Atomically resolved imaging of the conformations and adsorption geometries of individual β-cyclodextrins with non-contact AFM

Article https://doi.org/10.1038/s41467-024-53555-0 Atomically resolved imaging of the conformations and adsorption geometries of individual β-cyclodextrins with non-contact AFM Márkó Grabarics 1,2,8 ,BenjamínMallada 3,4,8 , Shayan Edalatmanesh 3,4,8 , Alejandro Jiménez-Martín 3,4,5 , Martin Pykal 4 ,MartinOndráček 3 , Petra Kührová 4 ,WestonB.Struwe 2,6 ,PavelBanáš 4 , Stephan Rauschenbach 1,2 ,PavelJelínek 3,4 & Bruno de la Torre 4,7 Glycans, consisting of covalently linked sugar units, are a major class of biopolymers essential to all known living organisms. To better understand their biological functions and further applications in fields from biomedicine to materials science, detailed knowledge of their structure is essential. However, due to the extraordinary complexity and conformational flexibility of glycans, state-of-the-art glycan analysis methods often fail to provide structural information with atomic precision. Here, we combine electrospray deposition in ultra-high vacuum with non-contact atomic force microscopy and theoretical calculations to unravel the structure of β-cyclodextrin, a cyclic glucose oligomer, with atomic-scale detail. Our results, established on the singlemolecule level, reveal the different adsorption geometries and conformations of β-cyclodextrin. The position of individual hydroxy groups and the location of the stabilizing intramolecular H-bonds are deduced from atomically resolved images, enabling the unambiguous assignment of the molecular structure and demonstrating the potential of the method for glycan analysis. Glycans are present in all cells, playing key roles ina range of biological processes and contributing greatly to the molecular and functional diversity of living organisms1,2. Structurally, they consist of monosaccharide building blocks, such as glucose, mannose, and galactose, which are connected to each other via their OH groups, forming glycosidic linkages. Much ofthe structuraldiversity of glycans stems from variations in the orientation of these OH groups. Such variations lead to the isomerism of monosaccharide units, and the variability in the orientation of the glycosidic bonds between them. Combined with differences in linkage position, these small variations lead to the intrinsic and significant structural complexity of glycans. As minute structural differences can fine-tune biological function, analytical approaches sensitive to such small structural details and capable of revealing the underlying structure of glycans are highly sought after3. This immense complexity of glycans impedes complete structural assignments. Their elaborate regioand stereochemistry pose Received: 22 June 2024 Accepted: 15 October 2024 Check for updates 1 Department of Chemistry, University of Oxford, OX1 3QU Oxford, UK. 2 Kavli Institute for Nanoscience Discovery, University of Oxford, OX1 3QU Oxford, UK. 3 Institute of Physics, Czech Academy of Sciences, 16200 Prague, Czech Republic. 4 Czech Advanced Technology and Research Institute, Palacký University Olomouc, 78371 Olomouc, Czech Republic. 5 Faculty of Nuclear Sciences and Physical Engineering, Czech Technical University in Prague, 115 19 Prague, Czech Republic. 6 Department of Biochemistry, University of Oxford, OX1 3QU Oxford, UK. 7 Nanomaterials and Nanotechnology Research Center, CSIC-UNIOVI-PA, 33940 El Entrego, Spain. 8 These authors contributed equally: Márkó Grabarics, Benjamín Mallada, Shayan Edalatmanesh. e-mail: [email protected];[email protected];[email protected];[email protected] Nature Communications | (2024) 15:9482 1 1234567890():,; 1234567890():,; significant challenges to state-of-the-art mass spectrometric methods4,5. Isomers with identical masses aredifficult to distinguish as they usually display very similar mass spectrometric fragmentation patterns, and the de novo assignment of linkage positions is hindered by the scarcity of diagnostic cross-ring fragments. In addition, atomicresolution imaging of glycans by ensemble-averaged microscopic and diffraction methods is often not possible due to their conformational flexibility. Consequently, our knowledge of the structure of glycans is often incomplete, which represents a major bottleneck in better understanding their functions and facilitating their applications, from vaccine and drug development to materials science6–10. An alternative approach to fully characterize the structure of glycans relies on the combination of electrospray ionization-based deposition methods and scanning tunneling microscopy (STM), which enables the gentle deposition of the thermolabile glycans onto clean single-crystal surfaces, and subsequent real-space imaging at the single-molecule level11–16. To obtain images with sufficient quality that allow for visualizing the connectivity of monosaccharide units within glycans, measurements must be performed under highly controlled conditions, requiring ultrahigh vacuum (UHV), cryogenic temperatures, and atomically flat surfaces. However, STM images obtained with bare metal tips reveal the overall shape of individual monosaccharide building blocks, which also depends on their adsorption conformation11,17, while internal