Mol Gen Genet (1996) 251:338-346

© Springer-Verlag 1996

(~agatay Giines • B. Miiller-Hill

Mutants in position 69 of the Trp repressor of Escherichia coil i(12 with altered DNA-bindingspecificity

Received: 11 August 1995 / Accepted: 14 November 1995

Abstract Structural analysis by X-ray crystallography has indicated that direct contact occurs between Arg69, the second residue of the first helix of the helix-turnhelix (HTH) motif of the Trp repressor, and guanine in position 9 of the c~-centred consensus trp operator. We therefore replaced residue 69 of the Trp repressor with Gly, Ile, Leu or Gln and tested the resultant repressor mutants for their binding to synthetic symmetrical ~- or /?-centred trp operator variants, in vivo and in vitro. We present genetic and biochemical evidence that Ile in position 69 of the Trp repressor interacts specifically with thymine in position 9 of the a-centred trp operator. There are also interactions with other bases in positions 8 and 9 of the a-centred trp operator. In vitro, the Trp repressor of mutant RI69 binds to the consensus c~-centred trp operator and a similar trp operator variant that carries a T in position 9. in vivo analysis of the interactions of Trp repressor mutant RI69 with symmetrical variants of the /%centred trp operator shows a change in the specificity of binding to a /?centred symmetrical trp operator variant with a gua-nine to thymine substitution in position 5, which corresponds to position 9 of the e-centred trp operator. Key words Trp repressor . Consensus trp operator .

Helix-turn-helix motif - Protein-DNA interaction Introduction

The X-ray structure of the co-crystal of the Trp repressor with a 19 bp consensus trp operator fragment that was derived from natural trp operators has been solved Communicated by J. W. Lengeler ~. Giines 1 • B. Miiller-Hitl ( 9 ) Institut fiir Genetik, Universit~it zu K61n, Weyertat 121, D-50931 K61n, Germany Present address:

Institut ftir Molecularbiologie II, UniversitM Zi.irich, Winterthurer Str. 190, CH-8057 Ziirich, Switzerland

(Otwinowski et al. 1988). All but one of the proteinDNA contacts uses water molecules for hydrogen bonding between bases and amino acid side chains of the HTH (helix-turn-helix) motif. The only direct contact occurs between the side chain of Arg69, the second residue of the first helix of the HTH motif, and guanine in position 9 of the operator. The authors proposed a new binding mode, indirect readout, where irregular positions of the phosphates in the DNA backbone play an important role in specific recognition. These results prompted the analysis of the interaction of the Trp repressor with the trp operator by genetic and biochemical methods: the solution of the NMR (nuclear magnetic resonance) structure of the Trp repressor-trp operator complex (Zhang et al. 1994) revealed contact NOEs (nuclear Overhauser effects) between protons of the side chain of Ile79 and protons of thymine 8. The genetic analysis of Trp repressor-trp operator interactions showed (Pfau et al. 1994) that Trp repressor mutants with substitutions in position 1 of the recognition helix have new binding specificities for trp operator variants with substitutions in position 7. The results of the NMR study (Zhang et al. 1994) and genetic analysis (Pfau et al. 1994) suggested hydrophobic interactions between Ala80, the second residue of the recognition helix of the Trp repressor, and thymine in position 4 of the trp operator. With respect to position 1 of the recognition helix of the Trp repressor, our in vivo and in vitro results (Giines et al. 1995) corroborate these NMR and genetic data. We have shown that Trp repressor mutants with various side chains at residue t of the recognition helix specifically recognize base pairs 7 and 8 of a-centred trp operator variants. We suggested direct hydrophobic interactions between Ile79 and thymines in these positions, especially with thymine in position 8. Based on our results with Trp repressor mutant AG80 we suggested that the side chain of Ala80, the second residue of the recognition helix of the Trp repressor, is involved in hydrophobic interactions with C6 and T5 of the c~-centred trp

