engineering a responsive, heterologous mevalonate …...engineering a responsive, heterologous...

Engineering a responsive, heterologous mevalonate pathway in Escherichia coli using microarrays

By

Robert Harlan Dahl II

A dissertation submitted in partial satisfaction of the

requirements for the degree of

Doctor of Philosophy

In

Chemical Engineering

In the

Graduate Division

Of the

University of California, Berkeley

Committee in charge:

Professor Jay D. Keasling, Chair

Professor David Schaffer

Professor Adam Arkin

Fall 2011

1

Abstract

Engineering a responsive, heterologous mevalonate pathway in Escherichia coli using microarrays

By

Robert Harlan Dahl II

Doctor of Philosophy in Chemical Engineering

University of California, Berkeley

Professor Jay D. Keasling, Chair

Isoprenoids are a highly diverse class of natural compounds that have received wide spread commercial use as flavors, fragrances, and pharmaceuticals. However, their supply is often limited due to their low abundance in their natural sources and difficult routes of chemical synthesis. To alleviate this limitation, the Keasling Lab has engineered Escherichia coli to produce large quantities of the ultimate precursors, the isoprenyl diphosphates, by introducing the mevalonate pathway from Saccharomyces cerevisiae. Coupled with engineered terpene synthases and tailoring enzymes from their native sources, we now have the ability to make these molecules from simple sugars in a controlled environment.

The engineered heterologous pathway however bypassed the native regulation of E. coli’s isoprenoid pathway and the unregulated expression of the pathway resulted in growth inhibition due to the accumulation of the isoprenyl diphosphates. Microarray analysis of the inhibited cells identified the PhoPQ regulon as a concerted response to the toxicity. The PhoPQ regulon is E. coli’s regulatory two component system to respond to magnesium concentrations in the environment. Further analysis of the growth conditions revealed that the toxicity occurred when the magnesium and glucose concentrations in the media were low, validating the microarray results and leading to the hypothesis that the accumulation of the negatively charged diphosphate molecules disrupted the divalent cation’s normal role in the cell. Enzyme assays with luciferase, which uses ATP and magnesium as cofactors, confirmed that farnesyl diphosphate was inhibitory at an enzyme level.

Fluorescent sensors were designed using promoters of these genes upregulated and downregulated in response to farnesyl diphosphate accumulation. The sensors were tested for their dose dependent response to farnesyl diphoshpate accumulation and these promoters were then used to regulate the heterologous pathway.

2

In the absence of other methods to regulate the pathway in response to the intermediates’ accumulation, we have used standard inducible promoters that give a static output in response to an exogenous inducer and have no connection to the flux through the pathway. However, at industrial scale production the use of these exogenous inducers is not feasible. Using the responsive promoters return dynamic regulation to the pathway and obviates the need for an exogenous inducer. When the responsive promoters were cloned 5’ of the isoprenyl diphsophates’ downstream enzyme and the pathway was still regulated by the inducible promoters, production was less than the inducible promoter regulating the downstream enzyme. However, when the entire pathway was regulated by the responsive promoters, production doubled relative to the inducible pathway. In addition, the responsive pathway reduced acetate accumulation, allowed faster glucose consumption, and increased cell growth.

i

Contents List of Figures .................................................................................................................................. iv

List of Tables .................................................................................................................................. vii

Acknowledgments......................................................................................................................... viii

Chapter 1: Introduction, Motivation, Outline ................................................................................ 1

Motivation and Outline ............................................................................................................................. 1

Background ............................................................................................................................................... 4

Isoprenoids in E. coli ............................................................................................................................. 4

Microarrays and their uses in metabolic engineering .......................................................................... 5

Microarrays in the Keasling Lab ............................................................................................................ 6

Regulation and dynamic control of heterologous pathways ................................................................ 8

Chapter 2: The toxicity of farnesyl diphosphate accumulation ................................................... 11

Table of Contents .................................................................................................................................... 11

Introduction ............................................................................................................................................ 11

Materials and Methods ........................................................................................................................... 12

Strains, plasmids, and media .............................................................................................................. 12

Growth conditions .............................................................................................................................. 13

Transcript analysis sample preparation .............................................................................................. 13

Microarray hybridization .................................................................................................................... 13

Microarray data analysis ..................................................................................................................... 14

Enzyme inhibition by FPP .................................................................................................................... 14

Results ..................................................................................................................................................... 14

Accumulation of FPP causes growth reduction, but only in certain media ........................................ 14

Microarrays of FPP accumulation in different media ......................................................................... 17

Varying [Mg2+] in media ...................................................................................................................... 23

ATP impact on growth......................................................................................................................... 25


Discussion................................................................................................................................................ 30

Chapter 3: Creating a FPP sensor and control of the downstream enzyme................................ 33


Introduction ............................................................................................................................................ 33

ii


Strains and Growth ............................................................................................................................. 34

Promoter selection ............................................................................................................................. 35

Cloning/biobricking ADS ..................................................................................................................... 35

Fluorescence and Amorphadiene (and other sesquiterpene) production ......................................... 36

qPCR, Proteomics, and Metabolites ................................................................................................... 37

Results ..................................................................................................................................................... 38

PhoPQ promoters respond to Mg2+ .................................................................................................... 38

PrstA responds to FPP levels ................................................................................................................. 39

Expression of ADS with PrstA ................................................................................................................ 43

Expanding the promoters used from the microarray ......................................................................... 46

Discussion................................................................................................................................................ 48

Chapter 4: Full feedback and feed-forward control of the pathway ........................................... 51


Introduction ............................................................................................................................................ 51



Cloning ................................................................................................................................................ 53


Proteomics, HPLC ................................................................................................................................ 54

Results ..................................................................................................................................................... 55

Construction of Inverters and Regulated mevalonate pathway ......................................................... 55

PgadE-MevT-MBIS maintained low acetate concentrations ................................................................. 57

Testing gadE promoter region truncations for amorphadiene production........................................ 61

Testing other strength promoters with differential expression ......................................................... 62

Bisabolene and isopentenol production ............................................................................................. 66

Discussion................................................................................................................................................ 67

Chapter 5: Scale up and other applications of the pathway ....................................................... 69


Introduction ............................................................................................................................................ 69


iii

Media and Growth .............................................................................................................................. 70


Proteomics .......................................................................................................................................... 71

Results ..................................................................................................................................................... 71

Glucose feeding ................................................................................................................................... 71

Minimal Media .................................................................................................................................... 72

Discussion................................................................................................................................................ 76

Chapter 6: Perspectives and Future Directions ............................................................................ 78


Conclusions ............................................................................................................................................. 78

Understanding the Toxicity ................................................................................................................. 78

Responsive Promoter .......................................................................................................................... 79

Future Work ............................................................................................................................................ 80

Bibliography .................................................................................................................................. 81

Appendix ....................................................................................................................................... 97

iv

List of Figures Figure 1. Isoprenoid biosynthetic pathways of Escherichia coli and Saccharomyces cerevisiae. Dxs, 1-

deoxy-D-xylulose-5-phosphate synthase; IspC, 1-deoxy-D-xylulose-5-phosphate reductoisomerase; IspD,

4-diphosphocytidyl-2-C-methyl-D-erythritol synthase ................................................................................. 3

Figure 2. The overall goal to create a heterologous mevalonate pathway that no longer requires IPTG as

an inducer, but directly responds to FPP to regulate the pathway. ............................................................. 4

Figure 3. Growth of strains accumulating FPP in different media A) LB and B) M9 MOPS 1% glucose. The

cultures were grown in 20 ml media with 20 mM mevalonate and 0.5 mM IPTG added. ......................... 15

Figure 4. Addition of supplements to media leads to different growth when FPP accumulates. A) LB with

1% glucose added. B) M9 MOPS 1% glucose with the individual components of LB, 10 g/L peptone, 5 g/L

yeast extract, 10 g/L NaCl. .......................................................................................................................... 16

Figure 5. Samples were collected throughout growth in LB to study the toxicity of FPP using microarrays.

10 mM mevalonate was used in LB. ........................................................................................................... 17

Figure 6. Samples collected throughout growth in M9 glucose medium to study the toxicity of FPP using

microarrays. 20 mM mevalonate was used. .............................................................................................. 18

Figure 7. Expression profile during microarray time course in LB. The expression in pADSmut is plotted

versus the expression in pADS. Each gene is represented by a single point. ............................................ 19

Figure 8. Expression profile during microarray time course in M9 MOPS 1% glucose. The expression in

pADSmut is plotted versus the expression in pADS. Each gene is represented by a single point. ............ 20

Figure 9. The expression pattern for the 140 genes identified in Weber et al. as being regulated by RpoS.

The red line is the average of the 140 genes at each time point during the LB microarray time course. . 22

Figure 10. The effect of Mg2+ on FPP toxicity in LB .................................................................................... 24

Figure 11. The effect of Mg2+ on FPP toxicity in M9 MOPS 1% glucose medium. Samples were collected

at the indicated time points for microarray analysis. ................................................................................. 24

Figure 12. The microarray results of FPP toxicity in LB and M9 with low magnesium conditions. The

highlighted sections are A) ribosomal proteins, B) acid resistance, and C) PhoPQ regulon....................... 25

Figure 13. Impact of atpEF knockouts on FPP toxicity with varying glucose concentrations in LB (A) and

M9 MOPS (B). .............................................................................................................................................. 27

Figure 14. Growth of the wildtype DH1 (A) and atpEF knockouts (B) in M9 MOPS 1% glucose media. ... 28

Figure 15. Enzyme inhibition of luciferase by FPP. Two separate experiments showing representative

results. ......................................................................................................................................................... 30

Figure 16. A cartoon summarizing the conditions where FPP toxicity occurs. Low magnesium/Low ATP

conditions are the most toxic, while the addition of either ATP/glucose or Mg2+ offsets or delays the

toxicity. When both Mg2+ and glucose are present in adequate quantities, FPP toxicity is not apparent.

.................................................................................................................................................................... 32

Figure 17. The response of the PhoPQ regulon promoters driving a fluorescent reporter with varying

magnesium in the M9 medium. RFU is relative fluorescent units, fluorescence per OD. ......................... 39

Figure 18. A) The growth of the strains accumulating FPP with the PrstA sensor in LB and M9 MOPS

medium and B)the fluorescent output of the PrstA sensor. 10 mM mevalonate was added to the media.

.................................................................................................................................................................... 40

v

Figure 19. A) The growth of the FPP accumulating strain with the PrstA sensor in M9 MOPS medium

with increasing amounts of mevalonate added to the medium. B) The relative fluorescent output of the

sensor. Solid bars are induced with IPTG, while shaded bars are uninduced. .......................................... 41

Figure 20. A) The growth of the strains accumulating FPP with the PrstA sensor. Strains with the extra

copy of MK were shown to improve amorphadiene production. B) The fluorescent output of the sensor

with and without the extra copy of MK. 10 mM mevalonate was added to the media............................ 42

Figure 21. The effect of using the PrstA sensor with varying terpene synthases. The cultures were fed 5

mM mevalonate and production and fluorescence was measured after 48 hours. Each spot represents a

different variant of a terpene synthase. The fluorescence is relative to pADSmut. Black is pADS, Yellow

are pADS variants, Blue are pBisabolene variants, Brown is pVetispiradiene, and Red is pFarnesene. .... 42

Figure 22. The growth and production of the responsive PrstA-ADS compared to pADS and pADSmut in EZ

RDM 1% glucose. pADS and pADSmut were induced with IPTG, while PrstA-ADS was not. A) The growth

in shake flasks. B)Amorphadiene production measured as a ratio of peak areas to the internal standard

caryophyllene at time points in stationary phase. C)mRNA levels of ADS in each strain, relative to pADS.

D)mRNA levels of ADS in each strain, relative to preinduction. E)Measurement of pathway enzyme

levels at 28 hr by multiple reaction monitoring LC-MS/MS. F)Measurement of intracellular pathway

intermediates at 28 hr by LC-TOF. .............................................................................................................. 44

Figure 23. The effect of overexpressing PhoP and PhoQ. A)The response to the accumulation of FPP.

The different sensors were transformed with pMBIS and pADS or pADSmut and fed mevalonate.

Fluorescence was measured after a day of growth. B)The different constructs were transformed with

pMevT and pMBIS and amorphadiene was measured after a day of growth. ........................................... 46

Figure 24. The effect of increasing the ribosome binding site for ADS. Amorphadiene was measured

after a day of growth. ................................................................................................................................. 46

Figure 25. The 34 additional promoters chosen plus rstA represented in a heatmap of their A)log ratio

expression (ADSmut/ADS). The promoter responses are hierarchically clustered. B)The response of the

promoters, log ratio of fluorescent output (ADSmut/ADS), when cloned in front of the fluorescent

reporter and tested with FPP accumulation (mevalonate fed with pMBIS and pADS or pADSmut). ........ 47

Figure 26. The thirty four promoters plus rstA were cloned 5’ of ADS and tested compared to pADS.

Amorphadiene was measured after 24 hours of growth. .......................................................................... 48

Figure 27. The amorphadiene production of a select group of promoters in EZ RDM + 1% glucose,

measured after 24 hours. ........................................................................................................................... 48

Figure 28. A representative diagram of an inverter to up regulate the downstream gene in response to

FPP. In addition it would up regulate the repressor protein for the lac promoter driving the mevalonate

pathway, thus down regulating expression of the pathway. ..................................................................... 56

Figure 29. The production of amorphadiene from the responsive promoter with the addition of LacI to

repress the mevalonate pathway. 28 hour production in EZ RDM + 1% glucose. ..................................... 56

Figure 30. Production of amorphadiene from the responsive promoters. The promoters were compared

to the three and the two plasmids IPTG induction system. ....................................................................... 57

Figure 31. Production of amorphadiene from the responsive promoters in the 2 plasmid system. 72

hours also monitoring glucose and acetate. ............................................................................................... 59

vi

Figure 32. Protein expression during growth of the PgadE-MevT-MBIS/pADS strain and PlacUV5-MevT-

MBIS/pADS. The proteins are normalized to an internal standard and are measured by MRM on an LC-

MS/MS. Time zero is induction. ................................................................................................................. 60

Figure 33. Production of amorphadiene from the responsive promoters in the wild type, DH1, or acetate

knockout strain, DP. 68 hour production EZ RDM. All strains have IPTG added. ..................................... 61

Figure 34. Shortening the gadE promoter to test which are the essential parts. Production measured in

EZ RDM at 89 hour. ..................................................................................................................................... 62

Figure 35. Testing additional promoters for fluorescence and amorphadiene production. Some from the

earlier tests were redesigned. A)Expression of RFP normalized to the trc promoter after a day of growth.

B)Each promoter is 5’ of MevT-MBIS and is paired with pADS. Amorphadiene measured after 30 hours

growth and sorted in the order of promoter strength. .............................................................................. 64

Figure 36. Testing several constitutive promoters of varying strengths for amorphadiene production.

A)Expression of RFP from the promoters measured 24 hours after growth. B)Each promoter 5’ of MevT-

MBIS coupled with pADS. Measuring amorphadiene at the times indicated............................................ 65

Figure 37. Production of isopentenol and bisabolene from the gadE promoter. 40 hour production

bisabolene, 70 hr isopentenol. IPTG added to all tubes except the PgadE/PrstA combination. .................... 66

Figure 38. Production in minimal medium, M9 MOPS + 1% glucose of A)amorphadiene and B)

bisabolene by the responsive mevalonate pathway. Measured at 50 hours, both cultures have IPTG

added .......................................................................................................................................................... 72

Figure 39. Expression of RFP from the promoters in M9 MOPS + 1% glucose behaves similarly to EZ RDM

+ 1% glucose. ............................................................................................................................................... 73

Figure 40. Protein data from A) EZ RDM + 1% glucose and B)M9 MOPS + 1% glucose for AtoB as a

representative sample of the difference in protein between the strains in the different media. ............. 73

Figure 41. Growth of strains in RAMOS system in KORZ media + 1.5% glucose + 1% Casein amino acids.

A) Growth as a function of time and B)Production of Bisabolene during growth. ..................................... 75

Figure 42. Production of Bisabolene from the acclimated strains in Korz medium (15 g/L glucose, 4 g/L

(NH4)2SO4 with or without 10 g/L casein amino acids in the Ramos shake flasks (300 rpm, 20 ml media, 2

ml dodecane). ............................................................................................................................................. 76

Figure 43. The base vector for the sensors used in this study. AatII and EcoRI sites are highlighted as

well to indicate the removal of LacI and promoter to make pBbA0k-rfp. ................................................ 108

Figure 44. A representative promoter sensor. gadE promoter inserted 5’ of rfp in the pBbA0k-rfp

vector. ....................................................................................................................................................... 109

vii

List of Tables Table 1. The differentially changing genes in the M9 microarrays. Significantly changing genes are bold

with the Z-scores in parentheses. ............................................................................................................... 20

Table 2. The list of strains and plasmids used in this work. ....................................................................... 97

Table 3. The list of all promoters and their sequences used in this work. ................................................ 99

Table 4. The list of ribosome binding sites used in the study. The BglII restriction enzyme site is the 5’

end and is italicized. The start codon of ADS is highlighted in red. ......................................................... 107

Table 5. Constitutive promoters tested for RFP strength and amorphadiene production when cloned 5’

of RFP or MevT-MBIS operon, respectively. Strengths listed are reported from the MIT parts registry.107

viii

Acknowledgments Well, I finally made it after being in College for 12 years. It has been a great time and I’ve learned so much along the way. First, I should give thanks to my advisor. Jay is a rock star and it has been an honor being a part of his lab. It’s almost incomprehensible how he manages such a large lab while fulfilling his many responsibilities and crazy travel schedule. Hopefully some of his awesomeness managed to rub off on me and I’ll be able to one day manage a group as well as he. I always tell potential incoming students I don’t know how he keeps track of everyone’s progress and still manages to give insightful help and guidance when you need it. He also has provided the lab with an incredible facility where we are never limited for resources or smart people and we have the fanciest equipment.

Second, my great lab mates have made it an enjoyable time and have provided me with entertainment and made those long hours and late nights manageable (because regardless of what time you are there, there is usually at least one other person in lab). I should specifically thank the elder grad students who were there when I got started and helped get me running. Doug Pitera, Brian Pfleger, Syd Withers, Wes Marner, Kai Wang, Bernardo da Costa, and Eric Paradise were all great. Lance Kizer ushered me into the lab and showed me how to do microarrays and got me going on my project. My lab mates that started about the same time I got going also were great. Eric Steen, Jeff Dietrich, and Howard Chou are rock stars in their own right, winning business plan competitions and doing world class science. Mario Ouellet has also been a great help and practically knows everything about molecular biology. Will Holtz has also been a valuable lab mate and has done a crazy amount of cloning and helped me out along the way by giving me some of his plasmids and helping out with synthetic biology questions. Becky Rutherford, CJ Joshua, Chris Petzold, and Edward Baidoo have also helped me out along the way and have collaborated together on several projects. Finally, Helcio Burd taught me a lot of the fed-batch fermentation techniques and has been a great help for scaling my strains up.

My family also deserves thanks for their support over the years. Being teachers, my parents infused me with a great thirst for knowledge. My brother and sisters have been there along the way always offering their encouragement. My family out here in the Bay Area has been the cycling team. Without the outlet of bike racing and rides, I think I would have gone crazy (although maybe I could have been done a couple years earlier ). The team has given me the opportunity to travel around the state and see parts I probably wouldn’t have visited otherwise (the Central Valley is nicer than you think, especially in the Spring when the almond trees are blooming and the hills are green). Finally, I met my fiancé through the cycling team. Thank you Joanna, you’ve been a wonderful partner these last couple years and I know you’ll be able to write your thesis faster than I have.

1

Chapter 1: Introduction, Motivation, Outline Motivation and Outline ............................................................................................................................. 1

Background ............................................................................................................................................... 4

Isoprenoids in E. coli ............................................................................................................................. 4

Microarrays and their uses in metabolic engineering .......................................................................... 5

Microarrays in the Keasling Lab ............................................................................................................ 6

Regulation and dynamic control of heterologous pathways ................................................................ 8

Motivation and Outline

The Keasling lab has been at the forefront of metabolic engineering of E. coli and S. cerevisiae and development of synthetic biology tools to quickly implement changes to the native metabolism of these organisms to produce new chemicals, pharmaceuticals, biofuels, and endless other potential molecules. We take our cues from nature and a common workflow is to find the enzymes that perform desired chemical reactions to make compounds of interest and then move them to our well characterized hosts, after which we work on improving their performance by a variety of means. A common analogy is to call these hosts our chassis into which we can integrate a pathway resulting in a chemical factory conveniently contained inside a cell which we can be grown easily at moderate temperature and pressure. Many of these products we wish to make are found in their native sources at extremely low quantities and make commercial development prohibitively expensive in many cases. Thus the potential to produce these molecules at higher titers, after many rounds of improvement in their new host, from simple starting materials (sugar, ammonia, and mineral salts) is boundless. We hope to transform the traditional routes to industrial chemicals, based on the petroleum industry, to a cleaner, renewable, green route based on our advancements.

When I started in the Keasling Lab, our focus was on the production of a group of compounds called the isoprenoids, Figure 1, specifically the production of the anti-malaria drug, artemisinin. Our lab had engineered E. coli to express the heterologous mevalonate pathway from S. cerevisiae and a terpene synthase from Artemisia annua, amorphadiene synthase, to produce amorphadiene, the ultimate precursor to the drug. We have since shifted our focus to the production of biofuels from the isoprenoid pathway by stopping the pathway at shorter molecules or by expressing different terpene synthases. The isoprenoids in their native environments are highly functionalized and tailored compounds on a hydrocarbon backbone, so the base molecule itself could be a starting point for a fuel-like molecule.

The expression of this new heterologous pathway in E. coli was toxic when some of the pathway intermediates accumulated. Previous work in the lab focused on understanding the toxicity associated with the accumulation of HMG-CoA, the second intermediate in the pathway. However, the final intermediates, the isoprenyl diphosphates, also caused toxicity when they accumulated, and had yet to be understood. While the expression of amorphadiene synthase was enough to prevent their accumulation and remove toxicity, other terpene

2

synthases don’t express as well in E. coli, given that many of them originate in plants and other higher organisms. It was with this knowledge that I set out to try to understand the toxicity using a systems biology approach.

The field of systems biology has developed over the years and has made significant progress as technologies have improved. Realizing that the cell’s physiology is an interconnected network of interactions at many different hierarchies, a holistic view of the cell is required to understand how our engineering impacts its normal physiology. The study of the cell at the various levels—DNA with genome sequencing, RNA with microarrays, proteins with proteomics, and metabolic pathways with metabolomics and flux analysis—gives a complete picture and allows us as engineers to make more intelligent changes to affect our improved phenotypes. I joined the lab and started studying the cell’s response to these heterologous pathways at the RNA level. With this knowledge, I wanted to move beyond the typical application of knocking out or upregulating significantly changed genes to reduce the toxicity phenotype. I wanted to apply this knowledge to the pathway itself and reengineer responsiveness to the heterologous pathway, Figure 2.

The typical approach to engineers when they add a new heterologous pathway to a cell is to express the pathway under the control of a tightly regulated promoter so they can express the pathway only when they want. As these pathways progress to the large scale production, these inducible promoters are often switched to constitutive promoters of a defined strength because the addition of an exogenous inducer is economically prohibitive. The use of these promoters that are either inducible or constitutive loses a connection to the cell’s native physiology. While steps are taken to find the optimal expression level, the enzymes are expressed at a given level, and do not respond to the cell’s changing environment or growth conditions. The response is static. I wanted to introduce dynamic control to the pathway regulation by reconnecting the levels of pathway intermediates to the pathway enzyme expression.

The following chapters highlight the work to move from understanding the toxicity to controlling the pathway expression. The rest of this chapter will provide background into current advances into microarrays to gain understanding of E. coli and other engineered microbes’ physiology and highlight new advances in the engineering of heterologous pathways. The microarrays offered an easy method to identify genes that respond to the pathway intermediates’ accumulation. Chapter 2 will highlight the responses to the intermediate toxicity and explain why under different growth conditions the toxicity abates. Chapter 3 takes the knowledge gained from the microarray studies and develops a sensor for intermediate accumulation and then uses the response to express the downstream enzyme. Chapter 4 completes the entire dynamic regulation, which improves production over the inducible promoters commonly used. And finally, chapter 5 will highlight the first attempts at scaling the responsive pathway up to a fed-batch fermentation to test its robustness and applicability to larger scale production.

3

Figure 1. Isoprenoid biosynthetic pathways of Escherichia coli and Saccharomyces cerevisiae. Dxs, 1-deoxy-D-

xylulose-5-phosphate synthase; IspC, 1-deoxy-D-xylulose-5-phosphate reductoisomerase; IspD, 4-

diphosphocytidyl-2-C-methyl-D-erythritol synthase; IspE, 4-diphosphocytidyl-2-C-methyl-D-erythritol kinase; IspF,

2C-methyl-D-erythritol 2,4-cyclodiphosphate synthase; IspG, 1-hydroxy-2-methyl-2-(E)-butenyl 4-diphosphate

synthase; IspH; atoB, acetoacetyl-CoA thiolase; ERG13, hydroxymethylglutaryl-CoA synthase (HMGS); HMG1,

hydroxymethylglutaryl-CoA reductase (HMHR); ERG12, mevalonate kinase (MK); ERG8, phosphomevalonate kinase

(PMK); ERG19, mevalonate pyrophosphate decarboxylase (MPD); Idi, isopentenyl pyrophosphate isomerase; IspA,

farnesyl pyrophosphate synthase; ADS, amorphadiene synthase

CO2H

O

OHC

OH

O OH

OH

HO

OH

HO

OP(OH)2O

OP(OH)2O

OP(OH)2O

OPO(OH)OP(OH)2O

SCoA

O

HSCoA

SCoA

OO

HSCoA

S-CoA

OOH

HO2C

HSCoA

OH

OH

HO2C

ATP

ADP

OP(OH)2O

OH

HO2C

ATP

ADP

CO2

Pi

O P

O-

O

O

HO

HO

HO

P

O

O

O- O

HO

HO

N

NNH2

O

O

P

HO

HO

O

O-

OP

O

O

O-

PPi

CTP

ATP

ADP

CMP

MK

PMK

MPD

O P

O-

O

O

HO

HO

O

P

O

O

O- O

HO

HO

N

NNH2

O

P

O

-O

-O

OPO(OH)OP(OH)2O

OH

OPO(OH)OP(OH)2O

Dxs

IspC

AtoB

Acetyl-CoA

HMGS Acetyl-CoA

HMGR

2 NADPH

2 NADP

IspD

IspE

IspF

IspG

IspH

1 : 5

ATP

ADP

OPO(OH)OP(OH)2O

OH

HO2C

Idi

IspA2X

OPO(OH)OP(OH)2O

Amorphadiene

FPP

DMAPP

IPP

ADS

Glyceraldehyde-3-phosphate

Acetyl-CoA

Pyruvate

Acetoacetyl-CoA

HMG-CoA

Mevalonate

Mevalonate-5-phosphate

Mevalonate-5-diphosphate

1-deoxy-D-xylulose-5-phosphate

2-C-methy-D-erythritol-4-phosphate

Non-Mevalonate Pathway Mevalonate Pathway

4

Figure 2. The overall goal to create a heterologous mevalonate pathway that no longer requires IPTG as an inducer, but directly responds to FPP to regulate the pathway.

Background

Isoprenoids in E. coli

Isoprenoids are a class of compounds that are derived from the universal five carbon precursor, isopentenyl diphosphate (IPP). IPP and its isomer, dimethylallyl diphosphate (DMAPP), are condensed in subsequent head to tail reactions to create longer chain molecules. At the various stages of extension, additional enzymes cyclize then functionalize the molecules, creating a diverse group of natural products (McCaskill & Croteau, 1997). Throughout nature the isoprenoids play important and diverse roles in many organisms for electron transport, plant defense and signaling, membrane fluidity, and pigments among many others; however, due to their limited abundance in their native sources and difficult routes of chemical synthesis (Nicolaou et al., 1994) they are in short supply. Regardless, due to their chemical properties, they have found many industrial and medicinal uses as fragrances, flavors, nutraceuticals, and pharmaceuticals, yet alternate production methods are needed.

In E. coli, IPP and DMAPP are synthesized via the methylerythritol phosphate (MEP) pathway then extended to form a component of quinones, functioning as electron carriers in the cytoplasmic membrane, and sugar-carrying lipids in the peptidoglycan layer. The Keasling Lab has introduced the S. cerevisiae mevalonate pathway into E. coli to bypass the regulation of

5

the native MEP pathway to increase the supply of the universal isoprenoid precursors (V. J. J. Martin, Pitera, Withers, J. D. Newman, & Jay D Keasling, 2003), thus unlocking the door to the largest class of natural compounds. The primary effort of our group had been the production of amorphadiene, the sesquiterpene precursor to the anti-malaria drug, artemisinin. Other groups have made recent progress for the production of precursors to the anti-cancer drug, taxol (Ajikumar et al., 2010). With the expression of the functionalization enzymes, the realization of producing these molecules in a heterologous host is getting closer and closer (M. C. Y. Chang, Eachus, Trieu, Ro, & Jay D Keasling, 2007; Ro et al., 2006), although partial chemical synthesis remains an option from the ultimate precursors, while minimizing the need to rely on natural resources for these valuable chemicals.

In addition to their medicinal uses, isoprenoids have recently been investigated for their use in industrial chemicals and biofuels (S. K. Lee, H. Chou, Ham, T. S. Lee, & Jay D Keasling, 2008). Due to their polymer-like features of creating a 5, 10, 15, and longer carbon molecules, isoprenoids can fill the role of a variety of fuel-like molecules (after hydrogenation to remove the double bonds). Shorter chain molecules are more gasoline-like, while the longer chain molecules are more diesel-like. The regular branching can help increase the octane, but lower the cetane number, while also improving the fluidity at lower temperatures. Recently, our lab has created/joined the Joint Bioenergy Institute to push isoprenoids towards this future and Amyris has several patents identifying isoprenoids for their potential fuel or chemical properties (N. Renninger, J. Newman, & Reiling, 2010).

Microarrays and their uses in metabolic engineering

Microarrays are an important tool of systems biology, studying how every gene is expressed in parallel and thus connecting DNA to expressed proteins on a cell-wide level. They were first developed in the early 1990’s (J. L. DeRisi, Iyer, & P O Brown, 1997; Lashkari, 1997; C. S. Richmond, Glasner, Mau, Jin, & F. R. Blattner, 1999; Schena, Shalon, Davis, & P. O. Brown, 1995) as an extension of Southern Blot technology. The most important advance over the older technology, done on nylon membranes, was depositing the DNA probes on a solid substrate. This allowed easier handling, more reproducible results, less background signal, and higher density of probes. These initial arrays printed full length PCR products of each gene and as the years have progressed the length of the printed probe has decreased. Current designs feature a minimum oligomer length of 25 nucleotides (Affymetrix), while Nimblegen and Agilent arrays typically have 60-mers. Further, the use of oligonucleotides only relies on a sequence without a priori knowledge of open reading frames and removes the tedious step of PCR.

Another important advancement has been the increase in printing density. The first arrays had 1000s of probes printed. With the use of robotic printers, 104 features could be spotted on a slide. Further advances of lithographic features and in situ printing of nucleotides allowed 105 features on a slide (Lockhart et al., 1996; E. F. Nuwaysir et al., 2002). With increased number of probes in an array, you are afforded more replicates for each probe and multiple probes per gene (C.-C. Chou, C.-H. Chen, T.-T. Lee, & Peck, 2004; Kreil, R. R. Russell, & S. Russell, 2006). Not only is this important for statistical analysis, but the increased density allows for probes covering the entire genome at high resolution, not just open reading frames. This advance has led to tiling arrays, which have been used to detect novel small peptides and

6

RNAs (Thomassen et al., 2010) and determine post transcriptional processing patterns (Stead et al., 2010), among others. Tiling arrays have also been coupled with chromatin immunoprecipitation (ChIP), termed ChIP-chip, to detect protein-DNA interactions across the entire genome (B.-K. Cho, Knight, & B. Ø. Palsson, 2008).

