Using bifunctional polymers presenting vancomycin and fluorescein groups to direct anti-fluorescein antibodies to self-assembled monolayers presenting D-alanine-D-alanine groups

Article abstract of DOI:10.1021/ja030045a

This paper describes the synthesis of bifunctional polyacrylamides containing pendant vancomycin (Van) and fluorescein groups, and the use of these polymers to direct antibodies against fluorescein to self-assembled monolayers (SAMs) presenting D-alanine-D-alanine (DADA) groups. These polymers bind biospecifically to these SAMs via interactions between the DADA and Van groups and serve as a molecular bridge between the anti-fluorescein antibodies and the SAM. The binding events were characterized using surface plasmon resonance spectroscopy and fluorescence microscopy. The paper demonstrates that polyvalent, biospecific, noncovalent interactions between a polymer and a surface can be used to tailor the properties of the surface in molecular recognition. It also represents a first step toward the design of polymers that direct arbitrarily chosen antibodies to the surfaces of cells.

Full text of DOI:10.1021/ja030045a

Published on Web 03/20/2003
Using Bifunctional Polymers Presenting Vancomycin and
Fluorescein Groups To Direct Anti-Fluorescein Antibodies to
Self-Assembled Monolayers Presenting D-Alanine-D-Alanine
Groups
Steven J. Metallo, Ravi S. Kane, R. Erik Holmlin, and George M. Whitesides*
Contribution from the Department of Chemistry and Chemical Biology, HarVard UniVersity,
12 Oxford Street, Cambridge, Massachusetts 02138
Received January 23, 2003; E-mail: gwhitesides@gmwgroup.harvard.edu
Abstract: This paper describes the synthesis of bifunctional polyacrylamides containing pendant vancomycin
(Van) and fluorescein groups, and the use of these polymers to direct antibodies against fluorescein to
self-assembled monolayers (SAMs) presenting D-alanine-D-alanine (DADA) groups. These polymers bind
biospecifically to these SAMs via interactions between the DADA and Van groups and serve as a molecular
bridge between the anti-fluorescein antibodies and the SAM. The binding events were characterized using
surface plasmon resonance spectroscopy and fluorescence microscopy. The paper demonstrates that
polyvalent, biospecific, noncovalent interactions between a polymer and a surface can be used to tailor
the properties of the surface in molecular recognition. It also represents a first step toward the design of
polymers that direct arbitrarily chosen antibodies to the surfaces of cells.
Bertozzi and Bednarski13,14 have synthesized a biotinylated
Introduction
C-glycoside of mannose. This conjugate was allowed to bind
to avidin, and the resulting aggregate was then exposed to anti-
avidin antibodies. The resulting complex bound to cells of
Escherichia coli expressing type 1 pili (these pili contain
mannose-specific receptors) via the mannosides present and
displayed the antibody portion of the complex; this region could
be recognized by components of the immune system. Li et al.15
have synthesized polymers presenting mannose and GalR1f3Gal,
and have demonstrated the ability of these polymers to bind
human anti-Gal antibodies and to prevent the agglutination of
yeast by E. coli (expressing type 1 pili mannose-binding sites
on their surface). They did not, however, demonstrate the ability
of these polymers to target anti-Gal antibodies to E. coli.
Polyvalent interactionssinteractions that involve the simul-
taneous binding of multiple ligands on one entity to multiple
receptors on anothersare ubiquitous in biology.1 We,2-5 and
others,6-10 have synthesized polymers that present multiple
copies of suitable ligands and that bind polyvalently to the
surfaces of pathogens. If polymers presenting multiple copies
of a ligand directed toward a specific targetspathogen or cell
(a cell endogenous to the organism, a tumor cell, a virally
infected cell, etc.)swere to incorporate a pendant antigen that
recruited antibodies to the surface of the target after the polymer
had bound to it, then it might be possible to direct or modulate
the interaction of the target with the immune system. Polymers
possessing both multiple target-binding sites and ligands that
interacted with elements of the immune system would be loosely
functionally analogous, but structurally dissimilar, to antibod-
ies.11,12
As a first step toward the development of a general strategy
for targeting antibodies to the surfaces of cells and to other
biological surfaces, we have synthesized bifunctional polymers
presenting both vancomycin (Van) and an antigen (fluorescein)
that is recognized specifically by an antibody (anti-fluorescein
IgG, IgGF). Vancomycin is a broad-spectrum glycopeptide
antibiotic that inhibits the growth of Gram-positive bacterial
cells by inhibiting the synthesis of the cell wall.16-18 It functions
(1) Mammen, M.; Choi, S.-K.; Whitesides, G. M. Angew. Chem., Int. Ed. Engl.
1998, 37, 2755-2794.
(2) Spaltenstein, A.; Whitesides, G. M. J. Am. Chem. Soc. 1991, 113, 686-
687.
(3) Kingery-Wood, J. E.; Williams, K. W.; Sigal, G. B.; Whitesides, G. M. J.
Am. Chem. Soc. 1992, 114, 7303-7305.
(4) Sigal, G. B.; Mammen, M.; Dahmann, G.; Whitesides, G. M. J. Am. Chem.
Soc. 1996, 118, 3789-3800.
(5) Choi, S.-K.; Mammen, M.; Whitesides, G. M. J. Am. Chem. Soc. 1997,
119, 4103-4111.
