UC Berkeley
UC Berkeley Previously Published Works
Title
Molecular recognition using corona phase complexes made of synthetic polymers adsorbed
on carbon nanotubes.
Permalink
https://escholarship.org/uc/item/1072p670
Journal
Nature nanotechnology, 8(12)
ISSN
1748-3387
Authors
Zhang, Jingqing
Landry, Markita P
Barone, Paul W
et al.
Publication Date
2013-12-01
DOI
10.1038/nnano.2013.236
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
Molecular recognition using corona phase
complexes made of synthetic polymers adsorbed
on carbon nanotubes
Jingqing Zhang
1
‡
, Markita P. Landry
1
‡
, Paul W. Barone
1
‡
,Jong-HoKim
1,2
‡
et al.*
Understanding molecular recognition is of fundamental importance in applications such as therapeutics, chemical catalysis
and sensor design. The most common recognition motifs involve biological macromolecules such as antibodies and
aptamers. The key to biorecognition consists of a unique three-dimensional structure formed by a folded and constrained
bioheteropolymer that creates a binding pocket, or an interface, able to recognize a specific molecule. Here, we show that
synthetic heteropolymers, once constrained onto a single-walled carbon nanotube by chemical adsorption, also form a new
corona phase that exhibits highly selective recognition for specific molecules. To prove the generality of this phenomenon,
we report three examples of heteropolymer–nanotube recognition complexes for riboflavin,
L
-thyroxine and oestradiol. In
each case, the recognition was predicted using a two-dimensional thermodynamic model of surface interactions in which
the dissociation constants can be tuned by perturbing the chemical structure of the heteropolymer. Moreover, these
complexes can be used as new types of spatiotemporal sensors based on modulation of the carbon nanotube
photoemission in the near-infrared, as we show by tracking riboflavin diffusion in murine macrophages.
M
olecular recognition and signal transduction are the two
main challenges in sensor design
1,2
. Frequently, scientists
and engineers borrow from nature to gain analyte speci-
ficity and sensitivity, using natural antibodies as vital components
of the sensors
1,3–6
. However, antibodies are expensive, fragile,
easily lose biological activity on external treatment (such as immo-
bilization) and exhibit batch-dependent variation, limiting their use
in widespread applications
1,6,7
. This has driven the search for
methods to synthesize or discover artificial antibodies or their ana-
logues from polymeric materials, leading to molecularly imprinted
polymers
6,8
and DNA-aptamers
2,3,6
.
Single-walled carbon nanotubes (SWNTs), rolled cylinders of
graphene, are ideal materials for molecular recognition and signal
transduction. They have near-infrared bandgap fluorescence
9
with
no photobleaching threshold
10,11
, and are sensitive to the surround-
ing environment. Electron-donating or -accepting groups can
increase or decrease emission
12–14
with single-molecule sensi-
tivity
15–17
. Adsorbates can also screen the one-dimensional confined
exciton
18,19
, causing solvatochromic shifts in emission
20,21
, particu-
larly with polymer wrappingsthat change conformation. Our labora-
tory and others have developed SWNT fluorescence sensors for
detecting b-
D-glucose
12
, DNA hybridization
21
, divalent metal
cations
20
, assorted genotoxins
22
, nitroaromatics
23
, nitric oxide
13,17
,
pH
16
and the protein avidin
14
. These efforts have invariably exploited
conventional molecular recognition entities, such as enzymes
24
, oli-
gonucleotides
21
or specific functional groups with known affinity for
the target molecule
13
. In this work, we instead demonstrate a new
recognition motif consisting of specific polymers adsorbed onto a
SWNT, whereby the interface recognizes certain small-molecule
adsorbates, resulting in modulation of nanotube fluorescence
(Fig. 1). We show that the recognition results from the combination
of analyte adsorption to the graphene surface and lateral interactions
with molecules in the adsorbed phase in a predictable manner. We
introduce this concept as corona phase molecular recognition.
