- Email: [email protected]
Virulence characteristics of Escherichia coli strains causing acute cystitis in young adults in Iran
Journal of Infection (2005) 50, 312–321
www.elsevierhealth.com/journals/jinf
Virulence characteristics of Escherichia coli strains causing acute cystitis in young adults in Iran M. Katoulia,*, A. Braunerb, L.K. Haghighic, B. Kaijserd, V. Muratova, ¨llbya R. Mo a
Microbiology and Tumorbiology Centre, Karolinska Institute, S-171 77 Stockholm, Sweden Department of Microbiology, Karolinska Hospital, S-171 76 Stockholm, Sweden c Department of Microbiology, School of Medicine, University of Shiraz, Shiraz, Iran d ¨teborg University, S-413 46 Go ¨teborg, Sweden Department of Clinical Bacteriology, Go b
Accepted 29 May 2004 Available online 21 July 2004
KEYWORDS Escherichia coli; Cystitis; Virulence properties; Antibiotic resistance
Summary Background. Escherichia coli strains that cause cystitis posses virulence properties that facilitate their colonisation and persistence in the bladder. In Iran, despite the high number of the urinary tract infections, very few studies has been done to determine the role of these virulence properties in the pathogenesis of E. coli cyctitis. Patients and methods. Eighty-seven strains of E. coli, isolated from young adults with cystitis in Shiraz, Iran, were examined for the expression of type 1 and P-fimbriae, mannose resistant haemagglutination, haemolysin production, aerobactin-mediated iron uptake, O:K serotypes, biochemical phenotypes (BPTs) and their antibiotic susceptibility patterns. Results. Seventy-six percent of the strains expressed multiple virulence properties. There was a significant correlation between the presence of aerobactin and the expression of type 1 fimbriae. All P-fimbriated strains produced aerobactin with 50% of them also coexpressing haemolysin. Of the 29 different O:K serotypes identified, 42% belonged to serotypes not commonly found among European serotypes associated with UTI. Strains of O groups 4 and 6 expressed more virulence factors than the others. A high resistance against ampicillin, trimethoprim and cotrimoxasol was observed among the isolates with 53% of the isolates showing multiresistance to these three antibiotics. Certain BPTs were also found among O:K serotypes with some containing strains of the same virulence profile. Conclusion. We conclude that certain colonal groups of E. coli are commonly associated with cystitis in young adults in Iran with strains possessing a combination of aerobactin and type 1 fimbriae being the dominant ones and belonging to serotypes not commonly found in Europe. We also conclude that the multiple antibiotic resistant E. coli strains causing cyctitis are highly prevalent in this part of the country. Q 2004 The British Infection Society. Published by Elsevier Ltd. All rights reserved.
*Corresponding author. Address: Faculty of Science, University of the Sunshine Coast, Maroochydore D.C., Qld. 4558, Australia. Tel.: þ 61-7-5430-2845; fax: þ 61-7-5430-2887. E-mail address: [email protected]
Introduction Urinary tract infections (UTI) are among the most
0163-4453/$30.00 Q 2004 The British Infection Society. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.jinf.2004.05.021
E. coli cystitis in Iran
common infections of humans. The majority of cases are caused by a limited number of bacterial species among which Escherichia coli accounts for more than three-quarters of all UTI.1 Infections are assumed to occur via the ascending of bacterial strains present in the host faeces,2 but it is now clear that such strains possess properties which enable them to adhere to urinary tract epithelial cells,3 – 5 a property which is not frequently observed among the normal faecal flora.6 The capacity of uropathogenic E. coli to adhere is mediated by a filamentous proteinaceous appendage on the surface of the bacteria capable of agglutinating human erythrocytes in the presence of D -mannose (i.e. P-fimbriae)3,7 or of mediating a mannose-sensitive haemagglutination of guinea-pig erythrocytes (type 1 fimbriae).4,8 Other potential virulence determinants have also been reported to be involved in the pathogenesis of E. coli causing UTI. These include certain outer membrane proteins,9 a complete (smooth) lipopolysaccharide layer,10 certain O serotypes,11 K112,13 and other K capsules,14,15 production of haemolysin,16,17 cytotoxic necrotising factor,18,19 certain colicins,20 cell surface hydrophobicity,21,22 and iron sequestration through the production of aerobactin.23,24 Some E. coli strains may cause asymptomatic bacteriuria which may reflect colonisation along the urinary tract, or symptomatic infection of either lower urinary tract and bladder (cystitis) or the upper urinary tract (pyelonephritis).25 It has been shown that strains isolated from each of these types of infections belong to certain O:K serotypes and possess certain virulence properties with varying prevalence.26 These pathogenic phenotypes have further been shown to belong to special clonal groups which have often been identified by their sero-biotypes.26 In this study, we investigated the role and importance of different virulence properties in E. coli strains causing acute cystitis in young adults in Iran. We also investigated their susceptibility to antibiotics commonly used to treat UTI and used a biochemical fingerprinting method, specifically developed for typing of E. coli strains,27 to investigate phenotypic variations among these strains.
Materials and methods Patients During October 1992 to March 1993, a total of 87 adult patients with acute cystitis caused by E. coli
313
were received in an out-patient clinic in Shiraz, Iran. These included 24 (28%) male (mean age 36 years, range 24 – 51) and 63 (72%) female (mean age 26 years, range 18 – 36). All patients had typical clinical signs and symptoms of UTI with urgency, dysuria and . 100,000 (67 patients), or . 10,000 (20 patients) E. coli/ml urine. None of the patients had a urine catheter.
Bacteriological investigation Numbers of bacteria in the urine were estimated from a clean-catch midstream by the standard loop technique and E. coli isolates were identified on the basis of their biochemical profiles by the methods of Edwards and Ewing28 after an initial isolation on MacConkey agar plates. Strains were then stored in deep agar at 4 8C and transferred to the Karolinska Institute, Stockholm, Sweden and tested for purity before storing at 2 80 8C.
Tests for virulence factors Adhesive properties. Bacteria were grown overnight at 37 8C on colonisation factor (CFA) agar and tested for P-fimbriae according to Svenson et al.29 using Gala1-4Galb coated latex beads (Orion Diagnostics, Finland). Presence of type 1 fimbriae was detected according to the method of Duguid et al.30 using erythrocytes from laboratory-bred guinea-pigs as described before.22 Using the same cultures of bacteria, strains were also tested for mannose-resistant haemagglutination (MRHA) of human erythrocytes of blood group A (Rh þ ) as described before.22 Haemolysin production. E. coli strains were tested for the production of haemolysin on blood agar plates. Defibrinated sheep blood was washed three times with 0.01 M tris chloride buffer (pH ¼ 7.5) containing 0.135 M NaCl and added to blood agar base no. 2 (Oxoid) at a final concentration, of 5%. Haemolysis was read after inoculation of bacteria and incubation at 37 8C overnight. Strains with at least 1 mm haemolysis zone around an isolated colony were regarded as haemolysin positive. Aerobactin production. Test strains were applied on agar plates covered with a suspension of the aerobactin requiring E. coli strain LG1522 as described previously.24 Determination of O antigens. E. coli strains were tested in microtiter plates for the presence of 68 somatic (O) antigens selected due their high prevalence among E. coli strains from UTI according to Lidin-Janson et al.31 Some strains (designated
314
non-typeable; ON) could not be typed with the antisera available. Determination of K antigens. Agar plates containing antisera were used according to the method of Kaijser.32 Twenty-three different antisera were used. Some strains (designated non-typeable; KN) could not be typed with the tested antisera. Antibiotic susceptibility testing. Strains were tested for their susceptibility to 10 commonly used antibiotics using the method of Ericsson and Sherris.32 The antibiotics included were, ampicilin, mecillinam, cefaroxim, trimetoprim, cotrimoxazol, nitrofurantoin, cefotaxime, nalidixic acid, imipenem and norfloxacin. Biochemical fingerprinting. E. coli strains were typed based on their biochemical phenotypes using the PhP-EC plates according to the methods described earlier (Katouli et al., 1991). Similarities between the strains were calculated as correlation coefficients as described before33 and clustered according to the unweighted pair group method with arithmetic averages (UPGMA)34 yielding a dendrogram. In the dendrogram, each strain is represented by a horizontal line. Different strains are connected with vertical lines at the similarity level they show to each other and thus the higher this line is, the more similar are the strains. An identity level of 0.975 was set based on reproducibility of the system as described before.33 Strains showing similarities to each other higher than this value were regarded as identical and assigned to the same biochemical phenotype (BPT). BPTs with more than one isolate were called common (C) and those with only one isolate were called single (S) BPTs. Statistical analysis. Fisher’s exact two-tailed test was used for statistical analysis.
