Appl Microbiol Biotechnol (2004) 65: 56–60
DOI 10.1007/s00253-004-1560-3
APPLIED GENE TICS AN D MO LECULA R BIO TECH NOLOGY
M. Lee . G. M. Smith . M. A. Eiteman . E. Altman
Aerobic production of alanine by Escherichia coli aceF ldhA
mutants expressing the Bacillus sphaericus alaD gene
Received: 23 October 2003 / Revised: 8 December 2003 / Accepted: 18 December 2003 / Published online: 4 February 2004
# Springer-Verlag 2004
Abstract Alanine was produced from glucose in an
Escherichia coli aceF ldhA double mutant strain that
contained the pTrc99A-alaD plasmid expressing Bacillus
sphaericus alanine dehydrogenase. The aceF gene en-
codes one of the proteins of the pyruvate dehydrogenase
complex, and therefore this strain required acetate as an
additional carbon source. The ldhA gene encodes
fermentative lactate dehydrogenase, a competitor of ala-
nine dehydrogenase for the substrate pyruvate. Fermenta-
tions included an oxygenated growth phase followed by an
oxygen-limited alanine production phase. The lowest
value for the mass transfer coefficient (kLa) studied during
the production phase yielded the greatest alanine. With
feeding of glucose and NH4Cl, 32 g/l alanine accumulated
in 27 h with a yield of 0.63 g alanine generated per gram
glucose consumed.
Introduction
L-Alanine is used commercially as a food additive, and can
be produced from L-aspartate via an enzymatic process
using L-aspartic-β-decarboxylase (Chibata et al. 1969).
Several recent microbial studies have also sought to
generate D- and L-alanine through fermentation. For
example, strain DAN 75 of the genus Arthrobacter,
selected for an inability to grow on D-alanine, accumulated
L-alanine to 75.6 g/l (with 1.2 g/l D-alanine) in 120 h with
a mass yield of over 50% from glucose (Hashimoto and
Katsumata 1998). Anaerobically, Zymomonas mobilis
transformed with the alaD gene from Bacillus sphaericus
produced 7.5 g/l L-alanine when the essential cofactor
thiamine PPi was limited (Uhlenbusch et al. 1991).
Limiting this cofactor reduced carbon flux through
M. Lee . G. M. Smith . M. A. Eiteman (*) . E. Altman
Center for Molecular BioEngineering, Driftmier Engineering,
University of Georgia,
Athens, GA 30602, USA
e-mail: eiteman@engr.uga.edu
Tel.: +1-706-5420833
Fax: +1-706-5428806
pyruvate decarboxylase, a pathway that competes with
alanine generation in this organism. A study using
Escherichia coli transformed with the Arthrobacter sp.
HAP1 L-alanine dehydrogenase gene resulted in the
accumulation of 2.9 g/l DL-alanine under aerobic condi-
tions and 8.1 g/l under oxygen-limited conditions in shake
flasks (Katsumata and Hashimoto 1996).
The biochemical pathways relevant to the synthesis of
alanine by the model organism E. coli are shown in Fig. 1.
The starting substrate glucose is metabolized via glycol-
ysis to pyruvate, which is ultimately generated from
phosphoenol pyruvate both by the phosphotransferase
glucose transport system and pyruvate kinase. The con-
version of pyruvate to alanine is most likely mediated by a
glutamate-pyruvate transaminase; however, this enzyme
has yet to be identified. Several other enzymes compete
for the substrate pyruvate under aerobic conditions: the
pyruvate dehydrogenase complex generates acetyl CoA,
lactate dehydrogenase generates lactate, pyruvate oxidase
generates acetate, and malic enzyme generates malate.
In other bacteria, such as numerous Bacillus species, the
conversion of pyruvate to alanine is mediated by alanine
dehydrogenase. We wanted to investigate whether signif-
icant alanine could be produced in E. coli by over-
producing an alanine dehydrogenase. Given the numerous
competing enzymes, however, any strategy to produce
alanine in E. coli would have to eliminate other key
enzymatic steps. The most significant carbon flux from
pyruvate is to acetyl CoA via the pyruvate dehydrogenase
complex. This activity could be eliminated using an aceF
mutant that knocks out the dihydrolipoyl transacetylase
component of this complex. A previous study has shown
that an aceF mutant of E. coli will accumulate nearly 40 g/
l pyruvate under aerobic conditions (Tomar et al. 2003).
