Modulation of tyrT promoter activity by template supercoiling in vivo.

4 downloads 143 Views 2MB Size Report
Aug 30, 1994 - in topological environment between topA and top' hosts. We therefore decided to re-examine the tyrT promoter system in two respects. First, we ...
The EMBO Journal vol.13 no.23 pp.5647-5655, 1994

Modulation of tyrT promoter activity by template supercoiling in vivo

Richard P.Bowater, Dongrong Chen and David M.J.Lilley1 CRC Nucleic Acid Structure Research Group, Department of Biochemistry, The University, Dundee DDI 4HN, UK 'Corresponding author Communicated by D.M.J.Lilley

We have found that initiation of RNA synthesis at the tyrT promoter of Escherichia coli can be stimulated on a plasmid by a factor of 4-6 by elevation of DNA supercoiling in vivo. Increased unconstrained plasmid supercoiling was achieved by inserting the tyrT promoter upstream of the tetracycline resistance gene tetA and transformation into a topA host. Under these conditions there is marked oversupercoiling of the plasmid DNA and we have shown previously that this can lead to increased promoter activity in the topological domain created. A critical element in the formation of this domain is the coupled transcription, translation and membrane insertion of tetA and we show that all of these events are important in the stimulation of tyrT promoter activity. The magnitude of the stimulation is in reasonable agreement with that measured in vitro as a function of plasmid supercoiling, if the unconstrained level of negative supercoiling in vivo is increased from a basal level of -a 0.022 to -a -0.052 by transcription-induced supercoiling. The induced supercoiling is very efficient, indicating that the tyrT promoter is itself contributing to the steady-state level despite a total lack of membrane anchorage for the tyrT transcription unit. This study provides a new example of the topological coupling of promoters. Key words: DNA topology/RNA polymerase/transcription

Introduction Initiation of transcription involves changes in the local winding of DNA as RNA polymerase binds to, and then unpairs, the DNA (Amouyal and Buc, 1987). Any events that lead to alteration in the winding of the DNA will be coupled to the topology of the molecule and thus the efficiency of transcriptional initiation could be affected by the prevailing level of negative supercoiling in the template. It has been known for many years that transcription in general, and some promoters in particular, are affected by DNA topology in vitro (Botchan et al., 1973; Richardson, 1974; Yang et al., 1979; Bertrand-Burggraf et al., 1984; Borowiec and Gralla, 1987) and in vivo (with topology modulated chiefly by inhibitors of DNA gyrase) (Javor, 1974; Sanzey, 1979; Kano et al., 1981; Engle et al., 1982). This is exemplified by the promoters of the K Oxford University Press

genes encoding DNA topoisomerases in Escherichia coli (Menzel and Gellert, 1983; Tse-Dinh, 1985; Tse-Dinh and Beran, 1988). A general role of DNA supercoiling as part of the cellular regulatory repertoire of gene expression has been discussed (Smith, 1981; Higgins et al., 1988; Pruss and Drlica, 1989; Lilley and Higgins, 1991). It is relatively difficult to demonstrate an effect of DNA topology on the function of a promoter inside the cell, particularly if the use of potentially pleiotropic inhibitors of DNA gyrase is avoided. However, transcription-induced supercoiling provides a way in which the topological state of plasmid DNA can be manipulated in vivo. Liu and Wang (1987) proposed that hindrance to rotation of elongating transcription complexes could lead to induction of positive and negative domains of supercoiling, ahead of and behind the RNA polymerase respectively. In a topA mutant of eubacteria such as E.coli there is reduced relaxation of the negative supercoiling, which may lead to an increased steady-state level of negative supercoiling in certain circumstances. Transcriptional supercoiling can be very significant when the transcriptional complex is anchored, by coupled transcription, translation and membrane insertion of the nascent polypeptide, as occurs with the tetracycline resistance gene tetA (Lodge et al., 1989; Chen et al., 1992; Lynch and Wang, 1993). We have shown that such supercoiling effects can exert a significant influence on promoter function in vivo in the case of the leu-500 promoter (Chen et al., 1992). The mutant promoter leu-500 was originally isolated from a leucine auxotroph of Salmonella typhimurium (Mukai and Margolin, 1963; Dubnau and Margolin, 1972; Gemmill et al., 1984). We have shown that this promoter can be activated on a plasmid in topA E.coli or S.typhimurium hosts, provided that it is located upstream of a functional tetA gene (Chen et al., 1992, 1993, 1994). The activation of the leu-500 promoter correlates with an elevation of unconstrained DNA supercoiling in the plasmid template generated by the twin supercoiled domain mechanism, which is not relaxed in topA hosts. The presence of the tetA gene anchors the transcribing RNA polymerase and so provides a very powerful induction of DNA supercoiling. Plasmids containing a functional tetA gene exhibit increased levels of unconstrained negative supercoiling, demonstrated by in situ probing of reporter sequences (Bowater etal., 1994). Moreover, these plasmids display unusual topoisomer profiles that include a fraction that is highly oversupercoiled when extracted from the cell (Pruss and Drlica, 1986; Lodge et al., 1989; Chen et al., 1992, 1994). Interference with either transcription or translation of tetA or with membrane insertion of its gene product reduces the activity of the leu-500 promoter in cis, together with the level of unconstrained negative supercoiling of the cellular plasmid (Bowater et al., 1994) and superhelix densities of extracted plasmids (Chen et al., 5647

R.P.Bowater, D.Chen and D.M.J.Lilley

A PtyrT

X bla

ga/K

bla

tetA

pTYRTtetAdiv tetA

B

-35

-10 _

_

AACACTTTACAGCGGCGCGTCATTTGATATGATGCGCCCCGCTTCCCGATAAGGGAGCAGGCCAGTAAAAA

TTGTGAAATGTCGCCGCGCAGTAAACTATACTACGCGGGGCGAAGGGCTATTCCCTCGTCCGGTCATTTTT

Fig. 1. Schematic maps of the plasmids used in these studies, and the sequence of the tyrT promoter. (A) Plasmid maps. pTyr2 (Lamond and Travers, 1983) is the plasmid used as a source of the tyrT promoter. This was used to construct the plasmids pTYRTtetApar and pTYRTtetAdiv, containing the tyrT promoter upstream of the tetA gene, with either the same or divergent orientation respectively. The black boxes show the location of transcriptional terminators, without which it is not possible to clone the very strong tyrT promoter. (B) Sequence of the tyrT promoter, with the transcriptional start site and -10 and -35 sequences highlighted.

