Generation and in situ evaluation of libraries of poly(acrylic acid) presenting sialosides as side chains as polyvalent inhibitors of influenza- mediated hemagglutination

Article abstract of DOI:10.1021/ja963519x

This paper describes a simple, microscale method for generating and evaluating libraries of derivatives of poly(acrylic acid) (pAA) that present mixtures of side chains that influence their biological activity. The method is based on the one-step conversion of poly(acrylic anhydride) (pAAn) to linear polymers presenting multiple units of R on side chains, pAA(R): the polymers are obtained by ultrasonication of a suspension of pAAn and aqueous RNH₂ contained in a 250-μL well of a microtiter plate. Using this method, derivatives of pAA having N-acetylneuraminic acid (NeuAc-L-NH₂) as a side chain, pAA(NeuAc-L), were generated and assayed for ability to inhibit hemagglutination (HAI) of chicken erythrocytes by influenza virus A (X-31); the constant (K(i)/(HAI)) describing this inhibition is calculated on the basis of the concentration of NeuAc groups in solution, rather than the concentration of polymer molecules. Co-polymeric pAA(NeuAc-L(n);L(n) = different linking groups) with a range of mole fractions of NeuAc-L-NH₂ (χ(NeuAc-L) = 0.02-0.11) exhibited HAI activities with K(i)/(HAI) values between 27 and 0.30μM. Using combinations of NeuAc-L-NH₂ and one of 26 different primary amines RNH₂, a variety of ter-polymeric pAA(NeuAc-L; R) (χ(NeuAc-L) ~ 0.05; χ(R) ~ 0.06) were also generated and assayed. Certain terpolymers yielded values of K(i)/(HAI) that were lower by a factor of ~10⁴ than that of the parent co-polymeric pAA(NeuAc-L): the most active inhibitor was pAA(NeuAc-L; L-3-(2'-naphthyl)alanine)) (K(i)/(HAI) ? 0.5 nM). Typically, the incorporation of hydrophobic-especially aromatic-side chains enhanced activities. These polymers (pAA(NeuAc-L; R)) belong to a new class of polymeric, polyvalent sialosides that are potent inhibitors of the adsorption of influenza virus to erythrocytes. They were active with only low to moderate levels of incorporation of functional groups into the side chains: χ(NeuAc)-L ~ 0.05; χ(R) ~ 0.06.

Full text of DOI:10.1021/ja963519x

J. Am. Chem. Soc. 1997, 119, 4103-4111
4103
Generation and in Situ Evaluation of Libraries of Poly(acrylic
acid) Presenting Sialosides as Side Chains as Polyvalent
Inhibitors of Influenza-Mediated Hemagglutination
Seok-Ki Choi, Mathai Mammen, and George M. Whitesides*
Contribution from the Department of Chemistry and Chemical Biology, HarVard UniVersity,
12 Oxford Street, Cambridge, Massachusetts 02138
ReceiVed October 8, 1996X
Abstract: This paper describes a simple, microscale method for generating and evaluating libraries of derivatives of
poly(acrylic acid) (pAA) that present mixtures of side chains that influence their biological activity. The method is
based on the one-step conversion of poly(acrylic anhydride) (pAAn) to linear polymers presenting multiple units of
R on side chains, pAA(R): the polymers are obtained by ultrasonication of a suspension of pAAn and aqueous
RNH2 contained in a 250-µL well of a microtiter plate. Using this method, derivatives of pAA having
N-acetylneuraminic acid (NeuAc-L-NH2) as a side chain, pAA(NeuAc-L), were generated and assayed for ability to
inhibit hemagglutination (HAI) of chicken erythrocytes by influenza virus A (X-31); the constant (KHi AI) describing
this inhibition is calculated on the basis of the concentration of NeuAc groups in solution, rather than the concentration
of polymer molecules. Co-polymeric pAA(NeuAc-Ln; Ln ) different linking groups) with a range of mole fractions
of NeuAc-L-NH2 (øNeuAc-L ) 0.02-0.11) exhibited HAI activities with KHi AI values between 27 and 0.30 µM.
Using combinations of NeuAc-L-NH2 and one of 26 different primary amines RNH2, a variety of ter-polymeric
pAA(NeuAc-L; R) (øNeuAc-L ∼ 0.05; øR ∼ 0.06) were also generated and assayed. Certain ter-polymers yielded
values of KHi AI that were lower by a factor of ∼104 than that of the parent co-polymeric pAA(NeuAc-L): the most
active inhibitor was pAA(NeuAc-L; L-3-(2′-naphthyl)alanine)) (KHi AI ≈ 0.5 nM). Typically, the incorporation of
hydrophobicsespecially aromaticsside chains enhanced activities. These polymers (pAA(NeuAc-L; R)) belong to
a new class of polymeric, polyvalent sialosides that are potent inhibitors of the adsorption of influenza virus to
erythrocytes. They were active with only low to moderate levels of incorporation of functional groups into the side
chains: øNeuAc-L ∼ 0.05; øR ∼ 0.06.
Introduction
of new ter-polymers active in this assay at concentrations less
of than 1 nM in sialic acid groups.
We describe a convenient microscale strategy that allows
rapid generation and in situ biological evaluation of libraries
of polymers and polymer sequences based on poly(acrylic acid)
(pAA) that present a variety of functional groups as amide side
chains.1 The use of the word “library” here differs slightly from
that common in combinatorial synthesis of small molecules: it
refers to different sets of sequences and configurations along
polymer backbones, rather than to different, uniquely defined
chemical compounds. The method is based on the conversion
of poly(acrylic anhydride) (pAAn) to derivatives of pAA by
reaction with various amines RNH2 in water. We prepare these
derivatized polymers by ultrasonication of the reactants directly
in the wells of a microtiter plate; we then assay them in the
same plate with no further manipulation. We illustrate the
application of this methodology by screening derivatives of pAA
presenting multiple copies of N-acetylneuraminic acid (NeuAc
or sialic acid)2 as polyvalent inhibitors of the agglutination of
erythrocytes induced by influenza virus A and by the discovery
Influenza initiates infection by adsorption to the surface of
mammalian cells through the multiple interactions of the viral
protein hemagglutinin (HA) and clusters of N-acetylneuraminic
acid (NeuAc; Figure 1) moieties expressed on the cellular
surface.3-5 HA is a major surface glycoprotein of the virus
and exists as a homotrimeric complex (HA3; MW ∼ 225 kDa):
each subunit of the complex contains a binding site for NeuAc
at its outermost portion.6-8 The virus presents approximately
200-300 copies of HA3 on the surface; this number constitutes
∼80-90% of the total protein on the surface. Influenza also
presents another surface protein, neuraminidase (NA or siali-
dase), which is an enzyme capable of catalyzing the hydrolysis
of the R-O-glycosidic linkage of NeuAc connected to a variety
of aglycons on the cellular surface.9,10 NA exists as a
homotetrameric complex (NA4; MW ∼ 240 kDa) on the surface
of the virion (∼20-40 copies of NA4 per virion) and constitutes
(2) Corfield, A. P.; Schauer, R. In Sialic Acids, Chemistry, Metabolism
and Function; Schauer, R., Ed.; Springer Verlag: Wien, New York, 1982;
Vol. 10, pp 5-39.
(3) Voet, D.; Voet, J. Biochemistry; John Wiley & Sons, Inc.: New York,
1995.
* Author to whom correspondence should be addressed. Telephone:
(617)-495-9430. FAX: (617)-495-9857. E-mail: gwhitesides@gmwgroup.
harvard.edu.
X Abstract published in AdVance ACS Abstracts, May 1, 1997.
(1) For reviews of biomedical applications of synthetic polymers, see:
(a) In Polymers of Biological and Biomedical Significance (ACS Symposium
Series 540); Shalaby, S. W., Ikada, Y., Langer, R., Williams, J., Eds.;
American Chemical Society: Washington, DC, 1994. (b) Tsuruta, T. In
Biopolymers Liquid Crystalline Polymers Phase Emulsion (AdVances In
Polymer Science); Tsuruta, T., O’Brien, D. E., Armitage, B. A., Bennett,
D. E., Lamparski, H. G., Sato, T., Teramoto, A., Cameron, N. R.,
Sherrington, D. C., Eds.; Springer Verlag: Berlin, Heidelberg, 1996; Vol.
126, pp 1-52.
(4) Kaplan, M. M.; Webster, R. G. Sci. Am. 1977, 237, 88.
(5) Paulson, J. C. In The Receptors; Conn, P. M., Ed.; Academic Press:
Orlando, FL, 1985; Vol. 2, pp 131-219.
(6) Wiley, D. C.; Skehel, J. J. Ann. ReV. Biochem. 1987, 56, 365.
(7) Weis, W.; Brown, J. H.; Cusack, S.; Paulson, J. C.; Skehel, J. J.;
Wiley, D. C. Nature 1988, 333, 426.
