Polymorphism of Kdo-based Glycolipids: the

1

Polymorphism of Kdo-based Glycolipids: the

Elaborately Determined Stable and Dynamic Bicelles

Yingle Feng,† ,§ Long Li,† Qiqige Du,† Lu Gou,⊥ Lei Zhang,⊥ Yonghai Chai,§ Ran Zhang,*‡

Tongfei Shi,‡ Guosong Chen*†,±

†The State Key Laboratory of Molecular Engineering of Polymers, Department of Macromolecular

Science, Fudan University, Shanghai 200433

‡Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun 130022

§Key Laboratory of Applied Surface and Colloid Chemistry, Ministry of Education and School

of Chemistry and Chemical Engineering, Shaanxi Normal University, Xi’an, Shaanxi 710119

⊥ MOE Key Laboratory for Nonequilibrium Synthesis and Modulation of Condensed Matter,

School of Science, Xi’an Jiaotong University, Xi’an 710049

±Multiscale Research Institute of Complex Systems, Fudan University, Shanghai, 200433

*Corresponding author: [email protected]; [email protected].

Page 1 of 27 CCS Chemistry

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 This article presented here has been accepted for publication in CCS Chemistry and is posted at the

© 2021 Chinese Chemical Society.request of the author prior to copyediting and composition.

mailto:[email protected]

mailto:[email protected]

2

ABSTRACT: Understanding how the diversity of glycolipids including their chemical structures

and composition affect their biological functions, is a remarkable fundamental challenge. In this

work, we employed a rare monosaccharide, Kdo (3-deoxy-D-manno-2-octulosonic acid) to build

up a simple and biomimetic model, for understanding the diversity of glycolipids from the

viewpoint of supramolecular chemistry. Kdo was chosen not only because its unusual 8-carbon

acidic carbohydrate backbone, which is very different from common hexoses; but also considering

its key structural role in lipopolysaccharides and wide existing nature in bacteria, plant and algae.

It was found that although both of the two Kdo-lipids S-Kdo and Kdo-S derived from the same

carbohydrate backbone and gave bicelles as their self-assembled morphology, experimental results

revealed that the self-assembly showed pathway complexity. Bicelle is the thermodynamic product

of S-Kdo, while for Kdo-S, bicelle is only a kinetically trapped state, which finally transforms to

ribbon. Molecular simulation clearly revealed the different packing of Kdo-lipids in the bicelles

with different contribution from hydrogen bonds and electrostatic interactions.

Keywords: Glycolipid, Kdo, self-assembly, bicelles

Introduction

Page 2 of 27CCS Chemistry



3

Achieving biomimetic complexity and diversity currently is aimed by ambitious supramolecular

chemists.1-3 It is known that lipid bilayer provides a functional barrier between subcellular

compartments and between cells and their environment.4,5 While the requirements for barrier

functions are not enough to explain the enormous polymorphism of lipid, which comes from not

only the chemical structure and composition, but also the structure and property of self-assembled

biomembranes.4 To understand such diversity and explore the generated different functions is a

fundamental challenge not only in biology, but also in material science.6-9

As a significant component of membrane, glycolipid holds a large proportion of diversity among

the lipid family, due to the structural diversity of carbohydrates. These carbohydrates play

indispensable roles from maintaining the membrane scaffold to mediating cell-cell

communications.9-11 The glycolipid structure can be even more diverse in the world of bacteria,

where the most significant example is the existence of a rare monosaccharide Kdo (3-deoxy-D-

manno-2-octulosonic acid).12 As an unusual 8-carbon acidic carbohydrate, Kdo is an essential

constituent of the cell wall lipopolysaccharides of Gram-negative bacteria, linking the lipid chain

with the outer membrane polysaccharides.13 As a nonrepeating core oligosaccharide, Kdo is highly

conserved in different bacteria, and even found in plant and algae.14 Concerning the structural

uncommonness of Kdo compared to other regular hexose, as well as the importance of Kdo in

kinds of biological processes, exploration and understanding of the self-assembly behavior of Kdo-

based glycolipids will provide us valuable reflection of polymorphism of glycolipid and reveal the

pathway complexity of biomembrane formation based on glycolipid.

Herein, we demonstrate the self-assembly behavior of Kdo-based glycolipids. Although

glycolipids have been employed to form various self-assembled structures,10, 15-17 as far as we

know, rare saccharides have not been explored as glycolipids for self-assembly due to the known




4

remarkable difficulty in their isolation or synthesis.5 In our study, by combination of experimental

and simulation results, we figure that the glycolipid structure significantly effects the self-assembly

pathway and the thermodynamic stable product of Kdo-based glycolipids, which indicates the

complexity of biomimetic self-assembly. The stability and mobility of bicelles formed by Kdo

derivatives can be greatly controlled by the structural changes, demonstrating the contribution of

this rare saccharide in biomembrane formation to an unprecedented level. This study will shed

light on the molecular mechanisms of rare saccharides in biomembrane formation, improving our

understanding on their dynamic behaviors and bringing inspiration to the design and preparation

of new glycolipid-based functional materials.

