A 72-Base Pair AT-rich DNA Sequence Element Functions as a Bacterial Gene Silencer*

We have previously demonstrated that sequential activation of the bacterial ilvIH-leuO-leuABCD gene cluster involves a promoter-relay mechanism. In the current study, we show that the final activation of the leuABCD operon is through a transcriptional derepression mechanism. TheleuABCD operon is transcriptionally repressed by the presence of a 318-base pair AT-rich upstream element. LeuO is required for derepressing the repressed leuABCD operon. Deletion analysis of the repressive effect of the 318-bp element has led to the identification of a 72-bp AT-rich (78% A+T) DNA sequence element, AT4, which is capable of silencing a number of unrelated promoters in addition to the leuABCD promoter. AT4-mediated gene silencing is orientation-independent and occurs within a distance of 300 base pairs. Furthermore, an increased gene-silencing effect was observed with a tandemly repeated AT4 dimer. The possible mechanism of AT4-mediated gene silencing in bacteria is discussed.

The leu-500 mutation is an A to G transition in the Ϫ10 region of the promoter of the Salmonella typhimurium leuABCD operon (1). The transcriptional activity of the mutant promoter is DNA supercoiling-dependent (2). The mechanism whereby the leu-500 promoter (pleu-500) is activated in the topA mutants is intriguing (3)(4)(5)(6)(7). Previous studies using a plasmid system have demonstrated that activation of plasmidborne pleu-500 in topA mutants requires an upstream transcriptional activity transcribing away from pleu-500 (8 -11). This notion has been confirmed in a recent study using the chromosomal setting (12). Transcriptional activation of the il-vIH promoter (pilvIH) located 1.9 kilobases upstream of pleu-500 was shown to be responsible for pleu-500 activation (5). Transcription-driven DNA supercoiling (13) has been suggested to play a role in this long-range promoter-promoter interaction.
The intervening promoter that relays the distant interaction between pilvIH and pleu-500 is the leuO promoter (pleuO). In addition to transcriptional activity from pleuO, the leuO gene product, LeuO, is also required to provide a trans-acting function for activation of pleu-500 (6). It appears that the functional pleuO (or other replaced promoter) and LeuO are coupled in activating pleu-500. The molecular basis for pleu-500 activation by the combined action of pleuO and LeuO is still a mystery.
There is a stretch of 434 base pairs (bp) 1 that is AT-rich DNA flanked by the divergently arrayed leuO and leuABCD (14). Besides the promoter sequences of the flanking genes, the function of the remaining 318-bp AT-rich (69% AϩT) DNA is unknown (illustrated in Fig. 1). By monitoring pleu-500 activation, we found that the 318-bp AT-rich intervening DNA appears to repress the short-range interaction (11) between the two flanking promoters. Interestingly, LeuO relieves the repression. The repressive effect of the AT-rich intervening DNA on the short-range promoter-promoter interaction (pleuO and pleu-500) could potentially be due to anchoring of the AT-rich DNA to a large mass, which restricts DNA rotation and thereby abolishes short-range promoter-promoter interaction via DNA supercoiling. However, detailed analysis to search for DNA rotation blockage has ruled out this anchorage possibility.
The repressive activity of the 318-bp AT-rich intervening DNA has been narrowed down to a 72-bp AT-rich (78% AϩT) DNA, referred to as AT4 in this work. AT4 is located at the pleuO end of the 318-bp AT-rich DNA. AT4 can repress promoter activity within a 300-bp distance. This repression is independent of the orientation of AT4. AT4-mediated repression of the promoter activity appears to be promoter nonspecific, because all promoters tested are repressed by AT4. These results support a role for AT4 as a gene silencer in bacteria.
