Selective degradation of integrated murine leukemia proviral DNA by deoxyribonucleases

Active gene sequences are undermethylated

The degree of methylation of active regions of the chromosome has been investigated by several techniques. DNase I (deoxyribonuclease I, EC 3.1.21.1) was used to introduce nicks in the active regions of the nucleus and thereby specifically label these areas. By using the methylation-specific restriction enzymes Hpa II and Hha I it could be shown that active genes are more sensitive to these probes than are other parts of the genome. In order to measure the amount of methylation at all CpG residues, DNA was nick-translated in the presence of [alpha-32P]dGTP as the sole nucleotide source and the methylated cytosine was detected by the standard nearest-neighbor analysis. Using this assay, we found that about 70% of all CpG sequences in animal cell DNA are methylated. In active nuclear regions that are sensitive to DNase I, only 30-40% of the CpG residues are methylated. This method was also employed to study the gene sequences that are complementary to cellular RNA. By this criterion expressed gene sequences are only 20-30% methylated. These data suggest that undermethylation is a general phenomenon in all actively transcribed genes.

Methylation of chromosomal DNA by two alkylating agents differing in carcinogenic potential

The affinity of 2 methylating agents, N-methyl-N-nitrosourea (MNU), a potent carcinogen, and dimethyl sulfate (DMS), a very weak or non-carcinogen, for specific structural or functional regions of DNA has been studied in an in vitro system using rat liver nuclei. The release of alkylated nucleotides from nuclei following limited nuclease digestion was measured. Under restrictive conditions, pancreatic DNase I preferentially digests DNA sequences active in RNA transcription while micrococcal nuclease digests spacer DNA between nucleosome cores. Nucleotides methylated by methylnitrosourea were preferentially released early during the digestions, suggesting a localization in both actively transcribing regions and spacer DNA. DMS alkylation, on the other hand, showed a random distribution in chromosomal DNA as measured by micrococcal nuclease and only limited accumulation in transcribing DNA.

Chromatin fine structure of active and repressed genes

Study of the structural organization of chromatin during transcription and replication may reveal important aspects of these processes. At the lowest level of organization, chromatin consists of a repeating subunit, the nucleosome (for reviews see refs 1-3). Electron microscopy indicates that the nucleosomes are arranged helically or form discrete superbeads, generating the familiar 250 A-300-A fibre. It has been suggested that this fibre is further folded into loops containing up to several hundred nucleosomes. Despite extensive study, the significance and fate of these nucleosomes remain obscure. We have used here micrococcal nuclease digestion to compare the structures of actively transcribing and inert chromatin of the genes coding for the major heat-shock protein of Drosophila melanogaster. The repressed hsp 70 genes were considerably more resistant to cleavage by micrococcal nuclease than their flanking regions and the bulk of chromatin. The active genes, previously shown to be more sensitive than the repressed genes, are also more susceptible to the nuclease than their 3'-flanking regions and bulk chromatin.

Nuclease sensitivity of active chromatin

The active regions of chicken erythrocyte nuclei were labeled using the standard DNase I directed nick translation reaction. These nuclei were then used to study the characteristics and, in particular, the nuclease sensitivity of active genes. Although DNase I specifically attacks active genes, micrococcal nuclease solubilizes these regions to about the same degree as the total DNA. On the other hand micrococcal nuclease does selectively cut the internucleosomal regions of active genes resulting in the appearance of mononucleosomal fraction which is enriched in active gene DNA. A small percentage of the active chromatin is also released from the nucleus by low speed centrifugation following micrococcal nuclease treatment. The factors which make active genes sensitive to DNase I were shown to reside on individual nucleosomes from these regions. This was established by showing that isolated active mononucleosomes were preferentially sensitive to DNase I digestion. Although the high mobility group proteins are essential for the maintenance of DNase I sensitivity in active regions, these proteins are not necessary for the formation of the conformation which makes these genes preferentially accessible to micrococcal nuclease. The techniques employed in this paper enable one to study the chromatin structure of the entire population of actively expressed genes. Previous studies have elucidated the structure of a few special highly prevalent genes such as ovalbumin and hemoglobin. The results of this paper show that this special conformation is a general feature of all active genes irregardless of the extent of expression.