structural details at the atomic level remain unresolved. Here we employ dynamic non-contact atomic force microscopy (nc-AFM) with CO-functionalized tips—enabled by electrospray deposition (ESD) of the molecules—to reveal the internal structure of monosaccharide units and resolve atomic-scale features that reflect individual functional groups and chemical bonds within the glycans. We investigate β-cyclodextrin (β-CD) as a model compound, a cyclic oligosaccharide consisting of seven D-glucopyranose units connected via α−1,4-glycosidic bonds (Fig. 1a). The shape of this macrocycle resembles a truncated cone where the glucose units enclose a central hydrophobic cavity18. Fourteen secondary OH groups form a wider rim at the cone’s base (secondary face), while seven primary OH groups form a narrower rim at the top (primary face). Its three-dimensional shape renders β-CD a challenging molecule for nc-AFM and expands the scope of this technique, which has so far been applied mainly to conjugated planar molecular structures19–23. By combining experiments with nc-AFM simulations that also take electrostatic tipmolecule interactions into account, we show that the obtained images contain sufficient information to distinguish and unambiguously assign the different OH groups within single β-CD molecules, and to confidently identify the atomic structure and conformation of this complex biomolecule. Results Single-molecule imaging of β-cyclodextrin conformers Incompatible with conventionally used thermal sublimation methods, β-CD was deposited onto an atomically clean Au(111) single-crystal substrateatroomtemperatureusingESD.Low-temperatureSTM topographs of the samples reveal irregular two-dimensional islands with average topographic heights between 250 and 350 pm (Fig. 1eand Supplementary Fig. 1). Within the islands, single β-CD molecules emerge as bright, doughnut-shaped objects. They consist of seven clearly resolved rounded lobes that encircle a central depression, which can be tentatively assigned to the individual glucose building blocks and the molecule’s cavity, respectively. The intact β-CDs are surrounded by a lower bed that consists of distorted macrocycles (and potentially some smaller fragments), discussed later in more detail (additional images of thestructuresformingthebedareshowninSupplementaryFig.2). Although STM reveals overall shapes and monomer units, a much more precise assignment can be achieved using high-resolution ncAFM. Primarily, three dominant types of objects are identified on the surface, all illustrated in Fig. 1b–d. The first type (Fig. 1b) exhibits sevenfold symmetry with an outside diameter of 12.6 ± 0.5Å. It features a central heptagon with bright, distinct borders enclosing a darker interior. A narrow ray points outwards from each corner of the heptagon, with darker regions between them, resulting in a starshaped structure. This structure is surrounded by a faint, diffuse halo that gradually dissipates into the substrate’s background. The second type (Fig. 1c) appears larger across (15.6 ± 0.1 Å). Seven round features, separated by narrow ribs, are arranged in a circular pattern around a central cavity. The differences in their diameter (12.6 vs. 15.6Å), along with their similar absorption height relative to the substrate (roughly 7.5–8Å, see Supplementary Fig. 3), suggest that they correspond to two β-CD geometries on the surface. The truncated cone shape of β-CD with its two distinct faces allowsfortwoorientationsonthesurfacewhereallseven glucoseunits are equally visible by STM and nc-AFM. In one case, β-CD is adsorbed on the surface via its secondary face, the primary face with its narrow rim pointing upwards (primary face up or p↑orientation). In the other case, the secondary face with the wide rim faces the probe, while the primary face points towards the surface (secondary face up or s ↑). Accordingly, based on the different diameters observed, the starshaped objects are assigned to p ↑, while the larger ones to the s↑ orientation. The assignment is further confirmed by simulated nc-AFM Distorted structures 5 Å Δf (Hz)-18.2 5.6 Δf (Hz) Secondary face up -4.5-15.0 5 Å Au(111) Probe bPrimary face up 5 Å Δf (Hz) 4.8 -19.3 cde 5 nm z (pm) 627 0 a Primary face (narrow rim) Secondary face (wide rim) Fig. 1 | Structure and real-space images of β-cyclodextrin molecules by lowtemperature STM and nc-AFM. a Chemical formula of β-CD and side-view of a 3Dmodel highlighting the molecule’s truncated cone shape. b–dNc-AFM images of β-CD on Au(111), recorded at constant height using frequency modulation detection. Representative examples are shown for the primary (b) and secondaryface-up geometries (c), as well as for distorted molecules (d). eSTM overview showing an irregular island formed upon electrospray deposition of β-CD onto Au(111) (bias = 1 V, tunneling current = 1 pA). The macrocycles stand out as doughnut-shaped structures from a lower background of fragments; the insert shows a higher resolution STM image of a single macrocycle (scale bar in the insert is 5Å). Article https://doi.org/10.1038/s41467-024-53555-0 Nature Communications | (2024) 15:9482 2 images of both geometries (discussed further in the next section). Additional STM and nc-AFM images of single β-CDmoleculesinp↑and s↑orientation are