Comparative Study of Aryl O-, C-, and S-Mannopyranosides as Potential Adhesion Inhibitors toward Uropathogenic E. coli FimH

A set of three mannopyranoside possessing identical 1,1′-biphenyl glycosidic pharmacophore but different aglyconic atoms were synthesized using either a palladium-catalyzed Heck cross coupling reaction or a metathesis reaction between their corresponding allylic glycoside derivatives. Their X-ray structures, together with their calculated 3D structures, showed strong indicators to explain the observed relative binding abilities against E. coli FimH as measured by a improved surface plasmon resonance (SPR) method. Amongst the O-, C-, and S-linked analogs, the C-linked analog showed the best ability to become a lead candidate as antagonist against uropathogenic E. coli with a Kd of 11.45 nM.


Introduction
Multidrug resistant bacteria constitutes a serious public health burden with approximately 700,000 people dying each year from infections caused by antibiotic resistant organisms. This number is expected to reach 10 million people annually by 2050 with related cost ranging $100 trillion US dollars [1]. Although medicinal chemistry has allowed the relatively fast discoveries of diverse families of potent antibiotics working under a wide range of bactericidal or bacteriostatic mechanisms [2], bacteria have similarly developed a variety of ingenious defense mechanisms [3]. Therefore, there is an urgent need to fight bacterial infections by mechanisms that alleviate multidrug resistances. One of these approaches, showing increasing successes, is directed at preventing the early step of bacterial adhesion and colonization, and biofilm formation pathways [4]. Since several infection mechanisms are initiated by molecular recognition between the host cell surface receptors and virulence factors on the bacterial surfaces [5], such strategies are likely to be unavoidable without impeding the bacteria's own capacity to recognize and adhere to the targeted tissues. Urinary tract infections (UTIs) represent one of the most common bacterial infections, affecting one out of two women during their lifetimes. UTIs are serious health problems as they reach close to $2.5 billion in annual healthcare costs in the US alone [6] which are caused by Gram-negative uropathogenic Escherichia coli (UPEC) strains [7].
UPEC are the major cause of urinary tract infections and intestinal bowel diseases (IBD) including Crohn's disease (CD) [1,6]. These strains enter urinary tracts to form colonies. The adhesions of various pathogenic E. coli strains to host cells are primarily mediated through carbohydrate-protein interactions involving bacterial fimbriae which can recognize specific glycoconjugate receptors on host cells. Of particular interests are the FimH and the PapG fimbriae that bind to mannosylated glycoproteins or glycolipids, respectively [8].
The FimH bacterial lectins adhere to the highly mannosylated glycoprotein uroplakin 1a [9,10] on the bladder cells and on glycoprotein 2 (GP2) exclusively expressed in the small intestine [7,11]. Once the infection process is initiated, it is usually followed by thick biofilm formation [12,13]. Biofilms have a pronounced impact in clinical settings because they render bacterial infections resilient to antibiotics [14]. Given that bacterial adhesion to host cells is the preliminary step toward the release of toxic proteins, the design of potent E. coli FimH antagonists has been the target of several efforts [15,16]. Several glycomimetic analogs of the naturally occurring complex oligomannosides have been synthesized [8,10] that provided potent α-D-mannopyranoside-based antiadhesins. Since small molecule antagonists represent the foremost choice of the pharmaceutical industry, these glycomimetics constitute most promising candidates for the replacement of the natural lead oligomannosides onto which E. coli adhere to [10,17].
To circumscribe the effect of anomeric linkages upon the design of uropathogenic E. coli FimH antagonists based on mannopyranosides, we compare herein three candidates possessing aromatic aglycones capable to fit within the well-established hydrophobic tyrosine gate present in the mannopyranoside binding pocket. Hence, the tridimensional structures, relative binding affinities of O-, C-, and S-linked mannopyranosides harboring a common 1,1 -biphenyl pharmacophore are described.