(8) Cusack, S.; Ruigrok, R. W.; Krygsman, P. C. J.; Mellema, J. E. J.
Mol. Biol. 1985, 186, 565.
(9) Air, G. M.; Laver, W. G. Proteins 1989, 6, 341.
(10) Colman, P. M. Protein Sci. 1994, 3, 1687.
S0002-7863(96)03519-6 CCC: $14.00 © 1997 American Chemical Society

4104 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Choi et al.
virus and the ability of the adsorbed polymers to form gel layers
that inhibit the approach of virus to the erythrocyte sterically
(steric stabilization).21-23 Practically, the effectiveness of
polymers in inhibiting hemagglutination may be a function of
multiple variables, including the structure and net charge of
individual NeuAc moieties and non-sialoside side chains, the
mole fraction of NeuAc-containing side chains in the polymer
(øNeuAc), and the length and the flexibility of the polymer
chains.21-23
Here, we describe a simple strategy that streamlines the
generation of derivatives of pAA and the evaluation of biological
activities of these polymers by carrying out both synthesis and
assay in the wells of microtiter plates. The method allows
convenient screening of libraries25 of polymers and polymer
sequences presenting multiple combinations of side chains at
controlled mole fractions. Since microtiter plate assays are
routine in biology and medical sciences,26 this method may serve
generally for screening and obtaining leads for a range of
agglutination interactions and other processes that might be
influenced by polyvalent inhibitors.27
Results and Discussion
Figure 1. Structures of N-acetylneuraminic acid (NeuAc, or sialic acid),
and synthetic R-C- and R-O-sialosides.
Method of Generation of Libraries of Polymer. Figure 2
describes our strategy for rapid and efficient generation of
libraries of potentially bioactive polymers. We synthesized
poly(acrylic acid) having multiple R groups as side chains, pAA-
(R), by sonicating a suspension (0.12 mg/µL) of poly(acrylic
anhydride)28,29 (pAAn) and an aqueous solution of an amine
RNH2 (0.1 M) contained in a 250-µL well of a microtiter plate
(Figure 2). We characterized the resulting pAA(R) by examin-
ing 1H-NMR (D2O) spectra of lyophilized reaction mixtures as
both a crude mixture and a purified form (following dialysis;
∼10% of the total protein on the surface. This enzyme is
usually considered to be unimportant in binding to target cells,
but perhaps involved in allowing newly assembled virions to
escape from the host cell and in helping the virus to escape
from NeuAc-containing, protective proteins (such as R2-mac-
roglobulin11)11 in the mucosal secretions of the respiratory tract.
Monomeric sialosides are, in general, weak inhibitors of the
adsorption of virus (as measured by the minimum concentration
required to inhibit hemagglutination, KHi AI ∼ 10-3 M),12,13 but
multivalent sialosides that present multiple copies of sialosides
tethered to liposomes,14,15 dendrimers,16 glycoproteins,17-19 or
1
MW cutoff of ∼3.5 kDa). The H-NMR signals of R from
pAA(R) were distinguished readily from those of free, unreacted
RNH2 by their shape (the lines due to polymer-attached species
are relatively broad) and by their chemical shift (the δ values
of CH2 or CH groups adjacent to the amide group are shifted
downfield relative to those adjacent to the amine group).
Molecular weights of purified pAA prepared using this method
were estimated from gel permeation chromatography using
polysaccharide standards: Mw ) 39.5 kDa; polydispersity (Mw/
Mn) ) 1.91. We emphasize that our strategy is designed for
rapid discovery of promising leads of bioactivity; once a polymer
lead, pAA(R′), becomes known, one might synthesize the same
pAA(R′) by use of other, more conventional methods of polymer
synthesis. Further purification and fractionation may be neces-
sary to yield an active polymer with a narrow molecular weight
distribution.30
polymers17,20-24 can be highly effective inhibitors (KHi AI
∼
10-5 - 10-11 M; expressed in terms of sialic acid groups). In
particular, derivatives of poly(acrylamide) presenting NeuAc
as side chains are now well established as potent inhibitors of
hemagglutination.20-24 These compounds are more effective
than the most effective natural inhibitors (for example, equine
R2-macroglobulin) by factors up to ∼103.21 We have ascribed
the high activities of these polyvalent sialosides to both a high-
affinity binding of the polymers to multiple receptors on the
(11) Pritchett, T. J.; Paulson, J. C. J. Biol. Chem. 1989, 264, 9850.
(12) Sauter, N. K.; Bednarski, M. D.; Wurzburg, B. A.; Hanson, J. E.;
Whitesides, G. M.; Skehel, J. J.; Wiley, D. C. Biochemistry 1989, 28, 8388.
(13) Sauter, N. K.; Hanson, J. E.; Glick, G. D.; Brown, J. H.; Crowther,
R. L.; Park, S.-J.; Skehel, J. J.; Wiley, D. C. Biochemistry 1992, 31, 9609.
(14) Kingery-Wood, J. E.; Williams, K. W.; Sigal, G. B.; Whitesides,
G. M. J. Am. Chem. Soc. 1992, 114, 7303.
(15) Spevak, W.; Nagy, J. O.; Charych, D. H.; Schaefer, M. E.; Gilbert,
J. H. J. Am. Chem. Soc. 1993, 115, 1146.
(16) Roy, R.; Tropper, F. D. J. Chem. Soc., Chem. Commun. 1988, 1058.
(17) In Methods in Enzymology (Neoglycoconjugates. Part B. Biomedical
Applications); Lee, Y. C., Lee, R. T., Eds.; Academic Press, Inc.: San Diego,
CA, 1994; Vol. 247.
(18) Roy, R.; Laferriere, C. A. Can. J. Chem. 1990, 68, 2045.
(19) Sabesan, S.; Duus, J. Ø.; Neira, S.; Domaille, P.; Kelm, S.; Paulson,
J. C.; Bock, K. J. Am. Chem. Soc. 1992, 114, 8363.
(20) Spaltenstein, A.; Whitesides, G. M. J. Am. Chem. Soc. 1991, 113,
686.
(21) Mammen, M.; Dahmann, G.; Whitesides, G. M. J. Med. Chem. 1995,
38, 4179.
(22) Choi, S.-K.; Mammen, M.; Whitesides, G. M. Chem. Biol. 1996, 3,
97.
(23) Sigal, G. B.; Mammen, M.; Dahmann, G.; Whitesides, G. M. J.
Am. Chem. Soc. 1996, 118, 3789.
By comparing the integrated intensity of NMR signals of
reacted RNH2 (as RNH-CO- of the polymer) relative to that
of unreacted free RNH2, we estimated the yields (Figure 2) of
incorporation of RNH2. Figure 3 summarizes selected results
with 4-aminobenzoic acid, 6-amino-1-hexanoic acid, and NeuAc-
(25) For an example of a combinatorial approach towards the construction
of libraries of polymeric catalysts, see: Menger, F. M.; Eliseev, A. V.;
Migulin, V. A. J. Org. Chem. 1995, 60, 6666.
(26) Adler, F. L.; Adler, L. T. In Methods in Enzymology (Immunochemi-
cal Techniques); Vunakis, H. V., Langone, J. J., Eds.; Academic Press:
New York, 1980; Vol. 70, pp 455-466.
(27) For reviews of polyvalency in biological systems, see: (a) Kiessling,
L. L.; Pohl, N. Chem. Biol. 1996, 3, 71. (b) Lee, Y.; Lee, R. Acc. Chem.
Res. 1995, 28, 321. (c) Matrosovich, M. N. FEBS Lett. 1989, 252, 1.
(28) Jones, J. F. J. Polym. Sci. 1958, 33, 15.
(29) Brotherton, T. K.; Smith, J., Jr.; Lynn, J. W. J. Org. Chem. 1961,
26, 1283.
(30) The development of drugs based on non-biodegradable polymers
is in its infancy, and the requirement for regulating clearance of such
materials is not known.
(24) Lees, W. J.; Spaltenstein, A.; Kingery-Wood, J. E.; Whitesides, G.
M. J. Med. Chem. 1995, 37, 3419.

Libraries of DeriVatiVes of Poly(acrylic acid)
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4105
Figure 2. Quasi-solid-phase strategy for synthesis of derivatives of
poly(acrylic acid), pAA. Co-polymer, poly(acrylic acid) presenting R1
group as amide side chains, pAA(R1), was generated in a well of a
96-microtiter plate by ultrasonicating a mixture of solid poly(acrylic
anhydride) and an aqueous solution of a primary amine, R1NH2.