shown in Supplementary Fig. 9. The third frequently observed object corresponds to distorted βCD molecules (Fig. 1d). In this case, only a few of the round features observed in molecules in s↑orientation can be resolved. Upon comparison with simulated nc-AFM images (see Supplementary Fig. 5), it becomes evident that these correspond to bent β-CD molecules, where only outward-facing features are visible in the constant-height nc-AFM images. Molecular dynamics (MD) simulations revealed that such distortion arises from the rotation of two α−1,4-glycosidic bonds, breaking the symmetry of the β-CD molecules (for more details see Supplementary Note 4 and Supplementary Fig. 4). This rotation causes two adjacent glucose units to twist, with their glucopyranose rings stacked on the surface. The remaining five glucose units remain in the original conformation. This conformational substate of the β-CD molecule was found to be a minor population in simulations in a solvent environment without including the surface. However, it is significantly stabilized when the β-CD molecule is adsorbed on the surface in the s↑orientation (discussed in more detail in Supplementary Note 4.2). The nc-AFM images were recorded in constant-height mode using frequency modulation detection. In this detection mode, more positive Δffrequency shifts (depicted in bright gray) reflect enhanced repulsive forces between the CO molecule attached to the tip and the molecule on the surface24,25. The repulsion arises mainly from the steric Paulirepulsion at closetip-sample distances and alternatively from the electrostatic interaction. Thus, in the first approximation, we may interpret bright areas as regions of higher electron density within the molecules, often indicating the location of atoms and chemical bonds26. Dark areas, in contrast, correspond to regions with attractive interaction driven by long-range van der Waals and, alternatively, electrostatic interactions dominating over short-range steric repulsion. Considering these contrast mechanisms in nc-AFM, some tentative structural assignments may be made based on the experimental micrographs. In p ↑, for instance, the bright borders of the heptagon, which outline a sharp ridge on the potential energy surface above the molecule, indicate the presence of a closed H-bonding network along the narrow rim of β-CD.Inthecaseofs↑,theseven round features and ribs must arise from corresponding atomic groups of the seven glucose building blocks, which very likely involve the secondary OH groups. However, unambiguous identification of the exact atoms giving rise to these features or determining the ring puckers of the glucopyranose units is not possible based solely on the singular constant-height images. The characteristics of nc-AFM images are dominated by forces that change with the tip-sample distance. Thus, for more definite structural assignments, experimental micrographs acquired at different tip-sample distances were compared to simulated images for a set of low-energy structural candidates. Structural assignment—comparison of experiment and theory The precise structural arrangement of different β-CD conformers from high-resolution nc-AFM imaging was elucidated through comparison with simulated nc-AFM images using the probe-particle (PP) model24. In essence, the probe particle (here a CO molecule) positioned at the tip apexdemonstratessensitivity tospatialfluctuations in the potential energy landscape of the molecule, arising from the interplay among Pauli, electrostatic, and van der Waals interactions27–29.Atshortdistances between the tip and sample, the probe particle adjusts in response to the potential energy surface, generating a distinct submolecular contrast. For modeling the sample, a set of low-energy gasphase β-CD conformers were generated by classical MD simulations and subsequently relaxed on the Au(111) surface in both p↑and s↑ geometry. Equilibrated molecules on the Au(111) surface were then used for additional total energy density functional theory (DFT) calculations using FHI-aims30. Figure 2compares theoretical and experimental nc-AFM images acquired for the p↑geometry. The optimized atomic model of the conformer of β-CD using totalenergy DFT calculations, whichgave the most accurate match between simulations and experimental images, is shown from the side of its primary face in Fig. 2a (atomic coordinates provided as Supplementary Data 1, 2). The most important functional groups are highlighted in the figure in color, and the numbering of atoms is explained in Supplementary Fig. 8. The structure in Fig. 2a exhibits C 7 symmetry with all glucopyranose units adopting a 4C 1 ring pucker. Along the narrow rim of the primary face, a closed, homodromic H-bonding network is formed by the seven primary C6-OH groups (OH···O distances at 1.79 Å), all serving both as donor and acceptor. In both experiment and simulation, the resulting ridges in the interaction potential energy surface above the molecule in p↑ orientation become visible at close tip-sample distances as the sharp borders of the central heptagon (Fig. 2b, c). The gradual emergence, size, and symmetry of the seven rays observed in experiments are also accurately reproduced in the simulations, along with the position and contrast of the darker