Design of catalytic polypeptides and proteins.

The first part of this review article lists examples of complete, empirical de novo design that made important contributions to the development of the field and initiated challenging projects. The second part of this article deals with computational design of novel enzymes in native protein scaffolds; active designs were refined through random and site-directed mutagenesis producing artificial enzymes with nearly native enzyme- like activities against a number of non-natural substrates. Combining aspects of de novo design and biological evolution of nature's enzymes has started and will accelerate the development of novel enzyme activities.

Fusion proteins from artificial and natural structural modules.

The purpose of preparing fusion proteins from designed and natural sequences is mainly twofold; it aims at the stabilization of structure and at the modification of biological activity. Fusion with beta-galactosidase, for example, can increase the intracellular stability and DDT-degrading activity of an artificial DDT-binding peptide, and fusions with a leucine zipper produce mono- and bifunctional single-chain variable domain antibody fragments or homodimeric and heterodimeric DNA-binding proteins like an artificial homodimeric HIV-1 enhancer-binding protein with increased binding specificity and repressor activity. Of importance are also short leader sequences that mediate the translocation of proteins across the cytoplasmic and the nuclear membrane. An interesting by-product of the leucine zipper-mediated dimerization of an HIV-1 enhancer-binding protein was the synthesis and the structural as well as functional characterization of a retro-leucine zipper.

The retro-GCN4 leucine zipper sequence forms a stable three-dimensional structure.

The question of whether a protein whose natural sequence is inverted adopts a stable fold is still under debate. We have determined the 2. 1-A crystal structure of the retro-GCN4 leucine zipper. In contrast to the two-stranded helical coiled-coil GCN4 leucine zipper, the retro-leucine zipper formed a very stable, parallel four-helix bundle, which now lends itself to further structural and functional studies.

Inhibition of HIV-1 enhancer-controlled transcription by artificial enhancer-binding peptides derived from bacteriophage 434 repressor.

An artificial HIV-1 enhancer-binding 42-residue peptide (R42) that had been derived from bacteriophage 434 repressor inhibited the cell-free in vitro transcription of HIV-1 enhancer-containing plasmids [Hehlgans, T., Stolz, M., Klauser, S., Cui, T., Salgam, P., Brenz Verca, S., Widmann, M., Leiser, A., Städler, K. & Gutte, B. (1993) FEBS Lett. 315, 51-55; Caderas, G. (1997) PhD Thesis, University of Zürich]. Here we show that, after N-terminal extension of R42 with a viral nuclear localization signal, the resulting nucR42 peptide was active in intact cells. NucR42 could be detected immunologically in nuclear extracts and produced a 60-70% reduction of the rate of transcription of an HIV-1 enhancer-carrying plasmid in COS-1 cells that had been cotransfected with the HIV enhancer plasmid, an expression plasmid for nucR42, and a control. NucR42 was also synthesized chemically and the synthetic product characterized by HPLC, mass spectrometry, and quantitative amino acid analysis. Band shift, footprint, and in vitro transcription assays in the presence of exogenous NF-kappaBp50 indicated that the binding sites of nucR42 and NF-kappaB on the HIV enhancers overlapped and that a relatively small excess of nucR42 sufficed to displace NF-kappaBp50. Band shift and in vitro transcription experiments showed also that exchange of the 434 repressor-derived nine-residue recognition helix of nucR42 for four glycines abolished the HIV enhancer binding specificity whereas leucine zipper- or retro-leucine zipper-mediated dimerization of R42 analogues increased it suggesting the potential application of such dimeric HIV enhancer-binding peptides as intracellular inhibitors of HIV replication.

Synthesis, physicochemical characterization, and crystallization of a putative retro-coiled coil.

