Contacts between Tet repressor and tet operator revealed by new recognition specificities of single amino acid replacement mutants

J. Mol. Biol. (1992)

226,

1257-1270

Contacts between Tet Repressor and tet Operator Revealed by New Recognition Specificities of Single Amino Acid Replacement Mutants Ralf Baumeister, Vera Helbl and Wolfgang Hillen? Lehrstuhl ftir Mikrobiologie Fried&h-AlexanderUniversittit Staudtstr. 5, D-8520 Erlangen, Germany (Received

6 January

1992; accepted 8 May

1992)

We have analyzed the DNA binding properties of Tet-repressor mutants with single amino acid residue replacements at eight positions within the a-helix-turn-a-helix DNA-binding motif. A saturation mutagenesis of Gln38, Pro39, Thr40, Tyr42, Trp43 and His44 in the second a-helix was performed; in addition, several substitutions of Thr27 and Arg28 in the first a-helix were constructed. The abilities of these mutant repressors to bind a set of 16 operator variants were determined and revealed 23 new binding specificities. All repressor mutants with DNA-binding activity were inducible by tetracycline, while mutants lacking binding activity were trans-dominant over the wild-type. All mutant proteins were present at the same intracellular steady-state concentrations as the wild-type. These results suggest the structural integrity of the mutant repressors. On the basis of the new recognition specificities, five contacts between a repressor monomer and each operator half-site and the chemical nature of these repressor-operator interactions are proposed. We suggest that Arg28 contacts guanine of the G. C base-pair at operator position 2 with two H-bonds, Gln38 binds adenine of the A* T base-pair at position 3 with two H-bonds, and the methyl group of Thr40 participates in a van der Waals’ contact wit’h cytosine of the G. C base-pair at position 6 of tet operator. A previously unrecognized type of interaction is proposed for Pro39, which inserts its side-chain between the methyl groups of the thymines of T. A and A. T base-pairs at positions 4 and 5. Computer modeling of these proposed contacts reveals that they are possible using the canonical structures of the helix-turn-helix motif and R-DNA. These contacts suggest an inverse orientation of the Tet repressor helix-turn-helix with respect to the operator center as compared with non-inducible repressor-operator complexes, and are supported by similar contacts of other repressor-operator complexes. Keywords:

Tet repressor; helix-turn-helix recognition specificity;

1. Introduction The sequence-specific recognition of DNA plays a central role in the control of transcription in all organisms. Despite the structural and functional diversity of DNA-binding proteins, they contain only a limited number of different DNA-binding motifs (Steitz, 1990; Harrison & Aggarwal, 1990; Harrison, 1991). Of these, the a-helix-turn-a-helix (HTHf) motif has been most extensively studied. It consists of 21 amino acid residues which fold into

7 Author to whom all correspondence should addressed. $ Abbreviations used: HTH, a-helix-turn-a-helix sequence motif: bp, base-pair(s).

be

motif; regulatory

protein-DNA proteins

interaction;

two a-helices oriented almost perpendicular to each other and which are separated by a short surface turn of three amino acid residues (Pabo & Sauer, 1984). Detailed analyses of protein-DNA complexes by X-ray crystallography and nuclear magnetic resonance spectroscopy have provided insights into t,he interactions between the different HTH motifs and their target sequences, revealing several common aspects of DNA recognition (Jordan & Pabo, 1988; Aggarwal et al., 1988; Mondragon et al., 1989a,b, 1991; Pabo, 1990). The carboxy-terminal “recognition” a-helix lies in the major groove and is held in place by an amino-terminal a-helix via a “hydrophobic brace”. Many of the sequence-specific contacts to DNA are made by amino acid residues

1257 0

1992 Academic

Press

Limited

1258

R. Baumeister et al.

(a)

(b)

T A t *123456789

ACTCTATCATTGATAGAGT TGAGATAGTAACTATCTCA 987654321*

a

T (c) Figure

1. Structures

of the HTH motif of Tet repressor and tet operator. (a) Schematic presentation of the HTH motif of TnlO-derived Tet repressor between amino acid residues 27 and 47. The 2 cc-helices are drawn as tubes with circles indicating the positions of residues mutated in

in the recognition helix. These specific contacts include hydrogen bonds and hydrophobic interactions. Although the overall mode of binding is very similar in most of these complexes, the available data do not allow us to identify a code for recognition or to predict sequence-specific interactions from the primary structures. It is therefore necessary to study a variety of examples of proteinDNA recognition. The TnlO-encoded Tet repressor has an HTH motif between amino acid residues 27 and 47, as supported by sequence comparisons and genetic studies (Postle et al., 1984; Isackson & Bertrand, 1985; Wissmann et al., 1991b; Baumeister et al., 1992). Tet repressor binds a palindromic tet operator of 19 base-pairs. A detailed analysis of t’he tet operator has shown that base-pairs (bp) at positions 2 to 6 (for nomenclature, see Fig. l(c)) in each halfsite are essential for TetR binding (Sizemore et al., 1990). Previously, we examined the binding of mutant repressors in which each amino acid residue in the HTH motif was changed, in turn, to alanine. The ability of each mutant repressor to bind the wild-type and 15 variant operators suggested that Thr27 and Arg28 in the first m-helix and Gln38, Pro39, Thr40, Tyr42, Trp43 and His44 in the second a-helix are critical for operator binding (Wissmann et al.. 1991b) (see Fig. l(a)). Furthermore, these results predict three specific contacts between amino acid residues of repressor and particular basepairs: Gln38 and 3 A. T, Pro39 and 4 T ’ A, and Thr40 and 6 G-C. Whereas the mechanism of Gln38 recognition of bp 3 is precedented, we have no hints how Pro and Thr can interact with DNA. Furthermore, no repressor contacts to the essential bp 2 in the tet operator (Heuer & Hillen, 1986; Hillen et al., 1984; Wissmann et al., 1988) have been identified. Here, we describe the phenotypes of mutant Tet repressors with single amino acid residue changes of Thr27 and Arg28 in the first a-helix and of amino acid residues 38 through 40 and 42 through 44 in the second a-helix (see Fig. l(a)) and identify a large number of changes that affect the binding specificity of Tet repressor. On the basis of these results, we present a molecular model of the Tet repressoroperator interface. this study. The turn between the helices is indicated by a solid line. (b) Helical wheel projection of residues 38 to 45 in the DNA-recognition helix starting from the aminoterminal end. Residues mutated in this study are shown in inverse print with white letters on black background. (c) Nucleotide sequence of tet operator 0, and its mutant 2T. Both palindromic half-sites are marked by lines and the numbering of base-pairs is indicated. The stars indicate the central bp. Operator mutants used in this analysis contain symmetric mutations in both half-sites. An example is shown with operator 2T. The nucleotide sequence of operator O2 differs from 0, only in the central G-C bp and an A.T to G.C exchange at position 7 in 1 half-site. The central bp shows no, and bp7 only a small contribution to repressor recognition (Wissmann et al., 1988).

