Trp RNA-Binding Attenuation Protein-5 Stem-Loop RNA Interaction Is Required for Proper Transcription Attenuation Control Of

Home , Terminator (genetics), Trp operon

trp RNA-Binding Attenuation Protein-5Ј Stem-Loop RNA Interaction Is Required for Proper Transcription Attenuation Control of the Bacillus subtilis trpEDCFBA Operon

HANSEN DU,† ALEXANDER V. YAKHNIN, SUBRAMANIAN DHARMARAJ,‡ AND PAUL BABITZKE* Department of Biochemistry and Molecular Biology, The Pennsylvania State University, University Park, Pennsylvania 16802

Received 25 October 1999/Accepted 10 January 2000

The trp RNA-binding attenuation protein (TRAP) regulates expression of the Bacillus subtilis trpEDCFBA operon by a novel transcription attenuation mechanism. Tryptophan-activated TRAP binds to the nascent trp leader transcript by interacting with 11 (G/U)AG repeats, 6 of which are present in an antiterminator structure. TRAP binding to these repeats prevents formation of the antiterminator, thereby promoting formation of an overlapping intrinsic terminator. A third stem-loop structure that forms at the extreme 5؅ end of the trp leader transcript also plays a role in the transcription attenuation mechanism. The 5؅ stem-loop increases the afﬁnity of TRAP for trp leader RNA. Results from RNA structure mapping experiments demon- ,strate that the 5؅ stem-loop consists of a 3-bp lower stem, a 5-by-2 asymmetric internal loop, a 6-bp upper stem and a hexaloop at the apex of the structure. Footprinting results indicate that TRAP interacts with the 5؅ stem-loop and that this interaction differs depending on the number of downstream (G/U)AG repeats present -in the transcript. Expression studies with trpE؅-؅lacZ translational fusions demonstrate that TRAP-5؅ stem loop interaction is required for proper regulation of the trp operon. 3؅ RNA boundary experiments indicate that .the 5؅ structure reduces the number of (G/U)AG repeats required for stable TRAP-trp leader RNA association Thus, TRAP-5؅ stem-loop interaction may increase the likelihood that TRAP will bind to the (G/U)AG repeats in time to block antiterminator formation.

Expression of the Bacillus subtilis tryptophan biosynthetic While it is not known how TRAP initially interacts with the genes is regulated in response to changes in the intracellular nascent trp leader transcript, the interaction must occur quickly level of tryptophan by the trp RNA-binding attenuation protein to prevent formation of the antiterminator structure. During (TRAP) (4, 16). The trpEDCFBA operon is regulated by attenuation regulation of the Escherichia coli trp operon, tran- TRAP-mediated transcription attenuation (5, 10, 17, 19, 24, scriptional pausing allows the regulatory ribosome to bind to 25) and translational control mechanisms (14, 19, 22). TRAP the leader transcript at an appropriate time (20). Since leader also regulates expression of the unlinked trpG gene at the peptide synthesis is not involved in transcription attenuation of translational level (8, 15, 30). TRAP exists as a complex con- the B. subtilis trp operon, nor has RNA polymerase pausing sisting of 11 identical subunits arranged in a single ring termed been demonstrated to play a role in this regulatory mechanism, the ␤-wheel (1, 3). Tryptophan cooperatively activates TRAP we were interested in determining if any factor besides TRAP by binding between every adjacent TRAP subunit (3, 6). and the (G/U)AG repeats were involved in TRAP interaction The 203-nucleotide untranslated trp operon leader transcript with the nascent trp leader transcript. can fold into three distinct RNA secondary structures that We recently demonstrated that, in addition to the antiter- participate in transcription attenuation (Fig. 1). When TRAP minator and terminator, an RNA structure predicted to form is activated by tryptophan, 11 KKR motifs that outline the at the extreme 5Ј end of the nascent trp leader transcript periphery of the TRAP complex can bind to 11 closely spaced participates in the transcription attenuation mechanism (28). (G/U)AG repeats present in the nascent trp leader transcript, Deletion or disruption of this putative structure resulted in a thereby wrapping the RNA around the periphery of the TRAP dramatic increase of trp operon expression in vivo and in- complex (2, 8, 31). TRAP binding blocks formation of the creased transcriptional readthrough in vitro. This previous antiterminator since six of the (G/U)AG repeats are present study also demonstrated that the 5Ј stem-loop functions pri- within this RNA structure (5, 8). Thus, TRAP binding pro- marily in TRAP-dependent regulation of the trp operon and motes formation of the overlapping intrinsic terminator which that overexpression of TRAP suppressed the defect associated results in transcription termination before RNA polymerase with the 5Ј stem-loop deletion. Moreover, we showed that the can reach the trp operon structural genes. In the absence of presumed 5Ј structure increased the afﬁnity of TRAP for trp TRAP binding, formation of the antiterminator permits tran- leader RNA (28). Thus, it was possible that the 5Ј stem-loop scription of the entire operon (5). participated in the attenuation mechanism by interacting with TRAP. In the present study we determined the secondary structure * Corresponding author. Mailing address: Department of Biochem- of the 5Ј stem-loop and found that TRAP interacts with this istry and Molecular Biology, The Pennsylvania State University, Uni- structure. We also established that the 5Ј stem-loop reduces versity Park, PA 16802. Phone: (814) 865-0002. Fax: (814) 863-7024. the number of (G/U)AG repeats required for stable TRAP-trp E-mail: [email protected]. Ј † Present address: Department of Biology, MS008, Brandeis Uni- leader RNA association and that the TRAP-5 stem-loop in- versity, Waltham, MA 02454. teraction differs depending on the number of downstream (G/ ‡ Present address: Ambion, Inc., Austin, TX 78744. U)AG repeats that are present in the transcript. Our results

1819 1820 DU ET AL. J. BACTERIOL.

