Mutational properties of the primary aflatoxin B1-DNA adduct.

The mutagenic activity of the major DNA adduct formed by the liver carcinogen aflatoxin B1 (AFB1) was investigated in vivo. An oligonucleotide containing a single 8,9-dihydro-8-(N7-guanyl)-9-hydroxyaflatoxin B1 (AFB1-N7-Gua) adduct was inserted into the single-stranded genome of bacteriophage M13. Replication in SOS-induced Escherichia coli yielded a mutation frequency for AFB1-N7-Gua of 4%. The predominant mutation was G --> T, identical to the principal mutation in human liver tumors believed to be induced by aflatoxin. The G --> T mutations of AFB1-N7-Gua, unlike those (if the AFB1-N7-Gua-derived apurinic site, were much more strongly dependent on MucAB than UmuDC, a pattern matching that in intact cells treated with the toxin. It is concluded that the AFB1-N7-Gua adduct, and not the apurinic site, has genetic requirements for mutagenesis that best explain mutations in aflatoxin-treated cells. While most mutations were targeted to the site of the lesion, a significant fraction (13%) occurred at the base 5' to the modified guanine. In contrast, the apurinic site-containing genome gave rise only to targeted mutations. The mutational asymmetry observed for AFB1-N7-Gua is consistent with structural models indicating that the aflatoxin moiety of the aflatoxin guanine adduct is covalently intercalated on the 5' face of the guanine residue. These results suggest a molecular mechanism that could explain an important step in the carcinogenicity of aflatoxin B1.

Sequence scanning: A method for rapid sequence acquisition from large-fragment DNA clones.

A strategy of "sequence scanning" is proposed for rapid acquisition of sequence from clones such as bacteriophage P1 clones, cosmids, or yeast artificial chromosomes. The approach makes use of a special vector, called LambdaScan, that reliably yields subclones with inserts in the size range 8-12 kb. A number of subclones, typically 96 or 192, are chosen at random, and the ends of the inserts are sequenced using vector-specific primers. Then long-range spectrum PCR is used to order and orient the clones. This combination of shotgun and directed sequencing results in a high-resolution physical map suitable for the identification of coding regions or for comparison of sequence organization among genomes. Computer simulations indicate that, for a target clone of 100 kb, the scanning of 192 subclones with sequencing reads as short as 350 bp results in an approximate ratio of 1:2:1 of regions of double-stranded sequence, single-stranded sequence, and gaps. Longer sequencing reads tip the ratio strongly toward increased double-stranded sequence.

Extending the chemistry that supports genetic information transfer in vivo: phosphorothioate DNA, phosphorothioate RNA, 2'-O-methyl RNA, and methylphosphonate DNA.

DNA and RNA are the polynucleotides known to carry genetic information in life. Chemical variants of DNA and RNA backbones have been used in structure-function and biosynthesis studies in vitro, and in antisense pharmacology, where their properties of nuclease resistance and enhanced cellular uptake are important. This study addressed the question of whether the base(s) attached to artificial backbones encodes genetic information that can be transferred in vivo. Oligonucleotides containing chemical variants of DNA or RNA were used as primers for site-specific mutagenesis of bacteriophage f1. Progeny phage were scored both genetically and physically for the inheritance of information originally encoded by bases attached to the nonstandard backbones. Four artificial backbone chemistries were tested: phosphorothioate DNA, phosphorothioate RNA, 2'-O-methyl RNA and methylphosphonate DNA. All four were found capable of faithful information transfer from their attached bases when one or three artificial positions were flanked by normal DNA. Among oligonucleotides composed entirely of nonstandard backbones, only phosphorothioate DNA supported genetic information transfer in vivo.

Transcribing of Escherichia coli genes with mutant T7 RNA polymerases: stability of lacZ mRNA inversely correlates with polymerase speed.

