Immunofluorescence Study of Fimbrial Phase Variation in Escherichia coli KS71

Transcription

1 JOURNAL OF BACTERIOLOGY, Nov. 1984, p /84/ $02.00/0 Copyright X) 1984, American Society for Microbiology Immunofluorescence Study of Fimbrial Phase Variation in Escherichia coli KS71 Vol. 160, No. 2 BOGDAN NOWICKI,t MIKAEL RHEN, VUOKKO VAISANEN-RHEN, AULI PERE, AND TIMO K. KORHONEN* Department of General Microbiology, University of Helsinki, SF-00280, Helsinki 28, Finland Received 14 May 1984/Accepted 14 August 1984 An immunofluorescence assay was developed to study fimbrial phase variation in a pyelonephritogenic Escherichia coli strain, KS71. By using fluorochrome-labeled antibodies specific for either P, type-1c, or type- 1 fimbriae of strain KS71, it was shown that in a broth culture of strain KS71 the fimbrial types mostly occured on different cells. Only 9% of the cells carried more than one fimbrial type. The KS71 cell population was fractionated into subpopulations expressing only one of the fimbrial types or lacking fimbriae. Immunofluorescence assay of the subpopulations revealed a rapid phase variation in fimbrial synthesis. Kinetic analyses of a nonfimbriated cell population suggested that a change from one fimbrial phase to another was not totally random. Enterobacterial fimbriae are protein filaments that function as binding organelles (4, 14, 28). It has become evident that single enterobacterial species, e.g., Escherichia coli, have many different types of fimbrial antigens. The receptor structures for only some of them are known. Most E. coli strains possess type 1 fimbriae, which are characterized by their ability to bind mannosides (4, 13, 18, 28). E. coli strains associated with human pyelonephritis carry P fimbriae (16, 33) which bind to P-blood-group-specific glycosphingolipids on human uroepithelial cells and erythrocytes (11, 17). A few human pathogenic strains have S fimbriae, which recognize sialyl galactosides (16a, 22), or M fimbriae, which recognize human-blood-group-m-specific antigens on human erythrocytes (32). Additionally, E. coli strains causing diarrhea in humans and animals possess fimbriae, such as colonization factor antigens or K88 or K99 antigens, which bind the bacteria to host intestinal epithelium (7). It has been shown that a particular enterobacterial strain is able to synthesize many fimbrial antigens (4, 9, 16, 31, 33). Brinton (1) and Duguid and Gillies (3) showed that type-1 fimbriae have phase variation, i.e., bacterial cells rapidly change from a fimbriated to a nonfimbriated phase. The genetic analysis of this on-off switch has been started (5). However, it has remained largely unresolved whether single bacterial cells express many fimbriae simultaneously. As a first step in studying this problem, we recently described fimbrial phase variation in KS71, a pyelonephritogenic E. coli strain (26). This strain has four fimbrial antigens, A, B, C, and D. The A and B fimbriae are serologically crossreactive (27) and recognize P-blood-group antigens (25); hence, they both are P fimbriae. The D fimbriae of strain KS71 correspond to type 1 fimbriae; they bind mannosides and are formed only after cultivation in static broth (27). The C fimbriae resemble pseudotype 1, or type 1C, fimbriae (12, 24). They are serologically distinct from the other KS71 fimbriae and lack demonstrable binding properties. The fimbrial composition of strain KS71 is very similar to that of E. coli C1212, the model strain for F7 fimbrial antigens (12, 20). We showed that P (A and B) and C fimbriae of agar- * Corresponding author. t Present address: Department of Microbiology, Medical School in Gdansk, Gdansk, Poland. 