|Home | About | Journals | Submit | Contact Us | Français|
A novel, site-specific, DNA backbone S-modification (phosphorothioation) has been discovered, but its in vivo function(s) have remained obscure. Here, we report that the enteropathogenic Salmonella enterica serovar Cerro 87, which possesses S-modified DNA, restricts DNA isolated from Escherichia coli, while protecting its own DNA by site-specific phosphorothioation. A cloned 15-kb gene cluster from S. enterica conferred both host-specific restriction and DNA S-modification on E. coli. Mutational analysis of the gene cluster proved unambiguously that the S-modification prevented host-specific restriction specified by the same gene cluster. Restriction activity required three genes in addition to at least four contiguous genes necessary for DNA S-modification. This functional overlap ensures that restriction of heterologous DNA occurs only when the host DNA is protected by phosphorothioation. Meanwhile, this novel type of host-specific restriction and modification system was identified in many diverse bacteria. As in the case of methylation-specific restriction systems, targeted inactivation of this gene cluster should facilitate genetic manipulation of these bacteria, as we demonstrate in Salmonella.
The high G+C Gram-positive bacterium Streptomyces lividans 66 contains DNA that is degraded by oxidative cleavage during electrophoresis in Tris-acetate buffer [the so-called DNA degradation (Dnd) phenotype] (1). Double-strand cleavage occurs at specific sites which contain a non-bridging S atom attached to backbone phosphorus at rare but specific sites (2,3). We have cloned and sequenced the S. lividans dnd gene cluster responsible for DNA S-modification. The proposed biochemical pathway leading to DNA S-modification has been shown to involve five putative proteins encoded by the dnd gene cluster (4–6), but the physiological function of DNA backbone S-modification in vivo has remained obscure. Previously, post-replicative host-specific cytosine and adenine modification (7,8) of DNA by methylation, hydroxymethylation and glucosyl-hydroxymethylation has been shown to be involved in DNA replication and repair, control of genomic imprinting, regulation of gene expression, and host-specific DNA restriction–modification (RM) systems in almost all organisms (9–12).
Here, we selected one of at least 10 Salmonella enterica serovar strains among more than 100 strains (13,14) from phylogenetically diverse bacteria which exhibited the Dnd phenotype (15) and contained homologs of the S. lividans dnd gene cluster to investigate whether a corresponding DNA restriction system related to S-modification system could exist in S. enterica, functioning like DNA methylation in prokaryotes for the protection of homologous DNA against host-specific restriction of foreign DNA (12,16).
Bacterial strains and plasmids used in this study are described in Table 1. Salmonella enterica serovar Cerro 87 isolated from an egg-producing farm was a non-pathogenic strain displaying the Dnd phenotype.
Degenerate oligonucleotide primers p1/p2 (p1: 5′ CTACTCGTTTCCGGCTATHCGNGG 3′; p2: 5′ ATCCTAGTTGCCCAAGNGCNTGCA 3′), p3/p4 (p3: 5′ GCGTNGNCAGCAGATGTTCGCCGA 3′; p4: 5′ GGAAGAATCTTTNCCNCCGCTGTA 3′) and p7/p8 (p7: 5′ TAGTTTCCGTTGGTGCACNGANCG 3′; p8: 5′ GCAATTTTCTCGCCRTCGAARAAG 3′), designed based on the highly conserved putative dnd genes in S. lividans 66, Streptomyces avermitilis NRRL8165, Pseudomonas fluorescens Pf0-1, Nostoc PCC7120, Nostoc punctiforme and Bacillus cereus E33L, were synthesized and used to amplify the corresponding fragments from S. enterica serovar Cerro 87 genomic DNA. The specific primers of p10/p11 (p10: 5′ GATTTGGCCGAATACGAAGA 3′; p11: 5′ TTTCCATCGCTTTATCTTTG 3′), according to the sequences of fragments p3p4 and p7p8, were used as a probe to amplify a 0.9-kb fragment and cloned into pMD-18T, giving pJTU1231 and sequenced (Supplementary Figure S1B). The association of this cloned probe with the Dnd phenotype was confirmed by the construction of mutant XTG101 through homologous recombination, which interrupted the dptC region and abolished the Dnd phenotype (Supplementary Figure S1D).
