Controlling Chiral Self-Sorting in Truxene-Based Self-Assembled Cages

Abstract

Coordination driven self-assembly of achiral components, i.e., hexa-alkylated truxene ligands (L) with bis-metallic complexes (M₂), afforded three chiral face-rotating stereoisomer polyhedra (M₆L₂). By tuning the length of the alkyl chains as well as the distance between both ligands facing each other in the self-assemblies (M₆L₂), one can control the diastereomeric distribution between the expected homo- and hetero-chiral structures.

1. Introduction

The labile nature of the coordination bond has been extensively exploited for the self-assembly, under thermodynamic control, of simple building blocks into sophisticated 2D or 3D architectures [1,2,3]. The latter can be used to target applications such as molecular recognition, drug delivery, molecular separation or even catalysis in confined spaces [4,5,6,7,8,9,10,11,12,13,14]. In line with the work developed in the field of molecular machines, researchers are particularly interested in designing stimuli-responsive self-assembled structures, whose properties or shape can be tuned through light irradiation, application of a current or addition of a chemical [15,16,17,18,19,20,21,22,23]. Nevertheless, most of these self-assembled structures are achiral. Developing new synthetic strategies to convert symmetric ligands into chiral self-assemblies is highly desirable since the latter can be in principle useful for enantioselective sensing or asymmetric transformations [24,25,26,27,28,29]. While the most common synthetic strategy relies on the use of one chiral building block along the self-assembly process [27,30,31], examples in which achiral components are self-sorted [32,33] to form either homochiral [34,35,36,37,38,39,40,41] or heterochiral [34,35,42,43] self-assemblies are also depicted.

In this context, C₃ symmetrical cages have emerged as an interesting class of supramolecular compounds for studying chirality dynamics [44]. Among others, this phenomenon was recently observed with the family of face-rotating polyhedral [45,46,47,48,49,50,51]. They are obtained by the combination of chiral or achiral linkers with face-rotating moieties such as triazatruxene [45,46] or truxene [40,48,49,50,51,52] for example. Hexa-alkylated truxene derivatives exhibit a C_3h-symmetry and show both clockwise (C, green) and anti-clockwise (A, blue) faces as defined by the rotation sense of the three sp3 bridges along the C₃ axis (Figure 1a). Functionalization of the latter with three pyridine units allows for producing prochiral ligands able to self-assemble with bis-metallic complexes into M₆L₂ cages (Figure 1b). Upon self-assembling, the ligands lose their mirror symmetry, which results in several possible stereoisomers, i.e., an enantiomers couple (CC/AA) and the meso form (AC). We showed recently that the hexabutyl truxene ligand LBu reacts with the hydroxynaphtoquinonato diruthenium complex Ru to afford, thanks to a chiral self-sorting process, only the CC/AA enantiomers couple [53]. We hypothesized this selectivity was due to the through space interactions occurring between the butyl chains located within the cavity, which self-organize in an alternated way to minimize the constraints. Herein, we investigate further the role of those interactions while studying other combinations of new hexa-alkyl truxene-based ligands and bis-metallic complexes of various lengths.

2. Results and Discussion

Ligands LEt and LBu (Scheme 1) were synthetized from previously described hexaalkyl truxene derivatives 1a and 1b respectively [54,55,56]. After a selective tri-bromination of the truxene moiety using Br₂ that affords compounds 2a and 2b (>80% yields), pallado-catalysed Suzuki-Miyaura cross-coupling reactions with 4-pyridinylboronic were carried out in a mixture of toluene and ethanol in basic conditions. The target ligands LEt and LBu were obtained in good yields considering that three sites are functionalized during the reaction (60% and 46%, respectively).

Single crystals of ligand LEt were obtained by slow evaporation of its solution in chloroform. The corresponding crystallographic structure is depicted in Figure 2 and compared to the previously described ligand LBu. Both ligands show nearly similar structural characteristics with a planar truxene core which places the three peripherical nitrogen atoms in the same plane. To minimize H-H interactions, the pyridine shows an average rotation angle of 38.6(1)° (LEt) and 40.8(1)° (LBu) with the central truxene core [57]. In accordance with the C₃ symmetry of both ligands, an angle of 120° is measured between each pyridine axis. Finally, the n-alkyl chains are arranged on both opposite sides, almost perpendicularly to the truxene plane.

