Biology of cell killing by 1-beta-D-arabinofuranosylcytosine and its relevance to molecular mechanisms of cytotoxicity.

Cells of the Chinese hamster ovary cell line were used to study the process of cell death induced by pulse treatment with 1-beta-D-arabinofuranosylcytosine (ara-C). Cells were synchronized by mitotic selection and pulse treated in early S phase with a concentration of ara-C (1 mM) which was sufficient to reduce plating efficiency to a few percentages of the control. The process of when and how the lethally damaged cells die was studied using a series of techniques in parallel. These included time-lapse microcinematography, flow microfluorimetry, and chromosome morphology in both anaphases/telophases and Colcemid-arrested metaphases. Most of the lethally damaged Chinese hamster ovary cells progressed through one, and many through two, cell cycles before death occurred. The cell death and abnormal divisions can be accounted for by the chromosome aberrations observed in Colcemid metaphases and anaphases/telophases. Death without any attempted division occurred between 3 and 9 normal cell cycle times after ara-C treatment. Chinese hamster ovary cells were also treated continuously with 1 mM ara-C. Under these conditions, cell death was still primarily division related. We argue that these data are not consistent with the actual incorporation of ara-C moieties into DNA being the primary cause of cell death. The data are discussed in relation to the postulated molecular mechanisms of toxicity of this drug.

Reversal of the multidrug resistance phenotype with cremophor EL, a common vehicle for water-insoluble vitamins and drugs.

A polyethoxylated castor oil, Cremophor EL, which is used as a vehicle for p.o. and i.v. administration of water-insoluble compounds in humans, can reverse the multidrug resistance (MDR) phenotype at doses which are likely to be readily achievable clinically. Using flow cytofluorometric analysis of daunorubicin (DNR) uptake as a measure of the expression of the MDR phenotype, Cremophor EL (1:10(3] in the growth medium increased intracellular DNR in an MDR cell line, R100 cells, to levels similar to that observed in the drug-sensitive parental cells, CCRF-CEM. A similar Cremophor EL-induced increase in DNR uptake was also observed in an unrelated MDR cell line derived from K562 cells. Cremophor EL (less than or equal to 3:10(4] did not inhibit the growth of CCRF-CEM cells or its vinblastine-resistant derivative, R100 cells, but would significantly increase the sensitivity of R100 cells to both vinblastine and DNR. Also Cremophor EL did not increase the sensitivity of normal bone marrow progenitor cells cultured in vitro to high concentrations of vinblastine. Cremophor EL may prove to be a relatively pharmacologically inactive addition to chemotherapeutic protocols which may be able to reverse the MDR phenotype in tumors and also help to prevent the selection of MDR cell variants from within a tumor cell population during chemotherapy.

Sequence specificity of cytosine methylation in the DNA of the Chinese hamster ovary (CHO-K1) cell line.

We have determined the DNA renaturation kinetics for those DNA sequences of the Chinese hamster ovary (CHO-K1) cells in which enzymatic cytosine methylation occurred immediately after strand synthesis and for those in which methylation was delayed after strand synthesis. DNA sequences showing immediate or delayed methylation were found to be distributed throughout all repetition classes of the DNA of these cells, with a slight concentration of immediate methylation in moderately repetitive sequences and with delayed methylation being slightly over-represented in the highly repetitive fraction. However, DNA sequences showing both classes of methylation were represented equally in unique DNA sequences. We interpret these data to mean that the methylase acting near the replication forks (the 'immediate' methylase) is a relatively inefficient enzyme, missing some 20% of hemimethylated sites produced by DNA replication in these cells. We suggest that the methylase performing maintenance methylation at sites remote from the replication forks (the 'delayed' methylase) is simply a back-up enzyme for the first and that it has no true sequence specificity. The implications of this for the function(s) of DNA methylation in mammalian cells are discussed.

DNA sequences showing a delay in cytosine methylation after replication. Time course of methylation in synchronized mammalian cell populations and relationship to DNAase I sensitive domains.

