You are on page 1of 27

MICROREVIEW

DOI: 10.1002/ejoc.201100407

The Bacterial Lectin FimH, a Target for Drug Discovery Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion
Mirja Hartmann*[a] and Thisbe K. Lindhorst*[a]
Keywords: Lectins / Cell adhesion / Carbohydrates / Cluster effect / Inhibitors
Adhesion is a prerequisite for bacteria to colonize cell surfaces. To accomplish cellular adhesion, many bacteria use carbohydrate-specific lectins, which are expressed as part of capillary protein appendages expanding from their surface, called fimbriae or pili. For bacteria, colonization of cell surfaces offers advantageous conditions to persist and multiply. For the host, however, bacterial colonization can be affiliated with severe health problems such as inflammation. Therefore, to combat bacterial adhesion and inflammatory diseases, investigation of the molecular and biophysical details of the relevant lectincarbohydrate interactions is important. Understanding molecular carbohydrate recognition can lead to the development of high-affinity inhibitors of bacterial lectins. That way, interfering with the bacterial attachment to surfaces proves the vision of an antiadhesion therapy, among others, against uropathogenic E. coli (UPEC). One of the most important and best investigated bacterial lectins is the mannose-specific protein FimH, which is expressed on the tips of type 1 fimbriae. During the last 30 years, many natural as well as synthetic mannosidic ligands of FimH have been designed and tested for their inhibitory potencies. We report key results and comment on key problems and perspectives of this research.

Contents
1. Introduction 2. Early studies on the carbohydrate specificity of type 1 fimbriae 3. Multivalent glycomimetics as inhibitors of type 1 fimbriae-mediated bacterial adhesion 4. Assays to test antagonists of type 1 fimbriae-mediated bacterial adhesion [a] Otto Diels Institute of Organic Chemistry, Christiana Albertina University of Kiel Otto-Hahn-Platz 3/4, 24098 Kiel, Germany Fax: +49-431-880-7410 E-mail: mhartmann@oc.uni-kiel.de tklind@oc.uni-kiel.de

5. Structure of the type 1 fimbrial lectin FimH 6. Rational design of carbohydrate ligands for the type 1 fimbrial lectin FimH 7. Special approaches to the inhibition of mannose-specific bacterial adhesion

1. Introduction
There is a class of proteins that reversibly bind carbohydrates. These proteins are neither carbohydrate-specific enzymes nor antibodies. Today, the understanding of glycoscientists is that this type of proteins has evolved to recognize specific carbohydrates and select them for binding to

Mirja Hartmann, M.Sc., was born in 1983. She studied Molecular Life Science at the University of Erlangen-Nuremberg with main focus on Drug Discovery, Molecular Biology and Molecular Synthesis. For her master thesis, she developed strategies for the synthesis of glycoconjugates and their biological testing with Prof. Mikael Elofsson at Ume University in Sweden. In 2007 she received her Master of Science degree from the University of Erlangen-Nuremberg and joined the research group of Prof. Thisbe Lindhorst at Christiana Albertina University of Kiel for her Ph.D. studies. Currently, she is working on biochemical assays and biophysical setups to test type 1 fimbriae-mediated bacterial adhesion.

Thisbe Lindhorst is Full Professor at the Faculty of Mathematics and Natural Science of Christiana Albertina University of Kiel since 2000. She studied chemistry at the Universities of Munich and Mnster, received her diploma in chemistry/ biochemistry in 1988 and her Ph.D. in Organic Chemistry in 1991 at the University of Hamburg in the group of Prof. J. Thiem. After a postdoctoral stay at the University of British Columbia with Prof. S. Withers, she worked on her habilitation and became Private Docent in 1998 at the University of Hamburg. In 1997 she was a Visiting Professor at the University of Ottawa in Canada in Prof. R. Roys laboratory. Since 2000, she holds a chair in Organic and Biological Chemistry in Kiel. Her scientific interests are in the field of synthetic organic chemistry and in biological chemistry, and in particular in glycochemistry and glycobiology. Current research is focussed on the study of cell adhesion to glycosylated surfaces. She is the author of the textbook Essentials of Carbohydrate Chemistry and Biochemistry.

Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3583

MICROREVIEW
form carbohydrateprotein complexes. While, apparently, carbohydrateprotein complexation has no obvious consequence for the structural integrity of the complexed and subsequently released saccharide ligand, such molecular recognition of carbohydrates can trigger biological response of different kinds in all types of organisms, such as cell recognition, signalling and cell adhesion. Lectins The carbohydrate-complexing proteins called lectins comprise a remarkable variety of structures, folds, functions and occurrence. The history of lectins goes back to the end of the 19th century.[1] At that time, work with extracts from seeds of the castor tree (Ricinus communis) led to the discovery that certain plant proteins have the ability to agglutinate erythrocytes. The agglutinating protein isolated from Ricinus communis was called ricin. Rapidly, many more proteins with the same feature were discovered in various plants and were called phytoagglutinins, phytohemagglutinins and somewhat later hemagglutinins. Soon they were employed in immunological studies, leading to the finding that specific hemagglutinins react specifically with human red blood cells of certain blood groups (A, B, or O).[2] In 1954, the term lectin was proposed by Boyd and Shapleigh for these and other antibody-like substances with blood-group-specific agglutination properties.[3] The authors quoted the name to be derived from the Latin lego, to choose or pick out; (in fact the Latin verb legere means to gather, choose, collect, select, pass through, read, and its perfect passive participle is lectus, meaning selected and picked out, or read out.) In the 1970s Sharon and Lis[4] suggested the term lectin as a general name for all proteins of nonimmune origin that possess the ability to agglutinate erythrocytes and other cell types. It was discovered very early that the interactions of lectins with cells can be inhibited by specific carbohydrates, mono- or oligosaccharides. Consequently, it was concluded that lectins are specific saccharide-binding proteins. This was first shown for the well-known plant lectin concanavalin A.[5] As a consequence, the first attempt to classify the fast-growing class of lectins was based on their carbohydrate specificity. Today, lectins are being classified on the basis of their structural features and especially the relatedness of their carbohydrate-binding sites, which are called carbohydrate recognition domains, in short CRDs.[6] It was Drickamer who pointed out the importance of sequence homologies in the carbohydrate recognition domain motifs of proteins for their carbohydrate-binding properties.[7,8] The carbohydrate specificity of lectin CRDs has remained a key criterion for the assessment of lectins so far.[9] Fimbrial Lectins of Bacteria Early in the history of lectins, it had become evident that, besides numerous plant lectins, most, if not all, organisms express this class of proteins, for very different purposes,
3584
www.eurjoc.org

M. Hartmann, T. K. Lindhorst

however. Already during the first half of the 20th century, it was discovered that also many bacteria, in particular those of the Enterobacteriaceae family, have the ability to agglutinate erythrocytes. This hemagglutination activity of bacteria is almost always associated with the presence of multiple filamentous protein appendages projecting from their surface, called fimbriae (from the Latin word for thread) or pili (from the Latin word for hair) (Figure 1). Fimbriae are adhesive organelles, comprising lectin subunits, which mediate carbohydrate-specific adhesion to cell surfaces as well as cell agglutination. Thus, bacteria utilize the sugar decoration of cells, the glycocalyx, to colonize the cell surface, wherever cells are in contact with the outside environment, as for example in the case of epithelial cells.

Figure 1. Transmission electron micrograph of a type 1 fimbriated E. coli cell. Some hundreds of the adhesive organelles cover the bacterial surface.

Bacterial Adhesion and Diseases Initial bacterial adhesion to cell surfaces is in turn amplified, leading to the development of a well-organized superstructure, called a biofilm. Biofilm formation is highly advantageous for the colonizing microbes. It facilitates firm and irreversible adhesion to a surface, interlinking bacteria of different species, which produce a carbohydrate mucus to maintain the biofilm.[10] Through this exopolysaccharide layer,[11] bacteria can achieve chemical communication and profit from favourable coordination (quorum sensing). Biofilm formation can form the basis of a beneficial symbiosis between a microorganism and its host such as in the gut, where Escherichia coli (E. coli) bacteria produce vitamin K in the large intestine. However, as soon as microorganisms invade another habitat or only slightly change their genes, disorders such as inflammation, or even apoptosis or uncontrolled cell growth can arise.[12] Typically, bacterial colonization is accompanied by infectious diseases, and this constitutes a major global health problem.[13] In addition, biofilm formation on medical devices and implants frequently causes complications[14] and significantly contributes to the pathogenesis of implant-related infections. Among the most prevalent inflammatory diseases that are caused by pathogens are urinary tract infections. The predominant pathogens in this case are uropathogenic E.
Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion

coli (UPEC). UPEC can attach to specific niches in the urinary tract by virtue of the interaction of their surface fimbriae with carbohydrate receptors on the luminal surface of the bladder epithelium, leading to bladder infection and inflammation (cystitis).[15]

1E,[24] but are simply referred to as type 1 fimbriae in most studies, as well as in this account. Because of the importance of type 1 fimbriae for bacterial adhesion, numerous studies on their carbohydrate specificity were undertaken, in order to design effective inhibitors of bacterial adhesion to mucosa.

Carbohydrate Antagonists and Antiadhesion Therapy Adhesion to the surface of the host cell is a prerequisite for colonization and successful reproduction conditions especially for enteric, oral and respiratory bacteria. Adherence protects bacteria from being swept away by the normal cleansing mechanism operating on mucosal surfaces such as urinary flow. Thus, successful adhesion increases the ability of the bacteria to colonize epithelial cells, multiply and eventually also invade the host. As bacterial adhesion is mediated by interactions with cell surface carbohydrates, the intriguing idea to prevent bacterial colonization by carbohydrate inhibitors of adhesion arises. Thus, carbohydrates could be developed as antiadhesive drugs according to a strategy against pathogens that is an alternative to antibiotics. An antiadhesive therapy[1618] could help to fight especially those bacteria with multi-antibiotic resistance, constituting an increasing problem in medicine.[19] It can be considered as rather unlikely that bacteria develop resistances against antiadhesive drugs. Hence, studies on the carbohydrate-specificity of bacterial lectins have been largely motivated by the importance of carbohydrate-mediated bacterial adhesion in infection, and this research has led to the vision to prevent bacterial adhesion by suitable carbohydrate antagonists of natural lectin ligands. There are many different lectins of Gramnegative bacteria known with various carbohydrate specificities. In E. coli, P fimbriae (specific for Gal1,4Gal), S fimbriae (specific for Neu5Ac2,3Gal), and type G fimbriae (specific for GalNAc) are found, as well as the mannosespecific type 1 fimbriae, which are one of the most abundant surface structures both in pathogenic and nonpathogenic Gram-negative bacteria.[20]

2. Early Studies on the Carbohydrate Specificity of Type 1 Fimbriae


Interest in type 1 fimbriae started relatively early. Approximately since the 1970s it was known that hemagglutination mediated by type 1 fimbriae can be inhibited by mannose, methyl -d-mannoside (MeMan) and mannan.[20] Then, evidence that type 1 fimbriae are major virulence factors of UPEC motivated a number of more in-depth investigations on the carbohydrate-specificity of type 1 fimbriae. Ofek and colleagues showed that the carbohydrate specificity of the E. coli type 1 fimbrial lectin can be studied quantitatively by examination of the inhibitory potency of mono- and oligosaccharides on the agglutination of yeast cells by bacteria. In this agglutination inhibition assay, the concentration of MeMan that is required to inhibit 50 % of yeast agglutination by E. coli 364 (a strain which predominately expresses type 1 fimbriae) was determined to range between 0.15 and 0.40 mm.[25] A number of oligosaccharides were tested as inhibitors of yeast agglutination and compared to the inhibitory potency of MeMan, which was arbitrarily set to 1 (Figure 2). Mannobiosides Man1,2Man (1), Man1,3Man (2) and Man1,6Man (3), as well as the branching trimannoside Man1,6[Man1,3]Man1-OMe (4), formed a first group of tested saccharides, of which only trisaccharide 4 showed a significantly higher inhibitory potency than MeMan [RIP(4)MeMan = 10.5]. In 2006, very similar saccharides were investigated as inhibitors of the adhesion of type 1 fimbriated E. coli to mannan in an enzymelinked immunosorbent assay (ELISA).[26] Lindhorst and co-workers employed allyl mannobiosides Man1,2Man1OAll (5), Man1,3Man1-OAll (6), Man1,6Man1-OAll (7) and Man1,4Man1-OAll (8), and the allyl trimannoside Man1,6[Man1,3]Man1-OAll (9), and largely found relative inhibitory potencies similar to those reported by Ofek and his colleagues. However, the 1,3-linked disaccharide 6 performed significantly better in the ELISA, giving a RIP value of 11 (Figure 2). It should be noted, however, that the standard deviation in this assay was determined as 7, which was rather high. Also the allyl trisaccharide 9 [RIP(9)MeMan = 20] was a better inhibitor than the analogous methyl trisaccharide 4 [RIP(4)MeMan = 10.5], though in different assays. When Ofek and co-workers tested more complex oligosaccharides, the strict specificity of the type 1 fimbrial lectin for -mannosides was confirmed. However, it was also shown that the way an -mannosyl ligand is scaffolded on a particular oligosaccharide is crucial. While for trisaccharide 10, Man1,3Man1,4GlcNAc, a RIP value of 21 was determined, its isomer 11, Man1,6Man1,4GlcNAc, had
www.eurjoc.org

