A cyclin D1/cyclin-dependent kinase 4 binding site within the C domain of the retinoblastoma protein.

Phosphorylation of the retinoblastoma protein (Rb) by the cyclin D1/cyclin-dependent kinase (cdk) 4 complex (cdk4/D1) is a key regulatory step for maintaining the orderly progression of the cell cycle. The B domain of Rb contains a site that recognizes and binds the LXCXE motif found in D-type cyclins. This interaction is important for phosphorylation of Rb by cdk4/D1, although in vitro the Rb C domain alone is efficiently phosphorylated by cdk4/D1. A mutation in the C domain of Rb, L901Q, has been identified that completely abolishes cdk4/D1 phosphorylation of the isolated C domain. By contrast, the L901Q mutation has no effect on phosphorylation by either cyclin E/cdk2 or cyclin B/cdk1, suggesting that the interaction between L901Q and cdk4/D1 is specific. Introduction of the L901Q mutation into Rb containing the A, B, and C domains results in phosphorylation becoming predominantly dependent on the LXCXE binding region. However, when the LXCXE binding region of Rb is mutated, phosphorylation becomes dependent on the L901 site within the C domain. The L901 binding site can supplant the LXCXE binding site for the cdk4/D1-dependent phosphorylation of S780 and S795 but not S807/S811. Despite the limited homology between C domains of Rb, p107, and p130, the L901 site is conserved and introduction of the L925Q mutation into the isolated C domain of p107 also inhibits phosphorylation by cdk4/D1. These data support a model for cdk4/D1 recognizing two independent binding sites in Rb and suggests a conservation of this C domain binding motif for cyclin D1/cdk4 kinase among the Rb family of proteins.

Evolution of binding affinity in a WW domain probed by phage display.

The WW domain is an approximately 38 residue peptide-binding motif that binds a variety of sequences, including the consensus sequence xPPxY. We have displayed hYAP65 WW on the surface of M13 phage and randomized one-third of its three-stranded antiparallel beta-sheet. Improved binding to the hydrophobic peptide, GTPPPPYTVG (WW1), was selected in the presence of three different concentrations of proteinase K to simultaneously drive selection for improved stability as well as high-affinity binding. While some of the selected binders show cooperative unfolding transitions, others show noncooperative thermal unfolding curves. Two novel WW consensus sequences have been identified, which bind to the xPPxY motif with higher affinity than the wild-type hYAP65 WW domain. These WW domain sequences are not precedented in any natural WW domain sequence. Thus, there appear to be a large number of motifs capable of recognizing the target peptide sequence, only a subset of which appear to be used in natural proteins.

Defining the substrate specificity of cdk4 kinase-cyclin D1 complex.

cdk4 kinase-cyclin D1 complex (cdk4/D1) does not phosphorylate all of the sites within retinoblastoma protein (Rb) equally. Comparison of five phosphorylation sites within the 15 kDa C domain of Rb indicates that Ser795 is the preferred site of phosphorylation by cdk4/D1. A series of experiments has been performed to determine the properties of this site that direct preferential phosphorylation. For cdk4/D1, the preferred amino acid at the third position C-terminal to the phosphorylated serine/threonine is arginine. Substitution of other amino acids, including a conservative change to lysine, has dramatic effects on the rates of phosphorylation. This information has been used to mutate less favorable sites in Rb, converting them to sites that are now preferentially phosphorylated by cdk4/D1. A conserved site at Ser842 in the related pocket protein p107 is also preferentially phosphorylated by cdk4/D1. Although Rb and p107 differ significantly in sequence, the Rb Ser795 site can replace the p107 Ser842 site without affecting the rate of phosphorylation. These results suggest that although a determinant of specificity resides in the sequences surrounding the phosphorylated site, the structural context of the site is also a critical parameter of specificity.

Efficient display of an HCV cDNA expression library as C-terminal fusion to the capsid protein D of bacteriophage lambda.

