Transient and dynamic DNA supercoiling potently stimulates the leu-500 promoter in Escherichia coli

The inactive prokaryotic leu-500 promoter (Pleu-500) contains a single A-to-G point mutation in the −10 region of the leucine operon promoter, which causes leucine auxotrophy. This promoter can be activated by (−) DNA supercoiling in Escherichia coli topA strains. However, whether this activation arises from global, permanent, or transient, dynamic supercoiling is still not fully understood. In this article, using a newly established in vivo system carrying a pair of divergently coupled promoters, i.e. an IPTG-inducible promoter and Pleu-500 that control the expression of lacZ and luc (the firefly luciferase gene), respectively, we demonstrate that transient, dynamic (−) DNA supercoiling provided by divergent transcription in both wild-type and topA strains can potently activate Pleu-500. We found that this activation depended on the promoter strength and the length of RNA transcripts, which are functional characteristics of transcription-coupled DNA supercoiling (TCDS) precisely predicted by the twin-supercoiled domain model of transcription in which a (+) supercoiled domain is produced ahead of the RNA polymerase and a (−) supercoiled domain behind it. We also demonstrate that TCDS can be generated on topologically open DNA molecules, i.e. linear DNA molecules, in Escherichia coli, suggesting that topological boundaries or barriers are not required for the production of TCDS in vivo. This work demonstrates that transient, dynamic TCDS by RNA polymerases is a major chromosome remodeling force in E. coli and greatly influences the nearby, coupled promoters/transcription.

DNA supercoiling is a fundamental property of chromosomal DNA in living cells and greatly affects the efficiency of many essential DNA metabolisms, such as transcription, DNA replication, and recombination (1,2). In Escherichia coli and many other organisms, DNA is typically (Ϫ) supercoiled (3). DNA supercoiling homeostasis of E. coli cells is primarily set by counter actions of DNA topoisomerase I and gyrase (4,5). Inhibition of either enzyme causes the production of hypernegatively or (ϩ) supercoiled DNA in vivo (6,7). Recent genomic and bioinformatics studies also demonstrated that DNA supercoiling substantially affects expression of many genes in E. coli (8 -11). For instance, a study by Blot et al. (8) showed that half of all genes in E. coli are sensitive to DNA supercoiling (9), suggesting that DNA supercoiling is a global transcriptional regulator for bacterial growth (10,12).
Transcription can induce localized DNA supercoiling in vitro (13)(14)(15)(16)(17) and in vivo (6, 7, 18 -22). Liu and Wang (23) formulated an elegant twin supercoiled domain model of transcription to explain these effects. They pointed out that a transcribing RNA polymerase becomes increasingly more difficult to rotate around the axis of the DNA double helix as the size of the growing RNA transcript increases. At a critical point, energetically, it is more feasible for the DNA molecule to rotate around its own helical axis to produce a (ϩ) supercoiled domain ahead of the RNA polymerase and a (Ϫ) supercoiled domain behind it. Although these two transient supercoiled domains may be relaxed by DNA topoisomerases or cancel each other by diffusion (13,14,16,24), they should have great influence on nearby DNA transactions, such as transcription. Indeed, several cases (25)(26)(27)(28)(29) demonstrated that transcription coupled DNA supercoiling (TCDS) 2 plays important roles in divergently coupled promoters/transcriptions in which two neighboring promoters initiate transcription in opposite directions. For example, in the ilvYC operon of E. coli, the ilvY promoter is divergently coupled to the ilvC promoter (28). It was shown that transcriptional activities of the ilvY and ilvC promoters are dependent on the localized superhelical density around the promoter region (28,30). Another well characterized example is the activation of Salmonella typhimurium leu-500 promoter (P leu-500 ) by TCDS in topA strains (25)(26)(27)(31)(32)(33)(34)(35)(36)(37)(38). P leu-500 contains a single A-to-G point mutation in the Ϫ10 region of the promoter controlling the leu operon, which results in leucine auxotrophy (39,40). The AT-to-GC mutation is expected to increase the energy barrier for the formation of a functional transcription open complex and, as a result, requires (Ϫ) supercoiling for its activation (41). Although it has been demonstrated that TCDS around the promoter region plays an essential role in the activation of P leu-500 (25)(26)(27)(31)(32)(33)(34)(35)(36), a detailed mechanism explaining how localized TCDS regulates gene expression is still lacking. This problem may be caused by This work was supported in whole or in part by Grant 1R15GM109254-01A1 from the National Institutes of Health (to F. L.). The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This article contains supplemental Tables S1-S3 and Figs. S1-S5. 1 To whom correspondence should be addressed: Dept cro ARTICLE the fact that previous studies relied on the use of small circular plasmid templates in which transient supercoiled domains produced by transcription can diffuse along the plasmid DNA and therefore cancel each other (16). Additionally, transcription drives a significant amount of plasmid DNA templates into a hypernegatively superhelical status in topA strains (7,21). In this case, it is difficult to determine whether the transient or global supercoiling contributes to the activation. A "better" methodology is needed to directly examine how the transient and dynamic supercoiling induced by transcription regulates the coupled gene expression.
Furthermore, whether fixed topological boundaries or barriers are required for the production of transcription-driven supercoiled domains in vivo is still an open question. Because the twin-domain model of transcription in the original form suggested that fixed topological boundaries or barriers are required to generate supercoiled domains around the RNA polymerase (23), a transcribing RNA polymerase along a topologically open DNA molecule, such as a linear DNA template, should not generate sufficient friction force to supercoil the DNA molecule. However, a more recent theoretical study by Nelson (42) suggested that small natural bends in the DNA molecule could significantly increase transcription-induced friction torsional stress and therefore allow the RNA polymerase to supercoil a linear unanchored DNA molecule. Indeed, Levens and co-workers (43) demonstrated that transcription of linear DNA molecules containing a single active promoter was able to supercoil upstream region of the promoter in vitro. Nevertheless, whether transcription is capable of supercoiling upstream of a promoter on a linear DNA template in vivo has not been examined yet.
In this article, utilizing different DNA molecules, i.e. circular plasmids, linear plasmids, and the E. coli chromosome, we demonstrate that transient and dynamic (Ϫ) DNA supercoiling provided by transcription from an IPTG-inducible promoter is able to potently activate the divergently coupled supercoilingsensitive promoter P leu-500 in E. coli cells. More importantly, our results showed that TCDS could be generated on topologically open DNA molecules, such as linear DNA templates, to activate P leu-500 promoter in E. coli. These results suggest that topological boundaries or barriers are not required for the production of supercoiled domains around the RNA polymerase in vivo. We also found that activation of P leu-500 by TCDS is dependent on the promoter strength and the length of RNA transcripts, functional properties of TCDS precisely predicted by the twin-domain mechanism (23).