The genomic fosmid library of S. enterica serovar Cerro 87 was constructed using CopyControl™ Fosmid Library Production Kit from EPICENTER® Biotechnologies.
Six fosmids were picked using polymerase chain reaction (PCR) sib-selection with primers p10/p11 (Supplementary Figure S1B). Restriction patterns and sequence analysis localized the dpt gene cluster on a 7.8-kb core region comprised of a 4.4-kb and a 3.4-kb EcoRI fragments, which were digested from fosmid 8F4 and inserted into the EcoRI site of pUC18 for the generation of pJTU1232 and pJTU1233, respectively.
A 2.5-kb EcoRI–ScaI fragment from pJTU1233 was ligated into EcoRI–EcoRV-digested pBluescript II SK (+) to generate pJTU1237. A 4.2-kb BglII–EcoRI fragment from pJTU1232 was inserted into the BamHI–EcoRI sites of pJTU1237, generating pJTU1238, carrying a 6.7-kb functional dptB-E gene cluster.
Fosmid 6G12 was selected to sequence based on the restriction phenotype (Figure 2). A c. 20-kb BglII fragment carrying the modification and restriction gene cluster from 6G12 was ligated into BamHI-digested pCC1FOSTM to generate pJTU3818.
A 5.9-kb PvuII/ScaI fragment (modification genes) from pJTU1238 was cloned into the SmaI site of pUC18, generating pJTU2489. An 8.9-kb NsiI-digested fragment (restriction genes) from 6G12 was then ligated into the PstI site of pJTU2489, giving pJTU2492, carrying the whole cluster except for ORF5. pJTU2492 was digested with SalI and self-ligated to generate pJTU3809 with partial deletion of dptH.
A 0.7-kb EcoRI–EcoRV fragment from pJTU1231, carrying a 0.9-kb PCR product originated from using primers p10/p11, was cloned into the corresponding sites of pKD46, resulting in pJTU1239. After introduction by transformation into S. enterica serovar Cerro 87 at 30°C, mutant XTG101 was obtained by single crossover at 42°C selection, which disrupted dptC (Supplementary Figure S1).
The following were the strategies to construct mutants XTG102, XTG104 and XTG105: first, amplifying left and right arms using total DNA isolated from S. enterica serovar Cerro 87 as template and introducing the corresponding restriction sites; second, amplifying the recombinant fragment using the mixture of left and right arms in the proper ratio as templates, overlapping by 40bp, and then cloning into the thermo-sensitive plasmid pKOV-Kan digested with BamHI–SalI; third, introducing the generated plasmids by transformation into Escherichia coli DH5α (pJTU1238) to be phosphorothioated; fourth, introducing the phosphorothioated plasmids into S. enterica serovar Cerro 87 at 30°C and obtaining single crossover intermediates at 43°C; finally, obtaining the mutant through double crossing over on an LA plate with 15% sucrose at 43°C.
For XTG102, dptLL/dptLR (dptLL: 5′ GACCTCGAGTTGGTTTTCAATA 3′, XhoI underlined; dptLR: 5′ AAGCAACCGTGTCAAGGTAATTGGCGTTGCTGCGTGGTTA 3′) were used for the 752-bp left arm, dptRL/dptRR (dptRL: 5′ TAACCACGCAGCAACGCCAATTACCTTGACACGGTTGCTT 3′; dptRR: 5′ ATAGGATCCACGACGCCTACAAAT 3′, BamHI underlined) for the 819-bp right arm and dptLL/ dptRR for the 1531-bp entire homologous recombination region with introduced BamHI and XhoI sites, generating pJTU2468 with deletion of the modification genes dptB-E.
For XTG104, FLL/FLR (FLL: 5′ AATGTCGACTATCGTCCGAATAAATA 3′, SalI underlined; FLR: 5′ AGCACCAGCACTTTCCCTGCTGCCGTAGATCCCGCACTTA 3′) were used for the 535-bp left arm, FRL/FRR (FRL: 5′ TAAGTGCGGGATCTACGGCAGCAGGGAAAGTGCTGGTGCT 3′; FRR: 5′ CTTGGATCCTGTTAATCAGGGTATCG 3′, BamHI underlined) for the 580-bp right arm and FLL/FRR for the 1075-bp entire homologous recombination region with introduced BamHI and SalI sites, generating pJTU3828 with a 921-bp deletion in dptF.
