Integrative self-sorting is a programming languagefor high level self-assemblyWei Jiang and Christoph A. Schalley1

Institut fur Chemie und Biochemie, Freie Universitat Berlin, Takustrasse 3, 14195 Berlin, Germany

Edited by Julius Rebek Jr., The Scripps Research Institute, La Jolla, CA, and approved January 1, 2009 (received for review September 24, 2008)

Starting from the basis of a simple 4-component self-sortingsystem of crown ethers and ammonium ions, we design 6 buildingblocks in which 2 identical or different binding sites are incorpo-rated. These building blocks can be mixed in many different waysto yield quite distinctly different pseudorotaxane assemblies. Theself-sorting process integrates all building blocks in specific placesso that this approach permits us to exert positional control and canwidely influence the resulting assemblies with respect to thedetails of their structures. At maximum, we report quadruplyinterlocked species with up to 5 subunits that form specific assem-blies. Although NMR methods are limited to the analysis of simplercomplexes, ESI-MS and, in particular, tandem mass spectrometry ishighly useful to analyze the assemblies’ connectivities.

interlocked molecules � pseudorotaxanes � supramolecular chemistry �tandem mass spectrometry � crown ether

The ‘‘bottom-up’’ approach to the construction of moleculardevices and machines through molecular self-assembly has

attracted extensive attention in the past 2 decades (1–4). Mo-lecular self-assembly (5–7) harnesses noncovalent interactions toorganize discrete components into well-organized architectures.To achieve this, the necessary instructions must be written intothe structures of the building blocks by chemical synthesis. Mostartificial self-assembled architectures known today incorporatemultiple copies of the same building blocks and thus are oftenhighly symmetrical structures.

In the macroscopic world, however, devices and machines areassembled from many different components cooperating witheach other to perform complex functions. The same holds truefor most biomolecules. Nature efficiently uses the principles ofnoncovalent self-assembly together with self-sorting (8, 9) phe-nomena to generate complex, functional architectures. DNAmay serve as a prominent example whose function is informationstorage. Four bases (adenine (A), thymine (T), guanine (G), andcytosine (C)) self-sort to form base pairs (AT and GC) in theirmixture. Integrating these bases to the polymer backbones ofDNA in a specific sequence not only stores information but alsoallows the cell to replicate and transcribe the stored information.

When complex functions are to be achieved in artificialmolecular devices or machines, the self-assembly of highlysymmetrical architectures from multiple identical buildingblocks is a severe restriction that can be overcome by applyingthe same strategy used by nature: The integration of differentsubunits into the architecture with precise positional control bycombining the principles of self-assembly and self-sorting. Wevery recently coined this strategy, integrative self-sorting (10),and are convinced that it will not only enrich the world ofsupramolecular architectures from the point of view of structuralvariability but will also form the basis for the implementation ofhigh-level functions in future assemblies.

Herein, we present several strategies about how to writeinstructions into molecular components for constructing com-plex pseudorotaxane assemblies by using the programming lan-guage of integrative self-sorting. Starting from a very simple4-component self-sorting system consisting of 2 crown ethers and2 ammonium ions, pseudorotaxane assemblies can be generated

by merging 2 of these components. These pseudorotaxanes areup to quadruply interlocked and combine up to 5 building blockswith high spatial control in 1 assembly. Although NMR exper-iments can, in principle, be used for the characterization of theassemblies up to a certain level of complexity, tandem massspectrometry turns out to be particularly useful to analyze theassemblies’ connectivities.

Results and DiscussionDesign Concept. Conceptually, interesting integrative self-sortingsystems can be derived from a very basic simple self-sortingsystem containing several discrete complexes by merging 2 of thecomponents into 1. Recently, we reported (10) a 4-componentself-sorting system (Fig. 1) that owes its high fidelity to thefollowing facts: (i) benzo-21-crown-7 (C7) is not able to pass overa phenyl group under the conditions of the experiment. Thus, theformation of a pseudorotaxane with 1-H�PF6 is kinetically hin-dered; (ii) the complexation of 2-H�PF6 with C7 is thermody-namically more stable than that with dibenzo-24-crown-8 (C8);(iii) C8 thermodynamically prefers 1-H�PF6 over 2-H�PF6.