We have shown that in several mammalian cell lines a minor fraction of cytosine methylation is delayed for up to several hours after strand synthesis and that different methylases performed the immediate and the delayed classes of DNA methylation. To investigate the time course of this delayed methylation we have used three different cell lines, two of human and one of hamster origin. These were synchronized by two different methods: mitotic detachment and double hydroxyurea blocks. A uniform picture was obtained with all three cell lines. Delayed methylation of early replicating sequences occurred while cells were still in mid-S-phase, with the maximum rate of delayed methylation occurring in cells in the second half of S and in G2. Delayed methylation seems to be complete before cells entered the next G1-phase. Limited DNAase I hydrolysis of cell nuclei was used to test whether the delay in methylation in some DNA sequences was due to high levels of transcriptional activity. However, DNA sequences exhibiting delayed methylation showed no preferential concentration in or exclusion from DNAase I hypersensitive regions.

A sizing artifact of DNA restriction fragments with unusually long single-stranded termini.

The restriction endonucleases (ENases) BstNI (CCATGG) and EcoRII (CCATGG) both cleave DNA at the same time sequences, but only EcoRII produces 5-nucleotide (nt) cohesive ends and is inhibited by 5-methylation of the inner cytosine. The low-Mr fragments in digests of mouse DNA made with these two ENases exhibit different mobilities during agarose-gel electrophoresis. The difference in the mobilities of the BstNI and EcoRII fragments from mouse DNA was not due to closely spaced, differentially methylated sites, or to alternate mechanisms such as circularization of the long cohesive ends of the EcoRII fragments, or to residual bound protein. Rather, it was due to the unusually long 5-nt single-stranded (ss) ends of fragments produced by EcoRII digestion, since the slower mobility of the EcoRII fragments was abolished by treatment with ss-specific nuclease. Similar mobility differences between BstNI and EcoRII fragments which could be removed by ss nuclease were also observed in digests of simian virus 40 DNA.

Effects of mcr restriction of methylated CpG islands of the L1 transposons during packaging and plating stages of mammalian genomic library construction.

The use of optimally methylation-tolerant mcrA- mcrB- strains has been shown to produce an over tenfold increase in the plating efficiencies of mammalian genomic libraries, compared to a superior conventional phage host strain LE392 which is mcrB+. However, there is an even more significant effect of mcr restriction. Amongst the recombinants recovered with an mcrB+ host, we have found that there is an additional 30-fold reduction in the frequencies of clones containing the heavily methylated 5'-CpG island sequences of both the human and rat L1 repetitive elements. The mcrA product was also found to restrict clones of these methylated genomic segments, but not as strongly as mcrB. However, the use of packaging extracts made from mcrA+ lysogens did not result in convincing reductions in the recoveries of these dispersed methylated elements. The magnitude of mcr restriction during plating due to methylated dispersed elements is sufficient to make a significant proportion of mammalian genomes unclonable from genomic libraries constructed previously using conventional mcr+ hosts.

Red cell nucleotide levels in patients with thalassaemia, iron deficiency, and renal failure.

Red cell nucleotides, predominantly adenine nucleotides, were measured spectrophotometrically in normal individuals and in patients with iron deficiency, beta thalassaemia and renal failure following dialysis or transplantation. There was a significant correlation between the mean red cell haemoglobin (MCH) or the mean cell volume (MCV) and the nucleotide levels in red cells (P less than 0.001). Patients with beta thalassaemia did not have increased nucleotide levels compared to red cells having the same MCH or MCV. Patients with renal failure had significantly increased levels of nucleotides compared to normal individuals, and this level remained high after renal dialysis but returned to normal following renal transplantation.

The majority of methylated deoxycytidines in human DNA are not in the CpG dinucleotide.

The only natural postsynthetic modification known to occur in mammalian DNA is the methylation in the 5 position of deoxycytidines. Of the four 5'-CpN-3' dinucleotides (ie. CpG, CpC, CpA, and CpT), the dinucleotide which contains the highest proportion of deoxycytidines methylated is CpG, with 40 to 80% methylation in different mammalian genomes. It has also been shown that CpA, CpT, and CpC are methylated as well but to a much lower extent. Here we report the result of a full nearest neighbour analysis (together with quantitation of methylation levels in the 4 CpN dinucleotides) for DNA from human spleen. Using the values we have calculated the overall frequencies for all the methylated dinucleotides in the human genome. Because of the relative underrepresentation (by 7 to 10 fold) of the CpG dinucleotide, only 45.5% of total mC was present in mCpG, with 54.5% in mCpA, mCpT plus mCpC. These calculations have implications for studies into the function and significance of DNA methylation in mammalian cells.

Aberrant replication of chromosomal DNA segments in CHO cells as a manifestation of a DNA repair system.