In parallel to the development of microarrays, a burgeoning field of statistics has emerged to handle the immense amount of data created with each experiment (Bolstad, Collin, K. M. Simpson, Irizarry, & T P Speed, 2004; Churchill, 2002; Dharmadi & R. Gonzalez, 2004; M. L. Lee, Kuo, Whitmore, & Sklar, 2000). The raw fluorescence intensities extracted from each probe undergo normalization to correct for local artifacts due to the spotting process (Colantuoni, Henry, Zeger, & Pevsner, 2002) or to combine the multiple probes for each gene into a single data point (Bolstad, Irizarry, Astrand, & T P Speed, 2003). After a log ratio (because typically we are concerned with relative expression changes) is obtained for each gene, the significance is determined by comparing the replicate probes within each slide and the technical or biological replicate slides. The analysis of a single gene in comparison to all the other genes requires a modification beyond the simple T-test to control the amount of false positives, termed the false discovery rate (Benjamini & Hochberg, 1995; Ge, Dudoit, & Terence P. Speed, 2003; Tusher, Tibshirani, & Chu, 2001).

Apart from determining significance, patterns in the data are difficult to distinguish without extra analysis. Hierarchical clustering the data groups genes with similar expression patterns by stepwise calculating the distance between each pair (Eisen, Spellman, P O Brown, & Botstein, 1998). Other clustering algorithms to order data include K-means clustering, principle component analysis, self-organizing maps, and support vector machines. Online tools available at www.microbesonline.org and www.ecocyc.org also aid in grouping based on biological information (Dehal et al., 2010; Karp, 1996). MicrobesOnline.org also serves as a data repository and a tool to compare expression changes across multiple experiments. Most publications require that microarray data be submitted to a data repository, such as MicrobesOnline or the Gene Expression Omnibus (GEO) (Edgar, 2002), and meet certain guidelines (termed MIAME, minimal information about a microarray experiment) before publication.

The Keasling Lab has used the full spectrum of arrays, and this work spans the use of printed oligonucleotide arrays synthesized in house to commercial high density arrays. But with the advance of next generation sequencing, the future of microarrays might be replaced by these technologies as they can provide much better sensitivity, dynamic range, and even give absolute quantification (Z. Wang, Gerstein, & Snyder, 2009). Nanostring technologies have also developed a technology that more accurately quantifies gene expression and does not require any enzymatic reactions (Geiss et al., 2008).

Microarrays in the Keasling Lab

In E. coli, microarrays have been used to characterize the gene expression response to heat shock, carbon substrate, and nitrogen limitation, and growth phase transitions among others (Hua, C. Yang, Oshima, Mori, & Shimizu, 2004; M.K. Oh, Cha, S. G. Lee, Rohlin, & J.C. Liao, 2006; C. S. Richmond, Glasner, Mau, Jin, & F. R. Blattner, 1999; Tani, Khodursky, Blumenthal,

http://www.microbesonline.org/

http://www.ecocyc.org/

7

Patrick O Brown, & Matthews, 2002). In addition, microarrays have been used to characterize multi-gene and multi-operon regulation by LexA (response to UV), CRP (catabolite repression), OxyR (response to oxidative stress), as well as LRP (coordination of anabolic pathways) (Courcelle, Khodursky, Peter, Patrick O. Brown, & Hanawalt, 2001; Hung, Baldi, & Hatfield, 2002; D. Zheng, Constantinidou, Hobman, & Minchin, 2004; M. Zheng et al., 2001). Many of these efforts elucidated the complex regulatory mechanisms of E. coli as well as identified functions to previously unknown genes. While by no means exhaustive, the above list of experiments has helped contribute to the complex regulatory and physiological responses of E. coli. Additional gains are often realized when microarray studies are combined with studies of other cellular components to provide a holistic view of the complex interactions inside the cell (B.-K. Cho et al., 2009; Covert, Knight, Reed, Herrgard, & B. O. Palsson, 2004; Faith et al., 2007; Sang Yup Lee, 2009).

Microarrays have also been used to characterize E. coli for metabolic engineering

purposes. The response to isopropyl--D-thiogalactopyranoside (IPTG) induction for overexpression of recombinant proteins (M K Oh & J C Liao, 2000), high cell density fermentation (R T Gill, DeLisa, Valdes, & Bentley, 2001), and the metabolic burden of maintaining plasmids (Ow, Nissom, Philp, S. Oh, & Yap, 2006) are all relevant to the engineering of exogenous pathways. Microarrays have also been used to identify genes that correlate with improved tolerance or production of a metabolite. Gill et al. used microarrays to determine which genes were responsible for increased tolerance to Pine-Sol (R T Gill, Wildt, Y. T. Yang, Ziesman, & G Stephanopoulos, 2002), while Gonzalez et al. studied the genes responsible for ethanol tolerance by comparing a strain evolved to overproduce ethanol with an ethanol resistant strain (R. Gonzalez et al., 2003). Gill and coworkers have gone on to expand their work to create a technology called SCALEs and to interrogate the effects of cDNA libraries or up- or down-regulation of entire genomes to effect a certain phenotype (M. Lynch, T. Warnecke, & R. Gill, 2007; T. E. Warnecke et al., 2010; Warner, Reeder, Anis Karimpour-Fard, L. B. A. Woodruff, & Ryan T Gill, 2010; L. B. Woodruff & Ryan T Gill, 2011).

In the Keasling Lab we have used microarrays to study E. coli’s response to the exogenous mevalonate pathway as well as E. coli’s response to alcohol exposure in addition to the study of a poorly characterized organism. Previously, L. Kizer et al. used microarrays to study the response of E. coli to the toxicity associated with the accumulation of the mevalonate pathway intermediate, HMG-CoA (Kizer, Pitera, Pfleger, & Jay D Keasling, 2008). This study found that the accumulation of HMG-CoA upregulated fatty acid biosynthesis, and the authors hypothesized that this was a result of the exogenous CoA molecule inhibiting these fatty acid biosynthesis steps.

In work I contributed to, we studied the response of E. coli to the exogenous addition of several potential biofuels: isopentenol and butanol. While the exogenous addition might trigger one stress, the internal production of these molecules might pose additional problems. We sought to study the differences between the production and exposure to these molecules; however, the current production levels were too low to see any changes in growth or transcription. The exogenous addition of butanol to E. coli did cause a dramatic decrease in growth and transcriptional response (Rutherford et al., 2010). The primary response was a

8

disruption in membrane stability resulting in respiratory genes, oxidative stress, and heat shock and cell envelope stress genes being activated. Similar work with isobutanol by another lab also showed similar transcription patterns (Brynildsen & James C Liao, 2009). Follow up work is being done by B. Rutherford to use these responding genes to over express and down regulate expression to mitigate the stress in advance.

We also studied a less characterized organism, the archaeon Sulfolubus acidocaldarius, with microarrays. This organism has received increased attention in our lab because it can grow at high temperatures and low pH. Furthermore, fermentation and ethanol production at these higher temperatures could bypass the costly distillation steps associated with production in yeast (with the caveat that ethanol production can still be engineered in this organism). This organism has been observed to not exhibit typical diauxic behavior when fed multiple sugars, another beneficial characteristic for biofuel production from biomass. We used microarrays to elucidate this unusual behavior by comparing cells grown on a single sugar (either glucose or xylose) to cells grown on the mixture of sugars. While we were not able to find a specific regulator responsible for the synchronous catabolism, putative glucose transporters were identified that had been annotated as maltose transporters (Joshua, R. Dahl, Benke, & Jay D Keasling, 2011). Further experiments are underway to better characterize the synchronous metabolism.

Regulation and dynamic control of heterologous pathways

The introduction of heterologous pathways into a new host typically removes the native transcriptional control. Initially, metabolic engineers and synthetic biologists regulate the heterologous pathway in its new host with inducible promoter systems, where the expression can be initiated by the addition of a chemical such as IPTG, arabinose, or anhydrotetracycline (Lutz & Bujard, 1997). While each of these inducers and corresponding promoter systems has their characteristic properties, they are well characterized and can meet the engineer’s requirements in the laboratory. However, apart from the pharmaceutical industry, where the economic margins are great; the burgeoning industrial chemical and biofuels industry cannot afford the addition of expensive components to the media. There are alternative promoter options that have been used industrially (primarily in the more mature pharmaceutical industry still), such as constitutive or promoters responding to environmental signals, for example temperature, oxygen, pH, or a medium component (Neubauer & Winter, 2001; Sawers & Jarsch, 1996).

While these alternate methods of induction are popular for their economics and simple expression of a single enzyme, they lack the ability to respond to dynamic changes in the environment, growth phase of the cell, or the pathway metabolite levels. Furthermore, the expression of an entire heterologous pathway presents additional problems such as unbalanced expression of enzymes leading to accumulation of an intermediate. Thus, the expression of an entire pathway requires additional levels of control. Synthetic biologists have developed these methods and have created increasingly complex genetic circuits to impart novel functions to E. coli, yet very few examples of dynamic control of a heterologous pathway exist, reviewed by my lab mate (Holtz & Jay D Keasling, 2010).

9

There are several control points at which the relative expression level can be adjusted to fine tune expression of the heterologous pathway. At the promoter level, synthetic promoters have been designed to allow fine tuning of gene expression (Alper, Fischer, Nevoigt, & Gregory Stephanopoulos, 2005; Cox, Surette, & Michael B Elowitz, 2007; Hammer, Mijakovic, & Jensen, 2006; Miksch et al., 2005). These approaches typically give constitutive expression, but allow the engineer to screen the library to find the optimal expression level for each gene. At the post-transcriptional stage, the mRNA stability can be varied to tailor the relative expression levels of enzymes (Babiskin & Christina D Smolke, 2011; Pfleger, Pitera, Christina D Smolke, & Jay D Keasling, 2006; C D Smolke, V. J. Martin, & J D Keasling, 2001). At the translational level, the ribosome binding strength can be adjusted to manipulate the amount of protein synthesized for a given mRNA amount. Using a thermodynamic model, the ribosome binding site can be modified to fine tune translation initiation (H. M. Salis, Mirsky, & C. A. Voigt, 2009). Finally, at the post-translational level, the protein stability can be modified by adding a degradation tag to alter the half-life (Andersen, Sternberg, & Poulsen, 1998).

E. coli has a variety of sensing mechanisms that have been engineered to impart novel functions. In many cases, these sensing devices are naturally connected to gene expression, either through direct interaction with DNA or RNA, or via another protein. Allosteric protein interactions are a common regulatory mechanism for metabolic pathways. A pathway intermediate can bind a transcription factor, which then undergoes a conformational change which alters its DNA binding properties. While engineering a protein to detect a novel metabolite is difficult, recent advancements in computational design and protein engineering have facilitated construction of DNA binding proteins with novel functions (Dietrich, McKee, & Jay D Keasling, 2010; Looger, Dwyer, J. Smith, & H. Hellinga, 2003; Marvin & H. W. Hellinga, 2001).

Naturally occurring RNA secondary structures called aptamers are also able to bind metabolites (Noeske, Richter, Stirnal, Schwalbe, & Wöhnert, 2006). Because these structures are already composed of nucleotides, they can be integrated directly into the mRNA to cause rapid changes in gene expression in response to the metabolite (Win & Christina D Smolke, 2007; Winkler, Nahvi, Roth, J. A. Collins, & Breaker, 2004). When the aptamer binds the metabolite, a conformational change occurs which can reveal or occlude the ribosome binding site, thus actuating gene expression. Unlike their DNA binding allosteric protein counterparts, aptamers allow direct actuation of gene expression without the need for protein expression and folding, increasing the speed of response to change. Additionally, they are easier to engineer due to the less complex chemistry and number of nucleotides versus amino acids.

An additional type of sensing and expression system is two component systems. They comprise of a sensor kinase and response regulator, and unlike the previous two mechanisms typically straddle the membrane and can sense extracellular signals as well (E. A. Groisman, 2001). Upon sensing a stimulus, the sensor kinase transfers a phosphoryl group to the response regulator. The modified response regulator changes its affinity for DNA and affects transcriptional initiation. Typically, the sensor kinase only recognizes its cognate response regulator through protein-protein interactions, however recent work has focused on rewiring

10

the connection to impart novel functions to the machinery (Groban, Clarke, H. M. Salis, S. M. Miller, & C. a Voigt, 2009; Ninfa et al., 2007; Skerker et al., 2008).

The best example of dynamic control of a metabolic pathway to date has used a two component system’s machinery. Liao et al. increased production of lycopene by expressing rate limiting enzymes behind the glnAp2 promoter (Farmer & James C Liao, 2000). The glnAp2 promoter recognizes the nitrogen utilization regulator, NTR. By knocking out the regulator’s cognate sensor kinase, the regulator was able to respond to acetyl phosphate accumulation, which is an indicator of glycolytic overflow. When acetyl phosphate began to accumulate, the response regulator increased expression of the rate limiting steps (which were native E. coli enzymes, not part of the heterologous lycopene pathway), diverting excess carbon to the product of interest rather than the waste product acetate.

Despite this success, most non-native and even many native metabolic intermediates do not have regulatory systems (allosteric, aptamer, or two component system) that respond to concentration and that could be used to regulate a heterologous metabolic pathway. None the less, synthetic biologists have built increasingly complex genetic circuits and the ability to engineer aptamers or proteins to meet these demands is just around the corner. Reviewed in these papers (Boyle & Silver, 2009; McDaniel & Weiss, 2005; H. Salis, Tamsir, & C. Voigt, 2009; C. A. Voigt, 2006), the last 10 years have seen the creation of oscillating circuits (M B Elowitz & Leibler, 2000; Fung et al., 2005; Stricker et al., 2008), complex logic gates (Anderson, C. A. Voigt, & Arkin, 2007; Yokobayashi, Weiss, & Arnold, 2002), and cells that perform novel tasks such as detecting spatial patterns (Basu, Gerchman, C. H. Collins, Arnold, & Weiss, 2005), recording images of light (Levskaya et al., 2005), and detecting and invading cancer cells (Anderson, Clarke, Arkin, & C. a Voigt, 2006). Coupling these advances in controlling the cell to the advances in pathway engineering and natural product discovery will usher in a new era of chemical, pharmaceutical, and commodity production in microbes that will bring about great changes to the petroleum and synthetic chemistry industries and hopefully make for a greener planet and exciting future for our field.

11

Chapter 2: The toxicity of farnesyl diphosphate accumulation

Table of Contents Introduction ............................................................................................................................................ 11


Strains, plasmids, and media .............................................................................................................. 12

Growth conditions .............................................................................................................................. 13

Transcript analysis sample preparation .............................................................................................. 13

Microarray hybridization .................................................................................................................... 13

Microarray data analysis ..................................................................................................................... 14


Results ..................................................................................................................................................... 14

Accumulation of FPP causes growth reduction, but only in certain media ........................................ 14

Microarrays of FPP accumulation in different media ......................................................................... 17

Varying [Mg2+] in media ...................................................................................................................... 23

ATP impact on growth......................................................................................................................... 25


Discussion................................................................................................................................................ 30

Introduction

The introduction of heterologous pathways into new organisms is fraught with challenges and often results in reduced levels of growth. However, to produce potential fuels, chemicals, or pharmaceutical products at commercial titers, it is often necessary to transfer the native biosynthetic pathways to a more tractable host and express the enzymes at high levels. The expression of many heterologous enzymes from multi-copy plasmids often leads to a host of toxicity problems, even from maintaining the plasmids themselves (Ow et al., 2006). High level expression of a single protein in E. coli often leads to growth reduction (M K Oh & J C Liao, 2000), so the expression of entire pathways is a challenging endeavor.

Often expressing the pathway at lower levels results in improved performance (Jones, Kim, & J D Keasling, 2000); however, the unbalanced expression of the pathway enzymes can also lead to the accumulation of inhibitory molecules (Ajikumar et al., 2010; Barbirato, Grivet, Soucaille, & Bories, 1996; Zhu, Lawman, & Cameron, 2002) The heterologous pathway often loses its native regulation when expression is moved to an inducible promoter. This gives the engineer ultimate control of when the enzymes are expressed, and often some control of the strength of the expression, but the enzyme levels are expressed at a static level, without the native response to balance expression.

12

To improve production of isoprenoids in E. coli the mevalonate pathway from S. cerevisiae was introduced under control of IPTG inducible promoters. Growth reduction was observed with the expression of the enzymes to convert acetyl-CoA to mevalonate. It was observed that distinct levels of the two enzymes producing and consuming HMG-CoA were required for optimal growth (Pitera, Chris J Paddon, J. D. Newman, & Jay D Keasling, 2007). Microarray analysis was performed and the conclusion was that the accumulation of HMG-CoA caused increased expression in the fatty acid biosynthesis pathway (Kizer, Pitera, Pfleger, & Jay D Keasling, 2008). Further experiments were performed and the accumulation of HMG-CoA was shown to inhibit fatty acid biosynthesis and the accumulation of the native substrate, malonyl-CoA, which then caused a cascade of downstream effects due to the changes to the cell membrane.

Additionally, the exogenous mevalonate pathway caused toxicity when the isoprenyl diphosphate intermediates were allowed to accumulate (V. J. J. Martin, Pitera, Withers, J. D. Newman, & Jay D Keasling, 2003). Each intermediate gave a different reduction in growth, with the most severe being IPP, then DMAPP, FPP, and finally mevalonate phosphate. This was alleviated when an active amorphadiene synthase was expressed that consumed the intermediates. However, expression of other terpene synthases with lower expression, a problem often encountered when expressing enzymes originating in plant species, resulted in much lower production titers, and thus accumulation of these intermediates could still pose a problem (V. J. Martin, Yoshikuni, & J D Keasling, 2001). The root cause of the toxicity was never investigated, so microarray analysis was performed to understand the transcriptional response to the accumulation. For the purpose of this work, we only focused on the toxicity of the 15 carbon intermediate, FPP, for the production of sesquiterpenes. Follow up work was performed with enzyme assays and media supplementation to reduce the toxicity and ultimately the knowledge was used in further chapters to create a responsive, dynamic mevalonate pathway.

Materials and Methods

Strains, plasmids, and media

The strains and plasmids used in this study are described in the appendix in Table 2. The bottom half of the mevalonate pathway, pMBIS, from Saccharomyces cerevisiae was used along with the codon optimized amorphadiene synthase, pADS, from Artemisia annua (V. J. J. Martin, Pitera, Withers, J. D. Newman, & Jay D Keasling, 2003). To accumulate FPP, an inactive version of ADS was created by mutating three catalytic aspartate residues in the first aspartate rich motif (A. Chen, Kroon, & C D Poulter, 1994). The acetate producing pathways, ackA, pta, poxB,

and a subunit of the ATP synthase, atpEF, were deleted using the red protocol described by Datsenko Wanner (Datsenko & Wanner, 2000).

Medium components and chemicals were purchased from Sigma-Aldrich, EMD chemicals, and Difco. Mevalonolactone (Sigma) was converted to mevalonate by adding 1.02 volumes 2 M KOH to 1 volume 2 M mevalonolactone in water. The mixture was incubated at 37 °C for 30 minute, and then filter sterilized. Luria Broth Miller (LB) (EMD) and a modified M9 MOPS medium were used for cell growth and microarray studies. Magnesium sulfate and

13

glucose were added to the media at varying levels as indicated. The modified M9 MOPS medium consists of 1x M9 salts (Difco) (1 g/L NH4Cl, 0.5 g/L NaCl, 6.6 g/L Na2HPO4 (anhydrous),

3 g/L KH2PO4), 2 mM MgSO4, 100 M CaCl2, 40 mM MOPS, 1x trace metal containing (3x10-6 mM (NH4)6(MO7)24, 4x10-4 mM HBO3, 3x10-5 mM CoCl2, 10-5 mM CuSO4, 8x10-5 mM MnCl2, 10-5

mM ZnSO4), 4.5 g/ml thiamine, 0.01 mM FeSO4 (heptahydrate), and glucose (Neidhardt, Bloch,

& D. F. Smith, 1974). Tetracycline (10 g/ml) and carbenicilin (100 g/ml) were added as appropriate. Cell cultures were grown at 30 °C.

Growth conditions

Starter cultures of E. coli containing pMBIS and pADS or pADSmut were inoculated from single colonies and incubated overnight at 30 °C. Before growth analysis in M9 medium, the strains were acclimated to the minimal medium by successively passing the cultures into fresh medium until overnight growth reached an optical density at 600 nm (OD) > 2. Acclimated

cultures were aliquoted into single-use vials and frozen with 15% glycerol. 50 l of thawed, acclimated cultures were used to inoculate 5 ml of fresh media to begin experiments.

Overnight starter cultures were diluted in fresh media to an OD ~0.05, incubated at 30

°C with continuous shaking, and induced with 500 M isopropyl -D-thio-galactoside (IPTG) and mevalonate as indicated at an OD ~0.2-0.3. Growth curves and microarray analysis studies were cultured in baffled flasks. Samples were taken throughout exponential phase growth and stationary phase. The OD of samples was measured using a UV spectrophotometer (Beckman). Experiments were repeated at least twice to confirm trends in growth.

Transcript analysis sample preparation

At specific times during growth, approximately 1x 1010 cells were harvested by centrifugation and flash frozen in liquid nitrogen and stored at -80 °C until further processing. Total RNA was isolated from the cell pellets using Qiagen RNEasy Midi kits. The RNA quality was checked using an Agilent Bioanalyzer and the quantity was measured with a Nanodrop

(Thermo). cDNA was synthesized following the enzyme manufacturer’s instructions in 40 l

reactions using 20 g RNA, 12.5 g random hexamers (Invitrogen), 1x reverse transcription buffer (Invitrogen), 0.5 mM (each) dATP, dGTP, dCTP, 0.1 mM dTTP, and 0.4 mM amino-allyl dUTP (Ambion), 0.01 mM DTT (Invitrogen), 40 units Superase-In (Ambion), and 600 units Superscript III (Invitrogen). The cDNA was base hydrolyzed in 100 mM NaOH-10 mM EDTA at 65 °C for 10 minutes and then neutralized in HEPES, pH 7.0, at a final concentration of 500 mM. The remaining tris in the cDNA samples was removed by three buffer exchange spins in Microcon YM-30 columns (Millipore) and eluted in water. The cDNA was dried in a speed vac to

5 l and labeled with either Alexa Fluor 555 or 647 dyes (Invitrogen) following the manufacturer’s protocol.

Microarray hybridization

The E. coli microarrays were produced by spotting 70-mer oligonucleotides (Operon) onto SuperAmine2 substrates (Telechem) using a robotic arrayer (Gene Machines). The oligonucleotides were designed to probe every open reading frame of E. coli MG1655. Oligonucleotides were spotted in triplicate on the surface in different locations to prevent local

hybridization effects (NCBI GEO# GPL8811). Approximately 5 g of labeled cDNA per channel of

14

detection were pooled in Ambion Slide Hybridization 3 buffer and hybridized to the microarrays in a Tecan hybridization station. The hybridization included a prehybridization step (5x SSC [1x SSC is 0.15 M NaCl plus 0.015 M sodium citrate]-0.2% sodium dodecyl sulfate [SDS], 1% bovine serum albumin, 42 °C, 60 min), 17 hour hybridization (40 °C, medium agitation), two low stringency washes (1x SSC, 0.2% SDS, 42 °C, 2 min each), two high stringency washes (0.1x SSC, 0.2% SDS, 25 °C, 2 min each), and two final washes (0.1x SSC, 25 °C, 2 min each). Following

hybridization, slides were scanned at 10 m resolution with a GenePix 4200A microarray scanner.

Microarray data analysis

Images were analyzed using GenePix Pro to determine fluorescence intensity for each spot in both channels and globally normalize. Local hybridization and intensity-dependent artifacts across the array were removed by loess normalization using Standardization and Normalization of Microarray Data (SNOMAD) (Colantuoni et al., 2002). The log ratio of each gene was calculated using the normalized expression intensities and averaging the triplicate data points on each slide and replicate hybridization data sets. The Z score for each gene was

calculated using the formula,

√ ∑ , where 0.25 is a pseudovariance term

(Mukhopadhyay et al., 2006). The data sets are stored online at www.microbesonline.org (Dehal et al., 2010) and NCBI’s Gene Expression Omnibus (GEO) (Edgar, 2002)

To visualize the data sets, www.microbesonline.org was used to map to genes with common functions and metabolic pathways. Pathway Tools and www.ecocyc.org were used to highlight regulatory architecture (Karp, 1996). The data were hierarchically clustered to group genes with similar expression patterns across time-points (Eisen et al., 1998) using Genesis (Sturn, Quackenbush, & Trajanoski, 2002).

Enzyme inhibition by FPP

The enzymatic inhibitory properties of FPP were investigated using different enzyme assays. Luciferase from the Enliten kit (Promega) was used to produce light in response to varying amounts of ATP. Light was measured on a Spectramax M2 (Molecular Devices) at 560

nm. Thirty-five l of reconstituted Luciferin reagent was added to 100 l reaction mixtures. The reaction mixtures contained water, ATP at various concentrations, and various potential inhibitors.

Results

Accumulation of FPP causes growth reduction, but only in certain media

Previous work in our lab showed that the isoprenyl diphosphate intermediates caused reductions in growth compared to an empty vector control and also to a strain that expressed an efficient enzyme that converted the intermediates to a volatile product that could diffuse through the cell membrane and escape the cell. This initial work was performed in Luria Broth (LB), a rich, undefined medium, with no glucose added, and repeated in Figure 3A. An undefined medium isn’t ideal for microarray analysis, so the strains were grown in a defined M9 glucose medium.




15

E. coli DH1 was transformed with the bottom half of the mevalonate pathway (pMBIS) and either a plasmid with an active amorphadiene synthase (pADS) or a plasmid with a catalytically inactive mutant (pADSmut). The resulting strains were grown in the defined, minimal medium with glucose and 20 mM mevalonate added to the medium. Unlike in the rich, undefined LB, the two strains grew equally well, with no growth reduction, Figure 3B. Upon further investigation, the differences in growth were not seen until well into stationary phase (after 30 hours). To assess the change in phenotype in the different media, the individual components of LB (peptone, yeast extract (YE), and NaCl), were added to the defined M9 glucose medium and glucose was added to LB, Figure 4.

A

B

Figure 3. Growth of strains accumulating FPP in different media A) LB and B) M9 MOPS 1% glucose. The cultures

were grown in 20 ml media with 20 mM mevalonate and 0.5 mM IPTG added.

16

The addition of glucose to LB had a pronounced impact. Without glucose, the strain accumulating FPP abruptly stopped growing and the OD decreased after 10 hours. When glucose was added, the growth was slightly slower compared to growth on LB alone, but the strains with and without FPP accumulation grew equally well. The strain accumulating FPP stopped growing just before the strain producing amorphadiene, thus the overall difference in OD was much smaller compared to the media without glucose. The addition of the LB components to M9 glucose medium had varying impacts. The addition of NaCl to the medium had no effect. However, the additions of yeast extract and peptone changed the growth patterns. Interestingly the addition of yeast extract resulted in the strain accumulating FPP reaching higher density (but this was probably due to a crack in the flask). The addition of peptone resulted in FPP causing a reduction in growth.

A

B

Figure 4. Addition of supplements to media leads to different growth when FPP accumulates. A) LB with 1% glucose added. B) M9 MOPS 1% glucose with the individual components of LB, 10 g/L peptone, 5 g/L yeast extract, 10 g/L NaCl.

17

Microarrays of FPP accumulation in different media

Given the difference in growth between the two strains depending on the media, microarrays were performed in LB and M9 MOPS medium to characterize the transcriptional response to FPP accumulation. Since the metabolic burden of producing heterologous proteins is well documented (Hoffmann & Rinas, 2004; Ow et al., 2006) and the production of amorphadiene is non-toxic, the full pathway producing amorphadiene was used as the control rather than an empty vector. The catalytically inactive ADSmut was also useful to accumulate FPP while maintaining equal protein production between the two strains. The strains with the mevalonate pathway (pMBIS) and an active ADS or inactive ADSmut were grown and samples collected throughout growth and into stationary phase in LB and M9 MOPS, Figure 5 and Figure 6, respectively. The RNA was isolated from the cell pellets and converted to cDNA, fluorescently labeled, and hybridized to microarrays synthesized in house. The data are stored in NCBI GEO under accession numbers GSE29267 and GSE30403, respectively.

Figure 5. Samples were collected throughout growth in LB to study the toxicity of FPP using microarrays. 10 mM mevalonate was used in LB.

18

Figure 6. Samples collected throughout growth in M9 glucose medium to study the toxicity of FPP using microarrays. 20 mM mevalonate was used.

Although it isn’t necessary to see big changes in growth to see changes in the transcription patterns, the transcription profile in response to FPP accumulation followed this pattern. The first time point in the LB study where the growth had not diverged had zero genes differentially regulated (|Z|>1.96, |log2ratio|>1), while at the subsequent time points 189, 100, and 147 genes were changing. This was evident in the scatter plots of gene expression, Figure 7. The gene expression in M9 at the points analyzed showed little difference in gene expression, Figure 8. Only 7 and 4 genes met the significantly changing criteria, Table 1, with no

apparent connection between the genes differentially expressed.

19

A

B

C

D

Figure 7. Expression profile during microarray time course in LB. The expression in pADSmut is plotted versus the expression in pADS. Each gene is represented by a single point.

20

A

B

Figure 8. Expression profile during microarray time course in M9 MOPS 1% glucose. The expression in pADSmut is plotted versus the expression in pADS. Each gene is represented by a single point.

Table 1. The differentially changing genes in the M9 microarrays. Significantly changing genes are bold with the Z-scores in parentheses.

Gene Function Time 1 Time 2

iscR (b2351)

Iron-sulfur cluster regulator -0.99 (-1.95) -1.50 (-2.52)

cspB Cold shock protein ND -1.35 (-2.51)

fimA Major type 1 subunit fimbrin -1.27 (-2.47)

-1.58 (-2.66)

fimI Fimbrial protein -1.22 (-2.35)

-1.20 (-1.82)

glcF Glycolate oxidase, predicted iron sulfur subunit 1.21 (2.01)

1.56 (1.82)

glmS L-glutamine:D-fructose-6-phosphate aminotransferase

-1.51 (-2.94)

-0.56 (-0.98)

osmE Osmotically inducible protein 1.24 (2.16)

0.72 (0.92)

tra5_4 Transposon element -1.14 (-2.20)

-0.34 (-0.54)

yehI Conserved protein -1.13 (-1.33)

-1.51 (-2.18)

yejB Component of oligopeptide ABC transporter -1.17 (-2.17)

-0.29 (-0.34)

However, in LB the main transcriptional response appear to fit into several categories: transcriptional machinery, response to stress and stationary phase, metabolism in LB, and the

21

PhoPQ regulon. The ribosomal proteins (rpl, rpm, and rps genes) were all downregulated at time point 2. This is a common response to a slowdown in growth rate (D.-E. Chang, Smalley, & Conway, 2002; Kurland & Henjiang Dong, 1996) and could be a result of the strain without FPP accumulation continuing to grow. To further this point, at the later time points these genes moved towards equal expression in each and were no longer differentially regulated. The ribosome modulation factor (rmf), rpsV, and yfiA are associated with ribosome stability in stationary phase and the set show increased expression at time point 2. YhbH, an additional protein associated with ribosome stability in stationary phase, was upregulated as well, but did not meet the significantly changing criteria (Izutsu, A. Wada, & C. Wada, 2001; Maki, Yoshida, & A. Wada, 2000). Along with the ribosomal proteins, genes associated with ATP synthesis (atpIBEFHADC), purine biosynthesis (purACDHLMRT), and the TCA cycle (gltA, sdhCDAB, sucABCD) were downregulated, all of which have been identified in the past with a slowing of growth.