DNA 2. Materials

Recognition Speci$cities

and Methods

(a) Materials and general methods Materials, enzymes, media and general methods used in this study have been described in detail (Baumeister et al., 1992). Bacterial strains are derivatives of Escherichia coli K12. E. coli WH207 (Wissmann et al., 1991a) served as host of j-galactosidase assays. The plasmid pWHlO12, its 15 derivatives containing mutant operators (Sizemore et al., 1990), pWH510 (Altschmied et al., 1988), pWH1401, pWH1411 (Baumeister et al.. 1992), and the lysogenic bacteriophage &et50 (Bertrand et aE., 1984; Wissmann et aZ.. 1991a) have been described. pWH806 and pWH853 (Wissmann et al., 1991a) were used to probe trunsdominant phenotypes of repressor mutants as described. Tet-repressor mutants of amino acid residues at positions 38, 39, 40, 42, 43 and 44 were constructed by cassette mutagenesis detailed by Baumeister et al. (1992). Mutants that had not been obtained by cassette mutagenesis and mutants of amino acid residues 27 and 28 were introduced by oligonucleotide-directed mutagenesis according to the method of Su & El-Gewely (1988). (b) b-Gulactosidase

assays

The binding of Tet repressors to tet operators and inducibility by tetracycline were determined in E. coli WH207 transformed with plasmid pWH1012 or its derivatives containing tet-operator variants, and compatible plasmid pWH1411 derivatives encoding mutant Tet repressors. Cultures for j?-galactosidase assays were grown in Luria broth supplemented with appropriate antibiotics to an absorbance at 600 nm of @4 at 37°C. The specific j?-galactosidase activities were determined as described by Miller (1972). The inducibility of Tet-repressor variants was tested using @2 pg tetracycline/ml in both overnight and log-phase cultures. The trans-dominant phenotype of repressor mutants was assayed in E. co2i WH207(&et50) transformed with pWH853. This plasmid harbors a wildtype tetR gene, which is constitutively expressed at a low level. Details of the /Y-galactosidase assays have been described (Wissmann et al., 1991b; Baumeister et al., 1992). (c) Probing

the steady-state level8 of mutant

repressor

Total cell protein from E. coli WH207/pWH1012/ pWH1411 harboring derivatives of the tetR gene was prepared and analyzed as described by Wissmann et al. (1991b). TetR protein bands in Western blot experiments were visualized using TetR specific polyclonal antibody (a gift from Dr C. F. Beck). (d) Model buildiw The amino acid nomenclature was used according to the ILJPAC-IUB Commission on Biochemical Nomenclature (1970). Model building was done on an Evans $ Sutherland PS390 graphic system hosted by a MicroVAX 350 using Insight (Biosym Inc., San Diego, CA). Protein data for 434 Cro repressor were from the Brookhaven protein data bank (Bernstein el al., 1977).

3. Results and Discussion (a) Mutants of Tet repressor in the helix-turn-helix

motif

We have constructed 121 different mutants of Tet repressor encoded by plasmid pWH1411 with single

of Tet-Repressor Mutants

1259

amino acid residue substitutions at positions 27, 28, 38 to 40 and 42 to 44. Thr27 and Arg28 are located at the N terminus of the first a-helix. where amino acid residues in some HTH proteins contribute to sequence-specific binding. A helical wheel projection of the second a-helix shows that, the side-chains of residues 38 to 40 and 42 to 44 define the solventexposed face of this helix (see Fig. 1 (b)) (Hansen Dz Hillen, 1987; Hansen et al., 1987). Consistent with this idea, we have shown that, to maintain binding, Leu41 on the other side can be replaced by only hydrophobic residues (Baumeister et a,Z., 1992). (b) In vivo systems of the analysis repressor-operator interactions

oj

The mutants were analyzed for their ability to repress fl-galactosidase expression of tetA-la& fusions in vivo. For this purpose, we used the two plasmids pWH1411 and pWH510, which constitutively express different amounts of Tet repressor. pWH1411 produces TetR at more than fourfold higher steady-state concentrations than does pWH510 (Wissmann et al., 1991b). These properties lead to different efficiencies of repression of tetA1acZ fusions in trans, provided either on the compatible plasmid pWHlOl2 (Sizemore et al., 1990) or on a single copy 2 lysogen in E. coli WH207(&et50) (Wissmann et al., 1991a). Two combinations used in this work are depicted in Figure 2(a) and (1)). Using the pWH14ll/pWHl012 combination (Fig. 2(a)), it was possible to analyze repressor mutants with severe reductions of operator affinity. In addition, we have used 15 derivatives of pWH 1012 containing tet operator variants to ident)ify new recognition specificities. We will refer to this system as the “high” expression system. The pWH5lO/,%tet50 combination (Fig. 2(b)) was employed to examine repressor mutants that gave nearly wild-type levels of binding affinity in the “high” expression system. We call this combination the “low” expression system. When appropriate, a combination of plasmids pWH510 and pWH1012 was also used. The abilities of several mutant proteins to repress expression of p-galactosidase were determined in both systems. A comparison of the results is shown in Figure 2(c). In the “high” expression system, repression values of tightly binding mutants cannot be distinguished, whereas this is possible in the “low” expression system. Therefore. mutants yielding /?-galactosidase activity levels below 1.5 y. (relative to wild-type) in the “high” expression system were analyzed further using the “low” expression system. Quantitative studies with TetR mutants have shown that a lOOO-fold decrease in binding affinity measured in vitro (Hansen & Hillen, 1987) corresponds to relative /?-galactosidase activities of 14% in the “high” and 80% in the “low” expression system. The tetA-ZacZ fusions derived from E. coli WH207(11tet50) or from pWH1012 and its derivatives direct the expression of different amounts of B-galactosidase in the absence of Tet repressor

f ? 0

1260

R. Baumeister

et al

Table 1 P-galactosidase expression obtained with Tet repressor mutants with high afinity for operator variants

tetR

IacZ

pwH1411

TetR

cat

\

01

WY43 WH43 HW44 HY44 wt TS40 TA40 TA40 TS40 wt

wt wt, wt wt wt wt wt 6A 6A 6A

lO@O( f4+)t 52( + o.a)t 231(k2.4)? 12.9( f 2.2)? 152( + 1.6)t 14q + 1.5)7 243( k 1.3)t 6@9( + 1.5)x 7CO( * 2.0)s 965( k57)$ 664( k 06)$ 7@4( +0.4)$ 82.9( + 1.0):

All activities are expressed relative to the wild-type tet operator defined as 100%. Abbreviation used: wt, wild-type. t Determined in the “low” expression system (pWHFilO/ Itet50: see Fig. 2(b)). fDetermined in the mixed system (pWH510/pWH1012: see Fig. 2(a)).

n

Met50

YF42

wt wt wt

wt (a)

Relative fi-galactosidase activity

tet operator variant

mutant

L

02

IacZ

(Sizemore

ib)

A

I

et al.,

1990).

To

compare

the

binding

efficiencies of Tet repressor to the set of operator mutants, we have defined the constitutive fl-galactosidase expression in the absence of TetR as 100 %, and /Cgalactosidase activities in the presence of mutant Tet repressors are given relative t’o this value.

I

(c) Binding

of Tet-repressor variants

mutants

to

operator

Each

IA

-..I

0

20 40 60 80 100

/3-galaciosidase activity pWH5lO/Xtet50

(%)

for

(cl Figure 2. In viva systems used in the analysis of Tet repressor-tet operator interactions. (a) “High” expression system for the analysis of repressor mutants with a severe reduction of their operator affinity. Circles symbolize the plasmid DNA and large boxes the genes for tetR and la&. Their directions of transcription are indicated by arrows. Small boxes represent tet operators O1 and OZ. The Tet-repressor dimers are drawn as double spheres. (b) “Low” mutants repressor.

expression system for the with operator affinity similar Phage 1 DNA is represented

analysis of repressor to that of wild-type by a line and the

other symbols P-galactosidase

are as described in (a). (c) Comparison of activities obtained in both in vivo

systems.

/3-galactosidase

The

activities

obtained

pWH1012)

activities

obtained

with

the

variant

was tested

for binding

to

with the levels of wild-type (data not shown). Therefore,

Tet repressor at 37°C the differences in the repressors to tet operators

binding of these mutant should not be due to different proteins

amounts

of these

in the cell.

