SEMI-SYNTHESIS OF AROMATIC DIACIDS AND BIOSYNTHESIS OF KANOSAMINE IN 
ESCHERICHIA COLI 
By Kelly Miller 
A DISSERTATION 
Submitted to 
Michigan State University 
in partial fulfillment of the requirements 
for the degree of 
ChemistryÑDoctor of Philosophy 
2018 ABSTRACT 
SEMI-SYNTHESIS OF AROMATIC DIACIDS AND BIOSYNTHESIS OF KANOSAMINE IN 
ESCHERICHIA COLI 
By Kelly Miller 
 Microbial synthesis of chemicals from biobased feedstocks is an alternative to 
manufacture of materials derived from petroleum or unconventional gases, such as shale gas and 
coalbed methane.  Aromatic diacids, terephthalic acid, isophthalic acid , and phthalic acid, are 
valuable compounds for the polymers and plasticizers industry.  These diacids are currently 
manufactured via Amoco Mid-Century oxidation of petroleum-derived 
para
-, meta- and 
ortho-
xylenes, respectively; a process not without environmental and economic challenges. 
 In this work, a semi-synthesis of terephthalic, isophthalic, and phthalic acids is outlined 
using 2-hydroxymuconic acid as a common synthetic precursor.  Recombinant 
Escherichia coli
 was utilized combining protocatechuic acid anabolic and catabolic pathways to achieve 1.2 g/L 
2-hydroxymuconic acid from 
D-glucose in fed-batch fermentation.  Lewis-acid catalysis was 
explored in a cycloaddition of 2-hydroxymuconic acid with ethylene to produce terephthalic 
acid.  Cyclization of 2-hydroxymuconic acid afforded 6-carboxy-2-pyrone which was reacted 
with propiolic acid to afford mixtures of isophthalic and phthalic acids.  Propiolic acid from 
biogas methane and ethylene from bioethanol could maximize the amount of renewable carbon 
in this semi-synthetic strategy, producing all three petroleum-derived aromatic diacids without 
the need for Amoco Mid-Century oxidation process.   
 In an additional research trajectory, the production of kanosamine (3-amino-3-deoxy-
D-glucose) is evaluated in fed-batch fermentation of recombinant 
E. coli
.  Two biosynthesis 
pathways are reported in the literature: one native to 
Bacillus subtilis
 and the other reported in 
Amycolatopsis mediterranei 
and 
Bacillus pumilus
.  Genes encoding kanosamine biosynthetic 
enzymes from 
B. subtilis
, A. mediterranei
, and 
B. pumilus 
are expressed in 
Escherichia coli 
to 
determine whether heterologous expression results in kanosamine accumulation and if the 
system can be manipulated to maximize kanosamine production.  Recombinant 
E. coli
 expressing genes from 
B. subtilis
 168 produces 12.7 ± 0.6 g/L kanosamine in a 6% mol/mol yield 
from 
D-glucose.  Kanosamine yields were increased to 18 ± 1% mol/mol by blocking the 
Embden-Meyerhoff-Parnas pathway through a mutation in 
pgi-encoded phosphoglucose 
isomerase.  Enzymes native to 
A. mediterranei 
proved to be challenging to efficiently express in 
E. coli
, prompting examination of 
B. pumilus
 kanosamine biosynthesis.  Expression of 
B. 
pumilus
 SH-B11 genes in 
E. coli
 results in 6.3 g/L kanosamine titers in 4.4% mol/mol yield from 
D-glucose in fed-batch fermentation.  
In vitro
 feeding experiments suggest a departure from what 
is in the literature regarding kanosamine production in 
Bacillus pumilus
. TABLE OF CONTENTS 
LIST OF TABLES 
LIST OF FIGURES 
KEY TO ABBREVIATIONS 
CHAPTER ONE: SYNTHESIS OF RENEWABLE TEREPHTHALIC ACID, 
ISOPHTHALIC ACID, AND PHTHALIC ACID FROM MICROBE-
SYNTHESIZED 2-HYDROXYMUCONIC ACID 
INTRODUCTION 
1.  Chemicals manufacture from natural hydrocarbon sources: shale gas and 
petroleum 
2.  Commercial synthesis of terephthalic acid 
3.  Routes to biobased terephthalic acid 
PART I.  SYNTHESIS OF AROMATIC DIACIDS FROM 2-
HYDROXYMUCONIC ACID 
1.  Overview 
2.  Chemical synthesis of 2-hydroxymuconic acid and 6-carboxy-2-pyrone 
3.  Synthesis of terephthalic acid via cycloaddition of 2-hydroxymuconic acid and 
ethylene 
4.  Isophthalic acid and phthalic acid from cycloaddition of 6-carboxy-2-pyrone with 
propiolic acid 
5.  Discussion 
PART II.  BIOCATALYTIC SYNTHESIS OF 2-HYDROXYMUCONIC ACID 
FROM 
D-GLUCOSE 
1.  Overview 
2.  Host strain 
3.  Plasmid construction 
4.  Fed-batch fermentation conditions 
 4.1.  Glucose-rich conditions 
 4.2.  Glucose-limited conditions 
5.  Biosynthesis of 2-hydroxymuconic acid in 
E. coli
  5.1.  Fermentation of 
E. coli
 WN1/pKM6.280 
 5.2.  Investigation into first half of biosynthesis pathway 
 5.3.  Enzymatic tautomerization of 2-hydroxymuconic acid 
  5.3.1.  Deletion of 
pptA
 locus from 
E. coli
 WN1 
  5.3.2.  Fermentation of 
E. coli
 KM1/pKM6.280 
6.  Isolation of 2-hydroxymuconic acid from fermentation broth 
 6.1.  Organic extraction of 2-hydroxymuconic acid isomers 
 6.2.  Conversion to 6-carboxy-2-pyrone and subsequent isolation 
viii 
x xviii 
1 1 1 5 9 14 14 16 23 32 36 40 40 42 44 57 57 58 58 58 60 66 67 70 72 73 75 !iv7.  Discussion 
PART III: CONCLUSIONS 
REFERENCES 
CHAPTER TWO: HETEROLOGOUS EXPRESSION OF KANOSAMINE 
BIOSYNTHESIS PATHWAYS IN 
ESCHERICHIA COLI
 INTRODUCTION 
PART I.  KANOSAMINE PRODUCTION VIA EXPRESSION OF 
Bacillus subtilis
 GENES 1.  Variations in gene arrangement and promoter strength 
2.  Competition with glycolysis for 
D-glucose-6-phosphate availability 
 2.1.  Activity of 
D-glucose-6-phosphate 3-dehydrogenase 
 2.2.  Improving product yields by blocking a glycolytic pathway 
 2.3.  Media supplementation 
3.  Kanosamine isolation from fermentation broth and inhibitory effects on 
E. coli
 RB791serA PART II.  KANOSAMINE BIOSYNTHESIS FROM UDP-
D-GLUCOSE 
1.  Overview 
2.  Kanosamine production in fed-batch fermentations of 
E. coli 
constructs carrying 
Amycolatopsis mediterranei 
genes 
3.  Activity of UDP-
D-glucose 3-dehydrogenase 
4.  Bacillus pumilus 
as source of kanosamine biosynthetic genes 
 4.1.  Substrate specificity of NtdC
pum PART III.  CONCLUSIONS 
REFERENCES 
CHAPTER THREE: EXPERIMENTAL 
GENERAL METHODS 
 General chemistry 
 Analytical methods 
BACTERIAL STRAINS AND PLASMIDS 
PLASMID CONSTRUCTION VIA THE BglBrick ASSEMBLY STANDARD 
CULTURE MEDIUM 
FERMENTATION CONDITIONS 
 General 
 Glucose-rich fermentations 
 Glucose-limited fermentations 
 General procedure for processing fermentation broth 
 Analysis of fermentation cultures 
GENETIC MANIPULATIONS 
 General methods 
 Preparation and transformation of electrocompetent 
E. coli 
  77 80 83 94 94 104 104 135 135 138 145 147 152 152 153 164 166 179 181 187 194 194 194 194 197 198 200 202 202 203 205 206 206 207 207 208 !v Small-scale purification of plasmid DNA 
 Large-scale purification of plasmid DNA 
 Restriction enzyme digestion of DNA 
 Ligation of DNA 
 Agarose gel electrophoresis 
 Polyacrylamide gel electrophoresis (SDS-PAGE) 
CHAPTER ONE 
SYNTHETIC PROCEDURES 
 Diethyl 2-hydroxyhexa-2,4-dienedioate 
22  6-Carboxy-2-pyrone 
19  2-Hydroxymuconic acid  
 Derivatization of 2-hydroxymuconic acid to 2-hydroxydimethylmuconate 
 General procedure for cycloaddition of 2-hydroxymuconic acid with ethylene 
 2   General procedure for cycloaddition of propiolic acid 
18 and 6-carboxy-2- 
 pyrone 
19  Conversion of 2-hydroxymuconic acid isomer mixture to 6-carboxy-2- 
 pyrone 
19 in fermentation broth 
 D-Erythrose-4-phosphate 
ISOLATION OF 
Paenibacillus 
sp. JJ-1b (ATCC 35889) GENOMIC DNA 
BACTERIAL STRAINS AND PLASMIDS 
 Strain generation: 
E. coli
 KM1 
 BglBrick parts and plasmids 
 Plasmid pKM6.274 
 Plasmid pKM6.232 
 Plasmid pKM6.280 
ENZYME ASSAYS 
 Transketolase assay 
 DAHP synthase (AroF
FBR
) assay 
2-HYDROXYMUCONIC ACID EXTRACTION SOLVENT SCREENING 
CHAPER TWO 
PLASMID CONSTRUCTION 
 BglBrick parts and plasmids 
 Plasmid pSN1.292 
 Plasmid pKM10.220ABC 
 Plasmid pKM10.214CA 
 Plasmid pKM10.214CB 
 Plasmid pKM10.214AB 
 Plasmid pKM10.82A 
 Plasmid pKM11.9 
 Plasmid pKM11.46 
 Plasmid pKM11.135 
 Plasmid pKM10.284 
209 210 210 211 
211 
212 213 213 213 214 215 215 216 216 217 217 218 219 219 220 221 221 222 222 222 223 224 225 225 225 227 227 227 228 228 228 229 229 230 230 !vi Plasmid pKM9.279C 
RB791serA/pSN1.292 GROWTH STUDY WITH EXOGENOUS KANOSAMINE 
CELL LYSATE PREPARATIONS  
PROTEIN PURIFICATIONS 
 B. subtilis
 ntdC-
encoded 
D-glucose-6-phosphate 3-dehydrogenase 
 Amycolatopsis mediterranei rifL
-encoded UDP-
D-glucose 3-dehydrogenase 
ENZYME ASSAYS 
 D-glucose-6-phosphate 3-dehydrogenase assay (NtdC
sub
 and NtdC
pum)  UDP-
D-glucose 3-dehydrogenase assay (RifL) 
PURIFICATION OF KANOSAMINE FROM FERMENTATION BROTH 
REFERENCES 
  230 231 231 232 232 233 234 234 234 235 236 !vii
LIST OF TABLES 
Table 1.
  Composition of Marcellus shale gas play in Appalachia. 
Table 2.
  Characteristic data of a Middle-Eastern petroleum reserve. 
Table 3.  
Uncatalyzed cycloaddition between 2-hydroxymuconic acid and ethylene 
2. Table 4
.  Selected catalysts from screen of catalyzed reaction between 2-
hydroxymuconic acid and ethylene 
2. Table 5.  
Attenuating Cu
2+ electrophilicity. 
Table 6.  
Effects of temperature on yield of PTA. 
Table 7.  
Effects of catalyst loading on yield of PTA. 
Table 8.  
Effect of Br¿nsted acid on Cu(OTf)
2-catalyzed cycloaddition. 
Table 9.  
Copper(I)-catalyzed cycloaddition of 2-hydroxymuconic acid and ethylene. 
Table 10.  
Initial Lewis acid-catalyzed reactions between propiolic acid 
18 and 6-
carboxy-2-pyrone 
19. Table 11.
  Strains and plasmids used in Chapter 1 of this study. 
Table 12.
  Specific activity of DAHP synthase in cellular lysate of WN1/pKM6.274 
during glucose-limited fermentation.  
Table 13.
  Specific activity of DAHP synthase in cellular lysate of WN1/pKM6.280 
during glucose-limited fermentation. 
Table 14.
  Effect of exogenous 2-hydroxymuconic acid and 2-hydroxymuconate 6-
semialdehyde on DAHP synthase activity of 
E. coli
 WN1/pKM6.274. 
Table 15
.  2-Hydroxymuconic acid extraction solvent screening. 
Table 16.
  Extraction solvent screening for 6-carboxy-2-pyrone 
19 from 
fermentation broth. 
2 3 26 27 28 29 29 30 32 35 43 64 64 65 74 76 !viii
Table 17.
  Minimal inhibitory concentrations of kanosamine against various bacteria 
and yeast. 
Table 18.
  Strains and plasmids used in Chapter 2 of this study.
 Table 19.
  Kanosamine production in fed-batch fermentation of 
E. coli
 RB791serA(DE3)/pKM10.82A. 
Table 20.
  Kanosamine production with incomplete constructs. 
Table 21.  
D-glucose-6-phosphate 3-dehydrogenase specific activity. 
Table 22.
  Nitrogen supplementation during fed-batch fermentation of 
E. coli
 RB791serA/pSN1.292. 
Table 23.
  Specific activity of UDP-
D-glucose 3-dehydrogenase in cell-free extract 
and with purified protein. 
Table 24. 
 Activity of 
D-glucose-6-phosphate 3-dehydrogenase from 
B. subtilis
 and 
B. pumilus 
when expressed in 
E. coli
. Table 25.
  Plasmids used and constructed in Chapter 1 of this work. 
Table 26.
  Plasmids used and constructed in Chapter 2 of this work. 
97 104 126 129 137 146 165 180 221 225 !ixLIST OF FIGURES 
Figure 1.
  Map of shale gas reserves in the contiguous United States.   
Figure 2.  
Schematic representation of petroleum fractionation.
 Figure 3.  
Oxidation of petroleum-derived xylene isomers. 
Figure 4.  
Reaction conditions of Amoco Mid-Century oxidation of 
p-xylene 
1 to 
PTA. 
Figure 5.  
Degradation of acetic acid via Kolbe oxidation. 
Figure 6.
  Biobased routes to 
p-xylene 
1 from ethylene 
2, isobutanol 
3, sorbitol 
4, and 5-hydroxmethylfurfural. 
Figure 7.
  Biobased routes to terephthalic acid from succinic acid 
5, cis,cis-muconic 
acid 
7, limonene 
10, malic acid 
12, 5-hydroxymethylfurfural, and isoprene 
14 and 
acrylic acid 
15 Figure 8.  
Reaction network to biobased terephthalic acid, isophthalic acid, and 
phthalic acid via microbe-synthesized 2-hydroxymuconic acid. 
Figure 9.  
Biosynthesis of 2-hydroxymuconic acid via protocatechuate 2,3-cleavage 
pathway described in 
Paenibacillus 
sp. JJ-1b. 
Figure 10. 
 Synthesis of 2-hydroxymuconic acid and 6-carboxy-2-pyrone 
19. Figure 11.  
1H NMR spectrum of 2-hydroxyhexa-2,4-dienedioate 
22 (CD3OD).
 Figure  12
.  13C NMR spectrum of 2-hydroxyhexa-2,4-dienedioate 
22 (CD3OD).
 Figure 13.  
1H NMR spectrum of 6-carboxy-2-pyrone 
19 (D2O).
 Figure 14
.  13C NMR of 6-carboxy-2-pyrone 
19 (DMSO-
d6). Figure  15.  
1H NMR spectrum of 2-hydroxymuconic acid (CD
3OD). 
Figure 16.
  1HMR spectrum of 2-hydroxymuconic acid (DMSO-
d6). Figure 17.
 Possible geometries about the C2-C3 double bond.
 2 4 6 7 8 10 12 15 16 17 18 19 20 21 22 23 24 !xFigure 18.
  Relative energy levels of the frontier molecular orbitals of 2-
hydroxymuconic acid and ethylene 
2. Figure 19.  
Putative mechanism for copper(I)-catalyzed protodecarboxylation of 
aromatic carboxylates. 
Figure 20.  
Synthesis of isophthalic acid (IPA) and phthalic acid (PA) from 6-
carboxy-2-pyrone 
19 and renewable propiolic acid 
18. A. a) 1500-1900 ¡C, b) Cu
+. Figure 21.
  Relative energy levels of the frontier molecular orbitals of 6-carboxy-2-
pyrone 
19 and propiolic acid 
18. Figure 22.
  Degradation of 4-hydroxybenzoic acid via the PCA 2,3-cleavage 
pathway in 
Paenibacillus
 sp. JJ-1b. 
Figure 23.  
De novo biosynthesis of 2-hydroxymuconic acid combining 
protocatechuic acid anabolism and catabolism.
 Figure 24.
  Schematic representation of BglBrick DNA assembly method. 
Figure 25. 
 Preparation of 
pra
 BglBrick parts: plasmids pKM6.196, pKM6.200, and 
pKM6.201. 
Figure 26.  
Preparation of 
serA BglBrick part, plasmid pKM6.240. 
Figure 27.  
Preparation of 
aroF
FBR
 BglBrick part, plasmid pKM6.268. 
Figure 28.  
Preparation of 
ParoF
 BglBrick part, plasmid pKM6.241. 
Figure 29.  
Preparation of plasmid pKM6.208. 
Figure 30.  
Preparation of plasmid pKM6.232. 
Figure 31.  
Preparation of plasmid pKM6.261. 
Figure 32.  
Preparation of plasmid pKM6.274. 
Figure 33.  
Preparation of plasmid pKM6.280. 
Figure 34
.  Biosynthesis of 2-hydroxymuconic acid by 
E. coli
 WN1/pKM6.280 
under glucose-rich conditions.
 25 31 33 34 41 42 46 48 49 50 51 52 53 54 55 56 59   !xiFigure 35.
  Biosynthesis of 2-hydroxymuconic acid by 
E. coli
 WN1/pKM6.280 
under glucose-limited conditions.   
Figure 36.  
Biosynthesis of PCA by 
E. coli
 WN1/pKM6.274 under glucose-rich 
conditions. 
Figure 37.  
Biosynthesis of PCA by 
E. coli
 WN1/pKM6.274 under glucose-limited 
conditions.  
Figure 38.  
Biosynthesis of PCA and DHS by 
E. coli 
WN1/pKM6.274 in the 
presence of externally added 2-HMA (1 g/L).  
Figure 39.
  Structures of 2-hydroxymuconic acid (2-HMA) and 
!
-keto acid 
tautomers. 
Figure 40.  
Action of 4-oxalocrotonate tautomerase. 
Figure 41.
  Workflow for homologous recombination using FRT-flanked kanamycin 
resistance cassette and phage 
"
 Red recombinase system. 
Figure 42.
  Verification of 
pptA
 deletion via PCR. 
Figure 43.
  Biosynthesis of 2-hydroxymuconic acid by 
E. coli
 KM1/pKM6.280 
under glucose-rich conditions.   
Figure 44.
  Biosynthesis of 2-hydroxymuconic acid by 
E. coli
 KM1/pKM6.280 
under glucose-limited conditions.   
Figure 45.
  Biologically active lipopeptides, siderophores, and volatile organic 
compounds produced by 
Bacillus 
spp. 
Figure 46.
  Structure of kanosamine (3-amino-3-deoxy-
D-glucose). 
Figure 47.
  Kanosamine biosynthesis from 
D-glucose-6-phosphate described in 
Bacillus subtilis
. Figure 48.
  Kanosamine biosynthesis pathway via UDP-
D-glucose as proposed by 
Umezawa. 
Figure 49.
  Representative members of the ansamycin and mitomycin families of 
antibiotics. 
59 60 61 62 66 67 69 70 71 71 95 96 98 99 100 !xii
Figure 50.
  Synthesis of kanosamine from UDP-
D-glucose and the formation of 3-
amino-5-hydroxybenzoic acid via the aminoshikimate pathway in 
Amycolatopsis 
mediterranei. 
Figure 51.
  Construction of 
ntd
 BglBrick parts: plasmids pSN1.296, pSN1.295, and 
pSN1.294. 
Figure 52.
  Construction of plasmid pSN1.119. 
Figure 53.
  Construction of plasmid pSN1.139. 
Figure 54.
  Construction of plasmid pSN1.292. 
Figure 55. 
 Construction of plasmids pKM10.204A, pKM10.204B, and 
pKM10.204C. 
Figure 56.
  Construction of plasmid pKM10.214BC. 
Figure 57.
  Construction of pKM10.220ABC. 
Figure 58.
  Kanosamine production by 
E. coli 
RB791serA/pSN1.292 under glucose-
rich conditions. 
Figure 59.
  Kanosamine production by 
E. coli 
RB791serA/pSN1.292 under glucose-
limited conditions. 
Figure 60.
  Kanosamine production by 
E. coli 
RB791serA/pKM10.220ABC under 
glucose-rich conditions. 
Figure 61.
  Kanosamine production by 
E. coli 
RB791serA/pKM10.220ABC. 
Figure 62.
  Construction of plasmid pKM10.82A. 
Figure 63.
  Kanosamine production by 
E. coli 
RB791serA(DE3)/pKM10.82A under 
glucose-rich conditions. 
Figure 64.
  Kanosamine production by 
E. coli 
RB791serA(DE3)/pKM10.82A under 
glucose-limited conditions. 
Figure 65.
  Construction of plasmid pKM10.214CA. 
Figure 66.
  Construction of plasmids pKM10.214AB and pKM10.214CB. 
101 108 109 110 
111 
112 
113 
114 
119 
119 
121 121 123 124 124 127 128 !xiii
Figure 67.  
Byproducts formed during fermentation of 
E. coli
 RB791serA/pSN1.292 
under glucose-rich conditions.   
Figure 68.  
Byproducts formed during fermentation of 
E. coli
 RB791serA/pSN1.292 
under glucose-limited conditions. 
Figure 69. 
 Stacked 
1H NMR spectra of partially purified kanosamine with 
increasing amounts of added 
D-glucosamine. 
Figure 70. 
 Stacked 
13C NMR spectra corresponding to partially purified 
kanosamine with and without added 
D-glucosamine. 
Figure 71.
  D-Glucosamine synthesis from 
D-glucose-6-phosphate through 
transamination of 
D-fructose-6-phosphate catalyzed by 
glmS-
encoded 
L-glutamine:
D-fructose-6-phosphate aminotransferase followed by phosphate ester 
hydrolysis.  
Figure 72.
  Kanosamine production by 
E. coli 
RB791serA pgi
::KanR/pSN1.292 
under glucose-rich conditions with a 
D-glucose-only feed. 
Figure 73.
  Byproducts formed in fed-batch fermentation of 
E. coli 
RB791serA 
pgi::KanR/pSN1.292 under glucose-rich conditions with a 
D-glucose-only feed. 
Figure 74.
  Kanosamine production by 
E. coli 
RB791serA pgi
::KanR/pSN1.292 
under glucose-rich conditions with a mixed feed of 
D-glucose/
D-fructose (1:1 mol/
mol). 
Figure 75.
  Byproducts formed in fed-batch fermentation of 
E. coli 
RB791serA 
pgi::KanR/pSN1.292 under glucose-rich conditions with a 
D-glucose-only feed. 
Figure 76.
  The Krebs cycle indicating the glyoxylate shunt from 
D-threo
-isocitrate 
to glyoxylate. 
Figure 77.
  Glutamate dehydrogenase-catalyzed transamination of 2-ketoglutarate to 
L-glutamic acid. 
Figure 78.
  1H NMR spectrum of kanosamine isolated from fermentation broth. 
Figure 79.
  13C NMR spectrum of kanosamine isolated from fermentation broth. 
Figure 80.
  HSQC NMR spectrum of kanosamine isolated from fermentation broth. 
130 131 133 133 135 139 139 140 140 143 144 148 149 150 !xiv
Figure 81.
  Exponential phase growth of RB791
serA/pSN1.292 with exogenously 
added kanosamine. 
Figure 82.  
Kanosamine biosynthesis pathways reported in the literature.  Pathway A 
described in 
Bacillus subtilis
; Pathway B described in 
B. pumilus
, B. cereus
, and 
A. 
mediterranei. 
Figure 83.
  Construction of 
rif BglBrick parts: plasmids pKM10.254K, 
pKM10.254L, and pKM10.254L. 
Figure 84.
  Construction of plasmid pKM10.268M. 
Figure 85.
  Construction of plasmid pKM10.278KM. 
Figure 86.
  Construction of plasmid pKM11.9. 
Figure 87.
  Construction of plasmid pKM11.46. 
Figure 88.
  Products formed in fed-batch fermentation of 
E. coli 
RB791serA/pKM11.9 under glucose-rich conditions. 
Figure 89.
  Products formed in fed-batch fermentation of 
E. coli 
RB791serA/pKM11.9 under glucose-limited conditions. 
Figure 90.
  Products formed in fed-batch fermentation of 
E. coli 
RB791serA/pKM11.9 with 20 mg/L uracil supplementation under glucose-rich conditions. 
Figure 91.
  Products formed in fed-batch fermentation of 
E. coli 
RB791serA(DE3)/
pKM11.46 with 200 mg/L uracil supplementation under glucose-limited conditions. 
Figure 92.
  Products formed in fed-batch fermentation of 
E. coli 
KL3/pKM11.9 
glucose-rich conditions. 
Figure 93.
  Products formed in fed-batch fermentation of 
E. coli 
KL3/pKM11.9 
glucose-limited conditions. 
Figure 94.
  Assay conditions for RifL-catalyzed conversion of UDP-
D-glucose to 3-
keto-UDP-
D-glucose. 
Figure 95.
  Kanosamine biosynthesis pathway via UDP-
D-glucose as proposed by 
Umezawa. 
151 153 155 156 157 158 159 160 160 161 162 163 163 165 167 !xvFigure 96. 
 Construction of 
ntd
 BglBrick parts: plasmids pKM11.121A, 
pKM11.121B, and pKM11.121C. 
Figure 97.
  Construction of plasmid pKM11.125. 
Figure 98.
  Construction of plasmid pKM11.130. 
Figure 99.
  Construction of plasmid pKM11.135. 
Figure 100.
  Kanosamine production by 
E. coli 
RB791serA/pKM11.135 under 
glucose-rich conditions.  
Figure 101.
  Kanosamine production by 
E. coli 
RB791serA/pKM11.135 under 
glucose-limited conditions. 
Figure 102.  
Byproducts formed during fermentation of 
E. coli
 RB791serA/pKM11.135 under glucose-rich conditions. 
Figure 103.  
Byproducts formed during fermentation of 
E. coli
 RB791serA/pKM11.135 under glucose-limited conditions.  
Figure 104.
  Kanosamine production in fed-batch fermentation of RB791
serA/pKM11.135 in glucose-rich conditions with 6 mg IPTG addition every 6 h beginning 
at 12 h into the fermentation run.   
Figure 105.
  Kanosamine production in fed-batch fermentation of RB791
serA/pKM11.135 in glucose-limited conditions with 6 mg IPTG addition every 6 h 
beginning at 12 h into the fermentation run.  
Figure 106.
  BglBrick plasmid assembly workflow. 
Figure 107.
  BglBrick vector pBbA1a-RFP. 
Figure 108. 
 Mass flow controller settings for first stage of glucose-rich 
fermentations. 
Figure 109. 
 Mass flow controller settings for second stage of glucose-rich 
fermentations. 
Figure 110. 
 Mass flow controller settings for second stage of glucose-rich 
fermentations. 
Figure 111
.  BglBrick plasmid pBbA7k-RFP. 
170 171 172 173 174 175 176 177 178 178 199 200 204 205 206 229 !xviFigure 112.
  Beginning gene sequence for plasmid pKM10.284. 
Figure 113.
  Beginning gene sequence for plasmid pKM9.279C.  
230 231 !xvii
KEY TO ABBREVIATIONS 
 AHBA   3-amino-5-hydroxybenzoic acid 
 aminoDAHP    4-amino-3,4-dideoxy-
D-arabino
-heptulosonate-7-phosphate 
 aminoDHQ    5-amino-5-deoxy-3-dehydroquinate 
 aminoDHS    5-amino-5-deoxy-3-dehydroshikimate 
 aminoF6P    3-amino-3-deoxy-
D-fructose-6-phosphate 
 AMP   2-amino-2-methyl-1-propanol 
 ATP     adenosine triphosphate 
 BTX     benzene, toluene, xylenes 
 cAMP   cyclic adenosine monophosphate 
 4-CBA    4-carboxybenzaldehyde 
 CHMS    5-carboxy-2-hydroxymuconic semialdehyde 
 CoA   coenzyme A 
 DAHP    3-deoxy-
D-arabino
-heptulosonate 7-phosphate 
 DEHP    di-(2-ethylhexyl) phthalate 
 DEPT     distortionless enhancement by polarization transfer 
 DINCH    diisononyl cyclohexane-1,2-dicarboxylate 
 DHQ     3-dehydroquinate 
 DHS     3-dehydroshikimate 
 DMF     dimethylfuran 
 DMT     dimethylterephthalate 
!xviii
 DNA     deoxyribonucleic acid 
 D.O.     dissolved oxygen 
 E4P     
D-erythrose-4-phosphate 
 F6P     
D-fructose-6-phosphate 
 FDCA    2,5-furandicarboxylic acid 
 GA   gallic acid 
 G6P     
D-glucose-6-phosphate 
 GC   gas chromatography 
 Glu     
L-glutamate 
 Gln     
L-glutamine 
 GlcN   glucosamine (2-amino-2-deoxy-
D-glucose) 
 h     hour 
 2-HMA    2-hydroxymuconic acid 
 HMF     hydroxymethylfurfural 
 HMS     2-hydroxymuconic semialdehyde 
 HOMO    highest occupied molecular orbital 
 HPLC   high performance liquid chromatography 
 IED DA    inverse electron demand Diels-Alder 
 iminoE4P    1-imino-1-deoxy-
D-erythrose-4-phosphate 
 IPA    isophthalic acid 
 IPTG   
#
-D-isopropylthiogalactopyranoside 
 k     rate constant 
!xix
 kb   kilobase 
 kD   kilodalton 
 
!
KG   
!
-ketoglutarate 
 K6P     kanosamine-6-phosphate 
 LUMO    lowest occupied molecular orbital 
 M     molar 
 mC7N   six-membered carbocycle with 
meta-substituted carbon and  
    nitrogen groups 
 mg     milligram 
 mL     milliliter 
 mM     millimolar 
 min     minute 
 NAD
+    nicotinamide adenine dinucleotide, oxidized form 
 NADH   nicotinamide adenine dinucleotide, reduced form 
 NMR     nuclear magnetic resonance 
 NOE     nuclear Overhauser effect 
 NTA   nitrilotriacetic acid 
 OD     optical density 
 PA     phthalic acid 
 PCA     protocatechuic acid 
 PCR     polymerase chain reaction 
 PEP     phosphoenolpyruvate 
!xx PET     poly(ethylene terephthalate) 
 PLP   pyridoxal phosphate 
 PMP   pyridoxamine phosphate 
 psi   pounds per square inch 
 PTA     terephthalic acid 
 RNA   ribonucleic acid 
 R5P   ribose-5-phosphate 
 SDS-PAGE  sodium dodecylsulfate polyacrylamide gel electrophoresis 
 TCA   citric acid 
 U   unit; 
$
mol/min 
 UDP     uridine diphosphate 
 UDPGlc  UDP-3-keto-
D-glucose 
 UTP     uridine triphosphate 
!xxiCHAPTER ONE: SYNTHESIS OF RENEWABLE TEREPHTHALIC ACID, 
ISOPHTHALIC ACID, AND PHTHALIC ACID FROM MICROBE-SYNTHESIZED 2-
HYDROXYMUCONIC ACID 
INTRODUCTION 
1. Chemicals manufacture from natural hydrocarbon sources: shale gas and petroleum 
 Commercial manufacture of fuels and chemicals relies heavily on non-renewable natural 
resources such as natural gas and petroleum.  Reserves of these fossil fuels are widespread and 
the composition of each varies from one location to another.
1  In North America, natural gas 
trapped in pores of underground shale rock beds is an abundant source of hydrocarbons.  A map 
of the shale gas reserves in the contiguous United States presented by the United States Energy 
Information Administration is shown in Figure 1.
2  Though the precise composition of shale gas differs from one wellhead to another, the 
primary component is always methane.
3  Ethane, propane, and higher hydrocarbons may be 
present as well as compounds such as carbon dioxide, nitrogen, and hydrogen sulfide.  The 
Marcellus shale gas play is the largest in the U.S. and one of the most productive in the world.
4  The composition of the Marcellus shale gas play is described in Table 1.
1  Because shale gas is 
trapped much deeper than conventional natural gas, new technologies were required to access it.  
Horizontal drilling and hydraulic fracturing, or fracking, were developed explicitly for this 
purpose.
5 %1%   Figure 1.
  Map of shale gas reserves in the contiguous United States. 
  Table 1.
  Composition of Marcellus shale gas play in Appalachia. 
 Briefly, a well bore drills vertically downward until it reaches a layer of shale rock.  The 
bore is then turned horizontally and drilling continues through the rock, providing much greater 
Component
Mol percentage
Methane
76.9Ethane
13.4Propane
4.5Butanes
1.6Pentanes
1.3CO21.0N21.0%2recovery of gas trapped in rock pores than vertical drilling.  A steel casing equipped with 
explosive charges surrounds the horizontal portion of the well.  Detonation of the charges creates 
large fractures in the shale rock which are kept open by highly pressurized water.  The water 
contains fine-grained silica (SiO
2) and other chemicals that prevent the fractures collapsing once 
the water pump is shut off.  Reducing the water pressure allows the shale gas to flow through the 
open fractures and into the well.
5  Petroleum contains a wider distribution of hydrocarbons than natural or shale gas.  When 
organic compounds such as fatty acids and steroids are pressurized for millions of years within 
the earth, a crude oil forms in which molecules containing forty or more carbon atoms may be 
present.
6,7  In order for crude oil to serve as a useful source of hydrocarbons for chemicals 
manufacture, it must be separated into useful fractions.  Table 2 describes some characteristic 
properties of petroleum fractions from a Middle-Eastern crude oil.
8  Each fraction contains a 
different distribution of paraffins, aromatics, and cycloaliphatics (naphthenes).    
Table 2.
  Characteristic data of a Middle-Eastern petroleum reserve. 
  Fraction
Volume 
percentage
Boiling 
point range  
(¡C)
Paraffins  
(vol%)
Naphthenes  
(vol%)
Aromatics  
(vol%)
Light naphtha
7.920-10089.69.50.9Heavy naphtha
6.8100-15070.321.48.3Kerosene
12.5150-23558.023.718.3Light gas oil
16.4235-343ÑÑHeavy gas oil
26.3343-565ÑÑResidual oil
26.8>565ÑÑ%3 Petroleum refining begins by desalting crude oil via thermal or electrical separations.
7  Contaminants such as sodium, calcium, and magnesium occupy an aqueous phase that is 
coproduced with the crude oil.  These salts must be removed in order to prevent corrosion of 
pipelines and reactor equipment.  Atmospheric distillation (Figure 2) follows removal of 
inorganic contaminants.
7  The fractionation tower separates light hydrocarbons from heavier gas 
oils which are subjected to further fractionation at pressures below 1 bar.  Side streams of 
atmospheric distillation, heavy naphtha, kerosene, and light and heavy gas oils, are separated in 
stripping columns with use of superheated steam.  Gas oils and fractions isolated from vacuum 
distillation are commercially used for lubricating oils and bunker fuels while commodity 
chemicals are made from atmospheric distillation fractions.
7,8 % Figure 2.  
Schematic representation of petroleum fractionation.
 Crude
oilAtmospheric
distillation
column
Light naphtha  20-100¡C
Heavy naphtha  100-150¡C
Gas oils  235-565¡C
Heavy distillate  >565¡C
Vacuum
distillation
column
Lubricating oil
Asphalt/Residue
%4 Though petroleum and natural gas allow access to a plethora of different hydrocarbons 
for manufacture of fuels and chemicals, the cost and availability of such resources are subject to 
substantial and unpredictable variation.  Alternative feedstocks must be made available to 
stabilize costs associated with commodity chemical synthesis.  In times of increased petroleum 
or natural gas prices, synthetic routes involving biobased starting materials are advantageous.  
Terephthalic acid is an example of a commodity chemical that would greatly benefit from a 
diverse pool of available feedstocks.  
2.  Commercial synthesis of terephthalic acid 
 Since its discovery in 1847,
9 purified terephthalic acid (PTA) has become an important 
large volume commodity chemical.  Nearly all of the 47 million tons consumed globally in 2012 
went into the manufacture of poly(ethylene terephthalate) (PET).
10  With applications ranging 
from clear beverage bottles, molded packaging containers, and polyester films and fibers, PET is 
widely used and highly valuable.
10   Current commercial production of PTA relies on the Co
2+/Mn2+-catalyzed oxidation of 
petroleum-derived 
para
-xylene developed by the Mid-Century Corporation (Amoco).
11  The 
xylene isomers, 
ortho
-, meta
-, and 
para
- (Figure 3), are dimethylated aromatic compounds which 
are part of a highly valued cut of petroleum-derived aromatics called BTX (benzene, toluene, 
xylenes).  Products of xylene oxidation are important players in the polymers and plasticizers 
industry (Figure 3). 
%5% Figure 3.  
Oxidation of petroleum-derived xylene isomers. 
   Oxidation of 
ortho
-xylene affords phthalic acid (PA) which can be dehydrated to phthalic 
anhydride.  This compound is used as a scaffold for plasticizers such as di-(2-ethylhexyl) 
phthalate (DEHP) and diisononyl cyclohexane-1,2-dicarboxylate (DINCH) (Figure 3).  The 
products of 
meta- and 
para
-xylene oxidation, isophthalic acid (IPA) and terephthalic acid (PTA), 
respectively, are condensed with ethylene glycol to afford PET.  Amoco Mid-Century liquid-
phase oxidation of 
meta- and 
para
-xylenes is commercially performed in a high pressure reactor 
charged with substrate and catalyst mixture in acetic acid.  The catalyst system is a mixture of 
cobalt and manganese acetates with a bromine compound such as sodium bromide as a radical 
chain propagator.  The oxidation occurs at pressures of 15-30 bar and temperatures of 
175-225¡C.
12 HO2CCO2HHO2CCO2HOOOdiisononyl cyclohexane-1,2-dicarboxylate
DINCHOOOOOOOOPoly(ethylene terephthalate)
PETortho
-xylene
meta
-xylene
para
-xylene
phthalic anhydride
isophthalic acid
IPA
terephthalic acid
PTA
OOOOdi-(2-ethylhexyl) phthalate
DEHP
OOOO%6 Figure 4.  
Reaction conditions of Amoco Mid-Century oxidation of 
p-xylene 
1 to PTA. 
 Oxidation of 
p-xylene 
1 to PTA (Figure 4) follows the Hammett structure-reactivity 
relationship (equation 1): 
 (1)    log(
k/k0) = (!)(!) where 
k = the rate constant for the consumption of a compound, 
k0 = the rate constant for the 
consumption of toluene, 
! = the characteristic constant for a ring substituent, and 
! = the 
characteristic constant for a set of reaction conditions.  The Hammett equation (eq. 1) describes a 
linear free-energy relationship relating reaction rates and equilibrium constants for many organic 
reactions involving benzene derivatives.  The equation relates the equilibrium constant (
k) for a 
given reaction with substituent, R, and the equilibrium constant for a reference compound (
k0) where R is a hydrogen atom.  Only two parameters are utilized: a substituent constant (
&
) which 
depends only on the specific substituent and the reaction constant (
!) which depends on the type 
of reaction but not the substituent used.  Reported 
! values for metal/bromine-catalyzed 
oxidations of aryl methyl groups typically fall in the range of -0.6 to -1.3.
12  Industrial oxidation 
conditions are conventionally 90% acetic acid solutions in water, which offer 
! values of -0.95.
12  Equation 1 explains why the complete oxidation of one methyl group occurs before the other.  
Once the first methyl group reaches the carboxylic acid stage, the 
! value for the 
para
-methyl 
OHOHOOHOOHOOHOOHOHOONaBr, CBr
4Mn(OAc)
2Co(OAc)
2AcOH
175-225 ¡C
15-30 bar
+H2OH2, Pd/C
285 ¡C90 bar1PTA
4-CBA
PTA
2%7group rises from -0.3 to +0.4, thus indicating a reduction in the ring electron density.  This 
reduction causes the second methyl oxidation to proceed 4.9 times slower than the first.
12,13  Overoxidative burning of 
p-xylene 
1 to carbon oxides represents a major challenge 
associated with the Amoco Mid-Century process.  Additional carbon oxides are produced along 
with methane via Kolbe decomposition of acetic acid (Figure 5).
12,13  These side reactions 
attendant with radical chain propagation and harsh reaction conditions greatly impact process 
economics.  For every 1000 kg of PTA produced, 165 kg of CO
2 is generated with 60% due to 
oxidation of acetic acid and 40% due to 
p-xylene 
1 overoxidation.
14  Solvent degradation and 
oxidative burning of crude PTA embody an unnecessary waste of carbon and increase the costs 
associated with PTA manufacture. 
%  Figure 5.  
Degradation of acetic acid via Kolbe oxidation. 
 Ironically, the most problematic contaminant formed during the Amoco Mid-Century 
oxidation is a product of incomplete oxidation, 4-carboxybenzaldehyde (4-CBA, Figure 4).  
During the condensation polymerization of IPA, PTA, and ethylene glycol to afford PET, any 4-
CBA present in the crude PTA product mixture would serve as a chain terminator, thus inhibiting 
PET chain growth.  As such, significant effort has been expended to eliminate 4-CBA from the 
product stream.  Crude PTA typically containing around 0.3% by weight of 4-CBA is dissolved 
in water at 285 ¡C and 90 bar and sent into a trickle-bed reactor.  The PTA crude flows down a 
Pd/C catalyst bed while a gaseous H
2 stream moves upward.  Hydrogen reduces 4-CBA to 
p-OHOOOCO2CH3%8toluic acid 
2 which has a higher solubility in water than PTA.  Terephthalic acid of 99.99% purity 
precipitates from solution as the aqueous phase exits the reactor while 
p-toluic acid 
2 remains in 
solution.
13    3.  Routes to biobased terephthalic acid 
 Challenges accompanying Amoco Mid-Century oxidation of aryl methyl groups 
encourage the development of alternative substrates for PTA production.  Synthetic routes 
involving biobased substrates offer an attractive means to supplement fossil fuel-derived PTA.  
Synthesis of PTA from alternative feedstocks can be broadly categorized into two groups: those 
with and those without 
p-xylene 
1 intermediacy.  Biobased materials including
 5-hydroxymethylfurfural (HMF)
15, ethylene 
2,16,17 isobutanol 
3,17-21 and sorbitol 
422 (Figure 6) 
offer a means of access to 
p-xylene 
1 which are easier to integrate into the current chemical 
manufacture infrastructure than novel routes to PTA.  These routes, however, do not avoid aryl 
methyl group oxidation.  Despite its current utility, the Amoco Mid-Century oxidation requires 
large amounts of solvent, expensive reactor equipment, arduous purifications, and results in 
overoxidative burning of the crude product. 
%9%   Figure 6.
  Biobased routes to 
p-xylene 
1 from ethylene 
2, isobutanol 
3, sorbitol 
4, and 5-hydroxymethylfurfural. 
 High yields and selectivities are reported for different steps in many of the syntheses of 
biobased 
p-xylene 
1.  The cycloaddition of ethylene 
2 with HMF-derived dimethylfuran (DMF), 
for example, affords 
p-xylene 
1 with >99% conversion and 90% selectivity.
23  Vapor phase 
dehydration of isobutanol 
3 over 
'
-Al
2O3 at 450 ¡C and 4 bar affords isobutylene with >99% 
conversion with 95% selectivity.
17-20  Currently, no route reports high yields and selectivities for 
every step of the synthesis.  Moreover, pathways from ethylene 
2 and sorbitol 
4 require 
expensive catalysts, such as the 10 wt% Pt(Re)/C catalyst used in the conversion of sorbitol 
422 to 
p-xylene 
1 or the iridium pincer complex used in the second step of the route from biobased 
OOHOHOOOHOHOHOHOHHOO1CO2HOCO2HCHOHMF
DMF
234H2O%10ethylene 
2.16,17  Industrially viable syntheses should use inexpensive catalysts and afford >90% 
selectivities.  Low conversions could be tolerated assuming any unreacted species are able to be 
recycled.  More development is necessary if any of the routes to biobased 
p-xylene 
1 can be 
considered for commercialization.  Even so, Amoco Mid-Century oxidation will be required in 
order to afford PTA. 
 Several synthetic routes to biobased PTA have been proposed that do not require 
p-xylene  
intermediacy and hence avoid the challenges associated with the Amoco Mid-Century process 
(Figure 7).
24-37  Succinic acid 
5, muconic acid 
7, malonic acid 
12, and 2,5-furandicarboxylic acid 
(FDCA) already contain the requisite carboxylic acid moieties present in the structure of PTA, 
thus avoiding the need for aryl methyl group oxidation.  Routes from succinic 
5, muconic 
7, and 
malonic acids 
12 require an esterification step to achieve reasonable yields in subsequent 
cycloadditions.  These pathways necessitate hydrolysis of dimethylterephthalate (DMT) to afford 
PTA.  Introducing additional steps makes a synthesis less commercially attractive.  Use of 
limonene as a substrate for PTA synthesis involves an oxidation with stoichiometric potassium 
permanganate.
36  This reaction would likely by supplanted with an Amoco Mid-Century 
oxidation as would oxidation of 
p-toluic acid to PTA in the route from isoprene 
14 and acrylic 
acid 
15.37 %11% Figure 7.
  Biobased routes to terephthalic acid from succinic acid 
5, cis,cis-muconic acid 
7, limonene 
10, malic acid 
12, 5-hydroxymethylfurfural, and isoprene 
14 and acrylic acid 
15.  Several challenges must be addressed before PTA production from renewable resources is 
commercially established.  Low conversions and selectivities preclude the viability of most of 
the proposed routes to biobased PTA.  High selectivities may reduce waste if unreacted starting 
material could be recycled.  Without high conversion or selectivity at every step of a synthesis, 
the likelihood is low that it could be adopted into practice.  Stoichiometric use of catalysts such 
as those employed for limonene 
10 and the oxidation of HMF to FDCA are also undesirable for 
industrial processes.  
 Finally, there is only one proposed pathway to biobased PTA that addresses the need for 
3-5% (w/w) IPA for PET manufacture.  Regioselective cycloaddition of isoprene 
14 and acrylic 
acid 
15 catalyzed by 2 mol% TiCl
4 affords a 24:1 molar ratio of 
para
- and 
meta-
cycloadducts.  
After aromatization and oxidation, PTA is produced along with small amounts of IPA which can 
OOHOFDCAOHOHOOO10OROOOR12OHOOHOHO1113OOOOCO2CH3H3CO2CCO2RRO2CR = H, PTA
R = CH
3, DMT
CO2CH3OH3CO2COR = H, 
5R = CH
3, 6R = H, 
8R = CH
3, 9OROOORROOORO7HMF
OOHOHOOHO1415%12be used to manufacture PET.  This is a beneficial feature of this particular synthesis, however 
Amoco Mid-Century oxidation would still be required.  A synthetic route to PTA from biobased 
starting materials that also affords small amounts of IPA allows PET manufacture to remain even 
if the BTX stream of petroleum refining disappeared.  The work reported herein demonstrates 
biobased 2-hydroxymuconic acid could be converted to mixtures of terephthalic, isophthalic, or 
phthalic acids without the need for Amoco Mid-Century oxidation, thus establishing a proof-of-
concept synthetic strategy that is not reliant on BTX. 
