Independent and tight regulation of transcriptional units in
Escherichia coli
via the LacR/O, the TetR/O and AraC/I
1
-I
2
regulatory elements
Independent and tight regulation of transcriptional units in Escherichia coli via the LacR/O, the TetR/O and AraC/I 1 -I 2 regulatory elements
Rolf
Lutz
and
Hermann
Bujard*
ZMBH Zentrum für Molekulare Biologie der Universität Heidelberg, Im Neuenheimer Feld 282, D-69120
Heidelberg
,
Germany
Received November 21, 1996;
Revised and Accepted January 7, 1997
DDBJ/EMBL/GenBank accession nos U66308-U66313
ABSTRACT
Based on parameters governing promoter activity and using regulatory elements of
the
lac
,
ara
and
tet
operon transcription control sequences were composed which permit the
regulation in
Escherichia coli
of several gene activities independently and quantitatively. The novel promoter
P
LtetO-1
allows the regulation of gene expression over an up to 5000-fold range with anhydrotetracycline (aTc) whereas with IPTG and arabinose
the activity of P
lac/ara-1
may be controlled 1800-fold.
Escherichia coli
host strains which produce defined amounts of the regulatory proteins,
Lac
and
Tet
repressor as well as AraC from chromosomally located expression units provide
highly reproducible
in vivo
conditions. Controlling the expression of the genes encoding luciferase, the
low abundance
E.coli
protein DnaJ and restriction endonuclease
Cfr
9I not only demonstrates that high levels of expression can be achieved but also
suggests that under conditions of optimal repression only around one mRNA every
3rd generation is produced. This potential of quantitative control will open up
new approaches in the study of gene function
in vivo
, in particular with low abundance regulatory gene products. The system will
also provide new opportunities for the controlled expression of heterologous
genes.
INTRODUCTION
Genetic switches which permit the control of individual gene activities quantitatively and specifically will greatly facilitate the study of gene
function
in vivo
. They would be particularly useful for the analysis of phenotypes which arise
through small perturbations of sensitive equilibria. The signalling pathway of
the heat shock response (
1
) or the control of cell division (
2
) may be just two of many examples.
In the past, regulated promoters of the
Escherichia coli
system such as P
L
of phage lambda and the promoter of the
lac
operon as well as some of its derivatives have been widely used to control gene expression (
3
-
5
). Moreover, the specialized RNA polymerase/promoter system of phages T7 and T3 was applied when particularly tight control
appeared to be required (
6
,
7
). While useful in a great number of applications, these systems have serious limitations. Thus, P
L
is commonly induced by inactivating the repressor cI 857 via a temperature shift. This induction principle does not
permit quantitative control over time and, in addition, causes pleiotropic
effects. Similar limitations exist for experimental schemes where the
introduction of, for example, phage T7 RNA polymerase into a cell via phage
infection activates a gene (
8
). On the other hand, the promoter of the
lac
operon, P
lac
, a well regulatable promoter of intermediate strength depends on the activation by CRP/cAMP. This activating complex affects, however, many additional operons and thus profoundly changes the
metabolic state of the cell when switched into its active form by cAMP. The P
lac
derivative, P
tac
(
4
) and similar constructs like P
trc
or P
tic
(
9
) which do not depend on activation are repressed to a reasonable extent only at
Lac
repressor concentrations which hardly allow full induction. The more recently
described systems where promoters of the
ara
(
10
) and the (Tn
10
)
tet
operon (
11
) were employed are useful alternatives; their range of regulation and their
tightness in the repressed state may, however, fall short when compared with
the system described here although we have not performed direct comparisons.
Here we describe a system for the quantitative and independent control of two
transcription units in
E.coli
. The centerpiece of the system are regulatable promoters which were developed
following principles described earlier (
12
). They are controlled by elements of the
lac
,
ara
or
tet
(Tn
10
) operon and, accordingly, promoter activities are sensitive towards IPTG,
arabinose or tetracycline, respectively. These promoters are tightly
repressible and can be regulated over an up to 5000-fold range. By varying the plasmid copy number the regulatory range of
these promoters can be shifted to span different windows.