Some of the mutant proteins did not show any repressor activity. A representative set of those nonbinding mutants and the previously described GE21 mutant (Isackson & Bertrand, 1985) were tested for trans-dominant described a phenotype as

with

several TetR mutants in the “low” expression system (pWH510/ltet50) are shown on the horizontal axis and the P-galactosidase mutants in the

repressor

wild-type operator in the “high” expression system. The results are presented in Figure 3. Several mutants that directed repression levels similar to those of wild-type were recloned into pWH510, and repression was determined in the “low” expression system (see Table 1). Western blot analyses of total cell proteins from E. coli WH207/pWH1012 transformed with derivatives of pWH1411 (see Fig. 2(a)) showed that the steady-state levels of Tet repressor proteins in the cells are the same for all mutants, and are identical

respective

“high” expression system (pWH1411/ are given on the vertical axis. The bars

indicated by the filled a lOOO-fold difference

WF43-wild-type Hillen.

1987).

triangles of K,,

tet operator

show the range as determined

interaction

covered

by

by the TetR (Hansen &

DNA

a II K 1 s

03.1 90.0 91.0 93.0 91.9

t * f + t

1.5 2.1 4.0 3.4 6.0

1261

Recognition fhpeci$cities of Tet-Repressor Mutants

A 0 w P K Y 1 c R F 0 H L I E ” Y

23.7 57.6 82.3 85.5 94.3 95.6 95.8 96.2 95.5 95.4 100.0 ICQ.? 101.5 1025 1026 103.2 105.1

t 0.3 t 1.3 * 1.8 t 3.4 t 1.n * 1.4 t 1.9 * I.0 f 20 * 1.8 * 11 * 1.5 * 3.4 * 1.1 * 2.6 f 2.9 t 1.1

A a 1 K II Y C R 0 w D ”

2.2 35.7 81.0 96.0 96.1 an.0 90.4 se.5 90.5 91.7 99.1 90.4

* * t f * t * * t t f t

0.1 1.1 0.1 4.0 0.1 0.4 2.6 3.1 3.9 2.4 20 0.9

A Q c ” N K u 0 I w P L

11.3 24.1 72.1 85.8 89.3 90.3 91.5 93.9 94.3 94.6 94.9 95.1

,. t * f t t * t * t f *

4.2 0.6 10.5 1.9 3.4 28 3.9 2.3 4.0 5.5 8.1 2.2

I 925 t L 94.5 t w w.6 t R 91.1 t A 98.1 * w 90.4 t K 98.5 * v an.9 * H 99.4 t c 100.2 * P 1W.8 * E

100.9

t

1.0 27 5.4 2.2 1.8 3.9 1.8 3.6 1.5 3.5 6.2 2.9

S 101.2 t a 101.6 * a 101.1 t

2.3 1.4 5.0

1

28

101.9

*

D 102.1 t

F

14.1

t

Q 12.2 t s 79.5 d A 81.4 * c 84.7 * N 55.3 t R 97.5 t a 98.3 * E

101.2

t

1.5

1.1 4.0 3.3 3.1 3.2 5.a 2.8 3.4

D 102.9 f I 103.8 * 1 104.2 *

3.a 1.3 3.5

K P

105.3 lW.4

t *

22 3.9

L

lOs.5

t

Y 101.8 t 2.5 ] 1V 109.2 t

3.3

3.3 4.5

R K

so.8 * 11.4 t

4.9 3.5

F

83.5

t

2.9

0 0 s I A 1

94.5 B5.4 95.5 aa8 91.0 91.2

t * * f t *

5.3 3.4 5.4 3.0 1.1 5.5

L

97.7

*

3.9

P 98.9 t 5.2 v 99.1 t 5.5 M 99.3 t 3.2 E 99.8 e. 7.0 c 100.4 t 3.4 N

lW.5

t

2.3

Figure 3. In ,uivo operator binding affinities of mutant Tet repressors in the “high” expression system. The HTH motif is depicted at the top of the Figure, amino acid residues are given in the one letter code. Large boxes represent the 2 cc-helices. the small box the turn between them. Residues mutated in this study are printed in white on black background. Single substitutions for each residue are shown in the columns together with the p-galactosidase activities and standard deviations obtained for every mutant repressor. The mutants are aligned in order of decreasing affinity of wild-type operator. (Wissmann et al., 1991a). The results shown in Table 2 indicate that all mutants but one are dominant in bans to the wild-type, indicating that they can form heterodimers with wild-type repressor. The only mutant with a weak trans-dominant phenotype is QP38, probably because it shows residual DNA-binding activity. Therefore, the amino acid residue replacements of all mutants most likely do not change the overall structure of the protein. This hypothesis is also supported by the fact that the mutant proteins are present in the same steadystate levels as is the wild-type. Furthermore, all mutants with detectable binding activity are fully inducible by tetracycline (data not shown). Thus, it is reasonable to assume that the phenotypes of these changes can be attributed either to local structural perturbations or to their direct effects on operator recognition. In the next set of experiments, we tested whether any of the repressor mutations could be functionally suppressed by operator mutations. The repression levels of combinations of Tet-repressor mutants in pWH1411 with tet-operator variants in pWH1012 were determined. The results obtained with 202 combinations are presented in Table 3. To facilitate the comparison to other complexes of HTH motifs with their recognized DNA sequences, the primary structures of eight HTH motifs and their target DNA sequences are compared in Figure 4. They are oriented with respect to base-pairs 2 to 6 of tet operator, and are designated according to the tet

nomenclature (see Fig. 1). The crystal structures of five systems have been analyzed at high resolution, except for the ;1 Cro complex, for which only a lowresolution structure is available (Brennan et al., 1990). Extensive biochemical data and nuclear magnetic resonance data are available for the LacI system, (Boelens et al., 1987; Lamerichs et al., 1990; Lehming et al., 1987, 1988, 1990; Sartorius et al., 1989, 1991; deVlieg et al., 1989; Kaptein, et al., 1985, 1990). We also compared our results with the “indirect readout” proposed for the Trp repressoroperator complex (Otwinowski et al., 1988), although the specificity of this complex has been questioned (Staacke et al., 1990). (d) RoZe of Thr27

for

operator binding

Thr27 is the first residue of the HTH motif. The TA27 mutation of Tet repressor leads to a severe reduction in recognition of the wild-type operator (Wissmann et al., 1991b). In several HTH motifs, the first residue is involved in contacts with the phosphate backbone. The Gln residue of ;1 cl and 434 Cro contribute, in addition, to an extensive network of protein internal H-bonds defining the position of the Gln side-chains at’ the aminoterminal ends of the second helix (Pabo et aZ., 1990). However, the equivalent Glnl7 in 434 CI does not contribute to such a protein-protein interaction (Aggarwal et al., 1988). We therefore constructed Tet-repressor mutants of Thr27 and measured their

1262

9. Baumeister

et al.

