13 ACTA BIOLOGICA SLOVENICA LJUBLJANA 2006 Vol. 49, [t. 2: 13–21 Sprejeto (accepted): 2006-09-28 Colicins of the Escherichia coli uropathogenic strain collection Kolicini zbirke uropatogenih bakterij Escherichia coli Marjanca STARČIČ ERJAVEC, Matija RIJAVEC, Darja ŽGUR-BERTOK Department of Biology, Biotechnical Faculty, University of Ljubljana, Večna pot 111, 1000 Ljub- ljana, Slovenia, Fax: 00386 1 257 3390, E-mail: Marjanca.Starcic.Erjavec@bf.uni-lj.si, r_matija@ yahoo.com, Darja.Zgur@bf.uni-lj.si. Abstract. 110 uropathogenic Escherichia coli (UPEC) strains were screened for colicin production and 42 (38%) of the tested UPEC strains were found to be colicinogenic. The ColA, ColB, ColD, ColE2, ColE3, ColE4, ColE5, ColE6, ColE7, ColIa, ColIb, ColK, ColN, MccB17, ColS4, MccC7 and ColE6-J colicin producer strains from Pugsley’s collection of colicinogenic strains were lysed by all colicinogenic UPEC strains, the ColM and ColE1 producer strains by 93% of the UPEC colicinogenic strains and the ColV producer strain by only 81% of the UPEC colicinogenic strains. 67% of the colicinogenic UPEC strains were able to lyse all 20 used coli- cin producer strains and 33% of the colicinogenic UPEC strains were able to lyse 19 Pugsley’s strains. Hence, a majority (67%) of the studied UPEC strains encode and produce either more than one colicin, or a colicin not tested. Colicins of UPEC strains producing only one colicin were identifi ed; 8 strains (19% of the colicinogenic strains) produced only ColV, 3 strains (7%) ColM and 3 strains (7%) ColE1. Plasmids were found in 88% of the colicinogenic strains. 11 DL strains were found to harbour conjugative plasmids encoding antibiotic resistance(s) and colicinogenicity. Further, 19% of the haemolytic UPEC strains and 44% of non-haemolytic strains were also colicinogenic, 28% of the cnf encoding strains and 41% of the strains not encoding cnf were colicinogenic, while 40% of ibeA encoding strains and 38% of strains not encoding ibeA were colicinogenic. Key words: colicin, uropathogenic Escherichia coli, UPEC, plasmid, haemolysin, hly, cytotoxic necrotising factor, cnf, invasin, ibeA. Izvleček. 110 uropatogenih sevov bakterije Escherichia coli (UPEC) smo s pomočjo 20 kolicinogenih sevov iz Pugsleyeve zbirke testirali za produkcijo kolicinov. 42 testiranih sevov (38%) je bilo kolicinogenih. Vsi kolicinogeni sevi UPEC so povzročili propad producentskih sevov kolicinov ColA, ColB, ColD, ColE2, ColE3, ColE4, ColE5, ColE6, ColE7, ColIa, ColIb, ColK, ColN, MccB17, ColS4, MccC7 in ColE6-J iz Pugsleyeve zbirke. 93% kolicinogenih sevov UPEC je povzročilo propad ColM in ColE1 producentskega seva in 81% kolicinogenih sevov UPEC je povzročilo propad producentskega seva kolicina ColV. 67% kolicinogenih sevov UPEC je povzročilo propad vseh 20 kolicinogenih sevov Pugsleyeve zbirke in 33% kolicinogenih sevov UPEC je povzročilo propad 19 sevov iz Pugsleyeve zbirke. Večina kolicinogenih sevov (67%) je sintetizirala vsaj dva kolicina, ali pa kolicin, ki ni bil testiran. Kolicine sevov UPEC, ki sintetizirajo samo 1 kolicin, smo prepoznali; 8 sevov UPEC (19%) je sintetiziralo samo ColV, 3 sevi (7%) so sintetizirali samo ColM in 3 sevi (7%) so sintetizirali samo ColE1. V 88% koli- cinogenih sevov smo našli plazmide. V 11 sevih DL smo našli konjugativne plazmide z zapisi za odpornost proti antibiotikom in za sintezo kolicinov. 