Characterization of a dam Mutant of Serratia marcescens and Nucleotide Sequence of the dam Region

Abstract

The DNA of Serratia marcescens has N₆-adenine methylation in GATC sequences. Among 2-aminopurine-sensitive mutants isolated from S. marcescens Sr41, one was identified which lacked GATC methylation. The mutant showed up to 30-fold increased spontaneous mutability and enhanced mutability after treatment with 2-aminopurine, ethyl methanesulfonate, or UV light. The gene (dam) coding for the adenine methyltransferase (Dam enzyme) of S. marcescens was identified on a gene bank plasmid which alleviated the 2-aminopurine sensitivity and the higher mutability of a dam-13::Tn9 mutant of Escherichia coli. Nucleotide sequencing revealed that the deduced amino acid sequence of Dam (270 amino acids; molecular mass, 31.3 kDa) has 72% identity to the Dam enzyme of E. coli. The dam gene is located between flanking genes which are similar to those found to the sides of the E. coli dam gene. The results of complementation studies indicated that like Dam of E. coli and unlike Dam of Vibrio cholerae, the Dam enzyme of S. marcescens plays an important role in mutation avoidance by allowing the mismatch repair enzymes to discriminate between the parental and newly synthesized strands during correction of replication errors.

In Escherichia coli, the Dam enzyme (DNA adenine methyltransferase) catalyzes the methylation of adenine at N₆ in the sequence GATC in duplex DNA (20, 21), a reaction in which S-adenosylmethionine is the methyl group donor and also an allosteric effector (7). The methylation occurs with a delay at replication forks rendering the newly synthesized DNA strand temporarily unmethylated (6, 12). The first dam mutants of E. coli were isolated by Marinus and his coworkers (27, 28). From studies with such mutants and strains which overproduce the Dam enzyme, several important roles of the methylation status of DNA have been deduced. First, dam hemimethylation of DNA at the replication fork is required for parental strand-directed mismatch repair by the mutHLS system (for a review, see reference 33). This was concluded from the observations that dam mutants display increased spontaneous mutability (29) and are hypermutable by and sensitive to various mutagens, including base analogues (16, 17). Overproduction of the Dam enzyme increases the spontaneous mutation frequency, presumably as a consequence of a reduced time span of the hemimethylated status (21, 30). Second, in cells which lack or overproduce the Dam enzyme, the timing and control of DNA replication are disturbed (9, 31). Apparently, the methylation of GATC sequences in the origin region is critical for replication (32, 40). Finally, the methylation status of GATC sequences in −35 and −10 regions of the promoters of several genes appears to have an effect on gene expression (6). It has further been established that the dam gene of E. coli is expressed under growth rate control (36, 37).

In E. coli, the dam gene is located at 74 min of the chromosome as a cistron within an operon of seven genes (10, 22, 24). dam mutants are known to be sensitive to the base analogue 2-aminopurine (2-AP; 17). Besides the E. coli mutants, dam-deficient strains have been isolated from Salmonella typhimurium (39, 45) and Vibrio cholerae (3, 4). While the dam mutant of S. typhimurium had a phenotype rather similar to that of the E. coli mutants, the V. cholerae mutant was 2-AP and UV sensitive but did not display enhanced mutability. The amino acid sequence of the Dam enzyme of V. cholerae was rather different from that of the enzyme of E. coli (3). In order to establish whether the dam gene and its function observed in E. coli and the closely related bacterium S. typhimurium represent a more general type among members of the family Enterobacteriaceae, we have isolated a dam mutant of Serratia marcescens. Here we describe some of its properties with respect to spontaneous and induced mutability. We have also identified, cloned, and sequenced a DNA fragment of S. marcescens with the dam gene and neighboring sequences. Our results suggest that the dam gene of S. marcescens differs in structure and function from the V. cholerae gene and is more like those of E. coli and S. typhimurium.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The strains and plasmids used are listed in Table 1. If not stated otherwise, TBY broth medium (10 g of Bacto Tryptone, 5 g of Bacto Yeast Extract, 5 g of NaCl per 1,000 ml) was used. TBY plates contained 15 g of Difco agar per 1,000 ml. The incubation was at 30°C. If required, the media contained antibiotics or other supplements as detailed later in the text.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Size (kb)	Relevant genotype or alternate designation	Source or reference
S. marcescens strains
Sr41		Wild type	44
WA736		Same as Sr41 but dam-1	This work
E. coli K-12 strains
WA112		K-12s	2
DH5α		recA1 endA1 gyrA96 thi-1 supE44 relA1 hsdR17 (rK−mK+)	19
GM2159		dam-13::Tn9; Cmr	27
Plasmids
pBR322	4.4	AprTcr	8
pBR328	4.9	AprCmrTcr	43
pRE432	13.5	Mini-F cosmid vector; Apr Cmr Tcr	13
pBluescript KS+	2.9	Apr	Stratagene, La Jolla, Calif.
pTO1B	41.0	pRE432 with genomic insert of 27.5 kb of Sr41 DNA; AprCmrdam+	This work
pTO3C	8.1	pBR328 with 3.2-kb ClaI fragment from pTO1B; Apr Cmrdam+	This work
pTO6A	4.3	pBluescript with 1.35-kb EcoRV-SacI fragment from pTO3C; Apr	This work
pTO11A	4.5	pBluescript with 1.5-kb BamHI-ClaI fragment from pTO3C; Apr	This work