339

operator. Residues with longer side chains in position of the Trp repressor and bases of the trp operator. 2 of the recognition helix broaden the specificity of the The phenotypes of the Trp represssor mutants studied Trp repressor with trp operator variants with substitu- support and extend our proposal. tions in positions 3 and 4 (Giines et al. 1995). In a recent review of the trp system (Youderian and Arvidson 1994), arguments in favour of direct interactions of side Materials and methods chains of the Trp repressor with bases of the trp operator are summarized. Media, chemicals and general methods The Trp repressor regulates the expression of at least five operons by binding to their upstream operator Media, chemicals and general methods have been described presites: trpEDCBA (Rose et al. 1973; Bennett and viously (Staacke et al. 1990). Oligonucleotides were synthesized on Applied Biosystems 380A DNA Synthesizer and purified on Yanofsky 1978), trpR (Gunsalus and Yanofsky 1980) an denaturing potyacrylamide gels prior to cloning. All trpR gene c~roH (Zurawski et al. 1981; Grove and Gunsalus 1987), variants that encode Yrp repressor mutants and all trp operator mtr (Heatwole and Somerville 1991) and aroL (Heat- variants that were derived from the consensus c~- or /?-centred wote and Somerville 1992). The consensus sequence consensus trp operators were verified by DNA sequencing (Sanger for these operators (Bass et al. t987; Kumamoto et al. et at. 1977). 1987) is large in comparison with the consensus sequences of most known prokaryotic operators. Two Strains and plasmids additional axes occur left and right of the central axis a (Bennett et al. 1976). Haran et al. (1992) have demon- Strain DC41-3 of Escherichia coli [trpR- A(lac pro) thi- 9alE Smr strated that an a-centred 19 bp target binds one Trp recA] was used to determine specific fi-galactosidase activities in the repressor dimer whereas a fi-centred target of the same presence of wild-type Trp repressor or its mutants or in the presence length binds two Trp repressor dimers. Lawson and of an inactive Trp repressor mutant that lacks most of the HTH Carey (1993) solved the co-crystal structure of a 16 bp, motif (Staacke et al. 1990). Plasmid pWBI0t, which carries the Trp repressor mutants, and fi-centred target with the Trp repressor. Their structure plasmid pWB7000, which carries the lacZ reporter system with the supports the binding of two Trp repressor dimers to fi-centred trp operator variants, are described in Staacke et al. (1990). fi-centred targets. In this co-crystal, the binding of pWBt01 contains a unique SacI site in codons 74 and 75 of the trpR two Trp repressor dimers is enhanced by intermolecu- gene. It was introduced by Walter (1989) by site-directed mutagenlar protein-protein interactions (Lawson and Carey esis (Hutchison et al. 1978). We introduced a second SacI site 42.bp upstream of the first site (between codons 60 and 61 of the Trp 1993). repressor) by site-directed mutagenesis by the same method. The In the light of our results and the findings of several plasmid was then digested with SacI. To clone the Trp repressor groups who showed that more than one Trp repressor mutants, we synthesized complementary oligonucleotides with the dimer can bind to the trp operator in vitro depending respective base pair substitutions encoding Gly, Ile, Leu or Gln in 69 of the Trp repressor and overhanging SacI ends. They on the length or sequence of the operator fragments or position were purified on denaturing polyacrylamide gels. The complementthe concentration of Trp repressor (Carey et al. 1991; ary oligonucleotides were eluted together overnight at 37° C. The Haran et al. 1992; Beckmann et al. 1993; Liu and resulting double-stranded molecules were ligated into the SacI sites Matthews 1993, 1994; Staacke 1993), we proposed of pWB101. The ligation products were transformed into E. coli that the natural c~-centred trp operators can be seen to strain DH5c~ IF', endA1, hsdR17, (r~, m+), supE44, thi, )~- fecal, reIA1, ~80 dIacZAM15]. Single colonies were picked and consist of two imperfect half-sites with the axes fi 9yrA96, screened by DNA sequence analysis of the 5' end of the trpR genes (Giines et al. 1995). Depending on the sequence of the on the plasmid DNAs. pWB10I is a pBR322 derivative (about 30 natural trp operator and on the concentration of Trp copies per cell) and pWB7000 is a pACYC184 derivative (about 10 repressor, one or two Trp repressor dimers might bind copies per cell) (Sambrook et al. 1989). OIigonucleotides that contained the respective trp operator varito the operator to repress the expression of the respectants were cloned in place of the tacZ operator upstream of the lacZ ive genes. Similar proposals had been made on the reporter gene into an AsnI site close to the - 10 region of the Iac basis of DNase I footprinting and methylation promoter (Gfines et al. 1995). pWB7300 contains an optimized, fully protection studies (Kumamoto et al. 1987). Such symmetrical synthetic c~-centredTrp repressor binding site, derived cooperative binding might allow fine regulation of gene from the consensus sequence of all fullqength trp operators. pWB7200 carries the synthetic/?-centred Trp repressor binding site expression. that was derived from the right half-site of the consensus sequence Arg69, the second residue of the first helix of the and optimized by base pair substitutions (Staacke et al. 1990). All HTH motif of the Trp repressor, was shown to contact trp operator variants used in this study were derived from these guanines in position 9 of the a-centred trp operator optimized trp operators by single or double symmetrical base pair by direct hydrogen bonding (Otwinowski et al. 1988; substitutions. All plasmids were constructed according to standard Marmorstein et al. 1991; Smith et al. 1994; Zhang et al. methods (Sambrook et al. 1989). 1994). Here, we present results obtained with Trp repressor mutants with substitutions in position 69 and e- The test system and fi-centred trp operator variants. We used gel retardation and an in vivo repression assay to identify A two-plasmid system (Lehming et al. 1987; Gtines et al. 1995) was specific interactions between amino acid side chains used to determine the level of repression of fi-galactosidase synthesis

340 by Trp repressor. The system consists of two plasmids with different origins of replication and an ampicillin or a tetracycline resistance gene, respectively. One of the plasmids contains the lacZ gene under the control of a synthetic promoter with a unique AsnI site into which synthetic trp operators can be cloned (pWB7000 and its derivatives). The other plasmid contains the erpR gene into which specific mutations can be introduced (pWB101 and its derivatives). The level of expression of fl-galactosidase depends on the free or occupied state of the respective trp operator variant by wild-type or mutant Trp repressors, The degree of repression of fi-galactosidase synthesis under the control of a given trp operator variant reflects the affinity of the Trp repressor protein (wild-type or mutant) for this particular trp operator variant (for details see Giines et al. 1995). All values for specific fi-galactosidase activities are based on at least four determinations. They are the averages of two measurements of each of at least two independent transformants. Specific fl-galactosidase activities were determined as described by Miller (1972).

Preparation of cell extracts We used crude extracts of E. coli strain DC41-3 for gel retardation assays (Staacke et al. 1990). In this strain, the trpR gene has been inactivated by an internal deletion of at least 200 bp (Staacke 1993) and thus encodes an inactive Trp repressor, We transformed the cells with pWB10t or one of its derivatives that carries the gene for the wild-type or a mutant Trp repressor. The cells were grown to saturation at 37°C in 400 ml rich yeast medium (dYT) in the presence of ampicillin. Cell pellets were resuspended in storage buffer (10 mM TRIS-HCI, pH 7.0; 100 mM NaC1; 1 mM EDTA; 0.5 mM L-tryptophan in 30% glycerol) and sonicated. Cell debris was separated by centrifugation and aliquots of the clear supernatant were stored at - 7 0 ° C (Giines et al. 1995). The protein concentrations were determined by the Biuret assay (Miller 1972) and adjusted to 25 mg protein/ml.

