Difluorobenzocyclooctyne: Synthesis, reactivity, and stabilization by β-cyclodextrin

Article abstract of DOI:10.1021/ja105005t

Highly reactive cyclooctynes have been sought as substrates for Cu-free cycloaddition reactions with azides in biological systems. To elevate the reactivities of cyclooctynes, two strategies, LUMO lowering through propargylic fluorination and strain enhancement through fused aryl rings, have been explored. Here we report the facile synthesis of a difluorobenzocyclooctyne (DIFBO) that combines these modifications. DIFBO was so reactive that it spontaneously trimerized to form two asymmetric products that we characterized by X-ray crystallography. However, we were able to trap DIFBO by forming a stable inclusion complex with β-cyclodextrin in aqueous media. This complex could be stored as a lyophilized powder and then dissociated in organic solvents to produce free DIFBO for in situ kinetic and spectroscopic analysis. Using this procedure, we found that the rate constant for the cycloaddition reaction of DIFBO with an azide exceeds those for difluorinated cyclooctyne (DIFO) and dibenzocyclooctyne (DIBO). Cyclodextrin complexation is therefore a promising approach for stabilizing compounds that possess the high intrinsic reactivities desired for Cu-free click chemistry.

Full text of DOI:10.1021/ja105005t

Published on Web 07/28/2010
Difluorobenzocyclooctyne: Synthesis, Reactivity, and
Stabilization by ꢀ-Cyclodextrin
Ellen M. Sletten,† Hitomi Nakamura,† John C. Jewett,† and Carolyn R. Bertozzi*,†,‡,§,|
Departments of Chemistry and Molecular and Cell Biology and Howard Hughes Medical
Institute, UniVersity of California, Berkeley, California 94720, and The Molecular Foundry,
Lawrence Berkeley National Laboratory, Berkeley, California 94720
Received June 8, 2010; E-mail: crb@berkeley.edu
Abstract: Highly reactive cyclooctynes have been sought as substrates for Cu-free cycloaddition reactions
with azides in biological systems. To elevate the reactivities of cyclooctynes, two strategies, LUMO lowering
through propargylic fluorination and strain enhancement through fused aryl rings, have been explored.
Here we report the facile synthesis of a difluorobenzocyclooctyne (DIFBO) that combines these modifications.
DIFBO was so reactive that it spontaneously trimerized to form two asymmetric products that we
characterized by X-ray crystallography. However, we were able to trap DIFBO by forming a stable inclusion
complex with ꢀ-cyclodextrin in aqueous media. This complex could be stored as a lyophilized powder and
then dissociated in organic solvents to produce free DIFBO for in situ kinetic and spectroscopic analysis.
Using this procedure, we found that the rate constant for the cycloaddition reaction of DIFBO with an azide
exceeds those for difluorinated cyclooctyne (DIFO) and dibenzocyclooctyne (DIBO). Cyclodextrin com-
plexation is therefore a promising approach for stabilizing compounds that possess the high intrinsic
reactivities desired for Cu-free click chemistry.
rapid and selective cyclooaddition reactions with azides,6
enabling their use as substrates for “Cu-free click chemistry”
Introduction
Growing interest in bioorthogonal reactions has prompted
many chemists to revisit exotic structures previously considered
to be of purely theoretical interest.1 The roots of many
bioorthogonal reactions are found in the physical organic
chemistry literature from the middle of the 20th century, when
chemists were studying the boundaries between reactivity and
stability of molecules.2 It was during this time that now-famous
molecules such as tetrahedrane, cubane, and Dewar benzene
were envisioned and many clever, strain-inducing reactions were
developed.3 Also of interest to physical organic chemists was
cyclooctyne, the smallest readily isolable cycloalkyne.4 Cy-
clooctynes possess ∼18 kcal/mol of strain energy, primarily as
a result of the distorted bond angles surrounding their sp-
hybridized carbon atoms.5
in biological settings.7 The applications of this chemistry have
proven to be quite broad,1 some examples being protein8 and
lipid9 labeling and imaging of glycans in cells,7c,d,f Caenorhab-
ditis elegans,10 and developing zebrafish.11
Figure 1. A panel of selected cyclooctynes and their second-order rate
constants for the reaction with benzyl azide in acetonitrile-d3 (1, 2, 4) or
methanol-d4 (3).
A theme that has emerged from these studies is the need for
rapid kinetics to achieve the desired reaction under conditions
where the azide-functionalized biological target is present in
relatively low abundance. In earlier work, we found that the
addition of a gem-difluoro group at the propargylic position,
embodied in the compound we named DIFO (1; Figure 1A),
increased the rate constant for the reaction with benzyl azide
by almost 2 orders of magnitude relative to the parent cyclooc-
tyne (OCT, 2).7a,c Boons and co-workers achieved a similar rate
enhancement by fusing two aryl rings to the cyclooctyne
scaffold, as reflected in the compound named DIBO (3).7d
Interest in cyclooctynes has reemerged in recent years for
reasons beyond their odd structures. The reagents engage in
† Department of Chemistry, University of California, Berkeley.
‡ Department of Molecular and Cell Biology, University of California,
Berkeley.
§ Howard Hughes Medical Institute.
| The Molecular Foundry.
(1) For reviews, see: (a) Sletten, E. M.; Bertozzi, C. R. Angew. Chem.,
Int. Ed. 2009, 48, 6974–6798. (b) Jewett, J. C.; Bertozzi, C. R. Chem.
Soc. ReV. 2010, 39, 1272–1279.
(2) (a) Liebman, J. F.; Greenberg, A. Chem. ReV. 1976, 76, 311–365. (b)
Hoffmann, R.; Hopf, H. Angew. Chem., Int. Ed. 2008, 47, 4474–4481.
(3) (a) Maier, G. Angew. Chem., Int. Ed. 1988, 27, 309–322. (b) Griffin,
G. W.; Marchand, A. P. Chem. ReV. 1989, 89, 997–1010. (c) Schafer,
W.; Hellmann, H. Angew. Chem., Int. Ed. 1967, 6, 518–525.
(4) Krebs, A.; Wilke, J. Top. Curr. Chem. 1983, 109, 189–233.