Strong activation of P leu-500 by TCDS at circular and linear plasmid levels
Recently, we have established an in vivo system to study the activation of supercoiling-sensitive P leu-500 by TCDS from an IPTG-inducible, divergent promoter-transcription unit. The system consists of E. coli topA strain VS111(DE3)⌬lacZ or wildtype strain MG1655(DE3)⌬lacZ and a circular plasmid or a linear plasmid. The DE3 on the chromosome carries a lacI gene under the control of the lacI q promoter to express and provide sufficient amounts of LacI for the IPTG-inducible promoter. For plasmid DNA templates, P leu-500 is divergently coupled to the strong IPTG-inducible P T7A1/O4 (Fig. 1). The distance between these two promoters is 81 bp (the distance was calculated between the Ϫ35 regions of two promoters; Fig. 1B). Additionally, two sets of four Rho-independent, rrnB T1 transcription terminators are used to stop transcription from P T7A1/O4 and P leu-500 , respectively (Figs. 2A and 3A). In this way, transcription is restricted to selected regions of the plasmids (21). We also cloned a luc gene (to express firefly luciferase) under the control of P leu-500 and a lacZ gene under the control of P T7A1/O4 . Because lacZ deletion mutants were used, we were able to determine transcription levels from P T7A1/O4 by measuring ␤-galactosidase activities. These two plasmids were transformed into VS111(DE3)⌬lacZ or MG1655(DE3)⌬lacZ. After addition of IPTG to the cell culture in early log phase, the activation of P leu-500 was monitored by measuring luciferase activities and also by determining transcription levels using RT-PCR assays. The results in Figs. 2 and 3 unambiguously demonstrate that TCDS provided by the divergent transcription was able to potently activate the supercoiling-sensitive P leu-500 . For instance, TCDS by E. coli RNA polymerase on the circular plasmid pZXD133 was capable of activating P leu-500 ϳ7.3and 2.5fold in VS111(DE3)⌬lacZ and MG1655(DE3)⌬lacZ, respectively (Fig. 2C). TCDS on the linear plasmid pZXD143 was also able to activate P leu-500 ϳ9.7and 3.5-fold in VS111(DE3)⌬lacZ and MG1655(DE3)⌬lacZ, respectively ( Fig. 3C; only one copy of linear plasmid exists in each cell according to our estimate). Nevertheless, a careful inspection of these results demonstrates that the activation level of P leu-500 is only correlated with the transcription level from P T7A1/O4 , regardless of the genetic backgrounds of host strains (Figs. 2 and 3). In other words, a higher expression of ␤-galactosidase always corresponds to a higher expression of firefly luciferase (Figs. 2 and 3). The difference of activation fold between MG1655 and VS111 partially stems from the basal expression of lacZ (␤-galactosidase) (44). In the absence of IPTG, the basal expression of ␤-galactosidase is higher for the wild-type strain (MG1655) than that for a topA mutant strain (VS111) (44). Design of divergently coupled transcription to study transcription activation of P leu-500 by TCDS in E. coli. A, divergently coupled promoters P T7A1/O4 and P leu-500 were used to control the expression of ␤-galactosidase (lacZ) and firefly luciferase (luc), respectively. The red oval represents the lac O1 operator. B, the DNA sequence of the divergently coupled promoters, P T7A1/O4 and P leu-500 . Boxed are Ϫ10 and Ϫ35 regions of P leu-500 and P T7A1/O4 . The arrowhead indicates the mutated G in the Ϫ10 region of P leu-500 .