Another significant route of pyruvate metabolism is via
lactate dehydrogenase (Futai and Kimura 1977). More-
over, like alanine dehydrogenase, this enzyme uses NADH
as a cosubstrate and is thus a direct competitor for alanine
production. Fortunately, lactate dehydrogenase could be
eliminated using an ldhA mutant. Pyruvate formate lyase is
active only during anaerobic conditions (Knappe and

Fig. 1 Biochemical pathways in Escherichia coli at the pyruvate
node that impact the accumulation of alanine. Key enzymatic steps
include: 1 probable glutamate-pyruvate transaminase, 2 pyruvate
oxidase, 3 pyruvate dehydrogenase complex, 4 malic enzyme, 5
phosphotransferase system, 6 pyruvate kinase, 7 lactate dehydroge-
nase
Sawers 1990; Kessler and Knappe 1996), and therefore
would not compete with alanine dehydrogenase when
oxygen is present. For malic enzyme, the direction of
malate to pyruvate is slightly favored thermodynamically
(Harary et al. 1953) and, although this route may divert
pyruvate in cases of pyruvate accumulation, it is not likely
to be a major competitor with alanine dehydrogenase for
the substrate pyruvate. Relatively little is known about the
role that pyruvate oxidase plays in pyruvate assimilation,
and this enzyme might be a significant competitor with
alanine dehydrogenase under certain growth conditions,
but this was beyond the scope of this initial study. The
goal of this study therefore was to investigate whether
significant alanine accumulated in E. coli aceF ldhA under
oxygen-limited conditions.
Materials and methods
Bacterial strains
E. coli CGSC6162 [DC80, F+ aceF10 fadR2000 tyrT58 (AS)
adhE80 mel-1] was the parent strain used in this study (Clark and
Cronan 1980). The ldhA mutant was introduced into CGSC6162 by
P1 transduction using E. coli strain CGSC7726 [F+λ− rpoS396(Am)
rph-1 ldhA::Kan Δ(pflAB::Cam)] as the donor (Bunch et al. 1997).
The resulting strain was designated ALS887 (CGSC6162 ldhA::
Kan).
Construction of pTrc99A-alaD
Because it is well-characterized (Ohashima and Soda 1979;
Ohshima et al. 1990), the B. sphaericus alanine dehydrogenase
gene was selected and amplified using the polymerase chain reaction
(PCR). The enzyme has been shown to have high activity in Bacillus
species (Ohashima and Soda 1979) with Km values of 10 μM
(NADH), 1.7 mM (pyruvate) and 28.2 mM (ammonium). Pfu DNA
polymerase was used instead of Taq DNA polymerase and
chromosomal DNA prepared from B. sphaericus IFO3525
(ATCC10208, DSM5019) served as the DNA template. Primers
were designed based on the published B. sphaericus alanine
57
dehydrogenase gene sequence (Kuroda et al. 1990) and contained
a BamHI (GGATCC) restriction site and Shine-Dalgarno sequence
at the beginning of the amplified fragment, and a HindIII
(AAGCTT) restriction site at the end of the amplified fragment;
forward primer 5′ TAC TAT GGA TCCAGG AGG AAC AGC
TATGAA GAT TGG TAT TCC AAA GGA AAT TAA AAA C 3′;
reverse primer 5′ ATA GCG ATC GAT AGC GGT AAG CTT ATT
ATT GGA TTA ATT CAT CCA CAT TCA CAT ATG 3′ (the
BamHI, Shine Dalgarno, ATG start, and HindIII sites are under-
lined). The resulting 1.2 kb PCR product was gel isolated, digested
with BamHI and HindIII and ligated into the pTrc99A expression
vector digested with the same two enzymes. Initially, the B.
sphaericus alanine dehydrogenase gene was cloned and sequenced,
but not designated (Kuroda et al. 1990). Subsequently, the gene was
named alaD (Uhlenbusch et al. 1991), the designation we use in this
study. In Bacillus subtilis, the alanine dehydrogenase gene has been
named ald (Siranosian et al. 1993).
