Dynamic Metalloporphyrin‐Based [2]Rotaxane Molecular Shuttles Stimulated by Neutral Lewis Base and Anion Coordination

Abstract A series of dynamic metalloporphyrin [2]rotaxane molecular shuttles comprising of bis‐functionalised Zn(II) porphyrin axle and pyridyl functionalised macrocycle components are prepared in high yield via active metal template synthetic methodology. Extensive variable temperature 1H NMR and quantitative UV‐Vis spectroscopic titration studies demonstrate dynamic macrocycle translocation is governed by an inter‐component co‐ordination interaction between the macrocycle pyridyl and axle Zn(II) metalloporphyrin, which serves to bias a ‘resting state’ co‐conformation. The dynamic shuttling behaviour of the interlocked structures is dramatically inhibited by the addition of a neutral Lewis base such as pyridine, but can also be tuned via post‐synthetic rotaxane demetallation of the porphyrin axle core to give free‐base, or upon subsequent metallation, Ni(II) [2]rotaxane analogues. Importantly, the Lewis acidic Zn(II) porphyrin axle component is also capable of coordinating anions which induces mechanical bond shuttling behaviour resulting in a novel optical sensing response.


Introduction
[17][18] The exquisite topological control afforded by the mechanical bond in MIMs frequently affords enhanced host-guest binding behaviour, with increased selectivity and binding affinities over their non-interlocked analogues. [19]The prevalence of anions in biological and environmental processes has led to marked interest in the sensing of anions.[27][28][29][30][31][32][33] Typically, such dynamic motion arises from an anion being bound in the cavity between macrocycle and axle components.Herein, we report the high yielding synthesis and characterisation of a series of dynamic metalloporphyrin based [2]rotaxanes, comprising of a pyridyl functionalised macrocycle and an axle with a central zinc(II) metalloporphyrin.NMR and UV-visible spectroscopic investigations reveal such an interlocked structure possesses an inter-component macrocycle pyridyl•••Zn(II) porphyrin axle co-ordination interaction, imparting significant bias towards a 'resting state' co-conformation in which the macrocycle co-ordinates to the axle.The disruption of this binding interaction through the competitive binding of either a neutral pyridine ligand or co-ordinating halide anion guest species, leads to displacement, and therefore translocation, of the macrocycle.Importantly, the metalloporphyrinanion binding event results in an optical response in the rotaxane axle's porphyrin-based absorption which serves as a novel transduction mechanism for signalling mechanical bond dynamic shuttling behaviour and detecting anions (Figure 1).
The requisite bis-azide appended Zn(II) porphyrin precursors were synthesised according to Scheme 1b.For both meta-and para-bis-azide phenyl substituted porphyrin precursors, the free-base porphyrin was synthesised via an acid-catalysed condensation reaction between the appropriately functionalised azide appended benzaldehyde and dipyrrole methane, followed by a DDQ mediated oxidation.Purification by column chromatography of the crude products afforded the regioisomeric meta-and para-bis-azide phenyl substituted free base porphyrins m-3 • H 2 and p-4 • H 2 in yields of 50 % and 45 %, respectively. [52]Subsequent zinc complexation was achieved through stirring the free-base porphyrins with Zn(OAc With the required components in hand, a general AMT MIM CuAAC synthetic method [48][49][50][51][52] was employed for rotaxane construction, in which either m-3 • Zn or p-4 • Zn and terphenyl functionalised stopper alkyne component [53] were added to a solution of macrocycle precomplexed with a catalytically active Cu(I) source, Cu(MeCN) 4 PF 6 , in 1,2-dichloroethane (DCE) and the reaction mixture left to stir at 60 °C (Scheme 1c).
In both cases, after 5 days, analysis of the crude reaction mixture by TLC indicated complete consumption of the bisazide starting materials and ESI-MS analysis of the crude reaction mixtures revealed signals at m/z = 2274 corresponding to the target isomeric [2]rotaxanes.Subsequent purification by preparative thin layer chromatography gave the desired target meta-and para- [2]rotaxanes, m-5 • Zn and p-6 • Zn, in yields of 30 % and 76 %, respectively, in addition to the corresponding non-interlocked axle components (m-9 • Zn, p-10 • Zn).The isolated [2]rotaxanes and non-interlocked axles were characterised by 1 H, 13 C{ 1 H} NMR spectroscopy and high-resolution electrospray ionisation (HR-ESI) mass spectrometry (See Supporting Information).
The interlocked nature of both [2]rotaxanes was confirmed through comparison of their 1 H NMR spectra in CDCl 3 with the constituent non-interlocked components, a representative example of which for p-6 • Zn is shown in Figure 2. Notably, the axle phenyl protons H 4 and H 5 display a downfield shift compared to the free axle, likely due to ring current effects from proximal macrocycle aromatic protons and a corresponding significant upfield shift (Δδ � 0.25 ppm) was observed in the macrocycle phenyl protons.Perhaps most notable is the significant broadening of the axle methylene signals H 6 and H 8, and the complete disappearance of the triazole signal H 7 .
Interestingly, another species of higher polarity than the desired [2]rotaxanes was also consistently observed during preparative TLC purification.Evidently containing porphyrin, these additional products isolated from the CuAAC-AMT reaction mixtures gave highly symmetrical and well resolved 1 H NMR spectra displaying proton signals from a porphyrin axle and a macrocycle component, which upon integration revealed a 1 : 2 ratio respectively.ESI-MS analysis confirmed the products were the higher order [3]rotaxanes where the porphyrin containing axles are encircled by two macrocycles.It is noteworthy that the isolated yields of the [3]rotaxanes was considerable (8 % and 24 % for m-7 • Zn and p-8 • Zn respectively) which translates to an impressive near quantitative total interlocked product yield of > 99 % in the para case (Scheme 1c).