(11) Ravetch, J. V.; Kinet, J. Annu. ReV. Immunol. 1993, 9, 457-492.
(12) Antibodies are at least bivalent and have domains (for example, the Fc
region) that mediate the clearance of target pathogens or cells by
macrophages.
(6) Matrosovich, M. N.; Mochalova, L. V.; Marinina, V. P.; Byramova, N. E.;
Bovin, N. V. FEBS Lett. 1990, 272, 209-212.
(7) Mochalova, L. V.; Tuzikov, A. B.; Marinina, V. P.; Gambaryan, A. S.;
Byramova, N. E.; Bovin, N. V.; Matrosovich, M. N. AntiViral Res. 1994,
23, 179-180.
(13) Bertozzi, C. R.; Bednarski, M. D. J. Am. Chem. Soc. 1992, 114, 2242-
2245.
(14) Bertozzi, C. R.; Bednarski, M. D. J. Am. Chem. Soc. 1992, 114, 5543-
5546.
(8) Roy, R.; Laferriere, C. A. Carbohydr. Res. 1988, 177, C1-C4.
(9) Roy, R.; Zanini, D.; Meunier, S. J.; Romanowska, A. J. Chem. Soc., Chem.
Commun. 1993, 213-217.
(15) Li, J.; Zacharek, S.; Chen, X.; Wang, J.; Zhang, W.; Janczuk, A.; Wang,
P. G. Bioorg., Med. Chem. 1999, 7, 1549-1558.
(10) Sabesan, S.; Duus, J. O.; Neira, S.; Domaille, P.; Kelm, S.; Paulson, J. C.;
Bock, K. J. Am. Chem. Soc. 1992, 114, 8363-8375.
(16) Barna, J. C.; Williams, D. H. Annu. ReV. Microbiol. 1984, 38, 339-357.
(17) Perkins, H. R. Biochem. J. 1969, 111, 195-205.
9
4534
J. AM. CHEM. SOC. 2003, 125, 4534-4540
10.1021/ja030045a CCC: $25.00 © 2003 American Chemical Society

Synthesis of Bifunctional Polymers
A R T I C L E S
interaction between the polymer and DADA to be investigated.19
SAMs of alkanethiolates on gold are also compatible with
measurements of binding by surface plasmon resonance spec-
troscopy (SPR).19-23 We used SPR to study both the binding
of the polymer24-26 to the SAM and the subsequent binding of
antibodies to the surface comprising the SAM and a film of
adsorbed polymer. Finally, we confirmed that the polymer bound
to live bacteria at a concentration and a time scale consistent
with observations of binding to the model surface.
Scheme 1. Illustration of the Use of Bifunctional Polymers
Presenting Vancomycin and Fluorescein Groups to Direct
Anti-fluorescein Antibodies to a Surfacea
Results and Discussion
Synthesis of Polymers. Poly(N-acryloyloxysuccinimide)
(PNAS)27 was available from a previous study. Polyacrylamide
(pA) presenting 1 mol %28 fluorescein groups (pA-F) was
synthesized by the reaction of PNAS with 5-((5-aminopentyl)-
thioureidyl) fluorescein (fluorescein cadaverine) in anhydrous
N,N-dimethylformamide (DMF) in the presence of triethylamine,
followed by a quenching of the reaction mixture with aqueous
ammonium hydroxide. Polyacrylamide presenting 5 mol %
vancomycin and 1 mol % fluorescein groups (pA-V-F) was
synthesized as shown in Scheme 2. Although the vancomycin
derivative 1 used for the synthesis of pA-V-F has three
potentially reactive amines, the reaction with the hindered
secondary or primary amine does not compete significantly with
the reaction with the primary amine from the diaminobutane
linker.29
Preparation of Surfaces. Mixed SAMs presenting DADA
groups and tri(ethylene glycol) groups (which resist the non-
specific adsorption of biomolecules) were generated as described
by Lahiri et al.30 A mixed SAM was generated from ethanolic
solutions of a mixture of HS(CH2)11(OCH2CH2)3OH (1.8 mM)
and HS(CH2)11(OCH2CH2)6OCH2COOH (0.2 mM).31 Reaction
of N-R-acetyl-L-lysine-D-alanine-D-alanine (N-R-Ac-KDADA)
with the carboxylic acid groups of the SAM (activated as
N-hydroxysuccinimidyl esters) generated SAMs presenting
DADA groups. To confirm the coupling of N-R-Ac-KDADA to
the SAM, we used polarized infrared external reflectance
spectroscopy (PIERS). PIERS provides evidence for the pres-
(19) Rao, J.; Yan, L.; Xu, B.; Whitesides, G. M. J. Am. Chem. Soc. 1999, 121,
2629-2630.
(20) Mrksich, M.; Sigal, G. B.; Whitesides, G. M. Langmuir 1995, 11, 4383-
4385.
a
The model surface used in this study consisted of SAMs presenting
DADA groups in a background of tri(ethylene glycol) groups. Passing a
solution of the bifunctional polymer over the SAM resulted in the adsorption
of the polymer on the surface. The adsorbed polymer film presented
fluorescein groups that enabled the binding of IgGF but not of other
antibodies, to the surface. The SAM, the antibody, and the pendant
vancomycin and fluorescein groups are drawn approximately to scale. The
scheme shows only a part of a polymer chain.
(21) Sigal, G. B.; Mrksich, M.; Whitesides, G. M. Langmuir 1997, 13, 2749-
2755.