Escherichia coli
strains which produce defined amounts of
Lac
and
Tet
repressor (
Lac
R,
Tet
R) as well as of AraC ensure reliable intracellular conditions. The tightness of
the system is demonstrated by quantitative control of a low abundance protein
of
E.coli
as well as by the stable maintenance of a gene encoding a restriction
endonuclease. This endonuclease is upon induction efficiently overproduced
despite of the immediate growth arrest of the culture.
MATERIALS AND METHODS
Construction of the pZ vector system
Construction of promoters
Promoters P
LlacO-1
, P
A1lacO-1
, P
LtetO-1
and P
lac/ara-1
were obtained by total synthesis. The
lac
operator
O1
upstream of P
lac/ara-1
was introduced via a PCR primer with the corresponding overhang and cloned as a 5'-
Aat
II-
Xho
I-3' fragment upstream of the promoter. The intervening sequence
between the promoter and the upstream operator was derived from the human
c-myc
gene (
23
) to minimize recombination and potential transcriptional signalling. After cloning, all promoter sequences were verified by dideoxy
sequencing (
24
).
Cloning of the restriction endonuclease
Cfr
9I
The gene encoding restriction endonuclease
Cfr
9I was amplified from vector pCfr9I2.3X (
25
) by PCR and cloned into vectors pZS*24 and pZA24, respectively, via
Kpn
I/
Xba
I or
Eco
RI/
Xba
I. The utilization of the
Kpn
I cleavage site resulted in a mRNA with a strong RBS, whereas the RBS generated
via
Eco
RI was ~10 times less efficient.
Construction of
E.coli
strain DH5
[alpha]
Z1
Escherichia coli
strain DH5[alpha]Z1 was obtained following the description of Diederich
et al
. (
19
). For integration of plasmids of the pZ series into the chromosome, the lambda
attP
site pLDR8 was cloned into the
Avr
II site of pZS4Int. For integration, the origin of replication was removed by
cleavage with
Spe
I and
Avr
II (generating compatible cohesive ends) and the religated fragment was transferred to
E.coli
DH5[alpha]pLDR8 by electroporation. Cells were incubated for 2 h at 42oC and then at 37oC overnight and transformants were selected on LB Sp[50 [mu]g/ml] plates.
Determination of
in vivo
promoter activities
Promoters P
LlacO-1
, P
LtetO-1
and P
lac-ara-1
, respectively, were inserted into plasmids of the pZ series and the expression
of the eukaryotic
luciferase
gene of
Photinus pyralis
(
17
,
26
) was measured by monitoring its enzymatic activity. Overnight cultures of
E.coli
cells DH5[alpha]Z1 grown at 37oC in LB medium containing the appropriate antibiotics were diluted
1:100 in LB medium in presence or absence of various inducers [1 mM IPTG, L(+)-arabinose, anhydrotetracycline] at concentrations indicated. After 3 h,
the OD
600
was measured and the cultures were kept at room temperature for 15 min. To
determine luciferase activities in crude extracts of logarithmically growing
cultures 3 ml cells were sedimented, resuspended in 50 [mu]l lysis buffer (1 mM EDTA, 1 mg/ml lysozyme) and incubated at room temperature for 15 min. Upon addition of 300 [mu]l H
2
O and 300 [mu]l buffer I (100 mM KH
2
PO
4
, 1 mM DTT, pH 7.8) 35 [mu]l were mixed with 250 [mu]l buffer II (15 mM MgSO
4
, 25 mM glycylglycin, 2.5 mM ATP) and luciferase activity was measured (10 s,
delay 0 s) in a Berthold Lumat type LB9501. Activities are given as `relative
light units' (RLU) after subtraction of the instrumental background and
normalization to the number of viable cells (
27
).
Enzymes, antibodies, media and chemicals
Standard DNA manipulations were carried out as described (
30
). All enzymes were purchased from Boehringer Mannheim. DNA sequencing reactions
were performed using the Pharmacia T7 sequencing kit. Synthetic oligonucleotides and sequencing primers were supplied by the inhouse facility. Antibiotics were added to the growth medium at the following concentrations: 100 [mu]g/ml ampicillin, 40 [mu]g/ml kanamycin; 25 [mu]g/ml chloramphenicol and 50 [mu]g/ml spectinomycin. Luciferin, IPTG and standard chemicals, p.a. grade, were purchased from AppliChem, l(+)-arabinose from Sigma while anhydrotetracycline was
obtained from Acros. Radiochemicals were purchased from Amersham & Buchler.