The extraction by micrococcal nuclease of glucocorticoid receptors and mouse mammary tumor virus DNA sequences is dissociated

Glucocorticoid receptors (RG) and mammary tumor virus (MM-TV) DNA sequences were extracted by micrococcal nuclease digestion from the nuclei of C3H mouse mammary tumor cells in order to specify their relative distribution in chromatin. RG was labelled and translocated into the nuclei by incubating cells with 3H Dexamethasone (3H Dex). The purified nuclei were then treated at 2 degrees C with micrococcal nuclease. Three chromatin fractions were successively obtained: an isotonic extract (ne3H1), ahypotonic extract (ne2) and the residual pellet (P). The Dex-RG complexes were measured by the hydroxyapatite technique. The MMTV DNA sequences were titrated by molecular hybridization with an excess of MMTV radioactive cDNA probe. Up to 75% of the nuclear 3H Dex and the MMTV radioactive cDNA probe. Up to 75% of the nuclear 3H Dex and MMTV DNA sequences were extracted in a concentration dependent manner while only 10-15% of nucleic acids became soluble in 10% perchloric acid. The extracted 3H Dex-RG complex was found to be partly bound to soluble chromatin and partly free. The free complex displayed similar sedimentation constants (4S, 7S) and DNA binding ability to the cytosol receptor. The 3H Dex-RG complexes were 2 to 8 fold more concentrated in ne1, which is known to be enriched in active chromatin, than in ne2. Conversely, the concentration of MMTV DNA sequences per microgram DNA was the same in the three nuclear fractions. These results suggest that the Dex-RG complexes are concentrated in an active fraction of chromatin. We propose that, among the 20-30 copies of MMTV genes per haploid genome, only a small proportion are transcribed or regulated.

Subnuclear fractionation by mild micrococcal-nuclease treatment of nuclei of different transcriptional activities causes a partition of expressed and non-expressed genes

Extremely mild treatment with micrococcal nuclease of isolated nuclei yields subnuclear fractions in which the majority of RNA polymerase II transcriptional complexes formed in vivo are segregated [Tata & Baker (1978) J. Mol. Biol. 118, 249-272]. We now describe different approaches followed to established whether or not the nuclei are thus resolved into transcribed and non-transcribed DNA. First, we have compared the sensitivity to deoxyribonuclease I, which is known to digest preferably expressed genes as present in nuclei or chromatin, of three micrococcal-nuclease-derived fractions from nuclei of different transcriptional activities. In transcriptionally active nuclei (rat liver, hen liver and oviduct, and Xenopus liver), the DNA in a polynucleosomal fraction comprising 6-15% of DNA and the majority of template-engaged RNA polymerase II (fraction P2) was 10-50 times as sensitive to deoxyribonuclease I as the DNA in the other two fractions (fractions P1 and S, comprising 78-88% of total nuclear DNA as large polynucleosomal aggregates and 2-6% of DNA mostly as mononucleosomes, respectively). In transcriptionally inactive nuclei obtained from hen erythrocytes, micrococcal nuclease did not separate DNA into fractions exhibiting such differential sensitivities. Second, we have monitored the partition of an expressed gene. Hybridization of complementary DNA to Xenopus albumin mRNA revealed a 5-10-fold enrichment of the albumin (but not the globin) gene in the P2 fraction of nuclei from Xenopus liver in which this gene is fully expressed. Third, a large part of the nascent rapidly labelled RNA synthesized in vivo in rat liver nuclei was recovered in the micrococcal-nuclease-derived fraction that is more susceptible to digestion with deoxyribonuclease I. It is concluded that mild micrococcal-nuclease treatment of nuclei causes their separation into transcribed and non-transcribed DNA as determined by a number of very different criteria.

