A kit for the investigation of live Escherichia coli cell adhesion to glycosylated surfaces

Abstract

A combination of microtiter plate functionalization techniques and two facile bacterial adhesion inhibition assays form a flexible toolbox for the investigation of bacterial adhesion mechanisms on glycosylated surfaces.

---

Bacterial infections constitute a major global health problem, especially threatening the health of young children.¹ The most common serious neonatal infections involve bacteraemia, meningitis, and respiratory tract infections.² Key pathogens in these diseases are Escherichia coli, Klebsiella sp., Staphylococcus aureus and Streptococcus pyogenes. To cause infection, bacteria often need to adhere to target cells and to colonize the glycosylated cell surface. For the attachment to cells, most bacteria depend on the expression of specialized adhesive organelles, which are hair-like, 1–2 µm long and ∼7 nm wide protein structures on the bacterial cell surface. They are referred to as fimbriae (or pili). Much studied examples include P fimbriae and especially type 1 fimbriae, which are common throughout the Enterobacteriaceae³ and provide uropathogenic E. coli (UPEC) with the ability to bind to carbohydrate receptors in the urinary tract.⁴

In order to investigate the mechanisms of carbohydrate-specific bacterial adhesion, convenient testing systems are required. For in vitro-studies with type 1 fimbriated E. coli a classic hemagglutination assay employing guinea pig erythrocytes⁵ or an ELISA (enzyme-linked immunosorbent assay) on mannan-coated 96-well microtiter plates has been described.⁶ To study the complex and highly multivalent scenario of type 1 fimbriae-mediated adhesion of live E. coli to carbohydrates more in depth, assaying in a microtiter plate formate can be improved in two prime aspects: (i) systematic modification of the microtiter plate surface employing tailor-made glycosides would allow us to study the parameters of carbohydrate specificity more precisely; and (ii) advancement of assaying bacterial adhesion would result in easier, faster and more reliable testing.

Here, both techniques, which are ideally required to facilitate evaluation of carbohydrate-specific adhesion of E. coli, were elaborated and applied. Tailor-made carbohydrate-decorated microtiter plates were prepared and utilized to determine bacterial adhesion in an optimized assay employing a GFP-tagged E. coli strain. When an ELISA is used to measure bacterial adhesion, the procedure comprises the addition of a monoclonal antibody, followed by treatment with a horseradish peroxidase (HRP)-conjugated secondary antibody and subsequent staining with the HRP co-substrate ABTS [2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid)] to allow an optical density (OD) readout.⁶ One general drawback of the ELISA technique is the need of the primary monoclonal antibody, which has to be produced exclusively for one specific application in most of the cases. A second disadvantage is that the binding event is detected indirectly. Here, the ELISA detection protocol was advanced by biotinylation of type 1 fimbriated E. coli⁷ according to a known procedure⁸ to allow their detection relying on the well-investigated biotin–streptavidin system.⁹ Streptavidin exhibits a total of four binding sites for biotin binding with an equilibrium constant of 10¹⁵ mol⁻¹. Thus, biotin-labelled bacteria can be practically anchored to a HRP–streptavidin conjugate through the very strong biotin–streptavidin interaction and subsequently detected after ABTS oxidation (Fig. 1, left). To allow direct detection of bacterial adhesion, bacteria were sought that contain a promptly detectable moiety. As fluorescing bacteria ideally meet this demand, green fluorescent protein (GFP)¹⁰ was transfected into E. coli. The GFP-tagged strain PKL1162 was constructed by introduction of the plasmid pPKL174 into strain SAR18¹¹ and applied in binding assays on microtiter plates. In this case readout of bacterial adhesion could be performed directly with a fluorescence intensity reader at 485 nm (Fig. 1, right). Thus, incubation with peroxidase-conjugated reagents and the following staining steps could be omitted.

Fig. 1 Two advanced methods for detection of bacterial adhesion on a carbohydrate-coated surface. On the left: biotin-labelled, type 1 fimbriated E. coli are allowed to bind to a mannoside-exposing surface or an inhibitor in solution (not shown); adhered bacteria are incubated with streptavidin–HRP conjugate and detected by enzymatic staining. On the right: the same adhesion assay is performed with green fluorescent protein (GFP)-tagged bacteria that can be detected directly by fluorescence readout.

