). Charge transfer to the electrochemical bioreactors were determined from

the Bio-logic EC-Lab software used to control the Bio-logic VMP20 Multi-channel potentiostat. The

VMP logged cell current every 120s. Total charge, Q, was determined as the integral of the current

transferred over the batch time and converted to mmol e-. γ, defined as the change in the degree of

reduction between the standard and electrical conditions was defined according to the following

equation:

𝛾 = ∑ 𝜀𝑖𝑥𝑖

𝑒𝑙𝑒𝑐𝑡𝑟𝑖𝑐𝑎𝑙

− ∑ 𝜀𝑖𝑥𝑖

𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑

where 𝜀𝑖 is the degree of reduction of each fermentation product and 𝑥𝑖 is molar yield of that

fermentation product. 𝛾 represents the total change in the degree of reduction of the fermentation

products and thus captures whether more or less electrons are present in the products normalized to

the total glucose consumed at the end of the batch.


45 | P a g e

3.2.5 Strains Used in this Study

Table 3-1 Strains Used for Microbial Electrosynthesis Studies

Genotype Source

E. coli BW21153 wild-type Coli Genetic Stock Center, Yale

BW21153 ∆ldhA::km Baba et al, 2006

BW21153 ∆adhE::km Baba et al, 2006

3.3 Results

3.3.1 Construction of Novel Bioreactor

Our first objective was to establish a working system for microbial electrosynthesis and to identify the

physiological response of neutral red mediated electrosynthesis. To simplify our experimental system,

we used an innovative approach to bioreactor design that allowed us to leverage the existing

fermentation vessels that we use for traditional fermentation studies with a little modification. We

used an Applikon miniBioreactor (500 mL) and used existing DO connections and acid/base addition

ports on the head plate to connect the electrodes to a multi-channel potentiostat. The largest ports,

typically used for dissolved oxygen probes were used, instead, to connect to a dialysis tubing bag that

was submerged in the fermentation vessel. The 1kD dialysis tubing contained the graphite counter

electrode, and provided a barrier to the electroactive cells with neutral red. The dialysis tubing was

submerged into the working volume of the cathode compartment (the reactor vessel). Figure 3-1a

and 3-1b show the electrochemical reactor setup.

3.3.2 Wild-type cells for succinate production

To understand the physiological response of fermenting E. coli cells in our system, we grew wild-type

BW21153 cells anaerobically under a reducing potential (polarized at -600 mVAg/AgCl). This

provided us with a baseline for yield that we could compare future experiments against and allowed

us to determine whether our bioreactor design was suitable for electrosynthesis applications.

The results of the fermentation are shown in Table 3-1. As a control, we grew the same strain

under the same reactor conditions, but in the absence of any reducing potential. Final yields of the

fermentation products are shown in Figure 3-2. We found that the end-point yields of ethanol and

succinate increased by 26% and 89% in the wild-type, respectively. This resulted in a net increase in


46 | P a g e

succinate yield of 0.08 mol-succinate/mol-glucose in the strain grown under a reducing potential.

Lactate yields decreased by 77% as a fraction of the fermentation end products. Acetate end-point

titers were not affected by the reducing potential of the electrode. Therefore, more reduced

fermentation products increased in concentration while more oxidized fermentation products

decreased in endpoint concentrations. Acetate, produced largely by a redox independent pathway,

was unchanged. The results are in line with previous studies using of neutral red mediated

electrobiosynthesis4,23. These results indicated that our system can be suitably used for microbial

electrosynthesis, and the cells’ fermentation profiles are affected by a reducing potential. In total,

succinate titers reached 2.1 g/L for the wild-type strain grown under a reducing potential.

As previously determined in the literature22, we also found that stoichiometric accounting of

electron transfer did not account for the shift in fermentation products to a more reduced state. The

cumulative charge transfer to the bioreactor was determined to be 0.036 mmol e-, or 0.018 NADH

equivalents. By comparison, the total shift in reducing equivalents as measured by the degree of

reduction was determined to be 0.35 (electron equivalents/C-mol).

Strain Condition Succinate Lactate Formate Acetate Ethanol µ Q γ

WT Electrical 0.17 0.05 0.94 0.56 0.53 0.34 0.036 0.35

Standard 0.09 0.22 1.12 0.54 0.42 0.37 - -

∆ldhA Electrical 0.11 - 1.01 0.30 0.74 0.26 0.018 0.07

Standard 0.08 - 1.22 0.31 0.72 0.24 - -

∆adhE Electrical* 0.049 1.98 0 0.02 - 0.10 0.023 -.10

Standard 0.045 2.0 0 0.02 - 0.10 - -

Table 3-2 Fermentation summary showing the molar yields of the products. Completed in biological duplicate.

Molar yields were calculated based on the results presented in Figure 3-2. The yields were affected by the gene

deleted. µ - growth rate in h-1, Q – charge transfer in mmol e-, γ – difference in degree of reduction per mol

glucose between electrical and standard conditions. Yields of succinate, lactate, formate, acetate and ethanol in

mol product/mol glucose. *Indicates data from single trial.

3.3.3 Single Mutant Study

We found that, as expected from previous studies, growing cells in the presence of a reducing current

and a charge carrying mediator can shift the fermentation products to a more reduced state. Since the

change in the end-point fermentation profile is likely caused by regulatory changes resulting from a

reduced menaquinone pool and a direct transfer of electrons to the menaquinone pool, we wondered

what the impact would be on succinate production in cells that have disruptions in their fermentation

metabolism, since the final step of the succinate producing pathway is catalyzed by a menaquinone


47 | P a g e

dependent reductase. Since lactate dehydrogenase and alcohol dehydrogenase are common targets

for genetic knockouts for the production of succinate, we characterized E. coli strains that had ldhA

and adhE deleted from their genome. Cells engineered for chemical production, and especially

succinate production, are known to have their co-factor pools shifted to a more reduced state. Hence,

we wondered whether the intermediate strains in which the genes that produce lactate and ethanol are

deleted would enhance or detract from the ability of electrosynthesis to produce more succinate. We

used single gene mutants from the Keio collection. .

The results from these fermentations is summarized in Table 3-1. They show that the ldhA

deficient mutant grown under a reducing potential produced succinate at molar yield 0.11 mol/mol

compared to 0.08 mol/mol for the control strain, without electrical input, a 40% increase. The adhE

deficient mutant showed less than a 10% change in succinate yield compared to the control strain

which had a yield of 0.045 mol/mol. The ldhA strain had a shift in the degree of reduction of the

fermentation products of 0.07 mol e-/C-mol towards more reduced products. Interestingly, while the

adhE deficient strain had similar yields of succinate and lactate under standard and electrically reduced

potentials, the degree of reduction γ of the fermentation products was 0.10 less than the control

conditions that had no reducing potential, despite having a positive current flow into the reactor.

Finally, we found that the total electric charge (Q) delivered to the cells was not demonstrably

correlated with the shift in the degree of reduction.


48 | P a g e

Figure 3-3 Growth Characteristics of cells. Shows the distribution of fermentation products between cells

growing under normal conditions and cell growing under a reducing potential. A clear shift towards the reduced

products ethanol and succinate is seen while less lactate is produced. (Top) Wild-type cells (Middle) Growth

curve wild-type cell growing under a reducing potential. Cumulative charge transferred to cells is shown in

blue. (Bottom) Distribution of fermentation products for ldhA mutant. The error bars represent standard error

of two replicates.

3.4 Discussion

This short study provides insight into the role that microbial electrosynthesis has on effecting the

physiology of E. coli cells with genetic mutations to its fermentative metabolism. Specifically, we

looked at three the strains with and without mutations that are common for anaerobic products, with

a focus on production of succinic acid.


49 | P a g e

In agreement with a recent study22,24, we found that calculating the degree of reduction in the

fermentation end products, the existing mechanisms for electron transfer mediated by neutral red

were insufficient to explain the changes in product profile. Work by Harrington et al. has since then

elucidated the underlying mechanism by which neutral red is able to increase production of reduced

fermentation metabolites. Their model of the mechanism of charge transfer was ascribed to changes

in the regulatory cascade involving arcB and charge transfer directly to the menaquinone pool. In

general, we found the total current delivered to be in line with previous results from the literature24.

Cells absent in the ldhA and adhE genes have a reduced pool of NAD (high NADH/NAD+

ratio). As a consequence, we can examine the role of mediator driven electron transfer on the synthesis

of succinic acid. We present results suggesting that mediator based electron transfer to the

menaquinone pool may show some ability to further drive the synthesis of succinate in engineered

strains. These conclusions are based on the finding that cells known to have a greater pool of reduced

NAD (∆ldhA > WT) continued to show the ability to produce more succinic acid relative to the

controls. However, the relative changes between strains are likely correlated with the internal redox

state of the cell. The increase in succinate in ∆ldhA strains was substantially less than the increase in

succinate in the wild-type cells while the adhE mutant showed negligible change. The total current

delivered to the cells was also not correlated with fermentation end-production concentrations across

the different strains.

Finally, we observed that the total charge delivered to the bioreactors was largely invariant

between the strains and that the total succinate titres were lowest for the ∆ldhA strain and highest for

the wild-type strains. The results reinforce that the total electron recovery is not correlated to the

direct transfer of electrons but likely by a regulatory mediated processes. Additional studies would be

required to conclusively determine the current efficiency of charge transfer to the cell and distinguish

regulatory effects from those of direct electron transfer.

With hindsight, results are expected since efficient production of succinate requires further

genetic manipulation of anaplerotic enzymes to drive flux towards succinate in addition to gene

deletions of competing metabolic pathways. In essence, neutral-red mediated electron transfer to the

menaquinone pool has a limited ability to “pull” flux from fumarate to succinate. Therefore, a further

study of microbial electrosynthesis using an engineered succinate producing strain with high

expression of anaplerotic metabolism in a high CO2 environment would likely be beneficial in


50 | P a g e

elucidating the ability to increase succinate production. A previous study has found that the presence

of fumarate in the fermentation media increased total delivered current28. A fumarate overproducing

strain may have achieved similar results without supplementation.

Organisms in nature are capable of electron transfer by using mediators. Shewanella is one

example of a microbe that secretes flavin molecules to mediate extracellular electron transfer29.

However, other compounds such as humic acids have also been shown to be implicated in a similar

process30. Diversity in the types of humic acids present in nature means that there is also variety in

the mid-point potential of electrons donors for NAD or menaquinone. Hence, a worthwhile future

experiment might to be examine the efficiency of different mediators in reducing either the

menaquinone pool or even NAD directly.

Taken together, the results of this study suggest that mediator driven electrosynthesis may be

promising for delivering current for biochemical production linked to menaquinone oxidoreductases

if the current exchange density can be increased. Neutral red has the ability to be a charge mediator

in cells with elevated NADH/NAD redox ratios. However, the work here raises the possibility that

other mediators generated electrochemically may also be suitable since the reduction of the

menaquinone pool is the driver for changes in fermentation end product concentrations. One notable

mediator is formate which may be suitable for reducing the menaquinone pool25–27. In conclusion,

even in the absence of direct electron transfer to intracellular NAD+, microbial electrosynthesis can

still provide a means to drive flux towards product linked by menaquinone oxidoreductases. However,

any meaningful application would require identification of a mechanism to deliver charge at higher

rates.


51 | P a g e

3.5 References

1. de Graef, M. R., Alexeeva, S., Snoep, J. L. & Teixeira de Mattos, M. J. The steady-state internal redox state (NADH/NAD) reflects the external redox state and is correlated with catabolic adaptation in Escherichia coli. J. Bacteriol. 181, 2351–7 (1999).

2. Tremblay, P.-L., Angenent, L. T. & Zhang, T. Extracellular Electron Uptake: Among Autotrophs and Mediated by Surfaces. Trends Biotechnol. xx, 1–12 (2016).

3. Park, S. M., Sang, B. I., Park, D. W. & Park, D. H. Electrochemical reduction of xylose to xylitol by whole cells or crude enzyme of Candida peltata. J. Microbiol. 43, 451–5 (2005).

4. Park, D. H. & Zeikus, J. G. Utilization of electrically reduced neutral red by Actinobacillus succinogenes: physiological function of neutral red in membrane-driven fumarate reduction and energy conservation. J. Bacteriol. 181, 2403–10 (1999).

5. Ross, D. E., Flynn, J. M., Baron, D. B., Gralnick, J. a & Bond, D. R. Towards electrosynthesis in shewanella: energetics of reversing the mtr pathway for reductive metabolism. PLoS One 6, e16649 (2011).

6. Nevin, K. P., Woodard, T. L., Franks, A. E., Summers, Z. M. & Lovley, D. R. Microbial Electrosynthesis: Feeding Microbes Electricity To Convert Carbon Dioxide and Water to Multicarbon Extracellular Organic Compounds. MBio 1, e00103-10-e00103-10 (2010).

7. Li, H. et al. Integrated electromicrobial conversion of CO2 to higher alcohols. Science 335, 1596 (2012).

8. Kracke, F., Vassilev, I. & Krömer, J. O. Microbial electron transport and energy conservation - The foundation for optimizing bioelectrochemical systems. Front. Microbiol. 6, 1–18 (2015).

9. Rabaey, K. & Rozendal, R. a. Microbial electrosynthesis - revisiting the electrical route for microbial production. Nat. Rev. Microbiol. 8, 706–16 (2010).

10. Peguin, S. & Soucaille, P. Modulation of Metabolism of Clostridium acetobutylicum Grown in Chemostat Culture in a Three-Electrode Potentiostatic System with Methyl Viologen as Electron Carrier. Biotechnol. Bioeng. 51, 342–348 (1996).

11. Strycharz, S. M. et al. Graphite electrode as a sole electron donor for reductive dechlorination of tetrachlorethene by Geobacter lovleyi. Appl. Environ. Microbiol. 74, 5943–7 (2008).

12. Flynn, J. M., Ross, D. E., Hunt, K. A., Bond, D. R. & Gralnick, J. A. Enabling unbalanced fermentations by using engineered electrode-interfaced bacteria. MBio 1, e00190–10 (2010).

13. Shi, L. et al. Molecular Underpinnings of Fe(III) Oxide Reduction by Shewanella Oneidensis MR-1. Front. Microbiol. 3, 50 (2012).

14. Coursolle, D. & Gralnick, J. a. Modularity of the Mtr respiratory pathway of Shewanella oneidensis strain MR-1. Mol. Microbiol. 77, 995–1008 (2010).


52 | P a g e

15. Cordova, C. D. (Stanford U. No Title Molecular Basis of Respiratory Plasticity in Shewanella Oneidensis MR-1. (2010).

16. Straub, M., Demler, M., Weuster-Botz, D. & Dürre, P. Selective enhancement of autotrophic acetate production with genetically modified Acetobacterium woodii. J. Biotechnol. 178, 67–72 (2014).

17. Nevin, K. P. et al. Electrosynthesis of organic compounds from carbon dioxide is catalyzed by a diversity of acetogenic microorganisms. Appl. Environ. Microbiol. 77, 2882–6 (2011).

18. Marshall, C. W., Ross, D. E., Fichot, E. B., Norman, R. S. & May, H. D. Electrosynthesis of Commodity Chemicals by an Autotrophic Microbial Community. Appl. Environ. Microbiol. (2012). doi:10.1128/AEM.02401-12

19. Mahadevan, R., Palsson, B. Ø. & Lovley, D. R. In situ to in silico and back: elucidating the physiology and ecology of Geobacter spp. using genome-scale modelling. Nat. Rev. Microbiol. 9, 39–50 (2011).

20. Jensen, H. M. et al. Engineering of a synthetic electron conduit in living cells. Proc. Natl. Acad. Sci. (2010). doi:10.1073/pnas.1009645107

21. Park, D. H. & Zeikus, J. G. Electricity generation in microbial fuel cells using neutral red as an electronophore. Appl. Environ. Microbiol. 66, 1292–7 (2000).

22. Harrington, T. D. et al. The mechanism of neutral red-mediated microbial electrosynthesis in Escherichia coli: menaquinone reduction. Bioresour. Technol. 192, 689–695 (2015).

23. Mi, S. U. N., Kang, H. Y. E. S. U. N., Park, D. A. E. W. O. N. & Park, D. O. O. H.

Electrochemical Control of Metabolic Flux of Weissella kimchii sk10 : Neutral Red Immobilized in Cytoplasmic Membrane as Electron Channel. 15, 80–85 (2005).

24. Harrington, T. D. et al. Neutral red-mediated microbial electrosynthesis by Escherichia coli, Klebsiella pneumoniae, and Zymomonas mobilis. Bioresource Technology (2015). doi:10.1016/j.biortech.2015.06.005

25. Harris, D. M., Van Der Krogt, Z. A., Van Gulik, W. M., Van Dijken, J. P. & Pronk, J. T. Formate as an auxiliary substrate for glucose-limited cultivation of Penicillium chrysogenum: Impact on penicillin G production and biomass yield. Appl. Environ. Microbiol. 73, 5020–5025 (2007).

26. Berríos-Rivera, S. J., Bennett, G. N. & San, K.-Y. Metabolic Engineering of Escherichia coli: Increase of NADH Availability by Overexpressing an NAD+-Dependent Formate Dehydrogenase. Metab. Eng. 4, 217–229 (2002).



53 | P a g e

28. Park, D. H., Laivenieks, M., Guettler, M. V., Jain, M. K. & Zeikus, J. G. Microbial utilization of electrically reduced neutral red as the sole electron donor for growth and metabolite production. Appl. Environ. Microbiol. 65, 2912–2917 (1999).

29. Marsili, E. et al. Shewanella secretes flavins that mediate extracellular electron transfer. Proc. Natl. Acad. Sci. U. S. A. 105, 3968–3973 (2008).

30. Zhou, S., Chen, S., Yuan, Y. & Lu, Q. Influence of Humic Acid Complexation with Metal Ions on Extracellular Electron Transfer Activity. Sci. Rep. 5, 17067 (2015).

Engineering Utilization of Formate in Escherichia coli

54 | P a g e

Parts of this chapter were submitted to Biochemical Engineering Journal in an article titled: Microbial and Electrochemical Routes for the Production of Chemicals from Carbon dioxide.

Abstract

The modern day bioprocess for making fuels and chemicals has as its input a carbon feedstock that is

derived from a renewable carbon source from which cells derive their energy. Microbial

electrosynthesis seeks to replace this conventional electron donor in favour of one that is derived

electrochemically. In this work we study the ability for E. coli to assimilate formate as a carbon source

for growth by expressing formate tetrahydrofolate ligase (fhs). We find that cells are not capable of

utilizing formate in a wild-type background. However, by engineering a formate auxotrophic strain

through a serA, gcvT double mutant, we show that formate is required for biomass synthesis. The

growth rate of the mutant strain was determined to be 0.33 h-1 ± 0.004 compared to the wild-type µ

= 0.46 h-1 ± 0.04. Further thermodynamic analysis of the formate utilization pathway suggested a

Max-min driving force (MDF) of less than 0.5 kJ/mol. Overall, these results lead us to conclude

that the reductive glycine pathway may be inefficient for supporting growth without significant

adaptive laboratory evolution and additional pathways for supporting production of NAD(P)H from

formate.


55 | P a g e

4.1 Introduction

Methods for transferring reducing equivalents are not limited to direct electron transfer from a

cathode1. Organic substrates, which are derived from the electrochemical reduction of CO2, have

been suggested as potentially overcoming the challenges associated with the expression of functional

electrical conduits in E. coli and low current exchange densities2–4. Formic acid is an example of a

mediator that can be derived from electrochemically reduced CO2.5,6 And indeed, it has been

demonstrated that production of succinate is increased in the presence of formate and the expression

of formate dehydrogenase3,7 because of its ability to supply electrons. It has also been demonstrated

for a variety of other products and applications8–11. Hence, there appears to be some viability in the

method of using formate as a charge carrying mediator that transfers electrons to NAD+ and the leaves

the cell as CO2. The merits of this approach, as opposed to direct electron transfer, lies in the

perceived difficulty of expressing membrane bound proteins as opposed to soluble cytoplasmic

proteins for mediator based electrosynthesis2.

E. coli however lacks any natural pathways for assimilation of formate as a growth substrate. Moreover,

while the CO2 produced can be assimilated by phosphoenolpyruvate carboxylase, this reaction has

limited capacity towards supporting cell growth in the absence of any traditional carbon sources such

as glucose. This occurs because the viability of carbon fixation via phosphoenolpyruvate carboxylase

is directly correlated to the requirement to generate phosphoenolpyruvate from glucose in a one to

one ratio since phosphoenolpyruvate cannot be regenerated by a cycle. This requirement to regenerate

the starting molecule of a carbon fixing pathway important because it underscores a common feature

of all natural carbon fixing pathways with the exception of the Wood-Ljungdahl pathway: they are

cyclical in nature12–14. Cycles require the coordinated expression of many enzymes in the pathway to

balance flux within the cycle. Another complication that arises from this task is that many of these

pathways go through highly connected and tightly regulated metabolite pools such as serine15 or

pyruvate16–19 which has implications for the metabolic engineering of substrate utilization pathways.

To date, heterologous expression of fully functioning cycles in the absence of another carbon and

electron donor has not been demonstrated. Hence, it is hypothesized linear pathways are, as a starting

point, much more amenable to heterologous expression of carbon fixing pathways.


56 | P a g e

In Chapter 3 I examined the feasibility and identified the limitations of using direct electrical current

and increase the yields of succinic acid, a compound that utilized CO2 as a substrate. The efficient

delivery of electrons to the cell is necessary to drive carbon fixation. To that end, the study of formate

as a mediator is useful because it can be generated in electrochemical cells when carbon dioxide is

reduced. This section describes work performed in attempting to engineer E. coli to use formic acid

as a carbon source. In concluding this chapter, the significant challenges are discussed related to the

assimilation of these mediators. Recommendations are made as to the next steps required.

Figure 4-1 The formate utilization pathway. First suggested by Bar-Even et al, for its supposed superior thermodynamics and its linear topology, this pathway was investigated as the route to support growth using formate alone. The sole enzyme non-native to E. coli is the formate tetrahydrofolate ligase (Fhs) catalyzing the reaction inside the green box. Formate is assimilated by this pathway to produce serine as the last metabolite. Serine can be used by the cell as a carbon and nitrogen source. The pathway requires 1 NADH, 2 NADPH and 2 ATP. Naturally, serine biosynthesis occurs from 3PG where glycolytic flux is channeled through this pathway. In many organisms this pathway can provide a substantial portion of ATP and NADPH requirement and is known as the SOG (serine, one-carbon cycle, glycine synthesis) pathway. Here, the carbon flow is reversed.

4.2 Results

4.2.1 Engineering Formate Assimilation by Formate Activation

The first metabolic step towards engineering formate assimilation is its activation by ligating it to a

high energy tetrahydrofolate cofactor (Figure 4-1). This reaction is carried out reversibly by an enzyme


57 | P a g e

known as formate tetrahydrofolate ligase (Fhs) which requires ATP to drive the reaction forward. To

engineer this first step and establish in vivo formate consumption, two Fhs enzymes were taken from

Staphylococcus aureus and Clostridium ljungdahlii. The genes were then cloned in the plasmid pTrc99a and

transformed into wildtype (WT) E. coli MG1655. The strains were then grown in the presence of

approximately 20 mM formate and 0.4% glucose supplemented with 0.2% yeast extract, aerobically.

However, rather than consuming formate, it was observed strains carrying the fhs gene produced small

quantities of formic acid. The results of this assay are summarized in Table 4-1. The total

concentration of formate increased as much as 10% for the strain with the Clostridium ljungdahli fhs

gene. The results suggested that the favoured in vivo direction of the reaction was the cleavage of

methylene-tetrahydrofolate to produce the THF carrier compound and formate. It was hypothesized

that since naturally glycolytic flux is directed towards serine and glycine biosynthesis15, then the

thermodynamic driving force this pathway in vivo is towards the cleavage of glycine. Indeed, it has

been suggested in several studies that fhs serves an important role in ATP generation20–22. Hence, in

the next steps I attempted to re-engineer the folate the metabolism of E. coli to redirect the driving

force from glycine cleavage to formate assimilation.

Strain Formate Concentration

to t = 24 hrs t = 48 hours

pTrc99a vector 22 ± 0 mM 22 ± 0 mM 22 ± 0 mM

SAV1732+ 22 mM 22 mM 23 mM

YP00293+ 22 mM 24 mM 25 mM

Table 4-1 Formate concentrations at 24 and 48 hours of the batch. Formate was found to increase in strains carrying the fhs gene (single experiment from biological duplicates). Control strains carrying the empty vector showed to increase in formate through the batch (duplicate). This data was additionally supported by multiple experiments across an expanded biological data set encompassing a total of four different fhs enzymes that were tested (See Raw data section) showing no utilization.

4.2.2 Rewiring Folate Metabolism by Deletion of Serine Biosynthesis

Pathways

The production of formate during the expression of fhs was unexpected since we were hoping to

observe formate consumption. To overcome this challenge we employed a series of genetic

interventions to reroute cellular flux.

Deleting D-3-phosphoglycerate dehydrogenase (serA) blocks glycolytic flux into the serine and glycine

biosynthesis pathways and turn E. coli into a serine auxotroph (Figure 4-3). Thus, without either serine


58 | P a g e

or glycine, the cell is incapable of growth. This deletion has a second role as well. By blocking

glycolytic flux to serine biosynthesis, it was thought that it might be possible to create a

thermodynamic driving force in the direction of formate activation and its assimilation since it is

expected that intracellular concentrations of the serine and glycine pools would decrease. Therefore

the serA mutant was transformed with the fhs containng plasmid from Staphylococcus aureus and grown

in the presence of 2 mM glycine in M9 minimal media. Glycine, it was hypothesized, could be

converted to serine once formate in the media entered the cell, was activated to 10-fTHF and ligated

to glycine after by glyA.

