A terphenyl scaffold for π-stacked guanidinium recognition elements

Article abstract of DOI:10.1021/ol7027042

(Chemical Equation Presented) A new synthetic model of arginine- carboxylate-aromatic triads-common motifs at sites of protein-protein interactions-is reported. Binding studies in mixed methanol/water solvent systems suggest that the carboxylate-binding a

Full text of DOI:10.1021/ol7027042

ORGANIC
LETTERS
2008
Vol. 10, No. 2
297-300
A Terphenyl Scaffold for
Guanidinium Recognition Elements
π-Stacked
Xing Wang, Olga V. Sarycheva, Bryan D. Koivisto, Aaron H. McKie, and
Fraser Hof*
Department of Chemistry, UniVersity of Victoria, P.O. Box 3065, Victoria, BC,
V8W 3V6 Canada
fhof@uVic.ca
Received November 7, 2007
ABSTRACT
A new synthetic model of arginine
−
carboxylate
−
aromatic triads
s
common motifs at sites of protein−protein interactionssis reported. Binding
studies in mixed methanol/water solvent systems suggest that the carboxylate-binding ability of
relative to a non-stacked control.
π-stacked guanidinium ions is improved
Arginine side chains are key mediators of a variety of protein
binding events, frequently forming charged hydrogen bonds
with phosphate, sulfate, and carboxylate binding partners.
Arginine’s guanidinium group also frequently interacts with
nearby aromatic side chains via π-stackingssurveys of
protein structures show that up to 74% of all arginine side
chains are in close contact with aromatic residues,1 and that
among these contact pairs the parallel π-stacked and offset-
stacked orientations are predominant.2-4 Arginines also
commonly participate in guanidinium-carboxylate-aromatic
triads in which both of the above modes of interaction operate
simultaneously. These motifs exist within monomeric pro-
teins and also as “hot-spots” of several protein-protein
interactions (Figure 1).5-8 Despite the importance of these
π-stacked salt bridges, the energetic influence of these two
interactions on each other is not well understood. Does
π-stacking increase or decrease the hydrogen bonding
potential of a guanidinium ion? Studies of guanidinium
hydrogen bonding and π-stacking individually are numerous,
but synthetic models for the study of guanidinium ions
participating in both hydrogen bonding and π-stacking are
few.9,10 We report herein the creation of a new synthetic
model in which guanidinium ions and aromatic rings are pre-
arranged in a π-stacked geometry and studies on its com-
plexation of carboxylate partners.
(1) Gallivan, J. P.; Dougherty, D. A. Proc. Natl. Acad. Sci. U.S.A. 1999,
96, 9459-9464.
(2) Flocco, M. M.; Mowbray, S. L. J. Mol. Biol. 1994, 235, 709.
(3) Brocchieri, L.; Karlin, S. Proc. Natl. Acad. Sci. U.S.A. 1994, 91,
9297-9301.
(4) Mitchell, J. B. O.; Nandi, C. L.; McDonald, I. K.; Thornton, J. M.;
Price, S. L. J. Mol. Biol. 1994, 239, 315.
(5) Peter B.; Crowley, P. B.; Golovin, A. Proteins: Struct. Funct. Bioinf.
2005, 59, 231-239.
Figure 1. Examples of guanidinium-aromatic-carboxylate triads
that mediate protein-protein interactions. (a) Residues at the
interface of a complex between Importin-â (blue) and Ran (orange)
(PDB code 1IBR).7 (b) Residues at the interface of the complex of
the Fyn SH3 domain (blue) and HIV-1 Nef (orange) (PDB code
1AVZ).8
(6) Inglis, S. R.; Stojkoski, C.; Branson, K. M.; Cawthray, J. F.; Fritz,
D.; Wiadrowski, E.; Pyke, S. M.; Booker, G. W. J. Med. Chem. 2004, 47,
5405-5417.
(7) Vetter, I. R.; Arndt, A.; Kutay, U.; Gorlich, D.; Wittinghofer, A.
Cell 1999, 97, 635.
(8) Arold, S.; Franken, P.; Strub, M. P.; Hoh, F.; Benichou, S.; Benarous,
R.; Dumas, C. Structure 1997, 5, 1361-1372.
10.1021/ol7027042 CCC: $40.75
© 2008 American Chemical Society
Published on Web 12/20/2007