Targets for the gel retardation assay The consensus 23 bp e-centred ~rp operator and its variants were used for gel retardation assays (Fig. 3). The consensus e-centred target has the same sequence as the one used by Haran et al. (1992) who showed that this target is occupied by one Trp repressor dimer. The oligonucleotide contains five consecutive cytosines to allow duplex formation. The long (31 bp) e-centred consensus target was derived from this target by adding base pairs to the left and to the right side. The short/~-centred consensus target (Fig. 3) is identical to the one used by Haran et al. (1992) who showed that two Trp repressor dimers bind to this target. We extended the /Lcentred targets by adding base pairs to the left and to the right of the ]~-axis (31 bp/~-centred) target.

Gel retardation assay Oligonucleotides carrying synthetic e- or /?-centred trp operator sequences were synthesized on an Applied Biosystems 380A DNA Synthesizer. They were purified on denaturing polyacrylamide gels with 7 M urea (Sambrook et al. 1989) and eluted overnight at room temperature. The hairpin structures were filled in at the protruding 5' ends with [e3ZP]dATP and Klenow fragment of DNA polymerase I. The fragments were purified again on native polyacrylamide gels and were resuspended in binding buffer (10 mM TRIS-HC1, pH 7.0; 100 mM NaCt; 1 mM EDTA; 0.5 mM L-tryptophan). Binding reactions were allowed to proceed at room temperature. Each sample contained 4.5 lag sonicated salmon sperm DNA. About 4 ng of trp operator DNA fragments (30-100 counts/s) were incubated with maximally 25 lag protein of crude extract in a total volume of

10 gl binding buffer. Trp repressor target-Trp repressor complexes were separated on pre-run native 10% polyacrylamide gels (19 : 1 polyacrylamide: bisacrylamide) in 10 mM TRIS-acetate, pH 7.0 and 0.1 mM L-tryptophan. Gels were run at 10 V/cm. Buffer was recirculated by an aquarium pump (Giines et al. 1995).

Results

Repression of ~- and ]3-centred trp operator variants by Trp repressor mutants with substitutions in position 69 The Trp repressor carries an arginine in the second position of the first helix of the HTH motif. We constructed four Trp repressor mutants - RG69, RI69, RL69 and RQ69 (Materials and methods) - and tested them with ct- and/3-centred trp operator variants with symmetrical substitutions in positions 1-9 with respect to e (Fig. 1) and in positions 1-5 with respect to /3 (Fig. 2) in repression tests (see Materials and methods; Ebright 1991; Giines et al. 1995). Wild-type Trp repressor represses the expression of /~-galactosidase under the control of synthetic e-centred symmetrical trp operator variants with substitutions in position 9 up to 350-fold less well than the e-centred consensus trp operator (Fig. 1). Mutant RG69 was used to test for loss of contact (Gtines et al. 1995; Ebright 1991). We indeed find that repression of trp operator variants with substitutions in position 9 (pWB7391, pWB7392, and pWB7394) by the mutant RG69 is two- to tenfold stronger than with wild-type Trp repressor and these same trp operator variants. This is not the case with any other trp operator variant (Fig. 1). Trp repressor mutants RI69 and RL69 repress /?-galactosidase synthesis under the control of trp operator variants with T/A base pairs in position 9 (pWB7394) about 100-fold more strongly than does wild-type Trp repressor. However, if thymine is present on the complementary strand (pWB7391), repression with these Trp repressor mutants is only about fivefold stronger than with wild-type Trp repressor and the respective trp operator variant (Fig. 1). Repression with RI69 and RL69 and a trp operator variant with C/G base pairs in position 9 (pWB7392) is more than tenfold stronger than repression with wild-type Trp repressor and this erp operator variant. Repression with mutant RQ69 is similar to that obtained with mutants RI69 and RL69 except that repression with RQ69 and the trp operator variant with A/T in position 9 (7391) is stronger than with RI69 and RL69 and this trp operator variant (Fig. 1). We also tested the binding of these Trp repressor mutants with/~-centred trp operator variants under the same conditions as with e-centred trp operator variants (Fig. 2). When we changed the GC base pair in position 5 of the /3-centred trp operator to AT, CG or TA, repression with wild-type Trp repressor was lowered by

341 repression values with a-centered trp operator variants and Trp repressor mutants carrying substitutions in position 69 pla~mid pwu 7300 731t 7312 7313 ~21 73~ 7323 7331 7332 73~ 7342 7343 7344 7351 7352 7353 7~1 7363 7~4 73~ 7373 73~ 7381 7382 7383 7391 7~2 7~4

~ centered ~ p operator sequence

R

1o987654321 T GTACTAGT TAACTAGTAC A ACATGATCAATTGATCATGT 12345678910 T GTACTAGTATACTAGTACA CG GC T GTACTAGATATCTAGTACA C G G C TGTACTAATTAATTAGTACA C G T A TGTACTCATTAATGAGTACA G C T A TGTACAAGTTAACTTGTACA C G G C T GTAATAGTTAACTATTAC A G C T A TGTCCTAGTTAACTAGGACA G C T A T GAACTAGTTAACTAGT TC A C G G C T ATACTAGTTAACTAGTAT A C G T A

a factor of a b o u t 100-500 (Fig. 2). The m u t a n t RG69 represses these trp o p e r a t o r variants (pWB7251, 7252 and 7254) as well as wild-type T r p repressor, indicating loss of contact (Ebright 1991; Giines et al. 1995). Mutants RI69 and RL69 repress the consensus /?-target (pWB7200) to a similar degree: about 40 times less than wild-type T r p repressor. RQ69 represses the consensus /?-target a b o u t 20-fold less well than wild-type T r p repressor (Fig. 2). Yet all three mutants repress trp o p e r a t o r variants with A T or T A in position 5 better than does wild-type T r p repressor. The m u t a n t RI69 shows a m a r k e d specificity change by our definition (Giines et al. 1995): first, repression of the/?-centred trp o p e r a t o r variant with T A in position 5 by m u t a n t RI69 is a b o u t 50-fold stronger than repression of the same trp o p e r a t o r variant by wild-type T r p repressor; second, repression of this ~rp o p e r a t o r variant (pWB7254) by RI69 is a b o u t 4-fold stronger than repression of the

