Programming the Drosophila embryo 2: from genotype to phenotype.

Although development is a single hierarchical process, scientists tend to study only one level at a time: molecular, cellular, or organismal. The data and theory are available to integrate molecular, cellular, and organismal levels into a series of maps for development of the Drosophila embryo. These maps link the transcriptional cascade with mitotic and phenotypic fate maps to trace hierarchical mechanisms of development from the genotype in the egg to the phenotype in the larva.

Deciphering the language of the genome.

The non-coding DNA in eukaryotic genomes encodes a language which programs organismal growth and development. We show that a linguistic and cryptographic approach can be used to deduce the syntax of this programming language for gene regulation and to compile a dictionary of enhancers which form its words.

Programming the Drosophila embryo.

A critical step in understanding the mechanisms of development is in defining the steps at the molecular, cellular, and organismal levels in the developmental program for a given organism-so that given the egg one can predict not only how the embryo will develop but also how that embryo evolved from its ancestors. Using methods employed by chemists and engineers in modeling hierarchical systems, I have integrated current theory and experiment into a calculational method that can model early Drosophila embryogenesis on a personal computer. This quantitative calculation tool is simple enough to be useful for experimentalists in designing experiments yet detailed enough for theoreticians to derive new insights on the evolution of developmental genetic networks. By integrating the strengths of theoretical and experimental methods into a single engineering model that can compute the cascade of genetic networks in a real organism, I provide a new calculational tool that can apply current theory to current experimental data to study the evolution of developmental programs.

A chromatin switch.

Cellular and molecular biological approaches that study eukaryotic gene regulation have led to separate models which describe structure and mechanism with differing precision. Using principles of combinatorial and cooperative interactions inherent in both models, we have extended concepts derived from a "genetic switch" for prokaryotes to a chromatin switch for eukaryotes composed of DNA, transactivators, nucleosomes and the nuclear matrix. We present a consensus model for gene regulation that uses a simple Monte Carlo method for simulating condensation and extension of chromatin. Such a chromatin switch can be modulated by known biochemical and molecular modifications, and the transactivator binding sites or enhancers within DNA domains can be organized into a hierarchy to control cell cycling and differentiation.

A nuclear matrix attachment region organizes the Epstein-Barr viral plasmid in Raji cells into a single DNA domain.

The extrachromosomal Epstein-Barr virus (EBV) plasmid in the Burkitt lymphoma cell line, Raji, is stably associated with the nuclear matrix. This association is effected by a nuclear matrix attachment region (MAR) located in the BamHI C fragment of the viral genome; no other region of EBV DNA was found to be attached to the nuclear matrix with high affinity. The MAR was mapped to 5.2 kbp of DNA, greater than 80% of which is found on the nuclear matrix in unsynchronized cells expressing only viral latent cycle products. Thus the majority of viral plasmids in Raji cells use the same MAR. The MAR of EBV DNA contains the origin of latent viral DNA replication (oriP), the genes for the small viral RNAs (EBERs) and a 500 bp region immediately upstream of the EBER-1 gene. The clustering of the latent viral replication origin and the nearby enhancer and promoters for latent viral transcription on the nuclear matrix is likely to be crucial for regulation of the latent viral genome.

Replication of minute virus of mice DNA in adenovirus-infected or adenovirus-transformed cells.

The effect of adenovirus infection or transformation on the DNA replication of Minute Virus of Mice (MVM) was studied in human fibroblast cell lines. In WI38, HeLa, and 293 cells MVM infection allowed production of viral NS-1 and capsid proteins with or without adenovirus 2 (Ad2) co-infection. However, MVM DNA replication varied markedly. In HeLa cells MVM DNA was replicated weakly in host nucleoli, and replication was increased markedly by Ad2 co-infection as well as recompartmentalized to Ad2 replication factories. In Ad-transformed 293 cells MVM DNA was replicated very efficiently when infected alone or with Ad2 co-infection although recompartmentalization from nucleoli to replication factories was also seen. In WI38 cells MVM DNA was not replicated under any conditions. The variation in DNA replication in WI38, HeLa, and 293 cells despite viral protein production in all cases suggests that MVM DNA replication is uncoupled from viral gene expression and that host factors required for MVM DNA replication are induced or recompartmentalized by adenovirus infection or transformation.

Adenovirus and minute virus of mice DNAs are localized at the nuclear periphery.

The localization of adenovirus 2 (Ad2) and Minute Virus of Mice (MVM) DNAs was studied in situ in infected HeLa cell nuclei using fluorescent DNA probes and confocal microscopy. Ad2 DNA was found in multiple foci which were localized along the periphery of the infected cell nuclei. MVM DNA was found in HeLa cell nucleoli which are associated with the nuclear envelope, and when co-infected with Ad2 MVM DNA was compartmentalized to multiple foci which again were localized at the nuclear periphery. The data are discussed in terms of a model for the role of intranuclear compartmentalization in eukaryotic DNA structure and function.