Treatment of cells with N-methyl-N-nitro-N-nitrosoguanidine.

Log-phase cells were mutagenized for 10 min in phosphate buffer (0.04 M Na₂HPO₄, 0.02 M KH₂PO₄, 0.07 M NaCl, 0.002 M MgSO₄, pH 7.4) containing N-methyl-N-nitro-N-nitrosoguanidine (75 μg/ml) at 30°C. The cells were then sedimented, washed twice in prewarmed TBY, and incubated further in TBY at 30°C. Cell survival was about 20%.

DNA manipulations, cloning, and sequencing.

Isolation of plasmids, restriction analysis of DNA, and cloning of DNA fragments were done by following standard procedures (26). The recipient in cloning experiments was E. coli DH5α. DNA sequencing was performed on both strands with the dideoxy-chain termination method (42).

Bacterial transformations.

S. marcescens cells were transformed by the Ca²⁺ method as described by Takagi and Kisumi (44), and E. coli cells were transformed by the method of Hanahan (19). dam mutants of E. coli and S. marcescens were transformed by electroporation. Electrocompetent cells were prepared from log-phase cultures (1 × 10⁸ to 2 × 10⁸ cells/ml) as described by Dower et al. (14) and stored at −80°C in 10% glycerol. For electroporation (14) employing a Gene Pulser (Bio-Rad Laboratories, Richmond, Calif.), the parameters were 12,500 V/cm, a capacitance of 25 μF, and a parallel resistance of 200 Ω.

Determination of spontaneous mutation frequencies.

Overnight cultures in TBY (5 ml) were started from single colonies. After 18 h of growth, the cultures were diluted in phosphate buffer and samples were plated after appropriate dilution on selective media (TBY with chloramphenicol, tetracycline, kanamycin, ampicillin, or rifampin) and on TBY (viable count). The mutation frequency is the ratio of the resistant-cell titer to the viable-count titer.

Treatment of cells with UV, 2-AP, or ethyl methanesulfonate (EMS).

The following experiments were done at least in triplicate with cultures each started from a single colony.

(i) UV.

Log-phase cells grown in TBY were resuspended in phosphate buffer (10 ml, 2 × 10⁸ cells/ml) and irradiated with a UV dose of either 0, 3, 6, or 9 J/m² (dose rate, 0.6 J/m² s). Samples of 0.1 ml were diluted in phosphate buffer and plated for measurement of survival. The remaining cells were sedimented, resuspended in TBY, and aerated at 30°C for 4 h. The cells were then sedimented and resuspended in 1 ml of TBY. Undiluted or appropriately diluted samples were then plated on selective (TBY with chloramphenicol or rifampin) and nonselective (viable count) media to determine the frequencies of resistant mutants.

(ii) 2-AP.

