Hydroaromatic equilibration during biosynthesis of shikimic acid

Article abstract of DOI:10.1021/ja0109444

The expense and limited availability of shikimic acid isolated from plants has impeded utilization of this hydroaromatic as a synthetic starting material. Although recombinant Escherichia coli catalysts have been constructed that synthesize shikimic acid from glucose, the yield, titer, and purity of shikimic acid are reduced by the sizable concentrations of quinic acid and 3-dehydroshikimic acid that are formed as byproducts. The 28.0 g/L of shikimic acid synthesized in 14% yield by E. coli SP1.1/pKD12.138 in 48 h as a 1.6:1.0:0.65 (mol/mol/mol) shikimate/quinate/dehydroshikimate mixture is typical of synthesized product mixtures. Quinic acid formation results from the reduction of 3-dehydroquinic acid catalyzed by aroE-encoded shikimate dehydrogenase. Is quinic acid derived from reduction of 3-dehydroquinic acid prior to synthesis of shikimic acid? Alternatively, does quinic acid result from a microbe-catalyzed equilibration involving transport of initially synthesized shikimic acid back into the cytoplasm and operation of the common pathway of aromatic amino acid biosynthesis in the reverse of its normal biosynthetic direction? E. coli SP1.1/pSC5.214A, a construct incapable of de novo synthesis of shikimic acid, catalyzed the conversion of shikimic acid added to its culture medium into a 1.1:1.0:0.70 molar ratio of shikimate/quinate/dehydroshikimate within 36 h. Further mechanistic insights were afforded by elaborating the relationship between transport of shikimic acid and formation of quinic acid. These experiments indicate that formation of quinic acid during biosynthesis of shikimic acid results from a microbe-catalyzed equilibration of initially synthesized shikimic acid. By apparently repressing shikimate transport, the aforementioned E. coli SP1.1/pKD12.138 synthesized 52 g/L of shikimic acid in 18% yield from glucose as a 14:1.0:3.0 shikimate/quinate/dehydroshikimate mixture.

Full text of DOI:10.1021/ja0109444

J. Am. Chem. Soc. 2001, 123, 10173-10182
10173
Hydroaromatic Equilibration During Biosynthesis of Shikimic Acid
‡
†
†
†
David R. Knop, K. M. Draths, Sunil S. Chandran, Jessica L. Barker,
§
§
,†,‡
Ralph von Daeniken, Wolfgang Weber, and J. W. Frost*
Contribution from the Departments of Chemistry and Chemical Engineering, Michigan State UniVersity,
East Lansing, Michigan 48824-1322, and F. Hoffmann-La Roche, Ltd., Pharmaceutical Research,
CH-4070 Basel, Switzerland
ReceiVed April 12, 2001
Abstract: The expense and limited availability of shikimic acid isolated from plants has impeded utilization
of this hydroaromatic as a synthetic starting material. Although recombinant Escherichia coli catalysts have
been constructed that synthesize shikimic acid from glucose, the yield, titer, and purity of shikimic acid are
reduced by the sizable concentrations of quinic acid and 3-dehydroshikimic acid that are formed as byproducts.
The 28.0 g/L of shikimic acid synthesized in 14% yield by E. coli SP1.1/pKD12.138 in 48 h as a 1.6:1.0:0.65
(mol/mol/mol) shikimate/quinate/dehydroshikimate mixture is typical of synthesized product mixtures. Quinic
acid formation results from the reduction of 3-dehydroquinic acid catalyzed by aroE-encoded shikimate
dehydrogenase. Is quinic acid derived from reduction of 3-dehydroquinic acid prior to synthesis of shikimic
acid? Alternatively, does quinic acid result from a microbe-catalyzed equilibration involving transport of initially
synthesized shikimic acid back into the cytoplasm and operation of the common pathway of aromatic amino
acid biosynthesis in the reverse of its normal biosynthetic direction? E. coli SP1.1/pSC5.214A, a construct
incapable of de novo synthesis of shikimic acid, catalyzed the conversion of shikimic acid added to its culture
medium into a 1.1:1.0:0.70 molar ratio of shikimate/quinate/dehydroshikimate within 36 h. Further mechanistic
insights were afforded by elaborating the relationship between transport of shikimic acid and formation of
quinic acid. These experiments indicate that formation of quinic acid during biosynthesis of shikimic acid
results from a microbe-catalyzed equilibration of initially synthesized shikimic acid. By apparently repressing
shikimate transport, the aforementioned E. coli SP1.1/pKD12.138 synthesized 52 g/L of shikimic acid in 18%
yield from glucose as a 14:1.0:3.0 shikimate/quinate/dehydroshikimate mixture.
Shikimic acid is a hydroaromatic intermediate (Scheme 1)
Scheme 1a
1
in the common pathway of aromatic amino acid biosynthesis.
2
Although the common pathway has been observed in plants,
microbes, and parasites,4 this wide distribution does not
translate into abundant availability of the pathway’s biosynthetic
intermediates such as shikimic acid. A tedious, multistep process
3
5
is used for isolation of shikimic acid from Illicium spp., a plant
whose cultivation has never benefited from large-scale monoc-
ulture. The carbocyclic array of chirality displayed by shikimic
acid has attracted attention as the basis for synthesis of a large
6
combinatorial library. However, synthetic utilization of shikimic
acid has been much more restricted in comparison to the more
abundantly available and less expensive quinic acid (Scheme
7
1
). Shikimic acid’s expense and limited availability has also
†
Department of Chemistry.
Department of Chemical Engineering.
F. Hoffmann-La Roche, Ltd.
‡
§
(
1) Haslam, E. Shikimic Acid: Metabolism and Metabolites; Wiley &
Sons: New York, 1993.
2) Herrmann, K. M.; Weaver, L. M. Annu. ReV. Plant Physiol. Plant
Mol. Biol. 1999, 50, 473.
3) Pittard, A. J. In Escherichia coli and Salmonella: Cellular and
Molecular Biology; Neidhardt, F. C., Ed.; ASM Press: Washington, DC,
996; Chapter 28.
4) Roberts, F.; Roberts, C. W.; Johnson, J. J.; Kyle, D. E.; Krell, T.;
(
a
Biosynthetic intermediates (abbreviations): phosphoenolpyruvic
acid (PEP), D-erythrose 4-phosphate (E4P), 3-deoxy-D-arabino-heptu-
losonic acid 7-phosphate (DAHP), shikimate 3-phosphate (S3P).
(
1
FBR
(
Enzymes (encoding genes): (a) DAHP synthase (aroF ), (b) 3-de-
hydroquinate synthase (aroB), (c) 3-dehydroquinate dehydratase (aroD),
d) shikimate dehydrogenase (aroE), (e,f) shikimate kinase (aroK, aroL).
Coggins, J. R.; Coombs, G. H.; Milhous, W. K.; Tzipori, S.; Ferguson, D.
J. P.; Chakrabarti, D.; McLeod, R. Nature 1998, 393, 801.
(
(
5) Ref 1, pp 40-42.
(6) Tan, D. S.; Foley, M. A.; Shair, M. D.; Schreiber, S. L. J. Am. Chem.
complicated its use as the starting material for the synthesis of
neuraminidase inhibitors such as the antiinfluenza drug marketed
as Tamiflu.
Soc. 1998, 120, 8565.
7) Barco, A.; Benetti, S.; De Risi, C.; Marchetti, P.; Pollini, G. P.;
Zanirato, V. Tetrahedron: Asymmetry 1997, 8, 3515.
(
8
1
0.1021/ja0109444 CCC: $20.00 © 2001 American Chemical Society
Published on Web 09/28/2001

10174 J. Am. Chem. Soc., Vol. 123, No. 42, 2001
Knop et al.
9
Table 1. Restriction Enzyme Maps of Plasmidsa
As part of a larger effort to replace isolation of scarce natural
products with microbe-catalyzed syntheses from abundant and
inexpensive carbohydrates, shikimic acid has recently been
synthesized from glucose using a recombinant Escherichia coli
strain under fermentor-controlled conditions.10 In addition to
shikimic acid, the hydroaromatics quinic acid and 3-dehy-
10
droshikimic acid were synthesized (Scheme 1) as byproducts.
Formation of quinic and 3-dehydroshikimic acids reduces the
yield and concentration of microbe-synthesized shikimic acid.
Quinic acid formation is particularly problematic, as this
hydroaromatic significantly complicates purification of shikimic
acid. Methodology has been reported for reducing the formation
1
0
of quinic acid during microbial synthesis of shikimic acid.
Unfortunately, this methodology does not suppress quinic acid
formation when tktA-encoded transketolase is overexpressed.
Amplified expression of transketolase has been demonstrated
to increase the yield and titer of 3-dehydroshikimic acid (Scheme
) synthesized from glucose.2
3a
1
In this account, the mechanism underlying the formation of
quinic acid and 3-dehydroshikimic acid during biosynthesis of
shikimic acid under glucose-limited conditions is further
delineated. All experiments rely on cultivation of E. coli in
fermentors in order to control pH, temperature, dissolved O2
concentrations, and the concentration of glucose in cultures. An
intriguing picture emerges of the common pathway operating
in the reverse of its normal biosynthetic direction with transport
of initially synthesized shikimic acid back into the cytoplasm
of E. coli playing a significant role in the formation of quinic
a
Restriction enzyme sites are abbreviated as follows: B ) BamHI,
Bg ) BglII, E ) EcoRI, H ) HindIII, K ) KpnI, N ) NcoI, P ) PstI,
S ) SacI, Sl ) SalI, Sm ) SmaI, Sp ) SphI, X ) XbaI. Parentheses
indicate that the designated enzyme site has been eliminated. Lightface
line indicates vector DNA; boldface line indicates insert DNA.
