Alanine-scanning mutagenesis of Bacillus subtilis trp RNA-binding attenuation protein (TRAP) reveals residues involved in tryptophan binding and RNA binding1

J. Mol. Biol. (1997) 270, 696±710

Alanine-scanning Mutagenesis of Bacillus subtilis trp RNA-binding Attenuation Protein (TRAP) Reveals Residues Involved in Tryptophan Binding and RNA Binding Min Yang, Xiao-ping Chen, Kevin Militello, Robert Hoffman Bruce Fernandez, Chris Baumann and Paul Gollnick* Department of Biological Sciences, State University of New York at Buffalo, Buffalo NY, 14260, USA

In Bacillus subtilis, expression of the trp genes is negatively regulated by an RNA binding protein called TRAP (trp RNA-binding Attenuation Protein), which is activated to bind RNA by binding L-tryptophan. TRAP contains 11 identical subunits assembled in a symmetric ring. We have used alanine-scanning mutagenesis to analyze the functions of surface amino acid residues of TRAP. The in vivo regulatory activity of each mutant TRAP was analyzed in a B. subtilis reporter strain containing a trpE0 -0 lacZ fusion. Mutant TRAP proteins with defective in vivo regulatory activities were characterized in vitro by measuring their tryptophan binding and RNA binding activities. Most of the mutant proteins with altered tryptophan binding, either af®nity or cooperativity, contained substituted residues located on two loops formed by residues 25 to 33 and residues 49 to 52, as well as on the b-strand and b-turn contiguous with these loops. Substitution of three residues (Lys37, Lys56 and Arg58) with alanine resulted in signi®cant decreases in the RNA binding activity of TRAP without altering tryptophan binding. Structural analysis shows that these three residues are directly aligned on the outer edge of TRAP. Further mutagenic analysis of these three residues revealed that only lysine or arginine residues at positions 37 or 58 allow proper TRAP function, whereas at position 56, only lysine is functional. Residue Asn20 is the only other residue in TRAP that is located on the line formed by residues 37, 56 and 58, and virtually any amino acid residue is functional at position 20. We propose that RNA wraps around TRAP by interacting with residues Lys37, Lys56 and Arg58. # 1997 Academic Press Limited

*Corresponding author

Keywords: tryptophan biogenesis; TRAP; tryptophan binding; RNA binding; site-directed mutagenesis

Introduction In Bacillus subtilis, an RNA binding protein called TRAP (trp RNA-binding attenuation protein) Present addresses: K. Militello, Department of Microbiology, State University of New York at Buffalo, Buffalo, NY 14260, USA. C. Baumann, NCI-LRBGE, Bldg 41/Room B602, 41 Library Dr. MSC 5055, Bethesda, MD 20892, USA. Abbreviations used: b-gal, b-galactosidase; IPTG, isopropy-b,D-thiogalactopyramoside; EmR, erythromycin resistance; PCR, polymerase chain reaction; TRAP, trp RNA-binding attenuation protein; WT, wild-type; X-gal, 5-bromo-4-chloro-3-indolyl-b,D-galactopyranoside. 0022±2836/97/300696±15 $25.00/0/mb971149

negatively regulates expression of the tryptophan biosynthetic (trp) genes in response to the intracellular levels of L-tryptophan (Shimotsu et al., 1986; Kuroda et al., 1988; Gollnick, 1994). By binding to several speci®c RNA targets in a tryptophan dependent manner, TRAP regulates both transcription of the trpEDCFBA operon (Shimotsu et al., 1986) as well as translation of trpE (Merino et al., 1995) and trpG (Yang et al., 1995). Transcription of the trp operon is controlled by an attenuation mechanism involving competition between two mutually exclusive RNA stem-loop structures in the leader transcript, termed the terminator and antiterminator (Shimotsu et al., 1986). In the presence of excess tryptophan, TRAP is activated to # 1997 Academic Press Limited

Alanine-scanning Mutagenesis of TRAP

bind to a target site within the trp leader RNA (Otridge & Gollnick, 1993; Babitzke et al, 1994). This binding facilitates formation of the terminator, which halts transcription of the operon. In limiting tryptophan, TRAP does not bind, the antiterminator forms and transcription continues into the structural genes. Translation of trpE is controlled by TRAP binding to the same target site in the leader region of read-through trp mRNAs that have escaped termination (Shimotsu et al., 1986; Merino et al., 1995). This binding is proposed to induce formation of a secondary structure that sequesters the trpE ribosome binding site, thereby inhibiting translation initiation. TRAP also regulates translation of trpG, the only trp gene in B. subtilis that is not in the trp operon, and is instead located within a folic acid operon (Slock et al., 1990). In this case, TRAP binds to a site that overlaps the trpG ribosome binding site (Yang et al., 1995). A mechanism involving direct competition between ribosomes and tryptophan-activated TRAP for binding to trpG mRNA has been proposed (Yang et al., 1995) and recent in vitro studies by Du et al. (1997) support this hypothesis. A number of studies both in vivo (Antson et al., 1995) and in vitro (Babitzke et al., 1994, 1995, 1996) have demonstrated that TRAP recognizes singlestranded RNA targets containing multiple repeats of GAG and UAG trinucleotides. The B. subtilis trp leader contains 11 G/UAG repeats. Eight G/UAGs and one AAG are found near the translational start site of trpG (Slock et al., 1990) and footprinting studies (Du et al., 1997) indicate that the AAG also interacts with TRAP. The G/UAG repeats of TRAP binding sites are always separated from each other by several nonconserved ``spacer'' residues. Several studies have shown that proper spacing of the G/UAG repeats is crucial for TRAP binding and that two nucleotides is the optimum spacer length (Antson et al., 1995; Babitzke et al., 1994, 1995). Recently, in vitro selection (SELEX) experiments con®rmed that the optimum spacer length is two nucleotides (Baumann et al., 1997). Moreover, over 90% of the spacer nucleotides in the selected RNAs were pyrimidines. This bias appears to be related to the reduced ability of pyrimidines to stack in the RNA as Baumann et al. (1997) showed that when trp leader RNA binds to TRAP the bases become unstacked. Thus the spacers appear to both maintain proper spacing between the G/UAG repeats and provide ¯exibility in the RNA. The three dimensional crystal structure of TRAP complexed with L-tryptophan contains 11 identical subunits arranged in a symmetric ring (Antson et al., 1995). The TRAP 11-mer is composed of 11 seven-stranded antiparallel b-sheets, each sheet composed of four b-strands from one subunit and three strands from the adjacent subunit (see Figure 1). The two sides of the protein ring are asymmetric, with one side being narrower than the other. We have designated the narrower side of the

697 protein ring as the ``front'' side of TRAP and the larger, ¯atter side as the ``back''. The RNA binding activity of TRAP is activated by the binding of 11 tryptophans in a positively cooperative manner to the TRAP 11-mer (Antson et al., 1995). Each tryptophan binding site of TRAP consists of a hydrophobic pocket that surrounds the indole ring, and nine hydrogen bonds that form between the bound tryptophan and amino acid residues of TRAP. Each tryptophan binds between two loops consisting of residues 25 to 33 (designated as loop BC since it connects b-strands B and C; see Figure 1) on one subunit and residues 49 to 52 (loop DE) on the adjacent subunit of TRAP. These two loops appear to be ¯exible since they do not interact with other parts of TRAP. We therefore proposed previously that conformational changes in these loops would be involved in activating TRAP to bind RNA (Antson et al., 1995). Based on the ®ndings that TRAP has 11 identical subunits and that its RNA targets contain multiple trinucleotide repeats, including 11 in the trp leader, we proposed a model in which RNA wraps around the TRAP protein with each trinucleotide repeat recognized by one subunit, or a combination of adjacent subunits (Antson et al., 1995). TRAP does not have sequence homology to other characterized RNA binding motifs. To investigate the functional domains of TRAP and, in particular, to search for residues involved in binding RNA, we have used alanine-scanning mutagenesis (Cunningham & Wells, 1989), substituting each solvent-exposed amino acid residue in TRAP with alanine. We examined surface exposed residues for several reasons: ®rst these represent likely candidates for residues that would interact with RNA, and second, substitutions of these residues is less likely to interfere with the folding and the overall structure of TRAP. Each mutant TRAP was initially characterized in vivo for the ability to regulate trp gene expression as assayed using a trpE0 -0 lacZ fusion in B. subtilis driven by the trp promoter and regulatory region. Mutant TRAP proteins that showed altered regulation in vivo were puri®ed and tested in vitro for tryptophan binding and RNA binding. Most of the alanine substitutions that altered the tryptophan binding characteristics of TRAP are located on, or near, loop BC and loop DE supporting the role of these loops in tryptophan binding (Antson et al., 1995). Three residues, Lys37, Lys56 and Arg58, were found that when substituted with alanine, had little or no effect on tryptophan binding but severely reduced the ability of TRAP to bind to trp leader RNA. In the crystal structure, these three residues are aligned on the outer edge of the front side of TRAP. These ®ndings support the model in which the RNA triplet motifs separated by spacers wrap around the TRAP oligomer (Antson et al., 1995; Babitzke et al., 1994, 1995) and suggest that each G/UAG trinucleotide interacts with a binding site consisting of residue 37 of one

698 subunit and residues 56 and 58 of the adjacent subunit.