Investigation of rotaxane inter-component interactions
The observed broadness of the 1 H NMR spectra for [2]rotaxanes m-5 • Zn and p-6 • Zn was attributed to a significant barrier to macrocycle translocation across the length of the axle.It was postulated that the internal pyridine moiety of the macrocycle was co-ordinating to the zinc (II) Lewis acidic centre of the centrally situated metalloporphyrin-containing axle. [54]Indeed, such a MIM inter-component interaction was noted by Mullen and co-workers [55] in their report of Zn(II) porphyrin-axle stoppered rotaxanes containing a pyridyl functionalised macrocycle and in analogous gel resin functionalised [2]rotaxane materials.
To gain further evidence for this co-ordination-induced conformational bias, preliminary UV-Vis measurements of the MIMs and their non-interlocked axle components were recorded in CH 2 Cl 2 solution.Notably for both regioisomeric [2]rotaxanes m-5 • Zn and p-6 • Zn the Soret bands are bathochromically shifted in comparison to their corresponding axles.Such a shift is typically observed upon co-ordination of the zinc(II) metalloporphyrin centre with a Lewis base (Figure 3, p-6 • Zn Soret band λ max = 413 nm, compared with the axle λ max = 410 nm.) [56][57] The addition of excess pyridine to a CH 2 Cl 2 solution of either of the [2]rotaxanes or axles induced similar bathochromic Soret band shifts in λ max .However, the relative magnitude of shift in λ max of the respective rotaxane is less than  that of the corresponding axle, which is consistent with an already present inter-component macrocycle-pyridyl-Zn(II) metalloporphyrin axle interaction in the interlocked systems.
Variable temperature 1 H NMR spectroscopic studies of the [2]rotaxanes were undertaken in CDCl 3 to investigate their dynamic properties.The 1 H NMR spectra of the [2]rotaxanes were recorded across a temperature range of -30 to + 50 °C.In all cases at lower temperatures the splitting of several proton signals, including axle porphyrin derived H 2 and H 3 and macrocycle H c environments, was observed, whilst at higher temperatures these signals coalesced and became increasingly resolved (Figure 4).These observations are consistent with the hypothesis that the [2]rotaxane adopts, at low temperatures, a co-conformation in which the macrocycle resides on one side of the central porphyrin axle core, thereby de-symmetrising the chemical environments of the axle component.At higher temperatures, macrocycle shuttling across the length of the axle is fast on the 1 H NMR timescale.
To elucidate the potential role of the Zn(II) centre in the dynamic properties of [2]rotaxane p-6 • Zn, the free base porphyrin containing [2]   Table 1.Activation parameters for dynamic intramolecular macrocycle shuttling in [2]rotaxanes as determined by variable temperature 1 H NMR spectroscopy (500 MHz, CDCl 3 ).Numbers in parentheses represent calculated errors in activation parameter final digit.The contrasting ΔS � values may be rationalised by taking account of the conformational freedom of the respective rotaxane.For p-6 • Ni and p-6 • H 2 , it is likely that the dominant contribution to the entropy of activation is the reduction in the conformational freedom of the macrocycle as it passes over the widest section of the axle's porphyrin motif.Such an effect was reported by Silvi and co-workers, [58] who observed a similar negative entropy of activation for the translocation of a dibenzo-24-crown-8 macrocycle over an axle amide group which was attributed to the rigidification of the crown ether macrocycle.
In the case of p-6 • Zn, shuttling of the macrocycle causes deviation from the dominant co-conformation in which the metalloporphyrin is axially ligated by the macrocycle pyridyl nitrogen, leading to greater conformational freedom, and a positive ΔS � .Indeed, the negative ΔS � value observed for m-5 • Zn, with a differing axle geometry, demonstrates the interplay between various dynamic processes involved in macrocycle shuttling but is consistent with the weaker Zn(II)-pyridyl interaction in the meta isomer.To further quantify the extent of this inter-component preorganising interaction, quantitative UV-visible spectroscopic titration experiments were conducted to determine the proportion of [2]rotaxanes existing in the inter-component ligated 'resting state' conformation, at room temperature (represented in Figure 7a). [58]The addition of pyridine to solutions of the [2]rotaxanes m-5 • Zn and p-6 • Zn, or the corresponding noninterlocked axles m-9 • Zn and p-10 • Zn, effected bathochromic shifts in the porphyrin Soret-band absorptions.Fitting of the resulting binding isotherms to a 1 : 1 host-guest stoichiometry using Bindfit [59] determined binding constants of pyridine to both the rotaxanes and the non-interlocked axle (K rot and K ax , Figure 7 b and c, respectively).Knowledge of both binding constants enabled the calculation of the equilibrium constant for formation of the 'resting state' in accordance with Equation (1) (Table 2), 54 and therefore the determination of the % of the rotaxane existing in the 'resting state' co-conformation.As expected from the earlier VT NMR studies, calculation of the percentage of [2]rotaxane in the 'resting state' coconformation reveals significant inter-component interaction in p-6 • Zn, with a calculated 'resting state' occupancy of 63 %, and a reduced interaction affording 15 % occupancy in m-5 • Zn.This clear difference in co-conformational bias is consistent with the different activation enthalpies for shuttling, as elucidated by variable-temperature NMR.Accordingly, the significantly higher ΔH � for p-6 • Zn is attributed to the enthalpic penalty of disrupting the 'resting state'.The significantly lower ΔH � for m-5 • Zn, is concordant with the lower enthalpic penalty of disrupting a weaker inter-component interaction.It is notable that ΔH � for m-5 • Zn approaches that of p-6 • H 2 , in which no inter-component interaction occurs and hence negligible bias for the 'resting state' is observed.
where K ax is the binding constant for pyridine to the axle, and K rot for pyridine to the [2]rotaxane (Figure 7).
With convincing evidence for the presence of a strong intercomponent pyridyl•••Zn(II) interaction in p-6 • Zn, the significant higher order [3]rotaxane p-8 • Zn product yield of 24 % (Scheme 1c) may be tentatively rationalised by a pseudo [2]rotaxane intermediate in which one of the two CuAAC stoppering reactions has been catalysed through the macrocycle interior.The strength of the inter-component pyridyl•••Zn(II) interaction retards macrocycle translocation, trapping the pseudo [2]rotaxane such that the second and final triazole forming stoppering reaction is completed by a separate Cu(I)complexed macrocycle, resulting in [3]rotaxane p-8 • Zn formation (Figure 8). [51,59]