The entry into stationary phase and general response to stress is controlled by the

alternative sigma factor, RpoS or S. Under normal growth conditions, 70 is dominant and S is quickly degraded by proteases (Lange & Hengge-Aronis, 1994). However, upon entry into

stationary phase or in response to stress, S accumulates due to many factors and regulates transcription of approximately 10% of all E. coli genes (Weber, Polen, Heuveling, Wendisch, & Hengge, 2005). The 140 genes identified in Weber et al. were plotted to show the expression pattern, Figure 9. The general trend is an upregulation at Time 2, followed by a decrease in the log ratio to the point where most genes are downregulated at Time 4. Genes associated with acid stress (gadABCWX), a regulator for cell membrane morphology in stationary phase (bolA), and osmotic stress (osmCY) are included on this list. Other genes previously identified as stationary phase genes, but not in the 140 genes in Weber et al., also follow this pattern; including genes associated with curli formation (csgABDEFG), more genes in acid stress (hdeAB and gadE), and the starvation lipoprotein (slp), among others.

Proteases and chaperones associated with the heat shock response, regulated by 32 (Bukau, 1993), are upregulated at time 2 and remain upregulated at time 4 (dnaJK, grpE, groELS, ibpAB and lon). Another operon that was strongly upregulated (the largest fold changers at time 3) was the phage shock protein operon (pspABCDE). The PSP operon responds

to a variety of stresses relating to a loss in the proton motive force, but it is not activated by 32

or S (Huvet et al., 2011; Weiner, Brissette, & Model, 1991). The proton motive force could be disrupted by damage to the membrane, which would also typically trigger oxidative stress (Farr & Kogoma, 1991). However, oxyR (one of the regulators along with soxS) is only upregulated at time 4, and other members of the regulon were not significantly changed.

22

Figure 9. The expression pattern for the 140 genes identified in Weber et al. as being regulated by RpoS. The red line is the

average of the 140 genes at each time point during the LB microarray time course.

Another response appeared to be the PhoPQ regulon. The PhoPQ regulon is a two-component system that regulates many genes in response to the magnesium levels in the medium (E. A. Groisman, 2001; PAPP-WALLACE & MAGUIRE, n d). Many genes in the regulon are significantly upregulated at time points 2 and 3. Of the 50 highest upregulated genes at time 3, four are members of the PhoPQ regulon (rstA, slyB, mgtA, and rstB) with ybjG and mgrB just outside this range. There is also a putative PhoP binding site in the promoter region of the transcription factor cysteine B (cysB). CysB regulates cysDNC, cysPUWAM, and cysJIH, all of which are upregulated at time point 3 except cysU, W, and M.

Catabolic genes had mixed responses, perhaps indicating the varying metabolism in the undefined LB. Genes associated with maltose transport and consumption (malK-lamB-malM,

malEFG) and mannose transport (manXYZ) were strongly downregulated. While fatty acid -oxidation (fadAB, fadL, fadIJ, maoC), phenylacetate degradation (paaABCDEFGHIJK), and peptide transporters (sapAB, dppABCD) were upregulated.

We also looked for changes in the genes in the native isoprenoid pathway to see if there was any transcriptional response to FPP accumulation, as well as changes in phosphatases or the phosphate regulon to indicate a response to the higher pyrophosphate levels. No genes in the methylerythritol phosphate pathway were differentially expressed (dxs, dxr, ispD, ispEFGH, idi); however, ispU, one of the enzymes that converts FPP to longer chain isoprenoids for the

23

formation of peptidoglycans, was downregulated, but not significantly (Z score 1.76<1.96). The other gene, ispB, was not differentially expressed. E. coli has several genes with undecaprenyl pyrophosphate phosphatase activity (bacA, lpxT, ybjG, and pgpB) (El Ghachi, Derbise, Bouhss, & Mengin-Lecreulx, 2005), and one of them, ybjG, was significantly upregulated at time 3. Interestingly, ybjG has a putative PhoP binding site in its promoter region (Eguchi et al., 2004) and was upregulated at the same time point as the other members of the PhoPQ regulon. Another protein with phosphatase activity, but without a known substrate (Kuznetsova et al., 2006), yedP, was upregulated at time 2. Finally, the PhoBR regulon, responsible for the response to phosphate starvation, was searched for differentially expressed genes, but only ugpB (but not the rest of its operon) and phoH were upregulated.

Varying [Mg2+] in media

The upregulation of the PhoPQ regulon during growth implied that the magnesium concentration in the LB was affecting the toxicity. Magnesium sulfate was present at 2 mM in M9 medium, whereas magnesium was not explicitly added to LB. Its concentration in the

tryptone and yeast extract has been approximated to be 50 M (PAPP-WALLACE & MAGUIRE, n d). However, the microarray was measuring the relative ratio of the gene expression between the strain containing the inactive ADS to the active ADS, thus the presence of FPP seemed to exacerbate the response. To test if more magnesium could alleviate the toxicity or delay as the glucose did, magnesium was added to the LB medium, Figure 10. The additional magnesium slightly increased the OD in the strain with ADS, but had a larger effect in the strain accumulating FPP. It didn’t remove the growth difference; in fact the growth rate decreased at the same point with or without magnesium. However, the big impact occurred in stationary phase. The magnesium stabilized the OD, reducing the large decrease in OD seen in LB.

The magnesium concentration was also reduced in the M9 medium from 2 mM Mg2+ to

10 M and samples were collected for microarray analysis, Figure 11. Unlike in Figures 1B and 4, where FPP accumulation caused minimal changes in growth, lowering the magnesium concentration had a significant effect. Similar to the growth in LB, with the lower magnesium concentration in M9, there was a drastic reduction in cell growth into stationary phase. The bigger changes in growth were also evident in the number of genes differentially expressed; 62, 186, and 126 genes were differentially expressed at the three time points (|Z|>1.96, |log2ratio|>1).

In general, the response in M9 was not very similar to the response in LB. This is highlighted by the ribosomal proteins which were significantly downregulated at time 2 in LB (then slightly upregulated at later time points), but in M9 they were upregulated at each time point, Figure 12A. Many of the stress response proteins seen upregulated in LB (dnaJK, ibpAB, oxyR, lon) were not changing in M9. The phage shock protein operon was significantly upregulated in LB, but downregulated in M9. Additionally, most of the genes profiled in Weber et al as stationary phase genes and plotted in Figure 9, did not change in M9. However, other genes associated with stationary phase, biofilm formation, and stress did change in M9, but not in LB (ycfR, yhcN, yhdA).

24

Unlike many of the genes above that did not show the same trends in LB and M9, the genes involved with acid resistance and the PhoPQ regulon showed similar trends in both conditions, Figure 12B and C, respectively. The acid resistance genes highlighted in Weber et al were upregulated at time 2 in LB, but downregulated at all other time points. The genes in the PhoPQ regulon (Minagawa et al., 2003) were upregulated at each time point, except for mgtA, phoP, and nagA, which were downregulated at time 2 and 3 in M9. YebO was included in Figure 12C, but had not been previously included in the PhoPQ regulon; however, its expression pattern was similar to the PhoPQ regulated gene upstream, mgrB (74 bases to the 3’ end). YbjG which has a PhoP binding site in its promoter, is also upregulated in M9, as it was in LB.

Figure 10. The effect of Mg2+

on FPP toxicity in LB

Figure 11. The effect of Mg2+

on FPP toxicity in M9 MOPS 1% glucose medium. Samples were collected at the indicated time points for microarray analysis.

25

Figure 12. The microarray results of FPP toxicity in LB and M9 with low magnesium conditions. The highlighted

sections are A) ribosomal proteins, B) acid resistance, and C) PhoPQ regulon

ATP impact on growth

Given the changes in growth with the media additions and the results of the microarray study, a hypothesis was formed about the cause of the FPP toxicity. While growth on an undefined LB medium and a defined minimal medium was very different, the addition of glucose to the medium appeared to be the main factor in removing the toxicity. With this in mind, we hypothesized that the addition of glucose greatly increased the levels of ATP in the cell due to the activity of glycolysis and the TCA cycle, and the ratio of FPP to ATP was the key

26

factor in the amount of growth reduction. We next wanted to test if the toxicity would be present when fed glucose if we reduced the cell’s ability to make ATP. Two of the components of ATP synthase, atpEF, were knocked out and the strains with the mevalonate pathway and ADS or ADSmut were grown in media with and without glucose. The atpEF knockouts were

made in two strains, wildtype DH1 and DH1 ackA, pta, poxB (DP), however, the ATP knockouts

in DH1 did not reliably grow, so the comparisons were between DP and DPA (DP atpEF).

When grown in LB with or without 1% glucose, Figure 13A, the toxicity was present in DH1 and DP in the LB media, but not in LB + glucose (similar to above). However, in DPA, the toxicity was present in both conditions. Similarly in the defined M9 MOPS medium, the amount of glucose present had no impact on toxicity in DH1 and DP, but in the ATP knockout strain the toxicity again was present, Figure 13B. This pattern repeats with growth in shake flasks, Figure 14.

Enzyme inhibition by FPP

Beyond, the role of ATP and glucose, the actual cause of reduction in growth due to FPP accumulation was still not known. However, the importance of Mg2+ to the growth in LB and M9, in addition to the upregulation of the PhoPQ regulon, added more weight to the interaction between FPP and ATP. Mg2+ is the common cofactor for ATP; the divalent cation stabilizes the negative phosphate charges (Maguire & J. a Cowan, 2002). The pyrophosphate bond of FPP also binds Mg2+ and the divalent cation is a common cofactor for many enzymes (J. A. Cowan, 2002), including isoprenyl diphosphate using enzymes (Wouters et al., 2003). Thus, it seemed likely that the amount of FPP accumulating caused toxicity when the ATP levels were low because FPP was effectively chelating Mg2+ and thus competing for cellular enzymes that require Mg2+-ATP to function.

To test if FPP would inhibit enzymes, luciferase was used as a test enzyme which required ATP for activity. Luciferase is the enzyme purified from firefly that oxidizes luciferin with O2 and ATP to produce oxyluciferin, AMP, pyrophosphate, CO2 and light (DeLuca & McElroy, 1978). The addition of increasing amounts of FPP to the enzyme reaction significantly decreased light output, while the equivalent amount of FPP’s buffer, 70:30 methanol/100 mM NH4OH, had no impact, Figure 15A. The amount of FPP added was important too. The lowest

level of FPP, 0.1 M, had a negligible impact, but 10 and 100 M had increasing levels of inhibition.

27

A

B

Figure 13. Impact of atpEF knockouts on FPP toxicity with varying glucose concentrations in LB (A) and M9 MOPS (B).

28

A

B

Figure 14. Growth of the wildtype DH1 (A) and atpEF knockouts (B) in M9 MOPS 1% glucose media.

The luciferase kit had a smaller linear range for ATP than the vendor’s literature provided, so the reaction was repeated within the linear range. Additionally, other isoprenyl diphosphate molecules were added to test if FPP’s impact was unique, Figure 15B. GPP (10 carbons) and DMAPP (5 carbons) were added to the mixture at the same concentration as FPP

(100 M), but had no effect. Farnesol (the dephosphorylated alcohol of FPP), had a significant impact, but much smaller than FPP. Finally, more magnesium was added to the reaction mixture with FPP to test if the extra magnesium could buffer the FPP and remove the inhibition. However, magnesium up to 1 mM had no impact at these concentrations of ATP and FPP. Thus it seemed that the size of the molecule had an important impact on the inhibition (as the smaller GPP and DMAPP had no effect, while the 15-carbon farnesol did), but the pyrophosphate group of FPP also was important.

The potential inhibition of enzymes that use ATP was observed with the fire fly luciferase enzyme; however, to test E. coli metabolic enzymes for this inhibition, several enzymes were tested for their inhibition. Enzymes were chosen that used ATP and Mg2+ as cofactors, just used Mg2+, and that didn’t use either to see if there were different patterns of inhibition based on cofactor usage. The enzymes were purchased from Sigma and coupled with a pyruvate kinase/lactate dehydrogenase assay to monitor the reaction by following NADPH consumption using a plate reader. However, these assays have yet to be completed.

30

B

Figure 15. Enzyme inhibition of luciferase by FPP. Two separate experiments showing representative results.

Discussion

The accumulation of FPP caused different growth behaviors depending on the media. When grown in LB compared to LB + 1% glucose, the strain accumulating FPP stopped growth much earlier and the difference between the strain producing amorphadiene was large, Figure 4. LB is a carbohydrate-poor medium; the cell is forced to grow on the peptides present in peptone and yeast extract as its primary carbon source (Sezonov, Joseleau-Petit, & D'Ari, 2007; C. H. Wang & Koch, 1978). As a result of the large consumption of amino acids and peptides, the pH of the spent medium after growth in LB increases. Adding glucose to LB would remove the need to consume amino acids as carbon sources, thus resulting in normal trends in pH (where it decreases due to E. coli’s preference for excreting organic acids with excess glucose). To rule out the pH effect on toxicity, the strains could be tested in buffered LB.

Alternatively, the accumulation of FPP had no impact on growth in M9 MOPS + 1% glucose with appropriate levels of magnesium, thus it appears that the addition of glucose is the most important factor in reducing the toxicity, Figure 3. However, reducing the magnesium in the buffered medium with glucose resulted in the toxicity phenotype, Figure 11. In addition to LB being a carbohydrate-poor medium, it is estimated that the concentration of divalent cations can also be limiting. The concentration of Mg2+ in LB has been approximated to be

between 50 and 200 M(PAPP-WALLACE & MAGUIRE, n d; Wee & Wilkinson, 1988). The addition of magnesium to the LB medium did not remove the toxicity, but it did reduce the growth differences in stationary phase, Figure 10.

31

The improved growth in media containing glucose and the impact of magnesium also supports the role of ATP in the toxicity. When grown in strains with the ATP synthase knocked out, the toxicity is present independent of glucose addition. Furthermore, FPP was shown to inhibit enzymes that use ATP, previously shown for the pathway enzyme mevalonate kinase (Hinson, Chambliss, Toth, Tanaka, & Gibson, 1997) and unpublished data by D. Garcia (personal communication). Thus, the model of toxicity appears to be that when ATP levels are low, the increasing amount of FPP interferes with ATP and Mg2+ broad usage in the cell. However, when ATP is plentiful the amount of FPP is non-inhibitory until Mg2+ becomes limiting. The conditions for FPP toxicity are summarized in Figure 16.

While the enzyme assays showed that FPP can inhibit enzymes that use ATP, the exact mechanism of the reduction in growth is not known. The microarrays in general gave insight into the downstream effects of the reduction in growth. The primary response appeared to be

activation of the alternative sigma factor, S. Sigma S is the holoenzyme responsible for RNA transcription into stationary phase and in response to many stresses (Weber et al., 2005); however the transcriptional pattern did not match up particularly well in all conditions.

Given magnesium’s role in impacting the toxicity of FPP, it was reassuring to observe the upregulation of genes in the PhoPQ regulon. The PhoPQ regulon is E. coli’s two-component system responsible for maintaining magnesium homeostasis in the cell (E. A. Groisman, 2001). PhoP is the cytosolic response regulator and PhoQ is the inner membrane bound sensor kinase. Magnesium is an important component of the cell, stabilizing the ribosomes and cell membrane. E. coli has a well documented response of changes to the cell membrane and degradation of ribosomes to magnesium starvation (Fiil & Branton, 1969; Lusk, Williams, & Kennedy, 1968; MCCARTHY, 1962; St John & Goldberg, 1980; Zundel, Basturea, & Deutscher, 2009) and it has been implicated in increased ethanol toxicity in yeast due to an increase of membrane fluidity (Dombek & Ingram, 1986). Thus it seems possible that the FPP toxicity could be a result of these effects if enough FPP accumulates to reduce the Mg2+ available for these roles. Unfortunately, the microarrays don’t provide insight into what is actually causing the increased transcription from the regulon; it isn’t clear whether the triggering of the PhoPQ response is a result of direct binding of Mg2+ by FPP or a result of changes in the cellular membrane triggered by the stress response. Another consistent response was the downregulation of the acid resistance genes. Recently connections have been identified between the PhoPQ regulon and acid resistance (via binding to their promoters), envelope stress (via the small RNA, micA), and rpoS (via a target of PhoP, ycgW, which helps stabilize rpoS) (Bougdour, Cunning, Baptiste, Elliott, & Gottesman, 2008; Coornaert et al., 2010; Zwir et al., 2005).

While there could be metabolic engineering roles in reducing FPP toxicity, the easiest appear to be adding glucose to the medium or expressing a functional terpene synthase to prevent the accumulation. Thus, the knowledge gained from the microarray experiments were used to engineer regulation into the heterologous pathway in the following chapters.

32

Figure 16. A cartoon summarizing the conditions where FPP toxicity occurs. Low magnesium/Low ATP conditions are the most toxic, while the addition of either ATP/glucose or Mg

2+ offsets or delays the toxicity. When both Mg

2+

and glucose are present in adequate quantities, FPP toxicity is not apparent.

33

Chapter 3: Creating a FPP sensor and control of the downstream enzyme




Promoter selection ............................................................................................................................. 35

Cloning/biobricking ADS ..................................................................................................................... 35

Biobrick strategy ............................................................................................................................. 35

Construction of reporters ............................................................................................................... 35

Construction of ADS constructs ...................................................................................................... 36

Cloning of PhoPQ ............................................................................................................................ 36


qPCR, Proteomics, and Metabolites ................................................................................................... 37

qPCR ................................................................................................................................................ 37

Proteomics ...................................................................................................................................... 37

Metabolite Analysis......................................................................................................................... 38

Results ..................................................................................................................................................... 38

PhoPQ promoters respond to Mg2+ .................................................................................................... 38

PrstA responds to FPP levels ................................................................................................................. 39

Expression of ADS with PrstA ................................................................................................................ 43

Expanding the promoters used from the microarray ......................................................................... 46

Discussion................................................................................................................................................ 48

Introduction

The previous chapter highlighted the problems associated with FPP accumulation and targeted several key responses the cell makes to mitigate FPP toxicity. The next step was to use this knowledge to create a pathway that responds to FPP accumulation to regulate enzyme expression. This could have several advantages over a standard inducible pathway. Theoretically, a standard inducible expression system could be tuned to give an optimal protein output that maximizes production, but does not harm growth by excessive protein burden. This isn’t as easy as it sounds as often promoters don’t have a linear output to added inducer or the expression is not homogeneous throughout the population (S. K. Lee & Jay D Keasling, 2005). Alternatively, the promoter could be set at a given output using a designed promoter library (Alper et al., 2005; Hammer et al., 2006) and the individual enzyme protein levels could

34

be adjusted by affecting the mRNA stability or translation rates (Pfleger, Pitera, Christina D Smolke, & Jay D Keasling, 2006; H. M. Salis, Mirsky, & C. A. Voigt, 2009; C D Smolke, V. J. Martin, & J D Keasling, 2001). However, these solutions are still static in their response to pathway flux; if every cell could respond to its internal FPP concentrations dynamically, each cell could perform at its optimal level.

The first thing we needed was a sensor for the accumulating intermediate. But without a native sensor available, the plan was to use the transcriptional response to toxicity to highlight promoters that responded to the accumulation. This could potentially have detrimental effects for the expression of the downstream enzyme if the response is an indirect result of the stress. This could result in a delay of expression, and potentially cause growth reduction due to increased levels of intermediate before the downstream enzyme could consume them. Additionally, an indirect response could lead to other stresses or conditions triggering the response, reducing the robustness of the pathway and potentially lowering productivity.

Several other methods have been tried to design sensors for intracellular metabolites. These primarily consist of engineering RNA aptamers or protein engineering ligand binding sites (Looger et al., 2003; Shamah, Healy, & Cload, 2008; Wieland, Benz, Klauser, & Hartig, 2009). While there has been some success in this field, typically the engineered devices have worked best when coupled with their native ligand (Nomura & Yokobayashi, 2007; Win & Christina D Smolke, 2007). Thus, for a molecule like FPP without a known sensor, alternative, yet complementary, methods are needed.

Based on the response to the toxicity of the PhoPQ regulon and the potential link to the toxicity, these promoters were chosen to act as the sensing and actuating devices. The PhoPQ regulon is well characterized and has a direct method of testing the functionality of devices built using the promoters (by varying the magnesium concentration in the medium). The key residues in PhoP and Q are also understood ( a Kato, Tanabe, & R Utsumi, 1999; Sauer, 1996), as are the binding sites in the promoters (Minagawa et al., 2003), so further engineering of the response was feasible. Furthermore, using a two component system gives a natural link between the sensing moiety and the regulation of gene expression. Given that, several promoters in the regulon were chosen and analyzed for their response to magnesium to make sure a functional sensor was created, then tested for their response to FPP. Finally, the promoters were used to drive amorphadiene synthase in the first attempt to make a responsive, inducer free mevalonate pathway.


Strains and Growth

The strains and plasmids used in this study are described in the appendix in Table 2. The mevalonate pathway from Saccharomyces cerevisiae, cloned into two separate plasmids, pMevT and pMBIS, was used along with the codon optimized amorphadiene synthase, pADS, from Artemisia annua (V. J. J. Martin, Pitera, Withers, J. D. Newman, & Jay D Keasling, 2003). To accumulate farnesyl diphosphate (FPP), an inactive version of ADS was created by mutating

35

three catalytic aspartate residues in the first aspartate rich motif (A. Chen, Kroon, & C D Poulter, 1994).

Medium components and chemicals were purchased from Sigma-Aldrich, EMD chemicals, Difco, and TekNova. Mevalonolactone (Sigma) was converted to mevalonate by adding 1.02 volumes 2 M KOH to 1 volume 2 M mevalonolactone in water. The mixture was incubated at 37 °C for 30 minutes, and then filter sterilized. Luria Broth Miller (LB) (EMD), EZ rich defined medium (RDM) (TekNova), and a modified M9 MOPS medium were used for cell growth. Magnesium sulfate and glucose were added to the media at varying levels as indicated. The modified M9 MOPS medium consisted of 1x M9 salts (Difco) (1 g/L NH4Cl, 0.5

g/L NaCl, 6.6 g/L Na2HPO4 (anhydrous), 3 g/L KH2PO4), 2 mM MgSO4, 100 M CaCl2, 40 mM MOPS, 1x trace metal containing (3x10-6 mM (NH4)6(MO7)24, 4x10-4 mM HBO3, 3x10-5 mM CoCl2,

10-5 mM CuSO4, 8x10-5 mM MnCl2, 10-5 mM ZnSO4), 4.5 g/ml thiamine, 0.01 mM FeSO4

(heptahydrate), and glucose (Neidhardt et al., 1974). Chloramphenicol (35 g/ml), Tetracycline

(10 g/ml), carbenicilin (100 g/ml), and kanamycin (50 g/ml) were added as appropriate. Cells were grown at 30 °C to improve production.

Promoter selection

The genes were selected based on the previous microarray results. The intergenic region upstream of each gene was analyzed for known transcription factor binding sites based on www.ecocyc.org and literature searches. The region was chosen to include all binding sites and in cases where no known binding site existed, the region extended into the 3’ end of the upstream gene. The 3’ end of the promoter region was picked downstream of the transcription start site. The nucleotide sequences for each promoter are included in the Appendix in Table 3. Promoters were cloned from E. coli DH1 genomic DNA with forward primers containing EcoRI and BglII restriction enzyme sites in the forward primer and BamHI site on the reverse primer. The cloned promoter regions were verified by DNA sequencing.

Cloning/biobricking ADS

Biobrick strategy

Most cloning reactions were performed using the BglBrick standard (Anderson et al., 2010). This strategy used the four restriction enzyme sites EcoRI and BglII 5’ of the part and BamHI and XhoI 3’. BglII and BamHI create compatible sticky ends which when ligated together create a scar site that neither enzyme recognizes. This allows the assembly of multiple parts in a sequential manner. New England Biolabs and Fermentas restriction enzymes and reagents were used. High fidelity Phusion (Finnzymes) was used for PCR reactions off of genomic DNA.

Construction of reporters

The first four promoter reporters used to monitor the cell’s response to magnesium were a gift from Weston Whitaker. They were designed based on the promoter probe vectors, pProbe (W. G. Miller, Leveau, & Lindow, 2000). The reporter plasmid used for all other tests was based on a modular vector designed in the lab called pBbA5k-RFP, where Bb stood for BglBrick, A stood for p15a origin (E stood for ColE1), 5 stood for lacUV5 promoter (1 stood for trc), and k stood for kanamycin resistant (a stood for ampicillin and c stood for


36

chloramphenicol). The mRFP gene was bracketed by the BglBrick restriction enzyme sites and was 3’ of the lacUV5 promoter, Figure 43.

To remove the lacUV5 promoter and LacIQ, the plasmid was sequentially digested with AatII and EcoRI at 37 °C for 1 hour each. The sticky ends were not compatible, so T4 DNA

polymerase was used to blunt the ends in a 30 L reaction with NEB Buffer 2, 100 M dNTPs, 100 ng digested plasmid, and 1 unit T4 DNA polymerase, for 15 minutes at 12 °C. The reaction was stopped by adding 10 mM EDTA and heating 75 °C for 20 minutes. The reaction mixture was ligated with T4 DNA ligase and transformed. The selected promoters were PCR’d from genomic DNA and the PCR products were doubly digested with EcoRI/BamHI. The plasmid backbone was doubly digested with EcoRI/BglII and the two products were ligated together with T4 DNA ligase. The resulting plasmid without promoter/repressor was called, pBbA0k. A representative plasmid map is in the Appendix, Figure 44. A similar procedure was used to create an empty backbone for expression of amorphadiene synthase on a ColE1, ampicillin resistant backbone, pBbE0a.

Construction of ADS constructs

Before cloning ADS with the BglBrick strategy, three internal EcoRI and one BglII sites were removed by PCR SOEing (splicing by overlapping extension). Forward and reverse primers were designed to mutate each site. The gene was PCR’d in five separate reactions and the products pooled and PCR’d with the flanking primers (which contained the restriction enzyme sites and the ribosome binding site, listed in Table 4). The full length PCR product was verified by sequencing and doubly digested with EcoRI/BamHI. The RFP was removed from pBbE0a-RFP by similar digestion and the ADS fragment was ligated into the vector. The resulting plasmid was doubly digested with EcoRI/BglII and promoters described above were ligated into the site.

The ribosome binding site (RBS) calculator, http://voigtlab.ucsf.edu/software/ (H. M. Salis, Mirsky, & C. A. Voigt, 2009) was used to design RBS with larger translation initiation rates. The new RBS were designed into the 5’ primer and the entire gene was PCR’d with the flanking primers. The resulting gene was cloned into the biobrick vector as above. The gene was also cloned into ptrc99a by doubly digesting the plasmid with EcoRI/BamHI.

Cloning of PhoPQ

PhoP and PhoPQ were cloned from genomic DNA with primers to include the native promoter. However, the EcoRI and BglII restriction enzyme sites were included on the reverse primer (and the BamHI on the forward primer) so the gene or operon could be cloned into the EcoRI/BglII site in the reverse direction of the rstA promoter.

Fluorescence and Amorphadiene (and other sesquiterpene) production

The sensor plasmids were transformed with pMBIS and pADS or pADSmut, while the amorphadiene producing constructs were transformed with pMevT and pMBIS. Overnight cultures were picked form single colonies and used to inoculate 5 ml fresh media (1%

inoculation). Mevalonate and IPTG (500 M) were added as indicated. Dodecane (20%) was added as an overlay for amorphadiene production. Cell growth (OD) was measured at 600 nm in a Beckman spectrophotometer. Fluorescence was measured in a safire plate reader (TECAN).

http://voigtlab.ucsf.edu/software/

37

GFP was measured at 400 nm/510 nm (excitation/emission) and RFP was measured at 558/583

nm (excitation/emission). Amorphadiene was measured by diluting 5 L of the dodecane layer

into 495 L of ethyl acetate spiked with the internal standard trans-caryophyllene. The samples were run on an Agilent GC-MS as previously described (V. J. J. Martin, Pitera, Withers, J. D. Newman, & Jay D Keasling, 2003).

qPCR, Proteomics, and Metabolites

pMevT, pMBIS, and either PrstA-ADS, pADS, or pADSmut were transformed and single colonies picked into EZ RDM + 1% glucose. One percent of the overnight was used to inoculate 50 ml of media in 250 ml baffled flasks. 5 ml of dodecane was added as an overlay. The strains

with pADS and pADSmut were induced with 500 M IPTG at an OD ~0.3. Samples were collected during growth for qPCR, amorphadiene production, protein, and metabolite analysis.

qPCR

To measure the mRNA levels of pathway enzymes, one ml samples of culture were spun at max speed for 1 minute in a bench top centrifuge and the pellets snap frozen in liquid nitrogen. The cell pellets were extracted with the Qiagen RNEasy mini kit including the on column DNase digestion following manufacturer’s instructions. Following RNA isolation, an

additional DNase treatment was used to reduce background amplification. Ten g RNA was

treated in a 50 L reaction with 2 L of Turbo DNase (Ambion) at 37 °C for 1 hour. The treated RNA was then reverse transcribed using randomly primed reactions and Superscript III (Invitrogen) following manufacturer’s instructions. Following reverse transcription, the remaining RNA was removed by RNase H treatment at 37 °C for 20 minutes. The reverse

transcription products were diluted 1:10 and 1 L cDNA was used in qPCR reactions.

In 25 L reactions, 1x Dynamo HS SYBR green master mix (Finnzymes), 500 nM primers, 1x ROX, and H2O were added to the diluted cDNA. Samples were run in triplicate on a Step One Plus (Applied Biosystems) qPCR machine. After a 15 min hot start, 40 cycles of 95 °C for 15 sec, 52 °C for 30 sec, and 72 °C for 30 sec, were followed by 72 °C for 10 min and a melt curve cycle. The primers for ADS and arcA as an internal control, cgactaccgaagaacatgac/cagcccattcaacagattcc and gctgaactgctgaagaaaatg/cggatcgtcacgtctacagt respectively, amplified ~150 nucleotides. The amplification thresholds were calculated using

the Step One software. The log2 ratios were calculated using the 2-Ct formula.

Proteomics

Samples were collected 28 hours after inoculation for targeted proteomics via selected reaction monitoring (SRM) as described previously (Redding-Johanson et al., 2011). Briefly, 20 OD*ml cell pellets were frozen in liquid nitrogen and then extracted with chloroform/methanol and dried in a speed vac. The dried protein extract was resuspended in 50 mM ammonium bicarbonate in 10% acetonitrile. After quantification with a Bradford assay, 1 mg/ml of protein was reduced in 5 mM tris(2-carboxyethyl)phosphine (TCEP) at room temperature for 30 minutes. The proteins were alkylated with 200 mM iodoacetic acid in 100 mM NaOH for 30

minutes in the dark. The proteins were digested with Trypsin (1 g/l) at 37 °C overnight. After addition of pure bovine serum albumin digests as an internal standard, samples were

38

analyzed on a LC/MS-MS (Eksigent TEMPO nanoLC-2D coupled to an Applied Biosystems 4000Q-Trap mass spectrometer) as described previously.

Metabolite Analysis

Samples were collected (20 ml) 28 hours after inoculation, pelleted, and snap frozen in liquid nitrogen. The pellet was extracted twice with two ml methanol, followed by two washes

with two ml water. The supernatant was lyophilized and resuspended in 500 L 50:50 methanol:water. The extract was passed through a 3000 molecular weight cut off filter and analyzed. The samples were run on an Agilent LC connected to a time of flight (TOF) mass spectrometer using methods described previously.