%13PART I
 SYNTHESIS OF AROMATIC DIACIDS FROM 2-HYDROXYMUCONIC ACID 
1.  Overview
  Microbe-synthesized 2-hydroxymuconic acid (2-HMA) is an attractive candidate as a 
precursor to biobased PTA (Figure 8).  Cycloaddition with bioethanol 
16-derived ethylene 
2 would provide a cyclohexene intermediate that could undergo facile dehydration and 
aromatization to afford PTA without the need for harsh, wasteful alkyl group oxidations.  This is 
similar to the previously reported synthesis of renewable PTA via reaction of ethylene 
2 with 
biobased muconic acid 
7.29  A key advantage to using 2-HMA as a synthetic intermediate over 
muconic acid 
7 is its associated intramolecular cyclization reaction leading to 6-carboxy-2-
pyrone 
19 (Figure 8).  A Diels-Alder reaction with methane-derived propiolic acid 
18 followed 
by aromatization affords IPA and PA.  All of the products from oxidation of petroleum-derived 
xylenes could thereby be prepared from 2-HMA obviating the need for the Amoco Mid-Century 
process.  Successful cycloadditions using unprotected carboxylic acids would be highly desirable 
as they circumvent additional esterification/hydrolysis steps. 
%14% Figure 8.  
Reaction network to biobased terephthalic acid, isophthalic acid, and phthalic acid via microbe-
synthesized 2-hydroxymuconic acid. 
 2-Hydroxymuconic acid is an intermediate in microbial catabolic pathways of aromatic 
compounds, such as catechol and protocatechuic acid (PCA).  Biocatalytic synthesis of 2-HMA 
would require a microbe capable of biosynthesis of either catechol or PCA that also expressed 
the respective degradation pathway enzymes.  Both of these aromatics have been previously 
produced in engineered 
E. coli
.38,39  The lower toxicity of PCA relative to catechol towards 
E. 
coli
 makes it a more attractive precursor to 2-HMA.  Recently, the PCA 2,3-cleavage pathway 
was elucidated in 
Paenibacillus 
sp. JJ-1b
40,41 (Figure 9).  Expression of PCA catabolic enzymes 
would enable the complete biosynthesis of 2-HMA in recombinant 
E. coli. 
OOHOHOHOHHOOHCO2HCO2HOHCO2HHOCO2HCO2HCO2HCO2HCO2H2-HMA
PTA
OOCO2HCO2HOOCO2HCO2HCO2HPACO2HIPA
CO2HCO2HOCO2HOCO2HCO22 CH43 H2H2OH2CO2CO2H2O16191817D-glucose
%15" Figure 9.  
Biosynthesis of 2-hydroxymuconic acid via protocatechuate 2,3-cleavage pathway described in 
Paenibacillus 
sp. JJ-1b. 
2.  Chemical synthesis of 2-hydroxymuconic acid and 6-carboxy-2-pyrone
  Chemically synthesized 2-HMA and 6-carboxy-2-pyrone 
19 were used to develop 
reactions leading to PTA, IPA, and PA.  Condensation of ethyl crotonate 
20 with diethyl oxalate 
21 affords diethyl 2-hydroxyhexa-2,4-dienedioate 
22 (Figure 10).  This allows direct access to 
either 2-HMA or 6-carboxy-2-pyrone 
19 upon ester hydrolysis or tandem hydrolysis-
lactonization, respectively.  Synthesis of 2-hydroxyhexa-2,4-dienedioate 
22 likely proceeds via 
deprotonation of a terminal allylic proton of ethyl crotonate 
20 to form a resonance stabilized 
enolate.  Nucleophilic attack upon diethyl oxalate 
21 affords a keto intermediate that undergoes 
rapid deprotonation at the 
!
-carbon to the ketone moiety. 
 CO2HOHOHCO2HCHOCO2HOHCO2HCHOOHOHCO2HCO2HPraA
O2PraH
CO2PraB
NAD
+PCA
2-HMS
2-HMA
5-CHMS
%16% Figure 10. 
 Synthesis of 2-hydroxymuconic acid and 6-carboxy-2-pyrone 
19.  (a)  K
0, (CH
3)3COH, toluene, 0 ¡C-rt, 
71%; (b) 1.  2 M NaOH, 2.  conc. H
2SO4, 4 ¡C, 59-64%; (c) H
2SO4, AcOH, H
2O, 100 ¡C, 85%.  
 H2SO4 Ethyl crotonate 
20 and diethyl oxalate 
21 were condensed at 0 ¡C in dry toluene 
with freshly prepared KOC(CH
3)3.  The reaction was allowed to warm to rt to afford 2-
hydroxyhexa-2,4-dienedioate 
22 in 71% isolated yield.  Ester hydrolysis to 2-HMA was 
performed in 2 M NaOH followed by treatment with concentrated H
2SO4 at 4 ¡C.  Acid 
treatment afforded 2-HMA as a beige precipitate which was washed in hot water and dried.  2-
HMA was collected in 59% isolated yield.  Lactonization and ester hydrolysis of 2-
hydroxyhexa-2,4-dienedioate 
22 to afford 6-carboxy-2-pyrone 
19 was conducted at 100 ¡C in a 
solvent system consisting of 1:4 H
2SO4:AcOH in water.  Extraction into EtOAc followed by 
solvent removal in vacuo afforded 6-carboxy-2-pyrone 
19 in 85% isolated yield.  Interconversion 
of 2-HMA and 6-carboxy-2-pyrone 
19 employed the same reaction conditions and achieved 
similar yields as synthesis via 2-hydroxyhexa-2,4-dienedioate 
22.  Isolated products were 
OOOOOOOOOHOOa+OOOHOObcHOOOHOHOOOCO2Hbc2-HMA
1922202122%17characterized by NMR spectroscopy (Figures 11-16)  All three products matched previously 
reported syntheses.
42   % Figure 11.  
1H NMR spectrum of 2-hydroxyhexa-2,4-dienedioate 
22 (CD3OD).
 O1O2OH3456OO91078%18% Figure  12
.  13C NMR spectrum of 2-hydroxyhexa-2,4-dienedioate 
22 (CD3OD). O1O2OH3456OO91078%19% Figure 13.  
1H NMR spectrum of 6-carboxy-2-pyrone 
19 (D2O). 
432O65OCO2H1%20% Figure 14
.  13C NMR of 6-carboxy-2-pyrone 
19 (DMSO-
d6). 432O65OCO2H1%21% Figure  15.  
1H NMR spectrum of 2-hydroxymuconic acid (CD
3OD).
 HO1O2OH3456OHO%22! Figure 16.
  13C NMR spectrum of 2-hydroxymuconic acid (DMSO-
d6). 3.  Synthesis of terephthalic acid via cycloaddition of 2-hydroxymuconic acid and ethylene 
 To understand the geometry of the conjugated 
(
 system in chemically synthesized 2-
HMA, NOE (nuclear Overhauser effect) data was collected.  This effect results from the transfer 
of spin polarization from one atom to another via cross-relaxation across interatomic space.  
Examination of vicinal proton coupling of protons on C-4 and C-5 suggested a 
trans
-geometry 
about the 
(
 bond (J3,4 = 15.6 Hz).  This offered two possible structures for 2-HMA (Figure 17).  
Transfer of spin polarization in NOE spectroscopy occurs at short interatomic distances.  Thus, 
differentiation between a 
2E- or 2Z-(
 bond could be discerned by exciting the proton on C-5.  
Carboxylic acid protons can be difficult to observe in 
1H NMR spectra.  To be able to clearly see 
HO1O2OH3456OHO%23any NOE effect of the carboxylate moiety by excitation at C-5, 2-HMA was derivatized to 
dimethyl 2-hydroxymuconate with trimethylsilyldiazomethane.  No NOE effect was observed for 
the methyl protons on C-1 when C-5 was irradiated indicating that the two proton environments 
are relatively far away in space.  This implies that the 
(
 bond between C2 and C3 likely has 
Z stereochemistry. 
% Figure 17.
 Possible geometries about the C2-C3 double bond. 
 HOMO and LUMO energy levels of 2-HMA and ethylene 
2 were calculated using 
Spartan Ô08 software and M¿ller-Plesset 6-31+G* basis set from AM1 initial geometry (Figure 
18)104.  The relative energies between the frontier molecular orbitals reveal that a cycloaddition 
would best proceed from overlap of the HOMO of ethylene 
2 and the LUMO of 2-HMA (Figure 
18) in an inverse electron demand Diels-Alder (IED DA) reaction.
43  Lewis acid catalysis can 
modulate the relative orbital energy levels of a substrate, facilitating overlap between the HOMO 
of one reaction partner and the LUMO of another.  Coordination of a Lewis acid to a substrate 
has two broad effects.  Firstly, relative energies of the LUMO and HOMO decrease according to 
the electron withdrawing strength of the Lewis acid with a concomitant decrease in the gap 
between the HOMO and LUMO energy levels.  An additional effect of Lewis acid coordination 
is a change in the sizes of the orbital coefficients at each carbon atom.  For example, a 
&
-CO2HHCO2HOH(2Z,4E)-2-HMA
CO2HHOHCO2H(2E,4E)-2-HMA
%24complex between the carbonyl oxygen at C-6 of 2-HMA and a classic Lewis acid such as TiCl
4 would result in an increase in the LUMO coefficients at C-2, C-4, and C-6, while decreasing the 
LUMO coefficients at C-1, C-3, and C-5.  In asymmetrical reactions, the sizes of the orbital 
coefficients often influences the regioselectivity of the products. 
%  Figure 18.
  Relative energy levels of the frontier molecular orbitals of 2-hydroxymuconic acid and ethylene 
2.  Catalysis of the reaction between 2-HMA and ethylene 
2 could involve either 
&
- or (
-electrophilic Lewis acids.
44  Classic Lewis acids such as BCl
3, TiCl
4, and AlCl
3 form strong 
&
-complexes with heteroatoms of carbonyl and imine groups.  On the other hand, salts of transition 
metals can behave as bifunctional Lewis acids, activating either carbon-carbon multiple bonds 
via 
(
-bonding or forming 
&
-complexes in the same manner as conventional Lewis acids.  
CO2HCO2HOH0 eV0.6 eV! 9.6 eV4.9 eV! 10.1 eV%25Platinum(II), gold(I), and copper(I) salts, in particular, have high propensities to form 
(
-complexes.
44,45,46  These metals may be able to lower the LUMO energy of ethylene 
2 sufficiently to promote reaction with the HOMO of 2-HMA in a normal electron demand Diels-
Alder reaction.  
 2-Hydroxymuconic acid and ethylene 
2 were first reacted in the absence of Lewis acid 
catalysis to determine the effect of thermal cycloaddition (Table 3).  Previous reaction conditions 
for the cycloaddition of 
2E,4E-muconic acid and ethylene 
2 were initially investigated (Entries 
1-3, Table 3).
47  A 20 mL titanium high-pressure reactor was charged with 2-HMA and a solvent.  
The reactor head was secured and the chamber was flushed with ethylene gas (3x) while stirring.  
After pressurizing with ethylene 
2, the reactor was heated over a period of 1-2 h to the reaction 
temperature after which the temperature was held constant.  Thermal cycloaddition resulted in 
low yields likely due to the large difference (10.7 eV) in HOMO-LUMO energy levels of the two 
substrates.  Significant charring was also observed resulting in poor mass balance.   
Table 3.  
Uncatalyzed cycloaddition between 2-hydroxymuconic acid and ethylene 
2.    The uncatalyzed cycloaddition appeared to be complete in 1-3 h as reaction times up to 
24 h did not afford substantially improved yields.  Solvent choice was the most significant 
Entry
Solvent
Reaction 
time  (h)
Temperature  
(¡C)
Ethylene 
pressure  (psi)
Yield PTA  
(mol %)
1bis(methoxypropyl) ether
2415025042water320050043m-xylene
3150250141,4-dioxane
120050055acetic acid
2420050016N,N-dimethylacetamide
242005001%26variable with 1,4-dioxane, bis(methoxypropyl) ether, and water affording >1% yields.  Variations 
of temperature and ethylene 
2 pressure had little effect on the yield of PTA.   
 Several catalysts were screened in the reaction of 2-HMA and ethylene 
2 using 1,4-
dioxane as a noncomplexing solvent (Table 4).  Conventional Lewis acids such as ScCl
3 and 
TiCl
4, which is known to catalyze cycloadditions with unprotected carbocylic acids,
37,48-51 afforded only marginal amounts of PTA (Entries 1-2, Table 4).  Similarly, La(OTf)
3, which has 
proven to be an effective Diels-Alder catalyst and can be quantitatively recovered and reused 
upon aqueous workup,
52,53 resulted in minor amounts of PTA (Entry 3).  Previous literature 
reports both Cu(OTf)
2- and ammonium zeolite-beta (NH
4-BEA)-catalyzed cycloadditions of 
ethylene 
2 and DMF at 250 ¡C with conversions of >90% and high selectivity.
23  Copper(II) 
triflate was the first Lewis acid tested to show appreciable activity in the screen (Entry 5, Table 
4).  Other base metal triflates such as Fe
2+ and Fe
3+ did not exhibit similar catalytic activity 
(Entries 6 and 7, Table 4). 
Table 4
.  Selected catalysts from screen of catalyzed reaction between 2-hydroxymuconic acid and ethylene 
2.  Lewis acidity can be modulated by altering the ligands surrounding a metal center.
54 Triflate groups are strongly electron withdrawing and produce hard, electrophilic metal atoms.
55  Entry
Catalyst
Yield PTA  (mol %)
1ScCl
3trace
2TiCl
413La(OTf)
314NH4-BEA
trace
5Cu(OTf)
2196Fe(OTf)
217Fe(OTf)
31%27Attempts to modulate the electrophilicity of the Cu
2+ metal center were made by varying the 
counterion of the copper salt.  Acetate was selected to provide a softer metal center while CuCl
2 was combined with AgPF
6 in an effort to generate a harder, electrophilic center.  A silver cation 
would form an insoluble salt with a chloride anion leaving a weakly coordinating 
hexafluorophosphate anion to complex with the Cu
2+ ion.  Neither catalyst exhibited pronounced 
activity (Table 5), indicating PTA formation is sensitive to electrophilicity about the Cu
2+ atom.  
It is possible that a precise complexation is required between Cu(OTf)
2 and 2-HMA which must 
be strong enough to lower the LUMO energy level but weak enough to allow dissociation and 
catalyst turnover.  Too electrophilic of a Cu
2+ may not be able to dissociate from the catalyst-
substrate complex, while not electrophilic enough would not be able to achieve sufficient LUMO 
energy level adjustment to promote cycloaddition. 
Table 5.  
Attenuating Cu
2+ electrophilicity. 
   
 Alterations of catalyst loading and reaction temperature during Cu(OTf)
2-catalyzed 
cycloadditions resulted in locating an optimal range of reaction conditions (Tables 6 and 7).  
Temperatures above 100 ¡C are required for PTA formation however significant decomposition 
of reaction components at temperatures above 200 ¡C preclude elevated yields.  At 200 ¡C, the 
highest yields of PTA were achieved using substoichiometric amounts of catalyst, though catalyst 
Entry
Catalyst
Yield PTA  
(mol %)
1Cu(OAc)
212CuCl2 " AgPF
60.5%28loading of at least 10% was required.  Charring of the reaction mixture was still observed under 
these conditions.    
Table 6.  
Effects of temperature on yield of PTA. 
Table 7.  
Effects of catalyst loading on yield of PTA. 
    
 The most interesting result from Cu(OTf)
2-catalyzed cycloadditions was the effect of a 
catalytic amount of Br¿nsted acid (Table 8, Entry 2).  Pronounced improvement in yield of PTA 
was observed when a drop of H
2SO4 was added to the reaction.  Previous research demonstrated 
that the addition of a small amount of Br¿nsted acid in Cu(OTf)
2-catalyzed benzannulations of 
alkylacetylenes with 
o-alkynyl(oxo)benzenes suppressed the formation of benzoic acid and 
benzoic anhydride byproducts and increased yields of desired naphthalene products.
111   The 
presence of a Br¿nsted acid in Cu(OTf)
2-catalyzed cycloaddition of 2-HMA and ethylene 
2 would enhance protonolysis of a Cu-
&
-complex and increase catalyst turnover.  An acidic 
Entry
Temperature  
(¡C)
Yield PTA  
(mol %)
13008220019310014500Entry
Mol %
Yield PTA  
(mol %)
110042501631019413%29medium would also promote dehydration of the cyclohexene intermediate formed after 
cycloaddition occurs.  A drop of water was added to the Cu(OTf)
2-catalyzed reaction as a control 
which resulted in a substantially decreased yield and the formation of benzoic acid along with a 
brick-red precipitate.   
Table 8.  
Effect of Br¿nsted acid on Cu(OTf)
2-catalyzed cycloaddition. 
   The Cu(OTf)
2-catalyzed reaction appeared highly sensitive to the presence of oxygen and 
water.  When oxygen was not rigorously excluded, reduced yields were observed in tandem with 
the formation of a brick-red precipitate visibly similar to Cu
2O.  In these reactions, small 
amounts of benzoic acid were seen as a byproduct.  Cuprous aromatic carboxylates readily 
undergo decarboxylations at temperatures of 150-250 ¡C.
58,59  Though the mechanism of 
copper(II) reduction during the cycloaddition of 2-HMA and ethylene 
2 was not experimentally 
verified, it is possible that ethylene 
2 reacts with trace amounts of  oxygen and water to afford 
ethylene glycol via ethylene oxide.  Heating ethylene glycol in the presence of oxygen affords 
glycolaldehyde, which is used as a reductant in the synthesis of metal nanoparticles including Pd, 
Ag, and Cu.
105-108  Incomplete reduction of copper in the presence of oxygen could afford a Cu
1+ species, catalyzing a protodecarboxylation of either terephthalic acid or a reaction intermediate.  
The catalytically active Cu
1+ species could be Cu
2O or another compound that is ultimately 
converted to Cu
2O.  Previous research has demonstrated that Cu
1+ compounds including Cu
2O, 
Entry
Additive
Yield PTA  
(mol %)
Yield benzoic 
acid  (mol %)
1none19trace
2H2SO42603H2O33%30are effective catalysts for the protodecarboxylation of various aromatic and heteroaromatic 
carboxylic acids.
60,61  Prior observation  of the necessity for unsaturation in the substrate acid, 
increasing rate with increasing polarizability, and 
(
-complexation propensity of the cuprous ion 
suggests a mechanism whereby a 
(
-bonded cuprous ion inductively stabilizes the resultant anion 
which develops as the carbon dioxide is lost.
61   % Figure 19.  
Putative mechanism for copper(I)-catalyzed protodecarboxylation of aromatic carboxylates. 
 Trapping experiments have suggested a 
&
-bonded organocopper intermediate, facilitated 
by the inherent stability of vinyl copper compounds.
62  More recent research suggests 
complexation of the cuprous ion to the C-CO
2 bond of the carboxylate via 
(
-complexation with 
the aromatic ring.
60  Oxidative addition would form a ternary copper(III) intermediate
63 which 
would undergo rapid reductive elimination to the arylcopper species and carbon dioxide (Figure 
19).  Insertion of copper into the C-C bond would be facilitated by 
(
-backbonding into the 
&
* orbital of the C-C bond.
60,61 To investigate the possible role of a putative copper(I) species in the 
cycloaddition of 2-HMA with ethylene 
2, CuOTf was employed in several reactions (Table 9).  
CuOTf-catalyzed cycloadditions required slightly elevated temperatures relative to Cu(OTf)
2-catalyzed reactions to achieve comparable yields.  An interesting result of CuOTf-catalyzed 
reactions was a significant amount of benzoic acid byproduct and accompanying brick-red 
precipitate.  These reactions indicate that protodecarboxylation of either terephthalic acid or 
Cu1+OOOOCu1+Cu1+OOCu3+OOCO2Cu1+%31another reaction intermediate likely occurs in the presence of a Cu
1+ species though the identity 
of such a species has not been demonstrated.      
Table 9.  
Copper(I)-catalyzed cycloaddition of 2-hydroxymuconic acid and ethylene. 
 Terephthalic acid was heated at 225¡C in 1,4-dioxane in the presence of 10 mol% CuOTf 
to determine whether decarboxylation of terephthalic acid was responsible for the formation of 
benzoic acid observed during the cycloaddition.  Despite the formation of a small amount of 
brick-red precipitate, no benzoic acid was detected.  Therefore, benzoic acid must form as a 
result of decarboxylation of a reaction intermediate rather than terephthalic acid. 
4.  Isophthalic acid and phthalic acid from cycloaddition of 6-carboxy-2-pyrone with 
propiolic acid 
 Ring closure of 2-HMA provides access to renewable IPA and PA upon reaction with 
methane-derived propiolic acid 
18 (Figure 20).  In 2013, annual biogas production was estimated 
to be 1.2 
) 1010 m3.64  This represents a massive amount of available carbon.  
Dehydrodimerization of methane at temperatures of 1500-1900 ¡C affords acetylene.  
Carboxylation of acetylene to propiolic acid 
18 utilizes copper catalyst in the presence of carbon 
dioxide.
65,66,67 Entry
Catalyst
Temperature 
(¡C)
Yield PTA  
(mol %)
Yield benzoic 
acid  (mol %)
1CuOTf
22511192CuOTf
17540.53CuOTf/
Cu(OTf)
220042%32% Figure 20.  
Synthesis of isophthalic acid (IPA) and phthalic acid (PA) from 6-carboxy-2-pyrone 
19 and renewable 
propiolic acid 
18..  Synthesis of aromatic dimethyl dicarboxylates from thermal cycloaddition of the methyl 
ester of 6-carboxy-2-pyrone 
19 with methyl propiolate has been established in the literature.
68  Reflux in mesitylene affords a 1.7:1 ratio of dimethylphthate:dimethylisophthalate in 
quantitiative yield.  No method has been established that utilizes the free acid form of 6-
carboxy-2-pyrone 
19 and propiolic acid 
18.  Nor has selectivity been driven toward the more 
useful 
meta-
substituted IPA.  Lewis acid catalysis offers a possible means to achieve each of 
these features. 
 Cursory examination of the energy levels of the frontier molecular orbitals of propiolic 
acid 
18 and 6-carboxy-2-pyrone 
19 using Spartan Ô08 software and M¿ller-Plesset 6-31+G* basis 
set from AM1 initial geometry,
104 reveals that the closest overlap would result from the LUMO 
of the propiolic acid 
18 dieneophile with the HOMO of the 6-carboxy-2-pyrone 
19 diene (Figure 
21).  This mode of orbital overlap represents a normal electron demand Diels-Alder reaction.  
2 CH43 H2CO2CO2HCO2HOOCO2HOOHO2CCO2HOOHO2CCO2HCO2HCO2HCO2HCO2HCO2181819IPA
PA1500-1900¡CCu+Previously established:
65,66,67This work:%33Both TiCl
437 and BCl
369 have been successfully used in catalyzing Diels-Alder cycloadditions 
with unsaturated carboxylic acid dieneophiles.  With precedent established, these Lewis acids 
were initially investigated for their catalytic propensity in the cycloaddition between 6-
carboxy-2-pyrone 
19 and propiolic acid 
18. %  Figure 21.
  Relative energy levels of the frontier molecular orbitals of 6-carboxy-2-pyrone 
19 and propiolic acid 
18.  Refluxing in toluene in the presence of 20 mol% TiCl
4 afforded a 2:1 ratio of PA:IPA in a 
combined yield of 30% mol/mol from 6-carboxy-2-pyrone 
19.  Charring resulted in significant 
amounts of insoluble black solids and low mass balance.  These solids were not observed in the 
BCl3-catalyzed reaction which appeared to facilitate decomposition of propiolic acid 
18 without 
promoting cycloaddition with the pyrone.  The results were disappointing as neither catalyst 
afforded significant amounts of product and selectivity in the TiCl
4-catalyzed reaction favored 
PA rather than the more useful IPA.  
0 eV0.9 eV! 9.9 eV1.6 eV! 12.1 eVOOCO2HCO2H%34Table 10.  
Initial Lewis acid-catalyzed reactions between propiolic acid 
18 and 6-carboxy-2-pyrone 
19 using 1:1 
mol:mol ratio propiolic acid 
18:6-carboxy-2-pyrone 
19.  Screening of Lewis acid catalysts was performed by another group member.  Interesting 
catalysts were those that exhibited selectivity toward IPA and those that afforded high 
conversions.  Thermal cycloaddition in the presence of a 7:1 molar excess of propiolic acid 
18, resulted in a 1.6:1 ratio of IPA:PA in a combined yield of 49% mol/mol relative to 6-carboxy-2-
pyrone 
19.  Selectivity was enhanced to a 3.3:1 ratio using 5 mol% SnCl
4 in a combined yield of 
32%.  Interestingly, the copper(II) salts, CuCl
2 and Cu(OTf)
2, exhibited 100% selectivity toward 
IPA albeit only in 7% yield.   
 Charring and decomposition continued to be a challenge for the cycloaddition of 
propiolic acid 
18 and 6-carboxy-2-pyrone 
19, precluding yields above 30-40%.  Furthermore, the 
appearance of phthalic anyhydride in reactions catalyzed by strong Lewis acids suggests facile 
dehydration of phthalic acid.  The water formed during the dehydration may react with the Lewis 
acid resulting in decreased activity over time.  The highest yields of aromatic products were 
observed with ZrCl
4 (81%), phenylboronic acid (PhB(OH)
2, 83%), and tetraacetyl diborate 
(BOB(OAc)
4, 90%).  However, none of the catalysts affording high yields also achieved high 
selectivities. 
 Lewis 
acid
Remaining 6-
carboxy-2-
pyrone
Remaining 
propiolic 
acid
Yield 
isophthalic acid  
(mol %)
Yield phthalic 
acid  (mol %)
TiCl
410%0%10%20%BCl395%2%2%0%%355.  Discussion 
 The results described above demonstrate that 2-HMA can serve as a synthetic precursor 
to all three aromatic diacids that are currently manufactured via oxidation of petroleum-derived 
xylenes.  The compound is stable as the 
2Z,4E stereoisomer as suggested by spin coupling of 
vicinal protons and nuclear Overhauser effect experiments, which is ideal for a Diels-Alder 
cycloaddition.  At 200 ¡C in dioxane, 10 mol% Cu(OTf)
2 and a catalytic amount of H
2SO4 affords PTA in a 26% yield with no benzoic acid byproduct.  
 Interestingly, Cu(OTf)
2 was the only catalyst examined to afford appreciable amounts of 
PTA.  Copper(II) triflate has demonstrated high conversions and selectivities in the reaction 
between ethylene 
2 and DMF to afford 
p-xylene 
1.  Both in the synthesis of 
p-xylene and in the 
work reported herein, attenuating the electrophilicity of Cu
2+ by varying the counterion resulted 
in decreased yields relative to Cu(OTf)
2.  This suggests that PTA and 
p-xylene 
1 formation via 
cycloaddition with ethylene 
2 is sensitive to the hardness of Cu
2+.  It is possible that the precise 
electron density about Cu
2+ in Cu(OTf)
2 allows coordination which is sufficient to lower the 
LUMO of 2-HMA to promote cycloaddition with ethylene 
2 while remaining labile enough for 
facile protonolysis.   
 The catalyst screen performed in this work was not exhaustive.  A more thorough 
investigation of different metals and appropriate counterions may succeed in identifying a system 
with greater catalytic potential than the Cu(OTf)
2-H2SO4 system presented here.   Indeed, such a 
system would be required before synthesis of PTA from 2-HMA garners commercial interest.  
Low conversions are acceptable only if unreacted starting material can be recovered and 
recycled.  This cannot currently be done.  Charring and decomposition is prevalent throughout 2-
%36HMA cycloadditions.  This is the principal challenge that needs to be solved before any synthesis 
from 2-HMA becomes industrially viable.   
 Synthesis of PTA from 2-HMA using Cu(OTf)
2 catalysis must be performed in a 
rigorously oxygen-free environment.  Despite drying 1,4-dioxane over Na
0 and preparing the 
reaction in a glove bag under N
2 atmosphere, trace amounts of O
2 and water are occasionally 
introduced.  In these reactions, decreased yields of PTA are obtained with concomitant formation 
of a brick red precipitate and benzoic acid.  It is likely that these solids are Cu
2O which has been 
reported to catalyze protodecarboxylation of aromatic carboxylates.  Conversion of ethylene 
2 to 
ethylene glycol followed by oxidation affords glycolaldehyde which is a known metal reductant.  
Reduction of Cu
2+ to Cu
1+ generates a species with high propensity for decarboxylation.  
Whether that species is Cu
2O or a different Cu
1+ species that is converted to Cu
2O during the 
reaction is not clear.  When CuOTf was added to the reaction instead of Cu(OTf)
2, benzoic acid 
was observed as the major product and significant amounts of red precipitate formed.  This 
supports our hypothesis that the formation of Cu
1+ initiates decarboxylation of a reaction 
intermediate.  Preventing the introduction of O
2 and water in a Cu(OTf)
2-catalyzed cycloaddition 
should be enough to avoid Cu
2+ reduction and accompanying decarboxylation.  
 Cyclization of 2-HMA to afford 6-carboxy-2-pyrone 
19 provides access to IPA and PA 
via Diels-Alder cycloaddition with propiolic acid 
18.  The cyclization reaction is performed 
using Br¿nsted acid catalysis in good yields (85%).  Cycloaddition of 6-carboxy-2-pyrone 
19 with propiolic acid requires different catalysis than the reaction of 2-HMA with ethylene.  While 
the HOMO-LUMO energies of 2-HMA and 6-carboxy-2-pyrone 
19 are very similar, the frontier 
molecular orbitals of ethylene 
2 and propiolic acid 
18 are not.  Propiolic acid 
18 bears an 
%37electron deficient alkyne and hence has a low energy LUMO, whereas ethylene 
2 is more 
electron rich with higher energy orbitals.  This difference in energy delineates a change from an 
inverse electron demand Diels-Alder reaction in the case of 2-HMA and ethylene 
2, and a normal 
electron demand Diels-Alder reaction for 6-carboxy-2-pyrone 
19 and propiolic acid 
18.    Conventional Lewis acids like BCl
3 and TiCl
4 form strong 
&
-complexes with carbonyl 
moieties.  Because both propiolic acid 
18 and 6-carboxy-2-pyrone 
19 possess carboxylic acid 
groups, such a catalyst would likely partition binding between both substrates and thus fail to 
effectively lower orbital energies for an entire population of a single compound.  Softer metals in 
lower oxidation states such as AuCl or PtCl
2 could be used to selectively complex the alkyne 
group of propiolic acid 
18.  These metals are known to form stable 
(
-complexes  with alkynes.  
This mode of binding would be selective towards propiolic acid 
18 and could promote sufficient 
LUMO energy lowering to facilitate the cycloaddition. 
 High yields of IPA and PA products are observed with PhB(OH)
2 (83%), and BOB(OAc)
4 (90%) but selectivities are low.  Selectivity toward IPA is desired because of its use in PET 
manufacture.  However, developing selectivity in a cycloaddition with 6-carboxy-2-pyrone 
19 is 
challenging due to the similarity in size of the HOMO coefficients about C-3 and C-6.  It is a 
difference in the orbital coefficients that often drives regioselectivity in Lewis acid-catalyzed 
reactions.  When the sizes of the orbital coefficients are similar, other factors such as steric 
hindrance must influence selectivity. 
 Though 2-HMA was used to afford PTA, IPA, and PA in cycloadditions with ethylene 
2 and propiolic acid 
18, several challenges must be addressed before the proposed route can be of 
commercial use.  Charring and decomposition of reaction mixtures must be prevented.  Nearly 
%38all Lewis acids examined for either cycloaddition involved some degree of degradation.  This is 
not sustainable as it precludes attempts to recycle unreacted substrates and requires fresh catalyst 
for each conversion.  Controlling the observed charring would dramatically increase the mass 
balance for each conversion and enable attempts to recover both catalyst and substrate.  
Increased yields of PTA and increased selectivity toward IPA are also highly desirable.  If a 26% 
yield was achieved during each conversion, unreacted starting material would need to be 
recycled over 15 times to reach the same yield of PTA as is currently achieved from a single 
Amoco Mid-Century oxidation of 
p-xylene 
1.   Despite high yields from BOB(OAc)
4-catalyzed reaction of 6-carboxy-2-pyrone 
19 with 
propiolic acid 
18, further development should enhance selectivity toward IPA.  While PA is used 
for manufacture of plasticizers such as DINCH and DINP, it is not manufactured to the same 
scale as IPA, which is used for the production of PET.  Therefore it would be useful to achieve 
selectivities that match the market value of the products.   
 Though challenges remain, this work demonstrates that the synthesis of all three aromatic 
diacids that are currently manufactured from petroleum can be achieved from a single synthetic 
precursor, 2-HMA.  The proposed route does not require Amoco Mid-Century oxidations and 
does not involve esterification/hydrolysis steps.  The catalysts used are relatively inexpensive 
and all substrates can be prepared from renewable starting materials.  Ethylene 
2 is synthesized 
from bioethanol, propiolic acid 
18 is produced from biogas methane, and the synthesis of 2-
HMA from 
D-glucose will be discussed in Part II. 
%39PART II
 BIOCATALYTIC SYNTHESIS OF 2-HYDROXYMUCONIC ACID FROM 
D-GLUCOSE 
1.  Overview 
 Protocatechuic acid is a central intermediate in several degradation pathways of aromatic 
compounds including phthalates, hydroxybenzoates, and lignin-derived aromatics.  Aromatic 
ring cleavage catalyzed by one of three distinct dioxygenases, PCA 3,4-dioxygenase,
70 PCA 4,5-
dioxygenase,
71 and PCA 2,3-dioxygenase,
72 initiates different degradation pathways.  Ring 
fission catalyzed by PCA 3,4-dioxygenase affords 2-carboxy-
cis,cis-muconic acid which enters 
the 
#
-ketoadipate pathway reported in several organisms.
70,73  In the 4,5-cleavage pathway, PCA 
conversion to 4-carboxy-2-hydroxymuconate-6-semialdehyde precedes further degradation 
which ultimately affords 4-carboxy-4-hydroxy-2-oxoadipate before entering the citric acid cycle.
71a  Prior research identifies PCA 4,5-cleavage in 
Sphingomonas
,71a Comamonas
,74 Pseudomonas
,75 and
 Arthrobacter
76 strains.    
 Paenibacillus 
sp. JJ-1b (formerly 
Bacillus macerans
) exploits 4-hydroxybenzoate (4-
HBA) catabolism as a sole source of carbon and energy via the PCA 2,3-cleavage pathway 
(Figure 22).
72a  In the early stages of the pathway, PCA 2,3-dioxygenase catalyzes extradiol ring 
cleavage of PCA to (2
E,4E)-5-carboxy-2-hydroxymuconate-6-semialdehyde (5-CHMS).  Rapid 
disappearance of the immediate ring-fission product in cell extracts and accumulation of (2
E,4E)-2-hydroxymuconate-6-semialdehyde (2-HMS) suggested rapid enzymatic decarboxylation 
of the former compound.  By 1993, only one enzyme, PCA 2,3-dioxygenase, had been purified 
and characterized (
KmO2 = 142 $
M, 
KmPCA
 = 24 $
M, 
kcat = 210 s*1).77 %40%   Figure 22.
  Degradation of 4-hydroxybenzoic acid via the PCA 2,3-cleavage pathway in 
Paenibacillus
 sp. JJ-1b.
  In 2009, all of the genes involved in 4-HBA degradation via the PCA 2,3-cleavage 
pathway were identified from 
Paenibacillus 
sp. JJ-1b (Figure 22).
40  Based on 
in vitro
 experiments using cell extracts of 
E. coli
 carrying 
praI
, praA
, praH
, praB
, praC
, or praD
 these 
genes were discovered to encode 4-HBA 3-hydroxylase, PCA 2,3-dioxygenase, 5-carboxy-2-
hydroxymuconate-6-semialdehyde decarboxylase, 2-hydroxymuconate-6-semialdehyde 
dehydrogenase, 4-oxalocrotonate tautomerase, and 4-oxalocrotonate decarboxylase, respectively.  
Additional genes associated with the 
pra
 operon were reported to catalyze the degradation of the 
resultant 2-oxopent-4-enoate (2-OP) to pyruvate and acetyl CoA. 
 Engineering recombinant 
E. coli 
for de novo biosynthesis of 2-HMA from 
D-glucose 
(Figure 23) requires expression of PCA anabolic and catabolic enzymes.  Biosynthesis of PCA 
via the shikimate pathway has been previously reported using 
E. coli 
strain WN1.
38,39,47  This 
strain lacks a catalytically active shikimate dehydrogenase encoded by 
aroE
 and expresses DHS 
dehydratase from 
Klebsiella pneumoniae
 encoded by 
aroZ. 
 Biosynthesis of 2-HMA can be 
CO2HOHCO2HOHOHNAD(P)H
O2PraI
CO2HCHOCO2HOHCO2HCHOOHOHCO2HCO2HPraA
O24-HBA
PraH
CO2PraB
NAD
+PCA
2-HMS
2-HMA
5-CHMS
OHCO2HCO2H2-HMA
PraC
OCO2HCO2HPraD
OCO2H2-OP
CO2PraE
H2OHOCO2HOPraF
CO2HOHOPraG
NAD
+coenzyme A
CoAO%41realized by combining PCA anabolism from WN1 with PCA catabolic enzymes from 
Paenibacillus 
sp. JJ-1b. 
% Figure 23.  
De novo biosynthesis of 2-hydroxymuconic acid combining protocatechuic acid anabolism and 
catabolism.
 2.  Host strain 
 Strains and plasmids used in this work are presented in Table 11.  
E. coli 
WN1 was used 
as the host strain for 2-HMA synthesis.  The host strain was derived from 
E. coli
 AB2834 and 
lacks a catalytically active shikimate dehydrogenase.
79  This mutation prevents hydrogenation of 
PCA
CO2HHOOHNAD
+HOCO2HCO2HPraB
CO2PraH
O2PraA
2-HMS
2-HMA
5-CHMS
OHOHOOHOHHOOHOH2O3POOHOHOHH2O3PO+E4P
PEPOCO2HHOOHOHH2O3PODAHP
CO2HHOOHOHODHQOHOHODHSCO2HOHOHHOSACO2HAroF
FBR
H2OAroB
PiAroD
H2OAroE
NADH
CO2HHO2COHOCHOHO2COHH2OAroZ
D-glucose
PiNAD
+%423-dehydroshikimate (DHS) to shikimate in the common pathway of aromatic amino acid 
biosynthesis.  Further modification of the host genome included homologous recombination of 
an 
aroBaroZ
 cassette into the 
serA locus
79 and insertion of a 
tktAaroZ
 cassette into the 
lacZ 
locus.
80  Additional genomic copies of 
aroB 
and 
aroZ
 prevent the accumulation of 3-deoxy-
D-arabino
-heptulosonate 7-phosphate (DAHP) and DHS, respectively.  An additional copy of 
tktA on the genome helps to direct carbon flux from the pentose phosphate pathway into the shikimate 
pathway and increases the intracellular availability of erythrose 4-phosphate (E4P).  Since 
inactivation of the shikimate dehydrogenase precludes de novo biosynthesis of aromatic amino 
acids and aromatic vitamins, the growth of all constructs in minimal media required 
supplementation with 
L-phenylalanine, 
L-tryptophan, and 
L-tyrosine along with 
p-hydroxybenzoic acid, 
p-aminobenzoic acid, and 2,3-dihydroxybenzoic acid. 
Table 11.
  Strains and plasmids used in Chapter 1 of this study. 
Strain/plasmid
Relevant characteristics
Source or reference
Escherichia coli 
DH5
!FÕ 
"80lacZ
+M15 +(lacZYA-argF)U169
 deoR recA1 endA1 
hsdR17
(rk*, mk+) phoA 
#* supE44 thi-1 gyrA96 relA1
Invitrogen
Escherichia coli
 AB2834
tsx-352 glnV42 aroE353 malT352 
#$Coli Genetic Stock 
Center
78Escherichia coli 
KL7AB2834 
serA::aroBaroZ
lab79Escherichia coli
 WN1KL7 
lacZ
::tktAaroZ
lab80Escherichia coli
 KM1WN1 
pptA
::FRT
this work
Paenibacillus 
sp. 
JJ-1b
source for 
praApraBpraH
ATCC 35889
pBbA1a-RFP
ApR, lacI
q, Ptrc
rfp p15A replicon
Addgene #35334
92pWN1.028A
aroF
FBR
 source
lab80%43  3.  Plasmid construction 
 2-HMA-synthesizing constructs employed plasmid pKM6.280 (Table 11, Figure 33) 
which carried 
praBpraHpraAaroF
FBR
serAParoF
 under the control of a 
trc
 promoter.  We 
anticipated that arranging the PCA 2,3-cleavage enzymes in reverse order relative to the 
reactions they catalyze would ensure carbon is pulled entirely through the pathway.  The 
serA 
gene encodes 3-phosphoglycerate dehydrogenase, which is necessary for biosynthesis of 
L-serine 
in 
E. coli.  
Host strain WN1 possesses a mutated genomic copy of 
serA 
and is therefore incapable 
of growth in media lacking 
L-serine supplementation.  The inclusion of a functional 
serA gene on 
a plasmid restores the ability of the cell to synthesize 
L-serine, allowing growth in minimal salts 
medium only if the cell maintains the plasmid.  Nutritional rather than antibiotic selection is an 
extensively used plasmid maintenance strategy.
38,79-87 Table 11 (contÕd)
pWN1.162A
serA source
lab80pMF63A
ParoF
 source
lab88pKM6.196
Ptrc
praA
 in pBbA1a
this work
pKM6.200
Ptrc
praB
 in pBbA1a
this work
pKM6.201
Ptrc
praH 
in pBbA1a
this work
pKM6.208
Ptrc
praHpraA
 in pBbA1a
this work
pKM6.232
Ptrc
praBpraHpraA
 in pBbA1a
this work
pKM6.240
Ptrc
serA in pBbA1a
this work
pKM6.241
Ptrc
ParoF
 in pBbA1a
this work
pKM6.261
Ptrc
serAParoF
 in pBbA1a
this work
pKM6.268
Ptrc
aroF
FBR
 in pBbA1a
this work
pKM6.274
Ptrc
aroF
FBR
serAParoF
 in pBbA1a
this work
pKM6.280
Ptrc
praBpraHpraAaroF
FBR
serAParoF
 in pBbA1a
this work
%44 Efficient synthesis of PCA from 
D-glucose requires a mutant isozyme of DAHP synthase, 
an enzyme which catalyzes the condensation of erythrose-4-phosphate and phosphoenolpyruvate 
to DAHP.
89  Three DAHP synthase isozymes, AroF, AroG, and AroH, are sensitive to repression 
by one of three aromatic amino acids, 
L-tyrosine, 
L-phenylalanine, or 
L-tryptophan, respectively.  
Because of the mutation in 
aroE
-encoded shikimate dehydrogenase, the 
E. coli
 strains used in 
this study cannot synthesize aromatic amino acids, which necessitated supplementing the culture 
medium with these compounds.  At concentrations above 10 
$
M, 
L-tyrosine exhibits 
noncompetitive inhibition of native 
E. coli 
AroF with respect to erythrose-4-phosphate and 
competitive inhibition with respect to phosphoenolpyruvate.
89b  To direct carbon flux into the 
shikimate pathway, plasmids included 
aroF
FBR
, encoding a tyrosine feedback resistant isozyme 
of AroF.
89a  In addition, transcription of 
aroF
FBR
 is repressed by a complex between the native 
E. 
coli tyrR
 gene product and 
L-tyrosine.
89a  The binding sites for this repressor complex overlap 
with the 
aroF 
promoter sequence, 
ParoF
.  To mitigate transcriptional repression, an extra copy of 
ParoF
 was included on the plasmid.
88  Additional binding sites can titrate away the tyrosine 
repressor complex and improve transcription of the 
aroF
FBR
 gene product.
89,90,91  The construction of pKM6.280 was performed using the BglBrick assembly standard, 
developed by Anderson and Keasling (Figures 25-33).
92,93  This method of DNA assembly 
expedites the process of creating the desired construct and facilitates construct modifications 
such as changing the order of genes in a cassette and the removal or addition of various genetic 
elements.  The creation of BglBrick parts requires genetic elements including, coding sequences 
for the open reading frames, promoters, and terminators to be individually inserted into a vector 
with standardized prefix and suffix sequences (Figure 24).  These sequences encode specific 
%45restriction enzyme sites.  The 5Õ prefix sequence encodes 
EcoRI and 
Bgl
II sites, while the 3Õ 
suffix sequence encodes 
Bam
HI and 
Xho
I restriction sites.  Each open reading frame also 
included a 5Õ Shine-Dalgarno sequence (AAGGAGG).  
  % Figure 24.
  Schematic representation of BglBrick DNA assembly method.  Part B is digested with 
Bgl
II and 
Xho
I restriction enzymes while part A is digested with 
Bam
HI and 
Xho
I restriction enzymes.  Ligation of part B into part 
A results in a glycine-serine scar sequence and a new part comprising the original parts A and B.  This mode of 
insertion is called suffix-insertion. Figure adapted from reference 109.
    The compatible DNA ends of 
Bgl
II and 
Bam
HI restriction enzymes promotes 
concatenation of any two genetic elements, creating a glycine-serine scar.  Insertion of one part 
into another generates a new part that retains the original four restriction enzyme sites on the 5Õ 
and 3Õ ends.  In this way, assembling an arbitrary number of parts in any desired arrangement is 
no more difficult than ligating two together.  A further advantage of the BglBrick assembly 
method is that once a genetic element is inserted into the standard vector to create a BglBrick 
part, there is never a need to PCR-amplify that particular element which eliminates the 
probability of PCR-derived mutations during a cloning project.
92,93   %46 The genes, 
praA
, praH
, and 
praB
, were PCR amplified from 
Paenibacillus
 sp. JJ-1b 
genomic DNA and extended with 5Õ 
Bgl
II and Shine-Dalgarno sequences and 3Õ 
Bam
HI and 
Xho
I sites (Figure 25).  PCR amplified genes were inserted into the 
Bgl
II and 
Xho
I sites of 
BglBrick vector, pBbA1a.
93  The 
serA gene, 
aroF
FBR
 gene, and 
ParoF
 sequence were similarly 
amplified from pWN1.162, pWN1.028A, and pMF63A, respectively (Figures 26-28).  
Concatenation of 
praBpraHpraA
 genes resulted in plasmid pKM6.232 (Figure 30) while 
assembly of 
aroF
FBR
serAParoF
 afforded pKM6.274 (Figure 32).  Insertion of 
praBpraHpraA
 into 
pKM6.274 followed the prefix insertion method to afford pKM6.280 (Figure 33). 
%47! Figure 25. 
 Preparation of 
pra
 BglBrick parts: plasmids pKM6.196, pKM6.200, and pKM6.201.
 0.8 kB
1.5 kB
1.0 kB
Paenibacillus 
sp. JJ-1b
1)  PCR
2)  
Bgl
II, Xho
I digest
Bgl
II, Xho
I digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IpraB
praH
AmpRlacI
qp15Aori
mRFP-1
pBbA1a-RFP
4.3 kb
Ptrc
AmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
praA
praB
praH
pKM6.196
4.4 kB
pKM6.200
5.0 kB
pKM6.201
4.5 kB
praA
Bgl
IIXho
IBgl
IIXho
IBgl
IIXho
I%48% Figure 26.  
Preparation of 
serA BglBrick part, plasmid pKM6.240. 
1.2 kB
pWN1.1621)  PCR
2)  
Bgl
II, Xho
I digest
Bgl
II, Xho
I digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
mRFP-1
pBbA1a-RFP
4.3 kb
Ptrc
AmpRlacI
qp15Aori
Ptrc
serA
pKM6.240
4.8 kB
serA
Bgl
IIXho
I%49% Figure 27.  
Preparation of 
aroF
FBR
 BglBrick part, plasmid pKM6.268. 
1.2 kB
pWN1.028A1)  PCR
2)  
Bgl
II, Xho
I digest
Bgl
II, Xho
I digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
mRFP-1
pBbA1a-RFP
4.3 kb
Ptrc
AmpRlacI
qp15Aori
Ptrc
aroF
FBR
pKM6.268
4.8 kB
aroF
FBR
Bgl
IIXho
I%50% Figure 28.  
Preparation of 
ParoF
 BglBrick part, plasmid pKM6.241. 
0.1 kB
pMF63A
1)  PCR
2)  
Bgl
II, Xho
I digest
Bgl
II, Xho
I digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
mRFP-1
pBbA1a-RFP
4.3 kb
Ptrc
AmpRlacI
qp15Aori
Ptrc
ParoF
pKM6.241
3.7 kB
ParoF
Bgl
IIXho
I%51% Figure 29.  
Preparation of plasmid pKM6.208. 
