ABSTRACT
We have screened a panel of tetracycline (tc)-like compounds for their potential use with tc-repressor (tetR) based gene switches. The interaction between tc and
tetR appears quite specific, as only tc itself and its close homologues anhydro-tc and doxycycline strongly inhibited DNA binding. However, a single tc-like compound, GR33076X, increased DNA binding of the tetR-VP16 fusion protein, both in eukaryotic cells and in
bacteria. We provide evidence that this antagonist of tetracycline is
potentially useful for accelerated gene switching, especially in whole animals.
One of the most powerful, regulatable gene switches that has been developed for
eukaryotic cells is based on the
Escherichia coli
Tn10 tetracycline (tc) operon (
1
-
3
). Briefly, the system is based on a fusion protein (tTA) which contains the DNA-binding tc-repressor (tetR) and the C-terminal domain of the herpes simplex virus VP16
trans
-activator. In absence of tc, tTA binds and
trans-
activates minimal promoters carrying multiple
tet
-operator sites. In the presence of tc, DNA binding does not occur and the
gene of interest is silent. The utility of this system has been abundantly
demonstrated in cell lines and in transgenic animals and plants (
4
-
16
). Recently, a mutant of tTA, rtTA was described that displays a `reversed'
phenotype,
i.e. it requires a tc analogue, doxycycline, for
trans
-activation, and has a silent default state (
17
).
The successful use of these regulatory systems for the production of recombinant
protein, the study of gene function or in gene therapy protocols is partly
dependent on the availability of pharmaceutical compounds tailored to specific
applications of the switch. In particular, the long half-lives of tc (4-12 h) and doxycycline (12-24 h) are an obstruction to fast on/off switching in whole
animals (
18
, and our unpublished results). We therefore set out to explore the effect of a
number of tc-like compounds on both the tTA and rtTA-mediated genetic switches.
HeLa cells were co-transfected in duplicate 3 cm wells with a 1:1 ratio of regulatable
luciferase reporter pUHC13-3 (
1
) plus tTA expression vector pUHG15-1 (
11
), or a derivative of pUHG15-1 in which we had introduced the four point mutations of rtTA (
17
). A total of 1.25 [mu]g DNA was used per well, complexed with DOTAP (Boehringer Mannheim),
according to the manufacturer's instructions. The tc-like compounds were added from a 0.25 mg/ml stock solution in
dimethylsulfoxide (DMSO) along with the DNA, until harvesting 2 days later.
Extracts were prepared as described (
19
), except that buffer without BSA was used. The protein concentration was
determined (
20
), and samples containing 20 [mu]g protein were tested for luciferase activity (
19
) in microtiter plates (FluoroNunc
®
, Nunc) using a MicroLumat LB96P luminometer (10 s delay after mixing with the
substrate followed by 10 s counting).
The Tn10
tetA
promoter plus first codons was PCR-amplified using oligonucleotides ATA GGT CTC ACC ACC GAA TGG CCA GAT GAT
TAA TTC C and CAT TTT TTG CCC TCG TTA TCT AGA TTT TTG TCG AAC TAT TCA TTT CAC,
with pASK85-D1.3 (
21
) as template. A modified green fluorescent protein (GFP) cDNA was amplified
from pBAD-GFP/AC2 (Crameri, Whitehorn, Tate and Stemmer, submitted) using
oligonucleotides AGA TAA CGA GGG CAA AAA ATG GCT AGC AAA GGA GAA GAA CT and ATA
GGT CTC ACC ACG ACA AAA AAA ATG TCG CAC AAT GTG CGC CAT TTT TCA CGA ATT CAT TAT
TTG TAG AGC TCA TCC ATG CC. This latter oligo includes the transcriptional stop
site also used in pASK85-D1.3. Finally, these two overlapping PCR fragments were mixed and
`ligated' in a PCR using the first and the last oligonucleotides mentioned
above. The resulting gene was
Bsa
I digested and cloned into the pBK-CMV (Stratagene)
Bsa
I site. Bacteria carrying this construct are strongly UV fluorescent. Into this
construct, tTA or rtTA cDNA was cloned as a
Pst
I-
Not
I fragment, PCR-amplified using oligonucleotides ATA CTG CAG GAT GTC TAG ATT AGA TAA AAG
TAA AGT GAT TAA C and GGA GGA GCG GCC GCC CCC TAC CCA CCG TAC TC, and pUHG15-1 or the derivative carrying rtTA as template.
XL1-Blue bacteria (Stratagene) were transformed with the ptetA-GFP vector. Overnight cultures were diluted 400* in Luria Broth (LB) and grown overnight at 37oC in duplicate 200 [mu]l wells in a flat-bottom 96-well plate fitted in a microtiter plate
shaker (Heidolph), in presence of kanamycin (30 [mu]g/ml), 1 mM IPTG and 1 [mu]g/ml of the tc-like compounds. The cultures were diluted 1000-fold (to ~10
6
c.f.u./ml) in filtered (40 nm pore size) LB and 10
5
particles were analysed by FACS.
A panel of tc-like compounds was composed (Fig.
1
) to examine their activities in both types of tc-regulatable switches. HeLa cells were transiently transfected with a
regulatable luciferase reporter plasmid plus a plasmid expressing either tTA or
rtTA, in the presence or absence of compounds (50 or 500 ng/ml). In absence of
compounds, tTA (Fig.
