Translation of rod-like template sequences into homochiral assemblies of stacked helical oligomers.

At the molecular level, translation refers to the production of a new entity according to a template that has a different chemical composition. In this way, chemical information may be translated from one molecule to another. The process is useful to synthesize structures and thus functions that might be difficult to create otherwise, and it reaches exquisite levels of efficiency in biological systems, as illustrated by protein expression from mRNA templates or by the assembly of the tobacco mosaic virus capsid protein according to the length of its RNA. In synthetic systems, examples of template-directed syntheses are numerous, but general and versatile schemes in which a non-natural sequence actually encodes the information necessary to produce a different sequence are few and far from being optimized. Here we show a high-fidelity enzyme-free translation of long rod-like alkylcarbamate oligomers into well-defined sequences of stacked helical aromatic oligoamides. The features present in the rods, which include the number and distance between carbamate functions and stereogenic centres, template the self-assembly of complementary stacks of helices that each have a defined right (P) or left (M) handedness, length and single or double helicity. This process enables the production of very large (>20 kDa) abiotic artificial folded architectures (foldamers) that may, for example, serve as scaffolds to organize appended functional features at positions in space defined with atomic precision across nanometric distances.

Foldaxanes are helix-rod host-guest complexes that have been shown to form upon winding aromatic oligomers around para-phenylenes,12,13 poly-alkylammoniums,14 or dicarbamates derived from α,ω-diaminoalkanes.15,16 In the latter case, arylamide foldamer hosts17 may be either single15 or double16 helical once wound around the rod (Figure 1a, Supplementary Figure 8, Supplementary Table 1). Our set of building blocks comprises short oligomers 1, 2 and 4 along with newly synthesised longer sequences 3, 5, 6 (Figure 1b, Supplementary Figure 1) and rods 7-29 (Figure 1b, Supplementary Figures 2-7). Single station rods 7-14 served to assess the stability of host-guest complexes as a function of guest length. Association constants in CDCl3 were systematically determined by 1H NMR titrations (Supplementary Table 2, Supplementary Figures 9 to 27). In addition, a Van’t Hoff plot allowed us to calculate thermodynamic parameters for the formation of (6)2⊃15 (the sign ⊃ stands for “include”, meaning 15 is included in (6)2). The obtained values (Ka = 590 L.mol–1 at 293 K; ΔH = –41.3 kJ.mol–1, ΔS = –89.0 J.K–1, Supplementary Figure 28) reveal a strongly enthalpically driven process with a large entropic barrier. Complex formation involves hydrogen bonding between carbonyl groups of the guest and amide protons of 2,6-pyridinedicarboxamide units of the host, located at the two ends of the single helices, or at one end of each strand of the double helices. Consequently, a strict match is required between on one hand the distance between hydrogen bond donors on the helices (i.e. the number of helix turns) and on the other hand the length of the alkyl chains connecting hydrogen bond acceptors on the rods.15 In single helical complexes, shrinking the rod by a single CH2 unit may result in a large or even complete loss of stability. In contrast, screw motions within the double helices allow them to adjust their length and to bind to rods differing by 3 to 4 CH2 units with comparable affinities.16 This prior knowledge hinted at the possibility of loading numerous single and/or double helices, each having a defined length, on rods possessing the complementary binding stations arranged in a chosen sequence. The sequence would thus template the assembly of an organised multi-helical supramolecular polymer arrangement (Figure 1a). Templation would be facilitated by the facts that: (i) complex formation does not have to involve a threading mechanism which would require the stepwise introduction of the helices in the order of their arrangement on the rod. Instead, it may also occur via an unfolding of the helix and its refolding around the rod, allowing error correction in the process. Thermodynamic products may thus form regardless of the order in which components are assembled; (ii) both single and double helices may slide along the rods and find their best binding station without dissociating.15,16

Figure 1

Principle of rod-to-foldaxane translation and molecules used.

(a) Schematic representation of the controlled homochiral assembly of single and double stranded helices of various length onto a long dumbbell shaped template possessing complementary binding stations for each helix and terminal stereogenic centres. (b) Formulae of aromatic oligoamides forming single helical (1-3) and double helical (4-6) host-guest helix-rod complexes. In the absence of guest, all oligomers exist predominantly as double helices at thermodynamic equilibrium in 1 mM CDCl3 solutions. (c) Formulae of oligocarbamate guest template sequences.