691 grown strain KS71 do not occur on the same cells and that there is a rapid phase variation between P-fimbriated and C- fimbriated cells (26). In this communication we describe a simple and fast immunofluorescence method for assessing the expression of fimbrial antigens on single bacterial cells and extend our previous findings by demonstrating also that type 1 fimbriae are involved in this phase variation. MATERIALS AND METHODS Bacteria. E. coli KS71 and its fimbriae have already been described by us (16, 24-27). For the immunofluorescence assay the strain was first cultured overnight on colonization factor antigen agar plates (6) at 37 C, and 109 cells were then transferred to 20 ml of Luria broth and cultured statically at 37 C (15). Nonfimbriated cells were obtained by subculturing the strain on colonization factor antigen agar plates at room temperature. Antisera. Antisera against purified A and C fimbriae of strain KS71 were prepared as described (15) by immunizing with fimbriae purified from recombinant strains with the structural genes for A or C fimbriae only (25). Anti-KS71A serum (anti-a serum) cross-reacts strongly with B fimbriae of strain KS71 (25, 27) and hereafter will be called anti-p serum. Antiserum to the type 1 fimbriae of E. coli 2131 was available from previous work (16, 24, 27); it reacts strongly only with the D fimbriae of strain KS71 (16, 24-27). Preparation of fluorochrome-labeled antibodies. Isolation of the immunoglobulin G (IgG) fraction from the various antisera and its conjugation with fluorescein isothiocyanate (FITC) or tetramethylrhodamine B isothiocyanate (TRITC) (Sigma Chemical Co., St. Louis, Mo.) were performed as described (8), except that IgG was further purified by ionic exchange chromatography in a DEAE-cellulose column (Pharmacia Fine Chemicals, Uppsala, Sweden) before conjugation. In each case 5 ml of hyperimmune serum was used for purification. Unbound stain was removed from conjugates by gel filtration in a Sephadex G25 column (Pharmacia) with phosphate-buffered saline (PBS) (ph 7.1) as eluant. The conjugates were then dialyzed against PBS and adjusted to 3 ml. The fluorochrome/protein ratio was 2.0 for anti-type-1 FITC conjugate, 2.1 for anti-p FITC conjugate, 2.3 for anti-p TRITC conjugate, and 1.9 for anti-c TRITC conjugate. Immunofluorescence assay. Samples from bacterial cultures were stained with the various conjugates (Table 1) as

2 692 NOWICKI ET AL. J. BACTERIOL. TABLE 1. Immunofluorescence staining of E. coli KS71 cells Percentage of cells with: Antifimbrial sera (conjugate) used in staining No. ofe Only type 1 Only P Only C More than No counted Oltye1 OlP OnyC 1 fimbrial fmra fimbriae fimbriae fimbriae type fimbriae Anti-type-i (FITC) + anti-p (TRITC) 1, Anti-type-i (FITC) + anti-c (TRITC) 1, v Anti-type-i (FITC) + anti-p (TRITC) + anti-c (TRITC) 1, Anti-P (FITC) + anti-c (TRITC) 2, follows. The bacterial culture (10 Ixl) was pipetted onto a glass slide, dried at room temperature, and fixed for 10 min with 3.5% (wt/vol) paraformaldehyde (E. Merck AG, Darmstadt, Federal Republic of Germany) in PBS. The slides were then washed with PBS and incubated with the conjugates for 30 min in a moist chamber at room temperature. The slides were then washed with PBS, and a drop of 50% (wt/vol) PBS-buffered glycerol was added to the sample before topping with a glass cover. The samples were examined in a Zeiss fluorescence microscope with epiilluminator IV FL (Osram HBO 50-W high-pres'sure mercury lamp; excitation filters KP 500 and 490, dichroic mirrors FT 510, emission filter 520). Three to five fields of each sample were analyzed. A field was first photographed through a normal phasecontrast microscope to evaluate the total number of cells and then separately for FITC and TRITC stainings; finally, a double exposure of both stains was taken. The number of bacteria in each micrograph (phase-contrast, FITC stained, TRITC stained, or double exposure) of a certain field were then estimated. We used black-and-white prints, color prints, and color slides. Cell numbers were estimated by drawing the images of all micrographs of a given field on top of one another. Before routine assays, each conjugate was tested in twofold dilutions (in PBS) by immunofluorescence assay using strain KS71 from broth culture (37 C) and from agar plates (room temperature) and a type-1-fimbriated strain, 2131 (16). A dilution of an anti-p or anti-c conjugate that reacted strongly with strain KS71 from culture at 37 C but not at all with strain KS71 from cultures at room temperature and with