When in Escherichia coli the host RNA polymerase is replaced by the 8-fold faster bacteriophage T7 enzyme for transcription of the lacZ gene, the beta-galactosidase yield per transcript drops as a result of transcript destabilization. We have measured the beta-galactosidase yield per transcript from T7 RNA polymerase mutants that exhibit a reduced elongation speed in vitro. Aside from very slow mutants that were not sufficiently processive to transcribe the lacZ gene, the lower the polymerase speed, the higher the beta-galactosidase yield per transcript. In particular, a mutant which was 2.7-fold slower than the wild-type enzyme yielded 3.4- to 4.6-fold more beta-galactosidase per transcript. These differences in yield vanished in the presence of the rne-50 mutation and therefore reflect the unequal sensitivity of the transcripts to RNase E. We propose that the instability of the T7 RNA polymerase transcripts stems from the unmasking of an RNase E-sensitive site(s) between the polymerase and the leading ribosome: the faster the polymerase, the longer the lag between the synthesis of this site(s) and its shielding by ribosomes, and the lower the transcript stability.

Analysis of RNA-binding proteins by in vitro genetic selection: identification of an amino acid residue important for locking U1A onto its RNA target.

An in vitro genetic system was developed as a rapid means for studying the specificity determinants of RNA-binding proteins. This system was used to investigate the origin of the RNA-binding specificity of the mammalian spliceosomal protein U1A. The U1A domain responsible for binding to U1 small nuclear RNA was locally mutagenized and displayed as a combinatorial library on filamentous bacteriophage. Affinity selection identified four U1A residues in the mutagenized region that are important for specific binding to U1 hairpin II. One of these residues (Leu-49) disproportionately affects the rates of binding and release and appears to play a critical role in locking the protein onto the RNA. Interestingly, a protein variant that binds more tightly than U1A emerged during the selection, showing that the affinity of U1A for U1 RNA has not been optimized during evolution.

Phage display selection of ligand residues important for Src homology 3 domain binding specificity.

The Src homology 3 (SH3) domain is a 50-aa modular unit present in many cellular proteins involved in intracellular signal transduction. It functions to direct protein-protein interactions through the recognition of proline-rich motifs on associated proteins. SH3 domains are important regulatory elements that have been demonstrated to specify distinct regulatory pathways important for cell growth, migration, differentiation, and responses to the external milieu. By the use of synthetic peptides, ligands have been shown to consist of a minimum core sequence and to bind to SH3 domains in one of two pseudosymmetrical orientations, class I and class II. The class I sites have the consensus sequence ZP(L/P)PP psi P whereas the class II consensus is PP psi PPZ (where psi is a hydrophobic residue and Z is a SH3 domain-specific residue). We previously showed by M13 phage display that the Src, Fyn, Lyn, and phosphatidylinositol 3-kinase (PI3K) SH3 domains preferred the same class I-type core binding sequence, RPLPP psi P. These results failed to explain the specificity for cellular proteins displayed by SH3 domains in cells. In the current study, class I and class II core ligand sequences were displayed on the surface of bacteriophage M13 with five random residues placed either N- or C-terminal of core ligand residues. These libraries were screened for binding to the Src, Fyn, Lyn, Yes, and PI3K SH3 domains. By this approach, additional ligand residue preferences were identified that can increase the affinity of SH3 peptide ligands at least 20-fold compared with core peptides. The amino acids selected in the flanking sequences were similar for Src, Fyn, and Yes SH3 domains; however, Lyn and PI3K SH3 domains showed distinct binding specificities. These results indicate that residues that flank the core binding sequences shared by many SH3 domains are important determinants of SH3 binding affinity and selectivity.

Human white blood cells contain cyclobutyl pyrimidine dimer photolyase.

Although enzymatic photoreactivation of cyclobutyl pyrimidine dimers in DNA is present in almost all organisms, its presence in placental mammals is controversial. We tested human white blood cells for photolyase by using three defined DNAs (supercoiled pET-2, nonsupercoiled bacteriophage lambda, and a defined-sequence 287-bp oligonucleotide), two dimer-specific endonucleases (T4 endonuclease V and UV endonuclease from Micrococcus luteus), and three assay methods. We show that human white blood cells contain photolyase that can photorepair pyrimidine dimers in defined supercoiled and linear DNAs and in a 287-bp oligonucleotide and that human photolyase is active on genomic DNA in intact human cells.