Fig. 2 Repression values for//centred trp operator variants and Trp repressor mutants carrying substitutions at residue 2 of the first helix (position 69) of the HTH motif. The Trp repressor mutants were tested with all single symmetrical/?-centred trp operator variants substituted in base pairs 1-5. All/~-centred trp operators are derivatives of pWB7200, which carries the optimized/~-centred target (Staacke et al. 1990). The numbering of the plasmids is as in Fig. 1

(wt)

G

I

L

Q

7000

130

1500

1300

1100

3900

220 100 215 1 1 1 1 1 1 1 1 1 1 1

950 350 200 4 4 1 3 2 2 1 1 1 t 2

680 460 230 2 4 1 1 1 2 2 I 1 1 1

1170 730 200 4 8 1 1 2 4 6 1 1 1 5

1 I

1 I

I I

I I

I I

I I

I I

I I

2700

2900 270 650 980 64 500 260 450 65 12 t 500 1 19 3 2 110

180 13 560 3000 250 270 50 22

1

1

1 1 160 780 90 550 350 290

1

4 1 700 1100 600 1700 900 2500

1

1 1 660 1100 430 1470 630 2350

3 1 670 1150 700 1980 540 1110

Fig. 1 Repression values for e-centred trp operator variants and Trp repressor mutants carrying substitutions at residue 2 (position 69) of the first helix of the helix-turn-helix (HTH) fi motif. The Trp repressor mutants were tested with all possible single symmetrical c~centred trp operator variants substituted in basepairs 1-9. All c~centred trp operator variants are derivatives of pWB7300, which carries the optimized e-centred target (Giines et al. 1995). In the trp operator variants, the last number of the plasmid designation stands for: 1, adenine; 2, cytosine; 3, guanine and 4, thymine. The second last number of a variant stands for the position of the altered base. Thus in the trp operator variant pWB7331, base 3 to the left of the centre of symmetry is an adenine. The IacZ gene is under the control of the trp operator variant indicated. Repression was determined with wild-type and mutant Trp repressors. The standard one-letter code is used for amino acids. The accuracy of each measurement of /?-galactosidase activity is ___20%

consensus /?-target (pWB7200) by RI69, whereas repression of the consensus /?-target by wild-type T r p repressor is about 500-fold stronger than repression

repression values with ~ - c e n t e r e d trp operator variants and T r p repressor mutants carrying substitutions in position 6 9 pl~mid pWB

~

~ centered operator s e q u e n c e

R (wt)

G

I

L

Q

7200

7654321 ATGTACTAGTACAT TACATGATCATGTA 1234567

t100

5

25

23

51

~11 ~12 7213 7~1 7223 7224 7232 7233 7234 7241 ~42 ~ 7~I 7~2 7254

ATGTACATGTACAT CG GC ATGTAATATTACAT G C T A ATGTCCTAGGACAT G C T A ATGAACTAGTTCAT C G G C ATATACTAGTATAT C G T A

1 1 1 2 2 2 2 18 1 120 53 39 9 3 2

1 1 1 1 1 1 1 1 1 53 17 7 13 5 4

1 t 1 1 1 1 1 1 1 6 9 4 52 7 110

1 1 1 t 1 1 1 1 1 8 38 6 46 12 38

1 1 1 1 1 I 1 1 1 10 23 7 90 16 44

342 of the /?-centred trp operator variant with thymine in position 5 by wild-type trp repressor (Fig. 2). Mutants RL69 and RQ69 broaden the specificity of the Trp repressor for all the trp operator variants with substitutions in position 5 (pWB7251; pWB7252; pWB7254). Some Trp repressor mutants studied here with substitutions in position 69 repress some of the trp operator variants with substitutions in position 8 of the c~-centred and the corresponding position 4 of the /?-centred trp operator better than the consensus c~and/?-centred trp operators. Thus, RG69 represses the c~-centred trp operator variant with C/G in position 8 about sixfold more strongly than it does the consensus sequence (Fig. 1). Yet, wild-type Trp repressor represses this trp operator variant about fourfold better than does mutant RG69. Similarly, mutant RG69 represses trp operator variants with C/G or A/T base pairs at the corresponding position in the /~-centred target (position 4) three- to tenfold better than the consensus /?-target (Fig. 2). But, these /?-centred trp operator variants are more efficiently repressed by wild-type Trp repressor than by RG69. These results indicate some loss of contact at position 8 of the a-centred target and the corresponding position 4 of the/?-centred target. Repression of c~-centred trp operator variants with A/T and G/C substitutions in position 8 by Trp repressor mutants RI69, RL69 and RQ69 is slightly stronger than repression of these trp operator variants by wild-type Trp repressor (Fig. t). These Trp repressor mutants repress the consensus c~-centred target more effectively than the respective variants with base pair substitutions in position 8. Gel retardation assays with Trp repressor mutant RI69 and trp operator variants We used the short (23 bp) and long (31 bp) e- and /?-centred trp operator hairpin targets (Haran et al. 1992) and mutant RI69 for gel retardatio n experiments (Materials and methods). Wild-type Trp repressor does not bind to a-centred trp operator variants with T/A in position 9. In contrast, Trp repressor mutant RI69 binds the consensus c~-centred trp operator and its variant with a T/A substitution in position 9 (Fig. 3.1 and Fig. 3.2). Mutant RI69 forms less stable complexes with short a-centred targets, which results in a smear (Fig. 3.1). The short a-centred consensus target binds one wild-type Trp repressor dimer whereas the short /?-centred consensus target binds two wild-type Trp repressor dimers (Haran et al. 1992). Long ~- and /?centred consensus targets form 2:1 complexes with wild-type Trp repressor (Fig. 3.1, lane 3 and Fig. 3.2, lanes 9 and 17). We used the relative position of these complexes in the gel as reference markers for 1 : 1 or 2 : 1 occupation (Fig. 3.1, lanes 9 and 6 and Fig. 3.2, lanes 3 and 6). We obtained a greater band shift with mutant