(22) Lahiri, J.; Isaacs, L.; Grzybowski, B.; Carbeck, J. D.; Whitesides, G. M.
Langmuir 1999, 15, 7186-7198.
(23) Chapman, R. G.; Ostuni, E.; Yan, L.; Whitesides, G. M. Langmuir 2000,
16, 6927-6936.
(24) Green, R. J.; Tasker, S.; Davies, J.; Davies, M. C.; Roberts, C. J.; Tendler,
S. J. B. Langmuir 1997, 13, 6510-6515.
(25) Sigl, H.; Brink, G.; Seufert, M.; Schulz, M.; Wegner, G.; Sackmann, E.
Eur. Biophys. J. Biophys. Lett. 1997, 25, 249-259.
(26) Rudolph, J.; Patzsch, J.; Meyer, W. H.; Wegner, G. Acta Polym. 1993, 44,
230-237.
by binding to terminal DADA residues in the precursors of the
cell wall of Gram-positive bacteria.16-18 Fluorescein is a low
molecular weight antigen representative of many that might be
used to recruit antibodies. This paper describes a test of the
ability of bifunctional polymers presenting vancomycin and
fluorescein groups to direct anti-fluorescein antibodies to
surfaces in a model system based on self-assembled monolayers
(SAMs) presenting D-alanine-D-alanine (DADA) groups (Scheme
1). SAMs of alkanethiolates on gold allow molecular control
over the properties of solid-liquid interfaces and allow the
(27) Mammen, M.; Dahmann, G.; Whitesides, G. M. J. Med. Chem. 1995, 38,
4179-4190.
(28) We define the mole percent of a group as the percentage of side chains in
the polymer containing that group.
(29) Sundram, U. N.; Griffin, J. H. J. Org. Chem. 1995, 60, 1102-1103.
(30) Lahiri, J.; Isaacs, L.; Tien, J.; Whitesides, G. M. Anal. Chem. 1999, 71,
777-790.
(31) The composition of the SAM formed from an ethanolic solution of
alkanethiols with polar tail groups is similar but not necessarily identical
to the solution composition: Bain, C. D.; Evall, J.; Whitesides, G. M. J.
Am. Chem. Soc. 1989, 111, 7155-7164. Lahiri et al.30 have observed that
the mole fraction, ø(COOH), of the carboxylic acid-terminated alkanethi-
olate incorporated into SAMs along with HS(CH2)11(OCH2CH2)3OH, was
higher than the solution ø(COOH), with the maximum deviation occurring
at a solution ø(COOH) ) 0.4. The mole fractions of the SAMs reported in
this paper refer to the mole fractions of the solutions used to prepare the
SAM.
(18) Williams, D. H.; Bardsley, B. Angew. Chem., Int. Ed. Engl. 1999, 38, 1173-
1193.
9
J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003 4535

A R T I C L E S
Metallo et al.
Scheme 2. Synthesis of the Bifunctional Polymer PA-V-Fa
a
Conditions: (a) diaminobutane, HBTU, HOBT, DIEA; (b) 0.02 equiv of fluorescein cadaverine, triethylamine; (c) 0.05 equiv of 1; (d) aqueous NH4OH
(30%).
ence of specific functional groups on the surface of the SAM.
After the reaction of N-R-Ac-KDADA with the SAM, we
observed the appearance of bands at 1550 and 1660 cm-1
A solution of PBS was passed over the sensing surface for 15
min. Figure 1A demonstrates that neither pA nor pA-F bind to
SAMs presenting DADA. We also observed that pA-V-F does
not bind to single-component SAMs presenting tri(ethylene
glycol) groups (sensorgram not shown). These controls indicate
that this polymer does not bind to the SAM nonspecifically when
either pendant vancomycin or surface-displayed DADA is absent.
Figure 1A also demonstrates that pA-V-F binds to SAMs
presenting DADA. The polymer desorbed from the surface very
slowly in PBS buffer (koff ≈ 2 × 10-6 s-1) or in 10 mg/mL (35
mM) sodium dodecyl sulfate (SDS) (koff ≈ 3 × 10-6 s-1) (Figure
1B). The concentration of SDS is well above the critical micelle
concentration (cmc ) 8 mM)33 and is a concentration typically
used to dissociate proteins from a sensor chip. The rate of
dissociation of the polymer increased by a factor of ∼50 (to
koff ) 10-4 s-1) in the presence of the soluble ligand (0.5 mM
N-R,N-ꢀ-diAc-LKDADA). These sensorgrams indicate that pA-
V-F forms a tightly bound film on the surface of SAMs
presenting DADA groups, presumably through polyvalent in-
teractions between the multiple vancomycin groups on the
polymer and multiple DADA groups on the SAM. The increase
in the rate of dissociation of the polymer in the presence of
soluble ligand relative to the rate of dissociation in PBS suggests
that vancomycin groups on the polymer rebind to the surface
during the dissociation in the absence of the soluble DADA.34
We also measured the thickness of the layer of adsorbed pA-
V-F by using ellipsometry. The measured thickness (65 Å) lies
within the range reported for monolayers of adsorbed poly-
,
corresponding to the NH bending modes and the CdO stretch
of the amide bonds.30 We also estimated the extent of reaction
of DADA groups with the activated esters of the SAM crudely
by using ellipsometry. We observed an increase in the thickness
of the SAM of 2 ( 1 Å after the reaction with N-R-Ac-KDADA.