Figure 1. (a) Illustration of Gram-negative bacteria membrane and partial structure of LPS. (b)

Design and reaction condition of Kdo-based glycolipids and their use for membrane mimic.

Reaction condition: a) LiOH/H2O, MeOH; b) Stearylamine/myristylamine, EDCI, HOBt, DIPEA;

c) Pd/C, H2, MeOH; d) 0.1 M HCl, MeOH. e) PPh3, THF/H2O; f) Pentafluorophenyl

stearate/pentafluorophenyl myristate, DIPEA, DMF; g) NaOH/H2O, MeOH.




5

Results and discussion

Molecular design and synthesis of Kdo-based glycolipid. The monosaccharide 3-deoxy-D-

manno-oct-2-ulosonic acid (Kdo) was selected to construct the model glycolipid because it is a

very special sugar connecting lipid A and polysaccharide in bacteria with an unusual C8 backbone.

Meanwhile, the incorporation of Kdo is a vital step in the assembly of the protective outer

membrane of Gram-negative bacteria.12 From Figure 1, one may find the ‘polarity’ of Kdo, i.e.

different structures (lipid A and polysaccharides) are connected to its two different sides. The

‘polarity’ character of Kdo was further optimized here in the construction of our model glycolipid.

The alkyl chains were designed to be linked at the head (C-1) or end (C-8) of Kdo. To introduce

alkyl chain at the C-8 position, amine was designed to be installed for easy modification affording

the Kdo-Cn set of glycolipids. To make the Cn-Kdo compatible to Kdo-Cn, the amine modification

was remained as chloride salt form in the former set.

According to synthetic procedures, briefly, the two sets of Kdo-based glycolipids were prepared

by using compound I (Figure 1) as an important intermediate, which was firstly synthesized from

regular hexose mannose over 8 steps according to our previous work (Figure 1b).18 From the

structures shown in Figure 1, one may tell the great structural difference between the backbone of

mannose and Kdo. The carboxylic ethyl ester of I was firstly hydrolyzed with aqueous lithium

hydroxide (0.5 mol/L), then amidation with stearylamine or myristylamine followed by catalytic

hydrogenation and hydrochlorination provided positive charged glycolipid mimics S-Kdo and M-

Kdo. Meanwhile, the azide group of I was reduced using the Staudinger reaction, and saturated

fatty acid with different carbon chain (stearic acid, myristic acid) was then installed at the C-8

position. Ethyl ester of the resulting glycolipids were hydrolyzed with 0.5 M aqueous sodium




6

hydroxide to give Kdo-S and Kdo-M. The detailed synthetic procedures are reported in the

supporting information.

Bicelles formation and characterizations of assembled S-Kdo and Kdo-S. The self-assembly

behavior of glycolipids with long tail lengths, i.e. S-Kdo and Kdo-S were first investigated. Briefly,

the aqueous solution of samples with the concentration of 1 mg/mL were heated to 80 °C and then

cooled down to room temperature (details in the supporting information). The plate-shaped

assemblies were observed for both of the two compounds as confirmed by the transmission

electron microscopy (TEM) observation (Figure 2a-d). The radius of the plates for S-Kdo (Figure

2a, 2b) and Kdo-S (Figure 2c, 2d) varied from hundreds of nanometers to micrometers. As shown

in Figure 2a and 2c, no appreciable electron density difference could be found at the rim or in the

interior of the plate. Within the plate assemblies, the folded rim can be clearly seen from the TEM

images (pointed by the red arrow in Figure 2c). By using atomic force microscope (AFM), the

height of the plates was measured as 3.97 nm for S-Kdo (Figure 2b) and 3.95 nm for Kdo-S (Figure

2d). This height could be easily explained by the bilayer model, as the molecular length of the two

glycolipids is about 3.5 nm using Visual molecular dynamics (VMD)19 (Figure S1a), and the

bilayer with interleaved pattern yielding a layer thickness of around 4 nm (Figure S1b) matches

the experimental data (3.95 nm). The combination of TEM and AFM observation suggested that

these plates most likely were bicelles, instead of other morphologies like vesicles etc. Under the

same condition, the self-assembly of M-Kdo and Kdo-M in water was also investigated. The short

carbon chain could not support the regular packing as that in the bicelles, only spherical and

irregular micelles were afforded by M-Kdo and Kdo-M, respectively (Figure S2 and S3).