RI69 and the short and long a-centred targets. This suggests that possibly three Trp repressor dimers are bound to these targets (Fig. 3.1, lanes 10 and 14 and Fig. 3.2, lanes 10 and 14). With Trp repressor mutant RI69 and the 31 bp long/?-centred targets with a T/A substitution in position 5, only slight retardation is visible as a less shifted band (Fig. 3.1). Short/?-centred targets were not retarded by mutant Trp repressor RI69 (data not shown).

Discussion

In vivo analysis of binding of a-centred trp operator variants to Trp repressor mutants with substitutions in position 69 Trp repressor mutant RG69 represses the expression of/?-galactosidase under the control of c~-centred trp operator variants with substitutions at position 9 even better than does wild-type Trp repressor and exhibits minor repression of all other operator variants. This indicates loss of contact between residue 69 and base 9 (Ebright 1991; Giines et al. 1995). Trp repressor mutants RI69, RL69 and RQ69 broaden the specificity of the Trp repressor. They recognize all trp operator variants with substitutions in position 9 more efficiently than wild-type Trp repressor and recognize these c~-centred trp operator variants (pWB7391, pWB7392 or pWB7394) slightly better than the consensus a-centred trp operator (pWB7300). Due to the nature of their side chains, Trp repressor mutants display varying preferences for bases present in position 9 of the a-centred trp operator variants. The hydrophobic side chains of isoleucine and leucine may interact with methyl groups of thymines in a hydrophobic manner. Thus, these two mutants bind the a-centred trp operator variant with T/A in position 9 (pWB7394) better than trp operator variants with A/T or C/G in position 9 (pWB7391 or pWB7392). On the other hand, the side chain of glutamine of RQ69 is better suited to making hydrogen bonds with the adenine of the A/T base pairs in position 9 of the a-centred trp operator (pWB7391). Gln69, thus, binds this trp operator better than the trp operator variant with T/A in position 9 (pWB7394). Gln69 might also contact the adenine presented on the opposite strand in position 9 (pWB7394) whereas Ile69 and Leu69 could contact thymine presented on the opposite strand of the A/T base pair of pWB7391. The relatively high repression values obtained with the a-centred trp operator variant with C/G in position 9 (pWB7392) and Trp repressor mutants RI69 or RL69 are probably due to van der Waals contacts with ring CH atoms of cytosine and the hydrophobic side chains of isoleucine or leucine (Suzuki 1994). The interaction of Gln69 with cytosine of C/G9 possibly occurs by hydrogen bonding.

343 I

I 9 AGTTtAACT A G T A C G A T GGA*A* C 0L C A G G ~ T A G C 0~A T G A~T C A A I T T G AI~T C A TI~G C T A~C C T T C I C

A

31

bp

a consensus

C

TCC

ATCG

9

TACT

I

C

B

C

23 bp

23 bp

consensus

a consensus

CC

C

C C

C C C

D

23 bp

a 9T

C

C

I

5 I AGCG TACTtAGTA

TCT

5 CGCA

TGA*A*

A G A~T C G C(~A T G AI~T C A TI~G C G T~A C T T f

9 TCG TACT

I

fl

I AGTTIAACT

AGTA

9 CGA*A*

A G CC~A T G APT C A At('T~ T G APT C A T°~G C T T I i I

9

I

9

e T ON: A C TCG,,,AAC , A G , ,Z1 G,'A" c C

C

AGALAJATGA T C A A I T T G A I I

C

In the co-crystal (Otwinowski et al. 1988), Arg69 binds to G9 by bidentate hydrogen bonds. A recently refined solution of the NMR structure of theTrp repressor-trp operator complex (Zhang et al. 1994) and chemical studies (Marmorstein et al. 1991; Smith et al. 1994) corroborate this observation. Our genetic data obtained with synthetic trp operator variants and Trp repressor mutants with substitutions in position 69 are in agreement with these results. The contributions of individual bases in the trp operator to specific inter-actions with the Trp repressor have been described by several groups (Marmorstein et al. 1991; Czernik et al. 1994; Giines et al. 1995) whereas Bass et al. (1987) found, in a genetic analysis, less than a fivefold contribution of G/C base pairs in position 9 to specificity. We do not understand the reason for this discrepancy. Finally, our in vivo results suggest weak interaction of the side chain of Arg69 with bases in position 8 of the c~-centrcd trp operator (Fig. 1). The hydrophobic stem of arginine could interact with thymine in position 8.