(5) (a) Petersen, H.; Kolshorn, H.; Meier, H. Angew. Chem., Int. Ed. 1978,
17, 461–462. (b) Turner, R. B.; Jarrett, A. D.; Goebel, P.; Mallon,
B. J. J. Am. Chem. Soc. 1973, 95, 790–792.
(6) Recently, cyclooctynes have also been reacted with nitrones in 1,3-
dipolar cycloaddition reactions. See: (a) McKay, C. S.; Moran, J.;
Pezacki, J. P. Chem. Commun. 2010, 46, 931–933. (b) Ning, X.;
Temming, R. P.; Dommerholt, J.; Guo, J.; Ania, D. B.; Debets, M. F.;
Wolfert, M. A.; Boons, G.-J.; van Delft, F. L. Angew. Chem., Int. Ed.
2010, 49, 3065–3068.
9
10.1021/ja105005t  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 11799–11805 11799

A R T I C L E S
Sletten et al.
Scheme 1. Synthesis of DIFBO
In parallel with these experimental studies, theoretical work
aimed at understanding the basis of the rate enhancements for
DIFO and DIBO relative to OCT has been performed.12 Density
functional theory (DFT)-based methods have consistently
predicted the rate-enhancing impact of fluorination, which has
been attributed to its effects on transition-state distortion/
interaction energies.12a,b The fused aryl rings of DIBO were
proposed by Goddard and co-workers to exert two opposing
effects on the rate of cycloaddition with azides. On the one hand,
the aryl rings could promote the cycloaddition reaction by
augmenting the ring strain in the cyclooctyne. However, this
effect was predicted to be tempered by unfavorable steric
interactions (i.e., A-1,3 strain) between the alkyl substituent of
the azide and the ortho hydrogen atoms of the aryl rings.
Goddard further speculated that removal of one aryl ring could
enhance the cycloaddition rate relative to DIBO, at least for
one triazole regioisomer.12c
enabling long-term storage (on the benchtop) and use as a
reagent for rapid cycloaddition with azides.
Results and Discussion
Synthesis of DIFBO. In previous cyclooctyne syntheses, we
found it prudent to install the alkyne toward the end of the
synthetic route using mild conditions. Fluoride-mediated elimi-
nation of a silyl enol triflate precursor, which proceeds rapidly
at room temperature and in high yield,7f seemed ideal for
generating the alkyne in DIFBO. Thus, we envisioned trimeth-
ylsilyl ketone 7 as an intermediate in the synthesis (Scheme 1).
Installing a silyl group R to a carbonyl functionality is
notoriously difficult, as the conditions necessary for carbon-
silicon bond formation often lead to silyl enol ether formation
instead.13 We therefore looked to disconnections other than
carbon-silicon bond formation. This led us to propose a
homologation strategy in which 7 could be produced by Lewis
acid-mediated ring expansion of difluorobenzosuberone 6 with
trimethysilyl diazomethane. Homologation with trimethylsilyl
diazomethane is well-precedented,14 although not with R-fluoro
ketones.15 Trimethylsilyl diazomethane is usually chosen as a
diazomethane surrogate on the basis of safety considerations
and not because the silyl functionality is desired.16,17
As shown in Scheme 1, 6 was synthesized by treatment of
the hexylamine imine of 1-benzosuberone (5) with Selectfluor,
a method reported by Stavber and co-workers.18 Treatment of
6 with trimethylsilyl diazomethane and a stoichiometric amount
of trimethylaluminum resulted in clean homologation to yield
compound 7 with the trimethylsilyl group intact. Immediate enol
triflate formation by deprotonation with KHMDS followed by
trapping with trifluoromethane sulfonic anhydride yielded
compound 8. This intermediate was converted to DIFBO (4)
by reaction with cesium fluoride (CsF); however, the desired
product could not be isolated in pure form. Rather, DIFBO
underwent spontaneous homotrimerization to form a mixture
of two compounds, an early indication of its extreme reactivity.4,19
After extensive NMR analysis (Figures S1 and S2 in the
Motivated by this growing body of experimental and theoreti-
cal work, we propose that compounds combining the rate-
enhancing features of DIFO and DIBO might possess superior
kinetics that are desirable for labeling of azides in biological
systems. Here we report the synthesis and characterization of
difluorobenzocyclooctyne (DIFBO) (4; Figure 1), a hybrid of
DIFO and DIBO. This new cyclooctyne was more reactive than
either parent compound, to an extent that it underwent spontane-
ous homotrimerization in solution. However, we were able to
stabilize DIFBO in an inclusion complex with ꢀ-cyclodextrin,
(7) (a) Agard, N. J.; Prescher, J. A.; Bertozzi, C. R. J. Am. Chem. Soc.
2004, 126, 15046–15047. (b) Agard, N. J.; Baskin, J. B.; Prescher,
J. A.; Lo, A.; Bertozzi, C. R. ACS Chem. Biol. 2006, 1, 644–648. (c)
Baskin, J. M.; Prescher, J. A.; Laughlin, S. T.; Agard, N. J.; Chang,
P. V.; Miller, I. A.; Lo, A.; Codelli, J. A.; Bertozzi, C. R. Proc. Natl.
Acad. Sci. U.S.A. 2007, 104, 16793–16797. (d) Ning, X.; Guo, J.;
Wolfert, M. A.; Boons, G.-J. Angew. Chem., Int. Ed. 2008, 47, 2253–
2255. (e) Debets, M. F.; van Berkel, S. S.; Schoffelen, S.; Rutjes,
F. P. J. T.; van Hest, J. C. M.; van Delft, F. L. Chem. Commun. 2010,
46, 97–99. (f) Jewett, J. C.; Sletten, E. M.; Bertozzi, C. R. J. Am.
Chem. Soc. 2010, 132, 3688–3690. (g) Codelli, J. A.; Baskin, J. M.;
Agard, N. J.; Bertozzi, C. R. J. Am. Chem. Soc. 2008, 130, 11486–
11493. For a recent review, see: (h) Debets, M. F.; van der Doelen,
C. W. J.; Rutjes, F. P. J. T.; van Delft, F. L. ChemBioChem 2010, 11,
1168–1184.
(13) Sampson, P.; Wiemer, D. F. J. Chem. Soc., Chem. Commun. 1985,
1746–1747.