regions. Overall, the agreement between experimental and simulated nc-AFM images is highly accurate for the p↑geometry. Overlay of the atomic model with a corresponding simulated image for p↑also enables direct visual assignment (Fig. 2d). As mentioned above, the structural elements underlying the central heptagon are the primary C6-OH groups thatform a closed network of H-bonds.Thisalsosuggeststhatβ-CD may be present as neutral Fig. 2 | Comparison of experimental and simulated nc-AFM images of the βcyclodextrin primary face. a Atomic model of β-CD, representing the lowestenergy conformer at the DFT PBE level of theory, viewed from its primary face. More prominent functional groups are highlighted in color for ease of discussion. bExperimentaland csimulated nc-AFM imagesof β-CD in the p↑geometry, thetipsample distance decreasing from left to right. In the experiment, z 0 is defined as +250 pm relative to an initial STM setpoint above the molecular ring with the feedback at 1 V and 3 pA; Δzdesignates the difference in tip-sample distance relative to z 0 .dOverlayofatomicmodel andcorrespondingsimulatedimage ofβ-CD in the p↑geometry; the exact atomicgroupsgiving rise to the various features can be readily identified. All scale bars are 5 Å. Article https://doi.org/10.1038/s41467-024-53555-0 Nature Communications | (2024) 15:9482 3 species on the Au(111) surface following electrospray deposition, as coordination to small metal ions—potentially present in the electrosprayed solutions—generally disrupts the regular intramolecular H-bonding pattern in isolated cyclodextrins31.Thesevenrayscanbe assigned mainly to a protruding hydrogen atom of methylene bridges that connect the C6-OH groups with the six-membered pyranose rings of the glucose units (see optimized DFT structure in Supplementary Data 1, 2). The ring oxygens located an atomic layer below, and the glycosidic oxygens, buried even deeper, are not directly visible in the images. These results show the superior resolution and additional structural information provided by nc-AFM as compared to conventional STM, where such assignments would be impossible due to lack of high-resolution, sub-monomer detail. A similar comparison of experiment and simulation for the s↑ geometry is shown in Fig. 3. Here, the structural elements most accessible to the tip are the secondary OH groups, highlighted in color in Fig. 3a. These groups point anti-clockwise when seen from the secondary face, forming H-bonds in a pairwise manner along the wide rim where the C3-OH oxygens (O3) serve as donors and the C2-OH oxygens (O2) as acceptors. Simulated images for s↑are also in excellent agreement with their experimental counterparts (Fig. 3b, c). The central cavity, theround features, and the narrow ribs observed experimentally are all correctly reproduced by theory at different tip-sample distances. By overlaying the atomic model on the corresponding simulated images, the seven round features and the seven ribs can be assigned to the C2-OH and C3OH groups, respectively. Overall, the agreement between experiment and simulation is very convincing for both adsorption geometries, enabling a confident assignment of β-CD’s atomic structure. It needs mentioning that the β-CD molecules in s↑orientation appear slightly distorted in the experimental images. Variations in the brightness of the round features and ribs indicate that, upon adsorption via its primary face, β-CD can undergo slight deformation that leadstodifferences intheheightofthesecondaryOHgroups(typically withinthefractionofanÅ).Whilecomparablevariationsinconformation may prevent ensemble-averaged imaging methods from obtaining high-resolution structural information, nc-AFM allows for the imaging of single molecules in real space, thereby revealing the structural heterogeneity present in the sample. Besides enabling the assignment of β-CD’s atomic structure, simulations also helped provide a rationale behind the observed contrast in the nc-AFM images. We found improved agreement between theoretical and experimental nc-AFM images when the electrostatic interaction between tip and sample was taken into account. The impact of electrostatics on the simulated images is shown in Fig. 4for the s↑orientation. The influence of electrostatics is also recognizable for p↑(see Supplementary Fig. 7), but is much more prominent in the case of the s↑geometry. The greater impact of the electrostatic interaction on the latter reflects the stronger polar character of the secondary face. This results mainly from differences in the orientation and bonding of the C2-OH and C3-OH groups, which also contribute to their different contrast. The hydrogens of the C2-OH groups in the atomic model point away from the plane of the O2 atoms, protruding slightly from the secondary face. The hydrogens in the C3-OH groups, on the other hand, arelocated in the plane of the O2atoms, with which they engage in H-bonds. Thus, the hydrogen atoms in the C2-OH groups are located somewhat closer to the plane of the tip. As theO–H bonds are highly polar, the difference in the orientation of the C2-OH and C3-OH groups leads to an alternating pattern of positive and negative electrostatic potential values above these moieties (see the electrostaticpotentialmapsinFig. 