Solvent-exposed Residues in the Tet repressor (TetR) Four-helix Bundle Contribute to Subunit Recognition and Dimer Stability*

Dimerization specificity of Tet repressor (TetR) can be altered by changes in the core of the four-helix bundle that mediates protein-protein recognition. We demonstrate here that the affinity of subunit interaction depends also on the solvent-exposed residues at positions 128 and 179′–184′, which interact across the dimerization surface. TetR(B) and (D), two naturally occurring sequence variants, differ at position 128 with respect to the monomer-monomer distances in the crystal structures and the charge of the amino acids, being glutamate in TetR(B) and arginine in TetR(D). In vivoanalysis of chimeric TetR(B/D) variants revealed that the single E128R exchange does not alter the dimerization specificity of TetR(B) to the one of TetR(D). When combined with specificity mutations in α10, it is, however, able to increase dimerization efficiency of the TetR(B/D) chimera with TetR(D). A loss of contact analysis revealed a positive interaction between Arg-128 and residues located at positions 179′–184′ of the second monomer. We constructed a hyperstable TetR(B) variant by replacing residues 128 and 179–184 by the respective TetR(D) sequence. These results establish that in addition to a region in the hydrophobic core residues at the solvent-exposed periphery of the dimerization surface participate in protein-protein recognition in the TetR four-helix bundle.

Protein-protein complexes play crucial roles in many biological processes such as gene expression, replication, DNA repair, signal transduction, enzyme regulation, and immune response (1)(2)(3)(4)(5)(6)(7)(8)(9). The rules determining specificity in protein-protein recognition are therefore of fundamental biological importance. Among the structural motifs mediating protein-protein recognition, assemblies of ␣-helices are probably most widespread (10). Oligomerization of transcription factors, e.g. is often accomplished by ␣-helical coiled coils (11), and four-helix bundles form dimerization surfaces in many proteins (12). The mechanistic details governing recognition specificity and binding affinity of four-helix bundle mediated protein oligomerization are only poorly understood. We investigated subunit recognition in the four-helix bundle of Tet repressor (TetR) 1 dimers with the goal of determining generally applicable rules for this process.
TetR sequence variants regulate tetracycline-induced ex-pression of seven naturally occurring tetracycline-resistance determinants (classes A-E, G, and H; for a review, see Ref. 13).
In the absence of tetracycline, dimeric TetR binds to tet operator (tetO) repressing transcription of the tet promoters. Nanomolar concentrations of tetracycline lead to dissociation of TetR from tetO to induce transcription. This transcriptional switch is exceptionally sensitive to low inducer concentrations, making it the system of choice for regulation of gene expression in many higher organisms, including plants, transgenic mice, and human cells (for two recent reviews, see Refs. 14 and 15). Crystal structures of TetR(D) in complex with [Mg-tetracycline] ϩ revealed a small N-terminal and a large C-terminal domain (16,17). The latter mediates tetracycline binding and dimerization via a four-helix bundle formed by the helices ␣8 and ␣10 of each subunit. Disruption of the dimeric structure of TetR by urea concomitantly leads to denaturation of the subunits (18).
We have established that the sequence variants TetR(B) and TetR(D) do not form heterodimers because amino acids located in the core of the four-helix bundle in helix ␣10 of TetR(B) and TetR(D) lack structural complementarity (19). Adjusting all amino acids in ␣10 yielded a change in subunit recognition specificity but did not restore the full dimerization efficiency of the wild type. We demonstrate here that this requires only one additional exchange, E128R, located in helix ␣8 at the edge of the four-helix bundle. Furthermore, we construct a hyperstable TetR(B) variant by improving the interaction of Arg-128 with residues of the second monomer. Thus, interactions between partially solvent-exposed, hydrophilic residues can profoundly influence the specificity of subunit recognition in protein hetero-oligomers and improve the stability of homo-oligomers.

EXPERIMENTAL PROCEDURES
Materials and General Methods-Chemicals were obtained from Merck (Darmstadt), Serva (Heidelberg), Sigma (Mü nchen), or Roth (Karlsruhe) and were of the highest purity available. Tetracycline was purchased from Fluka (Buchs). Enzymes for DNA restriction and modification were obtained from Boehringer Mannheim, Life Technologies, Inc. (Eggenstein), New England Biolabs (Schwalbach), or Pharmacia (Freiburg). Sequencing was carried out according to the protocol provided by Perkin Elmer for cycle sequencing.