Results Of 87 E. coli strains tested, 83 (95%) produced one or more of the virulence properties tested with aerobactin in 61 (70%) and type 1 fimbriae in 63 (72%) strains being the virulence factors most commonly expressed (Table 1). The prevalence of these virulence factors did not differ significantly between the male and female patients (Table 1). Altogether, 29 different O:K serotypes were identified among the strains. The O typeable strains were distributed among 12 serogroups with strains belonging to O groups 4, 6, 7 and 17 comprising 76% of the O typeable strains (Table 1). Biochemical fingerprinting of the strains showed the presence of different biochemical phenotypes
M. Katouli et al.
(BPTs) within each O group. All strains of O groups 4 and 6 but not 7 or 17, expressed multiple virulence factors (Figs. 1 and 2(A)). Production of P-fimbriae, haemolysin and aerobactin among strains of O groups 4 and 6 were significantly higher than among those of other prevalent serogroups (Figs. 1 and 2(A)). Generally, strains belonging to the same BPTs (i.e. C-BPTs) and having the same O:K serotypes, had similar virulence factors (Figs. 1 and 2(A)). Nineteen (22%) strains were O antigen nontypeable (ON) and belonged to diverse BPTs with 84% having K antigen non-typeable (KN) (Fig. 2(A)). Of eight strains in O group 17, six cross-reacted with the antisera against O 77 in our serotyping assay. Biochemical fingerprinting of these strains showed the presence of one common BPT (i.e. C1) containing both O17 and O77 strains all being K antigen nontypeable (Fig. 2(B)). Four strains were negative for any of the virulence markers tested and 17 strains produced only one virulence factor. In contrast, 66 (76%) strains expressed two or more virulence factors ðP , 0:0001Þ (Table 2). Coexpression of these virulence factors among the strains and their distribution among O:K serotypes did not significantly differ from each other (Table 2). Production of P-fimbriae was more pronounced among the strains coexpressing 4 or more virulence markers (18/21 strains) than those producing 3 or less (13/44 strains) ðP ¼ 0:0255Þ (Table 2). There was no difference between the number of strains producing aerobactin with or without the expression of Pfimbriae (32 vs. 29 strains). Contrary to that, all P-fimbriae producing strains ðn ¼ 32Þ expressed aerobactin (Table 3). Of 61 aerobactin producing strains, 45 (74%) strains also carried type 1 fimbriae (Table 3). There was a negative correlation between the presence of aerobactin and the expression of haemolysin or MRHA. Similarly, negative correlations were found between the presence of type 1 fimbriae and expression of MRHA or P-fimbriae (Table 3). Strains having K1, K5 and K12 comprised 63% of the strains with known capsular types (Table 4). The prevalence of virulence factors among these strains did not differ significantly. However, hemolysin and MRHA tend to be slightly higher, although significancy not different, among the K12 strains. Likewise, production of aerobactin and type 1 fimbriae was also more pronounced among these strains than those having other K antigens (Table 4). Eighty-five out of the 87 strains were tested for their patterns of susceptibility to 10 antibiotics commonly used for treatment of UTI. Resistance were found against ampicillin in 65 strains (76%), trimethoprim in 53 strains (62%), cotrimoxazol in 52
E. coli cystitis in Iran
315
Table 1 Prevalence of different virulence factors and serotypes among the Escherichia coli strains isolated from 87 adult patients with acute cystitis in Iran Virulence factor
None Haemolysin Aerobactin P-fimbriae MRHA Type 1 fimbriae MSHA (human A)c
a b c
Number (%) isolates M ðn ¼ 24Þ
F ðn ¼ 63Þ
Total
1 (4) 8 (33) 15 (63) 5 (21) 6 (25) 18 (75) 6 (25)
3 (5) 28 (44) 46 (73) 27 (43) 21 (33) 45 (71) 21 (33)
4 (5) 36 (41) 61 (70)a 32 (37) 27 (31) 63 (72)b 25 (29)
Serotypes (number)
Number (%) of isolates
O6:K5 (7), K13 (5), K2 (1), K14 (1), KN (7) O4:K12 (7), K3 (2), K1 (1), KN (1) O7:K1 (5), KN (2) O17 (77):K1 (1), K10 (1), KN (6) O21:K11 (1), KN (3) O8:KN O15:K2 O2:K1 O18:KN O19:KN O1:KN O75:KN OR:K12 (1), K1 (1), KN (4) ON:KN (16), K2 (2), K12 (1)
21 (24.0) 11 (12.6) 7 (8.0) 8 (9.2) 4 (4.6) 3 (3.4) 2 (2.3) 2 (2.3) 1 (1.1) 1 (1.1) 1 (1.1) 1 (1.1) 6 (6.9) 19 (21.8)
Aerobactin vs. haemolysin ðP ¼ 0:04Þ; P-fimbriae ðP ¼ 0:02Þ and MRHA ðP ¼ 0:004Þ: Type 1 fimbriae vs. haemolysin ðP ¼ 0:03Þ; P-fimbriae ðP ¼ 0:01Þ and MRHA ðP ¼ 0:002Þ: Strains showing only mannose-sensitive haemagglutination (MSHA) with erythrocytes from human blood group A. MRHA, mannose resistant haemagglutination; M, male; F, female; ON, O antigen non-typeable; OR, rough strain; KN, K antigen non-typeable.
Table 2 Coexpression of different virulence factors and their distribution among O:K serotypes of Escherichia coli strains isolated from adult patients with cystitis in Iran Coexpression of virulence factors
Number of isolates
Serotype (number of isolates)
None
4
ON:KN (2), OR:KN (1), O21:KN (1) a
Only 1 Hly MRHA Abn Type 1
17 (6) 3 1 7 6
2 Hly/Abn Hly/Type 1 P-Fim/Abn MRHA/Type 1 Abn/Type 1
24 (13)a 1 8 2 1 12
3 Hly/MRHA/Type 1 Hly/Abn/Type 1 P-Fim/Abn/Type 1
20 (5)a 3 1 11
MRHA/Abn/Type 1
5
4 Hly/P-Fim/MRHA/Abn Hly/P-Fim/Abn/Type 1 P-Fim/MRHA/Abn/Type 1 Hly/MRHA/Abn/Type 1
16 (1)a 6 5 2 3
O15:K2 (2), O6:K5 (1), O4:K3 (2), O17 (77): K1 (1) ON:KN (1), O21:KN (1), O6:K5 (1), O6:KN (1), O4:KN (1) O4:K1 (1), ON:KN (1) O6:KN (1), O4:K12 (1), O18:KN (1)
All 5
6
O4:K12 (3), O6:K12 (1), ON:KN (1), OR:K12 (1)
O6:K13 (1), O6:KN (1), O2:K1 (1) O8:KN (1) O7:K1 (3), O4:K12 (1), ON:KN (1), OR:KN (1), O17 (77):KN (1) O2:K1 (1), O17:KN (1), OR:KN (2), ON:K2 (1), O17 (77):KN (1) O7:KN (1) O6:KN (2), O6:K13 (4), O6:K2 (1), ON:K2 (1) ON:KN (1), O7:KN (1) O17:K10 (1) O6:K14 (1), O6:K5 (1), ON:KN (5), OR:KN (1), O8:KN (2), O7:K1 (1), OR:K1 (1) O4:K12 (2), ON:K12 (1) ON:KN (1) O75:KN (1), ON:KN (2), O6:K5 (4), O6:KN (1), O7:K1 (1), O17 (77):KN (2) O21:K11 (1), O21:KN (1), O1:KN (1), O19:KN (1), O17 (77): KN (1)
P-Fim, P-fimbriae; Hly, haemolysin; Abn, aerobactin; MRHA, mannose-resistant haemagglutination; Type 1, Type 1 fimbriae; ON, O antigen non-typeable; OR, rough strain; KN, K antigen non-typeable. a Numbers in bracket represent strains that showed only mannose-sensitive haemagglutination of human red blood cells.
316
M. Katouli et al.
Figure 1 Biochemical phenotypes (BPTs), K antigens and virulence properties of Escherichia coli strains of serogroup O6 (21 strains) (dendrogram A), serogroup O4 (11 strains) (dendrogram B) and serogroup O7 (7 strains) (dendrogram C) isolated from adult patients with cystitis in Iran. ID-level: Identity level above which strains are regarded as identical. Hly, haemolysin; MRHA, mannose-resistant haemagglutination; P-Fim, P-fimbriae; Type 1, type 1 fimbriae; KN, K antigen non-typeable; NT, not tested; S1-7, single BPTs 1 to 7; C1-3, common BPTs 1 to 3. Numbers of strains expressing different virulence factors have been shown for each serogroup. Significance of differences between O groups has been calculated only for the expression of P-Fim, Hly and Abn. O6 vs. O7 ðP ¼ 0:0113Þ; O4 vs. O7 ðP ¼ 0:0031Þ:
strains (61%), mecillinam in three strains (3%), cefaroxim in two strains (2%), and nitrofurantoin and nalidixic acid (in one strain each). 55 strains (65%) were resistant to two or more antibiotics with 45 strains (53%) resistant to trimethoprim, cotrimoxasol and ampicillin (Table 5). The latter three antibiotics constitute the highest rate of resistant among the isolates. Eleven strains were sensitive to all antibiotics tested (Table 5).
