Escherichia coli to Isolated Rabbit Intestinal Brush Borders

Transcription

1 INFECTION AND IMMUNITY, Nov. 1979, p /79/11-736/8$2./ Vol. 26, No. 2 Quantitation of the Adherence of an Enteropathogenic Escherichia coli to Isolated Rabbit Intestinal Brush Borders CHRISTOPHER P. CHENEY,* EDGAR C. BOEDEKER, AND SAMUEL B. FORMAL Departments of Gastroenterology and Bacterial Diseases, Walter Reed Army Institute of Research, Washington, D.C. 212 Received for publication 17 August 1979 Two assays were developed to quantitate the adherence of an Escherichia coli strain (RDEC-1) known to colonize the mucosal surface of the small intestine of rabbits to brush borders isolated from rabbit intestinal epithelial cells. In the first assay, the mean adherence per rabbit brush border was determined by counting the number of organisms adhering to each of 4 brush borders under phase microscopy. The mean adherence of RDEC-1 (11.5 ±.7 per rabbit brush border) was significantly greater than adherence of two nonpathogenic strains: HS (2.7 ±.4 per rabbit brush border) and 64 (.8 ±.1 per rabbit brush border). A similar distinction between the adherence of RDEC-1 and the control (nonadherent) organisms could be made more rapidly by determining the percentage of the total number of brush borders which had 1 or more adherent organisms; this second assay was used to define the optimum conditions for adherence. Maximum adherence was seen within 15 min. Adherence was temperature dependent, with adherence after 1 min at 37 C being fourfold greater than that at 4 C. The ph optimum for adherence was between 6.5 and 7., and adherence was abolished below ph 5.. With the first, more sensitive assay, the effect of electrolytes and a number of hexoses and hexosamines on adherence was analyzed. RDEC-1 adherence was inhibited at high ionic strengths; however, adherence was not influenced at moderately high concentrations (2 mg/mil) by either D-mannose or L-fucose, in contrast to the case for other reported enteric pathogens. These two quantitative in vitro assays for adherence produce consistent results and have been used to partially characterize the adherence of RDEC-1 to rabbit brush borders. Adherence of enteropathogenic bacteria to the mucosal surface of intestinal epithelial cells is an important determinant of virulence for some enteric organisms. Adherence of pathogenic Escherichia coli appears to facilitate the ability of the organism to colonize the small bowel by allowing the pathogens to replicate while resisting the possible clearing action of intestinal peristalsis. Enteric organisms already included among this group of adherent pathogens are porcine K88 (19, 22, 28, 38) and bovine K99 (4) strains of E. coli, Vibrio cholerae (2, 21), and possibly the human enterotoxigenic E. coli, H147 (12, 34). Mucosal adherence is not an exclusive property of enteric infections, but has also been described with gonococcal (3) and E. coli (1, 23) infections of the urogenital tract and with streptococcal infections of the oral cavity (2) and heart valves (17). In most of these cases, bacterial adherence results from the mutual recognition and interaction of surface structures from both the bacteria and the host cells. Recent investigations have succeeded in identifying some of the adhesins on the surface of pathogenic organisms (12, 13, 18, 19, 28, 3, 31, 34) and to a lesser extent in identifying the constituents of the receptor on the host epithelial cells (1, 15, 29; C. P. Cheney, E. C. Boedeker, and S. B. Formal, Fed. Proc. 37:1727, 1978). To study the optimal conditions for adherence and to further probe the biochemical nature of the involved structures by studying the effect of inhibitors on this interaction, an in vitro adherence model is desirable. A suitable in vitro model should correlate with the known characteristics of the infective process in vivo and in particular should possess the same host and tissue specificity. For our studies in adherence, we chose the rabbit model of diarrhea reported by Cantey and Blake (5). They described a fatal diarrhea in young rabbits caused by a piliated enteropathogenic E. coli, RDEC-1 (15:NM). RDEC-1 appears to be an unusual organism among virulent enteric E. coli strains in that it is a noninvasive organism which does not synthesize either a 736 Downloaded from on December 29, 217 by guest