From this simple 4-component self-sorting system, numerousintegrative self-sorting systems can be constructed by suitableprogramming. The key is the design of appropriate integrativecomponents that can bring together all of the other subunits into1 product complex with positional control. The components inour basic self-sorting system have a specific dimensionality. Forexample, the ammonium salts are thread-like molecules that canbe considered as ‘‘1-dimensional’’ structures—the axles of thepseudorotaxanes. The 2 crown ethers can simply be regarded as‘‘2-dimensional’’ structures. They can slip onto an ammoniumion guest as the pseudorotaxane wheel. Many different optionsfor the construction of more complex architectures arise: (i) Wecan connect the 2 guests ‘‘linearly’’ (11–14). (ii) The 2 crownether hosts can be incorporated into dimers and—if connectedthrough the anthracene core chosen here—are coplanar witheach other. (iii) One of the 2 guests can be cross-connected withthe other host (15–20). (iv) Several reasonable combinations andrepetitions of the first 3 strategies can be realized (21–23). In thisarticle, we will focus on strategies i, ii, and iv. The molecularstructures of the building blocks used throughout this article areshown in Fig. 2. They comprise 3 ditopic diammonium axles(3-2H�2PF6, 4-2H�2PF6, 5-2H�2PF6) with spacers of differentlengths between the 2 binding sites A and B, an anthracene-spacered crown ether heterodimer (6) bearing 1 C8 and 1 C7motif and 2 anthracene-spacered crown ether homodimers 7 and8. The syntheses and full analytical data for all these compoundsare provided in the supporting information (SI) Appendix.

Author contributions: W.J. and C.A.S. designed research; W.J. performed research; W.J.analyzed data; and W.J. and C.A.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

1To whom correspondence should be addressed. E-mail: schalley@chemie.fu-berlin.de.

This article contains supporting information online at www.pnas.org/cgi/content/full/0809512106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0809512106 PNAS � June 30, 2009 � vol. 106 � no. 26 � 10425–10429

CHEM

ISTR

YSP

ECIA

LFE

ATU

RE

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

Merging 2 Monotopic Axles into Ditopic Ones. Ditopic axle3-2H�2PF6 inherits the binding motifs from 1-H�PF6 and 2-H�PF6without interference between them. The sequence of these 2motifs in this guest determines the sequence of C7 and C8 in thehetero[3]pseudorotaxane 9-2H�2PF6 (Fig. 3) formed upon mix-ing all 3 components in equimolar amounts. As shown earlier(10), the sequence of crown ethers in 9-2H�2PF6 is fixed becauseof the presence of the bulky anthracenyl stopper and the centralphenyl spacer incorporated in the axle. They admit only C8 tobinding site A, whereas only site B remains for C7. In noncom-petitive solvents such as dichloromethane (DCM), mixingequimolar amounts of C7 and C8 with 3-2H�2PF6 indeed resultsin the virtually quantitative formation of the sequence-definedhetero[3]pseudorotaxane 9-2H�2PF6. In 9-2H�2PF6, the integra-tive component 3-2H�2PF6 thus integrates 2 distinct hosts thatmay finally be equipped with different functional groups thatprovide functionality.

As a variation of the theme, we synthesized 2 similar guests4-2H�2PF6 and 5-2H�2PF6 with spacers between 2 binding siteslonger than that in 3-2H�2PF6. The 1H NMR spectrum (Fig. 4,line b) of the equimolar mixture of 4-2H�2PF6, C7, and C8 showsimilar chemical shifts for Ha, Hb, He, Hf, and Hg as thoseobserved for 9-2H�2PF6. Consequently, site B is occupied by C7as clearly indicated by the shift of He to the same positionobserved in Fig. 4, line a. Site A is bound to C8, as characterizedby downfield shifts of Ha and Hb identical to those observed inFig. 4, line c. Therefore, hetero[3]pseudorotaxane 10-2H�2PF6 isthe predominant species in this equimolar mixture of 4-2H�2PF6,C7, and C8.

The conclusion from 1H NMR spectroscopic data were easilyconsolidated by electrospray-ionization mass spectrometry (ESI-MS) results obtained from the equimolar solution of 4-2H�2PF6,C7, and C8 in DCM. The MS results are quite similar to thoseobserved for 9-2H�2PF6 (10). A very clean mass spectrum withonly 1 intense peak for [10-2H�PF6]� at m/z 1,500 is obtained.The following infrared-multiphoton dissociation (IRMPD) ex-periments (MS/MS; Fig. 5) (24, 25) performed with mass-selected [10-2H�PF6]� show the loss of C8 merely to occur at

higher laser intensities as a consecutive fragment after losing C7and HPF6. This experiment confirms the sequence of crownethers: C8 cannot leave without losing C7 first.