However, a major unsolved problem was the control of helix handedness. In the absence of chiral information transfer from the rod into a preferred handedness of the helix hosts, and in the absence of end-to-end helix-helix handedness communication between contiguous helices on the rod, the loading of numerous helices would only yield complex mixtures of diastereomeric helix arrangements all having a different spatial organization. Thus, we first aimed to control helix handedness using stereogenic centres on the rod. A series of rods having one helix binding station and bearing one or two terminal chiral groups were prepared (16-20, Supplementary Figure 2) and their host-guest complexes with single or double helical hosts of matching length were investigated (Supplementary Figures 29-34 and 36-37). As illustrated by the emergence of a major species in 1H NMR spectra (Figure 2a), the two chiral phenethyl groups on guest 16 efficiently induced the handedness of 2 (diastereomeric excess d.e. = 93%). A crystal structure of the 2⊃16 complex (Figure 2e, Supplementary Figure 42, Supplementary Table 3) allowed us to unambiguously assign M helicity as being favoured by (R,R) chirality on the rod. This resulted in a strong induced negative circular dichroism (CD) band at 342 nm whilst the (S,S) enantiomer induced a positive band (Figure 2c). Diastereoselectivity was moderate when the rod possessed a single chiral group (3⊃18, see Figure 2d), and also when the host was a double helix ((5)2⊃17, see: Figure 2f, Supplementary Figure 43, Supplementary Table 4). Using a chiral terminal naphthylethyl group (3⊃19 and (6)2⊃20) gave rise to similar handedness induction. Altogether, the five entries shown in Figure 2d indicate subtle variations of handedness induction as a function of the nature (phenethyl or naphthylethyl) and the number (one or two) of chiral groups, the length of the binding station and the single or double helicity of the host. Although a more thorough study would be required to decipher the interplay between all these parameters, the several cases we report already show efficient handedness induction.

Figure 2

Foldaxane assembly, diastereoselectivity and helix handedness induction.

Part of the 700 MHz 1H NMR spectra in CDCl3 at 298K of: (a) 2⊃16-(S,S) with [2] = 1 mM and [16-(S,S)] = 2 mM; (b) (6)2⊃20-(R) with [(6)2] = 1 mM and [20-(R)] = 2 mM. Signals of major diastereomeric complexes are denoted in blue whereas minor complexes are denoted in red and pointed to with red arrows. (c) Circular dichroism (CD) spectra of 2 (20 µM, 313K) in CDCl3 at different time intervals (5 min., 30 min., 60 min., 90 min., 120 min, 150 min. and 210 min.) after the addition of 3 equiv. of 16-(S,S) (blue) or 16-(R,R) (red). (d) d.e. values of helix-rod complexes defined as d.e. = ([M helix] - [P helix]) / ([P helix] + [M helix]) x 100. The preferred helical sense is indicated next to d.e. values. (e) Tube (single helix) and space-filling (rod) representation of the crystal structure of P-2⊃16-(S,S). (f) Tube (double helix) and space-filling (rod) representation of the crystal structure of P-(5)2⊃17-(S,S). Isobutoxy side chains and included solvent molecules have been omitted for clarity.