Screening of a polymer–SWNT library
To generate polymer–nanotube interfaces in aqueous solution,
candidate amphiphilic polymers were synthesized with both hydro-
phobic and hydrophilic domains, enabling SWNT surface adsorp-
tion and entropic stabilization, respectively. For simplicity, only
non-ionic or weakly ionic polymer phases were used in this study,
to assist structural understanding. For screening, each polymer–
SWNT construct was exposed to a library of 36 small molecules,
and the resulting SWNT fluorescence response was monitored in
triplicate using a near-infrared fluorescence spectrometer, auto-
mated to acquire data in a 96-well-plate format (Supplementary
Fig. 4). Each spectrum was deconvoluted into eight SWNT chiral-
ities using a custom spectral fitting program (Supplementary
section ‘Materials and Methods’) that also calculates standard
errors. The resulting intensity and wavelength modulations of
each construct before and after the addition of each analyte in the
library were recorded (for all unprocessed spectra see
Supplementary Figs 23–46).
Molecular recognition by polymer corona phase on SWNTs
From the polymer–SWNT library screened to date we identified
three distinct examples of polymer–nanotube-mediated molecular
recognition. The simplest example is provided by a rhodamine iso-
thiocyanate-difunctionalized poly(ethylene glycol)–SWNT complex
(RITC-PEG-RITC–SWNT; Fig. 2a, 5 or 20 kDa), which exhibits flu-
orescence quenching exclusively in response to oestradiol (both
alpha- and beta-oestradiol). Two similar synthetic variants—fluor-
escein isothiocyanate-difunctionalized PEG (FITC-PEG-FITC;
Fig. 2b) and distearyl phosphatidylethanolamine PEG (PE-PEG;
Fig. 2c)—result in non-selective response profiles. The simple
polymer design allows for elucidation of the adsorbed phase struc-
ture. The selectivity of this response (Fig. 2a) is distinct from
schemes using principle component analysis
22
or differential
sensor responses
25
for analyte recognition, and we therefore assign
*A full list of authors and their affiliations appears at the end of the paper.
ARTICLES
PUBLISHED ONLINE: 24 NOVEMBER 2013 | DOI: 10.1038/NNANO.2013.236
NATURE NANOTECHNOLOGY | VOL 8 | DECEMBER 2013 | www.nature.com/naturenanotechnology 959
© 2013 Macmillan Publishers Limited. All rights reserved.
it as the first demonstration of molecular recognition from the
adsorbed phase itself. In the polymer–SWNT complex, the FITC
fluorescence is quenched, providing evidence that the hydrophobic
ends of the polymer are adsorbed onto the SWNT surface
(Supplementary Fig. 8). The hydrophilic PEG chain is expected to
form a loop extending into solution. Such a partitioning of hydro-
phobic and hydrophilic domains is consistent with earlier studies
that have utilized PEG-conjugated hydrophobic molecules for
SWNT suspension
26–32
, and is supported by the experimental obser-
vation that PEG itself is incapable of suspending SWNTs. Although
we do not have direct evidence of the formation of the loop
structure, single-molecule fluorescent imaging of the polymer
co-localized with the fluorescent SWNT is consistent and further
supports both this structure and the interaction of oestradiol
with the hydrophobic anchors (Supplementary Figs 9–12,
Supplementary Tables 4–6).
