Cassette mutagenesis: an efficient method for generation of multiple mutations at defined sites

A method for introducing random single point deletions in specific DNA target sequences using oligonucleotides

We describe a method for the generation of random point deletions in any target DNA sequence using synthetic mixed oligonucleotides. A mixed pool of oligonucleotides, which contain single nucleotide deletions randomly distributed throughout the full length, was generated by a modification of the synthesis cycle of an automated DNA synthesiser that allowed the inefficient incorporation of nucleotide monomers during each cycle of synthesis. A family of oligonucleotides was used to prime in vitro synthesis of the complementary strand of a cloned DNA fragment in an M13 vector which had previously been passaged through a dut-, ung- Escherichia coli host. Strong selection for progeny from the newly synthesised strand is provided by transforming the heteroduplex into a dut+, ung+ host. This procedure introduced point deletions at 10-25% efficiency. It has been used to introduce point deletions into operator sequences which bind the yeast regulatory proteins encoded by MATa1 and MAT alpha 2.

Structural plasticity broadens the specificity of an engineered protease

The substrate specificity of alpha-lytic protease has been changed dramatically, with a concomitant increase in activity, by replacing an active-site Met with Ala. The substrate specificity of both this mutant and another similar mutant are extraordinarily broad. X-ray crystallographic analysis shows that structural plasticity, a combination of alternate side-chain conformations and binding-site flexibility, allows both large and small substrates to be well accommodated.

Receptor and antibody epitopes in human growth hormone identified by homolog-scanning mutagenesis

A strategy, termed homolog-scanning mutagenesis, was used to identify the epitopes on human growth hormone (hGH) for binding to its cloned liver receptor and eight different monoclonal antibodies (Mab's). Segments of sequences (7 to 30 residues long) that were derived from homologous hormones known not to bind to the hGH receptor or Mab's, were systematically substituted throughout the hGH gene to produce a set of 17 chimeric hormones. Each Mab or receptor was categorized by a particular subset of mutant hormones was categorized by a particular subset of mutant hormones that disrupted binding. Each subset of the disruptive mutations mapped within close proximity on a three-dimensional model of hGH, even though the residues changed within each subset were usually distant in the primary sequence. The mapping analysis correctly predicted those Mab's which could or could not block binding of the receptor to hGH and further suggested (along with other data) that the folding of these chimeric hormones is like that of HGH. By this analysis, three discontinuous polypeptide determinants in hGH--the loop between residues 54 and 74, the central portion of helix 4 to the carboxyl terminus, and to a lesser extent the amino-terminal region of helix 1--modulate binding to the liver receptor. Homolog-scanning mutagenesis should be of general use in identifying sequences that cause functional variation among homologous proteins.

p21-ras effector domain mutants constructed by "cassette" mutagenesis

A series of mutations encoding single-amino-acid substitutions within the v-rasH effector domain were constructed, and the ability of the mutants to induce focal transformation of NIH 3T3 cells was studied. The mutations, which spanned codons 32 to 40, were made by a "cassette" mutagenesis technique that involved replacing this portion of the v-rasH effector domain with a linker carrying two BspMI sites in opposite orientations. Since BspMI cleaves outside its recognition sequence, BspMI digestion of the plasmid completely removed the linker, creating a double-stranded gap whose missing ras sequences were reconstructed as an oligonucleotide cassette. Based upon the ability of the mutants to induce focal transformation of NIH 3T3 cells, a range of phenotypes from virtually full activity to none (null mutants) was seen. Three classes of codons were present in this segment: one which could not be altered, even conservatively, without a loss of function (codons 32 and 35); one which retained detectable biologic activity with conservative changes but which lost function with more drastic substitutions (codons 36 and 40); and one which retained function even with a nonconservative substitution (codon 39).

