Pieter J.
Gilissen‡
a,
Annemiek D.
Slootbeek‡
a,
Jiangkun
Ouyang
a,
Nicolas
Vanthuyne
b,
Rob
Bakker
a,
Johannes A. A. W.
Elemans
*a and
Roeland J. M.
Nolte
*a
aInstitute for Molecules and Materials, Radboud University, Heyendaalseweg 135, 6525 AJ Nijmegen, The Netherlands. E-mail: r.nolte@science.ru.nl; j.elemans@science.ru.nl
bAix Marseille Univ, CNRS, Centrale Marseille, iSm2, Marseille, France
First published on 13th January 2021
The construction of macromolecular hosts that are able to thread chiral guests in a stereoselective fashion is a big challenge. We herein describe the asymmetric synthesis of two enantiomeric C2-symmetric porphyrin macrocyclic hosts that thread and bind different viologen guests. Time-resolved fluorescence studies show that these hosts display a factor 3 kinetic preference (ΔΔG‡on = 3 kJ mol−1) for threading onto the different enantiomers of a viologen guest appended with bulky chiral 1-phenylethoxy termini. A smaller kinetic selectivity (ΔΔG‡on = 1 kJ mol−1) is observed for viologens equipped with small chiral sec-butoxy termini. Kinetic selectivity is absent when the C2-symmetric hosts are threaded onto chiral viologens appended with chiral tails in which the chiral moieties are located in the centers of the chains, rather than at the chain termini. The reason is that the termini of the latter guests, which engage in the initial stages of the threading process (entron effect), cannot discriminate because they are achiral, in contrast to the chiral termini of the former guests. Finally, our experiments show that the threading and de-threading rates are balanced in such a way that the observed binding constants are highly similar for all the investigated host–guest complexes, i.e. there is no thermodynamic selectivity.
Our research aims at encoding information into synthetic polymers using a macrocyclic porphyrin catalyst based on the glycoluril framework, i.e.Mn1 (Fig. 1) that can thread onto a polymer containing alkene double bonds, e.g. a polybutadiene chain, and processively epoxidize these double bonds.16–18 Since we intend to write a binary code on said polymer in the form of chiral epoxides ((R,R)-epoxide = digit 0, (S,S)-epoxide = digit 1), our desired writing system requires a chiral variant of Mn1 that is capable of performing sequential processive threading and catalysis. Two important requirements for sequential processive catalysis are spatiotemporal control and unidirectionality,19 both of which are determined by kinetic factors. We hypothesized that unidirectional threading of polymer chain through a catalytic machine might be accomplished by aligning chiral structural information in both the catalyst and the polymer. Therefore, it is crucial that the catalytic machine displays a kinetic preference for one of the enantiomers of a chiral (polymeric) guest. In earlier work, we reported on the post-modification of the achiral porphyrin cage H2120 by providing it with a nitro function, yielding a racemic mixture of planar chiral nitro-functionalized porphyrin cage H22.21 The enantiomers of H22 could be resolved by chiral HPLC22 and by installing a chiral auxiliary on the corresponding amino-functionalized host.23 The enantiopure hosts investigated earlier were shown to have small differences in affinity for chiral viologen guests (up to 3-fold difference in binding constant Kassoc).22,23 In this paper we report an alternative approach to synthesize a chiral C2-symmetric porphyrin cage compound, i.e. a host in which the porphyrin macrocycle is linked to the glycoluril framework via chiral linkers (H23, Fig. 1). Furthermore, we show that the enantiomers of H23 display a kinetic preference for the threading and binding of viologen guests equipped with a chiral head group.
