Induced-fit catalysis of corannulenebowl-to-bowl inversionMichal Jurıcek1, Nathan L. Strutt1, Jonathan C. Barnes1, Anna M. Butterfield2, Edward J. Dale1,

Kim K. Baldridge2,3, J. Fraser Stoddart1 and Jay S. Siegel2,3*

Stereoelectronic complementarity between the active site of an enzyme and the transition state of a reaction is one of thetenets of enzyme catalysis. This report illustrates the principles of enzyme catalysis (first proposed by Pauling and Jencks)through a well-defined model system that has been fully characterized crystallographically, computationally and kinetically.Catalysis of the bowl-to-bowl inversion processes that pertain to corannulene is achieved by combining ground-statedestabilization and transition-state stabilization within the cavity of an extended tetracationic cyclophane. This syntheticreceptor fulfils a role reminiscent of a catalytic antibody by stabilizing the planar transition state for the bowl-to-bowlinversion of (ethyl)corannulene (which accelerates this process by a factor of ten at room temperature) by an induced-fitmechanism first formulated by Koshland.

Enzymatic catalysis depends strongly on what Pauling1 referredto as specific complementarity between the active-site structureof a host and the activated state of a guest. This level of ligand

specificity of an enzyme led Jencks2 to propose the concept of anti-body catalysis, in which a specific antibody can be generated inresponse to a hapten that resembles the transition state of the reac-tion in question. Employing this methodology of a transition-stateanalogue (TSA), the groups of Lerner3 and Schultz4 developed anti-body hydrolases. Since then, many antibodies5–17 have been ‘pro-grammed’ to behave like enzymes for standard synthetic organicreactions (for example, Diels–Alder cycloadditions18–23) usingstrongly binding TSAs. These amazing achievements notwithstand-ing, there remains a need for a simple kinetically and structurallywell-defined and well-characterized model system to illustrate thePauling–Jencks model. Corannulene24–41 (Fig. 1a), a C5v symmetricbowl-shaped polycyclic aromatic hydrocarbon (bowl depth42

0.87 Å) undergoes degenerate bowl-to-bowl inversion along awell-defined reaction path that passes through a planar D5htransition state. ExBox 41, a synthetic tetracationic cyclophane43

(Fig. 1b), comprises two p-electron-poor extended bipyridiniumunits that selectively bind planar aromatics comparable to thecorannulene transition state (Fig. 1a). Herein, ExBox 41 is shownto catalyse the bowl-to-bowl inversion process of corannulene;kinetic, crystallographic and computational studies support thissystem as a fiduciary one that displays catalysis consistent with thePauling–Jencks model.

The energy barrier for bowl-to-bowl inversion of nascentcorannulene has been estimated38 to be 11.5 kcal mol21 from itsground-state bowl form to its transition-state planar form.Corannulene’s ‘bowl depth’ (for example, 0.87 Å from the bestplane of the hub carbon atoms to the best plane of rim carbonatoms) serves38 as a composite reaction coordinate for the bowl-to-bowl inversion process. Substituents in the peri positions ofcorannulene cognates demonstrate that the barrier for bowl-to-bowl inversion depends38 on the fourth power of this bowl depth;thus, a 10% modulation in bowl depth results in a 103 change in

rate, which makes the system ideal for studying catalysis inducedby stereoelectronic binding.

Free ExBox 41 is ideally suited43 for binding planar aromaticguests. Its cavity is 3.5 Å wide and 11.3 Å long after taking intoconsideration van der Waals radii, and it forms strong one-to-onearene , ExBox 41 complexes with an array of polycyclic aromatichydrocarbons that range in size from two to seven fused rings (coroneneis among the most strongly bonded substrates). In general, greatersurface areas and orbital overlap lead to larger binding constants.

Corannulene’s ground-state bowl form, with van der Waals dimen-sions of 10× 4.3 Å, is too thick to fit into ExBox41 with any appreci-able overlap of the p-surface area (Fig. 1). Complexation could occur if

a

b

6.9 Å 3.5 Å

4.3 Å

0.87 Å

4.3 Å

0.87 Å

3.4 Å

0.0 Å

N N

N N

+

+

+

+

Figure 1 | Structural formulae and solid-state structures. a,b, The structural

formulae and solid-state structures, obtained by X-ray crystallography, of

corannulene (a) and ExBox41 (b). The optimized geometry of the planar

transition state of the bowl-to-bowl inversion process of corannulene (a)

was obtained by DFT (B97D/Def2-TZVPP).

1Center for the Chemistry of Integrated Systems, Department of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, Illinois 60208, USA,2Organic Chemistry Institute (OCI), University of Zurich, Winterthurerstrasse 190, Zurich, CH-8057, Switzerland, 3School of Pharmaceutical Science andTechnology, Tianjin University (A210/Building 24), 92 Weijin Road, Nankai District, Tianjin, 300072 PRC, China. *e-mail: dean_spst@tju.edu.cn

ARTICLESPUBLISHED ONLINE: 26 JANUARY 2014 | DOI: 10.1038/NCHEM.1842

NATURE CHEMISTRY | VOL 6 | MARCH 2014 | www.nature.com/naturechemistry222

© 2014 Macmillan Publishers Limited. All rights reserved.

ExBox41 and corannulene undergo ‘induced-fit’ conformationalchanges (as first proposed by Koshland44–46) and the final bindingstrength would result from an optimization of the energy cost ofinduced fit against the energy benefit of a better p-surface overlap.The energy ‘stored’ in the form of strain in the widened ExBox41 canbe viewed as the potential energy that forces corannulene to adopt aflatter conformation inside ExBox41, which effectively decreases theenergy barrier for the bowl-to-bowl inversion process. Consequently,ExBox41 distorts the bowl-shaped ground state and stabilizes theplanar transition state of corannulene, as it adapts its shape continu-ously to fit the substrate tightly at each intermediate state.

