0022-534 7/90/1432-0386$02.00/0 Vol. 143, February Printed in U.S.A.
THE JOURNAL OF UROLOGY Copyright© 1990 by AMERICAN UROLOGICAL ASSOCIATION, INC.
OUTER MEMBRANE PROTEINS OF E.COLI IN THE HOST-PATHOGEN INTERACTION IN URINARY TRACT INFECTION JAIME A. ROBLEDO, ALIX SERRANO
AND
GERALD J. DOMINGUE*
From the Department of Urology and Department of Microbiology and Immunology, Tulane University School of Medicine, New Orleans, Louisiana
ABSTRACT
Outer membrane protein patterns (Omp) of Escherichia coli obtained directly from the urine of bacteriuric patients without passage on artificial culture media (ACM) were studied by polyacrylamide gel electrophoresis (SDS-PAGE) in an effort to determine whether in vivo conditions of growth affected the expression of these bacterial surface structures. Seventeen strains studied showed two distinct Omp patterns: one protein band appeared at the level of porin proteins (40 kDa) in both patterns, but Omp A protein was at the level of 36 kDa in the first pattern and a new protein was observed at 21.5 kDa in the second pattern suggesting that it is a fragment of Omp A. High molecular weight proteins were also observed in most of the strains and this finding was related to lack of free iron when the same strains were grown under iron restricted conditions in vitro. The same strains grown in pooled urine from normal females showed the first pattern mentioned above. Comparative growth on ACM of urinary strains and E. coli strains isolated from blood, feces and wounds showed an increase in the number of porins expressed (from 1 to 2 or 3, with some variability observed between strains). Differences in osmolality between pooled urine and ACM used, plus in vitro studies varying the osmolality of culture media, showed that osmolality accounted for differences in the number of porins expressed: porin expression decreased in urine and ACM of high osmolality, suggesting that the same phenomena occurred in vivo. It is concluded that host factors including low availability of iron and high osmolality present in the urinary tract influence the expression of several E. coli surface proteins. These proteins may relate to the ability of E. coli to colonize and invade the urinary tract by regulating the physiologic and/or metabolic state of the bacterial cell favoring survival of the organism in a hostile environment. Specific immune responses directed against porins could influence the outcome of this host-parasite interaction. (J. Ural., 143: 386-391, 1990) E. coli is the most frequent cause of urinary tract infection in humans. 1 Several bacterial virulence factors have been implicated in the host-pathogen interation, yet despite extensive research, no single virulence factor has completely accounted for the urovirulence of E. coli in pyelonephritis. Most of the recent efforts have been devoted to the role of adhesins (fimbriae) of E. coli recognizing specific receptors on the mucosal epithelium. Strains having P fimbriae adhesin have been linked to major inflammatory changes in the urinary tract, 2 and are more often isolated from patients with pyelonephritis, 3·4 suggesting that P fimbriae are important in the pathogenesis of the disease. Additionally, bacterial capsules and hemolysin have been reported to be significant as virulence factors. 5 ' 6 Recent data, however, from our laboratory, utilizing an experimental murine model of pyelonephritis indicated that wild type E. coli strains devoid of fimbriae, hemolysin, acidic capsules and sensitive to human serumcidal activity were capable of causing incipient and acute pyelonephritis. 7 Furthermore, even among identical serotypes and biotypes, the presence of fimbriae did not appear to be critical factors in urovirulence, nor did the presence of several other positive characteristics (hemolysin, K capsule, flagella, serum resistance) in a given strain enhance uorpathogenicity. These phenotypic characteristics may simply represent associated or serologic markers with the host serving as the dominant determinant of bacterial virulence and susceptibility to urinary infection. Accepted for publication September 14, 1989.