2 VOL. 26, 1979 classical heat-stable or heat-labile E. coli toxin. This organism was found to adhere and multiply on the surface of the ileum and the colon (36). Furthermore, this organism has also been shown to bind to and cause agglutination of ileal brush borders and microvillus membranes isolated from rabbit ileal epithelial cells but not to similar membranes prepared from rat, guinea pig, or human tissue (E. C. Boedeker, Y. K. Higgins, S. B. Formal, and A. Takeuchi, Gastroenterology 72:1154, 1977) and not to cells isolated from other rabbit tissue (bladder and erythrocytes). Thus, the RDEC-1 rabbit diarrheal model is both convenient and highly appropriate for the in vitro study of bacterial diarrhea associated with species- and tissue-specific mucosal adherence. The present studies were conducted to develop sensitive in vitro assays for the quantitation of bacterial adherence which in turn were utilized for determining the optimal conditions for adherence and for testing the influence of ionic strength on this process. In addition, we tested the influence on RDEC-1 adherence of several hexoses and hexosamines, including D- mannose and L-fucose, which have been shown to inhibit mucosal adherence in other bacterial systems. MATERIALS AND METHODS Brush border preparation. Male New Zealand White rabbits weighing between 1 and 3 kg were anesthetized by intravenous injection of pentobarbital (65 mg/kg). A 5-cm segment of distal ileum was excised and immediately rinsed with cold (4C) ethylenediaminetetraacetic acid (EDTA) buffer (5 mm EDTA, 5 mm Na2H2PO4, ph 7.5). All further procedures were done at 4C. The ileum was opened, the mucosal surface was scraped with glass slides, and the scrapings were placed in EDTA buffer. Rabbit brush borders (RBB) were prepared from these mucosal scrapings by the method of Donaldson et al. (8) by differential centrifugation. Purity of the brush border preparations was assessed by examination under phase microscopy (Fig. 1A) and by following the enrichment of maltase activity by the method of Dahlquist (7). Brush border preparations were subjected to repeated slow-speed centrifugation until nuclear contamination was essentially eliminated. Maltase activity of the final rabbit ileal brush borders was enriched between 9- and 13-fold compared with the homogenates. The final brush border fraction was resuspended in phosphatebuffered saline (PBS) (.145 M NaCl,.1 M Na2HPO4-NaH2PO4, ph 7.). Brush borders were prepared fresh daily for each assay. Protein was estimated by the method of Lowry et al. (26), and brush borders were diluted to a final concentration of 1 mg/ml with PBS. The number of RBB per milligram of brush border protein was determined by counting in a hemacytometer under phase microscopy and found to average 2.65 (±.25) x 17 RBB/mg. ADHERENCE TO RABBIT BRUSH BORDERS 737 Bacterial strains. Three strains of E. coli were examined for their ability to adhere to rabbit brush borders. The RDEC-1 strain (15:K?:NM) was isolated by J. R. Cantey (Veterans Administration Hospital, Charleston, S.C.) from laboratory rabbits that spontaneously developed diarrhea. Two strains of E. coli were used as controls: 64 (6:H5), isolated from the intestinal flora of a normal rabbit, and HS (O undetermined:h4), isolated from a normal human fecal sample and used earlier as a nonpathogenic control strain (25). Neither strain agglutinated RBB in vitro or caused diarrheal disease in rabbits in vivo (C. P. Cheney, E. C. Boedeker, M. K. Diadato, and S. B. Formal, Gastroenterology 76:1112, 1979). Bacterial cultures were grown overnight at 37 C in 1 ml of Penassay broth (Difco Laboratories, Detroit, Mich.) in screw-topped culture tubes. Bacteria from a single tube were pelleted at 2,4 x g at 23 C for 5 min, washed twice in PBS, and resuspended in 2 mil of PBS. Final concentrations of the bacteria were measured by plating serial 1-fold dilutions of the PBS suspensions on MacConkey agar plates and determining colony-forming units per milliliter after overnight incubation at 37 C. The final concentrations of bacteria used in this study were 2 x 19 to 4 x 19/ml. In vitro bacterial adherence assays. A 5-p1 amount of purified RBB (5 ytg of protein, 16 brush borders) was mixed with a 2-,ul suspension of bacteria (18 E. coli) and 3,l of PBS and allowed to incubate for up to 3 min with shaking at 23 C. The tubes were examined visually for agglutination, and phase microscopic examination at 6x revealed large aggregates of coagglutinated brush borders and RDEC-1 (Fig. 1B). Initially this reaction was evaluated by rating the aggregates from to +4, but such scoring was not sufficiently quantitative. However, it was found that mixing for 1 s on setting no. 5 with a Vortex-Genie (Scientific Industries