(8) (a) Kim, C. U.; Lew, W.; Williams, M. A.; Liu, H.; Zhang, L.;
Swaminathan, S.; Bischofberger, N.; Chen, M. S.; Mendel, D. B.; Tai, C.
Y.; Laver, W. G.; Stevens, R. C. J. Am. Chem. Soc. 1997, 119, 681. (b)
Rohloff, J. C.; Kent, K. M.; Postich, M. J.; Becker, M. W.; Chapman, H.
H.; Kelly, D. E.; Lew, W.; Louie, M. S.; McGee, L. R.; Prisbe, E. J.;
Schultze, L. M.; Yu, R. H.; Zhang, L. J. Org. Chem. 1998, 63, 4545.
acid. This mechanistic insight has, in turn, resulted in the
identification of methodologies that can be conveniently used
to minimize formation of quinic acid during microbial synthesis
of shikimic acid from glucose even when expression of tktA-
encoded transketolase is amplified. The result is a significant
improvement in the yield and titer of shikimic acid synthesized
from glucose.
(
9) (a) Li, K.; Frost, J. W. J. Am. Chem. Soc. 1998, 120, 10545. (b)
Hansen, C. A.; Dean, A. B.; Draths, K. M.; Frost, J. W. J. Am. Chem. Soc.
999, 121, 3799. (c) Kambourakis, S. K.; Draths, K. M.; Frost, J. W. J.
1
Am. Chem. Soc. 2000, 122, 9042. (d) Kambourakis, S. K.; Frost, J. W. J.
Org. Chem. 2000, 65, 6904.
(
10) Draths, K. M.; Knop, D. R.; Frost, J. W. J. Am. Chem. Soc. 1999,
Results
1
21, 1603.
Shared Genetic Elements. Both shikimate kinase isozymes
(Scheme 1) encoded by aroK and aroL were inactivated in all
E. coli constructs used in this study by successive P1 phage-
(
(
11) Løbner-Olesen, A.; Marinus, M. G. J. Bacteriol. 1992, 174, 525.
12) Ogino, T.; Garner, C.; Markley, J. L.; Herrmann, K. M. Proc. Natl.
Acad. Sci. U.S.A. 1982, 79, 5828.
(
(
13) Weaver, L. M.; Herrmann, K. M. J. Bacteriol. 1990, 172, 6581.
14) Snell, K. D.; Draths, K. M.; Frost, J. W. J. Am. Chem. Soc. 1996,
R
mediated transductions of aroL478::Tn10 and aroK17::Cm into
1
1
1
18, 5605.
the appropriate E. coli host. While ensuring that carbon flow
directed into the common pathway did not proceed beyond
synthesis of shikimic acid, inactivation of the shikimate kinases
also precluded de novo biosynthesis of aromatic amino acids
and aromatic vitamins. Growth of all constructs therefore
required supplementation with L-phenylalanine, L-tyrosine,
L-tryptophan, p-hydroxybenzoic acid, p-aminobenzoic acid, and
2,3-dihydroxybenzoic acid. These supplements could potentially
create a problem in that feedback inhibition of 3-deoxy-D-
arabino-heptulosonic acid 7-phosphate (DAHP) synthase by
aromatic amino acids plays a prominent role in controlling
(
(
15) Dell, K. A.; Frost, J. W. J. Am. Chem. Soc. 1993, 115, 11581.
16) (a) Draths, K. M.; Frost, J. W. J. Am. Chem. Soc. 1990, 112, 1657.
(
b) Draths, K. M.; Pompliano, D. L.; Frost, J. W.; Berry, A.; Disbrow, G.
L.; Staversky, R. J.; Lievense, J. C. J. Am. Chem. Soc. 1992, 114, 3956. (c)
Patnaik, R.; Liao, J. C. Appl. EnViron. Microbiol. 1994, 60, 3903. (d) Lu,
J.-L.; Liao, J. C. Biotechnol. Bioeng. 1997, 53, 132.
(17) (a) Konstantinov, K. B.; Nishio, N.; Yoshida, T. J. Ferment. Bioeng.
1
990, 70, 253. (b) Konstantinov, K. B.; Nishio, N.; Seki, T.; Yoshida, T. J.
Ferment. Bioeng. 1991, 71, 350. (c) Kleman, G. L.; Strohl, W. R. Appl.
EnViron. Microbiol. 1994, 60, 3952.
(18) (a) Pittard, J.; Wallace, B. J. J. Bacteriol. 1966, 92, 1070. (b) Brown,
K. D.; Doy, C. H. Biochim. Biophys. Acta 1976, 428, 550. (c) Whipp, M.
J.; Camakaris, H.; Pittard, A. J. Gene 1998, 209, 185
12
(19) Saier, M. H., Jr.; Ramseier, T. M.; Reizer, J. In Escherichia coli
carbon flow directed into the common pathway. Accordingly,
and Salmonella: Cellular and Molecular Biology; Neidhardt, F. C., Ed.;
ASM Press: Washington, DC, 1996; Chapter 85.
an isozyme of DAHP synthase encoded by plasmid-localized
FBR
aroF
that was insensitive to feedback inhibition by aromatic
(20) (a) Postma, P. W.; Roseman, S. Biochim. Biophys. Acta 1976, 457,
2
13. (b) Misset, O.; Blaauw, M.; Postma, P. W.; Robillard, G. T.
amino acids was carried as a plasmid-localized insert (Table 1:
pKD12.112, pKD12.138, and pJB5.291; details of strains and
plasmids can be found in Table 2).
Biochemistry 1983, 22, 6163. (c) Grenier, F. C.; Waygood, E. B.; Saier,
M. H., Jr. J. Cell. Biochem. 1986, 31, 97. (d) Meadow, N. D.; Roseman, S.
J. Biol. Chem. 1982, 257, 14526. (e) Stock, J. B.; Waygood, E. B.; Meadow,
N. D.; Postma, P. W.; Roseman, S. J. Biol. Chem. 1982, 257, 14543.
1
3
With increased carbon flow resulting from amplified expres-
FBR
(21) Bonner, C. A.; Jensen, R. A. Biochem. J. 1994, 302, 11.
(22) Geck, M. K.; Kirsch, J. F. Biochemistry 1999, 38, 8032.
(23) (a) Li, K.; Mikola, M. R.; Draths, K. M.; Worden, R. M.; Frost, J.
sion of aroF , DAHP is not converted into 3-dehydroquinic
acid by aroB-encoded 3-dehydroquinate synthase at a rate
sufficiently rapid to avoid dephosphorylation of DAHP and
export of 3-deoxy-D-arabino-heptulosonic acid (DAH) into the
W. Biotechnol. Bioeng. 1999, 64, 61. (b) Li, K.; Frost, J. W. Biotechnol.
Prog. 1999, 15, 876.

Biosynthesis of Shikimic Acid
J. Am. Chem. Soc., Vol. 123, No. 42, 2001 10175
Table 2. Bacterial Strains and Plasmids Used in This Study
Table 3. Hydroaromatics Biosynthesized under Glucose-Limited
Culture Conditions
strain/
source/
plasmid
relevant characteristics
lacZ∆M15 recA
reference
E. coli:
strain
SP1.1/pKD12.112
0.1 0.8
26 14
SP1.1/pKD12.138
0.1 0.8
28 12
DH5R
RB791
AL0807
JP11123
CL451
KL3
Gibco BRL
ATCC 53622
Marinus
Pittard
Stocker
lab strain
this study
a
K
c
q
lacl
-
R
SAb
F aroL478::Tn10 aroK17::Cm
aroD shiA354::KanR
b
QA
8.1
6.5
3.1
1.2
4.6
12
19
11
1.6
9.4
5.7
1.4
DHSb
aroD25::Tn10
AB2834 aroE353 serA::aroB
c
SA/QA
SA yieldd
13%
20%
10%
14%
14%
29%
8%
18%
SP1.1
RB791 serA::aroB aroL478::Tn10
aroK17::CmR
total yielde
R
SP1.1shiA SP1.1 shiA354::Kan
JB4
plasmid
pSU18
pMAK705 Cm , lacZ', ts-pSC101 replicon
pKK223-3 Ap , Ptac, pBR322 replicon
pKL3.82A Cm , serA::aroB in pMAK705
pIA321
pD2625
SP3
this study
a
Proportional gain of glucose PID control loop. b g/L of shikimic
-
R
KL3 aroL aroK17::Cm aroD25::Tn10 this study
c
acid (SA), quinic acid (QA), 3-dehydroshikimic acid (DHS). (mol SA)/
(mol QA). d (mol SA)/(mol glucose). e (mol SA + QA + DHS)/(mol
R
Cm , PlaclacZ′, p15A replicon
Bartolome
Kushner
Pharmacia
lab plasmid
Coggins
Genencor Intl.
Bonner
R
glucose).
