Polymerization amplified detection for nanoparticlebased biosensing
Adam J. Gormley‡, Robert Chapman‡, Molly M. Stevens
Department of Materials, Department of Bioengineering, and Institute for Biomedical
Engineering, Imperial College London, London, United Kingdom
ABSTRACT: Efficient signal amplification processes are key to the design of sensitive assays
for biomolecule detection. Here, we describe a new assay platform that takes advantage of both
polymerization reactions and the aggregation of nanoparticles to amplify signal. In our design, a
cascade is set up in which radicals generated by either enzymes or metal ions are polymerized to
form polymers that can entangle multiple gold nanoparticles (AuNPs) into aggregates, resulting
in a visible color change. Less than 0.05% monomer to polymer conversion is required to initiate
aggregation, providing high sensitivity towards the radical generating species. Good sensitivity
of this assay towards horseradish peroxidase, catalase, and parts per billion concentrations of
iron and copper is shown. Incorporation of the oxygen consuming enzyme glucose oxidase,
enables this assay to be performed in open air conditions at ambient temperature. We anticipate
that such a design will provide a useful platform for sensitive detection of a broad range of
biomolecules through polymerization-based amplification.
1
KEYWORDS: Gold nanoparticles, polymerization based amplification, biosensing, glucose
oxidase, catalase, horseradish peroxidase
Graphical Abstract
In recent years, a number of optical biosensors have been developed based on the controlled
growth and assembly of gold nanoparticles (AuNPs).1-3 These systems take advantage of distinct
changes in localized surface plasmon resonance (LSPR)
that occur when nanoparticles
aggregate, resulting in red-to-blue color shifts of the bulk solution. Due to the high extinction
coefficient of AuNPs and the ease of surface functionalization, these sensors can made to be
highly sensitive to the presence of a range of target molecules. A number of aggregation
mechanisms have been developed including DNA hybridization,4 peptide folding,5 streptavidinbiotin binding,6 and aptamer-target complexation.7 Polymers are frequently used to control the
stability and assembly of nanoparticle dispersions. Dense coatings of polymers such as
poly(ethylene glycol) (PEG) are often used to prevent protein adsorption, immune detection and
colloidal instability.8 However, exposure to very small concentrations of polymer is also capable
of bridging nanoparticles together resulting in entanglement and aggregation.9 This
2
nanoparticle/polymer assembly process can be driven by charge-charge interactions, hydrogen
bonding or more specific molecular interactions. Such assemblies have useful applications in
drug delivery, biosensing and catalysis.10
Enzymes are another convenient tool for signal amplification. In most cases, as in the enzymelinked immunosorbent assay (ELISA), enzymes catalyze the oxidation/reduction of colored
molecules to generate signal. Coupling enzymatic assays with nanoparticle based systems such
that nanoparticle growth or aggregation is altered by the enzyme, has emerged as a highly
attractive signal amplification route.11-13 Previous work by our group has shown that coupling
nanoparticle growth assays with enzymes such as glucose oxidase (GOx) and catalase can
provide sensitivities of ~10-18 g.ml-1 of target protein in whole serum.14, 15 With these systems,
the enzyme is used to modulate the peroxide concentration, which dramatically alters the kinetics
or mode of crystal growth and therefore the size, dispersity and color of the resulting
nanoparticle suspension. Similarly, some systems have also used nanoparticles as sensors for
hydrogen peroxide.16-18 Other enzyme based systems cleave or dimerize crosslinks such as
peptides to initiate aggregation, dispersion or fluorescence resonance energy transfer (FRET)
between dye molecules and the nanoparticle.19, 20
Free radical polymerization reactions offer an attractive signal amplification platform because
of their sensitivity to the presence of very low concentrations of radicals.21, 22 When a free radical
is generated and transferred to a vinyl containing monomer, rapid step-growth polymerization
results in the formation of large polymer chains. This cascade, therefore, may act to amplify any
events that result in the generation of radicals. However, limitations in appropriate readout
mechanisms and the sensitivity of these reactions towards small amounts of oxygen have
restricted the utility of these approaches to date.
3
In the work presented here we take advantage of these amplification techniques by using
enzymes to generate polymer via free-radical polymerization of a cationic monomer to control
gold nanoparticle aggregation (Figure 1). Through the use of GOx, polymerizations can be
performed in open well plates under normal atmosphere. The polymer produced is used to trigger
the aggregation of negatively charged AuNPs, resulting in a visible color change. Such a design
takes advantage of polymerization-based signal amplification wherein a single enzyme event
results in the dynamic growth of macromolecular chains.21 As only a small amount of polymer is
required to initiate aggregation, this assay is highly sensitive to the presence of any radical
generating species such as enzymes.23-27 We demonstrate the use of this assay to detect low
concentrations of two different enzymes, horseradish peroxidase (HRP) and catalase, as well as
parts per billion concentrations of iron and copper with the naked eye (Figure 1).