Anion binding studies
We next sought to exploit the interruption of the rotaxane inter-component interaction, via an anion coordinating stimulus wherein competing halide anions induce macrocycle translocation, resulting potentially in an optical response.
To this end, UV-visible spectroscopic halide titration experiments were performed on rotaxanes, m-5 • Zn, p-6 • Zn, p-6 • Ni, p-6 • H 2, m-7 • Zn and p-8 • Zn, and the non-interlocked axles m-9 • Zn and p-10 • Zn.Typically, successive additions of the tertbutyl ammonium (TBA) halide salts, TBAX, X=Cl, Br, to 2 μM acetone solutions of the hosts resulted in significant perturbation in the Soret and Q bands of the UV-visible spectra of hosts containing a Zn(II) metalloporphyrin centre with bathochromic shifts greater than 10 nm observed (Figure 9b). [60]A clear isosbestic point was observed in all cases, indicating a 1 : 1 host: guest binding stoichiometry.No such perturbations were observed for iodide, which may be rationalised by the lower charge density, and hence lower Lewis basicity, of the halide anion.Binding isotherms were plotted from the intensity of the Soret band absorption (Figure 9 c) and 1 : 1 host:guest association constants were determined by Bindfit analysis (Table 3). [61]he trend in determined anion association constants mirrors that of halide charge density, namely Cl À > Br À > I À . [39]As expected, in all cases the binding constants for the [2]rotaxanes are significantly reduced over those for the non-interlocked axles.This phenomenon is attributed to the competing rotaxane co-ordination at the Zn(II) metalloporphyrin centre by the macrocycle pyridyl group.The lower chloride anion binding

