The extent to which ATP demand controls the glycolytic flux depends strongly
on the organism and conditions for growth.
Brian J. Koebmann1, Hans V. Westerhoff2, Jacky L. Snoep2,3, Christian Solem1, Martin B. Pedersen4
Dan Nilsson5, Ole Michelsen1, and Peter R. Jensen1*
1 Section of Molecular Microbiology, BioCentrum-DTU, Technical University of Denmark,
Building 301, DK-2800 Kgs. Lyngby, Denmark.
2Department of Molecular Cell Physiology, Vrije Universiteit, De Boelelaan 1087, NL-1081 HV
Amsterdam, The Netherlands
3Department of Biochemistry, University of Stellenbosch, Private Bag X1, Matieland 7602, South
Africa
4 Department of Genomics and Strain Development, Chr. Hansen A/S, Bøge Allé 10-12, DK-2970
Hørsholm, Denmark.
5 CH Bio Ingredients, Chr. Hansen A/S, Bøge Allé 10-12, DK-2970 Hørsholm, Denmark
Running title: Control of glycolysis in Escherichia coli and Lactococcus lactis
Key words: ATPase, coli, glycolysis, lactococcus, MCA, metabolic control analysis
*Correspondent footnote. Mailing address: Section of Molecular Microbiology, BioCentrumDTU, Technical University of Denmark, Building 301, DK-2800 Kgs. Lyngby. Tel +45 45252510.
Fax. +45 45932809. E-mail: [email protected]
Abstract
Using molecular genetics we have introduced uncoupled ATPase activity in two different
bacterial species, Escherichia coli and Lactococcus lactis, and determined the elasticities of the
growth rate and glycolytic flux towards the intracellular [ATP]/[ADP] ratio. During balanced growth
in batch cultures of E. coli the ATP demand was found to have almost full control on the glycolytic
flux (FCC=0.96) and the flux could be stimulated by 70%. In contrast to this, in L. lactis the control
by ATP demand on the glycolytic flux was close to zero. However, when we used non-growing cells
of L. lactis (which have a low glycolytic flux) the ATP demand had a high flux control and the flux
could be stimulated more than two fold. We suggest that the extent to which ATP demand controls
the glycolytic flux depends on how much excess capacity of glycolysis is present in the cells.
Introduction
The control of the glycolytic flux in living cells has been investigated for several decades.
Most of the glycolytic enzymes have been overexpressed individually or in combinations of several
enzymes together with virtually no effect on the glycolytic flux (Schaaff et al., 1989; Snoep et al.,
1996; Müller et al., 1997; Hauf et al., 2000). This could be due to distribution of the flux control
over many enzymes in the pathway, which would then require the simultaneous overexpression of
many enzymes in order to achieve a higher glycolytic flux. Alternatively, the flux control could also
reside outside the glycolytic reactions, for instance in the transport of sugar into the cell or in the
reactions that consume the products of glycolysis. Indeed, early computer simulations indicated that
the ATP consuming reactions could play an important role in controlling the glycolytic flux in
erythrocytes (Rapoport et al., 1976; Heinrich and Schuster, 1996). Such a distribution where most
flux control lies outside the pathway has also been favoured from a functional point of view
(Hofmeyr and Cornish-Bowden, 2000).
In this paper we discuss experimental data on the importance of the ATP consuming
reactions for control of glycolysis in two microbial systems. We show that the ATP consumption
does indeed have almost full control over the glycolytic flux in aerobic E. coli while in anaerobic L.
lactis it has virtually no control at all.
Materials and Methods
The materials and methods used throughout the current paper are described in appendix I
Results
Introduction of uncoupled ATPase activity in bacteria.
