Plasmid DNA of six Escherichia fergusonii colicinogenic strains (three producers of colicin E1, two of Ib and one of Ia) was isolated and the colicin-encoding regions of the corresponding Col plasmids were sequenced. Two new variants of colicin E1, one of colicin Ib, and one of colicin Ia were identified as well as new variants of the colicin E1 and colicin Ib immunity proteins and the colicin E1 lysis polypeptide. The recombinant Escherichia coli producer harboring pColE1 from E. fergusonii strain EF36 (pColE1-EF36) was found to be only partially immune to E1 colicins produced by two other E. fergusonii strains suggesting that pColE1-EF36 may represent an ancestor ColE1 plasmid.
Colicinogeny (i.e. the ability of bacterial strains to produce colicins) is one of the frequent characteristics of Escherichia coli strains isolated from both natural and clinical isolates. Approximately 40% of E. coli strains of human origin have previously been shown to be colicinogenic .
Within the group of 50 Escherichia fergusonii strains mostly of human origin, six strains (12%) were identified as colicinogenic (J. Šmarda, unpublished results). E. fergusonii was established as a new species of the genus Escherichia in 1985 and is most closely related to E. coli and Shigella sp. and more distantly related to other bacterial species of the family Enterobacteriaceae.
To test whether the close relatedness of E. fergusonii strains to strains of E. coli and Shigella sp. also reflects similarities in colicin types produced by E. fergusonii we characterized six individual Col plasmids isolated from E. fergusonii colicinogenic strains. In this communication we describe colicin-encoding regions of E. fergusonii-derived Col plasmids and compare the corresponding colicin, immunity and lysis proteins to those described in E. coli.
Materials and methods
Bacterial strains and growth conditions
All E. fergusonii colicinogenic strains were provided by the National Reference Laboratory for E. coli and Shigellae, Center of Epidemiology and Microbiology, National Institute of Public Health, Prague, Czech Republic. All colicinogenic E. fergusonii strains were of human origin. E. coli TOP10F′ (Invitrogen, San Diego, CA, USA) was used as a colicin indicator. Bacterial strains were grown at 37°C in TY medium containing 8 g Bacto-tryptone (Difco Laboratories, Sparks, MD, USA), 5 g yeast extract and 5 g NaCl per liter (pH 7). For selection and maintenance of plasmids, 25 μg of chloramphenicol or 25 μg of kanamycin per ml of liquid medium and 1.5% TY agar (w/v) were added.
Crude colicin preparations and colicin activity assays
Cells from the TY cultures of colicinogenic E. fergusonii or recombinant E. coli strains (producers of colicins E1, Ib and Ia) induced by mitomycin C (Sigma, St. Louis, MO, USA; 0.5 μg ml−1) were harvested, resuspended in distilled water, washed and sonicated. The sonicates were used as crude colicins. Colicin activity was tested by spotting 10-fold dilutions of colicin-containing crude cell lysates on agar plates seeded with sensitive bacteria; TY agar plates were overlaid by 3 ml of 0.75% (w/v) TY agar with 100 μl of an overnight culture of indicator bacteria. Each experiment was performed at least three times and the data represent the average of three independent measurements.
In vitro transposition
To isolate the plasmid governing colicin synthesis and to sequence the colicin-encoding region, two subsequent in vitro transposition steps were performed with Tn5- and Tn7-based in vitro transposition systems. For Tn7 plasmid mutagenesis, an in vitro Tn7 transposition system (GPS™-1 Genome Priming System, New England Biolabs) was used according to the manufacturer's recommendations. An EZ::TN™ pMOD™-2 Transposon Construction Vector (Epicentre Technologies, Madison, WI, USA) with a cloned kanamycin resistance gene served as the DNA template for amplification of the Tn5 transposon, and transposition was performed according to the EZ::TN™ TET-1 Insertion Kit (Epicentre Technologies) protocol. The DNA sequencing reactions were directed from both ends of the Tn5 transposon using pMODFseq (5′-GCCAACGACTACGCACTAGCCAAC-3′) and pMODRseq (5′-GAGCCAATATGCGAGAACACCCGAGAA-3′) primers. Alternatively, Tn7RN and Tn7LS primers recognizing the ends of Tn7 transposon were used for sequencing (Tn7RN: 5′-ACTTTATTGTCATAGTTTAGATCTATTTTG-3′; Tn7LS: 5′-TATTAGGAATTTTTGAGGTAAAGGTGGGGA-3′).