19% hemolitičnih sevov UPEC in 44% nehemolitičnih sevov UPEC je bilo kolicinogenih; 28% sevov z zapisom cnf in 41% sevov brez 14 Acta Biologica Slovenica, 49 (1), 2006 zapisa cnf je bilo tudi kolicinogenih, medtem ko je bilo 40% z zapisom ibeA in 38% brez tega zapisa tudi kolicinogenih. Ključne besede: kolicin, uropatogena Escherichia coli, UPEC, plazmid, hemolizin, hly, citotoksičen nekrotizirajoč dejavnik, cnf, invazin, ibeA. Introduction Colicins are bacteriocins produced by Escherichia coli (E. coli) strains. As other bacteriocins, produced by different types of Eubacteria and Archaebacteria (RILEY & GORDON 1999), colicins are extracellular bacterial toxic proteins, that are active against the same species, or closely related species of the producer cell (DAW & FALKINER 1996). The mechanism of action of these compounds involves adsorption to specifi c receptors located on the external surface of sensitive bacteria followed by killing via one of three primary mechanisms: the formation of channels in the cytoplasmic membrane, the degradation of cellular DNA or the inhibition of protein synthesis (RILEY & GORDON 1999). Because of their narrow range of activity, it has been proposed that the primary role of bacteriocins is to mediate intraspecifi c, or population level, interactions (RILEY 1998). However, bacteriocins have been also implicated in virulence determination, since many patho- genic strains harbour plasmid-encoded bacteriocins, for example ColV (WATERS & CROSA 1991). A relatively high frequency of colicin producing strains is found in isolates of pathogenic E. coli (VAN DER WAL & al. 1995), for example approximately 80% of enterohemorrhagic E. coli strains studied by Bradley and Howard were colicinogenic (BRADLEY & HOWARD 1991). In the presented study a collection of 110 uropathogenic E. coli (UPEC) strains, isolated at the Institute of Microbiology and Immunology of the Medical Faculty of Ljubljana, Slovenia, was examined for the ability to produce colicins and lyse other E. coli strains of the Pugsley collection of colicinogenic strains. Further, the association of colicinogenicity and some established virulence factors was also analysed. The results of the study show, that almost 40% of the UPEC strains are colicinogenic and that all colicinogenic strains are very effi cient in lysing E. coli strains of the Pugsley collection. Further, a non-proportional distribution of the ability to produce colicins among the hly and cnf coding versus non-coding strains was determined. In 88% of the colicinogenic strains plasmids were found, 11 colicinogenic strains harboured conjugative plasmids encoding antibiotic resistances and colicin production. Methods Bacterial strains, plasmids and growth conditions The 110 uropathogenic Escherichia coli DL strains used in this study were isolated from urine of patients with urinary tract infections at the Institute of Microbiology and Immunology of the Medical Faculty of Ljubljana. The strain AB1133 [thr1 leuB6 proA2 his argE2 thi ara lacY galK2 xyl mtl rpsL α− supE (B. Bachmann)], which is sensitive to all colicins, was used to identify colicinogenic UPEC strains. The strains of Pugsley’s collection of colicinogenic strains are listed in Tab. 1. For mating experiments the following strains were used: DH5α [Φ80dlacZ∆M15 ∆(lacZYA-argF)U169 endA1 recA1 hsdR17 deoR thi-1 supE44 gyrA96 relA1 (BRL Life)], HB101 [hsdR hsdM recA13 supE44 leuB6 lacZ proA2 (D. Ehrlich)], RU4404 [MM294::Tn1725 Cmr thi endA hsdR (R. Schmitt)], RU4406 [MM294::Tn1732 Knr thi endA hsdR (R. Schmitt)] and TR51 [araD139 ∆(argF-lac)U169 rpsL150 relA1 fl bB5301 ptsF25 deoC1 cpxR::spc Spcr (T. J. Silhavy)]. Bacteria were grown in Luria-Bertani (LB) medium with aeration at 37°C or on LB plates without aeration. Ampicillin (Ap, 100 µg/ml), tetracycline (Tc, 10 µg/ml), kanamycin (Kn, 30 µg/ml), chloramphenicol (Cm, 50 µg/ml), spectinomycin (Spc, 20 µg/ml), streptomycin (Sm, 150 µg/ml), trimethoprim (Tp, 10 µg/ml) and nalidixic acid (Nal, 25 µg/ml) were added to the growth media, when appropriate. 