Stimulation of the leu-500 promoter by transient DNA supercoiling
We also found another interesting phenomenon. The expression level of both ␤-galactosidase and firefly luciferase is higher in MG1655 than that in VS111 for circular plasmids (Fig.  2). However, the expression pattern is reversed for linear plasmids, i.e. higher in VS111 (because of the lower copy number of linear plasmids in E. coli (1 copy/cell for linear plasmids versus ϳ20 copies/cell for circular plasmids; data not shown), the overall expression levels are much lower for linear plasmids (Figs. 2 and 3)). We believe that (Ϫ) DNA supercoiling causes the expression difference between circular and linear plasmids. As shown previously (22), in the early log phase when we measured the activities of ␤-galactosidase and firefly luciferase, circular plasmids with an actively transcribing unit become hypernegatively supercoiled. Usually, promoters including P T7A1/O4 and P leu-500 are sensitive to supercoiling status of DNA templates and have highest activities at certain (Ϫ) superhelical densities (45). The hypernegative supercoiling of the circular plasmids is likely to inhibit the expression of ␤-galactosidase and firefly luciferase in VS111 and cause the expression level of ␤-galactosidase and firefly luciferase to be higher in MG1655 than that in VS111. Linear plasmids are different. As shown previously, linear plasmids cannot be globally supercoiled (46). As a result, only transient, localized TCDS may be produced. Because VS111 does not contain DNA topoisomerase I to remove (Ϫ) DNA supercoils produced during transcription elongation, more localized (Ϫ) supercoils are "accumulated" on linear plasmids in VS111 than in MG1655 to provide an optimal superhelicity for P T7A1/O4 and P leu-500 . This is the reason why the expression level is higher for linear plasmids in VS111.

Potent activation of P leu-500 by TCDS on the E. coli chromosome for MG1655 and VS111
Next, we examined whether TCDS is able to activate P leu-500 on the chromosome for both the wild-type strain MG1655 and the topA strain VS111. For this purpose, we adopted a procedure for site-specific insertion of transgenes into the E. coli chromosome using transposon Tn7 (47). Using this method, we inserted the divergently coupled P leu-500 and P T7A1/O4 promoters with the luc and lacZ genes (Fig. 4A) to the attTn7 site of the E. coli chromosome (84 min of the chromosome; (48)). Because of technical difficulties, we were not able to include the four terminators in these constructs. As expected, TCDS provided by E. coli RNA polymerase was able to strongly activate P leu-500 on the chromosome. For instance, TCDS was able to activate P leu-500 ϳ18and 6-fold in VS111 and MG1655, respectively (Fig. 4B). In contrast, IPTG was not able to activate the P leu-500 Figure 2. Activation of the supercoiling-sensitive P leu-500 by TCDS provided by E. coli RNA polymerase for circular plasmid pZXD133. Activities of ␤-galactosidase and firefly luciferase were measured as described under "Experimental Procedures." Briefly, overnight cultures of E. coli cells were diluted 100-fold and grown in the presence of various amounts of IPTG to an A 600 of 0.5 and assayed for ␤-galactosidase (Miller's units) and luciferase activities (RLU). A, plasmid map for pZXD133. Winged triangles represent a set of four Rho-independent E. coli rrnB T1 terminators. The directions of P T7A1/O4 , P leu-500 , luc, lacZ, and T1 terminators are indicated. B and C, activities of ␤galactosidase (␤-gal, B) and firefly luciferase (C) for MG1655(DE3)⌬lacZ (black squares and lines) and VS111(DE3)⌬lacZ (red solid circles and lines) that carry pZXD133. Blue open circles and lines represent results from VS111(DE3)⌬lacZ carrying the circular plasmid pZXD95 that does not contain a P T7A1/O4 as controls. D, real-time RT-PCR analyses for VS111(DE3)⌬lacZ (columns 1, 2, 5, and 6) or MG1655(DE3)⌬lacZ (columns 3, 4, 7, and 8) carrying pZXD133 in the absence (columns 1, 3, 5, and 7) and presence (columns 2, 4, 6, and 8) of 500 M of IPTG. The standard deviation was calculated according to three independent experiments. Figure 3. Activation of the supercoiling-sensitive P leu-500 by TCDS provided by E. coli RNA polymerase for linear plasmid pZXD143. Activities of ␤-galactosidase and firefly luciferase were measured as described under "Experimental Procedures" and also in Fig. 2. A, plasmid map for pZXD143. Winged triangles represent a set of four Rho-independent E. coli rrnB T1 terminators. The directions of P T7A1/O4 , P leu-500 , luc, lacZ, and T1 terminators are indicated. B and C, activities of ␤-galactosidase (␤-gal, B) and firefly luciferase (C) for MG1655(DE3)⌬lacZ (black squares and lines) and VS111(DE3)⌬lacZ (red solid circles and lines) that carry pZXD143. The standard deviation was calculated according to three independent experiments.