For XTG105, GLL/GLR (GLL: 5′ TGGGTCGACTTGCGAAGTCATCTT 3′, SalI underlined; GLR: 5′ GGACGATAGCCATTATTCAACTGGAATTTAGGGTAGTGTT 3′) were used for the 537-bp left arm, GRL/GRR (GRL: 5′ AACACTACCCTAAATTCCAGTTGAATAATGGCTATCGTCC 3′; GRR: 5′ CGCGGATCCAATTGGCAAACAGCGTAC 3′, BamHI underlined) for the 604-bp right arm and GLL/GRR for the 1101-bp entire homologous recombination region with introduced BamHI and SalI sites, generating pJTU3829 with a 762-bp deletion in dptG.
The concentrations of plasmid DNA for restriction test, purified using Qiagen Plasmid Midi Kit, were determined by the absorbance at 260nm and the purity was estimated with the ratio of A260/A280.
Electroporation-competent cells of S. enterica strains were prepared as described in Sambrook et al. Electroporation was done using a Cell-Porator with a voltage booster and 1-mm gap cuvette according to the manufacturer's instructions (GIBCO/BRL) by using 50μl of cells and 5µl plasmid DNA (~10ng). Shocked cells were added to 1ml LB, incubated at 30°C for 30min, and then spread onto agar containing antibiotic to select transformants.
The CaCl2-heat shock method was used for introducing plasmids into E. coli. The competent cells were prepared according to Sambrook et al. 5µl (~10ng) of plasmid DNA and 95µl of competent cells were mixed well and kept in ice for 30min before being heat-shocked for 90s at 42°C and placed in ice for another 5min till adding 900µl of LB. The mixture was incubated at 30°C for 30min and plated on L-agar containing antibiotic(s).
In bacteria, site-specific DNA modification (usually methylation) is often, but not always, associated with a sequence-specific endonuclease. These host-specific restriction systems prevent the introduction of heterologous DNA from phages and plasmids that lack the specific DNA modification. We first tried to test whether pUC18 DNA isolated from E. coli was restricted by S. enterica serovar Cerro 87. Figure 1A shows the results of transformations of S. enterica- or E. coli-competent cells using equal amounts of pUC18 isolated from E. coli (S− DNA) or S. enterica (S+ DNA). Salmonella enterica reproducibly yielded about 100 times fewer colonies with S− DNA from E. coli than with S+ homologous DNA from S. enterica, while about the same number of transformants could be obtained when E. coli DH5α was used as the host.
Plasmids from E. coli that had escaped restriction by S. enterica were no longer restricted by S. enterica. When these plasmids were passaged through E. coli, they were again restricted by S. enterica, indicating that the plasmids were not mutants lacking the S-modification site. These results suggested that S. enterica might specifically restrict DNA lacking the S-modification.
To investigate whether this restriction phenotype was related to DNA S-modification, we cloned the dnd homologous gene cluster of the enteropathogenic S. enterica serovar Cerro 87. The S. lividans gene cluster was not suitable as a hybridization probe because it has a higher G+C content (66%) than Salmonella (48%). We designed degenerate oligonucleotide primers according to the most highly conserved region of the S. lividans dnd genes and amplified a fragment of the expected size (0.9kb) within the S. enterica DptC coding sequence. Using the fragment, an S. enterica mutant strain that had S-free (unmodified) DNA was obtained by single crossover integration of a temperature-sensitive plasmid. This proved that the amplified fragment originated from the dnd homologous dpt (DNA phosphorothioation) gene cluster responsible for the S-modification of S. enterica DNA (Supplementary Figure S1A and B). The dpt gene cluster probe was then used to screen a fosmid library from the genome of S. enterica serovar Cerro 87. PCR sib-selection identified six overlapping clones covering four genes (dptB-E) in the same order as the dndB-E genes of S. lividans, whose encoded proteins were significantly similar.
