Carbohydrate recognition in water by a tricyclic polyamide receptor

Article abstract of DOI:10.1002/anie.200461409

A sweet center: Most carbohydrates are supremely comfortable in water. Can they be tempted to leave? Yes, if their new home satisfies all their needs. The watersoluble receptor 1 (R = polycarboxylate) provides hydrophobic surfaces (purple) for CH groups a

Full text of DOI:10.1002/anie.200461409

Communications
Molecular Recognition
Carbohydrate Recognition in Water by a Tricyclic
Polyamide Receptor**
Emmanuel Klein, Matthew P. Crump, and
Anthony P. Davis*
Carbohydrate recognition in aqueous solution remains an
important challenge in supramolecular chemistry.[1] Saccha-
rides are complex, irregular, and multifunctional, while their
hydroxylated exteriors blend well with water. Even nature has
difficulty; binding constants of proteins with monosaccha-
rides peak at approximately 107 mꢀ1, a remarkably low value
for biological molecular recognition.[1a,2] Synthetic receptors
ꢀ
that exploit covalent B O bonds have been relatively
successful,[1b,3] but there are still few reports of binding in
water through noncovalent interactions.[1a,4] Such systems are
of special interest as they shed light on natural carbohydrate
recognition, a process of fundamental biological signifi-
cance.[5]
A few years ago, we reported the octa-amide 1a as a
receptor for octyl glycosides in organic media.[6] The tricyclic
core 1 was specifically targeted at b-glucosyl derivatives 2. It
ꢀ
was supposed that the axial hydrogens in 2 would make C H–
p contacts with the biphenyl surfaces, while the equatorial
substituents would form hydrogen bonds to the isophthala-
mide spacers. Accordingly, 1a showed high affinities for 2
[*] Dr. E. Klein, Dr. M. P. Crump, Prof. A. P. Davis
School of Chemistry
University of Bristol
Cantock’s Close
Bristol BS81TS (UK)
Fax: (+44)117-929-8611
E-mail: anthony.davis@bristol.ac.uk
[**] Financial support was provided by the European Commission
(network contract HPRN-CT-2002-00190). Assistance from Dr. C.
Dempsey (Department of Biochemistry, University of Bristol) and
the EPSRC National Mass Spectrometry Service Centre, Swansea, is
gratefully acknowledged.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
298
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200461409
Angew. Chem. Int. Ed. 2005, 44, 298 –302

Angewandte
Chemie
(R = octyl) and was less effective for the a
anomer or octyl b-d-galactoside. Our prime
objective, however, was recognition in water. In
this more-challenging medium, hydrogen bond-
ing would provide less (if any) impetus for
ꢀ
binding, but, in compensation, the C H–p
interactions should be reinforced by the hydro-
phobic effect. Tetracarboxylate 1b could not be
employed owing to aggregation in water.
Attempts to use a tris(hydroxymethyl)methyl
solubilizing group also foundered, although
precursor 1c proved useful for phase-transfer
experiments.[7] We now describe the synthesis
of dodecacarboxylate 1d and report that this
polyanion provides, at last, an opportunity to
study carbohydrate recognition by 1 in homo-
geneous aqueous solution.
Scheme 1. Synthesis of 6: a) tert-butyl acrylate, NaOH, DMSO, H2O; b) 1,3,5-benzenetricarbonyl
trichloride (excess), iPr2NEt, THF; c) aq. NaOH (2m); d) aq. HCl (3.5m); e) C6F5OH, DCC, THF.
DMSO=dimethylsulfoxide, DCC=N,N’-dicyclohexylcarbodiimide.
Receptor 1d was prepared as shown in
Schemes 1 and 2. The active ester 6 was synthe-
sized from tris(hydroxymethyl)aminomethane
3 through the aminotriester 4[8] and diacid 5 (Scheme 1). The
biphenyl unit 7 was prepared as previously reported.[7] Acid
deprotection of 7 followed by cyclization with 6 under high
dilution gave macrocycle 8 (Scheme 2). Hydrogenolysis of 8
followed by a second [2+2] macrolactamisation gave the
protected macrocycle 9. The sodium salt of dodecacarbox-
ylate 1d was obtained through deprotection of 9 with
trifluoroacetic acid followed by treatment with sodium
Preliminary NMR titration studies on the dodeca-tert-
butyl ester 9 confirmed that, as expected, this macrocycle
binds carbohydrates in the organic solvent system CDCl3/
CD3OD (92:8). The Ka value for octyl b-d-glucopyranoside
was 600mꢀ1, which is similar to those obtained previously for
1a[6] and 1c.[7] Variations in the chemical shifts were also
similar to those for the earlier systems. More significantly, the
addition of glucose 10 to 1d in D2O/H2O yielded qualitatively
similar NMR effects. In particular, the signal for the inward-
directed spacer protons, B (for labeling, see Figure 1), moved
downfield, whereas that for the annular amide protons, NHa,
moved upfield and split into two (Figure 2). The signal for the
protons C of the biphenyl group also split and moved
1
hydroxide. H NMR spectra of solutions in D2O/H2O con-
sisted of sharp signals (Figure 1) and did not vary with
concentration in the range 1–4 mm. Receptor 1d was there-
fore assumed to be monomeric in aqueous solution at
approximately 1 mm.
Scheme 2. Synthesis of 1d: a) TFA, CH2Cl2; b) 6, iPr2NEt, THF, high dilution; c) H2, 10% Pd/C, THF/methanolic ammonia (1:1); d) 6, iPr2NEt,
THF, high dilution; e) TFA, CH2Cl2; f) aq. NaOH (1m). Boc=tert-butoxycarbonyl, Cbz=benzyloxycarbonyl, TFA=trifluoroacetic acid.
Angew. Chem. Int. Ed. 2005, 44, 298 –302
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
299