Media and growth conditions
Cells of E. coli ALS887 pTrc99A-alaD were first grown at 20 ml
volume in an agitated screw-top test tube with medium composed of
(per liter) 15.0 g glucose, 3.0 g acetic acid, 6.0 g succinic acid, 2.5 g
tryptone, 2.5 g NaCl, and 1.25 g yeast extract. After 3 h growth,
10 ml was used to inoculate 100 ml medium composed of (per liter)
15.0 g glucose, 3.0 g acetic acid, 6.0 g succinic acid, 10.0 g
tryptone, 2.5 g yeast extract, 3.0 g KH2PO4, 6.0 g Na2HPO4, 6.0 g
NH4Cl, 0.14 g CaCl2·H2O, and 0.25 g MgSO4·7H2O, in a 250 ml
baffled shake flask. Cells were grown at 250 rpm (19 mm radius of
orbit) for 6 h and then used to inoculate a fermenter of the same
composition as the shake flask except 40 g/l glucose. Fermentations
of 1.5 l initial volume were conducted using a BioFlow 2000 (New
Brunswick Scientific, New Brunswick, N.J.). Air was supplied
continuously at 1.0 l/min. After 3–4 h of growth in the fermenter,
1.0 mM isopropyl-β-D-thiogalactopyranoside (IPTG) was added for
gene induction. During the first 11 h of a fermentation, the agitation
was 1,000 rpm, a rate that insured that the dissolved oxygen
remained above 20% of saturation. At 11 h the alanine production
phase was initiated by reducing the agitation rate to a lower constant
value as described in the text. Oxygen mass transfer coefficients
(kLa) for each experimental agitation rate were determined in a
separate experiment using the static sparging method (Wise 1951)
with identical medium and fermenter system. At 15 h, additional
glucose and NH4Cl was added as described in the text to replenish
those components that had been consumed for the generation of cell
mass and alanine. All media contained 100 mg/l ampicillin,
fermentations were carried out at 37°C and the pH was maintained
at pH 7.0 throughout.
Analyses
Cell growth was measured using optical density (OD) at 550 nm
(DU-650 UV-vis spectrophotometer, Beckman, San Jose, Calif.). We
found dry cell weight (DCW, g/l) to correlate with OD according to
DCW =0.4×OD. Samples were centrifuged (10,000 g for 10 min at
25°C), and the supernatant analyzed for glucose, succinate, lactate,
and acetate, pyruvate by high pressure liquid chromatography
(HPLC) as previously described (Eiteman and Chastain 1997).
Alanine was analyzed by HPLC using an Aminex HPX-87C
carbohydrate column with 0.1 M Ca(NO3)2 effluent and a refractive
index detector (Waters 2410, Millipore, Milford, Mass.). Ammoni-
um ion was determined for selected samples using the Berthelot
reaction (Chaney and Marback 1962).
58
Enzyme assay
Cell-free extracts were prepared by centrifuging twice (5,000 g for
20 min at 4°C), decanting the supernatant and resuspending the cells
in 10 mM pH 7.0 potassium phosphate buffer. The cell material was
disrupted using an SLM-Aminco French pressure cell (Spectronic
Instruments, Rochester, N.Y.), the cell debris removed by centrif-
ugation (20,000 g for 20 min at 4°C), and the extract used for
measuring alanine dehydrogenase activity (Ohashima and Soda
1979). One unit of enzyme activity is the quantity of enzyme
required to produce 1.0 μmol pyruvate in 1 min. The total protein in
the cell-free extract was determined using a BCA protein assay kit
(Pierce, Rockville, Ill.).