15 Table 1: Bacterial strains of Pugsley’s collection. Tabela 1: Bakterijski sevi Pugsleyeve zbirke. Strain Relevant features BZB2101 ColA producer BZB2102 ColB producer BZB2103 ColD producer BZB2104 ColE1 producer BZB2125 ColE2 producer BZB2106 ColE3 producer BZB2107 ColE4 producer BZB2108 ColE5 producer BZB2109 ColE6 producer BZB2110 ColE7 producer BZB2114 ColIa producer BZB2115 ColIb producer BZB2116 ColK producer BZB2123 ColN producer BZB2283 MccB17 producer PAP1 ColM producer PAP2 ColS4 producer PAP54 MccC7 producer PAP222 ColV producer PAP247 ColE8-J producer Overlay test The overlay test was based on the method described by Pugsley and Oudega (1987). The UPEC strains were inoculated on LB plates (20 colonies grid) using toothpicks. Following overnight incubation the cells were lysed with chloroform, to release the colicins. After aeration, the plates were overlaid with 4 ml of soft agar with 0,2 ml of an overnight culture of either AB1133 or one of the strains from Pugsley’s collection. The plates were then incubated overnight and next day examined for zones of colicin activity (clear zones around the colonies). Plasmid isolation Plasmid DNA was prepared by the alkaline lysis method described in SAMBROOK et al. (1989). Mating assay Conjugation experiments were performed overnight on LB plates. The mating mixture was transferred to LB plates supplemented with appropriate antibiotics to select for transconjugants. The conjugative plasmids encoding antibiotic resistances were initially, depending on the antibiotic resistance profi le, transferred to either DH5α, HB101, RU4404, RU4406 or TR51 and further to a another laboratory strain to confi rm the self conjugative transfer ability. All conjugative plasmids were fi nally transferred to strain DH5α. Results Colicinogenic UPEC strains and their colicins In order to identify the colicinogenic UPEC strains, strain AB1133, which is known to be sensitive to all colicins, was used in an overlay test as described above. 42 UPEC strains (38%) were able to lyse strain AB1133, hence exhibited colicinogenic activity. To attempt to characterise colicin encoded by M. Starčevič Erjavec & al.: Colicins of the Escherichia coli uropathogenic strain collection 16 Acta Biologica Slovenica, 49 (1), 2006 the individual UPEC strains, the 20 strains of Pugsley’s collection of colicin producing strains were used in overlay tests. The tests showed, that most of the colicinogenic UPEC strains (67%) produced either more than one colicin, or a colicin, not harboured by strains of the Pugsley collection. However, 14 UPEC strains were shown to produce only one colicin, namely 8 strains (19% of the colicinogenic strains) produced only ColV, 3 strains (7%) ColM and 3 strains (7%) produced only ColE1 (Tab. 2). Table 2: Colicinogenic DL strains. Tabela 2: Kolicinogeni sevi DL. Strain Number of lysed Pugsley’s strains Identifi ed colicin Plasmid detected Conjugative plasmid DL2 20 + + DL3 20 + – DL6 20 + + DL10 20 + – DL14 20 + + DL22 20 + + DL24 20 + – DL27 19 ColV + – DL35 20 + – DL37 20 + + DL40 19 ColM – – DL46 20 + + DL48 20 + + DL49 20 + – DL51 20 + – DL52 19 ColV + – DL53 20 + – DL56 20 + + DL57 19 ColV + – DL58 20 + – DL59 20 – – DL60 20 + – DL62 19 ColV – – DL63 19 ColE1 + – DL64 19 ColV + – DL66 19 ColV + – DL67 19 ColV + – DL71 19 ColE1 + – DL72 19 ColE1 + – DL75 20 + – DL76 20 + + DL83 19 ColV + – DL89 19 ColM + – DL91 19 ColM + – DL92 20 – – DL93 20 + – DL95 20 – – DL99 20 + – DL104 20 + – DL107 20 + – DL108 20 + + DL110 20 + + 17 Sensitivity of Pugsley’s colicinogenic strains for the colicinogenic UPEC strains The sensitivity of Pugsley’s colicinogenic