Construction of tetR Variants-The mutation E128R was introduced into tetR(B) by polymerase chain reaction according to a three-primer method (24). The conditions of the polymerase chain reactions were as described (25). The products of the second polymerase chain reaction * This work was supported by the Deutsche Forschungsgemeinschaft through SFB 473 and the Fonds der Chemischen Industrie. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed.  [179][180][181][182][183][184]. The corresponding pWH853 derivatives were digested with XbaI and NcoI, and the tetR fragments were purified and cloned into likewise digested pWH1950. DNA of positive candidates was analyzed by digestion with restriction enzymes and sequencing of tetR.
␤-Galactosidase Assays-Repression and induction efficiencies of the TetR variants as well as the negative transdominance efficiencies of tetR(B)⌬26-53 and tetR(D)⌬26-53 were assayed in E. coli WH207(tet50). The phage tet50 contains a tetA-lacZ transcriptional fusion integrated as single copy into the WH207 genome (22). Bacteria were grown in LB supplemented with the appropriate antibiotics. Quantification of induction efficiencies was done with 0.2 g/ml tetracycline in overnight and log-phase cultures. ␤-Galactosidase activities were determined as described by Miller (26). Three independent cultures were assayed for each strain, and measurements were repeated at least twice.
Purification of TetR Variants-pWH1950 or pWH1950 derivatives were transformed into E. coli RB791. Cells were grown in 3 liters of LB at 28°C in shaking flasks. Tet repressors were overexpressed by adding isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 1 mM at an A 600 of 0.7-1.0. Cells were pelleted, resuspended in buffer A (50 mM NaCl, 2 mM dithiothreitol, and 20 mM Na 3 PO 4 , pH 6.8), and broken by sonication, and TetR was purified by cation exchange chromatography and gel filtration as described (23).
Determination of the in Vitro Stability of TetR Variants-Circular dichroism (CD) measurements were performed on a Jasco J715 spectropolarimeter at protein concentrations of 5 M in 0.5-cm cells. All measurements were carried out at a temperature of 22°C. Equilibrium denaturation was performed by incubating protein samples overnight at the indicated urea concentration. We used F-buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol) for all spectroscopic measurements. Urea was obtained from ICN Biochemicals (Eschwege), and urea solutions were prepared every day. Renaturation was achieved by incubating the samples overnight at 8 M urea and then diluting them 200-fold with F-buffer. Thermodynamic calculations were done as described before (18).

Tet(B/D) Repressors-Using chimeric TetR(B/D) repressors,
we demonstrated previously that residues in helix ␣10 are critical for the different dimerization specificities of TetR(B) and TetR(D); however, further changes are required to achieve WT affinity of subunit binding (19). The monomer-monomer distances in the TetR(D)/[(Mg-tetracycline) ϩ ] 2 crystal structure and a structural model of TetR(B) are different at residues in ␣10 and at position 128 in ␣8 (19). Thus, we constructed five new TetR(B/D) repressor dimers to analyze the influence of mutations at position 128 on dimer formation. The sequences of these five dimeric TetR(B/D) repressors and those from the previous study used here again are depicted in Fig. 1 In Vivo Repression and Induction Efficiencies-The tetR variants were transformed into E. coli WH207tet50 to test in vivo operator binding activity and inducibility. pWH853 derivatives, which constitutively produce low levels of TetR (22), were used as expression plasmids. Repression of the tetA-lacZ transcriptional fusion and inducibility by tetracycline were quantified at 37°C. The results are shown in Table I Table I). We performed a loss of contact analysis by mutating Arg-128 of TetR(B/D)51-178 to A128 in TetR(B/D)51-127,129-178,A128 (Fig. 1). As a control, we also constructed TetR(B/D)51-127,129-178, which contains the TetR(B) amino acid Glu-128. Dimerization with TetR(D)⌬26-53 is neither detectable for TetR(B/ D)51-127,129-178,A128 nor for TetR(B/D)51-127,129-178. In contrast, the dimerization efficiency of TetR(B)⌬26-53 with these TetR variants is not affected (see Table I (Table I) (Table II). We conclude that denaturation under these conditions is reversible. The urea-induced unfolding of the TetR(B/D) chimera as followed by the change of the CD shows a monophasic, sigmoidal