Control analysis as a tool to understand the formation
of the las operon in Lactococcus lactis
Brian Koebmann, Christian Solem and Peter Ruhdal Jensen
Microbial Physiology and Genetics, BioCentrum-DTU, Technical University of Denmark, Kgs Lyngby, Denmark
Over the last three decades increasing attention has
been paid to how metabolic pathways are controlled.
Metabolic control analysis [1,2] has been applied suc-
cessfully to determine the flux control of many single
enzymes [3–7], but much less attention has been paid
to determine flux control by individual enzymes
cotranscribed in prokaryotic operons.
In Lactococcus lactis, an industrially important
organism used extensively in the fermentation of dairy
products, the three glycolytic enzymes phosphofructo-
kinase (PFK), pyruvate kinase (PK) and lactate dehy-
drogenase (LDH) are clustered in the so-called las
operon [8]. This organization of glycolytic genes is
unique and has given rise to speculation that the three
enzymes might play an important role in the control
and regulation of lactic acid production by this organ-
ism. We have previously shown that small changes
in the activity of PFK result in pronounced changes in
metabolite pools, glycolytic flux and growth rate in
L. lactis, but control by PFK has not been quantified
[9]. LDH was shown to have no control over either
growth or glycolytic flux at wild-type levels, but a
strong negative control over the minor flux to mixed
acids via pyruvate formate lyase (PFL) [10].
In this study, the activities of PFK and PK were
modulated individually by changing expression of the
Keywords

and acetate production (C
J
acetate
PK
¼ 0:8 À 1:0), whereas PFK exerts no control
over these fluxes at increased expression. Decreased expression of the entire
las operon resulted in a strong decrease in the growth rate and glycolytic
flux; at 53% expression of the las operon glycolytic flux was reduced to
44% and the flux control coefficient increased towards 3. Increased las
expression resulted in a slight decrease in the glycolytic flux. At wild-type
levels, control was close to zero on both glycolysis and the pyruvate bran-
ches. The sum of control coefficients for the three enzymes individually
was comparable with the control coefficient found for the entire operon;
the strong positive control exerted by PK almost cancels out the negative
control exerted by LDH on formate production. Our analysis suggests that
coregulation of PFK and PK provides a very efficient way to regulate gly-
colysis, and coregulating PK and LDH allows cells to maintain homolactic
fermentation during glycolysis regulation.
Abbreviations
LDH, lactate dehydrogenase; PFK, phosphofructokinase; PFL, pyruvate formate lyase; PK, pyruvate kinase.
2292 FEBS Journal 272 (2005) 2292–2303 ª 2005 FEBS
corresponding genes. We measured the control exerted
by each of the las enzymes on the glycolytic flux,
growth rate and product formation. We also studied
strains with modulated expression of the entire las
operon, and show that the data fit well with the indi-
vidual determination of flux control coefficients by
PFK, PK and LDH. The role of the las operon is dis-
cussed on the basis of the distribution of flux control
for PFK, PK and LDH.

cribed in Experimental procedures (Fig. 2D). From
these data it is clear that at the wild-type level PFK
has no control over the glycolytic flux (C
J
glucose
PFK
% 0)or
growth rate (C
J
l
PFK
% 0), and no control over the fluxes
to lactate (C
J
lactate
PFK
% 0), formate (C
J
formate
PFK
% 0) or acetate
(C
J
acetate
PFK
% 0) at the wild-type level and above.
Fig. 1. Glycolysis and the las operon in
Lactococcus lactis.The las operon in L. lactis
consists of the three genes pfk, pyk and
ldh, coding for phosphofructokinase (PFK),