Efficiency of the reaction was optimal when the ratio of molecular
equivalents (mol equiv) of R1NH2 to anhydride groups of pAAn was
<0.2, and the pH of the aqueous solution used was either ∼7 (for
aromatic amines) or ∼12 (for aliphatic amines). Ter-polymers, pAA-
(R1; R2), were prepared similarly using an aqueous solution containing
two amines: R1NH2 and R2NH2.
Figure 3. Dependence of the efficiency of the quasi-solid-phase
reaction on mol equiv of reactants and pH of aqueous medium. The
variables (yield, mol equiv, øR) are defined in Figure 2. (left) Plot of
yields of incorporation (%) versus mol equiv of RNH2 to anhydride
groups of pAAn as a function of pH (2, 7, 12). (right) Plot of mole
fraction of R (øR) of pAA(R) versus mol equiv of RNH2 to anhydride
groups of pAAn: this plot was derived from the previous plot (left),
and the straight line is a theoretical reference line that represents
quantitative reaction (where yield of incorporation is 100%). øR ) 0.5
(mol equiv of RNH2 to anhydride groups of pAAn)(incorporation yield
L1-NH2 (1). Because amide formation and hydrolysis of
anhydride groups were occurring competitively, the efficiency
of the former was affected by the relative reactivity of each
RNH2 and was also sensitive to both the pH of the aqueous
solutions of RNH2 and the ratio of molecular equivalents of
RNH2 to anhydride groups of pAAn. The optimal pH for amide
formation using the aqueous amine solutions was either ∼7 (for
aromatic amines) or ∼12 (for aliphatic amines); these values
are consistent with the values of pKa for these groups. The
efficiency of amide formation was greatest when the ratio of
molecular equivalents (mol equiv) of RNH2 to anhydride groups
of pAAn was less than 0.2. The average value of the optimized
yield of incorporation of RNH2 groups into the polymer under
these optimized conditions was ∼90% ((5) from experiments
using five different primary amines RNH2 (4-aminobenzoic acid,
6-amino-1-hexanoic acid, N-methylhydroxylamine, L-arginine,
and 1 (NeuAc-L1-NH2)).
of RNH2)/100.36 (a) RNH2 ) 4-aminobenzoic acid. (b) RNH2
6-amino-1-hexanoic acid. (c) RNH2 ) NeuAc-L1-NH2 (1).
)
well produced a library of pAA(R1; R2) with R1 and R2 as amide
side chains; we believe the sequence of R1 and R2 groups was
approximately random. These polymers, pAA(R1; R2), are thus
heterogeneous in terms of the sequence of side chains. Synthetic
polymers prepared by other methods are probably also
heterogeneous: direct polymerization (CH2dCHCO2H,
CH2dCHCONHR1, CH2dCHCONHR2); modification of a
preformed polymer (R1NH2, R2NH2, poly(N-(acryloyloxy)-
succinimide) in DMF; then OH-).
It was straightforward to extend this method to generate
libraries of ter-polymers, pAA(R1; R2), presenting both R1 and
R2 groups (Figure 2). These materials were prepared simply
by sonicating a three-component mixture of R1NH2, R2NH2,
and pAAn as an aqueous suspension. Here, the reaction in each
Throughout this work, we choose poly(acrylic anhydride),
pAAn,28,29 as the precursor polymer because it was easily
available, formed amides by reaction with primary amines in

4106 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Choi et al.
Table 1. Hemagglutination Inhibition Activities of pAA(NeuAc-L)
aqueous solution under conditions of quasi-solid-phase reaction,
and could be converted rapidly to derivatives of pAA in good
yield. We also applied this strategy to other precursor polymers
(reactive to amines) such as poly(methyl vinyl ether-co-maleic
anhydride), pMEMAn, and poly(N-(acryloyloxy)succinimide),21
pNAS. When applied to pMEMAn, the procedures yielded
polymers with acceptable solubilities in water and allowed us
to generate poly(methyl vinyl ether-co-maleic acid) presenting
the side chains of R as amide groups, pMEMA(R). The same
quasi-solid-phase reaction with pNAS, however, did not proceed
well when tested using either of two primary amines (4-
aminobenzoic acid and 6-amino-1-hexanoic acid); the sonication
left pNAS as a mostly unreacted, white suspension. In addition,
the reaction with pNAS had the disadvantage that it generated
a side product, N-hydroxysuccinimide (NAS), in reaction wells;
this reaction product may interfere with some bioassays. Our
strategy is, therefore, most convenient when used with anhydride-
based, precursor polymers, which produce no interfering side
products.
and Libraries of pAA(NeuAc-L; R)a
mol equivb
of RNH2
mol equivb of
NeuAc-L-NH2
KHi AI
polymer
(µM)c
pAA(1)
0
15000d
0.04 (1)
0.06 (1)
0.08 (1)
0.10 (1)
0.11 (1)
0.12 (1)
0.14 (1)
0.17 (1)
0.21 (1)
0.11 (2)
0.11 (3)
0.11 (4)
0.10 (1)
27
13
3.9
3.4
4.4
1.1
1.1
0.50
0.30
0.80
0.20
3.1
pAA(2)
pAA(3)
pAA(4)
pAA(1; R)
RNH2
0.12 (RNH2)
3-aminobenzoic acid
3-amino-5-hydroxybenzoic acid
4-aminobenzoic acid
1.5
3.1
3.1
4-amino-2-hydroxybenzoic acid
4-aminobenzenesulfonic acid
2-aminonicotinic acid
N-methylhydroxylamine
D-2-amino-2-deoxyglucose
D-2-amino-2-dexoymannose
1-amino-1-cyclopropanecarboxylic acid
1-amino-1-cyclopentanecarboxylic acid
1-amino-1-cyclohexanecarboxylic acid
aminocyclohexane
L-arginine
L-glutamate
L-histidine
D-4-hydroxyproline
DL-leucine
L-phenylalanine
L-4′-nitrophenylalanine
L-phenylalanine methyl ester
1-amino-2-phenylethane
L-3-(2′-naphthyl)alanine
L-tryptophan
3.1
1.5
1.5
1.5
2.2
0.055
1.1
0.20
0.028
0.0043
1.5
Generation and in Situ Bioassay of Co-polymeric Sialo-
sides. We applied this quasi-solid-phase synthetic method to
derivatives of NeuAc having different linking groups L (1-4;
NeuAc-Ln-NH2)31-33 to generate co-polymeric derivatives of
pAA presenting NeuAc-L as a side chain (pAA(1)-pAA(4)).
Following sonication, the crude solutions of polymers were
evaluated immediately for hemagglutination inhibition (HAI)
activities using an assay based on chicken erythrocytes and
influenza virus A (X-31).34,35 Table 1 gives the values of
KHi AI (the minimum concentration of NeuAc-L groups from
pAA(NeuAc-L) in solution that prevents hemagglutination) at
various mol equiv of NeuAc-L-NH2 to anhydride groups of
pAAn (for definition of mol equiv, see Figure 2). Here, we
assume that the mol equiv of NeuAc-L-NH2 to anhydride groups
of pAAn is related directly to the mole fraction of NeuAc-
containing side chains in the polymer (øNeuAc-L); øNeuAc-L can
be deduced from an equation including mol equiv and yield of
incorporation.36 The HAI activities of these crude pAA(1) (mol
equiv of 1 to anhydride groups of pAAn ) 0.1-0.2) are
comparable to that of purified poly(acrylamide) presenting 1 (
KHi AI ∼ 0.3 µM at ø1 ∼ 0.0521). Table 1 also shows three other
derivatives of pAA (pAA(2) - pAA(4)) with HAI activities in
the (sub)micromolar range. We used 3 and 4 for incorporation
into side chains of the polymer because aromatic moieties in
the middle of the linkage have been previously shown to
enhance the binding affinity of monomeric NeuAc-L to the HA
site.37 The HAI activities of all monomeric sialic acids (1-4)
were, however, low, KHi AI g 5 mM. We rationalize the
2.5
1.5
1.5
0.30
0.024
0.048
0.024
0.0021
0.00050
0.0043
3.1
L-Gly-L-Gly-L-Gly
L-Gly-L-Phe
1.5
pAA(3; R)
0.13 (RNH2)
0.11 (3)
RNH2
1-amino-2-phenylethane
L-3-(2′-naphthyl)alanine
0.0015
0.00070
a The hemagglutination inhibition (HAI) assay was performed as
described previously using erythrocytes from 2-week-old chickens and
influenza virua A (strain X-31).21,34,35 b Mol equiv refers to the ratio of
molecular equivalents of RNH2 (or NeuAc-L-NH2) used to anhydride
groups of pAAn in a reaction well. c Each reported value of KHi AI
represents an average value from at least five independent trials; the
experimental uncertainty in each value is approximately (50%. d Refers
to the lowest concentration of carboxylic acid groups from pAA in
solution that prevented influenza-mediated hemagglutination of eryth-
rocytes: this low HAI activity of underivatized pAA is our control.