(14) (a) Hashimoto, N.; Aoyama, T.; Shioiri, T. Tetrahedron Lett. 1980,
21, 4619–4622. (b) Maruoka, K.; Concepcion, A. B.; Yamamoto, H.
Synthesis 1994, 1283–1290. (c) Maruoka, K.; Concepcion, A. B.;
Yamamoto, H. Synlett 1994, 521–523.
(8) (a) Link, A. J.; Vink, M. K. S.; Agard, N. J.; Prescher, J. A.; Bertozzi,
C. R.; Tirrell, D. A. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 10180–
10185. (b) Fernandez-Suarez, M.; Baruah, H.; Martinez-Hernandez,
L.; Xie, K. T.; Baskin, J. M.; Bertozzi, C. R.; Ting, A. Y. Nat.
Biotechnol. 2007, 25, 1483–1487. (c) Nessen, M. A.; Kramer, G.;
Back, J.; Baskin, J. M.; Smeenk, L. E.; Van Maarseveen, J.; de Koning,
L. J.; de Jong, L.; Bertozzi, C. R.; Hiemstra, H.; de Koster, C. G. J.
Proteome Res. 2009, 8, 3702–3711.
(15) Homologation with TMSCHN2 has been used in the synthesis of a
cyclooctyne, although the silyl group was not retained. See: (a)
Chaffins, S.; Brettreich, M.; Wuld, F. Synthesis 2002, 1191–1194. (b)
Kornmayer, S. C.; Rominger, F.; Gleiter, R. Synthesis 2009, 2547–
2552.
(9) Neef, A. B.; Schultz, C. Angew. Chem., Int. Ed. 2009, 48, 1498–1500.
(10) Laughlin, S. T.; Bertozzi, C. R. ACS Chem. Biol. 2009, 4, 1068–1072.
(11) (a) Laughlin, S. T.; Baskin, J. M.; Amacher, S. L.; Bertozzi, C. R.
Science 2008, 320, 664–667. (b) Baskin, J. M.; Dehnert, K. W.;
Laughlin, S. T.; Amacher, S. L.; Bertozzi, C. R. Proc. Natl. Acad.
Sci. U.S.A. 2010, 107, 10360–10365.
(16) (a) Shioiri, T.; Aoyama, T. J. Synth. Org. Chem. Jpn. 1996, 54, 918–
928. (b) Hashimoto, N.; Aoyama, T.; Shioiri, T. Chem. Pharm. Bull.
1982, 30, 119–124.
(17) While this manuscript was in preparation, Li et al. reported the
homologation of aldehydes with TMSCHN2 to generate R-silyl
ketones. See: Li, J.; Sun, C.; Lee, D. J. Am. Chem. Soc. 2010, 132,
6640–6641.
(12) (a) Ess, D. H.; Jones, G. O.; Houk, K. N. Org. Lett. 2008, 10, 1633–
1636. (b) Schoenebeck, F.; Ess, D. H.; Jones, G. O.; Houk, K. N.
J. Am. Chem. Soc. 2009, 131, 8121–8133. (c) Chenoweth, K.;
Chenoweth, D.; Goddard, W. A., III. Org. Biomol. Chem. 2009, 7,
5255–5258. (d) Bach, R. D. J. Am. Chem. Soc. 2009, 131, 5233–
5243.
(18) Pravst, I.; Zupan, M.; Stavber, S. Synthesis 2005, 3140–3146.
(19) Trimerization is a well-established reaction pathway for highly reactive
cycloalkynes. See: (a) Buhl, H.; Gugel, H.; Kolshom, H.; Meier, H.
Synthesis 1978, 536. (b) Wittig, G.; Krebs, A. Chem. Ber. 1961, 94,
3260–3275.
9
11800 J. AM. CHEM. SOC. VOL. 132, NO. 33, 2010

Difluorobenzocyclooctyne
A R T I C L E S
Scheme 2. Asymmetric Trimerization of DIFBO
Scheme 4. Formation of an Inclusion Complex between DIFBO
and ꢀ-Cyclodextrin
during freeze-drying but readily dissociate upon suspension in
organic solvents. Cyclodextrin complexation has previously been
exploited to enhance the stability of lipophilic drugs.20,21 On
the basis of this precedent, we sought to test whether cyclo-
dextrins would form stable inclusion complexes with DIFBO,
thereby permitting long-term storage and facilitating kinetic
characterization (Scheme 4).
Supporting Information) and further confirmation by X-ray
crystallography (Figure 2), we determined that the two trimer
products were asymmetric diastereomers 9 and 10 (Scheme 2).
Interestingly, a trimeric product that would be derived from three
DIFBO molecules coming together in a C3-symmetric manner
was not observed.
We combined a crude solution of DIFBO in acetonitrile,
generated as shown in Scheme 1, with aqueous solutions of R-,
ꢀ-, or γ-cyclodextrin. Thin-layer chromatography and mass
spectrometry analyses showed the persistence of DIFBO in the
ꢀ-cyclodextrin-containing mixture, whereas side products formed
in the mixtures containing R- or γ-cyclodextrin (see below).
Encouraged by this observation, we developed a procedure for
generating and isolating the DIFBO-ꢀ-cyclodextrin complex.
We found that the reaction solution in which DIFBO was
generated could be directly loaded onto a silica gel column
without concentration. Upon elution with hexanes, pure DIFBO
was obtained. The DIFBO-ꢀ-cyclodextrin complex cannot be
formed in the presence of hexanes, and consequently, we
exchanged the solvent by dilution with acetonitrile followed by
careful room-temperature evaporation of the hexanes. An
aqueous solution of ꢀ-cyclodextrin was then added to the
DIFBO/acetonitrile solution, and immediately a white precipitate
formed. This mixture was lyophilized to afford a white powder.
