Serological point-of-care and label-free capacitive diagnosis of dengue virus infection

Highlights

• We demonstrate that the electrochemical capacitance can be used as tool in the analysis real patient sample.

• Just a simple drop of the patient sample into the electrochemical cell is required for the diagnostics.

• We demonstrate the use of the concepts for diagnostic of dengue febrile patients.

• The minimal need for manipulation of biological samples is attributed to the reagentless nature of capacitive assays.

Abstract

Dengue non-structural protein 1 (NS1 DENV) is considered a biomarker for dengue fever in an early stage. A sensitive and rapid assay for distinguishing positive from negative dengue infection samples is imperative for epidemic control and public health in tropical regions because it enables the development of instantaneous updatable databases and effective surveillance systems. Presently, we successfully report, for the first time, the use of the electrochemical capacitive method for the detection of NS1 DENV biomarker in human serum samples. By using a ferrocene-tagged peptide modified surface containing anti-NS1 as the receptor, it was possible to differentiate positive from negative samples with a p < 0.01 in a reagentless and label-free capacitive format. This capacitive assay had a cut-off of 1.36% (confidence interval of 99.99%); it therefore opens new avenues for developing miniature label-free electrochemical devices for infectious diseases.

Introduction

Presently, the diagnosis of Zika (ZIKV) and Dengue (DENV) virus infections include nucleic acid amplification tests (NAAT) (Balm et al., 2012) to detect viral RNA, as well as additional enzyme-linked immunosorbent assay (ELISA), tests to detect virus antibodies (IgM) or non-structural protein 1 (NS1) in serum. These tests are very expensive, unspecific and ineffective for field-based purposes, where low-cost point-of-care methods are needed (Darwish et al., 2018).

IgM for ZIKV or DENV is typically detectable around 3–5 days after infection. There is also a possibility for cross-reactivity with closely related dengue, yellow fever (YFV), Japanese encephalitis (JEV), and West Nile viruses. These cross-reactive results are more common in patients that show signs of previous flavivirus infections than patients with a primary virus infection. In countries where there is an active circulation of YFV and JEV or high immunization coverage for such viruses, cross-reactivity can impose an extra challenge for accurate diagnosis. On the other hand, despite the specificity of NAAT, which should be conducted within 7 days of the onset of the illness, false positive results may be reported requiring additional serological tests. In addition to expenses, trained staff is required to perform these assays. The use of such diagnostic algorithms slows down disease reporting, impairs control measures and promotes epidemiological scenarios where incidences are high. Central laboratory tests take longer because of inefficient logistics for epidemiologists, linked to the disadvantages already mentioned. The best diagnostic practice would be to analyze serum samples as early as possible with a second test 2–3 weeks later (Nascimento et al., 2018).

Several ideas for new diagnostic devices for flavivirus infection have recently emerged in the literature (Anusha et al., 2019; Cabral-Miranda et al., 2018; Cecchetto et al., 2017; Faria and Zucolotto, 2019). Most of these ideas focus on one specific virus identification, opposed to the complete serological differentiation between ZIKV and DENV (Song et al., 2018). DNA-based fluorescence detection is the preferred approach (Vina-Rodriguez et al., 2017), but it requires the sample to be heated for DNA amplification. Isothermal amplification protocols (Priye et al., 2017) have also been followed to reduce the heating requirements of the DNA-based fluorescence devices. However, these protocols still require complex equipment for the optimal quantification of the results. The Veredus Labs diagnostic test can differentiate between 26 globally significant tropical pathogens (Tan et al., 2014) with increased sensitivity and specificity; nonetheless, owing to the technology that the device employs, each test costs a few hundred dollars. Additionally, it requires a bulky fluorescence detection instrumentation that costs a few thousand dollars.

Researchers have recently focused on immunosensor development, which targets viral proteins to avoid excessive power-consumption for the heating stage in DNA amplification and other complex DNA-based analysis systems. Electrochemical biosensors seem to be the preferred approach in this case (Kaushik et al., 2018), owing to their minimal instrumentation and lower power requirements. Yet, there are currently no commercially feasible label-free diagnostic devices incorporating flavivirus electrochemical immunosensors even though such devices could be helpful for dengue real-time screening and epidemiological control.

Our research group has demonstrated the possibility of developing assays for the diagnosis of DENV by sensing NS1 in serum samples (or the combined quantification of NS1 and IgG) by utilizing electrochemical methods based on impedance-derived capacitance formats (Cecchetto et al., 2015, Cecchetto et al., 2017; Santos et al., 2018). The capacitive format has the advantage of being a reagentless assay (Cecchetto et al., 2017), i.e. it does not require adding electrochemical (redox-pair) probes to the biological samples. The redox probe (see Fig. 1, Fig. 2), contained in the biosensing receptive surface, promotes a pseudo-capacitive transducer signal (Bueno et al., 2017; Cecchetto et al., 2017; Fernandes et al., 2013) that varies depending on the selective binding of biomarkers at the interface.