Results Alanine scanning of the surface residues of TRAP In the three dimensional crystal structure of TRAP complexed with L-tryptophan, 44 of the 75 amino acid residues in each subunit have sidechains that are exposed to the solvent (Antson et al., 1995). These surface residues present likely candidates for amino acid residues that interact with RNA and are unlikely to be involved in subunitsubunit interaction. Indeed, we did not identify any of our alanine-substituted TRAP mutants that appeared to be defective in subunit assembly. We systematically replaced 38 of these surface residues with alanine to examine the role of each in the functions of TRAP. Alanine was chosen in order to minimize the possibility of drastically interfering with the folding and/or the conformation of the protein (Cunningham & Wells, 1989). Four glycine residues (residue 18, 27, 41 and 68) were not changed due to the difference in backbone ¯exibility between alanine and glycine. In addition, we did not replace Thr49 because the crystal structure shows that the side-chain hydroxyl group of this residue forms a hydrogen bond with the bound tryptophan (Antson et al., 1995). Regulatory functions of mutant TRAP proteins in B. subtilis To quickly assess the effect of each alanine substitution on the function of TRAP, we tested the ability of each mutant protein to regulate, in response to tryptophan, a single-copy trpE0 -0 lacZ translational gene fusion under the control of the trp promoter and leader region. This analysis was carried out in B. subtilis PGBS4233, which contains a deletion of the mtrB gene that encodes TRAP, and therefore shows constitutively high b-galactosidase (b-gal) levels (Table 1). Transforming this strain with the vector pHB201 had little or no effect on b-gal expression. However, expressing wildtype TRAP in this strain resulted in 42-fold downregulation of the reporter fusion in response to tryptophan (Table 1). The ®vefold reduction of b-gal expression in this strain grown in the absence of exogenous tryptophan (380 U, Units de®ned as in Miller, 1972) as compared to the parental strain PGBS4233 (1840 U) likely re¯ects the intracellular level of tryptophan in the strain, which is trp‡. We expressed each alanine-substituted TRAP in PGBS4233 and compared regulation of the trpE0 0 lacZ fusion to that when wild-type TRAP is expressed. Mutant proteins that showed similar regulation (greater than 13-fold) to the wild-type were generally not investigated further. We puri®ed the mutant TRAP proteins that showed reduced regulation ability in vivo. Tryptophan binding and RNA

Alanine-scanning Mutagenesis of TRAP

binding activities of these proteins were then tested in vitro to determine which of these functions were affected by the alanine substitution. Among the 38 alanine mutant proteins we analyzed, 15 (D8A, T25A, R26A, K31A, H33A, H34A, E36A, K37A, K40A, E42A, F48A, H51A, K56A, R58A and E60A) showed signi®cant changes in the ability to regulate the trpE0 -0 lacZ fusion, whereas the other 23 mutant proteins retained regulatory functions similar to those of wild-type TRAP (Table 1). Of the 15 mutant proteins with altered regulatory function, ®ve (T25A, R26A, K31A, H33A and H51A) replace residues located on the two loops (loop BC and loop DE; Figure 1) that form part of the tryptophan binding site (Antson et al., 1995). Another seven (H34A, E36A, K37A, K40A, E42A, K56A and R58A) contain alanine substitutions on the two b-strands (strand C and strand E) and the b-turn CD (residues 39 to 42) connecting these two loops. The remaining three regulatory-defective mutant TRAP proteins contain substitutions in residues lining the central channel (D8A and F48A) and the back (E60A) of the protein. The ®nding that most (12/15) of the alanine-substitutions that alter the in vivo regulatory functions of TRAP are located on, or directly linked to, loop BC and loop DE con®rms the importance of these two loops for proper TRAP function. Therefore, in addition to the 15 proteins with defective regulatory function in vivo, we also puri®ed and analyzed nine mutant TRAP proteins (N20A, I22A, T28A, D29A, T30A, F32A, S35A, D39A and E50A, labeled { in Table 1) containing alanine substitutions of residues on, or linked to, these two loops, even though these mutant proteins demonstrated normal regulatory functions in vivo. Tryptophan binding We measured the tryptophan binding activities of the 24 puri®ed mutant TRAP proteins mentioned above (Table 1). Wild-type TRAP binds L-tryptophan with an S0.5 of 11 mM and a Hill coef®cient (n) of 1.5 (Table 1). The S0.5 value is the concentration of tryptophan required to reach 50% of the saturation level of bound tryptophan, and re¯ects the overall af®nity of the protein for tryptophan. Ten of the mutant proteins (N20A, T28A, T30A, S35A, K37A, E42A, E50A, K56A, R58A and E60A) had similar S0.5 values to the wild-type TRAP. Eight mutant TRAP proteins (D8A, T25A, R26A, K31A, H33A, H34A, E36A and F48A) showed no detectable tryptophan binding. Three mutant TRAP proteins (I22A, K40A and H51A) showed increased S0.5 values of 46 mM, 33 mM and 78 mM, respectively, while three others (D29A, F32A and D39A) have lower S0.5 values than wildtype TRAP (Table 1). We also observed changes in the cooperativity of tryptophan binding in nine alanine-substituted mutant proteins (I22A, D29A, F32A, S35A, D39A, K40A, E42A, E50A and E60A).

699

Alanine-scanning Mutagenesis of TRAP Table 1. Characteristics of alanine mutant TRAP proteins in vivo

in vitro

b-Galactosidase activities (U)a

PGBS4233 ‡pHB201 ‡TRAP WT D8A K13A V15A E16A D17A {N20A {I22A T25A R26A {T28A {D29A {T30A K31A {F32A H33A H34A H34A {S35A E36A K37A {D39A K40A E42A Q47A F48A {E50A H51A K56A R58A E60A L62A Q64A Y67A E69A M70A K71A E73A K74A K75A

ÿTrp

‡Trpe

1840 1910

1850 1920

1 1

380 1300 420 440 230 250 340 670 1280 1210 400 480 240 1100 180 1770

9 1190 18 12 16 16 10 9 1510 1000 10 38 11 970 7 1920 1920 1260 18 1250 1060 10 190 240 15 1820 11 120 2360 480 280 11 9 9 18 6 18 9 11 9

42 1 23 37 14 16 34 74 1 1 40 13 22 1 26 1 1 1 34 1 1 40 3 2 27 1 17 3 1 1 3 29 30 41 26 23 14 14 19 16

1290 610 1220 890 400 550 420 410 1890 190 330 2330 590 840 320 270 370 470 140 260 130 210 140

Trp bindingb S0.5 (mM) n

(ÿTrp/ ‡ Trp)

11

13 46 12 5 16 5

10 12 7 33 14 15 78 10 12 12

NBf

NB NB 1.3 NB NB NB NB NB

NB

RNA bindingc Kd (nM)

1.5

0.8

1.6 1.9

0.7 1.0

1.6 2.1 0.7

1.0 0.1 f

1.2

0.7

2.8 1.5 2.1 3.3 1.8

495 0.1 106 1.5

1.8 1.6 1.6 1.3 1.9

0.5 5.8 626 677 0.8

Location

c b b b b f f f f f f f f f f f f f f f b f c c f f f f b b b b b b b b b f

a b-Galactosidase activity units de®ned as by Miller (1972). Each value is the average of at least two different experiments, each done in triplicate and standard deviations were less than 20% of the values shown. b The tryptophan binding data were ®t to the Hill equation (see Materials and Methods). Each value is the average of at least two different experiments. Standard deviations were less than 20% of the values shown. S0.5 is de®ned as the concentration of free L-tryptophan at which the concentration of bound L-tryptophan reaches 50% of the saturation level. The Hill coef®cient, n, describes the cooperativity of the interaction.d c The dissociation constant Kd was measured using ®lter binding in the presence of 1 mM L-Tryptophan. Each value is the average of three different experiments, each done in duplicate. Standard deviations were less than 25% of the values indicated. d The locations of the surface residues are divided into front side (f), back side (b) and central channel (c). e L-tryptophan (50 mg/ml) was present in the growth medium. f NB indicates no detectable tryptophan binding activity in the equilibrium dialysis assay. {Nine mutant TRAP proteins with normal in vivo regulatory functions and containing alanine substitutions of residues on, or linked to, loop BC and loop DE.