The self-assembly reactions were proceeded between two equivalents of ligands and three equivalents of bis-metallic complexes, with the objective of forming M₆L₂ cages which associate two truxene entities face to face (Scheme 1). The resulting cavity is thus intended to be filled by the n-alkyl groups which are intercalated between the two aromatic platforms. Our previous work showed the importance of through space interactions between n-alkyl chains to drive the relative spatial organization of the ligands within the self-assembly. With the aim of controlling the ratio between AA/CC and AC species, we studied new ligand-complex combinations, i.e., LEt with Ru and LBu with Rh. Based on the X-ray structures of LEt and LBu, we extracted the distance “a” between the sp³ carbon of the Ar-Ar bridge in the truxene moiety and the sp³ carbon of the terminal methyl group of the alkyl chain (Figure 2). On the other hand, “b” is defined as the intermetallic distance within the Ru and Rh bimetallic complexes as extracted from literature data. From these values, we calculated “c” (= b − 2a) which reflects the minimal distance between terminal methyl groups of opposite dangling alkyl chains as function of the dinuclear bridge (

Table 1. a, b and c values within the self-assembled structures BuRu, EtRu and BuRh.

	Alkyl Chain Length a (Å)	Metal-Metal Distance b (Å)	c = b − 2a (Å)
BuRu	5.0	8.4	−1.6
EtRu	2.6	8.4	2.2
BuRh	5.0	12.9	2.9

). While this value is negative (c = −1.6 Å) for BuRu cage, for which only the CC and AA enantiomers are observed, it increases to 2.9 Å for BuRh, meaning the alkyl chains are too far away to interact in this case. The situation for EtRu appears intermediate.

All reactions were carried out in methanol-d₄ at C = 10⁻³ M and well-resolved ¹H NMR spectra, suggesting the formation of one discrete species, were observed in the three cases after 4 h at 50 °C (Figures S7, S10 and S14). The corresponding cages BuRu, EtRu and BuRh were isolated by precipitation with diethyl ether in ca. 80% yield. ESI-FTICR-HRMS spectrometry experiments were carried out in methanol at C = 10⁻⁴ M on each sample (Figure 3). These measurements confirm the expected M₆L₂ stoichiometry in the three cases, with characteristic multi-charged isotopic patterns localized at m/z = 789.65155, 1024.30257 and 1415.38762 for compound BuRu (main contributions, Figure S17), at m/z = 721.97871, 939.71148 and 1302.59947 for EtRu (main contributions, Figure S18) and at m/z = 735.20408, 882.73054, 1089.06712 and 1398.57199 for BuRh (main contributions, Figure S19).

The ¹H NMR, ¹H COSY NMR and ¹H DOSY NMR spectra of the three cages are shown in Figure 4 (high-field region) and Figures S7–S16. The diffusion measurements revealed in each case one single diffusion value (D). BuRu and EtRu self-assemblies exhibit a similar diffusion coefficient close to 3.0 × 10⁻¹⁰ m²·s⁻¹ while this value decreases for BuRh cage (D = 2.3 × 10⁻¹⁰ m²·s⁻¹), as expected for a larger edifice. Hydrodynamic radii of ca. 13 Å and 17 Å were calculated from the Stokes-Einstein equation [58], in agreement with the formation of M₆L₂ species. The case of BuRu was deeply investigated in a previous work, establishing the exclusive formation of the D₃ symmetric CC and AA enantiomers [53].

This was assigned to through-space van der Waals interaction between butyl chains facing each other inside the cavity (c = −1.6 Å). While in this case only one set of signals for the truxene backbone (Figure S7, protons α, β, a, b and c) and for methyl groups from the alkyl chains is observed, these signals are splitted in the cases of EtRu (Figures S10, S11 and Figure 4d,d’) and BuRh (Figure S14 and Figure 4e), suggesting the presence of the mixture of diastereoisomers. The relative integrals of protons 4 (BuRu and BuRh) and protons 2 (EtRu) were used to determine the proportion of enantiomers AA/CC vs. the meso form AC. On this basis, a 50/50 statistical mixture of the diastereoisomers is determined for BuRh (Figure 4e), which confirms the absence of interaction between both opposite faces in the structure (c = 2.9 Å). A similar ratio is observed for EtRu at short reaction time (Figure 4d) but the latter evolves to 66/33 after several hours (Figure 4d’). This indicates that the reaction is thermodynamically driven and confirms the crucial role of the inter-ligand interactions inside the cavity. Regarding (i) the evolution of the ratio over time, (ii) the selectivity observed for BuRu and (iii) the statistical mixture observed for BuRh, the dominant species in EtRu should be the couple of enantiomers AA/CC.