strain 2131 was used in subsequent assays. Similarly, antitype-1 conjugates were used at dilutions reacting with strains 2131 and KS71 from culture at 37 C but not with KS71 from culture at room temperature. Most conjugates were used in 1:10 to 1:20 dilutions. Fractionation of KS71 cells. To test the reversibility of fimbrial phase variation, we isolated bacterial subpopulations expressing only one fimbrial type, P (A + B), C, or D (type-1), and a subpopulation of nonfimbriated cells. This was done in principle as described by us (26) and Deneke et al. (2). A type-i-fimbriated subpopulation was obtained by adsorbing bacteria to yeast cells (13), washing the aggregates five times with PBS at 4 C, and eluting bound bacteria with PBS containing 1.5% (wt/vol) a-methyl-d-mannoside (Sigma). The bacterial cells obtained were finally treated with anti-c and anti-p sera as described (26) to remove cells carrying these fimbrial antigens. A P-fimbriated subpopulation was prepared by adsorbing bacterial cells onto human OP1 erythrocytes in the presence of a-methyl-d-mannoside (2% [wt/vol]), washing the agglutinated erythrocytes five times with PBS, and eluting thermally at 370C (2). The subpopulation obtained was finally precipitated with anti-c serum. The yield of P-fimbriated cells was generally much less than that of type-1-fimbriated cells, and usually some erythrocytes contaminated the suspension. C-fimbriated cells were obtained by adsorbing the culture first onto human erythrocytes and then onto yeast cells as described above, except that no a-methyl-d-mannoside was added. Cells remaining in suspension were precipitated with anti-c serum and centrifuged as described (26). The precipitates were washed five times with PBS, and the resulting C- fimbriated cells were used after mixing in a whirly mixer. The cells remaining in suspension after precipitation with anti-c serum constituted the nonfimbriated subpopulation. Each of the fractionated subpopulations was immediately tested for hemagglutination, yeast cell agglutination, and agglutination with antifimbriae sera, as well as stained with conjugated IgG preparations to evaluate contamination by superflously fimbriated or nonfimbriated cells. The percentage of such cells was mostly less than 1% and at most 2% (usually ca. 500 cells were examined). Agglutination tests. Bacterial agglutination with yeast cells, human OP1 erythrocytes, or various antifimbriae sera were performed as described previously (13, 25). RESULTS Immunofluorescence assay. Figure 1 shows a typical result of immunofluorescence assay, obtained by staining a 35-h culture with anti-type-1 or anti-p and anti-c sera. Figure 1A is an underexposed micrograph of cells stained with antitype-1 serum; notably is the strength of fluorescence on the edges and its weakness at the center of the cells. The same phenomenon was observed with all conjugates and was considered to result from the staining of fimbriae protruding from the cell surface. Figure 1B is a phase-contrast micrograph where the staining reactions of each cell are shown for clarity. Figure 1C shows the same field photographed for a FITC conjugate (anti-type-1 IgG), Fig. 1D is for a TRITC conjugate (anti-p and anti-c IgG), and Fig. 1E is a double exposure for both fluorescein conjugates. Twelve cells (37%) were stained with anti-type-i conjugate; of these, eight (25% of all the cells; marked by a 1 in Fig. 1B) were stained with anti-type-1 conjugate only, and four (12%; marked by 1PC) were stained also with the mixture of anti-p and anti-c sera. The latter represented cells carrying more than one fimbrial type (i.e., type 1 and P or C, or both). Six cells (19%; marked by PC in Fig. 1B) were stained only with the mixture of anti-p and anti-c conjugates. The same fluorescent cells are visible in the double exposure of all conjugates. Fourteen (44%) of the cells shown in Fig. 1B (marked by an E) were not stained with any of the conjugates and were considered to represent nonfimbriated cells. The results shown in Fig. 1 suggested to us that most of type-1 and other fimbriae on a KS71 population might occur on separate cells. For further analysis we assayed all pairs of IgG conjugates by immunofluorescence. A culture grown in static broth for 35 h was stained with two or three conjugates