Table 2 Transdominunce

of mutant

tetR allele TetR

wt

tetRt -wt fwt fwt +wt +wt +wt +wt +wt twt +wt twt +wt Cwt +wt +wt +wt twt +wt

wt GE21 TR27 RL28 RS28 QK38 QL38 QP38 QW38 PK39 PL39 PW39 TK40 TL40 TP40 TW40

Table 3 Tet repressors

/?-Galactosidase activity (%)I lOO.O( k2.1) 6.8( f 02) 1.2( + 0.0) 22.0( + 0.8) 23.0( k 0.8) 21.0( k@5) 196( +@2) 204( &- 0.3) 185( + 1.3) 1@1(+@3) 2@2(f@6) 2@9( +@l) 21.0( f0.8) 201( f04) 204( & @O) 155( f 1.1) 21.6( +@8) 21.8( +@8)

Derepression compared to wt&

57 1.0 183 192 17.5 163 17.0 154 84 168 17.4 17.5 168 17.0 129 18.0 182

For every position a positively charged, a hydrophobic (L), an aromatic (W) residue and a proline residue replacement mutant were tested. Transdominance of these mutants at positions 42,43 and 44 gave identical results. wt, wild-type. t E. coli WH207jket50 was used as a source of the tetd-la& transcriptional fusions. The tetR alleles are all derivatives of pWH1411. In addition to these plasmids E. coli cells contained a 2nd compatible plasmid, which was either pWH806 (-wt tetR) or pWH853 (+ wt tetR). For further description see Materials and Methods. $ 100% /l-galactosidase activity corresponds to 2630( f 100) units as defined by Miller (1972). 0 Given as ratio of P-galactosidase activity obtained with the respective tetR allele over that obtained with wt t&R.

abilities to bind the wild-type operator and variant operators with changes at positions 1 and 2 (see Fig. 3 and Table 3). The TS27 mutant does not show a large reduction in affnity of the wild-type operator. In contrast, a change to Gln results in a great loss of binding affinity, whereas a change to Arg completely abolishes binding to the wild-type operator. None of the three mutant repressors can recognize operator variants with changes at positions 1 and 2. These results suggest that the hydroxyl function of Thr may be more important for operator binding than is the methyl group. Our model of TetR-tet operator interaction (see below) would allow H-bonding of Thr27 to the phosphate groups. The results obtained with TS27 and TA27 and the proximity of the Thr27 side-chain to the operator backbone in the model of the complex would be consistent with this interpretation, but no direct evidence of this interaction is available. (e) Contact

of

Arg28

with

operator

Repression of fi-galactosidase expression with repressor mutants and tet operator

bp 2 G ’ C

Arg28 is located at the second position of the HTH motif. The RA28 repressor does not bind to wild-type operator (Wissmann et al., 1991b). All replacement mutants at this position that we have tested in this study also exhibit a largely reduced binding to the wild-type operator (see Fig. 3). None of these mutants recognizes operator variants lG,

Operator Repressor mutant

wt

variants?

1G

2A

2C

2T

3C

3G

3T

100 40 93 100 105 105 104 98 87 103 95

100 99 105 100 103 96 98 102 95 97 95 1:

100 43 93 95 99 96 61 72 4 51 51 105 114 101 104 96 96

100 75

100 94

100 87

21 56 33 91 99 73

2148 rs 90 95 73 97

!I! 91 43 91 19

82 96 97 103

~~

97 100 97 96

100 100 102 100 100 99 93 99 94 98 103 98 97 101 93 98 98

wt

4A

4C

4G

5C$

5G

5T

6A

6C

wt PS39 PA39 PG39 PT39 PI39 PL39 PN39 P&39 PV39

100 I 2 2 36 81 100 102 96 99 99

100 36 72 71 60 101 101 99 96 106 101

100 61 14 54 * 101 102 100 97 54 97

100 76 87 106 1 100 110 105 108 120 100

100 97 92 98 101 96 101 100 99 99 98

100 98 48 @ sz 103 98 99 &j 95 103

100 16 38 92 71 91 88 103 119 114 88

100 2 ~~

100 55

~~

~~ ~~

TA40 TS40 TC40 TN40 T&40 TV40

111 72 89 94 86

103 95 101 98 105 99

102 98 99 93 93 96

89 19 95 96 111 89

d~ 4 4 98 74

wt TS27 T&27 TR27 RQ28 RN28 RK28 RT28 RL28 RS28 QS38 &A38 QN38 QT38 QR38 QV38

100 1 1 47 105 84 90 91 93 103 98 2:

obtained

variants

~~ ~~ ~

~~

~~ ~ ~~ ~~ ~~

~~ :z 103 102 103 100

Results are given in percent of the maximal /l-galactosidase expression obtained in the absence of tetR (pWH1401). The standard deviation of each value was typically lower than 5%. wt, wild-type. -1 not determined; t Combinations of Tet repressor mutants with tet operator variants that yielded lower B-galactosidase activities than the combination of the respective operator variant with wt repressor are indicated by bold numbers and are underlined. $ dam methylated operator variant (see the text).

2A and 2C (Table 3). Repressors R&28 and RN28 show a largely reduced affinity for the wild-type operator with 2 G. C (84% and 90%, respectively) and no repression with operator variants lG, 2A and 2C. In addition, RQ28 does not bind operator 2T (96%), which is bound to 61 o/o efficiency (relative to wild-type) by RN28. RK28, with a mutation of Arg28 to Lys, the other positively charged amino acid, has a low binding activity for tested. This is not every operator we have surprising, since the side-chain of Arg has the potential of making up to five H-bonds, while the single amino group in the side-chain of Lys cannot form an extensive network of H-bonds. An RK mutation also leads to a loss of DNA binding of the bZIP domain of opaque-2 and loss of RNA binding of

DNA TrpR

OR

I

CAP

RO

RET

A Cl

GE 01

OSG OSA

Lad

L Y

434 Cl

OA 0 T T R

YQT 00s

ACm

434~10

TOW

AT

OR GAL NKA

I

SRC EOL QL

I

YWH

. . ii;;;;

434 cm

TACAA..

.

t

434 Cl

TACAA..

.

t

TATCA., TATCA..

. .

t t

UP

TGTGA..

.

t

trp

GTACT...,.

kc

l

.

Awe

Act

CTCAC

(f) Contacts

Figure 4. Comparison of 8 HTH motifs and their operator sequences. Residues that occupy identical positions in the HTH motif of 8 proteins are given at the top. Their location in the HTH motif is indicated by the relative numbering according to the system of Pabo & Sauer (1984). Only residues at positions mentioned in the

text are shown. At the bottom, the respective nucleotide sequences of the operator half-sites are presented. asterisk represents the center of each operator.

An

HIV-1 Tat protein (Aukerman et al., 1991; Calnan et al., 1991). The same is true for Lac repressor, where the RK17 mutation results in a complete loss of operator affinity (Kleina & Miller, 1990). TetR RT28 shows a largely reduced binding to the wild-type operator (93 y. /I-galactosidase activity), but has a new recognition specificity; it binds operator 2T (4 y. fl-galactosidase activity) better than does the wild-type repressor (42%). The C?’ methyl group of Thr must contribute to this interaction, because RS28 represses operator 2T only to 51%. From these data, we propose a sequence-specific interaction of Arg28 with bp 2 of tet operator. The only other HTH protein with an Arg at relative position 2 in Figure 4 is TrpR, where Arg69 contacts bp 2 G. C directly (Otwinowski et al., 1988) forming two H-bonds to N7 and 06 atoms of the guanine residue. A similar chemical interaction of Arg28 in TetR with bp 2 G. C is proposed here. This model also explains the methylation protection of the N7 atom of guanine in the presence of repressor protein (Hillen et al., 1984). In the 2T operator, the C5 methyl group of the thymine protrudes into the major groove and would sterically interfere with the large

side-chain

of

Arg28.

In

contrast,

1263

not result in a favorable contact. A similar interaction is predicted between Thrl7 and thymine at the same relative operator position (see Fig. 4) in the II Cro complex (Brennan et al., 1990)). Leu in TetR RL28 could make a similar contact, but shows only partial repression at operator 2T, which is perhaps due to its large hydrophobic side-chain located in an otherwise solvent-exposed environment. In four other HTH proteins shown in Figure 4, the thymine residue of a T. A bp at this position is involved in hydrophobic interactions with residue 1 (Leu in LacI: Kaptein et al., 1990), residue 12 (Gln in 434 cI,1 c1: Pabo et il., 1990) or residue 2 (Glu in I c1: Jordan & Pabo, 1988). Thus, there is no common mode of interaction between different HTH proteins and this bp in various operators.