Figure 1. Schematic illustration of the assay design which takes advantage of polymerization
based signal amplification. Enzymes commonly used for biosensing applications such as HRP
are capable of generating free radicals that can be used to trigger the dynamic growth of polymer
chains. Because of the extreme sensitivity of AuNPs to aggregation by cationic polymers, low
monomer to polymer conversion (<0.05%) is able to generate a visible color change. So that the
assay can function in open, oxygen exposed well plates, GOx is used to degas the solution and
4
provide the peroxide required by the HRP. TEM images are shown of dispersed (top) and
aggregated (bottom) nanoparticles. Scale bar 100 nm.
Since this assay design relies on the use of polymer to trigger AuNP aggregation in the
presence of unconverted monomer, we began by comparing the ability of various polymers of 3aminopropyl methacrylamide (APMA) to aggregate AuNPs with that of free monomer (Figure
2). Aggregation was observed by changes in the absorbance spectra and confirmed by
transmission electron microscopy (TEM, see Figure 1 and S1). Aggregation of both 5 nm (Figure
S2) and 20 nm (Figure 2b) citrate capped AuNPs was observed at very low polymer
concentrations (>10-4 mg/ml). By controlling the polymer molecular weight using reversible
addition chain transfer (RAFT) polymerization, the concentration at which aggregation by the
polymer occurred was determined to be independent of the polymer molecular weight. Even
short polymers of 27 monomer units were able to aggregate the AuNPs at similar concentrations
to polymers generated by uncontrolled free radical polymerization (FRP). The monomer was
also able to cause aggregation of the AuNPs, but only at three orders of magnitude higher
concentrations.
5
Figure 2. (a) Absorbance spectra of dispersed and aggregated AuNPs due to the presence of
pAPMA. (b-c) Aggregation of citrate capped AuNPs as monitored by a ratio of absorbance
(600/530 nm) in the presence of varying concentrations of monomer and polymers of different
lengths (DP) (b) or generated by HRP over time (c). While the monomer is only able to induce
aggregation at high concentrations, aggregation by polymer occurs at extremely low
concentrations independent of chain length. (d) Sensitivity of AuNPs functionalized with peptide
(C(GEP)3), polymer (pAA), or DNA to aggregation by monomer and polymer. Steric protection
against monomer driven aggregation was provided in all cases while maintaining sensitivity to
polymers, particularly in the case of DNA.
It is well known that enzymes such as HRP and laccase are capable of generating free radicals
that can initiate polymerization.23-30 These enzymes act by oxidation of small molecule mediators
such as acetylacetone (acac) in the presence of hydrogen peroxide.31 Therefore, to test if
polymers generated by this mechanism would also aggregate the AuNPs, HRP was used to
polymerize APMA in water deoxygenated by argon bubbling. Peroxide and acac were included
as the enzyme substrates, and the reaction was carried out at 30 °C for 1 - 24 h prior to
incubation with the AuNPs. With this method, 60% conversion was achieved within 6 h (Figure
S3). Even at low conversions (<10% at 1 h), the unpurified reaction mixture was able to
aggregate AuNPs at concentrations four orders of magnitude lower than the un-polymerized
control (Figure 2c). This data complements that seen in Figure 2b suggesting that aggregation is
possible at < 0.1% polymer conversion in the presence of monomer.
Since the concentration of monomer required to cause aggregation of the citrate capped AuNPs
(10-2 mg/ml, 5.6 x 10-5 M) is low compared to that required to produce any reasonable rate of
6
polymerization (0.25 - 1 M), it was necessary to modify the surface of the AuNPs to prevent
monomer driven aggregation. Monomer driven aggregation is most likely due to surface charge
neutralization, so to provide steric protection the AuNPs were functionalized with either a
negatively charged peptide (C(GEP)3), polymer (thiolated poly(acrylic acid) or pAA) or DNA.
