A simple, economical, sensitive, and specific diagnostic test is required to help to control flavivirus infections. The serological approaches currently available are limited due to the cross-reactivity between the members of the Flavivirus genus. Molecular diagnosis, although effective, requires expensive laboratory facilities frequently unavailable in endemic regions. Therefore, it is necessary to develop new technologies able to distinguish with precision the medically important viruses of this genus. The aim of the present study was to obtain DNA aptamers for the ZIKV NS1 protein, capable of providing a specific virus identification for application in diagnostic biosensors. In this context, biosensor platforms based on aptamers (aptasensors) are commonly used to assess and quantify in real-time, with high sensitivity, the presence of an analyte, such as a protein [26]. Aptasensors for hepatitis C, H5N1 avian influenza, and H1N1 viruses, among others, have been developed [27,28]. Furthermore, aptamers have the potential to overcome the lacking functional and storage stability of most biosensors exploiting antibodies.

In the present work, CE-SELEX was used for aptamer selection. This procedure uses CE in the SELEX separation step. A representative CE electropherogram is shown in Figure 1. The elution peaks corresponding to the NS1–aptamer complexes and the oligonucleotides that did not bind to the target occurred at different positions. Five runs were performed to accumulate enough DNA for the PCR amplification step and these runs are superimposed in Figure 1. Counterselections with NS1 proteins of DENV (serotypes 1, 2, 3, and 4) and YFV were performed after each positive selection with ZIKV NS1 protein, as described in the CE-SELEX selection schematic diagram of Figure S2 (Supporting Information File 1) as well as in the Experimental section.

Figure 1: Representative CE electropherogram of CE-SELEX separation step. Peak A corresponds to the NS1–aptamers complexes, while peak B refers to oligonucleotides that did not bind to the target and were discarded. Five runs, which are displayed by distinctly colored lines, were performed to accumulate enough DNA for PCR amplification. These runs are superimposed in the graph.

After the third selection cycle, the obtained ssDNA pools were cloned and sequenced. Seventy-five sequences were identified. The most prevalent sequences were chosen for further characterization. Among these, the sequence ZIKV60 presented the best results in terms of affinity and specificity. Figure S3 (Supporting Information File 1) illustrates the secondary structure of ZIKV60 aptamer predicted by Mfold software.

The aptamer affinity to the target is indicated by the dissociation constant at equilibrium (Kd). For the determination of Kd of the aptamers, a saturation-binding curve to the target was measured with an increasing series of ZIKV60 concentrations. The aptamer quantification was performed by ELONA [23]. ZIKV60 presented a high binding affinity to ZIKV NS1 protein with a Kd value of 2.28 ± 0.28 nM. Figure 2 illustrates the ZIKV60 aptamer saturation binding curve to ZIKV NS1.

Figure 2: Dissociation constant (Kd) of the ZIKV60 aptamer measured by ELONA. The saturation curve was obtained by plotting the aptamer concentration as a function of the total mass of ssDNA bound to ZIKV NS1. The value of Kd was calculated by non-linear regression analysis (one-site specific binding with Hill slope, software GraphPad Prism 5), resulting in Kd = 2.28 ± 0.28 nM (R2 = 0.9958).

The aptamer specificity was evaluated by qPCR [29-33]. The ZIKV60 binding to the NS1 protein of DENV (serotypes 1, 2, 3, and 4) and YFV was also determined. The results are shown in Figure 3. The specificity assay confirmed that ZIKV60 binds preferentially to the NS1 ZIKV protein relative to NS1 of the other viruses, with the following ratios: DENV1 14.2, DENV2 14.8, DENV3 13.4, DENV4 14.8, and YFV 14.2 (Table 1).

Figure 3: Binding of ZIKV60 aptamers to NS1 protein of different flaviviruses performed by qPCR. The graphs show the ssDNA amount (ng) recovered after incubation of 100 nM of ZIKV60 with NS1 protein (40 pmol) of the different flaviviruses: ZIKV, DENV (serotypes 1, 2, 3, and 4), and YFV. Statistical analyses were performed using the GraphPad Prism 5 software. *p < 0.05 and **p < 0.01.

Table 1: ZIKV60 aptamer characteristics.