R
R
phosphate pathway catalyzes interchange of C-6 and C-4
carbohydrates. Plasmid pKD12.112, which lacked a tktA insert
Table 1), and pKD12.138, which carried a tktA insert (Table
), were constructed for the purpose of evaluating the impact
P
tacaroE source
serA source
(
1
N. tabaccum aroD‚E source
R
FBR
pKD12.112 Ap , aroF , PtacaroE, serA in pSU18 this study
of transketolase overexpression on the biosynthesis of shikimic
acid. Plasmids pSC5.214A and pJB5.291 also carried a tktA
insert (Table 1). Strategies for increasing in vivo phospho-
enolpyruvate availability were not pursued in this study.
Amplified Expression of Transketolase. The glucose ad-
dition rate to microbe-catalyzed syntheses is a critical consid-
eration. Excessive rates of glucose addition result in accumu-
lation of byproducts such as acetic acid, which can lead to
inhibition of microbial growth and reduction in the yields of
pKD12.138 tktA in pKD12.112
this study
this study
this study
R
pSC5.214A Ap , PtacaroE, serA in pSU18
R
FBR
pJB5.291
Cm , aroF , tktA, serA,
N. tabaccum aroD‚E in pSU18
culture medium (Scheme 1). By inserting a cassette consisting
of aroB flanked with serA nucleotide sequences into the genomic
serA locus, 3-dehydroquinate synthase activity was increased
and DAH accumulation avoided because of the expression of a
second genomic copy of aroB.14 Complementation of the
disrupted genomic serA locus with plasmid-localized serA
provided the basis for plasmid maintenance (Table 1: pKD12.112,
pKD12.138, pSC5.214A, and pJB5.291) in all E. coli constructs
used in this study. The serA locus encodes 3-phosphoglycerate
dehydrogenase, an enzyme necessary for biosynthesis of L-
serine. Growth in culture medium lacking L-serine supplementa-
tion was only possible by expression of plasmid-localized serA.
In addition to aroB-encoded 3-dehydroquinate synthase,
feedback inhibition of shikimate dehydrogenase by shikimic acid
has previously been demonstrated to be a significant impediment
to the flow of carbon through the common pathway and is
reflected by accumulation of 3-dehydroshikimic acid in the
culture medium.1 Shikimate dehydrogenase activity was
therefore increased either by expression of plasmid-localized
E. coli aroE (Table 1: plasmids pKD12.112, pKD12.138, and
pSC5.214A) or N. tabacum aroD‚E (Table 1: plasmid pJB5.291)
from a strong Ptac promoter. A potential complication of
shikimate dehydrogenase overexpression was this enzyme’s
ability to reduce both 3-dehydroquinic acid and 3-dehy-
droshikimic acid.1 Increasing shikimate dehydrogenase activity,
while compensating for the feedback inhibition of this enzyme
by shikimic acid, could also interfere with synthesis of shikimic
acid by reducing intermediate 3-dehydroquinic acid to quinic
acid.
17
the desired product. For this reason, E. coli SP1.1/pKD12.112
and E. coli SP1.1/pKD12.138 were cultured under glucose-
limited conditions at 33 °C, pH 7.0, with dissolved oxygen
maintained at a set point of 10% air saturation. Glucose addition
was controlled by dissolved O2 concentration and was a critical
control parameter during syntheses of shikimic and quinic acids
by E. coli SP1.1/pKD12.112 and E. coli SP1.1/pKD12.138.
When dissolved oxygen levels exceeded the set point value
indicating decreased microbial metabolism, the rate of glucose
addition was increased. When dissolved oxygen levels declined
below the set point value indicating increased microbial
metabolism, the rate of glucose addition was decreased.
For E. coli SP1.1/pKD12.112 and E. coli SP1.1/pKD12.138,
the rate of glucose addition was governed by a proportional-
integral-derivative (PID) control loop. When the proportional
gain (Kc) on the glucose PID control loop was set to 0.1 for
culturing E. coli SP1.1/pKD12.112, substantial concentrations
of quinic acid were synthesized (Table 3). Increasing Kc to 0.8,
and thus increasing the average glucose concentration during
cultivation of E. coli SP1.1/pKD12.112, significantly reduced
5
0
10
the formation of quinic acid (Table 3). Higher levels of quinic
acid were synthesized by E. coli SP1.1/pKD12.138 relative to
E. coli SP1.1/pKD12.112 when cultured with a Kc of 0.1 (Table
3). Unlike E. coli SP1.1/pKD12.112, a substantial concentration
of quinic acid was still formed as a byproduct as evidenced by
the 1.4:1.0:0.68 molar ratio of shikimate/quinate/dehydroshiki-
mate synthesized after 42 h by E. coli SP1.1/pKD12.138 when
a Kc of 0.8 was employed during cultivation (Table 3).
A Kc of 0.1 led to comparatively constant glucose addition
rates during cultivation of E. coli SP1.1/pKD12.112 and E. coli
SP1.1/pKD12.138. By contrast, fermentor runs for these same
constructs that utilized a Kc of 0.8 were difficult to control
beyond 42 h because of oscillations in glucose addition rates
and glucose concentrations in the culture medium. Yields and
titers of synthesized hydroaromatics were also lower (Table 3)
for both E. coli SP1.1/pKD12.112 and E. coli SP1.1/pKD12.138
In addition to feedback-inhibition by aromatic amino acids,
the in vivo activity of overexpressed DAHP synthase is limited
by the availability of its substrate phosphoenolpyruvic acid and
D-erythrose 4-phosphate. A significant body of research has
demonstrated that expression levels of DAHP synthase are
reached at which point further increases in specific activity do
not result in increases in the flow of carbon directed into the
common pathway until the availability of D-erythrose 4-phos-
phate is first increased.16 This requires increasing expression
16a-c
levels of either tktA-encoded transketolase
or talA-encoded
transaldolase1 to increase the rate that the nonoxidative pentose
6d

10176 J. Am. Chem. Soc., Vol. 123, No. 42, 2001
Knop et al.
Figure 2. Equilibration of shikimic acid added to the culture medium
of E. coli SP1.1shiA/pSC5.214A. Legend: shikimic acid (white bar),
quinic acid (black bar), 3-dehydroshikimic acid (gray bar), dry cell
weight (b).
Figure 1. Equilibration of shikimic acid added to the culture medium
of E. coli SP1.1/pSC5.214A. Legend: shikimic acid (white bar), quinic
acid (black bar), 3-dehydroshikimic acid (gray bar), dry cell weight
(b).
when cultured at a Kc of 0.8 relative to a Kc of 0.1. This
deleterious impact of higher Kc values on hydroaromatic
biosynthesis and the failure to control quinic acid formation
during cultivation of E. coli SP1.1/pKD12.138 clearly indicated
that a different strategy was needed to control hydroaromatic
equilibration in constructs expressing amplified levels of trans-
ketolase activity. As a first step, a more detailed understanding
of the mechanism underlying the formation of quinic acid during
shikimic acid biosynthesis was needed when a Kc of 0.1 for
glucose addition was employed. These culture conditions
maintain a steady-state concentration of glucose of approxi-
mately 0.20 mM and are referred to as glucose-limited condi-
tions.
1) was observed. The 1.1:1.0:0.70 molar ratio of shikimate/
quinate/dehydroshikimate formed by 36 h (Figure 1) clearly
illustrated transport of shikimic acid from the culture medium
into the microbial cytoplasm and subsequent operation of the
common pathway in the reverse of its normal biosynthetic
direction.
Inactivation of ShiA-Encoded Shikimate Transport. Mu-
tagenic inactivation of shikimic acid transport provided an
approach to differentiate between de novo biosynthesis of quinic
acid from glucose and formation of quinic acid resulting from
transport and equilibration of shikimic acid. Pittard and co-
workers have characterized a system encoded by a locus
designated as shiA that is apparently responsible for transport
1
8
Hydroaromatic Equilibration under Glucose-Limited Con-
ditions. Microbe-catalyzed conversion of shikimic acid into
of shikimic acid in E. coli. This group has also constructed
E. coli JP11123, which carries catalytically inactive ShiA linked
1
8c
3
-dehydroshikimic and quinic acids had previously been ex-
to a transposable element. P1 phage-mediated transduction
R
amined by harvesting fermentor-grown E. coli SP1.1/pKD12.112
of shiA::Kan from E. coli JP11123 into E. coli SP1.1 produced
1
0
and resuspending these cells in minimal salts medium.
E. coli SP1.1shiA.
Equilibration of shikimic acid added to the culture medium was
then examined under shake-flask conditions.10 The goal in this
study was to examine hydroaromatic equilibration under fer-
mentor conditions similar to those used for de novo synthesis
of shikimic acid from glucose. To ensure that quinic acid
formation was due to equilibration of shikimic acid and not due
to reduction of 3-dehydroquinic acid synthesized from glucose,
E. coli SP1.1/pSC5.214A was constructed (Table 1). Plasmid
pKD12.138A was linearized by digestion with BglII restriction
enzyme, treated with Klenow fragment, and the blunt ends
subsequently ligated to form pSC5.214A. After transformation,
the resulting E. coli SP1.1/pSC5.214A expressed DAHP syn-
thase activity only from its chromosomal aroF, aroG, and aroH
loci. All of the DAHP synthase activity expressed by E. coli
SP1.1/pSC5.214A was thus sensitive to feedback inhibition by
L-phenylalanine, L-tyrosine, or L-tryptophan.3
E. coli SP1.1/pSC5.214A was first cultured under glucose-
limited conditions. L-Phenylalanine (0.7 g), L-tyrosine (0.7 g),
and L-tryptophan (0.35 g) were added to the culture medium (1
L) at the beginning of the fermentor run as required for growth
and at 18 and 30 h to inhibit DAHP synthase activity. The
absence of quinic acid, 3-dehydroshikimic acid, and shikimic
acid in the culture medium after 48 h indicated an absence of
DAHP synthase activity. Shikimic acid (10 g) was subsequently
added at 12 h (Figure 1) to E. coli SP1.1/pSC5.214A cultured
again with aromatic amino acid supplements to negate DAHP
synthase activity and de novo hydroaromatic biosynthesis.