Isolation of plasmid DNA, restriction analysis and sequencing
Standard methods were used for plasmid isolation, restriction endonuclease analysis and agarose gel electrophoresis . DNA was sequenced using the Taq Dye-deoxy Terminator method and a model 377 DNA sequencing system (Applied Biosystems, Foster City, CA, USA). The complete sequence of colicin activity, immunity and lysis genes was finished using specifically designed synthetic oligonucleotides. Computer-assisted sequence analysis was performed using the LASERGENE program package (DNASTAR, Madison, WI, USA).
Nucleotide sequence accession numbers
The nucleotide sequences reported in this study were deposited in the GenBank under the accession numbers AF453410–AF453415.
Results and discussion
Among 50 strains of E. fergusonii investigated, six colicinogenic strains were identified (J. Šmarda, unpublished results). Plasmid DNA from all six colicinogenic strains was isolated and colicin-encoding regions were identified and sequenced using in vitro transposon insertions.
E. fergusonii EF43
Tn5 insertions into plasmid DNA isolated from the E. fergusonii EF43 strain and screening for colicinogenic transformants resulted in an E. coli TOP10F′ strain harboring colicinogenic plasmid pDS455. The original colicinogenic plasmid of E. fergusonii EF43 strain was named pColE1-EF43. The activity and immunity genes encoding colicin E1 (cea) and immunity protein (imm) encoded by pColE1-EF43 were completely identical (Fig. 1) to cea and imm reported previously . However, the kil gene on pColE1-EF43 had one nucleotide change (C71G) resulting in a one-amino acid replacement (P24R) in the corresponding polypeptide when compared to the peptide encoded by kil on pColE1 . Restriction digestion of pColE1-EF43 revealed its molecular size to be 6.7 kb (data not shown), which is close to the size of the sequenced pColE1 (6646 bp; ), indicating that those two ColE1 plasmids are closely related.
E. fergusonii EF3
Plasmid pColE1 from strain E. fergusonii EF3 (pColE1-EF3) was isolated after insertion of Tn7 into the plasmid backbone, resulting in the colicinogenic plasmid DS300. The colicin E1 cea gene encoded by this plasmid had a one-nucleotide replacement (G211A) when compared to cea of pColE1-EF43, resulting in a one-amino acid change (A71T) in colicin E1 protein (Fig. 1). pColE1-EF3 imm and kil gene sequences were identical to those of pColE1-EF43. Consistent with this finding, E. coli strain TOP10F′ with pDS455 was immune to colicin E1 produced by EF3 strain and vice versa (Table 1). The size of pColE1-EF3 was determined to be approximately 9 kb (data not shown), which is considerably more than for pColE1-EF43.
|E. coli strain||Relevant genotype||Colicin E1 from E. fergusonii strain|
|E. coli strain||Relevant genotype||Colicin E1 from E. fergusonii strain|
aThe numbers indicate the exponents of the highest colicin dilution that resulted in a clear zone of growth inhibition and the last dilution that resulted in turbid zones (in parentheses) on the lawn of sensitive bacteria, e.g. 3=103.
E. fergusonii EF36
pDS503, a result of Tn7 insertion into pColE1 from strain EF36 (pColE1-EF36), encoded a cea gene that was most closely related to cea of Shigella sonnei pKY-1 plasmid (97.2% identity; ), differing by 16 nucleotide changes and insertion of 7 bp and 1-bp deletion. However, the EF36 colicin E1 protein sequence was most closely related (96.6% identity) to colicin E1 encoded by pColE1-EC39 isolated from E. coli EC39 , differing in 12 amino acid replacements and insertion of two amino acid residues. Cea of pColE1-EF36 is more distantly related to Cea encoded by pKY-1 (92.9% identity; Fig. 1). This discrepancy was because the 90-bp region between bp 342 and 432 of pKY-1 cea is translated from a different reading frame. Cea of EF36 differs in 68 amino acid residues from Cea encoded on pColE1-EF43 (corresponding genes differ in 205 nucleotides) and the amino acid replacements are preferably localized in the middle third of the colicin E1 molecule. The pColE1-EF36 imm gene was most closely related to the imm from E. coli EC39 strain , differing in four nucleotides that resulted in a one-amino acid replacement (E81V). The Imm proteins from EF36 and EF43 strains were different in eight amino acid residues (92.9% identity). The E. coli TOP10F′ pDS455 producer of colicin E1 was completely immune against colicin from the EF36 strain, but the protection of the E. coli TOP10F′ pDS503 producer is at least two orders of magnitude lower (Table 1). These data indicate that the immunity proteins encoded on pColE1-EF36 and pColE1-EF43 differ in the recognition of colicin E1 variants. In the interaction between colicin E1 and its cognate immunity protein, four critical amino acids were identified . Since colicins E1 encoded by pColE1-EF43 and pColE1-EF36 differ in one of these four amino acid residues (A472 and V473 in colicin E1 of EF43 and EF36, respectively), it is possible that this residue is involved in different recognition by immunity protein encoded on pColE1-EF36. The kil genes encoded on both pColE1-EF36 and pColE1-EC39 were identical  and the Kil of EF36 differed from the EF43 Kil in two amino acid residues (I31V, A43S). The plasmid size of pColE1-EF36 was estimated to be 7.4 kb (data not shown).