4e, j). The resulting difference in the electrostatic interaction, combined with variations in steric repulsion, is largely responsible for the different contrast of the two sets of secondary OH groups. These findings highlight the sensitivity of nc-AFM to small differences in the orientation and chemical environment of otherwise very similar functional groups within a complex glycan structure. Supplementary Fig. 6 shows simulated nc-AFM images for other stable and symmetric β-CDconformersfromMDsimulations,which display a different orientation of the hydroxy groups. However, these simulated images show significantly different sub-molecular contrast compared to the experiments, which further demonstrates the power of nc-AFM to identify atomic structural differences, in particular molecular conformations. Considering these findings, it is tempting to envisage similar sensitivity in identifying conformers and isomeric species from other glycan classes by nc-AFM, or even other biomolecules of comparable size. As a method capable of providing atomiclevel structural detail for complex, three-dimensional biomolecular species on the single-molecule level, as demonstrated herein, nc-AFM, in combination with ESD, has the potential to overcome many of the key challenges associated with glycan analysis, complement established analytical methods, and significantly improve our understanding of the structure of glycans and other biomolecules. Discussion This work demonstrates the high-resolution imaging of single glycan molecules using nc-AFM with CO-functionalized probes, providing atomic-level structural information on β-CD, a complex biomolecule with a distinctive three-dimensional shape. Compared to conventional STM methods, nc-AFM displays superior spatial resolution that reveals the internal structure of the monosaccharide building blocks. Fig. 3 | Comparison of experimental and simulated nc-AFM images of the βcyclodextrin secondary face. a Atomic model of β-CD, representing the lowestenergy conformer at the DFT PBE level of theory, viewed from its secondary face. More prominent functional groups are highlighted in color. bExperimental and csimulated (bottom) nc-AFM images of β-CD in the s↑geometry, the tip-sample distance decreasing from left to right. In the experiment, z 0 is defined as +190 pm relative to an initial STM setpoint above the molecular ring with the feedback at 0.5 V and 1 pA; Δzdesignates the difference in tip-sample distance relative to z 0 . dOverlay of atomic model and corresponding simulated image of β-CD in the s↑ geometry. All scale bars are 5Å. Article https://doi.org/10.1038/s41467-024-53555-0 Nature Communications | (2024) 15:9482 4 Individual OH groups on both faces of the molecule are resolved with atomic-scale detail, and the location of intramolecular H-bonds could be unambiguously determined from the atomically resolved nc-AFM images. By combining nc-AFM experiments with first-principles DFT simulations, the strong dependence of the imaging contrast on minor variations in the position and chemical environment of the secondary OH groups could be rationalized, allowing the unambiguous distinction and assignment of these key structural elements. The strong agreement between experiment and theory, taking into account the different adsorption geometries of β-CD, allowed for the confident assignment of the molecule’s atomic structure. The structure of intact β-CD on the weakly interacting Au(111) surface displays 7-fold symmetry. Every OH group in the molecule participates in intramolecular H-bonds: the C6-OH groups form a closed homodromic H-bonding network on the primary face, while the secondary OH groups are H-bonded in a pairwise manner. This structure, determined in the absence of solvent in UHV, is significantly different from the high-resolution models obtained via X-ray crystallography, where co-crystallized solvent molecules disrupt the intramolecular H-bonds of β-CD, altering its intrinsic conformation. NcAFM also reveals the different adsorption geometries of β-CD, as well as small structural variations between individual molecules on the surface. Thus, the present work provides novel insight into the structure βCD that has so far not been attainable by established imaging methods. This study represents the first application of high-resolution ncAFM for glycan analysis and serves as a significant step toward extending this technique beyond the analysis of conjugated planar molecules. Our results may also facilitate further nc-AFM studies exploring more complex glycans from both synthetic and natural sources. Determining conformations, distinguishing isomers, and assigning linkage position and configuration in unknown structures are key challenges where nc-AFM has the potential to complement established mass spectrometric and spectroscopic methods. The combination of nc-AFM with electrospray ion-beam deposition (ESIBD)32,33 to enable the soft landing of mass-selected ionic species would open further possibilities, including the analysis of complex glycan mixtures or studies on fragments generated in gas-phase reactions. Based on the findings presented herein, we anticipate that nc-AFM will play an important role in advancing our understanding of the structure of glycans and other complex biomolecules. Methods