Repertoire Selection of Variant Single-Chain Cro: Toward Directed

12038

Biochemistry 2004, 43, 12038-12047

Repertoire Selection of Variant Single-Chain Cro: Toward Directed DNA-Binding Specificity of Helix-Turn-Helix Proteins† Mikael T. I. Nilsson and Mikael Widersten* Department of Biochemistry, Uppsala UniVersity, Biomedical Center, Box 576, SE-751 23 Uppsala, Sweden ReceiVed April 30, 2004; ReVised Manuscript ReceiVed July 27, 2004

ABSTRACT: A single-chain derivative of the lambda Cro repressor (scCro) has been randomly mutated in amino acid residues critical for specific DNA recognition to create libraries of protein variants. Utilizing phage display-afforded affinity selection, scCro variants have been isolated for binding to synthetic DNA ligands. Isolated scCro variants were analyzed functionally, both in fusion with phage particles and after expression of the corresponding free proteins. The binding properties with regard to specificity and affinity in binding to different DNA ligands were investigated by inhibition studies and determination of equilibrium dissociation constants for formed complexes. Variant proteins with altered DNA-sequence specificity were identified, which favored binding of targeted synthetic DNA sequences over a consensus operator sequence, bound with high affinity by wild-type Cro. The specificities were relatively modest (2-3-fold, as calculated from KD values), which can be attributed to the inherent properties in the design of the selection system; one half-site of the synthetic DNA sequences maintains the consensus operator sequence, and one “subunit” of the variant single-chain Cro dimers was conserved as wild-type sequence. The anticipated interaction between the wild-type subunit and the consensus DNA half-site of target DNA ligands is, hence, expected to contribute to the overlap in sequence discrimination. The binding affinity for the synthetic DNA sequences, however, was improved 10-30-fold in selected variant proteins as compared to “wild-type” scCro.

DNA-binding proteins that utilize the helix-turn-helix (HTH)1 motif or variants thereof for DNA sequence recognition are found in organisms from all kingdoms as well as in bacteriophages and other viruses (1-4). Although there is substantial knowledge regarding structure/function relationships in DNA binding of several HTH proteins, the available data describe evolutionary snapshots of DNA-binding proteins and are examples of structural solutions to existing biological problems, for example, regulation of gene expression. Because maximum affinity and recognition specificity in DNA binding is not necessarily benficial to an organism, the evolutionary pressure toward such properties is expected to be weak. It is, hence, not possible to directly derive general structural criteria applicable to predictions of binding properties of a given HTH protein by extrapolating results gathered in studies of naturally occurring protein/DNA complexes. To address such general aspects of HTH protein-DNA recognition, we have pursued an approach of repertoire selection for novel HTH proteins with new and predefined DNA target specificities. These proteins are derivatives of a single-chain form of the lambda Cro repressor (scCro) (5). Large protein libraries of scCro variants of diverse DNAbinding properties have been constructed by randomized mutagenesis of amino acid residues critical for specific DNA † The work was financially supported by The Carl Trygger Foundation. * Corresponding author [fax +46 (18) 55 8431; telephone +46 (18) 471 4992; e-mail [email protected]]. 1 Abbreviations: EMSA, electrophoretic mobility shift assay; HTH, helix-turn-helix; scCro, single-chain Cro.

recognition (Figure 1a) (6). Affinity-based selection afforded by phage display (7, 8) toward predesigned synthetic DNA targets (Figure 1b) has enabled the isolation of scCro variants of distinct functional properties, which can be linked to corresponding primary structures. Previous studies on the DNA binding of zinc-finger proteins have demonstrated the feasibility of the approach (9) with regard to addressing fundamental issues of zinc-finger-DNA recognition as well as in the construction of novel zinc-finger proteins of predefined binding properties (10). The DNA molecules applied as target ligands in the present study have been designed to probe different aspects of scCro protein-DNA binding interactions: (i) To explore amino acid sequences favoring selection of a given variant, the nucleotide sequences of the ligands have been drastically altered from a known Cro high-affinity ligand, denoted ORC (11), while maintaining the overall nucleotide composition (Figure 1b). (ii) To investigate the importance of bendability of the target DNA during protein binding, two different ligands were synthesized (Figure 1b), which differ in their nucleotide compositions at positions (-4) and (-5) in the respective ORas sequences (Figure 1b). In ORas11, nucleotides (-4) and (-5) are guanosines. In ORas12 the corresponding nucleotides are substituted by inosines lacking the C-2 amino group (Figure 1b). Removal of the C-2 amino groups is expected to increase the tendency of conformational freedom due to a lower degree of interchain base pairing and steric hindrance between neighboring groups in the minor groove (12). These C-2 amino groups in ORas11 are not expected to interact with Cro directly because they are