The S-modification cluster was cloned into pBluescriptII SK(+) to give pJTU1238, which is compatible with pKOV-Kan, a vector with a kanamycin resistance marker. The presence of this plasmid conferred S-modification on pKOV-Kan and prevented its restriction by S. enterica (Figure 1B-1 and B-2). The S. enterica mutant strain XTG102, from which the genomic dptB-E operon was removed by gene replacement, gave S-free (Dnd−) pKOV-Kan DNA that was 100-fold restricted by the wild-type strain (Figure 1B-3 and B-4). These experiments proved that the S-modification and not another, unknown, DNA modification protected heterologous DNA from the as-yet unknown S. enterica serovar Cerro 87 restriction system.
It was surprising that S. enterica XTG102, which lacks the entire gene cluster responsible for DNA S-modification, was viable because it was thought to contain restriction genes without the protective S-modification genes, or a mutation might have removed the restriction function when the strain was generated by double-crossover recombination. Lack of restriction was confirmed by transformation (Figure 1B-5), and the cloned dptB-E gene cluster in pJUT1238 did not confer restriction on E. coli (Figure 1B-7). However, XTG102 became normally restricting after the cloned dptB-E gene cluster was introduced using pJUT1238 (Figure 1B-6). This suggested that the S. enterica restriction system may require the presence of the S-modification proteins for activity, and it seemed likely that additional genes required for restriction alone could be found on the fosmid clones near the dptB-E genes.
The six fosmids mentioned earlier carrying dptB-E were introduced by transformation into E. coli DH5α. Two conferred the restriction phenotype on E. coli DH5α (Figure 2). A 20-kb fragment from fosmid 6G12 was cloned into pUC18 to generate pJTU3818, which retained the ability to confer the restriction phenotype on E. coli DH5α. This suggested that the restriction genes were located on this fragment (Figure 3). The 20-kb region of pJTU3818 was sequenced (Genbank Accession number GQ863484), and eight open reading frames (ORFs), including dptB-E, related to DNA S-modification were identified. Four other genes, ORF5, dptF, dptG and dptH, showed no convincing similarity to any proteins in the databases, nor any pfam matches. In order to precisely identify these genes, a series of chromosomal in-frame (to avoid polar effects) deletion mutants were generated in S. enterica serovar Cerro 87. The strains XTG104 and XTG105, with deletions in dptF and dptG, respectively, lost the restriction phenotype. Surprisingly, pJTU2492, carrying a DNA fragment within which ORF5 was deleted, still restricted foreign plasmids when expressed in E. coli DH5α, suggesting that ORF5 was not essential for restriction. Escherichia coli, DH5α containing pJTU3809 (ORF5 and dptH deleted), did not restrict foreign plasmids. These results unambiguously demonstrated that dptF-H were absolutely essential for restriction in S. enterica serovar Cerro 87 (Figure 3). In summary, this is a complex restriction system that depends for its activity on the presence of S-modification (phosphorothioation) genes.
DNA from many bacteria is degraded during normal electrophoresis or pulsed-field gel electrophoresis in the presence of Tris, which is diagnostic for DNA S-modification (15). Nineteen dpt-like gene clusters from diverse strains were found in the DNA databases (Figure 4). Escherichia coli B7A was selected for further tests: it possesses the Dnd phenotype typical of S-modification in its DNA (Supplementary Figure S5A) and, in addition, S-free plasmids from S. enterica serovar Cerro 87 mutant XTG102 transformed E. coli B7A with 100-fold lower efficiency than S-modified plasmid DNA from S. enterica serovar Cerro 87 (Supplementary Figure S5B). This confirmed the prediction from sequence comparisons that a host-specific restriction system associated with DNA S-modification exists in E. coli B7A, and probably in many other bacteria of diverse origin.