We describe the construction and characterization of a hepatitis C virus (HCV) cDNA expression library displayed as a fusion to the carboxy terminus of the capsid protein D of bacteriophage lambda. cDNA inserts were obtained by tagged random-priming of the HCV genome and cloned into a lambda vector from which chimeric phage bearing both wild-type D protein and D fusion products on the capsid surface were produced. The resulting library was affinity-selected with anti-HCV human monoclonal antibodies recognizing linear or conformational epitopes, and human sera from HCV-infected patients. Selection was monitored by immuno-screening experiments, ELISA, and sequence analysis of positive clones. The performance of this library was compared with two additional HCV cDNA display libraries generated as N-terminal fusions to the III and VIII capsid proteins of filamentous phage M13. The results obtained demonstrate the great potential of the lambda display system for constructing complex cDNA libraries for natural ligand discovery.

Defining the minimal portion of the retinoblastoma protein that serves as an efficient substrate for cdk4 kinase/cyclin D1 complex.

We have determined the minimal portion of the retinoblastoma protein (Rb) that can serve as an efficient substrate for in vitro phosphorylation by cdk4 kinase-D1 cyclin. Kinetic measurements indicate that in vitro, a 15-kDa fragment that represents the C-terminus of Rb can serve equally well as a substrate when compared with the larger 56-kDa fragment of Rb, which contains the A, B and C domains. By comparison, peptide substrates appear to be 1000-fold less efficient. Furthermore, mutational analysis indicates that not all of the five phosphorylation sites within this minimal C domain are phosphorylated equally by cdk4/D1. Ser795 is the preferred phosphorylation site, whereas the four remaining sites Ser807, Ser811, Thr821 and Thr826 are phosphorylated to a much lesser degree. Truncations of the C domain from the carboxy terminus indicate that almost all of this domain is required for efficient phosphorylation. These data suggest that the structural context of the phosphorylation site within the substrate is critical for its phosphorylation by the cdk4/D1 kinase.

Generation of Cre recombinase-specific monoclonal antibodies, able to characterize the pattern of Cre expression in cre-transgenic mouse strains.

Transgene-encoded Cre recombinase can target alteration of loxP-tagged genes to specific cell types and developmental stages in mice, depending on the pattern of transgene expression. To facilitate determination of the latter, we have generated monoclonal anti-Cre antibodies which are specific for distinct epitopes on the recombinase and detect Cre both on immunoblots and intracellularly by immunofluorescence. We demonstrate the usefulness of these antibodies by an analysis of Cre expression in mice carrying a cre-transgene under B cell-specific control.

In vitro evolution of thermodynamically stable turns.

To determine the role of primary structure in specifying turns, random sequences (guests) were substituted for the native turn sequences in a series of proteins (hosts) of differing thermodynamic stabilities.The fraction of inserts that result in active proteins is measured as a function of the stability of the host and temperature. With a highly stable host, more than half of the inserts give functional proteins. However, a smaller fraction of sequences supports folding as the stability of the host decreases, and the temperature increases. The sequences of many of the selected inserts resemble the wild-type turn, and those that diverge match other established turn preferences. Thermodynamic measurements show that turn sequences selected under stringent conditions result in the most stable proteins. Thus, beta-turns appear to be under evolutionary pressure favouring thermodynamically stable structures.

Phage display: protein engineering by directed evolution.

Phage display of proteins has become an important tool for protein engineering. Over the past year, the versatility of the technology has expanded to include the development of DNA-binding proteins with novel specificities, energetics of protein folding and directed evolution of antibodies. In addition, display of expressed cDNA libraries opens an exciting opportunity for studying protein-protein interactions.

Tendamistat as a scaffold for conformationally constrained phage peptide libraries.