1H NMR analysis of the powder dissolved in DMSO-d6
(Figure S5) confirmed the presence of pure DIFBO, providing
strong evidence for a successful complexation process. However,
an inclusion complex of this type would be expected to
dissociate in DMSO,22 and indeed, our NMR analysis suggested
that ꢀ-cyclodextrin and DIFBO were both free in solution
(Figure S5). To obtain direct spectral evidence of the DIFBO-ꢀ-
cyclodextrin complex, we performed solid-state cross-polariza-
tion magic-angle-spinning (CPMAS) 13C NMR analysis of the
lyophilized powder in comparison with pure ꢀ-cyclodextrin
(Figure S6). The resonance corresponding to the anomeric
carbon of glucose in the putative DIFBO-ꢀ-cyclodextrin
complex was shifted downfield from the corresponding reso-
nance in pure ꢀ-cyclodextrin. In addition, all of the cyclodextrin
13C resonances were broadened in the putative complex. Both
Figure 2. Thermal ellipsoid plots for trimer products 9 and 10 shown at
50% probability. Gray atoms correspond to carbon and green atoms to
fluorine. Hydrogen atoms have been omitted for clarity. See Figures S12
and S13 for ORTEP diagrams with labeled atoms.
To confirm that DIFBO had indeed been formed prior to
trimerization, we added benzyl azide to the reaction of
compound 8 with CsF and obtained a mixture of triazole
products 11 and 12 (1:1.6 ratio; Figures S3 and S4) in 93%
yield (Scheme 3). Thus, the rate of reaction of DIFBO with
benzyl azide appeared to be faster than that of trimerization.
Scheme 3. Formation of DIFBO and In Situ Trapping with Benzyl
Azide
of these effects have been previously observed in CPMAS 13
C
NMR spectra of ꢀ-cyclodextrin inclusion complexes,23 sup-
porting our conclusion that DIFBO was indeed complexed inside
the ꢀ-cyclodextrin cavity.
Complexation with ꢀ-Cyclodextrin. The trimerization side
reaction created several problems with regard to the character-
ization of DIFBO. The compound was impossible to isolate in
pure form at convenient temperatures and could not be stored
for future use. Thus, we sought a means of stabilizing DIFBO
for detailed spectroscopic and kinetic characterization. Cyclo-
dextrins, cyclic oligomers of R-1,4-linked glucose, are well-
known for their ability to bind small, nonpolar molecules within
their bowl-shaped cavities.20 The three commonly used cyclo-
dextrins, R, ꢀ, and γ, which comprise six, seven, and eight
glucose residues, respectively, possess different cavity dimen-
sions that can be selected to match the small-molecule guest of
interest.20 The inclusion complexes are stable in water and
(20) (a) Dodziuk, H. Cyclodextrins and Their Complexes; Wiley-VCH:
Weinheim, Germany, 2006; pp 1-489. (b) Saenger, W. Angew. Chem.,
Int. Ed. 1980, 19, 344–362. (c) Wenz, G. Angew. Chem., Int. Ed. 1994,
33, 803–822.
(21) (a) Brewster, M. E.; Loftsson, T. AdV. Drug DeliVery ReV. 2007, 59,
645–666. (b) Uekama, K. Chem. Pharm. Bull. 2004, 52, 900–915. (c)
Uekama, K.; Hirayama, F.; Irie, T. Chem. ReV. 1998, 98, 2045–2076.
(22) Tarasezwska, T. J. Inclusion Phenom. Mol. Recognit. Chem. 1991,
10, 69–78.
(23) Koontz, J. L.; Marcy, J. E.; O’Keefe, S. F.; Duncan, S. E. J. Agric.
Food Chem. 2009, 57, 1162–1171.
9
J. AM. CHEM. SOC. VOL. 132, NO. 33, 2010 11801

A R T I C L E S
Sletten et al.
Figure 4. (A) Chemical structures of dimers 13 and 14. (B) Thermal
ellipsoid plots of dimers 13 and 14 shown with 50% probability. Gray atoms
correspond to carbon, green atoms to fluorine, and red atoms to oxygen.
Hydrogen atoms have been omitted for clarity. See Figures S14 and S15
for ORTEP diagrams with labeled atoms.
absence of trimers 9 and 10. We observed neither significant
trimer formation nor any other degradation products over the
course of 2 months during storage on the benchtop at room
temperature and open to air (Figures S8 and S9). Therefore,
ꢀ-cyclodextrin appears to be a convenient host for stabilization
and storage of DIFBO. DIFBO can be readily extracted from
the cyclodextrin simply by suspending the complex in an organic
solvent and removing the insoluble free cyclodextrin by
filtration.
In contrast, in the absence of any additive (Figure 3B) or in
the presence of water (Figure 3C) or glucose (Figure 3D),
DIFBO formed trimers 9 and 10 prior to resuspension with
benzyl azide. Similarly, addition of R-cyclodextrin had no
stabilizing effect on DIFBO, and only trimers were observed,
indicating either that DIFBO cannot fit into the smaller
cyclodextrin cavity or that its complex with R-cyclodextrin is
relatively unstable (Figure 3E).
Figure 3. Reversed-phase HPLC analysis of lyophilized DIFBO mixtures
reacted with benzyl azide. A solution of DIFBO in acetonitrile was combined
with (A) ꢀ-cyclodextrin (1 equiv), (B) no additive, (C) water, (D) glucose
(1 equiv), (E) R-cyclodextrin (1 equiv), or (F) γ-cyclodextrin (1 equiv)
and concentrated. The resulting solids were resuspended in a 1:1 acetonitrile/
water mixture containing 1 equiv of benzyl azide (final concentration 600
µM). The reaction mixture was analyzed by reversed-phase HPLC with
detection by absorbance at 254 nm. Peak assignments were made through
comparison to synthetic standards (Figure S7).
The products formed from the DIFBO/γ-cyclodextrin mixture
were more complicated (Figure 3F). Trimers 9 and 10 were the
major species observed, although the presence of small amounts
of triazoles 11 and 12 indicated that some DIFBO had been
preserved through the lyophilization process. However, multiple
new peaks in the HPLC trace suggested that γ-cyclodextrin
alters the reactivity of DIFBO. We isolated the two major new
species (HPLC peaks designated with asterisks in Figure 3F)
and used mass spectrometry, NMR spectroscopy and X-ray
crystallography to identify them as oxidized DIFBO dimers 13
and 14 (Figure 4). Both 13 and 14 comprise two DIFBO
moieties as well as a molecule of oxygen, with 14 being
hydrated as well. We speculate that the two compounds are
derived from a common intermediate, cyclobutadiene 15, formed
by combination of two DIFBO molecules (Scheme 5).24 This
unstable antiaromatic intermediate25 presumably reacts rapidly
with molecular oxygen, resulting either in exocyclic (15a leading
to 14) or endocyclic (15b leading to 13) C-C bond cleavage
with respect to the cyclooctyne rings (Scheme 5). Compound
We next performed a more detailed study of DIFBO’s
stability and reactivity with azides in the presence of various
additives, including the three cyclodextrins as well as monomeric
glucose and water alone. A solution of DIFBO in acetonitrile
was combined with each of the above additives, and the aqueous
mixtures were lyophilized. The resulting powders were resus-
pended in a solution of benzyl azide in acetonitrile, and the
products were analyzed by HPLC (Figure 3) alongside synthetic
standards (Figure S7).
