High yielding synthesis of 2,2′-bipyridine macrocycles, versatile intermediates in the synthesis of rotaxanes

We present a simple approach to bipyridine macrocycles in remarkable yields (typically >65%) and demonstrate their application in efficient rotaxane synthesis.


Introduction
The synthesis of mechanically interlocked molecules has progressed signicantly since early reports. 1,2 This was largely made possible by the development of passive template methodologies, inspired by Sauvage and co-workers' seminal work, 3 in which non-covalent interactions direct mechanical bond formation. 4 Thus, there exists a range of methodologies for the synthesis of complex interlocked structures in excellent yield in the mechanical bond-forming step.
Despite innovations in the preparation of interlocked molecules, 5 the synthesis of the requisite functionalised macrocyclic components is oen a challenge that can in turn limit the scalability of their synthesis, hindering their investigation for applications such as molecular machines, 6 materials, 7 drug delivery agents, 8 and catalysts. 9 Indeed the synthesis of macrocyclic molecules more generally remains an important area of investigation due to their unusual properties 10 combined with the inherent problem of selecting the desired ring closure event over oligomerisation. 11 A number of strategies have been developed to overcome this including the use of templating interactions to pre-organise the acyclic precursor for ring closure, the preparation of rigid acyclic precursors with restricted rotational degrees of freedom, and expansion of smaller rings. 12 Where this is not possible, the most general approach is to carry out the ring-closing step under high dilution, conditions with signicant consequences for the scalability of the process. 13 The contrast in efficiency between the synthesis of the macrocycle and the formation of the mechanical bond is perhaps nowhere more acute than in our small bipyridine macrocycle modication 14 of Leigh's active template Cumediated alkyne-azide cycloaddition 15 (AT-CuAAC) 16 reaction. 17,18 Although AT-CuAAC reactions with small bipyridine macrocycles have been described as "amongst the highest yielding and readily accessible routes to stable interlocked structures", 4c the synthesis of the macrocyclic bipyridine precursor is extremely inefficient, proceeding in just 4-11% yield. Given the proven utility of the AT-CuAAC reaction for the synthesis of functionalised rotaxanes, 14a,19 including examples with multiple mechanical bonds, 14d,20 stabilised organometallic species, 14b stereochemically complex interlocked molecules, 14c interlocked catalysts, 9g and molecular machines, 6a the limitations placed on the application of this powerful methodology by the poor availability of the most efficacious macrocyclic precursors 19 is a signicant barrier to further developments.
Here we report an operationally simple method for the scalable synthesis of bipyridine macrocycles from readily available substrates that effectively removes this limitation. All of the bipyridine macrocycles reported are efficiently converted to [2]rotaxanes under standard conditions. We demonstrate the long term potential of the AT-CuAAC approach with these now readily available substrates through the gram-scale synthesis of a [2]rotaxane catalyst precursor, and a novel synthesis of Sauvage-type molecular shuttles.