Results and discussionIn the solid-state superstructure (Fig. 2a and Supplementary Fig. 24)of corannulene , ExBox 41, corannulene is positioned withinExBox 41 such that the middle five-membered ring partially over-laps with both para-phenylene rings of ExBox 41. Corannulene isnot centred precisely within ExBox 41, but rather is offset and pro-trudes on one side. The width of ExBox 41 in the host–guestcomplex increases by 0.87 Å (from 6.92 to 7.79 Å) and the bowlwidth of corannulene decreases by 0.03 Å (from 0.87 to 0.84 Å),supporting the induced-fit hypothesis.

The same geometrical trends are observed (Fig. 2c–f) when theoptimized (B97D/Def2-TZVPP) geometries for the ground andtransition states of the corannulene , ExBox 41 1:1 complex arecompared47–57. The width of ExBox 41 increases by 0.37 and0.29 Å in the gas phase and Me2CO, respectively. The bowl depthof corannulene in the ground state decreases by 0.03 and 0.02 Åin the gas phase and Me2CO, respectively.

In the present context, coronene can serve as a conceptual TSAfor the bowl-to-bowl inversion of corannulene. The planar confor-mation of coronene is a consequence of its central six-memberedring and is to be contrasted, in the case of corannulene, with itscentral five-membered ring, which mandates its bowl-shaped con-formation. Indeed, the crystal superstructure43 of coronene ,ExBox 41 (Fig. 2b) displays an analogous superstructure to that ofthe complex between corannulene in its transition state andExBox 41. The length and width of ExBox 41 in coronene ,ExBox 41 have values very close to those obtained for the optimizedgeometries of the transition state—that is, approximately 14.5 Å inlength and 7.1 Å in width (Fig. 2d,f ).

The binding affinities (Ka) of ExBox 41 with corannulene weredetermined (Supplementary Figs 1, 2 and 18–23) as roughly103 M21: Ka¼ 3.12 × 103+41 M21 (1H NMR spectroscopy,MeCN-d3), 6.45 × 103+0.66 × 103 M21 (isothermal titrationcalorimetry (ITC), MeCN), 863+25 M21 (1H NMR spectroscopy,Me2CO-d6) and 5.02 × 103+3.1 × 103 M21 (ITC, Me2CO). Thehigher binding affinity in MeCN(-d3) compared with that inMe2CO(-d6) can be attributed to corannulene’s greater solubilityin Me2CO(-d6) than in MeCN(-d3). The binding affinity betweenExBox 41 and corannulene arises from multiple sources, includingcharge-transfer interactions, surface-area binding phenomena58,van der Waals force and solvophobic effects59.

The binding affinity of an aromatic guest towards ExBox 41

increases43 exponentially according to the number of p electrons.Perylene, like corannulene, has 20 p electrons in thearomatic system, but perylene is flat. The Ka value of perylene(1H NMR spectroscopy, MeCN-d3) is 88.1 × 103+67 × 103 M21

7.79 Å

14.14 Å

0.84 Å

a c

0.84 Å

7.61 Å

14.35 Å

0.85 Å

7.28 Å

14.48 Å

e

b f

0 Å

7.05 Å

14.49 Å

0 Å

14.49 Å

7.24 Å

0 Å

14.58 Å

6.99 Å

d

Figure 2 | Ground state versus transition state. a,b, Top and side-on views of the solid-state superstructures, obtained by X-ray crystallography, of

corannulene , ExBox41 (a) and coronene , ExBox41 (b) 1:1 complexes. c–f, Top and side-on views of the optimized geometries, obtained by DFT

(B97D/Def2-TZVPP), of the ground and transition states of the corannulene , ExBox41 complex in the gas phase (c,d, respectively) and in the solvent

(Me2CO) model (e,f, respectively).

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1842 ARTICLES

NATURE CHEMISTRY | VOL 6 | MARCH 2014 | www.nature.com/naturechemistry 223

© 2014 Macmillan Publishers Limited. All rights reserved.

(DG¼26.74+0.45 kcal mol21)43, approximately 30 times thatobtained for corannulene (3.12 × 103+41 M21 (1H NMR spectro-scopy, MeCN-d3), DG¼24.76+0.077 kcal mol21). The energyloss (1.98+0.46 kcal mol21) that lowers the overall affinity ofcorannulene towards ExBox 41 is, again, the result of the straingenerated in the components on complexation.

Variable-temperature 1H NMR experiments were performed on2:1 (Supplementary Fig. 4), 1:1 (Supplementary Fig. 5) and 1:2 mix-tures (Fig. 2 and Supplementary Fig. 6) of ExBox 41 and corannu-lene-d10 (ref. 60), and on a 1:1 mixture of ExBox 41 andcorannulene (Supplementary Fig. 3) in Me2CO-d6. Depending onthe rate of complexation equilibrium (in/out) and the bowl-to-bowl inversion processes on the 1H NMR timescale, one or multiplesets of signals for ExBox 41 and corannulene were observed. At162 K in Me2CO-d6, the complexation equilibrium (in/out) andthe bowl-to-bowl inversion processes are slow relative to the 1HNMR timescale. In the 1H NMR spectrum (Supplementary Figs 1and 2) of the 1:1 mixture of ExBox 41 and corannulene, the 32hydrogens of ExBox 41 give rise to five signals and the ten hydrogensof corannulene to one signal. As a result of the face-to-face orien-tation of corannulene inside ExBox 41, the b and g protons ofExBox 41, as well as the protons of corannulene, shift upfield,in contrast to the C6H4 ( para-phenylene) signals of ExBox 41,which shift downfield. Signals for the a and CH2 protons positionedin the corners of ExBox 41 are not affected by the shielding effectof the guest and do not exhibit significant shifts. In a sample witha ratio of 2:1 ExBox 41 to corannulene-d10, two sets of signals,which arise from bound and unbound ExBox 41, are observedbelow the coalescence temperature (Tc) (Supplementary Figs 4and 9), and two sets of signals for bound and unbound corannuleneare observed below Tc for a sample with a 1:2 ratio of ExBox 41 tocorannulene-d10 (Supplementary Fig. 6). In all mixtures, two setsof signals are observed additionally for bound ExBox 41 (b and g

protons) below Tc because the signals for the ‘top’ b and g

protons that face towards the inside of the bowl and for the‘bottom’ b and g protons that face towards the outside of thebowl each become heterotopic when corannulene does notundergo bowl-to-bowl inversion, or when the process is slow relativeto the 1H NMR timescale (Fig. 3). One set of signals is observed, ineach case, for the bound and unbound corannulene, which indicatesthat its rotation, both inside and outside ExBox 41, is fast on the 1HNMR timescale, even at low temperatures.