Single crystals were obtained for self-assemblies BuRu and BuRh from slow diffusion of methyl tert-butyl ether in their methanolic solutions (Figure 5). The BuRu cage crystallized in the non-centrosymmetric P3₁21 space group with a Flack parameter of 0.44 indicating enantio-enriched crystal (enriched in AA) and the BuRh cage in the centrosymmetric C 2/c space group. Interestingly, if BuRh solution speciation revealed a statistical mixture of all stereoisomers, only AA/CC enantiomeric pair was observed in the crystalline phase. Despite numerous attempts, no single crystals of EtRu suitable for XRD measurements could be obtained. We have therefore modeled the corresponding compound by MM+ calculations using the CC-BuRu structure as a starting-point. Unlike topologically similar metalla-structures in which a large tilt of the bis-ruthenium complexes allows the facing ligands planes to be closer in order to maximize the π-π interactions [59,60], the alkyl chains present inside the cavity of BuRu, EtRu and BuRh strongly limits structural deviation from right prismatic geometry and maintain both truxene moieties at distances close to the intermetallic one. These are however slightly shorter than the M-M distances. As a result, the trigonal prisms are only slightly distorted with average Bailar angles ranging from 7.9° to 16° (Figure S20). Moreover, in each case, the pyridyl rings are tilted by ca. 20° out of the plane of the truxene moiety. All these parameters generate two types of chirality, a double rosette (M/P) and a propeller isomerism (Δ/Λ). Both are dependent one other in these structures since only the enantiomers (AA, M, Δ) and (CC, P, Λ) are observed [60,61].

3. Materials and Methods

3.1. Chemicals

Compounds 1b [56], 2b and LBu [53], as well as compounds 1a and 2a [62], and complexes Ru [63] and Rh [64] were synthesized using procedures described in the literature. All reagents were of commercial reagent grade and were used without further purification. Silica gel chromatography was performed with a SIGMA Aldrich Chemistry SiO₂ (pore size 60 Ã, 40–63 µm technical grades).

3.2. Instrumentation

Characterizations and NMR experiments were carried out on a NMR Bruker Avance III 300 spectrometer at 298 K, using perdeuterated solvents. ¹H DOSY NMR spectra were analyzed with MestReNova software. MALDI-TOF-MS spectra were recorded on a MALDI-TOF Bruker Biflex III instrument using a positive-ion mode. ESI-FTICR mass spectra at very high resolution were performed in positive detection mode on a 7T Solarix 2xR (Bruker Daltonics, Marne la Vallée, France).

3.3. Experimental Procedure and Characterization Data

3.3.1. Ligand LEt

To a stirred solution of 2a (0.50 g, 0.669 mmol) in toluene (25 mL) and EtOH (12 mL) was added 4-pyridine boronic acid (0.33 g, 2.68 mmol, 4 equivalents) at room temperature. Then K₂CO₃ (1.94 g, 14.05 mmol, 21 equivalents) in water (8 mL) was added to the solution at room temperature. The solution was degassed with argon for 40 min at room temperature. Pd(PPh₃)₄ (0.23 g, 30%) was then added to the solution. The solution was stirred at 90 °C. After 64 h, the mixture was cooled to room temperature and extracted with dichloromethane. The organic extracts were washed with water, dried over magnesium sulfate and the solvent were evaporated. The residue was purified by chromatography on silica gel using ethyl acetate/dichloromethane/methanol/triethylamine (from 75/25/0/0 to 42/42/15/1) as an eluant to give ligand LEt as a yellow powder (369 mg, 74%). ¹H NMR (300 MHz, 298 K, CDCl₃): δ 8.74–8.72 (m, 6H), 8.50–8.47 (d, ³J = 8.9 Hz, 3H), 7.76 (s, 3H), 7.75 (d, ³J = 7.1 Hz, 3H), 7.68–7.65 (m, 6H), 3.12–3.05 (m, 6H), 2.31–2.17 (m, 6H), 0.29 (t, ³J = 7.1 Hz, 18H). ¹³C NMR (76 MHz, CDCl₃): δ 153.85, 150.43, 148.23, 145.27, 141.45, 138.50, 136.52, 125.45, 125.20, 121.56, 120.67, 57.18, 29.65, 8.65. HRMS (MALDI-TOF): found: 741.4110; Calculated: 741.4083.