DFT calculations and simulation of nc-AFM images DFT calculations for all free-standing molecules were performed using the FHI-aims package30. All geometry optimizations were carried out in the gas phase, using the generalized gradient approximation Perdew–Burke–Ernzerhof exchange–correlation functional34. In all the calculations, the light settings for the numerical atomic basis sets were employed. The Brillouin zone was sampled using the Gamma-point. Systems were allowed to relax until the Hellmann-Feynman forces reached values below 10–2eV Å–1. Theoretical nc-AFM maps were calculated with the Probe-Particle package24 for a CO-like tip, using the DFT-calculated structure and electric potential as an input. The AFM probe was modeled as the outermost atom of the metal tip to which a CO molecule (the probe particle) is covalently attached via its C atom. Simulated images were obtained by laterally scanning this probe with a step size of 0.1 Å above the sample. At each lateral position, the probe was approached towards the sample in steps of 0.1 Å and allowed to relax. Following relaxation at each step, the vertical component of the force exerted on the probe by the sample was calculated and subsequently converted into Δfvalues. The simulated Δf(x,y)mapswerefinally compared to those recorded experimentally. Electrospray deposition of β-cyclodextrin For electrospray deposition (ESD), β-cyclodextrin (from SigmaAldrich, purity ≥97%) was dissolved in a 50:50 (v/v) mixture of water (Milli-Q®) and methanol (HPLC-grade, ≥99.9%) at a concentration of 1mgmL −1. The solution was then loaded into a gastight Hamilton syringe, connected to the emitter of a commercially available UHV4i ESD source35 (Molecularspray Ltd.) through a PEEK capillary. To achieve stable spray conditions, the solution was pumped at a constant flow rate of 300 µLh −1using a syringe pump, with the emitter held at a potential of +2.3 kV. The molecules were deposited onto the (111) surface of an Au single crystal (MaTecK GmbH), held at room temperature in the UHV chamber of the microscope. During deposition, the pressure in the preparation chamber increased to the 10−7mbar range from a base pressure of 10−9mbar. Low-temperature STM and nc-AFM experiments The experiments were conducted at a temperature of 4.2 K using a commercial STM/nc-AFM microscope from CreaTec Fischer & Co. Fig. 4 | Nc-AFM simulations and the electrostatic potential of β-cyclodextrin with its secondary face upwards. The lateral relaxations of CO-like probeparticles at tip-sample distances corresponding to the closest approach of the probe during the oscillation acquiring nc-AFM images shown in b and g are presented asred dots in panels aand f, respectively. Panels b–dhold the nc-AFM simulations without including the electrostatic potential, and sections g–icontain the simulations that include the electrostatic interaction between probe and sample. Panels e,jinclude the top (viewed from the secondary face) and side views of β-CD’s electrostatic potential map, respectively. The Δzvalues define the distance between the outermost atom of the molecule and the probe particle along the z-axis (perpendicular to the surface plane). Article https://doi.org/10.1038/s41467-024-53555-0 Nature Communications | (2024) 15:9482 5 GmbH. Pt/Ir tips, sharpened by focused ion-beam (FIB), were utilized and were further cleaned and shaped by gentle indentations (≈1nm)in the bare metallic substrate. The Au(111) substrate was prepared through repeated cyclesof argon ion (Ar+) sputtering at 1 keV, followed by annealing at ≈800 K. STM topography was acquired in constant current mode with the bias voltage applied to the sample. Due to the low conductivity of the β-cyclodextrins, constant current images were typically performed with very low tunneling currents between 500 fA and 3 pA. In nc-AFM imaging, a qPlus sensor36 (resonant frequency ≈30 kHz; stiffness ≈1800 N m−1) was operated in frequency modulation mode with an oscillation amplitude of 50 pm. The nc-AFM images were obtained in constant height mode using CO-terminated tips. The CO functionalization was performed by moving the tip in constant current to a CO molecule on the Au(111) surface with a setpoint of 20pA tunneling current and 50 mV bias voltage. Once over the CO molecule, the STM feedback loop is opened, and the tip is approached towards the CO until a sudden drop in current and a corresponding change in frequency shift were observed. The STM/nc-AFM images were processed using Gwyddion37, WSxM38, and SpmImage Tycoon39 software. MD simulations MD simulations were performed using the Gromacs 5.1.4 software package40 with the Plumed extension41.