1o

.. . a. e: V... V . . V . V V
2.33.323... 3; V . V V V . V V V V
V. ... V V V V . V VV V V.. .
rv V , 0- g.

.1 V V V V . V

.1. ...
A

r

.1 .9
1V V. I”
:khﬂﬁuv. .

 

V 92%.
.3 . V V V . . . V J.. :35...“
..»..V....V........ . V V . . V V . a“...
g; in... , V V . V . 33.

43:1.
\

I... .53.
.. is...
éStInal-t 7

.. V V
Vii... A... . . . . V
win-yo" .u'a ' 1, Li!
.‘I \ ’1 I. in: a . , . g:
D in... r“... . . . V . . . 131 1!
. V . . . V . .ﬁnfh
. «an. kn...

7» up III.
.229. .
.... hr... 29.?

z... hr.“

1
5.37..
.534: V . V V
3:1... 35‘ . . It:
.. .......a..§.«....£ V V ﬂame... .
J. V V . .2.
”it

4". .
.32».
.. a
. . \i I!“
.- ~.i 3014. 53’“ .
Kevan. .3... V V 1
1m .ttvzl...v3rmw.1ﬂ¢ V . 1.5."...

.,u.¢!. .n:

 

 

 

 

.a. .. an
.V....:....ht...nn...ﬁ.....d.y. : r
at! 15:! 1.1.3 ‘3'»: .
Yvnlj‘NO-In 75" he, “1‘
- 1.1.... «.704...
4f

a 5

. .

Lt. .i... .0!“
. 3-1.x. .3...

. (.nv 3‘. .1

. s V 1......ng in
an... ”.39....

g . -

i!
gun l...
.AI‘NIMWOILJ .v”.'-l
. . .
3.93:...

pm... .4
4 . 5 ialip r
y .33 3...... “TI... w .
. V earn». $u.ﬂéﬁ>a‘ . - .
. . VV~ Vsiwwnl. unhuVmeV Hurt. J» n... {.1 1 V. ..
“43. $2. 4V . 2. . .

 

 

 

This is to certify that the
thesis entitled

INSIGHTS INTO ACTINOBACILLUS SUCCINOGENES
FEHMENTATIVE METABOLISM

presented by

James B. McKinlay

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Microbiology and Molecular
Genetics

 

 

9/7 41.2%».

 

ﬂ Map/Professor’s Signature

/1ﬂ;hé

Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

‘— _.

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 pziClRC/DaleDueVinddpJ

 

INSIGHTS INTO AC TINOBACILLUS S U C CINOGENES FERMENTATIVE

METABOLISM

By

James B. McKinlay

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Microbiology and Molecular Genetics

2006

ABSTRACT
INSIGHTS INTO ACTINOBA CILLUS SUCCINOGENES FERMENTATIVE
METABOLISM
By

James B. McKinlay

Bio-based succinate production is receiving increasing attention as a potential
intermediary feedstock for replacing a large petrochemical-based bulk chemical market.
The economical and environmental beneﬁts of a bio-based succinate industry have
motivated research and development of organisms that produce succinate. Actinobacillus
succinogenes is one of the best succinate-producing organisms ever reported, however it
also produces unwanted forrnate, acetate, and ethanol. Thus, it is desirable to engineer A.
succinogenes metabolism to produce succinate as the sole fermentation product.
Metabolic engineering of A. succinogenes has been impeded by a relatively limited
knowledge of its physiology and metabolism, and a lack of genetic tools.

l3C-metabolic flux analysis methods were chosen to obtain a detailed knowledge
of A. succinogenes metabolic pathways, the ﬂuxes through these pathways, and how
these ﬂuxes respond to perturbations. l3C metabolic flux analysis is informative when
different pathways result in different labeling patterns in metabolic products (e. g.,
proteinaceous amino acids). A chemically deﬁned A. succinogenes growth medium,

which included only the required amino acids (i.e., cysteine, methionine and glutamate),
was developed to ensure that amino acids are labeled from 13C-labeled substrate.

Glutamate auxotrophy was determined to be due to an inability to made a-ketoglutarate,

indicating at least two missing tricarboxylic acid cycle-associated activities. A 13C-
labelling experiment was performed with [1-13C] glucose in the deﬁned medium to
delineate A. succinogenes metabolic pathways and quantify in vivo ﬂuxes. N ADPH was
produced by pyruvate and forrnate dehydrogenases coupled with transhydrogenase and
by malic enzyme rather than by the oxidative pentose phosphate pathway. A C4-
decarboxylating activity shunted ﬂux from the succinate-producing C4 pathway to the
formate-, acetate-, and ethanol-producing C3 pathway, likely through both malic enzyme
and oxaloacetate dccarboxylase. l3C-labelling experiments were then conducted under
either N2 or H2 atmospheres with a mixture of [l-'3C] glucose, [U-BC] glucose, and either
25 or 100 mM NaHCOg. The labeling conditions revealed exchange ﬂuxes in the C3
pathway and between the C3 and C4 pathways. The effects of NaHCO3 and H2
concentrations on A. succinogenes fermentations went beyond their roles as C4 pathway
substrates for succinate production. NaI-IC03 concentration affected the amount of ﬂux
shunted from the C4 to the C3 pathway. Fluxes within the C3 pathway changed to
compensate for different reductant demands at different NaHC03 concentrations. H2 also
affected the C3 pathway ﬂuxes by alleviating the need for N ADH production by pyruvate
and forrnate dehydrogenases. l3C-metabolic ﬂux analyses were complemented by A.
succinogenes draft genome sequence information. The genome sequence was also
analyzed for its industrially relevant features. In summary, A. succinogenes metabolism is
complex with ﬂux distribution to succinate versus alternative fermentation products
occurring at four different nodes. The metabolism is also ﬂexible as ﬂuxes change to

meet the reductant demands at different NaHCO3 and H2 concentrations.

ACKNOWLEDGEMENTS

This dissertation owes much to interactions I have had with the kind and helpful
individuals that embody the cooperative, creative, and productive atmosphere at MSU. I
acknowledge past and present Zeikus lab members, with whom I shared the daily joys
and frustrations that accompany the pursuit of discoveries. I am grateful to Pil Kim, the
postdoc from whom I inherited this project, for constantly discussing his results and ideas
with me despite my lack of knowledge and experience at the time. He was a true friend
and a wonderfully amusing, though somewhat seedy, ambassador of Korean culture. I am
also grateful to Maris Laivenieks for his advice, training and friendship. Harini
Krishnarnurthy was a great friend with whom I discussed all matters of life and research.
She put up with my jokes, which were usually at her expense, and my positive experience
in the lab owes much to her. I am grateful to Sui] Kang, Hoon Song, Yoshi Sasaki, Karla
Ziegelman-Fjeld, and Bryan Schindler for their friendship in the lab. I was fortunate to
have worked with bright undergraduate researchers Andy Boyle, Kirk Burkart, Greg
Costakes, Adam Edmunds and Vicki Harkins, from whom I learned more in attempting
to train them than they likely learned from me. The independence given to me by my
mentor Greg Zeikus, and his constant support of my research directions/tangents, have
been an important part of my professional development. I am honored to make a
contribution to his rich history of scientiﬁc achievements. I am indebted to my co-mentor
Claire Vieille, who allowed me to work independently and always gave thoughtful
consideration to even the smallest of my ideas or concerns, even when she was burdened

by the enormous task of writing multiple grants. She has been an ideal mentor and a great

iv

friend and I am conﬁdent that her versatility, creativity, and leadership will earn her a
well-deserved position to continue the exciting and applied projects the lab is known for.
I thank my committee members, Drs. Michael Bagdasarian, John Breznak, C.
Adinarayana Reddy, and Yair Shachar-Hill for their guidance, advice, and thought-
provoking questions. I additionally acknowledge Dr. Breznak whose notoriously
infectious enthusiasm for research somehow had its effect on me during a technical
conversation about gas chromatography at a time of low morale during my Masters
project. I have carried his enthusiasm ever since. I also acknowledge Dr. Shachar-Hill
who devoted considerable time and thought to develop my metabolic ﬂux analysis skills.
I attribute my success in securing a postdoc position largely to his thoughtful advice. I am
indebted to the Shachar-I-Iill and Ohlrogge labs, particularly to a group of inspired and
resourceful postdocs: Drs. Doug Allen, Ana Alonso, Eva Collakova, Fernando Goffman,
Igor Libourel, and J org Schwender. I especially thank Dr. Schwender, who devoted hours
of his time on short notice to help me prepare for my comprehensive exam, and Dr.
Allen, who gave every one of my emails or concerns much thought and a prompt reply. I
thank Peter Bergholz for being a good friend and for helpful advice on the manual
genome sequence annotation. I thank Trevor Wagner for sharing the knowledge
accumulated from his tireless trials in making gene knockouts. I thank all the friends I
have made at MSU, whose names and attributes deserve mention but would add a volume
to this dissertation. I acknowledge my family of self-proclaimed homebodies, who endure
my anomalous travels from home to study strange bacteria. Finally, I thank Nastya
Gridasova, who brought renewed joy to things in life I had long taken for granted. What a

perfect time to learn to take a second look at what’s in front of you.

TABLE OF CONTENTS

 

 

LIST OF TABLES __ ................................. ix
LIST OF FIGURES ............................................................. x
SYMBOLS AND ABBREVIATIONS _, xii

 

Chapter 1. Prospects for an Actinobacillus succinogenes-based succinate industry

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1.1 Introduction "1
1.2 The succinate market and the petrochemical competition ___________________________________ 2
1.3 Succinate-producing bacteria ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 7
1.3.1 Escherichia coli ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 8
1. 3. 2 Corynebacterium glutamicum .14
1.3.3 Anaerobiospirillum succiniciproducens 18

l. 3. 4 Rumen Pasteurellaceae: Actinobacillus succinogenes and Mannheimia
succiniciproducens ,2 1
1.4 Backgound on l3C-metabolic ﬂux analysis ,,,,,,,,,,,,,,,,,,,,,, 26
1.5 Objectives ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 34
1.6 References ______ 37

Chapter 2. Insights into Actinobacillus succinogenes fermentative metabolism in a
chemically deﬁned growth medium 2 _ _ ________________ 46
2.1 ABSTRACT ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 47
2.2 INTRODUCTION _ ________________________ 48
2.3 MATERIALS AND METHODS __________________________ 50
2.3.1 Chemicals, bacteria, and culture conditions ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 50
2.3.2 Identiﬁcation of a deﬁned growth medium, AM3 "50
2.3.3 Growth of A. succinogenes on AM3 solid agar medium ___________________ 52

2.3.4 Determination of growth trends and fermentation balances in AM3
and Medium A 52

 

2.3.5 Determination of fermentation balances, growth rates, and product
formation rates in AM3 with different NaHCO3 concentrations 53

 

 

2.3.6 Preparation of crude cell extracts and enzyme assays 53
2.3.7 Test of potential glutamate precursors to support growth _________________ 54
2.4 RESULTS AND DISCUSSION ,,,,,,,,,,, 56
2.4.1 Creation of the deﬁned growth medium, AM3 _________________________________ 56

2.4.2 Growth trends and fermentation balances in AM3 and Medium A_.57
2.4.3 Effect of AM3 NaHCO3 concentration on fermentation balances,

 

 

 

growth rates, and metabolic rates ....57

2.4.4 A. succinogenes is missing at least two TCA cycle-associated
enzyme activities ........................ 61
2.5 ACKNOWLEDGMENTS _ __ ............................ 66
2.6 REFERENCES 67

vi

Chapter 3. Determining Actinobacillus succinogenes metabolic pathways and ﬂuxes
by NMR and GC- MS analyses of 3C-labeled metabolic product isotopomers 69

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.1 ABSTRACT ,,,,,,,,,,,,,,,,,,, 70
3.2 INTRODUCTION ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 71
3.3 MATERIALS AND METHODS _ ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 76
3. 3. 1 Chemicals, bacteria, and culture conditions ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 76
3. 3. 2 Growth and sampling conditions 1n the presence of
[1- C] glucose ,,,,,,,,,,,,,,,,,,,,,,,,, 76
3.3.3 HPLC analysis of glucose and products and GC analysis of
headspace gas ,,,,,,,,,,,,,,,,, 77
3. 3. 4 Determination of extracellular ﬂuxes 77
3. 3. 5 GC- MS analysis of amino acids and organic acids 78
3. 3. 6 NMR analysis of organic acids, alanine, and glucose monomers from
glycogen ______ 79
3. 3. 7 Metabolic modeling and ﬂux analysis .. ________________ 80
3. 3. 8 Analytical techniques for determining A. succinogenes cellular
composition ____________________________________________ 87
3.3.9 Enzyme assays ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 90
3.4 RESULTS ,,,,,,,,,,,,,,,,,,,,,,, 92
3.4.1 Determining A. succinogenes’ cellular composition ,,,,,,,,,,,,,,,,,,,,,,,,,, 92
3.4.2 Conﬁrming a pseudo-metabolic steady state ______________________________________ 92
3.4.3 Metabolic pathway delineation and ﬂux quantiﬁcation ,,,,,,,,,,,,,,,,,,,,, 94
3.4.4 The glyoxylate cycle _ ___________________________ 98
3.4.5 The oxidative pentose phosphate pathway ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 99
3.4.6 The Entner-Doudoroff pathway _____________________ 102
3.4.7 C3 pathway dehydrogenases and transhydrogenase 102
3.4.8 Malic enzyme and OAA decarboxylase , , 104
3.5 DISCUSSION _______________________ 108
3.6 ACKNOWLEDGEMENTS ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 113
3.7 REFERENCES _ _________________ 114
Chapter 4. l3C-metabolic flux analysis of Actinobacillus succinogenes fermentative
metabolism at different NaHCO3 and H2 concentrations ____________________________________________ 120
4.1 ABSTRACT ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 120
4.2 INTRODUCTION _____________________________________ 121
4.3 MATERIALS AND METHODS __________________________________ 125
4.3.1 Chemicals, bacteria, and culture conditions ...................................... 125
4.3.2 Growth and sampling conditions in the presence of l3C-glucosem126
4.3.3 Analytical procedures __ 126
4.3.4 Enzyme assays ......................... 127
4.3.5 Metabolic modeling and ﬂux analysis _________________ 128
4.4 RESULTS 130
4.4.1 Conﬁrmation of pseudo-metabolic steady state and choice of
substrate isotopomers ,,,,,,,,,,,,,,,,,,,,, 130

 

vii

4.4.2 Modifying the A. succinogenes ﬂux model based on results from

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

growth condition C ,,,,,,,, 131
4.4.3 Effects of NaHCO3 and H2 concentrations on metabolic ﬂuxes _____ 139
4.5 DISCUSSION 149
4.6 REFERENCES "157
Chapter 5. Genomic perspective on the potential of Actinabacillus succinogenes for
industrial succinate production _ _____ 161
5.1 ABSTRACT _________________________________ 161
5.2 INTRODUCTION ______________________________ 162
5.3 MATERIALS AND METHODS _______________________________________ 164
5.3.1 Source strain and genomic DNA puriﬁcation ___________________________________ 164
5.3.2 Sequencing and automated annotation ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 164
5.3.3 Manual annotation .. _______________________ 165
5.4 RESULTS AND DISCUSSION _______________________________ 166
5.4.1 General features of the draft genome sequence ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 166
5.4.2 Metabolic reconstruction ,,,,,,,,,,,,,,,,,,,,,,,,, 166
5.4.3 Potential for engineering by natural transformation ,,,,,,,,,,,,,,,,,,,,,,,,, 178
5.4.4 Potential for non-pathogenicity ................ 181
5.5 REFERENCES ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 187
Chapter 6. CONCLUSIONS and FUTURE DIRECTIONS _____ 194
6.1 The current understanding of A. succinogenes fermentative metabolism ,,,,, 194
6.2 Future research and development of A. succinogenes metabolism _________________ 200
6. 2.1 Oxaloacetate decarboxylase and malic enzyme __________ 200
6. 2. 2 The effects of CO2 concentration on PEPCK and C4-decarboxylat1n g
ﬂuxes ________________________ 203
6.2.3 Pyruvate dehydrogenase and forrnate dehydrogenase ______________________ 204
6.2.4 The C3 pathway ,,,,,,,,,,,,,,,,,,,,,,,, 205
6.2.5 Importance of the A. succinogenes genome sequence ______________________ 208
6.2.6 Developing genetic tools for replacing A. succinogenes
chromosomal DNA segments ______________________ 210
6.3 REFERENCES 214

 

viii

LIST OF TABLES

1.1 Comparison of succinate fermentations from glucose by different bacterial speciesul6
2.1 Log phase fermentation balances of A. succinogenes in AM3 and Medium A ____________ 58

2.2 Effect of NaHCO3 concentration on endproduct distribution and growth rate in
AM3 60

 

2.3 Effect of N aHCO3 concentration on speciﬁc metabolic rates and estimated ﬂuxes___.62
2.4 Ability of glutamate precursors to support growth of A. succinogenes in AM3 _________ 64
3.1 A. succinogenes metabolic intermediate and cofactor requirements for biosynthesis_82

3.2 Mass isotopomer distributions of TBDMS-amino acid and -organic acid
fragments 84

3.3 Growth and metabolic parameters ___________________________________________________________________________________ 94

3.4 Measured and simulated percent l3C-enrichments in alanine, glucose, and organic

 

 

acids as determined by NMR ______ 95
3.5 Enzyme activities in cell extracts of A. succinogenes grown in AM3 with 150 mM
NaHC03 __ ,,,,,,,,,,,,,, 103
4.1 Selected TBMDS-amino acid and -succinate fragments obtained from A.
succinogenes grown without H2 in AM3 with 100 mM NaHCO3 (Condition C) _____________ 133
4.2 Growth parameters observed for the different growth conditions _______________________________ 140
4.3 Effect of NaHCO3 concentration and H2 on net and exchange ﬂuxes ________________________ 143

4.4 Enzyme activities in cell extracts of A. succinogenes grown in different NaHC03 and
H2 conditions 144

5.1 Central metabolic enzymes expected and identiﬁed in the draft genome sequenced 67
5.2 Dicarboxylate transporters identiﬁed in the draft genome sequence ,,,,,,,,,,,,,,,,,,,,,,,,,,, 175

5 .3 A. succinogenes ORFs encoding uptake and degradation pathways for sugars other
than glucose , 177

5 .4 A. succinogenes homologs of H. inﬂuenzae competency proteins ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 181

ix

LIST OF FIGURES

1.1A Simpliﬁed metabolic map of wild-type E. coli mixed acid fermentation __________________ 9

1.1B Simpliﬁed metabolic map of engineered E. coli anaerobic succinate-producing
metabolism 11

---------------------

 

1.1C Simpliﬁed metabolic map of engineered E. coli aerobic succinate-producing
metabolism____ 1 3

 

1.2 Simpliﬁed map of wild-type C. glutamicum succinate-producing metabolism __________ 17

1.3 Simpliﬁed map of wild-type A. succiniciproducens succinate-producing
metabolism 18

 

1.4 Simpliﬁed map of wild-type A. succinogenes and M. succiniciproducens succinate-
producing metabolism 23

1.5 Determining relative EMP and OPPP ﬂuxes from GC-MS measurements of serine._3l

1.6 Estimation of intermediary metabolic ﬂuxes from product isotopomers using 13C-

 

FLUX ,,,,,,,,,,,,,,,,,,, 33
2.1 Simpliﬁed metabolic map of A. succinogenes central metabolism _______________________________ 49
2.2 Possible enzyme activities leading to glutamate synthesis ____________________________________________ 64
3.1 A. succinogenes metabolic pathways addressed in this study ________________________________________ 73
3.2 A. succinogenes batch fermentation characteristics in AM3 ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 93

3.3 A. succinogenes metabolic ﬂuxes in AM3 with 150 mM NaHCO3 obtained after
ﬁtting of experimental and simulated data sets in I 3C—F lwr ________________________________________________ 96

3.4 Acetyl-CoA and OAA isotopomers expected from forward glyoxylate cycle ﬂux ..... 99
3.5 Isotopomers expected from OPPP and ED pathway ﬂuxes _________________________________________ 101
3.6 Two-step illustration of the ﬂuxes resulting in [2-‘3C]pyruvate __________________________________ 105

4.1 Simpliﬁed central metabolic pathways known to be active during A. succinogenes
mixed acid fermentation, 123

 

 

4.2 Two-step illustration of ﬂuxes leading to isotopomers that provide information on C4
pathway reversibility and on the C4-decarboxylating ﬂux 132

 

4.3 Glycine synthesis and C1 metabolism used in the metabolic ﬂux model ____________________ 136

4.4 Modiﬁed metabolic ﬂux model of A. succinogenes mixed acid fermentative
metabolism___ ,,,,,,,,,,,,,,,,,,,,,, _ 1 38

 

5.1 Central metabolism based on identiﬁed enzymes in the draft genome sequence ....... 168

5.2 Physical map of the OAA decarboxylase operon region in the A. succinogenes (top)
and K. pneumoniae (bottom) genomes .. _ 172

ooooooooooooooooooooooooooooo

 

5.3 Uptake and degradation pathways for sugars other than glucose based on genes
identiﬁed in the draft genome sequence _ _ 178

 

5.4 Nucleotide frequencies in 1,454 A. succinogenes uptake signal sequences _______________ 180

xi

SYMBOLS AND ABBREVIATIONS
AcCoA, acetyl-coenzymeA
Ace, acetate
ADH, alcohol dehydrogenase
AK, acetate kinase
Cit, citrate
CL, citrate lyasc
C4dec, C4—decarboxylating ﬂux
ED, Entner Doudoroff pathway
EMP, Embden-Meyerhoff-Pamas or glycolysis
EtOH, ethanol
E4P, erythrose—4-phosphate
For, formate
ForDH, formate dehydrogenase
Fm, fumarase
FR, fumarate reductase
Fum, fumarate
F6P, fructose-6-phosphate
Glc, glucose
Glxt, glyoxylate
G3P, g1yceraldehyde-3-phosphate
G6P, glucose-6-phosphate

ICL, isocitrate lyase

xii

Mal, malate

MDH, malate dehydrogenase

ME, malic enzyme

Msyn, malate synthase

OAA, oxaloacetate

OAAdec, oxaloacetate decarboxylase
OPP, oxidative pentose phosphate pathway
PEP, phosphoenolpyruvate

Pyr, pyruvate

PEPCK, PEP carboxykinase

PFL, pyruvate formate-lyase

PK, pyruvate kinase

pppl and 2, transketolase

ppp3, transaldolase

PyrDH, pyruvate dehydrogenase

PTS, PEstugar phosphotransferase system
R5P, pentose-phosphates

Suc, succinate

S7P, sedoheptulose-7-phosphate

upt, glucose phosphorylation by hexokinase and PTS.

xiii

Chapter 1

Prospects for an Actinobacillus succinogenes-based succinate industry

1.1 Introduction

Most bulk and specialty chemicals are derived from crude oil. Declining reserves,
rising prices, and concerns over the environmental impacts of oil-based industries have
prompted a movement towards bio-based processes. Succinate is currently produced
petrochemically to satisfy a specialty chemical market, but it could also be produced by
microbial fermentations. More importantly, succinate could serve as the starting material
for producing bulk chemicals such as 1,4-butanediol (a precursor to “stronger-than-steel”
and biodegradable plastics), ethylene diamine disuccinate (a biodegradable chelator),
diethyl succinate (a “green” solvent replacement for methylene chloride), and adipic acid
(a nylon precursor) (104). Bio-based succinate production also involves CO2 ﬁxation.
Thus, in addition to being based on renewable resources, succinate production also has
the environmental beneﬁt of using a greenhouse gas as a substrate. For bio-based
succinate to serve as an intermediary chemical feedstock for bulk chemicals, its
production cost must be made competitive with petrochemical alternatives (e.g., maleic
anhydride). Creating a bio-based succinate industry will require advances on several
fronts, including strain engineering, medium optimization, and succinate puriﬁcation.

For reasons discussed later in this chapter, Actinobacillus succinogenes is one of
the best candidate organisms for industrial succinate-production. However, it also
produces unwanted forrnate, acetate, and ethanol. Eliminating these fermentation

byproducts would allow for efﬁcient substrate conversion into succinate, and for less

complicated succinate puriﬁcation procedures. Thus, A. succinogenes must be
metabolically engineered towards a homosuccinate fermentation. The purpose of this
chapter is four-fold: (i), to describe the bulk chemical market targeted by bio-based
succinate production, and the obstacles that must be overcome; (ii), to compare A.
succinogenes to other industrially relevant succinate-producing bacteria; (iii), to provide
a background on l3C-labeling studies and their usefulness for understanding microbial
metabolism and guiding metabolic engineering; and (iv), to outline the objectives of the

work described in this dissertation.

1.2. The succinate market and the petrochemical competition

Succinate is currently used as a surfactant, an ion chelator, and in the
pharmaceutical and food industries. However, this specialty chemical market is relatively
small at 33 M lb in the US (69). The research and development of bio-based succinate
production is targeting a much larger commodity chemical market. In 1999, Zeikus et al.
described a $15 billion network of bulk chemicals, mostly solvents, resins, and plastics,
that could be produced from succinate as an intermediary feedstock (104). Here I revisit a
number of these chemicals that could be derived from succinate with respect to their

current markets:

1,4-Butanediol (BDO). The demand for BDO (equaling production plus imports minus
exports) increased from 700 M lb in 1999 to 900 M lb in 2004, and it is projected to reach
1,055 M lb in 2008 (2). BDO’s price has also increased from 1999-2004 from $1.00 to

$1.25 per lb. Fifty-one percent of BDO is used to make tetrahydrofuran (THF); twenty

percent is used to make 7 -butyrolactone (GBL); while another twenty percent is used to
make polybutylene terephthalate high-performance resins for the automotive and

electronics industries. BDO can be produced from succinate in a two-step process (69).

THF. Most THF is used to manufacture thermoplastic urethane elastomers and
polyurethane ﬁbers, but it is also an important solvent (3). A process for making a

mixture of BDO, THF, and GBL from succinate in a single step has been described (69).

GBL. New regulations on volatile organic chemical usage have resulted in new markets
for solvents such as GBL and its primary derivative, N-methyl-2-pyrrolidone, for
replacing methylene chloride. The US market size for N-methy1-2-pyrrolidone is 80 M lb.
GBL can also be converted to 2-pyrrolidone, a plasticizer and solvent with a US market
size in excess of 65 M lb (69). Succinate can be converted to GBL as the sole product or

to GBL-containing mixtures as mentioned above (69).

Succinate salts. About 210 M lb of salt are used each year as deicers. Succinate salts
could replace current salts by offering improved performance characteristics and more

neutral effects on surfaces and the environment (69).

Adipic acid. Adipic acid is primarily used to make nylon 6/6. The current US market for
adipic acid is ~ 2 M lb (69). Benzene is currently the main feedstock for adipic acid
production and benzene prices have increased dramatically from $1.53 per gallon in 2003

(89) to $3.85 in 2005 (5). Two alternative methods for making adipic acid involve

succinate. One method uses succinate as a ligand in a ‘green’ production method that
uses hydrogen peroxide instead of nitric acid, to avoid toxic nitrous oxide emissions (13,
18). The other method, although not yet competitive with the petrochemical manufacture,
hydrogenates succinate to BDO, which is then carbonylated to adipic acid (13, 69, 104).
Succinate, adipic acid, BDO, and ethylene glycol can also be combined to make the
biodegradable plastic, Bionelle (24).

Succinate can currently be produced by fermentation for about 25-50 ¢/Ib (69). At
this production cost, the succinate-derived chemicals described above (with the exception
of succinate salts and adipic acid), are competitive or nearly competitive with the current
petrochemical manufacture costs (69). As a result, succinate-producing facilities are
beginning to emerge. Mitsubishi Chemical (Japan) is planning to make enough bio-based
succinate to produce 30,000 M tons/yr of the biodegradable polymer polybutylene
succinate (100). Another Japanese initiative, the Research Institute of Innovative
Technology for the Earth and Showa Highpolymer Co., Ltd. is moving towards industrial
succinate production using a genetically modiﬁed strain of Corynebacterium glutamicum.
The group plans to produce 50,000 tons succinate/yr using wastepaper as the feedstock
(www.tifac.org.in/offer/tsw/japenv.htm). DSM (Netherlands) and Diversiﬁed Natural
Products (East Lansing, MI) have also agreed to commercialize bio-based succinate-
derived products (100).

These bio-based succinate-production facilities are targeting specialized markets,
but large-scale production for bulk chemical synthesis is still a distant prospect. Unlike
other bio-based industrial chemicals (e.g., citrate, glutamate, and lysine), for which

biological routes had clear economic advantages over petrochemical routes (99), bio-

based succinate targets a thriving petrochemical market, particularly that of maleic
anhydride. Rising oil prices are resulting in record high petrochemical prices, including
that of butane from which maleic anhydride is derived. However, maleic anhydride is in
high demand thanks to strong housing and construction markets as well as increasing
utilization by the auto and marine craft industries (8). In response to this demand,
Huntsman Corp. (Salt Lake City, UT), which currently operates a 235 M lb maleic
anydride plant, is increasing its production capacity by building a new 100 M 1b capacity
plant (8, 9). Lanxess corp. (Germany) is also increasing its US maleic anhydride capacity
to 160 M 1b (4). Thus, while high oil prices are drawing interest to bio-based chemicals,
bio-based succinate production has yet to be made competitive with a booming
petrochemical market.

Bio-based succinate production facilities will have to be constructed where the
feedstock (e. g., corn or agricultural waste) is produced to minimize transport costs.
Further savings could be made by producing hi gh-value co-fermentation products.
Succinate was suggested as a suitable co-product to improve the economics of industrial
ethanol fermentations (58). An alternative process that is more focused on large-scale
succinate production is to produce ethanol and succinate in separate fermentations but in
the same facility. This process would allow CO2 waste from the ethanol fermentation to
be used as substrate for succinate production (104).

Considerable production cost is associated with succinate puriﬁcation. Several
methods have been described to purify succinate from fermentation broth. Succinic acid
was purifyied from 80-1Anaerobiospirillum succiniciproducens cultures by

electrodialysis (6) followed by processing on a bipolar, water-splitting membrane stack to

produce succinic acid from succinate, then ion exchange chromatographies (21). The
recovered succinic acid (80% of recovered dry wt.) was almost entirely free of
nitrogenous material, but acetic acid was present at 20% of the recovered dry weight (21).
Electrodialysis methods are also inherently expensive (7). Another succinate puriﬁcation
method from 80-1 A. succiniciproducens cultures resulted in higher succinic acid purity
(94% of recovered dry wt.) but still contained acetic acid (5% recovered dry wt.) and
protein (1% of recovered dry wt.) (14). The method involved adding Ca(OH)2 to control
the fermentation pH and precipitate succinate. Succinic acid was released from the
precipitate by adding sulfuric acid, then puriﬁed by ion exchange chromatographies (14).
While this method resulted in a higher succinate purity, it also produced commercially
unusable CaSO4 (gypsum) (7). A cheaper method for purifying succinic acid focused on
using lower quantities of reagents, and thereby produced less byproducts (7). This
method involved producing di-ammonium succinate while maintaining the fermentation
pH (e.g., by adding NH40H), then reacting the material with ammonium bisulfate to form
ammonium sulfate and succinic acid. Sulfuric acid was added to decrease the pH below
2, causing succinic acid to crystallize. After collecting by ﬁltration, succinic acid was
dissolved in methanol, which also precipitates contaminating sulfates. The methanol was
evaporated, condensed, and recycled, and the ammonium sulfate was converted to
ammonium bisulfate for reuse by cracking at 300°C. This method was demonstrated
using Escherichia coli AFP-l 1 1, and succinic acid accounted for over 90% of the
recovered dry weight, close to the theoretical yield of ~ 95% (7). Reactive extraction is
another method that was used to purify succinic acid from fermentation broths. Tri-n-

octylamine extracts only undissociated carboxylic acids. Thus, by controlling the pH of

the extraction, undesired organic acids (e.g., acetic and pyruvic acids) were extracted,
leaving succinic acid and residual glucose behind. The residual glucose was removed by
a low pH crystallization at 4°C. This technique was demonstrated on Mannheimia
succiniciproducens fermentation broths (31, 32) achieving a ﬁnal succinic acid yield of
73.1% and a purity of 99.8% (32). The large-scale economics of using trim-octylamine,
and whether it can be reused, have not yet been addressed, or at least not reported.
Regardless of the method, purifying succinic acid is a signiﬁcant portion of the total cost.
Thus, using an organism that produces succinate at near-maximum theoretical yields with

little or no byproducts will contribute to cost-effectiveness.

1.3. Succinate-producing bacteria
Bio-based succinate will be produced from the most abundant sugars in plant.

biomass (i.e., glucose, fructose, arabinose, and xylose). Glucose is the most commonly
used substrate for research related to industrially relevant fermentations. Theoretically,
~1.71 mol succinate could be produced per mol glucose, based on the available electrons
and given additional CO2 (i.e., 24 electrons in glucose divided by 14 electrons in
succinate):

glucose + 0.86 HCO3' —> 1.71 succinatez' + 1.74 H2O + 2.58 11+ AG°' -173 kJ/mol
In the presence of both CO2 and additional reducing power, (e. g., H2), the theoretical
yield increases to 2 mol succinate per mol glucose:

glucose + 2 HCOg' + 2H2 -) 2 succinatez' + 2H2O + 2 H+ AG°' -317 kJ/mol
These theoretical yields are the targets sought for bacterial succinate production.

Experimental yields will be constrained by the pathways used and by the carbon diverted

to biomass and alternative products. Succinate titers and productivities aim to be similar
to those obtained in industrial glutamic acid production: ~150g 1'1 at 5 g l'1 h".

A wide variety of bacteria and fungi excrete succinate (86), but only a few
produce enough to be considered for industrial succinate production. These organisms
have either been engineered to produce high succinate concentrations or do so naturally.
This section focuses on ﬁve of the most promising bacteria for industrial succinate
production. For each bacterial species, the metabolic routes used for succinate production

are described along with their advantages and disadvantages.

1.3.1 Escherichia coli

E. coli naturally produces succinate from the reductive branch of the tricarboxylic
acid (TCA) cycle when fermenting glucose, but only as a minor product of a mixed acid
fermentation (Fig. 1.1A). Because E. coli is one of the best-understood bacteria, and
because it can be genetically engineered with relative ease, there has been considerable
effort to engineer E. coli for succinate production. Early efforts focused on increasing
phosphoenolpyruvate (PEP) and pyruvate carboxylations to redirect carbon ﬂux to the
reductive TCA branch. By overexpressing E. coli PEP carboxylase (PEPC), succinate
yields were improved 3.75 times to 10.7 g l'1 (61). Succinate production was also
enhanced in strains overexpressing Sorghum vulgare PEPC (56). Overexpressing E. coli
PEP carboxykinase (PEPCK) was originally shown to have no effect on succinate
production in a medium containing 15 g MgC03/1 (61), but when NaHC03 (more soluble
than MgC03) was used, succinate concentrations increased (45). Overexpressing A.

succinogenes PEPCK in a ppc' E. coli strain also improved succinate production (40).

ATP
Glc A——» GGP

Gch
EMF
2 mom/i PTS
C02
Lac
PEPC PEP PM ATP MD"
NADH %H

OAA Pyr
MDH CO2 FDH
For

PFL
Mal

Fm H2

Fum AcCoA

OCH 2 NM)“
Frd AD” ATP
AK

Suc EtOH Ace

Figure 1.1A. Simpliﬁed metabolic map of wild-type E. coli mixed acid fermentation.
The pentose phosphate pathway is not shown. Metabolites: AcCoA, acetyl-CoA; Ace,
acetate; EtOH, ethanol; For, forrnate; Fum, fumarate; G6P, glucose-6-phosphate; Glc,
glucose; Lac, lactate; Mal, malate; OAA, oxaloacetate; PEP, phosphoenolpyruvate; Pyr,
pyruvate; Suc, succinate; and QOH, menaquinol. Reactions: ADH, alcohol
dehydrogenase; AK, acetate kinase; EMP, Embden-Meyerhoff—Parnas pathway; FDH,
formate dehydrogenase; Fm, fumarase; F rd, fumarate reductase; Gch, glucokinase;
LDH, lactate dehydrogenase; MDH, malate dehydrogenase; PEPC, PEP carboxylase;
PFL, pyruvate formate-lyase; PTS, PEP: glucose phosphotransferase system; and Per,
pyruvate kinase.

Increased pyruvate carboxylation leading to improved succinate production was achieved
by overexpressing pyruvate carboxylase (PYC) from Lactococcus lactis (56) or
Rhizobium etli (23). Succinate production in strains overexpressing PEPC or PYC was
further enhanced by overexpressing E. coli pantothenate kinase and supplying
pantothenic acid to increase intracellular acetyl-CoA levels (57). Acteyl-CoA is an
activator of PEPC and PYC. Increased pyruvate carboxylation was also achieved using

I the E. coli mutant NZNl 11. In this strain, pyruvate is a fermentation product due to

insertional mutations in the pyruvate formate-lyase and lactate dehydrogenase genes (29,
62, 88). When E. coli NAD+-dependent malic enzyme is overexpressed in NZN] l 1,
pyruvate is carboxylated, and succinate is produced as the major fermentation product
(29, 62, 88). Unfortunately, NZN 1 11 does not grow fermentatively on glucose and
produces succinate slowly when overexpressing malic enzyme (88).

A major breakthrough in E. coli succinate production was the isolation of a
spontaneous mutant of NZN 1 1 1, AFPl 11, that grows fermentatively on glucose,
produces succinate as the main fermentation product, and consumes glucose and other
sugars simultaneously (10). The AFPI 11 mutation was found in the ptsG gene, encoding
part of the glucose-speciﬁc phosphotransferase system (10). This mutant produces up to
50 g succinate/l using light steep water as an inexpensive alternative to yeast extract and
tryptone (62). Using AFPl 11 expressing R. etli PYC, the effects of fermentation
conditions on succinate production were explored by using an aerobic growth phase
followed by an anaerobic succinate production phase (94). From these studies, Vemuri et
al. discovered a new route for succinate production involving the glyoxylate shunt, which
was activated during aerobic growth (94). The glyoxylate shunt can partially alleviate the
reductant limitation associated with using the reductive TCA branch for succinate
production. If all reductant comes from glycolysis (i.e., 2 NADH/glucose) then only one
succinate can be produced per glucose by the reductive TCA branch (Fig. 1.1A).
Glyoxylate shunt ﬂux complements reductive-TCA branch ﬂux by using less reductant.
Use of the glyoxylate shunt might even generate reductant through if acetyl-CoA
production (used by the glyoxylate shunt for making citrate) is accompanied by forrnate

oxidation or if pyruvate dehydrogenase is used in place of pyruvate-formate lyase (Fig.

10

1.1B). After an aerobic growth phase, AFPl 11 produced 99.2 g succinic acid 1'1 in an
anerobic fed batch fermentation (93), one of the highest succinate titers ever reported

(Table 1.1).

were .127 cap

 

Fum
0" {Fm
Suc Gm '0" EtOH Ace
ICL

Figure 1.18. Simplified metabolic map of engineered E. coli anaerobic succinate-
producing metabolism (80). Grey arrows indicate missing enzyme activities due to gene
deletions. The double-lined arrow indicates introduced activity. Metabolites: Glxt,
glyoxylate; and Icit, citrate/isocitrate. Reactions: CS, citrate synthase and aconitase; ICL,
isocitrate lyasc; Msyn, malate synthase; and PYC, pyruvate carboxylase. See Fig. 1A for
other abbreviations.

With a knowledge of complementary pathways for succinate production, wild-
type E. coli was re-engineered to eliminate fermentation byproducts and guarantee
glyoxylate shunt ﬂux (Fig. 1.18). The ldhA, adhE, and ack-pta deletions prevented

lactate, ethanol and acetate productions, and deleting iclR (encodes the transcriptional

repressor of the glyoxylate cycle) ensured the expression of glyoxylate cycle enzymes

11

(80). As was observed with AFPl 11 (94), though, the iclR deletion was not required for
glyoxylate shunt ﬂux (80). Finally, L. lactis PYC was overexpressed to divert ﬂux from
pyruvate to oxaloacetate (OAA), a precursor to both succinate producing pathways (80).
This strain, SBSSSOMG/pHIA13, produced 66 g succinate/l with a 1.6 mol succinate per
mol glucose yield (80). This yield is the maximum yield predicted by elementary ﬂux
mode analysis of the engineered metabolic network (12). One drawback of using the
glyxoylate shunt to produce succinate is that it wastes carbon through CO2 production.
Three PEP are required to generate the l OAA + 2 acetyl-CoA needed to produce 2
succinate using glyoxylate shunt, as opposed to 1 PEP for l succinate using the reductive
TCA branch, given ample reductant and CO2. Sanchez et al. showed that, for succinate
production, the optimal ﬂux partitioning at OAA to the glyoxylate (i.e., to form citrate)
and reductive TCA pathways is 0.32: 0.68 (79). Thus, while the theoretical yield
decreases to 1.6 succinate/glucose (12), the optimal ﬂux through this metabolic network
still results in net CO2 consumption.

E. coli was also engineered to produce succinate aerobically (55) (Fig. 1.1C). By
deleting sdhA, encoding succinate dehydrogenase, succinate became the terminal product
of the oxidative TCA-branch. By deleting iclR (the transcriptional repressor of the
glyoxylate cycle), the glyoxylate cycle formed a second route to succinate. Deleting poxB
and ackA-pta prevented acetate production. When ptsG was inactivated and S. vulgare
PEPC was overexpressed, a 1 mol succinate per glucose yield was achieved, the
maximum yield predicted from elementary ﬂux mode analysis of this metabolic network
(12). In fed batch aerobic cultures, the strain produced 58.3 g succinate l‘1 at a rate of 1.1

g 1‘1 h", opening the possibility for aerobic succinate production (54). This aerobic

12

system has several disadvantages, though. First, the maximum theoretical yield of 1
succinate per glucose is less efﬁcient than anaerobic processes. And second, maximum
succinate yields result in a net CO2 production, rather than CO2 ﬁxation (12). Thus, one

of the environmental beneﬁts of succinate production is not fulﬁlled.

ATP
Glc ﬂ?» GGP

EMP
2NAD
co,

   

co2

MDH UqOH
Mal
ﬂ...
Fum “‘1’"
“90" 51.111 "1"”
Glxt ATP
Sue '0“ Ace
co2 ’CL co2
MG D H ICDH
NADH NADPH
ATP aKG

Figure 1.1C. Simplified metabolic map of engineered E. coli aerobic succinate-
producing metabolism (54). Grey arrows indicate missing enzyme activities due to gene
deletions. The double-lined arrow indicates introduced activity. Metabolites: aKG, a-
ketoglutarate and UqOH, ubiquinol. Reactions: aKGDH, aKG dehydrogenase and
succinyl-CoA synthetase; ICDH, isocitrate dehydrogenase; PyrDH, pyruvate
dehydrogenase; and Perx, pyruvate oxidase. See Fig. 1.1A for other abbreviations.

The progress made in engineering E. coli to produce succinate is an excellent

- example of successful metabolic engineering. E. coli strains can now produce succinate at

13

concentrations comparable to those attained by natural succinate producers, and with high
succinate product ratios (Table 1.1). However, E. coli-based succinate production has
several disadvantages. First, the speciﬁc productivities of the engineered strains are
generally lower than those of natural succinate producers (Table 1.1). One of the lower E.
coli speciﬁc productivities is actually that of the aerobic succinate producing strain,
HL27659k/pKK313 (Table 1.1), which had been engineered to overcome the low
productivities of anaerobic fermentations (54). Second, dual-phase fermentations and
medium aeration introduce additional production costs. Third, with the exception of
AFPI 11 in light steep water (62), all E. coli strains have been studied in media containing
tryptone and yeast extract, which are too expensive for large-scale use. And fourth, all
strains overexpressing enzymes have used isopropyl-beta-D-thiogalactopyranoside to
induce expression, which is also too expensive for large-scale use. Continued research
and development of E. coli-based succinate production should focus on increasing
speciﬁc productivity, be characterized in industrially relevant media, and use constitutive

expression systems or inexpensive induction methods to overexpress enzymes.

1.3.2 Corynebacterium glutamicum

C. glutamicum is primarily known as the bacterium used for industrial amino acid
production. Intensive research from several groups has aimed to understand and improve
its amino acid production abilities. As a result, robust genetic tools exist for C.
glutamicum, and a wealth of knowledge is available, including genome sequence,
microarray, proteomic, and ﬂux analysis data (95). Although not traditionally considered

as a succinate producer, C. glutamicum is now poised to produce succinate on an

14

industrial scale (www.tifac.org.in/offer/tsw/japenv.htm). This advance was not the result
of genetic engineering but rather resulted from modifying its environment. In anoxic
conditions with bicarbonate, non-growing C. glutamicum converts glucose to succinate
and lactate with acetate as a minor product (33). This mixed acid fermentation reached
succinate and lactate titers of 23 g 1'1 and 95 g 1", respectively (65). At high cell densities
(60 g dry cell weight [DCW] 1“) succinate was produced at 11.7 g 1‘1 h" (65), with a
speciﬁc productivity of ~200 mg g DCW'l h". To increase the range of lignocellulose-
based sugars to be used as feedstock, C. glutamicum was engineered to consume xylose.
By completing the partial xylose catabolizing pathway with E. coli xylose isomerase, and
by overexpressing E. coli xylulokinase, C. glutamicum concomitantly converted glucose
and xylose to succinate and lactate (39). Succinate titer and speciﬁc productivity from
glucose and xylose were similar to those from glucose alone.

The C. glutamicum succinate-producing pathway (Fig. 1.2) was determined from
in vitro enzyme assays and multiple gene knockouts (33). OAA is converted to succinate
by the reductive TCA cycle branch. The glyoxylate shunt was not necessary for succinate
production, but it was not ruled out as a possible contributing route (33). OAA is
produced primarily from PEP by PEPC, with PEPCK and PYC perhaps making minor
contributions to OAA generation. When lactate dehydrogenase activity was impaired,
succinate and acetate were the sole endproducts but were not produced at higher titers.
By overexpressing PYC in an ldhA mutant, succinate and acetate production increased
~two—fold (33).

Guided by the vast knowledge on its metabolism and its reliable genetic tools, C.

glutamicum succinate production has moved from a research setting to an industrial

15

05 w. E .EocEEooA—oo “c3583. 5 owEEo 05 mm each .38 Eaton—veg 82:83 2.28% 2: ﬂ 25 20:3 ”a: x $138» u can qum: cease—«U

ANNV 0; Mo econ—O an cow 1 wE O? «o 2:? a main 622328 8:33» :8 >5 M

.25228 oEEo< _

omega couoscoa 0523.5 2 0223 539m 659% Eo¢ mchEmcmb Ba< c
82:8 .30» .o> 625% :on ”enough—H .EH mocoaombom ES :26: moo; E8 .30 ”88:0ch 05: :80 .EU 583-com .mi EBB .m c
22605 cocﬂcoEﬂ .550 2 22:83 «0 23M 0
.ﬁwv A oE: Am mmmEos «o anEn 2: E a was .oE: E omcaso

n
53835.5 oEoEEo> c

.8327. 3.582. E955 .3 88:.» Ea...“ $833558 0358.5 «o EamiamEoU €— 033—.

 

 

 

A V PM «mmustu
F— a” m I - c c d .
mm > m m «N oNé w _ cbo coo w _ VNm -5.» «E.U .53 EM. #33 SEA
Ami can 6% Mm v.m|o 34 34191.0 0mm m; @2102 chat—.3 .mmgmm—z
ESEEESSEE .3
. . 0.: . . . . 33.62% EMAUS‘E
E o . i «
Ana 9 r em ce :2: 3A A: A o: 2 Ned {50333.3cae<aehezrﬁ<
. when 6.2% .593 .963 .SA
3.0 EH 0%” mi am ”2 .6 3.0 8.2 _._ .wn + . . .
a w m -593 aged mSMMab—mmchNA:
. +93 EU. 6:333 .563
88 Eh..o>. mi ma Nd QNTSW hm; OMNAK a; no w.wo «
c m 10 SAVE “.333 .2339020mmmmm
:8 H
Ame to Acmd c am do 26 or 3 8 252:» Am
EEEESEM U
GNV 0% .30 um— NmIVm «:3 o; delmﬁo «:3 m N- o. N 0945 ENE ME
88 Eco Wm B c. To 5o Po EN A. 0 cc 252:, "Mom
ANNV A; .30 Am mmbm JED Au— nmdlwod «A5 A. A I N. A 3. do 0992:: MNem—
uwzuueﬁouau .<
$8 30 um mi. «.5 RB cod «.5 A. Tm. o Adm A.. 67$.
:8 .6 Am 3 3.5 2 85 as A .N 3n 252:,
Eng . . .. a .«A
A 1: 2:: A wwv A -c -38me A A3
_. c . _. _ A A . w _.
«om c6382 EV .250 6.605 22» Ego—68¢ T: A _ V 55 Boa 093950 ":Ebm
”cozﬁcoEuom oEE. . n 85 . .
:00 29:83 332 @3585 £625 a 2:625

 

16

0°" 4 Frd

Sue
K

 

Figure 1.2. Simplified map of wild-type C. glutamicum succinate-producing
metabolism. Dotted lines indicate enzyme activities with unknown in vivo contributions.
MEnz, malic enzyme; OAAdec, OAA decarboxylase, PEPCK, PEP carboxykinase. See
Fig. 1.1A for additional abbreviations.
setting within a relatively short time frame. Despite this rapid utilization of C.
glutamicum-based succinate production technology, there is still much to be understood
and improved.

As shown in Figure 1.2, there are several pathways whose contribution to organic
acid production is unknown. Speciﬁc succinate productivities by C. glutamicum are also
relatively low (Table 1.1). Most importantly, there are no reports of a C. glutamicum

strain that efﬁciently produces succinate in great majority to alternative products. As a

result, succinate titers and product ratios are relatively low (Table 1.1). The resulting

l7

product mixture may also require more complicated and expensive puriﬁcation

procedures, the details of which have not been reported.

Lac
PEPCK PEP PM ATP man
mm "P LDH
OAA Pyr co2
MDH PFR
Mal
Fm F“red
Fum AcCoA

2 NAD+
oou
Frd AD” ATP
AK

Suc EtOH Ace

Figure 1.3. Simpliﬁed map of wild-type A. succiniciproducens succinate-producing
metabolism. PFR, pyruvate ferredoxin oxidoreductase; Fermi, reduced ferredoxin. See
Fig. MA and Fig. 1.2 for additional abbreviations.

1.3.3 Anaerobiospirillum succiniciproducens.

A. succiniciproducens, isolated from the feces of humans and animals (85),
naturally produces succinate as its major fermentation product. The earliest metabolic
studies on A. succiniciproducens demonstrated the effects of pH and C02 availability on
the succinatezlactate ratio (21, 78). High C02 concentrations and a pH near 6, rather than
7, shifted metabolism entirely away from lactate to succinate and acetate at a ratio of
1.821 (21, 78). Enzyme activities detected in cell extracts suggested that glucose was

metabolized to PEP by glycolysis. Pyruvate-ferredoxin oxidoreductase was found to

convert pyruvate to acetyl-CoA. PEPCK was implicated as the major COz-ﬁxing enzyme,

18

generating OAA that was then converted to succinate by the reductive TCA branch (Fig.
1.3). Both A. succiniciproducens pyruvate-ferredoxin oxidoreductase and PEPCK
activities responded to pH and C02 concentration changes (78). The energy-conservin g
role of PEPCK is unlike its more commonly known role in gluconeogenesis and, at the
time, there were few examples of PEPCK having this role, except during propionate
fermentation by Bacteroides species from the rumen (82) and human intestine (59). A.
succiniciproducens PEPCK was puriﬁed and characterized (73). It has been characterized
extensively by site-directed mutagenesis and x-ray crystallography (11, 34-38). A.
succiniciproducens can also use H2 as an additional electron source, and supplying H2
favors succinate production (102). Pyruvate-ferredoxin oxidoreductase activity also

decreased signiﬁcantly in the presence of H2 (102).