Inc., Bohemia, N.Y.) would disperse the large aggregates into individual brush borders without eliminating the brush border-bacterial interactions, thus permitting quantitation of the number of organisms adhering to each brush border. In the first assay (assay I), bacterial adherence to dispersed brush borders was quantitated by observation under phase microscopy at a magnification of 6x and reported as the number of organisms adhering per RBB. To more rapidly monitor bacterial adherence, a second adherence assay (assay II) was established with the same incubation conditions. Adherence in this assay was expressed as the percentage of RBB possessing 1 or more adherent E. coli within a given area (.4 mm2) of a hemacytometer. An average of 2 RBB were counted in this assay. Effect of ionic strength. RBB (5 td), RDEC-1 (2 il), and portions (3 til) from stock solutions of varying concentrations of either KCl, NaCl, or KH2PO4 made up in PBS were mixed together and incubated for 15 min at 23 C. In assay I, RDEC-1 adherence to RBB was quantitated at a range of final concentrations of KCl (3 to 6 mm) and at a single final concentration of 6 mm for NaCl and KH2PO4 in at least five separate brush border preparations. The data were normalized to control values for RDEC- 1 adherence in PBS alone, and the results were expressed as percent of control adherence +1 standard Downloaded from on December 29, 217 by guest

3 738 CHENEY, BOEDEKER, AND FORMAL error. The reaction mixture was maintained at ph 7., and the osmolality of the solutions at their highest ionic strength was between 1,192 and 1,3 mosm as determined with a Digimatic-Osmometer (model 3D, Advanced Instruments Inc., Needham, Mass.). Carbohydrate inhibition. Portions (3 jil) from standard stock solutions (66 mg/mi of PBS) of the hexoses and hexosamines D-mannose, yeast mannan, a-methyl-mannoside, L-fucose, D-glucose, D-galactose, N-acetyl-D-galactosamine, or N-acetyl-D-glucosamine were added to 5 pl of RBB (5 jtg of protein) and 2 td of RDEC-1. The final carbohydrate concentration was 2 mg/ml or approximately 125 mm for each hexose. The mixture was incubated for 15 min, and the average number of adherent RDEC-1 on 4 RBB was determined (assay I). The data were normalized to control values for RDEC-1 adherence in the absence of carbohydrate, and the results were expressed as percentage of control adherence +1 standard error. RESULTS Microscopic appearance of E. coli interactions with isolated RBB. When the three test strains of E. coli, RDEC-1, 64, and HS, were incubated with isolated rabbit intestinal brush borders, visual agglutination was noted within 15 min with RDEC-1 but not with either of the control strains, HS or 64. At this stage, large aggregates of coagglutinated RDEC-1 and brush borders were observed under phase mi- INFECT. IMMUN. croscopy (Figure 1B), whereas 64 remained uniformly dispersed, and only occasional clumping of up to four brush borders with HS could be found. After blending in a Vortex mixer, the large aggregates of RDEC-1 and brush borders were dispersed, and individual brush borders with adherent RDEC-1 were observed (Fig. 1). The cytoplasmic side of the brush borders was essentially free of any adherent RDEC-1 organisms as far as could be detected by phase microscopy at this magnification. In each RDEC-1 incubation, a number of brush borders remained free of adherent RDEC-1 organisms. Strain 64 continued to show an absence of adherence to RBB even after mixing (Fig. id). Quantitation of in vitro RDEC-1 adherence to RBB. Assay I: mean number of adherent organisms per RBB. To demonstrate quantitatively that the adherence of RDEC-1 was significantly greater than that of the control E. coli strains, the number of adherent organisms on each of 4 brush borders after a 15-min incubation at 23C was counted in at least three separate studies (different brush border preparations) per strain. From these data, the mean number of adherent organisms per brush border standardd error) was calculated. Consistent with the microscopic appearance, Downloaded from c- AI N. F C X elli,~ il AP..e :,ff.:p4- :f:...!.." 4! 4 4g. on December 29, 217 by guest S 5 ~. ^ 4~~~ FIG. 1. Phase-contrast photomicrographs (x6) of the reactions of RBB with RDEC-1 or 64 E. coli after 15 min of incubation at 23C. (A) Control RBB preparation without added organisms. (B) Aggregates of coagglutinated RDEC-1 and RBB. (C) Vortex dispersion ofrdec-1 and RBB aggregates resulting in solitary RBB containing adherent RDEC-1. (D) Nonadherent control strain of E. coli (64) with RBB.

