On the conservation of protein sequences in evolution.

The amino acid sequences of enzymes like alcohol dehydrogenase and glyceraldehyde-3-phosphate dehydrogenase are strongly conserved across all phyla. We suggest that the amino acid conservation of such enzymes might be a result of the fact that they function as part of a multi-enzyme complex. The specific interactions between the proteins involved would hinder evolutionary change of their surfaces.

Dimerisation mutants of Lac repressor. II. A single amino acid substitution, D278L, changes the specificity of dimerisation.

Assembly of the lactose repressor tetramer involves two subunit interfaces, the C-terminal heptad repeats, and the monomer-monomer interface. Dimerisation between two monomers of Lac repressor of Escherichia coli lacking the two C-terminal heptad repeats occurs through the interactions between three alpha-helices of each monomer, which form a highly hydrophobic interface. Residues possibly involved in specific dimer formation are known from X-ray studies and from the phenotypes of more than 4000 single amino acid substitutions. During the examination of numerous mutants within the dimerisation interface of Lac repressor, we found that substitution of one amino acid, D278 to leucine, is sufficient to change the specificity of dimerisation. Analysis of this single substitution indicates that D278L mutant Lac repressor represses like wild-type. However, it no longer forms heterodimers with wild-type Lac repressor.

Dimerisation mutants of Lac repressor. I. A monomeric mutant, L251A, that binds Lac operator DNA as a dimer.

Dimer formation between monomers of the Escherichia coli Lac repressor is substantially specificed by the interactions between three alpha-helices in each monomer which form a hydrophobic interface. As a first step in analysing the specificity of this interaction, we examined the mutant L251A. LacR bearing this mutation in a background lacking the C-terminal heptad repeats is completely incapable of forming dimers in solution, with a dimer-monomer equilibrium dissociation constant, or Kd, higher than 10(-5)M. This correlates with a 200-fold decrease in its ability to repress the lac operon in vivo compared to dimeric LacR. Surprisingly, the mutant is still capable of forming dimers upon binding to short operator DNA in vitro. Analysis of the kinetic parameters of binding of the mutant to operator DNA reveals a 2000 to 3000-fold increase in the equilibrium dissociation constant (Kd) of the mutant-DNA complex in comparison to dimeric LacR-operator complexes, with the change almost entirely due to a greater than 1000-fold decrease in association rate. The dissociation rate varies only by a factor of about two, in comparison to dimeric LacR. This change reflects a kinetic pathway in which dimer formation, in solution or on DNA, is the rate-limiting step. These findings have implications for the specificity and stability of the protein-protein interface in question.

A novel feature of DNA recognition: a mutant Gcn4p bZip peptide with dual DNA binding specificities dependent of half-site spacing.

Homodimeric DNA-binding proteins with relaxed half-site spacing requirements for their DNA targets have been described. As an example, the yeast transcriptional activator Gcn4p binds in vitro equally well to the AP1 site (5'A4T3G2A1C0T1'C2'A3'T4'3') and the ATF/CREB site (5'A4T3G2A1C0G0'T1'C2'A3'T4'3'), which have identical but differently spaced half-site blocks. We describe a novel feature for the bZip class of DNA-binding proteins. The N-14 mutant of a Gcn4p-derived bZip peptide shows a diametrically opposed base-pair recognition specificity depending on the half-site spacing of its DNA target: on pseudo-palindromic, AP1 site-like binding sites, guanine is required in position 2 for proper binding; in contrast, on palindromic, ATF/CREB site-like targets, position 2 must be cytosine to prevent a loss of binding. Modeling studies suggest that the different base-pair requirements on differently spaced DNA targets are due to minimal alterations of the distances between the relevant atoms of the N-14 side-chain and the corresponding target groups on the DNA. Although the N-14 peptide does not have a natural counterpart, its behavior hints at the possibility that dual binding modi dependent on half-site spacing may occur also for natural homodimeric DNA-binding proteins.

The two homeodomains of the ZmHox2a gene from maize originated as an internal gene duplication and have evolved different target site specificities.