Plasmid pHD34 contains the trp promoter and nucleotides 6 to 203 ⌬(ϩ1toϩ5) of the trp leader. This plasmid was constructed by a two-step process using overlap extension PCR. The final PCR product was digested with EcoRI and HindIII and subcloned into the EcoRI and HindIII sites of PTZ18U. pHD40 carries the trp promoter and a mutant trp leader in which nucleotides 6 to 9 were replaced with a T residue, while pHD46 carries the trp promoter and a leader containing nucleotides 16 to 203 ⌬(ϩ1toϩ15). Both of these plasmids were constructed in the same manner as pHD34. The B. subtilis integration vector, ptrpBG1-PLK, used for the generation of trpEЈ-ЈlacZ translational fusions was described previously (22). The plasmids pHD52, pHD53, and pHD54, which contain trpEЈ-ЈlacZ fusions, were constructed by subcloning the trp promoter and leader region from pHD34, pHD40, and pHD46 into the EcoRI and HindIII sites of the ptrpBG1-PLK polylinker, respectively. The three plasmids pHD52, pHD53, and pHD54 were linearized with SalI and separately integrated into the amyE locus of B. subtilis W168. The resulting strains are PLBS138, PLBS139, and PLBS140. ␤-Galactosidase assay. Cells were cultured in minimal Spizizen salts medium (27) containing 0.2% acid-hydrolyzed casein, 0.2% glucose, and 5 ␮g of chlor- amphenicol per ml in the presence or absence of 50 ␮g of tryptophan per ml. Cells were harvested in mid-exponential phase, and cell suspensions were prepared as previously described (28). ␤-Galactosidase activity was subsequently assayed by the method of Miller (23). FIG. 1. Nucleotide sequence of the B. subtilis trp leader transcript showing In vitro transcription. Gel-purified transcripts used in this analysis were syn- Ј the 5 stem-loop and the mutually exclusive antiterminator and terminator struc- thesized by using the Ambion MEGAscript in vitro transcription kit. Templates tures. Boxed nucleotides mark overlapping segments of the competing secondary consisted of various plasmids that had been linearized with BamHI or HindIII. structures. The (G/U)AG repeats known to be involved in TRAP-RNA recog- 5Ј-End-labeled RNAs were generated by treating in vitro-generated transcripts nition are indicated by boldface type. Numbering is from the start of transcrip- with calf intestinal phosphatase and subsequently with polynucleotide kinase and tion. RNA secondary structure predictions were performed using MFOLD (29, ␥ 32 Ј [ - P]ATP. The unlabeled and labeled RNA was gel purified as previously 32). Note that the 5 stem-loop is modified from the structure predicted by described (14). MFOLD due to the RNA secondary structure mapping data obtained during the Gel mobility shift assay. The binding affinity between TRAP and trp leader course of these studies. RNA was estimated by using gel mobility shift assays by modifying a previously published procedure (28). TRAP was purified as described earlier (5). Tran- scripts used in the analysis were generated from pPB77 (wild type), pPB310 (5Ј Ј stem-loop deletion), or pHD68 (5Ј stem-loop only) that had been linearized with suggest that the TRAP-5 stem-loop interaction increases the BamHI. Binding reactions (8 ␮l) containing 0.2 nM 5Ј-end-labeled RNA, various probability that TRAP will bind to the (G/U)AG repeats be- concentrations of TRAP (TRAP excess), 1 mM tryptophan in 50 mM Tris- fore the antiterminator can form, thereby increasing the like- acetate (pH 8.0), 4 mM magnesium acetate, 5 mM dithiothreitol, 10% glycerol, lihood that transcription termination occurs before RNA poly- 0.2 mg of E. coli tRNA per ml, 0.1 mg of xylene cyanol per ml, and 400 U of merase can reach the trp operon structural genes. RNasin (Promega) per ml were incubated at 25°C for 20 min. Aliquots of reaction mixtures were fractionated through native polyacrylamide gels containing 375 mM Tris-HCl (pH 8.8), 5% glycerol, and 1 mM EDTA. Electrophoresis MATERIALS AND METHODS was performed at room temperature in running buffer containing 25 mM Tris- glycine (pH 8.3) and 1 mM EDTA. Gels were dried, and the bound and free Bacterial strains and plasmids. All of the B. subtilis strains used in this study RNA bands were quantified by using a PhosphorImager (Molecular Dynamics) are listed in Table 1. The plasmids pTZ18U (Stratagene) and pPB77, pPB78, and the ImageQuant software package. Modifications of the standard reaction pPB82, and pPB83 (8) have been described. Plasmid pPB310 contains nucleo- are described in the text or the appropriate figure legend. The binding data were tides 32 to 111 of the B. subtilis trp leader region and was constructed by PCR. ϭ ϩ fit to the simple binding equation: RNAb a[TRAP]f/(Kd [TRAP]f), where a The resulting PCR product was digested with EcoRI and BamHI and subcloned is the maximal fraction of bound RNA (RNAb) that is approximately equal to 1; into the EcoRI and BamHI sites of the pTZ18U polylinker. Plasmid pHD55, Kd is defined as the concentration of free protein, [TRAP]f, at which the RNAb which contains nucleotides 1 to 36 of the B. subtilis trp leader, was also con- reaches 50% saturation; RNAb is the fraction of RNA bound between 0 and 1; structed by PCR. In this case the PCR product was digested with EcoRI and KpnI and [TRAP]f is the concentration of free TRAP 11-mer which was assumed to be and ligated into the EcoRI and KpnI sites of the pTZ18U polylinker. Plasmid the concentration of total TRAP added since the total TRAP concentration was pHD68 was constructed by digesting pHD55 with EcoRI and treating it with in at least 12-fold molar excess over RNA. mung bean nuclease to remove the cohesive ends followed by self-ligation. RNA structure mapping. RNA structures were predicted by using the MFOLD program (29, 32). RNA structure mapping using unlabeled transcripts followed previously published procedures (14). The unlabeled transcripts used in TABLE 1. B. subtilis strains used in this study this analysis were generated from pPB83 linearized with HindIII as template. Titrations of RNases and chemical reagents were routinely performed to deter- Source or mine the amount of each reagent that would prevent multiple cleavages or Strains Genotypea reference chemical modifications in any one transcript so that we could minimize the potential of secondary rearrangements in short RNA segments. RNA samples c W168 Prototroph BGSC were partially digested with RNase T1 (Gibco-BRL) or RNase V1 (Pharmacia) PLBS44 amyE::[trpP(Ϫ412 to ϩ203) 28 and recovered as described earlier (14). CMCT and DMS modification reactions, trpEЈ-ЈlacZ Cmr] as well as the subsequent recovery of RNA samples, followed a previously PLBS104 amyE::[trpP(Ϫ412 to ϩ203) 28 published procedure (14). RNA samples were resuspended in primer extension buffer and hybridized to a ␥-32P-end-labeled primer, and the primers were ⌬(ϩ3toϩ32)trpEЈ-ЈlacZ Cmr] Ϫ ϩ extended with Moloney murine leukemia virus (MMLV) reverse transcriptase PLBS138 amyE::[trpP( 412 to 203) This study (U.S. Biochemicals) as described elsewhere (14). After 10 min at 42°C, reactions r ⌬(ϩ1toϩ5)trpEЈ-ЈlacZ Cm ] were terminated by the addition of 3 ␮l of standard sequencing stop solution. PLBS139 amyE::[trpP(Ϫ412 to ϩ203) This study Samples were fractionated through 6% denaturing polyacrylamide gels. Control ⌬(ϩ6toϩ9)Tb trpEЈ-ЈlacZ Cmr] sequencing reactions were carried out using the same plasmids and end-labeled PLBS140 amyE::[trpP(Ϫ412 to ϩ203) This study primer as described above. Ј ⌬(ϩ1toϩ15)trpEЈ-ЈlacZ Cmr] 5 -end-labeled RNA (see above) generated from various templates (pPB77, pPB78, pPB82, pPB83, or pHD68 digested with BamHI) was renatured by a trpP denotes the trp promoter. Prime indicates truncation of the gene. Ϫ412 heating at 95°C for 1 min, followed by a 10-min incubation at 37°C. RNA was to ϩ203 indicates the DNA fragment containing the trp promoter and neighbor- digested with 0.07 U of RNase V1 per ml for 10 min at 37°C in 40 mM Tris-HCl ⌬ ing regions that was incorporated relative to the transcription start site. “ ” (pH 8.0)–250 mM KCl–4 mM MgCl2 (TKM buffer). Samples were fractionated designates the portion of the leader region that was deleted. through 6% denaturing polyacrylamide gels. The G sequencing ladder was gen- b T was substituted for nucleotides 6 to 9. erated by partial RNase T1 digestion under denaturing conditions as described c BGSC, Bacillus Genetic Stock Center, Ohio State University, Columbus, previously (11). Alkali digestion ladders were prepared as described elsewhere Ohio. (13) from the same end-labeled transcripts. VOL. 182, 2000 TRAP-5Ј STEM-LOOP RNA INTERACTION 1821