Next, we assessed helix-helix end-to-end handedness communication when multiple helices were loaded on a multistation rod. Initial attempts using the double helix (4)2 as a host showed that its affinity for guests of matching length was too low to achieve quantitative binding of several stations on a single rod at low mm concentration (Supplementary Table 2). Longer oligomer 5 was thus prepared and shown to bind as a double helix to single station rod 21 with a Ka of 1700 m–1 in CDCl3 at 298 K (Figures 3b, 4b, Supplementary Figures 40, 44, Supplementary Table 5). Rods 22-24 possess two, three and five binding stations identical to that of 21 and may in principle bind to two, three or five (5)2 duplexes, respectively. The stations are separated by an ethylene glycol spacer which plays a critical role in the design. Too short a spacer may cause steric hindrance between duplexes (5)2 bound to contiguous stations resulting in negative binding cooperativity and eventually in unoccupied stations. Too long a spacer and the absence of contacts between adjacent duplexes may result in the absence of helix handedness communication, giving rise to complex mixtures of diastereomeric aggregates. Indeed, [5]foldaxane (5)4⊃22 may exist as a pair of enantiomers PP/MM or as a PM meso species. Similarly, [7]foldaxane (5)6⊃23 may exist as three distinct pairs of enantiomers and [11]foldaxane (5)10⊃24 as ten pairs of enantiomers (Supplementary Figure 38). Upon mixing (5)2 with 22, 23 or 24, 1H NMR initially showed complex patterns which simplified over time, eventually resulting in the emergence of a major species. Integration of the rod and helix signals established that the final aggregate stoichiometry corresponds to binding of helices to all stations on each rod (Figures 4c-e, Supplementary Figure S39). In the case of 22 and 23, the major species was unambiguously identified in the solid state by x-ray crystallography as being racemic homohelical (5)4⊃22 and (5)6⊃23, in which all helices on a given rod have the same handedness (Figure 3d, 3e, Supplementary Figures 45, 46, Supplementary Tables 6, 7). Thus, contacts between helices having a “like” handedness are more favourable than contacts between helices having an “unlike” handedness, resulting in the translation of the sequence of stations on the rods into a well-defined arrangement of helical aromatic oligoamides. It was assumed that the same rule holds true for (5)10⊃24 of which a molecular model was built showing a 9 nm long structure (Figure 3f, Supplementary Movie 1). Indeed, given the behaviour of (5)4⊃22 and (5)6⊃23, we find no reason to suggest the exclusive formation of a conformational isomer of (5)10⊃24 other than the all P or all M. Single crystals of this very large complex (22 kDa) could also be obtained, but diffraction intensity was too weak to resolve the structure. The crystal structures of (5)4⊃22 and (5)6⊃23 revealed direct face-to-face π-π contacts between the helices. At each station, the alkyl moiety of the rod is slightly bent which makes its two ends protrude from the helix at an angle (see Figure 3c). In contrast, the rods adopt a linear conformation at helix-helix junctions where a “like” handedness permits the same tilt angle of each helix with respect to the rod. This alternation of linear and bent segments of the rods gives rise to an undulated shape of the multi-helical aggregates.

Figure 3

Foldaxane structure elucidation.

Solid state structures in the solid state elucidated by x-ray crystallography of (a) dumbbell rod 21; (b) side views (left and center) and top view (right) of (5)2⊃21. The rod is shown in space-filling representation. Either one or both strands of the double helix are shown in tube representation.; (c) rod 22; (d) side view (left) and top view (right) of (5)4⊃22 in space-filling and tube/space-filling representations for the rod and the double helix, respectively; (e) side views of (5)6⊃23. The rod is shown is space-filling representation. The helices are shown in tube (left) or space-filling (right) representations. (f) Energy minimised molecular model using the Merck Molecular Force Field static (MMFFs) of the structure of (5)10⊃24. Isobutoxy side chains and included solvent molecules were omitted for clarity. Only the all P helical isomers are shown. The structures belong to centrosymmetrical space groups and thus also contain the all M isomers.

Figure 4

Solution evidence of the uniqueness of self-assembled foldaxanes.

Part of the 700 MHz 1H NMR in CDCl3 (0.25 mM) showing the amide and some aromatic resonances of: (a) (5)2; (b) (5)2⊃21; (c) (5)4⊃22; (d) (5)6⊃23 and (e) (5)10⊃24. Red dashes illustrate the shielding of terminal aromatic amide protons.

The formation of homomeric multi-helical complexes was further studied in solution using double station rod 25 and double helix (6)2 which forms the most stable complexes and was thus the least prone to dissociate upon heating (Supplementary Figure 35). Upon saturation of the two-station rod, two duplexes are bound to form (6)4⊃25. Aside from the major PP/MM species, minor signals assigned to a PM diastereomer allowed to calculate a d.e. value above 90%. This value did not change significantly with temperature suggesting that more stable PP/MM diastereomers are favoured by enthalpy, with negligible involvement of entropy.