Hence, we designed a stepwise protocol for the introduction of the chiral centers. The Mitsunobu reaction of 2-benzyloxyphenol with ethyl (S)-lactate or ethyl (R)-lactate afforded esters (R)-5 and (S)-5 in 62% and 72% yield, respectively. After quantitative removal of the benzyl protecting group, the other chiral center was installed with a second Mitsunobu reaction. In this way, diesters (R,R)-7 and (S,S)-7 were obtained in reasonable yields (57% and 74%, respectively) and with excellent diastereomeric ratios (>98:2). Both diesters 7 were reduced with lithium aluminum hydride, to afford the corresponding primary alcohols 8 in excellent yields. The primary alcohols were then transformed into tosylate leaving groups, affording (R,R)-9 (83%) and (S,S)-9 (99%). According to the established procedure for the synthesis of achiral cage compound H21,28 the N,O-acetals of compound 10 were activated with zinc chloride in the presence of thionyl chloride, and the resulting N-acyliminium intermediates were reacted with electron-rich arenes (R,R)-9 and (S,S)-9 to afford chiral clip molecules (R,R,R,R)-11 (38%) and (S,S,S,S)-11 (36%). The syntheses of the two enantiomers of H23 were completed by reacting chiral clips 11 under highly dilute basic conditions with 1 equivalent of porphyrin tetraol H212. After two chromatographic purification steps (alumina followed by 60H silica), the enantiopure cage compounds (R,R,R,R)-H23 (8%) and (S,S,S,S)-H23 (5%) were obtained as purple solids. Electronic circular dichroism (ECD) measurements (Fig. 2B) showed that the chirality of the flexible spacers is clearly transferred to the porphyrin, as opposite signs of the CD signals were observed for the Soret bands (λmax = 416 nm) of the enantiomeric hosts (R,R,R,R)-H23 and (S,S,S,S)-H23. 1H NMR analysis (Fig. 2C) revealed that the methyl groups of the C2-symmetric macrocycles are in uncommon – highly shielded – chemical environments, as their protons resonate at negative chemical shift (assigned as protons a and b). This effect can be explained by the close proximity of the shielding area of the aromatic porphyrin ring, indicating that the methyl groups are pointing into the cavity of the cage. 2D ROESY experiments indicated the close proximity of the methyl groups a and b, as we observed mutual NOE contacts. DFT calculations (Fig. 2D) confirmed that the minimum energy structure of H23 is the one with the methyl groups pointing into the cavity of the porphyrin macrocyclic host.
Entry | Guest | Host | k on (M−1 s−1) | ΔG‡on (kJ mol−1) | ΔΔG‡on (kJ mol−1) | k off (s−1) | ΔG‡off (kJ mol−1) | K assoc (M−1) | ΔGassoc (kJ mol−1) |
---|---|---|---|---|---|---|---|---|---|
a Values taken from ref. 20. b Determined by fluorescence titration. c Values taken from ref. 30. d Estimated error 20%. e K assoc = kon/koff, unless stated otherwise, estimated error 30%. See the ESI for details. | |||||||||
1a | 13 | H21 | — | — | — | — | — | 6.0 × 105 | −33.0 |
2b | 13 | (R,R,R,R)-H23 | — | — | — | — | — | 1.9 × 106 | −35.8 |
3b | 13 | (S,S,S,S)-H23 | — | — | — | — | — | 1.6 × 106 | −35.4 |
4c | 14 | H21 | 4.0 × 104 | 46.7 | — | 1.4 × 10−3 | 89.3 | 2.9 × 107 | −42.5 |