Functioning as pseudo-capacitive assays (Fernandes et al., 2013; Lehr et al., 2014) within redox entities that can be designed to generate a specific faradaic mode of biosensing detection (using redox compounds such as Prussian Blue (Oliveira et al., 2019) or electroactive monolayers (Fernandes et al., 2013; Piccoli et al., 2018)), electrochemical capacitive interfaces are quite functional in the development of label-free detection for a range of proteins, including NS1 (Cecchetto et al., 2015, Cecchetto et al., 2017; Santos et al., 2018), known to be applicable for the diagnosis of DENV.

Our research group designed electrochemical capacitive interfaces by applying mesoscopic physical concepts (Bueno, 2018a; Bueno et al., 2017; Fernandes et al., 2013; Hudari et al., 2018; Lehr et al., 2014; Oliveira et al., 2019; Piccoli et al., 2018; Santos et al., 2014, Santos et al., 2015). Mesoscopic physics can be applied to systems in which the mechanical properties lie in between classical and quantum mechanics (Bueno, 2018a, b; Bueno et al., 2016; Bueno et al., 2015; Miranda and Bueno, 2016). This is characteristic of electrochemical interfaces where electric charges accumulate of length typically between 0.5 to 5 nm (Bueno, 2018a, b). Fig. 1 illustrates a type of mesoscopic interface constituted of an ensemble of mesoscopic capacitive moieties (redox centers) in which each moiety is electronically coupled to an electrode, and each of them contains molecular receptors. Upon the binding of targets to receptors, variation in the potential $V_G$ associated with the redox activity of the redox moieties can be attained by measuring the variation of capacitance $C_{\bar{\mu }}$ as a function of target concentration (see Fig. 2). The electrochemical capacitance $C_{\bar{\mu }}$ is a type of mesoscopic capacitance where both the classic and quantum characteristics of the capacitance observed at the interface determines its properties (Bueno, 2018a, b). The mesoscopic capacitance at the interface can be theoretically understood and analyzed by solving a density functional Hamiltonian of the interface, and the resulting output is the energy of the interface (Bueno, 2018b; Bueno et al., 2015), which is proportional to two series capacitive components having $C_{\bar{\mu }}$ as the equivalent capacitance expressed as (Bueno, 2018a, b; Bueno et al., 2016; Bueno and Davis, 2014b; Bueno et al., 2017):

$$E\propto \frac{1}{C_{\bar{\mu }}}=\left(\frac{1}{C_i}+\frac{1}{C_q}\right),$$

where, $C_i$ (associated with non-Faradaic processes) is the ionic capacitance at the interface (typically represented by double layer processes) and $C_q$ is the quantum capacitive term associated with the occupation of redox states (Bueno, 2018a, b; Bueno and Davis, 2014b; Bueno et al., 2017). Changes in the electronic states and ionic structures at the interface due to chemical changes caused by environmental variations can be attained by measuring $C_{\bar{\mu }}$. The mesoscopic states accessible by measuring $C_{\bar{\mu }}$ of the interface are, therefore, quite sensitive to minimal environmental changes. In summary, the energy variation per number of electric charge $\mathcal{N}$ or variations in the electron density which is measured as changes in $C_{\bar{\mu }}$ at the interface can be experimentally obtained through the quantification of $C_{\bar{\mu }}$ with the equation: $dE/d\mathcal{N}=\mathcal{N}(e^2/C_{\bar{\mu }})$, where $e$ accounts for the elementary charge (Bueno, 2018a, b; Bueno et al., 2017; Piccoli et al., 2018). Note that $\mathcal{N}$ can be represented as a function of the electron density at the interface, expressed as $\mathcal{N}={\int }_{\mho }\text{Δ}\rho \left(\overrightarrow{r}\right)d\overrightarrow{r}$) (Bueno, 2018a, b), which varies in the biosensing interface as a function of the recognition of the biological target (Bueno et al., 2013, Bueno et al., 2012; Fernandes et al., 2013; Miranda and Bueno, 2016; Santos et al., 2015). In other words, the mechanism of signal transduction is based on the change of the chemical energy of the interface per number of varied particles, that is $dE/d\mathcal{N}$, which is directly proportional to $1/C_{\bar{\mu }}$ (see Fig. 1S in SI document for more details). Accordingly, with the binding of the biological target, in the present case the NS1 DENV, to the anti-NS1 attached to the ferrocene-tagged peptide monolayer there is a variation in the number of the chemical particles in the surface of the sensors. This variation of the number of chemical particles changes the electrochemical occupancy of ferrocene-tagged peptide centers and thus decreases $C_{\bar{\mu }}$ and increases $1/C_{\bar{\mu }}$. The plot of $1/C_{\bar{\mu }}$ against the logarithmic of target concentration (shown in the SI document, Fig. 1S) is a linear plot that can be used as an analytical curve. Analytical curves are useful in quantifying the concentration of the target when this is required.