However, when the cells were grown in minimal media supplemented with glycine, they showed no

growth. There were two causes that were hypothesized for no growth. (1) Without pathways for

serine biosynthesis and in the presence of high protein demand from the pTrc plasmid and strong trc

promoter, there was an abnormally high protein burden on the cell. (2) Expression of fhs was too high

that it tied up all free THF co-factors to their bound state as 10-Formyl-THF (10-fTHF). The total

size of the folate co-factor pools in E. coli is approximately 50 µM. Hence, without sufficient

expression of the pathway enzymes, serine biosynthesis from glycine which requires reduced 10-fTHF

could not proceed, resulting in a serine and glycine auxotroph. Indeed, the serine glycine biosynthesis

is a highly regulated system at the transcriptional level and at the protein level to ensure that glycine

cleavage and C1 unit production is balance15. For example, overexpression of the glycine cleavage

system (GCV) is reported to create partial glycine auxotrophs23.

4.2.3 Addressing Cell Regulation and Development of Formate Assay

The glycine cleavage system and the serine biosynthesis pathways are a highly regulated node of the

amino acid metabolism of the cell. To address the challenges above, a straight-forward approach was

sought that could be used to engineer formate assimilation. Two strategies were developed.

The first was based on the hypothesis that the overexpression of fhs was converting free THF into an

unusable form for the cell. The goal was then to increase the supply of free THF by deleting the gene

purR, a repressor gene that controls the purine nucleotide biosynthesis pathway and synthesis of the

GCV operon24,25. Hence, deletion of purR it was believed would increase biosynthesis of the glycine

cleavage system to overcome low problems associated in low expression. However, the ∆serA-

∆purR::cm double mutant strain expressing the fhs gene from S. aureus also showed no growth. Hence,


59 | P a g e

it was determined that rather than targeting expression levels of individual genes one by one, it was

better to develop a formate auxotrophy and screen for growth. Hence, a double mutant was created.

The ∆serA-∆gcvT strain requires glycine and formate for growth in minimal media to generate the

necessary 10-fTHF co-factor that can be used to synthesize serine when grown with glycine in the

media. This approach is summarized in Figure 4-2.

Figure 4-2 Strategy showing the approach to engineering formate utilization in E. coli. Initial approach relied only on deleting the serA gene that converts 3PG to 3PHP resulting in a serine auxotrophy. In a second approach, the glycine cleavage system was disrupted through the deletion of the gcvT gene resulting in a requirement for glycine and serine. Conversion of formate to 5,10-mTHF could then complement the serine auxotrophy in the presence of glycine confirming functional expression the formate utilization pathway.

Initial experiments resulted in no cellular growth. Therefore, the strain was adapted by serial dilution

from minimal media initially supplemented with 0.2% yeast extract, 1mM IPTG, 20mM formate and

2mM glycine. The final media composition contained no yeast extract and 0.2 mM IPTG and 20 mM

formate and 2 mM glycine. Figure 4-3 shows the growth curve for the wild-type and the double

mutant utilizing formate as a rescue carbon source. The inset in Figure 4-3 shows the OD

measurements for the control strains with only glycine and only formate, and no glycine or formate

that showed no growth after 24 hours.

4.2.4 Modelling Formate Pathway Using a Lumped Kinetic Model


60 | P a g e

Having established a screen for formate utilization in E. coli, we sought to understand why initial

approaches to simply overexpress formate-THF ligase was insufficient to see a consumption of the

formate in the media and why, unexpectedly increases in formate concentration were observed for

some strains. To investigate this problem a thermodynamic modelling framework was used to

understand the flux directionality of the pathway reactions.

Figure 4-3 Growth based screening for formate utilization. Growth curves for WT strain with pTrc99a (blue) and ∆serA,gcvT::cm pTrc99a-fhs (green). It was possible to engineer formate assimilation by using serine auxotrophy a selection criteria for formate utilization. Growth media contained 20 mM formate and 2 mM glycine. Growth rates were determined to be 0.46 h-1± 0.04 (standard error) for the wild-type strain and 0.33 h-1 ± 0.005 (standard error) for the double mutant. Inset: Controls contained that contained either glycine, formate or no supplement showed no growth after 24 hours (duplicate experiments, absolute mean deviation).

We had an initial hypothesis that the reason for the absence of formate utilization and in some

cases for formate production in the non-auxotrophic strains was either because of a low

[NADPH]/[NADP] ratio which is necessary to drive the pathway flux in the direction of serine

synthesis or that the folate concentration as free [THF]/[5,10-mTHF] was not optimized to support

synthesis of serine from glycine. The standard Gibbs energy change was plotted:

∆𝐺𝑟𝑥𝑛′ = ∆𝐺° + 𝑅𝑇 ∙ 𝑙𝑛𝑄

Formate Glycine to t24

0.036 ± 0 0.026 ± 0.01

x 0.036 ± 0 0.03 ± 0

x 0.034 ± 0.005 0.024 ± 0.01


61 | P a g e

where Q was modelled as the intracellular ratios of either [NADP]/[NADPH] or [Ser]/[Gly] at

constant THF co-factor ratios. The standard free energy values were calculated based using a group

contribution method estimated by eQuilibrator. Two reactions were examined: serinehydroxymethyl

transferase (SHMT) which is the last step and most thermodynamically unfavourable reaction of the

pathway and methylenetetrahydrofolate dehydrogenase with requires NADPH as a co-factor.

Figure 4-4a models the reduction of 10-fTHF to 5,10-mTHF which has a negative standard Gibbs

free energy change. However, while the other steps in the pathway require the folate co-factor pools

to be shifted towards 5,10-mTHF in order to drive the flux in the direction of serine biosynthesis, this

reaction is inhibited by large concentrations of 5,10-mTHF. Hence, to understand this relationship

better the flux through the reaction was modelled for several ratios of the THF co-factor. The results

show that it should remain thermodynamically favourable even when [5,10mTHF]/[10fTHF]

approaches 10 since typical intracellular ratios of [NADP]/[NADPH] are in the range of 0.05-0.226–28

are sufficient to offset high 5,10-mTHF concentrations.

Figure 4-4b models the thermodynamic driving force of the last step (glycine to serine, glyA) in the

reductive glycine pathway and shows the large unfavourable driving force in the direction of serine

biosynthesis. The reaction naturally occurs in the direction of serine cleavage. Positive free energy

values except for the most extreme values of folate or NADPH values suggests this is the bottleneck

of this pathway. To overcome this thermodynamic bottleneck, the relative concentration of glycine

needs to be much higher than that of serine. As can be seen from this figure, the folate ratios

([THF]/[m5,10THF]) needs to be much smaller than one in order to generally support concentrations

of serine greater than glycine. Under physiological conditions, intracellular folates are around 50 µM

with a ratio of 0.12-0.2. In other words, to compensate for a low [THF]/[m5,10THF] co-factor ratio,

the concentration of glycine relative to serine needs to be substantially higher. Given a theoretical

limit in glycine concentration of around 10 µM, this would create an upper limit on serine

concentration of ~7.5 µM. Moreover, since the first step of the reductive glycine pathway also requires

a free THF which is the product of this reaction, it creates a co-factor cycle that needs to be optimized

and well balanced. Although the analysis of Figure 4-4 is sheds light on the role of serine and glycine

concentration as being pivotal for establishing pathway directionality, it is important to note that the

co-factors are shared across various enzymes of this pathway. Hence the interplay between the driving


62 | P a g e

forces of all participating reactions, which is not accounted for by simple Gibbs free energy

calculations in Figure 4-4, is an important consideration as well.

To deconvolute the various effects of the different metabolites and co-factors from the pathway, a

global analysis was applied to the pathway to understand the maximum thermodynamic driving force

for the pathway that could be expected under physiologically relevant metabolite concentrations. This

analysis was performed by solving a linear programing problem that maximizes the ∆G of the least

favourable reaction of the driving force constrained by a set of feasible metabolite concentrations.

The results are shown in Figure 4-4.

Figure 4-4 Thermodynamic analysis of THF co-factor utilizing reactions of the Reductive Glycine Pathway. (A) Shows the change in Gibbs free energy as a function of the intracellular NADP(H) redox ratio for various concentrations of α represents the ratio [5,10mTHF]/[10fTHF]. (B) Shows the change in free energy for the reaction catalyzed by folD that converts glycine to serine. α represents the ratio [THF]/[5,10mTHF].

The results in Figure 4-4a show that metabolic pathway is actually thermodynamically unfavourable

for all but the most extreme conditions where the free NADPH concentration needs to be at least 15

fold greater than the free NADP concentration and the cell needs to be growing in a ATP rich

environment with an ATP concentration ~13 fold greater than ADP. Figure A-4b shows the

concentrations used as constraints by the model. The results show that under the most feasible set of

conditions, the concentration of the folate co-factors are at their physiological limits. Together, these

results suggest that the in vivo application of this pathway has substantial limitations as a replacement

for glucose. It also partially explains why the difficult we had earlier in engineering the cells to utilize

A B


63 | P a g e

formate and lends insight into why cells use the reductive glycine pathway to generate ATP at the

expense of carbon from glycine as opposed to utilizing it as a method for assimilating carbon.

Figure 4-5 Calculation of the max-min driving force (MDF) for the reductive glycine pathway. A positive MDF indiciate the pathway is favourable in the direction specified. (A) Shows the max-min free energy change for the pathway under various ATP and NADPH concentrations. Both are necessary to drive flux through the pathway. (B) Shows the concentrations that were used to determine the max-min driving force. Blue circles are the concentration at the most thermodynamically favourable conditions. Several metabolites need to be present at their most physiologically extreme concentration for the pathway to function in the serine synthesis direction.

4.2.5 Measuring Intracellular Concentration of Energy Metabolites and

Cofactors

Work towards validating the thermodynamic prediction of the earlier model was begun by developing

and testing a protocol for measuring the intracellular concentration. While not statistically significant

A

B


64 | P a g e

since these were not done it replicate, the preliminary metabolomics do show difference between the

test and control strains in ATP/ADP ratio. The results seem to support modelling showing that

intracellular pools of are shifted towards NADPH in the strain containing formate tetrahydrofolate

ligase. It is speculated that the increase in ATP is necessary to drive the first step of the pathway.

Figure 4-6. Raw mass-spec data showing changes peak intensity of the intracellular redox co-factors NADP(H) and phosphate co-factors ATP, ADP and AMP can be detected by sampling methodology. The methodology was not able to determine concentrations of serine or glycine which had masses that were too small to detect accurately. The data shows a clear shift in the NADPH concentration relative to NADP for the strains containing fhs. There is also a shift in ATP relative to ADP and AMP for the formate auxotroph. F – wildtype, 3 – wildtype pTrc99a-fhs, A - ∆serA,gcvT::cm pTrc99a-fhs

4.3 Discussion

Experimental evidence based on auxotrophic selection suggests that formate can be assimilated by E.

coli as a carbon source by ligating the C1 moiety onto glycine and producing serine. Hypothetically,

serine can then be deaminated to generate pyruvate which can feed into the central carbon metabolism.

Since initial screening of four different fhs enzymes showed no formate consumption by HPLC, an

auxotrophic approach was utilized that was eventually successful. This solution is attractive because

it offers a potentially useful methodology for selecting for a complete pathway in E. coli to utilize

formate.

However, it is worthwhile noting that in silico thermodynamic modelling of the pathway suggests

several bottlenecks that represent a barrier for establishing the reductive glycine pathway for biomass

formation. Specifically, the modelling reveals that the thermodynamic bottleneck, the ligation of

5,10m-THF onto glycine to produce serine, requires that most of the metabolites in the cell be present


65 | P a g e

at their physiologically extreme concentrations. This is especially true of the THF co-factors as well

as the [NADH]/[NAD] ratio which was determined to be four under the most favourable conditions.

By comparison, this ratio typically hovers between 1-10-1 under physiologically relevant conditions.

One way around this might be to engineer the specificity of the glycine cleavage system for NADPH

instead of NADP which is known to exist in a more reduced state. Formate dehydrogenase coupled

with a transhydrogenase could then be used to reduce NAD to NADH and subsequently generate

NADPH.

The reductive glycine pathway is one of several pathways in the literature that has been suggested for

engineering synthetic growth on formate. The formalose pathway is one additional example30. In that

pathway, the formate assimilation was demonstrated by in vitro proof-of-concept through the

reduction of formate to formaldehyde, and its subsequent condensation to form dihydroxyacetone.

While a thermodynamic analysis of the pathway was performed by the researchers and showed an

MDF greater than 12kJ/mol, poor enzyme kinetics made in vivo validation of the pathway not possible.

The results suggest that the poor kinetics of individual enzymes creates a bottleneck that substantially

hinders total pathway flux. Interestingly, while the reductive glycine pathway suffers from a poor

MDF, it is worthwhile noting that a high MDF such as the one for the formalose pathway, is also not

sufficient to improve pathway flux and that enzyme engineering for more active proteins may be

needed to support growth in all cases.

The thermodynamic analysis also suggested that efficient utilization of formate by the reductive glycine

pathway requires co-factors such as folate to be at their extreme concentrations. Folates, such as those

in the form of THF, are synthesized from chorismate, which is converted to pABA (Figure 4-6). To

improve the flux through the reductive glycine pathway, further pathway engineering targeting the

concentrations of the co-factor pools is suggested. For example, it was shown that by overexpressing

the panB gene in E. coli, the total folate pools could be increased by 46%31. Overexpressing the

shikimate biosynthesis pathway to produce more chorismate might also have similar affects to increase

the pools of precursor metabolites. Together, targeting these genes in the secondary metabolism

might help to address some of the co-factor limitations so as to improve the flux through formate

utilization pathway.


66 | P a g e

Figure 4-6 The classical THF synthesis pathway and a hypothetical folate-cleaving side-reaction mediated by PanB. This figure was copied from Thiaville et al, 2016 under the Creative Commons License.

4.4 Conclusions

Together the experimental work along with the computational thermodynamic modelling suggests

that while an adaptive laboratory evolution approach might be able to produce an E. coli strain capable

of converting formate to biomass, that they pathway is not as attractive for the biological production

of chemicals and fuels.


67 | P a g e

4.5 Material and Methods



(HPLC). We used an Aminex 87H column with 5 mM H2SO4 as the eluent and a flowrate of 0.4

mL/min at 50°C. Organic acids were detected at 210 nm. Cell densities of the cultures were

determined by measuring optical density at 600 nm (GENESYS™ 20 Visible). Cell density samples

were diluted as necessary so as to fall within the linear range. A differential refractive index detector

(Agilent, Santa Clara, CA) was used for analyte detection and quantification. Yields were calculated

between two time points, whereas the cumulative yield was calculated between the initial and final

measurements.

4.5.2 Plasmids and Strains

The fhs genes were cloned from three organisms Clostridium perfringens (GHAW-2805), Staphylococcus

aureus (SAV1732), Clostridium ljungdahlii (YP_003781893). A fourth fhs gene from Moorella thermoacetica

(MOTH0109) was codon optimized and ordered as gBlocks in three parts and assembled using

Gibson Assembly. Cloned fragments ligated to a pTrc99a plasmid by restriction endonucleases NcoI

and PstI and ligated into pTrc99a treated with the same endonucleases. pTrc99a’s native RBS was

used to drive expression of the genes.

4.5.3 Media and Cultivation Conditions

Cells were grown using lysogeny broth (LB) as per manufacturer’s instructions (Bioshop, Burlington,

ON) for all strain construction and fermentation pre-cultures. When characterizing strains, cell were

grown under M9 minimal media with the following compositions: 1.0 g/L NH4Cl, 3.0 g/L KH2PO4,

6.8 g/L Na2HPO4, 0.50 g/L NaCl. Supplements of yeast extract were added to minimal media ad

described. Glucose was used as the carbon source as concentrations described in the text. IPTG was

used at a concentration of 1mM when necessary or as described. A trace metal solution was prepared

according to the following composition prepared in 0.1 M HCl per litre and added at a concentration

of 1/1000: 1.6 g FeCl3, 0.2 g CoCl2•6 H2O, 0.1 g CuCl2, 0.2 g ZnCl2•4H2O, 0.2 g NaMoO4, 0.05 g

H3BO3. 1 M MgSO4 and 1 M CaCl2 was also added to the media at a concentration of 1/500 and

1/10,000, respectively. For all cultures, carbenicillin was added as appropriate at 100 µg/mL. Cells


68 | P a g e

were grown in 250 mL shake-flasks. 1 M Na-formate was used as a stock solution and diluted to the

appropriate concentration as described in the text.

4.5.4 Max-min Driving Force Thermodynamic Modelling

Thermodynamic modelling was carried out using the framework provided by Noor and co-workers29.

Briefly, the following linear program was solved in Matlab (Mathwork, 2015).

𝐺𝑖𝑣𝑒𝑛 𝑆, 𝐺°, 𝑅𝑇, 𝐶𝑚𝑖𝑛, 𝐶𝑚𝑎𝑥

𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒 𝐵

𝑆𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 − (∆𝐺° + 𝑅𝑇𝑆𝑇 ∙ 𝑥) ≥ 𝐵

ln(𝐶𝑚𝑖𝑛) ≤ 𝑥 ≤ 𝐶𝑚𝑎𝑥

In the above program, B represents a tight lower bound (i.e., the minimum) on the driving force of all

reactions. The solution to the problem yields B, which is defined as the Max-min Driving Force of

the pathway in kJ/mol. When B is maximized, all possible reactions are as far from equilibrium as

possible within the defined concentration ranges.

4.5.5 Sampling Methodology for Mass-Spec

1. Bacterial cultures were grown in 250 mL baffled flasks.

2. 10 mL of sample was collected as spun down in 15 mL falcon tubes at room temperature, 5000g.

3. The supernatant was removed and cell were immediately quenched with in 1.5mL of -20°C

extraction solution in 15mL falcon tube, vortex and place at -20°C for 30min.

4. The extraction solution was transferred to 2mL eppendorf and spin at max speed, 4°C for 5min.

5. The supernatant was transferred to new 2mL eppendorf (store at -20°C). The pellet was

resuspended in 300uL of fresh extraction solution and placed at -20°C for 30min. It was then spun

at max speed, 4°C for 5min.

6. The supernatant was added to the previous 2mL eppendorf (store at -20°C), and the pellet was

suspended in 200uL of fresh extraction solution, placed at -20°C for 30min. It was then spun at max

speed, 4°C for 5min.

7. Add supernatant to previous 2mL eppendorf, add 7uL Ammonium hydroxide per mL of extraction

solution and mix thoroughly (to neutralize).


69 | P a g e

8. Split into two equal fractions and store at -80°C overnight.

9. Concentrate each sample under vacuum and re-suspend in 100uL of H2O.

Extraction solution: 0.1M formic acid in MeOH:Acetonitrile:H2O (40:40:20)

4.6 References

1. Li, H. & Liao, J. C. Biological conversion of carbon dioxide to photosynthetic fuels and electrofuels. Energy Environ. Sci. 6, 2892–2899 (2013).


3. Berríos-Rivera, S. J., Bennett, G. N. & San, K.-Y. Metabolic Engineering of Escherichia coli: Increase of NADH Availability by Overexpressing an NAD+-Dependent Formate Dehydrogenase. Metab. Eng. 4, 217–229 (2002).

4. Yishai, O., Lindner, S. N., Gonzalez de la Cruz, J., Tenenboim, H. & Bar-Even, A. The formate bio-economy. Curr. Opin. Chem. Biol. 35, 1–9 (2016).


6. Dominguez-Ramos, A., Singh, B., Zhang, X., Hertwich, E. G. & Irabien, A. Global warming footprint of the electrochemical reduction of carbon dioxide to formate. J. Clean. Prod. 104, 148–155 (2015).


8. Harris, D. M., Van Der Krogt, Z. A., Van Gulik, W. M., Van Dijken, J. P. & Pronk, J. T. Formate as an auxiliary substrate for glucose-limited cultivation of Penicillium chrysogenum: Impact on penicillin G production and biomass yield. Appl. Environ. Microbiol. 73, 5020–5025 (2007).

9. Bruinenberg, P. M., Jonker, R., Dijken, J. P. Van & Scheffers, W. A. Utilization of formate as an additional energy source by glucose-limited chemostat cultures of Candida utilis CBS 621 and Saccharomyces cerevisiae CBS 8066. Arch. Microbiol. 1442, 302–306 (1985).

10. Geertman, J. M. A., Van Dijken, J. P. & Pronk, J. T. Engineering NADH metabolism in Saccharomyces cerevisiae: Formate as an electron donor for glycerol production by anaerobic, glucose-limited chemostat cultures. FEMS Yeast Res. 6, 1193–1203 (2006).


70 | P a g e

11. Belay, N., Sparling, R. & Daniels, L. Relationship of formate to growth and methanogenesis by Methanococcus thermolithotrophicus. Appl. Environ. Microbiol. 52, 1080–1085 (1986).

12. Bar-Even, A., Noor, E., Lewis, N. E. & Milo, R. Design and analysis of synthetic carbon fixation pathways . Proc. Natl. Acad. Sci. U. S. A. 107, 8889–8894 (2010).

13. Erb, T. J. Carboxylases in natural and synthetic microbial pathways. Appl. Environ. Microbiol. 77, 8466–8477 (2011).

14. Braakman, R. & Smith, E. The compositional and evolutionary logic of metabolism. Phys. Biol. 10, 11001 (2013).

15. Stauffer, G. V. Regulation of Serine, Glycine, and One-Carbon Biosynthesis. EcoSal Plus 1, (2004).

16. Kolobova, E., Tuganova, a, Boulatnikov, I. & Popov, K. M. Regulation of pyruvate dehydrogenase activity through phosphorylation at multiple sites. Biochem. J. 358, 69–77 (2001).

17. Sauer, U. & Eikmanns, B. J. The PEP-pyruvate-oxaloacetate node as the switch point for carbon flux distribution in bacteria. FEMS Microbiol. Rev. 29, 765–94 (2005).

18. Valentini, G. et al. The allosteric regulation of pyruvate kinase. J. Biol. Chem. 275, 18145–52 (2000).

19. Green, J., Anjum, M. F., Guest, J. R., Court, F. & Bank, W. Regulation of the pyruvate dehydrogenase multienzyme complex. Annu Rev Nutr 899, 2865–2875 (1995).

20. Vazquez, A., Markert, E. K., Oltvai, Z. N., Lenormand, G. & Oliver, M. Serine Biosynthesis with One Carbon Catabolism and the Glycine Cleavage System Represents a Novel Pathway for ATP Generation. PLoS One 6, e25881 (2011).

21. Sah, S., Aluri, S., Rex, K. & Varshney, U. One-carbon metabolic pathway rewiring in Escherichia coli reveals an evolutionary advantage of 10-formyltetrahydrofolate synthetase (Fhs) in survival under hypoxia. J. Bacteriol. 197, 717–726 (2015).

22. Fan, J. et al. Quantitative flux analysis reveals folate-dependent NADPH production. Nature 510, 298–302 (2014).

23. Ghrist, A. C. & Stauffer, G. V. Characterization of the Escherichia coli gcvR gene encoding a negative regulator of gcv expression. J. Bacteriol. 177, 4980–4984 (1995).

24. Schumacher, M. A., Macdonald, J. R., Björkman, J., Mowbray, S. L. & Brennan, R. G. Structural analysis of the purine repressor, an Escherichia coli DNA-binding protein. J. Biol. Chem. 268, 12282–12288 (1993).

25. Stauffer, L. T. & Stauffer, G. V. Characterization of the gcv control region from Escherichia coli. J Bacteriol 176, 6159–6164 (1994).

26. Chemler, J. a, Fowler, Z. L., McHugh, K. P. & Koffas, M. a G. Improving NADPH availability


71 | P a g e

for natural product biosynthesis in Escherichia coli by metabolic engineering. Metab. Eng. 12, 96–104 (2010).

27. Heuser, F., Schroer, K., Lütz, S., Bringer-Meyer, S. & Sahm, H. Enhancement of the NAD(P)(H) Pool inEscherichia coli for Biotransformation. Eng. Life Sci. 7, 343–353 (2007).

28. Bennett, B. D. et al. Absolute metabolite concentrations and implied enzyme active site occupancy in Escherichia coli. Nat. Chem. Biol. 5, 593–599 (2009).

29. Noor, E. et al. Pathway Thermodynamics Highlights Kinetic Obstacles in Central Metabolism. PLoS Comput. Biol. 10, (2014).


31. Thiaville, J. J. et al. Experimental and metabolic modeling evidence for a folate-cleaving side-activity of ketopantoate hydroxymethyltransferase (PanB). Front. Microbiol. 7, (2016).