An artificial HIV enhancer-binding polypeptide has recently been dimerized by covalently linking it to the leucine zipper motif of the yeast transcriptional activator GCN4 (Liu N et al., 1997, Eur Biophys J 25:399-403). Although it seemed that the dimerization of this peptide could be best achieved by the use of the retro sequence of the leucine zipper, this approach was not implemented in the original construct. As the first step toward the synthesis of a basic region-retro leucine zipper HIV enhancer-binding fusion protein, we have now prepared the retro version of the leucine zipper (r-LZ35) and performed initial physicochemical characterization. Circular dichroism and sedimentation equilibrium studies showed that, at concentrations < 100 microM, the retro peptide was an unstructured monomer. At higher concentrations, however, the monomer was in equilibrium with a tetramer and, at 1 mM, the retro peptide was almost fully helical. N-terminal extension of the retro peptide by the tripeptide Cys-Gly-Gly resulted in a 38-residue polypeptide that could be covalently dimerized by forming a disulfide bond between two chains to give the peptide (r-LZ38)2. Even in the low micromolar concentration range peptide (r-LZ38)2 formed a stable, noncovalent, helical dimer as revealed by circular dichroism and sedimentation equilibrium in the presence and absence of guanidinium chloride. (r-LZ38)2 has been crystallized and X-ray structural analysis is under way. The disulfide-crosslinked retro-leucine zipper may lend itself to interesting protein structural studies, including protein design. The present work also highlights the structural and functional potential of retro proteins in general.

An artificial HIV enhancer-binding peptide is dimerized by the addition of a leucine zipper.

A 42 residue artificial peptide that binds to the HIV-1 enhancers has been described previously. The specificity of interaction of the peptide with its target DNA sequence has been demonstrated by a variety of techniques. Naturally occurring regulatory proteins frequently bind to DNA as dimers, thereby increasing the strength and specificity of the interaction, the dimer interface often being provided by a leucine zipper type coiled coil. As a suitable binding site for this kind of system is located to the 5' end of the HIV enhancer region, it was decided to design and synthesize a fusion peptide that not only contained the DNA binding sequence of the original 42 residue peptide but also incorporated a leucine zipper based on that of the GCN4 transcriptional activator, that should, therefore, be capable of dimerizing. The resultant peptide, LZ66, has now been shown to be fully active in band shift and in vitro transcription assays and to exhibit about double the inhibitory activity of the parent 42 residue peptide. Preliminary CD measurements revealed that the peptide has a high alpha-helical content and that it adopts a stable conformation down to the low micromolar peptide concentration range. Sedimentation equilibrium studies confirmed that the principles involved in the design of the peptide are valid and that the peptide is indeed dimeric in solution.

Design, synthesis, and characterization of HIV-1 enhancer-binding polypeptides derived from bacteriophage 434 repressor.

We have designed and synthesized HIV-1 enhancer-binding polypeptides that were derived from bacteriophage 434 repressor. These peptides were 39-54 residues long and contained either the recognition helix or the entire helix-turn-helix motif of the DNA-binding domain of 434 repressor. The dissociation constant of the complex formed between the standard peptide (R42) and a synthetic 70-bp HIV enhancer DNA was ca. 10(-8) M. The specificity of the interaction of R42 with the two HIV enhancers was demonstrated by competitive band shift assays, stepwise displacement of the p50 subunit of transcription factor NF-kappa B from its two HIV enhancer binding sites, and DNase I footprinting; R42 seemed to protect best the two TTTCC sequences of the HIV enhancers against digestion by DNase I. R42 analogues with mutated recognition helix had lower DNA binding specificity. It remains to be investigated whether our artificial HIV enhancer-binding polypeptides are active in vivo.

The DNA-binding properties of an artificial 42-residue polypeptide derived from a natural repressor.

Bacteriophage 434 repressor recognizes the operator sequences ACAAG and ACAAT. As the same or similar sequences occur in the enhancer region of HIV-1, 434 repressor was a potential HIV enhancer-binding protein. We found that the interaction of the DNA-binding domain of 434 repressor with a 57-bp HIV enhancer DNA was very weak whereas a 42-residue construct, comprising the recognition helix and four copies of a positively charged segment of the repressor, bound strongly. The results of footprint and cell-free in vitro transcription studies showed that the 42-residue peptide bound preferably to the enhancer region of HIV-1 and acted as an artificial repressor. Replacement of an essential glutamine of the recognition helix by glutamic acid resulted in a partial shift of the sequence specificity of the 42-residue peptide.

Carbohydrate masking of an antigenic epitope of influenza virus haemagglutinin independent of oligosaccharide size.