Aptamer ΔG (kcal/mol) Kd (nM) Target/non-target ratioa ZIKV60 −21.75 2.28 ± 0.28 DENV1: 14.2       DENV2: 14.8       DENV3: 13.4       DENV4: 14.8       YFV: 14.2

aRatio between the binding of ZIKV60 to the NS1 protein of the Zika virus in comparison to the NS1 protein of the other tested flaviviruses

Applying CE to SELEX (CE-SELEX) was an important advance. The separation efficiency of CE contributed to the reduction of the number of cycles necessary for the selection of high-affinity aptamers. As the molecules remain in solution during the separation process, there is no steric hindrance for aptamers binding to the target [26]. In addition, the native conformation of the protein targets is retained and is the same as found in biological samples. To obtain aptamers specific for the ZIKV NS1 protein, counterselections with NS1 proteins of DENV (serotypes 1, 2, 3, and 4) and YFV were included. Carrying out counterselections with homologous proteins is advantageous to avoid cross-reactivity. By this approach, it was possible to obtain the ZIKV60 aptamer that, in addition to the high affinity (Kd = 2.28 ± 0.28 nM), showed a high specificity for ZIKV NS1. The ZIKV60 aptamer binding to ZIKV NS1 protein was around 14 times higher than the binding to NS1 proteins of the other flavivirus.

According to the selectivity assay results, the specificity of ZIKV60 to ZIKV NS1 protein is considerably higher than that to NS1 of the other viruses. Concerning the development of aptasensors for ZIKV NS1 protein detection, the results show that the biosensors based on ZIKV60 aptamer would yield improved protein recognition. Following this concept, we functionalized multiple graphene field-effect transistor devices with ZIKV60 aptamers to demonstrate the feasibility of constructing graphene-based aptasensors for ZIKV NS1 protein detection. Importantly, these ZIKV60 aptamers feature a pyrene moiety for their direct immobilization on graphene via π–π stacking [34].

The qualitative confirmation of aptamer immobilization on graphene is illustrated in Figure 4a. For a representative device, Figure 4a exhibits curves of graphene resistance (RSD) as a function of the gate voltage (VG) before (black line) and after (red line) functionalization with ZIKV60 aptamers. In this experiment, we used 100 mM PBS as electrolyte for gating. The π-conjugated units of the pyrene-modified ZIKV60 aptamers transfer electrons to graphene, resulting in the left-shift of the graphene transfer curve because of the aptamer adsorption. Several studies support this electron transfer from pyrene-modified molecules to graphene as the binding mechanism in π–π interactions between such compounds [35-37]. The association of pyrene-modified ZIKV60 aptamers with graphene may also be mediated by charge transfer that assists the interaction between the pyrene moiety of ZIKV60 and the π orbitals of graphene [38,39]. Consequently, the electron transfer to graphene after functionalization reveals a factual immobilization of ZIKV60 aptamers on its surface. Similar results were obtained for four additional graphene devices. See Figure S4 (Supporting Information File 1) for more details.

Figure 4: Functionalization of graphene with ZIKV60 and detection of ZIKV NS1 protein. (a) Representative curves of the normalized graphene resistance (RSD) as a function of the gate voltage (VG) before (black line) and after (red line) functionalization with ZIKV60 aptamers. These measurements were carried out using 100 mM PBS as electrolyte for gating. (b) Normalized graphene transfer curves of a single device functionalized with ZIKV60 aptamers before (red line) and after sequential addition of ZIKV NS1 protein with the following concentrations: 0.01 (orange line), 0.1 (yellow line), 1 (green line), 10 (cyan line), 100 (blue line), and 1000 (purple line) pg/mL. The inset highlights the portion of each graphene transfer curve extending from 0.00 to 0.20 V. In each measurement, ZIKV NS1 protein was diluted in human serum, which was used as electrolyte for gating.

In the following, we address the capability of ZIKV60-functionalized graphene devices to detect distinctively the ZIKV NS1 protein. We conducted electrical measurements of graphene transfer curves to evaluate the binding of ZIKV NS1 diluted in human serum to ZIKV60 ligands. The response of a representative ZIKV60-functionalized graphene device to the sequential addition of increasing amounts of NS1, from 0.01 up to 1000 pg/mL, are shown in Figure 4b. For each specific concentration, we first exposed the graphene device to the electrolyte containing the NS1 protein for 10 min. Then, we measured the corresponding electrical response. The characterization of ZIKV60 alone (red line in Figure 4b) was carried out after 10 min of incubation with human serum not containing NS1 to provide a basis for comparison to upcoming receptor–target conjugations occurring in protein-rich environments. From 0.01 to 100 pg/mL, the graphene transfer curve left-shifts successively as a result of progressive additions of five specific protein dilutions. This denotes a cumulative electron transfer to graphene as NS1 binds to ZIKV60. However, this trend is interrupted at the cutoff value of 100 pg/mL. In fact, the inset in Figure 4b evidences only minimal alterations of the graphene transfer curve corresponding to 100 pg/mL (blue line) after adding 1000 pg/mL of NS1 (purple line). Physically, this outcome is sustained by a complete saturation of the active ZIKV60 binding sites with NS1 due to exposure to protein dilutions of higher concentrations, such as 100 pg/mL. Consequently, further protein incorporation results in inappreciable specific recognition as well as negligible charge transfer to graphene.