Results and discussion
A new Ni-mediated macrocyclisation Bipyridine macrocycles suitable for AT-CuAAC reactions such as 2a are typically synthesised via a double Williamson cyclisation of a pre-formed bipyridine precursor. 14,16b,21 However, the preferred trans-rotamer of the bipyridine unit 22 disfavours the desired cyclisation, leading to low yields, particularly in the case of smaller macrocycles. 23 To overcome this, we resolved to form the ligand during the macrocyclisation reaction.
Previous routes to bipyridine macrocycles have made use of the Ni-mediated reductive coupling 24 of 2-halo pyridines 25 to produce the bipyridine moiety prior to the macrocyclisation step. 14,16b Although Ni-mediated couplings have not been employed as a macrocyclisation reaction for the preparation of bipyridine macrocycles for incorporation into interlocked architectures, 26 similar reactions have been successful in the key cyclisation step towards cyclic and bicyclic oligoparaphenylenes, 27,28 as well as limited examples of bi-phenyl and bibenzyl natural products. 29 Accordingly, dibromo precursor 1a was synthesised from known intermediates in a succinct manner (see ESI †) and, under high dilution, subjected to conditions based on those developed by the groups of Tiecco and Iyoda for the Ni-mediated intermolecular homo-coupling of 2-halo-pyridines. 25a,c No reaction was observed (Table 1, entry 1).
Repeating the same reaction at higher concentration (entry 2) led to consumption of 1a to give 2a, albeit in low yield. Combining these observations, slow addition of 1a to the preformed Ni 0 species under otherwise identical conditions gave macrocycle 2a in 37% yield (entry 3). A brief screen of reaction conditions revealed that replacing NiCl 2 $6H 2 O with NiBr 2 -(PPh 3 ) 2 resulted in an increased yield of 2a (entry 4), whereas conducting the reaction in THF, commonly employed in such couplings, 25a led to a less efficient reaction (entry 5). Increasing the equivalents of Zn reduced the yield of product, with signicant proto-dehalogenation observed in the crude mixture (entry 6). On the hypothesis that this was due to halogen-zinc exchange competing with the desired Ni-mediated process, Mn was substituted for Zn leading to an improved yield (entry 7). 30 Finally, increasing the equivalents of Mn (entry 8) to improve the efficiency of reductive steps during the catalytic cycle, 24 gave 2a in 70% yield.

Preparative scale synthesis of bipyridine macrocycles
Under our optimised conditions, the preparative scale (2 mmol) reaction of 1a provided macrocycle 2a in an excellent 71% isolated yield (Fig. 1) and so we extended our investigation to a selection of previously reported and novel macrocycles. An ether link proximal to the pyridine unit did not affect the outcome; previously disclosed macrocycles 2b and 2c were produced in excellent yields, demonstrating that, unlike Williamson ether methods, the Ni-mediated cyclisation is relatively insensitive to ring size. Furthermore, when the dichloro analogue of 1b was employed, macrocycle 2b was produced in 56% yield, indicating that 2-chloro pyridines are also viable substrates. The reaction also tolerates heteroatom substitution on the pyridine ring (2d) although in this case an elevated temperature was required due to the lower reactivity of the more electron-rich substrate. 31 Similarly, the reaction tolerates 6-arylsubstitution; sterically hindered macrocycle 2e, a smaller analogue of Sauvage's iconic phenanthroline macrocycle, 3 was produced in excellent 73% yield. We then extended the Ni-mediated coupling to the synthesis of unsymmetrical bipyridine macrocycles, important intermediates in the synthesis of mechanically chiral molecules. 14c Macrocycle 2f, a hybrid of macrocycles 2a and 2b in which the two pyridine moieties are differentiated by a proximal ether unit, was produced in excellent 76% yield. Initial attempts to extend our approach to previously disclosed macrocycle 2g were hampered by the poor solubility of 1g in DMF. This was readily overcome by employing a DMF-THF solvent mixture, leading to 2g in 66% yield, an order of magnitude improvement over the previous synthesis (4%). When the exible alkyl chain of 2g was replaced by a p-xylyl moiety to give 2h, the yield of the macrocyclisation reaction fell to 41%, presumably due to the increased strain of the smaller, more rigid macrocycle. However, the yield of 2h remains signicantly higher than previous syntheses (6%; see ESI †) and demonstrates that the Nicoupling can be used to generate more challenging structures.
Finally, we extended our approach to an aryl-pyridine crosscoupling. Initial attempts with the corresponding dibromo precursor delivered 2i in a low 16% yield. However, replacing the bromo-benzene moiety with an iodo-benzene starting material raised the yield to 24%, suggesting that there is potential to extend our efficient Ni-mediated ring closure to other biaryl macrocyclic motifs in future.
The Ni-mediated synthesis of bipyridine macrocycles is obviously a dramatic improvement over established procedures in terms of simple reaction yield, by as much as an order of magnitude in some cases. Other key benets of this methodology include: (i) the Ni-mediated coupling appears relatively insensitive to substrate structure (ring size, pyridine substituents) and tolerates simple modications (temperature, solvent) where required; (ii) the reaction is capable of producing unsymmetrical bipyridine macrocycles such as 2f-i through a formal intramolecular crosscoupling reaction; (iii) bis-halopyridine precursors 1 are readily available on scale using simple chemistry and commercially available precursors; (iv) isolation of the product is simplied both by the improved impurity prole of the crude product and by the use of pseudo-high dilution conditions. This latter point renders the Ni-mediated macrocyclisation highly scalable; whereas previous syntheses of small bipyridine macrocycles 2a, 2b, 2g and 2h required between 3 and 9 litres of DMF per mmol of product, only $30 ml of DMF is required to produce each mmol of macrocycle using the method presented here. Using James' EMAC measure of reaction efficiency, 13 a logarithmic scale taking into account the yield and reaction concentration, this corresponds to an increase in reaction efficiency from EMAC ¼ 3 to 7. Furthermore, the excellent reaction yield is maintained across a range of scales; repeating the synthesis of 2a (Scheme 1) with 5 mmol of 1a delivered 1.6 g of macrocycle 2a (69% isolated yield).