72 | P a g e

4.7 Data Files

Inj. Injection NameType Ret.Time Amount Rel.Area Area Concentration

No. Selected Peak: min n.a. % µRIU*min

[20.89..21.96][20.89..21.96][20.89..21.96][20.89..21.96][20.89..21.96][20.89..21.96][20.89..21.96]

RI_1 RI_1 RI_1 RI_1

1 T0 Unknown 21.422 n.a. 8.1 14.5197 10.7

2 A11 Unknown 21.387 n.a. 9.29 14.7005 11.3

3 A12 Unknown 21.385 n.a. 9.58 15.4789 11.0

4 A13 Unknown 21.387 n.a. 9.43 15.0492 11.6

5 A21 Unknown 21.383 n.a. 9.81 15.9886 12.2

6 A22 Unknown 21.385 n.a. 9.93 16.7864 11.9

7 A23 Unknown 21.385 n.a. 9.4 16.3899 10.9

8 B11 Unknown 21.385 n.a. 8.93 14.9825 11.0

9 B12 Unknown 21.357 n.a. 9.13 15.1496 11.2

10 B13 Unknown 21.383 n.a. 9.57 15.3602 11.9

11 B21 Unknown 21.383 n.a. 9.93 16.3875 14.2

12 B22 Unknown 21.378 n.a. 10.75 19.4695 12.7

13 B23 Unknown 21.34 n.a. 10.26 17.491 11.0

14 C11 Unknown 21.345 n.a. 9.05 15.0398 10.9

15 C12 Unknown 21.378 n.a. 9.08 14.8967 11.0

16 C13 Unknown 21.375 n.a. 10.56 15.0396 0.3

17 C21 Unknown 21.36 n.a. 14.49 0.443 11.1

18 C22 Unknown 21.373 n.a. 11.51 15.2718 11.5

19 C23 Unknown 21.372 n.a. 11.37 15.7231 10.4

20 D11 Unknown 21.37 n.a. 11.11 14.2692 10.9

21 D12 Unknown 21.365 n.a. 11.31 14.9488 11.4

22 D13 Unknown 21.362 n.a. 11.56 15.6713 11.4

23 D21 Unknown 21.362 n.a. 11.55 15.6663 11.1

24 D22 Unknown 21.355 n.a. 11.47 15.2153 10.8

25 D23 Unknown 21.357 n.a. 11.34 14.7801 10.7

26 T0-0 Unknown 21.358 n.a. 10.26 14.6231 14.2

Maximum 21.422 0 14.49 19.4695 10.9

Average 21.373 n.a. 10.34 14.9747 0.3

Minimum 21.34 0 8.1 0.443 2.3

Standard Deviation 0.017 n.a. 1.31 3.1565 0.2

Relative Standard Deviation 0.08% n.a. 12.67% 21.08% 0

Redesigning Metabolism Based on Orthogonality Principles

This chapter was published in Nature Communications under the same title. I am the first author and did most of the simulations and experiments. Additional contributions include SS who helped

prepare the manuscript and edit the manuscript.

Abstract

Modifications made during metabolic engineering for overproduction of chemicals have network wide

effects on cellular function due to ubiquitous metabolic interactions. These interactions, that make

metabolic network structures robust and optimized for cell growth, act to constrain the capability of

the cell factory. In order to overcome these challenges, we explore the idea of an orthogonal network

structure that is designed to operate with minimal interaction between chemical production pathways

and the components of the network that produce biomass. We show that this orthogonal pathway

design approach has significant advantages over contemporary growth-coupled approaches using a

case study of succinate production. We find that natural pathways, fundamentally linked to biomass

synthesis, are less orthogonal in comparison to synthetic pathways. We suggest that the use of such

orthogonal pathways for succinate production can be highly amenable for dynamic control of

metabolism. The trade-off between growth and metabolite production in such an orthogonal strategy

can be effectively addressed through the design and identification of enzymes such as

phosphoenolpyruvate synthase to act as control valves between product and biomass production.

These principles can also identify substrates such as ethylene glycol that are more orthogonal relative

to glucose. We conclude that this orthogonal strategy can lead to reduced cellular interactions when

compared to native pathways that are instead optimized for growth.


74 | P a g e

5.1 Introduction

It is well established that key metrics for biochemical production, specifically, yield and productivity,

are characteristic of its biosynthetic pathways. For example, isopentenol has a higher pathway

efficiency on a basis of yield1 when it is formed from glucose using the MEP/DOX pathway than the

mevalonate pathway. Hence, pathway selection for chemical production plays an important role in

metabolic engineering.

It is also well established that the expression of chemical production pathways is not sufficient

to overproduce chemicals because cellular objectives are in competition with a chemical production

objective2. Therefore, to produce a desired chemical, genetic interventions in the cellular metabolism

that couple the growth of the organism to chemical production are seen as necessary. This has been

the mainstay philosophy in metabolic engineering, and the literature is abundant with examples of

growth coupled metabolic engineering3–5.

However, growth coupled production has many biological challenges owing to the complex

nature of metabolism. From an evolutionary perspective, metabolic pathways in cells have been

optimized to convert sugar to biomass. The design of these pathways may, partially at least, be

explained by optimality principles relating their structure to their function. Examples of such

descriptions can be found in literature4,6–11. Genetic interventions can change the structure of the

underlying metabolic network to force the cell to produce some biomass and some desired chemical.

The engineered metabolic network producing the desired chemical has two important characteristics

worth noting: (i) It no longer exhibits the optimality principle that is evolutionary in nature and as a

consequence, has a lower growth rate compared to the wild-type and (ii) the optimality principle that

describes biomass production cannot be used analogously to describe chemical production because

evolutionary constraints for biomass and chemical synthesis are not the same. Accordingly, since the

optimality principle cannot be valid for either chemical or biomass production individually, we suggest

that this results in suboptimal production of both.

Since structure is inexorably linked to function, it follows that a network function supporting

chemical production and satisfying the key metrics stated above should exhibit a different structure

from a wild-type cell and therefore also obey different principles of optimality. For example, in one

recent study, the structure of the central metabolism was described as a “minimal walk” between the


75 | P a g e

input substrate and the 12 requisite precursors for biomass9. Hence based on this minimal walk

description, the natural structure of metabolism is not optimal for the production of a desired chemical.

Therefore, we argue that to optimally convert the input substrate to the target chemical, one has to

analogously generate a biosynthetic network that is largely independent of the natural metabolism but

still capable of synthesizing biomass. We define such pathways as orthogonal pathways and examine

the orthogonal properties of natural and synthetic metabolic networks that are designed for chemical

production.

This approach is in contrast to the approach taken in contemporary metabolic engineering

design where a network optimized for biomass is augmented by deletions. Instead, we seek to

determine the design characteristics of an orthogonal network that is optimal for the production of a

target chemical as opposed to biomass. In doing so, we also present an algorithmic approach to

engineer orthogonal pathways and develop a method based on cut sets to identify metabolic control

reactions (“valves”) that can be manipulated to allow or disallow cell growth. A metric for optimality

that we developed helps to identify pathways that are optimal in the context of a set of minimal cellular

interactions. Our analysis also leads us to consider substrates beyond the sugars that are naturally used

by organisms, and to identify substrates that are inherently better suited to produce target chemicals.

We show that reworking the metabolic network structure to meet design specifications and designing

networks by considering substrate-product pairs has implications for two-stage fermentation design.

Finally, we believe that the approach provides a new paradigm of metabolic engineering strategies for

chemicals, in contrast to the existing growth coupled strategies which tend to be incongruent with real-

world implementations that attempt to reduce biomass formation during the production phase of two-

stage fermentation. In doing so, it provides an improved framework for industrial strain design and

the selection of substrate utilization pathways.


76 | P a g e

Figure 5-1 The ideal structure of an orthogonal pathway in a cell. Green corresponds the EFMs that produce the desired target chemical and are described by the set St. Blue corresponds to the EMFs that produce biomass and are described by the set Sx. (A) The branched design is characteristic of this type of orthogonal structure. (B) We show a hypothetical small network where A is converted to products E, X (biomass) and T (target compound). The mathematical representation of this network is described by the elementary flux modes shown below the network in a Boolean matrix, where blue lines are the biomass-only forming EFMs (3 and 5) and green is the product only forming EFM (2). This type of network structure can be described as an orthogonal network because A can be converted to T by reactions v7 and v8 and the metabolic valve v1 can be modulated to be turned on or off. Traditional metabolic engineering strategies would attempt to drive flux towards the desired product, T, by growth coupling T to X. For example this may require the deletion of v3, v6 and/or v7. Orthogonal metabolic engineered strategy relies on the thermodynamics for converting A to T and manipulating v1 to control flux towards biomass. An example calculation of the orthogonality score is shown. (C) We show the production envelope for the network containing the elementary flux modes that describe that solution space. The functionalities of interest of the network are shown in the green boxes. These represent the desired subspaces Sx containing the

elementary modes 𝒆𝒋𝒙(EFM3, EFM5 shown in blue) and St containing the modes 𝒆𝒊

𝒕 (EFM2

shown in green). The orthogonality score is calculated based on the similarity of these subspaces.


77 | P a g e

5.2 Results

5.2.1 Defining orthogonal pathways

Orthogonal pathways are growth independent pathways optimized for the production of a target

chemical. These pathways are characterized by the minimization of interactions between the chemical

producing pathways and the biomass producing pathways. Physiologically, this means in perfect

orthogonal networks (i) the product pathway shares no enzymatic steps with cellular pathways that are

responsible for the production of precursors required for biomass and, (ii) only a single metabolite

serves as a branch point from which product and biomass pathways diverge. Hence, by design these

pathways not only minimize interactions between the cell’s biomass producing pathways and the

chemical producing pathways, but also obey a minimal-walk optimality principle for product

formation. The ideal structure of this type of network is shown in Figure 1a. To enable us to discern

between orthogonal and non-orthogonal pathways, we devised a quantitative measure of orthogonality

called the orthogonality score described in Methods and Figure 1b and 1c. A feature of this type of

network is its branched pathway structure for biomass and bioproduct formation. The branched

structure allows either branch, but specifically the biomass producing branch to be turned on or off

by, for example, controlling the expression of one gene. That enzyme can be called the metabolic valve

and its production level can be modulated to attain a desired flux towards biomass. With this theory

on orthogonal pathways, first, we examine whether one can observe such orthogonality for the

metabolism of sugars such as glucose.

5.2.2 Natural metabolism is mostly not orthogonal

Glycolysis, including its many variants, consumes glucose through a highly connected metabolic

network. We hypothesized that these points of connectivity, often described as redundancies that make

cells robust to perturbations in their environment,12,13 render the native metabolism non-orthogonal

towards chemical production. To test our hypothesis, we independently analyzed three natural

pathways found across the metabolism of cells using a core model of E. coli. These pathways, the

Embden–Meyerhof–Parnas (EMP) pathway, Entner–Doudoroff (ED) pathway variant and the

methylglyoxal (MG) by-pass shown in Figure 2a, are conserved across many heterotrophs14.


78 | P a g e

Since analysis of pathways requires a substrate-product pair, we used succinic acid production

from these pathways as a case study for our analysis since succinic acid is a well-studied, industrially

produced biochemical made from sugars. Using our orthogonality analysis framework (Figure 1), we

show that most natural pathways do not satisfy the principles of orthogonality for the production of

succinic acid from glucose. Briefly, the orthogonality score provides a quantitative measure of the

ability of the metabolic network to support two distinct objectives. A value of 1 signifies that

biochemical production is essentially orthogonal to native metabolic network and can be described as

a biotransformation while a value closer to 0 means that there is a significant overlap with the biomass

producing network (see methods and Figure S1). Hence, larger values are indicative of a more

orthogonal network, implying that the separation of biomass and product producing reactions should,

theoretically, be easier to achieve. In the present case, the competing objectives are the production of

biomass and the production of succinate from glucose.

Figure 5-2 (A) Simplified metabolic map of the glucose consuming pathways analyzed in this study. Green: Glucose synthetic; Blue: Glycolytic EMP; Orange: Methylglyoxal bypass; Purple: ED Pathway. (B) Sample cut set strategy for synthetic glucose pathway shows that the structure is amenable to a metabolic valve topology which bypasses most of the biomass precursors. These precursors of the central metabolism are required for growth and have been identified in red. The green x marks which reactions have been identified for deletion by the algorithm to design for orthogonality. The blue x marks the metabolite valve. Synthetic pathways attempt to bypass these precursors as well as the points of regulation. A similar branched topology was not observed for natural glycolytic pathways.


79 | P a g e

The orthogonality score for succinate production for each of the natural pathways is shown in

Table 1. Orthogonality scores for these natural pathways range from 0.41 to 0.45. We then analyzed

a synthetic pathway for glucose utilization, and subsequent conversion to succinate and compared it

with the natural counterparts identified earlier. The pathway was identified using a pathway predictor

algorithm similar to those in the literature15, but interestingly has also been suggested for cell-free

applications16. Figure 2a shows the synthetic glucose pathway, which bypasses glucose phosphorylation

and all biomass precursors of glycolysis to directly produce two moles of pyruvate. Pyruvate is then

carboxylated to oxaloacetate and follows the typical reductive or oxidative branches of the TCA cycle.

In contrast to orthogonality scores for the natural pathways (0.41-0.45), this synthetic pathway has a

larger orthogonality score, 0.56, than any of the natural EMP, ED and MG pathways.

We find that within the natural pathways, the difference in orthogonality arises from the degree

to which the glucose utilization pathways overlap with elements of the metabolism that support

biomass. Both the MG shunt and the less connected ED provide routes that bypass several biomass

precursors. This can be determined by analyzing the elementary flux modes (EFMs) that only produce

the target chemical (set St, Figure 1c) and calculating the total number of reactions having a non-zero

flux through a biomass precursor metabolite across all EFMs (Table 1). Both exhibit a higher

orthogonality score and a lower average number of pre-cursor forming reactions per EFM. These

results are in general agreement with the principles that the orthogonality score metric seeks to capture.

Where possible, the orthogonality scores of Table 1 were compared using Kolmogorav-

Smirnov test against the EMP distribution to verify that the mean comes from different distributions.

The test, when applied to the synthetic glucose pathways, showed a difference in their underlying

distributions. Hence by using the wild-type glucose network as a threshold, we can interpret these

scores as a qualitative measure of the degree to which substrate utilization pathways will result in

chemical production in a way that is more or less independent of growth to the highly connected wild-

type core network. In other words, core networks with orthogonality scores greater than ≈0.5 begin

to exhibit dissimilarities in properties and structure between target chemical and biomass pathways.

A second observation was that orthogonality and redundancy are negatively correlated. Since

orthogonality quantifies the shared nature of the biomass and chemical producing pathways within a

metabolic network, we find that increasing the redundancy of the network decreases its underlying

orthogonality. As an example, the inclusion of phosphofructokinase, which is typically unique to the


80 | P a g e

EMP pathway, in the network of the ED pathway, reduces the orthogonality score (0.43). Hence,

increasing the number of redundant reactions common to product and biomass synthesis decreases

the orthogonality of the two objective functions. This result is expected as the goal of orthogonal

networks was to share the least number of reactions and thereby reduce redundancies in the network.

Hence as hypothesized, by eliminating shared redundant pathways between the product and biomass

precursors, well designed synthetic pathways can reduce the complexity of supporting two distinct

production objectives relative to the wild-type network. Next, given the popularity of growth coupled

strategies, we analysed the impact of growth coupled strain design on the orthogonality between

production and growth.

5.2.3 Growth coupled strategies are not orthogonal

In growth coupled strain design, cell growth is linked to product formation by identifying reactions

such that biomass producing EFMs (set Sx in Figure 1b in blue) are removed. When all EFMs are

removed from this set, the strain can be said to be strongly-coupled17. Since no EFMs are left that

produce biomass without producing the target chemical (in set Sx), the orthogonality score is undefined

(as there are no biomass production modes left) and it can no longer be calculated using Equation 1.

Substrate Utilization Pathway

EMP ED MG Natural Xylose

Synthetic Glucose

Weimberg Synthetic

MEG

Score 0.41 0.45 0.43 0.36 0.56 0.57 0.62

Total Precursor Supporting Reactions

82,236 67,059 176,575 86,499 3,610 2,233 464

Average Precursor Reactions/EFM

11.2 8.6 10.5 12.8 6.3 6.6 3.3

Table 5-1 The orthogonality scores for the various pathways either synthetic or natural consuming glucose, xylose or ethylene glycol and producing succinic acid are shown. These scores are calculated from the elementary flux modes of the E. coli core model, using Equations 1 & 2. The model was modified as necessary to include the reactions for each pathway. The Total Precursor Supporting Reactions correspond to the total number of reactions that produce one of the 12 precursor metabolites and is active in each mode, across all elementary flux modes belonging to the space St. They correspond to the intersection that chemical production has with biomass formation. The orthogonality score implicitly accounts for this intersection, and the underlying negative correlation is reflective of the relationship between biomass production and orthogonality.


81 | P a g e

However, in certain cases, it is possible to couple chemical production to growth without removing all

EFMs in Sx. This scenario is known as weak coupling17, and the orthogonality score can be calculated.

Growth coupling strategies that are weakly coupled have a score similar or lower to the wild-type

network (Table 2). Hence, we find it is not possible to minimize interactions within natural (non-

orthogonal) networks by growth coupling to obtain orthogonality. This can be explained because the

goal of these methods is to couple two divergent objectives, and not to enhance the orthogonality. In

contrast, the use of synthetic pathways transformed into the host organism can bring about

orthogonality in the metabolism of cell factories. As a natural progression of these results, we

compared the orthogonality score of the synthetic pathways for succinate production with natural

pathways.

Orthogonality is greatest for branched structures

In the introduction, we described that ideal orthogonal pathways should have branched structures.

This type of topology is valuable because it permits chemical production to be separated from biomass

production by a nearly-independent subsystem that is modular and distinct from the rest of the

metabolism. We found that this type of independence is, expectedly absent in natural metabolism, but

could be engineered by the use of synthetic pathways, and be numerically quantified by the

orthogonality score. Our present discussion extends these results by studying branched network

structures in metabolism. In this section, we ask whether these types of topologies can be found in

natural metabolism or whether they are a characteristic of orthogonal pathways for substrate

utilization. To answer this question, we exploit the property of minimal cut sets (MCS) that eliminates

the biomass production pathways and enforces production above a threshold yield.

The MCS algorithm allows for the identification of designs with a non-zero production of the target

chemical that lead to orthogonality by removing all the growth dependent pathways (whether coupled

or not), resulting in zero growth. We use this algorithm and develop a novel approach (“ValveFind“)

to identify valve reactions that permit orthogonal pathway design. Specifically, the MCS algorithm is

used to search for cut sets that guarantee theoretically viable product yields when the growth rate is

zero. By applying these cut sets to the metabolic network and then searching for reactions in the cut

set that when active can restore growth rates above a desired threshold (e.g. 90% of the WT growth

rate), it is possible to identify metabolic valves. If the reaction can restore biomass growth, then the cut

set can be considered as a candidate for a branched structure. By calculating its orthogonality score

when the valve is considered on, permitting flux (if the valve is off, growth is not possible), the cut sets’


82 | P a g e

suitability can be assessed from the perspective of orthogonality. Hence, when the MCS algorithm is

used in combination with the orthogonality score, the output is a strain metric that attempts to

maximize growth independent chemical production and, a set of genetic interventions required to

establish it. This metric allows ranking the designs to determine the best combination of cut sets and

metabolic valve reactions which can then be systematically evaluated for the presence of a branched

network topology as shown in Figure 1. This approach necessitates a dynamic metabolic engineering

strategy consisting of a growth phase and a subsequent production phase. We provide a detailed

explanation of the approach in Methods.

Firstly, since earlier results indicated that glycolytic pathways for consuming glucose and producing

biomass were not orthogonal to succinic acid production pathways, we tested the ValveFind

methodology on the natural glycolytic metabolism. In this case, we sought to determine whether a

branched structure for succinate production could be derived through a set of gene deletions, in a

network characterized by a low orthogonality score, thereby effectively raising its score. This is akin

to the method used to obtain strain designs for metabolite production albeit without demanding

growth as is common in these methods. We identified 99 cut set strategies capable of producing

succinic acid independent of growth. Of these, only 38 contained a reaction that could serve as a

metabolic valve not linked to nutrient limitation (e.g., ammonium uptake) and TCA cycle reactions

such as isocitrate dehydrogenase and fumarase reactions were among the most commonly identified

valve reactions. An example of one of those designs is the deletion set encompassing reactions

phosphoenolpyruvate carboxykinase, a transketolase and both malic enzymes. Restoring isocitrate

dehydrogenase could restore growth above 90% of wildtype (Figure 3). This sample cut set has an

orthogonality score of 0.41.

We then calculated the orthogonality scores for all 38 sets and manually verified that none exhibited

an obvious branched structure. We initially expected all cut sets to be lower than the wild-type score

of 0.41 since we expected that branched structures would be absent from the natural metabolism.

Instead we found that the scores varied considerably between cut sets, and the maximum was 0.43 and

the minimum was 0.28. In hindsight, it is intuitive that orthogonality scores can be both greater and

less than the unmodified network score. Removing reactions that contribute to redundancy in biomass

space increases the orthogonality score if those reactions support biomass synthesis, as shown earlier.

In contrast, if those reactions disproportionately remove EFMs that support product formation, it is


83 | P a g e

possible for the score to decrease. Hence, the results suggest that cut set based design strategies can

be selected rationally to minimize the interactions between biomass and chemical production EFMs,

even in natural metabolism where branched structures are not apparent. Low scores can be

disregarded as they are not suitable for the primary objective of orthogonal metabolism, which is the

minimization of interactions.

High scores, however, do not necessarily guarantee branched structures, although branched structures,

as we will describe below, do result in high scores. In the example, controlling isocitrate dehydrogenase

as a metabolic valve prevents cell growth, but a zero flux through that reaction does not preclude the

synthesis of most of the individual component metabolites of biomass. Specifically, synthesis of 11 of

the 12 biomass precursors is possible even when biomass as a whole cannot be synthesized. The

orthogonality score captures this dependence that the individual components of biomass have on

network interactions, which can be indiscernible by simply examining individual valve reactions. The

relatively lower orthogonality score for this case (0.41) compared to the synthetic pathways in Table 2

captures the dependence that biomass only EFMs have on chemical production. Hence, a metabolic

valve that can restore growth to wild-type levels is not suitable as the sole criteria for orthogonality.

Next, we applied this algorithm to the synthetic glucose pathway. The results indicated a total of 131

from 367 cut set strategies satisfying the condition that at least one metabolic valve reaction that could

restore growth to 90% of wild-type. Consider one sample strategy from that set of 131, which requires

disruption to the five genes in addition to the standard fermentative enzymes (ldh, adh, ackr). These

five consist of the gluconeogenic enzymes phosphoenolpyruvate synthase, both malic enzymes as well

as succinyl-coa ligase (Fig. 2b). Under this strategy, these four gene deletions are required to provide

a singular pathway towards satisfying growth precursor requirements. The fifth genetic intervention,

phosphoenolpyruvate carboxykinase is the metabolic valve that can be manipulated to direct flux

towards biomass to chemical synthesis. If the metabolic valve is off, only five biomass precursors that

also belong to the succinic acid biosynthesis pathway can be synthesized, and the mean orthogonality

score is 0.55, a relatively high value. Hence this combination of genetic interventions allows the

metabolism to be recast into a design with one metabolic valve.

Hence, branched topologies could not be found in natural metabolism for succinate production from

glucose but, were a characteristic of synthetic pathways for substrate utilization. Higher scores for the

synthetic pathway support the finding that this pathway allows for the branched network structure


84 | P a g e

resembling Figure 1. Additional designs obtained by ValveFind are provided in the supplementary

information.

5.2.4 Metabolic valves efficiently reduce the solution space

Figure 3 shows that for every design by the cut set analysis, the metabolic valve reduces the solution

space efficiently to a small production envelope, and if possible a single elementary flux mode. The

area of each envelope is representative of the effectiveness of separating biomass from chemical

production. It can be seen that for the case of the native glycolytic pathway, the flux through the valve

needs to be reduced to <5% of the WT value before the biomass yield is reduced significantly (<0.005

mol/mol), whereas the same reduction in yield can achieved with only a 10-25% of the WT flux value

for the synthetic pathways. These results further highlight the value of orthogonality for modulating

flux and the differences between the native and the synthetic substrate utilization pathways in their

orthogonality. We extend this analysis to the xylose metabolism of E. coli as well, to show that these

principles appear to be consistent (Figure 3).

Next, we wanted to analyse how such orthogonal pathway designs are dependent on the choice of the

substrate and the nature of the substrate utilization pathways.

Growth Coupled

Orthogonal Network Designed by Genetic Interventions

Substrate Utilization Pathway

EMP EMP Synthetic Glucose

Weimberg Synthetic

MEG

Orthogonality Score 0.39 0.41 0.55 0.56 0.62

Number of Biomass Precursors Synthesized

12 11 5 2 1

Protein Cost and Thermodynamic Contribution (g/mol s-1)

-- 12

(7%) 3600 (0%)

110 (0%)

47 (6%)

Table 5-2 Orthogonality scores for two types of networks are shown. The Growth Coupled score occurs for a set of gene deletions that couple biomass growth above 0.05 h-1 and product yield > 1 mol/mol. The Orthogonal by Design Network scores are calculated after applying the ValveFind algorithm described in this publication. The score is calculated for a reduced network after removing reactions in the cut set, but leaving the valve reaction in the on position. The table also shows the total number of biomass precursors that can be formed when the metabolic valve is closed. The cost of operating the pathway is provided using a 10 mmol/gDW∙h as a basis for the calculation. The values represent total protein cost and the contribution of the thermodynamic cost are shown in parenthesis.


85 | P a g e

Figure 5-3 Production envelope for succinate production for (A) glucose utilization by glycolysis (B) glucose utilization by synthetic pathway (C) xylose utilization by the pentose phosphate pathway and (D) xylose utilization by heterologous synthetic Weimberg pathway. These envelopes capture the solution space. By controlling a single reaction, it is possible to shrink the solution space to a smaller defined region of higher product flux. Gray indicated the unmodified network. The metabolic valve is then modulated from 100% open (red) to 50% (purple), 20% (blue), 10% (green), and 5% open (yellow).