Anti-DnaJ rabbit serum for immunoblots, prepared in house, was diluted 1:7500
for the preparation of immunoblots. Specific antibody-DnaJ complexes were detected with alkaline phosphatase-conjugated anti-rabbit IgG (Promega) as described (
29
).
RESULTS
Rational of promoter designs
The decisive parameter for the efficient repression of promoters where
repressors interfere directly with the binding of RNA polymerase is the rate of
complex formation (
k
ON
) between RNA polymerase and promoter (
15
). Promoters which bind RNA polymerase at low rates are well repressed since
they give the binding of the repressor a competitive advantage. Such promoters, however, remain weak upon induction unless they are activated as, for example,
is the case for P
lac
. By contrast, promoters which are strong in the absence of any activator bind
RNA polymerase efficiently and can in general not be well repressed. We have
developed two classes of repressible promoters: those which, after combination
with operators, still initiate RNA synthesis efficiently and those which
require activation in the derepressed state. The first class is derived from
strong phage promoters such as P
L
of phage lambda (
31
) and P
A1
of phage T7 (
32
). Members of the second class are derivatives of P
lac
. Sequences of the
lac
or
tet
operator were inserted within the various promoters at positions previously
shown to be most effective (
15
), particularly in the downstream or within the spacer region, position III and IV, respectively
(Fig.
1
). Moreover, in some constructs the effect of auxiliary operators of the
lac
system was exploited by placing a third
lac
operator sequence in position VI upstream of the promoter (Fig.
1
). For activating `low k
ON
promoters', AraC has been utilized which in contrast to CRP/cAMP acts highly
specifically.
.
(
a
)
Induction and repression of P
LtetO-1
, P
LlacO-1
and P
A1lacO-1
in
E.coli
DH5[alpha]Z1. (
b
) Induction and repression of P
lac/ara-1
in
E.coli
DH5[alpha]Z1
Promoters were inserted upstream of the luciferase gene in the pZ vectors containing the origin of replication indicated. The various constructs were
transfered into
E.coli
DH5[alpha]Z1. Overnight cultures of such transformants were diluted 1:100 in LB
medium and grown up in presence or absence of aTc or IPTG, respectively. The concentration of aTc was 100 ng/ml, of IPTG 1 mM. At OD
600
= 0.5, cells were harvested and luciferase activity was determined. The luciferase activities given are the mean values of five independent
experiments (standard deviation <10%). The intracellular copy numbers were determined by comparing luciferase
activity of cells harbouring the respective plasmids with the activity in cells
containing only a single luciferase expression unit integrated in the chromosome (data not shown). They agree well with previously published data derived from direct copy number measurements (14). In Table 1b, P
lac/ara-1
was induced either by 1 mM IPTG alone or by 1 mM IPTG and l(+)-arabinose (0.05%).
Construction of promoters controlled by
Tet
R or
Lac
R
Promoter P
L
of phage lambda has a low homology score and binds RNA polymerase with a
moderate forward rate constant of 1.1 * 10
7
M
-1
s
-1
(
31
). It is a strong promoter
in vivo
which, nevertheless, can be tightly repressed by cI, the lambda repressor. We
have replaced the cI binding sites with sequences encoding the operator 2 (
tet
O2) of the Tn
10
tetracycline resistance operon (
33
). The resulting 74 bp promoter-operator sequence, P
LtetO-1
, obtained by oligonucleotide synthesis contains a
tet
O2 sequence in position V and a 18 bp
tet
O2 core sequence in the spacer region (Fig.
1
). P
LtetO-1
is tightly repressible by the
Tet
repressor and can be regulated over an up to 5000-fold range by supplying anhydrotetracycline (aTc) to the culture (Table
1
a). In an analogous way,
lac
O1 sequences were integrated into P
L
(Fig.
1
): an 18 bp sequence in the spacer region (overlapping by 1 bp with the -33 hexamer) and a 22 bp sequence upstream of the promoter centred around
position -43 (overlapping by 2 bp with the -33 hexamer). The activity of the resulting promoter P
LlacO-1
can be regulated over a >600-fold range by IPTG in
E.coli
DH5[alpha]Z1 (Table
1
a).