We have previously demonstrated that transient blocking of DNA synthesis with drugs such as 1-beta-D-arabinofuranosylcytosine results in some segments of the chromosomal DNA being replicated more than once in a single cell cycle. One explanation for this phenomenon might be that it is due to a perturbation of the cellular mechanism which normally controls the process of initiating replication of the chromosomal DNA and which ensures that each DNA segment is normally replicated once only in each cell cycle. To examine an alternative explanation, we have used the Chinese hamster ovary CHO-K1 cell line to test whether UV irradiation induces aberrant double replication of chromosomal DNA segments. We present data which show that UV irradiation induces a linear increase (regression coefficient, 0.996) in aberrant reinitiation of DNA replication in DNA segments replicated earlier in the same cell cycle. This was shown by DNA strands labelled with bromodeoxyuridine (BrdU) being synthesized off 3H-labelled template strands which were themselves synthesized during a [3H]thymidine pulse shortly before the UV irradiation. We suggest that blocked replication forks trigger an emergency response to unreplicated DNA segments in which abnormal origins of replication are used to circumvent this damage. This results in abnormal patterns of DNA replication, and it occurs whether DNA replication forks are blocked metabolically (such as with an inhibitor of DNA polymerase) or physically (as in the case of pyrimidine dimers).

Levels and stability of DNA methylation in random surviving cell clones derived from a Chinese hamster cell line after prolonged treatment with 5-aza-2'-deoxycytidine.

SCC30 cells (derived from a single cell from the Chinese hamster ovary CHO-K1 cell line, selected on the basis of a stable chromosome complement) were used to select cell variants with hypomethylated DNA. Cells were treated with 5-aza-2'-deoxycytidine (5azadCyd) at 0.1, 1, or 5 microM for two weeks with the medium and drug renewed twice weekly. From the few surviving cells, 25 random single cell-derived clones were grown for freezing cell stocks, and for DNA isolation for 5-methyldeoxycytidine (5medCyd) estimations. After a minimum of one month's recovery from the drug, these cells showed a continuum of 5medCyd levels ranging from ones with the same as the parental clone (2.93%) to ones having lost almost 50% of their DNA methylation. The modal value corresponded to a loss of one third to one quarter of methylated sites. Five subclones with hypomethylated DNA were grown from the frozen stocks. These cells were shown not to be 5azaCyd-resistant cell variants. By the time sufficient cells had been grown to determine DNA methylation levels, the average percentage of 5medCyd had increased to 76% of the SCC30 value compared to 67% at the time of freezing cell stocks. However, this level of DNA hypomethylation remained constant over two months of continuous culture. Cells of one of these hypomethylated subclones were subjected to a second cycle of 5azaCyd treatment. Six random clones from the survivors showed a further decrease averaging 11% in the level of DNA methylation but, by two months in continuous culture, 5medCyd levels had returned to that present before the second cycle of selection. Hence, cell variants can be readily obtained which have lost some 8-10 million methylated sites (pairs of methylated deoxycytidines), and this loss does not compromise cell viability in in vitro culture. This is consistent with mammalian genomes containing a high level of background methylation in non-essential sites. The usefulness of such single cell-derived clones with stably hypomethylated genomes is discussed in relation to understanding the functions of deoxycytidine methylation in mammalian DNA.

Delayed DNA methylation is an integral feature of DNA replication in mammalian cells.

In the majority of sites of methylation in the DNA of mammalian cells, the symmetry of methylation is restored within a few minutes of the passage of a replication fork. However, it has been shown that daughter strand methylation in immortalised cell lines is delayed in a substantial minority of sites for up to several hours after replication. We report here the results of two new approaches to the determination of the functional significance of delayed DNA methylation in mammalian cells. Firstly, we demonstrate that normal, nontransformed cells (human peripheral lymphocytes in short-term primary culture) have comparable proportions of delayed DNA methylation to many immortalised cell lines, showing that delayed DNA methylation is not just a secondary consequence of abnormally high methionine requirements commonly observed in transformed cells and that delayed DNA methylation would be unlikely not to occur in vivo. Secondly, we have used 5-aza-2'-deoxycytidine (5azadCyd) to derive subclones of cells from the Chinese hamster ovary cell line which have stably hypomethylated DNA. In three of these subclones which had lost on average one fourth of the methylation sites from their genomes, the proportion of daughter strand methylation which was delayed after replication was reduced by less than 10%. If delayed DNA methylation were site-specific, this implies that of the order of twice the number of "immediate" methylation sites than delayed methylation sites had been lost from the genomes of these hypomethylated subclones. Thus, delayed DNA methylation is an integral part of the process whereby replicating mammalian cells maintain the pattern of methylation in their genomes. These observations are discussed in relation to the significance of delayed DNA methylation for the accurate maintenance of methylation patterns in the genome and the consequent implications for the possible role of methylated deoxycytidines in mammalian gene control.