We infer that the yield for the coupling of N-R-Ac-KDADA to
the SAM was in the range 50-100%, a range similar to that
reported by Lahiri et al.30 for the coupling of a benzenesulfona-
mide ligand to a mixed SAM having the same composition.
Characterization of the Binding of Polymers to SAMs
Presenting DADA Groups. The binding of the polymers pA,
pA-F, and pA-V-F to SAMs presenting DADA was characterized
using SPR.20-23,32 The BIAcore 1000 instrument used in this
study reports changes in the resonance angle in resonance units
(RU). For most proteins, a change in the resonance angle of 1
RU corresponds to a change of ∼0.1 ng/cm2 in the quantity of
protein adsorbed to the surface.
Figure 1 shows SPR sensorgrams for the interaction of the
polymers (pA, pA-F, pA-V-F) with SAMs presenting DADA
groups (ø(DADA) ) 0.1). Each of these binding experiments
had three stages: (a) buffer (phosphate-buffered saline solution,
PBS) was passed over the sensing surface for 10 min. (b) A
solution of the polymer in PBS was passed over the sensing
surface for 30 minsa period of time that was sufficient for the
amount of polymer adsorbed to approach a constant value. (c)
(32) Chapman, R. G.; Ostuni, E.; Takayama, S.; Holmlin, R. E.; Yan, L.;
(33) Jacquier, J. C.; Desbene, P. L. J. Chromatogr., A 1996, 743, 307-314.
(34) Glaser, R. W. Anal. Biochem. 1993, 213, 152-161.
Whitesides, G. M. J. Am. Chem. Soc. 2000, 122, 8303-8304.
9
4536 J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003

Synthesis of Bifunctional Polymers
A R T I C L E S
Figure 2. SPR data for the binding of IgGF(60 nM) to pA-V-F adsorbed
onto a SAM presenting DADA (ø(DADA) ) 0.1). The error bars correspond
to our estimate of the error ((200 response units) in the amount of IgGF
adsorbed. This value represents half of the maximum range in the amount
of IgGF adsorbed (400 response units). The line is a fit to the data (Kd
)
m
8 µM, Kdd ) 14 nM). The inset shows sensorgrams for the binding of IgGF
(60 nM) in the presence of fluorescein (0, 1, 10, and 100 µM) and for the
binding of IgEDNP (60 nM) to the pA-V-F-coated SAM. The amount of
IgGF adsorbed to a pA-V-F-coated SAM in the absence of fluorescein (4100
RU) is close to the amount of IgGF adsorbed (3700 RU) on SAMs presenting
fluorescein groups (ø(fluorescein) ) 0.1) and corresponds to a monolayer
of IgGF.
d
stants for the monovalent (Kdm) and divalent (Kd ) binding of
IgGF to the fluorescein groups displayed on the surface.40,41 To
simplify the analysis, we assumed that the equilibrium constant
for the monovalent binding of IgGF to the fluorescein groups
displayed on the surface (Kdm) has the same magnitude as the
equilibrium constant for the monovalent interaction between
IgGF and soluble fluorescein (Kdm,soln). We also assumed that
there is no cooperativity between the binding sites of IgGF for
the binding to soluble fluorescein. Analysis of the data, subject
to these assumptions, yielded the line illustrated in Figure 2.
Figure 1. (A) SPR sensorgram showing binding of pA-V-F (10 µM
vancomycin, 2 µM fluorescein), pA-F (2 µM fluorescein), and pA to SAMs
presenting DADA groups. (B) SPR sensorgram showing the dissociation of
pA-V-F from a SAM presenting DADA groups in the presence of buffer,
buffer with sodium dodecyl sulfate (SDS), and buffer with 0.5 mM soluble
ligand N-R,Ν-ꢀ-diacetyl-L-lysine-D-alanine-D-alanine (N-R,N-ꢀ-diAc-LKDA-
DA).
mers.35-39 The presence of an adsorption plateau in the SPR
sensorgram for the adsorption of pA-V-F (Figure 1A) is also
consistent with the adsorption of a monolayer of pA-V-F.
Characterization of the Binding of Antibodies to pA-V-F
Adsorbed onto SAMs Presenting DADA Groups. Figure 1
indicates that pA-V-F adsorbs strongly to the surface of SAMs
presenting DADA groups. The pendant fluorescein groups of
the adsorbed pA-V-F should then be available for binding to
anti-fluorescein antibodies. SPR sensorgrams for the binding
of anti-fluorescein IgG (IgGF) to adsorbed pA-V-F are shown
in the inset to Figure 2. We observed that this antibody binds
to the adsorbed pA-V-F, and the binding of the antibody can
be inhibited by using soluble fluorescein. Anti-dinitrophenyl IgE
(IgEDNP) did not bind to the adsorbed pA-V-F (Figure 2). These
results indicate that the binding of the antibody is a result of
the biospecific interaction between IgGF and the fluorescein
groups displayed on the exposed region of the adsorbed polymer.