The cytoplasmic F1-domain of the F1F0-H+ -ATPase contains the catalytic site for
synthesis/hydrolysis of ATP. We previously described the cloning of atpAGD genes encoding the
a-, g-, and b-subunits of the F1-part, respectively downstream synthetic constitutive promoters,
which expressed the genes to different extents (Fig. I; Koebmann et al., 2002a, 2002b). The E. coli
and L. lactis atp genes were cloned in transcriptional fusions with lacLM coding for b-galactosidase,
which allowed for an indirect measurement of transcription of the ATPase genes in both organisms
E. coli BOE270
L. lactis MG1363
1
0,8
0,6
0,4
0,2
0
-0,2
-0,4
-0,6
-0,8
-1
Elasticity of glyc. flux towards [ATP]/[ADP]
1
Elasticity of glyc. flux towards [ATP]/[ADP]
Elasticity of growth towards [ATP]/[ADP]
Elasticity of growth towards [ATP]/[ADP]
0,8
0,6
Elasticities
Elasticities
(A)
0,4
0,2
0
-0,2
-0,4
-0,6
-0,8
-1
0
2
4
6
8
10
5
6
7
8
9
[ATP]/[ADP] ratio
[ATP]/[ADP] ratio
(B)
1
Control by ATP demand on
glycolytic flux
Control by ATP demand on
glycolytic flux
1
0,9
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
0
2
4
6
8
[ATP]/[ADP] ratio
10
0
-1
5
6
7
8
9
[ATP]/[ADP] ratio
Figure 1: Elasticities and flux control by ATP demand on the glycolytic fluxes in E. coli and L.
lactis. (A) Elasticities of the glycolytic flux and the growth rate towards the intracellular
[ATP]/[ADP] ratio. The elasticities are the slopes of the scaled fluxes towards the [ATP]/[ADP]
ratio in (Fig. IIIB (Appendix III)), calculated from the fitted equations. (B) Flux control by the
demand for ATP on the glycolytic flux as a function of the [ATP]/[ADP] ratio, calculated from
eq. 1. (From Koebmann et al., 2002a, 2002b)
(Koebmann et al., 2002a). The gradual increase in ATPase activity resulted in a concomitant gradual
decrease in biomass yield. In combination with direct in vitro measurements of the ATPase activity
in E. coli (Koebmann et al., 2002a) this showed that ATP was indeed being hydrolyzed in the
bacterial cells.
Impact of increased ATP demand on the energy state.
We then studied how the extra ATP consumption affected the energy state of the cells. In E.
coli the concentration of ATP decreased slightly, with increased ATPase activity to 25% lower
concentration at the highest ATPase activity, while the ADP concentration increased by more than
65% (Fig. II (Appendix III); Koebmann et al., 2002a). This was associated with an approximately
18% decrease in total ATP+ADP concentration and a drop in [ATP]/[ADP] ratio from 11 to 5 (Fig.
II). The intracellular energy level was also affected in L. lactis, where the [ATP]/[ADP] ratio
dropped from 9 to 5 (Fig. II (Appendix III); Koebmann et al., 2002b). But in contrast to E. coli,
there was no significant change in the total ATP+ADP concentration in L. lactis.
Impact of uncoupled ATPase activity on anabolic and catabolic fluxes.
The increased ATP demand had different impacts on the glycolytic fluxes in E. coli and L.
lactis. In both organisms we observed a decrease in growth rate to 76% (E. coli) and 69% (L. lactis)
of wild-type level (Table I (Appendix II)). The uncoupled ATPase activity resulted in a significant
decrease in biomass yield in E. coli to 45% of the wild-type yield (Koebmann et al., 2002a) and in L.
lactis the yield decreased to 69% of the wild-type yield (Koebmann et al., 2002b). The glycolytic
fluxes were measured as the steady state consumption rate of glucose during exponential growth.
Interestingly, the glycolytic flux in E. coli increased gradually to 170% of wild-type flux (Koebmann
et al., 2002a), whereas in L. lactis no change in the glycolytic flux was observed (Koebmann et al.,
2002b).