A. succiniciproducens’s fermentation characteristics were analyzed in industrially
relevant media. Taking advantage of A. succiniciproducens ability to consume lactose,
whey was examined as a cheap substrate source in batch, fed-batch, and continuous
culture conditions (50, 77). Com-steep liquor (77), or both yeast extract and peptone (50)
were added to the whey media as nitrogen sources. Under fed-batch and batch conditions,
succinate yields over 91 % were achieved (50, 77). When glucose was included in whey
fermentations, lactose consumption did not occur until glucose was ﬁrst consumed,
indicating that A. succiniciproducens will not naturally co-metabolize sugars (50). Wood
hydrolysate was also explored as a cheap substrate source with yeast extract, peptone, or
com-steep liquor as nitrogen sources. Succinate yields near 90% were achieved in all
cases (49). Glycerol is a byproduct of fatty-acid producing industries, including bio-

, diesel production. Although glycerol is more expensive than glucose, its price is expected

19

to drop as bio-diesel production increases. It is also an attractive substrate for succinate
production because it has the same number of available electrons as succinate (both
glycerol and succinate yield 14 electrons when completely oxidized to CO2), thus no
additional source of reducing power would be needed to produce succinate from glycerol
and CO2. A. succiniciproducens can naturally convert glycerol to succinate with little
acetate production. Succinatezacetate ratios from glycerol were 26:1 compared to 4:1
from glucose (51). Glycerol and glucose could also be co-utilized resulting in improved
succinate productivity, but lower succinatezacetate ratios (51).

A. succiniciproducens is an attractive candidate for bio-based succinate
production because it naturally produces high succinate concentrations, and its
fermentation characteristics have been examined in several industrially relevant media.
However, there are several disadvantages to A. succiniciproducens-based succinate
production. There is only one mention of an improved succinate-producin g strain. Strain
FA- 10 was obtained by using ﬂuoroacetate as selective pressure and displayed
signiﬁcantly less acetate production (25). However, pyruvate was produced in place of
acetate (25). FA-lO produced succinate titers between 25 and 55 g l'1 (25). Developing
genetic tools for A. succiniciproducens to eliminate fermentation byproducts and allow
for co-utilization of different sugars would greatly beneﬁt its fermentation characteristics.
Furthermore, relatively little is known about its metabolism. The number of reactions
shown in Figure 1.3 merely reﬂects the amount of knowledge accumulated on A.
succiniciproducens metabolism and does not necessarily reﬂect the simplicity of its
metabolic pathways. A. succiniciproducens is also a strict anaerobe, so special procedures

are required to avoid exposure to oxygen. Another potential disadvantage is that A.

20

succiniciproducens can cause bacteremia in humans (85). Although infections are rare
and typically occur in immuno—compromised patients, the potential virulence of A.

succiniciproducens cannot be ignored.

1.3.4 Rumen Pasteurellaceae: Actinobacillus succinogenes and Mannheimia
succiniciproducens

Succinate is an important metabolic intermediate in the rumen. Several rumen
bacteria obtain energy by decarboxylating succinate to propionate, which then serves as a
nutrient for the ruminant (77, 83, 92). Many succinate-producing bacteria have been
isolated from the rumen. The two currently receiving the most attention for industrial
succinate production are A. succinogenes and M. succiniciproducens.

A. succinogenes and M. succiniciproducens were both isolated from the bovine
rumen, A. succinogenes from a Michigan State University cow and M.
succiniciproducens from a Korean cow. Both organisms belong to the Pasteurellaceae
and their genome sequences are more similar to each other than to any other genome
sequence (Chapter 5). Not surprisingly, the two bacteria share many metabolic traits.
Both consume a wide variety of carbon and energy sources, including the most abundant
plant sugars: glucose, fructose, xylose (28, 48, 91), and arabinose (28, 91) (arabinose
consumption has not yet been reported for M. succiniciproducens). As with A.
succiniciproducens, succinate production is enhanced in each strain by increased
availability of CO2 and H2 (30, 91). Providing electrically-reduced neutral red, as an
alternative to H2 improved A. succinogenes succinate yields and product ratios during

glucose fermentations (67). It was shown that neutral red, a redox dye, could serve as an

21

electron donor for fumarate reductase (67). Unlike with A. succiniciproducens, A.
succinogenes and M. succiniciproducens fermentation product proﬁles are remarkably
stable at different pH values (30, 91), and at least A. succinogenes can simultaneously
consume multiple sugars (26). Furthermore, A. succinogenes can tolerate glucose
concentrations as high as 150 g l'1 (27) whereas A. succiniciproducens does not grow
well above carbohydrate concentrations exceeding 100 g l'1 (21). A. succinogenes and M.
succiniciproducens also show metabolic differences. For example, A. succinogenes
consumes sorbitol whereas M. succiniciproducens does not, and M. succiniciproducens
produces lactate whereas A. succinogenes does not (48, 91).

An understanding of the A. succinogenes pathways responsible for succinate
production was obtained primarily from in vitro enzyme assays (91), whereas M.
succiniciproducens pathways were largely inferred from its genome sequence (30). In
both species, succinate production occurs in a manner similar to that in A.
succiniciproducens, with PEP being carboxylated to OAA by PEPCK, then OAA being
converted to succinate by the reductive TCA branch (Fig. 1.4). PEPC does not appear to
be important for converting PEP to OAA. Indeed, insertional inactivation of PEPC in M.
succiniciproducens did not affect product distributions, whereas the insertional
inactivation PEPCK severely impaired growth and production of all organic acids (53).
Unlike A. succiniciproducens, pyruvate-ferredoxin oxidoreductase is not responsible for
converting pyruvate to acetate and ethanol in either A. succinogenes or M.
succiniciproducens. Instead, pyruvate formate-lyase was implicated in this role (30, 91).
Despite often being depicted as a simple branched metabolism from PEP to succinate and

alternative products, other enzyme activities suggest that flux distribution to products

22

could occur at metabolite nodes other than PEP. Malic enzyme and OAA decarboxylase
activities were detected in A. succinogenes cell extracts (91). These enzymes were also
identiﬁed in genomic and proteomic analyses of M. succiniciproducens (30, 46).
Insertional inactivation of malic enzyme did not affect M. succiniciproducens product

distributions, but compensating alternative pathways cannot be ruled out (53).

Art:
Glc ---‘-'----> GGP
C
EMP ‘, /
2 "AD \ ::>ET-S< ”
’ Lac
PEPCK OszP Per m, ,‘quH
W. /H
................ Pyr 0:02
MD” co ’1‘ PFL For "':
Mal . , 2 x H
‘x NADPH ,/ 2
/Fm “~-____=f __________ 11" MEnz
Fum AcCoA
4.
con Frd 2 MD AD” ATP
AK
Suc EtOH Ace

Figure 1.4. Simpliﬁed map of wild-type A. succinogenes and M. succiniciproducens
succinate-producing metabolism. Dotted lines indicate enzyme activities with unknown
in vivo contributions. The grey LDH arrow indicates that lactate production is unique to
M. succiniciproducens. See Fig. l.lA and Fig. 1.2 for abbreviations.

Both bacterial species have been characterized in industrially relevant media. M.
succiniciproducens was grown in batch and continuous cultures with whey and corn steep
liquor together. Succinate yields and productivities were 71% at 1.2 g 1’I h'1 and 69% at
3.9 g l'l h'1 in batch and continuous cultures, respectively (47). In a medium containing
corn steep liquor, yeast extract, and glucose, A. succinogenes produced 80 g succinate 1"

(27), a remarkable titer for a wild-type strain. Many of the described succinate

‘ puriﬁcation procedures involve making succinate salts (see section 2). A. succinogenes

23

will grow and ferment in the presence of 96 g l'1 disodium succinate or 130 g 1'1
magnesium succinate, whereas A. succiniciproducens will not (27).

A. succinogenes and M. succiniciproducens naturally produce high concentrations
of succinate and have among the highest speciﬁc productivities reported (Table 1.1). A.
succinogenes has a speciﬁc succinate productivity of 430 mg g’I l’1 in batch
fermentations with 5 g l'1 yeast extract (McKinlay et al. unpublished data). Therefore,
these bacterial species are excellent starting points for engineering an industrial strain. Of
the two strains, A. succinogenes has the highest succinate yields reported. It is worth
noting, though, that M. succiniciproducens has not been grown under the same
conditions, and no side-by-side comparative fermentation with these two bacteria has
been reported. Only a few genetic engineering efforts have been reported for these two
species. Fluoroacetate resistant mutants of A. succinogenes were obtained that, like A.
succiniciproducens FA-lO, produce pyruvate (26). The ﬂuoroacetate resistance and
altered metabolism of these strains is thought to be due to an inactive pyruvate formate-
lyase (66). Despite producing pyruvate, the mutant strains generally had improved
succinate yields, with one strain, F253, reaching a ﬁnal titer of 106 g succinate/l in a
medium containing corn steep liquor, yeast extract, and glucose (26). This is the highest
succinate titer ever reported from a fermentation. Because of a lack of genetic tools, the
fluoroacetate-resistant strains are the only known A. succinogenes mutants. On the other
hand, several genes have been insertionally inactivated in M. succiniciproducens using a
sacB-based sucrose counter selection system (53). This knock-out system was used to
disrupt flux to lactate, formate, and acetate by insertionally inactivating lactate

dehydrogenase, pyruvate formate-lyase, phosphotransacetylase, and acetate kinase,

24

creating the mutant LPK7. Lactate, forrnate, and acetate productions were severely
limited, but pyruvate and malate were produced instead and batch succinate titers only
improved from 10.5 g l'1 to 13.4 g l'l as a result of the extensive engineering. However,
the mutants were only tested under electron-limiting conditions with glucose, yeast
extract, and peptone as electron sources. If all reductant is formed from glycolysis, there
is only enough NADH produced to form one succinate per glucose. Interestingly, LPK7
only achieved one succinate per glucose (53). LPK7, and the other mutants described in
the study, should be characterized in electron rich conditions (e.g., by adding H2). Under
electron limiting fed-batch conditions, a succinate titer of 52.4 g l’1 was achieved by M.
succiniciproducens LPK7 (Table 1.1).

Overall, both rumen bateria are important organisms to be studied and engineered
for industrial succinate production. However, several hurdles must be overcome. In
contrast to E. coli and C. glutamicum, little is known about the physiology and genetics
of these bacteria, and there is still much to be learned about their metabolism. Progress in
engineering A. succinogenes has been relatively slow, hindered by a lack of genetic tools
and system-wide analyses. Progress with M. succiniciproducens, on the other hand, has
been rapid with the genome sequence (30), a proteomic analysis (46), and gene knockout
technology (53) described within ﬁve years of the ﬁrst published characterization of the
organism (48). Despite the availability of genetic tools and system-wide approaches, the
engineering efforts with M. succiniciproducens have thus far focused on inherently
obvious targets, namely those activities leading to alternative endproducts (53). Only
recently were oxaloacetate decarboxylase and an acetate-forming aldehyde

dehydrogenase identiﬁed as potentially important targets after a M. succiniciproducens

25

proteomics study (46). These genes were either overlooked or ignored during the M.
succiniciproducens genome annotation, and the in vivo contribution of these enzymes to
flux distribution remains unknown. It is also important to note that both bacteria are
related to several well-known pathogens, including A. actinomycetemcomitans, A.
pleuropneumoniae, Haemophilus inﬂuenzae, M. haemolytica, and Pasteurella multocida.
While M. succiniciproducens is reported to have no leukotoxin or capsule production
genes, a detailed analysis of M. succiniciproducens or A. succinogenes for virulence traits
has not been performed. Non-pathogenicity should be established before these strains are

used on an industrial scale.

1.4. Backgound on l3C-metabolic ﬂux analysis

Successful metabolic engineering of microbes is usually reported as being based
on a logical rationale for the genetic manipulations made that result in the expected
metabolic effects. Usually not reported are the majority of engineering efforts that result
in unexpected or undesired phenotypes (20, 87). However, these unpublished results are
important to highlight the current limitations in accurately predicting the metabolic
effects of genetic perturbations.

Even with E. coli the metabolic effects of genetic manipulations can be
unexpected. For example, to improve succinate production, E. coli’s metabolism was
engineered to resemble that of the natural succinate-producer, M. succiniciproducens
(52). After several gene disruptions, the resulting mutants had few favorable
characteristics over the wild type strain. The authors then used a computer modeling

approach to predict the effects of different combinatorial gene knockouts. The

26

simulations correctly predicted the ineffectiveness of some of the mutations already
made. They also predicted that knocking out pykA, pku, and ptsG would beneﬁt
anaerobic succinate production. These genes were knocked out, resulting in a seven fold
increase in succinate titer (52).

Using computer simulations in an effort to guide E. coli metabolic engineering is
possible owing to a well-annotated genome and knowledge of active pathways. However,
such detailed information is not available for all industrially relevant organisms. For
example, the A. succinogenes draft genome sequence was only recently made available
(http://genomeoml.gov/microbial/asucl). Furthermore, the robust genetic tools available
for E. coli, make it practical to test the effects of multiple mutations in a relatively short
period of time. These mutations are important not only for engineering production strains,
but also for understanding metabolism and physiology. Other industrially relevant
organisms, including A. succinogenes, are not so easily engineered. Thus, when genetic
engineering is time-consuming and laborious, it is especially important to have other
means by which to study metabolism, and to be able to accurately predict the metabolic
effects of genetic manipulations.

Part of the current limitation in predicting the metabolic effects of genetic
manipulations is due to past inabilities to assess the response of intermediary fluxes to
genetic and environmental perturbations. For example, under certain conditions, both E.
coli pyruvate kinase isoenzymes can be insertionally inactivated with no observable
effect on growth parameters (74). In other conditions, inactivating one or both enzymes
seemingly results only in a decreased growth rate (74). These results were surprising

because pyruvate kinase is a central metabolic enzyme as the last step in glycolysis. By

27

using PTS deﬁcient mutants it was inferred that the FT S could supply adequate pyruvate
to support growth in the absence of pyruvate kinase activity (74). However, other
metabolic changes could not be observed from this approach. Experiments with [1-
'4C] glucose, showed that pyruvate kinase knockout mutants had increased oxidative
pentose phosphate pathway (OPPP) ﬂux (75). However, it was not until '3 C-labelin g
studies were performed that it was realized that PEP carboxylase and malic enzyme were
compensating for the pyruvate kinase deﬁciency (1, 19) and that the OPPP and TCA
cycle fluxes varied with growth conditions (19). These discoveries were made without
relying on genomic analyses or further genetic manipulations. This example demonstrates
the usefulness of l3C-labeling experiments for revealing the global metabolic effects of
perturbations, which could not otherwise be observed by measuring extracellular ﬂuxes.
With knowledge of the enzymes involved in the metabolism under study, and
with global analyses of how metabolism responds to genetic or environmental
perturbations, one should be able to better predict the metabolic effects of genetic
manipulations. As illustrated above, 13C-labeling studies are an effective method for
obtaining global metabolic data. These techniques are often referred to as l3C—metabolic
ﬂux analysis (MFA), because, in addition to identifying which intermediary metabolic
pathways are active, ﬂuxes through these pathways can be quantiﬁed. MFA evolved from
a mass balance approach, or ﬂux balance analysis (FBA). FBA relies on a metabolic
steady state, where the ﬂuxes into a metabolite pool equal the ﬂuxes out of that pool.
Intermediary metabolic ﬂuxes are estimated within the constraints of a stoichiometric
metabolic network model using measurements of extracellular ﬂuxes (i.e., substrate

consumption and product formation rates, where products include biomass components).

28

Because FBA only uses consumption and production rate measurements, it has several
limitations (84, 96). Fluxes through alternative pathways to a common product (without
the production of a unique byproduct from one of those pathways) cannot be
distinguished. Fluxes through cyclic pathways and bi-directional ﬂuxes also cannot be
determined. FBA analyses must therefore rely on assumptions about metabolism to
estimate intermediary ﬂuxes. For example, cofactor balances or demands, such as those
for NADH and NADPH, can be used to estimate ﬂuxes through various pathways, but the
presence of enzymes like transhydrogenase make these assumptions ineffective.

Using l3C-labeled substrates in combination with FBA overcomes these
limitations. In addition to requiring metabolic steady state, MFA also requires (at least
until recently [63, 64]) an isotopic steady state, where the proportion of labeled atoms in
subtrate and metabolites is constant with time. The equations used in MFA to determine
ﬂuxes from labeling patterns describe metabolic and isotopic steady states so it is
important that these conditions are met (97). Metabolic ﬂux information is obtained from
MFA when a labeled substrate is processed by different pathways, resulting in different
labeling patterns, or isotopomers, in metabolic intermediates (96). These labeling patterns
are imprinted on the metabolic products (e. g., proteinaceous amino acids) that are made
from these metabolic intermediates and that can be measured by gas chromatography-
mass spectrometry (GC-MS) and nuclear magnetic resonance spectroscopy (NMR). The
ratios of the different labeling patterns can then be compared to determine the relative '
ﬂuxes through the pathways that produced them. An example of this approach is
illustrated in Figure 1.5, where forward glycolytic ﬂuxes produce singly labeled 3-

phosphoglycerate from [1-13C] glucose, whereas OPPP ﬂux produces unlabeled 3-

29

phosphoglycerate. The labeling pattern is imprinted on serine, which is then incorporated
into protein. After protein hydrolysis, serine is derivatized and analyzed by GC-MS.
After correcting for natural abundances of mass isotopes the relative ﬂuxes through each
pathway can be determined from the ratio of labeled to unlabeled serine.

Complex labeling patterns and reversible ﬂuxes through multiple pathways can
make manual interpretation of labeling patterns time-consuming and difﬁcult. As a result,
sophisticated software suites, such as I3C—Flux (98), have been developed to estimate
ﬂuxes from isotopomer data (76). The process by which 13C-Flux, and similar software,
obtain values for metabolic ﬂuxes from an experimental data set is illustrated in Figure
1.6. I3C—Flux uses a metabolic network, substrate isotopomers conditions, and an
arbitrary set of ﬂuxes, to simulate an isotopomer data set. This simulated data set is
compared to the measured data set, and the similarity between the two is scored by a sum

of squared residuals (SSms), weighted by the standard deviation of each measurement:

measured value - simulated value 2
S Sres = X ( measured standard deviation )

The arbitrary ﬂuxes are then adjusted in an iterative process until a set of ﬂuxes is found
that results in a simulated isotopomer data set that closely resembles the measured data
set; an optimized ﬁt. The quality of the ﬁt can be statistically assessed by a )8 test.

Statistical assessment of individual ﬂuxes can also be performed (98).

3O

[1 -13C]G6P OPP co2
IEIEIIZJEIEI

 

 

 

 

l EMP asp
[BENEHZIE]
EMP converts GaP
Egiliﬂittts. ‘1"?- Ga" °ﬁ§P°°m
1- C Glc to
sinlglg' i1": 63:3 SEE. 390. ENE-IE] Lnlabelled G3P
un a e e \ EE. /

 

 

EJIZIEI

. “J “'4‘" Labeling pattern is
4:” 59f
EH3]. E, imprinted on Ser
which can be obtained
GENE] by protein hydrolysis

 

 

 

 

 

 

 

 

 

 

After derivatization
60"” with TBDMS, Ser
mass isotopomers are
observed by GC-MS
7000 l TBDMScSerine
m ..
Fragmentation 5000

 

 

 

 

 

. . 2000 .
Information 1000 [g

prmi/tides 'some :> g 4000-
pos none 4
labeling § 3000
§§ §

§

 

Natural isotope abundances are corrected and positional isotopic enrichments obtained:
9.9., 45% of Ser is [3-‘3C]Ser
Stoichiometry of network is used to calculate relative fluxes:
EMP, 1 vs. OPPP, 0.14

 

 

 

Figure 1.5. Determining relative EMP and OPPP ﬂuxes from GC-MS measurements
of serine. 3PG, 3-phosphoglycerate; G3P, glyceraldehyde-3-phosphate; G6P, glucose-6-
phosphate; R5P, ribulose-S-phosphate TBDMS, tert-butyldimethylsilsyl.

31

MFA has been used to guide the engineering of industrially important organisms,
such as lysine-producing C. glutamicum, as reviewed by Koffas and Stephanopoulos
(41). Lysine is synthesized from OAA and pyruvate. FBA studies indicated that
anaplerotic ﬂux to OAA was limiting lysine production (90). Illustrating the limitationsof
FBA, ﬂuxes from PEP and pyruvate to OAA could not be distinguished, and OAA was
thought to originate entirely from PEP carboxylation (90). Later, ”C-labeling studies
indicated that a pyruvate carboxylating activity was the principal C3-carboxylating ﬂux
(68, 72). The PYC responsible for this ﬂux was cloned, and conﬁrmed to play a role in
metabolism through genetic manipulations (42, 43, 70, 71). Overexpressing PYC favored
either growth or lysine production depending on the aspartate concentration in the cell.
This trend suggested that aspartate kinase ﬂux was also important for balancing the lysine
and biosynthesis demands on OAA (42). A metabolic control analysis study also
indicated that aspartate kinase activity had a large inﬂuence on lysine production (101).
By overexpressing both PYC and aspartate kinase, lysine speciﬁc productivity was
increased by 44% on glucose, without growth defects. The increased PYC ﬂux likely
compensated for the drain of OAA from the TCA cycle to aspartate kinase. Thus, there
was enough OAA supply the lysine synthesis pathway and the TCA cycle, which
provides intermediates for biosynthesis (44).

Improving riboﬂavin-production by Bacillus subtilis is another example of
metabolic engineering being guided by knowledge obtained from MFA. MFA performed
under different growth conditions showed that different OPPP ﬂuxes did not affect the
riboﬂavin yield, indicating that neither precursor supply nor NADPH limited riboﬂavin

production (16, 17, 81). MFA also identiﬁed a signiﬁcant PEPCK ﬂux from OAA to

32

 

 

 

 

 
 
 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

i 02
can —4 n59
FGP H E4P
G3P
3PG
co
2 PEP'
w 1' x .
[OAA Pyr ‘1
Fum For AcCoA
I / \
Suc EtOH Ace
13C-Excreted 1i’C-Proteinaceous
organic acids amino acids
NMR .GC-MS
Measured
isotopomers ' "
Substrate ssms
isotopomer \
or
Simulated
I -=-)
F ux model isotopomers
Ammaw j Optimized
fluxes R fit
Adjust arbitrary fluxes

Figure 1.6. Estimation of intermediary metabolic ﬂuxes from product isotopomers
using 13C-FLUX. E4P, erythrose-4-phosphate; F6P, fructose-6—phosphate; and SSm,
weighted sum of squared residuals. See Figure 1.1A and Figure 1.5 for other
abbreviations. Dark gray ovals signify intermediates used for amino acid synthesis. Light
. gray ovals signify excreted organic acids.

33

PEP, completing an ATP-dissipating futile cycle from PEP -) pyruvate -) OAA -) PEP
(15-17, 81). Flux analyses carried out with carbon sources known to affect the ATP/ADP
balance in the cell indicated that excess ATP was mostly being consumed in cell
maintenance and energy-dissipating reactions (16). Those substrate mixtures resulting in
the highest ATP/ADP ratios also resulted in the highest riboﬂavin yields (16). Guided by
this understanding of the riboﬂavin-producing metabolism, the engineering focused on
respiration efﬁciency and maintenance metabolism. By genetically disrupting the electron
transport chain bd oxidase, which only pumps one H+ per electron transfer, the bacterium
had to rely on the more efﬁcient aa3 oxidase for respiration, which pumps two H+ per
electron transfer (103). The more efﬁcient respiration decreased the energetic
maintenance coefﬁcient by ~40%, and subsequently improved riboﬂavin yields by ~30%
(103). These examples with C. glutamicum and B. subtilis demonstrate the insights into
microbial metabolism that can be gained from MFA, and how the information obtained
can be used to guide successful metabolic engineering strategies. These examples also
illustrate how carbon ﬂuxes, determined in the absence of cofactor balance constraints,

can provide an understanding of potentially important cofactor ﬂuxes.

5. Objectives

A. succinogenes is a promising candidate for industrial succinate production,
having the highest succinate titer, the highest substrate and product tolerance, and one of
the highest speciﬁc succinate productivities (430 mg g'1 l"; McKinlay et al. unpublished
data). However, relatively little is known about the organism, including its metabolism.
This dissertation gives insight into A. succinogenes metabolism by accomplishing three

objectives: (i), determine active metabolic pathways in A. succinogenes; (ii), determine

34

the ﬂuxes through these pathways; and (iii), determine how ﬂuxes respond to
environmental perturbations. The resulting information is expected to be essential for
guiding the metabolic engineering of A. succinogenes. l3C-MFA was an attractive
approach for studying A. succinogenes metabolism, because A. succinogenes is currently
difﬁcult to genetically manipulate, and until recently, no genome sequence was available.

l3C-MFA analyzes the labeling patterns in metabolic products. A chemically-
deﬁned medium was required to ensure that metabolic products (e. g., amino acids) were
synthesized from l3C—glucose rather than being obtained from unlabeled carbon sources
(e. g., yeast extract). Chapter 2 describes the development of a chemically deﬁned A.
succinogenes growth medium, containing only the amino acids required for growth. This
medium was also used without 13 C to address the existence of A. succinogenes TCA-
associated ﬂuxes. Chapter 3 describes the use of [1-‘3C]glucose in the deﬁned medium to
establish the ﬁrst A. succinogenes metabolic ﬂux model. The A. succinogenes metabolic
map was reﬁned, and an understanding of the pathways responsible for generating
anabolic reducing power and for distributing ﬂux to succinate and alternative products
was obtained. Chapter 4 describes MFA experiments performed under different NaHCO3
and H2 conditions with a more informative substrate isotopomer than was used in Chapter
3. Exchange ﬂuxes were identiﬁed that could not be observed from using [1-‘3C] glucose
alone. A new understanding of how NaHC03 and H2 perturbations affect A. succinogenes
intermediary metabolic ﬂuxes was obtained. In all cases, the MFA studies were
complemented by A. succinogenes draft genome sequence information. An analysis of
the industrially relevant features of the genome is described in Chapter 5. Chapter 6

summarizes the insights made into A. succinogenes metabolism. The results are used to

35

discuss possible future research for understanding and engineering A. succinogenes
metabolism. Chapter 6 also emphasizes the importance of A. succinogenes gene knock-

out and knock-in technologies and discusses possibilities for developing these tools.

36

1.6 References.

l.

10.

11.

12.

13.

Al Zaid Siddiquee, M. J. A.-B. K., and K. Shimizu. 2004. Metabolic ﬂux analysis
of [1ku gene knockout Escherichia coli based on l3C-labeling experiments together

with measurements of enzyme activities and intracellular metabolite concentrations.
Appl. Microbiol. Biotechnol. 63:407-417.

Anonymous. 2005. 1,4-Butanediol, p. 42, Chemical Market Reporter, vol. 268.

Anonymous. 2002. BASF increases prices for 1,4-butanediol and derivatives, p. 8,
Chemical Market Reporter, vol. 262.

Anonymous. 2005. Lanxess ups maleic anhydride capacity, p. 4, Chemical Market
Reporter, vol. 267.

Anonymous. 2005. Prices & People, p. 16, Chemical Market Reporter, vol. 267.

Berglund, K. A., P. Elankovan, and D. A. Glassner. 1991. Carboxylic acid
puriﬁcation and crystallization process. US. patent 5,034,105.

Berglund, K. A., S. Yedur, and D. D. Dunuwila. 1999. Succinic acid production and
puriﬁcation. US. patent 5,958,744.

Brown, R. 2005. Housing sector boosts maleic anhydride, p. 24, Chemical Market
Reporter, vol. 267.

Chang, J. 2006. Huntsman picks La. site, p. 4, Chemical Market Reporter, vol. 269.

Chartteijee, R., C. S. Millard, K. Champion, D. P. Clark, and M. I. Donnelly. 2001.
Mutation of the ptsG gene results in increased production of succinate in
fermentation of glucose by Escherichia coli. Appl. Environ. Microbiol. 67: 148-154.

Cotelesage, J. J ., L. Prasad, J. G. Zeikus, M. Laivenieks, and L. T. Delbaere. 2005.
Crystal structure of Anaerobiospirillum succiniciproducens PEP carboxykinase
reveals an important active site loop. Int. J. Biochem. Cell. Biol. 37:1829-1837.

Cox, S. J ., S. Shale] Levanon, A. M. Sanchez, H. Lin, B. Peercy, G. N. Bennett, and
K. Y. San. 2006. Development of a metabolic network design and optimization
framework incorporating implementation constraints: a succinate production case
study. Metab. Eng. 8.

Dake, S. B., R. V. Gholap, and R. V. Chaudhari. 1987. Carbonylation of 1,4-

butanediol diacetate using rhodium complex catalyst: a kinetic study. Ind. Eng.
Chem. Res. 26: 1513-1518.

37

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

Datta, R., D. A. Glassner, M. K. Jain, and J. R. Vick Roy. 1992. Fermentation and
puriﬁcation process for succinic acid. US. patent 5,168,055.

Dauner, M., J. E. Bailey, and U. Sauer. 2001. Metabolic ﬂux analysis with a

comprehensive isotopomer model in Bacillus subtilis. Biotechnol. Bioeng. 76: 144-
156.

Dauner, M., M. Sonderegger, M. Hochuli, T. Szyperski, K. Wuthrich, H. P.
Hohmann, U. Sauer, and J. E. Bailey. 2002. Intracellular carbon ﬂuxes in
riboﬂavin-producing Bacillus subtilis during growth on two-carbon substrate
mixtures. Appl. Environ. Microbiol. 68: 1760-1771.

Dauner, M., T. Stomi, and U. Sauer. 2001. Bacillus subtilis metabolism and

energetics in carbon-limited and excess-carbon chemostat culture. J. Bacteriol.
183:7308-7317.

Deng, Y., Z. Ma, K. Wang, and J. Chen. 1999. Clean synthesis of adipic acid by
direct oxidation of cyclohexene with H202 over peroxytungstate-organic complex
catalysts. Green Chem. 275:275-276.

Emmerling, M., M. Dauner, A. Ponti, J. Fiaux, M. Hochuli, T. Szyperski, K.
Wuthrich, J .E. Bailey, and U. Sauer. 2002. Metabolic ﬂux responses to pyruvate
kinase knockout in Escherichia coli. J. Bacteriol. 184: 152-164.

Fell, D. 1997. Understanding the control of metabolism. Portland Press, London
and Miami.

Glassner, D. A., and R. Datta. 1992. Process for the production and puriﬁcation of
succinic acid. US. patent 5,143,834.

Gokarn, R. R., M. A. Eiteman, and E. Altman. 2000. Metabolic analysis of
Escherichia coli in the presence and absence of the carboxylating enzymes

phosphoenolpyruvate carboxylase and pyruvate carboxylase. Appl. Environ.
Microbiol. 66: 1844-1850.

Gokarn, R. R., J. D. Evans, J. R. Walker, S. A. Martin, M. A. Eiteman, and E.
Altman. 2001. The physiological effects and metabolic alterations caused by the

expression of Rhizobium etli pyruvate carboxylase in Escherichia coli. Appl.
Microbiol. Biotechnol. 56: 188-195.

Gross, R. A., and B. Kalra. 2002. Biodegradable polymers for the environment.
Science 297:803-807.

Guettler, M. V., and M. K. Jain. 1996. Method for making succinic acid,

Anaerobiospirillum succiniciproducens varients for use in process and methods for
obtaining varients. US. patent 5,521,075.

38

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

Guettler, M. V., M. K. Jain, and D. Rumler. 1996. Method for making succinic acid,
bacterial variants for use in the process, and methods for obtaining variants. US
Patent 5,573,931 patent 5,573,931.

Guettler, M. V., M. K. Jain, and B. K. Soni. 1996. Process for making succinic acid,
microorganisms for use in the process and methods of obtaining the
microorganisms. US Patent 5,504,004 patent 5,504,004.

Guettler, M. V., D. Rumler, and M. K. Jain. 1999. Actinobacillus succinogenes sp.
nov., a novel succinic-acid-producing strain from the bovine rumen. Int. J. Syst.
Bacteriol. 49:207-216.

Hong, S. H., and S.Y. Lee. 2001. Metabolic ﬂux analysis for succinic acid
production by recombinant Escherichia coli with ampliﬁed malic enzyme activity.
Biotechnol. Bioeng. 74:89-95.

Hong, S. H., J. S. Kim, S. Y. Lee, Y. H. In, S. S. Choi, J. K. Rih, C. H. Kim, H.
Jeong, C. G. Hur, and J. J. Kim. 2004. The genome sequence of the capnophilic
rumen bacterium Mannheimia succiniciproducens. Nat. Biotechnol. 22: 1275-1281.

Huh, Y. S., Y. K. Hong, W. H. Hong, and H. N. Chang. 2004. Selective extraction
of acetic acid from the fermentation broth produced by Mannheimia
succiniciproducens. Biotechnol. Lett. 26: 1581-1 584.

Huh, Y. S., J. Y-S., Y. K. Hong, H. Song, S. Y. Lee, and W. H. Hong. 2006.
Effective puriﬁcation of succinic acid from fermentation broth produced by
Mannheimia succiniciproducens. Process Biochem. 41 : 1461-1465.

Inui, M., S. Murakami, S. Okino, H. Kawaguchi, A. A. Vertes, and H. Yukawa.
2004. Metabolic analysis of Corynebacterium glutamicum during lactate and
succinate productions under oxygen deprivation conditions. J. Mol. Microbiol.
Biotechnol. 7.

J abalquinto, A. M., F. D. Gonzalez-Nilo, M. Laivenieks, M. Cabezas, J. G. Zeikus,
and E. Cardemil. 2004. Anaerobiospirillum succiniciproducens

phosphoenolpyruvate carboxykinase. Mutagenesis at metal site 1. Biochimie 86:47-
51.

Jabalquinto, A. M., M. Laivenieks, M. Cabezas, J. G. Zeikus, and E. Cardemil.
2002. The effect of active site mutations in the oxaloacetate decarboxylase and
pyruvate kinase-like activities of Anaerobiospirillum succiniciproducens
phosphoenolpyruvate carboxykinase. J. Protein. Chem. 21:443-445.

Jabalquinto, A. M., M. Laivenieks, F. D. Gonzalez-Nilo, M. V. Encinas, G. Zeikus,
and E. Cardemil. 2003. Anaerobiospirillum succiniciproducens

39

37.

38.

39.

40.

41.

42.

43.

45.

46.

phosphoenolpyruvate carboxykinase: mutagenesis at metal site 2. J. Protein. Chem.
22:5 15-5 19.

Jabalquinto, A. M., M. Laivenieks, F. D. Gonzalez-Nilo, A. Yevenes, M. V.
Encinas, J. G. Zeikus, and E. Cardemil. 2002. Evaluation by site-directed
mutagenesis of active site amino acid residues of Anaerobiospirillum

succiniciproducens phosphoenolpyruvate carboxykinase. J. Protein. Chem. 21 :393-
400.

Jabalquinto, A. M., M. Laivenieks, J. G. Zeikus, and E. Cardemil. 1999.
Characterization of the oxaloacetate decarboxylase and pyruvate kinase-like
activities of Saccharomyces cerevisiae and Anaerobiospirillum succiniciproducens
phosphoenolpyruvate carboxykinases. J. Protein. Chem. 18:659-664.

Kawaguchi, H., A. A. Vertes, S. Okino, M. Inui, and H. Yukawa. 2006.
Engineering of a xylose metabolic pathway in Corynebacterium glutamicum. Appl.
Environ. Microbiol. 72:3418-3428.

Kim, P., M. Laivenieks, C. Vieille, and LG. Zeikus. 2004. Effect of overexpression
of Actinobacillus succinogenes phosphoenolpyruvate carboxykinase on succinate
production in Escherichia coli. Appl. Environ. Microbiol. 70: 1238-1241.

Koffas, M., and G. Stephanopoulos. 2005. Strain improvement by metabolic
engineering: lysine production as a case study for systems biology. Curr. Opin.
Biotechnol. 16:361-366.

Koffas, M. A., G. Y. Jung, J. C. Aon, and G. Stephanopoulos. 2002. Effect of
pyruvate carboxylase overexpression on the physiology of Corynebacterium
glutamicum. Appl. Environ. Microbiol. 68:5422-5428.

Koffas, M. A., R. Ramamoorthi, W. A. Pine, A. J. Sinskey, and G. Stephanopoulos.
1998. Sequence of the Corynebacterium glutamicum pyruvate carboxylase gene.
Appl. Microbiol. Biotechnol. 50:346-352.

Koffas, M. A. G., G. Y. Jung, and G. Stephanopoulos. 2003. Engineering
metabolism and product formation in Corynebacterium glutamicum by coordinated
gene overexpression. Metab. Eng. 5:32-41.

Kwon, Y. D., S. Y. Lee, and P. Kim. 2006. Inﬂuence of gluconeogenic
phosphoenolpyruvate carboxykinase (PCK) expression on succinic acid

fermentation in Escherichia coli under high bicarbonate condition. J ‘. Microbiol.
Biotechnol. 16: 1448-1452.

Lee, J. W., S. Y. Lee, H. Song, and J. S. Yoo. 2006. The proteome of Mannheimia
succiniciproducens, a capnophilic rumen bacterium. Proteomics 6:3550-3566.

40

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

Lee, P. C., S. Y. Lee, S. H. Hong, and H. N. Chang. 2003. Batch and continuous
cultures of Mannheimia succiniciproducens MBEL55E for the production of
succinic acid from whey. Bioprocess. Biosyst. Eng. 26:63-67.

Lee, P. C., S. Y. Lee, S. H. Hong, and H. N. Chang. 2002. Isolation and
characterization of a new succinic acid-producing bacterium, Mannheimia

succiniciproducens MBELSSE, from bovine rumen. Appl. Microbiol. Biotechnol.
58:663-668.

Lee, P. C., S. Y. Lee, S. H. Hong, H. N. Chang, and S. C. Park. 2003. Biological
conversion of wood hydrolysate to succinic acid by Anaerobiospirillum
succiniciproducens. Biotechnol. Lett. 25:11 1-1 14.

Lee, P. C., W. G. Lee, S. Kwon, S. Y. Lee, and H. N. Chang. 2000. Batch and
continuous cultivation of Anaerobiospirillum succiniciproducens for the production
of succinic acid from whey. Appl. Microbiol. Biotechnol. 54:23-27.

Lee, P. C., W. G. Lee, S. Y. Lee, and H. N. Chang. 2000. Succinic acid production
with reduced by-product formation in the fermentation of Anaerobiospirillum

succiniciproducens using glycerol as a carbon source. Biotechnol. Bioen g. 72:41-
48.

Lee, S. J ., D.-Y. Lee, T. Y. Kim, B. H. Kim, J. Lee, and S. Y. Lee. 2005. Metabolic
engineering of Escherichia coli for enhanced production of succinic acid, based on

genome comparison and in silico gene knockout simulation. Appl. Environ.
Microbiol. 71 :7880-7887.

Lee, S. J ., H. Song, and S. Y. Lee. 2006. Genome-based metabolic engineering of

Mannheimia succiniciproducens for succinic acid production. Appl. Environ.
Microbiol. 72: 1939-1948.

Lin, H., G. N. Bennett, and K. Y. San. 2005. Fed-batch culture of a metabolically
engineered Escherichia coli strain designed for hi gh-level succinate production and
yield under aerobic conditions. Biotechnol. Bioeng. 90:775-779.

Lin, H., G. N. Bennett, and K. Y. San. 2005. Metabolic engineering of aerobic
succinate production systems in Escherichia coli to improve process productivity
and achieve the maximum theoretical succinate yield. Metab. Eng. 7:116-127.

Lin, H., K. Y. San, and G. N. Bennett. 2005. Effect of Sorghum vulgare
phosphoenolpyruvate carboxylase and Lactococcus lactis pyruvate carboxylase

coexpression on succinate production in mutant strains of Escherichia coli. Appl.
Microbiol. Biotechnol. 67:515-523.

Lin, H., R. V. Vadali, G. N. Bennett, and K. Y. San. 2004. Increasing the acetyl-
CoA pool in the presence of overexpressed phosphoenolpyruvate carboxylase or

41

58.

59.

60.

61.

62.

63.

65.

66.

67.

68.

69.

pyruvate carboxylase enhances succinate production in Escherichia coli.
Biotechnol. Prog. 20: 1599-1604.

Lynd, L. R., C. Wyman, M. Laser, D. Johnson, and R. Landucci. 2002. Strategic
bioreﬁnery analysis: analysis of bioreﬁneries. National Renewable Energy
Laboratory.

Macy, J. M., L. G. Ljungdahl, and G. Gottschalk. 1978. Pathway of succinate and
propionate fromation in Bacteroidesfragilis. J. Bacteriol. 134:84-91.

McKinlay, J. B., Y. Shachar-Hill, J. G. Zeikus, and C. Vieille. 2006. Determining
Actinobacillus succinogenes metabolic pathways and ﬂuxes by NMR and GC-MS
analyses of l3C-labeled metabolic product isotopomers. Metab. Eng. In press.

Millard, C. S., Y.-P. Chao, J. C. Liao, and M. I. Donnelly. 1996. Enhanced
production of succinic acid by overexpression of phosphoenolpyruvate carboxylase
in Escherichia coli. Appl. Environ. Microbiol. 62: 1808-1810.

Nghiem, N., M. Donnelly, C.S. Millard, and L. Stols. 1999. Method for the
production of dicarboxylic acids. US. patent 5,869,301.

Noh, K., A. Wahl, and W. Wiechert. 2006. Computational tools for isotopically
instationary l3C labeling experiments under metabolic steady state conditions.
Metab. Eng. 8:554-577.

Noh, K., and W. Wiechert. 2006. Experimental design principles for isotopically
instationary '3 C labeling experiments. Biotechnol. Bioeng. 94:234-251.

Okino, S., M. Inui, and H. Yukawa. 2005. Production of organic acids by
Corynebacterium glutamicum under oxygen deprivation. Appl. Microbiol.
Biotechnol. 68:475-480.

Park, D. H., M. Laivenieks, M. V. Guettler, M. K. Jain, and J. G. Zeikus. 1999.
Microbial utilization of electrically reduced neutral red as the sole electron donor
for growth and metabolite production. Appl. Environ. Microbiol. 65:2912-2917.

Park, D. H., and J. G. Zeikus. 1999. Utilization of electrically reduced neutral red
by Actinobacillus succinogenes: physiological function of neutral red in membrane-
driven fumarate reduction and energy conservation. J. Bacteriol. 181:2403-2410.

Park, S. M., C. Shaw-Reid, A. J. Sinskey, and G. Stephanopoulos. 1997.
Elucidation of anaplerotic pathways in Corynebacterium glutamicum via l3C-NMR
spectroscopy and GC-MS. Appl. Microbiol. Biotechnol. 47:430-440.

Paster, M., J. L. Pellegrino, and T. M. Carole. 2003. Industrial bioproducts: today
and tomorrow. Energetics, Incorporated.

42

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

Peters-Wendisch, P. G., C. Kreutzer, J. Kalinowski, M. Patek, H. Sahm, and B. J.
Eikmanns. 1998. Pyruvate carboxylase from Corynebacterium glutamicum:

characterization, expression and inactivation of the pyc gene. Microbiology 144
915-927.

Peters-Wendisch, P. G., B. Schiel, V. F. Wendisch, E. Katsoulidis, B. Mockel, H.
Sahm, and B. J. Eikmanns. 2001. Pyruvate carboxylase is a major bottleneck for
glutamate and lysine production by Corynebacterium glutamicum. J. Mol.
Microbiol. Biotechnol. 3:295-300.

Petersen, S., A. A. de Graaf, L. Eggeling, M. Mollney, W. Wiechert. and H. Sahm.
2000. In vivo quantiﬁcation of parallel and bidirectional ﬂuxes in the anaplerosis of
Corynebacterium glutamicum. J. Biol. Chem. 275 :35932-35941.

Podkovyrov, S. M., and J. G. Zeikus. 1993. Puriﬁcation and characterization of
phosphoenolpyruvate carboxykinase,a catabolic COz-ﬁxing enzyme, from
Anaerobiospirillum succiniciproducens. J. Gen. Microbiol. 139:223-228.

Ponce, E., N. Flores, A. Martinez, F. Valle, and F. Bolivar. 1995. Cloning of the
two pyruvate kinase isoenzyme structural genes from Escherichia coli: the relative
roles of these enzymes in pyruvate biosynthesis. J. Bacteriol. 177:5719-5722.

Ponce, E., A. Martinez, F. Bolivar, and F. Valle. 1997. Stimulation of glucose
catabolism throught the pentose pathway by the absence of the two pyruvate kinase
isoenzymes in Escherichia coli. Biotechnol. Bioeng. 58:292-295.

Ratcliffe, R. G., and Y. Shachar-Hill. 2006. Measuring multiple ﬂuxes through
plant metabolic networks. Plant J. 45:490-51 1.

Samuelov, N. S., R. Datta, M. K. Jain, and J. G. Zeikus. 1999. Whey fermentation
by Anaerobiospirillum succiniciproducens for production of a succinate-based
animal feed additive. Appl. Environ. Microbiol. 65:2260-2263.

Samuelov, N. S., R. Lamed, S. Lowe, and J. G. Zeikus. 1991. Inﬂuence of C02-
HCO3 levels and pH on growth, succinate production, and enzyme activities of
Anaerobiospirillum succiniciproducens. Appl. Environ. Microbiol. 57:3013-3019.

Sanchez, A. M., G. N. Bennett, and K. Y. San. 2006. Batch culture characterization

and metabolic ﬂux analysis of succinate-producing Escherichia coli strains. Metab.
Eng. 82209-226.

Sanchez, A. M., G. N. Bennett, and K. Y. San. 2005. Novel pathway engineering

design of the anaerobic central metabolic pathway in Escherichia coli to increase
succinate yield and productivity. Metab. Eng. 7:229-239.

43

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

91.

92.

Sauer, U., V. Hatzimanikatis, J. E. Bailey, M. Hochuli, T. Szyperski, and K.
Wuthrich. 1997. Metabolic ﬂuxes in riboﬂavin-producing Bacillus subtilis. Nat.
Biotechnol. 15:448-452.

Scardovi, V., and M. G. Chiappini. 1966. Studies in rumen bacteriology. V.
Carboxylation of phosphoenolpyruvate in some rumen bacterial strains and in the
cell-free extract of the total rumen ﬂora. Ann. Microbiol. Enzymol. 16:119-127.

Scheiﬁnger, C. C., and M. J. Wolin. 1973. Propionate formation from cellulose and
soluble sugars by combined cultures of Bacteroides succinogenes and Selenomonas
ruminantium. Appl. Environ. Microbiol. 26:789-795.

Schmidt, K., A. Marx, A.A. de Graaf, W. Wiechert, H. Sahm, J. Nielsen, and J.
Villadsen. 1998. 13C tracer experiments and metabolite balancing for metabolic ﬂux
analysis: comparing two approaches. Biotechnol. Bioeng. 58:254-257.

Secchi, C., V. V. Cantarelli, S. Pereira Fde, H. H. Wolf, T. C. Brodt, M. C. Amaro,
and E. Inamine. 2005. Fatal bacteremia due to Anaerobiospirillum
succiniciproducens: ﬁrst description in Brazil. Braz. J. Infect. Dis. 9: 169-172.

Song, H., and S. Y. Lee. 2006. Production of succinic acid by bacterial
fermentation. Enzyme Microbial Techno]. 39:353-361.

Stephanopoulos, G., A. A. Aristidou, and J. Nielsen. 1998. Metabolic Engineering:
Principles and Methodologies. Academic Press, London.

Stols, L., and M. I. Donnelly. 1997. Production of succinic acid through
overexpression of NAD+-dependent malic enzyme in an Escherichia coli mutant.
Appl. Environ. Microbiol. 63:2695-2701.

Tilton, H. 2003. Prices & People, p. 20, Chemical Market Reporter, vol. 264.

Vallino, J. J ., and G. Stephanopoulos. 1993. Metabolic ﬂux distributions in

Corynebacterium glutamicum during growth and lysine overproduction. Biotechnol.
Bioeng. 41:633-646.

van der Werf, M. J ., M. V. Guettler, M. K. Jain, and J. G. Zeikus. 1997.
Environmental and physiological factors affecting the succinate product ratio during
carbohydrate fermentation by Actinobacillus sp. 130Z. Arch. Microbiol. 1672332-
342.

van Gylswyk, N. O. 1995. Succiniclasticum ruminis gen. nov., sp. nov., a ruminal
bacterium convering succinate to propionate as the sole energy-yieldin g
mechanism. Int. J. Syst. Bacteriol. 45:297-300.

44

93.

94.

95.

96.

97.

98.

99.

100.

101.

102.

103.

104.

Vemuri, G. N., M. A. Eiteman, and E. Altman. 2002. Succinate production in dual-
phase Escherichia coli fermentations depends on the time of transition from aerobic
to anaerobic conditions. J. Ind. Microbiol. Biotechnol. 28:325-332.

Vemuri, G. N., M.A. Eiteman, and E. Altman. 2002. Effects of growth mode and
pyruvate carboxylase on succinic acid production by metabolically engineered
strains of Escherichia coli. Appl. Environ. Microbiol. 68: 1715-1727.

Wendisch, V. F., M. Bott, J. Kalinowski, M. Oldiges, and W. Wiechert. 2006.
Emerging Corynebacterium glutamicum systems biology. J. Biotechnol. 124:74-92.

Wiechert, w. 2001. 13C Metabolic ﬂux analysis. Metab. Eng. 3: 195-206.

Wiechert, W., and A. A. de Graaf. 1997. Bidirectional reaction steps in metabolic

networks: I. Modeling and simulation of carbon isotope labeling experiments.
Biotechnol. Bioeng. 55:101-117.

Wiechert, W., M. Mollney, S. Petersen, A. A. de Graf, and 2001. A universal
framework for 13C metabolic ﬂux analysis. Metab. Eng. 3:265-283.

Wilke, D. 1999. Chemicals from biotechnology: molecular plant genetics will
challenge the chemical and fermentation industry. Appl. Microbiol. Biotechnol.
52: 135-145.

Wood, A. 2004. Bioprocessing, p. 15-17, Chemical Week, vol. 166.

Yang, Q, Q. Hua, and H. Shimizu. 1999. Development of a kinetic model for L-
lysine biosynthesis in Corynebacterium glutamicum and its application to metabolic
control analysis. J. Biosci. Bioeng. 88:393-403.

Yoo, J. Y., and J. G. Zeikus. 1996. Modulation of phosphoenolpyruvate metabolism
of Anaerobiospirillum succiniciproducens ATCC 29305. J. Microbiol. Biotechnol.
6:43-49.