Stimulation of the leu-500 promoter by transient DNA supercoiling
in the absence of a divergently coupled P T7A1/O4 . Again, the activation level of P leu-500 is correlated with the transcription level from P T7A1/O4 , regardless of the genetic backgrounds of host strains (Fig. 4, A and B). Similar results were obtained when IPTG-inducible E. coli ribosomal rrnB P1 and P2 promoters were used to replace P T7A1/O4 for MG1655 and VS111 (Fig.  4, C-E). Because all seven E. coli ribosomal operons carry similar P1 and P2 promoters (49), these results suggest that TCDS from ribosomal promoters has great effects on the nearby promoters/transcriptions.
Because of the potent effects of TCDS from rRNA promoters on the divergently coupled promoters in our model system, we decided to examine whether TCDS from the seven rRNA promoters affect the transcriptional levels of the divergent coupled genes on the chromosome directly using quantitative RT-PCR. As a first step, we searched promoters or transcription units divergently coupled to the seven ribosomal promoters or operons and found that three genes, purH, yieP, and yrdA, are divergently coupled to the P1 and P2 promoters of rrnE, rrnC, and rrnD operons, respectively (50). However, only one 70 promoter, i.e. the purH promoter, is divergently coupled to the rrnE operon (50). No 70 promoters are associated with yieP and yrdA (only transcription starting sites are found) (50). These results suggest that E. coli may eliminate promoters divergently coupled to the rrn operons because of unnecessary transcription activation by TCDS from the rRNA promoters. Because the rRNA promoters are highly active in the early log phase and greatly depressed in the stationary phase (51), if our hypothesis is correct, the transcriptional levels of purH, yieP, and yrdA should be greatly inhibited by TCDS from rRNA transcription in the early log phase compared with those in the stationary phase. Fig. 4F shows our results. Indeed, the transcriptional levels of purH, yieP, and yrdA in the stationary phase are significantly higher than those in the early log phase for both MG1655 and VS111. There are some differences between these two strains. Although the transcriptional level of purH, yieP, and yrdA in the early log phase is slightly lower for VS111 than that for MG1655, the transcriptional level of these three genes is much higher for VS111 than that for MG1655 in the stationary phase. These results suggest that topA or DNA supercoiling plays an important role in regulating their transcription during these two phases in E. coli. Nevertheless, these results also suggest that TCDS from the seven rRNA promoters greatly affects the transcription level of the divergently coupled promoters or genes in the early log phase. E. coli strains FL1130 (MG1655(DE3)⌬lacZ attTn7::P T7A1/O4 lacZ-P leu-500 luc) and FL1131 (VS111(DE3)⌬lacZ attTn7::P T7A1/O4 lacZ-P leu-500 luc) that carry a pair of divergently coupled promoters, P T7A1/O4 and P leu-500 on the chromosome were created by using a Tn7-mediated method. The ␤-galactosidase (␤-gal) and luciferase activities were measured according to the methods as described under "Experimental Procedures" and also in Fig. 2. A and B, the activation of P leu-500 by TCDS provided by E. coli RNA polymerase on the chromosome for FL1130 (black squares and lines) and FL1131 (red solid circles and lines). The activities of ␤-galactosidase (A; Miller's units) and firefly luciferase (B; RLU) were plotted against the IPTG concentration added to the cell cultures. Controls represent results from E. coli strains carrying the same construct without P T7A1/O4 . C, E. coli strains FL1261 (MG1655(DE3)⌬lacZ attTn7::rrnBP 1 , P 2/lacO1 lacZ-P leu-500 luc) and FL1262 (VS111(DE3)⌬lacZ attTn7:: rrnBP 1 , P 2/lacO1 lacZ-P leu-500 luc) that carry a pair of divergently coupled promoters, IPTG-inducible ribosomal rrnB P1 and P2 promoters, and P leu-500 on the chromosome were created by using a Tn7-mediated method. D and E, E. coli strains FL1261 (black squares and lines) and FL1262 (red circles and lines) were used. The activities of ␤-galactosidase (D; Miller's units) and firefly luciferase (E; RLU) were plotted against the IPTG concentration added to the cell cultures. Red ovals represent the lac O1 operators. F, real-time RT-PCR analyses of transcription levels of purH, yrdA, and yieP in the early log phase and stationary phase for MG1655 and VS111. RNA samples were purified from the early log phase (A 600 ϭ ϳ0.5) and the stationary phase (the overnight culture) as described under "Experimental Procedures." MG1655L, MG1655S, VS111L, and VS111S represent results of the log phase (L) and the stationary phase (S) of MG1655 and VS111, respectively. The standard deviation was calculated according to three independent experiments.

Stimulation of the leu-500 promoter by transient DNA supercoiling Activation of P leu-500 by TCDS is dependent on the promoter strength
Because previous studies showed that TCDS is dependent on the promoter strength in E. coli topA strains (22), we decided to test whether the activation of P leu-500 by TCDS is also dependent on the promoter strength of the divergently coupled promoters. For this experiment, we utilized four IPTG-inducible promoters with different strengths, i.e. P T7A1/O4 , P tac , P lacUV5 , and P lac . These promoters were placed divergently to P leu-500 and used to control the transcription and expression of lacZ (supplemental Fig. S1). We examined TCDS from these promoters on P leu-500 at three different levels: circular plasmid, linear plasmid, and chromosome. Fig. 5 and supplemental Figs. S2 and S3 show that transcription or TCDS initiated from all promoters were able to substantially activate the divergently coupled P leu-500 in both topA and wild-type strains. The activation level is proportional to the promoter strength: the stronger the promoter, the more activation ( Fig. 5 and supplemental Fig.  S3). These results further demonstrate that TCDS is able to potently activate the divergently coupled P leu-500 in E. coli.
Interestingly, the activation of P leu-500 is more potent at the chromosomal level than at the plasmid levels for both strains ( Fig. 5D and supplemental Fig. S3D). Nevertheless, there are some differences between these two E. coli strains. For instance, in the topA strain, the activation level is almost linearly proportional to the promoter strength (Fig. 5D). However, TCDS from the strong artificial P tac only slightly activated the divergently coupled P leu-500 in the wild-type strain (supplemental Fig. S3), suggesting that other factors may also be involved. We again found that a higher expression of ␤-galactosidase always corresponds to a higher expression of firefly luciferase, independent of the host genetic background, although the fold of activation could be substantially different (Fig. 5 and supplemental Figs. S2 and S3).