Modulation of the affinity of the single-stranded DNA-binding protein of Escherichia coli (E. coli SSB) to poly(dT) by site-directed mutagenesis

A vector for site-directed mutagenesis and overproduction of the Escherichia coli single-stranded-DNA-binding protein (E. coli SSB) was constructed. An E. coli strain carrying this vector produces up to 400 mg pure protein from 25 g wet cells. The vector was used to mutate specifically the Phe60 residue of E. coli SSB. Phe60 had been proposed to be located near the single-stranded-DNA-binding site. Substitution of the Phe60 residue by Val, Ser, Leu, His, Tyr and Trp gave proteins with no or only minor conformational changes, as detected by NMR spectroscopy. The affinity of the mutant E. coli SSB proteins for single-stranded DNA decreased in the order Trp greater than Phe (wild-type) greater than Tyr greater than Leu greater than His greater than Val greater than Ser, leading to the conclusion that position 60 is a site of hydrophobic interaction of the protein with DNA.

Generation of multiple independent substitution mutants by M13 in vitro mutagenesis using a single mutagenic oligonucleotide

A 56-nucleotide mutagenic oligomer containing six mismatches with the wild-type template was used to construct multiple transversion mutations in the putative heparin binding region of the rat neural cell adhesion molecule (NCAM) cDNA sequence. Mutants were screened by hybridization to the 56-mer. The relative stability of a mutant DNA:56-mer duplex correlated with the number of base substitutions present in the mutant sequence. Five independent categories of mutants carrying from two to five of the expected nucleotide substitutions were isolated. No mutations other than those directed by the 56-mer were observed. These results suggest a method for generating sets of related predefined substitution mutants.

Simple and highly efficient site-specific mutagenesis, by ligation of an oligodeoxyribonucleotide into gapped heteroduplex DNA in which the template strand contains deoxyuridine

A simple and highly efficient procedure for oligodeoxynucleotide (oligo)-directed mutagenesis has been developed. In this procedure, a gapped heteroduplex DNA is first constructed and purified. The gapped heteroduplex consists of a circular 'template' strand of DNA, which contains some misincorporated deoxyuridine nucleotides, and a complementary strand which does not contain deoxyuridine, and which lacks a defined segment. Making a specific change in the sequence of the DNA within the gapped region then only requires ligation and transformation. An oligo, exactly the same length as the gap, and with the desired sequence, is synthesized, purified, and ligated directly into the gap in the heteroduplex. When this DNA is used to transform wt (ung+) Escherichia coli, about 80% of the resulting plasmids contain the sequence determined by the synthetic oligo. One gapped heteroduplex preparation can be used for many mutagenesis experiments, so that this procedure is well-suited for producing a series of defined mutations within a defined target region flanked by sites for restriction enzyme cleavage. As the method does not require a polymerase, the effects of primer displacement and polymerase infidelity are avoided.

Inosine induced mutations

Two complementary 24 base single stranded oligonucleotides containing randomly located inosine residues were synthesized in vitro. Once annealed, the two oligonucleotides were cloned into derivatives of ColE1 and transformed into Escherichia coli. Sequence analysis of 157 clones yielded 305 mutations. The pattern of the mutations revealed the following: (1) The frequency of inosine induced mutations was significantly less than predicted from its content in the oligonucleotides; (2) Inosine incorporation resulted almost exclusively in base changes to guanine; (3) The mutation distribution is biased towards A/T to G/C substitutions; (4) There were reproducible position biases; and (5) There was a reproducible strand bias which was independent of the cassette orientation with respect to the plasmid origin of replication.

ADP-ribosyltransferase mutations in the catalytic S-1 subunit of pertussis toxin

The ADP-ribosyltransferase activity of pertussis toxin resides within the S-1 subunit of the toxin. Deletion mapping of a recombinant S-1 subunit produced in Escherichia coli showed that amino acids 2 through 180 are required for ADP-ribosylation of Gi protein. Mutants of the S-1 subunit which lacked either amino acids 2 through 22 or amino acids 153 through 180 failed to express enzyme activity, implicating a functional or structural role for these residues in catalysis. The catalytic carboxy-terminal S-1 deletion, C-180, was found to be more soluble than the recombinant S-1 subunit, making it a useful construct for future structure-function studies on enzyme catalysis. Four independent single-amino-acid substitutions which decreased ADP-ribosyltransferase activity were constructed in the recombinant S-1 subunit. Substitution of Asp-11 by Ser, Arg-13 by Leu, or Trp-26 by Ile decreased enzyme activity to below detectable levels (less than 1.0% of that of the recombinant S-1 subunit). The Glu-139-to-Ser substitution reduced ADP-ribosyltransferase activity to 15% of that of the recombinant S-1 subunit. Both the oxidized and reduced forms of the recombinant S-1 subunit and recombinant S-1 subunits containing single-amino-acid substitutions were degraded through identical immunoreactive tryptic peptides, suggesting that the conformations of the mutants are similar to that of the recombinant S-1 subunit. Identification of noncatalytic forms of the S-1 subunit of pertussis toxin which have conserved protein structure is an initial step in the generation of a recombinant noncatalytic form of pertussis toxin which may be tested as a candidate for an acellular vaccine against Bordetella pertussis.