Zamboni, N., N. Mouncey, H. P. Hohmann, and U. Sauer. 2003. Reducing
maintenance metabolism by metabolic engineering of respiration improves
riboﬂavin production by Bacillus subtilis. Metab. Eng. 5:49-55.

Zeikus, J. G., M. K. Jain, and P. Elankovan. 1999. Biotechnology of succinic acid

production and markets for derived industrial products. Appl. Environ. Microbiol.
51:545-552.

45

Chapter 2

Insights into Actinobacillus succinogenes fermentative metabolism in a chemically

defined growth medium

McKinlay, J. B., J. G. Zeikus, and C. Vieille. 2005. Insights into Actinobacillus

succinogenes fermentative metabolism in a chemically deﬁned growth medium. Applied

and Environmental Microbiology. 71:6651-6656.

46

2.1 ABSTRACT

Chemically deﬁned media allow for a variety of metabolic studies that are not
possible in undeﬁned media. A deﬁned medium, AM3, was created to expand the
experimental opportunities for investigating the fermentative metabolism of succinate-
producing Actinobacillus succinogenes. AM3 is a phosphate-buffered medium containing
vitamins, minerals, NH4Cl as the main nitrogen source, and glutamate, cysteine, and
methionine as required amino acids. A. succinogenes growth trends and endproduct
distributions in AM3 and rich medium fermentations were compared. The effect of
N aHC03 concentration in AM3 on endproduct distribution, growth rate, and metabolic
rates were also examined. The A. succinogenes growth rate was 1.3 to 1.4 times higher at
25 mM NaHCO3 than at any other NaHCO3 concentration, likely because both energy-
producing metabolic branches (i.e., the succinate-producing branch and the forrnate,
acetate, and ethanol-producing branch) were functioning at relatively high rates in the
presence of 25 mM bicarbonate. To improve the accuracy of the A. succinogenes
metabolic map, the reasons for A. succinogenes glutamate auxotrophy were examined by
enzyme assays and by testing the ability of glutamate precursors to support growth.
Enzyme activities were detected for glutamate synthesis that required glutamine or a-
ketoglutarate. The inability to synthesize a-ketoglutarate from glucose indicates that at
least two tricarboxylic acid cycle-associated enzyme activities are absent in A.

succinogenes.

47

2.2 INTRODUCTION

Bio-based chemical production is a growing multi—billion dollar industry converting
renewable resources into valuable products (20, 21). A US $15 billion market could be
based on succinate for producing bulk chemicals such as 1,4-butanediol (a precursor to
“stronger-than-steel” plastics), ethylene diamine disuccinate (a biodegradable chelator),
diethyl succinate (a green solvent for replacement of methylene chloride), and adipic acid
(nylon precursor) (24). However, the cost of bio-based succinate is not yet competitive
with petrochemical-based alternatives such as maleic anhydride. The development of a
cost-effective industrial succinate fermentation will rely on organisms able to produce
high concentrations of succinate and at high rates.

Actinobacillus succinogenes is a capnophilic, facultatively anaerobic, gram negative
bacterium that naturally produces high concentrations of succinate as a fermentation
endproduct in addition to formate, acetate, and ethanol (4-6, 18). A. succinogenes
converts glucose to phosphoenolpyruvate (PEP), at which point metabolism splits into
two branches: (i) the formate, acetate and ethanol-producin g C3 pathway, and (ii) the
succinate producing C4 pathway (Fig. 1). Metabolic engineering of A. succinogenes has
begun, with the aim of achieving a homosuccinate fermentation. The most notable
success has arisen from inactivation of pyruvate-forrnate lyasc (PFL) by selecting
mutants resistant to ﬂuoroacetate (4, 13). A. succinogenes PFL mutants have increased
succinate yields, however signiﬁcant amounts of pyruvate are also formed.

Modem, efﬁcient metabolic engineering strategies rely on a thorough understanding
of the metabolism under study, and of how metabolism responds to environmental and

genetic perturbations (2, 16). This understanding can be obtained by using l3C-labeling

48

 

Glc ——>(;6P *
(I) (3) A (2) I

€02
(4) PEP (8)
A/ of} OAA

For A?” :0. (9)

 

_"“(12)
(6 ) ACCOA (‘7) "'§ TCA cycle
/ \ Cit 10) S c
EtOH Ace 9.1112, aKG‘(“"‘

Figure 2.1. Simplified metabolic map of A. succinogenes central metabolism. Thin
black arrows are glucose uptake, pentose phosphate pathway (PPP), and Embden-
Meyerhoff-Parnas pathway (EMP) reactions. Grey arrows are C3 pathway reactions.
Thick black arrows are C4 pathway reactions. Dashed arrows are TCA-associated
reactions that have not been tested. 1. hexokinase or PEP: glucose phosphotransferase; 2.
PPP; 3. EMP; 4. pyruvate kinase and PEP: glucose phosphotransferase; 5. pyruvate-
formate lyase; 6. acetaldehyde dehydrogenase and alcohol dehydrogenase; 7.
phosphotransacetylase and acetate kinase; 8. PEP carboxykinase; 9. malate
dehydrogenase, fumarase, and fumarate reductase; 10 succinyl-CoA synthetase, aKG
dehydrogenase, and aKG synthase; 11. isocitrate dehydrogenase and aconitase; 12. citrate
lyase and citrate synthase. Metabolites: Glc, glucose; G6P, glucose-6-phosphate; Pyr,
pyruvate; For, forrnate; AcCoA, acetyl-CoA; EtOH, ethanol; Ace, acetate; OAA,
oxaloacetate; Suc, succinate; Cit, citrate.

experiments to measure intracellular metabolic ﬂuxes. These experiments require a
deﬁned growth medium so that cell components (e.g., amino acids) are synthesized from
labeled substrate (e.g., l3C-glucose) and not from complex media components such as
yeast extract. Furthermore, an accurate metabolic map is essential for metabolic ﬂux
analyses. A deﬁned medium, AM3, for growing wild-type A. succinogenes is described.
A common experiment for succinate-producing capnophiles is conducted in AM3 with
different NaHCO3 concentrations that provides new insights into A. succinogenes
metabolism. Finally, we improve the A. succinogenes metabolic map in the poorly
characterized region of its TCA-cycle by using experiments made possible by one of A.

succinogenes amino acid auxotrophies and by the advent of AM3.

49

2.3 MATERIALS AND METHODS

2.3.1 Chemicals, bacteria, and culture conditions. All chemicals were purchased from
Sigma-Aldrich (St. Louis, MO) unless otherwise stated. Escherichia coli K12 (ATCC
10798) and A. succinogenes type strain 1302 (ATCC 55618) were obtained from the
American Type Culture Collection. All liquid cultures were incubated at 37°C and shaken
at 250 rpm. Cultures were inoculated with cell suspensions that were harvested in late log
phase, washed twice in sterile saline, and resuspended in an appropriate volume of sterile

saline to give a starting OD660 of 0.1 after inoculation.

2.3.2 Identification of a defined growth medium, AM3. The deﬁned medium was
based on the phosphate buffer of the rich medium, Medium A, commonly used to grow
A. succinogenes (13, 14, 18), and it contained per liter: 15.5 g K2HPO4, 8.5 g
Na2HPO4*H2O, 1 g NaCl, 2 g NH4C1, and 10 ml mineral mix. This basal solution was
aliquoted into 28-ml anaerobic test tubes. The tubes were then sealed with rubber bungs
and aluminum crimps, and repeatedly ﬂushed and evacuated with N2. After autoclavin g,
each tube received ﬁlter-sterilized vitamin mix, kanamycin, amino acids, glucose, and
NaHCO3 to ﬁnal respective concentrations of 2 ml/L, 10 ug/ml, 0.08 %, 50 mM, and 30
mM. Concentrations of basal solution and supplement stocks were adjusted to give total
culture volumes of 10 ml. Soluble NaHC03 was used instead of insoluble MgCO3 (4, 5,
18) to facilitate growth measurements by optical density. One M NaHC03 stock solutions
under 100 % CO2 atmosphere were prepared as described (19). An equal volume of

sterile 100 % CO2 was used to replace any volume of NaHCO3 taken from the stock

50

solutions. The ﬁnal medium pH was 6.9 - 7.1, depending on the amount of NaHCO3
added. The mineral mix was based on Lovley (11) and contained per liter: 1.5 g
nitrilotriacetic acid, 3 g MgSO4*7H2O, 0.5 g MnSO4*H20, 0.1 g FeSO4*7H2O, 0.1 g
CaCl2*2H2O, 0.1 g CoCl2*6H2O, 13 mg ZnCl2, 10 mg CuSO4*5H20, 10 mg
A1K(SO4)2*12H2O, 10 mg H3BO3, 25 mg Na2MoO4, 25 mg NiCl2*6H2O, 25 mg
Na2WO4*2H2O, and 10 mg NaSeO3. The vitamin mix was based on Wolin et a1. (22) and
contained per liter: 10 mg biotin, 10 mg folic acid, 50 mg pyridoxine HCl, 25 mg
thiamine HCl, 25 mg riboﬂavin, 25 mg nicotinic acid, 25 mg pantothenic acid, 0.5 mg
cyanocobalamin, 25 mg P-aminobenzoic acid, and 25 mg thioctic acid.

The inoculum was A. succinogenes 1302 grown from glycerol frozen stocks in 10
ml of BBL trypticase soy broth (TSB; Becton Dickinson, Sparks, MD) containing 50 mM
glucose, 30 mM NaHCO;;, and 10 ug/ml kanamycin in 15-ml screw cap glass tubes with
air as headspace. The defined medium was inoculated with 0.5 ml of washed cell
suspension. The original deﬁned medium supporting growth contained 12 amino acids
(i.e., glutamate, aspartate, cysteine, tyrosine, phenylalanine, serine, alanine, isoleucine,
valine, arginine, leucine, and methionine) that were chosen based on literature for
Haemophilus inﬂuenzae deﬁned media (7, 8). Cells grown in this medium were washed
and used to inoculate various deﬁned media containing 11 amino acids, each medium
missing one of the initial 12 amino acids. This procedure was repeated with fewer and
fewer amino acids until the amino acids required for growth were identiﬁed. The defined

medium with the fewest amino acids still supporting growth was called AM3.

51

2.3.3 Growth of A. succinogenes on AM3 solid agar medium. AM3 agar was prepared
as for the liquid medium, with the addition of 1.5% Bacto agar (Becton Dickinson) and
with or without 10 g/L MgCO; prior to autoclaving. N aHCO3 was added to some
preparations after autoclaving to a ﬁnal concentration of 30 mM. A. succinogenes was
grown in liquid AM3 and washed as described. After aerobic inoculation, plates were

incubated at 37°C in an anaerobic jar with a C02 headspace.

2.3.4 Determination of growth trends and fermentation balances in AM3 and
Medium A.

Anoxic Media A and AM3 (11 ml ﬁnal volume in 28-ml test tubes) were inoculated with
0.25 ml of washed cells grown in identical media. Medium A differs from AM3 by
having 5 g/L yeast extract in place of the vitamins, minerals, amino acids, NaCl, and
NH4C1 in AM3. Both media contained 150 mM NaHCOg. Growth was monitored
throughout log phase by measuring OD660 with a Spectronic 20 (Bausch and Lomb,
Rochester, NY), which does not require culture sampling. Growth rates were determined
from 4 — 6 measurements. Samples (< 1 ml) were collected at the beginning of
incubation, once during log phase (0.6 — 1.0 OD660), and at the end of log phase. The
optical densities of these samples were determined using a DU 650 spectrophotometer
(Beckman, Fullerton, CA). These OD560 values (more precise than those obtained with
the Spectronic 20) were used to calculate ﬁnal OD, and in carbon and electron balances.
Glucose and metabolic endproducts in the sample supematants were separated by HPLC
(Waters, Milford, MA) on a 300 x 7.8 mm Aminex HPX-87H column (Bio-Rad,

Hercules, CA) at 23°C with 4 mM H2804 as the eluent, at a ﬂow rate of 0.6 ml/min.

52

Glucose and ethanol were quantiﬁed using a Waters 410 differential refractometer, and

organic acids were quantiﬁed using a Waters 2487 UV detector at 210 nm.

2.3.5 Determination of fermentation balances, growth rates, and product formation
rates in AM3 with different NaHCO3 concentrations. Anoxic AM3 was prepared as
described but with NaHC03 concentrations ranging from 5 to 150 mM. The inoculum
was 0.25 ml of washed culture grown in AM3 of identical NaHCO3 concentration.
Sample collection and determination of cell densities, growth rates, and endproduct
concentrations were performed as described above. CO2 was detected by transferring 1
ml of culture headspace and 0.3 ml of liquid cultures to separate bung-sealed 13-ml
serum vials. The liquid sample was acidiﬁed with 50 ul of 3.2 N H2SO4. Vial headspaces
were sampled using a pressure syringe and injected into a Series 750 gas chromatograph
(GOW-MAC, Bethlehem, PA) equipped with a Carbosphere column, methanizer, and
ﬂame ionization detector. Speciﬁc rates were calculated as described for batch cultures

( 15, 16). For example, to calculate a speciﬁc product formation rate the equation rp =
Yxpp, was used, where rp is the speciﬁc product formation rate, Yxp is the amount of

product produced per gram of biomass, and ,u is the growth rate.

2.3.6 Preparation of crude cell extracts and enzyme assays. A. succinogenes 130Z was
grown in 450 ml Medium A containing 33 mM glucose and 15 mM NaHC03 in l L
spherical ﬂasks with an N2 headspace. Cultures were harvested in log phase by
centrifugation, washed once with 200 ml 0.1 M Tris-HCl (pH 7.7), and resuspended in 20

ml 0.1 M Tris-HCl (pH 7.7). Cells were lysed by two passages through a French press at

53

1,200 -1,400 lb under an N2 headspace. Cell extracts were stored at -20°C before assays.
The cell extract protein concentration (i.e., 1.8 mg protein /ml) was quantiﬁed by the
bicinchoninic acid assay (Pierce, Rockford, IL) with bovine serum albumin as the
standard (12).

Enzyme activities were assayed by measuring the oxidation or reduction of
NADP(H) using a Cary 300 spectrophotometer (Varian, Palo Alto, CA). An extinction
coefﬁcient of 6.23 cm'1 mM'1 at 340 nm was used for NADPH (18). Reagents were
dissolved in 0.1 M Tris-HCl (pH 8.0), and reactions were carried out in triplicate in 1 ml
volumes at 37°C. The reaction mixture for the glutamate dehydrogenase assay contained
40 mM NH4C1, 5 mM a-ketoglutarate (aKG), 0.3 mM NADPH, 1 mM CaCl2, and 25 ul
cell extract. The reaction was started by the addition of aKG. Glutamate synthase activity
was tested in the presence of 5 mM glutamine, 5 mM aKG, 0.3 mM NADPH, 1 mM
CaCl2, and 25 ul cell extract. The reaction was started by the addition of glutamine.
Isocitrate dehydrogenase was assayed using anoxic reagents in rubber-stoppered cuvettes
that were evacuated and ﬂushed with N2 as described (23). The reaction mixture
contained 0.1 M NaCl, 5 mM MgCl2, 1 mM dithiothreitol, 0.3 mM NADP“, 50 ul cell
extract, and 5 mM isocitrate (17). Cell extracts (15.2 mg protein/ml) from E. coli K12
aerobically grown in LB with 25 mM glucose were used as a positive control. The
reaction was started with the addition of isocitrate. No enzyme activity was detected in

any assay when NAD(H) was used in place of NADP(H).

2.3.7 Test of potential glutamate precursors to support growth. Anoxic defined

medium was prepared as described for AM3, but glutamate and/or NH4Cl were omitted

54

where appropriate. Filter-sterilized stock solutions of potential glutamate precursors (i.e.,
aKG, glutamine, aspartate, isocitrate, and citrate) were added to the autoclaved medium
to 15 mM ﬁnal concentration. The inoculum was 0.25 ml of A. succinogenes grown in
AM3 and washed as described. The culture volume was 12 ml. The turbidity was
monitored until stationary phase was reached or for ﬁve days using a Spectronic 20. If
growth occurred, cells were washed as before and used to inoculate identical medium to

ensure that growth was not due to nutrient carry over.

55

2.4 RESULTS AND DISCUSSION

2.4.1 Creation of the defined growth medium, AM3. A. succinogenes grew slowly
(0.06 hr") when ﬁrst transferred from TSB to deﬁned medium containing the initial 12
amino acids, with ﬁnal OD660 values ranging from 0.7 to 1.1. After several transfers in
deﬁned medium, growth rates increased to 0.14 hr'l. This improvement could be due to a
slow response in gene regulation to suit the new growth conditions or to genetic drift.
After removing amino acids from the deﬁned medium one at a time, the amino acid
requirements of A. succinogenes were determined to be cysteine, glutamate, and
methionine. To improve the A. succinogenes growth rate and ﬁnal OD in deﬁned
medium, concentrations of amino acids, NH4C1, vitamin mix, and mineral mix were
varied, and their effects on growth rate and ﬁnal OD were determined. Mineral mix,
vitamin mix, and amino acids were required for anaerobic growth on glucose. Increasing
the vitamin concentration from 2 mllL to 10 ml/L doubled the growth rate and tripled the
ﬁnal OD. A. succinogenes grew without NH4Cl when glutamate, cysteine, and
methionine were present but the growth rate (0.03 :I: 0.00 hr '1) and ﬁnal OD660 (0.44 :I:
0.07) were poor. The improved medium, called AM3, contained per liter: 15.5 g K2HPO4,
8.5 g Na2HPO4*H2O, 1 g NaCl, 2 g NH4Cl, 0.15 g L-glutamate, 0.08 g L-cysteine—HCl,
0.08 g L-methionine, 10 ml mineral mix, 10 ml vitamin mix, 30 mmol NEIHCO}, and 50
mmol glucose.

A. succinogenes also grew on solid AM3 agar. One-mm sized colonies developed
after 2 - 4 days of incubation under CO2 gas phase at 37°C. Colonies developed with and

without MgC03 or NaHCO3.

56

2.4.2 Growth trends and fermentation balances in AM3 and Medium A. In a defined
medium, bacteria are forced to synthesize a number of cellular building blocks that would
otherwise be available from rich medium components. For this reason, growth rates were
lower in AM3 (0.24 1 0.01 hr") than in Medium A (0.43 :e 0.01 hr"). The ﬁnal on660 in
AM3 (2.82 :1: 0.05) was slightly lower than in Medium A (3.03 i 0.14). Since most of the
succinate is produced during log phase, fermentation balances were based on log phase
samples. While carbon and electron recoveries for cultures grown in AM3 were near
100%, recoveries for cultures grown in Medium A exceeded 100% (Table I) likely
because carbon and electron recoveries only take into account the glucose consumed. The
yeast extract carbon in Medium A is ~50% that of the supplied glucose, according to the
BD Diagnostic Systems website
(http://www.bd.com/diagnostics/microservices/regulatory) and Doyle et al. (1). Thus,
there is ample carbon in yeast extract to explain a 117% carbon recovery in Medium A.
Yeast extract may also have contributed to the higher formate and acetate yields and to
the lower succinate product ratio in Medium A compared to AM3. With no undeﬁned
carbon sources to track in AM3, the comparison of fermentation balances in AM3 and

Medium A illustrates how a chemically deﬁned medium facilitates metabolic studies.

2.4.3 Effect of AM3 NaHCO3 concentration on fermentation balances, growth rates,
and metabolic rates. Succinate production by A. succinogenes requires CO2, presumably
as a substrate for PEP carboxykinase (9, 18) (Fig.1). We previously showed that A.

succinogenes produces more succinate and less alternative endproducts when the MgCO3

57

002088005. .0. 08 0005 0000 00» 00: 00:

£5 00530; 53800 0000038300 000283 00 0:0 00 x08 3:080 0:0 0000000 c.0 83 05 :08 000808 000. 00 80800005 408050
0:0 0000000 00 8:0 05 8 .50 00800000 8 0:0 HE 808 008800 8 000808 :0 005 0080000 00>» = .200080 + 00000000 \ 0008026 0
.Amv :0w082 03200 :0 00000 0.00 00803000 00802.".— 0

0080800 000003 80: 003000 00080000 00 003000800 008050 05 m0 .000: 003 «00:00 60000005. .3: 000808

00080000 038 00a 0008 003 «DU 00 0.08 0:0 008 0008 003 00:08:30 =< 0080800 000008 8 80800 \ 003005 8 80800 n
.05 0250 $3.00 27020an0

h.0 002000800 :00 0 0:0 00090 000 _8 \ £803 :00 b0 «8 Sn 00 m020> 0080000 waded 008800000 003 000805 _088 0

00.80.00 0000288 800“. 0080300 200080 H 8008 000 000Q

< 8.5002 6:0 m2< 8 08003050030 .< «0 00080—0.— =0=8=0800u 000nm mod ._.N 030,—.

 

 

 

2 880020
modﬁmﬁo can: 10>: mHoE 0H2 mﬂow 0H3 ﬁnch :03
$25
modﬁgd chE NHS mHoE mﬁo «Hg mHE _Ho\l
0080C
0800 003005 A000 >00>000M €er b0>000m 0000805 _0:0£m 000003.. 8080"“ 0008005
0 e 0 88002

0008005 00800:.”— =0€00 0080800 00002» _088 9: 203005 .088

 

58

concentration was increased in Medium A (18). To conﬁrm that this trend holds in AM3,
we compared endproduct distributions in AM3 with different NaHCO3 concentrations.
Indeed, increasing the NaHCO3 concentration in AM3 increased the succinate product
ratio (Table 2). In addition to conﬁrming this trend, we also wanted to determine an
optimal NaHCO3 concentration for succinate production in AM3. In our previous study,
the maximum MgCO3 concentration tested was equimolar to that of the supplied glucose
(i.e., 55 mM) (18). Here we tested NaHCO3 concentrations up to 3 times the molar
concentration of supplied glucose. As seen in Table 2, the succinate production ratio
plateaued at NaHC03 concentrations above 75 mM. Although not quantitatively precise,
CO2 was detected in all cultures at the time of log phase sampling (data not shown).
Therefore, the reported succinate yields are not due to complete CO2 consumption. While
CO2 was not completely consumed, the endproduct distributions suggest that NaHCO3, at
5 and 25 mM, may be limiting succinate production.

As shown in Table 2, the average growth rate was statistically higher at 25 mM
NaHCO3 than at any other NaI-IC03 concentration (t-test, 2-tail, equal variance, p =
0.001). This high growth rate can be explained by the metabolic rates shown in Table 3.
At 5 mM NaHCO3, C3 ﬂux (estimated from speciﬁc acetate and ethanol formation rates)
is high, while C4 ﬂux (i.e., the speciﬁc succinate formation rate) is low. At 25 mM
N aHCOg, C3 ﬂux remains high, with a maximum acetate formation rate, and C4 ﬂux
reaches a maximum. At 75 mM NaHC03 and above, C4 ﬂux remains high, while C3 ﬂux
decreases by about half. ATP can be derived from PEP conversion to succinate via the C4
pathway (1.67 mol ATP), and from ethanol (1 mol ATP) or acetate (2 mol ATP)

productions via the C3 pathway. With high C3 and C4 ﬂuxes at 25 mM NaHCO3, more

59

000.. 000.008. 0.00.0000 .000 00000000 00000.0 50300. 000000 000805 0080.00 0.00.3.0 800. 0000300 0000003 0 00008 0.0 0.00.

05.... 8 0.0.. 5.50..» .80 00.80.80... .0.—008.000 :0 00308000000 060—007. 00 80.5— .N.N 030,—.

.. 0.00... 0. 00000000 00 0000.00.00 0003

 

 

.0.000N.0 N0.0000.0 N000. N00 0000. N00 N00 00.0 000.. 00.
.0.000N.0 00.00000 N00: N000. .000. 000. .000 N000 .0.0 N.
00.00 0N.0 00.00000 000.. 00 N0. 0. 000. .00. 00 .0 0000 .000 00
.0.00N0.0 00.00000 N00.. N0 .0. 0000. 000N .000 .0 .0 N0Nm 0N
.0.000N.0 N0.00NN.0 N000. .0 .0. 0000. 00 .0 .0N0 .00.. .000 0
00. 00.
3-05 0000 000.0000. 00080.0. .000...m. 0.0000. 0.0800”. 0.00.026 9.8.
80000.0. >00>000m

0.0. 0.3000 0.08005 .00000 0000.02

000005

000000

 

00800000 00000.w .088 00. 200000. .088

 

0
6

ATP is generated for biosynthesis, explaining the high growth rate in these conditions.
The maximum succinate formation rate at 25 mM NaHCO;; also indicates that NaHCO3 is
not limiting succinate production, unlike what Table 2 endproduct distributions suggest.
It is also worth noting that the high speciﬁc product formation rates at 25 mM
NaHC03 are allowed by a high glucose consumption rate (Table 3). This high glucose
uptake rate (and expected corresponding high glycolytic ﬂux) should be able to support a
succinate formation rate higher than the observed 4 mmol x g biomass'l x hr'l. Not
enough is known at this stage to identify the regulatory mechanism or mechanisms
behind this observation. Among the possible explanations are that (i) high NaHCO3
concentrations inhibit the C3 pathway and that the C4 pathway cannot process substrate
faster than 4 mmol x g biomass'l x hr". This bottleneck would in turn inhibit glucose
uptake and glycolysis. (ii) Another possibility is that NaHCO3 (or CO2) inhibits glucose
uptake and/or glycolysis. Any of these mechanisms would have important implications
for the metabolic engineering of A. succinogenes succinate production. Future metabolic
ﬂux analyses, in which genetic or environmental perturbations individually affect
individual pathway branches, will focus on identifying what limits the succinate
production rate in A. succinogenes.
2.4.4 A. succinogenes is missing at least two TCA cycle-associated enzyme activities.
A. succinogenes was found to be auxotrophic for cysteine, methionine and glutamate.
Glutamate auxotrophy was initially surprising since A. succinogenes cell extracts have
aspartate: glutamate transaminase activity (18). Figure 2 shows possible enzyme

activities leading to glutamate, not all of which are known to be present in A.

61

0.0... 00 00.08000 + 0.0. 0000000... 0.00.0000 0.0.0000 n 0.0... 0..0.000.w 00.08..0m. e

.00.. 000000. .0.. 0.0.... 00.0 000000... 000.0000. 0.0.080.

.>.. 000 .0000. . 00000.0 0000 0000.00.80.00 mmn. 000 .0000... 0.0.000 .0000... 0.08.00 0.... .0000.0.0.00...0..n.00..n. 3 00800000

0. 0.0.0. . A... .0000... 0.08.00 3 0000000... 0.00. w0..00>0.n. 80.000 00800000000002... 00000.?me 3 .0 00080.80; m0.N....0
.000. .3 00.....0 0.0a. . 00800000 00.0.00 00000.w 0.. 030:0. 00 0.0 00030.00 0:30.08 .0008 3 000088 000 0000800000 0.00. .0
0000080000 0...... 802000.08 .00000 0. .8... 0000800000 n....< 00.08000 I 000080. 0.00. 00.08..0m. .00..08..0.. 0.0.0. .00 00.08..0m. 0

0.0. 0000800. .8080 00.6000 + 0.0. 00008.0. 0.0.000 0....0000 n 0.0... 00 00.08..0m. 0

.0008... 00.08.00 000 00.0.. 00000.08 0.0.02.0 :0 00.22.002.00 0000.07. ..0 .0000... .0.N 0.00.0

 

 

0.00 N... 0.00 0.0. N00 N.0 .00 0.0 ..00 0.0 .00 0.0 .00 0.0 0.00 0.0 00.
N0 0 0.0 0.0 0 0.0. N0 0 0.0 ..0 0 0.0 ..0 0 0.0 .d 0 N0 ..0 0 0.0 N... 0 0.0 0N.
.00 N.0 N00 0N. ..00 0.0 .00 0.0 0.00. .0 .00 0.N .00 0.0 N.o00.0 00
0.0 0 0... 0.0 0 0.0. N0 0 0.0 N0 0 N.N ..0 0 0.0 ..0 0 0.0 N0 0 0.0 0.0 0 0.0 0N
0.0 0 0.0. 0.0 0 0.0. 0.0 0 N... 0.0 0 0.0 0.0 0 0.0 ..0 0 0.0 ..0 0 N.N 0.0 0 0.0 0
00.0... n.00..08..0. 0.0.... 00 0.28.
0..0.000.m 0.0.40. .00 00.0.. 00 00008.8 00008.0. 00008.8 .0 00008.0. 0000800000 .00000
00.08..0m. 00.0800... 00.08..0m. .000...m. 0.0.00< 0.08.0". 0.00.0000 00000.0 000:0 Z

 

0.- 0.. x T 00080... 0 x .088. 00.0. 0.0.0000

62

succinogenes. Several glutamate precursors (i.e., aKG, isocitrate, citrate, succinate) are
TCA cycle intermediates. It is still unclear whether A. succinogenes has a complete TCA
cycle (Fig. 1). Because a complete TCA cycle would mean at least two pathways for
succinate production and/or consumption, we used A. succinogenes's glutamate
auxotrophy to our advantage to study a poorly characterized region of the A.
succinogenes central metabolic map. Table 4 shows that aKG can replace glutamate in
the growth medium when NH4Cl is present, indicating in vivo glutamate dehydrogenase
activity. Aspartate plus aKG also supported growth, while aspartate alone did not. These
results suggest that aspartate: glutamate transaminase is functional in vivo. Alternatively,
aspartase activity could convert aspartate to fumarate and NHX, and then NH4+ used with
aKG by glutamate dehydrogenase to produce glutamate. Growth on glutamine indicates
the presence of a glutamine deaminating activity (e.g., glutamine synthetase or carbamoyl
phosphate synthetase). In vitro enzyme activity assays suggested that glutamate
dehydrogenase (1,100 :l: 180 nmol x min'l x mg protein!) and glutamate synthase (30 d:
10 nmol x min'l x mg protein") are also functional in A. succinogenes. Taken together,
these results suggest that all the enzyme activities (i.e., 3, 6, 7, and 8) below aKG in
Figure 2 are present in A. succinogenes.

These results also point to a single reason for A. succinogenes's glutamate
auxotrophy: A. succinogenes cannot synthesize aKG from glucose. This inability means
that enzymes are absent or inactive in two pathways: (i) between succinate and aKG in
the reverse TCA cycle (especially since A. succinogenes produces ample succinate), and
(ii) in the TCA cycle from acetyl-CoA and oxaloacetate to citrate to aKG (Fig. 1). This

conclusion is supported in part by the fact that no in vitro isocitrate dehydrogenase

63

Cit Suc

ATP + CoA
(l) (4)
ADP
NADP S-CoA
NADPH+ C02
NADH
4* C02
NAD+
+ CoA
NADPH NH4
+ (3)
NADP
NADP
NADPH
4' aKG
ATP
Gln

Figure 2.2. Possible enzyme activities leading to glutamate synthesis. 1, aconitase; 2,
isocitrate dehydrogenase; 3, glutamate dehydrogenase; 4, succinyl-CoA synthetase; 5,
aKG dehydrogenase; 6, aspartate: glutamate transaminase; 7, glutamate synthase; 8,
glutamine synthetase. Metabolites: Ict, isocitrate; S-CoA, succinyl-CoA.

 

 

 

Supplement
NH4+ NH4+ Asp
N114+ + + Gln Asp +
Gln aKG aKG
Growtha - + + + - +

 

Table 2.4. Ability of glutamate precursors to support growth of A. succinogenes in
AM3. a Growth (+) is deﬁned as > 1.5 OD660 within 24 hours. No growth (-) is deﬁned as
no OD660 increase within 5 days after inoculation. Tests were performed at least in
duplicate.

activity could be detected in anaerobically and aerobically grown A. succinogenes cell
extracts, while it was detected in E. coli cell extracts as a positive control (70 :t 10 nmol
NADP(H) min'l mg protein'l). Growth experiments on citrate or isocitrate were not
informative (A. succinogenes did not grow when citrate or isocitrate were supplied with
NH4Cl or aspartate, data not shown) for at least two reasons: (i) it is not known whether
citrate and isocitrate are taken up by A. succinogenes cells; and (ii) citrate prevented A.
succinogenes growth at concentrations above 3 mM in the presence of glutamine or
glutamate (data not shown). This inhibition was countered by adding extra minerals (data
not shown), suggesting that citrate binds essential minerals (e. g., iron) and prevents
mineral acquisition.

A. succinogenes is a promising catalyst for bio-based production of succinate and
potentially other chemicals (e.g., malate, fumarate, 5-aminolevulinate, aKG, and
glutamate). We have described a chemically deﬁned medium for growing A.
succinogenes and studying its metabolism. NaHCO3 concentration had pronounced
effects on fermentation end product distributions between 5 and 75 mM but not at higher
concentrations. A. succinogenes had an optimal growth rate at 25 mM NaHCO3, where
both energy producing pathways displayed their highest ﬂuxes. aKG could be used in
place of glutamate to support growth, indicating that at least two TCA cycle-associated
enzyme activities are absent. The deﬁned medium made testing growth on glutamate
precursors possible. The discovery that A. succinogenes lacks a full TCA cycle is a key
information for the construction of an accurate A. succinogenes metabolic map that will
be essential in future metabolic flux analyses and practical metabolic engineering designs

for A. succinogenes-based chemical production.

65

2.5 ACKNOWLEDGEMENTS

This work was supported by the National Science Foundation grant #BES-0224596.
James McKinlay was supported by a Michigan State University Quantitative Biology and
Modeling Initiative fellowship. We thank Dr. John Breznak, Dr. Yair Shachar-Hill,

Harini Krishnamurthy, and Maris Laivenieks for helpful discussions.

66

2.6 REFERENCES

1.

10.

11.

12.

Doyle, A., M. N. Weintraub, and J. P. Schimel. 2004. Persulfate digestion and
simultaneous colorimetric analysis of carbon and nitrogen in soil extracts. Soil Sci.
Soc. Am. J. 68:669-676. ‘

Fell, D. 1997. Understanding the control of metabolism. Portland Press, London.
Gottschalk, G. 1986. Bacterial Metabolism, 2nd ed. Springer-Verlag, New York.

Guettler, M. V., M. K. Jain, and D. Rumler. 1996. Method for making succinic acid,
bacteria] variants for use in the process, and methods for obtaining variants. US.
patent 5,573,931.

Guettler, M. V., M. K. Jain, and B. K. Soni. 1996. Process for making succinic acid,
microorganisms for use in the process and methods of obtaining the microorganisms.
US. patent 5,504,004.

Guettler, M. V., D. Rumler, and M. K. Jain. 1999. Actinobacillus succinogenes sp.
nov., a novel succinic-acid-producing strain from the bovine rumen. Int. J. Syst.
Bacteriol. 49:207-216.

Herroitt, R. M., E. M. Meyer, and M. Vogt. 1970. Deﬁned nongrowth media for stage
H development of competence in Haemophilus inﬂuenzae. J. Bacteriol. 101:517—524.

Herroitt, R. M., E. Y. Meyer, M. Vogt, and M. Modan. 1970. Deﬁned medium for
growth of Haemophilus inﬂuenzae. J. Bacteriol. 101:513-516.

Kim, P., M. Laivenieks, C. Vieille, and J. G. Zeikus. 2004. Effect of overexpression
of Actinobacillus succinogenes phosphoenolpyruvate carboxykinase on succinate
production in Escherichia coli. Appl. Environ. Microbiol. 70:1238-1241.

Kroger, A., S. Biel, J. Simon, R. Gross, G. Unden, and C. R. D. Lancaster. 2002.
Fumarate respiration of Wolinella succinogenes: enzymology, energetics and
coupling mechanism. Biochim. BiOphys. Acta. 1553:23-28.

Lovely, D. 2000. Dissimilatory Fe(III)-and Mn(IV)-Reducing Prokaryotes. In Martin
Dworkin (ed.), The Prokaryotes. Springer-Verlag [Online.]
http://141 . 150.157. 1 17:8080/prokPUB/index.htm. Accessed December 2003.

McKinlay, J. B., and J. G. Zeikus. 2004. Extracellular iron reduction is mediated in

part by neutral red and hydrogenase in Escherichia coli. Appl. Environ. Microbiol.
70:3467-3474.

67

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

Park, D. H., M. Laivenieks, M. V. Guettler, M. K. Jain, and J. G. Zeikus. 1999.
Microbial utilization of electrically reduced neutral red as the sole electron donor for
growth and metabolite production. Appl. Environ. Microbiol. 65:2912-2917.

Park, D. H., and J. G. Zeikus. 1999. Utilization of electrically reduced neutral red by
Actinobacillus succinogenes: physiological function of neutral red in membrane-
driven fumarate reduction and energy conservation. J. Bacteriol. 181:2403-2410.

Sauer, U., D. R. Lasko, J. Fiaux, M. Hochuli, R. Glaser, T. Szyperski, K. Wiithrich,
and J. E. Bailey. 1999. Metabolic flux ratio analysis of genetic and environmental

modulations of Escherichia coli central carbon metabolism. J. Bacteriol. 18126679-
6688.

Stephanopoulos, G., A. A. Aristidou, and J. Nielsen. 1998. Metabolic Engineering:
Principles and Methodologies. Academic Press, London.

Thorsness, P. E., and D. E. Koshland, Jr. 1987. Inactivation of isocitrate

dehydrogenase by phosphorylation is mediated by the negative charge of the
phosphate. J Biol Chem 262: 10422-10425.

van der Werf, M. J ., M. V. Guettler, M. K. Jain, and J. G. Zeikus. 1997.
Environmental and physiological factors affecting the succinate product ratio during
carbohydrate fermentation by Actinobacillus sp. 1302. Arch. Microbiol. 167:332-
342.

Widdel, F., and F. Bak. 1992. Gram-negative mesophilic sulfate-reducing bacteria, p.
3358-3378112 A. Balows, H. G. Truper, M. Dworkin, W. Harder, and K. H. Schleifer
(ed.), The Prokaryotes, 2nd ed, vol. 4. Springer-Verlag, New York.

Wilke, D. 1999. Chemicals from biotechnology: molecular plant genetics will
challenge the chemical and fermentation industry. Appl. Microbiol. Biotechnol.
52:135-145.

Wilke, D. 1995. What should and what can biotechnology contribute to chemical
bulk production? FEMS Microbiol. Rev. 16:89-100.

Wolin, E., M. Wolin, and R. Wolfe. 1963. Formation of methane by bacterial
extracts. J. Biol. Chem. 238:2882-2886.

Zeikus, J. G., G. Fuchs, W. KeneaJy, and R. K. Thauer. 1977. Oxidoreductases
involved in cell carbon synthesis of Methanobacterium thermoautotrophicum. J.
Bacteriol. 132:604-613.

Zeikus, J. G., M. K. Jain, and P. Elankovan. 1999. Biotechnology of succinic acid

production and markets for derived industrial products. Appl. Environ. Microbiol.
51:545-552.

68

Chapter 3
Determining Actinobacillus succinogenes metabolic pathways and fluxes by NMR

and GC-MS analyses of l3C-labeled metabolic product isotopomers

McKinlay, J. B., Y. Shachar-Hill, J. G. Zeikus, and C. Vieille. 2006. Determining
Actinobacillus succinogenes metabolic pathways and ﬂuxes by NMR and GC-MS

analyses of 13C-labeled metabolic product isotopomers. Metabolic Engineering. In press,

doi:10.1016/j.ymben.2006.10.006.

69

3.1 ABSTRACT

Actinobacillus succinogenes is a promising candidate for industrial succinate
production. However, in addition to producing succinate, it also produces forrnate and
acetate. To understand carbon ﬂux distribution to succinate and alternative products we
fed A. succinogenes [1-‘3C] glucose and analyzed the resulting isotopomers of excreted
organic acids, proteinaceous amino acids, and glycogen monomers by gas
chromatography-mass spectrometry and nuclear magnetic resonance spectroscopy. The
isotopomer data, together with the glucose consumption and product formation rates and
the A. succinogenes biomass composition, were supplied to a metabolic ﬂux model.
Oxidative pentose phosphate pathway ﬂux supplied, at most, 20% of the estimated
NADPH requirement for cell growth. The model indicated that NADPH was instead
produced primarily by the conversion of NADH to NADPH by transhydrogenase and/or
by NADP-dependent malic enzyme. Transhydrogenase activity was detected in A.
succinogenes cell extracts, as were forrnate and pyruvate dehydrogenases, which the
model suggested were contributing to NADH production. Malic enzyme activity was also
detected in cell extracts, consistent with the ﬂux analysis results. Labeling patterns in
amino acids and organic acids showed that oxaloacetate and malate were being
decarboxylated to pyruvate. These are the ﬁrst in vivo experiments to show that the
partitioning of flux between succinate and alternative fermentation products can occur at
multiple nodes in A. succinogenes. The implications for designing effective metabolic
engineering strategies to increase A. succinogenes succinate production are discussed.
Keywords: Actinobacillus succinogenes, succinate, metabolic ﬂux analysis, biomass

composition, oxaloacetate decarboxylase, malic enzyme, transhydrogenase

70

3.2 INTRODUCTION

Declining oil reserves, rising petrochemical prices, and the environmental impact
of oil-based industries have prompted the development of bio-based processes for fuel
and chemical production (64, 65). Industrial-scale microbial processes are meeting the
global demands for a variety of amino acids, organic acids, vitamins, and antibiotics (64).
Succinate is produced petrochemically from butane to satisfy a specialty chemical
market, but it can also be produced from microbial fermentations (69). More importantly,
bio-based succinate could replace a large petrochemical-based commodity chemical
market for making bulk chemicals including 1,4-butanediol (a precursor to “stronger-
than-steel” and biodegradable plastics), ethylene diamine disuccinate (a biodegradable
chelator), diethyl succinate (a “green” solvent replacement for methylene chloride), and
adipic acid (a nylon precursor) (69). In addition to being based on renewable resources,
bio-based succinate production has the additional environmental beneﬁt of using C02, a
greenhouse gas, as a substrate. Developing a more cost-effective industrial succinate
fermentation to replace the butane-based commodity maleic anhydride will require
advances on several fronts. In particular, it will require developing organisms that

produce high succinate concentrations at high rates.

The capnophilic bacterium, Actinobacillus succinogenes, is a promising
biocatalyst for industrial succinate production. While A. succinogenes produces some of
the highest succinate concentrations ever reported (15, 16), it does so as part of a mixed
acid fermentation, producing high concentrations of formate and acetate as well.
Therefore, it is desirable to genetically engineer A. succinogenes to produce succinate as

the sole fermentation end product.

71

Metabolic engineering is most effective when it is based on an understanding of
the pathways involved and of how the ﬂuxes through those pathways are controlled (12,
51). Knowledge of A. succinogenes metabolism comes from fermentation balances, from
in vitro enzyme assays (34, 55), and, more recently, from its genome sequence
(McKinlay et al., manuscript in preparation., http://genomeornl.gov/microbial/asucl). A
simpliﬁed map of A. succinogenes central metabolism is depicted in Figure 1. Previous
data suggest that A. succinogenes ferments glucose to phosphoenolpyruvate (PEP) by
glycolysis and the oxidative pentose phosphate pathway (OPPP) (55). PEP is thought to
serve as the branchpoint between the formate-, acetate-, and ethanol-producing pathway
(C3 pathway), and the succinate-producing (C4) pathway. Malic enzyme and oxaloacetate
(OAA) decarboxylase activities measured in vitro (55) suggest that malate and OAA can
also serve as branch points between the C3 and C4 pathways. The presence of these
activities and others in vivo, such as those of the Entner Doudoroff (ED) and glyoxylate
pathways (for which questionable in vitro activities were detected [(55)]), must be

determined to develop effective A. succinogenes engineering strategies.

‘3 C-labeling studies are a useful approach for understanding the in vivo workings of
metabolism (42, 60). Knowledge resulting from these techniques has been used to guide
the metabolic engineering of industrial microbes, such as vitamin-producing Bacillus
subtilis (7, 46, 68) and amino acid-producing Corynebacterium glutamicum (26, 38, 40).
13C-labeling studies can distinguish ﬂuxes through different pathways when these ﬂuxes
result in different positional isotopic enrichments in metabolic intermediates. These
labeling patterns are imprinted on metabolic products (e. g., proteinaceous amino acids

and excreted organic acids) and can be analyzed by gas chromatography-mass

72

Figure 3.1. A. succinogenes metabolic pathways addressed in this study. Solid lines:
pathways or reactions for which enzyme activity was detected in vitro; dotted lines:
pathways or reactions where no activity or uncertain activity was detected in vitro (55).
Unidirectional arrows: ﬂuxes considered to be unidirectional (all other ﬂuxes are
considered to be reversible). The C4 pathway is deﬁned as: PEP 9 OAA 9 Mal 9 Fum
9 Suc. The C3 pathway is deﬁned as: PEP 9 Pyr 9 AcCoA 9 Ace + EtOH.
Alternative PPP reactions, cysteine and methionine degradation pathways, and amino
acid synthesis pathways are not shown (Supplementary material). It was assumed that
0.67 ATP is produced per fumarate reductase reaction based on data from Wolinella
succinogenes (27). Metabolites: AcCoA, acetyl-coenzymeA; Ace, acetate; Cit, citrate;
EtOH, ethanol; E4P, erythrose—4-phosphate; For, forrnate; Fum, fumarate; F6P, fructose-
6-phosphate; Glc, glucose; Glxt, glyoxylate; G3P, glyceraldehyde-3-phosphate; G6P,
glucose-6-phosphate; Mal, malate; OAA, oxaloacetate; PEP, phosphoenolpyruvate; Pyr,
pyruvate; R5P, pentose—phosphates, Suc, succinate; S7P, sedoheptulose-7-phosphate.
Pathways and reactions: ADH, alcohol dehydrogenase; AK, acetate kinase; CL, citrate
lyasc; ED, Entner Doudoroff pathway; empl, 2, and 3, Embden-Meyerhoff—Pamas (EMP)
or glycolytic reactions; Fm, fumarase; FR, fumarate reductase; ICL, isocitrate lyase and
aconitase; MDH, malate dehydrogenase; ME, malic enzyme; Msyn, malate synthase;
OAAdec, oxaloacetate decarboxylase; OPP, oxidative pentose phosphate pathway;
PEPCK, PEP carboxykinase; PFL, pyruvate forrnate-lyasc; PK, pyruvate kinase and
PEP: glucose phosphotransferase system (PT S); pppl and 2, transketolase; ppp3,
transaldolase; PyrDH, pyruvate dehydrogenase or PFL coupled with forrnate

dehydrogenase; upt, glucose phosphorylation by hexokinase and PTS.

73

Figure 3.1 (cont’d)

 

 

Glc
upt k ATP C02
2 NADPH
----> GSP 1‘ map
OPP
,.;. empl? 0
ED :
O S7P 15

, F VP 4......"........,....ppp3..‘ kpppj.

:\v NADPH V... ”II-11;. ”:33...“

3 ATP ,° " 7° y
empz E4P G3P asp
* .9 9

Pyr 3‘ G3P + G3P< ' — --------------
pppZ
NADH
oemp3 ATP

PEPCK
COPA::C\:<$:TP

MD. MDH OAA I,——-\o2 Pyr

NADPH

 

 

Fum ‘
3Q GIXt
MAI)“ A.
213 ATP
Suc ' [CL

74

spectrometry (GC-MS) and nuclear magnetic resonance spectroscopy (NMR). We
recently developed a chemically deﬁned growth medium that makes l3C-labeling
experiments with A. succinogenes possible (34). Here we describe the use of [1-

13C] glucose to obtain an A. succinogenes metabolic ﬂux map based on analyses of its cell
composition, extracellular ﬂuxes, and isotopomers of amino acids, organic acids, and

glycogen monomers.

75

3.3 MATERIALS AND METHODS

3.3.1 Chemicals, bacteria, and culture conditions. All chemicals were purchased from
Sigma-Aldrich (St. Louis, MO) unless stated otherwise. A. succinogenes type strain 1302
(ATCC 55618) was purchased from the American Type Culture Collection (Manassas,
VA) and adapted to growth in AM3 as described (34). AM3 is a chemically deﬁned
medium containing Cys, Met, and Glu, which are required for A. succinogenes growth.
Cultures were grown in 11-ml volumes of AM3 in 28-ml anaerobic test tubes or in 60-ml
volumes in ISO-ml serum vials. Anoxic AM3 contained 150 mM NaHC03 and 50 mM
glucose, and was prepared as described (34). Cultures were incubated at 37°C with

shaking at 250 rpm.

3.3.2 Growth and sampling conditions in the presence of [1-13C]glucose. A starter
culture was grown from a glycerol frozen stock in AM3 to an OD660 of 1.0. The starter
culture was harvested, washed once, and resuspended in an equal volume of the AM3
phosphate buffer. Nine ll-ml volumes of AM3 containing 50 mM [1-13C] glucose (99%,
Cambridge Isotope Laboratories, Cambridge, MA), instead of unlabeled glucose, were
each inoculated with 0.1 ml of the cell suspension to give a starting OD660 of ~0.01.
Unlabeled biomass accounted for only 0.5 — 1.8% of the total biomass at the time of
harvesting. Growth was monitored using a Spectronic 20 spectrophotometer (Bausch and
Lomb, Rochester, NY) to calculate growth rates. A l-ml sample was taken 4.5 hours into
growth to serve as a ‘0 time point’ sample, rather than at the time of inoculation to
account for any lag phase. This sample was used for a more accurate OD660 measurement

using a Beckman DU 650 spectrophotometer (Fullerton, CA). It was also analyzed by

76

high performance liquid chromatography (HPLC) to determine glucose and product
concentrations. Culture growth was stopped by chilling on ice as follows: three tubes
were stopped at 0.5 0D660, three at 1.0 0D660, and three at 1.5 0D560. Cell densities were
recorded, then cultures were harvested by centrifugation at 4°C. The supematants were

stored at —70°C, as was the biomass after two washes with 0.9% NaCl at 4°C.

3.3.3 HPLC analysis of glucose and products and GC analysis of headspace gas.
HPLC was performed on culture supematants as described (34). A negative HC03' peak
interfered with ethanol peak integration. To improve the ethanol measurement accuracy,
subsamples were mixed 1:1 with 150 mM H2304 to release HC03' as C02 before
analysis. Untreated samples were used to quantify glucose, succinate, forrnate, acetate,
and fumarate. GC was used to analyze culture headspace for H2 using a Shimadzu GC-
2014 (Columbia, MD) equipped with an 80/100 Porapak Q column and a thermal

conductivity detector.

3.3.4 Determination of extracellular ﬂuxes. Speciﬁc glucose consumption and product
formation rates (i.e., extracellular ﬂuxes) were calculated as described (48) using the
equation rp = Yxpp, where rp is the speciﬁc product formation rate, Yxp is the amount of
product produced per gram of biomass, and y is the growth rate. The same equation was
used for glucose consumption. All ﬂux data were obtained during exponential growth,

during which the growth rate and extracellular ﬂuxes are constant.