The energy barrier for the bowl-to-bowl inversion process of cor-annulene inside ExBox 41 was investigated by dynamic 1H NMRspectroscopy. The difference in chemical shifts for the two sets ofsignals of bound ExBox 41 below Tc can be used to determine theenergy barrier and rate for the bowl-to-bowl inversion process(Fig. 3). The chemical shift of the g protons of ExBox 41 is themost sensitive to the change of chemical environment, as theseprotons are positioned right above and below the centre of thebound corannulene core. The signals for the g protons were there-fore used to determine the energy barrier associated with bowl-to-bowl inversion in Me2CO-d6. The sharp singlet for the g protonsat 6.86 ppm broadens (Fig. 3b) and shifts upfield slightly as thetemperature is decreased. When the temperature reaches 190+5 K,the signal for the g protons becomes broadened into the baselineand then, at 165 K, appears again as two signals at 7.45 and5.08 ppm, which correspond to the ‘bottom’ g2 protons that facetowards the outside and the ‘top’ g1 protons that face towards theinside of the corannulene bowl, respectively.

Dynamic 1H NMR line-shape simulations were conducted todetermine the values of the rate constants (k) at various tempera-tures and to estimate the value (190+5 K) of Tc. The Eyringequation, DG‡

c ¼2RT ln(kch/kBTc), was then used to calculateDG‡

c values at Tc and the Eyring plot (Supplementary Fig. 16) wasused to determine DH‡ and DS‡ values, which were used to calculate

δ (ppm)

C20HD9Hβ CH2C6H4Hα Hγ

300

290

280

270

260

250

240

230

220

210

200

190

180

170

165

HαHβ1

Hγ1

CH2

C6H4

Hγ2

Hβ2

C6H4

CH2

CH2

CH2

Hα

Hα Hβ1

Hα Hβ2

5678910

T (K)

4PF6–

δ (ppm) δ (ppm)

200

190

180

165

Sim

Sim

Sim

Simk = 470 s–1

k = 1,200 s–1

k = 3,190 s–1

k = 5,400 s–1300

280

260

230

Sim

Sim

Sim

Sim

T (K) T (K)

CH2

k = 46,500 s–1

k = 251,160 s–1

k = 642,000 s–1

k = 1,580,000 s–1

CH2

Hγ1Hγ2

Hγ1Hγ2

Hγ1 = Hγ2

Hγ1 = Hγ2

Hγ1 = Hγ2

C20HD9

b

a

c

Hγ2Hβ2

CH2

Hα

Hα

Hβ2

C6H4

C6H4

Hα

Hβ1Hγ1

CH2

CH2

Hα

Hβ1

CH2

4PF6–

T > 190 KHγ1

= Hγ2

T < 190 KHγ1

≠ Hγ2

567 567

Hγ1Hγ2

Hβ1

Hβ2

Figure 3 | NMR spectroscopy with corannulene-d10. a, Bowl-to-bowl

inversion process of corannulene inside ExBox41. The signals for the ‘top’ g1

protons that face towards the inside of the bowl and the ‘bottom’ g2 protons

that face towards the outside of the bowl each become heterotopic below

Tc (,190 K). b, Stacked spectra for the variable-temperature 1H NMR

spectroscopic measurements of a 1:2 mixture of ExBox†4PF6 and corannulene-d10

(C20D10) in Me2CO-d6. Tc of the 1H NMR resonances for g protons of

ExBox†4PF6 was determined as 190+5 K. c, Measurements of the rate

constants (k) for the in/out equilibrium process of corannulene and ExBox41

were conducted by dynamic 1H NMR line-shape simulations (Sim) using the

program iNMR (version 3.2.1). The obtained values of the kinetic parameters

were: DG‡c¼ 7.91+0.22 kcal mol21, DH‡¼ 5.62+0.19 kcal mol21,

DS‡¼211.9+0.91 cal mol21 K21, DG‡190K¼ 7.88+0.26 kcal mol21.

ExBox† 4PF6 spectra, green; corannulene spectra, black.

ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1842

NATURE CHEMISTRY | VOL 6 | MARCH 2014 | www.nature.com/naturechemistry224

© 2014 Macmillan Publishers Limited. All rights reserved.

DG‡T values by using the equation DG‡

T¼ DH‡ 2 TDS‡. The DG‡T

values at 190 K (DG‡190K) were used to compare energy barriers of

different processes.The energy barrier (DG‡

190K) (corresponding to either the in/outequilibrium or the bowl-to-bowl inversion process) was found to be7.88+0.26 kcal mol21 (DG‡

c ¼ 7.91+0.22 kcal mol21) for the 1:2mixture of ExBox41 and corannulene-d10 (Fig. 3 and SupplementaryFig. 7). For a comparison, the value of the energy barrier (DG‡

c) forthe 2:1 (Supplementary Fig. 4) and 1:1 (Supplementary Fig. 5)mixtures of ExBox 41 and corannulene-d10 was found to be7.91+0.22 kcal mol21 on average. The 7.88+0.26 kcal mol21

value is smaller, by 3.6 kcal mol21, than that (11.5 kcal mol21) ofthe estimated38 energy barrier for free corannulene. The bowl-to-bowl inversion process of unbound parent corannulene cannotbe observed in this system because of symmetry degeneracy38.