Results
Construction of ALS887 pTrc99A-alaD
An ldhA mutant was introduced into E. coli CGSC6162
and designated ALS887. We did not detect lactate
dehydrogenase activity in this strain. The B. sphaericus
alanine dehydrogenase gene was cloned into the pTrc99A
expression vector and then transformed into ALS887. The
resulting strain, ALS887 pTrc99A-alaD, was used to
examine the production of alanine in E. coli. Because this
strain contains an aceF mutant, and thus has no pyruvate
dehydrogenase activity, it must be supplemented with
acetate in addition to glucose. Succinate is added to
Fig. 2A, B Production of alanine in E. coli aceF ldhA pTrc99A-
alaD. Cells were grown for 11 h under fully oxygenated conditions,
then provided oxygen at a constant kLa of 18 h−1; 30 g glucose and
7.5 g NH4Cl were added into the 1.5 l fermenter at 15 h. A ◯
Glucose, ■ alanine, □ pyruvate. B ● OD, △ succinate, ▽ acetate
improve growth (Tomar et al. 2003). Furthermore, because
alanine dehydrogenase consumes ammonium in the con-
version of pyruvate to alanine, we fed NH4Cl solutions at
various intervals during fermentation.
Effect of oxygen mass transfer
We first studied alanine accumulation during an oxygen-
limited production phase in which the oxygen mass
transfer coefficient (kLa) was held constant by maintaining
the agitation speed. This phase commenced after 11 h of
growth under fully oxygenated conditions. Because the
original glucose (and NH4Cl) could become depleted, at
15 h we added 110 ml of a solution containing 30 g
glucose and 7.5 g NH4Cl into the 1.5 l medium. Several
values of kLa were examined ranging from about 7 to
109 h−1, and Fig. 2 shows an example result with a kLa of
18 h−1. For all the fermentations, an OD of 9–15 was
reached at the time that the production phase commenced
when a substantial amount of pyruvate had been formed
(between 10 and 22 g/l). About 50–70% of the succinate
and acetate supplied originally had been consumed during
the growth phase, and the concentration of these two
compounds remained unchanged during the production
phase. Although some alanine (usually about 3 g/l) was
generated during the growth phase, the majority of the
alanine accumulated after 11 h. The activity of alanine
dehydrogenase measured at 15 h and 21 h was always
between 0.20 and 0.35 U/mg protein. The alanine yield
based on the alanine generated and glucose consumed
after 15 h was a strong function of kLa (Fig. 3). At lower
kLa values (7 h−1, 18 h−1) the majority of glucose carbon
was converted to alanine (yield on glucose of 0.53 g/g–
0.69 g/g) and pyruvate concentration and OD decreased
slowly. At the higher two values of kLa (52 h−1, 109 h−1)
cells continued to grow slowly and accumulated some
pyruvate. Because the greatest alanine concentration of
15.7 g/l, with the greatest yield, occurred using the lowest
value of kLa, the corresponding agitation rate was selected
for subsequent studies.
Fig. 3 Alanine yield from glucose at different oxygen mass transfer
rates during the production phase of the fermentation

Fig. 4A, B Production of alanine in E. coli aceF ldhA pTrc99A-
alaD. Cells were grown for 11 h under fully oxygenated conditions,
then provided oxygen at a constant kLa of 7 h−1; 30 g glucose and
22.5 g NH4Cl were added into the 1.5 l fermenter at 15 h and again
at 23 h. A ◯ Glucose, ■ alanine, □ pyruvate. B ● OD, △ succinate, ▽
acetate
Effect of ammonium addition
In our studies on the effect of kLa on alanine accumulation,
a consistent result was that alanine generation occurred at
the highest rate immediately following the addition of
glucose and NH4Cl at 15 h. For example, for the
fermentation shown in Fig. 2, the volumetric rate of
alanine generation was 1.54 g l−1 h−1 between 15 h and
17 h, 1.27 g l−1 h−1 between 17 h and 19 h, but essentially
zero after 19 h even though the glucose concentration then
was 4–7 g/l. In order to determine whether NH4Cl was
limiting the conversion of pyruvate to alanine via alanine
dehydrogenase, we determined the ammonium ion con-
centration at the end of these fermentations, and found the
ammonium concentration to be a minimum of 20 mmol/l.
We then repeated those fermentations with the lowest
value of kLa (7 h−1). In this case, however, we provided
three times the NH4Cl (22.5 g) with the 30 g glucose at
15 h and then both materials again at 23 h. The resulting
alanine production rate was consistently above 2.0 g l−1 h
−1
between 15 h and 27 h (Fig. 4), significantly greater
than previously when less NH4Cl was added. In duplicate
experiments, the alanine concentration reached 32 g/l in
27 h, but the alanine concentration did not increase further
regardless of whether additional glucose and NH4Cl were
added. The overall alanine yield on glucose averaged
59
0.63 g/g, and the alanine yield on glucose after 15 h
averaged 0.81 g/g.