strains for the colicinogenic UPEC strains was also demonstrated in the overlay tests. The ColA, ColB, ColD, ColE2, ColE3, ColE4, ColE5, ColE6, ColE7, ColIa, ColIb, ColK, ColN, MccB17, ColS4, MccC7 and ColE6-J colicin producer strain from Pugsley’s collection were lysed by all colicinogenic UPEC strains, the ColM and ColE1 producer strains by 93% of the UPEC colicinogenic strains and the ColV producer strain by only 81% of the UPEC colicinogenic strains (Graph 1).The high sensitivity of the Pugsley colicinogenic strains for the colicinogenic UPEC strains is also evident from Tab. 2, which shows that 67% of the colicinogenic UPEC strains were able to lyse all 20 of the used colicin producer strains and 33% of the colicinogenic UPEC strains were able to lyse 19 of the Pugsley strains. The ability to produce colicins and plasmids Preparation of plasmid DNA with alkaline lysis from colicinogenic strains revealed that 37 (88%) of the colicinogenic strains harboured plasmids. In only 5 strains (DL40, DL59, DL62, DL92 and DL95) no plasmid DNA was detected (Tab. 2). Colicinogenic proprieties encoded on conjugative plasmids In the mating assays, 19 DL strains (DL2, DL6, DL7, DL8, DL14, DL17, DL22, DL37, DL41, DL43, DL46, DL48, DL56, DL76, DL81, DL84, DL108, DL109 and DL110) were found to harbour conjugative plasmids encoding antibiotic resistances. 11 of these strains (DL2, DL6, DL14, DL22, DL37, DL46, DL48, DL56, DL76, DL108 and DL110) were also found to be colicinogenic (Tab. 2). To determine whether the colicinogenicity is encoded on a conjugative plasmid (pDL), the DH5α laboratory strains harbouring the conjugative plasmids from the DL strains were subjected to overlay tests. When the DH5α strains harbouring pDL conjugative plasmids were overlaid with the Pugsley strains, all 20 of the Pugsley strains were lysed. Further, when the original DL strains were overlaid by DH5α strains harbouring pDL conjugative plasmids, the DH5α pDL strains were not lysed by the original DL strain. The obtained results clearly demonstrated that colicinogenicity in all of these strains was encoded by a conjugative plasmid and hence, was transferable to other strains. The ability to produce colicins and some other virulence factors The obtained data concerning colicinogenicity of the tested strains presented in this paper were compared with our unpublished data on the prevalence of virulence factors; namely haemolysin (hly), cytotoxic necrotising factor (cnf) and invasin (ibeA). Our results showed that 19% of the haemolytic UPEC strains and 44% of the non-haemolytic strains were also colicinogenic, 28% of the cnf encoding strains and 41% of the strains not encoding cnf were colicinogenic, while 40% of the ibeA encoding strains and 38% of the strains not encoding ibeA were colicinogenic (Fig. 1). M. Starčevič Erjavec & al.: Colicins of the Escherichia coli uropathogenic strain collection 18 Acta Biologica Slovenica, 49 (1), 2006 Fig. 1: Incidence of colicinogenic strains in correlation with some virulence factors. The incidence of colicino- genic strains is expressed in percentage of strains encoding or not encoding haemolysin (hly), cytotoxic necrotising factor (cnf) and invasin (ibeA). Slika 1: Pogostnost kolicinogenih sevov v povezavi z nekaterimi virulenčnimi dejavniki. Pogostnost kolicinoge- nosti je izražena v odstotkih sevov z oziroma brez zapis za hemolizin (hly), citotoksičen nekrotizirajoč dejavnik (cnf) in invazin (ibeA). Discussion The DL strain collection of uropathogenic E. coli strains was screened for colicin production. 