Effect of pyruvate carboxylase overexpression on the physiology of Corynebacterium glutamicum.

Abstract

Pyruvate carboxylase was recently sequenced in Corynebacterium glutamicum and shown to play an important role of anaplerosis in the central carbon metabolism and amino acid synthesis of these bacteria. In this study we investigate the effect of the overexpression of the gene for pyruvate carboxylase (pyc) on the physiology of C. glutamicum ATCC 21253 and ATCC 21799 grown on defined media with two different carbon sources, glucose and lactate. In general, the physiological effects of pyc overexpression in Corynebacteria depend on the genetic background of the particular strain studied and are determined to a large extent by the interplay between pyruvate carboxylase and aspartate kinase activities. If the pyruvate carboxylase activity is not properly matched by the aspartate kinase activity, pyc overexpression results in growth enhancement instead of greater lysine production, despite its central role in anaplerosis and aspartic acid biosynthesis. Aspartate kinase regulation by lysine and threonine, pyruvate carboxylase inhibition by aspartate (shown in this study using permeabilized cells), as well as well-established activation of pyruvate carboxylase by lactate and acetyl coenzyme A are the key factors in determining the effect of pyc overexpression on Corynebacteria physiology.

Full text

Page 1

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Nov. 2002, p. 5422–5428
0099-2240/02/$04.00?0DOI: 10.1128/AEM.68.11.5422–5428.2002
Copyright © 2002, American Society for Microbiology. All Rights Reserved.
Vol. 68, No. 11
Effect of Pyruvate Carboxylase Overexpression on the Physiology of
Corynebacterium glutamicum
Mattheos A. G. Koffas, Gyoo Yeol Jung, Juan C. Aon, and Gregory Stephanopoulos*
Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
Received 29 April 2002/Accepted 5 August 2002
Pyruvate carboxylase was recently sequenced in Corynebacterium glutamicum and shown to play an important
role of anaplerosis in the central carbon metabolism and amino acid synthesis of these bacteria. In this study
we investigate the effect of the overexpression of the gene for pyruvate carboxylase (pyc) on the physiology of
C. glutamicum ATCC 21253 and ATCC 21799 grown on defined media with two different carbon sources, glucose
and lactate. In general, the physiological effects of pyc overexpression in Corynebacteria depend on the genetic
background of the particular strain studied and are determined to a large extent by the interplay between
pyruvate carboxylase and aspartate kinase activities. If the pyruvate carboxylase activity is not properly
matched by the aspartate kinase activity, pyc overexpression results in growth enhancement instead of greater
lysine production, despite its central role in anaplerosis and aspartic acid biosynthesis. Aspartate kinase
regulation by lysine and threonine, pyruvate carboxylase inhibition by aspartate (shown in this study using
permeabilized cells), as well as well-established activation of pyruvate carboxylase by lactate and acetyl
coenzyme A are the key factors in determining the effect of pyc overexpression on Corynebacteria physiology.
Corynebacteria belong to the large group of gram-positive
bacteria with a high G?C content, which constitutes the Acti-
nomyces subdivision together with genera such as Streptomyces,
Propionibacteria, and Arthrobacter. Certain saprophytic coryne-
bacteria, such as Corynebacterium glutamicum and its close
relative Brevibacterium flavum are used predominantly in in-
dustrial fermentation processes, such as lysine production, be-
cause of the almost complete lack of regulation of their lysine
biosynthesis metabolic pathway and their high capability for