1994). Thus there is a good correlation between the in vivo function of this promoter and the local level of template supercoiling. We have coined the term topological coupling to cover the modulation of one promoter by the activity of another by means of changes in local levels of DNA supercoiling. A promoter whose supercoiling dependence has been

well studied is that of the tyrT gene (Berman and Landy, 1979; Travers et al., 1982, 1983; Lamond and Travers, 1985), which encodes one of the two major tRNATyr species of E.coli. This promoter has a high rate of transcriptional initiation in vivo, commensurate with the highly efficient rRNA promoters (Lamond and Travers, 1983). At physiological salt concentrations, in vitro transcription initiated at a plasmid-borne tyrT promoter only when the DNA template was negatively supercoiled (Lamond, 1985), whereas the promoter was inactive on linear DNA. The supercoiling dependence was a property of the primary promoter sequences, but this sensitivity to supercoiling could be removed by reduction of the salt concentration below 50 mM KCl. The in vitro results with the tyrT promoter provided a striking example of the modulation of promoter activity by the level of DNA supercoiling in the template. However, when Lamond (1985) sought to demonstrate a corresponding effect in vivo the results were disappointing. No difference in the amount of transcription in E.coli strains that carried mutations in either the DNA gyrase or topoisomerase I genes could be detected. Thus there seemed to be a discrepancy between the in vitro data, which clearly showed that the tyrT promoter was sensitive to template topology, and the experiments in vivo, which suggested that there was no effect of supercoiling on the plasmidborne promoter. This led to the suggestion that DNA supercoiling may not be used by the cell to regulate synthesis from this tRNA promoter (Lamond, 1985). In the light of the studies of the leu-500 promoter, 5648

possible explanations for the differences between the results on the activity of the tyrT promoter in vivo and in vitro suggest themselves. Although the tyrT promoter exhibited a large activation in vitro when the supercoiling was changed from fully unconstrained (i.e. linear DNA) to native levels of negative supercoiling, the topological changes wrought in vivo by genetic means may have been rather less extreme. Indeed, since the plasmid used by Lamond and Travers (1983) did not carry tetA or equivalent genes to provide membrane anchorage, any supercoiling arising from transcription may have been relaxed by superhelical diffusion, even in a topA host. Thus the tyrT promoter may not have experienced a significant difference in topological environment between topA and top' hosts. We therefore decided to re-examine the tyrT promoter system in two respects. First, we have studied the in vitro transcription of the tyrT promoter as a function of negative supercoiling, using a range of plasmid topoisomer distributions from relaxed to above physiological levels of negative supercoiling. Second, we constructed new plasmids containing the tyrT promoter located upstream of the tetA gene. We show that in this situation, the activity of the tyrT promoter is elevated and that the increase is consistent with that observed in vitro over the expected range of DNA supercoiling.

Results Construction of plasmids containing the tyrT promoter adjacent to the tetA gene Previous studies have shown that the tetA gene is important in the generation of very highly supercoiled plasmids in bacterial strains lacking a functional topoisomerase I (Pruss, 1985; Lodge et al., 1989; Lynch and Wang, 1993; Bowater et al., 1994; Chen et al., 1994). Plasmids lacking a gene that encodes an inner membrane protein, such as TetA, do not exhibit this marked oversupercoiling. We

DNA supercoiling and the tyrT promoter

A -,

C

C.

-

-,

'-

-

..1 .

..

B

C r-.

1:7 -

7, C

z

ZI"

c

c

Y-

*. 2

S

..

Fig. 2. In vitro transcription of the tyrT promoter as a function of plasmid superhelix density. (A) Transcription in 100 and 200 mM KCI at 30°C. A series of topoisomer distributions of pTYRTtetAdi, of different mean superhelix density were subjected to in vitro transcription by Ecoli RNA polymerase for 15 min. RNA was isolated and used as a template for reverse transcription from a 5'-32P-labelled DNA primer. The extent and site of initiation of RNA synthesis were then determined by electrophoresis on a sequencing gel. The autoradiograph is shown and the cDNA corresponding to initiation of RNA synthesis at the tyrT promoter is indicated by the arrow (right). Dideoxy sequencing reactions were obtained using the template DNA with the same primer, electrophoresed in tracks 8-11. Transcription reactions were performed in the presence of either 100 mM (tracks 1-7) or 200 mM (tracks 12-18) KCI. Linear DNA was also transcribed using the same protocol (tracks 7 and 12). The circular plasmid samples had mean negative superhelix densities of 0, i.e. relaxed (tracks 6 and 13), 0.012 (tracks 5 and 14), 0.026 (tracks 4 and 15), 0.036 (tracks 3 and 16), 0.049 (tracks 2 and 17) and 0.10 (tracks I and 18). (B) Transcription in 0 mM KCI at 30°C. The same topoisomer distributions of pTYRTtetAd,V used in (A) were subjected to in vitro transcription in the absence of added KCI. The DNA had mean negative superhelix densities of 0 (track 1), 0.012 (track 2), 0.026 (track 3), 0.036 (track 4), 0.049 (track 5) and 0.10 (track 6). (C) Quantification of the dependence of in vitro transcription on superhelix density and KCI concentration. Extent of initiation of transcription at the tyrT promoter after a fixed period of transcription was quantified by phosphorimaging and is plotted as a function of mean negative superhelix density. Each set of data have been normalized to the maximum for those conditions. The absolute level of maximal transcription obtained at 50 mM KCI was -2-fold lower than that at the higher salt concentrations. The data for transcription in 50 mM (U), 100 mM (*) and 200 mM KCI (-) are shown. The data for 0 mM KCI are closely similar to those at 50 mM KCI and so were omitted for clarity.

have shown that the leu-SOO promoter responds to such oversupercoiling, allowing it to be activated on a plasmid (Chen et al., 1992). In these experiments we wished to exploit this large increase in in vivo levels of plasmid supercoiling to study the response of the tyrT promoter in

ciu.

75

0

0

vivo.