A possible explanation for the lack of stimulation of the glycolytic flux in L. lactis could be
that glycolysis in growing L. lactis is already running close to its maximal capacity. We therefore
also measured the effect of ATPase activity in non-growing cells of L. lactis. L. lactis cells
containing different F1-ATPase activities were resuspended in SA medium without vitamins and
amino acids, which lead to a decrease in the ATP demanding anabolic reactions as a result of the lack
of essential nutrients. Thus, in non-growing wild-type cells the glycolytic flux is reduced to 37% of
steadily growing cells (Koebmann et al., 2002b). The introduction of F1-ATPase activity stimulated
the glycolytic flux until the flux in growing cells was approached, but not above this flux (data not
shown).
Flux control of the ATP demanding processes.
In order to quantity the extent of which ATP demand controls glycolysis we simplified the
cellular free-energy metabolism into a supply module and a demand module for ATP, by assuming
that the fluxes are only linked via ATP:
e1
Substrate
e2
DGp
Growth
When Metabolic Control Analysis is applied the flux control coefficients can be expressed in
terms of the block elasticities as:
CeJ2 =
- e ep1
e ep2 - e ep1
(eq. 1)
The elasticities of the blocks can be obtained from the slopes of the fluxes as functions of the
energy level (Fig. IIIA (Appendix III)). After the data points have been fitted to algebraic equations
in a log-log plot (Fig. IIIB (Appendix III)), the elasticities can be calculated (Fig. 1A). For E. coli the
elasticity of the catabolic block was quite high, -0.89, in the absence of ATPase and dropped to
–0.42 at the highest ATPase activity. In contrast, the elasticity of the anabolic block was rather low,
0.04, in the absence of ATPase and increased gradually to 0.65 at the highest ATPase activity
(Koebmann et a., 2002a). For L. lactis different curves (logarithmic, linear, exponential, power) were
fitted to the data points with small changes in the [ATP]/[ADP] ratio as compared to the wild type.
The elasticity of the catabolic block was estimated to be close to zero, -0.02, which indicates that
glycolysis is very insensitive towards changes in the energy level. The elasticity of the anabolic
block (growth rate) ranged from 0.22 to 0.26 in the absence of ATPase, depending on the applied
curve fit, and increased slightly with higher ATPase expression.
The control by the ATP demanding processes on the glycolytic flux was then calculated
from eq. 1 (Fig. 1B). For E. coli essentially all control resided in the ATP demanding processes
( CeJ21 =0.96). Even if calculation of the anabolic flux is based on a linear fit (which leads to
overestimation of the anabolic elasticity in the wild-type cell) flux control by ATP demand is still
CeJ21 =0.75. In L. lactis the ATP consuming reactions were found to have less than 10% of flux
control on glycolysis ( CeJ21 =0.1).
Discussion
We have measured to what extent ATP demand contributes to the control of glycolytic flux in
bacteria. By expressing three subunits, a, b and g, of the catalytic part (F1) of the H+ -ATPase from
a series of promoters with increasing strength we varied the ATP hydrolysis reaction. The
introduction of uncoupled ATPase activity in the cytoplasm resulted in a gradual decrease in the
intracellular [ATP]/[ADP] ratio and in a decrease of the biomass yield.
In aerobic E. coli cells growing in minimal medium with glucose as the sole carbon- energy
source, the ATP demand turned out to have almost full control over the glycolytic flux with a flux
control coefficient of 0.96. The high flux control was the result of a high (negative) elasticity of
glycolysis and a low elasticity of the anabolic reactions towards the phosphorylation potential. The
results also demonstrate that the growth rate of E. coli is controlled mainly by anabolic reactions and
not by ATP production. Experiments have been performed earlier with E. coli cells in which the
activity of the coupled H+ -ATPase (ATP synthase) was being modulated (genetically), which
amounts to modulating the ATP supply (Jensen et al., 1993b). In those experiments it was found
that the ATP synthase had zero control on the growth rate, which is therefore in good agreement
with the results reported in the current paper.