Comparison of the haemagglutinins (HA) of the pathogenic avian influenza viruses A/FPV/Dutch/27 (H7N7) and A/FPV/Rostock/34 (H7N1) revealed 94.7% nucleotide and 93.8% amino acid sequence homologies. Six of the seven N-glycosidic oligosaccharides of the Rostock HA are at the same positions as the six carbohydrates of the Dutch strain. The additional oligosaccharide side chain of the Rostock strain, which is of the complex type, is attached to asparagine149 in antigenic epitope B. The accessibility of this antigenic epitope has been analysed by using rabbit antisera raised against synthetic peptides comprising amino acids 143-162. The carbohydrates of the HA of the Rostock strain have been modified (i) to truncated cores by expression in insect cells using a baculovirus vector, (ii) to oligomannosidic side chains by growth in the presence of the trimming inhibitor methyldeoxynojirimycin and (iii) to a single N-acetylglucosamine residue by removal of the oligomannosidic sugar with endo-beta-N-acetylglucosaminidase H. Neither the authentic nor the modified oligosaccharides allowed antibody binding, as indicated by enzyme-linked immunosorbent assay (ELISA) and Western blot analyses. Reactivity was observed, however, after complete removal of the carbohydrate from HA of the Rostock strain by digestion with peptide-N-glycosidase F. HA of the Dutch strain was reactive without prior peptide-N-glycosidase F treatment. These results demonstrate that a single N-acetyl-glucosamine at asparagine149 is sufficient to prevent recognition of the peptide epitope.

Structure-function studies of designed DDT-binding polypeptides.

An artificial 24-residue DDT-binding polypeptide (Moser, R., Thomas, R.M., and Gutte, B. (1983) FEBS Lett. 157, 247-251) and several analogues of this peptide were characterized by ligand binding, spectroscopic, and immunological studies. Comparison of dissociation constants showed that Phe14 and His16 were important for DDT binding and that the designed peptide had noticeable ligand specificity. Measurement of the circular dichroism of the artificial DDT-binding peptide revealed a high proportion of beta-structure which was abolished only partly by 8 M urea. When Tyr15, Tyr17, and Phe3 whose side chains were on the same side of the proposed beta-sheet were replaced by non-aromatic amino acids, the cross-reactivity with antibodies against the original DDT-binding peptide decreased stepwise. In summary, the results of this study support essential features of our structural model of the designed 24-residue DDT-binding peptide.

Preparation and characterization of polyclonal and monoclonal antibodies against the insecticide DDT.

A synthetic DDT derivative in which the molecular structure of DDT was completely retained was coupled to bovine serum albumin. Animals were immunized with the DDT-bovine serum albumin conjugate and polyclonal and monoclonal antibodies against the insecticide were isolated. These antibodies seemed to be the first true anti-DDT antibodies and distinguished much better between DDT and DDT metabolites than previously prepared anti-DDT antisera. In competitive solid phase radioimmunoassays, DDT concentrations as low as 10 nM or 0.0035 mg/1 were detectable. The anti-DDT antibodies can be used for environmental analyses and lend themselves to the elucidation of the structure of the DDT binding site.

Rapid partial degradation of DDT by a cytochrome P-450 model system.

Heme compounds, in combination with a reducing agent and oxygen, can express various activities of cytochrome P-450 enzymes. In the present study it was found that a mixture of hemin and excess cysteine was able to degrade the insecticide DDT partially. The major products were three water-soluble, non-toxic conjugates of DDT metabolites with cysteine which had lost two or three of the five chlorine atoms of DDT per molecule and whose structures were elucidated by gas chromatography/mass spectrometry. In 0.05 M NH4HCO3, pH 7.7/ethanol (5:6, by vol.), the degradation reaction catalyzed by the hemin-cysteine model system was at least 8 x 10(4) times faster than the uncatalyzed reaction. In the presence of a designed 24-residues polypeptide or beta-casein, two DDT-binding proteins, an additional fourfold increase in the rate of DDT degradation was observed. Although the concentrations of DDT and cysteine occurring in an organism would be expected to be lower than those in the experiments described, the formation of water-soluble conjugates of DDT metabolites with cysteine (and other amino acids) could also play a role in metabolism and excretion of DDT in vivo.

Expression of the synthetic gene of an artificial DDT-binding polypeptide in Escherichia coli.