* Requests for reprints: Dept. of Urology, Tulane University School of Medicine, 1430 Tulane Ave., New Orleans LA 70112. Presented at the 88th Annual Meeting of the American Society for Microbiology, Abstract Number B205, May 8, 1988, Miami, Florida. 386
Bacterial virulence, therefore, should be examined in the context of the relationship of both microorganism and host: bacterial survival in a hostile environment, and host factors which intervene to overcome initiation of bacterial colonization and further infection of the urinary tract. This complex relationship must involve not only the above mentioned virulence factors but the entire bacterial cell structure. In fact, bacteria growing in vivo may be different metabolically and structurally from organisms traditionally studied by in vitro conditions on artificial culture medium. 8 Urine is a complex body fluid containing a variety of excreted products. 9 It does support bacterial growth although several urine factors have been implicated as inhibitory. For example, the lack of free iron has been related to the expression of iron regulated membrane proteins (IRMP) on bacteria isolated from the human urinary tract, 10• 11 indicating a restrictive condition for bacterial growth and a consequent adaptational change. Although other factors (pH, concentration of urea in urine and urine osmolality) have been demonstrated as inhibitory for bacterial growth under certain conditions, 12• 13 the nature of the structural changes which bacteria undergo while growing in the urinary tract is mostly unknown. Major outer membrane proteins are part of the external surface of the bacterial cell. Porins belong to a particular group of outer membrane proteins that have been termed peptidoglycan-associated general diffusion pore proteins, which besides serving as receptors for certain bacteriophages also act as nonspecific pores that allow the entrance of molecules smaller than 600 M.W. into the bacterial cell. 14 Since the expression in vitro of this group of proteins is related to the composition of the culture medium or environmental conditions, 15- 17 it appears
E. COLI MEMBRANE PRODUCTS
warranted to study the nature of these proteins expressed by urinary pathogens in vivo in an effort to define their role in the host-pathogen interaction. Specifically, the aims of the present study were to determine the expression of E. coli outer membrane proteins (Omp) in vivo when bacteria were obtained directly from human bacteriuric urines (without passage on artificial culture media) and to compare this expression to that observed on the same strains grown on artificial culture media and in pools of human urine in vitro. In order to determine whether there are differences between bacteriuric isolates of E. coli and isolates from other sources, several E. coli strains from feces, blood and wounds were compared. MATERIALS AND METHODS
E. coli strains. Seventeen E. coli strains isolated from bacteriuric urines and with counts greater than 105 CFU /ml. were collected from the hospital/clinic of Tulane University Medical Center. Positive urines were kept under refrigeration at 4C until laboratory results indicated that E. coli was the only isolate and the quantitative count was confirmed. Additionally, seven E. coli strains isolated from blood, eleven from feces of normal individuals and two from wounds were obtained from the hospital/clinic laboratory. The strains isolated from patients with bacteriuria were also characterized according to 0 serotype, mannose resistant agglutination, mannose sensitive agglutination and hemolysin production according to procedures previously described. 7 Strains isolated from urine as well as strains from blood, feces and wounds were inoculated into trypticase soy broth plus 5% glycerol, grown overnight and kept at -70C until used. Culture media and growth of the bacterial strains. Luria broth and a pool of urine from normal females (morning urine, filtered through a 0.22 micron filter) were utilized for growing the E. coli strains in vitro. Strains isolated from urine were studied without passage on artificial culture media. Osmolalities were not performed on these individual urines. Also, these urine strains were grown overnight as static cultures in Luria broth, four hour subcultures served as inocula (108 CFU/ml. in 10 ml. as final concentration) for fresh Luria broth and urine pools. Broth and urine cultures were subsequently incubated for four hours at 35C. Strains from blood, feces and wounds were isolated in pure culture on routine artificial culture media (for no more than one to two passages) prior to inoculating in fresh Luria broth and urine pools. Extraction, electrophoresis and identification of outer membrane proteins. Omp extraction methods were based on procedures described by Achtman et al. 18 Strains isolated directly from urine which had E. coli as the only bacterial isolate and with counts> 10 5 CFU/ml. were subjected to differential centrifugation (1,000 RPM for 10 minutes, followed by 3,500 RPM for 15 minutes) in order to obtain a bacterial pellet free of epithelial cells and leukocytes. This bacterial pellet plus bacterial pellets obtained from strains grown in Luria broth and pooled urine were suspended in 50 mM TRIS-HCL pH 8.0 to a final concentration of 108 CFU/ml., and disrupted by sonication (40 seconds, two to three times, at half intensity in the cold). Unbroken cells and bacterial debris were removed by centrifugation (1,800 g, 4C for 20 minutes) and bacterial membranes were pelleted from the supernate (45,650 g, 4C for 60 minutes). The pelleted membranes were resuspended in 150 microliters of distilled H 2 0. Eight volumes of N-Laurylsarcosine (1.67% N-Laurylsarcosine, Sigma Chemical Co., St. Louis, MO, 11.1 mM TRIS-HCL pH 7.6) were added to 50 and 100 microliters aliquots of the membrane suspension to solubilize the inner membrane proteins; outer membrane proteins were precipitated by centrifugation (in eppendorftubes using adaptors in a Beckman rotor Model JA-20, at 20,000 rpm, 20C for 90 minutes). The 50 tnicroliter aliquots were assayed for protein concentra-