Having established rod-helix and helix-helix chiral communication independently, we challenged ourselves to integrate all types of information that a rod may contain to be translated into an organised sequence of helices: chiral groups to control absolute helix handedness, several stations to bind to several helices, different stations to bind to different helices, e.g. a single and a double helix (Figure 5e). Rods 26 and 27 possess two distinct and long stations to ensure a high thermodynamic stability of their complexes with the matching helices, separated by an ethylene glycol spacer to induce helix-helix end-to-end handedness communication. A terminal phenethyl group is placed next to a station complementary to the single helix of 3, and a terminal naphthylethyl group is placed next to a station complementary to the duplex (6)2. Rods 26 and 27 differ from the relative stereochemistry of their two stereogenic centres which may favour the binding of two hosts having opposite handedness in the case of 26 or the same handedness in the case of 27, thus acting either antagonistically or synergistically with respect to helix-helix handedness communication. Indeed, upon mixing 3 and (6)2 with either 26-(R,S) or 27-(R,R), very different outcomes resulted. With 26-(R,S), a complex NMR spectrum formed corresponding to a mixture of four possible P(P)2, M(M)2, P(M)2 and M(P)2 (3.(6)2)⊃26 complexes none of which had a strong prevalence (Figure 5c). In contrast, with 27-(R,R), a sharp 1H NMR spectrum indicated the presence of a dominant species (Figure 5d) whose structure was identified in the solid state as the favoured homohelical arrangement (Figure 5e,f, Supplementary Figure 47, Supplementary Table 8). The structure showed that the helix-helix end-to-end homohelical contact is similar for single helix-double helix communication to that found for double helix-double helix communication. Thus, the stereogenic centres at each end of the rod as well as the helix-helix contact, although all remote from each other,18 cooperatively contribute to the emergence of a complex multi-helical aggregate through the translation of information contained on a multistation guest.

Figure 5

Formation of heteromeric stacks of oligomers on heteromeric rods.

Part of the 400 MHz 1H NMR spectra showing resonances of the pivaloyl end groups of the helices in CDCl3 at 298K of: (a) 3⊃27-(R,R) (1 mM); (b) (6)2⊃27-(R,R) (1 mM); (c) 3 (1 mM) and (6)2 (1 mM) in the presence of rod 26-(R,S) (1 equiv.); (d) 3 (1 mM) and (6)2 (1 mM) in the presence of rod 27-(R,R) (1 equiv.). White diamonds denote M-3⊃27-(R,R) whereas M-(6)2⊃27-(R,R) are marked with black diamonds. Red circles denote the homochiral complex (M-3.M-(6)2)⊃27-(R,R). Side view of: (e) the structure of rod 27-(R,R); (f) the crystal structure of (M-3.M-(6)2)⊃27-(R,R). 3 is shown in grey and (6)2 in purple and light purple. (g-i) Details of the crystal structure of (M-3.M-(6)2)⊃27-(R,R) showing different views of the stacked pyridine trimers of 3 (grey tube) and (6)2 (purple tube) which both hydrogen bond to rod 27-(R,R) shown in space-filling representation (g) or in tube representation (i, j). The volumes of the pyridine trimers are shown as grey taupe and purple isosurfaces, respectively, in quad mesh representation. Side chains (OiBu groups) and included solvent molecules have been omitted for clarity.

The various helices 1-6 at our disposal selectively bind guests according to their length (Supplementary Table 2) in such a way that three host-guest complexes may form with negligible cross-association: (6)2⊃13, 3⊃9, and 2⊃7 where 13, 9 and 7 contain undecylene, heptylene and pentylene chains, respectively. For these three complexes, Ka values in CDCl3 are above 104 L.mol–1 and cross association is inferior to 0.2 %. Complexes between 1 and shorter guests would not cross-associate but their use would be hampered by lower binding. Creating additional orthogonal helix-rod association would entail the preparation of longer helices and longer guests, or the use of other helix-rod binding modes.12–14,19 With these complexes in hand, we tested the possibility of increasing information content during multi-foldaxane formation. Rods 28 and 29 were prepared which possess three and five helix-binding stations, respectively, of the three kinds mentioned above. The formation of the complexes was monitored by electrospray ion mobility mass spectrometry and confirmed a perfect match (no outlier was detected) between the nature and number of stations on the rod, and the nature and number of helices assembled on it, which supports the conclusion that each helix locates on the station that it is meant to bind, thus resulting in a high fidelity translation of the sequence of binding stations of the rod. Thus, the sequential loading of 3, (6)2, and 2 yielded each expected intermediate up to final complex 3.62.2⊃28(Figure 6a-6c). Because of its large size (24kDa), the observation of 3.62.2.62.3⊃29 required to push instrumentation to its limits but was successful as well (Figure 6d). Provided the concentration is high enough to achieve quantitative binding, the formation of these sequences is quantitative when equilibrium is reached (see Methods section). In addition, ion mobility mass spectrometry gave access to the collision cross-sections of these objects in the gas phase. Calculated and measured values of foldaxanes from rods 21-24, 28 and 29 are in excellent agreement (Figure 6e,f and Supplementary Figures 40-41). The linear relation between CSSs and molecular mass is consistent with a rigid cylinder model20 and suggests a long persistence length of the multihelix-rod complexes.