The energy barrier for the bowl-to-bowl inversion of corannuleneinside ExBox 41 was also calculated using density functional theory(DFT) (B97D/Def2-TZVPP)47–57. The calculated values for theenergy barrier associated with this process are 8.77 kcal mol21 inthe gas phase and 8.47 kcal mol21 using a continuum solvent model(Me2CO), both of which are in reasonable agreement with the valueobtained experimentally in Me2CO-d6 (7.88+0.26 kcal mol21).

The bowl-to-bowl inversion process of ethylcorannuleneinside ExBox 41 was studied by dynamic 1H NMR spectroscopy to

evaluate more accurately whether the 7.88+0.26 kcal mol21 valuefor the energy barrier corresponds to the bowl-to-bowl inversionor rather to the in/out equilibrium process. Ethylcorannuleneshows a very similar binding affinity to ExBox41 in bothMeCN-d3 (Ka¼ 2.86 × 103+0.72 × 103 M21) and Me2CO-d6(Ka¼ 530+31 M21) compared to those obtained for corannulene(Supplementary Figs 10 and 11). Moreover, the sterically non-demanding ethyl substituent should not alter36 the bowl depth ofthe corannulene core significantly and, thus, the energy barrier forthe bowl-to-bowl inversion process should be very close to thatobserved for corannulene. The 1H NMR signals for the CH2protons of ethylcorannulene (which become heterotopic when thebowl-to-bowl inversion is slow relative to the 1H NMR timescale)were used to determine the energy-barrier values (DG‡

190K) inMe2CO-d6. These values were found to be 10.8+0.62 kcal mol21

for nascent ethylcorannulene (Fig. 4a,c and Supplementary Figs12 and 13) and 8.71+0.36 kcal mol21 for the 2:1 mixture ofExBox 41 and ethylcorannulene (Fig. 4b,d and SupplementaryFigs 14 and 15). This decrease of 2.09+0.72 kcal mol21 in theenergy barrier demonstrates the ability of ExBox 41 to catalyse thebowl-to-bowl inversion of ethylcorannulene, and accelerate thisprocess by a factor of ten at room temperature based on the rateconstants obtained from 1H NMR line-shape simulations(see Supplementary Figs 13 and 15).

δ (ppm)

2.32.52.72.9

H1 H2

220

210

195

170

Sim

Sim

Sim

Simk = 26 s–1

k = 530 s–1

k = 2,650 s–1

k = 6,000 s–1

T (K)H1 = H2

3.03.13.23.33.4

δ (ppm)

280

260

240

Sim

Sim

Sim

T (K)

k = 335 s–1

k = 2,000 s–1

k = 14,580 s–1

H1 = H2

H1 H2

220

Sim

k = 90 s–1

c d

300

280

270

260

250

240

230

220

210

200

190

180

170

290

H1 = H2

195

H1 H2

T (K)

δ (ppm)

2.32.52.72.93.1

300

280

270

260

250

245

240

235

230

220

210

200

190

180

170

290

H1 H2

H1 = H2T (K)

3.03.13.23.33.4

δ (ppm)

a b

HCor-Ar

H1H2

HCor-Ar

H1H2

* *

HCor-Ar

H1 H2

HCor-Ar

H1 H2

Figure 4 | NMR spectroscopy with ethylcorannulene. a,b, Bowl-to-bowl inversion process (top) and stacked spectra (bottom) for the variable-temperature1H NMR spectroscopic measurements of ethylcorannulene (a) and a 2:1 mixture of ExBox† 4PF6 and ethylcorannulene (b) in Me2CO-d6. Tc values of the 1H

resonances for CH2 protons of ethylcorannulene were determined as 240+5 K for nascent ethylcorannulene (a) and 195+5 K for bound ethylcorannulene

(b). c,d, Measurements of the rate constants (k) for the bowl-to-bowl inversion processes of nascent ethylcorannulene (c) and ethylcorannulene inside

ExBox41 (d) were conducted by dynamic 1H NMR line-shape simulations (Sim) using the program iNMR (version 3.2.1). The obtained values of the kinetic

parameters were: DG‡c¼ 11.2+0.12 kcal mol21, DH‡¼ 9.95+0.49 kcal mol21, DS‡¼24.35+2.0 cal mol21 K21, DG‡

190K¼ 10.8+0.62 kcal mol21 (c), and

DG‡c¼ 8.82+0.12 kcal mol21, DH‡¼ 7.16+0.27 kcal mol21, DS‡¼28.14+1.2 cal mol21 K21, DG‡

190K¼ 8.71+0.36 kcal mol21 (d). ExBox† 4PF6 spectra,

green; corannulene spectra, black, HDO (2.81 ppm at 300 K), H2O (2.77 ppm at 300 K) and CHD2COCD3 (2.05 ppm) spectra, pink; ,ExBox41,

green asterisks; Ar, aryl.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1842 ARTICLES

NATURE CHEMISTRY | VOL 6 | MARCH 2014 | www.nature.com/naturechemistry 225

© 2014 Macmillan Publishers Limited. All rights reserved.