(31) Sparks, M. A.; Williams, K. W.; Whitesides, G. M. J. Med. Chem.
1993, 36, 778.
(32) Ogura, H.; Furuhata, K.; Itoh, M.; Shitori, Y. Carbohydr. Res. 1986,
158, 37.
relative activities of these polymers using the structures of the
monomers. For instance, pAA(3) was active at a 4-fold lower
concentration than pAA(2); this increased activity is probably
due to the increased hydrophobicity of 3 relative to 2.
(33) Lees, W. J.; Spaltenstein, A.; Kingery-Wood, J. E.; Whitesides, G.
M. J. Am. Chem. Soc. 1994, 37, 3419.
(34) Tech. Rep. Ser. W. H. O. 1953, 64, 1.
(35) Kilbourne, E. D. Bull. W. H. O. 1969, 41, 643.
(36) The mole fraction of R-containing side chains in pAA(R) (øR), which
is defined below, can be deduced from an equation including mol equiv
and yield of incorporation (%):
Bioassay of Libraries of Ter-polymeric Sialosides. We
extended this method to generate libraries of ter-polymers, pAA-
(NeuAc-L; R), presenting both NeuAc-L and one other R group,
simply by sonicating a three-component mixture of NeuAc-L-
NH2, RNH2, and pAAn. Application to more than one RNH2
group should also be possible. Table 1 summarizes the values
of KHi AI of pAA(NeuAc-L; R), obtained from combination of
NeuAc-L-NH2 (mol equiv to anhydride groups of pAAn ) 0.10)
and one of 26 different RNH2 (mol equiv to anhydride groups
[-CONHR]
[-COOH] + [-CONHR]
øR
)
)
{no. of moles of -CONHR}
2{no. of moles of anhydride groups of pAAn}
øR ) 0.5 (mol equiv of RNH2 to anhydride groups of pAAn) ×
(incorporation yield (%) of RNH2)/100.
(37) Watowich, S. J.; Skehel, J. J.; Wiley, D. C. Structure 1994, 2, 719.

Libraries of DeriVatiVes of Poly(acrylic acid)
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4107
of pAAn ) 0.12). We chose a variety of commercially available
primary amines with combinations of functional groups: hy-
drophobic, hydrophilic, charged (negatively; positively), aro-
matic, aliphatic, amino sugars, amino acids, and peptides. The
wide selection of the second R group was intended to suggest
a structure-activity relationship for the various ter-polymers;
the incorporation of R group influences the activity of the ter-
polymer by affecting multiple factors such as conformation,
flexibility, and solubility of the polymer as well as by displaying
independently intrinsic biospecific activity of the R group.
Several pAA(1; R) (and pAA(3; R)) were 100 to ∼7000 more
active than the parent co-polymeric pAA(1) (mol equiv of 1
(and 3) to anhydride groups of pAAn ) 0.11) in which there is
no R group.38 Although we measured the activities directly from
crude pAA(NeuAc-L; R), several control experiments confirmed
that these ter-polymers were indeed responsible for the high
activities. Co-polymers that did not contain sialic acid groups,
pAA(R), were inactive at ∼15 mM (except pAA(cyclohexyl)
and pAA(L-phenylalanine methyl ester), which gave weak HAI
activities KHi AI, ∼1.6 and ∼3.2 mM, respectively). The mix-
ture of pAA(1) and RNH2 gave the same result as that obtained
with pAA(1) alone. Finally, the activities of pAA(1; R) (RNH2
) L-3-(2′-naphthyl)alanine, L-tryptophan), either crude or puri-
fied (after dialysis; MW cutoff of ∼3.5 kDa), were the same
within a factor of 2: the purified polymers were characterized
1
by H-NMR spectroscopy (D2O).
Typically, the incorporation of derivatives of hydrophobic
or aromatic amino acids enhanced the activities greatly. The
results are related closely to the previous observation that the
incorporation of benzylamine (øbenzyl ∼ 0.05 or 0.1 as amide
side chains) into poly(acrylamide) presenting 1 (ø1 ∼ 0.2)
increased HAI activities 2- to ∼4-fold.21 In addition, the results
suggest that there is some specificity in the ability of the R
groups to increase the activity of pAA(1; R): (i) the incorpora-
tion of D-2-amino-2-deoxymannose as R group of pAA(1; R)
gave an activity, 40-fold higher than that from D-2-amino-2-
deoxyglucose; (ii) the HAI activities of pAA(1; R) (R )
1-amino-1-cycloalkanecarboxylic acid) are correlated to the ring
size of cycloalkane groups (the value of KHi AI of pAA(1;
1-amino-1-cyclohexanecarboxylic acid) was less than that of
pAA(1; 1-amino-1-cyclopentanecarboxylic acid) < that of pAA-
(1; 1-amino-1-cyclopropanecarboxylic acid). The present work
thus demonstrates that specific structural features (and not
hydrophobicity alone) may be responsible for some of the
Figure 4. Structure of monovalent derivatives of adamantane and
polymeric polyvalent adamantane. Amantadine (1-aminoadamantane)
and its structural analogue, rimantadine, are used commercially as two
of several agents effective against Influenza virus. These molecules
bind to M2 protein on the surface of Influenza virus and prevent H+
flux through the proton-conducting channel formed by M2.
membrane-bound proton-conducting channel,39 and its surface
density (∼20-60 copies per virion) is however small, ∼50-
and 5-fold lower than those of HA and NA, respectively.
Antiviral agents such as amantadine (1-aminoadamantane) and
its structural analogue, rimantadine (Figure 4), interfere with
replication of virus by binding to the M2 site (by diffusing into
the lumen of the channel and preventing H+ flux across
membrane).40,41 We tested the above hypothesis (the possibility
of involvement of hydrophobic surface sites of the virus in
polyvalent attachment of pAA(NeuAc; R)) using poly(acryla-
mide) presenting derivatives of amantadine as an amide side
chain, pA(Ad-1) to pA(Ad-4) (øAd ) 0.1; Figure 4). The
polymers were synthesized by reacting poly(N-(acryloyloxy)-
succinimide) (pNAS) sequentially with 0.1 equiv of adaman-
tane-2 (and -4)-NH2 to activated NHS ester of pNAS, and with
excess ammonia. The values of KHi AI of purified pA(Ad-2) (ø
) 0.1) and pA(Ad-4) (ø ) 0.1) are ∼0.5 and ∼0.3 mM,
respectively; here, KHi AI refers to the lowest concentration of
adamantane groups from pA(Ad) in solution that prevented
hemagglutination. The values of KHi AI of the polymers are
lower at least ∼5-10-fold than those of corresponding monova-
lent derivatives of adamantane. We interpret the result by
assuming that pA(Ad) attaches to the surface of virus by binding
of adamantane ligand to hydrophobic sites (like M2) and
contributes to inhibiting the attachment of virus to erythrocyte,
sterically (steric stabilization; Figure 5).
observed differences in values of KiHAI
.
Mechanism of Inhibition of Viral Adhesion by Polymeric
Polyvalent Sialosides. We previously showed that polyvalent
presentation of NeuAc as a polymer greatly increased its ability
to inhibit viral attachment to the surface of the erythrocyte (as
measured by KHi AI) and that potent HAI activity might be
explained by a combination of two proposed mechanisms: (i)
polyvalent, affinity-enhanced binding of polymeric NeuAc to
viral hemagglutinin (HA) and (ii) steric stabilization of viral
surface by gel layer of adsorbed, water-swollen polymer (Figures
4 and 5).21-23 Regarding the high activity of several of the
new pAA(NeuAc; R) demonstrated here, we hypothesize that
the non-sialoside groups (R) may be involved in (non)specific
binding to hydrophobic sites on the surface of virus. One of
the plausible sites for binding of the hydrophobic R group is
nonfunctional, local hydrophobic clefts on the surface of the
virus, and another is the viral M2 protein. The M2 protein is a
(39) Pinto, L. H.; Holsinger, L. J.; Lamb, R. A. Cell 1992, 65, 517.
(40) Hay, A. J. In Concepts in Viral Pathogenesis III; Notkins, A. L.,
Oldstone, M. B. A., Eds.; Springer-Verlag: New York, 1989; pp 361-
367.
(41) Davies, W. L.; Grunert, R. R.; Haft, R. F.; McGahen, J. W.;
Neumayer, E. M.; Paulshock, M.; Watts, J. C.; Wood, T. R.; Herman, E.
C.; Hoffman, C. E. Science 1964, 144, 862.
(38) The HAI assay requires the use of a finite amount of virus and
cannot measure accurately the effectiveness of inhibitors with KHi AI < 1
nM: Mammen, M.; Helmerson, K.; Kishore, R.; Choi, S.-K.; Phillips, W.
D.; Whitesides, G. M. Chem. Biol. 1996, 3, 757.

4108 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Choi et al.