β-CD was modeled with the GLYCAM04 force field42–45. The Au(111) surface, approximately 10 × 10 nm in size, was represented using a GolP force field, which explicitly accounts for induced polarization46. The Au(111) atoms were fixed in place, except for the rigid rod model of dipoles. The structure of β-CD was obtained from the Protein Data Bank (PDB ID: 3CGT). The β-CD molecule was positioned on the Au(111) surface with either its primary face up (p ↑) or secondary face up (s ↑). To enhance hydrogen bond stability on the primary face,a 1 kcal mol−1potential (referred to as HBfix) was applied between the heavy oxygen atoms47.Nonbonded interactions were truncated at 1 nm. Electrostatic interactions were treated using the Particle-Mesh Ewald method (PME) with the 1 nm cutoff for the real space term. The LINCS algorithm was used to constrain hydrogen atoms48. The equations of motion were integrated with a 2-fs time step. Each system was minimized, followed by a 5 ns NpT (in the case of solventsimulation) thermalization using a V-rescale thermostat49 with a gradual temperature rise from 10 to 300 K and a coupling constant of 0.1 ps, and an isotropic Berendsen barostat50 with a reference pressure of1 barand a couplingconstant of 1.0 ps.Thefinal production run was performed in the NVT ensemble for 1.1 µs(orfor 100 ns for a set of 10 β-CDs and simulations with HBfix). Finally, the equilibrated molecules were cooled to 5 K in 2 ns to obtain low-energy gas-phase β-CD conformers for further ab initio simulations. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Data availability The data that support the findings of this study are available from Zenodo51 and from the corresponding authors upon request. Optimized DFT structures are provided as Supplementary Data 1, 2. References 1. Dwek, R. A. Glycobiology: toward understanding the function of sugars. Chem. Rev. 96,683–720 (1996). 2. Varki, A. Biological roles of glycans. Glycobiology 27,3–49 (2017). 3. Gray, C. J. et al. Advancing solutions to the carbohydrate sequencing challenge. J. Am. Chem. Soc. 141, 14463–14479 (2019). 4. Ruhaak,L.R.,Xu,G.,Li,Q.,Goonatilleke,E.&Lebrilla,C.B.Mass spectrometry approaches to glycomic and glycoproteomic analyses. Chem. Rev. 118,7886–7930 (2018). 5. Grabarics, M. et al. Mass spectrometry-based techniques to elucidate the sugar code. Chem. Rev. 122, 7840–7908 (2022). 6. Ernst, B. & Magnani, J. L. From carbohydrate leads to glycomimetic drugs. Nat. Rev. Drug Discov. 8,661(2009). 7. Hudak, JasonE. & Bertozzi, CarolynR. Glycotherapy: new advances inspire a reemergence of glycans in medicine. Chem. Biol. 21, 16–37 (2014). 8. Smith, B. A. H. & Bertozzi, C. R. The clinical impact of glycobiology: targeting selectins, Siglecs and mammalian glycans. Nat. Rev. Drug Discov. 20,217–243 (2021). 9. Seeberger, P. H. Discovery of semiand fully-synthetic carbohydrate vaccines against bacterial infections using a medicinal chemistry approach. Chem. Rev. 121,3598–3626 (2021). 10. Djalali,S.,Yadav,N.&Delbianco,M.Towardsglycanfoldamersand programmable assemblies. Nat. Rev. Mater. 9,190–201 (2024). 11. Abb, S. et al. Carbohydrate self-assembly at surfaces: STM imaging of sucrose conformation and ordering on Cu(100). Angew. Chem. Int. Ed. 58, 8336–8340 (2019). 12. Wu, X. et al. Imaging single glycans. Nature 582,375–378 (2020). 13. Anggara, K. et al. Exploring the molecular conformation space by soft molecule–surface collision. J. Am. Chem. Soc. 142, 21420–21427 (2020). 14. Anggara, K. et al. Identifying the origin of local flexibility in a carbohydrate polymer. Proc.NatlAcad.Sci.USA118, e2102168118 (2021). 15. Anggara, K. et al. Direct observation of glycans bonded to proteins and lipids at the single-molecule level. Science 382, 219–223 (2023). 16. Seibel, J. et al. Visualizing chiral interactions in carbohydrates adsorbed on Au(111) by high-resolution STM imaging. Angew. Chem. Int. Ed. 62, e202305733 (2023). 17. Abb, S. et al. Polymorphism in carbohydrate self-assembly at surfaces: STM imaging and theoretical modelling of trehalose on Cu(100). RSC Adv. 9,35813–35819 (2019). 18. Szejtli, J. Introduction and general overview of cyclodextrin chemistry. Chem. Rev. 98,1743–1754 (1998). 19. Gross, L. et al. Organic structure determination using atomicresolution scanning probe microscopy. Nat. Chem. 2, 821–825 (2010). 20. Iwata, K. et al. Chemical structure imaging of a single molecule by atomic force microscopy at room temperature. Nat. Commun. 6, 7766 (2015). 21. Patera, L. L., Queck, F. & Repp, J. Imaging charge localization in a conjugated oligophenylene. Phys. Rev. Lett. 125,176803(2020). 22. Xu, J. et al. Determining structural and chemical heterogeneities of surface species at the single-bond limit. Science 371,818–822 (2021). 23. de la Torre, B. et al. Submolecular resolution by variation of the inelastic electron tunneling spectroscopy amplitude and its relation to the AFM/STM signal. Phys.Rev.Lett.119, 166001 (2017). 24. Hapala, P. et al. Mechanism of high-resolution STM/AFM imaging with functionalized tips. Phys. Rev. B 90, 085421 (2014). 25. Jelínek, P. High resolution SPM imaging of organic molecules with functionalized tips. J. Phys. Condens. Matter 29,343002(2017). 26. Pavliček, N. & Gross, L. Generation, manipulation and characterization of molecules by atomic force microscopy. Nat. Rev. Chem. 1, 0005 (2017). 27. Moll, N., Gross, L., Mohn, F., Curioni, A. & Meyer, G. A simple model of molecular imaging with noncontact atomic force microscopy. N. J. Phys. 14,083023(2012). Article https://doi.org/10.1038/s41467-024-53555-0 Nature Communications | (2024) 15:9482 6 28. Hapala, P., Temirov, R., Tautz, F. S. & Jelínek, P. Origin of highresolution IETS-STM images of organic molecules with functionalized tips. Phys. Rev. Lett. 113, 226101 (2014). 29. Hapala, P. et al. Mapping the electrostatic force field of single molecules from high-resolution scanning probe images. Nat. Commun. 