p21-ras effector domain mutants constructed by "cassette" mutagenesis

A series of mutations encoding single-amino-acid substitutions within the v-rasH effector domain were constructed, and the ability of the mutants to induce focal transformation of NIH 3T3 cells was studied. The mutations, which spanned codons 32 to 40, were made by a "cassette" mutagenesis technique that involved replacing this portion of the v-rasH effector domain with a linker carrying two BspMI sites in opposite orientations. Since BspMI cleaves outside its recognition sequence, BspMI digestion of the plasmid completely removed the linker, creating a double-stranded gap whose missing ras sequences were reconstructed as an oligonucleotide cassette. Based upon the ability of the mutants to induce focal transformation of NIH 3T3 cells, a range of phenotypes from virtually full activity to none (null mutants) was seen. Three classes of codons were present in this segment: one which could not be altered, even conservatively, without a loss of function (codons 32 and 35); one which retained detectable biologic activity with conservative changes but which lost function with more drastic substitutions (codons 36 and 40); and one which retained function even with a nonconservative substitution (codon 39).

DNA sequence determinants of CAP-induced bending and protein binding affinity

The sites of DNA bending induced by binding catabolite activator protein are identified and shown to coincide with positions where DNA grooves face the protein. The bendability of DNA with different sequences at these bend centres parallels the bending preference of the sequences in nucleosomal DNA. Anisotropic DNA bendability significantly affects the structure and strength of regulatory protein-DNA complexes.

Total chemical synthesis of a gene for hepatitis B virus core protein and its functional characterization

We have chemically synthesized a DNA duplex of 560 nucleotides that codes for the hepatitis B virus (HBV) core protein. The synthetic gene contains 27 unique internal restriction sites. Thereby, it can easily be mutagenized by replacement of rather short restriction fragments. A number of restriction recognition sequences are in common between the synthetic and the authentic gene, thus allowing for the transfer of synthetic segments into the cloned viral genome. Several unexpected mutations in the synthetic gene were readily corrected utilizing the multiple unique restriction sites. In Escherichia coli, the expression level of the synthetic gene product amounts to about 4% of the total soluble protein. It forms particles closely resembling native HBV cores. After transfer of the synthetic gene into the viral genome, transient expression in a hepatoma cell line yields proteins indistinguishable from the native gene products. The synthetic gene thus provides a useful tool for studies on the structure and function of the isolated HBV core protein as well as the gene and its various products in the viral life-cycle.

Dissecting the catalytic triad of a serine protease

Serine proteases are present in virtually all organisms and function both inside and outside the cell; they exist as two families, the 'trypsin-like' and the 'subtilisin-like', that have independently evolved a similar catalytic device characterized by the Ser, His, Asp triad, an oxyanion binding site, and possibly other determinants that stabilize the transition state (Fig. 1). For Bacillus amyloliquefaciens subtilisin, these functional elements impart a total rate enhancement of at least 10(9) to 10(10) times the non-enzymatic hydrolysis of amide bonds. We have examined the catalytic importance and interplay between residues within the catalytic triad by individual or multiple replacement with alanine(s), using site-directed mutagenesis of the cloned B. amyloliquefaciens subtilisin gene. Alanine substitutions were chosen to minimize unfavourable steric contacts and to avoid imposing new charge interactions or hydrogen bonds from the substituted side chains. In contrast to the effect of mutations in residues involved in substrate binding, the mutations in the catalytic triad greatly reduce the turnover number and cause only minor effects on the Michaelis constant. Kinetic analyses of the multiple mutants demonstrate that the residues within the triad interact synergistically to accelerate amide bond hydrolysis by a factor of approximately 2 X 10(6).