Chromatin conformation of integrated Moloney leukemia virus DNA sequences in tissues of BALB/Mo mice and in virus-infected cell lines

The technique of preferential DNase I digestion of transcriptionally active chromatin regions was used to study the structural organization of integrated Moloney murine leukemia virus (M-MuLV) proviral sequences in various cells carrying integrated viral genomes. BALB/Mo mice, which carry M-MuLV as an endogenous virus at a single Mendelian locus, were used to examine the genetically transmitted viral genome copy and additional M-MuLV sequences acquired somatically during leukemogenesis. It has been shown previously that M-MuLV genome expression in these mice is restricted to lymphatic target tissues. In young homozygous BALB/Mo mice carrying one M-MuLV genome copy per haploid mouse genome in all cells we found that the genetically transmitted viral genome copy was in a preferentially DNase I-sensitive conformation in lymphatic target tissues, whereas in nontarget tissues the same sequence was not preferentially DNase I sensitive. This suggests that the chromatin conformation and the transcriptional activity of the integrated proviral genome are related to and probably determined by the state of cellular differentiation. In target tissues from BALB/Mo mice examined at different ages and in different stages of leukemogenesis the majority of the new somatically acquired M-MuLV sequences were preferentially DNase I digestible. A very similar pattern of DNase I digestibility was observed in target tissues from BALB/c mice exogenously infected with M-MuLV. This shows that in these tissues somatically acquired proviral sequences integrate preferentially or exclusively at sites of the host genome in which they are in a transcriptionally active chromatin conformation. Alternatively, the chromatin structure of the respective host genome region may be changed after the integration of viral DNA. In nontarget tissues from BALB/Mo mice the M-MuLV-specific sequences remained DNase I resistant throughout the lives of the animals. A different pattern of DNase I digestibility was observed in virus-infected cell lines which had been produced by low-multiplicity infection, cloned, and selected for virus production. When cell lines harboring different numbers of M-MuLV proviral copies were examined, it was found that a minority of the proviral sequences (on the average only one M-MuLV genome copy per haploid mouse genome) were preferentially digestible by DNase I, independent of the total number of proviral genome copies present. This suggests that the chromatin conformation of newly acquired proviral sequences is influenced by the state of differentiation of the infected cell or the way infected cells are selected or both.

Limited DNase I nicking as a probe of gene conformation

We have extended the use of pancreatic DNase I as a probe of chromatin structure by exploring the accessibility of an active gene to the introduction of the first single-stranded nick. We show by a target analysis that the beta-globin gene is about 25-fold more sensitive to single-site nicking than is an average sequence in the chicken erythrocyte nucleus or the nontranscribed albumin gene. The sites of initial DNase I nicking are shown to cluster within the transcribed sequence of the beta-globin gene.

Reconstitution of a deoxyribonuclease I-sensitive structure on active genes

Chicken erythrocyte nuclei have been labeled in the active regions of the chromosome by using the nick translation reaction. In this procedure, accessible areas of the genome are preferentially nicked by the action of pancreatic DNase I and subsequently labeled by using DNA polymerase I from Escherichia coli. These nuclei were employed as a substrate for studying the factors responsible for maintaining the special chromatin conformation of the overall population of active genes. Treatment of nuclei with 0.35 M NaCl resulted in the loss of DNase I sensitivity in the active genes, but this sensitivity could be restored when nuclei were reconstituted with the NaCl eluate. Further purification of the released factors revealed that the HMG (high-mobility group) proteins HMG-14 and HMG-17 are involved in maintaining the conformation of the active regions. These factors are not tissue specific and seem to be involved in the chromosomal structure of most of the active genes.