Formation of quinic acid and 3-dehydroshikimic acid (Figure
E. coli SP1.1shiA was then transformed with plasmid
pKD12.138 and the resulting E. coli SP1.1shiA/pKD12.138
cultured under glucose-limited conditions. Synthesis of quinic
acid and 3-dehydroshikimic acid along with shikimic acid was
observed throughout the fermentor run. E. coli SP1.1shiA/
pKD12.138 synthesized 23 g/L of shikimic acid, 22 g/L of
quinic acid, and 9.6 g/L of 3-dehydroshikimic acid over a 48 h
period. To determine if hydroaromatic equilibration was oc-
curring, E. coli SP1.1shiA was transformed with pSC5.214A
to generate a construct lacking shiA-encoded shikimate transport,
DAHP synthase activity, and the attendant capacity for de novo
biosynthesis of hydroaromatics when cultured with aromatic
amino acid supplements. Shikimic acid was added at 12 h to E.
coli SP1.1shiA/pSC5.214A cultured under conditions identical
to those employed to study hydroaromatic equilibration in E.
coli SP1.1/pSC5.214A. Formation of quinic acid and 3-dehy-
droshikimic acid (Figure 2) was observed. Although hydroaro-
matic equilibration was slower than for SP1.1/pSC5.214A, the
0.85:1.0:0.57 molar ratio of shikimate/quinate/dehydroshikimate
formed by 42 h (Figure 2) clearly indicated that transport and
equilibration of shikimic acid were occurring in E. coli
SP1.1shiA/pSC5.214A.
Glucose-Limited Cultures and Methyl-R-D-glucopyrano-
side. As the average glucose concentration increased with
increasing Kc value, a concentration of glucose may have been
19
realized that resulted in catabolic repression of the transcription
of genes involved in shikimic acid transport. This may explain
the success of utilizing increased Kc values to control quinic

Biosynthesis of Shikimic Acid
J. Am. Chem. Soc., Vol. 123, No. 42, 2001 10177
Table 4. Hydroaromatics Biosynthesized under Glucose-Limited
Culture Conditions with Added Methyl-R-D-glucopyranoside and
under Glucose-Rich Culture Conditions
E. coli
SP1.1/pKD12.112
SP1.1/pKD12.138
glucose-
limited
glucose-
rich
glucose-
limited
glucose-
richa
a
b
a
SAc
QA
DHS
SA/QA
SA yield
total yield
27
38
35
52
4.1
11
14
18%
24%
c
0.0
5.3
-
15%
18%
2.0
6.4
20
12%
15%
2.8
8.8
14
19%
25%
c
d
e
f
a
K
of 0.1 with 1 mM methyl-R-D-glucopyranoside. b 25 g/L glucose.
c
c
g/L of shikimic acid (SA), quinic acid (QA), 3-dehydroshikimic acid
DHS). (mol SA)/(mol QA). (mol SA)/(mol glucose). (mol SA +
QA + DHS)/(mol glucose).
d
e
f
(
Figure 4. Biosynthesis of shikimic acid under glucose-limited condi-
tions by E. coli SP1.1/pKD12.138 in the presence of 1 mM methyl-
R-D-glucopyranoside. Legend: shikimic acid (white bar), quinic acid
(black bar), 3-dehydroshikimic acid (gray bar), dry cell weight (b).
SP1.1/pKD12.112 and SP1.1/pKD12.138 and contrasts with the
inability to maintain control of glucose addition typically
encountered after 42 h for cultivation of these shikimate-
synthesizing E. coli at a Kc of 0.8. Most importantly, use of
methyl-R-D-glucopyranoside to inhibit formation of quinic acid
established (Table 4) that amplified expression of transketolase
by E. coli SP1.1/pKD12.138 increased the yield of shikimic
acid, the titer of shikimic acid, and the total yield of hydroaro-
matics synthesized from glucose.
Glucose-Rich Cultures. Perhaps the simplest strategy for
1
9
catabolite repression of shikimate transport is to culture the
microbe under glucose-rich conditions. Such culture conditions
were initially avoided because of precedent indicating that
abundant glucose availability can increase formation of byprod-
ucts such as acetic acid and lower yields in E. coli-catalyzed
conversions. To determine how glucose-rich culture conditions
compared with addition of methyl-R-D-glucopyranoside in
conjunction with glucose-limited conditions, E. coli constructs
SP1.1/pKD12.112A and SP1.1/pKD12.138A were cultured at
33 °C and pH 7.0 with glucose concentrations maintained at
25 g/L (140 mM). Dissolved oxygen levels were controlled at
a set point of 10% air saturation by allowing the impeller speed
to vary over the range 750-1600 rpm. Baffles were also
introduced into the fermentor vessel to enhance mixing and
attendant aeration of cultures.
Under glucose-rich conditions, constructs synthesized shikimic
acid for a longer period of time (60 h) relative to cultivation
under glucose-limited conditions in the absence (42 h) or in
the presence (48 h) of methyl-R-D-glucopyranoside. E. coli
SP1.1/pKD12.112 synthesized 38 g/L of shikimic acid in 12%
yield after 60 h of cultivation (Table 4, Figure 5). The molar
ratio of shikimate/quinate/dehydroshikimate at 60 h was 21:
1.0:3.6. Although its formation was not completely eliminated
under the glucose-rich conditions, less quinic acid was formed
relative to 3-dehydroshikimic acid (Table 4, Figure 5). The yield
of shikimic acid and total yield of hydroaromatics synthesized
by E. coli SP1.1/pKD12.112 under glucose-rich conditions was
only marginally reduced relative to cultivation of this same strain
under glucose-limited conditions in the presence of methyl-R-
D-glucopyranoside (Table 4).
Figure 3. Biosynthesis of shikimic acid under glucose-limited condi-
tions by E. coli SP1.1/pKD12.112 in the presence of 1 mM methyl-
R-D-glucopyranoside. Legend: shikimic acid (white bar), quinic acid
17
(black bar), 3-dehydroshikimic acid (gray bar), dry cell weight (b).
acid formation during biosynthesis of shikimic acid in the
absence of overexpressed transketolase. An alternate approach
might be to add some concentration of a glucose mimic to the
culture medium of a shikimate-synthesizing E. coli. Methyl-R-
D-glucopyranoside was chosen as the glucose mimic, because
it is hydrolytically stable under the solution conditions used to
culture shikimate-synthesizing E. coli constructs. Although
methyl-R-D-glucopyranoside is transported and phosphorylated
by E. coli, it is not otherwise metabolized.20
Cultivation of SP1.1/pKD12.112 with 1 mM methyl-R-D-
glucopyranoside under glucose-limited conditions with a Kc of
0
.1 resulted in the synthesis of 27 g/L of shikimic acid in 15%
yield after 48 h (Table 4, Figure 3). Although 3-dehydroshikimic
acid was also synthesized (Table 4, Figure 3), there was no
detectable formation of quinic acid. The 1 mM concentration
of methyl-R-D-glucopyranoside that inhibited quinic acid forma-
tion was apparently sufficiently low to avoid the reduction in
yield of shikimic acid associated with increasing the Kc to 0.8.
Cultivation of SP1.1/pKD12.138 with 1 mM methyl-R-D-
glucopyranoside under glucose-limited conditions using a Kc
of 0.1 led to the synthesis of 35 g/L of shikimic acid in 19%
yield after 48 h (Table 4, Figure 4). Although quinic acid
formation was not completely eliminated, less quinic acid was
formed than 3-dehydroshikimic acid (Table 4, Figure 4). The
molar ratio of shikimate/quinate/dehydroshikimate was 14:1.0:
Hydroaromatic equilibration was also inhibited when trans-
ketolase-overexpressing E. coli SP1.1/pKD12.138 was cultured
under glucose-rich conditions (Table 4, Figure 6). E. coli SP1.1/
pKD12.138 synthesized 52 g/L of shikimic acid in 18% yield
after cultivation for 60 h (Table 4, Figure 6). The molar ratio
3.5 at 48 h. No control loss of glucose addition was encountered
with methyl-R-D-glucopyranoside present. This allowed fer-
mentations to be controlled for 48 h during cultivation of both

10178 J. Am. Chem. Soc., Vol. 123, No. 42, 2001
Knop et al.
Figure 5. Biosynthesis of shikimic acid under glucose-rich conditions
by E. coli SP1.1/pKD12.112. Legend: shikimic acid (white bar), quinic
acid (black bar), 3-dehydroshikimic acid (gray bar), dry cell weight
Figure 7. Biosynthesis of shikimic acid under glucose-limited condi-
tions by E. coli JB4/pJB5.291. Legend: shikimic acid (white bar), quinic
acid (black bar), 3-dehydroshikimic acid (gray bar), dry cell weight
(b).
(b).
extensively studied.23 In addition to a mutation in its aroE locus
inactivating shikimate dehydrogenase activity, E. coli KL3
contains a second copy of aroB inserted into its serA locus.