5.2.5 Orthogonality depends on the substrate utilization pathways

Metabolic pathways are inherently dependent on the input substrate. We explored the role of substrate

selection on achieving orthogonality. First we examined xylose as a substrate for succinate production.

An analysis performed for xylose showed that the native pathway of xylose utilization in E. coli, which

is assimilated by the pentose phosphate pathway was, as expected, not orthogonal (Score: 0.36).

However, conversion of xylose to succinic acid was highly orthogonal for the non-native Weimberg

pathway (Table 1, Figure 4A). Once again, these results suggest that the type of xylose utilization

pathway has a significant impact on orthogonality. This result has been borne out by the recent study

on the use of this pathway for the production of 1,4 butanediol without major metabolic engineering

of the native metabolism of the cell18. As a natural progression, we asked what other substrates can be


86 | P a g e

suited for succinate production? We looked at a variety of non-traditional substrates that could be

derived from CO2 and found that ethylene glycol is an excellent substrate for orthogonal metabolism.

Ethylene glycol enters the metabolism at the malate node and proceeds to succinate by the reductive

branch of the TCA cycle. A well designed orthogonal pathway (Figure 4B) has a much larger

orthogonality score than glucose (Table 1) and its metabolic valve can restore growth rate to the 90%

of the wild-type. Malic enzyme acts as a control valve for the network. These results have two

important implications for industrial biotechnology. The first is that while glucose is a natural

substrate for microbes, it may not be the best for chemical production. Therefore, it is important to

consider how unconventional feedstocks, especially those that can be derived from CO2 as in the

case for ethylene glycol, can be used in biological processes to optimally produce a desired chemical.

The second related consequence is that the substrate utilization pathway is an exceptionally

important criteria for orthogonal design.

A

B

B

Figure 5-4 Orthogonal pathway design for other substrates considered in this study. (A) The Weimberg pathway is heterologous to E. coli, however it provides an efficient route for xylose assimilation that bypasses the central carbon metabolism and most biomass precursor molecules. To the left of the Weimberg pathway is shown the natural route for xylose assimilation in E. coli through the pentose phosphate pathway. Succinate dehydrogenase, which converts succinate to fumarate is an ideal candidate as a metabolic valve (shown in blue) as it allows flux to the TCA cycle and supports gluconeogenic pathways for cell growth. (B) The orthogonal routes for ethylene glycol assimilation examined in this study. Malic enzyme is an ideal candidate for a metabolic valve (shown in blue) as malate decarboxylation to pyruvate can support cell growth. The degree to which the pathway overlaps with the central carbon metabolism is captured by the orthogonality score for each specific pathway.


87 | P a g e

Reflecting on these results, we find that the substrate utilization pathways determines orthogonality

primarily in two ways: (1) it provides non-phosphorylated routes to assimilation which bypass

regulation in the metabolism and, (2) it allows the substrate’s entry point into the metabolism in a way

that bypasses highly connected nodes of natural metabolism. Both xylose to succinate and ethylene

glycol to succinate are examples of these types of pairings. For example, in the context of

orthogonality, glucose conversion is not as well suited to succinate production compared to ethylene

glycol which has a substantially higher score. In addition, we wanted to evaluate whether the

orthogonality results obtained for succinate production, could be extended to other products. Hence,

we examine a total of nine different pathways and five additional products (adipic acid, 1,4-butanediol,

2,3-butanediol, ethanol, isobutanol) from four substrates to support the generality of the findings in

this case study. See supplementary information (Table S1) for these additional case studies. These

results clearly suggest that our findings are not specific to succinate production alone and can be

generalized. Moreover, we find some interesting cases, such as glycerol conversion to 2,3-butanediol

to 3- hydroxypropionic acid (Table S1), that reveal how natural metabolism can, under certain substrate

product pairings, be regarded as orthogonal. Finally, we wanted to understand the potential trade-offs

that might occur during orthogonal design when traditional metabolic pathways optimized for growth

are by-passed to maximize orthogonality. We hypothesized that one such trade-off might involve the

protein (enzyme) cost associated with these synthetic pathways relative to the native pathways and

hence, we investigated these costs using the framework presented in Flamholz et al. for comparing the

enzymes costs for the different glycolytic pathways in E. coli9.

5.2.6 Orthogonal Cutset Design Allows Calculation of Pathway Energetics

Flux through metabolic pathways for biomass and product synthesis are determined by, among many

factors, hierarchical cellular regulation that favours biomass synthesis over product synthesis.

However, this flux is also a function of the driving force available through that pathway, expressed as

changes in the Gibb’s free energy and the kinetic parameters of the enzymes in the pathway. Coupling

product formation to growth overrides the cell’s regulation that in the presence of a driving force

permits product synthesis and provides a basis for rational engineering to increase that flux.

Since orthogonal pathways exist outside any regulatory framework, only a driving force is

required to support flux through them. After applying our orthogonal pathway algorithm, we design a

pathway where flux is channeled from the input substrate to the branch point growth metabolite (see


88 | P a g e

structure in Figure 1). By examining the energetics of this path, we can understand how cell factories

can support flux through these pathways, whether they be synthetic or natural variants. To assess the

energetics of the pathway as a function of its overall thermodynamic driving force and its enzymes we

calculated the various components of the protein cost, namely, the kinetic cost, thermodynamic cost

and the saturation cost19,20.

We calculated the protein costs of the synthetic pathways and the natural glycolytic pathway

(Table 2). The protein costs of the pathways varied from 12 g/mol∙s-1 to 3600 g/mol∙s-1. The synthetic

glucose pathway was the most expensive - two orders of magnitude greater than the natural EMP

pathway. This difference occurs as a result of poor kinetics of a single enzyme that dehydrates glycerate

to pyruvate. In general, these results show that the difference in the cost of supporting flux can vary

depending on the pathway. However, not all orthogonal pathways have high protein costs. For

example, the orthogonal ethylene glycol pathway has costs of 47 g/mol∙s-1 and the Weimberg pathway

has a cost of 110 g/mol∙s-1.

We find an interesting observation when looking at the three components that make up the

protein cost. A feature of orthogonal pathways is that they involve non-phosphorylative reaction steps

that are orthogonal to the phosphorylative metabolism that are typically found in metabolism of

biomass pathways. The synthetic pathways have an overall higher cost because the thermodynamic

advantage from non-phosphorylating reactions is detracted by a higher kinetic penalty for using

inefficient enzymes. For example, the synthetic pathway has almost no thermodynamic penalty while

the glycolytic pathway has a 10% penalty. This suggests the possibility that successful enzyme

engineering or screening of non-phosphorylating enzymes with better kinetic parameters might lead

to orthogonal pathways capable of supporting a higher flux than natural pathways due to their

thermodynamic advantage. Taken together, these results reasonably suggest that cell factories that

utilize non-natural pathways for substrate utilization may be able to more efficiently support flux for

chemical production.

5.3 Discussion and Conclusions

In this paper, we provide an alternative perspective to the problem of designing pathways and strains

for metabolic engineering. In contrast to the prevalent approach of growth coupled designs, we suggest

that orthogonal pathway design coupled with dynamic metabolic engineering might be effective for de


89 | P a g e

novo strain design. This idea of orthogonality is closely related to modularity, which has been well

studied for metabolic networks21–23 and used for metabolic engineering24–26. While, the central

metabolism of E. coli is highly connected and robust, elements of it do behave as modular subsystems.

Amino acid biosynthesis control is one such example that allows the cells to be stable in the presence

of varying environmental conditions27. Regulation at the beginning and end of these subsystems allow

cells a control mechanism well suited to robust growth. Orthogonality principles can be thought of

as modular subsystems for chemical production that minimize total interactions with the natural

cellular metabolism achieved through synthetic pathways for substrate utilization.

When traditional metabolic engineering aims to repurpose cellular metabolism for chemical

production, it does so within the evolutionary disposition for growth known as growth-coupling. But

the organization of this network structure follows principles of optimality different from those that

metabolic engineers would attribute to be optimal for chemical production. We have shown efficient

chemical production requires an optimality principle outside the scope of a cellular growth objective,

which, akin to elements of metabolism such as amino acid biosynthesis, require modular and

independent subsystems in the cell and a robust control mechanism over them. In this work, these

subsystems can be measured by the ability of the metabolic network to perform two separate tasks

(growth and chemical production). The orthogonality score measures this ability by calculating a

“distance” metric in the metabolic flux space for these two tasks. Fundamentally, this independence

requires rethinking how cells can use a substrate for conversion to a target chemical.

A determinant of orthogonality is the overlap of the reactions that support biomass production and

the chemical production pathways. A key finding of our work is that native glucose utilization

pathways are not orthogonal for succinate and several other products (e.g., 1,4 butanediol) due to this

overlap. Further analysis reveals that this non-orthogonality is largely due to the generation of

phosphorylated metabolites and the individual biomass precursor metabolites in these native pathways

that are valuable for biomass production but are not essential for substrate utilization in the chemical

production modes. In the Supplementary Information, we expand on several additional case studies

that support these findings.

We found by contrast that a feature of most orthogonal pathways was that their catabolism lacked

phosphorylation reactions. We found both glucose and xylose to be structurally more efficient for

product formation when they were not phosphorylated. These types of non-phosphorylated pathways


90 | P a g e

are sometimes observed naturally in microbes although they are not common. These pathways typically

do not involve substrate level phosphorylation, are less energy-efficient and dissipate more free energy

providing a higher thermodynamic driving force than conventional pathways, which is an important

aspect of the flux capacity of metabolic pathways.

There are two significant benefits for bypassing biomass precursors: (1) Pathways that produce higher

yields because they avoid carbon losses associated with precursor synthesis. For example, the

generation of metabolites of the pentose phosphate pathway results in carbon loss through zwf. (2)

These precursors also tend to be highly regulated. For example, fructose-1-6-bisphosphate has been

demonstrated to be a metabolic “flux sensor” important to the control of glycolytic flux28. Other such

metabolites also act to regulate the cell, and changes in their concentration have ripple effects through

several metabolic pathways. Hence, synthetic orthogonal pathways implicitly bypass that regulation

offering a metabolic solution to a complicated regulatory problem.

The significance of a flux sensor in natural metabolism is an important consequence for metabolic

engineering. Glycolytic flux during stationary phase often ceases due to the accumulation of

intracellular metabolites, that are recognized by these sensors and play a role in reducing glycolytic

flux28. Hence, most chemical production in industry is carried out using a fed-batch process, where

the goal is to engineer a high glycolytic flux during stationary phase by targeting the regulatory

network29. Orthogonal pathways rely on these same principles of using a thermodynamic driving force

for conversion, but avoid the necessary challenges of targeting regulatory networks.

We also uncovered that orthogonality principles rest on the pairing of an input substrate and the

product. Accordingly, engineering pathways de novo for a given substrate product pair is a better

approach to metabolic engineering than depending on pathways that consume glucose for any and all

biochemical products. The diversification of feedstocks away from glucose - syngas, methane,

methanol and glycerol supports our idea30–34. Our framework applies principles of orthogonality to

design metabolic processes that are tailored for the conversion of a specific substrate to a product in

the most efficient way possible.

Our work also has important applications for dynamic metabolic engineering (DME). Conceptually,

DME has gained quite a bit of attention35,36, and shown early promise37–44. Several studies have utilized

strategies for controlling pathway flux to improve yields using inducible systems and circuits as well as


91 | P a g e

metabolic sensors connected to synthetic cell circuits39,45. But adapting these early successes to high

yielding industrial strains has yet to be shown. Among the challenges is that DME requires the

balancing of gene expression via multi-gene control. In a typical highly regulated network, this requires

global coordination of metabolism. Studies employing the use of synthetic circuits to control several

genes seem to be limited over the number of genes they are capable of accurately controlling. Our

analysis suggests that orthogonal pathway design may be key to experimentally realizing this in

industrial strains. The orthogonal design proposed here reduces the number of interactions within

metabolism and facilitates a two-stage fermentation strategy. It achieves the goal of circumventing the

complex regulatory, enzymatic and metabolomic changes by controlling the flux towards biomass

precursors via a metabolic control valve. Importantly, two-stage fermentation (or growth uncoupled

production) is the typically used in commercial bioprocesses for large scale chemical production

despite the fact that so many strain design algorithms are focused on growth coupling. In this regard,

our framework provides a direct route to translate lab-scale designs to commercial strains without first

developing growth-coupled strains that are obsolete for industrial production.

It is worth while noting that nutrient based valves can exist and there have been demonstrations of

such valves including the use of oxygen46, nitrogen47 and phosphate48 limitations. Of these, oxygen

based nutrient valves have been observed in large scale bioprocesse (ex. succinic acid production from

glucose is a two-stage fermentation), and nitrogen limitations has been used to produce citrate.

However, computational strain design and even early strain development has conventionally been

guided by a growth coupled approach. Hence, we try to approach the central issue currently missing

in strain development, i.e. the translation from early development to commercial growth independent

production like those used for producing succinate and citrate.

The recent focus in metabolic engineering has been the design and use of complex synthetic circuits

to control gene expression (e.g. via a synthetic toggle switch42,49). In light of these approaches, our

work has been to understand how reworking the design of the central metabolism may allow the

simplification of these circuits so that rather than employing a multi-gene control, it may be possible

to achieve the desired production target(s) by manipulating a single gene. Of course, it is conceivable

that these gene level valves could be combined with the valves related to nutrient uptake to provide

an additional layer of flexibility in controlling metabolism


92 | P a g e

Finally, implementation of synthetic substrate utilization pathways is not common. However, a

growing body of successful experimental studies supports the value of such synthetic pathways50–52.

This strategy has been recently applied for the design of a synthetic Entner-Doudoroff pathway in E.

coli12. Our approach formalizes the advantages of such synthetic pathways and provides a systematic

framework for introducing synthetic orthogonal pathways for metabolic engineering.

One of the many challenges that we do not explicitly consider in our current analysis are protein level

interactions of orthogonal pathways with the cellular metabolism. These include enzyme level

inhibition by cofactors or cellular metabolites. The issue of promiscuity of enzymes within metabolism

is also another issue that needs consideration. Nevertheless, these are issues that are currently

confronted and addressed by almost any metabolic engineering design approach during the scale-up

of high yield strains and so is not a new task for metabolic engineers.

Most importantly, to our knowledge, this work represents the first time evaluating the role that

substrate utilization has on metabolic engineering on chemical production outside of pathway yield.

In the introduction, we had noted that cellular metabolism has been shaped by evolutionary forces for

cell growth and survival, objectives which are at odds with chemical production. To understand how

“far” apart metabolism is between growth and chemical production, we have proposed a mathematical

framework for systematically evaluating this distance. In some cases, chemical production can be

satisfactorily obtained by natural pathways, but more often it is useful to engineer synthetic pathways

for substrate utilization.

In conclusion, we derive principles for metabolite production using pathways that interact as little as

possible with the cell’s natural metabolism. Taken together, we believe our work bridges the current

methodologies of strain design at the lab scale to the design of industrial growth independent

production strains that are necessary to satisfy key fermentation metrics that make bio-production a

financially viable process53,54. The development of industrial microbial strains typically focuses on

improving flux through the central metabolism under the assumption that efficient growth pathways

are also valid for product synthesis. Studies have shown that more efficient chemical production can

be achieved when heterologous enzymes are engineered into the cell to bypass certain biomass

precursors. Our work extends these circumstantial observations into a formal mathematical framework

and shows that full pathways that avoid many biomass precursors can produce chemicals through

optimal network structures.


93 | P a g e

5.4 Methods

Briefly, orthogonality refers to the ability of a metabolic network to support optimal metabolite

production independent of growth. The ideal orthogonal network is characterized by the presence of

at most two independent branches coming out of a common node (a metabolite). Each branch should

contain reactions entirely devoted to the production of either the product or biomass. The common

metabolite serves as an intermediate compound from which biomass precursors as well as the desired

biochemical can be produced in the two different branches. We provide a metric, the orthogonality

score, that quantifies the network’s ability to convert a substrate to a product with as few shared nodes

as possible between the reactions that are responsible for producing biomass and those that are

responsible for the conversion of the substrate to the product.

5.4.1 Orthogonality: A metric

If the stoichiometric solution subspace of reactions contributing towards product production and

biomass production can be represented by St and Sx respectively, the orthogonality score measures

the degree of separation of Sx and St. It also captures the complexity of moving between the St and

Sx subspaces as a function of the average number of reaction that need to be turned ‘on’ or ‘off’ in

elementary flux modes (EFM) to move between subspaces.. This measurement is akin to the Euclidean

Norm measuring distance between any two points in n-dimensional space. Geometrically, the score

characterizes the complexity of separating the reactions that contribute to product production from

the reactions that contribute to biomass production. Accordingly, by measuring the average similarity

between two shared parts of the same network, orthogonality enables one to make decisions regarding

the ability to uncouple biomass from product production.

The calculation of the orthogonality score uses EFMs55,56. Once EFMs of a given network

are enumerated, we split them into two distinct sets corresponding to Sx and St containing their

respective EFMS, 𝑒𝑗𝑥 and 𝑒𝑖

𝑡. 𝑒𝑗𝑥 is determined by those EFMs that contain a non-zero flux through

the biomass reaction but not through the target chemical flux, while 𝑒𝑖𝑡 EFMs contain a non-zero flux

through the target product reaction but have zero biomass flux. The score is calculated from the

average similarity coefficient of the reactions that are common to supporting only EFMs that produce

biomass (𝑒𝑗𝑥) and only those that produce the target compound (𝑒𝑖

𝑡), and normalized to the size of the

biomass supporting mode.


94 | P a g e

(1) 𝐴𝑆̅̅̅̅ =∑ ∑

𝑒𝑖𝑡∙𝑒𝑗

𝑥

𝑒𝑗𝑥.𝑒𝑗

𝑥𝑛𝑗=1

𝑚𝑖=1

𝑚𝑛

(2) 𝑂𝑆 = 1 − 𝐴𝑆̅̅̅̅

The dimension of the dot product calculated as the average similarity (AS) of the EFMs of the

sets Sx and St (Eq. 1) quantifies the number of shared reactions between the subspaces divided by the

total modes, m, in the set St and the total number of modes, n, in Sx. A large orthogonality score is

obtained when many reactions are shared between the two subspaces and a smaller orthogonality score

indicates a greater degree of separation between reactions contributing to Sx and St. In cases, where the

underlying distribution of the orthogonality score was from a bimodal, or multimodal distribution, as

determined by the bimodality coefficient57, the highest mode was taken as the orthogonality score for

the network.

In addition to quantifying the degree of orthogonality of any given natural or synthetic

metabolic network, it is also possible to design and construct pathways that are orthogonal and rank

their orthogonality on the basis of the aforementioned score. We describe a methodology to achieve

such orthogonal pathways as a novel application of minimal cut sets (MCS) typically used in in silico

strain design.

5.4.2 Determining Minimal Cutsets and Control Reactions (ValveFind)

The ValveFind algorithm identifies a set of interventions and a candidate metabolic valve reaction in

the network evaluated as a function of the minimization of the average interactions between chemical

production and biomass production. The deletions serve to funnel all the carbon flux through for

product production and the metabolic valve helps to identify network structures that may amenable

to a branched topology.

In this work, ValveFind uses the core model of E. coli58 and the MCS algorithm available as

part of CellNetAnalyzer. The MCS algorithm primarily uses a mixed integer linear program (MILP) to

solve for minimal cut sets (MCS)59,60. To identify the set of genetic interventions, we search for MCS

reactions that are required to be removed to guarantee a yield greater than a desired product yield

threshold without demanding growth. This method retains EFMs on the µ = 0 hyperplane above a

yield threshold, and may also retain EFMs contained within the production envelope which are growth

coupled. Then, the ValveFind algorithm ranks cutset designs by their orthogonality score by exploiting


95 | P a g e

our theory on the overlap of EFMs. Efficient designs that reduce elementary flux modes in the

hyperplane to a single point if possible (depending on the threshold product yield demanded17) and

will have a high score. However, those designs that retain growth coupled EFMs present in a small

high yield region of the production envelope will have a lower score.

For the sugar substrates in this study, we used 1 mol/mol as the minimal yield threshold.

Substrate uptake reactions were considered at 20 mmol∙gDW-1h-1. Due to the dependency of the cut

sets on initial exchange reactions permitted in the model, these were included or excluded on a case

by case basis, and is described further in the sample results of the Supplementary.

With MCS identified, the ValveFind algorithm applies flux variability analysis (FVA) on each

reaction in each cut set to determine the ability of every reaction within every cut set to restore growth.

Within a given cut set, a reaction that can restore the maximum growth (closest to the wild type growth

rate) is designated as a candidate for the metabolic valve. The cut set is then identified as a possible

candidate for having a branched network topology since it contains metabolic valve reaction. Its

orthogonality score is then calculated. The scores calculated for the resulting network(s) can help

confirm their orthogonality and eventually rank them as such. These sets are then manually curated

to determine their suitability towards orthogonality and branched design.

5.4.3 Thermodynamic and Protein Cost Estimations

We used the methodology described by Flamholz et al.9 to calculate the thermodynamic driving force

for each reaction and the corresponding protein costs. The protein cost of a reaction represents the

amount of energy required to be expended by the cell such that a non-zero net flux is possible through

the reaction.

The estimated cost accounts for the thermodynamics and the kinetics of the enzyme with

respect to its interaction with the substrates and products of the reaction and is also a function of the

forward flux flowing through the reaction. We used the LP formulation9 to estimate the minimum

protein cost for every reaction in a network with physiological, thermodynamic, and kinetic constraints

on metabolite concentrations and reaction fluxes. A Michaelis-Menten kinetic rate law formulation

was used to describe reaction fluxes using parameters kcat and Km obtained from in vitro enzyme assays

data in the literature.


96 | P a g e

5.5 References

1. Dugar, D. & Stephanopoulos, G. Relative potential of biosynthetic pathways for biofuels and bio-based products. Nat. Biotechnol. 29, 1074–1078 (2011).

2. Burgard, A. P., Pharkya, P. & Maranas, C. D. Optknock: a bilevel programming framework for identifying gene knockout strategies for microbial strain optimization. Biotechnol. Bioeng. 84, 647–57 (2003).

3. Machado, D. & Herrgård, M. Co-evolution of strain design methods based on flux balance and elementary mode analysis. Metab. Eng. Commun. 2, 85–92 (2015).

4. Lewis, N. E., Nagarajan, H. & Palsson, B. Ø. Constraining the metabolic genotype-phenotype relationship using a phylogeny of in silico methods. Nat. Rev. Microbiol. 10, 291–305 (2012).

5. Long, M. R., Ong, W. K. & Reed, J. L. Computational methods in metabolic engineering for strain design. Curr. Opin. Biotechnol. 34, 135–41 (2015).

6. Heinrich, R., Montero, F., Klipp, E., Waddell, T. G. & Meléndez-Hevia, E. Kinetic and thermodynamic constraints for the structural design of glycolysis. Nonlinear Anal. Theory, Methods Appl. 30, 1793–1804 (1997).

7. Court, S. J., Waclaw, B. & Allen, R. J. Lower glycolysis carries a higher flux than any biochemically possible alternative. Nat. Commun. 6, 1–8 (2015).

8. Biologie, S. & Biophysik, B. Minimization of intermediate concentrations as a suggested optimality principle for biochemical networks. 1991, 443–455 (1991).

9. Flamholz, A., Noor, E., Bar-Even, A., Liebermeister, W. & Milo, R. Glycolytic strategy as a tradeoff between energy yield and protein cost. Proc. Natl. Acad. Sci. U. S. A. 110, 10039–44 (2013).

10. Helnrlch, R., Montero, F., Klipp, E., Waddell, T. G. & Melendez-Hevia, E. Theoretical Approaches to the Evolutionary Optimization of Glycolysis. Thermodynamic and Kinetic Constraints. Eur. J. Biochem. 243, 191–201 (1997).

11. Pfeiffer, T. & Schuster, S. Game-theoretical approaches to studying the evolution of biochemical systems. Trends Biochem. Sci. 30, 20–25 (2005).

12. Stelling, J., Sauer, U., Szallasi, Z., Doyle, F. J. & Doyle, J. Robustness of cellular functions. Cell 118, 675–685 (2004).

13. Csete, M. & Doyle, J. Bow ties, metabolism and disease. Trends Biotechnol. 22, 446–450 (2004).

14. Caetano-Anollés, G. et al. The origin and evolution of modern metabolism. Int. J. Biochem. Cell


97 | P a g e

Biol. 41, 285–297 (2009).

15. Soh, K. C. & Hatzimanikatis, V. DREAMS of metabolism. Trends Biotechnol. 28, 501–8 (2010).

16. Guterl, J.-K. et al. Cell-free metabolic engineering: production of chemicals by minimized reaction cascades. ChemSusChem 5, 2165–72 (2012).

17. Klamt, S. & Mahadevan, R. On the feasibility of growth-coupled product synthesis in microbial strains. Metab. Eng. 30, 166–78 (2015).

18. Tai, Y.-S. et al. Engineering nonphosphorylative metabolism to generate lignocellulose-derived products. Nat Chem Biol 12, 247–253 (2016).

19. Liebermeister, W. & Klipp, E. Bringing metabolic networks to life: convenience rate law and thermodynamic constraints. Theor. Biol. Med. Model. 3, 41 (2006).

20. Liebermeister, W., Uhlendorf, J. & Klipp, E. Modular rate laws for enzymatic reactions: Thermodynamics, elasticities and implementation. Bioinformatics 26, 1528–1534 (2010).