pWU812T was derived from pWU802T. To construct pWU812T, a 320-bp promoterless non-AT-rich (47% AϩT) DNA was generated by polymerase chain reaction (PCR) from the coding region of human cathepsin B gene (19), with the primers introducing HincII and BstXI sites at the ends. The digested HincII-BstXI fragment was then used to replace the 318-bp HincII-BstXI segment containing the native AT-rich sequence (69% AϩT) between the divergently transcribing ptac and pleu-500 in pWU802T. The native 318-bp AT-rich intervening DNA was also PCR-amplified with primers containing AatII restriction sites. The AatII-digested AT DNA fragment was inserted into the unique AatII site on pAO to yield pAO-AT and pAO-ATR. The plasmid carrying an AT DNA insert, with the leuO end of the DNA orientated proximal to pbla, was designated pAO-AT. The plasmid carrying the DNA insert in the opposite orientation was designated as pAO-ATR. Similarly, the 72-bp DNA located near the pleuO end of the 318-bp AT-rich DNA was PCR-amplified with primers so that the 72-bp AT4 DNA was flanked by AatII restriction sites on both ends. The AatII-digested AT4 DNA was inserted at the unique AatII site on pAO to yield pAO-AT4 (the leuO end of the DNA insert was proximal to pbla) and pAO-AT4R (the opposite orientation). The AT4 DNA sequence was 5Ј-CACAATCATACACCAA-* This work was supported by a National Institutes of Health Grant GM-53617. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  GTGAATGATCATTTAAGTTTCAATTAAATGTTTATATTATTAATAG-CTAAAAAGTT-3Ј. The nucleotide sequence of the rest of the intervening DNA between the divergently arrayed leuO and leuABCD genes can be obtained from the GenBank data base (accession number AF106956). Other testing plasmids were all derived from the above-described plasmids and were individually described in the experiments. The following 72-bp DNA sequence from the lacZ coding region was used to replace the AT4 DNA on pWU802T as described in Fig. 6. 5Ј-AACCATCGAAGTGACCAGCGAATACCTGTTCCGTCATAGCGAT-AACGAGCTCCTGCACTGGATGGTGGTACC-3Ј.
The following four synthetic DNA oligomers, consisting of nucleotide sequences of the lacZ coding region, were used to sequentially extend the distance between the AT4 insert and pbla in pAO-AT4 as described in Fig. 8: 5Ј-GCGATCTTCCTGAGGCCGATACTGTCGTCGTCCCC-TCAAACTGGCGGATCCAA-3Ј; 5Ј-GATCCACGGTTACGATGCGCCC-ATCTACACCAACGTAACCTATCCCATGAATTC-3Ј; 5Ј-AATTCAATC-CGCCGTTTGTTCCCACGGAGAATCCGACGGGTTGTTACTGGTAC-C-3Ј; and 5Ј-CATTTAATGTTGATGAAAGCTGGCTACAGGAAGGCCA-GACGCGAAAGCTTGTAC-3Ј. CH601(topA ϩ ) and CH582(topA Ϫ ), an isogenic pair of S. typhimurium strains, were described previously (3) and provided by Dr. David Lilley. AS19, an Escherichia coli B strain that is permeable to drugs such as novobiocin (20) was previously used in studies involving gyrase inhibition (15,21). The leuO Ϫ strain, MF1, was derived from MC4100, an E. coli K-12 strain (22). TO2 is a ⌬leuO strain due to the replacement of the BstXI-ApaI fragment of the leuO coding region with a DNA fragment carrying the cam (cm r ) gene (23). MF1 was prepared by introducing the leuO::cam mutation into MC4100 (recipient strain) by P1 transduction using TO2 as a donor strain. Disruption of the leuO in the chromosome in MF1 was confirmed by the restriction enzyme cleavage pattern and the size of the DNA product of polymerase chain reaction (PCR). Bacteria were grown in Luria-Bertani (LB) medium at 37°C with aeration. 50 g/ml ampicillin or 12.5 g/ml tetracycline was added as needed.
Primer Extension-Primer extension was carried out as previously described (11). Several DNA oligomers were used as primers in the study. 5Ј-TCTGGGTGAGACAAAACAGGAAGGC-3Ј was used for detecting pbla-mediated transcripts; 5Ј-AGAATTCTCATGTTTGACAGC-TTATCATCG-3Ј was used for detecting pleu-500-mediated transcripts; 5Ј-CCTATAAAAATAGGCGTATCACGAGGCCCT-3Ј was used for detecting ptac-mediated transcript; 5Ј-CTACAGCATCCAGGGTGACGGT-GCC-3Ј was used for detecting ptetA-mediated transcripts. All primers hybridize only with plasmid DNA sequences so that no transcripts from the chromosomal genes would interfere with the primer extension results. With the exception of Figs. 1 and 6 (see below), an individual primer was mixed with 100 g of total RNA in the primer extension reactions. Two primers were mixed in the primer extension reactions for simultaneous detection of the bla and leu-500 transcripts in Fig. 1 and the leu-500 and tac transcripts in Fig. 6. The initiation sites of RNA transcripts were verified based on the specific sizes of the primer extension DNA products that run off the 5Ј-end of RNA transcripts. A DNA sequence ladder was prepared using each individual primer for verifying the initiation site at a DNA sequence level. The radioactivity of primer extension DNA product was visualized and quantified using a Storm imaging system (model 840, Molecular Dynamics). The reported quantification is the average of at least two experiments. The signals were normalized based on the total plasmid DNA content in the harvested cells. 1.5-ml aliquots of bacterial culture were saved at the harvest and used to prepare total plasmid DNA. The total plasmid DNA in the 1.5-ml sample was analyzed by agarose gel electrophoresis followed by Southern blotting using a 32 p-labeled probe that specifically hybridizes with plasmid DNA.
