INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Formation of spores from vegetative bacteria by Bacillus subtilis is a primitive system of cell differentiation. A hallmark of B. subtilis sporulation is an asymmetric division that produces two cells of unequal size, the smaller prespore and the larger mother cell. The prespore develops into the mature, heat-resistant spore, whereas the mother cell ultimately lyses. These radically different developmental fates are in part determined by the action of a series of sporulation-specific RNA polymerase  factors: ^F and ^G in the prespore and ^E and ^K in the mother cell. Of these, ^F is the first to become active (reviewed in references 17 and 36).

Sigma factors determine the promoter recognition specificity of RNA polymerase. Comparative analysis of different promoters and analysis of promoter mutations have revealed that two regions centered at approximately 35 and 10 nucleotides upstream of the transcription start point are of prime importance for promoter recognition directed by the majority of  factors. They are generally referred to as the 35 and 10 regions. By far the most extensively studied promoters are those for the major vegetative  factors ⁷⁰ of Escherichia coli and ^A of B. subtilis. For both of these  factors, the consensus 35 sequence is TTGACA and the 10 consensus is TATAAT. The distance between these two sequences is also important and varies mostly between 16 and 18 nucleotides. Although the sequences at these two regions are not strictly conserved, B. subtilis promoters conform much more closely to the consensus than do those of E. coli (11). The different promoters for the sporulation-associated  factors of B. subtilis are less firmly defined (9).

The complex regulation of ^F activation is increasingly becoming understood (16, 17). However, much less is known about promoters recognized by RNA polymerase containing ^F (E-^F). Few genes have been identified that are transcribed by E-^F (9). The two genes of known function that are exclusively under the control of E-^F are spoIIR (13, 20) and spoIIQ (19); P_spoIIQ is much the stronger of the two promoters. Several E-^F-transcribed genes are also transcribed by E-^G (9), whose promoter specificity overlaps that of E-^F (9), whereas the specificity of one overlaps that of E-^B (28).

In order to investigate the features of promoters that are recognized by E-^F, we have employed a method based on that developed by Oliphant and Struhl (23, 24) to study ^A promoters of E. coli. The method utilizes chemically synthesized DNA sequences that are degenerate in a predefined region. In the present study, the randomized sequences were for the 10 and 35 regions of the promoter for spoIIQ. Randomized promoter regions were cloned in front of a promoterless copy of lacZ in a vector designed for insertion by double crossover of single copies of the promoter-lacZ fusions into the amyE region of the B. subtilis chromosome. This system made it possible to test for transcription of lacZ by E-^F in vivo. The results indicate a weak ^F-specific 10 consensus and a rather stronger 35 consensus. The looseness of the ^F promoter requirement contrasts with the strict requirement for ^A promoters (11) and suggests that additional factors may determine the specificity of promoter recognition by E-^F in vivo.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains. The E. coli strain used for all cloning and clone bank amplification was DH5 [F endA1 hsdR17(r_K m_K⁺) supE44 thi-1 recA1 gyrA96 relA1 (lacZYA-argF)U169 80dlacZM15]. The parental B. subtilis strain was BR151. The B. subtilis strains used are listed in Table 1.

Plasmids. All plasmids were maintained in E. coli DH5. Plasmid pEIA84 is a derivative of integrative plasmid pDH32 (35), in which the spoIIQ promoter region extending from 200 to +9 was amplified by PCR and inserted into the EcoRI site (made blunt with the Klenow fragment of DNA polymerase) of pDH32, which is located immediately upstream of the lacZ ribosome binding site. Plasmid pEIA96 is a derivative of pDH32 (35) in which the cat cassette was removed by digestion with EcoRI and StuI and replaced with a neo cassette from a derivative of pBEST501 (12). pEIA98 was constructed by cloning the region upstream of SpoIIQ, extending from 200 to 38 (obtained by PCR), into the EcoRI site (made blunt with Klenow) of pEIA96.

Media. E. coli was maintained on Luria-Bertani (LB) broth or LB agar supplemented with ampicillin (100 µg/ml) when required. B. subtilis was maintained by using Schaeffer's sporulation agar (SSA) or modified Schaeffer's sporulation medium (MSSM) (30). When appropriate, B. subtilis was grown in the presence of antibiotics at the following concentrations: neomycin, 5 µg/ml; spectinomycin, 50 µg/ml. When required, 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal) was added to SSA to a concentration of 100 µg/ml.