-3؅-boundary analysis. The 3Ј-boundary analysis followed a published proce dure (11). 5Ј-end-labeled transcripts (see above) generated from pPB77 (wild- type trp leader) or pPB310 (5Ј stem-loop deletion trp leader) were treated with alkali to generate an RNA ladder. Then, 100-␮l RNA samples (10 pmol) were incubated for 5 min at 95°C in alkaline hydrolysis buffer (100 mM NaHCO3- ␮ ␮ Na2CO3 [pH 9.0]–2 mM EDTA–0.5 gofE. coli tRNA per l) and then recovered by ethanol precipitation. Hydrolyzed RNAs were mixed with 50 ␮gof TRAP and incubated at 25°C for 20 min in TKM buffer. The reaction mixtures were fractionated through 6% native polyacrylamide gels. Bound and unbound transcripts were visualized by autoradiography, excised from the gel, and subsequently eluted from the gel. RNAs were ethanol precipitated and fractionated through 6% denaturing polyacrylamide gels. RNase T1 and alkali digestion ladders of the same 5Ј-end-labeled transcripts were used as molecular size stan- dards.

RESULTS

Gel mobility shift analysis of TRAP and trp leader RNA. Results from previous in vivo experiments demonstrated that overexpression of mtrB, the gene encoding TRAP (17), suppressed the defect associated with deletion of the 5Ј stem-loop (28). Using gel mobility shift assays we further showed that the 5Ј stem-loop increases the affinity of TRAP for trp leader RNA approximately fivefold (28). In the previous study (28) we observed a TRAP-dependent band that migrated just behind the free RNA. We assumed that this band resulted from TRAP-trp leader RNA complex dissociation soon after loading the gel. When we repeated the analysis using a modified gel shift procedure (see Materials and Methods) the presence of this band was eliminated, confirming our previous assumption. As was previously observed (28), the presence of the 5Ј structure in transcripts that contained all 11 (G/U)AG repeats increased the affinity of TRAP for trp leader RNA (Fig. 2). Binding to the Ј wild-type trp leader transcript was detectable at 2.5 nM TRAP FIG. 2. Gel mobility shift analysis of TRAP complexed with wild-type or 5 stem-loop deletion trp leader transcripts. 5Ј-end-labeled trp leader transcripts and saturated at approximately 320 nM TRAP (Fig. 2A). With (0.2 nM) were incubated with 1 mM tryptophan and the concentration of TRAP the 5Ј stem-loop deletion transcript, comparable binding was indicated at the bottom of each lane (nanomolar). Each transcript contained the detected at 5 nM TRAP but did not reach saturation even at a 11 (G/U)AG repeats between nucleotides 36 and 91. Bands corresponding to ء concentration of 1.28 ␮M TRAP (Fig. 2B). In each case we free (F) and bound (B or ) RNA are indicated on the left. (A) Wild-type trp leader transcripts. (B) 5Ј stem-loop deletion trp leader transcripts. observed a prominent shifted complex. Note that we also observed two additional shifted complexes for each of these tran- while the other ,(ء) scripts. One of these complexes is shown extremely faint complex is not. Note that these complexes were RNA interaction does not involve the KKR motifs known to not observed in our previous study (28). While the most prom- interact with the (G/U)AG repeats (2, 31). -inent shifted species probably consists of complexes containing 5؅ stem-loop structure mapping. To determine if the pre one TRAP 11-mer bound to a single trp leader transcript, the dicted 5Ј structure actually formed in the trp leader transcript, composition of the other shifted species is not known. We fit we probed the structure of a transcript containing the first 68 these data to a simple binding equation by using nonlinear nucleotides of the trp leader in vitro with structure-specific least-squares analysis. This method yielded estimated Kd val- enzymatic and chemical reagents. This transcript contained the ues of 26 Ϯ 5 nM TRAP for the wild-type transcript and 280 Ϯ predicted 5Ј stem-loop and the first six (G/U)AG repeats 50 nM for the 5Ј stem-loop deletion transcript. The small known to interact with TRAP, as well as four upstream and difference in these values from those observed previously (28) downstream nucleotides derived from the vector. Note that probably reflects the different binding and gel-running condi- computer predictions indicated that these additional residues tions used in the current study. do not interfere with 5Ј stem-loop formation. trp leader tran- The finding that the 5Ј stem-loop increased the affinity of scripts were subjected to partial digestion or chemical modifi- TRAP for trp leader RNA approximately 10-fold suggested cation using RNase T1, RNase V1, DMS, or CMCT. The sites that the 5Ј stem-loop interacted with TRAP. When we per- of nuclease cleavage or chemical modification were mapped by formed gel shift experiments with transcripts derived from primer extension using an end-labeled primer and MMLV plasmid pHD68 that only contained the 5Ј stem-loop, we did reverse transcriptase. Cleavage or chemical modification not detect any evidence of TRAP binding (data not shown). would give rise to a primer extension product one nucleotide We also performed RNA competition experiments with wild- shorter than the corresponding band in the sequencing lane. type trp leader transcripts and transcripts that only contained The results of the structure mapping experiments are shown the 5Ј stem-loop. While the unlabeled wild-type trp leader in Fig. 3 and summarized in Fig. 4. The computer-predicted transcript was able to compete for TRAP binding to labeled structure of the 5Ј stem-loop is identical to the experimentally wild-type and 5Ј stem-loop deletion trp leader transcripts, the determined structure except that U5 and A29 are predicted to Ј transcript that only contained the 5 stem-loop only competed pair as are A16 and U21. RNase T1 cleaves following unpaired ء away the higher-shifted complexes ( ) (data not shown). Since G residues. We observed prominent RNase T1 cleavage fol- the RNA that only contained the 5Ј stem-loop was an ineffec- lowing the G residues at positions 18, 20, 38, and 42, indicating tive competitor, these results suggest that TRAP–5Ј stem-loop that these residues are single stranded. Note that bands cor- 1822 DU ET AL. J. BACTERIOL.

FIG. 4. Summary of the 5Ј stem-loop structure mapping results. This figure is adapted from the data presented in Fig. 3. Positions of cleavage by the single- stranded probe RNase T1 are indicated by arrows. Positions of cleavage by the double-stranded probe RNase V1 are indicated by arrowheads. Positions of RNA modification using the single-stranded probe DMS (circles) or CMCT (squares) are also indicated. Filled arrowheads, circles, or squares indicate strong modification or cleavage, whereas open arrowheads, circles, or squares indicate weak modification or cleavage. Numbering is from the start of transcription.