Type 1 Fimbriae Type 1 fimbriae serve as extremely efficient adhesion tools for bacteria inhabiting diverse environments, including biotic and abiotic surfaces.[21] They are uniformly distributed on Enterobacteriaceae, commonly between 100 and 400 fimbriae per cell. Their length varies between 0.1 to 2 micrometer, their width is approximately 7 nm (Figure 1). Fimbriae are present in at least 90 % of all known UPEC strains, which are the main cause of urinary tract infections in humans, and they have been shown to be important virulence and pathogenicity factors.[22] Type 1 fimbriae mediate agglutination of guinea pig red blood cells in an -mannose-inhibitable manner and are responsible for bacterial binding to a wide range of glycoproteins carrying one or more N-linked high-mannose type oligosaccharide.[23] Type 1 fimbriae of the E. coli type have been classified as type
Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3585

MICROREVIEW

M. Hartmann, T. K. Lindhorst

Figure 2. Structures of oligosaccharides tested as inhibitors of mannose-specific adhesion of E. coli together with their relative inhibitory potencies based on the reference mannoside methyl -d-mannoside (MeMan). IP: inhibitory potency; RIP: relative inhibitory potency (with IP of MeMan 1); SD: standard deviation. [a] Values from inhibition of yeast agglutination with E. coli 346;[25] [b] values from inhibition of adhesion of type 1 fimbriated E. coli HB101 pPKL4 to the polysaccharide mannan, measured by ELISA;[26] [c] Oligomannose-9 and -3 are named according to a suggestion made in the literature,[28] where they were tested as ligands for the fimbrial lectin FimH (vide infra). 3586
www.eurjoc.org 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Eur. J. Org. Chem. 2011, 35833609

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion

a RIP of 0.7, thus performing worse than simple MeMan as inhibitor of mannose-specific yeast agglutination. Furthermore, while the branched oligomannosides 12 and 13 were very good inhibitors, 14 and 15 were significantly weaker. The data obtained in yeast agglutination assays were confirmed with guinea pig erythrocytes a little later.[27] Some of these results, obtained in the 1980s, can be rationalized today on the basis of more modern testing results with oligosaccharides 16 and 17, which have been termed oligomannose-9 and -3, respectively, by Bouckaert et al. (Figure 2).[28] At the time, however, researchers had no idea about the structure of the carbohydrate recognition domain of the type 1 fimbrial lectin and had to rely on models, which were deduced on the basis of the determined testing results. The model of the type 1 fimbrial carbohydrate combining site, which was finally suggested by Ofek and Sharon, was therefore based on the findings obtained with a series of oligomannosides and, in addition, based on structureactivity relationships obtained with a series of simple mannosides with varying aglycon moieties (Figure 3).

patterns on the aromatic ring showed a favourable effect on the inhibitory potency of the respective mannnoside. For p-nitrophenyl-o-chlorophenyl -d-mannoside (pNoClMan), for example, a RIP value of approximately 720 was determined. As mentioned earlier, this and some other testing results of the early days can today be decoded on the basis of the crystal structure of the fimbrial lectin (vide infra). Finally, Sharon and Ofek established a model for the carbohydrate combining site of the type 1 fimbrial lectin and postulated that it should correspond to the size of a trisaccharide (having three subsites) with a hydrophobic binding region adjacent to the carbohydrate-binding site, which has a strict specificity for the -configuration of the bound mannoside. Today, crystallographic studies have partially approved Sharons model, while the three-subsite model could not be confirmed so far. Nevertheless, at present there seems to be some mystery hidden behind type 1 fimbriae-mediated binding of mannosides that has not been unravelled yet. Therefore, Sharon s model might contain a relevance, which is not reflected in the X-ray analysis of the type 1 fimbrial lectin, called FimH, and thus has not yet been appreciated and remains underestimated.

3. Multivalent Glycomimetics as Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion


After the early investigations with natural mannosides and oligosaccharides, a series of studies has employed multivalent glycomimetics as inhibitors of type 1 fimbriaemediated bacterial adhesion. The multivalent presentation of -d-mannosides was anticipated as a promising approach to high-affinity (avidity, respectively) inhibitors of mannose-specific bacterial adhesion based on multivalency effects that had been observed in other carbohydratelectin interactions. Characteristically, lectins can contain not only one but two or more carbohydrate-binding sites, and their interaction with carbohydrates is therefore often multivalent. Multiple lectincarbohydrate contacts can lead to agglutination on one hand, but on the other hand, they also may lead to increased affinities as well as to a higher specificity of the interaction. When the affinity of a lectin to a carbohydrate ligand is increased by multivalent carbohydrateCRD contacts, the term avidity is used to describe the strength of the overall interaction. To achieve an avidity effect by employing multivalent glycomimetics as lectin antagonists is especially attractive, as carbohydrate binding of lectins is typically associated with only moderate affinity. Thus, the typical stabilities of many lectinmonosaccharide complexes are low and lie in the millimolar to high micromolar range.[30] Indeed, studies on the inhibition of type 1 fimbriae-mediated bacterial adhesion and hemagglutination have revealed that rather high concentrations of saccharides are required for effective inhibition. In the 1970s, Y. C. Lee and coworkers approved weak binding of galactose and GalNAc binding to the asialoglycoprotein receptor (ASGPR). However, when multivalent glycomimetics were employed, they
www.eurjoc.org

Figure 3. The nature of the aglycon moiety of simple -d-mannosides is decisive for their potency as inhibitors of type 1 fimbriaemediated bacterial adhesion.[29] RIP: relative inhibitory potency (with inhibitory potency of MeMan 1).

Strikingly, mannosides with an aromatic aglycon, such as p-nitrophenyl -d-mannoside (pNPMan) and 4-methylumbeliferyl -d-mannoside (MeUmbMan), were shown to exceed the inhibitory potency of MeMan significantly (Figure 3),[29] performing better than all oligosaccharides that had been examined earlier. In addition, certain substitution
Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3587

MICROREVIEW
discovered that a linear increase in the number of monosaccharide epitopes, presented in a specific glycomimetic, led to a logarithmically increased avidity to the ASGPR. Lee and co-workers, at that time, concluded that this finding strongly suggests that lectins possess more than one carbohydrate-binding site, hence are bi- or even multivalent. The resulting effect in binding multivalent carbohydrates they called the cluster effect, which today is often referred to as multivalency effect.[31] Inspired by Y. C. Lees results, chemists have strived after multivalent glycomimetics to achieve high-affinity carbohydrate ligands for lectins. Multivalent glycomimetics have been designed and synthesized along a large variety of architectures, and this field has been extensively reviewed.[32,33] It has, however, not been easy to simply reproduce the results Y. C. Lee observed in the case of carbohydrate binding to ASGPR with other lectins. Today it can be concluded that multivalency effects, observed in carbohydrateprotein interactions, can result from quite different biological mechanisms and according to quite different thermodynamic and kinetic conditions.[3438] Thus, the effect of variable multivalent glycomimetics in a particular testing system can not necessarily be compared in the sense of quantitative structureactivity relationships (QSARs). Multivalent glycopolymers,[3942] for example, might interact with a lectin according to fundamentally different models than a multivalent glycocluster, which is based on a carbohydrate core, for instance.[26] Among multivalent glycomimetics, glycodendrimers are an especially promising class of molecules, which have been evaluated as antiadhesives. Their synthesis and testing results as inhibitors of UPEC has recently been reviewed in detail.[43] Mannose Glycoclusters with Noncarbohydrate Core In 1998, the effect of multivalency on the inhibition of type 1 fimbriae-mediated bacterial adhesion was studied for the first time,[44] by using a hemagglutination inhibition assay (vide infra). Branched and hyperbranched amines were employed as multivalent scaffold molecules and reacted with glycosyl isothiocyanates to achieve di-, tri-, tetra-, and hexavalent -mannose clusters 18 to 21 by thiourea bridging (Table 1). The inhibitory potencies determined with these glycoclusters and glycodendrimers[4547] turned out to be not strictly valency-dependent.[44] Whereas divalent molecule 18 and tetravalent cluster 20 performed 31 and 39 times better than the reference inhibitor MeMan, respectively, trivalent cluster 19 had a 106-fold higher inhibitory potential. When the determined inhibitory potencies are considered as valency-corrected values (RIPvc), trivalent glycocluster 19 was superior over all other multivalent glycomimetics tested in this series. In valency correction, the inhibitory potency determined for a specific glycocluster is divided by the number of clustered mannosyl moieties. Thus, a RIP value of 106, determined for a trivalent glycocluster, results in a RIPvc of 35. The hexavalent molecule 21, for example, was shown to have the same RIP value as the trivalent cluster 19, but its valency-corrected inhibitory potency, RIPvc, is only 18.
3588
www.eurjoc.org

M. Hartmann, T. K. Lindhorst

Besides the valency of a glycocluster, the structure of the aglycon or scaffolding unit was shown to play a crucial role for lectin-binding properties. Trivalent cluster 22, which, apart from its O-glycosidic linkage, offers great structural similarity to thiourea-bridged glycocluster 19 [RIP(19)MeMan = 106], performed only 10 times better than MeMan.[44] Thus, thiourea bridging of mannosyl residues was especially effective in inhibiting mannose-specific bacterial adhesion. This was once again confirmed in a study with tetravalent clusters 23 and 24,[48] which are based on tetraazamacrocycle scaffolds. While cyclen 23 showed a 195-fold increased inhibitory potency in a hemagglutination assay, when compared to MeMan, the more flexible cyclam analogue 24 exceeded the inhibitory potency of MeMan by 780-fold. These findings showed that tri- and tetravalent glycoclusters perform particularly well as inhibitors of type 1 fimbriae-mediated bacterial adhesion, and this led to the development of a series of further small glycoclusters and cluster mannosides (Table 1). A new ELISA (vide infra) was developed to determine the antiadhesive properties of such triand tetravalent cluster mannosides, all O-glycosidically attached to the respective noncarbohydrate cores.[49] In addition, the idea was born to cluster mannosides via the 6position of the sugar ring. This concept was considered in order to combine a favourable aglycon effect with the success of trivalent clustering of mannosidic ligands. Thus, in the late 1990s mannoside donors (instead of the classical mannosyl donors) were designed, each with a functionalized 6-position to achieve a library of analogous cluster mannosides with varied aglycon moiety. For clustering peptide coupling to branched triacids, or thiourea bridging to branched triamines was utilized. This approach to high-affinity ligands for type 1 fimbrial lectin led to cluster glycosides such as 25 and 26, which were quite poor inhibitors, however. Nevertheless, direct comparison of two structurally similar C-6-clustered inhibitors with aliphatic [RIP(25)MeMan = 0.7] and aromatic aglycon [RIP(26)MeMan = 8] moieties showed that the aromatic character of the aglycon affected binding. On the other hand, the anomerically linked trivalent cluster mannnoside 27 tested in the same assay and lacking an aromatic moiety, produced a RIP of 90. Overall, the concept of mannoside clustering via the 6position of the sugar ring cannot be recommended. Any good testing results with this type of 6-modified cluster mannosides must be considered false positive results[49] in light of the FimH X-ray structure,[50] which indicates that complexation of mannosides within the FimH CRD is not possible once a scaffolding unit is connected to the 6-position of the sugar ring (cf. Figure 6). The first crystal structure of the type 1 fimbrial lectin FimH was published in 1999.[50] It proved that this lectin is monovalent and possesses just one CRD, thus favourable effects with multivalent ligands could not necessarily be expected on the basis of this information. In spite of this, many more multivalent cluster mannosides were prepared in the following years up to the present, and they were tested as inhibitors of type 1 fimbriae-mediated bacterial
Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion

adhesion and even the isolated lectin FimH. Such research was motivated and reasoned by numerous experimental fin-

dings, which showed that especially tri- and tetravalent cluster mannosides are indeed potent inhibitors of type 1 fim-

Table 1. A selection of representative mannose glycoclusters with noncarbohydrate core and their testing results. As many of the structures are rather complex, -d-mannosyl residues have been simplified and represented as six-membered heterocycles shaded in grey, here and likewise in all other following tables.

Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjoc.org

3589

MICROREVIEW
Table 1. (Continued)

M. Hartmann, T. K. Lindhorst

3590

www.eurjoc.org

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Eur. J. Org. Chem. 2011, 35833609

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion Table 1. (Continued)

[a] SD = standard deviation, included in brackets if literature-reported. [b] RIP = relative inhibitory potency, based on MeMan with IPMeMan 1. [c] RIPvc = valency-corrected RIP (rounded down). [d] = RIP based on d-mannose with IPMan 1.
Eur. J. Org. Chem. 2011, 35833609 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.eurjoc.org

3591

MICROREVIEW

M. Hartmann, T. K. Lindhorst

Table 2. A selection of representative carbohydrate dendrimers and carbohydrate-centred cluster mannosides and their testing results.

3592

www.eurjoc.org

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Eur. J. Org. Chem. 2011, 35833609

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion Table 2. (Continued)

[a] SD = standard deviation, included in brackets if literature-reported. [b] RIP = relative inhibitory potency, based on MeMan with IPMeMan 1. [c] RIPvc = valency corrected RIP (rounded down). [d] K abbreviates l-lysine as branching element.

briae-mediated bacterial adhesion and good antagonists of FimH. Thus, a collection of structurally related tri- and tetravalent cluster mannosides with a C3-aliphatic aglycon, 28 to 31, were investigated.[51,52] Here it was again confirmed that a linear increase in valency can result in a much more than linear enhancement in affinity, as observed by comparison of 28 and 31. However, the attempt to improve binding to FimH by incorporation of an aromatic moiety adjacent to the clustered mannosides, such as in the case of 28 and 29,[51] was not successful. Trivalent cluster mannoside 30 performed approximately 6 times better than its analogue 29 with a phenyl ring incorporated at its focal point. On the other hand, the importance of the nature of the scaffolding unit for the affinity of clustered mannosides was also revealed. Four clusters with ethylene glycol (EG) spaEur. J. Org. Chem. 2011, 35833609

cers were tested (Table 1): trivalent clusters 32 and 33,[53] both yielding a high RIP of 1063, and tetravalent molecules 34 and 35,[54] which had a 100 times lower RIP in the same assay system. Thus, in spite of the fact that the same type of linkers were used for clustering in all four cases, their inhibitory behaviour differed significantly. In recent years, also additional chemistries such as click chemistry,[55] Sonogashira coupling[56] and squaric acid conjugation[57] were employed for the synthesis of cluster mannosides, leading, for example, to triazole-linked clusters 36, 37, 41 and 42, alkyne-linked multivalent mannosides 38 to 40 and clusters such as 43, linked by squaric acid. With the isolated lectin FimH on hand, such tetraand hexavalent cluster mannosides were tested in surface plasmon resonance (SPR) experiments (vide infra).[58] The
www.eurjoc.org

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3593

MICROREVIEW
obtained RIP values vary from 157 for cluster 36 to 41 for cluster 37, and even more drastically, from 4889 for tetravalent cluster 38 to 8 for the similar tetravalent 39. This tremendous discrepancy points out that the distance between the mannoside ligand and the aromatic moiety as part of the spacer is extremely critical for binding efficiency to FimH. An interesting observation was made, when the valency of alkyne-linked clusters was changed while the principal architecture of these clusters was maintained. When tetravalent molecule 39 was expanded to hexavalent structure 40, the relative inhibitory potencies remained the same. The triazole-linked cluster mannosides 41 and 42, on the other hand, led to a significant improvement, such that the RIP value of 41 was increased by a factor of 4.7 when compared to that of 36. The synthesis of cluster 42[59] was inspired by the structure of the highly potent heptyl -d-mannoside (cf. Table 3) and indeed seems to be a very promising candidate for inhibition of bacterial adhesion. Glycocluster 42 was shown to exhibit a relative inhibitory potency of 2670 (based on mannose) when tested in a hemagglutination inhibition (HAI) assay. An impressive example for a cluster effect in FimH binding was shown by applying mannosylated lysine-based dendrimers.[60] With growing generations, the relative inhibitory potencies of the di-, tetra-, octa- and hexadecavalent dendrimers 44, 45, 46 and 47 increased from 455 over 2000 and 3571 to 11111, respectively (Table 2). Though the avidity did not grow logarithmically, it rose to a much greater extent than it would by a linear increase with respect to the increasing valency.

M. Hartmann, T. K. Lindhorst

Carbohydrate-Centred Cluster Mannosides and Carbohydrate Dendrimers Carbohydrate dendrimers were obtained by employing a glycosyl donor as the branching element.[61] This synthesis starts from a 3,6-di-O-allyl mannoside (Scheme 1), which can be converted into a respective diol by hydroboration or radical addition of 2-mercaptoethanol. Then, mannosylation with a 3,6-di-O-allylated mannosyl donor delivers the next generation of carbohydrate dendrimers. When hydroboration is employed to produce the diol acceptor for glycosylation, mannose moieties are linked by a C3 spacer, such as in the case of 48 and 49 (Table 2). When 2-mercaptoethanol was added upon the allylic double bonds, more flexible dendritic structures such as 50 resulted. Interestingly, thioethers like 50 could be oxidized to the respective sulfones (51), to increase the hydrophilicity of the spacers. Adhesion inhibition assays showed that, relative to the first generation precursor 48, a duplication of mannoside residues in 49 led to a fourfold higher inhibitory potency [RIP(48)MeMan = 10; RIP(49)MeMan = 42]. Secondly, again the importance of spacer properties for proteinligand interactions was apparent. Sulfone-bridged cluster 51 surpassed the inhibitory potency of MeMan by 25-fold, whereas its less polar analogue 50 was 110 times a better inhibitor than the reference mannoside. Finally, carbohydrate-centred cluster mannosides were introduced and tested.[26,62] They were designed to closely resemble the chemical nature of natural oligosaccharides. For steric reasons, C3 spacers were included between the

Scheme 1. Carbohydrate dendrimers such as 50 (Table 2) can be formed in an iterative synthetic sequence employing a glycosyl donor as the branching element.[61] 3594
www.eurjoc.org 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Eur. J. Org. Chem. 2011, 35833609

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion

glucose core and the scaffolded mannoside ligands. It was anticipated that, once the synthetic sequence to glucosecentred cluster glycosides is established, it can be transferred to others, for example, di- and trisaccharide core glycosides. Thus, a small library of carbohydrate-centred clusters such as 52, 53[26] and 54[62] (Table 2) could be produced, which gave RIP values of 180 for pentavalent cluster 52, 230 for octavalent 53 and 190 for dodecavalent mannoside 54.[62] Strikingly, these values are all in the same range, pointing out that the enhancement in molecule size and valency had no significant effect on inhibition properties of the respective cluster mannosides in this case. Reasoning of Multivalency Effects in Carbohydrate Binding of Type 1 Fimbriated Bacteria The multivalency effects observed with many different types of mannoside clusters as inhibitors of type 1 fimbriaemediated adhesion of E. coli have not been fully understood. Generally, a cluster effect can be explained with the clustering of multiple CRDs,[31,34,37,63] but the type 1 fimbrial lectin FimH possesses only one CRD. Interfimbrial clustering of FimH CRDs might occur with large enough molecules. In an average glycocluster having about ten CC single bonds between two mannoside residues, the distance between these two ligands is in the range of 1.5 nm. As the bacterial fimbriae measure about 7 nm in diameter, binding of one cluster molecule to more than one fimbrial lectin at a time is very unlikely with cluster mannosides such as those depicted in Table 1. Intramolecular binding is a much more sensible explanation of avidity enhancement observed with multivalent cluster glycosides in this adhesion system. Hence, several putative carbohydrate-binding sites on FimH, in addition to the CRDs known from X-ray studies, were suggested (vide infra).[64] Multiple binding to more than one carbohydrate-binding site on FimH could account for strengthened binding of multivalent ligands. A most obvious explanation for the observed multivalency effects is the elevated effective concentration of mannoside residues in the proximity of the carbohydrate recognition domain when a multivalent ligand is employed. Quick rebinding of a mannoside ligand is therefore statistically favoured once a bound ligand is released from the CRD. Another probable explanation for multivalency effects in inhibition of type 1 fimbriae-mediated adhesion could be that cooperative effects between ligand and protein play a role. For example, intramolecular preorganization of free ligands could arise when the first ligand of a multivalent glycomimetic is bound to the lectin. This induced spatial organization could enhance binding of the other ligands (positive cooperativity) or reduce their affinity (negative cooperativity). Moreover, binding of one ligand might induce a change in lectin conformation, which enforces the subsequent binding event of the next mannoside moiety of a cluster mannoside. This mechanism would correspond to an allosteric activation of the protein achieved by a multivalent ligand.
Eur. J. Org. Chem. 2011, 35833609

Several studies indicated that especially potent clustered inhibitors of type 1 fimbriae-mediated bacterial adhesion were tri- or tetravalent. To explain the variant potencies of these, the chemical properties of the used linkers are decisive. Tuning of the linker chain length and its conformational flexibility makes the exposed carbohydrates more or less easily accessible for lectin binding. Furthermore, the conformational stability of the chosen linker in water (or buffer), the surrounding solvent in biological systems, is an important property. Very hydrophilic spacers may dissolve the carbohydrate cluster so well that binding to a partially hydrophobic protein would be disfavoured by enthalpy considerations. In contrast, hydrophobic linkers might lead to supramolecular assemblies of the clusters, in course lowering their accessibility for lectin binding. This is a reasonable explanation of the finding that higher valencies in many cases did not lead to drastically increased inhibitory potencies of cluster mannosides and mannose glycodendrimers. Higher numbers of mannoside residues on a scaffolding core molecule of limited size does not necessarily mean better ligand availability. Individual mannoside moieties might be clustered too closely, so that enough space for the lectin to bind multiple copies is not available.