4 VOL. 26, 1979 mean RDEC-1 adherence was significantly greater than that of control strains with an average of 11.5 ±.7 RDEC-1 bound per brush border as compared with a mean 2.7 ±.4 HS per brush border (P <.1) and.8 ±.1 64 per brush border (P <.1). Complete enumeration of adherence in a single tube required approximately 5 to 7 min and therefore severely limited the number of reactions which could be studied in a single day. Assay II: percentage of RBB with 1 adherent E. coli. The enumeration data from the above studies were tabulated to display the numerical distribution of RBB which had the following ranges of adherent E. coli per RBB:, 1 to 4, 5 to 9, 1 to 14, 15 to 19, 2 to 24, and greater than 24 (Table 1). Analysis of the data in this table reveals that 55% (22/4) of the RBB possessed 1 or more adherent E. coli when incubated with RDEC-1 as compared with 5% (2/4) when incubated with HS or % (/4) when incubated with 64. Therefore, determining the percentage of RBB with 1 or more adherent organisms within a given area of the hemacytometer (.4 mm2) provided an assay which was sensitive enough to distinguish between strongly adherent and either weakly adherent or nonadherent E. coli strains and yet could be performed in approximately 2 min per sample or in one-third the time of the previous assay. Assay II was subsequently used as our rapid method of quantitation of RDEC-1 adherence. Characteristics of RDEC-1 adherence to rabbit leal brush borders. (i) Time course. Observations at 1, 5, and 15 min with assay II revealed that binding of RDEC-1 occurs rapidly at 23C and reaches a plateau (71.6 ± 3.5% of RBB with 21 adherent organisms) within 15 min (Fig. 2). A slight drop in adherence of 7% (not shown on the graph) was observed when ADHERENCE TO RABBIT BRUSH BORDERS 739 incubation was continued up to 3 min. The 64 strain failed to adhere throughout the entire time span studied. (ii) Effect of temperature. Incubations performed at 4, 23, and 37C demonstrated that RDEC-1 adherence is temperature dependent (Fig. 3). Maximal adherence at 37C (82.6 ± 2.%) was observed at 15 min. This degree of adherence was not achieved at 4C within 15 min; however, continued incubation to 6 min revealed that adherence at 4C had reached a level comparable to that seen at 37C after 15 min. The half time for maximal adherence was less than 1 min at 37C, 2 min at 23C, and 7 min at 4C. As shown in Fig. 3, binding within the first minute was fourfold greater at 37C than at 4C (P <.1). w ax co I a - O co ws:n a: c cr n c RDEC 64 T T l 1 I TIME (min) FIG. 2. Timeprofile of bacterial adherence to RBB at 23C. Adherence was monitored under phase microscopy by determining the percentage of brush borders containing 1 or more adherent bacteria. Each point represents the mean of four experiments done in duplicate ± standard error. TABLE 1. Distribution of rabbit ileal brush borders with varying numbers of adherent E. coli Adherent E. coli per No. of brush borders with adherent E. coli brush border RDEC-1a (7)b HS (4) 64 (3) 2.3 ±.9c (5.7)d 16.8 ± 3.4 (42.) 21. ±.6 (52.5) ± 1. (15.5) 15.5 ± 1.4 (38.8) 19. ±.6 (47.5) ± 1.3 (22.) 5.5 ± 2.5 (14) ± 1.2 (23.) 1.8 ± 1.4 (4) ± 1.1 (9.5).5 ±.5 (1) (9.5) < ±.6 (4.) Total brush borders counted a Strain tested. b Number of times strain was tested with separate brush border preparations. 'Mean number standardd error) of brush borders with adherent E. coli. d Percentage of total brush borders counted with adherent E. coli. Downloaded from on December 29, 217 by guest

5 74 CHENEY, BOEDEKER, AND FORMAL LJ 1- a AI\ ' C -a, 6 w) CID/ - T 4 Iw (n a- co co a~~~~~~~~~~~~~~~ 2- co4-1 UA TIME (MIN) FIG. 3. Effect of temperature on the rate of RDEC- I adherence to RBB. Each point is the mean of four assays performed in duplicate ± standard error. The assay conditions were the same as those used in Fig. 2. (iii) Effect of changes in ph. The ability of a change in ph to reverse RDEC-1 adherence was assessed. Samples of brush borders and RDEC-1 were incubated at 23 C for 15 min in PBS at ph 7., and the percentage of brush borders with 1 or more RDEC-1 was determined. At this point, all of the samples demonstrated comparable adherence values. Each sample was then centrifuged at 75 x g for 1 min at 4 C, the supernatants were aspirated, and the pellets containing RBB with adherent RDEC-1 were resuspended in buffers of varying test phs. The samples were then incubated at the test ph for an additional 15 min. RDEC-1 adherence, determined by assay II, was plotted against ph (Fig. 4). No change in