In summary, certain non-sialoside groups enhanced the
activities of pAA(NeuAc-L; R) in hemagglutination inhibition
by factors up to ∼104, even though they showed no HAI activity
by themselves. We suggest that the large increases in the
potency of these ter-polymers relative to pAA(NeuAc) are partly
due to binding of non-sialoside groups to non-HA sites and thus
to enhancing the affinity of polymer for the viral surface. The
results of assay with several co-polymeric poly(acrylamide) and
poly(acrylic acid) presenting non-sialoside side chains suggest
that steric stabilization may play a role in inhibiting adhesion
of pathogen to cell and may be useful as a concept to designing
new types of pharmaceuticals. The best of these pAA(NeuAc-
L; R) belongs to a new class of hemagglutination inhibitors with
unusually high activities at relatively modest mole fractions of
NeuAc-L (ø ∼ 0.05) and R (ø ∼ 0.06): the value in mole
fraction of NeuAc-L or R is equivalent to ∼ 30-40 side chains
of NeuAc-L or R per polymer molecule. This finding empha-
sizes the importance of combinations of side chains in modulat-
ing the activities of these polyvalent inhibitors.
Conclusion
In short, this procedure is a rapid, economical method for
synthesizing and screening polyvalent, polymeric inhibitors for
bioactivity. Particularly when minor structural modifications
of side chains and/or polymer backbone can induce unexpectedly
large changes in activities,27 the present method may improve
the efficiency of screening candidate compounds. We believe
that this strategy should be broadly applicable to studies of
polymeric modulators involved in a range of biological processes
including pathogen-cell interaction,42,43 tumor metastasis and
immunomodulation,44-46 and cell migration and adhesion.46-49
Experimental Section
General Procedure. Reagents were, unless otherwise specified,
used as received. Microtiter plate (96 conically-shaped wells) equipped
with cover plate was purchased from ICN, Flow. Ultrasonication was
performed on ultrasonic bath-type cleaner (Fisher). Dialysis membranes
(cellulose; MW cutoff of ∼3.5, or 12-14 kDa) were purchased from
Spectrum. Hemagglutination inhibition (HAI) assays were performed
as described,34,35 using Influenza virus (strain X-31) and chick red blood
cells (RBCs, or erythrocytes). A suspension of erythrocytes from
2-week-old chicks was purchased from Spafas Co. Thin layer
chromatography (TLC) was performed on silica gel precoated glass
plates (E. Merck, Darmstadt). Flash column chromatography was
performed on silica gel 60F254 (230-400 mesh, E. Merck).
Figure 5. Scheme of influenza virus-cell association (a) and of the
proposed mechanism of inhibition of the viral adhesion by polymeric
polyvalent inhibitors (b). Influenza virus attaches tightly to the surface
of cell using multivalent binding of its surface hemagglutinins to
multiple copies of sialosides clustered on the surface of the target cell.
Ter-polymers, pAA(NeuAc; R), are believed to inhibit the adsorption
of influenza to the surface of erythrocyte by polyvalent binding of
polymeric NeuAc to HA sites (affinity-enhanced, competitive binding)
and by stabilizing the surface of virus with a layer of adsorbed, water-
swollen polymer gel (steric stabilization). Certain polymers presenting
non-sialoside groups as side chains, pA(adamantane), also inhibited
agglutination of virus to cell, although the inhibitory effect was weak.
These polymers might attach to the surface of virus by binding of non-
sialoside ligands (adamantane) to non-HA sites of virus (such as M2
site) or to adventitious hydrophobic sites, and inhibit the attachment
of virus to cell sterically.
Synthesis of Poly(acrylic anhydride).28,29 A solution (120 mL) of
anhydrous benzene containing acrylic anhydride29 (11.8 g, 93.56 mmol)
and AIBN (azobis(isobutyronitrile), 500 mg) was degassed for 5 min
in Vacuo and saturated with N2 gas for 30 min. The solution was heated
slowly to reflux under a stream of N2 gas and allowed to reflux for 8
h. White precipitate was collected on a Bu¨chner funnel, washed quickly
with anhydrous benzene (∼150 mL) and anhydrous THF (∼100 mL),
and immediately dried in Vacuo. The polymerization reaction yielded
11.5 g of poly(acrylic anhydride), pAAn, as colorless solid. The
molecular weight distribution of pAAn was determined after hydroly-
In proposing steric stabilization of influenza virus as the
mechanism of action of pA(adamantane), we are suggesting
attaching a polymer to the surface of virus by binding of non-
NeuAc ligands to non-HA sites on the viral surface; these sites
would normally be unimportant in adhesion of the virus to the
cell. The substantial HAI activity of pA(adamantane), along
with those of other co-polymers, pAA(cyclohexyl) and pAA-
(L-phenylalanine methyl ester), suggests that the polymer might
inhibit virus-cell adhesion (HA-NeuAc mediated), sterically by
attaching to non-HA sites on the surface of the virus. Further-
more, it implies that the potent activities of certain ter-polymeric
sialosides may be due to binding of a second non-sialoside group
to the viral surface (in addition to binding of NeuAc to HA)
and thus to increased binding affinity of these ter-polymeric
sialosides to the virus.
(42) Karlsson, K.-A. Curr. Opin. Struct. Biol. 1995, 5, 622.
(43) Gaffar, A.; Nathoo, S.; Miller, S. Chem. Br. 1996, May, 51.
(44) Teodorczyk-Injeyan, J. A.; Falk, J. A.; Makowka, L.; Moffat, F.
L.; Filion, L.; Falk, R. E. In Immune Modulation Agents and Their
Mechanisms; Fenichel, R. L., Chirigos, M. A., Eds.; Marcel Dekker, Inc.:
New York, 1984; pp 261-288.
(45) Bovin, N. V.; Gabius, H.-J. Chem. Soc. ReV. 1995, 413.
(46) Rossen, S. D.; Bertozzi, C. R. Curr. Opin. Cell Biol. 1994, 6, 663.
(47) Kelm, S.; Pelz, A.; Schauer, R.; Filbin, M. T.; Tang, S.; Bellard,
M.-E. d.; Schnaar, R. L.; Mahoney, J. A.; Hartnell, A.; Bradfield, P.;
Crocker, P. R. Curr. Biol. 1994, 4, 965.
(48) Spevak, W.; Foxall, C.; Charych, D.; Dasgupta, F.; Nagy, J. O. J.
Med. Chem. 1996, 39, 1018.
(49) Roy, R.; Park, W. K. C.; Srivastava, O. P.; Foxall, C. Bioorg. Med.
Chem. Lett. 1996, 6, 1399.

Libraries of DeriVatiVes of Poly(acrylic acid)
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4109
sis: (i) pAAn was hydrolyzed to poly(acrylic acid), pAA, by ultra-
sonicating a mixture of pAAn and 0.1 M NaOH for 0.5 h; (ii) crude
pAA was purified by dialysis (MW cutoff of ∼3.5 kDa) against 0.1 M
NH4Cl and deionized water; (iii) molecular weights of purified pAA
were estimated by gel permeation chromatography (GPC) using
polysaccharide standards: Mn ) 20.7 kDa, Mw ) 39.5 kDa, polydis-
persity ) Mw/Mn ) 1.91. 1H-NMR (300.1 MHz, purified pAA in
D2O): δ (ppm) 2.7-2.4 (br s), 2.4-2.1 (br s), 2.0-1.7 (br s), 1.7-1.5
(br s), 1.5-1.2 (br s).
MHz, D2O): δ (ppm) 7.6-7.4 (br m), 7.4-7.1 (br m), 3.7-3.5 (br
m), 3.6-3.4 (br m), 3.3-3.1 (br), 3.1-2.8 (br m), 2.6-1.9 (br m),
1.8-1.0 (br m).
Hemagglutination Inhibition (HAI) Assay.21,34,35 Influenza virus
(strain X-31) was kept as a suspension in a super stock solution (∼15
mg of protein/mL of PBS, pH 7.2) at 4 °C. A solution of erythrocytes
from 2-week-old chicks (which was provided as a suspension (∼5%
v/v) in Elsevers solution) was washed with PBS, pH 7.2 (4 times), as
described21 and then resuspended in PBS (∼0.5% v/v). The HAI assay
of polymer was performed at room temperature (∼20 °C), using 2-fold
serially diluted solutions of polymeric NeuAc: (i) each well (50 µL)
containing certain amount of a polymer was mixed with 50 µL of a
suspension of X-31 virus (∼0.025 µg protein/mL PBS); (ii) after 30
min of incubation at ∼20 °C, 100 µL of a suspension of chicken
erythrocytes (∼0.5%) was added to each well, followed by gentle
agitation and incubation for 1 h at room temperature (rt). The end
point of the HAI assay is the last well in which an agglutinated pellet
is observed. The value of HAI activity, KHi AI is defined as the lowest
concentration of NeuAc of polymer in solution (at this end point) that
inhibited the agglutination of erythrocytes induced by influenza virus.