DNase I sensitivity of ribosomal genes in isolated nucleosome core particles

The level of chromatin structure at which DNase I recognizes conformational differences between inert and activated genes has been investigated. Bulk and ribosomal DNA's of Tetrahymena pyriformis were differentially labeled in vivo with [14C]- and [3H]-thymidine, respectively, utilizing a defined starvation-refeeding protocol. The 3H-labeled ribosomal genes were shown to be preferentially digested by DNase I in isolated nuclei. Staphylococcal nuclease digested the ribosomal genes more slowly than bulk DNA, probably owing to the higher GC content of rDNA. DNase I and staphylococcal nuclease digestions of purified nucleosomes and of nucleosome core particles isolated from dual-labeled, starved-refed nuclei were indistinguishable from those of intact nuclei. We conclude from these studies that DNase I recognizes an alteration in the internal nucleosome core structure of activated ribosomal genes.

Simple isolation of DNA hydrophobically complexed with presumed gene regulatory proteins (M3)

Chromatin from chicken reticulocytes and mouse Ehrlich ascites tumor cells has been extracted with 2 M NaCl, leaving a portion of the DNA still complexed with a fraction of nonhistones (designated M3, since it can be dissociated from DNA in solutions of 3 M NaCl containing 5 M urea). The DNA complexed with M3, separated from the bulk DNA by centrifugation, was found to contain sequences poorly represented in bulk DNA. Specifically we found that DNA--M3 complexes isolated from chicken reticulocyte chromatin were enriched in globin gene sequences by 20-fold relative to unfractionated DNA and by over 1000-fold relative to DNA rendered free of protein following the extraction of chromatin with 2 M NaCl. We have therefore isolated DNA fractions complexed with M3 which are enriched in specific sequences as may be determined by M3.

Selective activation of human beta-but not gamma-globin gene in human fibroblast x mouse erythroleukaemia cell hybrids

The human alpha- and beta-globin genes have been activated in MEL X human fibroblast cell hybrids. However, even though the human gamma- and beta-globin genes are closely linked and were shown in these hybrid clones to be present in approximately equal numbers, no human gamma-globin mRNA was produced. Thus, the human beta- and gamma-globin genes in these cells are differentially regulated apparently by a positive regulatory factor(s) specific for individual globin genes.

Nucleosome structure of Xenopus oocyte amplified ribosomal genes

The chromatin subunit or nucleosome structure of the amplified, extrachromosomal, ribosomal genes of oocytes of the amphibian Xenopus laevis has been investigated during stages of growth when these genes are markedly changing their rates of transcriptional activity. Nucleic acid hybridization studies involving micrococcal nuclease derived monomer nucleosome DNA fragments and purified ribosomal RNAs indicate that the apparent degree of accessibility of the ribosomal genes to short-term nuclease hydrolysis varies as a function of the rate of ribosomal RNA (rRNA) transcription. However, at no stage during oocyte development are all of the amplified ribosomal genes completely accessible to nuclease hydrolysis, even in those stages with maximal rates of rRNA transcriptional activity. These results suggest that the transcriptionally active ribosomal genes of oocytes are partially, or perhaps transiently, associated with histones in the form of nuclease releasable nucleosomes but that the degree of this association may change with varying rates of rRNA synthesis. Additionally, the present data indicate that the average size of the double-stranded ribosomal DNA associated with monomer nucleosomes is the same (about 200 base pairs) in all of the oocyte stages examined regardless of the rates of rRNA synthesis in these stages.