10.1021/bi049122k CCC: $27.50 © 2004 American Chemical Society Published on Web 08/28/2004

Phage Display Selection of Novel HTH Proteins

Biochemistry, Vol. 43, No. 38, 2004 12039

FIGURE 1: Schematic representation of scCro gene libraries and DNA ligands used during phage display-afforded affinity selection or in electrophoretic mobility shift assays (EMSA). (a) scCro gene library constructs. Random mutagenesis of codons for residues making basespecific contacts in the wild-type Cro protein was performed in the 5′-proximal cro gene (cro-N). Three different libraries, denoted A, B, or C containing different numbers of randomly (NNS) substituted codons, were constructed in which the cro-N genes (red) were linked with 24 nucleotides to the cro-C gene (blue) encoding the wild-type Cro protein. The cro-C gene was further linked to geneIII of phage M13. The proteins expressed were hence covalently linked via short peptide linkers. (b) Nucleotide sequences of dsDNA ligands used during selection by phage display (“B” denotes biotin) or in EMSA. Nucleotides on yellow background indicate positions making specific contacts with wild-type Cro (15). Boxed nucleotides in ORas12 indicate the positions of guanosine (in ORas11) to inosine replacements in ORas12. Nucleotides in red indicate positions for labeling with [R-32P]-dATP by Klenow fill-in.

with different relative affinities. Earlier results have shown that it is indeed possible to isolate phage clones displaying scCro variant proteins of distinct primary structures with binding properties favoring recognition for the different ORas ligands (6). The present work describes the outcome of affinity selections of DNA-binding proteins from two new libraries of scCro variants. One of these new libraries was designed to test the significance of a protein sequence motif found in a majority of clones isolated after a previous selection toward ORas12 (6) (“B”-clones in Table 1). Selected protein variants have been analyzed for primary structure and characterized for their DNA-binding properties, targeting the structural determinants for their DNA sequence specificity. FIGURE 2: Image of the Cro/ORC complex. Specific protein-DNA base contacts are indicated by dotted lines (red, hydrogen bonds; blue, van der Waals interactions). Ser-28 interacts via its side-chain hydroxyl with O-6 and N-7 of G (-4) in ORC. The amino group at C-2 (purple within yellow circle) points toward the minor groove, away from the bound protein. Key: Protein R-helices are represented as red cylinders, loops in green, and side chains subjected to random mutagenesis in, carbon, yellow; oxygen, red; and nitrogen, blue. The strands of cocrystallized ORC DNA are shown in gold and cyan. Image produced in InsightII (Accelrys, San Diego, CA) from the atomic coordinates in 6CRO (15).

positioned within the minor groove, hence pointing away from the protein in the binding complex but presenting the same groups for hydrogen bonding and van der Waals contacts toward the major groove (Figure 2). Hence, if DNA bending is a component in sequence discrimination, different scCro variants are expected to bind the two ORas ligands

EXPERIMENTAL PROCEDURES Bacteria and Phages. Escherichia coli strains XL1-Blue (Stratagene, La Jolla, CA) and JM109(DE3) (Promega Corp., Madison, WI) were used as host cells during phage-display selection or in overexpression of scCro proteins, respectively. Phage M13K07 (KanR) was used as helper phage during phage-display expression. Annealing, Labeling, and Purification of Double-Stranded DNA Ligand. Oligonucleotide concentrations in water were determined by UV absorption at 260 nm using calculated extinction coefficients. For annealing of double-stranded ligands, oligonucleotides were diluted to 100 µM in water followed by mixing at 1:1 molar ratios in 10 mM Tris-HCl and 50 mM NaCl to final concentrations of 10 µM. Mixtures were heated to 96 °C for 10 min followed by cooling to room temperature over a period of 140 min. Double-stranded DNA was stored at -20 °C until used.