The reported value of KHi AI represents an average of at least five
independent measurements.
Typical Procedure for Generation of pAA(R) Using Quasi-Solid-
Phase Reaction (Figures 2 and 3). Solutions of co-polymers pAA-
(R) were prepared by reacting of RNH2 (4-aminobenzoic acid, 6-amino-
1-hexanoic acid, or NeuAc-L1-NH230) with poly(acrylic anhydride)
(pAAn) using different numbers of molar equivalents of RNH2 to
anhydride groups of pAAn (mol equiv ) {number of moles of NeuAc-
L-NH2}/{number of moles of anhydride groups of pAAn}) and using
aqueous solutions of amines adjusted to different pHs (2, 7, or 12).
The polymer for which the mol equiv is 0, is a homo-polymeric pAA
obtained from sonication (hydrolysis) of pAAn alone in phosphate-
buffered saline (PBS) solution (137 mM NaCl, 2.7 mM KCl, 7.7 mM
Na2HPO4, 1.5 mM KH2PO4, 0.05% NaN3), pH 12. Co-polymeric pAA-
(R; RNH2 ) 4-aminobenzoic acid) for which the mol equiv is >0 was
generated in microtiter plates with 96 conically-bottomed wells as
follows: (i) placing an amount of pAAn powder (2.8, 3.3, 4.4, 6.7,
9.3, 11, 19 mg) into a well; (ii) soaking the powder with 100 µL of
aqueous 4-aminobenzoic acid (0.2 M in PBS buffer, pH 7); (iii)
immediately sealing the plate (taping four sides of a plate with Parafilm
before placing a cover plate tightly), and then ultrasonicating (Fisher
ultrasonic bath-type cleaner) the mixture for 0.5 h: the sonication also
increased the temperature of water in the bath (and accordingly, the
reactants), slowly up to ∼50 °C. Each of the crude polymer solutions
was neutralized with 1.0 M NaOH, lyophilized to dryness, and
Synthesis of Monovalent and Polymeric Polyvalent Adamantanes
(Figure 4). Adamantane-1-NH2 was synthesized from 1-aminoada-
mantane using two steps of reaction: (i) Michael addition to acrylonitrile
(EtOH, 24 h, reflux, 55%) and (ii) reduction of nitrile to amino group
(LiAlH4, ether, 5 h, 0 °C, 74%). 1H-NMR (300.1 MHz, CD3OD): δ
(ppm) 2.68 (t, J ) 7.1 Hz, 2H), 2.61 (t, J ) 7.1 Hz, 2H), 2.06 (br s,
3H), 1.75-1.69 (br d, 12H), 1.63-1.59 (quin, J ) 7.1 Hz, 2H). After
a coupling reaction of adamantane-1-NH2 and 6-NHCbz-1-hexanoic
acid N-hydroxysuccinimide (NHS) ester (MeOH, rt, 24 h, 32%) and
hydrogenolytic deprotection of N-Cbz group (H2 (1 atm), 10% Pd/C,
MeOH, rt, 24 h, ∼90%), the product (adamantane-2-NH2) was obtained
as a pale yellow oil. Rf ) 0.45 in i-PrNH2/MeOH/CH2Cl2 (1:5:15).
1H-NMR (250.1 MHz, CD3OD/CDCl3): δ (ppm) 3.16 (s, 4H), 2.63-
2.57 (t, J ) 7.1 Hz, 2H), 2.10-2.04 (t, J ) 7.3 Hz, 2H), 1.91 (br s,
3H), 1.69-1.68 (br s, 6H), 1.59 (br s, 6H), 1.53-1.41 (m, 4H), 1.25-
1.22 (quin, J ) 6.72 Hz, 2H). FAB-MS (NBA): m/z 344 [M + Na]+.
Adamantane-3-NH2 was synthesized by a coupling reaction of ada-
mantane-1-carboxylic acid and 1,2-diaminoethane (DCC, NHS, CH2-
Cl2; then the diamine, MeOH, 84%). 1H-NMR (300.1 MHz, CDCl3):
δ (ppm) 3.23 (m, 2H), 2.75 (t, J ) 7.0 Hz, 2H), 1.97 (br s, 3H), 1.78
(m, 6H), 1.65 (m, 6H). After a coupling reaction of adamantane-3-
NH2 and 6-NHCbz-1-hexanoic acid NHS ester (DMF, 24 h, rt, 68%)
and deprotection of N-Cbz group (H2 (1 atm), 10% Pd/C, MeOH, rt,
24 h, ∼90%), the product (adamantane-4-NH2) was obtained as a white
solid. Rf ) 0.62 in i-PrNH2/MeOH/CH2Cl2 (1:2:14). 1H-NMR (300.1
MHz, CD3OD/CDCl3): δ (ppm) 3.24-3.19 (t, J ) 6.7 Hz, 2H), 2.74-
2.62 (m, 4H), 2.20-2.15 (t, J ) 7.4 Hz, 2H), 2.08 (br s, 3H), 1.68-
1.58 (m, 14 H), 1.48-1.44 (quin, J ) 7.3 Hz, 2H), 1.37-1.30 (quin,
J )7.3 Hz, 2H). 13C-NMR (100.6 MHz, CD3OD): δ (ppm) 181.08,
176.39, 51.37, 41.73, 41.26, 39.96, 37.57, 36.79, 30.43, 29.73, 27.22,
26.49. FAB-MS (NBA): m/z 358 [M + Na]+. HRMS: calcd for
C19H33N3O2Na 358.2468, found 358.2470. To a solution of DMF (2
mL) containing poly(N-(acryloyloxy)succinimide) or pNAS (50 mg,
equivalent to 0.3 mmol of NAS) was added adamantane-4-NH2 (10.1
mg, 29 µmol) dissolved in DMF (1 mL), followed by addition of Et3N
(0.1 mL). After stirring of the mixture (3 d, rt; then 6 h, 50 °C), aqueous
NH3 (10% w/w; 1 mL) was added to the mixture, followed by stirring
for additional 6 h at rt. At the conclusion of reaction, the mixture was
transferred into a dialysis bag (cellulose membrane, MW cutoff of
∼12-14 kDa) and was dialyzed at rt over 3 days: water (2 × 4 L),
5% (w/w) NH4Cl (4 L), and water (2 × 4 L). After dialysis, the content
of the bag was frozen, and lyophilized to afford 25 mg of pA(Ad-4) as
a fluffy white solid (ca. yield ) 83%). 1H-NMR (300.1 MHz, D2O):
δ (ppm) 3.13-3.12 (br s), 2.25-1.95 (br m), 1.83 (br s), 1.7-1.01 (br
m), 1.6 (s). pA(Ad-2) was prepared similarly from adamantane-2-NH2.
1H-NMR (300.1 MHz, D2O): δ (ppm) 3.14 (br s), 2.98 (br m), 2.84
(br m), 2.4-1.9 (br m), 1.7 (s), 1.6-1.0 (br m).
1
characterized by H-NMR spectroscopy. The yield of incorporation
was estimated by comparing the intensity of integrated signal of reacted
RNH2 (as RNH-CO- of the polymer) relative to that of free unreacted
RNH2. Crude pAA(R) (RNH2 ) 4-aminobenzoic acid; mol equiv of
RNH2 to anhydride groups of pAAn ) 0.38; pH ∼7.0): yield of
incorporation ) 62%. 1H-NMR (300.1 MHz, D2O): δ (ppm) 7.8-
7.5 (br), 7.7-7.6 (d, J ) 7.5 Hz; 2H from unreacted RNH2), 7.5-6.9
(br), 6.7-6.6 (d, J ) 7.5 Hz; 2H from unreacted RNH2), 2.7-2.0 (br
m), 1.9-1.3 (br m). Crude pAA(R) (RNH2 ) 6-amino-1-hexanoic acid;
mol equiv of RNH2 to anhydride groups of pAAn ) 0.26; pH ∼12):
yield of incorporation ) 55%. 1H-NMR (300.1 MHz, D2O): δ (ppm)
3.2-2.9 (br s), 2.9-2.8 (t, J ) 7.0 Hz; 2H from unreacted RNH2),
2.5-2.0 (br m), 1.8-1.1 (br m). Crude polymer solutions were purified
by dialysis (MW cutoff of ∼12-14 kDa; against water): 1H-NMR
spectra of pure pAA(R) showed no signals for free RNH2. The effect
of the pH on the extent of incorporation was studied by performing
the reaction using aqueous solutions of primary amines with pH 2, 7,
or 12.