Previously we have modified P
A1
of phage T7 in a similar fashion (Lanzer and Bujard, unpublished) by inserting
two
lac
operator sequences into position III and IV, i.e. into the spacer and the
downstream region (Fig.
1
). This strong promoter binds RNA polymerase with a relatively high forward rate
constant (
34
) and although the
lac
operator sequence in position III reduces the rate of promoter clearance (
22
), P
A1lacO-1
remains a strong promoter
in vivo
when derepressed. At
Lac
repressor concentrations as in
E.coli
DH5[alpha]Z1, this promoter can be regulated over a 350-fold range by IPTG (Table
1
a).
A promoter controlled by
Lac
R and AraC
P
lac
, when derepressed by IPTG and activated by CRP/cAMP, is a promoter of
intermediate activity
in vivo
(
32
). Some mutants of P
lac
show increased activity
in vivo
but remain susceptible to repression as well as activation. One of these mutant
promoters is P
lac-8A
. It differs from the wild type by a single T to A base change at position -8 (Fig.
1
) and has a 3-fold higher promoter strength
in vivo
when compared with P
lac
(
35
). Since P
lac-8A
has still a low homology score and consequently binds RNA polymerase rather slowly,
it is tightly repressible and can be activated by CRP/cAMP (
35
). To convert P
lac-8A
into a well regulatable promoter, we have introduced three modifications.
First, a symmetrical 20 bp
lac
operator sequence (Os, Fig.
1
) was placed in the spacer region. Second, a 35 bp wild-type operator sequence (
lac
O1) was integrated upstream of the promoter at position -448 following principles described previously (
36
-
38
). Third, the CRP/cAMP binding site was deleted and replaced by the I1/I2
recognition site of AraC, the repressor-inducer of the BAD promoter of the
ara
operon (
39
). To maintain the -35 hexamer of P
lac-8A
and to centre the I
1
/I
2
sequence around -53 as in the
ara
operon, 5 bp of the I
2
site were abolished. The resulting P
lac/ara-1
was examined for its regulatory potential in
E.coli
as described below. As shown in Table
1
b, this promoter can be regulated over an ~1800-fold range whereby derepression via IPTG causes an ~100-fold and activation via arabinose a 15-20-fold increase in promoter activity.
The pZ vector system
The vectors depicted in Figure
2
emerged from our earlier developments, the pBU (
40
), pDS (
14
) and pUH (Lanzer and Bujard, unpublished) series. The salient feature of the pZ
plasmids is their modular structure. Module I contains all the regulatory
elements which control the expression of a gene of interest i.e. the
regulatable promoter, a ribosomal binding site (RBS) and a transcriptional
terminator. In the basic pZ plasmid, this module harbours multiple cloning
sites (MCS). The transcriptional signals as well as the RBS can be exchanged
using unique cleavage sites. Module II contains an origin of replication which
is protected from outside transcriptional readthrough (
14
) by two terminators. Four origins of replication were adjusted to fit into the
system via unique cleavage sites. This permits the variation of the plasmid
copy number as well as of the compatibility group. Thus, when the ColE1, the
p15A or the pSC101 origin of replication is used, intracellular copy numbers of
50-70, 20-30 and 10-12, respectively, are established. Particularly low copy
numbers are achieved with the origin of pMPP6 (
20
), a derivative of pSC101 which gives rise to only three to four plasmids per
cell and is referred to in our system as pSC101* origin. Finally, module III
carries a resistance marker and the genes encoding ampicillin, kanamycin,
chloramphenicol and spectinomycin resistance together with their genuine
transcription and translation signals were again adjusted to fit into the
constructs via unique cleavage sites. The nomenclature of the pZ plasmid family
is explained in Figure
2
. Of particular interest for the study here were the plasmids which contain a
modified luciferase gene (
17
) as reporter for promoter activity. The effect of using different origins of
replication led to a 15-20-fold shift of the regulatory window. This is most clearly
demonstrated by comparing the luciferase activities in the repressed state of P
LtetO-1
and P
lac/ara-1
when integrated into pZE, pZA and pZS* (Table
1
).