387
tion by using a modification of the Lowry method described by Markwell et al., 19 the remaining aliquots were resuspended in SDS-PAGE sample buffer (0.0625 M TRIS-HCL pH 6.8, 10% glycerol, 2% SDS, 5% 2-mercaptoethanol, 0.001% bromophenol blue), and adjusted to a protein concentration of 0.5 to one mg./ml. The samples were heated to 100 C for five minutes before being applied to the gel, each well contained 30 micrograms of protein. Discontinuous SDS-PAGE was performed by a modification of the Laemmli procedure, 18 in which four M urea was added to the separation gel and 0.15% of SDS was used in the electrode buffer. The gels were stained with coomassie blue R-250 (BioRad Laboratories, Richmond, CA) for 30 minutes, followed by overnight destaining in 40% methanol-10% acetic acid, and further soaking in 50% methanol, 5% glycerol before storage. Major outer membrane proteins were defined as proteins with apparent molecular weights between 30 kDa nd 40 kDa. 18• 20 Omp A was identified by its heat modifiability, appearing at approximately 35 to 36 kDa after exposure to SDS at lOOC and showed an increased mobility to approximately 31 kDa when the samples were heated to 37C in the presence of SDS. 18020 Porins were identified as the protein bands that appeared at approximately 40 kDa when the samples were heated to lOOC in the presence of SDS and were not solubilized when heated to 37C; 18• 20 further characterization of individual porins were not made. Iron regulated membrane proteins (IRMP) were identified as the proteins that appeared in the range between 66.2 kDa to 92.5 kDa when the individual strains were grown in TSB (Trypticase soy broth; BBL, Becton, Dickinson and Co.) supplemented with ovotransferrin (0.5 mg./ml.; type 1, Sigma Chemical Co., St. Louis, MO 0.07 M NaHC0 3 ), and IRMP were not expressed when grown in the same medium supplemented with iron (0.02 mM FeS0 4 ). 10• 11 • 21 Osmolality studies. To determine whether osmolality affected the expression of the major outer membrane proteins in vivo several E. coli strains isolated from human urine were grown in Luria broth and pooled urine at different osmolalities with a range between 100 to 1000 mOsM/kg. Lyophilized urine was reconstituted and diluted with various volumes of distilled water until the desired osmolality was obtained. Luria broth was likewise diluted. Increasing the osmolality of Luria broth was accomplished by adding sodium chloride to the medium. Omp preparations and SDS-PAGE were performed as previously described as well as the growth conditions. RESULTS
Strain characteristics. Results summarized in table 1 indicate that only six of the strains were typeable using O antisera commonly employed for classifying E. coli strains causing urinary tract infections. Nine exhibited mannose resistant agglutination of human erythrocytes (P fimbriae), fourteen strains showed mannose sensitive agglutination of guinea pig erythrocytes (type 1 fimbriae, additionally confirmed by mannose sensitive agglutination of Saccharomyces cerevisiae cells), and seven strains produced hemolysin in 5% sheep blood agar plates. One strain, serotype 018, and three non typeable strains were positive for all characteristics tested. Omp patterns in SDS-PAGE. Strains exhibited individual variabilities in their major outer membrane protein patterns by SDS-PAGE. However, all strains (including the bacteriuric isolates and strains isolated from other sources) could be grouped according to common similarities in their expression of major outer membrane proteins. The seventeen strains isolated from urine and analyzed by SDS-PAGE, without passage on artificial culture media, showed two distinct patterns at the level of major outer mem brane proteins. The first one was found in a group of seven strains which showed one band at approximately 40 kDa corresponding to porins, and one band at approximately 36 kDa corresponding to Omp A (fig. 1). The second pattern was
388
ROBLEDO, SERRANO AND DOMINGUE TABLE
1. Characteristics of 17 E. Coli strains isolated from
human urine Strain
0 Serotype
U-2333 U-2336 U-2312 U-2309 U-2263 U-2308 U-1773 U-2882 U-2928 U-2929 U-3029 U-3040 U-3060 U-3061 U-ST U-3162 U-3271
NT** NT NT NT NT NT ND*** 018 NT 075 02 01 NT NT 06 ND 018
MR*
MS*
Agglut.
Agglut.
+
+
+
+ + + + + +
ND
ND
ND
+
+ + + +
+ + + + + + +
ND ND
ND ND
+
+
+ +
Hemolysin
+
KDa 92. 5
IRMP
4S. 0+ + + +
* Mannose resistant agglutination of human erythrocytes and mannose sensitive agglutination of guinea pig erythrocytes indicative of P fimbriae and type 1 fimbriae respectively. ** Non-typeable utilizing 026, 055, 0111, 0127, 086, 0119, 04, 011, 0124, 0125, 0126, 044, 0112, 028, 020, 018, 08, 09, 015, 022, 075, 083, 02, 01, 06 antisera *** Not done.