An interesting observation was that the flux control by ATP demand remained relatively
high as the activity of F1-ATPase increased and an overall increase in the glycolytic flux by 70%
could be achieved. This result shows that there is a relatively large excess of glycolytic capacity,
which can be mobilized upon demand.
In anaerobic Lactococcus lactis cells growing in defined medium supplemented with glucose,
the control by ATP demand was found to be very close to zero. The anabolic reactions in L. lactis
had a low elasticity towards the phosphorylation potential similarly to what was observed for E.
coli but in the case of L. lactis the elasticity of glycolysis was virtually zero. These data suggest that
the rate of glycolysis in growing L. lactis is already close to its maximal capacity. Indeed, when the
experiment was repeated with non-growing L. lactis cells (cells resuspended in buffer which have 3
fold lower glycolytic flux compared to growing cells) the uncoupled ATPase resulted in more than
two fold increase in the glycolytic flux. Thus, in non-growing cells the ATP demand does appear to
have a high control on the glycolytic flux, also in L. lactis. Apparently, the extent to which the ATP
demand controls glycolysis depends on how much excess capacity of glycolysis is present in the
cells.
Acknowledgement: This work was supported by The Danish Academy of Technical Sciences
(ATV), The Danish Research Agency, and Chr. Hansen A/S. We thank Regina Schürmann for expert
technical assistance, Jannie Hofmeyr, David Fell, and Reinhardt Heinrich for discussions.
References
Andersen, H.W., Solem, C., Hammer, K. and Jensen, P.R., J. Bacteriol. 183 (2001) 3458.
Boogerd, F.C., Boe, L., Michelsen, O. and Jensen, P.R., J. Bacteriol. 180 (1998) 5855.
Gasson, M. J., J. Bacteriol. 154 (1983) 1.
Hauf, J., Zimmermann, F.K. and Muller, S., Enzyme Microb. Technol.26 (2000) 688.
Heinrich, R. and Schuster, S., In The regulation of cellular systems, p.177-188. Chapman & Hall, New
York, 1996, pp 177-188.
Hofmeyr J.-H.S and Cornish-Bowden, A. FEBS Letters 467 (2000), 47.
Jensen, P.R. and Hammer, K. Appl. Environ. Microbiol. 59 (1993a), 4363.
Jensen, P.R., Michelsen, O. and Westerhoff, H.V. Proc. Natl. Acad. Sci. USA 90 (1993b), 8068.
Jensen, P. R. and Hammer, K. Appl. Environ. Microbiol. 64 (1998), 82.
Koebmann, B.J., Westerhoff, H.V., Snoep, J.L., Nilsson, D. and Jensen, P.R.. J. Bacteriol. 184 (2002a),
3909.
Koebmann B.J., Solem, C., Pedersen, M.B., Nilsson, D. and Jensen, P.R. Appl. Environ. Microbiol.
(2002b), Submitted.
Müller, S., Zimmermann, F. K., Boles, E. Microbiology 143 (1997), 3055.
Neidhardt, F.C., Bloch, P.L. and Smith, D.L. J. Bacteriol. 119 (1974), 736.
Rapoport,T.A., Heinrich, R. and Rapoport, S.M. Biochem. J. 154 (1976), 449.
nd
Sambrook, J., Fritsch, E.F. and Maniatis, T., Molecular cloning: a laboratory manual, 2
ed. Cold
Spring Harbor Laboratory Press, Cold Spring Harbor, N. Y., 1989.
Schaaff, I., Heinisch, J. and Zimmermann, F.K. Yeast 5 (1989), 285.
Snoep, J. L., Arfman, N., Yomano, L.P., Westerhoff, H.V., Conway, T. and Ingram, L.O. Biotechnol.
Bioeng. 51 (1996), 190.
APPENDICES
Appendix I
Materials and Methods
Bacterial strains: The effect of uncoupled ATPase activity was studied in E. coli K-12 strain
BOE270 (Boogerd et al., 1998) and in the plasmid-free Lactococcus lactis subsp. cremoris MG1363
(Gasson, 1983).