The energy barrier in free corannulene determined by DFT(B97D/Def2-TZVPP)47–57 was found to be 10.92 kcal mol21 inMe2CO, and an energy barrier of 8.47 kcal mol21 was calculatedfor corannulene , ExBox41 in Me2CO. The calculated energy-barrier decrease (2.45 kcal mol21) for corannulene in Me2CO(Fig. 5a) is in excellent agreement with that (2.09+0.72 kcal mol21)measured for ethylcorannulene in Me2CO-d6. Moreover, the2.09+0.72 kcal mol21 energy-barrier decrease is in a very close agree-ment with the value of the ‘energy loss’ (2.03+0.47 kcal mol21;binding energy (DG) of perylene43 versus ethylcorannulene(Supplementary Fig. 11) inside ExBox 41 in MeCN-d3) that occurson complexation as a result of induced-fit conformationalchanges, demonstrating that this ‘stored’ potential energyeffectively decreases the energy barrier of the bowl-to-bowl inver-sion process. The 7.88+0.26 kcal mol21 value obtained forcorannulene , ExBox 41 in Me2CO-d6 must, therefore, correspondto the energy barrier of the ‘out’ process, a conclusion that wassupported by dynamic 2H NMR measurements and line-shapeanalysis for corannulene-d10 , ExBox41 in Me2CO (SupplementaryFig. 17, DG‡

190K (‘out’)¼ 8.16+0.21 kcal mol21). The value ofthe energy barrier for the bowl-to-bowl inversion process ofcorannulene , ExBox 41 (‘energy loss’ of 1.98+0.46 kcal mol21)is expected to be very close to that obtained for ethylcorannulene(8.71+0.36 kcal mol21).

It has been shown38 that decreasing the bowl depth of corannuleneby modifying its periphery (and so, effectively, destabilizing itsground state) lowers the energy barrier for the bowl-to-bowl inver-sion process. In the corannulene , ExBox 41 complex, corannulene’s

bowl depth was found to be 0.84 Å (X-ray diffraction, Fig. 2a) and0.85 Å (DFT, Fig. 2e). The portion of the energy-barrier decreaserelated to the ground-state destabilization of corannulene insideExBox 41 has been calculated by DFT (B97D/Def2-TZVPP)47–57 tobe 0.5 kcal mol21. Considering that the energy-barrier decrease(DDG‡

catalysis) for the bowl-to-bowl inversion of corannulene insideExBox 41 was calculated to be 2.5 kcal mol21 and for ethylcorannu-lene inside ExBox 41 it was found to be 2.09+0.72 kcal mol21,DDG‡

catalysis is not merely the result of corannulene ground-statedestabilization (0.5 kcal mol21), but mainly of stabilization(2.0 kcal mol21) of the planar transition-state structure in thebowl-to-bowl inversion process of corannulene (Fig. 5b).

ConclusionIn summary, we have identified a catalytic biomimetic system inwhich the energy-barrier decrease of the bowl-to-bowl inversionof corannulene and ethylcorannulene is achieved by combiningthe effects of the ground-state destabilization and transition-statestabilization within the cavity of a synthetic receptor, which com-prises two extended bipyridinium units joined end-to-end by twopara-xylylene linkers. This tetracationic cyclophane adopts a rolesimilar to that of a catalytic antibody, wherein the planar transitionstate of the guest is stabilized through the stereoelectronic reorgan-ization of the host in the host–guest complex from a strained to anenergetically favourable conformation. These experimental obser-vations, in conjunction with DFT calculations, provide an alterna-tive induced-fit mechanism (transition-state stabilization versusground-state destabilization) for lowering the energy barrier of thebowl-to-bowl inversion process of corannulene. In this simple text-book example, catalysis of the inversion process in corannuleneinduced by stereoelectronic binding of corannulene inside a syn-thetic receptor can be followed along one ‘reaction’ coordinate,where the reactant and the product are identical.

MethodsComputational data. Four pdb files (Supplementary pdbs 1–4) of the optimizedgeometries (B97D/Def2-TZVPP) of the ground and transition states of corannulene, ExBox 41 complex in the gas phase and in Me2CO are available in theSupplementary Information.

X-ray crystallography. The crystallographic data for corannulene ,ExBox † 4PF6(MeCN)7 are available free of charge from the CambridgeCrystallographic Data Centre (CCDC) via www.ccdc.cam.ac.uk/data_request/cif.CCDC Number: 950390.

Method. Corannulene (0.7 mg, 3 mmol) was added to a solution of ExBox † 4PF6(3.0 mg, 2.4 mmol) in MeCN (0.8 ml) and, after it had dissolved, the mixture waspassed through a 0.45 mm filter equally into three 1 ml tubes. The tubes wereplaced together in one 20 ml vial that contained i-Pr2O (�3 ml) and the vial wascapped. Slow vapour diffusion of i-Pr2O into the 1.25:1 solution of corannulene andExBox † 4PF6 in MeCN over a period of three days yielded yellow single crystals(0.27 × 0.22 × 0.07 mm) of corannulene , ExBox † 4PF6(MeCN)7 (space groupP, Supplementary Section S6b). The solid-state superstructure of corannulene ,ExBox † 4PF6(MeCN)7 is shown in Supplementary Fig. 24.

Crystal parameters. [C20H10 , C48H40N4† (PF6)4] † (CH3CN)7: yellow block

(0.27 × 0.22 × 0.07 mm), triclinic, P, a¼ 17.8249(5), b¼ 20.7626(6),c¼ 24.4682(7) Å, a¼ 84.8192(14), b¼ 86.0993(15), g¼ 66.3786(14)8,V¼ 8,257.8(4) Å3, Z¼ 4, T¼ 100.05 K, rcalc¼ 1.440 g cm23, m¼ 1.782 mm21.Of a total of 75,420 reflections that were collected, 27,233 were unique. FinalR1(F2 . 2sF2)¼ 0.0808 and wR2¼ 0.2272. Data were collected at 100 K on a BrukerKappa APEX CCD diffractometer equipped with a CuKa microsource with Quazaroptics. SADABS-2008/1 (Bruker, 2008) was used for the absorption correction.wR2(int) was 0.0478 before and 0.0387 after correction. The ratio of minimum tomaximum transmission was 0.8801. The half correction factor was 0.0015. Groupanisotropic displacement parameters were refined for the disordered MeCNmolecules. Distance restraints were refined for the disordered nitrogen atom N(14)in the MeCN molecule. Rigid bond restraints (estimated standard deviation (e.s.d.)0.01) were imposed on the displacement parameters as well as restraints on similaramplitudes (e.s.d. 0.05) separated by less than 1.7 Å on the disordered PF6

