Hormone-PAMAM dendrimer conjugates: Polymer dynamics and tether structure affect ligand access to receptors

Article abstract of DOI:10.1002/anie.200601923

(Figure Presented) An open and closed case: Estradiol was conjugated to a PAMAM dendrimer (gray circle) through a short (17α-phenylethynyl) or a long (hexaethyleneglycol) tether to evaluate how ligand access to a receptor is affected. With the short tethe

Full text of DOI:10.1002/anie.200601923

Angewandte
Chemie
Dendrimers
DOI: 10.1002/anie.200601923
Hormone–PAMAM Dendrimer Conjugates:
Polymer Dynamics and Tether Structure Affect
Ligand Access to Receptors**
Sung Hoon Kim and John A. Katzenellenbogen*
The development of ligand–polymer conjugates as diagnostic
and therapeutic agents raises an interesting question: In what
way do the conformational and dynamic features of the
macromolecular carrier influence ligand access to its target
receptor? Poly(amide)polyamine (PAMAM) dendrimers, in
particular, are being widely adapted to biomedical applica-
[*] Dr. S. H. Kim, Prof. J. A. Katzenellenbogen
Department of Chemistry
University of Illinois
600 South Mathews Avenue, Urbana, IL 61801 (USA)
Fax: (+1)217-333-7325
E-mail: jkatzene@uiuc.edu
[**] We are grateful for support of this research through grants from the
National Institutes of Health (PHS 5R37 DK15556). We thank
Kathryn Carlson for determining the receptor binding affinities.
NMR spectra were obtained in the Varian Oxford Instrument Center
for Excellence in NMR Laboratory. Funding for this instrumentation
was provided in part by the W. M. Keck Foundation, the National
Institutes of Health (PHS 1 S10 RR104444-01), and the National
Science Foundation (NSF CHE 96-10502). Mass spectra were
obtained on instruments supported by grants from the National
Institute of General Medical Sciences (GM 27029), the National
Institutes of Health (RR 01575), and the National Science Founda-
tion (PCM 8121494). PAMAM=poly(amide)polyamine.
Supporting information for this article (details of the preparation
and characterization of estrogen–dendrimer conjugates) is avail-
able on the WWW under http://www.angewandte.org or from the
author.
Angew. Chem. Int. Ed. 2006, 45, 7243 –7248
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
7243

Communications
tions,[1–5] yet little is known about how their complex, pH-
The EDCs were prepared by reductive amination of the
appropriate estrogen aryl carboxaldehyde with the primary
amine termini of the G6 PAMAM. Imine formation pro-
ceeded spontaneously, and reduction with borohydride was
quantitative; these steps could be followed by 1H NM R
spectroscopy. The final ligand:dendrimer ratio (18:1–20:1,
Scheme 1), determined by MALDI MS, simply reflected the
reaction stoichiometry. The resulting EDCs were purified by
ultrafiltration.[9]
Although conformationally mobile, G6 PAMAMs are
believed to be roughly spherical up to the generation five
layer.[6,7] Conformation and flexibility are predicted to be pH-
dependent,[6,8] with low pH values favoring rigid, extended
forms (because of Coulombic repulsion) and high pH values
producing more dense compact forms. At neutral pH, the
peripheral amine groups are partially protonated, with the
outer layers constituting a relatively flexible framework
having pockets,[6–8] a characteristic through which PAMAMs
can be used as drug and gene delivery systems.[14,15] In our
EDCs we have added a peripheral hydrophobic group
(estrogen) to a macromolecule that has a relatively hydro-
philic surface, but which is also porous with a more hydro-
phobic interior. Thus, the disposition of the ligand with
respect to the surface—Is it extended outward and thus
dependent conformation and flexibility[6–8] affects the acces-
sibility of covalently attached ligands. We have used spectro-
scopic and physical methods to characterize ligand dynamics
and access to the receptor in four estrogen–dendrimer
conjugates (EDCs) that were prepared to study pathways of
estrogen signaling.[9] Curiously, EDCs having long or hydro-
phobic tethers engender considerable ligand shielding that
results in poor access to the receptor, whereas those with
short tethers expose the ligand so it binds to the receptor with
little impediment from the dendrimer carrier.
The four EDCs (Scheme 1) are based on a generation six
(G6) PAMAM dendrimer, with a molecular weight of
approximately 58000,[1] comparable to that of bovine serum
albumin (67000) used to construct other estrogen–protein
conjugates.[10,11] Three are linked to estradiol, either through
17a-ethynyl (Ia and IIa) or 7a sites (IVa), which are known to
tolerate substitution,[12] while the fourth is based on the
nonsteroidal ligand cyclofenil (IIIa).[13] The linking tethers are
short (Ia and IIIa), moderate (IVa), or long (IIa, hexaethylene
glycol). Four compounds (Ib–IVb) in which the tether chain
was terminated by an (acetamido)ethylamine function, which
mimicks the first dendrimer attachment site, were prepared
for reference.
Scheme 1. Estrogen–dendrimer conjugates (EDCs, Ia–IVa) and reference compounds (Ib–IVb).
7244 www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 7243 –7248