Nucleic acid manipulation. DNA was prepared as described previously (39). DNA sequencing was done by the dideoxy-chain termination method of Sanger et al. (33), using the Circum Vent Thermal Cycle Dideoxy DNA Sequencing Kit (New England Biolabs, Beverly, Mass.). The primers used for sequencing of cloned promoter inserts were LacZ2 (reverse) (GGGTTTTCCCGGTCG) and IIQ12 (forward) (CTATGTTCAGCAAGACGC). Cellular RNA was extracted essentially as described by Penn et al. (27). Primer extension analyses were performed as described previously (39).

Transformation. B. subtilis transformation was performed as described previously (31). When large numbers of colonies were being screened as donors in transformations, donor DNA was not purified. Instead, a modification of the method of Ephrati-Elizur (6) was used essentially as described previously (29). Donor strains were grown in LB broth to stationary phase, at which stage they secreted DNA onto the cell surface or into the medium (6). A volume of 0.05 ml of these cultures was used as a donor and added to 0.5 ml of the culture of a competent recipient. This method required counterselection against the donor.

Induction of the P_spac promoter by IPTG. Overnight cultures were diluted 100-fold into prewarmed MSSM and incubated at 37°C with aeration. Induction of P_spac by addition of 1 mM isopropyl--D-thiogalactopyranoside (IPTG) was performed as described previously (34)

-Galactosidase activity. B. subtilis strains grown in MSSM at 37°C were assayed with o-nitrophenyl--D-galactopyranoside as substrate essentially as described by Nicholson and Setlow (22). -Galactosidase specific activity is expressed as nanomoles of o-nitrophenyl--D-galactopyranoside hydrolyzed per minute per milligram (dry weight) of bacteria.

Construction of randomized spoIIQ promoter clone banks. Clone banks were constructed as described by Oliphant and Struhl (23), with some modifications. Approximately 50 pmol of oligonucleotides R10IIQup and R35IIQup was phosphorylated with 5 U of T4 polynucleotide kinase and ATP for 1 h at 37°C in separate 1.5-ml microcentrifuge tubes. These oligonucleotides consisted of the spoIIQ promoter region extending from 37 to +9, with positions 14 to 7 (R10IIQup) or 35 to 31 (R35IIQup) randomized. The phosphorylated oligonucleotides were cleaned by precipitation and resuspended in 10 µl of sterile distilled water. Next, 50 pmol of oligonucleotide IIQext was added, separately, to the phosphorylated R10IIQup and R35IIQup oligonucleotides. Primer IIQext has 15 bases, corresponding to spoIIQ positions +9 to 6, that hybridize to R10IIQup and R35IIQup and also a 12-nucleotide extension downstream of +9 that will form a BamHI restriction site. The mixtures were heated to 90°C for 5 min and then cooled to room temperature slowly. After the annealing, Klenow buffer (1× final concentration), dNTPs (0.25 mM final concentration of each dNTP), and 5 U of Klenow fragment were added in a final volume of 30 µl. The mixtures were incubated in 37°C for 1 h and then fractionated by electrophoresis in a 4% NuSieve GTG agarose gel. The appropriate double-stranded DNA (dsDNA) band was excised from the gel and purified by using the QIAEX II (Qiagen, Chatsworth, Calif.) gel extraction kit. Approximately 2 µg of dsDNA from each preparation was digested with BamHI and purified. The resulting oligonucleotides, with a blunt end at 37 and a BamHI site at +9, were ligated to pEIA98 previously digested with StuI and BamHI. As the yield of dsDNA was somewhat variable, the amount of insert to be added to a given amount of vector was determined empirically in order to optimize the ligation reactions. In this way, random 10 (R10) and random 35 (R35) clone banks were obtained by using the framework of P_spoIIQ.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Test system for analysis of E-^F-transcribed promoters. The two well-characterized loci that are known to be transcribed only by E-^F are spoIIR (13, 20) and spoIIQ (19). Primer extension analysis was used to identify the 5' end of the spoIIQ transcript and hence to infer the transcription start point and promoter. RNA was isolated from B. subtilis strain SL5618 4 h after the start of sporulation. Primer extension analysis was performed, and the same 5' end was identified with two different primers (Fig. 1). It is inferred that this represents the transcription start point. The sequence of the spoIIQ promoter is indicated in Fig. 2. The deduced spoIIQ promoter 10 and 35 regions are indicated in bold. A similar analysis was performed for spoIIR (data not shown). The deduced spoIIR promoter agreed with that predicted previously (Fig. 2) (13). Also shown are the promoter regions for genes known to be transcribed in vivo by E-^F and whose transcription start point has been inferred from primer extension analysis.