The maize ZmHox2a gene encodes two homeodomains which originated by a 699 bp duplication within an ancestral precursor. The sequences of the two ZmHox2a homeodomains are highly diverged in the N-terminal arm, while residues in the helical part have mostly been conserved. We show here that both ZmHox2a homeodomains are functional DNA-binding motifs but exhibit different target site specificities. CASTing experiments reveal a TCCT motif recognized by HD1 but a GATC tetranucleotide as the recognition sequence of HD2. Mutation of the central nucleotides in both tetranucleotide core motifs abolishes DNA binding. A domain swap experiment indicates that target site specificity is achieved in a combinatorial manner by the contributions of the diverged N-terminal arms together with the slightly different recognition helices. Computer modelling suggests that K47 and H54 in the recognition helices preferentially contact the bases at the 3'-terminus of the tetranucleotide target sequences.

Cre mutants with altered DNA binding properties.

The recombinase Cre of bacteriophage P1 is a member of the family of site-specific recombinases and integrases that catalyze inter- and intramolecular DNA rearrangements. To understand how this protein specifically recognizes its target sequence, we constructed Cre mutants with amino acid substitutions in different positions of the presumptive DNA binding region. Here we present the results of in vitro DNA binding and in vivo recombination experiments with these Cre mutants. Most substitutions of presumptive DNA-binding amino acids in in vitro tests resulted either in the loss of target binding or in a broadening of target recognition specificity. Of the mutations resulting in a broadening of target specificity, one, N317A, results in a reduced recombination efficiency with the wild-type loxP target but recombines, in contrast to wild-type Cre, in in vivo experiments, with a symmetric variant of the wild-type target sequence. This target variant differs from wild-type loxP by the symmetric C to A replacement in position 6 of the inverted repeats. We propose a common multihelical DNA binding motif for the family of integrases and recombinases. This model implies a major structural rearrangement for the DNA binding region of lambda integrase, analogous to the structural rearrangements of the DNA binding motifs of other proteins when contacting their target DNA.

Genetic studies of the Lac repressor. XV: 4000 single amino acid substitutions and analysis of the resulting phenotypes on the basis of the protein structure.

Each amino acid from position 2 to 329 of Lac repressor was replaced by 12 or 13 of the 20 natural occurring amino acids. The resulting phenotypes are discussed on the basis of (1) the recently published structure of the Lac repressor core complexed with the inducer IPTG and (2) a model of the dimeric Lac repressor built by homology modelling from the X-ray structure of the purine repressor-corepressor-operator complex. This phenotype analysis, based on 4000 well-defined mutants, yields a functional description of each amino acid position of Lac repressor. In most cases, mutant effects can be directly correlated with the structure and function of the protein. This connection between the amino acid position and the structure and function of the protein is in most cases direct and not complicated: amino acids which are directly involved in sugar binding are affected in Lac repressor mutants of the Is type; small amino acids which can only be replaced by other small acids are located in the core of the protein; positions at which nearly all amino acids are tolerated are in most cases located on the surface of the protein. Amino acids which are highly conserved throughout the LacI family of repressors, and not directly involved in specific functions of the protein like DNA recognition or sugar binding, form a network of contacts with other amino acids. Such amino acids are either located inside one subunit, mostly at the interface between secondary structure elements, or are involved in the dimerisation interface.

Mutant bZip-DNA complexes with four quasi-identical protein-DNA interfaces.

The complex between the yeast transcriptional activator GCN4 and the palindromic ATF/CREB site 5'- A4T3G2A1C0*G0'T1'C2'A3'T4'-3' shows dyad symmetry. The basic region of GCN4 contains a segment of 18 amino acids with a partially palindromic sequence: N-LKRARNTEA*ARRSRARKL-C. Symmetric residues are underlined. Apart from the ATF/CREB site, GCN4 also binds well to the symmetric variants with guanine in position 4 (5'-G4T3G2A1C0*G0'T1'C2'A3'C4'-3') or thymine in position 0 (5'-A4T3G2A1T0*A0'T1'C2'A3'T4'-3'). The half-sites of these sequences can be regarded as short pseudo-palindromes with central guanine 2/cytosine 2' base pairs. We investigated whether the geometry of the peptide of the basic region of GCN4 could be functionally related to the pseudo-palindromic character of some target half-sites. Since inspection of the X-ray structures of GCN4-DNA complexes reveals that several amino acid-DNA interactions are symmetric within the wild-type half-complexes, we introduced mutations into a GCN4 bZip peptide that improve the symmetry of the peptide. We found that most of the constructs retain specific DNA recognition. For one mutant, we conclude that it is not only capable of forming DNA complexes showing the well-known overall dyad symmetry, but that the protein-DNA interface of each half-complex can be divided further into two quasi-identical, quasi-symmetric substructures.