This observation has been made consistently with six additional graphene devices, whose detection range extends from 0.1 to 100 pg/mL, see Figure S5 (Supporting Information File 1) for more details. We therefore analyzed in depth the general dose-dependent responses of all seven aptasensors. Figure 5a exhibits a calibration curve for ZIKV NS1 detection, reflecting the average behavior of this group of sensing devices with respect to protein addition. We evaluated the sensor responses as percentage changes in graphene resistance (ΔRSD) at VG = 0.2 V caused by aptamer–protein binding. For more details on how the variations in graphene resistances were measured, see Table S1 and Table S2 (Supporting Information File 1). For each specific ZIKV NS1 dilution, the blue filled circle in Figure 5 represents the mean percentage increment in graphene resistance for the seven different devices, while the error bars give the standard error of the measurements. The pattern that emerges from Figure 5 clearly reveals that the collective response of the graphene sensors as a function of the ZIKV NS1 concentration follows a sigmoidal trend. Given that, we interpolated a Hill–Langmuir curve, ΔR%(c) ∝ cn (kn + cn)−1, to the experimental data. This fit curve is represented by the red line Figure 5 [40]. The percentage changes in graphene resistance, ΔR%, as a function of the ZIKV NS1 concentration, c, yielded a microscopic dissociation constant k = 1.4 ± 0.3 pg/mL and a Hill coefficient of n = 0.8 ± 0.2. The microscopic dissociation constant provides an estimate for the extent of target molecules producing half-occupation on ligand binding sites. Correspondingly, the value obtained for k corroborates the hypothesis that our aptasensors are fully capable of selective and sensitive recognition of ZIKV NS1 protein at picogram per milliliter concentrations.

Figure 5: Calibration curve for ZIKV NS1 protein detection. Percentage change in graphene resistance (ΔRSD) at VG = 0.2 V as a function of the ZIKV NS1 protein concentration. For each specific dilution, the blue filled circle represents the mean percentage increment in graphene resistance accounted for seven different devices; the error bars show the standard error of the measurements. The red line is a Hill–Langmuir curve fitted to the experimental data (R2 = 0.9969).

The levels of NS1 in individuals infected with Zika virus are not completely clear yet. However, Yap et al. recently reported that the normal values for ZIKV NS1 antigen found in clinical samples during acute infection are extremely low or undetectable by ELISA in most cases [41]. In that report, the authors verified that the degree of ZIKV NS1 in 60% of the ill patients was lower than 0.1 ng/mL, which was the limit of detection of their assay. In contrast, the platform reported here surpassed considerably the limitations of ELISA concerning the detection of ZIKV NS1 antigen. Our aptasensors recognized significantly small protein quantities, as low as 0.01 pg/mL, coexisting with large amounts of blood components. For comparison, Afsahi et al. described a biosensor based on graphene utilizing monoclonal antibodies as recognition elements for ZIKV NS1 that could detect 500 ng/mL of the target protein diluted in human serum [42]. Therefore, the potential of ZIKV60 aptamer in association with graphene to detect the NS1 protein of Zika virus on a clinical relevant scale offers the implementation of this hybrid system for early diagnosis of ZIKV infection in real samples in the future.

Aptamers for ZIKV NS1 protein have been reported in the literature in only two other studies so far. Lee and Zeng selected an aptamer (APT2) with high affinity (Kd = 24 pM) but exhibiting cross-reactivity for the NS1 protein of all DENV serotypes. They also obtained a second aptamer (APT10) with negligible cross-reactivity, but less affinity (Kd = 134 nM) [43]. In another study, Morais et al. performed a DNA aptamer selection by immobilizing ZIKV NS1 on ELISA plates [44]. Seven selection cycles were used, including a counterselection step for YFV NS1 in the last cycle. Nonetheless, data concerning aptamer affinity and specificity were not reported in that study. In the present study, an aptamer that combines high affinity and specificity to ZIKV NS1 was successfully obtained. The binding capacity of ZIKV60 to the target protein was exhaustively demonstrated using different methods, namely ELONA (Kd determination assay), qPCR (specificity assay), and biosensing with functionalized graphene devices. To the best of our knowledge, our work is the first to demonstrate aptamer selection for ZIKV NS1 and, simultaneously, its application on a biosensing platform, creating possible routes for the future development of point-of-care diagnostic systems for Zika virus disease.