Fig. 1. A 4-component self-sorting system constructed from an equimolarmixture of 1-H�PF6, 2-H�PF6, C8, and C7.

Fig. 2. Structures of building blocks 3–2H�2PF6, 4-2H�2PF6, 5-2H�2PF6, 6, 7,and 8.

Fig. 3. Molecular structures of integrative self-sorting complexes involved inthis research.

Fig. 4. Partial 1H NMR spectra (500 MHz, 298 K, CDCl3:CD3CN � 2:1, 10.0 mM)of equimolar mixtures of 4-2H�2PF6 and C7 (line a); 4-2H�2PF6 and C8 (line c);and 4-2H�2PF6, C7, and C8 (line b). Complexed and uncomplexed species aredenoted by ‘‘c’’ and ‘‘uc’’ in parentheses, respectively. Asterisks indicatesolvent impurities.

10426 � www.pnas.org�cgi�doi�10.1073�pnas.0809512106 Jiang and Schalley

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

The predominance of 11-2H�2PF6 in the equimolar solution of5-2H�2PF6, C7, and C8 can also unambiguously be demonstratedby analogous 1H NMR and MS experiments (see Figs. S1 and S2in SI Appendix. Therefore, the elongation of the spacers between2 binding sites in the integrative components 4-2H�2PF6 and5-2H�2PF6 does not affect the fidelity of the resulting integrativeself-sorting systems 10-2H�2PF6 and 11-2H�2PF6. Thus, thedistance between C7 and C8 in this kind of integrative self-sorting system can be adjusted according to requirements.

Merging 2 Monotopic Wheels into a Heteroditopic One. As anexample for the second strategy, 2 hosts, C7 and C8, can bemerged together into 1 integrative component. For this purpose,heteroditopic host 6 with one 21-crown-7 motif and one 24-crown-8 motif was synthesized. Based on the results above, thepseudorotaxane 12-2H�2PF6 is expected to prevail in an equimo-lar solution of 1-H�PF6, 2-H�PF6, and 6. The first piece ofevidence for self-sorting in this mixture came from 1H NMRexperiments (Fig. 6). The Ha and Hb protons on 1-H�PF6

experience significant downfield shifts with respect to free1-H�PF6. The shift differences are similar to those observed for[1-H@C8]�PF6, 9-2H�2PF6, 10-2H�2PF6, and 11-2H�2PF6 and

identical to those monitored in the 1:1 mixture of 1-H�PF6 and6. The only reasonable explanation for this result is that axle1-H�PF6 is bound in the cavity of 24-crown-8 section in 6 because21-crown-7 unit lacks the ability to pass over phenyl group underthe experimental conditions. Thus, the 21-crown-7 part is left forguest 2-H�PF6. The Hd protons in 2-H�PF6 are shifted uponbinding, as expected. Furthermore, the resonances for He, Hg,and Hf are broadened, likely because of the existence of 2isomers (‘‘parallel,’’ 12p-2H�2PF6 and ‘‘anti-parallel,’’ 12ap-2H�2PF6 in Fig. 3) for 12-2H�2PF6. Consequently, the NMRexperiment demonstrates the 2 isomers of 12-2H�2PF6 to dom-inate in the mixture—even though the evidence does not seemto be as clear-cut as at first glance.

ESI-MS experiments with the equimolar solution of 1-H�PF6,2-H�PF6, and 6 in DCM support this conclusion. The ESI massspectrum (Fig. 7) shows only 1 intense peak at m/z 728, corre-sponding to [12-2H]2�. No 2:1 complexes of 1-H�PF6 and 6, or2-H�PF6 and 6, are observed. The tandem MS experiment onmass-selected [12-2H]2� remarkably shows the phenyl-stoppered axle [1-H]� to dethread first from the parent ion. Onlyat higher laser intensity, the [2-H@6]� fragment starts todecompose further into: (i) neutral 6 and [2-H]�; (ii) neutral 2and [6�H]�. If one considers the narrower axle 2-H�PF6 to bindto the wider crown ether (24-crown-8), whereas the thicker axle1-H�PF6 is bound in the narrower crown ether (21-crown-7), onewould certainly predict [2-H]� to be lost first from the corre-sponding parent ion. Because the opposite is observed, we cansafely rule out this isomer. The only alternative 12-2H�2PF6 musttherefore be formed. Consequently, the MS results provideunambiguous evidence for the 2 axles being threaded throughthe appropriate crown ethers as expected from the resultsdiscussed before.