2 ions andthe disordered atoms of the para-phenylene ring in ExBox †4PF6. Refinement of F2

was performed against all reflections. The weighted R-factor wR and goodness of fit Sare based on F2, and conventional R-factors R are based on F, with F set to zero for

FlatBowl Bowl

FlatBowl Bowl

E (

kcal

mol

–1)

ΔGbinding

ΔΔG‡catalysis

0

4.0

14.9

a

8.5

ΔG‡unbound

ΔG‡unbound

ΔG‡unbound

Unbound Unbound

Bound Bound

b

ΔG‡bound

ΔG‡bound

ΔG‡bound

ΔΔG‡GS destabilization

ΔΔG‡TS stabilization

10.9

kca

l mol

–1

2.5 kcal mol–1

2.0 kcal mol–1

0.5 kcal mol–1

Figure 5 | Energy profile. a, Comparison of the relative-energy (E) profiles

for the bowl-to-bowl inversion of corannulene (black) and corannulene ,

ExBox41 (green) in Me2CO calculated by DFT (B97D/Def2-TZVPP). The

value of the energy-barrier decrease (DDG‡catalysis) was determined as

2.5 kcal mol21. b, Absolute contributions of ground-state (GS) destabilization

(0.5 kcal mol21) and transition-state (TS) stabilization (2.0 kcal mol21) to the

overall energy-barrier decrease (DDG‡catalysis) of the bowl-to-bowl inversion

process of corannulene inside ExBox41 calculated by DFT (B97D/Def2-TZVPP).

ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1842

NATURE CHEMISTRY | VOL 6 | MARCH 2014 | www.nature.com/naturechemistry226

© 2014 Macmillan Publishers Limited. All rights reserved.

negative F2. The threshold expression of F2 . 2sF2 is used only to calculateR-factors(greater than), and so on, and is not relevant to the choice of reflections forrefinement. R-factors based on F2 are statistically about twice as large as thosebased on F, and R-factors based on all data will be even larger.

1H NMR spectroscopy. The 1H NMR (298 K, 500 or 600 MHz) titrations wereperformed by adding small volumes of a solution of (ethyl)corannulene inMe2CO-d6 or MeCN-d3 to a solution of ExBox † 4PF6 in Me2CO-d6 or MeCN-d3,respectively. The upfield shifts of the 1H NMR resonances for g protons wereobserved and used to determine the association constants (Ka). The Ka values werecalculated using Dynafit61, a program that employs nonlinear least-squaresregression on host–guest binding data. Measurements of the rate constants (k) forthe bowl-to-bowl inversion process of (ethyl)corannulene were conducted62

by dynamic 1H NMR line-shape simulations using the program iNMR(version 3.2.1). The Eyring equation, DG‡

c ¼2RT ln(kch/kBTc), was used tocalculate63 the DG‡

c values at Tc. The DH‡ and DS‡ values were obtainedthrough an Eyring plot (1/T against ln(k/T), Supplementary Fig. 16) generatedusing rate constants (k) determined by dynamic 1H NMR line-shapesimulations (Supplementary Figs 7, 13 and 15). The DG‡

190K values wereobtained using equation DG‡

T¼ DH‡ 2 TDS‡.

ITC. The ITC measurements were performed in dry, degassed Me2CO and MeCN at298 K. A solution of ExBox † 4PF6 was used as the host solution in a 1.8 ml cell.Solutions of corannulene were added by injecting 10 ml of titrant successivelyover 20 seconds (25×) with intervals of 300 seconds between each injection.Experiments were repeated three times. Thermodynamic information wascalculated using a one-site binding model to utilize data, from which the heat ofdilution of the guest was subtracted, with the average of three runs reported(Supplementary Figs 18–23).

Received 15 August 2013; accepted 9 December 2013;published online 26 January 2014

References1. Pauling, L. Nature of forces between large molecules of biological interest.

Nature 161, 707–709 (1948).2. Jencks, W. P. Catalysis in Chemistry and Enzymology (McGraw-Hill, 1969).3. Tramontano, A., Janda, K. D. & Lerner, R. A. Catalytic antibodies. Science 234,

1566–1570 (1986).4. Pollack, S. J., Jacobs, J. W. & Schultz, P. G. Selective chemical catalysis by an

antibody. Science 234, 1570–1573 (1986).5. Janda, K. D., Schloeder, D. & Benkovic, S. J. Induction of an antibody

that catalyzes the hydrolysis of an amide bond. Science 241,1188–1191 (1988).

6. Kang, A. S., Barbas, C. F., Janda, K. D., Benkovic, S. J. & Lerner, R. A. Linkage ofrecognition and replication functions by assembling combinatorial antibodyfab libraries along phage surfaces. Proc. Natl Acad. Sci. USA 88,4363–4366 (1991).

7. Lerner, R. A., Benkovic, S. J. & Schultz, P. G. At the crossroads of chemistryand immunology: catalytic antibodies. Science 252, 659–667 (1991).

8. Benkovic, S. J. Catalytic antibodies. Annu. Rev. Biochem. 61, 29–54 (1992).9. Janda, K. D., Shevlin, C. G. & Lerner, R. A. Antibody catalysis of a disfavored

chemical transformation. Science 259, 490–493 (1993).10. Janda, K. D. et al. Direct selection for a catalytic mechanism from combinatorial

antibody libraries. Proc. Natl Acad. Sci. USA 91, 2532–2536 (1994).11. Tawfik, D. S., Eshhar, Z. & Green, B. S. Catalytic antibodies: a critical assessment.

Mol. Biotechnol. 1, 87–103 (1994).12. Stewart, J. D. & Benkovic, S. J. Transition-state stabilization as a measure of

the efficiency of antibody catalysis. Nature 375, 388–391 (1995).13. Schultz, P. G. & Lerner, R. A. From molecular diversity to catalysis: lessons from

the immune system. Science 269, 1835–1842 (1995).14. Wagner, J., Lerner, R. A. & Barbas, C. F. III. Efficient aldolase catalytic

antibodies that use the enamine mechanism of natural enzymes. Science 270,1797–1800 (1995).

