Growth media and temperature effects on biofilm formation by serotype O157:H7 and non-O157 Shiga toxin-producing Escherichia coli
Gaylen A. Uhlich,
Molecular Characterization of Foodborne Pathogens Research Unit, Eastern Regional Research Center, Agricultural Research Service, U.S. Department of Agriculture, Wyndmoor, PA, USA
Correspondence: Gaylen A. Uhlich, Molecular Characterization of Foodborne Pathogens Research Unit, Eastern Regional Research Center, Agricultural Research Service, U.S. Department of Agriculture, 600 East Mermaid Lane, Wyndmoor, PA 19038-8542, USA. Tel.: 215 233 6740;
Biofilm formation in most Escherichia coli strains is dependent on curli fimbriae and cellulose, and the production of both varies widely among pathogenic strains. Curli and cellulose production by colonies growing on agar are often identified by their affinity for Congo red dye (CR). However, media composition and incubation temperature can affect dye affinity and impose limitations on red phenotype detection by this method. In this study, we compared different Shiga toxin-producing E. coli for CR affinity and biofilm formation under different media/temperature conditions. We found strain and serotype differences in CR affinities and biofilm formation, as well as temperature and media requirements for maximum CR binding. We also constructed strains with deletions of curli and/or cellulose genes to determine their contributions to the phenotypes and identified two O45 strains with a medium-dependent induction of cellulose.
Escherichia coli biofilm formation typically requires the production of adhesive curli fimbriae and the exopolysaccharide, cellulose (Olsén et al., 1989; Zogaj et al., 2001). Curli are encoded by csgBA, and cellulose production is directed from the bcsABZC gene cluster in which bcsA (cellulose synthase) and bcsC are essential (Römling et al., 1998; Zogaj et al., 2001). Cellulose and curli are often identified using their affinity for Congo red (CR) dye. In certain Salmonella and E. coli, growth on agar containing CR produces a brown colony when producing curli, a pink phenotype when producing cellulose, and a dark red and dry phenotype when producing both (Zogaj et al., 2001; Solano et al., 2002). Strains producing neither remain white. Cellulose is also detected by its affinity for calcofluor, which produces fluorescence under a UV light source (Zogaj et al., 2001; Solano et al., 2002). In a study of commensal E. coli, biofilm formation was strongest among strains that produced both cellulose and curli and reduced in strains producing only one factor (Bokranz et al., 2005). A few strains formed some biofilm in the absence of both. Curli and cellulose expression requires the transcriptional regulator CsgD. CsgD directly activates the curli structural operon, csgBAC, and indirectly stimulates cellulose expression by increasing adrA transcription. The diguanylate cyclase activity of AdrA produces cyclic(c) di-GMP that is essential for cellulose production (Gerstel & Römling, 2003; Barnhart & Chapman, 2006). CsgD regulation is complex requiring protein transcription factors, GGDEF/EAL proteins that control c-di-GMP levels, and various sRNAs (Gerstel & Römling, 2003; Sommerfeldt et al., 2009; Ogasawara et al., 2010a, b; Thomason & Storz, 2010; Jørgensen et al., 2012). Such regulation allows csgD transcription to respond to extracellular stimuli such as osmolarity, pH, O2 concentrations, catabolite levels, and membrane alterations (Vidal et al., 1998; Gerstel & Römling, 2001; Jackson et al., 2002; Ferrières & Clarke, 2003; Dorel et al., 2006; Ogasawara et al., 2007). Strain and/or serotype differences in csgD expression would be expected due to the high number of regulators and the complexity of the network. We previously showed that Shiga toxin-producing E. coli (STEC) varied greatly in their ability to express curli and generate biofilm (Chen et al., 2013; Uhlich et al., 2013). Curli and biofilm deficiencies in serotype O157:H7 likely stem from bacteriophage insertions in mlrA, a DNA-binding protein that enhances RpoS-driven csgD transcription (Uhlich et al., 2013). In non-O157:H7 STEC, strong biofilm formation was more readily observed. The lowest expression was seen in strains with motility deficiencies, RpoS mutations, or prophage insertions in mlrA (Chen et al., 2013). These studies were performed using Congo red indicator (CRI) agar for CR affinity studies (Hammar et al., 1996), Luria–Bertani (LB) broth with no salt (LB-NS) to maximize biofilm expression (Uhlich et al., 2006), and 23–25 °C incubations to simulate environmental conditions. However, other STEC studies have described CR-binding differences dependent on temperature and culture media (Saldaña et al., 2009). In this study, we compared CR affinity and biofilm formation under different media/temperature conditions. In addition, we tested strains with deletions of curli and/or cellulose genes to determine their contributions to the phenotypes observed.