Most of these (8/9) showed higher Hill coef®cients (n), indicative of greater cooperativity in tryptophan binding, whereas only one mutant protein, F32A, showed less cooperativity (Table 1). As mentioned above, we puri®ed and characterized nine mutant TRAP proteins that showed wildtype-like levels of in vivo regulation. Almost all

(8/9) of these mutant proteins bound tryptophan as well as (or better than) wild-type TRAP. In addition, all of these mutant TRAP proteins bound trp leader RNA with af®nities similar to or better than wild-type TRAP (Table 1; see below). These ®ndings indicate that measuring the in vivo regulatory activity is a reliable assay to screen our mutant proteins. The only exception, I22A, had a

700

Alanine-scanning Mutagenesis of TRAP

Figure 1. Ribbon diagrams of TRAP. A, One complete subunit and parts of the two adjacent subunits in TRAP are shown. The b-strands are depicted as arrows and the bound tryptophans as ball and stick models. The three different subunits are shown in gray, white and black respectively. The seven b-strands of the middle subunit are labeled A to G as, by Antson et al. (1995). Loop BC (residues 25 to 33), loop DE (residues 49 to 52) and b-turn CD (residues 39 to 42) of the middle subunit are indicated. B, Diagram of TRAP 11-mer structure. Individual subunits are shown in gray, white or black. The 11 bound tryptophan molecules are shown as van der Waals spheres. Both diagrams were generated by MOLSCRIPT (Kraulis, 1991).

fourfold decreased af®nity for tryptophan. This decreased af®nity may account for the nearly twofold increase in b-gal expression in cells grown in the absence of tryptophan but apparently is not large enough to affect expression in cells grown in excess tryptophan (Table 1). The results of these tryptophan binding studies indicate that most of the in vivo defective mutant TRAP proteins have lost regulatory ability due to decreased af®nity for tryptophan. Moreover, most (13/18) of the alanine substitutions that affected tryptophan binding, either af®nity or cooperativity, or both, are located on loop BC and loop DE, or on the b-strands and turn connecting these two loops (Figure 2) con®rming the importance of these structures in binding tryptophan. RNA binding Because the RNA binding activity of TRAP depends on binding L-tryptophan as a cofactor, we only measured the RNA binding activity of the 16 alanine-substituted mutant proteins that showed detectable tryptophan binding. We used both gel mobility-shift and ®lter-binding assays to analyze RNA binding to these proteins. The mobility shift assay allows visualization of speci®c complexes, whereas the ®lter-binding assay provides more quantitative data and was used to measure the af®nity of each mutant TRAP protein for trp leader RNA in the presence of excess tryptophan (Table 1). Of the 16 mutant proteins we analyzed, nine

(N20A, I22A, T28A, T30A, F32A, S35A, E42A, E50A and E60A) bind trp leader RNA with nearly the same af®nity (Kd values from 0.5 nM to 1.5 nM; Table 1) as wild-type TRAP, ®ve (K37A, K40A, H51A, K56A and R58A) showed decreased RNA binding af®nity and two mutant proteins (D29A and D39A) bind RNA with higher af®nity (Table 1). Two alanine-substituted TRAP proteins, D29A and D39A, were found to have nearly tenfold higher af®nity for trp leader RNA than wild-type TRAP, each with a Kd of 0.1 nM (Table 1). Interestingly, both of these alanine-substituted TRAP proteins regulate the trpE0 -0 lacZ fusion in vivo similarly to wild-type TRAP. While aspartic acid residues are replaced with alanine in both of these proteins, residues 29 and 39 are rather far apart on the surface of TRAP (see Figure 2). It is interesting that both of these mutant proteins formed two distinct complexes with trp leader RNA, seen as two different shifted bands in mobility-shift gels (Figure 3). Wild-type TRAP (Figure 3), as well as all the other alanine-mutant proteins tested (data not shown), formed only one complex with trp leader RNA. Previous studies have indicated that wild-type TRAP from either B. subtilis (Baumann et al., 1996) or from Bacillus stearothermophilus (A. Antson & P. Gollnick, unpublished results) binds to trp leader RNA with a stoichiometry of 2 TRAP:1 RNA. When gel slices containing the two different D29A TRAP-RNA complexes were excised and analyzed (see Materials and Methods), the molar ratios of TRAP to trp leader RNA were found to be

Alanine-scanning Mutagenesis of TRAP

701

Figure 3. Mobility shift analysis of D29A and D39A mutant TRAP proteins binding to trp leader RNA 2-138. Each lane contained 50 nM of 32P-labeled RNA and the indicated amount of TRAP in the presence of 1 mM Ltryptophan. Bands labeled C correspond to the wildtype TRAP-RNA complex whereas C1 and C2 correspond to D29A or D39A TRAP-RNA complexes. RNA indicates the position of the unbound trp leader RNA.

Figure 2. Location of TRAP residues important for tryptophan binding. The backbone traces of three adjacent subunits of TRAP are shown with the top and bottom subunits in dark blue and the middle subunit shown in light blue. The b-strands C and D in the middle subunit are shown as rectangular ribbons and loops BC, DE and b-turn CD are highlighted as oval ribbons. The bound tryptophans are shown in red ball and stick models. The eight residues shown in yellow, Asp8, Thr25, Arg26, Lys31, His33, His34, Glu36 and Phe48, have no detectable tryptophan binding when replaced by alanine. The nine residues shown in orange, Ile22, Asp29, Phe32, Ser35, Asp39, Lys40, Glu50, His51, and Glu60, have altered tryptophan binding af®nity and/or cooperativity when replaced by alanine. Residues 43 to 47 on b-strand D were not substituted because they are the buried subunit interface; nor was Leu38, because its side-chain is not solvent exposed. Glycine residues Gly27 and Gly41 were also not replaced in this study. Alanine substitution of residues Thr28, Thr30 and Lys37 did not alter tryptophan binding. The Figure was generated by the Insight1 II molecular modeling system from Biosym/MSI, San Diego (Insight1II User Guide, 1995).

0.8  0.2 to 1 for C1 and 3.7  0.3 to 1 for C2 (Figure 3). While we did not analyze the complexes formed with D39A TRAP, we expect they are similar to those with D29A, given the similar patterns seen in mobility-shift gels (Figure 3). These results, together with the tryptophan-binding data (Table 1), indicate that replacing aspartic acid residues 29 or 39 with alanine affects both tryptophan binding and RNA binding such that both mutant proteins bind both ligands better than wild-type TRAP. Most intriguingly these substitutions result in the formation of two different complexes and, at

least for D29A, changes the stoichiometry of the TRAP-RNA complexes. Five mutant proteins, K37A, K40A, H51A, K56A and R58A, showed decreased af®nity for trp leader RNA (Table 1). Alanine substitution of His51 had only a small (sevenfold) effect, replacement of Lys40 had a moderately large (130-fold) effect, while substitutions of Lys37, Lys56 and Arg58 had more drastic effects (500 to 700-fold). In the H51A and K40A mutant proteins, tryptophan binding is also affected in addition to RNA binding with S0.5 values of 78 mM and 33 mM, respectively. However, because we assayed RNA binding in the presence of excess (1 mM) L-tryptophan, the reduced af®nity for RNA of proteins H51A and K40A cannot be simply attributed to inadequate tryptophan binding. Residue His51 is located on loop DE and its side-chain may interact with loop BC, both of which are involved in tryptophan binding. The slight decrease in af®nity for RNA seen for H51A (Table 1) may therefore result from suboptimal activation upon tryptophan binding. Lys40 is loÊ from the cated on the b-turn CD and is 18 A bound tryptophan (Figures 1 and 2). It is not clear whether the reduced af®nity of K40A for RNA is a direct or an indirect effect (see Discussion). One of the objectives of this study was to locate residues in TRAP that are speci®cally involved in interacting with RNA. Replacing such a residue with alanine should result in a mutant TRAP protein that (1) is unable to properly regulate the trpE0 -0 lacZ reporter fusion in vivo, (2) still binds tryptophan normally and (3) has decreased af®nity for trp leader RNA. Of the 38 alanine substitutions we created, three, K37A, K56A and R58A (shown in bold in Table 1, ®t these criteria. These results imply that the side-chains of these three residues are speci®cally involved in interacting with the bound RNA. Examination of the positions of resi-

702

Alanine-scanning Mutagenesis of TRAP

Figure 4. Representation of the K37K56R58 RNA binding motif on the surface of the TRAP 11-mer ring using CPK models. Side (on the left) and front (on the right) views are shown. Lys37, Lys56 and Arg58 residues are shown in red, Asn20 residues are shown in magenta and all the other surface residues of TRAP are shown in blue. Adjacent subunits of TRAP in the side view (left) are distinguished from each other as dark blue and light blue. This Figure was generated by the Insight1 II molecular modeling system from Biosym/MSI, San Diego (Insight1II User Guide, 1995).

