Coordination Chemistry of a Molecular Pentafoil Knot

The binding of Zn(II) cations to a pentafoil (51) knotted ligand allows the synthesis of otherwise inaccessible metalated molecular pentafoil knots via transmetalation, affording the corresponding “first-sphere” coordination Co(II), Ni(II), and Cu(II) pentanuclear knots in good yields (≥85%). Each of the knot complexes was characterized by mass spectrometry, the diamagnetic (zinc) knot complex was characterized by 1H and 13C NMR spectroscopy, and the zinc, cobalt, and nickel pentafoil knots afforded single crystals whose structures were determined by X-ray crystallography. Lehn-type circular helicates generally only form with tris-bipy ligand strands and Fe(II) (and, in some cases, Ni(II) and Zn(II)) salts, so such architectures become accessible for other metal cations only through the use of knotted ligands. The different metalated knots all exhibit “second-sphere” coordination of a single chloride ion within the central cavity of the knot through CH···Cl– hydrogen bonding and electrostatic interactions. The chloride binding affinities were determined in MeCN by isothermal titration calorimetry, and the strength of binding was shown to vary over 3 orders of magnitude for the different metal-ion–knotted-ligand second-sphere coordination complexes.


■ INTRODUCTION
Molecular knots and entanglements occur in DNA, 1 some proteins, 2 and form spontaneously in polymer chains 3 of sufficient length and flexibility. 4 Synthetic routes to several small-molecule knots have been developed, 5,6 or serendipitously discovered, 7 and physical and chemical consequences of knotting have been demonstrated, 4 including anion binding, 7h,8 asymmetric, 9 and allosteric regulation 10 of catalysis, and the securing of a threaded structure by the increase in steric bulk that accompanies tying a knot in a molecular strand. 11 However, despite metal template synthesis being a common route to molecular knots, 4a,c,f,h,5a,c−k,7d,8 effects on metal-ion coordination induced by knotting of the ligand strand have rarely 10,12 been described.
We recently reported that a pentafoil (5 1 ) knot could be prepared by ring-closing olefin metathesis (RCM) of a pentameric circular Fe(II) 5 -helicate. 10 The framework of this 5 1 knot contains only kinetically robust covalent bonds, and so, unlike previous examples, 5e,f the metal ions can be removed from a molecular pentafoil knot without destroying the knot topology. Direct remetalation of the knotted ligand proved only possible with Zn(II) cations, presumably because the coordination dynamics of other metal(II) cations is too slow to allow "mistakes" regarding which bipyridine moieties coordinate to which metal ion to be corrected. However, complexation of the five Zn(II) cations holds the knotted ligand in the conformation necessary to bind to five metal ions. We reasoned that this might enable sequential substitution of other metal(II) cations 12 for Zn(II) without permitting the knotted ligand to adopt conformations that can bind to incoming metal ions through "wrong" combinations of bipyridine units. 13 Here, we show that this enables the efficient synthesis of the otherwise inaccessible Co(II), Ni(II), and Cu(II) metalated knots, enabling the exploration of the coordination chemistry of the knotted ligand. The unknotted ligand monomer only forms circular helicates with Fe(II) and Zn(II), 10 and so the transmetalation strategy, and resulting breadth of coordination chemistry, is only achievable with a knotted ligand.
■ RESULTS AND DISCUSSION Direct Metalation of Knotted Ligand 1. The coordination chemistry of pentafoil knot 1 was investigated by first probing the rate of metalation of 1 with different zinc(II) salts (Scheme 1, steps a,b). The rate and efficiency of introduction of the zinc(II) ions proved to be highly dependent on the salt used (Table 1). Treatment of 1 with Zn(BF 4 ) 2 in MeOH/ CH 2 Cl 2 (most conveniently followed using deuterated solvents) generated [Zn 5 1·Cl](BF 4 ) 9 in 98% yield after 2 h at 40°C followed by work-up with 1 equiv of Bu 4 NCl (  9 quantitatively in less than 5 min at room temperature, followed by exchange of the nine noncavity bound Cl − anions for BF 4 − for solubility reasons (Table 1, entry 3).