TCDS by T7 RNA polymerase also strongly activates P leu-500
Previously we demonstrated that transcription by T7 RNA polymerase significantly supercoiled plasmid DNA templates containing a T7 promoter in topA strains (21). We therefore decided to examine whether TCDS by T7 RNA polymerase is also able to activate the divergently coupled P leu-500 in E. coli. For this purpose, we constructed a circular (pZXD99) template and a linear (pZXD103) plasmid DNA template (supplemental Fig. S4) that carry a T7 promoter divergently coupled to P leu-500. Additionally, pZXD99 and pZXD103 carry a lacZ gene whose codon usage was optimized to eukaryotic cells. In this case, very little ␤-galactosidase was expressed from the open reading frame of lacZ after IPTG induction because the high level of expression of ␤-galactosidase under the control of T7 promoter is toxic to the host cell (data not shown; we were not able to obtain a plasmid that contain a lacZ gene whose codon usage was optimized to E. coli under the control of T7 promoter in VS111(DE3) or MG1655(DE3)). These two plasmids were transformed into VS111(DE3) or MG1655(DE3). After the addition of IPTG to the cell culture in the early log phase, the activation of the leu-500 promoter was monitored by measuring the luciferase activities and also by determining the transcription level using RT-PCR assays. The results in Fig. 6 clearly demonstrate that TCDS by T7 RNA polymerase was able to activate the supercoiling-sensitive P leu-500 . For example, TCDS by T7 RNA polymerase on the circular plasmid pZXD99 was capable of activating the leu-500 promoter ϳ8.0and 2.2-fold in VS111(DE3) and MG1655(DE3), respectively (Fig. 6A). TCDS by T7 RNA polymerase on the linear plasmid pZXD103 was able to activate P leu-500 ϳ4.0and 1.6-fold in VS111(DE3) and MG1655(DE3), respectively (Fig. 6B). These results suggest that the activation of P leu-500 by TCDS is independent of the RNA polymerase employed, and transcription alone is sufficient to produce significant amounts of localized (Ϫ) DNA supercoils and therefore activates P leu-500 .
Because TCDS by T7 RNA polymerase is dependent on the length of RNA transcripts in topA strains (21), we wondered whether the activation of P leu-500 by T7 RNA polymerase is also dependent on the length of RNA transcripts. For these experiments, we systematically truncated the lacZ of pZXD99 so that four transcripts with different lengths were generated after IPTG induction. The results in supplemental Fig. S5, A and B clearly demonstrate that the activation of P leu-500 by TCDS

Stimulation of the leu-500 promoter by transient DNA supercoiling
from T7 RNA polymerase is proportional to the RNA transcript length: the longer the RNA transcripts, the more activation. In this paper, we also examine whether the expression of a membrane-insertion protein, such as tetA, affects the activation level of P leu-500. For these experiments, we constructed a circular plasmid pZXD118 and a linear plasmid pZXD123 in which tetA is under the control of the T7 promoter (supplemental Fig. S4). Additionally, we constructed another linear plasmid pZXD124 that carries a 1.2-kb fragment of lacZ under the control of a T7 promoter for comparison (supplemental Fig. S4). These plasmids were introduced into VS111(DE3)⌬lacZ. After IPTG induction, TCDS from the transcription of tetA by T7 RNA polymerase is able to activate P leu-500 for both circular and linear plasmids (supplemental Fig. S5, C and D). The activation level of P leu-500 is higher for tetA than that for the 1.2-kb fragment of lacZ (supplemental Fig. S5, C and D). At 50 M of IPTG, the activation level of P leu-500 for tetA dramatically decreased, especially for cells carrying the circular plasmid pZXD118. This decrease may attribute to the toxic effect of overexpressing tetA because cells started dying at 50 M of IPTG (52). These results suggest that co-transcription and translation of a membraneinsertion protein is indeed able to induce more localized (Ϫ) supercoils behind the transcribing RNA polymerase and therefore enhance the activation of P leu-500 .