Replacement of invariant bZip residues within the basic region of the yeast transcriptional activator GCN4 can change its DNA binding specificity.

Two residues are invariant in all bZip basic regions: asparagine -18 and arginine -10 (we define the first leucine of the leucine zipper of GCN4 as +1). X-ray structures of two specific GCN4-DNA complexes (Ellenberger et al., Cell, 71, 1223-1237, 1992; König & Richmond, J. Mol. Biol., 233, 139-154, 1993) demonstrate the involvement of both residues in specific base pair recognition. We replaced either asparagine -18 or arginine -10 with all other amino acids and tested the DNA binding properties of the resulting mutant peptides by gel mobility shift assays. Peptides with histidine -18 or tyrosine -10 bind with changed specificities to variants of the ATF/CREB site 5'A4T3G2A1C0*G0'T1'C2'A3'T4'3' with symmetric exchanges in positions 2/2' or 0/0', respectively. The double mutant with histidine -18 and tyrosine -10 combines the features of the parental single mutants and binds specifically to the respective double exchange target. Furthermore, the tyrosine -10 mutant clearly prefers the palindrome 5'ATGATATCAT3' over the corresponding pseudo-palindrome 5'ATGATTCA-T3', whereas the lysine -10 mutant binds better to the pseudo-palindromic AP1 site 5'ATGACTCAT3' than to the palindromic ATF/CREB site. Thus, although invariant within natural bZip proteins, asparagine -18 or arginine -10 can be functionally replaced by other amino acids, and their replacement can lead to new DNA binding specificities.

Creating new DNA binding specificities in the yeast transcriptional activator GCN4 by combining selected amino acid substitutions.

The specificity of the GCN4/DNA complex is mediated by a complicated network of interactions between the basic regions of both GCN4 monomers and their target halfsites. According to X-ray analyses (1, 2) one particular thymine of the target sequence is recognized by serine -11 and alanine -15 (we define the leucine in the first d-position of the heptad repeats as +1). We replaced serine -11 or alanine -15 with all other amino acids and analysed the DNA binding properties of the resulting stable GCN4 derivatives by electrophoretic mobility shift assays. Among these, mutants with tryptophan in position -11, or glutamic acid and glutamine in position -15, differ significantly from GCN4 in their DNA binding specificities. We then constructed selected double mutants, which differ from GCN4 in positions -11, -15 or -14 (3) of the basic region. The double mutants with tryptophan in position -11 and asparagine or serine in position -14 show drastically altered DNA binding specificities, presumably due to additive effects.

Single exchanges of amino acids in the basic region change the specificity of N-Myc.

We exchanged specific amino acids in the basic region of the murine N-Myc protein and tested the mutant proteins for their DNA binding specificity. The amino acids we exchanged were chosen in analogy to residues of the homologous basic regions of bHLH and bZIP proteins. Mutant N-Myc peptides were expressed in Escherichia coli and specific DNA binding was monitored by gel shift experiments. For this we used palindromic target sequences with systematic base pair exchanges. Several mutants with altered DNA binding specificity were identified. Amino acid exchanges of residues -14 or -10 of the basic region lead to specificity changes (we define leucine 402 of N-Myc as +1; comparable to GCN4 see (1)). The palindromic N-Myc recognition sequence 5'CACGTG is no longer recognized by the mutant proteins, but DNA fragments with symmetrical exchanges of the target sequence are. Exchanges at position -15 broaden the binding specificity. These data were used to build a computer based model of the putative interactions of the N-Myc basic DNA binding region with its target sequence.