RDEC-1 adherence occurred between ph 6.5 and 7.; however, a dramatic reversal of adherence occurred at test phs below ph 6.. At ph 4.3, RDEC-1 adherence was abolished. Changes in adherence at low phs were associated with changes in the appearance of brush borders which became contracted and opaque. In addition, at very low phs (below 3.) RDEC-1 was observed to precipitate out of suspension. Since the above experiment actually represents a desorption or reversal of RDEC- 1 adherence, a second experiment was performed to investigate whether a similar effect on adherence occurred when the ph of the reaction mixture was altered at zero time. The profile of the ph curve for RDEC-1 adherence was identical to Fig. 4, again displaying (not shown) an optimum at ph 7.. (iv) Effect of salt concentration. The influence of increasing salt concentration on RDEC-1 adherence to RBB was assessed. Depression of RDEC-1 adherence was observed with increasing concentrations of KCl in the reaction mixture (Fig. 5). At a KCl concentration of.6 M, yielding a total salt concentration of.75 M because of the contribution of PBS, the adherence of RDEC to brush borders was significantly decreased to 34.7 ± 7.9% below control values (P <.1). Experiments to determine whether the inhibition was attributable to either the K+ or Cl- ions were performed by replacing either of these ions with others. In doing so, a similar degree of adherence inhibition was seen at a salt concentration of.75 M when NaCl replaced KCl (35.1 ± 6.3%), whereas somewhat greater suppression below control levels was detected when KH2PO4 replaced KCl (48.6 ± 5.4%). (v) Carbohydrate inhibition of adherence. The hexoses D-glucose, D-galactose, D- mannose, and L-fucose plus the hexosamines N- acetyl-d-glucosamine and N-acetyl-D-galactosamine, which are frequent constituents of the glycoproteins and glycolipids found on mammalian plasma membranes, were added to the incubation medium to yield a final concentration of 2 mg/ml (approximately 125 mm for the hexoses), and RDEC-1 adherence was quantitated (Table 2). Known inhibitors of K-12 E. coli adherence, D-mannose, yeast mannan, and a-methyl-mannoside (1, 33) failed to inhibit RDEC-1 adherence. Similarly, L-fucose, a known inhibitor of V. cholerae adherence to RBB (2), had no significant effect on RDEC-1-8 (n C- L 1-6- cr 6 M a: o n 4- m QX W z 2- le , P H INFECT. IMMUN FIG. 4. Effect ofphon RDEC-I adherence to RBB. Assay tubes containing RDEC-1 and RBB in PBS (ph 7.) were incubated for 15 min at 23'C. The reactants were sedimented by centrifugation and resuspended at test phs in the following buffers: citric acid-na2hpo4 for the ph range 3 to 7; NaH2PO4- Na2HPO4 for the ph range 6 to 8; and glycine-naoh for the ph range 9 to 1. Incubation was continued for 15 min at test ph, and the adherence was determined. Downloaded from on December 29, 217 by guest

6 VOL. 26, %1 Z 8%- I cr 4 6%- Z 4%- U.) 2%- I KCI CONCENTRATION (MOLAR) FIG. 5. Effect of increasing KCI concentration in the incubation medium on RDEC-1 adherence to RBB. Adherence is expressed as the percentage of control adherence (absence of any KCl). The final salt concentration in the incubation media is the sum of the number of millimoles of KCI added plus 155 mmol attributable to PBS. Each point represents mean of five assays performed in duplicate (± standard error). adherence. None of the other carbohydrates significantly influenced RDEC-1 adherence. DISCUSSION Since a major goal in studying bacterial adherence to mucosal surfaces is to identify the determinants involved with the hope of finding inhibitors of adherence, quantitative assays for in vitro adherence are needed. Such assays can be used to establish conditions for optimum adherence and to test whether isolated bacterial adhesins or brush border receptors, their analogs, or their constituents can inhibit the reaction. We have developed two in vitro assays for RDEC-1 adherence, the first, a sensitive but time-consuming assay in which the number of adhering organisms per brush border was enumerated, and the second, a more rapid assay which provides the ability to test numerous experimental conditions on any given day. These assays were used to demonstrate the adherence of RDEC-1 to brush borders from rabbit ileal cells. Frequently, as many as 3 organisms were observed adhering to the mucosal surface of individual brush borders, indicating the presence of RDEC-1 receptors on the brush borders. However, the actual number of RDEC-1 receptors may be far larger than the number of adhering