Communications
Figure 1. Partial 1H NMR spectrum (600 MHz, 93:7 H2O/D2O) of
receptor 1d.
&
^
Figure 3. Theoretical ( ) and experimental ( ) binding curves for
a) NMR titration of 1d+10 (proton B) and b) fluorescence emission
titration of 1d+10 (cps=counts per second; lex =266 nm,
lem =475 nm).
values (especially when Ka ꢁ 2mꢀ1), but selectivities are
clearly significant. The receptor shows a notable preference
for b-glucosyl units in accordance with the original design and
with previous work on 1a and 1c. Thus the highest Ka values
are to methyl b-d-glucoside 15 and cellobiose 22. The
selectivity for 22 relative to lactose 23 and maltose 24 is
especially striking and mirrors the results in organic media for
an “extended”, terphenyl-based analogue of 1a.[10] Other
units with all-equatorial substitution (2-deoxy-d-glucose 17
and d-xylose 18) are also bound well by the receptor.
Remarkably, the removal of hydroxyl groups does not
enhance binding—17 is no more strongly bound than glucose,
while rhamnose 20 and fucose 21 are poor substrates.
The formation of the complexes was also detected by
means of fluorescence spectroscopy. On addition of substrates
to 1d, the fluorescence emission was enhanced and shifted
slightly in wavelength (see Supporting Information). Analysis
by curve fitting (see Figure 3b) yielded the binding constants
shown in Table 2 which are in good agreement with the values
obtained through 1H NMR titrations.
Figure 2. 1H NMR spectra (600 MHz, 93:7 H2O/D2O) of receptor 1d
after addition of increasing amounts of d-glucose 10: a) 0, b) 20,
b
&
*
c) 50, d) 200, and e) 1000 equivalents. =NH , =protons B,
~
=protons C.
downfield. The splitting of the eight equivalent NHa and C
protons into groups of 4 is expected for a D2h receptor that
interacts with an asymmetrical substrate.[6] As illustrated in
Figure 3a the motions could be fitted, to high accuracy, to a
1:1 binding model. The resulting binding constants varied
according to the history of the glucose solution. For freshly
dissolved glucose with an anomeric ratio a/b of 72:28 a Ka
value of 4.6mꢀ1 was obtained in D2O. For glucose that had
been equilibrated overnight (a/b = 40:60), the binding con-
stant was 9.2mꢀ1. The figures imply substantial selectivity for
the b anomer; indeed, they are consistent with Ka ꢁ 14mꢀ1 for
b-glucopyranose and < 1mꢀ1 for the a anomer.[9]
Table 2: Association constants (Ka) between receptor 1d and carbohy-
drates as measured in H2O by fluorescence titration.[a]
Carbohydrate
Ka [mꢀ1
]
Carbohydrate
Ka [mꢀ1
]
d-glucose
9.5
methyl a-d-glucoside
5.7
(10; a/b 40:60)
methyl b-d-glucoside
(15)
(16)
d-xylose
(18)
32
3.8
Binding to other carbohydrates (11–24) was followed by
NMR spectroscopy to yield the association constants shown
in Table 1, page 301. Errors may be high for the smaller Ka
[a] T=298 K. For binding curves and further details, see Supporting
Information.
300
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 298 –302