Figure 6

Detection of heteromeric foldaxane sequences on heteromeric rods using ion mobility mass spectrometry.

Electrospray mass spectra (ESI-MS) of: (a) 3⊃28; (b) 3.62⊃28; (c) 3.62.2⊃28; and (d) 32.64.2⊃29, ([foldaxane] = 150 µM in CHCl3). The peak annotation [n]z+ indicates the number and nature of the foldamer n and the charge z of the detected complexes. Only the expected stoichiometries of the oligomers loaded on each rod are detected. Peaks marked with a star correspond to the expected masses +625 from an unidentified contaminant. (e,f) Comparison of collisional cross sections (DTCCSHe, nm2) either experimental (●) or calculated assuming a rigid cylinder model (▲): (e) homomeric assemblies 52⊃21, 54⊃22, 56⊃23, 510⊃24; (f) heteromeric assemblies 3⊃28, 3.62⊃28, 3.62.2⊃28, 32.64.2⊃29.

In summary, we have established a robust and versatile scheme to produce well-defined homochiral arrangements of helical oligomers wound around multistation rod-like template guest sequences (Supplementary Movies 1-3). Templates having up to ten urethane functions were used to form the largest (up to 24 kDa) abiotic folded architectures known to date. Efficient helix-helix end-to-end handedness communication in discrete aggregates is conceptually different from earlier complexes between polymers and helical14 or macrocyclic21–25 hosts in which no communication between the hosts takes place. It can be related to the piling up of helices into columns with preferred “like”26–28 or “unlike”29 contacts as observed in the solid state. As other helix-rod recognition patterns are being identified,19 template driven helical assemblies may be further expanded and allow the organization in space of various functional groups attached to each helical component. Another extension would consist in optimizing the covalent capture30–31 of these non-covalent assemblies to convert the stack of helices into a single molecular polymeric chains.

Methods section

Preparation and purification of homo- and heteromeric stacks of aromatic amide oligomers on oligo-urethane rods

Conditions differ depending on whether or not the rod possesses two bulky stoppers. Rods devoid of at least one stopper may directly thread themselves into the helix cavity through a much faster process than unfolding-refolding around dumbbell-shaped rods. Conversely, the kinetic stability of complexes between helices and dumbbell-shaped rods allow their purification using gel permeation chromatography. CDCl3 was the preferred medium for easy monitoring by 1H NMR. Prior to use, CDCl3 was made free from water and traces of HCl to optimize foldaxane formation, by filtration through a pad of activated basic alumina and distillation from calcium hydride.

Rods devoid of at least one stopper

Aliquots of a stock solution of the urethane guest were added to an NMR tube containing the helical host(s) ([host] > 4 mM) in the same solvent. The sample tube was shaken carefully after each addition and 1H NMR spectra were recorded after ten minutes of incubation. Fast threading of the helix on the rod allowed the immediate observation of the corresponding foldaxane. Upon the addition of a stoichiometric amount of guest, the observed yield is quantitative (yield > 99%) as far as NMR can detect) for affinity constants larger than 104 L.mol–1. Note that for multiple station guests and/or chiral guests, a longer time (i.e. several hours) is required to reach stereochemical equilibria (P vs. M helicity).

Rods equipped with two stoppers

In an NMR tube, a stoichiometric amount of helices and dumbbell-shaped guest were dissolved in a minimum of chloroform. Typically a volume of 500 µL was used to dissolve the powder and was subsequently reduced to 50 µL to reach a concentration above 50 mM. High concentrations enhance the kinetics of binding via unfolding-refolding of the helices around the rod. Under these conditions, mass spectrometry shows that each binding station is occupied by the helix that binds it best within one day, but not always with the preferred arrangement of helix handedness, resulting in complicated NMR spectra due to the presence of multiple diastereomers. To reach sterochemical equilibrium, incubation at >50 mM for at least five days at 35°C is required for deca-urethanes. The process may be interrupted, and the sample cooled to room temperature and diluted to monitor advancement by 1H NMR, and resumed if necessary. As for rods devoid of at least one stopper, the yield is the outcome of thermodynamic equilibrium and may be quantitative for affinity constants larger than 104 L.mol–1. The final product may be subjected to purification by recycling gel permeation chromatography (GPC), for example to eliminate an excess amount of helices. For those foldaxanes that yield crystals (sometimes directly in the NMR tube by liquid-liquid diffusion after layering with hexane as a precipitant), crystallization also proved to be efficient at purifying the final product from excess helices.