The alpha-amylase inhibitor Tendamistat (Hoe-467), a 74 amino acid beta-sheet protein from Streptomyces tendae has been expressed on the surface of the filamentous bacteriophage M13. Phage displaying Tendamistat inhibit the hydrolysis of starch by alpha-amylase, indicating that the displayed protein is functional. The displayed Tendamistat has been used as a molecular scaffold for the presentation of constrained random peptides. Two loops, comprising residues 38 to 40 and 60 to 65 of Tendamistat, were randomized using PCR mutagenesis. Libraries of approximately 10(8) different mutant Tendamistat molecules were tested for binding to monoclonal antibody A8, which recognizes endothelin. After three cycles of biopanning, phage were isolated that specifically bound the monoclonal antibody. Loop swapping and alanine replacement mutagenesis indicated that residues in the 60 to 65 loop are responsible for binding to the monoclonal antibody. This work demonstrates the use of relatively small non-antibody protein scaffolds for the presentation of constrained random peptide sequences to select for novel binding molecules.

Display of peptides and proteins on the surface of bacteriophage lambda.

The display of peptides or proteins on the surface of viruses is an important technology for studying peptides or proteins and their interaction with other molecules. Here we describe a display vehicle based on bacteriophage lambda that incorporates a number of features distinct from other currently used display systems. Fusions of peptides or protein domains have been made to the amino terminus of the 11-kDa D protein of the lambda capsid. These fusions assemble onto the viral capsid and appear to be accessible to ligand interactions, based on the ability of a monoclonal antibody to recognize an epitope fused to the D protein on phage heads. To produce large D fusion display libraries and yet avoid the cumbersome task of cloning many fragments into lambda DNA, we have used the Cre-loxP site-specific recombination system in vivo to incorporate plasmids encoding the D fusions into the phage genome. Finally, we show that D fusion proteins can be added in vitro to phage lacking D protein and be assembled onto the viral capsid.

Thermodynamic genetics of the folding of the B1 immunoglobulin-binding domain from streptococcal protein G.

A method has been developed to select proteins that are thermodynamically destabilized yet still folded and functional. The DNA encoding the B1 IgG-binding domain from Group G Streptococcus (Strp G) has been fused to gene III of bacteriophage M13. The resulting fusion protein is displayed on the surface of the phage thus enabling the phage to bind to IgG molecules. In addition, these phage exhibit a small plaque phenotype that is reversed by mutations that destabilize the Strp G domain. By selecting phage with large plaque morphology that retain their IgG-binding function, it is possible to identify mutants that are folded but destabilized compared with wild-type Strp G. Such mutants can be divided into three general categories: 1) those that disrupt packing of hydrophobic side chains in the protein interior; 2) those that destabilize secondary structure; and 3) those that alter specific hydrogen bonds involving amino acid side chains. A number of the mutants have been physically characterized by circular dichroism and nuclear magnetic resonance and have been shown to have structures similar to wild-type Strp G but stabilities that were decreased by 2-5 kcal/mol.

Constrained peptide libraries as a tool for finding mimotopes.

The monoclonal antibody (mAb) KAA8 recognizes the peptide angiotensin II (AII). The KAA8 mAb was biopanned using two phage-displayed peptide libraries, one unconstrained, the other constrained by a disulfide bond. After several cycles of biopanning, both libraries showed enrichment for phage that bind KAA8. Phage isolated from the unconstrained library contain a consensus sequence that matches the sequence of AII. A consensus sequence was also identified from the constrained library that does not resemble the AII sequence, and represents a mimotope of AII. We have also demonstrated that monovalent phage display can be used to discriminate between modest and high-affinity binding peptides.

Identification of a structural epitope by using a peptide library displayed on filamentous bacteriophage.

The screening of phage-displayed random peptide libraries has recently emerged as a powerful technique for probing Ab-Ag interactions. We have used this method to identify the epitope recognized by a mAb, CB5B10, raised against plasminogen activator inhibitor type-1 (PAI-1). Two phage libraries, displaying random hexapeptides with or without flanking cysteine residues, were screened for binding to mAb CB5B10. The selected phages were shown to contain similar peptide sequences, all of which were flanked by cysteines. When compared with the crystal structure of PAI-1, the selected peptides closely resemble the sequence of a solvent-exposed loop connecting the COOH-terminal of an alpha-helix at Phe114 to a beta-sheet at Ser119. Because of the constraints imposed by the flanking cysteine residues, the selected peptides appear to mimic the structure and the sequence of the PAI-1 epitope. Specific contacts between the amino acids displayed by the phage and the mAb were explored using site-directed mutants of the phage peptide. The effects of these substitutions on binding to the mAb correlated well with the accessibility of the corresponding residues in the PAI-1 epitope. This is the first example of the use of phage-displayed peptide libraries to identify a structural epitope.