dues 37, 56 and 58 in the crystal structure of tryptophan-activated TRAP shows that these three residues are all on the front side of TRAP and form a line around the outer edge of the protein (Figure 4). The positions of these putative RNA binding amino acid residues are consistent with our model, in which the RNA wraps around TRAP to encircle the protein ring. One possible explanation for the loss of RNA binding seen for K37A, K56A and R58A TRAP could be that these mutant proteins fail to fold properly or fail to undergo the necessary conformational changes induced by tryptophan binding in order to activate TRAP to bind RNA. Because all three mutant proteins bind tryptophan with nearly identical af®nity and cooperativity as wildtype TRAP, it appears that these polypeptides fold well and associate into 11-mers. Moreover, circular dichroism (CD) spectra of all three of these mutant proteins, in the absence and presence of tryptophan, are nearly identical to those of wild-type TRAP (data not shown) indicating that there are no large changes in the secondary structure of any of these TRAP mutants as compared to wild-type TRAP. Together, these data suggest that the loss of

RNA binding activity of these three alanine-substituted proteins is due to removal of speci®c contacts between amino acid residue side-chains and the RNA. Mutagenic analysis of positions 37, 56 and 58 As described above, substituting alanine for residues Lys37, Lys56 or Arg58 appears to remove amino acid residue side-chains that interact with trp leader RNA. To better understand the nature of the interaction between TRAP and RNA at these positions, we tested which amino acid residues at each position allow proper TRAP function. Random nucleotide sequences were introduced into the three positions of each of these three codons in mtrB by site-directed mutagenesis. This approach created three libraries of plasmids containing codons for all 20 amino acids at each position. Plasmids that produced functional TRAP proteins were selected in PGBS4233 as white colonies on X-gal plates containing tryptophan (see Materials and Methods). We then sequenced the plasmids from all the white colonies representing each position (Table 2). At positions 37 and 58, we found only arginine or lysine codons, while at position 56

Table 2. Codons at positions 37, 56 and 58 making TRAP active in vivo Position 37 13  CGG 1  CGA 4  CGT 1  AGA 4  AGG 3  AAAa 2  AAG 28b a b

Position 56 Arg Arg Arg Arg Arg Lys Lys

5  AAAa 11  AAG 16b

Position 58

Lys Lys

1  CGA 2  CGT 8  AGAa 2  AGG 1  AAA 1  AAG 15b

Codons present in the wild-type mtrB gene. The total number of sequenced plasmids at each individual position.

Arg Arg Arg Arg Lys Lys

703

Alanine-scanning Mutagenesis of TRAP Table 3. Characteristics of mutant TRAP proteins with changes at position Lys37, Lys56 and Arg58 in vivo

a

TRAP

ÿTrp

WT K37Aa K56Aa R58Aa K37R R58K K56R K37RR58K K37Q R58Q

380 890 2330 590 600 600 1520 570 1320 1790

in vitro

b-Galactosidase activities (U) ‡Trp (ÿTrp/ ‡ Trp) 9 1060 2360 480 10 10 1500 12 1210 1550

42 1 1 1 60 60 1 48 1 1

Trp binding S0.5 (mM)

n

RNA binding Kd (nM)

11 12 12 12

1.5 1.5 1.3 1.3

0.8 495 677 677

9

2.6

578

12 10

1.4 1.5

181 133

These data are reproduced from Table 1 and are included for comparison.

only lysine codons were present. In agreement with these ®ndings, we found that the mutant proteins K37R, R58K and the double-mutant protein K37RR58K regulated the trpE0 -0 lacZ fusion well in vivo whereas K56R did not (Table 3). Moreover, the ®nding that puri®ed K56R TRAP protein binds tryptophan similar to wild-type TRAP but has signi®cantly lost RNA binding activity (Table 3) con®rms the side-chain of Lys56 is involved in RNA binding but not tryptophan binding and that the lysine at position 56 is critical for interacting with trp leader RNA. The ®nding that both basic amino acid residues, lysine and arginine, function at positions 37 and 58 suggests that a positive charge is important at these two positions. To test this hypothesis, we substituted glutamine at these two positions. Glutamine has a similar size as lysine and arginine, and like these amino acids it can donate hydrogen bonds; however, glutamine lacks the positive charge of the basic amino acids. Both K37Q and R58Q TRAP failed to regulate the trpE0 -0 lacZ fusion in vivo (Table 3). Both mutant proteins bound tryptophan as well as wild-type TRAP but bound trp leader RNA 150 to 200-fold less well (Table 3). These data imply that the positive charge on Lys37 and Arg58 does play a role in binding RNA. However, hydrogen bonding and/or hydrophobic interactions may also be involved, since substituting glutamine at either position allowed three to fourfold better RNA binding than the corresponding alanine-substituted proteins (Table 1). Analyzing the role of Asn20 The alignment of residues Lys37, Lys56 and Arg58 on the surface of TRAP (Figure 4) suggests that RNA binds by wrapping around TRAP and interacts with the protein along this line. Asn20 is the only other amino acid residue that falls on this line around TRAP (Figure 4). The location of this residue suggested it might also interact with RNA, although substitution with alanine did not affect TRAP activities (Table 1). To characterize further the role of position 20 in RNA binding, we tested which amino acid residues are functional at this

position. In this case, we used site-directed mutagenesis to introduce all possible codons into this position, introduced these plasmids into PGBS4233 and selected colonies containing non-functional mutant TRAP proteins as blue colonies on X-gal plates. The only missense mutations at position 20 that resulted in inactive TRAP encoded lysine. When we attempted to purify N20K TRAP, the protein was insoluble, suggesting that this substitution resulted in a misfolded protein. We con®rmed that several of the changes including N20E, N20V and N20F all resulted in functional TRAP proteins, although N20F does show slightly decreased regulation in vivo (Table 4). Thus, in stark contrast to positions 37, 56 and 58, position 20 can be occupied by nearly any amino acid residue.

Discussion Importance of the front surface of TRAP The crystal structure shows that TRAP has a toroidal shape (Antson et al., 1995). The protein ring formed by 11 identical subunits is asymmetric and can be divided into three different surfaces: a narrower ``front'' side, a broader ``back'' side and a Ê diameter. The back side of central channel of 23 A TRAP is rather ¯at, whereas the front side has a more varied surface structure. TRAP is activated to bind RNA by the binding of 11 tryptophans to pockets on the front of the molecule. Two loops, BC and DE, on adjacent subunits on the front of TRAP make up part of the tryptophan binding site (Figure 1) and we proposed previously that these loops participate in the activation of TRAP to bind RNA (Antson et al., 1995). In agreement with inferences from the structural data, most (18/21) of the alanine substitutions on the front side of TRAP altered its functions. Among these 18 mutant proteins, 15 changes affected tryptophan binding, and the three substitutions that speci®cally decrease RNA binding, K37A, K56A and R58A, are also all on the front of TRAP. Thus, our mutational analysis indicates that the front surface of TRAP is most important for both tryptophan and RNA binding.

704

Alanine-scanning Mutagenesis of TRAP

Table 4. Analysis of mutant TRAP proteins with changes at position 20 in vivo

a

TRAP

ÿTrp

WT N20A N20E N20F N20V N20Ka

380 340 230 190 270 1700

in vitro

b-Galactosidase activities (U) ‡Trp (ÿTrp/ ‡ Trp) 9 10 11 25 12 1700

42 34 21 8 23 1

S0.5 (mM)

Trp bindingRNA binding n

Kd (nM)

11 13

1.5 1.6

0.8 0.7

The protein was found to be insoluble in E. coli.