77

3.3.5 GC-MS analysis of amino acids and organic acids. Amino acid preparation and
GC-MS analysis were based on the methods of Schwender et a1. (50) and Dauner and
Sauer (6). All drying and evaporation steps were performed under a stream of N2. Each
cell pellet from 8.4 ml of culture was resuspended in 1 m1 of distilled water and lysed by
six freeze-thaw cycles (—70°C/30°C). The lysates were centrifuged, then 600 pl aliquots
of supernatant were then transferred to 1.5-ml screw-cap conical tubes and concentrated
to 100 pl under vacuum. Proteins were hydrolyzed by adding 333 pl of 6 N HCl to each
sample and incubating at 110°C for 24 h under Ar. HCl was then evaporated to near-
dryness at 55°C, then samples were diluted with 1 ml of 0.01 N HCl, and applied to a
Dowex 50Wx8-100 cation exchange column. Amino acids were eluted with 1 N NH40H
and dried at 55°C. Samples were then dissolved in 0.1 ml of 0.01 N HCl. Amino acid
concentrations were estimated by spotting 10- and 100-fold dilutions of the samples and
an amino acid standard onto silica gel plates and comparing the color intensities after
staining with ninhydrin spray solution and incubating at 80°C for 10 min. Volumes
containing 100—200 pg of amino acids were transferred to glass vials and dried at 25°C.
To derivatize the amino acids for GC-MS analysis, 800 pl of methylene chloride was
added to each sample and evaporated to remove trace amounts of water. Acetonitrile and
N—(tert-butyldimethylsilyl)-N-methyl-triﬂuoroacetamide (MTBSTFA) (200 p1 each) were
then added to the samples under Ar. Vials were capped and incubated at 25°C for 30 min,
then at 120°C for 1 h. Samples were evaporated to near dryness at 25°C before being
diluted with 400 pl of 20:1 acetonitrile: MTBSTFA, and run on GC-MS. Succinate and
fumarate in supematants were derivatized by the same method but without column

puriﬁcation, and concentrations were determined by HPLC.

78

GC-MS was performed using an Agilent 5973 inert MSD benchtop quadrapole
mass spectrometer (Palo Alto, CA) at the MSU Mass Spectrometry facility. Derivatized
compounds in 1-pl injections were separated on a 60-m DBSMS column. The inlet
temperature was 280°C. The oven temperature started at 100°C for 4 min, was then
raised by 5°C/min to 200°C, then by 10°C/min to 300°C, and held at 300°C for 10 min.
Tert-butyldimethylsilyl (TBDMS)-amino acid and organic acids were identiﬁed using

described fragmentation patterns (6) and unlabeled standards.

3.3.6 NMR analysis of organic acids, alanine, and glucose monomers from glycogen.
Prior to NMR analysis, organic acids were separated from glucose and ethanol on a
Dowex 1X8-200 anion exchange column. Organic acids were eluted with 1 M HC104.
The eluate was made alkaline by adding K2C03 until C02 bubbles stopped forming. The
KC104 precipitate was pelleted by centrifugation and the supernatant was lyophilized and
dissolved in D20. Amino acids were obtained from hydrolyzed proteins as described
above but, prior to hydrolysis, the proteins were precipitated with trichloroacetic acid (25
% ﬁnal concentration) on ice for 30 min, pelleted by centrifugation, washed twice with
ice-cold acetone, air-dried, and resuspended in 100 pl water. After hydrolysis, amino
acids were partially puriﬁed, dried, and dissolved in D20. Glucose monomers. were
prepared from glycogen by resuspending the pellets from centrifuged cell lysates in 50
mM sodium acetate (pH 4.5), boiling for 5 min, then digesting with Pseudomonas
amyloderamosa isoamylase (EC. 3.2.1.68) and Aspergillus niger amyloglucosidase (EC.
3.2.1.3) for 24 h at 40°C with occasional mixing. Cell debris was pelleted by
centrifugation, and the supernatant was applied to a Dowex 1X8-200 anion exchange

column. The ﬂow-through was lyophilized and redissolved in D20.

79

All compounds were analyzed by 1H-NMR and 1H-decoupled l3C-NMR using a
Varian VXR 500 MHz spectrometer with a 5-mm l3C-lH switchable probe at the MSU
Max T. Rodgers NMR facility. For 13C-NMR, lH-decoupling was applied only during
acquisition, and recycle times were 60 -120 s to allow for full relaxation. After
determining that there were no statistical differences in the relative lH-NMR peak
intensities between samples originating from different cell concentrations (see Results),
the organic acid samples originating from all 0D660 0.5 and 0D56o 1.0 cultures were
combined into a single sample to decrease l3C-NMR run times. Organic acid samples
originating from the 0D660 1.5 cultures were not pooled for 13C.NMR. All samples were
pooled for glucose and Ala measurements. NMR peak assignments were based on spectra
found at www.aist. go.jp/R10DB/SDBS/ (Japanese National Institute of Advanced

Industrial Science and Technology, Dec 2005) and on spectra of unlabeled standards.

3.3.7 Metabolic modeling and flux analysis. The initial A. succinogenes metabolic
model (Fig.1) is based on A. succinogenes enzyme activities (55), and on sequences in
the A. succinogenes draft genome sequence. The tricarboxylic acid (TCA) cycle was
omitted because it was shown that A. succinogenes cannot synthesize a—ketoglutarate
when grown in AM3, indicating at least two missing TCA cycle enzyme activities (34).
Glucose uptake was not constrained to the conversion of PEP to pyruvate (i.e., by the
PEP: glucose phosphotransferase system [PTS]) to allow for glucokinase activity. Both
PTS and glucokinase activities were detected in A. succinogenes (55), and the PTS does
not appear to be the main glucose uptake mechanism at high glucose concentrations (24).

Pyruvate carboxylating activities were omitted, because no pyruvate carboxylase gene

80

was found in any Pasteurellaceae genome sequence, and A. succinagenes pyruvate
formate-lyase mutants excrete pyruvate (15), suggesting that A. succinogenes cannot
carboxylate pyruvate. Redox balance constraints were omitted to allow for
transhydrogenase activity.

AM3 contains unlabeled Glu, Cys, Met, and vitamins that could potentially enter
amino acid or central metabolic pathways and affect our interpretation of the isotopomer
data set. Glu cannot affect the isotopomer data set for several reasons. First, it is normally
catabolized via a-ketoglutarate, which is disconnected from central metabolism in A.
succinogenes (34). Second, there is only enough Glu supplied to meet the biosynthetic
requirements for Glu and its derivatives, Gln, Pro, Arg, and polyamines (estimated from
Table 1 values). While unlabeled Glu directly affects the isotopomer distribution of Gln,
Pro, and Arg (Glx and Pro are ~100% unlabeled, Table 2), these amino acid isotopomers
are not used in our ﬂux analysis, and not enough Glu is provided to affect other metabolic
pathways. The amount of vitamins supplied is also too low to impact our isotopomer
measurements, representing only 0.28% of the carbon consumed as glucose by the lowest
cell density culture and 0.07% of that consumed by the highest cell density culture. Cys
and Met are supplied at 4- and 5-fold excess, respectively, and together account for 1.3%
of the total supplied carbon. For this reason, we included Cys and Met degradation
pathways in our model based on those found at Metacyc (i.e., Cys 9 pyruvate; Met 9
oxobutanoate 9 Thr 9 Asp 9 0AA; www.metacvc.com).

The model is simpliﬁed for ﬂux analysis by grouping metabolites that are
assumed to be in rapid equilibrium (e. g., pentose phosphates in the pentose phosphate

pathway [PPP]), or that are undistinguishable by labeling (e.g., all metabolites in

81

Table 3.1. A. succinogenes metabolic intermediate and cofactor requirements for
biosynthesis.

a ATP needs for polymerization reactions were assumed to be the same as in E. coli (22),
but adjusted for percentage of dry cell weight.

b Gln and Asn were oxidized to Glu and Asp during hydrolysis and Cys and Trp were
destroyed. Cys and Trp values were based on E. coli values (22). A correction factor
was applied to Tyr to account for average recoveries of 70%.

° The values were derived using an RNA NTP composition based on that of A.
succinogenes ribosomal RNA.

d The values were derived using a DNA dNT P composition based on the GC content of A.
succinogenes (17).

° The phospholipid composition was assumed to be 25 % phosphatidyl glycerol and 75%
phosphatidylethanolamine. The values were derived using the A. succinogenes fatty
acid composition, which was measured to be (in percent of total lipid mass): 14:0, 11;
3-0H-14:0, 3; 16:0, 35; 16:1, 37; 18:0, 1; 18:1, 3; and 18:2, 10.

fThe polyamine requirement was assumed to be the same as for E. coli (22).

g A stoichiometry of 1 ATP per NH3, P042", or amino acid transported was assumed. The
ATP requirement for K+ transport was assumed to be 190 pmol per g biomass'l (52).

h Unaccounted for biomass is assumed to be composed of metabolites and inorganic ions.

Table 1 values were corrected for Glu, Cys, and Met, which are supplied in AM3.

82

 

New; 09:“ 3nd wmm Ohm; om 3N £0 02 c; 2w.» amt? SAN- 9843 Woo A<HOF
<<O <oUu< in Ami 0mm mmO mvm Amy— mcm moo mmz Ian—<2 EQ<Z ACE.

 

as £2. 8:5.
co. b.
we; «.8
23 m2 Hagan:
an a _ a: to maﬁaboa
an R m: mm R as sea 5 nm on
5. mm mm mm mm a. a. m... :w o E: 3 an:
M33 8 8 3 £8 8- $3 3 92a:
8... 8m 3 Escape
2 3 we me. EN 92. 2.3 3. .229
18 2.: m; on? :m an- 33. w: o<zm
E as as 3. _a>
oﬂ ms 2 cm. 3. ﬂ .2 at.
cm 2 cm 8 8 8- a: ed are
an as as Na. 3 E.
:N :m :N :N- N.N am
we 3... we 3 8m
E a: 92 E a: 2 ea
2 .02
SN new man one mam mm «.3
E... E. E 5. E. 3. .3
as SN 9% SN. 2:. E 0:
me as 3 Eu. 3. 2 am
was was was we»- as :0
8a SN 3 .30
ad nmhu
3. v8 ms. 30 on 25......
3. a? 3;- 83 2 wi
:0 E. E. 3 «2
23.8 wen 522a

 

<<o <82 .5 Aim 0.: a8 %m .5. new .60 £2 Eo<z :92 a? as, =8 5.53

 

39:05 w com 2081 be .8 3080.; ﬁn 035.

83

Table 3.2. Mass isotopomer distributions of TBDMS-amino acid and -organic acid
fragments

a Negative values are the result of correcting for natural abundances and can be treated as
having zero value. The distributions shown were corrected as described in the methods
and also for natural l3C abundance in the amino and organic acid backbones using an
excel macro developed by J oerg Schwender (Brookhaven National Laboratory, Upton,
NY) based on correction matrices described by Lee et al. (30). Simulated values
resulting from the optimized set of ﬂuxes (i.e., Figure 3 ﬂuxes) are in parentheses.

b Data for Phe’s highest mass fragments were omitted from the table.

* Values were omitted from the ﬁtting procedure.

84

 

 

 

*md H bd- ...Nd H _d- 88 dd H .d- 8.88 Nd H w.8v 828 vd H :m v; 532
add H dd- *dd H dd 8.8 dd H .d 8.8: Nd H N.8v 8. _ 8 Nd H Edd v.-— m T2 cam
*vd H _.d *md H 8d- 88 dd H bd- 888 dd H _.dm 828 d._ H WE v; 8-2 :58
*md H dd *d.m H Em 8N8 d._ H YNN 8.83 82 H wdv ANNNV dd H ddN m-N 8m 72
.38 H Nd- 1d H dd *Nd H w... 8de _.d H ”N 8.88 dd H :3 2.88 Nd H _.dN W. 8-2 _m>
*Nd H wd 2.8 Nd H d._ 8.83 vd H «:3 8.de md H 8dm v-N mw-2
*dd H dd- *Nd H _d- 8.8 Nd H dd- 8.88 md H 8.; 828 dd H mNm v; 8-2 EH
8.8 _.d H md 8.: dd H N2 8.88 .d H w.w8 N-_ Ndmd
8.8 .d H .2 8.88 dd H ddv 8.28 8d H 8.8-d m-N 82-2
8.8 _d H _.d 283 N. H Ede 8.28 N. H N.mm m-N 3-2
.Ld H mN 2.: _d H .d 883 wd H mdv 8.di vd H ”lam m2 88.-2 com
vd H Nd Nd H Nd Nd H vd dd H Nd _._ H 2.88 m-N 82-2 sci
8.8 _d H _d 8.: _.d H w._ 8.8V md H _.w8 N-_ Ndmd
*Nd H dd- 8.8 d.N H d.—- 8.8 Nd H d.v 2.68 m._ H ~.wN 8d$ _._ H 8.3. 8N8 dd H véN 8-N mw-2
*md H m._ 8.8 _.d H _d 8.8 Nd H wem 8.88 vd H d.dN 8d$ 8d H N.Nv 8N8 vd H w.NN 8-2 8-2 no.2
Nd H vd _d H Nd _d H md- dd H vd dd H v.88 m-N mw-2
md H N._ Nd H dd md H dd Nd H vd- dd H vd 8d H d.h8 m; 8-2 102
*Nd H md *dd H Nd 8.8 _.d H v.2 8&8 md H 3% 2.88 md H Ndm 86$ vd H d.d_ d-N 82-2
1d H _.d *dd H dd 8.8 _.d H dd_ 8.38 vd H 5mm 2.88 vd H mdm 86: dd H 8.: d-N 3-2 so..—
sdd H dd- *dd H .d 2.8 dd H Nd 8de Nd H WMN 8.83 Nd H 8.8» 2.88 md H vNN d-N 82-2
*md H _.d Lad H dd. 2.8 Nd H _.d 868 md H ﬁmN 8.88 vd H _.8v 2.2-Nd dd H wNN d-N 3.2 o:
8.8 _.d H Nd 2.88 md H m.88 N 3.2
8.8 dd H dd 8.: Nd H N._ 8.8V md H :8 N-_ 8-2 .20
.d H _.d- dd H dd _d H md dd H dd Nd H ”88 m-N 82-2
md H md- md H _.d Nd H vd- dd H Nd- Nd H d._d_ m-N 3.2
_.d H .d d._ H Nd v.N H m2- 8d H Nd- dd H _.d _._ H N28 m2 332 .220
8.8 _d H md- 8.8 dd H d.d 8.8V N2 H 5.8 N-_ Ndmd
_. _ .d H dd 2.8 _d H Nd- 88$ md H wdv 8.de dd H 8.Nm v-N 82-2
Ld H _.d 2.8 Nd H _d 868 vd H v.3. 8.de md H mNm v-N mw-2
*dd H Nd Ld H vd 8.8 Nd H dd 8.8; vd H W? 828 nd H v.8 v; 8-2 xm<
8.8 _d H vd 8.2-8 N._ H d.dv 8.de m._ H w.mm m-N 3-2
a 2 .d H _.d 8.8 dd H 2d 8.88 md H wdv 8.de vd H dém m; 8-2 a?
m+x v+x m+x N+X _+X d+x mcoﬁau Eugen“ 8:35
Eons—=88 .2688 do 88 «22.3256 3808902 mma2 29.233 Nd 035—-

 

85

glycolysis between g1yceraldehyde-B-phosphate [G3P] and 3-phosphoglycerate).
Anabolic ﬂuxes were based on the A. succinogenes biomass composition (Table 1). The
standard deviation for all anabolic ﬂuxes was set to 10% of the ﬂux value, which is the
average standard deviation determined from replicate samples for all measured biomass
components, and which takes into account the standard deviation for the dry cell weight
determination. The biosynthetic origin of amino acid carbon atoms from their metabolic
precursors was based on constraints described by Szyperski (53).

Extracellular ﬂuxes, and NMR and GC-MS data sets used were averages of
measurements on all nine cultures with the exception of a few samples that were pooled
for NMR analyses (see N MR methods section). Mass isotopomer standard deviations for
individual amino acid fragments from all nine cultures were typically < 1%. To more
closely reﬂect biological variability, standard deviations for mass isotopomers were
raised to 2.6%. This value was the standard deviation between mass isotopomers that are
expected to be identical (i.e., between Ala, Asp, Ser, and Thr M-57 fragment x+0 mass
isotopomers). The averaged data were applied to several versions of an A. succinogenes
metabolic model (Fig. 1) using 13C-Flux software (62). The model variations were: the
presence and absence of ED and glyoxylate pathways, and alternative PPP reactions (25,
56), as well as grouping or separating 0AA and malate pools. I3C—Flux uses a metabolic
network (speciﬁed by the user) and an arbitrary set of starting values for the free ﬂuxes to
simulate a data set of isotopomers and extracellular ﬂuxes. The simulated data set is then
compared to the experimental data set, and the ﬁt between simulated and experimental
data is determined by a sum of squared residuals (SS-.2) weighted by their respective

standard deviations. The free ﬂux values are then varied until the weighted SSres is

86

minimized, indicating an optimized ﬁt between the two data sets. The choice of free ﬂux
starting values can affect the resulting SSms value if local minima are encountered in the
optimization process. We used thousands of starting values to determine if multiple
solutions existed. This process was automated using a per] script, ‘autoﬂux3.pl,’ written
by Hart Poskar (University of Manitoba, Canada). Natural l3C abundance in the supplied
glucose was included in the model. Mass distribution data were therefore corrected only
for natural isotope abundances in all TBDMS atoms and all amino acid heteroatoms, and
for unlabeled inoculum biomass, using software described by Wahl et al. (58). The
heaviest mass isotopomers were omitted from the ﬁtting process, since their occurrence is
highly improbable when starting from a singly-labeled isotopomer (Table 2) and they are
therefore likely to be overestimated due to contaminants. NMR alanine and glycogen data

were corrected for the inoculum’s unlabeled biomass as described (13).

3.3.8 Analytical techniques for determining A. succinogenes cellular composition.
A. succinogenes was grown as described above and harvested at 0.6—2.1 0D660. Samples
were centrifuged and washed at least once in 0.9% NaCl at 4°C prior to analysis, except
for dry cell weight determination. All sample measurements were compared to the
corresponding 0D660 values by linear regression.

Dry cell weight was determined from thirteen cell suspensions, each composed of
3-5 pooled cultures. Suspensions were ﬁltered through pre-weighed 0.45 pm HA ﬁlters
(Millipore, Billerica, MA) and washed with 10 ml of 0.9% NaCl or water. The biomass-
containing ﬁlters were dried to constant weight at 85°C for 24 h, and the weights

recorded.

87

Total protein and RNA contents were determined from twelve cultures each. For
protein content, cell pellets from l-ml culture samples were resuspended in 0.5 ml water,
lysed by sonication, and lysate protein (20 pl) was quantiﬁed by the bicinchoninic acid
assay (Pierce) with bovine serum albumin as the standard (33). Total cell RNA
quantiﬁcation was based on the method of Benthin et al. (2). Cell pellets from 9-ml
culture samples were resuspended in 3 ml of 0.3 M KOH and digested for 1 h at 37°C,
then chilled on ice. Ice-cold 3 M HC104 (1 ml) was added to the suspension, then KC104
and insoluble cell components were pelleted by centrifugation. The supernatant was
saved, and RNA was extracted from the pellet twice more with 4 ml of ice-cold 0.5 M
HC104. The supematants were pooled and brought up to 13 ml with 0.5 M HC104.
Samples were neutralized with 2 M Tris-HCl (pH 7.4) and the absorbance was measured
at 260 and 280 nm (A260:A230 ratios were between 1.9-2.0). A correspondence of 1.0 A260
to 40 mg RNA/l was used (1). The accuracy of the method was conﬁrmed by performing
the RNA and dry weight determinations on six Escherichia coli cultures grown
aerobically in modiﬁed-M9 without casamino acids (33) and obtaining the expected RNA
content of 20.8% of the dry cell weight (36).

Total cell lipid was determined from six cultures using a transmethylation
protocol (31). Cell pellets were resuspended in 1 ml water and volumes containing ~ 25
mg of biomass were transferred to glass screw cap tubes and dried by lyophilization.
Cells were resuspended in 2 ml of fresh 5% H2S04 in methanol. To this solution were
added 25 pl of 0.2% butylated hydroxy toluene in methanol and 24 pl of internal standard
containing 1.74 mg/ml glyceryl triheptadecanoate in a 2:1 mixture of 2,2,4-

trimethylpentaneztoluene. The reaction mixtures were heated and quenched, and FAMES

88

were extracted and analyzed by GC as described (31). Peak identities were determined
from fatty acid standards GLC-10 and GLC-SO.

Total glycogen was determined from eighteen cultures. Cell pellets from 6—10 ml
of culture were resuspended in 1.4 ml of 50 mM sodium acetate buffer (pH 4.5) and lysed
by sonication, followed by 5 min of boiling. Lysate samples (0.75 ml) were used to
determine the initial glucose concentrations by HPLC. To ensure total glucose release,
6,700 U of isoamylase and 3 U of amyloglucosidase were incubated with the lysates for
20—90 h at 40°C with occasional mixing. At various time points, cell debris was pelleted
by centrifugation and glucose in the supernatant was quantiﬁed by HPLC. Similar
glucose yields were observed in 20 and 90 h digests. Isoamylase was required for
complete digestion during this time, as determined by comparing glucose released from
digests of type H glycogen from oyster with and without isoamylase (data not shown).

Peptidoglycan (PG) and lipopolysaccharide (LPS) levels were estimated from the
difference in the surface areazvolume ratios of A. succinogenes and E. coli, using the
same PGzLPS ratio for A. succinogenes as for E. coli (36). The dimensions of 25 A.
succinogenes and 43 E. coli exponential phase cells were measured by phase contrast
microscopy. Volumes and surface areas were calculated assuming that each cell pole was
a hemi-sphere and that the section between the hemi-spheres was a right cylinder.

To determine A. succinogenes’ amino acid composition, lysates were prepared as
described for total protein determination and centrifuged. Supematants were dried by
lyophilization, washed four times with ice-cold 80% ethanol in water to remove free

amino acids, and dried under vacuum. Proteins were hydrolyzed under vacuum with 6 N

89

HCl at 110°C for 24 h. Amino acid analysis was performed at the MSU Macromolecular
Structure Facility using a Hitachi L-8800 amino acid analyzer (San Jose, CA).

To determine the A. succinogenes C and N contents, cultures were harvested and
washed three times with 0.9% NaCl. Pellets were resuspended in 1 ml water, boiled for
10 min, and dried by lyophilization. The C and N composition was then determined by

the Duke Environmental Stable Isotope Laboratory, Duke University as described (31).

3.3.9 Enzyme assays. Cell extracts were prepared from A. succinogenes grown in AM3.
Cell pellets were washed in 0.9% NaCl, resuspended in 100 mM Tris-HCI (pH 7.4), and
lysed by sonication. Soluble cell extracts were the supernatant of centrifuged lysate while
particulate cell extracts were not centrifuged. Cell extract protein was quantiﬁed using
the bicinchoninic acid assay. All enzyme activities were measured in at least triplicate
using a Cary 300 spectrophotometer (Varian, Palo Alto, CA) in ﬁnal volumes of 1 ml at
37°C. All reactions were started by adding the substrate, except for 0AA decarboxylase,
which was started by adding cell extract after the spontaneous 0AA degradation rate was
measured. Any background activity before adding the substrate was subtracted from the
activity detected after adding the substrate. Extinction coefﬁcients (mM'l cm") used
were: 3-acetylpyridine adenine dinucleotide, 6.1 (57); benzyl viologen (BV), 8.65; 0AA,
0.95; NAD+ and NADP‘”, 6.23.

Formate dehydrogenase was assayed by monitoring BV reduction at 578 nm or
NAD“ reduction at 340 nm in a mixture containing 100 mM Tris-HCl (pH 7.4), 2 mM
BV or NAD”, 5 mM sodium forrnate, and 0.2 pg/ml particulate cell extract protein. The

reaction was made anoxic by ﬂushing solutions with N2 for 10 min, sealing cuvettes with

90

rubber stoppers, and injecting the substrate through the stopper. Malic enzyme was
assayed by monitoring NADP” reduction at 340 nm in a mixture containing 75 mM Tris-
HCl (pH 8.1), 4 mM MnCl2, 1 mM dithiothreitol, 1 mM NADP+, 40 mM disodium
malate, 20 pg/ml soluble cell extract protein, and in the presence or absence of 2 mM
NH4C1. To conﬁrm that NH4C1 did not activate other NADPH-producing activities,
reactions were stopped by cooling on ice, then products were analyzed by HPLC. 0AA
decarboxylase was assayed by monitoring 0AA removal at 265 nm (8) in a mixture
containing 75 mM Tris-HCl (pH 7.4), 1 mM potassium 0AA, 85 pg/ml particulate or 20
pg/ml soluble cell extract protein, in the presence or absence of 15 mM NaCl. 0AA
decarboxylation to pyruvate was conﬁrmed by analyzing the reaction products by HPLC.
Pyruvate dehydrogenase was assayed as described (35) in a mixture containing 750 pg/ml
soluble cell extract protein. Transhydrogenase was assayed as described (44) in a mixture

containing 170 pg/ml particulate cell extract protein.

91

3.4 RESULTS

3.4.1 Determining A. succinogenes’ cellular composition. Knowing the requirements
for metabolic intermediates for biosynthesis is essential for accurately quantifying
metabolic ﬂuxes and for estimating ATP and redox balances. This knowledge can be
obtained by quantifying cell components or by assuming a cell composition similar to
that of E. coli, and relating the values to central metabolism through anabolic pathway
stoichiometries (19, 22, 28, 32, 47). We determined the A. succinogenes glycogen, lipid,
protein, and RNA levels, which together account for ~ 86% of the dry cell weight. At an
0D660 of 1.0 in AM3, A. succinogenes had a concentration of 535 :t 47 mg dry cell wt/l.
A. succinogenes amino acid and fatty acid compositions were also determined (Table 1).
DNA levels were estimated by assuming the same genome copy number for A.
succinogenes and E. coli and adjusting for cells per gram of dry biomass (A.
succinogenes, 3.3 x 1015; E. coli, 1.1 x 1015 [(36)]) and for genome sizes (A.
succinogenes, 2.1 Mb; E. coli K12, 4.6 Mb). LPS and PG were estimated from the
differences in surface area:volume ratios between A. succinogenes and E. coli. With cell
dimensions of 0.63 X 0.63 x 1.21 pm (A. succinogenes) and 0.80 x 0.80 x 2.63 pm (E.
coli), A. succinogenes has a surface area: volume ratio 1.39 times that of E. coli. Table 1
summarizes A. succinogenes’ cell composition and its metabolite and cofactor

requirements.

3.4.2 Confirming a pseudo-metabolic steady state. Since industrial fermentations
typically occur as batch processes, we decided to perform ﬂux analysis with A.
succinogenes under batch conditions. Figure 2, shows batch A. succinogenes growth and

fermentation characteristics in AM3, during which the pH decreases from 7.2 to 6.4.

92

50~ 223.5
45
4.3

402 A
BA E
3235-1 *"2.5g
85 8
avsm D
-s .29
O:
9g254 ’3‘
£520 1.5:
030‘ 0
.DC '0
3015~ _1=
(0° a)

10« o

+0.5
5.-
0- 0

 

 

 

 

0 3 6 9 12 15 18
Time (h)

Figure 3.2. A. succinogenes batch fermentation characteristics in AM3. Data were

obtained from triplicate 60-ml cultures rather than ll-ml cultures to avoid signiﬁcant

volume changes during the course of the fermentation. Data obtained from parallel ll-ml

cultures sampled showed nearly identical fermentation characteristics (data not shown).

Standard deviations are less than or equal to the size of the symbols. Legend: glucose, +;
succinate, e; forrnate, A ; acetate, C1; ethanol, 0; growth, x.

Metabolic ﬂux analysis uses equations that describe a metabolic steady state, where the
ﬂuxes into a metabolite pool equal the ﬂuxes out of that pool. In batch cultures, however,
changing cell, substrate, product, and proton concentrations can potentially affect
metabolic ﬂuxes. To conﬁrm that A. succinogenes ﬂuxes were constant during
exponential growth, we ﬁrst grew ll-ml cultures with unlabeled glucose and compared
extracellular ﬂuxes (speciﬁc rates of glucose consumption and product formations) at
different cell densities in duplicate (i.e., at 0D660 values of 0.5, 1.0, and 1.5) as described
in the methods. The pH at the highest cell concentration sampled had only decreased to
7.0, and fermentation balances are known to be similar at pH values between 6.0 and 7.4
(55). The extracellular ﬂuxes were similar at all cell densities (data not shown), so the

experiment was repeated with nine cultures grown on [1-‘3C] glucose as described in the

93

Methods section. Extracellular ﬂuxes and positional isotopic enrichments were compared
in the cultures harvested at the different cell densities using Student’s t-tests (95 %
conﬁdence, equal-variance). A statistical difference occurred about once in every 19
comparisons, which is indistinguishable from the expected frequency for false positive
occurrences. Based on these results we felt conﬁdent that our batch cultures exhibited a
pseudo-metabolic steady state during exponential growth, and we averaged the data

across all 9 cultures for use in our metabolic ﬂux analysis.

3.4.3 Metabolic pathway delineation and flux quantification. To better understand A.
succinogenes in vivo pathway utilization we used our biomass (Table l), extracellular
ﬂux (Table 3), and isotopomer data sets (Tables 2 and 4) in several variations of the

metabolic model shown in Figure 1. This process was done both manually and by using

 

 

(h 1) mmol g.l h.I Carbon Electron
recove recovery
I“ rGlC rSuc rFum r For r Ace rEtOH ((70) ry (£70)

 

0.28 :t 0.01 7.9 t 0.5 5.8 i 0.3 0.2 t 0.0 3.5 :t 0.3 4.4 :t: 0.2 1.2 :t 0.2 98 :t 4 104 i 4

Table 3.3 Growth and metabolic parameters. The carbon and electron recoveries were
calculated using an assumed elemental composition of CH200.5N0.2. This typical

microbial composition (51) agrees with the C:N ratio for A. succinogenes (Cng).
I3C-Flux software. Where possible, the values used were the average from all nine
cultures, sampled at different cell densities. The set of ﬂuxes giving the best ﬁt between
the simulated and experimental data sets (weighted SSm = 148) is shown in Figure 3.
This model contains 28 free net ﬂuxes and 11 free exchange ﬂuxes (39 degrees of
freedom). With a SS“=3 > 54, our ﬁtting fails the )6 test at the 95% conﬁdence interval, but

this result is observed with most ﬂux analysis studies of biological systems (61).

94

 

l3C enrichment (%) a

 

 

Compound / Carbon IH-NMR I3C-NMR
Acetate / (:1 5.1 :1: 0.5 (4.7)b 4.9 a 0.8 (4.7)
Acetate / cz 42.5 :1: 1.0 (43.1) 42.5 :1: 1.0 (43.1)d
Alanine / c2 4.9 :1: 0.4 (4.7)“c 5.4 :1: 0.7 (4.7)
Alanine/ c3 43.9 :1: 0.9 (43.1)c 43.9 :1: 0.9 (43.1)d
Formate 0.0 :t 0.0 (0.3) ND.
Glucose / CI 97.6 a 3.3 (97.2) 97.6 a 3.3 (97.2)d
Glucose / C6 N.D. 1.3 :t 0.2 (1.4)
Succinate / C1 or C4 N.D. 0.0 :1: 0.0 (0.3)
Succinate / C2 or c3 24.0 :1: 0.2 (23.9) 24.0 a: 0.2 (23.9)d

 

Table 3.4. Measured and simulated percent 13C-enrichments in alanine, glucose, and

organic acids as determined by NMR

3 Simulated values resulting from the optimized set of ﬂuxes (i.e., Figure 3 ﬂuxes) are in
parentheses. ND, not determined.

b Long range coupling between 1H and neighboring l3C (i.e., acetate, 2J'Hm-HC; alanine,
2 J‘HB-BCG)
C While there is considerable overlapping of peaks in the 1H-NMR spectra of amino acids,

the peaks representing the alanine C3 protons are sufﬁciently resolved for accurate
quantiﬁcation.

d Since l3C-NMR only gives relative peak intensities, the lH-NMR—determined valueror
this carbon was used to represent the C-NMR peak area. The label represented by C-
NMR peaks for other carbons was then determined from the relative peak intensity.

N on-oxidative PPP models usually consist of three reactions (pppl-3; Fig. 1).
Additional reactions may exist, though, that can affect the resulting global ﬂux values.
We tested the effects of including additional PPP reactions (56) and of describing all
possible PPP reactions as a simpliﬁed set of half reactions (25). These models gave
nearly identical ﬂuxes and weighted SSms values (152 when including additional
reactions and 154 when using half-reactions) as when only traditional PPP reactions were
included. Because alternative PPP reactions have not been solidly established in vivo, and

because they did not affect the resulting ﬂuxes, we did not use alternative PPP reactions

in our ﬁnal model.

95

Figure 3.3. A. succinogenes metabolic fluxes in AM3 with 150 mM NaHCO3
obtained after fitting of experimental and simulated data sets in 13C-Flux. All flux
values have units of mmol g'1 h". Single parameter 90% conﬁdence intervals were
calculated as described (63). Net ﬂux directions are indicated by an enlarged arrowhead.
Exchange ﬂux values are in parentheses. Exchange ﬂuxes that could not be quantiﬁed are
indicated by a bi-directional arrow without an exchange ﬂux value. Anabolic ﬂuxes are
based on Table 1 values. Anabolic flux requirements for 3-phosphoglycerate, Tyr, and
Trp are accounted for in other anabolic ﬂuxes. The estimated ATP, N ADH, and NADPH
ﬂuxes for biosynthesis and transport were —l3.36, 0.78, and —3.83, respectively.
Reversible citrate lyasc was omitted because there was no glyoxylate shunt ﬂux and

citrate was not observed in culture supematants.

96

Glc

 

 

 

 

 

 

 

 

Figure 3.3
(cont’d) 7.61 :1: 0.29 C02
0.39 :1: 0.10 /
GGP % RSP
710:0.28i
0. 08 :1: 0. 03 MS“;
F6P"—""\/1|
0. 08 a 0. 03
7 19 :1: 0. 28
E4P (\GIBP R5P
0.04 :1: 0.03
G3P + G3P< /
14.00 1 0.56
(27.6 :1: 16.7)
11.40 :1: 2. 04
(48:23) 2.511199
CAN 4. 90i 1 2. 01
P r
Mal V 3.49 a 0.49
6.08 a 0.43 Sco,
(4.411.” 3.14:0.65 For
Fum AcCoA
1..24:l:020 4..52:1:038
5.93.843 CO/ \
Sue EtOH Ace
Additional net ﬂuxes.
ATP NADH CO2 uptake Furn production
8.23 3.43 2.47 :1: 0.92 0.15 :1: 0.02
Net Anabolic ﬂuxes.
G6P F6P asp E4P GBP PEP Pyr AcCoA OAA
0.12 0.04 0.19 0.04 0.42 0.10 0.78 0.86 0.42

 

97

Our focus was to identify pathways involved in distributing flux to succinate and
alternative fermentation products as well as those involved in redox balance. For this
reason, and because the ATP required for cell maintenance is unknown, we did not focus
on why the net ATP production is 1.6 times greater than the estimated requirement.
Figure 3 is described in detail below, focusing on the pathways of interest, and explaining

the progression from the metabolic model shown in Figure 1 to that shown in Figure 3.

3.4.4 The glyoxylate cycle. The glyoxylate cycle could act as a shunt between the C3 and
C4 pathways. Because low in vitro isocitrate lyasc activity was reported for A.
succinogenes (55), we investigated the presence of a functioning glyoxylate cycle in our
l3C-labeling experiment. Glyoxylate shunt ﬂux was ﬁrst addressed by inspection of the
isotopomer data. As seen in Figure 4, the [3-13C]OAA and [2-13C]acetyl-CoA produced
from [1—13C] glucose would generate double-labeled OAA through an active glyoxylate
cycle. Table 2 shows insigniﬁcant fractions of double-labeled OAA-derived products
(i.e., Asp, Thr, Fum, and Suc). These results suggest that the glyoxylate cycle, if present
in A. succinogenes, is not active in cultures grown in AM3. Conﬁrming our manual
inspection of the data, all ﬁttings of simulated to experimental data in 13C-Flux indicated
no forward or reverse glyoxylate cycle ﬂux. These results agree with the fact that no
Pasteurellaceae genome sequenced to date (complete or incomplete) contains genes

showing similarity to known malate synthase and isocitrate lyase genes.

98

PEP

E—>”" MEI-M11121.

39°\ VF"

OAA/Mal OAA/Mal AcCoA

EEEE BEBE EH

Glx AcCoA

Ell Ell e1.
EEEBE

 
   

Suc

EEEE
EEEE

Figure 3. 4. AcetyI-CoA an3d OAA isotopomers expected from forward glyoxylate
cycle ﬂux. Black squares, l3;C white squares, 2C. The shaded oval highlights the double—
labeled OAA that would be generated by glyoxylate cycle ﬂux.

3.4.5 The oxidative pentose phosphate pathway. Since succinate production from PEP
requires reductant (Fig. 1), it is important to understand A. succinogenes’ redox balance.
The OPPP was thought to be important for NADPH production based on previous
enzyme activity data (55). Also, the A. succinogenes genome sequence encodes the OPPP
enzymes, glucose-6-phosphate (G6P) dehydrogenase (AsucDRAFI‘_0146) and 6-
phosphogluconate dehydrogenase (AsucDRAFI‘_014l). We ﬁrst addressed OPPP ﬂux by
inspecting the isotopomer data. While glycolysis converts [l-l3C]G6P into equal
proportions of unlabeled and single-labeled G3P, the OPPP produces 13C02 and

unlabeled G3P (Fig. Sa). Between 45.5% (in Ser) and 50.1% (in Fum) of each

downstream product of G3P (i.e., Ala, Asp, Fum, Ser, Suc, and Thr) contain a single

99

label (Table 2). These values are close to the 50% expected from glycolytic ﬂux alone.
However, to accurately quantify the glycolytic and OPPP ﬂuxes, the existence of [6-
l3C]G6P should be considered. [6-‘3C]G6P is produced by reverse ﬂux from [6—
l3C]fructose-6-phosphate, which is generated by triose-phosphate cycling (50) and
reverse PPP reactions. NMR analysis of glycogen’s glucose monomers shows that only
1.4% of G6P is labeled at carbon 6 (Table 4), indicating very little OPPP ﬂux.

We then used 13C-Flux to quantify the OPPP ﬂux. Since there is a net C02
consumption by A. succinogenes fermentations, we originally only included a C02 inﬂux
in our model. This ﬁrst model did not account for the possibility of 13C02 produced by
the OPPP being diluted by the large amount of 12C02 in the medium. Without a C02
efﬂux, all ﬁttings indicated no OPPP ﬂux. Since most metabolites, with the exception of
fumarate and succinate, indicate a small 13C loss from the system (Tables 2 and 4), we
included a C02 efﬂux in the model. The OPPP ﬂux was 5% that of the glucose uptake
rate, only enough to supply 20% of the estimated NADPH requirements (Fig. 3).

Like OPPP ﬂux, Cys and Met uptake would contribute to unlabeled isotopomers.
However, the ﬁtting shown in Figure 3 indicates no Cys and Met uptake. To determine if
our adjustment of the mass isotopomer standard deviations (see Methods) prevented the
detection of Cys and Met uptake, we used unmodiﬁed standard deviations in the ﬁtting
process. The best ﬁt in this case (SS.cs of 384) showed essentially the same set of ﬂuxes,
including no Met uptake and a very small Cys uptake (1% of the glucose uptake rate). As
expected, OPPP ﬂux was lower (4% of the glucose uptake rate) in the ﬁt showing Cys

uptake. If our model underestimates the extent of Cys and Met uptakes, then it

100

A.

[1 -13C]GGP OPP C02
IEEEEE

l EMP magma
G3P 1

EE. 639
[Ill EEE

B.

[1 -‘3C]G|c

IEEEEE

GSP

IEEEEE

Pyr (Ala) Gap

EEE EEE

For AcCoA

ll EE

Figure 3.5. Isotopomers expected from OPPP and ED pathway fluxes Black squares,
l3C; white squares, 12C. A. G3P isotopomers expected from forward glycolytic and OPPP
ﬂuxes. B. Alanine and formate isotopomers expected from ED pathway ﬂux.

101

overestimates the OPPP ﬂux as well. Thus, 20% represents the upper limit of the

NADPH requirements that can be supplied by the OPPP.

3.4.6 The Entner-Doudoroff pathway. With low OPPP ﬂux, other pathways must be
responsible for N ADPH production. Previously, low ED pathway enzyme activities were
detected in A. succinogenes extracts (55). The A. succinogenes genome encodes two
proteins showing 36% and 37% identity to E. coli 2-keto-3-deoxy-6-phosphogluconate
aldolase (AsucDRAFI‘_0727 and 1070) and a protein that is 27% identical to E. coli
phosphogluconate dehydratase (AsucDRAFI‘_0513). For these reasons, we investigated
the presence of ED pathway ﬂux in our labeling experiment. Flux through the ED
pathway would produce [1-13C]pyruvate, resulting in [1-13C]Ala and l3C-formate (Fig
5b). As seen in Table 2, Ala M-57 fragments (containing carbons 1-3) and M-85
fragments (containing carbons 2 and 3) have almost identical mass isotopomer
distributions. Since losing Ala carbon 1 does not affect the mass distribution, carbon 1
must be 100% unlabeled. Furthermore, no labeled forrnate was observed by lH-NMR
(Table 4). These results indicate that there is negligible ED pathway ﬂux. All I3C-Flux

ﬁttings using models containing the ED pathway conﬁrmed this conclusion.

3.4.7 C3 pathway dehydrogenases and transhydrogenase. HPLC analysis of excreted
products showed that 1.6 times more acetate + ethanol were excreted than forrnate. The
expected ratio is < 1.0 since pyruvate forrnate-lyasc produces equal amounts of forrnate
and acetyl-CoA (Fig. 1) and some acetyl-CoA is required for biomass (Fig. 3). This

observation can be explained if formate dehydrogenase (ForDH) consumes some forrnate

102

or pyruvate dehydrogenase (PyrDH) converts some pyruvate to acetyl—CoA and C02. The
A. succinogenes genome encodes both PyrDH (AsucDRAFI‘_001-003) and ForDH

(AsucDRAFI‘_l405-1408), and in vitro activity was detected for both enzymes (Table 5).

 

 

Enzyme assayed Speciﬁc activity (nmol mg protein'1 min")

Transhydrogenase 14 i 1
Pyruvate dehydrogenase 5 :1: 1
Formate dehydrogenasea 7 :1: 2
Malic enzyme (- NI-IJ)b 120 :1: 10
Malic enzyme (+ NHJ)b 620 :1: 50
Soluble 0AA decarboxylase (- Na+)c 260 t 60
Soluble 0AA decarboxylase (+ Na+)c 80 t 1 10
Particulate 0AA decarboxylase (+ NaJ')c 1,080 i 1 10

 

Table 3.5. Enzyme activities in cell extracts of A. succinogenes grown in AM3 with
150 mM NaHCO3

8 Activity was not detected when NAD+ was supplied in place of BV.

b Since NH: is known to stimulate malic enzyme activity (43), NHaCl was added to the
reaction to conﬁrm that the measured activity was due to malic enzyme. The amount of
NADPH produced was 94 :1: 3 % that of the pyruvate produced.

° HPLC analysis conﬁrmed that pyruvate was the main product formed.

ForDH is often coupled to hydrogenase in bacterial fermentations, producing H2 rather

than NADH. Here, the electron balance was 104 :1: 4% (Table 3) and GC analysis showed

that H2 was absent from culture headspaces, indicating either that H2 was not produced or
that it was quickly consumed. In any case, pyruvate oxidation to acetyl-CoA without

forrnate or H2 production yields reductant. Our metabolic ﬂux model (Fig. 3) indicates a

3.14 mmol g"1 h'1 ﬂux through the PyrDH/ForDH grouped reaction, giving a total N ADH

production rate of 17.92 mmol g‘1 h"1 (including NADH produced by G3P dehydrogenase

and anabolism). Succinate and ethanol productions require a total of 14.49 mmol NADH

g“1 h'1 (Fig. 3), leaving 3.43 mmol NADH g"1 h’1 to be consumed by other reactions. One

possibility for an NADH-consuming reaction is transhydrogenase. The A. succinogenes

103

genome encodes a membrane-associated transhydrogenase (AsucDRAFI‘_0847-848). In
vitro transhydrogenase activity was previously reported as diaphorase activity (37) and it
was also detected in this study (Table 5). If the excess NADH is consumed entirely by

transhydrogenase, it could supply 90% of the estimated NADPH requirements.

3.4.8 Malic enzyme and OAA decarboxylase. While transhydrogenase and OPPP
ﬂuxes could account for up to 110% of the estimated NADPH requirements, other
NADH-consuming (i.e., malate dehydrogenase) and NADPH-producing (i.e., malic
enzyme) reactions are possible. Together, the reactions catalyzed by malate
dehydrogenase and malic enzyme are an alternative to transhydrogenase, resulting in the
net oxidization of NADH and reduction of NADP'( (Fig. 1). Besides producing NADPH,
malic enzyme also bridges the C4 and the C3 pathways (Fig. 1). Therefore, the existence
of a malic enzyme ﬂux could dramatically inﬂuence metabolic engineering strategies.
Malic enzyme activity was detected in extracts of A. succinogenes grown in several
conditions (55), including in AM3 (Table 5). To distinguish ﬂuxes between PEP and
pyruvate from ﬂuxes between OAA/malate and pyruvate, a unique isotopomer of
pyruvate or OAA/malate must be present. As illustrated in Figure 6, a unique isotopomer,
[2-‘3C]fumarate, is formed when scrambling caused by the symmetry of fumarate
generates 25% each of [3-‘3C]fumarate and [2—‘3C]fumarate (50% of the fumarate is
unlabeled). Exchange ﬂux in the C4 pathway then generates 7% [2-‘3C]0AA and 2% [2-
l3C]PEP (determined from GC-MS analyses of Asp’s and Phe’s f302 fragments,
respectively; Table 2). If all pyruvate were generated from PEP, as would be expected in

a simple branched fermentation, pyruvate and PEP would have identical positional

104

co2 PEP

AEtyEEI\
OAA/Mal PVT
EEIE
F£m For AcCoA
EEIE '3 El-
EIIIEHZI Al...
EH
602 2 % PEP
B El EIE
V \
7 % OAA/Mal 5% Pyr

EIEE T» EIE

I .9 A

25 % Fum 2 For AcCoA
EIEE E IE
5%‘Ace
IE

Figure 3.6. Two-step illustration of the ﬂuxes resulting in [2-13C]pyruvate. Black

squares, l3C; white squares, 12C. A. Step 1: Generation of [2-13C]fumarate (highlighted by

the shaded oval) by forward ﬂuxes through the C4 pathway and fumarate’s symmetry. B.

Step 2: Isotopomers resulting from [2-13C]fumarate by reverse C4 pathway ﬂux, C4-

decarboxylating ﬂux, and forward C3 pathway ﬂux.

105

isotopic enrichments. NMR analyses of Ala (Table 4) show that 5%, rather than 2%, of
pyruvate is labeled at the C2 position. This result indicates that a C4-decarboxylating
enzyme (i.e., malic enzyme or 0AA decarboxylase) is active during growth in AM3. This
conclusion is also supported by the detection of 5% [1-‘3C]acetate (Table 4), a
downstream product of [2-13C]pyruvate.

Because ﬂuxes from 0AA, malate, and PEP to pyruvate give identical pyruvate
isotopomers, and because different combinations of these three ﬂuxes can result in
identical pyruvate isotopomer distributions, 0AA and malate decarboxylating ﬂuxes
cannot be distinguished. For this reason, malate and 0AA were grouped into a single
pool, as in other metabolic ﬂux models (3, 32, 40, 59, 66). In Figure 3, PEPCK and PK
net ﬂux values were dependent on the OAAdec/ME ﬂux (set as a free variable),
explaining why PEPCK, PK, and OAAdec/ME ﬂuxes have similar conﬁdence intervals.
These conﬁdence intervals are large, and different values for ﬂuxes between PEP,
OAA/malate, and pyruvate were obtained in multiple ﬁttings with little effect on the SSms
value. However, our ﬁttings consistently showed that the OAAdec/ME ﬂux was larger
than the PK ﬂux. The best ﬁt indicates that the C4-decarboxylating ﬂux is 4.9 mmol g'1 h'
1, which is greater than the ﬂux needed to convert excess NADH to NADPH (3.43 mmol
g'l h'l). Therefore, the C4-decarboxylating flux must represent a combination of malic
enzyme and OAA-decarboxylatin g ﬂuxes, since 0AA decarboxylation to pyruvate does
not yield reductant. An operon encoding a membrane-bound, Na+-pumping 0AA
decarboxylase was identiﬁed in the genome sequence (AsucDRAFF_1557-1559). In
addition to detecting a Na+-independent 0AA decarboxylase activity in soluble cell

extracts, we also detected signiﬁcant Na+-dependent activity in particulate cell extracts

106

but not in soluble cell extracts (Table 5). From the ﬂux analysis and enzyme activities
detected, it is clear that ﬂux distribution to succinate and alternative products can occur at

0AA and malate, rather than just at PEP.

107

3.5 DISCUSSION

This study has described important aspects of A. succinogenes metabolism. Flux
through the glyoxylate and ED pathways was ruled out. OPPP ﬂux was too low to meet
NADPH requirements. PyrDH and/or ForDH contribute to a net NADH production that
could be converted to NADPH by transhydrogenase. Malic enzyme may also contribute
to NADPH production. In addition to PEP, OAA and malate are important nodes for ﬂux
distribution to succinate and alternative fermentation products.

We assume that the excess ATP not needed for biosynthesis and transport (Table
1) is used for cell maintenance. The estimated ATP required for biosynthesis alone (10.48
mmol g'1 h!) is 49 % of the total ATP produced, which is not unreasonable. It was
estimated that 51% of the energy produced by anaerobically growing E. coli is used for
transport and maintenance processes (52).

Optimized ﬁts showed a greater OAAdec/ME ﬂux than PK ﬂux. PK ﬂux is
composed of pyruvate kinase and the PTS. If glucose phosphorylation occurred through
the PTS alone, the PK ﬂux would be at least equal to the glucose uptake rate. Because the
estimated PK ﬂux is only 33% of the glucose uptake rate, and because most of the ﬂux
into the C3 pathway is likely from OAA/malate rather than from PEP (Fig. 3), it is
unlikely that the PTS is the main glucose phosphorylation mechanism. A. succinogenes
was previously shown to have hexokinase activity in addition to PTS activity (55).
Uptake assays with MC-glucose and a l4C-glucose analogue indicated that the PTS is not
the main A. succinogenes glucose uptake mechanism at high glucose concentrations (24).

This study identiﬁed the mechanisms responsible for NADPH production in

glucose-grown A. succinogenes in AM3. Product isotopomer data indicated no ED

108

pathway ﬂux and low OPPP ﬂux. The lower ﬂux to formate than to acetyl-CoA
indicated, in addition to pyruvate forrnate-lyasc activity, a pyruvate oxidizing activity
(i.e., PyrDH) and/or a forrnate oxidizing activity (i.e., ForDH). PyrDH is often considered
to be an enzyme of aerobic metabolism. However, it is expressed in anaerobically-grown
Haemophilus inﬂuenzae (41), a bacterium closely related to A. succinogenes. PyrDH and
ForDH activities were also detected in A. succinogenes cell extracts (Table 5). In either
case, there is enough excess NADH to meet the NADPH requirements through
transhydrogenase and/or through NADH-oxidizing malate dehydrogenase plus NADP—
reducing malic enzyme. Product isotopomer data conﬁrmed the existence of a C4-
decarboxylating ﬂux, suggesting that malic enzyme contributes to NADPH production.
NADP-dependent malic enzyme and transhydrogenase activities were detected in cell
extracts (Table 5). Given the low OPPP ﬂux, transhydrogenase and/or malic enzyme are
the main N ADPH-producing mechanisms, and account for at least 80% of the NADPH
requirements.