Bipyridine macrocycles for the scalable synthesis of [2] rotaxanes in excellent yield
Macrocycles 2a-c and 2g are well precedented in the high yielding AT-CuAAC synthesis of interlocked molecules. 9g,14,16b Pleasingly, preliminary investigations conrmed that macrocycles 2d-f and 2h are also excellent rotaxane precursors; under AT-CuAAC conditions with simple alkyne and azide coupling partners, 2d-f and 2h are effectively converted (>90% in all cases) to interlocked products (see ESI †). 32 Furthermore, with signicant quantities of macrocycles 2 in hand we were able for the rst time to demonstrate the scalability of the AT-CuAAC reaction. Previous reports have been limited to small scales (typically 0.025 mmol) by the availability of the macrocycle component. Pleasingly, when the synthesis of interlocked phosphine oxide 5, an advanced precursor of a stimuli-responsive interlocked Au catalyst, 9g was carried out at an 80-fold larger scale using 0.96 g of macrocycle 2a (2 mmol) the isolated yield of the interlocked product was increased relative to our previous report, delivering 1.8 g (95%) of 5 in a single synthetic operation.

An efficient AT-CuAAC approach to Sauvage-type molecular shuttles
Now that bipyridine macrocycles are far more readily available their application in more complex mechanically interlocked molecules and molecular machines is a far more attractive proposition. To capitalise on this, we developed a streamlined synthesis of two Sauvage-type molecular shuttles using macrocycles 2a and 2e and compared their behaviour. Such shuttles are usually based on phenanthroline macrocycles and a thread containing a terpyridine and a bipyridine station, assembled using a passive template threading approach. 3c Replacing the phenanthroline macrocycle with its bipyridine analogue and the terpyridine and bipyridine stations with bistriazolylpyridine (btp) and monotriazolylpyridine (mtp) units respectively, 33,34 greatly facilitates shuttle synthesis using an AT-CuAAC approach. Thus, shuttles 9a and 9e were synthesised in a concise manner from readily available ethynyl pyridine derivatives 6 and 8 through sequential CuAAC couplings of 1,6 bis-azido hexane, delivering 9a and 9e in 73% and 52% isolated yields respectively in the mechanical bond-forming step (Scheme 2). 35 The corresponding non-interlocked thread was also synthesised for comparison. With shuttles 9 in hand, we examined their co-conformational behaviour by 1 H and ROESY NMR (Fig. 2 and ESI respectively †).
Analysis of shuttles 9a and 9e by 1 H NMR (Fig. 2c and e respectively) suggests that in both cases the macrocycle predominantly occupies the tridentate btp station; signicant shis of thread triazole protons H p and H t were observed relative to the non-interlocked thread (Fig. 2d) associate with the bidentate mtp unit remain largely unaffected by the formation of the mechanical bond. ROESY NMR analysis supports this assignment with cross-peaks observed between macrocycle protons and H q , H r and H s of the thread. Given the signicant shis of protons H p and H t and our previous observation of C-H/N hydrogen bonding in AT-CuAAC derived rotaxanes, 14a the localisation of the macrocycle over the btp station in rotaxanes 9 is tentatively attributed to the presence of two C-H/N hydrogen-bonding interactions compared with only one in the case of the mtp station.