21. Ravasz, E., Somera, A. L., Mongru, D. A., Oltvai, Z. N. & Barabási, A.-L. Hierarchical Organization of Modularity in Metabolic Networks. Science (80-. ). 297, 1551–1555 (2002).

22. Kashtan, N. & Alon, U. Spontaneous evolution of modularity and network motifs. Proc. Natl. Acad. Sci. U. S. A. 102, 13773–8 (2005).

23. Kreimer, A., Borenstein, E., Gophna, U. & Ruppin, E. The evolution of modularity in bacterial metabolic networks. Proc. Natl. Acad. Sci. U. S. A. 105, 6976–81 (2008).

24. Xu, P. et al. Modular optimization of multi-gene pathways for fatty acids production in E. coli. Nat Commun 4, 1409 (2013).

25. Sheppard, M. J., Kunjapur, A. M., Wenck, S. J. & Prather, K. L. J. Retro-biosynthetic screening of a modular pathway design achieves selective route for microbial synthesis of 4-methyl-pentanol. Nat Commun 5, (2014).

26. Qin, J. et al. Modular pathway rewiring of Saccharomyces cerevisiae enables high-level production of L-ornithine. Nat Commun 6, (2015).

27. Smith, R. B. and E. The compositional and evolutionary logic of metabolism. Phys. Biol. 10, 11001 (2013).

28. Kochanowski, K. et al. Functioning of a metabolic flux sensor in Escherichia coli. Proc. Natl. Acad. Sci. U. S. A. 110, 1130–5 (2013).

29. Chubukov, V. & Sauer, U. Environmental Dependence of Stationary-Phase Metabolism in Bacillus subtilis and Escherichia coli. Appl. Environ. Microbiol. 80 , 2901–2909 (2014).

30. Travick, J. D., Burk, M. & Burgard, A. P. Microogranisms and Methods for Conversion of Syngas and Other Carbon Sources to Useful Products. (2010). doi:20100304453A1


98 | P a g e

31. Sonntag, F. et al. Engineering Methylobacterium extorquens for de novo synthesis of the sesquiterpenoid α-humulene from methanol. Metab. Eng. 32, 82–94 (2015).

32. Kalyuzhnaya, M. G. et al. Highly efficient methane biocatalysis revealed in a methanotrophic bacterium. Nat. Commun. 4, 2785 (2013).

33. Murarka, A., Dharmadi, Y., Yazdani, S. S. & Gonzalez, R. Fermentative utilization of glycerol by Escherichia coli and its implications for the production of fuels and chemicals. Appl. Environ. Microbiol. 74, 1124–35 (2008).

34. da Silva, G. P., Mack, M. & Contiero, J. Glycerol: a promising and abundant carbon source for industrial microbiology. Biotechnol. Adv. 27, 30–39 (2009).

35. Anesiadis, N., Cluett, W. R. & Mahadevan, R. Dynamic metabolic engineering for increasing bioprocess productivity. Metab. Eng. 10, 255–66 (2008).

36. Anesiadis, N., Kobayashi, H., Cluett, W. R. & Mahadevan, R. Analysis and Design of a Genetic Circuit for Dynamic Metabolic Engineering. ACS Synth. Biol. 2, 442–452 (2013).

37. Venayak, N., Anesiadis, N., Cluett, W. R. & Mahadevan, R. Engineering metabolism through dynamic control. Curr. Opin. Biotechnol. 34, 142–52 (2015).

38. Brockman, I. M. & Prather, K. L. J. Dynamic metabolic engineering: New strategies for developing responsive cell factories. Biotechnol. J. 10, 1360–1369 (2015).

39. Xu, P., Li, L., Zhang, F., Stephanopoulos, G. & Koffas, M. Improving fatty acids production by engineering dynamic pathway regulation and metabolic control. Proc. Natl. Acad. Sci. U. S. A. 111, 11299–304 (2014).

40. Zha, W., Rubin-Pitel, S. B., Shao, Z. & Zhao, H. Improving cellular malonyl-CoA level in Escherichia coli via metabolic engineering. Metab. Eng. 11, 192–198 (2009).

41. Solomon, K. V., Sanders, T. M. & Prather, K. L. J. A dynamic metabolite valve for the control of central carbon metabolism. Metab. Eng. 14, 661–671 (2012).

42. Soma, Y., Tsuruno, K., Wada, M., Yokota, A. & Hanai, T. Metabolic flux redirection from a central metabolic pathway toward a synthetic pathway using a metabolic toggle switch. Metab. Eng. 23, 175–84 (2014).

43. Zhang, F., Carothers, J. M. & Keasling, J. D. Design of a dynamic sensor-regulator system for production of chemicals and fuels derived from fatty acids. Nat. Biotechnol. 30, 354–359 (2012).

44. Brockman, I. M. & Prather, K. L. J. Dynamic knockdown of E. coli central metabolism for redirecting fluxes of primary metabolites. Metab. Eng. 28, 104–13 (2015).

45. Soma, Y. & Hanai, T. Self-induced metabolic state switching by a tunable cell density sensor for microbial isopropanol production. Metab. Eng. 30, 7–15 (2015).

46. Vemuri, G. N., Eiteman, M. a. & Altman, E. Succinate production in dual-phase Escherichia


99 | P a g e

coli fermentations depends on the time of transition from aerobic to anaerobic conditions. J. Ind. Microbiol. Biotechnol. 28, 325–332 (2002).

47. Anastassiadis, S., Aivasidis, A. & Wandrey, C. Citric acid production by Candida strains under intracellular nitrogen limitation. Appl. Microbiol. Biotechnol. 60, 81–87 (2003).

48. Youngquist, J. T., Rose, J. P. & Pfleger, B. F. Free fatty acid production in Escherichia coli under phosphate-limited conditions. Appl. Microbiol. Biotechnol. 97, 5149–5159 (2013).

49. Collins, J. J., Gardner, T. S. & Cantor, C. R. Construction of a genetic toggle switch in Escherichia coli. Nature 403, 339–342 (2000).


51. Bogorad, I. W., Lin, T.-S. & Liao, J. C. Synthetic non-oxidative glycolysis enables complete carbon conservation. Nature 502, 693–697 (2013).

52. Radek, A. et al. Engineering of Corynebacterium glutamicum for minimized carbon loss during utilization of d-xylose containing substrates. J. Biotechnol. 192, Part A, 156–160 (2014).

53. Zhuang, K., Yang, L., Cluett, W. R. & Mahadevan, R. Dynamic strain scanning optimization: an efficient strain design strategy for balanced yield, titer, and productivity. DySScO strategy for strain design. BMC Biotechnol. 13, 1–15 (2013).

54. Van Dien, S. From the first drop to the first truckload: commercialization of microbial processes for renewable chemicals. Curr. Opin. Biotechnol. 24, 1061–1068 (2013).

55. Klamt, S., Stelling, J., Ginkel, M. & Gilles, E. D. FluxAnalyzer: exploring structure, pathways, and flux distributions in metabolic networks on interactive flux maps. Bioinformatics 19, (2003).

56. Gagneur, J. & Klamt, S. Computation of elementary modes: a unifying framework and the new binary approach. BMC Bioinformatics 5, 1–21 (2004).

57. Pfister, R., Schwarz, K. A., Janczyk, M., Dale, R. & Freeman, J. B. Good things peak in pairs: a note on the bimodality coefficient. Front. Psychol. 4, 700 (2013).

58. Orth J, F. R. P. B. Reconstruction and Use of Microbial Metabolic Networks: the Core Escherichia coli Metabolic Model as an Educational Guide. EcoSal Plus (2010). at <http://www.asmscience.org/content/journal/ecosalplus/10.1128/ecosalplus.10.2.1>

59. von Kamp, A. & Klamt, S. Enumeration of Smallest Intervention Strategies in Genome-Scale Metabolic Networks. PLoS Comput. Biol. 10, (2014).

60. Klamt, S. & Gilles, E. D. Minimal cut sets in biochemical reaction networks. Bioinformatics 20, 226–234 (2004).


100 | P a g e

5.6 Extended Data Set: Redesigning metabolism based on orthogonality principles

5.6.1 Synthetic pathway design

Pathway design requires consideration for three specific aspects of any metabolic pathway; the desired product to be produced, the input substrate and the kinetics of enzyme associated with the pathway. In addition, they also need to consider network energy and redox constraints. We cover these criteria in detail below.

One important consideration is that the pathways we model are a simplification of the cell’s metabolism and we purposefully at this stage neglect the complexity that arises by considering the metabolism in more detail. For example, we largely neglect amino acid biosynthesis pathways in the cell. We do this because calculation of the orthogonality score is a computationally burdensome problem. This simplification does not preclude the design of orthogonal pathways and identification of valves in genome-scale networks as the underlying algorithm is based on minimal cut sets that have been extended to genome-scale networks(von Kamp & Klamt 2014)

5.6.2 Substrate selection

The first step in the re-design of metabolic pathways is the selection of a substrate. Our analysis began by identifying substrates that would make good candidates for biotransformation into our target compound(s). Any compound that is renewable, inexpensive, non-toxic and capable of being transported into the cytoplasm is a suitable candidate. Apart from glucose, we found ethylene glycol to satisfy all the above criteria.

5.6.3 Selection of intermediate precursor(s)

In the case of orthogonal pathway design, the ability to attain the orthogonal structure in Figure 1a for metabolic pathways is also dependent on the chosen intermediate precursor used as the branching point for metabolic control. Subsequent to identifying a suitable substrate, a suitable precursor metabolite, we need to identify a common metabolite for both product and biomass production pathways. This precursor metabolite should serve as a possible growth substrate for the cell. Pyruvate as an example, is a key precursor from which for instance, succinate can be synthesized. It can also be used as a growth substrate. Since in fact most industrial compounds are produced from a small subset of key metabolic precursors from the central carbon metabolism, the list of suitable precursor metabolites is often very small. Other examples include the production of 1,4-butanediol from acetyl-coA, malonic acid from pyruvate, and isoprenoids from erythrose-4-phospate. These key precursors serve as branch points for building synthetic pathways for natural or non-native compounds. We use a pathway predictor algorithm to identify the synthetic pathways to this precursor metabolite.


101 | P a g e

5.6.4 Redox and ATP Cost.

The conversion of the input substrate must produce reducing equivalents in excess of what is consumed by product formation. This criterion guarantees that sufficient energy is available for cell growth and maintenance requirements. The criteria arise because energy requirements cannot be met by substrate level phosphorylation. Since very few enzymes are capable of producing ATP from substrate level phosphorylation, and the metabolites involved in these reactions are also well connected to other parts of the cell’s metabolic network. This requirement hinders the orthogonality of the network. Hence, in order to support orthogonal pathways, cellular ATP requirements need to be met by enzymes not involved in substrate level phosphorylation reactions. Oxidative phosphorylation to generate ATP becomes the alternate choice to satisfy cellular ATP needs. Finally, we apply all these criteria in arriving at a suitable synthetic pathway for a given substrate-product pairing.

5.6.5 Analysis of a Simple Branched Structure

The main text of the manuscript provides a short analysis of a toy network to provide an understanding of how the orthogonality score is calculated and applied. Here, we provide a short derivation of the theory behind the orthogonality score based on the ideal network structure.

Given an ideal branched structure (Figure S1), independent from one another and producing either a product P or biomass X, then its elementary flux modes are described by EFM1 and EFM2. This independence arises when no input or co-factor from one branch is required by another. It is readily shown that the orthogonality score for this network is 0.6, as calculated below. Furthermore, in a simple, ideal structure, it can be seen that the score for this network is function of the total length of EFM2 and the substrate utilization reactions, v1 and v2, up to the branch point metabolite, C. Hence, if there was one more common reaction present between A and C (for a total of 3) then the orthogonality score would decrease to 0.5 or if one less reaction was present between A and C (for a total of 1) then the orthogonality score would increase to 0.75. Thus, this example provides a simplified model of how orthogonality can be calculated and how it is a function of the shared elements (reactions) of the P and X forming EFMs. Hence, in summary, an orthogonality score greater than 0.5 means that fewer than half the reactions are shared relative to the length of the biomass producing EFMs.

Figure S1. Orthogonality Calculation for a Branched Structure. EFM1 is indicated in green. EFM2 is indicated in blue.


102 | P a g e

Let us extend this analysis a little further however to understand how this toy model might apply to a cellular network – and why the orthogonality score as calculated is a suitable metric for larger models.

We begin by noting that that for the branched network structure with two independent arms that there are only two EFMs. The optimal solution has flux through only one branch but the flux distribution, however, can be defined by a linear combination of these two EFMs by

𝒗 = ∑ 𝛼𝑘𝒆𝑘

𝐾

𝑘=1

where every flux v is a non-negative sum of EFMs: 𝒆1 … 𝒆𝑘 multiplied by its corresponding weighting factor

𝛼𝑘 and ∑ 𝛼𝑘 = 1. In the above example, there are only two EFMs and each is an optimal solution EFM for the production of either P or X. Thus, in the above example, the orthogonality score is determined solely by these two EFMs.

We can also observe that for any network that can produce P or X independently, there will always exist at

least two elementary flux modes 𝒆𝑘 each that is characterized by the smallest set of interactions between

the independent production of P and of X. Let these EFMs be 𝒆1 and 𝒆2 (EFM1 and EFM2 in the above case). Therefore, the largest orthogonality score for the network will exist between these two elementary

flux modes and will occur when 𝒆1 is shortest.

Let us consider that the network is modified in such a way that there is now one additional EFM present in

the network, 𝒆3, such that there are two EFMs producing P and still one producing X. Therefore, the

orthogonality score is now also a function of the dot product (shared elements) between 𝒆2 and 𝒆3. Since

𝒆3 is by definition different than 𝒆1, then by considering this new EFM, the orthogonality score can either increase or decrease. If however, the orthogonality score increases, then this can only occur if the number

of reactions shared by 𝒆3 with 𝒆2 is less than 𝒆1 with 𝒆2. However, if this was the case, then 𝒆1 is no longer

the most orthogonal elementary flux mode as this violates our starting condition. Therefore, 𝒆3 must have

more shared elements with 𝒆2 than 𝒆1 has with 𝒆2 which will necessarily reduce the orthogonality score. It follows, that the addition of any elementary flux mode outside the ideal branched network structure must always reduce the orthogonality score.

Now recognize that the most orthogonal P forming 𝒆𝑖 to B is not necessarily optimal flux (i.e. highest yield) EFM. Hence, an FBA solution which only finds the optimal flux condition EFM is not sufficient to fully characterize the network interaction space. In natural metabolism, the optimal flux EFM is not necessarily the most orthogonal pathway possible when considering the global repertoire of known metabolic enzymes.

In our present work, we are concerned about the design of substrate utilization pathways and their impact on strain design. In this endeavor, since we are engineering pathways a priori we can ignore from

consideration those 𝒆𝑖 that form P but are also low yield since this does not meet design criteria for yield. Hence, it follows that for synthetic pathways like those examples in our case study, or in the simplified branch structure above, the highest yield EFM must also be the most orthogonal. And our central task is to engineer pathways in which the optimal flux EFM corresponds to the most orthogonal EFM from the entire set of possible pathways.


103 | P a g e

Supplementary II

Natura

l Glucos

e (EMP)

Natural Glucose (ED

)

Synthetic

Glucose

Natural

Xylose

Synthetic Xylose

Glycerol

Ethylene

Glycol 1

Ethylene Glycol

2

Ethylene Glycol

3

Succinic Acid

0.41 82236 11.2

0.45 67059

8.6

0.56 3610 3.6

0.36 86499 12.8

0.57 2233 6.6

0.48 17943

9.1

0.62 464 3.3

0.54 1119 5.2

0.36 34437 11.9

Isobutanol

0.48 38202 10.5

0.47 29451

7.7

0.54 2126 4.9

0.37 24798 12.3

0.55 2974 5.9

0.49 4095 8.3

0.61 436 2.5

0.61 145 3.7

0.38 16663 11.0

Adipic Acid

0.44 26672 11.0

0.45 24114

8.2

0.54 1287 5.5

0.35 20468 12.5

0.54 4025 5.9

0.47 4602 4.7

0.57 396 3.3

0.52 493 5.2

0.34 18663 12.0

Ethanol 0.44 72974 11.0

0.45 79170

8.4

0.58 2319 4.7

0.36 67148 12.4

0.54 4203 6.0

0.47 11510

8.4

0.59 1635 3.8

0.61 263 4.0

0.39 13967 11.1

1,4-Butanediol

0.46 319359

11.7

0.44 120741

8.3

0.54 6356 6.7

0.36 198596

13.3

0.55 8813 6.7

0.46 37430

9.5

0.57 1086 4.2

0.53 1719 5.6

0.35 85438 12.4

2,3-Butanediol

0.47 24006 10.5

0.48 19096

7.5

0.56 1756 3.0

0.40 16751 11.9

0.55 2974 5.9

0.53 2652 7.2

0.62 436 2.5

0.66 91 3.4

0.39 25088 10.7

Table S1. Orthogonality scores for a variety of substrates, products and pathways are shown in bold text. The Total Precursor Supporting Reactions are shown in italics while the Average Precursor Reactions/EFM appears as normal text. Broadly, the results support our finding that substrate selection can play in effectively engineering growth independent chemical production. Finally, Figure S2 shows the correlation between the orthogonality scores from Table S1 and the Average Precursor Reactions/EFM that describe the cell’s ability to produce a biomass precursor. The individual pathways used in the modelling along with their metabolic pathways are described in Supplementary V.


104 | P a g e

Figure S2: Correlation between Orthogonality Score and the Average Precursor Reactions per EFM. The following reactions were used to determine the number of biomass precursor forming reactions. {'PGI', 'FBP', 'TKT2', 'TPI', 'FBA', 'GAPD', 'PGK', 'ENO', 'PPCK', 'PPS', 'ME1', 'PYK', 'PDH', 'PPC', 'MDH', 'SUCOAS', 'AKGDH', 'ICDHyr', 'RPI', 'TKT1'};

5.6.6 Results of orthogonality score and biomass supporting reactions are

generalizable

Production of succinic acid by various different substrates and pathways was used in the main text as a case study to explore concepts of orthogonality and convey the importance that engineering substrate utilization pathways has on various aspects of metabolic engineering. In Supplementary II, our goal is to show that the conclusions derived from the case study is broadly generalizable to various different substrates, their catabolic pathways and across many different products. To that end, we have expanded our analysis to five additional products that have been routinely cited in the literature and are of commercial interest. These products have been explored in four substrates across a total nine different pathways. The orthogonality score for the various combinations of substrates and products are shown above in Table S1. Included in this table is also the total number of elementary flux modes that support biomass pre-cursor forming reactions as well as the average number of biomass pre-cursor forming reactions. Together, these results are meant to mirror the analysis in Table 1 of the main text. In general, we find consistency between the results in the main text and Table S1. Below we highlight four observations of interest.

5.6.7 A Study of Counter Examples

One benefit of the orthogonality metric for evaluating the predisposition of a metabolic structure to produce a desired chemical is that it reveals, (i) inherent dependency that substrate selection has on growth independent chemical production, and (ii) that the substrate product pairing is not pathway agnostic for ideal networks. These observations, which are made using an unbiased and quantifiable measure, provides a rational basis of design for the metabolic engineer. Two surprising exceptions


105 | P a g e

help to provide a rationale for why several of the observations made in the central text cannot necessarily be derived intuitively for all cases, but rather that a metric provides a basis for understanding how we can achieve orthogonal design of metabolism and rework metabolism in a way that is consequential for metabolic engineering. Hence, first we examine a case where native metabolism is orthogonal towards chemical production and in a second case where a synthetic metabolism is not orthogonal.

5.6.8 Exception to Natural Metabolism is Not Orthogonal

We described in the Section 2.2 that natural metabolism was not orthogonal towards succinic acid production. We also described in Section 2.4 of the main text that the substrate product pairing is an essential component of determining how permissible network structures are towards two independent production tasks (biomass and chemical). Those general observations are largely invariant in the presence of the further analysis presented in Table S1. However, Table S1 does show in interesting result for the production of 2,3-butanediol from glycerol. The orthogonality score for this paring was determined to be 0.53 – greater than any of the natural sugar pathways, though still less than any of the synthetic pathways. Nonetheless a value greater than 0.5 indicates for us a greater ability of the network to support growth independent production. Hence, while instances of natural metabolism exhibiting orthogonal behaviour is uncommon, this exception underscores the role of an unbiased metric rather than intuition in assessing metabolic network structures.

5.6.9 Exception to Non-native metabolism as obligate orthogonal pathways

Through the text we describe how synthetic pathways can be used to achieve orthogonal metabolic structures. In the case study on succinic acid, orthogonality was achieved on by way of synthetic pathways but at the same time, it was also dependent on the substrate selection (ex. xylose). Here we provide another exception to the idea that non-native metabolism and pathways are always orthogonal pathways which arises from Table S1. As in the last case, we believe this exception strengthens our case in the text.

The first is presented by examining the three different pathways of ethylene glycol utilization. Despite being a non-native substrate for E. coli, the degree to which ethylene glycol is capable of supporting biomass independent chemical production is highly dependent on the pathway and the location the pathway enters the cell’s natural metabolism. Ethylene glycol variant #3 has substantially lower scores than variant #2 and #1 because it enters the pentose phosphate pathway. Hence, this finding mirrors our work in the central text that examines xylose utilization through the pentose phosphate pathway or the Weimberg pathway. It supports the idea that not all substrate utilizing pathways evoke a similar behaviour in the cell towards chemical production and that substrate utilization needs to be considered on a case by case basis.


106 | P a g e

5.6.10 Co-factor and other network effects are not captured by a simple

branched reaction network

Orthogonality was determined by the elementary flux modes of the network and we found that branched structures exhibited the highest scores. This can lead one, falsely, to the conclusion that it is self-evident that synthetic pathways with a branched topology are orthogonal. To dissuade the reader from making generalizations on the network topology without examining the underlying characteristics of the network, we looked for a counter example that showed a branched structure, but upon determining elementary flux modes and calculating the orthogonality score led one to the opposite conclusion. We find that glycerol conversion to 1,3-propanediol to be an ideal example of the case (Figure S3).

Glycerol can be converted to the product or to biomass by either branch of this ideal network structure. Then one would expect that the orthogonality score for this network to produce 1,3-PDO would be very high and the clear branching should allow the network two function independently and performing either tasks. Instead, what we find is that the orthogonality score for this network is only 0.49. In other words, there is considerable overlap, more than to be expected at least by quick inspection, between the pathways and that this score is far smaller than for networks that appear far more complex (ex. ethylene glycol to succinic acid). Why is this the case? The lower score arises from the network interactions caused by NADH requirements. Thus, this pathway is a simplified example of how co-factor interactions, which are global influencers of the metabolism, might have wide effects that may not be discernable by a cursory look at the pathway. When we take away the NADH requirement for 1,3-PDO synthesis and instead produce 3-hydroxypropionic acid which is more oxidized, the orthogonality score jumps to 0.54.

Figure S3. Simplified pathway showing production of 1,3-propanediol (1,3-PDO) and 3-hydroxypropionic acid (3HP) from glycerol. 1,3-PDO requires excess NADH while 3HP does not.

2.2.4 The Pentose Phosphate Pathway is not suitable for growth and production independence


107 | P a g e

From Table S1 we find in general that natural pathways that utilize xylose for the biological production of chemicals are substantially worse in their ability to uncouple product and biomass pathways. These results provide a theoretical basis for understanding xylose or other substrate assimilation pathways that do not pass through the pentose phosphate pathway. The natural xylose pathway and the ethylene glycol variant #3 are both examples of these types of pathways and exhibit the lowest orthogonality scores.

5.6.11 Valve Selection is also a determinant of metabolic independence

A complete list of the cut-sets determined by the ValveFind algorithm for glucose conversion of succinic acid by the synthetic pathway studied in this publication is shown in Supplementary III. Additionally, Table S2 shows a sample cut-set for the most orthogonal cases of xylose and ethylene glycol utilization for the different product pairing identified in Table S1 as evidence that the approaches and the conclusions laid out in the case study on succinic acid are generalizable to a variety of products, substrates and pathways for substrate assimilation. Out of this analysis, we wanted to highlight a particular finding that sheds novelty in our approach to designing cells with a focus on substrate utilization, namely the selection of reactions suitable as metabolic valves.

Any given cut-set contains a set of genetic deletions as well as the identification of a metabolic valve reaction. In many of these cases, however, more than one reaction is identified as a candidate metabolic valve for a different set of deletions. The question then arises which reaction is more suitable? For example, consider the following designs from Supplementary III.

1) Score: 0.53, Valve: G6PDH2r, Cutset: TKT2 ME1 NADTRHD PPCK

2) Score: 0.51, Valve: TKT2, Cutset: G6PDH2r ME1 NADTRHD PPCK

Despite the similarity in the design, G6PDH2r is suggested to be a slightly better valve that TKT2. Given that recent developments by Prather et al. in controlling glucose utilizing reactions2 in metabolic engineering, this example is also grounded in physiological reality. The results point to the notion that metabolic control based in rational design may be an essential component of practically realizing these types of dynamic strategies in industrial strains. In another case cut-sets (3) and (4) are reasonably similar in that the deletion set differs by only 1 reaction (PYK vs PPS) yet the metabolic valve selection and the orthogonality score differ substantially supporting the earlier conclusions that network interactions not immediately discernible are systemic and influence strain design.