Materials and methods
Bacterial strains and mutant construction
Bacterial strains are listed in Table 1. Non-O157:H7 isolates from the seven important serogroups (O26, O45, O103, O111, O113, O121, and O145) were prescreened to include strains with a range of biofilm phenotypes (Chen et al., 2013). Strain 43894OR (OR), a constitutive curli-producing strain, was used as a positive control for biofilm and Congo red (CR) dye affinity studies (Uhlich et al., 2001). Strain 43895 (43895OW in certain previous studies), shown to be a nonbiofilm-forming isolate on different surfaces in various media, was used as a negative biofilm control (Uhlich et al., 2006, 2013). Cultures from frozen glycerol stocks were recovered on LB agar at 30 °C. When appropriate, kanamycin (25 mg L−1) and chloramphenicol (10 mg L−1) were added. Gene deletions were constructed by replacing the ORF with a neomycin resistance cassette (neo) as previously described (Uhlich, 2009); csgBA deletions removed sequence from the csgB start codon to the csgA stop codon. All deletions in the FCL1 strain were generated using a chloramphenicol resistance cassette (Gene Bridges GmbH). The csgB gene in strains PA5 and PA46, and csgA in strain SJ7 were deleted instead of csgBA; bcsC was deleted in strain SJ7. Deletions were verified by PCR, and resistance cassettes were not cured.
ARS, Agricultural Research Service, Eastern Regional Research Center; ATCC, American Type Culture Collection, Manassas, VA, USA; CDC, US Centers for Disease Control and Prevention, Atlanta, Georgia, USA; MSU, Michigan State University Department of Microbiology and Molecular Genetics STEC Center, East Lansing, Michigan, USA; PADH, Pennsylvania Department of Health; PHAC, Public Health Agency of Canada, Winnipeg, Manitoba, Canada.
CR affinity was assayed on CRI or T-medium agar (Collinson et al., 1991) containing 20 mg L−1 CR dye and 10 mg L−1 Coomassie brilliant blue G-250 (TA). Calcofluor (Fluorescent brightener 28; Sigma) was added to T-medium agar at 200 mg L−1 to detect cellulose (Uhlich et al., 2006). Biofilm was assayed as previously described using YESCA broth (medium base for CRI) and T-medium at 25 and 30 °C; selected strains were tested at 37 °C (Hammar et al., 1996; Uhlich et al., 2013). The mean optical densities at 590 nm (OD590 nm) of eluted crystal violet (CV) from four wells of each strain were calculated. At temperatures of 25 and 30 °C, the means of all strains, calculated for a particular media type, were compared with control strain 43895 using a Dunnett's test. Strains with mean OD values significantly greater (P < 0.05) than strain 43895 were interpreted as having developed biofilm. Their frequencies at 25 and 30 °C in YESCA broth and T-medium were subjected to logistical regression, and odds ratios were calculated for parameters showing a significant effect (sas/stat 9.3, Cary, NC).
Medium and temperature effects on CR affinity in serotype O157:H7 STEC
Strain appearance on CRI and TA after 48-h incubation at different temperatures is shown in Fig. 1. Light reflection off the dry surface of strain OR masked its underlying color in many images, but its true red color is apparent at the colony borders, particularly on TA at 25 and 30 °C (Fig. 1a). There was little CR binding by the O157:H7 strains on CRI at 25 °C with only PA32 and PA46 showing red color. At 30 °C, PA32 and PA46 became darker red and most other strains appeared pale brown. At 37 °C, most strains showed elevated brown color. PA6 and PA46 were the only strains showing dark red staining. Strain PA1 was noteworthy in being the only strain whose red color intensity diminished with each temperature increase. In strain PA32, red color was most intense at 30 °C but diminished at 37 °C. In general, color was darker on TA than CRI. At 25 °C, PA32 and PA46 stained the darkest red, but strains such as PA1, PA51, and B6-914 also showed low CR affinity. Some additional strains showed red coloring on TA at 30 °C, but PA32 and PA46 remained the darkest.