Stationary-phase cells grown in M9 medium (26) were diluted in M9 medium to give a titer of 2 × 10⁷ cells/ml. To 10-ml samples, 2-AP from an aqueous stock solution (5 mg/ml) was added to final concentrations of 0, 10, 50, and 100 μg/ml. The cultures were aerated at 30°C until the sample without 2-AP reached a titer of 4 × 10⁸ cells/ml (about 3 h). The cells of all samples were then sedimented, washed twice in phosphate buffer, and each resuspended in 1 ml of phosphate buffer. Appropriate dilutions were plated on selective media (TBY with chloramphenicol or rifampin) and on TBY (viable count) to determine the frequencies of resistant mutants.

(iii) EMS.

EMS was added to 10-ml samples of a log-phase culture grown in TBY to give final concentrations of 0, 0.2, 0.5, and 1%. After 10 min at 30°C in a shaking incubator, the cells were sedimented and washed with TBY and samples were plated after appropriate dilution on TBY to determine survival. Of the remaining suspensions, 2.5 ml was added to prewarmed TBY and incubated for 4 h at 30°C. The cells were then sedimented and resuspended in 1 ml of TBY, and appropriately diluted samples were plated on selective (TBY with chloramphenicol or rifampin) and nonselective (viable count) media to determine the frequencies of resistant mutants.

Nucleotide sequence accession number.

The sequence described in this report has been deposited in the EMBL database under accession no. X78412.

RESULTS

Isolation of a dam mutant.

Preliminary experiments had shown that chromosomal DNA of S. marcescens Sr41 and pBR322 DNA isolated from a plasmid-bearing derivative of this strain were cleaved in vitro by restriction endonuclease Sau3AI and were refractory to digestion with MboI. Both enzymes recognize the sequence GATC. Sau3AI cleaves it irrespective of methylation at A, whereas MboI cleaves only the nonmethylated sequence. Corresponding to previous studies on DNA methylation by restriction analysis (15), it was concluded that in S. marcescens Sr41, the GATC sequences are methylated as in E. coli by the Dam enzyme (28). This corresponds to a previous finding of dam methylation in S. marcescens (5). Since it is known that dam mutants of E. coli are sensitive to 2-AP (17), we decided to use 2-AP sensitivity in screening for a dam mutant of S. marcescens. In a sensitivity test with S. marcescens Sr41, it was found that 1,000 μg of 2-AP/ml of TBY agar was the highest concentration which did not suppress the growth of the strain. About 3,000 colonies from a mutagenized culture were replica plated on 2-AP agar (1,000 μg/ml) and six 2-AP-sensitive clones were identified. The six mutants were found to have resistance to UV irradiation and mitomycin C treatment (growth on TBY agar with 1 μg of mitomycin C/ml) similar to that of the parental strain, indicating that they were not recA mutants, which are known to be highly sensitive to these DNA-damaging agents. After all six 2-AP-sensitive strains were transformed with pBR322, the plasmid DNA isolated from only one of the mutants was cleavable by both Sau3AI and MboI. This indicated that the mutant lacked dam-specific DNA methylation. The organism was termed the dam-1 mutant and characterized further.

Characterization of the dam-1 mutant.

In E. coli, the deficiency of dam activity results in increased mutability. To examine the phenotype of the S. marcescens dam-1 mutant, the frequencies of various spontaneous forward mutations were determined (Table 2). Compared to the wild type, the frequencies of mutation to rifampin resistance (Rif^r), chloramphenicol resistance (Cm^r), and ampicillin resistance (Ap^r) were increased about 30-fold and the frequency of kanamycin resistance (Km^r) was increased about 8-fold. No strong increase in the frequency of tetracycline resistance (Tc^r) was observed (Table 2). Hypermutability of the dam-1 strain was also found when the frequencies of induced mutations (Rif^r and Cm^r) were determined after treatment of cells with various mutagens. The dam-1 strain was slightly more UV sensitive than the wild type, and the number of induced Rif^r and Cm^r mutants increased much more with increasing UV doses in the dam-1 strain than in the wild type (Fig. 1a). In the presence of 0.5% EMS, the survival of the dam mutant was 32% (wild type, 90%) and the frequency of Rif^r was 3 × 10⁻⁶ (wild type, 1 × 10⁻⁸). A result similar to that obtained with EMS was also obtained with 2-AP (Fig. 1b). The remarkable sensitivity of the dam-1 strain to 2-AP parallels observations with E. coli and S. typhimurium (17, 39) and allowed the isolation of the mutant. The increase in Rif^r mutant numbers with increasing 2-AP doses was stronger than that which could be expected if selection of preformed Rif^r occurred. The increased spontaneous and induced mutability of the dam-1 strain suggests that a mechanism of mutation avoidance is blocked, a phenotype also seen in E. coli dam mutants (17).

