The large number of phenotypic differences associated with Dam deficiency in E. coli is consistent with multiple roles for dam methylation (8, 11). These roles can broadly be defined as affecting (i) DNA mismatch repair (12), (ii) alterations in gene expression (15), (iii) the initiation of chromosome replication (1), and (iv) the chromosome structure (6). The phenotypes of dam mutants have been explored using presumed missense (e.g., dam-3) and insertion and deletion (e.g., dam-13::Tn9 and dam-16::Kan, respectively) mutations. In spite of numerous attempts, no temperature-sensitive dam alleles have been isolated from E. coli, although such alleles would be very useful in functional studies to turn Dam methylation on and off. These alleles would also allow testing of the effects of partially methylated or hemimethylated DNA on cell functions.

In an alternative approach, we placed the dam gene borne on a chromosomal XbaI-HindIII fragment from pYIN2 (13) under the control of the araBAD promoter in vector pBAD18 (5), creating pMQ400. In the absence of inducer, however, the cells still showed a Dam⁺ phenotype, as was determined by the action of DpnI (digests methylated dam DNA), DpnII (digests unmethylated dam DNA), and Sau3AI (digests both methylated and unmethylated dam DNA) on total DNA extracted from cultures using a MasterPure DNA purification kit (Epicentre). To reduce the amount of Dam produced from the plasmid, we removed the ribosome binding site downstream of the araBAD promoter in pMQ400. This was achieved by removing the native DNA upstream of the ATG of the dam gene and fusing the translation initiation codon (an NlaIII site, ↓CATG) directly to the SphI site (G↓CATG) in the polylinker sequence of pBAD18. The resulting construct was designated pMQ430.

Log-phase cultures of strain GM3819 (dam-16::Kan [16]) (Table ​(Table1)1) bearing pMQ430 growing in L broth show no detectable methylation of chromosomal DNA in the absence of inducer (Fig. ​(Fig.1).1). Upon exposure to 0.2% arabinose, however, chromosomal DNA was completely methylated within one generation (40 min), as was determined with the restriction enzymes DpnI, DpnII, and Sau3AI (Fig. ​(Fig.1).1). Removal of the inducer after a 40-min exposure and its replacement with 0.2% glucose led to a slow reduction in chromosomal methylation requiring 4.5 h or nine generations to effect complete loss of detectable methylation (Fig. ​(Fig.2).2). It should be noted that neither DpnI nor DpnII digests hemimethylated DNA. The cells showed no decrease in growth rate (30-min generation time) or cell number during this period. The time required to demethylate chromosomal DNA was longer than expected, and we suspect that this may be due to the slow degradation of the inducer. Consequently, we have not quantitated it further by the use of Southern or Western blots.

We have also tested the effect of sequentially turning Dam production on and off during cell growth. Cultures were exposed to 0.2% arabinose for 60 min, diluted, and then allowed to grow logarithmically in L broth with 0.2% glucose for 14 generations. The culture was again exposed to inducer, and the cycle was repeated two more times. We found that there was no change in growth rate or numbers of CFU during these cycling experiments. We conclude that alternating cycles of methylation and demethylation have no deleterious effects on cell growth.

Several proteases have been described for E. coli (4), and we wanted to determine if any were active in degrading Dam methyltransferase. We constructed pMQ430/dam-13::Tn9 derivatives of a series of protease-deficient strains and determined the state of DNA methylation 14 generations after a 60-min exposure to inducer. The results in Table ​Table22 show that under these conditions, methylated DNA was found only in a Lon-deficient strain, thus implicating it in Dam methyltransferase stability. This is the first direct evidence for proteolytic degradation of Dam and confirms previous indirect data to this effect (7).

We tested the utility of pMQ430 to produce Dam⁺ or Dam⁻ conditions in conjugal crosses between a dam Hfr strain (GM2807 Kan^r Str^s [Table ​[Table1])1]) and a recipient strain (GM7428 Str^r [Table ​[Table1])1]) bearing a ruvABC::Cam deletion. Logarithmic-phase cells were mixed in a ratio of 1 donor to 10 recipients (at about 10⁸ per ml), mated for 60 min, and plated on selective media with or without arabinose. As shown in Table ​Table3,3, no dam ruvABC recombinants (Kan^r Str^r) were recovered in the absence of arabinose, confirming previous results indicating that this combination of mutations is lethal (10). A few recombinant colonies did appear on the selective plates, but although these were dam by their kanamycin-resistant phenotype, they were also chloramphenicol sensitive, indicating the loss of the ruvABC allele and its replacement by the wild-type genes. The same low frequency of colonies was obtained when the mating mixture was plated in the presence of 0.002 and 0.02% arabinose. At a concentration of 0.2% arabinose, however, we observed a high frequency of recombinants which were sensitive to UV light and chloramphenicol resistant, indicating a ruvABC defect. These recombinants were resistant to ampicillin, indicating retention of pMQ430. They were unable to grow when they were patched onto the same selective medium without arabinose, thereby confirming that the dam ruvABC combination is inviable.