3 VOL. 160, 1984 A.;-. B 9 ~~~~~~3va.!Xi*J ;y-;4to this particular population (inoculated from an agar plate FIMBRIAL PHASE VARIATION 693 once, mostly on separate cells, and that only a fraction of the cells possess more than o6ne fimbrial type. Variation in the fimbrial-type frequencies shown in Table 1 was 10 to 15%; this was considered to reflect the accuracy of the immunofluorescence method and variability of the phase variation. Shown are the mean values for three ;, :t-o'-:?t....i;*a ;j- independent experiments. From the percentages in Table 1 we estimated the following composition for a KS71 cell population: 42% without fimbriae, 20% with type-i fimbriae, ~ ~~ ~~k 17% with P fimbriae, 16% with C fimbriae, and 5% with more s@s3 than one fimbrial type. The percentages probably apply only and cultured for 35 h in static L broth); subculturing in static broth is known to increase the number of type-l-fimbriated cells (3, 4). Reversibility of fimbrial expression. To test the stability of C the subpopulations expressing different fimbrial types, we fractionated the KS71 cell population into nonfimbriated cells and cells carrying only one of the fimbrial types. The proportion of superflously fimbriated or nonfimbriated cells in each subpopulation was mostly less than 1%, as estimated by immunofluorescence staining of the cells immediately after fractionation. The subpopulations were also tested for hemagglutination, yeast cell agglutination, and agglutination by the antifimbriae sera; none of them showed unexpected reactions in the tests; e.g., the type-i-fimbriated subpopula- D p tion showed yeast cell agglutination but lacked hemagglutination and was agglutinable by anti-type-1 but not by anti-p or anti-c antisera. It was concluded that the subpopulations were homogeneous enough for tests of reversibility in fimbrial expression. The existence of reversibility was tested in two ways. First, cells from the fractionated subpopulations were transferred to agar plates to obtain single colonies, and the E colonies were tested for hemagglutination, yeast cell aggluti- nation, and agglutination by the antifimbriae sera. Yeast cell agglutination was further confirmed by subculturing a fraction of the colonies in static broth; however, positive reac- 100 * FIG. 1. Immunofluorescence staining of E. coli KS71. (A) Underexposed micrograph of cells stained with anti-type-1 FITC e conjugate; note the strong staining at cell edges. (B) Phase-contrast */ micrograph of cells that are also shown for anti-type-1 FITC staining F. / (C), anti-p- and anti-c TRITC staining (D), and double exposure of Z 50.I FITC and TRITC conjugates (E). In (B) nonfimbriated cells are c ic marked by an E, type-l-fimbriated cells are marked by a 1, C- and/or Kl,.*-* P-fimbriated cells are marked by PC, and 1PC denotes cells stained W, with both (FITC and TRITC) conjugates. (A) x2800; (B through E) / 0o- x c<. / ) I - o/ \. o~~~~~~~~ as described in Table 1, and the number of cells stained with each conjugate was estimated. The results were similar to those shown in Figure 1: about half of the cells were not TIME (h) FIG. 2. Kinetics of fimbrial-phase E. coli stained at all, and most of the fluorescent cells were stained fimbriatrplate atroo temperature, KS71. with withonlyone one off te the cojugaes. conjugates. The percentage ercetageof clls cells Nonfimbriated were transferredcells, to static grown broth on at agar 370Cplates and assayed at roomevery temperature, hour for stained with more than one conjugate varied from 3 to 9% fimbriation by immunofluorescence. About 2,500 cells were examand, as expected, was highest when all three antifimbriae- ined each hour. Symbols: 0*.- 0, nonfimbriated cells; O, C-fimbri- IgG conjugates were used. It was concluded that in a KS71 ated cells; 0, P-fimbriated cells; -0, type-l-fimbriated cells; *, cell population type-i, P (A + B), and C fimbriae occur, at cells having both P and C fimbriae.