We analyzed the data (Figure 2) for the inhibition of the
binding of IgGF to the adsorbed pA-V-F in the presence of
soluble fluorescein to obtain equilibrium dissociation con-
m
d
The values of the dissociation constants (Kd ) 8 µM, Kd
)
14 nM) for the interaction between IgGF and the adsorbed pA-
m
V-F were close to the values of the dissociation constants (Kd
d
) 8 µM, Kd ) 2 nM) for the binding of IgGF to SAMs
presenting fluorescein groups (ø(fluorescein) ) 0.1; data not
shown). The enhancement in the affinity of IgGF for adsorbed
pA-V-F due to divalency is a factor of ∼570. This enhancement
in affinity lies in the range (100-1000) usually observed for
the interaction of divalent molecules with surface-bound
ligands.19,42-46
Sensorgrams for the dissociation of bound IgGF from the
adsorbed pA-V-F are shown in Figure 3. The rate of dissociation
of IgGF (koff ) 1.1 × 10-5 s-1 in PBS) increased in the presence
of soluble ligand (koff ) 3.6 × 10-4 s-1 in 10 µM fluorescein
in PBS; koff ) 3.9 × 10-3 s-1 in 1 mM fluorescein in PBS).
The increase in the rate of dissociation in the presence of soluble
(40) Rao, J.; Whitesides, G. M. J. Am. Chem. Soc. 1997, 119, 10286-10290.
(41) Pisarchick, M. L.; Thompson, N. L. Biophys. J. 1990, 58, 1235-1249.
(42) Karulin, A. Y.; Dzantiev, B. B. Mol. Immunol. 1990, 27, 965-971.
(43) Hornick, C. L.; Karush, F. Immunochemistry 1972, 9, 325-340.
(44) Greenbury, C. G.; Moore, D. H.; Nunn, L. C. Immunology 1965, 8, 420-
431.
(35) Shiratori, S. S.; Rubner, M. F. Macromolecules 2000, 33, 4213-4219.
(36) Killmann, E.; Eckart, R. Makromol. Chem. 1970, 132, 239-258.
(37) Klein, J. J. Colloid Interface Sci. 1986, 111, 305-313.
(38) Leermakers, F. A. M.; Gast, A. P. Macromolecules 1991, 24, 718-730.
(39) Shiratori and Rubner35 have reported that the thickness of a single layer of
adsorbed poly(acrylic acid) or poly(allylamine hydrochloride) can range
between 5 and 80 Å depending on the conditions under which deposition
is carried out.
(45) Karush, F. Compr. Immunol. 1978, 5, 85-116.
(46) Defrees, S. A.; Kosch, W.; Way, W.; Paulson, J. C.; Sabesan, S.; Halcomb,
R. L.; Huang, D.-H.; Ichikawa, Y.; Wong, C.-H. J. Am. Chem. Soc. 1995,
117, 66-79.
9
J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003 4537

A R T I C L E S
Metallo et al.
Figure 3. SPR sensorgram showing the dissociation of IgGF from pA-
V-F adsorbed onto a SAM presenting DADA in the presence of fluorescein
(0 µM, 10 µM, and 1 mM in PBS).
fluorescein suggests that the antibody rebinds to ligands on the
surface during the dissociation.
Figure 4. Detection of the binding of IgGF to the bifunctional polymer by
immunofluorescence after laminar flow patterning. (A) Top view of the
capillary network. The dotted line in the expanded view indicates the
interface between two streams flowing laminarly through the channel. The
arrows indicate the direction of the flow. (B) Fluorescence micrograph (green
fluorescence mode) illustrates that the polymer adsorbed uniformly in the
channel. (C) Fluorescence micrograph (red fluorescence mode) illustrates
that IgGFcAM has adsorbed to the surface only in the region where IgGF
was allowed to flow.
From the data shown in Figures 1-3, we conclude that pA-
V-F forms a molecular bridge between IgGF and SAMs
presenting DADA groups as a result of two independent, high-
affinity interactionssthe polyvalent interaction between pA-
V-F and DADA groups of the SAM and the divalent interaction
between IgGF and the adsorbed pA-V-F.
Immunofluorescence. The binding of antibodies to the
bifunctional polymer was also visualized by immunofluorescent
staining after laminar flow patterning.47,48 A poly(dimethyl-
siloxane) (PDMS) stamp having 300-µm channels molded in
its surface (Figure 4A) was placed in conformal contact with a
self-assembled monolayer presenting DADA groups (ø(DADA)
) 0.1). A solution of pA-V-F (10 µM vancomycin, 2 µM
fluorescein) was added to inlets 1 and 2 and was allowed to
flow through the channel (Figure 4B). The channel was rinsed
with PBS buffer. PBS buffer was then added to inlet 1, a solution
of IgGF was added to inlet 2, and these solutions were allowed
to flow through the channel. Finally, a solution of tetrameth-
ylrhodamine isothiocyanate (TRITC)-labeled Fc-specific, anti-
mouse, secondary antibody (IgGFcAM) was added to both inlets
and allowed to flow through the channel (Figure 4C). The
PDMS stamp was removed and the pattern of fluorescence
characterized by fluorescence microscopy. The fluorescence
micrograph (green fluorescence detection) shown in Figure 4B
illustrates that pA-V-F adsorbed uniformly in the channel. The
fluorescence micrograph (red fluorescence detection) shown in
Figure 4C illustrates that IgGFcAM adsorbed to the surface only
in the region where IgGF had been allowed to flow. This
experiment demonstrates that IgGF binds to pA-V-F adsorbed
on SAMs presenting DADA groups and that the bound IgGF is
a living bacterium presented sufficient numbers of accessible
LKDADA moieties in the cell wall to allow polyvalent interaction
with a vancomycin-presenting polymer. We determined the
ability of the bifunctional polymer to bind to the Gram-positive
bacteria E. faecalis (Figure 5A). By using the fluorescein
attached to pA-V-F as a fluorescent marker instead of an antigen,
we directly observed binding of the polymer to the target
bacteria.4,49 The bacteria were exposed to a red-fluorescent,
intercalating dye (SYTO 63) and to pA-V-F in PBS. After
several washings with PBS, the presence of the fluorescent
polymer on the bacteria was determined by fluorescence
microscopy. Comparison of the red fluorescence due to the
nucleic acid stain (Figure 5B) with the green fluorescence due
to fluorescein (Figure 5C), which indicates the presence of pA-
V-F, shows that the large majority of bacteria have detectable
amounts of bifunctional polymer bound to them. Some bacteria
either do not have polymer bound to them or have substantially
less polymer bound than the majority of the bacteria. Exposure
of E. faecalis to the control polymer, pA-F, yielded nonfluo-
rescent bacteria. When the Gram-negative bacterium E. coli was
exposed to pA-V-F, no fluorescence was observed (data not
shown). These experiments indicate that the bifunctional
polymer binds specifically to Gram-positive bacteria through
interactions with the vancomycin moiety.