Discussion
We presented clear and convincing evidence demonstrating that TCDS is able to potently activate the divergently coupled, supercoiling-sensitive promoter P leu-500 . The activation was so potent that in certain cases it reached 18-fold comparing with that in the absence of TCDS (Fig. 4). These results mostly favor the twin supercoiled domain model of transcription (23). Not only do (Ϫ) supercoils generated behind the RNA polymerase provide sufficient free energy to overcome the energy barrier for the formation of a functional transcription open complex and thus activate P leu-500 , but also the activation of P leu-500 by TCDS has functional properties consistent with those predicted for the twin-domain mechanism (21,23). For instance, the activation of P leu-500 is dependent on the promoter strength ( Fig. 5 and supplemental Fig. S3) and the length of RNA transcripts (supplemental Fig. S5), and independent of the RNA polymerase used (supplemental Fig. S5). Additionally, the activation of P leu-500 by TCDS was enhanced by the expression of a membrane-insertion protein tetA (supplemental Fig. S5). All evidence suggests that the twin-supercoiled domain mechanism is at work in vivo.
Previously it was demonstrated that TCDS was able to activate P leu-500 in topA strains (25)(26)(27)(31)(32)(33)(34)(35)(36). However, it was difficult to determine whether transient or global supercoiling contributes to the activation because the small circular plasmid DNA templates were usually used in these studies in which the twin supercoiled domains produced by transcription can diffuse along the plasmid DNA and therefore cancel each other. Additionally, transcription drives a significant amount of plasmid DNA templates into a hypernegatively superhelical status in topA strains (7,21,22). In this way, the activation might result from the hypernegative supercoiling that was introduced into the DNA templates. Our results presented here unambiguously demonstrate that transient and dynamic TCDS is responsible for the activation of P leu-500 . First, TCDS was able to strongly activate P leu-500 in the wild-type strain MG1655 (Figs. 2, 4, and 6 and supplemental Fig. S3). Because MG1655 has all four DNA topoisomerases, the global supercoiling level of DNA templates is not dramatically fluctuated (4,5). Additionally, hypernegatively supercoiled DNA cannot be generated in MG1655 (44). This rules out the possibility of building up sufficient permanent supercoiling on the DNA templates to activate P leu-500 . In other words, the activation of P leu-500 must stem from transient and dynamic (Ϫ) supercoiling generated by the divergently coupled transcription unit in MG1655. Second, TCDS was able to potently activate P leu-500 on the linear plasmids in E. coli (Figs. 3, 5, and 6 and supplemental Fig. S3). Because linear DNA templates cannot be globally supercoiled (46), the use of linear plasmid DNA templates in E. coli wildtype strains unequivocally demonstrated that transient and dynamic TCDS, rather than global supercoiling, activates the divergently coupled P leu-500 . Intriguingly, our results showed that the activation of P leu-500 by TCDS is more potent on the chromosome than on plasmid DNA templates (Figs. 5 and supplemental Fig. S3). A possible reason is that unlike plasmid DNA templates, DNA supercoiling domains on the chromosome cannot cancel each other through merging supercoils of

Stimulation of the leu-500 promoter by transient DNA supercoiling
opposite signs. In this way, the average lifetime of transient and dynamic TCDS on the chromosome is longer. As a result, the activation of P leu-500 will be more potent on the chromosome. Nevertheless, this result suggests that the local environment between these two types of DNA templates is quite different.
Our results also showed that TCDS could be produced on topologically open DNA molecules, i.e. linear DNA molecules in E. coli cells (Figs. 3 and 5 and supplemental Fig. S3). This result suggests that DNA molecules do not need to anchor to big cellular structures to generate TCDS in vivo, which is not in accord with the classical Liu-Wang twin-supercoiled-domain model (23). In a theoretical study to discuss TCDS, Nelson (42) considered the role of small natural bends in the DNA double helix backbone and showed that transcribing RNA polymerases could induce sufficient friction force on linear DNA molecules and therefore produce significant amounts of TCDS. This theoretical work suggested that DNA molecules are not required to be anchored to produce TCDS (42). Consistent with Nelson's prediction, Levens and co-workers (43) demonstrated that small RNA polymerases, such as T7 and T3 RNA polymerase, were able to produce TCDS on linear, unanchored DNA molecules in an in vitro defined protein system. Furthermore, they demonstrated that TCDS was able to drive the supercoiling sensitive far upstream element of the human c-myc into singlestranded DNA (43). Here we went a step further and demonstrated that in E. coli cells, RNA polymerases are able to produce sufficient amount of TCDS to activate the divergently coupled P leu-500 on linear, unanchored DNA molecules. Apparently, DNA topological boundaries or barriers are not required.
RNA polymerases are powerful motor proteins that rapidly move along the E. coli chromosome (53,54). During the exponential phase when RNA polymerases actively transcribe genes on the chromosome, these motor proteins should produce significant amount of TCDS and remodel the chromosome structure. Indeed, our results showing that the rrnB P1 and P2 promoters were able to greatly activate the divergently coupled P leu-500 (Fig. 4, C-E) are consistent with this hypothesis. Our quantitative RT-PCR results in Fig. 4F also support this hypothesis. If not regulated, the large amounts of TCDS would be a problem for E. coli cells, especially for regions around actively transcribed operons such as the seven ribosomal operons (51). In other words, how do E. coli cells control the localized DNA supercoiling around these ribosomal operons during the exponential phase? As mentioned above, in E. coli, DNA supercoiling status is usually set by the counter actions of DNA topoisomerase I and gyrase (4,5). DNA topoisomerase I should be responsible for removing excess (Ϫ) supercoils behind a transcribing RNA polymerase (23). However, during rapid transcription phase, is DNA topoisomerase I alone able to manage TCDS from RNA polymerases transcribing seven ribosomal operons? Our results showed that strong promoters are still able to potently activate the neighboring weak promoters even in the wild-type strain MG1655 (Fig. 4), suggesting that E. coli topoisomerases are not the only regulator to control DNA supercoiling in E. coli. Possibly E. coli cells also utilize other regulators, such as protein-mediated topological barriers (55), to control DNA supercoiling at localized levels. For these ribosomal operons, the upstream region of each P1 promoter car-ries several tandem copies of the FIS binding sites for the activation of the transcription from the P1 promoters (56). However, the activation mechanism is not still well understood. One possibility is that FIS upon binding to the FIS-binding sites forms topological barriers (8,55,57) and blocks supercoiling (TCDS) diffusion. In this case, TCDS generated from the strong P1 and P2 promoters is utilized to activate the P1 promoters and make more rRNA, which is beneficial to cell growth. It is also possible that the FIS-mediated topological barriers prevent TCDS from activating or inhibiting other nearby promoters. In this scenario, the FIS-mediated topological barriers give DNA topoisomerases more time to remove the excess, harmful supercoiling. So far, our experimental results are consistent with this hypothesis. Nevertheless, additional experiments that test this hypothesis are still needed. Furthermore, other DNAbinding proteins, such as histones and histone-like proteins, may use the same mechanism to modulate localized DNA supercoiling on eukaryotic chromosomes.