CMCT modifies unpaired G and U residues at the N1 and N3 positions, respectively. These DMS- and CMCT-modified residues are unable to serve as templates for reverse transcriptase. We observed DMS signals at the A and C residues corresponding to positions 1, 6, 8, 9, 16, 17, 19, 28, 29, 30, 32, 33, 34, 37, 39, and 40, suggesting that these residues are single stranded. FIG. 3. 5Ј stem-loop structure mapping. RNA containing nucleotides 1 to 68 However, the relatively weak DMS signals at positions 1, 30, of the trp leader transcript was used in this analysis (Fig. 1). trp leader RNA was and 32 suggest that these residues can be paired or unpaired. treated with RNase T1, RNase V1, DMS, or CMCT. Residues that were cleaved The absence of DMS modification of the remaining A and C by RNase T1 or RNase V1 or modified by DMS or CMCT were detected by residues suggests that these residues are base paired. With the primer extension by using MMLV reverse transcriptase. The mock-treated control lane without enzymatic or chemical treatment is indicated. Note that the exception of A16 and A29, these results are consistent with the bands observed in the treated lanes are one nucleotide shorter than the corre- predicted structure (Fig. 3 and 4). We observed prominent sponding bands in the A, C, G, or U sequencing lanes. The positions of the CMCT signals at the U and G residues corresponding to po- nucleotides corresponding to the lower stem, the internal loop, the upper stem, sitions 5, 7, 18, 20, 21, 35, 36, 38, and 41, indicating that these the hexaloop, and the first (G/U)AG repeat (UAG) are indicated at the right. Numbering at the left corresponds to the DNA sequencing ladder and is from the residues are single stranded. The absence of CMCT modifica- start of transcription. tion of the remaining U and G residues suggests that these nucleotides are base paired (Fig. 3 and 4). With the exception of G7, which was not cleaved by RNase T1, the CMCT results are consistent with the other reagents tested. Taken together, responding to the G residues at positions 2, 7, 22, 24, and 31 the results of the structure mapping experiments are consistent were not detected, suggesting that these residues were base with the structure shown in Fig. 4, although it appears that the paired (Fig. 3 and 4). With the exception of G7, these results lower stem is relatively unstable. The structure that we deter- are consistent with the computer predicted secondary struc- mined differs from the predicted structure by two base pairs. ture. RNase V1 is generally specific for base-paired residues; The predicted U5-A29 and the A16-U21 base pairs were not however, this enzyme does not cleave all paired residues, and detected in the 5Ј stem-loop secondary structure, suggesting it sometimes cleaves the first few bases in a single-stranded the existence of a larger asymmetric internal loop and hairpin RNA segment that is adjacent to an RNA duplex (26). We loop, respectively (Fig. 4). When taken together, our structure Ј observed prominent RNase V1 cleavage following A1, U4, mapping results indicate that the 5 stem-loop consists of a U11, A12, and U23, as well as weak RNase V1 cleavage fol- 3-bp lower stem, a 5-by-2 asymmetric internal loop, a 6-bp lowing C3, A10, G24, C32, and A33, suggesting that these upper stem, and a hexaloop at the apex of the structure. residues are base paired (Fig. 3 and 4). These results are TRAP interacts with the 5؅ stem-loop. Our gel shift analysis consistent with the predicted 5Ј stem-loop structure. Note that indicated that the 5Ј stem-loop increases the affinity of TRAP the cleavage of A1 and A33 is likely due to their position for trp leader RNA but provided little evidence that TRAP immediately adjacent to the lower stem of the structure. interacts with the 5Ј structure. We performed TRAP-trp leader To determine the structure of the 5Ј stem-loop more pre- RNA footprint experiments to determine if TRAP interacts cisely, chemical modification experiments with DMS and with the 5Ј stem-loop (Fig. 5). We used the same in vitro- CMCT were carried out. DMS methylates N1 of adenine and generated trp leader transcript (positions 1 to 68), chemical N3 of cytosine when the residues are single stranded, whereas and enzymatic probes, and 5Ј-end-labeled primer used for the VOL. 182, 2000 TRAP-5Ј STEM-LOOP RNA INTERACTION 1823

TABLE 2. Effect of 5Ј stem-loop mutations on trp operon expression

␤-Gal activityb 5Ј stem-loop ␤-Gal ratio Strain (Miller units) mutation (ϪTrp/ϩTrp) ϩTrp ϪTrp PLBS44 Wild type 0.2 Ϯ 0.05 52 Ϯ 5 260 PLBS104 ⌬(ϩ3toϩ32) 12 Ϯ 0.5 274 Ϯ 11 23 PLBS138 ⌬(ϩ1toϩ5) 0.4 Ϯ 0.07 75 Ϯ 2 188 PLBS139 ⌬(ϩ6toϩ9) 4.2 Ϯ 0.4 190 Ϯ 15 45 replaced with Ua PLBS140 ⌬(ϩ1toϩ15) 3.7 Ϯ 0.2 152 Ϯ 12 41

a Nucleotides 6 to 9 (AGAA) were replaced by U in the trp leader transcript. b ␤-Galactosidase (␤-Gal) activity expressed from the trpEЈ-ЈlacZ fusion is given in Miller units (23).