Analysis of DNA Topoisomers by Two-dimensional Gel Electrophoresis-DNA samples were separated by 1% agarose gel electrophoresis in the first dimension in 1ϫ Tris-phosphate-EDTA (TPE) buffer (containing 80 mM Tris phosphate and 8 mM EDTA, pH 8.0), the 20-ϫ 20-cm gel was then soaked in 4 M chloroquine (Sigma) for 2 h. The soaked gel was turned 90°and electrophoresed in the second dimension in 1ϫ TPE buffer containing 4 M chloroquine. The two-dimensional gels were subjected to in situ Southern hybridization (described in Ref. 21) using pAO-specific 32 p-labeled probe so that the coexisting pEV101 was not visualized.
Immunoblotting Detection of Overexpressed LeuO in a leuO Ϫ Strain-A LeuO-specific antiserum was raised by injecting the purified overexpressed S. typhimurium His-tagged LeuO into a rabbit. The affinity-purified IgG (1.4 mg/ml) from the antiserum was used at a dilution factor of 1:5000 as the primary antibody to detect the cellular LeuO protein. The secondary antibody was anti-rabbit IgG conjugated to alkaline phosphatase. The blot was developed by ECL in a modification of a previous procedure (24) using an ECF Western blotting kit (Amersham Pharmacia Biotech). The chemifluorescence signal was detected and quantified by the Storm 840 imaging system (Molecular Dynamics).

The 318-bp AT-rich Intervening DNA Represses Short-range
Promoter-promoter Interaction-The 318-bp AT-rich DNA is located between the two divergently arrayed promoters, pleuO and pleuABCD. Previous studies have demonstrated that this region is involved in the promoter relay mechanism for pleu-500 activation (5,6). To test the effect of the 318-bp AT-rich intervening DNA on the interaction between pleuO and pleuABCD, we have constructed two plasmids, pWU802T and pWU812T (Fig. 1). pWU802T was constructed from a plasmid that contains the entire region of pleuO, 318-bp AT-rich DNA, and pleuABCD in their corresponding chromosomal context, by replacing pleuO with an IPTG-inducible tac promoter, ptac (25). The leu-500 mutant was included so that a nearly on-off pleu-500 activity change (6) could be used as indication of the short-range ptac-pleu-500 interaction (11). pWU812T was constructed from pWU802T by replacing the 318-bp AT-rich intervening DNA with a 320-bp promoterless and non-AT-rich DNA sequence from human cathepsin B cDNA (19). The leuO coding region in both plasmids was truncated so that no functional LeuO was generated in cis. Upon IPTG induction, LeuO was produced in trans from a coexisting expression vector, pEV101 (6). The absence of the 318-bp AT-rich DNA in pWU812T resulted in an IPTG-inducible pleu-500 activation in S. typhimurium CH582 (topA Ϫ ), regardless of the presence or absence of LeuO (Fig. 1, lanes 6 and 8). In contrast, with the 318-bp native AT-rich intervening DNA in place, pleu-500 failed to be activated in pWU802T in the absence of LeuO (Fig. 1, compare lanes 3 and 4). Consistent with the previous results (6), IPTG induction resulted in activation of pleu-500 on pWU802T when LeuO was provided in trans ( Fig. 1, compare lanes 1 and 2). These results demonstrate a repressive effect of the 318-bp AT-rich intervening DNA on the short-range pleuO(ptac)-pleu-500 interaction. The trans-acting LeuO relieves the repression.