4. Assays to Test Antagonists of Type 1 Fimbriae-Mediated Bacterial Adhesion


The development of new drugs is typically based on reliable estimation of QSAR studies with potential leads. In addition to rational ligand design, which is based on the knowledge of the structural properties of the drug target, in vitro methods to screen for potential lead compounds are indispensable tools. Therefore, any test system has to be particularly suited for the biological question that is investigated. In an in vitro assay, which is typically used for QSAR studies, three major components determine the situation during testing: the biological receptor or target, the native ligand and the test compound, which can be an agonist or an antagonist. If this three-compartment model is translated to the screening for potent inhibitors of type 1 fimbriae-mediated bacterial adhesion (Figure 4), the bacterial lectin FimH is the receptor, the native ligand is approximated by the glycosylated surface (e.g. human bladder epithelium cells, or mannan as its mimic), and the test compound is the putative inhibitor of bacterial adhesion, typically an -d-mannoside derivative. All possible mutual interactions of the three components have to be considered to influence the outcome of an assay. In this system, a number of parameters are important, and they are discussed in the following. (1) Presentation of the lectin: How the lectin is applied in an assay plays a crucial role for the outcome. Three states of complexity of the receptor system have been employed. The lectin can be applied as recombinantly expressed, isolated protein or protein conjugate. Sheared-off bacterial type 1 fimbriae can be used, or an assay can be performed with whole bacteria. Even without any knowledge of the
www.eurjoc.org

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3595

MICROREVIEW

M. Hartmann, T. K. Lindhorst

Figure 4. Three different types of interactions play key roles in FimH-mediated bacterial adhesion to mammalian cells. In the development of assays to test and optimize putative adhesion inhibitors, all interactions and their mutual interplay have to be considered: the bacterial lectin FimH can either bind to a diversely glycosylated surface, a cell or cell mimetic (interaction I) or to an inhibitor in solution, a multivalent -d-mannoside in this case (interaction II); additionally, an interplay of the carbohydrates on the surface with those in solution (interaction III) has to be considered. Each change in one of the shown interactions will influence the others, and likewise changes in the conditions of the binding process have an impact on the whole system. Thus, the results of any FimH binding assay or adhesion inhibition assay are highly dependent on the setup of the applied in vitro test system.

lectin structure, it can easily be understood that the binding conditions can differ a lot in these three different setups. (2) Glycosylated surfaces: Not only the presentation of the lectin FimH but also the glycosylated surface used in the assay influences the results. The ligand system can be applied in many different stages of complexity. Whole cells can be used to test adhesion at native interfaces. On the other hand, by using synthetic surfaces it is possible to tune the complexity and structural features of a glycosylation pattern in great detail. In this way, structural refinement allows to test for very specific patterns in a ligand surface, if needed. (3) Presentation of the inhibitor: Structural details of the test compound that enhance or lower its affinity are the object under investigation and can naturally not be varied to tune an assay. But still, they play a role in the way a FimH antagonist is applied. As mentioned earlier, inhibitors that are clustered can be much more effective than their respective monovalent precursors. Immobilized on some surface or other rigid device, the inhibitors properties can change and thus lead to an inhibitory effect that differs from the properties of the inhibitor in solution. In Vitro Assay Systems to Screen for Potent FimH Inhibitors Different types of assays have been developed to test the quality of a potential new inhibitor of type 1 fimbriae-medi3596
www.eurjoc.org

ated bacterial adhesion. The data that are generated for the same FimH ligand can vary significantly when obtained in different assay systems. To be able to understand some of the observed effects and apparent discrepancies, it is useful to categorize the tests applied according to the conditions under which inhibition of adhesion is investigated. Assays in a Three-Dimensional Setup The first assays that were performed to screen the interactions between type 1 fimbriated bacteria and eukaryotic cells were aggregation assays.[25] In the 1980s Sharon and colleagues could show that the agglutination of yeast cells by type 1 fimbriated bacteria, such as, among others, E. coli and some species of Salmonella, can be inhibited by addition of -mannosides. This agglutination assay using bacteria and yeast cells is still used today.[65,66] The assay setup is rather close to physiological conditions regarding the glycosylated eukaryotic cell as well as in view of the fact that the lectin-bearing bacteria can move freely in all three dimensions. A closely related testing system that has nearly replaced the yeast aggregation assay due to its higher sensitivity is the hemagglutination inhibition assay.[44,49,58,67,68] In this assay, an inhibition titre (IT) is determined, which reflects the concentration of a tested inhibitor that is needed to prevent type 1 fimbriated bacteria from agglutinating guinea pig erythrocytes. As direct IT values generally differ
Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion

much from one screen to the other even if the same bacterial strain is used, relative IT values are commonly calculated, which are referenced to a known inhibitor tested in parallel. In most cases, the reference glycoside is methyl d-mannoside (MeMan). In 2002, along with hemagglutination inhibition data of type 1 fimbriated E. coli, Lee and co-workers detected the binding event by radioactivity assays.[60] Mannosides coupled with 125I-labelled human serum albumin (HSA) were incubated together with E. coli bacteria and different dilutions of the test substance. After centrifugation, the remaining radioactivity of the E. coli cells was an inverse measure for the inhibitory effect of the tested carbohydrate ligand. As the tested inhibitors have to compete with the mannosylated HSA surface for bacterial binding, half maximal inhibitory concentration (IC50) values can be derived by plotting the % inhibition against the inhibitor concentration on a logarithmic scale to obtain a sigmoidal dose response curve. The IC50 value of a test substance reflects the inhibitor concentration at which 50 % of the binding is inhibited. In dose response curves the IC50 value is the point of inflection. A second 3D approach to determine inhibitory potencies of FimH ligands was reported in 2001.[69] Human cells (neutrophilic granulocytes) and a dilution of potential adhesion inhibitors were added to a solution of biotinylated type 1 fimbriae. After incubation and washing, fluorescein-labelled streptavidin facilitated the readout by fluorescence-activated cell sorting (FACS). In 2002, it was shown that the binding of mannose-encapsulated gold nanoparticles (m-AuNP) to type 1 fimbriated E. coli cells is inhibited by addition of MeMan. M-AuNPs were thereby introduced as tools for the imaging of FimH mediated bacterial binding in transmission electron microscopy (TEM).[70] Recently, mannosylated nanoparticles, which can be used in imaging applications were introduced again. In this study, the adhesion of E. coli to mannose-functionalized hematite (iron oxide) nanoparticles was visualized by TEM.[71] Detection has been performed with 3H-labelled mannose. After incubation of isolated FimH together with 3H-mannose and a competing inhibitor, the assay mixture was filtered, and the remaining radioactivity could be read out. When using different inhibitor dilutions, Kd values could be derived for the tested substances.[72] In recent work, two additional three-dimensional assays were applied to test the binding between FimH and potential carbohydrate ligands. One of these is the investigation of the binding of fluorescent-dye-labelled mannosides to isolated FimH in a fluorescence polarization (FP) assay.[68] In this assay, the fluorophore on the FimH ligand is excited by polarized light, so it consequently emits polarized light. The more freely the molecule can move in solution the more disturbed the polarization of the emitted light. Bound ligands partially freeze and thus move more slowly; therefore, the polarization of the emitted light increases upon binding, which can be measured. As ligand binding is not competitive in this setup, the inhibitory effect of a substance is not detected directly. Nevertheless, the binding event per se can be monitored precisely. Thus, the characteristic measure for a tested
Eur. J. Org. Chem. 2011, 35833609

ligand is not an IT, but the half maximal effective concentration, abbreviated EC50. The EC50 can again be deduced from a dose response curve, a plot of the gained effect against the corresponding ligand concentration applied. Isothermal titration calorimetry (ITC) is a different and well-known test method that was first utilized for the detection of FimH binding mannosides by Bouckaert and colleagues in 2011.[73] Mannosylated ligands can be titrated into a FimH solution to measure the temperature changes upon saccharide addition and calculate a Kd value to characterize the binding of the ligand. Measurement of inhibitory effects was not performed, but binding constants for the ligandFimH interaction were deduced. Assays on Surfaces Besides the three-dimensional test systems, a number of assay setups have been published that depend on the immobilization of one binding partner and subsequent detection of the binding event to the manipulated surface. As only one of the binding partners is in solution and can thus move freely, the setups can be categorized as two-dimensional. The prototype of 2D assays is probably the classical ELISA that depends on the detection of a molecule or protein bound to a predefined surface by incubation with a specific antibody against the presumably surface-bound compound. Lindhorst and co-workers used microwell plates coated with mannan (mannose polymer) for ELISA to bind type 1 fimbriated E. coli cells.[49] These could be detected with an antibody against FimA, the predominant protein of the fimbrial shaft and a peroxidase-labelled secondary antibody. Upon addition of mannoside solutions in different concentrations, dose response curves of the inhibitory effect of the tested ligand against the applied concentration could be plotted, which led to IC50 values as characteristic for each tested ligand. As these ligands compete with the mannan-coated surface for FimH binding, a direct inhibitory potency is measured in this case. To be able to compare two such ELISA measurements with one another, a standard inhibitor has to be tested on the same test plate, so that the poorly comparable IC50 values from different experiments can be uniformly referenced. Different ELISA experiments have been performed by using whole E. coli cells,[26,44,48,5154,61,62,7476] type 1 fimbriae[77] and the isolated FimCH protein complex.[78] Assays similar to the ELISA method have been developed. They all share the competitive 2D binding situation, but the detection method is nonimmunological. By using type 1 fimbriated E. coli cells, two different competitive adhesion inhibition assay setups have been introduced.[79] These assays basically resemble the ELISA and were performed on the same mannan-coated test plates, but either the bacteria were biotinylated to be detected with a streptavidinperoxidase conjugate, or directly detectable self-fluorescing bacteria expressing the green fluorescent protein (GFP) were used. In addition to mannan-coated plates, covalently glycosylated assay surfaces were used to screen the affinity of bacteria to a surface-bound ligand. Both assays resulted in IC50 values for the tested substances. Covalent
www.eurjoc.org

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3597

MICROREVIEW
immobilization of saccharides on a test surface is also needed for current research and carbohydrate array fabrication. A competitive binding assay on mannosylated chips was published in 2004.[80] Type 1 fimbriated bacteria were stained with intercalating fluorescent dyes and subsequently, together with a reference mannoside solution, applied to a mannoside array to competitively bind to the microchip. The fluorescence readout was then used to quantify the surface-bound bacteria. A more flexible carbohydrate surface can be built up by self-assembled monolayer (SAM) formation on a gold surface. In 2002, an adhesion inhibition assay on mannosylated SAMs in a 96-well formate was performed.[81] The adhesion of type 1 fimbriated fluorescence dye-stained bacteria to the glycosylated SAM was shown to be inhibitable by addition of a mannoside solution. As this inhibition again turned out to be dependent on inhibitor concentration, IC50 values for the used inhibitors could be reported. Radioactively labelled bacteria were also used to detect binding to GP2 (a membrane glycoprotein) in a competitive manner.[82] To test the inhibitory potency of a glycopeptide from soybean, Guo and coworkers[83] used an adhesion inhibition assay, in which they let E. coli and Salmonella cells adhere to eukaryotic LoVo cells (colon carcinoma cells). This adhesion was inhibited by glycopeptides. To characterize the glycoconjugate GP2 as inhibitor, the remaining bacteria cells on the LoVo cell surface were counted. A different 2D test on human cells was performed in 2008.[28] A human urothelial cell line was grown in 12-well plates. Type 1 fimbriated E. coli were preincubated with mannoside solutions and then transferred to the cell surface. After incubation and washing, the bladder cells were lysed by addition of trypsin and the lysate was transferred to agar culture plates to determine the remaining amount of colony-forming units. Turning the setup upside down, the FimH CRD can also be immobilized to microtitre wells, to which the inhibitor dilution and a biotinylated polyacrylamide glycopolymer are applied and incubated, generating a competitive binding situation. By addition of peroxidase-coupled streptavidin, the amount of bound mannoside-functionalized polymer is detected. By plotting the corresponding dose response curves, IC50 values for the tested substances can be obtained.[84] Assays under Flow One parameter that plays an important role in physiological systems is the flow that the binding event has to withstand. Though in some of the reported assays the samples are shaken during incubation, the shear stress that the lectin is exposed to is not regulated. This is different in flow-regulated SPR measurements. For SPR, two different general setups have been applied. In the first SPR measurements with FimH, either an anti-FimH antibody or a BSAmannoside conjugate was bound to the Biacore sensor chip. When isolated FimH was coursed over the immobilized antibody together with a soluble adhesion inhibitor, only a small percentage of FimH could be attached to the surface, depending on the binding affinity between FimH and the
3598
www.eurjoc.org