TCAT

T~CTT

D

Fig. 3.1 Gel retardation assays with Trp repressor mutant RI69 and the short (23 bp) e-centred consensus target or variants carrying symmetrical basepair substitutions at base pair 9. The corresponding DNA sequences (A-D) are given below. A The 65-mer oligonucleotide is similar to that used by Haran et al. (1992), forming an asymmetrical e-centred 31-mer duplex D N A with a short loop formed by five cytosines. The Klenow fragment of D N A polymerase I was used to label the DNA by a fill-in reaction at the protruding 5' end. Binding conditions are described in Materials and methods. B The target is the 49-mer oligonucteotide that forms the 23-mer hairpin/~-centred consensus target, which was shown to bind two Trp repressor dimers (Haran et al. 1992). C The target is the 23-mer, e-centred trp operator, which binds one Trp repressor dimer (Haran et al. 1992). D The guanines at position 9 of the e-centred consensus target are replaced by thymines. Substituted base pairs are boxed. e- and//-axes are indicated. Aliquots (25 gg) of protein of a crude extract containing wild-type or mutant Trp repressor were used for complex formation. ( - ) indicates free D N A without crude extract. Triangle indicates crude extract with an inactive Trp repressor mutant as a negative control (this Trp repressor mutant lacks almost the whole HTH motif; see Materials and methods); wt indicates crude extract with wild-type Trp repressor; RI69 indicates the Trp repressor mutant with isoleucine in position 69. All complexes were run on the same gel

344 Fig. 3.2 Gel retardation assays with Trp represser mutant RI69 and long (31 bp) e- and/~-centred consensus targets and their variants. The D N A sequences (A-F) are indicated below. A The 23-mer e-centred consensus target. B The 23-mer/~-centred consensus target. C The consensus c~-centred 3liner duplex (for A, B and C, see Fig. 3.1.). D In the 31 bp c~centred target, the guanines at position 9 are replaced by thymines. E The 31-met consensus fl-centred target. F The 31-mer fi-centred target with guanine to thymine substitutions in position 5. Substituted basepairs are boxed. RI69 indicates the Trp represser mutant with isoleucine in position 69. All complexes were separated on the same gel. For further details see Fig. 3.1

!

C

A

23 bp

C~consensus

CC

C

9 TCG TACT

I

C

AGCCLATGA~TCAAI~TTGA~TCAT~GCT I

AGTT

AACT

AGTA

9 CGA*A* T

I

B

23 bp

C C TCT consensus

CC

5 II AGCG TACT AGTA

5 CGCA

TGA*A*

A G A~T C G C(IA T G A~T C A T~G C G T~A C T T C I I

C

31 bp

consensus

C

C C

C

D

E

31 bp

31 bp

o~ 9T

C

C

C C C C

~consensus

TCC

ATCG

9

TACT

~,

AGTT

AACT

AGTA

9

CQAT

GGA*A*

A G G~T A G C (IA T G A~T C A AO~T T G AI~T C A T0~G C y A~C C T y I I 1

TCC Tell ACT,GTT'AACTCGT

CAT,CA'A"

AGG[3~AGL~ ATGA TCAAe~TTGA TeATITICT AI~CCTT t I

C

C

C

TCC

5

li

5

G G C G A T C G T A C T A G T A C G A T A G C C GGA*A" 17 C A G G C C G C~T A G CaA T G A~3T C A TEtG C T A~T C G GC~CC T T C I t I

F

31 bp

~3 5T

In vivo analysis of/%centred trp operator variants with Trp represser mutants with substitutions in position 69 We found strong evidence for direct interaction with the side chain of residue 69 and bases in position 5 of the/?-centred trp operator. Mutant RI69 changed the specificity (for definition see Giines et al. 1995) of the Trp represser for the variant of the fi-centred trp operator with thymine in position 5 (pWB7254) (Fig. 2). Mutants RL69 and RQ69 broadened the specificity (for definition see Giines et al. 1995) of the Trp represser for trp operator variants with substitutions in position 5.

C

C

TCOIGGGG

C

AGG CCGC~TAGLAJATGA

C

ATcF~2ACT

AGT I

GA

TCATITICTA

CCcIGGA*A* TCGG

CCTT

I

fi-centred trp operators and their variants can be seen to the left and to the right of the central axis cc If so, positions _+5 of the fl-centred trp operator correspond to positions +_9 and +_1 of the e-centred trp operator (Fig. 4). Thus, it is not surprising that the specificity changes obtained with the fl-centred trp operator variants and the Trp represser mutants with substitutions in position 69 correspond to those obtained with a-centred trp operator variants. This result is in agreement with our previous observations, which showed that Trp represser mutant IW79 recognizes optimally an e-centred trp operator variant that can be seen to consist of two truncated fl-centred trp operator

345 plasmid pWB 7254

~t 5T

7300

~ consensus

7254

13 5T

65 4321 1234 TT*t ACT'AGTA*aA

56

109

8765 432t t234 5678 810 T G * T A C T * A G T T "A A C T * A G T A" C A ¢z cL ~ T t-TACT*AGT a*AA 65 4321 1234 56

Fig. 4 Alignment of the fl-centredtrp operator variant with T in position 5 (pWB7254)with the right and lefthalf-siteof the c~-centred consensustrp operator (pWB7300).All DNA sequencesare givenin the 5' to 3' direction. The base pairs are numberedfrom the centres of symmetryin the 3' to 5' direction. For simplicity,only position 1-10 of the e-centred operator and position 1-6 of the fl-centred operator are shown (see Figs, 1 and 2). Positions 1 and 9 in each half-siteof the e-centredtrp operator and position 5 in each halt-site of the fl-centredtrp operators are indicatedin bold letters. Bases in the/~-centredtrp operator variant that correspondexactlyto bases in the e-centred consensus trp operator are indicated in capital letters