38% of the examined strains exhibited in the performed overlay tests a phenotype consistent with colicin production. Since the overlay tests were performed without mitomycin C treatment, which is known to induce temperate bacteriophage, it can be assumed that the obtained results have not been misin- terpreted due to bacteriophage. However, to completely rule out the possibility of misinterpretation of bacteriophage activity as colicin production further tests would be needed. Typically, 25–50% of E. coli isolates are colicinogenic (RILEY & GORDON 1996). Usually, the percentages are higher in pathogenic than commensal strains, they are also higher in human compa- red to animal strains (RILEY & GORDON 1996). In this study the obtained percentage (38%) of colicin producing strains in the DL collection is comparable to that of other studies, MCGEACHIE (1965) found 36% of colicinogenic strains among uropathogenic E. coli strains and O’BRIEN et al. (1996) found 41% of colicin producers among uropathogenic E. coli strains. The obtained percentage is also com- parable to the incidence of colicinogenicity (41%) in strains from colons of healthy persons (ŠMARDA & OBDRŽALEK 2001). It is established that the frequency of different colicin types varies substantially between populations (RILEY & GORDON 1996), hence the colicin pattern in the DL collection should differ from patterns in other E. coli collections. The applied tests allowed the assignment of colicins of those strains producing only one colicin, namely ColV by 19% of the colicinogenic strains, ColM and ColE1 by 7% of the colicinogenic strains. Colicins of all other colicinogenic strains (67%) could not be assigned as these strains produce more than one colicin or they produce a colicin type not presented in the Pugsley collection of colicinogenic strains. However, the combination of ColV, ColM and ColE1 colicins in one collection is, to our knowledge, unique. High levels of colicin resistance are known to occur in natural E. coli populations (FELDGARDEN & RILEY 1998), hence the high sensitivity of Pugsley’s collection for the colicinogenic strains (the majority, 67%, of colicinogenic strain were able to lyse all 20 strains of Pugsley’s collection), is a bit surprising. But it is explainable with the assumption, that the UPEC strain encode more than one colicin or a colicin not produced by the Pugsley strains collection. 19 It is known, that the genetic determinants of most of the colicins are located on the plasmids, apart from few, which are chromosomally encoded (DAW & FALKINER 1996). Hence, it is not surprising that in 88% of colicinogenic strains we were able to detect plasmid DNA. A very similar percentage, 86%, of plasmid harbouring colicin producing strains was found by Riley and Gordon (1992) in the ECOR collection, a collection of strains representing clinical and non-clinical isolates from man, domestic and zoo animals. To determine whether the colicins are really encoded on the detected plasmids, the plasmids should be transferred to laboratory strains and subsequently tested for colicin production. Unfortunately, the capabilities of transferring the plasmids to laboratory strains are limited due to large plasmid size and lack of selection possibilities. However, several conjugative plasmids encoding antibiotic resistances were found in the DL collection and were successfully transferred. The overlay tests performed on these strains showed that all such plasmids (11) also encoded colicin production. The role of colicins in microbial communities is still not clear (RILEY & WERTZ 2002). Bacteriocins may serve as anti-competitors enabling the invasion of a strain into an established microbial community or they may play a defensive role and act to prohibit the invasion of other strains or species into an occupied