lysine excretion (7, 8). Lysine is used in increasing volumes as
a food additive for poultry and pig breeding (2), a clear reflec-
tion of the improvement of living conditions around the world.
Extensive studies during the past 15 years have enhanced
our understanding of metabolic flux distribution and control in
C. glutamicum during lysine fermentations. These studies led
to the identification of phosphoenolpyruvate/pyruvate as a crit-
ical branch point, controlling the supply of anaplerotic carbon
for the biosynthesis of aspartic acid family amino acids. Ox-
aloacetate replenishment in particular was determined to be a
critical step in lysine production (38, 42, 43). The first anaple-
rotic enzyme to be investigated in this context was phos-
phoenolpyruvate carboxylase (PEPC), whose presence has
been well established in C. glutamicum previously (9, 10, 27).
Using a PEPC-deficient mutant strain (ppc?), it was shown
that PEPC is dispensable in lysine production and has little
effect on cell growth (11, 30). Furthermore, a double mutant
(ppc?pyk?) showed a substantial reduction of both growth and
lysine productivity (28, 29), pointing to the importance of pyru-
vate in supplying anaplerotic carbon for growth and lysine
production. These results suggested that alternative anaple-
rotic pathways, which most probably use pyruvate as a sub-
strate, operate in corynebacteria. Further evidence for the
direct carboxylation of pyruvate was provided by the develop-
ment of an independent enzymatic assay that utilized perme-
abilized cells (31, 41). The presence of pyruvate carboxylase
was finally confirmed by sequencing the pyc gene in coryne-
bacteria (18, 32). Studies with the pyc deletion mutant as well
as the pyc ppc double mutant showed that pyruvate carboxylase
is the essential anaplerotic pathway and that no further anaple-
rotic pathways exist in C. glutamicum (32). However, despite
the important role of this enzyme in growth and product for-
mation, no results on its physiological effects have been re-
ported.
In the present study, we have investigated the effect of pyc
overexpression on C. glutamicum physiology, especially in
terms of growth and lysine production. Overexpression of
pyruvate carboxylase must be studied in conjunction with other
enzymes synthesizing or depleting metabolites that regulate
their respective activities. Of particular importance is aspartate
kinase and, by extension, the genetic background of the strain
determining the regulation of these enzymes. We report here
our findings with pyc overexpression in two different strains: C.
glutamicum ATCC 21253, which has a regulated aspartate ki-
nase, and C. glutamicum ATCC 21799, which has a deregulated
aspartate kinase (15), and for two different carbon sources,
glucose and lactate.
MATERIALS AND METHODS
Strains and media. Two strains, C. glutamicum ATCC 21253, auxotrophic for
L-homoserine (or L-threonine plus L-methionine) and L-leucine, and C. glutami-
cum ATCC 21799, auxotrophic for L-leucine and pantothenate and aminoethyl-
cysteine resistant (AECr), were used in this study. For convenience, C. glutami-
cum ATCC 21253 and ATCC 21799 are referred to simply as 21253 and 21799,
respectively. Other strains are summarized in Table 1. The defined medium
developed by Kiss and Stephanopoulos (17) was used as a basal medium sup-
plemented with 10 mg of pantothenate per lites for 21799. For some experiments
focusing on examining the effect of the carbon source, 20 g of lactate per liter was
used as a carbon source instead of glucose.
Analysis of cell mass, carbon sources, and intra- and extracellular amino
* Corresponding author. Mailing address: Department of Chemical
Engineering, MIT, Room 56-469, Cambridge, MA 02139. Phone: (617)
253-4583. Fax: (617) 253-3122. E-mail: gregstep@mit.edu.
5422