TABLE 2.

Spontaneous mutability of wild-type S. marcescens and the dam-1 mutant to resistance to various antibiotics

Antibiotic (conc, μg/ml)	Mutation frequency/107 cellsa		Mutant/wild-type ratio
	Wild-type	dam-1 mutant
Kanamycin (25)	7.8 ± 2.8	59.0 ± 27.3	7.6
Rifampin (100)	0.2 ± 0.1	6.1 ± 3.1	28.1
Tetracycline (75)	111.4 ± 29.9	188.7 ± 63.6	1.7
Chloramphenicol (50)	1.6 ± 0.5	47.7 ± 11.7	30.5
Ampicillin (100)	2.2 ± 0.6	67.0 ± 12.6	29.9

^aThe data are means and standard deviations of three independent determinations. 

FIG. 1.

Effect of UV irradiation (a) or 2-AP treatment (b) on survival (top) and mutation frequency (Cm^r, middle; Rif^r, bottom) in dam⁺ (○) and dam-1 mutant (●) strains. The data are means and standard deviations of three independent experiments.

Cloning of the dam gene.

Since the expression of the dam⁺ gene provides resistance of cells to 2-AP in S. marcescens and E. coli, we used this phenotype to screen a monocopy cosmid genomic library of S. marcescens (13) for a plasmid carrying the dam gene. For this, the E. coli dam-13 mutant was transformed with the library cosmids (selection was for Ap^r) and the transformant colonies were replica plated on TBY supplemented with 900 μg of 2-AP/ml. At this concentration, the E. coli dam-13 strain no longer grew. Several 2-AP-resistant transformants were identified. Their plasmids were isolated and tested for dam-specific DNA methylation by treatment with MboI and Sau3AI, respectively. The DNA of all plasmids was restricted by Sau3AI and not by MboI, indicating a Dam⁺ phenotype of the cells. The plasmid of one of the transformants was termed pTO1B. Its restriction map (Fig. 2) indicated that it contained a 27.5-kb insert of DNA.

FIG. 2.

The dam region of S. marcescens. The top line is a restriction map of the 27.5-kb insert of S. marcescens DNA covering dam⁺ cloned into the BamHI site of mini-F cosmid vector pRE432 (13). The second line from the top is a restriction map of the 3.2-kb ClaI fragment expressing dam⁺ cloned into the ClaI site of pBR328 (PTO3C). The thin horizontal lines are the fragments cloned in pBluescript for sequencing (pTO6A and pTO11A), and the dotted line depicts the sequenced DNA. The arrows indicate the positions and direction of ORFs. The filled boxes represent vector DNA at the cloning sites.

Restriction fragments of the insert of pTO1B were subcloned into pBR328 by using E. coli DH5α. Lack of cleavage by MboI of a plasmid with the 3.2-kb ClaI fragment after passage through E. coli dam-13 indicated that the dam gene was expressed from this fragment (pTO3C; Fig. 2).

Nucleotide sequence of the dam region.