View larger version (66K): [in this window] [in a new window]   FIG. 1.   Determination of the 5' end of spoIIQ mRNA by primer extension analysis. RNA was extracted from strain SL5618 (spoIIIG1) 4 h after the end of exponential growth. Primer extension analysis was done with primer IIQ2 (TCAGCCAACGGATCCTTTACC, positions +161 to +181 from the inferred transcription start point) (A) and primer IIQ10 (CTGATTGATACCAAAGGAC, positions +128 to +147 from the inferred transcription start point) (B). Sequencing ladders using the same primers are also shown. The likely transcription start site is indicated by an asterisk.

View larger version (26K): [in this window] [in a new window]   FIG. 2.   Sequence alignment of promoter regions for E-F-transcribed genes for which the transcription start site has been inferred by primer extension analysis. Of these, only spoIIQ, spoIIR, and katX are not also recognized by E-F (1-3, 5, 7, 8, 13, 34, 37-39). Asterisks indicate spaces introduced to adjust the spacing between the 10 and 35 consensuses. The underlined bases indicate the inferred transcription start points.

Construction and analysis of a clone bank containing a randomized 10 ^F promoter region. The procedure used to construct a library with the randomized 10 region is described in Materials and Methods. Positions 14 to 7 were randomized, with the rest of P_spoIIQ kept constant from 200 to +9. In particular, the G at 15 was kept constant. Sun et al. (37) had shown that a G in this position in other promoters appeared to be necessary for ^F recognition (although one weak ^F /^G promoter, for ywhE, has recently been described that has a T at this position [26]). The ligation mixtures were used to transform E. coli, and in all, approximately 250,000 transformant colonies were obtained. Plasmids were isolated from eight of these clones, and the promoter regions were sequenced. Three or four different nucleotides were found at each of the randomized positions in the eight clones, indicating that the randomization was successful; none of the eight clones displayed ^F promoter activity in B. subtilis (data not shown). The rest of the transformant colonies were recovered from the agar plates, and pooled plasmid DNA was prepared from them. This preparation was designated the R10 clone bank.

View larger version (24K): [in this window] [in a new window]   FIG. 3.   Effect of induction of F on expression of lacZ associated with different spoIIQ promoter regions. Synthesis of F was induced by addition of IPTG to strains containing the structural gene for F, spoIIAC, under the control of the Pspac promoter (41). Shown are strains with alterations to the 10 region (A) and the 35 region (B) of the promoter directing lacZ. Cultures were grown in MSSM to an optical density at 600 nm of 0.3. They were then divided in two, and 1 mM IPTG was added to one of each pair. None of the cultures displayed significant -galactosidase activity in the absence of IPTG. (A) SL7472/EIA100 (PspoIIQ [200 to 38]) symbols: , no IPTG; , 1 mM IPTG. SL7472/R10-103 symbols: , no IPTG; , 1mM IPTG. SL7472/R10-333 symbols: , no IPTG; , 1 mM IPTG. SL7472/R10-435 symbols: , no IPTG; , 1 mM IPTG. SL7472/R10-901 symbols: ×, no IPTG; *, 1 mM IPTG. SL7472/R10-1024 symbols: +, no IPTG; , 1 mM IPTG. Samples grown in the absence of IPTG produced little or no -galactosidase, and symbols for the different cultures largely mask each other. (B) SL7472/R35-10 symbols: , no IPTG; , 1 mM IPTG. SL7472/R35-20 symbols: , no IPTG; , 1 mM IPTG. SL7472/R35-43 symbols: , no IPTG; , 1 mM IPTG. SL7472/R35-53 symbols: , no IPTG; , 1 mM IPTG. wt, weight.