Glycomimetic Synthesis
Since O-substituted mannopyranosides are susceptible to hydrolysis by mannosidases and low pHs of the gut, there is a need to identify alternative candidates suitable under in vivo conditions. N- [13], C- [10,18,19], or S-linked [20] glycomimetics have the potential to possess the required stabilities and appropriate pharmacokinetic and pharmacodynamic parameters.
Analogues of Oand C-linked α-D-mannopyranosides with hydrophobic and aryl substituents have already been identified as potent E. coli FimH antagonists having low nanomolar Kds or IC 50 s [5,10,18,21,22]. Para-substituted biphenyl derivatives were shown to be particularly appealing owing to their numerous favorable binding interactions within the identified tyrosine gate formed between Tyr48 and Tyr137 [18,21,[23][24][25]. In turn, some derivatives were shown to bind better when the Tyr-gate is either open, half-open, or close, further pointing toward rationally design glycol therapeutics [20,22,[26][27][28]. In fact, using the centroid positions of each of the Tyr48 and Tyr137 phenyl groups in crystal data sets, we calculated that the Tyr-gates are 8.77 Å, 7.79 Å, and 6.99 Å apart in the open (PDB 4AV0) [29], half-open (PDB 4BUQ) [29], and close states (PDB 4AUY) [30], respectively. Since this family of α-D-mannopyranoside derivatives showed high promises as antagonists, we decided to further evaluate their comparative 3-D structures, relative binding capacities, and biophysical characteristics, by exploring the roles played by substituting the anomeric linkages with heteroatoms, while keeping constant the para-biphenyl moieties. Toward this goal, we first accessed a series of α-D-mannopyranosides 1-6 ( Figure 1) known to serve as comparative standards. Compounds 1 (MeαMan) and 2 (PNPαMan) are commercial products, while 3 (HeptylαMan), the golden standard with a reported Kd of 5 nM (SPR), together with compounds 4 (see SI) were prepared by literature procedures. The 1, 1 -biphenyl derivatives with an anomeric O-linkage (5) and the corresponding C-linked analog (6), albeit with one atom shorter, were also prepared by published procedures [18]. Note that we successfully crystallized compounds 5 and 6 for the first time (see discussion below). We describe herein the synthesis of the related S-linked α-D-mannopyranoside (12) (Scheme 1). Similarly to its O-and C-linked congeners, we chose the S-allyl α-D-mannopyranoside 9 as key starting materials. However, rather than using a palladium catalyzed Heck reaction and a series of aryl iodides for the cross-coupling as before [18,21], we decided to extend its alkenyl functionality by a metathesis reaction using 4-vinyl-1,1′-biphenyl (10) and Grubbs 2 nd generation catalyst. This alternative strategy was considered equally appealing owing to its potential to access a diversified library of analogs. In this way, we can demonstrate the versatility of these two complementary chemical strategies leading to these important families of O-, C-, and S-linked derivatives.

Structural and Conformational Analyses
Having secured entry to our three heteroatom-linked α-D-mannopyranosides 5, 6, and 12, we evaluated their detailed 3D structures as it was anticipated that the relative orientation of their pbiphenyl pharmacophores with the E. coli FimH binding site would be determinant for their relative binding abilities. This is particularly true given the difference in the width of the Tyr gate in the open, half-open, and close states, as mentioned above and that they may individually bind to one form or another unpredictably. Fortuitously, we accessed to two crystalline structures, namely, those of the O-and C-linked analogues 5 and 6, respectively. In addition, we previously resolved fourteen X-ray co-crystal data of several O-linked analogues bound to the E. coli FimH, thus giving us access to several docking opportunities [10,18,21,29,30]. One of them in particular (7) is of interest (PDB 4AV5) We describe herein the synthesis of the related S-linked α-D-mannopyranoside (12) (Scheme 1). Similarly to its Oand C-linked congeners, we chose the S-allyl α-D-mannopyranoside 9 as key starting materials. However, rather than using a palladium catalyzed Heck reaction and a series of aryl iodides for the cross-coupling as before [18,21], we decided to extend its alkenyl functionality by a metathesis reaction using 4-vinyl-1,1 -biphenyl (10) and Grubbs 2 nd generation catalyst. This alternative strategy was considered equally appealing owing to its potential to access a diversified library of analogs. In this way, we can demonstrate the versatility of these two complementary chemical strategies leading to these important families of O-, C-, and S-linked derivatives. We describe herein the synthesis of the related S-linked α-D-mannopyranoside (12) (Scheme 1). Similarly to its O-and C-linked congeners, we chose the S-allyl α-D-mannopyranoside 9 as key starting materials. However, rather than using a palladium catalyzed Heck reaction and a series of aryl iodides for the cross-coupling as before [18,21], we decided to extend its alkenyl functionality by a metathesis reaction using 4-vinyl-1,1′-biphenyl (10) and Grubbs 2 nd generation catalyst. This alternative strategy was considered equally appealing owing to its potential to access a diversified library of analogs. In this way, we can demonstrate the versatility of these two complementary chemical strategies leading to these important families of O-, C-, and S-linked derivatives.