The mouse Enhancer trap locus 1 (Etl-1): a novel mammalian gene related to Drosophila and yeast transcriptional regulator genes.

A novel mouse gene, Enhancer trap locus 1 (Etl-1), was identified in close proximity to a lacZ enhancer trap integration in the mouse genome showing a specific beta-galactosidase staining pattern during development. In situ analysis revealed a widespread but not ubiquitous expression of Etl-1 throughout development with particularly high levels in the central nervous system and epithelial cells. The amino acid sequence of the Etl-1 protein deduced from the cDNA shows strong similarity, over a stretch of 500 amino acids, to the Drosophila brahma protein involved in the regulation of homeotic genes and to the yeast transcriptional activator protein SNF2/SWI2 as well as to the RAD54 protein and the recently described helicase-related yeast proteins STH1 and MOT1. Etl-1 is the first mammalian member of this group of proteins that are implicated in gene regulation and/or influencing chromatin structure. The homology to the regulatory proteins SNF2/SWI2 and brahma and the expression pattern during embryogenesis suggest that Etl-1 protein might be involved in gene regulating pathways during mouse development.

Substitutions of hydrophobic amino acids reduce the amyloidogenicity of Alzheimer's disease beta A4 peptides.

The deposition of amyloid protein aggregates in brain is the main pathological feature of Alzheimer's disease. Their principal constituent is a peptide termed beta A4, which comprises up to 43 amino acid residues. It is highly insoluble under physiological conditions and aggregates into filaments that form very dense clusters in vivo and in vitro. Based on a beta A4 prototype sequence spanning residues 10 to 42 or 43, we have designed analogues in which hydrophobic amino acid residues in position 17 to 20 were substituted by more hydrophilic residues. Depending on the kind of newly introduced amino acids and their position within the sequence, the substitution of only two residues led to variants exhibiting a broad spectrum of different properties. Common to them was a reduced beta-sheet content after solubilization in water and in the solid state. Some of the variants showed significantly reduced amyloidogenicity: although still forming filaments, they did not aggregate into the highly condensed depositions that are typical for amyloid. In addition, they could be solubilized in 200 mM-NaCl and KCl. When mixed with beta A4 peptides bearing the natural sequence, two of the analogues could inhibit the formation of filaments in vitro. These results demonstrate that a well-preserved hydrophobic core around residues 17 to 20 of beta A4 is crucial for the formation of beta-sheet structure and the amyloid properties of beta A4. The introduction of structural alterations within this region may guide the development of reagents for the therapy of Alzheimer's disease.

Lac repressor with the helix-turn-helix motif of lambda cro binds to lac operator.

Lac repressor, lambda cro protein and their operator complexes are structurally, biochemically and genetically well analysed. Both proteins contain a helix-turn-helix (HTH) motif which they use to bind specifically to their operators. The DNA sequences 5'-GTGA-3' and 5'-TCAC-3' recognized in palindromic lac operator are the same as in lambda operator but their order is inverted form head to head to tail to tail. Different modes of aggregation of the monomers of the two proteins determine the different arrangements of the HTH motifs. Here we show that the HTH motif of lambda cro protein can replace the HTH motif of Lac repressor without changing its specificity. Such hybrid Lac repressor is unstable. It binds in vitro more weakly than Lac repressor but with the same specificity to ideal lac operator. It does not bind to consensus lambda operator.

Human and rodent sequence analogs of Alzheimer's amyloid beta A4 share similar properties and can be solubilized in buffers of pH 7.4.

The filamentous amyloid protein aggregates found in the brain of patients affected with Alzheimer's disease principally consist of a peptide termed beta A4, according to its secondary structure of beta-pleated sheets and its molecular mass of about 4 kDa. It has a length of up to 42 or 43 residues. By chemical means, we have synthesized peptide analogs corresponding to the human and rodent beta A4 sequences. We describe structural and functional properties of peptides spanning residues 1-43, 10-23, 1-27 and 4-27 of beta A4. The peptides have been tested for their ability to form filaments in vitro. Their solubilities and secondary structures in solution and in the solid state have been used to detect differences between the properties of human and rodent beta A4 sequences. We show that mouse and rat beta A4 homologs are as amyloidogenic as the human sequence. The absence of amyloid deposits in the brain of aged rats and mice is therefore not due to the three amino acid substitutions identified within the sequence which is homologous to beta A4 of humans. Moreover, peptides corresponding to residues 1-27 of human and rodent beta A4 are solubilized under physiological conditions; thus they are very unlikely to form stable filaments in vivo.