Comparison of the deduced amino acid sequences of TRAP from three different Bacillus species also supports the conclusion that the front surface of TRAP is most important for its function. Overall, these three proteins from B. subtilis (Gollnick et al., 1990), Bacillus pumilus (Hoffman & Gollnick, 1995) and B. stearothermophilus (X. C., C. B. & P. G., unpublished results), are highly homologous, with over 70% identity between them. However, comparing the conservation of surface exposed residues in TRAP, we found 90% identity among residues on the front side of TRAP, whereas there is only 40% and 67% identity for residues on the back side and in the central channel, respectively. In contrast to the results seen for the front side of TRAP, only one (E60A) of the 13 alanine substitutions on the back side of TRAP signi®cantly altered regulation of the trpE0 -0 lacZ fusion in vivo (Table 1). Furthermore, the in vitro RNA binding and tryptophan binding activities of puri®ed E60A protein were nearly the same as those of wild-type TRAP (Table 1). The slight increase in the cooperativity of tryptophan binding is not likely to cause the reduced in vivo regulation. These results suggest that the poor in vivo regulation seen with this mutant protein may be due to a low steadystate level of this protein. The crystal structure of the TRAP-tryptophan complex shows that two residues (Gln47 and Thr49) exposed in the central channel form hydrogen bonds with the bound tryptophan (Antson et al., 1995). In addition, the two alanine substitutions (D8A and F48A) in the central channel that alter regulation in vivo, fail to bind tryptophan in vitro even though their side-chains do not directly contact the bound tryptophan in the crystal structure (Table 1). Together these results demonstrate that residues in the central channel are important for tryptophan binding but provide no evidence of any direct role for this part of TRAP in RNA binding. Based on the size of the central channel, we had proposed previously that the DNA double helix might be threaded through the channel and that this arrangement could be important for proper attenuation control of the trp operon by TRAP (Antson et al., 1995). Our alanine-scanning results provide no evidence either for or against this hy-

pothesis. However, we have recently determined the crystal structure of TRAP from B. stearothermophilus complexed with L-tryptophan (Antson et al., unpublished results). The characteristics of the central channels of these two homologous proteins are quite different. The hole is smaller in B. stearotherÊ diameter), making it too mophilus TRAP (17 A small for a double-stranded B-form DNA to ®t through. In addition, the electrostatic characteristics of the residues surrounding the channel are opposite in the two proteins, being predominately acidic in B. subtilis TRAP and basic in B. stearothermophilus TRAP. Thus, it seems unlikely that TRAP function involves passing the DNA double helix through the central channel, such as occurs with several other ring-like proteins (Kong et al., 1992; Krishna et al., 1994; Lima et al., 1994; Naktinis et al., 1996). Rather, our ®ndings indicate that residues lining the central channel of TRAP play a role only in tryptophan binding. Tryptophan binding and activation of TRAP TRAP is activated to bind RNA by binding 11 tryptophan molecules, each of which binds between two loops, termed BC and DE, on adjacent subunits (Figure 1). Of the alanine-substituted TRAP proteins we characterized, 18 (D8A, I22A, T25A, R26A, D29A, K31A, F32A, H33A, H34A, S35A, E36A, D39A, K40A, E42A, F48A, E50A, H51A and E60A) showed altered tryptophan binding properties, either af®nity, cooperativity, or both (Table 1). All of these substitutions alter amino acid residues whose side-chains do not directly interact with the bound tryptophan in the crystal structure (Antson et al., 1995). We substituted only one residue, Thr30, whose side-chain directly interacts with the bound tryptophan. The hydroxyl group of Thr30 forms a hydrogen bond with the amino group of tryptophan. Replacing this residue with alanine had little or no effect on tryptophan binding (Table 1). This result is not surprising given that the amino group of tryptophan forms three additional hydrogen bonds to other amino acid residues in TRAP. Most (15/18) of the alanine substitutions that affected tryptophan binding replace amino acid residues on the two loops (BC and DE) mentioned previously as well as b-strand C (residues 34 to

Alanine-scanning Mutagenesis of TRAP

38), and the b-turn CD, (residues 39 to 42; Figure 2). Thus, in addition to the two loops that directly interact with tryptophan, the structures connecting these two loops are also important for proper interaction with tryptophan. In particular, b-turn CD, which connects the two b-strands between loops BC and DE appears to be important for both tryptophan and RNA binding. Alanine substitutions of Asp39 or Lys40 both resulted in altered tryptophan binding and RNA binding activities (Table 1; Figure 2). One possibility is that this turn may serve as a ``hinge'' with one end ``®xed'' to b-strand D, which is buried in the interface between adjacent subunits, while the other end of the turn is connected to b-strand C, whose conformation may change upon binding tryptophan, thus activating TRAP to bind RNA (note Lys37, one of the three residues identi®ed as being involved in RNA binding is on b-strand C). Gly41, which is conserved in the three TRAP proteins from B. subtilis, B. pumilus and B. stearothermophilus (Hoffman & Gollnick, 1995; X. C., C. B. & P. G., unpublished results), is also on this turn, and might allow the ¯exibility of this turn necessary for the conformational change to occur. None of the alanine-substituted TRAP mutants in this study was constitutively active in down-regulating the trpE0 -0 lacZ fusion in vivo in the absence of tryptophan. This may be a limit of the alaninesubstitution approach. We have found one TRAP mutant, H51Y, that binds trp leader RNA equally well in the absence or presence of tryptophan with a Kd of 5 nM (M. Y. & P. G., unpublished results). H51Y TRAP retains tryptophan binding activities, both af®nity (S0.5 ˆ 11 mM) and cooperativity (n ˆ 1.6), that are nearly identical to wild-type TRAP. Surprisingly, this TRAP mutant does not show greater repression of the trpE0 -0 lacZ fusion in the absence of tryptophan in vivo. RNA binding: the KKR structural motif We have identi®ed three residues in TRAP, Lys37, Lys56 and Arg58, as being speci®cally involved in RNA binding. Eleven clusters of this KKR motif, containing Lys37 from one subunit, and Lys56 and Arg58 from the adjacent subunit, fall on a single line encircling the TRAP oligomer (Figure 4). This spatial arrangement supports the model we previously proposed in which the trp leader RNA wraps around the 11-mer of TRAP (Antson et al., 1995). Furthermore, we now propose that the RNA encircles TRAP along the line formed by Lys37, Lys56 and Arg58, directly contacting these three residues. This putative RNA binding motif consists of 11 KKR clusters each separated by one amino acid residue, Asn20. Similarly, the optimum RNA target for TRAP is composed of 11 G/UAG repeats each separated by two spacer nucleotides (Babitzke et al., 1994, 1995). Virtually any amino acid can be substituted at position 20 in TRAP without affecting RNA binding, and the identities of the bases in the RNA spacers are not

705 crucial for binding to TRAP (Babitzke et al., 1995; Baumann et al., 1997). These observations suggest that in the TRAP-RNA complex, Asn20 may be juxtaposed to, but not form any speci®c interactions with, the spacer nucleotides. The distance between the a-carbon atoms of Ê . Based Lys37 residues on adjacent subunits is 20 A on transfer RNA coordinates (Rould et al., 1989), we have found that the phosphate-to-phosphate length of ®ve nucleotides (G/UAG trinucleotide recognition motif plus a dinucleotide spacer) of Ê and single-stranded RNA can vary between 8.5 A Ê depending on its conformation, with an 29.2 A Ê (Antson et al., 1995). average distance of 20.2 A Thus it is reasonable to propose that an RNA target of TRAP consisting of 11 G/UAG repeats each separated by two nucleotides could wrap around TRAP and interact with these KKR motifs. Although we have identi®ed three residues in TRAP as being speci®cally involved in RNA binding, we can not exclude the possibility that there are additional residues involved in the interaction of TRAP with RNA. Using this mutagenic approach, we are not able to detect residues where the peptide backbone contacts the RNA. As long as the mutant protein does not have an altered backbone conformation as the consequence of alanine substitution, these replacements will not change RNA binding. Furthermore, if a mutant protein does not bind tryptophan, such as any of the eight alanine-substituted mutant proteins that show no detectable tryptophan binding (Table 1), it is not possible to assess the role of these residues in RNA binding activity. The K40A mutant protein has reduced af®nity for both tryptophan and trp leader RNA (threefold and 100-fold, respectively; Table 1). Ê from the a-carThe a-carbon of Lys40 is only 9 A bon of Lys37; thus it seems Lys40 might also participate directly in RNA binding. However, given the effect of this substitution on tryptophan binding, it is also possible that the reduced af®nity for RNA is an indirect effect. We have previously identi®ed eight amino acid residues in TRAP as likely candidates for residues that interact with RNA (Antson et al., 1995). These residues (Asp8, Asn20, Ser35, Lys37, Asp39, Lys40, Lys56 and Arg58) are conserved in TRAP from several Bacillus species but have no apparent role in subunit assembly or tryptophan binding based on examination of the crystal structure of the TRAP-tryptophan complex. We have now identi®ed three of these residues, Lys37, Lys56 and Arg58, as being directly involved in RNA binding. Moreover, Lys40 appears to play a role, directly or indirectly, in RNA binding. The other three residues from this set, Asp8, Ser35 and Asp39 are all involved in tryptophan binding and may play a role in activating TRAP to bind RNA. Several sequence motifs have been found in RNA binding proteins, including the RNP domain/RRM (Dreyfuss et al., 1993; Kenan et al., 1991), the RGG box (Kiledjian & Dreyfuss, 1992; Ghisol® et al., 1992), the Arg-rich motif (Lazinski

706 et al., 1989; Tan et al., 1993), and the KH motif (Siomi et al., 1993; Gibson et al., 1993a,b). Even though TRAP does not show signi®cant sequence homology to any of these motifs, there are interesting similarities between the RNA binding domain of TRAP and other RNA binding proteins. The three residues in TRAP (Lys37, Lys56 and Arg58) that we have identi®ed as being speci®cally involved in RNA binding do not form an apparent primary sequence motif. Instead these residues are juxtaposed in the three-dimensional structure of TRAP (Figure 4) on the outer edge of the protein ring and are exposed on the surface of an antiparallel b-sheet. The antiparallel b-sheet appears to be an important structural motif in RNA recognition (Nagai, 1995). This is particularly true for interactions with single-stranded RNA where the bsheet furnishes a large surface for interaction with the nucleotide bases. For example, the RNP domain, which is present in many RNA binding proteins, folds into a four-stranded antiparallel bsheet (Nagai, 1995). Moreover, in the crystal structure of the RNP domain of U1A snRNP complexed with a fragment of U1 RNA, the RNA binds across the b-sheet making many speci®c contacts with amino acid residues on this structure (Oubridge et al., 1994). A ten-stranded antiparallel b-sheet in the MS2 bacteriophage coat protein is also involved in binding to its RNA target (Valegard et al., 1994).