recognized by soluble IgGFcAM
.
Binding of pA-V-F to Enterococcus faecalis. The SAM
model provided a reproducible, controlled system for experi-
mentation with biologically relevant polymer-surface interac-
tions that would be difficult to systematically study otherwise.
We wanted to confirm the relevance of the SAM model to
biological surfaces; specifically, we wanted to establish whether
Although the surface of a cell wall is much more complex
than the surface of the model SAMs, the bifunctional polymer
bound to the bacteria at densities similar to those found in the
experiments using SAMs and SPR, and with similar rates.
Furthermore, the polymer remained bound to the bacterial
surface through multiple rounds of washing; this adherence
(47) Kenis, P. J. A.; Ismagilov, R. F.; Whitesides, G. M. Science 1999, 285,
83-85.
(48) Takayama, S.; McDonald, J. C.; Ostuni, E.; Liang, M. N.; Kenis, P. J. A.;
Ismagilov, R. F.; Whitesides, G. M. Proc. Natl. Acad. Sci. U.S.A. 1999,
96, 5545-5548.
(49) Gordon, E. J.; Gestwicki, J. E.; Strong, L. E.; Kiessling, L. L. Chem., Biol.
2000, 7, 9-16.
9
4538 J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003

Synthesis of Bifunctional Polymers
A R T I C L E S
antibodies to the bacterial surface (Metallo and Whitesides,
unpublished results).
More generally, our work demonstrates that biospecific,
polyvalent interactions can be used to engineer molecular
recognition at a surface. The use of polyvalent interactions to
cause the adsorption of polyelectrolytes on charged surfaces is
well-established.50-52 Since most biological surfaces carry a net
charge, it is difficult to modify specific surfaces selectiVely by
adsorption of polyelectrolytes. The change in the types of
molecular recognition that occur at a surface, based on the
polyvalent strategy outlined in this paper, is selective, stable,
and can be carried out under mild conditions.
Modification using multifunctional polymers might enable
us to change the chemistry of various kinds of surfaces,
including those of mammalian cells.53 While our preliminary
efforts have been focused on designing polymers that adsorb
on bacteria, it should also be possible to design multifunctional
polymers that direct antibodies toward other types of cells.
Experimental Section
Materials. All materials and reagents were used as received.
N-hydroxysuccinimide (NHS), 1-(3-dimethylaminopropyl)-3-ethylcar-
bodiimide hydrochloride (EDC), O-benzotriazol-1-yl-N,N,N′,N′-tetra-
methyluroniumhexafluorophosphate (HBTU), 1-hydroxybenzotriazole
(HOBT), diaminobutane, and N,N-diisopropylethylamine (DIEA)were
purchased from Aldrich (Milwaukee, WI). Vancomycin (V-2002), N-R-
Ac-KDADA (A-6950), IgGF (F-5636), IgGFcAM (T-7657), and IgEDNP
(D-8406) were purchased from Sigma (St. Louis, MO). Fluorescein
cadaverine was purchased from Molecular Probes (Eugene, OR).
Anhydrous DMF and dimethyl sulfoxide (DMSO) were purchased from
EM Science (Gibbstown, NJ) and were used without further purification.
Silicon wafers were purchased from Silicon Sense (Nashua, NH).
Synthesis of the Vancomycin Conjugate 1. The synthesis of 1 was
carried out according to the method of Sundram et al.29 Diaminobutane
(5.3 mg, 60 µmol) was dissolved in 0.5 mL of DMSO and added to a
solution of vancomycin (50 mg, 34 µmol) in 0.5 mL of DMF. The
combined solution was cooled to 4 °C, and HBTU (19.2 mg, 50 µmol)
and HOBT (6.6 mg, 49 µmol) were added in 0.2 mL of DMF. DIEA
(10 µL, 57 µmol) was added, and the reaction mixture was allowed to
warm to room temperature and was stirred overnight. The precipitate
obtained by adding methylene chloride to the reaction mixture was
filtered, washed, and purified by RP-HPLC on a C18 column. The 1H
NMR showed resonances attributable to vancomycin as well as the
amide and methylenes expected for the linker. The electrospray MS
showed an (M + Na) ion at m/z ) 1540 (expected (M + Na): 1540).