observed in a group of seven strains that showed one band at approximately 40 kDa at the level of porins and a second band at approximately 21.5 kDa that may correspond to either Omp A or one of its fragments, since this is the location for the tryptic fragment of Omp A when the protein is exposed to trypsin (fig. 2). In three strains, the patterns observed were unclear (with faint bands) and were not assigned to any of the groups. The initial concentration of bacteria for those samples was low, yielding less protein in the final sample. The procedure could not be repeated since the bacteriuric urine was no longer available. Figures 1 and 2 also show the patterns observed when the strains from both groups were grown in vivo; in pooled urine all showed a pattern similar to the first one mentioned above, one band corresponding to porins and one band corresponding to Omp A. A different pattern also common to all strains however, was observed when the bacteriuric strains were grown in artificial culture medium (Luria broth); at approximately 40 kDa, two to three bands (some variability in the number observed between strains) were demonstrated with characteristics compatible with porins, and one band was observed at approximately 36 kDa, the level of Omp A. The strains isolated from blood, feces and wounds, despite showing individual variability in the expression of outer membrane proteins (as observed with the urinary strains) gave similar results to the urinary isolates when grown in vitro (Luria broth and pooled urine). In artificial culture medium, these strains expressed two to three bands at the porin levels (approximately 40 kDa) and one band at Omp A level (approximately 36 kDa), and in pooled urine, the band at the Omp A level and only one band at the porin level (fig. 3). It was observed that the osmolality of standard Luria broth and the pooled urine samples used to grow all strains was markedly different (400 mOsm/kg. and 800 mOsm/kg. respectively). Experiments varying the osmolality of both the Luria broth and pooled urine were done with certain strains in order to determine if variation in the expression of porins related to differences in the osmolality of the media. Strains grown in serially diluted Luria broth or pooled urine showed that the number of porin bands increased from one to three when the osmolality of the culture medium decreased, and an increasing osmolality decreased the number of porins observed to one in all strains tested (figs. 4, 5). In addition to the major outer membrane protein patterns demonstrated, most of the strains processed directly from urine showed several protein bands of high molecular weight in the
PORINS OMP A
31. 021. S -
1
2
3
4
FIG. 1. Omp pattern 1 in SDS-PAGE found in E. coli strains isolated from patients with bacteriuria (> 105 CFU/ml.). Lanes: 1: molecular weight standards; 2: Omp patterns from bacteria processed directly from urine showing band at porin levels (40 kDa) and band at Omp A level (36 kDa); 3: same strain grown in Luria Broth showing two bands at porin levels and one band at Omp A level; 4: same strain grown in pooled urine from normal females showing pattern similar to lane 2. Location of iron regulated membrane proteins is indicated by IRMP.
range between 66.2 kDa to 92.5 kDa, levels at which iron regulated membrane proteins (IRMP) were expressed. The relationship of these bands to the absence of free iron was confirmed when individual strains grown in low iron conditions expressed several high molecular weight protein bands (usually five in the range between 62.5 kDa to 92.5 kDa). These bands were not present when the medium was supplemented with iron. Figure 6 demonstrates IRMP expression in the presence and absence of iron when strains were tested after four hours of growth. DISCUSSION
E. coli is a microorganism with a remarkable ability to adapt to changes in growth medium making it suitable as a successful colonizer in various environments. In the urinary tract, despite adverse conditions (pH, osmolality, urea concentration) 12• 13 limiting, almost totally the growth of many organisms, E. coli is able to survive, colonize and ultimately invade the kidney causing pyelonephritis. The seventeen bacteriuric isolates demonstrated a variety of strain characteristics, with the presence of mannose sensitive agglutination (Type 1 fimbriae) being more prevalent, a property which has been reported as a common finding in E. coli uroisolates. 3 •4 It is of interest that most strains did not belong to the usual O serotypes classified as uroisolates, 22 • 23 and that
K
.5
R p
66.
JePORIN
:JOM
3L
2 Ea coli strains isolated FIG. 2. Omp pattern 2 in SDS-PAGE isolated from patients with bacteriuria (> molecular weight standards; 2: pattern from directly from urine showing band porin levels (40 kDa) band at approximately 21 kDa level; 3: same strain grown in pooled urine from normal females showing band at porin levels and band at A level bands (36 kDa); 4: same strain grown in Luria Broth showing at porin levels and band at Omp A leveL Location of iron regulated membrane proteins is indicated by IRMP.
mannose resistant agglutination (P E""'Jao," and production were not present in all strains. Expression of porins was different among strains isolated directly from urine or m urine from those grown in artificial culture 111t,u1u11,, reflecting differences in osmolality found between the standard (400 and the of mine c,sed (800 24 Because of human quantitative the most evident difference in the expression of porins ,Nas seen when of -~,,.,-..-,, (high to 861 to 439 and 1020 to 489 mOsm/kg. (Luria reports which have demonstrated that expression of E. coi.i porins in vitro is subjected to variation in environmental changes, osmolality and temperature. Porin protein C is F porin in favored in a medium of high osmolality and low osmolality. 16• 17 Since our results demonstrated that all strains expressed one porin band in vivo and in pooled urine (high osmolality), this band should correspond to Omp C. On the other hand, Omp F should be one of the additional porin bands which was present during growth in Luria broth (low osmolality). The position of Omp C and Omp F bands in our gels are reversed in relation to that which has been reported in the literature. 16· 20 We utilized 4M urea in order to obtain a better resolution of the major outer membrane proteins. 18 Urea has been shown to alter the
Lanes: L rnolecular strains grown in. and one urine fro111
1 FIG. 4. pooled urine
SDS-PAGE f:rom E. coli strain gTuw11 -n different osmolalities Lanes:
dilution ,,ccL,CKe
THE JOURNAL OF UROLOGY Copyright© 1990 by AMERICAN UROLOGICAL ASSOCIATION, INC.