Bicelles are a very important kind of morphology mimicking the natural lipid bilayer membranes.20

By using bicelles formed by phospholipid as a model, structure and function of various membrane




7

proteins have been investigated.21-23 Although the bicelles based on phospholipid have made a

great success,24 bicelles made by glycolipid have not been reported as far as we know. Thus, it is

natural for us to explore the formation mechanism of the above bicelles. We traced the bicelles

formation process by TEM. Excitingly, it was found that although the bicelles from S-Kdo and

Kdo-S looked very similar to each other, their formation kinetics were quite different. As can be

seen from Figure 2e-g, in the case of S-Kdo, it takes a very long time for us to observe high

conversion of the well-formed bicelles. After incubation for 3 h, micelles can be found from the

aqueous solution (Figure 2e), the hydrodynamic diameters of the micelles were about 7 nm

according to number distribution of dynamic light scattering (Figure S4). Then, the micelles were

found aggregated into bicelles (Figure 2f). Although intact bicelles have been found within a day

(Figure 1a and 1b), this process was found to be completed as long as several months. As shown

in Figure 2f, the bicelles at a large magnification showed the aggregated micelles just forming the

bicelle shapes. Even after 6 months, micelles can still coexist with well-defined bicelles (Figure

2g and S5). In contrast, for the glycolipid Kdo-S, the micelles and original plate-shaped bicelles

simultaneously emerged within 5 min (Figure 2h, 2i). More excitingly, after only 15 min, almost

all of the micelles were already found to convert into bicelles (Figure 2j). In this process, round

bicelles with voids (Figure S6a) were observed from AFM in 10 minutes and then gradually

became an integral part within 18 h (Figure S6b, 6c). These results revealed that the glycolipids,

i.e. S-Kdo and Kdo-S appeared to share the same carbohydrate backbone, similar assembled

morphology and formation pathway, but differed strikingly in formation dynamics, ranging from

dozens of minutes to several months. The formation process of bicelles by S-Kdo and Kdo-S was

illustrated in Figure 2B. This exciting result inspires us to reveal the molecular packing structure

by molecular dynamics simulation, in order to explain such dramatic kinetic difference.




8

Figure 2. (A) Self-assembly of S-Kdo and Kdo-S. (a) TEM and (b) AFM images of bilayer

structures formed by S-Kdo. (c) TEM and (d) AFM images of bilayer structure formed by Kdo-

S. (e~g) TEM images of bilayer structure formed by S-Kdo at different time point. (h~j) TEM

images of bilayer structures formed by Kdo-S at different time point. (B) Diagram of the formation

process of the bicelles by S-Kdo and Kdo-S.

Molecular dynamics simulations. Briefly, the molecular dynamics (MD) simulation was

performed by CHARMM36 force field25,26 for the molecules and TIP3P model27 for water. Lipid

bilayer systems of 1240 S-Kdo/Kdo-S and roughly 94,000 water molecules were constructed in a

cubic box (22 nm × 22 nm × 8 nm) with 3D periodic conditions (the detailed information is in

supporting information). The preliminary simulations shown that the common construction

patterns of dialkyl lipid bilayers such as LC phase16 clearly deviate from the experimental results

(Figure S7). Therefore, by mimicking the equilibrated interleaved bilayer structure (Figure S1b),

we constructed the initial structures (Figure S8 and S9) of all lipid bilayers. The snapshots of the




9

self-assembled bilayers after 100 ns simulation (the potential diagram see Figure S10) were

presented in Figure 3 serving as an anatomy of the assembled structures. For S-Kdo, an overview

(top view, Figure 3a) of bilayer structure showed very regular arrangement of the monosaccharide

Kdo, and the snapshot of the bilayer from side view (Figure 3b) verified antiparallel arrangement

of the glycolipids. From the enlarged view (Figure 3c, 3d) and that from different azimuth angles

(Figure 3e), obvious hydrogen bonding occur in intermolecular amide bonds as well as intra- and

intermolecular OH/NH3+ groups of Kdo moiety. It is noteworthy that both the amide bonds and

sugar-rings were orderly arranged despite of the electrostatic repulsion from NH3+ groups.




10

Figure 3. Molecular dynamics simulationsof bicelles formed by S-Kdo and Kdo-S. Hydrogen

bonds between amide groups are showed as red dotted lines, other hydrogen bonds are showed as

blue dotted lines, Cl- in S-Kdo and all alkyl chains are showed as transparent. (a) Top view of

bicelles formed by S-Kdo at a simulation time of 100 ns. The different colors were used to

distinguish the different structural units, each one corresponds to the color in structure of S-Kdo




11

in Figure 3d. (b) Side view of bicelles formed by S-Kdo. (c) Snapshots of partial enlarged top view

of bicelles formed by S-Kdo. (d) Structure of S-Kdo with different colors for distinguishing

different structural units. (e) Partial enlarged view of bicelles formed by S-Kdo from different

azimuth angles. (f) Top view of bicelles formed by Kdo-S at a simulation time of 100 ns. The

different colors were used to distinguish the different structural units, each one corresponding to

the color in structure of Kdo-S in Figure 3i. (g) Side view of bicelles formed by Kdo-S. (h)

Snapshots of partial enlarged top view of bicelles formed by Kdo-S. (i) Structure of Kdo-S with

different colors for distinguishing different structural units. (j) Partial enlarged view of bicelles

formed by Kdo-S from different azimuth angles. (k) The statistical results of hydrogen number

involved in bicelles formed by S-Kdo and Kdo-S. (l) The statistical results of SASA involved in

bicelles formed by S-Kdo and Kdo-S.