The X-ray crystal structure of [Zn 5 1·Cl](PF 6 ) 9 features a chloride anion tightly bound in the central cavity as a consequence of electrostatic interactions and 10 CH···Cl − hydrogen bonds (Figure 2c and f). 10 Given the very different rates of metalation with the different zinc salts, it seems likely that halide binding in the central cavity plays a significant role in facilitating the metalation of the knotted ligand, encouraging rapid rearrangement of wrongly coordinated bipyridine residues. In the 1 H NMR spectrum of the titration of ZnCl 2 into knot 1, free ligand and [Zn 5 1·Cl] 9+ signals dominate. Asymmetric species, that is, partially metalated pentafoil knots [Zn 1−4 1·Cl] 1−7+ , are only present in very minor amounts ( Figure S13). Coordination of the first zinc cations to the knotted ligand appears to preorganize the remaining empty coordination pockets, expediting the subsequent coordination events.
In contrast to the complexation of the knotted ligand with Zn(II), direct metalation of 1 with some other first row transition metal salts (Fe(II), Co(II), Ni(II), or Cu(II)) failed to generate isolable quantities of fully metalated knots, even over extended reaction times (up to 60 h) at elevated temperatures (up to 80°C). 14 Rather than start from metalfree ligand 1, we reasoned that as the Zn(II) cations in [Zn 5 1· Cl](BF 4 ) 9 organize the binding sites of 1 in the correct arrangement, the labile Zn(II) ions of that complex might be exchanged for a less labile metal (introduced in large excess) in a stepwise process. Such stepwise transmetalation could allow the knotted ligand to retain the relative positions of its
1 H NMR ( Figure S14) and ESI−MS ( 1·Cl](BF 4 ) 9−n } n+ followed. No species featuring four or less metal ions were detected. The results are consistent with a stepwise transmetalation process where one zinc cation is substituted by an iron center at a time without the need for major reorganization of the knotted ligand. 15 In analogous fashion, knots [Co 5 1·Cl](BF 4 ) 9 , [Ni 5 1·Cl]-(BF 4 ) 9 , and [Cu 5 1·Cl](BF 4 ) 9 were formed by the transmetalation of [Zn 5 1·Cl](BF 4 ) 9 with the respective metal tetrafluoroborate salts, followed by work-up with 1 equiv of Bu 4 NCl (Scheme 1, steps c,d). Longer reaction times were required for the complete exchange with Zn(II) in the order: Co(BF 4 ) 2 (2 days) < Ni(BF 4 ) 2 ≈ Cu(BF 4 ) 2 (4 days). Reaction progress was monitored by mass spectrometry until no heterometallic species were detected. The paramagnetic nature of Co(II), Ni(II), and Cu(II) precluded obtaining structural information for those pentametallic knots by NMR spectroscopy, but ESI−MS confirmed the isolation of homometallic coordination complexes (e.g., Figure 1  The three crystal structures are broadly similar ( Figure 2) but with slightly different conformations of the knotted ligand to accommodate the different sized metal ions. In each structure, the 190-atom-long pentafoil knotted ligand wraps around each of the metal ions in a closed loop double helicate. Five of the 15 bipyridine groups form an inner cavity lined with 10 electron-poor hydrogen atoms that form an array of hydrogen bonds with the chloride anion located inside the cavity (CH··· Cl − distances given in Table 3). The cavity diameters vary depending on the metal cation: Fe(II), Zn(II), and Ni(II), 3.3(1) Å cavities; Co(II), 3.5(1) Å. The cobalt pentafoil knot dimensions are also more distorted from that of a regular pentagon than the other metalated knots: standard deviations of the metal−metal distances 0.6 Å for the cobalt knot as