Angewandte
Chemie
1
Table 1: Association constants (Ka) between receptor 1d and carbohydrates as measured in D2O by H NMR titration.[a]
Carbohydrate
Ka [mꢀ1
]
Carbohydrate
Ka [mꢀ1
]
d-glucose
4.6
2-deoxy-d-glucose[c]
7.2
(a/b 72:28)[b]
10
d-glucose
9.2
d-xylose[c]
d-lyxose[c]
l-rhamnose[c]
l-fucose[c]
4.6
(a/b 40:60)[c]
d-galactose[c]
d-mannose[c]
d-arabinose[c]
2.1
v.s.[d]
v.s.[d]
2.1
v.s.[d]
2.2
d-ribose[c]
3.1
27.3
6.9
d-cellobiose[c]
d-lactose[c]
16.6
methyl
b-d-glucoside
v.s.[d]
v.s.[d]
methyl
a-d-glucoside
d-maltose[c]
[a] T=296 K. Data were analyzed by using a nonlinear least-squares curve-fitting program implemented within MS Excel 2000. Quoted figures were
derived from the peak for proton B (see Figure 1); peaks for protons C and (sometimes) A could also be followed and gave closely consistent results.
For binding curves and further details, see Supporting Information. [b] Freshly dissolved in D2O. Anomeric ratio was monitored by 1H NMR
spectroscopy and was found to remain roughly constant throughout the titration. [c] Solution in D2O allowed to equilibrate before use. For further
information on anomeric ratios, see Supporting Information. [d] Very small. Minor signal movements, almost linear with substrate concentration.
Information on the geometry of the 1d·glucose complex
was provided by NOESY experiments. A portion of the
spectrum is shown in Figure 4. Cross-peaks are observed
between glucose protons (b anomer only) and receptor
protons B, C, and NHa, whereas connections from substrate
to externally directed protons A, NHb, and D are weak or
Figure 4. Partial NOESY spectrum of receptor
1d and 1000 equivalents of d-glucose (10) in
H2O/D2O 93:7 (600 MHz). The horizontal axis
covers the region 7.0–9.0 ppm, which includes
aromatic and NH protons of the receptor (for
atom labeling, refer to Figure 1). The vertical
axis covers 2.8–3.8 ppm, which includes pro-
tons 2–6 of glucose. The region plotted verti-
cally also includes protons F and E of the recep-
tor which are invisible in the 1D spectrum
owing to the disparity in concentrations. Intra-
molecular cross-peaks that involve these sig-
nals are framed by dotted lines (g). All other
cross-peaks represent intermolecular contacts.
For further details see Supporting Information.
Angew. Chem. Int. Ed. 2005, 44, 298 –302
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
301