5 | 14 | (R,R,R,R)-H23 | 2.0 × 103 | 54.1 | +0.0 | 1.3 × 10−4 | 95.2 | 1.5 × 107 | −41.0 |
6 | 14 | (S,S,S,S)-H23 | 2.0 × 103 | 54.1 | 1.2 × 10−4 | 95.3 | 1.7 × 107 | −41.2 | |
7b | (R,R)-15 | H21 | — | — | — | — | — | 8.6 × 106 | −39.6 |
8 | (R,R)-15 | (R,R,R,R)-H23 | 4.2 × 104 | 46.6 | +0.1 | 4.0 × 10−3 | 86.7 | 1.1 × 107 | −40.1 |
9 | (R,R)-15 | (S,S,S,S)-H23 | 4.4 × 104 | 46.5 | 3.9 × 10−3 | 86.7 | 1.1 × 107 | −40.2 | |
10b | (S,S)-15 | H21 | — | — | — | — | — | 1.1 × 107 | −40.1 |
11 | (S,S)-15 | (R,R,R,R)-H23 | 4.4 × 104 | 46.5 | +0.1 | 3.7 × 10−3 | 86.9 | 1.2 × 107 | −40.4 |
12 | (S,S)-15 | (S,S,S,S)-H23 | 4.5 × 104 | 46.4 | 4.1 × 10−3 | 86.6 | 1.1 × 107 | −40.2 | |
13 | (R,R)-16 | H21 | 3.6 × 104 | 47.0 | — | 2.3 × 10−3 | 88.0 | 1.5 × 107 | −41.0 |
14 | (R,R)-16 | (R,R,R,R)-H23 | 7.8 × 103 | 50.8 | −1.1 | 4.7 × 10−4 | 92.0 | 1.7 × 107 | −41.2 |
15 | (R,R)-16 | (S,S,S,S)-H23 | 4.9 × 103 | 51.9 | 3.7 × 10−4 | 92.6 | 1.3 × 107 | −40.6 | |
16 | (S,S)-16 | H21 | 3.7 × 104 | 46.9 | — | 2.4 × 10−3 | 87.9 | 1.5 × 107 | −41.0 |
17 | (S,S)-16 | (R,R,R,R)-H23 | 4.7 × 103 | 52.0 | +1.0 | 3.4 × 10−4 | 92.8 | 1.4 × 107 | −40.8 |
18 | (S,S)-16 | (S,S,S,S)-H23 | 7.0 × 103 | 51.0 | 4.4 × 10−4 | 92.1 | 1.6 × 107 | −41.1 | |
19 | (R,R)-17 | H21 | 1.8 × 103 | 54.4 | — | 1.5 × 10−4 | 94.9 | 1.2 × 107 | −40.4 |
20 | (R,R)-17 | (R,R,R,R)-H23 | 7.2 × 101 | 62.4 | +0.0 | 3.1 × 10−5 | 98.7 | 2.3 × 106 | −36.3 |
21 | (R,R)-17 | (S,S,S,S)-H23 | 7.1 × 101 | 62.4 | 2.9 × 10−5 | 98.9 | 2.5 × 106 | −36.5 | |
22 | (S,S)-17 | H21 | 1.9 × 103 | 54.3 | — | 1.4 × 10−4 | 94.9 | 1.3 × 107 | −40.6 |
23 | (S,S)-17 | (R,R,R,R)-H23 | 7.1 × 101 | 62.4 | +0.0 | 2.8 × 10−5 | 99.0 | 2.5 × 106 | −36.5 |
24 | (S,S)-17 | (S,S,S,S)-H23 | 7.2 × 101 | 62.4 | 3.0 × 10−5 | 98.8 | 2.4 × 106 | −36.4 | |
25 | (R,R)-18 | H21 | 4.5 × 103 | 52.1 | — | 2.6 × 10−4 | 93.4 | 1.7 × 107 | −41.3 |
26 | (R,R)-18 | (R,R,R,R)-H23 | 7.2 × 101 | 62.4 | +3.0 | 1.1 × 10−5 | 101.3 | 6.6 × 106 | −38.9 |
27 | (R,R)-18 | (S,S,S,S)-H23 | 2.5 × 102 | 59.3 | 2.3 × 10−5 | 99.4 | 1.1 × 107 | −40.1 | |
28 | (S,S)-18 | H21 | 5.1 × 103 | 51.8 | — | 2.4 × 10−4 | 93.6 | 2.1 × 107 | −41.8 |
29 | (S,S)-18 | (R,R,R,R)-H23 | 2.2 × 102 | 59.6 | −2.9 | 2.4 × 10−5 | 99.3 | 9.2 × 106 | −39.7 |
30 | (S,S)-18 | (S,S,S,S)-H23 | 6.7 × 101 | 62.6 | 1.4 × 10−5 | 100.7 | 4.8 × 106 | −38.1 |
We then studied the threading of achiral polymer 14 through macrocyclic hosts H2130 and H23 (Table 1, entries 4–6). The kon-value for the threading of polymer 14 through H21 was 20 times higher than that for the threading through the two enantiomers of H23. In addition, the koff-value for the de-threading of polymer 14 from H21 was 14 times higher than that for the de-threading of 14 from the enantiomers of H23. Even though the kinetic processes of both threading through and de-threading from H21 are a lot faster compared to those determined for H23, the on- and off-rates are balanced, resulting in similar Kassoc values. The NMR spectra of the host–guest complexes of H23 with 14 again showed deshielding of methyl protons a and b by ∼2 ppm units (see the ESI, Fig. S106†).