whereas the relative LDH activity was reduced to 80%
of the wild-type level (Fig. 3A).
In order to study the control exerted over the meta-
bolic fluxes by PK, strains with PK activities altered
around the wild-type level were grown in defined SAL
medium supplemented with glucose. A slight decrease
in growth rate and glycolytic flux was observed at
increased PK activities (Figs 3B and 5). For strain
CS1929 we found a strong decrease in growth rate and
glycolytic flux, almost proportional to the change in
PK activity. The data points for growth and glucose
flux were then fitted against the PK activities in order
to determine the control exerted by PK over the
growth rate (Fig. 3B) and glycolytic flux (Fig. 5) from
which we conclude that PK exerted no significant con-
trol over either growth rate or glycolytic flux at the
wild-type level. However, reducing the PK activity to
37% enhances the control exerted by PK over growth
rate to C
J
l
PK
% 1.
Product formation changed significantly as the PK
activity was modulated. At increased PK activity we
found an almost proportional increase in formate and
acetate production and a decrease in lactate produc-
Fig. 2. Modulation of PFK activity and the effects on growth and fluxes. (A) Library of strains with modulated PFK activities. The PFK activit-
ies were measured in extracts from strains in which an additional pfk gene transcribed from synthetic promoters was integrated on the chro-
mosome by site-specific recombination in a phage attachment site. The specific PFK activity in MG1363 was determined to 0.55 UÆmg

PK
¼ 0:8 À 1:0 (Fig. 5).
Modulation of the entire las operon
Strains with altered expression of the entire las operon
were previously obtained by replacing the native las
promoter with synthetic promoters in a single cross-
over event [11]. From this library consisting of 50
strains with altered expression of the las operon, the
enzyme activities of PFK, PK and LDH were deter-
mined and eight strains with enzyme activities 0.5–3.5
times the wild-type level were selected for further
analysis (Fig. 6A). Good correlation among relative
enzyme activities of the three enzymes was found.
These strains then allowed us to study the control
exerted by all three las enzymes simultaneously. The
growth rate and metabolic fluxes for the strains were
determined and we also found that growth rate and
glycolytic flux were highest when the activities of the
las enzymes were at wild-type levels (Figs 6B and 7).
The data points were fitted to the equations described
in Experimental procedures and are presented in
Figs 6B and 7 for calculations of flux control coeffi-
cients. The sum of flux control on glycolysis and
growth rate by the las enzymes at wild-type levels is
close to 0 (C
J
glucose
las
% 0 and C
J

modulation of PK and LDH (Fig. 7). Strong negative
flux controls on formate production (C
J
formate
las
%
ðÀ1:4ÞÀðÀ1:7Þ) and acetate production (C
J
acetate
las
%
ðÀ1:7ÞÀðÀ2:0Þ) were observed at reduced activities of
the las enzymes to 50–60% of wild-type level (Fig. 7).
When the activities of the las enzymes were increased
three times we find a flux control coefficient at
C
J
formate
las
%ðÀ0:4Þ for the formate flux and
C
J
acetate
las
%ðÀ0:4Þ for the acetate flux.
A
B
Fig. 3. Modulation of PK activity and the effect on growth rate. (A)
Enzyme activities of PFK, PK and LDH relative to the wild-type level
in strains with modulated PK activities. The enzyme activities were

error bars, are based on measurement of three individual cultures.
Control analysis of the las operon B. Koebmann et al.
2296 FEBS Journal 272 (2005) 2292–2303 ª 2005 FEBS
AB
Fig. 6. Modulation of the las operon. (A) Enzyme activities of PFK, PK and LDH relative to the wild-type level. The enzyme activities were
measured in extracts from strains in which the native las promoter was replaced by a library of synthetic promoters with different strengths
[11]. (B) Growth rates of selected strains (including flux control coefficients). Standard deviations, indicated by error bars, are based on meas-
urement of three individual cultures.
Fig. 7. Flux control coefficients for the las enzymes on metabolic fluxes. A selection of strains were analysed with respect to glycolytic flux
and metabolic fluxes. Flux control coefficients with respect to glycolysis, lactate, acetate and formate production were determined from the
fitted equations as described in Experimental procedures. Standard deviations, indicated by error bars, are based on measurement of three
individual cultures.
B. Koebmann et al. Control analysis of the las operon
FEBS Journal 272 (2005) 2292–2303 ª 2005 FEBS 2297
Comparison of control by the las enzymes
Based on the data presented here and on earlier
data for LDH [10], it is possible to compare the flux
controls of the individual enzymes with that of a
simultaneous modulation of all the las enzymes, i.e. to
test whether: C
J
las
¼ C
J
PFK
þ C
J
PK
þ C
J