M. Hartmann, T. K. Lindhorst

added inhibitor.[58,72] Recently, the antibody or BSA conjugate was replaced by low-molecular-weight mannosides, which were covalently attached to the Biacore chip.[73] When a mixture of FimH and soluble inhibitors mannosylated fullerenes in this case passes the glycosylated chip surface, the lectin can either bind to the Biacore chip and be detected or stick to the glycosylated fullerenes and be washed away. Both of these setups for SPR measurements resulted in Kd values for the FimHmannoside interaction. In most cases, noncovalent binding of proteinligand interactions is shorter-lived under shear stress. Such interactions are referred to as slip bonds.[85] Nevertheless, for some interactions of cells with surfaces, it was first proposed in 1988[86] that a critical applied tensile force can enhance the binding strength rather than weakening the adhesion.[85] FimH was the first bacterial protein for which such catch bonds were discovered.[87] Since 2010, the way bacterial lectin FimH tightens the binding to its sugar ligand under tensile force is understood in great detail. As long as the FimH protein is in a relaxed conformation, the N-terminal lectin domain and the C-terminal pilin domain (vide infra) are in close proximity, and the lectin domain is compressed. When shear force is applied, the two domains separate, and the lectin domain is elongated by 11 . This elongation process stretches the conformation of the -sheets of the lectin domain in such a way that the CRD constricts. This, in turn, tightens the binding of the ligand to the binding site. The same mechanism is in effect when a FimH truncate is isolated in pure form. As the FimHtr isolates lack the attachment to the C-terminal pilin domain, the lectin domain is frozen in its elongated high-affinity state.[87] This finding is particularly valuable for the interpretation of the results gained in different adhesion or binding assays with type 1 fimbriated bacteria under flow. In addition, the affinity state in which the lectin FimH is applied in different assays has a great impact on the measured binding strength. Considering these recent findings about the catch bond mechanism of lectin, many testing results with the various FimH antagonists might appear in different light and deserve an updated interpretation. The evaluation of testing results can be complex and difficult. Even the interpretation of inhibitory potencies within one assay system can already be challenging. In many cases, it can be very helpful to keep in mind the three-compartment model of an assay (Figure 4). To be able to compare the effect of different compounds, it is essential to test a reference compound in the same assay and batch, so that all results can be referenced reliably. In FimH binding studies, MeMan (Table 3) or pNPMan (Table 4) are mostly used as reference glycosides. In particular, these relative values are needed to compare the results of different assays. As the conditions under which binding is measured vary from system to system, the same inhibitor might perform quite differently in two apparently similar assays. Illustrative examples are the highly diverging values measured for pNPMan in many different test systems (Table 3). Usually, the calculated relative values of the inhibitory potencies are in good agreement, though.
Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion

5. Structure of the Type 1 Fimbrial Lectin FimH


Crystal structures of carbohydrate-binding proteins provide valuable information about the modes of interactions that mediate carbohydrate recognition.[88,89] It has been frequently observed in X-ray diffraction studies with lectins that a carbohydrate ligand is complexed by a well-defined array of hydrogen bonds, involving hydrogen-bond donor and acceptor groups of the ligand and the side chains of asparagine or glutamine residues, carboxylate groups from aspartates or glutamates, OH groups in serine side chains and NH-groups from lysine, tryptophan or histidine residues of the lectin CRD. Water molecules can mediate these hydrogen bonds, and in some cases divalent metal ions such as Ca2+ or Mn2+ are additionally involved in carbohydrate binding. Also, sandwiching of the carbohydrate ligand between aromatic amino acid side chains is an important binding motif, leading to a relevant contribution of CH interactions in carbohydrate binding.[90,91] The type 1 fimbrial lectin is called FimH and is assembled at the tips of type 1 fimbrial composite structures according to the chaperone/usher pathway.[50,92] Nine fim genes are required for fimbriae assembly. The tip-fibrillar structure contains the mannose-specific adhesin FimH, whereas the fimbrial rod is mainly composed of FimA subunits (Figure 5). Type 1 fimbriae are constructed by a donor strand exchange (DSE) process, in which the Ig fold of every subunit is completed by an amino-terminal extension from the following subunit.[21]

FimH in 1999,[50] in a complex with the FimC chaperone. Cyclohexylbutanoyl-N-hydroxyethyl-d-glucamide (CHEGA), which had been added for crystallization, was found to be bound to the CRD of FimH, where it adopts a conformation that relatively closely resembles the structure of a mannose pyranoside ring. Thus, it was revealed that FimH has two domains, a pilus domain, FimHP, and a lectin domain, FimHL. The N-terminal mannose-binding lectin domain FimHL comprises residues 1156, and the C-terminal pilin domain, which is used to anchor the adhesin to the pilus, comprises residues 160279. There is one single CRD to be seen in the FimH crystal structure, which is located at the tip of the lectin domain. It is capable of accommodating a mannose monosaccharide in its -configuration (Figure 6).

Figure 6. A mannosidic ligand is complexed within the FimH CRD such that the aglycon of an -d-configured mannoside points towards the entrance of the binding pocket. This allows terminal mannose residues on complex oligosaccharides to be complexed by FimH. -d-Mannosides cannot be complexed for steric reasons.

Figure 5. The type 1 fimbrial rod is assembled from different Fim proteins, which attach to each other by donor strand exchange (DSE). The fimbrial lectin, FimH, terminates the fimbrial fibre at its tip, and the most abundant protein constituent, FimA, attaches the fimbriae to the bacterial outer membrane.

FimH is a 29 kDa protein with a length of 279 amino acid residues. Its exact structure has remained unknown for long, because FimH cannot be crystallized in pure form as it is missing a strand to complete its fold. In addition, it is proteolytically degraded when it is expressed alone. Knight and co-workers reasoned that, according to the two-component system that is required for fimbriae assembly (according to the DSE process), the C-terminal part of FimH requires interaction with another protein. For fimbriae assembly, FimH is complexed with the chaperone FimC, and this led to the attempt to stabilize FimH by addition of FimC. This idea delivered the first crystal structure of
Eur. J. Org. Chem. 2011, 35833609

In 2002, the crystal structure of the FimC/FimH chaperoneadhesin complex bound to its physiologically relevant ligand -d-mannose was published, and the amino acids that are important for mannose complexation were identified in great detail.[78] Mannose was found to be buried in a deep and negatively charged pocket. The mannose ring makes ten direct hydrogen bonds to the FimH binding site, and in addition indirect water-mediated hydrogen bonds are formed. All hydroxy groups of the sugar ring, other than the anomeric position, interact extensively with the CRDs of FimH, in particular with residues Phe1, Asn46, Asp47, Asp54, Gln133, Asn135, Asp140 and Phe142 (Figure 7). The mannose-binding pocket is surrounded by a hydrophobic ridge comprising Ile13, Tyr48, Ile52, Tyr137 and Phe142. The side chains of Tyr48 and Tyr137 are positioned such that they form a gate-like structure that has been named the tyrosine gate.[21] Thus, mannoside ligands with an aromatic aglycon, such as pNPMan, can establish interactions with the tyrosine gate flanking the entrance of the FimH CRD (Figures 7 and 8), leadwww.eurjoc.org

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3599

MICROREVIEW

M. Hartmann, T. K. Lindhorst

Figure 7. Both graphics show the amino acid residues in an orientation revealed by X-ray diffraction analysis (PDB ID: 1KLF),[78] depicted in ball-and-stick form. Left: The bottom site of the FimH CRD with docked p-nitrophenyl -d-mannoside (pNPMan). The CRD contains the N-terminal Phe1 amino acid of the protein. Prominent hydrogen bonds between the carbohydrate-binding site and the ligand are depicted as grey dotted lines. Right: The amino acid residues at the entrance of the FimH CRD. The aromatic side chains of Tyr48 and Tyr137 form interactions with docked pNPMan, leading to an increased affinity of the ligands with an aromatic aglycon.[93]

ing to increased affinities. Thus, the early findings with ligands such as pNPMan or MeUmbMan (Figure 3) can now be understood on the basis of the FimH structure.

Figure 8. Structures of the CRD of the bacterial lectin FimH crystallized with the tyrosine gate (residues Tyr48 and Tyr137) in an open or closed conformation. The conformation of the tyrosine gate depends on the complexed ligand. The open gate structure (PDB ID: 1KLF) was obtained with FimH in complex with mannose (top),[78] while the closed gate conformation (PDB ID: 1UWF) arose from complexation of BuMan (bottom).[72] The FimH receptor structure is depicted as a Connolly surface, and to highlight the tyrosine gate it is meshed in each case. 3600
www.eurjoc.org

In 2005, another crystallographic study was published, which employed two FimH proteins named FimHtr1 and FimHtr2 from two different bacterial strains.[72] These FimHtr proteins were truncated to only the FimH lectin domain, built up by residues 1158. In both cases, although no sugar was included in the crystallization setups, it turned out that butyl -d-mannoside (BuMan) was bound in the CRD. It was shown that this mannoside originated from the LB (Luria-Bertani) medium used to grow the bacteria during expression of the protein. The butyl moiety of bound BuMan extends out of the mannose-binding pocket towards Tyr48 and Tyr137, making van der Waals contacts to both tyrosine rings and Ile52. In case of FimHtr2 the Tyr48 and Tyr137 side chains were found in an almost parallel orientation as in the earlier FimC/FimH structures.[50,78] In the FimHtr1 structure, on the other hand, the parallel orientation of the Tyr48 ring is prevented and instead it is packed edge-to-face with Tyr137 (Figure 8). Thus, there is a conformational flexibility of the tyrosine gate at the entrance of the FimH CRD, which has implications on ligand design and the interpretation of testing results (vide supra). In 2008, a truncated version of FimH, FimHtr comprising residues 1158, was finally crystallized in a complex with a natural ligand, oligomannose-3 (Figure 2).[28] It was shown that the 1,3-linked mannose residue is complexed within the CRD and that the Man1,4GlcNAc1, 4GlcNAc portion of oligomannose-3 interacts with an extended region of the binding site. Interestingly, for the central mannose unit, a strong stacking interaction with the aromatic ring of Tyr48 can be observed, whereas Tyr137 interacts with the mannose-bound GlcNAc moiety. Thus, the FimH tyrosine gate can be likewise utilized by oligosacEur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion

charide ligands for tighter binding as it has been observed for synthetic ligands with an aromatic aglycon moiety (Figure 3, Table 4).

6. Rational Design of Carbohydrate Ligands for the Type 1 Fimbrial Lectin FimH
As the structures of the type 1 fimbrial lectin FimH and its CRD are known today from several crystallographic studies,[28,50,68,72,78] a rational, computer-aided design of high-affinity ligands to develop inhibitors of bacterial adhesion to mucosal surfaces should be greatly facilitated.

Only recently, a number of such more rational studies have been published, leading to three classes of FimH ligands with relatively high affinity: (i) long-chain alkyl mannosides, (ii) mannosides with variously substituted aromatic aglycon moieties, and (iii) mannosides with extended aglycon moieties.

Long-Chain Alkyl -D-Mannosides Hydrophobic Interactions with the Tyrosine Gate In case of the crystallization studies with FimHtr published in 2005,[72] exogenous BuMan was found to be com-

Table 3. A selection of representative alkyl -d-mannosides and their testing results.

[a] SD = standard deviation, included in brackets if literature-reported. [b] RIP = relative inhibitory potency, based on MeMan with IPMeMan 1.
Eur. J. Org. Chem. 2011, 35833609 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.eurjoc.org

3601

MICROREVIEW
plexed within the CRD. As even repeated dialysis could not remove BuMan from the protein, it was concluded that this simple alkyl mannoside undergoes a stable interaction with the protein. In fact, van der Waals contacts of the butyl aglycon to the phenyl rings of the lectins Tyr48 and Tyr137 and to the side chain of Ile52 were found. Today, it can be reasoned that in its complexed conformation, BuMan can mimic the hydrophobic face of an oligosaccharide such as oligomannose-3 (Figure 2).[28] To screen the ligand properties of further mannosides with a more extended hydrophobic alkyl aglycon, a series of alkyl -mannosides was synthesized and tested in SPR experiments and a displacement assay using 3H-labelled mannose (vide supra).[72] As indicated in Table 3, the affinity of the various alkyl -dmannosides for FimH increases with the length of the aglycon alkyl chain. The only irregularity in this study was a comparably high RIP value of 440 for heptyl -d-mannoside (HeptMan), determined in SPR studies. The strong binding to FimH is assumed to result from interactions with the heptyl aglycon. These high affinities of HeptMan for FimH were approved in other studies.[68,84] However, the high value of the MeMan-based RIP(HeptMan)MeMan = 440, determined in the initial SPR study,[72] has not been confirmed in any other test system to date (cf. Table 3).