To investigate whether enantiomeric viologens with chiral N-substituents would exhibit stereoselective threading behavior through the cavities of the enantiomers of H23, we examined the threading and de-threading processes of chiral viologens (R,R)-15 and (S,S)-15 (Table 1, entries 7–12).23 The threading experiments (Fig. 4A and B) revealed that the enantiomers of H23 display no kinetic preference for the binding of either (R,R)-15 or (S,S)-15. Also the de-threading (see the ESI, Fig. S60–S67†) of these guests from both enantiomers of H23 occurred at the same rate. As a consequence, there is no thermodynamic preference for the binding of the chiral guests. The binding constants of all complexes H23·15 were verified independently by fluorescence titrations and the values were similar to those of H21·15 (see the ESI, Tables S7–S13†). The 1H NMR spectra of all combinations of host–guest complexes H23·15 were nearly fully superimposable (see the ESI, Fig. S107†). The spectra again showed deshielding of the methyl groups of the chiral spacers of the hosts, as we also observed for the host–guest complexes of H23 with achiral guests. We propose that the lack of kinetic selectivity of chiral hosts H23 for threading onto chiral guests 15 arises from the absence of chirality at the termini of the viologen substituents, i.e. the part of the guest molecule that has to initiate the threading process. Previously, we have shown that the threading of polymeric viologens through H21 involves a kinetically favorable ‘entron’ effect, which involves an internal filling of the cavity of the porphyrin macrocyclic host by the first 5–8 atoms of the polymer chain.30 Starting from the chain termini, the stereogenic centers of viologens 15 only appear at the 6th chain atom. We hypothesize that the remote location of the chirality in the chain inhibits the kinetic stereoselectivity of the threading process, and that the initial entron effect is only governed by the threading of the achiral isopropyl termini of the chains of 15, which are identical for both enantiomers. Alternatively, the lack of stereoselectivity could be due to the lack of steric bulk near the chiral centers of guest 15.
To investigate these hypotheses further, we designed and synthesized chiral guest molecules 16–18 (see the ESI, Scheme S1†). Guest 16 contains small sec-butoxy chiral groups at the chain termini and guest 17 contains bulky 1-(4-octylphenyl)ethoxy chiral groups in the center of the chains. Guest 18 contains a viologen binding station appended at the chain termini with chiral 1-phenylethoxy moieties.
We found that guests 16 (Table 1, entries 13–18) display a kinetic preference of a factor 1.6 (ΔΔG‡on = 1 kJ mol−1) for threading through the enantiomers of H23, i.e. (R,R)-16 prefers (R,R,R,R)-H23 over (S,S,S,S)-H23 and (S,S)-16 prefers (S,S,S,S)-H23 over (R,R,R,R)-H23 (Fig. 4C and D). The small selectivity is attributed to the location of the small chiral moieties at the termini of the guests.
Then, we investigated guests 17 (Table 1, entries 19–24), which contain bulky chiral moieties in locations remote from the chain termini. We found that threading of hosts H23 onto guests 17 proceeded very slowly, i.e. 1–3 orders of magnitude slower than measured for the other investigated guests. This deceleration is caused by the steric bulk in the chains of the guests. The threading experiments (Fig. 4E and F) also indicated that guests 17 do not display kinetic selectivity, which is line with the experiments involving guests 15. Hence, for such selectivity the location of the chiral moiety at the terminus of the guest is pivotal.
In a final set of experiments, we investigated the threading of guests 18 (Table 1, entries 25–30), in which the bulky chiral moieties are positioned at the chain termini. The threading rates of guests 17 and 18 were similar, which is expected based on the similar degree of steric bulk. More interestingly, (R,R)-18 showed a factor 3 (ΔΔG‡on = 3 kJ mol−1) kinetic preference for threading host (S,S,S,S)-H23 over (R,R,R,R)-H23 (Fig. 4G). The opposite kinetic selectivity was observed when guest (S,S)-18 was threaded (Fig. 4H). The results indicate that the chiral environment of the chain termini attached to the viologen moiety determines the extent of kinetic selectivity. Interestingly, de-threading of guests 16–18 from hosts H23 (see the ESI, Fig. S68–S103†) followed the same trends as observed for the corresponding threading processes, i.e. a kinetic preference in the threading process is associated with a kinetic preference in the de-threading process. As a result, the thermodynamic selectivity for all diastereomeric complexes of guests 16–18 with hosts H23 is low, i.e. all binding constants are similar. Moreover, all the host–guest complexes H23·16, H23·17, and H23·18 display similar 1H NMR spectra (see the ESI, Fig. S108–S110†). The methyl protons a and b are again deshielded by ∼2 ppm units. In addition, the aromatic viologen protons are shielded by 2–4 ppm units. The complexation induced shifts (CIS-values) for the protons of the macrocyclic hosts H23 and the protons of the guests 16–18 near the viologen binding station are nearly identical for all host–guest complexes H23·16, H23·17, and H23·18 (see the ESI, Tables S65–S67†). Finally, for all the investigated guests 16–18, the kinetic events of threading and de-threading involving H21 occur faster than those involving H23, but the resulting thermodynamic equilibria remain mostly unaffected.
Footnotes |
† Electronic supplementary information (ESI) available: Experimental procedures, characterization data, kinetic data, DFT calculations, and copies of NMR spectra of all new compounds. See DOI: 10.1039/d0sc05233g |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2021 |