þ C
J
formate
PK
þ C
J
formate
LDH
¼À0:3. Interestingly, when
all enzymes from the las operon were modulated simul-
taneously we found a control of C
J
formate
las
¼À0:26 on the
formate flux, which again fits very well with the sum
of the control by the individual enzymes.
A similar comparison of flux control was not poss-
ible for the acetate flux because this was not measured
in the earlier study on LDH [10]. However, we expect
the sum of the individual flux control coefficients to
add up to that found for the combined change of the
las enzymes, because mixed acid metabolism under
anaerobic conditions is expected to result in equal
amounts of formate and acetyl-CoA and the resulting
acetyl-CoA is then metabolized into equal amount of
ethanol and acetate to maintain the redox balance.
Discussion
In this study we quantified the control exerted by the
las enzymes on the metabolic fluxes under conditions

the fitted equations, this amounts to a flux control
coefficient approaching 3! This is significantly higher
than the flux control coefficients for the individual las
enzymes at comparable levels. From the data for PFK
given in Andersen et al. [9], the flux control coefficient
on the glycolytic flux of PFK activity at 50% of wild-
type level can be estimated to 0.45 by fitting the data
to a linear curve. According to Andersen et al. [10],
the flux control coefficient on the glycolytic flux for
LDH at 50% of wild-type activity was found to be
around 0.1–0.2. In this study we found the flux control
coefficient on the glycolytic flux by PK at 50% of
wild-type activity to be around 1.0. Thus, the sum of
the individual enzymes amounts to only 1.6–1.7.
The dramatic reduction in growth rate and glycolytic
flux at reduced las enzyme activity may be explained
by perturbations in metabolite pools. In the previous
study by Andersen et al. it was suggested that the
strong effect on growth and glycolytic flux observed
when reducing PFK activity could be due to an accu-
mulation of hexose phosphates [9]. The stronger effect
on the growth rate and glycolytic flux observed in this
study when all the las enzymes were reduced to 50%
of wild-type levels may then be the result of decreased
PK activity which would result in an increased
phosphoenolpyruvate pool, which in turn would
enhance the activity of the PTS system and thereby
result in further increases in hexose phosphate pools.
The decreased LDH activity may contribute further to
this effect by causing an accumulation of pyruvate and

increase in the flux to mixed acid products. In a recent
study by Ramos et al. [17] it was found that the
fermentation pattern in a PK-overproducing strain
showed a typical homolactic metabolism under anaer-
obic conditions. At first, this seems to contradict our
results. However, in practice, the flux to formate at the
wild-type level amounts to only 3.5% of the pyruvate
metabolism, and a doubling in formate flux would
amount to only 7%, which would still be considered to
be homolactic fermentation.
The magnitude of the control exerted by PK
(C
J
formate
PK
¼ 0:9 À 1:1) over formate production was
almost comparable but of the opposite sign compared
with the negative control found previously for LDH
(C
J
formate
LDH
%À1:3) [10]. Because the control by PFK on
the flux to formate was found to be 0, the sum of
control on the formate flux was only slightly negative
(C
J
formate
las
%À0:3), which explains why changing expres-

ment by changing the expression of only a few genes
and using the protein-synthesizing capacity to express
these genes when needed. Here we have studied a set
of enzymes that are needed by these cells under all
growth conditions, because glycolysis is the energy-
producing pathway. Indeed, in contrast to many other
systems, only a few fold regulations of the genes have
been shown to take place [15].
Metabolic control analysis has helped us to charac-
terize the role of the individual genes in an operon
and, to some extent, explain why L. lactis may benefit
from the way in which the las operon is organized. We
believe that such analysis would not have been possible
using traditional functional analysis with gene knock-
outs and overexpression of enzymes from a plasmid.
Experimental procedures
Bacterial strains and plasmids
For cloning purposes was used Escherichia coli strain
ABLE-C {E. coli C lac(LacZ
–
)[Kan
r
McrA
–
McrCB
–
McrF
–
Mrr
–