Conclusions
In summary, a series of dynamic metalloporphyrin axle containing [2]-and [3]-rotaxanes were prepared via AMT mechanical bond synthesis in high yields, wherein the nature of the Lewis acidic metallocentre dictates the strength of inter-component axle-macrocycle shuttling behaviour and halide anion coordinating capability.
The rotaxanes contained either Zn(II), Ni(II) metallo-, or freebase porphyrin axle and pyridyl functionalised macrocycle components.Variable temperature 1 H NMR and UV-Vis spectroscopic studies demonstrated the [2]rotaxane macrocycle trans-location or shuttling behaviour is principally governed by the Lewis acidity of the porphyrin metal centre.Specifically, the macrocycle pyridyl co-ordination interaction with the Zn(II) centre of the axle porphyrin, in the 'resting state' coconformation.The resting state was disrupted by addition of a competing neutral Lewis base pyridine or by halide anions, both of which stimulate dynamic shuttling behaviour via coordination to Zn(II) and displacement of the macrocycle.
Importantly, Lewis base pyridine or in particular halide binding to the Zn(II) metalloporphyrin centre was accompanied by marked isosbestic changes to the porphyrin Soret absorption band, affording an optical sensory response that signals mechanical bond dynamic shuttling and anion detection.

Figure 1 .
Figure 1.Cartoon representation of competitive neutral Lewis base or anion co-ordination inducing dynamic shuttling behaviour signalled via an optical response in the metalloporphyrin-containing [2]rotaxane.