The
E.coli
host strain DH5
[alpha]
Z1
To ensure stable and defined conditions for the synthesis and maintenance of the
regulatory proteins
Tet
R and
Lac
R, the genes encoding these two repressor molecules were placed under the
control of the two constitutive promoters P
N25
and the
laci
q
promoter P
i
q (
41
), respectively, and integrated in tandem into the chromosome of
E.coli
strain DH5[alpha] at the phage lambda attachment site (
42
) as outlined in Figure
3
. Analysis of several spectinomycin-resistant colonies by Southern blot analysis (data not shown) showed that
the two transcription units encoding
Tet
R and
Lac
R as well as the spectinomycin resistance marker were stably integrated in the
DH5[alpha] genome. The resulting strain, DH5[alpha]Z1, produces ~3000 molecules of
Lac
R and around 7000 molecules of
Tet
R per cell as determined by ELISA and Western blot (data not shown). Since
E.coli
DH5[alpha]Z1 is a genuine producer of AraC, all regulatory proteins required are constitutively synthesized in the cells which were used throughout the experiments
described here. The entire unit encoding
Lac
R,
Tet
R and Sp
r
can be readily transferred to other
E.coli
strains by phage
P1
transduction as exemplified for the widely used W3110 strain which led to
E.coli
W3110Z1 (data not shown).
Regulation of the activity of promoters P
LtetO-1
, P
LlacO-1
, P
A1lacO-1
and P
lac/ara-1
Promoter P
LtetO-1
is controlled by the operator repressor system of the Tn
10
-derived
tet
resistance operon. Accordingly it is induced by tetracyclines of which anhydrotetracycline is presently the most suitable one. The other promoters are all induced by IPTG and P
lac/ara-1
can, in addition, be activated by L(+)arabinose.
The regulatory range of all promoters was determined using the luciferase gene
as reporter unit in absence and presence of the respective inducers. To assess
the influence of the intracellular plasmid copy number, P
LtetO-1
and P
lac/ara-1
were inserted in plasmids of the pZ-family containing the replication origin of plasmids ColE1, p15A and
pSC101*, respectively. The host strain in all experiments was DH5[alpha]Z1. The most highly repressed state and the largest regulation factor
exceeding a 5000-fold range was found with P
LtetO-1
when carried on a low copy number plasmid (Table
1
a). Higher intracellular plasmid numbers increased the luciferase activity accordingly (7-fold for p15A and 15-fold for ColE1). The luciferase activities in the repressed state
did, however, correlate only qualitatively with the copy number. Both P
LlacO-1
and P
A1lacO-1
are repressed to about the same level. However, since upon induction P
LlacO-1
produces twice the amount of luciferase, its regulation factor is higher (620- versus 350-fold). Examining P
lac/ara-1
, a regulatory range of 1700-1800-fold is found irrespective of the intracellular plasmid copy number which,
nevertheless, affects the absolute values of repression and induction (Table
1
b). The lowest luciferase activity in the repressed state was again observed with plasmids of the pZS* series as
expected. For all promoters, the activity in the fully induced state was
identical to their activity in the repressor-free strain DH5[alpha] (data not shown).
The potential to quantitatively control a gene activity with the promoters
described is exemplified by experiments depicted in Figure
4
. The luciferase gene as well as the gene encoding the low abundance
E.coli
chaperone DnaJ were placed under the control of P
LtetO-1
, P
LlacO-1
and P
lac/ara-1
, respectively and the activity of the promoters was analyzed at various
concentrations of inducers. The dose response curves show that partial
induction can be achieved with all promoters and that P
lac/ara-1
can be tuned particularly well since induction with IPTG and activation with
arabinose allows a high degree of differentiation. The induction curve of
promoter P
LtetO-1
suggests a strong cooperative effect in the binding of the inducer aTc to the
Tet
repressor. The same phenomenon was observed with several other
Tet
R regulated constructs (data not shown). The lower part of Figure
4
shows the controlled expression of DnaJ. Western blots demonstrate that the
repressed state is hardly different from the cellular background (~100 DnaJ molecules/cell) (
43
) whereas full induction yields high levels of expression with all three
promoters.