Scheme 1. Synthesis of Hosts 3 and 4
The preorganization of two aryl groups into offset-stacked
geometries was previously accomplished using structures
such as 1 (Figure 2a).11-13 We used molecular modeling of
compound 2 to examine the effect of replacing one of the
aryl residues with a guanidinium ion (Figure 2b). The
covalent tether preserves the offset-stacked geometry (i) in
gas-phase energy minimizations (HF/6-31+G*).14 In a
second set of calculations, structure (i) was soaked in a 30
Å diameter droplet of explicit water molecules and the whole
structure was minimized at the MMFFaq level. The offset-
stacked geometry is preserved.
To allow experimental studies of this system, we designed
terphenyl receptor 3, which consists of two copies of this
stacked guanidinium-benzene motif on either side of a
central benzene spacer. Modeling of 3 in complex with
glutarate shows that both guanidinium ions can adopt the
stacked geometry of interest while each engaging one of the
guest’s carboxylate ions (Figure 2c). Comparing the behavior
of 3 with the binding of similar guests by control receptor
4,15 which lacks π-stacking elements but is predicted to
bind in a similar geometry, will provide insight into the
beneficial or deleterious effects of π-stacking on guani-
dinium-carboxylate interactions.
To synthesize this novel terphenyl-derived receptor, 2′,5′-
dimethyl-p-terphenyl 516 is first brominated using NBS,
giving bis(bromomethyl) terphenyl 6 in 80% yield (Scheme
(9) Thompson, S. E.; Smithrud, D. B. J. Am. Chem. Soc. 2002, 124,
442-449.
(10) Rensing, S.; Arendt, M.; Springer, A.; Grawe, T.; Schrader, T. J.
Org. Chem. 2001, 66, 5814-5821.
(11) Rashkin, M. J.; Waters, M. L. J. Am. Chem. Soc. 2002, 124, 1860-
1861.
(12) Rashkin, M. J.; Hughes, R. M.; Calloway, N. T.; Waters, M. L. J.
Am. Chem. Soc. 2004, 126, 13320-13325.
Figure 2. (a) The appended benzene rings of compound 1 are
known to adopt an offset-stacked geometry. (b) Calculations (see
text) predict that compounds such as 2 can also preorganize an
offset-stacked geometry (i) for a guanidinium-benzene pair. (c)
A model of receptor 3 (HF/6-31+G*) showing that each guani-
dinium binding element can simultaneously participate in π-stacking
and salt bridge formation when binding to glutarate.
(13) Martin, C. B.; Mulla, H. R.; Willis, P. G.; Cammers-Goodwin, A.
J. Org. Chem. 1999, 64, 7802-7806.
(14) Spartan ’04; Wavefunction, Inc.: Irvine, CA, 2004.
(15) Hamilton has conducted a thorough study of dicarboxylate binding
by a non-stacked receptor analogous to control host 4 but constructed using
closely related 2-amino-1,3-imidazoline binding elements. Linton, B. R.;
Goodman, M. S.; Fan, E.; van Arman, S. A.; Hamilton, A. D. J. Org. Chem.
2001, 66, 7313-7319.
(16) Merlet, S.; Birau, M.; Wang, Z. Y. Org. Lett. 2002, 4, 2157-2159.
298
Org. Lett., Vol. 10, No. 2, 2008