A variant of lambda repressor with an altered pattern of cooperative binding to DNA sites.

The bacteriophage lambda repressor binds cooperatively to pairs of adjacent sites in the lambda chromosome, one repressor dimer binding to each site. The repressor's amino domain (that which mediates DNA binding) is connected to its carboxyl domain (that which mediates dimerization and the interaction between dimers) by a protease-sensitive linker region. We have generated a variant lambda repressor that lacks this linker region. We show that dimers of the variant protein are deficient in cooperative binding to sites at certain, but not all, distances. The linker region thus extends the range over which carboxyl domains of DNA-bound dimers can interact. In particular, the linker is required for cooperative binding to a pair of sites as found in the lambda chromosome, and thus is essential for the repressor's physiological function.

Efficient recovery of infectious vesicular stomatitis virus entirely from cDNA clones.

Infectious vesicular stomatitis virus (VSV), the prototypic nonsegmented negative-strand RNA virus, was recovered from a full-length cDNA clone of the viral genome. Bacteriophage T7 RNA polymerase expressed from a recombinant vaccinia virus was used to drive the synthesis of a genome-length positive-sense transcript of VSV from a cDNA clone in baby hamster kidney cells that were simultaneously expressing the VSV nucleocapsid protein, phosphoprotein, and polymerase from separate plasmids. Up to 10(5) infectious virus particles were obtained from transfection of 10(6) cells, as determined by plaque assays. This virus was amplified on passage, neutralized by VSV-specific antiserum, and shown to possess specific nucleotide sequence markers characteristic of the cDNA. This achievement renders the biology of VSV fully accessible to genetic manipulation of the viral genome. In contrast to the success with positive-sense RNA, attempts to recover infectious virus from negative-sense T7 transcripts were uniformly unsuccessful, because T7 RNA polymerase terminated transcription at or near the VSV intergenic junctions.

Cre-mediated site-specific translocation between nonhomologous mouse chromosomes.

Chromosome rearrangements, such as large deletions, inversions, or translocations, mediate migration of large DNA segments within or between chromosomes, which can have major effects on cellular genetic control. A method for chromosome manipulation would be very useful for studying the consequences of large-scale DNA rearrangements in mammalian cells or animals. With the use of the Cre-loxP recombination system of bacteriophage P1, we induced a site-specific translocation between the Dek gene on chromosome 13 and the Can gene on chromosome 2 in mouse embryonic stem cells. The estimated frequency of Cre-mediated translocation between the nonhomologous mouse chromosomes is approximately 1 in 1200-2400 embryonic stem cells expressing Cre recombinase. These results demonstrate the feasibility of site-specific recombination systems for chromosome manipulation in mammalian cells in vivo, breaking ground for chromosome engineering.

Operator binding by lambda repressor heterodimers with one or two N-terminal arms.

The first 6 amino acids (NH2-Ser1-Thr2-Lys3-Lys4-Lys5-Pro6) of bacteriophage lambda cI repressor form a flexible arm that wraps around the operator DNA. Homodimeric lambda repressor has two arms. To determine whether both arms are necessary or only one arm is sufficient for operator binding, we constructed heterodimeric repressors with two, one, or no arms by fusing the DNA binding domain of lambda repressor to leucine zippers from Fos and Jun. Although only one arm is visible in the cocrystal structure of the N-domain-operator complex, our results indicate that both arms are required for optimal operator binding and normal site discrimination.

Stringent chemical and thermal regulation of recombinant gene expression by vaccinia virus vectors in mammalian cells.