A simple and efficient procedure for generating random point mutations and for codon replacements using mixed oligodeoxynucleotides

A very simple and highly efficient procedure for the generation of single and multiple substitutions in segments of DNA is described which has no requirements for conveniently placed restriction sites, and allows all DNA sequences to be equally accessible. A mixed pool of oligodeoxynucleotides is synthesized by contaminating the monomeric nucleotides with low levels of the other three nucleotides such that the full-length oligonucleotide contains on the average one to two changes per molecule. This pool is used in priming in vitro synthesis of the complementary strand of cloned DNA fragments in M13 or pEMBL vectors which have previously been passed through a dut-, ung- Escherichia coli host. Strong selection for the newly synthesized strand is provided by transforming the heteroduplex into a dut+, ung+ host. Single and multiple substitutions in the carboxy-terminal coding region of the MATa1 gene of Saccharomyces cerevisiae are introduced at high efficiency (25-55%) and the changes are identified by direct sequencing alone. The same principle can be used to generate multiple sets of changes at any specified codon.

A knowledge-based system for cassette mutagenesis experimental design

A knowledge-based system for the design and planning of cassette mutagenesis experiments has been developed for scientists working in the field of structural biology and protein engineering. The system applies domain-specific knowledge to manage the menial details and automate most of the decision-making steps involved in the design process. This allows scientists to work at a high abstraction level, and results in significant time savings and increased productivity. The system also includes an automated documentation facility to improve the efficiency and accuracy of record keeping.

Efficient saturation mutagenesis of a pentapeptide coding sequence using mixed oligonucleotides

Site-directed mutagenesis using oligonucleotides that are degenerate at a specific codon was employed to construct a set of mutations in a pentapeptide sequence targeting cytosolic proteins to lysosomes during serum withdrawal. Low-temperature annealing of the mixed oligonucleotides to single-stranded phage DNA and a genetic selection for the DNA strand carrying the mutations were utilized. The use of mixed oligonucleotides by this technique provides an economical means of generating a large set of substitution mutations. A single codon can be changed to codons for most other amino acids in one step. This approach eliminates the need for restriction enzyme cleavage sites flanking the target for mutagenesis and, therefore, is useful for targeting mutations to any DNA fragment cloned into an appropriate single-stranded bacteriophage.

Recruitment of substrate-specificity properties from one enzyme into a related one by protein engineering

The Bacillus licheniformis and Bacillus amyloliquefaciens subtilisins differ by 31% in protein sequence and by factors of greater than 60 in catalytic efficiency, kcat/Km, toward various substrates. Despite large differences in sequence and substrate specificity for these serine proteases, only two amino acid substitutions (residues 156 and 217) occur within 4 A (contact distance) of modeled substrates, and a third substitution (residue 169) is within 7 A. The three B. licheniformis substitutions (Ser-156/Ala-169/Leu-217) were introduced into the wild-type B. amyloliquefaciens subtilisin (Glu-156/Gly-169/Tyr-217) by site-directed mutagenesis. The substrate specificity of the triple mutant approaches that of B. licheniformis enzyme when assayed with seven different substrates that vary in charge, size, and hydrophobicity. Thus, specificity properties of distantly related and functionally divergent enzymes can be exchanged by limited amino acid replacements, in this case representing less than 4% of the sequence differences.

Designing substrate specificity by protein engineering of electrostatic interactions

Protein engineering of electrostatic interactions between charged substrates and complementary charged amino acids, at two different sites in the substrate binding cleft of the protease subtilisin BPN', increases kcat/Km toward complementary charged substrates (up to 1900 times) and decreases kcat/Km toward similarly charged substrates. From kinetic analysis of 16 mutants of subtilisin and the wild type, the average free energies for enzyme-substrate ion-pair interactions at the two different sites are calculated to be -1.8 +/- 0.5 and -2.3 +/- 0.6 kcal/mol (1 cal = 4.18 J) [at 25 degrees C in 0.1 M Tris X HCl (pH 8.6)]. The combined electrostatic effects are roughly additive. These studies demonstrate the feasibility for rational design of charged ligand binding sites in proteins by tailoring of electrostatic interactions.