Under the same simulation protocol, the structure investigation and statistics of hydrogen bonds

were also performed for glycolipid Kdo-S with ideally protonated COOH groups. According to

the snap-shots from top view (Figure 3f) and side view (Figure 3g), neat arrangement of amide

bonds was also obviously observed. However, in contrast to the orderly arrangement of the

carbohydrate backbone in the bilayer structure of S-Kdo, the Kdo moiety in bilayer of Kdo-S was

alternatively or randomly positioned on both sides of the ordered amide groups (Figure 3j). The

random position is quite obvious along the aligned molecules. This can be attributed to the

difference in covalent connections between the amide bond and the Kdo moiety, since with -

CH2CHOH- (square part in Figure 3i), the corresponding conformational rigidity of the Kdo-S

molecule is greatly decreased compared with S-Kdo.

Besides the difference from packing of Kdo moieties in the two bicelles, the hydrogen bond

numbers involved in the S-Kdo and Kdo-S bicelles also provided distinguishable information. As




12

shown in Figure 3k and Table S1, the total number of the hydrogen bonds originated from bicelles

of Kdo-S was much larger than that from bicelles formed by S-Kdo. When the hydrogen bonds

(H-bonds) are broken down to different origins (Figure 3k), one may find the difference comes

from the H-bonds between the amide bond, those between the OH from Kdo, and those participated

by NH3+ or COOH. This result can be mainly explained by the disordered arrangements of Kdo

backbone of Kdo-S on the surface of the bilayer, which facilitated the formation of hydrogen

bonds. Meanwhile, the charged NH3+ brought repulsion between S-Kdo, which also decreased the

number of H-bonds. The detailed information of hydrogen bonding was shown in Table S1 in the

SI.

Moreover, the ultimate solvent accessible surface area (SASA) for the total and the individual

molecules that calculated from the simulation results are shown in Figure 3l and Table S2. Briefly,

the SASAs from hydrophobic parts of S-Kdo and Kdo-S are quite similar, indicating their Kdo

backbones are in similar hydrophobic environments. While the difference comes from the SASA

involved by NH3+ and COOH, the former exposed to water more than the latter, even resulting

similar trend of SASA of the hydrophilic part of Kdo. From all above results, one may find that

although S-Kdo and Kdo-S share the same carbohydrate backbone, due to the unique structure of

Kdo, the packing of the glycolipid in the bicelle can be so much different. The regular packing of

Kdo in S-Kdo indicates its higher stability than that of Kdo-S, which also explains the slow

formation kinetics that we observed in experiments. The high SASA from the chemical structure

especially the charged NH3+ may also contribute to the stability. While the large number of H-

bond contributes to the enthalpy of the self-assembly process, indicating the self-assembly of Kdo-

S is relatively enthalpy favored, resulting in fast formation kinetics. Thus, we may predict the




13

properties of the two bicelles can be quite different, as we found from the formation kinetics, which

inspire us to explore the stability of the bicelles in a relatively long time scale.

Figure 4. TEM and AFM images of bicelles formed by Kdo-S at different time interval. (a) TEM

image of Kdo-S assemblies after incubation for 2 weeks. (b, c) TEM and AFM images after

incubation for 3 days. (d) AFM image of ribbons (e, f) TEM images of sample without shaking for

2 weeks.

Morphology transition of Kdo-S from bicelles to ribbons. According to the aforementioned

MD result, the stability of the bicelles was investigated to demonstrate their difference in

aggregation state. As shown in Figure 2g and S4, although the formation of S-Kdo was quite slow,

the formed bicelles were found capable of stably survive in aqueous solution for more than half a




14

year. This result is consistent to the quite regular packing of Kdo backbone and relatively high

SASA found in simulation. On the contrary, the disordered packing of carbohydrate backbone in

bicelles of Kdo-S indicated morphology transition possibility. Interestingly, when the morphology

of the sample was verified again 2 weeks later, instead of bicelles, well-defined and long ribbons

featured by aligned structure were observed with TEM (Figure 4a) and cryo-EM (Figure S11a).

To reveal the transition of the morphology, the self-assembly behavior was tracked by TEM and

AFM again. Further experimental results showed that although the round bilayer bicelles

seemingly have no obvious change according to TEM image (Figure 4b), these planar bicelles

gradually dissociated, bicelles with jagged surface (Figure 4c, S11b) was observed from AFM

images only 3 d later. And small amount of ribbons were observed from TEM and AFM images

(Figure S11c, 11d). The structure showed in Figure 4c is quite similar to the ‘fused micelle’ state

shown in Figure 2i, indicating that there is a fast equilibrium between the bicelles and the ‘fused

micelles’. Then the ‘fused micelles’ state was supposed to have a strong tendency to transform

into ribbons. The height of these ribbons was measured as ~4 nm and 8 nm (Figure 4d), implying

the formation of multilayer structures. The small-angle X-ray scattering (SAXS) results (Figure

S12) of Kdo-S also demonstrated the space of the bilayer structure is 3.83 nm according to the

Bragg equation (2d sin θ = nλ or d = 2nπ/q, λ = 1.5418 Å), the value is in consistent with the height

of the ribbon according to the AFM results. Interestingly, when the solution was standing without

shaking for 2 weeks, ‘cinnamon roll’-shaped structures in clockwise direction were captured with

TEM (Figure 4e, 4f, S11e-h). The intermittent spiral structure at a larger magnification indicated

that they are most likely evolved from the ‘fused micelles’ (Figure 4f).