Discussion
This study focuses on alanine production from glucose in
E. coli overexpressing alanine dehydrogenase. Theoreti-
cally, 2 mol alanine can be generated biochemically from
1 mol glucose (mass yield of 0.978), and the pathway is
redox balanced; i.e., the complete conversion of 1 mol
glucose into 2 mol alanine would result in no net exchange
between the cofactors NADH and NAD. Thus, to achieve
the maximum theoretical yield, the NADH generated in
glycolysis is necessary for the final reduction of pyruvate
to alanine via alanine dehydrogenase. Since in normal
growth oxygen will consume NADH via both oxidative
phosphorylation and the enzyme NADH oxidase, the rate
of oxygen transfer should have a significant influence on
alanine generation. Our results support this hypothesis,
with the lowest value of kLa resulting in the greatest
alanine production in terms of yield and productivity. We
did not observe the accumulation of formate, which might
be generated via pyruvate formate lyase. Since this
enzyme is very sensitive to oxygen (Kessler and Knappe
1996), activation of pyruvate formate lyase was avoided
by the presence of a small quantity of oxygen. An
improved method for alanine generation might be to delete
the pfl gene in order to prevent the action of this enzyme
even under anaerobic conditions. The consequence of such
a mutation on cell growth and alanine production is
unknown, but another pfl strain of E. coli has been
successfully used for succinate accumulation in a similar
aerobic cell growth/anaerobic production process (Chat-
terjee et al. 2001; Vemuri et al. 2002).
The rate of alanine production slowed as alanine
accumulated in the fermentations with “low” ammonium
concentration. Hashimoto and Katsumata (1998) observed
a similar phenomenon and attributed the decreasing rate
with the amount of alanine accumulating to changes in
glucose concentration. The equilibrium between pyruvate
and alanine strongly favors alanine with an equilibrium
constant (Keq = [NH4+][pyruvate][NADH][H+]/[alanine]
[NAD]) of about 10−14–10−15 M2 (Grimshaw and Cleland
1981; Ohshima et al. 1990). The lowest concentration of
ammonium we measured in any fermentation was
20 mmol/l. Since the NADH/NAD ratio in aerobic cultures
of wild-type E. coli is about 0.02 (De Graef et al. 1999),
by the equilibrium relationship one might have expected
an alanine concentration at least 1,000 times greater than
pyruvate. Nevertheless, we observed a significantly great-
er alanine production rate with a greater concentration of
ammonium ion (but the same concentration of glucose).
Moreover, this rate of alanine production was maintained
as glucose approached depletion and alanine accumulated.
Zymomonas has been shown to concentrate alanine
intracellularly to over 50 times the external alanine
concentration, although this ratio of internal to external
alanine decreased with time (Uhlenbusch et al. 1991). The
60
internal concentrations of the several metabolites involved
in the alanine-pyruvate equilibrium were not determined in
this study. The production of alanine may be limited by a
transport barrier that is overcome by a greater concentra-
tion of ammonium.
During the fermentations in which two additions of
NH4Cl were made into the fermenter, alanine production
eventually ceased. After these NH4Cl additions, the
concentration of the chloride ion in the medium reached
about 0.6 M, a concentration which itself may have
prevented further generation of alanine. An alternative
method of supplying cells with the substrate ammonium
without accumulation of chloride would establish whether
this ion limited the cells in alanine accumulation.
This study recorded the highest concentration of alanine
generated by E. coli, and the highest productivity and yield
on glucose reported for any organism via a microbial
fermentation process.
Acknowledgements The authors thank Sarah Lee, Kris DeWitt
and Patrick Reeves for assistance with the experimental aspects of
this project. Financial support from the United States Department of
Energy Biobased Products Industry Education Program and the
Georgia Experiment Station is gratefully acknowledged.