Page 2

acids. Cell mass was measured by using a UVIKON 930 spectrophotometer
(Research Instruments, San Diego, Calif.). Dry cell weight was estimated based
on the correlation 1 OD600unit ? 0.28 g dry cell weight (DCW)/liter. The
amount of L-lysine accumulated in the medium and the amount of remaining
L-threonine were quantitatively measured as o-phtalaldehyde derivatives (16) on
a Hewlett-Packard (Waldronn, Germany) 1050 high-performance liquid chro-
motography HPLC system equipped with an AminoQuant column (Agilent
Technologies, Wilmington, Del.). The levels of intracellular amino acids were
measured in the extracts of the cells inactivated by silicon oil centrifugation to
produce a cytosolic fraction of metabolites based on the method developed by
Rottenberg (35) using silicon oil (Sigma, St. Louis, Mo.) and a glass bead beater
(Biospec, Washington, N.C.). Glucose and lactate levels in the medium were
measured by using the corresponding kits for each carbon source purchased from
Boehringer (Mannheim, Germany).
Plasmid construction. Three different plasmids were constructed, allowing
insertion of upstream and downstream untranslated regions of different sizes
along with the structural pyc gene. The Escherichia coli-C. glutamicum vector
pMAGK (?) was used for gene cloning in C. glutamicum. This vector was
constructed based on the multiple-cloning site of the vector pMAL-p2X (New
England Biolabs, Beverly, Mass.) and the broad-host-range replication site pEP2
(24).
For the first construct, cosmid IIIF10, which was used to obtain the pyc
sequence (18), was digested with HindIII and the pyc gene was cloned as a 9-kb
fragment into the E. coli vector pCR-Script as well as the E. coli-C. glutamicum
vector pMAGK(?), giving rise to plasmids pCR9pc-Script and pMAGK9pc,
respectively. Plasmid pMAGK9pc was then transformed into strain 21799. For
the second construct, plasmid pCR9pc-Script was digested with 18 restriction
enzymes that do not affect the pyc gene based on its restriction map: AflII, ApaI,
AvrII, BspHI, DraI, EcoRV, HindIII, KasI, NcoI, NdeI, NheI, NspI, SacI, SpeI,
SphI, SspI, XbaI, and XmnI. Using DNA electrophoresis, several bands were
detected after the digest, one of which (a 4-kb DNA fragment) contained the pyc
gene (identified by PCR). The 5? and 3? ends of this DNA fragment (possible
sticky) were blunted using Pfu DNA polymerase as described by the manufac-
turer (Stratagene Inc., La Jolla, Calif.), and the fragment was then introduced
into vector pMAGK(?), giving rise to plasmid pMAGK4pc. This plasmid was
also introduced into strain 21799.
For the third construct, cosmid IIIG7, one of the four cosmids containing the
pyc gene (18) was first digested with HindIII, yielding a 12-kb DNA fragment that
contains the pyc gene. This fragment was introduced into vector pCR-Script,
generating plasmid pCR12pc-Script. The latter was next digested with ScaI-SspI,
and the resulting 7-kb DNA fragment containing the pyruvate carboxylase gene
was introduced into vector pMAGK(?), generating plasmid pKD7. This plasmid
was transformed into strains 21799 and 21253, generating strains 21799(pKD7)
and 21253(pKD7), respectively. The resulting three constructs of plasmids are
described briefly in Fig. 1.
DNA preparation and transformation. Restriction enzymes, T4 DNA ligase,
and reagents were purchased from New England Biolabs or Boehringer Mann-
heim (Indianapolis, Ind.). Cosmid and plasmid DNA were prepared using Qia-
prep spin columns (Qiagen, Valencia, Calif.), and DNA was extracted from
agarose gels with the Qiaex kit (Qiagen). Plasmid DNA from C. glutamicum was
isolated as described by Colon (6). E. coli was transformed by the CaCl2method
(36), and C. glutamicum was transformed by electroporation as described by
Liebl et al. (22).
Pyruvate carboxylase assay. The activity of pyruvate carboxylase was deter-
mined using permeabilized C. glutamicum cells by the method of Uy et al. (41).
Cells collected from the culture, cultivated for 9 h (just before the depletion of
threonine in the medium), were washed twice with 20 ml of 50 mM Tris-HCl
buffer (pH 6.3) containing 50 mM NaCl, resuspended to an optical density at 600
nm (OD600) of about 150 in 10 mM EDTA buffer (pH 7.4) containing 20%
(vol/vol) glycerol, and then frozen at ?20°C.
For the permeabilization step, the frozen cells were slowly thawed on ice and
then mixed with a solution of 2.5% (wt/vol) hexadecyltrimethylammonium bro-
mide (CTAB), resulting in a 0.3% (wt/vol) final CTAB concentration. The
duration of permeabilization was about 1 min, which was previously reported to
be the optimal permeablization duration (41).
The permeabilized-cell suspension was immediately used to assay pyruvate
carboxylase activity from the measured transformation rate of pyruvate into
oxaloacetate. A proper amount of each permeabilized-cell suspension was added
to 1 ml of a reaction mixture containing 100 mM Tris-HCl (pH 7.3), 25 mM
NaHCO3, 5 mM MgCl2, 3 mM pyruvate, and 4 mM ATP. To find the appropriate
activity of each cell five different amounts of permeabilized cells, 0.5, 1, 2, 3, and
4 mg, were incubated for three different reaction periods, 0.5, 1, and 2 min, at
FIG. 1. Three constructs containing the pyruvate carboxylase gene and different sizes of untranslated regions used in this study. The C.
glutamicum-E. coli plasmid pMAGK(?) was used as the vector for the constructs shown. Thick arrows describe the structural genes of pyruvate
carboxylase, and thin lines represent the untranslated regions. Numbers in parentheses are the positions of the restriction sites from the beginning
of each construct.
TABLE 1. Strains used in this study
Strain Genotype and characteristicsReference or source
C. glutamicum
21253
21253(pKD7)
21799
21799(pKD7)
hom
hom pyc::kan
AECr
AECr, pyc::kan
26
This study
19
This study
E. coli
DH5?
hsdR recA12
VOL. 68, 2002EFFECT OF PYC OVEREXPRESSION IN C. GLUTAMICUM5423