Enzymes : Restriction enzymes, Calf Intestine Phosphatase, and T4 DNA ligase from Pharmacia
and New England Biolabs were used as recommended by the manufacturers.
DNA techniques: DNA isolation, were carried out as described by Sambrook et al., (1989).
Plasmids with atpAGD under the control of synthetic constitutive promoters: Details of the
construction work is given elsewhere (Koebmann et al., 2002a, 2002b). Briefly, the F1 genes
atpAGD coding for the a-, g-, and b-subunits, respectively, were cloned downstream a range of
synthetic promoters with different strengths in a plasmid library in which synthetic constitutive
promoters have been placed upstream to the lacLM genes (Figure 1) (Jensen and Hammer, 1998).
Growth experiments: E. coli and L. lactis strains were inoculated from overnight cultures at 30°C
into 100 ml of agitated MOPS (morpholine-propanesulfonic acid) (Neidhardt et al., 1974)
supplemented with 5 mg of thiamin per ml, 50 mg each of leucine, isoleucine, and valine per ml, 200
mg/ml of erythromycin and 0.4 g/l of glucose (E. coli) or into 100 ml of SA medium (Jensen and
Hammer, 1993a) supplemented with 1.0 g/L of glucose and 5 mg/ml of erythromycin (L. lactis).
Regular measurements of OD450 were made and samples were withdrawn for determination of
[ATP] and [ADP] concentrations, and for measurement of extracelular metabolite concentrations on
HPLC.
Growth of resuspended L. lactis cells: The growth experiment was carried out as described in
Koebmann et al. (2002b). Briefly, L. lactis was grown exponentially in SA medium supplemented
with glucose to an OD450 of 0.9. The cultures were then cooled, washed and resuspended in SA
medium supplemented with 2 g/L of glucose but without amino acids or vitamins. Samples were
taken for HPLC determination of glucose and product concentrations at 10, 30, 70, 150, and 310
minutes after resuspension.
Measurement of [ATP]/[ADP ]: Samples for ATP and ADP measurements were taken and
measured using a luciferin-luciferase ATP monitoring kit (LKB) as described in Koebmann et al.
(2002a).
Measurement of glucose consumption and product formation by HPLC: Samples of 2 ml were
withdrawn from the cultures at different time intervals and quickly filtered through a 0.22 mm filter
and stored at –20°C. The separation was performed as described in (Andersen et al., 2001).
Appendix II
TABLE I. Physiology of derivatives of strain E. coli BOE270 and L. lactis MG1363 with uncoupled
F1-ATPase activity (a-, g-, b-subunits). (From Koebmann et al., 2002a, 2002b)
a
Strain
BOE270
PJ4454
BK1032
BK1017
BK1036
MG1363
Plasmid
pCP44
pCP41::atpAGD
pCP34::atpAGD
pCP44::atpAGD
Specific b- Biomass
yield
gal
Yg
activity
Growth rate
µ
Glucose
flux
J gluc
Biomass
Yield
%
Growth
rate
%
Glucose
flux
%
Miller units
(MU)
gdw/mmol
glucose
h- 1
mmol glucose/h/gdw
--7.5
22
44
0.075
0.050
0.041
0.033
0.48
0.45
0.40
0.37
6.5
8.9
9.9
11.0
100
67
55
45
100
93
83
76
100
137
153
170
BK1010
pAK80
0
0.0305
0.74
24.2
100
100
100
BK1546
pCPC69::atpAGD
0.2
0.0302
0.73
24.1
99
99
100
BK1540
pCPC63::atpAGD
5.9
0.0301
0.73
24.4
99
100
101
BK1094
pCP34::atpAGD
95
0.0280
0.71
25.3
92
96
105
BK1506
PCPC7::atpAGD
137
0.0278
0.67
24.2
91
92
100
BK1542
pCPC65::atpAGD
173
0.0266
0.66
24.9
87
90
103
BK1552
PCPC75::atpAGD
177
0.0264
0.65
24.6
87
88
102
BK1503
pCPC4::atpAGD
307
0.0234
0.58
25.0
77
79
103
BK1557
PCPC80::atpAGD
405
0.0239
0.56
23.5
78
76
97
BK1502
PCPC3::atpAGD
714
0.0210
0.51
24.2
69
69
100
The expression of the lacLM genes in operon with the atpAGD are given in the table as specific b-gal activities (Miller
units) (Miller, 1972), and can be related to the expression of atpAGD.