Implications regarding the forces involved in the TRAP-RNA interaction Two pieces of evidence suggest that ionic interactions between the phosphate backbone of RNA and amino acid side-chains are involved in RNA binding to TRAP. First, the three residues of TRAP that are speci®cally involved in RNA binding are all positively charged. Second, we have found that incorporation of either guanosine 50 -phosphorothioate (GTPaS) or adenosine 50 -phosphorothioate (ATPaS), but not CTPaS or UTPaS, into trp leader RNA inhibits binding to TRAP (C. B. & P. G., unpublished results). These modi®ed nucleotides contain a sulfur atom in place of one of the non-bridging oxygen atoms of the a-phosphate group and alter the charge distribution around this group. These results suggest that TRAP contacts the phosphate groups 50 to the Gs and As in the G/UAG repeats. However, we have also found that the stability of the interaction between TRAP and trp leader RNA is insensitive to changes in ionic strength between 0.1 and 0.7 M monovalent salt, suggesting that ionic interactions are not the major stabilizing force in the interaction between TRAP and trp leader RNA (Baumann et al., 1996). Hydrogen bonding is also implicated in the TRAP-RNA interaction based upon the mutagenic analysis of positions 37 and 58. Glutamine substitutions at positions 37 and 58 of TRAP re-

Alanine-scanning Mutagenesis of TRAP

sulted in stronger RNA binding as compared to alanine substitutions at these positions, although their RNA binding af®nities are still lower than wild-type (Table 3). In contrast to alanine, glutamine has an aliphatic neutral side-chain capable of forming hydrogen bonds, thus hydrogen bonding may also be occurring at these two positions. It is interesting to note that alanine substitution of Phe32, the only surface amino acid residue with an aromatic side-chain on the front side of TRAP, does not alter RNA binding. This result argues against the involvement of ring stacking force contributing to the interaction of TRAP with trp leader RNA. Although positions 37 and 58 at the ends of the recognition KKR motif (Figure 4), can be occupied by either lysine or arginine, only lysine is functional at the central position 56. One possible explanation for this restriction is that the size of the side-chain may be more critical at position 56 than at positions 37 or 58. While both lysine and arginine are positively charged and are both able to donate hydrogen bonds, the arginine side-chain is one carbon longer than that of lysine, and the guanidino group of arginine is more bulky. Recent studies comparing the RNA binding activity of wild-type TRAP and the R58K mutant protein have shown that, whereas wild-type TRAP binds RNAs containing GAG repeats better than UAG-containing RNAs, R58K has a much weaker preference for GAG over UAG repeats (S. Xirasagar & P. G., unpublished results). These results suggest that Arg58 interacts with the ®rst residue (G or U) of the trinucleotide repeats in the RNA target of TRAP. Moreover, given that changes at position 58 in¯uence the speci®city of the TRAP-RNA interaction, it appears that Arg58 provides more than just a salt link to the phosphodiester backbone, more likely forming hydrogen bonds with the guanine or uracil. This ®nding also suggests an orientation of the bound RNA on TRAP with the 50 end of the G/UAGNN repeats interacting with Arg58 of the K37K56R58 binding motif of TRAP and then wrapping counterclockwise around the protein ring when viewed from the front (as depicted in Figure 4).

Materials and Methods Site-directed mutagenesis Site-directed mutagenesis was performed by PCR using mutagenic oligonucleotides (Merino et al., 1992). PCR reactions were carried out for 35 cycles (94 C for one minute, 50 C for two minutes and 60 C for one minute) in 50 mM KCl, 10 mM Tris-Cl (pH 8.3), 1.5 mM MgCl2, 0.01% (w/v) gelatin, 0.2 mM deoxynucleoside triphosphates with 0.3 units of Taq DNA polymerase. First, two overlapping half molecules of the mtrB gene were generated in two separate reactions using the wild-type mtrB-gene cloned in pBlueScriptKS(ÿ) (Stratagene) as template. One reaction contained the mutagenic

707

Alanine-scanning Mutagenesis of TRAP oligonucleotide and the universal reverse primer while the other reaction contained the universal ÿ20 primer and an oligonucleotide that annealed to positions 1228 to 1250 at the 30 end of mtrB (numbering as given by Gollnick et al., 1990). The two PCR products were mixed, and denatured by boiling for ten minutes. The overlapping regions between two PCR products were allowed to anneal by slowly cooling the mixture to room temperature, and the hybrid DNA fragment was then used as the template for another PCR reaction using the universal ÿ20 and reverse primers. The PCR products, which contained 50% wild-type and 50% mutant sequences, were cloned into a multicopy Escherichia coli/ Bacillus subtilis shuttle plasmid, pHB201 (S. Bron, unpublished results). The resulting recombinant plasmids were sequenced using Sequenase (Amersham) and those that contained the desired mutations were transformed into the B. subtilis reporter strain PGBS4233 (see below) using natural competence (Anagnostopoulos & Spizizen, 1961; Kuroda et al., 1988). In vivo regulation activity assay of TRAP in B. subtilis To allow us to quickly assess the affects of each alanine substitution on the regulatory activities of TRAP in vivo, we created the B. subtilis strain PGBS4233 (EmR, (
mtrB, argC4, amyE::[Ptrp-trpL-trpE0 -0 lacZ]). B. subtilis BG4233 (
mtrB, argC4) was transformed with chromosomal DNA of CYBS12 (EmR, argC4, amyE::[Ptrp-trpL-trpE0 0 lacZ]; Kuroda et al., 1988) and blue colonies were selected on minimal agar plates (Vogel & Bonner, 1956) supplemented with 0.2% (w/v) acid-hydrolyzed casein, 0.2% (w/v) glucose, 10 mg/ml L-arginine, 50 mg/ml 5-bromo-4-chloro-3-indolyl-bD-galactopyranoside (X-gal) and 1 mg/ml erythromycin. The amyEÿ phenotype was con®rmed on starch plates (Shimotsu & Henner, 1986) and the mtrBÿ phenotype was checked using 5-¯uorotryptophan resistance (Xu et al., 1989). The deletion in mtrB was found to be 176 base-pairs (1001 to 1176; numbering as given by Gollnick et al., 1990) by PCR ampli®cation of the mutant mtrB gene and sequencing the resulting PCR product. The trpE0 -0 lacZ fusion is constitutively overexpressed in this strain due to the lack of TRAP (
mtrB). Plasmid pHB201 was used to express individual mutant mtrB genes in this reporter strain under the control of the p59 promoter (Van der Vossen et al., 1987). Functional TRAP results in the regulation of trpE0 -0 lacZ expression in response to tryptophan. We then compared the regulatory activity of each mutant TRAP protein to wild-type TRAP using b-galactosidase assays (see below). b -galactosidase assays PGBS4233 cells containing mutant mtrB genes in pHB201 were grown in minimal salts (Spizizen, 1958) with 0.2% glucose, 0.2% acid-hydrolyzed casein, 5 mg/ml L-arginine and 5 mg/ml chloramphenicol in the absence or presence of 50 mg/ml L-tryptophan at 37 C until the absorbance of the culture at 600 nm reached between 0.4 and 0.8. b-Galactosidase activity was assayed as described previously (Kuroda et al., 1988). Values reported are the average from three different colonies, each analyzed in duplicate. Standard deviations were less than 20% of the mean.