Page 3

30°C. The reaction was stopped by addition of 80 ?l of 30% (wt/vol) o-phospho-
ric acid, and cell debris was removed by centrifugation (25,000 ? g at 4°C for 15
min). In each sample, the unconverted pyruvate concentration was determined
by a colorimetric method using a pyruvate assay kit purchased from Sigma
Diagnostics Inc. One unit of pyruvate carboxylase activity was defined as the
amount of enzyme converting 1 nmol of pyruvate per min.
Detection of in vivo-biotinylated proteins. Protein extracts from different
strains (C. glutamicum and E. coli) were first separated by sodium dodecyl
sulfate-polyacrylamide gel electrophoresis SDS-PAGE (7% [wt/vol] acrylamide)
and transferred to a nitrocellulose membrane. Biotinylated proteins were then
detected with avidin-alkaline phosphatase conjugate (Bio-Rad, Hercules, Calif.)
directly on the membrane as previously described (18). Two biotinylated en-
zymes were detected: one in the region of 70 kDa, which has previously been
shown to correspond to the biotinylated subunit of acetyl coenzyme A (CoA)
carboxylase (31, 32), and one in the region of 120 kDa, corresponding to pyruvate
carboxylase (18).
Cell cultures. All bacterial cultures in this study were conducted with 50 ml of
defined medium (described above) in 300-ml shake flasks incubated at 30°C with
rotation at 220 rpm. To measure the cell mass and extracellular amino acid
production, samples of 1 ml each were taken from the culture broth at the proper
cultivation time during the incubation. For the measurement of pyruvate car-
boxylase activity and intracellular amino acid concentrations, cells were har-
vested 9 h after the start of the cultivation (just before the depletion of threonine
in the culture broth) and measurements were made as described above.
RESULTS
Overexpression of pyruvate carboxylase in corynebacteria.
Overexpression of pyruvate carboxylase in C. glutamicum was
achieved using three different plasmid constructs, each con-
taining the pyc gene in a different-size DNA fragment. Pyru-
vate carboxylase polypeptides were detected by Western blot
analysis (Fig. 2). For cell cultivation on glucose (Fig 2a), all
three recombinant strains showed similar amount of polypep-
tide synthesis. However, when lactate was used as the carbon
source, different amounts of polypeptides were detected for
the different plasmid constructs transformed into strain 21799
(Fig. 2b). The strain harboring plasmid pKD7 [21799(pKD7)],
containing the pyc gene together with a 2.3-kb upstream DNA
region, showed the largest amount of pyruvate carboxylase
polypeptide. Due to the significant difference in pyc overex-
pression obtained with plasmid pKD7, strains harboring this
plasmid were used in the subsequent pyruvate carboxylase
overexpression studies.
Overexpression of the pyc gene was also obtained by trans-
forming strain 21253, as shown in Fig. 3.
Pyruvatecarboxylase activities
corynebacteria. Table 2 summarizes the pyruvate carboxylase
activities for the two recombinant strains cultivated on two
carbon sources, glucose and lactate. It can be seen that the use
of lactate invariably produces higher pyruvate carboxylase ac-
tivities, which is in accordance with observations reported be-
fore (31, 41). We found that when the recombinant strains
were grown on lactate, their pyruvate carboxylase activity was
greater regardless of the type of host cell used. While the same
behavior was observed with strain 21799 on glucose, no differ-
ence in pyruvate carboxylase activity was observed between the
recombinant and parental strains in strain 21253 grown on
glucose. A major difference between strains 21253 and 21799
was the feedback inhibition of aspartate kinase in strain 21253,
as a result of which it is possible that aspartate accumulates
in
pyc-overexpressing
FIG. 2. Western blot of pyruvate carboxylase expressed in the pa-
rental and recombinant 21799 strains harboring three different plasmid
constructs of pyc genes. Avidin-alkaline phosphatase was used for the
detection of the biotinylated enzymes. (a) Cells grown on glucose; (b)
cells grown on lactate. Lanes: 1, parental strain; 2, pMAGK4pc; 3,
pMAGK9pc; 4, pKD7. AcCoACase, acetyl-CoA carboxylase.
FIG. 3. Western blot of pyruvate carboxylase expressed in the pa-
rental and recombinant 21253 strains harboring pKD7 grown on glu-
cose (a) or lactate (b). Lanes: 1, parental; 2, pKD7. AcCoACase,
acetyl-CoA carboxylase.
TABLE 2. Comparison of pyruvate carboxylase activities in C.
glutamicum ATCC 21253 with those in C. glutamicum ATCC 21799
on different carbon sources
Strain
Pyruvate carboxylase activity
(U/mg DCW ? SD) on:
Glucose Lactate
21253
21253(pKD7)
21799
21799(pKD7)
58 ? 5
58 ? 6
30 ? 4
53 ? 7
97 ? 7
161 ? 10
36 ? 3
89 ? 8
5424KOFFAS ET AL.APPL. ENVIRON. MICROBIOL.