Combination/Repetition of Integrative Self-Sorting Strategies. Toconstruct even more complex assemblies, the strategies dis-cussed above can be combined. In the following, we construct themore complex integrative self-sorting complexes 13-4H�4PF6,14-4H�4PF6, 15-4H�4PF6, and 16-4H�4PF6 (Fig. 3). Connectingseveral of the integrative components in 1 assembly increases thenumber and variety of components incorporated in the finalcomplex, and we therefore anticipate increasing difficulties tocharacterize these architectures in detail. Indeed, the 1H NMRspectra (see Figs. S3–S6 in SI Appendix) are often broadened andcomplicated, so that a detailed analysis of the resulting assem-blies is not straightforward anymore. However, MS and MS/MSexperiments are a powerful tool to study complex systems andelucidate the complexes structures (24, 25). Therefore, we willfocus on the MS and MS/MS results of integrative self-sortingcomplexes 13-4H�4PF6, 14-4H�4PF6, 15-4H�4PF6, and 16-4H�4PF6 in the following discussion.

Fig. 5. Mass spectra of 10-2H�2PF6. (Upper) ESI-FTICR mass spectrum of a 1:1:1DCM solution of 4-2H�2PF6, C7, and C8. (Lower) Infrared-multiphoton disso-ciation (IRMPD) experiments (MS/MS) of mass-selected [10-2H�PF6]� at differ-ent laser intensities.

Fig. 6. Partial 1H NMR spectra (500 MHz, 298 K, CDCl3:CD3CN � 2:1, 10.0 mM)of equimolar mixtures of 1-H�PF6 and 6 (line a); 2-H�PF6 and 6 (line c); and1-H�PF6, 2-H�PF6, and 6 (line b). Complexed and uncomplexed species aredenoted by ‘‘c’’ and ‘‘uc’’ in parentheses, respectively. Asterisks indicatesolvent impurities.

Fig. 7. Mass spectra of 12-2H�2PF6. (Upper) ESI-FTICR mass spectrum of a 1:1:1DCM solution of 1-H�PF6, 2-H�PF6, and 6. (Lower) IRMPD experiments (MS/MS)of mass-selected [12-2H]2�.

Jiang and Schalley PNAS � June 30, 2009 � vol. 106 � no. 26 � 10427

CHEM

ISTR

YSP

ECIA

LFE

ATU

RE

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

In the mass spectrum (Fig. 8) obtained from a 1:2:2 DCMsolution of 7, C8, and 3-2H�2PF6, [13-4H�2PF6]2� was identifiedby the corresponding signal at m/z 1417. In the structure of13-4H�4PF6, 5 components are assembled together through weakinteractions, so that even a soft ionization method such as ESImay destroy this assembly to some extent during ionization.Indeed, we observed 2 fragments of [13-4H�2PF6]2� at m/z 1,007and 1,827, which also appear in the tandem MS spectra. How-ever, at m/z 1,456 an additional complex of 2 C8 on 3-2H�2PF6is observed that indicates incomplete self-sorting. As for thesimpler cases above, tandem mass spectrometry is capable ofrevealing the details of the location of the 5 components in13-4H�4PF6 (Fig. 8). From both, the parallel and antiparallelisomers 13p-4H�4PF6 and 13ap-4H�4PF6 in Fig. 3, one wouldexpect that the lowest-energy dissociation pathway is loss of anaxle/C8 complex. Losses of just a neutral C8 should instead notbe possible without inferring complex rearrangements that areunlikely to compete. The fragments at m/z 1,007 and 1,827already detected immediately after isolation of [13-4H�2PF6]2�

correspond to the expected charge-separating fragmentationinto [3-2H@C8�PF6]� and [3-2H@(7�C8)�PF6]�, whereas, in-deed, no C8 loss was observed. Consecutive fragmentationreactions are indicated in the lower spectrum in Fig. 8.

Similarly, we can provide evidence for the formation of14-4H�4PF6 in a 1:2:2 DCM solution of 8, C7, and 3-2H�2PF (seeFig. S4 in SI Appendix). From the structures of both isomers14p-4H�4PF6 and 14ap-4H�4PF6 (Fig. 3), one would predict thatnow losses of 1 or even 2 C7 molecules should correspond to thelowest-energy fragmentation reactions. The observed sequentialloss of first 2 C7, then of 2 HPF6, and only finally of 1 of the axles[3-H]� from [14-4H�2PF6]2� is perfectly in line with the pro-posed structure.

As discussed, parallel and antiparallel isomers exists for13-4H�4PF6 and 14-4H�4PF6 and it would certainly be moresatisfying to have control even on the alignment of the axlesrelative to each other not only on the connectivities of thebuilding blocks. The problem is the 1-point connection betweenall subunits. Applying 2-point connections (26) as in 15-4H�4PF6and 16-4H�4PF6 will not only provide positional control, but alsostabilize the assemblies so that we can expect them to formexclusively in contrast to 13-4H�4PF6 and 14-4H�4PF6.

From 3-2H�2PF6 and 6, 15-4H�4PF6 is expected to be exclu-sively formed, and indeed, a single peak at m/z 1,522, corre-sponding to [15-4H�2PF6]2�, is observed in the ESI mass spec-trum obtained from this sample (Fig. 9). The absence of peaksfor self-assembled oligomers indicates the closed structure of15-4H�4PF6 to be formed rather than an open-chain analogue.As expected from the structure of 15-4H�4PF6, the MS/MSspectra (Fig. 9) of mass-selected [15-4H�2PF6]2� show the dica-tion to fragment symmetrically into 2 monocations

[3-2H@6�PF6]� at m/z 1,522, in line with the symmetry in thedication’s structure. Because the dicationic parent ion and thesingly charged fragment both appear at the same m/z, only acareful isotope pattern analysis reveals this process (Insets in Fig.9). Although the isotope pattern of the mass-selected parentclosely matches that calculated for [15-4H�2PF6]2�, the primaryproduct appears with a peak spacing of 1 amu indicating amonocation. In the structure of 15-4H�4PF6, full control over theantiparallel orientation of both the axles as well as the crownether dimers is thus achieved.

From the set of our building blocks, a parallel arrangement ofthe 2 axles can also be generated by using 2 homoditopic hosts7 and 8 that replace 6 in 15-4H�4PF6 to yield 16-4H�4PF6. Again,a quite clean mass spectrum with only 1 intense peak at m/z 1,522assigned to [16-4H�2PF6]2� is obtained from the 1:1:2 DCMsolution of 7, 8, and 3-2H�2PF6, which also suggests the predom-inance of 16-4H�4PF6 in this solution. No oligomer peak wasobserved, indicating the closed structure of 16-4H�4PF6. Thedifferent arrangement of building blocks, however, changes thefragmentation pattern significantly: Two fragmentation path-ways are observed for mass-selected [16-4H�2PF6]2�: (i) losingneutral 21-crown-7 dimer 7; (ii) breaking 1 macrocycle of24-crown-8 in 8 and distributing 2 charges to 2 fragments. Thefirst one is clearly within expectation and indicates 7 to belocated near the open end of both axles.

To further confirm the closed structures of 15-4H�4PF6 and16-4H�4PF6 and to strengthen our line of reasoning, each of themwas mixed with 2 equivalents of 9-2H�2PF6. If either one of the2 assemblies were an open chain with unoccupied binding sitesat its 2 ends, one should see a partial transfer of C7, C8, or theaxle from 9-2H�2PF6 to 15-4H�4PF6 or 16-4H�4PF6. This is,however, not the case (Fig. S10 in SI Appendix). Assembly9-2H�2PF6 does not interfere with 15-4H�4PF6 or 16-4H�4PF6 atall, providing evidence for them to be closed structures.

ConclusionsConceptually derived from a very simple 4-component self-sorting system, 6 ditopic key compounds have been developedthat can be used to integrate several different building blocksinto more complex assemblies with high positional control. Upto 5 components were assembled into up to quadruply inter-locked pseudorotaxane assemblies. Different strategies weredeveloped and evidence for the successful assembly of thedesired structures presented. NMR experiments provide astraightforward structure analysis for the less complex assem-blies but are no great help when complexity increases. Instead,ESI mass spectrometry and, in particular, tandem mass spec-trometric experiments were capable of unraveling the details of

Fig. 8. Mass spectra of 13-4H�4PF6. (Upper) ESI-FTICR mass spectrum of a 1:2:2DCM solution of 7, C8, and 3-2H�2PF6. (Lower) IRMPD experiments (MS/MS) ofmass-selected [13-4H�2PF6]2�. Fig. 9. Mass spectra of 15-4H�4PF6. (Upper) ESI-FTICR mass spectrum of a 1:1

DCM solution of 3-2H�2PF6 and 6. (Lower) IRMPD experiments (MS/MS) ofmass-selected [15-4H�2PF6]2�.

10428 � www.pnas.org�cgi�doi�10.1073�pnas.0809512106 Jiang and Schalley

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

the structures even for the most complex assemblies presentedhere. Integrative self-sorting thus turns out to be a highly capableprogramming language. Together with self-assembly, it is apowerful strategy to assemble even highly unsymmetrical su-pramolecular architectures without repetitively using multiplecopies of the same subunits. Provided these building blocks are,in future, equipped with suitable functional groups, the assemblyof supramolecules with higher-level functions should becomemore easily designable. The concept of integrative self-sorting iscertainly not limited to the systems based on weak interactionsbut also can, for example, be extended to self-assembly based onmetal coordination (27) or reversible covalent bonds. It is thusa general means to create molecular architectures with a higherdegree of complexity than most that are known so far. Theconcept of integrative self-sorting describes the borderline be-tween self-assembly and self-sorting, where the self-sortingphenomena are used to program the self-assembly process.

Materials and MethodsElectrospray-Ionization Fourier-Transform Ion-Cyclotron-Resonance (ESI-FTICR)MS. Because the tandem MS experiments are most important to the presentarticle, we describe them here. All other experimental details can be found in

SI Appendix. The ESI-FTICR MS experiments were performed with a Varian/IonSpec QFT-7 FTICR mass spectrometer equipped with a superconducting7-Tesla magnet and a micromass Z-spray ESI ion source using a stainless steelcapillary with a 0.65-mm inner diameter. The solutions of samples wereintroduced into the source with a syringe pump (Harvard Apparatus) at a flowrate of �2.0 �L�min�1. Parameters were adjusted as follows: source temper-ature, 40 °C; temperature of desolvation gas, 40 °C; parameters for capillaryvoltage, extractor cone, and sample cone are optimized for maximum inten-sitites. No nebulizer gas was used for the experiments. The ions were accu-mulated in the instrument’s hexapole long enough to obtain useful signal-to-noise ratios. Next, the ions were introduced into the FTICR analyzer cell,which was operated at pressures �10�9 mbar, and detected by a standardexcitation and detection sequence. After mass selection of the ions of interest,a 25-W CO2 laser (10.6-�m wavelength) was used to irradiate and fragmentthe ions in the IR region (infrared multiphoton dissociation; IRMPD). The laserintensities can be adjusted by the control software. Several experiments atdifferent laser intensities provide information on primary and consecutivefragmentation reactions. For each measurement, 20 scans were averaged toimprove the signal-to-noise ratio.

ACKNOWLEDGMENTS. We thank Dr. Andreas Springer for help with theESI-FTICR experiments. This work was supported by Deutsche Forschungsge-meinschaft (DFG) Grant SFB 765. C.A.S. thanks the Fonds der ChemischenIndustrie (FCI) for a Dozentenstipendium and the DFG and FCI for support.

1. Balzani V, Credi A, Raymo FM, Stoddart JF (2000) Artificial molecular machines. AngewChem Int Ed 39:3348–3391.

2. Schalley CA, Beizai K, Vogtle F (2001) On the way to rotaxane-based molecular motors:Studies in molecular mobility and topological chirality. Acc Chem Res 34:465–476.

3. Balzani V, Credi A, Venturi M (2003) Molecular devices and machines: A journey intothe nano world (Wiley–VCH, Weinheim, Germany).

4. Kay ER, Leigh DA, Zerbetto F (2007) Synthetic molecular motors and mechanicalmachines. Angew Chem Int Ed 46:72–191.

5. Whitesides GM, Mathias JP, Seto CT (1991) Molecular self-assembly and nanochemistry:A chemical strategy for the synthesis of nanostructures. Science 254:1312–1319.