Discussion The success or failure of a bacterial invasion is determined by the contest between the host defence mechanisms and the virulence properties of the invading bacteria. Iron starvation is one of the major barriers that virulent bacteria must overcome in order to proliferate in the host.35 One of the most efficient systems for obtaining iron
E. coli cystitis in Iran
317
Figure 2 Biochemical phenotypes (BPTs), K antigens and virulence properties of Escherichia coli strains having O antigen non-typeable (ON; 19 strains) (dendrogram A) and of serogroup O17 (or O77) (8 strains) (dendrogram B). ID-level: identity level above which strains are regarded as identical. Hly, haemolysin; MRHA, mannose-resistant haemagglutination; P-Fim, P-fimbriae; Type 1, type 1 fimbriae; KN, K antigen non-typeable; S1-17, single BPTs 1 to 17; C1, common BPT 1. Numbers of strains expressing different virulence factors have been shown for each serogroup. Significance of differences between O groups has been calculated only for the expression of P-Fim, Hly and Abn. O6 (see Fig. 1) vs. ON ðP ¼ 0:0204Þ; and O17 (or O77) ðP ¼ 0:0424Þ: O4 (see Fig. 1) vs. ON ðP ¼ 0:0049Þ; and O17 (or O77) ðP ¼ 0:0083Þ:
by E. coli strains causing septicaemia or UTI is the production of siderophore aerobactin.36 In the present study, we found a higher incidence of aerobactin producing strains among our cystitis patients than those reported elsewhere.23,24,37
Aerobactin has often been reported to be associated with P-fimbriae in strains causing cystitis,37,38 however, we found that only half of the aerobactin producing strains in this study expressed P-fimbriae. Contrary to that, all P-fimbriated strains
Table 3 Correlation between the expression of different virulence factors among Escherichia coli strains isolated from 87 adult patients with cystitis in Iran Virulence factor
Aerobactin Haemolysin P-fimbriae MRHA Type 1 fimbriae
Number of strains
61 36 32 27 63
Number (%) of strains coexpressed Aerobactin
Haemolysin
P-fimbriae
MMA Type 1
Fimbriae
– 22 32 22 45
22 (36.1)a – 17 (53.1) 18 (66.7) 26 (41.3)
32 17 – 14 24
22 (36.1)a 18 (50.0) 14 (43.7) – 20 (31.7)e
45 (73.8)a 26 (72.2)b 24 (75.0)c 20 (74.1)d –
(61.1) (100)c (81.5)d (71.4)e
(52.4) (47.0) (51.8) (38.1)e
MRHA, mannose-resistant haemagglutination; Hly, haemolysin; Abn, aerobactin; Type 1, Type 1 fimbriae. a P ¼ 0:08 for Hly2 vs. Hlyþ strains and MRHA2 vs. MRHAþ strains; P ¼ 0:002 for Type 1þ vs. Type 12 strains. b P ¼ 0:04 for Type 1þ vs. type 12 strains. c P ¼ 0:00001 for Abnþ vs. Abn2 strains; P ¼ 0:03 for Type 1þ vs. Type 12 strains. d P ¼ 0:008 for Abnþ vs. Abn2 strains; P ¼ 0:05 for Type 1þ vs. Ttype 12 strains. e P ¼ 0:008 for Abnþ vs. Abn2 strains; P ¼ 0:02 for MRHAþ vs. MRHA2 strains and P ¼ 0:16 for P-Fim2 vs. P-fimþ strains.
318
M. Katouli et al.
Table 4 Distribution of different virulence properties among K-antigens found in Escherichia coli strains isolated from 87 adult patients with cystitis in Iran K antigen
K1 K2 K3 K5 K10 K11 K12 K13 K14 KN Total no.
Number of strains (%)
10 (11.5) 5 (5.7) 2 (2.3) 7 (8.0) 1 (1.1) 1 (1.1) 9 (10.3) 5 (5.7) 1 (1.1) 46 (52.9) 87 (100)
Number of strains producing Hemolysin
P-fimbriae
MRHA
Aerobactin
Type 1 fimbriae
2a 4 2 2a – – 8a 5 – 13
3 2 2 6 – – 4 – – 15
2a 2 2 1a 1 1 8a – – 10
8b 2 2 7b – 1 6b – 1 34
5 3 – 6b 1 1 8b 4 1 34
36
32
27
61
63
MRHA, mannose resistant haemagglutination; KN, K antigen non-typeable. a (P ¼ 0:1260; K12 vs. K1 and K5) for both virulence factors. b Not significant.
coexpressed aerobactin with 50% also producing haemolysin. This finding indicates that the iron sequestering system may be more important than Pfimbriation for pathogenicity of our strains and that P-fimbriated strains may carry more than one iron acquiring mechanisms. Association of aerobactin with a particular serotype has also been described for clonal groups of human invasive strains of E. coli K1.39 In the present study, aerobactin production was more pronounced among the strains of K antigens 1, 5 and 12 compared to others. It has been shown that certain O serotypes such as O1, O2, O4, O6, O7, O8, O16, O18, O25 and O75
are strongly associated with UTI among European studies.40,41 Fifty-four percent of our isolates belonged to these so called European Uropathogenic clones. In contrast, 42% of the strains (including non-typable strains) belonged either to serotypes different from those found in Europe or antisera against them were not available. A surprisingly high incidence of males were observed in the present study. The possibility that some of our patients had prostatitis was not appealing to us as none of the patients had classical symptoms associated with prostatitis. Besides obtaining samples from prostatitis patients requires
Table 5 Pattern of antibiotic resistance among 85 out of 87 Escherichia coli strains isolated from adult patients with cystitis in Iran Resistant to antibiotics
Pattern of resistance
None
Number of strains 11 (14%)
Only 1 Amp Trim
18 (21%) 1 (1%)
Amp/Trim Trim/Ctx Cfx/Mcl
2 (2%) 6 (7%) 1 (1%)
Amp/Trim/Ctx Trim/Ctx/Nfn
42 (49%) 1 (1%)
Amp/Trim/Ctx/Mcl Amp/Trim/Ctx/Nx
1 (1%) 1 (1%)
Amp/Trim/Ctx/Mcl/Cfx
1 (1%)
2
3
4
5 Amp, ampicillin; Trim, trimetoprim; Ctx, cotrimoxazol; Mcl, mecillinam; Nfn, nitrofurantoin; Cfk, cefaroxim; Nx, nalidixic acid.
E. coli cystitis in Iran
a vigorous procedure, which was not done in this study. One possible explanation, which may partly explain the high ratio of male over female with cystitis in this study is that young females experiencing UTI may not, due to cultural differences, seek clinical help as often as males do. It is well known that many women first try house-hold remedies which are common among traditional families. During the last few years, considerable progresses have been made to identify factors specifically or non-specifically involved in the pathogenesis of UTI. It is now clear that pathogenic bacteria attach and then enter the host tissue, proliferating and eventually cause a definite damage. The role of type 1 fimbriae as a potential adhesive factor in the pathogenesis of UTI has been clearly demonstrated in experimental animals.4,5 In our study, type 1 fimbriae was produced by majority of the strains and most of them coexpressed aerobactin, suggesting that these two virulence factors play an important role in pathogenesis of E. coli strains causing cystitis in Iran. A key feature of uropathognesis is the ability to adhere to bladder mucosa. P-fimbriae is a major type of adhesive fimbriae in strains causing pyelonephritis.3,42 The number of P-fimbriated strains in our study was much higher than those reported for cystitis patients by others.24,42,43 Its prevalence was more pronounced among strains expressing type 1 fimbriae, indicating that the fimbriation status of E. coli strains causing cystitis in our study is varying from predominantly type 1 fimbriae to a mixture of type 1 and P fimbriated cells, a result which has also been found by others.5 Production of haemolysin by E. coli strains is a mechanism of scavenging iron under iron-limited conditions. However, it was interesting to note that more than 61% of haemolysin-producing strains also produced aerobactin. It has been postulated that bacteria containing haemolysin have no need to produce aerobactin since they can obtain the required iron from lysis of red blood cells.38 The high association of haemolysin and aerobactin among our strains, therefore, indicates that these strains have a greater iron-sequestering ability than those producing either of these two factors alone. Furthermore, our finding that almost 50% of Pfimbriated strains also produced haemolysin suggests a potential role for combination of these three virulence factors in pathogenesis of E. coli cystitis in this part of the country. Metabolic fingerprinting of the strains indicated the presence of different biochemical phenotypes (BPTs) among different O serogroups with one BPT being dominant in each group. Using a combination
319
of biochemical fingerprinting, O serogroup and virulence factors, we previously characterised enteropathogenic E. coli strains isolated from Iran and found that strains carrying similar virulence properties had similar BPTs and belonged to the same O serogroup although, different BPTs were also observed within each group.44 In the present study, common BPTs of serogroups O6, O4 and O7 generally carried strains with either identical Ktype or non-typeable K antigen (see Fig. 1). Often there was variation among the type of virulence factors expressed by some strains within each O:K serotype. Bacterial clones which exist within a population for a long time may undergo changes in phenotypic characters and virulence properties.45 Therefore, it is possible that some of our strains had lost or gained one or more of their virulence factors or had gone through changes in their O or K antigens, a phenomenon which can be induced by factors such as antibodies of the host.24 We also found that strains of the same O:K serotype (e.g. O6:K13 in Fig. 1(a) or O4:K12 in Fig. 1(b)) belonged to different BPTs which is not surprising since biochemical reactions among E. coli strains vary widely. Nonetheless, the fact that most strains belonging to a common BPT and having the same O:K serotype, carried the same virulence factors, is in agreement with the clonal concept of pathogenic bacteria.45 The presence of a high percentage of both O and K non-typeable strains in our study is rather surprising. Both biochemical fingerprinting and virulence profile of these isolates showed that they belonged to diverse groups of E. coli strains ruling out the possibility that they belonged to a common and/or newly emerged clonal group causing cystitis among these patients. We postulate that these strains had been members of the normal flora that have picked up the ability to colonise urinary tract and cause infection, since exchange of genetic materials among bacteria of the gut is common. To our knowledge, this is the first report describing the prevalence of different virulence properties associated with E. coli cystitis in Iran. Several other important virulence properties such as cytotoxic necrotising factor,19 and different classes of adhesins such as Dr family of adhesins have been shown to play an important role in pathogenesis of UTI.46 In this study, we investigate the presence of some virulence attributes, and we here show that a high portion of E. coli strain causing cystitis belong to non-typeable serotypes (both O and K types) which are not common to those found in Europe and that majority of them contain a combination of different virulence properties and are resistant to three commonly used antibiotics.