M. Hartmann, T. K. Lindhorst

Monochloro-substituted mannosides, oClPMan, mClPMan and pClPMan did not perform as well as pNoClPMan when tested in a hemagglutination inhibition assay. Also in a cyano-substituted series of mannosides, oCNPMan, mCNPMan and pCNPMan, it was confirmed that mannnosides with o-substituted phenyl aglycon performed best as hemagglutination inhibitors, whereas the p-substituted compounds were the worst inhibitors (Table 4). The high inhibitory potency of MeUmbMan, which had been reported by Sharon three decades ago, was confirmed in three independent assay systems.[68,72,74] From the X-ray structure of FimH it can be seen that the aromatic system in MeUmbMan perfectly fits between the phenyl rings of the tyrosine gate to build up strong stacking interactions. A similar effect could also be achieved with other fused ring systems, such as naphthalene (for 55). Mannoside 55 has a fivefold higher inhibitory potency than pNPMan. Interestingly, in the applied assay, mannosides with an ester-substituted thiophene ring fused to the phenyl aglycon (mannoside 56) produced a RIP value of 16 (based on pNPMan), just as MeUmbMan, though lacking the extended aromatic system of MeUmbMan. Mannosides with Extended Aromatic Aglycon Moieties Interactions beyond the Tyrosine Gate? When MeUmbMan was docked into the FimH binding site in its closed-gate conformation,[72] the pyrone part of the conjugated system was located exactly between the two phenyl rings of the CRD tyrosine gate. This complexation mode explains the high affinity measured for MeUmbMan (Table 4). Extension of the aromatic system by another planar substituent with the ability to establish additional or tighter interactions with the tyrosine gate seemed a promising approach towards low-molecular-weight mannosides with very high affinities for FimH. Thus, a series of mannosides with varied biphenyl aglycon were synthesized and tested in hemagglutination experiments,[68] a selection of which is shown in Table 5. The biphenyl analogue of pNPMan, 57, and its m-nitro-substituted analogue 58 supported the initial assumption. The inhibitory potencies of these biphenyl mannosides exceeded that of pNPMan by a factor of 16 for 57, a value that was also found for MeUmbMan in the same assay (Table 4). The m-nitro-substituted biphenyl mannoside 58 exceeded the inhibitory potency of pNPMan even by a factor of 62. To identify a substitution pattern on biphenyl mannosides that is optimal for FimH binding, the influence of o-, m- and p-cyano-substitution (compounds 59 to 61) was investigated.[68] Other than in the case of phenyl mannosides (cf. Table 4), the inhibitory potency could be best enhanced, namely by a factor of 31 relative to pNPMan, by the introduction of a cyano group into the m-position of the biphenyl system (60). CN residues in the o-position led to a 16-fold inhibitory potency, while the p-substituted product 61 had a RIP value of 4 (Table 5). The same inhibitory effect could be achieved, when the biphenyl system was replaced by a benzophenone aglycon (62).
Eur. J. Org. Chem. 2011, 35833609

Mannosides with Aromatic Aglycon Moieties Interactions with the Tyrosine Gate It has been known since the 1980s that aromatic aglycon moieties can enhance the affinity of the respective mannosides for FimH by a factor of 600 or more.[29] Later, on the basis of the FimH crystal structures, such findings could be rationalized. Thus, stacking interactions of aromatic rings with the amino acid side chains of Tyr48 and Tyr137 of the FimH protein improve the affinity of a carbohydrate ligand. Also oligosaccharides, the natural ligands for FimH, can interact with this tyrosine gate by means of their hydrophobic sites.[28] To further improve the affinity of pNPMan, extensive studies on variation of the aromatic aglycon of mannoside ligands of FimH have been performed, and some of the results are summarized in Table 4. Variation of the p-substituent of phenyl mannosides had no big effect. It was shown that neither the reduction of the nitro group in pNPMan (leading to pAPMan), nor its deletion (PMan) changed the compounds inhibitory potency to a great extent.[67,68] While Sharon and co-workers had reported a difference between PMan [RIP(PMan)MeMan = 40] and pNPMan [RIP(pNPMan)MeMan = 70], somewhat later this difference could not be confirmed (cf. Table 4). Only an Nacetylamino substituent in the p-position of phenyl mannoside (pNAcPMan) increased the inhibitory potency by a factor of 4 compared to that of pNPMan. On the other hand, the favourable effect of o-substitution was confirmed, as introduction of a chlorine substituent in the o-position of the phenyl ring to yield pNoClPMan increased the affinity for FimH significantly by a factor of 200[74] and 720.[29]
3602
www.eurjoc.org

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion Table 4. A selection of representative mannosides with aromatic aglycon moieties and their testing results.

[a] SD = standard deviation, included in brackets if literature-reported. [b] RIP = relative inhibitory potency, based on MeMan with IPMeMan 1. [c] RIP based on pNPMan with IPpNPMan 1.
Eur. J. Org. Chem. 2011, 35833609 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.eurjoc.org

3603

MICROREVIEW

M. Hartmann, T. K. Lindhorst

Table 5. A selection of representative mannosides with extended aromatic aglycon moieties and their testing results.

[a] SD = standard deviation, included in brackets if literature-reported. [b] RI = relative inhibitory potency, based on MeMan with IPMeMan 1. [c] RIP based on pNPMan with IPpNPMan 1.

When pNPMan was extended by a squaric acid moiety (mannoside 63), also a very high inhibitory potency relative to pNPMan resulted, with a RIP value of 58 as determined by ELISA. Introduction of a chloro substituent in the oposition (mannoside 64) led to a RIP value of 223 (Table 5), and this could be rationalized by molecular modelling.[74] Somewhat different results for 63 and 64 were obtained later by employing a competitive binding study.[84] When the aglycon portion of mannoside 63 was further extended by short peptide chains, the affinities decreased [RIP(65)pNPMan = 7 and RIP(66)pNPMan = 14]. It has therefore been argued that the potentially reactive squaric acid monoamide moiety in 63 might block the binding pocket permanently by covalent attachment to the Nterminus of the lectin, Phe1. This would imply that the affinity for the corresponding squaric acid diamide 67 should
3604
www.eurjoc.org

be lower than the one for 63. Only recently, Lindhorst et al. could show that this is not the case.[93] On the contrary, the squaric acid diamide 67 exceeded the RIP value of its corresponding monoamide derivative 63 by a factor of 3.

Conformational Flexibility of the Ligand and the Lectin According to the induced fit model of ligandreceptor binding, it must be considered that both a carbohydrate ligand and its lectin receptor change their conformation upon formation of a carbohydratelectin complex. Such conformational considerations have, however, not yet been extensively investigated for carbohydrate binding of FimH. For a focussed library of cluster mannosides and glycodendrimers, molecular dynamic simulations have been performed that
Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion

provide an impression of the conformational availability of the included mannose moieties.[94,95] Interestingly, predictions made in these theoretical studies can be aligned with experimental results obtained from inhibition adhesion measurements. When -d-mannosides where clustered as multivalent glycomimetics, the affinities of the respective glycoclusters and glycodendrimers for type 1 fimbriated bacteria (cf. Tables 1 and 2) showed moderate to good improvement.[44,58,60,61] However, when p-nitrophenyl -d-mannoside moieties were analogously assembled in order to increase the favourable affinity of pNPMan, testing results were disappointing.[44,96] In fact, clustering of pNPMan hardly led to inhibitory potencies that exceeded that of monovalent pNPMan when tested with type 1 fimbriated E. coli. This finding can be explained by results obtained in molecular modelling.[94] Molecular dynamic studies with pNPMan glycoclusters suggest that there is a pronounced intramolecular stacking of the included phenyl moieties and that the conformational space that is populated by the scaffolded pNPMan moieties is thus considerably limited. If the sugar moiety is less available for complexation with FimH in pNPMan glycoclusters than in monovalent pNPMan, consequently favourable multivalency effects cannot be expected in this case, and they have not been measured either. Thus, the conformational features of lectin ligands certainly influence their receptor binding properties. On the other hand, the fimbrial lectin FimH will also experience conformational changes upon ligand binding.[97] Most obviously, speculations can be made about the meaning of different conformations of the tyrosine gate at the entrance of the FimH CRD. In crystal structures, the tyrosine gate was in an either closed[68,72] or open conformation.[78] Consequently, the predictions about ligand binding to FimH, deduced from docking studies, are different depending on which FimH conformation was taken as the basis of the modelling.[74] Some mannosides show higher affinity to the closed gate conformation of FimH, others show the opposite preference. It can be speculated that, depending on the ligand that is available for FimH binding in vivo, the lectin adopts an appropriate conformation in order to bind as tightly as possible. The tyrosine gate in this situation might adopt any conformation between the two extremes closed and open. Furthermore, it is intriguing to speculate that the conformational flexibility of the entrance domain of the FimH CRD (including the tyrosine gate) might have a function in selecting and/or preorganizing carbohydrates for binding. Future research might address such hypotheses.

7. Special Approaches to the Inhibition of Mannose-Specific Bacterial Adhesion


From X-ray diffraction analysis and studies on the assembly of fimbriae, it seems to be evident that FimH is the type 1 fimbrial lectin, located at the type 1 fimbrial termini and displaying one monovalent carbohydrate recognition domain (CRD) in its lectin domain (FimHL). Nevertheless, type 1 fimbriae-mediated bacterial adhesion is a multifacEur. J. Org. Chem. 2011, 35833609

eted process, which is not yet understood in all of its details. Consequently, there is also room for novel conceptual approaches both to better understand the mechanisms of bacterial adhesion as well as to further improve the affinities and other advantageous features of inhibitors of bacterial adhesion. As mentioned earlier, higher-valent inhibitors effect an increased avidity compared to that of their monovalent counterparts. These results, obtained with multivalent glycomimetics as ligands of FimH and as inhibitors of type 1 fimbriae-mediated adhesion of E. coli,[32,33,43] are not in accordance with the monovalent nature of the fimbrial lectin, though. In addition, clustering of several FimH CRDs by typical multivalent glycomimetics is unlikely, given the size of fimbriae and their distance dimensions on the bacterial surface. In order to take advantage from the reported multivalency effects and combine them with the ability of carbohydrate conjugates to cluster the bacterial lectins, mannose residues can be coupled to a more extended scaffold. A common technique in biochemistry is to display the ligand on a protein. Some examples of serum albumins functionalized with 8 to 35 mannose residues have been tested in a competitive binding assay with type 1 fimbriated E. coli bacteria.[60] Generally, it was shown that the binding efficiency increased with the degree of mannosylation. Besides, as reported for glycoclusters (vide supra), the length and the hydrophilic and steric properties of the linker connecting the mannose moiety and the protein play a crucial role for the inhibitory potency of the tested compound. The valency-corrected relative inhibitory potencies based on the values obtained for MeMan were in the range of results typically measured for glycoclusters. Most of the tested neoglycoconjugates had a RIPvc of about 50.[60] Hence, increased inhibitory effects, which might arise from lectin clustering or agglutination, were not observed. Only recently, the idea of particle-bound mannosides was resumed. Fullerenes were functionalized with twelve mannoside residues, by using two different linker strategies.[73] It turned out that the valency-corrected inhibitory potencies were unexpectedly low, in the range of 3 (compared to those of the monovalent counterparts). A considerable multivalency effect would have increased the avidity by a much higher factor. On the other hand, it was reported in this study that the dodecamannosylated fullerenes could bind up to seven isolated FimH proteins. Clustering of the isolated form of the lectin domain was thus achieved by this approach. An even more extended scaffold molecule was employed, when mannosylated pseudopolyrotaxanes were introduced in 2010.[98] Three, five or ten cucurbit[6]uril-based mannosylated wheel structures were threaded on polyviologen strings with approximately eleven viologen units. These pseudorotaxane inhibitors were tested in a hemagglutination inhibition assay. The relative inhibitory potencies of the threaded mannose wheels were reported to be 180 to 300. This is a value typically obtained for a potent mannose cluster. Hence, the desired prominent inhibitory effects of these interesting compounds could not be measured,
www.eurjoc.org