TRG

OQS OPT

,*t

Recognition Specijicities of Tet-Repressor Mutants

the

new

recognition specificity obtained with a combination of RT28 and operator variant 2T is interpreted as a van der Waals’ contact between the methyl groups of Thr28 and thymine in bp 2 T. A, because operator variant 2C or repressor mutant RS28 do

of GZn38 with 3 A . T of the operator

Gln38 is the first amino acid residue in the second a-helix. From the new specificities of the &A38 repressor for operator variants 3C, 3G and 3T, it was proposed that the side-chain of Gln38 contacts adenine in bp 3 with two H-bonds (Wissmann et al., 1991b). Therefore, we made mutants with each of the possible changes to the 19 amino acid residues at this position and tested their abilities to repress the wild-type operator (Fig. 3). Only when Gln was replaced by amino acid residues with smaller sidechains, was repression of wild-type operator detected. Even the Asn38 side-chain with the same functional groups as the wild-type Gln38, spaced only one methylene group closer to the a-carbon atom, leads to a severely reduced repression of the mutant QN38 for wild-type operator. The QS38 repressor binds the wild-type operator almost as tightly as does the wild-type repressor. &A38 and QG38 recognize wild-type operator less well, resulting in repression to 24% and 58 %, respectively. We also tested six mutant repressors for binding to mutant operators with changes of bp 2 and bp 3 (Table 3). None of these repressors binds any of the variants of bp 2, but every mutant shows a new binding specificity for operator variants at bp 3. Both QS38 and &A38 bind operator variants 3C, 3G and 3T better than does the wild-type repressor, while wild-type repressor recognizes wild-type operator best. &A38 and QS38 also bind the wildtype operator and, therefore, exhibit a general loss of specificity in binding to operator variants at bp 3. Such a loss of specificity was also reported for the CAP mutants RG180 and RA180 (Zhang et aZ., 1990) (same relative position 12 in Fig. 4). The binding effects of TetR mutant &A38 had already been described (Wissmann et al., 1991b). It has been proposed that the C5 methyl group of thymine interacts with the methyl group of Ala38. This hypothesis is supported by the result that QV38 is the second-best binding protein for the 3T operator (see Table 3). This is very remarkable, because QV38 does not bind to wild-type operator. When a

1264

R. Baumeister et al.

Cy methyl group of valine is changed to a hydroxyl group in QT38, repression is reduced. In 434 c1, mutation &A28 also resulted in a changed operator specificity from an A. T to a T. A bp at position 3 (Wharton et al., 1987). A thymine base at operator bp 3 is also contacted via its C5 methyl group by the hydrophobic side-chain of Ile79 in TrpR (Otwinowski et al., 1988) or by the hydrophobic part of the large side-chain of Tyr17 in LacI at the same relative positions (Boelens et al., 1987). The results presented here clearly suggest that Gln38 contacts bp 3, which is in good agreement with Gln contacts in other protein-DNA complexes. For example, in 434 Cro, 434 ~1, 1 c1 and I Cro (Mondragon et al., 1989a,b; Mondragon & Harrison, 1991; Anderson et al., 1987; Jordan & Pabo, 1988; Brennan et al., 1990), the residue Gln in the N terminus of the respective recognition helices (see Fig. 4) contacts the adenine residue by two H-bonds. CAP has Arg at the same position and the operator has guanine, which is contacted by two H-bonds (Schultz et al., 1991). The new specificity of the Tet-repressor mutant QR38 for operator 3G may be explained by an analogous contact of Arg38. Although a repression to 73% indicates a poor affinity, this is nevertheless higher than that of wildtype repressor (94% /I-galactosidase activity), illustrating a change of binding specificity from Gln with adenine to Arg with guanine. The new specificity of QN38 for operator 3C (33% P-galactosidase activity, versus 75% obtained with wild-type repressor) may also be interpreted as an alternative contact of Asn38 with the bp 3 C*G. (g) Contacts of Pro39 with base-pairs 4T.Aand5A.T The second amino acid residue of recognition helices from six HTH proteins (434 ~1, 434 Cro, 1 Cro, LacI, TrpR, CAP) makes sequence-specific contacts with two base-pairs following that contacted by the first residue. The loss-of-contact analysis had revealed contacts of only Pro in TetR to bp 4 of the tet operator (Wissmann et al., 1991b). Here, we show that replacements of Pro39 by Ser, Ala, Gly or Thr lead to TetR mutants with binding activity for wild-type operator. The ability of nine mutant repressors to bind operators with mutations of bp 4 or 5 was also tested (see Table 3). PA39, PG39 and PS39 have new specificities for variant operators with changes of both bp 4 and bp 5. The highest repression is obtained for PS39 with operator 4C (14%), which is repressed to only 61 y. by wild-type TetR. PS39 also recognizes operator 5G, which is not bound by wild-type repressor. Furthermore, PS39 is the only mutant with residual binding to operator 5C. This operator variant contains a GATC sequence, which is subject to dam methylation of both adenine residues in bp 3 and 4 at N6 and prevents wild-type TetR binding (Sizemore et al., 1999). These results suggest that bp 4 and 5 are contacted by Pro39.

The amino acid residue Gln is located at equivalent positions in 434 Cro, 434 c1 repressor and Lac repressor. In all three repressors this residue recognizes neighboring C. G and A. T base-pairs. In the cocrystals of the 434 complexes, the Gln side-chain curls in a proline-like fashion around the surface of the a-helix backbone towards its own m-carbon atom (Mondragon et al., 1989a,b; Mondragon & Harrison, 1991; Anderson et al., 1987; Aggarwal et al., 1988). This structural property might be suggested for the Gln side-chains at this position in other repressors and would explain a functional similarity to Pro in Tet repressor. The P&39 mutant in Tet repressor has completely lost the affinity for wild-type operator with T. A and A. T pairs at positions 4 and 5. It is a striking result that the only operator bound by P&39 is 4C, which places neighboring C. G and A. T base-pairs near Q39. We therefore suggest that the Gln side-chain in P&39 has the same conformation and interacts with both operator base-pairs in a manner similar to that described for Gln29 in 434 Cro and 434 c1. This hypothesis agrees very well with the location of Pro39 in front of bp 4 and 5 as deduced from the contacts of Arg28 with bp 2 and Gin38 with bp 3, which were described above. We propose that the aliphatic ring of Pro39 is inserted between two methyl groups of thymine bp 4 and 5. A model of this interaction is shown in Figure 5 and illustrates the snug tit of the Pro side-chain in between the methyl groups. It is consistent with this model that operator variants 4A and 5T have the highest affinities for the wild-type Tet repressor. In both operator variants, the two thymine residues would still be located close to Pro. However, a hydrophobic side-chain is not the only prerequisite for amino acid residue 39. The mutation of Pro39 to any of the three aliphatic residues Ile, Leu and Val does not result in repressor mutants with binding activity at operator variants of position 4 and 5. The recognition helices of 1 Cro and 1 cl both contain serine at position 2 of the recognition helix (Fig. 4). Ser28 in A Cro is predicted to bind to adenine in bp 4 T-A (Ohlendorf et al., 1982; Benson & Youderian, 1989) and contacts guanine in operator bp 5 C.G (Hochschild & Ptashne, 1986). In 1 cl, it is involved in H-bonding to guanine in operator bp 5 C. G and is in van der Waals’ contact with the thymine residues in bp 6 A. T and 4 T. A in the consensus half-site (Jordan & Pabo, 1988; Sauer et al., 1990). According to our model, the Ser39 residue in TetR PS39 would be able to make H-bonds to guanine by 5 C. G in a manner similar to that found for A c1 and Cro. The first published example of a repressor mutant with an altered operator specificity involves a His to Pro mutation in the anti-parallel p-sheet motif of P22 Mnt repressor (Youderian et al., 1983). This mutant repressor binds an operator with a G. C to A-T exchange lOOO-fold better than it does to the wild-type. It was suggested that Pro might contact a hydrophobic pocket formed by four methyl groups in the major groove of DNA. Our model of t,he

DNA

Recognition SpecQficities of Tet-Repressor Mutants

1265

of Pro39 with operator bp 4 T. A and 5 A .T. Pro39 (yellow) is located in the F :igure 5. Model of the interaction fore :ground. The operator haves are painted in cyan and are located in the background. The van der Waals’ radii of the Pro side-chain and the C5 methyl groups of thymine residues are dotted.

interaction with this

between suggestion.