320
References 1. Roberts AP, Phillips R. Bacteria causing symptomatic urinary tract infection or bacteriurea. J Clin Pathol 1979;32:492—6. ˚, Kaijser B, et al. Comparison of 2. Lidin-Janson G, Hanson AA Escherichia coli from bacteriuric patients with those from feces of healthy school children. J Infect Dis 1977;136: 346—53. 3. Hagberg L, Jodal LA, Korhonen T, Lidin-Janson G, Lindberg U, Svanborg-Eden C. Adhesion, hemagglutination and virulence of Escherichia coli causing urinary tract infections. Infect Immun 1981;31:564—70. 4. Iwahi T, Abe Y, Nakao M, Imada A, Tsuchiya K. Role of type 1 fimbriae in pathogenesis of ascending urinary tract infection induced by Escherichia coli in mice. Infect Immun 1983;39: 1307—15. 5. Kisielius PV, Schwan WR, Amundsen SK, Duncan JL, Schaeffer AJ. In vivo expression and variation of Escherichia coli type 1 and P pili in the urine of adults with acute urinary tract infections. Infect Immun 1989;57:1656—62. 6. Svanborg-Eden C, Lidin-Janson G, Lindberg U. Adhesiveness to urinary tract epithelial cells of fecal and urinary Escherichia coli isolates from patients with symptomatic urinary tract infections or asymptomatic bateriuria of varying duration. J Urol 1979;122:185—8. 7. Ka ¨llenius G, Mo ¨llby R, Winberg J. In vitro adhesion of uropathogenic Escherichia coli to human periuretheral cells. Infect Immun 1980;28:972—80. 8. Schaeffer A, Schwan WR, Hultgren SJ, Duncan JL. Relationship of type 1 pilus expression in Escherichia coli to ascending urinary tract infections in mice. Infect Immun 1987;373—80. ¨ quero ME, Aron L, DeLuca AG, Timmis KM, Cabello FC. A 9. A plasmid-encoded outer membrane protein, tra T, enhances resistance of Escherichia coli to phagocytosis. Infect Immun 1984;46:740—6. 10. Taylor P. Bacteriocidal and bacteriolytic activity of scrum against gram-negative bacteria. Mirobiol Rev 1983;47: 46—83. 11. Hughes C, Phillips R, Roberts AP. Serum resistance among Escherichia coli strains causing urinary tract infection in relation to O type and carriage of hemolysin, colin and antibiotic resistance determinants. Infect Immun 1982;35: 270—5. 12. Cross AS, Gemski P, Sadoff JC, Ørskov F, Ørskov I. The importance of the K1 capsul in invasive infections of Escherichia coli. J Infect Dis 1984;149:184—93. 13. Kaijser B. Immunology of Escherichia coli: K antigen and its relation to urinary tract infection. J Infect Dis 1973;127: 670—7. ˚, Jodal U, Lidin-Janson G, Robbins JB. 14. Kaijser B, Hanson LA Frequency of Escherichia coli K antigens in various forms of urinary tract infection in children. Lancet 1977;I:663—4. 15. Opal S, Cross A, Gemski P. K antigen and serum sensitivity of rough Escherichia coli. Infect Immun 1982;37:956—60. 16. Hughes C, Hacker J, Roberts A, Goebel W. Hemolysin production as virulence marker in symptomatic and asymptomatic urinary tract infections caused by Escherichia coli. Infect Immun 1983;39:546—51. 17. Smith HW, Huggins MB. The toxic role of alpha hemolysin in pathogenesis of experimental Escherichia coli infection in mice. J Gen Microbiol 1985;131:395—403. 18. Alonso P, Blanco J, Blanco M, Gonzales EA. Frequent production of toxins by Escherichia coli strains from urinary tract infections: relation with haemagglutination. FEMS Microbiol Lett 1987;48:391—6.
M. Katouli et al.
19. Caprioli A, Falbo V, Ruggeri FM, Baldassarri L, Bisicchi R, Ipollito G, Romoli E, Donelli G. Cytotoxic nectrtizing factor production by hemolytic strains of Escherichia coli causing extraintestinal infections. J Clin Microbiol 1987;25:146—9. 20. Williams PH, Warner PJ. Col V plasmid-mediated, colicin Vindependent iron uptake system of invasive strains of Escherichia coli. Infect Immun 1980;29:411—6. 21. Jacobson SH, Tullus K, Baruner A. Hydrophobic properties of Escherichia coli causing acute pyelonephritis. J Infect 1989; 19:17—23. 22. Brauner A, Katouli M, Tullus K, Jacobsson SH. Cell surface hydrophobicity, adherence to Hela cell cultures and haemagglutination pattern of pyelonephritogenic Escherichia coli strains. Epidemiol Infect 1990;105:255—63. 23. Carbonetti NH, Boonachi S, Parry SH, Va ¨isa ¨nen-Rhen V, Korhonen TK, Williams PH. Incidence of aerobactin iron uptake by Escherichia coli isolates from human extraintestinal infections. Infect Immun 1986;51:966—8. ¨ sterberg E, 24. Jacobson SH, Hammarlind M, Lidefeldt KJ, O Tullus K, Brauner A. Incidence of aerobactin-positive Escherichia coli strains in patients with symptomatic urinary tract infection. Eur J Clin Microbiol Infect Dis 1988;7:630—4. 25. Clarridge JE, Pezzlo MT, Vosti KL, Weissfeld AS. Laboratory diagnosis of urinary tract infections. Cumitech, 2A. Washington, DC: American Society for Microbiology; 1987. 26. Ørskov I, Ørskov F. Escherichia coli in extra-intestinal infections. J Hyg Camb 1985;95:551—75. 27. Ku ¨hn I. Biochemical fingerprinting of Escherichia coli: a simple method for epidemiological investigations. J Microbiol Methods 1985;3:159—70. 28. Edwards PR, Ewing WH. Identification of Enterobacteriacaea. 3rd ed. Minneapolis: Burgrass Publishing Co; 1971. 29. Svenson SB, Ka ¨llenitis G, Mo ¨llby R, Hultberg H, Winberg J. Rapid identification of P-fimbriated Escherichia coli by a receptor-specific particle agglutination test. Infection 1982; 10:209—14. 30. Duguid JP, Clegg S, Wilson MI. The fimbrial and non-fimbrial haemagglutinations of Escherichia coli. J Med Microbiol 1979;12:213—27. 31. Lidin-Janson G, Falsen E, Jodal U, Kaijser B, Lincoln K. Characteristics of antibiotic resistant Escherichia coli in the rectum of healthy school-children. J Med Microbiol 1977;10: 299—308. 32. Kaijser B. A simple method for typing of acidic polysaccharide K antigens of Escherichia coli. FEMS Microbiol Lett 1977; 1:285—8. 33. Mo ¨llby R, Ku ¨hn I, Katouli M. Computerized biochemical fingerprinting—a new tool for typing of bacteria. Rev Med Microbiol 1993;4:231—41. 34. Sneath PHA, Sokal RR. Numerical taxonomy. San Francisco: W.H. Freeman; 1973. 35. de Lorenzo V, Martinez JL. Acrobactin production as a virulence factor: a reevaluation. Eur J Clin Microbiol Infect Dis 1988;7:621—9. 36. Martinez LJ, Delgado-Iribarren A, Baquero F. Mechanisms of iron acquisition and bacterial virulence. FEMS Microbiol Rev 1990;75:45—56. 37. Ørskov I, Svanborg-Eden C, Ørskov F. Aerobatin production of serotyped Escherichia coli from urinary tract infection. Med Microbiol Immunol 1988;177:9—14. 38. Johnson JR, Mosley SL, Roberts PL, Stamm WE. Aertobactin and other virulence factor genes among strains of Escherichia coli causing urosepsis: association with patients characteristics. Infect Immun 1988;56:405—12. 39. Valvano MA, Silver RP, Crosa JH. Occurence of chromosomeor plasmid-mediated aerobactin iron transport system and hemolysin production among clonal groups of human
E. coli cystitis in Iran
40.
41.
42.
43.
invasive strains of Escherichia coli K1. Infect Immun 1986; 52:192—9. Vaisanen-Rhen V, Elo J, Vaisanen E, et al. P-fimbriated clones among uropathogenic Escherichia coli strains. Infect Immun 1984;43:149—55. Truck M, Petersdorf RG, Fournier MR. The epidemiology of non-enteric Escherichia coli infections: prevalence of serological groups. J Clin Invest 1962;41:1760—5. Ka ¨llenius G, Mo ¨llby R, Svenson SB, et al. Occurence of Pfimbriated Escherichia coli in urinary tract infections. Lancet 1981;2:1369—72. Winberg J, Bollgren I, Ka ¨llenius G, Mo ¨llby R, Svenson S. Bacterial virulence and host defence in pathogenesis of renal infection and renal scaring. In: Francois B, Perrin B, editors.