Results

PhoPQ promoters respond to Mg2+

Several PhoPQ promoters were chosen based on the microarray response to FPP toxicity. These promoters were cloned in front of a fluorescent reporter and their function was tested by varying the magnesium concentration in a M9 medium. The concentration was varied over 4 orders of magnitude and the fluorescent output was measured. The functionality of the reporters was confirmed as they each increased expression as magnesium was lowered, Figure 17. The threshold confirmed previous literature reports that PhoPQ activates in micromolar magnesium conditions (E. A. Groisman, 2001). The different promoters gave different responses however. rstA had the biggest on state and a low off state, while others had smaller activations, and yrbL had high expression in repressing millimolar magnesium concentrations. The different responses could be indicative of the differential binding strengths of PhoP in the promoter region or that they need different responses for different roles (Minagawa et al., 2003). The different responses in the millimolar region could also be a result of other transcription factors initiating transcription. Based on these responses and the fact that rstA had the largest fold change of the PhoPQ regulon in the microarray study, it was chosen to test for responsiveness to FPP.

39

Figure 17. The response of the PhoPQ regulon promoters driving a fluorescent reporter with varying magnesium in the M9 medium. RFU is relative fluorescent units, fluorescence per OD.

PrstA responds to FPP levels

The bottom half of the pathway, pMBIS, was transformed with the active or inactive ADS used in the previous chapter, and the PrstA sensor to test for responsiveness to FPP. Exogenous mevalonate was added to the media and the fluorescent output was measured after a day of growth. The tests were performed in LB and defined M9 medium to rule out growth differences impacting the final fluorescent reading and to test its versatility to work in a variety of media. The growth in the different media was similar to previous results where the strain accumulating FPP had much lower growth in LB, but the growth difference was much smaller in M9 + glucose. In both cases, the strain with the inactive ADS had higher fluorescent outputs (indicative of the higher response in the microarray), Figure 18. To test if PrstA could detect increasing levels of FPP, mevalonate was varied and the output increased when the pathway was induced, Figure 19. The output saturated above 5 mM mevalonate, which could be indicative of diffusion problems across the membrane or saturation/inhibition of the mevalonate pathway enzymes, a problem demonstrated with the human mevalonate kinase (Hinson et al., 1997).

40

A

B

Figure 18. A) The growth of the strains accumulating FPP with the PrstA sensor in LB and M9 MOPS medium and B)the fluorescent output of the PrstA sensor. 10 mM mevalonate was added to the media.

To further test for its response to FPP, strains with different pathway enzymes were used. First, we added extra copies of a limiting enzyme that was shown to improve amorphadiene production. Previous work in our lab found that an extra copy of mevalonate kinase (MK) improved amorphadiene production (J. R. Anthony et al., 2009). The extra copy of MK was added to the plasmid with the inactive ADS to give a strain that accumulated more FPP. When transformed with the fluorescent reporter, the strains accumulating FPP had higher fluorescence than the strains with the active ADS as above, and as expected the strain with the extra copy of MK had the highest output, exceeding the output of the strain without the extra copy, Figure 20.

41

A

B

Figure 19. A) The growth of the FPP accumulating strain with the PrstA sensor in M9 MOPS medium with increasing amounts of mevalonate added to the medium. B) The relative fluorescent output of the sensor. Solid bars are induced with IPTG, while shaded bars are uninduced.

Finally, the sensor was tested for its ability to detect varying amounts of FPP when different terpene synthases were expressed. Terpene synthases are known to have expression problems in E. coli and thus will use FPP at varying rates (V. J. Martin et al., 2001). Variants of synthases to make amorphadiene, bisabolene, vetispiradiene, and farnesene were used. The PrstA sensor, the bottom half of the pathway, and these terpene synthases were transformed and fed 5 mM mevalonate. The amount of sesquiterpene product produced was measured as well as the fluorescent output after 48 hours. The fluorescence output is relative to pADSmut and the slope of the trend is negative because FPP was consumed with more terpene production, Figure 21. Based on this response, the sensor could detect the varying outputs from the different terpene synthases, however some of the error bars were a little large. Also, some of the active terpene synthases produced very little product, yet registered decreased fluorescence. While the general performance is ok, the sensor was not used as a screening tool without further improvements to increase the dynamic range, as currently it is only about two, and reduce the variability in the response.

42

A

B

Figure 20. A) The growth of the strains accumulating FPP with the PrstA sensor. Strains with the extra copy of MK were shown to improve amorphadiene production. B) The fluorescent output of the sensor with and without the extra copy of MK. 10 mM mevalonate was added to the media.

Figure 21. The effect of using the PrstA sensor with varying terpene synthases. The cultures were fed 5 mM mevalonate and production and fluorescence was measured after 48 hours. Each spot represents a different variant of a terpene synthase. The fluorescence is relative to pADSmut. Black is pADS, Yellow are pADS variants, Blue are pBisabolene variants, Brown is pVetispiradiene, and Red is pFarnesene.

43

Expression of ADS with PrstA

The rstA promoter showed it could detect increasing amounts of FPP in the previous experiments, both in LB and M9 media. The next step was to see if the promoter could drive amorphadiene synthase and regulate enzyme expression in response to the internal FPP conditions. This would allow the cell to regulate the downstream enzyme and consume the toxic FPP, and do so without adding an inducer. Amorphadiene synthase was cloned behind the rstA promoter on a ColE1 plasmid without LacIQ or Ptrc, PrstA-ADS. When transformed with pMevT and pMBIS, production occurred without IPTG addition because neither of the mevalonate pathway plasmids contained LacIQ. However, when compared to production from pADS (which used the Ptrc promoter), the responsive promoter typically produced between 30 and 70% less (without the addition of IPTG). The strain would also often lag and grow slower than the fully inducible pathway.

To investigate the differences in growth and production, pADS, pADSmut, and PrstA-ADS were transformed with pMevT and pMBIS and grown in EZ RDM 1% glucose (pADS and pADSmut containing strains induced, while PrstA-ADS was not) and sampled for intracellular metabolites, amorphadiene, RNA and protein levels. The growth and amorphadiene production of the two strains containing the active ADS were similar to previous experiments, Figure 22A and B. The mRNA levels of the pathway enzymes were measured by qPCR, Figure 22C and D. The responsive rstA promoter was consistently lower than the trc promoter throughout growth, despite all increasing over time. It was about 25 lower, or about 32 fold, which could easily explain the slower growth and lower production if FPP was allowed to accumulate. The protein levels of the pathway enzymes were measured at 28 hours by tandem LC-MS/MS with multiple reaction monitoring (MRM), Figure 22E. As expected by the qPCR results, the responsive PrstA had much lower ADS protein levels, ~4-5 fold lower. Interestingly, the rest of the pathway enzymes were slightly higher. This could be a result of the responsive pathway not having a LacIQ, and thus the amount of IPTG added to the inducible strains was not enough to fully derepress the pathway enzymes. With lower pADS levels, but higher pathway enzyme levels, it was not surprising the pathway metabolites in the bottom half of the pathway accumulated compared to the other strains, Figure 22F. Most noticeably, the mevalonate levels were much higher than the isoprenyl diphosphate intermediates. In fact, the isoprenyl diphosphate intermediates were just above the level of detection of the instrument. Surprisingly, the strain with the inactive ADS, only accumulated FPP to 9 nM, two-fold higher than PrstA-ADS, while the active ADS accumulated more IPP.

With expression from PrstA much lower than Ptrc, we tried to improve expression by improving the transcription and translation levels from the promoter. We had been relying on the chromosomal copies of the two component system to initiate transcription, thus the extra copies of the promoter could have been titrating out phosphorylated PhoP and preventing full transcription. PhoP and PhoPQ were cloned with their native promoter divergent of PrstA-ADS (and PrstA-RFP to monitor the sensor’s response). The different sensor constructs were transformed with pMBIS and pADS or pADSmut and fed mevalonate. The fluorescence was measured after 24 hours of growth, Figure 23A. The addition of PhoP increased the overall output, but worsened the response to FPP by increasing the level in the strain producing

44

A

B

C

D

E

F

Figure 22. The growth and production of the responsive PrstA-ADS compared to pADS and pADSmut in EZ RDM 1% glucose. pADS and pADSmut were induced with IPTG, while PrstA-ADS was not. A) The growth in shake flasks. B)Amorphadiene production measured as a ratio of peak areas to the internal standard caryophyllene at time points in stationary phase. C)mRNA levels of ADS in each strain, relative to pADS. D)mRNA levels of ADS in each strain, relative to preinduction. E)Measurement of pathway enzyme levels at 28 hr by multiple reaction monitoring LC-MS/MS. F)Measurement of intracellular pathway intermediates at 28 hr by LC-TOF.

45

amorphadiene as well. The addition of the complete PhoPQ operon returned the output to the same level as without the extra copies and the response to FPP stayed the same.

The higher output in the sensor with extra copies of only PhoP was not observed when the strain was producing amorphadiene, Figure 23B. PhoPQ had no impact as well, in fact both constructs had lower production than the original PrstA. Thus, this strategy for improving the output did not work and PrstA remained below pADS and the trc promoter.

The next strategy to improve production from the PrstA was to improve the strength of the ribosome binding site (RBS) with the hopes that it would produce more protein with the lower transcript levels. This was accomplished using the ribosome binding site calculator (H. M. Salis, Mirsky, & C. A. Voigt, 2009). First, the strength of the RBS used in the PrstA-ADS construct was calculated using the reverse engineering tool. Similarly, the strength of pADS was also calculated as a comparison. Interestingly, the PrstA RBS registered at 7713, while the pADS registered 180. Next, the forward engineering tool was used to design a RBS with a translation rate of 50,000. Five designs were chosen because of the success rate quoted in the paper (the list of designed ribosome binding sites is listed in Table 4). The modified RBS was cloned in front of ADS and the constructs were inserted behind the two promoters (rstA and trc). The eight constructs were transformed with pMevT and pMBIS and the amorphadiene was measured after 24 hours of growth, Figure 24. This course of action did not improve production either. Both PrstA and Ptrc with the stronger RBS produced the same amount as their counterpart without the increased strength RBS (only one construct of each is shown as a representative sample).

Expanding the promoters used from the microarray

The efforts to improve the performance of PrstA by increasing the amount of transcription and translation were unsuccessful. The responsive promoter still produced less than pADS so a different approach was taken. The FPP toxicity profile was used to pick out alternative promoters beyond the PhoPQ regulon. Thirty four other promoters were chosen to have a variety of outputs and expression profiles. Their responses in the LB microarray time course are hierarchically clustered below, Figure 25A. The promoters were cloned 5’ of the fluorescent reporter and transformed with pMBIS and pADS or pADSmut. The strains were fed 5 mM mevalonate and the fluorescence was measured in each strain after 24 hr, Figure 25B. In general, there was fairly good agreement between the sensors and the microarray, although the responses were typically much smaller. The heat maps also fail to highlight the relative strength of each promoter, apart from the samples without a signal (noted in grey). For instance, pspA registered one of the largest log ratios in the sensor, but this was because the overall expression was low and thus the fold change was bigger.

46

A

B

Figure 23. The effect of overexpressing PhoP and PhoQ. A)The response to the accumulation of FPP. The different sensors were transformed with pMBIS and pADS or pADSmut and fed mevalonate. Fluorescence was measured after a day of growth. B)The different constructs were transformed with pMevT and pMBIS and amorphadiene was measured after a day of growth.

Figure 24. The effect of increasing the ribosome binding site for ADS. Amorphadiene was measured after a day of growth.

47

Figure 25. The 34 additional promoters chosen plus rstA represented in a heatmap of their A)log ratio expression (ADSmut/ADS). The promoter responses are hierarchically clustered. B)The response of the promoters, log ratio of fluorescent output (ADSmut/ADS), when cloned in front of the fluorescent reporter and tested with FPP accumulation (mevalonate fed with pMBIS and pADS or pADSmut).

The promoters were also cloned 5’ of ADS (in the same backbone as PrstA-ADS) because amorphadiene production was the ultimate goal. These constructs were each transformed with pMevT and pMBIS and tested for amorphadiene production in LB as a screen to identify the best targets for follow up work, Figure 26. The candidates that had low expression of RFP also had low production of amorphadiene (dxr, purH, mdtE, sucA, bolA). The candidates with the best production compared to pADS were gadE, atpI, paaA, yrbL, and reassuringly rstA. sdhC and ibpA were also chosen for follow up work because they had good expression of RFP and some of the highest ratios of activation with FPP accumulation. These seven promoters were tested again in EZ RDM + 1% glucose, Figure 27. gadE, yrbL, and rstA were the best producers, although they were still producing below pADS. paaA, which had produced the most in LB, produced the least in the EZ RDM + 1% glucose. Similarly, atpI had lower production in this media and the larger activation by FPP of sdhC and ibpA did not improve their production.

48

Figure 26. The thirty four promoters plus rstA were cloned 5’ of ADS and tested compared to pADS. Amorphadiene was measured after 24 hours of growth.

Figure 27. The amorphadiene production of a select group of promoters in EZ RDM + 1% glucose, measured after 24 hours.

Discussion

The knowledge gained from the FPP toxicity microarray study was used to create sensors for FPP and drive the downstream enzyme to produce amorphadiene in an inducer-free manner. Several promoters of genes in the PhoPQ regulon were tested for their response to magnesium and the promoter with the largest dynamic range and biggest differential expression in the microarray study, rstA, was used to make an FPP sensor and drive amorphadiene synthase. The sensor responded to FPP accumulation in a dose-dependent manner; however the response saturated above 5 mM mevalonate added to the medium. Recent work in our lab has shown that the S. cerevisiae mevalonate kinase is inhibited by FPP (similar to reports of the human mevalonate kinase (Hinson et al., 1997) as well as by mevalonate, thus the saturation could be a result of this inhibition. Alternatively, the

49

chromosomal copy of PhoP could have been saturated preventing further binding to the promoter region; however, overexpression of the two component system did not improve the response.

The sensor was also used to detect FPP levels as a result of different terpene synthase activities. The dynamic range when used as a sensor was only two, which would need to be improved before using this device in high throughput screening to find improved terpene synthases. An alternative approach to improve the output and dynamic range would have been to increase the binding site strength of PhoP and vary its binding location in the promoter region (Cox, Surette, & Michael B Elowitz, 2007), however these approaches have not been taken yet. Of the promoters in the PhoPQ regulon, PhoP had a moderate binding strength in the rstA promoter relative to other members, so there is room for improvement (Minagawa et al., 2003).

Despite the shortcomings of the current design of the sensor there is still value due to our inability to measure FPP reliably in extracted cell lysates. The detected peaks reported in Figure 22F, were just above the detection limit of the highly sensitive time of flight mass spectrometer. The detection of these molecules has been challenging mainly because of their instability due to the nature of the pyrophosphate bonds. Previous approaches in our lab to measure the isoprenyl diphosphate intermediates was to use a protocol where the extract was treated with a phosphatase to remove the phosphates and measure the alcohols on HPLC (with radioactive labeling) (D. Zhang & C D Poulter, 1993). However, recent publications have demonstrated efficient methods to extract and detect many metabolites in a high throughput manner, so this bottleneck could be alleviated in the future (Bajad et al., 2006; Ewald, Heux, & Zamboni, 2009).

Regardless of the shortcomings of the sensors, the promoters were able to produce significant amounts of amorphadiene in an inducer-free manner. The rstA promoter was the most studied and showed production ranging from 30 to 70% of the stronger trc promoter. The inducer-free system, grew slightly slower than the original constructs, but still reached the same final density. Analysis of the transcript and protein levels of the pathway enzymes revealed that the rstA promoter was much weaker, approximately 30 fold according to the transcripts and five-fold by protein analysis. Coupled with a now constitutive mevalonate pathway (without LacIQ), FPP accumulated to higher levels than the inducible system. This accumulation could cause a slowing of growth while ADS is being made in response. Despite the much lower protein levels, production was still comparable; perhaps indicating that the ADS made from the trc promoter is in excess.

After expanding the screen to include more promoters from the microarray studies, several promising candidates were selected which produced amorphadiene in LB to levels comparable to rstA. The selected promoters were tested again in EZ RDM + 1% glucose and the paaA promoter, which produced the most in LB, now produced much less. The paaA promoter is crp regulated, so the promoter could be catabolite repressed with the addition of glucose (Ferrández, García, & Díaz, 2000). This is an important consideration to choose promoters that are robust under many conditions (although the promoter region could also be engineered to

50

limit this response). The other promoters that performed well were PyrbL and PgadE, members of the PhoPQ regulon and the acid response regulator respectively, two of the most prominent responses to the toxicity. The top three producers were still lower than pADS, thus in the next chapter the responsive promoters were used to regulate the entire pathway.

51

Chapter 4: Full feedback and feed-forward control of the pathway




Cloning ................................................................................................................................................ 53


Proteomics, HPLC ................................................................................................................................ 54

HPLC ................................................................................................................................................ 54

Proteomics ...................................................................................................................................... 54

Results ..................................................................................................................................................... 55

Construction of Inverters and Regulated mevalonate pathway ......................................................... 55

PgadE-MevT-MBIS maintained low acetate concentrations ................................................................. 57

Testing gadE promoter region truncations for amorphadiene production........................................ 61

Testing other strength promoters with differential expression ......................................................... 62

Bisabolene and isopentenol production ............................................................................................. 66

Discussion................................................................................................................................................ 67

Introduction

The previous chapters highlighted the microarray response to FPP toxicity and then used the response to create a sensor using the rstA promoter that responded in a dose dependent manner to accumulation. When the promoter was driving amorphadiene synthase, the strain was able to produce amorphadiene without the addition of IPTG; however, the production was consistently lower than pADS. The transcription and translation from the responsive promoter was much lower, yet efforts to improve these rates were unsuccessful. The number of promoters used was expanded to try a broader collection of E. coli promoters. These showed a variety of responses to FPP accumulation and amorphadiene production. Several new highly producing promoters were chosen for follow up work, but the production was never better than the inducible pADS.

Thus, a new approach was required to improve inducer-less production. Up to this point, the response to FPP was only used to drive the downstream enzyme. However, most biosynthetic pathways have regulation on the whole pathway. For instance, the mevalonate pathway in its native host, Saccharomyces cerevisiae, is regulated extensively by allosteric feedback regulation as well as transcriptional control (Maury, Asadollahi, Møller, Clark, & Nielsen, 2005). Another recent example of dynamic control was recently elucidated in the

52

leucine biosynthetic pathway in yeast (Chin, Chubukov, Jolly, J. DeRisi, & Li, 2008). The pathway is feedback regulated by a transcription factor that responds to leucine. The enzyme downstream of the committed step of the pathway responds strongly and abruptly to starvation, but the entire pathway is also regulated by the transcription factor and also has a small increase in transcription. Thus, my approach to this point had neglected to control the upstream pathway.

Achieving dynamic control of a metabolic pathway is not a trivial task. Most efforts to date have relied on fine tuning static promoters to achieve the desired output. This has been done by replacing promoters with promoters of a desired strength (Alper et al., 2005; De Mey et al., 2010; L. Z. Yuan, Rouvière, Larossa, & Suh, 2006) or creating libraries of known outputs (Hammer et al., 2006; Miksch et al., 2005). Alternatively, promoters have been chosen that respond to the growth phase or environmental conditions to trigger a response (Kang, Q. Wang, H. Zhang, & Qi, 2008; K. H. Lee et al., 2003). As our ability to make genetic circuits becomes more pronounced, multiple signals can be combined to generate a unique output (Clarke & C. A. Voigt, 2011). To this end, a bacterium has been engineered to invade tumor cells in response to multiple conditions indicative of a cancer environment, yet the multiple signals have yet to be combined into a single output (Anderson, Clarke, Arkin, & C. a Voigt, 2006).

Perhaps the best example of dynamic metabolic control is by the Liao lab (Farmer & James C Liao, 1997; 2000; 2001). They used the 2 component system responsive to nitrogen levels, NtrBC, to respond to internal levels of acetyl phosphate. The system only worked when the sensor kinase, NtrB, was knocked out; under these conditions, NtrC was phosphorylated by accumulation of acetyl phosphate. This only happened under conditions of excess glycolysis flux when acetate would accumulate. They overexpressed key enzymes in the DXP pathway under control of a NtrC regulated promoter so that excess glycolysis flux was diverted to lycopene production rather than acetate. The dynamic responding promoter produced more lycopene and minimized acetate accumulation compared to the static inducible promoter.

A fully responsive pathway was created in two ways. First, an inverter was built with LacI under the control of the FPP responsive promoter to repress the lac promoters driving the mevalonate pathway. Second, responsive promoters replaced the inducible lac promoters in front of the mevalonate pathway. The designs were tested and the strains with the fully responsive pathways grew just as well as the ones with the inducible systems, an improvement over the pathways with only the downstream enzyme responsive. The responsive pathways produced varying amounts of amorphadiene, with the first method doing much worse and the second method doing much better than the original constructs. This work demonstrated the benefits of using a responsive pathway, not only by eliminating the need for an exogenous inducer, but also by minimizing acetate accumulation and allowing full glucose consumption.


Strains and Growth

The strains and plasmids used in this study are described in the appendix in Table 2.

53

Medium components and chemicals were purchased from Sigma-Aldrich, EMD chemicals, Difco, and TekNova. Mevalonolactone (Sigma) was converted to mevalonate by adding 1.02 volumes 2 M KOH to 1 volume 2 M mevalonolactone in water. The mixture was incubated at 37 °C for 30 minute, and then filter sterilized. Luria Broth Miller (LB) (EMD), EZ rich defined medium (RDM) (TekNova), and a modified M9 MOPS medium were used for cell growth. Magnesium sulfate and glucose were added to the media at varying levels as indicated. The modified M9 MOPS medium consisted of 1x M9 salts (Difco) (1 g/L NH4Cl, 0.5

g/L NaCl, 6.6 g/L Na2HPO4 (anhydrous), 3 g/L KH2PO4), 2 mM MgSO4, 100 M CaCl2, 40 mM MOPS, 1x trace metal containing (3x10-6 mM (NH4)6(MO7)24, 4x10-4 mM HBO3, 3x10-5 mM CoCl2,

10-5 mM CuSO4, 8x10-5 mM MnCl2, 10-5 mM ZnSO4), 4.5 g/ml thiamine, 0.01 mM FeSO4

(heptahydrate), and glucose (Neidhardt et al., 1974). Chloramphenicol (35 g/ml), Tetracycline

(10 g/ml), carbenicilin (100 g/ml), and kanamycin (50 g/ml) were added as appropriate. Cells were grown at 30 °C to improve production.

Cloning

The BglBrick strategy was again used for most constructs and was described in the previous chapter. The fluorescent reporter inverters were constructed by piecing together successively larger parts. Briefly, GFP, LacI, and a terminator were ligated in order, in parallel to the trc promoter and RFP. The two parts were cloned together to make a GFP-LacI-terminator-Ptrc-RFP part and then ligated 3’ of PrstA or PBAD. The constructs were tested by varying the magnesium or arabinose in the medium, respectively.

The one plasmid mevalonate pathway inverter was built by cloning LacI and a terminator behind PrstA-ADS. The Plac promoter was cloned in front of a biobrick MBIS part, and the entire construct was cloned behind PrstA-ADS-LacI-terminator. The construct was fed mevalonate. The PrstA-ADS-LacI-terminator construct was also tested as a separate plasmid for amorphadiene production.

The one plasmid biobrick mevalonate pathways (MevT, MevT-MevB, MevT-MBIS) were constructed by TS Lee on a pBbA5c backbone (p15a, lacUV5 promoter, chloramphenicol). The pathway enzymes were cloned into the backbone with the LacIQ/lacUV5 promoter removed (pBbA0c-RFP) in place of RFP, to make the plasmid pBbA0c-MevT, pBbA0c-MevT-MevB, or pBbA0c-MevT-MBIS. The selected promoter regions were cloned 5’ of the mevalonate pathways. The promoter regions were also cloned 5’ of the reporter plasmid (pBbA0k-RFP). The shortened gadE promoter regions were designed to include different transcription factor binding sites and promoters. The list of all promoter regions is listed in Table 3.

The constitutive promoters were designed by Chris Anderson and were selected from the MIT Parts registry, http://partsregistry.org/Promoters/Catalog/Anderson. The promoters covered a range of outputs as measured by Chris Anderson and are listed in Table 5. The promoters were designed in one oligonucleotide. A complementary oligonucleotide was ordered with EcoRI compatible overhangs on the 5’ and BamHI overhangs on the 3’ end so the annealed oligonucleotides could be ligated into an EcoRI/BglII digested plasmid (pBbA0c-MevT-

MBIS or pBbA0k-RFP). The complementary oligonucleotides (10 M of each) were annealed with 50 mM NaCl in a 30 mL reaction by placing the tube in boiling water for 2 minutes. After

http://partsregistry.org/Promoters/Catalog/Anderson

54

cooling, 4 L of the annealed primer mix was added to 2 L of digested plasmid and ligated with T4 DNA ligase.

The isopentenol (3-methyl-3-buten-1-ol) and bisabolene producing constructs were made by cloning the C9b fragment and codon optimized bisabolene synthase (constructs made by Syd Withers and Pamela Peralta-Yaha, respectively) from the pC9b and pBISco constructs (ptrc99a backbone with trc promoter). The genes were cloned with a BglII site in the 5’ primer and XhoI in the 3’ primer and cloned 3’ of PrstA in a BamHI/XhoI digested pBbE0a-PrstA vector.


The promoter strength was tested using the sensor plasmids without the mevalonate pathway plasmids present. Overnight cultures were picked form single colonies and used to inoculate 5 ml fresh media, EZ RDM + 1% glucose (1% inoculation). Cell growth (OD) was measured at 600 nm in a Beckman spectrophotometer. Fluorescence was measured in a Safire plate reader (TECAN). RFP was measured at 558/583 nm (excitation/emission).

Amorphadiene and bisabolene were produced using the mevalonate plasmids with the lacUV5 or selected promoters as indicated and the amorphadiene and bisabolene synthases 3’ of the trc or rstA promoter. Overnight cultures were picked from single colonies and used to

inoculate 5 ml fresh media, EZ RDM + 1% glucose. IPTG (500 M) was added as indicated to the culture at inoculation. Dodecane (20%) was added as an overlay. Amorphadiene and

bisabolene were measured by diluting 5 L of the dodecane layer into 495 L of ethyl acetate spiked with the internal standard trans-caryophyllene. The samples were run on an Agilent GC-MS as previously described (V. J. J. Martin, Pitera, Withers, J. D. Newman, & Jay D Keasling, 2003). The quantity was determined from a standard curve using pure standards produced from large scale shake flask production.

Isopentenol was produced using PlacUV5-MevT-MevB or PgadE-MevT-MevB and pC9b or PstA-C9b. Overnight cultures were picked from single colonies and used to inoculate 5 ml fresh

medium, EZ RDM + 1% glucose. IPTG (500 M) was added as indicated to the culture at inoculation. Isopentenol was extracted by adding 150 mL culture to 750 mL 4:1 chloroform:methanol. After 20 minutes of shaking, the organic layer was transferred to a GC vial and measured on a GC-FID. A standard curve was prepared from commercially available isopentenol (SIGMA).

Proteomics, HPLC

HPLC

Glucose and acetate concentrations were measured from the cell culture supernatant by HPLC. An Agilent 1100 Series Binary LC System equipped with a Bio-Rad Aminex HPX-87H 300x7.8 mm column (with Micro-Guard Cation H Refill Cartridges, 30 x 4.6 mm) and an Agilent 1200 Series refractive index detector. The samples were run through the column at 50°C and a flow of 0.6 ml/min with 4 mM H2SO4 as eluent. Standard curves were prepared with commercially available standards (SIGMA).

Proteomics

55

Samples were collected throughout growth for targeted proteomics via selected reaction monitoring (SRM) as described previously (Redding-Johanson et al., 2011). Briefly, 20 OD*ml cell pellets were frozen in liquid nitrogen and then extracted with chloroform/methanol and dried in a speed vac. The dried protein extract was resuspended in 50 mM ammonium bicarbonate in 10% acetonitrile. After quantification with a Bradford assay, 1 mg/ml of protein was reduced in 5 mM tris(2-carboxyethyl)phosphine (TCEP) at room temperature for 30 minutes. The proteins were alkylated with 200 mM iodoacetic acid in 100 mM NaOH for 30

minutes in the dark. The proteins were digested with Trypsin (1 g/l) at 37 °C overnight. After addition of pure bovine serum albumin digests as an internal standard, samples were analyzed on a LC/MS-MS (Eksigent TEMPO nanoLC-2D coupled to an Applied Biosystems 4000Q-Trap mass spectrometer) as described previously.

Results

Previously, 35 promoters identified in the microarray response to FPP accumulation were cloned 5’ of RFP and ADS. The promoters driving ADS produced a wide range of amorphadiene, but the best of the lot never were able to produce as much as the inducible trc promoter, despite efforts to improve their transcription or translation rates. Thus a new approach was taken, more closely mimicking natural biosynthetic pathways where the entire pathway is coordinately regulated. We wanted to regulate the pathway so that the enzymes downstream consuming FPP turn on when it accumulates, and the enzymes upstream creating it turn off when too much is made. The pathway couldn’t be down regulated too much, or else amorphadiene production would be detrimentally impacted.

Construction of Inverters and Regulated mevalonate pathway

Much time was spent trying to build the entire pathway in an inverter design on one plasmid. The FPP responsive promoter, PrstA, would upregulate ADS and LacI in response to increased amounts of FPP. In turn, LacI would repress the Plac driven mevalonate pathway preventing further accumulation of FPP until ADS could convert it to amorphadiene, Figure 28. Similar constructs were made using fluorescent reporters in place of ADS and the mevalonate pathway to test the design. The constructs with the fluorescent reporters using PrstA and varying magnesium typically were only able to modulate the downstream enzymes, but the inversion never worked consistently. Tests showed that the “off” state of the rstA promoter was not low enough to have LacI present in low enough quantities to have “induced” behavior from the upstream promoter, a common problem with engineered circuits (H. Kobayashi et al., 2004). Degradation tags were used on LacI (Andersen et al., 1998), but with little success. When a commonly used, inducible promoter (PBAD) was used, inversion was successful (data not shown).

Ultimately, the design on one plasmid was never able to produce amorphadiene (data not shown). The downstream portion of the inverter was tested separately for amorphadiene production on a similar vector to PrstA-ADS. PrstA-ADS-LacI was created by biobricking LacI behind ADS in an operon, with a small gap and its own ribosome binding site after the stop codon of ADS. The new inverter version was tested for production with pMevT and pMBIS along with pADS and PrstA-ADS. The addition of the LacI gene lowered production relative to the

56

construct without it, and PrstA-ADS was again lower than pADS, Figure 29. Interestingly, the construct with LacI did not produce the same as PrstA-ADS when IPTG was added.

Figure 28. A representative diagram of an inverter to up regulate the downstream gene in response to FPP. In addition it would up regulate the repressor protein for the lac promoter driving the mevalonate pathway, thus down regulating expression of the pathway.

Figure 29. The production of amorphadiene from the responsive promoter with the addition of LacI to repress the mevalonate pathway. 28 hour production in EZ RDM + 1% glucose.

With the lower production from the inverter constructs and the difficulty tuning the fluorescent inverters, a different approach followed. The responsive promoters highlighted in previous chapters were tested for their ability to express the mevalonate pathway. Initially no constraints were added about how the promoters needed to respond (i.e. up or down regulated). The three promoters highlighted with good production when producing amorphadiene, Figure 27, PyrbL, PgadE, and PrstA, were the first to be tested.