This structural configuration is also corroborated by molecular
dynamics simulations (Supplementary section ‘Materials and
Methods’) for the three polymers (Fig. 2). Oestradiol contains an
aromatic group that can adsorb to the SWNT surface if the
pendant hydroxyl groups are hydrogen-bonded to the adjacent
RITC amides. The molecular weight of the PEG chain does not
influence the selectivity (Supplementary Figs 13–15), but the end-
group structure has a substantial effect (Fig. 2a–c). The molecular
recognition of the RITC-PEG-RITC–SWNT to oestradiol results
from the strong interaction between the oestradiol and RITC
anchors. A molecular dynamics simulation at a starting coverage
of 50 polymers per 20 nm of SWNT reveals that, at equilibrium,
80% of the SWNT surface is covered by RITC anchors. We conclude
that the intermolecular spacing of hydrophobic groups on the
SWNT determines selectivity. Different characterization techniques
have been applied here: dry-state atomic force microscopy (AFM)
estimated a mean radius of 2.5 nm for RITC-PEG-RITC–SWNT,
consistent with the 2.7 nm estimated from molecular dynamics
results, but smaller than the 8.1 nm that is estimated from single-
particle tracking experiments (Supplementary Table 3). Under the
hydrated conditions of single-particle tracking, the hydrodynamic
radius can be increased. While AFM provides a direct measure of
the complex radius, single-particle tracking can be influenced by
the variations in nanotube length. Transmission electron
microscopy (TEM) images (Supplementary Fig. 17a) indicate a
range of radii from bare surface (0.47 nm) to approximately twice
the AFM average value (4.9 nm), reflecting variable surface coverage
along the length.
We also compared this with responses obtained from SWNTs
suspended with two commonly used dispersing agents. Sodium
cholate suspended SWNT (Fig. 2d) is non-responsive to the analytes
tested, confirming the tight surface packing reported previously
33
.
In contrast, d(GT)
15
DNA-wrapped SWNT (Fig. 3d) shows poor
selectivity, suggesting a more porous interface composed of
consecutive DNA helices
34
. This type of profile, which informs
the accessibility of molecules to the SWNT surface, as presented
in Fig. 2, allows for a unique ‘fingerprinting’ of polymer adsorbed
phases in a manner inaccessible to other analytical techniques such
as NMR, Fourier transform infrared (FTIR) spectroscopy and
ultraviolet absorption. Structural schematics of the nanotube
complexes were deduced from a combination of polymer mole-
cular structure and fingerprinted response profile, further sup-
ported by single-particle tracking data and molecular dynamics
results (Fig. 2).
A more complex example is provided by the PEG brush
described in Fig. 3. One brush segment is alkylated for hydrophobi-
city while the remaining sites can display a variety of functionalities.
When these sites contain Fmoc
L-phenylalanine (Fmoc-Phe-
PPEG8), we find that a seven-membered brush structure recognizes
L-thyroxine (Fig. 3a), whereas a three-membered analogue does not
(Fig. 3b). Replacing the Fmoc
L-phenylalanine with amine groups
also results in a loss of selectivity (NH
2
-PPEG8, Fig. 3c) due to
Polymer
Nanotube
Hydrophilic
Hydrophobic
a
b
Phenoxy-functionalized dextran
Hydrophobic
Hydrophilic
Analyte
O
OH
OH
O
O
O
OH
OH
O
O
O
OH
O
OH
OH
O
OH
HN
O
B
OH
OH
m
p
n
Figure 1 | Schematic of the molecular recognition concept. a, A polymer with an alternating hydrophobic and hydrophilic sequence adopts a specific
conformation when adsorbed to the nanotube. The polymer is pinned in place to create a selectiv e molecular recognition site for the molecule of interest,
leading to either a wavelength or intensity change in SWNT fluorescence. b, An example of a hydrophobic–hydr ophilic alternating sequence for bor onic acid-
derivatized phenylated dextran.
ARTICLES
NATURE NANOTECHNOLOGY DOI: 10.1038/NNANO.2013.236
NATURE NANOTECHNOLOGY | VOL 8 | DECEMBER 2013 | www.nature.com/naturenanotechnology960
© 2013 Macmillan Publishers Limited. All rights reserved.
the decrease in polymer hydrophobicity. For Fmoc-Phe-PPEG8–
SWNT, both the aliphatic chain and the Fmoc group adsorb onto
the SWNT. We propose that the strong interaction between Fmoc
and
L-thyroxine leads to the molecular recognition. Analogous to
the previous example, we hypothesize that the PEG groups form
loops, extending into the aqueous phase, imparting colloidal
stability. Additionally, steric hindrance between the PEG arms
influences how the Fmoc ends pack on the SWNT surface,
providing additional control of the footprint on the SWNT
surface. For NH
2
-PPEG8–SWNT, only the aliphatic chain is
adsorbed on the surface of the SWNT, while the amine group,
mostly protonated at pH 7.4, will extend into the aqueous phase.