Revised genomic consensus for the hypermethylated CpG island region of the human L1 transposon and integration sites of full length L1 elements from recombinant clones made using methylation-tolerant host strains.

Efficient recovery of clones from the 5' end of the human L1 dispersed repetitive elements necessitates the use of deletion mcr- host strains since this region contains a CpG island which is hypermethylated in vivo. Clones recovered with conventional mcr+ hosts seem to have been derived preferentially from L1 members which have accumulated mutations that have removed sites of methylation. We present a revised consensus from the 5' presumptive control region of these elements. This revised consensus contains a consensus RNA polymerase III promoter which would permit the synthesis of transcripts from the 5' end of full length L1 elements. Such potential transcripts are likely to exhibit a high degree of secondary structure. In addition, we have determined the flanking sequences for 6 full length L1 elements. The majority of full length L1 clones show no convincing evidence for target site duplication in the insertion site as commonly observed with truncated L1 elements. These data would be consistent with two mechanisms of integration of transposing L1 elements with different mechanisms predominating for full length and truncated elements.

The effect of E. coli host strain on the consensus sequence of regions of the human L1 transposon.

We have used highly methylation tolerant host strains to clone hyper- and hypo-methylated genomic elements from different regions of the same family of long interspersed repetitive elements from human DNA, specifically the 1.8 kilobase (kb) and 1.2kb KpnI fragments from members of the L1 family of transposable elements in which respectively some 18% and 2.7% of cytosines are methylated in vivo in human spleen DNA. The consensus of the DNA sequences of the ends of 13 clones from the hypomethylated region of human L1 agreed exactly with the consensus derived previously from clones made using conventional host strains. However the sequences of 18 of our clones from the 5' end of the hypermethylated region differed significantly from the sequences of clones made using conventional hosts (P less than 0.0001). The 5' region of the 1.8kb L1 region is a CpG island which, in human somatic tissue, appears to be maintained in a highly methylated state, including methylation at sites other than CpG dinucleotides. The consensus sequence of this region also has features suggestive of a previously unrecognized open reading frame.

RglB facilitated cloning of highly methylated eukaryotic DNA: the human L1 transposon, plant DNA, and DNA methylated in vitro with human DNA methyltransferase.

In vitro methylation of Bluescribe plasmid DNA (pBS) with human placental DNA methyltransferase to 6% 5-methylcytosine (mC) reduced transformation efficiencies in rglB+ host strains C600 and DS410 by almost 2 orders of magnitude. By contrast, the rglB- derivative of DS410 showed no reduction in transformation efficiency with methylation while the rglB- derivative of C600 was partially tolerant to methylation. Further, we show that the 1.8 kilobase (kb) and 1.2 kb KpnI fragments derived from the human L1 repeat have respectively 18.3% and 2.3% mC in vivo. Using these hyper- and hypo-methylated genomic segments ligated into the pBS plasmid, transformants with the highly methylated 1.8 kb L1 insert were recovered at 17 to 40 fold higher frequency with the rglB- host strains than with the rglB+ hosts. In addition, recombinant phage (lambda 2001) containing inserts of plant genomic DNA with 26.7% mC (from Petunia hybrida) when plated on rglB- hosts gave titres up to 222 times higher than on the rglB+ strains.

Quantitative evaluation of Escherichia coli host strains for tolerance to cytosine methylation in plasmid and phage recombinants.

Many strains of E. coli K12 restrict DNA containing cytosine methylation such as that present in plant and animal genomes. Such restriction can severely inhibit the efficiency of cloning genomic DNAs. We have quantitatively evaluated a total of 39 E. coli strains for their tolerance to cytosine methylation in phage and plasmid cloning systems. Quantitative estimations of relative tolerance to methylation for these strains are presented, together with the evaluation of the most promising strains in practical recombinant cloning situations. Host strains are recommended for different recombinant cloning requirements. These data also provide a rational basis for future construction of 'ideal' hosts combining optimal methylation tolerance with additional advantageous mutations.