4 694 NOWICKI ET AL. tions were already visible on the agar plates. Fifty colonies from each subpopulation were tested, and all of the colonies reacted positively. Second, fractionated subpopulations were transferred into broth, cultured for 8 to 10 h, and stained with the three fluorescence conjugates (anti-p, anti- C, and anti-type-1). Again, a fraction of each cultured subpopulation was stainable with all three conjugates. These results show that fimbrial synthesis in strain KS71 is reversible and that the cells may change from one fimbrial phase to another. Kinetics of fimbrial phase variation. To learn whether a change from one fimbrial subpopulation to another occured randomly, i.e., whether cells in a particular phase changed randomly to all other fimbrial phases, we subcultured strain KS71 on agar plates at room temperature, a condition where none of the fimbriae are expressed. The cells were then transferred into a static broth culture at 37 C, and two samples were taken every hour for 26 h. One sample was stained with anti-type-1 FITC, anti-p TRITC, and anti-c TRITC conjugates, and the other was stained with anti-p FITC and anti-c TRITC conjugates, and the percentage of cells with each fimbrial type was estimated. Another culture, inoculated in an identical manner, was used to determine the number of bacterial cells in the culture. Viable counts (data not shown) revealed that the number of cells in the broth culture increased logarithmically from 8.4 x 107/ml to 7.8 x 108/ml in 3 h, remained fairly constant up to 15 h (8.2 x 108/ml), and rose to 2.2 x 109/ml at 26 h. The last increase probably resulted from surface growth becoming observable at that time. Cells from the agar plates kept at room temperature were not stained with any of the fluorescence conjugates (0 h in Fig. 2), nor were such cells hemagglutinative or agglutinable with any of the antifimbriae sera. C-fimbriated cells were first detected at 1 h; their percentage increased rapidly to 68% by 4 h and then decreased gradually to 24% by 26 h. Also, cells having both P and C fimbriae were detected at 1 h; their percentage remained low (1 to 7%) throughout the test. P-fimbriated cells were detected at 2 h, and their percentage rose gradually to 21% by 8 h and remained fairly constant through the rest of the experiment. Type-l-fimbriated cells were not detected until 9 h. Their percentage remained low until it increased to 21% at 20 h, and it remained above 10%. The percentage of cells having all three fimbrial types was also estimated; these cells were detected at 9 h and, their frequency remained low (2 to 6%; data not shown) throughout the experiment. Accordingly, the percentage of nonfimbriated cells decreased rapidly to 24% at 4 h and then fluctuated between 21 and 42%. The percentages at 26 h in Fig. 2 are in close agreement with those obtained with the 35-h-culture described in Table 1. We conclude that fimbrial phase variation is not totally random; a random variation would not explain the rapid increase in the percentage of C-fimbriated cells or the late appearance of type-l-fimbriated cells. DISCUSSION We recently described phase variation between P and C fimbriae of agar-grown E. coli KS71 (26). Our present communication supplements those findings and shows also that type-1 fimbriae are involved in such variation. Of importance in this communication are the findings that P, C, and type-1 fimbriae occur mostly on separate KS71 cells and that extremely rapid phase variation takes place between different fimbrial types (Table 1; Fig. 2). The fimbrial composition of strain KS71 is very similar to J. BACTERIOL. that of E. coli C1212, which is the model strain for F7 and type-ic fimbrial antigens (12, 19, 20). Thus, the A and B fimbriae of strain KS71 probably correspond to F71 and F72, respectively, and the C fimbriae of strain KS71 correspond to type 1C of strain C1212 (12, 19, 24-27). Both strains express type-i fimbriae in broth culture only. It is interesting to note that 0rskov et al. (21) found that adsorption of a C1212 cell population onto human erythrocytes removed cells carrying F7 antigens, whereas type-ic fimbrial antigens could still be detected on cells remaining in suspension. This is in agreement with our previous findings on phase variation between P and C fimbriae on strain KS71 (26) and can be explained by the fact that the two fimbrial types reside on different cells. 0rskov et al. (20) also provided immune electronmicroscopic evidence which suggested that F7 and type-1 fimbriae on broth-cultured strain C1212 could occur on the same cell, but our present findings (Fig. 1; Table 1) show that this is true for only a small fraction of KS71 cells, probably those changing from one fimbrial phase to another. An intriguing aspect of the fimbrial-phase variation under discussion is its extreme rapidity; an overnight culture of fractionated subpopulations produced colonies that were heterogeneous in respect to fimbrial antigens, and variation in fimbrial synthesis could be observed during a 26-h broth culture of the strain (Fig. 2). Thus, fimbrial phase variation in strain KS71 is much faster than flagellar phase variation in Salmonella typhimqrium (29), the on-off switch of type-1 fimbrial synthesis in E. coli (5), or fimbrial phase variation in Neisseria gonorrhoeae (30). All these variations can be followed by enumeration of colonies, although single colonies expressing different fimbrial types were observed by Swanson and Barrera in N. gonorrhoeae (30). Fimbrial phase variation in gonococci also involves changes in the apparent molecular weights of the fimbrillin subunits; we have not observed such changes in strain KS71. The results shown in Figure 2 indicate that fimbrial phase variation is not totally random. The rapid increase in the percentage of C-fimbriated cells can only be explained by the fact that after transfer of cells to broth culture, most of the new cells expressed C fimbriae. Cells having both C and P fimbriae were detected 1 h before the appearance of P- fimbriated cells; maybe some of the C-fimbriated cells had changed into a P-fimbrial phase. The late appearance of type-l-fimbriated cells also speaks against random change. It is an established fact that manye. coli strains do not express type-1 fimbriae on agar plates, whereas other fimbriae, like P or type 1C, are expressed (3, 4, 16, 20). The molecular mechanisms responsible for control of this phenomenom, as well as of fimbrial phase variation, are not known. We cloned separately the four fimbrial genes of strain KS71 (25; M. Rhen, unpublished data) to study the molecular basis for phase variation; so far our results show that there are regulatory interactions between the fimbrial locuses. It should be stressed that the kinetics of phase variation shown in Fig. 2 probably hold for only that particular phase, i.e., nonfimbriated cells grown on agar plates at room temperature. The immunofluorescence assay described in Fig. 1 is a useful tool for studying rapid variation in surface antigens. It may also be helpful for analysis of fimbrial synthesis in vivo, e.g., in patients or in test animals challenged with fimbriated bacteria. E. coli KS71 has been isolated from a case of acute pyelonephritis, and similar fimbrial compositions occur frequently among E. coli strains causing urinary tract infections (16, 19). Thus, it may be that the fimbrial-phase variation described here is a general property of pyelonephritogenic E.