Malic enzyme’s contribution to NADPH production in A. succinogenes and other
bacteria is currently unclear. Under ammonia-limiting conditions, or when pyruvate
kinase is knocked out, malic enzyme ﬂux increases in E. coli, butthis results in excess
NADPH that is oxidized by transhydrogenase (11). Transhydrogenase, on the other hand,
is a signiﬁcant NADPH producer in several cases. In fact, E. coli was recently shown to
rely on transhydrogenase for up to 45% of its NADPH requirements in aerobic batch
cultures (44). Transhydrogenase is even more important for E. coli NADPH production
when the OPPP enzyme, G6P dehydrogenase, is disrupted (21, 44). Transhydrogenase is

also important for NADPH production in Rhodospirillum rubrum and Rhodobacter

109

sphaeroides (14, 20), suggesting an important role for this enzyme across a diverse range
of bacteria. While we have shown that transhydrogenase and/or malic enzyme are
important for NADPH production in A. succinogenes, it remains to be seen how ﬂux
through these enzymes and through the OPPP respond to environmental stimuli.

Understanding the N ADPH-producing mechanisms in A. succinogenes may also
be important for devising metabolic engineering strategies. It is well known that the CO2
concentration affects the ﬂux distribution into the C4 and C3 pathways (34, 55). Here we
have shown that ﬂux to the C3 pathway (i.e., malic enzyme) and ﬂux through the C3
pathway (i.e., PyrDH and ForDH) are important for NADPH production. Therefore, it is
plausible that the need for NADPH also controls ﬂux distribution between the C4 and C3
pathways. Under the conditions studied here, a limited quantity of reductant must be
shared by the C4 pathway and anabolism. Adding H2 shifts the A. succinogenes
fermentation balance toward succinate in rich media (55). Adding H2 also, at least
partially, alleviates the need for ﬂux through C3 pathway dehydrogenases, as shown by
the fact that adding H2 increases the forrnate:(acetate + ethanol) ratio in AM3-grown
cultures (McKinlay eta1., unpublished data). It will be important to determine whether a
surplus of electrons (e. g., as H2) will be sufﬁcient to suppress the C3 pathway or if
genetic modiﬁcations are necessary.

The discovery of an in vivo C4-decarboxylating ﬂux is particularly important
because it reveals that ﬂux distribution to succinate and alternative products can occur at
multiple nodes (i.e., PEP, OAA, malate, and possibly pyruvate). This discovery also
implicates PEPCK as a major ﬂux-carrying enzyme that feeds both product-formin g

branches rather than just the C4 pathway. Thus, at least under the growth conditions

110

tested here, OAA and malate could be more important nodes than PEP in controlling ﬂux
distribution to succinate and alternative products. There is currently much interest in the
ﬁelds of bacterial physiology and metabolic engineering in understanding ﬂux
distribution at these nodes (45). In the case of A. succinogenes, it now seems likely that
redirecting ﬂux towards succinate production will face the engineering challenges of
dealing with alternative routes (e.g., pyruvate kinase vs. PEPCK and a C4-decarboxylase)
and redundant activities. In this study, malic enzyme and two OAA-decarboxylating
activities were identiﬁed. The Na+-dependent OAA decarboxylase activity was only
observed in particulate cell extracts, which is consistent with the Na+-pumping,
membrane-bound OAA decarboxylase encoded in the A. succinogenes genome.
Membrane-bound OAA decarboxylase is also expressed by closely related Mannheimia
succiniciproducens grown anaerobically on glucose (29). While this enzyme is required
for growth on citrate by other bacteria (5, 9, 67), its activity during glucose fermentation
is an intriguing ﬁnding that suggests an alternative role. A role for OAA decarboxylase in
N a+-linked energetics has been suggested for several bacteria (10, 18), including in Na+-
dependent glutamate transport and drug efﬂux in closely related A.
actinomycetemcomitans (18). A Na+ gradient could also be involved in fumarate and
succinate transport as was shown for Wolinella succinogenes (54). However, given that
AM3 contains ~230 mM Na“, the membrane-bound OAA decarboxylase may simply
function to pump N a+ out of the cell. A N a+-independent OAA-decarboxylating activity
was detected in the soluble crude extract (Table 5). Malic enzyme (43), PEPCK (23),
pyruvate kinase (4), and 2-keto-3-deoxy-6-phosphogluconate aldolase (39) can also

decarboxylate OAA, at least in vitro. While the in vivo contribution of these OAA

111

decarboxylating activities is unknown, the potential for redundant activities is
considerable, and may present a challenge for metabolic engineering. On the other hand,
a C4—decarboxylatin g ﬂux that only meets anabolic pyruvate and acetyl-CoA needs could
be beneﬁcial for developing a succinate-producing strain with an acceptable growth rate.
This study has provided key insights into A. succinogenes NADPH production
and ﬂux distribution to major fermentation products, and has raised important
considerations for its metabolic engineering. l3C—labeling studies of industrial organisms
have not only led to strain improvements but also to a deeper understanding of
metabolism. They have led to the discovery of new pathways (3, 49), and to a better
understanding of anaplerosis (38, 40, 45) and of the controls effected by cofactor
production and cell energetics (7, 46, 68). The effects of CO2 and alternative carbon and
electron sources on A. succinogenes’ fermentation balance make this bacterium an
excellent system for analyzing the metabolic control effects of CO2 and reducing power
on ﬂux distribution between PEP, pyruvate, OAA, and malate using l3C-labeling and
other studies. While A. succinogenes has industrial potential, it also has unique metabolic
traits whose study could contribute to a broader understanding of bacterial metabolism

and its diversity.

112

3.6 ACKNOWLEDGEMENTS

This work was supported by the National Science Foundation grant BBS-0224596
and by a grant from the Michigan State University (MSU) Research Excellence Fund.
James McKinlay was supported in part by MSU Quantitative Biology and Modeling
Initiative and Marvin Hensley Endowed fellowships and by an Annie’s Homegrown
fellowship for environmental studies.

We are indebted to Dr. Douglas Allen for valuable discussions and technical
assistance. We also thank Drs. John Ohlrogge, Ana Alonso, Fernando Goffman, Igor
Libourel, and J oerg Schwender for valuable discussions. We are grateful for the use of
Drs. Tom Schmidt’s and John Ohlrogge’s GC equipment. We are also grateful for the use
of the autoﬂux3 program written by Dr. Hart Poskar. We thank Drs. Beverly Chamberlin,
Daniel Holmes, and Joe Leykam for assistance with GC-MS, NMR, and amino acid

analysis, respectively.

113

3.7 REFERENCES

1.

10.

11.

12.

13.

Ausubel, F. M., R. Brent, R. E. Kingston, D. D. Moore, J. G. Seidman, J. A. Smith,
and K. Struhl. (ed.). 1998. Current Protocols in Molecular Biology. John Wiley and
Sons, Inc.

Benthin, S., J. Nielsen, and J. Villadsen. 1991. A simple and reliable method for the
determination of cellular RNA content. Biotechnol. Tech. 5:39—42.

Christensen, B., and J. Nielsen. 2000. Metabolic network analysis of Penicillium
chrysogenum using l3C-labeled glucose. Biotechnol. Bioeng. 68:652-659.

Creighton, D. J ., and I. A. Rose. 1976. Oxalacetate decarboxylase activity in muscle
is due to pyruvate kinase. J. Biol. Chem. 251:69-72.

Dahinden, P., Y. Auchli, T. Granjon, M. Taralczak, M. Wild, and P. Dimroth. 2005.
Oxaloacetate decarboxylase of Vibrio cholerae: puriﬁcation, characterization, and
expression of the genes in Escherichia coli. Arch. Microbiol. 183: 121-129.

Dauner, M., and U. Sauer. 2000. GC-MS analysis of amino acids rapidly provides
rich information for isotopomer balancing. Biotechnol. Prog. 16:642-649.

Dauner, M., M. Sonderegger, M. Hochuli, T. Szyperski, K. Wuthrich, H. P.
Hohmann, U. Sauer, and J. E. Bailey. 2002. Intracellular carbon ﬂuxes in riboﬂavin-

producing Bacillus subtilis during growth on two-carbon substrate mixtures. Appl.
Environ. Microbiol. 68: 1760-1771.

Dimroth, P. 1981. Characterization of a membrane-bound biotin-containing enzyme:
oxaloacetate decarboxylase from Klebsiella aerogenes. Eur. J. Biochem. 1151353-
358.

Dimroth, P. 1980. A new sodium-transport system energized by the decarboxylation
of oxaloacetate. FEBS Lett. 122:234-236.

Dimroth, P., and B. Schink. 1998. Energy conservation in the decarboxylation of
dicarboxylic acids by fermenting bacteria. Arch. Microbiol. 170:69-77.

Emmerling, M., M. Dauner, A. Ponti, J. Fiaux, M. Hochuli, T. Szyperski, K.
Wuthrich, J.E. Bailey, and U. Sauer. 2002. Metabolic ﬂux responses to pyruvate
kinase knockout in Escherichia coli. J. Bacteriol. 184:152-164.

Fell, D. 1997. Understanding the control of metabolism. Portland Press, London and
Miami.

Fischer, E., and U. Sauer. 2003. Metabolic ﬂux proﬁling of Escherichia coli mutants
in central carbon metabolism using GC-MS. Eur. J. Biochem. 270:880-891.

114

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

Grammel, H., E. D. Gilles, and R. Ghosh. 2003. Microaerophilic cooperation of
reductive and oxidative pathways allows maximal photosynthetic membrane
biosynthesis in Rhodospirillum rubrum. Appl. Environ. Microbiol. 69:6577-6586.

Guettler, M. V., M. K. Jain, and D. Rumler. 1996. Method for making succinic acid,
bacteria] variants for use in the process, and methods for obtaining variants. US.
patent 5,573,931.

Guettler, M. V., M. K. Jain, and B. K. Soni. 1996. Process for making succinic acid,
microorganisms for use in the process and methods of obtaining the microorganisms.
US. patent 5,504,004.

Guettler, M. V., D. Rumler, and M. K. Jain. 1999. Actinobacillus succinogenes sp.
nov., a novel succinic-acid-producing strain from the bovine rumen. Int. J. Syst.
Bacteriol. 49:207-216.

Hase, C. C., N. D. Fedorova, M. Y. Galperin, and P. A. Dibrov. 2001. Sodium ion
cycle in bacterial pathogens: evidence from cross-genome comparisons. Microbiol.
Mol. Biol. Rev. 65:353-370.

Henriksen, C. M., L. H. Christensen, and J. Nielsen. 1996. Growth energetics and

metabolic ﬂuxes in continuous cultures of Penicillium chrysogenum. J. Biotechnol.
45: 149-164.

Hickman, J. W., R. D. Barber, E. P. Skaar, and T. J. Donohue. 2002. Link between
the membrane-bound pyridine nucleotide transhydrogenase and glutathione-
dependent processes in Rhodobacter sphaeroides. J. Bacteriol. 184:400-409.

Hua, Q., C. Yang, T. Baba, H. Mori, and K. Shimizu. 2003. Responses of the central
metabolism in Escherichia coli to phosphoglucose isomerase and glucose-6-
phosphate dehydrogenase knockouts. J. Bacteriol. 185:7053-7067.

Ingraham, J. L., O. Maaloe, and F. C. Neidhardt. 1983. Growth of the bacterial cell.
Sinauer Associates, Sunderland, MA.

J abalquinto, A. M., M. Laivenieks, J. G. Zeikus, and E. Cardemil. 1999.
Characterization of the oxaloacetate decarboxylase and pyruvate kinase-like activities
of Saccharomyces cerevisiae and Anaerobiospirillum succiniciproducens
phosphoenolpyruvate carboxykinases. J. Protein. Chem. 18:659-664.

Kim, P., and C. Vieille. Glucose phosphorylation mechanisms in succinate-producing
Actinobacillus succinogenes. submitted.

115

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

Klei'n, R. J., W. A. van Winden, W. M. van Gulik, and J. J. Heijnen. 2005. Revisiting
the 3C-label distribution of the non-oxidative branch of the pentose phosphate
pathway based upon kinetic and genetic evidence. FEBS J. 272:4970—4982.

Koffas, M. A. G., G. Y. Jung, and G. Stephanopoulos. 2003. Engineering metabolism
and product formation in Corynebacterium glutamicum by coordinated gene
overexpression. Metab. Eng. 5:32-41.

Kroger, A., S. Biel, J. Simon, R. Gross, G. Unden, and C. R. D. Lancaster. 2002.
Fumarate respiration of Wolinella succinogenes: enzymology, energetics and
coupling mechanism. Biochim. Biophys. Acta. 1553:23-28.

Lange, H. C., and J. J. Heijnen. 2001. Statistical reconciliation of the elemental and

molecular biomass composition of Saccharomyces cerevisiae. Biotechnol. Bioeng.
75:334-344.

Lee, J. W., S. Y. Lee, H. Song, and J. S. Yoo. 2006. The proteome of Mannheimia
succiniciproducens, a capnophilic rumen bacterium. Proteomics 6:3550-3566.

Lee, W. N., L. O. Byerley, E. A. Bergner, and J. Edmond. 1991. Mass isotopomer
analysis: theoretical and practical considerations. Biol. Mass. Spectrom. 20:451-458.

Li, Y., F. Beisson, M. Pollard, and J. Ohlrogge. 2006. Oil content of Arabidopsis
seeds: the inﬂuence of seed anatomy, light and plant-to-plant variation.
Phytochemistry 67:904-9 15.

Marx, A., A. A. de Graaf, W. Wiechert, L. Eggeling, and H. Sahm. 1996.
Determination of ﬂuxes in the central metabolism of Corynebacterium glutamicum by

nuclear magnetic resonance spectroscopy combined with metabolic balancing.
Biotechnol. Bioeng. 49:] 11-129.

McKinlay, J. B., and J. G. Zeikus. 2004. Extracellular iron reduction is mediated in

part by neutral red and hydrogenase in Escherichia coli. Appl. Environ. Microbiol.
70:3467-3474.

McKinlay, J. B., J. G. Zeikus, and C. Vieille. 2005. Insights into Actinobacillus
succinogenes fermentative metabolism in a chemically deﬁned growth medium.
Appl. Environ. Microbiol. 71 :6651-6656.

Millar, A. H., C. Knorpp, C. J. Leaver, and S. A. Hill. 1998. Plant mitochondrial
pyruvate dehydrogenase complex: puriﬁcation and identiﬁcation of catalytic

components in potato. Biochem. J. 334:571-576.

Neidhardt, F. C. 1987. Escherichia coli and Salmonella typhimurium cellular and
molecular biology, vol. 1. American Society for Microbiology, Washington, DC.

116

37.

38.

39.

40.

41.

42.

43.

45.

46.

47.

48.

Park, D. H., and J. G. Zeikus. 1999. Utilization of electrically reduced neutral red by
Actinobacillus succinogenes: physiological function of neutral red in membrane-
driven fumarate reduction and energy conservation. J. Bacteriol. 181:2403-2410.

Park, S. M., C. Shaw-Reid, A. J. Sinskey, and G. Stephanopoulos. 1997. Elucidation
of anaplerotic pathways in Corynebacterium glutamicum via l3C-NMR spectroscopy
and GC-MS. Appl. Microbiol. Biotechnol. 47:430-440.

Patil, R. V., and E. E. Dekker. 1992. Cloning, nucleotide sequence, overexpression,
and inactivation of the Escherichia coli 2-keto-4-hydroxyglutarate aldolase gene. J.
Bacteriol. 174:102-107.

Petersen, S., A. A. de Graaf, L. Eggeling, M. Mollney, W. Wiechert, and H. Sahm.
2000. In vivo quantiﬁcation of parallel and bidirectional ﬂuxes in the anaplerosis of
Corynebacterium glutamicum. J. Biol. Chem. 275:35932-35941.

Raghunathan, A., N. , D. Price, M. Y. Galperin, K. S. Makarova, S. Purvine, A. F.
Picone, T. Chemy, T. Xie, T. J. Reilly, J. R. Munson, R. E. Tyler, B. J. Akerley, A. L.
Smith, B. O. Palsson, and E. Kolker. 2004. In silica metabolic model and protein
expression of Haemophilus inﬂuenzae strain Rd KW20 in rich medium. Omics 8:25-
41.

Ratcliffe, R. G., and Y. Shachar-Hill. 2006. Measuring multiple ﬂuxes through plant
metabolic networks. Plant J. 45:490-51 1.

Sanwal, B. D., and R. Smando. 1969. Malic enzyme of Escherichia coli. Diversity of
the effectors controlling enzyme activity. J. Biol. Chem. 244: 1817-1823.

. Sauer, U., F. Canonaco, S. Heri, A. Perrenoud, and E. Fischer. 2004. The soluble and

membrane-bound transhydrogenases UdhA and PntAB have divergent functions in
NADPH metabolism of Escherichia coli. J. Biol. Chem. 279:6613-6619.

Sauer, U., and B. J. Eikmanns. 2005. The PEP-pyruvate-oxaloacetate node as the
switch point for carbon ﬂux distribution in bacteria. FEMS Microbiol. Rev. 29:765-
794.

Sauer, U., V. Hatzimanikatis, J. E. Bailey, M. Hochuli, T. Szyperski, and K.
Wuthrich. 1997. Metabolic ﬂuxes in riboﬂavin-producing Bacillus subtilis. Nat.
Biotechnol. 15:448-452.

Sauer, U., V. Hatzimanikatis, H. P. Hohmann, M. Manneberg, A. P. van Loon, and J.
E. Bailey. 1996. Physiology and metabolic ﬂuxes of wild-type and riboﬂavin-
producing Bacillus subtilis. Appl. Environ. Microbiol. 62:3687-3696.

Sauer, U., D. R. Lasko, J. Fiaux, M. Hochuli, R. Glaser, T. Szyperski, K. Wiithrich,
and J. E. Bailey. 1999. Metabolic ﬂux ratio analysis of genetic and environmental

117

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

modulations of Escherichia coli central carbon metabolism. J. Bacteriol. 181:6679-
6688.

Schwender, J ., F. Goffman, J. B. Ohlrogge, and Y. Shachar-Hill. 2004. Rubisco
without the Calvin cycle improves the carbon efﬁciency of developing green seeds.
Nature 432:779-782.

Schwender, J ., J. B. Ohlrogge, and Y. Shachar-Hill. 2003. A ﬂux model of glycolysis
and the oxidative pentose phosphate pathway in developing Brassica napus embryos.
J. Biol. Chem. 278:29442-29453.

Stephanopoulos, G., A. A. Aristidou, and J. Nielsen. 1998. Metabolic Engineering:
Principles and Methodologies. Academic Press, London.

Stouthamer, A. H. 1979. The search for correlation between theoretical and
experimental growth yields. In J. R. Quayle (ed.), International review of
biochemistry. Microbial biochemistry, vol. 21. University Park Press, Baltimore.

Szyperski, T. 1995. Biosynthetically directed fractional l3C-labeling of proteinogenic
amino acids. An efﬁcient analytical tool to investigate intermediary metabolism. Eur.
J. Biochem. 232:433-448.

Ullmann, R., R. Gross, J. Simon, G. Unden, and A. Kroger. 2000. Transport of C4-
dicarboxylates in Wolinella succinogenes. J. Bacteriol. 182:5757-5764.

van der Werf, M. J ., M. V. Guettler, M. K. Jain, and J. G. Zeikus. 1997.
Environmental and physiological factors affecting the succinate product ratio during
carbohydrate fermentation by Actinobacillus sp. 1302. Arch. Microbiol. 167:332-342.

van Winden, W., P. Verheijen, and S. Heijnen. 2001. Possible pitfalls of ﬂux
calculations based on l3C-labeling. Metab. Eng. 3: 151-162.

Venning, J. D., and J. B. Jackson. 1999. A shift in the equilibrium constant at the
catalytic site of proton-translocating transhydrogenase: signiﬁcance for a 'binding-
change' mechanism. Biochem. J. 341:329-337.

Wahl, S. A., M. Dauner, and W. Wiechert. 2004. New tools for mass isotopomer data
evaluation in 13C ﬂux analysis: mass isotope correction, data consistency checking,
and precursor relationships. Biotechnol. Bioeng. 85:259-268.

Wendisch, V. F., A. A. de Graaf, H. Sahm, and B. J. Eikmanns. 2000. Quantitative
determination of metabolic ﬂuxes during coutilization of two carbon sources:

comparative analyses with Corynebacterium glutamicum during growth on acetate
and/or glucose. J. Bacteriol. 182:3088-3096.

Wiechert, w. 2001. 13c Metabolic ﬂux analysis. Metab. Eng. 3: 195-206.

118

61. Wiechert, W., and A. A. de Graaf. 1997. Bidirectional reaction steps in metabolic

networks: I. Modeling and simulation of carbon isotope labeling experiments.
Biotechnol. Bioeng. 55: 101-1 17.

62. Wiechert, W., M. Mollney, S. Petersen, and A. A. de Graaf. 2001. A universal
framework for 13C metabolic ﬂux analysis. Metab. Eng. 3:265-283.

63. Wiechert, W., C. Siefke, A. A. de Graaf, and A. Marx. 1997. Bidirectional reaction
steps in metabolic networks: II. Flux estimation and statistical analysis. Biotechnol. ‘
Bioeng. 55:118-135.

64. Wilke, D. 1999. Chemicals from biotechnology: molecular plant genetics will
challenge the chemical and fermentation industry. Appl. Microbiol. Biotechnol.
52:135-145.

65. Wilke, D. 1995. What should and what can biotechnology contribute to chemical bulk
production? FEMS Microbiol. Rev. 16289-100.

66. Wittmann, C., and E. Heinzle. 2001. Application of MALDI-TOF MS to lysine-
producing Corynebacterium glutamicum: a novel approach for metabolic ﬂux
analysis. Eur. J. Biochem. 268:2441-2455.

67. Woehlke, G., and P. Dimroth. 1994. Anaerobic growth of Salmonella typhimurium on
L(+)- and D(-)-tartrate involves an oxaloacetate decarboxylase Na+ pump. Arch.
Microbiol. 162:233-237.

68. Zamboni, N., N. Mouncey, H. P. Hohmann, and U. Sauer. 2003. Reducing
maintenance metabolism by metabolic engineering of respiration improves riboﬂavin
production by Bacillus subtilis. Metab. Eng. 5:49-55.

69. Zeikus, J. G., M. K. Jain, and P. Elankovan. 1999. Biotechnology of succinic acid

production and markets for derived industrial products. Appl. Environ. Microbiol.
51:545-552.

119

Chapter 4.
13 C-metabolic ﬂux analysis of Actinobacillus succinogenes fermentative metabolism
at different NaHCO3 and H2 concentrations
4.1 ABSTRACT
Actinobacillus succinogenes naturally produces high concentrations of succinate;

a potential intermediary feedstock for bulk chemical productions. High C02 and H2
concentrations make A. succinogenes produce more succinate and less forrnate, acetate,
and ethanol. To determine how intermediary ﬂuxes in A. succinogenes respond to CO2
and H2 perturbations, l3C-metabolic ﬂux analysis was performed in batch cultures at two
different NaHCO3 concentrations, with and without H2, using a substrate mixture of [1-
13C] glucose, [U-13C]glucose, and unlabeled NaHCO3. The resulting amino acid, organic
acid, and glycogen isotopomers were analyzed by GC-MS and NMR. In all conditions,
exchange ﬂuxes were observed through pyruvate formate-lyase, formate dehydrogenase
and/or pyruvate dehydrogenase, and through malic enzyme and/or oxaloacetate
decarboxylase. The detection of exchange ﬂux between oxaloacetate, malate, and
pyruvate indicated that pyruvate is also a node for ﬂux distribution to succinate versus
alternative fermentation products. High NaHC03 concentrations decreased the amount of
ﬂux shunted by Ca-decarboxylating activities from the succinate-producing C4 pathway
to the forrnate-, acetate-, and ethanol-producing C3 pathway. Correspondingly, pyruvate
carboxylating ﬂux increased in response to high NaHCO3 concentrations. C3-pathway
dehydrogenase ﬂuxes responded to the different redox demands imposed by the different
NaHCO3 and H2 concentrations. Overall, the metabolic ﬂux changes allowed A.

succinogenes to maintain a constant growth rate and biomass yield in all conditions.

120

4.2 INTRODUCTION

Succinate is produced as a fermentation product by a variety of bacteria and can
be converted into many important commodity chemicals, including 1,4-butanediol,
tetrahydrofuran, and y —butyrolactone, which are largely used to make plastics, resins,
and solvents (25, 43). These chemicals are currently produced from petrochemical
feedstocks, namely from butane through maleic anhydride. By producing these chemicals
from succinate, a crude oil-based industry could be replaced by one based on renewable
resources. Succinate produced by fermentation has the additional environmental beneﬁt
of using CO2, a greenhouse gas, as a substrate. To create a bio-based succinate industry,
production costs must be made competitive with that of maleic anhydride, which is in
increasing demand despite rising oil prices (1). Obtaining an organism that produces high
succinate concentrations at high rates, with little byproduct formation, would allow for
more cost effective substrate utilization and succinate puriﬁcation processes.

Actinobacillus succinogenes is a capnophilic, gram-negative bacterium that
naturally produces succinate as a major fermentation endproduct. After extensive
engineering, Escherichia coli can produce nearly 100 g/l succinate in fed-batch
conditions, but it requires transitioning from an aerobic growth phase to an anaerobic
production phase (35). In contrast, wild-type A. succinogenes can grow and produce
nearly 80 g/l succinate in a single batch fermentation (10). Fluoroacetate-resistant A.
succinogenes mutants can produce over 100 g/l succinate in batch fermentations; the
highest succinate concentrations ever reported (9). With evidence that the already high

succinate titers obtained with the wild-type A. succinogenes can be improved through

121

genetic modiﬁcations, A. succinogenes is a logical starting point for engineering an
industrial succinate-producing bacterium.

To engineer A. succinogenes effectively it is important to know which enzymes
and pathways are involved in distributing ﬂux to succinate and alternative fermentation
products, and to understand how ﬂuxes through these pathways respond to perturbations.
A. succinogenes metabolism was ﬁrst studied using in vitro enzyme assays and
fermentation balances (34). Enzyme activities levels indicated that A. succinogenes
metabolizes glucose to phosphoenolpyruvate (PEP) via glycolysis with the involvement
of the oxidative pentose phosphate pathway (OPPP). PEP was thought to serve as the
branch point to the forrnate-, acetate-, and ethanol-producing C3 pathway and the
succinate-producing C4 pathway (Fig. 4.1).

Succinate production from PEP involves one CO2 ﬁxation step and two reduction
steps (Fig. 4.1). Succinate yields increase with increasing CO2 and reductant (e.g., H2)
concentrations. The reason for these increased yields was thought to simply be the
increased availability of these C4 pathway substrates (34). Oxaloacetate (OAA)
decarboxylase and malic enzyme activities were also detected in A. succinogenes cell
extracts, suggesting that, in addition to PEP, OAA and malate were nodes for ﬂux
distribution between the C4 and C3 pathways (34). A recent l3C-metabolic ﬂux analysis
conﬁrmed in vivo ﬂux from 0AA and malate to pyruvate (Pyr) in A. succinogenes, and it
also indicated that more ﬂux to Pyr came from OAA and malate than from PEP (18).
This large C4-decarboxylating ﬂux suggests a model in which PEPCK feeds both the C4

and C3 pathways. In this model, CO2 availability could affect ﬂux through both

122

Glc
PEP ATP
up! 002
Pyr 2 NADPH

GSP 1‘ - RSP

empl I

F6P ppp3 37" pppl
ATP \j) <

 

 

“”2 E4P G3P RSP
G3P + G3P<— pppZ j
“”3 NADH
C 02 ATP

ATP ATP F orDH
OAAdec FOI’Q 002

MDH :
"ADV OAA \ Pyr PFL NADH
NADPH CO2 41(1)”
02

 

 

Mal J MEnz NADH
Fm CO2 C
AcCoA
Fum 2 MAI)+ ATP
NR ADH AK
NAo+
2'3 an» Suc EtOH Ace

Figure 4.1. Simplified central metabolic pathways known to be active during A.
succinogenes mixed acid fermentation (18)). Black bold lines: succinate-producing C4 pathway.
Gray bold lines: formate-, acetate-, and ethanol-producing C3 pathway. Unidirectional arrows:
ﬂuxes considered to be unidirectional (all other ﬂuxes are considered to be reversible). The
assumption that 213 ATP is produced per fumarate reductase reaction is based on data from
Wolinella succinogenes (14). Metabolites: AcCoA, acetyl-coenzyme A; Ace, acetate; EtOH,
ethanol; E4P, erythrose-4-phosphate; For, forrnate; Fum, fumarate; F6P, fructose-6-phosphate;
Glc, glucose; G3P, glyceraldehyde—3-phosphate; G6P, glucose-6-phosphate; Mal, malate; 0AA,
oxaloacetate; PEP, phosphoenolpyruvate; Pyr, pyruvate; R5P, pentose-phosphates; Suc,
succinate; and S7P, sedoheptulose-7-phosphate. Pathways and reactions: ADH, alcohol
dehydrogenase; AK, acetate kinase; empI, 2, and 3, Embden-Meyerhoff—Parnas or glycolytic
reactions; Fm, fumarase; ForDH, forrnate dehydrogenase; FR, fumarate reductase; MDH, malate
dehydrogenase; MEnz, malic enzyme; OAAdec, oxaloacetate decarboxylase; OPPP, oxidative
pentose phosphate pathway; PEPCK, PEP carboxykinase; PFL, pyruvate forrnate-lyasc; PK,
pyruvate kinase and PEP: glucose phosphotransferase system; pppl and 2, transketolase; ppp3,
transaldolase; PyrDH, pyruvate dehydrogenase; upt, glucose phosphorylation by hexokinase and
PEP: glucose phosphotransferase system.

123

pathways, rather than just the C4 pathway. This recent l3C-labeling study also indicated
that little flux occurs through the OPPP (l8). NADPH was instead generated by ﬂuxes to
and through the C3 pathway (i.e., by malic enzyme and by forrnate and pyruvate
dehydrogenases [ForDH and PyrDH] coupled with transhydrogenase) (18). Therefore,
reductant availability may exert some control over C3 pathway ﬂuxes, in addition to
directly affecting C4 pathway ﬂux. Here we examine the effects of CO2 and reductant
concentrations on A. succinogenes intermediary metabolism using l3C-metabolic ﬂux
analysis in batch fermentations. NaHCO3 and H2 were used as C02 and reductant sources,
respectively. The N aHCO3 concentrations chosen were based on those shown to affect A.
succinogenes fermentation balances in the deﬁned medium, AM3 (20). A mixture of [1-
l3C]glucose and [U-UC] glucose was used to increase the amount of information obtained
from amino acid and organic acid isotopomers over our previous ﬂux analysis that used

[1-13C]glucose alone (18).

124

4.3 MATERIALS AND METHODS

4.3.1 Chemicals, bacteria, and culture conditions. All chemicals were purchased from
Sigma-Aldrich (St. Louis, MO) unless stated otherwise. A. succinogenes type strain 1302
(ATCC 55618) was purchased from the American Type Culture Collection (Manassas,
VA) and adapted to growth in AM3 medium as described (20). AM3 is a chemically
deﬁned medium containing Cys, Met, and Glu, which are required for A. succinogenes
growth. For this paper’s experiments, AM3 was modiﬁed by omitting the 1 g/l NaCl and
adjusting Cys-HCl, Met, and monosodium Glu concentrations to 17, 21, and 240 mg/l,
respectively, to more closely reﬂect their requirements for biosynthesis (18). AM3 was
prepared by distributing 8.3 ml aliquots of a 1.2X AM3 basal solution containing the '
phosphate buffer, NH4Cl, monosodium Glu, and minerals, to 28-ml anaerobe tubes
(Bellco, Vineland, NJ). Tubes were sealed with rubber bungs and aluminum crimps, then
repeatedly evacuated and ﬂushed with N2 to create anaerobic conditions. In tubes to be
used for growth conditions with H2, N2 was replaced by H2. After autoclaving, excess
pressure was released, then tubes were pressurized with 3 ml of either N2 or H2 at 1
ATM. The total H2 supplied was ~0.8 mmols. CO2-saturated 1 M NaHCO3 was prepared
as described (38), and 0.25 or 1 ml was then injected into each tube. Tubes receiving 0.25
ml 1M NaHCO3 also received 0.75 ml 1M NaCl. Each tube also received 0.5 ml of an
anoxic (under N2) nutrient mix containing 1 M glucose, 200 Hill] vitamin mix (20), 0.2
mg/ml kanamycin, 0.34 g/l Cys-HCl, and 0.42 g/l Met. After adding 0.2 ml of inoculum,
all ﬁnal volumes were 10 ml with 50 mM glucose, and satisﬁed one of four growth

conditions: A, 25 mM NaHCO3 under N2; B, 25 mM NaHCO3 under H2; C, 100 mM

125

N aHCO3 under N2; and D, 100 mM NaHCO3 under H2. Cultures were incubated at 37°C

with shaking at 250 rpm.

4.3.2 Growth and sampling conditions in the presence of 13C-glucose. The AM3
medium contained [l-BC] glucose and [U-13C] glucose (99% each; Cambridge Isotope
Laboratories, Cambridge, MA) at a molar ratio of 1.97: 1 .00. Starter cultures from
glycerol frozen stocks were grown in each growth condition to an OD660 of 0.7 - 1.1.
Starter cultures were harvested by centrifugation, and washed once in an equal volume of
anoxic (under N2) 1X basal solution. Cells were resuspended in 1X basal solution to an
OD660 of ~ 2, and 0.2 ml was inoculated into four replicates of each growth condition.
Growth was monitored using a Spectronic 20 spectrophotometer (Bausch and Lomb,
Rochester, NY) to calculate growth rates. A l-ml sample was taken 2 — 2.5 hours into
growth, rather than at the time of inoculation, to serve as a ‘0 time point’ sample and
account for any lag phase. This sample was used for a more accurate OD660 measurement
using a Beckman DU 650 spectrophotometer (Fullerton, CA). It was also analyzed by
high performance liquid chromatography (HPLC) to determine glucose and product
concentrations. Culture growth was stopped by chilling on ice at 1.1 - 1.2 OD660
(conditions A and B) and at 1.5 — 1.7 OD660 (conditions C and D). Cell densities were
recorded, then cultures were harvested by centrifugation at 4°C. Supematants were stored
at -70°C. The biomass was washed twice with 0.9% NaCl at 4°C, resuspended in 0.5 ml

distilled water at 4°C, then stored at -70°C.

4.3.3 Analytical procedures. Glucose and product concentrations were determined by

HPLC as described (18, 20). These values along with growth rates were used to calculate

126

extracellular ﬂuxes using the Monod model as described (18, 29, 31). To obtain l3C-
labeled amino acids for analysis by gas chromatography-mass spectrometry (GC-MS),
harvested biomass was thawed on ice, then sonicated three times. Lysates were then
diluted to 1 ml with distilled water at 4°C and centrifuged. Supernatant samples (100 pl)
were transferred to 1.5-ml screw cap-conical tubes and diluted to 700 111 with distilled
water. Proteins were precipitated with trichloracetic acid and washed twice with acetone
as described (18). Protein pellets were hydrolyzed, and amino acids partially puriﬁed,
derivatized, and analyzed by GC-MS as described (18). Organic acids and glucose
monomers of glycogen were prepared and analyzed by 1H-NMR and lH-decoupled 13C-

NMR as described (18).

4.3.4 Enzyme assays. Cell extracts for enzyme assays were prepared from A.
succinogenes grown in each of the four conditions with unlabeled glucose. Cultures were
harvested at an OD660 of 1.0 — 1.2 by centrifugation at 4°C. Pellets were washed once in
an equal volume of 100 mM Tris-HCl (pH 7.4), then resuspended in 2 ml of the same
buffer. Cells were lysed by sonicating three times and lysates were stored at -70°C. After
thawing, lysate subsamples were centrifuged to obtain soluble cell extracts, while
particulate cell extracts were not centrifuged. Cell extract protein was quantiﬁed using
the bicinchoninic acid assay (19). All assays were performed in l-ml volumes with 3-6
replicates. All reactions were started by adding substrate or cell extract. Extinction
coefﬁcients (mM'l cm") used were: 3-acetylpyridine adenine dinucleotide, 6.1 (36);
benzyl viologen, 8.65; OAA, 0.95; and NADH and NADPH, 6.23. ForDH was assayed

under anoxic conditions by monitoring benzyl viologen reduction as described (18), but

127

with 1 mM dithiotheitol and 0.1 mg/ml particulate cell extract protein. Hydrogenase was
assayed using the ForDH protocol, except that forrnate was excluded and reactions were
initiated by replacing the N2 headspace with H2. Malic enzyme was assayed as described
(18) with 2 mM NI-14Cl and ~20 ug/ml soluble cell extract protein. Na+-dependent OAA
decarboxylase was assayed as described (18) with ~0. lmg/ml particulate cell extract
protein. PyrDH was assayed as described (22) in a mixture containing ~0. 12 mg/ml
soluble cell extract protein, except that Tris-HCl (pH 7 .4) was used as the buffer.
Transhydrogenase was assayed as described (28) in a mixture containing ~0.1 mg/ml

particulate cell extract protein.

4.3.5 Metabolic modeling and flux analysis. The initial A. succinogenes metabolic
model was based on the results obtained from a previous l3C-labeling experiment (18).
Modiﬁcations were then made to this model (e.g., setting some reactions as reversible)
based on information obtained by including [U-BC] glucose in the substrate isotopomer
mixture. Thermodynamic considerations on reaction reversibility made use of AG"
values obtained from Metacyc (2) or from the Thermodynamics of Enzyme-catalyzed
Reactions Database (8). Cys uptake was removed from the model because the medium
contained only enough Cys for protein synthesis. All modeling and ﬁtting of ﬂuxes to the
experimental data sets were performed using 13C-Flux software (40). I3C-Flux uses a
metabolic network, substrate isotopomer conditions, and an arbitrary set of ﬂuxes to
simulate an isotopomer data set. This simulated data set is then compared to the measured
data set, and the similarity between the two is scored by a sum of squared residuals

(SSms), weighted by the standard deviation of each measurement:

128

(measured value — simulated yalue)2
2 measured standard deviation

SSies=

A ﬁtting algorithm is then used to adjust the arbitrary ﬂuxes in an iterative process until a
set of ﬂuxes is found that results in a simulated isotopomer data set that closely resembles
the measured data set; an optimized ﬁt. Because the substrate isotopomer mixture used in
the model included natural l3C abundance, amino acid and succinate fragment mass
isotopomers were corrected for natural abundances in all atoms of the derivatization
agent, all amino acid heteroatoms, and for unlabeled inoculum biomass. This correction
procedure was performed using software described by Wahl et al. (37). For succinate
mass isotopomers, there was no initial unlabeled succinate to correct for. Glucose
monomers from glycogen were manually corrected for unlabeled inoculum biomass as
described (7). Although mass isotopomer standard deviations were less than 1% of the
mean, standard deviations between mass isotopomers that should give identical
information (i.e., M-85 and M-159 fragments of the same amino acid) were usually
greater (i.e., ~1.2% of the mean). To take this variability into account, mass isotopomer
values with standard deviations less than 1.2% of the mean were raised to 1.2%. This
adjustment did not signiﬁcantly affect the ﬂuxes nor did it greatly increase conﬁdence
intervals (data not shown). Anabolic ﬂuxes used in the model were based on the A.
succinogenes biomass composition (18). Because the choice of free ﬂux starting values
can affect the resulting 88",, value, we used ~100 different starting values for each model
to determine if multiple solutions existed. This process was automated using a per] script,

‘autoﬂux3.pl,’ written by Hart Poskar (University of Manitoba, Canada).

129

4.4 RESULTS

4.4.1 Conﬁrmation of pseudo-metabolic steady state and choice of substrate
isotopomers. Metabolic ﬂux analysis requires a metabolic steady state. Previously,
metabolic steady state in A. succinogenes batch cultures was conﬁrmed by comparing
extracellular ﬂuxes and speciﬁc isotopic enrichments at different cell densities during
exponential growth (18). We measured extracellular ﬂuxes for the four growth conditions
(A, 25 mM NaHCO3; B, 25 mM NaHCO3 + H2; C, 100 mM NaHCO3; and D, 100 mM
NaHCO3 + H2) at three different cell densities, in duplicate (i.e., two biological replicates
per cell density per condition for a total of six cultures per condition). Only 2 of the 60
comparisons showed a statistical difference (2-tailed t-test, equal variance, p<0.05), as
would be expected when taking into account false positive frequencies. Based on these
extracellular ﬂux data, and on our previous investigation into product isotopomer
distributions at different cell densities in batch cultures (18), we felt conﬁdent that the
batch cultures exhibited pseudo-metabolic steady state.

Previous l3C-labeling experiments revealed a C4-decarboxylating ﬂux, forming a
shunt from OAA and malate to Pyr (18). However, the conﬁdence interval for this ﬂux
was large. To determine whether the C4-decarboxylating ﬂux could be measured with
greater conﬁdence we ran simulations with different glucose and CO2 isotopomers using
the previously determined set of ﬂuxes (18). We then examined the resulting sensitivity
matrices using the Estimatestat program in I3C-Flux (40). The simulations indicated that
no mixture would greatly improve our ability to quantify the C4-decarboxylating ﬂux. On
the other hand, a practical and economical substrate isotopomer mixture consisting of [1-

13C] glucose, [U-BC] glucose, and unlabeled NaHCO3, was predicted to greatly increase

130

the number of isotopomers carrying information on the C4-decarboxylating ﬂux. Using
[1-‘3C] glucose alone, 28 measurable product isotopomers provide information on the C4-
decarboxylating ﬂux. With the described substrate isotopomer mixture, 59 product
isotopomers can be used to quantify this ﬂux, and 56 of these isotopomers were used in
this study. As before, the information on the Cadecarboxylating ﬂux depends on
isotopomers generated by the symmetry of fumarate and by the C4-pathway exchange
ﬂux (Fig. 4.2). When using [U-BC] glucose, unlabeled CO2 and [U-13C]PEP combine to
form [1,2,3-‘3C]OAA, which is then converted to fumarate by the C4 pathway. Label
scrambing by fumarate symmetry generates a unique C4 isotopomer that can be tracked
throughout metabolism. To maximize isotopomer levels from [1-‘3C] glucose and [U-

13C] glucose, they were supplied at approximately a 2:1 molar ratio.

4.4.2 Modifying the A. succinogenes ﬂux model based on results from growth
condition C. Growth condition C most closely resembles the 150 mM NaHCO3
conditions used for the previous A. succinogenes l3C-labeling study (18). A. succinogenes
fermentation balances and extracellular ﬂuxes were shown to be similar in AM3 between
75 and 150 mM N aHCO3 (20). The previously established model (18) was used to ﬁt
ﬂuxes to the condition C data set. Surprisingly, a high SSm (>4500) was obtained. The
weighted squared residuals (Sm) of each measurement (i.e., [measured value — simulated
value]2 -:— [standard deviationlz) were examined to identify which isotopomer
measurements contributed the most to the SSms (40). The highest Sm values were from
isotopomers that provide information on C4 pathway reversibility (i.e., mass isotopomers

of Asx, Suc, and Thr M-57 fragments and Asx and Phe f302 fragments). Table 4.1 shows

131

From [U-‘acJGlc
PEP co2

E

Y

From [1 -‘3C]Glc l E
: OAA/Mal E

PEP co2

EEI E

Y

OAA/Mal

mgaml IEEE

1

Fum

EEIE
EBEE

---_-------------------------------------1

EIIII EII

Fum For

EIEE E. E
EIIII E. I

Figure 4.2. Two-step illustration of ﬂuxes leading to isotopomers that provide
information on C4 pathway reversibilityC and on the Ca-decarboxylating ﬂux. White
boxes, 12C; Black boxes, l3C. Step 1: [2- C]fumarate and [2,3 ,4-‘3C]fumarate originate
in C4 pathway ﬂuxes from forward C4 pathway ﬂuxes and fumarate symmetry. Step 2:
Isotopomers resulting from [2-‘3C]fumarate and [2, 3, 4- l3C]fumarate through C4 pathway
ﬂux and the C4 decarboxylating ﬂux. Abbreviations are the same as in Figure 4.1.

l OAA/Mal .
EIEE 2 EEE g

132

that ~12-13% of the C4 compounds, Asx, Suc, and Thr M-57 fragments were double-
labeled. Using multiple different starting values for C4 pathway exchange ﬂuxes did not
result in signiﬁcantly lower SS,“ values. However, setting the C4-decarboxylatin g
reaction as reversible decreased the SSM value from >4500 to 1520, indicating that the
double-labeled C4 isotopomers must come from Pyr. Ala M-57 fragment measurements
show that there is also an unexpectedly high percentage of double-labeled Pyr (Table
4.1). We previously only considered irreversible ﬂux from OAA and malate to Pyr
because an uncharacterized A. succinogenes mutant that is deﬁcient in PFL activity (23),
accumulates and excretes Pyr (9), suggesting that it cannot carboxylate Pyr. The ﬂux
model indicates that, in addition to PEP, OAA, and malate, Pyr is a node for ﬂux

distrubution to succinate and alternative products.

 

Mass isotgomer distributionsb (% of product population)

 

 

Fragment Carbons X+0 X+ l X+2 X+3 X+4 X+5
Ala M-57 1-3 29.4 1 0.1 32.51 0.1 20.0 1 0.4 18.1 1 0.3
Asxc M-57 1-4 30.3 1 0.2 32.8 1 0.1 12.2 1 0.3 24.1 1 0.1 0.6 1 0.0
1302 1-2 58.3 1 0.2 17.1 1 0.3 24.7 1 0.2
Gly M-57 1-2 64.8 1 0.4 5.8 1 0.4 29.4 1 0.1
M-85 2 67.3 1 0.2 32.7 1 0.2
Met M-57 1-5 94.11 0.5 3.2 1 0.3 0.6 1 0.8 1.9 1 0.1 0.0 1 0.1 0.1 1 0.1
M-85 2-5 95.9 1 1.1 1.9 1 0.2 1.9 1 0.9 0.3 1 0.6 0.0 1 0.4
Phe f302 1-2 64.7 1 0.1 5.2 1 0.1 30.1 1 0.1
Ser M-57 1-3 31.2 1 0.1 34.5 1 0.2 5.2 1 0.1 29.1 1 0.1
M-85 2-3 31.5 1 0.1 37.2 1 0.1 31.3 1 0.2
Thr M-57 1-4 30.3 1 0.3 32.4 1 0.1 12.4 1 0.3 24.4 1 0.4 0.5 1 0.1
1 4

Suc M-57 31.2 1 0.4 32.3 1 0.1 13.3 1 0.3 22.7 1 0.1 0.5 1 0.0

 

Table 4.1. Selected TBMDSa-amino acid and succinate fragments obtained from A.
succinogenes grown without H2 in AM3 with 100 mM NaHCO3 (Condition C).

a tert-butyldimethylsilyl. Details of fragmentation patterns for TBDMS-amino acids are
described by Dauner and Sauer (4), and Wahl et al. (37).

b The distributions shown were corrected as described in the methods as well as for
natural '3 C abundance in the amino and organic acid backbones using an excel macro
developed by Dr. Joerg Schwender (Brookhaven National Laboratory, Upton, NY) based
on correction matrices described by Lee et al. (17).

c Combined data for Asp and Asn since Asn is oxidized to Asp during hydrolysis.

133

Previously we considered the possibility of reversible PFL and PyrDH ﬂuxes, but
using [l-‘3C] glucose alone provided no information on these exchange ﬂuxes (18). The
new isotopomer mixture indicates that both PFL and PyrDH ﬂuxes can be reversible.
These exchange ﬂuxes allow for recombination of labeled and unlabeled forrnate or CO2
with different acetyl-CoA isotopomers, generating the 20% of double-labeled Pyr
(observed in Ala M-57 fragments; Table 1). PFL must be reversible to obtain low SSms
values. SSms values are lowest when both PFL and PyrDH are made reversible. While the
original model depicts PyrDH activity, it is important to note that ForDH was included in
this activity. Both PyrDH and ForDH activities were detected in cell extracts, but the in
vivo ﬂuxes through these enzymes cannot be distinguished by the labeling experiments
(18), thus only one of the two reactions is represented. When ForDH is substituted in the
model for PyrDH, similar SS,“ values are obtained. Furthermore, setting PyrDH or
ForDH exchange ﬂuxes to zero in models with either PyrDH or ForDH affects the Sres
values of the same isotopomer measurements. Since the new labeling conditions are
sensitive to the reversibility of these reactions, we used ForDH in the model instead
PyrDH. ForDH and PFL, with AG“ values of -15 and —17 kJ/mol, respectively, are more
likely to be reversible than PyrDH, with a AG" value of —33 kJ/mol.

The original model (18) used Ser and Gly mass isotopomer measurements as
exact imprints of G3P (the model was simpliﬁed by including 3-phosphoglycerate in the
G3P pool). Fitting to the condition C data set resulted in high S,Les values for Ser M-57
and M-85 fragment measurements. Forward ﬂuxes from G3P to Ser should generate
approximately equal percentages of Ser M-85 x+0, x+1, and x+2 mass isotopomers, even

when taking into account label scrambling from the C4 pathway. Table 4.1 shows that the

134

percentage of Ser M-85 x+1 mass isotopomer is higher than expected. This pattern could
result from a reversible ﬂux from Gly to Ser, combining ”Q with double-labeled G1 y
and 13C. with unlabeled Gly. The C1 intermediate in Gly synthesis is 5,10—methylene-
tetrahydrofolate, which can potentially inter-convert with forrnate through N-formyl-
tetrahydrofolate or with 5-methyl-tetrahydrofolate from Met degradation. Surprisingly, a
small percentage of Met is labeled (Table 4.1). Not only is there a signiﬁcant percentage
of Met x+1 mass isotopomer, as might be expected through Cl exchange, but there is also
a signiﬁcant percentage of Met M-57 x+3 mass isotopomer. The A. succinogenes draft
genome sequence encodes a homocysteine transmethylase (EC 2.1.1.13; Asuc_DRAFI‘
1212) suggesting that Cl exchange with methionine is possible. The presence of Met x+3
mass isotopomers may indicate ﬂux from Thr to Met. However, neither homoserine O-
acetyltransferase (EC 2.3.1.31) nor homoserine O-succinyltransferase (EC 2.3.1.46) are
encoded in the A. succinogenes draft genome sequence, thus separating methionine from
central metabolism. To date, we have been unable to grow A. succinogenes in AM3 in the
absence of Met (20). Met x+3 mass isotopomers may be an artifact. At best, ~10-20 times
less Met mass isotopomers are obtained from hydrolyzed protein compared to most other
amino acid mass isotopomers. Thus, Met mass isotopomer measurements may be
particularly susceptible to the effects of contaminating fragments.