1). Conversion to the bis(azidomethyl) terphenyl 7 by
treatment with NaN3 is followed by LiAlH4 reduction to give
the diamine 8 (65% for two steps). Treatment with 1,3-bis-
(tert-butoxycarbonyl)-2-methyl-2-thiopseudourea (9) fol-
lowed by deprotection with TFA gives the bis(guanidinium)
terphenyl compound 3 as the TFA salt. Finally, the trifluo-
roacetate counterions are exchanged with Cl- by repeated
cycles of treatment with excess HCl (3 M in dioxane) and
evaporation (36% yield, three steps). The non-stacked control
compound 4 is similarly synthesized from p-xylylenediamine
in three steps with an overall yield of 29%.
Table 1. Binding Constants of Dicarboxylate Guests for Hosts
3 and 4 in Mixtures of CD3OD and D2O as Determined by H
NMR Titration
1
Compound 3 binds various dicarboxylate guests (tetrabu-
tylammonium glutarate (10), tetrabutylammonium Cbz-
glutamate (11), and tetrabutylammonium glutamate (12) were
investigated) in mixtures of CD3OD and D2O, as evidenced
by NMR titrations (Figure 3). In all cases, we obtain good
a Receiving solutions and titrants contained host at 1-3 mM. Titrant
solutions additionally contained guest at 30-50 mM. Shifts of host CH2
protons were tracked and fit to a 1:1 binding isotherm in order to determine
values for K. All K values are the result of 2-3 repetitions with an estimated
error of (10%. See Supporting Information for details. b Small or nonexist-
ent chemical shifts indicate binding is weaker than the lower limit for
determination by NMR (20 M-1).18
explained by the repulsive electrostatic influence and in-
creased desolvation potential of glutamate’s ammonium
functionality. The stacked guanidinium host 3 binds glutarate
(10) and Cbz-protected glutamate (11) 1.6-fold to 4.7-fold
more strongly than does non-stacked host 4. The weaker
association constants observed for glutamate (12) (150 M-1
for host 3 and 170 M-1 for host 4) agree within experimental
error. In the more competitive 50:50 CD3OD/D2O solvent
system, the stacked host again wins out, this time binding
Cbz-glutamate 1.5-fold more strongly and glutarate 20-fold
more strongly than non-stacked host 4.
What can account for the increased affinity of 3 for
carboxylate guests relative to 4? The preorganizing influence
of the newly introduced phenyl ring may contribute, but in
our hands, other diguanidinium hosts preorganized by
adjacent methyl groups do not show a significant effect
(unpublished data). Stereoelectronic considerations would
have the acidity of the guanidinium ions decreased by the
nearby electron density of the aromatic π-cloud, thereby
decreasing their hydrogen bonding ability;19,20 this effect,
though possibly a minor contributor in this system, runs
opposite to the trends observed here. The most likely
explanation is rooted in a solvation effect, in which the
nearby aromatic surface21 shields the salt bridge from
disruption by competitive solvent. In two prior experimental
Figure 3. 1H NMR titration data for the complexation of
(n-Bu4N+)2 glutarate by stacked host 3 (9, host concn ) 1.2 mM)
and non-stacked control host 4 (b, host concn ) 2.4 mM) in 90:
10 (v/v) CD3OD/D2O at 298 K. The lines represent fitted 1:1
binding isotherms. See the Supporting Information for details.
Inset: a Job plot (90:10 (v/v) CD3OD/D2O; [3] + [10] ) 5 mM;
T ) 298 K) demonstrates the formation of a 1:1 complex. The two
curves in the Job plot track the movement of two different signals
on host 3.
fits of titration data to a 1:1 binding isotherm, and analysis
by the method of continuous variation (Job plot) shows that
binding between host 3 and glutarate occurs in the proposed
1:1 stoichiometry (Figure 3 inset).17 The resulting equilibrium
constants are shown in Table 1 alongside values for the non-
stacked control receptor 4.15
(18) The exchangeable guanidinium protons that participate directly in
hydrogen bonding are not observed in the deuterated aqueous mixtures used
here. In these systems, the chemical shifts of the observable neighboring
methylene protons are inherently small even for strong binding events,
leading us to set a lower limit of 20 M-1 for binding constant determination
in this setting.
(19) Dvornikovs, V.; Smithrud, D. B. J. Org. Chem. 2002, 67, 2160-
2167.
(20) Inoue, Y.; Sugio, S.; Andzelm, J.; Nakamura, N. J. Phys. Chem. A
1998, 102, 646-648.
In 90:10 CD3OD/D2O, both hosts have dramatically lower
affinity for glutamate than for glutarate and Cbz-glutamate,
(21) The proximity of guanidinium and aromatic elements in 3 is
supported by modeling and by upfield chemical shifts (∼0.3 ppm in DMSO;
see Supporting Information) of guanidinium NH protons in 3 relative to
those in 4. Determining the exact structure of the guanidinium-aromatic
interactions of 3 and related compounds in aqueous media will be the focus
of future work.
(17) This does not rule out the formation of oligomeric n:n complexes,
but the binding interactions of a single carboxylate group to receptor 3 on
which oligomer formation would rely is extremely weak (K < 10 M-1) in
the methanol/water mixtures employed here. We assume that the contribution
of non-1:1 structures to the observed binding isotherms is negligible.
Org. Lett., Vol. 10, No. 2, 2008
299

studies of guanidinium-carboxylate9 and ammonium-
carboxylate19 pairs near aromatic surfaces in mixed organic/
aqueous solvents, the presence of an aromatic surface has
also driven the formation of a stronger salt bridge. However,
hydration effects are greatly perturbed by organic cosolvents
such as those used here and in other studies.9,15,22,23 Com-
pound 3 is unable to bind dicarboxylate guests in pure water,
so our study leaves the role of hydration in the formation of
guanidinium-carboxylate-aromatic triads open to discus-
sion. The current results provide a new model of these
important motifs, and we are now pursuing variations on
this synthetic model system that will allow us to carry out
studies in the medium of lifeswarm, salty water.
Acknowledgment. This research was supported by the
University of Victoria and NSERC. A.H.M. is a Pacific
Century Scholar. F.H. acknowledges a Career Scholar Award
from the Michael Smith Foundation for Health Research.
Supporting Information Available: Synthetic proce-
dures, characterization of new compounds, and a detailed
description of NMR titrations. This material is available free
of charge via the Internet at http://pubs.acs.org.
(22) Marcus, Y. Chem. ReV. 2007, 107, 3880-3897.
(23) Kalidas, C.; Hefter, G.; Marcus, Y. Chem. ReV. 2000, 100, 819-
852.
OL7027042
300
Org. Lett., Vol. 10, No. 2, 2008