Many O157:H7 strains showed more intense red staining on TA at 37 °C than at 30 °C, but a few strains (PA32, PA1, PA7, PA52, and B6-914) stained less at 37 °C than at 30 °C, indicating that the optimal temperature for CR binding varies among strains. It should be noted that few strains showed dark red staining under optimal conditions, none had a dry appearance, and only strains PA46 (at 30 and 37 °C) and PA32 (at 30 °C) showed red color comparable with the 43894OR control. Even under more favorable media and temperature conditions, CR affinity in serotype O157:H7 is relatively low.
Medium and temperature effects on CR affinity in non-O157:H7 STEC
When tested on CRI, eight non-O157:H7 STEC strains showed red color but none approached the color intensity of strain OR at 25 °C (Fig. 1b). At 37 °C, there was a dramatic reduction in color (six strains lost nearly all CR affinity) and the dry phenotype (FCL1, SJ10, and SJ13). Only strain 04-1450 showed dark staining, which was relatively consistent at all three temperatures. The results of non-O157:H7 STEC on TA was overall similar to those seen on CRI, except that the intensity was increased and the color of some strains possessed a purple hue. In contrast to the O157:H7 strains, CR affinity was repressed at 37 °C on both CRI and TA in those non-O157 STEC strains capable of strong CR binding, except for strain 04-1450 whose color remained strong at all three temperatures. Strains SJ29 and 96-3285 showed enhanced CR binding at 30 and 37 °C but not at 25 °C. Similar to the O157:H7 strains, TA enhanced CR affinity compared with that displayed on CRI, but it did not affect CR affinity in strains that showed little or no affinity on CRI. Strains SJ7 and 05-6545 were exceptions, showing strong color on TA but little color on CRI.
Curli- and cellulose-defective mutants
Deletions in the csg or bcs operons were constructed in selected CR-binding strains to study the role of curli and cellulose in CR affinity (Fig. 2). For the O157:H7 strains, csgBA deletion resulted in complete loss of CR affinity, indicating an essential role for curli. Deletion of bcsA had little effect on CR affinity in these strains.
For non-O157:H7 mutant strains, CR binding was more evident at 25 °C than 37 °C. While bcs deletion resulted in either no effect or complete loss of CR affinity, depending on strain background, deletion of the csgB/A genes resulted in reduction in CR affinity and changes in hue, from purple to pink. Deletion of csgD, the essential regulator of both curli and cellulose, completely abolished CR affinity in all non-O157 strains tested.
For SJ7 and 05-6545 (previously shown to be curli-deficient on CRI; Chen et al., 2013), which showed CR affinity on TA but not on CRI, deletion of bcs genes eliminated CR affinity at 25 °C, indicating that cellulose was responsible for CR binding. However, deletion of csg genes reduced CR affinity, suggesting that a low level of curli expression may augment CR binding by cellulose in those strains. This concurs with earlier findings where extracellular curli production by strain 05-6545 could be detected by anti-CsgA antibodies but not by protein staining (Chen et al., 2013). Interestingly, at 37 °C where most CR affinity of the non-O157 strains was repressed, deletion of csgBA allowed greater CR affinity in strain 05-6545.
At 25 °C, curli-producing strains 05-6544 and FCL1 (Chen et al., 2013) showed complete loss of red staining when csgD was deleted and partial loss when csgBA was deleted, indicating that curli and cellulose were both required for full CR affinity. However, bcsA deletion had little effect on strain color, suggesting that cellulose either played a very minor role or its contributions were masked by strong curli expression. At 37 °C, CR affinity was low on both media for all four non-O157:H7 parent strains. However, deletion of csgBA genes enhanced CR binding in strain 05-6544 and 05-6545 on both media. It was not clear why the suppression of CR binding in certain non-O157 STEC strains was reversed by csg deletions at 37 °C.
Calcofluor staining for cellulose production
Calcofluor binding of select strains is also shown in Fig. 2. Calcofluor binding was minimal in O157:H7 strains with no difference in staining between bcsA mutants and parents at any temperature, indicating that cellulose was not expressed. A barely discernible fluorescence reduction between csgBA mutants and the PA32 and PA46 parents suggests a slight calcofluor affinity for curli in those strains, as is apparent in the Salmonella Enteritidis control strain 942 (curli+, cellulose−).