Each of these AuNPs were found to be stable against up to 10 mg/ml of the monomer while
retaining good sensitivity to the polymers (Figure 2d). Based on this difference in sensitivity, a
theoretical detection limit of less than 0.05% conversion was estimated. In the case of the citrate,
peptide and polymer functionalized AuNPs, polymer concentrations in excess of 10-2 mg/ml
conferred stability presumably due to steric stabilization. Such stabilization was not seen with the
DNA functionalized AuNPs at the concentration ranges tested. For this reason we used the DNA
coated AuNPs in all further experiments.
7
Figure 3. (a) Modeled oxygen concentration (mmol.ml-1) as a function of time and depth in a
standard 96 well plate. b) Expected oxygen concentration at 200, 100 and 50 nM GOx, and (c) as
a function of time at 200 nM. (d) Absorbance spectra of the AuNPs after addition to the assay
reaction mixture at varying concentrations of GOx and HRP, providing experimental validation
of the model. Polymerization and therefore complete degassing is only possible when the GOx
concentration is above 50 nM.
As oxygen is a potent radical quencher, it was necessary to incorporate a mechanism for
scrubbing all oxygen from solution so that sensitivity at low HRP concentrations in an open air
assay format was possible. For this reason, GOx was added to the reaction mixture to
simultaneously deoxygenate the solution and provide peroxide for the HRP (Figure 3a). By
modeling the kinetics of this reaction in one-dimension against time and distance from the
solution surface, it was determined that the consumption of oxygen by GOx at concentrations of
100-200 nM should be much faster than the diffusion of oxygen from the top of the solution
(Figure 3b). At 200 nM GOx, almost all of the dissolved oxygen is expected to be consumed
after only 5 min. At this point, steady state is reached and only the top third of the well volume
should contain more than 10 µM of oxygen (Figure 3c). Thus in the absence of stirring or
excessive convection it should be possible to perform the polymerization in an open well plate.
To test these calculations, polymerization reactions were performed in 96-well plates
containing 300 µl of GOx, APMA (0.25 M) and acac (0.4 mM) in a MES buffer (20 mM, pH
7.5) at 30°C for 30 min. To stop the reaction, the solution was diluted in oxygenated water (1:1
in most cases) and immediately added to a suspension of DNA-AuNPs. A large excess of
glucose (100 mM) was also provided to ensure that oxygen was at all times the limiting reagent
8
for the GOx. Sealing the top of the well plate with plastic tape was not found to affect the
reaction, and so all polymerizations were carried out in wells open to the atmosphere. Under
these conditions we observed that 200 nM GOx was able to sufficiently deoxygenate the solution
for HRP (1 µg/ml) to generate enough polymer to aggregate the AuNPs (Figure 3d). Similar
aggregation and therefore polymer generation was observed in the presence of 100 nM GOx,
though the assay was found to be less robust at this concentration. When GOx concentrations
less than or equal to 50 nM were tested, aggregation was entirely inhibited indicating that these
concentrations are insufficient to properly deoxygenate the solution as was expected from the
oxygen diffusion. Similarly, incubating the polymerization solution with 200 nM GOx for less
than 30 min resulted in significantly weaker AuNP aggregation (Figure S4) indicating a lag
phase of 10-20 min before sufficient polymer is generated.
Next we sought to understand the influence of substrate concentration on the assay. As
expected, GOx activity and peroxide generation was found to be dependent on glucose. The
activity of the GOx was sufficient to fully deoxygenate the solution at glucose concentrations
above 1 mM as seen by the aggregation of the AuNPs (Figure S5). At this concentration and
above, the amount of peroxide generated (1 – 2 mM) is similar to that used in standard HRP
based assays. Because the concentration of peroxide is dependent on the GOx activity, it cannot
be reduced and so the assay should always be kinetically dependent on the concentration of HRP.
Addition of more peroxide was found to result in suicide inactivation of the HRP (Figure S6).
Interestingly, we found that under certain conditions GOx was able to initiate polymerization
and cause AuNP aggregation even in the absence of HRP. Two separate mechanisms for this
polymerization by GOx were determined. The first source is from the redox degradation of
peroxide into hydroxyl radicals by iron and copper via the Fenton reaction.22, 32-34 When trace
9
metals were removed by prior treatment with a metal chelating resin (chelex), no aggregation
was observed. However, when 2 – 50 ppb iron was reintroduced into the reaction mixture, strong
aggregation was observed (Figure 4a). Thus, the GOx / monomer system provides a highly
sensitive, naked eye sensor for iron. We observed no aggregation from any of the other transition
metals, except for copper, at concentrations up to 50 ppb (Figure S7). Though this background
signal was present in this assay design, it is not expected that the same background would be a
problem in other GOx based nanoparticle assays as monomer is a key component.