3.3.2. Self-Assembly EtRu

A mixture of LEt (10.00 mg, 13.5 μmol, 2 equiv.) and complex Ru (19.35 mg, 20.0 µmol, 3 equiv.) in methanol (2 mL) was stirred 20 h at 50 °C. Then, diethyl ether (5 mL) was added and the resulting suspension was centrifuged. The resulting solid was washed twice with diethyl ether to give EtRu (24.02 mg, 4.5 μmol, 83%) as a dark solid. ¹H NMR (300 MHz, MeOD): δ 8.50–8.46 (m, 12H), 8.32–8.26 (m, 6H), 7.90–7.77 (m, 24H), 7.32–7.31 (m, 12H), 5.92–5.88 (m, 12H), 5.69–5.65 (m, 12H), 2.92–2.88 (m, 12H), 2.70–2.50 (m, 6H), 2.17–2.16 (m, 24H), 2.02–1.95 (m, 6H), 1.37 (d, ³J = 6.9 Hz, 36H), 0.10–(−0.02) (m, 18H), −0.32–(−0.49) (m, 18H). ¹H DOSY NMR (300 MHz, MeOD) D = 3.03 × 10⁻¹⁰ m²·s⁻¹. FTICR-HRMS (m/z), [EtRu − 3TfO⁻]³⁺: found: 1302.59965, calculated 1302.59947, [EtRu − 4TfO⁻]⁴⁺: found: 939.71154, calculated 939.71148, [EtRu − 5TfO⁻]⁵⁺: found: 721.97871, calculated 72197870.

3.3.3. Self-Assembly BuRh

A mixture of LBu (10.00 mg, 10.6 μmol, 2 equiv.) and complex Rh (23.98 mg, 15.9 µmol, 3 equiv.) in methanol (2.5 mL) was stirred 1 h at 50 °C. Then, diethyl ether (5 mL) was added and the resulting suspension was centrifuged. The resulting solid was washed twice with diethyl ether to give BuRh (25.14 mg, 3.92 μmol, 74%) as a yellow solid. ¹H NMR (300 MHz, 298 K, MeOD): δ 10.38–10.35 (m, 12H), 9.93 (d, ³J = 7.3 Hz, 12H), 8.72–8.68 (m, 12H), 8.47–8.43 (m, 12H), 8.33–8.30 (m 6H), 7.95–7.75 (m, 24H), 2.90–2.86 (m, 6H), 2.71–2.67 (m 6H), 2.13–2.08 (m, 6H), 1.95–1.82 (m, 96H), 0.95–0.87 (12H), 0.50–0.05 (m, 42H), −0.06–(−0.20) (m, 12H), −0.36 (t, ³J = 7.2Hz), −0.44 (t, ³J = 7.3 Hz, 9H). ¹H DOSY NMR (300 MHz, MeOD): D = 2.34 × 10⁻¹⁰ m²·s⁻¹. FTICR-HRMS (m/z): [BuRh − 4OTf⁻]⁴⁺: found: 1398.57256; calculated 1398.57199, [BuRh − 5Otf⁻]⁵⁺: found: 1089.06707; calculated 1089.06712, [BuRh − 6Otf⁻]⁶⁺: found: 882.72998; calculated: 882.73054, [BuRh − 7Otf⁻]⁷⁺: found: 735.20397; calculated: 735.20408.