Electron microscopic evidence for splicing of Moloney murine leukemia virus RNAs

Poly (A) containing RNA extracted from Moloney murine leukemia virus infected mouse cells was hybridized with long single-stranded complementary DNA, prepared in detergent disrupted virions. Visualization of the hybrids in the electron microscope revealed among the structures, circles and circles with tails. Measurements performed on the circular molecules revealed two major species with circumferences corresponding to 3 and 8.2 kilobases. The latter structures had identical size to circles obtained after annealing of cDNA with the viral genome, 35S RNA. Circularization of a small viral RNA (3 kb) from infected cells in the RNA-cDNA hybrids is a direct evidence that like the 35S RNA it shares similar nucleotide sequences at both the 5' and 3' ends. The presence of 5' end sequences common to the two RNA species indicates the existence of a spliced viral RNA. Furthermore, based on the circularization of viral RNA in the hybrids, we suggest a new way to quantitate and determine the lengths of spliced RNA in retrovirus infected cells.

Regulation of expression and chromosomal subunit conformation of avian retrovirus genomes

We have investigated the copy number, chromosomal subunit conformation and regulation of expression of integrated avian retrovirus genomes. Our results indicate that there are approximately two copies of the endogenous viral genomes (RAV-O) per haploid cell genome in uninfected chick embryo fibroblasts (CEF) and red blood cells (RBC). The copy number and subunit conformation (as measured by DNAasel sensitivity) of the RAV-O genomes are independent of the level of expression of these viral DNA sequences. In cells isolated from embryos of the V+, gs-chf- and gs+chf+ phenotypes, approximately one of the two viral genomes is in a DNAase l-sensitive conformation. Upon infection with an exogenous Rous sarcoma virus (PR-RSV-C), one new viral genome is integrated per haploid CEF genome. The newly integrated RSV genome is completely sensitive to DNAase l, and the subunit conformation of the endogenous viral genomes is not altered by the integration of additional exogenous proviruses. Both the endogenous and newly integrated exogenous viral genomes are present in "nu-body" structures, and the selective sensitivity of these proviral DNA sequences to DNAase l is maintained in isolated nucleosomes. Our experiments revealing the DNAase l sensitivity of one of the two RAV-O genomes in gs-chf-CEF led us to reexamine the level of viral specific RNA in CEF of various GS genotypes. We find that GS/GS CEF contain approximately 100 copies of viral RNA per cell, gs/gs CEF contain no detectable viral RNA, and the heterozygote GS/gs CEF contain approximately 50 copies of viral specific RNA per cell. These results suggest that the GS gene controls production of RAV-O RNA sequences in CEF in a "cis" fashion. In RBCs, however, the expression of the RAV-O genome is independent of the GS gene, with both GS/GS and gs/gs RBCs containing roughly equivalent amounts of viral specific RNA. Our results suggest that the chromosomal structure of the endogenous viral genes is independent of the GS gene, and that the GS gene is cis-acting and tissue-specific.

Organization of 5-methylcytosine in chromosomal DNA

The 5-methylcytosine residues of L-cells have been labeled with [methyl-3H]-L-methionine and their chromatin localization studied using deoxyribonucleases. The kinetics of micrococcal nuclease digestion showed that the methylated cytosine residues are concentrated within regions resistant to nuclease digestion and preferentially missing from those regions between nucleosomes which are nuclease sensitive. Using DNA hybridization kinetic analysis, it is shown that 5-methylcytosine is abundant in highly repeated sequences but is also present in middle repetitive and unique sequence DNA.

DNA associated with hyperacetylated histone is preferentially digested by DNase I

Butyrate-treated cells give rise to massive hyperacetylation of histones and have been used to test the idea that regions of DNA in association with hyperacetylated histones are preferentially solubilized upon digestion with DNase I. Such hyperacetylated histones can be derived from both pre-existing histones or from histone newly synthesized in the presence of butyrate which leads to extreme modification. The DNA in association with both types of hypermodified histone is equally and selectively digested.

Chromatin

The approximate shape of the chromatin subunit called the nucleosome is now known, but its internal architecture is not well understood. Recent studies reveal details of the organisation of DNA within the nucleosome, and show that the arginine-rich histones are essential to DNA folding. Nucleosomes or structures related to them seem to be present at points of DNA replication and transcription; interactions within and between nucleosomes are likely to play a critical part in these processes.