Nuclear Magnetic Resonance

NMR spectra were recorded on 3 different NMR spectrometers: (1) an Avance II NMR spectrometer (Bruker Biospin) with a vertical 7,05T narrow-bore/ultrashield magnet operating at 300 MHz for 1H observation and 75 MHz for 13C observation by means of a 5-mm direct BBO H/X probe with Z gradient capabilities; (2) an Avance 400 NMR spectrometer (Bruker Biospin) with a vertical 9.4T narrow-bore/ultrashield magnet operating at 400 MHz for 1H observation by means of a 5-mm direct QNP 1H/13C/31P/19F probe with gradient capabilities; (3) an Avance III NMR spectrometer (Bruker Biospin) with a vertical 16.45T narrow-bore/ultrashield magnet operating at 700 MHz for 1H observation by means of a 5-mm TXI 1H/13C/15N probe with Z gradient capabilities. Chemical shifts are reported in parts per million (ppm, δ) relative to the 1H residual signal of the deuterated solvent used. 1H NMR splitting patterns with observed first-order coupling are designated as singlet (s), doublet (d), triplet (t), or quartet (q). Coupling constants (J) are reported in hertz. Data processing was performed with Topspin 2.0 software. Samples were not degassed. CDCl3 from Eurisotop was used after filtration through an alumina pad followed by a distillation over calcium hydride.

Circular Dichroism

Circular dichroism studies were carried out in distilled chloroform using 2 or 10 mm pathlength cell. Homogenization and sample equilibration were performed after each addition of guest and CD spectra were recorded on a Jasco J-815 spectropolarimeter at 298 K.

Mass Spectrometry

Electrospray ion mobility mass spectrometry (ESI-IMMS) experiments were performed on a Agilent 6560 DTIMS-Q-TOF spectrometer (Agilent Technologies) equipped with a dual-ESI source operated in positive ion mode. The oligomers were analysed at a concentration of 150 µM in CHCl3. The syringe pump flow rate was 240 µl hr-1. Ion mobility experiments were performed in helium (P = 3.89 torr). The trap entrance grid delta was set to 1 V to improve the softness and the funnels radio frequency amplitudes was set to 200 V to improve the transmission of high m/z ions. The data were analysed using Agilent MassHunter software (B.07) and IM-MS Browser B.07.01. Ion mobility give us access to the collisional cross section which is the momentum transfer collision integral averaged over all possible ion collision geometries with the helium gas. Collisional cross sections in helium (DTCCSHe) of the ions have been obtained using stepped field experiment (5 electric fields).

Molecular modeling and theoretical collision cross-section (CCS calculation)

Starting structures of (5)2⊃21, (5)4⊃22 and (5)6⊃23 were built from their X-ray structures, respectively. Structures of (5)10⊃24, 3.62.2⊃28 and 32.64.2⊃29 were energy minimised using the MMFFs force field implemented in MacroModel version 8.6 via Maestro version 6.5 (Schrödinger Inc.). All simulations have been carried out using Hyperche 8.0.10 software (hypercube inc.) and the AMBER force field (parm99) has been used. The structures are relaxed and optimised using Polak-Ribiere algorithm until a RMS energy gradient of 0.05 kcal/ A.mol is obtained. The RMSD on all atoms compared to the crystal structure obtained for 3⊃28 and 3.62⊃28 were lower than 0.9 Å which indicate that this force field can perform satisfactorily. The structures are then submitted to unrestrained molecular dynamics in vacuo. Gas phase MD simulations were run in canonical ensemble (at T =296 K) for 1000 ps using an integration time-step of 1 fs. Structures were collected every 25 ps to calculate their CCS. Theoretical CCS calculation have been obtained using the exact hard sphere scattering model (EHSSrot)32 with the Siu’s atom parameters3. 840000 trajectories are computed per ion structure and the average value is used for comparison with the experimental CCS.