Cloning and expression of a gene encoding restriction endonuclease
Cfr
91
Based on earlier results (
44
,
45
) it can be estimated that under repression conditions promoters like P
LtetO-1
and P
lac/ara-1
when placed on a low copy number plasmid such as pZS* produce less than one
mRNA per cell. This should permit the cloning of genes encoding highly toxic
products. To test this prediction, the gene of the
Cfr
9I restriction endonuclease was cloned in absence of its cognate
methyltransferase. The coding sequence of
Cfr
9I was placed under the control of P
lac/ara-1
in plasmids pZS*24[Delta]RBS, pZA24[Delta]RBS (where RBSII was deleted) and pZA24. In all three plasmids,
the gene could be stably maintained in DH5[alpha]Z1 and growth rates of cells harbouring pZS*24[Delta]RBS-cfr were indistinguishable from cells without any plasmid
(Fig.
5
a). However, cells containing pZA24-cfr formed colonies with a mucoid phenotype. Induction of transcription by
IPTG or by IPTG and arabinose led to immediate growth arrest of the culture
(Fig.
5
b and c). Since in
E.coli
protein synthesis can continue for some time after the destruction of
chromosomal DNA (
46
) the feasibility of producing
Cfr
9I endonuclease in DH5[alpha]Z1 was examined. Indeed using pZA24-cfr, the endonuclease could be produced to a level corresponding to
~2% of the total cellular protein, despite immediate growth arrest of the
culture upon induction (Fig.
5
d).
DISCUSSION
The transcription control systems described here expand our capabilities of
studying gene function
in vivo
. First, gene activities can be regulated over a wide range spanning more than
three orders of magnitude but more importantly they can be repressed extremely tightly. This opens up the possibility of varying the
concentrations of regulatory proteins which, under physiological conditions,
are present at very low levels. Examples for such proteins may be the central
heat shock regulator of
E.coli
[sigma]
32
, the chaperone DnaJ or ftsZ, a crucial component in the signalling pathway of
cell division. Second, by exploiting the three regulatory principles,
Lac
R/O,
Tet
R/O and AraC/I
1
-I
2
, several gene activities can be independently regulated. This will allow the
analysis of intracellular equilibria by varying the concentrations of
participants and elucidate their contribution to a phenotype.
The crucial developments for the expression system described here were the
promoter-operator combinations which were conceived following principles described earlier (
12
,
15
). Accordingly, promoters were selected which exhibit low or intermediate rates of complex formation with RNA polymerase. Moreover, operators were
positioned in regions shown to be most effective. Thus, provided a 17-19 bp operator sequence binds a repressor sufficiently tightly, it can be
accommodated in the spacer region of a promoter where it interferes with RNA
polymerase binding most efficiently (
15
) and where it perturbs least the functional program of a promoter. The second
best choice for placing an operator is position III where the
lac
operator sequence, however, diminishes promoter clearance by RNA polymerase (
22
).
For the first class of regulatable promoters, P
L
of phage lambda served as a paradigm. It is a strong and highly repressible
promoter
in vivo
which, however, binds RNA polymerase with a moderate forward rate constant. By
combining this promoter with
tet
operators, P
LtetO-1
was obtained whose activity can be controlled via
Tet
R and anhydrotetracycline. It is a strong promoter
in vivo
and can, nevertheless, be repressed up to 5000-fold in
E.coli
DH5[alpha]Z1. This is the widest range of regulation measured for any
E.coli
promoter so far using the Luciferase reporter system. Partial induction of P
LtetO-1
is achieved by varying the concentration of aTc (Fig.
4
a). In contrast to tetracycline, anhydrotetracycline is a particularly useful
inducer. It binds
Tet
R with an ~35-fold higher binding constant and thus allows to operate at very low
concentrations. At the same time, its antibiotic activity is ~100-fold lower (
47
) and concentrations of <50 ng/ml as required for the full induction of P
LtetO-1
have no effect on the growth of
E.coli.
The finding that repression is less effective at higher plasmid copy numbers
may be due to the different ratio of operators to repressors as well as to the
increase in unspecific binding sites which affects the concentration of free
repressor. Following the same strategy but using the
lac
operator sequences, P
LlacO-1
was constructed. It is a strong promoter which can be regulated over a >600-fold range. From the results shown in Table
1
, we anticipate that placing this promoter into low copy number plasmids, it
will permit a similar tight repression of transcription as P
LtetO-1
and will therefore also be suitable for controlling gene products at very low
intracellular levels. Promoter P
A1lacO-1
contains one of its
lac
operators in position III (Fig.