It was important to evaluate how the three different adhesion assays, ELISA, biotin–streptavidin-based and GFP-based assay, respectively, compare in analogous set-ups. Thus, adhesion inhibition studies were performed employing serial dilutions of well-known inhibitors of type 1 fimbriae-mediated bacterial adhesion, namely methyl α-D-mannoside (MeMan) and p-nitrophenyl α-D-mannoside (pNPMan) (Table 1). In this assay, the mannoside-equipped microtiter plate surface competes with mannosides in solution for binding to the bacteria, so that bacterial adhesion is partly inhibited. From the resulting inhibition curves, IC₅₀-values were deduced that reflect the inhibitor concentration, which causes 50% inhibition of bacterial binding to the polysaccharide mannan. While the obtained absolute IC₅₀-values differ significantly depending on which of the three assays was used, very similar RIP-values were determined in all cases.¹² From a series of independent tests it was evident that reproducibility in the case of the biotin–streptavidin- and the GFP-based assay was much higher than in the case of the ELISA. The assay using GFP-tagged bacteria is especially fast, taking less than 1.5 hours on pre-prepared microtiter plates. Moreover it is cheaper, when compared to other assays, which require an enzyme conjugate and staining and it is highly reproducible and more robust than the ELISA. Almost no experience is needed to produce reliable results, in particular employing the GFP-based assay, which works almost like a “kit”.

Table 1 Comparison of the three assays in testing bacterial adhesion to the polysaccharide mannan

	ELISA	Biotin–streptavidin-based assay^a	GFP-based assay
a Control experiment: biotin showed no effect as inhibitor of E. coli adhesion. b As mannan-coated and blocked microtiter plates can be stored at 4 °C, the time required for their preparation was not counted in; biotin labelling of E. coli has to be performed directly before use and was therefore included into ‘time needed’.^1 c The GFP-based assay proved to be successful with inexperienced users. d Average values are given, deduced from at least ten independent assays. e RIP = relative inhibitory potency, referenced on MeMan with IP ≡ 1; RIP-values allow comparison of different inhibitors regardless of the performed assay.
Time needed^b	3.5 h	6.5 h	1 h, 25 min
Detection	Indirect	Indirect	Direct
Robustness^c	+	++	+++
IC50MeMan^d/µM	5370	1210	6530
IC50pNPMan^d/µM	119	12.5	71.5
RIP^d^,^epNPMan	45	97	91

---

Then, chemical modification of polystyrene microtiter plates was performed. This allows us to investigate the effect of various parameters of ligand presentation, such as, i.a., influence of the ligation chemistry, conformational flexibility of the linker, and especially the nature of the aglycon moiety of the immobilized mannoside. Thus it was interesting to compare two opposing polystyrene surfaces, matrix 2 having aromatic units bound to the surface and matrix 4 in which aliphatic linkers were employed (Scheme 1). In both cases, active ester-activated microtiter plates were incubated with amino-functionalized mannosides, p-aminophenyl α-D-mannoside 1 for fabrication of 2, and the aminothiaalkyl mannoside 3, which was employed to make 4.† The phenyl mannoside 1 was obtained by catalytic reduction of pNPMan,¹³ and the alkyl mannoside 3 was derived from allyl α-D-mannoside¹⁴ in an almost quantitative radical addition reaction with cysteamine hydrochloride.¹⁵†

Scheme 1 Fabrication of mannoside-functionalized assay matrices: the amino-functionalized mannosides 1 and 3 were immobilized onto active ester-modified microtiter plates; the squaric acid-modified mannoside 5 was reacted with amino-functionalized plates; (a) carbonate buffer, pH 9.6; (b) carbonate buffer, pH 8.2.