View this table: [in this window] [in a new window]   TABLE 2.   Promoter sequences from the R10 clone bank that responded to E-F

Construction and analysis of a clone bank containing a randomized 35 ^F promoter region. The 35 clone bank was constructed in a way similar to that used for the R10 clone bank, as described in Materials and Methods. Five bases, from 35 to 31 in the spoIIQ promoter, were randomized. Ligation mixes were used to transform E. coli, and about 7,700 transformants were obtained. The transformant colonies were recovered, and pooled plasmid DNA was prepared from them; this preparation is referred to as the R35 clone bank. The same general procedure was used to screen the R35 and R10 banks. Transformation of EIA87 with a portion of the R35 bank yielded 3,300 colonies on SSA containing neomycin and X-Gal; of these, 64 appeared blue after 2 days at 37°C. These were presumed to represent distinct clones. All 64 clones were screened for ^F-responsive promoter activity, as described for the R10 bank; 44 passed the screens and were presumed to contain ^F-controlled promoters. Four of the 44 presumptive ^F-controlled promoter clones were also tested for their response to artificial induction of ^F by addition of IPTG to a strain (SL7542) that had the structural gene for ^F, spoIIAC, placed under the control of the P_spac promoter. -Galactosidase was measured to indicate promoter activity. The results are shown in Fig. 3B. Addition of IPTG resulted in induction of -galactosidase in all four clones.

View this table: [in this window] [in a new window]   TABLE 3.   Promoter sequences from the R35 bank that responded to E-F

Distinguishing promoters recognized only by E-^F from those recognized by both E-^F and E-^G. As mentioned previously, ^F and ^G have overlapping promoter specificities (9, 37). The positive, sequenced clones from the R10 and R35 banks were tested for -galactosidase expression in a B. subtilis strain, SL5037, that lacked ^F but had ^G activity. This strain has the spoIIA4 mutation that deletes most of the spoIIA operon, including spoIIAC, which encodes ^F, and spoIIAB, which encodes a negative regulator of ^G (15, 31); apparently as a consequence of the deletion of spoIIAB, ^G becomes active even though ^F is absent (V. K. Chary and P. J. Piggot, unpublished observations). Promoter-lacZ fusions were introduced into SL5037 by transformation. As controls, dacF-lacZ (responding to ^F and ^G) and spoIIQ-lacZ (responding only to ^F) were also introduced into SL5037. Transformants were selected on SSA containing neomycin with X-Gal. Transformants were scored after a 2-day incubation at 37°C. Colonies of the positive control (dacF-lacZ) were blue; colonies of the negative control (spoIIQ-lacZ) were white. By this test, 10 of the 35 R10 promoter-lacZ fusions responded to ^G. None of the R35 promoter-lacZ fusions was responsive. The failure of the R35 clones to respond to ^G is thought to result from their having the 10 sequence of the spoIIQ promoter, which is presumed to prevent recognition by E-^G. The responses of the different R10 clones to ^G are indicated in Table 2. Seven of the natural promoters responded to ^G, while three did not (Fig. 1). The frequencies of bases at different positions are shown in Table 4 for promoters that responded to ^G and for those that did not; data for natural promoters and for the R10 bank are combined.

View this table: [in this window] [in a new window]   TABLE 4.   Frequencies of occurrence of different nucleotides in the 10 region of promoters recognized by E-Fa

Search for ^F-controlled genes by computer-assisted analysis. The putative ^F consensus sequence, as defined above, was used to search the B. subtilis database for any potential promoter within 100 nucleotides upstream of an open reading frame by utilizing the search tools provided by the Subilist database (http://genolist.pasteur.fr/SubtiList). For the search, four different 35 sequences (GTTTD, GTATD, GCTTD, and GCATD, where D is A, G, or T) were used, with a spacer of 14 to 16 bp and GKNNANNNT (K is G or T; N is G, A, T, or C) as the 10 sequence. The search gave 193 hits, including all known ^F-controlled genes that conformed to the search sequence. Of the potential promoter regions, the following seven were chosen for further analysis: yhcR, ycdF, yveA, ykqC, yxlJ, lonB, and yfhF. The potential promoter regions were amplified by PCR and cloned into pEIA96 as a prelude to testing in strain SL7542 (P_spac-spoIIAC). The expression of only one, lonB, was induced by ^F, indicating that lonB is transcribed by E-^F in vivo. LonB is involved in the proteolysis of ^H at low pH (18). However, a knockout mutation of lonB was generated and it had no discernible effect on growth or sporulation (data not shown). Serrano et al. have also identified lonB as an E-^F-transcribed gene (M. Serrano, S. Hövel, C. P. Moran, Jr., A. O. Henriques, and U. Völker, personal communication).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The RNA polymerase factor ^F has a pivotal role in the sporulation of B. subtilis. Its activation is highly regulated. It is the first  factor to display compartmentalized expression, during sporulation (10). It is made in large quantities (21), and inappropriate expression is toxic (4, 42). However, fewer promoters have been identified as specifically requiring only ^F for their expression than requiring any of the other sporulation-associated  factors (1-3, 5, 7, 8, 13, 34, 37-39). Sun et al. originally identified four promoters that responded to E-^F. The most striking feature that they noted was G at 15 (37) (to avoid confusion, all numbering is relative to the spoIIQ promoter [Fig. 2]). Subsequently, Londoño-Vallejo et al. (19) identified spoIIQ, which was found to have the strongest promoter known to be recognized only by E-^F. We determined the 5' end of ^F spoIIQ mRNA and inferred the transcription start point and hence the promoter. The promoter deviated somewhat from the previously suggested consensus but retained the G at 15. Here we report the analysis of ^F 10 and 35 region recognition determinants by using a method similar to that developed by Oliphant and Struhl (23), with the spoIIQ promoter as the base.