The 318-bp AT-rich Intervening DNA-mediated Repression Is Not Due to Restriction of the Rotational Motion of the DNA Helix-The short-range interaction between the two divergently arrayed promoters, ptac and pleu-500, is almost certainly due to transcription-driven DNA supercoiling (11,13). The repressive effect of the 318-bp AT-rich intervening DNA on the short-range promoter-promoter interaction is intriguing. One possible explanation is that the AT-rich intervening DNA may be organized into a large structure (e.g. anchored to a large mass) so that DNA rotation along its helical axis is restricted (schematically illustrated in Fig. 2). To test this possibility, a DNA supercoiling assay based on previously established methodology (15) was used. For these experiments, the 318-bp ATrich DNA was inserted at the unique AatII site in pAO so that the major transcription unit, ␤-lactamase gene (bla), is transcribing away from the AT-rich DNA insert (illustrated in Fig.  2). The resultant plasmid, pAO-AT, and the parental plasmid, pAO, were used in the assay.
This assay was designed based on the theory that if an anchor was formed at the inserted AT-rich DNA, the accumulation of transcription-driven supercoiling would be intensified due to the lack of a diffusional pathway to dissipate DNA supercoils (15). Under such a condition, inhibition of DNA gyrase would result in a relatively more rapid increase in the DNA linking number (positive DNA supercoiling) in both monomeric and dimeric plasmid DNAs as previously demonstrated using lac operator as a model system (15). When such an experiment was performed in cells harboring pAO-AT, which contained the 318-bp AT-rich DNA instead of the lac operator, no accumulation of the positively supercoiled DNA topoisomers was observed (Fig. 2C). In fact, the topoisomer distribution of pAO was almost identical to that of pAO-AT (Fig. 2, compare A and C), suggesting that the 318-bp AT-rich DNA insert did not significantly restrict DNA helix rotation and therefore could not block DNA supercoils generated by transcription of the bla gene. In addition, IPTG induction (the production of LeuO protein from pEV101) did not affect the topoisomer distribution of either pAO or PAO-AT (Fig. 2, B and  D). These results suggest that the 318-bp AT-rich DNA-mediated repression is unlikely to be due to restriction of DNA helix rotation.
The 318-bp AT-rich DNA Represses Transcription of Adjacent Promoters-The 318-bp AT-rich DNA could repress transcription from one of the flanking promoters (either ptac or pleu-500) and thereby abolishes short-range promoter-promoter interaction. To test this possibility, we examined the transcription activity of ptac in pWU802T and pWU812T. Primer extension results indicated that the ptac activity ( Fig.  3) strikingly correlated with the pleu-500 activity (Fig. 1). The ptac functioned normally if the 318-bp AT-rich DNA was replaced with a neutral DNA sequence of similar size as in pWU812T (lanes 6 and 8 in Fig. 3). In the presence of the native 318-bp AT-rich DNA, the ptac activity on pWU802T was severely impaired (Fig. 3, lane 4). On the same plasmid, the ptac activity was partially restored if LeuO was provided in trans (Fig. 3, lane 2). This result strongly supports the notion that the 318-bp AT-rich intervening DNA is a negative regulatory element for transcription.
A 72-bp DNA Element Within the 318-bp AT-rich DNA Silences Adjacent Genes-To test whether or not the repressive effect of the 318-bp AT-rich DNA element on transcription can be observed with other promoters, the 318-bp DNA was inserted at the unique AatII site located 99 bp upstream of the bla promoter (pbla) in pAO. In either orientation (pAO-AT and pAO-ATR), the 318-bp AT-rich DNA insert caused an ϳ45% reduction of the pbla activity (Fig. 4A, compare lanes 2 and 3 with lane 1). Deletion analysis using pAO-AT had located a predominant gene-silencing effect (more than 80% reduction on pbla activity) in AT4, a 72-bp AT-rich DNA located near the pleuO end of the 318-bp DNA (lane 6 in Fig. 4A). Those DNA inserts containing the 72-bp AT4 plus all or part of the rest of 318-bp AT-rich DNA (AT, ATR, and AT2) exerted lesser genesilencing effects (lanes 2, 3, and 4 in Fig. 4A). Furthermore, the AT1 DNA segment, which represents the 146-bp pleu-500 end of the 318-bp AT-rich DNA, enhanced the pbla activity (lane 5 in Fig. 4A). These results indicate that, although the 72-bp AT4 exhibits a clear gene-silencing effect, a complex transcriptional effect is present in the rest of the 318-bp AT-rich DNA. The stronger silencing effect that is associated with AT4 may be due to elimination of other complex and opposing effects within the 318-bp AT-rich intervening DNA.