OUTER MEMBRANE PROTEINS OF E.COLI IN THE HOST-PATHOGEN INTERACTION IN URINARY TRACT INFECTION JAIME A. ROBLEDO, ALIX SERRANO
AND
GERALD J. DOMINGUE*
From the Department of Urology and Department of Microbiology and Immunology, Tulane University School of Medicine, New Orleans, Louisiana
ABSTRACT
Outer membrane protein patterns (Omp) of Escherichia coli obtained directly from the urine of bacteriuric patients without passage on artificial culture media (ACM) were studied by polyacrylamide gel electrophoresis (SDS-PAGE) in an effort to determine whether in vivo conditions of growth affected the expression of these bacterial surface structures. Seventeen strains studied showed two distinct Omp patterns: one protein band appeared at the level of porin proteins (40 kDa) in both patterns, but Omp A protein was at the level of 36 kDa in the first pattern and a new protein was observed at 21.5 kDa in the second pattern suggesting that it is a fragment of Omp A. High molecular weight proteins were also observed in most of the strains and this finding was related to lack of free iron when the same strains were grown under iron restricted conditions in vitro. The same strains grown in pooled urine from normal females showed the first pattern mentioned above. Comparative growth on ACM of urinary strains and E. coli strains isolated from blood, feces and wounds showed an increase in the number of porins expressed (from 1 to 2 or 3, with some variability observed between strains). Differences in osmolality between pooled urine and ACM used, plus in vitro studies varying the osmolality of culture media, showed that osmolality accounted for differences in the number of porins expressed: porin expression decreased in urine and ACM of high osmolality, suggesting that the same phenomena occurred in vivo. It is concluded that host factors including low availability of iron and high osmolality present in the urinary tract influence the expression of several E. coli surface proteins. These proteins may relate to the ability of E. coli to colonize and invade the urinary tract by regulating the physiologic and/or metabolic state of the bacterial cell favoring survival of the organism in a hostile environment. Specific immune responses directed against porins could influence the outcome of this host-parasite interaction. (J. Ural., 143: 386-391, 1990) E. coli is the most frequent cause of urinary tract infection in humans. 1 Several bacterial virulence factors have been implicated in the host-pathogen interation, yet despite extensive research, no single virulence factor has completely accounted for the urovirulence of E. coli in pyelonephritis. Most of the recent efforts have been devoted to the role of adhesins (fimbriae) of E. coli recognizing specific receptors on the mucosal epithelium. Strains having P fimbriae adhesin have been linked to major inflammatory changes in the urinary tract, 2 and are more often isolated from patients with pyelonephritis, 3·4 suggesting that P fimbriae are important in the pathogenesis of the disease. Additionally, bacterial capsules and hemolysin have been reported to be significant as virulence factors. 5 ' 6 Recent data, however, from our laboratory, utilizing an experimental murine model of pyelonephritis indicated that wild type E. coli strains devoid of fimbriae, hemolysin, acidic capsules and sensitive to human serumcidal activity were capable of causing incipient and acute pyelonephritis. 7 Furthermore, even among identical serotypes and biotypes, the presence of fimbriae did not appear to be critical factors in urovirulence, nor did the presence of several other positive characteristics (hemolysin, K capsule, flagella, serum resistance) in a given strain enhance uorpathogenicity. These phenotypic characteristics may simply represent associated or serologic markers with the host serving as the dominant determinant of bacterial virulence and susceptibility to urinary infection. Accepted for publication September 14, 1989.