The newly developed assay using GPF-tagged bacteria was applied for inhibition adhesion measurements on both matrices, 2 and 4. Again, MeMan and pNPMan were used as inhibitors of type 1 fimbriae-mediated adhesion of E. coli. Highly reproducible binding curves were obtained in all cases, from which IC₅₀-values were deduced (Table 2). Interestingly, almost equal concentrations of MeMan were needed to inhibit 50% of bacterial adhesion in both cases. Also inhibition of bacterial adhesion with pNPMan led to very similar IC₅₀-values. In other words, bacterial adhesion to surface 2 and to surface 4 is comparably strong.¹⁶ This is surprising because it is known that mannosides with an aromatic aglycon can undergo favorable ππ-stacking interactions with two tyrosine residues at the entrance of the carbohydrate recognition domain of the type 1 fimbrial adhesin FimH.¹⁷ This is reflected in lower IC₅₀-values when glycosides such as pNPMan are used as inhibitors of type 1 fimbriae-mediated bacterial adhesion to mannan in comparison to MeMan (cf.Table 1).

Table 2 Inhibition of bacterial adhesion on fabricated microtiter plate surfaces: IC₅₀-values of standard mannosides were compared in independent assays

Inhibitor	Matrix 2^a		Matrix 4^a		Matrix 6^b
	IC50^c/µM	RIP	IC50^c/µM	RIP	IC50^c/µM	RIP
a With amide-linked surfaces 2 and 4 the GFP-based assay was used. b With the squaric-acid-linked surface 6 biotin-labelled bacteria were applied. c IC50-values are average values from three independent experiments.
MeMan	20 000	1	20 700	1	300	1
pNPMan	500	40	313	66	0.87	345

---

In order to understand the observation made with matrices 2 and 4, mannoside 3, which was used for fabrication of 4, was tested as inhibitor of bacterial adhesion to mannan using GFP-tagged E. coli. This revealed a RIP-value of 38 referenced to MeMan. With these values available, two surprising results could be deduced: (i) it became clear that type 1 fimbriae-mediated bacterial adhesion to an amide-ligated molecular layer of 3 (matrix 4) is more effective than adhesion to mannan (more inhibitor is needed to effect 50% of inhibition to the respective surface); (ii) apparently, pNPMan is a considerably better inhibitor of type 1 fimbriae-mediated bacterial adhesion than mannoside 3, when applied in solution [RIP(pNPMan) = 91, RIP(3) = 31], whereas the corresponding surfaces 2 and 4 show less difference in their potency to bind these bacteria (cf.Table 2). These results exemplify the discrepancy between ligand binding in solution versus binding to immobilized ligands.

To test the influence of the ligation chemistry, which is used for matrix fabrication, in addition to active ester-modified plates, amine-coated plates were employed and functionalized with the squaric acid monoester mannoside 5 (Scheme 1), which was derived from 3. According to a known squaric ester-ligation protocol,¹⁸ matrix 6 was obtained. Because, the transparent amino-functionalized plates are incompatible with fluorescence readout, matrix 6 had to be tested with the biotin–streptavidin-based assay. Therefore, the measured data cannot be directly compared to the IC₅₀-values, which were obtained with the GFP-tagged bacteria. However, because it is known how the different assays compare among each other (cf.Table 1), it can be stated that bacterial adhesion to matrix 6 is significantly weaker than to matrix 4 as it can be inhibited by relatively low inhibitor concentrations (Table 2). This finding indicates that the ligation chemistry, which is used for derivatization of the microtiter plates, can have an important influence on its adhesive potential and this observation has to be considered in future studies.

In summary, the two new bacterial adhesion assays together with tailor-made functionalization of microtiter plates form a tool box for facile in-depth studies of bacterial adhesion to carbohydrate surfaces. It was shown that GFP-tagged bacteria allow fast and reliable measurement of fimbriae-mediated bacterial adhesion. Biotinylation of bacteria, on the other hand, is convenient to determine adhesion of any bacterial strain, which was not GFP-tagged before, and, in addition, biotin-labelled bacteria can be detected on transparent microtiter plates. A combination of adhesion measurements and adhesion inhibition assays, using ligands in solution as well as immobilized on the plate surface, has led to the conclusion that type 1 fimbriae-mediated binding to ligands in solution differs significantly from binding to an immobilized ligand layer. This finding has implications in cell–cell interactions which depend on carbohydrate recognition on the glycosylated cell surface. It can be envisioned that the herein exemplified methodology will be used to learn more about the multi-facetted scenario of bacterial adhesion to more or less complex surfaces, resembling some of the complexity of cell adhesion to the glycocalyx in vivo.