3) Score: 0.41, Valve: PGK, Cutset: AKGDH ME1 ME2 PYK 4) Score: 0.55, Valve: PPCK, Cutset: AKGDH ME1 ME2 PPS

Engineering Escherichia Coli for Utilization of Ethylene Glycol

108 | P a g e

This chapter has been submitted as a manuscript to Biotechnology and Bioengineering

Abstract

A considerable challenge in the development of bioprocesses for producing chemicals and fuels has

been the high cost of feedstock relative to oil prices that make these processes uncompetitive with

their conventional petrochemical counterparts. Hence, in the absence of high oil prices for a

foreseeable future, which was the main driver for white biotechnology, there has a shift in the industry

to instead produce higher value compounds such as fragrances for cosmetics. Yet still, there is a need

to address climate change and develop biotechnological approaches for producing large market, lower

valued chemicals and fuels. In this work, we study ethylene glycol, a novel feedstock that we believe

has promise to address this challenge. We engineer E. coli to consume ethylene glycol and as a case

study, for chemical production, examine glycolate production. The best fermentation performance led

to the production of 10.4 g/L of glycolate after 112 hours of production time. Our experiments lead

us to realize that oxygen concentration is an important factor in assimilation of MEG as a substrate.

We also find that the uptake rates for ethylene glycol are sufficient to satisfy commercial benchmarks

for productivity and yield. Finally, our use of metabolic modelling sheds light on the intracellular

distribution through the central metabolism implicating flux to 2-phosphoglycerate as the primary

route for MEG assimilation. Overall, our work leads us to conclude that ethylene glycol is a useful

platform for commercial synthesis of fuels and chemicals that may achieve economic parity with

petrochemical feedstocks while sequestering carbon dioxide.


109 | P a g e

6.1 Introduction

Biotechnological approaches to addressing climate change and the need to sequester carbon dioxide

have focused on the development of microbial strains engineered to produce chemicals and fuels

derived from renewable sources of sugar. Despite the considerable success at engineering these strains,

failure at the commercial scale belies the immense challenge in the financial viability of these

technologies in the face of low oil prices and expensive feedstock costs. In response, non-sugar

feedstocks have been put forward as alternatives to compete efficiently with glucose based

bioprocesses. For example, methane and syngas fermentations are currently under intense study and

are also the focus of commercial development1–3. Formate is another chemical that has been suggested

as a replacement for glucose since it can be produced from carbon dioxide and because of its inherent

compatibility with biological processes4,5. However, its utility as feedstock for biological processes

suffers from a number of drawbacks. The most evident drawback is the absence of pathways for its

assimilation in the metabolism of traditional workhorse organisms such as yeast or E. coli. The oxidized

nature of the substrate also results in carbon loss to enable synthesis of NAD(P)H co-factors that

support product and ATP formation, and the requirement for high transport rates into the cell to

achieve productivities similar to glucose or xylose fermentation. Hence, while appealing, the technical

challenges are numerous.

Nonetheless, this appeal arises from the fact that formic acid can be generated

electrochemically from CO2. A one electron pair reduction of one mole of CO2 produces one mole

of formic acid. However by tailoring the catalyst and the reduction potential, multi-electron reduction

can be achieved and it is possible to produce a variety of different reduced carbon species6. Not

surprisingly, biological processes have been used to produce many of these same chemicals that are

typically produced by the petrochemical industry including 1-propanol, acetate, ethylene, etc7–9. Our

work, here, is motivated by the observation that like formate, these other carbon containing compound

derived by the electrochemical reduction of CO2 are feasible growth substrates for biological processes

and this should merit their consideration as alternative feedstocks for bioprocesses.

In evaluating these substrates as potential replacements for glucose, it is important to recognize

that many cannot be naturally catabolized by traditional industrial workhorses. Hence, similar to

formate, the metabolic engineering of substrate utilization pathways is necessary. Additionally, many


110 | P a g e

of the potential replacements are toxic and not compatible with bioprocesses. Others, while technically

feasible as inputs to biological processes, suffer from poor faradic efficiency in electrochemical

reactors. Hence, after screening from a list of products that can be generated electrochemically, it

becomes apparent that only a few can be realized as practical substitutes for glucose6. Finally, beyond

toxicity and efficiency, which can be evaluated in a relatively straightforward manner, evaluating the

feasibility of a new substrate for bio-based chemical production can be obfuscated by how its

utilization is linked to the highly interconnected metabolic network. Indeed, refactoring large

metabolic pathways into heterologous hosts has proven challenging in the past10. One method that

may help to explain why a new substrate performs poorly examines the metabolic pathway that

supports a substrate for chemical production in relation to the cell’s entire metabolism.

In an earlier study22 we characterized this relationship by calculating the interactions between

two competing objectives of cellular systems, growth and chemical production. The theory laid out

how the underlying network structure gives way to growth independent chemical production. That

relationship was captured by a mathematical framework using elementary flux modes to measure the

interconnectedness of the cell system and the desired objectives. Hence, we defined a metric to

measure the orthogonality of the chemical production pathways with respect to biomass production.

We found that the organization of ideal metabolic structures designed to minimize cell-wide

interactions had a characteristic branched topology. This type of orthogonal structure could be

exploited for two stage fermentation. Furthermore, an important finding from that study was that

glucose, while a common substrate for industrial fermentation, is not ideally suited for chemical

production objectives. Instead, substrate selection should be based on the chemical targeted for

production. Among the various substrates and products, we identified that ethylene glycol was a highly

promising substrate for orthogonal production of a variety of chemicals because it minimized the

interactions between biomass and chemical producing pathways.

Therefore, among the variety of different chemicals that can be produced electrochemically,

ethylene glycol is a promising, unconventional feedstock. It is produced today primarily by the

petrochemical industry from ethylene; however, a process for making ethylene glycol from CO2 has

shown early promise, and is currently the focus of industrial scale up. In this regard, its utilization as

a feedstock for biological processes is important because it can serve as a replacement for glucose in

the modern bioprocess.


111 | P a g e

Motivated by these considerations, we engineer and characterize E. coli as a biocatalyst capable

of consuming ethylene glycol as a carbon source and explore its application as a novel substrate for

industrial bioprocesses.

We do note though, that while ethylene glycol to date as largely been studied as a target for

production in metabolic engineering applications27-30 because of its importance in manufacturing

plastics and polyesters, we believe its consideration as a feedstock for producing other chemicals also

merits consideration. This platform for growth and chemical production is then applied to a case

study for glycolic acid production. This case study attempts to validate our orthogonal approach for

chemical production, relating the network topology and two-stage fermentation. Conventional

approaches to glycolic acid in E. coli have instead focused on using glucose as the substrate, and

implementing genetic strategies that couple production to growth. Several studies have been published

that have examined glycolic acid production from glucose and xylose. The highest of these reports

achieves titers of 56.44 g/L and a yield of 0.62 g/g11. To our knowledge, only three studies have

examined ethylene glycol conversion to glycolic acid as a biotransformation12–14. However, in this

work, we have thoroughly characterized the metabolism and growth physiology of E. coli growing on

ethylene glycol. We find that while growth rate is markedly slow relative to growth on glucose, with a

doubling time of 3.85 hours on ethylene glycol, the substrate uptake rate is sufficiently high at up to 5

mmol/gDW-h to be relevant for industrial production. Glycolate, which required micro-aerobic

conditions, reached titres of 10.4 g/L at a maximum theoretical yield of 66%. Overall, we find that

understanding the growth characteristics of the cell and a model on glycolate production shows that

using ethylene glycol has potential for replacing glucose in industrial bioprocesses in applications where

CO2 streams and renewable electricity are available.

6.2 Materials and Methods

6.2.1 Media and Cultivation Conditions

Cells were grown using lysogeny broth (LB) as per manufacturer’s instructions (Bioshop, Burlington,

ON) for all strain construction and fermentation pre-cultures. When characterizing strains, cell were

grown under M9 minimal media with the following compositions: 1.0 g/L NH4Cl, 3.0 g/L KH2PO4,

6.8 g/L Na2HPO4, 0.50 g/L NaCl. Supplements of yeast extract at 2 g/L were added to minimal


112 | P a g e

media. Ethylene glycol was used as the carbon source as concentrations described in the text. IPTG

was used at a concentration of 1mM when necessary. A trace metal solution was prepared according

to the following composition prepared in 0.1 M HCl per litre and added at a concentration of 1/1000:

1.6 g FeCl3, 0.2 g CoCl2•6 H2O, 0.1 g CuCl2, 0.2 g ZnCl2•4H2O, 0.2 g NaMoO4, 0.05 g H3BO3. 1 M

MgSO4 and 1 M CaCl2 was also added to the media at a concentration of 1/500 and 1/10,000,

respectively. For all cultures, carbenicillin was added as appropriate at 100 µg/mL. Cells were grown

in 250 mL shake-flasks for all characterization experiments and in bioreactors as described.

6.2.2 Culturing Techniques in Reactors

Pre-cultures were grown in LB rich media in 10 mL test tube cultures overnight and transferred fresh

shake-flaks containing LB, 1 mM IPTG and 10 g/L ethylene glycol. After 24 hours, these cells were

harvested by centrifugation, re-suspended in 2mL of residual supernatant and used as inoculum for

bioreactor or minimal media shake-flasks for characterization at 37°C.

Applikon MiniBio500 fermentation vessels were used for cultivating strains in bioreactors.

Dissolved oxygen and pH probes were used in accordance with the manufacturers operating

guidelines. M9 minimal media was used for cultivation in the bioreactor. pH was maintained at 7 with

the addition of 3N KOH. Growth conditions were maintained at 37°C. Dissolved oxygen was

maintained as described in the text. Flowrate was controlled as described using a Books Instruments

mass flow controllers (GF Series) and gas was analyzed using Thermo Scientific™ Sentinel dB mass

spectrometer for online gas measurement.



(HPLC). We used an Aminex 87H column with 5 mM H2SO4 as the eluent and a flowrate of 0.4

mL/min at 50°C. Organic acids were detected at 210 nm. Cell densities of the cultures were

determined by measuring optical density at 600 nm (GENESYS 20 Visible Spectrophotometer). Cell

density samples were diluted as necessary so as to fall within the linear range. A differential refractive

index detector (Agilent, Santa Clara, CA) was used for analyte detection and quantification. Yields were

calculated between two time points, whereas the cumulative yield was calculated between the initial

and final measurements.


113 | P a g e

6.2.4 Plasmids and Strains

fucO and aldA were cloned from E. coli MG1655 genomic DNA and assembled using Gibson

Assembly onto a pTrc99a vector. RBS sequences were placed onto the overhang of the forward

primer. AACAAAATGAGGAGGTACTGAG was the RBS sequence used in front of aldA.

AAGTTAAGAGGCAAGA was the RBS sequence used in front of fucO. The Trc promoter was

used to drive expression. Wild-type strains of E. coli MG1655 were obtained from the Coli Genetic

Stock Centre (Yale).

Table 6-1 Strain and Plasmid Table for Ethylene Glycol Study

Strain or Plasmid Relevant Characteristic Reference or source

E. coli MG1655 Wild-type strain Coli Genetic Stock Center

LMSE11 MG1655 harbouring pfucO1 This study

LMSE12 MG1655 harbouring pfucO2 This study

pTrc99a AmpR, E. coli shuttle vector for regulated gene

expression; Ptrc, pUC18 ori

Amersham

pfucO1 pTrc99a derivative containing the fucO I7L, L8V

and aldA genes from E. coli.

This study

pfucO2 pTrc99a derivative containing the fucO L8M

and aldA genes from E. coli.

This study

6.2.5 Flux Balance Analysis

Flux balance analysis (FBA) was performed using MATLAB R2015a installed with COBRA 2.0

toolbox and using the GLPK linear solver (GNU Project). The genome scale model iAF1260 was

used to perform all modelling. The ATP maintenance reaction was left unchanged at a value of 8.9

mmol/gDW-h. The model was modified by adding a reaction for converting ethylene glycol to

glycolaldehyde using NAD cofactors. Transport of ethylene glycol was modelled as free diffusion and

no proton translocation was included as part of its exchange reaction. Initial characterization of the

cell to model the respiratory quotient was only constrained by its substrate uptake rate which was

measured at 5 mmol/gDW-h. More detailed intracellular flux data were extrapolated by constraining


114 | P a g e

substrate uptake rate as well as glycolate production rates and oxygen uptake rates as determined by

analysis of the off-gas from the process mass-spec during bioreactor cultivation.

6.3 Results

6.3.1 Ethylene glycol is a preferred substrate over formate

In an earlier study, we identified orthogonality as a metric to assess and design efficient metabolic

networks for the production of chemicals. That study defined orthogonality as a quantitative measure

of the interconnectedness between pathways that produce a target chemical and biomass. The

principle focus of that work was to examine how metabolic pathway organization influences chemical

production. In this first section, we apply that methodology to compare formate and ethylene glycol

utilization, both of which can be synthesized electrochemically. We assess the specific role that

substrate selection has on five different chemicals that are important to industry found in Table 4-1.

This analysis allows us to implicitly account for metabolic constraints such as redox and ATP. Glycolic

acid showed the highest orthogonality score between all the substrate product pairs, and hence was

selected as the demonstration product for production of ethylene glycol.

Table 6-1 shows the orthogonality score for these chemicals using ethylene glycol and formate as

carbon sources. Glucose and xylose are also included in the calculations as they provide a reference

against the conventional bio-process. For all chemicals, the orthogonality score is larger for ethylene

glycol than formate and less substrate is required to produce the same quantity of product as well.

The orthogonality metric is a mathematical measure of the set of interactions that each substrate

assimilation pathway has to the cell components outside their pathways. Hence, it implicitly measures

the biological complexity one might expect to ensure that the biomolecular machinery of that pathway

can concurrently function within the cell’s natural metabolism to support biological and chemical

production objectives. Analysis of the metabolism of formate shows its low score arises from its low

degree of reduction which requires flux through the TCA cycle to generate the necessary reducing

equivalents for growth and energy, irrespective of what chemical is produced. The low degree of

reduction is also the reason for low product yields. Hence, this line of network analysis suggests

ethylene glycol is a superior substrate to formate in E. coli. Given higher scores for ethylene glycol


115 | P a g e

utilization, we resolved that ethylene glycol utilization was promising and we compared these results

to sugar metabolism in E. coli for glycolic acid production.

Glycolate is an alpha-hydroxyacid used in the synthesis of a variety of different plastics and polymers,

cosmetics and industrial detergents. Currently, metabolic engineering has established routes to glycolic

acid from glucose and from xylose. Theoretical yields have been dependent on both the substrate

selected as well as the biosynthetic pathway used for production. Examples of glycolate production

from glucose in literature has primarily been demonstrated by the activation of the glyoxylate shunt.

Figure 6-1 shows glycolate production from three different pathways. Production by the glucose is

highly coupled to biomass synthesis, and exhibits the lowest orthogonality score, 0.41. Glycolate

production using xylose has also been demonstrated by the use of a synthetic pathway for xylose

assimilation in E. coli. While this pathway fits partly into an orthogonal criteria for glycolate

production, the concomitant production of pyruvate for every mole of glycolate requires the use of

the cells highly interconnected glyoxylate cycle, to reach theoretical yields. The orthogonality score,

for this reason, is comparatively smaller. The largest orthogonality score was determined to be for

ethylene glycol conversion as a substrate was 0.67. Bioconversion of ethylene glycol to glycolate fits

into the ideal network architecture that follows a branched pathway. Under oxygen limiting conditions,

the reaction that consumes glycolate, glycolate oxidase, can be limited, and the cell can accumulate

glycolate. These results show that ethylene glycol as a substrate is more orthogonal than traditional

substrates and hence suitable for validating as a concept of orthogonal pathways based design.

Succinate Ethanol Glycolate 2,3-Butanediol

Score Yield Score Yield Score Yield Score Yield

Formate 0.47 0.29 0.50 0.14 0.48 0.33 0.49 0.18

Ethylene glycol 0.54 0.95 0.61 0.62 0.67 1.22 0.66 0.66

Glucose 0.41 1.12 0.44 0.51 0.41 0.85 0.47 0.50

Xylose 0.36 1.11 0.36 0.65 0.34 0.65 0.40 0.64

Table 6-2 Yield and orthogonality metrics for chemical production from different substrates. The orthogonality

scores for various products are shown comparing two substrates that can be generated electrochemically against

conventionally used substrates by their natural pathways. Formate has orthogonality scores similar to many sugar

consuming pathways, indicating a relatively complex and inter-connectedness for its utilization. The highest scores are

those for ethylene glycol with yields as are better than sugars glucose and xylose. Yield is given as g of product per g of

substrate.


116 | P a g e

6.3.2 Ethylene Glycol Utilization by E. coli

There exist pathways in nature that allow microorganisms to consume ethylene glycol as a carbon

source15–18. While not commonly reported in metabolic engineering applications, these organisms use

one of two types of metabolic pathways. The first pathway utilizes a diol-dehydratase, resulting in the

dehydration of ethylene glycol to acetaldehyde. Acetaldehyde is then activated to acetyl-coa by an

acetaldehyde dehydrogenase enzyme which provides the cell with the key pre-cursor metabolite to

support growth via the TCA cycle and gluconeogenic pathways. The production of one mole of acetyl-

coa from one mole of ethylene glycol concomitantly produces one NADH. This pathway is most

commonly found in some Clostridium species and a few other anaerobic organisms owing to the oxygen

sensitivity of the diol-dehydratase15,17. The second mode of ethylene glycol degradation utilizes a

pathway wherein ethylene glycol is successively oxidized using nicotinamide cofactors and oxygen to

produce glyoxylate. Glyoyxlate, which is a gluconeogenic carbon substrate, can then be used as the

growth metabolite as it enters lower glycolysis at the 2-phosphoglycerate node as well as the TCA cycle

via the glyoxylate shunt.

Wildtype E. coli MG1655 cannot naturally grow on or degrade ethylene glycol. However, it is

possible to select for this strain, and to our knowledge, only one study has ever reported ethylene glycol

utilization by E. coli.19 That strain was selected from derivatives of propylene glycol utilizing mutants.

Researchers identified increased activities of glycolate oxidase, glycolaldehyde dehydrogenase and

propanediol oxidoreductase as the necessary components required for its assimilation. More generally,

a survey of the literature shows that enzyme promiscuity is an essential element of the utilization of

alcohols22,23. In this specific case, enzymes regarded as being essential for propanediol or even glycerol

utilization across many organisms have shown activity on ethylene glycol and are regarded as the key

methods for degradation, irrespective of the dehydratase route or the oxidative route via glyoxylate16–

18. Hence, in this study, to engineer E. coli we overexpressed the native gene fucO and aldA that have

been established as key enzymes supporting propanediol utilization in E. coli. Since FucO has

previously been shown to be sensitive to oxygen via metal catalyzed oxidation that results in the

inactivation of Fe2+ dependent propanediol oxidoreductases, we designed two variants of the pathway

to consume ethylene glycol. In variant 1 (strain LMSE11), the mutated version of fucO was used

wherein I7L and L8V based on earlier mutagenesis studies20. In the second variant (strain LMSE12),

L8M was used because it was also suggested to play a role in alleviating metal catalyzed oxidation

(MCO) toxicity in propanediol assimilation by E. coli. Both variants had the same ribosome binding


117 | P a g e

site and trc promoter upstream of the start codon. Cells were grown aerobically in M9 minimal media

with ~10 g/L ethylene glycol, supplemented with 0.2% yeast extract in 250 mL shakeflasks.

Fermentation profiles between the two strains constructed were markedly different. LMSE11

completely consumed ethylene glycol in 47 hours while LMSE12 had consumed only ~10% of the

initial substrate in same time period with 10 g/L as residual MEG. These results are shown in Figure

4-2.


118 | P a g e

Figure 6-1 Glycolate can be produced by a variety of different substrate. These pathways are shown in the panels. The chemical structures for the metabolites in ethylene glycol and xylose utilization pathways are also shown. The two most commonly studied substrates for production are xylose (B) and glucose (C). To efficiently produce glycolate from glucose or xylose, genetic interventions are required to the central metabolism to couple growth and glycolate synthesis. The focus of this study examines ethylene glycol consumption. Limiting oxygen provides a mechanism to permit glycolate accumulation. Under fully aerobic conditions, glycolate is converted to glyoxylate and channeled to the central metabolism for growth via the glycerate metabolism. Under oxygen limiting conditions, glycolate accumulates.


119 | P a g e

Growth yield for LMSE11 was calculated to be 0.28 ± 0.05 gDW/g MEG (± indicates

standard error). Flux balance analysis via in silico simulations of the core model of E. coli revealed the

theoretical yield to be 0.35 gDW/g MEG. These results seemed to be in reasonable agreement with

theoretical yields for biomass synthesis, suggesting that two genes are sufficient to efficiently convert

ethylene glycol to biomass using E. coli’s natural biosynthetic pathways. The substrate uptake rate in

shake-flasks was determined to be 5.2 ± 1.1 mmol/gDW-h (± indicates standard error). The

experimental growth rate was calculated to be 0.18 h-1 corresponding to a 3.85 hour doubling time.

Figure 4-2 shows the growth curve and substrate utilization of for both variants. LMSE12 consumed

substantially less ethylene glycol and had residual ethylene glycol concentrations just under 10 g/L in

the same time period.

Figure 6-2 Cell growth curves and their substrate consumption profiles for the strains constructed in this study. The oxygen variants of fucO showed a marked difference in growth rate and substrate utilization in shake-flask experiments. Ethylene glycol consumption is shown by the dashed lines and OD600 is depicted by the solid lines. Yellow (light) shows strain LMSE11 while green shows LMSE12. Error bars indicate standard deviation of triplicate experiments.


120 | P a g e

Analysis of the fermentation media by HPLC showed the absence of fermentation products

like acetate or lactate, and the intermediate metabolites glycolaldehyde and glycolate. However, since

LMSE11 showed higher utilization rates, we decided to pursue that variant further.

6.3.3 Orthogonal Production of Glycolate by E. coli

Having established ethylene glycol consumption by an engineered strain of E. coli, we explored the use

of ethylene glycol as an orthogonal substrate for the production of glycolic acid. E. coli strain LMSE11

was grown in bioreactors with minimal media, supplemented with yeast extract at 2 g/L and sparged

with air to maintain oxygen at 1 v/vm (300 mL/min). This ensured that oxygen saturation above 50%.

Cells were initially grown overnight for 18 hours for growth in LB rich media supplemented with

ethylene glycol and induced with IPTG. After overnight growth, they were centrifuged, washed and

suspended in minimal media and inoculated to bioreactors at an OD ~ 0.4 (approx. 0.23 gDW/L).

The bioreactors contained 1 mM IPTG to maintain induced expression of MEG utilization genes to

support biomass.

At 20 hours, the aeration was reduced to 150 mL/min (0.5 v/vm) and 50 mL/min (0.16 v/vm)

to simulate high and low aeration rates, and the impeller agitation was dropped to 500 rpm. We

observed that cell growth continued until approximately 40 hours reaching approximately 5 gDW/L

at which point cells in both reactors appeared to reach a stationary phase. Production of glycolate,

however, was continued for 30 hours more after the beginning of stationary phase at which point the

fermentation was stopped. Cells grown at a higher rate of aeration accumulated more glycolate by the

end of the batch. Counter-intuitively, the lower aeration led to lower glycolate titers. We believed this

to be related to the ability of the cell to regenerate intracellular NAD+.

The final glycolate titres for the two treatments were 2.5 ± 0.19 g/L and 4.1 ± 0.39 g/L (±

indicates mean absolute deviation). Using flux balance analysis to approximate carbon loss from

respiration and accounting for cell growth and other products, we were able to close the carbon balance

at 88% and 91%, respectively. Average mass yield for glycolate on MEG measured during the

production phase was 0.18 ± 0.01 g/g and 0.32 ± 0.005 g/g (± indicates mean absolute deviation).


121 | P a g e

6.3.4 Dissolved Oxygen and Control Over Metabolism

During the glycolate production stage, we detected acetate as well as trace amounts of other

fermentation products such as formate (less than 0.1 g/L) that are more typically found during

anaerobic growth conditions. These results were unexpected given the limited flux we anticipated in

conversion of glycolate to 2PG. However, the results suggest that with further metabolic engineering

and an efficient control system for dissolved oxygen tension in the fermenter, the cell can produce

anaerobic products while allowing the oxygen dependent enzymes to be active. This highlights a trade-

Figure 6-3 Influence of aeration on glycolate production. To assess the impact of oxygen transfer in bioreactors, cells were grown under two aeration rates during the micro-aerobic phase of the fermentation. (Top) High aeration had a flow rate of 150 mL/min. (Bottom) Low aeration was characterized by flow at 50 mL/min. Experiments were conducted in duplicate. Error bars indicate range of the measured values.


122 | P a g e

off around oxygen concentration. At higher oxygen concentrations, aerobic products from ethylene

glycol are permissible while at lower oxygen concentration, it could be possible to produce anaerobic

products. Secondly, observation of these anaerobic products reveals that intracellular dissolved

concentrations should be less than the bulk reactor concentrations to permit the function of oxygen

dependent metabolic pathways. This has important implications for using FucO as the protein to

oxidize ethylene glycol since it is known to be oxygen labile. It is known that increased tolerance of

FucO towards oxygen decreases its kinetic activity20. These results suggest that optimization of the

bioreactor conditions might allow for the use of the more active and more labile protein. We saw in

later experiments that cell growth is inhibited during high oxygen flowrates (Supplementary).

To gain further insight into control of the cell’s metabolism using oxygen and refine our

approach to glycolate production, we used flux balance analysis (FBA) to simulate the intracellular flux

through the central metabolism at 5 mmol/gDW-h which was determined with the shake-flask

experiments. The simulations were constrained using the substrate uptake rate to approximate E. coli

growth during the early exponential growth phase measured in shake flasks. The ATP maintenance

flux was approximated at 8.9 mmol/gDW-h, a value experimentally used for glucose metabolism. The

simulated flux distributions revealed a highly re-organized central metabolism of E. coli using

gluconeognic pathways.

