TI - Location of NarL binding sites at the dmsA promoter . AB - Under anaerobic conditions , dmsA expression is repressed approximately 10-fold by NarL when nitrate is present [4] . Although three putative NarL sites have been proposed based on their similarity to a NarL consensus recognition sequence , TACYYMT ( Y = C or T , M = A or C ) [24-26] , no in vivo or in vitro information is available regarding the location of the NarL sites within the dmsA regulatory region . To evaluate where NarL binds , DNase I footprinting experiments were performed using a dmsA promoter fragment corresponding to -127 to + 62 relative to the start of transcription at P1 . When NarL-phosphate was incubated with the coding strand of DNA , a 76 bp region was protected that extends from position -48 to +28 relative to the start of dmsA transcription ( Figure 2 , open rectangle ) . DNase I hypersensitive cutting sites were seen at positions +32 , +31 , +30 , +18 , +4 , -18 , -19 , -41 , and -53 relative to the start of transcription . The DNase I footprint of the non-coding strand of dmsA with PHOSphorylated NarL revealed an 83 bp protected region that extends from -51 to +32 relative to the start of transcription ( Figure 3 ) . DNase I hypersensitive cutting was observed at positions +27 , +4 , +3 , -10 , -20 , -21 , -22 and -32 . For the non-coding DNA strand , the size of the DNase I footprint pattern appears to increase when higher levels of NarL-phosphate are present . DNase I protection first occurs within the +10 to -15 region followed by an extension to the -25 to -50 region and then finally the +20 region . The size of the protected region did not change further when up to five-fold higher amounts of NarL-phosphate were used ( 10 muM , data not shown ) . Thus , the sizes of the protected regions on each strand concur . Furthermore , non-PHOSphorylated NarL gave no DNase I protections in the dmsA promoter region under the same conditions for either strand ( up to 10 muM ) , suggesting that NarL-phosphate can act as a transcriptional repressor for dmsA expression . The protections for both strands are consistent with the binding of multiple molecules of NarL to the DNA . Hydroxyl radical footprinting of the NarL interactions with dmsA promoter DNA was also performed for both strands of DNA ( Figure 2 and 3 ) . In hydroxyl radical footprinting , the small , highly reactive hydroxyl radical (*OH) attacks the deoxyribose sugars along the DNA backbone with no sequence or base specificity , thereby providing a high resolution of structural information [27,28] . A total of eight to nine distinct hydroxyl radical protected regions were observed of three to five basepairs in size for each DNA strand that extended over a 97 bp region ( Figure 4 ) . These NarL-phosphate protections extended from position -59 to +38 , consistent with the results of the DNase I footprinting experiments described above . The hydroxyl radical protected regions for the dmsA strands of DNA were offset by 3 bp in the 3' direction ( Figure 4 ) . This offset suggests that the NarL protein either occludes the minor groove of DNA , or that the DNA conformation is distorted upon NarL binding . In the former case , the DNA backbone sites located closest to one another are across the minor groove and separated in sequence by 3 bp [28-30] . The second proposal is supported by a recent 2.2 angstrom NarL-DNA structure for NarL-C-terminal domain complexed to a synthetic 7-2-7 NarL consensus binding site where the protein causes a conformational change of B-DNA to A-DNA ( Ann Maris , personal communication ) . Furthermore , the hydroxyl radical protected regions occur at ten bp intervals , a regular phasing of the helix repeat . This signifies that NarL-phosphate binds to only one side of the DNA molecule dmsA promoter region [27-30] . The hydroxyl radical data are also consistent with the binding of multiple molecules of NarL-phosphate to the dmsA promoter region as suggested by the DNase I data . Therefore , a simple model that accounts for the footprint data is the assembly of multiple NarL-phosphate molecules onto one face of the DNA that somehow protect the minor groove from hydroxyl radical attack [28-30] . In an alternative model , NarL binds only at the three proposed NarL heptamer consensus sites spaced at 20 bp intervals within the dmsA regulatory region ( Figure 4 ) . However , this model is difficult to envision since the DNase I and hydroxyl radical cleavage patterns extend over seven to nine turns of DNA . By either of the above models , the location of the NarL-phosphate protected regions suggests that NarL may compete with Fnr and/or RNA polymerase