Conversion of KL3 from a 3-dehydroshikimate-synthesizing
strain to a shikimate-synthesizing strain began with successive
R
P1 phage-mediated transductions of aroK17::Cm and aroL478::
1
1
R
Tn10. The resulting KL3 aroK::Cm aroL::Tn10 was then
cultured in the presence of D-cycloserine to select for strains
that had lost transposon Tn10 inserts to become sensitive to
S 24
tetracycline resistance (Tc ) while retaining the aroK aroL
phenotype. P1 phage-mediated transduction of an aroD25::Tn10
2
5
R
S
allele into KL3 aroK::Cm aroL::Tc afforded E. coli JB4,
which possessed the desired aroD aroE aroK aroL phenotype.
E. coli JB4 was then transformed with plasmid pJB5.291, which
carried N. tabacum aroD‚E expressed from a Ptac promoter along
Figure 6. Biosynthesis of shikimic acid under glucose-rich conditions
by E. coli SP1.1/pKD12.138. Legend: shikimic acid (white bar), quinic
acid (black bar), 3-dehydroshikimic acid (gray bar), dry cell weight
FBR
with aroF , tktA, and serA inserts (Table 1).
(
b).
Although no reduction in the amount of 3-dehydroshikimic
acid formed during synthesis of shikimic acid was observed
for E. coli JB4/pJB5.291, quinic acid formation was surprisingly
reduced (Figure 7). E. coli JB4/pJB5.291 was cultivated in the
absence of methyl-R-D-glucopyranoside under glucose-limited
conditions at 36 °C, pH 7.0 and with dissolved O2 maintained
at 10% of air saturation. This resulted in the synthesis of 34
g/L of shikimic acid in 15% yield after 66 h with a 29:1.0:5.7
molar ratio of shikimate/quinate/dehydroshikimate. E. coli JB4/
pJB5.291 was cultured at a higher temperature because of the
of shikimate/quinate/dehydroshikimate synthesized by E. coli
SP1.1/pKD12.138 at 60 h under glucose-rich conditions was
1
4:1.0:3.0. Once again, the yield of shikimic acid and total yield
of hydroaromatics synthesized by E. coli SP1.1/pKD12.138
under glucose-rich conditions was not significantly reduced
relative to cultivation of this construct under glucose-limited
conditions in the presence of methyl-R-D-glucopyranoside (Table
4
).
1
sluggish growth characteristics of this construct at 33 °C. H
Nicotiana tabacum aroD‚E. With the success in controlling
NMR analysis of N. tabacum aroD‚E incubated with NADPH
hydroaromatic equilibration under glucose-limited culture condi-
tions by adding methyl-R-D-glucopyranoside or by use of
glucose-rich culture conditions, 3-dehydroshikimic acid became
the major contaminant in E. coli-catalyzed syntheses of shikimic
acid from glucose. In plants, 3-dehydroquinate dehydratase and
(25 mM) and 3-dehydroquinic acid (25 mM) in phosphate buffer
(100 mM, pH 7.5) at 30 °C for 6 h led to a 54:1:6 molar ratio
of shikimate/quinate/dehydroshikimate. Inhibition by shikimic
acid of the binding of 3-dehydroshikimic acid to the shikimate
dehydrogenase portion of N. tabacum aroD‚E was also mea-
sured. Shikimic acid was a competitive inhibitor with Ki ) 1
mM.
shikimate dehydrogenase activities are associated with a single
bifunctional enzyme.2 The gene encoding this bifunctional
enzyme has been isolated from a Nicotiana tabacum cDNA
21
library and designated as aroD‚E. To test the possibility that
metabolite channeling22 in AroD‚E might reduce accumulation
of 3-dehydroshikimic acid, a shikimate-synthesizing E. coli
construct expressing N. tabacum aroD‚E was needed that was
deficient in native aroD-encoded 3-dehydroquinate dehydratase
and native aroE-encoded shikimate dehydrogenase.
Discussion
Hydroaromatic Equilibration. Formation of quinic acid
results from the reduction of 3-dehydroquinic acid. Delineating
the source of the 3-dehydroquinic acid that is reduced thus
becomes central to understanding the mechanism responsible
(
24) Cobos, A.; Fernandez, M. F.; Hernandez, P. E.; Sanz, B. Curr.
Microbiol. 1990, 20, 13.
25) Berson, A. E.; Hudson, D. V.; Waleh, N. S. Mol. Microbiol. 1991,
5, 2261.
E. coli KL3 was selected as the starting point for construction
of the host strain for expression of N. tabacum aroD‚E.
Synthesis of 3-dehydroshikimic acid by this mutant has been
(

Biosynthesis of Shikimic Acid
J. Am. Chem. Soc., Vol. 123, No. 42, 2001 10179
for byproduct formation during shikimic acid biosynthesis. Two
mechanisms are possible. 3-Dehydroquinic acid synthesized
from glucose may be reduced by aroE-encoded shikimate
dehydrogenase prior to its conversion to 3-dehydroshikimic acid
catalyzed by aroD-encoded 3-dehydroquinate dehydratase
sources. Consideration of shikimic acid as a source of carbon
for growth and metabolism also leads to a possible strategy for
repression of shikimate transport. When another carbon source
in addition to glucose is available at any one time for growth
of microbes such as E. coli, glucose is frequently consumed
1
9,28
(Scheme 1). Alternatively, microbe-synthesized shikimic acid
first.
Expression of genes required for transport and
that accumulates in the culture supernatant may be transported
back into the microbial cytoplasm followed by microbe-
catalyzed equilibration. Conversion of transported shikimic acid
into 3-dehydroquinic acid would require shikimate dehydrogen-
ase-catalyzed oxidation of shikimic acid to 3-dehydroshikimic
acid followed by hydration of this hydroaromatic catalyzed by
metabolism of non-glucose carbon sources can be prevented
by catabolite repression.
1
9
Both small (1 mM) concentrations of methyl-R-D-glucopy-
ranoside under glucose-limited culture conditions (0.20 mM
glucose) and glucose-rich (140 mM glucose) culture conditions
successfully inhibited quinic acid formation (Table 4, Figures
3-6) during shikimic acid biosynthesis. These observations are
consistent with catabolic repression of all systems employed
by E. coli to transport shikimic acid. Inhibition of quinic acid
formation attendant with repression of shikimic acid transport,
in turn, indicates that quinic acid formation during shikimic acid
biosynthesis is best viewed as an equilibration of initially
synthesized shikimic acid as opposed to reduction of intermedi-
ate 3-dehydroquinic acid prior to formation of shikimic acid.
In essence, the common pathway is operating in an intact
organism in the reverse of its normal biosynthetic direction.
3-Dehydroshikimic Acid and Shikimate Dehydrogenase
Activity. Formation of quinic acid and 3-dehydroshikimic acid
when shikimic acid was added to the culture medium of E. coli
SP1.1/pSC5.214A (Figure 1) also provided insights into forma-
tion of 3-dehydroshikimic acid during shikimic acid biosyn-
thesis. Accumulation of 3-dehydroshikimic acid has typically
been attributed to the inability of shikimate dehydrogenase to
catalyze the turnover of its substrate into shikimic acid at a rate
sufficiently rapid to avoid accumulation of 3-dehydroshikimic
acid when the flow of carbon into the common pathway is
3
-dehydroquinate dehydratase.
Demonstrating (Figure 1) the ability of E. coli to catalyze
the formation of 3-dehydroshikimic acid and quinic acid from
shikimic acid under fermentor-controlled, glucose-limited condi-
tions was essential to establishing that hydroaromatic equilibra-
tion was a possible mechanism explaining the formation of
quinic acid during shikimic acid biosynthesis. However, dem-
onstration of hydroaromatic equilibration does not necessarily
establish that this is the mechanism involved in formation of
quinic acid during microbial synthesis of shikimic acid from
glucose. Delineation of the role of shikimic acid transport in
quinic acid formation provided an essential second line of
mechanistic information. Continued formation of quinic acid
in a shikimate-synthesizing E. coli construct deficient in shikimic
acid transport would be consistent with a mechanism involving
shikimate dehydrogenase-catalyzed reduction of intermediate
3-dehydroquinic acid prior to synthesis of shikimic acid. A pro-
nounced decline or the absence of quinic acid generated by this
same E. coli construct lacking the ability to transport shikimic
acid from its growth medium to its cytoplasm would be
indicative of equilibration of initially synthesized shikimic acid.
1
4,15
increased.
However, it is also possible that accumulation
of 3-dehydroshikimic acid may reflect rate-limiting hydration
of 3-dehydroshikimic acid by 3-dehydroquinate dehydratase
when the common pathway is operating in the reverse of its
normal biosynthetic direction. Such a mechanism is consistent
with the accumulation of 3-dehydroshikimic acid formed during
equilibration of shikimic acid catalyzed by E. coli SP1.1/
pSC5.214A (Figure 1) and E. coli SP1.1shiA/pSC5.214A (Figure
Transport of shikimic acid in E. coli has been traced to the
shiA locus.18 The availability18c of a transposon-linked mutation
in shiA made construction of a shikimate-synthesizing E. coli
shiA strain a relatively straightforward exercise. One such
construct deficient in shiA-encoded transport of shikimic acid,
E. coli SP1.1shiA/pKD12.138, produced quinic acid and 3-de-
hydroshikimic acid during de novo biosynthesis of shikimic acid
from glucose. This result, upon first inspection, appeared to be
consistent with a mechanism for quinic acid formation involving
reduction of intermediate 3-dehydroquinic acid prior to forma-
tion of shikimic acid. However, subsequent evaluation of the
ShiA-deficient E. coli SP1.1shiA/pSC5.214A revealed that
quinic acid and 3-dehydroshikimic acid were formed when
shikimic acid was added to this construct’s culture medium
2
).