We tested the effects of randomizing positions 14 to 7 of the spoIIQ promoter while keeping the rest of the region from 200 to +9 fixed. Surprisingly, almost 1% of 56,000 clones with this randomized region displayed E-^F promoter activity. Of 35 active promoters that were sequenced, only 2 were the same and only 1 was identical to the spoIIQ promoter (Table 2). This variability suggests that a large number of possible 14 to 7 sequences could be recognized by E-^F within the context of the spoIIQ promoter region. Data for the 10 natural promoters are combined with data for the R10 sequences in Table 4. A consensus, GG/tNNANNNT, from 15 to 7 was discernible. However, the ANNNT portion, from 11 to 7, is common to all sporulation-associated  factors, as well as to ^A (Table 5). The 10 natural ^F promoters had T or A at 16 (Fig. 2), and this position might also contribute to promoter recognition. The G at 15 had previously been considered to be essential for ^F activity (37), although one exception, the ywhE promoter, has recently been described (26). Nucleotides G are T are favored at 14. Site-directed mutatgenesis of the spoIIQ promoter confirmed the importance of the A at 11 and the T at 7 and also confirmed that a C is tolerated at 14 (E. Amaya and P. J. Piggot, unpublished observations).

View this table: [in this window] [in a new window]   TABLE 5.   Alignment of 10 consensus sequences for B. subtilis  factorsa

The limited nature of the 10 consensus identified as specific for ^F promoters was surprising. Comparison of ^A-controlled promoters has indicated that they conform much more closely to the consensus in B. subtilis than do the comparable ⁷⁰-controlled promoters in E. coli (11). The analysis by Oliphant and Struhl (23) of ⁷⁰-controlled promoters indicated that the results obtained with their method agreed very well with those obtained by analyzing natural promoters. Our preliminary analysis of B. subtilis ^A promoters by the randomized-oligonucleotide method is entirely consistent with a tighter consensus than for ⁷⁰ promoters of E. coli (A. Khvorova and P. J. Piggot, unpublished observations). It may be that the context of the spoIIQ promoter permitted the flexibility in the 10 region of the E-^F promoters analyzed. However, the natural E-^F promoters (Fig. 2) did not suggest a more restricted 10 region, with the possible exception of additional conformity at 9.

Analysis of the effects of randomizing five nucleotides in the 35 region suggested greater ^F specificity than at 10. Of 3,300 clones tested with a randomized 35 to 31 pentamer, 44 displayed ^F promoter activity. From these, 18 were sequenced, and 17 had the motif GTAT (35 to 32), with 15 being GTATA/T (Table 3); analysis of nonexpressing clones in the R35 bank had indicated that the 35 to 31 region was indeed randomized in the bank. The 10 natural ^F promoters display greater variability, with T, C, or G being found at position 34 and T or A at 33 (Fig. 2). This greater flexibility in the natural promoters may indicate that the spoIIQ promoter framework somehow restricts the flexibility in recognition of the 35 region. As with the 10 consensus, much of the 35 consensus, GNATA, is recognized by other sporulation-specific  factors (Table 6). The natural promoters also suggested that positions flanking the region tested might contribute to promoter specificity; these positions were not investigated in the present study. The natural promoter sequences are combined with the randomization data to illustrate the conformity to the 35 consensus (Table 7).