The use of a selectable FokI cassette in DNA replacement mutagenesis of the R388 dihydrofolate reductase gene

FokI, a class-IIS restriction endonuclease, cleaves double-stranded DNA to produce a protruding 5' end consisting of four nucleotides, 10-13 residues 3' from the nonpalindromic recognition sequence, GGATG. Cassettes which utilize this separation of cleavage and recognition site have been constructed for the purpose of linker mutagenesis and DNA replacement experiments. The cassettes are flanked by FokI recognition sequences oriented such that the FokI cleavage sites are several nucleotides beyond the cassette/vector fusion sites. FokI excises the cassette and several base pairs of the neighboring vector sequence. The ends produced in the vector by FokI cleavage are generally noncomplementary and suitable for the insertion of a segment of synthesized double-stranded replacement DNA. A cassette which contains a tyrosine tRNA suppressor gene (supF) is selectable by the suppression of amber mutations in the recipient host. A vector containing a pBR322-derived origin of replication, the Escherichia coli xanthine-guanine phosphoribosyl transferase gene as a selectable marker, and no FokI sites has been constructed for use with the FokI cassettes. An experiment which utilized the FokI/supF cassette to modify the N-terminal coding region of the R388 dihydrofolate reductase gene is described.

Expression of Bacillus amyloliquefaciens extracellular ribonuclease (barnase) in Escherichia coli following an inactivating mutation

An inactivated gene for Bacillus amyloliquefaciens extracellular ribonuclease (barnase) has previously been cloned and sequenced following transposon mutagenesis. The intact gene could not be assembled in Escherichia coli and is presumed to be lethal. Therefore, we introduced specific mutations into the barnase gene to prevent its lethal effect. A Gln-73 mutant gene was stable in E. coli but only produced low amounts of barnase antigen. Mutants containing Asp, Gln or Arg, instead of His-102, at the active site were identified by immunological screening for barnase antigen. None of the mutant proteins with alterations at aa residue 102 possessed RNase activity. The level of barnase (Asp-102) was higher in E. coli than in B. subtilis but the protein was not processed to the correct size in E. coli. To obtain correct processing, the barnase (Asp-102) structural gene was fused to the E. coli alkaline phosphatase promoter and signal sequence (phoA). Cells containing this construct secreted correctly processed barnase (Asp-102) into the periplasmic space and culture supernatant at a level of 20 mg/l. Barnase (Asp-102) was purified and found to have an identical N-terminus and a thermal unfolding curve that was nearly identical to that of active barnase (His-102). The cloning and expression of barnase in E. coli will allow detailed analysis of barnase protein folding by molecular genetic approaches.

Oligonucleotide-directed mutagenesis: a simple method using two oligonucleotide primers and a single-stranded DNA template

The important features of the protocol described here are as follows: First, the procedure consists of a few simple steps and results in a reasonably high frequency of mutagenesis. Second, using two primers, there is no need to isolate covalently closed double-stranded molecules as in our previous method. Third, the use of vectors derived from single-stranded phage facilitates template preparation, mutagenesis efficiency, screening, and DNA sequencing. Fourth, the same basic steps can be directly applied when using the single-stranded pUC derivatives.

Site-directed mutagenesis

Images

Secretion and autoproteolytic maturation of subtilisin

The sequence of the cloned Bacillus amyloliquefaciens subtilisin gene suggested that this secreted serine protease is produced as a larger precursor, designated preprosubtilisin [Wells, J. A., Ferrari, E., Henner, D. J., Estell, D. A. & Chen, E. Y. (1983) Nucleic Acids Res. 11, 7911-7925]. Biochemical evidence presented here shows that a subtilisin precursor is produced in Bacillus subtilis hosts. The precursor is first localized in the cell membrane, reaching a steady-state level of approximately equal to 1000 sites per cell. Mutations in the subtilisin gene that alter a catalytically critical residue (i.e., aspartate +32----asparagine), or delete the carboxyl-terminal portion of the enzyme that contains catalytically critical residues, block the maturation of this precursor. This block occurs when these mutant genes are expressed in B. subtilis hosts where the chromosomal subtilisin gene has been deleted. When the mutant B. amyloliquefaciens subtilisins are expressed in B. subtilis hosts that contain an intact chromosomal subtilisin gene, the mutant precursors are processed to a mature form and released to the medium. Such processing, in trans, of the precursor is also demonstrated in vitro by addition of active subtilisin. Thus, the release of subtilisin from the cell membrane is dependent on an autoproteolytic process that appears to be novel among secreted proteins.