* Requests for reprints: Dept. of Urology, Tulane University School of Medicine, 1430 Tulane Ave., New Orleans LA 70112. Presented at the 88th Annual Meeting of the American Society for Microbiology, Abstract Number B205, May 8, 1988, Miami, Florida. 386
Bacterial virulence, therefore, should be examined in the context of the relationship of both microorganism and host: bacterial survival in a hostile environment, and host factors which intervene to overcome initiation of bacterial colonization and further infection of the urinary tract. This complex relationship must involve not only the above mentioned virulence factors but the entire bacterial cell structure. In fact, bacteria growing in vivo may be different metabolically and structurally from organisms traditionally studied by in vitro conditions on artificial culture medium. 8 Urine is a complex body fluid containing a variety of excreted products. 9 It does support bacterial growth although several urine factors have been implicated as inhibitory. For example, the lack of free iron has been related to the expression of iron regulated membrane proteins (IRMP) on bacteria isolated from the human urinary tract, 10• 11 indicating a restrictive condition for bacterial growth and a consequent adaptational change. Although other factors (pH, concentration of urea in urine and urine osmolality) have been demonstrated as inhibitory for bacterial growth under certain conditions, 12• 13 the nature of the structural changes which bacteria undergo while growing in the urinary tract is mostly unknown. Major outer membrane proteins are part of the external surface of the bacterial cell. Porins belong to a particular group of outer membrane proteins that have been termed peptidoglycan-associated general diffusion pore proteins, which besides serving as receptors for certain bacteriophages also act as nonspecific pores that allow the entrance of molecules smaller than 600 M.W. into the bacterial cell. 14 Since the expression in vitro of this group of proteins is related to the composition of the culture medium or environmental conditions, 15- 17 it appears
E. COLI MEMBRANE PRODUCTS
warranted to study the nature of these proteins expressed by urinary pathogens in vivo in an effort to define their role in the host-pathogen interaction. Specifically, the aims of the present study were to determine the expression of E. coli outer membrane proteins (Omp) in vivo when bacteria were obtained directly from human bacteriuric urines (without passage on artificial culture media) and to compare this expression to that observed on the same strains grown on artificial culture media and in pools of human urine in vitro. In order to determine whether there are differences between bacteriuric isolates of E. coli and isolates from other sources, several E. coli strains from feces, blood and wounds were compared. MATERIALS AND METHODS
E. coli strains. Seventeen E. coli strains isolated from bacteriuric urines and with counts greater than 105 CFU /ml. were collected from the hospital/clinic of Tulane University Medical Center. Positive urines were kept under refrigeration at 4C until laboratory results indicated that E. coli was the only isolate and the quantitative count was confirmed. Additionally, seven E. coli strains isolated from blood, eleven from feces of normal individuals and two from wounds were obtained from the hospital/clinic laboratory. The strains isolated from patients with bacteriuria were also characterized according to 0 serotype, mannose resistant agglutination, mannose sensitive agglutination and hemolysin production according to procedures previously described. 7 Strains isolated from urine as well as strains from blood, feces and wounds were inoculated into trypticase soy broth plus 5% glycerol, grown overnight and kept at -70C until used. Culture media and growth of the bacterial strains. Luria broth and a pool of urine from normal females (morning urine, filtered through a 0.22 micron filter) were utilized for growing the E. coli strains in vitro. Strains isolated from urine were studied without passage on artificial culture media. Osmolalities were not performed on these individual urines. Also, these urine strains were grown overnight as static cultures in Luria broth, four hour subcultures served as inocula (108 CFU/ml. in 10 ml. as final concentration) for fresh Luria broth and urine pools. Broth and urine cultures were subsequently incubated for four hours at 35C. Strains from blood, feces and wounds were isolated in pure culture on routine artificial culture media (for no more than one to two passages) prior to inoculating in fresh Luria broth and urine pools. Extraction, electrophoresis and identification of outer membrane proteins. Omp extraction methods were based on procedures described by Achtman et al. 18 Strains isolated directly from urine which had E. coli as the only bacterial isolate and with counts> 10 5 CFU/ml. were subjected to differential centrifugation (1,000 RPM for 10 minutes, followed by 3,500 RPM for 15 minutes) in order to obtain a bacterial pellet free of epithelial cells and leukocytes. This bacterial pellet plus bacterial pellets obtained from strains grown in Luria broth and pooled urine were suspended in 50 mM TRIS-HCL pH 8.0 to a final concentration of 108 CFU/ml., and disrupted by sonication (40 seconds, two to three times, at half intensity in the cold). Unbroken cells and bacterial debris were removed by centrifugation (1,800 g, 4C for 20 minutes) and bacterial membranes were pelleted from the supernate (45,650 g, 4C for 60 minutes). The pelleted membranes were resuspended in 150 microliters of distilled H 2 0. Eight volumes of N-Laurylsarcosine (1.67% N-Laurylsarcosine, Sigma Chemical Co., St. Louis, MO, 11.1 mM TRIS-HCL pH 7.6) were added to 50 and 100 microliters aliquots of the membrane suspension to solubilize the inner membrane proteins; outer membrane proteins were precipitated by centrifugation (in eppendorftubes using adaptors in a Beckman rotor Model JA-20, at 20,000 rpm, 20C for 90 minutes). The 50 tnicroliter aliquots were assayed for protein concentra-