Pro and DNA

is in agreement

(h) Contact of Thr40 with operator bp 6 G. C Thr40 is the third amino acid residue in the recognition helix. It has previously been shown that the TA40 mutation leads to a change in specificity from G. C to A. T at position 6 of tet operator (Altschmied et al., 1988; Wissmann et al., 1991b). The nature of the interaction of Thr40 with bp 6 G. C and Ala40 with bp 6 A. T has been unclear. We show here that Thr can be substituted by Ser without an apparent loss of wild-type operatorbinding activity in the “high” expression system. Substitutions by Ala and Gly lead to pronounced losses of activity. Mutant TC40 is only partially functional and the other possible replacements are at the lower limit of detectable binding. TS40 and TA40 show new specificities for operator 6A, which is also bound efficiently by TC40, TN40 and wildtype repressor. The comparison of quantitative binding with these combinations yielded the best resolution in a mixed in vivo system containing pWH510 and pWHlOl2 derivatives (see Fig. 2). The results are displayed in Table 1 and indicate that removal of the Cy methyl group of Thr in TS40 has a negative effect on binding to wild-type operator that is not detected in the “high” expression system. TA40 recognizes operator 6A better than does TS40 and both bind operator 6A better than they bind wild-type repressor. Thus, both the hydroxyl and methyl groups in the Thr40 side-chain seem to be involved in binding to wild-

type and mutant operators. The importance of the hydroxyl group is clearly seen by the effect of mutant TV40, where an exchange of Thr40 to Val results in a dramatic loss of wild-type (86 y. B-galactosidase activity) and 6A operator binding (74% fi-galactosidase activity) compared to binding by wild-type repressor (1 and 2%, respectively). In the proposed structure of the Tet repressoroperator complex (Fig. 6), Thr40 would be located very close to bp 6 G. C of the wild-type operator. The hydroxyl group of Thr40 may be involved in H-bonds to either the protein backbone (e.g. the peptide CO-bond of Glu37) or to the phosphate backbone in the opposite DNA strand, while we propose a van der Waals’ contact between the CY methyl group of Thr and C5 atom of cytosine. With operator variant 6A, this methyl group is in steric conflict with the methyl group of thymine in bp 6 A. T. The steric hindrance would be reduced in the case of TS40, but it would be partially compensated by the ability of Ser to form H-bonds. An Ala sidechain would be located within a favorable van der Waals’ distance from the C5 methyl group of thymine, explaining the new specificity of mutant TA40 (see Table 1). A similar contact is found in several other complexes. Five out of eight operators shown in Figure 4 contain A. T base-pairs at position 6. In 1 c1, 434 Cro, 434 c1 and CAP the methyl groups of the thymine residues are in van der Waals’ proximity with amino acid residues at relative positions 13 and 14 of the respective HTH motif and form a compact hydrophobic segment. The tet operator 6C lacks a protruding methyl group and is most efficiently recognized by wild-

1266

R. Baumeister et al.

type repressor. We have no explanation why all other mutants we tested showed only low or no affinity for this operator. Perhaps as yet unidentified amino acid residues in the DNA-binding domain of TetR contribute to the recognition of bp 6. (i) Roles of Tyr42, Trp43 and His44 Tyr42, Trp43 and His44 are located at positions 5, 6 and 7, respectively, in the recognition helix. Ala substitutions of these residues resulted in the largest decrease of binding to wild-type operator among all residues in the HTH motif (Wissmann et al., 199lb). However, no base-pair-specific effects have been observed for these residues. This result is supported by the randomization of each residue leading to repressors with only small or no sequence-specific binding effects. We therefore suggest that these residues are involved in indirect readout of the operator sequence by the repressor. The smaller effects of new binding specificities at these positions will be described elsewhere (our unpublished results). The aromatic amphothere Tyr42 can be replaced best by Phe, which lacks only the hydroxyl group. YM42 is the only mutant with at least partial operator-binding activity, while Y142 shows only residual activity (see Fig. 3). This may indicate that Tyr42 prefers a hydrophobic environment in the protein-DNA complex. Met, with a flexible sidechain, seems to be the best aliphatic residue for filling the space of the Tyr side-chain. The hydroxyl function may contribute only slightly to the complex, since the YF42 mutant binds operator almost as well as does wild-type. The small number of functional substitutions at this position is reminiscent of the result obtained for randomization of Leu41 (Baumeister et al., 1992), which is thought to be involved in formation of the hydrophobic layer of the HTH motif. The helical wheel projection in Figure l(b) indicates that Tyr42 may also be partially involved in the formation of this layer. Trp43 has been used previously as a fluorescent probe to study Tet repressor-operator binding. The results suggested a contact of Trp43 with the DNA (Hansen $ Hillen, 1987). The aromatic amino acid residues Tyr and Phe and the imidazole ring system of His yield functional replacement mutants, while the complete loss of the side-chain in Gly leads to a largely reduced activity. An aromatic side-chain or a partially aromatic imidazole group is obviously needed at this position. In addition, the hydroxyl function of Tyr results in much better binding compared to that obtained with Phe in this position. In several other HTH motifs, the residue at the relative position 17 contributes to the interaction at bp 6. For example, in A c1 and in 434 Cro an Ala and a Leu, respectively, are in van der Waals’ distance to the C5 methyl group of thymine in bp 6 A. T, in LacI, two H-bonds of Arg22 with guanine in bp 6 C. G have been predicted (Kisters-Woike et al.,