AT4-mediated Gene Silencing Is Additive, Orientation-independent, and topA Genetic Background-independent-Inversion of the AT4 DNA insert did not significantly affect gene silencing (compare lanes 8 and 10 in Fig. 4A). The reduction of pbla activity was ϳ80% with either orientation. Furthermore, the gene-silencing effect was additive. In either orientation, an ϳ95% reduction was achieved when the AT4 DNA insert was tandemly repeated (lanes 9 and 11 in Fig. 4A). Thus far, characterization of the 72-bp gene silencer had been carried out in the S. typhimurium topA Ϫ strain, CH582, where pleu-500 activation was originally studied. The topA Ϫ genetic background has been shown to enhance short-range promoter-promoter interaction such as activation of a plasmid-borne pleu-500 (11). To examine whether AT4-mediated gene-silencing effect was dependent on the topA Ϫ genetic background, pAO-AT4 was tested in an S. typhimurium topA ϩ strain, CH601, which is the parental strain of CH582. The same degree of gene silencing (ϳ80% reduction of the pbla activity) was observed in both the topA ϩ and the topA Ϫ strains (Fig. 4B).
LeuO Protein Negates AT4-mediated Gene Silencing-The trans-acting LeuO protein was shown to relieve 318-bp AT-rich DNA-mediated repression of the short-range pleuO (ptac)-pleu-500 interaction in pWU802T (Fig. 1). To test whether or not pWU802T containing the native 318-bp AT-rich intervening DNA and pWU812T containing a 320-bp non-AT-rich DNA replacement were assayed for pleu-500 activation in CH582 using primer extension. Upon IPTG induction, the coexisting pACYC-based plasmid, pEV101, provided both LeuO and lac repressor (LacI), whereas pSO1000 provided only LacI in trans. LacI is required for controlling expression from the IPTG-inducible ptac in pWU802T, pWU812T, and pEV101 in the lac repressor-free CH582. The presence of either one of the coexisting plasmids and the IPTG treatment is indicated with "ϩ" sign. The "Ϫ" sign or blank indicates the absence of the treatment or coexisting plasmid in the experiment. The DNA sequence ladders are included to locate the transcription initiation sites of leu-500 and bla transcripts. The bla transcript serves as an internal control.
LeuO can also suppress AT4-mediated gene silencing, the effect of LeuO on AT4-mediated silencing of pbla was examined in an E. coli leuO Ϫ strain, MF1 (Fig. 5). A slightly stronger gene-silencing effect (ϳ88% reduction of the pbla activity) was found in the LeuO-free strain (Fig. 5A, compare lanes 1 and 2). When LeuO was provided in trans in MF1 from a coexisting expression vector, pEV101, AT4-mediated gene silencing was nearly abolished even without IPTG induction (Fig. 5A, lane 3). This was probably due to the leakage of LeuO from the expres-sion vector, pEV101. Such a leakage was evidenced from immunoblotting analysis (Fig. 5C, lane 2; 8.6 ng of LeuO was detected in the 100 g of total protein loaded). Upon further increase of cellular LeuO due to IPTG induction (Fig. 5C, lanes  3-6), AT4-mediated gene silencing was completely eliminated (Fig. 5A, lanes 4 -7). The pbla activity was fully restored with 50 M IPTG treatment (Fig. 5A, compare lanes 5 and 1). In a control experiment using the parental plasmid pAO, the pbla activity was unaffected by IPTG treatment (data not shown). These results indicate that LeuO negates AT4-mediated gene silencing.
However, the effect of LeuO on transcription could be nonspecific. To test whether or not the effect of LeuO is specific for AT4, a 72-bp DNA consisting of a DNA sequence from the lacZ coding region was synthesized and used to replace the 72-bp AT4 DNA in pWU802T (illustrated in Fig. 6). In the absence of LeuO (i.e. the absence of pEV101), this replacement resulted in a LeuO-independent pleu-500 activation on the mutant plasmid (Fig. 6B, lane 2). A significant ptac-mediated transcription activity was also detectable in the mutant plasmid (Fig. 6A,  lane 2). With the native 72-bp AT4 in place, a significantly reduced ptac activity was detected in pWU802T (Fig. 6A, lane  1). Apparently, the reduced ptac activity was too weak to activate pleu-500 at this distance (Fig. 6B, lane 1). Our study, thus far, has clearly indicated that LeuO relieves 318-bp intervening DNA-mediated repression (Fig. 1) by specifically negating AT4-mediated gene silencing.