5 VOL. 160, 1984 coli strains, which are often known to carry multiple fimbrial antigens (9, 16). P fimbriae are needed for bacterial attachment to uroepithelial cells (14) and invasion of kidneys (10). However, pyelonephritis is associated with an intense immune respose (23), and perhaps bacteria are able to rapidly change from one fimbrial type to another after reaching the kidneys. ACKNOWLEDGMENTS B.N. was a postdoctoral fellow supported by the Sigrid Jusdlius Foundation. This study was also supported by the Academy of Finland, the Yrjo Jahnsson Foundation, and the Magnus Ehrnrooth Foundation. We thank Tuula Taskinen for skilled technical assistance. LITERATURE CITED 1. Brinton, C. C., Jr The structure, function, synthesis and genetic control of bacterial pili and a molecular mechanism for DNA and RNA transport in gram-negative bacteria. Trans. N.Y. Acad. Sci. 27: Deneke, C. F., G. M. Thorne, and S. K. Gorbach Attachment pili from enterotoxigenic Escherichia coli pathogenic for humans. Infect. Immun. 26: Duguid, J. P., and R. R. Gillies Fimbriae and adhesive properties in dysentry bacilli. J. Pathol. Bacteriol. 74: Duguid, J. P., and D. C. Old Adhesive properties of Enterobacteriaceae, p In E. H. Beachey (ed.), Bacterial adherence. Receptors and recognition, series B, vol. 6. Chapman and Hall, London. 5. Eisenstein, B. I Phase variation of type-1 fimbriae in Escherichia coli is under transcriptional control. Science 214: Evans, D. G., R. P. Silver, D. J. Evans, Jr., D. G. Chase, and S. L. Gorbach Plasmid-controlled colonization factor associated with virulence in Escherichia coli enterotoxigenic for humans. Infect. Immun. 12: Gaastra, W., and F. K. de Graaf Host-specific fimbrial adhesins of noninvasive enterotoxigenic Escherichia coli strains. Microbiol. Rev. 46: Hudson, L., and F. C. Hay Practical immunology. Blackwell Scientific Publications, Oxford. 9. Jann, K., B. Jann, and G. Schmidt SDS polyacrylamide gel electrophoresis and serological analysis of pili from Escherichia coli of different pathogenic origin. FEMS Microbiol. Lett. 11: Killenius, G., S. B. Svenson, H. Hultberg, R. Mollby, J. Winberg, and J. A. Roberts P-fimbriae of pyelonephritogenic Escherichia coli: significance for reflux and renal scarring-a hypothesis. Infection 11: Kallenius, G., S. B. Svenson, R. Mollby, B. Cedergren, H. Hultberg, and J. Winberg Structure of carbohydrate part of receptor on human uroepithelial cells for pyelonephritogenic Escherichia coli. Lancet ii: Klemm, P., I. 0rskov, and F. 0rskov F7 and type-l-like fimbriae from three Escherichia coli strains isolated from urinary tract infections. Infect. Immun. 36: Korhonen, T. K Yeast cell agglutination by purified enterobacterial pili. FEMS Microbiol. Lett. 6: Korhonen, T. K., S. Eden, and C. Svanborg-Eden Binding of purified Escherichia coli pili to human urinary tract epithelial cells. FEMS Microbiol. Lett. 7: Korhonen, T. K., E.-L. Nurmiaho, H. Ranta, and C. Svanborg- Eden New method for isolation of immunologically pure pili from Escherichia coli. Infect. Immun. 27: FIMBRIAL PHASE VARIATION Korhonen, T. K., V. Vaisanen, H. Saxen, H. Hultberg, and S. B. Svenson P-antigen-recognizing fimbriae from human uropathogenic Escherichia coli strains. Infect. Immun. 