FEBS Journal 272 (2005) 2292–2303 ª 2005 FEBS 2299
on the TP901-1 attB locus on the chromosome of MG1363.
The replication-thermosensitive plasmid pG
+
host8, which
contains a gene for tetracycline resistance [23], was used to
delete the pyk gene from the las operon on the chromo-
some.
Growth media and growth conditions
E. coli strains were grown aerobically at 37 °C in Luria–
Bertani broth [24]. L. lactis strains were routinely cultivated
at 30 °C without aeration in M17 broth [25] or in chemic-
ally defined SA medium [26] modified by exclusion of ace-
tate and inclusion of 2 lgÆ mL
)1
lipoic acid (SAL medium).
The media were supplemented with 1 or 10 gÆL
)1
glucose
and appropriate selective antibiotics.
L. lactis growth experiments were performed as batch
cultures (flasks) at 30 °C in 100 mL of SAL medium [26]
supplemented with 0.12% (w ⁄ v) of glucose when determin-
ing biomass yield on glucose, Y
g
, or else 1% (w ⁄ v) of glu-
cose. Antibiotics were only used in precultures and not in
the growth experiments. Enzyme activities and product for-
mation were determined by using the same cultures thereby
assuring that genetic constructions were intact. A slow stir

All manipulations were performed as described by
Sambrook et al. [24]. Taq DNA polymerase (New England
Biolabs, Frankfurt am Main, Germany) was applied
for analytical purposes and PCR products intended for
cloning were generated using Elongase
R
enzyme mix (Invi-
trogen, Ta
˚
strup, Denmark). Chromosomal DNA from
L. lactis was isolated using a method described previously
[27] with the modification that cells were treated with
20 lg lysozyme per mL for 2 h before lysis. Digestion with
restriction enzymes (Fermentas, St Leon, Germany; Amer-
sham, Hillerød, Denmark), treatment with T4 DNA ligase
(Fermentas) and shrimp alkaline phosphatase (Fermentas)
were carried out as prescribed by the manufacturers. DNA
fragments were purified from agarose gels using GFX
PCR DNA and Gel Band Purification Kit (Amersham).
E. coli was transformed by electroporation. Cells were
plated on Luria–Bertani plates supplemented with appro-
priate antibiotics. Plasmid DNA was isolated from E. coli
by using Qiaprep Spin Miniprep Kit (Qiagen, Hilden, Ger-
many). Cells of L. lactis were made electrocompetent by
growth in GM17 medium containing 1% glycine, and
DNA was introduced by electroporation as previously des-
cribed by Holo and Nes [28]. After electroporation cells
were plated on GM17 supplemented with appropriate anti-
biotics.
Enzyme measurements

10 mm NH
4
Cl, 0.3 UÆmL
)1
triose phosphate isomerase,
1UÆmL
)1
glycerol 3-phosphate dehydrogenase and 0.3 U
aldolase. PK was assayed as described by Crow and
Pritchard [30]. Final concentrations in assay was: 1 mm
GDP, 1 mm PEP, 1 mm fructose 1,6-bisphosphate, 10 mm
MgCl
2,
0.2 mm NADH and 6.3 UÆmL
)1
LDH. LDH was
measured according to Crow and Pritchard [31]. Final con-
centrations in assay was: 10 mm pyruvate, 0.2 mm NADH,
1mm fructose 1,6-bisphosphate. All measured enzyme
activities were related to the A
280
of the extract, for the
purpose of determining relative activities. The specific activ-
ities of PFK and PK and LDH in MG1363 were deter-
mined as UÆmg
)1
of protein, where a unit (U) is defined as
the amount of enzyme producing 1 lmol of NADH per
Control analysis of the las operon B. Koebmann et al.
2300 FEBS Journal 272 (2005) 2292–2303 ª 2005 FEBS