In vivo screening for active or inactive TRAP after random-codon mutagenesis Oligonucleotides containing equal representations of all four nucleotides at the positions corresponding to a particular codon in mtrB were used in mutagenic PCR as described above. This created a library of mutant mtrB genes encoding all possible substitutions at a particular amino acid residue in TRAP. The PCR products of these reactions were ligated into pHB201 and transformed into E. coli K802. Transformants were selected on LB plates containing 12.5 mg/ml chloramphenicol. Several thousand individual transformant colonies were scraped from the plates and pooled, and plasmid DNA was extracted from the pool (Birnboim & Doly, 1979). The plasmids were then transformed into B. subtilis PGBS4233 by natural competence (Anagnostopoulos & Spizizen, 1961; Kuroda et al., 1988) and selected on plates containing 0.2% acid-hydrolyzed casein, 50 mg/ml X-gal and 50 mg/ ml L-tryptophan as described above. Colonies containing non-functional mutant TRAP proteins were blue due to unregulated overexpression of b-galactosidase from the trpE0 -0 lacZ fusion, whereas colonies containing functional TRAP were white. Over 1500 individual transformants were screened in each case to ensure that all 64 possible codons at each position were represented (P > 0.999; Frischauf, 1988).

Expression and purification of mutant TRAP proteins Mutant mtrB genes were expressed in E. coli using the T7 RNA polymerase expression system. Plasmids (pTZ18U, US Biochemical) containing the desired mutations were transformed into E. coli SG62052/pGP1-2 (Tabor & Richardson, 1985) and selected on LB plates containing 100 mg/ml ampicillin and 50 mg/ml kanamycin. Transformants were grown, TRAP expression was induced, and cell extracts were prepared as described previously (Antson et al., 1994). In some cases, TRAP was produced using the T7 expression plasmid pET9a (Novagen) in which the BamHI site in the polylinker was replaced with a NotI site (Barry Hurlburt, University of Arkansas, Little Rock). In these cases, expression in E. coli BL-21(DE3) (Studier et al., 1990) was induced by the addition of isopropyl-b,D-thiogalactopyranoside (IPTG) to 1 mM. In all cases, TRAP was puri®ed by immunoaf®nity chromatography as described by Otridge & Gollnick (1993) and quanti®ed by UV absorbance at 280 nm using an extinction coef®cient of 1280 Mÿ1cmÿ1.

Analysis of tryptophan binding to TRAP Tryptophan binding to puri®ed TRAP was analyzed by equilibrium dialysis using [14C]L-tryptophan (Du Pont NEN) as described previously (Antson et al., 1995). After incubating for 16 to 24 hours in an equilibrium microvolume dialyzer (Hoefer, EMD101) at 4 C, samples were analyzed by scintillation counting. Tryptophan binding to TRAP is positively cooperative, therefore the data were ®t to the Hill equation (Segel, 1975): ‰TrpŠb ˆ

a  …‰TrpŠf =S0:5 †n 1 ‡ …‰TrpŠf =S0:5 †n

where a is the saturation level of bound tryptophan ([Trp]b); S0.5 is de®ned as the concentration of free tryptophan, [Trp]f, at which the [Trp]b reaches 50% of

708 saturation, and is used to describe the binding af®nity. The Hill coef®cient, n, is used to describe the cooperativity of binding: n is 1.0 in the case of no cooperativity, >1.0 for positive cooperativity and <1.0 for negative cooperativity. Analysis of RNA binding to TRAP For RNA binding assays, 32P-labeled RNA containing residues ‡2 to ‡138 (RNA 2-138) or ‡36 to ‡92 (RNA 36-92) of the trp leader as de®ned by Shimotsu et al., (1986), was prepared by in vitro transcription using T7 RNA polymerase and [a-32P]UTP (Du Pont NEN) as described previously (Otridge & Gollnick, 1993; Baumann, et al., 1996). Labeled RNAs were gel puri®ed using the crush and soak method (Milligan & Uhlenbeck, 1989) from 6% (W/V) denaturing polyacrylamide gels. Two assays, mobility-shift gels and ®lter binding, were used to analyze the RNA binding activity of puri®ed TRAP in vitro. RNA mobility-shift assays (Otridge & Gollnick, 1993) have the advantage of allowing visualization of speci®c complexes formed between TRAP and trp leader RNA. This assay was used to assess the speci®city of the interaction between TRAP and trp leader RNA, particularly with regard to the number of complexes formed. The stoichiometry of the interaction of D29A TRAP (single-letter amino acid code indicating replacement of Asp29 by Ala) and trp leader RNA was also analyzed using mobility-shift gels. In this case, in vivo 35S-labeled D29A TRAP, prepared as described previously (Baumann et al., 1996), was used with 32P-labeled trp leader RNA 2-138. Gels containing N, N0 -bis-acrylylcystamine (BioRad) as the crosslinker were prepared and run as described previously (Baumann et al., 1996; Otridge & Gollnick, 1993). Gel fragments containing the TRAPRNA complexes were excised and dissolved by incubating overnight in 300 ml of 1 M dithiothreitol. The molar ratios of TRAP and RNA were analyzed as described previously (Baumann, et al., 1996). Nitrocellulose ®lter binding, as described by Baumann et al. (1996) was used as a more quantitative assay to measure the af®nity of mutant TRAP proteins for trp leader RNA 36-92. Filters were counted and analyzed as described previously (Baumann et al., 1996).

Acknowledgements This work was supported by National Science Foundation grants MCB-9118654 and MCB-9603594. P. Gollnick is a Pew Scholar in the Biomedical Sciences. We thank Dr Sierd Bron for providing plasmid pHB201 and Michael Kesten for helping to make mutant mtrB. We thank Charley Yanofsky, Gerald Koudelka, Margaret Hollingsworth, Paul Babitzke and Sandhya Xirasagar for the critical reading of the manuscript, Jim Stamos and Alan Siegel for great help making Figures, and Alfred Antson for help in making Figure 1.

References Anagnostopoulos, C. & Spizizen, J. (1961). Requirements for transformation in Bacillus subtilis. J. Bacteriol. 81, 741 ± 746.

Alanine-scanning Mutagenesis of TRAP Antson, A. A., Brozozwski, A., Dodson, E., Dauter, E., Wilson, K., Kurecki, T., Otridge, J. & Gollnick, P. (1994). 11-Fold symmetry of the trp RNA-binding attenuation protein (TRAP) from Bacillus subtilis determined by X-ray analysis. J. Mol. Biol. 244, 1 ± 5. Antson, A. A., Otridge, J., Brzozowski, A. M., Dodson, E. J., Dodson, G. G., Wilson, K. S., Smith, T. M., Yang, M., Kurecki, T. & Gollnick, P. (1995). The structure of trp RNA-binding attenuation protein. Nature, 374, 693± 700. Babitzke, P., Gollnick, P. & Yanofsky, C. (1992). The mtrAB operon of Bacillus subtilis encodes GTP cyclohydrolase I (MtrA), an enzyme involved in folic acid biosynthesis, and MtrB, a regulator of tryptophan biosynthesis. J. Bacteriol. 174, 2059± 2064. Babitzke, P., Stults, J. T., Shire, S. J. & Yanofsky, C. (1994). TRAP, the trp RNA-binding attenuation protein of Bacillus subtilis, is a multisubunit complex that appears to recognize G/UAG repeats in the trpEDCFBA and trpG transcripts. J. Biol. Chem. 269, 16597± 16604. Babitzke, P., Bear, D. G. & Yanofsky, C. (1995). TRAP, the trp RNA-binding attenuation protein of Bacillus subtilis, is a toroid-shaped molecule that binds transcripts containing GAG or UAG repeats separated by two nucleotides. Proc. Natl Acad. Sci. USA. 92, 7916± 7920. Babitzke, P., Yealy, J. & Campanelli, D. (1996). Interaction of the trp RNA-binding attenuation protein (TRAP) of Bacillus subtilis with RNA: effects of the number of GAG repeats, the nucleotides separating adjacent repeats, and RNA secondary structure. J. Bacteriol. 178, 5159± 5163. Baumann, C., Otridge, J. & Gollnick, P. (1996). Kinetic and thermodynamic analysis of the interaction between TRAP (trp RNA-binding attenuation protein) of Bacillus subtilis and trp leader RNA. J. Biol. Chem. 271, 12269± 12274. Baumann, C., Xirasagar, S. & Gollnick, P. (1997). The trp RNA-binding Attenuation Protein (TRAP) from Bacillus subtilis binds to unstacked trp leader RNA. J. Biol. Chem. In the press. Birnboim, H. C. & Doly, J. (1979). A rapid alkaline extraction procedure for screening recombinant plasmid DNA. Nucl. Acids Res. 7, 1513± 1523. Cunningham, B. C. & Wells, J. A. (1989). High-resolution epitope mapping of hGH-receptor interactions by alanine-scanning mutagenesis. Science, 244, 1081± 1085. Dreyfuss, G., Matunis, M. J., Pinol-Roma, S. & Burd, C. G. (1993). hnRNP proteins and the biogenesis of mRNA. Annu. Rev. Biochem. 62, 289± 321. Du, H., Tarpey, R. & Babitzke, P. (1997). TRAP regulates TrpG synthesis by binding to the trpG ribosome binding site of Bacillus subtilis. J. Bacteriol. 179, 2582± 2586. Frischauf, A.-M. (1988). Digestion of DNA: size fractionation. Screening of Genomic Clone. In Methods in Enzymology: Guide to Molecular Cloning Techniques (Berger, S. L. & Kimmel, A. R., eds), pp. 183± 189, Academic Press, San Diego. Ghisol®, L., Joseph, G., Amaltic, F. & Erard, M. (1992). The glycine-rich domain of nucleolin has an unusual supersecondary structure responsible for its RNA-helix-destabilizing properties. J. Biol. Chem. 267, 2955± 2959. Gibson, T. J., Thompson, J. D. & Heringa, J. (1993a). The KH domain occurs in a diverse set of RNA-binding