387
tion by using a modification of the Lowry method described by Markwell et al., 19 the remaining aliquots were resuspended in SDS-PAGE sample buffer (0.0625 M TRIS-HCL pH 6.8, 10% glycerol, 2% SDS, 5% 2-mercaptoethanol, 0.001% bromophenol blue), and adjusted to a protein concentration of 0.5 to one mg./ml. The samples were heated to 100 C for five minutes before being applied to the gel, each well contained 30 micrograms of protein. Discontinuous SDS-PAGE was performed by a modification of the Laemmli procedure, 18 in which four M urea was added to the separation gel and 0.15% of SDS was used in the electrode buffer. The gels were stained with coomassie blue R-250 (BioRad Laboratories, Richmond, CA) for 30 minutes, followed by overnight destaining in 40% methanol-10% acetic acid, and further soaking in 50% methanol, 5% glycerol before storage. Major outer membrane proteins were defined as proteins with apparent molecular weights between 30 kDa nd 40 kDa. 18• 20 Omp A was identified by its heat modifiability, appearing at approximately 35 to 36 kDa after exposure to SDS at lOOC and showed an increased mobility to approximately 31 kDa when the samples were heated to 37C in the presence of SDS. 18020 Porins were identified as the protein bands that appeared at approximately 40 kDa when the samples were heated to lOOC in the presence of SDS and were not solubilized when heated to 37C; 18• 20 further characterization of individual porins were not made. Iron regulated membrane proteins (IRMP) were identified as the proteins that appeared in the range between 66.2 kDa to 92.5 kDa when the individual strains were grown in TSB (Trypticase soy broth; BBL, Becton, Dickinson and Co.) supplemented with ovotransferrin (0.5 mg./ml.; type 1, Sigma Chemical Co., St. Louis, MO 0.07 M NaHC0 3 ), and IRMP were not expressed when grown in the same medium supplemented with iron (0.02 mM FeS0 4 ). 10• 11 • 21 Osmolality studies. To determine whether osmolality affected the expression of the major outer membrane proteins in vivo several E. coli strains isolated from human urine were grown in Luria broth and pooled urine at different osmolalities with a range between 100 to 1000 mOsM/kg. Lyophilized urine was reconstituted and diluted with various volumes of distilled water until the desired osmolality was obtained. Luria broth was likewise diluted. Increasing the osmolality of Luria broth was accomplished by adding sodium chloride to the medium. Omp preparations and SDS-PAGE were performed as previously described as well as the growth conditions. RESULTS
Strain characteristics. Results summarized in table 1 indicate that only six of the strains were typeable using O antisera commonly employed for classifying E. coli strains causing urinary tract infections. Nine exhibited mannose resistant agglutination of human erythrocytes (P fimbriae), fourteen strains showed mannose sensitive agglutination of guinea pig erythrocytes (type 1 fimbriae, additionally confirmed by mannose sensitive agglutination of Saccharomyces cerevisiae cells), and seven strains produced hemolysin in 5% sheep blood agar plates. One strain, serotype 018, and three non typeable strains were positive for all characteristics tested. Omp patterns in SDS-PAGE. Strains exhibited individual variabilities in their major outer membrane protein patterns by SDS-PAGE. However, all strains (including the bacteriuric isolates and strains isolated from other sources) could be grouped according to common similarities in their expression of major outer membrane proteins. The seventeen strains isolated from urine and analyzed by SDS-PAGE, without passage on artificial culture media, showed two distinct patterns at the level of major outer mem brane proteins. The first one was found in a group of seven strains which showed one band at approximately 40 kDa corresponding to porins, and one band at approximately 36 kDa corresponding to Omp A (fig. 1). The second pattern was
388
ROBLEDO, SERRANO AND DOMINGUE TABLE
1. Characteristics of 17 E. Coli strains isolated from
human urine Strain
0 Serotype
U-2333 U-2336 U-2312 U-2309 U-2263 U-2308 U-1773 U-2882 U-2928 U-2929 U-3029 U-3040 U-3060 U-3061 U-ST U-3162 U-3271
NT** NT NT NT NT NT ND*** 018 NT 075 02 01 NT NT 06 ND 018
MR*
MS*
Agglut.
Agglut.
+
+
+
+ + + + + +
ND
ND
ND
+
+ + + +
+ + + + + + +
ND ND
ND ND
+
+
+ +
Hemolysin
+
KDa 92. 5
IRMP
4S. 0+ + + +
* Mannose resistant agglutination of human erythrocytes and mannose sensitive agglutination of guinea pig erythrocytes indicative of P fimbriae and type 1 fimbriae respectively. ** Non-typeable utilizing 026, 055, 0111, 0127, 086, 0119, 04, 011, 0124, 0125, 0126, 044, 0112, 028, 020, 018, 08, 09, 015, 022, 075, 083, 02, 01, 06 antisera *** Not done.