We developed a stringently regulated expression system for mammalian cells that uses (i) the RNA polymerase, phi 10 promoter, and T phi transcriptional terminator of bacteriophage T7; (ii) the lac repressor, lac operator, rho-independent transcriptional terminators and the gpt gene of Escherichia coli; (iii) the RNA translational enhancer of encephalomyocarditis virus; and (iv) the genetic background of vaccinia virus. In cells infected with the recombinant vaccinia virus, reporter beta-galactosidase synthesis was not detected in the absence of inducer. An induction of at least 10,000- to 20,000-fold occurred upon addition of isopropyl beta-D-thiogalactopyranoside or by temperature elevation from 30 to 37 degrees C using a temperature-sensitive lac repressor. Regulated synthesis of the secreted and highly glycosylated human immunodeficiency virus 1 envelope protein gp120 was also demonstrated. Yields of both proteins were approximately 2 mg per 10(8) cells in 24 hr. Plasmid transfer vectors for cloning and expression of complete or incomplete open reading frames in recombinant vaccinia viruses are described.

The DnaJ chaperone catalytically activates the DnaK chaperone to preferentially bind the sigma 32 heat shock transcriptional regulator.

In Escherichia coli the heat shock response is under the positive control of the sigma 32 transcription factor. Three of the heat shock proteins, DnaK, DnaI, and GrpE, play a central role in the negative autoregulation of this response at the transcriptional level. Recently, we have shown that the DnaK and DnaJ proteins can compete with RNA polymerase for binding to the sigma 32 transcription factor in the presence of ATP, by forming a stable DnaJ-sigma 32-DnaK protein complex. Here, we report that DnaJ protein can catalytically activate DnaK's ATPase activity. In addition, DnaJ can activate DnaK to bind to sigma 32 in an ATP-dependent reaction, forming a stable sigma 32-DnaK complex. Results obtained with two DnaJ mutants, a missense and a truncated version, suggest that the N-terminal portion of DnaJ, which is conserved in all family members, is essential for this activation reaction. The activated form of DnaK binds preferentially to sigma 32 versus the bacteriophage lambda P protein substrate.

A single residue in DNA polymerases of the Escherichia coli DNA polymerase I family is critical for distinguishing between deoxy- and dideoxyribonucleotides.

Bacteriophage T7 DNA polymerase efficiently incorporates a chain-terminating dideoxynucleotide into DNA, in contrast to the DNA polymerases from Escherichia coli and Thermus aquaticus. The molecular basis for this difference has been determined by constructing active site hybrids of these polymerases. A single hydroxyl group on the polypeptide chain is critical for selectivity. Replacing tyrosine-526 of T7 DNA polymerase with phenylalanine increases discrimination against the four dideoxynucleotides by > 2000-fold, while replacing the phenylalanine at the homologous position in E. coli DNA polymerase I (position 762) or T. aquaticus DNA polymerase (position 667) with tyrosine decreases discrimination against the four dideoxynucleotides 250- to 8000-fold. These mutations allow the engineering of new DNA polymerases with enhanced properties for use in DNA sequence analysis.

Structural variety of arginine-rich RNA-binding peptides.

Arginine-rich domains are used by a variety of RNA-binding proteins to recognize specific RNA hairpins. It has been shown previously that a 17-aa arginine-rich peptide from the human immunodeficiency virus Rev protein binds specifically to its RNA site when the peptide is in an alpha-helical conformation. Here we show that related peptides from splicing factors, viral coat proteins, and bacteriophage antiterminators (the N proteins) also have propensities to form alpha-helices and that the N peptides require helical conformations to bind to their cognate RNAs. In contrast, introducing proline mutations into the arginine-rich domain of the human immunodeficiency virus Tat protein abolishes its potential to form an alpha-helix but does not affect RNA-binding affinity in vitro or in vivo. Based on results from several peptide-RNA model systems, we suggest that helical peptides may be used to recognize RNA structures having particularly wide major grooves, such as those found near loops or large bulges, and that nonhelical or extended peptides may be used to recognize less accessible grooves.