niche or limit the advance of neighbouring cells (RILEY & WERTZ 2002). Further it was also suggested that colicins, at least ColV, more exactly the virulence factors also encoded by ColV pla- smids might play a role in pathogenesis (WATERS & CROSA 1991). In a study of UPEC strains ŠMARDA & OBDRŽALEK (2001) observed that the incidence of colicinogenicity signifi cantly differs between haemolytic and non-haemolytic strains (42% of non-haemolytic and only 22% of haemolytic strains were colicinogenic). A comparable difference was also observed in the presented study, as 44% of the non-haemolytic and 19% of the haemolytic strains were colicinogenic. Further, a similar even though less distinctive difference, was observed with the cnf encoding (28%) and cnf non encoding strains (41%). These results suggest, that the production of colicins is associated with some virulence factors, since otherwise the incidence of colicinogenic strains should be equal in both groups, as it is in the presented study for the ibeA encoding (40%) and non ibeA encoding strains (38%). Even though the obtained results could indicate that colicinogenic strains are less virulent, further studies are needed to establish the role colicins play in natural populations. Conclusions To summarise and conclude: 1. 110 uropathogenic Escherichia coli (UPEC) strains were screened for colicin production; 2. 38% of the tested UPEC strains have been found to be colicinogenic; 3. 67% of the colicinogenic UPEC strains were able to lyse all 20 used colicin producer strains, hence these UPEC strains encode and produce either more than one colicin, or a colicin not tested; 4. 19% of the colicinogenic strains produced only ColV, 7% only ColM and 7% only ColE1; 5. plasmids were found in 88% of the colicinogenic strains; 6. 26% of colicinogenic strains harboured conjugative plasmids encoding antibiotic resistance(s) and colicinogenic proprieties and 7. 19% of haemolytic UPEC strains and 44% of non-haemolytic strains were also colicinogenic, 28% of cnf encoding strains and 41% of strains not encoding cnf were colicinogenic, while 40% of ibeA encoding strains and 38% of strains not encoding ibeA were colicinogenic. M. Starčevič Erjavec & al.: Colicins of the Escherichia coli uropathogenic strain collection 20 Acta Biologica Slovenica, 49 (1), 2006 Acknowledgement The authors thank B. Bachmann for providing the strain AB1133, A. P. Pugsley for providing the colicinogenic strain collection, D. Ehrlich for strain HB101, R. Schmitt for strains RU4404 and RU4406 and T. J. Silhavy for strain TR51. Povzetek V predstavljeni raziskavi smo 110 uropatogenih sevov bakterije Escherichia coli (UPEC), ki so jih na Inštitutu za mikrobiologijo in imunologijo Medicinske fakultete v Ljubljani izolirali iz urina bolnikov z urinarno infekcijo, s pomočjo 20 kolicinogenih sevov iz Pugsleyeve zbirke testirali za produkcijo kolicinov. Testiranje je pokazalo, da je bilo 42 sevov UPEC (38%) kolicinogenih. Ker smo testiranje izvedli brez indukcije z mitomicinom C, za katero je znano, da sproži lizogene bakteriofage, lahko predpostavljamo, da dobljeni rezultati temeljijo samo na aktivnosti kolicinov in ne bakteriofagov. Dobljen odstotek kolicinogenih sevov (38%) se ujema tudi z rezultati objavljenimi v podobnih študijah. Vsi kolicinogeni sevi UPEC so povzročili propad producentskih sevov kolicinov ColA, ColB, ColD, ColE2, ColE3, ColE4, ColE5, ColE6, ColE7, ColIa, ColIb, ColK, ColN, MccB17, ColS4, MccC7 in ColE6-J iz Pugsleyeve zbirke. 