321
Urinary tract infection insights and prospects. London: Butterworths Publications; 1983. p. 19—33. 44. Katouli M, Ku ¨llby R. Biochemical phenotypes of ¨hn I, Mo enteropathogenic Escherichia coli common to Iran and Sweden. J Med Microbiol 1991;35:270—7. 45. Ørskov F, Ørskov I. Summary of workshop on the clone concept in epidemiology, taxanomy and evolution of Entrobacteriacae and other bacteria. J Infect Dis 1983; 148:346—57. 46. Nowicki B, Labigne A, Moseley S, Hull R, Hull S, Moulds J. The Dr hemagglutinin, afimbrial adhesins AFA-I and AFA-III, and F1845 fimbriae of uropathogenic and diarrhoea-associated Escherichia coli belong to family of hemagglutinins with Dr receptor recognition. Infect Immun 1990;58:279—81.
www.elsevierhealth.com/journals/jinf
Virulence characteristics of Escherichia coli strains causing acute cystitis in young adults in Iran M. Katoulia,*, A. Braunerb, L.K. Haghighic, B. Kaijserd, V. Muratova, ¨llbya R. Mo a
Microbiology and Tumorbiology Centre, Karolinska Institute, S-171 77 Stockholm, Sweden Department of Microbiology, Karolinska Hospital, S-171 76 Stockholm, Sweden c Department of Microbiology, School of Medicine, University of Shiraz, Shiraz, Iran d ¨teborg University, S-413 46 Go ¨teborg, Sweden Department of Clinical Bacteriology, Go b
Accepted 29 May 2004 Available online 21 July 2004
KEYWORDS Escherichia coli; Cystitis; Virulence properties; Antibiotic resistance
Summary Background. Escherichia coli strains that cause cystitis posses virulence properties that facilitate their colonisation and persistence in the bladder. In Iran, despite the high number of the urinary tract infections, very few studies has been done to determine the role of these virulence properties in the pathogenesis of E. coli cyctitis. Patients and methods. Eighty-seven strains of E. coli, isolated from young adults with cystitis in Shiraz, Iran, were examined for the expression of type 1 and P-fimbriae, mannose resistant haemagglutination, haemolysin production, aerobactin-mediated iron uptake, O:K serotypes, biochemical phenotypes (BPTs) and their antibiotic susceptibility patterns. Results. Seventy-six percent of the strains expressed multiple virulence properties. There was a significant correlation between the presence of aerobactin and the expression of type 1 fimbriae. All P-fimbriated strains produced aerobactin with 50% of them also coexpressing haemolysin. Of the 29 different O:K serotypes identified, 42% belonged to serotypes not commonly found among European serotypes associated with UTI. Strains of O groups 4 and 6 expressed more virulence factors than the others. A high resistance against ampicillin, trimethoprim and cotrimoxasol was observed among the isolates with 53% of the isolates showing multiresistance to these three antibiotics. Certain BPTs were also found among O:K serotypes with some containing strains of the same virulence profile. Conclusion. We conclude that certain colonal groups of E. coli are commonly associated with cystitis in young adults in Iran with strains possessing a combination of aerobactin and type 1 fimbriae being the dominant ones and belonging to serotypes not commonly found in Europe. We also conclude that the multiple antibiotic resistant E. coli strains causing cyctitis are highly prevalent in this part of the country. Q 2004 The British Infection Society. Published by Elsevier Ltd. All rights reserved.
*Corresponding author. Address: Faculty of Science, University of the Sunshine Coast, Maroochydore D.C., Qld. 4558, Australia. Tel.: þ 61-7-5430-2845; fax: þ 61-7-5430-2887. E-mail address: [email protected]
Introduction Urinary tract infections (UTI) are among the most
0163-4453/$30.00 Q 2004 The British Infection Society. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.jinf.2004.05.021
E. coli cystitis in Iran
common infections of humans. The majority of cases are caused by a limited number of bacterial species among which Escherichia coli accounts for more than three-quarters of all UTI.1 Infections are assumed to occur via the ascending of bacterial strains present in the host faeces,2 but it is now clear that such strains possess properties which enable them to adhere to urinary tract epithelial cells,3 – 5 a property which is not frequently observed among the normal faecal flora.6 The capacity of uropathogenic E. coli to adhere is mediated by a filamentous proteinaceous appendage on the surface of the bacteria capable of agglutinating human erythrocytes in the presence of D -mannose (i.e. P-fimbriae)3,7 or of mediating a mannose-sensitive haemagglutination of guinea-pig erythrocytes (type 1 fimbriae).4,8 Other potential virulence determinants have also been reported to be involved in the pathogenesis of E. coli causing UTI. These include certain outer membrane proteins,9 a complete (smooth) lipopolysaccharide layer,10 certain O serotypes,11 K112,13 and other K capsules,14,15 production of haemolysin,16,17 cytotoxic necrotising factor,18,19 certain colicins,20 cell surface hydrophobicity,21,22 and iron sequestration through the production of aerobactin.23,24 Some E. coli strains may cause asymptomatic bacteriuria which may reflect colonisation along the urinary tract, or symptomatic infection of either lower urinary tract and bladder (cystitis) or the upper urinary tract (pyelonephritis).25 It has been shown that strains isolated from each of these types of infections belong to certain O:K serotypes and possess certain virulence properties with varying prevalence.26 These pathogenic phenotypes have further been shown to belong to special clonal groups which have often been identified by their sero-biotypes.26 In this study, we investigated the role and importance of different virulence properties in E. coli strains causing acute cystitis in young adults in Iran. We also investigated their susceptibility to antibiotics commonly used to treat UTI and used a biochemical fingerprinting method, specifically developed for typing of E. coli strains,27 to investigate phenotypic variations among these strains.
Materials and methods Patients During October 1992 to March 1993, a total of 87 adult patients with acute cystitis caused by E. coli
313
were received in an out-patient clinic in Shiraz, Iran. These included 24 (28%) male (mean age 36 years, range 24 – 51) and 63 (72%) female (mean age 26 years, range 18 – 36). All patients had typical clinical signs and symptoms of UTI with urgency, dysuria and . 100,000 (67 patients), or . 10,000 (20 patients) E. coli/ml urine. None of the patients had a urine catheter.
Bacteriological investigation Numbers of bacteria in the urine were estimated from a clean-catch midstream by the standard loop technique and E. coli isolates were identified on the basis of their biochemical profiles by the methods of Edwards and Ewing28 after an initial isolation on MacConkey agar plates. Strains were then stored in deep agar at 4 8C and transferred to the Karolinska Institute, Stockholm, Sweden and tested for purity before storing at 2 80 8C.
Tests for virulence factors Adhesive properties. Bacteria were grown overnight at 37 8C on colonisation factor (CFA) agar and tested for P-fimbriae according to Svenson et al.29 using Gala1-4Galb coated latex beads (Orion Diagnostics, Finland). Presence of type 1 fimbriae was detected according to the method of Duguid et al.30 using erythrocytes from laboratory-bred guinea-pigs as described before.22 Using the same cultures of bacteria, strains were also tested for mannose-resistant haemagglutination (MRHA) of human erythrocytes of blood group A (Rh þ ) as described before.22 Haemolysin production. E. coli strains were tested for the production of haemolysin on blood agar plates. Defibrinated sheep blood was washed three times with 0.01 M tris chloride buffer (pH ¼ 7.5) containing 0.135 M NaCl and added to blood agar base no. 2 (Oxoid) at a final concentration, of 5%. Haemolysis was read after inoculation of bacteria and incubation at 37 8C overnight. Strains with at least 1 mm haemolysis zone around an isolated colony were regarded as haemolysin positive. Aerobactin production. Test strains were applied on agar plates covered with a suspension of the aerobactin requiring E. coli strain LG1522 as described previously.24 Determination of O antigens. E. coli strains were tested in microtiter plates for the presence of 68 somatic (O) antigens selected due their high prevalence among E. coli strains from UTI according to Lidin-Janson et al.31 Some strains (designated
314
non-typeable; ON) could not be typed with the antisera available. Determination of K antigens. Agar plates containing antisera were used according to the method of Kaijser.32 Twenty-three different antisera were used. Some strains (designated non-typeable; KN) could not be typed with the tested antisera. Antibiotic susceptibility testing. Strains were tested for their susceptibility to 10 commonly used antibiotics using the method of Ericsson and Sherris.32 The antibiotics included were, ampicilin, mecillinam, cefaroxim, trimetoprim, cotrimoxazol, nitrofurantoin, cefotaxime, nalidixic acid, imipenem and norfloxacin. Biochemical fingerprinting. E. coli strains were typed based on their biochemical phenotypes using the PhP-EC plates according to the methods described earlier (Katouli et al., 1991). Similarities between the strains were calculated as correlation coefficients as described before33 and clustered according to the unweighted pair group method with arithmetic averages (UPGMA)34 yielding a dendrogram. In the dendrogram, each strain is represented by a horizontal line. Different strains are connected with vertical lines at the similarity level they show to each other and thus the higher this line is, the more similar are the strains. An identity level of 0.975 was set based on reproducibility of the system as described before.33 Strains showing similarities to each other higher than this value were regarded as identical and assigned to the same biochemical phenotype (BPT). BPTs with more than one isolate were called common (C) and those with only one isolate were called single (S) BPTs. Statistical analysis. Fisher’s exact two-tailed test was used for statistical analysis.