Proteins, chemicals, and reagents
Ethidium bromide, kanamycin, and lysozyme were purchased from Sigma-Aldrich. Ampicillin and BSA were obtained from Fisher Scientific. IPTG was obtained from Anatrace (Maumee, OH). All restriction enzymes, T4 DNA ligase, T4 polynucleotide kinase, and E. coli DNA gyrase were bought from New England Biolabs (Beverly, MA). Pfu DNA polymerase was purchased from Stratagene, Inc. (La Jolla, CA). All synthetic oligonucleotides were obtained from MWG-Biotech, Inc. (Huntsville, AL). QIAprep spin miniprep kit, QIAquick gel extraction kit, RNeasy mini kit, and QIAquick nucleotide removal kit were bought from Qiagen. ThermoScript RT-PCR System plus Platinum Taq DNA polymerase was purchased from Invitrogen. Power SYBR Green PCR Master Mix was obtained from Applied Biosystems. The luciferase assay system is a product of Promega Corporation. SYBR Gold nucleic acid gel stain was purchased from Life Technologies, Inc.

Plasmid DNA templates
Properties of plasmids used in this study are summarized in supplemental Table S1. All circular plasmids were derived from plasmid pBR322. Construction of plasmid DNA templates sometimes required DNA fusions between non-complementary cohesive termini. In this scenario, cohesive ends were converted before ligation to blunt ends by incubation of the DNA fragments with T4 DNA polymerase in the presence of dNTPs. Plasmids pZXD99, pZXD133, pZXD145, and pZXD146 were described previously (44). Plasmid pZXD95 is the self-ligation product of the 8,792-bp AatII fragment of pZXD77 (44). As a result, pZXD95 do not have a T7 promoter. Plasmids pZXD115 and pZXD116 are, respectively, self-ligation products of the 8,868-bp HpaI fragment and the 8,287-bp MluI fragment of pZXD99. Plasmid pZXD120 is the self-ligation production of the 7,663-bp HpaI fragment of pZXD116. Transcription of pZXD99, pZXD115, pZXD116, and pZXD120 by T7 RNA polymerase produces 3.1-, 2.4-, 1.9-, and 1.2-kb RNA transcripts, respectively. Plasmid pZXD118 was constructed by inserting a 1,280-bp DNA fragment carrying a tetA gene into the AgeI and

Stimulation of the leu-500 promoter by transient DNA supercoiling
BsmI sites of pZXD99. In this case, the lacZ gene was replaced by the tetA gene. Plasmid pZXD147 was made in two steps. First, an 87-bp synthetic DNA oligomer containing P lac was inserted between the EcoRI and XhoI sites of pZXD99 to generate pZXD108. Then a 3,093-bp PCR product containing the lacZ gene amplified from MG1655 genomic DNA was cloned into the AgeI and BsmI sites of pZXD108 to yield pZXD147.
All linear plasmids were derived from the coliphage N15based, linear plasmid pZXD4 described previously (22). Like circular plasmids, these linear plasmids are able to replicate independently from the host chromosome and maintain in E. coli. Plasmid pZXD91 was constructed by inserting the 6,884-bp BglII-SpeI fragment of pZXD86 into the BglII and NheI sites of pZXD4. Plasmid pZXD103 was made by the insertion of a 6,763-bp BglII-SpeI DNA fragment of pZXD99 into the BglII and NheI sites of pZXD4. Plasmid pZXD123 was generated by inserting the BglII-SpeI fragment of pZXD118 into the BglII and NheI sites of pZXD4. Plasmid pZXD124 was produced by inserting the BglII-SpeI fragment of pZXD120 into the BglII and NheI sites of pZXD4. Plasmid pZXD143 was constructed by inserting the 6,807-bp BglII-SpeI fragment of pZXD133 into the BglII and NheI sites of pZXD4. Plasmid pZXD150 was created by inserting a 6,817-bp BglII-SpeI DNA fragment of pZXD144 into the BglII and NheI sites of pZXD4. Plasmid pZXD151 was constructed by inserting a 6,837-bp BglII-SpeI fragment of pZXD145 into the BglII and NheI sites of pZXD4. Plasmid pZXD152 was made by inserting a 6,839-bp BglII-SpeI fragment of pZXD146 into the BglII and NheI sites of pZXD4. Plasmid pZXD153 was created by the insertion of a 6,839-bp BglII-SpeI fragment of pZXD147 into the BglII and NheI sites of pZXD4.