6. Lehn J-M (2002) Toward self-organization and complex matter. Science 295:2400–2403.

7. Lehn J-M (2002) Toward complex matter: Supramolecular chemistry and self-organization. Proc Natl Acad Sci USA 99:4763–4768.

8. Wu AX, Isaacs L (2003) Self-sorting: The exception or the rule? J Am Chem Soc125:4831–4835.

9. Mukhopadhyay P, Wu AX, Isaacs L (2004) Social self-sorting in aqueous solution. J OrgChem 69:6157–6164.

10. Jiang W, Winkler HDF, Schalley CA (2008) Integrative self-sorting: Construction of acascade-stoppered hetero[3]rotaxane J Am Chem Soc 130:13852–13853.

11. Ballardini R, et al. (2002) Photoinduced electron transfer in a triad that can beassembled/disassembled by two different external inputs. Toward molecular-levelelectrical extension cables. J Am Chem Soc 124:12786–12795.

12. Ooya T, et al. (2006) pH-Responsive movement of cucurbit[7]uril in a diblock polypseudo-rotaxane containing dimethyl a-cyclodextrin and cucurbit[7]uril. Org Lett 8:3159–3162.

13. Zou D-P, et al. (2007) Selective binding of cucurbit[7]uril and �-cyclodextrin with aredoxactive molecular triad Ru(bpy)3–MV2�–naphthol. Chem Commun 4734–4736.

14. Yang C, et al. (2007) Dynamic switching between single- and double-axial rotaxanesmanipulated by charge and bulkiness of axle termini. Org Lett 9:4789–4792.

15. Ferrer B, et al. (2006) Photoinduced electron flow in a self-assembling supramolecularextension cable. Proc Natl Acad Sci USA 103:18411–18416.

16. Liu Y, Ke C-F, Zhang H-Y, Wu W-J, Shi J (2007) Reversible 2D pseudopolyrotaxanesbased on cyclodextrins and cucurbit[6]uril. J Org Chem 72:280–283.

17. Wang F, et al. (2008) Self-sorting organization of two heteroditopic monomers tosupramolecular alternating copolymers. J Am Chem Soc 130:11254–11255.

18. Rogez G, et al. (2007) A molecular plug-socket connector. J Am Chem Soc 129:4633–4642.

19. Miyauchi M, Harada A (2004) Construction of supramolecular polymers with alternat-ing �-, �-cyclodextrin units using conformational change induced by competitiveguests J Am Chem Soc 126:11418–11419.

20. Miyauchi M, Hoshino T, Yamaguchi H, Kamitori S, Harada A (2005) A [2]rotaxanecapped by a cyclodextrin and a guest: Formation of supramolecular [2]rotaxanepolymer. J Am Chem Soc 127:2034–2035.

21. Rudzevich Y, et al. (2005) Self-assembly of programmed building blocks into structur-ally uniform dendrimers. J Am Chem Soc 127:14168–14169.

22. Feng D-J, Li X-Q, Wang X-Z, Jiang X-K, Li Z-T (2004) Highly stable pseudo[2]rotaxanesco-driven by crown ether–ammonium and donor–acceptor interactions. Tetrahedron60:6137–6144.

23. Li X-Q, Feng D-J, Jiang X-K, Li Z-T (2004) Donor–acceptor interaction-mediated ar-rangement of hydrogen bonded dimmers. Tetrahedron 60:8275–8284.

24. Schalley CA (2001) Molecular recognition and supramolecular chemistry in the gasphase. Mass Spectrom Rev 20:253–309.

25. Baytekin B, Baytekin HT, Schalley CA (2006) Mass spectrometric studies of non-covalentcompounds: Why supramolecular chemistry in the gas phase? Org Biomol Chem4:2825–2841.

26. Frey J, et al. (2008) Cyclic [2]Pseudorotaxane tetramers consisting of two rigid rodsthreaded through two bis-macrocycles: Copper(I)-templated synthesis and X-ray struc-ture studies. J Am Chem Soc 130:11013–11022.

27. Hiraoka S, Sakata Y, Shionoya M (2008) Ti(IV)-centered dynamic interconversionbetween Pd(II), Ti(IV)-containing ring and cage molecules. J Am Chem Soc 130:10058–10059.

Jiang and Schalley PNAS � June 30, 2009 � vol. 106 � no. 26 � 10429

CHEM

ISTR

YSP

ECIA

LFE

ATU

RE

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0