12040 Biochemistry, Vol. 43, No. 38, 2004 A. Ligands Used in Phage Display Selection. Biotinylated and unbiotinylated double-stranded ORC (ORC+, TGT ATC ACC GCG GGT GAT AGT with ORC-, ACT ATC ACC CGC GGT GAT ACA; ORC+B, biotin-GTG TGT GTA TCA CCG CGG GTG ATA GT and ORC-), ORas11 (ORas1+B, biotin-GTG TGT GAT ACC AAG CGG GTG ATA GT with ORas1-1, ACT ATC ACC CGC TTG GTA TCA), and ORas12 (ORas1+B with ORas1-2, ACT ATC ACC CGC TTI ITA TCA) DNA ligands were obtained by annealing as described above. B. Ligands for EMSA. Double-stranded DNA ligands (ORC, ORas11, ORas12) were prepared by mixing complementary oligonucleotides N-ORC+EMS, TTT TTA TCA CCG CGG GTG AT, with N-ORC-EMS, TTT TTA TCA CCC GCG GTG AT; 2-ORas1+EMS, TTT TGA TAC CAA GCG GGT GAT, with 2-ORas11-EMS, TTT TAT CAC CCG CTT GGT ATC; and 2-ORas1+EMS with 2-ORas12EMS, TTT TAT CAC CCG CTT IIT ATC, as described above. Protruding oligo-dT ends were filled in with [R-32P] dATP using Klenow fragment (Cloned, Exo-free, USB). Change of buffer and removal of unincorporated dATP were achieved by passing the labeled oligos through G-25 MicroSpin columns (Amersham Bioscience) equilibrated with 10 mM Tris-HCl, pH 7.4, 100 mM KCl, and 1 mM EDTA. Phage Display Library Construction. Library A: The tandem cro gene construct encoding scCro8 (5) was inserted into pC3∆NX (13). The 5′ cro gene was randomly mutated by PCR in codons 27-29 and 31-33 using primers cro8mut, TAA AGA TCT CGG CGT GTA TNN SNN SNN SAT CNN SNN SNN SAT CCA TGC CGG CCG AAA, and Cro8Lrev, GCC GCC AGA GCC ACC, as described previously (6). Library C: A gene library was constructed, as above, introducing random mutations in codons 27, 28, and 31. Codons 29 (TGC) and 32 (TAC) resulted in Ala-29 to Cys-29 and Lys-32 to Tyr-32 point mutations, respectively. Mutagenesis was performed by PCR as above with primers cro10mut, TAA AGA TCT CGG CGT GTA TNN SNN STG CAT CNN STA CGC CAT CCA TGC CGG CC, and Cro8Lrev. The phage display vector used in the construction of the C library was a derivative of pC3∆NX, in which the unique SpeI site flanking the scCro gene at the 3′-end had been replaced by an AatII site by silent mutagenesis (data not shown). Amplification and harvest of constructed phage display libraries of the different scCro constructs were performed as described earlier (6). For evaluation of library quality concerning sequence diversity, plasmid DNA from 24 transformants was isolated and the scCro gene regions were sequenced. Affinity Selection by Phage-Display Expression. Library A: Phages displaying scCro randomly mutated in residues 27-29 and 31-33 were challenged for binding to biotinylated ORas11 or ORas12 DNA under binding conditions as described previously (6) for six selection rounds. All incubations and washing steps during the affinity selection cycles were performed on ice. Library C: Two parallel selections of phage-displayed scCro C-variants under binding conditions, as above, were conducted at either 22 °C or on ice for four rounds, with the exception that counterselection by addition of unbiotinylated ORC was not included. Enrichment during selection was assessed from the ratio of phages eluted after a completed selection cycle divided by the number of phages allowed to bind to the biotinylated