Communications
[5] R. A. Dwek, T. D. Butters, Chem. Rev. 2002, 102, 283 (and
succeeding articles); C. R. Bertozzi, L. L. Kiessling, Science
2001, 291, 2357; T. Feizi, B. Mulloy, Curr. Opin. Struct. Biol. 2001,
11, 585 (and succeeding articles).
[6] A. P. Davis, R. S. Wareham, Angew. Chem. 1998, 110, 2397;
Angew. Chem. Int. Ed. 1998, 37, 2270.
[7] T. J. Ryan, G. Lecollinet, T. Velasco, A. P. Davis, Proc. Natl.
Acad. Sci. USA 2002, 99, 4863.
[8] C. M. Cardona, R. E. Gawley, J. Org. Chem. 2002, 67, 1411.
[9] This estimate is subject to large errors, and is therefore tentative.
[10] G. Lecollinet, A. P. Dominey, T. Velasco, A. P. Davis, Angew.
Chem. 2002, 114, 4267; Angew. Chem. Int. Ed. 2002, 41, 4093.
nonexistent. These data provide strong evidence that the
carbohydrate is bound in the interior of the macrotricyclic
structure. Both protons B and C appear to show cross-peaks
with all of the carbohydrate protons, but the cross-peaks
between protons B and proton H6 of glucose, and those
between proton C and the axial protons H3, H4, and H5, are
especially strong. Our proposed binding geometry, with
“axial!biphenyl” and “equatorial!isophthaloyl” orienta-
tions, thus receives some support.
In conclusion we have shown that, in the form of 1d, the
tricyclic cage 1 can bind carbohydrates in water. Affinities are
low, but selectivities are significant. The designed preference
for b-glucosyl which was previously demonstrated in organic
solvents is retained in the saccharidesꢀ natural environment.
The formation of complexes has been characterized by
1H NMR spectroscopy and confirmed by a second, independ-
ent technique, namely fluorescence spectroscopy. Tricycle 1 is
thus established as a carbohydrate receptor with consistent
behavior across a wide range of media. This property should
allow its use in studies which complement those on natural
receptors and further elucidate the principles underlying
biological carbohydrate recognition.
Received: July 23, 2004
Keywords: carbohydrates · hydrophobic effect · molecular
.
recognition · receptors · supramolecular chemistry
[1] a) A. P. Davis, R. S. Wareham, Angew. Chem. 1999, 111, 3160;
Angew. Chem. Int. Ed. 1999, 38, 2978; b) T. D. James, K. R. A. S.
Sandanayake, S. Shinkai, Angew. Chem. 1996, 108, 2038; Angew.
Chem. Int. Ed. Engl. 1996, 35, 1911; c) S. Striegler, Curr. Org.
Chem. 2003, 7, 81.
[2] F. A. Quiocho, Pure Appl. Chem. 1989, 61, 1293.
[3] For recent examples, see: W. Q. Yang, J. Yan, H. Fang, B. H.
Wang, Chem. Commun. 2003, 792; C. J. Ward, P. Patel, T. D.
James, Org. Lett. 2002, 4, 477; Z. L. Zhong, E. V. Anslyn, J. Am.
Chem. Soc. 2002, 124, 9014; J. N. Camara, J. T. Suri, F. E.
Cappuccio, R. A. Wessling, B. Singaram, Tetrahedron Lett. 2002,
43, 1139; H. S. Cao, D. I. Diaz, N. DiCesare, J. R. Lakowicz,
M. D. Heagy, Org. Lett. 2002, 4, 1503; W. Yang, H. He, D. G.
Drueckhammer, Angew. Chem. 2001, 113, 1764; Angew. Chem.
Int. Ed. 2001, 40, 1714.
[4] R. D. Hubbard, S. R. Horner, B. L. Miller, J. Am. Chem. Soc.
2001, 123, 5810; V. Krꢁl, O. Rusin, F. P. Schmidtchen, Org. Lett.
2001, 3, 873; N. Sugimoto, D. Miyoshi, J. Zou, Chem. Commun.
2000, 2295. For leading references to studies in nonaqueous
media, see ref [1a] and: T. Ishi-i, M. A. Mateos-Timoneda, P.
Timmerman, M. Crego-Calama, D. N. Reinhoudt, S. Shinkai,
Angew. Chem. 2003, 115, 2402; Angew. Chem. Int. Ed. 2003, 42,
2300; M. Segura, B. Bricoli, A. Casnati, E. M. Munoz, F.
Sansone, R. Ungaro, C. Vicent, J. Org. Chem. 2003, 68, 6296;
K. Wada, T. Mizutani, S. Kitagawa, J. Org. Chem. 2003, 68, 5123;
R. Welti, F. Diederich, Helv. Chim. Acta 2003, 86, 494; M. Mazik,
W. Radunz, W. Sicking, Org. Lett. 2002, 4, 4579; Y. H. Kim, J. I.
Hong, Angew. Chem. 2002, 114, 3071; Angew. Chem. Int. Ed.
2002, 41, 2947; K. Ladomenou, R. P. Bonar-Law, Chem.
Commun. 2002, 2108; J. H. Liao, C. T. Chen, H. C. Chou, C. C.
Cheng, P. T. Chou, J. M. Fang, Z. Slanina, T. J. Chow, Org. Lett.
2002, 4, 3107; J. Bitta, S. Kubik, Org. Lett. 2001, 3, 2637; S.
Tamaru, M. Yamamoto, S. Shinkai, A. B. Khasanov, T. W. Bell,
Chem. Eur. J. 2001, 7, 5270.
302
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 298 –302