decrease (Fig. 2), and the midpoint of the unfolding transition depends on the protein concentration (data not shown), as expected for a bimolecular reaction. Thus, as for TetR(B) (18), unfolding of the TetR(B/D) chimera is quantitatively described by a two-state model, in which only folded dimers and unfolded monomers exist at equilibrium in significant concentrations. Calculated thermodynamic stabilities of the TetR(B/D) variants are also given in Table II Dimerization of TetR(B) or TetR(B/D)179-184 with TetR(D) is not affected by mutating Glu-128 to Arg-128. The mutation E128R alone is, therefore, not sufficient for dimerization of TetR(B) with TetR(D). This is in agreement with previous results, which identified the amino acids 188, 192, 193, and 197 as the main determinants of the TetR(B)/(D) dimerization specificity (19). In agreement with the conclusions discussed so far, the stability of TetR(B) is profoundly increased by introducing Arg-128 and the TetR(D) residues contacting Arg-128 across the dimerization surface. The stability of TetR(B) is not changed by replacing either residue at 128 or 179Ј-184Ј by the corresponding TetR(D) amino acids, but combining the mutation in TetR(B/D)128,179-184 increases ⌬G by 12 kJ mol Ϫ1 . Hydrogen bond formation between Arg-128 and Gln-184Ј most likely is a Tetracycline binding obtained after de-and renaturation. b The identical results were obtained by following fluorescence. responsible for the increased stability of TetR(B/D)128,179-184, but the nature of that interaction cannot be concluded from the current data. The negative transdominance presently cannot be quantitatively related to the dissociation constant or ⌬⌬G. Because the negative transdominance correlates with the amount of heterodimer formed in vivo (19) and the thermodynamic stability in vitro, it provides a valid qualitative measure of dimerization efficiencies.
Recognition between polypeptides has been investigated in several proteins. One of the best analyzed examples is the complex formed between the human growth hormone and its receptor. A saturating alanine-scanning mutagenesis of the hormone-receptor interface supported the so-called "hot spot concept" of protein-protein recognition (27), where only a few amino acids of a larger interface account for the affinity in protein complexes. These residues are mostly hydrophobic and are located in the center of the interaction surface, whereas the primarily hydrophilic residues at the periphery of the interaction surface are unimportant (7,27). The main determinants of the TetR(B)/(D) dimerization specificity are four residues of helix ␣10 that are located in the center of the four-helix bundle (19). Therefore, the hot spot concept seems also to be valid for protein recognition during TetR dimerization. However, an additional mechanism must influence protein-protein recognition in TetR because even the exchange of the complete ␣10 is not sufficient to restore WT dimer formation with TetR(D) in a Tet(B/D) repressor. The results presented here establish that the additional recognition present in TetR only depends on the partially solvent-exposed residues at position 128 and at 179 -184. The arginine residue at position 128 is 95% solvent-exposed (calculated using GRASP (31)) compared with a surface accessibility of 89, 83, and 79% for the arginine residues at positions 28, 62, and 87, respectively, which are not involved in dimerization. The leucine residue at position 193 in the center of the four-helix bundle, where it contributes to dimerization specificity (19), shows 0% solvent accessibility. Thus, the specificity of protein-protein recognition in the TetR four-helix bundle is mediated by a dual mechanism that is determined by the buried residues in the center of the dimerization surface and also by partially solvent-exposed residues located at the edge of the dimerization surface. Dimerization of leucine zipper peptides is profoundly influenced by the interaction of hydrophilic residues at the Glu and Gly positions, and mutations at these positions can be sufficient to change dimerization specificity (28,29). However, recent results demonstrated that the patterns of charged residues at the Glu and Gly positions are not sufficient to predict dimerization specificities of many leucine zipper mutants (30). These results support the view that the dual mechanism of protein recognition established for the TetR four-helix bundle might as well apply for other protein complexes.