With the successful inhibition of quinic acid formation (Table
, Figures 3-6), most of the accumulated 3-dehydroshikimic
4
acid can be reasonably attributed to feedback inhibition of
1
5
shikimate dehydrogenase by shikimic acid. The standard
approach to reducing accumulation of 3-dehydroshikimic acid
entails overexpression of shikimate dehydrogenase by plasmid-
localization of the aroE gene. However, 3-dehydroshikimic acid
accumulated as the major byproduct during shikimic acid
biosynthesis when hydroaromatic equilibration was inhibited.
This prompted the evaluation of the aroD‚E gene from N.
tabacum. The proximity of 3-dehydroquinate dehydratase and
shikimate dehydrogenase activities in AroD‚E might allow for
metabolic channeling of enzyme-bound 3-dehydroquinic acid
directly into shikimic acid formation and avoid accumulation
(
Figure 2). This observation indicated that E. coli employed at
least one system for shikimic acid transport that was not encoded
by shiA.
Additional insight into the role of shikimic acid transport and
formation of quinic acid came from consideration of shikimic
acid’s role as a source of carbon for microbial growth and
metabolism. Numerous microbes can use hydroaromatics such
21
2
6
as shikimic acid as a sole source of carbon during growth.
2
of 3-dehydroshikimic acid in the cytoplasm. Minimizing or
One of these species, Klebsiella spp., is evolutionarily related
to E. coli.27 Although unable to utilize shikimic acid as a sole
completely avoiding accumulation of 3-dehydroshikimic acid
9
c
may be important for plants, given the precedented enzymatic
source of carbon for growth, shikimic acid transport in E. coli
may be an evolutionary remnant of a former capacity to
catabolize shikimic acid and may also reflect the multiple
transport systems microbes often possess for transport of carbon
29
and nonenzymatic conversion of this hydroaromatic into gallic
acid. Heterologous expression of N. tabacum aroD‚E had
(28) Gottschalk, G. Bacterial Metabolism, 2nd ed.; Springer-Verlag: New
York, 1985; p 185.
(
26) Giles, N. H.; Case, M. E.; Baum, J.; Geever, R.; Huiet, L.; Patel,
V.; Tyler, B. Microbiol. ReV. 1985, 49, 338.
27) Jensen, R. A. Mol. Biol. EVol. 1985, 2, 92.
(29) (a) Richman, J. E.; Chang, Y.-C.; Kambourakis, S.; Draths, K. M.;
Almy, E.; Snell, K. D.; Strasburg, G. M.; Frost, J. W. J. Am. Chem. Soc.
1996, 118, 11587.
(

10180 J. Am. Chem. Soc., Vol. 123, No. 42, 2001
Knop et al.
previously been demonstrated with the successful complemen-
tation of E. coli aroD and E. coli aroE auxotrophs.2
relative to the shikimic acid synthesized by E. coli SP1.1/
pKD12.138 was 2-fold higher than that realized with E. coli
JB4/pJB5.291, this higher level of quinic acid contamination
was still readily removed from product shikimic acid by
crystallization.
1
E. coli JB4/pJB5.291, which lacked wild-type E. coli aroD
and aroE activities and expressed N. tabacum aroD‚E, still
accumulated substantial quantities of 3-dehydroshikimic acid.
N. tabacum aroD‚E-encoded shikimate dehydrogenase activity
was also just as sensitive to feedback inhibition as wild-type E.
coli aroE-encoded shikimate dehydrogenase. An unanticipated
surprise when culturing E. coli JB4/pJB5.291 under glucose-
limited conditions was the low quantities of quinic acid that
were formed during shikimic acid biosynthesis under glucose-
limited culture conditions. Reduced formation of quinic acid
was ultimately traced to the selectivity of N. tabacum AroD‚E
favoring reduction of 3-dehydroshikimic acid to shikimic acid
over reduction of 3-dehydroquinic acid to quinic acid. This
selectivity adds a new twist to the list of enzymes displaying
dehydrogenase activity toward intermediates in the common
pathway of aromatic amino acid biosynthesis.
The first reported microbial synthesis of quinic acid relied
on heterologous expression in a 3-dehydroquinate-synthesizing
E. coli construct of the Klebsiella pneumoniae qad gene
encoding quinate dehydrogenase.30 Qad catalyzed the reduction
of 3-dehydroquinic acid to quinic acid. In K. pneumoniae, Qad
oxidizes quinic acid to 3-dehydroquinic acid, which is the first
step in the catabolism of quinic acid.31 Synthesis of quinic acid
by E. coli strains such as SP1.1/pKD12.112 and SP1.1/
pKD12.138 (Table 3, Figure 1) is thus noteworthy given that
E. coli has not previously been reported to biosynthesize quinic
acid or to exploit this hydroaromatic as a sole source of carbon
for growth. The ability of aroE-encoded shikimate dehydrogen-
ase to catalyze the reduction of 3-dehydroquinic acid in addition
to its reduction of 3-dehydroshikimic acid (Scheme 1) to
shikimic acid had apparently evaded the intense mechanistic
scrutiny afforded enzymes in the common pathway. With N.
tabacum AroD‚E, a true shikimate dehydrogenase has been
identified given this enzyme’s selective reduction of 3-dehy-
droshikimic acid over 3-dehydroquinic acid.
Peak shikimic acid concentrations using E. coli JB4/pJB5.291
took significantly longer to achieve (60 h) relative to use of E.
coli SP1.1/pKD12.138 under glucose-rich (54 h) or glucose-
limited (48 h) culture conditions. This increased length of time
for synthesis of shikimic acid reflects the slower growth
characteristics of E. coli JB4/pJB5.291, which had to be cultured
at 36 °C relative to the 33 °C culture conditions where vigorous
growth was observed for E. coli SP1.1/pKD12.138. The 18-
19% (Table 4) yield of shikimic acid and the 24-25% (Table
4) combined yield of shikimic acid, quinic acid, and 3-dehy-
droshikimic acid were essentially unaffected (Table 4) by the
use of glucose-rich versus glucose-limited conditions for
cultivation of E. coli SP1.1/pKD12.138. However, the 52 g/L
of shikimic acid in fermenation broths synthesized by E. coli
SP1.1/pKD12.138 when cultured under glucose-rich conditions
was substantially higher relative to the 35 g/L synthesized by
this same construct under glucose-limited conditions in the
presence of methyl-R-D-glucopyranoside.
Another outcome of controlling hydroaromatic equilibration
during shikimic acid biosynthesis is the ability to take full
advantage of the improvement in carbon flow directed into the
common pathway (Scheme 1) by overexpression of transketolase
and the attendant increase in the availability of D-erythrose
4-phosphate for aromatic amino acid biosynthesis.16 E. coli
SP1.1/pKD12.112, which expresses wild-type expression levels
of transketolase, synthesized 38 g/L of shikimic acid in 12%
yield. This titer and yield contrast sharply with the 52 g/L of
shikimic acid synthesized in 18% yield under identical culture
conditions by E. coli SP1.1/pKD12.138, which overexpresses
transketolase. Prior to delineating the mechanism associated with
hydroaromatic equilibration and developing methodologies
appropriate for controlling this phenomenon, it was not possible
to evaluate the impact of transketolase overexpression on the
yields and titers of shikimic acid synthesized by E. coli from
glucose. With the ability to control hydroaromatic equilibration
when D-erythrose 4-phosphate availability is increased, microbe-
catalyzed synthesis of shikimic acid from glucose can now be
used as a paradigm to study the impact on product yields and
titers when the in vivo availability of phosphoenolpyruvic acid
is increased.
Purity, Titer, and Yield. Three methods are now available
for controlling hydroaromatic equilibration during shikimic acid
biosynthesis. Under glucose-limited culture conditions, shiki-
mate-synthesizing E. coli constructs can be cultivated in the
presence of low concentrations (1 mM) of methyl-R-D-glucopy-
ranoside. Hydroaromatic equilibration can also be avoided under
glucose-limited culture conditions by expressing N. tabacum
aroD‚E in an E. coli host lacking native aroE-encoded shikimate
dehydrogenase activity. Finally, cultivating shikimate-synthesiz-
ing E. coli constructs under glucose-rich conditions inhibits
hydroaromatic equilibration. These methods can be individually
evaluated and compared with one another on the basis of the
purity, titer, and yield of shikimic acid synthesized by E. coli
from glucose.
