[2]Catenane Synthesis via Covalent Templating

Abstract After earlier unsuccessful attempts, this work reports the application of covalent templating for the synthesis of mechanically interlocked molecules (MiMs) bearing no supramolecular recognition sites. Two linear strands were covalently connected in a perpendicular fashion by a central ketal linkage. After subsequent attachment of the first strand to a template via temporary benzylic linkages, the second was linked to the template in a backfolding macrocyclization. The resulting pseudo[1]rotaxane structure was successfully converted to a [2]catenane via a second macrocyclization and cleavage of the ketal and temporary linkages.

from passivem etal templating, Godt et al. developedacarbonate template for the synthesis of [2]catenanesa nd polymeric catenanes. [12][13][14] Hçger et al. successfully modified its terephthalic ester macrocyclization template to obtain MiMs, and our group later investigated this method as well. [15][16][17][18] Other functional groups used for templating include esters [19][20][21][22] and imines. [23,24] For this work however,w ew eres pecially inspiredb yS chill and colleagues, who,b yt he synthesis of ac atenane,p repared aM iM for the first time and pioneered the field of covalently templated MiM synthesis. [25][26][27] In one study (Scheme 1, route A), they took advantage of ad irectingk etal group to join am acrocycle (in red) to ap roperly functionalized linear thread( in blue) in ap erpendicular fashion. [28] The resulting intermediate can adopt two conformations, ap rerotaxane-like one 1,w ith the threadp ositioned within the ring, and at rivial one 1',w hich are in an equilibrium lying far to the left. Upon alkylation with bulky stopper groups, this equilibrium was frozen, and the two distinct compounds could be separated.
Acidic hydrolysis of the ketal moiety then gave am ixture of the separate ring and axle for the major intermediate and a [2]rotaxane for the very minor one (0.08-0.12% yield). Despite the poor yields, it wasn ot only provent hat the prerotaxane conformation is possible, but also that it can in principle lead to an interlocked species.
We recentlyd evised as omewhat similar approacht oM iMs termed" templated backfolding" (Scheme 1, route B). Starting from two linear strands joined together at the centerb ya ketal, as uitable template is connected via temporary linkages (dashed lines) tog ive 6.N ext the first macrocycle is formed by linkingt he template to its opposings trand (in red) in a" backfolding" fashion. Thanks to the covalentt emporary linkages, the unfavored pseudorotaxane conformation of 1 is the only one available to 7.
From 7,asecond macrocycle (in blue) is formed,a nd final cleavageo ft he temporary linkages and the ketal selectively results in a [ 2]catenane.T he efficacy of the backfolding approach was proven by the synthesis of both aq uasi [1]rotaxane and a quasi [1]catenane, featuring irreversible bonds between the axle and ring fragments. [29] The next step was to make the bond to the central quaternary spiro-carbon reversible, by introducing ak etal group. With this purpose,t arget [2]catenane 10 a (Scheme 2) was addressed.
By followingt he backfolding templated strategy as outlined in Scheme 1B andu sing the powerful Cu I -catalyzed azidealkynec ycloaddition (CuAAC) andr ing-closing metathesis (RCM)a st he key macrocyclization steps, precatenane 9a was successfully obtained. However,t oo ur surprise but even more disappointment, all attempts at hydrolyzingt he acid labile ketal group failed. [30] Initiallyt his was attributed to steric shieldingw ithin the very congestedp recatenane architecture. In order to gather experimentale videnceo ft his, model compound 11 was synthesized,which closely matches the electronic environment of precatenane 9a (Scheme 3). Treatment of ketal 11 with concentrated aqueous HCl in MeOH at room temperature showede ven after 28 ho nly trace amounts of the hydrolysis products.H ydrolysis of ketal 11 could only be accomplished after stirring in concentrated HCl and MeOH at 50 8Cf or severalh ours. This indicates that besides the catenane effect that clearly plays ar ole in the remarkable stability of ketal 9a,o ther factorsp lay ar ole. In contrast, ketal 12,a n early intermediate in synthesis of 9a,c ould be hydrolyzed at room temperature in only 4hunder otherwise identical conditions. The positive influence of amide groupso nt he stability of nearby ketals has been reported before [31] and this observation inspired us to pursuet he synthesis of 10 b,a na nalog of catenane 10 a in which the amides have been replaced by ethylidene groups.