The mevalonate pathway plasmids had been cloned into BglBrick compatible vectors, easing the cloning of the three promoters into new constructs. Additionally, the new mevalonate plasmids used the lacUV5 promoter in place of the lac promoter because it was 2x

57

stronger and improved production in another study (J. R. Anthony et al., 2009). Additionally, the mevalonate pathway was combined into one operon, MevT-MBIS, and placed on a single plasmid to ease cloning. The gadE promoter was tested 5’ of MevT and MevT-MBIS and compared to the three plasmids version (pMevT/pMBIS/pADS or with PrstA-ADS) or the newer plasmids all with PrstA-ADS, Figure 30. The original three plasmid version with pADS and PrstA-ADS behaved similarly to previous experiments, with pADS producing more than the responsive promoter. The constructs with PgadE or lacUV5 5’ of MevT and pMBIS as a separate plasmid, followed a similar trend where the inducible promoter produced slightly more. However, the biggest difference occurred when PgadE was 5’ of MevT and MBIS in a single operon. The strain with PgadE-MevT-MBIS now produced twice as much as the one containing PlacUV5-MevT-MBIS, and more than any of the other combinations.

Figure 30. Production of amorphadiene from the responsive promoters. The promoters were compared to the three and the two plasmids IPTG induction system.

PgadE-MevT-MBIS maintained low acetate concentrations

The strain with the responsive promoter 5’ of MevT-MBIS produced twice as much amorphadiene as the strains with the inducible lacUV5 or lac promoters. The gadE promoter consistently produced at these levels. During these tests, we observed that the PgadE containing strains reached higher cell densities and that the color of the cell culture was different after a day or more of growth. The supernatant pH was measured and the PgadE containing media were near 7, but the PlacUV5 containing media were near 5 (despite being grown in buffered EZ RDM). To test this, the cultures were grown and samples of the supernatant were saved and run on the HPLC to measure glucose and acetate levels. The other two promoters, yrbL and rstA, were also tested in front of the mevalonate operon, Figure 31.

58

The two plasmid systems, PlacUV5-MevT-MBIS and the responsive promoters PgadE-, PyrbL-, and PrstA-MevT-MBIS, were transformed with either pADS or PrstA-ADS and grown for several days to monitor amorphadiene production as well as glucose and acetate concentrations. The third day of production is represented below. PgadE-MevT-MBIS continued to outperform the lacUV5 pathway by two fold. When coupled with the responsive PrstA-ADS, production without IPTG was still more than the inducible pathway with IPTG. The other responsive promoters driving MevT-MBIS produced less than PgadE, but similar amounts to lacUV5, Figure 31A. After 72 hours, all the strains except PgadE-MevT-MBIS/pADS when induced had glucose remaining. PgadE-MevT-MBIS/PrstA-ADS had the next lowest amount, and all other strains had over 1 g/L glucose remaining in the supernatant, Figure 31B. The acetate levels followed a similar trend. PgadE-MevT-MBIS/pADS kept acetate levels to a minimum (when induced), while all other strains had acetate accumulation resulting in the change in pH, Figure 31C. Thus, the PgadE-MevT-MBIS/pADS and to a lesser extent PrstA-ADS were able to produce more amorphadiene because the strain minimized acetate accumulation, allowing the cells to maintain at viable pH and fully consume the glucose. In all other strains, acetate accumulation reduced the cells’ ability to consume glucose and produce amorphadiene. When calculating the yield on glucose, PgadE-MevT-MBIS still produced more than these other strains, with higher productivities too.

The higher producing gadE promoter and the inducible lacUV5 pathway (both coupled with pADS) were tested for its enzyme protein concentrations. Samples were collected throughout growth and analyzed by MRM on a LC-MS/MS. The mevalonate pathway enzyme levels were measured relative to an external standard (bovine serum albumin, BSA), Figure 32A-I. The general trend was that the pathway enzymes expressed from the gadE promoter started at time 0 (before induction) higher than the uninduced lacUV5 promoter, then had a slight decrease before increasing expression after 5 hours. The enzymes from the lacUV5 promoter increased slightly over time, but were far lower than the gadE promoter. Interestingly, the gadE and lacUV5 promoters expressed RFP equally as well, data shown below.

The strain with PgadE maintained acetate at low concentrations and had higher cell viability than the strain with PlacUV5. The impact of acetate was tested by transforming the

strains into DP, the acetate knockout ackA,pta,poxB, Figure 33. In DH1, the mevalonate pathway driven by PgadE maintained high production rates compared to PrstA and PlacUV5. In DP however, the strain with PlacUV5 recovered and had the highest production. PrstA, which also had acetate accumulation in DH1, increased production as well, while PgadE maintained the same level. Thus it appeared that the acetate accumulation in DH1 had a negative impact on its ability to maintain amorphadiene production. The promoters expressing a fluorescent reporter behaved similarly in DH1 or DP, regardless of exogenous acetate addition to the media (data not shown).

59

A

B

C

Figure 31. Production of amorphadiene from the responsive promoters in the 2 plasmid system. 72 hours also monitoring glucose and acetate.

60

A

B

C

D

E

F

G

H

I

Figure 32. Protein expression during growth of the PgadE-MevT-MBIS/pADS strain and PlacUV5-MevT-MBIS/pADS. The proteins are normalized to an internal standard and are measured by MRM on an LC-MS/MS. Time zero is induction.

BLUE REDPgadE-MevT-MBIS/pADS PlacUV5-MevT-MBIS/pADS

61

Figure 33. Production of amorphadiene from the responsive promoters in the wild type, DH1, or acetate knockout strain, DP. 68 hour production EZ RDM. All strains have IPTG added.

Testing gadE promoter region truncations for amorphadiene production

The gadE promoter region expressing the mevalonate pathway consistently produced more than the IPTG inducible lacUV5 pathway in DH1. GadE is the regulator for many genes in the acid response network of E. coli and its promoter region is 750 nucleotides with 4 promoters annotated on ecocyc.org (Sayed & Foster, 2009). There are binding sites for many transcription factors, all of which result in GadE being an important regulator for the cell to respond to a variety of conditions. Given the large size of the region, number of promoters, and the number of transcription factor binding sites, truncated segments were cloned in front of the mevalonate pathway to determine if there is a minimal unit required for improved amorphadiene production.

Ten truncated promoter regions were designed to include different promoters and transcription factor binding sites, Figure 34. Promoters labeled Px, were designed from the 5’ end of the region to extend just beyond the transcription start site of the 4 annotated promoters in the region. Promoters indicated by a number, 50, 100, etc., were designed from the 3’ transcription start site in the 5’ direction to include increasing numbers of transcription factor binding sites. Although the error bars are higher than we would have liked, the main trend is that all the promoter truncations work just as well as the full length version except for the two shortest promoters. The 50 and 100 base pair truncations from the 3’ end do not reach the gadEp1. The full length gadE promoter was used for all further work.

62

Figure 34. Shortening the gadE promoter to test which are the essential parts. Production measured in EZ RDM at 89 hour.

Testing other strength promoters with differential expression

The gadE promoter showed better performance partially because it was able to stay active longer. The protein expression profile showed that expression was higher than lacUV5 throughout growth and into stationary phase. To test if the improved function was a result of its strength or its dynamic expression profile, more promoters were selected from the microarray results. Because gadE had increased expression in stationary phase and is the acid response regulator, more promoters with similar functions were selected (osmY, yahO, xasA – which is also known as gadC, slp, and bolA was redesigned). Additionally, genes without differential expression in response to FPP, but of varying expression intensity (low – hyfD,

63

nemA, and thiM, and high – b1592, yhdT, and yihE), were selected as a control for gene strength vs. responsiveness. In all, 20 more promoters were tested, including some redesigned versions of the previous set which had no expression of RFP or production of amorphadiene.

The new promoters were cloned 5’ of RFP and the MevT-MBIS operon. The promoter strength was tested for these constructs by monitoring fluorescence, Figure 35A. PgadE, PlacUV5, and Ptrc were tested as well to characterize their respective strengths. The promoters were sorted for strength, relative to Ptrc. PgadE and PlacUV5 were the same strength and Ptrc was about twice as strong. The other promoters covered a range of outputs, with other stationary phase promoters ranking near the top. Of the promoters that were redesigned, bolA, improved expression and now ranked among the top. BolA , XasA, Slp, and GadE are all implicated in the acid response and stationary phase expression, and all rank at the top. ThiM and EntC were also near the top of the expression strength, but were not changing in the microarray study. All of the other redesigned promoters retained their low expression.

The nine strongest promoters were selected as well as five of the weaker promoters to test for amorphadiene production. These promoters driving MevT-MBIS were all transformed with pADS and the strains were induced with IPTG. Amorphadiene was measured after 30 hours of growth, Figure 35B. The stationary phase promoters all produced more than the lacUV5 promoter, even osmY which had lower RFP expression. The weaker promoters all produced less than lacUV5, including the non-changers, thiM and entC. Thus, it seems that the strength of the promoter was important as well as the fact that they were expressing in stationary phase.

To further test the promoter strength’s impact on amorphadiene production, seven constitutive promoters were picked covering the range of strength. The promoters were designed by Chris Anderson and are listed on the MIT parts registry, with their arbitrary strengths from the registry summarized in Table 5 (http://partsregistry.org/Promoters/Catalog/Anderson). The promoters were cloned 5’ of RFP and the MevT-MBIS operon and tested as above. The strongest promoter was equivalent to the induced trc promoter, #2 and #3 were similar in strength to gadE and lacUV5, and the other 4 were of weaker strength, Figure 36A. However, when producing amorphadiene, all the constitutive promoters produced nearly zero amorphadiene, Figure 36B. In fact, the strongest constitutive promoter was never successfully cloned 5’ of MevT-MBIS. There was a general trend of the weaker constitutive promoters producing more amorphadiene, however the total amounts were so small that this was negligible compared to the gadE and the inducible promoters. The stronger inducible trc promoter was included for comparison. It produced just slightly less amorphadiene than the gadE promoter. The strength of induction mattered; however, the trc promoter was also roughly two times stronger when producing RFP, yet didn’t produce more amorphadiene.


64

A

B

Figure 35. Testing additional promoters for fluorescence and amorphadiene production. Some from the earlier tests were redesigned. A)Expression of RFP normalized to the trc promoter after a day of growth. B)Each promoter is 5’ of MevT-MBIS and is paired with pADS. Amorphadiene measured after 30 hours growth and sorted in the order of promoter strength.

65

A

B

Figure 36. Testing several constitutive promoters of varying strengths for amorphadiene production. A)Expression of RFP from the promoters measured 24 hours after growth. B)Each promoter 5’ of MevT-MBIS coupled with pADS. Measuring amorphadiene at the times indicated.

66

Bisabolene and isopentenol production

To broaden the number of products made with the responsive mevalonate pathway, several biofuel targets were selected. After hydrogenation, isopentenol would be a short chain alcohol alternative to ethanol or butanol, and bisabolene would be more of a diesel or jet fuel. They are both derivatives of the isoprenoid pathway; isopentenol is a five-carbon molecule and bisabolene is a 15-carbon molecule. The pyrophosphatase needed to make isopentenol from DMAPP was isolated from Bacillus subtilis and the genomic fragment was cloned behind the trc promoter, akin to amorphadiene synthase (pC9b) (Withers, Gottlieb, Lieu, J. D. Newman, & Jay D Keasling, 2007). The fragment was also cloned behind PrstA. Bisabolene synthase was codon optimized (thanks to P.P-Yaha) and cloned behind the trc promoter as well as PrstA, pBIS and PrstA-BIS respectively. The gadE promoter was cloned 5’ of the shortened MevT-MevB operon for the production of isopentenol.

The isopentenol production from the inducible lacUV5 promoter was finicky, Figure 37A. With IPTG induction, production was only measured with the PrstA-C9b construct. The production from the pC9b construct was also low when produced from the PgadE-MevT-MevB mevalonate pathway. The pC9b construct should be IPTG inducible, but the –IPTG sample produced as much as the induced sample. However, when C9b is regulated by PrstA, production from the PgadE-MevT-MevB operon is significantly higher. The improved production from the weaker rstA promoter could be indicative of protein expression toxicity by the pyrophosphatase. The rstA promoter was not tested for responsiveness to DMAPP, but if the toxicity of the 5 carbon isoprenyl diphosphate molecule mimics FPP, the improved production could be a result of the promoter responding to the same stress and expressing C9b as necessary. The gadE promoter also proved to be superior to lacUV5 again.

Bisabolene production from PlacUV5-MevT-MBIS behaved similarly to amorphadiene production. PgadE-MevT-MBIS produced more than PlacUV5-MevT-MBIS. The improvement in production was less than the two fold improvement seen with amorphadiene production, but it was consistent and significantly different regardless. When coupled with PrstA-BIS, IPTG independent production was achieved, and it was at similar levels to the stronger trc promoter.

A

B

Figure 37. Production of isopentenol and bisabolene from the gadE promoter. 40 hour production bisabolene, 70 hr isopentenol. IPTG added to all tubes except the PgadE/PrstA combination.

67

Discussion

The use of the responsive promoters controlling the entire pathway was wildly successful. When solely the downstream enzyme was controlled by a responsive promoter in the previous chapter, the responsive pathways grew slower and produced less than the inducible systems. However, by mimicking natural biosynthetic pathways and regulating the entire pathway with a responsive promoter, growth and production was improved. The responsive pathway was able to minimize acetate accumulation, and thus consume all the glucose in the medium, resulting in twice as much amorphadiene production. Once the accumulation of acetate was removed, the lacUV5 promoter was able to produce more amorphadiene.

The gadE promoter and lacUV5 promoter were roughly the same strength when expressing RFP; however, when expressing the pathway enzymes their expression changed drastically. The gadE promoter produced roughly five times more pathway enzymes, and also importantly there was more protein around in stationary phase when acetate would normally accumulate. By having more protein later in the growth phase, the gadE driven pathway was able to redirect the flux of carbon from acetate to more production of amorphadiene.

To test the importance of the strength of the promoter driving the pathway, the stronger trc promoter was used in place of the lacUV5 promoter. The trc promoter improved production over the lacUV5 promoter, but did not produce as much as the gadE promoter (despite being twice as strong when expressing RFP, pathway enzyme levels have not been measured from this strain). Even worse were constitutive promoters. The strongest constitutive promoter (equivalent to trc) was not able to be cloned in front of the pathway and the other constitutive promoters of varying strengths produced far less amorphadiene than the inducible or the responsive pathways. During the tests with the constitutive promoters, IPTG was added from the beginning to express ADS and prevent the accumulation of FPP and resulting growth problems.

The strength of the promoter seemed less important than the timing of expression. There have been many reports about the timing of induction for protein over expression in E. coli (C. Lee et al., 1997; Neubauer, Hofmann, Holst, Mattiasson, & Kruschke, 1992; Studier, 2005). Typically in these approaches, maximizing the yield of protein is the most important factor; protein expression begins and cell growth ceases due to immense cellular burden the expression system causes to the cell. However, for the production of a metabolite as a final product, the expression system cannot shut down cell metabolism. While the implementation of dynamically responding heterologous pathways is in its infancy, one common feature has been the timing of expression to coincide with stationary phase and the accumulation of acetate (Farmer & James C Liao, 2000; Kang et al., 2008). Additionally, recent studies which optimized expression through the isoprenoid pathway, but with traditional promoters, have also resulted in minimal acetate accumulation (Ajikumar et al., 2010; Tabata & Hashimoto, 2004). The reduction in competing pathways is a common strategy for traditional metabolic engineering (Lin, George N Bennett, & K.-Y. San, 2005; Sánchez, George N Bennett, & K.-Y. San, 2006; Y. T. Yang, G N Bennett, & K. Y. San, 1999), so it is no surprise that an improved pathway

68

also results in lower acetate accumulation, and may suggest that knockouts are not needed if the pathway performs at optimal levels (we have observed this as well with another pathway and strong enzymes). Interestingly, the mevalonate pathway currently used when expressed from the lacUV5 promoter in DH1 results in large acetate accumulation and poor performance, but when expressed in the acetate pathway knockout strain performs much better. The use of responsive promoters however removed this problem using the same enzymes.

The success of the gadE promoter led to the study of other stationary phase and acid responsive promoters controlling the mevalonate pathway. This class of promoters also produced high levels of amorphadiene, more so than promoters that didn’t change (thiM, yihE) or a differentially expressed stress response promoter (pspA). The stationary phase and acid responsive promoters were also higher strength when expressing RFP. It remains to be seen if this trend is specific to this stress or a general strategy to have stationary phase promoters controlling expression. As discussed in previous chapters, the acid stress response was consistent across the microarray conditions while other typical stress responses were not. To confirm the specificity to this response, other commonly identified stationary phase and stress response promoters should be tested (including those that are not differentially expressed). Alternatively, a library of stationary phase promoters, which could control for expression strength, has also been created (Miksch et al., 2005) and could be used to control the pathway. The actual dynamics of the pathway expression in response to FPP also need to be studied to determine why this class of promoter excels and how they are actually responding to FPP (are they downregulating pathway expression in response to FPP accumulation or are they upregulating into stationary phase).

While the responsive promoters driving the pathway worked well, the inverter strategy never worked. The problem with the strategy was that the balancing of the outputs of the two promoters was never matched. Additionally, the rstA promoter might not be a strong enough promoter in response to FPP (the dynamic range of the promoter in response to Mg2+ is much bigger than the response to FPP) to successfully build an inverter. The tuning of the promoter outputs is a crucial component of making a working genetic circuit (Anderson, Clarke, Arkin, & C. a Voigt, 2006; Anderson et al., 2007; H. Kobayashi et al., 2004), and these strategies have yet to be applied in this case. Thus despite these limitations, a construct using only one responsive promoter could still be designed.

Nonetheless, the responsive promoter strategy worked well for the production of amorphadiene as well as other isoprenoids. The five carbon isopentenol and 15 carbon bisabolene were produced from the gadE driven pathway and rstA driven synthase at higher levels than the inducible systems. The production of these molecules in an IPTG independent manner bodes well for the production of these molecules at large scale without the addition of an expensive inducer. The robustness and scale up of the responsive pathway is discussed in the following chapter.

69

Chapter 5: Scale up and other applications of the pathway



Media and Growth .............................................................................................................................. 70


Proteomics .......................................................................................................................................... 71

Results ..................................................................................................................................................... 71

Glucose feeding ................................................................................................................................... 71

Minimal Media .................................................................................................................................... 72

Discussion................................................................................................................................................ 76

Introduction

The creation of a responsive induction system for the production of sesquiterpenes without the need of an exogenous inducer resulted in the discovery of the gadE promoter to drive the mevalonate pathway. The production of amorphadiene, bisabolene, and isopentenol from these promoters in shake flasks worked exceptionally well. The final goal of the project was to test the responsive production of these molecules in more industrial-like conditions in fed-batch high cell density fermentation. This will be an important characterization for the responsive promoters as the cells will be under different stresses and the robustness of the approach will be tested.

E. coli has been successfully grown to high cell densities (>200 g/L DCW) in fed-batch fermentation (S Y Lee, 1996) primarily for the production of recombinant therapeutics (Yee & Blanch, 1992), although there have been advancements for the production of low value, high volume products, 1-3 propanediol (Genencor and DuPont, (Nakamura & Whited, 2003)) and polyhydroxybutyrate (Metabolix-ADM), among others (Shiloach & Fass, 2005). More recently Amyris demonstrated a process to produce the anti-malaria drug precursor, amorphadiene, at titers approaching commercial feasibility (Amyris (Tsuruta et al., 2009)).

The standard approach for induction of heterologous pathways in shake flask studies are tightly controlled, inducible promoters. At industrial scale, the use of expensive inducers is cost prohibitive, except for the products with the highest margins (therapeutic proteins and other pharmaceuticals). The traditional approach has been to use constitutive promoters or promoters that respond to media components such has sugars, phosphates, metals, or environmental conditions such as temperature. For the reasons discussed previously, we are interested in creating dynamic promoters that respond to the pathway intermediates to remove the need for exogenous inducers.

70

The use of the endogenous promoters identified in this work, while showing a response to FPP, also respond to many different signals. While the responsiveness and production were tested in several different media conditions, the environment in a fermentor will be much different. In a fed-batch process, after exhausting the initial batch media components, the feed media will begin at a rate that allows E. coli to grow to high cell densities in a non-limited (nutrients) manner. This is unlike the shake flask growths, where after the cells reach stationary phase, cell growth ceases. Additionally, the cells are continuously aerated at much higher levels than the passive transfer that occurs through the opening in the shake flask. At higher cell densities, pure oxygen can also be supplemented to maintain the respiratory requirements. Finally, the pH of the culture can be maintained by the addition of base.

Given these differences, the mevalonate pathway with the responsive pathway was tested for amorphadiene and bisabolene production in minimal medium, more suitable to use in fed-batch fermentation. The initial work was done in shake flasks, using conditions to more accurately mimic the conditions in a fed batch fermentor by using slow release glucose beads and increasing shaking speed and decreasing media fill volume. Media components were also changed before scaling the process up to fed-batch fermentation.


Media and Growth

EZ RDM and M9 MOPS media were used as described in the previous chapters. Another defined minimal medium was used that had been scaled up to a high cell density, fed-batch fermentor (Korz, 1995). The Korz medium contained 15 g/L glucose, 4 g/L (NH4)2SO4, 1.7 g/L citrate, 4.2 g/L KH2PO4, 12 g/L K2HPO4, 5 g/L MgSO4, 5 mg/L thiamine, 8 mg/L EDTA, 2.5 mg/L CoCl2*6H2O, 15 mg/L MnCl2*4H2O, 1.5 mg/L CuCl2*2H20, 3 mg/L H3BO3, 2.5 mg/L Na2MoO4*2H2O, 100 mg/L Fe(III)citrate and 50 mM MOPS. Fed-batch cell growth was approximated in a shake flask by growing in EZ RDM without glucose, (Jeude et al., 2006). Polymer FeedBeads® (Kuhner), which slowly release glucose to the medium, were added to the culture.

The oxygen transfer rate was measured using the RAMOS system (Kuhner). Twenty mL of culture was added to the shake flasks fitted with oxygen sensors and the cultures were shaken at 300 rpm. The oxygen transfer rate was monitored using the associated software.


The promoter strength was tested using the sensor plasmids without the mevalonate pathway plasmids present. Overnight cultures were picked form single colonies and used to inoculate 5 ml fresh media, EZ RDM + 1% glucose (1% inoculation). Cell growth (OD) was measured at 600 nm in a Beckman spectrophotometer. Fluorescence was measured in a Safire plate reader (TECAN). RFP was measured at 558/583 nm (excitation/emission).

Amorphadiene and bisabolene were produced using the mevalonate plasmids with the lacUV5 or selected promoters as indicated and the amorphadiene and bisabolene synthases 3’ of the trc or rstA promoter. Overnight cultures were picked from single colonies and used to

71

inoculate 5 ml fresh media, EZ RDM + 1% glucose. IPTG (500 M) was added as indicated to the culture at inoculation. Dodecane (20%) was added as an overlay. Amorphadiene and

bisabolene were measured by diluting 5 L of the dodecane layer into 495 L of ethyl acetate spiked with the internal standard trans-caryophyllene. The samples were run on an Agilent GC-MS as previously described (V. J. J. Martin, Pitera, Withers, J. D. Newman, & Jay D Keasling, 2003). The quantity was determined from a standard curve using pure standards produced from large scale shake flask production.

Proteomics

Samples were collected throughout growth for targeted proteomics via selected reaction monitoring (SRM) as described previously (Redding-Johanson et al., 2011). Briefly, 20 OD*ml cell pellets were frozen in liquid nitrogen and then extracted with chloroform/methanol and dried in a speed vac. The dried protein extract was resuspended in 50 mM ammonium bicarbonate in 10% acetonitrile. After quantification with a Bradford assay, 1 mg/ml of protein was reduced in 5 mM tris(2-carboxyethyl)phosphine (TCEP) at room temperature for 30 minutes. The proteins were alkylated with 200 mM iodoacetic acid in 100 mM NaOH for 30

minutes in the dark. The proteins were digested with Trypsin (1 g/l) at 37 °C overnight. After addition of pure bovine serum albumin digests as an internal standard, samples were analyzed on a LC/MS-MS (Eksigent TEMPO nanoLC-2D coupled to an Applied Biosystems 4000Q-Trap mass spectrometer) as described previously.

Results

The growth in a high cell density, fed-batch fermentor is a very different environment from cell culture in shake flasks. The fermentation is typically done in a defined, minimal medium to maximize reproducibility absent in an undefined medium and minimize the costly additions of more expensive media components, inducers, or antibiotics. The robustness of the responsive mevalonate pathway was tested in conditions similar to a fed-batch fermentation several ways. First, the feed rate of glucose was simulated by using slow-release glucose beads. Second, the strains were tested in a minimal medium, where the macronutrient composition was varied, and finally the tests were carried out in shake flasks with oxygen sensors attached to the head space to ensure the process was carried out in non-oxygen limiting conditions.

Glucose feeding

The feed phase of the fed-batch process adds high concentrations of glucose and other medium components to the vessel in a controlled manner, at a rate that ensures the cells can consume the glucose without the production of acetate. Under these conditions the cells are not limited by a medium component; unlike in stationary phase in shake flasks, where typically due to a medium or oxygen limitation the cell is sitting in a pool of glucose. Thus, polymer discs from Kuhner which slowly release glucose into the medium were used to mimic this process. The strains were grown in EZ RDM with the beads as the carbon source. Growth and amorphadiene and bisabolene production were monitored. Unfortunately, the strains with the beads failed to grow or produce. After several experiments, it was determined that that this was a result of the addition of dodecane to the culture. The beads were light enough that they

72

floated in the interface between the media and dodecane. Kuhner reformulated their polymer composition, but the heavier beads did not work either (data not shown).

Minimal Media

The next test for the responsive pathway was in a minimal medium. Both a M9 MOPS medium (the same as used previously) and a medium commonly (referred here as Korz) used in high cell density fermentation (Korz, 1995). The strains were tested for production of amorphadiene and bisabolene in these media after several days of acclimation. The tests were carried out in culture tubes to check for production before transferring to a fermentor. In both amorphadiene and bisabolene production, the pathway driven by the gadE promoter no longer produced as much as the lacUV5 promoter in M9 MOPS + 1% glucose, Figure 38, or Korz medium (data not shown). This was unexpected as the rstA promoter was able to respond to FPP in M9 medium and the gadE had differential expression in the low magnesium M9 FPP microarray experiment.

The promoter expression in the various media was tested by measuring the fluorescence from the constructs with RFP. Interestingly, the trc, lacUV5, gadE, and rstA reporters all had the same relative expression ratios in the M9 medium, Figure 39. The absolute fluorescence was also essentially the same as in EZ RDM. The pathway enzyme levels were also tested in the different medium by targeted proteomics. Unlike the fluorescent reporters, the pathway enzymes behaved much differently in the rich and minimal medium. In EZ RDM, the gadE driven pathway has higher expression than lacUV5 (represented by AtoB) as seen before, Figure 40A. However, in the M9 MOPS medium, the expression profile was the inverse, Figure 40B. The gadE promoter had low expression while the lacUV5 expression was high (and roughly the same as the gadE in EZ RDM).

A

B

Figure 38. Production in minimal medium, M9 MOPS + 1% glucose of A)amorphadiene and B) bisabolene by the responsive mevalonate pathway. Measured at 50 hours, both cultures have IPTG added

73

Figure 39. Expression of RFP from the promoters in M9 MOPS + 1% glucose behaves similarly to EZ RDM + 1% glucose.

A

B

Figure 40. Protein data from A) EZ RDM + 1% glucose and B)M9 MOPS + 1% glucose for AtoB as a representative sample of the difference in protein between the strains in the different media.

The lower production in minimal medium was troubling. The plasmid stability and structure were tested after growth in the minimal medium to rule out degradation or loss of the plasmid. The plasmid retained the same size and cells remained viable on the antibiotic containing agar plates. The promoter region was also sequenced to rule out mutations which would terminate expression, but no mutations were found.

The media composition was also studied. The magnesium levels in M9 were lowered

from 2 mM to 100 M to increase expression from the rstA promoter and increase the chance

74

of toxicity, however no changes improved production. The composition of EZ RDM was also investigated because most of the production we measure occurred in stationary phase and the gadE promoter has high expression in stationary phase. EZ RDM is balanced for growth with 0.2% glucose; however we’ve added 1% glucose to increase the amount of production. With the extra glucose the other media components become limiting, resulting in large amounts of glucose remaining in stationary phase. This can be seen in Figure 31, where it takes 3 days to consume 10 g/L glucose in EZ RDM, but in Korz medium it is consumed in less than 24 hours (data not shown).

Using the RAMOS system (Guez, Müller, Danze, Büchs, & Jacques, 2008), the oxygen transfer rate of the cultures was measured. Using this information, we were able to monitor the limitations in the media. By adjusting the NH4 levels in the minimal medium, we were able to determine the optimal amount of NH4 to glucose ratio was roughly 0.45 mol NH4/mol C or 1:6 g (NH4)2SO4:g glucose (which agrees with previously reported media compositions (Neidhardt et al., 1974)). However, the ratio in EZ RDM with 1% glucose is ~0.17 mol NH4/mol C (depending if you count the NH4 and C from the added amino acids). The NH4 levels were lowered in the minimal medium to approximate the ratio in EZ RDM, however this also did not improve the production of PgadE-MevT-MBIS (data not shown).

Ultimately, the problem (although maybe not the only one) with PgadE-MevT-MBIS expression in minimal medium was attributed to the acclimation process (plating directly on minimal agar plates was also unproductive). Freezer stocks were saved each day of the acclimation progress (typically lasting over 4 days). Production was observed to decrease with each successive pass. We decided to forgo the typical acclimation process and grew our strains in a minimal medium, but added amino acids to the mixture to improve growth. The responsive pathway and PlacUV5-MevT-MBIS/pBIS were tested in KORZ medium + 1.5% glucose + 1% casein amino acids in the RAMOS flasks, Figure 41. The responsive pathway grew slightly slower, but the overall cell density was the same. The glucose and acetate profiles were the same throughout growth in the two strains. We observed no acetate accumulation (data not shown). The production in the two strains was similar until the second day of growth, but at least the responsive pathway was now producing (although no longer twice as much as PlacUV5-MevT-MBIS/pBIS). Interestingly, the glucose levels in the medium are below the limit of detection after 20 hours; however, the production continued to increase after this point. The use of colonies from the plate might not be ideal for reproducibility in a fermentor (necessitating the use of freezer plugs), but multiple colonies (tested up to four) grew and produced identically in the non-limiting conditions of the RAMOS flasks. At this point, although the conditions weren’t perfect to show the ultimate utility of the responsive mevalonate pathway, the strains were transferred to a fed-batch fermentation to test its performance at a larger scale.

75

A

B

Figure 41. Growth of strains in RAMOS system in KORZ media + 1.5% glucose + 1% Casein amino acids. A) Growth

as a function of time and B)Production of Bisabolene during growth.

The use of amino acids in the fermentor proved to be challenging and ultimately a bad idea. Initially, the addition of amino acids was not combined with a similar reduction in (NH4)2SO4 in the batch or feed. This resulted in the cells not growing up to high cell densities, presumably from ammonia toxicity. We also observed problems with higher amounts of precipitation, thus a solution without the addition of amino acids was in order. The acclimation problem seemed to be a good choice to target, and stocks were taken at earlier acclimating passes. We found a point (using fresh transformants helped) where the cells still grew well and produced similarly in shake flasks. The production of this strain in Korz medium without amino acids was roughly the same as with amino acids in both PgadE-MevT-MBIS/PrstA-BIS and PlacUV5-MevT-MBIS/pBIS, Figure 42. Similar to the tests above, PgadE-MevT-MBIS produced similar amounts to PlacUV5 in the first 20 hours, but then PlacUV5 kept increasing.

76

Figure 42. Production of Bisabolene from the acclimated strains in Korz medium (15 g/L glucose, 4 g/L (NH4)2SO4 with or without 10 g/L casein amino acids in the Ramos shake flasks (300 rpm, 20 ml media, 2 ml dodecane).