The tyrT-carrying plasmid (pTyr2) studied by Lamond and Travers (1983) did not contain a functional tetA gene. We therefore created two new plasmids containing the tyrT promoter inserted in either orientation upstream of the complete tetA gene. A DNA fragment containing the tyrT promoter was excised from pTyr2 and cloned into a modified form of the plasmid pAT1 53 (Twigg and Sherratt, 1980) (for complete details refer to Materials and methods). As found previously (Lamond and Travers, 1983), to enable the tyrT promoter to be cloned it was necessary to include a transcription terminator adjacent to the insertion site. Two forms of the resulting 4 kb plasmids were obtained, in which the tyrT promoter was divergent (pTYRTtetAdiV) or parallel (pTYRTtetApar) with respect to the tetA promoter (Figure 1). The level of transcription from the tyrT promoters was analysed by using reverse transcriptase to produce runoff cDNA from the transcribed RNA produced either in vitro or in vivo (Figueroa and Bossi, 1988; Chen et al., 1992). The sites of transcription initiation can be found by sequencing gel electrophoresis of the cDNA next to sequence markers and the relative amount of transcription initiation was quantified by measuring the amount of cDNA by phosphorimaging. We used a primer that hybridized to sequences between the promoter start site and the terminator, from +82 to +101 relative to the transcription start site.

50z Er

25-

I~ 0

0.025

0.05

0.075

0. 125

Fig. 3. In vitro transcription of the tyrT promoter in 150 mM KCI at 37°C as a function of plasmid superhelix density. Topoisomer distributions of pTYRTtetAdi. were subjected to in vitro transcription for 15 min and the extent of initiation at the tyrT promoter quantified as before by reverse transcription. Extent of initiation of transcription at the tvrT promoter is plotted as a function of mean negative superhelix density.

In vitro transcription of the tyrT promoter depends on negative supercoiling of the template and salt concentration Earlier experiments by Lamond (1985) had shown that a plasmid-bome tyrT promoter was dependent upon the topological state of the template when assayed in vitro at physiological salt concentrations, but this sensitivity was removed if the concentration of KCI was reduced below 50 mM. We prepared a range of topoisomer distributions of pTYRTtetAd1V with mean negative superhelix densities ranging from -cY = 0 (relaxed) to -a = 0.1 (highly negatively supercoiled) and examined transcriptional initi5649

R.P.Bowater, D.Chen and D.M.J.Lilley

ation at the tyrT promoter under the conditions employed by Lamond (1985). cDNA generated from RNA transcripts by reverse transcriptase was analysed on a sequencing gel (Figure 2A and B); the degree of initiation was quantified by phosphorimaging, shown graphically in Figure 2C. Closely similar results were obtained using the plasmid pTYRTtetApar At KCI concentrations of 50 mM and below, the level of transcription at 30°C was insensitive to the level of supercoiling. In the presence of 100 mM KCI or more, we observed a marked dependence of promoter activity upon the level of DNA supercoiling, the profile of which depended on the prevailing salt concentration. At all salt concentrations (except 200 mM KCI) we found that linear DNA was significantly poorer as a template than relaxed closed circular DNA, which we did not investigate further. We also performed in vitro transcription reactions under the more physiological conditions of 37°C and 150 mM KCI. Under these conditions we observed a dependence of tyrT transcription on supercoiling (Figure 3) that was similar to that seen at 30°C and 100 mM KCI. At very low levels of negative supercoiling there was little transcriptional initiation at the tyrT promoter. Initiation increased substantially at a superhelical density more negative than -a = 0.01, reaching a plateau at approximately -aY = 0.05. As the level of supercoiling was increased to very high values (- a = 0.1) there was a reduction in the amount of initiation of transcription. A similar phenomenon was observed at 30°C in 100 mM KCI, but at 200 mM KCI the amount of transcription was still increasing at the highest level of supercoiling used (Figure 2).

Initiation of transcription of the tyrT promoter in vivo is increased by the presence of an adjacent tetA gene To test the effect of in vivo supercoiling on the tyrT promoter, the plasmids were transformed into E.coli DM800, which contains a deletion of the topoisomerase I gene (Sternglanz et al., 1981; Di Nardo et al., 1982; Pruss et al., 1982). We have previously shown that this strain can be used in the analysis of the effects of transcription-induced supercoiling upon the leu-500 promoter (Chen et al., 1994). RNA was isolated from cells in exponential growth and radioactively labelled cDNA was prepared by reverse transcription and analysed by denaturing gel electrophoresis (Figueroa and Bossi, 1988; Chen et al., 1992). Since the sequence of the primer is not contained within chromosomal sequences, we would not observe RNA synthesis initiated at the chromosomal tyrT promoter and we detected zero background in cells lacking the plasmid. Cellular RNA was isolated from E.coli DM800 containing pTyr2, pTYRTtetAdiV or pTYRTtetApar For both plasmids containing the tetA gene there was an increase in transcription by -4-6-fold compared with pTyr2 (Figure 4A, compare tracks 6 with 5 and 11 with 12). From our past experience with the leu-500 promoter (Chen et al., 1992), this suggests that the tetA gene may be stimulating the function of the tyrT promoter by topological coupling. Modulation of tyrT promoter activity by tetA expression In our earlier studies we showed that the activation of a plasmid-borne leu-500 promoter was dependent on the 5650