Results Of 87 E. coli strains tested, 83 (95%) produced one or more of the virulence properties tested with aerobactin in 61 (70%) and type 1 fimbriae in 63 (72%) strains being the virulence factors most commonly expressed (Table 1). The prevalence of these virulence factors did not differ significantly between the male and female patients (Table 1). Altogether, 29 different O:K serotypes were identified among the strains. The O typeable strains were distributed among 12 serogroups with strains belonging to O groups 4, 6, 7 and 17 comprising 76% of the O typeable strains (Table 1). Biochemical fingerprinting of the strains showed the presence of different biochemical phenotypes
M. Katouli et al.
(BPTs) within each O group. All strains of O groups 4 and 6 but not 7 or 17, expressed multiple virulence factors (Figs. 1 and 2(A)). Production of P-fimbriae, haemolysin and aerobactin among strains of O groups 4 and 6 were significantly higher than among those of other prevalent serogroups (Figs. 1 and 2(A)). Generally, strains belonging to the same BPTs (i.e. C-BPTs) and having the same O:K serotypes, had similar virulence factors (Figs. 1 and 2(A)). Nineteen (22%) strains were O antigen nontypeable (ON) and belonged to diverse BPTs with 84% having K antigen non-typeable (KN) (Fig. 2(A)). Of eight strains in O group 17, six cross-reacted with the antisera against O 77 in our serotyping assay. Biochemical fingerprinting of these strains showed the presence of one common BPT (i.e. C1) containing both O17 and O77 strains all being K antigen nontypeable (Fig. 2(B)). Four strains were negative for any of the virulence markers tested and 17 strains produced only one virulence factor. In contrast, 66 (76%) strains expressed two or more virulence factors ðP , 0:0001Þ (Table 2). Coexpression of these virulence factors among the strains and their distribution among O:K serotypes did not significantly differ from each other (Table 2). Production of P-fimbriae was more pronounced among the strains coexpressing 4 or more virulence markers (18/21 strains) than those producing 3 or less (13/44 strains) ðP ¼ 0:0255Þ (Table 2). There was no difference between the number of strains producing aerobactin with or without the expression of Pfimbriae (32 vs. 29 strains). Contrary to that, all P-fimbriae producing strains ðn ¼ 32Þ expressed aerobactin (Table 3). Of 61 aerobactin producing strains, 45 (74%) strains also carried type 1 fimbriae (Table 3). There was a negative correlation between the presence of aerobactin and the expression of haemolysin or MRHA. Similarly, negative correlations were found between the presence of type 1 fimbriae and expression of MRHA or P-fimbriae (Table 3). Strains having K1, K5 and K12 comprised 63% of the strains with known capsular types (Table 4). The prevalence of virulence factors among these strains did not differ significantly. However, hemolysin and MRHA tend to be slightly higher, although significancy not different, among the K12 strains. Likewise, production of aerobactin and type 1 fimbriae was also more pronounced among these strains than those having other K antigens (Table 4). Eighty-five out of the 87 strains were tested for their patterns of susceptibility to 10 antibiotics commonly used for treatment of UTI. Resistance were found against ampicillin in 65 strains (76%), trimethoprim in 53 strains (62%), cotrimoxazol in 52
E. coli cystitis in Iran
315
Table 1 Prevalence of different virulence factors and serotypes among the Escherichia coli strains isolated from 87 adult patients with acute cystitis in Iran Virulence factor
None Haemolysin Aerobactin P-fimbriae MRHA Type 1 fimbriae MSHA (human A)c
a b c
Number (%) isolates M ðn ¼ 24Þ
F ðn ¼ 63Þ
Total
1 (4) 8 (33) 15 (63) 5 (21) 6 (25) 18 (75) 6 (25)
3 (5) 28 (44) 46 (73) 27 (43) 21 (33) 45 (71) 21 (33)
4 (5) 36 (41) 61 (70)a 32 (37) 27 (31) 63 (72)b 25 (29)
Serotypes (number)
Number (%) of isolates
O6:K5 (7), K13 (5), K2 (1), K14 (1), KN (7) O4:K12 (7), K3 (2), K1 (1), KN (1) O7:K1 (5), KN (2) O17 (77):K1 (1), K10 (1), KN (6) O21:K11 (1), KN (3) O8:KN O15:K2 O2:K1 O18:KN O19:KN O1:KN O75:KN OR:K12 (1), K1 (1), KN (4) ON:KN (16), K2 (2), K12 (1)
21 (24.0) 11 (12.6) 7 (8.0) 8 (9.2) 4 (4.6) 3 (3.4) 2 (2.3) 2 (2.3) 1 (1.1) 1 (1.1) 1 (1.1) 1 (1.1) 6 (6.9) 19 (21.8)
Aerobactin vs. haemolysin ðP ¼ 0:04Þ; P-fimbriae ðP ¼ 0:02Þ and MRHA ðP ¼ 0:004Þ: Type 1 fimbriae vs. haemolysin ðP ¼ 0:03Þ; P-fimbriae ðP ¼ 0:01Þ and MRHA ðP ¼ 0:002Þ: Strains showing only mannose-sensitive haemagglutination (MSHA) with erythrocytes from human blood group A. MRHA, mannose resistant haemagglutination; M, male; F, female; ON, O antigen non-typeable; OR, rough strain; KN, K antigen non-typeable.
Table 2 Coexpression of different virulence factors and their distribution among O:K serotypes of Escherichia coli strains isolated from adult patients with cystitis in Iran Coexpression of virulence factors
Number of isolates
Serotype (number of isolates)
None
4
ON:KN (2), OR:KN (1), O21:KN (1) a
Only 1 Hly MRHA Abn Type 1
17 (6) 3 1 7 6
2 Hly/Abn Hly/Type 1 P-Fim/Abn MRHA/Type 1 Abn/Type 1
24 (13)a 1 8 2 1 12
3 Hly/MRHA/Type 1 Hly/Abn/Type 1 P-Fim/Abn/Type 1
20 (5)a 3 1 11
MRHA/Abn/Type 1
5
4 Hly/P-Fim/MRHA/Abn Hly/P-Fim/Abn/Type 1 P-Fim/MRHA/Abn/Type 1 Hly/MRHA/Abn/Type 1
16 (1)a 6 5 2 3
O15:K2 (2), O6:K5 (1), O4:K3 (2), O17 (77): K1 (1) ON:KN (1), O21:KN (1), O6:K5 (1), O6:KN (1), O4:KN (1) O4:K1 (1), ON:KN (1) O6:KN (1), O4:K12 (1), O18:KN (1)
All 5
6
O4:K12 (3), O6:K12 (1), ON:KN (1), OR:K12 (1)
O6:K13 (1), O6:KN (1), O2:K1 (1) O8:KN (1) O7:K1 (3), O4:K12 (1), ON:KN (1), OR:KN (1), O17 (77):KN (1) O2:K1 (1), O17:KN (1), OR:KN (2), ON:K2 (1), O17 (77):KN (1) O7:KN (1) O6:KN (2), O6:K13 (4), O6:K2 (1), ON:K2 (1) ON:KN (1), O7:KN (1) O17:K10 (1) O6:K14 (1), O6:K5 (1), ON:KN (5), OR:KN (1), O8:KN (2), O7:K1 (1), OR:K1 (1) O4:K12 (2), ON:K12 (1) ON:KN (1) O75:KN (1), ON:KN (2), O6:K5 (4), O6:KN (1), O7:K1 (1), O17 (77):KN (2) O21:K11 (1), O21:KN (1), O1:KN (1), O19:KN (1), O17 (77): KN (1)
P-Fim, P-fimbriae; Hly, haemolysin; Abn, aerobactin; MRHA, mannose-resistant haemagglutination; Type 1, Type 1 fimbriae; ON, O antigen non-typeable; OR, rough strain; KN, K antigen non-typeable. a Numbers in bracket represent strains that showed only mannose-sensitive haemagglutination of human red blood cells.
316
M. Katouli et al.
Figure 1 Biochemical phenotypes (BPTs), K antigens and virulence properties of Escherichia coli strains of serogroup O6 (21 strains) (dendrogram A), serogroup O4 (11 strains) (dendrogram B) and serogroup O7 (7 strains) (dendrogram C) isolated from adult patients with cystitis in Iran. ID-level: Identity level above which strains are regarded as identical. Hly, haemolysin; MRHA, mannose-resistant haemagglutination; P-Fim, P-fimbriae; Type 1, type 1 fimbriae; KN, K antigen non-typeable; NT, not tested; S1-7, single BPTs 1 to 7; C1-3, common BPTs 1 to 3. Numbers of strains expressing different virulence factors have been shown for each serogroup. Significance of differences between O groups has been calculated only for the expression of P-Fim, Hly and Abn. O6 vs. O7 ðP ¼ 0:0113Þ; O4 vs. O7 ðP ¼ 0:0031Þ:
strains (61%), mecillinam in three strains (3%), cefaroxim in two strains (2%), and nitrofurantoin and nalidixic acid (in one strain each). 55 strains (65%) were resistant to two or more antibiotics with 45 strains (53%) resistant to trimethoprim, cotrimoxasol and ampicillin (Table 5). The latter three antibiotics constitute the highest rate of resistant among the isolates. Eleven strains were sensitive to all antibiotics tested (Table 5).