organisms since the steric hindrance of the organisms may permit only partial saturation of the total number of receptors that might be present on the brush border surface. Of interest in this context, calculations using the average dimensions of an RDEC-1 organism (1.5 by 4.,tm) and the approximate ADHERENCE TO RABBIT BRUSH BORDERS 741 area at the apical end of an epithelial cell (166,tm2) indicated that the number of RDEC-1 needed to form a continuous monolayer of contigous bacteria lying lengthwise on the surface of each brush border would be approximately 28 organisms. In contrast to RDEC-1, both of the nonpathogenic control E. coli strains, 64 and HS, failed to adhere or did so only to a minor extent. This finding could indicate either that there are significantly fewer receptors for these nonpathogenic bacteria on the brush borders or that the receptors present have a markedly lower binding affinity for the nonpathogens than for RDEC-1. Using the adherence assays, experiments performed to determine the optimal conditions for RDEC-1 adherence showed that the RDEC-1 bound to isolated ileal brush borders rapidly (within 15 min) at both 23 and 37C. However, adherence at 4C was fourfold lower than that at 37CC at 1 min. This temperature dependence of binding suggests a possible enzymatic step in the adherence phenomenon. Alternatively, the lower temperature may have decreased the fluidity of host receptor sites in the RBB membranes, thereby delaying the rate of formation of a critical receptor concentration which may be needed for binding of RDEC-1 to the brush border (3, 14). The optimal ph for RDEC-1 adherence occurred between ph 6.5 and 7., with inhibition of RDEC-1 adherence observed below ph 6.. At ph of 4.3, RDEC-1 adherence was completely abolished. This inhibition of adherence may have been attributable to inhibition of enzymatic activity or conformational changes in the adhesins on the surface of the bacteria or in receptors on the brush borders. One possibility is a physical change at the isoelectric point (pi) of the structure involved in adherence. Recently, surface antigens on other enteric E. coli strains TABLE 2. Effect of carbohydrates on RDEC-1 adherence to isolated RBB % of control adherence' Carbohydrates tested (+SE) D-Mannose 92.8 ± 8.6 Yeast mannan 88.9 ± 8. a-methyl-mannoside 13.7 ± 6.2 L-Fucose D-Glucose D-Galactose N-Acetyl-D-glucosamine ± 1.1 N-Acetyl-D-galactosa ± 7.5 mine athe average number of RDEC-1 adhering per RBB in the presence of the test carbohydrate divided by the average number of RDEC-1 adhering in the absence of carbohydrate x 1. SE, Standard error. Downloaded from on December 29, 217 by guest

7 742 CHENEY, BOEDEKER, AND FORMAL have been correlated with their in vitro adherence. In particular, the K88 and K99 antigens of porcine and bovine enteric E. coli are thought to be the bacterial adhesins. These antigens have been isolated, and their pis were found to be between 4. and 5.3 (27, 35). Although the adhesin on RDEC-1 has not been identified, it is possible that the inhibition of RDEC-1 adherence below ph 6. may be associated with a change in the biological activity of the adhesins in that ph range. Depression of adherence observed in the presence of increasing salt concentration suggests that an initial step in RDEC-1 adherence might involve an electrostatic attraction of oppositely charged receptors on the host and bacteria. The presence of a high salt concentration may have neutralized the electrostatic force which normally governs the attraction of bacterial adhesins to host receptors. In each in vitro RDEC-1 adherence assay, the number of RDEC-1 adhering to the brush borders varied from to 3. The observation that some brush borders have few or no adhering organisms might relate to Cantey and Blake's original finding (5) that RDEC-1 preferentially adhered to villus tip cells in the ileum. The reason for such preferential adherence remains to be determined; however, a possible explanation may lie in the biochemical differences of the mucosal surfaces of epithelial cells along the intestinal villi. It is well known that the carbohydrate composition of the mucosal surface changes because of glycosylation of mucosal glycoproteins during the migration of epithelial cells from the crypts to the villus tips (24, 37). This difference in carbohydrate composition is thought to be responsible for demonstrated differences in lectin binding to cells from villus tips and crypts (11). Recent studies on adherence of various bacterial strains to host