Typical Procedure for Generation of Libraries of pAA(NeuAc)
and pAA(NeuAc; R). Solutions of co-polymers pAA(NeuAc-L) were
prepared by reaction of NeuAc-L-NH2 (1, 2, 3, or 4) with pAAn using
different numbers of mol equiv of NeuAc-L-NH2 to anhydride groups
of pAAn (Figure 2). Co-polymeric pAA(NeuAc-L) for which the mol
equiv is >0 was generated in microtiter plates with 96 conically-
bottomed wells as follows: (i) placing 6 mg of pAAn into a well; (ii)
soaking the powder with a variable amount (19-100 µL) of 0.1 M
NeuAc-L-NH2 in PBS buffer, pH 12; (iii) immediately sealing the plate,
and then ultrasonicating the mixture for 0.5 h. Each solution of co-
polymers (pH ∼ 3) generated in a well was neutralized to pH ∼ 7 by
adding 60 µL of 1.0 M NaOH and adjusted to 100 or 200 µL (total
volume) with PBS, pH 7.2, before the HAI assay. pAA(1) (mol equiv
of 1 to anhydride groups of pAAn ) 0.08; pH ∼ 12). 1H-NMR (300.1
MHz, D2O): δ (ppm) 3.8-3.5 (m), 3.5-3.3 (m), 3.3-3.1 (br), 2.6-
2.3 (br m), 2.3-1.8 (br m), 1.7-1.0 (br m). The above protocol was
extended similarly to the preparation of ter-polymers pAA(NeuAc-L;
R); here, a three-component mixture (6 mg of pAAn, 50 µL of 0.1 M
NeuAc-L-NH2 (1 or 3) and 30 µL of 0.2 M RNH2) was sonicated.
Purified (dialyzed) pAA(1; R) (RNH2 ) L-3-(2′-naphthyl)alanine; mol
equiv of 1 to anhydride groups of pAAn ) 0.10; mol equiv of RNH2
to anhydride groups of pAAn ) 0.12; pH ∼ 12): 1H-NMR (300.1
Synthesis of Derivatives of N-Acetylneuraminic Acid (Figure 6).
The R-C- and R-O-sialosides containing an amine-terminated linker

4110 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Choi et al.
Figure 6. Summary scheme for synthesis of R-C- and R-O-sialosides. (a) Nꢀ-Cbz-NR-dansyllysine, COIm2, Et3N, DMF, rt, 93%. (b) H2 (1 atm),
10% Pd/C, MeOH, rt, ∼100%. (c) hν (254 nm), 4-mercapto-1-butanoic acid, 4,4′-azo-bis(isocyanovaleric acid), H2O, 10 h, 62-74%. (d) NaN3,
DMF, rt, ∼100%. (e) H2 (1 atm), 10% Pd/C, MeOH, rt, 60%. (f) 4-NHBoc-butyric acid N-hydroxysuccinimide ester, DMF, 80 °C, 2 d, 35%. (g)
COIm2, DMF, 80 °C, 2 d, 42%. (h) LiOH‚H2O, H2O, rt, 24 h, ∼100%. (i) TFA, CH2Cl2, 0 °C, 1 h, ∼100%.
group, NeuAc-L1-NH2 (1)31 and NeuAc-L2-NH2 (2),32,33 were synthe-
sized according to literature methods.
71.12, 70.17, 69.34, 64.56, 63.93, 58.65, 57.77, 54.15, 42.23, 40.34,
40.27, 33.65, 33.34, 27.86, 23.24, 22.90, 22.72. FAB-MS (glycerol):
m/z 758 [M + H]+, 780 [M + Na]+.
NeuAc-L3-NH2 (3). A mixture of 1,1′-carbonyldiimidazole (53 mg,
0.33 mmol) and Nꢀ-Cbz-NR-dansyllysine (160 mg, 0.31 mmol) in DMF
(2 mL) was stirred for 3 h at rt, followed by addition of NeuAc-L2-
NH2 (2 as TFA salt; 120 mg, 0.24 mmol) and Et3N (0.15 mL, 1.08
mmol). After stirring (24 h, rt), the mixture was concentrated in Vacuo
to yield a pale yellow oily residue. This residue was dissolved in MeOH
(15 mL) containing 5 g of ion exchange resin (Dowex-50W, H+-form).
After shaking for 5 min, the resin-containing solution was filtered
through filter paper and was concentrated prior to flash column
chromatography (50 g of silica gel; 10% MeOH/CH2Cl2 to 5% HCO2H/
30% MeOH/CH2Cl2). The coupling product was obtained as a pale
yellow oil (195 mg, 93%) from fractions with Rf of 0.5 (5% HCO2H/
30% MeOH/CH2Cl2). 1H-NMR (500.1 MHz, CD3OD): δ (ppm) 8.53-
8.50 (m, 1H), 8.38-8.36 (br d, J ) 8.6 Hz, 1H), 8.21-8.19 (br d, J )
8.6 Hz, 1H), 7.57-7.52 (m, 2H), 7.32-7.24 (m, 5H; -COOCH2C6H5),
7.22-7.18 (m, 1H), 5.03 (s, 2H; -COOCH2C6H5), 3.88-3.80 (m, 3H),
3.75-3.48 (m, 9H), 3.27-3.26 (m, 2H), 3.08-3.07 (m, 2H), 2.84 (s,
6H; N(CH3)2), 2.71-2.67 (m, 3H), 2.0 (s, 3H; CH3CONH), 1.63-
1.58 (t, J ) 12.0 Hz, 1H; H3R), 1.49-1.48 (br s, 2H), 1.1-1.0 (m,
4H); FAB-MS (glycerol): m/z 914 [M + Na]+. A suspension of the
above N-Cbz protected, derivative of sialic acid (180 mg, 0.20 mmol)
and 10% Pd/C (100 mg) in MeOH (20 mL) was stirred under an
atmosphere of H2 (1 atm) for 1 d at rt. After removal of the palladium
catalyst by filtering through a pad of Celite, the filtrate was evaporated
in Vacuo to yield a crude product (NeuAc-L3-NH2 (3)) as a pale yellow
oil. We estimated the yield of hydrogenolysis to be approximately
quantitative on the basis of a 1H-NMR spectrum that indicates the
complete removal of the N-Cbz group. 1H-NMR (399.9 MHz, CD3-
OD): δ (ppm) 8.56-8.54 (t, J ) 8.69 Hz, 1H), 8.38-8.35 (dd, J )
3.27, 8.69 Hz, 1H), 8.22-8.20 (d, J ) 8.7 Hz, 1H), 7.62-7.52 (m,
2H), 7.27-7.23 (m, 1H), 3.91-3.77 (m, 4H), 3.76-3.61 (m, 4H), 3.58-
3.56 (m, 4H), 3.48-3.45 (t, J ) 4.75 Hz, 2H), 3.01-2.98 (m, 2H),
2.86 (s, 6H; N(CH3)2), 2.72-2.62 (m, 3H), 2.0 (s, 3H; CH3CONH),
1.64-1.56 (m, 3H), 1.43-1.40 (m, 2H), 1.18 (m, 2H). 13C-NMR (100.6
MHz, CD3OD): δ (ppm) 175.47, 175.10, 173.72, 129.34, 124.42,
120.56, 119.59, 119.18, 116.68, 116.45, 113.77, 101.72, 74.34, 73.03,
NeuAc-r-C-CO2H (6). Compound 6 was prepared using the
radical-initiated addition reaction of a thiol to an olefinic group of R-C-
allyl sialoside (5), according to literature methods.31,50,51 A suspension
of R-C-allyl-sialic acid methyl ester52,53 (0.6 g, 1.73 mmol), 4-mercapto-
1-butanoic acid54 (0.8 g, 6.67 mmol), and 4,4′-azobis(4-cyanovaleric
acid) (0.4 g, 1.43 mmol) in water (10 mL) was degassed for 10 min in
Vacuo prior to being saturated with N2 (by bubbling N2 gas through
the solution for 1 h). A reaction flask containing the mixture was placed
in a photochemical reactor (Rayonet) and was irradiated at 254 nm for
10 h. Evaporation of the irradiated mixture afforded a pale yellow
oil, which was purified with flash silica gel chromatography (10%
MeOH/CH2Cl2 to 5% HCO2H/30% MeOH/CH2Cl2). The adduct was
obtained as an oil from fractions with Rf ) 0.6 (5% HCO2H/30%
MeOH/CH2Cl2). The combined oily material was dissolved in MeOH
(5 mL), and it was poured slowly into ether (50 mL), which led to
instant precipitation of the product as a white solid (0.5-0.6 g, 62-
74%). 1H-NMR (500.1 MHz, CD3OD): δ (ppm) 3.84-3.76 (m, 3H),
3.78 (s, 3H; COOCH3), 3.70-3.48 (m, 4H), 2.60-2.55 (dd, J ) 4.62,
13.16 Hz, 1H; H3â), 2.53-2.47 (m, 4H; CH2SCH2), 2.36-2.35 (t, J )
7.24 Hz, 2H; CH2COOH), 2.04 (s, 3H; CH3CONH), 1.87-1.80 (m,
4H; CH2CH2SCH2CH2), 1.73-1.70 (m, 1H), 1.60-1.55 (t, J ) 11.89
Hz, 1H; H3R), 1.51-1.41 (m, 1H); FAB-MS (glycerol; negative ion
mode): m/z 466 [M - H]-; HRMS: calcd for C19H32N1O10S1 466.1745,
found 466.1747.
