Article pubs.acs.org/ac
Sensitive Affimer and Antibody Based Impedimetric Label-Free Assays for C‑Reactive Protein Anthony Johnson,† Qifeng Song,‡ Paul Ko Ferrigno,‡,¶ Paulo R. Bueno,*,§ and Jason J. Davis*,† †
Department of Chemistry, University of Oxford, South Parks Road, Oxford, OX1 3TA, United Kingdom Wellcome Trust/EPSRC/Leeds Medical Engineering Centre, Leeds Institute of Molecular Medicine, Wellcome Trust Brenner Building, St James’s University Hospital, Leeds LS9 7TF, United Kingdom § Instituto de Química (Institute of Chemistry, Physical Chemistry Department), Universidade Estadual Paulista (São Paulo State University, UNESP), CP 355, 14800-900, Araraquara, São Paulo, Brazil ‡
S Supporting Information *
ABSTRACT: C-reactive protein (CRP) is an acute phase protein whose levels are increased in many disorders. Levels greater than 3 μg/mL serum have hitherto been considered to indicate pathology, but there is increasing interest in assessments between 0.1 and 10 μg/mL, which have been found to correlate with severity of risk for cardiovascular disease. We report herein the generation of both antibody and Affimer based impedance immunoassays for CRP that are substantially more sensitive than clinically utilized immunonephelometry and immunoturbidity assessments. Significant in this study is not only the use of a constrained peptide to detect a clinically important target but also that derived electrochemical impedance assays can be highly sensitive even with probes whose relatively weak (μM) affinities are not amenable to target detection by surface plasmon resonance (SPR). Key to this finding is acknowledging that receptive surfaces of comparatively low initial steric bulk and charge transfer resistance are especially primed to be highly responsive to target binding in electroanalytical assays of this type.
I
There are currently several tests for C-reactive protein used in clinical practice, the highest sensitivity of which uses laser nephelometry. This is a well-established technique in immunology and, with immunoturbidetry assessments, is capable of target protein detection down to the low nM level, though lower limits are more usually in the tens of nM regime.1 These assays, though fairly robust, are both relatively imprecise and relatively insensitive.4 Since the median level of CRP in the blood of healthy individuals is 50% change at the Affimer interface (see below and Figure 5). Binding of the sterically bulky, pentameric CRP is associated, in all cases, with an increase in Rct dependent on concentration of target enabling quantification. Interestingly, the relationship between Rct and the concentration of CRP for Affimer/antibody assays is logarithmic rather than linear (Figures 4 and 5). Such a relationship has been noted in prior impedimetric assays15,36 and can be related to the “steric consequences” of CRP binding at the receptive
CRP specific antibody showed quantifiable binding at the concentrations assayed, with the Affimer interfaces consistently unresponsive over the target concentrations utilized (despite showing strong interactions with its scaffold-specific anticystatin antibody, data not shown). For the CRP antibody interface, the interaction KD was resolved at 75 ± 10 nM (a value in good agreement with EIS derived values) with an Rmax of 19.6 m°. The negligible response above noise of the P6P4 or P7i22 interfaces even at hundreds of nM target is consistent with KD’s in the μM regime at these interfaces. Though target binding was disappointingly weak at these Affimer interfaces, we were interested in whether the markedly smaller receptor size would be advantageous in an electrochemical format (where changes in interfacial bulk or capacitance are detected). EIS assays were, accordingly, implemented with both the antibody and the P7i22 Affimer, shown qualitatively to exhibit the greatest target binding affinity in microarrays. Redox probe cyclic voltammograms (CVs) and EIS Nyquist analyses (Figure 3) were used to initially monitor receptive layer formation (Figures S2 and S3, Supporting Information) and were characterized, on initial PEGylation, by changes typical for the successful formation of a thiolated adlayer (a dramatic change in both faradaic activity and interfacial capacitance). Modification of the gold electrode with the PEGylated thiol leads to the introduction of an additional capacitive relaxation process related to the dielectric characteristics of this layer in addition to monolayer resistor and capacitance terms (represented by a series resistive and capacitive terms denoted as Rt and Ct, respectively, and in parallel to capacitance of the monolayer Cm) (Figure 2). The
Figure 2. (a) Schematic representation of the interfacial impedance of the CRP receptive interfaces used herein. (b) The equivalent circuit used to model impedance data. Capacitance is dominated by Cm with Rt and Ct terms quantifying the thiol (SAM) dipolar relaxation characteristics. The Warburg element (Zw) accounts for the bulk diffusion characteristics of the redox probe, and Rct is the redox charge transfer resistance.The electrolyte resistance (Rs) is modeled in series with the above total interfacial impedance (see (a)) and is subtracted from analyses (since this is a series resistance, it does not affect the data interpretation but could influence the visualization of impedance spectra). Note that interfacial capacitance of a monolayer dielectric modified electrode is defined by two series capacitances, those of the monolayer (Cm) and of the double layer (Cdl) where Cdl ≫ Cm, meaning Cm dominates in analyses and is, therefore, the only capacitance represented in the equivalent circuit showed here.