Ser, Gly, C1, and Met pathways were remodeled to account for these observations,
including Met x+3 mass isotopomers (Fig. 4.3). The modiﬁed model was ﬁt to the
condition C data set, and the Ser fragments’ Sres values decreased signiﬁcantly. The
optimized ﬁt indicated that little, if any, 5,10-methylene-tetrahydrofolate was derived

from forrnate or from 5-methyl-tetrahydrofolate. Therefore, even if some Met originates

135

from Thr, Ser isotopomers are not affected because there is insigniﬁcant Cl exchange
between Met and Ser. The label scrambling observed in Ser is due almost entirely to the
reversibility of ley alone and to the large amount of labeled Ser C3 (~ 66% of Ser C3 is

expected to be labeled from forward ﬂuxes from the glucose isotopomer mixture).

EMP + PPP
I 4.05 1 0.54 3.16 1 0.52
69 (52, 89 14 (11, 17)

G3P 1 Ser T Gly
I K 11"" Ct <1 0.52
PEP 0.02 H3 4 For

Asp

0.28 1 0.05/y Met \oio 1 0.05
5 6

OAA Met_out Protein

Figure 4.3. Glycine synthesis and C1 metabolism used in the metabolic flux model.
Flux units are percent of glucose uptake rate 1 90% conﬁdence intervals. Exchange
ﬂuxes that could be determined are shown in italics with unsymmetrical exchange ﬂux
90% conﬁdence intervals in brackets. Conﬁdence intervals were calculated as described
(41). Exchange ﬂuxes between OAA to Thr could not be quantiﬁed from the available
isotopomers. Although the model allowed for exchange ﬂuxes between Thr and Met and
between C1 and For (as indicated by the bidirectional arrows), the exchange ﬂuxes were
not statistically different from zero. Reactionszl, vSer (simpliﬁed serine biosynthesis
pathway); 2, ley (glycine hydroxymethyltransferase); 3 vMetl(simpliﬁed methionine
biosynthesis pathway); 4, vCl (simpliﬁed pathway from 5,10-methylene-tetrahydrofolate
to forrnate); 5, vMet_in (methionine uptake); 6, vMet_prot (methionine incorporated into
protein). Other abbreviations are the same as in Figure l.

The described modiﬁcations to the A. succinogenes metabolic map are illustrated
in Figure 4.4. This new model was used to ﬁt ﬂuxes to the growth condition C data set.
The resulting ﬂuxes were, for the most part, very similar to those obtained for growth

with 150 mM NaHCO3 using [1-13C]glucose alone (18). Again, OPPP ﬂux is low at ~3%

136

of the glucose uptake rate, compared with 5%, obtained previously. All product
formation rates were similar, relative to the glucose uptake rate. The most signiﬁcant
differences between the two ﬂux maps are the ﬂuxes between PEP, OAA/Mal, and Pyr,
which are all dependent on each other. The new ﬂux map shows a C4-decarboxylating
flux (here-on referred to as C4dec, comprised of both malic enzyme and OAA
decarboxylase activities) that is 38% of the glucose uptake rate, compared to 64%
obtained previously (18). [1-13C]Glucose alone cannot be used to detect reversible C4dec
ﬂux, but it does produce [3-13C]Ser, which allows for label scrambling from exchange
ﬂux between Ser and Gly. To determine if the Ser and Gly pathway modiﬁcations affect
the determination of the C4dec ﬂux, we included the reversible vSer and ley reactions
in the original model and ran the ﬁtting algorithm with the data set obtained using [1-
l3C]glucose alone (18). The optimized ﬁt had an SS”,gs of 182 (compared to 148 obtained
previously) and the C4dec net ﬂux decreased to 41% of the glucose uptake rate, a value
similar to that obtained in this study. When vSer exchange ﬂux was set to zero in the new
model and used with the growth condition C data, the obtained C4dec net ﬂux values
were between 63% and 81% of the glucose uptake rate (SSres from 355-358). Therefore
the modiﬁcations to the Ser and Gly pathway may have resulted in an over-estimated
C4dec ﬂux in the previous model. However, even using the new model (Fig. 4.4) C4dec is

still a major catabolic ﬂux in growth condition C.

137

Glc

 

 

  
  

 

up: ‘ 002
Gly GSP 0””
CHIP];
O, my F6P‘—>< PPP3 $7P pppl
em 2
Ser ” A? E4P G3P asp
G3P + G3P
vSer PPPZ
lemp3 002
ForDV
PEPCK For F . F or
O A Al 66 or_out
Fm Mal : C4d \ Py:
., PFL
f. CO2 '5"..,...P,w/)H
Fum F—bFum 002‘ “ AcCoA
um out
FR; — ADH/ \AK
Sue EtOH Ace

Figure 4.4. Modiﬁed metabolic flux model of A. succinogenes mixed acid
fermentative metabolism. Net ﬂux directions are indicated by an enlarged arrow head.
Gray arrows highlight the modiﬁcations that were made. PyrDH is shown by dotted
arrows to indicate that ForDH replaced it. Most amino acid synthesis pathways are not
shown. See Figure 4.3 for more details on the modeling of C. metabolism. Reactions:
C4dec, OAA decarboxylase and malic enzyme; F 0r_0ut, forrnate excretion; F um_out,

fumarate excretion. Other abbreviations are the same as in Figures 4.1 and 4.3.

138

4.4.3 Effects of NaHC03 and H2 concentrations on metabolic ﬂuxes. C02 and
reductant were thought to affect fermentation balances through their role as substrates for
succinate production in the C4 pathway (Fig. 4.1) (20, 34). 13C metabolic ﬂux analysis of
A. succinogenes in a single growth condition suggested that the mechanisms by which
C02 and reductant affect fermentation balances are more complex (18). To deﬁne the
effects of C02 and reductant on metabolism, we performed l3C-metabolic ﬂux analysis
on A. succinogenes fermentative metabolism at two different NaHCO3 concentrations in
the presence and absence of H2. Table 4.2 shows the growth parameters for the four
growth conditions. Despite product formation rate changes, glucose uptake rates, growth
rates, and biomass yields were similar for all conditions. During the experiment, ~ 4%
and 10% of the H2 was estimated to be consumed in conditions B and D, respectively. H2
oxidation rates were less than succinate formation rates indicating that succinate
production was only supported in part by H2 oxidation (Table 4.2). Speciﬁc rates were
normalized to the glucose uptake rate and used in 13C-Flux ﬂux models for each
condition. The SSres values for each model were: condition A, 192; condition B, 150;
condition C, 233; and condition D, 163. Since all models have 42 degrees of freedom,
and all SSres values are >58, none of the models passed the )6 test at the 95% conﬁdence
interval, a common result for biological systems (39). However, these SS“,s values are
similar to that obtained previously (SSm = 148) for A. succinogenes grown with 150 mM
NaHCO3 (18). Higher SS“,s values were expected in this study because: (i) the minimum
mass isotopomer standard deviation was set to 1.2% of the mean, compared to 2.6% of
the mean used previously (18); and (ii) 197 measurements were used versus 127

previously (18). The estimated ﬂuxes in each growth condition are shown in Table

139

.Aomooiw do 88% .8: 8255 be 3:33 539w 335:0me mctsd 22» 2385 8

Am: 2620 do 2:: ZHU ”mamweﬁoozm .< 05 .33 80:88? E a: :5 20

828988 3520:: 283.: m m_ :02? .Ndzmdommu do 0.8805 3::st :8 m5? tea—:23 803 8538: 5:83 :5 59:6 5

9260.5 :82on

:5 3385 E @8820 2d 3 owﬁseoa 05 33:22: .3 89:83 me £9: 85 3:028 83 25:83 3088 do omﬁﬁéom 2:.

.2 rwov 33:8 m: :5 Am $.01: .23 836:8 a 5253 3538.. :obooB do one 2: 80¢ 3:358 803 mean: E5258 a: a

6:53:53 539.» 59.3.56 2: :8 328:: 9.82523 535.5 .N... «San.

 

 

 

_dd H mad _dd H dmd N H dd— N H mm dd H 5.3 _d H v._ mddd H Ngd md H .1» vd H Nd Vd H as. vd H vd + 8— Q
5d H mNd _dd H and N H 02 N H ad ad H N._ mddd H mddd _d H 06 _d H Ev Nd H _.b md H ed . 9: U
5d H vmd 5d H dvd m H we N H 3 Nd H d.: _d H md Edd H Edd Nd H d6 Nd H v.5 Nd H m6 md H .d + mm m
ddd H vmd Nod H 3d _ H 02 _ H 3 ad H N.N Nddd H dmdd vd H dd vd H ad vd H w.m vd H nd . mN <
32:;
UAW.” wv ATMSV 9?de nAokov «mm: roam: :5": 8<L 8n: gawk oak N: mOUIdZ
> on: -o co: .0
3.: 7w 38.5 8:: 23:0on £55ng

 

140

4.3. Unless stated otherwise, all ﬂux descriptions refer to net ﬂuxes. Net ﬂux corresponds
to the forward ﬂux minus the reverse ﬂux, and the exchange ﬂux is the remainder (39,
40). Exchange ﬂuxes reﬂect the reversibility of a reaction. Since all ﬂux values in this
study are reported as being positive in the directions indicated in Figure 4.4, an exchange
ﬂux is equal to the reverse ﬂux, and a forward ﬂux is the sum of the net and exchange
ﬂuxes.

In addition to similar a glucose uptake rate, growth rate, and biomass yield for all
conditions (Table 4.2), the glycolytic and OPPP ﬂuxes were also similar (Table 4.3). The
one exception is the emp3 exchange ﬂux. Large exchange ﬂuxes (i.e. that for emp3) can
generally only be determined within an order of magnitude (40, 41). The ﬂux changes
that allow A. succinogenes to maintain its growth parameters in the different growth
conditions occur downstream of PEP. The effects of NaHCO3 and H2 on these ﬂuxes are

described in detail in the following sections.

NaHC03 effects. As expected, higher succinate formation rates (FR) and larger C42C3
ﬂux ratios were observed at 100 mM NaHCO3 (C and D) than at 25 mM NaHCO3 (A and
B) (Table 4.3). Interestingly, fumarate formation rates were also signiﬁcantly higher in
conditions C and D than in conditions A and B (2-tailed t-test, equal variance, p<0.01;
Table 4.2). Although ~100 times less fumarate is excreted than the major fermentation
products, fumarate can be accurately quantiﬁed by HPLC because it strongly absorbs at
210 nm. NaHC03 did not drastically affect the Fm reverse ﬂux, which was approximately
proportional to the forward ﬂuxes (i.e., 24% and 28% that of the forward ﬂux rate in

conditions A and C, respectively).

141

PFL ﬂux, the central C3 pathway ﬂux in this model, is lower in conditions C and
D than in conditions A and B, as are C3 pathway product formation rates (i.e., forrnate,
acetate, and ethanol; Table 4.3). PFL exchange ﬂuxes were also lower in conditions C
and D than in conditions A and B. This decreased reverse ﬂux could be due in part to
lower forrnate concentrations in conditions C and D, but it could also be due to lower
PFL activity in general, since the reverse ﬂux decreased proportionately to the forward
ﬂux, in all conditions. The reverse PFL ﬂux was 72%, 72%, 69%, and 67% that of the
forward ﬂux in conditions A, B, C, and D, respectively.

It was previously suggested that the C02 concentration affects the ﬂux
distribution through the C4 and C3 pathways by increasing PEPCK ﬂux (34). Contrary to
this hypothesis, PEPCK ﬂuxes are similar at high and low NaHC03 concentrations
(Table 4.3). However, comparing PEPCK ﬂuxes is complicated by large conﬁdence
intervals. PEPCK conﬁdence intervals are large because PEPCK ﬂux is dependent on the
C4dec ﬂux, which is not well determined. PEPCK exchange ﬂuxes are also similar in all
conditions (32% to 35% that of the forward ﬂux), and they are better determined than the
net ﬂuxes. These similar exchange ﬂuxes indicate that the C02 concentration at least does
not affect PEPCK reversibility. An alternative (or additional) mechanism for CO2 to
affect the fermentation balance is by inﬂuencing the amount of ﬂux shunted from the C4
to the C3 pathway. The C4dec ﬂux is lower in conditions C and D than in conditions A
and B, but again, comparisons are complicated by large conﬁdence intervals (Table 4.3).

Unlike C4dec net ﬂuxes, C4dec exchange ﬂuxes were relatively well determined. C4dec

142

 

Fluxes (Percent of glucose uptake rate 1 90% conﬁdence intervals) a

 

 

A B C D
Fig. 4 Reactions 25 mM Hco3' 25 mM Hco3' + H2 100 mM Hco3' 100 mM Hco3' + H2
Glycolysis and OPPP
upt 100.0 1 4.0 100.0 1 2.5 100.0 1 2.3 100.0 1 4.2
OPPP 2.6 1 1.3 6.9 1 5.9 2.7 1 1.2 3.7 1 1.2
empl 95.6 1 4.0 91.4 1 5.5 95.7 1 2.5 94.6 1 4.2
emp2 94.5 1 3.9 93.4 1 2.6 94.8 1 2.3 94.3 1 4.1
emp3 182.0 1 7.8 181.5 1 4.7 182.8 1 4.6 182 1 8.3
emp3 1143 (616, 4606) 1926 (1024, 10185) 970 (641, 1827) 457 (261, 1118)
C3 and C4 pathway
Per 64.01210 51.11472 60.91151 64.01 30.1
PFL 101.8 1 6.6 97.9 1 3.9 87.2 1 3.6 78.0 1 5.6
PFL 268 (249, 288) 250 (232, 270) 190 (179, 203) 158 (145, I72)
ForDH 33.1 1 9.0 21.3 1 4.7 40.5 1 4.1 16.8 1 9.4
ForDH 30 (22, 39) 11(7, 15) 29 (21, 37) 22 (16, 29)
For__out 72.0 1 6.2 79.8 1 2.6 49.8 1 2.0 64.4 1 7.6
ADH 27.0 1 3.5 32.0 1 2.5 16.2 1 2.5 18.6 1 2.8
AK 62.1 1 5.6 53.6 1 2.7 58.5 1 2.2 46.9 1 4.7
PEPCK 115.8 1 21.6 128.2 1 47.2 119.8 1 15.4 115.9 1 30.6
PEPCK 55 (44, 67) 69 (58, 82) 61 (52, 71) 58 (47, 71)
Pm 59.9 1 5.5 63.5 1 2.7 75.6 1 3.1 83.9 1 7.0
Fm 19 (17, 21) 15 (12, 17) 30 (28, 32) 17(15, 20)
Fum_out 0.30 1 0.03 0.20 1 0.01 0.99 1 0.07 0.50 1 0.05
FR 59.6 1 5.5 63.3 1 2.7 74.6 1 3.1 83.5 1 7.0
Other ﬂuxes
C4dec 49.6 1 21.3 58.4 1 47.2 38.0 1 15.2 25.6 1 30.2
C4dec 105 (92, 121) 124 (96, 160) 170(154, 187) 158 (130, 192)
vSer 67 (48, 91 ) 83 (47, 141) 69 (52, 89) 75 (58, 95)
ley 12 (9, 15) 14 (8, 21) 14 (11, 17) 16 (13, 20)
(:02 uptake 24.0 1 3.6 35.0 1 9.9 32.1 1 2.6 63.1 1 14.9
H; oxidation 30.2 1 0.4 50.0 1 2.4
ATP” 67.2 1 30.9 61.3 1 52.0 79.1 1 25.5 73.5 1 37.7
NADHC 53.4 1 13.1 23.7 1 7.9 52.3 1 8.2 5.5 1 14.9
Ratios
ForDH:PFL 0.33 0.22 0.46 0.22
ADHzAK 0.43 0.60 0.28 0.40
C 4 (Fm):C3 (PFL) 0.59 0.65 0.87 1.08

 

Table 4.3. Effect of NaHCO3 concentration and H2 on net and exchange fluxes.

3 Exchange ﬂuxes are italicized with unsymmetrical 90% conﬁdence intervals in parentheses.
Conﬁdence intervals were calculated as described (41). Anabolic ﬂuxes out of metabolic
intermediates were within the following ranges for all conditions: G6P, 1.7-1.8; F6P, 0.5; R5P, 2.8-
3.1; E4P, 0.9; G3P, 6.1-6.2; PEP, 2.2; Pyr, 11.6-11.8; AcCoA, 12412.7; and OAA, 6.2-6.4. Net
ﬂuxes involved in Ser and Gly syntheses (see Fig. 3) were similar in all conditions (data not shown).
Non-oxidative pentose phosphate pathway net and exchange ﬂuxes are not shown.

b Equals the net ATP ﬂux produced by central metabolic ﬂuxes minus the estimated ATP ﬂux for
biosynthesis and transport (~200% of the glucose uptake rate; based on the A. succinogenes biomass

composition [18]).

c Includes the estimated NADH ﬂux from biosynthesis (~11.6% of the glucose uptake rate). Net
NADH is assumed to be converted to NADPH by transhydrogenase or by malate dehydrogenase
coupled with malic enzyme. The estimated NADPH ﬂux for biosynthesis was ~57.2% of the glucose

uptake rate.

143

exchange ﬂuxes were higher in conditions C and D than in conditions A and B, indicating
that high N aHCO3 concentrations stimulate reverse C4dec ﬂux. In vitro OAA
decarboxylase activity levels were similar in all conditions (Table 4.4). Higher malic
enzyme activity levels were detected in extracts from cells grown with 100 mM NaHCO3
(C and D) than in those of cells grown with 25 mM NaHCO3 (A and B) (Table 4.4). This
malic enzyme activity trend correlates with the reverse C4dec ﬂux, suggesting that malic

enzyme may be involved in pyruvate carboxylation.

 

Speciﬁc activity (nmol mg protein'1 min'l)

 

 

Enzyme assayed n A B C D
25 mM 25 mM 100 mM 100 mM

NaHCO3 NaHCO3 + H2 NaHCO3 NaHC03 + 112
Transhydrogenase 4 61 1 5 64 1 2c 62 1 7 54 1 48b
Pyruvate dehydrogenase 3 l6 1 0 14 1 0c 18 1 1a 15 1 1b
Formate dehydrogenase 3 36 1 4 37 1 5 27 1 4a 31 1 4‘16
Hydrogenase 3 400 1 60 290 1 120c 350 1 50 600 1 60”
Malic enzyme 6 460 1 20 430 1 20b 520 1 40a 460 1 303"
0AA decarboxylase 3 1060 1 110 990 1 150 980 1 1 10 970 1 80

 

Table 4.4. Enzyme activities in cell extracts of A. succinogenes grown in different
NaHCO3 and H2 conditions

a Statistically different from identical N2/H2 condition at 25 mM NaHCO3 (2-tailed t-test,
equal variance, p < 0.01).

b Statistically different from identical NaHCO3 condition without H2 (2-tailed t-test, equal
variance, p < 0.01).

c Statistically different from identical N aHCO3 condition without H2 (2-tailed t-test, equal
variance, p < 0.05)

d Statistically different from identical N2/H2 condition with 25 mM NaHCO3 (2-tailed t-
test, equal variance, p < 0.1).

c Statistically different from identical NaHCO3 condition without H2 (2-tailed t-test, equal
variance, p < 0.1).

Regardless of the mechanism by which CO2 affects ﬂux distribution between the

C4 and the C3 pathways, the increased succinate production at high CO2 concentrations

144

requires reductant. Glycolytic ﬂux and biomass yields were nearly identical in all growth
conditions (Tables 4.2 and 4.3). Therefore, the extra reductant needed for succinate
production in condition C was not generated by an increased glycolytic ﬂux or by the C4
pathway out-competing anabolism for reducing power. The main reductant sources for
biomass and high succinate producing rate in condition C are in the C3 pathway. In
condition C, the ethanol-producing ADH ﬂux decreased ~ 40% from that in condition A.
Producing 1 ethanol from 1 AcCoA requires 2 N ADH (Fig. 4.1). The ﬂux through
ForDH (and/or PyrDH) was higher in condition C than in condition A. Enzyme activity
data suggest that the increased succinate producing ﬂux in condition 3 most likely
coincides with increased ﬂux through PyrDH rather than through ForDH (Table 4.4).
ForDH activity may actually be inhibited by high CO2 conditions (Table 4.4). The
combination of decreased NADH consumption and increased NADH production in the
C3 pathway meets the high electron-demand in condition 3. In fact, the net NADH
production by the C3 pathway increases just enough to compensate succinate production
increase from conditions A to C, as illustrated by the similar NADH net ﬂuxes in
conditions A and C (Table 4.3). As noted previously (18), this excess NADH is likely
converted to N ADPH by transhydrogenase, an especially important process for
biosynthesis given the low OPPP ﬂux. Similar transhydrogenase activity levels were
detected in extracts of cells grown in conditions A and C (Table 4.4). The similar net
NADH ﬂuxes in conditions 1 and 3 demonstrate the ﬂexibility of the C3 pathway
dehydrogenase ﬂuxes for generating enough reducing power to maintain a biomass yield

in conditions with different reductant demands for succinate production.

145

H2 effects. For H2 to affect metabolism, it must be oxidized. Hydrogenase activity was
previously reported in A. succinogenes cell extracts (23, 24). It was also detected in this
study, and activity levels were higher in extracts of cells grown under H2, at least in high
C02 conditions (Table 4.4). Interestingly, hydrogenase activity was present even in cells
grown in the absence of H2.

In agreement with previous observations (34), growing A. succinogenes with H2
increased succinate production rates (FR; Table 4.3), but more so at high NaHCO3
concentrations (C vs. D). At both NaHCO3 concentrations, the presence of H2 decreased
fumarate formation rates signiﬁcantly (2-tailed t-test, equal variance, p<0.01; Fum_out;
Table 4.3). Decreased fumarate excretion suggests that reductant availability is limiting
fumarate reductase ﬂux, and that this limitation can be relieved, at least in part, by
supplying H2. While adding H2 generally increased the net Fm ﬂux, it also decreased the
Fm exchange ﬂux. Adding H2 decreased the reversezforward Fm ﬂux ratio from 0.24 and
0.28 (conditions A and C, respectively) to 0.19 and 0.17 (conditions B and D,
respectively). Reverse C4 pathway ﬂux generates the isotopomers used to quantify the net
C4dec ﬂux. Correspondingly, the conditions exhibiting the least Fm exchange ﬂux (i.e.,
conditions B and D) have the largest C4dec conﬁdence intervals. Thus, the perturbations
used to study the changes in ﬂux distribution may also affect our ability to quantify a
metabolic ﬂux of central interest.

H2 was hypothesized to affect ﬂux distributions through its role as a substrate for
the C4 pathway (34). Recently, the C3 pathway was shown to be important for generating
reductant, thus leading to an additional hypothesis that H2 can decrease C3 pathway ﬂux

by serving as an alternative reductant source (18). Supporting the new hypothesis, there is

146

less central C3-pathway ﬂux, PFL ﬂux, in condition D than in condition C (Table 4.3).
Acetate formation rates decreased in the presence of H2, but ethanol and forrnate
formation rates increased (Table 4.2). H2 is expected to stimulate ethanol production
because ethanol production requires reductant (Fig. 4.1). At both NaHCO3 conditions, the
presence of H2 increased the ethanolzacetate ﬂux ratio 1.4 fold, due to the combined
decreased ﬂux to acetate and increased ﬂux to ethanol (Table 4.3).

More than causing a decrease in the total C3 pathway ﬂux (at least at 100 mM
NaHCO3), H2 partially alleviated the need for NADH production by ForDH and/or
PyrDH. The net NADH ﬂux is lower in the presence of H2, and even approaches zero in
condition D (Table 4.3). The decrease in ForDH ﬂux in response to H2 is greater than the
decrease in PFL ﬂux, as reﬂected by the ForDH2PFL ratios (Table 4.3), resulting in
higher forrnate formation rates in conditions B and D than in conditions A and C. In
agreement with the observed ﬂuxes, in vitro PyrDH activity was lower in extracts of cells
grown under H2 (Table 4.4). In contrast, ForDH activity was similar in conditions 1 and 2
(25 mM NaHCO3), and may have increased in the presence of H2 at 100 mM NaHCO3.
With a high succinate production rate and low CO2 production rate (i.e., through
ForDH/PyrDH and C4dec ﬂuxes) in condition D, the net CO2 consumption rate is 2.6
times greater than in condition A.

ForDH exchange ﬂuxes were lower in cultures grown under H2, and unlike what
was observed with PFL, the reversezforward ForDH ﬂux ratio varied widely from 0.34
(condition B) to 0.57 (condition D). While NaHCO3 concentration did not affect ForDH
exchange ﬂuxes in the absence of H2 (conditions A and C), the ForDH exchange ﬂux as

much lower in condition B than in condition D. What causes these exchange ﬂux

147

variations is not obvious, but could be due to disproportional effects of N aHCO3 and H2
concentrations on ForDH and PyrDH.

Malic enzyme ﬂux is also a potentially important source of NADPH (18).
According to this hypothesis, supplying H2 could alleviate the need for malic enzyme
ﬂux if H2 oxidation can be coupled to NADPH production. C4dec ﬂuxes, which contain
malic enzyme ﬂux, are best compared at high NaHCO3 concentrations, however
conﬁdence intervals remain large. In these conditions (100 mM NaHCO3), supplying H2
decreased the C4dec ﬂux by 33% (Table 4.3). Correspondingly, in vitro malic enzyme
activity was less in extracts of cells grown with H2 at both NaHCO3 concentrations (Table
4.4). The effects of H2 on C4dec exchange ﬂuxes were less pronounced than those of
NaHCO3. Assuming that the net C4dec ﬂuxes are accurate, H2 had no observable affect
on the C4dec exchange ﬂux at 25 mM NaHCO3, with the reverse ﬂux being 68% that of
the forward ﬂux in both conditions A and B. The same effect is observed at 100 mM
NaHCO3, where the reverse C4dec ﬂux is 82% and 86% that of the forward ﬂux in

conditions C and D, respectively.

148

4.5 DISCUSSION

This study has made several contributions to the understanding of A. succinogenes
fermentative metabolism: (i), the use [1-13C] glucose and [U-13 C] glucose together brought
new insights into the reversibility of some pathways; (ii), pyruvate was shown to
participate as a node in the ﬂux distribution between the C3 and C4 pathways; (iii)
decreasing the C4-decarboxylating ﬂux was suggested as a new mechanism for how
NaHCO3 affects ﬂux distribution; (iv) H2 was shown to affect C3 pathway ﬂuxes in
addition to C4 pathway ﬂuxes; and (v) C3 pathway dehydrogenase ﬂuxes were shown to
be ﬂexible to meet different reductant demands.

Product isotopomers generated from [U-‘3C] glucose metabolism provided
information on exchange ﬂuxes (e.g., PFL exchange ﬂux) that could not be quantiﬁed by
using [l-l3 C] glucose alone. Reversible PFL ﬂux in A. succinogenes is supported by the
fact that a ﬂuoroacetate-resistant A. succinogenes mutant has no PFL activity (23). While
ﬂuoroacetate itself is not toxic, metabolic products of ﬂuoroacetate, such as
ﬂuoropyruvate, are toxic. Since the ﬂuoroacetate-resistant strain has no detectable PFL
activity, it suggests that PFL is required for ﬂuoroacetate toxicity, likely by generating
ﬂuoropyruvate through reverse ﬂux. Reverse PFL ﬂux is likely inﬂuenced by internal,
rather than external, forrnate concentrations. Formate consumption was not detected, and
fermentation balances were not affected, when forrnate was added to A. succinogenes
cultures (34). The lowest SSres values were obtained when PyrDH or ForDH were
allowed to be reversible. Although usually considered to be an enzyme of aerobic
metabolism, PyrDH activity was detected in extracts of A. succinogenes grown

anaerobically (Table 4.4), and PyrDH is expressed in closely related, anaerobically

149

grown, Haemophilus inﬂuenzae (26). Based on reaction thermodynamics, ForDH (AG°'
= -15 kJ/mol) is more likely to be reversible than PyrDH (A0“ = -33 kJ/mol). However,
high CO2 concentrations could affect PyrDH reversibility. Isocitrate dehydrogenase is
commonly considered to be a unidirectional enzyme, having a A0" of ~30 kJ/mol.
However, in the high CO2 environment of developing Brassica napus embryos,
mitochondrial isocitrate dehydrogenase carries a net reverse ﬂux (30). While PyrDH
reversibility cannot be assessed from the current data set, it could be assessed if the same
methods were applied to an A. succinogenes PFL-knockout strain.

The C4-decarboxylating ﬂux, composed of OAA decarboxylase and malic
enzyme, was also shown to be reversible. The A. succinogenes OAA decarboxylase
(AsucDRAFl‘_1557-1559), is believed to use the energy from OAA decarboxylation
(A0“: -29 kJ/mol) to transport Na+ across the cytoplasmic membrane, as has been
shown for the homologous Klebsiella aerogenes enzyme (5). The high AG" value
suggests that OAA decarboxylase is unlikely to be reversible. However, inverted vesicles
containing K. aerogenes OAA decarboxylase accumulated slightly more 22Na* in the
presence of pyruvate, HCO3', and 22NaCl than with 22NaCl alone (5). Malic enzyme
typically functions as a malate decarboxylating enzyme because it has high Km values
for pyruvate (~5-15mM) and CO2 (~l3-25 mM) (www.brend§.uni-koeln.de/index4311434,
Accessed Nov. 2006). However, the pyruvate carboxylating direction is
thermodynamically favored under standard conditions (AG° '= —8.7 kJ/mol). In fact, net
pyruvate-carboxylating ﬂux through malic enzyme has been shown in a succinate-
producing E. coli mutant (32). While reverse ﬂux through OAA decarboxylase cannot be

ruled out, malic enzyme is likely responsible the majority of the observed C4dec

150

exchange ﬂux. The NaHCO3 concentrations used in this study are near or above the
reported Km values for malic enzyme. Furthermore, the C4dec reverse ﬂux values are
higher at 100 mM NaHCO3 than at 25mM NaHCO3 (Table 4.3), indicating that the
reverse ﬂux is inﬂuenced by the CO2 concentration. Potentially, malic enzymes’s
pyruvate carboxylating ability could be harnessed to create a net ﬂux to malate in A.
succinogenes, as was done for E. coli (32). However, observing this ﬂux is not possible if
there is signiﬁcant ﬂux through OAA decarboxylase.

Reverse ﬂux from Pyr to malate and possibly 0AA has several implications. First,
it indicates that, in addition to PEP, OAA, and malate, pyruvate can also serve as a node
for ﬂux distribution between the C3 and the C4 pathways. Second, reverse ﬂux from Pyr
to OAA/malate increased in response to elevated NaHCO3 concentrations relative to the
forward ﬂux, resulting in a decreased C4dec net ﬂux. Thus, CO2 concentration may affect
succinate production by determining the amount of ﬂux shunted to the C3 pathway. This
C02 effect on the C4-decarboxylating ﬂux may work alone or in conjuction with the CO2
effect on ﬂux through PEPCK, which has been the prevailing hypothesis for CO2’s
effects on succinate production to date.

While PEPCK ﬂux was similar in all conditions, it could not be concluded that
CO2 does not affect PEPCK ﬂux due to large conﬁdence intervals (Table 4.3). Elevated
CO2 concentrations are thought to increase PEPCK ﬂux by overcoming limiting CO2
substrate concentrations and allowing PEPCK to out compete pyruvate kinase and the
glucose:PEP phosphotransferase system for PEP. There is evidence in other bacterial
systems suggesting that CO2 affects succinate production through PEPCK ﬂux. In

succinate-producing Anaerobiospirillum succiniciproducens, PEPCK activity and

151

succinate production both increase in response to increased CO2 concentration and
decreased pH (27). However, A. succinogenes fermentation balances are relatively
constant at different pH values and in vitro PEPCK activities are similar in extracts of
cells grown with different CO2 concentrations (34). In an E. coli ppc‘ mutant,
overexpressing A. succinogenes PEPCK improves succinate yields, indicating that
PEPCK ﬂux is important for succinate production (13). Furthermore, E. coli succinate
production was enhanced by overexpressing E. coli PEPCK, but only at high NaHCO3
concentrations (15). While these E. coli studies did not address C4-decarboxylating
ﬂuxes, they suggest that the CO2 concentration affects succinate production through its
effect on PEPCK ﬂux. It would be interesting to test whether overexpressing PEPCK in
A. succinogenes would increase ﬂux to succinate at high NaHCO3 concentrations or
whether C4 dearboxylating ﬂuxes would counter the effect of increased PEPCK ﬂux.
Interestingly, the effects of H2 on metabolic ﬂuxes were more pronounced at 100 mM
NaHCO3 than at 25mM NaHCO3 (Table 4.3). This trend suggests that sufﬁcient CO2 is a
prerequisite for H2 to stimulate succinate production, either by increasing PEPCK ﬂux or
by decreasing the C4-decarboxylating ﬂux, to make OAA and malate available for
reduction to succinate.

Malic enzyme ﬂux is a potentially important NADPH source in A. succinogenes
(18). C4dec ﬂuxes and in vitro malic enzyme activity levels suggest that supplying H2
partially alleviates the need for NADPH-producing malic enzyme ﬂux, assuming that
malic enzyme carries a net decarboxylating ﬂux (Tables 4.4 and 4.3). However, the
C4dec conﬁdence interval in condition D was very large and the C4-decarboxylating ﬂux

decreased and the pyruvate carboxylating ﬂux increased in response to elevated NaHC03

152

concentrations, despite the increased reductant demands by the accompanied increased
succinate formation rate (A vs. C; Table 4.3). This decreased C4dec ﬂux suggests that
transhydrogenase is more important for NADPH production than malic enzyme.
However, even in the presence of H2 optimized ﬁttings suggest signiﬁcant C4dec ﬂux. In
vitro OAA decarboxylase activity was similar for all conditions (Table 4.4). While this
enzyme is required for growth on citrate by other bacteria (3, 5, 42), its role during
glucose fermentation is an intriguing ﬁnding that warrants further investigation,
especially given its ability to divert ﬂux away from succinate. OAA decarboxylase could
be important for Na+-linked energetics (6, 11). In closely related A.
actinomycetemcomitans, OAA decarboxylase is involved in Na+-dependent glutamate
transport and drug efﬂux (11). A Na+ gradient could also be involved in fumarate and
succinate transport, as was shown for Wolinella succinogenes (33). OAA decarboxylase
could also be important for maintaining osmostic balance by pumping Na+ out of the cell.
Since the C4dec ﬂux is composed of both malic enzyme and 0AA decarboxylase, a high
OAA decarboxylase ﬂux could mask a net pyruvate carboxylating ﬂux by malic enzyme.
Modifying Na+ concentrations in the growth medium could be an insightful perturbation
to understand OAA decarboxylase’s role in A. succinogenes and gage the contributions of
OAA decarboxylase and malic enzyme to the total C4-decarboxylating ﬂux.

C3 pathway dehydrogenases are important for generating NADPH in A.
succinogenes when coupled with transhydrogenase (18). Alcohol dehydrogenase and
ForDH/PyrDH ﬂuxes responded to changes in NaHCO3 concentration to maintain a net
NADH ﬂux that is likely coupled to NADPH production through transhydrogenase

(Table 4.3). Therefore, ﬂux distribution in the C3 pathway is important for maintaining

153

redox balance and appears to be one the most ﬂexible aspects of A. succinogenes
fermentative metabolism. Ultimately, metabolic engineering of A. succinogenes
metabolism aims to minimize C3 pathway ﬂux and maximize C4 pathway ﬂux. Given the
importance of PyrDH, ForDH, and possibly malic enzyme for generating reducing power,
it will be important to have an alternative reductant source if C3 pathway ﬂux is
disrupted. Supplying H2 partially alleviated the need for N ADH generation in the C3
pathway (Table 4.3). Supplying H2 also decreased the fumarate formation rate,
suggesting that fumarate reductase ﬂux is limited by reductant availability. Although H2
is an attractive electron source, H2 oxidation is likely coupled to menaquinone reduction,
since H2 oxidation drove fumarate reduction in puriﬁed A. succinogenes membranes (24).
Thus, although H2 oxidation can supply reduced menaquinone for succinate production,
C3 pathway ﬂux will remain important for biomass production if H2 oxidation cannot be
coupled to N ADH or N ADPH production. A. succinogenes can grow by fumarate
respiration with H2, a process that involves hydrogenase 2, rather than hydrogenase l, in
E. coli (16, 21) and the A. succinogenes hydrogenase operon is most similar to
hydrogenase 2. Therefore, it may be necessary to overexpress a ferredoxin reducing
hydrogenase to couple H2 oxidation to NADH production and relieve the need for NADH
production by C3 pathway. Estimated H2 oxidation rates were less than succinate
formation rates in both conditions B and D, indicating that reducing power for fumarate
reductase was still generated to a large extent from carbon-based reactions. Therefore it
may also be worthwhile to increasing the H2 oxidation rate by hydrogenase 2. The H2
oxidation rate may increase naturally if ForDH and PyrDH are knocked out, or

overxpressing hydrogenase 2 may be required. Alternatively, H2 oxidation rates may

154

increase naturally if A. succinogenes is grown in bioreactors with higher dissolved H2
concentrations. Of course, H2 use in an industrial succinate fermentation will require an
inexpensive source of H2, perhaps from another bio-based process.

The effects of the NaHCO3 and H2 concentrations on in vitro enzyme activities,
although sometimes signiﬁcant, were relatively subtle in this study (<1 .5 fold differences;
Table 4.4). In no case was an enzyme activity or a metabolic ﬂux shut down in response
to a NaHCO3 or H2 perturbation. Perhaps most interesting was the observation of high
hydrogenase activity in extracts from cells grown without H2 (Table 4.4). Citrate lyase
was detected in cell extracts of A. succinogenes grown in fermentative and fumarate
respiration conditions (34). This activity was also surprising because citrate was not
provided and citrate and isocitrate have never been observed in A. succinogenes culture
supematants (McKinlay et al. unpublished data). Pasteurellaceae, the family to which A.
succinogenes belongs, are host-associated organisms, with few free-living examples.
Even Pasteurella multocida, which can be isolated from water and soil, may associate
with amoebae to survive in these environments (12). It is possible that the natural A.
succinogenes ecology is restricted to the rumen, where CO2 and H2 are major products of
other microbial fermentations. Without drastic changes to its environment, A.
succinogenes may have evolved to have a relatively uniform metabolism.

This study has illustrated the complexities of A. succinogenes metabolism, with
exchange ﬂuxes in, and between, the C3 and C4 pathways, and four nodes involved in the
distribution of ﬂux between the two pathways. The methods presented in this paper,
when combined with genetic engineering, indicate that A. succinogenes could be used to

address issues of in vivo malic enzyme, OAA decarboxylase and pyruvate dehydrogenase

155

reversibility. This study also illustrated how A. succinogenes maintains a ﬁxed growth
rate and biomass yield in different growth conditions by modulating its metabolic ﬂuxes.
When the NaHCO3 concentration is raised from 25 mM to 100 mM, the C4dec reverse
ﬂux increases, and the PEPCK ﬂux may also increase. The resulting increase in C4
pathway ﬂux, must be channeled to succinate for the cell to compensate energetically for
the decreased ﬂux to acetate (fumarate reductase is linked to an electron transport chain).
The reductant demand from the increased succinate formation rate is met by increasing
ForDH and PyrDH ﬂuxes and decreasing the ﬂux to ethanol. When H2 is supplied,
PyrDH and ForDH ﬂuxes decrease in response to the higher reductant availability. These
discoveries will be important to consider when engineering A. succinogenes metabolism.

They should also be of general interest for the study of microbial metabolism.

156

4.6 REFERENCES

l.

10.

ll.

12.

Brown, R. 2005. Housing sector boosts maleic anhydride, p. 24, Chemical Market
Reporter, vol. 267.

Caspi, R., H. Foerster, C. A. Fulcher, R. Hopkinson, J. Ingraham, P. Kaipa, M.
Krummenacker, S. Paley, J. Pick, S. Y. Rhee, C. Tissier, P. Zhang, and P. D. Karp.

2006. MetaCyc: a multiorganism database of metabolic pathways and enzymes.
Nucleic Acids Res. 34:D511-D516.

Dahinden, P., Y. Auchli, T. Granjon, M. Taralczak, M. Wild, and P. Dimroth. 2005.
Oxaloacetate decarboxylase of Vibrio cholerae: puriﬁcation, characterization, and
expression of the genes in Escherichia coli. Arch. Microbiol. 183:121-129.

Dauner, M., and U. Sauer. 2000. GC-MS analysis of amino acids rapidly provides
rich information for isotopomer balancing. Biotechnol. Prog. 16:642-649.

Dimroth, P. 1980. A new sodium-transport system energized by the decarboxylation
of oxaloacetate. FEBS Lett. 122:234-236.

Dimroth, P., and B. Schink. 1998. Energy conservation in the decarboxylation of
dicarboxylic acids by fermenting bacteria. Arch. Microbiol. 170:69-77.

Fischer, E., and U. Sauer. 2003. Metabolic ﬂux proﬁling of Escherichia coli mutants
in central carbon metabolism using GC-MS. Eur. J. Biochem. 270:880-891.

Goldberg, R. N ., Y. B. Tewari, and T. N. Bhat. 2004. Thermodynamics of enzyme-
catalyzed reactions - a database for quantitative biochemistry. Bioinformatics
20:2874-2877.

Guettler, M. V., M. K. Jain, and D. Rumler. 1996. Method for making succinic acid,
bacteria] variants for use in the process, and methods for obtaining variants. US.
patent 5,573,931.

Guettler, M. V., M. K. Jain, and B. K. Soni. 1996. Process for making succinic acid,
microorganisms for use in the process and methods of obtaining the microorganisms.
US. patent 5,504,004.

Hase, C. C., N. D. Fedorova, M. Y. Galperin, and P. A. Dibrov. 2001. Sodium ion
cycle in bacterial pathogens: evidence from cross-genome comparisons. Microbiol.
Mol. Biol. Rev. 65:353-370.

Hundt, M. J., and C. G. Ruffolo. 2005. Interaction of Pasteurella multocida with free—
living amoebae. Appl. Environ. Microbiol. 71 :5485 - 5464.

157

13. Kim, P., M. Laivenieks, C. Vieille, and J. G. Zeikus. 2004. Effect of overexpression
of Actinobacillus succinogenes phosphoenolpyruvate carboxykinase on succinate
production in Escherichia coli. Appl. Environ. Microbiol. 70: 1238-1241.

14. Kroger, A., S. Biel, J. Simon, R. Gross, G. Unden, and C. R. D. Lancaster. 2002.
Fumarate respiration of Wolinella succinogenes: enzymology, energetics and
coupling mechanism. Biochim. Biophys. Acta. 1553:23-28.

15. Kwon, Y. D., S. Y. Lee, and P. Kim. 2006. Inﬂuence of gluconeogenic
phosphoenolpyruvate carboxykinase (PCK) expression on succinic acid fermentation

in Escherichia coli under high bicarbonate condition. J. Microbiol. Biotechnol.
16:1448-1452.

l6. Laurinavichene, T. V., N. A. Zorin, and A. A. Tsygankov. 2002. Effect of redox
potential on activity of hydrogenase l and hydrogenase 2 in Escherichia coli. Arch.
Microbiol. 178:437-442.

17. Lee, W. N., L. O. Byerley, E. A. Bergner, and J. Edmond. 1991. Mass isotopomer
analysis: theoretical and practical considerations. Biol. Mass. Spectrom. 202451-458.

l8. McKinlay, J. B., Y. Shachar-Hill, J. G. Zeikus, and C. Vieille. 2006. Determining
Actinobacillus succinogenes metabolic pathways and ﬂuxes by NMR and GC-MS

analyses of l3C-labeled metabolic product isotopomers. Metab. Eng.
doi: 10. 1016/j.ymben.2006. 10.006.

19. McKinlay, J. B., and J. G. Zeikus. 2004. Extracellular iron reduction is mediated in
part by neutral red and hydrogenase in Escherichia coli. Appl. Environ. Microbiol.
70:3467-3474.

20. McKinlay, J. B., J. G. Zeikus, and C. Vieille. 2005. Insights into Actinobacillus
succinogenes fermentative metabolism in a chemically deﬁned growth medium.
Appl. Environ. Microbiol. 71 :6651-6656.

21. Menon, N. K., C. Y. Chatelus, M. Dervartanian, J. C. Wendt, K. T. Shanmugam, H.
D. Peck, Jr., and A. E. Przybyla. 1994. Cloning, sequencing, and mutational analysis
of the hyb operon encoding Escherichia coli hydrogenase 2. J Bacteriol 176:4416-23.

22. Millar, A. H., C. Knorpp, C. J. Leaver, and S. A. Hill. 1998. Plant mitochondrial
pyruvate dehydrogenase complex: puriﬁcation and identiﬁcation of catalytic
components in potato. Biochem. J. 334:571-576.

23. Park, D. H., M. Laivenieks, M. V. Guettler, M. K. Jain, and J. G. Zeikus. 1999.

Microbial utilization of electrically reduced neutral red as the sole electron donor for
growth and metabolite production. Appl. Environ. Microbiol. 65:2912-2917.

158

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

Park, D. H., and J. G. Zeikus. 1999. Utilization of electrically reduced neutral red by
Actinobacillus succinogenes: physiological function of neutral red in membrane-
driven fumarate reduction and energy conservation. J. Bacteriol. 18122403-2410.

Paster, M., J. L. Pellegrino, and T. M. Carole. 2003. Industrial bioproducts: today and
tomorrow. Energetics, Incorporated.

Raghunathan, A., N. , D. Price, M. Y. Galperin, K. S. Makarova, S. Purvine, A. F.
Picone, T. Chemy, T. Xie, T. J. Reilly, J. R. Munson, R. E. Tyler, B. J. Akerley, A. L.
Smith, B. O. Palsson, and E. Kolker. 2004. In silica metabolic model and protein

expression of Haemophilus inﬂuenzae strain Rd KW 20 in rich medium. Omics 8:25-
41.

Samuelov, N. S., R. Lamed, S. Lowe, and J. G. Zeikus. 1991. Inﬂuence of CO2-HC03
levels and pH on growth, succinate production, and enzyme activities of
Anaerobiospirillum succiniciproducens. Appl. Environ. Microbiol. 57:3013-3019.

Sauer, U., P. Canonaco, S. Heri, A. Perrenoud, and E. Fischer. 2004. The soluble and
membrane-bound transhydrogenases UdhA and PntAB have divergent functions in
NADPH metabolism of Escherichia coli. J. Biol. Chem. 279:6613-6619.

Sauer, U., D. R. Lasko, J. Fiaux, M. Hochuli, R. Glaser, T. Szyperski, K. Wiithrich,
and J. E. Bailey. 1999. Metabolic ﬂux ratio analysis of genetic and environmental

modulations of Escherichia coli central carbon metabolism. J. Bacteriol. 18126679-
6688.

Schwender, J., Y. Schachar-Hill, and J. Ohlrogge. 2006. Mitochondrial metabolism in
developing embryos of Brassica napus. J. Biol. Chem. 281 :34040-34047.

Stephanopoulos, G., A. A. Aristidou, and J. Nielsen. 1998. Metabolic Engineering:
Principles and Methodologies. Academic Press, London.

Stols, L., and M. I. Donnelly. 1997. Production of succinic acid through
overexpression of NAD+-dependent malic enzyme in an Escherichia coli mutant.
Appl. Environ. Microbiol. 63:2695-2701.

Ullmann, R., R. Gross, J. Simon, G. Unden, and A. Kroger. 2000. Transport of C4-
dicarboxylates in Wolinella succinogenes. J. Bacteriol. 182:5757-5764.

van der Werf, M. J ., M. V. Guettler, M. K. Jain, and J. G. Zeikus. 1997.
Environmental and physiological factors affecting the succinate product ratio during
carbohydrate fermentation by Actinobacillus sp. 1302. Arch. Microbiol. 167:332-342.

Vemuri, G. N., M. A. Eiteman, and E. Altman. 2002. Succinate production in dual-

phase Escherichia coli fermentations depends on the time of transition from aerobic
to anaerobic conditions. J. Ind. Microbiol. Biotechnol. 28:325-332.

159

36.

37.

38.

39.

40.

41.

42.

43.

Venning, J. D., and J. B. Jackson. 1999. A shift in the equilibrium constant at the
catalytic site of proton-translocating transhydrogenase: signiﬁcance for a 'binding-
change' mechanism. Biochem. J. 341:329-337.

Wahl, S. A., M. Dauner, and W. Wiechert. 2004. New tools for mass isotopomer data
evaluation in 13C ﬂux analysis: mass isotope correction, data consistency checking,
and precursor relationships. Biotechnol. Bioeng. 85:259-268.

Widdel, F., and F. Bak. 1992. Gram-negative mesophilic sulfate-reducing bacteria, p.
3358-3378. In A. Balows, H. G. Truper, M. Dworkin, W. Harder, and K. H. Schleifer
(ed.), The Prokaryotes, 2nd ed, vol. 4. Springer-Verlag, New York.

Wiechert, W., and A. A. de Graaf. 1997. Bidirectional reaction steps in metabolic

networks: I. Modeling and simulation of carbon isotope labeling experiments.
Biotechnol. Bioeng. 55:101-117.

Wiechert, W., M. Mollney, S. Petersen, and A. A. de Graaf. 2001. A universal
framework for 13C metabolic ﬂux analysis. Metab. Eng. 3:265-283.

Wiechert, W., C. Sieﬂte, A. A. de Graaf, and A. Marx. 1997. Bidirectional reaction
steps in metabolic networks: 11. Flux estimation and statistical analysis. Biotechnol.
Bioeng. 55:118-135.

Woehlke, G., and P. Dimroth. 1994. Anaerobic growth of Salmonella typhimurium on
L(+)- and D(-)-tartrate involves an oxaloacetate decarboxylase Na+ pump. Arch.
Microbiol. 162:233-237.

Zeikus, J. G., M. K. Jain, and P. Elankovan. 1999. Biotechnology of succinic acid
production and markets for derived industrial products. Appl. Environ. Microbiol.
51:545-552.

160

Chapter 5.
Genomic perspective on the potential of Actinobacillus succinogenes for industrial

succinate production

5.1 ABSTRACT

Actinobacillus succinogenes naturally converts sugars and C02 into high
concentrations of succinic acid as part of a mixed-acid fermentation. Efforts are ongoing
to maximize carbon ﬂux to the succinate-producing pathway to generate industrial
amounts of succinate. Described herein is the 2.04 Mb A. succinogenes draft genome
sequence in the context of A. succinogenes’ potential for: (i) chemical production, (ii)
genetic engineering, and (iii) non-pathogenicity. The genome sequence indicates that A.
succinogenes lacks a complete tricarboxylic acid cycle as well as the glyoxylate pathway.
Nine potential dicarboxylate transporters and numerous sugar uptake and degradation
pathways were identiﬁed. The genome sequence contains 1,458 DNA uptake signal
sequence repeats and a nearly complete set of natural competence proteins, suggesting
that A. succinogenes is, or was, capable of natural transformation. The genomes of A.
succinogenes and its closest-known relative, Mannheimia succiniciproducens, were
analyzed for the presence of Pasteurellaceae virulence factors. Neither bacterium appears
to have virulence traits such as toxin production, sialic acid and choline incorporation

into lipopolysaccharide, and uptake of haemoglobin and transferin as iron sources.

