Ethics statement

Anonymized human serum samples (2016-2019) were obtained from the Laboratory of Tropical Diseases, San Vicente de Paul Hospital, Orán, Salta, Argentina. The protocol was approved by the Provincial Commission on Health Science Research (COPICSA), Ministry of Health of Salta, under reference number 321-196940/18, and complies with the Declaration of Helsinki. Sample information has been previously published [45].

NS1 protein production

NS1-containing supernatant from different serotypes was obtained by infecting C6/36 insect cells with DENV-1 (Hawaii), DENV-2 (New Guinea C), DENV-3 (Philippines/H87/1956), and DENV-4 (H241) at a multiplicity of infection (MOI) of 1. Five days post-infection (p.i.), the supernatant was harvested, clarified by centrifugation, and concentrated four times by ultrafiltration (30 kDa cut-off, Vivaspin). Culture media was then exchanged to phosphate buffer saline (PBS) and used to immunize a llama.

For cross-reactivity assays, Vero CCL-81 cells were infected with DENV-1 to -4, Chikungunya virus (CHIKV), and Yellow Fever virus (YFV) at a MOI of 0.1. Cell culture supernatants were collected between days 3 and 7 p.i., clarified by centrifugation, filtered, and stored at −80 °C until use. Viral titers were quantified by plaque assay and ranged from 1 × 10⁶ to 1 × 10⁷ plaque-forming units (PFU) per milliliter. Viral strains were kindly provided by Dr. Alicia Mistchenko from the Laboratory of Virology, Hospital de Niños Dr. Ricardo Gutiérrez, Buenos Aires, Argentina.

Recombinant NS1 was generated through PCR amplification, followed by cloning using either restriction enzyme-based or seamless approaches. Primers are listed in S1 Table. The cDNA template was prepared from RNA extracted from C6/36 insect cells infected with the dengue virus strains previously mentioned. Two groups of constructs were used to clone NS1 from all four DENV serotypes: a pCAGGS-ss vector containing the human serum albumin secretion signal and a His-tag [46], and a modified pFUSE vector (invivoGene) with an IL2 secretion signal, a Flag-tag at the N-terminus, and a His-tag at the C-terminus. Recombinant NS1 protein from Zika virus (MR766) was also cloned into the same vectors. For protein extraction, vectors were transfected into HEK-293T cells using polyethyleneimine (PEI, polyAR). The supernatant was collected 72 hours post-transfection, adjusted with phosphate buffer or Tris-NaCl, and purified by IMAC using a Ni-NTA resin. DENV-2, cloned into the pCAGGS-ss vector, was expressed in Vero cells for biopanning experiments. Transfection was carried out on a 96-well plate using Lipofectamine 2000. Each well was seeded with 1.5 × 10⁴ Vero cells. After 24 hours, 60 ng of plasmid DNA, diluted in 50 µL of Opti-MEM, was mixed with 0.2 µL of Lipofectamine 2000 (also diluted in 50 µL of Opti-MEM). The mixture was incubated at room temperature for 25 minutes to allow complex formation. Transfection complexes were then added to the cells, which were incubated for 48 hours at 37°C in a 5% CO₂ incubator to enable protein expression.

Selection of NS1-specific Nanobodies

Animal immunizations were approved by the Ethical Committee of the Faculty of Veterinary Sciences, University of Buenos Aires (updated protocol 2022/23). A llama was immunized with five injections of 250 µL of purified supernatant containing NS1 of each DENV strain. Complete Freund’s adjuvant was used for the first immunization and Incomplete Freund’s for boosters on day 28, 41, 58, 83 and 90. Serum samples were collected before the first immunization and two weeks after each injection.

Antibody titers were monitored by ELISA using 96-well plates coated with 100 ng of recombinant NS1 proteins. Plates were blocked with 3% skimmed milk in PBST (PBS + 0.05% Tween 20), then incubated with five-fold diluted serum. Horseradish peroxidase (HRP)-linked anti-llama IgG (Bethyl Laboratories, 1:15,000) was added, and the reaction was developed with 3,3’,5,5’-Tetramethylbenzidine (TMB) and read at 450 nm. The endpoint titer was considered as the highest sample dilution with an absorbance at 450 at least double that of the pre-immune serum sample.

Four days after the final immunization, 150 mL of blood was collected. Peripheral blood lymphocytes were isolated using Ficoll Paque Plus filled Leucosep tubes, and RNA was extracted using RNeasy Midi columns. The VHH-gene library was constructed following the protocol published by Pardon et al. [47].

Biopanning was conducted in 96-well plates using either recombinant NS1-1 (cloned from DENV-1, 0.2 μg/well), or Vero cells transfected with pCAGGS-ss-NS1-2 construct (cloned from DENV-2, 60 ng/well) and fixed with acetone/ethanol 24 hours post-transfection. Welles prepared for both selection strategies were incubated at room temperature for 1 hour with 2% skimmed milk in PBST. Recombinant phages were then subjected to three successive rounds of counter-selection, each involving a 15-minute incubation at room temperature of 200 µL of phage suspension (at a concentration of 10¹² phages/mL) in wells containing protein extracts derived from non-transfected HEK293T or Vero cells, as appropriate. Counter selection steps were included to diminish non-specific phage binding, as the llama was immunized with a supernatant containing several irrelevant proteins. After the final counter-selection step, 100 µL of the collected phage suspension was transferred to the selection wells containing either the recombinant NS1-1 protein or transfected Vero cells. The remaining 100 µL was added to wells containing protein extracts from non-transfected HEK293T or Vero cells, serving as negative controls. Plates were incubated for 2 hours at room temperature. Non-specifically absorbed phages were removed by increasing the number of washes with PBST after each round of panning. Bound phages were eluted a two-step elution process, using trypsin solution (T-elution) and exponentially growing TG1 cells (C-elution) as previously published [48]. The first elution step involved the incubation for 30 minutes at room temperature with 100 µL per well of TPCK-treated trypsin (0.25 mg/mL in PBS). Enzymatic activity was quenched by adding 5 µL per well of AEBSF protease inhibitor (4 mg/mL). A second elution step consisted of the incubation at 37°C for 30 minutes with 200 µL of exponentially growing E. coli TG1 cells. Serial dilutions of infected bacteria were seeded to control antigen-specific phage enrichment and to select single colonies for further characterization.

Screening and characterization of NS1-specific Nanobodies

A total of 192 individual colonies from the second and third rounds of panning, conducted using either T- or C-elution protocols with recombinant NS1 protein or transfected Vero cells, were analyzed by ELISA. Nanobodies were extracted after induction with 1 mM IPTG for 6 hours at 37°C. ELISA plates were coated overnight at 4°C with 0.2 μg/mL of recombinant NS1-1 or NS1-2 proteins. The next day, plates were blocked with 3% skimmed milk in PBST, then incubated with 100 µL of periplasmic extract for 2 hours at room temperature. Plates were washed three times, and 50 µL/well of HRP-conjugated anti-HA antibody (1:1,500) was added for 1 hour to detect specific binding. The reaction was developed with TMB and read at 450 nm.

Ninety-six clones (1–48 selected using recombinant NS1-1 protein and 49–96 using a transfected DENV-2 NS1 plasmid) were amplified via PCR with primers listed in S1 Table. The PCR fragments were then analyzed by RFLP, utilizing the HinfI restriction enzyme for digestion. Samples were run on a 2.5% agarose gel to determine fragment patterns.

After sequencing, genetic similarity was analyzed using a Neighbor-joining tree with 1000 bootstrap replicates (MEGA version 11) [49]. The sequence logo was plotted using WebLogo3. Selected Nb sequences were subcloned into the pHen6 vector using NcoI and NotI restriction enzymes, and WK6 cells were transformed for Nb expression and scale-up.

Nanobody modifications

To use the selected Nbs for NS1 protein capture or detection in the enzyme immunoassay, we made several modifications. Two polystyrene (PS) binding peptide sequences, rich in tryptophan (VHWDFRQWWQP, called PSW) and rich in arginine (RIIIRRIRR, called PSR) were selected to modify Nbs for NS1 capture [50,51]. Nanobody sequences were amplified by PCR using specific primers and cloned into a vector with the PSW or PSR sequence at the N-terminal. The modified Nbs with the PSR sequence encountered expression and purification issues during protein production scaling; therefore, in this study, only the Nbs modified with the PSW sequence were used. Bivalent Nbs were prepared by cloning the same Nb separated by a 20 Glycine-Serine linker (G₄S)₄. To produce the Nb Fc fusions, Nb sequences were subcloned into a modified pFUSEss eukaryotic expression vector (InvivoGen). This vector incorporates an IL-2 secretion signal and the murine IgG2a fragment, comprising the hinge, CH2, and CH3 domains.

Nanobodies for NS1 detection were modified by biotinylation applying two different techniques. In one, Nb sequences with an AviTag and His-tag at the C-terminus were cloned into a pET22-based vector with the BirA biotin ligase sequence under the T7 promoter (Addgene plasmid # 100817) [52]. The second method involved the use of a sortase A reaction after adding the LPETG target sequence at the C-terminus of each Nb via Gibson Assembly (Addgene plasmid # 75144) [53,54]. Nbs for detection were also modified by adding an Fc tag followed by the HRP coding sequence. The CH1 domain was deleted from a pFUSE vector, the stop codon was removed, and the HRP sequence was cloned into the same ORF. All constructs were made using In-Fusion seamless cloning with primers listed in S1 Table.

Expression and purification of Nanobodies

Production and purification of the selected clones (n = 11), as well as Nbs PSW and bivalent Nbs, were carried out at 37°C in Terrific Broth (TB) medium supplemented with 100 μg/mL ampicillin and 0.1% glucose. Nanobody expression was induced with 1 mM IPTG for 16 hours at 28°C. The proteins were purified from periplasmic extracts by IMAC using a Ni-NTA resin and eluted with 300 mM Imidazole.

To biotinylate Nb with BirA, plasmids were transformed into BL21 bacteria, expression was induced with IPTG at 28°C. The cell culture was centrifuged, and the pellet was freeze-thawed in the presence of 50 µM D-biotin (Sigma), to rupture cell membranes and allow interaction between Nbs and biotin ligase. Bacterial lysate was incubated at 37°C for 30 minutes to enhance biotinylation yield [55]. Biotinylated Nbs were then purified by Ni-NTA resin using chromatography columns. In the case of biotinylation with sortase A, purified Nbs (50-70 μM) were incubated with 5 μM sortase A and 5 μM Biotin-PEG-NH2 in a reaction buffer (150 mM NaCl, 50 mM Tris pH 7.5, 10 mM CaCl₂) for 4 hours on ice. Unreacted Nbs and sortase A were removed using Ni-NTA resin, while excess nucleophile was eliminated with a PD-10 desalting column [56].

Nanobodies Fc and Nbs Fc-HRP fusions were transiently expressed in HEK-293T. The supernatant containing recombinant proteins was harvested 72 hours post-transfection and purified by Protein G affinity chromatography. Fractions containing the recombinant proteins, as judged by the SDS-PAGE analysis, were pooled, dialyzed against PBS, and stored at −80°C.

Evaluation of Nanobodies properties for capture and detection

To assess the ability of the different Nbs to capture NS1, 96-well ELISA plates were coated with 10 μg/mL of Nbs overnight at 4°C. Plates were blocked for 1 hour at room temperature with 5% equine serum, 3% BSA, 1% BSA, or 3% milk in PBST. After three PBST washes, 100 ng/well of recombinant NS1-2 was added and incubated for 1 hour and 30 minutes at room temperature. Plates were later incubated with anti-Flag HRP-conjugated monoclonal antibody (1:10,000) for 1 hour and 30 minutes at room temperature to detect NS1. An irrelevant Nb (specific for the receptor-binding domain of SARS-CoV-2) was used as a negative control. The reaction was developed with TMB and read at 450 nm. To compare the capture efficiency of different Nbs, microplates were coated with a serial three-fold dilution of Nbs, starting from 800 nM, and 50 ng/well of recombinant NS1-2 was used as antigen. Detection of NS1 was assessed as described in the former assay. These experiments were performed in triplicates. All Nbs were also evaluated as capture antibodies—both in their monovalent form and as Fc-fusion constructs—using clinical serum samples, following the establishment of the ELISA protocol (see Optimization of the Nanobody-based ELISA).

To evaluate the NS1 detection capacity of selected Nbs, MaxiSorp 96-well ELISA plates were coated overnight at 4°C with 1 μg/mL recombinant NS1-2 protein in carbonate/bicarbonate buffer (pH 9.6). Plates were blocked with 10% skimmed milk in PBST for 1 hour at 37°C. Purified Nbs were adjusted to a starting concentration of 1 μM, then serially diluted 5-fold and added to the plates for 1 hour at 37°C. A monoclonal Anti-Dengue virus NS1 protein antibody (BEI resources NR-10121) and an irrelevant Nb were also serially diluted in a 5-fold dilution series and used as positive and negative controls, respectively. A commercial HRP-conjugated anti-VHH antibody (1:8,000) was added. The reaction was developed with TMB and read at 450 nm. The EC₅₀ was estimated from triplicates measurements by a four-parameter log-logistic regression model. Similar conditions were applied when Nbs coupled to HRP were used as detection antibodies.

To evaluate potential epitope overlap among Nb14, Nb48, and Nb66, a competitive ELISA was conducted using unlabeled monovalent Nbs as competitors and their corresponding Fc-fusion formats as detection probes. Ninety-six-well ELISA plates were coated overnight at 4°C with NS1-2 protein (2 µg/mL), followed by blocking with 3% (w/v) skim milk in PBST for 1 hour at room temperature. Serial two-fold dilutions of the unlabeled Nbs (starting at 60 µg/mL) were incubated with a fixed concentration of Nb-Fc (0.4 µg/mL) and added to the wells for 1 hour. After washing, bound Nb-Fc was detected using an HRP-conjugated anti-mouse antibody (1:5,000 dilution). Signal development was carried out using TMB substrate as previously described.

Optimization of the Nanobody-based ELISA

Optimal concentrations of capture and detection antibodies were determined by a chessboard titration experiment. Microplates were coated overnight at 4°C with serial two-fold dilutions of Nb14 Fc, starting from either 4 or 2 µg/mL. Plates were blocked with 1% BSA in PBST for 1 hour at 37°C. After three PBST washes, 50 ng/well of recombinant NS1-2 was added and incubated for 1 hour and 30 minutes at 37°C. Each capture antibody concentration was tested against six detection antibody concentrations (Nb48 or Nb66, in Fc-HRP or biotinylated format): 1, 5, 10, 15, 20 and 25 µg/mL, incubated for 1 hour and 30 minutes at 37°C. For biotinylated Nbs, an additional incubation with streptavidin-HRP (1:5,000) was performed for 1 hour at 37°C. The reaction was developed with TMB and read at 450 nm. The assay was run with duplicate wells for each sample.

Production of a rabbit-derived polyclonal anti-NS1 serum

Experiments with rabbits were conducted in collaboration with researchers from the Universidad Nacional de Quilmes (approved protocol 001/16, updated protocol 004/24). A rabbit was subjected to a five-dose immunization protocol over a four-month period. The first two immunizations (day 0 and day 28) used 200 μg of pCAGGS NS1-1, NS1-3, and NS1-4 plasmids. A protein booster (30 μg of recombinant NS1-1, NS1-3, and NS1-4) was administered at day 56. The final two immunizations (day 91 and day 119) used 200 μg of pCAGGS NS1-1 plasmid. Blood was collected, and serum was extracted and purified to isolate IgG produced in response to immunization. Purified rabbit IgG titration was performed in triplicate via ELISA on plates coated with 2 μg/mL of each NS1 protein. Specific recognition was detected using an anti-rabbit HRP-coupled antibody (1:3,000).

Development and statistical validation of a combined Nanobody ELISA assay

To assess the accuracy and performance of the combined Nb ELISA (cNb-ELISA), serum samples from patients, categorized as NS1-positive using PLATELIA DENGUE NS1 Ag (BioRad) (n=94) or epidemiologically negative (n=74), were analyzed. Each assay plate included two internal controls: a pool of highly reactive positive sera and a low-reactivity negative serum. Microplates were coated with 10 μg/mL of anti-NS1 purified rabbit IgG in carbonate/bicarbonate buffer and incubated overnight at 4°C. The next day, plates were blocked with 1% BSA in PBST for 1 hour at 37°C. After four PBST washes, 25 ng/well of recombinant NS1 from the four DENV serotypes and ZIKV in a negative serum matrix, a pool of positive sera (diluted 1:2), or individual patient samples (diluted 1:2) were added and incubated for 1 hour and 30 minutes at 37°C. After washing, 5 μg/mL of biotinylated Nb66 was added and incubated for 1 hour and 30 minutes at 37°C. An additional incubation with streptavidin-HRP (1:5,000) was performed for 1 hour at 37°C. The reaction was developed with TMB and read at 450 nm. The cNb-ELISA protocol was further employed to assess the specificity of the detection system against other circulating viruses. For this purpose, supernatants from Vero cells infected with DENV serotypes, ZIKV, CHIKV, and YFV were analyzed.

Statistical methods included in the combined Nanobody ELISA validation

Absorbance values at 450 nm were used to construct a receiver operating characteristic (ROC) curve to evaluate the assay’s diagnostic performance (cut-off and diagnosis sensitivity and specificity). The area under the ROC curve (AUC) and its 95% confidence interval (CI) were calculated to determine the assay’s discriminative ability. The optimal cut-off value was established based on the Youden index and corroborated using the MaxSpSe method. Diagnostic sensitivity and specificity were then computed at the determined cut-off, with their respective 95% CIs. Data analysis was performed using the easyROC package in RStudio (version 3.6.1).

Concordance or agreement between reference samples (positive detected by PLATELIA DENGUE NS1 Ag (BioRad) assay performed at Orán Hospital, and negative assigned by an epidemiological algorithms) and our developed assay was assessed by the calculation of the Cohen’s kappa index [57]. Samples giving discordant results were tested using Dengue virus NS1 Ag (DIA.PRO), according to the provider protocol.

To determine the limit of detection (LOD), two-fold serial dilutions of recombinant NS1 of the indicated serotypes were spiked into 100% human sera from non-infected patients. Every sample was measured in triplicates and data was fitted to a 4-parameter logistic equation:

The LOD was calculated based on the standard deviation of the response (Sy) of the curve and the slope of the calibration curve (S) at levels approximating the LOD according to the formula: LOD = 3.3*(Sy/S).

The analytical sensitivity of the cNb-ELISA was evaluated using a pooled set of positive sera, serially diluted twofold from 1:2 to 1:256. Absorbance values at 450 nm were plotted against the base-10 logarithm of the dilution factor, and nonlinear regression analysis was performed. The linear range of the assay was identified by linear regression analysis, evaluating the significance of the linear relationship through the calculation of the coefficient of determination (R²), along with the corresponding slope and intercept values to ensure the robustness of the system.

Production and characterization of DENV NS1 specific Nanobodies

Heavy-chain-only antibodies specific to the NS1 protein were raised in a llama immunized with purified supernatants obtained from C6/36 insect cells infected with four DENV serotypes. After five immunizations, anti-DENV NS1 specific antibody titers increased (S1A Fig), and an immune library of 1 × 10⁸ transformants was constructed using peripheral blood lymphocytes. Colony PCR on randomly picked colonies identified the presence of inserts in 78.5% of transformants (S1B Fig). Recombinant and transfected NS1-1 and NS1-2 specific binders were selected through three rounds of panning using two elution strategies established in our laboratory [48]. Although all dengue serotypes have been reported to circulate in Argentina, we selected serotypes 1 and 2 for panning, as they have shown the highest frequency of circulation in the country in recent years [58–61]. Cross-reactivity of Nbs to NS1-1 and NS1-2 was analyzed by ELISA using wells coated with recombinant NS1 proteins. As shown in S2 Fig, most clones exhibited varying degrees of reactivity against both NS1-1 and NS1-2. Ninety-six clones were subjected to RFLP sequence analysis, revealing 18 different restriction patterns (S3 Fig). Plasmids were sequenced and 11 distinct Nbs (Nb5, Nb7, Nb14, Nb22, Nb34, Nb40, Nb48, Nb51, Nb-4, Nb66, and Nb70) with conserved residues in the framework regions (FRs) and high variability in the complementarity-determining regions (CDRs) were identified (S4 Fig). Phylogenetic and bioinformatic analysis of the selected Nbs sequences revealed three main groups, providing insights into their germline origin (S5 Fig).

Six Nbs were selected for further characterization based on phylogenetic analysis: four representatives of the identified phylogenetic groups (Nb5, Nb22, Nb48, and Nb66) and two ungrouped Nbs (Nb14 and Nb40). Despite numerous attempts, Nb70 and Nb40 failed to achieve sufficient expression levels for downstream assays and were therefore excluded from most experiments.

The specificity and affinity of the selected Nbs was tested by ELISA against recombinant NS1 proteins from all DENV serotypes and ZIKV. All Nbs demonstrated specificity for DENV NS1 proteins, with varying degrees of relative affinity depending on the Nb and serotype. None of the Nbs showed reactivity toward ZIKV NS1 (S6 Fig). Notably, all Nbs, except Nb22, exhibited single digit nanomolar EC₅₀ values, indicating strong binding to DENV NS1 (Table 1). Nb14, Nb40, and Nb66 displayed the highest relative affinities, highlighting their potential as diagnostic reagents for assay development. A monoclonal Anti-NS1 protein antibody and an irrelevant Nb were included as positive and negative controls, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. EC₅₀ values of selected Nanobodies against DENV NS1 proteins.

https://doi.org/10.1371/journal.pntd.0013168.t001

To identify shared epitopes among the different Nbs under investigation, we performed a competition assay. In each reaction, we combined Fc-tagged Nbs—whose binding was detected using an HRP-conjugated anti-mouse antibody—with monovalent Nbs. This approach allowed us to determine that Nb48 and Nb14 compete for the same binding site on the NS1 protein, whereas the epitope recognized by Nb66 is located elsewhere on the protein (S7 Fig).

The selected Nbs were analyzed by ELISA to identify the best Nb for NS1 antigen capture. Based on binding data (S6 Fig and Table 1), Nb14 demonstrated strong suitability for NS1 capture and was used to establish assay conditions. Optimal assay performance was determined by comparing two microplate brands and three blocking reagents (3% and 1% BSA, 5% equine serum, and 3% skimmed milk). Results indicated that 1% BSA was the most effective blocking reagent for recombinant NS1 capture when paired with Greiner microplates, exhibiting the highest P/N ratio (Fig 1A). Under optimized conditions, the capture capacities of nine Nbs for NS1-2 were evaluated at two concentrations using an anti-Flag HRP antibody for detection. Nb14 demonstrated the highest capture efficiency, with absorbance values close to 1 at both concentrations (Fig 1B). Nanobody 48 also showed the ability to capture NS1-2, though with lower absorbance values compared to Nb14. These findings suggest that Nb14 and Nb48 are promising candidates as capture antibodies for the diagnostic assays.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Determination of optimal conditions for NS1 capture.

A-Various blocking agents and two distinct ELISA microplates were evaluated. The P/N ratio for NS1-2 capture indicated that 1% BSA, combined with Greiner microplates, yielded the highest P/N ratio. B-The ability of the Nbs under study to capture the NS1-2 protein was assessed using two different concentrations (10 and 20 μg/mL). Detection was performed using an anti-Flag HRP antibody. Nanobody 14 and Nb48 exhibited a stronger capture capacity compared to other tested Nbs.

https://doi.org/10.1371/journal.pntd.0013168.g001

Development of capture and detection Nb-based tools for dengue diagnosis

To address the high quantity of Nb required for effective capture and the critical role of proper orientation in improving detection, Nb14 was engineered to enhance capture efficiency while reducing the necessary amount of protein. Key modifications included the incorporation of a polystyrene-binding sequence (Nb-PSW) to improve microplate immobilization, the construction of a bivalent Nb to enhance avidity, and the fusion of the Fc domain (Nb Fc) to combine the benefits of enhanced avidity and improved immobilization through Fc-mediated orientation on plastic surfaces. Additionally, biotinylated Nbs and Fc-HRP fusions of Nb48 and Nb66 were developed for Nb-based detection tools.

NS1 capture efficiency was assessed after serial dilutions of each engineered Nb. The capturing Nb14 Fc showed a 63-fold increase in ELISA compared to the monovalent Nb14, with the lowest EC₅₀, confirming its binding capacity for NS1. The bivalent Nb also showed significantly improved capture capacity compared to the monovalent formats. In contrast, the monovalent Nb and Nb PSW displayed similar titration curves and EC₅₀ values, indicating that the addition of the polystyrene-binding sequence did not significantly improve capture efficiency (Fig 2, Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. EC₅₀ values of modified Nb14.

https://doi.org/10.1371/journal.pntd.0013168.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Comparison of the NS1 capture capacity of modified Nb14.

A-Titration curves of unmodified Nb14, Nb14 PSW, bivalent Nb14, and Nb14 Fc. Absorbance at 450 nm was plotted as a function of Nb concentration. B-Analysis of log EC₅₀ for each Nb14 format, indicating significant differences in capture capacity. Letters (a, b, c) indicate groups with significant differences (one-way ANOVA, Tukey’s multiple comparison test, p < 0.01).

https://doi.org/10.1371/journal.pntd.0013168.g002

Additionally, due to scale-up challenges, the PSR-modified Nbs were not tested in our assay. While several strategies exist to address low expression yields—such as fusion to Maltose Binding Protein or protocols tailored for recalcitrant Nbs—we opted not to pursue yield optimization [62–64]. This decision was based on the lack of significant improvement in capture performance following the addition of the PSW polystyrene-binding peptide.

Identification of the optimal Nanobody pair for dengue diagnosis

Nanobody 14 Fc was designated as the capture antibody for the Nb-ELISA development, and a chessboard titration was conducted to optimize both capture and detection antibody concentrations. Nanobody 48 and Nb66, representing two distinct Nb families, based on phylogenetic and germline analyses, were used as detection antibodies in Fc-HRP fusion (S8 Fig) and biotinylated (S9 Fig) formats.

Optimization applying HRP-fused Nbs revealed that lower concentrations of Nb14 Fc as the capture antibody improved signal strength, likely by increasing availability for NS1 binding and reducing surface saturation (S8 Fig). For Nb14 Fc/Nb48 HRP pair, the highest P/N ratio was achieved with 4 µg/mL capture antibody and 10 µg/mL detection antibody (S8B Fig). Similarly, for Nb14 Fc/Nb66 HRP, the optimal signal was observed with 4 µg/mL capture antibody and 20 µg/mL detection antibody, with a P/N ratio close to 6 (S8D Fig). Based on these results, the first system for NS1 detection, named system 1, was proposed: 4 µg/mL Nb14 Fc as capture antibody, 10 µg/mL Nb48 HRP as detection antibody, and 1% BSA as the blocking reagent.

Two biotinylation strategies were employed to modify detection antibodies. The biotinylation efficiency using the sortase A system was significantly higher than that of the AviTag system (S10 Fig). Therefore, Nbs labeled with biotin via the sortase A system were used in the subsequent determinations. When utilizing these modified Nbs, a trend observed with HRP-conjugated Nbs emerged, with higher absorbance values achieved at even lower Nb14 Fc concentration (2 µg/mL; S9 Fig). The highest P/N ratio for the Nb14 Fc/Nb48 biotin pair was obtained using 2 µg/mL capture antibody and 10 µg/mL detection antibody (S9B Fig). For Nb14 Fc/Nb66 biotin pair, the optimal combination was 2 µg/mL capture antibody and 5 µg/mL detection antibody (S9D Fig). These results demonstrate that biotinylated Nbs enhance signal amplification and improve the sensitivity of the sandwich ELISA. Therefore, another system, named system 2, was proposed: 2 µg/mLNb14 Fc as the capture antibody, 5 µg/mL Nb66 biotin as the detection antibody, and 1% BSA as the blocking reagent.

Evaluation of NS1 Detection Systems in a Serum Matrix

The two optimized systems were tested for their ability to detect recombinant DENV and ZIKV NS1 proteins in a serum matrix (serum from a DENV NS1 negative patient) that does not exhibit reactivity towards these proteins. A concentration of 0.5 µg/mL, within the reported range for NS1 in the serum of DENV-infected patients, was used [18,19]. Additionally, the detection of NS1 in pools of DENV-positive and -negative sera was explored to evaluate the clinical applicability of the systems. Both systems successfully detected recombinant NS1 proteins from DENV serotypes 1 to 4, with varying absorbance values (Fig 3). High sensitivity was observed for NS1-1 and NS1-2, while detection of NS1-3 and NS1-4 was less efficient. No cross-reactivity with ZIKV NS1 was detected, confirming the specificity of both systems for distinguishing DENV infections from ZIKV. However, neither system was able to detect the NS1 protein present in patient sera.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Detection of recombinant DENV and ZIKV NS1 proteins in serum matrix, and pools of DENV-positive and -negative sera using the proposed systems.

A-Schematic representation of the two proposed Nanobody-based ELISA systems. System 1: 4 µg/mL of Nb14 Fc as the capture antibody and 10 μg/mL Nb48 HRP as the detection antibody. System 2: 2 µg/mL of Nb14 Fc as the capture antibody and 5 μg/mL biotinylated Nb66 as the detection antibody. B-Reactivity of System 1 with NS1 variants from DENV serotypes 1–4, ZIKV NS1, and human sera. C-Reactivity of System 2 with the same antigen panel. Data is shown as mean absorbance at 450 nm ± SD (n=3). Black dashed line represents the mean OD of negative control samples plus 3 standard deviations.

https://doi.org/10.1371/journal.pntd.0013168.g003

The failure to detect NS1 in positive serum samples may be attributable to immune complex formation between NS1 and anti-NS1 antibodies. This interaction potentially masks NS1 epitopes, thereby preventing recognition by detection Nbs. Various treatments, including heat and pH modifications using acidic and alkaline buffers, were employed to dissociate these immune complexes and destabilize antigen-antibody interactions [65–67]. Despite these efforts, no improvement in NS1 detection was observed in the treated samples. Nanobodies from the initial panel, assessed in both their monovalent and Fc-fused formats, were evaluated for their ability to capture NS1 in clinical serum samples. Under the assay conditions employed, none of the candidates exhibited detectable NS1 capture (S11 Fig).

Development of a combined Nanobody ELISA for dengue diagnosis

Polyclonal anti-NS1 serum was produced in a rabbit following a five-dose immunization protocol. Indirect ELISA results demonstrated strong binding of the purified IgG to recombinant NS1 proteins from all four DENV serotypes, with absorbance values increasing in a dose-dependent manner (S12 Fig). Endpoint titration showed that a minimum IgG concentration of 0.2 µg/mL was sufficient to detect NS1 from all DENV serotypes. Mild cross-reactivity with ZIKV NS1 was observed, likely due to the polyclonal nature of the serum, but this was limited to high antibody concentrations and was significantly weaker than binding to DENV NS1 (S12 Fig).

The purified polyclonal IgG, combined with biotinylated Nb66 as a detection antibody, was employed to develop a cNb-ELISA for DENV NS1 (Fig 4A ). Nanobody 66 was selected for its cost-effectiveness, reduced usage concentration, and strong performance in Nb-based ELISA. The ability of purified IgG to capture recombinant NS1 from DENV serotypes 1–4 and ZIKV was assessed in a serum matrix devoid of anti-NS1 reactivity. At an antigen concentration of 0.5 µg/mL, the IgG successfully captured NS1-1 and NS1-2 with high sensitivity, while NS1-3 and NS1-4 exhibited lower absorbance values (Fig 4B). No cross-reactivity with ZIKV NS1 was detected when the purified IgG was used as the capture antibody, confirming the system’s specificity for distinguishing DENV infections from ZIKV. To further evaluate the specificity of our detection system for dengue virus NS1 protein, we tested supernatants from Vero cells infected with DENV serotypes 1–4, ZIKV, CHIKV, and YFV. The selected Nbs showed no cross-reactivity with other flaviviruses, including YFV and Zika virus, nor with the alphavirus CHIKV, which presents clinical symptoms that closely resemble those of dengue (Fig 4D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Evaluation of the combined Nanobody ELISA for the detection of DENV NS1.

A-Schematic representation of the combined Nanobody ELISA workflow. B-Detection of recombinant NS1 proteins from DENV and ZIKV in a negative serum matrix. C-Detection of NS1 in pooled human serum samples from DENV-positive and negative patients. Serum dilutions (1:5 and 1:10) were analyzed using biotinylated Nb66 as detector. D-Specificity assessment of the combined Nb ELISA using supernatants from Vero cells infected with different flaviviruses (DENV-1 to DENV-4, ZIKV, YFV) and the unrelated arbovirus CHIKV, which co-circulate in endemic regions and cause overlapping clinical symptoms. Mock-infected supernatant was used as a negative control.

https://doi.org/10.1371/journal.pntd.0013168.g004

The optimal concentrations of capture and detection antibodies for the cNb-ELISA were determined using a previously described chessboard titration method. Based on these findings, 10 µg/mL of purified polyclonal anti-NS1 serum and 5 µg/mL of biotinylated Nb66 were selected for the assay. In contrast to the previously developed Nb-only systems, the cNb-ELISA effectively detected NS1 in a pool of positive serum samples (Fig 4C). At both tested dilutions (1:5 and 1:10), the assay produced a clear and distinguishable signal for positive samples compared to negatives, underscoring its robustness and sensitivity. Its demonstrated ability to detect NS1 in clinical samples highlights the potential of this cNb-ELISA as a diagnostic tool, successfully addressing the limitations of earlier systems.

Validation of the combined Nanobody ELISA

To evaluate the performance of the cNb-ELISA, serum samples were collected from individuals presenting symptoms consistent with DENV infection. A flow diagram of the validation process for the cNb-ELISA is shown as supplementary information (S13 Fig). These samples were categorized at Orán Hospital as either NS1-positive (n=94, confirmed using PLATELIA DENGUE NS1 Ag from BioRad) or epidemiologically negative (n=74, not tested with by a commercial kit due to being collected 5 or more days after the acute phase). The ROC curve analysis yielded an area under the curve (AUC) of 0.9408 with a 95% confidence interval (CI) of 0.9078 to 0.9730. This indicates excellent discriminative capacity (AUC > 0.9; Fig 5A). Using the Youden index and the MaxSpSe method, a cutoff absorbance value of 0.126 at 450 nm was established. At this threshold, the diagnostic sensitivity was 93.6% (95% CI: 86.6–97.6%) and the diagnostic specificity was 81.1% (95% CI: 70.3–89.3%). Samples with absorbance values below 0.126 were classified as negative, while those exceeding this threshold were classified as positive (Fig 5B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Performance of the developed cNb-ELISA.

A-Receiver operating characteristic (ROC) curve analysis indicating the calculated AUC for the validation of the assay. B-Interactive dot plot showing the absorbance value of each serum sample. Dot line shows the determined cut-off point.

https://doi.org/10.1371/journal.pntd.0013168.g005

Agreement of the developed cNb-ELISA with results obtained after the application of the official algorithm executed at Orán Hospital (NS1-positive samples analyzed using the Bio-Rad test for specimens collected up to day five, along with epidemiologically negative samples) was calculated. Kappa calculated (K=0.75) using data from Table 3 indicates a substantial agreement according to the standard interpretation by Landis and Koch [68].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Concordance Analysis between the cNB-ELISA test and the results from the algorithm executed at Orán Hospital.

https://doi.org/10.1371/journal.pntd.0013168.t003

Regarding discordant results (S2 Table), 6 samples initially identified as positive were negative by the cNb-NS1 ELISA, while 14 samples identified as negative tested positive. To address this discrepancy, the samples were re-evaluated using the commercial ELISA Dengue virus NS1 antigen (DIA.PRO). Six samples undetected by our cNb-ELISA were confirmed as positive by the DIA.PRO results, highlighting a limitation in our assay. However, five positive samples were negative both for the cNb-ELISA and the DIA.PRO ELISA. It is likely that these positive results were inaccurate, as the unusually low cutoff reported for these samples using the Bio-Rad test stands out when compared to the cutoffs of the other samples measured with the same kit (S3 Table). Therefore, these group of samples were not included in this analysis.

Orán Hospital adheres to an “algorithm for the differential diagnosis of febrile syndrome in its initial phase”, established by the health authority, to analyze human samples. According to this algorithm, samples collected five days after symptom onset are not tested but can be classified as epidemiologically positive or negative. Fourteen samples epidemiologically negative from this group tested positive with our kit, and the results were corroborated by the DIA.PRO test (S2 and S3 Tables). This highlights the critical role of antigen detection in diagnosing dengue infection while emphasizing the effectiveness of our system in identifying NS1 in samples collected over five days after symptom onset.

Analytical sensitivity and limit of detection of the combined Nanobody ELISA

The analytical sensitivity of the cNb-ELISA was evaluated using recombinant NS1 protein spiked into human serum (Fig 6). Standard curves were performed for each serotype by using NS1 at different concentrations and data was fitted to a 4-parameter logistic model. The LOD values were determined to be 101.80 ng/mL, 133.68 ng/mL, 507.70 ng/mL, and 425.58 ng/mL for NS1-1, NS1-2, NS1-3, and NS1-4, respectively (Fig 6A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Determination of the limit of detection and the analytical sensibility.

A-ELISA to detect DENV1-4 NS1 at different concentrations to determine the limit of detection of the optimized cNb-ELISA. The developed detection system has variable LOD for the different DENV NS1. B-Estimation of the analytical sensitivity of the cNb-ELISA, absorbance values obtained based on 2-fold serial dilutions of the pool of positive sera in a range from 1/2 to 1/256. Inset: linear regression analysis of absorbance values based on antigen concentration.

https://doi.org/10.1371/journal.pntd.0013168.g006

Absorbance values above the cutoff point were obtained up to a 1/256 dilution of the pool of positive sera form infected patients, confirming the high sensitivity of the system. In contrast, the negative sera used as controls did not show absorbance values above the proposed cutoff point at any of the dilutions evaluated. From this assay, the linear range of the ELISA was determined, defined by the dilutions of the sera pool in which the absorbance is proportional to the antigen concentration. Within this range, a significant linear relationship was obtained, with a determination coefficient (R²) of 0.997, indicating an excellent correlation between the sera pool dilutions and the detected signal (Fig 6B). These results demonstrate the robustness of the developed system, as it can maintain high sensitivity and reproducibility over a wide range of dilutions of the positive sera pool.