A37 from DMS methylation, whereas bound TRAP enhanced modification of A1 (Fig. 5). In the case of CMCT, bound TRAP protected G7, G18, G20, U21, and G39 from CMCT modification, whereas modification of U4 was enhanced when TRAP was bound. It should be pointed out that the results with CMCT and RNase T1 are not in agreement. Whereas TRAP protected G18 and G20 from CMCT modification, bound TRAP did not significantly protect either of these residues from RNase T1 cleavage. The reason for this discrepancy is unknown. When taken together, the footprinting results are consistent with a TRAP-5Ј stem-loop RNA complex containing the internal loop, the upper stem, the hexaloop, and the 3Ј FIG. 5. TRAP–5Ј stem-loop RNA footprint. RNA containing nucleotides 1 to 68 of the trp leader transcript was used in this analysis (Fig. 1). trp leader RNA side of the lower stem. Note that in no case did we detect ϩ was treated with RNase T1, RNase V1, DMS, or CMCT in the presence ( )or TRAP binding to the trp leader transcript in the absence of Ϫ absence ( ) of bound TRAP. Residues that were cleaved by RNase T1 or RNase tryptophan (data not shown). V1 or modified by DMS or CMCT were detected by primer extension by using TRAP interaction with the 5؅ stem-loop is required for MMLV reverse transcriptase. The mock-treated control lane without enzyme or Ј chemical treatment is indicated. Note that the bands observed in the treated proper regulation of the trp operon. The TRAP–5 stem-loop lanes are one nucleotide shorter than the corresponding bands in the A, C, G, or footprint results suggested that TRAP does not interact with U sequencing lanes. The positions of the nucleotides corresponding to the lower the 5Ј side of the lower stem. To determine if the lower stem stem, the internal loop, the upper stem, the loop, and the first (G/U)AG repeat is important for 5Ј stem-loop function, we deleted the DNA (UAG) are indicated on the right. The numbering on the left corresponds to the DNA sequencing ladder and is from the start of transcription. region corresponding to the first five nucleotides of the trp leader transcript. We examined B. subtilis strains containing trpEЈ-ЈlacZ translational fusions that were controlled by the wild type (WTtrpL), the 5Ј stem-loop deletion ⌬(ϩ3toϩ32), structure-mapping experiments (see above). The cleavage pat- or the ⌬(ϩ1toϩ5) trp leader and analyzed ␤-galactosidase tern with RNase V1 was dramatically altered when TRAP was expression when each strain was grown in the presence or bound to the trp leader transcript. Bound TRAP reduced or absence of exogenous tryptophan. We observed minimal ex- Ј prevented RNase V1 cleavage at every 5 stem-loop residue pression in the WTtrpL strain PLBS44 grown in the presence that was cleaved in the absence of TRAP (Fig. 5). Since RNase of tryptophan (Table 2). The effect of exogenous tryptophan Ј Ј V1 is generally specific for double-stranded RNA, these results on the expression of the WTtrpL trpE - lacZ fusion can be suggested that TRAP bound to the 5Ј structure and prevented assessed from the ϪTrp/ϩTrp ratio, which was 260. Compa- ⌬ ϩ ϩ cleavage. In sharp contrast, the RNase T1 cleavage pattern rable experiments were performed with the ( 3to 32) within the 5Ј stem-loop was only slightly altered when TRAP strain PLBS104 and the ⌬(ϩ1toϩ5) strain PLBS138. As was was bound to the transcript (Fig. 5). Interestingly, the RNase previously observed (28), deletion of the entire 5Ј stem-loop T1 cleavage pattern suggests that the GAG sequence in the resulted in a dramatic increase in expression, especially when loop of the 5Ј structure does not interact with a TRAP KKR cells were grown in the presence of tryptophan. In this case the motif (Fig. 4 and 5). Previous results demonstrated that both G ϪTrp/ϩTrp ratio was only 23, significantly lower than that residues in GAG repeats are strongly protected from RNase observed for the WTtrpL strain (Table 2). Interestingly, the ⌬ ϩ ϩ T1 cleavage by bound TRAP (8, 15). This finding is consistent expression levels of the ( 1to 5) strain were similar to the with a previous in vivo study where it was determined that wild-type strain (Table 2), indicating that these residues are changing this sequence to GUG had little effect on trp operon not required for 5Ј stem-loop function (Table 2). This result is expression (28). Note that, with the exception of the first UAG consistent with the footprint analysis (Fig. 5). repeat, bound TRAP prevented or reduced cleavage of the G The footprint results presented above also indicated that the residues in the (G/U)AG repeats that were previously shown 5Ј side of the asymmetric internal loop is involved in TRAP-5Ј to interact with TRAP (data not shown) (18). stem-loop interaction. We replaced nucleotides 6 to 9 As was observed for RNase V1 cleavage, the DMS and (AGAA) of the internal loop with a single U residue. The CMCT RNA modification patterns were significantly altered predicted structure of this mutant transcript contained a con- when TRAP was bound to the trp leader transcript. We found tiguous 12-bp stem without an internal loop (structure not that bound TRAP protected A8, A9, A19, A30, C32, A34, and shown). We examined the effect of this trp leader mutation on 1824 DU ET AL. J. BACTERIOL. expression of a trpEЈ-ЈlacZ translational fusion (PLBS139). Compared to the wild-type strain, we found that ␤-galactosidase levels increased 20-fold when this strain was grown in the presence of tryptophan and 4-fold in its absence (Table 2). In this case the ϪTrp/ϩTrp ratio was 45, only twofold higher than that observed for the strain in which the entire 5Ј stem-loop was deleted. This result indicates that the 5Ј side of the asymmetric internal loop is important for 5Ј stem-loop function, which again is consistent with the footprint results. We also examined the effect of a deletion that extended from 1 to 15 and found that the expression levels in this strain (PLBS140) were similar to those of the 5Ј stem-loop deletion strain (PLBS139) (Table 2). The 5؅ stem-loop reduces the number of (G/U)AG repeats required for tight TRAP-trp leader RNA binding. Our footprint and gel shift results demonstrated that TRAP interacts with the 5Ј stem-loop and that this interaction increases the affinity of TRAP for trp leader RNA. We performed a 3Ј boundary analysis using wild type (nucleotides 1 to 111) and 5Ј stem-loop deletion (nucleotides 32 to 111) trp leader transcripts to determine if the 5Ј stem-loop reduced the number of (G/U)AG repeats that were required for tight TRAP-trp leader RNA binding. Note that these two transcripts were identical to those used in the gel shift analysis (Fig. 2). RNAs were 5Ј end labeled, hydrolyzed to obtain a ladder of 5Ј-end-labeled transcripts, and subsequently mixed with tryptophan-activated TRAP. Bound and unbound RNAs were separated by native gel electrophoresis, gel purified, and separated on a standard denaturing sequencing gel. We observed cutoffs between bound and unbound transcripts with both the wild-type and 5Ј stem-loop deletion trp leader transcripts (Fig. 6). The cutoff for FIG. 6. 3Ј boundary analysis of wild-type and 5Ј stem-loop deletion trp leader transcripts. Limited alkaline hydrolysis ladders of 5Ј-end-labeled wild-type (WT) the wild-type trp leader transcript was relatively sharp and or 5Ј stem-loop deletion trp leader transcripts were incubated with tryptophan- occurred at between seven and eight (G/U)AG repeats, with activated TRAP. TRAP-RNA complexes were separated from unbound RNA on bound and unbound lanes showing complementary cutoffs and a native gel and subsequently fractionated through a denaturing 6% polyacryl- cutons. Under the binding conditions employed here, this re- amide gel (shown). Labels for lanes are as follows: OHϪ, a limited alkaline hydrolysis ladder; T1, partial RNase T1 digest; B and U, bound and unbound are sult demonstrated that the first six (G/U)AG repeats were RNA fragments from the limited alkaline hydrolysis that either bound (B) or did required for stable TRAP-trp leader RNA complex formation not bind (U) TRAP. The numbers on the left (wild-type transcript) or right (5Ј when the 5Ј stem-loop was present in the transcript. However, stem-loop deletion transcript) indicate the relative positions of the (G/U)AG a small fraction of the transcripts that contained as few as three repeats, with 1 being closest to the 5Ј end of the transcript. repeats was also shifted. Interestingly, the corresponding cutoff for the 5Ј stem-loop deletion transcript occurred at between nine and ten (G/U)AG repeats, indicating that the first eight ing mechanism such as this might increase the probability that (G/U)AG repeats were required for comparable binding. In tryptophan-activated TRAP would bind to the trp leader this case a small fraction of the transcripts that contained as in time to block antiterminator formation. For this binding few as six repeats were also shifted. Note that the short tran- mechanism to have the greatest impact on trp operon expres- scripts containing fewer than five (G/U)AG repeats in the sion, one would predict that TRAP–5Ј stem-loop interaction unbound 5Ј stem-loop deletion sample were not gel purified in would occur in the absence of any downstream (G/U)AG re- this experiment (Fig. 6) since previous experiments indicated peats. that these transcripts were not gel shifted by TRAP. The re- We performed a TRAP-RNA footprint experiment using sults of the 3Ј boundary analysis demonstrate that the 5Ј stem- 5Ј-end-labeled trp leader transcripts that contained the 5Ј loop structure reduces the number of (G/U)AG repeats re- stem-loop in the absence of any downstream (G/U)AG repeats quired for stable TRAP association. (nucleotides 1 to 36) to determine if TRAP could interact with ؅ Ј The nature of the TRAP–5 stem-loop RNA interaction is a transcript that only contained the 5 structure. The RNase V1 dependent on the number of downstream (G/U)AG repeats. cleavage pattern in the absence of TRAP differed from the Results from a previous study demonstrated that the 5Ј stem- cleavage pattern when TRAP was present (Fig. 7). We ob- loop functions in the transcription attenuation mechanism that served appreciable RNase V1 cleavage in the presence or ab- controls expression of the trp operon (28). Furthermore, the sence of TRAP following U11, A12, U23, and G24. Surpris- results described above indicate that TRAP interacts with the ingly, cleavage following U25, A26, G31, and C32 was only 5Ј stem-loop and that this interaction increases the affinity of observed in the presence of TRAP. These results indicate that TRAP for trp leader RNA. Moreover, we found that the pres- TRAP can interact with the 5Ј stem-loop in the absence of the ence of the 5Ј structure reduces the number of (G/U)AG 11 (G/U)AG repeats and that this interaction was transient, repeats required for stable TRAP-trp leader RNA association. resulting in a 5Ј stem-loop that is more highly structured. The One possible explanation for these results is that the 5Ј stem- fact that our gel shift assay was unable to detect a complex loop might tether TRAP to the nascent trp leader transcript between TRAP and a transcript that only contained the 5Ј such that TRAP would be in position to bind to the (G/U)AG stem-loop (data not shown) is consistent with rapid TRAP-5Ј repeats as soon as they are transcribed. A multipartite bind- stem-loop RNA complex dissociation. VOL. 182, 2000 TRAP-5Ј STEM-LOOP RNA INTERACTION 1825