Structural and Conformational Analyses
Having secured entry to our three heteroatom-linked α-D-mannopyranosides 5, 6, and 12, we evaluated their detailed 3D structures as it was anticipated that the relative orientation of their pbiphenyl pharmacophores with the E. coli FimH binding site would be determinant for their relative binding abilities. This is particularly true given the difference in the width of the Tyr gate in the open, half-open, and close states, as mentioned above and that they may individually bind to one form or another unpredictably. Fortuitously, we accessed to two crystalline structures, namely, those of the O-and C-linked analogues 5 and 6, respectively. In addition, we previously resolved fourteen X-ray co-crystal data of several O-linked analogues bound to the E. coli FimH, thus giving us access to several docking opportunities [10,18,21,29,30]. One of them in particular (7) is of interest (PDB 4AV5) Scheme 1. Synthetic scheme leading to the S-linked mannopyranoside analog 12 using a metathesis reaction.

Structural and Conformational Analyses
Having secured entry to our three heteroatom-linked α-D-mannopyranosides 5, 6, and 12, we evaluated their detailed 3D structures as it was anticipated that the relative orientation of their p-biphenyl pharmacophores with the E. coli FimH binding site would be determinant for their relative binding abilities. This is particularly true given the difference in the width of the Tyr gate in the open, half-open, and close states, as mentioned above and that they may individually bind to one form or another unpredictably. Fortuitously, we accessed to two crystalline structures, namely, those of the Oand C-linked analogues 5 and 6, respectively. In addition, we previously resolved fourteen X-ray co-crystal data of several O-linked analogues bound to the E. coli FimH, thus giving us access to several docking opportunities [10,18,21,29,30]. One of them in particular (7) is of interest (PDB 4AV5) [29] as it also contains a 1,1 -biphenyl ring, albeit with a more rigid alkynyl linkage (see Figure S37 in SI for superimposed figure containing 5 and 7). Interestingly, while 7 showed a Kd of 405 nM, 5 was 135 fold better with a Kd of 3 nM (SPR) [18], again demonstrating the importance of pharmacophore orientation within the tyrosine gate. Figure S37 (SI) clearly shows compound 5 forming a closer contact with Tyr137 in the open form. It's docking within the FimH active site was done by superimposition of the mannoside residues with that of known 7.
The crystalline structure of 5 alone is shown in Figure 2. Also of interest is the fact that the two biphenyl rings are not tilted with respect to one another. This is in strike contrast with the situation observed from other p-biphenyl candidates including our own above (7, PDB 4Av5) [28,29,37]. In addition; it is also noteworthy that the conformation around the C5-C6 bond in the mannose ring is in the gg orientation, as it is the case of most crystal structures of mannosides within the binding pocket of FimH.
Molecules 2016, 21, x 4 of 5 [29] as it also contains a 1,1′-biphenyl ring, albeit with a more rigid alkynyl linkage (see Figure S.37 in SI for superimposed figure containing 5 and 7). Interestingly, while 7 showed a Kd of 405 nM, 5 was 135 fold better with a Kd of 3 nM (SPR) [18], again demonstrating the importance of pharmacophore orientation within the tyrosine gate. Figure S37 (SI) clearly shows compound 5 forming a closer contact with Tyr137 in the open form. It's docking within the FimH active site was done by superimposition of the mannoside residues with that of known 7.