Identification of a peptide which binds to the carbohydrate-specific monoclonal antibody B3.

The monoclonal antibody (mAb) B3 recognizes an antigen found on the surface of many adenocarcinoma cells. While the structure of the cellular antigen is unknown, epitope mapping using neoglycoproteins with known carbohydrate moieties indicates that the mAb B3 reacts with the LewisY (LeY) antigen [Pastan et al., Cancer Res. 51 (1991) 3781-3787]. We have used mAb B3 to select for peptides that mimic the carbohydrate structure using libraries of filamentous phage displaying random peptides on their surface. Phage that were selected coded for the sequence APWLYGPA. The corresponding peptide was synthesized and tested for its ability to bind to mAb B3. The peptide was found to inhibit specifically the binding of 111In-labeled mAb B3 to A431 adenocarcinoma cells, as well as to inhibit killing of these cells by a B3 immunotoxin. In addition, the LeY carbohydrate, lactodifucotetraose, was able to compete with the phage displaying this peptide for binding to mAb B3. Alanine-scanning mutagenesis of the sequence coding for this peptide indicates that four residues, PWLY, were critical for binding to the mAb. The sequence is similar to other sequences known to mimic carbohydrate structures.

Identification of novel peptide antagonists for GPIIb/IIIa from a conformationally constrained phage peptide library.

Methods have recently been developed to present vast libraries of random peptides on the surface of filamentous phage. To introduce a degree of conformational constraint into random peptides, a library of hexapeptides flanked by cysteine residues (capable of forming cyclic disulfides) was constructed. This library was screened using the platelet glycoprotein, IIb/IIIa, which mediates the aggregation of platelets through binding of fibrinogen. A variety of peptides containing the sequence Arg-Gly-Asp or Lys-Gly-Asp were discovered and synthesized. The cyclic, disulfide-bonded forms of the peptides bound IIb/IIIa with dissociation constants in the nanomolar range, while reduced forms or an analogue in which Ser replaced the Cys residues bound considerably less tightly. These results demonstrate the feasibility for introducing conformational constraints into random peptide libraries and also demonstrates the potential for using phage peptide libraries to discover pharmacologically active lead compounds.

Stoichiometry of the Cre recombinase bound to the lox recombining site.

The site-specific recombinase Cre from bacteriophage P1 binds and carries out recombination at a 34 bp lox site. The lox site consists of two 13 bp inverted repeats, separated by an 8 bp spacer region. Both the palindromic nature of the site and the results of footprinting and band shift experiments suggest that a minimum of two Cre molecules bind to a lox site. We report here experiments that demonstrate the absolute stoichiometry of the Cre-lox complex to be one molecule of Cre bound per inverted repeat, or two molecules per lox site.

Evidence for a second conserved arginine residue in the integrase family of recombination proteins.

This study was designed to search for new regions of similarity in the integrase family of recombination proteins which consists of 28 members found in bacteria and yeast. A computer method based on an information content analysis has been used to align local regions of homology in the set of unaligned protein sequences from this family. Among the aligned regions with high information content were those containing the known conserved histidine, arginine and tyrosine residues. In addition, a new region was identified containing another arginine residue that appears to be conserved in all members of the family. To test further the importance of this newly identified arginine residue, mutants in the Cre protein from phase P1, a member of this integrase family, have been constructed which alter this residue. The mutations which change arginine to lysine and arginine to cysteine depress catalytic activity but not site-specific binding to the lox site. This result is expected for a conserved active site residue. This computer analysis also provides a means for searching for new members of the integrase family.