3.4. Molecular Modelling

Molecular modelling was performed by using the molecular mechanics force field MM+ method from the HyperChem 8.0.3 program (Hypercube, Inc., Waterloo, ON, Canada,) configured in vacuo, with a RMS of 10⁻⁵ kcal/mole, a number of maximum cycles of 32,500, and a Polak-Ribiere algorithm. Counter anions were omitted to simplify the calculation.

3.5. X-ray Crystallographic Analysis

X-ray single-crystal diffraction data were collected at low temperature on a Rigaku Oxford Diffraction SuperNova diffractometer equipped with Atlas CCD detector and micro-focus Cu-Kα radiation (λ = 1.54184 Å). The structures were solved by dual-space algorithm, expanded and refined on F² by full matrix least-squares techniques using SHELX programs (G. M. Sheldrick, SHELXT 2018/2 and SHELXL 2018/3). All non-H atoms were refined anisotropically and multiscan empirical absorption was corrected using CrysAlisPro program (CrysAlisPro 1.171.40.45a and 1.171.41.118a, Rigaku Oxford Diffraction, 2019–2021). The H atoms were placed at calculated positions and refined using a riding model.

The structure refinement of BuRh showed disordered electron density which could not be reliably modeled. The program PLATON/SQUEEZE was used to add the corresponding scattering contribution to the calculated structure factors. This electron density can be attributed to solvent molecules (methyl tert-butyl ether (MTBE)) and missing triflate molecules (48 CF₃SO₃ anions). The assumed solvent composition and missing anions were included in the calculation of the empirical formula, formula weight, density, linear absorption coefficient, and F(000).

Crystallographic data for LEt: C₅₄H₅₁N₃, M = 741.97, T = 200K, colorless prism, 0.243 × 0.137 × 0.108 mm³, monoclinic, space group Cc, a = 16.8429(4) Å, b = 29.2254(7) Å, c = 8.8287(3) Å, β = 104.094(3)°, V = 4215.0(2) Å3, Z = 4, ρcalc = 1.169 g/cm³, μ = 0.513 mm⁻¹, F(000) = 1584, θmin = 3.024°, θmax = 72.419°, 15675 reflections collected, 6363 unique (Rint = 0.0661), parameters/restraints = 520/2, R1 = 0.0631 and wR2 = 0.1739 using 5960 reflections with I > 2σ(I), R1 = 0.0657 and wR2 = 0.1786 using all data, GOF = 1.047, −0.300 < Δρ < 0.237 e.Å-3. CCDC 2183860.

Crystallographic data for BuRh: C₃₇₆H₅₁₆F₃₆N₂₄O₅₆Rh₆S₁₂, M = 7954.28, T = 150K, yellow prism, 0.13 × 0.076 × 0.027 mm³, monoclinic, space group C2/c, a = 39.414(3) Å, b = 24.8816(17) Å, c = 40.900(6) Å, β = 98.08(1)°, V = 39711(7) Å3, Z = 4, ρcalc = 1.330 g/cm³, μ = 3.291 mm⁻¹, F(000) = 16696, θmin = 2.916°, θmax = 73.304°, 85378 reflections collected, 37770 unique (Rint = 0.1374), parameters/restraints = 1315/28, R1 = 0.1173 and wR2 = 0.3152 using 10726 reflections with I > 2σ(I), R1 = 0.2119 and wR2 = 0.3999 using all data, GOF = 0.927, −0.739 < Δρ < 0.594 e.Å-3. CCDC 2183861.

4. Conclusions

In summary, we synthetized a series of three chiral M₆L₂ metalla-cages from prochiral hexa-alkylated truxene ligands and achiral dinuclear metallic complexes. Remarkably, the diastereomeric ratio between enantiomers AA/CC and the meso AC form can be tuned thanks to the non-covalent interactions occurring between the internal alkyl chains upheld by two opposite facing ligands. While only the AA/CC enantiomers couple is observed in solution in the case of BuRu, the metalla-cage BuRh exists as the statistical mixture of the three possible stereoisomers. The latter are also observed for EtRu through preferentially producing the enantioenriched AA/CC forms. These results unambiguously confirm the crucial role of inter-ligand communication through the alkyl chains in the self-sorting process. Work is under progress to evaluate the binding abilities of these M₆L₂ chiral cages.