1
) which limits the rate of promoter clearance by RNA polymerase. Thus, it is a
somewhat weaker promoter which, nevertheless, is well regulatable (Table
1
a).
While P
L
is an example for a highly repressible, strong promoter with a moderate
k
ON
, P
lac
is an example for a promoter whose high repressibility is due to its low rate
of polymerase binding. This, however, limits its activity in the derepressed
state. For full activity it requires the upstream binding of CRP/cAMP. But even when fully activated, P
lac
remains a moderately strong promoter. Examining a number of P
lac
mutants, P
lac-8A
exhibited interesting features: its
in vivo
strength when derepressed but not activated was 16 times but its
k
ON
only three times higher than that of P
lac
. It also could still be activated by CRP/cAMP. The repressibility of this
promoter was optimized by introducing a symmetrical 20 bp
lac
O sequence into the spacer region (overlapping with the -10 and the -35 hexamer by 1 bp each) and by placing a third operator at
position VI (Fig.
1
). To avoid pleiotropic effects by CRP/cAMP activation, the AraC binding site I
1
-I
2
of the
ara
BAD promoter replaced the CRP/cAMP site. The resulting promoter (P
lac/ara-1
) is regulatable over an ~1800-fold range and when fully induced and activated it exceeds the
in vivo
strength of P
lac
6-fold. Thus, it is a strong and highly regulatable promoter. The fine
tuning of P
lac/ara-1
is facilitated by a two step mechanism: increasing the IPTG concentration in
the medium up to 0.2 mM leads to an ~100-fold induction which can be enhanced 15-20-fold by adding arabinose to a final concentration of
0.03% (Fig.
4
c).
Addition of glucose (0.6%) to the growth medium decreased the activation
potential of AraC 2-3-fold (data not shown). This is most likely due to the reduction of
araC
transcription which is controlled by CRP/cAMP (
48
). This glucose effect can of course be avoided by replacing glucose with
glycerol or other non-PTS sugars as a primary carbon source when, for example, minimal medium is
required for culturing.
Repression and induction depend on a number of parameters such as the
concentration of free repressor and the increment by which an inducer decreases
the affinity of a repressor to its operator. Free repressor concentration is
also a function of the number of unspecific (and specific) DNA binding sites
and may thus be affected by plasmid copy number and size although this is a
minor parameter with the plasmids described herein. A simple increase of the
intracellular repressor concentration on the other hand does not necessarily
compensate for this effect since the residual affinity of the repressor-inducer complex to the respective operator sequence prevents full induction as seen for both TetR and LacR (data not shown). Moreover, high repressor concentrations may be toxic for the cell as is the case for TetR (ref.
49
and our unpublished results). Incomplete induction is frequently encountered with the widely used tac or trc type promoter systems because these high `k
ON
' promoters are reasonably well repressed only at very high intracellular repressor concentrations. When examined under
conditions as defined in Table
1
repression of these promoters is only 10-20-fold (data not shown). It is therefore important to establish
stable intracellular conditions where the relevant regulatory proteins are
present in defined concentrations which warrant a reliable control of promoters
under various physiological conditions. This was achieved by integrating the
laci
as well as the
tetR
gene controlled by promoters of appropriate
in vivo
strength into the
E.coli
chromosome. The high `k
ON
' constitutive promoters P
i
q
and P
N25
ensure efficient transcription even under conditions of reduced concentration
of active [sigma]
70
RNA polymerase e.g. in stationary phase. The resulting
E.coli
strains DH5[alpha]Z1 and W3110Z1 produce constitutively around 3000 tetrameric
Lac
and 7000 dimeric
Tet
repressors per cell during logarithmic growth. Sufficient AraC is supplied by
its natural autoregulated pathway as its overproduction from a plasmid did not
lead to increased activation of P
lac/ara-1
(data not shown). Thus,
E.coli
strains of the DH5[alpha]Z1 type provide all regulatory proteins required in appropriate amounts
for tight repression and full induction (which is indistinguishable from
repressor-free host strains; data not shown) at different plasmid copy numbers. The
tight repression is maintained also in stationary phase and in overnight
cultures (data not shown). The placement of repressor encoding units onto the
chromosome has also simplified the vector constructs and increased the degree
of freedom of the system.