200 lgÆmL
)1
5-bromo-4-chloro-3-indolyl-beta-d-glucuronide
(X-gluc) (Biosynth AG, Switzerland).
Construction of a strain with reduced PK activity was
performed by deleting the native pyk gene in strain CS1897
which already contains an additional copy of the pyk gene
at the TP901-1 phage attachment site. PCR products
upstream to pyk using primer pyk1 (5¢-TGGTACTCGAG
CAATTTCTGAAGGTATCGAAG-3¢) and pyk2 (5¢-GG
AAGGATCCTTGTGTTTTTCTCCTATAATG-3¢) and
downstream to pyk using primer pyk3 (5¢-GGAAGGA
TCCTTTGTCAATTAATGATCTTAAAAC-3¢) and pyk4
(5¢-CTAGTCTAGATGAGCTCCAGAAGCTTCC-3¢) were
amplified. The PCR products were digested with XhoI ⁄
BamHI and BamHI ⁄ XbaI, respectively, and cloned in iden-
tical restriction sites in plasmid pG
+
host8, using E. coli
KW1 as cloning host. The resulting plasmid, pCS1919, was
used to delete pyk from the las operon by a double cross-
over event as previously described [11].
Curve fitting and calculation of control
coefficients
To estimate the control of PFK, PK and all las enzymes on
the glycolytic flux (J
glucose
), growth rate (J
l
) and on the

lactate
(a
PFK
) ¼ )0.414
*
a
PFK
+42.7, J
formate
(a
PFK
) ¼
)0.0806
*
a
PFK
+ 1.80: J
acetate
(a
PFK
) ¼ ) 0.0237
*
a
PFK
+
1.17, PK: J
l
ða
PK
Þ¼0:0298 Ãð18:7 À a

glucose
ða
PK
Þ¼À0:511 þ 56:2 Ã
a
PK
À 35:8 Ã a
2
PK
þ 6:90 Ã a
3
PK
(Polynomial Fit), J
glucose
ða
PK
Þ¼
ð52:4 Ã a
PK
À 0:533Þ=ð1 þ 0:249 Ã a
PK
þ 0:696 Ã a
PK
Þ
2
(Rat-
ional Function), J
lactate
(a
PK

3
PK
(Polynomial Fit), J
lactate
ða
PK
Þ¼
ð94:0 Ã a
PK
À 0:984Þ=ð1 þ 0:00179 Ã a
PK
þ 0:846 Ã a
2
PK
Þ (Ratio-
nal Function), J
formate
ða
PK
Þ¼3:58 Ã 0:535
a
PK
à a
1:67
PK
(Hoerl
model), J
formate
ða
PK

PK
(Polyno-
mial fit), J
acetate
ða
PK
Þ¼À0:0137 þ 1:636 à a
PK
À 0:284 Ã a
2
PK
(Quadratic fit), J
acetate
ða
PK
Þ¼1:91 Ã 0:701
a
PYK
à a
1:177
PK
(Hoerl
model), J
acetate
ða
PK
Þ¼ð0:0500 þ 1:034 à a
PK
Þ=ð1 À 0:350 Ã a
PK

las
Þ¼0:919 Ãð129 À a
las
ÞÃð1 À e
À6Ãa
2:1
las
ÞÀ75:2 (Us er
defined), J
acetate
ða
las
Þ¼0:1135 ÃðÀ30:3 À a
las
ÞÃð1 À e
À6Ãa
3:3
las
Þþ
4:66 (User defined), J
formate
ða
las
Þ¼0:173 ÃðÀ23:7 À a
las
ÞÃ
ð1 À e
À5:6Ãa
2:3
las