Overproduction of the Dam enzyme results in a hypermutable phenotype in E. coli. Therefore, the DNA for sequencing of the dam region was prepared from pBluescript multicopy plasmids in which only parts of the dam gene were cloned. As shown in Fig. 2, these plasmids were pTO6A (1.35-kb EcoRI-SacI fragment) and pTO11A (1.5-kb BamHI-ClaI fragment), both of which did not provide Dam activity (data not shown). The sequences of both inserts were determined and aligned. The 2,340 bp revealed three open reading frames (ORFs) of 642, 810, and 630 bp oriented in the same direction (Fig. 2). The first ORF turned out to be the 3′ terminal portion of a gene with 45% identity of the deduced amino acid sequence (within the 214 codons present in the sequence) to that deduced from the urf 74.3-nucleotide sequence of E. coli which is located in front of dam (22). The second ORF codes for a protein of 270 amino acids (aa) (molecular mass, 31.3 kDa) with 72% sequence identity to the E. coli Dam protein (278 aa; 10, 22, 24). This gene was termed dam of S. marcescens. The amino acid sequence encoded by the dam gene contains motifs I to VIII and X, which are conserved among members of the α group of N₆-adenine aminomethyltransferases; motifs I to III and X are involved in S-adenosylmethionine binding, and the others are responsible for catalysis (25). The third ORF codes for a protein of 210 aa (molecular mass, 22.8 kDa) and was termed dod (for downstream of dam). It has 84% amino acid sequence identity with the recently identified protein encoded by the rpe gene of E. coli, which is also located directly downstream of dam and codes for a d-ribulose-5-phosphate epimerase (24). The dam gene of S. marcescens is followed by an inverted repeat of 7 bp which could form a hairpin (−18.1 kcal/mol) and might function as a transcription terminator.

Effect of dam⁺ overexpression in S. marcescens.

In E. coli, increased Dam activity is correlated with high spontaneous mutability. When cells of S. marcescens Sr41 were transformed by the multicopy plasmid pTO3C containing the dam⁺ region of S. marcescens, the frequency of spontaneous Rif^r mutants increased almost 100-fold (Table 3). The plasmid raised the mutability of the dam-1 mutant to the same high level (Table 3), indicating that no regulatory stimulation of chromosomal dam gene expression but the presence of a high number of dam⁺ genes leads to the increased mutability. In these experiments, it was not clear whether dam⁺ was transcribed from its own promoter(s), from a promoter of the vector plasmid, or from both.

TABLE 3.

Spontaneous Rif^r mutation frequencies of various strains of S. marcescens and E. coli

Bacterial strain	Plasmid	Rifr mutation frequency/108 cellsa
S. marcescens
dam+	pBR322	3.3 ± 0.9 (1)
dam-1	pBR322	53.4 ± 18.5 (16)
dam+	pTO3C	288.3 ± 89.8 (94)
dam-1	pTO3C	297.3 ± 78.3 (96)
E. coli
dam+ (K-12s)		4.9 ± 1.2 (1)
dam-13::Tn9		55.0 ± 13.8 (11.3)
dam-13::Tn9	pTO1B	4.6 ± 1.3 (0.9)

^aThe data are means from three independent experiments and are given with standard deviations. The values in parentheses are relative levels. 

Complementation of an E. coli dam mutant.

The cloning of the dam gene of S. marcescens by screening of 2-AP-resistant E. coli dam-13 transformants (see above) already suggested that the dam gene of S. marcescens can complement a dam defect of E. coli. As shown in Table 3, the monocopy plasmid pTO1B with dam⁺ of S. marcescens also decreased the high mutability of an E. coli dam-13 mutant to the level seen in E. coli dam⁺ cells. This shows that the enzyme from S. marcescens can substitute for the E. coli enzyme in mutation avoidance.

DISCUSSION

The dam mutant of S. marcescens described here lacks N₆-adenine methylation of GATC sequences, is sensitive to the base analogue 2-AP and to EMS, shows increased spontaneous mutability, and is hypermutable by 2-AP, EMS, and UV. This phenotype is very similar to that of dam mutants of E. coli and S. typhimurium (17, 39) and therefore indicates that the Dam enzyme in S. marcescens functions in differential strand tagging at the replication fork for methyl-instructed mismatch repair of newly synthesized DNA. The mismatch repair is governed by the mutH, mutL, and mutS genes in E. coli (33), and a mutH gene has been detected in S. marcescens (47). Further support for the role of dam comes from the finding that overexpression of the dam gene strongly increased spontaneous mutability, probably by rapid methylation of newly synthesized DNA, eliminating strand discrimination. The sequencing of the dam gene indicated high amino acid sequence identity to the E. coli Dam enzyme. In addition, determination of adjacent nucleotide sequences revealed a similar embedding of the gene with the neighboring genes (urf 74.3-nucleotide sequence and dod) corresponding to those in E. coli. It has been estimated that E. coli and S. marcescens diverged about 200 million years ago (34). The identification in S. marcescens of a dam gene and its function establishes the role of dam in members of the family Enterobacteriaceae. Possibly, this holds true also for several other groups in the γ subdivision of Proteobacteria which show GATC methylation. It was proposed that they have acquired this trait rather recently in evolution (5). Since crystallization of the E. coli Dam protein appears to be problematic due to aggregation, structural characterization of the protein has not been achieved (41). It is hoped that this problem can be solved with the enzyme of S. marcescens.