Although the regulatory range of the promoters described is large, it may not
satisfy all needs. For example for the tight control of a low abundance or
toxic gene product, even the fully repressed P
LtetO-1
may generate a too high background when contained in a ColE1-type plasmid. The vector system therefore offers still another degree of
freedom. By utilizing different origins of replication, the intracellular
number of plasmids can be varied between ~4 and 60, which permits to shift the regulatory window of a promoter within
an ~15-fold range. Thus, by fully exploiting the potential of the system
using, for example P
LtetO-1
, a gene's activity can be controlled over an ~60 000-fold range. The controlled synthesis of a restriction endonuclease, a
low abundance
E.coli
protein and luciferase under different conditions as exemplified in Figures
4
and
5
illustrates some of these aspects. Needless to say that three of the
replication origins adjusted to fit the vector system belong to different
plasmid compatibility groups and thus permit to maintain two or even three
vectors within DH5[alpha]Z1 cells if required.
It may be of interest to speculate on the absolute tightness achieved, for
example, with P
LtetO-1
in DH5[alpha]Z1 cells. When fully induced, this promoter has an activity of ~30 P
bla
units (
45
) and is estimated to initiate transcription ~5-fold less frequently than the fully activated
rrnB
P1 promoter (
12
). The
rrnB
P1 promoter is estimated to initiate 1.5 mRNAs/s at maximal growth rates during
logarithmic growth (
44
). Hence, it can be estimated that P
LtetO-1
initiates 0.3 mRNAs/s. Given a generation time of 25 min for
E.coli
in log phase cultures a 5000-fold repression of this promoter would reduce this rate to 6.5 * 10
-5
mRNAs/s or in other words one mRNA every 10th generation would be synthesized
in a single copy situation. Thus, P
LtetO-1
located on a plasmid of the pZS*-type giving rise to three to four copies/cell will produce one mRNA about every 3rd generation. The luciferase activity
monitored with P
LtetO-1
in the repressed state which corresponds to an average of 12 enzyme molecules
per cell is not in disagreement with these estimates. This suggests that at the
repression levels achieved only a fraction of a cell population synthesizes a
given gene product at any one time. Populations would therefore survive if this
gene product was highly poisonous as for example the restriction enzyme Cfr9I
since only a minor portion of cells would die.
The tight control of transcription, the potential to regulate gene activities
quantitatively over wide ranges and the possibility to control independently
several transcription units in a cell are the main advantages of the system
described here when compared to other commonly used promoter/vector
combinations. It thus opens up new perspectives for the study of cellular
physiology as well as for the controlled expression of heterologous genes.
ACKNOWLEDGEMENTS
We thank Dr Messer for plasmids and host strains of the chromosomal integration
system, Dr Bukau for plasmid pBB1 harbouring the spectinomycin resistance gene
and for a plasmid encoding DnaJ and Dr Janulaitis for providing a plasmid harbouring the
Cfr
9I restriction system. We are grateful to Dr Frank for the synthesis of oligonucleotides. This work was supported by the Deutsche
Forschungsgemeinschaft SFB229, by the Fonds der Chemischen Industrie Deutschlands and in part by the Bündesministeriüm für Bildüng und Forschung (no. 0311146).
38 Shore,D. and Baldwin,R.L. (1983) J. Mol. Biol., 170, 4, 957-981.
39 Schleif,R. (1992) In Transcriptional Regulation. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 643-665.
40 Gentz,R., Langner,A., Chang,A.C.Y., Cohen,S.N. and Bujard,H. (1981) Proc. Natl. Acad. Sci. USA78, 4936-4940.MEDLINE Abstract
41 Müller-Hill,B., Crapo,L. and Gilbert,W. (1968) Proc. Nat. Acad. Sci. USA59, 1259-1264.
42 Weisberg,R.A. and Landy,A. (1983) In Hendrix,R.W., Stahl,F.W. and Weismann,R.A. (eds) Lambda II. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 211-250.
43 Neidhard,F.C. and VanBogelen,R.A. (1987) In Neidhard,F.C. (ed.) Escherichia coli and Salmonella typhimurium. Washington, DC, pp. 1334-1345.