Discussion

While the work on scaling the responsive pathway to a fed-batch process is on-going, it appears that many of the benefits seen in shake flask in EZ RDM are absent in the defined minimal media typically used in fermentors. I believe this is due to the nature of the gadE promoter being highly expressed in stationary phase when (due to our excess glucose and media imbalance) production can occur. Also troubling/interesting is why the lacUV5 promoter expresses so poorly in EZ RDM, but in minimal media the expression patterns are the inverse. Yet the pathway seems to play a role in this behavior as expression of RFP from either promoter in either media behaved identically. Much more work can be done on this aspect to see if the gadE promoter could be modified to work better in industrial conditions or if another responsive promoter would work better.

Also of note is that the best way to find a robust responsive promoter that works in industrial conditions is to do your initial screening under industrial conditions. It remains to be seen if the responsive promoter could also minimize acetate accumulation in the fed-batch process and allow faster glucose feeding and higher productivity rates. But from its performance in shake flasks, this appears unlikely without further improvements.

The use of the RAMOS shake flasks greatly improved our understanding of the different growth phases and the oxygen transfer and hopefully will allow us to scale up the process easier. With it, we were able to monitor the much worse oxygen transfer rates in EZ RDM due to the nutrient limitations as well as find the optimal level of nitrogen to maximize cell growth. By just monitoring growth, you could observe the increase in cell densities with additional

77

ammonia, but you lack the real time monitoring of the reduced oxygen transfer when the cell becomes limited. The highest production of amorphadiene in literature occurred under nitrogen restricted conditions (which would be similar to that seen in EZ RDM), but we have yet to try this ourselves in the fed-batch conditions. The idea behind this is the formation of biomass is a big sink for the carbon, so if you restrict biomass formation by limiting nitrogen, more carbon is available for your product. There are several tools available to monitor the ammonia levels in the medium off-line during growth (enzyme assay kits or analytically with a tool similar to the Nova Bioanalyzer), but we have yet to invest in these options. Most of our processes to date have been with excess ammonia, so this lever can still be pulled to improve our process.

78

Chapter 6: Perspectives and Future Directions

Table of Contents Conclusions ............................................................................................................................................. 78

Understanding the Toxicity ................................................................................................................. 78

Responsive Promoter .......................................................................................................................... 79

Future Work ............................................................................................................................................ 80

Conclusions The goals of this project were to understand the cause of the isoprenyl diphosphate toxicity, use that knowledge to build responsive pathways, and finally to test the responsive pathways in industrial conditions where their benefit of not requiring inducers would serve its greatest purpose. The project was successful in that we realized a new way to regulate heterologous pathways without needing to induce the cultures. Ultimately, the method might prove ineffective at large scale conditions because ideally the screening process should occur in conditions which more closely mimic the final industrial conditions. Regardless, the responsive pathways proved effective in improving isoprenoid production over standard inducible promoters by minimizing acetate accumulation.

While I accomplished many of the goals I set out to achieve, there remains some work to complete the understandings of the toxicity and there is always room for improvement in pathway engineering with the promoters I identified. Some suggestions for additional work on these projects will be discussed as well as future directions for this work and what I see as the best way to re-engineer responsiveness into the pathway.

Understanding the Toxicity

In Chapter 2, we described microarray experiments to identify genes that responded to the accumulation of FPP. The microarrays were useful when the toxicity was apparent by causing differences in growth, but many of the genes identified were hypothetical or with unknown function or general stress responders. Microarrays are limited somewhat in that it is difficult to determine what changes are in direct response to your target molecule versus which are downstream effects of other conditions in the cell. With more time points, the temporal pattern of expression becomes clearer, but in this experiment, earlier time points when there was no change in growth showed very few differentially expressed genes.

A common follow-up experiment to microarray studies involves selecting targeted genes to upregulate or downregulate to see how their expression affects the toxicity. This approach was not taken in this work, because ultimately the answer to the toxicity problem is to alter the media or to express a functional terpene synthase to prevent their accumulation. The microarrays were useful in highlighting the PhoPQ regulon which helped contribute to the toxicity hypothesis of the interplay between ATP, Mg, and FPP. This led to testing the ATP synthase knockouts which accentuated the toxicity. Enzyme assays using luciferase showed that FPP inhibited this ATP-using enzyme, but further enzyme assays to show the inhibition of

79

metabolic enzymes were not completed. This remains as one of the tasks to show how FPP could reduce growth.

The other task is tracking down the accumulation of FPP and other related metabolites. Our efforts to measure FPP intracellularly, even in the strain with the inactive terpene synthase, have been difficult, only measuring just above the detection limit of the instrument. We have previously demonstrated (V. J. J. Martin, Pitera, Withers, J. D. Newman, & Jay D Keasling, 2003) the ability to measure the accumulation of these molecules as their phosphatase-treated alcohols, but it would be better to detect the actual molecule. If the extraction process proves to be too difficult for the unstable diphosphate molecule, we could return to the use of 14C mevalonate. In a strain with the inactive terpene synthase fed the radio-labeled mevalonate, we would fractionate the cell pellet into its cellular components: lipids, proteins, cytosol, etc. Using scintillation counters, we could track the radioactivity to the specific component where FPP accumulates. Without using radioactivity we can study the targeted cellular component with mass spectrometry to identify the exact mass of FPP.

Responsive Promoter

In chapter 4, we found that the gadE promoter produced more than two times the amount of amorphadiene as the lacUV5 promoter. While the proteomics analysis showed that pathway expression was much stronger than the lacUV5 promoter, their strengths were equal when expressing RFP. Yet, the gadE promoter was downregulated in response to FPP accumulation in the microarray analysis. The exact cause of the improvement and dynamics of the system are somewhat muddled. Does the gadE promoter work well because it lowers the pathway expression (once FPP accumulates), thus finding optimal levels of enzyme expression in each cell? Or does it perform better because it expresses better in stationary phase and when coupled with the pathway its expression is amplified? To attempt to address this, the responsive promoters could be used with the inactive pathway enzymes, coupled with analysis of transcript, protein, and metabolite levels. The levels with the inactive enzymes could be compared to the levels with the active enzymes to determine if they are also downregulated (or upregulated in the case of the rstA promoter) as FPP accumulates. Ideally, this would be done in a time course to determine when the expression changes and would include the non-responsive, inducible promoters.

While the microarrays were a great screen to identify potential sensor targets, with high throughput screening and liquid handling robotics becoming more tractable, an entire promoter library could be used to exhaustively search all possibilities. The promoter library coupled with the fluorescent reporter could give each gene’s dynamic response to accumulating FPP, giving both the timing and magnitude of their response (Zaslaver et al., 2006). While the fluorescent output is easier to do in high-throughput, using the entire library 5’ of the mevalonate pathway enzymes and measuring amorphadiene production could be accomplished, albeit slower. Growth, extractions, and GC analysis protocols directly from the deep well 96 well plates would need to be developed, but would greatly improve the throughput of the search.

80

Without expanding the search, the gadE and rstA promoters identified in this work, could also be improved. Despite the graded response to increasing FPP levels by the rstA promoter, it was weakest of the studied promoters by far. The lower strength led to decreased levels of amorphadiene production compared to the stronger trc promoter. While efforts to increase the translation strength by increasing the ribosome binding site did not work, it should be possible to engineer the promoter region itself to improve expression strength (Alper, Fischer, Nevoigt, & Gregory Stephanopoulos, 2005; Cox, Surette, & Michael B Elowitz, 2007). The binding strength of RNA polymerase can be improved to increase transcription by mutating the -10 and -35 sites. Alternatively, the PhoPQ binding site could be moved to a different promoter to make a hybrid promoter. The gadE promoter and other stationary phase promoters performed the best in EZ RDM. While constitutive promoters of varying strengths performed much worse, I believe testing the constitutive promoter library constructed based on stationary phase promoters (Miksch et al., 2005) would be an ideal way to tune expression strength and determine if the stationary phase expression was the most important factor to PgadE’s success.

The transition to minimal media and fed-batch fermentation was the first attempt to test the robustness of the approach. The production could also be tested in other stressful conditions to determine what other conditions disrupt pathway expression from the promoters. Another anticipated change in the process for future biofuels production is the consumption of hydrolyzed biomass as the carbon source. The feedstock and deconstruction groups of JBEI have produced small batches, so testing this material would be an easy next step.

Future Work The benefits of using a responsive promoter to drive the heterologous pathway have been demonstrated. The approach taken to use endogenous promoters that respond to the FPP accumulation worked well, but ultimately the response is indirect and thus could be delayed or triggered by alternate, unrelated stresses. Ideally, there would be direct detection of FPP, combined with actuation of transcription or translation, this approach is briefly discussed in Chapter 4. An aptamer or transcription factor could be engineered to recognize FPP to give direct detection and control. There are examples of aptamers recognizing phosphorylated molecules as well as thiamine, a vitamin whose synthesis pathway branches off of the native DXP pathway (Noeske et al., 2006). There isn’t a known transcription factor that recognizes the isoprenyl diphosphate molecules; however, work by fellow graduate student, Howard Chou, has shown the ability to fuse the binding domain of the IPP recognizing idi to the DNA binding domain of araC to create a sensor for IPP. Using this chimeric transcription factor and its corresponding promoters to drive the mevalonate pathway could give a direct link between the isoprenyl diphosphates’ accumulation and transcription of the pathway enzymes.

81

Bibliography

Ajikumar, P. K., Xiao, W.-H., Tyo, K. E. J., Wang, Y., Simeon, F., Leonard, E., et al. (2010).

Isoprenoid Pathway Optimization for Taxol Precursor Overproduction in Escherichia coli.

Science, 330(6000), 70-74. doi: 10.1126/science.1191652.

Alper, H., Fischer, C., Nevoigt, E., & Stephanopoulos, Gregory. (2005). Tuning genetic control

through promoter engineering. Proceedings of the National Academy of Sciences of the

United States of America, 102(36), 12678-83. doi: 10.1073/pnas.0504604102.

Andersen, J., Sternberg, C., & Poulsen, L. (1998). New unstable variants of green fluorescent

protein for studies of transient gene expression in bacteria. Applied and, 64(6), 2240-6.

Anderson, J. C., Clarke, E. J., Arkin, A. P., & Voigt, C. a. (2006). Environmentally controlled

invasion of cancer cells by engineered bacteria. Journal of molecular biology, 355(4), 619-

27. doi: 10.1016/j.jmb.2005.10.076.

Anderson, J. C., Dueber, J. E., Leguia, M., Wu, G. C., Goler, J. A., Arkin, A. P., et al. (2010).

BglBricks: A flexible standard for biological part assembly. Journal of biological

engineering, 4(1), 1. doi: 10.1186/1754-1611-4-1.

Anderson, J. C., Voigt, C. A., & Arkin, A. P. (2007). Environmental signal integration by a

modular AND gate. Molecular systems biology, 3, 133. doi: 10.1038/msb4100173.

Anthony, J. R., Anthony, L. C., Nowroozi, F., Kwon, G., Newman, J. D., & Keasling, Jay D.

(2009). Optimization of the mevalonate-based isoprenoid biosynthetic pathway in

Escherichia coli for production of the anti-malarial drug precursor amorpha-4,11-diene.

Metabolic engineering, 11(1), 13-9. doi: 10.1016/j.ymben.2008.07.007.

Babiskin, A. H., & Smolke, Christina D. (2011). A synthetic library of RNA control modules for

predictable tuning of gene expression in yeast. Molecular Systems Biology, 7. EMBO and

Macmillan Publishers Limited. doi: 10.1038/msb.2011.4.

Bajad, S. U., Lu, W., Kimball, E. H., Yuan, J., Peterson, C., & Rabinowitz, J. D. (2006).

Separation and quantitation of water soluble cellular metabolites by hydrophilic interaction

chromatography-tandem mass spectrometry. Journal of chromatography. A, 1125(1), 76-88.

doi: 10.1016/j.chroma.2006.05.019.

Barbirato, F., Grivet, J., Soucaille, P., & Bories, A. (1996). 3-Hydroxypropionaldehyde, an

inhibitory metabolite of glycerol fermentation to 1,3-propanediol by enterobacterial species.

Appl. Envir. Microbiol., 62(4), 1448-1451.

Basu, S., Gerchman, Y., Collins, C. H., Arnold, F. H., & Weiss, R. (2005). A synthetic

multicellular system for programmed pattern formation. Nature, 434(7037), 1130-4. doi:

10.1038/nature03461.

82

Benjamini, Y., & Hochberg, Y. (1995). Controlling the False Discovery Rate: A Practical and

Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society. Series B

(Methodological), 57(1), 289-300. Blackwell Publishers.

Bolstad, B. M., Collin, F., Simpson, K. M., Irizarry, R. a, & Speed, T P. (2004). Experimental

design and low-level analysis of microarray data. International review of neurobiology, 60,

25-58. doi: 10.1016/S0074-7742(04)60002-X.

Bolstad, B. M., Irizarry, R. a, Astrand, M., & Speed, T P. (2003). A comparison of normalization

methods for high density oligonucleotide array data based on variance and bias.

Bioinformatics (Oxford, England), 19(2), 185-93.

Bougdour, A., Cunning, C., Baptiste, P. J., Elliott, T., & Gottesman, S. (2008). Multiple

pathways for regulation of σS (RpoS) stability in Escherichia coli via the action of multiple

anti-adaptors. Molecular Microbiology, 68(2), 298-313. doi: 10.1111/j.1365-

2958.2008.06146.x.

Boyle, P. M., & Silver, P. A. (2009). Harnessing natureʼs toolbox: regulatory elements for

synthetic biology. Journal of the Royal Society, Interface / the Royal Society, 6 Suppl

4(Suppl_4), S535-46. doi: 10.1098/rsif.2008.0521.focus.

Brynildsen, M. P., & Liao, James C. (2009). An integrated network approach identifies the

isobutanol response network of Escherichia coli. Molecular systems biology, 5, 277. doi:

10.1038/msb.2009.34.

Bukau, B. (1993). Regulation of the Escherichia coli heat-shock response. Molecular

Microbiology, 9(4), 671-680. doi: 10.1111/j.1365-2958.1993.tb01727.x.

Chang, D.-E., Smalley, D. J., & Conway, T. (2002). Gene expression profiling of Escherichia

coli growth transitions: an expanded stringent response model. Molecular microbiology,

45(2), 289-306.

Chang, M. C. Y., Eachus, R. A., Trieu, W., Ro, D.-K., & Keasling, Jay D. (2007). Engineering

Escherichia coli for production of functionalized terpenoids using plant P450s. Nature

chemical biology, 3(5), 274-7. Nature Publishing Group. doi: 10.1038/nchembio875.

Chen, A., Kroon, P. A., & Poulter, C D. (1994). Isoprenyl diphosphate synthases: protein

sequence comparisons, a phylogenetic tree, and predictions of secondary structure. Protein

science : a publication of the Protein Society, 3(4), 600-7. doi: 10.1002/pro.5560030408.

Chin, C.-S., Chubukov, V., Jolly, E. R., DeRisi, J., & Li, H. (2008). Dynamics and design

principles of a basic regulatory architecture controlling metabolic pathways. PLoS biology,

6(6), e146. doi: 10.1371/journal.pbio.0060146.

83

Cho, B.-K., Knight, E. M., & Palsson, B. Ø. (2008). Genomewide identification of protein

binding locations using chromatin immunoprecipitation coupled with microarray. Methods

in molecular biology (Clifton, N.J.), 439, 131-45. doi: 10.1007/978-1-59745-188-8_9.

Cho, B.-K., Zengler, K., Qiu, Y., Park, Y. S., Knight, E. M., Barrett, C. L., et al. (2009). The

transcription unit architecture of the Escherichia coli genome. Nature biotechnology,

27(11), 1043-9. Nature Publishing Group. doi: 10.1038/nbt.1582.

Chou, C.-C., Chen, C.-H., Lee, T.-T., & Peck, K. (2004). Optimization of probe length and the

number of probes per gene for optimal microarray analysis of gene expression. Nucleic

acids research, 32(12), e99. doi: 10.1093/nar/gnh099.

Churchill, G. a. (2002). Fundamentals of experimental design for cDNA microarrays. Nature

genetics, 32 Suppl(December), 490-5. doi: 10.1038/ng1031.

Clarke, E. J., & Voigt, C. A. (2011). Characterization of combinatorial patterns generated by

multiple two-component sensors in E. coli that respond to many stimuli. Biotechnology and

bioengineering, 108(3), 666-75. doi: 10.1002/bit.22966.

Colantuoni, C., Henry, G., Zeger, S., & Pevsner, J. (2002). SNOMAD (Standardization and

NOrmalization of MicroArray Data): web-accessible gene expression data analysis.

Bioinformatics, 18(11), 1540-1541.

Coornaert, A., Lu, A., Mandin, P., Springer, M., Gottesman, S., & Guillier, M. (2010). MicA

sRNA links the PhoP regulon to cell envelope stress. Molecular microbiology, 76(2), 467-

79. doi: 10.1111/j.1365-2958.2010.07115.x.

Courcelle, J., Khodursky, A., Peter, B., Brown, Patrick O., & Hanawalt, P. C. (2001).

Comparative Gene Expression Profiles Following UV Exposure in Wild-Type and SOS-

Deficient Escherichia coli. Genetics, 158(1), 41-64.

Covert, M. W., Knight, E. M., Reed, J. L., Herrgard, M. J., & Palsson, B. O. (2004). Integrating

high-throughput and computational data elucidates bacterial networks. Nature, 429(6987),

92-6. doi: 10.1038/nature02456.

Cowan, J. A. (2002). Structural and catalytic chemistry of magnesium-dependent enzymes.

BioMetals, 15(3), 225-235-235. Springer Netherlands. doi: 10.1023/A:1016022730880.

Cox, R. S., Surette, M. G., & Elowitz, Michael B. (2007). Programming gene expression with

combinatorial promoters. Molecular systems biology, 3, 145. doi: 10.1038/msb4100187.

Datsenko, K. a, & Wanner, B. L. (2000). One-step inactivation of chromosomal genes in

Escherichia coli K-12 using PCR products. Proceedings of the National Academy of

Sciences of the United States of America, 97(12), 6640-5. doi: 10.1073/pnas.120163297.

84

De Mey, M., Maertens, J., Boogmans, S., Soetaert, W. K., Vandamme, E. J., Cunin, R., et al.

(2010). Promoter knock-in: a novel rational method for the fine tuning of genes. BMC

biotechnology, 10(1), 26. doi: 10.1186/1472-6750-10-26.

Dehal, P. S., Joachimiak, M. P., Price, M. N., Bates, J. T., Baumohl, J. K., Chivian, D., et al.

(2010). MicrobesOnline: an integrated portal for comparative and functional genomics.

Nucleic acids research, 38(Database issue), D396-400. doi: 10.1093/nar/gkp919.

DeLuca, M., & McElroy, W. D. (1978). Purification and Properties of Firefly luciferase.

Methods in Enzymology, 57, 3-15. Elsevier. doi: 10.1016/0076-6879(78)57003-1.

DeRisi, J. L., Iyer, V. R., & Brown, P O. (1997). Exploring the metabolic and genetic control of

gene expression on a genomic scale. Science (New York, N.Y.), 278(5338), 680-6.

Dharmadi, Y., & Gonzalez, R. (2004). DNA microarrays: experimental issues, data analysis, and

application to bacterial systems. Biotechnology progress, 20(5), 1309-24. doi:

10.1021/bp0400240.

Dietrich, J. A., McKee, A. E., & Keasling, Jay D. (2010). High-Throughput Metabolic

Engineering: Advances in Small-Molecule Screening and Selection. Annual review of

biochemistry. Annual Reviews. doi: 10.1146/annurev-biochem-062608-095938.

Dombek, K. M., & Ingram, L. O. (1986). Magnesium limitation and its role in apparent toxicity

of ethanol during yeast fermentation. Applied and environmental microbiology, 52(5), 975-

81.

Edgar, R. (2002). Gene Expression Omnibus: NCBI gene expression and hybridization array

data repository. Nucleic Acids Research, 30(1), 207-210. doi: 10.1093/nar/30.1.207.

Eguchi, Y., Okada, T., Minagawa, S., Oshima, T., Mori, H., Yamamoto, K., et al. (2004). Signal

transduction cascade between EvgA/EvgS and PhoP/PhoQ two-component systems of

Escherichia coli. Journal of bacteriology, 186(10), 3006-14.

Eisen, M. B., Spellman, P. T., Brown, P O, & Botstein, D. (1998). Cluster analysis and display of

genome-wide expression patterns. Proceedings of the National Academy of Sciences of the

United States of America, 95(25), 14863-8.

El Ghachi, M., Derbise, A., Bouhss, A., & Mengin-Lecreulx, D. (2005). Identification of

multiple genes encoding membrane proteins with undecaprenyl pyrophosphate phosphatase

(UppP) activity in Escherichia coli. The Journal of biological chemistry, 280(19), 18689-95.

doi: 10.1074/jbc.M412277200.

Elowitz, M B, & Leibler, S. (2000). A synthetic oscillatory network of transcriptional regulators.

Nature, 403(6767), 335-8. doi: 10.1038/35002125.

85

Ewald, J. C., Heux, S., & Zamboni, N. (2009). High-throughput quantitative metabolomics:

workflow for cultivation, quenching, and analysis of yeast in a multiwell format. Analytical

chemistry, 81(9), 3623-9. American Chemical Society. doi: 10.1021/ac900002u.

Faith, J. J., Hayete, B., Thaden, J. T., Mogno, I., Wierzbowski, J., Cottarel, G., et al. (2007).

Large-scale mapping and validation of Escherichia coli transcriptional regulation from a

compendium of expression profiles. (A. Levchenko, Ed.)PLoS biology, 5(1), e8. Public

Library of Science. doi: 10.1371/journal.pbio.0050008.

Farmer, W. R., & Liao, James C. (1997). Reduction of aerobic acetate production by Escherichia

coli. Appl. Envir. Microbiol., 63(8), 3205-3210.

Farmer, W. R., & Liao, James C. (2000). Improving lycopene production in Escherichia coli by

engineering metabolic control. Nature biotechnology, 18(5), 533-7. doi: 10.1038/75398.

Farmer, W. R., & Liao, James C. (2001). Acetate-inducible protein overexpression from the

glnAp2 promoter of Escherichia coli. Biotechnology and bioengineering, 75(5), 504-9.

Farr, S. B., & Kogoma, T. (1991). Oxidative stress responses in Escherichia coli and Salmonella

typhimurium. Microbiological reviews, 55(4), 561-85.

Ferrández, A., García, J. L., & Díaz, E. (2000). Transcriptional regulation of the divergent paa

catabolic operons for phenylacetic acid degradation in Escherichia coli. The Journal of

biological chemistry, 275(16), 12214-22.

Fiil, A., & Branton, D. (1969). Changes in the plasma membrane of Escherichia coli during

magnesium starvation. Journal of bacteriology, 98(3), 1320-7.

Fung, E., Wong, W., Suen, J., Bulter, T., Lee, S., & Liao, J. (2005). A synthetic gene–metabolic

oscillator. Nature, 435(May), 118-122. doi: 10.1038/nature03503.1.

Ge, Y., Dudoit, S., & Speed, Terence P. (2003). Resampling-based multiple testing for

microarray data analysis. Test, 12(1), 1-77. doi: 10.1007/BF02595811.

Geiss, G. K., Bumgarner, R. E., Birditt, B., Dahl, T., Dowidar, N., Dunaway, D. L., et al. (2008).

Direct multiplexed measurement of gene expression with color-coded probe pairs. Nature

biotechnology, 26(3), 317-25. Nature Publishing Group. doi: 10.1038/nbt1385.

Gill, R T, DeLisa, M. P., Valdes, J. J., & Bentley, W. E. (2001). Genomic analysis of high-cell-

density recombinant Escherichia coli fermentation and “cell conditioning” for improved

recombinant protein yield. Biotechnology and bioengineering, 72(1), 85-95.

Gill, R T, Wildt, S., Yang, Y. T., Ziesman, S., & Stephanopoulos, G. (2002). Genome-wide

screening for trait conferring genes using DNA microarrays. Proceedings of the National

Academy of Sciences of the United States of America, 99(10), 7033-8. doi:

10.1073/pnas.102154799.

86

Gonzalez, R., Tao, H., Purvis, J. E., York, S. W., Shanmugam, K. T., & Ingram, L. O. (2003).

Gene array-based identification of changes that contribute to ethanol tolerance in

ethanologenic Escherichia coli: comparison of KO11 (parent) to LY01 (resistant mutant).

Biotechnology progress, 19(2), 612-23. doi: 10.1021/bp025658q.

Groban, E. S., Clarke, E. J., Salis, H. M., Miller, S. M., & Voigt, C. a. (2009). Kinetic buffering

of cross talk between bacterial two-component sensors. Journal of molecular biology,

390(3), 380-93. Elsevier Ltd. doi: 10.1016/j.jmb.2009.05.007.

Groisman, E. A. (2001). The Pleiotropic Two-Component Regulatory System PhoP-PhoQ †.

Journal of Bacteriology, 183(6), 1835-1842. doi: 10.1128/JB.183.6.1835.

Guez, J. S., Müller, C. H., Danze, P. M., Büchs, J., & Jacques, P. (2008). Respiration activity

monitoring system (RAMOS), an efficient tool to study the influence of the oxygen transfer

rate on the synthesis of lipopeptide by Bacillus subtilis ATCC6633. Journal of

biotechnology, 134(1-2), 121-6. doi: 10.1016/j.jbiotec.2008.01.003.

Hammer, K., Mijakovic, I., & Jensen, P. R. (2006). Synthetic promoter libraries--tuning of gene

expression. Trends in biotechnology, 24(2), 53-5. doi: 10.1016/j.tibtech.2005.12.003.

Hinson, D. D., Chambliss, K. L., Toth, M. J., Tanaka, R. D., & Gibson, K. M. (1997). Post-

translational regulation of mevalonate kinase by intermediates of the cholesterol and

nonsterol isoprene biosynthetic pathways. Journal of lipid research, 38(11), 2216-23.

Hoffmann, F., & Rinas, U. (2004). Stress Induced by Recombinant Protein Production in

Escherichia coli - Physiological Stress Responses in Bioprocesses -. Advances in

Biochemical Engineering/Biotechnology, 89, 73-92-92. Springer Berlin / Heidelberg. doi:

10.1007/b93994.

Holtz, W. J., & Keasling, Jay D. (2010). Engineering static and dynamic control of synthetic

pathways. Cell, 140(1), 19-23. doi: 10.1016/j.cell.2009.12.029.

Hua, Q., Yang, C., Oshima, T., Mori, H., & Shimizu, K. (2004). Analysis of gene expression in

Escherichia coli in response to changes of growth-limiting nutrient in chemostat cultures.

Applied and environmental microbiology, 70(4), 2354-66.

Hung, S.-pin, Baldi, P., & Hatfield, G. W. (2002). Global gene expression profiling in

Escherichia coli K12. The effects of leucine-responsive regulatory protein. The Journal of

biological chemistry, 277(43), 40309-23. doi: 10.1074/jbc.M204044200.

Huvet, M., Toni, T., Sheng, X., Thorne, T., Jovanovic, G., Engl, C., et al. (2011). The evolution

of the phage shock protein response system: interplay between protein function, genomic

organization, and system function. Molecular biology and evolution, 28(3), 1141-55. doi:

10.1093/molbev/msq301.

87

Izutsu, K., Wada, A., & Wada, C. (2001). Expression of ribosome modulation factor (RMF) in

Escherichia coli requires ppGpp. Genes to cells : devoted to molecular & cellular

mechanisms, 6(8), 665-76.

Jeude, M., Dittrich, B., Niederschulte, H., Anderlei, T., Knocke, C., Klee, D., et al. (2006). Fed-

batch mode in shake flasks by slow-release technique. Biotechnology and bioengineering,

95(3), 433-45. doi: 10.1002/bit.21012.

Jones, K. L., Kim, S. W., & Keasling, J D. (2000). Low-copy plasmids can perform as well as or

better than high-copy plasmids for metabolic engineering of bacteria. Metabolic

engineering, 2(4), 328-38. doi: 10.1006/mben.2000.0161.

Joshua, C. J., Dahl, R., Benke, P. I., & Keasling, Jay D. (2011). Absence of Diauxie during

Simultaneous Utilization of Glucose and Xylose by Sulfolobus acidocaldarius. Journal of

bacteriology, 193(6), 1293-1301. doi: 10.1128/JB.01219-10.

Kang, Z., Wang, Q., Zhang, H., & Qi, Q. (2008). Construction of a stress-induced system in

Escherichia coli for efficient polyhydroxyalkanoates production. Applied microbiology and

biotechnology, 79(2), 203-8. Springer Berlin / Heidelberg. doi: 10.1007/s00253-008-1428-z.

Karp, P. (1996). EcoCyc: an encyclopedia of Escherichia coli genes and metabolism. Nucleic

Acids Research, 24(1), 32-39. doi: 10.1093/nar/24.1.32.

Kato, a, Tanabe, H., & Utsumi, R. (1999). Molecular characterization of the PhoP-PhoQ two-

component system in Escherichia coli K-12: identification of extracellular Mg2+-responsive

promoters. Journal of bacteriology, 181(17), 5516-20.

Kizer, L., Pitera, D. J., Pfleger, B. F., & Keasling, Jay D. (2008). Application of functional

genomics to pathway optimization for increased isoprenoid production. Applied and

environmental microbiology, 74(10), 3229-41. doi: 10.1128/AEM.02750-07.

Kobayashi, H., Kaern, M., Araki, M., Chung, K., Gardner, T. S., Cantor, C. R., et al. (2004).

Programmable cells: interfacing natural and engineered gene networks. Proceedings of the

National Academy of Sciences of the United States of America, 101(22), 8414-9. doi:

10.1073/pnas.0402940101.

Korz, D. (1995). Simple fed-batch technique for high cell density cultivation of Escherichia coli.

Journal of Biotechnology, 39(1), 59-65. doi: 10.1016/0168-1656(94)00143-Z.

Kreil, D. P., Russell, R. R., & Russell, S. (2006). Microarray oligonucleotide probes. Methods in

enzymology, 410, 73-98. doi: 10.1016/S0076-6879(06)10004-X.

Kurland, C. G., & Dong, Henjiang. (1996). Bacterial growth inhibition by overproduction of

protein. Molecular Microbiology, 21(1), 1-4. doi: 10.1046/j.1365-2958.1996.5901313.x.

88

Kuznetsova, E., Proudfoot, M., Gonzalez, C. F., Brown, G., Omelchenko, M. V., Borozan, I., et

al. (2006). Genome-wide analysis of substrate specificities of the Escherichia coli haloacid

dehalogenase-like phosphatase family. The Journal of biological chemistry, 281(47),

36149-61. doi: 10.1074/jbc.M605449200.

Lange, R., & Hengge-Aronis, R. (1994). The cellular concentration of the sigma S subunit of

RNA polymerase in Escherichia coli is controlled at the levels of transcription, translation,

and protein stability. Genes & development, 8(13), 1600-12.

Lashkari, D. A. (1997). Yeast microarrays for genome wide parallel genetic and gene expression

analysis. Proceedings of the National Academy of Sciences, 94(24), 13057-13062. doi:

10.1073/pnas.94.24.13057.

Lee, C., Sun, W.-J., Burgess, B. W., Junker, B. H., Reddy, J., Buckland, B. C., et al. (1997).

Process optimization for large-scale production of TGF- ? -PE40 in recombinant

Escherichia coli : effect of medium composition and induction timing on protein expression.

Journal of Industrial Microbiology and Biotechnology, 18(4), 260-266. Springer Berlin /

Heidelberg. doi: 10.1038/sj.jim.2900382.

Lee, K. H., Cho, M. H., Chung, T., Chang, H. N., Lim, S. H., & Lee, J. (2003). Characterization

of an oxygen-dependent inducible promoter, the Escherichia coli nar promoter, in gram-

negative host strains. Biotechnology and bioengineering, 82(3), 271-7. doi:

10.1002/bit.10567.