The less selective response profile of NH
2
-PPEG8–SWNT suggests
that the nanotube surface is more exposed compared to the
Fmoc-Phe-PPEG8–SWNT and is also consistent with the polymeric
structure. In addition, single-particle tracking estimates a radius of
–1.0
–0.5
0.0
0.5
1.0
Relative intensity change
FITC-PEG-FITC–SWNT
–1.0
–0.5
0.0
0.5
1.0
Relative intensity change
RITC-PEG-RITC–SWNT
Control
DMSO
17-Alpha-oestradiol
2,4-Dinitrophenol
Acetylcholine chloride
Alpha-tocopherol
Adenosine
ATP
cAMP
Creatinine
Cytidine
D-Aspartic acid
D-Fructose
D-Galactose
D-Glucose
D-Mannose
Dopamine
Glycine
Guanosine
Histamine
L-Ascorbic acid
L-Citrulline
L-Histidine
L-Thyroxine
Melatonin
NADH
Quinine
Riboflavin
Salicylic acid
Serotonin
Sodium azide
Sodium pyruvate
Sucrose
Thymidine
Tryptophan
Tyramine
Urea
–1.0
–0.5
0.0
0.5
1.0
Relative intensity change
Sodium cholate–SWNT
a
b
c
–1.0
–0.5
0.0
0.5
1.0
Relative intensity change
PE-PEG–SWNT
Oestradiol
d
RITC-PEG-RITC
FITC-PEG-FITC
PE-PEG
Sodium cholate
HO
H
H
OH
H
H
N
O
H
N
H
N
S
H
N
S
n
O
–
–
+
+
O
O
N
N
O
N
N
O
O
H
N
O
H
N
H
N
S
O
O
OH
OH
O
H
N
S
O
O
O
HO
HO
n
H
O
O
O
O
P
O
–
O
OO
O
O
N
H
O
113
HO
H
H
HH
H
O
–
O
OH
OH
Figure 2 | Construct that selectively recognizes oestradiol, and non-selective mutants. a, RITC-PEG-RITC–SWNT enables selective recognition of
oestr adiol via a selective quenching response on exposur e to 100
m
M oestradiol. b–d, Chemical mutants FITC-PEG-FITC–SWNT (b), PE-PEG–SWNT (c),
and sodium cholate–SWNT (d) lose sensitivity, selectivity, or both, to oestradiol. To the left of the bar charts are the polymer structure (top left),
schematics of the polymer–SWNT complex (top right), and front (bottom left) and side (bottom right) views calculated from molecular dynamics
simulations. The schematics ar e deduced from a combination of polymer molecular structur e and fingerprinted response profile, further supported by
single-particle tracking data and molecular dynamics results. Colour coding for molecular dynamics simulation results are provided in the Supplementary
Information. Bar charts sho w intensity change of these complexes against a panel of 36 biological molecules. The error bars represent two standard
deviations of triplicate measurements. Data repr esent (7,5) SWNT species.
NATURE NANOTECHNOLOGY DOI: 10.1038/NNANO.2013.236
ARTICLES
NATURE NANOTECHNOLOGY | VOL 8 | DECEMBER 2013 | www.nature.com/naturenanotechnology 961
© 2013 Macmillan Publishers Limited. All rights reserved.
1.82 nm for the Fmoc-Phe-PPEG8–SWNT construct, consistent
with the molecular dynamics estimation of 2.5 nm. However, radii
estimated from TEM (Supplementary Fig. 17) and AFM
(Supplementary Fig. 16) are twice as large, suggesting free
polymer adsorption onto the SWNT surface during the drying
process for AFM and TEM sample preparation.