transcription and translation of tetA and the membrane insertion of the TetA polypeptide product (Chen et al., 1992, 1993, 1994). Based on these studies, we constructed a number of plasmids to examine the effect of translation of tetA and the membrane insertion of its polypeptide upon transcription initiation at the tyrT promoter. In vivo transcription initiation at the tyrT promoter was monitored as before. The data are shown in Figure 4A and quantified in Figure 4B. In a plasmid carrying a 5' deletion of tetA such that TetA lacks amino acids 2-30 [pTYRTA(230)tetA], which is likely to cause disruption to the membrane anchorage events, transcription initiation at the tyrT promoter was reduced to approximately half of that from the plasmid containing the complete tetA gene (Figure 4A, track 13). The effect of tetA translation upon transcription initiated at the tyrT promoter was studied using tyrTcarrying plasmids into which translation terminators were placed at various positions along tetA, by insertion into the NheI, BamHI, Sall and NruI restriction sites. These reduce the length of the 394 amino acid TetA polypeptide to 50, 98, 192 and 298 amino acids respectively. Termination of translation within the 5'-half of the tetA gene reduced the amount of transcription initiation at the tyrT promoter to a level similar to that of pTyr2 (the plasmid lacking tetA) (see Figure 4A, compare tracks 4 to 1 and 14 to 17). As longer lengths of the polypeptide were synthesized there was an approximately linear increase (r = 0.98) in transcriptional initiation at the tyrT promoter (Figure 4C). Correlation of the activity of the tyrT promoter with the generation of a fraction of highly supercoiled plasmid DNA It has previously been shown that plasmid DNA isolated from bacteria lacking a functional topoisomerase I (Pruss, 1985; Lodge et al., 1989; Lynch and Wang, 1993; Bowater et al., 1994) may include a fraction that is highly negatively supercoiled. This requires that the plasmid carries a gene encoding a membrane-anchoring protein, such as tetA. In our earlier studies we observed a correlation between the highly supercoiled fraction and activation of the leu-500 promoter (Chen et al., 1994) and that modifications in tetA that reduced the activity of the leu-500 promoter produced corresponding reductions in the fraction of oversupercoiled plasmid. pTyr2 and pTYRTtetApar and its derivatives were isolated from E.coli DM800 and their level of supercoiling was analysed by electrophoresis in agarose gels containing chloroquine (Figure 5). While the plasmid pTyr2, which lacks the tetA gene, has a normal distribution of topoisomers, pTYRTtetApar is almost totally composed of a highly oversupercoiled fraction of topoisomers that are not resolved at this concentration of chloroquine. Indeed, the proportion of the extracted plasmid DNA in the oversupercoiled fraction is greater than that for any plasmid we have studied previously, including one containing the strong tac promoter (Chen et al., 1994). Thus the very strong tyrT promoter seems to be contributing to the plasmid supercoiling to a significant extent. All of the modifications to tetA decreased the extent of this highly supercoiled fraction. The plasmids with the greatest fraction of oversupercoiled DNA were those initiating the largest amount of transcription from the tyrT promoter.

A

P~

p.*V

DNA supercoiling and the tyrT promoter

.2.-

termst .

1

Jra

2

3

-

Z..

'.-

z

-.. :

;

cz,

J

~

.-_.A ;f

A

I/

5

..

er .:!

toi cept des

e'-a

Fig. 4. Stimulation of in vivo transcriptional initiation at the tyrT promoter carried on tetA-bearing plasmids in AtopA Ecoli. (A) Analysis of initiation at the tyrT promoter of pTyr2, pTYRTtetAdj, and pTYRTtetApar in E.coli DM800. RNA was isolated from cultures in exponential growth phase and the site and extent of initiation of transcription at the tVrT promoter analysed by reverse transcription as before. The autoradiograph is shown. The location of the cDNA corresponding to initiation at the tvrT promoter is indicated by the arrow (right). Dideoxy sequencing reactions were obtained using the template DNA with the same primer, electrophoresed in tracks 7-10. In vivo transcription initiation was compared in the plasmids pTyr2 (tracks 6 and 11), pTYRTtetAdi, and derivatives (tracks 1-5) and pTYRTtetApar and derivatives (tracks 12-17). The stimulation of activity of the tyrT promoter by the adjacent tetA gene is apparent by comparison of tracks 6 with 5 and I I with 12. The effects of modifying the function of the tetA gene on the activity of the tvrT promoter were analysed (see scheme below the autoradiograph). A derivative of pTYRTtetApar in which the tetA coding sequence was modified to remove amino acids 2-30 of TetA was studied (track 13); note that this reduces the level of stimulation of the activity of the tyrT promoter. The effects of premature termination of translation of TetA were also analysed, by insertion of translational terminators in the NheI (tracks 4 and 14), BamHI (tracks 3 and 15), Sall (tracks 2 and 16) and NruI sites (tracks I and 17) of the tetA gene. (B) Extent of stimulation of the tvrT promoter of pTYRTtetApar The gel electrophoretic data were quantified by phosphorimaging and are presented in the form of a histogram with all values normalized to 1.0 for pTyr2. For pTYRTtetApar there is an -6-fold stimulation of the rate of initiation of RNA synthesis at the tyrT promoter by the insertion of the complete tetA gene. Interference of translation or membrane insertion of TetA reduces the extent of stimulation. (C) Stimulation of the tvrT promoter of pTYRTtetApar is linearly dependent on the length of the TetA polypeptide translated. Plot of extent of RNA synthesis initiated at the tyrT promoter (relative to that in pTyr2) against the length of TetA. The line was fitted by linear regression.

This correlation strongly suggests that the observed increase in transcription from the tyrT promoter in the presence of tetA is linked to the topology of the plasmid and is probably due to an elevation in the unconstrained level of supercoiling in such plasmids arising from the activity of the tetA gene.

Discussion Stimulation of the tyrT promoter by negative supercoiling in

vivo

These experiments have demonstrated that the activity of a plasmid-bome tyrT promoter can be stimulated in vivo by effects that we believe are topological in character. This stimulation depends critically on the presence of the tetracycline resistance gene, tetA. We and others have previously shown that expression of tetA on a plasmid in

topA eubacterial host leads to the generation of a highly supercoiled fraction of plasmid DNA (Pruss, 1985; Lodge et al., 1989; Chen et al., 1992, 1994; Lynch and Wang, 1993) and increased unconstrained supercoiling (Bowater et al., 1994). We have shown that the leu-500 promoter located upstream of the tetA promoter may be activated under these circumstances (Chen et al., 1992, 1993, 1994). Comparative studies have established beyond reasonable doubt that leu-500 promoter activation is due to local changes in unconstrained template supercoiling. In the present work, we have exploited this system to compare the activity of the tvrT promoter on plasmids with and without a functional tetA gene. Our results indicate that the tvrT promoter may be stimulated by a factor of about 4-6 due to supercoiling in vivo. The critical role of the tetA gene in providing the supercoiling that mediates the stimulation of promoter activity was shown by investigaa

5651

R.P.Bowater, D.Chen and D.M.J.Lilley

.1tjrr1 .,

5

co

a a.