7, 11560 (2016). 30. Blum, V. et al. Ab initio molecular simulations with numeric atomcentered orbitals. Comput. Phys. Commun. 180,2175–2196 (2009). 31. Stachowicz, A., Styrcz, A., Korchowiec, J., Modaressi, A. & Rogalski, M. DFT studies of cation binding by β-cyclodextrin. Theor. Chem. Acc. 130,939–953 (2011). 32. Rauschenbach, S. et al. Electrospray ion beam deposition: softlanding and fragmentation of functional molecules at solid surfaces. ACS Nano 3,2901–2910 (2009). 33. Rauschenbach,S.,Ternes,M.,Harnau,L.&Kern,K.Massspectrometry as a preparative tool for the surface science of large molecules. Annu. Rev. Anal. Chem. 9,473–498 (2016). 34. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys.Rev.Lett.77, 3865–3868 (1996). 35. Satterley, C. J. et al. Electrospray deposition of fullerenes in ultrahigh vacuum: in situ scanning tunneling microscopy and photoemission spectroscopy. Nanotechnology 18, 455304 (2007). 36. Giessibl, F. J. The qPlus sensor, a powerful core for the atomic force microscope. Rev. Sci. Instrum. 90, 011101 (2019). 37. Nečas, D. & Klapetek, P. Gwyddion: an open-source software for SPM data analysis. Open Phys. 10,181–188 (2012). 38. Horcas, I. et al. WSXM: a software for scanning probe microscopy and a tool for nanotechnology. Rev. Sci. Instrum. 78,013705(2007). 39. Riss, A. SpmImage Tycoon: organize and analyze scanning probe microscopy data. J. Open Source Softw. 7, 4644 (2022). 40. Van Der Spoel, D. et al. GROMACS: fast, flexible, and free. J. Comput. Chem. 26,1701–1718 (2005). 41. Tribello, G. A., Bonomi, M., Branduardi, D., Camilloni, C. & Bussi, G. PLUMED 2: new feathers for an old bird. Comput. Phys. Commun. 185,604–613 (2014). 42. Kirschner,K.N.&Woods,R.J.Solventinteractionsdeterminecarbohydrate conformation. Proc. Natl Acad. Sci. USA 98,10541–10545 (2001). 43. Basma, M., Sundara, S., Çalgan, D., Vernali, T. & Woods, R. J. Solvated ensemble averaging in the calculation of partial atomic charges. J. Comput. Chem. 22, 1125–1137 (2001). 44. Kirschner,K.N.&Woods,R.J.Quantummechanicalstudyofthe nonbonded forces in water−methanol complexes. J. Phys. Chem. A 105,4150–4155 (2001). 45. Cézard, C., Trivelli, X., Aubry, F., Djedaïni-Pilard, F. & Dupradeau, F.-Y. Molecular dynamics studies of native and substituted cyclodextrins in different media: 1. Charge derivation and force field performances. Phys. Chem. Chem. Phys. 13, 15103–15121 (2011). 46. Iori, F., Di Felice, R., Molinari, E. & Corni, S. GolP: an atomistic forcefield to describe the interaction of proteins with Au(111) surfaces in water. J. Comput. Chem. 30, 1465–1476 (2009). 47. Kührová, P. et al. Computer folding of RNA tetraloops: identification of key force field deficiencies. J. Chem. Theory Comput. 12, 4534–4548 (2016). 48. Hess,B.,Bekker,H.,Berendsen,H.J.C.&Fraaije,J.G.E.M.LINCS:a linear constraint solver for molecular simulations. J. Comput. Chem. 18, 1463–1472 (1997). 49. Bussi, G., Donadio, D. & Parrinello, M. Canonical sampling through velocity rescaling. J. Chem. Phys. 126,014101(2007). 50. Berendsen,H.J.C.,Postma,J.P.M.,vanGunsteren,W.F.,DiNola,A. & Haak, J. R. Molecular dynamics with coupling to an external bath. J. Chem. Phys. 81,3684–3690 (1984). 51. Grabarics, M. et al. Atomically resolved imaging of the conformations and adsorption geometries of individual β-cyclodextrins with non-contact AFM. Zenodo https://doi.org/10.5281/zenodo. 13897824 (2024). Acknowledgements This work was supported by the Research Infrastructure NanoEnviCz, supported by the Ministry of Education, Youth and Sports of the Czech Republic (MEYS CR) under Project No. LM2023066 (B.T.); the DFG Walter Benjamin Program (Project GR 6150/1-1) (M.G.); the Czech Science Foundation 23-06781M (B.M., A.J.-M., and B.d.l.T.); the Czech Science Foundation 20-13692X (P.J., M.O., and S.E.); the UKRI Future Leaders Fellowship [MR/V02213X/1] (W.B.S.); the UKRI BBSRC (project BB/ W017024/1) (W.B.S., S.R., and M.G.); the ERDF/ESF project TECHSCALE (No. CZ.02.01.01/00/22_008/0004587) (P.B., M.P., B.T., and P.K.). Computational resources provided by the e-INFRA CZ project (ID:90254), supported by MEYS CR, are acknowledged by M.O., S.E., and P.J. Author contributions B.T., P.J., S.R., W.B.S., and M.G. conceived the research. B.M., M.G., A.J.- M., and B.T. performed the experiments and analyzed the data. S.E., M.O., and P.J. performed the DFT calculations and nc-AFM simulations. P.K., M.P., and P.B. performed the MD simulations. B.T., P.J., S.R., and P.B. supervised the project. All authors contributed to the writing of the manuscript. These authors contributed equally: M.G., B.M., and S.E. Competing interests The authors declare no competing interests. Additional information Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41467-024-53555-0. Correspondence and requests for materials should be addressed to Pavel Banáš, Stephan Rauschenbach, Pavel Jelínek or Bruno de la Torre. Peer review information Nature Communications thanks Leo Gross, Shijing Tan, and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available. Reprints and permissions information is available at http://www.nature.com/reprints Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/ licenses/by/4.0/. © Crown 2024 Article https://doi.org/10.1038/s41467-024-53555-0 Nature Communications | (2024) 15:9482 7