Experimental Section
Culture Medium. All solutions were prepared in distilled, deionized
water. LB (Luria-Bertani) medium32 contained (1 L) Bacto tryptone
(
10 g), Bacto yeast extract (5 g), and NaCl (10 g). M9 salts (1 L)
contained Na HPO (6 g), KH PO (3 g), NH Cl (1 g), and NaCl (0.5
g). M9 minimal medium contained D-glucose (10 g), MgSO (0.12 g),
2
4
2
4
4
3
-Dehydroshikimic acid is separated from shikimic acid and
4
quinic acid in culture supernatants by acidification, heating, and
absorption of the resulting protocatechuic acid on activated
carbon. Removal of quinic acid requires a fractional recrystal-
lization of crude shikimic acid product after removal of
and thiamine hydrochloride (0.001 g) in 1 L of M9 salts. M9 medium
(1 L) was supplemented where appropriate with L-phenylalanine (0.040
g), L-tyrosine (0.040 g), L-tryptophan (0.040 g), p-hydroxybenzoic acid
(0.010 g), potassium p-aminobenzoate (0.010 g), and 2,3-dihydroxy-
benzoic acid (0.010 g). L-Serine and shikimic acid were added to a
final concentration of 40 mg/L where indicated. Antibiotics were added
where appropriate to the following final concentrations: ampicillin (Ap),
3-dehydroshikimic acid. Contamination levels of quinic acid thus
become important, because the presence of quinic acid in excess
of 10% (mol/mol) relative to shikimic acid is difficult to remove
by crystallization. Although the concentration of quinic acid
5
0 µg/mL; chloramphenicol (Cm), 20 µg/mL; kanamycin (Kan), 50
µg/mL; and tetracycline (Tc), 25 µg/mL. Solutions of M9 salts, MgSO
4
,
and glucose were autoclaved individually and then mixed. Solutions
of amino acids, aromatic vitamins, shikimic acid, thiamine hydrochlo-
(
30) Draths, K. M.; Ward, T. L.; Frost, J. W. J. Am. Chem. Soc. 1992,
1
14, 9725.
(
31) (a) Mitsuhashi, S.; Davis, B. D. Biochim. Biophys. Acta 1954, 15,
2
68. (b) Davis, B. D.; Gilvarg, C.; Mitsuhashi, S. Methods Enzymol. 1955,
(32) Miller, J. H. Experiments in Molecular Genetics; Cold Spring Harbor
2
, 300.
Laboratory: Plainview, NY, 1972.

Biosynthesis of Shikimic Acid
J. Am. Chem. Soc., Vol. 123, No. 42, 2001 10181
ride, and antibiotics were sterilized through 0.22-µm membranes. Solid
medium was prepared by addition of Difco agar to a final concentration
of 1.5% (w/v) to the liquid medium.
growth on M9 containing serine, growth on LB, and no growth on LB
containing Cm.
RB791 serA::aroB was subjected to two successive P1 phage-
R
The standard fermentation medium (1 L) contained K
2
HPO
4
(7.5
mediated transductions to transfer the aroL478::Tn10 and aroK::Cm
loci of ALO80711 into the genome. P1 phage was propagated from E.
coli ALO807, and transductions were carried out using protocols
described by Miller.32 RB791 serA::aroB aroL::Tn10 aroK::Cm was
selected on the basis of the following growth characteristics: no growth
on M9 containing aromatic amino acids and aromatic vitamins; no
growth on M9 containing serine; growth on M9 containing aromatic
amino acids, aromatic vitamins, and serine; and growth on LB
g), ammonium iron (III) citrate (0.3 g), citric acid monohydrate (2.1
g), L-phenylalanine (0.7 g), L-tyrosine (0.7 g), L-tryptophan (0.35 g),
R
and concentrated H
to pH 7.0 by addition of concentrated NH
following supplements were added immediately prior to initiation of
the fermentation: glucose, MgSO (0.24 g), p-hydroxybenzoic acid
0.010 g), potassium p-aminobenzoate (0.010 g), 2,3-dihydroxybenzoic
2
SO
4
(1.2 mL). Fermentation medium was adjusted
4
OH before autoclaving. The
4
(
R
acid (0.010 g), and trace minerals, including (NH
4
)
6
(Mo
7
O
24)‚4H
‚5H
2
O
O
containing Tc and Cm. RB791 serA::aroB aroL::Tn10 aroK::Cm was
renamed SP1.1.
(
0.0037 g), ZnSO
4
‚7H
2
O (0.0029 g), H
‚4H
3
BO
3
(0.0247 g), CuSO
4
2
(0.0025 g), and MnCl
2
2
O (0.0158 g). Methyl-R-D-glucopyranoside
SP1.1shiA. E. coli SP1.1shiA was prepared by transduction of the
R
18c
was added to a final concentration of 1 mM where indicated. Glucose,
MgSO (1 M), and methyl-R-D-glucopyranoside (100 mM) solutions
shiA354::Kan locus from JP11123
into SP1.1. SP1.1shiA was
4
selected on the basis of its ability to grow on LB containing Cm, Tc,
and Kan. The inability of SP1.1shiA to grow in the absence of aromatic
amino acid and aromatic vitamin supplementation and in the absence
of serine supplementation was confirmed separately.
were autoclaved separately, while solutions of aromatic vitamins and
trace minerals were sterilized through 0.22-µm membranes. Shikimic
acid was added to the fermentation medium where indicated. A solution
of shikimic acid (10 g) in water (10 mL) was adjusted to pH 7 by
addition of KOH and sterilized through a 0.22-µm membrane prior to
addition to the fermentation medium. In cases where additional aromatic
amino acids were added while the batch was in progress, L-phenylal-
anine (0.7 g), L-tyrosine (0.7 g), and L-tryptophan (0.35 g) were added
as dry powders. Antifoam (Sigma 204) was added as needed.
JB4. Construction of E. coli JB4 began with KL3,2 which already
contains the serA::aroB locus as well as a mutation in the aroE locus.
Two successive P1 phage transductions using phage propagated from
ALO80711 were carried out on KL3 to afford KL3 aroK::Cm aroL::
Tn10. To make the strain sensitive to tetracyline without losing the
mutation in aroL, colonies were selected for resistance to D-cyclo-
serine.24 A single colony of KL3 aroK::Cm aroL::Tn10was introduced
into 5 mL of LB and grown overnight at 37 °C with agitation. The
overnight culture was diluted (1:500) into LB, and the cells were grown
for an additional 5 h. The cells were then collected by centrifugation,
resuspended in LB containing Tc, and cultured for 1 h at 37 °C with
agitation. D-Cycloserine was added to the culture to a final concentration
of 6 mM, and cell growth was continued for 3.5 h. Cells were collected
from 1 mL of the culture, washed with 1 mL of LB, and finally
resuspended in 1 mL of LB. A portion (10 µL) of the suspension was
used to inoculate 5 mL of fresh LB, and the cells were grown overnight.
After a second identical treatment with D-cycloserine, the surviving
cells were plated onto LB. Colonies which grew were then replicate-
plated onto LB and LB/Tc plates. Approximately 10% of the colonies
that grew on LB were sensitive to Tc. One of these colonies was named
3a
R
R
Bacterial Strains and Plasmids. E. coli K-12 strain RB791 was
obtained from the American Type Culture Collection (ATCC strain
1
1
5
(
3622). E. coli AL0807 was provided by Professor M. G. Marinus
1
8c
University of Massachusetts), E. coli JP11123 was provided by
Professor A. J. Pittard (University of Melbourne), and E. coli CL451
25
was provided by Professor B. A. D. Stocker (Stanford University).
Plasmid constructions were carried out in E. coli DH5R, which is
available from Gibco BRL Products. Homologous recombinations
3
3
utilized plasmid pMAK705, which was obtained from Professor S.
R. Kushner (University of Georgia).
Genetic Manipulations. Standard protocols were used for construc-
3
4
tion, purification, and analysis of plasmid DNA. T4 DNA ligase and
Large Fragment of DNA polymerase I (Klenow fragment) were
purchased from Gibco BRL Products. Calf intestinal alkaline phos-
phatase was purchased from Roche Molecular Biochemicals. PCR
amplifications were carried out as previously described.34 Each
amplification reaction (0.1 mL) contained 10 mM KCl, 20 mM Tris-
R
S
KL3 aroK::Cm aroL::Tn . Subsequent transduction of KL3 aroK::
R
S
25
Cm aroL::Tn with P1 phage propagated from CL451 was followed
by selection of colonies for resistance to Tc. KL3 aroK::Cm aroL::
Tn aroD::Tn10 colonies were selected on the basis of the following
R
S
HCl (pH 8.8), 10 mM (NH
4
)
2
SO
4
, 2 mM MgSO
4
, 0.1% Triton X-100,
dATP (0.2 mM), dCTP (0.2 mM), dGTP (0.2 mM), dTTP (0.2 mM),
template DNA (0.02 µg-1 µg), 0.5 µM of each primer, and 2 units of
Vent polymerase. Primers were synthesized by the Macromolecular
Structure Facility at Michigan State University.
growth characteristics: no growth on M9 containing serine; no growth
on M9 containing shikimic acid and serine; growth on M9 containing
aromatic amino acids, aromatic vitamins, and serine; and growth on
R
S
LB containing Cm and Tc. KL3 aroK::Cm aroL::Tn aroD::Tn10 was
renamed JB4.
SP1.1. Construction of SP1.1 began with the homologous recom-
bination of the aroB gene into the serA locus of E. coli RB791 resulting
in RB791 serA::aroB. Plasmid pKL3.82A, which contained the serA::
3
6
pKD12.112A. Plasmid pKD12.112 is a pSU18-derived plasmid
that encodes aroF , aroE, serA, and Ap . Construction of pKD12.112
began with pKL4.20B, a 3.6-kb plasmid that was described elsewhere.
Plasmid pKL4.20B contains a 1.2-kb fragment that encodes aroF
FBR
R
23a
23a
aroB locus, was described previously. The procedure for homologous
FBR 13
3
3,35
.
recombination has also been described.