View this table: [in this window] [in a new window]   TABLE 6.   Alignment of 35 consensus sequences for B. subtilis  factorsa

View this table: [in this window] [in a new window]   TABLE 7.   Frequency of occurrence of different nucleotides in the 35 region of promoters recognized by E-Fa

E-^F has promoter specificity that overlaps that of E-^G (9). P_spoIIQ is one of the few E-^F promoters that are not recognized by E-^G. Ten of the promoters from the R10 library constructed in the P_spoIIQ background responded to ^G, and 25 did not (Table 2). None of the promoters with a T at 14 responded to ^G (Tables 2, and 5), suggesting that the T at this position discriminated against ^G. However, a G, C, or A at this position was not sufficient among ^F promoters to determine ^G recognition. No clear discriminator was found at any other position. Scanning of known natural ^F promoters outside the regions tested here (Fig. 2) shed no further light on the situation. An example of the complexity is at position 12, where all four nucleotides were found in both classes of promoters (Tables 4 and 5), and yet R10-333, which responded to ^G, differed only at that position from R10-301, which did not respond. None of the clones from the R35 library was responsive to ^G; this is presumably a consequence of the T at position 14 in the spoIIQ promoter discriminating against ^G.

The B. subtilis database was scanned for plausible ^F promoters located within 100 nucleotides of the start of open reading frames, as defined in the database. By using the ^F consensus defined from the R10 and R35 analysis, 193 possible promoters were identified. A set of seven was chosen from this group that also fulfilled the additional criterion of having an upstream AT-rich region. This set was tested to see if the genes were transcribed in vivo by E-^F. Only one of the seven, lonB, was transcribed by E-^F. The inadequacy of our database predictions indicates that something additional to the 10 and 35 regions is important for E-^F recognition. This same conclusion can be deduced from the lack of a strong ^F-specific promoter consensus sequence. It may be that the surrounding spoIIQ promoter context provides ^F recognition determinants additional to those tested in the 10 and 35 regions. However, no strong determinants were apparent when other natural ^F promoters (Fig. 2) were compared to the set of eight potential promoters. For example, A/T at position 16 (Fig. 2) was present in three of the seven nonfunctional test promoters and not present in the likely lonB promoter. However, comparison of known ^F promoters did not reveal any obvious shared sequence motif outside the 10 and 35 regions (Fig. 2).

Accessory transcription factors have been identified that strongly modulate (activate or repress) the activity of RNA polymerase containing each of the other sporulation-associated  factors: SpoIIID, SpoVT, and GerE are associated with ^E, ^G, and ^K, respectively (17). No factor has been identified that shows comparable strong regulation of E-^F-directed promoters, although it seems plausible that such factors exist. The transcription of spoIIIG is delayed by some 40 min compared to the transcription of other ^F-directed genes (14) and requires that ^E also be active (25), suggesting that for spoIIIG, at least, there is an absolute requirement for an accessory transcription factor. Wu and Errington (40) have reported that RsfA increases the expression of spoIIR and modulates, to some extent, the activity of E-^F on other promoters, although not transcription of spoIIQ, the base of the present study. A requirement for unidentified transcription factors may explain, in part, the looseness of the ^F promoter consensus and the inability of our search to identify novel ^F-directed genes. However, it is also possible that the promoter recognition requirements of E-^F are complex, with weak contributions from various positions outside the tested regions, as well as from the tested regions.

Selection of functional nucleotide sequences from randomized libraries has become a powerful technique for studying the structure and function of the nucleic acids (7). This technique (often called SELEX) has been used successfully to select in vitro the functional sequences in both RNA and DNA that are capable of binding to specific ligands or of performing a chemical reaction. The use of the randomized libraries for selection of the functional nucleic acid sequences is mostly restricted to in vitro applications, mainly because of physical limitations on the length of the library (about 15 nucleotides) imposed by the cloning and expression procedures. Although the randomized libraries are limited by length, selection from them in vivo can potentially be a very powerful tool with which to study functions of nucleic acids that involve the interactions of short nucleotide (DNA or RNA) motifs. This approach was successfully used to analyze the consensus sequence of E. coli E-⁷⁰ promoters (23). The present study illustrates the usefulness and the limitations of SELEX logic applied to an analysis of promoter recognition determinants of E-^F in B. subtilis.