Nilsson and Widersten ORas12 ligand. The scCro variant genes of picked isolates from rounds 2-6 (library A) or rounds 3 and 4 (library C) were analyzed by DNA sequencing. DNA Binding SelectiVity of Phage-Displayed ScCro Variants. Phage clones A12:415, C12:316, C12:401, C12:403, C12:405, C12:408, C12:414, and C12:404* were purified and amplified. Pure variant scCro8 phages (∼109 cfu) were subjected to a single round of affinity selection toward three different biotinylated DNA ligands, ORC, ORas11, and ORas12, as described previously (6). The relationship of the survival ratios obtained was used as a measure of binding specificity for the different DNA ligands tested. ScCro Variant Protein Expression and Purification. Genes encoding selected scCro variants were subcloned into vector pET∆B, a BglII site-deficient derivative of pET21a(+) (Novagen). Confirmation of successful subcloning was verified by DNA sequencing. The resulting plasmid constructs were denoted pETscCro. The expression plasmids were transformed into E. coli JM109 (DE3) for protein synthesis by electroporation. A. Analytical-Scale Expression. Selected clones of E. coli JM109 (DE3) harboring expression plasmids with scCro variants were inoculated in 2TY containing 100 µg of ampicillin/mL and grown overnight at 30 °C. The overnight cultures were diluted 1:50 into 25 mL of 2TY containing 50 µg of ampicillin/mL. Cultures were grown at 30 °C until mid-log phase was reached (OD600 of 0.7-1) when scCro overexpression was induced by the addition of 1 mM isopropyl β-D-thiogalactopyranoside. Incubation was continued for 3 h when cells were harvested by centrifugation. Bacteria were resuspended in 0.5 mL of 20 mM sodium phosphate, pH 7.4, containing protease inhibitor cocktail (Roche) and subsequently lysed by ultrasonication. Insoluble debris was removed by centrifugation at 16000g for 15 min. Finally, glycerol and 2-mercaptoethanol were added to final concentrations of 50% (v/v) and 12.5 mM, respectively. Total protein content of lysates was determined according to the method of Bradford using a commercial reagent (Bio-Rad, Richmond, CA). The relative concentration of scCro protein variants in the lysates was estimated by Western blotting of lysates using polyclonal rabbit anti-scCro antibodies (AgriSera, Va¨nna¨s, Sweden). Immunocomplexes were detected by enhanced chemiluminiscense. Final lysate preparations were stored at -20 °C until used in EMSA. B. PreparatiVe-Scale Expression and Purification. Growth and expression were performed as described for small-scale cultures with the exception that culture volumes were 1.5 L per expressed variant. The crude lysate was desalted by gel filtration through Sephadex G-25, equilibrated with 50 mM sodium-phosphate, pH 7.5. The G-25 eluate was subsequently cleared either by ultracentrifugation at 100000g for 1 h or by filtration through an 0.8/0.2 µm Acrodisc syringe filter (Pall, Ann Arbor, MI). The desalted lysate was loaded at 1 mL/min onto a 5 mL column prepacked with SP-HiTrap cation exchanger resin (Amersham Biosciences). The column, pre-equilibrated in 50 mM sodium-phosphate and 50 mM NaCl, pH 7.5, was washed in the same buffer at 1 mL/ min until baseline absorbance was reached. Bound protein was eluted with a linear gradient, 4 mM/mL, of NaCl from 50 mM to 0.7 M. Fractions containing scCro protein were pooled. Buffer was changed to 10 mM Tris-HCl, pH 7.4, and 100 mM KCl (EMSA binding buffer, “BB”) by passing