Under oxygen limiting conditions, FBA predicts the observed fermentative cell behavior and

glycolate accumulation. We then explored this observation further by modelling the production of the

glycolate (as yield) by the cell and its respiratory quotient as a function of the oxygen uptake rate. This

allowed us to implicitly correlate the flowrate of air into the reactor to the metabolite production yields

since the specific oxygen uptake is a function of air intake. Figure 4-4 shows that the increase in

glycolate yield and the onset of fermentation as oxygen uptake rate is reduced. These yields correlate

with the respiratory quotient that also decreases at a lower oxygen flux and increases with increasing


123 | P a g e

oxygen flux before it levels off at saturating conditions. The results suggest that RQ is an important

variable that can be monitored and controlled to optimize for glycolate production in real-time. Hence,

we used this approach to control glycolate production in subsequent runs.

6.3.5 Glycolate Production and Fed Batch Strategy

Finally, given that were we able to produce glycolate, we performed further experiments to attempt to

improve glycolate production yield and increase titres. Based on what we learned from the initial

fermentations, we sought to increase the glycolate production phase and reduce the biomass

production phase. This was achieved by increasing the aeration rate to 2 v/vm (600 mL/min) during

the growth phase of the batch to prevent glycolate accumulation and divert as much flux towards

biomass. In the second phase, the aeration rate was dropped to 100 mL/min. Results of this strategy

are shown in Figure 4-5a. Final glycolate titres reached 6.8 g/L after approximately 70 hours of

production time with an initial production phase biomass concentration of approximately 4 gDW/L,

Figure 6-4 Metabolic modelling glycolate production. Glycolate yield (glycolate, blue), the respiratory quotient (RQ, green) and the substrate specific productivity (SSP, red) are modelled using FBA. Glycolate production begins at the onset of oxygen limitation which occurs at approximately 8 mmol/gDW-h of oxygen. At greater values, the RQ plateaus as sufficient oxygen as available for complete respiration and FBA predicts no glycolate accumulation. The grey bar indicates the values at which RQ was controlled experimentally during the production phase in later batches.


124 | P a g e

corresponding to an average productivity of 0.1 g/L-h or approximately 0.32 mmol/gDW-h. The

initial yield of glycolate was 0.92 g/g after the first sample was taken; however, the cumulative yield

decreased during the production course of the batch with the final overall production yield of 0.75 g/g

or 61% of theoretical.

We observed from these conditions that while we produced significantly more product at a

higher yield, the cells took much longer to reach a concentration appropriate for a production phase.

When the aeration rate was 1 v/vm in earlier batch, the cells reached a concentration of 4 gDW/L

within approximately 30 hours. However, at 2 v/vm, it took almost 70 hours to reach the same

concentration. We hypothesized the longer time to reach a higher OD was likely due to increased

dissolved oxygen levels and faster oxygen mass transfer rates to the cells during early exponential

phase. Given the sensitivity of FucO to oxygen, in even the mutant variant, these two factors likely

created an oxygen toxicity on the cells resulting from the inactivation of these proteins by metal-

catalyzed oxidation and placing a high metabolic burden on the cell in regards to high protein demand

without a sufficient means to utilize ethylene glycol as a carbon source.

Oxygen requirements are also one of the factors that affects the industrial production of

biochemicals since it is a key component of operating costs which are determined by the energy inputs.

One of the significant energy inputs for a process is the energy needed to aerate a bioreactor. In an

earlier experiment, we found that counter-intuitively, a higher aeration resulted in higher glycolate titres

at a higher yield but that high aeration also retards cell growth. From a process perspective, it is

desirable to operate a reactor at a lower oxygen flow rate. Building on all of these earlier studies and

the various competing objectives, we attempted to produce glycolate at a high titre but at a lower

aeration rate. Hence, cells were grown under a constant aeration 0.16 v/vm (50 mL/min), but during

the production phase, the impeller speed in the reactor was dropped until the RQ, as measured by the

online mass-spec, read ~0.4. The working hypothesis based on FBA simulations was that this would

achieve a yield greater than 0.4 mol/mol and place the production phase near its maximum substrate

specific productivity. The shaded region in Figure 4-4 shows the range of the RQ measured during

the course of the production phase as determined by three standard deviations from the average value.

The average RQ was measured to be 0.37.

The results of this experiment are shown in Figure 4-5b. We were able to reduce the biomass

production phase to 26 hours, and produce 10.4 g/L of glycolate over a 112 hour production phase.


125 | P a g e

The overall yield was determined to be 0.8 g/g from ethylene glycol corresponding to a molar yield of

0.66 mol/mol. The productivity was comparable to the earlier experiment at 0.1 g/L-h. These

experimental results were in line with and correlated well with FBA predictions for using RQ as a

control variable. As the batch entered the glycolate production phase, we observed a drop in the RQ.

However, the measured RQ value of 0.37 corresponded to a production yield of 0.66 mol/mol –

higher than the expected yield of 0.40 mol/mol. The results imply that while the general agreement

Figure 6-5 Fermentation profiles for fed batch strategies. Fed batch studies were conducted to assess the long term stability of the production phase. The production phase is separated from the growth phase by grey shading. (A) Shows bioreactor conditions at 2 v/vm during the growth phase and 0.33 v/vm during the production phase at a cell density corresponding to 4 gDW/L. (B) Cells were grown at 0.167 v/vm air flow rate into the bioreactor with an average stationary phase cell density at 2.5 gDW/L. Cells were capable of robust glycolate production for well over 100 hours in the production phase.


126 | P a g e

between experimental data and FBA simulations is useful in establishing a control mechanism for

fermentation on ethylene glycol, further optimization of model parameters is required to accurately

predict the physiological response to the environmental conditions. In particular, we found that

substrate uptake rate was reduced substantially in vivo however (approximately 0.7 mmol/gDW-h),

which was not accurately captured by the FBA models (at 3.5 mmol/gDW-h).

6.3.6 Metabolic Flux Analysis Using E. coli Model

To gain insight into the intracellular fluxes of the cell, we used mass spec and HPLC data to constrain

a genome scale model of E. coli and perform flux balance analysis. The model was then used to estimate

the intracellular fluxes under ethylene glycol growth conditions to gain insight into the cellular

metabolism. We determined that ethylene glycol enters the metabolism at the glyoxylate node (Figure

4-6a). 70% of the glyoxylate production flux is channeled towards 2-phosphoglycerate (2PG) under

aerobic conditions which enters lower glycolysis. The remaining glyoxylate is used to generate malate

via malate synthase. It appears from the simulations that the majority of the malate and 2PG generated

by these pathways ends up in the TCA cycle. 65% of the total carbon entering the cell as ethylene

glycol gets channeled into acetyl-coa. Conversely, about a fifth of the total carbon get channeled by

gluconeogenic pathways towards upper glycolysis and the pentose phosphate pathways.

During the growth phase, we also observed small amounts of glycolate. The accumulation of glycolate

suggested insufficient oxygen and thus, the possibility that anaerobic pathways in the cell may be

induced. Indeed, trace amounts of formate were detected as peaks in the HPLC chromatogram.

Given that the 2PG pathway that assimilates ethylene glycol results in carbon loss via the

tartronate semi-aldehyde carboligase step, we performed simulations to determine whether the

glyoxylate cycle was sufficient for supporting cell growth by removing the reaction glyck2 (glycerate

kinase) from the model. Removal of glyoxylate carboligase from the genome scale model showed a

50% decrease in the in silico growth rate. In contrast, experimental work on gene deletions in the same

pathway show that it abolishes growth on glycolate. To reconcile these differences, we analyzed the

genome scale model to determine the specific reactions that support cell growth. We found that

without glyoxylate carboligase, cell growth could theoretically be supported by the threonine pathway

where oxaloacetate is converted to serine, homoserine and threonine. Theronine aldolase is capable

of cleaving the amino acid to glycine for growth, and acetaldehyde for providing the acetyl-coa


127 | P a g e

necessary to replenish the acetyl-coa that is consumed by malate synthase. Hence, it is the threonine

metabolism generated from oxaloacetate that provides the route to support biomass in silico. The

pathway is a cycle and has as its output glycine and has acetyl-coa as its starting molecule. However,

it is unlikely that these enzymes are expressed in sufficient quantities to carry enough flux to support

growth. Hence, the primary role of the secondary malate synthase pathway and flux split in glyoxylate

metabolism between the glyck2 and mals (malate synthase) reactions seems to be to replenish the TCA

cycle intermediates as opposed to assimilating ethylene glycol.

We applied a similar methodology to determine the intracellular flux distribution under the

micro-aerobic conditions. During the glycolate production phase (Figure 4-5b), oxygen flowrate into

the bioreactors was limited to create a micro-aerobic environment. The resulting drop in oxygen

concentration affected the metabolic flux distribution. The most notable change was a reduction in

the substrate uptake rate of ethylene glycol to ~0.7 mmol/gDW-hr, a quarter of what was observed

during aerobic growth. Secondly, in silico simulations predicted reduced glyoxylate utilization through

malate synthase and instead majority of the flux was diverted towards the TCA cycle through 2PG.

Whereas the molar ratio of flux through lower glycolysis versus malate synthase was almost 1:1 under

aerobic conditions, it was estimated to be 30:1 under micro-aerobic conditions. The decrease in the

substrate uptake, we speculate, is likely caused by a lower oxidation rate of NADH by oxygen leading

to an accumulation of reduced NAD co-factors and leaving fewer oxidized molecules available for

ethylene glycol catabolism. The production of acetate in the metabolism is a characteristic of over-

flow metabolism associated with fermentative metabolism. Trace amounts of formate, produced by

pyruvate formate lyase which is transcriptionally controlled by oxygen, is consistent with other studies

showing activation of anaerobic pathways in the transition to a fermentative metabolism.


128 | P a g e

6.4 Discussion

Conventional approaches to the bio-based production of chemicals have relied on using glucose, and

more recently xylose as feedstocks. Yet microorganisms tend to be very diverse in their ability to

metabolize different carbon sources. In this work, we proposed and examined the use of ethylene

glycol as a substrate to replace glucose in bioprocesses for growth and chemical production. Counter

to other studies, many pertaining to the synthesis of ethylene glycol from glucose, our motivation for

studying ethylene glycol as a substrate stems from the fact that it can also be derived from CO26,21.

Hence, its consideration as a feedstock that can potentially sequester carbon and lower greenhouse gas

emissions is akin to studies examining syngas fermentation of formate utilization.

To assess ethylene glycol utilization in the context of biochemical production, we examined glycolic

acid production – an alphahydroxy acid used in cosmetics and polymer applications. The results from

(A)

(B)

Figure 6-6 Flux distribution of the metabolism and key enzymes in the pathway. (A) The estimated intracellular flux distribution under aerobic conditions. (B) Under oxygen limiting conditions, the metabolic model estimates ethylene glycol flux ethylene glycol is primarily converted to glycolate. Values in brackets represent upper and lower values obtained from flux variability analysis. The flux ranges provides an estimation in the error on the reaction fluxes based on the constraints imposed for the above simulation. In this case, the relatively narrow ranges on the estimations are useful to attribute a physiologically meaningful interpretation to the data.


129 | P a g e

our study allows us to conclude that ethylene glycol is a suitable platform for growth and highly

efficient for producing glycolic acid. More generally we find that with further metabolic engineering,

ethylene glycol could be used to produce alcohols and other organic acids that are typically produced

during fermentative metabolism. This capability, we believe, can have an impact in industrial

biotechnology. We elaborate on these findings by examining three specific areas.

Our consideration of ethylene glycol as a substrate was driven, primarily, by challenges related

to the utilization of non-native substrates in E. coli. These interactions, which we described earlier as

orthogonality, help to identify pathways with high and low degrees of interactions. Computationally,

we find that ethylene glycol exhibits a lower level of interactions than many natural and some synthetic

pathways which we believe make it a more robust substrate than substrates such as formate or

methanol. Hence, these interactions provided a rational basis for selecting and engineering a novel

substrate utilizing pathway into E. coli. This work demonstrates the first de novo design of an orthogonal

pathway for metabolic engineering based on an orthogonality metric.

Our results demonstrate the applicability of E. coli to use a new and novel substrate that has

never been considered as a potential feedstock. Initial characterization of the cell growth determined

that the substrate uptake rate was approximately 5 mmol/gDW-h. At typical cell densities for

industrial processes (10 – 100 g/L)24, this corresponds to net flux of 3-30 g/L-h, well above the

required 3-4 g/L-h productivity for growth independent production typically needed25. Furthermore,

we believe that with adaptive laboratory evolution of the MEG utilizing strains, we can likely see

further increases in substrate uptake rate as the cell adjusts its proteome to become more efficient in

metabolizing ethylene glycol. Further characterization of these strains led us to determine that there

was some oxygen sensitivity, especially during early exponential phase. We believe that these are likely

caused by metal catalyzed oxidation of FucO in the presence of excess aeration and could be addressed

by using O2-tolerant Zn2+-dependent variants.

An important observation made during the course of these experiments was a reduction in the

substrate uptake rate during oxygen limiting conditions. We believe that the oxidation of NADH

resulted in a shift in reduced NAD pools and a decrease in the rates of reaction catalyzed by fucO and

aldA. This had a net effect of lowering the flux of ethylene glycol into the cell. This finding necessitates

a further study of cellular physiology under ethylene glycol utilization so as to understand the trade-

off in yield and productivity as a function of the dissolved oxygen feeding in the bioreactor. For


130 | P a g e

example, whereas we found increases in overall glycolate titres at 150 mL/min relative to 50 mL/min

further, on-line monitoring in the fed-batch studies via maintaining a target respiratory quotient helped

to increase product yields and titres at 50 mL/min relative to the earlier experimental conditions at

150 mL/min. Hence, optimization of aeration in the bioreactor would substantially improve economic

performance, both in terms of product formation but also in terms of the absolute cost of aeration.

For example, the operating conditions of the experiment in this bioreactor, correspond to a kLa of 120

h-1. Typical jet loop bioreactors26 are capable of delivering this design constraint at a mass transfer

power of 3 kW/m3. Therefore a typical reactor that is 350 m3 would consume 1000 kW of power or

160,000 kWh over the course of a typical fermentation. This corresponds to an energy cost (at

$0.10/kWh) of over $15,000 which represents 20%, a substantial fraction, of the final cost of the

product at 100 g/L at $2/kg in a typical 350,000L fermenter. Hence, the importance of optimizing

process conditions through genetic engineering is important to its financial viability. Further work

entailing a more detailed study of the oxygen transfer and glycolate titres is expected to more accurately

determine the optimum conditions.

Further computational modelling allowed us to infer ratios of key branch points within the

metabolism and identified glyoxylate carboligase as the central pathway for assimilating ethylene glycol,

with malate synthase playing a relatively small role in its assimilation. Results of this also showed that

much of the NADPH redox requirements for cell growth were obtained surprisingly obtained through

the pentose phosphate pathway and relatively little from the anaplerotic NADP dependent malic

enzyme, as might be initially expected. We also observed small amounts of acetate and trace amounts

of ethanol in the fermentation media during microaerobic glycolate production phase. FBA modelling

results predicted ethanol production during microaerobic conditions, but failed to predict acetate

production, without the adequate constraints. The observation of acetate and ethanol in the

fermentation medium, typical products of anaerobic growth, suggest that microaerobic conditions may

permit ethylene glycol as a suitable feedstock for the production of other anaerobic products despite

its requirement for oxygen. Finally, by extending the observations from flux balance analysis, we were

able to couple them to a process mass spec to measure in real-time the respiratory quotient and by use

of a simple model, show its applicability as a parameter to control glycolic acid production during the

course of the fermentation. This may open new opportunities for producing a variety of products

using ethylene glycol as a feedstock, provided the oxygen mass transfer rate can be efficiently

controlled.


131 | P a g e

One area that is currently unresolved is an understanding of how ethylene glycol gets taken up

by the cell. Some work has suggested that the uptake of ethylene glycol is a diffusion based processes19.

However, it would be worthwhile determining if the expression of transporters, such as those involved

in propanediol metabolism, could increase the ethylene glycol uptake rates.

6.5 Conclusions

The results described in this study establish a framework for future production of chemicals in E. coli

using ethylene glycol as a substrate. We describe, for the first time, the successful production of

glycolic acid from ethylene glycol as a feedstock for growth and for production. We believe this can

have important implications in the future for integrating biorefineries into industries where carbon

dioxide can be captured from point sources.


132 | P a g e

6.6 References

1. Erickson, B., Nelson & Winters, P. Perspective on opportunities in industrial biotechnology in renewable chemicals. Biotechnol. J. 7, 176–185 (2012).

2. Jiang, Z., Xiao, T., Kuznetsov, V. L. & Edwards, P. P. Turning carbon dioxide into fuel. Philos. Trans. A. Math. Phys. Eng. Sci. 368, 3343–3364 (2010).

3. Liao, J. C., Mi, L., Pontrelli, S. & Luo, S. Fuelling the future: microbial engineering for the production of sustainable biofuels. Nat. Rev. Microbiol. 14, 288–304 (2016).




7. Straub, M., Demler, M., Weuster-Botz, D. & Dürre, P. Selective enhancement of autotrophic acetate production with genetically modified Acetobacterium woodii. J. Biotechnol. 178, 67–72 (2014).

8. Shen, C. R. & Liao, J. C. Synergy as design principle for metabolic engineering of 1-propanol production in Escherichia coli. Metab. Eng. 17, 12–22 (2013).

9. Pirkov, I., Albers, E., Norbeck, J. & Larsson, C. Ethylene production by metabolic engineering of the yeast Saccharomyces cerevisiae. Metab. Eng. 10, 276–280 (2008).

10. Smanski, M. J. et al. Functional optimization of gene clusters by combinatorial design and assembly. Nat. Biotechnol. 32, 1241–1249 (2014).

11. Deng, Y., Mao, Y. & Zhang, X. Metabolic engineering of E. coli for efficient production of glycolic acid from glucose. Biochem. Eng. J. 103, 256–262 (2015).

12. Kataoka, M., Sasaki, M., Hidalgo, A. G. D. & Nakano, M. Glycolic Acid Production Using Ethylene Glycol- Oxidizing Microorganisms. 8451, 37–41 (2014).

13. Gao, X., Ma, Z., Yang, L. & Ma, J. Enhanced Bioconversion of Ethylene Glycol to Glycolic Acid by a Newly Isolated Burkholderia sp. EG13. Appl. Biochem. Biotechnol. 174, 1572–1580 (2014).

14. Wei, G. et al. High cell density fermentation of Gluconobacter oxydans DSM 2003 for glycolic acid production. J. Ind. Microbiol. Biotechnol. 36, 1029–1034 (2009).


133 | P a g e

15. Toraya, T., Honda, S. & Fukui, S. Fermentation of 1,2-propanediol with 1,2-ethanediol by some genera of Enterobacteriaceae, involving coenzyme B12-dependent diol dehydratase. J. Bacteriol. 139, 39–47 (1979).

16. Child, J. & Willetts, A. Microbial metabolism of aliphatic glycols bacterial metabolism of ethylene glycol. BBA - Gen. Subj. 538, 316–327 (1978).

17. Hartmanis, M. G. & Stadtman, T. C. Diol metabolism and diol dehydratase in Clostridium glycolicum. Arch. Biochem. Biophys. 245, 144–152 (1986).

18. Mückschel, B. et al. Ethylene glycol metabolism by Pseudomonas putida. Appl. Environ. Microbiol. 78, 8531–8539 (2012).

19. Boronat, A., Caballero, E. & Aguilar, J. Experimental evolution of a metabolic pathway for ethylene glycol utilization by Escherichia coli. J. Bacteriol. 153, 134–139 (1983).

20. Lu, Z. et al. Evolution of an Escherichia coli Protein with Increased Resistance to Oxidative Stress *. 273, 8308–8316 (1998).

21. Tamura, J. et al. Electrochemical reduction of CO2 to ethylene glycol on imidazolium ion-terminated self-assembly monolayer-modified Au electrodes in an aqueous solution. Phys. Chem. Chem. Phys. 17, 26072–26078 (2015).

22. Pandya, C., Farelli, J. D., Dunaway-Mariano, D. & Allen, K. N. Enzyme promiscuity: Engine of evolutionary innovation. Journal of Biological Chemistry 289, 30229–30236 (2014).

23. Khersonsky, O. & Tawfik, D. S. Enzyme promiscuity: a mechanistic and evolutionary perspective. Annu. Rev. Biochem. 79, 471–505 (2010).

24. Soini, J., Ukkonen, K. & Neubauer, P. High cell density media for Escherichia coli are generally designed for aerobic cultivations - consequences for large-scale bioprocesses and shake flask cultures. Microb. Cell Fact. 7, 26 (2008).

25. Van Dien, S. From the first drop to the first truckload: commercialization of microbial processes for renewable chemicals. Curr. Opin. Biotechnol. 24, 1061–1068 (2013).


27. Pereira, B. et al. Efficient utilization of pentoses for bioproduction of the renewable two-carbon compounds ethylene glycol and glycolate. Metab. Eng. 34, 80–87 (2016).

28. Chen, Z., Huang, J., Wu, Y. & Liu, D. Metabolic engineering of Corynebacterium glutamicum for the de novo production of ethylene glycol from glucose. Metab. Eng. 33, 12–18 (2016).

29. Liu, H. et al. Biosynthesis of ethylene glycol in Escherichia coli. Appl. Microbiol. Biotechnol. 97, 3409–3417 (2013).

30. Boronat, A., Caballero, E. & Aguilar, J. Experimental evolution of a metabolic pathway for ethylene glycol utilization by Escherichia coli. J. Bacteriol. 153, 134–139 (1983).


134 | P a g e

6.7 Supplementary Data to Chapter 4

Characterization of the effect of oxygen extremes on cell growth. (Top) Panel shows cell growth measured as a function of CO2 concentration (%) at the outlet of the bioreactor from respiration. Cell growth is inhibited by high flowrates of oxygen into the bioreactor and growth does not commence until oxygen flowrate is lowered. (Bottom) Panel shows the control experiment at constant aeration. Cell growth is inhibited for the duration of the batch as measured by the bioreactor outlet CO2 concentration (%).

Summary of the Fermentation Experiments Parameters and Results

Growth Production Growth Phase Impeller Speed

Production Phase RPM

[Cells] [Glycolate] Production

Time

1 1 vvm 300 mL/min

0.5 vvm 150 mL/min

1000 rpm 500 rpm 5 g/L 4.1 g/L 30 hours

2 1 vvm 300 mL/min

0.17 vvm 50 mL/min


3 2 vvm 600 mL/min

0.33 vvm 100 mL/min


4

0.17 vvm 50 mL/min

0.17 vvm 50 mL/min

Controlled using RPM of impeller near 20% up to max 1000 rpm

Impeller speed reduced until RQ reached ≈0.4

2.5 g/L 10.4 g/L 112 hours

5 O2 Tested at various flow rates

1500 rpm

Conclusions and Recommendations


136 | P a g e

7.1 General Discussion

The central work of this thesis was concerned with elucidating both the challenges and the

effectiveness of different approaches for microbial electrosynthesis in industrial applications. By using

a single organism as a model as well as a computational framework, the experiments helped us to

identify the quickest routes to practically realize electrosynthesis based bioprocesses at the commercial

scale. This work is important because it represents the first real comprehensive study of

electrosynthesis comparing these various approaches.

In Chapter 3, the first type of electrosynthesis approach was laid out using neutral red as a mediator

to conduct charge to E. coli. We found that while neutral red at concentration of 10 µM appeared to

mediate charge to the cell resulting in a shift in succinate production from 0.09 to 0.17 (molar yield) in

wild-type, that this similar affect could not be generalized to other strains in the study. The notable

observation was found for the ldhA mutant strain showing that even though a positive charge of

0.018C had been delivered to the bioreactor, the degree of reduction for the fermentation products

was found to be higher for the for the electrically enhanced fermentation compared to the standard

fermentation. The work, which was first presented in 2013 in Portland at the Sustainable Biofuels and

Chemicals Conference, was the first to demonstrate that the increase in reduced products by E. coli

undergoing electrically enhanced fermentation was not due to direct electron transfer, but rather had

to be accounted for by other means. This work hypothesized that cell regulation was the cause of this

change, and this was validated by Harrington et al in 2015. Hence, the clear challenge of using mediator

based electrosynthesis techniques highlighted by this work is the efficient delivery of electron to the

cytoplasm to reduce intracellular NAD+.

The work on direct electron transfer spurred the exploration of identifying more promising routes for

microbial electrosynthesis. In Chapter 4, we validated the approach for using a formate as a mediator.

In the presence of 20 mM formate and 2 mM glycine, we were able to engineer E. coli to utilize formate

as a C1 donor with an average growth rate of 0.33 h-1. However, the utilization of non-natural

feedstocks such as formate were found to be problematic in that it required substantial rewiring of the

glycine and serine biosynthesis pathways. We hypothesized this challenge was due to the strong

regulation of the C1 metabolism in E. coli and the thermodynamics of the reductive glycine pathways.

Under the most optimum conditions, thermodynamic modelling of the pathway showed that the


137 | P a g e

maximum thermodynamic driving force of the least favourable reaction was less than 0.5 kJ/mol. This

presents a substantial thermodynamic bottleneck requires high protein expression for the pathway to

have a high flux. Overall, the computational analysis supported by experimental work suggested that

the inherent challenges associated with the utilization of formate by the reductive glycine pathway as

an approach for electro-biosynthesis to be thermodynamic arise from low co-factor concentrations

that are required to drive the synthesis of carbon-carbon bonds.