The Effect of Adjacent Transcriptional Activity on AT4-mediated Gene Silencing-The 72-bp gene silencer, AT4, is located at the 5Ј-ends of the divergently arrayed leuO and leuABCD genes. Such a chromosomal organization may not be a coincidence, because transcription-generated negative DNA supercoiling is known to accumulate in such a topological domain (13,21). To test whether or not AT4 functions most effectively when placed between divergently arrayed promoters, we inserted AT4 at the unique AatII site in pBR322 DNA so that AT4 was flanked by the divergently arrayed bla and tetA genes (illustrated in Fig. 7). As expected, AT4 caused reductions in both pbla and ptetA activities (Fig. 7, A and B). However, the gene-silencing effect of AT4 in pBR322 (Fig. 7A, lanes 2) was lower than that in pAO-AT4 (Fig. 4). The anti-tet transcription activity (16) that read through the AT4 insert in pBR322 could be the reason for this reduction. Despite the reduced effect, AT4-mediated gene silencing on pBR322 was still additive in the presence of a tandemly repeated AT4 dimer (Fig. 7, A and  B, lanes 3). Due to the simultaneous gene-silencing effects on both flanking promoters, it remained unclear whether or not the bacterial gene silencer was affected by an adjacent transcription activity.
To test the effect of adjacent transcription on AT4-mediated gene silencing, pJW270-based plasmid constructs were used. pJW270-based plasmids are essentially the same as pBR322, except that the tetA promoter is replaced by an IPTG-inducible lacUV5 promoter and that an iq promoter-controlled laci gene was inserted at the 5Ј-end of the lacUV5 promoter (Refs. 17 and 21, and plasmid maps in Fig. 7). The opposite orientation of the laci gene in pJW270 and pJW270II was designed to test the effect of an adjacent transcription activity on AT4-mediated gene silencing. As expected, AT4-mediated gene silencing on pbla was observed in pJW270II-AT4 and pJW270II-AT4R when laci was transcribing away from the AT4 DNA inserts (Fig. 7C, lanes 4 -6). Strikingly, the gene-silencing effect was abolished when laci was inverted in pJW270-AT4 and pJW270-AT4R (Fig. 7C, lanes 1-3). With a location at the 3Ј-end of an adjacent transcription unit, AT4 not only exerted no genesilencing effect on the pbla, but the transcription activity of Primer extension was employed for detecting RNA transcripts initiated from ptac on pWU802T or pWU812T in a similar set of experiments as described in Fig. 1. The primer extension results were quantified and shown in arbitrary units. pbla was actually increased (Fig. 7C, lanes 2 and 3). This result demonstrates that a parallel (convergently) arrayed adjacent transcription unit abolished AT4-mediated gene silencing. Transcription-driven positive DNA supercoiling at the 3Ј-end of a transcription unit could be the reason why AT4 function was abolished at such a location. Another interpretation is that AT4 is functional only when it is located at the 5Ј-end of a transcription unit. For maximum gene-silencing activity, adjacent tran-scription activities should be transcribing away from the gene silencer. Read-through transcription activity and a 3Ј-end location of a transcription unit will either weaken or impair AT4-mediated gene silencing.
AT4-mediated Gene Silencing Is Effective Up to a 300-bp Distance-The chromosomal location of AT4 was centered at the Ϫ83 position of leuO and at the Ϫ351 position of the divergently arrayed leuABCD. In pBR322-AT4, AT4 was lo-FIG. 4. AT4 DNA sequence element is responsible for gene silencing. As illustrated, the whole or the various parts of the 318 AT-rich DNA was inserted in either orientation at the unique AatII site located at the Ϫ99 position of bla in pAO. Plasmids were assayed in CH582. Primer extension results were quantified and shown. All quantified results were compared with the pbla activity on pAO (A, lanes 1 or 7), which was arbitrarily set as 1. The identified AT4-mediated gene-silencing effect was tested in CH601(B). Again, pbla activity on pAO in either CH601 (lane 1) or CH582 (lane 3) was quantified and served as the standard for comparison with the pbla activity in pAO-AT4 in either CH601 (lane 2) or CH582 (lane 4), respectively.