References
Bunch PK, Mat-Jan F, Lee N, Clark DP (1997) The ldhA gene
encoding the fermentative lactate dehydrogenase of Escheri-
chia coli. Microbiology 143:187–195
Chaney AL, Marback EP (1962) Modified reagents for determina-
tion of urea and ammonia. Clin Chem 8:130–132
Chatterjee R, Millard CS, Champion K, Clark DP, Donnelly MI
(2001) Mutation of the ptsG gene results in increased
production of succinate in fermentation of glucose by Escher-
ichia coli. Appl Environ Microbiol 67:148–154
Chibata TI, Kakimoto T, Kato J (1969) Process for producing L-
alanine. US Patent 3 458 400
Clark D, Cronan JE Jr (1980) Escherichia coli mutants with altered
control of alcohol dehydrogenase and nitrate reductase. J
Bacteriol 141:177–183
De Graef MR, Alexeeva S, Snoep JL, de Mattos MJT (1999) The
steady-state internal redox state (NADH/NAD) reflects the
external redox state and is correlated with catabolic adaptation
in Escherichia coli. J Bacteriol 181:2351–2357
Eiteman MA, Chastain MJ (1997) Optimization of the ion-exchange
analysis of organic acids from fermentation. Anal Chim Acta
338:69–75
Futai M, Kimura H (1977) Inducible membrane-bound L-lactate
dehydrogenase from Escherichia coli. J Biol Chem 252:5820–
5827
Grimshaw CE, Cleland WW (1981) Kinetic mechanism of Bacillus
subtilisL-alanine dehydrogenase. Biochemistry 20:5650–5655
Harary I, Korey SR, Ochoa S (1953) Biosynthesis of dicarboxylic
acids by carbon dioxide fixation. VII. Equilibrium of “malic”
enzyme reaction. J Biol Chem 203:595–604
Hashimoto S, Katsumata R (1998) L-Alanine fermentation by an
alanine racemase-deficient mutant of the DL-alanine hyperpro-
ducing bacterium Arthrobacter oxydans HAP-1. J Ferment
Bioeng 86:385–390
Katsumata R, Hashimoto S (1996) Process for producing alanine.
US Patent 5 559 016
Kessler D, Knappe J (1996) Anaerobic dissimilation of pyruvate. In:
Escherichia coli and Salmonella. Neidhardt FC et al (eds) ASM
Press, Washington D.C.
Knappe J, Sawers G (1990) A radical-chemical route to acetyl CoA
—the anaerobically induced pyruvate formate-lyase system of
Escherichia coli. FEMS Microbiol Rev 75:383–398
Kuroda S, Tanizawa K, Sakamoto Y, Tanaka H, Soda K (1990)
Alanine dehydrogenase from two Bacillus species with distinct
thermostabilities: molecular cloning, DNA and protein se-
quence determination, and structural comparison with other
NAD(P)+-dependent dehydrogenases. Biochemistry 29:1009–
1015
Ohashima T, Soda K (1979) Purification and properties of alanine
dehydrogenase from Bacillus sphaericus. Eur J Biochem
100:29–39
Ohshima T, Sakane M, Yamazaki T, Soda K (1990) Thermostable
alanine dehydrogenase from thermophilic Bacillus sphaericus
DSM 462. Eur J Biochem 191:715–720
Siranosian KJ, Ireton K, Grossman AD (1993) Alanine dehydroge-
nase (ald) is required for normal sporulation in Bacillus subtilis.
J Bacteriol 175:6789–6796
Tomar A, Eiteman MA, Altman E (2003) The effect of acetate
pathway mutations on the production of pyruvate in Escher-
ichia coli. Appl Microbiol Biotechnol 62:76–82
Uhlenbusch I, Sahm H, Sprenger GA (1991) Expression of an L-
alanine dehydrogenase gene in Zymomonas mobilis and
excretion of L-alanine. Appl Environ Microbiol 57:1360–1366
Vemuri GN, Eiteman MA, Altman E (2002) Succinate production in
dual-phase Escherichia coli fermentations depends on the time
of transition from aerobic to anaerobic conditions. J Ind
Microbiol Biotechnol 28:325–332
Wise WS (1951) The measurement of the aeration of culture media.
J Gen Microbiol 5:167–177