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3605

MICROREVIEW
though mannosides were displayed on a voluminous scaffold. Another interesting approach to the development of functional FimH ligands addresses the topic of multivalency from a different point of view. It has been speculated that multiple binding sites might exist on the FimH adhesin.[99] This hypothesis has been encouraged by the finding that certain carbohydrate ligands exhibit stimulating allosteric effects on adhesion,[100] which might be explained by the existence of allosteric binding sites or by activation of carbohydrate complexation in an analogy to the catch bond. Although the sequence of the FimH adhesin is highly conserved, studies by Sokurenko and colleagues[101103] have indicated that allelic variation in FimH is correlated with different carbohydrate-binding profiles. However, none of the allelic variations, which gives rise to differences in mannose-binding, occurs within, or even close to the FimH mannose-binding pocket. Again, additional sugar binding sites on the FimH lectin domain could be an explanation for this finding. Multiple binding sites on FimH could aid in recognizing large and multivalent carbohydrate receptors on the host cell surface. Consequently, molecular modelling was performed to identify putative additional carbohydrate-binding pockets on FimH, and this has led to the suggestion of three locations, which might constitute new potential carbohydratebinding sites on the surface of the FimH lectin domain, in addition to the mannose pocket at the tip of the domain.[99] The hypothesis of multiple carbohydrate binding sites on FimH was tested with a bivalent carbohydrate ligand (Figure 9),[75] which was shaped such that it could concomitantly occupy the known FimH CRD and a putative second carbohydrate-binding site with a preference for high-mannose trisaccharides, as suggested by Knight.[99] This mannoside-trimannoside ligand was meant to clamp two carbohydrate-binding sites on FimH, but failed as effective inhibitor of type 1 fimbriae-mediated bacterial adhesion, when tested by ELISA. Thus, the hypothesis of multiple carbohydrate-binding sites on FimH remains unapproved to date. An alternative method to prove multiple binding sites on FimH is photoaffinity labelling of the protein. According to this methodology, mannoside ligands of FimH have to be equipped with a photolabile functional group that can be activated for covalent cross-linking to the lectin upon irradiation. Proteolytic degradation and mass spectrometric analysis can then lead to the identification of receptor loci, which complex carbohydrates. To allow affinity chromatography with the proteolytic digest of a lectin labelled in this way and to thus facilitate the analysis, biotin can be incorporated into a photolabile lectin ligand, to be selected by streptavidin-based affinity chromatography. Indeed, photoactive mannoside ligands for FimH have been made available and shown to be suited to covalently label FimH.[76] Future studies will reveal whether photoaffinity labelling is a suitable methodology to reliably identify carbohydrate-binding sites as well as allosteric sites on FimH and other lectins.
3606
www.eurjoc.org

M. Hartmann, T. K. Lindhorst

Figure 9. This bivalent glycopeptide ligand was designed to bridge two putative carbohydrate-binding sites on FimH and was synthesized by squaric acid conjugation of a branched trimannoside and a simple -mannoside unit.[75]

8. Conclusions
Lectins have been defined as a class of carbohydratebinding proteins, which are neither antibodies nor enzymes. They have been discovered in all organisms, from plants and microorganisms to vertebrates including humans. They have been allocated to manifold biochemical processes, and it must be assumed that our understanding of how lectins function is not yet comprehensive. Bacteria use lectins as part of adhesive organelles, called fimbriae or pili, to adhere to the glycocalyx of their target cells. Adherence offers a viability advantage for bacteria, as the cell surface forms an ideal site for multiplication and persistency, providing a reliable environment with stable pH and salt conditions and versatile nutrients to feed on. Preventing bacterial adhesion by suitable carbohydrate inhibitors has been envisaged as a means against bacterial colonization, infection by pathogens, inflammation and other diseases and medical problems caused by adhesive bacteria. Antiadhesives might offer an alternative to antibiotic treatment, which is threatened by a growing number of multiresistant bacterial strains.[19] The vision of an antiadhesion therapy has not yet led to pharmacological applications, but it has greatly motivated current programmes on the development of high-affinity inhibitors of bacterial lectins.[66,68,93,104] The type 1 fimbrial lectin FimH is crucial for mannosespecific adhesion of most enterobacteria including UPEC and thus forms a promising target for carbohydrate drug development. While at first various mannosides and natural oligosaccharides were investigated as FimH antagonists, mainly two approaches were taken to develop high-affinity ligands of FimH. Since the first crystal structure of FimH was solved in 1999,[50] FimH ligands have been made (i) according to rational design guided by the crystal structure of the FimH carbohydrate-binding site or (ii) more often, glycoclusters were designed for the multivalent presentation of -d-mannoside ligands for FimH. Though the bacterial lectin FimH is a monovalent lectin with a carbohydrate-binding site that accommodates just one -mannosyl residue, frequently favourable effects have been
Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion

observed with multivalent cluster mannosides of various architectures. Such multivalency effects have not yet been fully understood in the case of FimH. For steric reasons, it is unlikely that the rather small glycoclusters that have been employed for the inhibition of type 1 fimbriae-mediated bacterial adhesion can effectively cluster several FimH CRDs. It appears to be more reasonable to conclude on secondary binding sites on the lectin surface or interpret the effect of multivalent FimH ligands on the basis of statistics such as high ligand density in the close proximity of the FimH CRD. For a sensible interpretation of testing results, it is important to keep in mind the complexity of most testing systems and of the in vivo situation of bacterial adhesion to host cell surfaces. In addition to the lectincarbohydrate interaction, which is isolated from the more complex molecular scenario to be used as the basis for rational drug design, any carbohydrate inhibitor also interacts with the surrounding carbohydrate environment. Likewise, the fimbrial lectin is embedded into a possibly favourable environment before an interaction with an added carbohydrate inhibitor occurs. The thermodynamic, enthalpic and entropic, as well as kinetic circumstances of this multifaceted situation can hardly be sorted out in all detail. Furthermore, the biophysical parameters of carbohydrate binding will be very different whether a hemagglutination assay, an ELISA or SPR measurements under flow, among others, are being used to determine the inhibitory potency of a particular FimH ligand. Moreover, it was shown that shear forces even influence the structure of the type 1 fimbrial lectin FimH,[87] leading to tighter carbohydrate binding according to a mechanistic principle that has been termed catch bonds.[85] All together, it cannot be taken for granted that carbohydrate ligands for bacterial lectins can be successfully developed as antiadhesives for an antiadhesion therapy in vivo. The multidimensional in vivo scenario of cellular adhesion might prevent the success of the rather one-dimensional idea of an antiadhesion therapy. In addition, it could be a critical limitation for the concept to provide enough selectivity not to interfere with physiological cell adhesion. Given that most pathogens, bacteria and viruses, possess different kinds of lectins to use several carbohydrates for adhesion, selectivity issues will form a major challenge for any approach to an antiadhesion therapy. After all, research on lectin ligands is also highly motivated by fundamental questions on the mechanism of carbohydrate binding. Thus, it will continue to form an important field in the glycosciences and eventually will lead to new applications in life science. The development of functional glycomimetics and lectin ligands will assist in unravelling the secrets of glycobiology and to understand biology beyond genomics and proteomics. The key to understanding glycobiology will be to identify so far undiscovered intrinsic features of carbohydrates, as displayed in the supramolecular environment of living cells.
Abbreviations AIBN = azobis(isobutyronitrile), ASGPR = asialoglycoprotein receptor, 9-BBN = 9-borabicyclo[3.3.1]nonane, BuMan = butyl -dEur. J. Org. Chem. 2011, 35833609

mannoside, C-HEGA = cyclohexylbutanoyl-N-hydroxyethyl-d-glucamide, CRD = carbohydrate recognition domain, DSE = donor strand exchange, E. coli = Escherichia coli, EC50 = half maximal effective concentration, ELISA = enzyme-linked immunosorbent assay, FACS = fluorescence-activated cell sorting, FimH-CRD-ThHis = construct containing the FimH-CRD, thrombin cleavage site and 6His-tag; FP = fluorescence polarization, Gal = galactose, GalNAc = N-acetylgalactosamine, GFP = green fluorescent protein, HAI = hemagglutination inhibition, HSA = human serum albumin, IC50 = half maximal inhibitory concentration, IT = inhibition titre, ITC = isothermal titration calorimetry, LoVo cells = human colon adenocarcinoma cell line, Man = mannose, MeMan = methyl -d-mannoside, MeUmbMan = methlyumbelliferyl -dmannoside, Neu5Ac = 5-N-acetylneuraminic acid, pNPMan = para-nitrophenyl -d-mannoside, QSAR = quantitative structure activity relationship; RIP = relative inhibitory potency, SAM = self-assembled monolayer, SD = standard deviation, SPR = surface plasmon resonance, TEM = transmission electron microscopy, TMSOTf = trimethylsilyl trifluoromethanesulfonate, UPEC = uropathogenic Escherichia coli.

Acknowledgments
Our own work on carbohydrate binding of type 1 fimbriated bacteria was financed in its first period by the Deutsche Forschungsgemeinschaft (DFG) and later by Christiana Albertina University. Most valuable over the years have been the contributions of the Lindhorst group members, in particular those of Dr. Christoffer Kieburg, Dr. Sven Ktter, Dr. Michael Dubber, Dr. Anupama Patel, Dr. Oliver Sperling, Dr. Christoph Heidecke and for molecular modelling Dr. Andreas Fuchs and Jrn Schmidt-Lassen. Furthermore, we are very thankful to our collaborators in this field of our research, Dr. Ulrike Krallmann-Wenzel, Prof. Dr. Stefan Ehlers and Prof. Dr. Stefan Knight. [1] N. Sharon, H. Lis, Glycobiology 2004, 14, 53R62R. [2] W. C. Boyd, E. Shapleigh, M. McMaster, Arch. Biochem. Biophys. 1955, 55, 226234. [3] W. C. Boyd, E. Shapleigh, Science 1954, 119, 419. [4] N. Sharon, H. Lis, Science 1972, 177, 949959. [5] J. B. Sumner, S. F. Howell, J. Bacteriol. 1936, 32, 227237. [6] M. Ambrosi, N. R. Cameron, B. G. Davis, Org. Biomol. Chem. 2005, 3, 15931608. [7] K. Drickamer, Curr. Opin. Struct. Biol. 1993, 3, 393400. [8] K. Drickamer, Curr. Opin. Struct. Biol. 1999, 9, 585590. [9] H. Lis, N. Sharon, Chem. Rev. 1998, 98, 6376749. [10] M. J. Blaser, Sci. Am. 2005, 292, 3845. [11] D. Mack, A. P. Davies, L. G. Harris, J. K. M. Knobloch, H. Rohde, Top. Curr. Chem. 2009, 288, 157182. [12] R. J. Menaker, N. L. Jones, Microbes Infect. 2003, 5, 1149 1158. [13] J. K. Savjani, A. K. Gajjar, K. T. Savjani, Mini-Rev. Med. Chem. 2009, 9, 194205. [14] A. B. Estrela, M. G. Heck, W. R. Abraham, Curr. Med. Chem. 2009, 16, 15121530. [15] H. Connell, W. Agace, P. Klemm, M. Schembri, S. Marild, C. Svanborg, Proc. Natl. Acad. Sci. USA 1996, 93, 98279832. [16] N. Sharon, Biochim. Biophys. Acta 2006, 1760, 527537. [17] I. Ofek, D. L. Hasty, N. Sharon, FEMS Immunol. Med. Microbiol. 2003, 38, 181191. [18] R. J. Pieters, Med. Res. Rev. 2007, 27, 796816. [19] S. B. Levy, Adv. Drug Delivery Rev. 2005, 57, 14461450. [20] N. Sharon, FEBS Lett. 1987, 217, 145157. [21] S. D. Knight, J. Bouckaert, Top. Curr. Chem. 2009, 288, 67 107.
www.eurjoc.org