FIG. 7. TRAP–5Ј stem-loop RNA footprint analysis with transcripts containing various numbers of (G/U)AG repeats. 5Ј-end-labeled trp leader RNA containing Ј ϩ ϩ the 5 stem-loop and either 0, 3, 6, 9, or 11 (G/U)AG repeats was used in this analysis. trp leader RNA was treated with RNase V1 ( ) in the presence ( ) or absence (Ϫ) of tryptophan-activated TRAP. The positions of the nucleotides corresponding to the 5Ј stem-loop are indicated at the right. The relative positions of the 11 Ϫ (G/U)AG repeats, as well as G18, G24, and G31, are shown on the left. The lanes corresponding to partial alkaline hydrolysis (OH ) and partial RNase T1 digestion (T1) ladders generated from the transcript containing all 11 (G/U)AG repeats are indicated.

We then examined the effect TRAP binding had on 5Ј stem- DISCUSSION loop RNase V1 cleavage patterns when transcripts contained Ј The transcription attenuation mechanism that controls ex- the 5 structure and 3, 6, 9, or 11 (G/U)AG repeats. As expression of the B. subtilis trpEDCFBA operon in response to pected, in the absence of TRAP we found that the cleavage Ј tryptophan relies on TRAP and three RNA secondary struc- pattern within the 5 stem-loop was essentially identical in all tures. When TRAP binds to the 11 (G/U)AG repeats present of the transcripts tested (Fig. 7). However, the cleavage pattern in the nascent trp leader transcript the antiterminator structure of the various transcripts in the presence of bound TRAP cannot form. Instead, an overlapping intrinsic terminator can differed considerably. When the transcript contained the first form which results in transcription termination upstream of the three (G/U)AG repeats (nucleotides 1 to 51), we observed a trp operon structural genes (Fig. 1). A recent genetic study reduction in cleavage following U11, A12, U23, and G24, as demonstrated that the 5Ј stem-loop also participates in the well as increased cleavage following U25, A26, G31, C32, and transcription attenuation mechanism (28). U35 (Fig. 7). Note that the increase in cleavage following U25, In the current study we examined the molecular basis of 5Ј A26, G31, and C32 was not as substantial as that observed for stem-loop function. We determined the secondary structure of Ј the transcript that only contained the 5 stem-loop. The cleav- the 5Ј stem-loop and found that it consists of a relatively age pattern in the transcripts containing the first six (1 to 68) unstable 3-bp lower stem, a 5-by-2 asymmetric internal loop, a or nine (1 to 84) (G/U)AG repeats were similar to one 6-bp upper stem, and a hexaloop at the apex of the structure another, although they differed from the other transcripts (Fig. 3 and 4). It is interesting to note that while both RNase tested. RNase V1 cleavage was essentially absent following T1 and CMCT are single-stranded specific G probes, only U11, A12, U23, G24, A26, G31, and C32 (Fig. 7). Note that CMCT detected G7 in the structure-mapping experiments. there was no increase in cleavage following U25, A26, G31, One possible explanation for this difference is that G7 partic- and C32 (Fig. 7). Remarkably, the RNase V1 cleavage pattern ipates in a non-Watson-Crick base-pairing interaction that pre- 1 in the transcript containing all 11 (G/U)AG repeats (1 to 111) vents RNase T1 cleavage but leaves the N position available was essentially identical to the pattern observed for the for CMCT modification. Our footprinting results suggest that transcript that only contained the 5Ј stem-loop. When taken TRAP interacts with both sides of the asymmetric internal together, these results indicate that TRAP can interact with loop, the upper stem, the hexaloop, and the 3Ј side of the lower 5Ј stem-loop in the absence of any downstream (G/U)AG stem (Fig. 5). It is interesting that the hexaloop contains a repeats and that the TRAP-5Ј stem-loop complex differs de- GAG sequence (nucleotides 18 to 20), while a single AAG pending on the number of (G/U)AG repeats following the 5Ј sequence is present in the residues comprising the 3Ј side of structure. the asymmetric loop and the 3Ј side of the lower stem (nucle- 1826 DU ET AL. J. BACTERIOL.