Synthesis of the Polymers. PNAS was available from a previous
study.27 PNAS (50 mg, 296 µmol in monomer units) was dissolved in
1.5 mL of anhydrous DMF. Triethylamine (100 µL, 720 µmol) and a
solution of fluorescein cadaverine (3.8 mg, 5.8 µmol) in anhydrous
DMF were added. The reaction was stirred overnight in the dark. A
part (75%) of the reaction mixture was removed and quenched with
aqueous ammonium hydroxide (30%) to give pA-F. To the remainder
(74 µmol in monomer units) of the reaction mixture was added 5.6 mg
(3.7 µmol) of the vancomycin conjugate 1 in anhydrous DMF. The
reaction mixture was stirred overnight in the dark, and then 200 µL of
aqueous ammonium hydroxide (30%) was added to give pA-V-F.
Unfunctionalized polyacrylamide (pA) was synthesized by the addition
of ammonium hydroxide to a solution of PNAS in DMF. The polymers
were dialyzed against deionized water, 0.5 M NH4Cl, and deionized
Figure 5. (A) Schematic of polymer binding to the surface of a Gram-
positive bacterium (not to scale) (B) Fluorescence micrograph (red
fluorescence mode) of E. faecalis bacteria stained with the membrane-
permeable, nucleic acid dye SYTO 63; box indicates enlarged region (C)
Fluorescence micrograph (green fluorescence) of the same field as in (B),
in this case indicating fluorescence due to the presence of pA-V-F on the
bacteria; box indicates enlarged region.
indicates that the strong binding of the polymer to a polyvalent
substrate that was observed in the model system was retained
in the bacterial system.
Conclusions
We have synthesized bifunctional polymers (pA-V-F) that
bind biospecifically to SAMs presenting DADA groups and serve
as a molecular bridge between the SAM and an antibody (IgGF)
that does not otherwise bind to the surface. Experiments confirm
that, under conditions similar to those used in the model study,
the bifunctional polymer binds to the surface of E. faecalis.
Polymers with pendant antigens able to recruit antibodies to
the surfaces of Gram-positive bacteria (or other cells) might
influence the interaction of these cells with the immune system.
Of note was the extent and the stability of the interaction of
the IgG targeted to an antigen displayed by the polymer. The
antigen, fluorescein, was present on average every 100 polymer
units yet a surface on which this polymer was bound was able
to recruit the same amount of antibody with nearly the same
affinity as a surface presenting a high density (10%) of
fluorescein molecules. Perhaps due to the flexibility of the
polymer as well the large size and flexibility of the IgG
molecules it may not be necessary to display a dense array of
antigens at a surface in order to achieve a high degree of stable
binding by antibodies. Further work has indicated that the
bifunctional polymer has the ability to recruit anti-fluorescein
(50) Decher, G. Science 1997, 277, 1232-1237.
(51) Hammond, P. T. Curr. Opin. Colloid Interface Sci. 1999, 4, 430-442.
(52) Yoo, D.; Shiratori, S. S.; Rubner, M. F. Macromolecules 1998, 31, 4309-
4318.
(53) Elbert, D. L.; Hubbell, J. A. Chem., Biol. 1998, 5, 177-183.
9
J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003 4539

A R T I C L E S
Metallo et al.
water again and then stored frozen at -20 °C. While there is no
indication (from NMR of pA-V-F) that reaction between the nitrogen
of the hindered N-terminus or of the vancosamine moiety of 1 and
PNAS occurs, a low level of multipoint attachment to the polymer
backbone that would result from reaction at these additional sites cannot
be definitely excluded.
to proceed for 30 min at room temperature. The chips were removed
from the solution, washed with distilled deionized water and with
ethanol, and dried under a stream of nitrogen.
Surface Plasmon Resonance Spectroscopy. All SPR experiments
were carried out on a BIAcore 1000 instrument. Our substrates were
glued into the commercial BIAcore cartridges (after removing the
manufacturer’s substrate) and then docked into the BIAcore instrument.
PBS (10 mM phosphate, 138 mM NaCl, and 2.7 mM KCl, pH 7.4)
was prepared in distilled deionized water. Solutions of IgGF, pA, pA-
V-F, and pA-F were prepared in PBS.
Ellipsometry. Ellipsometric measurements were carried out on a
commercial ellipsometer (AutoEL II, Rudolph, Inc.). Substrates for
ellipsometry were prepared by evaporating titanium (1.5 nm, as an
adhesion layer) and then gold (100 nm) onto silicon wafers. We
determined the optical constants for each substrate. The substrates were
then immersed in ethanolic solutions of a mixture of HS(CH2)11(OCH2-
CH2)3OH (1.8 mM) and HS(CH2)11(OCH2CH2)6OCH2COOH (0.2 mM)
overnight. The coupling of N-R-Ac-KDADA to the mixed SAMs was
carried out as described above. The SAMs presenting DADA were
immersed in a solution of pA-V-F (10 µM vancomycin, 2 µM
fluorescein) in PBS for 6 h. The substrate was rinsed with PBS and
then with distilled water and dried under a stream of nitrogen.
Ellipsometric thicknesses were determined (at each step) using the
previously determined optical constants.