Crystallography

Crystallographic data for host-guest complexes 2⊃16, (5)2⊃21 and (5)4⊃22 were collected at the IECB X-ray facility (UMS CNRS 3033 – INSERM US001) on a Rigaku MM007 HF rotating anode (0.8 kW). Data were collected at the copper kα wavelength with a partial chi goniometer (AFC11). All data collection strategies were based on Omega scans. The X-ray source is equipped with high flux Osmic Varimax mirrors and a RAPID SPIDER semi-cylindrical image plate detector. The Rigaku CrystalClear suite version 1.36 and 2.1 were used to index, integrate and scale the data with a multi-scan absorption correction.

Crystallographic data for host-guest complexes (5)2⊃17, (5)6⊃23 and (3.(6)2)⊃27 were collected at the synchrotron ID29 (ESRF) beamline at 0.8 Å. ID29 is a fully automated macromolecular crystallography beamline, intended for high-energy-resolution anomalous dispersion phasing experiments and for high-resolution X-ray diffraction experiments. The beamline is equipped with a MicroDiffractometer (MD2) - which allows to tailor the beam sizes down to 75 (full beam), 50, 30, 20 and 10 microns in diameter. Diffraction data are recorded with a fast readout Pilatus 6M pixel detector detector (Dectris LTD) that allows data collection with a maximum frame rate of 12 images/s. Data were collected with 0.2° increments (180° or 360° in total) and processed with the XDS package34.

For all six crystal structures the resolution of the experimental data allowed phasing by direct methods or charge flipping. SUPERFLIP35, SHELXT36 and SHELXD37 were run in parallel on all dataset and the best solutions were chosen as a starting point for the modelling and refinement processes. The program that gave the best starting model was found to be SHELXT for (5)2⊃21, SHELXD for 2⊃16 and (5)4⊃22, and SUPERFLIP for (5)2⊃17, (5)6⊃23 and (3.(6)2)⊃27.

The starting model of the host-guest molecule (5)6⊃23 was really poor. To accelerate the initial refinement process a “molecular replacement” approach was combined. Superimposition of a refined model of (5)2 in the orthorhombic cell on top of the phasing model helped a lot in getting a full model ready to refine (the superimposition was made “by hand” using the coot software38 and Shelxle software39. All structures were refined by full-matrix least-squares method on F2 with SHELXL-201437. As a rule of thumb, every time where disorder could be modelled with partial occupation, it was so. However, more often than not, no simple models of the side chains and solvents can be obtained because multiple positions exists and because complexity is introduced by the swapping of side chains and solvent molecules at some positions, leading to unstable refinement. Introducing too many restraints on the side chains destabilizes the main chain and thus also has to be ruled out. Under such circumstances, an approach that many crystallographers would follow would be to squeeze out the disordered areas that cannot be modelled. However, we did not use this approach which amounts to removing information from our data and which does affect the main chain displacement parameters. Instead, the side chains are modelled using the EADP constraints (equal atomic displacement parameters) and in most cases refined with isotropic atomic displacement parameters. The outcome is a view of the side chains in an average position which, in our view represents a real information contained in the data. Such modelling is efficient but due to radiation induced damage the position of some atoms from side chains could not be established (only in structure 23⊃(5)6). For all non-hydrogen atoms attempts to introduce anisotropic displacement parameters were made. However, whenever the ellipsoids have adopted unrealistic shape isotropic model was employed. FVAR instruction was used to force some of the isotropic temperature parameters to be equal. In order to model anisotropic displacement parameters ISOR, DELU and RIGU instructions were used. If it was necessary geometry of molecules was improved using DFIX, FRAG, FLAT or AFIX66 and AFIX116 SHELX instructions. Hydrogen atoms were positioned in idealised positions and refined with a riding model, with Uiso constrained to 1.2 Ueq value of the parent atom (1.5 Ueq when CH3). The positions and isotropic displacement parameters of the remaining hydrogen atoms were refined freely.

The final cif file was checked using IUCR’s checkcif algorithm. A - Level and B - level alerts were detected. These alerts are unavoidable and inseparably connected with the data quality (weak intensities, moderate resolution), crystal composition (big size of the molecules, large, heavily disordered parts, large solvent content, etc.) and decisions made during data refinement (i.e. isotropic displacement parameters for non-H atoms, SQUZEEZE40 procedure used only for solvent molecules (and after all attempts of modelling the disordered side chains) and do not reflect errors. All the A and B alerts (except for data resolution) concern the disordered solvent molecules and side chains of the helices but not the main chains or rod-shaped (guests) molecules.