37: a.Korhonen, T. K., V. Vaisanen-Rhen, M. Rhen, A. Pere, J. Parkkinen, and J. Finne Escherichia coli fimbriae recognizing sialyl galactosides. J. Bacteriol. 159: Leffler, H., and C. Svanborg-Eden Chemical identification of a glycosphingolipid receptor for Escherichia coli attaching to human urinary tract epithelial cells and haemagglutinating human erythrocytes. FEMS Microbiol. Lett. 8: Ofek, I., D. Mirelman, and N. Sharon Adherence of Escherichia coli to human mucosal cells mediated by mannose receptors. Nature (London) 265: rskov, I., and F. 0rskov Serology of Escherichia coli fimbriae. Prog. Allergy 33: rskov, I., F. 0rskov, and A. Birch-Andersen Comparison of Escherichia coli fimbrial antigen F7 with type 1 fimbriae. Infect. Immun. 27: rskov, I., F. 0rskov, A. Birch-Andersen, M. Kanamori, and C. Svanborg Eden , K, H and fimbrial antigens in Escherichia coli serotypes associated with pyelonephritis and cystitis. Scand. J. Infect. Dis. Suppl. 33: Parkkinen, J., J. Finne, M. Achtman, V. Vaisanen, and T. K. Korhonen Escherichia coli strains binding to neuraminyl a2-3 galactosides. Biochem. Biophys. Res. Commun. 111: Ratner, J. J., V. L. Thomas, B. A. Sanford, and M. Forland Bacteria-specific antibody in the urine of patients with acute pyelonephritis and cystitis. J. Infect. Dis. 143: Rhen, M., P. Klemm, E. Wahlstrom, S. B. Svenson, G. Kfilenius, and T. K. Korhonen P fimbriae of Escherichia coli: immuno- and protein chemical characterization of fimbriae from two pyelonephritogenic strains. FEMS Microbiol. Lett. 18: Rhen, M., J. Knowles, M. Penttila, M. Sarvas, and T. K. Korhonen P fimbriae of Escherichia coli: molecular cloning of DNA fragments containing the structural genes. FEMS Microbiol. Lett. 19: Rhen, M., P. H. Miakela, and T. K. Korhonen P fimbriae of Escherichia coli are subject to phase variation. FEMS Microbiol. Lett. 19: Rhen, M., E. Wahlstrom, and T. K. Korhonen P fimbriae of Escherichia coli: fractionation by immune precipitation. FEMS Microbiol. Lett. 18: SaHlt, I. E., and E. C. Gotschlich Hemagglutination by purified type 1 Escherichia coli pili. J. Exp. Med. 146: Simon, M., J. Zieg, M. Silverman, G. Mandel, and R. Doolittle Phase variation: evolution of a controlling element. Science 209: Swanson, J., and 0. Barrera Gonococcal pilus subunit size heterogeneity correlates with transitions in colony piliation phenotype, not with changes in colony opacity. J. Exp. Med. 158: Vaisanen, V., J. Elo, L. G. Tallgren, A. Siitonen, P. H. Maikelai, C. Svanborg-Eden, G. Kfillenius, S. B. Svenson, H. Hultberg, and T. K. Korhonen Mannose-resistant haemagglutination and P-antigen-recognition are characteristic of Escherichia coli causing primary pyelonephritis. Lancet ii: Vaisainen, V., T. K. Korhonen, M. Jokinen, C. G. Gahmberg, and C. Ehnhohn Blood group M specific haemagglutinin in pyelonephritogenic Escherichia coli. Lancet i: Vaisanen-Rhen, V., J. Elo, E. Vaisanen, A. Siitonen, I. 0rskov, F. 0rskov, S. B. Svenson, P. H. Makela, and T. K. Korhonen P-fimbriated clones among uropathogenic Escherichia coli strains. Infect. Immun. 43: