Competitive ELISA for a serologic test to detect dengue serotype-specific anti-NS1 IgGs using high-affinity UB-DNA aptamers

Serologic tests to detect specific IgGs to antigens related to viral infections are urgently needed for diagnostics and therapeutics. We present a diagnostic method for serotype-specific IgG identification of dengue infection by a competitive enzyme-linked immunosorbent assay (ELISA), using high-affinity unnatural-base-containing DNA (UB-DNA) aptamers that recognize the four categorized serotypes. Using UB-DNA aptamers specific to each serotype of dengue NS1 proteins (DEN-NS1), we developed our aptamer–antibody sandwich ELISA for dengue diagnostics. Furthermore, IgGs highly specific to DEN-NS1 inhibited the serotype-specific NS1 detection, inspiring us to develop the competitive ELISA format for dengue serotype-specific IgG detection. Blood samples from Singaporean patients with primary or secondary dengue infections confirmed the highly specific IgG detection of this format, and the IgG production initially reflected the serotype of the past infection, rather than the recent infection. Using this dengue competitive ELISA format, cross-reactivity tests of 21 plasma samples from Singaporean Zika virus-infected patients revealed two distinct patterns: 8 lacked cross-reactivity, and 13 were positive with unique dengue serotype specificities, indicating previous dengue infection. This antigen-detection ELISA and antibody-detection competitive ELISA combination using the UB-DNA aptamers identifies both past and current viral infections and will facilitate specific medical care and vaccine development for infectious diseases.

. Dengue diagnostics using ELISA formats. (a) Schematic illustration of general detection sensitivity patterns for dengue viral (DENV) NS1 protein, virus (genome), and DENV-reactive IgM and IgG, in primary and secondary DENV infections 8,35 . (b) Direct DEN-NS1 detection in patient blood, by using our UB-DNA aptamers specific to each serotype NS1 protein as capture agents and an anti-DEN-NS1 antibody (Ab#D06) as a detector agent 35,49 . (c) Competitive IgG detection in patient blood, by modifying the DEN-NS1 detection method in (b). If a blood sample contains IgGs binding to DEN-NS1, then the DEN-NS1 detection by UB-DNA aptamers is competitively inhibited. Thus, the addition of each serotype DEN-NS1 and UB-DNA aptamer to the blood sample, followed by mixing with the detector antibody, enables the detection and quantification of IgGs specific to each DEN-NS1 serotype. (d) Typical conventional method for IgG detection, using the antigen (NS1)-coated plates to capture anti-NS1 IgGs in a blood sample. www.nature.com/scientificreports/ Results UB-DNA aptamers targeting each DEN-NS1 serotype and the ELISA format. We previously developed an ELISA format of aptamer-antibody sandwich systems (Fig. 1b), using UB-DNA aptamers targeting each serotype or sub-serotype of DEN-NS1 (Supplementary Table 1 and Fig. 2), for diagnostics in the early stage of dengue infection (Fig. 1a) 49 . These high-affinity UB-DNA aptamers (K D = 30 − 182 pM) were generated from five-letter Ds-DNA libraries by ExSELEX, involving the Ds −Px pair as a third base pair in PCR. In the aptamer generation, we used each DEN-NS1 serotype purchased from The Native Antigen Company. Three aptamers, AptD1, AptD3, and AptD4 targeting DEN1-NS1, DEN3-NS1, and DEN4-NS1, respectively, contained two Ds bases, while the isolated aptamer, AptD2, exhibiting the highest affinity to DEN2-NS1, contained two Ds and one Px bases. These three UBs in AptD2 are essential for the tight binding to DEN2-NS1. This Px base was incorporated into the aptamer by a mutation during PCR amplification of the DNA libraries in ExSELEX.
Since the Px nucleoside is unstable and thus not suitable for chemical DNA synthesis, instead of the Px nucleoside, we used the Pa nucleoside, in which the nitro group is replaced with the aldehyde group ( Fig. 2a) 49,52,53 , for the chemical synthesis of AptD2. Each of these four aptamers contains an extraordinarily stable mini-hairpin sequence [54][55][56] at the 3′-terminus, in which the loop region is useful as a modification site with biotin for aptamer immobilization 48,[57][58][59] .
For the ELISA format with aptamer and antibody sandwich systems, we employed an anti-DEN-NS1 monoclonal antibody (Ab#D06), which binds to all four serotypes of DEN-NS1 with 27-107 pM K D values 49 . In the ELISA format, the mixtures of DEN-NS1 samples and the antibody Ab#D06 were incubated on an aptamerimmobilized plate, and then the signal was detected by the colorimetric output, using a secondary anti-IgG HRP-conjugated antibody. We determined the limits of detection (LOD) in buffer, which were 1.19-2.36 ng/mL for the DEN-NS1 samples purchased from The Native Antigen Company 49 .
Development of serotype-specific anti-DEN-NS1 IgG detection by a competitive ELISA format. We next tested this ELISA format in the presence of human serum purchased from Sigma-Aldrich, for the direct detection of each DEN-NS1 from The Native Antigen Company (Fig. 3). Unexpectedly, we found that the NS1 detection was significantly inhibited in the serum (Fig. 3b), relative to that in buffer (Fig. 3a). One of the plausible causes is the presence of anti-DEN-NS1 IgG antibodies in the serum, which inhibited the binding of the aptamer to the additional NS1 proteins. To validate this IgG contamination theory, we removed the total IgG antibodies from the serum by treating it with protein A-immobilized resin, and confirmed the absence of inhibition with the treated serum (Fig. 3c). We also performed an ELISA using a serum sample from a Singaporean healthy volunteer who was currently not infected with dengue, to determine whether the serum inhibited the detection. Interestingly, the serum showed the serotype-specific inhibitions in the detection of DEN2-NS1, as well as DEN1-NS1 to some extent (Fig. 3d), suggesting that the person might have previously been infected with the dengue serotype 2 and/or serotype 1 viruses. Therefore, we obtained two other serum samples from healthy Japanese volunteers (PD0-1 and PD0-2) who have never been infected by DENV and performed an ELISA. As expected, the two serum samples did not inhibit the DEN-NS1 detection in our ELISA format (Fig. 3e,f). For further studies, we used these serum samples as a control without anti-DEN-NS1 IgGs. These results inspired us to develop a new method using the competitive ELISA format with each DEN-NS1 serotype sandwiched with each aptamer and Ab#D06, for the serotype-specific detection of anti-DEN-NS1 IgG antibodies in human serum samples (Fig. 1c).
For the serotype-specific IgG detection, we developed a simple quantification method for the anti-DEN-NS1 IgG activities ( Supplementary Fig. 1). To this end, we performed competitive-inhibition in ELISA using a series of different volumes (0.05, 0.1, 0.2, 0.5, and 5 μL) of patient serum, in the presence of a certain amount of each DEN-NS1 serotype (The Native Antigen Company). After the absorbance measurement at 450 nm (OD 450 ) in the ELISA format, the OD 450 values were plotted against the volume of serum, and we calculated the serum volume required to give an OD 450 of 1.0. We then defined the relative IgG activity (Activity), by the following formula: Activity = 5/(the serum volume required for an OD 450 of 1.0). To equalize the sensitivities of the DEN-NS1 detection among the four serotypes, we adjusted the amounts of immobilized aptamers and spiked DEN-NS1 for each serotype ( Supplementary Fig. 2).

Serotype-specific detection of anti-DEN-NS1 IgG antibodies in patient samples. Using this
competitive ELISA format and the IgG quantification method, we measured the longitudinal changes in the IgG production at the acute and convalescent phases and identified the serotype specificities of the archived blood samples from eleven Singaporean patients (PD1-1 to PD4-1) with acute DENV infection ( Fig. 4 and Supplementary Figs. [3][4][5][6]. The dengue serotype of the current infection in each sample was also identified by our antigen-detection ELISA and an FTD dengue differentiation RT-qPCR test from Fast Track Diagnostics, and by sequencing the RT-PCR amplicons of the clinical samples obtained within 3-5 days after fever onset 49 . In addition, we performed the DEN-NS1, IgM, and IgG detections in these samples with the commercially available LFA kits from SD BIOLINE for DEN-NS1 detection and from Panbio for IgM/IgG detection. Our competitive ELISA format successfully detected the IgGs even after one year in the recovered patients, as shown in the samples PD2-3, PD3-1, and PD3-3. The samples can be categorized into two groups by the IgG detection: one group included PD1-1, PD1-2, PD1-3, PD2-1, PD2-2, PD3-1, and PD3-2, in which the IgG was not detected within a week after fever onset, and the other group included PD2-3, PD3-3, PD3-4, and PD4-1, in which the IgG was detected within 3-5 days after fever onset (acute phase). These data suggested that the latter patients were previously infected by dengue (Fig. 1a). Thus, the first group most likely represented the primary infection, and the second group was a secondary or higher infection. www.nature.com/scientificreports/ There were some discrepancies in PD1-1, PD3-2, and PD3-3 between our IgG detection and the conventional LFA method for IgM/IgG detection. The visual judgement using the LFA format was often ambiguous, and all of the longitudinal IgG detection data supported the higher accuracy of our method over that of the LFA format. The sensitivity and specificity of the LFA kit reported in the literature are 71.9% and 95%, respectively 8 . Thus, our IgG detection might faithfully identify the primary or secondary infection of patients within 3-5 days after fever onset.   www.nature.com/scientificreports/ Our data indicate that the inhibition of the aptamer binding to each DEN-NS1 serotype resulted from the anti-NS1 IgG, rather than the anti-NS1 IgM. No inhibition of the DEN-NS1 detection in our competitive ELISA occurred within the primary infection samples in the first week after fever onset, and the inhibition became detectable at 17 days or thereafter in longitudinal studies. In the samples from the patients with the primary infection, IgM was detected in PD1-1, PD2-1, PD3-1, and PD3-2 in the early phase (3-5 days) by the Panbio test (Fig. 4). Although the Panbio test detects IgMs targeting DENV envelope proteins, IgMs to DEN-NS1 proteins might also be present in these samples. However, the majority of the IgM in the acute phase of DENV infection reportedly targets viral particles or envelope proteins 60 , and the low abundance of anti-NS1-IgMs might reduce the sensitivity of the NS1-specific IgMs. To confirm the IgG-specific detection of our competitive ELISA format, further examinations using a large number of DENV clinical samples will be required.
Using the four samples (PD1-2, PD1-3, PD2-2, and PD3-1) collected 17 or more days after fever onset in the primary infection, our competitive ELISA format identified the serotype of the current infection from the highest serotype-specific activity of each detected IgG. Our results were identical with those obtained by RT-qPCR and sequencing 49 . In the samples obtained six months later (PD1-2 and PD2-2), our data indicated that the serotype-specificity of IgGs broadened to other serotypes.
In the secondary infection samples (PD2-3, PD3-3, PD3-4, and PD4-1), the initial IgG level was not identical to the serotype-specificity of IgGs in the current infection, which might mainly reflect the serotypes of past infection 26,27,61 . Even after one week, the production of the IgG antibodies that predominantly recognized the serotype resulting from the presumable past infection increased sharply, as compared to the IgGs produced from the current secondary infection. Although the predominance of the past infection varied depending on the patients, the PD2-3 and PD3-3 patient samples revealed the massive production of the IgG antibodies to the past serotype infection, and thus the competitive ELISA format might identify the dengue serotype of the past infection.

Comparison with the typical conventional serologic test.
To evaluate the competitive ELISA format, we compared the specificity and sensitivity of the format with a conventional serologic test reported in the literature 50,62 . There are several available kits for the direct DENV IgM and IgG detections by ELISA formats, which cannot identify the serotype, and their sensitivities and specificities for IgG detection are 45-56% and 93-95%, respectively, with Panbio Dengue Virus IgG Capture ELISA, and 55-89% and 64-99% with SD BIOLINE Dengue IgG ELISA 8,63,64 . Thus, we performed the conventional serologic test by incubating clinical samples on the plate immobilized with each serotype DEN-NS1 (The Native Antigen Company), followed by detection with a secondary anti-IgG HRP-conjugated antibody (Fig. 1d).
First, we compared the inhibition specificity by the competitive ELISA ( Fig. 5a) and the IgG detection by the conventional method (Fig. 5b), using various volumes (0.05-5 μL) of a Singaporean clinical plasma sample, PD2-4. Although the IgGs in the sample were not detected by the LFA kit, our method and the conventional methods clearly detected anti-DEN-NS1 IgGs. In our competitive ELISA, the aptamer binding to serotype 3 of DEN-NS1 (DEN3-NS1) was mainly inhibited with the clinical sample (0.1-5 μL of PD2-4), indicating that the IgGs in the sample specifically bind to DEN3-NS1. However, in the conventional method, the IgGs equally bound to all four DEN-NS1 serotypes and the serotype-specificity of the IgGs was unobservable.
Next, we extensively tested the IgG detections using 23 Singaporean clinical samples (PD1-1 to PD4-1) obtained within 3-5 days after fever onset (Supplementary Table 2). The serotypes of the current infection of each sample were initially determined by RT-qPCR and sequencing 49 , and the IgM and IgG detections were also performed with the LFA Panbio kit. The LFA tests detected anti-DEN-NS1 IgGs in the PD1-1, 1-9, 1-14, 1-19, 2-3, 3-2, 3-4, and 4-1 samples, suggesting that these patients had secondary infections. However, there are some discrepancies (including PD2-4, as shown in Fig. 5) among the results obtained using the LFA Panbio kit and our conventional and competitive ELISA methods (Fig. 6). The conventional and/or competitive ELISA methods detected IgGs in the samples of PD1-5, 1-10, 1-11, 1-18, 2-4, and 3-2, which were not detected by the LFA Panbio kit. In addition, the PD1-1, 1-10, 1-18, and 3-2 samples were IgG-positive by the conventional method, but the competitive ELISA detected no IgGs in these samples. Therefore, further extensive research is necessary to assess the sensitivity and specificity of our method by comparisons to those of the conventional method, using well-characterized samples before and after DENV infection.
Comparison of the competitive ELISA formats between aptamer-antibody and antibodyantibody sandwich systems. Competitive or blocking ELISA formats using the inhibition between protein-protein or protein-antibody interactions are well-known methods 6,65,66 . Theoretically, the DENV-serotypespecific IgG detection by the competitive ELISA format is also possible by using an antibody-antibody sandwich system, even if these antibodies have no specificity to each DEN-NS1 serotype (Fig. 7a) 67 . We tested this antibody-antibody system using Ab#D06 and Ab#D25, which do not compete with each other for the DEN-NS1 binding, and compared the longitudinal changes of the IgG detections with the aptamer-antibody system in the PD1-3, 2-2, 2-3, 3-4, and 4-1 samples ( Fig. 7 and Supplementary Figs. 3-6). The anti-DEN-NS1 IgG in the patient serum also inhibited the DEN-NS1 ternary complex formation with the antibody-antibody sandwich pair, and the test exhibited similar patterns to those obtained by the aptamer-antibody sandwich pair. However, the antibody-antibody pair could not detect the IgG activities in the early stage of infection, such as the day 5 sample of PD2-3 and the day 3 sample of PD4-1.
One of the plausible explanations for the difference in the sensitivities between the antibody-antibody and aptamer-antibody systems is the variations in the affinities of antibodies and aptamers to each DEN-NS1 protein serotype. We used high-affinity anti-DEN-NS1 antibodies for the competitive ELISA.  (Fig. 7a). Ab#D25 tightly binds to DEN1-NS1, DEN2-NS1, and DEN3-NS1 and might reduce the inhibitory effect of the anti-DEN-NS1 antibodies, as shown in the negative results in the acute phase of the clinical samples PD2-3, PD3-4, and PD4-1, in Fig. 7c. As compared to Ab#D25, the K D values of the aptamers are less biased to each DEN-NS1 serotype (K D = 182 pM for AptD1 to DEN1-NS1, 104 pM for AptD2 to DEN2-NS1, 57 pM for AptD3 to DEN3-NS1, and 30 pM for AptD4 to DEN4-NS1) 49 . To identify the serotype specificities of anti-DEN-NS1 IgGs, the ideal competitive ELISA would require the unbiased affinities of antibodies and aptamers against each DEN-NS1 serotype.
Cross-reactivity of the competitive ELISA format using the aptamer  Table 3 and Fig. 8). Since the acute-phase samples (2-7 DPO) corresponding to these patients were not available for this experiment, we used the previously obtained IgG detection data against Zika or dengue NS1 proteins in the acute phase (Supplementary Table 3 and Fig. 8). The anti-ZIKV-NS1 IgGs were detected by the H-zMut2 ELISA format 74 , which confirmed the positive results with anti-ZIKV-NS1 IgGs in all 21 convalescent-phase samples (Supplementary Table 3). The DEN-NS1 and IgM/IgGs to DENV were also tested by the SD BIOLINE Dengue Duo rapid IgG/IgM test (Dengue IgGs, Acute phase, Fig. 8 and Supplementary Table 3). However, the SD BIOLINE test uses mouse monoclonal anti-human IgG & IgM as the capture reagent on the membrane and gold colloid coated with the dengue virus envelope protein as the detector reagent, which could cross-react with IgGs to ZIKV 69,75,76 . Nineteen samples in the acute phase were negative for DEN-NS1 with the SD BIOLINE kit, and the other two samples, PZ-01 and PZ-05, were not tested.
Eight of the 21 samples (5 μL) from ZIKV patients did not inhibit the aptamer-DEN-NS1 binding for all dengue serotypes, and no cross-reactivity with the dengue competitive ELISA test was observed (Fig. 8). In contrast, the other 13 samples were positive in our competitive ELISA format with unique dengue serotype specificities, mostly serotype 2. Even small amounts (0.005-0.05 μL) of samples strongly inhibited the aptamer binding to DEN-NS1. In Singapore, many people could have been previously infected with DENV, especially serotype 2. Thus, these 13 patients may have had a prior DENV infection. The clinical samples from patients with the primary DENV infection and the secondary ZIKV infection reportedly showed strong cross-reactivity between ZIKA and DENV in the IgG detection 71 . Among these 13 samples, eleven acute phase samples were positive for the detection of the IgGs to DENV envelope proteins, using the SD BIOLINE test. Furthermore, three samples, PZ-03, PZ-15, and PZ-17, were negative for anti-ZIKV-NS1 IgG detection. Two samples, PZ-01 and PZ-16, showed a discrepancy between our competitive ELISA (anti-DEN-NS1 IgG positive in the convalescent phase) and the SD BIOLINE anti-DENV IgG test (negative to IgGs against DENV envelope proteins in the acute phase). PZ-20 revealed another discrepancy between our competitive ELISA (anti-DEN-NS1 IgG negative in the convalescent phase) and the SD BIOLINE test (positive to IgGs against DENV envelope proteins in the acute phase). These differences might result from the low sensitivity of the commercially available paper-based assay formats and the differences in the targeting DENV antigens. Together, these results suggest that our dengue competitive ELISA is not cross-reactive with anti-ZIKV-NS1 IgGs and can detect prior DENV infections in patients currently infected by ZIKV. We will further validate our method by comparisons with conventional methods and kits, using wellcharacterized clinical samples with sufficient patient information.
To confirm the absence of cross-reactivity of the competitive ELISA with anti-ZIKV-NS1 antibodies, we examined the inhibition by monoclonal antibodies, Ab#Z11 and Ab#Z12, prepared in-house for an antibody-antibody sandwich ELISA format to selectively detect ZIKV NS1 proteins. Both antibodies (100 fmol) showed no inhibition (Supplementary Fig. 7, top and middle panels). In contrast, the monoclonal antibody, Ab#D25 (Fig. 7a, used as the capture agent for the antibody-antibody sandwich ELISA system), effectively inhibited the aptamer-DEN-NS1 binding ( Supplementary Fig. 7, bottom panel). . Anti-DEN-NS1 IgG detection in Singaporean clinical samples using the competitive Apt/Ab ELISA format. The clinical human serum or plasma samples (PD1-1 to PD4-1) were obtained in 2016 in Singapore. The direct DEN-NS1 detection of the samples at the early stage (3-5 days after fever onset) was performed by the SD BIOLINE Dengue NS1 Ag rapid test. The IgM and IgG detections were performed with a Panbio Dengue Duo Cassette. The quantitative IgG activities to each serotype DEN-NS1 in the clinical samples were determined by our competitive Apt/Ab ELISA format, which also identified the current infection status (primary or secondary) and the serotype of the past infection (past). The discrepancies with the IgG detection between the Panbio LFA and our competitive ELISA data are indicated in red. The RT-PCR analysis of the samples at the early stage also confirmed the DENV serotype caused by the current infection. www.nature.com/scientificreports/ We also examined the cross-reactivity of our competitive ELISA with antibodies resulting from COVID-19 vaccination. During this research, the two healthy volunteers (PD0-1 and PD0-2), whose serums were used as controls (Fig. 3e,f), were vaccinated with the Pfizer-BioNTech COVID-19 vaccine. The production of SARS-CoV2 neutralizing antibodies was detected with a GenScript cPass kit ( Supplementary Figs. 8a and b). Using these serum samples before and after vaccination, we tested our dengue competitive ELISA format and confirmed the absence of cross-reactivity with the COVID-19 vaccination in the competitive ELISA format ( Supplementary  Fig. 8c).

Discussion
We have presented serotype-specific detection methods as a serologic test for IgG antibodies to DEN-NS1 in human blood samples, by the competitive ELISA format using high-affinity UB-DNA aptamers. Only IgGs with strong affinity to each of DEN-NS1 serotypes inhibit the DEN-NS1 detection by the UB-DNA aptamer-antibody sandwich system, allowing for the serotype-specific IgG detection. This IgG detection provides valuable information for the dengue diagnostics and the development and use of the dengue vaccine. (1) The competitive ELISA format can identify the secondary infection within several days (during the febrile period) after fever onset. If anti-DEN-NS1 IgG antibodies are detected in patients within one week after fever onset, then this indicates a secondary infection that may warrant close monitoring. (2) The format can identify the serotypes in the late stage of the primary DENV infection and in the early stage of the secondary infection, in which the serotypes of the past infection can be determined. In combination with the antigen-detection ELISA format using the UB-DNA aptamer-antibody sandwich system 49 , the serotypes of both the current and past infections, as well as the infection status, could be determined in the early stage of DENV infection, enabling prompt and specific medical treatment. (3) The format can check whether a person has been previously infected with DENV and determine the serotype. In the currently used field test for dengue infection, the serotype identification is still optional. However, as shown by the increase in severity by secondary infection, rapid test kits that identify the serotype could facilitate the advancement of dengue epidemiology and pathology, by accumulating the data of  Table 2). (b) Direct IgG detection by a typical conventional ELISA method (Fig. 1d)  www.nature.com/scientificreports/ www.nature.com/scientificreports/ the relationships among the serotypes, antibody production, and symptoms of dengue patients. (4) The format could also identify the dengue infection history of ZIKV-infected patients. Our data about the cross-reactivity with ZIKV infection using convalescent-phase clinical samples (Fig. 8) revealed two distinct patterns: one showed no cross-reactivity between dengue and Zika, in which patients were infected by only ZIKV, and the other exhibited strong inhibition in our format, in which patients might have been infected with DENV previously. For further evaluation, we will examine our dengue format using both the acute-and convalescent-phase ZIKV patient samples. (5) The format could provide valuable information for the usage and analyses of the dengue vaccines, for which documentation of prior infection is important prior to administration, due to concerns about antibody-dependent enhancement (ADE). For example, Dengvaxia can only be used in people previously infected with DENV, and thus the competitive ELISA can identify previous DENV infections. (6) The format could be useful for vaccine development to identify the specificity to antigen variants. To this end, the UB-DNA aptamer generation targeting viral particle or envelope proteins is more helpful. Our tests using patient samples with secondary DENV infections (Fig. 4) revealed that the IgG antibodies responding to the past infection were produced predominantly, even upon secondary infections with different dengue serotypes. These results correlate with other reports [21][22][23][24]26,27,61 , supporting ADE where secondary heterologous infections occasionally result in severe symptoms and highlighting the risk of vaccinating denguenaïve individuals. Patients with a primary infection produced IgG antibodies that mainly targeted the infected serotype. In the secondary infection, the initially produced IgG antibodies reacted more to the NS1 serotype of the past infection, and did not effectively react with the targets of the secondary infection. The application of Figure 8. Serotype-specific anti-DENV-NS1 IgG detection using convalescent-phase ZIKV clinical samples by the competitive Apt/Ab ELISA format. For the competitive ELISA tests, 5 µL of diluted plasma samples (corresponding to the use of 0.0005-5 µL) were used and the amounts of the spiked DEN-NS1 were adjusted to obtain similar signal intensities, as shown in Supplementary Fig. 1. ZIKV IgG detection patterns in each acutephase sample are from the ELISA results reported previously 74  www.nature.com/scientificreports/ this test in a larger cohort of dengue patients will clarify the mechanism of dengue pathogenesis, through the serotype-specific sequence of the DENV infection. Interestingly, this observation of patients with a secondary dengue infection is very similar to the observations of ZIKV patients who might have been previously infected by dengue (Fig. 8). In the future, our method may be expanded to test the efficacy of vaccine development 50,51 , and to diagnose other diseases and allergies. This competitive ELISA format is a simple method using UB-DNA aptamers, which enable the quantitatively identification of serotype-specific IgGs. A similar IgG detection concept using conventional DNA aptamers was used to detect the IgG antibodies to the P48 protein of Mycoplasma bovis 77 . However, the affinities of the DNA aptamers to the target were relatively low (K D = 16-33 nM) due to the conventional 4-letter DNA aptamers, and thus the background in the IgG detection was high and the quantitative analysis was difficult. Furthermore, this competitive method using antibodies is also applicable to DNA aptamer generation, especially for cell-SELEX to selectively isolate aptamer candidates for desired targets on the cell-surface 78,79 .
For the serotype-specific detection by the competitive ELISA format, unbiased affinities and sensitivities of ligands to each serotype of DEN-NS1 proteins might be important. In the ExSELEX procedure, we generated four UB-DNA aptamers targeting each DEN-NS1 serotype by isolating each aptamer using the same selection pressure, in the combination of the ELISA-type sandwich system with the antibody (Ab#D06) pair 49 . Therefore, the ELISA formats for each DEN-NS1 serotype exhibit similar sensitivities: the limit of detection (LOD) values in 10% human serum is 6.97 ng/ml for DEN1-NS1, 6.91 ng/ml for DEN2-NS1, 11.06 ng/ml for DEN3-NS1, and 4.10 ng/ml for DEN4-NS1 49 .
Although the antibody-antibody sandwich system can also be used for the IgG detection, the sensitivity is lower than that of the UB-DNA aptamer-antibody sandwich system, especially in the early stage of infection, due to the biased affinities to each serotype (Fig. 7). In addition, this sensitivity difference between these systems might also result from the different binding modes between aptamer-target and antibody-target. In general, nucleic acid aptamers cover wider areas of target proteins, as compared to those covered by antibodies, and thus a series of IgGs that bind to a region within the wide area of the targets can inhibit the aptamer binding. Our competitive ELISA format using four serotype-specific UB-DNA aptamers can be used for the serotype-specific identification of IgGs with the lateral-flow assay (LFA) system, in which each aptamer is embedded in each slit on the single membrane as the capture agent.
As shown in Fig. 3, the commercially available human serum contains considerable amounts of anti-DEN-NS1 IgGs, presuming the widespread DENV epidemic (or endemic) where the samples were collected. This also draws attention to the use of serum as a control in epidemic and pandemic disease studies, such as the current COVID-19 pandemic. In our study, we used serum sample 2 (Fig. 3e) as the DENV-related IgG-free control serum.
The combination of this competitive ELISA for IgG detection with the original ELISA format for the serotypeor sub-serotype-specific DEN-NS1 detection 49 allows the diagnosis of both past and current viral infections and will facilitate early medical care. Our competitive ELISA method could be employed for a wide range of IgG detections for infectious diseases, allergenic reactions and sensitivities, and vaccine development. By generating UB-DNA aptamers targeting each antigen or its variants, the competitive ELISA format using each aptamer/target pair could identify the specificity of IgGs produced by responses to infection, allergens, and vaccine treatments. Recently, SARS-CoV-2 serological assays with competitive or blocking ELISA formats, using the interaction between ACE2 and the viral spike protein, have been utilized to detect neutralization antibodies 80,81 and are commercially available as a test kit (cPass from GenScript). However, there is a possibility that the serologic test kits have the problem of cross-reactivity between dengue and COVID-19 for both types of patients 9,10 . As noted above, UB-DNA aptamers cover a large area of target proteins, and our competitive ELISA format could detect not only the neutralization antibodies but also other IgGs that competitively bind to different target areas.

Methods
Oligonucleotides, proteins, and clinical samples. Biotinylated UB-DNA aptamers (AptD1, AptD2, AptD3, and AptD4, Fig. 2b and Supplementary Table 1) were chemically synthesized with an H8 DNA/RNA Synthesizer (K&A Laborgerate) in-house by using natural base, Biotin-dT, and Ds and Pa phosphoramidites. The natural base and Biotin-dT phosphoramidites were purchased from Glen Research and LGC Link Technologies. The Ds and Pa phosphoramidites were prepared in-house, as described previously 52,53 . The synthesized UB-DNA aptamers were purified by denaturing gel electrophoresis after deprotection. Recombinant DENV1-4 NS1 proteins (DEN1-NS1, DEN2-NS1, DEN3-NS1, and DEN4-NS1) were purchased from The Native Antigen Company. Control human serum was purchased from Sigma-Aldrich (Sigma #H4522, pooled) and obtained from a healthy Singaporean volunteer (human serum sample 1), or two different Japanese volunteers (human serum samples 2 and 3, corresponding to PD0-1 and PD0-2) who have never been infected by dengue viruses. Clinical samples (serum or plasma) used in this study were prepared from the whole-blood samples of the patients who consented to the study and provided written informed consent, referred by the Communicable Disease Center, TTSH. The study protocol was approved by the National Healthcare Group Domain Specific Review Board (references 2015/00528 and 2016/00076) and by the SingHealth Centralized Institutional Review Board (reference no. 2016/2219). Anti-dengue NS1 rabbit monoclonal antibodies, Ab#D06 and Ab#D25, and anti-Zika NS1 rabbit monoclonal antibodies, Ab#Z11 and Ab#Z12, were prepared in-house 49 . The Streptavidin-HRP conjugate (1 mg/mL) and Peroxidase-conjugated AffiniPure Goat Anti-Human IgG (H + L) (#109-035-003, 1 mg/mL) were obtained from Jackson ImmunoResearch. Streptavidin, Tween 20, BSA, and anti-rabbit IgG HRP conjugate (1 mg/mL) were obtained from Promega. The 10 × D-PBS solution was purchased from Nacalai Tesque.
The present experiments were approved by the institutions of A*STAR and NCID, and we confirm that all methods were performed in accordance with the guidelines and regulations of A*STAR and NCID/TTSH. Competitive IgG detection using Apt/Ab and Ab/Ab ELISA formats. For the assays by the Apt/Ab ELISA format, we used the wells coated with each UB-DNA aptamer as a capture agent. For the preparation of the loading samples, a serum sample (5 μL, directly or 10, 25, 50 or 100-fold diluted with dilution buffer) was first mixed with 0.5 μL of each NS1 protein (DEN1-NS1: 350 or 200 pg, DEN2-NS1: 350 or 500 pg, DEN3-NS1: 450 or 875 pg, DEN4-NS1 200 or 275 pg). The spiked NS1 protein amounts were adjusted to each DEN-NS1 protein lot used in the assays, so that the OD 450 values only overflowed in dilution buffer, since the detection sensitivity of the spiked NS1 varied depending on the purchased DEN-NS1's lot ( Supplementary Fig. 1). After a 30-min incubation at 25 °C, the solution was mixed with 45 µL of 11.1 nM Ab#D06 in dilution buffer 2 and incubated for 30 min, followed by loading into the aptamer-coated well (50 µL) and a 30-min incubation. The subsequent procedures were performed as described above for the Apt/Ab ELISA. For the assays by the Ab/Ab ELISA format, we used the wells coated with Ab#D25 (overnight) as a capture agent. For the preparation of the loading samples, a serum sample (5 μL, directly or 10, 25, 50 or 100-fold diluted with dilution buffer) was first mixed with 0.5 μL of each NS1 protein (DEN1-NS1: 400 pg, DEN2-NS1: 250 pg, DEN3-NS1: 400 pg, DEN4-NS1: 300 pg). The spiked NS1 protein amounts were adjusted by each DEN-NS1 protein so that the OD 450 values only overflowed in the 10% control human serum sample, since the background signal levels were different between dilution buffer and 10% human serum ( Supplementary Fig. 1) 49 . After a 30-min incubation at 25 °C, 45 µL of the solution was mixed with 11.1 nM biotinylated Ab#D06 in dilution buffer 2. The subsequent procedures were performed as described above for the Ab/Ab ELISA.
From the plots of OD 450 against the volume of human serum used in the ELISA, the relative IgG activity was calculated through normalization of the serum volume required to exhibit an OD 450 lower than 1.0 (5/[the serum volume required for an OD 450 of 1.0]), as shown in Supplementary Fig. 2. Typical conventional method for IgG detection by ELISA. For the assays, we prepared the wells coated with each DEN-NS1, by a 2-h incubation of 1 µg/mL DEN-NS1 (The Native Antigen Company, 50 µL/ well) in 0.1 M sodium carbonate buffer (pH 9.6) at 25 °C, followed by blocking with BSA. A serum sample (0.05 to 5 μL) in dilution buffer (50 µL) was loaded in the well and incubated for 30 min at 25 °C. After washing the wells once, 100 µL of secondary detector solution (Peroxidase-conjugated AffiniPure Goat Anti-Human IgG (H + L), diluted 1: 2,500 with dilution buffer) was added to each well and then incubated for 30 min, followed by TMB detection as described above in the Apt/Ab and Ab/Ab ELISA formats.