For this, ketone 17 was prepared first from known alcohol 15 [32] in three steps (Scheme 4). The alkene groups were converted to alkynes via ab romination-elimination protocol, which proceeds in good selectivity under anhydrous conditions.
This was followed by oxidation of the alcohol group by PCC. As already found in our previouss tudies, direct coupling of ketones such as 17 and (+ +)-dimethyl tartratep roceeds sluggishly. This could be overcome by transformationo ft he ketone to the dimethoxy ketal, followed by transketalization with (+ +)-dimethyl tartrate to give functionalized ketal 18 in 58 %y ield. Althoughm ore viable,t his procedure suffers from incomplete conversion,l ikely due to the difficulties in completely removing water from the reactants, but nonetheless allows for recovery of ketone 17 in high yield (90 %b rsm). Assembly of the second macrocycle follows, by saponification of the methyl esters and couplingo ft he resulting diacid with amine 19, which also includes the temporary linkages. At this stage, transesterification of template 21 with the two phenolic groups in 20 affords the backfolding macrocyclization precursor 22 b.T oi nitiate this key step using the CuAAC reaction, 22 b was stirred at high dilution with Cu(MeCN) 4 BF 4 as catalyst and TBTAa sl igand. Refluxing in CH 2 Cl 2 for 3days gave cage compound 23 b in 24 %y ield. Compared to 23 a, [30] 23 b is formed in lower yield and showed am uch simpler 1 H-NMR spectrum,w ith only minor splittingo ft he signals corresponding to chemically equivalent protons. These surprising differences suggest the presence of substantial interactions between the macrocycle amides and the tartrate core in 23 a,r esulting in rigidificationo ft he macrocycle (in red) and slowing down conformational exchange.
With the ketal and two macrocycles in place, the two terminal alkenes are positioned such as to allow the second macrocyclization to give the precatenane skeleton. This is done via RCM, performed in CH 2 Cl 2 at 40 8Cu sing Grubbs' 2 nd generation catalyst (Scheme 5). The resulting product, obtained in 57 %y ield as an inseparable mixture of Ea nd Zi somers, is then converted to as ingle compound 24 b-H 2 after saturation of the double bonds by catalytic hydrogenation. To liberate the [2]catenane, first the temporary linkages were broken via solvolytic transesterification of the lactone groups, followed by protolytic cleavage of the benzylic tertiary amides using TFAi nt he presenceo fE t 3 SiH as cation scavenger.F inally, the ketal core was hydrolyzed under strongly acidic conditions, liberating [2]catenane 10 b in 60 %y ield.
To further assess the influence of the catenane effect on the stabilityo ft he ketal group in precatenanes 9a and 9b,w es et out to synthesize their respective regular spiro topoisomers and subjected them to ketal hydrolysis conditions (Scheme 6). This was done by simply alteringt he order of the final steps that were used for the syntheses of precatenanes 9a and 9b. Thus starting from prerotaxane 23 a,i nstead of first carrying out the RCM macrocyclization reaction, the sequence of reactions started by solvolysis of the temporary ester linkages in 23 a to give macrocycle 25 a' as an intermediate, followed by a concomitant spontaneous unwinding process, yielding 25 a. Subsequents ubjection of 25 a to RCM, catalytic hydrogenation conditions andf inal protolytic removal of the benzylic appendages gave ac ompound with different spectralp roperties from precatenane 9a,s upporting the proposed regular spiro topology of 27 a thus obtained.