Journal of the American Chemical Society
Article compared to 0.1, 0.1, and 0.2 Å in the iron, nickel, and zinc complexes, respectively; standard deviations of the metal− metal−metal angles (from the ideal 108°of a pentagon) 2.2°f or the cobalt knot as compared to 0.7°(iron), 1.5°(nickel), and 1.7°(zinc) pentafoil knots. The chloride anion is located at different distances from the plane of the metal ions in the different knot complexes, illustrating how the size, electrostatics, and shape of the knotted ligand cavity change with coordination to the different metal cations (Figure 2d−f). The packing of the zinc pentafoil knot [Zn 5 1·Cl](BF 4 ) 9 differs from that of the other pentafoil knot coordination complexes. The crystal structure of [Zn 5 1·Cl](BF 4 ) 9 features pairs of knots close-packed through π-stacking of phenyl rings of each ligand (shown in turquoise in Figure 3), with the knots separated by a layer of six tetrafluoroborate anions (shown in space-filling representation in Figure 3). Five of the anions are located between the metal centers, forming an array of CH 1 ···F bonds with the electron-poor CH 1 atoms of the coordinated bipy ligands, with the sixth BF 4 − anion directly between the two chloride anions. The sandwiched complexes stack in the unit cell, forming cationic pillars with anionic cores.
Chloride Binding Affinity of Metalated Pentafoil Knots. Isothermal titration calorimetry (ITC) was used to compare the chloride binding properties of the various metalated pentafoil knots (Table 4). Tetrabutylammonium chloride (Bu 4 NCl) was titrated into solutions of each chloride-free knot, [M 5 1](BF 4 ) 10 , in dry acetonitrile. Each of the different metalated knots was found by ITC to strongly bind chloride ions (Figures S30 and 31). The knot:chloride stoichiometry in each case is 1:1, although the number of  9 . In each structure, carbon atoms are light gray (except for one building block strand in which the C atoms are colored turquoise); N, blue; Co, teal; Ni, dark green; Zn, deep blue; the central chloride ion is shown at 99% van der Waals radius as a green sphere. Other anions, residual solvent molecules, and hydrogen atoms are omitted for clarity.   9 complexes in the X-ray crystal structure, one shown with carbon atoms in gray, one with carbon atoms in red. Anions (chloride and tetrafluoroborate) depicted in space-filling representations. The phenyl rings shown in turquoise are involved in π-stacking between the two Zn 5 -pentafoil knots.

Journal of the American Chemical Society
Article chloride binding sites determined for each knot (N values, Table 4) was slightly lower than 1.0, probably as a result of scavenging of traces of chloride ions from glassware/solvents by the "empty" knots prior to the ITC measurements. In all cases, the chloride binding was both enthalpically and entropically favorable. The chloride affinities were high and similar for the iron(II), cobalt(II), and nickel(II) pentafoil knots (K a (2.3−3.3) × 10 7 M −1 ), an order of magnitude lower for the zinc(II) knot (K a 5 × 10 6 M −1 ), and significantly lower for the copper(II) knot (K a 8 × 10 4 M −1 ). The different Cl − affinities of the various metalated pentafoil knots presumably result from several factors, including (but not restricted to): how coordination to the different size and electronic geometries of the cations affects the conformation of the cavity, the strength of coordination bonds influencing the polarization of the H 1 atoms, the strength of the long-range M(II)···Cl − electrostatic attraction, and the tolerance of distortion of six-coordinate geometries by the different metal ions. In the absence of definitive structural data from X-ray crystallography, we cannot be sure as to the reasons why the Cu(II) pentafoil knot complex is a much more modest binder of chloride ions than the other metalated pentafoil knots. It may be that accommodating the size and shape of the five Cu(II) ions distorts the folded geometry of the pentafoil knotted ligand to such an extent that the cavity is no longer of appropriate size or shape (e.g., no longer directs all of the polarized H 1 protons toward a single point in space) to bind Cl − .

■ CONCLUSIONS
A pentafoil knotted ligand 1, containing 15 bipyridine chelate residues in a 190-atom-long closed loop, forms "first-sphere" pentanuclear coordination complexes with a range of first row transition metal dications. The resulting complexes exhibit strong second-sphere coordination to a single chloride anion. Zn(II) ions, 14 but not Fe(II), Co(II), Ni(II), or Cu(II), can be introduced directly into the knotted ligand, but the Zn(II) can be smoothly transmetalated with other metal(II) tetrafluoroborate salts to the pentanuclear Fe(II), Co(II), Ni(II), and Cu(II) knot complexes. The stepwise metal-ion-for-metal-ion substitution mechanism prevents the ligand from ever having sufficient vacant metal binding sites for incorrect binding modes to be adopted. The knotted ligands wrap around the five metal ions in a circular double helicate motif. Lehn-type circular helicates only form with tris-bipy strands and Fe(II) ions (and, in some cases, Ni(II) 16 and Zn(II) 10 ), and so this motif is only accessible with other metal(II) cations through such a knotted ligand. The X-ray crystal structures of the different pentanuclear knot complexes show conformational changes in the entangled ligand to accommodate the different sized metals. This results in a range of second-sphere coordination binding affinities for chloride ions that varies according to the metal ion over nearly 3 orders of magnitude, from K a = 8 × 10 4 M −1 (Cu 5 1) to K a = 3.3 × 10 7 M −1 (Fe 5 1) in acetonitrile. Knotting ligand strands imparts coordination chemistry that is inaccessible with constitutionally similar but unknotted ligands, a significant example 4a,8−11    Journal of the American Chemical Society Article