for occupancy on the DNA but only when the bacteria are grown anaerobically in the presence of nitrate , conditions where NarL is in the activated form . No hydroxyl radical or DNase I protected regions of DNA were observed when non-PHOSphorylated NarL protein was used at a concentration of 10 muM ( data not shown ) . In addition , beta-galactosidase assays revealed that the 10-fold nitrate dependent repression of dmsA-lacZ expression was unaffected by the deletion of upstream DNA sequence to -71 relative to the start of dmsA transcription , further pinpointing the location of the 5' end of the NarL recognition site for dmsA ( data not shown ) . Furthermore , the NarL footprint pattern does not extend into the dmsA P2 promoter region . Therefore , NarL does not appear to directly affect regulation at the P2 site , unless a large DNA/protein complex that involves multiple transcriptional regulators is involved ( ie Fnr , NarL , ModE and IHF in addition to RNAP ) . Future investigation of this complex regulatory region will be needed to ascertain such a matter . Finally , a similar hydroxyl radical footprint pattern of 8-9 protected regions of 3-4 bp spaced 10 nucleotides apart was also observed for NarL-phosphate at the promoter region of the frdA gene , another anaerobically induced gene that is repressed by NarL in the presence of nitrate ( data not shown ) . To establish if the entire NarL protected region is required for NarL-phosphate to bind DNA , a DNA fragment ( designated Fragment B , Figure 5 ) containing a truncated region of the dmsA regulatory sequence was constructed . The fragment extends from position -127 to -13 relative to the start of dmsA transcription at P1 ( Figure 4 ) . In Fragment B , two of the three consensus heptamer sites have been replaced by the multi-cloning region of pGEM-11Zf ( Methods ) . When examined by DNase I footprinting analysis , the altered dmsA Fragment B ( Figure 5 , lanes 6 to 9 ) revealed a 38 bp NarL-phosphate protected region extending from position -51 to -13 . This protected region spans only the wild-type dmsA DNA sequences but not the adjacent foreign DNA sequences . In contrast , the full-length dmsA fragment ( Fragment A , -127 to +62 ) showed a larger protected region from -51 to +32 ( lanes 2-5 ) . These findings demonstrate that the smaller dmsA region containing only one of the three consensus heptamer sites ( Figure 4 ) is sufficient for NarL binding . However , a somewhat weaker binding of NarL-phosphate to the DNA fragment containing the truncated dmsA region relative to the full-length region may suggest that NarL binds at the promoter in a weakly cooperative fashion . The protections are consistent with the proposal that NarL-phosphate recognizes and binds at multiple heptamer recognition sites within the dmsA P1 promoter region . Three putative NarL binding sites with the consensus heptameric sequence (TACYYMT) have been proposed for dmsA[26] . These sites , previously assigned at positions +8 , -14 and -34 , are now centered at positions + 15 , -7 , and -27 ( Figure 4 ) due to the reassignment of the dmsA P1 start site [5,6] . Since the size of the DNase I and hydroxyl radical footprints in this study show DNA protections between and beyond these three consensus sites , other NarL binding sites may be present in this region . As the three consensus NarL boxes flank the dmsA promoter and are spaced 20 bp apart ( Figure 4 ) , the spacing and orientation of the NarL protected regions make it tempting to speculate that NarL-phosphate binds at each site . Additional NarL-phosphate monomers then assemble on the DNA to form a stable DNA protein complex . Alternatively , molecules of NarL-phosphate may bind at both the consensus and at related NarL-box sequences that contain one or two mismatches from consensus ( Figure 4 ) . Inspection of the DNA reveals an additional NarL box within the protected region that has one mismatch from consensus ( Figure 4 , dashed arrows ) and nineteen NarL-like boxes with two mismatches ( dotted arrows , not all shown ) . We note that none of the three consensus heptamers ( solid arrows , Figure 4 ) are arranged in a 7-2-7 sequence , nor are any of the consensus heptamers paired with any of the mismatch heptamers in such an arrangement . This is noteworthy since a 7-2-7 sequence has been speculated for nucleating NarL interactions at other promoters [31] . Stoichiometry experiments are planned to ascertain the number of NarL molecules that bind the dmsA promoter region , as are studies to mutagenize one or more of the NarL binding sites to determine the importance of the NarL consensus binding sites at the dmsA promoter .