Phage Display Selection of Novel HTH Proteins through a PD10 column (Amersham Biosciences). Protein stocks for EMSA studies were subsequently diluted in BB containing 50% (w/v) glycerol and 2 mM dithiothreitol. Proteins were stored at -20 °C until analyzed. Protein concentration was determined by UV absorbance of the solutions using calculated molar extinction coefficients at 276 nm: scCro, 7800 M-1 cm-1; A12:407, C12:408, C12: 404*, 10150 M-1 cm-1. Electrophoretic Mobility Shift Assay (EMSA). A. Binding Competition Experiments. Equal volumes of diluted scCro protein from diluted bacterial lysates and 32P-labeled ORC, ORas11, or ORas12 were mixed to 100 nM (protein) and 20 pM (DNA) final concentrations in the presence of various concentrations (0.5-450 µg/mL) of unspecific competitor DNA (sonicated salmon sperm DNA) and incubated in 10% (w/v) glycerol, 0.06% (w/v) Nonidet-P40, 1 mM dithiothreitol, 10 µg/mL BSA, and 0.5 mM EDTA in BB, on ice for 60 min. Samples of 20 µL were loaded on a 10% acrylamide gel containing 0.5 × TBE and 50 mg/mL of glycerol running at 16 V/cm, at 4 °C. The gel was prerun for 20 min prior to sample loading. After loading, the gel was run for an additional 20-40 min at 32 V/cm. Following electrophoresis the gel was dried, and separated radioactivity was analyzed by phosphorimaging (Fuji BAS-2500). The fraction of formed scCro/DNA ligand complex was expressed as percent complex of total amount of ligand DNA and plotted versus the competitor DNA concentration. Measures of the inhibition were expressed as competitor concentrations resulting in 50% (IC50) or 90% (IC90) inhibition of complex formation. B. Determination of Dissociation Constants. Serial dilutions of purified protein were performed prior to mixing with 1:1 volumes of labeled DNA ligands resulting in the same concentrations of additives as described above and 10 µg/ mL sonicated salmon sperm DNA. Equal volumes of diluted protein and diluted DNA operator were mixed and incubated on ice for 60 min. Electrophoresis and analysis were performed as described above. The fractional binding of scCro/DNA complex was determined after phosphorimager analysis. Dissociation constants were determined after the fitting of a 1:1 binding isotherm by nonlinear regerssion using program SIMFIT (url:http://www.simfit.man.ac.uk) to the experimental data. Structure Modeling. The tertiary structures of scCro and three different variants, A12:407, B12:407, and C12:405, were modeled as follows. (1) Modeling of ScCro: The AGTGGSGG peptide linker connecting the two subunits in scCro was introduced using the program Biopolymer (Accelrys, San Diego, CA) and the atomic coordinates for the crystal structure of wild-type Cro in 5CRO (14). The crude scCro model was soaked in a 5 Å layer of water and subsequently relaxed by energy minimizations of, initially, the solvent only, followed by protein hydrogens, side chains, and finally the whole molecular system using the steepest descent algorithm in Discover (Accelrys). After this relaxation, a conjugate gradient minimization was conducted for 1000 iterations. Following energy minimization the molecular dynamics of the soaked scCro was simulated until the total energy of the system reached equilibrium (∼10 ps) at 283 K. An average equilibrium structure was calculated and finally energy-minimized by conjugate gradient until convergence was reached. (2) Modeling of Variant ScCro

Biochemistry, Vol. 43, No. 38, 2004 12041 Proteins: Side-chain replacements in positions 27, 28, 29, 31, and 32 were introduced into the relaxed crude model of scCro8 using Biopolymer. The subsequent calculations were performed as described for scCro. Comparison of the modeled structures was performed by superimposition of the polypeptide backbone heavy atoms using InsightII (Accelrys). RESULTS Library Construction and Phage-Display Selection. Library A: The first constructed library was designed for phagedisplay selection of scCro variants with random substitutions in residues 27-29 and 31-33 (Figure 1a). In wild-type Cro Gln-27, Ser-28, Ala-29, Asn-31, and Lys-32 make specific contacts with nucleotides in the target ORC DNA (15) (Figure 2). Ala-33 does not interact with bound DNA directly but was included in the mutagenesis scheme due to its vicinity to the DNA-binding motif. The number of individual scCro-phage clones in the final library was titrated to be 107, corresponding to a representation of 1% of all possible gene combinations after random mutagenesis using NNS codons (N ) G, A, T, or C and S ) G or C) (326 ≈ 109). The fraction of wild-type scCro genes, a consequence of vector self-ligation, was