1991), and in CAP, Argl85 contributes to the recognition of thymine in bp 6 A * T with one H-bond (Schultz et al., 1991). The respective Trp residue in Tet repressor is located very close to the edges of bp 5 and 6 in the model in Figure 6. This would agree with results of fluorescence quenching (Hansen & Hillen, 1987; Chabbert et al., 1992) indicating that the Trp residue could be involved in stacking interactions with the bases or base-pairs. This assumption is supported by the His, Tyr and Phe replacement mutants, since these are exactly the amino acid residues for which stacking interactions are possible (Kumar & Govil, 1984; Helene et al., 1982). The imidazole ring of His44 can be replaced by the large, aromatic side-chains of Trp and Tyr, yielding highly active mutants, and by the charged side-chains of Arg and Lys yielding less active mutants. Thus, the nature of the His44 contact to operator is not clear. 4. conclusions The HTH motifs of several repressors can be superimposed with less than I.0 A (1 A = @I nm) discrepancy (Brennan & Matthews, 1989; Steitz, 1990). We have shown, using a genetic approach, that the TetR HTH motif has a very similar structure (Baumeister et al., 1992). These results encouraged us to use the computer graphics to analyze the sterical properties of each interaction proposed above. Our model of the Tet-repressor HTH motif was built using the peptide co-ordinates of 434 Cro as described (Baumeister et al., 1992). The tet operator was assumed to be in the B-DNA structure. Repressor and operator were then docked manually. The recognition helix was positioned almost perpendicular to the longitudinal axis of the DNA in the major groove and is held in place by the first helix spanning the major groove and contacting the backbone with its amino-terminal end. This is in arrangement with the location of all HTH motifs on DNA. The contacts define a reverse orientation of the HTH motif with respect to the operator center as compared with other non-inducible repressoroperator systems. The same orientation has been proposed for the Lac-repressor-lac-operator complex (Boelens et al., 1987). The resulting structure of the tet-operator-HTH complex is displayed in Figure 6(a). Sketches of the proposed interactions are depicted in Figure 6(b). This Figure also shows the location of phosphate contacts as determined from ethylation interference studies (Heuer & Hillen, 1988). They are located at those positions in the model, where the HTH motif is close to the DNA phosphate backbone. It is important to mention that all chemical interactions described in the Results and Discussion section are sterically possible in this structure. In fact, energy minimization calculations of the protein-DNA contacts do not lead to large changes of the model depicted in Figure 6. Slight changes occur in the DNA structure. The two H-bonds predicted for the interaction

DNA

Recognition

of Tet-Repressor Mutants

Specijcities

1267

(a)

(b)

Figure 6. Model of the Tet repressor-operator complex. (a) Red: protein backbone of the HTH motif and side-chains of Thr2’7, Tyr42, Trp43 and His44; yellow: side-chains of Arg28, Gln38, Pro39 and Thr40 (from top to bottom); magenta: DNA of 1 operator half-site with bp 1 at the top and bp 7 at the bottom of the Figure; cyan: nucleotides of bp 2 to 6. (b) Schematic presentation of the interactions. Relevant amino acid side-chains are drawn as stick models and are numbered. Relevant methyl groups are depicted as spheres. The positions of strong and weak ethylation interference sites at the phosphates (Heuer BE Hillen, 1988) are given by filled and open circles, respectively. The methylation protection of guanine in bp 2 in the complex (Hillen et al., 1984) is represented by an asterisk. Suggested H-bonds are given by broken lines and hydrophobic interactions are indicated by dotted clouds.

1268

R. Baumeister

between Arg28 and adenine of operator bp 2 A. T are somewhat distorted. A much better fit is obtained when unwinding is included in the DNA structure. It has indeed been concluded from topoisomerase I studies that tet-operator DNA is unwound by about 20” in the complex (M. Wagenhijfer & W. Hillen, unpublished results). This indicates that structural modeling on the basis of in vivo binding studies is a powerful method to elucidate the chemical nature of protein-DNA interactions. The co-operative fit of each single interaction in this model adds significantly to the credibility of the structural proposals. It has been argued that Tet repressor may not be related to the common HTH family because it contains Pro at the second position of the recognition helix, which could lead to a different HTH structure, possibly similar to that proposed for the LexA repressor (Lamerichs et al., 1989). We have, therefore, tested a model of the repressor-operator contacts using the LexA backbone co-ordinates (M. Schnarr, personal communication) with the TetR side-chains, but were not able to explain our results with this structure. This supports the structural similarity of the HTH motif in TetR to those of other proteins (except LexA), which has been suggested previously by the functional analysis of mutations in the hydrophobic core of the TetR HTH motif (Baumeister et al., 1992). We thank M. Peschke for excellent technical assistance, Dr M. Schnarr for providing nuclear magnetic resonance data of LexA repressors and Drs W. Saenger and W. Hinrichs for their help with the computer modeling. This work was supported by the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie. V. Helbl and R. Baumeister were supported by a predoctoral fellowship from the FAU Erlangen-Niirnberg.

References Aggarwal, A. K., Rodgers, D. W., Drottar, M., Ptashne, M. & Harrison, S. C. (1988). Recognition of a DNA operator by the repressor of phage 434; a view at high resolution. Science, 242, 899-907. Altschmied, L., Baumeister, R., Pfleiderer, K. & Hillen, W. (1988). A threonine to alanine exchange at position 40 of Tet repressor alters the recognition of the sixth base pair of tet operator from GC to AT. EMBO J. 7,4011-4017. Anderson, J. E., Ptashne, M. & Harrison, S. C. (1987). Structure of the repressor-operator complex of bacteriophage 434. Nature (London), 326, 846-852. Aukerman, M. J., Schmidt, R. J., Burr, B. & Burr, F. A. (1991). An arginine to lysine substitution in the bZIP domain of an opaque-2 mutant in maize abolishes specific DNA binding. Genes Develop. 5, 310-320. Baumeister, R., Miiller, G., Hecht, B. & Hillen, W. (1992). Functional roles of amino acid residues involved in forming the a-helix-turn-a-helix operator DNA binding motif of Tet repressor from TnZO. Proteinx Struct. Fun&. Genet. In the press. Benson, N. & Youderian, P. (1989). Phage 1 Cro protein and c1 repressor use two different patterns of specific protein-DNA interactions to achieve sequence specificity in vivo. Genetics, 121, 5-12.

et al. Bernstein, F. C., Koetzle, T. F., Williams, G. J. B., Meyer, E. F., Jr, Brice, M. D., Rodgers, J. R., Kennard, O., Shimanouchi, T. & Tasumi, T. (1977). The protein data bank: a computer-based archival file for macromolecular structures. J. Mol. Biol. 112, 535-542. Bertrand, K. R., Postle, K., Wray, L. V., Jr & Reznikoff, W. S. (1984). Construction of a single-copy promoter vector and its use in analysis of regulation of the transposon TnlO tetracycline resistance determinant. J. Bacterial. 158, 910-919. Boelens, R., Scheek, R. M., van Boom, J. H. & Kaptein, R. (1987). Complex of lac repressor headpiece with a 14, base-pair lac operator fragment studied by twodimensional nuclear magnetic resonance. J Mol. Biol. 193, 213-216. Brennan, R. G. & Matthews, B. W. (1989). The helixturn-helix DNA binding motif. J. Biol. Chem. 264, 1903-1906. Brennan, R. G., Roderick, S. L., Takeda, Y. & Matthews, B. W. (1990). Protein-DNA conformational changes in the crystal structure of a lambda-Cro-operator complex. Proc. Nut. Acad. Sci., U.S.A. 87, 81658169. Calnan, B. J., Tidor, B., Biancalana, S., Hudson, 1). & Frankel, A. D. (1991). Arginine-mediated RNA recognition. Science, 252, 1167-l 17 1. Chabbert, M., Hillen, W., Hansen, D., Takahashi, M. & Bousquet, J.-A. (1992). Structural analysis of the operator binding domain of TnlO encoded Tet repressor: a time-resolved fluorescence and anisotropy study. Biochemistry, 31, 1951-1960. de Vlieg, J., Berendsen, H. J. C. & van Gunsteren, W. F. (1989). An NMR-based molecular dynamics simulation of the interaction of the lac repressor headpiece and its operator in aqueous solution. Proteins: Struct. Funct. Genet. 6, 1044127. Hansen, D. & Hillen, W. (1987). Tryptophan in a-helix 3 of Tet repressor forms a sequence-specific contact with tet operator in solution. J. Riol. Chem. 262, 12269912274. Hansen, D., Altschmied, L. & Hillen, W. (1987). Engineered Tet repressor mutants with single tryptophan residues as fluorescent probes. Solvent accessibilities of DNA and inducer binding sites and interaction with tetracycline. J. Biol. C’hem. 262, 14030-14035. Harrison, S. C. (1991). A structural taxonomy of DNA-binding domains. Nature (London), 353, 7 15719. Harrison, S. C. & Aggarwal, A. K. (1990). DNA recognition by proteins with the helix-turn-helix motif. Annu. Rev. Biochem. 59, 933-969. Helene, C. C Lancelot, G. (1982). Interaction between functional groups in protein-nucleic acid associations. Prog. Biophys. Mol. Biol. 39, l-68. Heuer, C. & Hillen, W. (1988). Tet repressor-tet operator contacts probed by operator DNA-modification interference studies. J. Mol. Biol. 202, 407-415. Hillen, W., Schollmeier, K. & Gatz, C. (1984). Control of expression of the T&O-encoded tetracycline resistance operon. II. Interaction of RNA polymerase and Tet repressor with the tet operon regulatory region. J. Mol. Biol. 172, 185-201. Hochschild, A. & Ptashne, M. (1986). Homologous interactions of A repressor and 1 Cro with the 1 operator. Cell, 44, 925-933. Isackson, P. J. & Bertrand, K. P. (1985). Dominant negative mutations in the TnlO Tet repressor: evidence for use of the conserved helix-turn-helix

DNA

Recognition

Specificities

motif in DNA binding. Proc. Nat. Acad. Sci., U.S.A. 82, 6226-6230. IUPAC-IUB Commission on Biochemical Nomenclature 1969. (1970). Abbreviations and symbols for the description of the conformation of polypeptide chains. Biochemistry, 9, 3471-3479. Jordan, S. R. & Pabo. C. 0. (1988). Structure of the lambda complex at 25 A resolution: details of the repressor-operator interactions. Science, 242, 893% 899. Kaptein, R., Zeiderweg, E. R. P., Scheek, R. M., Boelens, R. & van Gunsteren. W. F. (1985). A protein structure from nuclear magnetic resonance data: lac repressor headpiece. J. Mol. Biol. 182, 179-182. Kaptein, R., Boelens, R. & Lamerichs, R. M. J. N. (1990). NMR studies of protein-DNA recognition. The interaction of lac repressor headpiece with operator DNA. In Nucleic Acids and Molecular Biology (F. Eckstein & I). M. !J. Lilley, eds), vol. 4, pp. 35-59, Springer, Berlin. Kisters-Woike, B.. Lehming. N., Sartorius, .J.. von Wilcken-Bergmann. B. & Miiller-Hill, B. (1991). A model of the repressor-operator complex based on physical and genetic data. Eur. J. Biochem. 198.

411-419. I,. 0. & Miller,
of Tet-Repressor

219, 321-334. A..

Wolberger,

C. & Harrison,

S. C. (1989a).

1269

Structure of phage 434 Cro protein at 2.35 A resolution. J. Mol. Biol. 205, 179-188. Mondragon, A., Subbiah, S., Almo, S. C.. Drottar. M. & Harrison, S. C. (1989b). Structure of the aminoterminal domain of phage 434 repressor at 2.0 A resolution. J. Mol. Biol. 205, 189-200. Ohlendorf, D. H., Anderson, W. F., Fisher, F. G., Takeda, Y. & Matthews, B. W. (1982). The molecular basis of DNA-protein recognition inferred from the structure of cro repressor. Nature (London), 298, 718-723. Otwinowski, Z., Schevitz, R. W., Zhang. R.-c:., Lawson, C. L., Joachimiak, A., Marmorstein, R. Q., Luisi, B. F. & Sigler, P. B. (1988). Cryst’al structure of trp repressor/operator complex at atomic resolution. Nature (London), 335, 321-329. Pabo, C. 0. & Sauer, R. T. (1984). Protein--DNA recognition. Annu. Rev. Biochem. 53, 293-321. Pabo, C. O., Aggarwal, A. K., ,Jordan, S. R., Beamer, L. J., Obeysekare, U. R. & Harrison. S. (‘. (1990). Conserved residues make similar contacts in two repressor-operator complexes. Science, 247. 12 lop 1213. Postle, K., Nguyen, T. T. & Bertrand. K. P. (1984). Nucleotide sequence of the repressor gene of the TnlO tetracycline resistance determinant. nTuc1. Acids Res.

12, 4849-4863.

Kleina,

Mondragon,

Mutants

Sartorius, J., Lehming, N., Kisters, B., von Wilcken-Bergmann, B. 5, Miiller-Hill. B. (1989). Lac repressor mutants with double or triple exchanges in the recognition helix bind specifically to lac operator variants with multiple exchanges. EMBO J. 8, 12651270. Sartorius, J., Lehming, N., Kisters-Woike, B., von Wilcken-Bergmann, B. & Miller-Hill, B. (1991). The roles of residues 5 and 9 of the recognition helix of lac repressor in lac operator binding. .I. Mol. Biol. 218,

313-321. Sauer,

R. T., Jordan, S. R. & Pabo, (‘. 0. (1990). 1 Repressor: a model system for understanding protein-DNA interactions and protein stability. In Advances in Protein Chemistry (Anfinsen. C. B., Edsall, J. T., Richards, F. M. & Eisenberg, D. S., eds), pp. 1-61, Academic Pres Inc., San Diego. Schultz, S. C., Shields, G. C. & Steitz, T. A. (1991). Crystal structure of a CAP-DNA complex: the I)NA is bent by 90”. Science, 253, 1001-1007. Sizemore, C., Wissmann, A., Giilland, U. 6 Hillen, W. (1990). Quantitative analysis of TnlO Tet repressor binding to a complete set of tet operator mutants. Nucl. Acids Res. 18, 2875-2880. Staacke, D., Walter, B., Kister-Woike, B., von Wilcken-Bergmann, B. & Miiller-Hill, B. (1990). How Trp repressor binds to its operator. EMBO J. 9, 1963-1967. Steitz, T. A. (1990). Structural studies of protein nucleic acid interaction: the sources of sequence-specific binding. Quart. Rev. Biophys. 23, 205-280. SU, T.-Z. & El-Gewely, M. R. (1988). A multisite-directed mutagenesis using T7 DNA polymerase: application for reconstruction of a mammalian gene. Gene, 69, 81-89. Wharton. R. P. & Ptashne, M. (1987). A new-specificity mutant of 434 repressor that defines an amino acidbase pair contact. Nature (London), 326, 888-891. Wissmann, A., Meier, I. & Hillen, W. (1988). Saturation mutagenesis of the TnlO-encoded tet operator 0,. Identification of base pairs involved in Tet repressor recognition. J. Mol. Biol. 202, 397-406. Wissmann, A., Wray, L. V.,
1270

R. Baumeister

Baumeister, R., Geissendiirfer, M. & Hillen, W. (1991a). Selection for TnlO Tet repressor binding to tet operator in Escherichia coli: isolation of temperature-sensitive mutants and combinatorial mutagenesis in the DNA binding motif. Genetics, 128, 225-232. Wissmann, A., Baumeister, R., Miiller, G., Hecht, B., Helbl, V., Pfleiderer, K. BEHillen, W. (19916). Amino acids determining operator binding specificity in the helix-turn-helix motif of TnlO Tet repressor. EMBO J. 10. 4145-4152. Edited

et al. Youderian, P., Vershon, A., Bouvier, S., Sauer, R. T. 8: Susskind, M. M. (1983). Changing the DNA-binding specificity of a repressor. Cell, 35, 777-783. Zhang, X. $ Ebright, R. H. (1990). Identification of a contact between arginine-180 of the catabolite gene activator protein (CAP) and base pair 5 of the DNA site in the CAP-DNA complex. Proc. Nat. Acad. Sci., U.S.A. 87, 4717-4721.

by R. Schleif