Notes and references

1. E. K. Mulholland and R. A. Adegbola, N. Engl. J. Med., 2005, 352, 75–77.
2. D. Osrin, S. Vergnano and A. Costello, Curr. Opin. Infect. Dis., 2004, 17, 217–224.
3. (a) K. Ohlsen, T. A. Oelschlaeger, J. Hacker and A. S. Khan, Top. Curr. Chem., 2009, 288, 109–120; (b) P. Klemm and M. Schembri, Int. J. Med. Microbiol., 2000, 290, 27–35; (c) M. A. Mulvey, Cell. Microbiol., 2002, 4, 257–271.
4. A. L. Kau, D. A. Hunstad and S. J. Hultgren, Curr. Opin. Microbiol., 2005, 8, 54–59.
5. T. K. Lindhorst, C. Kieburg and U. Krallmann-Wenzel, Glycoconjugate J., 1998, 15, 605–613.
6. S. Kötter, U. Krallmann-Wenzel, S. Ehlers and T. K. Lindhorst, J. Chem. Soc., Perkin Trans. 1, 1998, 2193–2200.
7. For this application the bacterial strain HB101 pPKL4 was used, exposing exclusively type 1 fimbriae; M. A. Schembri, G. Christiansen and P. Klemm, Mol. Microbiol., 2001, 41, 1419–1430.
8. I. Ofek, H. S. Courtney, D. M. Schifferli and E. H. Beachey, J. Clin. Microbiol., 1986, 24, 512–516.
9. (a) E. P. Diamandis and T. K. Christopoulos, Clin. Chem., 1991, 37, 625–636; (b) M. Wilchek, E. A. Bayer and O. Livnah, Immunol. Lett., 2006, 103, 27–32.
10. R. Tsien, Annu. Rev. Biochem., 1998, 67, 509–544.
11. (a) A. Reisner, J. A. J. Haagensen, M. A. Schembri, E. L. Zechner and S. Molin, Mol. Microbiol., 2003, 48, 933–946; (b) Plasmid pPKL174 contains the same fim gene cluster as in HB101 pPKL4, which is required for type 1 fimbriae assembly and expression. The chromosome of strain SAR18 contains the GFP gene, controlled by a constitutive promoter.
12. IC₅₀-values measured in the biotin–streptavidin-based assay were always rather low, falling 4 to 6 times below the values obtained in the other two assays. This decreased signal intensity could be due to biotin labelling within the mannoside binding pocket of the type 1 fimbrial lectin FimH or in its close proximity. This should diminish the ability of the bacteria to adhere to the mannan surface.
13. R. Roy, F. D. Tropper, T. Morrison and J. Boratynski, J. Chem. Soc., Chem. Commun., 1991, 536–538.
14. F. M. Winnik, J. P. Carver and J. J. Krepinsky, J. Org. Chem., 1982, 47, 2701–2707.
15. (a) P. B. van Seeventer, J. A. L. M. van Dorst, J. F. Siemerink, J. P. Kamerling and J. F. G. Vliegenthart, Carbohydr. Res., 1997, 300, 369–373; (b) M. Dubber and T. K. Lindhorst, Carbohydr. Res., 1998, 310, 35–41; (c) C. D. Heidecke and T. K. Lindhorst, Chem.–Eur. J., 2007, 13, 9056–9067.
16. The results were interpreted on the assumption that tighter adhesion of E. coli cells to a particular surface corresponds to higher IC₅₀-values.
17. (a) S. D. Knight and J. Bouckaert, Top. Curr. Chem., 2009, 288, 67–107; (b) O. Sperling, A. Fuchs and T. K. Lindhorst, Org. Biomol. Chem., 2006, 4, 391–392.
18. S. Spjut, M. Pudelko, M. Hartmann and M. Elofsson, Eur. J. Org. Chem., 2009, 349–357.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental procedures, adhesion-inhibition assays, NMR and mass spectra. See DOI: 10.1039/b922525k

---