Figure 4. AuNP aggregation (absorbance at 600/530 nm) indicating the initiation of
polymerization due to the Fenton reaction (a) or GOx directly (b) in the absence of HRP. In the
presence of trace amounts of FeCl3, the peroxide is degraded into polymer producing hydroxyl
radicals. Similarly, in the presence of large amounts of acac, GOx is able to directly oxidize acac
and produce radicals. (c) HRP calibration curve showing AuNP aggregation vs [HRP] at two
different acac concentrations.
10
The second source of background was observed to be due to direct oxidation of acac by GOx.
Reactions with increasing concentrations of acac and 200 nM GOx but no HRP resulted in
increasing amounts of aggregation (Figure 4b). As the concentration of peroxide should be the
same in all cases, the dependence of aggregation on acac concentration suggests that radicals are
generated by the action of GOx on the acac directly, and not just by degradation of the peroxide
into hydroxyl radicals. This background reaction can be greatly minimized by reducing the
concentration of acac below 2.4 mM without significantly harming HRP initiated
polymerization. In this way it was possible to generate a calibration curve for HRP that showed
aggregation down to 250 ng/ml (Figure 4c). The sensitivity here is not limited by the ability of
HRP to generate radicals, but rather it is limited by the background generated from the GOx and
acac. To the best of our knowledge, this is the first report suggesting that GOx is also able to
directly initiate FRP.
The ability of GOx to generate radicals at high acac concentration allows the use of this system
to detect catalase. Catalase is a very active enzyme that is able to consume peroxide to produce
water and oxygen, and we hypothesized that the generation of oxygen should be able to inhibit
polymerization (Figure 5a). To test this, we performed a set of polymerizations with 200 nM
GOx and 2.4 mM acac at varying concentrations of catalase. When no catalase was present, the
high acac concentration led to a small amount of polymerization directly by the GOx and
aggregation was observed (Figure 5b-c). As the concentration of catalase was increased the
extent of aggregation was decreased due to inhibition of polymerization. This assay format was
extremely sensitive to the presence of catalase, and inhibition of aggregation was observed down
to 0.7 ng/ml enzyme.
11
Figure 5. a) Scheme showing inhibition of polymerization using a catalase inverse assay. In this
design, catalase is used to outcompete the GOx by oxygenating the solution and preventing
polymer from forming. b) Absorbance spectra and c) Absorbance ratios (600/530 nm) vs catalase
concentration, showing catalase sensitivity down to 1 ng/ml.
In conclusion, a new assay format is described here that uniquely takes advantage of
polymerization based signal amplification. We show that by specifically designing the surface of
AuNPs, nanoparticle aggregation can be triggered in response to very low monomer-to-polymer
conversion (<0.05%) making this system highly sensitive to the presence of radical generating
enzymes. To allow this assay to function under atmospheric conditions, GOx is introduced to
deplete all oxygen from the solution and supply peroxide. In addition, using an inverse assay
format in which the generation of oxygen from catalase inhibited polymerization, we were able
12
to detect 0.7 ng/ml of catalase. Given that polymerization based signal amplification is currently
limited by appropriate readout methods, it is anticipated that this new platform will provide new
tools for biosensing.
ASSOCIATED CONTENT
Supporting Information. Detailed experimental procedures and methods, as well as the support
figures are given in the supporting information. This material is available free of charge via the
Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
Prof. Molly M. Stevens. Email: m.stevens@imperial.ac.uk, Ph: +44 (0)20 7594 6804, Address:
Prince Consort Road, South Kensington, SW7 2AZ London, United Kingdom
Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval
to the final version of the manuscript. ‡These authors contributed equally.
Notes
The authors declare no competing financial interest.
Funding Sources
This work was supported by a Whitaker International Scholarship to AG and EPSRC grant
(EP/K020641/1).
13
ACKNOWLEDGMENT
The authors thank Roberto de la Rica and Michael Thomas for useful discussions during the
development of this work.
REFERENCES
1.
Zhao, W.; Brook, M. A.; Li, Y. ChemBioChem 2008, 9, 2363-2371.
2.
Cao, X.; Ye, Y.; Liu, S. Anal Biochem 2011, 417, 1-16.
3.
Howes, P. D.; Rana, S.; Stevens, M. M. Chem Soc Rev 2014, 43, 3835-3853.
4.
Mirkin, C. A.; Letsinger, R. L.; Mucic, R. C.; Storhoff, J. J. Nature 1996, 382, 607-609.
5.