FIG. 5. LeuO protein negates AT4mediated gene silencing. pAO or pAO-AT4 was assay for AT4-mediated gene silencing in E. coli leuO Ϫ strain, MF1. The presence of the coexisting pEV101 and the IPTG treatment is indicated with "ϩ" sign. The "Ϫ" sign indicates the absence of the treatment or coexisting plasmid in the experiment. 1.5-ml aliquots of bacterial culture from the IPTG titration experiment were saved at harvest and prepared for the total protein lysates. 100 g of total protein from each lysate was loaded along with the LeuO standard, 10  cated at the Ϫ135 position of bla and the Ϫ159 position of tetA. Thus it was unclear whether or not AT4 could directly silence pleuABCD at its more distal location (351 bp). If AT4-mediated gene silencing reaches a distance of 351 bp, LeuO may cause pleu-500 activation directly rather than indirectly via a subsequent short-range pleuO(ptac)-pleu-500 interaction. To address this issue, four promoterless DNAs consisting of DNA sequences from the coding region of lacZ were used to sequentially extend the distance between AT4 DNA insert and pbla in pAO-AT4 (illustrated in Fig. 8). The location of AT4 in pAO-AT4 was at the Ϫ135 position of bla. The four insertions resulted in plasmids with AT4 located at positions Ϫ188, Ϫ242, Ϫ296, and Ϫ350 of bla, respectively. Primer extension results indicated that AT4-mediated gene silencing was slightly reduced but remained effective up to a distance of 296 bp (Fig. 8,  lanes 3-5). The gene-silencing effect was completely abolished at the distance of 350 bp (Fig. 8, lane 6). DISCUSSION pleu-500 activation has been suggested to be regulated in a complex manner involving sequential promoter activation in the ilvIH-leuO-leuABCD gene cluster. Lilley and Higgins (26) have suggested that the transcriptional activity of leuO may be responsible for pleu-500 activation. The present study has experimentally demonstrated the importance of the leuO gene in pleu-500 activation. We have shown that both pleuO and the leuO gene product are required for pleu-500 activation. The function of LeuO is to reverse silencing mediated by the AT4 DNA sequence element, and the function of transcription from pleuO is to provide short-range promoter-promoter interaction for activation of pleu-500 (11).
The AT-rich DNA that is flanked by the divergently arrayed leuO and leuABCD is found in both S. typhimurium (Gen-Bank accession number AF106956) and E. coli (GenBank accession number AF 106955). These noncoding, AT-rich DNAs appear to be the regulatory regions for sequential gene activation of the ilvIH-leuO-leuABCD gene cluster (24). However, these AT-rich intervening DNAs show little DNA sequence homology (14). AT richness appears to be the only similarity between these DNAs. Previous studies have demonstrated that both the AT-rich DNA as well as the leuO gene are important for pleu-500 activation (6). However, the precise role(s) of the AT-rich DNA had been unclear. The present study has identified a 72-bp AT4 bacterial gene silencer in the AT-rich DNA.
How does a 72-bp A/T-rich DNA sequence element cause the transcriptional repression? AT4 is unlikely affecting transcription activity via a currently known regulatory mechanism, because bacterial transcription-negative regulation is usually functional at a short distance (e.g. within 100 bp). In fact, the binding site of the bacterial regulator often maps directly in the promoter region. For example, the lac repressor binding site (lac operator) overlaps with the Ϫ10 sequence of lac promoter (27). Hence, the bacterial repressor is usually gene-specific. Acting at a distance (e.g. 1000 bp) is possible via an intervening DNA looping mechanism. Two examples of this include DNA FIG. 6. AT4 represses the short-range interaction between ptac and pleu-500. pWU802T(AT4)R was derived from pWU802T by replacing the 72-bp AT4 DNA element with a promoterless neutral DNA of same size derived from the lacZ coding region. pWU802M(AT4)R was derived from pWU802T(AT4)R by deleting the 66-bp ptac. Plasmids were assayed in CH582, in which ptac activity is not required IPTG induction due to the lack of lac repressor in S. typhimurium. The transcription activities of both ptac and pleu-500 were simultaneously detected using primer extension (A and B).