Considering the above simulation results, it is quite rational to attribute this morphology transition

from bicelles to fused micelles and then ribbons, to the packing of Kdo-S and the deprotonation




15

of COOH. Thus, molecular dynamics simulation of deprotonated Kdo-S with sodium as counter-

ion (Kdo(s)-S) were performed by using the same model. Initial structures of lipid bilayers in

simulation of Kdo(s)-S was shown in Figure S13. From the snap shots during simulation, it was

found that with the deprotonation of COOH to COO-, the previous regular packing of amide bond

was broken (Figure 5b). The irregular packing of amide can be even more obvious when bicelles

formed by Kdo(s)-S were relaxed for 100 ns (Figure 5c). Although this transition resulted in the

decrease of H-bonds between amides (Figure 5e, Table S3), the number of H-bonds between the

OH of Kdo was increased due to accessible hydroxyl groups of the flexible Kdo backbone.

Moreover, the appearance of COO- also increased the possibility of H-bond formation between

OH and COO-. Meanwhile, the SASA of Kdo(s)-S was increased compared to that of Kdo-S

(Figure 5f, Table S4), which suggested strong demands of exposure to solvents of Kdo(s)-S by

decreasing the volume of the aggregate. Although the simulation was performed in a time and

space scale that failed to cover any disassembly events of the Kdo(s)-S bilayer, the simulation

results connected the deprotonation of COOH to the packing of amide bonds and the H-bond

formation possibilities, which finally led to the dissociation of the bicelles.




16

Figure 5. Molecular dynamics simulations of bicelles formed by Kdo(s)-S. Hydrogen bonds

between amide groups are showed as red dotted lines, other hydrogen bonds are showed as blue

dotted lines, the alkyl chain and Na+ are showed as transparent. (a) Partial enlarged view of initial

bilayer bicelles generated by Kdo(s)-S. The different colors were used to distinguish the different

structural units, each one corresponds to the color in structure of Kdo(s)-S in Figure 5a. (b) Partial

enlarged view of bicelles generated by Kdo(s)-S at a simulation time of 50 ns. (c, d) Partial

enlarged view and top view of bicelles generated by Kdo(s)-S after relaxation for 100 ns. (e) The

statistical results of hydrogen number involved in bicelles of Kdo-S and Kdo(s)-S. (f) The

statistical results of SASA involved in bicelles of Kdo-S and Kdo(s)-S.

With the above results, we speculate that the evolution of Kdo-S from bicelles to fused micelles

and then ribbons was triggered by the deprotonation of COOH in aqueous solution. Considering

the pKa of this compound, this process could be relatively slow, resulting in the intermediate

morphologies we observed from solution. Thus fast formation of COO- may directly bring us the

final ribbon morphology quickly. LiOH (0.1 M, 20 L, 2 equiv) was added to Kdo-S solution (1




17

mg/mL, 0.5 mL) to adjust the pH value to about 9. After ultrasonic treatment for about 10 min and

incubation for additional 40 min, it was found that the bicelles (Figure S15a) appeared to be similar

with the bicelles formed by Kdo-S. Only one day later, fine-long ribbons (Figure S15b) emerged.

Similar results were observed when NaOH was added (Figure S15d~S15f).

Direct Formation of Ribbons of Kdo-S. Compared to bicelles, it turns out that ribbons might be

the thermodynamically stable product of Kdo-S. To confirm this hypothesis, other self-assembly

conditions were explored. methanol-water (v/v, 1/1) mixture as solvent was used to study the self-

assemble behavior of Kdo-S at the same concentration. After 30 min, negative stained TEM

images showed that helical ribbons (Figure 6a, S16a) with a pitch of about 232 nm (Figure 6a)

were obtained. The height of the nanostructures was measured as about 17 nm (Figure 6b),

indicating stacking of the ribbons. Moreover, the helical ribbons with multilayer structure were

also observed directly from Cryo-EM (Figure S16c). Furthermore, the structural information of

the ribbons was investigated by Circular Dichroism (CD) spectroscopy. A strong negative band at

~192 nm and a weak positive band at ~ 210 nm with a crossover at around 203 nm (Figure S16d)

indicated the formation of a PPII helix-style conformation, which is far more flexible than -helix

and -sheet types.28




18

Figure 6. Self-assembly of Kdo-S in MeOH/H2O and Kdo(a)-S in H2O. (a, b) TEM and AFM

images Kdo-S in MeOH/H2O. (c, d) TEM and AFM images of assemblies formed by Kdo(a)-S in

H2O.