Angewandte
Chemie
available to a receptor, or is it buried within the G6 layer and
inaccessible to other proteins?—becomes an important issue.
NOESY spectra show interactions between the ligand and
the dendrimer backbone (and sometimes tether). In both the
short- and long-tether EDCs the aromatic signals (C-1, 2, 4;
a–d) show strong cross-peaks in D2O with 4/5 resonances that
can be assigned to the dendrimer backbone (f–k); these cross-
peaks are very weak in methanol (Figures 1 and 2, top versus
bottom spectrum). Significantly, the long-tether EDC in D2O
(IIa, Figure 2, bottom) shows there are strong interactions
between the resonances of the ethylene glycol spacer (eg) and
the aromatic resonances of the ligand (C-1, 2, 4), even though
1
1
Figure 1. Aliphatic and aromatic region of the H NMR NOESY spectra
Figure 2. Aliphatic and aromatic region of the H NMR NOESY spectra
of short-tether EDC Ia in CD3OD (top) and in D2O (bottom).
of long-tether EDC IIa in CD3OD (top) and in D2O (bottom).
Angew. Chem. Int. Ed. 2006, 45, 7243 –7248
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7245

Communications
they are at opposite ends of the ligand. Thus, in an aqueous
CD3OD to D2O, but the decrease was more pronounced for
the long-tether EDC (IIa). The T2 values for the monomeric
ligand analogues (Ib and IIb) are, as expected, much larger.
The rotational correlation times of the C-4 proton on the
A ring of the steroid, calculated from the T1 and T2 values,[20]
show that the tumbling motion of two monomeric ligands (Ib
and IIb) in CD3OD is very similar, whereas the tumbling
motion of the two EDCs showed a twofold difference in
methanol, but only a 1.3-fold difference in D2O. We interpret
the effect of solvent on the rotational correlation times to
mean the following: in CD3OD, the size difference between
the short- and long-tether EDCs is pronounced because the
long-tether EDC (IIa) has an
environment, but not in methanol, the ligand portion of the
long-tether EDC becomes enwrapped by the ethylene glycol
tether and buried within the dendrimer with sufficient
stability to generate large NOE signals. Although oligo-
ethylene glycol derivatives are water compatible and flexible,
they have a coiled structure, engendered by the gauche
preference of the glycol unit,[16–19] which probably contributes
to this ligand-tether-dendrimer interaction.
Spin–lattice and spin–spin relaxation times T1 and T2 for
the C-4 proton of the EDCs Ia and IIa, as well as their
reference compounds Ib and IIb, are given in Figure 3. The
extended structure; by contrast,
in water, the ligand of the long-
tether EDC becomes encapsu-
lated in the dendrimer interior,
so its molecular size is less
different from that of the
short-tether EDC.
To further evaluate the ap-
parent size of the EDCs, we
determined the hydrodynamic
diameter of the G6 PAMAM
and the EDCs in water and
20% aqueous methanol solu-
tion by dynamic light scattering
(DLS; Figure 3). In all cases,
the diameters are smaller in
water than in aqueous metha-
nol, but the two EDCs show
very different solvent effects.
The radius of the short-tether
EDC (Ia) in both solvents is
nearly the same as the sum of
the PAMAM radius and the
length of the ligand moiety
(ca. 1.5 nm, determined by
molecular modeling studies).
By contrast, the radius of the
long-tether EDC (IIa) in aque-
ous MeOH is much larger, con-
sistent with a stretched-out con-
formation, whereas in water its
radius is much smaller, similar
to that of the short-tether EDC.
Thus, the short-tether EDC (Ia)
Figure 3. 1H NMR relaxation and rotational correlation times (in seconds) and hydrodynamic diameter
measurements for the short-tether (Ia) and long-tether (IIa) EDCs: a) EDC Ia in water, b) EDC Ib in 20%
MeOH in water, c) EDC IIa in water, d) EDC IIb in 20% MeOH in water. The diameters, measured by
dynamic light scattering studies, are shown schematically; the inner circle represents the PAMAM and the
outer circle the EDC diameter. The spin–lattice and spin–spin relaxation times (T1 and T2) and the
rotational correlation time (tc) are given for the C-4 proton of the 17a-ethynylestradiol in the EDCs and
control compounds. The abbreviated structure of the ligand is meant to represent the degree to which the
ligand is extended or enveloped relative to the PAMAM.
T1 value for the C-4 ligand proton in the long-tether EDC
(IIa) in D2O is considerably less than that in CD3OD; in the
latter solvent, the T1 value is very similar to that of the
monomeric ligand analogue IIb. In contrast, there was little
difference in the T1 value for the short-tether EDC (Ia) in the
two solvent systems, both values being similar to those of the
monomeric ligand analogue Ib. This finding suggests that in
hydrophilic solvents, only the ligand in the long-tether EDC
(IIa) undergoes a change of environment, which causes its
motion to become more restricted (increasing spin–lattice
relaxation; smaller T1 values). The T2 values of both the
short- and long-tether EDCs decrease upon shifting from
appears to have a more rigid structure, with the ligand
projecting outward from the peripheral region of the dendri-
mer in a relatively solvent-insensitive manner. By contrast,
the long linker in EDC IIa allows the hydrophobic ligand to
reach back into the interior region of the PAMAM in aqueous
media.
The hydrodynamic diameters of the other two EDCs (IIIa
and IVa) were approximately 8.6 nm and 45 nm in water,
respectively. The cyclofenil-functionalized EDC (IIIa), with a
short linker, has a radius close to that of the dendrimer plus
the length of the cyclofenil moiety (1.4 nm), as was the case
with Ia. The unexpectedly large diameter measured for the
7246 www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 7243 –7248