Next we investigated shuttles 9 in the presence of diamagnetic metal ions Cu I and Zn II in order to monitor their coconformational behaviour by NMR, as previously reported by Sauvage in the case of phenanthroline derived shuttles. 3c, 36 Addition of [Cu(MeCN) 4 ](PF 6 ) to 9a led to large changes in the 1 H NMR (Fig. 2b) consistent with the expected shuttling of the macrocycle to the bidentate mtp station to form the preferred tetradentate coordination site for Cu I ; protons H c , H d and H e are shielded signicantly and ROESY cross-peaks are observed between mtp triazole proton H i of the thread and alkyl proton H I of the macrocycle. Replacing Cu I with Zn II resulted in a new species consistent with the expected complex in which the macrocycle occupies the btp station, providing a pentadentate binding site for Zn; signals associated with the mtp station return to values similar to that of the thread and cross-peaks are observed between the triazole protons H p and H t of the btp station and H I of the macrocycle. Thus it appears that shuttle 9a behaves as a simple bipyridine-mtp/btp analogue of Sauvage's phenanthroline-bipyridine/terpyridine shuttle. 3c The behaviour of shuttle 9e proved to be more complicated. Surprisingly, addition of Cu I led to the formation of two new species in an approximate 2 : 1 ratio (Fig. 2f). Two-dimensional exchange spectroscopy (2D-EXSY) 37 conrmed that these coconformational isomers are in slow exchange on the NMR timescale with a unimolecular rate constant at room temperature of the order of 10 À3 s À1 corresponding to an activation barrier of $21 kcal mol À1 . The major co-conformation was assigned as that in which the macrocycle, as initially expected, is localised over the bidentate mtp station, based on the shielding of signals corresponding to H e , H d and H c , and ROESY cross-peaks between H D and H E of the macrocycle and H e and H d of the thread.
Similar analysis conrmed that the minor co-conformation is that in which the macrocycle unexpectedly coordinates the Cu I ion at the nominally tridentate btp station, initially suggestive of a ve-coordinate Cu centre. However, Schmittel and co-workers have previously reported the formation of a heteroleptic Cu I complex derived from one tridentate and one bidentate ligand in which the metal ion adopts the expected four-coordinate geometry, with the h donor not involved in binding to the metal ion. 38 Although this phenomenon requires further investigation, we tentatively suggest that similar behaviour may account for the minor co-conformation of [Cu(9a)](PF 6 ), with subtle differences in the secondary interactions (C-H/p, p/p) between macrocycles 2a or 2e and the thread accounting for the differences observed.  4 ](PF 6 ); (g) 9e + Zn(OTf) 2 . Selected signals assigned with labels as in Fig. 1 (macrocycle) and Scheme 2 (thread). Finally, replacing Cu I with Zn II led to a much simpler outcome; a single new species was observed with 1 H and ROESY NMR conrming that, in the case of the Zn II complex [Zn(9e)](OTf) 2 , the macrocycle was located predominantly on the tridentate btp station, as expected (Fig. 2g).

Conclusions
In conclusion we have presented an extremely efficient, scalable and general Ni-mediated method for the synthesis of small bipyridine macrocycles for the AT-CuAAC reaction in high yield from readily available precursors under pseudo high dilution conditions. Although this preliminary study has focussed on macrocycles similar to those used previously in the AT-CuAAC reaction, Ni-mediated couplings are typically tolerant to a wide range of functional groups 24b and investigations are currently underway to determine the wider substrate scope. These now readily available macrocycles are proven versatile intermediates for the synthesis of interlocked molecules in excellent yield using both active 14,16b,18a-e and passive 39 template methods. Given the scalability of both the macrocycle synthesis and the mechanical bond forming step demonstrated here, and the clear potential to extend the approach to more complex molecules, the AT-CuAAC reaction mediated by bipyridine macrocycles clearly has a bright future in the synthesis of interlocked architectures for a variety of applications.