15. Janda, K. D. et al. Chemical selection for catalysis in combinatorial antibodylibraries. Science 275, 945–948 (1997).

16. Hoffmann, T. et al. Aldolase antibodies of remarkable scope. J. Am. Chem. Soc.120, 2768–2779 (1998).

17. Wentworth, P. Jr & Janda, K. D. Catalytic antibodies. Cell Biochem. Biophys.35, 63–87 (2001).

18. Hilvert, D., Hill, K. W., Nared, K. D. & Auditor, M-T. M. Antibody catalysisof the Diels–Alder reaction. J. Am. Chem. Soc. 111, 9261–9262 (1989).

19. Braisted, A. C. & Schultz, P. G. An antibody-catalyzed bimolecularDiels–Alder reaction. J. Am. Chem. Soc. 112, 7430–7431 (1990).

20. Gouverneur, V. E. et al. Control of the exo and endo pathways of theDiels–Alder reaction by antibody catalysis. Science 262, 204–208 (1993).

21. Romesberg, F. E., Spiller, B., Schultz, P. G. & Stevens, R. C. Immunologicalorigins of binding and catalysis in a Diels–Alderase antibody. Science279, 1929–1933 (1998).

22. Heine, A. et al. An antibody exo Diels–Alderase inhibitor complex at 1.95angstrom resolution. Science 279, 1934–1940 (1998).

23. Serganov, A. et al. Structural basis for Diels–Alder ribozyme-catalyzedcarbon–carbon bond formation. Nature Struct. Mol. Biol. 12, 218–224 (2005).

24. Barton, W. E. & Lawton, R. G. Dibenzo[ ghi,mno]fluoranthene. J. Am. Chem. Soc.88, 380–381 (1966).

25. Barth, W. E. & Lawton, R. G. Synthesis of corannulene. J. Am. Chem. Soc.93, 1730–1745 (1971).

26. Schulman, J. M., Peck, R. C. & Disch, R. L. Ab initio heats of formation ofmedium-sized hydrocarbons. 11. The benzenoid aromatics. J. Am. Chem. Soc.111, 5675–5680 (1989).

27. Scott, L. T., Hashemi, M. M. & Bratcher, M. S. Corannulene bowl-to-bowlinversion is rapid at room temperature. J. Am. Chem. Soc. 114,1920–1921 (1992).

28. Borchardt, A., Fuchicello, A., Kilway, K. V., Baldridge, K. K. & Siegel, J. S.Synthesis and dynamics of the corannulene nucleus. J. Am. Chem. Soc.114, 1921–1923 (1992).

29. Disch, R. L. & Schulman, J. M. Theoretical studies of the inversion barrier incorannulenes. J. Am. Chem. Soc. 116, 1533–1536 (1994).

30. Sygula, A. & Rabideau, P. W. Bowl-to-bowl inversion in polynuclear aromatichydrocarbons with curved surfaces: an ab initio study. Chem. Commun.1497–1499 (1994).

31. Sygula, A. & Rabideau, P. W. Structure and inversion barriers of corannulene, itsdianion and tetraanion. An ab initio study. J. Mol. Struct. Theochem. 333,215–226 (1995).

32. Seiders, T. J., Baldridge, K. K. & Siegel, J. S. Synthesis and characterization of thefirst corannulene cyclophane. J. Am. Chem. Soc. 118, 2754–2755 (1996).

33. Rabideau, P. W. & Sygula, A. Buckybowls: polynuclear aromatichydrocarbons related to the buckminsterfullerene surface. Acc. Chem. Res. 29,235–242 (1996).

34. Scott L. T. et al. Corannulene: a three-step synthesis. J. Am. Chem. Soc.119, 10963–10968 (1997).

35. Seiders, T. J., Elliot, E. L., Grube, G. H. & Siegel, J. S. Synthesis ofcorannulene and alkyl derivatives of corannulene. J. Am. Chem. Soc. 121,7804–7813 (1999).

36. Biedermann, P. U., Pogodin, S. & Agranat, I. Inversion barrier of corannulene. Abenchmark for bowl-to-bowl inversions in fullerene fragments. J. Org. Chem.64, 3655–3662 (1999).

37. Marcinow, Z., Sygula, A., Ellern, A. & Rabideau, P. W. Lowering inversionbarriers of buckybowls by benzannelation of the rim: synthesis and crystal andmolecular structure of 1,2-dihydrocyclopenta[b,c]dibenzo[ g,m]corannulene.Org. Lett. 3, 3527–3529 (2001).

38. Seiders, T. J., Baldridge, K. K., Grube, G. H. & Siegel, J. S. Structure/energycorrelation of bowl depth and inversion barrier in corannulene derivatives:combined experimental and quantum mechanical analysis. J. Am. Chem. Soc.123, 517–525 (2001).

39. Wu, Y-T. & Siegel, J. S. Aromatic molecular-bowl hydrocarbons: syntheticderivatives, their structures, and physical properties. Chem. Rev. 106,4843–4867 (2006).

40. Osuna, S. & Houk, K. N. Cycloaddition reactions of butadiene and 1,3-dipolesto curved arenes, fullerenes, and nanotubes: theoretical evaluation of therole of distortion energies on activation barriers. Chem. Eur. J. 15,13219–13231 (2009).

41. Butterfield, A. M., Gilomen, B. & Siegel, J. S. Kilogram-scale production ofcorannulene. Org. Process Res. Dev. 16, 664–676 (2012).

42. Hanson, J. C. & Nordman, C. E. The crystal and molecular structure ofcorannulene, C20H10. Acta Crystallogr. B B32, 1147–1153 (1976).