Stevens, M. M.; Flynn, N. T.; Wang, C.; Tirrell, D. A.; Langer, R. Adv Mater 2004, 16,
915-918.
6.
Aslan, K.; Luhrs, C. C.; Pérez-Luna, V. H. J Phys Chem B 2004, 108, 15631-15639.
7.
Liu, J.; Lu, Y. Nat Protoc 2006, 1, 246-252.
8.
Niidome, T.; Yamagata, M.; Okamoto, Y.; Akiyama, Y.; Takahashi, H.; Kawano, T.;
Katayama, Y.; Niidome, Y. J Control Release 2006, 114, 343-347.
9.
Boal, A. K.; Ilhan, F.; DeRouchey, J. E.; Thurn-Albrecht, T.; Russell, T. P.; Rotello, V.
M. Nature 2000, 404, 746-748.
10.
Shenhar, R.; Norsten, T. B.; Rotello, V. M. Adv Mater 2005, 17, 657-669.
11.
Willner, I.; Basnar, B.; Willner, B. Febs J 2007, 274, 302-309.
12.
Ghadiali, J. E.; Stevens, M. M. Adv Mater 2008, 20, 4359-4363.
13.
la Rica, R.; Aili, D.; Stevens, M. M. Adv Drug Deliver Rev 2012.
14.
de la Rica, R.; Stevens, M. M. Nat Nanotechnol 2012, 7, 821-824.
14
15.
Rodríguez-Lorenzo, L.; de La Rica, R.; Álvarez-Puebla, R. A.; Liz-Marzán, L. M.;
Stevens, M. M. Nat Mater 2012, 11, 604-607.
16.
Wen, F.; Dong, Y.; Feng, L.; Wang, S.; Zhang, S.; Zhang, X. Anal Chem 2011, 83, 1193-
1196.
17.
Wu, Z.-S.; Zhang, S.-B.; Guo, M.-M.; Chen, C.-R.; Shen, G.-L.; Yu, R.-Q. Anal Chim
Acta 2007, 584, 122-128.
18.
Xuan, J.; Jia, X.-d.; Jiang, L.-P.; Abdel-Halim, E.; Zhu, J.-J. Bioelectrochemistry 2012,
84, 32-37.
19.
Ghadiali, J. E.; Lowe, S. B.; Stevens, M. M. Angew Chem 2011, 123, 3479-3482.
20.
Laromaine, A.; Koh, L.; Murugesan, M.; Ulijn, R. V.; Stevens, M. M. J Am Chem Soc
2007, 129, 4156-4157.
21.
Li, Y.; Wu, Y.-F.; Yuan, L.; Liu, S.-Q. Chinese J Anal Chem 2012, 40, 1797-1802.
22.
Berron, B. J.; Johnson, L. M.; Ba, X.; McCall, J. D.; Alvey, N. J.; Anseth, K. S.;
Bowman, C. N. Biotechnol Bioeng 2011, 108, 1521-1528.
23.
Durand, A.; Lalot, T.; Brigodiot, M.; Marechal, E. Polymer 2000, 41, 8183-8192.
24.
Durand, A.; Lalot, T.; Brigodiot, M.; Maréchal, E. Polymer 2001, 42, 5515-5521.
25.
Emery, O.; Lalot, T.; Brigodiot, M.; Maréchal, E. J Polym Sci Pol Chem 1997, 35, 3331-
3333.
26.
Hollmann, F.; Arends, I. W. Polymers 2012, 4, 759-793.
27.
Kalra, B.; Gross, R. A. Biomacromolecules 2000, 1, 501-505.
28.
Johnson, L. M.; Fairbanks, B. D.; Anseth, K. S.; Bowman, C. N. Biomacromolecules
2009, 10, 3114-3121.
15
29.
Sigg, S. J.; Seidi, F.; Renggli, K.; Silva, T. B.; Kali, G.; Bruns, N. Macromol Rapid
Comm 2011, 32, 1710-1715.
30.
Derango, R. A.; Chiang, L.-c.; Dowbenko, R.; Lasch, J. G. Biotechnol Tech 1992, 6, 523-
526.
31.
Teixeira, D.; Lalot, T.; Brigodiot, M.; Maréchal, E. Macromolecules 1999, 32, 70-72.
32.
Zepp, R. G.; Faust, B. C.; Hoigne, J. Environ Sci Technol 1992, 26, 313-319.
33.
Gozzo, F. J Mol Catal A-Chem 2001, 171, 1-22.
34.
Sarac, A. Prog Polym Sci 1999, 24, 1149-1204.
16