Sequences of colicin E1-encoding regions for strains EF43, EF3 and EF36 were deposited in the GenBank under accession numbers AF453410, AF453411, AF453412, respectively.
E. fergusonii EF6 and EF24
E. fergusonii producers of colicin Ib, strains EF6 (pColIb-EF6) and EF24 (pColIb-EF24), were found to have identical sequences in colicin Ib activity and immunity genes on ColIb plasmids. When compared to the sequence of the colicin Ib structural gene encoded on E. coli plasmid pColIb-P9 , 16 nucleotide changes were detected out of 1881 nucleotides. The colicin Ib proteins encoded by pColIb-EF6 and pColIb-EF24 differed in eight amino acid residues (98.7% identity) from the colicin Ib encoded by pColIb-P9 (Fig. 2). Five out of the eight substitutions are localized within the C-terminal 200-amino acid pore-forming domain. The colicin Ib immunity genes on pColIb-EF6 and pColIb-EF24 strains differed in four nucleotides from that encoded on pColIb-P9, and the corresponding proteins differed in three amino acid residues (97.4% identity; I17S, N31S, L108F). Strains E. coli pColIb-P9 and TOP10F′ pColIb-EF6 were cross-immune. Despite the identical sequence of colicin activity and immunity genes on pColIb-EF6 and pColIb-EF24, the HindIII restriction pattern for both plasmids differed (data not shown) indicating that both strains harbor unique pColIb plasmids. Sequences of the colicin Ib-encoding region for pColIb-EF6 and pColIb-EF24 were deposited in the GenBank under the accession numbers AF453413 and AF453414, respectively.
E. fergusonii EF31
E. fergusonii strain EF31 was identified as a producer of colicin Ia. The colicin Ia structural gene was most related to that encoded by pColIa-EC3  differing only in one nucleotide (G910C), resulting in one amino acid replacement (Fig. 2) in the resulting colicin Ia protein (D304H, 99.8% identity). The same degree of sequence homology was found between colicin Ia from the EF31 strain when compared to that encoded by pColIa-CA53  and colicin Ia from EF31 showed one amino acid substitution when compared to colicin Ia from CA53 (R351Q). The immunity gene encoded by strain EF31 was found to be identical to those encoded by pColIa-EC3 and pColIa-EC34 . Sequences of the colicin Ia-encoding region of EF31 were deposited in the GenBank under accession number AF453415.
The colicin-encoding regions of the Col plasmids isolated from E. fergusonii closely resembled the coding determinants of those isolated from E. coli. However, six of the E. fergusonii Col plasmids yielded four new colicin variants, two new colicin immunity protein variants, and one new variant of the colicin lysis protein.
The different spectra of inhibitory effects of colicin E1 and colicin Ib variants produced by E. fergusonii strains (J. Šmarda, unpublished results) on E. fergusonii indicator strains cannot be explained solely by the differences in the colicin protein sequences. The E. coli TOP10F′ strain with introduced colicinogenic plasmids encoding E. fergusonii variants of colicins E1 and Ib showed identical spectra against E. fergusonii indicator strains (data not shown) as the original E. fergusonii producers. The differences in activity spectra of E. fergusonii producers thus seem to reflect different levels of colicin synthesis as a result of Col plasmid differences together with a different degree of sensitivity of E. fergusonii indicators to colicins.
This work was partly supported by Grant no. 310/98/0083 from the Grant Agency of the Czech Republic.