The reaction containing ꢀ-cyclodextrin produced two major
products, which we identified as triazoles 11 and 12, with only
trace amounts of trimer isomers 9 and 10 (Figure 3A). This
result further confirms that ꢀ-cyclodextrin prevents trimerization
of DIFBO. We then investigated the long-term stability of the
DIFBO-ꢀ-cyclodextrin complex. The dry powder was periodi-
cally suspended in 1:1 acetonitrile/water containing benzyl azide,
and the integrity of DIFBO was determined indirectly by
monitoring the formation of triazoles 11 and 12 as well as the
9
11802 J. AM. CHEM. SOC. VOL. 132, NO. 33, 2010

Difluorobenzocyclooctyne
A R T I C L E S
Scheme 5. Proposed Mechanism of DIFBO Dimerization and Trimerization
Scheme 6. Reaction of Monobenzocyclooctyne (MOBO) with
Benzyl Azide
15 may also be an intermediate in the formation of trimers 9
and 10, which constitute the two regioisomers formed by
addition of another molecule of DIFBO. Indeed, rapid formation
of 15 as the first step in the trimerization process would explain
why the C3-symmetric homotrimer was not observed.
The formation of dimers 13 and 14 only in the presence of
γ-cyclodextrin raises an interesting question regarding the
participation of this host molecule. It is possible that the
γ-cyclodextrin cavity can accommodate two DIFBO molecules
and thereby facilitate the dimerization reaction. Also, the host
might uniquely bind intermediate 15, thereby preventing its
reaction with another large DIFBO molecule but not with
molecular oxygen.
Reactivity of DIFBO in Comparison with Other Cyclooc-
tynes. The ability to extract pure DIFBO from the ꢀ-cyclodextrin
complex allowed us to test how the combination of fluorination
and a fused aryl ring affects the cycloaddition rate relative to
that for either rate-enhancing feature alone. The DIFBO-ꢀ-
cyclodextrin complex was dissolved in acetonitrile-d3, and the
free ꢀ-cyclodextrin was removed by filtration. A stoichiometric
amount of benzyl azide was added, and the reaction was
monitored by 1H NMR spectroscopy. We measured the second-
order rate constant to be 0.22 ( 0.01 M-1 s-1 (Figure S10),
which is larger than those for DIFO (k ) 0.08 M-1 s-1) and
DIBO (k ) 0.06 M-1 s-1).26 These results demonstrate that
fusion of an aryl ring to the DIFO scaffold enhances its reactivity
with azides by a significant margin and suggest that addition
of electron-withdrawing groups to the DIBO scaffold would
augment its reactivity as well. However, DIFBO and DIBO
cannot be directly compared, since according to Goddard’s
proposal,12c the second aryl ring may actually retard the reaction
rate as a result of unfavorable steric interactions between its
ortho hydrogen atom and the azide substrate in the transition
state. Thus, we synthesized monobenzocyclooctyne (MOBO,
16; Scheme 6)27 as a better substrate against which to judge
the relative contribution of fluorination to the reactivity of
DIFBO. Unlike DIFBO, MOBO was readily isolable and
displayed only minimal decomposition upon extended storage.
We measured a rate constant of 0.0095 ( 0.0003 M-1 s-1 for
its cycloaddition reaction with benzyl azide, which affords
cycloadducts 17 and 18 in a 1:2 ratio (Scheme 6, Figure S11).
This rate constant is 23-fold smaller than that observed for
DIFBO and also almost an order of magnitude lower than those
for DIFO and DIBO. This result confirms that both the electronic
effects of fluorination and the enhanced ring strain created by
the fused aryl ring contribute to DIFBO’s reactivity, though the
electronic effects seem to predominate. Also noteworthy is the
fact that DIBO reacts with benzyl azide considerably faster than
MOBO, contrary to the previous DFT-based predictions.12c
To further elucidate the rate-enhancing effects in Cu-free click
chemistry, we attempted to synthesize other substituted monoben-
zocycloalkynes using the synthetic strategy developed for DIFBO.28
Unfortunately, homologation with trimethylsilyl diazomethane did
not prove to be a general strategy for cycloalkyne synthesis (see
the Supporting Information for further details).
(24) (a) Kimling, H.; Krebs, A. Angew. Chem., Int. Ed. 1972, 11, 932–
933. (b) Krebs, A. H.; Kimling, H.; Kemper, R. Justus Liebigs Ann.
Chem. 1978, 431–439. (c) Wittig, G.; Weinlich, J. Chem. Ber. 1965,
98, 471–479. (d) Wittig, G.; Mayer, U. Chem. Ber. 1963, 96, 342–
348. (e) Krebs, A.; Byrd, D. Justus Liebigs Ann. Chem. 1967, 707,
66–74.
(25) (a) Maier, G. Angew. Chem., Int. Ed. 1974, 13, 425–490. (b) Bally,
T.; Masamune, S. Tetrahedron 1980, 36, 343–370. (c) Minkin, V. I.;
Glukhovtsev, M. N.; Simkin, B. Y. Aromaticity and Antiaromaticity:
Electronic and Structural Aspects.; Wiley: New York, 1994; pp 1-
336.
(26) The rate for DIBO was determined in MeOD and is not directly
comparable to the rates measured for DIFBO and DIFO. For a rate
constant for DIBO that is consistent with other experimental results,
see: Poloukhtine, A. A.; Mbua, N. E.; Wolfert, M. A.; Boons, G.-J.;
Popik, V. V. J. Am. Chem. Soc. 2009, 131, 15769–1171. For a direct
comparison of DIBO and DIFO, see ref 7f.