double layer capacitance of the electrode, Cdl, lies in series with Cm and is characteristically pushed to the outer surface of the thiol film. Since Cdl ≫ Cm, the latter dominates in capacitative analyses35 and in the observed restricted faradaic activity at the modified surface. Subsequent modification of the layer terminus with receptive Affimer or antibody leads to a highly reproducible, but 6556
dx.doi.org/10.1021/ac300835b | Anal. Chem. 2012, 84, 6553−6560
Analytical Chemistry
Article
Figure 3. Overlaid (a) Nyquist impedimetric diagrams and (b) Nyquist capacitive diagrams of the P7i22 Affimer-CRP interface at different concentrations of CRP. Measurements performed in 1 mM K3Fe(CN)6 in 10 mM PBS. The insets show the behavior at higher frequencies for (a) impedance and (b) capacitance and highlight the influence of pegylated thiol layer on the capacitance at higher frequency (the semicircle visible at around 300 Hz representing the effects of Ct and Rt). The impedance sampled at low frequency responds greatly to CRP, markedly less so at higher frequency. Capacitance responds little to CRP at any frequency, showing that after the formation of the receptive layer the interfacial response to target is dominated by Rct. As noted, Ct and Rt terms are considered in the equivalent circuit model in enabling a more precise calculation of the Rct term. Observations at the antibody derived surfaces are largely equivalent and depicted in (c) and (d). All the impedance and capacitance complex values shown here are mean values acquired from measurements across three different electrodes.
Figure 4. (Left) Rct plotted against concentration of CRP on a logarithmic scale for the P7i22 Affimer interface (values obtained from a fitting of mean values gained from three independent impedance data sets; the error bars represent errors associated with fits of impedance to the model of Figure 3). (Right) The variance of the real part of impedance with log CRP concentration; data obtained at a sampling frequency of 0.1 Hz, where Rct and Z’ are most correlated. The dependence of both parameters is equivalent to within 10%. The linear dependence of the Pearson coefficient is >96% in both cases and 98% for the real part of the impedance at this frequency.
6557
dx.doi.org/10.1021/ac300835b | Anal. Chem. 2012, 84, 6553−6560
Analytical Chemistry
Article
Figure 5. (Left) Rct plotted against concentration of CRP on a logarithmic scale for the antibody interface (values obtained from a fitting of mean values gained from three independent impedance data sets; the error bars represent errors associated with fits of impedance to the model of Figure 2). (Right) The variance of the real part of impedance with log CRP concentration; data obtained at a sampling frequency of 0.1 Hz, where Rct and Z’ are most correlated). The dependence of both parameters is equivalent to within 10%. The linear dependence of the Pearson coefficient is >96% in both cases and 98% for the real part of the impedance at this frequency.
Figure 6. Schematic representation of the antibody (lower) and P7i22 Affimer (upper) interfaces on a PEGylated gold electrode (to relative scale). The steric bulk of the antibody generates a receptive layer of high initial charge transfer resistance (lower figure) that is modulated by subsequent CRP binding. The much smaller constrained peptide receptor presents a surface of much lower initial charge transfer resistance (a large “gate” accessible to the redox probe; middle figure) that is more sharply reduced by binding of the sterically bulky CRP target protein.
interface, that is the highly sensitive dependence of redox probe access to the underlying electrode on target binding (and the resulting logarithmic dependence of faradaic activity and Rct.)37 Highly significant in these results is both that the Affimer and
antibody interfaces are highly regenerable (