epithelial cells have focused on the sensitivity of bacterial adherence to various monosaccharides. The hypothesis that adherence results from the ability of the bacterial adhesins to recognize and bind to carbohydrate receptors on the cell surface of the host has been INFECT. IMMUN. suggested. Initial evidence for this came from the work of Collier and DeMiranda (6), who demonstrated that D-mannose inhibited the ability of some enteric pathogens to adhere to mammalian cells. Later, Duguid and Gillies (9) were able to inhibit both agglutination of erythrocytes and attachment to intestinal epithelial cells by piliated bacteria with D-mannose, a- methyl-mannoside, and yeast mannan. Subsequently, the adherence of Salmonella typhi and a piliated K-12 strain of E. coli to host phagocytes was also shown to be markedly inhibited in the presence of these carbohydrates (1). Furthermore, binding of purified pili from a K-12 strain of E. coli was found to possess the identical sensitivity to D-mannose and its analogs as did intact organisms (32, 33). Adherence of V. cholerae to rabbit intestinal mucosa shows greater sensitivity to L-fucose than D-mannose (2). The present study shows that RDEC-1 adherence to ileal RBB is different from the above bacterial systems in that it is resistant to D-mannose and other monosaccharides. Other E. coli strains whose adherence to host cells is known to be mannose resistant include the human diarrheal pathogens possessing the CFA factor (colonization factor antigen) found on H147 (D. G. Evans, D. J. Evans, and T. R. Deetz, Abstr. Annu. Meet. Am. Soc. Microbiol. 1978, B8, p. 15), the K88 porcine (15) and K99 bovine enteric E. coli (4), and E. coli pathogens isolated from patients with urinary tract infections (1). Although the high concentrations of hexoses used in the reaction mixtures suggest that a lectin-like interaction is not involved in RDEC-1 adherence, the possibility remains that oligosaccharides with structures more complex than those of the hexoses tested serve as the bacterial receptors on host epithelial cells. In summary, RDEC-1 is a new example of an enteropathogenic organism whose virulence is associated with adherence to the intestinal mucosa both in vivo and in vitro. The quantitative and rapid in vitro adherence assays reported in this paper have been utilized to define the optimum conditions for adherence, to demonstrate the sensitivity of the adherence interaction to salt concentration, and to show its resistance to D-mannose and other monosaccharides. We anticipate that these adherence assays will be used in future studies to investigate critical determinants of adherence localized on bacterial and host cells. ACKNOWLEDGMENTS This project was supported by the U.S. Army Medical Research and Development command. This excellent technical assistance of Melissa K. Diodato and the secretarial expertise of Barbara A. Mozie are gratefully acknowledged. LITERATURE CITED 1. Bar-Shavit, Z., I. Ofek, R. Goldman, D. Mirelman, and N. Sharon Mannose residues on phagocytes as receptors for the attachment of Escherichia coli and Salmonella typhi. Biochem. Biophys. Res. Commun. 78: Bartelt, M. A., and J. L. Duncan Adherence of group A streptococci to human epithelial cells. Infect. Immun. 2: Bell, G. I Models for the specific adhesion of cells to cells. Science 2: Downloaded from on December 29, 217 by guest

8 VOL. 26, 1979 ADHERENCE TO RABBIT BRUSH BORDERS Burrows, M. R., R. Sellwood, and R. A. Gibbons Haemagglutinating and adhesive properties associated with the K99 antigen of bovine strains of Escherichia coli. J. Gen. Microbiol. 96: Cantey, J. R., and R. K. Blake Diarrhea due to Escherichia coli in the rabbit. A novel mechanism. J. Infect. Dis. 135: Collier, W. A., and J. C. DeMiranda Bacterienhaemagglutination. Antonie van Leeuwenhoek J. Microbiol. Serol. 21: Dahlqvist, A Method for assay of intestinal disaccharidases. Anal. Biochem. 7: Donaldson, R. M., L. L. MacKenzie, and J. S. Trier Intrinsic factor-mediated attachment of vitamin B,2 to brush borders and microvillous membranes of hamster intestine. J. Clin. Invest. 46: Duguid, J. P., and R. R. Gillies Fimbriae and adhesive properties in dysentery bacilli. J. Pathol. 74: Eden, C. S., and H. A. Hansson Escherichia coli pili as possible mediators of attachment to human urinary tract epithelial cells. Infect. Immun. 