Conclusions
The expanding field of bioorthogonal chemistry has derived
considerable inspiration from classic chemistries that were first
(27) MOBO (16) was synthesized previously via selenadiazole degradation.
See: Meier, H.; Layer, M.; Zetzche, A. Chem.-Ztg 1974, 98, 460–
461. We synthesized MOBO through syn elimination of an enol triflate
(see Scheme S1).
9
J. AM. CHEM. SOC. VOL. 132, NO. 33, 2010 11803

A R T I C L E S
Sletten et al.
with a hexane/EtOAc solvent system (30:1 to 15:1). This procedure
yielded pure difluorobenzosuberone 6 as a clear oil (0.92 g, 4.7
mmol, 70%). Rf ) 0.4 in 4:1 hexane/EtOAc. 1H NMR (400 MHz,
CDCl3): δ 7.72 (dd, J ) 7.7, 1.3 Hz, 1H), 7.49 (td, J ) 6.5, 1.4
Hz, 1H), 7.36 (td, J ) 8.4, 0.9 Hz, 1H), 7.27 (dd, J ) 7.7, 0.6 Hz,
1H), 3.04-3.07 (m, 2H), 2.35-2.46 (m, 2H), 2.04 (apparent quin,
J ) 6.4 Hz, 2H). 13C NMR (100 MHz, CDCl3): δ 194.2 (t, J ) 29
Hz), 141.7, 134.8, 133.0, 130.3, 129.9, 126.9, 119.0 (t, J ) 250
Hz), 34.4 (t, J ) 24 Hz), 33.6, 22.0 (t, J ) 5 Hz). 19F NMR (376
MHz, CDCl3): δ -99.3 (t, J ) 15 Hz, 2F). HRMS (EI): calcd for
reported as mechanistic curiosities. This work has expanded on
that theme with the development of a new substituted cycloc-
tyne, DIFBO, whose reactivity profile provides insight into how
1,3-dipolar cycloaddition kinetics can be modulated. DIFBO is
highly reactive with azides (k ) 0.22 M-1 s-1) but also prone
to decomposition via trimerization. X-ray crystallography
revealed that the two trimer products were asymmetric diaster-
eomers 9 and 10, which are likely formed from a common
cyclobutadiene intermediate, 15.
C11H10OF2 [M]+, 196.0700; found, 196.0705.
+
To control the reactivity of DIFBO, we looked to another
classic molecule used in host/guest chemistry: cyclodextrin.
ꢀ-Cyclodextrin was most effective in complexing and stabilizing
DIFBO, presumably because of its appropriate cavity dimen-
sions, enabling the isolation of a stable inclusion complex that
we characterized by CPMAS 13C NMR spectroscopy. ꢀ-Cy-
clodextrin protected DIFBO from oligomerization and allowed
it to be stored as a lyophilized powder. Interestingly, our
attempts to form an inclusion complex of DIFBO and the larger
γ-cyclodextrin promoted the formation of two dimer products
(compounds 13 and 14) in addition to trimer products 9 and
10. The formation of these dimer products can also be
rationalized on the basis of the proposed intermediate 15. We
suspect that DIFBO’s reactivity is rooted in its propensity to
form this putatively antiaromatic compound. Thus, in addition
to enhancing our understanding of cyclooctyne-azide cycload-
dition reactions, DIFBO offers a new tool for studying the
classical chemical principle of (anti)aromaticity, which is still
a topic of discussion today.29 Additionally, the use of ꢀ-cyclo-
dextrin to control the reactivity of DIFBO demonstrates a
promising strategy for harnessing highly reactive small-molecule
components in bioconjugation reactions, opening the door to
new, rapid bioorthogonal chemistries.
7,7-Difluoro-5-(trimethylsilyl)-7,8,9,10-tetrahydrobenzo[8]annulen-
6(5H)-one (7). Difluorobenzosuberone 6 (4.0 mmol, 1 equiv) was
dissolved in CH2Cl2 (50 mL), and the solution was cooled to -78
°C, after which trimethylaluminum (2 M solution in toluene, 2.0
mL, 4.0 mmol, 1 equiv) was added. After 15 min, trimethylsilyl
diazomethane (2 M solution in hexanes, 2.0 mL, 4.0 mmol, 1 equiv)
was added, and 5 min later the reaction was quenched with saturated
aqueous NH4Cl and warmed to 0 °C. Rochelle’s salt was added to
complex the aluminum salts. This solution was stirred at rt for 15
min and then extracted with CH2Cl2 (3 × 75 mL). The organic
layers were combined, dried with MgSO4, decanted, and evaporated
to dryness. This procedure afforded compound 7 with 33% toluene
remaining (1.2 g, 3.9 mmol, 97%). Rf ) 0.85 in 4:1 hexane/EtOAc.
1H NMR (600 MHz, CDCl3): δ 7.17-7.22 (m, 2H), 7.07 (dd, J )
6.8, 2.0 Hz, 1H), 7.04 (dd, J ) 6.9, 1.9 Hz, 1H), 3.67 (s, 1H), 2.75
(dt, J ) 14.7, 4.6 Hz, 1H), 2.45-2.51 (m, 1H), 2.04-2.12 (m,
1H), 1.90-1.97 (m, 1H), 1.64-1.76 (m, 1H), 1.56-1.63 (m, 1H),
0.12 (s, 9H). 13C NMR (150 MHz, CDCl3): δ 201.8 (t, J ) 27 Hz),
137.7, 135.1, 130.4, 130.1, 126.9, 126.7, 119.2 (t, J ) 252 Hz),
48.6, 33.1 (t, J ) 25 Hz), 30.2, 23.6 (t, J ) 5 Hz), -1.6. 19F NMR
(564 MHz, CDCl3): δ -100.95 (apparent t, J ) 17 Hz, 1F),
-100.99 (dd, J ) 23, 17 Hz, 1F). HRMS (EI): calcd for
C15H20OF2Si+ [M]+, 282.1253; found, 282.1255.