variants with corresponding base pair substitutions (Giines et al. 1995). But why do Trp repressor mutants RI69, RL69 and RQ69 bind the consensus e-centred target (pWB7300) with high affinity? Although repression with these Trp repressor mutants and the consensus ]?-centred target (pWB7200) is 20- to 40-fold weaker than with wild-type Trp repressor, repression with these Trp repressor mutants and e-centred consensus trp operator (pWB7300, Fig. 1) is only about five- to sixfold weaker than with wild-type Trp repressor. Since isoleucine, leucine and glutamine are not candidates for an interaction with guanine in position 9, it is surprising that Trp repressor mutants carrying these amino acids in position 69 yield high repression values with the consensus e-centred target. This discrepancy can be explained by the fact that the consensus e-centred target mimics two (albeit imperfect) ]?-centred trp operator variants with T/A in position 5 (Fig. 4). Thus, mutant Trp repressors could recognize the e-centred consensus trp operator (pWB7300) as fl-centred trp operator variants with T/A in position 5 (pWB7254). The relatively high repression values could be due to cooperative or tandem binding to the long e-centred consensus target in vivo. Cooperative binding of two wild-type Trp repressor dimers to the e-centred trp operator has been described (Carey et al. 1991; Haran et al. 1992; Liu and Matthews 1993, 1994; Beckmann et al. 1993). In vitro analysis with Trp repressor mutant RI69 and short or long e-centred targets In the gel retardation assay, Trp repressor mutant RI69 binds to an e-centred trp operator variant with thymine in position 9 when this is presented as short or long hairpin DNA fragments, whereas these targets are not bound by wild-type Trp repressor (Fig. 3.1 and 3.2).

Trp repressor mutant RI69 also binds well to the e-centred consensus target. Thus, the specificity of this Trp repressor mutant for the trp operator variant with thymine in position 9 is broadened. Gel retardation experiments were carried out with crude cell extracts. We are aware that the use of cell extracts in vitro can be misleading in some cases. However, as the results presented here corroborate the in vivo results (Fig. 1) with respect to binding specificity, we believe that crude cell extracts can be used for a qualitative analysis in the case of the Trp repressor. The binding of two wild-type Trp repressor dimers to the e-centred target is similar to binding to the /7centred target, if the DNA is long enough (Fig. 3.2.). Haran et al. (1992) and Staacke (1993) have shown that only the short (23 bp) e-centred target binds only one Trp repressor dimer, e-centred targets that are shorter do not bind wild-type Trp repressor in the gel retardation assay, whereas targets longer than the 23 bp ecentred target bind two wild-type Trp repressor dimers depending on repressor concentration (Staacke 1993). With Trp repressor mutant RI69 and short or long e-centred targets a greater band shift occurs, with more than two mutant Trp repressor dimers being bound per target (Fig. 3.1 and 3.2). By comparing the relative mobility of this protein-DNA complex with the mobility of the complexes with one or two wild-type Trp repressors bound to the e- or ]?-centred targets, we speculate that possibly three dimers of mutant Trp repressor RI69 are bound to the e-centred targets. On the other hand, we observe bands of higher mobility with mutant RI69 and the long ]?-centred trp operators (Fig. 3.2). We speculate that these fast-moving bands probably reflect tetrameric occupation of these/?-centred trp operators by two Trp repressor dimers (RI69) and the greater mobilities of these complexes are possibly due to the loss of four positive charges on arginines, one from each subunit. We have shown that mutants in position 69 alter the specificity of the Trp repressor for trp operator variants with substitutions in position 9 of the e-centred trp operator and with substitutions in position 5 of the ]?-centred trp operator. These results confirm X-ray (Otwinowski et al. 1988) and N M R (Zhang et al. 1994) data showing that Arg69, the second residue of the first helix of the HTH motif, interacts with guanines in position +9 of the e-centred target. Direct contacts with the second residue of the first helix of the HTH motif and the bases of the operator DNA were also demonstrated in the case of 2, Cro (Brennan et al. 1990) and Tet repressor (Baumeister et al. 1992). The side chain of the )~ Cro repressor contacts bases far from the centre of symmetry, as does the Trp repressor. Furthermore, the results presented here confirm our previous findings that base pair substitutions in corresponding position in the e- and ]?-centred targets have similar effects on Trp repressor mutants with respect to specific interactions (Giines et al. 1995). In vivo and

346

in vitro results support the idea that the c~-centred trp operator can be considered to consist of two overlapping/bcentred half-sites. Each fl half-site can be occupied by two wild-type Trp repressor dimers, whereas both half-sites are unlikely to be bound at the same time. Protein-protein contacts, as observed in the cocrystal with the/~-centred target (Lawson and Carey 1993), enhance the tetrameric binding of two wild-type Trp repressor dimers. Binding of two wild-type Trp repressor dimers either to the left or to the right halfsite of the central axis e seems to be in balance in vitro. We expect that such occupation of trp operators might also exist in vivo. Acknowledgements We thank Dr. Andrew Barker for critical reading of this manuscript and Karin Otto, Gudrun Zimmer and Daniela Tils for excellent technical help. This work was supported by GIF.