The previously described dam mutant of another member of the γ subdivision of Proteobacteria, V. cholerae, remains somewhat an enigma. The mutant is sensitive to 2-AP, methyl methanesulfonate, and UV but retains methylation of GATC sequences and normal mutability (4). The cloned gene, when highly overexpressed from a multicopy plasmid, can complement an E. coli dam mutant (alleviation of UV and 2-AP sensitivity and increased spontaneous mutability) but does not increase the spontaneous mutability of E. coli or V. cholerae (3), as seen with the overexpressed Dam enzymes of E. coli (21, 30) and S. marcescens (Table 3). The enzyme is relatively small (192 aa) and has no significant amino acid sequence identity with the Dam enzyme of E. coli or S. marcescens. Perhaps the described Dam of V. cholerae is part of a restriction-modification system and exists in V. cholerae in addition to another Dam enzyme related to mismatch correction. How lack of the identified Dam in V. cholerae would interfere with DNA repair remains unclear.

Spontaneous mutability to resistance to various antibiotics was differentially increased in the dam mutant of S. marcescens. Whereas the Rif^r mutation frequency was elevated about 30-fold, the Tc^r frequency was not different from that of the wild type (Table 2). It is possible that GATC sequences are not present in the corresponding chromosomal region or that existing GATC sequences are specifically not methylated (18, 38) so that methyl-directed mismatch repair is inefficient in that region. Sites with GATC refractory to complete methylation have been identified on the E. coli chromosome (38). This would be consistent with the already high Tc^r mutation frequency in wild-type cells and suggests that lack of methylated GATC sequences in that region would not keep mutability down. In the genome of E. coli, the frequency of GATC sequences is correlated with the efficiency of mismatch repair (23) and the probability of repair of a mismatch decreases with the distance of the mismatch from the next GATC site (11). In this context, it is interesting that overexpression of dam⁺ causes a five- to sixfold higher spontaneous mutation frequency than that resulting from dam deficiency (Table 3). Possibly, when dam is overexpressed, mismatch correction is close to zero. Compared to this, in wild-type cells, about 99% of replication errors are corrected, and in dam-deficient cells, substantial mismatch correction by the mutH LS system is still apparent (Table 3). This would be consistent with previous studies (30) and with the finding that MutH, which is the nuclease component of the mismatch repair system, cleaves DNA when hemimethylated or nonmethylated but not when fully methylated (1, 48). Another explanation for the lower mutability of dam mutants than that of dam⁺-overexpressing strains could be that frequent MutHLS-caused double-strand breaks leave mutants inviable (35).

A phenotypic difference between the dam mutant of S. marcescens and those of E. coli was observed with respect to UV-induced mutability. Much higher numbers of mutants relative to the wild type were induced per UV dose in the dam mutant strain of S. marcescens compared to the corresponding numbers in E. coli (Fig. 1; 17). Glickman et al. (17) postulated that the methyl-directed mismatch repair system would reduce direct mutagenesis resulting from wrong single-base incorporations during replication but not indirect mutagenesis caused by nonpairing DNA lesions like pyrimidine dimers, which obstruct DNA synthesis. Such lesions would lead to mutation via SOS induction of the error-prone repair pathway. However, the repair of UV damage in DNA requires extensive repair replication (46) in the form of short patches (12 bases per lesion) or long patches (up to 2,000 bases per lesion). Therefore, in a dam mutant, misincorporations during repair replication would also lead to direct mutagenesis. If the contributions to mutation avoidance of proofreading by DNA polymerase and of mismatch correction were more on the side of mismatch correction in S. marcescens than in E. coli, then a higher mutability of the dam mutant of S. marcescens after DNA damage requiring repair replication could be expected, as was found. Alternatively, the repair after UV irradiation could more frequently involve long-patch repair in S. marcescens than in E. coli.