Yet another example is provided by a 53 mol/mol boronic acid-
substituted phenoxy-dextran-wrapped SWNT (BA-PhO-Dex–
SWNT, Fig. 4). The emission maximum of the complex redshifts
11 nm on addition of riboflavin, but not with the other diol-
containing substrates that typically bind to boronic acids
(Supplementary Fig. 18). The response is reversed in particular by
the addition of a riboflavin-binding protein, which competitively
inhibits riboflavin–SWNT binding (Supplementary Fig. 19). The
optical modulation in this example is a wavelength redshift
instead of fluorescence quenching as above. This type of response,
attributed to solvatochromism
35
induced by a polymer dielectric
change, is rare among the systems examined to date. For instance,
poly(vinyl alcohol)-wrapped SWNT (PVA–SWNT, Fig. 4c) shows
an intensity response to a number of analytes, including dopamine,
–1.0
–0.5
0.0
0.5
1.0
Relative intensity change
Fmoc-Phe-PPEG8–SWNT
a
b
c
d
–1.0
–0.5
0.0
0.5
1.0
Relative intensity change
Fmoc-Phe-PPEG4–SWNT
–1.0
–0.5
0.0
0.5
1.0
Relative intensity change
NH
2
-PPEG8–SWNT
Control
DMSO
17-Alpha-oestradiol
2,4-Dinitrophenol
Acetylcholine chloride
Alpha-tocopherol
Adenosine
ATP
cAMP
Creatinine
Cytidine
D-Aspartic acid
D-Fructose
D-Galactose
D-Glucose
D-Mannose
Dopamine
Glycine
Guanosine
Histamine
L-Ascorbic acid
L-Citrulline
L-Histidine
L-Thyroxine
Melatonin
NADH
Quinine
Riboflavin
Salicylic acid
Serotonin
Sodium azide
Sodium pyruvate
Sucrose
Thymidine
Tryptophan
Tyramine
Urea
–1.0
–0.5
0.0
0.5
1.0
Relative intensity change
GT
15
–SWNT
L-Thyroxine
Fmoc-Phe-PPEG8
Fmoc-Phe-PPEG4
NH
2
-PPEG8
(GT)
15
O
OH
O
HO
NH
2
I
I
I
I
N
H
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
n
O
n
O
n
O
n
O
n
O
n
O
n
HN
HN
HN
HN
HN
N
H
Fmoc
O
O
O
O
O
N
H
Fmoc
N
H
Fmoc
N
H
Fmoc
N
H
O
O
n
12
Fmoc
Fmoc
HN
N
H
Fmoc
O
HN
N
H
Fmoc
O
N
H
O
O
O
O
O
n
12
n
NH
O
H
N
O
O
O
O
n
NH
O
NH
O
O
O
n
HN
O
NH
O
O
N
H
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
NH
2
n
O
NH
2
O
NH
2
O
NH
2
O
NH
2
O
NH
2
O
NH
2
12
n
n
n
n
n
n
n
NH
N
O
O
O
P
O
O
O
O
P
O
O
O
O
P
O
O
O
O
O
N
H
N
N
N
O
H
2
N
15
Figure 3 | Construct that selectively recognizes L-thyro xine, and non-selective mutants. a, Fmoc-Phe-PPEG8–SWNT enables selective recognition of
L-thyroxine via a selective quenching response on exposur e to 100
m
M L-thyr oxine. b–d, Chemical mutants Fmoc-Phe-PPEG4–SWNT (b), NH
2
-PPEG–SWNT
(c), and (GT)
15
–SWNT (d) lose sensitivity , selectivity , or both, to L-thyroxine, as shown in the bar charts showing the intensity response of these complexes
on exposur e to same libr ary of molecules. The error bars represent standard errors of triplicate measur ements.
ARTICLES
NATURE NANOTECHNOLOGY DOI: 10.1038/NNANO.2013.236
NATURE NANOTECHNOLOGY | VOL 8 | DECEMBER 2013 | www.nature.com/naturenanotechnology962
© 2013 Macmillan Publishers Limited. All rights reserved.