All non-O157 parent strains tested showed some fluorescence at 25 °C. Deletion of bcsA and csgD reduced the fluorescence, indicating that cellulose is binding calcofluor. There were no differences between the bcsA mutants and parents of the non-O157 strains at 37 °C. However, csgBA deletion caused a slight fluorescence increase in some strains, suggesting that curli may interfere with calcofluor binding by other cellular components. At 25 °C, csgD deletion eliminated all calcofluor affinity in the non-O157:H7 parent strains. Deletion of csgBA in 05-6544 and FCL1 did not affect the intense fluorescence of the parents; however, bcsA deletion severely decreased the intensity, indicating that cellulose was responsible for most calcofluor binding. The residual staining in bcsA mutants likely represents minor binding by either curli or a different CsgD-regulated factor. We also tested SJ7 and 05-6545 on calcofluor-containing CRI-base medium at 25 °C and found no fluorescence (results not shown). Considering the differences in color on CRI/TA and in fluorescence by calcofluor staining of the mutants and parents, we believe that the dark purple color (more apparent on TA) was due to curli binding, while the pink/magenta color was mostly attributed to cellulose binding.
Biofilm formation on polystyrene
Biofilm formation was tested using both YESCA broth and T-medium at 25 and 30 °C; a subset of strains was also tested at 37 °C (Table 2). None of the serotype O157:H7 strains formed significant amounts of biofilm using YESCA at 25 °C including PA32 and PA46, the two strains that bound CR on CRI at 25 °C. Using T-medium at 25 °C, only strain PA46 (OD590 nm = 0.69) and PA7 (OD590 nm = 0.19) formed significant biofilm. When cultured at 30 °C, PA46 formed biofilm in both YESCA broth (OD590 nm = 0.22) and T-medium (OD590 nm = 0.82). Strains PA14, PA15, and PA32 also formed small amounts of biofilm using T-medium (OD590 nm = 0.15–0.24). We also tested biofilm formation at 37 °C in 12 O157:H7 strains that showed higher CR affinities. Of the 12 strains, only PA46 formed trace amounts of stainable biofilm (CV OD590 nm = 0.18) in T-medium. Although 37 °C appears favorable for serotype O157:H7 CR affinity, it was the poorest of the three temperatures for supporting biofilm formation.
Table 2. 48-h biofilm formation on polystyrene in T-medium broth and YESCA broth at 25 °C
Values represent the mean optical density (OD) ± standard deviation (SD) measured at 590 nm of crystal violet (CV) dye eluted from four stained wells/sample. The experimental means calculated for each of the strains using the designated medium at either 25 °C or 30 °C were compared with that of control strain 43895 under the same conditions using the Dunnett's Test.
Strain 43894OR was not included in the analysis.
Strain mean different (P = 0.05) from the control strain 43895.
The non-O157 STEC were better biofilm formers than the O157:H7 strains at 25 and 30 °C, but they formed little biofilm at 37 °C (Table 2). Most strains that bound CR dye also tended to form biofilm; however, CR dye-binding intensity did not necessarily correspond with the amount of biofilm formation. There was considerable strain variation regarding the optimum temperature and medium for biofilm formation. Strains 05-6545 and SJ14 bound CR poorly on CRI but formed low to moderate biofilm on both media at 25 and 30 °C. Strains 98-8338, SJ15, SJ7, and 03-4064 bound visible amounts of CR dye under most conditions but generated little or no biofilm. Therefore, CR binding had limited value for predicting biofilm formation on polystyrene in individual strains.
Logistic regression analysis revealed a significant medium effect on biofilm formation with strains being 2.69 times more likely to form biofilms in T-medium compared with YESCA broth. There was no significant effect determined for temperature between 25 and 30 °C.
In this study, we compared non-O157 STEC from seven important O-serogroups to serotype O157:H7 strains for growth media and temperature effects on CR affinity and biofilm formation. Both O157:H7 and non-O157 STEC showed higher CR affinity on TA than on CRI. Comparison of curli isolations from selected strains grown on TA and CRI showed greater curli protein production on TA, indicating that CR binding increases were due to higher curli production rather than increased CR-binding efficiency (data not shown). This agrees with previous findings (Saldaña et al., 2009). We also showed that CR binding by O157:H7 strains was usually greater at 30 and 37 °C than at 25 °C. However, the optimal temperature varied among strains. In contrast, the non-O157 STEC tended to prefer 25 and 30 °C, and most strains showed little CR binding at 37 °C. Such findings suggest that certain adhesion factors have become more adapted for usage at host temperatures in O157:H7 strains than in other STEC. However, studies with more strains from each non-O157:H7 serogroup will be needed to buttress this conclusion.