4

0)

3

0

0.2

0.4

0.6

0.8

1

Fraction oversupercoiled DNA

Fig. 5. A highly oversupercoiled fraction of plasmid DNA is associated with stimulation of activity of the tyrT promoter. pTyr2 and pTYRTtetApar and its derivatives were extracted from Ecoli DM800 (AtopA) cells in exponential growth. The topoisomer populations of the extracted DNA were analysed by electrophoresis in a 1% agarose gel in the presence of 1.8 tg/ml chloroquine. pTyr2 (track 1) (which lacks the tetA gene) displays a typical Gaussian-shaped topoisomer profile about a mean negative superhelix density of -0.05. In contrast, the DNA of pTYRTtetApar (track 2) is almost totally comprised of a very highly oversupercoiled form that migrates ahead of the normal form and is not resolved at this concentration of chloroquine. Interference with the function of the tetA gene, either by premature termination of translation (tracks 4-7) or N-terminal deletion (track 3), reduced the proportion of the oversupercoiled fraction.

ting plasmids in which the translation of tetA or the membrane insertion of the TetA polypeptide was impaired. Two effects of these modifications were observed; there was a reduction in the fraction of oversupercoiled plasmid DNA extracted from the cell and a reduced stimulation of tyrT promoter activity. These two sets of data may be correlated (Figure 6) with a linear correlation coefficient (r) of 0.96, increasing our confidence that the stimulation of the tyrT promoter is topological in origin. Our results do not prove that template topology will normally be involved in regulation of the tyrT promoter. It is clear from this and many other studies that the tyrT promoter has a very complex system of regulation and it is perhaps to be expected that no single factor will have total control. However, recent studies on the chromosomal tyrT promoter suggest that it is affected by the superhelical density within the cell (Free and Dorman, 1994), in good agreement with the results reported here. Thus, supercoiling may be one of many control mechanisms available to the cell for the control of transcription initiated at the tyrT promoter.

Comparison between stimulation of tyrT promoter activity in vitro and in vivo In an earlier work, Lamond (1985) has shown that while the activity of the tyrT promoter was stimulated by a factor of >100 between linear and native supercoiled DNA templates in vitro, no differences could be observed in vivo as a function of topA background. We now show that the tyrT promoter does respond to changes in plasmid supercoiling in vivo, provided that they are induced by transcription of the tetA gene in a topA host. In the absence of a functional tetA gene, no change in tyrT promoter 5652

Fig. 6. Correlation of the stimulation of the in vivo activity of the tvrT promoter and the proportion of oversupercoiled plasmid extracted from the cell. Plot of the increase in promoter activity compared with that in pTyr2 against the amount of oversupercoiled pTYRTtetApar as a fraction of total DNA. The line was fitted by linear regression.

activity was observed, in agreement with Lamond (1985). Truncation of TetA translation resulted in intermediate levels of stimulation of the tyrT promoter, as discussed above. The magnitude of the promoter stimulation was much smaller than that observed in vitro by Lamond (1985), but is consistent with our own in vitro studies. The basal level of effective negative supercoiling in eubacteria has been estimated by a number of techniques as-a- 0.022-0.025 (Greaves et al., 1985; Lilley, 1986; Bliska and Cozzarelli, 1987; Zacharias et al., 1988) and we have estimated the level of unconstrained superhelix density in tetA-carrying plasmids in topA E.coli as -a 0.052 (Bowater et al., 1994). Under conditions approximating most closely to physiological (37°C, 150 mM KCI), in vitro transcription initiated at the tyrT promoter increased by a factor of three over this range. This is somewhat smaller than the observed stimulation in vivo (a factor of 4-6), which could be a result either of overestimating the basal superhelix density of the plasmid in vivo or of incomplete equivalence between the conditions of transcription in vitro and in vivo. Nevertheless, in view of the uncertainties involved, the agreement between in vitro and in vivo transcriptional initiation is good. Much larger stimulation is possible in vitro comparing native supercoiled (-iy 0.06) and relaxed (-i = 0) plasmids (and even greater if linear and supercoiled DNA are compared), but this extreme range is not covered -

in vivo. The results on the modulation of the activity of the tyrT promoter in vitro and in vivo can be viewed from another perspective. The tyrT promoter might be regarded as a reporter sequence for the level of unconstrained negative supercoiling in the plasmid DNA in vivo. The reasonable agreement (within a factor of two) between the stimulation of the tyrT promoter in vitro and that expected in vivo on the basis of our current estimates of the level of unconstrained supercoiling in cellular DNA provides a confirmation of the general magnitude of superhelix density changes suffered by plasmid DNA under different transcriptional conditions.

Topological coupling between promoters These experiments provide a second example of topological coupling between promoters, where the activity of

DNA supercoiling and the tyrT promoter

one promoter is affected by changes in local unconstrained template supercoiling arising from transcription initiated at another nearby promoter. In the case of the leu-500 promoter, this was activated by the tetA gene in cis in a topA host, from being virtually inactive in top' strains (Chen et al., 1992). In contrast, the tyrT promoter was stimulated by a factor of 4-6 from an already high level of transcriptional initiation. This high basal rate of initiation of RNA synthesis at the tyrT promoter may itself contribute to the prevailing level of DNA supercoiling in the local topological domain, as discussed below. Thus the tyrT promoter may influence its own activity by topological coupling effects, given the correct circumstances. Activation of the leu-500 promoter requires that the host is topA, such that negative supercoiling induced by transcription is poorly relaxed by enzyme action. It has proved more difficult to demonstrate an unequivocal role for topA in the stimulation of the activity of the tyrT promoter in vivo. One factor which had a large effect upon the amount of transcription from tvrT was the rate of cell growth, as would be expected from earlier studies on this promoter (Gourse et al., 1986; Travers et al., 1986), and this varied widely between different strains. It was therefore impossible to compare the function of the tyrT promoter in topA and topA+ hosts in a meaningful way.

Factors affecting transcription-induced plasmid supercoiling Induction of DNA supercoiling by transcription according to the twin supercoiled-domain model of Liu and Wang (1987) has been experimentally demonstrated in a number of systems (Lockshon and Morris, 1983; Pruss and Drlica, 1986; Liu and Wang, 1987; Wu et al., 1988; Tsao et al., 1989; Droge and Nordheim, 1991; Chen et al., 1992, 1994; Dayn et al., 1992; Rahmouni and Wells, 1992; Bowater et al., 1994). We have found that our studies of the leu-SOO promoter can best be understood in terms of the creation of a topological domain upstream of the tetA gene in which promoter strength can be modulated by transcriptional activity (Chen et al., 1993). A number of promoters can influence the steady-state level of DNA supercoiling, provided that they are located within this domain. A strong promoter, such as the tac promoter, increases negative supercoiling, while the effects of transcription-induced supercoiling are diluted by increasing the size of the domain. The single most important element in the creation of the domain appears to be provided by the tetA gene; even in the presence of the tac promoter, oversupercoiling disappears in the absence of tetA transcription. In the present experiments we have introduced an extremely strong promoter, tyrT, into this topological domain, with a clear effect on plasmid supercoiling. The topoisomer profiles of pTYRTtetApar (Figure 5) and pTYRTtetAdiV (data not shown) extracted from Ecoli DM800 (AtopA) demonstrate that effectively all the plasmid DNA has been converted into the oversupercoiled fraction. This is in contrast to all previous plasmids that we have analysed, where we obtained bimodal distributions of extracted plasmids containing a fraction of DNA that is normally supercoiled (-ay 0.05) and a fraction that is oversupercoiled (-a 0.09) (Chen et al., 1992). With =

=

the tac promoter inserted in the domain, the proportion of the oversupercoiled fraction was increased (Chen et al., 1994) and this effect has become exaggerated to the limit with the even stronger tvrT promoter present. The marked effect of the tyrT promoter on plasmid supercoiling raises some interesting points. The high level of oversupercoiling is independent of the orientation of the tyrT promoter; both pTYRTtetApar and pTYRTtetAdiv exhibit extreme oversupercoiling. In the former plasmid the tyrT promoter is not divergent with respect to tetA transcription, but transcribes the same strand. Yet the effect cannot arise from boosted tetA transcription, because RNA polymerase initiated at the tyrT promoter is terminated at the strong trpA transcription terminator. Therefore, it seems that it is sufficient to have very active transcription in either direction within the domain, with the imbalance of relaxation in the topA gyr+ host leading to net negative supercoiling of the template. The oversupercoiled fraction of these plasmids is not observed in topA+ hosts (data not shown). A point of membrane attachment (provided by coupled transcription, translation and membrane insertion of tetA) remains the sine qua non for the creation of the topological domain, presumably because it provides a topological barrier against superhelical diffusion. However, other transcription units that play an important role in the supercoiling do not appear to require this kind of anchoring effect, as we have discussed previously for the b/a gene (Chen et al., 1993). This is seen most clearly here for the tvrT promoter, where there can be no question of translation into polypeptide. Indeed, since the maximum length of the transcript is only 170 nucleotides, even the role of RNA bulk in hindering rotation of the transcribing complex is questionable in this case. Activation of the leu-500 promoter in topA S.typhimurium has recently been achieved with plasmids lacking the tetA gene (Tan et al., 1994), suggesting that other forms of topological barrier could exist. It has been suggested that R-loop formation could be an important element in induction of supercoiling during transcriptional elongation (Drolet et al., 1994). In conclusion The influence of transcription-induced supercoiling on the activity of the tyrT promoter in topA E.coli is clear and demonstrates that the phenomenon of topological promoter coupling first observed at the leu-500 promoter can be more general and is not limited to mutant promoters. Moreover, the use of tetA-carrying plasmids and their variants in topA E.coli provides a well-controlled way in which to manipulate the local superhelical environment in vivo and thus to test the response of other promoters to template topology.

Materials and methods Bacterial strains and growth conditions During this study the following E.coli strains were used: HB1O1 (topA+), DM800 (AtopA) and SD 108 (topA+). DM800, a A(topA-cysB)204 acrAJ3 gyrB225 derivative of E.coli K12 strain W3 110, and SD108, a DM800 AtrpE63topA+cysB+pyrF287 phage P1-transductant (DiNardo et al., 1982; Pruss et al., 1982), were obtained from K.Drlica. Bacteria were cultured at 37°C with aeration in LB medium or grown on 1.2% LB agar plates. Media were supplemented with antibiotics as required: ampicillin was used at 50 tg/ml. Plasmids were propagated in Ecoli

5653

R.P.Bowater, D.Chen and D.M.J.Lilley HB101, extracted and purified by the SDS-alkali method (Bimboim and Doly, 1979) and transformed into Ecoli DM800 and SDl08 by the calcium chloride procedure (Cohen et al., 1972). DNA used for in vitro transcription experiments was purified by two rounds of CsCI-ethidium bromide isopycnic centrifugation.

sodium acetate (pH 5.2), 2% SDS, 0.3 M sucrose. The sample was extracted twice with phenol -chloroform and nucleic acids were precipitated with ethanol. Redissolved RNA was incubated with reverse transcriptase as described below, to identify the sites and extent of transcription initiation.

Oligonucleotide synthesis

In vitro transcription reactions Purified Ecoli RNA polymerase (Promega) was used for in vitro transcription reactions on plasmids of different mean superhelical densities. The RNA polymerase was 395% pure and was ¢50% holoenzyme. DNA (0.2 ,ug) was transcribed by 2 units (1.3 ,ug protein) of RNA polymerase in 50 pl of 40 mM Tris-HCI (pH 8.0), 10 mM MgC92, 0.05% bovine serum albumin, 0.1 mM EDTA, 0.1 mM DTT, 0.15 mM ribonucleotide triphosphates and 5 units human placental ribonuclease inhibitor (BRL) for 15 min at the specified temperature. KCI was added to the reactions at the indicated concentrations. Reactions were terminated by extraction of the proteins with phenol-chloroform and nucleic acids were precipitated with ethanol. Redissolved RNA was incubated with reverse transcriptase as described below, to identify the sites and extent of transcription initiation.

Oligonucleotides were synthesized using ,-cyanoethylphosphoramidite chemistry (Beaucage and Caruthers, 1981; Sinha et al., 1984) implemented on an Applied Biosystems 394 DNA/RNA synthesizer. After complete deprotection, oligonucleotides used in cloning were purified by denaturing gel electrophoresis in polyacrylamide containing 7 M urea, followed by electroelution. Oligonucleotides were 5'-phosphorylated using T4 polynucleotide kinase. Primers were 5 _32P radioactively labelled using [y-32P]ATP and T4 polynucleotide kinase (Maxam and Gilbert, 1980).

Plasmid construction We wished to construct a plasmid containing the tyrT promoter upstream of the tetA gene. The tyrT promoter initiates RNA transcription at a high rate and is thus very difficult to clone on a plasmid without the presence of a strong transcriptional terminator. The plasmid was therefore constructed using two steps. First, the host vector was derived from pAT153 (Twigg and Sherratt, 1980) by cloning the required restriction sites for insertion of the tyrT promoter upstream of the trpA transcription terminator (Christie et al., 1981). The oligonucleotides 5'-AATTGAAAAAAAAGCCCGCTCATTAGGCGGGCTATGCATTCGTCAATGTCCTTAAG-3' (NsiI and AfllI restriction sites underlined) and 5'-AATTCTTAAGGACATTGACGAATGCATAGCCCGCCTAATGAGCGGGC1-1-TTTTTC-3' were hybridized by slow cooling from 70°C and ligated into the EcoRI site of pAT153. Clones were identified which had plasmids containing the oligonucleotides in both of the possible orientations. The 334 bp NsiI-AfIII fragment containing the tyrT promoter was excised from pTyr2 (Lamond and Travers, 1983) and ligated into the above plasmids digested by these restriction enzymes. The two plasmids thus created carry the tyrT promoter in each orientation, either divergent to tetA (pTYRTtetAdiv) or transcribing in the same direction as tetA (pTYRT-

tetApar).

A number of variant plasmids were created to examine the effect of expression of the tetA gene on transcription initiated at the tyrT promoter. Plasmids containing translation termination codons inserted at a number of sites within tetA were created by exchanging the corresponding HindIll-AvaI fragments from similar plasmids used in our studies on the leu-500 promoter (Chen et al., 1992). Plasmids pTYRTtetA ter NheI, pTYRTtetA ter BamHI, pTYRTtetA ter Sall and pTYRTtetA ter NruI contain transcription terminators inserted into the NheI, BamHI, Sall and NruI restriction sites respectively. To create a plasmid with disrupted membrane insertion of the TetA polypeptide, we cloned the HindIIl -AvaI fragment from pLEU500A(2-30)tetA (Chen et al., 1993) into pTYRTtetApa; this plasmid (pTYRTA(2-30)tetA) contains an in-frame deletion in tetA such that the modified TetA polypeptide lacks amino acids 2-30.

Preparation of plasmid topoisomer distributions of different mean superhelical densities Topoisomer distributions were obtained by incubating native supercoiled plasmid DNA with wheat germ topoisomerase I (Promega) in the presence of various concentrations of ethidium bromide (Bowater et al., 1991). After phenol -chloroform extraction of the protein, the DNA was recovered by ethanol precipitation, dissolved in 10 mM Tris-HCI (pH 7.5), 0.1 mM EDTA and stored at -20°C. The superhelical density (a) of each distribution was determined by gel electrophoretic analysis in a number of agarose gels containing different concentrations of chloroquine. By counting up to the linking difference (ALk) of the topoisomer with the highest intensity in each of the Gaussian distributions, mean superhelical densities were estimated. It was usual to use eight concentrations of ethidium bromide to cover a range of mean superhelical density from -a = 0 to 0.1 (Bowater et al., 1991). We calculate superhelical density from: a = 10.5 ALk/N where N is the size of the plasmid (bp).

Extraction of cellular RNA RNA was prepared using a method described previously (Chen et al., 1992). In brief, 200 ,ul cultures in exponential growth phase were heated in a boiling water bath for 1 min with an equal volume of 20 mM

5654

Reverse transcription analysis of transcription initiation After addition of 0.2 pmol 5'-32P-labelled DNA primer, the sample was heated to 90°C in 4.5 gl of 50 mM Tris-HCI (pH 8.0), 50 mM KCI and rapidly cooled. Five units of human placental ribonuclease inhibitor were added and the solution was incubated at 43°C for 20 min before addition of 12 gl of 70 mM Tris-HCI (pH 8.0), 70 mM KC1, 15 mM MgCl2, 15 mM dithiothreitol and 1.3 mM deoxyribonucleotide triphosphates, containing 50 units of MMLV reverse transcriptase (Superscript plus; BRL) and incubation at 42°C for 2 h. Transcripts were analysed by electrophoresis in 6-10% polyacrylamide gels in 90 mM Tris-borate (pH 8.3), 10 mM EDTA (TBE buffer) containing 7 M urea, next to sequence markers generated by dideoxy sequence reactions using the same primer (Sanger et al., 1977). The gels were dried and radioactive bands were observed by autoradiography at -70°C with intensifier screens or with storage phosphor screens and a 400S phosphorimager (Molecular Dynamics). Quantitation was performed upon the phos-

phorimage.

Analysis of linking number of extracted plasmid DNA Cells were grown in 30 ml of LB plus antibiotics to mid-exponential phase and the plasmid DNA was extracted by the SDS-alkali method (Birnboim and Doly, 1979). To remove cellular RNA in the sample, 10 jl of DNAase-free RNaseA (10 mg/ml) was added and incubated at 37°C for 15 min. After two phenol-chloroform extractions, the samples were centrifuged through Sephadex G-50 to remove salts and the plasmid DNA was precipitated with ethanol. The DNA was analysed by electrophoresis in 1% agarose in TBE containing different concentrations of chloroquine. After extensive washing in water, the gels were stained in 1 gg/ml ethidium bromide and photographed under UV illumination with red and green filters to remove background fluorescence. Photographic negatives were subjected to laser densitometry (Molecular Dynamics) for quantitation and presentation of the digitized image.

Acknowledgements We thank Andrew Travers for discussions and the provision of the plasmid pTyr2, Karl Drlica for provision of bacterial strains, Andrew Free and Charles Dorman for discussions and the MRC and CRC for financial support.

References Amouyal,M. and Buc,H. (1987) J. Mol. Biol., 195, 795-808. Beaucage,S.L. and Caruthers,M.H. (1981) Tetrahedron Lett., 22, 18591862. Berman,M.L. and Landy,A. (1979) Proc. Natl Acad. Sci. USA, 76, 4303-4307. Bertrand-Burggraf,E., Schnarr,M., Lefevre,J.F. and Daune,M. (1984) Nucleic Acids Res., 12, 7741-7752. Bimboim,H.C. and Doly,J. (1979) Nucleic Acids Res., 7, 15 13-1523. Bliska,J.B. and Cozzarelli,N.R. (1987) J. Mol. Biol., 194, 205-218. Borowiec,J.A. and Gralla,J.D. (1987) J. Mol. Biol., 184, 587-598. Botchan,P., Wang,J.C. and Echols,H. (1973) Proc. Natl Acad. Sci. USA, 70, 3077-3081.

DNA supercoiling and the tyrT promoter Bowater,R., Aboul-ela,F. and Lilley,D.M.J. (1991) Methods Enzvnmol., 212, 105-120. Bowater,R.P., Chen,D. and Lilley,D.M.J. (1994) Biochemistry, 33, 9266-9275. Chen,D., Bowater,R., Dorman,C. and Lilley,D.M.J. (1992) Proc. Natl Acad. Sci. USA, 89, 8784-8788. Chen,D., Bowater,R. and Lilley,D.M.J. (1993) Biochemiiistrv, 32, 13162-13170. Chen,D., Bowater,R. and Lilley,D.M.J. (1994) J. Bacteriol., 176, 3757-3764. Christie,G.E., Famham,P.J. and Platt,T. (1981) Proc. Natl Acad. Aci. USA, 78, 4180-4184. Cohen,S., Chang,A. and Hsu,L. (1972) Proc. Natl Ac-ad. Sci. USA, 69, 2110-2114. Dayn,A., Malkhosyan,S. and Mirkin,S.M. (1992) Nucleic Acids Res., 20, 5991-5997. Di Nardo,S., Voelkel,K.A., Sternglanz,R., Reynolds,A.E. and Wright,A. (1982) Cell, 31, 43-51. Droge,P. and Nordheim,A. (1991) Nucleic Acids Res., 19, 2941-2946. Drolet,M., Bi,X. and Liu,L.F. (1994) J. Biol. Chem., 269, 2068-2074. Dubnau,E. and Margolin,P. (1972) Mol. Gen. Genet., 117, 91-112. Engle,E.C., Manes,S.H. and Drlica,K. (1982) J. Bacteriol., 149, 92-98. Figueroa,N. and Bossi,L. (1988) Proc. Natl Acad. Sci. USA, 85, 9416942. Free,A. and Dorman,C.J. (1994) Mol. Microbiol., in press. Gemmill,R.M., Tripp,M., Friedman,S.B. and Calvo,J.M. (1984) J. Bacteriol., 158, 948-953. Gourse,R.L., de Boer,H.A. and Nomura,M. (1986) Cell, 44, 197-205. Greaves,D.R., Patient,R.K. and Lilley,D.M.J. (1985) J. Mol. Biol., 185, 461-478. Higgins,C.F., Dorman,C.J., Stirling,D.A., Waddell,L., Booth,I.R., May,G. and Bremer,A. (1988) Cell, 52, 569-584. Javor,G.T. (1974) J. Bacteriol., 120, 282-286. Kano,Y., Miyashita,T., Nakamura,H., Kuroki,K., Nagata,A. and Imamoto,F. (1981) Gene, 13, 173-184. Lamond,A.I. (1985) EMBO J., 4, 501-507. Lamond,A.I. and Travers,A.A. (1983) Nature, 305, 248-250. Lamond,A.I. and Travers,A.A. (1985) Cell, 40, 319-326. Lilley,D.M.J. (1986) Nature, 320, 14-16. Lilley,D.M.J. and Higgins,C.F. (1991) Mol. Microbiol., 5, 779-783. Liu,L.F. and Wang,J.C. (1987) Proc. Natl Acad. Sci. USA, 84,7024-7027. Lockshon,D. and Morris,D.R. (1983) Nucleic Acids Res., 11, 2999-3017. Lodge,J.K., Kazik,T. and Berg,D.E. (1989) J. Bacteriol., 171, 2181-2187. Lynch,A.S. and Wang,J.C. (1993) J. Bacteriol., 175, 1645-1655. MaxamA.M. and Gilbert,W. (1980). Methods Enzvmol., 65, 499-560. Menzel,R. and Gellert,M. (1983) Cell, 34, 105-113. Mukai,F.H. and Margolin,P. (1963) Proc. Natl Acad. Sci. USA, 50,

Yang,H.-L., Heller,K., Gellert,M. and Zubay,G. (1979) Proc. Natl Acad. Sci. USA, 76, 3304-3308. Zacharias,W., Jaworski,A., LarsonJ.E. and Wells,R.D. (1988) Proc. Natl Acad. Sci. USA, 85, 7069-7073. Received on August 12, 1994; revised on August 30, 1994

140-148. Pruss,G. (1985) J. Mol. Biol., 185, 51-63. Pruss,G.J. and Drlica,K. (1986) Proc. Natl Acad. Sci. USA, 83, 89528956. Pruss,G.J. and Drlica,K. (1989) Cell, 56, 521-523. Pruss,G.J., Manes,S.H. and Drlica,K. (1982) Cell, 31, 35-42. Rahmouni,A.R. and Wells,R.D. (1992) J. Mol. Biol., 223, 131-144. Richardson,J.P. (1974) Biochemistr, 13, 3164-3169. Sanger,F., Nicklen,S. and Coulson,A.R. (1977) Proc. Natl Acad. Sci. USA, 74, 5463-5467. Sanzey,B. (1979) J. Bacteriol., 138, 40-47. Sinha,N.D., Biernat,J., McManus,J. and Koster,H. (1984) Nucleic Acids Res., 12, 4539-4557. Smith,G.R. (1981) Cell, 24, 599-600.

Sternglanz,R., DiNardo,S., Voelkel,K.A., Nishimura,Y., Hirota,Y, Becherer,K., Zumstein,L. and Wang,J.C. (1981) Proc. Natl Acad. Sci. USA, 78, 2747-2751. Tan,J., Shu,L. and Wu,H.-Y. (1994) J. Bacteriol., 176, 1077-1086. Travers,A.A., LamondAl. and Mace,H.A.F. (1982) Nucleic Acids Res., 10, 5043-5057. Travers,A.A., Lamond,A.I., Mace,H.A.F. and Berman,M.L. (1983) Cell, 35, 265-273. Travers,A.A., Lamond,A.I. and Weeks,J.R. (1986) J. Mol. Biol., 189, 251-255. Tsao,Y.-P.,Wu, H.-Y. and Liu,L.F. (1989) Cell, 56, 111-118. Tse-Dinh,Y. (1985) Nucleic Acids Res., 13, 4751-4763. Tse-Dinh,Y.-C. and Beran,R.K. (1988) J. Mol. Biol., 202, 735-742. Twigg,A.J. and Sherratt,D. (1980) Naiture, 283, 216-218. Wu,H.-Y., Shyy,S., Wang,J.C. and Liu,L.F. (1988) Cell, 53, 433-440.

5655