Angewandte
Chemie
Table 1: Relative binding affinity (RBA) values for the EDCs and reference compounds for ERa and ERb.
EDC or reference compound
I
II
III
IV
a
b
a
b
a
b
a
b
RBA[a]
3.8Æ0.8 3.6Æ 0.2
1.9Æ0.3 2.3Æ0.4
83–106%
0.9Æ0.1
3.3Æ0.5
2.9Æ0.2
2.8Æ0.3
2.4Æ0.3
10.3Æ0.4
(ERa)
RBA[a]
0.2Æ0.1
1.2Æ0.4
5.8Æ0.1 10.2Æ0.4
2.7Æ0.3
14.4Æ3.8
(ERb)
ligand–receptor access ratio[b]
nature of tether explanation
17–27%
long
60–104%
short
19–23%
short
medium; hydrophobic
ligand access restricted by
EDC aggregation
Good ligand–receptor ligand access restricted by encap- good ligand–receptor
access sulation by PAMAM and tether access
[a] RBA=relative binding affinity determined in competitive radiometric binding assays.[21,22] RBA={IC50[estradiol]/IC50[compound]}ꢀ100. Values are
the mean Æ range or SD of 2 or more independent experiments. The RBA for estradiol is 100; Kd value for estradiol is 0.2 nm for ERa and 0.5 nm for
ERb. [b] Ligand–receptor access ratio is defined as {RBA[EDC]/RBA[reference compound]}ꢀ100.
EDC IVa, with a moderate length but hydrophobic penta-
methylene group linker, is most likely the result of aggrega-
tion.
of the ligand to the receptor. Our findings hold important
implications for the future design of drug– or hormone–
polymer conjugates where receptor interaction by the poly-
mer-bound ligand is the goal, and our investigation provides
approaches that can be used to evaluate the behavior of new
ligand–polymer conjugates. We believe that these principles
will help ensure success in the future development of such
novel polymer-based biological reagents.
We determined the ligand access to the receptor in all four
EDCs by comparing their binding affinities for the estrogen
receptors with those of their reference compounds, which
were designed to mimic the EDC ligand in all respects except
not having the dendrimer attached (Scheme 1). Binding
affinities, determined by a known method,[21,22] are expressed
in Table 1 as relative binding affinity (RBA) values, where the
affinity of estradiol is set at 100. The values for the EDCs are
based on the ligand-equivalent concentration, which corrects
for the fact that multiple ligands are attached to each
dendrimer.
All of the compounds have RBA values in the range 0.2–
14.4. Most notably, the short-tether EDCs (Ia, IIIa) have
RBA values for the estrogen receptors (ERs) that are very
close to those of the corresponding model compounds (Ib and
IIIb). In contrast, both the long and moderate tether length
EDCs (IIa and IVa) have much lower affinities than the
reference compounds. The ratio of EDC to reference com-
pound affinity, expressed in percent, can be termed the
“ligand–receptor access index” and provides a direct mea-
surement of ligand access to the receptor. The two short-
tether EDCs Ia and IIIa have ligand–receptor access indices
of 83–106% and 60–104%, respectively, whereas the EDCs
with either long- or medium-length tethers (IIa and IVa) have
indices of only 17–27% and 19–23%, respectively. Thus, as
expected from the spectroscopic and hydrodynamic measure-
ment studies, the short-tether EDCs (Ia and IIIa) provide