Page 4

intracellularly in this strain and feedback inhibits pyruvate
carboxylase. Such would not be the case for strain 21799, where
no negative regulation was exercised on aspartate kinase by
threonine and accumulating lysine. Since aspartate is a com-
mon inhibitor of pyruvate carboxylases, the in vivo pyruvate
carboxylase activity should be attenuated in strain 21253 com-
pared to strain 21799. To investigate this hypothesis, the effect
of aspartate on pyruvate carboxylase activity was also studied.
Aspartate inhibits pyruvate carboxylase activity in C. glu-
tamicum. Since differences in aspartate kinase regulation in
strains 21253(pKD7) and 21799(pKD7) could lead to different
intracellular aspartate levels, the effect of aspartate on pyru-
vate carboxylase activity was investigated by using the perme-
abilized-cell activity assay. As shown in Fig. 4, aspartate
strongly inhibited pyruvate carboxylase. The pyruvate carbox-
ylase activity decreased sharply as the aspartate concentrations
increased up to 20 mM, with a slower effect observed beyond
this point.
Intracellular aspartate accumulation was next measured in
various C. glutamicum strains (Table 3). For this purpose,
strains 21253, 21799, 21253(pKD7), and 21799(pKD7) were
grown on lactate or glucose minimal medium and cells were
harvested at late exponential phase, just before threonine was
depleted from the medium. A higher intracellular aspartate
level was generally detected in strains 21253 and 21253(pKD7),
which express the wild-type-regulated aspartate kinase en-
zyme, than in strains 21799 and 21799(pKD7), which express a
deregulated aspartate kinase. Furthermore, the use of lactate
as a carbon source resulted in generally higher intracellular
aspartate levels than those obtained using glucose. These re-
sults support the hypothesis that aspartate kinase regulation by
threonine and lysine in strain 21253 leads to intracellular as-
partate accumulation that, in turn, inhibits the pyruvate car-
boxylation activity in host cells of strain 21253.
Effect of pyruvate carboxylase overexpression and carbon
source on cell physiology. The previously established effect of
carbon sources and the genetic background of the host cell on
the pyruvate carboxylase activity was expected to affect mac-
roscopic cell physiology as well. To investigate such effects,
strains 21253, 21253(pKD7), 21799, and 21799(pKD7) were
grown in either glucose or lactate minimal medium and their
growth and lysine production profiles were examined.
When the strains were grown on glucose, no significant dif-
ferences in cell growth rate or lysine production were observed
between the parental 21253 strain and its recombinant deriv-
atives, except for an extended lag phase in the profile of lysine
accumulation(Fig.5a). On
21799(pKD7) grew faster than parental strain 21799 on glu-
cose and the final cell concentration of the recombinant strain
was twice that of the parental strain (Fig. 5b). The specific
lysine production by the recombinant 21799(pKD7) strain was
lower than that by the parental strain, even though the final
lysine concentration in the fermentation broth was approxi-
mately the same (2 g/liter) for both strains (Fig. 5b).
When lactate was used as carbon source, both recombinant
strains 21253(pKD7) and 21799(pKD7) grew faster but pro-
duced less lysine than the parental strains (Fig. 6). In terms of
maximum growth rate and maximum specific lysine production
rate, the effect of pyruvate carboxylase overexpression ap-