Competent RB791 was
3
7
PCR amplification of the Ptac aroE sequence from pIA321 afforded
a 1.2-kb fragment that was subsequently localized in pKL4.20B to
afford pKD12.036. A 1.9-kb serA-encoding fragment was liberated from
pD2625 by digestion with EcoRV and DraI. Insertion of the serA
fragment into pKD12.036 yielded pKD12.047. Resistance to chloram-
phenicol conferred by pKD12.047 was eliminated by inserting the gene
that confers ampicillin resistance into the NcoI site of pKD12.047. PCR
transformed with pKL3.82A and plated out onto LB plates containing
Cm. Colonies were observed after incubation of the plates at 44 °C for
1
2 h. The resulting cointegrates were inoculated into 5 mL LB and
grown at 30 °C for 12 h to allow excision of the plasmid from the
genome. Cultures were diluted (1:20 000) in LB, and two more growth
cycles were performed at 30 °C for 12 h to enrich the cultures for
more rapidly growing cells that had lost the temperature-sensitive
replicon. Cultures were then diluted (1:20 000) into LB and grown at
R
38
amplification of a 1.0-kb Ap -encoding fragment from pUC18
followed by insertion of the fragment into the NcoI site of pKD12.047
afforded the 7.7-kb plasmid pKD12.112.
pKD12.138. Insertion of a 2.2-kb tktA-encoding fragment16b,39 into
pKD12.112 afforded the 9.9-kb plasmid pKD12.138.
4
4 °C for 12 h to promote plasmid loss from the cells. Serial dilutions
of each culture were spread onto LB plates and incubated at 30 °C for
2 h. The resulting colonies were screened on multiple plates to select
1
the desired recombination product. RB791 serA::aroB was isolated on
the basis of the following growth characteristics: no growth on M9,
pSC5.214A. Plasmid pSC5.214A is identical to pKD12.138 except
FBR
that pSC5.214A does not contain a functional aroF
gene. To
(
33) Hamilton, C. M.; Aldea, M.; Washburn, B. K.; Babitzke, P.;
Kushner, S. R. J. Bacteriol. 1989, 171, 4617.
34) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular Cloning: A
Laboratory Manual; Cold Spring Harbor Laboratory: Plainview, NY, 1990.
35) Ohta, K.; Beall, D. S.; Mejia, J. P.; Shanmugan, K. T.; Ingram, L.
O. Appl. EnViron. Microbiol. 1991, 57, 893.
(36) Bartolome, B.; Jubete, Y.; Martinez, E.; de la Cruz, F. Gene 1991,
102, 75.
(37) Anton, I. A.; Coggins, J. R. Biochem. J. 1988, 249, 319.
(38) Yanisch-Perron, C.; Vieira, J.; Messing, J. Gene 1985, 33, 103.
(39) Sprenger, G. A. Biochim. Biophys. Acta 1993, 1216, 307.
(
(

10182 J. Am. Chem. Soc., Vol. 123, No. 42, 2001
Knop et al.
inactivate the aroFFBR gene of pKD12.138, advantage was taken of a
10% air saturation because of residual initial glucose in the medium.
This lasted for approximately 0.5-1 h before glucose (65% w/v) feeding
commenced. The glucose feed PID control parameters were set to 0.0
13
unique BglII recognition site located internal to the gene. BglII
digestion of pKD12.138 was followed by treatment of the linearized
plasmid with the Klenow fragment and a cocktail of dNTPs. Subsequent
religation of the plasmid afforded pSC5.214A.
s (off) for the derivative control (τ
D
) and 999.9 s (minimum control
was set to 950% to achieve a K
c
of 0.8 as required.
action) for the integral control (τ
of 0.1 and 125% to achieve a K
I
). X
p
c
pJB5.291. Plasmid pJB5.291 is a pSU18-derived plasmid that
FBR
.
encodes aroF , tktA, serA, and the N. tabacum aroD E locus.
Fed-Batch Fermentations (Glucose-Rich Conditions). For fer-
Construction of pJB5.291 began with pJB3.141, a 7.6-kb plasmid that
mentations that employed glucose-rich conditions, a stainless steel baffle
FBR
.
21
1
encoded aroF , tktA, serA. The N. tabacum aroDE locus was cloned
2
cage consisting of four / ” × 5” baffles was placed in the fermentation
40
as a 1.9-kb fragment behind the Ptac promoter of pKK223-3 to afford
vessel. Three staged methods were used to maintain the D.O.
concentration at 10% air. With the airflow at an initial setting of 0.06
L/L/min, the 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. With the impeller rate constant at 750 rpm, the mass flow
controller then maintained the D.O. concentration by increasing the
airflow rate from 0.06 L/L/min to a preset maximum of 1.0 L/L/min.
After the preset maxima of 750 rpm and 1.0 L/L/min were reached,
the third stage of the fermentation was initiated in which glucose (65%
w/v) was added to the vessel at a rate sufficient to maintain a glucose
concentration of 20-30 g/L for the remainder of the run. Airflow was
maintained at 1.0 L/L/min, and the impeller was allowed to vary in
order to maintain the D.O. concentration at 10% air saturation. The
impeller speed typically varied from 750 to 1600 rpm during the
remainder of the run.
.
pJB5.272A. Liberation of a 2.2-kb Ptac aroDE fragment from pJB5.272A
followed by insertion of the fragment into pJB3.141 afforded pJB5.291.
Fed-Batch Fermentations (General). Fermentations employed a
2
.0 L working capacity B. Braun M2 culture vessel. Utilities were
supplied by a B. Braun Biostat MD controlled by a DCU-1. Data
+
acquisition utilized a Dell Optiplex Gs 5166M personal computer (PC)
equipped with B. Braun MFCS/Win software (v1.1). Temperature and
pH were controlled with PID control loops. Glucose feeding was
controlled with a PID control loop unless otherwise indicated. Tem-
perature was maintained at 33 °C for all fermentations with the
exception of the JB4/pJB5.291, which was cultured at 36 °C. pH was
maintained at 7.0 by addition of concentrated NH
Dissolved oxygen (D.O.) was measured using a Mettler-Toledo 12 mm
sterilizable O sensor fitted with an Ingold A-type O permeable
4 2 4
OH or 2 N H SO .
2
2
membrane. D.O. was maintained at 10% air saturation. The initial
glucose concentration in the fermentation medium ranged from 20 to
Analysis of Fermentation Broth. Samples (5-10 mL) of fermenta-
tion broths were taken at the indicated timed intervals. Cell densities
were determined by dilution of fermentation broth with water (1:100)
followed by measurement of absorption at 600 nm (OD600). Dry cell
weight (g/L) was calculated using a conversion coefficient of 0.43 g/L/
OD600. The remaining fermentation broth was centrifuged to obtain cell-
free broth.
3
0 g/L.
Inoculants were started by introduction of a single colony picked
from an agar plate into 5 mL of M9 medium. For strains possessing a
plasmid that confers resistance to ampicillin, ampicillin was added to
the medium at each step of the inoculant preparation. Cultures were
grown at 37 °C with agitation at 250 rpm until they were turbid (12-
Glucose concentrations in cell-free broth were measured using the
Glucose Diagnostic Kit obtained from Sigma. Solute concentrations
36 h) and subsequently transferred to 100 mL of M9 medium. Cultures
1
were grown at 37 °C and 250 rpm for an additional 8-12 h. After the
culture reached an appropriate OD600 (1.0-3.0), the inoculant was
transferred into the fermentation vessel and the batch fermentation was
initiated (t ) 0 h).
Fed-Batch Fermentations (Glucose-Limited Conditions). Three
staged methods were used to maintain D.O. concentrations at 10% air
saturation during the fermentations. With the airflow at an initial setting
of 0.06 L/L/min, the D.O. concentration was maintained by increasing
the impeller speed from its initial set point of 50 rpm to its preset
maximum of 940 rpm. With the impeller rate constant at 940 rpm, the
mass flow controller then maintained the D.O. concentration by
increasing the airflow rate from 0.06 L/L/min to a preset maximum of
in the cell-free broth were quantified by H NMR. Solutions were
concentrated to dryness under reduced pressure, concentrated to dryness
one additional time from D
2 2
O, and then redissolved in D O containing
a known concentration of the sodium salt of 3-(trimethylsilyl)propionic-
1
4
2,2,3,3-d acid (TSP, Lancaster Synthesis Inc). All H NMR spectra
were recorded on a Varian VXR-300 FT-NMR Spectrometer (300
MHz). Concentrations were determined by comparison of integrals
corresponding to each compound with the integral corresponding to
TSP (δ ) 0.00 ppm). A standard concentration curve was determined
for each metabolite using solutions of authentic, purified metabolites.
Compounds were quantified using the following resonances: shikimic
acid (δ 4.57, d, 1 H); quinic acid (δ 4.16, m, 1 H); and 3-dehy-
droshikimic acid (δ 4.28, d, 1 H).
1
.0 L/L/min. At constant impeller speed and constant airflow rate, the
D.O. concentration was finally maintained at 10% air saturation for
the remainder of the fermentation by oxygen sensor-controlled glucose
feeding. At the beginning of this stage, the D.O. concentration fell below
Acknowledgment. Research was supported by a grant from
the National Institutes of Health.
(40) Brosius, J.; Holy, A. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 6929.
JA0109444