FIG. 8. Transcription attenuation model of the B. subtilis trp operon. (A) Conditions of tryptophan excess. (B) Limiting tryptophan conditions. The 5Ј and 3Ј ends of the transcript are indicated. TRAP is represented by the gray doughnut structure. See the text for details.

otides 29 to 31). The TRAP binding target in the trp leader ing the entire stem, while a C15G-G22C compensatory change contains four UAG and seven GAG repeats between nucleo- only partially restored expression to wild type-like levels (28). tides 36 and 91 (Fig. 1) (8), while the TRAP binding site in the This suggests that both the structure and the sequence of the unlinked trpG transcript consists of one AAG, one UAG, and upper stem are important for TRAP interaction. seven GAG repeats (15). Since it is known that 11 KKR motifs While our footprinting and 5Ј stem-loop mutation studies on TRAP interact with GAG, UAG, and AAG repeats (2, 5, demonstrated that TRAP interacts with the 5Ј structure and 15, 31), it is possible that KKR motifs contribute to the that this interaction is required for proper regulation of the B. TRAP–5Ј stem-loop complex by interacting with the GAG subtilis trp operon (Table 2), results from our boundary analysis and/or AAG present within the 5Ј structure (Fig. 1). However, indicate that TRAP–5Ј stem-loop interaction reduces the num- as pointed out in Results, substantial evidence suggests that the ber of downstream (G/U)AG repeats that are necessary for GAG sequence in the hexaloop interacts with a region of tight TRAP-trp leader RNA binding (Fig. 6). In addition, our TRAP that is distinct from the KKR motifs. If a KKR motif footprinting results demonstrate that TRAP can interact with interacts with the AAG sequence, then the spacing of four the 5Ј stem-loop without any downstream repeats (Fig. 7). nucleotides between the AAG and the first UAG (nucleotides While the nature of the specific interactions are not well un- 36 to 38) (Fig. 1) is suboptimal. The optimal spacing between derstood, it is particularly striking that TRAP interaction with repeats is two nucleotides (7), although it was determined that the transcript containing only the 5Ј stem-loop resulted in a 5Ј three-nucleotide spacers are tolerated if present in the appro- hairpin that was more highly structured. A qualitatively iden- priate context (9). Moreover, spacers of five and eight nucle- tical result occurred when TRAP interacted with the transcript otides were identified in the trpG transcript (15); thus, it is containing the 5Ј structure and all 11 downstream (G/U)AG possible that a TRAP KKR motif interacts with this AAG repeats (Fig. 7). Interestingly, the TRAP-dependent RNase V1 sequence. This would bring the number of triplet repeats in the cleavage pattern that occurred in the transcripts containing the B. subtilis trp leader TRAP target to 12, the same number 5Ј stem-loop and six or nine repeats were identical to each identified in the Bacillus stearothermophilus trp leader (12). other but clearly distinct from the cleavage pattern of the 0- Note that the UAG sequence (nucleotides 5 to 7) is unlikely to and 11-repeat transcripts. Note that the TRAP-dependent interact with a TRAP KKR motif since deletion of the first five cleavage pattern of the 5Ј stem-loop in the transcript that also residues had virtually no effect on trp operon expression (Table contained three downstream (G/U)AG repeats is intermediate 2). between the other two RNase V1 cleavage patterns. We believe Our results also indicate that TRAP interacts with the 5Ј that this static in vitro experiment captures the essence of the side of the internal loop and the upper stem (Fig. 5 and Table dynamic events taking place during transcription of the trp 2). Moreover, we previously showed that substitution of G7 leader in vivo. with A resulted in a 5Ј stem-loop defect (28). Thus, it appears Our current model of the events taking place during tran- that TRAP interaction with the 5Ј side of the internal loop scription attenuation of the B. subtilis trp operon is shown in and/or non-Watson-Crick base pairing within this RNA seg- Fig. 8. Soon after transcription initiates the 5Ј stem-loop forms ment is crucial for 5Ј stem-loop function. Furthermore, we (structure 1) (Fig. 8A). Tryptophan-activated TRAP subse- previously demonstrated that disruption of the upper stem by quently binds to structure 1, thereby promoting formation of a point mutations (C15G or G22C) had similar effects as delet- more highly structured 5Ј hairpin (structure 2). As transcrip- VOL. 182, 2000 TRAP-5Ј STEM-LOOP RNA INTERACTION 1827 tion proceeds, the KKR motifs on the TRAP perimeter inter- 9. Babitzke, P., J. Yealy, and D. Campanelli. 1996. Interaction of the trp act with the (G/U)AG repeats one at a time as they become RNA-binding attenuation protein (TRAP) of Bacillus subtilis with RNA: effects of the number of GAG repeats, the nucleotides separating adjacent available, thereby wrapping the RNA around the periphery of repeats, and RNA secondary structure. J. Bacteriol. 178:5159–5163. the TRAP complex. Once all of the (G/U)AG repeats are 10. Babitzke, P., P. Gollnick, and C. Yanofsky. 1992. The mtrAB operon of bound, the geometry of this TRAP-trp leader RNA complex is Bacillus subtilis encodes GTP cyclohydrolase I (MtrA), an enzyme involved such that the trp leader transcript encircles the entire TRAP in folic acid biosynthesis, and MtrB, a regulator of L-tryptophan biosynthesis. 11-mer (2). Once this occurs the 5Ј stem-loop can dissociate J. Bacteriol. 174:2059–2064. 11. Bevilacqua, P. C., C. X. George, C. E. Samuel, and T. R. Cech. 1998. Binding from TRAP and retain the conformation of stem-loop struc- of the protein kinase PKR to RNAs with secondary structure defects: role of ture 2 or remain bound. The ability of the 5Ј stem-loop to the tandem A-G mismatch and noncontiguous helixes. Biochemistry 37: remain bound is supported by the gel shift results, where we 6303–6316. observed increased TRAP affinity when the 5Ј stem-loop was 12. Chen, X.-P., A. A. Antson, M. Yang, P. Li, C. Baumann, E. J. Dodson, G. G. Dodson, and P. Gollnick. 1999. Regulatory features of the trp operon and the present in a transcript that contained all 11 (G/U)AG repeats crystal structure of the trp RNA-binding attenuation protein from Bacillus (Fig. 4), while dissociation is supported by the footprint anal- stearothermophilus. J. Mol. Biol. 289:1003–1016. ysis (Fig. 7). As a consequence of TRAP binding, the antiter- 13. Donis-Keller, H., A. M. Maxam, and W. Gilbert. 1977. Mapping adenines, minator structure cannot form, which promotes formation of guanines, and pyrimidines in RNA. Nucleic Acids Res. 4:2527–2538. the terminator structure and, hence, transcription termination. 