References Bass S, Sugiono P, Arvidson DN, Gunsalus RP, Youderian P (1987) DNA specificity determinants of Escherichic~ coli tryptophan repressor binding. Genes Dev 1 : 565-572 Baumeister R, Helbl V, Hillen W (1992) Contacts between Tet repressor and tet operator revealed by new recognition specifities of single amino acid replacement mutants. J Mot Biol 226:1-14 Beckmann P, Martin SR, Lane AN (1993) Interaction of the Trp repressor with the trp operator DNA fragments. Eur Biophys J 21:417-424 Bennett GN, Yanofsky C (1978) Sequence analysis of operator constitutive mutants of the tryptophan operon of Escherichia coll. J Mol Biol 121:179-192 Bennett GN, Schweingruber ME, Brown KD, Squiers C, Yanofsky C (1976) Nucleotide sequence of region preceding trp mRNA initiation site and its role in promoter and operator function. Proc Natl Acad Sci USA 73 : 2351--2355 Brennan RG, Roderick SL, Takeda Y, Matthews BW (1990) ProteinDNA conformational changes in the crystal structure of a )~Crooperator complex. Proc Natl Acad Sci USA 87:8165-8169 Carey J, Lewis DEA, Lavoie TA, Yang J (1991) How does Trp repressor bind to its operator? J Biol Chem 266:24509-24513 Czernik PJ, Shin DS, Hurlburt BK (1994) Functional selection and characterization of DNA-binding sites for Trp repressor of Eseherichia coli. J Biol Chem 269:27869-27875 Ebright RH (1991) Identification of amino acid-base pair contacts by genetic methods. Methods Enzymol 208 : 620-640 Grove CL, Gunsalus RP (1987) Regulation of the aroH operon of Escherichia coli by the Tryptophan repressor. J Bacteriol 169: 2158-2164 Gunsalus RP, Yanofsky C (1980) Nucleotide sequence and expression of Escherichia coli Trp repressor, the structural gene for the Trp aporepressor. Proc Natl Acad Sci USA 77:7117-71221 Giines ~, Staacke D, 'yon Wilcken-Bergmann B, Mi.iller-Hill B (1995) The possible roles of residues 79 and 80 of Trp repressor of Escherichia coIi K-12 in trp operator recognition. Mol Gen Genet 246:180-196 Itaran TE, Joachimiak A, Sigler PB (1992) The DNA target of the Trp repressor. EMBO J 11:3021 3030 Heatwole VM, Somerville RL (1991) The tryptophan-specific permease gene mtr is differentially regulated by the Tryptophan and Tyrosine repressors in Escherichia coli K-12. J Bacteriol 173 : 3601-3604

Heatwole VM, Somerville RL (1992) Synergism between the Trp repressor and Tyr repressor in repression of the aroL promoter of Escherichia coli K-12. J Bacteriol 174:331-335 Hutchison CA, Phillips S, Edgell MH, Gillam S, Jahnke PA, Smith M (1978) Mutagenesis at a specific position in a DNA sequence. J Biol Chem 253 : 6551-6560 Kumamoto AA, Miller WG, Gunsalus RP (1987) Escherichia coIi Tryptophan repressor binds multiple sites within the aroH and trp operators. Genes Dev 1 : 556-564 Lawson CL, Carey J (1993) Tandem binding in crystals of a trp repressor/operator half-site complex. Nature 366:t78-182 Lehming N, Sartorius J, Niem611er M, Genenger G, von WitckenBergmann B, Mfiller-Hill B (1987) The interaction of the recognition helix of Lac repressor with Iac operator. EMBO J 6:3141-3153 Liu Y-C, Matthews KS (1993) Dependence of Trp repressor-operator affinity, stoichiometry, and apparent cooperativity on DNA sequence and size. J Biol Chem 268 : 23239-23249 Liu Y-C, Matthews KS (1994) Trp repressor mutations alter DNA complex stoichiometry. J Biol Chem 269 : 1692-1698 Marnaorstein RQ, Sprinzl M, Sigler PB (1991) An alkaline phosphatase assay to investigate trp repressor/operator interactions. Biochemistry 30:1141-1148 Miller JH (1972) Experiments in molecular genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York Otwinowski Z, Schevitz RW, Zhang R, Lawson CL, Joachimiak A, Marmorstein RQ, Sigler PB (1988) Crystal structure of Trp repressor/operator complex at atomic resolution. Nature 335: 321-329 Pfau J, Arvidson DN, Youderian P, Pearson LL, Sigman DS (1994) A site specific endonuclease derived from a mutant Trp repressor with altered DNA-binding specificity. Biochemistry 33: 11931 11403 Rose JK, Squires CL, Yanofsky C, Yang H, Zubay G (1973) Regulation of in vitro transcription of the tryptophan operon by purified RNA polymerase in the presence of partially purified repressor and tryptophan. Nature New Biol 245:133-137 Sambrook J, Fritsch ET, Maniatis T (1989) Molecular cloning: a laboratory manual (2nd edn) Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York Sanger F, Nicklen S, Coulson AR (1977) DNA sequencing with chain-terminating inhibitors. Proc Natl Acad Sd USA 74: 5463-5467 Smith SA, Rajur SB, McLaughlin LW (1994) Specific purine N vnitrogens are critical for high affinity binding by the trp repressor. Struct Biol 1:18-22 Staacke D (1993) Kooperative Bindung yon Trp-Repressor an c~und /~-zentrierte trp Operatoren. PhD Thesis, University of Cologne Staacke D, Walter B, Kisters-Woike B, yon Wilcken-Bergmann B, Mfiller-Hill B (1990) How Trp repressor binds to its operator. EMBO J 9 : 1963-1967 Suzuki M (1994) A framework for DNA-protein recognition code of the probe helix in transcription factors: the chemical and stereochemical rules. Structure 2:317-326 Walter B (1989) Untersuchungen zur Interaktion des Trp repressors mit verschiedenen trp Operatoren. Diploma Thesis, University of Cologne Youderian P, Arvidson DN (1994) Direct recognition of the trp operator by Trp holorepressor - a review. Gene 150:1-8 Zhang H, Zhao D, Revington M, Lee W, Jia X, Arrowsmith J, Jardetzky O (1994) The solution structures of the trp repressoroperator DNA complex. J Mol Biol 238 : 592-614 Zurawski G, Gunsalus RP, Brown KD, Yanofsky C (1981) Structure and regulation of aroH, the structural gene for the tryptophanrepressible 3-deoxy-d-arabino-heptulosonicacid-7-phosphate synthetase of Escherichia coli. J Mol Biol 145 : 47-73