The terminal regions of adenovirus and minute virus of mice DNAs are preferentially associated with the nuclear matrix in infected cells.

The interaction of viral genomes with the cellular nuclear matrix was studied by using adenovirus-infected HeLa cells and minute virus of mice (MVM)-infected A-9 cells. Adenovirus DNA was associated with the nuclear matrix both early and late in infection, the tightest interaction being with DNA fragments that contain the covalently bound 5'-terminal protein. Replicative forms of MVM DNA were also found to be exclusively matrix associated during the first 16 to 20 h of infection; at later times viral DNA species accumulated in the soluble nuclear fraction at different rates, suggesting a saturation of nuclear matrix-binding sites. MVM DNA fragments enriched in the matrix fraction were also derived from the terminal regions of the viral genome. However, only the subset of fragments which possess a covalently bound 5'-terminal protein (i.e., DNA fragments in which the 5' palindromic DNA sequences are in the extended duplex rather than the hairpin conformation) were matrix associated. These observations suggest that the DNA-matrix interactions are, at least in part, mediated by the viral terminal proteins. Since these proteins have previously been shown to be intimately involved in viral DNA replication, our results further indicate that an association with the nuclear matrix may be important for viral genome replication and possibly also for efficient gene transcription.

Interactions of minute virus of mice and adenovirus with host nucleoli.

Biochemical evidence is presented that both minute virus of mice (MVM) and adenovirus interact with the nucleolus during lytic growth and that MVM can also target specific changes involving nucleolar components in adenovirus-infected cells. These virus-nucleolus interactions were studied by analysis of intranuclear compartmentalization of both viral DNAs and host nucleolar proteins: (i) MVM in mouse cells (its normal host) replicates its DNA in the host nucleoli; (ii) specific nucleolar proteins as well as small nuclear ribonucleoprotein antigens are recompartmentalized to multiple intranuclear foci in adenovirus-infected HeLa cells; and (iii) when adenovirus helps MVM DNA replication in a nonpermissive human cell (HeLa), the MVM DNA is also recompartmentalized for synthesis. The data suggest mechanisms for disruption of nucleolar function common to oncogenic or oncolytic virus lytic growth and cell transformation.

The domain model for eukaryotic DNA organization. 2: A molecular basis for constraints on development and evolution.

A model for eukaryotic DNA organization has been proposed in which DNA regulatory processes depend on multiple site-specific DNA-nuclear matrix interactions throughout a DNA domain. In this model gene regulation depends on combinations of a few control factors in a cell to activate cell type-specific genes. This model suggests simple molecular mechanisms for organismal development which can account for sequential activation of appropriate groups of genes throughout development and for specific constraints on developmental pathways. Additionally, these suggested developmental pathways are consistent with mechanisms of evolution in which gradualism and punctuated equilibrium are not exclusive of one another and are interrelated mechanisms of evolution that are both induced by specific chromosomal mutations.

Sequence organization in regulatory regions of DNA of minute virus of mice.

Analysis of the nucleotide sequence of minute virus of mice (MVM) DNA indicates that the DNA termini contain clusters of potential DNA regulatory elements and that there are repetitive DNA elements highly reiterated throughout the entire genome, which may also have a role in DNA function. The left end of MVM DNA, which contains the promoter for the nonstructural genes, has a cluster of DNA elements that includes homologies to the polyoma virus enhancer, three copies of an E1A-inducible transcription factor (ATF) binding site, and a potential Z-DNA element. The MVM right end, which contains the origin of DNA replication, has a cluster of DNA elements that includes several homologies to the polyoma virus replication origin and a potential Z-DNA element. In addition, oligonucleotide frequency analysis indicates the presence of highly recurring sequence elements throughout the entire MVM genome that may be involved in regulation. This computer-aided analysis suggests similarities and significant differences in regulatory sequence organization between MVM and polyoma virus, and identifies specific DNA elements for future genetic characterization.

A domain model for eukaryotic DNA organization: a molecular basis for cell differentiation and chromosome evolution.

A model for eukaryotic chromatin organization is presented in which the basic structural and functional unit is the DNA domain. This simple model predicts that both chromosome replication and cell type-specific control of gene expression depend on a combination of stable and dynamic DNA-nuclear matrix interactions. The model suggests that in eukaryotes, DNA regulatory processes are controlled mainly by the intranuclear compartmentalization of the specific DNA sequences, and that control of gene expression involves multiple steps of specific DNA-nuclear matrix interactions. Predictions of the model are tested using available biochemical, molecular and cell biological data. In addition, the domain model is discussed as a simple molecular mechanism to explain cell differentiation in multi-cellular organisms and to explain the evolution of eukaryotic genomes consisting mainly of repetitive sequences and "junk" DNA.