The crystalline structure of 5 alone is shown in Figure 2. Also of interest is the fact that the two biphenyl rings are not tilted with respect to one another. This is in strike contrast with the situation observed from other p-biphenyl candidates including our own above (7, PDB 4Av5) [28][29]37]. In addition; it is also noteworthy that the conformation around the C5-C6 bond in the mannose ring is in the gg orientation, as it is the case of most crystal structures of mannosides within the binding pocket of FimH. The C-linked analogue 6, having a C-O-C bond shorter, deserves some specific comments on its own because of the controversial arguments about the preferred conformational states of this family of glycomimetics. Indeed, several literature cases argued for the fact that C-linked α-Dmannopyranosides do not exist preferentially in the "usual" 4 C1 chair conformations, at least for simple aglycones [10,[38][39][40][41]. As already shown by us with O-linked allyl α-D-mannopyranoside [10], the 4 C1 conformation was the most stable form with the inverted 1 C4 chair being next in line with a difference of 8.10 kcal/mole difference. To see if this observation still holds for compound 6, we undertook a conformational search using the Monte Carlo method [42]. The retained structure was then optimized using molecular mechanics [43][44][45] and the density functional theory (DFT) with the hybrid functional B3LYP, base 6-31G* (see SI info). The result clearly showed, once again that this additional C-linked α-D-mannopyranoside 6 also predominates in the desired 4 C1 conformation.
Fortuitously again, we were able to crystallize compound 6 alone as they appear in Figure 3 as 6A and 6B. However, the situation was complicated by the fact that the crystal lattice showed two sugar residues stacking to one another (Figure 3, 6A and 6B). Apparently, both biphenyl moieties were forming strong CH-π interactions [10,46,47], and likely reinforced in that state since the C3-OH group (6A) was hydrogen bonded to the C6-O (6B, 3.914 Å). In addition, one sugar (6A) had the C5-C6 bond in the tg conformation while 6B was in the gg state. Moreover, both of the p-substituted biphenyl rings were flat, as opposed to most observed cases discussed above [23,24] when the sugars are bound to the FimH. This is also seen in the calculated conformer 4 C1 (Figure 3, 6C) where the two rings are tilted by -56.0° in close agreement to those observed for PDB 3MCY [23] (-36.21°) and PDB 5F2F [23] (-37.47°) [24]. Additionally, the C5-C6 bond was in the tg conformation.
Clearly, for the situation seen when 6 crystallized alone, the self-assembly into a dimeric form becomes a dominant factor which could not be seen in either the calculated values nor with the FimH binding site. Other significant differences were observed between the three forms denoted 6A, 6B, and 6C for the crystalline states and calculated one, respectively: ΦA = -35.84°, ΦB = -54.84°, ΦC = +54.48°; ΨA = -54.50°; ΨB = +65.63°, ΨC = -91.69, wherein the Φ torsion angle is define between H1-C1-C1′-C2′ and the Ψ torsion angle is define as C1-C1′-C2′-C2′H. The C-linked analogue 6, having a C-O-C bond shorter, deserves some specific comments on its own because of the controversial arguments about the preferred conformational states of this family of glycomimetics. Indeed, several literature cases argued for the fact that C-linked α-D-mannopyranosides do not exist preferentially in the "usual" 4 C 1 chair conformations, at least for simple aglycones [10,[38][39][40][41]. As already shown by us with O-linked allyl α-D-mannopyranoside [10], the 4 C 1 conformation was the most stable form with the inverted 1 C 4 chair being next in line with a difference of 8.10 kcal/mole difference. To see if this observation still holds for compound 6, we undertook a conformational search using the Monte Carlo method [42]. The retained structure was then optimized using molecular mechanics [43][44][45] and the density functional theory (DFT) with the hybrid functional B3LYP, base 6-31G* (see SI info). The result clearly showed, once again that this additional C-linked α-D-mannopyranoside 6 also predominates in the desired 4 C 1 conformation. Fortuitously again, we were able to crystallize compound 6 alone as they appear in Figure 3 as 6A and 6B. However, the situation was complicated by the fact that the crystal lattice showed two sugar residues stacking to one another (Figure 3, 6A and 6B). Apparently, both biphenyl moieties were forming strong CH-π interactions [10,46,47], and likely reinforced in that state since the C3-OH group (6A) was hydrogen bonded to the C6-O (6B, 3.914 Å). In addition, one sugar (6A) had the C5-C6 bond in the tg conformation while 6B was in the gg state. Moreover, both of the p-substituted biphenyl rings were flat, as opposed to most observed cases discussed above [23,24] when the sugars are bound to the FimH. This is also seen in the calculated conformer 4 C 1 (Figure 3, 6C) where the two rings are tilted by -56.0 • in close agreement to those observed for PDB 3MCY [23] (-36.21 • ) and PDB 5F2F [23] (-37.47 • ) [24]. Additionally, the C5-C6 bond was in the tg conformation.
Clearly, for the situation seen when 6 crystallized alone, the self-assembly into a dimeric form becomes a dominant factor which could not be seen in either the calculated values nor with the FimH binding site. Other significant differences were observed between the three forms denoted 6A, 6B, and 6C for the crystalline states and calculated one, respectively:  The fact that FimH is known to adopt three different conformations at the tyrosine gate (Tyr48-Tyr137; open, close, half open) depending to which ligands it is bound to, constitutes an additional example of molecular adaptation that has been previously observed with C-linked oligomannoside analogues [47,48], a process of conformational selection suggesting that the FimH likely binds these glycomimetics similarly to the way that it binds the natural analogues [24,49].
Contrary to the above situation for the O-and C-linked analogues 5 and 6, we could not obtain a crystalline structure for the S-linked analogue 12. However, it is well known that thioglycosides also obey to the endo-as well as the exo-anomeric effects [50]. Therefore; we undertook to perform the above Monte Carlo and DFT molecular dynamic calculation on 12 to glance at its most likely 3D structure. As anticipated, the C1-S and the S-C1′ bonds length were 1.862 and 1.853 Å, respectively, longer than the equivalent C1-O and O-C1′ bonds lengths in 5 (1.399 Å, 1.496 Å). Moreover, the Φ was calculated to be -45.93° in good agreement with the exo-anomeric effect for 5 (Φ= -57.69°) ( Figure  4).

Relative FimH Binding Affinity
It has been previously observed that the analogous O-, C-and S-linked heptyl α−Dmannopyranosides had IC50s in the low nanomolar range when measured by surface plasmon resonance (SPR) against the E. coli FimH [20]. However, the order was O-(IC50 = 160 nM), C-(IC50 = The fact that FimH is known to adopt three different conformations at the tyrosine gate (Tyr48-Tyr137; open, close, half open) depending to which ligands it is bound to, constitutes an additional example of molecular adaptation that has been previously observed with C-linked oligomannoside analogues [47,48], a process of conformational selection suggesting that the FimH likely binds these glycomimetics similarly to the way that it binds the natural analogues [24,49].
Contrary to the above situation for the Oand C-linked analogues 5 and 6, we could not obtain a crystalline structure for the S-linked analogue 12. However, it is well known that thioglycosides also obey to the endoas well as the exo-anomeric effects [50]. Therefore; we undertook to perform the above Monte Carlo and DFT molecular dynamic calculation on 12 to glance at its most likely 3D structure. As anticipated, the C1-S and the S-C1 bonds length were 1.862 and 1.853 Å, respectively, longer than the equivalent C1-O and O-C1 bonds lengths in 5 (1.399 Å, 1.496 Å). Moreover, the Φ was calculated to be -45.93 • in good agreement with the exo-anomeric effect for 5 (Φ = -57.69 • ) (Figure 4). The fact that FimH is known to adopt three different conformations at the tyrosine gate (Tyr48-Tyr137; open, close, half open) depending to which ligands it is bound to, constitutes an additional example of molecular adaptation that has been previously observed with C-linked oligomannoside analogues [47,48], a process of conformational selection suggesting that the FimH likely binds these glycomimetics similarly to the way that it binds the natural analogues [24,49].
Contrary to the above situation for the O-and C-linked analogues 5 and 6, we could not obtain a crystalline structure for the S-linked analogue 12. However, it is well known that thioglycosides also obey to the endo-as well as the exo-anomeric effects [50]. Therefore; we undertook to perform the above Monte Carlo and DFT molecular dynamic calculation on 12 to glance at its most likely 3D structure. As anticipated, the C1-S and the S-C1′ bonds length were 1.862 and 1.853 Å, respectively, longer than the equivalent C1-O and O-C1′ bonds lengths in 5 (1.399 Å, 1.496 Å). Moreover, the Φ was calculated to be -45.93° in good agreement with the exo-anomeric effect for 5 (Φ= -57.69°) ( Figure  4).

Relative FimH Binding Affinity
It has been previously observed that the analogous O-, C-and S-linked heptyl α−Dmannopyranosides had IC50s in the low nanomolar range when measured by surface plasmon resonance (SPR) against the E. coli FimH [20]. However, the order was O-(IC50 = 160 nM), C-(IC50 =

Relative FimH Binding Affinity
It has been previously observed that the analogous O-, Cand S-linked heptyl α−D-mannopyranosides had IC 50 s in the low nanomolar range when measured by surface plasmon resonance (SPR) against the E. coli FimH [20]. However, the order was O-(IC 50 = 160 nM), C-(IC 50 = 424 nM), and the S-linked (IC 50 = 547 nM), thus showing that the O-linked analogue was still unsurpassed and that the nature of the aglyconic heteroatoms had a critical effect on the binding affinities.
To evaluate the relative binding affinities of our own series of compounds 5, 6, and 12, we undertook surface plasmon resonance (SPR) measurements using an improved competition method. To this end, we first immobilized 6-aminohexyl α-D-mannopyranoside (4) on the CM5 gold chip using the conventional carbodiimide method to which we allowed FimH to bind with and without our competitive inhibitors (see sensorgram of 4 in the SI). The equilibrium dissociation constant (Kd) of the interaction was then normalized, the data sets are illustrated in Figure 5 and values given in Table 1 (see experimental details). This modification from the previous protocol [18] using 8-aminooctyl α-D-mannoside was found much more practical and reliable given its very high affinity to FimH which made the gold chip regeneration from sample to sample too complicated (very slow k off ). To evaluate the relative binding affinities of our own series of compounds 5, 6, and 12, we undertook surface plasmon resonance (SPR) measurements using an improved competition method. To this end, we first immobilized 6-aminohexyl α-D-mannopyranoside (4) on the CM5 gold chip using the conventional carbodiimide method to which we allowed FimH to bind with and without our competitive inhibitors (see sensorgram of 4 in the SI). The equilibrium dissociation constant (Kd) of the interaction was then normalized, the data sets are illustrated in Figure 5 and values given in Table 1 (see experimental details). This modification from the previous protocol [18] using 8aminooctyl α-D-mannoside was found much more practical and reliable given its very high affinity to FimH which made the gold chip regeneration from sample to sample too complicated (very slow koff).  The binding data of methyl- (MeαMan, 1), p-nitrophenyl α-D-mannopyranoside (PNPαMan, 2), and heptyl-(heptylαMan, 3), were also included for comparison purpose. Their relative binding affinities are in the same order as those usually measured by other SPR methods [46], thus justifying the validity of our new procedure. As can be seen, the C-linked analog 6 was a better ligand (Kd = 11.45 nM) than the corresponding O-linked (5) (Kd = 19.90) and the S-linked (12) (Kd = 94.4 nM). Given the subtle changes in the 3D structures between 5 and 12, the difference in their respective Kds  The binding data of methyl- (MeαMan, 1), p-nitrophenyl α-D-mannopyranoside (PNPαMan, 2), and heptyl-(heptylαMan, 3), were also included for comparison purpose. Their relative binding affinities are in the same order as those usually measured by other SPR methods [46], thus justifying the validity of our new procedure. As can be seen, the C-linked analog 6 was a better ligand (Kd = 11.45 nM) than the corresponding O-linked (5) (Kd = 19.90) and the S-linked (12) (Kd = 94.4 nM). Given the subtle changes in the 3D structures between 5 and 12, the difference in their respective Kds is somewhat surprising (about 5 fold). However, the situation is similar for the related series of Oand S-linked heptyl α-D-mannopyranosides (see above, 3.4 fold). The major difference here is that the C-linked analog 6 was better than the other two and close to some of the best known E. coli FimH antagonists [10,18]. This could be attributed to a better π-π stacking of the biphenyl substituent of 6 within the tyrosine gate.

Materials and Methods
Reactions were carried out under Nitrogen using commercially available ACS grade solvents which were stored over 4Ǻ molecular sieves. Solutions in organic solvents were dried over anhydrous Na 2 SO 4 , filtered, and concentrated under reduced pressure. Reagents were obtained from Sigma Aldrich (Toronto, ON, Canada). The FimH residues 1-158 were isolated from the E. coli strain UTI89. The source was obtained from a synthetic gene expressed into pET24a [51][52][53]. Methyl α-D-mannopyranoside (Me Man, 1) and para-nitrophenyl α-D-mannopyranoside (PNP Man, 2) are commercially available from Alfa Aesar (Tewksbury, MA, USA) and Sigma Aldrich, respectively. Heptyl α-D-mannopyranoside (Heptyl Man, 3) was prepared according to published procedure [18,53]. Melting points were measured on a Fisher Jones apparatus and are uncorrected. Reactions were monitored by thin-layer chromatography using silica gel 60 F254 coated plates (E. Merck). NMR spectra (Merck, Darmstadt, Germany). Were recorded on Varian Inova AS600 and Bruker Avance III HD 600 MHz spectrometer. Proton and carbon chemical shifts (δ) are reported in ppm relative to the chemical shift of residual CHCl 3 , which was set at 7.28 ppm ( 1 H) and 77.16 ppm ( 13 C{ 1 H}). Coupling constants (J) are reported in Hertz (Hz)

SPR Solution Affinity of Ligands for the E. coli FimH Adhesin
First, the equilibrium dissociation constant (K D ) of FimH against immobilized 6-aminohexyl α-D-mannopyranoside (4) was determined. The ligand 6-aminohexyl α-D-mannopyranoside (10 mM in H 2 O) was immobilized via standard amine carbodiimide coupling chemistry (and blocking with ethanolamine as per the manufacturer's procedure) to obtain 364 response units (RU) (see sensorgram in SI). The reference channel was treated similarly without ligand. A concentration series of the analyte FimH (1024 to 8 nM in running buffer, 2-fold dilutions) was then injected over the surface (700 s association time; 1200 s dissociation time; 30 µL/min flow rate; 25 • C temperature; running buffer: PBS pH 7.45, 3 mM EDTA supplemented with 0.005% Tween-20; regeneration: 30 s of 10 mM Glycine pH 1.5) and the response was double-referenced by subtracting the reference channel and the response with no FimH (running buffer only). Since equilibrium could not be reached during injection, K D was obtained from the extracted kinetic rate constants (K D = k off / k on ) from a non-linear globally fitted 1:1 Langmuir binding isotherm containing a mass transfer constant (t c ) and locally fitted refractive index constant (RI) (Biacore T200 evaluation software v1.0, GE Healthcare Life Sciences) of the obtained sensorgrams.
Second, the equilibrium dissociation constants (K D ) of FimH against the compounds of interest were determined by measuring the amount of free FimH in a series of concentrations of the compounds of interest at a constant concentration of FimH (close to the Kd of FimH: 6-aminohexyl α-D-mannopyranoside) after equilibrium in solution. A concentration series of the ligand of interest (65536 to 0.5 nM in running buffer, 2-fold dilutions) was equilibrated overnight with analyte FimH (40 nM) and then injected over the previous ligand 6-aminohexyl α-D-mannopyranoside-immobilized surface (300 s association time; 300 s dissociation time; 30 µL/min flow rate; 25 • C temperature; running buffer: 5% DMSO, PBS pH 7.45, 3 mM EDTA supplemented with 0.05% Tween-20; regeneration: 30 s of 10 mM Glycine pH 1.5) and the response was single-referenced by subtracting the reference channel and solvent corrected with a DMSO calibration curve. Since equilibrium could not be reached during injection, the responses (free [FimH], Y-axis) at the end of the injections were normalized with the response from FimH only ([compound of interest] = 0) and then plotted over the compounds of interest concentrations (X-axis). The obtained curves were then non-linearly fitted according to the following equation [53] using GraphPad Prism (v6.07, GraphPad Software, San Diego, CA, USA) to extract R max and K S : where: Y is the concentration of free FimH (RU); R max is the binding capacity of the SPR surface at saturation of all binding sites; K C and K S are the equilibrium dissociation constants (K D ) for binding of FimH to immobilized 6-aminohexyl α-D-mannopyranoside ligand on the sensor surface and with the compound of interest in solution, respectively; [FimH] is the constant concentration of FimH (40 nM); X is the concentration of the compound of interest (nM).

Conclusions
The work presented herein which describes the syntheses, tridimensional structures, binding parameters using a modified SPR procedure of a series of O-, C-, and S-linked mannopyranoside derivatives incorporating the identical 1,1 -biphenyl pharmacophore but diverse aglyconic atoms illustrates the necessity to explore this aspect when exploring novel E. coli FimH antagonists. The fact that our series of biphenyl analogues differs from those previously obtained from the corresponding heptyl mannopyranosides, clearly justify the deepen analysis undertaken in this report. Clearly, this study also confirms the fact that our C-linked α-D-mannopyranoside unambiguously exists in its preferred 4 C 1 chair conformation as opposed to several claims in the literature. It also further confirms that S-glycoside to possess an exo-anomeric effect. Even though, none of the compounds exhibited higher affinities than the golden O-linked heptyl α-D-mannopyranoside standard, this work expanded our knowledge on the E. coli FimH tyrosine gate which showed that further binding contacts are worth exploring. This fact on its own strongly argues for revisiting some of the recent best antagonists identified thus far and that they can be further optimized. Further studies in vitro and PK/PD evaluation will warrant deepen our approach toward truly therapeutic candidates. Allyl 2,3,4,6-tetra-O-acetyl-1-thio-α-D-mannopyranoside (9). To a solution of peracetylated mannose (8), (405 mg, 1.12 mmol) in 2 mL of dried methylene chloride was added, under N 2 , allyl mercaptan (374 µL, 4 equiv.). The mixture was cooled at 0 • C and BF 3 Et 2 O (1.4 mL, 10 equiv.) was added. The reaction mixture was stirred at room temperature for 20 h. The reaction was quenched with 2 mL of water under stirring, then diluted with methylene chloride and was washed with 30 mL of 5% aqueous solution of NaHCO 3 . The organic layer then dried over Na 2 SO 4 , and concentrated under reduced pressure. The residue was purified by silica gel column chromatography using Hexane/ EtOAc (7:3) to afford compound 9 as a colorless oil with Rf = 0.35 (Hexane/ EtOAc (7:3) (2E)-3-(1,1 -biphenyl-2-propen-1-yl) 2,3,4,6-tetra-O-acetyl-1-thio-α-D-mannopyranoside (11). To a solution of 9 (53 mg, 0.131 mmol) in dry DCM (2 mL) were added to 4-vinyl-1,1 -biphenyl (10) (28 mg, 1.2 equiv), and 10% tricyclohexylphosphine[1,3-bis(2,4,6-trimethylphenyl)imidazol-2-ylidene] [2-thienylmethylene]ruthenium(II) dichloride (Grubbs' catalyst) under N 2 . The solution was refluxed