Compound 10. 1,4-Bis(bromomethyl)naphthalene (7)55 was con-
verted to 1,4-bis(azidomethyl)naphthalene (NaN3, DMF, rt);56 the azido
group was then reduced to an amino group using catalytic hydrogenation
(H2 (1 atm), 10% Pd/C, MeOH). 1,4-Bis(azidomethyl)naphthalene (8).
1H-NMR (400 MHz, CDCl3): δ (ppm) 8.06 (m, 2H), 7.63 (m, 2H),
(50) Roy, R.; Laferriere, C. A. Carbohydr. Res. 1988, 177, C1.
(51) Lee, R. T.; Lee, Y. C. Carbohydr. Res. 1974, 37, 193.
(52) Nagy, J. O.; Bednarski, M. D. Tetrahedron Lett. 1991, 32, 3953.
(53) Paulsen, H.; Matschulat, P. Liebigs Ann. Chem. 1991, 487.
(54) Lochon, P.; Schoenleber, J. Tetrahedron 1976, 32, 3023.
(55) Harvey, R. G.; Pataki, J.; Cortez, C.; Raddo, P. D.; Yang, C. J.
Org. Chem. 1991, 56, 1210.
(56) Blumenstein, J. J.; Michejda, C. J. Tetrahedron Lett. 1991, 32, 183.

Libraries of DeriVatiVes of Poly(acrylic acid)
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4111
7.45 (s, 2H), 4.77 (s, 4H). 13C-NMR (100 MHz, CDCl3): δ (ppm)
132.32, 131.73, 126.97, 126.51, 124.33, 52.94. CI-MS (NH3): m/z
273 (M + NH3)+. 1,4-Bis(aminomethyl)naphthalene (9). 1H-NMR
(300 MHz, CDCl3): δ (ppm) 8.10 (m, 2H), 7.54 (m, 2H), 7.43 (s, 2H),
4.31 (s, 4H). 13C-NMR (100 MHz, CDCl3): δ (ppm) 138.31, 131.60,
126.06, 124.31, 124.08, 44.13. CI-MS (NH3): m/z 187 (M + H)+. A
solution of DMF (150 mL) containing 1,4-bis(aminomethyl)naphthalene
(1.4 g, 7.53 mmol) and 4-NHBoc-1-butyric acid NHS ester (2.17 g,
7.53 mmol) was stirred at 80 °C for 2 d under a stream of N2.
Evaporation of DMF afforded an oily residue which was purified with
flash silica (200 g) chromatography (10% MeOH/CH2Cl2, 5% i-PrNH2/
10% MeOH/CH2Cl2). A monoadduct, 1-[(4′-NHBoc-1′-oxo-1′-ami-
nobutyl)methyl]-4-(aminomethyl)naphthalene (10) (Rf ) 0.58 in 5%
i-PrNH2/10% MeOH/CH2Cl2), was obtained in 35% yield (0.98 g). 1H-
NMR (300.1 MHz, CD3OD/CDCl3): δ (ppm) 7.98-7.95 (m, 2H),
7.51-7.49 (m, 2H), 7.33-7.32 (s, 2H), 4.81-4.77 (two s, 4H), 3.01-
2.98 (t, J ) 6.84 Hz, 2H), 2.18-2.14 (t, J ) 6.84 Hz, 2H), 1.74-1.70
(q, 2H, J ) 6.84 Hz), 1.34 (s, 9H). FAB-MS. m/z 393 (M + Na)+.
The reaction also yielded a bis-adduct, 1,4-bis[(4′-NHBoc-1′-oxo-1′-
aminobutyl)methyl]naphthalene (1.47 g).
NeuAc-L4-NH2 (4). A solution of DMF (3 mL) containing NeuAc-
R-C-CO2H (6) (292 mg, 0.625 mmol) and 1,1′-carbonyldiimidazole
(105 mg, 0.654 mmol) was stirred for 2 h at rt, prior to addition of 10
(300 mg, 0.810 mmol). The final mixture was stirred for 2 d at 80 °C
under a stream of N2. Evaporation of DMF afforded a pale yellow
residue, which was then dissolved in MeOH (10 mL) containing 5 g
of ion exchange resin (Dowex-50W; H+ form). After the suspension
was stirred for 5 min, resins were removed by filtration through filter
paper. The concentrated residue of the filtrate was purified by flash
column chromatography (silica gel; 5% MeOH/CH2Cl2 to 20% MeOH/
CH2Cl2). The coupling product (218 mg, 42%) was obtained as a light
yellow oil (Rf ) 0.71 in 20% MeOH/CH2Cl2). 1H-NMR (500.1 MHz,
CD3OD): δ (ppm) 8.07 (m, 2H), 7.55-7.53 (m, 2H), 7.41-7.39 (m,
2H), 4.90-4.85 (s, 2H), 4.82 (s, 2H), 3.95-3.71 (m, 3H), 3.70-3.47
(m, 4H), 3.73 (s, 3H; COOCH3), 3.05-3.03 (t, J ) 6.82 Hz, 2H; CH2-
NHBoc), 2.55-2.45 (m, 7H), 2.25-2.23 (t, J ) 6.82 Hz, 2H; ArCH2-
NHCOCH2), 2.02 (s, 3H; CH3CONH), 1.85-1.65 (m, 7H), 1.62-1.58
(t, H3R, J ) 12.84 Hz), 1.40 (m, 10H). FAB-MS (glycerol): m/z 721
(M + H-Boc)+. To a solution of MeOH (5 mL) containing the above
product (120 mg, 0.146 mmol) was added LiOH‚H2O (20 mg, 0.477
mmol) in H2O (5 mL). After being stirred (24 h, rt), the mixture was
adjusted to pH ∼ 3 by adding ion exchange resin (Dowex-50W, H+
form). After filtration of resins, the filtrate was concentrated in Vacuo
to afford an oily residue: this material was pure enough to proceed to
next step without further purification. 1H-NMR (500.1 MHz, CD3-
OD): δ (ppm) 8.06 (m, 2H), 7.57-7.54 (m, 2H), 7.42-7.41 (m, 2H),
4.85-4.81 (two s, 4H), 3.85-3.70 (m, 3H), 3.68-3.50 (m, 4H), 3.05-
3.02 (t, J ) 6.7 Hz, 2H; CH2NHBoc), 2.62-2.59 (m, H3â), 2.55-2.48
(m, 4H; CH2SCH2), 2.40-2.38 (t, J ) 7.23 Hz, 2H; CH2CONH), 2.26-
2.22 (t, J ) 7.16 Hz, 2H; NHCOCH2), 2.04 (s, 3H); CH3CONH), 1.86-
1.82 (m, 5H), 1.76-1.74 (t, J ) 7.16 Hz, 2H), 1.59-1.58 (t, H3R, J )
11.0 Hz), 1.40 (m, 10H). FAB-MS (glycerol; negative ion mode): m/z
805 [M - H]-. HRMS: calcd for C39H57N4O12S1 805.3690, found
805.3694, calcd for C39H58N4O12S1Na1 829.3666, found 829.3670. To
an N-Boc protected 4 (50 mg, 0.062 mmol) suspended in CH2Cl2 (1
mL) at 0 °C was added a mixture of CF3CO2H (1 mL) and CH2Cl2 (1
mL). The mixture was stirred for 1 h at 0 °C, and it (a pale red solution)
was poured slowly into cold ether (50 mL), which led to formation of
a white precipitate. It was collected by centrifugation and washed with
ether (50 mL). The yield on deprotection of N-Boc was estimated to
be quantitative on the basis of 1H-NMR (CD3OD) spectrum of the
product (NeuAc-L4-NH2 (4) as TFA salt). FAB-MS (glycerol; negative
ion mode): m/z 705 [M - H]-.
Acknowledgment. The study has been supported by NIH
(GM 30367). M.M. was an Eli Lilly predoctoral fellow. We
thank Professor John J. Skehel for his generous gift of Influenza
virus A (strain X-31). We thank Andreas Walch for his
assistance in synthesis of 4. The NMR and mass spectroscopy
facilities at Department of Chemistry and Chemical Biology in
Harvard University have been supported by funds from NSF
(CHE88-140195/CHE-9020043) and NIH (SIO-RR06716/1-
SIO-RR04870-01).
JA963519X