161

5.2 INTRODUCTION

Actinobacillus succinogenes is a gram negative capnophilic bacterium that was
isolated from cow-rumen as part of a search for succinate-producing bacteria (26).
Succinate is an important metabolic intermediate in the rumen, where several rumen
bacteria obtain energy by decarboxylating succinate to propionate, which then serves as a
nutrient for the ruminant (51, 56, 64). Succinate is used as a specialty chemical in food,
agriculture, and pharmaceutical industries, but it has a much greater potential value for
making bulk chemicals that are currently petrochemical based (73). Thus bio-based
succinate production is both economically and environmentally attractive, with the
potential to replace a large crude oil-based industry with one based on renewable
resources. A further environmental beneﬁt is that succinate production via fermentation
uses CO2, a greenhouse gas, as substrate.

A. succinogenes is one of the best succinate producers ever described but it also
produces high forrnate and acetate concentrations. Flux distribution between succinate
and alternative fermentation products is affected by environmental conditions. For
example, increasing the available CO2 and reductant (e.g., by supplying H2 or by using
more-reduced carbon sources than glucose) shifts the fermentation toward higher
succinate yields (40, 63). However, adjusting the environmental conditions is not enough
to achieve a homosuccinate fermentation, thus genetic engineering is required. A.
succinogenes' metabolic engineering will beneﬁt from an understanding of the enzymes
and mechanisms controlling ﬂux distribution. The A. succinogenes genome sequence is

an important step in deﬁning A. succinogenes metabolic pathways. The genome sequence

162

also opens the door to functional genomic approaches to address the metabolic and
regulatory mechanisms controlling ﬂux distribution during succinate production.

A. succinogenes belongs to the Pasteurellaceae family, which includes the genera
Actinobacillus, Haemophilus, Mannheimia, and Pasteurella. Aside from A. succinogenes,
twelve Pasteurellaceae genome sequences are publicly available
(www.ncbi.nlm.nih.gov/genomes/MICROBES/microbial taxtreehtml; Accessed Nov.
2006). A. succinogenes and its close relative, M. succiniciproducens, collectively
referred to as ‘succinogens’ in this chapter, are studied for their industrially attractive
metabolisms. Other Pasteurellaceae are studied for their pathogenic traits. It will be
important to conﬁrm non-pathogenicity in the succinogens before they are used on an
industrial scale. Without any reports of disease from A. succinogenes or M.
succiniciproducens, their genome sequences are a logical starting point to assess their
potential for non-pathogenicity. If A. succinogenes is non-pathogenic, comparing its
genome sequence with those of pathogenic Pasteurellaceae could lead to further insights
into Pasteurellaceae virulence mechanisms.

Here we present the ﬁrst detailed analysis of the A. succinogenes draft genome
sequence. Special attention is given to genes encoding metabolic enzymes from a
biotechnological perspective. We also present possible opportunities for developing
genetic tools based on characteristics of the genome sequence. Finally, the A.
succinogenes and M. succiniciproducens genome sequences are examined for known

Pasteurellaceae virulence genes.

163

5.3 MATERIALS AND METHODS

5.3.1 Source strain and genomic DNA puriﬁcation. A. succinogenes type strain 1302
(ATCC 55618) was obtained from the American Type Culture Collection (Manassas,
VA). A. succinogenes was grown in 100 ml of tryptic soy glucose (TSG; Becton
Dickinson, Sparks, MD) broth with 25 mM NaHCO3 in a 160-ml anaerobic serum vial at
37°C. The culture was harvested in log phase (~7.7 x 1010 cells) and washed twice in 45
ml of phosphate buffer containing (g/l): 15.5 K2HPO4, 8.5 NaH2PO4*H2O, and 1 g/l
NaCl. Genomic DNA was puriﬁed using a Qiagen genomic tip protocol with a Qiagen
maxiprep column (Valencia, CA). Brieﬂy, the cell pellet was lysed in 10 ml Bl buffer at
37°C for 1 h, then 4 ml B2 buffer was added, and the solution was incubated at 50°C for 1
h. The lysate was vortexed, then applied to a maxiprep column that was pre-equilibrated
with QBT buffer. The column was washed twice with 15 ml QC buffer, then DNA was
eluted with QF buffer (~30 m1) until the eluent was no longer viscous. Isopropanol (0.7
vol at 25°C) was added to the eluent to precipitate DNA. DNA was then pelleted at
10,000 rpm for 15 min, washed with 70% ethanol, dried under a stream of N2, and

dissolved in 10 mM Tris-HCl (pH 8.0) containing 1 mM EDTA.

5.3.2 Sequencing and automated annotation. Sequencing was performed by the
Department of Energy Joint Genome Institute. Automated annotation was performed by
the Oak Ridge National Laboratory. Open reading frames (ORFs) were identiﬁed using
three gene caller programs: Critica, Generation, and Glimmer. Translated ORFs were
subjected to an automated basic local alignment search tool (BLAST) for proteins (3)

against Genbank’s non-redundant database. The translated ORFs were also subjected to

164

searches against KEGG, InterPro (incorporating Pfam, PROSITE, PRINTS, ProDom,

SmartHMM, and TIGRFam,), and Clusters of Orthologous Groups of proteins (COGS).

5.3.3 Manual annotation. BLAST alignments were examined to assess the correctness
of the start codon. DNA sequences upstream of each ORF were examined for a ribosomal
binding site (at least 3 nt of AAGGAGG, 5-10 nt upstream of the start codon) using the
web Artemis tool. To assign product names to each ORF, results from BLAST, HMM
searches (i.e., PFAM and TIGRFAM), and domain and motif searches were considered.
Most importantly, efforts were made to ﬁnd a citation of biological function for a
homologous gene. If a translated ORF was at least 75% identical to a protein of known
function, or if it belonged to a TIGRFAM equivalog, it was given the associated product
name. If a translated ORFH was less than 75% identical to a protein of known function,
the product name was modiﬁed as follows: 60-75% identity, putative product; 40-64%
identity, probable product; 25-39% identity, possible product. If a translated ORF was at
least 60% identical to a protein of unknown function, it was named a conserved
hypothetical protein. If there was no adequate alignment with any protein, the translated
ORF was named a hypothetical protein.

Uptake signal sequence (USS) 9 nt cores were counted and their surrounding
sequences reported using our perl script, ‘USS.pl.’ The output was pasted into a
Microsoft Excel (Redmond, WA) spreadsheet to calculate the frequency of each
nucleotide occurring at each position, upstream and downstream of the USS core. The
accuracy of USS.pl was conﬁrmed by the fact that it identiﬁed the same number of USS

in other Pasteurellaceae genome sequences as already reported (4).

165

5.4 RESULTS AND DISCUSSION

5.4.1 General features of the draft genome sequence. The A. succinogenes draft
genome sequence is 2.04 Mb, spread over 117 contigs, making it one of the smallest
industrially relevant genome sequences. The GC content is 44.9%. There is at least one
prophage, which is comprised of 32 ORFs spanning 26 kb and possibly extending beyond
the end of contig 166. Automatic annotation identiﬁed 1,884 ORFs, but failed to identify
more ORFs, usually due to truncated sequences and poor sequence quality at the ends of
contigs. Of the 1,884 ORFs, 91% were most similar to other proteobacterial genes, with

1,241 (66%) being most similar to M. succiniciproducens genes.

5.4.2 Metabolic reconstruction. A. succinogenes metabolism is of central interest to
understand the pathways involved in succinate production. The central metabolic
enzymes identiﬁed in the draft genome sequence are summarized in Table 5.1 and Figure
5.1. A. succinogenes metabolism has primarily been studied using fermentation balances,
in vitro enzyme assays, and l3C-metabolic ﬂux analyses with glucose as the main carbon
source (40, 41, 63). These studies indicated that glucose is catabolized to
phosphoenolpyruvate (PEP) via glycolysis, with little involvement of the pentose
phosphate pathway. PEP is then converted into fermentation products via the C3 and C4
pathways, with malic enzyme and OAA decarboxylase forming shunts between these
pathways. All glycolytic and pentose phosphate pathway enzymes are encoded in the
genome sequence. Two possible homologs to the Entner-Doudoroff enzyme 2-keto-3-

deoxy-6—phosphogluconate aldolase were identiﬁed (AsucDRAFI‘_0727 and 1070). A

166

 

 

ORF(S) a Function EC. # Figure 5.1b
854-856 PEP: glucose phosphotransferase 2.7.1.69 1
1877 Glucokinase 2.7.1.2 2
146 Glucose-6-phosphate dehydrogenase 1.1.1.49 3
145 6-Phosphogluconolactonase 2. l . 1 .3 l 4
141 6-Phosphog1uconate dehydrogenase 1.1.1.44 5
1544 Ribose-S-phosphate isomerase A 5.3.1.6 6
139 Ribulose-S-phosphate 4-epimerase 5.1.3.4 7
1787 Transketolase 2.2.1.1 8
413 Transaldolase 2.2.1.2 9
256 G6P isomerase 5.3.1.9 10
1282 6-Phosphofructokinase 2.7.1.1 1 1 1
458-9 Fructose-bisphosphate aldolase 4.1.2.13 12
899, 924, 1291 Triose phosphate isomerase 5.3.1.1 13
307 Glyceraldehyde-3-phosphate dehydrogenase 1.2.1.12 14
460 Phosphoglycerate kinase 2.7.2.3 15
373 Phosphoglycerate mutase 3.6.1.7 16
1030 Enolase 4.2.1.11 17
232 Pyruvate kinase 2.7.1.40 18
1849 Lactate dehydrogenase 1.1.1.28 19
C 126, ORF l Pyruvate formate-lyase 2.3.1.54 20
1404-1408 Formate dehydrogenase 1.2.1.2 21
l Pyruvate dehydrogenase El 1.2.4.1 22
2 Pyruvate dehydrogenase 132 2.3.1.12 22
3 Pyruvate dehydrogenase E3 1.8.1.4 22
405 Acetaldehyde dehydrogenase 1.2.1.10 23
405 Alcohol dehydrogenase l .1 . l .1 24
1768 Predicted aldehyde dehydrogenase 1.2.1.3 25
NA Phosphotransacetylase 2.3. l .8

NA Acetate kinase 2.7.2.1

1662 Predicted acylphosphatase 3.6.1.7 26
NA Phosphoenolpyruvate carboxykinase 4.1.1.49

1557-9 Oxaloacetate decarboxylase 4.1.1.3 27
NA Malic enzyme 1.1.1.40

1103 Malate dehydrogenase 1.1.1.37 28
1823 Fumarase 4.2.1.2 29
210-214 Fumarate reductase 1.3.99.1 30
889a,b Succinyl-CoA synthetase 6.2.1.5 31
391 a-Ketoglutarate dehydrogenase El 1-2-4-2 32
390 a-Ketoglutarate dehydrogenase E2 2-3-1-61 32
3 a-Ketoglutarate dehydrogenase E3 1-3-1-4 32
NA Isocitrate dehydrogenase 1.1.1.42

1809 Aconitase 4.2.1.3 33
388-90 Citrate lyasc 4.1.3.6 34
1869 Possible aspartate aminotransferase 2.6.1.1 35
898 Aspartate aminotransferase 2.6.1.1 35
716 Aspartate ammonia-lyasc 4.3.1.1 36
1393-7 Hydrogenase 1.12.7.2 37
393 Carbonic anhydrase 4.2.1.] 38
847-8 Transhydrogenase 1 .6. 1 .2 39

 

Table 5.1. Central metabolic enzymes expected and identiﬁed in the draft genome
sequence. a NA: no ORF found for the function. b Corresponding reaction in Figure 5.1.

167

Glc

PEP ATP
0
"w NADPH NADPHC 2
"" GGP —4—> 6P6 Rbu5P
H2 ‘43, 211+ 10 3 4 5

6
mFGP *——-t 9 .579 8 RbO5P)
co H o HCO' 11+ ’1 ><
“I 2 ‘E’ 3 * F1,6P A 7
’EA E4P Gap xsp

DHAPH G3P < j

13 8
14 I\> NADH

NADH + NADP" G1,3P

Is- 14...

+ 396‘.
NAD + NADPH 16 t

2PG

co 173 Lac
Asp «i:A AAP))\2 pg: 13 1/<:2 CO2 + I-l+
NH.+ c702\ I/NADH
NADH

 

 

 

,3“;

Fum AcCoA
. 30 1:27 \
ﬁ§iTg£ (“t 23
Sue AcAld p_ AcP
Y1 33/ NAD I 26
SucCoA CO: 001’ lclt 724 25

NADH NADH

Figure 5.1. Central metabolism based on identiﬁed enzymes in the draft genome
sequence. See Table 5.1 for reaction names. XH, unknown reduced cofactor. Q”,

menaquinone. Gray arrows indicate enzymes that were expected but not identiﬁed in the
draft genome sequence. Numbers in italics refer to enzymes listed in Table 5.1.

168

possible homolog to the other Entner-Doudoroff enzyme, phosphogluconate dehydratase,
was also identiﬁed (AsucDRAFI‘_0513). It is unlikely that this ORF encodes
phosphogluconate dehydratase, though, because this ORF is located in an isoleucine
synthesis operon and, correspondingly, is an equivalog for dihydroxy-acid dehydratase
(EC 4.2.1.9).

Most of the expected C3 and C4 pathway proteins were identiﬁed, including
pyruvate dehydrogenase and forrnate dehydrogenase, which l3C-ﬂux analyses indicated
were important for generating reducing power for succinate production and anabolism
(Chapters 3 and 4). A few genes were truncated at the ends of contigs including those
encoding pyruvate forrnate-lyasc (PFL) and succinyl-CoA synthetase. PEP carboxykinase
(PEPCK) and malic enzyme were not identiﬁed. A. succinogenes pckA (encodes PEPCK)
was previously cloned and sequenced (35), and cell extracts contain high levels of malic
enzyme activity (Chapters 3 and 4). Both enzymes are expected to be identiﬁed in the
complete genome sequence. Acetaldehyde dehydrogenase and alcohol dehydrogenase
activities are likely due to a single multifunctional alcohol dehydrogenase (33) encoded
by Asuc_DRAF1‘ 305. Asuc_DRAFI‘ 1306 is related to the H. influenzae and M.
succiniciproducens ORFs annotated as alcohol dehydrogenase, however while this
enzyme is classiﬁed as a class III alcohol dehydrogenase it functions as a formaldehyde
dehydrogenase (E.C. 1.1.1.284) (37). Phosphotransacetylase and acetate kinase were not
identiﬁed, but a conserved hypothetical protein with predicted acylphosphotransferase
activity was identiﬁed. Phosphotransacetylase and acetate kinase are expected to be
identiﬁed in the complete genome sequence, since their activities are detected in A.

succinogenes cell extracts (63). However, a conserved hypothetical protein with predicted

169

aldehyde dehydrogenase activity may form an alternative route from acetyl-CoA to
acetate without the formation of ATP (Fig. 5.1). All genes encoding TCA-cycle enzymes
were identiﬁed with the exceptions of isocitrate dehydrogenase and citrate synthase.
Isocitrate dehydrogenase activity was not detected in extracts of A. succinogenes grown
under fermentative conditions (Chapter 3). Rather than citrate synthase, a citrate lyase
gene was identiﬁed and is discussed in greater detail below. The a-ketoglutarate (aKG)
dehydrogenase lipoamide dehydrogenase (E3) subunit is shared with pyruvate
dehydrogenase. It is encoded in the pyruvate dehydrogenase operon, as it is in E. coli
(60). It was reported that A. succinogenes glutamate auxotrophy was due to the absence
or inactivity of TCA cycle-associated enzymes on either side of aKG. While aKG
dehydrogenase and succinyl-CoA synthetase connect succinate and aKG, aKG
dehydrogenase is generally considered to be a unidirectional enzyme going from aKG to
succinyl-CoA. The reductive-TCA cycle counterpart to aKG dehydrogenase, aKG
ferredoxin oxidoreductase, was not found in the draft genome sequence. As expected, the
glyoxylate pathway enzymes, isocitrate lyase and malate synthase, were not found.

The draft genome sequence was also examined for gluconeogenic enzymes.
Even though Asuc_DRAFI‘ 855 is 26% identical to the E. coli PEP synthase, it is more
likely a PEP phosphotransferase system (PTS) component, as it is part of a PT S operon
and is 76% identical to an E. coli PTS protein. There are three known types of fructose-
l,6-bisphosphatases. The draft genome sequence does not encode Type I or III fructose-
1,6—bisphosphatases, but Asuc_DRAFI‘ 1718 is 69% identical to E. coli fructose-1,6—
bisphosphatase H. This gene is found in other Pasteurellaceae including M.

succiniciproducens. In E. coli this type II enzyme is not the main gluconeogenic enzyme.

170

When the type I enzyme was knocked out in E. coli, the type H enzyme supported growth
in conditions requiring gluconeogenesis only when it was overexpressed (21). Even if
PEP synthase is not found in the complete genome sequence, A. succinogenes could
generate a gluconeogenic ﬂux by using PEPCK and fructose—1,6-bisphosphatase H.
Gluconeogenesis is expected to be functional in A. succinogenes, because A.
succinogenes can grow by anaerobic respiration using H2 or electrically reduced neutral
red as an electron source and using fumarate or malate as an electron acceptor/carbon
source (45, 46, 63). Assuming rapid equilibrium between G3P and dihydroxyacetone
phosphate, 13C-metabolic flux analysis indicates that there is little gluconeogenic ﬂux to
glucose in A. succinogenes during mixed acid fermentation on glucose (Chapter 3).
Succinate production by A. succinogenes involves carboxylation of PEP and
pyruvate, and fermentation balances are affected by C02 concentrations. No PEP—
carboxylating enzymes were identiﬁed in the draft genome sequence, and pckA had been
previously cloned and characterized (35). An ORF encoding pyruvate carboxylase was
identiﬁed. Carbonic anhydrase, which interconverts CO2 and HCOg' was identiﬁed. Since
A. succinogenes succinate production is affected by C02 and HCO3’ concentrations,
carbonic anhydrase could be an important enzyme for succinate production. A.
succinogenes’ fermentation balances are constant within a pH range that affects the
fermentation balance of the succinate-producing bacterium Anaerobiospirillum
succiniciproducens (52, 63). Carbonic anhydrase may be important for making CO2 or
HCO3‘ available at different pH values. A. succinogenes will also utilize H2 as a source of
electrons for succinate production. A membrane-bound hydrogenase and a membrane-

bound transhydrogenase were identiﬁed in the genome sequence (Table 5.1).

171

Na+-

Unk. Citrate dependent 0AA decarboxylase Amino acid
“9339 transporter transporter

FL) r_A_‘ A A A

f \ f \ f \
:m E (556 (.5.. <15“
< citF citE [a]. citS >c oadA oadB citA >citB)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\ Y J \_Y_) \ Y j \ Y J L r J
Citrate Citrate Na*- OAA decarboxylase Citrate-
Iyase "9336 dependent sensing
citrate regulatory
transporter system

Figure 5.2. Physical map of the OAA decarboxylase operon region in the A.
succinogenes (top) and K. pneumoniae (bottom) genomes. Genes in black highlight
similarities between the genome topologies. Genes in gray highlight possible similarities
in function (e. g., citrate fermentation) but differ between genomes by position or
sequence. White genes are thought to have unrelated functions.

A. succinogenes OAA decarboxylase is also a Na+-pump. Na+-dependent OAA
decarboxylase activity was detected in A. succinogenes extracts (Chapters 3 and 4). This
enzyme is required for growth on citrate by other bacteria (18, 20, 71), but its role during
A. succinogenes mixed acid fermentations is unclear. During citrate fermentation in other
bacterial systems, citrate is taken up by a Na+-dependent transporter, cleaved to OAA and
acetyl-CoA by citrate lyasc, then the Na+-dependent OAA decarboxylase produces
pyruvate for gluconeogenesis. The genes encoding the enzymes responsible for citrate
fermentation in Klebsiella pneumoniae are arranged in an operon, with citrate lyase and
ligase encoded furthest upstream, followed by the Na+-dependent citrate transporter,

OAA decarboxylase, and a two-component regulatory system involved in regulating

citrate metabolism (57) (Fig. 5.2). Interestingly, A. succinogenes OAA decarboxylase is

172

encoded immediately downstream from a Na+ symporter and a citrate ligase gene, but no
two-component regulatory system is encoded downstream of OAA decarboxylase (Fig.
5.2). The citrate lyase genes are encoded on a separate contig. Citrate ligase acetylates
and activates citrate lyase in an ATP-dependent reaction. The citrate ligase near the OAA
decarboxylase operon was the only citrate ligase found in the draft genome sequence.
This gene is in the opposite orientation to that in the K. pneumoniae genome (Fig. 5.2).
Because this citrate ligase is most similar to that from F usobacterium nucleatum (57%
identitical), and because it is only 37% identical to the most closely related
Pasteurellaceae homolog (H. ducreyi), it was likely acquired by horizontal gene transfer.
The A. succinogenes Na+-dependent symporter is not necessarily citrate-speciﬁc since it
is not homologous to that in K. pneumoniae. Still, it belongs to the COG for di- and tri-
carboxylic acid transport, CitT. Interestingly, OAA decarboxylase and citrate lyase
activities were detected in extracts of A. succinogenes that were grown with glucose or
with fumarate and H2 (63). None of these conditions contained citrate and neither citrate
nor isocitrate have been detected in A. succinogenes supematants (data not shown). With
some evidence of horizontal gene transfer and the lack of a two-component regulatory
system, it is possible that OAA decarboxylase and citrate lyase are constitutively
expressed in A. succinogenes. Both OAA decarboxylase and citrate lyases are most
closely related to other Pasteurellaceae genes, suggesting that their activities during
glucose fermentation and fumarate respiration are not unique to A. succinogenes.

A potential alternative route from OAA to fumarate was identiﬁed involving
aspartate aminotransferase and aspartate ammonia-lyasc (Fig. 5.1). While M.

succiniciproducens also encodes aspartate aminotransferase and aspartate ammonia-

173

 

lyasc, knocking out fumC, encoding fumarase, prevented growth in anaerobic conditions.
Furthermore, aspartate ammonia-lyasc activity was barely detectable in cell extracts from
A. succinogenes grown under fermentative conditions (63). These observations suggest
that this alternative route is not active under fermentative conditions.

A. succinogenes has never been reported to produce lactate, unlike its succinate-
producing counterpart M. succiniciproducens. Surprisingly, an A. succinogenes lactate
dehydrogenase was found that is 60% identical to the E. coli enzyme (1.1.1.28) that
couples lactate oxidation to the transport of amino acids and sugars (7). To test whether
A. succinogenes lactate dehydrogenase is active under anaerobic conditions, A.
succinogenes was grown in AM3 with 100 mM NaHCO3, 50 mM glucose, and 25 mM
D,L—lactate. No lactate consumption was observed (data not shown).

As a producer of the dicarboxylate, succinate, it is interesting to note that the A.
succinogenes draft genome sequence may encode at least nine anaerobic dicarboxylate
transporters (Table 5.2). The closest A. succinogenes homolog to E. coli’s aerobic
dicarboxylate transporter, DctA, is Asuc_DRAFT 6 (24% identical). Seven of the
anaerobic transporters are similar to the tripartite ATP-independent periplasmic
transporters (T.C. 2.A.56) encoded by dctPQM (32). These transporters have been
characterized in Rhodobacter capsulatus and Wolinella succinogenes for their roles
during fumarate respiration, where fumarate is transported by proton symport (24, 62).
The other three anaerobic dicarboxylate transporters are related to DcuA, B, and C (T.C.
2.A. l 3). These transporters have been characterized during fumarate respiration by
bacteria such as E. coli and W. succinogenes (59, 62, 75). They operate by exchanging an

intracellular dicarboxylate (e. g., succinate) for an extracellular dicarboxylate (e. g.,

174

 

Asuc_DRAFl‘

Dicarboxylate transporter

 

ORF#
901 Possible TRAP C4-dicarboxylate transporter, large permease component, DctM
902 Conserved hypothetical protein (TRAP, small permease component, DctQ)
903 Possible TRAP C4-dicarboxylate transporter solute receptor, DctP
1074 Possible TRAP C4-dicarboxylate transporter, large permease component, DctM
1075 Conserved hypothetical protein (TRAP, small permease component, DctQ)
1076 Possible TRAP C4-dicarboxylate transporter solute receptor, DctP
170 Possible TRAP C4-dicarboxylate transporter solute receptor, DctP
l7l Possible TRAP C4-dicarboxylate transporter solute receptor, DctP
172 Possible TRAP C4-dicarboxylate transporter, large permease component, DctM
173 Conserved hypothetical protein (TRAP, small permease component, DctQ)
238 Possible TRAP C4—dicarboxylate transporter, large permease component, DctM
239 Possible TRAP C4-dicarboxylate transporter, small permease component, DctQ
240 Possible TRAP C4-dicarboxylate transporter solute receptor, DctP
721 Possible TRAP C4-dicarboxylate transporter solute receptor, DctP
722 Possible TRAP C4-dicarboxylate transporter, small permease component, DctQ
723 Possible TRAP C4-dicarboxylate transporter, large permease component, DctM
733 Possible TRAP C4-dicarboxylate transporter, large permease component, DctM
734 Conserved hypothetical protein (TRAP, small permease component, DctQ)
735 Possible TRAP C4-dicarboxylate transporter solute receptor, DctP
1737 Possible TRAP C4-dicarboxylate transporter, large permease component, DctM
717 Probable anaerobic C4-dicarboxylate transporter DcuB or DcuA
125 Putative anaerobic C4-dicarboxylate transporter DcuB
l66b Possible anaerobic C4-dicarboxylate transporter DcuC

 

Table 5.2. Dicarboxylate transporters identiﬁed in the draft genome sequence.
TRAP: tripartite ATP-independent periplasmic.

fumarate, malate, or aspartate). DcuA and B may also transport Na+ in symport with the

dicarboxylates to avoid dissipating the proton motive force (62). DcuC may have

preferential succinate efflux activity, since a dcuC ' E. coli strain had increased

dicarboxylate exchange and fumarate uptake activities (74). Glucose did not repress dcuC

expression, suggesting that DcuC plays a role in succinate excretion during E. coli mixed

acid fermentations (74). A microarray study examining H. inﬂuenzae gene expression

175

changes that occur as it becomes competent suggested that DcuA and B are important for
Pasteurellaceae fermentation or fumarate respiration (50). In this study, 151 genes
showed > 4-fold increase in transcript levels as H. inﬂeunzae became competent,
including transcripts for the C4 pathway enzymes aspartate ammonia-lyasc, malate
dehydrogenase, fumarase, fumarate reductase, and a single dicarboxylate transporter.
This H. inﬂuenzae transporter is 84% and 44% identical to the DcuB-type transporters,
Asuc_DRAFT 717 and 125, respectively. A. succinogenes dcuA, B, and C are therefore
attractive genes to investigate the importance of dicarboxylate transport during fumarate
respiration and succinate fermentation.

A. succinogenes grows on a variety of industrially-relevant sugars including
glucose, fructose, xylose, arabinose, and sucrose (26). Figure 5.3 and Table 5.3
summarize the sugar uptake and degradation pathways that were identiﬁed in the draft
genome sequence. PTS systems were identiﬁed for glucose, fructose, mannose, and
sorbitol, but biochemical testing is required to conﬁrm their speciﬁcities. A. succinogenes
produces acid from B-glucosides, including amygdalin, aesulin, arbutin, cellobiose,
gentiobiose, and salicin (26). A B-glucoside PTS was identiﬁed but it is not clear whether
this single transporter transports all [3- glucosides. Galactose, maltose, arabinose, ribose,
and xylose are taken up by ATP-dependent transporters. Xylose can also be transported
by a H+-symport mechanism. Htsymport transporters were also identiﬁed for lactate,
gluconate, and idonate. Operons for some transporters (e. g., for arabinose, maltose,
ribose, xylose, and lactose or sucrose) include regulatory proteins, suggesting that uptake
of these sugars is regulated (Table 5.3). Transporters were not found for sucrose,

mannitol, or arabitol, which A. succinogenes can use, nor were some sugar degradation

176

enzymes (Fig. 5.3). These enzymes may be identiﬁed in the complete genome sequence

and some transporters may transport multiple different sugars.

 

 

“US$52?” Function BC. or T.C. # El?“
1833-4, 1667 Putative ATP-dependent galactoside transporter 3.A. 1.2.3 1
1693 Galactokinase 2.7.1.6 2
1694 Galactose-l-phosphate uridylyltransferase 2.7.7.12 3
1229 Probable lactose permease 2.A.1.5.1 4
1228 Possible B-galactosidase 3.2.1.23 5
1230 Probable sucrose operon repressor (Lacl?)
1571.7 Probable a-amylase 3-2-1-1
1578 Probable maltoporin 6
1572-6,9 Putative ATP-dependent maltose transporter 3.A.1.1 .1 7
1568 Amylomaltase 2.4.1.25 8
1569 Probable maltodextn'n phosphorylase 2.4.1.1 9
1877 Possible glucokinase 2.7.1.2 10
1570 Probable maltose regulon transcriptional regulator
1050 Possible L-idonate or gluconate transport protein 2.A.8.1.2 l 1
1368 Possible L-idonate or gluconate transport protein 2.A.8.1.2 12
1048 Probable L-idonate S-dehydrogenase 1.1.1.- 13
1049 Putative gluconate 5-dehydrogenase 1.1.1.69 14
1047 Probable gluconokinase 2.7.1.12 15
936-938 Putative high afﬁnity ATP-dependent L-arabinose transporter 3.A. l .2.2 16
941 Probable L-arabinose isomerase 5.3.1.4 17
940 Conserved hypothetical protein (possible ribulokinase) 2.7. 1.17 18
939 Probable arabinose operon regulatory protein
308-311 Putative high afﬁnity ATP-dependent ribose transporter 3.A.1.2.1 19
312 Probable ribokinase 2.7.1.15 20
313 Probable ribose operon repressor protein
943 Probable low afﬁnity D-xylose-proton symporter 2.A.1.1.3 21
944-946 Probable high afﬁnity ATP-dependent xylose transporter 3.A.1.2.4 22
947 Putative xylose isomerase 5.3.1.5 23
948 Xylulose kinase 2.7.1.17 24
942 Putative xylose operon regulatory protein
409-410, 53 Possible fructose-speciﬁc PTS 4.A.2. l .1 25
54 Probable l-phosphofructokinase 2.7.1.56 26
46-48 Probable glucitol/sorbitol-speciﬁc PTS 4.A.4.1.1 27
44,45 Putative sorbitol-6-phosphate 2-dehydrogenase 1.1.1.140 28
273-275 Putative mannose-speciﬁc PTS 4.A.6.1.1 29
277 Mannose-6-phosphate isomerase, class I 5.3.1.8 30
730 Possible mannitol 2-dehydrogenase 1.1.1.67 31
1503 Probable ATP-dependent fructokinase 2.7.1.4 32
1504 Possible sucrose hydrolase 3.2.1.26 33
876 Probable B-glucoside-speciﬁc PTS, IIABC component 4.A.1.2.2 34
878 Probable 6-phosphoﬂlucosidase 3.2.1.86 35

 

Table 5.3. A. succinogenes ORFs encoding uptake and degradation pathways for

sugars other than glucose. 3 Corresponding reaction in Figure 5.3.

177

 

 

Lac Mal Gte Ido

      
       
             
       
    
   

V
'0'!-
I I I

01. 631:5! Lac Ma. ldo
13
Gal1P —>G‘IP Glc 8MaIGP ete<’— 5K6 24.1w

  

Gal

 

’0: Base \\ 1‘1/7 /5 171

0.. ssp—éf—v esp —> — GtGP TortlusrmE Rlbu

[31 Frc {1 co?

7:?T' R\5P<22R1b [o

:Fsppojm :8

o.- MOH1P/ Xusp !
29

Xylu +2-3— Xyl

Rlb

Xyl

Xyl

 
  
 
  

DHAP + G3P

inner membrane
periplasm

  
 

outer membrane

(% = PTS 8 = ATP-dependent transporter U = Ht-symporter

Figure 5.3. Uptake and degradation pathways for sugars other than glucose based
on genes identiﬁed in the draft genome sequence. Bold arrows, glycolytic or pentose
phosphate pathways as described in Figure 5.1 and Table 5.1; Gray arrows or symbols,
no suitable sequence identiﬁed for function; dotted arrow, no sequence is known
mannitol-l-phosphatase (EC. 3.1.3.22); outer-membrane triangle, maltoporin. Numbers
in italics refer to enzymes or transporters in Table 5.3.

5.4.3 Potential for engineering by natural transformation. Engineering A.
succinogenes into an industrial biocatalyst requires genetic tools. Genetic tools for A.
succinogenes are currently limited to an expression vector and transformation by
electroporation (35). The ability to replace chromosomal DNA segments with engineered

DNA is invaluable for making gene knockouts (e.g., of those encoding enzymes leading

178

 

 

to undesired products) and knock-ins (e. g., of a modiﬁed promoter to decrease or increase

enzyme expression). One possible route for replacing A. succinogenes chromosomal
DNA segments is through natural transformation. Under certain conditions (e. g.,
starvation or in the presence of elevated CAMP levels) many Pasteurellaceae become
naturally competent, and can be transformed at frequencies as high as 10'3 to 10'2 (9, 11,
25, 29, 50, 66, 67). The mechanism behind Pasteurellaceae competence involves a
coordinated response of several proteins that selectively uptake DNA molecules
containing a speciﬁc uptake signal sequence (USS; AAGTGCGGT) (22, 50). If the DNA
taken up is homologous to a section of chromosomal DNA then recombination can occur.
This DNA uptake mechanism works best with linear DNA, making it well suited for
strain engineering, since constructs can be easily generated by PCR and double
recombination events are needed to integrate the linear DNA into the chromosome.
USS-containing DNA is also referred to as ‘self-DNA’ because the USS occurs as
a repeat in Pasteurellaceae genomes. IfA. succinogenes is capable of natural competence
then its genome should contain USS repeats. The draft sequence contains 1,458 USSs,
well within the range found in other Pasteurellaceae (i.e., from 927 in P. multocida to
1,760 in A. actinomycetemcomitans). Like with other Pasteurellaceae USS, the A.
succinogenes 9-nt core sequence is usually preceded by ‘AA’ and followed by an AT-
rich region (Fig. 5.4). One outstanding difference is that 70% of the A. succinogenes USS
9-nt cores are immediately followed by a ‘C’. The frequency of ‘C’ in this position in

other Pasteurellaceae USS is from 27% in H. somnus to 46% in P. multocida.

179

   

< mu

   

    
 

effiw°°CCCCCCT C9

Percent of USS

E5
i
i
1

.v.x;.._m..ﬂli1?ézl-$!
. 'ﬁwihiWWIkmn:1:Y-J
uﬁiﬁiills-zzmwsm.‘

($537

 

12 3 4 5 6 7 8 910111213141516171819202122232425262728.2930
Micleotide position

Figure 5.4. Nucleotide frequencies in 1,454 A. succinogenes uptake signal sequences.

Pasteurellaceae competence involves a speciﬁc set of proteins. A regulon of 25
genes encoding known competence proteins and proteins of unknown function was
recently identiﬁed in a microarray study of H. inﬂuenaze competency (50). The regulon
is regulated by the CAMP receptor protein, which was shown to bind to a newly
discovered consensus sequence associated with the promoters of the regulon (50). The A.
succinogenes draft genome sequence does not encode a CAMP receptor protein but most
of the other competence proteins were identiﬁed (Table 5.4).

The number of USS repeats and competence proteins found in the A.
succinogenes genome sequence suggests that it can be naturally transformed. However, it
should be noted that the level of sequence identity between the H. inﬂuenzae and A.
succinogenes competence proteins is often low. Of the nineteen A. succinogenes
competence proteins ideniﬁed, seven were assigned probable function and six were
assigned possible function (Table 5.4). It is also worth noting that, in a study of A.
actinomycetemcomitans competency, only one of the 17 strains examined could be

naturally transformed under the tested conditions (67).

180

 

 

 

A. succinogenes

 

H’ '31:: Z209 H. inﬂuenzae product description mug-1??? mosléifptt’ifﬁuct
H1006] Recombination protein 336 NA (length)
HIO435 Competence protein B 1218 Putative
H10436 Competence protein D 2125 Possible
H10437 Competence protein C 1279 Possible
H10438 Competence protein B 1278 Possible
HIO439 Competence protein A 1277 Possible
H10365 Hypothetical protein H10365 1009 Function
H10296 Type 4 prepilin-like protein speciﬁc leader peptidase 580 Possible
H10297 Protein transport protein 562 Probable
H10298 Protein transport protein 56] Putative
H10299 Prepilin peptidase-dependent protein D 560 Putative
H1060] DNA transformation protein 1672 Probable
HIO659 Hypothetical protein HIO659 327 NA (length)
H10660 Hypothetical H10660 NA (N 0 hits)
H10938 Hypothetical protein H10938 l7 Probable
H10939 Hypothetical protein H10939 16 Probable
HIO940 Hypothetical protein HIO940 15 Possible
H1094] Hypothetical protein H1094] 20 Probable
H10985 DNA processing chain A 379 Probable
H11117 Competence protein 952 Function

HI] 182/83 DNA ligase 337 Putative
H1163] Hypothetical protein H1163] 1163 NA (length)
H1060] DNA transformation protein Sxy 1672 Probable

 

Table 5.4. A. succinogenes homologs of H. inﬂuenzae competency proteins. a Putative,
60-75% amino acid identity; Probable, 40-59% amino acid identity; Possible, 25-39%
amino acid identity; NA, no suitable homolog was identiﬁed in the A. succinogenes draft
genome sequence either due to insufﬁcient alignment of length, or no hits were retrieved

from the BLAST search.

5.4.4 Potential for non-pathogenicity. The natural Pasteurellaceae ecology is in

association with a host (34). The host is usually mammalian, with some exceptions, such

as P. multocida colonizing birds (34) or even amoebae (30). Most Pasteurellaceae have

been isolated from healthy hosts and are considered to be part of the normal ﬂora.

However, in circumstances such as host stress, many Pasteurellaceae cause disease. As a

result, Pasteurellaceae are generally considered to be opportunistic pathogens. All

18]

 

members of the genus Actinobacillus are automatically classiﬁed as risk group 2 bacterial

agents by the National Institute of Health and require handling under biosafety level 2
guidelines (1). Most Pasteurellaceae are isolated from the respiratory tract and cause
respiratory type ailments (34). Others, though, have been isolated from the oral cavity
(e.g., A. actinomycetemcomitans causing peridontitis [(28)]), the genital tract (e. g., H.
ducreyi causing sexually transmitted chancroid (8)]), and cattle rumen (e. g., A. lignieresii
causing “wooden toungue (47)]”). The two Pasteurellaceae receiving attention as
potential industrial biocatalysts, A. succinogenes and M. succiniciproducens, collectively
referred to as succinogens, were isolated from cow rumen. Virulence is an undesirable
trait for an industrial organism, and the relatedness of A. succinogenes and M.
succiniciproducens to several pathogens cannot be ignored. With no reports of disease
caused by the succinogens, their genome sequences are a convenient and logical starting
point to assess their potential for non-pathogenicity.

Pasteurellaceae are intensively studied due to their effects on human and animal
health and, as a result, many virulence factors are known. Known Pasteurellaceae
virulence genes (~250) were manually compilied and aligned against the of A.
succinogenes and M. succiniciproducens protein databases. Global approaches used to
identify Pasteurellaceae virulence factors, such as microarrays (13, 14, 72) and
signature-tagged transposon mutagenesis (44, 58), show that global metabolic changes
occur (e. g., expression of amino acid transporters, purine and pyrimidine biosynthesis
enzymes, and metabolic enzymes for anaerobic metabolism). While expression of these
genes may affect host health, they are not clear indicators of virulence. These genes were

therefore excluded from the comparison. The comparison included gene products whose

182

 

 

 

only role appears to be in virulence (e. g., toxins) and gene products involved in the
synthesis of cell surface molecules and iron uptake. While most of the A. succinogenes
genome is accounted for in the draft genome sequence, the absence of some genes may

be due to sequence gaps.

Toxins. RTX leukotoxins are produced by many Pasteurellaceae, including A.
pleuropneumoniae (54), A. actinomycetemcomitans (28), M. haemolytica (19), and P.
multocida (55). The rumen bacterium A. lignieresii contains a full RTX toxin operon but
lacks a promoter region to express it (55). The toxins are encoded by four genes: A,
encoding the structural pretoxin; C, encoding the protein that activates the pretoxin; and
B and D, encoding a type I secretion system for the toxin (55). The pretoxin is not
encoded in either succinogen genome sequence, and only possible homologs to the other
proteins are found, likely due to conserved sequences for ATP-dependent transport. A.
actinomycetemcomitans also uses a cytolethal distending toxin that is encoded near a
characteristic virulence-associated region (39). Neither succinogen genome sequence

encodes a cytolethal distending toxin.

Cell surface structures. Virulence factors used by pathogenic Pasteurellaceae include
cell surface structures such as pili, adhesions, lipopolysaccharides (LPS), and capsules.
Even though the succinogens are rumen bacteria, large-scale culturing necessitates an
analysis of the risk from aerosols, particularly since many Pasteurellaceae are respiratory
pathogens. Adherence to respiratory epithelial cells is the ﬁrst stage of colonization by

respiratory Pasteurellaceae pathogens. Adherence requires a number of cell surface

183

It"

'11

 

radar-.1

 

.3

v: u w-QA 9 «‘Hr.)

a.

 

mechanisms (23). H. inﬂuenzae ’s hemagglutinating pili (hifA-E) are not found in the
succinogen genome sequences, with the exceptions of a possible chaperonin-like HifB
component in M. succiniciproducens and possible minor ﬁmbrial subunit HifD precursor
in A. succinogenes. Neither succinogen encodes H. inﬂuenzae ’s surface ﬁbril protein, Hsf
(61). Both succinogenes have putative, probable, and possible matches to all components
of the type IV pilus (pilABCD) involved in H. inﬂuenzae ’s adherence to nasopharyngeal
tissue (27). These pili are also part of the H. inﬂuenzae competence regulon (50). Both
competence and pathogenicity are regulated in part by stress. M. succiniciproducens also
has probable homologs of the A. actinomycetemcomitans pili needed for tight adherence,
while A. succinogenes has possible homologs (48). These pili are widespread in bacteria
and archaea (48). H. inﬂuenzae strains also use high molecular weight (HMW) adhesin
proteins to colonize respiratory tissue (6, 23, 65). Aside from putative Hme homologs
in each genome, most H. inﬂuenzae HMW adhesions do not have signiﬁcant sequence
identity to any succinogen ORF. Each succinogen has several large ORFs that could
encode HMW adhesions, including a possible homolog of the A. actinomycetemcomitans
collagen adhesin, EmaA (42). Neither succinogen encodes a homolog of the H.
inﬂuenzae adherence and penetration protein, Hap (23).

Gram-negative bacteria possess outer membrane LPS, which can act as endotoxin
and play a role in evading the immune system. As expected, the succinogen genome
sequences indicate that they make LPS, and that they are likely capable of LPS phase
variation. Nontypeable H. inﬂuenzae are able to evade host immune defenses by
incorporating host sialic acid and choline into their LPS, thereby mimicking host cell

surfaces (2, 70). Both succinogenes lack the genes for choline incorporation (licABCD),

184

 

_. J

with the exception of a possible licB gene in A. succinogenes. Both succinogens also lack
the genes for sialic acid incorporation (lic3A2, siaB) but they have a possible sialic acid
transporter (siaT).

Pasteurellaceae, such as typeable H. inﬂuenzae, P. multocida, A.
pleuropneumoniae, and M. haemolytica, produce capsule that is important for virulence
(15, 16, 31, 36, 53, 68, 69). Both succinogens have possible homologs to, at most, two of
the four capsule biosynthesis and export proteins, suggesting that they are not capsulated
bacteria. However, nontypeable H. inﬂuenzae are non-capsulated, so capsule is not

required for virulence in all Pasteurellaceae.

Iron uptake mechanisms. While iron acquisition is a well-documented Pasteurellaceae
virulence trait, it is a common trait of most bacteria making it difﬁcult to associate iron
uptake with virulence. Nonetheless, some insight may be gained from the form of iron
transported. For example, transferrin and hemoglobin are proteins that mammalian hosts
use to bind iron. Unlike A. pleuropneumoniae (5, 31), H. inﬂuenzae (27), M. haemolytica
(43), and P. multocida (10), the succinogens have possible homologs to only a few of the

proteins required for transferin or hemoglobulin uptake.

Other virulence proteins. Both succinogens encode GroEL homologs. While GroEL is
involved in bone resorption by A. actinomycetemcomitans (28), it is also a ubiquitous
chaperonin that is important for the proper folding of proteins (17). Both succinogens
have a putative homolog to an A. actinomycetemcomitans immunosuppressive factor (38)

and a putative homolog to the inner membrane protein, ImpA, involved in

185

'11

_=‘:'1r.o

 

"r
”q“?—

autoaggregation (28). Some Pasteurellaceae have urease activity, which is a known
virulence factor of gastroduodenal and urinary tract pathogens (12), but the succinogens
have no urease homologs, and A. succinogenes tested negative for urease activity (26).
There is also no homolog in either succinogen genome sequence to H. inﬂuenzae
immunoglobulin protease, Iga, which cleaves immunoglobulin, helping H. inﬂuenzae
avoid host defenses at mucosal surfaces (49).

Non-pathogenicity cannot be concluded from the analysis of a genome sequence.
However, it is encouraging that the succinogen genome sequences lack a considerable
number of the virulence genes used by their relatives. Most Pasteurellaceae cause
respiratory disease, while succinogens likely exist in the rumen. Therefore, if a
succinogen-caused disease exists, the virulence factors involved could be different from
those found in other Pasteurellaceae. Unfortunately, A. lignieresii, which can also be
isolated from rumen and is the causative agent of ‘wooden-tongue’ disease, is not well-
characterized (55). With the low prevalence of ‘wooden—tongue’ disease and no reports of
disease caused by A. succinogenes or M. succiniciproducens, it is plausible that the
succinogens are simply commensal organisms. Succinogen non-pathogenicity is an
exciting prospect, not just for industrial purposes, but because comparisons of
succinogens with other non-pathogenic (i.e., H. inﬂuenzae KW20) and pathogenic

Pasteurellaceae could lead to a better understanding of Pasteurellaceae virulence.

186

5.5 REFERENCES.

l.

10.

11.

2002. NIH guidelines for research involving recombinant DNA molecules (NIH
guidelines). National Institute of Health.

Allen, S., A. Zaleski, J. W. Johnston, B. W. Gibson, and M. A. Apicella. 2005. Novel
sialic acid transporter of Haemophilus inﬂuenzae. Infect. Immun. 73:5291-5300.

Altschul, S. F., T. L. Madden, A. A. Schaffer, J. Zhang, Z. Zhang, W. Miller, and D.
J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generation of protein
database search programs. Nucleic Acids Res. 25:3389-3402.

Bakkali, M., T. Y. Chen, H. C. Lee, and R. J. Redﬁeld. 2004. Evolutionary stability
of DNA uptake signal sequences in the Pasteurellaceae. Proc. Natl. Acad. Sci. USA
101:4513-4518.

Baltes, N ., I. Hennig-Pauka, and G.-F. Gerlach. 2002. Both transferrin binding
proteins are virulence factors in Actinobacillus pleuropneumoniae serotype 7
infection. FEMS Microbiol. Lett. 209:283-287.

Barenkamp, S. J ., and J. W. St Geme, 3rd. 1996. Identiﬁcation of a second family of
high-molecular-weight adhesion proteins expressed by non-typable Haemophilus
inﬂuenzae. Mol. Microbiol. 19:1215-1223.

Barnes, E. M., Jr., and H. R. Kaback. 1971. Mechanisms of active transport in
isolated membrane vesicles. 1. The site of energy coupling between D-lactic

dehydrogenase and beta-galactoside transport in Escherichia coli membrane vesicles.
J. Biol. Chem. 246:5518-5522.

Bauer, B. A., M. K. Stevens, and E. J. Hansen. 1998. Involvement of the
Haemophilus ducreyi gmhA gene product in lipooligosaccharide expression and
virulence. Infect. Immun. 66:4290-4298.

Bigas, A., M. E. Garrido, A. M. de Rozas, I. Badiola, J. Barbe, and M. Llagostera.
2005. Development of a genetic manipulation system for Haemophilus parasuis. Vet.
Microbiol. 105:223-228.

Bosch, M., M. E. Garrido, M. Llagostera, A. M. Perez De Rozas, I. Badiola, and J.
Barbe. 2002. Characterization of the Pasteurella multocida hgbA gene encoding a
hemoglobin-binding protein. Infect. Immun. 70:5955-5964.

Bosse, J. T., J. H. E. Nash, J. S. Kroll, and P. R. Langford. 2004. Harnessing natural

transformation in Actinobacillus pleuropneumoniae: a simple method for allelic
replacements. FEMS Microbiol. Lett. 233:277-28].

187

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

Bosse, J. T., and J. I. MacInnes. 1997. Genetic and biochemical analyses of
Actinobacillus pleuropneumoniae urease. Infect. Immun. 65:4389-4394.

Boyce, J. B., I. Wilkie, M. Harper, M. L. Paustian, V. Kapur, and B. Adler. 2004.

Genomic-scale analysis of Pasteurella multocida gene expression during growth
within liver tissue of chickens with fowl cholera. Microb. Infect. 6:290-298.

Boyce, J. B., I. Wilkie, M. Harper, M. L. Paustian, V. Kapur, and B. Adler. 2002.
Genomic scale analysis of Pasteurella multocida gene expression during growth
within the natural chicken host. Infect. Immun. 70:6871-6879.

Boyce, J. D., and B. Adler. 2000. The capsule is a virulence determinant in the
pathogenesis of Pasteurella multocida M1404 (B:2). Infect. Immun. 68:3463-3468.

Boyce, J. D., J. Y. Chung, and B. Adler. 2000. Genetic organisation of the capsule
biosynthesis locus of Pasteurella multocida M1404 (B:2). Vet. Microbiol. 72:121-
134.

Chapman, E., G. W. Farr, R. Usaite, K. Furtak, W. A. Fenton, T. K. Chaudhuri, E. R.
Hondorp, R. G. Matthews, S. G. Wolf, J. R. Yates, M. Pypaert, and A. L. Horwich.

2006. Global aggregation of newly translated proteins in an Escherichia coli strain
deﬁcient of the chaperonin GroEL. Proc. Natl. Acad. Sci. U S A 103:15800—15805.

Dahinden, P., Y. Auchli, T. Granjon, M. Taralczak, M. Wild, and P. Dimroth. 2005.
Oxaloacetate decarboxylase of Vibrio cholerae: puriﬁcation, characterization, and
expression of the genes in Escherichia coli. Arch. Microbiol. 183:121-129.

Davies, R. L., T. S. Whittam, and R. K. Selander. 2001. Sequence diversity and
molecular evolution of the leukotoxin (lktA) gene in bovine and ovine strains of
Mannheimia (Pasteurella) haemolytica. J. Bacteriol. 183: 1394-1404.

Dimroth, P. 1980. A new sodium-transport system energized by the decarboxylation
of oxaloacetate. FEBS Lett. 122:234-236.

Donahue, J. L., J. L. Bownas, W. G. Niehaus, and T. J. Larson. 2000. Puriﬁcation and
characterization of gle—encoded fructose 1,6-bisphosphatase, a new enzyme of the
glycerol 3-phosphate regulon of Escherichia coli. J. Bacteriol. 182:5624-5627.

Dubnau, D. 1999. DNA uptake in bacteria. Annu. Rev. Microbiol. 53:217-244.
Ecevit, I. 2., K. W. McCrea, M. M. Pettigrew, A. Sen, C. F. Marrs, and J. R. Gilsdorf.