Homoeo-domain homology in yeast MAT alpha 2 is essential for repressor activity

The MAT alpha locus of the yeast Saccharomyces cerevisiae encodes two regulatory proteins, alpha 1 and alpha 2, which are responsible for determining the alpha-cell type. MAT alpha 1 is a positive regulator of alpha-cell-type-specific genes, and MAT alpha 2 is a negative regulator of a-cell-type-specific genes. MAT alpha 2 also determines the a/alpha diploid cell type, in conjunction with the MATa product, a1, by repressing haploid cell-type-specific genes. The MAT alpha 2-encoded protein binds specifically in vitro to a DNA sequence found upstream of several a-specific genes and is thus thought to exert its control directly at the transcriptional level of target genes. In an initial attempt to understand the molecular basis of the interaction of alpha 2 with DNA, we have saturated with missense mutations the segment of alpha 2 that is weakly homologous to a conserved prokaryote DNA-binding structure and to a portion of the higher eukaryote homoeo domain to ascertain the possible functional significance of this homology in alpha 2. We report here that most of the amino-acid residues in alpha 2 which correspond to conserved amino acids in the prokaryote DNA-binding proteins and in the homoeo domain are essential for the two repressor activities of alpha 2, that is, the repression of a-specific genes and of haploid-specific genes. Mutations in a subset of these amino-acid residues more severely affect the ability to repress a-specific genes than haploid-specific genes.

Site-saturation studies of beta-lactamase: production and characterization of mutant beta-lactamases with all possible amino acid substitutions at residue 71

A mutagenic technique that "saturates" a particular site in a protein with all possible amino acid substitutions was used to study the role of residue 71 in beta-lactamase (EC 3.5.2.6). Threonine is conserved at residue 71 in all class A beta-lactamases and is adjacent to the active site Ser-70. All 19 mutants of the enzyme were characterized by the penam and cephem antibiotic resistance they provided to Escherichia coli LS1 cells. Surprisingly, cells producing any of 14 of the mutant beta-lactamases displayed appreciable resistance to ampicillin; only cells with mutants having Tyr, Trp, Asp, Lys, or Arg at residue 71 had no observable resistance to ampicillin. However, the mutants are less stable to cellular proteases than wild-type enzyme is. These results suggest that Thr-71 is not essential for binding or catalysis but is important for stability of the beta-lactamase protein. An apparent change in specificity indicates that residue 71 influences the region of the protein that accommodates the side chain attached to the beta-lactam ring of the substrate.

A complete library of point substitution mutations in the glucocorticoid response element of mouse mammary tumor virus

The glucocorticoid response element (GRE) of mouse mammary tumor virus (MMTV) was chemically synthesized as two complementary DNA strands bearing cohesive termini. During automated synthesis, random mutations were introduced into the DNA by "doping" each of the four nucleoside phosphoramidites (A, G, C, and T) with a low level of the other three. These preparations were annealed and cloned into an M13 phage vector to produce a library of GRE mutants. Mutations within the synthesized region were identified by sequencing phage isolates at random. All of the chemically distinct classes of transition and transversion mutations have been observed. Statistical considerations indicate that the library contains all of the possible 90 point substitution mutations within a 30-nucleotide mutagenic target. So far 88 of these substitutions have been isolated, 74 as single mutants. At least two of the three possible single mutants at each of the 30 positions have been identified.

Synthesis of mutant parathyroid hormone genes via site-specific recombination directed by crossover linkers

A synthetic 'crossover linker' technique has been designed for gene modification. The linker has a restriction end for an initial 'cohesive end' ligation with one terminus of a linearized plasmid, a middle section carrying modified sequence information, and an 'homology-searching' sequence of 20 bp at its other end, that is homologous to a specific region in the opposite terminus of the plasmid. Inside the Escherichia coli transformation host, intramolecular recombination between the homologous ends of the resultant plasmid intermediate completes the integration of the linker. Using different crossover linkers, a human parathyroid hormone gene which had previously been cloned into plasmid pUC8 was converted to mutant coding sequences via specific base substitution, sequence deletion and sequence insertion.