Characterization of Polymers. Poly(N-acryloyloxysuccinimide) was
hydrolyzed by a reaction with aqueous sodium hydroxide (0.1 N). The
polymer was dialyzed against deionized water, followed by lyophiliza-
tion, and the molecular weight distribution was characterized by GPC:
Mw ) 96 500; Mn ) 65 000; PDI ) 1.48; degree of polymerization
≈900. The fractional coupling of fluorescein and vancomycin on the
1
polymers was determined by H NMR and absorption spectroscopy.
Characteristic aromatic peaks of fluorescein were integrated and
compared with the integrated signal from the protons on the polyacryl-
amide backbone to determine the percent substitution with fluorescein.
The NMR signals from vancomycin on pA-V-F were too broad to
provide reliable integrated values, so the percent substitution was
determined based on the vancomycin-to-fluorescein ratio. Due to the
method of synthesis, the mole percent substitution of fluorescein on
the pA-V-F polymer was the same as on the pA-F polymer. The mole
percent vancomycin on pA-V-F was determined using its absorbance
at 280 nm, where both vancomycin and fluorescein absorb, and 450
nm, where only fluorescein absorbs. The molar absorptivity at 280 and
450 nm was determined from Beer’s law plots of vancomycin (ꢀ280
)
Polarized Infrared External Reflectance Spectroscopy. PIERS
spectra were measured using a DigiLab Fourier transform infrared
spectrometer (BioRad) equipped with a liquid N2-cooled mercury-
cadmium-telluride detector. The p-polarized light was incident at 80°
relative to the surface normal of the substrate. Substrates were prepared
in a manner identical to those used for ellipsometric measurements.
Bacterial Fluorescence. Bacteria (E. faecalis, ATCC 49332) were
grown overnight in brain-heart infusion broth at 37 °C, with shaking.
The bacteria were harvested by centrifugation and washed by resus-
pension in PBS and centrifugation. The bacteria were then resuspended
to an OD600 ) 1 in PBS containing 5 µM SYTO 63 (Molecular Probes)
and pA-V-F (15 µM in vancomycin, 3 µM in fluorescein). After 30
min at room temperature, 50 µL of the bacterial suspension was
transferred to an ELF spin filter (0.2-µm pore size, Molecular Probes)
and centrifuged to remove the polymer- and dye-containing solution.
The bacteria were washed by being resuspended in 200 µL of PBS
and centrifuged. Samples were washed three times. After the final
washing, the bacteria were resuspended in a one-to-one mixture of PBS
and Slowfade antifade reagent (Molecular Probes). Control samples
using SYTO 63 and pA-F, or only SYTO 63 were prepared using the
same procedure. Samples containing the Gram-negative bacteria E. coli
were prepared in the same manner as the E. faecalis samples. Bacterial
samples were imaged using a Leica confocal microscope equipped with
an Ar-Kr laser.
6030 M-1 cm-1; ꢀ450 ) 0 M-1 cm-1) and fluorescein cadaverine (ꢀ280
) 20 800 M-1 cm-1; ꢀ450 ) 20 600 M-1 cm-1). From these values and
the absorbance of pA-V-F at the two wavelengths, the concentration
of vancomycin and fluorescein was determined. The ratio of their
concentrations, in conjunction with the mole percent substitution of
fluorescein on the polymer, yielded the mole percent vancomycin. The
polymers used in this work (pA, pA-V-F, pA-F) are soluble in water
and PBS.
Preparation of Gold Substrates Presenting Mixed SAMs. Gold
substrates were prepared by evaporating titanium (1.5 nm, as an
adhesion layer) and then gold (38 nm) onto glass cover slips (0.2 mm
thick, No. 2, Corning).30 Gold substrates were immersed in ethanolic
solutions of a mixture of HS(CH2)11(OCH2CH2)3OH (1.8 mM) and HS-
(CH2)11(OCH2CH2)6OCH2COOH (0.2 mM) overnight.30
Coupling Ligands to Carboxylic Acid Groups in Mixed SAMs.
The tripeptide N-R-Ac-KDADA was coupled to mixed SAMs by amide
bond formation between the ꢀ-amino group of lysine and the carboxylic
groups on the surface of mixed SAMs. The procedure was described
previously by Lahiri et al.30 Briefly, the SAM-coated gold chips were
immersed in a solution of EDC (0.1 M) and NHS (0.4 M) in deionized
water for 7 min at room temperature; this treatment converted the
carboxylic acid groups into NHS active esters. The chips were removed
from the solution of EDC and NHS and immersed without rinsing in
a phosphate-buffered (25 mM; pH 8.1) aqueous solution of the tripeptide
(2 mg/mL); the coupling reaction was allowed to proceed for 30 min
at room temperature. The chips were removed from the solution of
peptide, washed with distilled, deionized water, and with ethanol, and
dried under a stream of nitrogen. Fluorescein cadaverine was coupled
to the mixed SAMs using a similar procedure. Briefly, after activating
the carboxylic acid groups of the mixed SAMs as NHS esters, the chips
were dipped into a solution of fluorescein cadaverine (2 mg/mL) and
triethylamine (10 µL/mL) in DMF. The coupling reaction was allowed
Acknowledgment. This research was supported by the
National Institutes of Health (NIH GM 30367), DARPA, and
Space and Naval Warfare Systems Center, San Diego. S.J.M.
and R.E.H. acknowledge postdoctoral fellowships from the NIH.
We thank Professor Paul Laibinis (MIT) for use of the FTIR
instrument in his laboratory.
JA030045A
9
4540 J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003