Lee, M. L., Kuo, F. C., Whitmore, G. a, & Sklar, J. (2000). Importance of replication in

microarray gene expression studies: statistical methods and evidence from repetitive cDNA

hybridizations. Proceedings of the National Academy of Sciences of the United States of

America, 97(18), 9834-9.

Lee, S Y. (1996). High cell-density culture of Escherichia coli. Trends in biotechnology, 14(3),

98-105. doi: 10.1016/0167-7799(96)80930-9.

Lee, Sang Yup (Ed.). (2009). Systems Biology and Biotechnology of Escherichia coli. Springer

Netherlands. doi: 10.1007/978-1-4020-9394-4.

Lee, S. K., Chou, H., Ham, T. S., Lee, T. S., & Keasling, Jay D. (2008). Metabolic engineering

of microorganisms for biofuels production: from bugs to synthetic biology to fuels. Current

opinion in biotechnology, 19(6), 556-63. doi: 10.1016/j.copbio.2008.10.014.

Lee, S. K., & Keasling, Jay D. (2005). A propionate-inducible expression system for enteric

bacteria. Applied and environmental microbiology, 71(11), 6856-62. doi:

10.1128/AEM.71.11.6856-6862.2005.

Levskaya, A., Chevalier, A. A., Tabor, J. J., Simpson, Z. B., Lavery, L. A., Levy, M., et al.

(2005). Synthetic biology: engineering Escherichia coli to see light. Nature, 438(7067),

441-2. doi: 10.1038/nature04405.

89

Lin, H., Bennett, George N, & San, K.-Y. (2005). Metabolic engineering of aerobic succinate

production systems in Escherichia coli to improve process productivity and achieve the

maximum theoretical succinate yield. Metabolic engineering, 7(2), 116-27. doi:

10.1016/j.ymben.2004.10.003.

Lockhart, D. J., Dong, Helin, Byrne, M. C., Follettie, M. T., Gallo, M. V., Chee, M. S., et al.

(1996). Expression monitoring by hybridization to high density oligonucleotide arrays.

Nature Biotechnology, 14, 1675-1680.

Looger, L., Dwyer, M., Smith, J., & Hellinga, H. (2003). Computational design of receptor and

sensor proteins with novel functions. Nature, 423(May), 185-190. doi:

10.1038/nature01578.1.

Lusk, J. E., Williams, R. J., & Kennedy, E. P. (1968). Magnesium and the growth of Escherichia

coli. The Journal of biological chemistry, 243(10), 2618-24.

Lutz, R., & Bujard, H. (1997). Independent and tight regulation of transcriptional units in

Escherichia coli via the LacR/O, the TetR/O and AraC/I1-I2 regulatory elements. Nucleic

Acids Research, 25(6), 1203-1210. doi: 10.1093/nar/25.6.1203.

Lynch, M., Warnecke, T., & Gill, R. (2007). SCALEs: multiscale analysis of library enrichment.

Nature Methods, 4(1). doi: 10.1038/NMETH946.

Maguire, M. E., & Cowan, J. a. (2002). Magnesium chemistry and biochemistry. Biometals : an

international journal on the role of metal ions in biology, biochemistry, and medicine,

15(3), 203-10.

Maki, Y., Yoshida, H., & Wada, A. (2000). Two proteins, YfiA and YhbH, associated with

resting ribosomes in stationary phase Escherichia coli. Genes to cells : devoted to molecular

& cellular mechanisms, 5(12), 965-74.

Martin, V. J., Yoshikuni, Y., & Keasling, J D. (2001). The in vivo synthesis of plant

sesquiterpenes by Escherichia coli. Biotechnology and bioengineering, 75(5), 497-503.

Martin, V. J. J., Pitera, D. J., Withers, S. T., Newman, J. D., & Keasling, Jay D. (2003).

Engineering a mevalonate pathway in Escherichia coli for production of terpenoids. Nature

biotechnology, 21(7), 796-802. doi: 10.1038/nbt833.

Marvin, J. S., & Hellinga, H. W. (2001). Conversion of a maltose receptor into a zinc biosensor

by computational design. Proceedings of the National Academy of Sciences of the United

States of America, 98(9), 4955-60. doi: 10.1073/pnas.091083898.

Maury, J., Asadollahi, M., Møller, K., Clark, A., & Nielsen, J. (2005). Microbial isoprenoid

production: an example of green chemistry through metabolic engineering. Adv. Biochem

Engin/Biotechnol, 100, 19-51. doi: 10.1007/b136410.

90

MCCARTHY, B. (1962). The effects of magnesium starvation on the ribosome content of

Escherichia coli. Biochimica et Biophysica Acta (BBA) - Specialized Section on Nucleic

Acids and Related Subjects, 55(6), 880-889. doi: 10.1016/0926-6550(62)90345-6.

McCaskill, D., & Croteau, R. (1997). Prospects for the bioengineering of isoprenoid

biosynthesis. (R. G. Berger, W. Babel, H. W. Blanch, C. L. Cooney, S.-O. Enfors, K.-E. L.

Eriksson, et al., Eds.)Advances in Biochemical Engineering/Biotechnology, 55, 107-146-

146. Springer Berlin Heidelberg. doi: 10.1007/BFb0102060.

McDaniel, R., & Weiss, R. (2005). Advances in synthetic biology: on the path from prototypes

to applications. Current opinion in biotechnology, 16(4), 476-83. doi:

10.1016/j.copbio.2005.07.002.

Miksch, G., Bettenworth, F., Friehs, K., Flaschel, E., Saalbach, A., Twellmann, T., et al. (2005).

Libraries of synthetic stationary-phase and stress promoters as a tool for fine-tuning of

expression of recombinant proteins in Escherichia coli. Journal of biotechnology, 120(1),

25-37. doi: 10.1016/j.jbiotec.2005.04.027.

Miller, W. G., Leveau, J. H., & Lindow, S. E. (2000). Improved gfp and inaZ broad-host-range

promoter-probe vectors. Molecular plant-microbe interactions : MPMI, 13(11), 1243-50.

The American Phytopathological Society. doi: 10.1094/MPMI.2000.13.11.1243.

Minagawa, S., Ogasawara, H., Kato, A., Yamamoto, K., Eguchi, Y., Oshima, T., et al. (2003).

Identification and molecular characterization of the Mg2+ stimulon of Escherichia coli.

Journal of Bacteriology, 185(13), 3696-3702. doi: 10.1128/JB.185.13.3696.

Mukhopadhyay, A., He, Z., Alm, E. J., Arkin, A. P., Baidoo, E. E., Borglin, S. C., et al. (2006).

Salt stress in Desulfovibrio vulgaris Hildenborough: an integrated genomics approach.

Journal of bacteriology, 188(11), 4068-78. doi: 10.1128/JB.01921-05.

Nakamura, C. E., & Whited, G. M. (2003). Metabolic engineering for the microbial production

of 1,3-propanediol. Current Opinion in Biotechnology, 14(5), 454-459. doi:

10.1016/j.copbio.2003.08.005.

Neidhardt, F. C., Bloch, P. L., & Smith, D. F. (1974). Culture medium for enterobacteria.

Journal of bacteriology, 119(3), 736-47.

Neubauer, P., Hofmann, K., Holst, O., Mattiasson, B., & Kruschke, P. (1992). Maximizing the

expression of a recombinant gene in Escherichia coli by manipulation of induction time

using lactose as inducer. Applied Microbiology and Biotechnology, 36(6), 739-744-744.

Springer Berlin / Heidelberg. doi: 10.1007/BF00172185.

Neubauer, P., & Winter, J. (2001). Expression and fermentation strategies for recombinant

protein production in E. coli. In O.-W. Merten & E. F. of Biotechnology. (Eds.),

Recombinant protein production with prokaryotic and eukaryotic cells (p. 400). Springer.

91

Nicolaou, K. C., Yang, Z, Liu, J. J., Ueno, H., Nantermet, P. G., Guy, R. K., et al. (1994). Total

synthesis of taxol. Nature, 367(6464), 630-4. doi: 10.1038/367630a0.

Ninfa, A., Selinsky, S., Perry, N., Atkins, S., Song, Q. X., Mayo, A., et al. (2007). Using Two-

Component Systems and Other Bacterial Regulatory Factors for the Fabrication of

Synthetic Genetic Devices. Methods in Enzymology, 422(06), 488-512. doi: 10.1016/S0076-

6879(06)22025-1.

Noeske, J., Richter, C., Stirnal, E., Schwalbe, H., & Wöhnert, J. (2006). Phosphate-group

recognition by the aptamer domain of the thiamine pyrophosphate sensing riboswitch.

Chembiochem : a European journal of chemical biology, 7(9), 1451-6. doi:

10.1002/cbic.200600151.

Nomura, Y., & Yokobayashi, Y. (2007). Reengineering a natural riboswitch by dual genetic

selection. Journal of the American Chemical Society, 129(45), 13814-5. American

Chemical Society. doi: 10.1021/ja076298b.

Nuwaysir, E. F., Huang, W., Albert, T. J., Singh, J., Nuwaysir, K., Pitas, A., et al. (2002). Gene

expression analysis using oligonucleotide arrays produced by maskless photolithography.

Genome research, 12(11), 1749-55. doi: 10.1101/gr.362402.

Oh, M K, & Liao, J C. (2000). DNA microarray detection of metabolic responses to protein

overproduction in Escherichia coli. Metabolic engineering, 2(3), 201-9. doi:

10.1006/mben.2000.0149.

Oh, M.K., Cha, M. J., Lee, S. G., Rohlin, L., & Liao, J.C. (2006). Dynamic gene expression

profiling of Escherichia coli in carbon source transition from glucose to acetate. Journal of

microbiology and biotechnology, 16(4), 543–549. Korean Society for Applied

Microbiology.

Ow, D., Nissom, P., Philp, R., Oh, S., & Yap, M. (2006). Global transcriptional analysis of

metabolic burden due to plasmid maintenance in Escherichia coli DH5α during batch

fermentation. Enzyme and Microbial Technology, 39(3), 391-398. doi:

10.1016/j.enzmictec.2005.11.048.

PAPP-WALLACE, K. M., & MAGUIRE, M. E. (n.d.). Magnesium Transport and Magnesium

Homeostasis. In and D. U. A. Böck, R. Curtiss III, J. B. Kaper, P. D. Karp, F. C. Neidhardt,

T. Nyström, J. M. Slauch, C. L. Squires (Ed.), EcoSal—Escherichia coli and Salmonella:

Cellular and Molecular Biology. ASM Press. doi: 10.1128/ecosal.5.4.4.2.

Pfleger, B. F., Pitera, D. J., Smolke, Christina D, & Keasling, Jay D. (2006). Combinatorial

engineering of intergenic regions in operons tunes expression of multiple genes. Nature

biotechnology, 24(8), 1027-32. Nature Publishing Group. doi: 10.1038/nbt1226.

92

Pitera, D. J., Paddon, Chris J, Newman, J. D., & Keasling, Jay D. (2007). Balancing a

heterologous mevalonate pathway for improved isoprenoid production in Escherichia coli.


Redding-Johanson, A. M., Batth, T. S., Chan, R., Krupa, R., Szmidt, H. L., Adams, P. D., et al.

(2011). Targeted proteomics for metabolic pathway optimization: Application to terpene

production. Metabolic engineering. doi: 10.1016/j.ymben.2010.12.005.

Renninger, N., Newman, J., & Reiling, K. (2010). Fuel components, fuel compositions and

methods of making and using same. USPTO.

Richmond, C. S., Glasner, J. D., Mau, R., Jin, H., & Blattner, F. R. (1999). Genome-wide

expression profiling in Escherichia coli K-12. Nucleic acids research, 27(19), 3821-35.

Ro, D.-K., Paradise, E. M., Ouellet, M., Fisher, K. J., Newman, K. L., Ndungu, J. M., et al.

(2006). Production of the antimalarial drug precursor artemisinic acid in engineered yeast.

Nature, 440(7086), 940-3. doi: 10.1038/nature04640.

Rutherford, B. J., Dahl, R. H., Price, R. E., Szmidt, H. L., Benke, P. I., Mukhopadhyay, A., et al.

(2010). Functional genomic study of exogenous n-butanol stress in Escherichia coli.

Applied and environmental microbiology, 76(6), 1935-45. doi: 10.1128/AEM.02323-09.

Salis, H. M., Mirsky, E. A., & Voigt, C. A. (2009). Automated design of synthetic ribosome

binding sites to control protein expression. Nature biotechnology, 27(10), 946-50. Nature

Publishing Group. doi: 10.1038/nbt.1568.

Salis, H., Tamsir, A., & Voigt, C. (2009). Engineering bacterial signals and sensors.

Contributions to microbiology, 16, 194-225. doi: 10.1159/000219381.

Sauer, R. T. (1996). Signal Detection by the PhoQ Sensor-Transmitter. CHARACTERIZATION

OF THE SENSOR DOMAIN AND A RESPONSE-IMPAIRED MUTANT THAT

IDENTIFIES LIGAND-BINDING DETERMINANTS. Journal of Biological Chemistry,

271(43), 26630-26636. doi: 10.1074/jbc.271.43.26630.

Sawers, G., & Jarsch, M. (1996). Alternative regulation principles for the production of

recombinant proteins in Escherichia coli. Applied Microbiology and Biotechnology, 46(1),

1-9. Springer Berlin / Heidelberg. doi: 10.1007/s002530050775.

Sayed, A. K., & Foster, J. W. (2009). A 750 bp sensory integration region directs global control

of the Escherichia coli GadE acid resistance regulator. Molecular microbiology, 71(6),

1435-50. doi: 10.1111/j.1365-2958.2009.06614.x.

Schena, M., Shalon, D., Davis, R. W., & Brown, P. O. (1995). Quantitative Monitoring of Gene

Expression Patterns with a Complementary DNA Microarray. Science, 270(5235), 467-470.

doi: 10.1126/science.270.5235.467.

93

Sezonov, G., Joseleau-Petit, D., & DʼAri, R. (2007). Escherichia coli physiology in Luria-

Bertani broth. Journal of bacteriology, 189(23), 8746-9. doi: 10.1128/JB.01368-07.

Shamah, S. M., Healy, J. M., & Cload, S. T. (2008). Complex target SELEX. Accounts of

chemical research, 41(1), 130-8. American Chemical Society. doi: 10.1021/ar700142z.

Shiloach, J., & Fass, R. (2005). Growing E. coli to high cell density--a historical perspective on

method development. Biotechnology advances, 23(5), 345-57. doi:

10.1016/j.biotechadv.2005.04.004.

Skerker, J. M., Perchuk, B. S., Siryaporn, A., Lubin, E. A., Ashenberg, O., Goulian, M., et al.

(2008). Rewiring the specificity of two-component signal transduction systems. Cell,

133(6), 1043-54. doi: 10.1016/j.cell.2008.04.040.

Smolke, C D, Martin, V. J., & Keasling, J D. (2001). Controlling the metabolic flux through the

carotenoid pathway using directed mRNA processing and stabilization. Metabolic

engineering, 3(4), 313-21. doi: 10.1006/mben.2001.0194.

St John, A. C., & Goldberg, A. L. (1980). Effects of starvation for potassium and other inorganic

ions on protein degradation and ribonucleic acid synthesis in Escherichia coli. Journal of

bacteriology, 143(3), 1223-33.

Stead, M. B., Marshburn, S., Mohanty, B. K., Mitra, J., Castillo, L. P., Ray, D., et al. (2010).

Analysis of Escherichia coli RNase E and RNase III activity in vivo using tiling

microarrays. Nucleic acids research, 39(8), 3188-203. doi: 10.1093/nar/gkq1242.

Stricker, J., Cookson, S., Bennett, M. R., Mather, W. H., Tsimring, L. S., & Hasty, J. (2008). A

fast, robust and tunable synthetic gene oscillator. Nature, 456(7221), 516-9. Macmillan

Publishers Limited. All rights reserved. doi: 10.1038/nature07389.

Studier, F. (2005). Protein production by auto-induction in high-density shaking cultures. Protein

Expression and Purification, 41(1), 207-234. doi: 10.1016/j.pep.2005.01.016.

Sturn, A., Quackenbush, J., & Trajanoski, Z. (2002). Genesis: cluster analysis of microarray

data. Bioinformatics (Oxford, England), 18(1), 207-8.

Sánchez, A. M., Bennett, George N, & San, K.-Y. (2006). Batch culture characterization and

metabolic flux analysis of succinate-producing Escherichia coli strains. Metabolic

engineering, 8(3), 209-26. doi: 10.1016/j.ymben.2005.11.004.

Tabata, K., & Hashimoto, S.-I. (2004). Production of mevalonate by a metabolically-engineered

Escherichia coli. Biotechnology letters, 26(19), 1487-91. Springer Netherlands. doi:

10.1023/B:BILE.0000044449.08268.7d.

Tani, T. H., Khodursky, A., Blumenthal, R. M., Brown, Patrick O, & Matthews, R. G. (2002).

Adaptation to famine: a family of stationary-phase genes revealed by microarray analysis.

94

Proceedings of the National Academy of Sciences of the United States of America, 99(21),

13471-6. doi: 10.1073/pnas.212510999.

Thomassen, G. O. S., Weel-Sneve, R., Rowe, A. D., Booth, J. A., Lindvall, J. M., Lagesen, K., et

al. (2010). Tiling array analysis of UV treated Escherichia coli predicts novel differentially

expressed small peptides. PloS one, 5(12), e15356. doi: 10.1371/journal.pone.0015356.

Tsuruta, H., Paddon, Christopher J, Eng, D., Lenihan, J. R., Horning, T., Anthony, L. C., et al.

(2009). High-level production of amorpha-4,11-diene, a precursor of the antimalarial agent

artemisinin, in Escherichia coli. PloS one, 4(2), e4489. doi: 10.1371/journal.pone.0004489.

Tusher, V. G., Tibshirani, R., & Chu, G. (2001). Significance analysis of microarrays applied to

the ionizing radiation response. Proceedings of the National Academy of Sciences of the

United States of America, 98(9), 5116-21. doi: 10.1073/pnas.091062498.

Voigt, C. A. (2006). Genetic parts to program bacteria. Current opinion in biotechnology, 17(5),

548-57. doi: 10.1016/j.copbio.2006.09.001.

Wang, C. H., & Koch, A. L. (1978). Constancy of growth on simple and complex media. J.

Bacteriol., 136(3), 969-975.

Wang, Z., Gerstein, M., & Snyder, M. (2009). RNA-Seq: a revolutionary tool for

transcriptomics. Nature reviews. Genetics, 10(1), 57-63. doi: 10.1038/nrg2484.

Warnecke, T. E., Lynch, M. D., Karimpour-Fard, A, Lipscomb, M. L., Handke, P., Mills, T., et

al. (2010). Rapid dissection of a complex phenotype through genomic-scale mapping of

fitness altering genes. Metabolic engineering, 12(3), 241-50. doi:

10.1016/j.ymben.2009.12.002.

Warner, J. R., Reeder, P. J., Karimpour-Fard, Anis, Woodruff, L. B. A., & Gill, Ryan T. (2010).

Rapid profiling of a microbial genome using mixtures of barcoded oligonucleotides. Nature

biotechnology, 28(8), 856-862. Nature Publishing Group, a division of Macmillan

Publishers Limited. All Rights Reserved. doi: 10.1038/nbt.1653.

Weber, H., Polen, T., Heuveling, J., Wendisch, V. F., & Hengge, R. (2005). Genome-wide

analysis of the general stress response network in Escherichia coli: sigmaS-dependent

genes, promoters, and sigma factor selectivity. Journal of bacteriology, 187(5), 1591-603.

doi: 10.1128/JB.187.5.1591-1603.2005.

Wee, S., & Wilkinson, B. J. (1988). Increased outer membrane ornithine-containing lipid and

lysozyme penetrability of Paracoccus denitrificans grown in a complex medium deficient in

divalent cations. J. Bacteriol., 170(7), 3283-3286.

Weiner, L., Brissette, J. L., & Model, P. (1991). Stress-induced expression of the Escherichia

coli phage shock protein operon is dependent on sigma 54 and modulated by positive and

negative feedback mechanisms. Genes & development, 5(10), 1912-23.

95

Wieland, M., Benz, A., Klauser, B., & Hartig, J. S. (2009). Artificial ribozyme switches

containing natural riboswitch aptamer domains. Angewandte Chemie (International ed. in

English), 48(15), 2715-8. doi: 10.1002/anie.200805311.

Win, M. N., & Smolke, Christina D. (2007). A modular and extensible RNA-based gene-

regulatory platform for engineering cellular function. Proceedings of the National Academy

of Sciences of the United States of America, 104(36), 14283-8. doi:

10.1073/pnas.0703961104.

Winkler, W. C., Nahvi, A., Roth, A., Collins, J. A., & Breaker, R. R. (2004). Control of gene

expression by a natural metabolite-responsive ribozyme. Nature, 428(6980), 281-6. doi:

10.1038/nature02362.

Withers, S. T., Gottlieb, S. S., Lieu, B., Newman, J. D., & Keasling, Jay D. (2007). Identification

of isopentenol biosynthetic genes from Bacillus subtilis by a screening method based on

isoprenoid precursor toxicity. Applied and environmental microbiology, 73(19), 6277-83.

doi: 10.1128/AEM.00861-07.

Woodruff, L. B., & Gill, Ryan T. (2011). Engineering genomes in multiplex. Current opinion in

biotechnology. doi: 10.1016/j.copbio.2011.04.019.

Wouters, J., Oudjama, Y., Barkley, S. J., Tricot, C., Stalon, V., Droogmans, L., et al. (2003).

Catalytic mechanism of Escherichia coli isopentenyl diphosphate isomerase involves Cys-

67, Glu-116, and Tyr-104 as suggested by crystal structures of complexes with transition

state analogues and irreversible inhibitors. The Journal of biological chemistry, 278(14),

11903-8. doi: 10.1074/jbc.M212823200.

Yang, Y. T., Bennett, G N, & San, K. Y. (1999). Effect of inactivation of nuo and ackA-pta on

redistribution of metabolic fluxes in Escherichia coli. Biotechnology and bioengineering,

65(3), 291-7.

Yee, L., & Blanch, H. (1992). Recombinant protein expression in high cell density fed-batch

cultures of Escherichia coli. Nature Biotechnology, 10.

Yokobayashi, Y., Weiss, R., & Arnold, F. H. (2002). Directed evolution of a genetic circuit.

Proceedings of the National Academy of Sciences of the United States of America, 99(26),

16587-91. doi: 10.1073/pnas.252535999.

Yuan, L. Z., Rouvière, P. E., Larossa, R. A., & Suh, W. (2006). Chromosomal promoter

replacement of the isoprenoid pathway for enhancing carotenoid production in E. coli.


Zaslaver, A., Bren, A., Ronen, M., Itzkovitz, S., Kikoin, I., Shavit, S., et al. (2006). A

comprehensive library of fluorescent transcriptional reporters for Escherichia coli. Nature

methods, 3(8), 623-8. doi: 10.1038/nmeth895.

96

Zhang, D., & Poulter, C D. (1993). Analysis and purification of phosphorylated isoprenoids by

reversed-phase HPLC. Analytical biochemistry, 213(2), 356-61.

Zheng, D., Constantinidou, C., Hobman, J. L., & Minchin, S. D. (2004). Identification of the

CRP regulon using in vitro and in vivo transcriptional profiling. Nucleic acids research,

32(19), 5874-93. doi: 10.1093/nar/gkh908.

Zheng, M., Wang, X., Templeton, L. J., Smulski, D. R., LaRossa, R. A., & Storz, G. (2001).

DNA microarray-mediated transcriptional profiling of the Escherichia coli response to

hydrogen peroxide. Journal of bacteriology, 183(15), 4562-70. doi:

10.1128/JB.183.15.4562-4570.2001.

Zhu, M. M., Lawman, P. D., & Cameron, D. C. (2002). Improving 1,3-propanediol production

from glycerol in a metabolically engineered Escherichia coli by reducing accumulation of

sn-glycerol-3-phosphate. Biotechnology progress, 18(4), 694-9. doi: 10.1021/bp020281.

Zundel, M. A., Basturea, G. N., & Deutscher, M. P. (2009). Initiation of ribosome degradation

during starvation in Escherichia coli. RNA (New York, N.Y.), 15(5), 977-83. doi:

10.1261/rna.1381309.

Zwir, I., Shin, D., Kato, A., Nishino, K., Latifi, T., Solomon, F., et al. (2005). Dissecting the

PhoP regulatory network of Escherichia coli and Salmonella enterica. Proceedings of the

National Academy of Sciences of the United States of America, 102(8), 2862-7. doi:

10.1073/pnas.0408238102.

97

Appendix

Table 2. The list of strains and plasmids used in this work.

Strain or plasmid Genotype or description Source or Reference

DH1 Wildtype E. coli: endA1 recA1 gyrA96 thi-1 glnV44 relA1 hsdR17(rK

- mK+) λ-

DP DH1 ( ackA, pta, poxB) This work, gift of C.K.

DHA DH1 ( atpE,atpF) This work, gift of C.K.

DPA DP ( ackA, pta, poxB) This work, gift of C.K.

Plasmids

pMevT Lac promoter-atoB-HMGS-HMGR, p15a, Cm

(V. J. J. Martin, Pitera, Withers, J. D. Newman, & Jay D Keasling, 2003)

pMBIS Lac promoter-MK-PMK-PMD-idi-ispA, pBBR1, Tet


pADS Trc promoter-ADS, ColE1, Amp


pADSmut Trc promoter-inactive ADS, ColE1, Amp

This work, gift of D.P.

PrstA-gfp rstA promoter-gfp, p15a, Cm

gift of W.W. (W. G. Miller et al., 2000)

PyrbL-gfp yrbL promoter-gfp, p15a, Cm


PphoPQ-gfp phoPQ promoter-gfp, p15a, Cm


PmgrB-gfp mgrB promoter-gfp, p15a, Cm


pADS-MK Trc promoter-ADS-MK, ColE1, Amp

(J. R. Anthony et al., 2009)

pADSmut-MK Trc promoter-ADSmut-MK, ColE1, Amp

This work

pADS-BIS Trc promoter-ADS-BIS, ColE1, Amp

This work, gift of P.P-Y.

pBIS-BIS Trc promoter-BIS-BIS, ColE1, Amp


pADS-ispA Trc promoter-ADS-ispA, ColE1, Amp


pMBP-BIS Trc promoter-MBP-BIS, This work, gift of P.P-Y.

98

ColE1, Amp

pFarnesene Trc promoter-farnesene synthase, ColE1, Amp


pBIS-Farnesene Trc promoter-BIS-farnesene synthase, ColE1, Amp


pADS-fps1 Trc promoter-ADS-fps1, ColE1, Amp


pBISco Trc promoter-codon optimized BIS, ColE1, Amp


pBISnco Trc promoter-non codon optimized BIS, ColE1, Amp


pBISQ404A Trc promoter-point mutation BIS, ColE1, Amp


pVetispiradiene Trc promoter-vetispiradiene synthase, ColE1, Amp


pBbA0k-rfp Rfp without a promoter, p15a, Kan

This work

pBbE0a-rfp Rfp without a promoter, ColE1, Amp

This work

pBbA0c-rfp Rfp without a promoter, p15a, Cm

This work

PrstA-rfp rstA promoter-rfp, pBbA0k-rfp

This work

PhoP-PrstA-rfp PhoP (divergent)-rstA promoter-rfp, pBbA0k-rfp

This work

PhoPQ-PrstA-rfp PhoPQ (divergent)-rstA promoter-rfp, pBbA0k-rfp

This work

PrstA-ADS rstA promoter-ADS, pBbe0a-ADS

This work

PhoP-PrstA-ADS PhoP (divergent)-rstA promoter-ADS, pBbe0a-ADS

This work

PhoPQ-PrstA-ADS PhoPQ (divergent)-rstA promoter-ADS, pBbe0a-ADS

This work

PrstA-50krbs-ADS rstA promoter-calculated rbs-ADS, pBbe0a-ADS

This work, using the rbs described below

Ptrc99a-50krbs-ADS Trc promoter-calculated rbs-ADS, This work, using the rbs

99

pADS described below

Ppromoters-rfp Promoters listed below-rfp, pBbA0k-rfp

This work, using the promoters described below

Ppromoters-ADS Promoters listed below-ADS, pBbe0a-ADS

This work, using the promoters described below

PrstA-ADS-LacI rstA promoter-ADS-LacI, pBbe0a-ADS

This work

pBbA5c-MevT-MBIS aka PlacUV5-MevT-MBIS Built by T.S.L

pBbA5c-MevT aka PlacUV5-MevT Built by T.S.L

PgadE-MevT gadE promoter-MevT, pBba0c-MevT

This work

PgadE-MevT-MBIS gadE promoter-MevT-MBIS, pBba0c-MevT-MBIS

This work

PrstA-MevT-MBIS rstA promoter-MevT-MBIS, pBba0c-MevT-MBIS

This work

PyrbL-MevT-MBIS yrbL promoter-MevT-MBIS, pBba0c-MevT-MBIS

This work

pBbA1c-MevT-MBIS aka Ptrc-MevT-MBIS biobricked trc promoter-MevT-MBIS, p15a, Cm

Built by T.S.L

Ppromoters-MevT-MBIS

Various promoters-MevT-MBIS, pBba0c-MevT-MBIS

This work

pC9b Trc promoter-C9b (B. subtilis fragment, isopentenol synthesis), ColE1, Amp

(Withers et al., 2007)

PrstA-C9b rstA promoter-C9b, pBbe0a-rfp

This work

pBbA5c-MevT-MevB aka PlacUV5-MevT-MevB Built by T.S.L

PgadE-MevT-MevB gadE promoter-MevT-MevB, pBbA0c-MevT-MevB

This work

PrstA-MevT-MevB rstA promoter-MevT-MevB, pBbA0c-MevT-MevB

This work

PrstA-BISco rstA promoter-BIS, pBbe0a-rfp

This work

Table 3. The list of all promoters and their sequences used in this work.