peared to be similar for both strains when grown on lactate
(Table 4). This trend was the same when the cells were grown
on glucose, except for strain 21253, where pyc overexpression
did not seem to have any effect on either growth or lysine
productivity.
theother hand, strain
DISCUSSION
The role of anaplerosis on lysine production, as this is de-
fined by the formation of oxaloacetate, has attracted significant
attention in earlier research. Two critical milestones were the
development of an enzymatic assay for the C. glutamicum pyru-
vate carboxylase and the sequencing of the pyc gene (18, 32),
which confirmed the existence and operation of this pathway.
The availability of the pyc gene sequence also made it possible
to investigate the effect of its overexpression on cell physiology,
which is the subject of this paper.
The first interesting observation comes from the fact that the
carbon source appears to have a major effect on the transcrip-
tional and translational control of pyruvate carboxylase, with
lactate being the best carbon source for achieving maximum
expression levels. This was verified by Western blot and enzy-
matic activity data and appears to be in accordance with similar
literature reports (31, 32). Such an effect has been observed in
the past in other organisms such as Pseudomonas citronellolis
and Saccharomyces cerevisiae. In S. cerevisiae, the expression of
both isoenzymes (pyc1 and pyc2) of pyruvate carboxylase is
influenced by the growth phase and the type of carbon source.
FIG. 4. Effect of aspartate on pyruvate carboxylase activity in per-
meabilized C. glutamicum cells.
TABLE 3. Intracellular aspartate concentration in different
Corynebacterium species
Strain
Intracellular aspartate concn
(?mol/g DCW ? SD) ona
GlucoseLactate
21253
21253(pKD7)
21799
21799(pKD7)
30.7 ? 4.70
48.4 ? 1.91
16.6 ? 4.84
18.3 ? 4.36
65.8 ? 4.57
83.0 ? 4.56
22.8 ? 4.32
23.6 ? 2.65
aIntracellular aspartate concentrations were measured in cells harvested at
9 h.
VOL. 68, 2002EFFECT OF PYC OVEREXPRESSION IN C. GLUTAMICUM 5425

Page 5

On glucose minimal medium, pyc1 has a constant level of
expression throughout the growth phase, in contrast to a high
level of expression of pyc2 only during the early growth phase.
In ethanol minimal medium, the growth-related patterns of
pyc1 and pyc2 expression were similar and showed a decline
from early log phase to mid-log phase. The expression of pyc1
plays an important anaplerotic role in maintaining fermenta-
tive growth and more notably in establishing gluconeogenic
growth. On the other hand, pyc2 expression seems to support
growth on a glycolytic carbon source (3, 23).
For P. citronellolis, it has also been shown that the activity of
pyruvate carboxylase is controlled by the carbon source (40).
The activity of the enzyme is highest in cells grown in lactate or
glucose and virtually absent in cells grown in malate or aspar-
FIG. 5. Growth of and lysine production by parental and recombinant strain 21253 (a) and 21799 (b) cells grown on glucose. The upper and
lower panels show cell growth and lysine production profiles, respectively. Solid and open symbols represent the parental and recombinant strains,
respectively.
FIG. 6. Growth of and lysine production by parental and recombinant strain 21253 (a) and strain 21799 (b) cells grown on lactate. The upper
and lower panels show cell growth and lysine production profiles, respectively. Solid and open symbols represent the parental and recombinant
strains, respectively.
5426KOFFAS ET AL.APPL. ENVIRON. MICROBIOL.