Alanine-scanning Mutagenesis of TRAP proteins that include the antiterminator NusA and is probably involved in binding to nucleic acid. FEBS Letters, 324, 361 ± 366. Gibson, T. J., Rice, P. M., Thompson, J. D. & Heringa, J. (1993b). KH domains within the FMR1 sequence suggest that fragile X syndrome stems from a defect in RNA metabolism. Trends Biochem. Sci. 18, 331± 333. Gollnick, P. (1994). Regulation of the Bacillus subtilis trp operon by an RNA-binding protein. Mol. Microbiol. 11, 991± 997. Gollnick, P., Ishino, S., Kuroda, M. I., Henner, D. J. & Yanofsky, C. (1990). The mtr locus is a two-gene operon required for transcription attenuation in the trp operon of Bacillus subtilis. Proc. Natl Acad. Sci. USA. 87, 8726± 8730. Hoffman, R. J. & Gollnick, P. (1995). The mtrB gene of Bacillus pumilus encodes a protein with sequence and functional homology to the trp RNA-binding attenuation protein (TRAP) of Bacillus subtilis. J. Bacteriol. 177, 839 ± 842. Insight II, User Guide (October 1995). Biosym/MSI, San Diego. Kenan, D. J., Query, C. C. & Keene, J. D. (1991). RNA recognition: towards identifying determinants of speci®city. Trends Biochem. Sci. 16, 214 ± 220. Kiledjian, M. & Dreyfuss, G. (1992). Primary structure and binding activity of the hnRNP U protein: binding RNA through RGG box. EMBO J. 11, 2655± 2664. Kong, X. P., Onrust, R., O'Donnell, M. & Kuriyan, J. (1992). Three-dimensional structure of the b subunit of E. coli DNA polymerase III holoenzyme: a sliding DNA clamp. Cell, 69, 425± 437. Kraulis, P. J. (1991). MOLSCRIPT Ða program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallog. 24, 946± 950. Krishna, T. S., Kong, X. P., Gary, S., Burgers, P. M. & Kuriyan, J. (1994). Crystal structure of the eukaryotic DNA polymerase processivity factor PCNA. Cell, 79, 1233± 1243. Kuroda, M. I., Henner, D. & Yanofsky, C. (1988). Cisacting sites in the transcript of the Bacillus subtilis trp operon regulate expression of the operon. J. Bacteriol. 170, 3080± 3088. Lazinski, D., Grzadzielska, E. & Das, A. (1989). Sequence-speci®c recognition of RNA hairpins by bacteriophage antiterminators requires a conserved arginine-rich motif. Cell, 59, 207± 218. Lima, C. D., Wang, J. C. & Mondragon, A. (1994). Three-dimensional structure of the 67 kD N-terminal fragment of E. coli DNA topoisomerase I. Nature, 367, 138 ± 146. Merino, E., Osuna, J., Bolivar, F. & Soberon, X. (1992). A general, PCR-based method for single or combinatorial oligonucleotide-directed mutagenesis on pUC/M13 vectors. BioTechniques, 12, 508± 510. Merino, E., Babitzke, P. & Yanofsky, C. (1995). trp RNAbinding attenuation protein (TRAP)-trp leader RNA interactions mediate translational as well as transcriptional regulation of the Bacillus subtilis trp operon. J. Bacteriol. 177, 6362 ±6370. Miller, J. (1972). In Experiments in Molecular Genetics, pp. 352± 355, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Milligan, J. F. & Uhlenbeck, O. C. (1989). Determination of RNA-protein contacts using thiophosphate substitutions. Biochemistry, 28, 2849± 2855.

709 Nagai, K. (1995). RNA-protein complexes. Curr. Opin. Struct. Biol. 6, 53 ± 61. Naktinis, V., Turner, J. & O'Donnell, M. (1996). A molecular switch in a replication machine de®ned by an internal competition for protein rings. Cell, 84, 137± 145. Otridge, J. & Gollnick, P. (1993). MtrB from Bacillus subtilis binds speci®cally to trp leader RNA in a tryptophan-dependent manner. Proc. Natl Acad. Sci. USA, 90, 128± 132. Oubridge, C., Ito, N., Evans, P. R., Teo, C. H. & Nagai, Ê resolution of K. (1994). Crystal structure at 1.92 A the RNA-binding domain of the U1A spliceosomal protein complexed with an RNA hairpin. Nature, 372, 432 ± 438. Rould, M. A., Perona, J. J., Soll, D. & Steitz, T. A. (1989). Structure of E. coli glutaminyl-tRNA synthetase Ê complexed with tRNAGln and ATP at 2.8 A resolution. Science, 246, 1135± 1142. Segel, I. H. (1975). Multisite & Allosteric Enzymes. In Enzyme Kinetics: Behavior and Analysis of Rapid Equilibrium and Steady State Enzyme Systems, pp. 360 ± 361, John Wiley & Sons Inc., New York. Shimotsu, H. & Henner, D. J. (1986). Construction of a single-copy integration vector and its use in analysis of regulation of the trp operon of Bacillus subtilis. Gene, 43, 85± 94. Shimotsu, H., Kuroda, M. I., Yanofsky, C. & Henner, D. (1986). Novel form of transcription attenuation regulates expression of the Bacillus subtilis tryptophan operon. J. Bacteriol. 166, 461± 471. Siomi, H., Matunis, M. J., Michael, W. M. & Dreyfuss, G. (1993). The pre-mRNA binding K protein contains a novel evolutionarily conserved motif. Nucl. Acids Res. 21, 1193± 1198. Slock, J., Stahly, D. P., Han, C. Y., Six, E. W. & Crawford, I. P. (1990). An apparent Bacillus subtilis folic acid biosynthetic operon containing pab, an amphibolic trpG gene, a third gene required for synthesis of para-aminobenzoic acid, and the dihydropteroate synthase gene. J. Bacteriol. 172, 7211± 7226. Spizizen, J. (1958). Transformation of biochemically de®cient strains of Bacillus subtilis by deoxyribonucleate. Proc. Natl Acad. Sci. USA, 44, 1072± 1078. Studier, F. W., Rosenberg, A. H., Dunn, J. J. & Dubendorff, J. W. (1990). Use of T7 polymerase to direct expression of cloned genes. Methods Enzymol. 185, 60 ± 89. Tabor, S. & Richardson, C. C. (1985). A bacteriophage T7 RNA polymerase/promoter system for controlled exclusive expression of speci®c genes. Proc. Natl Acad. Sci. USA, 82, 1074± 1078. Tan, R., Chen, L., Buettner, J. A., Hudson, D. & Frankel, A. D. (1993). RNA recognition by an isolated a helix. Cell, 73, 1031± 1040. Valegard, K., Murray, J. B., Stockley, P. G., Stonehouse, N. J. & Liljas, L. (1994). Crystal structure of an RNA bacteriophage coat protein-operator complex. Nature, 371, 623 ± 626. Van der Vossen, J. M., van der Lelie, D. & Venema, G. (1987). Isolation and characterization of Streptococcus cremoris Wg2-speci®c promoters. Appl. Environ. Microbiol. 53, 2452± 2457. Vogel, H. J. & Bonner, D. M. (1956). Acetylornithinase of Escherichia coli: partial puri®cation and some properties. J. Biol. Chem. 218, 97± 106. Xu, Z. J., Love, M. L., Ma, L. Y., Blum, M., Bronskill, P. M., Bernstein, J., Grey, A. A., Hofmann, T.,

710

Alanine-scanning Mutagenesis of TRAP Camerman, N. & Wong, J. T. (1989). TryptophantRNA synthetase from Bacillus subtilis. Characterization and role of hydrophobicity in substrate recognition. J. Biol. Chem. 264, 4304± 4311.

Yang, M., Saizieu, A., Loon, A. P. G. M. V. & Gollnick, P. (1995). Translation of trpG in Bacillus subtilis is regulated by the trp RNA-binding attenuation protein (TRAP). J. Bacteriol. 17, 4272± 4278.

Edited by J. A. Wells (Received 14 March 1997; received in revised form 12 May 1997; accepted 12 May 1997)