21: Etzler, M. E., and M. C. Branstrator Differential localization of cell surface and secretary components in rat intestinal epithelium by use of lectins. J. Cell. Biol. 62: Evans, D. G., D. J. Evans, Jr., W. S. Tjoa, and H. L. DuPont Detection and characterization of colonization factor of enterotoxigenic Escherichia coli isolated from adults with diarrhea. Infect. Immun. 19: Evans, D. G., R. P. Silver, D. J. Evans, D. G. Chase, and S. L. Gorbach Plasmid-controlled colonization factor associated with virulence in Escherichia coli enterotoxigenic for humans. Infect. Immun. 12: Frye, L. D., and M. Edidin The rapid intermixing of cell surface antigens after formation of mouse-human heterokaryons. J. Cell Sci. 1: Gibbons, R. A., G. W. Jones, and R. Sellwood An attempt to identify the intestinal receptor for the K88 adhesin by means of a haemagglutination inhibition test using glycoproteins and fractions from sow colostrum. J. Gen. Microbiol. 86: Gibbons, R. J., D. M. Spinell, and E. Skobe Selective adherence as a determinant of the host tropisms of certain indigenous and pathogenic bacteria. Infect. Immun. 13: Gould, K., C. H. Ramirez-Ronda, R. K. Holmes, and J. P. Sanford Adherence of bacteria to heart valves in vitro. J. Clin. Invest. 56: Isaacson, R. E K99 surface antigen of Escherichia coli: purification and partial characterization. Infect. Immun. 15: Isaacson, R. E., P. C. Fusco, C. C. Brinton, and H. W. Moon In vitro adhesion of Escherichia coli to porcine small intestinal epithelial cells: pili as adhesive factors. Infect. Immun. 21: Jones, G. W., G. D. Abrams, and R. Freter Adhesive properties of Vibrio cholerae: adhesion to isolated rabbit brush border membranes and hemagglutinating activity. Infect. Immun. 14: Jones, G. W., and R. Freter Adhesive properties of Vibrio cholerae: nature of the interaction with isolated rabbit brush border membranes and human erythrocytes. Infect. Immun. 14: Jones, G. W., and J. M. Rutter Role of the K88 antigen in the pathogenesis of neonatal diarrhea caused by Escherichia coli in piglets. Infect. Immun. 6: Kallenius, G., and J. Winberg Bacterial adherence to periurethral epithelial cells in girls prone to urinary-tract infections. Lancet ii: Kim, Y. S., J. Perdomo, and J. Nordberg Glycoprotein biosynthesis in small intestinal mucosa. I. A study of glycosyltransferases in microsomal subfractions. J. Biol. Chem. 246: Levine, M. M., D. R. Nalin, R. B. Hornick, E. J. Bergquist, D. H. Waterman, C. R. Young, and S. Sotman Escherichia coli strains that caused diarrhea but do not produce heat-labile or heat-stable enterotoxins and are non-invasive. Lancet ii: Lowry,. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall Protein measurement with the Folin phenol reagent. J. Biol. Chem. 193: Morris, J. A., A. E. Stevens, and W. J. Sojka Preliminary characterization of cell free K99 antigen isolated from Escherichia coli B41. J. Gen. Microbiol. 99: Nagy, B., H. W. Moon, and R. E. Isaacson Colonization of porcine intestine by enterotoxigenic Escherichia coli: selection of piliated forms in vivo, adhesion of piliated forms to epithelial cells in vitro, and incidence of a pilus antigen among porcine enteropathogenic E. coli. Infect. Immun. 16: Ofek, I., D. Mirelman, and N. Sharon Adherence of Escherichia coli to human mucosal cells mediated by mannose receptors. Nature (London) 265: Pearce, W. A., and T. M. Buchanan Attachment role of gonococcal pili. J. Clin. Invest. 61: Punsalang, A. P., and W. D. Sawyer Role of pili in the virulence of Neisseria gonorrhoeae. Infect. Immun. 8: Salit, L. E., and E. C. Gotschlich Hemagglutination by purified type I Escherichia coli pili. J. Exp. Med. 146: Salit, I. E., and E. C. Gotschlich Type I Escherichia coli pili: characterization of binding to monkey kidney cells. J. Exp. Med. 146: Satferwhite, T. K., H. L. Dupont, D. G. Evans, and D. J. Evans Role of Escherichia coli colonization factor antigen in acute diarrhea. Lancet ii: Stirm, S., F. Orskov, I. rskov and B. Mansa. Episome-carried surface antigen K88 of Escherichia coli. II. Isolation and chemical analysis. J. Bacteriol. 93: Takeuchi, A., L. R. Inman, P. D. O'Hanley, J. R. Cantey, and W. B. Lushbaugh Scanning and transmission electron microscopic study of Escherichia coli 15 (RDEC-1) enteric infection in rabbits. Infect. Immun. 19: Weiser, M. M Intestinal epithelial cell surface membrane glycoprotein synthesis. II. Glycosyltransferases and endogenous acceptors of the undifferentiated cell surface membrane. J. Biol. Chem. 284: Wilson, M. R., and A. W. Hohmann Immunity to Escherichia coli in pigs: adhesion of enteropathogenic Escherichia coli to isolated intestinal epithelial cells. Infect. Immun. 1: Downloaded from on December 29, 217 by guest