1.0 kB
pKM6.201
EcoR
I, BamH
I digest
EcoR
I, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IpraH
AmpRlacI
qp15Aori
Ptrc
praH
pKM6.208
5.3 kB
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
praA
pKM6.196
4.4 kB
praA
EcoR
IBamH
I%52% Figure 30.  
Preparation of plasmid pKM6.232. 
1.5 kB
pKM6.200
EcoR
I, BamH
I digest
EcoR
I, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IpraB
AmpRlacI
qp15Aori
Ptrc
praH
pKM6.232
6.8 kB
praA
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
praH
pKM6.208
5.3 kB
praA
praB
EcoR
IBamH
I%53% Figure 31.  
Preparation of plasmid pKM6.261. 
1.2 kB
pKM6.240
EcoR
I, BamH
I digest
EcoR
I, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ParoF
pKM6.241
3.7 kB
serA
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ParoF
pKM6.261
5.0 kB
EcoR
IBamH
IserA
%54% Figure 32.  
Preparation of plasmid pKM6.274. 
1.1 kB
pKM6.268
EcoR
I, BamH
I digest
EcoR
I, Bgl
II digest
T4 ligationaroF
FBR
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ParoF
pKM6.274
6.2 kB
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ParoF
pKM6.261
5.0 kB
aroF
FBR
serA
serA
EcoR
IBamH
I%55% Figure 33.  
Preparation of plasmid pKM6.280. 
3.2 kB
pKM6.232
EcoR
I, BamH
I digest
EcoR
I, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IpraBpraHpraA
AmpRlacI
qp15Aori
Ptrc
praH
pKM6.280
9.5 kB
serA
praB
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ParoF
pKM6.274
6.2 kB
aroF
FBR
serA
ParoF
praA
aroF
FBR
EcoR
IBamH
I%564.  Fed-batch fermentation conditions 
 Fed-batch fermentations were performed in a 2 L working volume bench-top bioreactor 
at 33 ¡C and pH 7.0.  The dissolved oxygen (D. O.) level was maintained at 10% air saturation.  
Antifoam (Sigma 204) was manually added as needed and pH maintained at 7.0 by addition of 
NH4OH and 2 N H
2SO4.  Initial glucose concentrations were 18 g/L for glucose-limited 
conditions or 25 g/L for glucose-rich conditions. 
 Fermentation inoculants were initiated by introduction of a single colony into 5 mL of 
M9 medium and grown at 37 ¡C with agitation at 250 rpm for 20-24 h.  This culture was used to 
inoculate 100 mL M9 medium to an initial OD
600 of 0.05 and grown for an additional 6 h at 37 
¡C and 250 rpm.  After reaching an appropriate OD
600 (1.0-1.5), the inoculant was transferred to 
the fermentation vessel (t = 0).  
4.1.  Glucose-rich conditions 
 Fermentations run under glucose-rich conditions employed an initial glucose 
concentration of 25 g/L.  
D-Glucose addition rates were adjusted manually such that the 
concentration of 
D-glucose in the medium was maintained in the range of 20-30 g/L.  
Fermentation runs are divided into three stages, each of which corresponds to a different control 
procedure for maintaining D. O. levels at 10%.  In the first stage, D. O. concentration was 
maintained by increasing the impeller speed from its initial set point of 50 rpm to a preset 
maximum of 750 rpm while airflow was held at its initial setting of 0.06 L/L/min.  In the second 
stage, the impeller speed was held constant at 750 rpm while the mass flow controller increased 
airflow to a preset maximum level of 1.0 L/L/min.  In the third stage, D. O. levels were 
%57maintained at 10% by allowing the impeller speed to vary between 400-1800 rpm while the 
airflow was held constant at 1.0 L/L/min. 
4.2.  Glucose-limited conditions 
 Fermentations run under glucose-limited conditions employed an initial glucose 
concentration of 18 g/L and a steady state concentration of less than 1 g/L of glucose was 
maintained after the preset maximum levels of airflow (1.0 L/L/min) and impeller speed (850 
rpm) were reached.  Glucose-limited fermentations involve three stages similar to glucose-rich 
runs.  Indeed, the first two stages are controlled identically.  After reaching the preset maximum 
levels of impeller speed and airflow, oxygen sensor-controlled glucose feeding began, which 
maintained D. O. levels at 10% throughout the remainder of the run while keeping impeller 
speed and airflow constant.  
5.  Biosynthesis of 2-HMA  in 
E. coli
 5.1.  Fermentation of
 E. coli 
WN1/pKM6.280
  Fermentations of 
E. coli
 WN1/pKM6.280 employed both glucose-rich and glucose-
limited conditions.  Beginning at 12 h and continuing every 6 h throughout the fermentation run, 
aliquots of 
#
-D-isopropylthiogalactopyranoside (IPTG, 12 mg, 0.05 mM) were added to induce 
gene transcription.  Glucose-limited fermentations afforded marginally higher titers of 2-HMA 
than glucose-rich conditions (60 mg/L and 30 mg/L, respectively) (Figures 40 and 41).  
Unexpectedly, 1-2 g/L of an 
!
-keto acid tautomer, (
E)-5-oxohex-2-enedioc acid 
23, were 
%58observed in the fermentation broth by 
1H NMR.  Small amounts (100-500 mg/L) an 
!
-keto 
tautomer, (
E)-2-oxohex-3-enedioic acid 
24, were also seen. 
%                     %  Figure 34
.  Biosynthesis of 2-hydroxymuconic acid by 
E. coli
 WN1/pKM6.280 under glucose-rich conditions.  
Legend:  2-HMA, black columns; tautomer 
23, grey columns; tautomer 
24, open columns; dry cell weight, closed 
circles. 
%                 % Figure 35.
  Biosynthesis of 2-hydroxymuconic acid by 
E. coli
 WN1/pKM6.280 under glucose-limited conditions.  
Legend:  2-HMA, black columns; tautomer 
23, grey columns; tautomer 
24, open columns; dry cell weight, closed 
circles. 
Dry cell weight  (g/L)0102030402-HMA isomers  (g/L)00.511.52Time  (h)121824303642CO2HHO2CHO2-HMA
CO2HHO2CO23CO2HHO2CO24Dry cell weight  (g/L)0102030402-HMA isomers  (g/L)00.511.52Time  (h)121824303642CO2HHO2CHO2-HMA
CO2HHO2CO23CO2HHO2CO24%59 Alongside 2-HMA isomers, DHS was observed in fermentation broth at concentrations of 
2.2 g/L and 1.4 g/L in glucose-rich and glucose-limited runs, respectively.  PCA was not 
observed indicating that any PCA formed was likely consumed by the expressed 
pra
 enzymes.  
To examine whether 2-HMA affects PCA anabolism, 
E. coli
 WN1/pKM6.274 was employed in 
subsequent fermentations both with and without exogenous 2-HMA. 
  5.2.  An examination of PCA biosynthesis from 
D-glucose 
 Plasmid pKM6.274 contains 
aroF
FBR
serAParoF
 under the control of a 
trc
 promoter.  Both 
glucose-rich and glucose-limited fermentations of 
E. coli
 WN1/pKM6.274 enabled evaluation of 
PCA biosynthesis productivity in the absence of any PCA cleavage enzymes (Figures 42 and 43). 
%   Figure 36.  
Biosynthesis of PCA by 
E. coli
 WN1/pKM6.274 under glucose-rich conditions.  Legend: 
protocatechuate (PCA), grey columns; 3-dehydroshikimate (DHS), black columns; dry cell weight, black circles. 
   
Dry cell weight  (g/L)010203040DHS, PCA  (g/L)0481216Time  (h)1218243036%60% Figure 37.  
Biosynthesis of PCA by 
E. coli
 WN1/pKM6.274 under glucose-limited conditions.  Legend: 
protocatechuate (PCA), grey columns; 3-dehydroshikimate (DHS), black columns; dry cell weight, black circles. 
 Glucose-limited conditions afforded PCA titers of 4.8 g/L while glucose-rich 
fermentations achieved 15 g/L titers of PCA.  Both conditions revealed maximum DHS titers by 
24 h.  Beyond 24 h, DHS titers slowly decreased throughout the remainder of the fermentation, 
likely due to either dilution effects or small amounts of cellular uptake.
38  Chemically 
synthesized 2-HMA was added to exponentially growing WN1/pKM6.274 to determine if 2-
HMA accumulation has an effect on PCA production (Figure 38). 
Dry cell weight  (g/L)010203040DHS, PCA  (g/L)0481216Time  (h)1218243036%61% Figure 38.  
Biosynthesis of PCA and DHS by 
E. coli 
WN1/pKM6.274 in the presence of externally added 2-HMA 
(1 g/L).  Legend:  protocatechuate (PCA), grey columns; 3-dehydroshikimate (DHS), black columns; dry cell 
weight, black circles. 
 Once cells reached exponential growth phase (18 h), chemically synthesized 2-HMA was 
added as a sterile solution in water (pH 7) to a concentration of 1 g/L.  Interestingly, no 2-HMA 
nor either 
!
-keto acid tautomers were observed in the fermentation broth after 6 h following 
addition.  To ensure that the exogenously added 2-HMA does not undergo chemical degradation, 
a 500 mL baffled flask containing 100 mL fermentation medium and 1 g/L 2-HMA was 
incubated at 37 ¡C with agitation at 250 rpm.  After 3 h, 
!
-keto acid tautomer 
23 predominated in 
a 9:1 ratio with 2-HMA.  This distribution remained unchanged after 24 h.  
!
-Keto acid tautomer 
24 did not accumulate.  These results indicate the tautomerization of 2-HMA to 
!
-keto acid 
23 is 
facile, but enzyme action is necessary for the production of 
!
-keto acid 
24.  All of the mass of the 
initially added 0.1 g of 2-HMA could be accounted for after 24 h, indicating that chemical 
degradation does not occur under these conditions.  An observed loss of 1 g/L 2-HMA over 6 h 
during fermentation of WN1/pKM6.274 suggests gradual cellular uptake.  Despite apparent 
Dry cell weight  (g/L)010203040DHS, PCA  (g/L)0481216Time  (h)1218243036%62incorporation into the cell, 2-HMA did not appear to have deleterious effects on PCA or DHS 
biosynthesis.  
E. coli 
WN1/pKM6.274 achieves similar titers of PCA and DHS both in the 
presence and absence of 2-HMA.  It appears that 2-HMA does not inhibit PCA biosynthesis; 
however, the possibility of inhibition due to CHMS or 2-HMS remains. 
 A critical difference between pKM6.280 and pKM6.274 is the location of 
aroF
FBR
 relative to the 
trc
 promoter.  Premature termination of 
E. coli
 RNA polymerase negatively effects 
the translation rate of genes placed far away from the promoter sequence.
94  In plasmid 
pKM6.274, the 
aroF
FBR
 gene is the first gene in the cassette, while in plasmid pKM6.280, it is 
over 3 kilobases away.  The difference in distance of 
aroF
FBR
 away from 
Ptrc
 may lead to a 
difference in translation rate and a corresponding difference in AroF
FBR
 expression between 
pKM6.274 and pKM6.280.  To determine if there is a difference in AroF
FBR
 expression, DAHP 
synthase activity was measured in both WN1/pKM6.274 and WN1/pKM6.280.   
 A comparison of DAHP synthase activities of WN1/pKM6.274 and WN1/pKM6.280 
employed a colorimetric assay described by Schoner.
95  Briefly, E4P and phosphoenolpyruvate 
(PEP) are converted into DAHP during incubation with cellular lysate at 37 ¡C over 5 min.  
Aliquots are removed at timed intervals and quenched with 10% trichloroacetic acid.  
Precipitated protein is removed and a pink chromophore is generated by reaction of DAHP with 
thiobarbituric acid.  Quantification of DAHP proceeded via extraction of the chromophore into 
cyclohexanone followed by monitoring absorbance at 549 nm (
,
 = 68,000 M*1 cm
*1).  One unit 
of DAHP synthase activity was defined as the formation of 1 
$
mol of DAHP per min at 37 ¡C. 
 Four aliquots of cells were taken from glucose-limited fermentation runs with WN1/
pKM6.274 and WN1/pKM6.280 (Tables 12 and 13).  Aliquots were taken every 2.5 h beginning 
%6330 min after addition of IPTG.  Cells were pelleted, lysed, and assayed for DAHP synthase 
activity.  Steady activity was detected in WN1/pKM6.274 cell-free extracts whereas WN1/
pKM6.280 samples exhibited lower initial DAHP synthase activity that increased over time.  
After 20 h, DAHP synthase activities for both strains were similar.  The reason behind the lower 
initial activity observed for WN1/pKM6.280 was not proven, but if it is a consequence of 
premature termination events during transcription of 
aroF
FBR
 by E. coli 
RNA polymerase, 
insertion of an additional 
trc
 promoter preceding the 
aroF
FBR
 gene in plasmid pKM6.280 could 
alleviate this initial lag and improve carbon flow into the shikimate pathway during the early 
stages of fermentation. 
Table 12.
  Specific activity of DAHP synthase in cellular lysate of WN1/pKM6.274 during glucose-limited 
fermentation.
  Gene transcription induced by addition of aliquots of IPTG (12 mg, 0.05 mM) at 12 h and 18 h into course of 
fermentation. 
Table 13.
  Specific activity of DAHP synthase in cellular lysate of WN1/pKM6.280 during glucose-limited 
fermentation.
 Gene transcription induced by addition of aliquots of IPTG (12 mg, 0.05 mM) at 12 h and 18 h into course of 
fermentation. 
Run time
Sp. activity  (
#
mol/min/mg)
12.5.h1215.h917.5.h1020.h8Run time
Sp. activity  (
#
mol/min/mg)
12.5.h115.h217.5.h420.h7%64 To investigate possible feedback inhibition of AroF
FBR
 by 2-HMA or 2-HMS, DAHP 
synthase activity of WN1/pKM6.274 cell free lysate was assayed in the presence of these 
compounds (Table 14).  2-Hydroxymuconic acid was synthesized as described above in Part I 
and isolated as a dry solid.  A known amount of this material was dissolved in water and filtered 
through a 0.22 
$
m membrane to generate a stock solution.  2-Hydroxymuconate 6-semialdehyde 
(2-HMS) was made 
in situ
 via action of 6xHis-tagged PraA and 6xHis-tagged PraH on PCA prior 
to addition of DAHP-containing sample.  Quantification of 2-HMS was performed by 
monitoring the increase in absorbance at 375 nm (
,
 = 35,000 M*1 cm
*1).40  Neither compound 
exhibited inhibition of AroF
FBR
 up to concentrations of 1 mM or 10 mM for 2-HMS or 2-HMA, 
respectively. 
Table 14.
  Effect of exogenous 2-hydroxymuconic acid and 2-hydroxymuconate 6-semialdehyde on DAHP synthase 
activity of 
E. coli
 WN1/pKM6.274. 
a 2-hydroxymuconic acid; 
b 2-hydroxmuconate 6-semialdehyde. 
Entry
[HMA] or 
[HMS]  (mM)
Sp. activity  
(U/mg)
1a032a10*443a10*334a10*275a10*156a157a1048b16%655.3.  Enzymatic tautomerization of 2-hydroxymuconic acid 
 After 48 h in glucose-limited fermentation conditions, 
E. coli 
WN1/pKM6.280 
accumulates a product mixture of 2-HMA, 
!
-keto acid 
23, and 
!
-keto acid 
24 (Figure 39) is 
observed.  When chemically synthesized 2-HMA was incubated in 100 mL minimal media at 37 
¡C, isomerization to the 
#
,'
-unsaturated 
!
-keto acid 
23 was observed by 
1H NMR, however the 
conjugated 
!
-keto acid 
24 was not observed.  This indicates that spontaneous tautomerization is 
selective toward the unconjugated 
!
-keto acid 
24 in aqueous solutions.  The appearance of 
predominance of the unconjugated 
!
-keto acid 
24 in fermentations of WN1/pKM6.280 must be 
due to biological activity. 
%   Figure 39.
  Structures of 2-hydroxymuconic acid (2-HMA) and 
!
-keto acid tautomers. 
 Enzymatic tautomerization of 2-HMA may be catalyzed by 4-oxalocrotonate tautomerase 
(4-OT).  This enzyme is involved in the lower branch of aromatic hydrocarbon degradation 
pathways in many organisms including 
Pseudomonas putida
, Bacillus subtilis
, and 
Heliobacter 
pylori
.  It catalyzes the 
(
-bond migration of unsaturated 
!
-keto acids via a dienol intermediate 
(Figure 40).
96-102  The isomerized 
!
-keto acids are then subjected to further enzymatic 
degradation, ultimately entering the citric acid cycle.  In 
E. coli
, 4-OT is encoded by 
pptA
, and 
may be at least partly responsible for the low titers observed in fermentations with WN1/
CO2HHO2CHO2-HMA
CO2HHO2CO23CO2HHO2CO24%66pKM6.280.  Homologous recombination was explored to remove the genomic 
pptA
 locus of 
WN1 in an effort to increase titers of 2-HMA. 
%   Figure 40.  
Action of 4-oxalocrotonate tautomerase. 
 A.
 Reactions catalyzed by 4-oxalocrotonate tautomerase on 2-
hydroxymuconic acid;  
B.  Catalytic mechanism of 4-oxalocrotonate tautomerase demonstrating conversion of 2-
hydroxymuconic acid to 
!
-keto acid tautomer 
24. 5.3.1.  Deletion of pptA locus 
 Deletion of the 
pptA 
locus from the genome of 
E. coli
 WN1 employed the phage 
"
 Red 
recombinase system.
103  The Red system utilizes three genes: 
%, &, and 
exo, whose products are 
called Gam, Bet, and Exo, respectively.  Gam inhibits the host RecBCD exonuclease V so that 
Bet and Exo can gain access to ends of linearized DNA and promote recombination.
103  The 
recombination procedure for deletion of 
pptA
 involved four steps (Figure 41).  A kanamycin 
resistance gene flanked by FLP recombinase recognition targets (FRT-Kan
R) was PCR amplified 
OOOOONHPro-1
HH2NArg-11
NH2OOOOONHPro-1
H2NArg-11
NH2HOOOOONHPro-1
H2NArg-11
NH2CO2HHO2CHOCO2HHO2COCO2HHO2COAB2-HMA
2324%67from plasmid pKD4 using primers extended with 
pptA-
homologous regions (H1, H2, Fig. 41).  
The PCR product was transformed into WN1 expressing the 
"
 Red recombinase genes from 
plasmid pKD46.  Growth at 30 ¡C promotes recombination of FRT-Kan
R into the 
pptA
 locus of 
WN1 to afford WN1
pptA::
KanR.    Elimination of the kanamycin resistance cassette employed plasmid pCP20 which shows 
thermal induction of FLP recombinase synthesis and carries ampicillin resistance.  After 
transformation into WN1
pptA::
KanR, ampicillin-resistant transformants were selected at 30 ¡C.  
These were colony-purified once nonselectively at 43 ¡C and tested for loss of all antibiotic 
resistance.  Loss of pCP20 and the FRT-flanked Kan
R occurred simultaneously.  Deletion of the 
pptA
 gene from WN1 afforded strain KM1.  Successful deletion was verified by PCR using the 
following primers: A) 5Õ-CTGGACGTGGTAATGAATGAAGACGACTGC, B) 5Õ-GTTGCATC 
ATTTAGGTGCTGGGACGTTTGT, C) 5Õ-GCCTGCTTGCCGAATATCATGGTGGAAAAT, 
and D) 5Õ-GCCATTTTCCACCATGATATTCGGCAAGCA.  Primers are represented in Figure 
42 as colored squares with the corresponding annealing locations on the genomic DNA of 
WN1pptA
::KanR, WN1, or KM1.  Primers C and D are specific to the kanamycin resistance gene 
and therefore do not lead to an amplification product using WN1 or KM1 genomic DNA as a 
template when added to the reaction mixture.  The absence of a PCR amplification product in the 
presence of primers C and D correlate to the absence of the kanamycin resistance gene.  In these 
reactions, a single PCR product was observed by agarose gel electrophoresis corresponding to a 
0.6 kb DNA fragment or 0.4 kb fragment, respectively.  When WN1
pptA
::KanR genomic DNA 
was used as a template, three amplification products were obtained which matched the expected 
%68sizes on the basis of the sequence of the kanamycin resistance gene and the genes flanking the 
pptA
 locus. 
%  Figure 41.
  Workflow for homologous recombination using FRT-flanked kanamycin resistance cassette and phage 
"
 Red recombinase system.
 H1H2FRT
FRT
Kanamycin resistance
Gene A
Gene BGene CH1H2FRT
FRT
Kanamycin resistance
Gene A
Gene CFRT
Gene A
Gene CABCD%69% Figure 42.
  Verification of 
pptA
 deletion via PCR.  Expected PCR product sizes using genomic DNA from 
WN1pptA
::KanR, WN1, or KM1 are described below each cartoon.  Sizes of DNA fragments observed using agarose 
gel electrophoresis are also presented. 
   5.3.2.  Fermentations of KM1/pKM6.280 
 After transformation, 
E. coli 
KM1/pKM6.280 was examined under glucose-rich and 
glucose-limited fed-batch fermentor conditions (Figures 43 and 44).  The results were 
encouraging as 2-HMA was the predominant product up to 30 h in glucose-rich fermentation and 
36 h in glucose-limited fermentation.  The formation of both 
!
-keto acid tautomers 
23 and 
24 indicates that 4-OT is not solely responsible for 2-HMA tautomerization in 
E. coli
 KM1.  
Nonetheless, significant improvement in titers of 2-HMA were achieved.  Peak titers in glucose-
rich fermentations increased from 30 mg/L to 2.1 g/L while glucose-limited fermentations 
increased from 60 mg/L to 1.8 g/L of 2-HMA. 
%70%                   % Figure 43.
  Biosynthesis of 2-hydroxymuconic acid by 
E. coli
 KM1/pKM6.280 under glucose-rich conditions.  
Legend:  2-HMA, black columns; tautomer 
23, grey columns; tautomer 
24, open columns; dry cell weight, closed 
circles. 
"                 % Figure 44.
  Biosynthesis of 2-hydroxymuconic acid by 
E. coli
 KM1/pKM6.280 under glucose-limited conditions.  
Legend:  2-HMA, black columns; tautomer 
23, grey columns; tautomer 
24, open columns; dry cell weight, closed 
circles. 
  Dry cell weight  (g/L)0102030402-HMA isomers  (g/L)01234Time  (h)12182430364248CO2HHO2CHO2-HMA
CO2HHO2CO23CO2HHO2CO24Dry cell weight  (g/L)0102030402-HMA isomers  (g/L)01234Time  (h)12182430364248CO2HHO2CHO2-HMA
CO2HHO2CO23CO2HHO2CO24%71 Deletion of 
pptA
 resulted in accumulation of 2.1 g/L 2-HMA after 30 h in glucose-rich 
fermentation, corresponding to a seventy-fold improvement over 
E. coli
 WN1.  After 48 h, the 
total amount of biosynthesized 2-HMA, tautomer 23, and tautomer 24 reached 5.1 g/L.  This 
corresponds to a two-fold increase over 
E. coli
 WN1.  The gradual decrease in titers of 2-HMA 
observed with concomitant increase in titers of both 
!
-keto acid isomers 
23 and 
24 suggests an 
additional mechanism of tautomerization besides 4-OT.  
E. coli
 possesses an additional 
isomerase which may also act upon 2-HMA, called 5-carboxymethyl-2-hydroxymuconate 
isomerase (5CHMI).  Kinetic parameters previously determined for 5CHMI revealed modest 
activity for the enzyme utilizing 2-HMA rather than its physiological substrate.
102  It is possible 
that the activity of 5CHMI in KM1/pKM6.280 contributes to the increase of 
!
-keto acid isomers 
of 2-HMA over the course of fermentations and the continued presence of 
!
-keto acid tautomer 
24, which was not observed during spontaneous tautomerization of 2-HMA.  
6.  Isolation of 2-hydroxymuconic acid from fermentation broth
  Cells were removed from fermentation broth via centrifugation at 8,000 rpm for 20 min.  
Protein was then removed using a tangential flow apparatus fitted with a 10 kD ultrafiltration 
cassette (Sartorius).  The cell-free/protein-free broth then concentrated to dryness under reduced 
pressure and redissolved in 1,4-dioxane in preparation for a Cu(OTf)
2-catalyzed cycloaddition 
with ethylene.  Only minor amounts of PTA were produced in a 0.5% yield from all three 
isomers of 2-HMA.  Though this result indicates that biobased 2-HMA can be used as a 
precursor to renewable PTA, the low yield indicates that certain components in the fermentation 
%72broth inhibit the synthesis of PTA.  Attempts to isolate 2-HMA from crude cell-free/protein-free 
fermentation broth were undertaken.   
 During the chemical synthesis of 2-HMA from 2-hydroxyhexa-2,4-dienedioate 
22, 2-HMA is isolated as a beige-colored solid upon acidification of the reaction mixture at 4 ¡C.  
Similar treatment of the 2-HMA-containing fermentation broth did not afford clean material.  
Organic extraction was explored as the principal strategy of recovering 2-HMA from 
fermentation broth. 
6.1.   Organic extraction of 2-hydroxymuconic acid isomers 
 Various organic solvents were examined for their efficacy in isolating clean 2-HMA.  A 
sample of fermentation broth containing 2 g/L 2-HMA isomers was acidified to pH 2 and 
concentrated to 1/3 of the original volume under reduced pressure.  Aliquots (0.5 mL) were 
combined with an equal volume of organic solvent and shaken overnight in a rotary incubator.  
Extraction efficiency was calculated as percent of 2-HMA isomers recovered from the 
fermentation broth (Table 15). 
%73Table 15
.  2-Hydroxymuconic acid extraction solvent screening.
 Acetone and the short-chain aliphatic alcohols did not result in distinct separation of aqueous and 
organic layers.  The highest extraction efficiencies were achieved with cyclohexanol, 2-butanol, 
and methyl ethyl ketone (MEK).  As MEK afforded the highest extraction efficiency, it was used 
in all subsequent extractions.  Interestingly, nearly all of 
!
-keto acid tautomer 
23 remained in the 
aqueous layer after treatment with organic solvents.  
 Pretreatment methods were explored in an attempt to increase recovery of all three 
isomers of 2-HMA.  Iterations of charcoal pretreatment, concentration under reduced pressure, 
and acidification (pH 2) were employed.  Regardless of other treatment methods, acidification 
proved to be essential for extraction into organic solvents.  At neutral pH, carboxylic acid 
moieties of 2-HMA are deprotonated and would likely form a salt with any available cation.  
Acidification, therefore, increases organic solubility.   
Solvent
Recovery (mol %)
methanol
no separation
ethanol
no separation
n-propanol
no separation
isopropanol
no separation
n-butanol
23%2-butanol
48%cyclohexanol
41%octanol
23%acetone
no separation
4-methyl-2-pentanone
0%methyl ethyl ketone
65%%74 Activated carbon (Sigma, Darco 20-20 mesh) was added to samples of fermentation broth 
(1 g per 100 mL) which were shaken at 250 rpm at rt for 1 h.  Charcoal treated samples did not 
afford significantly cleaner extracts but did affect the ratio of 2-HMA to 
!
-keto acid tautomers 
present in the organic extract. Without charcoal pretreatment, both 
!
-keto acids remained in the 
aqueous layer, whereas charcoal-treated samples afforded higher recoveries of all three isomers.  
Concentration of the fermentation broth to 1/4 original volume did not have a significant effect 
on extraction efficiency.  
 The primary challenge associated with isolation of 2-HMA from fermentation broth is 
achieving high recovery of all three 2-HMA tautomers.  Charcoal pretreatment prior to organic 
extraction influences the ratio of recovered isomers.  Without pretreatment, 2-HMA is recovered 
in greatest extent with >90% recovery into the organic layer.  With charcoal pretreatment, the 
!
-keto acids are predominant in the organic extract.  In all cases, over 50% of 
!
-keto acid 
23 remains in the aqueous layer regardless of pretreatment methods.  
6.2.  Conversion to 6-carboxy-2-pyrone and subsequent isolation  
 Efforts to extract all three tautomers of microbe-synthesized 2-HMA were largely 
unsuccessful due to the propensity of 
!
-keto acid 
23 to remain in the aqueous layer.  Recovery of 
2-HMA would be easier if the three isomers could be converted to a single structure.  No form of 
chemical treatment was found that could enable conversion of a mixture of 2-HMA tautomers to 
a single isomer.  Attempts to convert 2-HMA to 6-carboxy-2-pyrone 
19 in fermentation broth 
were therefore undertaken with conditions used for conversion of chemically synthesized 2-
HMA (1:4 H
2SO4:AcOH in water, 100 ¡C).  After 3 h, all three isomers of 2-HMA had been fully 
%75consumed however 6-carboxy-2-pyrone 
19 was afforded in only 30% yield.  The reaction crude 
had become very dark and slightly more viscous.  Evidently, the acidic reaction medium coupled 
with elevated temperatures resulted in uncontrolled side reactions, consuming much of the initial 
2-HMA.  Despite the poor yield of 6-carboxy-2-pyrone, higher extraction efficiencies were 
achieved than extractions of 2-HMA isomers (Table 16). 
Table 16.
  Extraction solvent screening for 6-carboxy-2-pyrone 
19 from fermentation broth. 
The primary challenge associated with organic extraction is cleanly isolating the desired product.  
Solvents with high extraction efficiencies tended to solubilize most of the components within 
crude fermentation broth.  Solvents with lower efficiencies resulted in less contamination but did 
not effectively recover useful amounts of 2-HMA or 6-carboxy-2-pyrone 
19.  Introduction of 
ion-exchange chromatography into the isolation method either prior to, or following organic 
extraction may help to provide cleaner extracts.  Cleaner material should afford higher yields of 
PTA with microbe-synthesized 2-HMA. 
Solvent
Recovery (mol %)
n-butanol
87%2-butanol
90%isobutanol
50%cyclohexanol
67%octanol
33%4-methyl-2-pentanone
43%methyl ethyl ketone
n/a
%767.  Discussion
  This work lays the foundation for the synthesis of renewable aromatic diacids from 2-
HMA.  Expression of protocatechuate (PCA) 2,3-cleavage enzymes from 
Paenibacillus 
sp. JJ-1b 
in 
E. coli
 WN1 afforded 2.3 g/L of 2-HMA isomers in glucose-rich fed-batch fermentation.  
However, less than 1% was produced in the conjugated enol form.  This is the only isomer with 
the proper geometry to react with ethylene to afford PTA.  The predominance of 
!
-keto acid 
isomers must therefore be reduced before biobased 2-HMA can be utilized in a cycloaddition.   
 Deletion of native 4-oxalocrotonase tautomerase (4-OT) activity resulted in a two-fold 
increase in titer of 2-HMA isomers with a significant prevalence of the enol structure in 
fermentations of KM1/pKM6.280.  The reaction catalyzed by 4-OT results in 
(
-bond migration 
of an unsaturated 
!
-keto acid via a conjugated enol intermediate prior to degradation through the 
citric acid cycle.  Though titers of 2-HMA improved, product distribution skewed toward 
!
-keto 
acid tautomers by the end of the fermentation with the continued formation of 
!
-keto acid 
23, indicating that 4-OT activity is not solely responsible for 2-HMA tautomerization.  Incubation of 
chemically synthesized 2-HMA in fermentation medium at 37 ¡C revealed uncatalyzed 
tautomerization of 2-HMA to the conjugated 
!
-keto acid 
23 to be negligible.  This suggests the 
rate of tautomerization seen in fermentations is likely due to additional biological activity.  
E. 
coli
 may possess another isomerase that can utilize 2-HMA as a non-native substrate.  5-
Carboxymethyl-2-hydroxymuconate isomerase has been previously shown to catalyze 2-HMA 
tautomerization in 
E. coli
 albeit with modest kinetics.  Perhaps this, in conjunction with other 
promiscuous enzymes, contributes to the ultimate predominance of the 
!
-keto acid isomers.  The 
tautomerase superfamily of enzymes is characterized by a catalytic N-terminal proline residue 
%77and a conserved 
#
-!
-#
 motif.  Homology searches could help identify additional tautomerases 
present in the 
E. coli
 genome.  Deletion of genes encoding additional tautomerases via 
homologous recombination, as was used for deletion of 
pptA
, should prevent enzymatic 
isomerization of 2-HMA.   
 If genomic modification to remove enzyme-catalyzed tautomerization does not 
completely prevent the formation of 
!
-keto acid isomers in fermentations, selective adsorption 
might be employed to remove 2-HMA from fermentation broth as it is produced.  This technique 
has been used with great efficiency in catechol biosynthesis from 
D-glucose.  Selective 
adsorption may also help alleviate any inhibitory effects from accumulation of 2-HMA or 
biosynthetic precursors.  Inhibition of DAHP synthase was not observed however accumulation 
of 100-500 mg/L DHS in fermentation broth may indicate a bottleneck due to insufficient 
aroZ
 activity.  Enzyme assays of 6xHis-tagged 
aroZ
 in the presence of 2-HMA or 2-HMS could 
indicate whether this repression occurs.  There is also a possibility that an 
!
-keto acid isomer is 
inhibitory to one or more enzymes in the biosynthesis pathway.  Selective adsorption may be a 
way to keep the levels of 2-HMA-associated metabolites low enough to prevent any inhibition. 
 The primary challenge to be addressed is the isolation of 2-HMA from fermentation broth 
and subsequent conversion into appreciable quantities of PTA.  Attempts to cleanly isolate 2-
HMA as a single isomer or via cyclization to 6-carboxy-2-pyrone 
19 have been ineffective.  The 
minor amount of PTA produced during Cu(OTf)
2-catalyzed cycloaddition of biobased 2-HMA 
with ethylene 
2 indicates pure 2-HMA is necessary.  Derivatization to 6-carboxy-2-pyrone 
19 is 
also precluded by impure starting material.  Organic extraction affords 2-HMA as a mixture of 
keto and enol tautomers along with extensive contaminants from fermentation broth.  Only the 
%78conjugated enol possesses the appropriate geometry for Diels-Alder cycloaddition with ethylene.  
Attempts to drive the 
!
-keto acids toward the enol have been unsuccessful.  Without isolation of 
clean 2-HMA from fermentation broth and subsequent conversion to PTA, this route will never 
be commercially viable. 
  %79PART III 
CONCLUSIONS 
 The results described herein demonstrate a proof-of-concept semisynthetic strategy 
toward renewable terephthalic, isophthalic, and phthalic acids from biobased feedstocks.  Current 
manufacture of aromatic diacids relies exclusively upon subjecting petroleum-derived xylenes to 
costly and wasteful aryl methyl group oxidation.  Materials made from PTA, IPA, and PA are 
valuable and widely used.  It is highly desirable to reduce costs associated with their manufacture 
by avoiding aryl methyl group oxidation.  Biobased compounds are attractive candidates to 
address this need. 
 Renewable syntheses of PTA have received the most attention as it is the most highly 
valued of the aromatic diacids.  Many proposed synthetic pathways produce 
p-xylene 
1, thus 
enabling simple integration into current PTA manufacture architecture.  The principal drawback 
of these routes is their inevitable reliance on the wasteful Amoco Mid-Century oxidation process.  
Other routes to biobased PTA circumvent the Amoco Mid-Century oxidation by taking advantage 
of diacid substrates such as malic acid, succinic acid, and muconic acid.  All of these syntheses 
require esterification of the diacid preceding cycloaddition to form the six-membered ring of 
PTA.  Aromatization of the cyclic intermediate affords dimethyl terephthalate (DMT) which 
must be hydrolyzed to produce PTA.  These steps are commercially unattractive as they not only 
lengthen a synthetic route but involve copious amounts of methanol which is toxic and highly 
flammable.  Industrial facilities which use methanol require fitting of specialty wiring and other 
precautions to decrease the risk of a fire caused by an inadvertent electrical spark.  These 
%80facilities are, of course, highly expensive to build precluding the use of methanol in commodity 
chemical manufacture. 
 Only one of the eleven proposed routes to biobased PTA provides access to IPA as well.  
Poly(ethylene terephthalate) (PET) requires 2-5 wt% IPA to inhibit crystallization and provide a 
transparent, flexible polymer.  Despite stabilizing PTA manufacture costs by diversifying 
available feedstocks, PET manufacture would still be reliant on petroleum-derived IPA.  Thus, a 
route which allows access to both aromatic components of PET would be highly desirable.  
Microbe-synthesized 2-hydroxymuconic acid (2-HMA) provides a means to both PTA and IPA 
along with PA via conversion to 6-carboxy-2-pyrone 
19.  Several plasticizers such as diisononyl 
cyclohexane-1,2-dicarboxylate (DINCH) and di-(2-ethylhexyl)phthalate (DEHP) (Figure 1, Part 
I), are derived from PA.  Access to all regioisomers of aromatic diacids is thereby possible 
through 2-HMA. 
 As it stands, use of microbe-synthesized 2-HMA is far from commercial application.  
Biocatalytic synthesis has been established in 
E. coli
 and affords modest amounts of 2-HMA.  
However, not all of the 2-HMA produced is suitable for a cycloaddition with ethylene
 2, and an 
inability to cleanly isolate any such material precludes its use as a substrate for PTA 
manufacture.  Catalysis of the cycloaddition of 2-HMA with ethylene 
2 and the reaction of 6-
carboxy-2-pyrone 
19 with propiolic acid 
18 also needs improvement.  Though the latter reaction 
achieves high yields with BOB(OAc)
4, low catalytic selectivity produces IPA and PA in a 1:1 
ratio.  Desirable selectivities favor IPA, which is more useful in chemicals manufacture.  Finally, 
charring of the reaction mixture during the cycloaddition of 2-HMA with ethylene 
2 precludes 
recovery of unreacted substrate.  This challenge, when coupled to a 26% yield per pass, 
%81considerably reduces process efficiency.  Optimizing reaction conditions and continued catalyst 
screening may improve selectivities.  Despite lingering challenges, the work presented herein 
demonstrates that synthesis of PTA, IPA, and PA from a common, biobased starting material is 
possible.    
%82REFERENCES 
%83REFERENCES 
1. Cutler, J. Presented at the West Virginia Marcellus to Manufacturing Conference, Charleston, 
WV, March 2013. 
2. Maps: Exploration, Resources, Reserves, and Production. 
https://www.eia.gov/maps/maps.htm
 (accessed March 14, 2018). 
3.  National Energy Technology Laboratory. Modern Shale Gas Development in the United 
States: An Update. US Department of Energy, September 2013. 
4. Berkowitz, N. 
Fossil hydrocarbons: chemistry and technology
; Academic: San Diego, 1997; 
pp 88-91. 5.  King, H. Hydraulic Fracturing of Oil & Gas Wells Drills in Shale. 
https://geology.com/
articles/hydraulic-fracturing/
 (Accessed March 11, 2018). 
6.  Berkowitz, N. 
Fossil hydrocarbons: chemistry and technology
; Academic: San Diego, 1997; 
pp 91-94. 7.  Elvers, B. Handbook of Fuels: Energy Sources for Transportation; Wiley-VCH: Weinheim, 
2008; pp 29-34. 
8.  Franck, H.-G.; Stadelhofer, J. W. 
Industrial Aromatic Chemistry
; Springer: Berlin, 1988; pp 
60-62. 9. Cailliot, A. Etudes sur LÕEssence de Terebenthine. Ann. Chim. Phys., 1847, 3, 27-40. 
10.  Collias, D. I.; Harris, A. M.; Nagpal, V.; Cottrell, I. W.; Schultheis, M. W.; Biobased 
Terephthalic Acid Technologies: A Literature Review. 
Ind. Biotechnol.
 2014, 10, 91-105. 11.  Saffer, A.; Barker, R. S. Preparation of Aromatic Polycarboxylic Acids. US 2,833,816, 1958. 
12.  Tomas, R. A. F.; Bordado, J, C. M.; Gomes, J. F. P. 
p-Xylene Oxidation to Terephthalic Acid: 
A Literature Review Oriented Toward Process Optimization and Development. 
Chem Rev. 
2013, 113
, 7421-7469. 13.  Partenheimer, W. Methodology and Scope of Metal/Bromide Autoxidation of Hydrocarbons. 
Catal. Today
, 1995, 23, 69-158 14. Tibbitt, J. Personal communication. 
%8415. Roman-Leshkov, Y.; Barrett, C. J.; Liu, Z.Y.; Dumesic J.A. Production of Dimethylfuran for 
Liquid Fuels from Biomass Derived Carbohydrates 
Nature
, 2007, 447, 982-985. 16.  Brookhart, M.; Findlater, M.; Guironnet, D.; Lyons T.W. Synthesis of 
para
-Xylene and 
Toluene US 2013/0237732 A1, 2013 
17.  Lyons, T.W.; Guironnet, D.; Findlater, M.; Brookhart, M. Synthesis of 
para
-Xylene from 
Ethylene. 
J. Am. Chem. Soc., 
2012, 134 15708-15711 
18.  Peters, M. W.; Taylor, J. D.; Jenni, M. M.; Manzer, L. E.; Henton, D. E. Integrated Process to 
Convert Renewable Isobutanol to 
p-Xylene. US 2011/044243 A1, 2011. 
19. Taylor, J. D.; Jenni, M. M.; Peters, M. W. Dehydration of Fermented Isobutanol for the 
Production of Renewable Chemicals and Fuels. 
Top. Catal.
 2010, 53 (15), 1224Ð1230. 20. Taylor, T. J.; Taylor, J. D.; Peters, M. W.; Henton, D. E. Variations on Prins-like Chemistry to 
Produce 2,5-Dimethylhexadiene from Isobutanol. US 8,742,187 B2, 2014. 
21. Manzer, L. E.; Kourtakis, K.; Herron, N.; McCarron, E. M.; VerNooy, P. D. Process for the 
Preparation of 
p-Xylene. US 7,067,708 B2, 2006. 
22. Kunkes, E. L.; Simonetti, D. A.; West, R. M.; Serrano-Ruiz, J. C.; G−rtner, C. A.; Dumesic, J. 
A. Catalytic Conversion of Biomass to Monofunctional Hydrocarbons and Targeted Liquid-Fuel 
Classes. 
Science
 2008, 322 (5900), 417Ð421. 23. Chang, C.-C.; Green, S. K.; Luke Williams, C.; Dauenhauer, P. J.; Fan, W. Ultra-Selective 
Cycloaddition of Dimethylfuran for Renewable 
p-Xylene with H-BEA. 
Green Chem.
 2014, 16 (2), 585Ð588. 24. Campbell, C. D.; Cole, D. T.; Harold Raymond Taylor, I. Process for the Preparation of 
Dialkyl Succinylsuccinates. US 5,783,723, 1998. 
25. Kruper, W., JR.; Rand, C. L.; Molzahn, D. C. Processes for Producing Terephthalic Acid and 
Terephthalic Esters. US 2013/0345467A1, 2013. 
26. Ashworth, I. W.; Bowden, M. C.; Dembofsky, B.; Levin, D.; Moss, W.; Robinson, E.; Szczur, 
N.; Virica, J. A New Route for Manufacture of 3-Cyano-1-Naphthalenecarboxylic Acid. 
Org. 
Process Res. Dev.
 2003, 7 (1), 74Ð81. 27. Lee, J. J.; Kraus, G. A. One-Pot Formal Synthesis of Biorenewable Terephthalic Acid from 
Methyl Coumalate and Methyl Pyruvate. 
Green Chem.
 2014, 16 (4), 2111Ð2116. 
28. Lee, J. J.; Kraus, G. A. Divergent DielsÐAlder Methodology from Methyl Coumalate toward 
Functionalized Aromatics. 
Tetrahedron Lett.
 2013, 54 (19), 2366Ð2368. %8529. Frost, J. W.; Miermont, A.; Schweitzer, D.; Bui, R.; Wicks, D. A. Terephthalic and Trimellitic 
Based Acids and Carboxylate Derivatives Thereof. US 8,367,858 B2, 2013. 
30. Miura, T.; Kakinuma, H.; Kawano, T.; Matsuhisa, H. Method for Producing Furan-2,5-
Dicarboxylic Acid. US 7,411,078 B2, 2008. 
31. Gong, W. H. Terephthalic Acid Composition and Process for the Production Thereof. US 
7,385,081 B1. 32 Gong, W. H. Terephthalic Acid Composition and Process for the Production Thereof. US 
9,299,278 B2, 2012 33. Hanke, P. D. Enzymatic Oxidation of Hydroxymethylfurfural. US 8,183,020 B2, 2012. 
34. Berti, C.; Binassi, E.; Colonna, M.; Fiorini, M.; Kannan, G.; Karanam, S.; Mazzacurati, M.; 
Odeh, I. Bio-Based Terephthalate Polyesters. 8946472B2. 
35. Colonna, M.; Berti, C.; Fiorini, M.; Binassi, E.; Mazzacurati, M.; Vannini, M.; Karanam, S. 
Synthesis and Radiocarbon Evidence of Terephthalate Polyesters Completely Prepared from 
Renewable Resources. 
Green Chem.
 2011
, 13 (9), 2543Ð2548. 36. Kamitsou, M.; Panagiotou, G. D.; Triantafyllidis, K. S.; Bourikas, K.; Lycourghiotis, A.; 
Kordulis, C. Transformation of 
!
-Limonene into 
p-Cymene over Oxide Catalysts: A Green 
Chemistry Approach. 
Appl. Catal. A
 2014, 474, 224Ð229. 37. Miller, K. K.; Zhang, P.; Nishizawa-Brennen, Y.; Frost, J. W. Synthesis of Biobased 
Terephthalic Acid from Cycloaddition of Isoprene with Acrylic Acid. 
ACS Sustainable Chem. 
Eng.
 2014, 2 (8), 2053Ð2056. 38. Li, W.; Xie, D.; Frost, J. W. Benzene-Free Synthesis of Catechol: Interfacing Microbial and 
Chemical Catalysis. 
J. Am. Chem. Soc.
 2005, 127 (9), 2874Ð2882. 39. Wang, J.; Shen, X.; Yuan, Q.; Yan, Y. Microbial Synthesis of Pyrogallol Using Genetically 
Engineered 
Escherichia coli
. Metab. Eng.
 2018, 45, 134Ð141. 40. Kasai, D.; Fujinami, T.; Abe, T.; Mase, K.; Katayama, Y.; Fukuda, M.; Masai, E. Uncovering 
the Protocatechuate 2,3-Cleavage Pathway Genes. 
J. Bacteriol.
 2009, 191 (21), 6758Ð6768. 41. MetaCyc
 Pathway: protocatechuate degradation III (
para
-cleavage pathway). 
https://
metacyc.org/META/NEW-IMAGE?type=NIL&object=PWY-6336&redirect=T
 (Accessed March 
14, 2018). %8642. Metanis, N.; Keinan, E.; Dawson, P. E. A Designed Synthetic Analogue of 4-OT Is Specific 
for a Non-Natural Substrate. 
J. Am. Chem. Soc.
 2005, 127 (16), 5862Ð5868. 43. Boger, D. L.; Patel, M. Chapter 2 - Recent Applications of the Inverse Electron Demand 
Diels-Alder Reaction. In 
Progress in Heterocyclic Chemistry
; Suschitzky, H., Scriven, E. F. V., 
Eds.; Elsevier, 1989; Vol. 1, pp 30Ð64. 
44. Yamamoto, Y. From Sigma- to Pi-Electrophilic Lewis Acids. Application to Selective 
Organic Transformations. 
J. Org. Chem.
 2007, 72 (21), 7817Ð7831. 45. Suzuki, T.; Noble, R. D.; Koval, C. A. Electrochemistry, Stability, and Alkene Complexation 
Chemistry of Copper(I) Triflate in Aqueous Solution. Potential for Use in Electrochemically 
Modulated Complexation-Based Separation Processes. 
Inorg. Chem.
 1997, 36 (2), 136Ð140. 46. Ridlen, S. G.; Wu, J.; Kulkarni, N. V.; Dias, H. V. R. Isolable Ethylene Complexes of 
Copper(I), Silver(I), and Gold(I) Supported by Fluorinated Scorpionates [HB{3-(CF
3),5-(CH3)Pz}
3]Ð and [HB{3-(CF
3),5-(Ph)Pz}
3]Ð. Eur. J. Inorg. Chem.
 2016, 2016 (15-16), 2573Ð2580. 47.  Frost, J. W.; Miermont, A.; Schweitzer, D.; Bui, V. Preparation of 
trans, trans
-Muconic Acid 
and 
trans, trans -
Muconates. US 8,426,639 B2, 2013. 
48. Miles, W. H.; Cohen, E. M.; Naimoli, B. J. DielsÐAlder Reactions of 
#
-Acylacrylic Acids. 
Synth. Commun.
 2013, 43 (14), 1980Ð1991. 49. Robinson, R.; Fray, G. I. Improved Process for Diels-Alder Additions. GB 835,840, 1960.  
50. Robinson, R.; Fray, G. I. Process for the Production of Diels-Alder Adducts. U.S. 3,067,244, 
1962.  51  Takenaka, S.; Yamamoto, K.; Nishida, T. Preparation of Cyclohexenecarboxylic Acid. JP 
S54157546 A, 1979. 
52. Kobayashi, S. Lanthanide Trifluoromethanesulfonates as Stable Lewis Acids in Aqueous 
Media. Yb(OTf)
3 Catalyzed Hydroxymethylation Reaction of Silyl Enol Ethers with Commercial 
Formaldehyde Solution. 
Chem. Lett.
 1991, 20 (12), 2187Ð2190. 53. Kobayashi, S.; Hachiya, I.; Takahori, T.; Araki, M.; Ishitani, H. Lanthanide 
Trifluoromethanesulfonates as Reusable Catalysts. Michael and Diels-Alder Reactions. 
Tetrahedron Lett.
 1992, 33 (45), 6815Ð6818. 54. Pindur, U.; Lutz, G.; Otto, C. Acceleration and Selectivity Enhancement of Diels-Alder 
Reactions by Special and Catalytic Methods. 
Chem. Rev.
 1993, 93 (2), 741Ð761. %8755. Corey, E. J.; Shibata, T.; Lee, T. W. Asymmetric Diels-Alder Reactions Catalyzed by a Triflic 
Acid Activated Chiral Oxazaborolidine. 
J. Am. Chem. Soc.
 2002, 124 (15), 3808Ð3809. 56. Li, P.; Yamamoto, H. Bifunctional Acid Catalysts for Organic Synthesis. In 
Bifunctional 
Molecular Catalysis
; Ikariya, T., Shibasaki, M., Eds.; Springer Berlin Heidelberg: Berlin, 
Heidelberg, 2011; pp 161Ð183. 
57. Yamamoto, H.; Futatsugi, K. ÒDesigner AcidsÓ: Combined Acid Catalysis for Asymmetric 
Synthesis. 
Angew. Chem. Int. Ed.
 2005, 44 (13), 1924Ð1942. 58. Cairncross, J. R.; Roland, R. M.; Henderson, W. A.; Sheppard, A. F. J. Am. Chem. Soc. 1970, 
92, 3187.  59 a) Shepard, A. F.; Winslow, N. R.; Johnson, J. R. The Simple Halogen Derivatives of Furan. 
J. 
Am. Chem. Soc.
 1930, 52 (5), 2083Ð2090. b) Shang, R.; Liu, L. Transition Metal-Catalyzed 
Decarboxylative Cross-Coupling Reactions. 
Sci. China Chem.
 2011
, 54 (11), 1670Ð1687. c) 
Goo§en, L. J.; Rodr™guez, N.; Goo§en, K. Carboxylic Acids as Substrates in Homogeneous 
Catalysis. 
Angew. Chem. Int. Ed Engl.
 2008, 47 (17), 3100Ð3120. d) Goo§en, L. J.; Goo§en, K.; 
Rodr™guez, N.; Blanchot, M.; Linder, C.; Zimmermann, B. New Catalytic Transformations of 
Carboxylic Acids. 
J. Macromol. Sci. Part A Pure Appl. Chem.
 2008, 80 (8), 2084. e) L. J. 
Goo§en, K. Goo§en, N. Rodriguez, M. Blanchot, C. Linder, B. Zimmermann, 
Pure Appl. Chem
. 2008, 80, 1725. f) Rodr™guez, N.; Goo§en, L. J. Decarboxylative Coupling Reactions: A Modern 
Strategy for CÐC-Bond Formation. 
Chem. Soc. Rev
 2011
, 40, 5030Ð5048. 60. Cahiez, G.; Moyeux, A.; Gager, O.; Poizat, M. Copper-Catalyzed Decarboxylation of 
Aromatic Carboxylic Acids: En Route to Milder Reaction Conditions. 
Adv. Synth. Catal.
 2013, 355 (4), 790Ð796. 61. Cohen, T.; Schambach, R. A. Copper-Quinoline Decarboxylation. 
J. Am. Chem. Soc.
 1970, 92 (10), 3189Ð3190. 62. Whitesides, G. M.; Casey, C. P. The Stereochemistry of the Thermal Decomposition of Viylic 
Copper(I) and Silver(I) Organometallic Compounds1. 
J. Am. Chem. Soc.
 1966, 88 (19), 4541Ð4543. 63. Burdett, J. K.; Sevov, S. Stability of the Oxidation States of Copper. 
J. Am. Chem. Soc.
 1995, 117
 (51), 12788Ð12792. 64. Biogas Potential in the United States; NREL/FS-6A20-60178, National Renewable Energy 
Laboratory, Energy Analysis: Golden, CO, 2013 
65. Bricker, J. C.; Chen, J. Q.; Coughlin, P. K. Production of Oxygenates from a Methane 
Conversion Process, US 8,933,275, 2015. 
%8866. Bricker, J. C.; Chen, J. Q.; Coughlin, P. K.; Majumder, D. Production of Vinyl Chloride from 
a Methane Conversion Process, US 9,308,513, 2015. 
67. Goo§en, L. J.; Garrido, N. R.; Costa, F. M.; Lange, P. P. Process for Preparing a Propiolic 
Acid or Derivative Thereof, US 9,073,844, 2015. 
68. Effenberger, F.; Ziegler, T. Diels-Alder-Reaktionen Mit 2H-Pyran-2-Onen: Reaktivit−t Und 
Selektivit−t. 
Chem. Ber.
 1987, 120 (8), 1339Ð1346. 69. Zhang, P.; Kriegel, R. M.; Frost, J. W. BÐOÐB Catalyzed Cycloadditions of Acrylic Acids. 
ACS Sustainable Chem. Eng.
 2016, 4 (12), 6991Ð6995.  70. Harwood, C. S.; Parales, R. E. The Beta-Ketoadipate Pathway and the Biology of Self-
Identity. 
Annu. Rev. Microbiol.
 1996, 50, 553Ð590. 71. a) Masai, E.; Katayama, Y.; Fukuda, M. Genetic and Biochemical Investigations on Bacterial 
Catabolic Pathways for Lignin-Derived Aromatic Compounds. 
Biosci. Biotechnol. Biochem.
 2007, 71 (1), 1Ð15. b) Noda, Y.; Nishikawa, S.; Shiozuka, K.; Kadokura, H.; Nakajima, H.; Yoda, 
K.; Katayama, Y.; Morohoshi, N.; Haraguchi, T.; Yamasaki, M. Molecular Cloning of the 
Protocatechuate 4,5-Dioxygenase Genes of Pseudomonas Paucimobilis. 
J. Bacteriol.
 1990, 172 (5), 2704Ð2709. 72. a) Crawford, R. L.; Bromley, J. W.; Perkins-Olson, P. E. Catabolism of Protocatechuate by 
Bacillus Macerans. 
Appl. Environ. Microbiol.
 1979, 37 (3), 614Ð618. b) Wolgel, S. A.; 
Lipscomb, J. D. Protocatechuate 2,3-Dioxygenase from Bacillus Macerans. 
Methods Enzymol.
 1990, 188, 95Ð101. 73.Harayama, S.; Rekik, M.; Ngai, K. L.; Ornston, L. N. Physically Associated Enzymes 
Produce and Metabolize 2-Hydroxy-2,4-Dienoate, a Chemically Unstable Intermediate Formed 
in Catechol Metabolism via 
meta-Cleavage in 
Pseudomonas putida
. J. Bacteriol.
 1989, 171 (11), 
6251Ð6258. 74. Providenti, M. A.; Mampel, J.; MacSween, S.; Cook, A. M.; Wyndham, R. C. Comamonas 
Testosteroni BR6020 Possesses a Single Genetic Locus for Extradiol Cleavage of 
Protocatechuate. 
Microbiology
 2001, 147, 2157Ð2167. 75. Maruyama, K.; Shibayama, T.; Ichikawa, A.; Sakou, Y.; Yamada, S.; Sugisaki, H. Cloning 
and Characterization of the Genes Encoding Enzymes for the Protocatechuate Meta-Degradation 
Pathway of Pseudomonas Ochraceae NGJ1. 
Biosci. Biotechnol. Biochem.
 2004, 68 (7), 1434Ð1441. 76. Eaton, R. W. Plasmid-Encoded Phthalate Catabolic Pathway in 
Arthrobacter keyseri
 12B. J. 
Bacteriol.
 2001, 183 (12), 3689Ð3703. %8977. Wolgel, S. A.; Dege, J. E.; Perkins-Olson, P. E.; Jaurez-Garcia, C. H.; Crawford, R. L.; 
Mk, E.; Lipscomb, J. D. Purification and Characterization of Protocatechuate 2,3-
Dioxygenase from Bacillus Macerans: A New Extradiol Catecholic Dioxygenase. 
J. Bacteriol.
 1993, 175 (14), 4414Ð4426. 78. Pittard, J.; Wallace, B. J. Distribution and Function of Genes Concerned with Aromatic 
Biosynthesis in 
Escherichia coli
. J. Bacteriol.
 1966, 91 (4), 1494Ð1508. 79. Li, K.; Frost, J. W. Synthesis of Vanillin from Glucose. 
J. Am. Chem. Soc.
 1998, 120 (40), 10545Ð10546. 80. Niu, W.; Draths, K. M.; Frost, J. W. Benzene-Free Synthesis of Adipic Acid. 
Biotechnol. 
Prog.
 2002, 18 (2), 201Ð211. 
81. Draths, K. M.; Knop, D. R.; Frost, J. W. Shikimic acid and quinic acid: replacing isolation 
from plant sources with recombinant microbial biocatalysis. 
J. Am. Chem. Soc.
 1999, 121, 1603-1604.  82. Knop, D. R.; Draths, K. M.; Chandran, S. S.; Barker, J. L.; von Daeniken, R.; Weber, W.; 
Frost, J. W. Hydroaromatic equilibration during biosynthesis of shikimic acid. J
. Am. Chem. Soc
. 2001, 123, 10173-10182.  843. Chandran, S. S.; Yi, J.; Draths, K. M.; von Daeniken, R.; Weber, W.; Frost, J. W. 
Phosphoenolpyruvate availability and the biosynthesis of shikimic acid. 
Biotechnol. Prog
. 2003, 19, 808-814. 84. Dell, K. A.; Frost, J. W. Identification and removal of impediments to biocatalytic synthesis 
of aromatics from 
D-glucose: Rate-limiting enzymes in the common pathway of aromatic amino 
acid biosynthesis. 
J. Am. Chem. Soc
. 1993, 115
, 11581-11589.  
85. Li, K.; Mikola, M. R.; Draths, K. M.; Worden, R. M.; Frost, J. W. Fed-batch fermentor 
synthesis of 3-dehydroshikimic acid using recombinant 
Escherichia coli
. Biotechnol. Bioeng.
 1999, 64, 61-73.  86. Draths, K. M.; Pompliano, D. L. Conley, D. L.; Frost, J. W.; Berry, A.; Disbrow, G. L.; 
Staversky, R. J.; Lievense, J. C. Biocatalytic synthesis of aromatics from 
D-glucose: The role of 
transketolase. 
J. Am. Chem. Soc
. 1992, 114
, 3956-3962.  87. Kambourakis, S.; Draths, K. M.; Frost, J. W. Synthesis of Gallic Acid and Pyrogallol from 
Glucose: Replacing Natural Product Isolation with Microbial Catalysis. 
J. Am. Chem. Soc.
 2000, 122 (37), 9042Ð9043. 88.  Farabaugh, M. A. Biocatalytic Production of Aromatics from 
D-glucose.  Master of Science, 
Michigan State University, 1996. 
%9089. a) Garner, C. C.; Herrmann, K. M. Operator Mutations of the 
Escherichia coli aroF
 Gene. 
J. 
Biol. Chem.
 1985, 260 (6), 3820Ð3825. b) Schoner, R.; Herrmann, K. M. 3-Deoxy-
D-arabino
-Heptulosonate 7-Phosphate Synthase. Purification, Properties, and Kinetics of the Tyrosine-
Sensitive Isoenzyme from 
Escherichia coli
. J. Biol. Chem.
 1976, 251 (18), 5440Ð5447. 90. Cobbett, C. S.; Delbridge, M. L. Regulatory Mutants of the 
aroF-tyrA
 Operon of 
Escherichia 
coli
 K-12. 
J. Bacteriol.
 1987, 169 (6), 2500Ð2506.  91. Li, K.; Mikola, M. R.; Draths, K. M.; Worden, R. M.; Frost, J. W. Fed-Batch Fermentor 
Synthesis of 3-Dehydroshikimic Acid Using Recombinant 
Escherichia coli
. Biotechnol. Bioeng.
 1999, 64 (1), 61Ð73. 92. Anderson, J.; Dueber, J. E.; Leguia, M.; Wu, G. C.; Goler, J. A.; Arkin, A. P.; Keasling, J. D. 
BglBricks: A Flexible Standard for Biological Part Assembly. 
J. Biol. Eng.
 2010, 4 (1), 1. 93. Lee, T. S.; Krupa, R. A.; Zhang, F.; Hajimorad, M.; Holtz, W. J.; Prasad, N.; Lee, S. K.; 
Keasling, J. D. BglBrick Vectors and Datasheets: A Synthetic Biology Platform for Gene 
Expression. 
J. Biol. Eng.
 2011
, 5, 12. 94. Mischo, H. E.; Proudfoot, N. J. Disengaging Polymerase: Terminating RNA Polymerase II 
Transcription in Budding Yeast. 
Biochim. Biophys. Acta
 2013, 1829 (1), 174Ð185. 95. Schoner, R.; Herrmann, K. M. 3-Deoxy-
D-arabino
-Heptulosonate 7-Phosphate Synthase. 
Purification, Properties, and Kinetics of the Tyrosine-Sensitive Isoenzyme from 
Escherichia coli.
 J. Biol. Chem.
 1976, 251 (18), 5440Ð5447. 96. Lian, H.; Czerwinski, R. M.; Stanley, T. M.; Johnson, W. H.; Watson, R. J.; Whitman, C. P. 
The Contribution of the SubstrateÕs Carboxylate Group to the Mechanism of 4-Oxalocrotonate 
Tautomerase. 
Bioorg. Chem.
 1998, 26 (3), 141Ð156. 97. Whitman, C. P. The 4-Oxalocrotonate Tautomerase Family of Enzymes: How Nature Makes 
New Enzymes Using a Beta-Alpha-Beta Structural Motif. 
Arch. Biochem. Biophys.
 2002, 402 (1), 1Ð13. 98. Almrud, J. J.; Kern, A. D.; Wang, S. C.; Czerwinski, R. M.; Johnson, W. H.; Murzin, A. G.; 
Hackert, M. L.; Whitman, C. P. The Crystal Structure of YdcE, a 4-Oxalocrotonate Tautomerase 
Homologue from 
Escherichia coli
, Confirms the Structural Basis for Oligomer Diversity,. 
Biochemistry
 2002, 41 (40), 12010Ð12024. %9199. Wang, S. C.; Johnson, W. H.; Czerwinski, R. M.; Whitman, C. P. Reactions of 4-
Oxalocrotonate Tautomerase and YwhB with 3-Halopropiolates: Analysis and Implications. 
Biochemistry
 2004, 43 (3), 748Ð758. 100. Whitman, C. P.; Aird, B. A.; Gillespie, W. R.; Stolowich, N. J. Chemical and Enzymic 
Ketonization of 2-Hydroxymuconate, a Conjugated Enol. 
J. Am. Chem. Soc.
 1991, 113
 (8), 3154Ð3162. 101. Huddleston, J. P.; Burks, E. A.; Whitman, C. P. Identification and Characterization of New 
Family Members in the Tautomerase Superfamily: Analysis and Implications. 
Arch. Biochem. 
Biophys.
 2014, 564, 189Ð196. 102. Whitman, C. P.; Hajipour, G.; Watson, R. J.; Johnson, W. H.; Bembenek, M. E.; Stolowich, 
N. J. Stereospecific Ketonization of 2-Hydroxymuconate by 4-Oxalocrotonate Tautomerase and 
5-(Carboxymethyl)-2-Hydroxymuconate Isomerase. 
J. Am. Chem. Soc.
 1992, 114
 (26), 10104Ð10110. 
103. Datsenko, K. A.; Wanner, B. L. One-Step Inactivation of Chromosomal Genes in 
Escherichia coli
 K-12 Using PCR Products. 
Proc. Natl. Acad. Sci. U. S. A.
 2000, 97 (12), 6640Ð6645. 104.Velders, G. J. M.; Feil, D. Comparison of the Hartree-Fock, M¿ller-Plesset, and Hartree-
Fock-Slater Method with Respect to Electrostatic Properties of Small Molecules. 
Theor. Chim. 
Acta 1993, 86 (5), 391Ð416. 105. Samiee, L.; Dehghani, M. M.; Karami, R.; Ayazi, M. Developing of Ethylene Glycol as a 
New Reducing Agent for Preparation of Pd-Ag/PSS Composite Membrane for Hydrogen 
Separation. 
2012. 106.Skrabalak, S. E.; Wiley, B. J.; Kim, M.; Formo, E. V.; Xia, Y. On the Polyol Synthesis of 
Silver Nanostructures: Glycolaldehyde as a Reducing Agent. 
Nano Lett.
 2008, 8 (7), 2077Ð2081. 107. Guzman, A.; Arroyo, J.; Verde, L.; Rengifo, J. Synthesis and Characterization of Copper 
Nanoparticles/Polyvinyl Chloride (Cu NPs/PVC) Nanocomposites. 
Procedia Materials Science
 2015, 9, 298Ð304. 108. Asao, N.; Nogami, T.; Lee, S.; Yamamoto, Y. Lewis Acid-Catalyzed Benzannulation via 
Unprecedented [4+2] Cycloaddition of O-Alkynyl(oxo)benzenes and Enynals with Alkynes. 
J. 
Am. Chem. Soc.
 2003, 125 (36), 10921Ð10925. 109.  The BioBrick approach
. https://j5.jbei.org/j5manual/pages/21.html
 (Accessed 20 Aug. 
2018). %92%93CHAPTER TWO: HETEROLOGOUS EXPRESSION OF KANOSAMINE 
BIOSYNTHESIS PATHWAYS IN 
ESCHERICHIA COLI
 INTRODUCTION 
 Soil-borne plant pathogens are responsible for causing many plant diseases.
1  Widespread 
crop infection can result in significant economic losses.  Bacteria angatonistic to these pathogens 
have been isolated from the rhizosphere and used either alone or in conjunction with chemical 
biocontrol agents to control major crop diseases caused by pathogenic fungi and bacteria.
1-18  In particular, microbes of the genus 
Bacillus
 have been of interest for use in biocontrol strategies.  
Bacillus
 species are attractive candidates for biocontrol agents in agricultural applications due to 
their ability to produce a wide variety of bio-active compounds (Figure 45) and tendency to 
induce plant defense and growth.   
 The many compounds produced and exported by members of the genus 
Bacillus
, some of 
which are depicted in Figure 45, have various modes of action.  Some lipopeptides directly 
inhibit fungal growth, such as iturin A
17 against 
Rhizoctonia solani
, while others including 
surfactin
8,9 and fengycin
8 induce systemic resistance in plants.  In bean and tomato plants, 
surfactin and fengycin induce plant cells to activate lipoxygenase and lipid hydroperoxidase 
enzymes.  These enzymes are crucial for degradation of linolenic and linoleic acids to 
biologically active small molecules. 
%94 % Figure 45.
  Biologically active lipopeptides, siderophores, and volatile organic compounds produced by 
Bacillus 
spp. 
 Organic compounds produced by 
Bacillus
 spp., including 2,5-dimethylpyrazine, 
benzothiazole, 4-chloro-3-methylphenol, directly inhibit growth of pathogenic fungi
17 but may 
possess additional bioactivity.  For example, 2,3-butanediol imparts resistance to 
Erwinia 
carotovora
 via an ethylene-dependent pathway and also induces growth responses in certain 
OHNHONHNH2OHNONOHNOONH2ONHH2NOONHOHNOHOHNH2Oiturin A
bacillibactin
OOOONHOHNOHNOHNOHOOHONHOOHOHONHOOHHOOOHHNOHNOHNONHOHNOOONHONHOOHOsurfactin
2,5-dimethylpyrazine
NNbenzothiazole
SN4-chloro-3-methylphenol
HOCl2,3-butanediol
OHOHHOHNOOHONHONH2HNONHOOHHNONHOOHONOHHNONH2OHNOONHOOHOfengycin
%95strains of the mustard plant, 
Arabidopsis thaliana
.15  Siderophores and phytohormones produced 
by Bacillus
 spp. have been found to impact the growth of various plants.  For example, red 
peppers treated with bacillibactin, a catecholic siderophore, grew 27-54% taller, bore fruit that 
was 37% heavier, and flowered 17 days sooner than untreated plants.  Treated pepper also had 
reduced instance of 
Fusarium 
wilt (12-57%) relative to untreated plants.
16  Products based on 
Bacillus
 spp. currently commercialized by phytopharmaceutical companies include EcoGuard¨ 
by Novozymes
11 (Bacillus lichenformis
 isolate SB3086), Serenade¨ by Bayer
12 (Bacillus subtilis
 isolate QST30002), and Kodiak¨ by Gustafson
13 (Bacillus subtilis
 isolate GB03). 
% Figure 46.
  Structure of kanosamine (3-amino-3-deoxy-
D-glucose). 
 Among the biologically active compounds produced and exported by 
Bacillus 
spp. is 
kanosamine (3-amino-3-deoxy-
D-glucose, Figure 46).  Unlike other aminosugars of natural 
origin which are produced as a structural component of larger compounds, kanosamine is the 
only aminosugar to be exported in a free, monomeric form.
19  Like the compounds described 
above, kanosamine also displays important biological activity in the environment.  For example, 
kanosamine inhibits the growth of the oomycete 
Phytophthora medicaginis
, a fungus-like 
eukaryote responsible for damping-off disease in alfalfa, as well as select phytopathogenic fungi 
including 
Venturia inaequalis 
(apple scab)
 and 
Ustilago maydis 
(corn smut).
20  Minimal 
OHONH2OHOHHOkanosamine
%96inhibitory concentrations of kanosamine against many bacteria and yeasts are high (Table 17), 
but in combination with other antibiotics produced by 
Bacillus 
spp. such as zwittermycin A, 
kanosamine displays additive or synergistic activity.
23,24  Equipping crops with the genes 
required for kanosamine biosynthesis or incorporating a kanosamine-overproducing organism 
into soil alongside 
Bacillus 
spp. could serve as biocontrol strategies that both enhance plant 
resistance to pathogenic fungi and strengthen the inhibitory effects of antibiotic compounds 
produced by 
Bacillus 
spp. currently utilized for biocontrol of phytopathogens  
Table 17.
  Minimal inhibitory concentrations of kanosamine against various bacteria and yeast. 
 Two pathways have been elucidated for kanosamine biosynthesis in different 
Bacillus 
strains.  In 
Bacillus subtilis 
(Figure 47) kanosamine is biosynthesized from 
D-glucose-6-
phosphate through a pathway beginning with dehydrogenation to afford 3-keto-
D-glucose-6-
Organism
MIC  (
#
g/mL)
Reference
Staphylococcus aureus
 FDA 209P
5021S. aureus 
FDA 209P JC-1
10021S. aureus 
FS-1227 (penicillin resistant)
10021Sarcina lutea
 PCI 1001
10021Cytophaga johnsonae
30021Escherichia coli
 NIHJ JC-2
>80021Salmonella typhimurium
>80021Pseudomonas aeruginosa
 P-3
>80021Candida albicans
 ATCC 26278
1,25022Escherichia coli 
IAM 1253
>2,00030Saccharomyces cerevisiae 
ATCC 9763
2,50022%97phosphate.  The C-3 keto functionality is converted to an amine by an 
L-glutamate-dependent 
aminotransferase, followed by hydrolysis of the phosphate monoester of kanosamine-6-
phosphate.  This pathway was first discovered in 2013 from 
in vitro
 experiments using purified 
Ntd proteins.
25  The 
ntdC
, ntdA
, and 
ntdB
 genes of 
B. subtilis
 encode, respectively, 
D-glucose-6-
phosphate 3-dehydrogenase, 3-keto-
D-glucose-6-phosphate:
L-glutamate aminotransferase, and 
kanosamine-6-phosphate kinase.  Crystal structures of NtdA
26 and kinetic studies of NtdC have 
been published. 
% Figure 47.
  Kanosamine biosynthesis from 
D-glucose-6-phosphate described in 
Bacillus subtilis
.  An alternate pathway was described in 
Bacillus cereus
28 and 
Bacillus pumilus
.29  The 
pathway (Figure 48) begins with UDP-
D-glucose, which undergoes a similar reaction sequence 
of oxidation, transamination, and hydrolysis to afford kanosamine.  This pathway was first 
described in 1967 based on in vitro experiments using cell-free extracts of 
B. pumilus
.30  Using a 
reaction mixture consisting of cell-free extract, uniformly labeled 
D-[U-
14C]-glucose, MgSO
4, NAD+, L-glutamine, ATP, and UTP in phosphate buffer the total radioactivity of the kanosamine 
produced was 2,800 counts per minute (cpm).  This value fell to 215 cpm if both ATP and UTP 
OHOHOOHOHH2O3POOHOOOHOHH2O3POOHOH2NOHOHH2O3POOHOH2NOHOHHONAD
+NADH
NtdCD-glucose-
6-phosphate
3-keto-
D-glucose-
6-phosphate
kanosamine-
6-phosphate
kanosamine
NtdANH2OHOOOHOOHOOOHH2OHOPO
32-NtdB%98were omitted from the reaction mixture.  The observed requirement for ATP and UTP led to the 
proposed pathway shown in Figure 48.
30        % Figure 48.
  Kanosamine biosynthesis pathway via UDP-
D-glucose as proposed by Umezawa.
30  In other bacterial species, kanosamine serves as a biosynthetic precursor to 3-amino-5-
hydroxybenzoic acid (AHBA), which serves as a core for the ansamycins, including the 
rifamycins, and the mitomycin families of antibiotics (Figure 49).
31  AHBA produced via the 
aminoshikimate pathway (Figure 50) is required in the biosynthesis of a variety of secondary 
metabolites with biological activity.
31 . NH2OH2NOOHNH2OHOOOHOHOHOOHOHHOD-glucose
ATP
ADP
Mg2+OHOHOOHOPO
3H2HOD-glucose-
1-phosphate
UTPPO32-OHOHOOHOUDPHOUDP-
D-glucose
NAD
+NADH
OHOOOHOUDPHO3-keto-UDP-
D-glucose
OHOH2NOHOUDPHOUDP-kanosamine
OHOH2NOHOHHOkanosamine
PO32-UMP
UDP = uridine diphosphate
OHOOHNHOOOPOOOHPOOH%99%   Figure 49.
  Representative members of the ansamycin and mitomycin families of antibiotics. 
 The AHBA biosynthesis gene cluster was described in the rifamycin-producing organism, 
Amycolatopsis mediterranei
, which is a Gram positive bacteria of the family 
Actinobacteria
 distinguished by their high genomic G+C content ranging from 51% to over 70% G+C.
32  The 
functions of many AHBA biosynthesis enzymes were assigned using feeding experiments with a 
variety of labeled precursors and genetic mutants.
33,34  The origin of 3-amino-3-deoxy-
D-arabino
-heptulosonate-7-phosphate (aminoDAHP) biosynthesis remained unknown for several 
years following sequencing of the AHBA biosynthesis gene cluster.  Progress was made 
following the discovery of an essential transketolase-encoding gene, 
orf15AB
, within the AHBA 
biosynthesis genes, suggesting that aminoDAHP may be produced from condensation of 
phosphoenolpyruvate (PEP) and a nitrogen-containing analog of erythrose-4-phosphate (E4P).
34   AnsamycinsMitomycins
rifamycin B
OHOONHOOOHOOHOOOOOHOHgeldanamycin
OOOOOHONH2ONHOOmitomycin C
ONNHOOOH2NOH2NHHmitomycin A
OOONONHONH2O%100% Figure 50.
  Synthesis of kanosamine from UDP-
D-glucose and the formation of 3-amino-5-hydroxybenzoic acid via 
the aminoshikimate pathway in 
Amycolatopsis mediterranei.
 This hypothesis was proven when incubation of 3-amino-3-deoxyfructose-6-phosphate 
(aminoF6P), PEP, and ribose-5-phosphate with purified 
rifH-encoded aminoDAHP synthase 
afforded aminoDAHP in 2% yield and DAHP in 35% yield.
36  An 
E. coli
 transketolase-catalyzed 
reaction of 3-[
15N]-6,6-[
2H2]-aminoF6P demonstrated that the framework and nitrogen atom of 
NH2OH2NOOHOOH2NOOHOHOHOOHOUDPHOUDP-
D-glucose
OHOOOHOUDPHO3-keto-UDP-
D-glucose
OHOH2NOHOUDPHOUDP-kanosamine
OHOH2NOHOHHOkanosamine
RifLNAD+
NADH
RifKRifMH2OUDPRifMATP
ADP
OHOH2NCH3OHH2O3POkanosamine
6-phosphate
isomerase
OCH2OHOHNH2HOH2O3POaminofructose
6-phosphate
transketolase
OOHOHOHOPO
3H2HNOHOHOPO
3H21-deoxy-1-imino-
D-erythrose-
4-phosphate
RifHCO2HOPO
3H2OH2O3POHOH2NOHCO2H3-amino-3-deoxy-
D-arabino
-heptulosonate
7-phosphate
RifGNAD
+OHOH2NOHCO2H5-amino-5-deoxy-
3-dehydroquinate
RifJH2OOHOH2N5-amino-5-deoxy-
3-dehydroshikimate
CO2HRifKPLP
OH3-amino-5-hydroxy
benzoic acid
(AHBA)
CO2HH2Nrifamycin B
OHOONHOOOHOOHOOOOOHOHaminoshikimate pathway
%101aminoF6P is incorporated into aminoDAHP.
35  The origin of aminoF6P in the aminoshikimate 
pathway was confirmed following incubation of kanosamine-6-phosphate, ribose-5-phosphate, 
and PEP with 
E. coli
 tktA-encoded transketolase, yeast phosphoglucose isomerase, and 
rifH-encoded aminoDAHP synthase.
37  Isomerization of kanosamine-6-phosphate to aminoF6P 
catalyzed by yeast phosphoglucose isomerase during the reaction resulted in accumulation of 
aminoDAHP and DAHP in 2% and 30% yields, respectively.  Repeating the incubation of 
kanosamine-6-phosphate with 
A. mediterranei
 cell-free extract in place of 
E. coli
 tktA-encoded 
transketolase, yeast phosphoglucose isomerase, and purified 
rifH-encoded aminoDAHP synthase 
afforded aminoDAHP and DAHP in 6% and 20% yields, respectively, along with AHBA (2% 
yield), tyrosine (3% yield), and phenylalanine (2% yield).
37   In this work, genes encoding kanosamine biosynthetic enzymes from 
B. subtilis
, A. 
mediterranei
, and 
B. pumilus 
are expressed in 
Escherichia coli 
to determine whether 
heterologous expression results in kanosamine accumulation and if the system can be 
manipulated to maximize kanosamine production.  Recombinant 
E. coli
 are evaluated in fed-
batch fermentation at 2 L scale using two substrate feeding regimes.  Kanosamine does not 
inhibit growth of 
E. coli
 strains used in the fermentations and can be isolated from fermentation 
broth.  Recombinant 
E. coli
 expressing genes from 
B. subtilis
 168 produces 12.7 ± 0.6 g/L 
kanosamine in a 4.4 ± 0.4% mol/mol yield from 
D-glucose via the 
D-glucose-6-phosphate 3-
dehydrogenation-pathway described above.  Kanosamine yields were increased to 18 ± 1% mol/
mol by blocking the Embden-Meyerhoff-Parnas pathway through a mutation in 
pgi-encoded 
phosphoglucose isomerase.  
Amycolatopsis mediterranei 
S699 enzymes prove to be challenging 
to express efficiently in 
E. coli
 and did not result in kanosamine accumulation via UDP-
D-%102glucose in fermentation broth.  In light of this, 
B. pumilus 
was examined as a source of 
kanosamine biosynthesis genes in order to evaluate kanosamine production from UDP-
D-glucose.  Expression of 
B. pumilus
 SH-B11 genes in 
E. coli
 results in 6.3 g/L kanosamine titers 
in 4.4% mol/mol yield from 
D-glucose in glucose-limited fermentations.  Surprisingly, our 
results indicate that 
B. pumilus
 does not utilize UDP-
D-glucose but rather 
D-glucose-6-
phosphate, which suggests a departure from what is in the literature regarding kanosamine 
production in 
Bacillus pumilus
.  %103PART I 
  KANOSAMINE PRODUCTION VIA EXPRESSION OF 
BACILLUS SUBTILIS 
GENES
 1.  Variations in gene arrangement and promoter strength
  Bacillus subtilis 
168 is a common strain employed during studies of biocontrol methods 
on various plants.  Kanosamine biosynthesis in 
B. subtilis
 has been reported to proceed via 
D-glucose-6-phosphate in a sequence of three reactions catalyzed by 
ntdC
-encoded 
D-glucose-6-
phosphate 3-dehydrogenase, 
ntdA
-encoded 3-keto-
D-glucose-6-phosphate:
L-glutamate 
aminotransferase, and 
ntdB
-encoded kanosamine-6-phosphate phosphatase (Figure 46).
25  For 
expression of these enzymes in 
E. coli
, the BglBrick assembly standard of cloning was used to 
build a plasmid which harbored 
ntdC
, ntdA
, and 
ntdB
, as well as 
serA.  
All bacterial
 strains and 
plasmids used in this Chapter are presented in Table 18.  As described in Chapter 1, 
E. coli
 strains used in this study carry a mutation in the 
serA locus of the genome which prevents the 
synthesis of 
L-serine.  Plasmid-localized 
serA serves as the basis of plasmid maintenance when 
cells are cultured in a defined minimal media.   
Table 18
.  Strains and plasmids used in Chapter 2 of this work. 
Strain/plasmid
Relevant characteristics
Source or reference
Escherichia coli 
DH5
!FÕ 
"80lacZ
+M15 +(lacZYA-argF)U169
 deoR recA1 endA1 
hsdR17
(rk*, mk+) phoA 
#* supE44 thi-1 gyrA96 relA1
Invitrogen
Escherichia coli
 AB2834
tsx-352 glnV42 aroE353 malT352 
#$Coli Genetic Stock 
Center
58Escherichia coli
 KL3
AB2834 
serA::aroB
lab54Escherichia coli 
W3110
F$ IN(rrnD-rrnE)1 
#$ rph1 
Escherichia coli
 RB791serAW3110
 lacL8I
q serA
lab40%104Table 18 (contÕd)
Escherichia coli
 RB791serA(DE3)
RB791serA #
(DE3)
lab56Escherichia coli
 RB791serA pgi
::KanRRB791serA pgi::KanRlab46Bacillus subtilis
 168source for 
ntdA
sub
ntdB
sub
ntdC
sub
ATCC 23857
pBbA1a-RFP
ApR, lacI
q, Ptrc
rfp p15A replicon
Addgene #35334
57pBbA7k-RFP
KanR, lacI
q, PT7rfp, p15A replicon
Addgene #35318
57pKM6.240
Ptrc
serA in pBbA1a
Chapter 1
pSN1.294
Ptrc
ntdC
sub
lab (described 
herein)
pSN1.295
Ptrc
ntdB
sub
lab (described 
herein)
pSN1.296
Ptrc
ntdA
sub
lab (described 
herein)
pSN1.119
Ptrc
ntdC
sub
ntdA
sub 
in pBbA1a
lab (described 
herein)
pSN1.139
Ptrc
ntdC
sub
ntdA
sub
ntdB
sub
 in pBbA1a
lab (described 
herein)
pSN1.292
Ptrc
ntdC
sub
ntdA
sub
ntdB
sub
serA in pBbA1a
lab (described 
herein)
pJG7.246
ApR, PT5, His
6, lacI
q, ColE1 replicon
lab49pKM9.279C
PT5His6-ntdC
sub
 in pJG7.246
this work
pKM10.82A
PT7ntdC
sub
ntdA
sub
ntdB
sub
serA 
in pBbA7k
this work
pKM10.204A
Ptrc
ntdA
sub
serA in pBbA1a
this work
pKM10.204B
Ptrc
ntdB
sub
serA in pBbA1a
this work
pKM10.204C
Ptrc
ntdC
sub
serA in pBbA1a
this work
pKM10.214CA
Ptrc
ntdC
sub
ntdA
sub
serA in pBbA1a
this work
pKM10.214CB
Ptrc
ntdC
sub
ntdB
sub
serA in pBbA1a
this work
pKM10.214AB
Ptrc
ntdA
sub
ntdB
sub
serA in pBbA1a
this work
pKM10.214BC
Ptrc
ntdB
sub
ntdC
sub
serA in pBbA1a
this work
pKM10.220ABC
Ptrc
ntdA
sub
ntdB
sub
ntdC
sub
serA in pBbA1a
this work
pUCIDT(Kan)-rifK
KanR, Plac
, codon optimized 
rifK, ColE1 replicon
IDT
pUCIDT(Kan)-rifL
KanR, Plac
, codon optimized 
rifL, ColE1 replicon
IDT
pUCIDT(Kan)-rifM
KanR, Plac
, codon optimized 
rifM, ColE1 replicon
IDT
%105 The 
ntd
 genes were PCR-amplified (Figure 51) from 
B. subtilis
 168 genomic DNA with 
5Õ 
EcoRI and 
Bgl
II and 3Õ 
Bam
HI and 
Xho
I sites.  Each gene was first inserted into the 
EcoRI
 and 
Xho
I sites of plasmid pBbA1a to afford a BglBrick part (Figure 51).  These parts were used 
in the BglBrick assembly standard by iteratively inserting one gene in front of another using a 
routine sequence of digestion and ligations.  Firstly, each 
ntd
 gene was inserted upstream of 
serA 
by digesting the BglBrick part containing 
serA, pKM6.240 (Figure 51), with 
EcoRI and 
Bgl
II to 
create a double-stranded break upstream of 
serA.  Digestion of the 
ntd-
containing BglBrick parts, 
pSN1.294, pSN1.295, and pSN1.296 (Figure 51), with 
EcoRI and 
Bam
HI followed by ligation of 
Table 18 (contÕd)
pKM10.254K
Ptrc
rifK in pBbA1a
this work
pKM10.254L
Ptrc
rifL in pBbA1a
this work
pKM10.254M
Ptrc
rifM in pBbA1a
this work
pKM10.268K
Ptrc
rifKserA
 in pBbA1a
this work
pKM10.268L
Ptrc
rifLserA in pBbA1a
this work
pKM10.268M
Ptrc
rifMserA
 in pBbA1a
this work
pKM10.278KM
Ptrc
rifKrifMserA
 in pBbA1a
this work
pKM10.284L
PT5His6-rifL in pJG7.246
this work
pKM11.9
Ptrc
rifLrifKrifMserA
 in pBbA1a
this work
pKM11.46
PT7rifLrifKrifMserA
 in pBbA7k
this work
pUCIDT(Kan)-ntdA
KanR, Plac
, ntdA
pum, ColE1 replicon
IDT
pUCIDT(Kan)-ntdB
KanR, Plac
, ntdB
pum, ColE1 replicon
IDT
pUCIDT(Kan)-ntdC
KanR, Plac
, ntdC
pum, ColE1 replicon
IDT
pKM11.121A
Ptrc
ntdA
pum in pBbA1a
this work
pKM11.121B
Ptrc
ntdB
pum in pBbA1a
this work
pKM11.121C
Ptrc
ntdC
pum in pBbA1a
this work
pKM11.125
Ptrc
ntdB
pumserA in pBbA1a
this work
pKM11.130
Ptrc
ntdC
pumntdA
pum in pBbA1a
this work
pKM11.135
Ptrc
ntdC
pumntdA
pumntdB
pumserA in pBbA1a
this work
%106the digested DNA afforded pSN1.288, pKM10.204A, and pKM10.204C.  During the ligation, the 
EcoRI site was reformed and the complementary cohesive 
Bam
HI and 
Bgl
II digested DNA ends 
created a glycine-serine scar sequence between 
ntdB
 and 
serA.  Importantly for the BglBrick 
assembly standard, all four restriction sites (
EcoRI, Bgl
II, Bam
HI, 
Xho
I) that were originally 
present in the pKM6.240 recipient plasmid remain in the new plasmid.  This allows repeated 
cycles of digesting the vector plasmid with 
EcoRI and 
Bgl
II, digesting the plasmid containing the 
desired insert with 
EcoRI and 
Bam
HI, and using T4 DNA ligase to reform the 
EcoRI site and 
create a glycine-serine scar sequence between the newly inserted gene and the gene at the 5Õ end 
of the operon.  Using this method, 
ntdC 
was inserted upstream of 
ntdA 
in plasmid pSN1.296 to 
afford pSN1.119 (Figure 52) which contained a 
Ptrc
ntdCntdA
 cassette.  The 
trc
 promoter is 
encoded on the original BglBrick vector, pBbA1a, and is optimally positioned for expression of 
the inserted genes.  The 
ntdCntdA 
fragment was inserted upstream of 
ntdB
 in plasmid pSN1.295 
to afford pSN1.139, which contained a 
Ptrc
ntdCntdAntdB
 cassette (Figure 53).  Finally, the 
ntdCntdAntdB
 gene fragment was inserted upstream of 
serA in plasmid pKM6.240 to afford 
plasmid pSN1.292, containing a 
Ptrc
ntdCntdAntdBserA
 cassette (Figure 54).  Similarly, a plasmid 
was created that contained a 
Ptrc
ntdAntdBntdCserA
 cassette by inserting 
ntdB 
from pSN1.295 
into pKM10.204C (Figure 55) to afford pKM10.214BC (Figure 56), followed by insertion of 
ntdA
 from pSN1.296 into pKM10.214BC to afford pKM10.220ABC, containing the full set of 
ntd
 genes (Figure 57). 
%107% Figure 51.
  Construction of 
ntd
 BglBrick parts: plasmids pSN1.296, pSN1.295, and pSN1.294 
1.3 kb
0.8 kb
1.0 kb
Bacillus subtilis 
168EcoRI, XhoI
 digest
EcoRI, Xho
I digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IntdBntdCAmpRlacI
qp15Aori
mRFP-1
pBbA1a-RFP
4.3 kb
Ptrc
AmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
ntdAntdBntdCpSN1.296
4.9 kb
pSN1.295
4.4 kb
pSN1.294
4.6 kb
ntdAEcoRIBam
HIEcoRIBam
HIEcoRIBam
HIXho
IXho
IXho
IBgl
IIBgl
IIBgl
II%108% Figure 52.
  Construction of plasmid pSN1.119. 
1.0 kb
pSN1.294
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IntdCAmpRlacI
qp15Aori
Ptrc
ntdCpSN1.119
5.8 kb
EcoRIBam
HIEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ntdApSN1.296
4.8 kb
ntdABgl
II%109% Figure 53.
  Construction of plasmid pSN1.139. 
pSN1.119
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ntdBpSN1.295
4.4 kb
2.3 kb
ntdCntdAEcoRIBam
HIBgl
IIEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ntdCpSN1.139
6.0 kb
ntdBntdA%110
% Figure 54.
  Construction of plasmid pSN1.292. 
pSN1.139
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligation3.3 kb
ntdCntdAntdB
EcoRIBam
HIBgl
IIEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ntdCpSN1.292
8.1 kb
serA
ntdAntdBEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
serA
pKM6.240
4.8 kb
%111% Figure 55. 
 Construction of plasmids pKM10.204A, pKM10.204B, and pKM10.204C. 
1.3 kb
0.8 kb
1.0 kb
pSN1.296
pSN1.295
pSN1.294
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IntdBntdCAmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
serA
ntdBntdCpKM10.204A
6.1 kb
pKM10.204B
5.7 kb
pKM10.204C
5.9 kb
ntdAEcoRIBam
HIEcoRIBam
HIEcoRIBam
HIEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
serA
pKM6.240
4.8 kb
ntdAserA
serA
Bgl
IIBgl
IIBgl
II%112
% Figure 56.
  Construction of plasmid pKM10.214BC. 
Bgl
II0.8 kB
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IntdBAmpRlacI
qp15Aori
Ptrc
ntdBpKM10.214BC
6.7 kb
EcoRIBam
HIserA
ntdCEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ntdCpKM10.204C
5.9 kb
serA
pSN1.295
%113
% Figure 57.
  Construction of pKM10.220ABC. 
Bgl
II1.3 kb
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IntdAAmpRlacI
qp15Aori
Ptrc
ntdApKM10.220ABC
8.1 kb
EcoRIBam
HIserA
ntdBEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ntdBpKM10.214BC
6.7 kb
serA
ntdCntdCpSN1.296
%114
 Fed-batch fermentation of RB791
serA constructs were conducted as described in Chapter 
1 following either glucose-rich or glucose-limited conditions.  Glucose-rich conditions maintain 
an excess (20-30 g/L) of 
D-glucose in the culture throughout the course of a fermentation run.  
Initial glucose was added to the fermentation vessel prior to inoculation at a concentration of 30 
g/L.  This concentration was chosen to ensure that the amount of 
D-glucose present in the culture 
medium at the end of the batch phase (approximately 12 h after inoculation) and before 
D-glucose feeding began, did not fall below 20 g/L.  Glucose-rich fermentations used three staged 
methods to maintain the dissolved oxygen (D.O.) concentration at 10% air saturation.  The first 
stage maintained airflow at an initial setting of 0.06 L/L/min and increased impeller speed from 
an initial set point of 50 rpm to a preset maximum of 750 rpm.  This maximum was reached 
approximately 9 h after inoculation.  With an impeller speed maintained at 750 rpm, the second 
stage began in which the mass flow controller maintained D.O. concentration by increasing 
airflow from 0.06 L/L/min to a preset maximum of 1.0 L/L/min.  Approximately 3 h were 
needed to reach maximum airflow which coincided with cells beginning exponential growth 
phase.  Between 11-12 h into a fermentation run after the maximum set points for airflow and stir 
rate were reached, the third stage of the fermentation was entered in which 
D-glucose (60% w/v) 
was added at a rate sufficient to maintain 
D-glucose concentration of 20-30 g/L throughout the 
run.  The start of 
D-glucose feeding corresponded to the beginning of exponential growth phase 
for RB791serA constructs.  At 12 h, isopropyl 
#
-D-1-thiogalactopyranoside 
 (IPTG, 12 mg unless otherwise stated) was added to induce transcription of plasmid-localized 
ntd
 genes.  After initial addition of IPTG at 12 h, addition occurred every 6 h until the end of the 
%115
fermentation run.  Yields of metabolites from 
D-glucose were calculated based on observed 
amount of glucose added to the fermentation culture.  
  Glucose-limited fermentation conditions utilized oxygen-sensor-controlled glucose 
feeding to maintain 
D-glucose concentrations well below 1 g/L throughout the course of a 
fermentation.  Precise concentrations were not be measured as 
D-glucose levels were too low to 
detect using a GlucCellª meter or to be detected by 
1H NMR.  Initial glucose was added to the 
fermentation medium prior to inoculation at a concentration of 20 g/L.  This concentration was 
selected to ensure the complete consumption of 
D-glucose corresponded to cells entering 
exponential growth phase and the beginning of 
D-glucose feeding.  Three staged methods were 
used during glucose-limited fermentations to ensure D.O. concentration was maintained at 10% 
air saturation.  The first two stages were similar to those described for glucose-rich fermentations 
with a different preset maximum impeller speed.  With airflow maintained at 0.06 L/L/min, 
impeller speed was allowed to increase during the first stage from 50 rpm to a preset maximum 
of 850 rpm.  This maximum value was reached approximately 9 h after inoculation of the 
fermentation vessel.  The second staged method maintained impeller speed at 850 rpm while 
increasing airflow to a preset maximum of 1.0 L/L/min.  Approximately 1 h was required after 
the maximum airflow was reached for the initial glucose to be completely consumed.  This was 
marked by a sudden increase in D.O. levels and a corresponding decrease of airflow.  As soon as 
the D.O. spike was observed, oxygen-sensor-controlled glucose feeding was initiated.  Impeller 
speed and airflow was manually set to the maximum levels of 850 rpm and 1.0 L/L/min, 
respectively, while the mass flow controller adjusted glucose feeding rate to maintain D.O. 
concentration at 10% air saturation.  As in glucose-rich fermentations, IPTG (12 mg unless 
%116
otherwise stated) was added beginning at 12 h and continued every 6 h until the end of a 
fermentation run. 
 Kanosamine concentrations were calculated by 
1H NMR analysis of cell-free 
fermentation broth.  Integrals corresponding to the 
!
-anomeric proton (4.73 ppm, d, 
J = 7.2 Hz) 
and the 
#
-anomeric proton (5.27 ppm, d, 
J = 3.6 Hz) were summed to estimate the molar 
concentration of kanosamine in the NMR sample.  This value was used to calculate the 
kanosamine concentration in the fermentation sample using a calibration curve with sodium 3-
(trimethylsilyl)propionate-
2,2,3,3-d4 (TSP) as an internal standard and a response factor of 1.013.  
The calibration curve was generated using 
D-glucosamine hydrochloride (>98%, Alfa Aesar), 
which is a 2-amino regioisomer of kanosamine.  The longitudinal relaxation time (
T1) for D-glucosamine calculated via an inversion-recovery experiment was found to be the same as 
kanosamine (5.2 s for the 
#
-anomeric proton).  A relaxation delay of 26 s was used to acquire 
1H NMR spectra. 
 The precise relationship between the arrangement of genes that comprise an operon and 
expression of those genes is not completely understood.  However, it was recently demonstrated 
using synthetic operons in 
E. coli
 that the expression of a given gene increases with the length of 
the operon and as its position moves closer to the promoter.
38  The authors suggested that the 
increased distance from the start of a given gene to the end of the operon provides more time for 
translation to occur during transcription, leading to increased expression.  To evaluate whether 
the arrangement of plasmid-localized 
ntd 
genes between the 
trc
 promoter and 
rrnB 
transcription 
terminator affects kanosamine production in RB791
serA, two plasmids were utilized in fed-batch 
fermentations with different arrangements of the same 
B. subtilis
 ntd
 genes.  Plasmid 
%117
pKM10.220ABC contains an 
ntdAntdBntdCserA 
cassette behind 
Ptrc
 to match the order of genes 
observed in the 
B. subtilis 
genome while plasmid pSN1.292 carries a 
Ptrc
ntdCntdAntdBserA
 cassette.  This second arrangement places the genes in a Òreaction sequence orderÓ with the 
ntdC
-encoded 
D-glucose-6-phosphate 3-dehydrogenase placed closest to 
Ptrc
 and the 
ntdB
-encoded kanosamine-6-phosphate phosphatase placed furthest away.  These two plasmids were 
utilized to determine if the particular arrangement of 
ntd
 genes has an impact on kanosamine 
production in fed-batch fermentations in 
E. coli
.          Fed-batch fermentations of RB791
serA/pSN1.292 resulted in 12.7 ± 0.6 g/L kanosamine 
(4.4 ± 0.4% mol/mol yield from 
D-glucose) under glucose-rich conditions (Figure 58) and 8.5 ± 
1.4 g/L kanosamine (6.2 ± 2.2% mol/mol yield from 
D-glucose) under glucose-limited conditions 
(Figure 59).  Glucose-rich conditions led to much higher cell-densities (40-50 g/L dry cell 
weight) compared to glucose-limited conditions (20-30 g/L dry cell weight).  This increased 
biomass is a result of excess available carbon present in the culture media.  Glucose-rich 
conditions also contribute to greater 
D-glucose consumption during a 48 h fermentation run than 
glucose-limited conditions.  After 48 h, an average of 293 g (1.64 mol) 
D-glucose was consumed 
during glucose-rich fermentations while an average of 144 g (0.80 mol) 
D-glucose was consumed 
during glucose-limited fermentations.  This difference in 
D-glucose consumption results in 
higher yields of kanosamine in glucose-limited fermentations despite lower overall titers.    
%118
% Figure 58.
  Kanosamine production by 
E. coli 
RB791serA/pSN1.292 under glucose-rich conditions.  Legend: 
kanosamine, green columns; dry cell weight, black circles.  Error bars represent one standard deviation away from 
the mean across triplicate runs. 
% Figure 59.
  Kanosamine production by 
E. coli 
RB791serA/pSN1.292 under glucose-limited conditions.  Legend: 
kanosamine, green columns; dry cell weight, black circles.  Error bars represent one standard deviation away from 
the mean across quadruplicate runs.
     A decrease in rate of kanosamine accumulation appears to correlate with stationary 
growth phase in both glucose-rich (Fig. 58) and glucose-limited (Fig. 59) fermentations of 
RB791serA/pSN1.292.  As cell growth slows, so does kanosamine production which may reflect 
Dry cell weight  (g/L)01020304050607080Products  (g/L)
0246810121416Time  (h)12182430364248Dry cell weight  (g/L)01020304050607080Products  (g/L)
0246810121416Time  (h)12182430364248%119
the lower metabolic rates for 
E. coli
 in stationary phase relative to exponential phase.
38  This 
feature of both glucose-rich and glucose-limited fermentations has an impact on the yields from 
D-glucose reported here.  During stationary phase, kanosamine production slows yet 
D-glucose 
continues to be taken in by the cells.  The highest yields of kanosamine would result from halting 
D-glucose feeding at the moment kanosamine production slows.  However, as this would lead to 
lower titers than those observed herein, 
D-glucose feeding continued into stationary phase for 
both glucose-rich and glucose-limited fermentations. 
 Fed-batch fermentations of RB791
serA/pKM10.220ABC (Figure 60 and Figure 61) 
afforded similar titers and yields of kanosamine relative to RB791
serA/pSN1.292.   In glucose-
rich conditions, RB791
serA/pKM10.220ABC achieved kanosamine titers of 13.7 g/L in 4.9% 
mol/mol yield from 
D-glucose while RB791
serA/pSN1.292 achieved titers of 12.7 ± 0.6 g/L in 
4.4 ± 0.4% mol/mol yield from 
D-glucose.  During glucose-limited fermentations, RB791
serA/pKM10.220ABC produced 7.9 g/L kanosamine in 5.6% mol/mol yield from 
D-glucose while 
RB791serA/pSN1.292 produced 8.5 ± 1.4 g/L kanosamine in 6.2 ± 2.2% mol/mol yield.  The 
similarities between fermentations employing the two constructs suggest that the particular 
arrangement of 
ntd
 genes behind 
Ptrc
 does not have a great impact on kanosamine production 
with RB791
serA.  Based on these results, alternate gene permutations were not explored and the 
Òreaction sequence orderÓ arrangement of 
ntdCntdAntdB 
was used for further experiments.  
%120% Figure 60.
  Kanosamine production by 
E. coli 
RB791serA/pKM10.220ABC under glucose-rich conditions.  Legend: 
kanosamine, grey columns; dry cell weight, black circles.  
% Figure 61.
  Kanosamine production by 
E. coli 
RB791serA/pKM10.220ABC under glucose-limited conditions.  
Legend: kanosamine, grey columns; dry cell weight, black circles.  
 Such experiments included placing the 
ntdCntdAntdBserA
 cassette from pSN1.292 
behind a T7 promoter, which is not recognized by native 
E. coli
 RNA polymerase.  Kanosamine 
production was therefore evaluated in RB791
serA(DE3), previously engineered by Hansen and 
Dry cell weight  (g/L)01020304050607080Products  (g/L)
0246810121416Time  (h)121824303642Dry cell weight  (g/L)01020304050607080Products  (g/L)
0246810121416Time  (h)1218243036%121Frost.
40  The DE3 designation indicates the respective strain contains the 
"
 prophage DE3 
lysogen.  The 
"
DE3 lysogen carries the gene for T7 RNA polymerase under control of an IPTG-
inducible lacUV5 promoter.  Strength of a promoter is determined by the relative frequency of 
transcription initiation, which is mainly affected by the affinity of the promoter sequence for 
RNA polymerase.  The bacteriophage 
"
 T7 RNA polymerase has a high frequency of initiation 
and is not easily subject to early termination events.  These features result in transcription rates 
approximately five-fold higher than 
E. coli
 RNA polymerase.
41  To build the T7 expression 
plasmid, the 4.5 kb 
ntdCntdAntdBserA
 fragment from pSN1.292 was liberated by digestion with 
EcoRI, Bam
HI double-digestion and inserted immediately behind 
PT7 into the 
EcoRI, Bam
HI 
sites of BglBrick vector, pBbA7k to afford pKM10.82A (Figure 62).  Plasmid pKM10.82A 
carries a kanamycin resistance gene and a p15A origin of replication.  The orientation of the 
ntdCntdAntdBserA
 cassette was in the same direction as the T7 promoter.   Transformation of 
pKM10.82A into RB791
serA(DE3) was followed by evaluation of kanosamine production in 
both glucose-rich (Figure 63) and glucose-limited fed-batch fermentation (Figure 64). 
%122% Figure 62.
  Construction of plasmid pKM10.82A.
 Bgl
II4.5 kb
EcoR
I, BamH
I digest
EcoR
I, Bam
HI digest
T4 ligationntdCntdAntdBserA
EcoR
IBgl
IIBamH
IXho
IlacI
qp15Aori
mRFP-1
pBbA7k-RFP
4.3 kb
EcoR
IBamH
IEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ntdApSN1.292
8.1 kb
serA
ntdBntdCEcoR
IBgl
IIBamH
IXho
IKan
RlacI
qp15Aori
ntdApKM10.82A
8.1 kb
serA
ntdBntdCKan
RPT7PT7%123% Figure 63.
  Kanosamine production by 
E. coli 
RB791serA(DE3)/pKM10.82A under glucose-rich conditions.  
Legend: kanosamine, green columns; dry cell weight, black circles.  Error bars represent one standard deviation 
away from the mean across duplicate runs. 
% Figure 64.
  Kanosamine production by 
E. coli 
RB791serA(DE3)/pKM10.82A under glucose-limited conditions.  
Legend: kanosamine, green columns; dry cell weight, black circles.  Error bars represent one standard deviation 
away from the mean across duplicate runs.
    Fermentation results shown in Fig. 63 and Fig. 64 utilized 12 mg IPTG added every 6 h 
beginning at 12 h.  This is in the same manner as fermentations of RB791
serA/pSN1.292, which 
carried the 
ntd
 genes under the control of 
Ptrc
.  Lower titers and yields of kanosamine were 
Dry cell weight  (g/L)01020304050607080Products  (g/L)
0246810121416Time  (h)121824303642Dry cell weight  (g/L)01020304050607080Products  (g/L)
0246810121416Time  (h)12182430364248%124observed using 
PT7 expression relative to 
Ptrc
 using the same method of induction, affording 7.4 
± 0.6 g/L (2.6 ± 0.3% mol/mol yield from 
D-glucose) under glucose-rich conditions and 2.0 ± 0.3 
g/L (1.4 ± 0.2% mol/mol yield from 
D-glucose) under glucose-limited conditions.  This 
corresponds to a 1.7-fold and 4.2-fold decrease in kanosamine titers relative to 
Ptrc
 expression 
under glucose-rich and glucose-limited conditions, respectively.   
 Titrating IPTG concentration resulted in improved kanosamine production using 
PT7 gene 
expression in 
E. coli
 RB791serA(DE3)/pKM10.82A (Table 19).  Repeated IPTG additions every 
6 h throughout a fermentation run resulted in better yields and higher titers relative to 
fermentations which involved a single addition of IPTG at the beginning of exponential growth 
phase (12 h after inoculation).  Decreasing IPTG concentration from 12 mg (0.05 mM) to either 
6 mg (0.025 mM) or 2.4 mg (0.01 mM) resulted in improved titers and slightly improved yields 
of kanosamine.  It appeared that 6 mg IPTG added every 6 h resulted in marginally improved 
kanosamine production as compared to 2.4 mg IPTG addition, however, as these were singleton 
experiments, the difference may not be statistically significant.  Even with the improvements 
resulting from decreasing IPTG concentration from 12 mg to 6 mg, the T7 expression system, 
RB791serA(DE3)/pKM10.82A, did not achieve the same titers and kanosamine yields as when 
the 
B. subtilis ntd
 genes were under control of 
Ptrc
 in either RB791
serA/pKM10.220ABC or 
RB791serA/pSN1.292.   
%125Table 19.
  Kanosamine production in fed-batch fermentation of 
E. coli
 RB791serA(DE3)/pKM10.82A.
  The requirement for each of the 
B. subtilis ntd
 genes for kanosamine production in 
E. coli
 was evaluated by fed-batch fermentation of RB791
serA with one of three different plasmids.  
Each plasmid contained two of the three kanosamine biosynthesis 
ntd
 genes and was constructed 
as described above using the BglBrick assembly standard.  Plasmid pKM10.214CA was 
constructed by inserting 
ntdC
 from plasmid pSN1.294 upstream of 
ntdA
 in plasmid 
pKM10.204A (Figure 65).  Insertion of 
ntdA
 from plasmid pSN1.296 upstream of 
ntdB
 in 
plasmid pSN10.204B afforded plasmid pKM10.214AB (Figure 66).  Construction of plasmid 
pKM10.214CB followed insertion of 
ntdC
 from pSN1.294 upstream of 
ntdB
 in plasmid 
pSN10.204B (Figure 66). 
Entry
Glucose 
conditions
Amount IPTG
Addition 
frequency
[Kanosamine]  
(g/L)
Yield   
(mol/mol)
1Rich
12 mg  [0.05mM]
Every 6 h
7.4 ± 0.6
2.6 ± 0.3%
2Limited12 mg  [0.05mM]
Every 6 h
2.0 ± 0.3
1.4 ±  0.2%
3Rich
6 mg  [0.025mM]
Every 6 h
10.54.2%4Rich
6 mg  [0.025mM]
Once at 12 h
1.20.6%5Rich
2.4 mg  [0.01mM]
Every 6 h
9.13.5%6Rich
2.4 mg  [0.01mM]
Once at 12 h
6.22.1%%126% Figure 65.
  Construction of plasmid pKM10.214CA.
 1.0 kB
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IntdCAmpRlacI
qp15Aori
Ptrc
ntdCpKM10.214CA
7.2 kb
EcoRIBam
HIserA
ntdAEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
serA
pKM10.204A
6.1 kb
ntdApSN1.294
Bgl
II%127% Figure 66.
  Construction of plasmids pKM10.214AB and pKM10.214CB.
 1.3 kb
1.0 kb
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IntdCAmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
serA
ntdCpKM10.214AB
7.0 kb
pKM10.214CB
6.7 kb
ntdAEcoRIBam
HIEcoRIBam
HIntdAserA
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ntdBpKM10.204B
5.7 kb
serA
pSN1.296
pSN1.295
ntdBntdBBgl
IIBgl
II%128 Results from fermentations of RB791
serA constructs transformed with pKM10.214CA, 
pKM10.214AB, or pKM10.214CB are summarized in Table 20.  Plasmid pKM10.214CA did not 
carry the 
ntdB
-encoded kanosamine-6-phosphate phosphatase.  Fed-batch fermentation of 
RB791serA/pKM10.214CA revealed whether a native 
E. coli
 phosphatase had sufficient 
promiscuity to accumulate kanosamine in fermentation broth.  Kanosamine concentrations did 
not exceed 0.11 ± 0.03 g/L, corresponding to a 0.04% mol/mol yield from 
D-glucose.  While the 
small amount of kanosamine observed  indicates that RB791
serA does possess some native 
promiscuous phosphatase activity, the titers and yields are far from those achieved with any of 
the previously described constructs expressing a dedicated kanosamine-6-phosphate phosphatase 
encoded by 
ntdB
.   Table 20.
  Kanosamine production with incomplete constructs. 
Error represents one standard deviation away from the mean across duplicate runs.  
 Promiscuous 
D-glucose-6-phosphate 3-dehydrogenase activity in 
E. coli
 was examined in 
fed-batch fermentation of RB791
serA/pKM10.214AB while promiscuous aminotransferase 
activity was evaluated in fermentations of RB791
serA/pKM10.214CB.  Neither of these 
constructs resulted in any observable amounts of kanosamine in either glucose-rich or glucose-
Entry
Plasmid
[Kanosamine] 
(g/L)
Yield   
(mol/mol)
1pKM10.214CA
0.11 ± 0.03
0.04 ± 0.01%
2pKM10.214AB
00%3pKM10.214CB
00%%129limited fermentations.  These results indicate that kanosamine biosynthesis in 
E. coli
 requires 
expression of all three 
B. subtilis 
enzymes. 
 Regardless of the 
E. coli
 strain used, glucose-rich fermentations consistently resulted in 
higher final cell-densities and greater byproduct accumulation than glucose-limited 
fermentations.  Figure 67 and  Figure 68 compare biomass and byproduct formation between 
glucose-rich and glucose-limited fermentations of RB791
serA/pSN1.292 using 
Ptrc
 expression of 
B. subtilis ntd
 genes.  Approximately 15-20 g/L more biomass was generated in glucose-rich 
fermentations than glucose-limited conditions while the most abundant byproduct observed was 
acetic acid.  Pyruvic acid and lactic acid were also observed in glucose-rich fermentations. 
% Figure 67.  
Byproducts formed during fermentation of 
E. coli
 RB791
serA/pSN1.292 under glucose-rich conditions.  
Legend: acetic acid, orange columns; formic acid, grey columns; pyruvic acid, red columns; lactic acid, blue 
columns; 
D-glucosamine, purple columns; dry cell weight, black circles. 
Dry cell weight  (g/L)01020304050607080Byproducts  (g/L)
01234567Time  (h)12182430364248%130% Figure 68.  
Byproducts formed during fermentation of 
E. coli
 RB791
serA/pSN1.292 under glucose-limited 
conditions.  Legend: acetic acid, orange columns; formic acid, grey columns; dry cell weight, black circles.
  Glucose-rich fermentations of RB791s
erA/pKM10.220ABC and RB791
serA/pSN1.292 
resulted in formation of 
D-glucosamine alongside the organic acid byproducts.  Resonances for 
D-glucosamine became more clear upon partial purification of kanosamine-containing 
fermentation broth (Figure 69) by cation exchange chromatography.  A mass spectrometry 
analysis conducted by another group member revealed a single molecular ion peak at 180 m/z, 
corresponding to the molecular mass of both kanosamine and 
D-glucosamine.   Though complete 
separation of 
D-glucosamine and kanosamine could not be achieved by ion exchange 
chromatography, partial purification of kanosamine-containing fermentation broth by cation 
exchange chromatography (Dowex 50WX8, H
+-form) afforded a sample with which the identity 
of D-glucosamine could be confirmed.   The partially purified sample containing a mixture of the 
two aminosugars was analyzed by 
1H and 
13C NMR and then spiked with >98% pure 
D-glucosamine hydrochloride (Alfa Aesar) (Figure 69).  Prior to this experiment, a calibration 
curve was generated using sodium (trimethylsilyl)-
2,2,3,3-d4-propionate (TSP, 
-
 = 0.00 ppm) as 
Dry cell weight  (g/L)010203040506070Byproducts  (g/L)
01234567Time  (h)12182430364248%131an internal standard.  Comparing the relative integrations of the signal corresponding to TSP and 
the 
#
-anomeric proton peak of 
D-glucosamine (5.42 ppm, 0.6 H, 
J = 3.6 Hz) in samples 
containing 
D-glucosamine hydrochloride across a range of concentrations (0.5-100 mM), a 
calibration curve was calculated: [mM sample] = 1.013(1H 
) mM TSP) + 0.16 (
r2 = 0.99).    This calibration curve was used to determine the concentration of 
D-glucosamine in a 
partially purified sample of RB791
serA/pSN1.292 fermentation broth by 
1H NMR (19.6 mM, 
Òwithout 
D-glucosamine spike,Ó Fig. 69).  An aliquot (21 
$
L) of 0.47 mM 
D-glucosamine in D
2O was added to the NMR sample in order to increase the amount of 
D-glucosamine by 10 mM.  The 
concentration of 
D-glucosamine in this spiked sample was also determined by 
1H NMR (31.4 
mM).  Thus, the observed increase peak size corresponding to the 
#
-anomer was observed to be 
proportional to the amount of 
D-glucosamine added to the sample.  No new peaks were observed.  
Increasing the amount of added 
D-glucosamine to 300 mM allowed previously masked peaks to 
become apparent (Figure 69).  Furthermore, 
13C NMR analysis of the unspiked sample and 
sample with 300 mM 
D-glucosamine added did not reveal the presence of new peaks when 
D-glucosamine was added (Figure 70).  These results provide strong evidence that 
D-glucosamine 
is produced alongside kanosamine in glucose-rich fermentations. 
%132% Figure 69. 
 Stacked 
1H NMR spectra of partially purified kanosamine with increasing amounts of added 
D-glucosamine. 
% Figure 70. 
 Stacked 
13C NMR spectra corresponding to partially purified kanosamine with and without added 
D-glucosamine. 
300 mMD-glucosamine
spike10 mMD-glucosamine
spikewithout
D-glucosamine
spike300 mMD-glucosamine
spikewithout
D-glucosamine
spike%133 The lack of 
D-glucosamine formation in glucose-limited fermentations of RB791
serA/pSN1.292 and RB791
serA/pKM10.220ABC implies that 
D-glucosamine is not formed as a result 
of promiscuous isomerase or aminotransferase activity in 
E. coli
 but is more likely due to 
D-glucose availability.  Evidence from fermentations of RB791
serA pgi::
KanR in which 
kanosamine is produced without 
D-glucosamine accumulation under glucose-rich conditions 
with either a 
D-glucose or 
D-glucose/
D-fructose mixed feed (described in a later section) offers 
further support to this hypothesis. 
 D-Glucosamine is produced in 
E. coli
 upon hydrolysis of 
D-glucosamine-6-phosphate, in 
turn formed from 
#
-D-fructose-6-phosphate by action of 
glmS-
encoded 
L-glutamine:
D-fructose-6-phosphate aminotransferase (Figure 71).  The smaller relative amount of 
D-glucosamine produced during glucose-rich (0.33 ± 0.2 g/L) fermentations of RB791
serA/pSN1.292 compared to other glycolytic byproducts such as acetic acid (5.0 ± 2.3 g/L), likely 
reflects competition between GlmS and the 6-phosphofructokinase enzymes, PfkA and PfkB, 
which convert 
#
-D-fructose-6-phosphate into 
#
-D-fructose-1,6-bisphosphate for further 
degradation through glycolysis.  Additional competition results from expression of 
B. subtilis
 kanosamine biosynthesis genes which compete with glycolysis for 
D-glucose-6-phosphate. 
Blocking the 
D-glucosamine biosynthesis pathway through a mutation in 
glmS
 may help reduce 
byproduct formation during kanosamine biosynthesis in 
E. coli
, however increased kanosamine 
production may more effectively come from increasing 
D-glucose-6-phosphate availability.   
%134% Figure 71.
  D-Glucosamine synthesis from 
D-glucose-6-phosphate through transamination of 
D-fructose-6-phosphate 
catalyzed by 
glmS-
encoded L-glutamine:
D-fructose-6-phosphate aminotransferase followed by phosphate ester 
hydrolysis.  GlmS must compete with phosphofructokinase isozymes, PfkA and PfkB, for substrate availability. 
2.  Competition with glycolysis for 
D-glucose-6-phosphate availability
  2.1.  Activity of 
D-glucose-6-phosphate 3-dehydrogenase 
 During fermentative metabolism in 
E. coli
, D-glucose is concomitantly transported and 
phosphorylated by the phosphoenolpyruvate-dependent phosphotransferase system. The resulting 
D-glucose-6-phosphate is metabolized via the Embden-Meyerhof-Parnas pathway and the 
OHOHOHHOOH2O3POD-glucose-6-phosphate
Pgi
!-D-fructose-6-phosphate
H2O3POOHOHOOHOHATP
ADP
!-D-fructose-1,6-bisphosphate
H2O3POOHOHOOHOPO
3H2OOHOpyruvic acid
NH2OH2NOOHNH2OHOOOHOHNH2OHHOOH2O3POD-glucosamine-
6-phosphate
H2OHOPO
32-OHNH2OHHOOHOD-glucosamine
PfkA
PfkB
GlmS
NtdCNtdANtdBOHOHNH2HOOHOkanosamine
*%135pentose phosphate pathway, affording ATP, reducing equivalents, and organic compounds for 
biosynthesis.  Under oxygen-limiting conditions, 
E. coli
 converts most of the 
D-glucose-6-
phosphate to a mixture of organic acids and carbon dioxide, which are largely generated from 
pyruvate.
42  The observation of several products of fermentative metabolism (i.e., acetic acid, 
formic acid, and lactic acid) alongside pyruvic acid in glucose-rich fermentations with various 
RB791serA 
constructs suggests that the glycolysis pathways compete for 
D-glucose-6-phosphate 
with kanosamine biosynthesis.  This competition results in low yields of kanosamine from 
D-glucose (e.g., 4.4 ± 0.4% mol/mol in glucose-rich fermentation of RB791
serA/pSN1.292).  
Kanosamine yields are slightly improved in glucose-limited conditions  
which afford reduced accumulation of fermentative byproducts and lower final cell-densities 
relative to glucose rich conditions (Fig. 67, Fig. 68).  Glucose-limited fermentations of 
RB791serA/pSN1.292 achieve 6.2 ± 2.2% mol/mol yields of kanosamine from 
D-glucose.    
 Kanosamine biosynthesis and glycolysis both utilize 
D-glucose-6-phosphate as a 
substrate.  In order for kanosamine production to compete with glycolysis pathways for 
D-glucose-6-phosphate availability in fed-batch fermentation of RB791
serA 
constructs, 
E. coli
 must be able to express active Ntd enzymes.  Cell samples withdrawn from glucose-limited 
fermentation of RB791
serA/pSN1.292 every 12 h throughout the course of the fermentation were 
lysed and assayed for 
ntdC
-encoded 
D-glucose-6-phoshpate 3-dehydrogenase activity. 
%136Table 21.  
D-glucose-6-phosphate 3-dehydrogenase specific activity 
Fed-batch fermentation samples from glucose-limited run with 
E. coli 
RB791serA 
/pSN1.292. 
 The activity of 
ntdC
-encoded 
D-glucose-6-phosphate 3-dehydrogenase was assayed in 
cell-free extracts of RB791
serA/pSN1.292 during incubation with NAD
+ and 
 D-glucose-6-
phosphate in 100 mM 2-amino-2-methyl-1-propanol (AMP) buffer, pH 9.5.  The rate of the 
NtdC-catalyzed dehydrogenation has been previously reported
27 to increase up to pH 9.5 before 
sharply falling at pH 
.
 10.  The absorbance at 340 nm was monitored for 5 min at ambient 
temperature and NtdC specific activity was calculated from the increase in absorbance due to 
formation of NADH (Table 21).  Because enzyme activity is determined by monitoring cofactor 
reduction, which is not enzyme-specific, the small amount of observed enzyme activity at 12 h 
before the first aliquot of IPTG had been added is likely due background activity by one or more 
native 
E. coli 
enzymes present in the cell-free extract.  After induction, 
ntdC
-encoded 
D-glucose-6-phosphate 3-dehydrogenase activity increased to 0.27 
$
mol/min/mg (U/mg) and 
remained steady throughout the course of the fermentation. 
 Specific activities are reported as a function of amount of total protein (mg
*1) and are 
therefore a reflection of enzyme purity.  Activities around 0.3 U/mg match those reported for 
other 
D-glucose-6-phosphate dehydrogenases assayed in cell-free extracts, such as the wild-type 
Entry
Time point
Specific activity   
(#
mol min
$1 mg$1)112 h0.003224 h0.27336 h0.34448 h0.33%137D-glucose-6-phosphate 1-dehydrogenase from 
Pseudomonas fluorecens
 (0.31 U/mg).
43  These 
assay results imply good 
in vivo
 activity of NtdC following induction of transcription with IPTG, 
which remains stable during a fermentation.   
2.2.  Improving product yields by blocking glycolysis pathway 
 One possible strategy to improve kanosamine production from expression of 
B. subtilis
 Ntd enzymes would be to block a competing pathway.  The presence of acetic acid, lactic acid, 
formic acid, and pyruvic acid in fermentation broth as major byproducts indicate that glycolysis 
pathways compete with kanosamine biosynthesis for 
D-glucose-6-phosphate availability.  Prior 
research has demonstrated that 
D-glucose fermentation of an 
E. coli
 mutant with an inactive 
pgi-encoded phosphoglucose isomerase results in increased intracellular availability of 
D-glucose-6-
phosphate relative to wild-type strains.
44  Deletions in 
pgi have previously been used to achieve 
increased titers of compounds such as 2-deoxy-
scyllo-inosose
45 from 
D-glucose. 
 To evaluate whether a 
pgi-deficient mutant would lead to greater kanosamine 
accumulation, pSN1.292 was transformed into RB791
serA pgi
::KanR, a strain previously 
engineered by Amy Dean in the Frost group.
46  The resulting strain was evaluated in fed-batch 
fermentations under glucose-rich conditions.  The phosphoglucose isomerase mutation resulted 
in unsustained growth when utilizing 
D-glucose as the sole carbon source and produced minor 
amounts of kanosamine (Figure 72).  Loss of an active phosphoglucose isomerase
 significantly 
alters the structure of the metabolic network in 
E. coli
 by disabling the upper pathway of 
glycolysis.
47  Prior research has shown this reduces the growth rate to <20% of wild-type levels 
in minimal media.  
E. coli 
remains viable because 
D-glucose metabolism is rerouted through the 
%138pentose phosphate pathway.  This introduces a redox imbalance as excess NADPH is produced 
which contributes to the growth characteristics of 
pgi mutants.
47   % Figure 72.
  Kanosamine production by 
E. coli 
RB791serA pgi
::KanR/pSN1.292 under glucose-rich conditions with a 
D-glucose-only feed.  Legend: kanosamine, green columns; dry cell weight, black circles.  Error bars represent one 
standard deviation away from the mean across duplicate runs. 
% Figure 73.
  Byproducts formed in fed-batch fermentation of 
E. coli 
RB791serA pgi
::KanR/pSN1.292 under glucose-
rich conditions with a 
D-glucose-only feed.  Legend: acetic acid, orange columns; 
L-glutamic acid, bright green 
columns; dry cell weight, black circles.  Error bars represent one standard deviation away from the mean across 
duplicate runs. 
Dry cell weight  (g/L)01020304050607080Kanosamine  (g/L)0246810121416Time  (h)12182430364248Dry cell weight  (g/L)01020304050607080Byproducts  (g/L)
01234567Time  (h)12182430364248%139% Figure 74.
  Kanosamine production by 
E. coli 
RB791serA pgi
::KanR/pSN1.292 under glucose-rich conditions with a 
mixed feed of 
D-glucose/
D-fructose (1:1 mol/mol).  Legend: kanosamine, green columns; dry cell weight, black 
circles.  Error bars represent one standard deviation away from the mean across duplicate runs. 
% Figure 75.
  Byproducts formed in fed-batch fermentation of 
E. coli 
RB791serA pgi
::KanR/pSN1.292 under glucose-
rich conditions with a 
D-glucose-only feed.  Legend: acetic acid, orange columns; formic acid, grey columns; 
L-glutamic acid, bright green columns; formic acid, grey columns; dry cell weight, black circles.  Error bars represent 
one standard deviation away from the mean across duplicate runs.
  Fermentation of RB791
serA pgi
::KanR/pSN1.292 on a 
D-glucose-only feed resulted in the 
highest yields of kanosamine observed thus far.  Kanosamine titers reached 3.6 ± 0.2 g/L in an 18 
Dry cell weight  (g/L)01020304050607080Kanosamine  (g/L)0246810121416Time  (h)12182430364248Dry cell weight  (g/L)01020304050607080Byproducts  (g/L)
01234567Time  (h)12182430364248%140± 1% mol/mol yield from 
D-glucose (Figure 72).  These fermentations resulted in a 14.7-fold 
decrease in biomass formation (2.9 ± 1.4 g/L dry cell weight) relative to RB791
serA/pSN1.292 
with wild-type phosphoglucose isomerase in glucose-rich fermentations (42.8 ± 1.7 g/L dry cell 
weight).  Furthermore, only 20 g of 
D-glucose was consumed during these runs as opposed to 
296 g in glucose-rich fermentations of RB791
serA/pSN1.292.  Though growth was inhibited by 
the mutation in phosphoglucoisomerase, kanosamine yield from 
D-glucose implies that more 
D-glucose-6-phosphate was availabile to the kanosamine biosynthesis enzymes relative to 
fermentations of RB791
serA/pSN1.292 or RB791
serA/pKM10.220ABC.   
 Cell growth was restored and 
D-glucose consumption increased with the use of a feed 
consisting of equimolar amounts of 
D-glucose and 
D-fructose.  Upon transport into the cell, 
D-fructose is phosphorylated to 
D-fructose-6-phosphate, which is catabolized through the Embden-
Meyerhoff-Parnas pathway of glycolysis.  Thus, a feed containing 
D-fructose was used to 
supplement levels of 
D-fructose-6-phosphate that would have otherwise been supplied by 
phosphoglucose isomerase-catalyzed isomerization of 
D-glucose-6-phosphate to 
D-fructose-6-
phosphate.  Final cell densities reached 37 ± 2.4 g/L dry cell weight with 150 g 
D-glucose 
consumed.  Kanosamine titers reached 8.7 ± 0.08 g/L in 5.9 ± 0.6% mol/mol yield (Figure 74), 
enabled by the increased 
D-glucose consumption relative to RB791
serA pgi
::KanR/pSN1.292 
grown on a 
D-glucose-only feed.  Though the observed titers are lower than glucose-rich 
fermentations of RB791
serA/pSN1.292 (12.7 ± 0.6 g/L, 4.4 ± 0.4% mol/mol yield), the overall 
yield from 
D-glucose is higher.  Fermentations of RB791
serA pgi
::KanR/pSN1.292 also resulted 
in reduced formation of the acetic acid and pyruvic acid relative to RB791
serA/pSN1.292 
fermentations (Figure 73 and Figure 75).  Acetic acid concentration decreased from 5.0 ± 2.5 g/L 
%141to 1.7 ± 0.4 g/L and the concentration of pyruvic acid decreased from 0.7 ± 0.2 g/L to 0.05 ± 0.02 
g/L.  These results indicate that blocking glycolysis may be an effective strategy to increase 
kanosamine production, but strain or feed conditions must be optimized to balance kanosamine 
production with cell viability.
   An unexpected observation during these fermentations was the appearance of 
L-glutamic 
acid in the fermentation broth.  Both the 
D-glucose-only feed and the mixed 
D-glucose/
D-fructose 
feed led to 
L-glutamic acid accumulation with 2.9 ± 0.7 g/L 
L-glutamic acid produced with a 
D-glucose-only feed (Fig. 73) and 3.9 ± 0.9 g/L 
L-glutamic acid produced during fermentations of a 
1:1 
D-glucose:
D-fructose feed
 (Fig. 75).  Accumulation of 
L-glutamic acid is likely a result of a 
metabolic network imbalance due to the 
pgi mutation.  Previous reports have demonstrated that 
the glyoxylate shunt of the Krebs cycle (Figure 76) is more active in 
+pgi mutant strains of 
E. 
coli
 than wild type.
47  Routing flux through the glyoxylate shunt rather than the full Krebs cycle 
avoids exacerbating the redox imbalance caused by increased carbon flux through the pentose 
phosphate pathway, as NADH biosynthesis normally arising from conversion of 2-ketoglutarate 
to succinyl-CoA would not occur.
47  This increased flux through the glyoxylate shunt may result 
in an increased intracellular concentration of 2-ketoglutarate, which is converted to 
L-glutamic 
acid by 
gdhA-encoded glutamate dehydrogenase (Figure 77).  It is possible that the kanosamine 
production observed in fermentations of RB791
serA pgi
::KanR/pSN1.292 are at least partially 
due to increased intracellular 
L-glutamate levels as more substrate would be available for the 
transamination of 3-keto-
D-glucose-6-phosphate to kanosamine-6-phosphate. 
%142% Figure 76.
  The Krebs cycle indicating the glyoxylate shunt from 
D-threo
-isocitrate to glyoxylate. 
CoAOOOHOOHOOHOOHOOHOHOOHOOHOOHOOCoAOHOOOHOOOHOOHOHOOHOHOHOOHOOOHHOOOHOOOHOOHOOOHHoxaloacetate
malate
fumarate
succinate
succinyl-CoA
2-ketoglutarate
D-threo
-isocitrate
cis
-aconitate
citrate
acetyl-CoA
glyoxylate
NAD
+NADH
CO2glyoxylate
shuntCoACO2CoA%143% Figure 77.
  Glutamate dehydrogenase-catalyzed transamination of 2-ketoglutarate to 
L-glutamic acid. 
 Increased intracellular 
L-glutamic acid may explain the lack of 
D-glucosamine production 
in the fermentation broth in fermentations of RB791
serA pgi
::KanR/pSN1.292.  Transamination 
of #
-D-fructose-6-phosphate by action of 
glmS-
encoded 
L-glutamine:
D-fructose-6-phosphate 
aminotransferase affords 
L-glutamate as a byproduct.  Increased 
L-glutamate levels relative to 
wild type concentrations would drive the equilibrium position of the 
glmS
-encoded 
transamination (Figure 71) in the reverse direction, resulting in reduced 
D-glucosamine 
production.  This provides additional support to the hypothesis that 
D-glucosamine is not derived 
directly from kanosamine but rather forms as a result of carbon flux through glycolysis and 
action of native 
E. coli
 enzymes such as GlmS.
  The observation of 
D-fructose remaining in the fermentation broth after 48 h while 
D-glucose was not detected indicates that the two carbohydrates were not metabolized at equal 
rates.  Carbon catabolite repression is well documented and often observed during fermentations 
of mixed carbohydrate feeds.
48  Because 
D-glucose is the preferred carbon source for 
E. coli
, it is 
preferentially metabolized in the presence of alternative carbon sources.  A mutation in 
zwf
-encoded 
D-glucose-6-phosphate 1-dehydrogenase in addition to a mutation in 
pgi-encoded 
phosphoglucose isomerase has been shown to block carbon catabolite repression during 
fermentation of 
E. coli
 on a 
D-glucose/
D-xylose mixed feed.  Cells were shown to grow 
OHONH2OHOOHOOOHONH4+NADPH
H2ONADP
+GdhA%144exclusively on 
D-xylose while utilizing 
D-glucose for biosynthesis of 
D-glucaric acid.
48  An 
RB791serA +pgi 
+zwf 
double-mutant may result in an additional increase in kanosamine yield 
from 
D-glucose while achieving the good growth and high kanosamine titers that were absent 
during fermentation of RB791
serA pgi
::KanR/pSN1.292.   
2.3.  Media supplementation 
 The increased kanosamine yields observed in fermentations of RB791
serA pgi
::KanR, especially in the case of a 
D-glucose-only feed, in conjunction with accumulation of 
L-glutamic 
acid, a substrate for 
ntdA
-encoded 3-keto-
D-glucose-6-phosphate:
L-glutamate aminotransferase, 
may suggest that increased
 L-glutamate availability may have a beneficial effect on kanosamine 
production in 
E. coli
.  Therefore, external 
L-glutamate (10 g/L) was added during glucose-limited 
runs (Table 22) of RB791
serA/pSN1.292.  Surprisingly, these fermentations resulted in greatly 
reduced 
D-glucose consumption relative to glucose-limited fermentation of RB791
serA/pSN1.292 in the absence of additional supplements (63 g relative to 145 ± 17.9 g).  The decrease 
in D-glucose metabolism led to lower titer and yield of kanosamine (2.3 g/L, 3.7% mol/mol yield 
relative to 8.5 ± 1.4 g/L, 6.2 ± 2.2% mol/mol yield). 
%145Table 22.
  Nitrogen supplementation during fed-batch fermentation of 
E. coli
 RB791serA/pSN1.292
 aNitrogen supplements added prior to inoculation at 10 g/L final concentration.  Error represents one standard 
deviation away from the mean across triplicate (glucose-rich) or quadruplicate (glucose-limited) runs.
  Previous reports indicate that single amino acids such as 
L-proline, 
L-arginine, and 
L-glutamate are poor nitrogen sources.
49  Saturating levels of one of these amino acids lead to an 
increase in intracellular levels of 
!
-ketoglutarate.  Nitrogen and carbon assimilation pathways are 
linked at a branch point in which 
!
-ketoglutarate serves as a regulatory molecule.  Under 
nitrogen-limiting conditions, carbon flux is directed out of the citric acid cycle and into amino 
acid biosynthesis, affording a rise in 
!
-ketoglutarate levels.  Increased concentrations of 
!
-ketoglutarate create a feedback signal for carbon assimilation at the level of adenylate cyclase 
which catalyzes synthesis of cAMP.  As a regulatory molecule involved in 
D-glucose transport 
into the cell, reduced levels of cAMP lead to repression of 
D-glucose uptake.
49  Previous research 
by Jintao Guo in the Frost group demonstrated that soy flour supplementation at 10 g/L increased 
the amount of kanosamine produced in fed-batch fermentations of 
Bacillus pumilus 
by four- 
fold.
50  We then tested soy flour as a complex media supplement and source of extra nitrogen at 
10 g/L under both glucose-rich and glucose-limited fermentations of RB791
serA/pSN1.292 
(Table 22).  Under these conditions, 
D-glucose consumption was restored to levels matching 
Entry
Glucose 
conditions
Glucose 
consumed  (g)
Nitrogen 
supplementation
a[Kanosamine]   
(g/L)
Yield 
(mol/mol)
1limited145 ± 17.9
none8.5 ± 1.4
6.2 ± 2.2%
2rich
296 ± 6.9
none12.7 ± 0.6
4.4 ± 0.4%
3limited63L-glutamate
2.33.1%4limited155soy flour
2.81.6%5rich
300soy flour
9.95.3%%146those of RB791
serA/pSN1.292 without additional supplementation.  However, kanosamine 
production under glucose-rich conditions (9.9 g/L, 5.3% mol/mol yield from 
D-glucose) was not 
enhanced beyond levels reached without an external nitrogen source (Table 22), suggesting that 
nitrogen is not the limiting factor in kanosamine production.   Though exogenous 
L-glutamate 
supplementation did not lead to an enhancement of kanosamine biosynthesis, it is possible that 
an engineered 
E. coli
 with overexpressed 
L-glutamate biosynthesis enzymes could be more 
effective at improving kanosamine production than external media additives. 
3.  Kanosamine isolation from fermentation broth and inhibitory affects on RB791
serA  Kanosamine displays antibiotic activity against a variety of different organisms with 
varying effectiveness (Table 17).  To determine whether it is toxic to RB791
serA, kanosamine 
was isolated from fermentation broth and added to shake-flask cultures of RB791
serA/pSN1.292 
in minimal media.  Kanosamine was isolated as the hydrochloride salt using a procedure 
involving activated charcoal treatment, ion-exchange chromatography, and selective 
solubilization of kanosamine to separate contaminating salts.  Cell-free, protein-free fermentation 
broth was treated with activated charcoal (10 g/L) at 37¡C for 1 h and filtered through a layer of 
Celite 545.  The clear, colorless filtrate was loaded onto a cation exchange column (Dowex 
50WX8, H
+-form) and washed with four column-volumes of water.  Kanosamine hydrochloride 
was eluted with a linear gradient of HCl (0-0.5 N).  Fractions containing kanosamine were 
pooled and brought to pH 7.0 with 5 N KOH.  Water was removed under reduced pressure and 
the pale yellow solids were dissolved into methanol.  Insoluble salts were removed by filtration.  
The filtrate was evaporated under reduced pressure to afford kanosamine hydrochloride as an 
%147off-white solid.  Isolated material was characterized by 
1H, 
13C and two-dimensional NMR 
(Figures 78-80).    
% Figure 78.
  1H NMR spectrum of kanosamine isolated from fermentation broth. 
4321O56HOH2NOHOHHO4321O56HOH2NOHOHHO!-D-kanosamine
"-D-kanosamine
%148% Figure 79.
  13C NMR spectrum of kanosamine isolated from fermentation broth. 
4321O56HOH2NOHOHHO4321O56HOH2NOHOHHO!-D-kanosamine
"-D-kanosamine
%149% Figure 80.
  HSQC NMR spectrum of kanosamine isolated from fermentation broth. 
 To investigate potential inhibitory effects of kanosamine on RB791
serA/pSN1.292, 
kanosamine isolated from fermentation broth was added to exponentially growing cells.  A 5 mL 
culture of RB791
serA/pSN1.292 in minimal media was grown overnight at 35¡C with agitation 
at 250 rpm.  This culture was used to inoculate 20 mL of the same media to an initial optical 
density (OD
600) of 0.05.  Cells were grown at 35¡C with agitation at 250 rpm to an OD
600 = 0.4 marking the beginning of exponential growth phase, after which an aqueous solution of 
kanosamine (pH 8.0) was added to a final concentration of 25 mM (4.5 g/L) and 50 mM (9.0 g/
4321O56HOH2NOHOHHO4321O56HOH2NOHOHHO!-D-kanosamine
"-D-kanosamine
%150L).  The optical density of each culture was measured at 1 h intervals during mid-exponential 
growth phase.  These values were used to calculate the number of cells per milliliter of culture  in 
order to determine growth rate of each culture (Figure 81).  Conversion of optical density at 600 
nm (OD
600) to number of cells per milliliter utilized the following conversion formula:  
OD600(1.0) = 1)108 cells/mL.  The logarithm of the cellular concentration was plotted and the 
growth rates for each culture compared based on the slopes of the log curves.  Exogenously 
added kanosamine did not inhibit growth of RB791
serA/pSN1.292 at the concentrations tested, 
evidenced by growth rates similar to a culture of RB791
serA/pSN1.292 without added 
kanosamine.  This is supported by previous results reporting poor antibiotic activity of 
kanosamine on 
E. coli
 strains.   
% Figure 81.
  Exponential phase growth of RB791
serA/pSN1.292 with exogenously added kanosamine at 0 mM 
(blue), 25 mM (green), and 50 mM (orange). 
log(cells/mL)67891011Time  (h)036912%151PART II   
KANOSAMINE BIOSYNTHESIS FROM UDP-D-GLUCOSE 
1.  Overview 
 Two pathways of kanosamine biosynthesis have been reported in the literature (Figure 
82).  The pathway beginning with UDP-
D-glucose (Pathway B, Fig. 82) has been implicated in 
kanosamine production both 
Bacillus cereus
28 and 
Bacillus pumilus
,29 as well as organisms such 
as Amycolatopsis mediterranei
.33,34  In A. mediterranei
, kanosamine is a biosynthetic precursor 
to 3-amino-5-hydroxybenzoic acid (AHBA) via the aminoshikimate pathway.  The key enzymes 
in this pathway responsible for kanosamine biosynthesis, 
rifL-encoded UDP-
D-glucose 3-
dehydrogenase, 
rifK-encoded 3-keto-UDP-
D-glucose:
L-glutamine aminotransferase, and 
rifM-encoded UDP-kanosamine phosphatase, have been assigned largely based on circumstantial 
evidence and sequence homology.
51  For instance, RifK, in addition to its well established role as 
AHBA synthase, has been assigned a second role in the aminoshikimate pathway as a 3-keto-
UDP-D-glucose:
L-glutamine aminotransferase.  This assignment was based on structural 
similarities between RifK and the family of aspartate transaminases.  The roles of RifL and RifM 
were assigned based on structural homology to known oxidoreductases and phosphatases,  
respectively.
51  %152A% B! Figure 82.  
Kanosamine biosynthesis pathways reported in the literature.  Pathway A described in 
Bacillus subtilis
; Pathway B described in 
B. pumilus
, B. cereus
, and 
A. mediterranei.  
 2.  Kanosamine production in fed-batch fermentation of 
E. coli 
constructs carrying 
Amycolatopsis mediterranei 
kanosamine biosynthetic genes 
   Kanosamine production resulting from expression of 
Amycolatopsis mediterranei
 gene 
expression in 
E. coli
 was evaluated in fed-batch fermentations following glucose-rich and 
glucose-limited conditions as described above.  Because of the phylogenetic gap between 
A. 
mediterranei
 and 
E. coli
 and the difference in genomic G+C content, codon-optimized genes 
were utilized.  Gene sequences encoding 
rifL, rifK, and 
rifM from 
Amycolatopsis mediterranei
 S699 were optimized for expression in 
E. coli
 using the web service OPTIMIZER¨.
52  Optimized genes were synthesized by Integrated DNA Technologies, Inc. (IDT, Coralville, IA), 
OHOHOOHOHH2O3POOHOOOHOHH2O3POOHOH2NOHOHH2O3POOHOH2NOHOHHONAD
+NADH
NtdCD-glucose
6-phosphate
3-keto-
D-glucose
6-phosphate
kanosamine
6-phosphate
kanosamine
NtdANH2OHOOOHOOHOOOHH2OHOPO
32-NtdBOHOHOOHOUDPHOOHOH2NOHOHHONAD
+NADH
RifLUDP-
D-glucose
kanosamine
RifKNH2OH2NOOHOOH2NOOHH2OUDPRifMOHOOOHOUDPHO3-keto-UDP-
D-glucose
OHOH2NOHOUDPHOUDP-
kanosamine
%153and inserted by IDT into a pUC18 vector with a 5Õ 
Bgl
II restriction site and 3Õ 
Bam
HI and 
Xho
I restriction sites.  The restriction sites flanking synthetic 
rif genes facilitated BglBrick assembly 
of two plasmids containing a 
rifLrifKrifMserA
 cassette either behind 
Ptrc
 (pKM11.9, Figure 86) 
or PT7 (pKM11.46, Figure 87).   
 Each gene was first inserted into the 
Bgl
II and 
Xho
I sites of plasmid pBbA1a to afford a 
BglBrick part (Figure 83).  Next, each 
rif gene was inserted upstream of 
serA 
by digesting the 
BglBrick part containing 
serA, pKM6.240 (Figure 84), with 
EcoRI and 
Bgl
II to create a double-
stranded break upstream of 
serA.  Digestion of the 
rifM-
containing BglBrick part, pKM10.254M 
(Figure 84), with 
EcoRI and 
Bam
HI followed by ligation of the digested DNA afforded plasmid 
pKM10.268M.  The 
rifK gene from pKM10.254K was then inserted upstream of 
rifM 
in plasmid 
pKM10.268M to afford plasmid pKM10.278KM (Figure 85).  Finally, 
rifL from plasmid 
pKM10.254L was inserted upstream of 
rifM 
in plasmid pKM10.278KM to afford plasmid 
pKM11.9 containing a 
Ptrc
rifLrifKrifMserA
 cassette (Figure 86).  Another plasmid was created 
that contained a 
PT7 cassette by inserting the 
rifLrifKrifMserA
 fragment
 from pKM11.9 into 
pBbA7k to afford pKM11.46 (Figure 87). 
%154% Figure 83.
  Construction of 
rif BglBrick parts: plasmids pKM10.254K, pKM10.254L, and pKM10.254L. 
1.2 kb
1.1 kb
0.7 kb
pUCIDT-rifK
pUCIDT-rifL
pUCIDT-rifM
Bgl
II, Xho
I digest
Bgl
II, Xho
I digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IrifL
rifM
AmpRlacI
qp15Aori
mRFP-1
pBbA1a-RFP
4.3 kb
Ptrc
AmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
rifK
rifL
rifM
pKM10.254K
4.7 kb
pKM10.254L
4.6 kb
pKM10.254M
4.3 kb
rifK
Bgl
IIXho
IBgl
IIXho
IBgl
IIXho
IBam
HIBam
HIBam
HI%155% Figure 84.
  Construction of plasmid pKM10.268M. 
Bgl
II0.7 kb
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IrifM
AmpRlacI
qp15Aori
Ptrc
rifM
pKM10.268M
5.5 kb
EcoRIBam
HIEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
serA
pKM6.240
4.8 kb
serA
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
rifM
pKM10.254M
4.3 kb
%156% Figure 85.
  Construction of plasmid pKM10.278KM. 
1.1 kb
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IrifK
AmpRlacI
qp15Aori
Ptrc
rifK
pKM10.278KM
6.6 kb
EcoRIBam
HIserA
rifM
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
rifM
pKM10.268M
5.5 kb
serA
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
rifK
pKM10.254K
4.6 kb
%157% Figure 86.
  Construction of plasmid pKM11.9. 
Bgl
II1.3 kB
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IrifL
AmpRlacI
qp15Aori
Ptrc
rifL
pKM11.9
7.8 kb
EcoRIBam
HIserA
rifK
rifM
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
rifL
pKM10.254L
4.6 kb
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
rifK
pKM10.278KM
6.6 kb
serA
rifM
%158% Figure 87.
  Construction of plasmid pKM11.46. 
 Plasmid pKM11.9 was transformed into RB791
serA and kanosamine production was 
examined in glucose-rich (Figure 88) and glucose-limited conditions (Figure 89).  Rather than 
accumulating kanosamine as was anticipated, glucose-rich conditions afforded substantial 
amounts (46 g/L) of acetic acid.  Formic acid, pyruvic acid, and lactic acid were also observed at 
4.3 kb
EcoR
I, BamH
I digest
EcoR
I, Bam
HI digest
T4 ligationrifLrifKrifMserA
EcoR
IBgl
IIBamH
IXho
IlacI
qp15Aori
mRFP-1
pBbA7k-RFP
4.3 kb
EcoR
IBamH
IEcoR
IBgl
IIBamH
IXho
IKan
RlacI
qp15Aori
rifK
pKM11.46
7.8 kb
serA
rifM
rifL
Kan
RPT7PT7pKM11.9
Bgl
II%159concentrations of 0.75 g/L, 0.35 g/L, and 0.26 g/L, respectively.  Glucose-limited fermentation of 
RB791serA/pKM11.9 (Figure 89) resulted in small amounts of acetic acid (0.09 g/L), formic 
acid (0.07 g/L), pyruvic acid (0.04 g/L), and lactic acid (0.05 g/L). 
% Figure 88.
  Products formed in fed-batch fermentation of 
E. coli 
RB791serA/pKM11.9 under glucose-rich 
conditions.  Legend: acetic acid, orange columns; formic acid, grey columns; lactic acid, blue columns; dry cell 
weight, black circles.  
% Figure 89.
  Products formed in fed-batch fermentation of 
E. coli 
RB791serA/pKM11.9 under glucose-limited 
conditions.  Legend: acetic acid, orange columns; formic acid, grey columns; dry cell weight, black circles.  
Dry cell weight  (g/L)01020304050607080Byproducts  (g/L)
01020304050Time  (h)12182430364248Dry cell weight  (g/L)01020304050607080Byproducts  (g/L)
01234567Time  (h)12182430364248%160 It was thought that the lack of kanosamine production in RB791
serA/pKM11.9 which 
contained 
rifLrifKrifMserA 
under the control of 
Ptrc
 was due to pyrimidine starvation.  Previous 
research reports 
E. coli
 K-12 wild type strains, W3110 and MG1655, suffer pyrimidine starvation 
when cultured in minimal media.
53  As RB791
serA is a derivative of W3110, uracil was added at 
20 mg/L in glucose-rich fermentation of RB791
serA/pKM11.9 and at 200 mg/L in glucose-
limited fermentation of RB791
serA(DE3)/pKM11.46, which carried the 
rif genes under control 
of PT7.  Neither set of conditions resulted in kanosamine accumulation (Figure 90 and Figure 91).   
% Figure 90.
  Products formed in fed-batch fermentation of 
E. coli 
RB791serA/pKM11.9 with 20 mg/L uracil 
supplementation under glucose-rich conditions.  Legend: acetic acid, orange columns; formic acid, grey columns;  
lactic acid, blue columns; dry cell weight, black circles.  
Dry cell weight  (g/L)01020304050607080Byproducts  (g/L)
01020304050Time  (h)12182430364248%161% Figure 91.
  Products formed in fed-batch fermentation of 
E. coli 
RB791serA(DE3)/pKM11.46 with 200 mg/L uracil 
supplementation under glucose-limited conditions.  Legend: acetic acid, orange columns; formic acid, grey columns;  
lactic acid, blue columns; dry cell weight, black circles. 
 Kanosamine production was then evaluated using 
E. coli
 KL3, which is not known to 
suffer pyrimidine starvation.  
E. coli
 KL3 is a derivative of AB2834 and contains a mutation in 
aroE
 encoding shikimate dehydrogenase.
54  Because of this mutation, it is unable to synthesize 
aromatic amino acids and aromatic vitamins.  To ensure viability in minimal media, 
L-tryptophan, 
L-phenylalanine, 
L-tyrosine, 2,3-dihydroxybenzoic acid, 4-hydroxybenzoic acid, and 
4-aminobenzoic acid were added to all culture and fermentation media when utilizing KL3.  
Transformation of plasmid pKM11.9, which contained a 
Ptrc
rifLrifKrifMserA
 cassette, into KL3 
was followed by both glucose-rich and glucose-limited fermentations (Figure 92 and Figure 93). 
Dry cell weight  (g/L)01020304050607080Byproducts  (g/L)
01020304050Time  (h)121824303642%162% Figure 92.
  Products formed in fed-batch fermentation of 
E. coli 
KL3/pKM11.9 glucose-rich conditions.  Legend: 
acetic acid, orange columns; formic acid, grey columns;  lactic acid, blue columns; gallic acid, brown columns; 3-
dehydroquinic acid, pink columns; 3-dehydroshikimic acid, teal columns; dry cell weight, black circles. 
% Figure 93.
  Products formed in fed-batch fermentation of 
E. coli 
KL3/pKM11.9 glucose-limited conditions.  
Legend: acetic acid, orange columns; formic acid, grey columns;  lactic acid, blue columns; gallic acid, brown 
columns; 3-dehydroquinic acid, pink columns; 3-dehydroshikimic acid, teal columns; dry cell weight, black circles. 
  As was observed in fermentations of RB791
serA/pKM11.9 and RB791
serA(DE3)/
pKM11.46, kanosamine did not accumulate in fermentation broth.  Acetic acid production was 
Dry cell weight  (g/L)01020304050607080Byproducts  (g/L)
0246810Time  (h)12182430364248Dry cell weight  (g/L)01020304050607080Byproducts  (g/L)
0246810Time  (h)12182430364248%163substantially lower with KL3/pKM11.9 (2.4 g/L and 0.03 g/L under glucose-rich and glucose-
limited conditions, respectively) than with RB791
serA/pKM11.9 either in the presence (33.9 g/
L) or absence (46.3 g/L) of exogenously added uracil.  The major product observed in 
fermentations of KL3/pKM11.9 was 3-dehydroshikimic acid which accumulated in both glucose-
rich and glucose-limited fermentations at a concentration of 7.1 g/L.  3-Dehydroquinic acid and 
gallic acid were also observed in both glucose-rich and glucose-limited conditions.  
Accumulation of shikimate pathway metabolites is due to the mutation in 
aroE
-encoded 
shikimate dehydrogenase. 
3.  Activity of UDP-
D-glucose 3-dehydrogenase
  A possible reason for lack of kanosamine production in fermentations of 
E. coli
 transformed with codon-optimized 
rif genes could be due to an inability of 
E. coli
 to express the 
enzymes in active forms.  Cell-free extracts of samples withdrawn from fermentation of 
RB791serA/pKM11.9 and His
6-tagged UDP-
D-glucose 3-dehydrogenase encoded by 
rifL were 
assayed for activity by monitoring absorbance at 340 nm due to formation of NADH during the 
dehydrogenation of UDP-
D-glucose in 10 mM sodium phosphate buffer, pH 7.0 at ambient 
temperature (Figure 94).  Enzyme activity was not observed in the cell-free extract with either 
UDP-D-glucose.  Activity of 0.4 U/mg was measured in incubations of UDP-
D-glucose with 
His6-RifL, which had been purified by Ni-NTA metal-affinity chromatography on an •KTA 
FPLC (GE Life Sciences) (Table 23).  Though an observed activity of 0.4 U/mg would be 
expected using cell-free extracts, observed activity of a given enzyme is proportional to the 
purity of that enzyme.  Therefore, an activity of 0.4 U/mg of Ni-NTA purified protein suggests 
%164low inherent catalytic activity.  The lack of activity in cell-free extracts and the small amount of 
activity observed with His
6-tagged protein suggests a difficulty in expressing active enzyme in 
E. 
coli
.  Assays using 
D-glucose-6-phosphate were conducted as a negative control.  
% Figure 94.
  Assay conditions for RifL-catalyzed conversion of UDP-
D-glucose to 3-keto-UDP-
D-glucose. 
 Floss and coworkers report similar challenges associated with expression the kanosamine 
biosynthesis enzymes from 
Amycolatopsis mediterranei 
in 
E. coli
.51  The authors were unable to 
observe phosphatase activity using His
6-RifM and report a lack of His
6-RifK-catalyzed 
aminotransferase activity using a variety of pseudosubstrates in either the forward or reverse 
direction.  They further describe expression of RifL in 
E. coli 
resulted only in formation of 
insoluble protein aggregates.
50  An organism that is more closely related to 
A. mediterranei
, such 
as Streptomyces
 spp. may be better suited for Rif enzyme expression.  
Table 23.
  Specific activity of UDP-
D-glucose 3-dehydrogenase in cell-free extract and with purified protein. 
a His
6-tagged protein purified by Ni-NTA chromatography; cell lysates collected from glucose-limited fermentation 
of RB791serA/pKM11.9.  n.d. = not detected. 
UDPOOHOHHOOHOUDPOOHOHOOHONAD
+NADH
RifL10 mM sodium phosphate 
pH 7.0Enzyme source
aSp. activity (
#
mol/min/mg) (UDPGlc)
Sp. activity (
#
mol/min/mg) (G6P)
His6-RifL
0.41not tested
12 h cell lysate
n.d.n.d.24 h cell lysate
n.d.n.d.36 h cell lysate
n.d.n.d.48 h cell lysate
n.d.n.d.%1654.  Bacillus pumilus
 as source of kanosamine biosynthetic genes 
 The lack of kanosamine accumulation in fermentation of 
E. coli 
harboring plasmid-
localized, codon-optimized 
rif genes from 
Amycolatopsis mediterranei
 and the poor catalytic 
activity associated with 
rifL-encoded UDP-
D-glucose 3-dehydrogenase in 
in vitro
 assays 
prompted a search for an alternative source of kanosamine biosynthesis genes.  
Bacillus pumilus
 was the first microbe reported to accumulate kanosamine in batch fermentations in a media 
consisting of 
D-glucose (1.0 wt%), soybean meal (1.5 wt%), and NaCl (0.3 wt%), at 34¡C for 76 
h.29  When fed universally labelled 
D-[U-
14C]-glucose, fermentation of 
Bacillus pumilus
 resulted 
in universially labelled [U-
14C]-kanosamine.  The same feeding experiment was performed using 
D-[1-14C]-glucose, 
D-[2-14C]-glucose, and 
D-[6-14C]-glucose, which resulted in the production of 
the associated 
14C-labelled kanosamine. In vitro conversion of 
D-[U-
14C]-glucose catalyzed by a 
B. pumilus
 cell-free extract resulted in greater amounts of isotopically labelled kanosamine when 
both ATP and UTP were present in the reaction mixture.  When these cofactors were absent, a 
92% reduction of kanosamine production was observed. These observations ultimately led to the 
proposed pathway presented in Figure 95.
30  %166% Figure 95.
  Kanosamine biosynthesis pathway via UDP-
D-glucose as proposed by Umezawa.
30  A Basic Local Alignment Search Tool (BLAST) query of Rif proteins against protein 
sequence data reported for 
Bacillus pumilus 
(Taxonomy ID: 1408) did not identify highly similar 
Rif analogs in 
Bacillus pumilus
.  The taxonomy ID 1408 group encompasses all 
Bacillus pumilus
 strains derived from type strain 
B. pumilus 
Meyer and Gottheil 1901 and is assigned by the 
National Center for Biotechnology Information (NCBI).  Protein sequence data in the reference 
protein (ref_seq) database was used and proteins were determined to be highly similar if 
sequences contained >30%  alignment with an expect-threshold value <10
*6.  The expect-
threshold (
E) is used to estimate statistical significance and represents the amount of positive 
alignments one ÒexpectsÓ to see by chance when searching a database.  In other terms, this value 
represents random background noise.  Lower expect-thresholds indicate higher degrees of 
significant alignment.  A threshold of 10
*6 indicates that one in one million alignments may be a 
random positive alignment. 
NH2OH2NOOHNH2OHOOOHOHOHOOHOHHOD-glucose
ATP
ADP
Mg2+OHOHOOHOPO
3H2HOD-glucose-
1-phosphate
UTPPO32-OHOHOOHOUDPHOUDP-
D-glucose
NAD
+NADH
OHOOOHOUDPHO3-keto-UDP-
D-glucose
OHOH2NOHOUDPHOUDP-kanosamine
OHOH2NOHOHHOkanosamine
PO32-UMP
UDP = uridine diphosphate
OHOOHNHOOOPOOOHPOOH%167 Protein BLAST queries did reveal sequences for an oxidoreductase, aminotranferase, and 
a phosphatase present in multiple strains of 
Bacillus pumilus
 (Taxonomy ID: 1408) with a high 
degree of similarity and statistical significance to the Ntd protein sequences from 
Bacillus 
subtilis
.  The oxidoreductase protein sequence had 57% identity (78% similar) to 
B. subtilis
 NtdC (
E = 2e*142) over 91% of the 348 amino acid sequence.  The aminotransferase contained 
58% identity (75% similar) to 
B. subtilis
 NtdA (
E = 0.0) over 99% of the 441 amino acid 
sequence.  Lastly, the phosphatase sequence had 55% identity (75% similar) to 
B. 
subtilis NtdB 
(E = 3e*110
) over 96% of the 288 amino acid sequence.  Based on these highly similar sequences 
we wondered if 
Bacillus pumilus
 contains putative 
ntd
 genes that encode kanosamine 
biosynthesis enzymes which might utilize the 
D-glucose-6-phosphate-dependent pathway 
described above for 
B. subtilis
.    Several strains of 
B. pumilus
 have complete genome sequence data.  These include 
B. 
pumilus 
ATCC 7061,
 B. pumilus 
SH-B11, 
B. pumilus
 SAFR-032, and 
B. pumilus 
NCTC 10337, 
among others.  All of these strains are catalogued by NCBI under the Taxonomy ID 1408 and 
were found to contain genes that encode the putative Ntd proteins discovered by BLAST queries.  
The putative 
ntd
 gene sequences from the 
B. pumilus 
SH-B11 genome were submitted to IDT for 
gene synthesis without prior codon optimization.  Codon optimization was not utilized in order 
to make a direct comparison between expression of 
B. pumilus
 genes in 
E. coli
, and the 
previously described expression of 
B. subtilis
 genes, which had been PCR-amplified from the 
B. 
subtilis
 168 genome and therefore not codon-optimized.   
 To distinguish between genes and enzymes native to 
B. subtilis
 and those native to 
B. 
pumilus
, superscript descriptors, sub or pum, will be hereafter included in the name of a gene or 
%168protein.  For example, 
ntdC
sub
 will indicate the gene encoding 
D-glucose-6-phosphate 3-
dehydrogenase native to 
B. subtilis
, while 
ntdC
pum will indicate the gene encoding 
D-glucose-6-
phosphate-3-dehydrogenase native to 
B. pumilus
.    The synthetic 
ntd
pum genes were received from IDT in separate pUC18 vectors with a 5Õ 
Bgl
II restriction site and 3Õ 
Bam
HI and 
Xho
I restriction sites.  Each vector was digested with 
Bgl
II and 
Xho
I and inserted into the 
Bgl
II and 
Xho
I sites of plasmid pBbA1a to create BglBrick 
parts for each 
ntd
pum gene (Figure 96).  A complete plasmid was constructed by using prefix 
insertion of one BglBrick part in front of another.  To begin, the BglBrick part containing 
ntdB
pum (plasmid pKM11.121B) was digested with 
EcoRI and 
Bam
HI while the 
serA-containing 
BglBrick part (plasmid pKM6.240) was digested with 
EcoRI and 
Bgl
II (Figure 97).  Ligation of 
the digested DNA with T4 DNA ligase resulted in plasmid pKM11.125 which contained 
ntdB
pum upstream of 
serA.  The 
EcoRI site was reformed in plasmid pKK11.125 while the 
Bgl
II and 
Bam
HI compatible DNA ends were joined to afford a glycine-serine scar sequence between 
ntdB
pum and 
serA.  Plasmid pKM11.125, contains the same restriction sites as plasmid 
pKM6.240, which allows iterative gene insertion without need to use varying cut sites or 
enzymes.  This method of gene insertion was repeated to insert 
ntdC
pum upstream of 
ntdA
pum affording plasmid, pKM11.130 (Figure 98), which carried a 
Ptrc
ntdC
pumntdA
pum cassette.  Finally, 
ntdC
pumntdA
pum was inserted into pKM11.125 upstream of 
ntdB
pum to afford plasmid pKM11.135 
(Figure 99), which contained the complete operon: 
Ptrc
ntdC
pumntdA
pumntdB
pumserA. 
%169" Figure 96. 
 Construction of 
ntd
 BglBrick parts: plasmids pKM11.121A, pKM11.121B, and pKM11.121C.
  1.3 kb
0.9 kb
1.0 kb
Bacillus pumilus 
SH-B11
EcoRI, Bam
HI digest
EcoRI, Bam
HI digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IntdBntdCAmpRlacI
qp15Aori
mRFP-1
pBbA1a-RFP
4.3 kb
Ptrc
AmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
ntdApumntdBpumntdCpumpKM11.121A
4.9 kb
pKM11.121B
4.4 kb
pKM11.121C
4.6 kb
ntdAEcoRIBam
HIEcoRIBam
HIEcoRIBam
HIBgl
IIBgl
IIBgl
II%170% Figure 97.
  Construction of plasmid pKM11.125. 
0.8 kb
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IntdBAmpRlacI
qp15Aori
Ptrc
serA
pKM11.125
5.7 kb
EcoRIBam
HIntdBpumEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
serA
pKM6.240
4.8 kb
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ntdBpumpKM11.121B
4.4 kb
Bgl
II%171% Figure 98.
  Construction of plasmid pKM11.130. 
1.0 kb
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IEcoR
IBgl
IIBamH
IXho
IntdCAmpRlacI
qp15Aori
Ptrc
AmpRlacI
qp15Aori
Ptrc
ntdApumntdCpumpKM11.130
6.0 kb
pKM11.121C
4.6 kb
EcoRIBam
HIEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ntdApumpKM11.121A
4.9 kb
ntdCpumBgl
II%172% Figure 99.
  Construction of plasmid pKM11.135. 
2.3 kb
EcoRI, Bam
HI digest
EcoRI, Bgl
II digest
T4 ligationEcoR
IBgl
IIBamH
IXho
IntdCntdAAmpRlacI
qp15Aori
Ptrc
ntdCpumpKM11.135
8.1 kb
EcoRIBam
HIserA
ntdApumntdBpumEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
serA
pKM11.125
5.7 kb
ntdBpumEcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
Ptrc
ntdApumpKM11.130
6.0 kb
ntdCpumBgl
II%173 Kanosamine production resulting from expression of 
ntd
pum genes in 
E. coli
 was 
evaluated in glucose-rich (Figure 100) and glucose-limited (Figure 101) fed-batch fermentation 
of RB791serA/pKM11.135.  As was the case during fermentations of RB791
serA/pSN1.292 and 
RB791serA/pKM10.220ABC which carried plasmid-localized 
ntd
sub
 genes, 12 mg IPTG was 
added every 6 h throughout the course of a fermentation beginning at 12 h. 
%   Figure 100.
  Kanosamine production by 
E. coli 
RB791serA/pKM11.135 under glucose-rich conditions. Legend: 
kanosamine, green columns; dry cell weight, black circles.  
Dry cell weight  (g/L)01020304050607080Kanosamine  (g/L)0246810121416Time  (h)12182430364248%174% Figure 101.
  Kanosamine production by 
E. coli 
RB791serA/pKM11.135 under glucose-limited conditions. Legend: 
kanosamine, green columns; dry cell weight, black circles.
  Glucose-rich fermentation of RB791
serA/pKM11.135 resulted in 2.4 g/L kanosamine in a  
0.6% mol/mol yield from 
D-glucose.  This is a seven-fold decrease in titer and yield relative to 
glucose-rich fermentation of RB791
serA/pSN1.292 which harbored kanosamine biosynthetic 
ntd
sub
 genes.  Glucose-limited fermentation also resulted in reduced kanosamine production 
relative to RB791
serA/pSN1.292 fermentation, affording 1.8 g/L kanosamine after 48 h in 1.2% 
mol/mol yield from 
D-glucose.  No new byproducts were observed by 
1H NMR in the 
fermentation broth of RB791
serA/pKM11.135 relative to RB791
serA/pSN1.292 or RB791
serA/pKM10.220ABC, with acetic acid, pyruvic acid, formic acid, and lactic acid forming in both 
glucose-rich and glucose-limited fermentations (Figure 102 and Figure 103).  The overall amount 
of organic acid byproducts was reduced but an increase in final cell density was observed in 
fermentation of RB791
serA/pKM11.135 relative to RB791
serA/pSN1.292.  These observations 
may reflect a difference in the fate of acetyl-CoA between the two constructs.  Described in the 
Dry cell weight  (g/L)01020304050607080Kanosamine  (g/L)0246810121416Time  (h)12182430364248%175literature as Òoverflow metabolismÓ
55 acetyl-CoA can convert to acetic acid during fermentation 
of E. coli
 on D-glucose.  This appears less pronounced in RB791
serA/pKM11.135 than 
RB791serA/pSN1.292.  Instead of conversion to acetic acid, acetyl-CoA may be more effectively 
channelled into the Krebs cycle resulting in greater biomass production in fermentation of 
RB791serA/pKM11.135 relative to RB7981
serA/pSN1.292.  This hypothesis has not been 
explored and is offered as a possible explanation for the observed differences between the two 
constructs regarding cell-densities and organic acid accumulation. 
% Figure 102.  
Byproducts formed during fermentation of 
E. coli
 RB791
serA/pKM11.135 under glucose-rich 
conditions.  Legend: acetic acid, orange columns; formic acid, grey columns; pyruvic acid, red columns; lactic acid, 
blue columns; 
D-glucosamine, purple columns; dry cell weight, black circles.
 Dry cell weight  (g/L)01020304050607080Byproducts  (g/L)
01234567Time  (h)12182430364248%176% Figure 103.  
Byproducts formed during fermentation of 
E. coli
 RB791
serA/pKM11.135 under glucose-limited 
conditions.  Legend: acetic acid, orange columns; formic acid, grey columns; pyruvic acid, red columns; lactic acid, 
blue columns; 
D-glucosamine, purple columns; dry cell weight, black circles.
  A 3.6-fold improvement in kanosamine production was observed in glucose-limited 
conditions following a reduction of IPTG added every 6 h from 12 mg to 6 mg.  Kanosamine 
titers reached 6.3 g/L in a 4.4% mol/mol yield from 
D-glucose (Figure 104).  This yield matches 
those observed during glucose-rich fermentations of RB791
serA/pSN1.292 which carried 
ntd
sub
 genes also under control of 
Ptrc
. (4.4 ± 0.4% mol/mol) though falls short of yields achieved in 
RB791serA/pSN1.292 glucose-limited fermentations (6.2 ± 2.2% mol/mol).  Unexpectedly, 
decreasing the amount of IPTG added every 6 h throughout a fermentation from 12 mg to 6 mg 
did not greatly impact kanosamine production under glucose-rich conditions (Figure 104).  
Kanosamine titers reached 3.0 ± 0.7 g/L in a  1.1 ± 0.3% mol/mol yield from 
D-glucose. 
Dry cell weight  (g/L)01020304050607080Byproducts  (g/L)
01234567Time  (h)12182430364248%177% Figure 104.
  Kanosamine production in fed-batch fermentation of RB791
serA/pKM11.135 in glucose-rich 
conditions with 6 mg IPTG addition every 6 h beginning at 12 h into the fermentation run.  Legend: kanosamine, 
green bars; cell dry weight, black circles.  Error bars represent one standard deviation away from the mean across 
duplicate runs. 
% Figure 105.
  Kanosamine production in fed-batch fermentation of RB791
serA/pKM11.135 in glucose-limited 
conditions with 6 mg IPTG addition every 6 h beginning at 12 h into the fermentation run.  Legend: kanosamine, 
green bars; cell dry weight, black circles. 
Dry cell weight  (g/L)01020304050607080Kanosamine  (g/L)0246810121416Time  (h)121824303642Dry cell weight  (g/L)01020304050607080Kanosamine  (g/L)0246810121416Time  (h)12182430364248%1784.1. Substrate specificity of NtdC
pum 
 Bacillus pumilus
 was first examined as a source of kanosamine biosynthesis enzymes as a 
result of challenges associated with heterologous expression of 
A. mediterranei 
proteins,
 with the 
goal of achieving kanosamine biosynthesis from UDP-
D-glucose in 
E. coli
.  Examination of 
protein sequences using the Basic Local Alignment Search Tool revealed three proteins, an 
oxidoreductase, an aminotransferase, and a phosphatase, with a high degree of similarity to the 
Ntdsub
 proteins native to 
Bacillus subtilis
, which have been reported to synthesize kanosamine 
from 
D-glucose-6-phosphate.  Expression of 
ntd
pum and 
ntd
sub
 genes in 
E. coli
 RB791serA have 
led to kanosamine accumulation of 6.3 g/L (4.6% mol/mol from 
D-glucose) and 8.5 ± 1.4 g/L 
kanosamine (6.2 ± 2.2% mol/mol yield from 
D-glucose) under glucose-limited conditions.  Both 
fermentation results and protein sequence alignments imply 
B. subtilis
 and 
B. pumilus
 contain 
functionally homologous kanosamine biosynthesis enzymes. 
 Evidence supporting homology between 
B. subtilis
 and 
B. pumilus
 kanosamine 
biosynthesis enzymes was found by examining the 
D-glucose-6-phosphate 3-dehydrogenation 
reaction catalyzed by NtdC
pum and NtdC
sub
.  Cell samples were withdrawn from a glucose-
limited fermentation of RB791
serA/pSN1.292 carrying plasmid-localized 
ntd
sub
 genes, as well as 
a glucose-limited fermentation of RB791
serA/pKM11.135, harboring 
ntd
pum genes and achieving 
6.3 g/L kanosamine titers.  Samples were taken at 12, 24, 36, and 48 h and were lysed and 
centrifuged to afford cell-free extracts.  These extracts were used to determine whether the 
oxidoreductase encoded by 
ntdC
 from both 
B. subtilis
 and 
B. pumilus 
was expressed in an active 
form and if so, whether it acts upon 
D-glucose-6-phosphate (G6P) or UDP-
D-glucose (UDPGlc) 
(Table 24).  The results demonstrate that both enzymes act upon 
D-glucose-6-phosphate and have 
%179no significant activity with UDP-
D-glucose.  The minor amount of observed activity is likely due 
to the use of crude cell-free extract which may contain one or more native 
E. coli 
oxidoreductases that acted upon the NAD
+ used in the assay.  The lower relative titers and yields 
observed in fermentation of RB791
serA/pKM11.135 compared to RB791
serA/pSN1.292 is likely 
due to the gradual observed loss of enzyme activity (Table 24).  Improving kanosamine 
production in RB791
serA/pKM11.135 might be achieved through further alteration of IPTG 
additions, performing the fermentation at different temperatures, or synthetic evolution of the 
NtdC
pum protein.  
 From the results presented herein, it is reasonable to conclude that based on sequence 
homology, kanosamine production in fed-batch fermentation, and NtdC substrate specificities, 
that both 
Bacillus subtilis
 and 
Bacillus pumilus 
utilize the same pathway for kanosamine 
biosynthesis via 
D-glucose-6-phosphate.  These results are a departure from the literature 
regarding the nature of kanosamine biosynthesis in 
Bacilus pumilus.
  Table 24. 
 Activity of 
D-glucose-6-phosphate 3-dehydrogenase from 
B. subtilis
 and 
B. pumilus 
when expressed in 
E. 
coli
. Reaction mixture (1 mL) consisted of cell-free extract (0.05-0.5 mg as protein), 
D-glucose-6-phosphate (5 mM) or 
UDP-D-glucose (5 mM), NAD
+ (5 mM), AMP-HCl (100 mM, pH 9.5).
 n.d. = not detected.
 Bacillus pumilus
 NtdC
Bacillus subtilis
 NtdC
Run time  
(h)
Sp. activity(G6P)  
(#
mol/min/mg)
Sp. activity(UDPGlc) 
(#
mol/min/mg)
Sp. activity(G6P)  
(#
mol/min/mg)
Sp. activity(UDPGlc) 
(#
mol/min/mg)
12n.d.n.d.n.d.n.d.240.20n.d.0.27n.d.360.14n.d.0.34n.d.480.091n.d.0.33n.d.%180PART III   
CONCLUSIONS 
 This work sought to address two questions: 1. Can 
E. coli
 produce kanosamine via 
expression of one or both reported biosynthesis pathways? 2. Can modifications be made to 
improve kanosamine production from 
D-glucose? The answer to the first question is clear.  
E. 
coli
 RB791serA accumulates kanosamine in both glucose-rich and glucose-limited fermentations 
when transformed with a plasmid carrying either 
B. subtilis
 or B. pumilus 
biosynthesis genes.  A 
surprising discovery was that both species of 
Bacillus 
appear to synthesize kanosamine via a 
common pathway.  It had been previously reported that 
B. subtilis
 utilizes 
D-glucose-6-phosphate 
while 
B. pumilus
 utilizes UDP-
D-glucose to biosynthesize kanosamine.  However, highly similar 
homologues to the Ntd enzymes native to 
B. subtilis
 were found across multiple strains of 
Bacillus pumilus 
(taxonomy ID: 1408) using the BLAST tool provided by NCBI.  When the Ntd 
homologues from 
Bacillus pumilus
 SH-B11 were expressed in 
E. coli
 RB791serA, 6.3 g/L 
kanosamine was produced in 4.6% mol/mol from 
D-glucose in glucose-limited fermentation.  
Though the titers are lower, yields are similar to those achieved in fermentations of RB791
serA/pSN1.292, which carried the 
ntd
sub
 genes (12.7 ± 0.6 g/L, 4.4 ± 0.4% mol/mol in glucose-rich 
fermentations; 8.5 ± 1.4 g/L, 6.2 ± 2.2% mol/mol in glucose-limited fermentations).  The 
strongest piece of evidence to support a common pathway between 
B. subtilis
 and 
B. pumilus
 came from in vitro assays using cell-free extracts obtained from fermentations of RB791
serA/pKM11.135, expressing 
B. pumilus 
genes under control of 
Ptrc
, and RB791
serA/pSN1.292, 
expressing 
B. subtilis
 genes under control of 
Ptrc
.  Monitoring absorption at 340 nm due to 
%181formation of NADH from NAD
+ revealed good activity (0.2-0.3 U/mg) using 
D-glucose-6-
phosphate and cell-free extract taken at 24 h during a fermentation run, but only trace activity 
with UDP-
D-glucose.  The amount of observed activity using UDP-
D-glucose is most likely due 
to presence of native 
E. coli
 oxidoreductase(s) in the cell-free extract that convert NAD
+ to 
NADH.  The combination of sequence data, kanosamine production in fed-batch fermentation, 
and the common substrate specificity of the oxidoreductase from 
B. subtilis
 and 
B. pumilus
 indicate that both 
Bacillus subtilis
 and 
Bacillus pumilus
 synthesize kanosamine from 
D-glucose-6-phosphate and present a correction to the literature regarding 
Bacillus pumilus
.  Kanosamine did not accumulate during fed batch fermentation of 
E. coli
 harboring 
plasmid-localized kanosamine biosynthesis 
rif genes from 
Amycolatopsis mediterranei
.  Previous 
research demonstrates that 
A. mediterranei 
utilizes kanosamine as a nitrogenous precursor to 3-
amino-5-hydroxybenzoic acid (AHBA) during the biosynthesis of aminoglycoside  
antibiotics.
31,33,36,37  The kanosamine biosynthesis enzymes in 
A. mediterranei 
were reported to 
require UDP-
D-glucose as a substrate.  It was initially thought that the lack of kanosamine 
accumulation was due to an inherent pyrimidine starvation of RB791
serA, but neither uracil 
supplementation nor a strain of 
E. coli
 not known to starve for pyrimidine led to detectable 
amounts of kanosamine in fermentation broth. 
 Cell-free extracts of cell samples taken during fermentation of 
E. coli
 containing plasmid-
localized, codon-optimized 
rif genes were incubated with NAD
+ and UDP-
D-glucose in 
phosphate buffer to detect UDP-
D-glucose-3-dehydrogenase activity.  Only a minor increase in 
absorbance at 340 nm due to formation of NADH was observed.  This was also seen when 
D-glucose-6-phosphate was used in place of UDP-
D-glucose, indicating that the absorbance 
%182increase is not due to kanosamine biosynthetic enzymes but is merely background activity 
resulting from native 
E. coli
 oxidoreductase.  In vitro assays of purified His
6-tagged protein 
resulted in low activity (0.41 U/mg) for the 
rifL-encoded UDP-
D-glucose 3-dehydrogenase.  It 
appears that despite codon-optimization of 
rif genes, expression of active enzymes is difficult in 
E. coli
.  This is supported by reports by the Floss group
51 who first sequenced the AHBA 
biosynthesis gene cluster in 
A. mediterranei
 that neither the 
rifM-encoded UDP-
D-kanosamine 
phosphatase nor the 
rifK-encoded dual-functional 3-keto-UDP-
D-glucose:
L-glutamine 
aminotransferase/AHBA synthase exhibited any activity when expressed in 
E. coli
.  The 
challenges associate with expression of catalytically active kanosamine biosynthesis enzymes 
from 
A. mediterranei 
may be avoided if an alternative host organism is employed.  
Streptomyces 
coelicolor 
has been previously used to produce rifamycin via expression of 
A. mediterranei 
genes and might be a suitable candidate for kanosamine production as well. 
 Despite unsuccessful attempts to accumulate kanosamine via expression of enzymes 
reported to utilize UDP-
D-glucose as a substrate in 
E. coli
, this work demonstrates that 
kanosamine biosynthesis in 
E. coli
 is viable when utilizing 
Bacillus
 genes.  The second question 
to be addressed by this report is whether kanosamine biosynthesis in 
E. coli
 can be improved to 
maximize production in fed-batch fermentation.  Progress has been made in that direction but 
further work is necessary to achieve substantial improvements.   
 The highest titers of kanosamine are observed during fermentation of RB791
serA constructs expressing 
B. subtilis
 genes in glucose-rich conditions: 12.7 ± 0.6 g/L (4.4 ± 0.4% 
mol/mol yield) using a plasmid carrying a 
Ptrc
ntdCntdAntdBserA
 cassette (pSN1.292) and  13.7 
g/L (4.9% mol/mol yield) using a plasmid carrying a 
Ptrc
ntdAntdBntdCserA
 cassette 
%183(pKM10.220ABC).  These fermentations result in accumulation of acetic acid, formic acid, 
pyruvic acid, and lactic acid, which are a result of carbon flux through glycolysis and have been 
attributed to an abundance of 
D-glucose in fermentation broth.  
D-Glucosamine was also 
observed in glucose-rich conditions and is likely due to 
D-glucose availability, as it is not 
observed in glucose-limited fermentations nor those utilizing RB791
serA pgi
::KanR.  Kanosamine yields above 4.9% are achieved by restricting 
D-glucose availability in glucose-
limited fermentations.  Titers of 8.5 ± 1.4 g/L kanosamine are observed in a 6.2 ±  2.2% mol/mol 
yield from 
D-glucose.   
 Because the major byproducts observed in fermentation of RB791
serA/pSN1.292 are 
result from carbon flow through glycolysis, a mutant strain was utilized in which the Embden-
Meyerhoff-Parnas pathway of glycolysis was blocked at 
pgi-encoded phosphoglucose isomerase.  
This was in effort to increase the intracellular availability of 
D-glucose-6-phosphate for 
kanosamine biosynthesis.  The mutant strain, RB791
serA pgi
::KanR, was transformed with 
pSN1.292 and kanosamine production was examined using a 
D-glucose-only feed or a mixed 
feed consisting of equimolar amounts of 
D-glucose and 
D-fructose.  The highest yields of 
kanosamine observed thus far were achieved using a feed consisting of only 
D-glucose.  
Kanosamine titers reached 3.6 ± 0.2 g/L in an 18 ± 1% mol/mol yield from 
D-glucose.  These 
fermentations resulted in only transient growth of RB791
serA pgi
::KanR/pSN1.292 with final 
cell densities reaching 2.9 ± 1.4 g/L dry cell weight.  Furthermore, only 20 g of 
D-glucose was 
consumed during these runs.  Growth was restored and 
D-glucose consumption increased with 
the use of a feed consisting of equimolar amounts of 
D-glucose and 
D-fructose.  Final cell 
%184densities reached 37 ± 2.4 g/L dry cell weight with 150 g 
D-glucose consumed.  Kanosamine 
titers reached 8.7 ± 0.08 g/L in 5.9 ± 0.6% mol/mol yield.   
 It appears that maximizing kanosamine production by blocking a major glycolysis 
pathway must be balanced with deleterious growth effects when feeding 
D-glucose.  A mixed 
carbohydrate feed restored cell viability but did not result in high titers or yields of kanosamine.  
A substantial amount of 
D-fructose was observed at the completion of the fermentations of 
RB791serA pgi
::KanR/pSN1.292 while 
D-glucose had been completely consumed.  This is 
indicative of carbon catabolite repression, whereby 
D-glucose is consumed by the cell 
preferentially over other carbohydrates.    
 Previous research has demonstrated high yields of 
D-glucaric acid from 
D-glucose by 
indroducing a mutation in 
zwf
-encoded 
D-glucose-6-phosphate dehydrogenase to a 
+pgi mutant 
E. coli
.48  With both major pathways of glycolysis blocked, the redox imbalance leading to 
reduced cell viability in 
+pgi E. coli
 was not observed, nor was competing carbon catabolite 
repression when cells were cultured on a mixed 
D-glucose/
D-xylose feed.  It was shown that cells 
grew exclusively on 
D-xylose while using most of the available 
D-glucose for 
D-glucaric acid 
biosynthesis.  This could be a possible strategy to achieve kanosamine yields of 18% or greater 
without sacrificing cell viability.   
 The observation of 
L-glutamic acid accumulation in fermentation broth of 
+pgi 
mutants 
with the concommitant production of 3.6 ± 0.2 g/L kanosamine with 
D-glucose-only feed may 
suggest increased 
L-glutamate levels have a beneficial effect on kanosamine production in 
E. coli
 expressing 
Bacillus
 genes.  
L-Glutamate is a substrate for the 
ntdA
-encoded 3-keto-
D-glucose-6-
phosphate:L-glutamate aminotransferase.  Exogenous 
L-glutamic acid did not improve 
%185kanosamine production, but increasing carbon flux through 
L-glutamate biosynthesis in 
E. coli
 might lead to greater substrate availability and improved kanosamine production. 
%186REFERENCES 
%187REFERENCES 
1.  Mehta, C. M.; Palni, U.; Franke-Whittle, I. H.; Sharma, A. K. Compost: Its Role, Mechanism 
and Impact on Reducing Soil-Borne Plant Diseases. 
Waste Manag.
 2014, 34 (3), 607Ð622. 2.  Handelsman, J.; Stabb, E. V. Biocontrol of Soilborne Plant Pathogens. 
Plant Cell
 1996, 8 (10), 1855Ð1869. 3.  Caulier, S.; Gillis, A.; Colau, G.; Licciardi, F.; Li”pin, M.; Desoignies, N.; Modrie, P.; 
Legr‘ve, A.; Mahillon, J.; Bragard, C. Versatile Antagonistic Activities of Soil-Borne 
Bacillus
 spp. and 
Pseudomonas
 spp. against 
Phytophthora infestans
 and Other Potato Pathogens. 
Front. 
Microbiol.
 2018, 9, 143. 4.  Liu, L.; Liang, M.; Li, L.; Sun, L.; Xu, Y.; Gao, J.; Wang, L.; Hou, Y.; Huang, S. Synergistic 
Effects of the Combined Application of 
Bacillus subtilis
 H158 and Strobilurins for Rice Sheath 
Blight Control. 
Biol. Control
 2018, 117
, 182Ð187. 5.  Mora, I.; Cabrefiga, J.; Montesinos, E. Cyclic Lipopeptide Biosynthetic Genes and Products, 
and Inhibitory Activity of Plant-Associated 
Bacillus
 against Phytopathogenic Bacteria. 
PLoS 
One
 2015, 10 (5), e0127738. 
6.  Shrestha, B. K.; Karki, H. S.; Groth, D. E.; Jungkhun, N.; Ham, J. H. Biological Control 
Activities of Rice-Associated 
Bacillus
 sp. Strains against Sheath Blight and Bacterial Panicle 
Blight of Rice. 
PLoS One
 2016, 11 (1), e0146764. 
7.  Jacobsen, B. J.; Zidack, N. K.; Larson, B. J. The Role of 
Bacillus
-Based Biological Control 
Agents in Integrated Pest Management Systems: Plant Diseases. 
Phytopathology
 2004, 94 (11), 
1272Ð1275. 8.  Ongena, M.; Jourdan, E.; Adam, A.; Paquot, M.; Brans, A.; Joris, B.; Arpigny, J.-L.; Thonart, 
P. Surfactin and Fengycin Lipopeptides of 
Bacillus subtilis
 as Elicitors of Induced Systemic 
Resistance in Plants. 
Environ. Microbiol.
 2007, 9 (4), 1084Ð1090. 9.  Jourdan, E.; Henry, G.; Duby, F.; Dommes, J.; Barth”lemy, J. P.; Thonart, P.; Ongena, M. 
Insights into the Defense-Related Events Occurring in Plant Cells Following Perception of 
Surfactin-Type Lipopeptide from 
Bacillus subtilis
. Mol. Plant. Microbe. Interact.
 2009, 22 (4), 456Ð468. 10.  Lim, J.-H.; Kim, S.-D. Synergistic Plant Growth Promotion by the Indigenous Auxins-
Producing PGPR 
Bacillus subtilis
 AH18 and 
Bacillus licheniforims
 K11. 
Hanguk Ungyong 
Saengmyong Hwahakhoe Chi
 2009, 52 (5), 531Ð538. %18811.  Drahos, D. J.; West, L. 
Bacillus lichenformis
 Biofungicide.US 6,569,425 B2 , 2003. 
12.  Guilhabert-Goya, M.; Zhu, H. Method of Increasing Abiotic Stress Resistance in a Plant. US 
2014/0066302 A1 , 2014. 
13.  Brannen, P. M.; Kenney, D. S. KodiakÑa Successful Biological-Control Product for 
Suppression of Soil-Borne Plant Pathogens of Cotton. 
J. Ind. Microbiol. Biotechnol.
 1997, 19, 169Ð171. 14.  Williams, J. A.; Guicherit, O. M.; Zaharian, B. I.; Xu, Y.; Chai, L.; Wichterle, H.; Kon, C.; 
Gatchalian, C.; Porter, J. A.; Rubin, L. L.; et al. Identification of a Small Molecule Inhibitor of 
the Hedgehog Signaling Pathway: Effects on Basal Cell Carcinoma-like Lesions. 
Proc. Natl. 
Acad. Sci. U. S. A.
 2003, 100 (8), 4616Ð4621. 15.  Ryu, C.-M.; Farag, M. A.; Hu, C.-H.; Reddy, M. S.; Kloepper, J. W.; Par”, P. W. Bacterial 
Volatiles Induce Systemic Resistance in 
Arabidopsis
. Plant Physiol.
 2004, 134 (3), 1017Ð1026. 16.  Yu, X.; Ai, C.; Xin, L.; Zhou, G. The Siderophore-Producing Bacterium, 
Bacillus subtilis
 CAS15, Has a Biocontrol Effect on 
Fusarium
 Wilt and Promotes the Growth of Pepper. 
Eur. J. 
Soil Biol.
 2011
, 47 (2), 138Ð145. 17.  Gao, Z.; Zhang, B.; Liu, H.; Han, J.; Zhang, Y. Identification of Endophytic 
Bacillus 
velezensis 
ZSY-1 Strain and Antifungal Activity of Its Volatile Compounds against 
Alternaria
 solani
 and 
Botrytis cinerea
. Biol. Control
 2017, 105, 27Ð39. 18.  Yu, G. Y.; Sinclair, J. B.; Hartman, G. L.; Bertagnolli, B. L. Production of Iturin A by 
Bacillus amyloliquefaciens
 Suppressing 
Rhizoctonia solani
. Soil Biol. Biochem.
 2002, 34 (7), 955Ð963. 19.  Skarbek, K.; Milewska, M. J. Biosynthetic and Synthetic Access to Amino Sugars. 
Carbohydr. Res.
 2016, 434, 44Ð71. 20.  Milner, J. L.; Silo-Suh, L.; Lee, J. C.; He, H.; Clardy, J.; Handelsman, J. Production of 
Kanosamine by 
Bacillus cereus
 UW85. 
Appl. Environ. Microbiol.
 1996, 62 (8), 3061Ð3065. 21.  Iwai, Y.; Tanaka, H.; Oiwa, R.; Shimizu, S.; Omura, S. III. 3-Amino-3-Deoxy-
D-Glucose, an 
Inhibitor of Bacterial Cell Wall Biosynthesis. 
Biochim. Biophys. Acta
 1977, 498, 223Ð228. 22.  Janiak, A. M.; Milewski, S. Mechanism of Antifungal Action of Kanosamine. 
Med. Mycol.
 2001, 39 (5), 401Ð408. %18923.  Silo-Suh, L. A.; Lethbridge, B. J.; Raffel, S. J.; He, H.; Clardy, J.; Handelsman, J. Biological 
Activities of Two Fungistatic Antibiotics Produced by 
Bacillus cereus 
UW85. 
Appl. Environ. 
Microbiol.
 1994, 60 (6), 2023Ð2030. 24.  Silo-Suh, L. A.; Stabb, E. V.; Raffel, S. J.; Handelsman, J. Target Range of Zwittermicin A, 
an Aminopolyol Antibiotic from 
Bacillus cereus
. Curr. Microbiol.
 1998, 37 (1), 6Ð11. 
25.  Vetter, N. D.; Langill, D. M.; Anjum, S.; Boisvert-Martel, J.; Jagdhane, R. C.; Omene, E.; 
Zheng, H.; van Straaten, K. E.; Asiamah, I.; Krol, E. S.; et al. A Previously Unrecognized 
Kanosamine Biosynthesis Pathway in 
Bacillus subtilis
. J. Am. Chem. Soc.
 2013, 135 (16), 5970Ð5973. 26.  a) van Straaten, K. E.; Ko, J. B.; Jagdhane, R.; Anjum, S.; Palmer, D. R. J.; Sanders, D. A. R. 
The Structure of NtdA, a Sugar Aminotransferase Involved in the Kanosamine Biosynthetic 
Pathway in 
Bacillus subtilis,
 Reveals a New Subclass of Aminotransferases. 
J. Biol. Chem.
 2013, 288 (47), 34121Ð34130.  b) van Straaten, K. E.; Langill, D. M.; Palmer, D. R. J.; Sanders, D. A. 
R. Purification, Crystallization and Preliminary X-Ray Analysis of NtdA, a Putative Pyridoxal 
Phosphate-Dependent Aminotransferase from 
Bacillus subtilis
. Acta Crystallogr. Sect. F Struct. 
Biol. Cryst. Commun.
 2009, 65 (4), 426Ð429. 27.  Vetter, N. D.; Palmer, D. R. J. Simultaneous Measurement of Glucose-6-Phosphate 3-
Dehydrogenase (NtdC) Catalysis and the Nonenzymatic Reaction of Its Product: Kinetics and 
Isotope Effects on the First Step in Kanosamine Biosynthesis. 
Biochemistry
 2017, 56 (14), 2001Ð2009. 28.  Milner, J. L.; Silo-Suh, L.; Lee, J. C.; He, H.; Clardy, J.; Handelsman, J. Production of 
Kanosamine by 
Bacillus cereus
 UW85. 
Appl. Environ. Microbiol.
 1996, 62 (8), 3061Ð3065. 29.  Umezawa, S.; Umino, K.; Shibahara, S.; Omoto, S. Studies of Aminosugars. XVII. 
Production of 3-Amino-3-Deoxy-
D-Glucose by 
Bacillus
 Species. 
Bull. Chem. Soc. Jpn.
 1967, 40 (10), 2419Ð2421. 30.  Umezawa, S.; Shibahara, S.; Omoto, S.; Takeuchi, T.; Umezawa, H. Studies on the 
Biosynthesis of 3-Amino-3-Deoxy-
D-Glucose. 
The Journal of Antibiotics
 1968, 21 (8), 485Ð491. 31.  Kim, C.-G.; Kirschning, A.; Bergon, P.; Zhou, P.; Su, E.; Sauerbrei, B.; Ning, S.; Ahn, Y.; 
Breuer, M.; Leistner, E.; et al. Biosynthesis of 3-Amino-5-Hydroxybenzoic Acid, the Precursor 
of mC7N Units in Ansamycin Antibiotics. 
J. Am. Chem. Soc.
 1996, 118
 (32), 7486Ð7491. 32.  Ventura, M.; Canchaya, C.; Tauch, A.; Chandra, G.; Fitzgerald, G. F.; Chater, K. F.; van 
Sinderen, D. Genomics of 
Actinobacteria
: Tracing the Evolutionary History of an Ancient 
Phylum. 
Microbiol. Mol. Biol. Rev.
 2007, 71 (3), 495Ð548. %19033.  Yu, T. W.; Muller, R.; Muller, M.; Zhang, X.; Draeger, G.; Kim, C. G.; Leistner, E.; Floss, H. 
G. Mutational Analysis and Reconstituted Expression of the Biosynthetic Genes Involved in the 
Formation of 3-Amino-5-Hydroxybenzoic Acid, the Starter Unit of Rifamycin Biosynthesis in 
Amycolatopsis mediterranei 
S699. 
J. Biol. Chem.
 2001, 276 (16), 12546Ð12555. 34.  August, P. R.; Tang, L.; Yoon, Y. J.; Ning, S.; Muller, R.; Yu, T.-W.; Taylor, M.; Hoffmann, 
D.; Kim, C.-G.; Zhang, X.; et al. Biosynthesis of the Ansamycin Antibiotic Rifamycin: 
Deductions from the Molecular Analysis of the Rif Biosynthetic Gene Cluster of 
Amycolatopsis 
mediterranei
 S699. 
Chem. Biol.
 1998, 5 (2), 69Ð79. 35.  Wilson, M. C.; Gulder, T. A. M.; Mahmud, T.; Moore, B. S. Shared Biosynthesis of the 
Saliniketals and Rifamycins in 
Salinispora arenicola
 Is Controlled by the sare1259-Encoded 
Cytochrome P450. 
J. Am. Chem. Soc.
 2010, 132 (36), 12757Ð12765. 36.  Guo, J.; Frost, J. W. Biosynthesis of 1-Deoxy-1-Imino-
D-Erythrose 4-Phosphate: A Defining 
Metabolite in the Aminoshikimate Pathway. 
J. Am. Chem. Soc.
 2002, 124 (4), 528Ð529. 37.  Guo, J.; Frost, J. W. Kanosamine Biosynthesis: A Likely Source of the Aminoshikimate 
PathwayÕs Nitrogen Atom. 
J. Am. Chem. Soc.
 2002, 124 (36), 10642Ð10643. 38.  Lim, H. N.; Lee, Y.; Hussein, R. Fundamental Relationship between Operon Organization 
and Gene Expression. 
Proc. Natl. Acad. Sci. U. S. A.
 2011
, 108 (26), 10626Ð10631. 39.  Chubukov, V.; Sauer, U. Environmental Dependence of Stationary-Phase Metabolism in 
Bacillus subtilis
 and Escherichia Coli. 
Appl. Environ. Microbiol.
 2014, 80 (9), 2901Ð2909. 40.  Frost, J. W.; Hansen, C. A. Synthesis of 1,2,3,4-Tetrahydroxybenzenes and 1,2,3,-
Trihydroxybenzenes Using 
myo-Inositol-1-Phosphate Synthase and 
myo-Inositol-2-
Dehydrogenase. US 6,750,049, 2004. 
41.  Tegel, H.; Ottosson, J.; Hober, S. Enhancing the protein production levels in 
Escherichia coli
 with a strong promoter. 
FEBS J. 
2011
, 278, 729-739. 42  Murarka, A.; Clomburg, J. M.; Moran, S.; Shanks, J. V.; Gonzalez, R. Metabolic Analysis of 
Wild-Type 
Escherichia coli
 and a Pyruvate Dehydrogenase Complex (PDHC)-Deficient 
Derivative Reveals the Role of PDHC in the Fermentative Metabolism of Glucose. 
J. Biol. 
Chem.
 2010, 285 (41), 31548Ð31558. 43.  Lessmann, D.; Schimz, K.-L.; Kurz, G. 
D-Glucose-6-Phosphate Dehydrogenase (Entner-
Doudoroff Enzyme) from 
Pseudomonas fluorescen
s. 
Eur. J. Biochem.
 1975, 59, 545Ð559. %19144.  Morita, T.; El-Kazzaz, W.; Tanaka, Y.; Inada, T.; Aiba, H. Accumulation of Glucose 6-
Phosphate or Fructose 6-Phosphate Is Responsible for Destabilization of Glucose Transporter 
mRNA in 
Escherichia coli
. J. Biol. Chem.
 2003, 278 (18), 15608Ð15614. 45.  Kogure, T.; Wakisaka, N.; Takaku, H.; Takagi, M. Efficient Production of 2-Deoxy-
Scyllo
-Inosose from 
D-Glucose by Metabolically Engineered Recombinant Escherichia Coli. 
J. 
Biotechnol.
 2007, 129 (3), 502Ð509. 46.  A. Dean, unpublished results 
47.  Charusanti, P.; Conrad, T. M.; Knight, E. M.; Venkataraman, K.; Fong, N. L.; Xie, B.; Gao, 
Y.; Palsson, B. ¯. Genetic Basis of Growth Adaptation of 
Escherichia coli
 after Deletion of 
pgi, a Major Metabolic Gene. 
PLoS Genet.
 2010, 6 (11), e1001186. 
48.  Brockman, I. M.; Prather - Biotechnology and, K.; 2015. Improving Product Yields on 
D-glucose in 
Escherichia coli
 via Knockout of 
pgi and 
zwf
 and Feeding of Supplemental Carbon 
Sources. 
Wiley Online Library
 2015. 49.  Bren, A.; Park, J. O.; Towbin, B. D.; Dekel, E.; Rabinowitz, J. D.; Alon, U. Glucose 
Becomes One of the Worst Carbon Sources for 
E.coli
 on Poor Nitrogen Sources due to 
Suboptimal Levels of cAMP. 
Sci. Rep.
 2016, 6, 24834. 49.  Guo, J. Biosynthesis of Precursors to the Aminoshikimate Pathway and Microbial Synthesis 
of 5-Amino-5-Deoxyshikimate. Doctor of Philosophy, Michigan State University, 2004. 
51.  Floss, H. G.; Yu, T.-W.; Arakawa, K. The Biosynthesis of 3-Amino-5-Hydroxybenzoic Acid 
(AHBA), the Precursor of mC7N Units in Ansamycin and Mitomycin Antibiotics: A Review. 
J. 
Antibiot. 
2011
, 64 (1), 35Ð44. 52  PuigbŸ, P.; Guzm⁄n, E.; Romeu, A.; Garcia-Vallv”, S. OPTIMIZER: A Web Server for 
Optimizing the Codon Usage of DNA Sequences. 
Nucleic Acids Res.
 2007, 35 (Web Server 
issue), W126ÐW131 
53.  Jensen, K. F. The 
Escherichia coli
 K-12 ÒWild TypesÓ W3110 and MG1655 Have an Rph 
Frameshift Mutation That Leads to Pyrimidine Starvation Due to Low 
pyrE
 Expression Levels. 
Journal of Bacteriology
 1993, 175 (11), 3401Ð3407. 
54.  Li, K.; Mikola, M. R.; Draths, K. M.; Worden, R. M.; Frost, J. W. Fed-Batch Fermentor 
Synthesis of 3-Dehydroshikimic Acid Using Recombinant 
Escherichia coli
. Biotechnol. Bioeng.
 1999, 64 (1), 61Ð73. 55.  Vemuri GN, Altman E, Sangurdekar DP, Khodursky AB, Eiteman MA. Overflow 
metabolism in 
Escherichia coli
 during steady-state growth: transcriptional regulation and effect 
of the redox ratio. 
Appl Environ Microbiol 
2006, 72, 3653Ð3661. %19256.  Hansen, C. Chemo-enzymatic synthesis of aromatics via non-shikimate pathway 
intermediates. Doctor of Philosophy, Michigan State University, 2002. 
57.  Lee, T. S.; Krupa, R. A.; Zhang, F.; Hajimorad, M.; Holtz, W. J.; Prasad, N.; Lee, S. K.; 
Keasling, J. D. BglBrick Vectors and Datasheets: A Synthetic Biology Platform for Gene 
Expression. 
J. Biol. Eng.
 2011
, 5, 12. 58.  Pittard, J.; Wallace, B. J. Distribution and Function of Genes Concerned with Aromatic 
Biosynthesis in 
Escherichia coli
. J. Bacteriol.
 1966, 91 (4), 1494Ð1508. %193CHAPTER THREE: EXPERIMENTAL 
General methods 
General chemistry 
 All reactions sensitive to air and moisture were carried out in oven-dried glassware under 
a positive atmosphere of nitrogen or in a Parr high pressure reaction vessel.  All reagents 
sensitive to air and moisture were handled under a positive nitrogen atmosphere and transferred 
to reaction flasks fitted with rubber septa via oven-dried syringes or cannula.  Unless otherwise 
specified, all reactions were carried out at room temperature.  Solvents were removed using a 
hi rotary evaporator at water aspirator pressure or under high vacuum (0.5 mm Hg).  
 Toluene, 
m-xylene, and bis(methoxypropyl) ether were dried over 4 
" molecular sieves.  
1,4-Dioxane was distilled from sodium metal under a positive nitrogen atmosphere.  Water was 
glass-distilled and deionized.  All other reagents and solvents were used as available from 
commercial sources.  Organic solutions of products were dried over anhydrous MgSO
4.  The 
sodium salt of 3-(trimethylsilyl)propionic-
2,2,3,3-d4 acid (TSP) was purchased from Sigma-
Aldrich.  BOB(OAc)
4 was prepared and characterized previously by this laboratory.
1     Analytical methods 
 1H NMR spectra were recorded on a 500 MHz spectrometer.  For 1H NMR experiments, 
the relaxation delay parameter (d
1) was adjusted to 10 s for 2-HMA-containing samples or 25 s 
for kanosamine-containing samples.  Chemical shifts are reported (in parts per million) relative 
to residual non-deuterated solvent in CD
3OD (
-
 = 3.31 ppm) and DMSO-
d6 (-
 = 2.50 ppm).  
%194When D
2O was used as the solvent, chemical shifts are reported (in parts per million) relative to 
internal sodium 3-(trimethylsilyl)propionate-
2,2,3,3-d4 (TSP, 
-
 = 0.00 ppm).  The following 
abbreviations are used to describe spin multiplicity: s (singlet), d (doublet), t (triplet), q (quartet), 
dd (doublet of doublets), ddd (doublet of doublet of doublets), dt (doublet of triplets) m 
(unresolved multiplet).  
13C NMR spectra were recorded at 126 MHz and chemical shifts are 
reported (in parts per million) relative to CD
3OD (
-
 = 49.0 ppm) or DMSO-
d6 (-
 = 39.5 ppm).  
When D
2O was used as the solvent, chemical shifts are reported (in parts per million) relative to 
internal sodium 3-(trimethylsilyl)propionate-
2,2,3,3-d4 (TSP, 
-
 = 0.0 ppm).  Concentrations of 
fermentation and cell-free reaction products were determined by comparison of the integrals 
corresponding to each compound with the integral corresponding to TSP (
-
 = 0.0 ppm) or maleic 
acid (
-
 = 6.32 ppm) in the 
1H NMR.  Response factors were determined as follows: A known 
quantity of each compound was dissolved in 5 mL D
2O to obtain a stock solution.  Compounds 
with exchangeable protons were concentrated to dryness under reduced pressure and redissolved 
in an equal volume of D
2O a total of three times.  Various known volumes were then 
concentrated to dryness and redissolved in 1 mL D
2O containing 1 mM TSP or maleic acid and 
their 
1H NMR spectra were recorded.  Compounds were quantified using the following 
resonances: 2-hydroxymuconic acid (2-HMA, 
-
 7.71, dd, 1H), 
!
-keto acid 
24 (-
 6.97, dt, 2H), 
!
-keto acid 
23 (-
 6.89, dt, 2H), formic acid (
-
 8.26, s, 1H), HOAc (
-
 1.92, s, 3H), pyruvic acid (
-
 2.41, s, 3H), lactic acid (
-
 1.35, s, 3H), 
L-glutamic acid (
-
 2.35, dt, 2H), 3-dehydroquinic acid (
-
 4.38, d, 1H) , 3-dehydroshikimic acid (
-
 4.28, d, 1H), gallic acid (
-
 7.02, s 2H).  The 
concentrations of formic acid, HOAc, pyruvic acid, lactic acid, and 
L-glutamic acid were 
calculated by application of the following response factors derived from standards taken using 
%195reagent-grade samples.  The concentrations of gallic acid, 3-dehydroshikimic acid, and 3-
dehydroquinic acid were calculated by application of the following response factors derived from 
laboratory stock samples: 
 [glutamic acid mM]
actual
 = 0.93[L-glutamic acid mM]
NMR 
 [acetic acid mM]
actual
 = 1.12[L-glutamic acid mM]
NMR 
 [pyruvic acid mM]
actual
 = 0.85[L-glutamic acid mM]
NMR 
 [formic acid mM]
actual
 = 0.91[L-glutamic acid mM]
NMR 
 [gallic
 acid mM]
actual
 = 1.36[L-glutamic acid mM]
NMR 
 [lactic
 acid mM]
actual
 = 1.21[L-glutamic acid mM]
NMR 
 [3-dehydroquinic
 acid mM]
actual
 = 0.95[3-dehydroquinic
 acid mM]
NMR 
 [3-dehydroshikimic
 acid mM]
actual
 = 1.31[3-dehydroshikimic
 acid mM]
NMR 
 The concentrations of 2-hydroxymuconic acid, 
!
-keto acid 
23, and 
!
-keto acid 
24 were 
calculated by application of a response factor derived from standards taken using laboratory-
synthesized 2-hydroxymuconic acid: [mM]actual = 1.02[mM]
NMR   The concentrations of glucosamine and kanosamine were calculated by application of a 
calibration formula derived from standards taken using reagent-grade 
D-glucosamine 
hydrochloride (Alfa Aesar):  [mM]
actual
 = 1.01[mM]
NMR  Kanosamine and glucosamine concentrations were calculated from integration of 
anomeric resonances.  The ratio of 
!
- and 
#
-anomers of each aminosugar varied somewhat 
between samples [kanosamine (
-
 5.27, d, 0.3-0.5H and 
-
 4.73, d, 0.5-0.7H), glucosamine (
-
 5.43, %196d, 0.3-0.5H and 
-
 4.96, d, 0.5-0.7H)].  Therefore, the integration of both anomeric peaks were 
combined and applied to the calibration formulae for the respective compounds.  When this was 
not possible due to overlap of the 
!
-anomeric proton peak, an average ratio was used: 
!
:
#
 = 0.45:0.55 for both kanosamine and glucosamine. 
 UV and visible measurements were recorded on an Agilent 8453 UV-Vis 
spectrophotometer with a Dell Dimension 8100 equipped with ChemStation operating software 
or a ThermoFisher NanoDropª One
C microvolume UV-Vis spectrophotometer. 
 HPLC analysis was performed on an Agilent 1100 instrument.  Terephthalic acid and 
phthalic acid was quantified by HPLC using an Agilent ZORBAX SB-C18 column (4.6 x 150 
mm, 5 
$
m particle size) with isocratic elution of 15/85 (v/v) CH
3CN/H
2O (100 mM NH
4HCO
2, pH 2.5).  The ZORBAX column was received in a solution of 85/15 (v/v) CH
3OH/H2O and was 
subsequently washed with 1 L HPLC-grade CH
3CN (0.2 mL/min) prior to use.  Analytical 
standards were obtained from Sigma-Aldrich.  Requisite buffers were prepared with degassed, 
deionized, distilled water which had been filtered through a DURAPORE¨ 0.45 
$
m HV filter. 
(MILLIPORE).  The eluent buffer and HPLC-grade CH
3CN was mixed online.  All samples 
analyzed by HPLC were previously filtered through 0.45 
$
m disc filters (Whatman). 
Bacterial strains and plasmids 
 E. coli
 DH5
!
( [F* endAl glnV44
 thi-1 recA1 gyrA96 deoR nupG purB20
 /
80dlacZ
+
M15 
+(lacZYA
-argF
)U169
 hsdR17
(rK*mK+) "*] was previously obtained by this laboratory from 
Invitrogen.  The laboratory strain 
E. coli
 WN1 [F
* tsx-352
 glnV42
 aroE353
 malT352
("R) "* %197serA::aroBaroZ
 lacZ
::tktAaroZ
]  was previously constructed by Wei Niu in the Frost group.
2  The 
laboratory strain RB791
serA (W3110 
lacL8I
qserA) was previously constructed by this  
laboratory.
3  E. coli
 KM1 (WN1 
pptA
::FRT) was prepared using 
"
 Red recombinase-mediated 
homologous recombination.
4  Paenibacillus 
sp. JJ-1b (ATCC 35889) was obtained from ATCC.
5  Plasmids pBbA1a-RFP and pBbA7k-RFP were obtained from Addgene.
6  Plasmids pWN1.262,
2 pWN1.28,
2 pMF63,
2 and pJG7.246
7 were constructed previously by this laboratory.  
 Plasmids were transformed into 
E. coli
 DH5
!
 for long-term storage.  All bacterial strains 
were stored at 
*80¡C in glycerol.  Glycerol samples were prepared by adding 0.75 mL of a 
stationary phase culture to a sterile vial containing 0.25 mL 80% (v/v) glycerol.  The solution 
was mixed by inversion, incubated at rt for 2 h, then stored at 
*80¡C. 
Plasmid construction via BglBrick assembly standard 
 All plasmids described in this work were constructed using the BglBrick Assembly 
Standard developed by Anderson and Keasling.
6  A general outline for the method of plasmid 
construction is shown in Figure 106 and described herein.  Before genes could be combined into 
an expression plasmid, a BglBrick part was created for each gene of interest.  This was done by 
inserting the PCR-amplified or synthetic genes between the 
Bgl
II and 
Bam
HI sites in a standard 
BglBrick cloning vector, pBbA1a (Figure 107). 
 Once each gene had been inserted, two genes were concatenated following the prefix-
insertion method described in Figure 105.  The gene to be inserted (Gene 1, Fig. 106) is liberated 
by digestion with 
EcoRI and 
Bam
HI while the recipient vector is digested with 
EcoRI and 
Bgl
II.  Gene 1 is inserted upstream of Gene 2 in the vector following ligation of the digested DNA 
%198fragments.  The 
EcoRI site is reformed upstream of Gene 1 in the product plasmid while 
compatible BamHI/BglII sticky ends form a two amino acid scar sequence between Gene 1 and 
Gene 2 encoding glycine-serine (Figure 106).  The final plasmid retains the 
EcoRI, Bgl
II, Bam
HI, and 
Xho
I sites that the original BglBrick parts contained.  Therefore, any gene can be 
inserted upstream of any other by following this method of digestion and ligation.  
% Figure 106.
  BglBrick plasmid assembly workflow. 
GENE 1
EcoRIBgl
IIXho
IBam
HIGENE 2
EcoRIBgl
IIXho
IBam
HICut withEcoRI and Bam
HICut withEcoRI and Bgl
IIGENE 2
EcoRIBgl
IIXho
IBam
HIGENE 1
EcoRIBgl
IIBam
HIMix and 
LigateGENE 2
EcoRIBgl
IIXho
IBam
HIGENE 1
GlySer
GAATTC
nnnAGATCT
GENE1
GGATCC
nnnCTCGAG
GAATTC
nnnAGATCT
GENE2
GGATCC
nnnCTCGAG
CTTAAG
nnnTCTAGA
GENE1
CCTAGG
nnnGAGCTC
5Õ3Õ5Õ3Õ3Õ5ÕCTTAAG
nnnTCTAGA
GENE2
CCTAGG
nnnGAGCTC
3Õ5ÕGAATTC
nnnAGATCT
GENE1
GGATCT
GENE2
GGATCC
nnnCTCGAG
GlySer
AATTC
nnnAGATCT
GENE1
GGnnnTCTAGA
GENE1
CCTAG
5Õ3ÕGATCT
GENE2
GGATCC
nnnCTCGAG
AGENE2
CCTAGG
nnnGAGCTC
3Õ5Õ5Õ3Õ3Õ5ÕCTTAAG
nnnTCTAGA
GENE1
GGATCT
GENE2
CCTAGG
nnnGAGCTC
%199% Figure 107.
  BglBrick vector pBbA1a-RFP. 
Culture medium 
 Bactoª
 tryptone, Difcoª yeast extract, and Bactoª agar were obtained from BD 
Diagnostic Systems.  Ampicillin, thiamine hydrochloride, 2,3-dihydroxybenzoic acid, 4-
aminobenzoic acid, and 4-hydroxybenzoic acid were obtained from Spectrum.  Isopropyl-
#
-D-thiogalactopyranoside (IPTG) was obtained from Gold Biotechnology.  Ammonium 
heptamolybdate tetrahydrate was obtained from Mallinckrodt.  Aromatic amino acids (L-
phenylalanine, L-tyrosine, and L-tryptophan), iron(III) ammonium citrate, and kanamycin were 
purchased from Sigma Aldrich.  All solutions were prepared in distilled, deionized water. 
 LB medium
8 (1 L) contained Bactoª tryptone (10 g), Difcoª yeast extract (5 g), and 
NaCl (10 g).  2
!Yeast-tryptone (2
!YT) (1 L) contained Bactoª tryptone (16 g), Difcoª yeast 
extract (10 g), and NaCl (5 g).  Modified SOB medium (1 L) contained Bactoª tryptone (20 g), 
Difcoª yeast extract (5 g), and NaCl (5 g).  These modifications deviated from the standard 
recipe
9 by omitting KCl (2.5 mM, 0.186 g/L), MgCl
2 (10 mM, 0.952 g/L), and MgSO
4 (10 mM, 
1.204 g/L).  Modified SOC medium was prepared from modified SOB medium (1 L) with the 
EcoR
IBgl
IIBamH
IXho
IAmpRlacI
qp15Aori
mRFP-1
pBbA1a-RFP
4.3 kb
Ptrc
%200additions of 250 mM KCl (10 mL), 2 M MgCl
2 (5 mL), and 1 M 
D-glucose (20 mL).  These 
modifications deviated from the standard recipe
9 by omitting MgSO
4 (10 mM, 1.204 g/L).
8  ATCC medium 1416 4-hydroxybenzoic acid medium, pH 7.2 (1 L) contained 4-hydroxybenzoic 
acid (1 g), K
2HPO403H2O (4.25 g), NaH
2PO40H2O (1 g), NH
4Cl (2 g), nitrilotriacetic acid (0.1 g, 
adjusted to alkaline pH with concentrated NaOH), MgSO
407H2O (0.2 g), FeSO
407H2O (0.012 g), 
MnSO
40H2O (0.003 g), ZnSO
407H2O (0.003 g), and CoSO
4 (0.001 g).  M9 salts
9 (1 L) contained 
Na2HPO4 (6 g), KH
2PO4 (3 g), NH
4Cl (1 g), and NaCl (0.5 g).  Modified M9 media
8 was 
prepared by adding 
D-glucose (10 g), MgSO
4 (0.24 g), and thiamine hydrochloride (0.001 g) to 1 
L M9 salts.  These modifications increased the percentage of 
D-glucose from 0.4% to 1% (w/v), 
increased the amount of MgSO
4 by two-fold, and omitted 0.1 mM CaCl
2 from the recipe 
described by Sambrook and Russell.
9 Where appropriate, modified M9 media (1 L) was 
supplemented with 
L-phenylalanine (0.040 g), 
L-tyrosine (0.040 g), 
L-tryptophan (0.040 g), 4-
hydroxybenzoic acid (0.010 g), 4-aminobenzoic acid (0.010 g), and 2,3-dihydroxybenzoic acid 
(0.010 g).  L-Serine was added to a final concentration of 40 mg/L where indicated.  Antibiotics 
were also added where required to the following concentrations: ampicillin (Ap), 50 
$
g/mL; 
kanamycin (Kan), 50 
$
g/mL;  chloramphenicol (Cm), 20 
$
g/mL (in 95% EtOH); and 
tetracycline (Tc), 12.5 
$
g/mL (in 1:1 H
2O:EtOH).  IPTG was added to culture medium of strains 
carrying inducible promoters: 
Ptrc
, PT7, PT5.  Solutions of LB media, M9 salts, MgSO
4 (1 M), 
D-glucose (20% w/v), initial glucose (18-30 g/L), feed glucose (60% w/v), 
D-fructose (20% w/v), 
mixed 
D-glucose/
D-fructose feed (60% w/v, 1:1 glucose:fructose), modified SOB media, KCl 
(250 mM), and MgCl
2 (1 M) were sterilized by autoclaving.  Solutions of aromatic vitamins 
(containing 4-hydroxybenzoic acid, 4-aminobenzoic acid, and 2,3-dihydroxybenzoic acid), 
%201aromatic amino acids (containing 
L-phenylalanine, 
L-tyrosine, and 
L-tryptophan), 
L-serine, 
antibiotics, and IPTG were sterilized by filtration through 0.22 
$
m membrane.  Solid media was 
prepared by adding 1.5% (w/v) Bactoª agar to liquid media. 
 The standard 
E. coli
 fermentation medium
10 (1 L) contained K
2HPO4 (7.5 g), anhydrous 
citric acid (1.92 g), ammonium iron (III) citrate (0.3 g), conc. H
2SO4 (1.2 mL), and concentrated 
NH4OH (4.1 mL).  Where appropriate, fermentation media was supplemented with 
L-phenylalanine (0.7 g), 
L-tyrosine (0.7 g), and 
L-tryptophan (0.35 g).  The following supplements 
were added to the medium immediately prior to inoculation of the vessel and initiation of 
fermentation: 
D-glucose (18 g, 20 g, or 30 g), MgSO
4 (0.24 g), and trace minerals including 
(NH
4)6(Mo
7O24)04H2O (0.0037 g), ZnSO
407H2O (0.0029 g), H
3BO3 (0.0247 g), CuSO
405H2O (0.0025 g), and MnCl
204H2O (0.0158 g).  For fermentations requiring aromatic vitamin 
supplementation, 4-hydroxybenzoic acid (0.01 g), 2,3-dihydroxybenzoic acid (0.01 g), and 4-
aminobenzoic acid (0.01 g) were added alongside initial glucose supplements. 
Fermentation conditions 
General 
 Fermentations were conducted in a B. Braun BIOSTAT¨ B-DCU system controlled by a 
DCU tower with 2 L working capacity fermentor vessel.  Data was acquired on a Digilink 
personal computer utilizing MFCS/Win 3.0 software.  PID control loops were used to maintain 
temperature (33¡C) and pH (7.0 ± 0.1).  pH was maintained via addition of 2 N H
2SO4 and 
concentrated NH
4OH.  Glucose was added as a 60% (w/v) solution using a Matson Marlow 
101U/R peristaltic pump.  A mixed carbohydrate feed consisting of equimolar amounts of 
D-%202glucose (30% w/v) and 
D-fructose (30% w/v) was similarly added as a 60% (w/v) aqueous 
solution.  Dissolved oxygen (D.O.) was monitored using a Hamilton OxyFerm FDA 225 
sterilizable O
2 sensor fitted with an Optiflow FDA O
2-permeable membrane.  pH was measured 
using a Hamilton EasyFerm Plus K8 200 sterilizable probe.  Glucose concentrations were 
measured with a GlucCellª meter and test strip (CESCO Bioengineering Co., Ltd.).  Antifoam 
204 (Sigma-Aldrich) was added manually as needed to mitigate foam accumulation. 
 Growth of an inoculant was initiated by introduction of a fresh, single colony into a 
culture tube (18 mm 
! 150 mm) containing 5 mL modified M9 medium.  Cultures were grown 
with agitation at 250 rpm for 18-24 h.  2-Hydroxymuconic acid (2-HMA)-producing strains were 
cultured at 37¡C while kanosamine-producing strains were cultured at 35¡C.  An appropriate 
amount of the 5 mL culture was used to inoculate 100 mL M9 medium in a 500 mL three-baffled 
Kimbleª KIMAXª culture flask to an initial OD
600 = 0.05.  The 100 mL cultures were 
incubated with agitation at 37¡C, 250 rpm for 2-HMA-synthesizing strains or 35¡C, 250 rpm for 
kanosamine-producing strains, for 6-8 h.  After an appropriate OD
600 was reached (1.0-1.5) the 
culture was transferred into the fermentation vessel.  The initial glucose concentrations in the 
fermentation medium was 18-20 g/L for glucose-limited conditions and 30 g/L for glucose-rich 
conditions.   
Glucose-rich fermentations 
 For fermentations which employed glucose-rich conditions, a stainless steel baffle cage 
containing four 1/4
1
 ! 41
 baffles was fitted to the fermentation vessel.  Fermentation runs used 
three staged methods to maintain D.O. concentrations at 10% air saturation.  Mass flow 
%203controller settings for the first stage are presented in Figure 108 while settings for the second 
stage are presented in Figure 109.  With the airflow at an initial setting of 0.06 L/L/min, the D.O. 
was maintained by increasing the impeller speed from an initial set point of 50 rpm to a preset 
maximum of 750 rpm.  With the impeller rate constant at 750 rpm, the mass flow controller then 
maintained D.O. concentration by increasing airflow from its initial setting to a preset maximum 
of 1.0 L/L/minutes. After the preset maxima of 750 rpm and 1.0 L/L/min were reached, the third 
stage of the fermentation was entered in which glucose (60% w/v) was added at a rate sufficient 
to maintain 20-30 g/L 
D-glucose concentration throughout the run.  During this stage, the 
impeller speed was allowed to vary between 400-1800 rpm in order to maintain D.O. 
concentration at 10% air saturation.   
Figure 108. 
 Mass flow controller settings for first stage of glucose-rich fermentations.  Stir settings: 2.5% (50 rpm), 
37.5% (750 rpm).  Airflow settings: 2.0% (0.06 L/L/min), 33.3% (1.0 L/L/min). 
Deadband 0.5% 
saturation
CASCADE1STIR
2AIRFLOW
Parameter
Minimum
Maximum
%P&I&DHysterese
Mode
STIR
2.5%37.5%150.0%100.0 s0.0 s01:00 m:s
auto
AIRFLOW
2.0%33.3%90.0%50.0 s0.0 s01:00 m:s
auto
%204Figure 109. 
 Mass flow controller settings for second stage of glucose-rich fermentations.  Stir settings: 37.5% (750 
rpm), 90.0% (1,800 rpm).  Airflow settings: 2.0% (0.06 L/L/min), 33.3% (1.0 L/L/min). 
Glucose-limited fermentations  
 All glucose-limited fermentations did not utilize a fermentation vessel outfitted with a 
stainless steel baffle cage.  During a fermentation, the following three stages were used to 
maintain D.O. concentration at 10% air saturation.  With the airflow at an initial set point of 0.06 
L/L/min, the D.O. concentration was maintained by varying the impeller speed from an initial 
setting of 50 rpm to a preset maximum of 850 rpm.  Once the preset maximum was reached, the 
airflow was allowed to increase from its initial setting to a preset maximum of 1.0 L/L/minutes. 
At a constant impeller speed of 850 rpm and airflow rate of 1.0 L/L/min, the D.O. concentration 
was maintained at 10% air saturation for the remainder of the run by oxygen-sensor-controlled 
glucose feeding.  PID control parameters for oxygen-sensor-controlled glucose feeding were set 
to 0.0 (off) for the derivative control (
#D) and 999.9 s (minimum control action) for integral 
control (
#I).  The proportional response (
Xp) was set to 950.0% to achieve a proportional gain 
(Kc) of 0.1.  Mass flow controller settings for stir speed and airflow are shown in Figure 110. 
Deadband 0.5% 
saturation
CASCADE1AIRFLOW
2STIR
Parameter
Minimum
Maximum
%P&I&DHysterese
Mode
STIR
37.5%90.0%150.0%100.0 s0.0 s01:00 m:s
auto
AIRFLOW
33.3%33.3%90.0%50.0 s0.0 s01:00 m:s
auto
%205Figure 110. 
 Mass flow controller settings for second stage of glucose-rich fermentations.  Stir settings: 2.5% (50 
rpm), 42.5% (850 rpm).  Airflow settings: 2.0% (0.06 L/L/min), 33.3% (1.0 L/L/min). 
General procedure for processing fermentation broth 
 Cells were removed from fermentation broth by centrifugation (6,000
g, 20 min, 4¡C).  
Proteins were removed by filtration through a Sartocon¨ ECO Hydrostart¨ membrane 10 kD 
cut-off cassette using a SartoJet¨ 4-piston Diaphragm Pump.  Tangential flow filtration was 
performed with 25 ± 5 psig toward the membrane and a 7 ± 1 psig back-pressure.  When 
approximately 200 mL broth remained unfiltered, the volume was insufficient to sustain flow and 
the filtration ceased. 
Analysis of fermentation cultures 
 Samples (5-15 mL) were removed at indicated intervals and cell densities of 
E. coli
 were 
determined by dilution of an aliquot of fermentation broth with water (1:50, 1:100, or 1:200; 1 
mL total volume of diluted sample) followed by measurement of absorption at 600 nm.  Dry cell 
weight for 
E. coli
 was obtained using a conversion coefficient of 0.43 g/L/OD
600.  The remaining 
Deadband 0.5% 
saturation
CASCADE1STIR
2AIRFLOW
Parameter
Minimum
Maximum
%P&I&DHysterese
Mode
STIR
2.5%42.5%175.0%100.0 s0.0 s01:00 m:s
auto
AIRFLOW
2.0%33.3%90.0%50.0 s0.0 s01:00 m:s
auto
%206fermentation broth was centrifuged (11,300 
g, 5 min, rt for 5 mL samples or 7,600 
g, 15 min, 4¡C 
for 15 mL samples) and the cell-free supernatant was used to determine solute concentrations.  
Cell pellets from 15 mL samples were frozen at 
*80¡C for later use. 
 Solute concentrations in cell-free broth were determined by 
1H NMR with application of 
formulae described above.  Kanosamine containing solutions were concentrated to dryness under 
reduced pressure, concentrated to dryness one additional time from D
2O, and redissolved in D
2O containing a known quantity of the sodium salt of 3-(trimethylsilyl)propionic-
2,2,3,3-d4-acid 
(TSP).  Solutions containing 2-hydroxymuconic acid were acidified to pH 4 with 10% (v/v) 
H2SO4 and concentrated to dryness under reduced pressure.  The yellow oil was redissolved in 
CD3OD containing a known amount of maleic acid. 
Genetic manipulations 
General methods 
 E. coli
 DH5
!
 was used as the host strain for plasmid manipulations.  Restriction 
enzymes, T4 DNA ligase, and Q5 DNA polymerase were purchased from New England BioLabs, 
Inc. (Ipswich, MA).  Agarose (electrophoresis grade) was purchased from Invitrogen and dNTPs 
were purchased from Promega.  PCR amplifications were performed following the recommended 
protocol as described by New England BioLabs (NEB).  Each reaction (50 
$
L) contained 5
! Q5 
reaction buffer (10 
$
L), dNTPs (0.2 mM), forward and reverse primers (0.5 
$
M), template DNA 
(100-500 ng), and 1 unit of Q5 DNA polymerase.  Primers were synthesized by Integrated DNA 
Technologies (Coralville, IA).  The PCR reaction was performed using a Bio-Rad DNA Engine¨ 
Peltier Thermal Cycler. 
%207 PCR products and digested DNA fragments were isolated and purified from 0.7% (w/v) 
agarose gel (TAE buffer, pH 8.3) using the Zymocleanª Gel DNA Recovery kit (Zymo 
Research) or the Monarch¨ DNA Gel Extraction kit (NEB) following manufacturerÕs protocol.  
TAE buffer contained trishydroxymethylaminomethane (Tris base,  40 mM), HOAc (20 mM), 
and sodium ethylenediaminetetraacetate (EDTA, 50 mM).  Ligation reaction mixtures were 
purified prior to transformation using the Zymo Research DNA Clean & Concentratorª-25 kit 
or the Monarch¨ PCR & DNA Cleanup kit (NEB) following manufacturerÕs protocol.  DNA 
concentration was determined by measuring the absorbance at 260 nm.  An absorbance of 1.0 at 
260 nm corresponds to 50 
$
g/
$
L of double-stranded DNA.  Protein contamination was estimated 
by measuring the absorbance of DNA sample at 280 nm.  A ratio of A
260:A280 of 1.8-2.0 indicates 
pure DNA.   
Preparation and transformation of electrocompetent 
E. coli
  Electrocompetent cells were prepared using modified procedures from Sambrook and 
Russell.
9  A single colony was used to inoculate LB media (5 mL).  The liquid culture was 
incubated overnight at 37¡C, 250 rpm.  An appropriate amount of overnight culture was used to 
inoculate a 1 L six-baffled Kimbleª KIMAXª culture flask containing 200 mL 2
!YT media to 
an initial OD
600 = 0.015.  Cells were cultured at 37¡C, 250 rpm for 2-3 h until mid-exponential 
growth phase (OD
600 = 0.5-0.7) at which point the flask containing the cell culture was chilled on 
ice for 30 minutes. Cells were harvested by centrifugation (5 min, 2,800
g, 4¡C) and the 
supernatant was discarded.  All subsequent manipulations were carried out on ice.  The cell pellet 
was resuspended in 200 mL sterile ice-cold water and harvested by centrifugation (10 min, 
%2082,800g, 4¡C).  The supernatant was discarded and cells resuspended in 100 mL sterile ice-cold 
water.  Cells were harvested again by centrifugation (10 min, 2,800
g, 4¡C).  The supernatant was 
discarded and the cells were resuspended in 100 mL sterile, ice-cold 10% (v/v) glycerol.  
Resuspended cells were chilled on ice for 30 min prior to centrifugation (10 min, 2,800
g, 4¡C).  
The supernatant was discarded and the cells were resuspended in 0.5 mL sterile, ice-cold 10% (v/
v) aqueous glycerol solution.  Aliquots (50 
$
L) were dispensed into sterile, ice-cold 
microcentrifuge tubes and flash-frozen in liquid nitrogen.  Frozen electrocompetent cells were 
stored at 
*80¡C. 
 Plasmid DNA (dissolved in sterile water, 1-3 
$
L; 10-100 ng) or purified, de-salted DNA 
ligation reaction solutions (3 
$
L) was mixed with 50 
$
L of thawed electrocompetent cells.  The 
mixture of cells and DNA was transferred to a Bio-Rad Gene Pulser cuvette (0.2 cm electrode 
gap).  Electroporation was performed using a Bio-Rad Gene Pulser II electroporator set to 2.5 
kV, 25 
$
F, and 200 
2
.  A single pulse was applied across the cuvette containing cells and DNA 
which resulted in a time constant of 5.16-5.30 ms.  SOC medium (950 
$
L) was added into the 
cuvette and the cell-containing mixture was transferred into a sterile glass culture tube (20 
! 150 mm).  Cells were cultured in a rotary shaker (37¡C, 250 rpm) for 1 h and plated onto appropriate 
medium.  Before plating onto M9 medium, cells were washed with M9 salts (1 mL, 3
!).  Washed, pelleted cells were resuspended in M9 salts before plating. 
Small-scale purification of plasmid DNA 
 A single colony of freshly transformed 
E. coli
 was used to inoculate 5 mL LB media 
containing the appropriate antibiotic.  Inoculated cultures were incubated with shaking at 37¡C, 
%209250 rpm overnight.  Cells were harvested from 3 mL of stationary-phase cultures by 
microcentrifugation (2,400
g, 3 min, rt).  Isolation and purification of plasmid DNA used the 
Qiagen QIAprep¨ Spin Miniprep kit following the manufacturerÕs protocol.  Plasmid DNA was 
eluted from the spin column with 50 
$
L sterile water.  
Large-scale purification of plasmid DNA 
 A single, fresh colony of 
E. coli
 was used to inoculate 100 mL LB media containing the 
appropriate antibiotic in a 500 mL three-baffled Kimbleª KIMAXª culture flask.  Inoculated 
cultures were incubated with shaking at 37¡C, 250 rpm overnight.  Cells were harvested by 
centrifugation (2,800
g, 5 min, 4¡C).  Isolation and purification of plasmid DNA used the Qiagen 
Plasmid Midiprep kit following the manufacturerÕs protocol.  The purified air-dried plasmid 
DNA pellet was dissolved in 200 
$
L sterile water and stored at 4¡C. 
Restriction enzyme digestion of DNA 
 A typical restriction enzyme DNA digest (20 
$
L) contained 2 
$
L restriction enzyme 
buffer (10
! concentration), 5-10 
$
L purified DNA stock (0.1-1.0 
$
g), 1 $
L of restriction 
enzyme, and 7-13 
$
L sterile water.  Double-digestions used 1 
$
L each of the required restriction 
enzymes.  When sequential restriction digests involved enzymes requiring buffers of varying salt 
concentrations, the first digest was performed in the low-salt buffer followed by addition of 1 
$
L 
1.7 M NaCl and 1 
$
L of high-salt restriction enzyme before the second digest.  Reactions were 
incubated a 37¡C for 1 h.  When pJG7.246 was used as the vector, 1 unit of calf intestinal 
alkaline phosphatase (CIAP) was added following 
Bam
HI digestion and the reaction mixture 
%210incubated for an additional hour at 37¡C.  Digests were terminated by addition of 4 
$
L DNA gel 
loading dye containing SDS (6
! concentration, NEB) and subsequently analyzed by agarose gel 
electrophoresis. 
Ligation of DNA 
 Most DNA ligations were designed so that the molar ratio of insert DNA to vector DNA 
was 3:1.  When inserting a fragment into pJG7.246, however, the molar ratio of insert DNA to 
vector DNA was increased to 10:1.  A typical ligation reaction contained 0.01-0.02 pmol of 
vector DNA and 0.03-0.1 pmol insert DNA.  This mixture was brought to 17 
$
L using sterile 
water before adding 2 
$
L 10
! T4 ligase buffer and 1 
$
L of T4 DNA ligase (2 units).  Ligations 
were conducted for 2 h at rt.  Reactions were terminated by incubation at 65¡C for 10 min before 
purifying the reaction mixture with a Zymo Research DNA Clean & Concentratorª-25 kit or the 
Monarch¨ PCR & DNA Cleanup kit (NEB) following the manufacturerÕs protocol.  Purified 
ligation mixtures were eluted from the spin column in 6 
$
L sterile water. 
Agarose gel electrophoresis 
 Agarose gel electrophoresis was performed in TAE buffer containing 
trishydroxymethylaminomethane (Tris base,  40 mM), HOAc (20 mM), and sodium 
ethylenediaminetetraacetate (EDTA, 50 mM).  A constant voltage of 97 V was applied for 1 h.  
Gels typically contained 0.7% (w/v) agarose in TAE buffer.  Lower concentrations of agarose 
(0.35 %) were used to resolve genomic DNA.  Ethidium bromide (0.5 
$
g/mL) was added to the 
agarose to allow visualization of DNA fragments under a UV lamp.  The size of the DNA 
%211
fragments were determined by comparison to DNA fragments present in 10 
$
L 1 kb DNA ladder 
(NEB, 500-10,000 bp).   