Discussion The success or failure of a bacterial invasion is determined by the contest between the host defence mechanisms and the virulence properties of the invading bacteria. Iron starvation is one of the major barriers that virulent bacteria must overcome in order to proliferate in the host.35 One of the most efficient systems for obtaining iron
E. coli cystitis in Iran
317
Figure 2 Biochemical phenotypes (BPTs), K antigens and virulence properties of Escherichia coli strains having O antigen non-typeable (ON; 19 strains) (dendrogram A) and of serogroup O17 (or O77) (8 strains) (dendrogram B). ID-level: identity level above which strains are regarded as identical. Hly, haemolysin; MRHA, mannose-resistant haemagglutination; P-Fim, P-fimbriae; Type 1, type 1 fimbriae; KN, K antigen non-typeable; S1-17, single BPTs 1 to 17; C1, common BPT 1. Numbers of strains expressing different virulence factors have been shown for each serogroup. Significance of differences between O groups has been calculated only for the expression of P-Fim, Hly and Abn. O6 (see Fig. 1) vs. ON ðP ¼ 0:0204Þ; and O17 (or O77) ðP ¼ 0:0424Þ: O4 (see Fig. 1) vs. ON ðP ¼ 0:0049Þ; and O17 (or O77) ðP ¼ 0:0083Þ:
by E. coli strains causing septicaemia or UTI is the production of siderophore aerobactin.36 In the present study, we found a higher incidence of aerobactin producing strains among our cystitis patients than those reported elsewhere.23,24,37
Aerobactin has often been reported to be associated with P-fimbriae in strains causing cystitis,37,38 however, we found that only half of the aerobactin producing strains in this study expressed P-fimbriae. Contrary to that, all P-fimbriated strains
Table 3 Correlation between the expression of different virulence factors among Escherichia coli strains isolated from 87 adult patients with cystitis in Iran Virulence factor
Aerobactin Haemolysin P-fimbriae MRHA Type 1 fimbriae
Number of strains
61 36 32 27 63
Number (%) of strains coexpressed Aerobactin
Haemolysin
P-fimbriae
MMA Type 1
Fimbriae
– 22 32 22 45
22 (36.1)a – 17 (53.1) 18 (66.7) 26 (41.3)
32 17 – 14 24
22 (36.1)a 18 (50.0) 14 (43.7) – 20 (31.7)e
45 (73.8)a 26 (72.2)b 24 (75.0)c 20 (74.1)d –
(61.1) (100)c (81.5)d (71.4)e
(52.4) (47.0) (51.8) (38.1)e
MRHA, mannose-resistant haemagglutination; Hly, haemolysin; Abn, aerobactin; Type 1, Type 1 fimbriae. a P ¼ 0:08 for Hly2 vs. Hlyþ strains and MRHA2 vs. MRHAþ strains; P ¼ 0:002 for Type 1þ vs. Type 12 strains. b P ¼ 0:04 for Type 1þ vs. type 12 strains. c P ¼ 0:00001 for Abnþ vs. Abn2 strains; P ¼ 0:03 for Type 1þ vs. Type 12 strains. d P ¼ 0:008 for Abnþ vs. Abn2 strains; P ¼ 0:05 for Type 1þ vs. Ttype 12 strains. e P ¼ 0:008 for Abnþ vs. Abn2 strains; P ¼ 0:02 for MRHAþ vs. MRHA2 strains and P ¼ 0:16 for P-Fim2 vs. P-fimþ strains.
318
M. Katouli et al.
Table 4 Distribution of different virulence properties among K-antigens found in Escherichia coli strains isolated from 87 adult patients with cystitis in Iran K antigen
K1 K2 K3 K5 K10 K11 K12 K13 K14 KN Total no.
Number of strains (%)
10 (11.5) 5 (5.7) 2 (2.3) 7 (8.0) 1 (1.1) 1 (1.1) 9 (10.3) 5 (5.7) 1 (1.1) 46 (52.9) 87 (100)
Number of strains producing Hemolysin
P-fimbriae
MRHA
Aerobactin
Type 1 fimbriae
2a 4 2 2a – – 8a 5 – 13
3 2 2 6 – – 4 – – 15
2a 2 2 1a 1 1 8a – – 10
8b 2 2 7b – 1 6b – 1 34
5 3 – 6b 1 1 8b 4 1 34
36
32
27
61
63
MRHA, mannose resistant haemagglutination; KN, K antigen non-typeable. a (P ¼ 0:1260; K12 vs. K1 and K5) for both virulence factors. b Not significant.
coexpressed aerobactin with 50% also producing haemolysin. This finding indicates that the iron sequestering system may be more important than Pfimbriation for pathogenicity of our strains and that P-fimbriated strains may carry more than one iron acquiring mechanisms. Association of aerobactin with a particular serotype has also been described for clonal groups of human invasive strains of E. coli K1.39 In the present study, aerobactin production was more pronounced among the strains of K antigens 1, 5 and 12 compared to others. It has been shown that certain O serotypes such as O1, O2, O4, O6, O7, O8, O16, O18, O25 and O75
are strongly associated with UTI among European studies.40,41 Fifty-four percent of our isolates belonged to these so called European Uropathogenic clones. In contrast, 42% of the strains (including non-typable strains) belonged either to serotypes different from those found in Europe or antisera against them were not available. A surprisingly high incidence of males were observed in the present study. The possibility that some of our patients had prostatitis was not appealing to us as none of the patients had classical symptoms associated with prostatitis. Besides obtaining samples from prostatitis patients requires
Table 5 Pattern of antibiotic resistance among 85 out of 87 Escherichia coli strains isolated from adult patients with cystitis in Iran Resistant to antibiotics
Pattern of resistance
None
Number of strains 11 (14%)
Only 1 Amp Trim
18 (21%) 1 (1%)
Amp/Trim Trim/Ctx Cfx/Mcl
2 (2%) 6 (7%) 1 (1%)
Amp/Trim/Ctx Trim/Ctx/Nfn
42 (49%) 1 (1%)
Amp/Trim/Ctx/Mcl Amp/Trim/Ctx/Nx
1 (1%) 1 (1%)
Amp/Trim/Ctx/Mcl/Cfx
1 (1%)
2
3
4
5 Amp, ampicillin; Trim, trimetoprim; Ctx, cotrimoxazol; Mcl, mecillinam; Nfn, nitrofurantoin; Cfk, cefaroxim; Nx, nalidixic acid.
E. coli cystitis in Iran
a vigorous procedure, which was not done in this study. One possible explanation, which may partly explain the high ratio of male over female with cystitis in this study is that young females experiencing UTI may not, due to cultural differences, seek clinical help as often as males do. It is well known that many women first try house-hold remedies which are common among traditional families. During the last few years, considerable progresses have been made to identify factors specifically or non-specifically involved in the pathogenesis of UTI. It is now clear that pathogenic bacteria attach and then enter the host tissue, proliferating and eventually cause a definite damage. The role of type 1 fimbriae as a potential adhesive factor in the pathogenesis of UTI has been clearly demonstrated in experimental animals.4,5 In our study, type 1 fimbriae was produced by majority of the strains and most of them coexpressed aerobactin, suggesting that these two virulence factors play an important role in pathogenesis of E. coli strains causing cystitis in Iran. A key feature of uropathognesis is the ability to adhere to bladder mucosa. P-fimbriae is a major type of adhesive fimbriae in strains causing pyelonephritis.3,42 The number of P-fimbriated strains in our study was much higher than those reported for cystitis patients by others.24,42,43 Its prevalence was more pronounced among strains expressing type 1 fimbriae, indicating that the fimbriation status of E. coli strains causing cystitis in our study is varying from predominantly type 1 fimbriae to a mixture of type 1 and P fimbriated cells, a result which has also been found by others.5 Production of haemolysin by E. coli strains is a mechanism of scavenging iron under iron-limited conditions. However, it was interesting to note that more than 61% of haemolysin-producing strains also produced aerobactin. It has been postulated that bacteria containing haemolysin have no need to produce aerobactin since they can obtain the required iron from lysis of red blood cells.38 The high association of haemolysin and aerobactin among our strains, therefore, indicates that these strains have a greater iron-sequestering ability than those producing either of these two factors alone. Furthermore, our finding that almost 50% of Pfimbriated strains also produced haemolysin suggests a potential role for combination of these three virulence factors in pathogenesis of E. coli cystitis in this part of the country. Metabolic fingerprinting of the strains indicated the presence of different biochemical phenotypes (BPTs) among different O serogroups with one BPT being dominant in each group. Using a combination
319
of biochemical fingerprinting, O serogroup and virulence factors, we previously characterised enteropathogenic E. coli strains isolated from Iran and found that strains carrying similar virulence properties had similar BPTs and belonged to the same O serogroup although, different BPTs were also observed within each group.44 In the present study, common BPTs of serogroups O6, O4 and O7 generally carried strains with either identical Ktype or non-typeable K antigen (see Fig. 1). Often there was variation among the type of virulence factors expressed by some strains within each O:K serotype. Bacterial clones which exist within a population for a long time may undergo changes in phenotypic characters and virulence properties.45 Therefore, it is possible that some of our strains had lost or gained one or more of their virulence factors or had gone through changes in their O or K antigens, a phenomenon which can be induced by factors such as antibodies of the host.24 We also found that strains of the same O:K serotype (e.g. O6:K13 in Fig. 1(a) or O4:K12 in Fig. 1(b)) belonged to different BPTs which is not surprising since biochemical reactions among E. coli strains vary widely. Nonetheless, the fact that most strains belonging to a common BPT and having the same O:K serotype, carried the same virulence factors, is in agreement with the clonal concept of pathogenic bacteria.45 The presence of a high percentage of both O and K non-typeable strains in our study is rather surprising. Both biochemical fingerprinting and virulence profile of these isolates showed that they belonged to diverse groups of E. coli strains ruling out the possibility that they belonged to a common and/or newly emerged clonal group causing cystitis among these patients. We postulate that these strains had been members of the normal flora that have picked up the ability to colonise urinary tract and cause infection, since exchange of genetic materials among bacteria of the gut is common. To our knowledge, this is the first report describing the prevalence of different virulence properties associated with E. coli cystitis in Iran. Several other important virulence properties such as cytotoxic necrotising factor,19 and different classes of adhesins such as Dr family of adhesins have been shown to play an important role in pathogenesis of UTI.46 In this study, we investigate the presence of some virulence attributes, and we here show that a high portion of E. coli strain causing cystitis belong to non-typeable serotypes (both O and K types) which are not common to those found in Europe and that majority of them contain a combination of different virulence properties and are resistant to three commonly used antibiotics.