observed in a group of seven strains that showed one band at approximately 40 kDa at the level of porins and a second band at approximately 21.5 kDa that may correspond to either Omp A or one of its fragments, since this is the location for the tryptic fragment of Omp A when the protein is exposed to trypsin (fig. 2). In three strains, the patterns observed were unclear (with faint bands) and were not assigned to any of the groups. The initial concentration of bacteria for those samples was low, yielding less protein in the final sample. The procedure could not be repeated since the bacteriuric urine was no longer available. Figures 1 and 2 also show the patterns observed when the strains from both groups were grown in vivo; in pooled urine all showed a pattern similar to the first one mentioned above, one band corresponding to porins and one band corresponding to Omp A. A different pattern also common to all strains however, was observed when the bacteriuric strains were grown in artificial culture medium (Luria broth); at approximately 40 kDa, two to three bands (some variability in the number observed between strains) were demonstrated with characteristics compatible with porins, and one band was observed at approximately 36 kDa, the level of Omp A. The strains isolated from blood, feces and wounds, despite showing individual variability in the expression of outer membrane proteins (as observed with the urinary strains) gave similar results to the urinary isolates when grown in vitro (Luria broth and pooled urine). In artificial culture medium, these strains expressed two to three bands at the porin levels (approximately 40 kDa) and one band at Omp A level (approximately 36 kDa), and in pooled urine, the band at the Omp A level and only one band at the porin level (fig. 3). It was observed that the osmolality of standard Luria broth and the pooled urine samples used to grow all strains was markedly different (400 mOsm/kg. and 800 mOsm/kg. respectively). Experiments varying the osmolality of both the Luria broth and pooled urine were done with certain strains in order to determine if variation in the expression of porins related to differences in the osmolality of the media. Strains grown in serially diluted Luria broth or pooled urine showed that the number of porin bands increased from one to three when the osmolality of the culture medium decreased, and an increasing osmolality decreased the number of porins observed to one in all strains tested (figs. 4, 5). In addition to the major outer membrane protein patterns demonstrated, most of the strains processed directly from urine showed several protein bands of high molecular weight in the
PORINS OMP A
31. 021. S -
1
2
3
4
FIG. 1. Omp pattern 1 in SDS-PAGE found in E. coli strains isolated from patients with bacteriuria (> 105 CFU/ml.). Lanes: 1: molecular weight standards; 2: Omp patterns from bacteria processed directly from urine showing band at porin levels (40 kDa) and band at Omp A level (36 kDa); 3: same strain grown in Luria Broth showing two bands at porin levels and one band at Omp A level; 4: same strain grown in pooled urine from normal females showing pattern similar to lane 2. Location of iron regulated membrane proteins is indicated by IRMP.
range between 66.2 kDa to 92.5 kDa, levels at which iron regulated membrane proteins (IRMP) were expressed. The relationship of these bands to the absence of free iron was confirmed when individual strains grown in low iron conditions expressed several high molecular weight protein bands (usually five in the range between 62.5 kDa to 92.5 kDa). These bands were not present when the medium was supplemented with iron. Figure 6 demonstrates IRMP expression in the presence and absence of iron when strains were tested after four hours of growth. DISCUSSION
E. coli is a microorganism with a remarkable ability to adapt to changes in growth medium making it suitable as a successful colonizer in various environments. In the urinary tract, despite adverse conditions (pH, osmolality, urea concentration) 12• 13 limiting, almost totally the growth of many organisms, E. coli is able to survive, colonize and ultimately invade the kidney causing pyelonephritis. The seventeen bacteriuric isolates demonstrated a variety of strain characteristics, with the presence of mannose sensitive agglutination (Type 1 fimbriae) being more prevalent, a property which has been reported as a common finding in E. coli uroisolates. 3 •4 It is of interest that most strains did not belong to the usual O serotypes classified as uroisolates, 22 • 23 and that
K
.5
R p
66.
JePORIN
:JOM
3L
2 Ea coli strains isolated FIG. 2. Omp pattern 2 in SDS-PAGE isolated from patients with bacteriuria (> molecular weight standards; 2: pattern from directly from urine showing band porin levels (40 kDa) band at approximately 21 kDa level; 3: same strain grown in pooled urine from normal females showing band at porin levels and band at A level bands (36 kDa); 4: same strain grown in Luria Broth showing at porin levels and band at Omp A leveL Location of iron regulated membrane proteins is indicated by IRMP.
mannose resistant agglutination (P E""'Jao," and production were not present in all strains. Expression of porins was different among strains isolated directly from urine or m urine from those grown in artificial culture 111t,u1u11,, reflecting differences in osmolality found between the standard (400 and the of mine c,sed (800 24 Because of human quantitative the most evident difference in the expression of porins ,Nas seen when of -~,,.,-..-,, (high to 861 to 439 and 1020 to 489 mOsm/kg. (Luria reports which have demonstrated that expression of E. coi.i porins in vitro is subjected to variation in environmental changes, osmolality and temperature. Porin protein C is F porin in favored in a medium of high osmolality and low osmolality. 16• 17 Since our results demonstrated that all strains expressed one porin band in vivo and in pooled urine (high osmolality), this band should correspond to Omp C. On the other hand, Omp F should be one of the additional porin bands which was present during growth in Luria broth (low osmolality). The position of Omp C and Omp F bands in our gels are reversed in relation to that which has been reported in the literature. 16· 20 We utilized 4M urea in order to obtain a better resolution of the major outer membrane proteins. 18 Urea has been shown to alter the
Lanes: L rnolecular strains grown in. and one urine fro111
1 FIG. 4. pooled urine
SDS-PAGE f:rom E. coli strain gTuw11 -n different osmolalities Lanes:
dilution ,,ccL,CKe