Our results have established, at least for S. enterica, a biological role for the previously identified DNA backbone phosphorothioation: in a cluster of eight genes, seven are evidently the determinants for host-specific R–M. The dptB-E genes are essential for phosphorothioation, and the dptF-H genes are all required for restriction of unmodified DNA. The novel host-specific restriction and S-modification system described here is similar to the earlier extensively studied R–M system as prokaryotic immune systems that attack foreign DNA entering the cell. The R–M systems are traditionally classified into four major groups: type I, II, III and IV on the basis of subunit composition, sequence recognition, cleavage position and cofactor requirements. In types I, II and III, foreign DNA is inactivated by endonucleolytic cleavage, the host DNA is normally methylated within specific sequence by the congnate methyltransferase and protected against restriction. In type IV, only modified foreign DNA is cleaved (22,23). Here, experiment results suggested the host-specific R–M system might be close to type I R–M system. The host-specific restriction system is composed of seven genes, dptBCDE are essential for phosphorothioation and the dptFGH genes are all required for restriction of unmodified DNA. Restriction activity required dptFGH genes in addition to at least four contiguous dptBCDE genes necessary for DNA S-modification. This functional overlap ensures that restriction of heterologous DNA occurs only when the host DNA is protected by phosphorothioation. The composition of genes and functional overlap are very similar to the type I R–M system, which are composed of three subunits that typically contain two REase subunits that are required for DNA cleavage, one specificity subunit that specifies the DNA sequence recognized, and two MTase subunits that catalyse the methylation reaction. The genes for DNA S-modification and restriction also seemed to form a complex which was more complicated than type I R–M system. In addition, genes responsible for DNA sequence specificity and DNA cleavage site still remained obscure.
In methylation-specific R–M systems, a restriction gene is often linked to a modification gene, forming a so-called restriction–modification gene complex, but often solitary methyltransferases that do not have a restriction-enzyme counterpart (7) could also be found. This also seems to be true for the phosphorothioation-specific restriction systems. Among 20 taxonomically diverse bacteria, including low G+C Gram-positive Bacillus species (Figure 4), identified as carrying the syntenous dptB-E homologous gene clusters, all possessed dptF-H homologs (required for restriction only) in same order, but many other bacteria contain only dptB-E (for DNA S-modification) without dptF-H homologs. We suspect that, as in methylation-specific restriction systems, DNA S-modification may act not only as a sort of immune system, allowing the bacteria to protect themselves from infection by bacteriophages, but also as an epigenetic signal for new biological function(s).
We do not know which of the dndA counterparts is functional for phosphorothioation in S. enterica serovar Cerro 87, but dndA counterparts, whose function could likely be recruited from shared pathway(s), were found to be separated from eight of the 20 syntenous dptB-H counterpart clusters. The high degree of synteny of the dptB-H counterparts therefore provides strong support for the idea that dptB-H counterparts might be a later addition by horizontal transfer of genomic island(s) (15). This agrees well with the fact that many restriction–modification gene complexes reside on plasmids and prophages, which lends mobility (24), and host-specific restriction systems may be part of a mobile element near to an integrase (15) and therefore able to move freely between genomes. Additionally, methylated DNA is known to be an essential factor in Salmonella virulence, and its absence causes severe attenuation (25), but it was not clear whether S-modified DNA could play a similar role in host–pathogen interactions. We also attempted in vitro restriction enzyme assays to detect the presence of a host-specific restriction system, but these enzymes did not produce any distinctive DNA fragments in the crude extracts.
The discovery described here is, to our knowledge, a first report of a host-specific restriction system associated with S-modification of DNA instead of methylation to protect homologous DNA. The systems might be useful for engineering host resistance against in vivo exo- or endo-nucleolytic degradation, and/or for transfer into heterologous cells to control the detrimental effects of phage contamination in the fermentation industry.
Supplementary Data are available at NAR Online.
The National Science Foundation of China; the Ministry of Science and Technology; China Ocean Mineral Resources R & D Association (DYXM11502203); the Shanghai Municipal Council of Science and Technology (05ZR14069); and Shanghai Leading Academic Discipline Project B203. Funding for open access charge: Ministry of Science and Technology of China (grant 2009ZX09501-008).
Conflict of interest statement. None declared.
The authors thank Dr. Tobias Kieser, Prof. Peter C. Dedon and Prof. Sir David Hopwood, FRS for valuable discussion and critical reading of the manuscript and Prof. Toshiyuki Murase for supplying S. enterica serovar Cerro 87.