The challenges in engineering microbial electrosynthesis of using direct electron transfer and formate

necessitated a different approach. Hence, we reasoned that the two key elements of microbial

electrosynthesis – an efficient transfer of electrons to NAD and the synthesis of the first carbon-

carbon bond were necessary. Hence, we hypothesized this required using carbon feedstocks that could

be generated through renewable technologies that were at least two carbon units in lengths and those

that had a minimal set of regulatory and metabolic interactions with the cell’s natural metabolism.

Hence, we developed a computational framework for analyzing and prioritizing metabolic pathways

that assimilate substrates. The framework, while having applications for general metabolic engineering,

was useful for showing that unconventional feedstocks can reduce the interactions of metabolic

pathways with their natural metabolism. Using this framework, we identified ethylene glycol as a

promising feedstock for chemical production based on its orthogonality score.

Finally, we applied the principles of orthogonality to engineering E. coli to utilize ethylene glycol as a

carbon source. While the growth rate was determined to be 0.18 h-1, relatively slow compared to cell

growth on glucose, the substrate uptake rate was 5 mmol/gDW-h. This uptake rate was comparable

to glucose uptake for aerobic conditions and hence was determined to be reasonable for the production

of chemicals from a bio-based process. Production of glycolate from ethylene glycol was then analyzed

and several oxygen flow rates and reactor conditions optimized. We found that E. coli was able to

accumulate 10.6 g/L of glycolate in the fermentation media by controlling the respiratory quotient

below 0.4. The importance of this work shows that ethylene glycol, which can be renewably produced

can be used as an efficiently as well by a cell factory.

In summary, the work presented in this thesis provides a basis for understanding the approaches for

microbial electrosynthesis in a single, model organism. This study is the first comprehensive work that

examines more than one approach. On balance, while we find that every type of modality of


138 | P a g e

electrosynthesis shows some degree of viability, based on the data we present, the use of ethylene

glycol shows the most promising route for developing metabolic engineering strategies around.

7.2 Conclusions

We approached the challenge of microbial electrosynthesis by studying the various strategies. The

following are the four studies reported in this thesis:

1. Characterization of E. coli mutant strains for the purpose of mediator driven microbial

electrosynthesis for understanding succinate production

2. Development of a strain of E. coli for utilizing ethylene glycol and the identification of the key

process variables affecting production of glycolate

3. Studying the role of substrate utilization pathways, and the impact that they have on chemical

production by microbial cell factories.

4. Engineering E. coli to utilize formate, and identification of the key parameters affecting the

efficient use of formate as a carbon source

This studies were performed to assess the four main hypotheses described in Chapter 1.

1. In Hypothesis 1, it was thought that E. coli growing in the cathode compartment of an

electrochemical cell, in the presence of a reduction potential and a mediator would produce

greater quantities of succinic acid. Work on neutral red as an electron source highlighted the

incredible challenge posed in engineering cell systems to utilize mediators for electron transfer.

While results were initially promising, we did uncover through electron accounting that

previous models to electron transfer were not capable of accurately describing the physiological

phenomena observed. During the course of this work, other groups corroborated our initial

work on mutant strains of E. coli and discovered the underlying mechanisms of electron

transfer to the cell. This work disproved the working hypothesis that electron mediators are

an efficient method to deliver current to the cell. Indeed it appears that substantial metabolic

engineering is required to express a functional electron conduit from the menaquinone pool

to the NADH cofactors that would ultimately drive physiological processes.


139 | P a g e

2. In Hypothesis 2 it was believed that heterologous pathways that assimilate of formate can be

used to improve fermentation processes used for producing valuable chemicals. During the

course of these experiments we attempted to engineer formate utilization (Chapter 4) While

formate assimilation was eventually shown using auxotrophic strategies, it was not possible to

engineer E. coli to solely rely on formate as carbon source. However, the experiments and

computational analysis of that study identified the significant physiological challenges of the

reductive glycine pathway. In conclusion, contrary to the hypothesis, it is now believed that

formate is not an efficient substrate for producing chemicals.

3. In Hypothesis 3 it was thought that the specific utilization pathways of natural and non-

natural substrates plays an important role in the metabolic engineering of cell factories for

producing fuels and chemicals and they can be quantified using metabolic modelling

techniques. The use metabolic modelling techniques to understand the role that substrate

utilization pathways (Chapter 5) have on the ability of a cell to produce chemicals was useful

in identifying alternative strategies to engineer microbial electrosynthesis platforms (Chapter

6). The work helped to guide later experiments and was experimentally validated in Chapter

6.

4. Finally in Hypothesis 4 it was thought at if E. coli can be engineered to consume ethylene

glycol, then because of its high degree of reduction it can efficiently produce fermentation

products. This hypothesis of successfully validated. Studies relating to ethylene glycol

established various factors include the role of oxygen because of its effect on FucO as well as

because of its impact on the trade-off between glycolate productivity and yield.

Based upon the results from research objectives of this thesis, the following conclusions can be made:

1. Strains of E. coli with elevated NADH/NAD+ ratios show marginal increases in

succinate yield. Succinate yield for the ldhA mutant strain increased from 0.08 mol/mol to

0.11 mol/mol glucose. Succinate yield for the wild-type showed much larger increases in yield

from 0.09 to 0.17 mol/mol.

2. Total charge delivered to the cell measured at the cathode does not correlate with

change in the degree of reduction of the fermentation products.


140 | P a g e

3. Ethylene glycol is an efficient platform for chemical production. While glycolate titres

were not optimized for commercial scale, titres reached over 10 g/L and achieved 66% of the

theoretical limit.

4. Despite using an oxygen tolerant variant of FucO, oxygen sensitivity was observed

when E. coli was grown at high aeration rates. Metabolic modelling established a trade-

off between the oxygen uptake rate and the glycolate productivity. Experimental data showed

an RQ of 0.37 corresponded to a glycolate yield of 0.66.

5. An orthogonality framework was developed to understand the role of substrate

utilization pathways chemical production. We showed that this orthogonal pathway

design approach has significant advantages over contemporary growth-coupled approaches

using a case study of succinate production. We found that natural pathways, fundamentally

linked to biomass synthesis, are less orthogonal in comparison to synthetic pathways. We

suggest that the use of such orthogonal pathways can be highly amenable for dynamic control

of metabolism and have other implications for metabolic engineering.

6. Formate was demonstrated to rescue a glycine and serine auxotrophic strain of E. coli.

While the recovered growth rate was reduced from 0.46 h-1 to 0.33 h-1, growth in formate and

glycine validated intracellular activity of folate tetrahydrofolate ligase to partially support the

flux through the reductive glycine pathway.

Hence, the overarching contribution of this work is a set of guiding principles for engineering novel

methods for substrate utilization that are primarily driven by a motivation to convert renewable

electrical energy and carbon dioxide to value added compounds using biological platforms. In short,

cell systems are incredibly challenging due to their highly complex nature and correlated pathways for

both carbon and electron utilization. While these interactions can be determined mathematically as

we showed, experimental evidence suggests that efficient delivery of electrons and carbon ought to be

done via a single, reduced substrate as we demonstrated for ethylene glycol utilization. Hence, given

that there is sufficient technology to convert CO2 to various reduced species, this work supports the

idea that ethylene glycol is a useful and beneficial substrate to circumvent various other challenges

posed by microbial electrosynthesis platforms.


141 | P a g e

7.3 Future Work

The natural next steps of this work would follow the further development of a biocatalyst capable of

efficiently converting ethylene glycol. To that end, the following specific recommendations are made:

1. Integration of the oxygen tolerant genes for ethylene glycol utilization into the cell genome

followed by adaptive laboratory evolution (ALE) in minimal media. This would allow the

researcher to build on the existing work for the purposes of metabolic engineering. Increased

growth rates and reduction of the cells, proteome to a core set would be the likely outcome of

this work. This would have beneficial impacts as it would allow the cell to exhibit a growth

phenotype more optimized for commercial studies. Metabolomic and transcriptomic studies

in this evolved catalyst would be beneficial for understanding the genetic interventions required

to divert flux for chemical production.

2. A multi-variate study using miniaturized bioreactors to examine the effects of oxygen partial

pressure, impeller speed and temperature on the ability of E. coli to use ethylene glycol as a

carbon source. Specifically as it relates to oxygen, mixtures of air and nitrogen should be made

while maintaining a constant inlet pressure to determine the impact of O2 partial pressure on

metabolism. This could ideally be performed on milliliter scale shaken plates. This work

would be extremely valuable to identify optimum growing conditions as well as maximum

conditions that enable high substrate uptake rates.

3. Engineering an alternative pathway for ethylene glycol utilization would be beneficial for

chemical production. It was suggested as part of the computational analysis of this work that

that ethylene glycol can be metabolized by the cell by various different pathways and that these

pathways have an inherently different ability to produce chemicals independent of cell growth.

At least two additional pathways can be tested without relative experimental complexity. These

pathways are consumption via an aldolase reaction that produces erythose-4-phosphate as the

growth metabolite, and by a dehydratase reaction that produces acety-CoA as the growth

metabolites. The dehydratase route is of particular significance because it can be operated

anaerobically, mitigating the need to aerate the bioreactor.


142 | P a g e

4. Orthogonality scores were developed as a methodology to determine whether cellular

networks are capable behaving in ways that maximize two different objectives. While the

specific purpose of the orthogonality score was to understand biomass and chemical

production objectives of networks, the methodology could be applied in theory to any type of

performance objective that a network could support. To that end, potential follow-up work

would be to understand natural objectives within cellular networks. For example, various

organisms are capable of biphasic shifts or employ complex regulatory networks to turn on or

off metabolic pathways. It would be an interesting question at ask whether the independence

of two specific tasks (ex. solventogenesis vs acetogenesis for Clostridium spp.) is controlled by

many or few regulator elements in the cell. Consider another example: it was recently reported

that when E. coli is grown on a nitrogen source other than ammonia, glucose provides the cell

the lowest growth rates from several different substrates. Researchers found cAMP levels, a

global metabolite regulator, to be the cause. It would be interesting study to understand

whether any structural properties, measured by the orthogonality score, can be predictive of

growth and the carbon/nitrogen pairing and the number of transcription factors controlling

that growth. This could hypothetically be extended to other growth environments.

Recommendations to other studies of this thesis are also made:

1. Investigate succinate production using neutral red in a strain of E. coli engineered for succinate

production. While it was found that the mutant strains showed some improvement in

succinate, it is likely that succinate production was limited by the expression of the

phosphoenolpyruvate carboxylase. Hence, it is possible that an engineered strain may be able

to take up more electrons.

2. As described in the literature review, Ajo-Franklin and co-workers recently developed an

engineered strain of E. coli capable of reducing nano-crystalline iron. Given that current

transfer across cytochromes has been shown to be bi-directional, it may be of interest to study

the effect that a reducing potential has on the growth characteristics of the E. coli strain

developed by the Ajo-Franklin lab.

3. Chapter 4 shows a protocol for measuring intracellular concentration of the NAD co-factor

pools and ATP was effective in determining differences between the formate utilizing strains.


143 | P a g e

It would be useful to further develop a protocol for measuring the folate co-factors accurately

in E. coli and validate the thermodynamic bottlenecks identified for the formate utilizing

pathway.

Expression of Outer Membrane Protein MtoA

144 | P a g e


145 | P a g e

Overview and Background

Part of the work performed during the course of this PhD was to create a synthetic electron conduit

from Sideroxydans lithotrophicus, an organism that is able to naturally use Fe(II) as an electron donor.

That work was stopped early on when other researchers were able to accomplish a similar task in E.

coli by using proteins from Shewanella as described in the Literature Review. This section describes the

extent of the work undertaken.

Sideroxydans lithotrophicus oxides Fe2+ from its environment by using cytochromes present on its outer

membrane and spannings its periplasmic space. The gene cluster required for synthesizing this

electron conduit is the Mto operon as well as the CymA gene. Hence, in this work we attempted to

express the Mto operon in E. coli.

Results

The Mto operon in Sideroxydans is made up of three genes that are responsible for the core function

of transferring a charge across the periplasm plus CymA. Hence to begin transferring a functioning

electron conduit to E. coli we began by expressing the individual protein in the cell to ensure they were

folding correctly and functionally. Plasmids were constructed for the individual genes MtoA, MtoB,

MtoC and CymA. MtoA was then expressed in BL21 strain of E. coli. Figure B-1 shows the cells

grown with pTrc99a-MtoA and the cytochrome maturation enzymes (ccm). Cells containing both

sets of genes turn red in colour indicating heme insertion into the overexpressed protein. The control

strains are unchanged. Strain with just the cytochrome maturation genes as pEC86 do not change

colour and without these genes the strain containing the MtoA gene cannot produce a functional

cytochrome.


146 | P a g e

Figure B-1. Red cell pellets showing heme incorporation in MtoA when both the cytochrome maturation genes in plasmid pEC86 and the mtoA gene in plasmid pVP101 are present. Pellets lacking either plasmid are not red.

To characterize expression and functionality, cell were induced with IPTG and grown overnight.

The cell were then harvested and the periplasmic faction was analyzed by LDS-PAGE for correct

localization and by UV spectroscopy for functionality. UV spectroscopy show the Soret bands of

the periplasmic fraction. Peaks at 408 nm are indicative of correct heme insertion in the

cytochromes. The reduced periplasmic space shows a shift in the band by about 10nm and alpha and

beta peaks at 526 and 550nm. These results are indicative of correct localization and functionally

active MtoA cytochromes. However attempts to further verify by LDS-PAGE were unsuccessful

because there were no bands observed.


147 | P a g e

Figure B-2. Soret Bands of the periplasmic space showing active MtoA.

Conclusions

The preliminary work done suggests that the Mto operon form Sideroxydans lithotrophicus should be

genetically portable in E. coli while remaining functional. Hence, these initial results that the genetic

protability of building electrical conduits in E. coli is not limited to Shewanella and that there exists the

prospect of expanding the diversity in the genetic backgrounds from which these extracellular charge

carrying proteins maybe be functionally expressed. Further research into this area may reveal some

systems to be inherently more expressible (ex. from codon bias) in E. coli than others.

Engineering Succinate Producing Triple Mutant

148 | P a g e


149 | P a g e

Overview and Background

Succinic acid is a valuable, industrial biochemical that is produced primarily by fermentation. It is a

four carbon di-acid that is used primarily to produce polyesters and is a fine chemical precursor for

compounds such as butanediol and maleic acid. The primary feedstocks for commercial processes

making succinic acid is glucose. The goal of our work in this section was to expand, E. coli ability to

make glycolate from ethylene glycol to succinate, as a proof-of-concept demonstration showing the

ability of ethylene glycol to produce a variety of different biochemical. To that extent, we

methodology was to introduce genetic disruptions in the cell to accumulate succinate and then

express the ethylene glycol catabolic pathways. We reasoned that an aerobic production strategy

would be required since ethylene glycol requires oxygen for consumption. Hence, three genetic

mutations were made in E. coli targeting the TCA cycle succinate dehydrogenase and both malic

enzymes. We reasoned that since succinate dehydrogenase has been demonstrated in a variety of

studies to be necessary for the accumulation of succinate, it would also be required for an ethylene

glycol consuming strain. Deletion of the malic enzymes it was hypothesized would reduce the drain

of malate, a TCA cycle metabolite to pyruvate since malic enzymes are upregulated on

gluconeogenic substrates. Figure C-1 shows a schematic of the disruptions to the metabolism.

Results

Cells were grown in 250mL shakeflasks containing M9 minimal media supplemented with 0.2%

yeast extract. Cells were grown under two substrate conditions: (1) containing 10 g/L of ethylene

glycol and (2) containing 10 g/L ethylene glycol and 10 g/L acetate. Results of the experiments are

summarized in Table C-1.

Cells grown on solely on ethylene glycol produced no detectable amount of succinate in the

fermentation media. Moreover, the results seems to indicate that no ethylene glycol was consumed.

It was hypothesized that this could be occurring for several reasons. The disruption of succinate

dehydrogenase (sdh) may be causing an accumulation in TCA cycle intermediates other than just

succinate. This increase in metabolite pools may be creating a kinetic bottleneck for ethylene glycol

catabolism. Previously, disruption of sdh has been accompanied by an overexpression of either

pyruvate or phosphoenolpyruvate carboxylase to drive carbon flux towards succinate and avoid an

accumulation of intermediate metabolites in the pathway.


150 | P a g e

In contrast, the experiments that had both acetate and ethylene glycol as substrates did exhibit small

amounts of succinate production at about 0.2 g/L. It is possible that partial succinate accumulation

are a result of an active malate synthase G supplied with a sufficient acetyl-coa pool derived directly

from acetate. Interestingly in both sets of experiments ethylene glycol appeared relatively

unchanged or actually increased.

Acetate + Ethylene Glycol Ethylene Glycol

Time

(h) MEG Succinate Glycolate Acetate MEG Succinate Glycolate Acetate

0 91.3±0.2 0.00±0 0.0±0 22.8±0.4 90.7±3.1 - 0.0±0 -

12 90.3±1.8 0.00±0.01 0.9±0.01 21.6±0.8 92.1±1.9 - 1.0±0.06 -

24 93.5±0.5 0.12±0 1.3±0.01 20.9±0.4 93.5±0.5 - 1.3±0.01 -

36 91.8±5.8 0.45±0.07 1.2±0.1 18.8±1.4 98.7±0.8 - 1.8±0.05 -

48 94.3±9.8 1.46±0.1 0.0±0 12.3±1.3 108.8±1.5 - 1.5±0.09 -

Table C-1 Shake-flask results of strain containing mutation in maeB, scfA, and sdhAB. When grown on acetate, cell produce small quantities of succinate. No consumption of ethylene glycol was observed in ethylene glycol strains. ± indicates standard deviation in triplicate experiments. Concentration measured in mM.

Figure C-1 PCR Confirmation of deletions.

Conclusions

Further genetic interventions are required for the production of succinate from ethylene glycol.

Using acetate as a substrate demonstrates that succinate can be accumulated and excreted by the cell

using a gluconeogenic substrate. However, the approach used in these experiments was not

maeB scfA sdhAB maeB scfA mdh


151 | P a g e

sufficient to produce succinate. Is suggested that a PEP carboxylase (feedback resistant) be

expressed to drive flux from PEP to oxaloacetate.

Bioreactor Designs

152 | P a g e

Bioreactor Designs

153 | P a g e

The reactor design went through several iterations before the robust set-up described in Chapter 3

was finalized. The images below show these design iterations. The top image shows a traditional

microbial fuel cell designed to be used with a Nafion proton exchange membrane. The bottom

image shows glass bioreactor modified with custom ports for gas sparging, pH measurements and

samples as well as base addition. It utilized dialysis tubing as the barrier between the cathode and

anode compartment. A stir bar was used for agitation. The final design used elements of this design

including the submersible dialysis tubing.

Bioreactor Designs

154 | P a g e

Copyright Acknowledgements

ELSEVIER LICENSE

TERMS AND CONDITIONS

Jul 20, 2017

This Agreement between Mr. Aditya Pandit ("You") and Elsevier ("Elsevier") consists of

your license details and the terms and conditions provided by Elsevier and Copyright

Clearance Center.

License Number 4153110532120

License date Jul 20, 2017

Licensed Content Publisher Elsevier

Licensed Content Publication Current Opinion in Chemical Biology

Licensed Content Title The formate bio-economy

Licensed Content Author Oren Yishai,Steffen N Lindner,Jorge Gonzalez de la Cruz,Hezi Tenenboim,Arren Bar-Even

Licensed Content Date Dec 1, 2016

Licensed Content Volume 35

Licensed Content Issue n/a

Licensed Content Pages 9

Start Page 1

End Page 9

Type of Use reuse in a thesis/dissertation

Portion figures/tables/illustrations

Number of figures/tables/illustrations

1

Format both print and electronic

Are you the author of this Elsevier article?

No

Will you be translating? No

Original figure numbers Figure 1

Title of your thesis/dissertation

On the effectiveness and challenges of electrosynthesis strategies in Escherichia coli

Expected completion date Sep 2017

Estimated size (number of pages)

161

Requestor Location Mr. Aditya Pandit 50 Viewmark Dr

Bioreactor Designs

155 | P a g e

Richmond Hill, ON L4S 1E6

Canada Attn: Mr. Aditya Pandit

Total 0.00 CAD

Terms and Conditions

NATURE PUBLISHING GROUP LICENSE


Jul 20, 2017

This Agreement between Mr. Aditya Pandit ("You") and Nature Publishing Group ("Nature

Publishing Group") consists of your license details and the terms and conditions provided by

Nature Publishing Group and Copyright Clearance Center.



Licensed Content Publisher Nature Publishing Group

Licensed Content Publication Nature Reviews Microbiology

Licensed Content Title Fuelling the future: microbial engineering for the production of sustainable biofuels

Licensed Content Author James C. Liao, Luo Mi, Sammy Pontrelli, Shanshan Luo

Licensed Content Date Mar 30, 2016


Licensed Content Issue 5

Type of Use reuse in a dissertation / thesis

Requestor type academic/educational

Format print and electronic


Number of

figures/tables/illustrations

1

High-res required no

Figures Figure 1c

Author of this NPG article no

Your reference number

Title of your thesis / dissertation



Bioreactor Designs

156 | P a g e

Estimated size (number of

pages)

161

Requestor Location Mr. Aditya Pandit 50 Viewmark Dr Richmond Hill, ON L4S 1E6 Canada

Attn: Mr. Aditya Pandit

Billing Type Invoice

Billing Address Mr. Aditya Pandit 50 Viewmark Dr

Richmond Hill, ON L4S 1E6 Canada Attn: Mr. Aditya Pandit

Total 0.00 CAD

ELSEVIER LICENSE


Jul 20, 2017

This Agreement between Mr. Aditya Pandit ("You") and Elsevier ("Elsevier") consists of

your license details and the terms and conditions provided by Elsevier and Copyright

Clearance Center.




Licensed Content Publication Biotechnology Advances

Licensed Content Title Bioconversion of natural gas to liquid fuel: Opportunities and

challenges

Licensed Content Author Qiang Fei,Michael T. Guarnieri,Ling Tao,Lieve M.L. Laurens,Nancy Dowe,Philip T. Pienkos

Licensed Content Date May–June 2014




Start Page 596

Bioreactor Designs

157 | P a g e

End Page 614


Intended publisher of new work

other


Number of


1



No



Title of your thesis/dissertation



Estimated size (number of

pages)

161

Requestor Location Mr. Aditya Pandit 50 Viewmark Dr Richmond Hill, ON L4S 1E6


Total 0.00 CAD

ROYAL SOCIETY OF CHEMISTRY LICENSE


Sep 27, 2017

This Agreement between Mr. Aditya Pandit ("You") and Royal Society of Chemistry ("Royal

Society of Chemistry") consists of your license details and the terms and conditions provided

by Royal Society of Chemistry and Copyright Clearance Center.


Bioreactor Designs

158 | P a g e

License date Sep 27, 2017

Licensed Content Publisher Royal Society of Chemistry

Licensed Content Publication Energy & Environmental Science

Licensed Content Title New insights into the electrochemical reduction of carbon dioxide on metallic copper surfaces

Licensed Content Author Kendra P. Kuhl,Etosha R. Cave,David N. Abram,Thomas F. Jaramillo

Licensed Content Date Feb 24, 2012



Type of Use Thesis/Dissertation

Requestor type academic/educational

Portion figures/tables/images

Number of figures/tables/images

1

Format print and electronic

Distribution quantity 1

Will you be translating? no

Order reference number

Title of the

thesis/dissertation

On the effectiveness and challenges of electrosynthesis strategies in

Escherichia coli


Estimated size 161

Requestor Location Mr. Aditya Pandit 50 Viewmark Dr

Bioreactor Designs

159 | P a g e



Billing Type Invoice

Billing Address Mr. Aditya Pandit 50 Viewmark Dr

Richmond Hill, ON L4S 1E6 Canada Attn: Mr. Aditya Pandit

Total 0.00 USD

Terms and Conditions

This License Agreement is between {Requestor Name} (“You”) and The Royal Society of Chemistry (“RSC”) provided by the Copyright Clearance Center (“CCC”). The license consists of your order details, the terms and conditions provided by the Royal Society of Chemistry, and the payment terms and conditions.

ELSEVIER LICENSE


Sep 27, 2017

This Agreement between Mr. Aditya Pandit ("You") and Elsevier ("Elsevier") consists of your

license details and the terms and conditions provided by Elsevier and Copyright Clearance

Center.


License date Sep 27, 2017


Licensed Content Publication Current Opinion in Chemical Biology

Licensed Content Title The formate bio-economy

Bioreactor Designs

160 | P a g e

Licensed Content Author Oren Yishai,Steffen N Lindner,Jorge Gonzalez de la Cruz,Hezi Tenenboim,Arren Bar-Even

Licensed Content Date Dec 1, 2016


Licensed Content Issue n/a


Start Page 1

End Page 9


Intended publisher of new work

other


Number of


1



No



Title of your

thesis/dissertation

On the effectiveness and challenges of electrosynthesis strategies in

Escherichia coli


Estimated size (number of pages)

161

Requestor Location Mr. Aditya Pandit

50 Viewmark Dr

Bioreactor Designs

161 | P a g e



Total 0.00 USD

on the effectiveness and challenges of … · aditya vikram pandit doctor of philosophy chemical...

Documents