Figure 3 .
Figure 3. Soret-band absorption spectra of p-6 • Zn and the analogous noninterlocked axle p-10 • Zn in CH 2 Cl 2 in the presence and absence of excess pyridine.
rotaxane p-6 • H 2 was prepared in excellent yield via a demetallation reaction using trifluoroacetic acid.A subsequent re-metallation with Ni(acac) 2 afforded p-6 • Ni in 87 % yield (Scheme 1d).Exchange of the Zn(II) centre of p-6 • Zn to p-6 • Ni or demetallation to p-6 • H 2 , was accompanied by a dramatic sharpening and resolution of the [2]rotaxane 1 H NMR signals.This observation strongly suggests that the [2]rotaxane inter-component macrocycle pyridyl•••Zn(II) metalloporphyrin interaction restricts the macrocycle's shuttling motion across the axle in p-6 • Zn.Quantitative determination of the [2]rotaxane shuttling rates by Eyring analysis (Figure 5) enabled the calculation of ΔH � and ΔS � values shown in Table 1.Inspection of the ΔH � values reveal the shuttling processes are universally endothermic with the largest ΔH � value observed for p-6 • Zn, consistent with the loss of the strongest pyridyl•••Zn(II) metalloporphyrin interaction upon shuttling.Interestingly, the determined ΔH � for m-5 • Zn was substantially less endothermic, approaching the ΔH � of p-6 • Ni and p-6 • H 2, which have no substantial Lewis acidic properties
It was of interest to investigate whether the addition of a competing Lewis base, such as pyridine, would affect the dynamic shuttling behaviour of rotaxane p-6 • Zn.Upon adding increasing amounts of pyridine-d 5 to a CDCl 3 solution of p-6 • Zn the resolution of several broad proton signals associated with the axle and macrocycle-based methylene environments improved and the splitting of these signals was observed (Figure 6).No equivalent behaviour was observed with analogous titration experiments conducted on the non-interlocked axle.These results are consistent with macrocycle shuttling being prohibited by a sterically blocking ligand, leading to axle desymmetrisation.

Figure 7 .
Figure 7. Cartoon depiction of the equilibria in Equation 1.
constant for[3]rotaxane p-8 • Zn compared with p-6 • Zn, may be attributed to the increased effective concentration of the competing pyridyl ligand when two macrocycles are present in the higher order interlocked system.Likewise, this increase in local pyridyl concentration may account for the K a (Cl À ) trend m-5 • Zn > m-7 • Zn.No binding was observed upon the addition of any halide tested to p-6 • H 2 nor to p-6 • Ni, consistent with the lack of Lewis acidic centre in the free base or Ni(II) [2]rotaxanes.The reduction in chloride binding constant observed between the [2]rotaxane and corresponding non-interlocked axles was significantly less for m-5 • Zn and m-9 • Zn than the regioisomeric analogues p-6 • Zn and p-10 • Zn.The relative similarity of the chloride binding constant values between axle isomers m-9 • Zn and p-10 • Zn, strongly suggests the difference is not the result of a changed mode of axle halide binding, nor due to significant changes in sterics upon changing the linker.It is more likely due to the diminished prevalence of the 'resting state' in m-5 • Zn, and hence reduced competition for the vacant axial co-ordination site of the Zn(II) metalloporphyrin.Hence, rotaxane host-Cl-guest binding induces an optical response which is directly correlated to the inter-component interaction afforded by the mechanical bond.

Figure 9 .
Figure 9. a) Cartoon depiction of anion binding mode of rotaxanes, b) Soret-and Q-band region absorption spectrum of a 2 μM acetone solution of p-6 • Zn upon successive addition of a TBACl solution in acetone.Black arrows indicate direction of change.c) Binding isotherms for the data shown in a).

Table 3 .
Host-guest binding constants for 2 μM indicated host and 100 mM TBAX, X=Cl, Br, I solutions in acetone.Deviations in Soret-band absorption upon binding fitted to a 1 : 1 host:guest stoichiometry.Numbers in parentheses represent error in final digit.[a] [a] NB = no binding observed.