In addition, to prove that ribbons can be formed directly as the thermodynamic self-assembly

product rather than bicelles upon weaking the electrostatic repulsion from carboxylic acid anion,

ammonium carboxylates form of Kdo-S, i.e. Kdo(a)-S was also prepared by adding ammonium

hydroxide to a solution of Kdo-S in MeOH.18 The mixture was kept stirring for 40 min, and the

solution was concentrated under reduced pressure, then the self-assembly behavior was

investigated. Only under ultrasonic treatment for 10 min in H2O, long ribbons (Figure 6c, 6d, S16e)

were quickly provided after incubation for 3 h. 7 weeks later, flexible ribbons were also observed

from TEM (Figure S16f) and verified by CD spectrum (Figure S16g). AFM measurement also

showed that these ribbons tend to pack with each other to provide multilayer structures (Figure

S16h).




19

Summary of the stable and dynamic bicelles formed by Kdo-based glycolipids. In short, the

two bicelles formed by glycolipids gave very different properties, although they shared similar

driving forces to assemble, including hydrophobic interactions of alkyl chains, and the hydrogen

bonds between neighboring amides and those bonds between Kdo moieties. The bicelles formed

by S-Kdo are fairly stable. The energy landscape of formation process was mapped in Figure 7A.

On cooling of a hot solution of glycolipid S-Kdo, micelles were firstly formed. As time went on,

bicelles were formed via fusion of micelles. In the case of Kdo-S, the relative irregular packing of

Kdo backbone brought great opportunity for hydroxyl groups from Kdo to form hydrogen bonds,

which may significantly allow the fast formation of bicelles as kinetically favored intermediate

(Figure 7B). Meanwhile, the irregular packing of Kdo backbone indicated the flexible nature of

the bicelles, in which the morphology transformation was triggered by the slow deprotonation of

COOH, which could be suppressed during the fast bicelles formation. With time went by, more

and more deprotonation took place, and the electrostatic repulsion of carboxylic acid anion induced

an increase in curvature of bicelles, and the spherical micelles with increased SASA were provided.

Furthermore, hydrophobic interactions of alkyl tails, and in situ spatial redistribution of other kinds

of weak interactions among micelles led to the formation of spiral ribbons. The similar process

was also undergone by dissociated micelles, and compactly arrangement of the glycolipids with

bilayer structures gave rise to the final long ribbons and even macrostructures.




20

Figure 7. Energy landscape of self-assembly process of (A) S-Kdo and (B) Kdo-S.

Antibacterial trials. Considering that glycolipids have the ability to disturb integrity and

permeability of membrane by inducing pore and ion channels,29 and Kdo derivates also have great

potential in antibacterial activity,12 the antibacterial trials of the assemblies were also explored.

Two Gram-positive strains (S. aureus, E. faecium) and two Gram-negative strains (E. coli and P.

aeruginosa) were selected to evaluate the anti-bacterial activities of above self-assembled

glycolipids. Commercially available Vancomycin and Gentamicin were selected as positive

control. Minimum inhibitory concentration (MIC) values were measured using the broth dilution

method 30-32(details shown in supporting information). As shown in Table 1, assembled S-Kdo

exhibited better antibacterial activities than Gentamicin, when it was treated with the two Gram-

positive bacteria with MIC value (Table S5) as 12 to S. Aureus and 6 to E. facium. However, no

activity was observed in antibacterial trials of S-Kdo against the two Gram-negative bacteria. The

most promising antibacterial activity came from the assembled M-Kdo, which displayed strong

inhibition to all of the four bacterial strains. The MIC values (Table S5) for E. coli, P. aeruginosa

and S. aureus are all around 24 g/mL, and 12 g/mL for E. facium. The TEM images (Figure

S17) showed that the cell membranes of the bacteria were dramatically damaged, implying the




21

positive glycolipids may intervene bacteria growth primarily by disrupting bacterial cell

membrane. It was known that most bacterial membranes are negatively charged, then cationic

molecules are attracted and the lipophilic chains could insert into the lipid bilayer to disrupt the

membrane integrity and stability, which might explain the inhibition activities of the positive

charged glycolipids, i.e. S-Kdo and M-Kdo to the relevant bacteria, while Kdo-S and Kdo-M

displayed no activity against all of the bacterial strains even when the concentration was increased

to 384 g/mL. According to the selectivity of S-Kdo, the protection from outer membranes of

Gram-negative bacteria, might prevent affection of S-Kdo with long alkyl chain as large

aggregates.