2004. Prevalence of the hifBC, hmwlA, hmw2A, hmw2C, and hid genes in
Haemophilus inﬂuenzae isolates. J. Clin. Microbiol. 42:3065-3072.

24. Forward, J. A., M. C. Behrendt, N. R. Wybom, R. Cross, and D. J. Kelly. 1997.

TRAP transporters: a new family of periplasmic solute transport systems encoded by

188

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

the dctPQM genes of Rhodobacter capsulatus and by homologs in diverse gram-
negative bacteria. J. Bacteriol. 179:5482-5493.

Gioia, J ., X. Qin, H. Jiang, K. Clinkenbeard, R. Lo, Y. Liu, G. E. Fox, 8.
Yerrapragada, M. P. McLeoad, T. Z. McNeil], L. Hemphill, E. Sodergren, Q. Wang,
D. M. Muzny, F. J. Homsi, G. M. Weinstock, and S. K. Highlander. 2006. The
genome sequence of Mannheimia haemolytica Al: Insights into virulence, natural
competence, and Pasteurellaceae phylogeny. J. Bacteriol. 18827257-7266.

Guettler, M. V., D. Rumler, and M. K. Jain. 1999. Actinobacillus succinogenes sp.
nov., a novel succinic-acid-producing strain from the bovine rumen. Int. J. Syst.
Bacteriol. 49:207-216.

Harrison, A., D. W. Dyer, A. Gillaspy, W. C. Ray, R. Mungur, M. B. Carson, H.
Zhong, J. Gipson, M. Gipson, L. S. Johnson, L. Lewis, L. O. Bakaletz, and R. S.
Munson Jr. 2005. Genomic sequence of an otitis media isolate of nontypeable

Haemophilus inﬂuenzae: Comparative study with H. inﬂuenzae serotype d, strain
KW20. J. Bacteriol. 187:4627-4636.

Henderson, B., S. P. Nair, J. M. Ward, and M. Wilson. 2003. Molecular pathogenicity
of the oral opportunistic pathogen Actinobacillus actinomycemcomitans. Annu. Rev.
Microbiol. 57:29-55.

Herroitt, R. M., E. M. Meyer, and M. Vogt. 1970. Deﬁned nongrowth media for stage
11 development of competence in Haemophilus inﬂuenzae. J. Bacteriol. 101:517-524.

Hundt, M. J ., and C. G. Ruffolo. 2005. Interaction of Pasteurella multocida with free-
living amoebae. Appl. Environ. Microbiol. 71:5485-5464.

Jacques, M. 2004. Surface polysaccharides and iron-uptake systems of Actinobacillus
pleuropneumoniae. Can. J. Vet. Res. 68:81-85.

Kelly, D. J ., and G. H. Thomas. 200]. The tripartite ATP-independent periplasmic
(TRAP) transporters of bacteria and archaea. FEMS Microbiol. Rev. 25:405-424.

Kessler, D., I. Leibrecht, and J. Knappe. 1991. Pyruvate-formate-lyase-deactivase and
acetyl-CoA reductase activities of Escherichia coli reside on a polymeric protein
particle encoded by adhE. FEBS Lett. 281:59-63.

Killian, M., W. Frederiksen, and E. L. Biberstein. 1981. Haemophilus, Pasteurella,~
and Actinobacillus. Academic Press Inc., New York.

Kim, P., M. Laivenieks, J. McKinlay, C. Vieille, and J. G. Zeikus. 2004. Construction
of a shuttle vector for the overexpression of recombinant proteins in Actinobacillus
succinogenes. Plasmid 5 1 : 108-1 15.

189

36.

37.

38.

39.

40.

41.

42.

43.

45.

46.

Kroll, J. S., B. Loynds, L. N. Brophy, and E. R. Moxon. 1990. The bex locus in
encapsulated Haemophilus inﬂuenzae: a chromosomal region involved in capsule
polysaccharide export. Mol. Microbiol. 4: 1853-1862.

Kummerle, N ., H. H. Feucht, and P. M. Kaulfers. 1996. Plasmid-mediated
formaldehyde resistance in Escherichia coli: characterization of resistance gene.
Antimicrob. Agents Chemother. 40:2276-2279.

Kurita-Ochiai, T., and K. Ochiai. 1996. Immunosuppressive factor from

Actinobacillus actinomycetemcomitans down regulates cytokine production. Infect.
Immun. 64:50-54.

Mayer, M. P. A., L. C. Bueno, E. J. Hansen, and J. M. DiRienzo. 1999. Identiﬁcation
of a cytolethal distending toxin gene locus and features of a virulence associated
region in Actinobacillus actinomycetemcomitans. Infect. Immun. 67: 1227- 1237.

McKinlay, J. B., Y. Shachar-Hill, J. G. Zeikus, and C. Vieille. 2006. Determining
Actinobacillus succinogenes metabolic pathways and ﬂuxes by NMR and GC-MS

analyses of l3C-labeled metabolic product isotopomers. Metab. Eng.
doi:10.1016/j.ymben.2006.]0.006.

McKinlay, J. B., J. G. Zeikus, and C. Vieille. 2005. Insights into Actinobacillus
succinogenes fermentative metabolism in a chemically deﬁned growth medium.
Appl. Environ. Microbiol. 71 :6651-6656.

Mintz, K. P. 2004. Identiﬁcation of an extracellular matrix protein adhesin, EmaA,

which mediates the adhesion of Actinobacillus actinomycetemcomitans to collagen.
Microbiol. 150:2677-2688.

Ogunnariwo, J. A., T. K. Woo, R. Y. Lo, G. C. Gonzalez, and A. B. Schryvers. 1997.
Characterization of the Pasteurella haemolytica transferrin receptor genes and the
recombinant receptor proteins. Microb. Pathog. 23:273-284.

. tha, S., M. Sirois, and J. I. Macinnes. 2005. Identiﬁcation of Actinobacillus suis

genes essential for the colonization of the upper respiratory tract of swine. Infect.
Immun. 7 3:7032-7039.

Park, D. H., M. Laivenieks, M. V. Guettler, M. K. Jain, and J. G. Zeikus. 1999.
Microbial utilization of electrically reduced neutral red as the sole electron donor for
growth and metabolite production. Appl. Environ. Microbiol. 65:2912-2917.

Park, D. H., and J. G. Zeikus. 1999. Utilization of electrically reduced neutral red by

Actinobacillus succinogenes: physiological function of neutral red in membrane-
driven fumarate reduction and energy conservation. J. Bacteriol. 181:2403-2410.

190

’1.

 

am”...

 

 

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

Phillips, J. E. 1981. The genus Actinobacillus, p. 1393-1398. In M. P. Starr, H. Stolp,
H. G. Truper, A. Balows, and H. G. Schlegel (ed.), The Prokaryotes, vol. H. Springer-
Verlag, Berlin.

Planet, P. J ., S. C. Kachlany, D. H. Fine, R. DeSalle, and D. H. Figurski. 2003. The
widespread colonization island of Actinobacillus actinomycetemcomitans. Nat. Genet.
34: 193-198.

Plaut, A. G. 1983. The IgAl proteases of pathogenic bacteria. Annu. Rev. Microbiol.
37:603-622.

Redﬁeld, R. J., A. D. Cameron, Q. Qian, J. Hinds, T. R. Ali, J. S. Kroll, and P. R.
Langford. 2005. A novel CRP—dependent regulon controls expression of competence
genes in Haemophilus inﬂuenzae. J. Mol. Biol. 347:735-747.

Samuelov, N. S., R. Datta, M. K. Jain, and J. G. Zeikus. 1999. Whey fermentation by
Anaerobiospirillum succiniciproducens for production of a succinate-based animal
feed additive. Appl. Environ. Microbiol. 65 :2260-2263.

Samuelov, N. S., R. Lamed, S. Lowe, and J. G. Zeikus. 1991. Influence, of CO2-HCO3
levels and pH on growth, succinate production, and enzyme activities of
Anaerobiospirillum succiniciproducens. Appl. Environ. Microbiol. 57:3013-3019.

Satola, S. W., P. L. Schirmer, and M. M. Farley. 2003. Complete sequence of the cap

locus of Haemophilus inﬂuenzae serotype b and nonencapsulated b capsule-negative
variants. Infect. Immun. 71:3639-3644.

Schaller, A., R. Kuhn, P. Kuhnert, J. Nicolet, T. J. Anderson, J. I. Maclnnes, R. P. A.
M. Segers, and J. Frey. 1999. Characterization of apleA, a new RTX determinant of
Actinobacillus pleuropneumoniae. Microbiol. 145:2105-21 16.

Schaller, A., P. Kuhnert, V. A. de la Puente-Redondo, J. N icolet, and J. Frey. 2000.
Apx toxins in Pasteurellaceae species from animals. Vet. Microbiol. 74:365-376.

Scheiﬁnger, C. C., and M. J. Wolin. 1973. Propionate formation from cellulose and
soluble sugars by combined cultures of Bacteroides succinogenes and Selenomonas
ruminantium. Appl. Environ. Microbiol. 26:789-795.

Schneider, K., C. N. Kastner, M. Meyer, M. Wessel, P. Dimroth, and M. Bott. 2002.
Identiﬁcation of a gene cluster in Klebsiella pneumoniae which includes citX, a gene

required for biosynthesis of the citrate lyase prosthetic group. J. Bacteriol. 184:2439-
2446.

Sheehan, B. J ., J. T. Bosse, A. J. Beddek, A. N. Rycroft, J. S. Kroll, and P. R.

Langford. 2003. Identiﬁcation of Actinobacillus pleuropneumoniae genes important
for survival during infection in its natural host. Infect. Immun. 71:3960-3970.

191

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

69.

Six, 8., S. C. Andrews, G. Unden, and J. R. Guest. 1994. Escherichia coli possesses
two homologous anaerobic C4-dicarboxylate membrane transporters (DcuA and

DcuB) distinct from the aerobic dicarboxylate transport system (Dct). J. Bacteriol.
176:6470-6478.

Spencer, M. B., and J. R. Guest. 1985. Transcription analysis of the sucAB, aceEF
and lpd genes of Escherichia coli. Mol. Gen. Genet. 200: 145-154.

St Geme, J. W., 3rd, D. Cutter, and S. J. Barenkamp. 1996. Characterization of the

genetic locus encoding Haemophilus inﬂuenzae type b surface ﬁbrils. J. Bacteriol.
17826281-6287.

Ullmann, R., R. Gross, J. Simon, G. Unden, and A. Kroger. 2000. Transport of C4-
dicarboxylates in Wolinella succinogenes. J. Bacteriol. 182:5757-5764.

van der Werf, M. J., M. V. Guettler, M. K. Jain, and J. G. Zeikus. 1997.
Environmental and physiological factors affecting the succinate product ratio during
carbohydrate fermentation by Actinobacillus sp. 130Z. Arch. Microbiol. 167:332-342.

van Gylswyk, N. O. 1995. Succiniclasticum ruminis gen. nov., sp. nov., a ruminal
bacterium convering succinate to propionate as the sole energy-yielding mechanism.
Int. J. Syst. Bacteriol. 45:297-300.

van Schilfgaarde, M., P. van Ulsen, P. Eijk, M. Brand, M. Stam, J. Kouame, L. van
Alphen, and J. Dankert. 2000. Characterization of adherence of nontypeable
Haemophilus inﬂuenzae to human epithelial cells. Infect. Immun. 68:4658-4665.

Wang, Y., J. Orvis, D. Dyer, and C. Chen. 2006. Genomic distribution and functions
of uptake signal sequences in Actinobacillus actinomycetemcomitans. Microbiol.
152:3319-3325.

Wang, Y., S.D. Goodman, R.J. Redﬁeld, and C. Chen. 2002. Natural transformation
and DNA uptake signal sequences in Actinobacillus actinomycetemcomitans. J.
Bacteriol. 184:3442-3449.

Ward, C. K., and T. J. Inzana. 1997. Identiﬁcation and characterization of a DNA
region involved in the export of capsular polysaccharide by Actinobacillus
pleuropneumoniae serotype 5a. Infect. Immun. 65:2491-2496.

Ward, C. K., M. L. Lawrence, H. P. Veit, and T. J. Inzana. 1998. Cloning and
mutagenesis of a serotype-speciﬁc DNA region involved in encapsulation and
virulence of Actinobacillus pleuropneumoniae serotype 5a: concomitant expression of

serotype 5a and 1 capsular polysaccharides in recombinant A. pleuropneumoniae
serotype 1. Infect. Immun. 66:3326-3336.

192

'11

r“:

 

70.

71.

72.

73.

74.

75.

West-Bamette, S., A. Rockel, and W. E. Swords. 2006. Bioﬁlm growth increases
phosphorylcholine content and decreases potency of nontypable Haemophilus
inﬂuenzae endotoxins. Infect. Immun. 74: 1828-1836.

Woehlke, G., and P. Dimroth. 1994. Anaerobic growth of Salmonella typhimurium on
L(+)- and D(-)-tartrate involves an oxaloacetate decarboxylase Na“ pump. Arch.
Microbiol. 162:233-237.

Wong, S. M., and B. J. Akerley. 2005. Environmental and genetic regulation of the
phosphorylcholine epitope of Haemophilus inﬂuenzae lipooligosaccharide. Mol.
Microbiol. 55:724-738.

Zeikus, J. G., M. K. Jain, and P. Elankovan. 1999. Biotechnology of succinic acid
production and markets for derived industrial products. Appl. Environ. Microbiol.
51:545-552.

Zientz, E., I. G. Janausch, S. Six, and G. Unden. 1999. Functioning of DcuC as the
C4-dicarboxylate carrier during glucose fermentation by Escherichia coli. J. Bacteriol.
181:3716-3720.

Zientz, B., S. Six, and G. Unden. 1996. Identiﬁcation of a third secondary carrier
(DcuC) for anaerobic G.,-dicarboxylate transport in Escherichia coli: roles of the three
Dcu carriers in uptake and exchange. J. Bacteriol. 178:7241-7247.

193

 

Chapter 6.

CONCLUSIONS and FUTURE DIRECTIONS.

6.1 The current understanding of A. succinogenes fermentative metabolism. A.
succinogenes produces some of the highest succinate titers ever reported, making it a
promising biocatalyst for industrial succinate production (3, 4). However, in addition to
producing succinate, A. succinogenes also produces unwanted forrnate, acetate, and
ethanol. While A. succinogenes produces more succinate with increasing CO2 or
reductant (i.e., H2 or a more-reduced carbon source than glucose) concentrations, forrnate
and acetate are always produced as major fermentation products. Therefore, A.
succinogenes must be engineered to increase ﬂux to succinate and decrease ﬂux to
alternative fermentation products.

Metabolic engineering strategies are unlikely to succeed without an understanding
of the metabolism under study. A. succinogenes metabolism was ﬁrst characterized by in
vitro enzyme assays and fermentation balances (5, 19). These data indicated that A.
succinogenes catabolizes glucose to phosphoenolpyruvate (PEP) by glycolysis and the
oxidative pentose phosphate pathway (OPPP) (19). Metabolism was then thought to split
at PEP to form the acetate-, forrnate-, and ethanol-producing C3 pathway, and the
succinate-producing C4 pathway. Succinate production from PEP involves one CO2
ﬁxation step by PEP carboxykinase (PEPCK), and two reduction steps by malate
dehydrogenase and fumarate reductase (Chapter 3, Fig.1). The positive effects of

increased CO2 and reductant concentrations on succinate production were therefore

194

thought to be due to the increased availability of these substrates for PEPCK and the C4
pathway dehydrogenases (19).

Metabolic engineering also requires genetic tools. An A. succinogenes-
Escherichia coli shuttle vector was developed to overexpress proteins in A. succinogenes
(10). A. succinogenes PEPCK was overexpressed and was expected to increase succinate
production because PEPCK is the CO2-ﬁxing, ﬁrst reaction of the C4 pathway. E. coli
NADP-dependent and Bacillus subtilis NAD-dependent malic enzymes were also
overexpressed, and were expected to increase succinate production through pyruvate
(Pyr) carboxylation, as they did for E. coli (8, 17). In each case, enzyme activity levels
increased 2 — 5 fold, but fermentation balances were unaffected (Kim et al., unpublished
data).

The inability to affect A. succinogenes fermentation balances by overexpressin g
PEPCK and malic enzymes indicated that there was insufﬁcient knowledge on A.
succingenes metabolism to predict the metabolic effects of genetic manipulations. In fact,
several enzyme activities detected in A. succinogenes cell extracts suggested a more
complex metabolism than a simple branched fermentation (19): (i) high citrate lyase
activity levels suggested that TCA-cycle associated ﬂuxes could bridge the C3 and C4
pathways; (ii) low Entner-Doudoroff pathway enzyme activity levels suggested an
alternative to the pentose phosphate pathway for N ADPH generation; (iii) low isocitrate
lyase activity levels suggested that the glyoxylate pathway could bridge the C3 and C4
pathways; and (iv) oxaloacetate (OAA) decarboxylase and malic enzyme activities

suggested shunts from the C4 pathway to the C3 pathway.

195

l3C-metabolic ﬂux analysis (MFA) methods were chosen to obtain a detailed
understanding of A. succinogenes fermentative metabolism. Unlike in vitro enzyme
activity levels, which do not necessarily correlate with in vivo ﬂuxes, results from MFA
can provide a global picture of active pathways and the ﬂuxes through these pathways.
Further insight into the roles of metabolic ﬂuxes can be obtained by repeating MFA
experiments under various growth conditions. MFA involves the analysis of '3 C
incorporated into metabolic products, with most information usually obtained from
labeling patterns in proteinaceous amino acids. To ensure that A. succinogenes proteins
are synthesized from l3C-labeled substrate, a chemically deﬁned growth medium had to
be developed.

In Chapter 2, A. succinogenes fermentations were characterized in a chemically
deﬁned growth medium, called AM3, and compared to those in a comlex medium.
Succinate yields were similar in each medium, but Cg-pathway product yields were
higher in the complex medium. These high C3-pathway product yields may be due to
decreased pyruvate and acetyl-CoA requirements for biosynthesis, since biosynthetic
precursors are abundant in yeast extract. More pyruvate and acetyl-CoA are used for
biosynthesis than any other intermediates during growth in AM3 (Chapter 3). The ratio of
succinate:(acetate plus ethanol) increased with the NaHC03 concentration between 5 mM
and 75 mM N aHCO3 in AM3, indicating that the NaHC03 effects observed in complex
media are also observed in AM3. Fermentation balances were similar between 75 and
150 mM NaHCO3 in AM3, deﬁning an upper range for the effects of NaHCO3 on the
fermentation balance. AM3 contained Cys, Met, and Glu, which were required for

growth. Aspartate: glutamate transaminase activity was previously detected in A.

196

succinogenes extracts (19), suggesting that A. succinogenes could synthesize Glu from a-
ketoglutarate (uKG). aKG supported growth in place of Glu when NH.,Cl or aspartate
was supplied as a nitrogen source. Therefore, Glu-auxotrophy was due to an inability to
synthesize aKG from glucose. This inability indicated that at least one TCA cycle
activity and one reductive TCA-cycle activity are missing or inactive under fermentative
conditions in AM3. Thus, while citrate lyasc is detected in A. succinogenes cell extracts
(19), it is not involved in bridging the C3 and C4 pathways by known TCA-cycle
associated reactions.

Chapter 3 described the ﬁrst 13C-labeling study of A. succinogenes metabolism, in
which A. succinogenes was fed [1-‘3C] glucose in batch AM3 fermentations with 150 mM
N aHCO3 to determine active (3, 13)metabolic pathways and to quantify metabolic ﬂuxes.
Pseudo-metabolic steady state was conﬁrmed by comparing extracellular ﬂuxes and
isotopic enrichments at different cell densities. When Entner-Doudoroff and glyoxylate
pathways were included in metabolic models, they were shown to carry no ﬂux. The
OPPP ﬂux was only 5% that of the glucose uptake rate, not nearly enough to meet
anabolic NADPH demands. More acetate and ethanol were produced than forrnate,
indicating formate dehydrogenase (ForDH) and/or Pyr dehydrogenase (PyrDH) activity.
Both activities were detected in vitro and the ﬂux through these enzymes contributed to a
net N ADH ﬂux. Transhydrogenase could convert this net NADH ﬂux to NADPH, and in
vitro transhydrogenase activity was detected. Pyruvate dehydrogenase activity may also
explain how A. succinogenes mutants with defective pyruvate forrnate lyase activity (3,
13) produce essential acetyl-CoA. Another potential NADPH source was malic enzyme.

The isotopomer data indicated a signiﬁcant C4-decarboxylating ﬂux forming a shunt from

197

the C4 to the C3 pathways. This ﬂux was probably due to both malic enzyme and OAA
decarboxylase, and both activities were detected in cell extracts. Since PEPCK can feed
both the C4 and C3 pathways in the presence of the C4-decarboxylating ﬂux,
overexpressing PEPCK alone is likely not enough to increase ﬂux to succinate. OAA
decarboxylase activity was an intriguing ﬁnding as its sequence and Na+-dependent
activity indicated that it was also a Na+-pump.

In Chapter 4, an informative substrate isotopomer mixture composed of [1-
13 C] glucose, [U-13C] glucose, and unlabeled NaHCO3, was used for a series of l3C-MFA
studies. This isotopomer mixture revealed that there is not only a C4-decarboxylating
ﬂux, but a Pyr-carboxylating ﬂux as well. Exchange ﬂux was also revealed in the C3
pathway, through pyruvate formate-lyase (PFL), ForDH and possibly through PyrDH.
Chapter 4 also provided further insight into the roles of the C4-decarboxylating ﬂux and
the C3 pathway dehydrogenases by examining the ﬂux changes that occur in response to
different Nal-ICO3 and H2 concentrations. While the environmental perturbations affected
the fermentation balance, the growth rate and biomass yield were not affected, indicating
that the resulting ﬂux changes were to maintain growth parameters. The results indicated
that CO2 concentration does not necessarily affect PEPCK ﬂux as was previously
hypothesized. Instead, increased N aHCO3 concentration decreased the ﬂux shunted from
the C4 to the C3 pathway and increased the ﬂux in the opposite direction. With more ﬂux
committed to the C4 pathway, more succinate was excreted. To meet the reductant
demands of the increased ﬂux to succinate, C3 pathway dehydrogenases responded by
decreasing the NADH-oxidizing ﬂux to ethanol and increasing the NADH-producing ﬂux

through ForDH and/or PyrDH. Supplying H2 partially alleviated the need for NADH-

198

production by the C3 pathway, as indicated by a slight decrease in the total C3 pathway
ﬂux, and a larger decrease in the ForDH and/or PyrDH ﬂux.

Chapter 5 described industrially relevant features of the A. succinogenes draft
genome sequence. The metabolic enzymes identiﬁed in the genome sequence generally
support the ﬁndings of the metabolic ﬂux analyses. Dicarboxylate transporters were
identiﬁed that could be involved in succinate excretion. Numerous transporters of
industrially relevant sugars were also identiﬁed. There is some evidence that A.
succinogenes is, or was, capable of natural transformation, as indicated by a nearly
complete set of competence proteins and 1,458 DNA uptake signal sequences. The A.
succinogenes genome sequence does not encode many of the virulence factors used by its
pathogenic relatives.

The results in this dissertation have re-shaped our understanding of A.
succinogenes fermentative metabolism. Instead of a simple branched fermentation, ﬂux is
distributed between the C3 and C4 pathways at four metabolic nodes: PEP, Pyr, OAA, and
malate, due to the activities of OAA decarboxylase and malic enzyme. NADPH is
produced by C3 pathway dehydrogenases coupled with transhydrogenase, and possibly by
malic enzyme, rather than by the OPPP. The mechanisms by which NaHCO3 and H2
affect A. succinogenes fermentation balances are not limited to their roles as substrates
for succinate production in the C4 pathway. NaHCO3 concentration may affect the
amount of ﬂux shunted between the C4 and the C3 pathways. By changing the metabolic
reductant demands, NaHCO3 also affects the distribution of ﬂuxes within the C3 pathway.
H2 also affects C3 pathway ﬂuxes by serving as an alternative source of reductant and

alleviating the need for reductant generation by ForDH and/or PyrDH.

199

'lI

 

 

 

6.2 Future research and development of A. succinogenes metabolism. The described
discoveries have raised important considerations for the metabolic engineering of A.
succinogenes and important questions about A. succinogenes metabolism. The following
sections address the future work needed to understand and engineer A. succinogenes

fermentative metabolism.

6.2.1 Oxaloacetate decarboxylase and malic enzyme. OAA decarboxylase and malic
enzyme are shunts between the C4 and the C3 pathways. Since these enzymes can divert
ﬂux away from succinate, it is desirable to decrease the ﬂux through these enzymes from
OAA and malate to Pyr). However, each enzyme potentially serves an important
physiological role, and could affect the outcome of engineering strategies if not taken into
accout.

As a Na+-pump, OAA decarboxylase may be important for maintaining a N a+
gradient used for cell processes such dicarboxylate transport (e. g., glutamate and
succinate) (6, 18). Furthermore, OAA decarboxylase ﬂux could be important for
maintaining osmotic balance. In AM3, with 150 mM NaHCO3, the total Na“
concentration is ~230 mM. If OAA decarboxylase functions to maintain osmotic balance,
then decreasing the N a+ concentration in the medium could result in decreased OAA
decarboxylase ﬂux. High N a+ concentrations are also harmful to industrial bioreactors.
To test whether OAA decarboxylase ﬂux responds to Na+ concentrations, a low Na+
medium must ﬁrst be developed. In AM3, the phosphate buffer (NaH2PO4) and NaHCO3

are the main Na+ sources. Using K+ media components is not suitable because A.

200

 

 

 

succinogenes growth rates decrease at a K+ concentration above 75 mM. A suitable
alternative might be to use poorly soluble MgCO3 with only enough phosphate buffer to
provide adequate PO42' for growth. A. succinogenes can grow and produce high succinate
titers with MgCO3 concentrations as high as 80 g/l (~950 mM if all were dissolved) (4).
Acid produced during the fermentation dissolves more MgCO3 and moderates the pH
decline. The main disadvantage to using MgCO3 is that growth measurements are less
convenient since they can no longer be performed by optical density. Samples would
have to be taken during growth to determine biomass concentrations by protein assays,
and taking multiple samples necessitates larger culture volumes. Preliminary experiments
would involve culturing A. succinogenes in this medium with different NaCl (treatment)
and KO (control) concentrations. The effects of cation concentrations on the
succinate:(acetate + ethanol) ratio would then be determined by HPLC, and OAA
decarboxylase activity would be quantiﬁed in cell extracts from each condition. l3C-
labeling experiments could be used to determine whether the C4-decarboxylating ﬂux
decreases in response to a decrease in Na+ concentrations. However, even in the absence
of H2, this ﬂux is not well determined and changes in OAA decarboxylase ﬂux may not
be observed against a high malic enzyme ﬂux. Ultimately, it would be desirable to knock
out OAA decarboxylase and determine whether the mutant strain has a decreased N a+
tolerance. In the absence of genetic tools, the importance of OAA decarboxylase for Na"
tolerance could potentially be assessed by using limiting amounts of biotin. OAA
decarboxylase is a biotin-containing enzyme. Biotin is also required for essential
enzymes such as acetyl-CoA carboxylase for lipid synthesis, and A. succinogenes may

require biotin (biotin is supplied in the AM3 vitamin mix). If A. succinogenes requires

20]

biotin, and if OAA decarboxylase is responsible for Na+ tolerance, then A. succinogenes
should grow to higher ﬁnal ODs at low Na+ concentrations than at high Na+
concentrations when biotin is limiting. These experiments should show whether OAA
decarboxylase and the Na+ concentration are important for distributing ﬂux between
succinate and alternative products. These experiments should also establish a Na"
concentration for maximum succinate production without adversely affecting cell growth.
Malic enzyme was hypothesized to be important for NADPH production based on
the C4-decarboxylating ﬂuxes observed. However, its contribution to NADPH production
could not be determined for several reasons: (i), C4 decarboxylating ﬂux conﬁdence
intervals were large; (ii) the individual contributions of OAA decarboxylase and malic
enzyme to the C4 decarboxylating ﬂux is unknown; and (iii) the NADPH requirement
could be satisﬁed entirely by OPPP ﬂux plus ForDH/PyrDH ﬂux coupled with
transhydrogenase. Since pyruvate carboxylating ﬂux increased and the C4dec ﬂux
decreased in conditions with high reductant demands (i.e., high succinate production rates
in the presence of 100 mM NaHCO3; Chapter 4), transhydrogenase is likely more
important than malic enzyme for NADPH production. If malic enzyme is important for
N ADPH production, then inactivating malic enzyme should be compensated for by other
ﬂuxes (i.e., increased PyrDH/ForDH or OPPP ﬂuxes). Knocking out OAA decarboxylase
or malic enzyme individually would also be useful for determining the contribution of
each enzyme to the total C4-decarboxylating ﬂux in a 13C-labeling experiment. It is
possible that malic enzyme carries a net pyruvate carboxylating ﬂux under the high CO2
concentrations used in this dissertation. Malic enzyme carboxylates pyruvate to malate in

succinate producing E. coli strains overexpressing malic enzyme (8, 17). With either

202

enzyme inactivated, in vivo exchange ﬂuxes for each enzyme could be determined with
reasonable conﬁdence (Chapter 4). With both C4-decarboxylating enzymes inactivated,
the in vivo contribution of enzymes like PEPCK and pyruvate kinase to the total C4-
decarboxylating ﬂux could be determined. These enzymes can have OAA decarboxylase

activity, as discussed in Chapter 3.

6.2.2 The effects of CO2 concentration on PEPCK and C4-decarboxylating ﬂuxes.
The prevailing hypothesis for how CO2 concentrations affect A. succinogenes succinate
production is through its role as a substrate for PEPCK. The results in Chapter 4 do not
show a signiﬁcant difference in PEPCK ﬂux at 25 and 100 mM NaHCO3, but it could not
be concluded that the N aHCO3 concentration does not affect PEPCK ﬂux due to large
conﬁdence intervals. The results in Chapter 4 did indicate that the NaHCO3 concentration
affects the amount of ﬂux shunted to the C3-pathway. These results provide an alternative
(or an additional) hypothesis for how CO2 concentration affects A. succinogenes
fermentation balances. OAA decarboxylase and malic enzyme knockouts would certainly
be useful for determining the extent to which CO2 concentration affects fermentation
balances in the absence of one or both C4-decarboxylating activities. The mechanisms by
which CO2 affects ﬂux distributions could also be addressed without genetic tools. When
A. succinogenes was grown in anaerobic respiration conditions with fumarate and H2,
signiﬁcant forrnate and acetate were still produced (19). In vitro OAA decarboxylase and
malic enzyme activities were also detected, leading to the hypothesis that these enzymes
divert ﬂux to the C3 pathway during fumarate respiration as well (19). During fumarate

respiration, PEPCK should carry gluconeogenic ﬂux, so the succinate formation rate

203

 

11

3
:1
Ii

wq'w—v m~ ”a"

 

fum ~.-. 4.

 

should not depend on PEP-carboxylating PEPCK ﬂux. A decrease in forrnate and acetate
yields in response to high NaHCO3 concentrations during fumarate respiration would
support the hypothesis that NaHCO3 concentrations affect the amount of ﬂux shunted
through malic enzyme and OAA decarboxylase. The main drawback to this experiment is
that the ﬁndings may not apply to fermentative conditions.

A potential experiment for testing the effect of NaHCO3 concentration on PEPCK
ﬂux would be to grow A. succinogenes on fructose at different NaHCO3 concentrations.
This experiment operates on the assumption that fructose is transported only by a
PEszructose phosphotransferase system (PT S). If high NaHCO3 concentrations cause
PEPCK ﬂux to increase, then PEPCK could potentially compete with the PTS for PEP,
resulting in a decreased growth rate. Alternatively, PTS ﬂux could increase in response to
the increased PEPCK ﬂux, resulting in an increased growth rate. Using l3C-fructose
isotopomers would be useful for this experiment to conﬁrm that the observed effects are
related to PEPCK ﬂux changes. Characterizing metabolic ﬂuxes on fructose is also
industrially relevant since fructose is a highly abundant plant sugar, especially in

sugarcane, which is a feedstock for industrial fermentations in tropical countries.

6.2.3 Pyruvate dehydrogenase and formate dehydrogenase. MFA experiments
indicated that PyrDH and/or ForDH ﬂuxes were important for generating reducing power
in the absence of H2. However, the ﬂuxes through each enzyme could not be
distinguished. Although the ﬁnal engineering goal is to decrease the ﬂux through both
enzymes, being able to quantify ﬂuxes through the individual enzymes is of interest to
the study of microbial metabolism. Chapter 4 presented methods for quantifying C3

pathway exchanges ﬂuxes. PyrDH is generally considered to be a unidirectional enzyme,

204

If

'11

 

,
.
EV -
.
i-aavw-‘xi --.- .

 

but the high CO2 environment used to grow A. succinogenes may inﬂuence PyrDH
reversibility. Knocking out PFL would prevent forrnate production, thereby preventing
ForDH ﬂuxes, and leaving PyrDH as the only enzyme activity connecting Pyr and acetyl-
CoA. The methods in Chapter 4 could then be applied to the mutant strain to determine

the extent to which N aHCO3 concentration affects PyrDH exchange ﬂux, if at all.

6.2.4 The C3 pathway. The C3 pathway has three important functions: (i) synthesis of
Pyr and acetyl-CoA for biosynthesis; (ii) ATP production (by acetate kinase); and (iii)
NADH production for succinate production and biosynthesis (or N ADH oxidation at low
CO2 concentrations). Knowing the C3 pathway functions is important for designing
effective metabolic engineering strategies to increase ﬂux to succinate, for example by
forcing A. succinogenes to fulﬁll the C3 pathway functions by other means. Since
pyruvate and acetyl-CoA synthesis is useful to avoid adding potentially expensive
alternatives, engineering efforts should focus on generating ATP and/or N ADH by other
means. For example, supplying H2 partially alleviated the need for reductant generation
in the C3 pathway (Chapter 4). Supplying H2 also decreased fumarate excretion rate,
suggesting that fumarate reductase ﬂux is limited by reductant availability. Achieving a
homosuccinate fermentation requires the addition of extra reductant (Chapter 1). H2 may
be a suitable source of reductant for industrial succinate production provided that it can
be obtained or produced inexpensively (e. g., from phostosynthetic bacteria). C3 pathway
ﬂux could decrease and ﬂux to succinate could increase by increasing the rate of H2
oxidation. Increased H2 oxidation could be achieved to some extent by using reactors that

can increase the dissolved H2 concentration. However, it may be more beneﬁcial to force

205

"t‘

 

 

 

high H2 oxidation rates in A. succinogenes. H2 oxidation in A. succinogenes is coupled to
quinone reduction (14). By overexpressing hydrogenase in A. succinogenes, high H2
oxidation rates could be achieved and would maintain a reduced quinone pool leading to
increased succinate production rates. Repeated culturing of hydrogenase-overexpressing
A. succinogenes strains, may lead to mutants with an increased ability to couple quinone
oxidation to succinate production, since these mutants would have the energetic
advantage of increased ATP formation rates by ATPase utilizing the W gradient
generated by quinone oxidation. Repeated culturing may also lead to mutants with
increased ﬂux to succinate through the C4 pathway by overcoming allosteric regulatory
mechanisms such as ATP concentration inhibiting PEPCK activity. Sequencing enzymes
such as PEPCK in evolved strains would then provide further insights into the amino acid
residues important for allosteric regulation. The energetic advantage of increasing ﬂux to
succinate to compensate for the increased H2 oxidation rates may be enough to
signiﬁcantly favor C4 pathway ﬂux over C3 pathway ﬂux without gene knockouts. High
H2 oxidation rates could be complemented by overexpressing NADH oxidase to further
reduce the quinone pool. The increased NADH oxidizing ﬂux may stimulate higher
glycolytic ﬂux to compensate, resulting in a higher glucose consumption rate, and a
higher succinate formation rate. Since ethanol production relies on NADH oxidation
rather than quinone oxidation, succinate production would be favored over ethanol
production in hydrogenase-overexpressing strains. However, to ensure that NADH from
glycolytic ﬂux is used for succinate production, and converted to NADPH for biomass, it
may be necessary to knockout acetaldehyde/alchohol dehydrogenase. Knocking out

acetate kinase would also force the bacterium to rely on the C4 pathway for ATP

206

production. Potentially, with the C3 pathway terminating at Acetyl-CoA, there would be
only enough C3 pathway ﬂux to supply pyruvate and acetyl-CoA for biosynthesis.
However, inactivating central C3 pathway enzymes may also be useful. For example,
inactivating PFL has already been shown to increase succinate yields for A. succinogenes
(3) and result in decreased growth rates, perhaps due to decreased ﬂux to acetyl-CoA
with pyruvate dehydrogenase acting alone. These decreased growth rates may be useful
for diverting more carbon to succinate rather than biomass.

The engineering strategies described above illustrate how knowledge of pathway
function can inﬂuence engineering strategies. However, strategies made with the current
understanding of A. succinogens metabolism could be ineffective since the physiological
importance of the C4-decarboxylating ﬂuxes is unknown. OAA decarboxylase ﬂux may
be inﬂuenced entirely by Na+ concentrations and not by reductant and ATP demands.
Malic enzyme may even carry a net pyruvate carboxylating ﬂux that can’t be observed
against a high OAA decarboxylase ﬂux. The strategies described above could result in
excess pyruvate as is observed for A. succinogenes mutants with inactive PFL (3). A
better understanding of how CO2 concentrations inﬂuence the direction of malic enzyme
could therefore be useful to divert excess pyruvate to succinate. It is currently unknown
which of malic enzyme, OAA decarboxylase, PTS, or pyruvate kinase carries the most
ﬂux to pyruvate. Thus, the beneﬁts of knocking out any single one of these enzymes for
succinate production cannot be reasonably predicted. Determining the functions of the
C4-decarboxylating ﬂuxes and the individual ﬂuxes leading into pyruvate will

undoubtedly be useful for developing effective metabolic engineering strategies.

207

 

'11-

 

 

 

6.2.5 Importance of the A. succinogenes genome sequence. While MFA measures
ﬂuxes through a metabolic network, it does not necessarily identify the enzymes
responsible for these ﬂuxes. In fact, the accuracy of MFA is dependent on the accuracy of
the metabolic model. A genome sequence is an excellent resource by which to shape and
constrain a metabolic network. A genomic sequence therefore complements MFA by
indicating which enzymes may be responsible for observed ﬂuxes, by suggesting
pathways that could generate unusual labeling patterns, and by decreasing the number of
labeling experiments needed to address the activity of certain pathways.

The analysis of sugar uptake and degradation pathways in Chapter 5 points to
potentially important targets for metabolic engineering. Xylose and arabinose are
abundant plant pentoses and as such are industrially relevant carbon sources. A.
succinogenes encodes two xylose transporters: (i) a low afﬁnity H+-symporter, and (ii) a
high afﬁnity, ATP-dependent transporter. To conserve energy, it may be beneﬁcial to
ensure xylose transport through the H+-symporter by knocking out the ATP-dependent
transporter. An E. coli PFL knockout strain did not grow fermentatively on xylose by
using an ATP-dependent transporter, but grew when a III-symporter was expressed (7).
Additionally, only an ATP-dependent arabinose transporter has been identiﬁed in the A.
succinogenes draft genome sequence. If an arabinose H+-symporter is not found in the
complete genome sequence, it may be important to introduce a gene encoding one.
Constitutively expressing the xylose H+-symporter could also lead to arabinose uptake
since pentose H+-symporters can have low substrate speciﬁcity (7).

To produce succinate industrially using A. succinogenes in an inexpensive

chemically deﬁned medium, cysteine auxotrophy must be eliminated. While glutamate

208

11*

FT”

 

I 1.--";

 

 

and methionine are relatively inexpensive, being produced from large scale
fermentations, cysteine is expensive. When the A. succinogenes genome sequence is
complete, it will be important to determine which cysteine pathway enzymes are not
encoded, and introduce them into the genome. If cysteine, or methionine, auxotrophy can
be eliminated by introducing a single enzyme activity, the genes encoding these enzymes
could also be useful as alternative A. succinogenes selective markers to antibiotic
resistance cassettes.

The A. succinogenes genome sequence opens the door to using functional
genomics for studying A. succinogenes metabolism. Microarray and proteomic analyses
can complement MFA studies to gain a better understanding of the factors controlling A.
succinogenes metabolism. By shifting A. succinogenes from one growth condition to
another, and sampling for microarrays at different time points during this transition,
coordinately regulated genes can be identiﬁed. Such a microarray experiment was done
for Haemophilus inﬂuenzae by transitioning it from growth conditions to starvation
conditions to induce competence (15). An interesting outcome from this study was that
many C4 pathway genes were coordinately expressed as part of a regulon regulated by the
CAMP receptor protein. This protein has not been identiﬁed in the A. succinogenes
genome sequence, but the idea of a regulatory protein controlling all the C4 pathway
genes as a regulon is an exciting prospect. Kacser and Acerenza (9) proposed a ‘universal
method’ for increasing ﬂux to a desired product, which, in short, involves overexpressing
every enzyme in a pathway leading to the desired product. This approach was tested by
overexpressing different combinations of the ﬁve enzymes involved in tryptophan

synthesis in Saccharomyces cerevisiae (12). The highest ﬂux to tryptophan was observed

209

.‘rr

 

 

when all ﬁve enzymes were overexpressed (12). If all enzymes in the C4 pathway are part
of the same regulon, then it may be possible to affect their expression through the
regulator (e. g., by constitutively expressing the regulator or by mutating allosteric
binding sites), or through engineering the promoters. This approach would be more
convenient, and likely more stable, than overexpressing all C4 pathway genes from a

plasmid.

6.2.6 Developing genetic tools for replacing A. succinogenes chromosomal DNA
segments. Many of the future directions described in the previous sections rely on the
ability to knock out genes. This technology has not been developed yet for A.
succinogenes. This technology is not only important for studying A. succinogenes
metabolism, but is essential for its metabolic engineering. Knocking out genes is the most
commonly used method for preventing ﬂux to unwanted products and diverting ﬂux to
the product of interest. However, in some cases, preventing ﬂux to an unwanted product
could be detrimental to cell health, or even lethal. Although the goal for developing an A.
succinogenes homosuccinate fermentation involves decreasing the C3 pathway ﬂux, the
C3 pathway produces pyruvate and acetyl-CoA, which are essential intermediates for
biosynthesis. Without C3 pathway ﬂux, the medium must contain alternative compounds
to pyruvate and acetyl-CoA to support growth. The most economical source of these
compounds would likely be an uncharacterized mixture of an agricultural product or
waste stream. The problem with using these uncharacterized mixtures is that they make
purifying succinate from fermentation broths more difﬁcult and more expensive. If

succinate is to be produced in an inexpensive deﬁned medium, then some C3 pathway

210

 

51‘ ,

 

 

 

 

ﬂux must remain to produce enough pyruvate and acetyl-CoA for growth. Achieving low
C3 pathway ﬂux for biosynthesis without product excretion will require genetic tools
capable of making minor ﬂux adjustments. Such genetic tools will likely rely on the same
technology as that for gene knockouts. Rather than replacing a gene for an enzyme with
an interrupted or partially deleted gene, a gene could be replaced with one encoding a less
active enzyme. Alternatively, promoters could be re-engineered to decrease enzyme
expression. Clearly there is much need for technology to make speciﬁc chromosomal
alterations in A. succinogenes.

Prospects for developing genetic tools for chromosomal alterations in A.
succinogenes are not discouraging. As described in Chapter 5, many Pasteurellaceae
have speciﬁc mechanisms taking up linear fragments of DNA. Natural transformation has
been used to replace chromosomal segments in at least four different Pasteurellaceae (1,
2, 15, 20). The A. succinogenes genome sequence has several indicators that it is capable
of natural transformation, or at least was capable in the past. However, it is important to .
note that, in a study of natural competence in A. actinomycetemcomitans, only one of
seventeen isolates tested could be naturally transformed (20).

A more promising method for replacing A. succinogenes chromosomal segments
is to use a sucrose counter selection system. This technology was recently used to knock
out seven different genes in M. succiniciproducens (11), the closest known relative of A.
succinogenes. This technology works by using a suicide vector containing the mutant
gene of interest (e. g., a gene interrupted with an ampicillin resistance cassette) and sacB,
encoding B. subtilis levansucrase. Sucrose is toxic to Gram-negative bacteria expressing

levansucrase, likely due the accumulation of levans in the persiplasmic space ( 16). The

211

suicide vector is introduced to the cells (e. g., by electroporation) then the cells are plated
on media with ampicillin. Cells incorporating the plasmid into their genomes, most likely
by a single-recombination event, will grow on the selective media. Single recombinants
are then grown in liquid media. The greater cell population increases the chance of a
second single-recombination event between the wild-type gene and the mutant gene that
can splice out the wild-type gene and the sacB-containing suicide vector. Cells are then
plated on media with ampicillin and sucrose, and only those cells that retained the mutant
gene and lost the sacB gene should survive.

In establishing a knockout method for A. succinogenes it is important that the
target he a non-essential enzyme. PFL is not expected to be essential because A.
succinogenes mutants with no PFL activity and M. succiniciproducens pﬂB::CmR strains
are viable (11, 13). Furthermore, a PFL knockout strain should have better ratios of
succinate to alternative endproducts making it industrially attractive (11, 13). A PFL
knockout strain would also be useful for assessing the reversibility of PyrDH, as
described in section 6.2.3.

In summary, this dissertation has reﬁned our understanding of A. succinogenes
metabolism by analyzing its fermentations with and without l3C-substrates, and by
analyzing the its draft genome sequence. A. succinogenes metabolism was revealed to be
complex, with multiple branchpoints for distributing ﬂux to succinate versus alternative
fermentation products. Furthermore, the classical environmental perturbations used to
affect A. succinogenes fermentation balances, CO2 and reductant concentrations, were
shown to affect metabolism by other mechanisms than just their roles as substrates for the

C4 pathway. The future of A. succinogenes research and development will beneﬁt from

212

'16

9.1

F

‘11
L»
.

 

 

 

 

 

further characterizations of its metabolism using environmental perturbations, such as
different Na+ concentrations, and different NaHCO3 concentrations during fermentative
growth on fructose or during respiration with fumarate. The A. succinogenes genome
sequence greatly expands the experimental opportunities for using system-wide
approaches to study its metabolism. The ability to make chromosomal modiﬁcations in
would further increase the experimental opportunities for studying A. succinogenes

metabolism and is crucial for engineering a homosuccinate fermentation.

213

6.3 REFERENCES

1.

10.

11.

Bigas, A., M. E. Garrido, A. M. de Rozas, I. Badiola, J. Barbe, and M. Llagostera.
2005. Development of a genetic manipulation system for Haemophilus parasuis. Vet.
Microbiol. 105:223-228.

Bosse, J. T., J. H. E. Nash, J. S. Kroll, and P. R. Langford. 2004. Harnessing natural
transformation in Actinobacillus pleuropneumoniae: a simple method for allelic
replacements. FEMS Microbiol. Lett. 233:277-281.

Guettler, M. V., M. K. Jain, and D. Rumler. 1996. Method for making succinic acid,
bacteria] variants for use in the process, and methods for obtaining variants. US.
patent 5,573,931.

Guettler, M. V., M. K. Jain, and B. K. Soni. 1996. Process for making succinic acid,
microorganisms for use in the process and methods of obtaining the microorganisms.
US. patent 5,504,004.

Guettler, M. V., D. Rumler, and M. K. Jain. 1999. Actinobacillus succinogenes sp.
nov., a novel succinic-acid-producing strain from the bovine rumen. Int. J. Syst.
Bacteriol. 492207-216.

Hase, C. C., N. D. Fedorova, M. Y. Galperin, and P. A. Dibrov. 2001. Sodium ion
cycle in bacterial pathogens: evidence from cross-genome comparisons. Microbiol.
Mol. Biol. Rev. 65:353-370.

Hasona, A., Y. Kim, F. G. Healy, L. 0. Ingram, and K. T. Shanmugam. 2004.
Pyruvate forrnate lyase and acetate kinase are essential for anaerobic growth of
Escherichia coli on xylose. J. Bacteriol. 186:7593-7600.

Hong, S. H., and S.Y. Lee. 2001. Metabolic ﬂux analysis for succinic acid production
by recombinant Escherichia coli with ampliﬁed malic enzyme activity. Biotechnol.
Bioeng. 74:89-95.

Kacser, H., and L. Acerenza. 1993. A universal method for achieving increases in
metabolite production. Eur. J. Biochem. 216:361-367.

Kim, P., M. Laivenieks, J. McKinlay, C. Vieille, and J. G. Zeikus. 2004. Construction
of a shuttle vector for the overexpression of recombinant proteins in Actinobacillus
succinogenes. Plasmid 51: 108-1 15.

Lee, S. J ., H. Song, and S. Y. Lee. 2006. Genome-based metabolic engineering of

Mannheimia succiniciproducens for succinic acid production. Appl. Environ.
Microbiol. 72: 1939- 1948.

214

 

12.

l3.

14.

15.

16.

17.

18.

19.

20.

Niederberger, P., R. Prasad, G. Miozzari, and H. Kacser. 1992. A strategy for
increasing an in vivo ﬂux by genetic manipulations. The tryptophan system of yeast.
Biochem. J. 287:473-479.

Park, D. H., M. Laivenieks, M. V. Guettler, M. K. Jain, and J. G. Zeikus. 1999.
Microbial utilization of electrically reduced neutral red as the sole electron donor for
growth and metabolite production. Appl. Environ. Microbiol. 65:2912-2917.

Park, D. H., and J. G. Zeikus. 1999. Utilization of electrically reduced neutral red by
Actinobacillus succinogenes: physiological function of neutral red in membrane-
driven fumarate reduction and energy conservation. J. Bacteriol. 181:2403-2410.

Redﬁeld, R. J ., A. D. Cameron, Q. Qian, J. Hinds, T. R. Ali, J. S. Kroll, and P. R.
Langford. 2005. A novel CRP-dependent regulon controls expression of competence
genes in Haemophilus inﬂuenzae. J. Mol. Biol. 347:735-747.

Reyrat, J. M., V. Pelicic, B. Gicquel, and R. Rappuoli. 1998. Counterselectable

markers: untapped tools for bacterial genetics and pathogenesis. Infect. Immun.
66:4011-4017.

Stols, L., and M. 1. Donnelly. 1997. Production of succinic acid through
overexpression of NAD+-dependent malic enzyme in an Escherichia coli mutant.
Appl. Environ. Microbiol. 63:2695-2701.

Ullmann, R., R. Gross, J. Simon, G. Unden, and A. Kroger. 2000. Transport of C4-
dicarboxylates in Wolinella succinogenes. J. Bacteriol. 182:5757-5764.

van der Werf, M. J ., M. V. Guettler, M. K. Jain, and J. G. Zeikus. 1997.
Environmental and physiological factors affecting the succinate product ratio during
carbohydrate fermentation by Actinobacillus sp. 1302. Arch. Microbiol. 167:332-342.

Wang, Y., S.D. Goodman, R.J. Redﬁeld, and C. Chen. 2002. Natural transformation

and DNA uptake signal sequences in Actinobacillus actinomycetemcomitans. J.
Bacteriol. 184:3442-3449.

215

   

111111111111111111