much better ligand access to the receptor than do the longer
tether EDCs (IIa and IVa), for the reasons summarized in the
Table 1.
Received: May 15, 2006
Revised: August 24, 2006
Published online: October 6, 2006
Keywords: dendrimers · estrogen · hormones · receptors ·
.
solvent effects
[1] R. Esfand, D. A. Tomalia, Drug Discovery Today 2001, 6, 427.
[2] C. C. Lee, J. A. MacKay, J. H. Frechet, F. C. Szoka, Nat.
Biotechnol. 2005, 23, 1517.
[3] T. D. McCarthy, P. Karellas, S. A. Henderson, M. Giannis, D. F.
OꢀKeefe, G. Heery, J. R. Paull, B. R. Matthews, G. Holan, Mol.
Pharm. 2005, 2, 312.
[4] S. Svenson, D. A. Tomalia, Adv. Drug Delivery Rev. 2005, 57,
2106.
[5] T. P. Thomas, I. J. Majoros, A. Kotlyar, J. F. Kukowska-Latallo,
A. Bielinska, A. Myc, J. R. Baker, Jr., J. Med. Chem. 2005, 48,
3729.
[6] I. Lee, B. D. Athey, A. W. Wetzel, W. Meixner, J. R. J. Baker,
Macromolecules 2002, 35, 4510.
[7] P. K. Maiti, T. Cagin, G. Wang, W. A. I. Goddard, Macromole-
cules 2004, 37, 6236.
[8] W. Chen, D. A. Tomalia, J. L. Thomas, Macromolecules 2000, 33,
9196.
[9] W. R. Harrington, S. H. Kim, C. C. Funk, Z. Madak-Erdogan, R.
Schiff, J. A. Katzenellenbogen, B. S. Katzenellenbogen, Mol.
Endocrinol. 2006, 20, 491.
[10] P. E. Stevis, D. C. Deecher, L. Suhadolnik, L. M. Mallis, D. E.
Frail, Endocrinology 1999, 140, 5455.
[11] J. L. Temple, S. Wray, Endocrinology 2005, 146, 558.
[12] G. M. Anstead, K. E. Carlson, J. A. Katzenellenbogen, Steroids
1997, 62, 268.
[13] R. S. Muthyala, S. Sheng, K. E. Carlson, B. S. Katzenellenbogen,
J. A. Katzenellenbogen, J. Med. Chem. 2003, 46, 1589.
[14] S. Kannan, P. Kolhe, V. Raykova, M. Glibatec, R. M. Kannan, M.
Lieh-Lai, D. Bassett, J. Biomater. Sci. Polym. Ed. 2004, 15, 311.
[15] C. Kojima, K. Kono, K. Maruyama, T. Takagishi, Bioconjugate
Chem. 2000, 11, 910.
We have shown that tethering ligands to polymeric
macromolecules can significantly affect ligand access to a
receptor. In our EDCs, where the ligand is hydrophobic (for
example, an estrogen) and is attached to a macromolecule
having a flexible structure with surface invaginations (for
example, PAMAM), ligand access becomes a function of the
tether: Short-tether EDCs are able to maintain ligand
exposure, thus providing essentially unimpeded access to
the ERs; in contrast, ligands attached through long or
hydrophobic tethers experience masking by burrowing into
the PAMAM or by aggregation and thus result in poor access
Angew. Chem. Int. Ed. 2006, 45, 7243 –7248
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7247

Communications
[16] H. J. Kreuzner, R. L. C. Wang, M. Grunze, New J. Phys. 1999, 1,
21.1.
[17] S. Wolfe, Acc. Chem. Res. 1972, 5, 102.
[18] K. B. Wiberg, M. A. Murcko, K. E. Laidig, P. MacDougall, J.
Phys. Chem. 1990, 94, 6956.
[19] M. Sundaralingam, Nature 1968, 217, 35.
[20] H. Pfeifer, NMR: Basic Principles and Progress, Vol. 7, Springer,
New York, 1972.
[21] K. E. Carlson, I. Choi, A. Gee, B. S. Katzenellenbogen, J. A.
Katzenellenbogen, Biochemistry 1997, 36, 14897.
[22] J. A. Katzenellenbogen, H. J. Johnson, Jr., K. E. Carlson, Bio-
chemistry 1973, 12, 4092.
7248 www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 7243 –7248