Appendix III
a
ori
g
atp
A
erm
pCPCx::atpAG
D
b
atpG
atp
D
Subunit
lacLM
Gene
CP
X
Figure I. Linear representation of the plasmids constructed for modulating ATPase activity in E. coli
and L. lactis (not drawn to scale). CPX:= different constitutive promoters (From Koebmann et al.,
2002a, 2002b).
L. lactis
E. coli
8
14
10
10
8
4
6
4
2
Concentration
(m mol/gdw)
Concentration
(m mol/gdw)
6
[ATP]/[ADP] ratio
12
8
6
6
4
4
9
2
2
[ATP]/[ADP] ratio
8
2
0
0
0
2
4
6
8
10
Specific ATPase activity (min-1)
[ATP]
[ATP]+[ADP]
Poly. ([ATP]/[ADP])
[ADP]
[ATP]/[ADP]
0
0
200
400
600
0
800
Specific b -galactosidase activity
[ATP]
ATP+ADP
Lineær (ATP+ADP)
Lineær ([ADP])
[ADP]
ATP/ADP
Lineær ([ATP])
Figure II. Effect of increased ATPase activity on the [ATP]/[ADP] ratios.
Correlation between specific ATPase activity or b-galactosidase activity with ATP, ADP, ATP+ADP
pools and [ATP]/[ADP] ratios. The experimental data for ATP, ADP, and ATP+ADP pools are fitted
to linear curves shown by dotted lines, and the [ATP]/[ADP] ratio is fitted by a curve shown by a full
line. Error bars indicate the standard deviations of the measurements. Data represent the average of
four to six measurements. (From Koebmann et al., 2002a, 2002b)
(A)
E. coli BOE270
L. lactis MG1363
120
Relative flux (%)
Relative flux (%)
200
100
Glyc. flux
Growth rate
100
80
60
40
Glycolytic flux
Growth rate
20
0
0
0
2
4
6
8
10
12
4
5
6
[ATP]/[ADP] ratio
2,2
2,1
2
1,9
1,8
log(Jg)
log (growth)
1,6
2
y = -0,8979x + 1,9111x + 0,9836
2
R =1
log (Jg) or log (m)
log (Jg) or log (m)
2,05
2
R = 0,9938
2,3
1,7
y = -0,0226x + 2,0262
2
1,95
y = 0,2175Ln(x) + 2,0087
1,9
1,85
log(Jg)
1,8
log (growth)
1,75
Series3
1,5
0
0,2
9
2,1
2
y = -0,6917x + 0,5506x + 2,1781
2,4
8
[ATP]/[ADP] ratio
(B)
2,5
7
0,4
0,6
0,8
log ([ATP]/[ADP])
1
1,2
1,7
0,6
0,7
0,8
Series4
0,9
Log. (Series3)
1
log ([ATP]/[ADP])
Linear
(Series4)
Figure III. Dependence of glycolytic flux and growth rate on the [ATP]/[ADP] ratio.
(A) The relative glycolytic fluxes (squares) and growth rates (triangles) are plotted as a function of
the [ATP]/[ADP] ratios for strains with modulated F1-ATPase activities. (B) Logarithmic (scaled)
relative glycolytic fluxes (open squares) and growth rates (open triangles) as functions of logarithmic
[ATP]/[ADP] ratios. The experimental data points indicated with full squares or triangles are fitted
to curves shown by penetrating lines and the equations indicated above or below the plots.