A model of the lac repressor-operator complex based on physical and genetic data.

Computer graphics were used to build a molecular model of the complex of Lac repressor and lac operator. The model is based (a) on the NMR data of the Kaptein group [Boelens, R., Lamerichs, R. M. J. N., Rullmann, J. A. C., van Boom, J. H. & Kaptein, R. (1988) Protein Sequence Data Anal. 1, 487-498] and (b) on our genetic and biochemical data including specificity changes [Lehming, N., Sartorius, J., Kisters-Woike, B., von Wilcken-Bergmann, B. & Müller-Hill, B. (1990) EMBO J. 9, 615-621]. Effects of amino acid exchanges in the recognition helix could be predicted by the model and were subsequently tested and confirmed by genetic experiments. Comparison of the modelled lac complex with the known crystallographic structures of several helix-turn-helix DNA complexes reveals striking similarities and suggests rules which govern the recognition between particular amino acid side chains and particular base pairs in these systems.

Aggregation and secondary structure of synthetic amyloid beta A4 peptides of Alzheimer's disease.

The deposition of amyloid beta A4 in the brain is a major pathological hallmark of Alzheimer's disease. Amyloid beta A4 is a peptide composed of 42 or 43 amino acid residues. In brain, it appears in the form of highly insoluble, filamentous aggregates. Using synthetic peptides corresponding to the natural beta A4 sequence as well as analog peptides, we demonstrate requirements for filament formation in vitro. We also determine aggregational properties and the secondary structure of beta A4. A comparison of amino-terminally truncated beta A4 peptides identifies a peptide spanning residues 10 to 43 as a prototype for amyloid beta A4. Infrared spectroscopy of beta A4 peptides in the solid state shows that their secondary structure consists of a beta-turn flanked by two strands of antiparallel beta-pleated sheet. Analog peptides containing a disulfide bridge were designed to stabilize different putative beta-turn positions. Limited proteolysis of these analogs allowed a localization of the central beta-turn at residues 26 to 29 of the entire sequence. Purified beta A4 peptides are soluble in water. Size-exclusion chromatography shows that they form dimers that, according to circular dichroism spectroscopy, adopt a beta-sheet conformation. Upon addition of salts, the bulk fraction of peptides precipitates and adopts a beta-sheet structure. Only a small fraction of peptides remains solubilized. They are monomeric and adopt a random coil conformation. This suggests that the formation of aggregates depends upon a hydrophobic effect that leads to intra- and intermolecular interactions between hydrophobic parts of the beta A4 sequence. This model is sustained by the properties of beta A4 analogs in which hydrophobic residues were substituted. These peptides show a markedly increased solubility in salt solutions and have lost the ability to form filaments. In contrast, the substitution of hydrophilic residues leads only to small deviations in the shape of filaments, indicating that hydrophilic residues contribute to the specificity of interactions between beta A4 peptides.

The roles of residues 5 and 9 of the recognition helix of Lac repressor in lac operator binding.

We constructed expression libraries for Lac repressor mutants with amino acid exchanges in positions 1, 2, 5 and 9 of the recognition helix. We then analysed the interactions of residues 5 and 9 with operator variants bearing single or multiple symmetric base-pair exchanges in positions 3, 4 and 5 of the ideal fully symmetric lac operator. We isolated 37 independent Lac repressor mutants with five different amino acids in position 5 of the recognition helix that exhibit a strong preference for particular residues in position 2 and, to a lesser extent, in position 1 of the recognition helix. Our results suggest that residue 5 of the recognition helix (serine 21) contributes to the specific recognition of base-pair 4 of the lac operator. They further suggest that residue 9 of the recognition helix (asparagine 25) interacts non-specifically with a phosphate of the DNA backbone, possibly between base-pairs 2 and 3.