14. Du, H., and P. Babitzke. 1998. trp-RNA binding attenuation protein-mediated long-distance RNA refolding regulates translation of trpE in Bacillus Since only a relatively short window of opportunity exists for subtilis. J. Biol. Chem. 273:20494–20503. TRAP to block antiterminator formation, it appears that this 15. Du, H., R. Tarpey, and P. Babitzke. 1997. The trp-RNA binding attenuation multipartite binding mechanism increases the probability that protein regulates TrpG synthesis by binding to the trpG ribosome binding TRAP associates with the nascent trp leader transcript in time site of Bacillus subtilis. J. Bacteriol. 179:2582–2586. 16. Gollnick, P. 1994. Regulation of the Bacillus subtilis trp operon by an RNA- to promote termination. When the concentration of trypto- binding protein. Mol. Microbiol. 11:991–997. phan is low, TRAP is not activated and does not bind to the 17. Gollnick, P., S. Ishino, M. I. Kuroda, D. J. Henner, and C. Yanofsky. 1990. nascent trp leader transcript. In this case, antiterminator for- The mtr locus is a two-gene operon required for transcription attenuation in mation prevents formation of the intrinsic terminator, result- the trp operon of Bacillus subtilis. Proc. Natl. Acad. Sci. USA 87:8726–8730. ing in transcription of the entire operon (Fig. 8B). The trp 18. Hoffman, R. J., and P. Gollnick. 1995. The mtrB gene of Bacillus pumilus encodes a protein with sequence and functional homology to the trp RNA- operon leader transcripts of Bacillus pumilus (18), Bacillus binding attenuation protein (TRAP) of Bacillus subtilis. J. Bacteriol. 177: caldotenax (31), and B. stearothermophilus (29) also contain 5Ј 839–842. stem-loops and multiple triplet repeats, as well as overlapping 19. Kuroda, M. I., D. Henner, and C. Yanofsky. 1988. cis-acting sites in the antiterminator and terminator structures. Thus, it appears that transcript of the Bacillus subtilis trp operon regulate expression of the operon. J. Bacteriol. 170:3080–3088. all four organisms control expression of the trp operon by 20. Landick, R., C. L. Turnbough, and C. Yanofsky. 1996. Transcription atten- essentially identical transcription attenuation mechanisms. uation, p. 1263–1286. In F. C. Neidhardt et al. (ed.), Escherichia coli and Salmonella: cellular and molecular biology. American Society for Microbi- ology, Washington, D.C. ACKNOWLEDGMENTS 21. Lowman, H. B., and D. E. Draper. 1986. On the recognition of helical RNA We thank Philip Bevilacqua, Craig Cameron, and Subita Suders- by cobra venom V1 nuclease. J. Biol. Chem. 261:5396–5403. Merino, E., P. Babitzke, and C. Yanofsky. hana for discussions throughout the course of this study. We also thank 22. 1995. trp RNA-binding attenuation protein (TRAP)-trp leader RNA interactions mediate translational as Philip Bevilacqua, Janell Schaak, and Charles Yanofsky for critical well as transcriptional regulation of the Bacillus subtilis trp operon. J. Bac- reading of the manuscript. teriol. 177:6362–6370. This work was supported by grant GM52840 from the National 23. Miller, J. H. 1972. Experiments in molecular genetics, p. 352–355. Cold Institutes of Health. Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 24. Otridge, J., and P. Gollnick. 1993. MtrB from Bacillus subtilis binds specif- REFERENCES ically to trp leader RNA in a tryptophan-dependent manner. Proc. Natl. Acad. Sci. USA 90:128–132. 1. Antson, A. A., A. M. Brzozowski, E. J. Dodson, Z. Dauter, K. S. Wilson, T. 25. Shimotsu, H., M. I. Kuroda, C. Yanofsky, and D. J. Henner. 1986. Novel Kurecki, J. Otridge, and P. Gollnick. 1994. 11-fold symmetry of the trp form of transcription attenuation regulates expression of the Bacillus subtilis RNA-binding attenuation protein (TRAP) from Bacillus subtilis determined tryptophan operon. J. Bacteriol. 166:461–471. by X-ray analysis. J. Mol. Biol. 244:1–5. 26. Shiratsuchi, A., and S. Sato. 1991. Nucleotide sequence of trpE, anthranilate 2. Antson, A. A., E. J. Dodson, G. Dodson, R. B. Greaves, X.-P. Chen, and P. synthase I gene, of Bacillus caldotenax. Biochim. Biophys. Acta 1090:348– Gollnick. 1999. Structure of the trp RNA-binding attenuation protein, TRAP, bound to RNA. Nature 401:235–242. 350. 3. Antson, A. A., J. Otridge, A. M. Brzozowski, E. J. Dodson, G. G. Dodson, 27. Spizizen, J. 1958. Transformation of biochemically deficient strains of Ba- K. S. Wilson, T. M. Smith, M. Yang, T. Kurecki, and P. Gollnick. cillus subtilis by deoxyribonucleate. Proc. Natl. Acad. Sci. USA 44:1072–1078. 1995. The Ј structure of trp RNA-binding attenuation protein. Nature 374:693–700. 28. Sudershana, S., H. Du, M. Mahalanabis, and P. Babitzke. 1999. A 5 RNA 4. Babitzke, P. 1997. Regulation of tryptophan biosynthesis: Trp-ing the TRAP stem-loop participates in the transcription attenuation mechanism that con- or how Bacillus subtilis reinvented the wheel. Mol. Microbiol. 26:1–9. trols expression of the Bacillus subtilis trpEDCFBA operon. J. Bacteriol. 5. Babitzke, P., and C. Yanofsky. 1993. Reconstitution of Bacillus subtilis trp 181:5742–5749. attenuation in vitro with TRAP, the trp RNA-binding attenuation protein. 29. Walter, A. E., D. H. Turner, J. Kim, M. H. Lyttle, P. Mueller, D. H. Mathews, Proc. Natl. Acad. Sci. USA 90:133–137. and M. Zuker. 1994. Coaxial stacking of helixes enhances binding of oligo- 6. Babitzke, P., and C. Yanofsky. 1995. Structural features of L-tryptophan ribonucleotides and improves predictions of RNA folding. Proc. Natl. Acad. required for activation of TRAP, the trp RNA-binding attenuation protein of Sci. USA 91:9218–9222. Bacillus subtilis. J. Biol. Chem. 270:12452–12456. 30. Yang, M., A. de Saizieu, A. P. G. M. van Loon, and P. Gollnick. 1995. 7. Babitzke, P., D. G. Bear, and C. Yanofsky. 1995. TRAP, the trp RNA-binding Translation of trpG in Bacillus subtilis is regulated by the trp RNA-binding attenuation protein of Bacillus subtilis, is a toroid-shaped molecule that binds attenuation protein (TRAP). J. Bacteriol. 177:4272–4278. transcripts containing GAG or UAG repeats separated by two nucleotides. 31. Yang, M., X.-P. Chen, K. Militello, R. Hoffman, B. Fernandez, C. Baumann, Proc. Natl. Acad. Sci. USA 92:7916–7920. and P. Gollnick. 1997. Alanine-scanning mutagenesis of Bacillus subtilis trp 8. Babitzke, P., J. T. Stults, S. J. Shire, and C. Yanofsky. 1994. TRAP, the trp RNA-binding attenuation protein (TRAP) reveals residues involved in tryp- RNA-binding attenuation protein of Bacillus subtilis, is a multisubunit com- tophan binding and RNA binding. J. Mol. Biol. 270:696–710. plex that appears to recognize G/UAG repeats in the trpEDCFBA and trpG 32. Zuker, M. 1989. On finding all suboptimal foldings of an RNA molecule. transcripts. J. Biol. Chem. 269:16597–16604. Science 244:48–52.