This paper reports the expression of an artificial functional polypeptide in bacteria. The gene of a designed 24-residue DDT-binding polypeptide (DBP) was inserted between the BamHI and PstI cleavage sites of plasmid pUR291. The hybrid plasmid, pUR291-DBP, was cloned in Escherichia coli JM109. After induction by isopropyl-beta-D-thiogalactopyranoside a fusion protein was expressed in which DBP was linked to the COOH-terminus of beta-galactosidase. DBP, which is stable to trypsin, was obtained by tryptic digestion of the fusion protein and subsequent fractionation of the tryptic peptides by reversed-phase h.p.l.c. Recombinant and chemically synthesized DBP showed identical chromatographic properties, amino acid composition, and chymotryptic digestion patterns. Both the beta-galactosidase-DBP fusion and isolated recombinant DBP bound DDT. The fusion protein was 25 times as potent as the designed 24-residue DBP in activating a cytochrome P-450 model system using equimolar catalytic amounts of the two proteins.

A crystalline synthetic peptide representing the epitope of a monoclonal antibody raised against synthetic interferon-alpha 1 fragment 111-166.

The antigenic determinant recognized by the monoclonal antibody that had been raised against synthetic human interferon-alpha 1 (IFN-alpha 1) fragment 111-166 [Arnheiter, H., Thomas, R.M., Leist, T., Fountoulakis, M., and Gutte, B. (1981) Nature (Lond.) 294, 278-280] and that cross-reacted with human IFN-alpha 1, IFN-alpha 2, and IFN-alpha A made in Escherichia coli, was localized to the region between residues 151 and 166 using synthetic COOH-terminal interferon fragments. In solid-phase radioimmunoassays neither the strongly hydrophilic COOH-terminal nonapeptide IFN 158-166 nor its mixtures with IFN 151-162 or IFN 149-158 showed any measurable interaction with the antigen binding site of the monoclonal antibody. For antibody binding, the full covalent structure of IFN 151-166 was required. Quantitatively very similar results were obtained with IFN 149-166 and IFN 143-166. The synthetic COOH-terminal hexadecapeptide of human IFN-alpha 1 (IFN 151-166) could be crystallized.

Orientation of a human leukocyte interferon molecule on its cell surface receptor: carboxyl terminus remains accessible to a monoclonal antibody made against a synthetic interferon fragment.

An 125I-labeled monoclonal antibody made against a synthetic 56-residue fragment of human leukocyte interferon (IFN) alpha 1 recognizes human, Escherichia coli-derived IFN alpha A bound to the surface of Madin-Darby bovine kidney cells. A major fraction of the antibody recognizes IFN specifically bound to the cells, because the number of bound antibody molecules corresponds to the number of cell-bound IFN molecules (as measured with radiolabeled ligand) and because the fraction of the IFN unspecifically bound to the cells is less than 10% of the total bound IFN. A synthetic carboxyl-terminal 16-residue IFN peptide, though not inhibiting binding of IFN to cells, inhibits binding of antibody to IFN. A recombinant IFN alpha A molecule with a carboxyl-terminal 13-residue deletion, though still able to compete for binding of IFN to cells, is not recognized by the antibody. Scatchard plot analysis of the binding data revealed apparent dissociation constants of 6.0 x 10(-10) M for the antibody-IFN interaction and of 4.0 x 10(-11) M for the IFN-cell receptor interaction. The antibody inhibits the binding of IFN to cells only weakly and neutralizes the antiviral activity of the ligand only when in a large molar excess. We conclude that the carboxyl-terminal 10-16 residues that are predicted from the cloned IFN cDNAs and that are present in some natural IFNs are not involved in binding to cells but are antigenic and hence exposed on the molecules' surface. That the carboxyl terminus is not directly involved in binding to cells is consistent with the observation that some IFNs with carboxyl-terminal deletions are biologically active.

Design, synthesis and characterisation of a 34-residue polypeptide that interacts with nucleic acids.

Based on secondary structure prediction rules and model building a neutral artificial 34-residue polypeptide with potential nucleic acid-binding activity was synthesised. This peptide and its covalent dimer showed strong interaction with cytidine phosphates and single-stranded DNA. The dimer had considerable ribonuclease activity with high preference for cleavage at the 3'-end of C.