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3607

MICROREVIEW
[22] T. A. Oelschlaeger, U. Dobrindt, J. Hacker, Curr. Opin. Urol. 2002, 12, 3338. [23] G. Zhou, W. J. Mo, P. Sebbel, G. W. Min, T. A. Neubert, R. Glockshuber, X. R. Wu, T. T. Sun, X. P. Kong, J. Cell Sci. 2001, 114, 40954103. [24] E. F. Boyd, D. L. Hartl, J. Bacteriol. 1999, 181, 13011308. [25] N. Firon, I. Ofek, N. Sharon, Carbohydr. Res. 1983, 120, 235 249. [26] M. Dubber, O. Sperling, T. K. Lindhorst, Org. Biomol. Chem. 2006, 4, 39013912. [27] J.-R. Neeser, B. Koellreutter, P. Wuersch, Infect. Immun. 1986, 52, 428436. [28] A. Wellens, C. Garofalo, H. Nguyen, N. Van Gerven, R. Slttegrd, J. P. Hernalsteens, L. Wyns, S. Oscarson, H. De Greve, S. Hultgren, J. Bouckaert, PLoS ONE 2008, 3, e2040. [29] N. Firon, S. Ashkenazi, D. Mirelman, I. Ofek, N. Sharon, Infect. Immun. 1987, 55, 472476. [30] J. Beuth, H. L. Ko, G. Pulverer, G. Uhlenbruck, H. Pichlmaier, Glycoconjugate J. 1995, 12, 16. [31] Y. C. Lee, R. T. Lee, Acc. Chem. Res. 1995, 28, 321327. [32] M. Lahmann, Top. Curr. Chem. 2009, 288, 1765. [33] Y. M. Chabre, R. Roy, Adv. Carbohydr. Chem. Biochem. 2010, 63, 165393. [34] M. Mammen, S.-K. Choi, G. M. Whitesides, Angew. Chem. 1998, 110, 29082953; Angew. Chem. Int. Ed. 1998, 37, 2754 2794. [35] C. R. Bertozzi, L. L. Kiessling, Science 2001, 291, 23572364. [36] L. L. Kiessling, R. A. Splain, Annu. Rev. Biochem. 2010, 79, 619653. [37] J. J. Lundquist, E. J. Toone, Chem. Rev. 2002, 102, 555578. [38] E. J. Toone, Curr. Opin. Struct. Biol. 1994, 4, 719728. [39] Y. Chen, G. Chen, M. H. Stenzel, Macromolecules 2010, 43, 81098114. [40] S. G. Spain, N. R. Cameron, Polym. Chem. 2011, 2, 6068. [41] J. E. Gestwicki, C. W. Cairo, L. E. Strong, K. A. Oetjen, L. L. Kiessling, J. Am. Chem. Soc. 2002, 124, 1492214933. [42] K. Godula, C. R. Bertozzi, J. Am. Chem. Soc. 2010, 132, 9963 9965. [43] M. Touaibia, R. Roy, Mini-Rev. Med. Chem. 2007, 7, 1270 1283. [44] T. K. Lindhorst, C. Kieburg, U. Krallmann-Wenzel, Glycoconjugate J. 1998, 15, 605613. [45] T. K. Lindhorst, Nachr. Chem. Tech. Lab. 1996, 44, 10731079. [46] N. Rckendorf, T. K. Lindhorst, Top. Curr. Chem. 2001, 217, 201238. [47] W. B. Turnbull, J. F. Stoddart, J. Biotechnol. 2002, 90, 231255. [48] B. Knig, T. Fricke, A. Wamann, U. Krallmann-Wenzel, T. K. Lindhorst, Tetrahedron Lett. 1998, 39, 23072310. [49] S. Ktter, U. Krallmann-Wenzel, S. Ehlers, T. K. Lindhorst, J. Chem. Soc. Perkin Trans. 1 1998, 21922200. [50] D. Choudhury, A. Thompson, V. Stojanoff, S. Langermann, J. Pinkner, S. J. Hultgren, S. D. Knight, Science 1999, 285, 1061 1066. [51] N. Rckendorf, O. Sperling, T. K. Lindhorst, Aust. J. Chem. 2002, 55, 8793. [52] T. K. Lindhorst, M. Dubber, U. Krallmann-Wenzel, Eur. J. Org. Chem. 2000, 20272034. [53] T. K. Lindhorst, S. Ktter, U. Krallmann-Wenzel, J. Chem. Soc. Perkin Trans. 1 2001, 823831. [54] M. M. K. Boysen, K. Elsner, T. K. Lindhorst, Eur. J. Org. Chem. 2003, 43764386. [55] M. Meldal, C. W. Tornoe, Chem. Rev. 2008, 108, 29523015. [56] R. Chinchilla, C. Najera, Chem. Rev. 2007, 107, 874922. [57] L. F. Tietze, M. Arlt, M. Beller, K. H. Glsenkamp, E. Jhde, M. F. Rajewsky, Chem. Ber. 1991, 124, 12151221. [58] M. Touaibia, A. Wellens, T. C. Shiao, Q. Wang, S. Sirois, J. Bouckaert, R. Roy, ChemMedChem 2007, 2, 11901201. [59] S. G. Gouin, A. Wellens, J. Bouckaert, J. Kovensky, ChemMedChem 2009, 4, 749755. 3608
www.eurjoc.org

M. Hartmann, T. K. Lindhorst [60] N. Nagahori, R. T. Lee, S.-I. Nishimura, D. Pag, R. Roy, Y. C. Lee, ChemBioChem 2002, 3, 836844. [61] C. Heidecke, T. K. Lindhorst, Chem. Eur. J. 2007, 13, 9056 9067. [62] O. Sperling, M. Dubber, T. K. Lindhorst, Carbohydr. Res. 2007, 342, 696703. [63] A. Mulder, J. Huskens, D. N. Reinhoud, Org. Biomol. Chem. 2004, 2, 34093424. [64] F. G. Sauer, M. Barnhart, D. Choudhury, S. D. Knight, G. Waksman, S. J. Hultgren, Curr. Opin. Struct. Biol. 2000, 10, 548556. [65] D. Abgottspon, G. Rlli, L. Hosch, A. Steinhuber, X. Jiang, O. Schwardt, B. Cutting, M. Smiesko, U. Jenal, B. Ernst, A. Trampuz, J. Microbiol. Methods 2010, 82, 249255. [66] T. Klein, D. Abgottspon, M. Wittwer, S. Rabbani, J. Herold, X. Jiang, S. Kleeb, C. Lthi, M. Scharenberg, J. Bezencon, E. Gubler, L. Pang, M. Smiesko, B. Cutting, O. Schwardt, B. Ernst, J. Med. Chem. 2010, 53, 47794792. [67] T. K. Lindhorst, S. Ktter, J. Kubisch, U. Krallmann-Wenzel, S. Ehlers, V. Krn, Eur. J. Org. Chem. 1998, 16691674. [68] Z. Han, J. S. Pinkner, B. Ford, R. Obermann, W. Nolan, S. A. Wildman, D. Hobbs, T. Ellenberger, C. K. Cusumano, S. J. Hultgren, J. W. Janetka, J. Med. Chem. 2010, 53, 47794792. [69] A. Horst, S. Ktter, T. K. Lindhorst, A. Ludwig, E. Brandt, C. Wagener, Med. Microbiol. Immunol. 2001, 90, 145149. [70] C.-C. Lin, Y.-C. Yeh, C.-Y. Yang, C.-L. Chen, G.-F. Chen, C.C. Chen, Y.-C. Wu, J. Am. Chem. Soc. 2002, 124, 35083509. [71] L.-H. Liu, H. Dietsch, P. Schurtenberger, M. Yan, Bioconjugate Chem. 2009, 20, 13491355. [72] J. Bouckaert, J. Berglund, M. Schembri, E. De Genst, L. Cools, M. Wuhrer, C.-S. Hung, J. Pinkner, R. Slttegrd, A. Zavialov, D. Choudhury, S. Langermann, S. J. Hultgren, L. Wyns, P. Klemm, S. Oscarson, S. D. Knight, H. De Greve, Mol. Microbiol. 2005, 55, 441455. [73] M. Durka, K. Buffet, J. Iehl, M. Holler, J.-F. Nierengarten, J. Taganna, J. Bouckaert, S. P. Vincent, Chem. Commun. 2011, 47, 13211323. [74] O. Sperling, A. Fuchs, T. K. Lindhorst, Org. Biomol. Chem. 2006, 4, 39133922. [75] T. K. Lindhorst, K. Bruegge, A. Fuchs, O. Sperling, Beilstein J. Org. Chem. 2010, 6, 801809. [76] T. K. Lindhorst, M. Mrten, A. Fuchs, S. D. Knight, Beilstein J. Org. Chem. 2010, 6, 810822. [77] D. Chessa, C. W. Dorsey, M. Winter, A. J. Bumler, J. Biol. Chem. 2008, 283, 81188124. [78] C.-S. Hung, J. Bouckaert, D. Hung, J. Pinkner, C. Widberg, A. DeFusco, C. G. Auguste, R. Strouse, S. Langermann, G. Waksman, S. J. Hultgren, Mol. Microbiol. 2002, 44, 903915. [79] M. Hartmann, A. K. Horst, P. Klemm, T. K. Lindhorst, Chem. Commun. 2010, 46, 330332. [80] M. D. Disney, P. H. Seeberger, Chem. Biol. 2004, 11, 1701 1707. [81] X. Qian, S. J. Metallo, I. S. Choi, H. Wu, M. N. Liang, G. M. Whitesides, Anal. Chem. 2002, 74, 18051810. [82] S. Yu, A. W. Lowe, BMC Gastroenterol. 2009, 9, 58. [83] B. Yang, X. Zhang, X. Bao, Y. Lv, J. Zhang, S. Guo, Food Res. Int. 2008, 41, 594599. [84] S. Rabbani, X. Jiang, O. Schwardt, B. Ernst, Anal. Biochem. 2010, 407, 188195. [85] W. Thomas, Annu. Rev. Biomed. Eng. 2008, 10, 3957. [86] M. Dembo, D. C. Torney, K. Saxman, D. Hammer, Proc. R. Soc. London Ser. B 1988, 234, 5583. [87] I. Le Trong, P. Aprikian, B. A. Kidd, M. Forero-Shelton, V. Tchesnokova, P. Rajagopal, V. Rodriguez, G. Interlandi, R. Klevit, V. Vogel, R. E. Stenkamp, E. V. Sokurenko, W. E. Thomas, Cell 2010, 141, 645655. [88] W. I. Weis, K. Drickamer, Annu. Rev. Biochem. 1996, 65, 441 473. [89] E. P. Mitchell, C. Sabin, L. Snajdrov, M. Pokorn, S. Perret, C. Gautier, C. Hofr, N. Gilboa-Garber, J. Koca, M. Wim
Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion merov, A. Imberty, Proteins Struct., Funct., Bioinf. 2005, 58, 735746. [90] K. Ramrez-Gualito, R. Alonso-Ros, B. Quiroz-Garca, A. Rojas-Aguilar, D. Daz, J. Jimnez-Barbero, G. Cuevas, J. Am. Chem. Soc. 2009, 131, 1812918138. [91] D. B. Walker, G. Joshi, A. P. Davis, Cell Mol. Life Sci. 2009, 66, 31773191. [92] H. Remaut, C. Tang, N. S. Henderson, J. S. Pinkner, T. Wang, S. J. Hultgren, D. G. Thanassi, G. Waksman, H. Li, Cell 2008, 133, 640652. [93] C. Grabosch, M. Hartmann, J. Schmidt-Lassen, T. K. Lindhorst, ChemBioChem 2011, 12, 10661074. [94] C.-W. von der Lieth, M. Frank, T. K. Lindhorst, Rev. Mol. Biotechnol. 2002, 90, 311337. [95] A. Fuchs, Dissertation, Kiel, 2005. [96] C. Kieburg, Dissertation, Hamburg, 1997. [97] M. Erikson, Dissertation, Kiel, 2010; M. Eriksson, T. K. Lindhorst, B. Hartke, to be published. [98] J. Kim, Y. Ahn, K. M. Park, D.-W. Lee, K. Kim, Chem. Eur. J. 2010, 16, 1216812173. [99] S. D. Knight, J. Berglund, D. Choudhury, Curr. Opin. Chem. Biol. 2000, 4, 653660. [100] A. J. Schaeffer, S. K. Amundsen, J. M. Jones, Infect. Immun. 1980, 30, 531537. [101] E. V. Sokurenko, V. Chesnokova, R. J. Doyle, D. L. Hasty, J. Biol. Chem. 1997, 272, 1788017886. [102] E. V. Sokurenko, V. Chesnokova, D. E. Dykhuizen, I. Ofek, X.-R. Wu, K. A. Krogfelt, C. Struve, M. A. Schembri, D. L. Hasty, Proc. Natl. Acad. Sci. USA 1998, 95, 89228926. [103] M. A. Schembri, E. V. Sokurenko, P. Klemm, Infect. Immun. 2000, 68, 26382646. [104] B. Ernst, J. L. Magnani, Nat. Rev. Drug Discovery 2009, 8, 661677. Received: March 24, 2011 Published Online: June 15, 2011

Eur. J. Org. Chem. 2011, 35833609

2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjoc.org

3609

You might also like