The expression of ␤-galactosidase
The expression level of ␤-galactosidase was measured by Miller's assay as described (58). Briefly, 100 ml of LB was inoculated with 1 ml of overnight bacterial cell culture until the A 600 level reached ϳ0.5. 100 l of bacterial cell culture was mixed with 900 l of Z buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 1 mM MgSO 4 , and 50 mM ␤-mercaptoethanol). The cells were lysed with 60 l of chloroform and 30 l of 0.1% SDS. After cell lysates were equilibrated at 30°C for 5 min, 200 l of 4 mg/ml ortho-Nitrophenyl-␤-galactoside (ONPG) was added to the cell lysates. After an additional 15 min of incubation at 30°C, the reactions were stopped by the addition of 500 l of 1 M Na 2 CO 3 . After cell debris was removed by centrifugation at 13,000 rpm for 1 min, the A 420 and A 550 were measured in a Cary 50 spectrophotometer. ␤-Galactosidase activities (E) were calculated using the following equation, where t and v represent reaction time and cell culture volume, respectively.

Luciferase assay
Luciferase assay was used to measure the expression of the luciferase gene in various E. coli strains carrying different plasmid DNA templates. Briefly, E. coli cells carrying different plasmids were grown overnight in LB. Antibiotics were added to LB as needed. The overnight culture was then diluted (1:100) in fresh LB in the presence of different concentrations of IPTG and grown until the A 600 reached ϳ0.5. Next, 50 l of cells were mixed with 10 l of 1 M K 2 HPO 4 (pH 7.8) and 20 mM EDTA, quickly frozen in liquid nitrogen for 3 min, and equilibrated to room temperature for 30 min to yield about 60 l of cell lysate. Then the cell lysate was added with 300 l of freshly prepared lysis mix containing 1ϫ cell culture lysis reagent, 1.25 mg/ml lysozyme, and 2.5 mg/ml BSA, and incubated for 10 min at room temperature. Finally, 100 l of luciferase assay reagent (Promega) was added to 20 l of the cell lysate and subsequently analyzed by a Promega GloMax 20/20 single-tube luminometer or a Biocounter luminometer (Titusville, FL).

RNA isolation, cDNA synthesis, and PCR
Total RNA was isolated from E. coli cells using a Qiagen RNeasy kit as described by the manufacturer. To determine the integrity of the total RNA samples, 16S and 23S rRNA were resolved by electrophoresis in a 1.2% agarose gel in 1ϫ MOPS buffer containing formaldehyde (20 mM MOPS, 8 mM sodium acetate anhydrous, and 1 mM EDTA, pH 7.0, and 1% formaldehyde). After electrophoresis, agarose gels were stained with ethidium bromide, destained, and photographed under UV light. cDNA was synthesized from total RNA samples using a ThermoScript RT-PCR system. 2.76 g of RNA was mixed with random hexamer primers (50 ng/l) and four dNTPs (final concentration, 1 mM). The mixtures were incubated at 65°C for 5 min and transferred on ice for another 5 min to remove secondary structures of RNA. The denatured RNA samples were then mixed with 1ϫ cDNA synthesis buffer with a total volume of 20 l containing 5 mM DTT, 40 units of RNaseOut, and 15 units of ThermoScript reverse transcriptase and incubated at 25°C for 10 min followed by 50°C for 50 min to synthesize cDNA. The cDNA synthesis mixtures were transferred to an 85°C water bath for 5 min to terminate the reactions. After the synthesis step, the reaction mixtures were added with 2 units of

Stimulation of the leu-500 promoter by transient DNA supercoiling
RNase H and incubated at 37°C for 20 min to remove the RNA templates.
PCRs were carried out using cDNA samples synthesized as described above. A 50-l PCR contains 1ϫ PCR buffer without Mg 2ϩ , 1 mM MgCl 2 , 0.2 mM dNTPs, 0.2 M of each primer, 0.5 l of cDNA, and 2 units of Platinum Taq DNA polymerase. The reactions were started at 94°C for 2 min; proceeded for 16 cycles (for linear plasmids, used 21 cycles instead) of 94°C for 30 s, 55°C for 30 s and 72°C for 1 min; and then terminated at 72°C for 10 min. Subsequently, the PCR products were analyzed by electrophoresis in a 12% polyacrylamide gel in 1ϫ TAE buffer. After electrophoresis, polyacrylamide gels were stained with ethidium bromide, destained, and photographed under UV light.

Real-time PCR assays
Real-time PCR assays were carried out using MiniOpticon real-time PCR system (Bio-Rad). A 20-l reaction contains 1 l of cDNA, 0.5 M of each primer, and 10 l of Power SYBR Green PCR Master Mix (2ϫ). The reaction started at 95°C for 10 min and continued for 40 cycles at 95°C for 15 s and 60°C for 1 min. The C q values (quantification cycle values) were calculated from exponential phase of each PCR amplification reaction as recommended by the manufacturer. The primers used in the RT-PCRs are summarized in supplemental Table S3.