320
References 1. Roberts AP, Phillips R. Bacteria causing symptomatic urinary tract infection or bacteriurea. J Clin Pathol 1979;32:492—6. ˚, Kaijser B, et al. Comparison of 2. Lidin-Janson G, Hanson AA Escherichia coli from bacteriuric patients with those from feces of healthy school children. J Infect Dis 1977;136: 346—53. 3. Hagberg L, Jodal LA, Korhonen T, Lidin-Janson G, Lindberg U, Svanborg-Eden C. Adhesion, hemagglutination and virulence of Escherichia coli causing urinary tract infections. Infect Immun 1981;31:564—70. 4. Iwahi T, Abe Y, Nakao M, Imada A, Tsuchiya K. Role of type 1 fimbriae in pathogenesis of ascending urinary tract infection induced by Escherichia coli in mice. Infect Immun 1983;39: 1307—15. 5. Kisielius PV, Schwan WR, Amundsen SK, Duncan JL, Schaeffer AJ. In vivo expression and variation of Escherichia coli type 1 and P pili in the urine of adults with acute urinary tract infections. Infect Immun 1989;57:1656—62. 6. Svanborg-Eden C, Lidin-Janson G, Lindberg U. Adhesiveness to urinary tract epithelial cells of fecal and urinary Escherichia coli isolates from patients with symptomatic urinary tract infections or asymptomatic bateriuria of varying duration. J Urol 1979;122:185—8. 7. Ka ¨llenius G, Mo ¨llby R, Winberg J. In vitro adhesion of uropathogenic Escherichia coli to human periuretheral cells. Infect Immun 1980;28:972—80. 8. Schaeffer A, Schwan WR, Hultgren SJ, Duncan JL. Relationship of type 1 pilus expression in Escherichia coli to ascending urinary tract infections in mice. Infect Immun 1987;373—80. ¨ quero ME, Aron L, DeLuca AG, Timmis KM, Cabello FC. A 9. A plasmid-encoded outer membrane protein, tra T, enhances resistance of Escherichia coli to phagocytosis. Infect Immun 1984;46:740—6. 10. Taylor P. Bacteriocidal and bacteriolytic activity of scrum against gram-negative bacteria. Mirobiol Rev 1983;47: 46—83. 11. Hughes C, Phillips R, Roberts AP. Serum resistance among Escherichia coli strains causing urinary tract infection in relation to O type and carriage of hemolysin, colin and antibiotic resistance determinants. Infect Immun 1982;35: 270—5. 12. Cross AS, Gemski P, Sadoff JC, Ørskov F, Ørskov I. The importance of the K1 capsul in invasive infections of Escherichia coli. J Infect Dis 1984;149:184—93. 13. Kaijser B. Immunology of Escherichia coli: K antigen and its relation to urinary tract infection. J Infect Dis 1973;127: 670—7. ˚, Jodal U, Lidin-Janson G, Robbins JB. 14. Kaijser B, Hanson LA Frequency of Escherichia coli K antigens in various forms of urinary tract infection in children. Lancet 1977;I:663—4. 15. Opal S, Cross A, Gemski P. K antigen and serum sensitivity of rough Escherichia coli. Infect Immun 1982;37:956—60. 16. Hughes C, Hacker J, Roberts A, Goebel W. Hemolysin production as virulence marker in symptomatic and asymptomatic urinary tract infections caused by Escherichia coli. Infect Immun 1983;39:546—51. 17. Smith HW, Huggins MB. The toxic role of alpha hemolysin in pathogenesis of experimental Escherichia coli infection in mice. J Gen Microbiol 1985;131:395—403. 18. Alonso P, Blanco J, Blanco M, Gonzales EA. Frequent production of toxins by Escherichia coli strains from urinary tract infections: relation with haemagglutination. FEMS Microbiol Lett 1987;48:391—6.
M. Katouli et al.
19. Caprioli A, Falbo V, Ruggeri FM, Baldassarri L, Bisicchi R, Ipollito G, Romoli E, Donelli G. Cytotoxic nectrtizing factor production by hemolytic strains of Escherichia coli causing extraintestinal infections. J Clin Microbiol 1987;25:146—9. 20. Williams PH, Warner PJ. Col V plasmid-mediated, colicin Vindependent iron uptake system of invasive strains of Escherichia coli. Infect Immun 1980;29:411—6. 21. Jacobson SH, Tullus K, Baruner A. Hydrophobic properties of Escherichia coli causing acute pyelonephritis. J Infect 1989; 19:17—23. 22. Brauner A, Katouli M, Tullus K, Jacobsson SH. Cell surface hydrophobicity, adherence to Hela cell cultures and haemagglutination pattern of pyelonephritogenic Escherichia coli strains. Epidemiol Infect 1990;105:255—63. 23. Carbonetti NH, Boonachi S, Parry SH, Va ¨isa ¨nen-Rhen V, Korhonen TK, Williams PH. Incidence of aerobactin iron uptake by Escherichia coli isolates from human extraintestinal infections. Infect Immun 1986;51:966—8. ¨ sterberg E, 24. Jacobson SH, Hammarlind M, Lidefeldt KJ, O Tullus K, Brauner A. Incidence of aerobactin-positive Escherichia coli strains in patients with symptomatic urinary tract infection. Eur J Clin Microbiol Infect Dis 1988;7:630—4. 25. Clarridge JE, Pezzlo MT, Vosti KL, Weissfeld AS. Laboratory diagnosis of urinary tract infections. Cumitech, 2A. Washington, DC: American Society for Microbiology; 1987. 26. Ørskov I, Ørskov F. Escherichia coli in extra-intestinal infections. J Hyg Camb 1985;95:551—75. 27. Ku ¨hn I. Biochemical fingerprinting of Escherichia coli: a simple method for epidemiological investigations. J Microbiol Methods 1985;3:159—70. 28. Edwards PR, Ewing WH. Identification of Enterobacteriacaea. 3rd ed. Minneapolis: Burgrass Publishing Co; 1971. 29. Svenson SB, Ka ¨llenitis G, Mo ¨llby R, Hultberg H, Winberg J. Rapid identification of P-fimbriated Escherichia coli by a receptor-specific particle agglutination test. Infection 1982; 10:209—14. 30. Duguid JP, Clegg S, Wilson MI. The fimbrial and non-fimbrial haemagglutinations of Escherichia coli. J Med Microbiol 1979;12:213—27. 31. Lidin-Janson G, Falsen E, Jodal U, Kaijser B, Lincoln K. Characteristics of antibiotic resistant Escherichia coli in the rectum of healthy school-children. J Med Microbiol 1977;10: 299—308. 32. Kaijser B. A simple method for typing of acidic polysaccharide K antigens of Escherichia coli. FEMS Microbiol Lett 1977; 1:285—8. 33. Mo ¨llby R, Ku ¨hn I, Katouli M. Computerized biochemical fingerprinting—a new tool for typing of bacteria. Rev Med Microbiol 1993;4:231—41. 34. Sneath PHA, Sokal RR. Numerical taxonomy. San Francisco: W.H. Freeman; 1973. 35. de Lorenzo V, Martinez JL. Acrobactin production as a virulence factor: a reevaluation. Eur J Clin Microbiol Infect Dis 1988;7:621—9. 36. Martinez LJ, Delgado-Iribarren A, Baquero F. Mechanisms of iron acquisition and bacterial virulence. FEMS Microbiol Rev 1990;75:45—56. 37. Ørskov I, Svanborg-Eden C, Ørskov F. Aerobatin production of serotyped Escherichia coli from urinary tract infection. Med Microbiol Immunol 1988;177:9—14. 38. Johnson JR, Mosley SL, Roberts PL, Stamm WE. Aertobactin and other virulence factor genes among strains of Escherichia coli causing urosepsis: association with patients characteristics. Infect Immun 1988;56:405—12. 39. Valvano MA, Silver RP, Crosa JH. Occurence of chromosomeor plasmid-mediated aerobactin iron transport system and hemolysin production among clonal groups of human
E. coli cystitis in Iran
40.
41.
42.
43.
invasive strains of Escherichia coli K1. Infect Immun 1986; 52:192—9. Vaisanen-Rhen V, Elo J, Vaisanen E, et al. P-fimbriated clones among uropathogenic Escherichia coli strains. Infect Immun 1984;43:149—55. Truck M, Petersdorf RG, Fournier MR. The epidemiology of non-enteric Escherichia coli infections: prevalence of serological groups. J Clin Invest 1962;41:1760—5. Ka ¨llenius G, Mo ¨llby R, Svenson SB, et al. Occurence of Pfimbriated Escherichia coli in urinary tract infections. Lancet 1981;2:1369—72. Winberg J, Bollgren I, Ka ¨llenius G, Mo ¨llby R, Svenson S. Bacterial virulence and host defence in pathogenesis of renal infection and renal scaring. In: Francois B, Perrin B, editors.
321
Urinary tract infection insights and prospects. London: Butterworths Publications; 1983. p. 19—33. 44. Katouli M, Ku ¨llby R. Biochemical phenotypes of ¨hn I, Mo enteropathogenic Escherichia coli common to Iran and Sweden. J Med Microbiol 1991;35:270—7. 45. Ørskov F, Ørskov I. Summary of workshop on the clone concept in epidemiology, taxanomy and evolution of Entrobacteriacae and other bacteria. J Infect Dis 1983; 148:346—57. 46. Nowicki B, Labigne A, Moseley S, Hull R, Hull S, Moulds J. The Dr hemagglutinin, afimbrial adhesins AFA-I and AFA-III, and F1845 fimbriae of uropathogenic and diarrhoea-associated Escherichia coli belong to family of hemagglutinins with Dr receptor recognition. Infect Immun 1990;58:279—81.