93% kolicinogenih sevov UPEC je povzročilo propad ColM in ColE1 producentskega seva in 81% kolicinogenih sevov UPEC je povzročilo propad producentskega seva kolicina ColV. 67% kolicinogenih sevov UPEC je povzročilo propad vseh 20 kolicinogenih sevov Pugsleyeve zbirke in 33% kolicinogenih sevov UPEC je povzročilo propad 19 sevov iz Pugsleyeve zbirke. Večina kolicinogenih sevov (67%) sintetizira vsaj dva kolicina, ali pa kolicin, ki ni bil zastopan v Pugsleyevi zbirki kolicinogenih sevov. Kolicine sevov UPEC, ki sintetizirajo samo 1 kolicin, smo prepoznali; 8 sevov UPEC (19%) je sintetiziralo samo ColV, 3 sevi (7%) so sintetizirali samo ColM in 3 sevi (7%) so sintetizirali samo ColE1. Sicer je znano, da različne populacije E. coli sintetizirajo različne kombinacije kolicinov, a kombinacija ColV, ColM in ColE1, do sedaj še ni bila ugotovljena. Iz 88% kolicinogenih sevov smo uspeli izolirati plazmidno DNA po metodi alkalne lize. S poskusi konjugacije smo v 11 sevih DL našli konjugativne plazmide z zapisi za odpornost proti antibiotikom in za sintezo kolicinov. Vloga kolicinov v naravnih populacijah še ni popolnoma razjasnjena. Za plazmid ColV je znano, da ima poleg zapisa za kolicin V tudi zapise za virulenčne dejavnike. V naši raziskavi smo ugotovili, da je bilo 19% hemolitičnih sevov UPEC in 44% nehemolitičnih sevov UPEC tudi kolicinogenih; 28% sevov z zapisom cnf in 41% sevov brez zapisa cnf je bilo kolicinoge- nih, medtem ko je 40% z zapisom ibeA in 38% brez tega zapisa bilo kolicinogenih. Čeprav dobljeni rezultati nakazujejo, da so kolicinogeni sevi manj virulentni kot nekolicinogeni sevi, so za potrditev te domneve potrebne še nadaljnje raziskave. Literature BRADLEY, D. E. & S. P. HOWARD 1991. Colicinogeny of O157:H7 enterohemorrhagic Escherichia coli and the shielding of colicin and phage receptors by their O-antigenic side chains. Can. J. Microbiol. 37:97–104. DAW, M. A. & F. R. FALKINER 1996: Bacteriocins: nature, function and structure. Micron 27:467– 479. FELDGARDEN, M. & M. A. RILEY 1998. High levels of colicin resistance in Escherichia coli. Evolution 52:1270–1276. MCGEACHIE, J. 1965: Bacteriocin typing in urinary infection: Zentralbl. Bakt. I Orig. 196:377–384. O’BRIEN, G. J., S. T. CHAMBERS, B. PEDDIE & H. K. MAHANTY 1996: The association between colicino- genicity and pathogenesis among uropathogenic isolates of Escherichia coli. Microb. Pathogen. 20:185–190. 21 PUGSLEY, A. P. & B. OUDEGA 1987: Methods for studying colicins and their plasmids. In Hardy, K. G. Plasmids, a practical approach. IRL Press, Oxford, Washington DC. RILEY, M. A. & D. M. GORDON 1992: A survey of Col plasmids in natural isolates of the Esche- richia coli and an investigation into the stability of Col-plasmid lineages. J. Gen. Microbiol. 138:1345–1352. RILEY, M. A. & D. M. GORDON 1996: The ecology and evolution of bacteriocins. J. Industrial. Micro- biol. 17:151–158. RILEY, M. A. 1998: Molecular mechanisms of bacteriocin evolution. Annu. Rev. Genet. 32:255– 278. RILEY, M. A. & D. M. GORDON 1999: The ecological role of bacteriocins in the bacterial competition. Trends in Microbiology 7:129–133. RILEY, M. A. & J. E. WERTZ 2002: Bacteriocin diversity: ecological and evolutionary perspectives. Biochemie 84:357–364. SAMBROOK, J., E. F. FRITSCH & T. MANIATIS 1989. Molecular cloning: a laboratory manual, 2nd edition. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. ŠMARDA, J. & V. OBDRŽALEK 2001: Incidence of colicinogenic strains among human Escherichia coli. J. Basic Microbiol. 41:367–374. VAN DER WAL, F. J., J. LUIRINK & B. OUDEGA 1995: Bacteriocin release proteins: mode of action, struc- ture, and biotechnological application. FEMS Microbiol Rev 17:318–399. WATERS, V. L. & J. H. CROSA 1991: Colicin V virulence plasmids. Microbiol. Rev. 55:437–450. M. Starčevič Erjavec & al.: Colicins of the Escherichia coli uropathogenic strain collection