43. Barnes, J. C. et al. Exbox: a polycyclic aromatic hydrocarbon scavenger. J. Am.Chem. Soc. 135, 183–192 (2013).

44. Koshland, D. E. Jr. Application of a theory of enzyme specificity to proteinsynthesis. Proc. Natl Acad. Sci. USA 44, 98–104 (1958).

45. Koshland, D. E. Jr. Correlation of structure and function in enzyme action.Science 142, 1533–1541 (1963).

46. Koshland, D. E. Jr. The key–lock theory and the induced fit theory. Angew.Chem. Int. Ed. 33, 2375–2378 (1994).

47. Schmidt, M. et al. General atomic and molecular electronic structure system.J. Comp. Chem. 14, 1347–1363 (1993).

48. Frisch, M. J. et al. Gaussian 09, Revision A.1 (Gaussian, Inc., Wallingford,Connecticut, 2009).

49. Grimme, S. Semiempirical GGA-type density functional constructed with along-range dispersion correction. J. Comput. Chem. 27, 1787–1799 (2006).

50. Peverati, R. & Baldridge, K. K. Implementation and performance of DFT-Dwith respect to basis set and functional for study of dispersion interactionsin nanoscale aromatic hydrocarbons. J. Chem. Theor. Comput. 4,2030–2048 (2008).

51. Becke, A. D. Density-functional exchange-energy approximation with correctasymptotic behavior. Phys. Rev. A 38, 3098–3100 (1988).

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1842 ARTICLES

NATURE CHEMISTRY | VOL 6 | MARCH 2014 | www.nature.com/naturechemistry 227

© 2014 Macmillan Publishers Limited. All rights reserved.

52. Weigend, F. & Ahlrichs, R. Balanced basis sets of split valence, triple zeta valenceand quadruple zeta valence quality for H to Rn: design and assessment ofaccuracy. Phys. Chem. Chem. Phys. 7, 3297–3305 (2005).

53. Klamt, A. & Schurmann, G. COSMO: a new approach to dielectric screening insolvents with explicit expressions for the screening energy and its gradient.J. Chem. Soc. Perkin Trans. 2 5, 799–805 (1993).

54. Baldridge, K. K. & Klamt, A. First principles implementation of solvent effectswithout outlying charge error. J. Chem. Phys. 106, 6622–6633 (1997).

55. Klamt, A., Jonas, V., Burger, T. & Lohrenz, C. W. Refinement andparametrization of COSMO-RS. J. Phys. Chem. 102, 5074–5085 (1998).

56. Baldridge, K. K. & Greenberg, J. P. Qmview: a computational chemistry three-dimensional visualization tool at the interface between molecules and mankind.J. Mol. Graph 13, 63–66 (1995).

57. WebMO, http://www.webmo.net/index.html.58. Toyota, S. et al. Tetranuclear copper(I)-biphenanthroline gridwork: violation

of the principle of maximal donor coordination caused by intercalation andCH-to-N forces. Angew. Chem. Int. Ed. 40, 751–754 (2001).

59. Baldridge, K. K., Cozzi, F. & Siegel, J. S. Basicity of (2,6-pyridino)paracyclophanes: lone pair–p, cation–p, and solvation effects. Angew. Chem.Int. Ed. 51, 2903–2906 (2012).

60. Duttwyler, S., Butterfield, A. M. & Siegel, J. S. Arenium acid catalyzeddeuteration of aromatic hydrocarbons. J. Org. Chem. 78, 2134–2138 (2013).

61. Kuzmic, P. Program DYNAFIT for the analysis of enzyme kinetic data:application to HIV proteinase. Anal. Biochem. 237, 260–273 (1996).

62. Bain, A. D., Rex, D. M. & Smith, R. N. Fitting dynamic NMR lineshapes.Magn. Reson. Chem. 39, 122–126 (2001).

63. Sutherland, I. O. The investigation of the kinetics of conformational changesby nuclear magnetic resonance spectroscopy. Annu. Rep. NMR Spectrosc.4, 71–235 (1971).

AcknowledgementsThe authors thank M. Stuparu for synthesizing bromocorannulene and C. L. Stern forperforming the X-ray crystallographic analysis. This research is part of the Joint Center ofExcellence in Integrated Nano-Systems (JCIN) at King Abdul-Aziz City for Science andTechnology (KACST) and Northwestern University (NU) (Project 34-947). The authorswould like to thank both KACST and NU for their continued support of this research. Wealso acknowledge support from the World Class University Program (R-31-2008-000-10055-0) in Korea. M.J. gratefully acknowledges The Netherlands Organisation forScientific Research and the Marie Curie Cofund Action (Rubicon Fellowship). N.L.S. andE.J.D. are supported by a Graduate Research Fellowship from the National ScienceFoundation. J.C.B. is supported by a National Defense Science and Engineering GraduateFellowship from the Department of Defense and gratefully acknowledges receipt of a RyanFellowship from the NU International Institute for Nanotechnology. K.K.B. and J.S.S.gratefully acknowledge the Swiss National Science Foundation, the Qian Ren ScholarProgram of China and the Synergetic Innovation Center of Chemical Science andEngineering (Tianjin).

Author contributionsM.J., N.L.S., J.C.B., J.F.S. and J.S.S. conceived the project and prepared the manuscript. M.J.,J.C.B., A.M.B. and E.J.D. synthesized the different molecules studied in this work. M.J. andN.L.S. carried out NMR studies. K.K.B. performed DFT calculations. M.J., N.L.S., J.C.B.,K.K.B., J.F.S. and J.S.S. investigated the bowl-to-bowl inversion process.

Additional informationSupplementary information and chemical compound information are available in theonline version of the paper. Reprints and permissions information is available online atwww.nature.com/reprints. Correspondence and requests for materials should beaddressed to J.S.S.

Competing financial interestsThe authors declare no competing financial interests.

ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1842

NATURE CHEMISTRY | VOL 6 | MARCH 2014 | www.nature.com/naturechemistry228

© 2014 Macmillan Publishers Limited. All rights reserved.