1 Kacser H & Burns JA (1973) The control of flux. In
Rate Control of Biological Processes (Davies DD, ed.),
Cambridge Symposium in Experimental Biology, Vol. 27,
pp. 65–104. Cambridge University Press, Cambridge.
2 Heinrich R & Rapoport TA (1974) A linear steady-state
treatment of enzymatic chains: general properties, con-
trol and effector-strength. Eur J Biochem 42, 89–95.
3 Niederberger P, Prasad R, Miozzari G & Kacser H
(1992) A strategy for increasing an in vivo flux by
genetic manipulations. The tryptophan system of yeast.
Biochem J 287, 473–479.
4 Snoep JL, Arfman N, Yomano LP, Westerhoff HV,
Conway T & Ingram LO (1996) Control of glycolytic
flux in Zymomonas mobilis by glucose-6-phosphate
dehydrogenase activity. Biotechn Bioeng 51, 190–197.
5 Jensen PR, Michelsen O & Westerhoff HV (1993) Con-
trol analysis of the dependence of Escherichia coli phy-
siology on the H
+
-ATPase. Proc Natl Acad Sci USA
90, 8068–8072.
6 Jensen PR, Westerhoff HV & Michelsen O (1993)
Excess capacity of H
+
-ATPase and inverse respiratory
control in Escherichia coli. EMBO J 12, 1277–1282.
7 Solem C, Koebmann BJ & Jensen PR (2003) Glyceral-
dehyde-3-phosphate dehydrogenase has no control over
glycolytic flux in Lactococcus lactis MG1363. J Bacteriol
185, 1564–1571.

sizes of CTP and dCTP in Lactococcus lactis. Eur J Bio-
chem 271, 2438–2445.
15 Luesink EJ, van Herpen RE, Grossiord BP, Kuipers OP
& de Vos WM (1998) Transcriptional activation of the
glycolytic las operon and catabolite repression of the gal
operon in Lactococcus lactis are mediated by the catabo-
lite control protein CcpA. Mol Microbiol 30, 789–798.
16 Westerhoff HV & van Workum M (1990) Control of
DNA structure and gene expression. Biomed Biochim
Acta 49, 839–853.
17 Ramos A, Neves AR, Ventura R, Maycock C, Lo
´
pez P
& Santos H (2004) Effect of pyruvate kinase overpro-
duction on glucose metabolism of Lactococcus lactis.
Microbiology 150, 1103–1111.
18 Thomas S & Fell DA (1998) A control analysis explora-
tion of the role of ATP utilisation in glycolytic-flux con-
trol and glycolytic-metabolite-concentration regulation.
Eur J Biochem 258, 956–967.
19 Wilson KJ, Gilles KE, Jefferson RA (1991) Beta-glucu-
ronidase (GUS) operon fusion as a tool for studying
plant–microbe interactions. In Advances in Molecular
Genetics of Plant–Microbe Interactions (Hennecke H &
Verma DPS, eds), pp. 226–229. Kluwer Academic,
Dordrecht.
20 Gasson MJ (1983) Plasmid complements of Streptococ-
cus lactis NCDO 712 and other lactic streptococci after
protoplast-induced curing. J Bacteriol 154, 1–9.
21 Brøndsted L & Hammer K (1999) Use of the integra-

29 Fordyce AM, Moore CH & Pritchard GG (1982) Phos-
phofructokinase from Streptococcus lactis. Methods
Enzymol 90, 77–82.
30 Crow VL & Pritchard GG (1982) Pyruvate kinase from
Streptococcus lactis. Methods Enzymol 90, 165–170.
31 Crow VL & Pritchard GG (1977) Fructose 1,6-dipho-
sphate-activated 1-lactate dehydrogenase from Strepto-
coccus lactis: kinetic properties and factors affecting
activation. J Bacteriol 131, 82–91.
FEBS Journal 272 (2005) 2292–2303 ª 2005 FEBS 2303
B. Koebmann et al. Control analysis of the las operon