7,7-Difluoro-5-(trimethylsilyl)-7,8,9,10-tetrahydrobenzo[8]annulen-
6-yl Triflate (8). A solution of compound 7 (8.5 mmol, 1 equiv) in
THF (110 mL) was cooled to -78 °C, and potassium bis(trimeth-
ylsilyl)amide (0.5 M solution in toluene, 20.4 mL, 10.2 mmol, 1.2
equiv) was added, after which the solution turned a dark orange/
brown color. After 1 h, triflic anydride (2.07 mL, 10.2 mmol, 1.2
equiv) was added, and the solution turned a lighter yellow color.
The solution was stirred for 3 h, warming to roughly -45 °C, at
which point it was quenched with MeOH and evaporated to dryness.
(CAUTION: Use of a large excesss of Tf2O or warming this reaction
to room temperature before quenching results in the formation of
a gel from which it is very difficult to separate compound 8.) The
crude product was purified via silica gel chromatography with a
hexane/toluene solvent system (75:1, 50:1, 25:1). This procedure
resulted in pure trimethylsilyl enol triflate 8 in 80% yield (2.8 g,
Experimental Procedures
2,2-Difluoro-1-benzosuberone (6).18 1-Benzosuberone (5, 1.0
mL, 6.7 mmol, 1 equiv) was dissolved in cyclohexane (13.5 mL),
after which hexylamine (1.2 mL, 9.1 mmol, 1.4 equiv) and
trifluoroacetic acid (5 drops) were added. The reaction mixture was
heated to reflux overnight with azeotropic removal of water
(Dean-Stark trap). The following morning, the reaction mixture
was evaporated to dryness, dissolved in Et2O (25 mL), and washed
with saturated NaHCO3 (1 × 15 mL) and brine (1 × 15 mL). The
organic solution was dried over MgSO4 and evaporated to dryness.
The resulting crude imine (1.8 g) was dissolved in CH3CN (67
mL). To this solution were added Selectfluor (5.01 g, 14.2 mmol,
2.1 equiv) and Na2SO4 (680 mg, 4.9 mmol). The reaction mixture
was heated to reflux overnight. The following morning, 3 M HCl
(4.5 mL) was added to hydrolyze the imine. After 10 min at reflux,
the solution was cooled to room temperature (rt) and evaporated
to dryness. The residue was dissolved in Et2O (25 mL) and then
washed with saturated NaHCO3 (1 × 15 mL) and brine (1 × 15
mL). The organic solution was dried over MgSO4 and evaporated
to dryness. The crude mixture was purified by flash chromatography
1
6.8 mmol). Rf ) 0.8 in 6:1 hexane/EtOAc. H NMR (400 MHz,
CDCl3): δ 7.24-7.29 (m, 2H), 7.16-7.19 (m, 1H), 7.03-7.07 (m,
1H), 2.77 (dt, J ) 13.1, 4.4 Hz, 1H), 2.56 (td, J ) 5.3, 1.3 Hz,
1H), 1.91-2.13 (m, 2H), 1.49-1.69 (m, 2H), 0.19 (s, 9H). 13C
NMR (100 MHz, CDCl3): δ 144.9 (t, J ) 28 Hz), 142.7 (t, J ) 4
Hz), 136.2 (t, J ) 3 Hz), 135.8, 128.7, 128.4, 126.7, 126.5, 119.0
(q, J ) 321 Hz), 118.3 (dd, J ) 246, 241 Hz), 32.6 (t, J ) 25 Hz),
30.2, 23.9 (dd, J ) 7, 3 Hz), -0.62. 19F NMR (376 MHz, CDCl3):
δ -70.1 (t, J ) 13 Hz, 3F), -84.5 (dsex, J ) 277, 11 Hz, 1F),
-92.0 (dm, J ) 278 Hz, 1F). HRMS (EI): calcd for C16H19O3F5SiS+
[M]+, 414.0744; found, 414.0642.
(28) The ketones shown below were synthesized, but their conversion to
cyclooctynes was prevented because they did not undergo homolo-
gation with TMSCHN2 (see the Supporting Information for further
details).
Difluorobenzocyclooctyne (DIFBO, 4). Trimethylsilyl enol
triflate 8 (34 mg, 0.082 mmol, 1 equiv) was dissolved in CD3CN
(3 mL), and CsF (75 mg, 0.50 mmol, 6 equiv) was added. The
reaction was stirred at rt for ∼30 min (the reaction mixture turned
yellow and had a foul odor), at which point it was transferred
directly onto a plug of silica gel and eluted with CD3CN (0.75 mL).
A portion of this solution was taken for NMR analysis. Rf ) 0.75
1
in 4:1 hexane/EtOAc. H NMR (500 MHz, CD3CN): δ 7.38 (t, J
(29) Bally, T. Angew. Chem., Int. Ed. 2006, 45, 6616–6619.
) 8.1 Hz, 1H), 7.33 (t, J ) 7.3 Hz, 1H), 7.26-7.31 (m, 2H), 2.97
9
11804 J. AM. CHEM. SOC. VOL. 132, NO. 33, 2010

Difluorobenzocyclooctyne
A R T I C L E S
(apparent t, J ) 5.0 Hz, 2H), 2.57-2.65 (m, 2H), 1.89 (bs, 2H). 4
was too unstable for 13C NMR analysis. 19F NMR (564 MHz,
CD3CN): δ -87.5 (bs, 2F). HRMS (EI): calcd for C12H10F2+ [M]+,
192.0751; found, 192.0754.
Analysis of DIFBO-Cyclodextrin Complexes (Figure 3).
DIFBO was generated, purified, and transferred into acetonitrile
as described above. The resulting DIFBO/acetonitrile solution was
divided into six portions (A-F) that were treated in different
manners. Portion B was evaporated to dryness and placed on the
lyophilizer. Portion C was combined with H2O (4 mL), after which
the acetonitrile was removed by rotary evaporation and the resulting
aqueous solution flash-frozen and lyophilized. Portion D was
combined with 1 equiv of glucose in H2O (4 mL) and treated in
the same manner as portion C. Portions A, E, and F were treated
in the same manner as portion C except that the glucose was
replaced by ꢀ-cyclodextrin, R-cyclodextrin, and γ-cyclodextrin
respectively.