Polyacrylamide gel electrophoresis (SDS-PAGE) 
 Polyacrylamide gels were run under denaturing conditions in SDS-PAGE buffer (pH 8.3) 
containing sodium dodecyl sulfate (SDS, 0.1% w/v), 
L-glycine (192 mM), and Tris base (25 
mM).  Protein analysis used 4-20% (w/v) acrylamide Mini-PROTEAN¨ TGXª precast gels 
purchased from Bio-Rad Laboratories, Inc.  Each protein sample (60 
$
L) was diluted with 
Laemmli sample buffer (60 
$
L) consisting of 4% (w/v) SDS, 20% (v/v) glycerol, dithiothreitol 
(100 mM), 0.004% (w/v) bromophenol blue, and 125 mM Tris-HCl (pH 6.8).  Diluted protein 
samples were incubated at 90-95¡C for 5 min and microcentrifuged (11,300 
g, 2 min, rt).  
Samples (30 
$
L) and Precision Plus Proteinª All Blue ladder (Bio-Rad) were loaded into 
sample wells and the gel was run under constant voltage at 200 V.  When the blue tracking dye 
reached the bottom of the gel, the gels were removed from the cassettes and stained in a solution 
containing 0.1% (w/v) Coomassie Brilliant Blue R, 45% (v/v) MeOH, and 10% (v/v) HOAc in 
H2O for at least 2 h.  Gels were destained in a solution of 45% (v/v) MeOH and 10% (v/v) HOAc 
in H
2O. 
%212CHAPTER ONE 
Synthetic procedures 
Diethyl 2-hydroxyhexa-2,4-dienedioate 
22  A 500 mL three-neck flask equipped with a thermometer was charged with 
t-butanol (39 
mL, 420 mmol) which was stirred under a nitrogen atmosphere.  Finely sliced potassium metal 
(3.9 g, 100 mmol) was added gradually.  The mixture was allowed to stir at rt for 1 h after which 
dry toluene (30 mL, 284 mmol) was added via syringe.  An ice bath was used to cool the reaction 
mixture to 0¡C where it was maintained for 30 min before adding a solution of diethyl oxalate 
(13.5 mL, 100 mmol) in dry toluene (8 mL, 76 mmol).  Diethyl oxalate was added slowly to 
ensure the temperature was maintained between 0-5¡C during which time the solution developed 
a bright yellow-orange color.  Upon complete addition of diethyl oxalate, the solution was stirred 
at 0¡C for 30 min before a solution of ethyl crotonate (12.4 mL, 100 mmol) in dry toluene (8 mL, 
76 mmol) was added via a syringe.  The ethyl crotonate solution was similarly added to ensure 
the temperature was maintained between 0-5¡C.  The resultant red-orange solution was stirred at 
0¡C for 1 h before the ice bath was removed and the reaction mixture allowed to warm gradually 
to rt.  A mustard yellow precipitate was observed beginning 1 h after addition of ethyl crotonate.  
The reaction mixture was allowed to stir at rt for 15 h after which time no further formation of 
yellow precipitate was observed.  
  The mustard-yellow solids were dissolved in 5% HOAc (250 mL) to afford a red 
solution.  Dilute H
2SO4 was used to acidify the solution to pH 5, whereupon crude diethyl 2-
hydroxyhexa-2,4-dienedioate 
22 was extracted into EtOAc (3 x 250 mL).  The organic extracts 
%213were pooled and dried with anhydrous MgSO
4 before concentrating under reduced pressure.  The 
pale yellow solids were dried overnight under reduced pressure at rt to afford diethyl 2-
hydroxyhexa-2,4-dienedioate 
22 (15.2 g, 71 mmol, 71% yield). 
1H NMR (500 MHz, CD
3OD) 
-
 7.72 (dd, J = 15.6, 11.7 Hz, 1H), 6.26 (dd, 
J = 11.7, 0.9 Hz, 1H), 6.01 (dd, 
J = 15.5, 0.9 Hz, 1H), 
4.29 (q, J = 7.1 Hz, 2H), 4.19 (q, 
J = 7.1 Hz, 2H), 1.33 (t, 
J = 7.1 Hz, 3H), 1.29 (t, 
J = 7.1 Hz, 
3H).  
13C NMR (126 MHz, CD
3OD) 
-
 168.7, 165.1, 147.9, 139.1, 122.1, 109.3, 63.1, 61.5, 14.6, 14.4. 6-Carboxy-2-pyrone 
19  Diethyl 2-hydroxyhexa-2,4-dienedioate 
22 (3.2 g, 15 mmol) was dissolved in an aqueous 
solution (17.5 mL total volume) of conc. H
2SO4 (2.5 mL, 47 mmol) and glacial HOAc (10 mL, 
175 mmol) in a 50 mL round-bottom flask equipped with a thermometer and reflux condenser.  
The bright yellow solution was heated to reflux for 3 h during which time a red-brown color 
developed.  After allowing to cool to rt, the acidic solution was neutralized with NaOH and 
extracted with EtOAc (3 x 20 mL).  The organic extracts were pooled and dried with anhydrous 
MgSO
4 before concentrating under reduced pressure.  After drying overnight under reduced 
pressure at rt, 6-carboxy-2-pyrone 
19 was obtained as light brown solids (1.8 g, 12.8 mmol, 85% 
yield).  
1H NMR (500 MHz, D
2O) 
-
 7.73 (dd, J = 9.4, 6.7 Hz, 1H), 7.27 (dd, 
J = 6.7, 1.0 Hz, 
1H), 6.66 (dd, 
J = 9.4, 1.0 Hz, 1H).  
13C NMR (126 MHz, DMSO-
d6) -
 160.4, 160.0, 149.5, 143.6, 120.0, 110.1.  
%2142-Hydroxymuconic acid  
 A rust-colored solution of 6-carboxy-2-pyrone 
19 (0.5 g, 3.6 mmol) dissolved in 2 M 
NaOH (18 mL) was stirred at rt for 12 h in a round bottom flask.  The solution was transferred to 
a beaker on ice and chilled to 0¡C.  Acidification of the ice-cold solution to pH 2 with conc. 
H2SO4 resulted in the formation of a pale-yellow precipitate.  Solids were removed by filtration 
and dried under reduced pressure to afford 2-hydroxymuconic acid (2-HMA, 0.32 g, 64% yield).  
An alternative method of preparation utilized a solution of 2-hydroxyhexa-2,4-dienedioate 
22 (7.0 g, 33 mmol) in 2 M NaOH (200 mL) following the described procedure resulting in 2-HMA 
in 59% isolated yield (2.77 g, 19.5 mmol).  
1H NMR (500 MHz, CD
3OD) 
-
 7.72 (dd, J = 15.6, 11.8 Hz, 1H), 6.26 (d, 
J = 11.7 Hz, 1H), 5.96 (d, 
J = 15.5 Hz, 1H).
 13C NMR (126 MHz, DMSO-
d6) -
 167.7, 165.0, 147.2, 137.5, 121.8, 107.9.  Derivatization of 2-hydroxymuconic acid to 2-hydroxydimethylmuconate 
 A solution of 2-hydroxymuconic acid (10 mg, 0.06 mmol) in MeOH (0.2 mL) was stirred 
in a pear-shaped flask at rt.  Trimethylsilyldiazomethane (TMSD, 2 M in diethyl ether, 0.3 mL) 
was added via syringe and the mixture was stirred at rt for 1 h.  Upon completion of the reaction, 
solvent was removed by passing a stream of nitrogen over mouth of reaction vessel.  The oily 
residue was dissolved in a solvent mixture of EtOAc:hexanes (1:4) and purified by silica gel 
chromatography.  Fractions containing 2-hydroxydimethylmuconate were pooled and used for 
nuclear Overhauser effect experiments.  
1H NMR (500 MHz, DMSO-
d6) -
 7.61 (dd, J = 15.4, 11.8 Hz, 1H), 6.29 (d, 
J = 11.8 Hz, 1H), 6.13 (d, 
J = 15.5 Hz, 1H), 0.95 (s, 3H), 0.94 (s, 3H). 
%215General procedure for cycloaddition of 2-hydroxymuconic acid with ethylene 
2  Cycloadditions of 2-HMA with ethylene were conducted in a Series 4560 Titanium Mini-
reactor Stirred Vessel (Parr Instrument Co.).  Solvent, catalyst (where indicated), and 2-HMA 
were added to the reactor vessel in a glove bag under a positive nitrogen atmosphere.  The 
reactor head was attached and secured by tightening the collar bolts finger-tight in the glove bag.  
The sealed reactor vessel was brought to a fume hood and secured to the instrument mount.  
Bolts were tightened further with a torque wrench to 15 psi.  While stirring at low speed (50 
rpm), the reactor vessel was flushed with ethylene gas (3x) before pressurizing with ethylene gas 
to the reaction pressure (500 psig unless otherwise indicated).  A heating mantle was lifted into 
place surrounding the reactor vessel and the impeller speed was increased to 200 rpm.  
Approximately 1.5 h was required to reach reaction temperature (t = 0) during which time the 
internal reactor pressure rose by 350 psig.  An additional 100 psig increase of pressure was 
observed during the course of the reaction.  To terminate the reaction, the heating mantle was 
removed and the reactor was allowed to cool to rt.  Approximately 2.5 h were required to cool 
reactor completely.  Pressure was released by slowly opening the vent valve and the reactor head 
was removed in air.  The reactor chamber, impeller, and thermoprobe were rinsed with acetone 
followed by copious amounts (200-300 mL) of 
N,N-dimethylacetamide (DMA).  Acetone and 
DMA solutions were analyzed separately by NMR and HPLC.  
General procedure for cycloaddition of propiolic acid 
18 and 6-carboxy-2-pyrone 
19  A three-neck flask equipped with a stir bar, thermometer, and jacketed condenser was 
charged with solvent (5 mL) and propiolic acid (0.9, 14 mmol).  Catalyst (20 mol%) was added 
%216and the mixture stirred at rt for 2 minutes. 6-Carboxy-2-pyrone (0.5 g, 3.6 mmol) was added and 
the reaction mixture was heated to reflux.  After 12 h, the reaction was cooled to rt and quenched 
with H
2O.  The solvent was removed under reduced pressure and the crude reaction material 
analyzed by NMR and HPLC.     
Conversion of 2-hydroxymuconic acid isomer mixture to 6-carboxy-2-pyrone 
19 in fermentation 
broth 
 Cell-free, protein-free fermentation broth containing 2-HMA, 
!
-keto acid 
23 and 
!
-keto 
acid 
24 (700 mL) was combined with HOAc (560 mL) and conc. H
2SO4 (136 mL) in a 2 L three-
neck flask equipped with stir bar, thermometer, and jacketed condenser.  The solution was stirred 
under a nitrogen atmosphere at 95¡C for 5 h.  Upon cooling to rt, the dark brown reaction crude 
was concentrated under reduced pressure and analyzed by 
1H NMR.     
D-Erythrose-4-phosphate 
 Synthesis of 
D-erythrose-4-phosphate followed procedures described by Sieben.
11  Glacial HOAc (500 mL) was stirred in a 1 L three-neck flask under a stream of nitrogen for 15 
min at rt.  A solution of 
D-glucose-6-phosphate, disodium salt hydrate (0.56 g, 2 mmol) in 4 mL 
H2O was added via syringe.  Dropwise addition of a suspension of Pb(OAc)
4 (1.51 g, 3.4 mmol) 
in HOAc (60 mL) with 6 N H
2SO4 (1.2 mL) employed an addition funnel affixed to the 1 L 
three-neck flask.  Complete addition of Pb(OAc)
4 required 1.5 h during which time a white 
precipitate developed in the reaction flask.  The precipitate was removed by filtration through 
Celite¨ 545.  The clarified solution was concentrated to 1/5 volume under reduced pressure and 
%217the excess HOAc was removed by azeotropic distillation with H
2O (5
!).  The resulting solution 
was applied to a cation exchange resin at rt (Dowex¨ 50WX8, 1.5 in 
! 6 in, 50 mL, hydrogen 
form) and eluted with H
2O (300 mL).  
D-Erythrose 4-phosphate (E4P) was quantified using a 
transketolase coupled enzyme assay described below. 
Isolation of 
Paenibacillus
 sp. JJ-1b (ATCC 35889) genomic DNA 
 Lyophilized 
Paenibacillus 
sp. JJ-1b (ATCC 35889) was rehydrated with 250 
$
L of ATCC 
#1416 media and streaked onto ATCC #1416 agar plates.  Cells were grown at 30¡C for 36 h.  A 
single colony was used to inoculate ATCC #1416 media (5 mL) and the culture was incubated at 
30¡C, 250 rpm overnight.  Genomic DNA was isolated from growing cells following a procedure 
described by Wecke.
12  A 250 mL three-baffle flask containing 50 mL ATCC #1416 media was 
inoculated with 1 mL overnight culture.  Cells were incubated at 30¡C, 250 rpm for 16 h to a 
final OD
600 = 1.3.  Cells were harvested by centrifugation (5 min, 2,000 
g, 4¡C) and the 
supernatant was discarded.  The cell pellet was suspended in 400 
$
L TEN buffer (10 mM Tris-
HCl, pH 8.0, 10 mM EDTA, 150 mM NaCl) and transferred into a microcentrifuge tube.  
Lysozyme (0.4 mg) was added and the mixture was incubated for 20 min at 37¡C.  RNase A 
(0.04 mg) was added and the mixture was incubated for 3 min at 65¡C.  SDS (0.004 mg) and 
proteinase K (0.15 mg) was added along with 550 
$
L TEN* buffer (10 mM Tris-HCl, pH 8.0, 1 
mM EDTA, 50 mM NaCl) and the mixture was incubated for 2 h at 60¡C.  Tris-buffered phenol 
(450 $
L) was added and the microcentrifuge tube vortexed briefly to mix.  Organic and aqueous 
layers were separated by microcentrifugation (5 min, 11,300 
g, rt).  The aqueous layer was 
decanted into a new microcentrifuge tube and phenol extraction was repeated.  After 
%218microcentrifugation (5 min, 11,300 
g, rt) phenol was discarded and the aqueous layer was 
washed with chloroform:isoamyl alcohol (24:1, 900 
$
L).  The aqueous phase was then 
transferred into a test tube containing 95% EtOH (10 mL, 
*20 ¡C).  Precipitated DNA was 
collected on the tip of a Pasteur pipette and dried in air at rt for 5 minutes. Genomic DNA was 
then dissolved in 100 
$
L sterile water.  The DNA concentration was 1.4 
$
g/
$
L based on the 
absorbance at 260 nm.   
Bacterial strains and plasmids 
Strain generation: 
E. coli
 KM1 
 E. coli
 KM1 was prepared by inactivation of genomic 
pptA
 from 
E. coli 
WN1 using 
"
 Red mediated homologous recombination following the procedure of Wanner.
4  Plasmid pKD46 
allows cell growth in the presence of ampicillin and carries a temperature-sensitive replicon and 
the 
"
 Red recombinase genes behind an arabinose-inducible promoter.  This plasmid was 
transformed into WN1 and electrocompetent WN1/pKD46 cells were prepared as described 
above.  A kanamycin resistance gene (Kan
R) flanked by flippase recognition targets (FRT) was 
PCR-amplified from plasmid pKD4 using Q5 high-fidelity DNA polymerase.  The FRT-flanked 
KanR PCR product was treated with DpnI to remove template DNA and purified using the Zymo 
Research DNA Clean & Concentratorª-25 kit.  Purified PCR product was transformed into 
electrocompetent WN1/pKD46.  Cells were plated onto LB/Kan and grown at 37¡C for 24 h, 
during which time homologous recombination events led to insertion the FRT-flanked Kan
R cassette into the genomic 
pptA
 locus.  Transformants were re-streaked onto LB agar and grown at 
37¡C for 24 h.  Single colonies from these LB plates were replicate plated onto LB, LB/Ap, and 
%219LB/Kan plates.  Colonies that grew in the presence of kanamycin and ampicillin were selected to 
prepare electrocompetent cells, WN1
pptA
::FRT-Kan
R/pKD46.  Plasmid pCP20, which carries a 
thermally-induced yeast Flp recombinase gene and a temperature-sensitive replicon, was 
transformed into electrocompetent WN1
pptA
::FRT-Kan
R/pKD46 and plated onto LB/Ap.  Cells 
were grown at 30¡C for 24 h.  Transformants were re-streaked onto LB agar before growth at 
43¡C.  Loss of antibiotic resistance was tested by replicate plating colonies from LB plates onto 
LB, LB/Kan, and LB/Cm.  An inability to grow in the presence of kanamycin, chloramphenicol, 
or ampicillin implied loss of FRT-flanked kanamycin resistance gene and 
"
 Red helper plasmids 
simultaneously.  Loss of genomic 
pptA
 was verified by PCR followed by agarose gel 
electrophoresis. 
BglBrick parts and plasmids 
 All plasmids used in this Chapter are presented in Table 23.  All ÒpartsÓ are genes 
inserted into pBbA1a (Figure 107).  All insertions followed the method of cloning described 
above (Figure 106).  The sequence for all BglBrick parts were confirmed by Sanger sequencing. 
%220Table 25.
  Plasmids used and constructed in Chapter 1 of this work. 
Plasmid pKM6.274 
 Plasmid pKM6.274 contained a 
Ptrc
aroF
FBR
serAParoF
 cassette.  It was created by first 
inserting 
serA from plasmid pKM6.240 upstream of 
ParoF
 in plasmid pKM6.241 to afford 
plasmid pKM6.261.  The 
aroF
FBR
 gene was then inserted upstream of 
serA to afford plasmid 
pKM6.274. 
Plasmid pKM6.232 
 Plasmid pKM6.232 contained a 
Ptrc
praBpraHpraA
 cassette.  It was created by first 
inserting 
praH
 from plasmid pKM6.201 upstream of 
praA 
in plasmid pKM6.196 to afford 
Strain/plasmid
Relevant characteristics
Source or reference
pBbA1a-RFP
ApR, lacI
q, Ptrc
rfp p15A replicon
Addgene #35334
6pWN1.028A
aroF
FBR
 source
lab2pWN1.162A
serA source
lab2pMF63A
ParoF
 source
lab13pKM6.196
Ptrc
praA
 in pBbA1a
this work
pKM6.200
Ptrc
praB
 in pBbA1a
this work
pKM6.201
Ptrc
praH 
in pBbA1a
this work
pKM6.208
Ptrc
praHpraA
 in pBbA1a
this work
pKM6.232
Ptrc
praBpraHpraA
 in pBbA1a
this work
pKM6.240
Ptrc
serA in pBbA1a
this work
pKM6.241
Ptrc
ParoF
 in pBbA1a
this work
pKM6.261
Ptrc
serAParoF
 in pBbA1a
this work
pKM6.268
Ptrc
aroF
FBR
 in pBbA1a
this work
pKM6.274
Ptrc
aroF
FBR
serAParoF
 in pBbA1a
this work
pKM6.280
Ptrc
praBpraHpraAaroF
FBR
serAParoF
 in pBbA1a
this work
%221plasmid pKM6.208.  The 
praB
 gene was then inserted upstream of 
praA
 to afford plasmid 
pKM6.232. 
Plasmid pKM6.280 
 Plasmid pKM6.280 contained a 
Ptrc
praBpraHpraAaroF
FBR
serAParoF
 cassette.  It was 
created by inserting the 
praBpraHpraA
 gene fragment from plasmid pKM6.232 upstream of 
aroF
FBR
 in plasmid pKM6.274. 
Enzyme assays 
Transketolase assay 
 Estimation of laboratory-synthesized E4P concentration employed the transketolase assay 
using the method of Paoletti.
14  A dilute solution of E4P was concentrated four-fold by rotary 
evaporation and the pH adjusted to 7.0 with 5 N KOH.  The assay mixture (1 mL) contained 150 
mM triethanolamine (pH 7.6), 5 mM MgCl
2, 0.1 mM thiamine pyrophosphate, 0.4 mM NADP
+, 0.4 mM 
#
-hydroxypyruvate, 50 
$
L E4P solution, glucose-6-phosphate dehydrogenase (8 units, 
Roche), and phosphoglucose isomerase (4 units, Sigma-Aldrich).  The solution was allowed to 
equilibrate at rt for 3-5 min during which time the unreacted 
D-glucose-6-phosphate remaining in 
solution from the synthesis of E4P had been consumed.  After no change in absorbance was 
observed, a solution containing purified 
E. coli
 transketolase (10 units, Sigma-Aldrich) was 
added and the increase in absorbance at 340 nm was monitored until the absorbance was no 
longer increasing (5 min). The difference in absorbance at 340 nm due to formation of NADH (
,
 = 6,220 M*1 cm
*1) was used to estimate the concentration of laboratory-synthesized E4P.  
%222DAHP synthase (AroF
FBR
) assay 
 DAHP synthase activity was assayed according to the procedure described by Schoner.
15  A dilute solution of E4P was concentrated to 12 mM under reduced pressure and neutralized with 
5 N KOH.  Two different solutions were prepared and incubated separately at 37¡C for 5 
minutes. The first solution (1 mL) contained E4P (6 mM), phosphoenolpyruvate (PEP, 12 mM), 
ovalbumin (1 mg/mL), and potassium phosphate buffer (25 mM, pH 7.0).  The second solution 
(0.5 mL) contained 1,3-propanediol (250 mM), PEP (0.5 mM), and an appropriate amount of 
protein sample diluted in potassium phosphate buffer (50 mM, pH 7.0).  After the two solutions 
were mixed (time = 0), aliquots (0.15 mL) were removed at timed intervals and quenched with 
0.1 mL trichloroacetic acid (10% w/v).  Precipitated protein was removed by microcentrifugation 
(11,300
g, 2 min, rt) and the amount of DAHP in each sample was determined using a 
thiobarbituric acid assay. 
  The reagents used in the thiobarbituric acid assay
16 include solution A: 0.2 M NaIO
4 in 
8.2 M H
3PO4; solution B: 0.8 M NaAsO
2 and 0.5 M Na
2SO4 in 0.1 M H
2SO4; and solution C: 
0.04 M thiobarbituric acid in 0.5 M Na
2SO4 (pH 7.0).  An aliquot (0.1 mL) of DAHP-containing 
sample was reacted with 0.1 mL of solution A in a glass test tube (13 mm 
! 100 mm) at 37¡C for 
5 minutes. The reaction was quenched by addition of 0.5 mL solution B and briefly vortexed 
until a brown color disappeared.  Upon addition of 3 mL solution C, the sample was heated at 
100¡C for 15 minutes. Samples were briefly cooled and the pink chromophore was extracted into 
4 mL of cyclohexanone.  The aqueous and organic layers were separated by centrifugation (2,000 
g, 15 min).  The absorbance of the organic layer was measured at 549 nm (
,
 = 68,000 M*1 cm
*1). %2232-Hydroxymuconic acid extraction solvent screening 
 Cell-free, protein-free fermentation broth containing 2-hydroxymuconic acid, 
!
-keto acid 
23 and 
!
-keto acid 
24 was concentrated to 1/4 original volume under reduced pressure and 
acidified to pH 2 with conc. H
2SO4.  Extraction screens were performed in 1 dram glass vials 
with screw caps.  Each vial contained 0.5 mL concentrated broth and 0.5 mL extraction solvent.  
The solvent and broth were mixed by continuously rotating (7 rpm) for 12 h at rt using a 
hybridization incubator.  The concentration of 2-HMA, 
!
-keto acid 
23 and 
!
-keto acid 
24 in the 
organic and aqueous layers were determined by 
1H NMR. 
%224CHAPTER TWO 
Plasmid construction 
BglBrick parts and plasmids 
 A list of the BglBrick parts used in this Chapter are presented in Table 26.  All ÒpartsÓ are 
genes inserted into pBbA1a (Figure 107).  All insertions followed the method of cloning 
described above (Figure 106).  The sequence for all BglBrick parts was confirmed by Sanger 
sequencing. 
Table 26.
  Plasmids used and constructed in Chapter 2 of this work. 
Strain/plasmid
Relevant characteristics
Source or reference
pBbA1a-RFP
ApR, lacI
q, Ptrc
rfp p15A replicon
Addgene #35334
6pBbA7k-RFP
KanR, lacI
q, PT7rfp, p15A replicon
Addgene #35318
6pKM6.240
Ptrc
serA in pBbA1a
Chapter 1
pSN1.294
Ptrc
ntdC
sub
lab (described 
herein)
pSN1.295
Ptrc
ntdB
sub
lab (described 
herein)
pSN1.296
Ptrc
ntdA
sub
lab (described 
herein)
pSN1.119
Ptrc
ntdC
sub
ntdA
sub 
in pBbA1a
lab (described 
herein)
pSN1.139
Ptrc
ntdC
sub
ntdA
sub
ntdB
sub
 in pBbA1a
lab (described 
herein)
pSN1.292
Ptrc
ntdC
sub
ntdA
sub
ntdB
sub
serA in pBbA1a
lab (described 
herein)
pJG7.246
ApR, PT5, His
6, lacI
q, ColE1 replicon
lab7pKM9.279C
PT5His6-ntdC
sub
 in pJG7.246
this work
pKM10.82A
PT7ntdC
sub
ntdA
sub
ntdB
sub
serA 
in pBbA7k
this work
pKM10.204A
Ptrc
ntdA
sub
serA in pBbA1a
this work
pKM10.204B
Ptrc
ntdB
sub
serA in pBbA1a
this work
%225Table 26 (contÕd)
pKM10.204C
Ptrc
ntdC
sub
serA in pBbA1a
this work
pKM10.214CA
Ptrc
ntdC
sub
ntdA
sub
serA in pBbA1a
this work
pKM10.214CB
Ptrc
ntdC
sub
ntdB
sub
serA in pBbA1a
this work
pKM10.214AB
Ptrc
ntdA
sub
ntdB
sub
serA in pBbA1a
this work
pKM10.214BC
Ptrc
ntdB
sub
ntdC
sub
serA in pBbA1a
this work
pKM10.220ABC
Ptrc
ntdA
sub
ntdB
sub
ntdC
sub
serA in pBbA1a
this work
pUCIDT(Kan)-rifK
KanR, Plac
, codon optimized 
rifK, ColE1 replicon
IDT
pUCIDT(Kan)-rifL
KanR, Plac
, codon optimized 
rifL, ColE1 replicon
IDT
pUCIDT(Kan)-rifM
KanR, Plac
, codon optimized 
rifM, ColE1 replicon
IDT
pKM10.254K
Ptrc
rifK in pBbA1a
this work
pKM10.254L
Ptrc
rifL in pBbA1a
this work
pKM10.254M
Ptrc
rifM in pBbA1a
this work
pKM10.268K
Ptrc
rifKserA
 in pBbA1a
this work
pKM10.268L
Ptrc
rifLserA in pBbA1a
this work
pKM10.268M
Ptrc
rifMserA
 in pBbA1a
this work
pKM10.278KM
Ptrc
rifKrifMserA
 in pBbA1a
this work
pKM10.284L
PT5His6-rifL in pJG7.246
this work
pKM11.9
Ptrc
rifLrifKrifMserA
 in pBbA1a
this work
pKM11.46
PT7rifLrifKrifMserA
 in pBbA7k
this work
pUCIDT(Kan)-ntdA
KanR, Plac
, ntdA
pum, ColE1 replicon
IDT
pUCIDT(Kan)-ntdB
KanR, Plac
, ntdB
pum, ColE1 replicon
IDT
pUCIDT(Kan)-ntdC
KanR, Plac
, ntdC
pum, ColE1 replicon
IDT
pKM11.121A
Ptrc
ntdA
pum in pBbA1a
this work
pKM11.121B
Ptrc
ntdB
pum in pBbA1a
this work
pKM11.121C
Ptrc
ntdC
pum in pBbA1a
this work
pKM11.125
Ptrc
ntdB
pumserA in pBbA1a
this work
pKM11.130
Ptrc
ntdC
pumntdA
pum in pBbA1a
this work
pKM11.135
Ptrc
ntdC
pumntdA
pumntdB
pumserA in pBbA1a
this work
%226Plasmid pSN1.292 
 Plasmid pSN1.292 contained a 
Ptrc
ntdC
sub
ntdA
sub
ntdB
sub
serA cassette.  It was created by 
first inserting 
ntdC
sub
 from plasmid pSN1.294 upstream of 
ntdA
 in plasmid pSN1.296 to afford 
plasmid pSN1119.  The 
ntdC
sub
ntdA
sub
 gene was then inserted upstream of 
ntdB
sub
 in plasmid 
pSN1.295to afford plasmid pSN1.139, which carried a 
Ptrc
ntdC
sub
ntdA
sub
ntdB
sub
 cassette.  
Finally, the 
ntdC
sub
ntdA
sub
ntdB
sub
 gene fragment
 was inserted upstream of 
serA in plasmid 
pKM6.240 to afford plasmid pSN1.292. 
Plasmid pKM10.220ABC 
 Plasmid pKM10.220ABC contained a 
Ptrc
ntdA
sub
ntdB
sub
ntdC
sub
serA cassette.  It was 
created by first inserting 
ntdC
sub
 from plasmid pSN1.294 upstream of 
serA in plasmid pKM6.240 
to afford plasmid pKM10.204C.  The 
ntdB
sub
 gene was then inserted upstream of 
ntdC
sub
 to 
afford plasmid pKM10.214BC, which carried a 
Ptrc
ntdB
sub
ntdC
sub
serA cassette.  Finally, 
ntdA
sub 
was inserted upstream of 
ntdB
sub
 to afford plasmid pKM10.220ABC. 
Plasmid pKM10.214CA 
 Plasmid pKM10.214CA contained a 
Ptrc
ntdC
sub
ntdA
sub
serA cassette.  It was created by 
first inserting 
ntdC
sub
 from plasmid pSN1.294 upstream of 
ntdA
sub
 in plasmid pKM10.204A to 
afford plasmid pKM10.214CA.  
%227Plasmid pKM10.214CB 
 Plasmid pKM10.214CB contained a 
Ptrc
ntdC
sub
ntdB
sub
serA cassette.  It was created by 
first inserting 
ntdC
sub
 from plasmid pSN1.294 upstream of 
ntdB
sub
 in plasmid pKM10.204B to 
afford plasmid pKM10.214CB.  
Plasmid pKM10.214AB 
 Plasmid pKM10.214AB contained a 
Ptrc
ntdA
sub
ntdB
sub
serA cassette.  It was created by 
first inserting 
ntdA
sub
 from plasmid pSN1.296 upstream of 
ntdB
sub
 in plasmid pKM10.204B to 
afford plasmid pKM10.214AB.  
Plasmid pKM10.82A 
 Plasmid pKM10.82A contained a 
PT7ntdC
sub
ntdA
sub
ntdB
sub
serA 
cassette.  The 
ntdC
sub
ntdA
sub
ntdB
sub
serA 
fragment was removed from pSN1.292 by 
EcoRI and 
Bam
HI 
digestion.  BglBrick plasmid pBbA7k (Figure 111) was linearized also with 
EcoRI and 
Bam
HI 
digestion.  Ligation of the two DNA fragments afforded pKM10.82A. 
%228% Figure 111
.  BglBrick plasmid pBbA7k-RFP.
 Plasmid pKM11.9 
 Plasmid pKM11.9 contained a 
Ptrc
rifLrifKrifMserA
 cassette.  It was created by first 
inserting 
rifM from plasmid pKM10.254M upstream of 
serA in plasmid pKM6.240 to afford 
plasmid pKM10.268M.  The 
rifK gene was then inserted upstream of 
rifM to afford plasmid 
pKM10.278KM, which carried a 
Ptrc
rifKrifMserA
 cassette.  Finally, 
rifL was inserted upstream of 
rifK to afford plasmid pKM11.9. 
Plasmid pKM11.46 
 Plasmid pKM11.46 contained a 
PT7rifLrifKrifMserA 
cassette.  The 
rifLrifKrifMserA 
fragment was removed from pKM11.9 by 
EcoRI and 
Bam
HI digestion.  BglBrick plasmid 
pBbA7k (Figure 114) was linearized also with 
EcoRI and 
Bam
HI digestion.  Ligation of the two 
DNA fragments afforded pKM11.46. 
EcoR
IBgl
IIBamH
IXho
IlacI
qp15Aori
mRFP-1
pBbA7k-RFP
4.3 kb
Kan
RPT7%229Plasmid pKM11.135 
 Plasmid pKM11.135 contained a 
Ptrc
ntdC
pumntdA
pumntdB
pumserA cassette.  It was created 
by first inserting 
ntdB
pum from plasmid pKM11.121B upstream of 
serA in plasmid pKM6.240 to 
afford plasmid pKM11.125.  The 
ntdC
pum gene was inserted upstream of 
ntdA
pum in plasmid 
pKM11.121A to afford plasmid pKM11.130.  Finally, the 
ntdC
pumntdA
pum 
fragment
 was inserted 
upstream of 
ntdB
sub
 in plasmid pKM11.125 to afford plasmid pKM11.135. 
Plasmid pKM10.284 
 The 
rifL gene was PCR-amplified from pKM10.254L without its native ATG start codon 
and the addition of 
Bam
HI sites at the 5Õ and 3Õ end.  The PCR product was inserted into the 
Bam
HI site of pJG7.246 to afford plasmid pKM10.284.  The relevant gene and vector sequences 
are shown in Figure 112. 
% Figure 112.
  Beginning gene sequence for plasmid pKM10.284.  Vector codons are in blue, the BamHI site is in 
black, 
rifL 
codon is in green.  The transcription start codon and the His
6 sequence on vector pJG7.246 are 
underlined. 
Plasmid pKM9.297C  
 The 
ntdC
sub
 gene was PCR-amplified from pSN1.294 without its native ATG start codon 
and the addition of 
Bam
HI sites at the 5Õ and 3Õ end.  The PCR product was inserted into the 
Bam
HI site of pJG7.246 to afford plasmid pKM9.279C.  The relevant gene and vector sequences 
are shown in Figure 113. 
5Õ  Ñ 
ATG
 AGA GGA TCG 
CAT CAC CAT CAC CAT CAC
 GGA TCC
 GCA 
Ñ  3Õ
%230% Figure 113.
  Beginning gene sequence for plasmid pKM9.279C.  Vector codons are in blue, the 
Bam
HI site is in 
black, 
ntdC
sub
 codon is in green.  The transcription start codon and the His
6 sequence on vector pJG7.246 are 
underlined. 
   
RB791
serA/pSN1.292 growth study with exogenous kanosamine 
 A culture of RB791
serA/pSN1.292 in 5 mL modified M9 media was grown at 35¡C with 
agitation at 250 rpm for 11 h.  An aliquot of this culture was used to inoculate three 20 mL 
cultures of the same media in 100 mL three-baffled flasks to an initial OD
600 = 0.05.  Growth 
continued at 35¡C with agitation at 250 rpm until OD
600 = 0.4.  An appropriate amount of a 300 
mM aqueous solution of kanosamine hydrochloride (pH 8.0) was added to each of the three 
flasks to achieve the desired final concentration of 0, 25, or 50 mM.  Growth continued at 35¡C 
with agitation at 250 rpm and was monitored once every hour until stationary phase.  Growth 
rates were determined during exponential phase by using the conversion factor of OD
600(1.0) = 1.0 x 108 cells/mL.  The logarithm of the number of cells measured between 4 h and 8 h was 
used to compare the rate of growth for each culture. 
Cell-lysate preparations 
 E. coli
 cells were harvested from flask cultures by centrifugation at 4,000
g for 5 min at 
4¡C.  Cell samples removed from fermentation runs were collected by centrifugation at 11,300
g for 20 min at 4¡C. Cell lysis was achieved by two passages through a French¨ pressure cell 
(SLM Aminco) at 17,400 psi.  Cellular debris was removed from the lysate by centrifugation 
(38,500g, 30 min, 4¡C).  Protein concentrations were determined using the Bradford dye-binding 
procedure
11 using a protein assay solution purchased from Bio-Rad (1
! concentration).  Assay 
5Õ  Ñ 
ATG
 AGA GGA TCG 
CAT CAC CAT CAC CAT CAC
 GGA TCC
 TCT 
Ñ  3Õ
%231solution (1 mL) was combined with 20 
$
L protein containing solution and incubated at rt for 3 
minutes. The absorbance was measured at 595 nm and the protein concentration determined by 
comparison to a standard curve prepared using bovine serum albumin. 
Protein purifications 
B. subtilis
 ntdC
-encoded 
D-glucose-6-phosphate 3-dehydrogenase 
 D-Glucose-6-phosphate (G6P) 3-dehydrogenase
17 was purified from 
E. coli 
DH5
!
/pKM9.279C (
PT5, lacO
, His6, ntdC
, Amp
R) on a Ni-NTA Agarose column (Qiagen).  Buffers 
used in the purification included buffer A: Tris-HCl (20 mM), NaCl (500 mM), imidazole (5 
mM), glycerol (20% v/v), pH 7.9; buffer B: Tris-HCl (20 mM), NaCl (500 mM), imidazole (60 
mM), glycerol (20% v/v), pH 7.9; buffer C: Tris-HCl (20 mM), NaCl (500 mM), imidazole (5 
mM), EDTA (50 mM), glycerol (20% v/v), pH 7.9; buffer D: Tris-HCl (25 mM), NaCl (150 
mM), glycerol (50% v/v), pH 8.0.   
 A single colony of 
E. coli
 DH5
!
/pKM9.279C was used to inoculate 5 mL LB containing 
ampicillin.  Cells were grown with agitation at 250 rpm, 37¡C for 14 h.  An appropriate amount 
of culture was used to inoculate 500 mL LB containing ampicillin to an initial OD
600 = 0.01.  Growth continued at 250 rpm, 37¡C for 3 h until the OD
600 reached 0.6.  IPTG was added to a 
final concentration of 1 mM.  After an additional 4 h of growth, cells (4.8 g) were harvested by 
centrifugation (5,500
g, 10 min, 4¡C), resuspended in buffer A (2 mL/g wet cells), and lysed by 
two passages through a French¨ pressure cell at 17,400 psi.  Cellular debris was removed by 
centrifugation (38,500
g, 30 min, 4¡C).  To the supernatant was added a 50% (w/v) slurry of Ni-
NTA agarose resin (1 mL resin per 4 mL crude lysate) that had been equilibrated in buffer A and 
%232the mixture was gently stirred at 4¡C for 1 h.  The lysateÐresin slurry was transferred to a 
polypropylene column (Qiagen) and the column was washed with buffer B (2 
! 4 mL per mL of 
Ni-NTA agarose resin).  The His
6-tagged protein was eluted from the column by washing with 
buffer C (4 
! 0.5 mL per mL of Ni-NTA agarose resin).  Fractions were analyzed by sodium 
dodecyl sulfateÐpolyacrylamide gel electrophoresis (SDSÐPAGE), and those containing pure 
protein were combined and dialyzed twice against buffer D at 4¡C.  Enzyme was stored in 100 
$
L aliquots at 
*20¡C.  
Amycolatopsis mediterranei rifL
-encoded UDP-
D-glucose 3-dehydrogenase 
 UDP-
D-glucose 3-dehydrogenase was purified from 
E. coli 
DH5
!
/pKM10.284 (
PT5, lacO
, His6, rifL, Amp
R) on an •KTA
start FPLC.
  Buffers used in the purification included buffer 
A: NaH
2PO4 (20 mM, pH 7.4), NaCl (500 mM); buffer B: NaH
2PO4 (20 mM, pH 7.4), NaCl (500 
mM), imidazole (500 mM) buffer C: Tris-HCl (25 mM), NaCl (150 mM), glycerol (50% v/v), 
pH 8.0.   
 A single colony of 
E. coli
 DH5
!
/pKM10.284 was used to inoculate 5 mL LB containing 
ampicillin (50 
$
g/mL).  Cells were grown with agitation at 250 rpm, 37¡C for 14 h.  An 
appropriate amount of culture was used to inoculate 500 mL LB containing ampicillin to an 
initial OD
600 = 0.01.  Growth continued at 250 rpm, 37¡C for 3 h until the OD
600 reached 0.6.  
IPTG was added to a final concentration of 1 mM.  After an additional 4 h of growth, cells (5.3 
g) were harvested by centrifugation (5,500
g, 10 min, 4¡C), resuspended in buffer A (2 mL/g wet 
cells), and lysed by two passages through a French¨ pressure cell at 17,400 psi.  Cellular debris 
was removed by centrifugation (38,500
g, 30 min, 4¡C).  The supernatant was loaded onto a 
%233HisTrap Fast Flow Sepharose column (16 x 25 mm, 1.0 mL) which had been equilibrated with 
buffer A.  The column was washed with buffer A (20 mL) and eluted with a linear gradient 
(4%-100%) of buffer A and buffer B.  Fractions were analyzed by sodium dodecyl sulfateÐ
polyacrylamide gel electrophoresis (SDSÐPAGE), and those containing pure protein were 
combined and dialyzed twice against buffer C at 4¡C.  Enzyme was stored in 100 
$
L aliquots at 
*20¡C.  
Enzyme assays
 D-glucose-6-phosphate 3-dehydrogenase assay (NtdC
sub
 and NtdC
pum)  Enzyme activity was determined spectrophotometrically by monitoring increase in 
absorption at 340 nm due to the formation of NADH (
,
 = 6,220 M*1 cm
*1). The reaction was 
carried out in 2-amino-1-methylpropanol (AMP, 100 mM) buffer (pH 9.5) using a mixture (1 
mL) consisting of 
D-glucose-6-phosphate (5 mM), NAD
+ (5 mM), and an appropriate amount of 
enzyme at rt. One unit of 
D-glucose-6-phosphate 3-dehydrogenase activity was defined as the 
amount which catalyzed the formation of 1 
$
mol of NADH per min.  
UDP-D-glucose 3-dehydrogenase assay (RifL) 
 Enzyme activity was determined spectrophotometrically by monitoring increase in 
absorption at 340 nm due to the formation of NADH (
,
 = 6,220 M*1 cm
*1).  The reaction was 
carried out in sodium phosphate buffer (50 mM, pH 7.0) using a mixture (1 mL) consisting of 
UDP-D-glucose (5 mM), NAD
+ (5 mM), and an appropriate amount of enzyme at rt.  One unit of  
%234UDP-D-glucose 3-dehydrogenase activity was defined as the amount which catalyzed the 
formation of 1 
$
mol of NADH per min.
 Purification of kanosamine from fermentation broth 
 After completion of a fermentation run, cells were harvested by centrifugation (6,400 
g, 4¡C, 20 min).  Cell-free broth was then passed through a tangential flow filtration apparatus (10 
kD) (Sartorius) to remove proteins. The clarified broth was then treated with activated charcoal 
(10 g L, Darco KB-B, 100 mesh) at 37¡C for 1 h and filtered through a layer of Celite 545. The 
colorless filtrate was loaded onto a cation exchange column (Dowex 50WX8, H
+-form, 50-100 
mesh) and washed with four column volumes of water. Kanosamine was eluted as the 
hydrochloride salt with a linear gradient of HCl (0-0.5 N). Kanosamine containing fractions were 
combined and adjusted to pH 7.0 with KOH (5 N). Water was removed under reduced pressure 
and the pale yellow solids extracted into methanol (2 x 30 mL). Insoluble salts were removed by 
filtration. The filtrate was evaporated under reduced pressure to afford kanosamine 
hydrochloride as a white solid. 
1H NMR (500 MHz, D
2O, TSP = 0.00 ppm): 
' 5.27 (d, J = 3.5 Hz, 1H), 4.73 (d, 
J = 7.7 Hz, 1H), 3.94-3.73 (m, 6H), 3.67 (dt, 
J = 9.9, 2.9 Hz, 2H), 3.58 (ddd,
 J = 9.7, 5.4, 2.2 Hz, 1H), 3.47-3.39 (m, 2H), 3.24 (t, 
J = 10.4 Hz, 1H). 
13C NMR (126 MHz, D
2O, 
TSP = 0.0 ppm): 
' 98.9, 94.0, 79.6, 74.1, 73.4, 71.0, 68.9, 68.8, 63.0, 62.8, 60.7, 57.9. %235REFERENCES 
%236REFERENCES 
1.  Zhang, P.; Kriegel, R. M.; Frost, J. W. BÐOÐB Catalyzed Cycloadditions of Acrylic Acids. 
ACS 
Sustainable Chem. Eng.
 2016, 4 (12), 6991Ð6995. 2.  Niu, W.; Draths, K. M.; Frost, J. W. Benzene-Free Synthesis of Adipic Acid. 
Biotechnol. Prog.
 2002, 18, 201Ð211. 
3.  Frost, J. W.; Hansen, C. A. Synthesis of 1,2,3,4-Tetrahydroxybenzenes and 1,2,3,-Trihydroxybenzenes 
Using Myo-Inositol-1-Phosphate Synthase and 
myo-Inositol-2-Dehydrogenase. US 6,750,049, 2004. 
4.  Datsenko, K. A.; Wanner, B. L. One-Step Inactivation of Chromosomal Genes in Escherichia Coli 
K-12 Using PCR Products. 
Proc. Natl. Acad. Sci. U. S. A.
 2000, 97 (12), 6640Ð6645. 5.  Kasai, D.; Fujinami, T.; Abe, T.; Mase, K.; Katayama, Y.; Fukuda, M.; Masai, E. Uncovering the 
Protocatechuate 2,3-Cleavage Pathway Genes. 
J. Bacteriol.
 2009, 191 (21), 6758Ð6768. 6.  a) Lee, T. S.; Krupa, R. A.; Zhang, F.; Hajimorad, M.; Holtz, W. J.; Prasad, N.; Lee, S. K.; Keasling, J. 
D. BglBrick Vectors and Datasheets: A Synthetic Biology Platform for Gene Expression. 
J. Biol. Eng.
 2011
, 5, 12.  b) Anderson, J.; Dueber, J. E.; Leguia, M.; Wu, G. C.; Goler, J. A.; Arkin, A. P.; Keasling, J. 
D. BglBricks: A Flexible Standard for Biological Part Assembly. 
J. Biol. Eng.
 2010, 4 (1), 1. 7.  Guo, J. Biosynthesis of Precursors to the Aminoshikimate Pathway and Microbial Synthesis of 5-
Amino-5-Deoxyshikimate. Doctor of Philosophy, Michigan State University, 2004. 
8.  Miller, J. H. 
Experiments in Molecular Genetics
; Cold Spring Harbor Laboratory: Cold Spring Harbor, 
NY, 1972. 
9.  Sambrook, J.; Russell, D. W. 
Molecular Cloning: A Laboratory Manual
 3rd ed.; Cold Spring Harbor 
Laboratory: Cold Spring Harbor, NY, 2001. 
10.  Knop, D. R.; Draths, K. M.; Chandran, S. S.; Barker, J. L.; von Daeniken, R.; Weber, W.; Frost, J. W. 
Hydroaromatic Equilibration During Biosynthesis of Shikimic Acid. 
J. Am. Chem. Soc.
 2001, 123 (42), 10173Ð10182. 11.  Sieben, A. S.; Perlin, A. S.; Simpson, F. J. An Improved Preparative Method for 
D-Erythrose 4-
Phosphate. 
Canadian Journal of Chemistry
 1966, 44 (6), 663Ð669. 12.  Wecke, T. Isolation of genomic DNA from Bacillus http://2012.igem.org/wiki/images/6/6b/LMU-
Munich_2012_Isolation_of_genomic_DNA_from_Bacillus_%28for_PCR....%29.pdf (accessed Jan 
2018). 13.  Farabaugh, M. A. Biocatalytic Production of Aromatics from 
D-glucose.  Master of Science, 
Michigan State University, 1996. 
14.  Paoletti, F.; Williams, J. F.; Horecker, B. L. An Enzymic Method for the Analysis of 
D-Erythrose 4-
Phosphate. 
Anal. Biochem.
 1979, 95, 250Ð253. 15.  Schoner, R.; Herrmann, K. M. 3-Deoxy-
D-arabino
-Heptulosonate 7-Phosphate Synthase. 
Purification, Properties, and Kinetics of the Tyrosine-Sensitive Isoenzyme from 
Escherichia coli
. J. Biol. 
Chem.
 1976, 251 (18), 5440Ð5447. 16.  Gollub, E.; Zalkin, H.; Sprinson, D. B. Assay for 3-Deoxy-
D-arabino
-Heptulosonic Acid 7-Phosphate 
Synthase. 
Methods in Enzymology
 1971, 17, 349Ð350. %23717.  Vetter, N. D.; Palmer, D. R. J. Simultaneous Measurement of Glucose-6-Phosphate 3-Dehydrogenase 
(NtdC) Catalysis and the Nonenzymatic Reaction of Its Product: Kinetics and Isotope Effects on the First 
Step in Kanosamine Biosynthesis. 
Biochemistry
 2017, 56 (14), 2001Ð2009. %238