7. An adjacent transcription activity affects AT4-mediated gene silencing. AT4-mediated gene silencing was analyzed by placing the AT4 DNA insert at the unique AatII site on pBR322, pJW270, and pJW270II, respectively. Both flanking pbla and ptetA activities on various pBR322 vectors in CH582 were detected using primer extension and shown in A and B, respectively. The constructs were: lane 1, pBR322 vector with no insert; lane 2, pBR322 with AT4 insert; lane 3, pBR322 with a tandemly repeated AT4 dimer. The pbla activity of pJW270 or pJW270II vector in CH582 with or without the AT4 insert in either orientation was detected by primer extension and shown in C.
looping-mediated transcriptional activation in nitrogen (ntr) regulation (28) and DNA looping-mediated transcriptional repression of the araBAD promoter (29). However, if DNA looping due to protein-protein interaction is important for AT4mediated gene silencing, the optimum distance for gene silencing should be about 500 bp (30). Using the lac operator as a model system, it has been experimentally demonstrated that, starting from a distance of ϳ150 bp, DNA looping mediated by repressor binding to the operator increased dramatically when the size of the intervening DNA increased. This effect peaked at a distance of 500 bp (30). In contrast, AT4 retained its transcriptional repressive effect up to a distance of 300 bp. The repressive effect was abolished at a distance longer than 350 bp (Fig. 8). Starting from a distance of 188 bp up to the 300-bp limit, the repressive effect was slightly reduced rather than dramatically enhanced as one might expect for the above-discussed protein-protein interaction at a distance scenario. This clear difference argues against the possibility that AT4 functions as a binding site for a repressor "acting at a distance," which represses transcription via a direct contact with the RNA polymerase complex at a distance.
What is the mechanism whereby AT4 silences an adjacent transcription activity at a distance of 300 bp? LeuO-mediated reversal effect (Fig. 5) has provided a possible clue. Overexpression or underexpression of LeuO has been linked to a number of hns Ϫ -associated phenotypes (23,31,32). Mizuno's group has shown that LeuO relieves bgl silencing in E. coli (23). Both H-NS and AT-rich DNA flanking the bgl promoter have been shown to be responsible for bgl silencing (33)(34)(35). In addition, using the pleu-500 activity as a reporter, we have shown genetically that H-NS plays a repressive role in the transcriptional regulation. 2 Together, these results suggest a possible involvement of H-NS in AT4-mediated gene silencing.
H-NS has been known to bind preferentially to curved DNA (36,37). Once H-NS is recruited to the local site (the AT-rich DNA sequence element, AT4), the binding cooperativity of H-NS may cause a cis-spreading (oligomerization) of H-NS to the promoter region. The H-NS oligomer may physically block RNA polymerase complex from accessing the promoter (Ϫ35 and Ϫ10 sequences). The binding cooperativity of H-NS may determine the size of the H-NS oligomer and hence the 300-bp distance limit of the AT4-mediated gene-silencing effect. The proposed mechanism is similar but distinct from the nucleoprotein filament model for the bacterial centromere site-mediated transcriptional silencing, which affects genes within several kilobases (38). Our model is also different from the DNA sequestration-mediated gene silencing model (39).
A similar direct transcriptional repressor role of H-NS was proposed by Ueguchi and Mizuno (40). They have shown in vitro that H-NS inhibits proV (proU)-mediated transcription by directly binding to the promoter region. The repressive H-NS complex is strikingly local and highly specific to the DNA sequence in the proV promoter, because H-NS does not affect transcription from ptac on the same DNA molecule (40). One or more cis-elements in the proV promoter must be responsible for H-NS recruitment to the local site. The 72-bp AT4 DNA may contain one or more similar elements that trigger H-NS localization in the proV promoter. Because DNA structure rather than the specific DNA sequence is known to be important for H-NS localization (36, 37 and reviewed in Ref. 41), no DNA sequence homology is expected between the promoters that utilize such a mechanism in their transcriptional silencing.
The idea that DNA structural elements could serve as signals for the formation of a transcriptional repressive nucleoprotein structure may also be applicable to explain the well known eukaryotic heterochromatic gene-silencing mechanism, because high AϩT composition and repetitiveness are the two common features for DNA structural elements involved in heterochromatin formation. The LINE-1 element in X chromosome inactivation (42 and reviewed in Ref. 43), the satellite DNAs in the centromeric-or telomeric-heterochromatin (reviewed in Ref. 44), and the silencers I and E in the yeast MAT loci (reviewed in Ref. 45) are all AT-rich. Combinations of any pair of the yeast silencers, HMR-E, HML-R, and HML-I, can result in inactivation of any gene activity flanked by the AT-rich DNA elements (46). However, the mechanistic link between DNA structural elements in the eukaryotic silencers and the formation of heterochromatin has been unclear. Detailed studies of the 72-bp AT-rich DNA-mediated gene silencing in bacteria could shed light on the mechanism of gene silencing in both prokaryotes and eukaryotes.