Cycloadducts of Benzyl Azide and DIFBO (11 and 12).
Compound 8 (202 mg, 0.49 mmol, 1 equiv) and CsF (440 mg, 2.9
mmol, 5.9 equiv) were combined in CH3CN (10 mL). After 15
min, BnN3 (66 µL, 0.53 mmol, 1.1 equiv) was added. This mixture
was allowed to stir overnight at rt. The following morning, the
reaction mixture was evaporated to dryness and purified by silica
gel chromatography (hexane/EtOAC system), affording two triazole
products, 11 (55 mg, 0.17 mmol, 35%) and 12 (92 mg, 0.28 mmol,
1
58%). Compound 11: Rf ) 0.2 in 6:1 hexane/EtOAc. H NMR
(600 MHz, CD3CN): δ 7.48 (td, J ) 7.5, 1.5 Hz, 1H), 7.44 (dd, J
) 7.7, 1.3 Hz, 1H), 7.39 (apparent td, J ) 7.5, 1.3 Hz, 1H), 7.30
(d, J ) 7.6 Hz, 1H), 7.20-7.22 (m, 3H), 6.88-6.90 (m, 2H), 5.56
(apparent d, J ) 3.1 Hz, 2H), 2.57 (apparent dd, J ) 13.4, 6.2 Hz,
1H), 2.06-2.19 (m, 3H), 1.73-1.85 (m, 1H), 1.61-1.67 (m, 1H).
13C NMR (150 MHz, CD3CN): δ 143.1 (t, J ) 30 Hz), 141.1, 136.6,
135.3 (t, J ) 6 Hz), 132.0, 130.5, 130.0, 129.7, 129.1, 128.2, 127.9,
126.6, 119.1 (dd, J ) 235, 231 Hz), 53.0, 31.9 (t, J ) 24 Hz),
30.6, 25.9 (d, J ) 9 Hz). 19F NMR (564 MHz, CD3CN): δ -91.8
For HPLC analysis of the resulting products, 0.6 µmol of each
powder was dissolved in 1:1 CH3CN/H2O (1 mL). To this solution
was added 1 equiv of BnN3 (15 µL of 0.04 mM solution in 1:1
CH3CN/H2O), and the mixture was allowed to react overnight. The
following morning, 150 nmol (250 µL) was analyzed by reversed-
phase HPLC (C-18 column) using a gradient of 50% CH3CN/50%
H2O to 100% CH3CN over 25 min followed by isocratic elution
with 100% CH3CN for 10 min before re-equilibration to 50%
CH3CN/50% H2O. The chromatography was monitored by absor-
bance at 210 and 254 nm.
+
(ddd, J ) 271, 34, 6 Hz, 2F). HRMS (ESI): calcd for C19H18N3F2
[M + H]+, 326.1463; found, 326.1461. Compound 12: Rf ) 0.6 in
1
6:1 hexanes/EtOAc. H NMR (600 MHz, CD3CN): δ 7.72 (dd, J
Acknowledgment. We thank Dr. A. DiPasquale for the X-ray
crystallography and Dr. C. Canlas for the CPMAS NMR experi-
ments. Additionally, we thank Dr. C. Canlas and Dr. R. Nunlist
for help with the 19F NMR experiments. E.M.S. was supported by
a predoctoral fellowship from the American Chemical Society
Division of Organic Chemistry (Genentech). H.N. was partially
supported by the Amgen Scholars Program. J.C.J. was supported
by a postdoctoral fellowship from the American Cancer Society.
This work was supported by a grant to C.R.B. from the National
Institutes of Health (GM58867). We thank K. Dehnert for critical
reading of the manuscript.
) 7.6, 1.3 Hz, 1H), 7.45 (td, J ) 7.5, 1.5 Hz, 1H), 7.36-7.41 (m,
3H), 7.32-7.35 (m, 4H), 5.80 (s, 2H), 2.50-2.68 (bm, 2H),
1.62-2.21 (bm, 4H). 13C NMR (150 MHz, CD3CN): δ 145.8 (t, J
) 6 Hz), 139.5, 137.0, 131.5, 130.8, 130.7 (t, J ) 131 Hz), 130.4,
130.2, 129.7, 129.2, 128.7, 127.8, 119.2 (t, J ) 235 Hz), 54.5 (t,
J ) 3 Hz), 32.4 (t, J ) 24 Hz), 30.8, 26.5 (t, J ) 5 Hz). 19F NMR
(564 MHz, CD3CN): δ -74.1 (d, J ) 270 Hz, 1F), -89.8 (d, J )
265 Hz, 1F). HRMS (ESI): calcd for C19H18N3F2 [M + H]+,
+
326.1463; found, 326.1460.
Generation of the DIFBO-ꢀ-Cyclodextrin Complex. DIFBO
precursor 8 (256 mg, 0.618 mmol, 1 equiv) was dissolved in CH3CN
(26 mL), and CsF (573 mg, 3.77 mmol, 6.1 equiv) was added. This
mixture was allowed to stir at rt until all of the 8 was converted
into DIFBO (∼25 min; the reaction mixture turned slightly yellow
in color and had a foul odor). This solution was directly loaded
onto silica gel and eluted with hexane. The fractions containing
DIFBO were combined with ∼75 mL of acetonitrile, and the hexane
was removed by careful rotary evaporation at rt. The remaining
DIFBO/acetonitrile solution was added to a solution of ꢀ-cyclo-
dextrin (recrystallized, 702 mg, 0.618 mmol, 1 equiv, in 50 mL of
deionized H2O). The solution turned cloudy. The acetonitrile was
removed by rotary evaporation, and the resulting aqueous solution
was flash-frozen in liquid N2 and lyophilized to a white powder
(688 mg, 0.518 mmol, 84%).
Supporting Information Available: Figures S1-S15; Scheme
S1; characterization data and CIF files for oligomer products
9, 10, 13, and 14; general experimental procedures; CPMAS
NMR procedure; synthesis of MOBO (16) and ketones
19-22; determination of the second-order rate constants for
DIFBO and MOBO; NMR spectra for all reported com-
pounds. This material is available free of charge via the
Internet at http://pubs.acs.org.
JA105005T
9
J. AM. CHEM. SOC. VOL. 132, NO. 33, 2010 11805