Conclusion

In this paper, by using glycolipids based on rare saccharide Kdo, we demonstrated the complexity

and diversity of self-assembled glycolipid structures via combination of experiments and

simulation. It was found that, although Kdo-S and S-Kdo share the same saccharide backbone,

their self-assembly behavior could be very different. The stabilities of the bicelles made by those

glycolipids, were greatly controlled by the packing of the saccharide moieties, i.e. regular packing

of S-Kdo gave stable bicelles as thermodynamic stable product, while flexible packing of Kdo-S

brought H-bond formation possibility, which was kinetically favored. Then the kinetic bicelles of

Kdo-S transformed to ribbons as thermodynamically stable structure. The above findings

significantly demonstrated the precise control on the assembled morphology of the carbohydrate

moiety of glycolipids, which could not be simply explained as a ‘hydrophilic head’. A subtle

structural change of the carbohydrate structure (including the charge, the attachment and the

attached positions) and external conditions (including pH, solvent) could greatly affect the packing

of the carbohydrate backbone, then affect the related H-bond formation and other non-covalent




22

interactions, finally to the property of the microscopic structure. This discovery and the obtained

regular glyco-structures of bicelles with antibacterial activities will not only deepen our

understanding on the diversity of glycolipids, but also shed light on the further exploration of

antibacterial glyco-materials.

Supporting Information

Information is available and includes the protocol for the synthesis of the glycolipids,

computational methods and results, characterization of the compounds and assemblies, data and

figure about antibacterial assay.

Conflict of Interest

The authors declare no competing financial interest.

Acknowledgments

G.C. acknowledge the financial support from the National Natural Science Foundation of China

(No. 51721002, 21861132012, 91956127, 21975047). R.Z. thanks NSFC/China (No. 21674114,

91956127) for financial support. This work is supported by Shanghai Municipal Science and

Technology Major Project (No.2018SHZDZX01) and ZJ Lab.

References

(1) Lehn, J.-M. Toward complex matter: Supramolecular chemistry and self-organization.

Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 4763-4768.




23

(2) Aida, T.; Meijer, E. W.; Stupp, S. I. Functional supramolecular polymers. Science 2012,

335, 813-817.

(3) Korevaar, P. A.; George, S. J.; Markvoort, A. J.; Smulders, M. M.; Hilbers, P. A.;

Schenning, A. P.; De Greef, T. F.; Meijer, E. W. Pathway complexity in supramolecular

polymerization. Nature 2012, 481, 492-496.

(4) Harayama, T.; Riezman, H. Understanding the diversity of membrane lipid composition.

Nat. Rev. Mol. Cell Biol. 2018, 19, 281-296.

(5) Flores, J.; White, B. M.; Brea, R. J.; Baskin, J. M.; Devaraj, N. K. Lipids: chemical tools

for their synthesis, modification, and analysis. Chem. Soc. Rev. 2020, 49, 4602-4614.

(6) Hendricks, M. P.; Sato, K.; Palmer, L. C.; Stupp, S. I. Supramolecular assembly of peptide

amphiphiles. Acc. Chem. Res. 2017, 50, 2440-2448.

(7) Liu, R.; Hudalla, G. A. Using self-assembling peptides to integrate biomolecules into

functional supramolecular biomaterials. Molecules 2019, 24, 1450.

(8) Xiao, C.-S.; Ding, J.-X.; He, C.-L.; Chen, X.-S. Glycopolypeptides: synthesis, self-

assembly and biomedical applications. Acta Polym. Sin. 2018, 45-55.

(9) Sherman, S. E.; Xiao, Q.; Percec, V. Mimicking complex biological membranes and their

programmable glycan ligands with dendrimersomes and glycodendrimersomes. Chem. Rev. 2017,

117, 6538-6631.

(10) Delbianco, M.; Bharate, P.; Varela-Aramburu, S.; Seeberger, P. H. Carbohydrates in

supramolecular chemistry. Chem. Rev. 2016, 116, 1693-1752.




24

(11) Jayaraman, N.; Maiti, K.; Naresh, K. Multivalent glycoliposomes and micelles to study

carbohydrate-protein and carbohydrate-carbohydrate interactions. Chem. Soc. Rev. 2013, 42,

4640-4056.

(12) Cipolla, L.; Gabrielli, L.; Bini, D.; Russo, L.; Shaikh, N. Kdo: a critical monosaccharide

for bacteria viability. Nat. Prod. Rep. 2010, 27, 1618-1629.

(13) Lundstedt, E.; Kahne, D.; Ruiz, N. Assembly and maintenance of lipids at the bacterial

outer membrane. Chem. Rev. 2021, 121, 5098–5123.

(14) Wen, L.; Zheng, Y.; Li, T.; Wang, P. G. Enzymatic synthesis of 3-deoxy-D-manno-

octulosonic acid (KDO) and its application for LPS assembly. Bioorg. Med. Chem. Lett. 2016, 26,

2825-2828.

(15) Corti, M.; Cantù, L.; Brocca, P.; Del Favero, E. Self-assembly in glycolipids. Curr. Opin.

Colloid Interface Sci. 2007, 12, 148-154.

(16) Hashim, R.; Zahid, N. I.; Velayutham, T. S.; Aripin, N. F. K.; Ogawa, S.; Sugimura, A.