In addition to temperature and media differences, variations in cellulose production were also identified. Serotype O157:H7 strains showed little cellulose production and complete dependence on curli for CR binding at each temperature and media combination tested. On the other hand, mutational analyses of a small number of non-O157:H7 STEC strains demonstrated that both cellulose and curli played a role in CR binding. It was unclear why serotype O157:H7 strains failed to produce cellulose and whether such failures resulted in optimum temperature differences or lowered CR affinities compared with most non-O157 STEC. Previously, we determined that a high incidence of prophage insertions in mlrA and heterogeneous rpoS mutations limited csgD-dependent phenotypes in O157:H7 strains (Uhlich et al., 2013). Decreased cellulose expression may be an additional consequence. However, curli production at 37 °C could also imply utilization of a different sigma factor at higher incubation temperatures. It remains to be seen whether rpoD replaces rpoS at higher temperatures, altering csgD dependence on mlrA and establishing a different regulatory pathway for both cellulose and curli production. We also identified O45:H2 strains showing a medium-dependent, differential expression of cellulose that resulted in CR and calcofluor binding on TA but not on CRI. This was likely due to substrate utilization differences. We are currently conducting experiments to identify the components involved.
At 30 °C, a few native O157:H7 strains bound CR and formed small amounts of biofilm, but at 37 °C where CR affinity was greatest, biofilm was not formed, except for the traces by PA46. In contrast to the native strains, control strain OR, which overexpresses csgD from a mutant promoter, produced high levels of biofilm at all temperatures. It is possible that differences in curli expression alone may have resulted in the drastic biofilm differences between the native strains and strain OR at 37 °C, given that hyperexpression of a regulatory factor from a mutant promoter may cause unique phenotypic presentations. However, it is unclear why native strains with low curli expression at 30 °C produced more biofilm than the same native strains producing even more curli at 37 °C. Clearly, other factors besides curli production contributed to the differential biofilm phenotypes. Our studies showing no bcsA/C-dependent calcofluor staining of the serotype O157:H7 strains at either temperature indicate that cellulose did not contribute to the biofilm differences in these strains.
Unlike the O157:H7 strains, many non-O157 strains produced abundant biofilms. Cellulose and curli both affected CR binding, but there was significant individual strain variation in the phenotypic consequences of curli and/or cellulose gene deletions, especially at various temperatures. Gualdi et al. (2008) found that cellulose and curli expression had variable effects on biofilm formation in a nonpathogenic E. coli strain. Low curli-expressing strain MG1655 was made biofilm proficient by overexpressing plasmid-encoded csgD. Cellulose was suppressed at 37 °C but expressed at 30 °C due to temperature-regulated adrA expression. Biofilm formation was suppressed at 30 °C when cellulose blocked curli adhesion, but was generated at 37 °C when cellulose was absent. In the absence of curli expression, cellulose acted as a weak adhesin, indicating that cellulose can both promote or repress biofilm formation, depending on the level of curli expression. Clearly, the regulation of curli and cellulose, their roles in biofilm formation, and their interactions postexpression are complex and contribute to high phenotypic variability among strains. More studies will be needed to understand their role in the STEC.
In this study, we defined certain media and temperature effects on CR affinity and biofilm formation that will aid in defining the optimum conditions for identifying these characteristics in different clinically important STEC strains. In addition, we described differences in cellulose production among STEC serotypes and identified serogroup O45 STEC strains with medium-dependent cellulose production. This study also indicates that the CR-binding phenotype may be a reasonable predictor of ambient-temperature biofilm formation in strains of serotype O157:H7, whose cellulose production is low. However, in certain other STEC serogroups, co-expression of curli and cellulose at different levels leads to high phenotypic variability that may confound the correlation between curli expression and biofilm formation.
We thank John Phillips (ARS, USDA) for assistance with statistical analyses and Carlos Gamazo (University of Navarra, Navarra, Spain) for the gift of Salmonella Enteritidis strains 3934, 942, and 1170/97.
G.A. Uhlich and C.-Y. Chen contributed equally to this report.
USDA is an equal opportunity provider and employer. Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture.