To Fold or Not to Fold: Diastereomeric Optimization of an α-Helical Antimicrobial Peptide

Membrane disruptive α-helical antimicrobial peptides (AMPs) offer an opportunity to address multidrug resistance; however, most AMPs are toxic and unstable in serum. These limitations can be partly overcome by introducing D-residues, which often confers protease resistance and reduces toxicity without affecting antibacterial activity, presumably due to lowered α-helicity. Here, we investigated 31 diastereomers of the α-helical AMP KKLLKLLKLLL. Three diastereomers containing two, three, and four D-residues showed increased antibacterial effects, comparable hemolysis, reduced toxicity against HEK293 cells, and excellent serum stability, while another diastereomer with four D-residues additionally displayed lower hemolysis. X-ray crystallography confirmed that high or low α-helicity as measured by circular dichroism indicated α-helical or disordered structures independently of the number of chirality switched residues. In contrast to previous reports, α-helicity across diastereomers correlated with both antibacterial activity and hemolysis and revealed a complex relationship between stereochemistry, activity, and toxicity, highlighting the potential of diastereomers for property optimization.


■ INTRODUCTION
Membrane disruptive antimicrobial peptides (AMPs), which occur naturally as part of the innate immune system, offer an opportunity to address multidrug-resistant (MDR) bacteria because of their unspecific mechanism of action, against which resistance does not occur easily. 1−3 Such AMPs are however unstable in serum and most often toxic owing to their membrane disruptive amphiphilic and usually α-helical structure triggering their antibacterial effect. Their properties can be improved by sequence optimization, 4−7 whereby the most versatile approach consists in introducing non-natural structural elements 8 such as D-amino acids, 9−13 non-natural residues, 14 βor γ-amino acids, 15,16 isopeptide bonds, 17 or entirely non-peptidic elements such as spermine 18 or fatty acids. 19,20 A complete redesign of AMPs is also possible in the form of dimers, 21 cyclic or bicyclic staples, 22−24 small molecules, 25 peptoids, 26,27 foldamers, 28 or dendrimers. 29,30 For α-helical AMPs and analogues, the toxicity reduction effect observed upon introducing D-residues or similar perturbations, often measured as lower lysis of red blood cells, is generally attributed to a reduced α-helical folding, which would block pore formation on the membrane surface as a trigger for hemolysis. On the other hand, coating and destabilization of the bacterial membrane, and therefore the antibacterial effect, would still be possible with the modified peptide in the absence of folding. 12,31−33 However, very little structural evidence or systematic studies support the hypothesis that reduced α-helical folding should generally preserve antibacterial activity while reducing toxicity.
In our own search for new antibacterial compounds, we have discovered several AMP dendrimers (AMPDs) with very low hemolysis and strong activity against Gram-negative bacteria including MDR clinical isolates. 34−37 By investigating stereorandomized sequences, which are obtained by solid-phase synthesis using racemic building blocks and consist of a mixture of all possible diastereomers, we found that stereorandomized (sr-) AMPDs also exhibit strong antibacterial effects and very low hemolysis, suggesting an intrinsically disordered bioactive conformation. 38,39 The same effect was observed with the intrinsically disordered AMP indolicidin 40 but not with α-helical linear AMPs such as DJK-5, 41 which lost their activity when stereorandomized. 38 In a separate series of experiments with antimicrobial bicyclic peptides, 22,42 we discovered a short membrane disruptive antibacterial but somewhat hemolytic linear undecapeptide, KKLLKLLKLLL (ln65), which did not appear, even as partial sequence, in databases of AMPs, 43,44 proteins, 45 or ChEMBL (Figure 1). 46 The activity of this AMP was preserved upon inverting its four lysine residues to Denantiomers to form kkLLkLLkLLL (ln69), while its hemolysis was strongly reduced. 47 Strikingly, both the all-L sequence ln65 and its diastereomer ln69 were strongly αhelical, as established by circular dichroism (CD) and X-ray crystallography, showing that in this case lowered hemolysis was not related to a reduced α-helical folding. Intrigued by this observation, we set out to prepare and test the stereorandomized version sr-ln65 as well as multiple diastereomers of ln65 in search for analogues with possibly improved activity and/or reduced toxicity. Systematic studies of multiple diastereomers have shown significant activity modulations in the case of short, non-helical arginine−tryptophan containing AMPs. 48−50 ■ RESULTS AND DISCUSSION Enantiomeric and Stereorandomized Sequences. We first investigated dln65 and dln69, the enantiomers of ln65 and its diastereomer ln69, to check that they displayed similar activities as expected for enantiomeric membrane disruptive AMPs (Table S1). CD spectra of dln65 and dln69 in aqueous phosphate buffer in the presence of either 5 mM dodecylphosphocholine (DPC), which forms micelles mimicking a membrane environment, 52 or 20% trifluoroethanol (TFE) as a folding inducer, 53,54 were mirror images from those of the L-enantiomers and confirmed their α-helical folding (Figure 2a,b). The enantiomeric pair ln65/dln65 gave essentially the same minimal inhibitory concentration (MIC) values against the five bacterial species used in this study (Pseudomonas aeruginosa, Klebsiella pneumoniae, Acinetobacter baumannii, Escherichia coli, and methicillin-resistant Staphylococcus aureus), as well as the same minimum hemolytic concentration (MHC) on human red blood cells (hRBCs) indicating significant hemolysis (125 μg/mL, Table 1, Figure  S1 and Table S2). In line with these activities, the membrane disruptive effects of both enantiomers on fluorescein-loaded vesicles 55 made of the anionic egg yolk phosphatidyl glycerol (EYPG) mimicking bacterial membranes as well as on vesicles made of zwitterionic egg yolk phosphatidyl choline (EYPC) mimicking eukaryotic membranes were comparably strong  . CD-spectra of ln65, dln65, ln69, dln69, sr-ln65, sr-ln65L 6 , HP5, HP7, and HP9, measured with 0.1 mg/mL peptide in phosphate buffer at pH 7.4 with 10 and 20% v/v 2,2,2-TFE and with 5 mM DPC of (a) ln65 (full lines) and dln65 (dashed lines), (b) ln69 (full lines) and dln69 (dashed lines), (c) sr-ln65 (full lines) and sr-ln65L 6 (dashed lines), (d) HP5, (e) HP7, and (f) HP9.
( Table 1, columns 10 and 11 and Figure S2). A similar behavior of ln65 and its enantiomer dln65 was consistent with membrane disruption as the primary mechanism of action for these α-helical AMPs. On the other hand, despite the mirror image CD-spectra and comparable vesicle leakage activities of ln69 and dln69, dln69 was four-fold more antibacterial and hemolytic than ln69, which might reflect an additional activity of dln69 unrelated to its membrane activity.
To further probe if α-helical folding was required for activity, we prepared the fully stereorandomized sequence sr-ln65, a racemic mixture of the 1024 possible diastereomers, as well as sr-ln65L 6 with pure L-leucine at position 6 of the sequence, containing all 1024 diastereomers with single chirality at position 6 such as to make a possible folding detectable by CD. Remarkably, both sr-ln65 and sr-ln65L 6 were as antibacterial as ln65 but much less hemolytic, an effect comparable to our previous observation with AMPDs and sr-AMDPs, suggesting that the antibacterial bioactive conformation of ln65 might be disordered while the hemolytic bioactive conformation would be α-helical. 38,39 However, while CD spectra of sr-ln65 were nearly flat as expected because the stereorandomized sequence is racemic, those of sr-ln65L 6 showed approximately 17% αhelix content in 5 mM DPC or with TFE, suggesting that a significant fraction of the 1024 possible diastereomers of ln65 might be α-helical ( Figure 2c). Therefore, the activity of sr-ln65 might also be explained by the presence of some highly active and α-helical diastereomers, such as ln69, mixed with inactive and possibly disordered diastereomers.
Diastereomers and Mutants of ln65. In view of these preliminary experiments, we set out to test a series of diastereomers of ln65 for their α-helicity and antibacterial and hemolytic effects. From the 1024 possible diastereomers, 11 (0.1%) sequences are possible with a single inverted chirality residue, 55 (5.4%) with two, 165 (16.1%) with three, 330 (32.2%, including ln69) with four, and 462 (45.1%) with five inverted chirality residues. Balancing our interest to investigate diastereomers with multiple D-residues related to ln69 with the expectation that α-helical folding was more likely to be preserved with only a few inverted chirality residues, 56,57 we selected 31 diastereomers HP1−HP31, one (3%) with a single D-residue, five (16%) with two D-residues, four (13%) with three D-residues, 13 (42%) with four D-resides, and eight (26%) with five D-residues, distributing D-residues in groups or scattered, at N-or C-termini, or in the middle of the sequence ( Table 1).
Diastereomers with four and five D-residues (HP11−HP31) were generally less active against bacteria, especially against A. baumannii, K. pneumoniae, and MRSA, although they all kept significant EYPG vesicle leakage activities, reflecting the fact that vesicle leakage activity is often not sufficient for antibacterial effects to occur due to the much more complex nature of bacteria compared to lipid vesicles. Furthermore, these diastereomers mostly lost their hemolytic activity in proportion to their low EYPC vesicle leakage activities, except for HP16, HP24, and HP29, which, like HP9, showed significant hemolysis despite being non-helical and inactive on EYPC vesicles. The least active peptides were HP13 with four D-residues and HP25 with five D-residues. Both peptides retained some activity against E. coli, P. aeruginosa, and A. baumannii (MIC = 4−32 μg/mL) but were inactive against K. pneumoniae and MRSA, were non-hemolytic, and were not αhelical (7 and 11% in 5 mM DPC). Gratifyingly, one peptide with four D-residues, HP19, was as strongly antibacterial and low hemolytic as the previously identified ln69 with four Dresidues. Another peptide with four D-residues, HP11 (MIC = 2−8 μg/mL), was even slightly more antibacterial than ln69, although slightly more hemolytic (MHC = 62.5 μg/mL). HP11 and HP19 were among the most α-helical in this set (52−55% in 5 mM DPC) although not as much as ln69 (61%).
To compare the effects of diastereomeric changes with more classical sequence variations, we performed conservative mutations in ln65 and ln69 by mutating all lysines to arginines, all leucines to isoleucines, or both, preserving their chirality pattern. In this series, the Lys → Arg exchanges (ln65 → HP32 and ln69 → HP33) preserved α-helicity, antibacterial activity, and EYPG vesicle leakage, but increased hemolysis and EYPC vesicle leakage, which might be related to the better cell-penetrating properties of poly-arginines versus poly-lysines attributed to stronger binding to phospholipids. 60 On the other hand, Leu → Ile exchanges (ln65 → HP34, ln69 → HP35, HP32 → HP36, HP33 → HP37) led to reduced antibacterial effects and in part lower hemolysis, accompanied by slightly lower α-helicity as expected since Leu stabilizes and Ile destabilizes α-helices. 61 Surprisingly, dimerization of ln65 to 2ln65 and ln69 to 2ln69 gave peptides that were strongly αhelical and hemolytic but entirely inactive against bacteria. Vesicle leakage activities were generally high for EYPG vesicles and partially followed hemolysis trends for EYPC.
Taken together, these experiments showed that diastereomers of ln65 featured new analogues with interesting activity profiles, while other simple modifications such as Lys → Arg, Leu → Ile mutations or dimerization were not as profitable. For further evaluation, we selected the most strongly antibacterial diastereomeric AMPs irrespective of their hemolytic properties, namely, ln65, ln69, dln69, and all diastereomers HP1−HP11 except HP6 and HP9. These AMPs showed good activities (MIC = 2−8 μg/mL) against additional Gram-negative and Gram-positive bacteria including several drug-resistant P. aeruginosa variants, 62 although none of them were active against Burkholderia cenocepacia, a Gramnegative bacterium which is naturally resistant to AMPs like colicin (Table 2). 63 Furthermore, most diastereomers were much more stable against serum degradation than the full Lsequence ln65 (Figure 3a). Interestingly, inverting the chirality of only the N-and C-termini (ln65 → HP5) was sufficient to entirely stabilize the peptide in line with the non-recognition of D-amino acids by proteases preventing the proteolysis from peptide extremities. On the other hand, dln69 with 7 D-leucine residues was entirely degraded due to proteolytic scission at the N-terminal L-lysine residue presumably from trypsin-like proteases ( Figure S4).
While most of these strongly antibacterial diastereomers were equally or more hemolytic than the full L-sequence ln65, they showed reduced cytotoxicity against human embryonic kidney HEK293 cells (Figure 3b). Diastereomer HP7, which was the most active AMP against bacteria, showed the lowest toxicity in the series (IC 50 = 128 ± 5 μM). Furthermore, diastereomers were generally toxic against A549 lung cancer cells, with HP11 showing the strongest toxicity (IC 50 = 3.6 ± 0.1 μM), in line with the fact that many AMPs are often active against cancer cells ( Figures S5 and S6). 64,65 The observed differences between diastereomers in hemolysis, toxicity against HEK293 cells or A549 lung cancer cells, are probably caused by diastereomeric interactions with the different membrane components of the different cell types and possibly proteins in the cell culture medium. 66 X-ray Crystallography. To establish whether the CD signal observed with diastereomeric AMPs was indeed caused by α-helical folding, we prepared derivatives with their Ntermini acylated with an α-C-fucosylacetyl group for crystallization as complexes with lectin LecB, 67 an approach which we have successfully used for oligonucleotides, 68 cyclic, 69 bicyclic, 47 and linear peptides, 70 as well as for peptide dendrimers. 71,72 We considered the nine most potent diastereomers detailed above, the Lys → Arg mutants HP32 and HP33, and the inactive, alternating chirality diastereomers HP30 and HP31. Crystallization screening provided good diffracting LecB crystals with well-resolved ligand electron density for complexes with the fucosylated analogues FHP5, FHP8, FHP30, and FHP31 (Table 3).
In the X-ray crystal structure of FHP5 in complex with LecB, the undecapeptide was visible in full α-helical conformation in two of the four different fucose-binding sites present in the asymmetric unit, while the other two fucosebinding sites only showed electron density for the fucosyl group, probably due to a disordered conformation (PDB 8AN9, 1.3 Å resolution, Figure 4a, Table S3 and Figure S7). The two α-helices in the well-resolved binding sites are superimposable and interact through intermolecular hydrophobic interactions between leucine side chains ( Figure 4b).
We observed a similar situation for the structure of FHP8 in complex with LecB (PDB 8ANO, 1.3 Å resolution, Figure 4c,d, Table S4 and Figure S8). Both structures were very similar to the previously reported structure of fucosylated dln69 with LecB. 47 Although the inactive undecapeptides HP30 and HP31 with alternating L-and D-residues in their sequences had almost the same number of L-and D-residues, their flat CD spectra most likely indicated a disordered conformation considering that an Table 2. Antimicrobial Activity of Diastereomeric AMPs e Cpd.   Table S5 and Figure S9). A similar situation was observed in the LecB complex with fucosylated HP31 containing four different asymmetric units. In this case, only two of them were completely resolved and showed unordered conformations interacting with LecB via H-bonds but also with symmetrical peptides (PDB 8AOO, 1.2 Å resolution, Table S6 and Figure S10). Molecular Dynamics. To further investigate the α-helical folding of our diastereomers, we performed molecular dynamics (MD) simulations over 250 ns using GROMACS 73 starting from a pre-folded α-helical structure in water with or without a DPC micelle. For active diastereomers such as HP5 in the presence of DPC micelles, the peptide first entered in contact with the micelle surface by salt bridges between lysine side chain ε-ammonium groups and phosphate groups of DPC and later remained in an α-helical conformation at the micelle surface (Figure 5a,b). The peptide did not deviate significantly from the starting α-helical conformation (Figure 5d) and retained the full set of backbone H-bonds (Figure 5e).
In water by contrast, the α-helix of HP5 completely and irreversibly unfolded to an unordered conformation ( Figure  5c). This unordered conformation strongly differed from the starting α-helix (Figure 5a) with complete loss of backbone Hbonds (Figure 5b). Similar results were obtained for the other active compounds (HP1, HP2, HP3, HP4, HP7, HP8, HP10, and HP11, Figures S11−S19). For the inactive, non-helical diastereomers HP16 and HP29 by contrast, the starting αhelical conformation rapidly unfolded to an unordered conformation with complete loss of backbone H-bonds even in the presence of the DPC micelle ( Figures S20 and S21).
Statistical Analysis. In view of the structural studies above showing in several cases that the degree of α-helicity of ln65 diastereomers as measured by CD corresponded to an observable α-helix or structural disorder, we assumed that the CD signal could be used as indication of folding across the entire series. Strikingly, increasing α-helicity (% in 5 mM DPC) was linearly correlated with increasing antibacterial activity measured as log 2 (MIC) against K. pneumoniae (r 2 = 0.57), A. baumannii (r 2 = 0.59), and MRSA (r 2 = 0.62), but to a lesser extent with activity against P. aeruginosa (r 2 = 0.41) and with hemolysis (r 2 = 0.37), and only quite poorly with activity against E. coli (r 2 = 0.29) against which most diastereomers were active (Figures 6a,b and S22).
To gain an overview of the series, we performed principal component analysis of the complete data set of antimicrobial activity, hemolysis, and folding under the different conditions measured. The first principal component PC1 covered 64% of data variance and reflected the variation of antimicrobial activities, hemolysis, and vesicle leakage activities with αhelicity measured in any of the four conditions (Figure 6c). The second principal component PC2 covered another 10% of data variance and reflected a modulation of antimicrobial activities with α-helicity and EYPG vesicle leakage independent of hemolysis and EYPC vesicle leakage. The distribution of the diastereomers on the (PC1 and PC2) plane separated active from inactive compounds from right to left and separated the two most active diastereomers identified, HP5 and HP7, from the majority of tested diastereomers (Figures 6d,e and S23). Both optimized AMPs stood out by their increased antimicrobial activity, which was particularly strong against A. baumannii, while keeping a moderate level of hemolysis, and reduced toxicity against HEK293 cells.  One letter code for amino acids. * = α-L-fucosyl-acetyl. Many high-quality LecB crystals were also obtained in complex with fucosylated ln65R and ln69r; however, in these two cases, electron density only revealed the L-fucose and the adjacent two arginine residues.

■ CONCLUSIONS
The above experiments with diastereomers of undecapeptide ln65 supported by X-ray crystallography show that this αhelical AMP preserves folding and activity across many of its diastereomers. In contrast to previous studies of diastereomers focused on cases with reduction in toxicity and preservation of antibacterial effects, our study across a broad set of diastereomers shows that introducing D-residues in an αhelical AMP can affect antibacterial effects at least as much as toxicity as measured by hemolysis. Although these activities were correlated, sufficient variability was available to identify two diastereomers with improved properties, HP5 and HP7, as two AMPs with increased antibacterial effects compared to the full L-AMP ln65, and moderate hemolysis and reduced toxicity against HEK293 cells.
In the present study, the preservation of folding and activity across many diastereomers of ln65 was anticipated by characterizing its stereorandomized version sr-ln65L 6 , which showed only modest reduction in antibacterial effects and significant α-helicity. Identifying active diastereomers might be more difficult for other α-helical AMPs if they lose their activity in the stereorandomized form as reported for DJK-5 (vqwrairvrvir) and SB1 (KYKKALKKLAKLL). 38 Testing the stereorandomized sequence might therefore be the first step to address diastereomeric optimization of other AMPs.  On-Beads Sugar Coupling and Deprotection for Fucosylated Compounds. Peracetylated α-L-fucosyl-acetic acid (3 equiv), Oxyma (3 equiv), and DIC (3 equiv) were dissolved in 6 mL of DMF. Double coupling (2 × 1 h) was performed at 50°C under nitrogen bubbling on resin. Deacetylation of sugar was performed directly onbead using a mixture of MeOH/H 2 O/NH 3 (8:1:1, v/v/v). Reaction was stirred overnight at room temperature. The reaction mixture was removed by filtration, and the resin was washed with DMF and MeOH before cleavage.
Cleavage from the Resin. Cleavage was carried out by treating the resins with 7 mL of a TFA/TIS/H 2 O (94:5:1, v/v/v) solution for 3 h at room temperature. The peptide solutions were precipitated with 25 mL of cold TBME, centrifuged for 10 min at 3500 rpm, evaporated, and dried with argon. Purification and Characterization. The dried crude was dissolved in a water/ACN mixture, filtered (pore size 0.22 μm), and purified by preparative RP-HPLC with gradients of 60 min. Fractions were analyzed by analytical LCMS. Peptides were obtained as white foamy solids after lyophilization and analyzed by both LCMS and HRMS. Yields were calculated for the TFA salts. All compounds are >95% pure by HPLC.
CD Spectroscopy. CD experiments were measured on a Jasco J-715 spectropolarimeter. All the experiments were performed using Hellma Suprasil 110-QS 0.1 cm cuvettes. For each peptide, the measurements were performed in phosphate buffer (PB, pH = 7.4, 7 mM), 10% TFE, 20% TFE, 5 mM DPC. The buffer was degassed for 10 min under high vacuum before each set of experiments. The concentration of the peptides was 0.100 mg/mL, and each sample was measured in one accumulation. The scan rate was 20 nm/min, pitch was 0.5 nm, response was 16 s, and bandwidth was 1.0 nm. The nitrogen flow was kept >8 L/min. After each measurement, the cuvettes were washed successively with milli-Q H 2 O and PB (pH 7.4).
The baseline was recorded under the same conditions and subtracted manually. Primary CD spectra were analyzed using DichroWeb 58 and Contin-LL method (set 4). 74 Antimicrobial Activity. Antimicrobial activity was determined for all peptides on E. coli W3110, P. aeruginosa PAO1, A. baumannii ATCC19606, K. pneumoniae NCTC418, and methicillin-resistant S. aureus COL and for selected peptides on P. aeruginosa PA14 and the polymyxin B-resistant derivatives PA14 4.13, PA14 4.18, and PA14 2P4 as well as the clinical isolates ZEM-1A and ZEM9A, K. pneumoniae OXA-48, Enterobacter cloacae, Stenotrophomonas maltophilia, B. cenocepacia, and Staphylococus epidermis. To determine the MIC, the broth microdilution method was used. 75 A single bacterial colony was grown in LB medium overnight at 37°C and 180 rpm shaking. The compounds were prepared as stock solutions of 2 mg/ mL in sterilized milliQ deionized water, added to the first well of 96well sterile, polypropylene round-bottom microtiter plates (TPP, untreated, Corning Incorporated, Kennebunk, USA), and diluted serially by 1/2. The concentration range tested was 0.5−64 μg/mL. The bacterial inoculum was prepared by measuring the absorbance of the overnight culture, which was diluted to concentration of the bacteria was quantified by measuring absorbance at 600 nm and diluted to an OD 600 of 0.022 in MH medium pH 7.4 (Sigma-Aldrich, Buchs, Switzerland). The sample solutions (150 μL) were mixed with 4 μL of diluted bacterial suspension with a final inoculation of about 5 × 10 5 CFU. For each test, two columns of the plate were kept for sterility control (MH medium only) and growth control (MH medium with bacterial inoculum, no compound). Positive control was carried out using either polymyxin B for Gram-negative or vancomycin for Gram-positive strains (starting with a concentration of 16 μg/mL) in MH medium. The plates were incubated at 37°C for 16−20 h under static conditions. 15 μL of 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide (MTT) 76 (1 mg/mL in sterilized milliQ deionized water) was added to each well, and the plates were incubated at room temperature until MTT staining was completed. The MIC was defined as the lowest concentration of the dendrimer that inhibits the visible growth of the tested bacteria (yellow) with the unaided eye.
Hemolysis Assay. To determine the MHC stock solutions, 8 mg/ mL peptide in PBS (pH 7.4) was prepared, and 50 μL was diluted serially by l/2 in 50 μL of PBS (pH 7.4) in a 96-well plate (Costar or Nunc, polystyrene, untreated). The concentration range tested was 15.6−2000 μg/mL. hRBCs were obtained by centrifugation of 1.5 mL of whole blood, from the blood bank of Bern, at 3000 rpm for 15 min at 4°C. Plasma was discarded, and the pellet was re-suspended in a 15 mL Falcon tube in 5 mL of PBS. The washing was repeated three times, and the remaining pellet was re-suspended in 10 mL of PBS. The hRBC suspension (50 μL) was added to each well, and the plate was incubated at room temperature for 4 h. MHC end points were determined by visual determination of the wells after the incubation period. Controls on each plate included a blank medium control (50 μL PBS + 50 μL of hRBCs suspension) and a hemolytic activity control (mQ-deionized water 50 μL + 50 μL hRBC suspension).
Vesicle Leakage Assay. 5(6)-carboxyfluorescein (CF) was purchased from Sigma. EYPC, EYPG, and a Mini-Extruder were purchased from Avanti Polar Lipids. Egg PC or Egg PG thin lipid layers were prepared by evaporating a solution of 100 mg of EYPC or EYPG in 4 mL of MeOH/CHCl 3 (1:1) on a rotary evaporator at room temperature and then dried in vacuo overnight. The resulting film was then hydrated with 2 mL of CF buffer (50 mM CF, 10 mM TRIS, 10 mM NaCl, pH 7.4) for 30 min at room temperature under stirring and then subjected to freeze−thaw cycles (7×) and extrusion (15×) through a polycarbonate membrane (pore size 100 nm). Extravesicular components were removed by gel filtration (Sephadex G-50) with 10 mM TRIS, 107 mM NaCl, pH 7.4 buffer. Final conditions: ∼2.5 mM PC or PG; inside: 50 mM CF, 10 mM TRIS, 10 mM NaCl, pH 7.4 buffer; outside: 10 mM TRIS, 107 mM NaCl, pH 7.4. PC or PG stock solutions (37.5 μL) were diluted to 3000 μL with a buffer (10 mM TRIS, 107 mM NaCl, pH 7.4) in a thermostated fluorescence cuvette (25°C) and gently stirred (final lipid concentration ∼31 μM). The CF efflux was monitored at λ em 517 nm (λ ex 492 nm) as a function of time after addition of the desired volume of peptide from 2 mg/mL stock in mQ water at t = 45 s, 10 and 50 μg/mL were monitored for both EYPC and EYPG. Finally, 30 μL of 1.2% Triton X-100 was added to the cuvette (0.012% final concentration) at t = 240 s to reach the maximum intensity. Fluorescence intensities were then normalized to the maximal emission intensity using I(t) = (I t − I 0 )/(I ∞ − I 0 ) where I 0 = I t at peptide addition, I ∞ = I t at saturation of lysis.
Time-Killing Kinetic Assay. A single colony of P. aeruginosa PAO1 was picked and grown overnight with shaking (180 rpm) in LB (Sigma-Aldrich, Buchs, Switzerland) medium 5 mL overnight at 37°C . The overnight bacterial culture was diluted to OD 600 0.002 (2 × 10 6 CFU/mL) in fresh MH medium. Stock solutions of AMPs in sterilized milliQ water were prepared in 1 mg/mL and were diluted to two times more than required concentration in fresh MH (Sigma-Aldrich, Buchs, Switzerland) medium at pH 7.4. 100 μL of prepared bacteria solution in MH and 100 μL of samples in MH were mixed in a 96-well microtiter plate (TPP, untreated, Corning Incorporated, Kennebunk, USA). Untreated bacteria at 1 × 10 6 CFU/mL were used as a growth control. 96-well microtiter plates were incubated in 37°C with shaking (180 rpm). Surviving bacteria were quantified at 0, 0.5, 1, 2, 3, 4, 5, and 6 h by plating 10-fold dilutions of the sample in sterilized normal saline on LB agar plates. LB agar plates were incubated at 37°C for 10 h, and the number of individual colonies was counted at each timepoint. The assay was performed in triplicate in the biosafety level 2 lab.
Serum Protein precipitates were pelleted under centrifugation, and supernatants were then sampled and analyzed by LC−MS. Experiment controls included two references, one known to be degraded and one known to be undegraded. Peaks corresponding to the internal standard and the undegraded peptides were integrated, with the ratio peptide/standard at t = 0 h as 100%.
Cell-Viability Assay. The viability of the cells was assessed with an AlamarBlue assay (ThermoFisher). Cells were seeded into 96-well plates, 4000 cells/well (HEK293) and 8000 cells/well (A549), the day before the experiment. Cells were then treated with the increasing concentration of the compound and incubated for 24 h at 37°C in the presence of 5% CO 2 . The next day, the medium was removed and replaced by a 10% AlamarBlue solution in full growth medium (DMEM or RPMI-1640). The cells were incubated for 3−5 h at 37°C with 5% CO 2 in a humidified atmosphere. The fluorescence was then measured on a Tecan Infinite M1000 Pro plate reader at λ ex 560 nm and λ em 590 nm. The value was normalized according to the untreated cells.
Crystallography Experiment and Data Acquisition. Suitable diffracting crystals were obtained via co-crystallization of the Cfucosylated derivatives with the bacterial lectin LecB. The sitting drop vapor diffusion method was used, screening 192 different conditions per compound. The lyophilized protein was dissolved in milli-Q water (5 mg/mL) in the presence of salts (6 mM CaCl 2 and MgCl 2 ). The peptides were added to the protein at a 5:1 molar excess related to the LecB lectin monomer. Crystals were obtained within 1−3 months after mixing 1.5 μL of the LecB ligand complex with 1.5 μL of reservoir solution and incubation at 18°C. All crystallization conditions were found in Index screens I/II (96 conditions) and Crystal Screen I/II (96 conditions) (Hampton Research, Laguna Niguel, CA, USA). Diffraction data were collected at the Paul Scherrer Institute (Villigen, Switzerland) on beamline X06DA PX-III using a DECTRIS PILATUS 2M-F detector and a multi-axis PRIGo goniometer. The structures were solved and visualized with the help of Phenix, 77 ccp4, 78 PyMol, 79 coot, 80 and XDS. 81 MD Simulation. MD simulations were performed using GROMACS 73 software version 2018.1 and the GROMOS53a6 force field. 82 The starting topologies were built from PyMOL. A dodecahedral box was created around the peptide 1.0 nm from the edge of the peptide and filled with extended simple point charge water molecules. Sodium and chloride ions were added to produce an electroneutral solution at a final concentration of 0.15 M NaCl. The energy was minimized using a steepest gradient method to remove Journal of Medicinal Chemistry pubs.acs.org/jmc Article any close contacts before the system was subjected to a two-phase position-restrained MD equilibration procedure. The system was first allowed to evolve for 100 ps in a canonical NVT (N is the number of particles, V the system volume, and T the temperature) ensemble at 300 K before pressure coupling was switched on, and the system was equilibrated for an additional 100 ps in the NPT (P is the system pressure) ensemble at 1.0 bar. MD in the Presence of DPC Micelle. MD simulations in the presence of a DPC (n-DPC) micelle were performed as follows. Parameters and references for the DPC molecule for the GROMOS53a6 forcefield are given in the Supporting Information. Peptides were manually placed at a distance from the pre-equilibrated micelle (of 65 DPC molecules) equal to the diameter of said peptide. Box, solvation, and NVT equilibration procedures were performed as explained previously. For each peptide/micelle system, 10 runs of 50 ns were generated to show the possibility for the peptide to either interact or diffuse away from the micelle. Then, runs of interest were extended up to 250 ns.
Clustering of Stable Structures. To obtain a representative conformer for each run, the last 100 ns (10001 frames) were clustered using an RMSD cutoff adapted to get a good balance between the number of clusters and the size of the main cluster. Many clusters combined with a very large percentage of structures in the top cluster is an indication of the stability of the one main conformer in each case. The PyMol Molecular Graphics System, version 1.8 (Schrodinger, LLC), was used to create structural models.
Peptides yields, MS and analytical data, CD spectra and analysis, vesicle leakage fluorescence curves, time-killing kinetics curves, full serum stability curves, full cytotoxicity data, data collection, refinement statistics, X-ray structures, topology of DPC molecules for MD simulations, MD simulation analyses, additional statistical analysis of ln65 diastereomers, structures, and HPLC−MS chromatograms and HRMS spectra for all compounds (PDF) SMILES and activity of all tested peptides (CSV)

Accession Codes
Coordinates and structure factors of refined LecB complexed with fucosylated peptides are available from the Protein data Bank with accession codes 8AN9 (FHP5), 8ANO (FHP8), 8ANR (FHP30), and 8AOO (FHP31 H.P. designed the project and carried out peptide synthesis, microbiological, hemolysis, serum stability and vesicle leakage assays, CD spectroscopy, MD simulations, and X-ray crystallography and wrote the paper. T.P. performed cytotoxicity assay and wrote the paper. S.F. and S.B. synthesized peptides, performed microbiology and hemolysis assays and CD spectroscopy. T.K. and C.v.D. supervised experiments with clinical and MDR strains. A.S. supervised Xray crystallography experiment. S.J. supervised MD studies and wrote the paper. J.-L.R. designed and supervised the study and wrote the paper.

Notes
The authors declare no competing financial interest. Raw data of LC−MS analyses for characterization and serum stability, HRMS, CD measurements, vesicle leakage assay, Xray crystallography, MD simulations, cytotoxicity assay, and pictures of 96-well plates for microbiological and hemolysis assays are accessible at: 10