Dry thermotropic glycolipid self-assembly: a review. J. Oleo. Sci. 2018, 67, 651-668.

(17) Brea, R. J.; Bhattacharya, A.; Bhattacharya, R.; Song, J. J.; Sinha, S. K.; Devaraj, N. K.

Highly stable artificial cells from galactopyranose-derived single-chain amphiphiles. J. Am. Chem.

Soc. 2018, 140, 17356-17360.

(18) Feng, Y.; Dong, J.; Xu, F.; Liu, A.; Wang, L.; Zhang, Q.; Chai, Y. Efficient large scale

syntheses of 3-deoxy-D-manno-2-octulosonic acid (Kdo) and its derivatives. Org. Lett. 2015, 17,

2388-2391.




25

(19) Humphrey, W.; Dalke, A.; Schulten, K. Vmd: Visual Molecular Dynamics. J. Mol.

Graphics Modell. 1996, 14, 33-38.

(20) Li, Q.; Han, X. Self-assembled "breathing" grana-like cisternae stacks. Adv. Mater. 2018,

30, e1707482.

(21) Sanders, C. R.; Landis, G. C. Reconstitution of membrane proteins into lipid-rich bilayered

mixed micelles for nmr studies. Biochemistry 1995, 34, 4030-4040.

(22) Tan, C.; Fung, B. M.; Cho, G. Phospholipid bicelles that align with their normals parallel

to the magnetic field. J. Am. Chem. Soc. 2002, 124, 11827-11832.

(23) Hutchison, J. M.; Shih, K. C.; Scheidt, H. A.; Fantin, S. M.; Parson, K. F.; Pantelopulos,

G. A.; Harrington, H. R.; Mittendorf, K. F.; Qian, S.; Stein, R. A.; Collier, S. E.; Chambers, M.

G.; Katsaras, J.; Voehler, M. W.; Ruotolo, B. T.; Huster, D.; McFeeters, R. L.; Straub, J. E.; Nieh,

M. P.; Sanders, C. R. Bicelles rich in both sphingolipids and cholesterol and their use in studies of

membrane proteins. J. Am. Chem. Soc. 2020, 142, 12715-12729.

(24) Barbosa-Barros, L.; Rodriguez, G.; Barba, C.; Cocera, M.; Rubio, L.; Estelrich, J.; Lopez-

Iglesias, C.; de la Maza, A.; Lopez, O. Bicelles: lipid nanostructured platforms with potential

dermal applications. Small 2012, 8, 807-818.

(25) Huang, J.; Rauscher, S.; Nawrocki, G.; Ran, T.; Feig, M.; de Groot, B. L.; Grubmuller, H.;

MacKerell, A. D., Jr. CHARMM36m: an improved force field for folded and intrinsically

disordered proteins. Nat. Methods 2017, 14, 71-73.




26

(26) Guvench, O.; Mallajosyula, S. S.; Raman, E. P.; Hatcher, E.; Vanommeslaeghe, K.; Foster,

T. J.; Jamison, F. W., 2nd; Mackerell, A. D., Jr. CHARMM additive all-atom force field for

carbohydrate derivatives and its utility in polysaccharide and carbohydrate-protein modeling. J.

Chem. Theory Comput. 2011, 7, 3162-3180.

(27) Jorgensen, W. L.; Tirado-Rives, J. Potential energy functions for atomic-level simulations

of water and organic and biomolecular systems. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 6665-

6670.

(28) Siligardi, G.; Drake, A. F. The importance of extended conformations and, in particular,

the pll conformation for the molecular recognition of peptides. Biopolymers (Peptide Science)

1995, 37, 281-292.

(29) Ines, M.; Dhouha, G. Glycolipid biosurfactants: potential related biomedical and

biotechnological applications. Carbohydr. Res. 2015, 416, 59-69.

(30) Engler, A. C.; Tan, J. P.; Ong, Z. Y.; Coady, D. J.; Ng, V. W.; Yang, Y. Y.; Hedrick, J. L.

Antimicrobial polycarbonates: investigating the impact of balancing charge and hydrophobicity

using a same-centered polymer approach. Biomacromolecules 2013, 14, 4331-4339.

(31) Hoque, J.; Konai, M. M.; Sequeira, S. S.; Samaddar, S.; Haldar, J. Antibacterial and

antibiofilm activity of cationic small molecules with spatial positioning of hydrophobicity: an in

vitro and in vivo evaluation. J. Med. Chem. 2016, 59, 10750-10762.

(32) Bai, P. Y.; Qin, S. S.; Chu, W. C.; Yang, Y.; Cui, D. Y.; Hua, Y. G.; Yang, Q. Q.; Zhang,

E. Synthesis and antibacterial bioactivities of cationic deacetyl linezolid amphiphiles. Eur. J. Med.

Chem. 2018, 155, 925-945.




27

Table of Contents Graphic




Polymorphism of Kdo-based Glycolipids: the

Documents