In a more theoretical approach, the signal transduction associated with $C_{\bar{\mu }}$ can be understood and analytically deduced using quantum mechanics, mainly using a DFT Hamiltonian analysis (Bueno, 2018b; Bueno et al., 2015). The ground state energies of quantum mechanics are defined at the zero-temperature approximation, whereas the electrochemical redox processes practically occur at room temperature. The finite-temperature situation requires thermal broadening in energetic terms associated with $C_{\bar{\mu }}$. The thermal energy is computed by considering $C_{\bar{\mu }}\propto [d⟨\mathcal{N}⟩/d\bar{\mu }]$, where $d\bar{\mu }=ed{V}_G$ (where, $V_G=e/C_{\bar{\mu }}$), $⟨\mathcal{N}⟩$ denotes an average state of $\mathcal{N}$ based on fermionic statistics which can be expressed as $⟨\mathcal{N}⟩={\left(1+\exp \left[-e{V}_G/k_{B}T\right]\right)}^{-1}$, where variations in $V_G$ are related to the potential of electrons in the electrode $V$ and the potential of the redox species $V_r$ contained in the molecules coupled to the electrode is expressed as $d{V}_G=(V-V_r)=-d\bar{\mu }/e$ (see Fig. 1c). The $[d⟨\mathcal{N}⟩/d\bar{\mu }]$ term, at a constant volume and finite-temperature, is resolved as $1/k_{B}T\mathcal{N}(1-⟨\mathcal{N}⟩)$ by appealing to the thermodynamics grand canonical ensemble. Therefore, $C_{\bar{\mu }}$ can be rewritten as

$$C_{\bar{\mu }}=\left[\frac{e^2\Gamma }{k_{B}T}\right]\mathcal{N}(1-⟨\mathcal{N}⟩).$$

When the potential of the electrode is poised at the half-wave or in its formal potential we note that $⟨\mathcal{N}⟩(1-⟨\mathcal{N}⟩)=1/4$ and Eqn. (2) turns into $C_{\bar{\mu }}=e^2\text{Γ}/4k_{B}T$, which is a constant. As demonstrated (Bueno, 2018a, b) this constant is identical to that defined by Laviron (Laviron, 1979a, b). Therefore, the electrochemical capacitance is the constant between the current density $j$ and scan rate $s$ and thus the constant depends on the characteristics of the molecular film, that controls the kinetics of electron transfer in typical voltammetry kinetic measurements of redox adsorbed species. Yet following Eqn. (2) it can be experimentally confirmed that when a redox film equivalent to that shown in Fig. 1a is fully oxidized or fully reduced $C_{\bar{\mu }}$ is minimum and the maximum value of $C_{\bar{\mu }}$ is attained at the half-wave potential (see Fig. 2a). Experimentally, Eqn. (2) is validated by obtaining the low-frequency value (see Fig. 2b and c) of $C_{\bar{\mu }}$ as a function of electrode potential, as shown in Fig. 2a.

To summarize, diagnostic assays based on biosensing important biomarkers using $C_{\bar{\mu }}$ can be very sensitive and specific because it allows sensing marked molecules due to variations in chemical energy at the interface depending on the binding of selective target molecules (Bueno, 2018a, b). Diagnostic assays do not need redox species to be added in the patient samples, as is the case with traditional potentiometric and other electrochemical type assays. The chemical energy at the interface is both electrostatic and dynamic, the latter associated with variations in the electronic structure and occupancy of redox states at the interface (Bueno et al., 2017; Fernandes et al., 2013; Piccoli et al., 2018; Santos et al., 2015) and the former with the spatial separation of the charges in the molecular film. The changes in the energy state at the interface are perceived by monitoring $C_{\bar{\mu }}$ as a function of biological reactivity of the target to specifically designed interfaces (Bueno et al., 2017; Fernandes et al., 2013; Piccoli et al., 2018; Santos et al., 2015). In other words, the electrochemical activity (both non-Faradaic $C_i$ and Faradaic $C_q$) at the interface can be accurately measured and eventually spectroscopically separated from undesirable non-specific molecular bindings associated exclusively to the electrostatics of the mesoscopic interface. Particularly, in the case of redox-tagged interfaces, the variations in $C_q$ component of $C_{\bar{\mu }}$ at the interface is preponderant (Bueno et al., 2017; Fernandes et al., 2013; Piccoli et al., 2018; Santos et al., 2015). For instance, see Fig. 3c, where it can be noticed that the non-Faradaic response of the interface (used in the present work) is only about 5 μF cm⁻² whereas the Faradaic response is more than 45 times higher, at about 240 μF cm⁻². The use of a $C_{\bar{\mu }}$ response, at the interface containing $C_q$ Faradaic terms, amplifies the signal response at the interface, implying that the interface is more sensitive to variations in the Faradaic components since it is considerably higher than non-Faradaic components.

Here we report, for the first time, that the above principles apply to serological samples in a direct, rapid and reagentless point-of-care format, observing that a quantification of NS1, though possible, is not necessary for assays aiming at only to distinguish between positive and negative patient's samples. The proof-of-concept that capacitive interfaces comprising ferrocene-tagged peptide containing anti-NS1 (used as the biological receptor) can be explicitly designed to be used in DENV assays is here demonstrated by detecting NS1 protein in serum samples of febrile patients. The importance of detecting NS1 DENV rely in outbreak investigations, where rapidity and specificity of the test is crucial (World Health, 2009). In summary, a capacitive label-free assay was developed for the specific diagnostic of DENV.