Highly recurring sequence elements identified in eukaryotic DNAs by computer analysis are often homologous to regulatory sequences or protein binding sites.

We have used computer assisted dot matrix and oligonucleotide frequency analyses to identify highly recurring sequence elements of 7-11 base pairs in eukaryotic genes and viral DNAs. Such elements are found much more frequently than expected, often with an average spacing of a few hundred base pairs. Furthermore, the most abundant repetitive elements observed in the ovalbumin locus, the beta-globin gene cluster, the metallothionein gene and the viral genomes of SV40, polyoma, Herpes simplex-1 and Mouse Mammary Tumor Virus were sequences shown previously to be protein binding sites or sequences important for regulating gene expression. These sequences were present in both exons and introns as well as promoter regions. These observations suggest that such sequences are often highly overrepresented within the specific gene segments with which they are associated. Computer analysis of other genetic units, including viral genomes and oncogenes, has identified a number of highly recurring sequence elements that could serve similar regulatory or protein-binding functions. A model for the role of such reiterated sequence elements in DNA organization and function is presented.

Identification and characterization of a protein covalently bound to DNA of minute virus of mice.

We identified a protein which is covalently linked to a fraction of the DNA synthesized in cells infected with minute virus of mice. This protein is specifically bound to the 5' terminus of the extended terminal conformers of the minute virus of mice replicative-form DNA species and of a variable fraction of single-stranded viral DNA. The chemical stability of the protein-DNA linkage is characteristic of a phosphodiester bond between a tyrosine residue in the protein and the 5' end of the DNA. The terminal protein (TP) bound on all DNA forms has a relative molecular weight of 60,000; it is also seen free in extracts from infected cells. Immunologic comparison of the TP with the other known viral proteins suggests that the TP is not related to the capsid proteins or NS-1.

Effect of nucleotide analogs on the cleavage of DNA by the restriction enzymes AluI, DdeI, HinfI, RsaI, and TaqI.

The cleavage of specific DNA sequences by the restriction endonucleases AluI, DdeI, HinfI, RsaI, and TaqI has been studied by monitoring the effect of various nucleotide modifications on the rate of DNA digestion. Bacteriophage fd DNA was completely substituted in one strand with a single nucleotide analog, using an in vitro primed DNA synthesis reaction on a single-stranded viral DNA template. Twelve deoxynucleotide analogs were incorporated into these DNA substrates: 2-aminopurine, 2,6-diaminopurine, deoxytubercidin, deoxyuridine, 5-bromodeoxyuridine, 5-allylamine deoxyuridine, 5-biotinyl deoxyuridine, deoxypseudouridine, deoxyinosine, 8-azadeoxyguanosine, 5-iododeoxycytidine, and 5-bromodeoxycytidine. The restriction enzymes tested varied considerably in their ability to digest hemi-substituted DNAs containing these modified nucleotides. Structural alterations in the base pairs immediately adjacent to the phosphodiester bonds cleaved by the enzyme reduced the rate of enzyme activity most dramatically, and in most cases more than a single determinant on each base pair altered activity. Interactions with nucleotides outside the recognition site seem to have little importance in the binding or catalytic activity of these enzymes.

Proteins tightly bound to HeLa cell DNA at nuclear matrix attachment sites.

DNA-protein complexes have been isolated from HeLa cell nuclei and nuclear matrix preparations. Two proteins, 55 and 66 kilodaltons in size, remain bound to HeLa DNA after treatment at 80 degrees C in 2% sodium dodecyl sulfate and purification by exclusion chromatography on Sepharose 2B-CL in the presence of 0.3% sodium dodecyl sulfate. These proteins appear to be tightly bound but not covalently linked to the DNA, and they are distributed over the DNA with an average spacing of 40 kilobase pairs. This spacing distribution remains essentially constant throughout the cell cycle. The proteins are bound to the residual 2% of HeLa cell DNA which remains attached to the nuclear matrix after extensive nuclease digestion, a condition which reduces the average size of the DNA to approximately 150 base pairs. Our results suggest that these tightly bound proteins are involved in anchoring cellular DNA to the nuclear matrix. These tightly bound proteins are identical by partial peptide mapping to proteins found tightly bound to the DNA of mammalian, plant, and bacterial cells (D. Werner and C. Petzelt, J. Mol. Biol. 150:297-302, 1981), implying that these proteins are involved in the organization of chromosomal domains and are highly conserved in both procaryotic and eucaryotic cells.