Promoter region

Nucleotide sequence

rstA gattaccagccccaactttttaccttaacacttccataacaagtcatcagtagaatacctgatgaaaacttgtttagaaacgattgatagtaagtaaaaacagcgcggtgtattgtgacgtttttatatctacc

100

yrbL tggccgccgttatcgccctctttaattctggttttgatgaagattaatcttcatcactttgacatacaactcccttcaaactccccccgacaataagaaaatcacgtactgaaatcgttctcaatcaacgtcatttgtacattttgtgcgcttttcactttcagaagaaccttaagaaaaccttaagaggcattgtttaggttttgtttaagttaatcgaccatactggagatcgtcagaaaatatttcc

phoPQ gagctatcacgatggttgatgagctgaaataaacctcgtatcagtgccggatggcgatgctgtccggcctgcttattaagattatccgctttttattttttcactttacctcccctccccgctggtttatttaatgtttacccccataaccacataatcgcgttacactattttaataattaagacagg

mgrB aatgatgaatcgcattacaacctcttctctttttatgttcgcttaatcgtagcggcaatatgcgctgaagcaagcgactcattccgaaaaagcacgaatatcgacatagttaggcgctgtttaactaacgcatgctagtttaatgacataaggtaggtgaaacggagattgga

ycjM ccccatagattatttgcgtcaccccatagattatttgcgtcagctcacaaatacgctttttccctggtaaaaaatgatttcctgcgtgactaaaacccttgtgctcaattgacagtttattttctgcggagtagtctctcgtttcatgggaccgctaccacggaaaggcaa

ompF aagaagattttgtgccaggtcgataaagtttccatcagaaacaaaatttccgtttagttaatttaaatataaggaaatcatataaatagattaaaattgctgtaaatatcatcacgtctctatggaaatatgacggtgttcacaaagttccttaaattttacttttggttacatattttttctttttgaaaccaaatctttatctttgtagcactttcacggtagcgaaacgttagtttgaatggaaagatgcctgcagacacataaagacaccaaactctcatcaatagttccgtaaatttttattgacagaacttattgacggcagtggc

yqjF cggttttggcgtatggagcgcctggcgtctggttaaaacgaccctcaagcagcaacagcttcgcggttaacttccctctggccggagccattccggccttatccctcaaattttttgaagatttttgacagttttccttgctaacaatcatcattcaccacgtttatgattctctccatcgacagcaacgacgctaataccgcgccattgcacaaaaaaacaatcagcagcctgagtggc

ibpA catcatcattacgtcgcactgtggcggctatcgcactttaacgtttcgtgctgccccctcagtctatgcaatagaccataaactgcaaaaaaaagtccgctgataaggcttgaaaagttcatttccagacccatttttacatcg

gadE ttaatactctctccgctacgcagtgttgtagatcaattgcgcactatcattgaaataattacctgctagtgattatttcaacctactgaatttcatctaatttttttcactctatggcaaattagccatttcaaacattatcatggctgatattttccgtagtcaggtttaatgttttaaaagtgctgtgggaaagtgaacaaagagttccgtaagcgttgatgctatgggcggttaaataagtaatccgggttcatttttttgcaactggcgttgattacattgcataaatatccgtgtctccagacgctatataaaaacctgaagacatgaatgcgttatttactcaggtaatttcaatgcgttaaaagaaagctggcaatccaattgccagcttaagtcgaaacaaggagactcgatatttaaatcggattacattttaactttagtaatattcttcagagatcacaaactggttattgataacttattcttgggcagtaatccgcaaacgttaactttttgtttgctatttacaagctgataacaaccaggaatcttacttaggatcaatatatggagtgcgtgatggataaatctgaagtattgattagtgttaatagacgtattagttcacgaagggtaaagttcttataggcgtttactatattgaacaacgattcggacaaggatgtaaataatgaaaaggatgacatattcgaaacga

mdtE gagggatgcagatgctgatttcattacccccggtgattactaaaggagaggctaaaacgactttattcccctggtatgtgtatccaccagtagaacccttcgttgcccgaatgctggcaggaactgttggcagaacggcaacattttttttgtcgttgacctcaccatgtcgatcactgtgcctgtatcccaccttactggctgacaaccccactatgccgctggtctgta

sdhC tgcctttcagcacatccagttcaacagctgtatccccgttgagggtgagttttgcttttgtatcagccatttaaggtctccttagcgccttattgcgtaagactgccggaacttaaatttgccttcgcacatcaacctggctttacccgttttttatttggctcgccgctctgtgaaagaggggaaaacctgggtacagagctctgggcgcttgcaggtaaaggatccattgatgacgaataaatggcgaatcaagtacttagcaatccgaattattaaacttgtctaccactaataactgtcccgaatgaattggtcaatacttccacactgttacataagttaatcttaggtgaaataccgacttcat

101

aacttttacgcattatatgcttttcctggtaatgtttgtaacaactttgttgaatgattgtcaaattagatgattaaaaattaaataaatgttgttatcgtgacctggatcactgttcaggataaaacccgacaaactatatgtaggttaattgtaatgattttgtgaacagcctatactgccgccaggtctccggaacaccctgcaatcccgagccacccagcgttgtaac

sucA tgcaacgtaatgcgtaaaccgtaggcctgataagacgcgcaagcgtcgcatcaggcaaccagtgccggatgcggcgtgaacgccttatccggcctacaagtcattacccgtaggcctgataagcgcagcgcatcaggcgtaacaaagaaatgcaggaaatctttaaaaactgcccctgacactaagacagtttttaaaggttccttcgcgagccactacgtagacaagagctcgcaagtgaa

paaA gctaactgctgcatcgctactctccagatgtttcacatttctgttgctaatagttaaatcgcgaatcataaaaagcaaaggatcttttaacgaaatgttaactatgcgatctgtatagcaactgcggaaaacattaatgcactgataaataatgatttataaaaatagggtgcgaaatccgtcacagttcaaacatacaaaatttgtgattttacttaactattgtgtaactttcataaaacaatgtgattcgtgtttttaattaattcacgaaaactggaatcgtaaaggtgatg

flgB tgtgcaattgcgatgtgagattgctcgccgtacttaacggactgaacagtatcgcgatgatcgccacgctacgttttattatcagcattttcgcccccagccatttctacaacgtgaattgtacctgtccgcaatgaccatcaacggcataaatagcgacccattttgcgtttattccgccgataacgcgcgcgtaaaggcatttaagctgatggcaga

ydcS aataccgcgtggcagtggggagagcgtgaagaacaggcggtaaaacaattaggcaaacttattcaagaacggctgtaatagcgtttaatttaattcctcttagattgggtaatatgaatttcgaatagcagtcatatttcctaactccttgactatactccagaagataaccttacagacggcataatgcgcggtagctcacaacctgaataaattttctcaggggcgaaggtgtgcctgcaagccgccgtctatggttaaacaaggagatatttttacggcacggcggctgaacaattaattacgacaggagtaagaccttATGAGCAAGACATTTGCCCGCAGCAGCCTGTGTGCGCTCAGCATGACAATAATGACCGCTCACGC

add aggtcgtggttttgtccgtctcaatgccggctgcccacgttcgaaactggaaaaaggtgtggctggattaattaacgccatccgcgctgttcgttaaccccaattgcgcaacgtaaaaaatcgttgcgcaatcgtggatttttaccctgctttgtttttataatggtgcgcacttttatatccagaaaaagagtgcgaccATGATTGATACCACCCTGCCATTAACTGATATCCATCGCCACCTTGATGGCAACATTCGTCCCCAGACCA

fadB catcgcgagtccgttcttgtaaggtagctatatgatttttatagagcgaggccagtgattccattttttacccttctgtttttttgaccttaagtctccgcatcttagcacatcgttcatccagagcgtgatttctgccgagcgtgatcagatcggcatttctttaatcttttgtttgcatatttttaacacaaaatacacacttcgactcatctggtacgaccagatcaccttgcggat

ychH gtcaacgaaccaggcaccagcattatgtcgcgttgcggcgtattcagcaccggggttcgccaggccgacaatcaatttaatcgtcacgtttttttgtcctgagtgtgtacataactggcgcgtagtttactggttgcggccccgcttgacaaaaaactgcgtatcaaatgcagataacgtaataattgcctgagtggactattagaaagtcaaggtgttcaggcgtttatttgtaaagttttgttgaaataagggttgtaattgtgatcacgcccgcacataacccactgggtgttgtctatactttacacataaggaagaggggtattccctgttacaacccagaaagttccgg

atpI atggcgaacgtcatctggtggtgattaaagcaaataaaatttaatttttatcaaaaaaatcataaaaaattgaccggttagactgttaacaacaaccaggttttctactgatataactggttacatttaacgccacgttcactcttttgcatcaacaagataacgtggctttttttggtaagcagaaaataagtcattagtgaaaatatcagtctgctaaaaatcggcgctaagaaccatcattggctgttaaaacattattaaaaatgtcaatgggtggtttttgttgtgtaaatgtcatttattaaaacagtatctgtttttagactgaaatatcataaacttgcaaaggcatcatttgccaagtaaataaatatgctgtgcgcgaacatgcgcaatatgtgatctgaagcacgctttatcaccagtgtttacgcgttatttacagtttttcatgatcgaacagggttagcagaaaagtcgcaattgtatgcactggaaaaatatttaaacatttattcaccttttggctacttattgtttgaaatcacgggggcgcaccgtataatttgaccgctttttgatgcttgactctaagccttaaagaaagttttatacgacacgcggcata

102

bolA tggtaaaaattcccgccatcataacattgccaacggcgaggggaagtgggtaaggcatgtaaattcatcatgttgacgaaataatcgcccctggtaaaagaaacactgatgcgaggcctgtgtttcaatctttaaatcagtaaacttcatacgcttgacggaaaaaccaggacgaaacctaaatatttgttgttaagctgcaatggaaacggtaaaagcggctag

mgtA ctttgatggtcagccgattttgcatcctgttgtcctgtaacgtgttgtttaattatttgagcctaacgttacccgtgcattcagcaatgggtaaagtctggtttatcgttggtttagttgtcagcaggtattatatcgccatagatgctacgaatattattggattctccttattatttgcggcgcttttttcacttaccggaggttatatggaacctgatcccacgcctctccctcgacggagattaaaacttttccggtaagcccgtcttttcacggcgttaccggatgcgtaaggccgtgacgttttaacgtccctgctcagctttattacct

yrbL tgatgttggattctgccaaaggacggcagattgaagttgaagcgaccggtccacaggaagaggaagcactggccgccgttatcgccctctttaattctggttttgatgaagattaatcttcatcactttgacatacaactcccttcaaactccccccgacaataagaaaatcacgtactgaaatcgttctcaatcaacgtcatttgtacattttgtgcgcttttcactttcagaagaaccttaagaaaaccttaagaggcattgtttaggttttgtttaagttaatcgaccatactggagatcgtcagaaaatatttcc

mgrB acagcgggcaatctgttatccccaaaaaaccacttttagtgtgcaagtattgtaccgtgctggtgcctctggcagtcagataggtacattgcaaacctaatcctgcggcattctctttgcttccaatcaaaacgccatatccgctgagtaataatcctatccataccagtgctatcagcataactgtgcgaatgatgaatcgcattacaacctcttctctttttatgttcgcttaatcgtagcggcaatatgcgctgaagcaagcgactcattccgaaaaagcacgaatatcgacatagttaggcgctgtttaactaacgcatgctagtttaatgacataaggtaggtgaaacggagattgga

slyB gtcgccaacacaacatcaacaccatcaaggctggtgcctgacataacgccaataaagcggcccgatttcatagttcatcctttttcaatctgacgtttgcgcaccactcaaacataaacttttcgtgaataccatgcggaatgaccgatttttaccgttggtagtaaaacattatcttcaaatcaataatcatcatgaatgttttgtttataattggttgatcctactttcattatgatttgctcatatttggtagaacatgtaaccatggattcacatatgccatatactttgaccatgagggatgcttgcgtggcgtttcatggtgaaca

purH tcgacttgcatgtgttaggcctgccgccagcgttcaatctgagccatgatcaaactcttcaatttaaaagtttgatgctcaaagaattaaacttcgtaatgaattacgtgttcactcttgagacttggtattcatttttcgtcttgcgacgttaagaatccgtatcttcgagtgcccacacagattgtctgataaattgttaaagagcagttgcgacgcgctttagcgcactgtcgcgaggtggcgtatattacgctttcctctttcagagtcaaccctgaatttcaggatttttctcttcaaccgaaccggctgtttgtgtgaagtgattcacatccgccgtgtcgatggaggcgcattatagggagttctccgcaggccgcaatagaaaaattgcagaaaaatgactgactgctgcattccccagcaaaagcccgctttatacctttttacgcacagagttatccacaatcatcaatgtaatttctgtattttgcccacggtaaccacagtcaaaattgtgatcaccattgaaagagaaaaattcgcgagcgttgcgcaaacgttttcgttacaatgcgggcgaaaaataaggatgccccgt

purM ttgctcacgcatcagtcccagcttgtgtttgacgagtgggtgtttgacttccacgatcttcatactctttctcctttgaggggcagccacaaaaaaaatcgacggattatacctcctttcttcaaggcggcaatattcttttcgttgactttagtcaaaatgataacggtttgagataaagttattttatattcagatggttatgaaagaagattattccatccgaaaactaacctttaccctggcacaagtcttctttcgccgcgcgcctggggaaaagacgtgcaaaaaggttgtgtaaagcagtctcgcaaacgtttgctttccctgttagaattgcgccgaattttatttttctaccgcaagtaacgcg

ispU agtcgccagaaaagaggtgatgcgtctcgcaagctgaggataatccggctacagagagtcgcgctatttgttagcgtagggcttcagtgatatagtctgcgccatctgatcgtaagtagttggctttataaggtcagatatgccgtggttttacacggcttttttttgtataggcttcagtattcctgagtaccgtaaaccctgtcag

103

dxr gccgttctcaggacgatgtacagaaactgactgatgctgcaatcaagaaaattgaagcggcgctggcagacaaagaagcagaactgatgcagttctgatttcttgaacgacaaaaacgccgctcagtagatccttgcggatcggctggcggcgttttgctttttattctgtctcaactctggatgtttc

xseB gcgcacagattggcttttgatggtgatttccgggaacaatttaatgataaacttcatggcggcaatggttcgttggcaagccttaagcgacttgtatagggaaaaatacagcagcccacacctgcggctgcatccaggcgcggaagtataccactaacatcgctttgctgtgcacatcaccttaccattgcgcgttatttgctatttgccctgagtccgttaccatgacggggcgaaaaatattgagagtcagacattcattATGCCG

ybjG gcctgacgtccaagcctggcaccggaagctaatttattttgcatgcaatttcttcgccaataataatcgcgcagagtttaataaaagcgcagctaacgagaaagcgaattttgtagctgaaaccacggttaagcacattcttacattattacgagtatagctacgctttctttaagttttatttaacctatgcccgttacaatcacccaccgtaaacaggccgcttgagggaaataagacgatgccgctttacccagtttaacctgcactttattctcaacgacttgcctgtattggct

miaA ccgaagaaaaatcggcattagaaaaagcgcagtctgccctggcggaattgggtattgatttccagtcagatgcacagcatgtgaccatcagggcagtgcctttacccttacgccaacaaaatttacaaatcttgattcctgaactgataggctacctggcgaagcagtccgtattcgaacctggcaatattgcgcagtggattgcacgaaatctgatgagcgaacatgcgcagtggtcaatggcacaggccataaccctgctggcggacgtggaacggttatgtccgcaacttgtgaaaacgccgcc

pspA aaacctgttccagcacttcgagaaagctgttcgcctcaccaagtaaattatctttgtattctgccatgatgaaattcgccacttgttagtgtaattcgctaactcatcctggcatgttgctgttgattcttcaatcagatctttataaatcaaaaagataaaaaattggcacgcaaattgtattaacagttcagcaggacaatcctgaacgcagaaatcaagaggaca

aroF catcataccagtcggtgctggtcaaggttggcgcgtgagtataaggatcaatctgacatccgctaaccagtaaagccaacaagggggcgataaactttttcatcatttctttctcctttttcaaagcatagcggattgttttcaaagggagtgtaaatttatctatacagaggtaagggttgaaagcgcgactaaattgcctgtgtaaataaaaatgtacgaaatatggattgaaaactttactttatgtgttatcgttacgtcatcctcgctgaggatcaactatcgcaaacgagcataaacaggatcgc

entC gcgagtctcacaaatcagcttcctgttattaataaggttaagggcgtaatgacaaattcgacaaagcgcacaatccgtcccctcgcccctttggggagagggttagggtgaggggaacagccagcactggtgcgaacattaaccctcaccccagccctcaccctggaagggagagggggcagaacggcgcaggacatcacattgcgcttatgcgaatccatcaataatgcttctcattttcattgtaaccacaaccagatgcaaccccgagttgcagattgcgttacctcaagagttgacatagtgcgcgtttgcttttaggttagcgaccgaaaatataaatgataatcattattaaagcctttatcattttgtggaggatgatATGGATACGTCACTGGCTGA

fabH agcgttggctacaaaagagcctgacgaggcgattcagtcacctcaaccccgaccagtataacggcgcctgtctgttaggattgcgcggcacggtgataaaaagtcatggtgcagccaatcagcgagcttttgcggtcgcgattgaacaggcagtgcaggcggtgcagcgacaagttcctcagcgaattgccgctcgcctggaatctgtatacccagctggttttgagctgctggacggtggcaaaagcggaactctgcggtagcaggacgctgccagcgaactcgcagtttgcaagtgacggtatataaccgaaaagtgactgagcgtacATG

fabA ccatctgggcagtagacgatgtgactgacgcactgccgttattattaaatctggtgtgggatggcgaaggccaaacgacgctgatgcaaaccatccaggaacgtatcgcgcaagcatcgcaacaggaaggacgtcaccgttttccatggccattacgttggctgaactggtttattccgaactgatcggacttgttcagcgtacacgtgttagctatcctgcgtgcttcaataa

fabB tcaaaatctcgggaaacaggtgtaccctcagcattaaattcgaggttggcaggttgtatggagtagtgtttcacgtaagttactcgtcttacaggcggtggctcgatcttagcgatgtgtgtaaggctgcgcaaatttctctattaa

104

atggctgatcggacttgttcggcgtacaagtgtacgctattgtgcattcgaaacttactctatgtgcgac

accD gatcgtcgggattattgtggtttctattttgccgggcgtcatcgaaataatccgtcacaaacgcgctgcggcacgcgccgcaaaataaaaagtaactccgcggttcgaccacttttttatccaaagtttcgggctgttatgttttaatgtgcaacattcatggtctgttgggggcaaaaatggcattatgcgtccccaaagataaaactggc

gadE-500 tcaggtttaatgttttaaaagtgctgtgggaaagtgaacaaagagttccgtaagcgttgatgctatgggcggttaaataagtaatccgggttcatttttttgcaactggcgttgattacattgcataaatatccgtgtctccagacgctatataaaaacctgaagacatgaatgcgttatttactcaggtaatttcaatgcgttaaaagaaagctggcaatccaattgccagcttaagtcgaaacaaggagactcgatatttaaatcggattacattttaactttagtaatattcttcagagatcacaaactggttattgataacttattcttgggcagtaatccgcaaacgttaactttttgtttgctatttacaagctgataacaaccaggaatcttacttaggatcaatatatggagtgcgtgatggataaatctgaagtattgattagtgttaatagacgtattagttcacgaagggtaaagttcttataggcgtttactatattgaacaacgattcggacaaggatgtaaataatgaaaaggatgacatattcgaaacga

gadE-400 aaacctgaagacatgaatgcgttatttactcaggtaatttcaatgcgttaaaagaaagctggcaatccaattgccagcttaagtcgaaacaaggagactcgatatttaaatcggattacattttaactttagtaatattcttcagagatcacaaactggttattgataacttattcttgggcagtaatccgcaaacgttaactttttgtttgctatttacaagctgataacaaccaggaatcttacttaggatcaatatatggagtgcgtgatggataaatctgaagtattgattagtgttaatagacgtattagttcacgaagggtaaagttcttataggcgtttactatattgaacaacgattcggacaaggatgtaaataatgaaaaggatgacatattcgaaacga

gadE-300 aaatcggattacattttaactttagtaatattcttcagagatcacaaactggttattgataacttattcttgggcagtaatccgcaaacgttaactttttgtttgctatttacaagctgataacaaccaggaatcttacttaggatcaatatatggagtgcgtgatggataaatctgaagtattgattagtgttaatagacgtattagttcacgaagggtaaagttcttataggcgtttactatattgaacaacgattcggacaaggatgtaaataatgaaaaggatgacatattcgaaacga

gadE-200 agtaatccgcaaacgttaactttttgtttgctatttacaagctgataacaaccaggaatcttacttaggatcaatatatggagtgcgtgatggataaatctgaagtattgattagtgttaatagacgtattagttcacgaagggtaaagttcttataggcgtttactatattgaacaacgattcggacaaggatgtaaataatgaaaaggatgacatattcgaaacga

gadE-100 gattagtgttaatagacgtattagttcacgaagggtaaagttcttataggcgtttactatattgaacaacgattcggacaaggatgtaaataatgaaaaggatgacatattcgaaacga

gadE-50 caacgattcggacaaggatgtaaataatgaaaaggatgacatattcgaaacga

gadE-P1 ttaatactctctccgctacgcagtgttgtagatcaattgcgcactatcattgaaataattacctgctagtgattatttcaacctactgaatttcatctaatttttttcactctatggcaaattagccatttcaaacattatcatggctgatattttccgtagtcaggtttaatgttttaaaagtgctgtgggaaagtgaacaaagagttccgtaagcgttgatgctatgggcggttaaataagtaatccgggttcatttttttgcaactggcgttgattacattgcataaatatccgtgtctccagacgctatataaaaacctgaagacatgaatgcgttatttactcaggtaatttcaatgcgttaaaagaaagctggcaatccaattgccagcttaagtcgaaacaaggagactcgatatttaaatcggattacattttaactttagtaatattcttcagagatcacaaactggttattgataacttattcttgggcagtaatccgcaaacgttaactttttgtttgctatttacaagctgataacaaccaggaatcttacttaggatcaatatatggagtgcgtgatggataaatctgaagtattgattagtgttaataga

gadE-P2 ttaatactctctccgctacgcagtgttgtagatcaattgcgcactatcattgaaataattacctgctagtgattatttcaacctactgaatttcatctaatttttttcactctatggcaaattagccatttcaaacattatcatggctgatattttccgtagtcaggtttaatgttttaaaagtgctgtgggaaagtgaacaaagagttccgtaagcgttgatgc

105

tatgggcggttaaataagtaatccgggttcatttttttgcaactggcgttgattacattgcataaatatccgtgtctccagacgctatataaaaacctgaagacatgaatgcgttatttactcaggtaatttcaatgcgttaaaagaaagctggcaatccaattgccagcttaagtcgaaacaaggagactcgatattt

gadE-P3 ttaatactctctccgctacgcagtgttgtagatcaattgcgcactatcattgaaataattacctgctagtgattatttcaacctactgaatttcatctaatttttttcactctatggcaaattagccatttcaaacattatcatggctgatattttccgtagtcaggtttaatgttttaaaa

gadE-P4 ttaatactctctccgctacgcagtgttgtagatcaattgcgcactatcattgaaataattacctgctagtgattatttcaacctactgaatttcatctaatttttttcactctatggcaaattagccatttcaaacattatcatggctgatattttccgtagtcaggtttaatgttttaaaagtgctgtgggaaagtgaacaaagagttccgtaagcgttgatgctatgggcggttaaataagtaatccgggttcatttttttgcaactggcgttgattacattgcataaatatccgtgtctccagacgctatataaaaacctgaagacatgaatgcgttatttactcaggtaatttcaatgcgttaaaagaaagctggcaatccaattgccagcttaagtcgaaacaaggagactcgatatttaaatcggattacattttaac

Constitutive #1 ttgacagctagctcagtcctaggtataatgctagc

Constitutive #2 ttgacagctagctcagtcctaggtattgtgctagc

Constitutive #3 tttacagctagctcagtcctaggtattatgctagc

Constitutive #4 tttacggctagctcagccctaggtattatgctagc

Constitutive #5 tttacggctagctcagtcctaggtacaatgctagc

Constitutive #6 ttgacagctagctcagtcctagggattgtgctagc

Constitutive #7 tttacagctagctcagtcctagggactgtgctagc

osmY cgcctggcacaggaacgttatccggacgttcagttccaccagacccgcgagcattaattcttgcctccagggcgcggtagccgctgcgccctgtcaatttcccttccttattagccgcttacggaatgttcttaaaacattcacttttgcttatgttttcgctgatatcccgagcggtttcaaaattgtgatctatatttaacaaagtgatgacatttctgacggcgttaaataccgttcaatgcgtagatatcagtatctaaagccgtcgattgtcattctaccgatattaataactgattcagaggctgtaatggtcg

yahO caaaaataatcgggtgcgagagagatcacaaagtgtcttatttccggttactggcgtttatgccctgactgaactaattattaatcaacccaataatgtgggtgggtgatagtgtgataacaactctgga

xasA caggagacacagaatgcgcataaaaataacagcataaaacaccttaccaccacccaagaatttcatattgtattgtttttcaatgaaaaaatattattcgcgtaatatctcacgataaataacattaggattttgttatttaaacacgagtcctttgcacttgcttactttatcgataaatcctacttttttaatgcgatccaatcattttaagg

slp tgatggatattcatgtcacgccccaaaattaactgagttcacctaaacagaaaggatataaacatcagacaggtttacgttactatcaggcatatcacctcagaatcagatgaaaactataaagaaatatctattatggttttaatatttgttgataaggatagtaacATGAACATG

csgB gcttgtgcgcaagacatatcgcagcaatcagcgacgggcaagaagaatgactgtctggtgctttttgatagcggaaaacggagatttaaaagaaaacaaaatatttttttgcgtagataacagcgtatttacgtgggttttaatactttggtatgaactaaaaaagaaaaatacaacgcgcgggtgagttattaaaaatatttccgcagacatactttccatcgtaacgcagcgttaacaaaatacaggttgcgttaacaacc

106

treA gctaatcacgaacccaacgccgctggctggctaataaacagacgttatttgatcactctatttcgccgttatttatccctttaattcccttttctaaaatgcctgacagttcgcagaatgagatttcgatcatgcagctagtgcgatcctgaactaaggttttctgatacttgaataccg

b1592 atctcttaaagcacgcactgcttttgcggctggcctcttttgccgcaaaatagtcgcccgtgtttcattgcccatttctgctcatgcatcatctacacatctatccggatctgcgcact

yhdT gatctgcagatccgcatcatgaatgacgagaacttccagcatggtggcactaacatccactatctggagaaaaaactcggtcttcaggaaaaataagactgctaaagcgtcaaaaggccggattttccggccttttttattactggggatcgacaacccccataaggtacaatccccgctttcttcac

yihE tgaaaatgcgtgattccgcgaaagatgcggtgatcccgggtttgcagaaagattatgaagaagacttcaaaacggcgctgttacgcgctcgcggcgtaattaaagagtaaaagcttgtaagcggcgccaccaaaatcatcgtgaaatgatatccttcgtcattcg

hyfA cattccgtgctggctgcgcttgcggccagcatacctcacttctcgtgatcaagatcacattctcgctttcccctgcgacacgggtgtcgaatccattttttgctgaacgttaatgaccatcatttttgtaccgttcagaatccagttaatacataacttattgaatatattgagttaatcagaatggcatcctttatgcaatatgaaatgcaatg

citA tgcaacgaatgttgttcaatgttgcaaactgataaccttttattttcacttgggagaaagggggtgatcgaggtatatctttttctcctttcgctatacatcctaaggagtatttcggcgtgaaattttgatttatttcacatagagttagtggttttttatttatttaatgattttaagttttttaattaatgtaattacgaaatgactcgcaggtttaagtgatttaattgatttaatgaataaaatttgccacgatcataattaatatctatgtattttgattcaacattttaattacatccgtc

thiM acagattgctgctatccgtggcgcggtaaacggtctgatgcgggaagtgattaaaggtcatctgacggaacacatcgttcaccagggggatgagctaaaacgtgaagaagatctggatgtcgttctgaaggtgctggattcatatatcaaataatttattaacgcgattgtaaaactgccgtttttcctcgtttacaacgcgtgcgctggacattaccatcctcctct

pykA gcgacaggatttactttgcgacgcggtgcaaaattcagagataacttgaagcgggtcaaagaagcgctgaaggaatcgcgttttgataagcagttacttaatttaagtgacgatcgctaaaaacgactgtcactgtcctaatcttatacgacatcc

mdtE –redo attgatggctggcatgacgagggatgcagatgctgatttcattacccccggtgattactaaaggagaggctaaaacgactttattcccctggtatgtgtatccaccagtagaacccttcgttgcccgaatgctggcaggaactgttggcagaacggcaacattttttttgtcgttgacctcaccatgtcgatcactgtgcctgtatcccaccttactggctgacaaccccac

bolA – redo agagcaactaacgggaagaggatttttttgaacatgttcgggctctcagagactcttaagcgtgtttggtaaaaattcccgccatcataacattgccaacggcgaggggaagtgggtaaggcatgtaaattcatcatgttgacgaaataatcgcccctggtaaaagaaacactgatgcgaggcctgtgtttcaatctttaaatcagtaaacttcatacgcttgacggaaaaaccaggacgaaacctaaatatttgttgttaagctgcaatggaaacggtaaaagcggctagtatttaaaggg

ispU – redo gtgaagtcgccagaaaagaggtgatgcgtctcgcaagctgaggataatccggctacagagagtcgcgctatttgttagcgtagggcttcagtgatatagtctgcgccatctgatcgtaagtagttggctttataaggtcagatatgccgtggttttacacggcttttttttgtataggcttcagtattcctgagtaccgtaaaccctgtcagggaataaaaaacgc

pspA – redo gtgcgagatgcgaaacctgttccagcacttcgagaaagctgttcgcctcaccaagtaaattatctttgtattctgccatgatgaaattcgccacttgttagtgtaattcgctaactcatcctggcatgttgctgttgattcttcaatcagatctttataaatcaaaaagataaaaaattggcacgcaaattgtattaacagttcagcagg

107

aroF – redo agcggattgttttcaaagggagtgtaaatttatctatacagaggtaagggttgaaagcgcgactaaattgcctgtgtaaataaaaatgtacgaaatatggattgaaaactttactttatgtgttatcgttacgtcatcctcg

entC – redo gagtctcacaaatcagcttcctgttattaataaggttaagggcgtaatgacaaattcgacaaagcgcacaatccgtcccctcgcccctttggggagagggttagggtgaggggaacagccagcactggtgcgaacattaaccctcaccccagccctcaccctggaagggagagggggcagaacggcgcaggacatcacattgcgcttatgcgaatccatcaataatgcttctcattttcattgtaaccacaaccagatgcaaccccgagttgcagattgcgttacctcaagagttgacatagtgcgcgtttgcttttaggttagcgaccgaa

Table 4. The list of ribosome binding sites used in the study. The BglII restriction enzyme site is the 5’ end and is italicized. The start codon of ADS is highlighted in red.

Ribosome binding

site Nucleotide sequence

Calculated strength

(H. M. Salis,

Mirsky, & C. A.

Voigt, 2009)

Ppromoter-ADS

AGATCTAGGAGGAAGGATCTATGGCCCTGAC 7713

50krbs #1 AGATCTCAACCAGCTTCAACAGACATAGAGAGGAGGTAGCATGGCCCTGAC 41647

50krbs #2 AGATCTTCCCCTATAGAATAGTAACGTAAGACAAGAGGTATGGCCCTGAC 23407

50krbs #3 AGATCTGAGAGCATAACAACTAGTAAGGAAGTACAACATGGCCCTGAC 50425

50krbs #4 AGATCTACCAACACAGATTACACCAAGGAGACCGATATATGGCCCTGAC 76015

50krbs #5 AGATCTGCAAAGATACAGCCCCACGCTGGAGGTAATTATGGCCCTGAC 36710

Table 5. Constitutive promoters tested for RFP strength and amorphadiene production when cloned 5’ of RFP or MevT-MBIS operon, respectively. Strengths listed are reported from the MIT parts registry.

Constitutive Promoter Alternative Name Strength Reference

Constitutive #1 BBa_J23119 n/a http://partsregistry.org/Promoters/Ca

talog/Anderson

Constitutive #2 BBa_J23104 0.72 http://partsregistry.org/Promoters/Ca

talog/Anderson


talog/Anderson


talog/Anderson









108

talog/Anderson


talog/Anderson


talog/Anderson

Figure 43. The base vector for the sensors used in this study. AatII and EcoRI sites are highlighted as well to indicate the removal of LacI and promoter to make pBbA0k-rfp.

lacI 1131...49

1 AatII (1)

lacUV5 promoter 1500...1575lac operator 1516...1579

1580 EcoRI (1)1589 BglII (1)RBS 1595...1614

RFP 1615...2292

p15a 3163...2452

dbl term 2317...24452302 XhoI (1)

2293 BamHI (1)

TO 3254...3359

Kan/neoR 4179...3385

pBbA5k-RFP

4315 bp




109

Figure 44. A representative promoter sensor. gadE promoter inserted 5’ of rfp in the pBbA0k-rfp vector.

RBS 749...768

gadE promoter region 16...742

10 BglII (1)1 EcoRI (1)

RFP 769...1446

1447 BamHI (1)1456 XhoI (1)

dbl\term 1471...1599

TO 2408...2513

p15a 2317...1606

Kan/neoR 3333...2539

2-gadE-RFP-p15a-Kan

3469 bp

engineering a responsive, heterologous mevalonate …...engineering a responsive, heterologous...

Documents