As shown in Figure 1, ac omparison of the 1 H-NMRs pectra of 27 a and 9a shows marked differencesf or severalkey peaks. These werea ssigned for 27 a based on integral and multiplicity,a nd their identity was confirmed by COSY and HSQC NMR (see supporting information). Most striking, is the splitting of the diastereotopic benzylic protons (d), which is significantly increased in precatenane 9a.S imilarly,t he aliphatic signals between 1.5 and 3ppm as well as the amide NH signals display much more complex patterns in 9a.T his is likelyt he result of slower conformational movements in the sterically more congested precatenane macrocycles. [33] In addition, the aromatic template signal (c)i ss hifted downfield by 0.34 ppm, indicating additional shielding in 27 a due to ring-current effects arising from the triazole moieties. [34] This interaction is reduced in 9a as the intra-annular fragment limits the ability of the ketone macrocycle (in red) to fold. These findings are consistent with observations made on similari nverted spiro architectures previously synthesized via our backfolding approach. [29,35] Remarkably,s olvolysis of the ester linkages in 23 b gave intermediate 25 b' as as table compound at room temperature, which required extensive heating to unwind to the thermodynamically favored conformer 25 b.M olecular mechanicsm odeling of structures 25 b' and 25 a' suggest that the benzylic groups (in black) could hamper this process, as they barely fit in the macrocycle cavity according to space filling representations (see supporting information). However,s of ar we have no sound explanation for the exceptional differences in the kinetics of this step between 25 a' and 25 b'.From unwound macrocycle 25 b,t he trivial spiro bismacrocycle 26 b waso btained after RCM and catalytic hydrogenation.
Similarly to what was observed for precatenanes 9a and 9b, the final hydrolysis of the linking ketals in the regulars piro compounds 26 b and 27 a gave ac ompletely different outcome (Scheme 6). Startingf rom 26 a,t he benzylic groups could be selectively cleaved in TFAi nt he presence of Et 3 SiH, giving 27 a,w hichf ailed to hydrolyze to the desired separate macrocycles even under forcing conditions (not shown). In stark contrast, 26 b partially split into the separatem acrocycles 29 and 28 b already by TFAa ddition. This processw as broughtt oc ompletionb yt reatment with HCl in MeOH/H 2 Oa t room temperature. The relatively mild conditions required underscore that the high stabilityo ft he ketal in precatenanes 9a,b emergest oagreat extent from the catenanee ffect. Due to the presence of the additional ketal stabilizing amide groups in 9a,i nt his case ketal cleavage is completely blocked.
Analysis by 1 H-NMR revealed marked differences between catenane 10 b and a1 :1 mixture of its separatea nd non-interlockedc omponents 28 b and 29,a ss hown in Figure 2. The benzylic protons (d)a ppear equivalent in 28 b,a nd become diastereotopic in 10 b.I na ddition, the overall chemical shifto f the aliphatic proton signals is significantly reduced. This points to increased shielding and is commonf or methylene chains entrappedw ithin macrocycles bearinga romatic rings. [33,36,37] However, the mosts triking differencei st he downfield shift of 0.56 and 0.11ppm, respectively,o ft he template aromatic (c) and methyl ester (g)p rotons. As was observed for precatenane 9a and trivial bicycle 27 a,t his is probablyd ue to the rigid conformation of the interlocked rings in [2]catenane 10 b preventingt he macrocyclesf rom collapsing. Furthermore, the NMR spectrum of catenane 10 b shows that the signals of both the terephthalate protons (c)a nd triazole protons (a)a ppear as double singlets ( Figure 3). This points to the presence of two diastereomeric forms of catenane 10 b,l ikely resulting  precatenane 9a [30] (II) in CDCl 3 .For proton assignments, see Scheme6. from ac ombination of central chirality,e merging from the tartrate moiety,a nd planar chirality of the red macrocycle. The AB system displayed by the benzylic protons (d)s uggests in fact that rotation of the p-cyclophane-type terephthalate moietyi s hindered,r esulting in two different conformationso ft he macrocycle, whichare mirror images of each other (Figure 3, III).
Conversely,t he non-interlocked macrocycle 28 b,w hich lacks the steric constraints of the mechanical bond, displays as harp singlet signal for the benzylic (d)p rotons, indicatingt hat the terephthalic moiety can rotate freely.S of ar,a ll attempts to separatet he eventual diastereomers of 10 b by severals ymmetric and asymmetric HPLC methods failed.
In conclusion, despite the strong stabilizing catenane effect experienced by the pivotal ketal link, smalls tructuralc hanges allowed its hydrolysis liberating the mechanicallyi nterlocked [2]catenane. As ac onsequence,t he covalent template backfolding approach was, for the first time, successfully employed in the synthesis of am echanically interlocked molecule. Currently,w ea re focusing on reducing the footprint of the template andf urther improving the versatility of the temporary linkagest oe xpand the structurald iversity within the class of mechanically interlocked molecules.