2
, black bars) induced high luciferase activity, whereas rtTA (white bars) showed
low induction. In presence of 5a,6-anhydro-tc or doxycyclin, the situation was reversed. As shown previously (
17
) and confirmed in Figure
2
, tc itself is a good co-repressor of tTA, but a poor co-activator of rtTA. Repression of tTA appears specific in that only
tc itself and its close homologues doxycyclin, anhydro-tc and compounds c, i, l and n gave good repression. Induction of rtTA
followed sometimes (c) but not always (f, o) the reverse pattern seen for tTA.
Surprisingly, one of the compounds, f, strongly increased rather than decreased
tTA
trans
-activation.
In order to obtain further insight into GR33076X's mode of action, the compound
was also tested in bacteria. A vector was constructed carrying an improved
green fluorescent protein (GFP) cDNA under the original
tetA
promoter, derived from Tn10. In addition, a tTA or rtTA cDNA was cloned into an
expression cassette in this plasmid (Fig.
4
A). In bacteria harbouring the plasmid with the tTA cDNA insert, tTA binds the
two
tet
operator sites in the
tetA
promoter. As VP16 does not transactivate transcription in bacteria, little GFP
was expressed in absence of any tc-like compound (Fig.
4
B, tTA bars); however, addition of tc or anhydro-tc resulted in increased fluorescence. Conversely, addition of GR33076X
significantly reduced the background fluorescence, by ~50%. The construct carrying rtTA showed the opposite effect with respect to
anhydro-tc, namely, a reduction of fluorescence when this compound is present.
Again, GR33076X reduced fluorescence, although its effect was relatively weaker
than for the tTA-containing construct. Constructs lacking rtTA or tTA inserts (Fig.
4
B, `none') showed high fluorescence, that was not affected by the presence of tc-like compounds. The overall fluorescence levels of the (r)tTA-carrying constructs is well below that of the `empty' vector. This
may be due to `leakiness' of the system, i.e. some binding of tetR may occur
even in presence of tc-like compounds. Alternatively, overexpression of tTA may result in cell
death; our FACS analysis cannot distinguish cell debris from true c.f.u.s.
The most straightforward explanation of our results, substantiated by their
close structural similarity (Fig.
3
A), is that GR33076X is a true antagonist of tc and its close homologues and
directly binds the tetR moiety in tTA, resulting in enhanced DNA-binding. The results in bacteria make a non-specific effect on eukaryotic transcription, or via VP16 unlikely.
Another possibility is that GR33076X somehow increases the stability of tTA,
perhaps by shielding it from proteolytic activity. A similar situation is
encountered when transcription factors are protected from proteolysis by DNA
binding (e.g
.
22
). Ligand binding may also stabilize dimerization or (in eukaryotes) nuclear
uptake.
The tetR and a number of its mutants have been studied in considerable molecular
detail, and the DNA- and tc-binding domains are well-known (
23
-
33
). Moreover, the 2.5 Å crystal structure of the [tetR
D
-tc-Mg]
+
complex has recently been elucidated (
34
). This complex has the same peptide folding as tetR
B
, which is the tetR variant used in tTA (
34
). Based on this structure it has been proposed that two strictly conserved His-residues (His
64
and His
100
) are both involved in tetR binding of the [tc-Mg]
+
complex. As two out of the three oxygen atoms (at C
2
and C
11
; Fig.
3
A) in tc that are involved in this interaction are also present in GR33076X, one
may speculate that GR33076X binds the same tetR `pocket' as tc.
On the other hand it has been observed that none of the four point mutations in
rtTA involve amino acids thought to be directly involved in tc-binding (
17
), and many of the point mutations that block anhydro-tc activation of the tetR turned out to be on the [alpha]6 helix, and not in the tc-pocket (
34
). These observations suggest that the `tetR-switch' involves amino acid residues that are not part of the tc pocket.
It is therefore also possible that GR33076X directly interacts with these
domains, instead of binding the tc pocket. The availability of tetR mutants and
structure data may be helpful in distinguishing between these possibilities.
Either way, and elaborating on the `see-saw' model for tetR conformation change upon its binding of the [tc-Mg]
+
complex (
34
), one may speculate that GR33076X somehow antagonizes the proposed displacement
of the [alpha]1-3 DNA-binding helices and `locks' the tetR instead in its tc-free conformation with DNA-binding domains exposed. If non-tc-ligated tetR is normally in dynamic
equilibrium between the two conformational states, this might explain how
GR33076X enhances tTA-mediated gene activation.
GR33076X seems highly unique in its action; compound g (Fig.
1
) is structurally very similar to it, yet it does not display the same activity.
Recently, variants of the tc-switch have been devised in which tetR alone (
35
-
37
), or fused with the KRAB repressor domain from the human Krox-1 transcription factor (
38
) was used to inhibit various promoters in which
tet
operator sites had been inserted. Since all these systems are based on the wild-type tetR, we anticipate GR33076X to function in those as well.
We observed no cytotoxicity for GR33076X in HeLa cells at the concentrations and
duration (3 d) tested. Also, GR33076X was not antibiotic for tc-sensitive bacteria at the highest concentration tested ( <= 12 [mu]g/ml; data not shown). On the basis of these results we believe
that GR33076X will prove useful for the accelerated and enhanced gene switching
with tetR based systems in both eukaryotes and bacteria.
We thank Dr A. Jaxa-Chamiec for searching the Glaxo-Wellcome database and providing us with tc-like compounds, Dr J.-P. Aubry for the FACS analyses, Dr P. Stemmer for
plasmid pBAD-GFP/AC2 and Dr A. Skerra for pASK85-D1.3.
REFERENCES
Return