A New Approach in DENV Diagnostic based on Linear Peptides Obtained by Phage Display

Dengue is a viral disease caused by any of the four distinct dengue virus (DENV) serotypes that circulate in many parts of the world. DENV now co-circulates with Zika and Chikungunya viruses (ZIKV and CHIKV) in many regions of the Americas. Having this in mind, plus the fact that DENV clinical diagnosis persists as a dicult task, due to the similarity in symptoms, as well as false-positive results by the cross-reactivity of the IgG and IgM against these three viruses, correct identication of DENV at an early stage of the disease is essential to minimise transmission and prevent potentially devastating sequelae. Here, by phage display, we isolate specic peptides for dengue virus NS1 protein. The specicity of the linear peptides as diagnostic tools for DENV NS1 protein in sera samples was investigated, and the selected peptides showed the ability to recognise DENV, and no cross-reactivity was shown. Moreover, in silico analysis was performed to assess the possible binding modes of these peptides to DENV-NS1, using a molecular docking approach. These peptides are suitable for use in an ELISA assay for dengue virus detection in human serum and the possibility to adapt these peptides to PoC platforms. and 6) Unusual Dengue (UD) 4 . Severe cases of DENV cause DHF or DSS (generally associated with a secondary DENV infection), resulting in approximately 22,000 deaths per year globally, mainly affecting children 5 . The characteristic of the infection course presents three stages: 1) febrile (1-7 days from symptom onset), 2) critical (from days 8-9), and 3) recovery (after day 10). The symptoms developed are fever, headache, retro-ocular pain, myalgia, arthralgia, dehydration, vomiting, lethargy, abdominal pain, loss of appetite (anorexia), and skin eruptions (rash) 6,7 . These results suggest that the phage clones NS1P3, NS1P6, and NS1P8 are specic for rNS1. The batch production of these clones was used to evaluate the recognition ability of viruses present in serum samples. The remaining three phage clones (NS1P4, NS1P9, NS1P14) were not specic for rNS1 protein (Figure 2B, Figure 2E, Figure 2F). These phage clones were discarded for this study and stored for further evaluations. The DENV serotype specicity using the peptide-display phages (NS1P3, NS1P6, and NS1P8) was evaluated in an ELISA sandwich, using sera samples of the four serotypes of DENV. The results revealed the binding anity of the peptide-display phages against the four serotypes of DENV, where the phage clone NS1P3 (Figure 3A) and NS1P6 (Figure 3B) shows a statistically signicant difference (p <0.0001, p <0.0001 and, p=0.0208, respectively) compared to the negative serum control. The same result is found in recognition of the DENV-2, DENV-3, and DENV-4 serotypes, respectively. However, in the NS1P8 (Figure 3C), no statistically signicant difference (p<0.05) was found to recognise DENV-3, and DENV-4 serotypes present in serum, compared to the negative serum control. These results indicated that NS1P3, NS1P6, and NS1P8 showed DENV recognition in general, but they do not show serotype specicity. However, the recognition of the DENV present in serum is relevant, suggesting all three peptide-display phages were reactive to the hexameric isoform of the NS1 protein. the linear peptide can be evaluated for the same and optical density (OD) was measured at 450 nm (measurement wavelength) using the microplate reader (BioRad xMark™ Microplate Absorbance Spectrophotometer). Signicance (p value) determined for rNS1 recognition against BSA were assessed appropriate arithmetic averaging. Detection of rNS1 with a p-value below 0.05 were considered differentially recognised by peptide-display phages. 100 µL of PBS-T. The detection was then performed with 50 µL of the anti-ZIKV (Fitzgerald, 10-2714) and anti-CHIKV (Fitzgerald, 10-2717) antibodies at 1 µg/mL and anti-NS1 (R&D Systems, MAB94441) at 2 µg/mL respectively, incubating 2 h at room temperature. Three washes were performed and 50 µL of mouse anti-IgG antibody, conjugated with HRP in 1: 10,000 dilution (Abcam, Ab97023) were added and incubated for 1 h at room temperature. After washing as described above, 50 µL of 1-Step Ultra TMB substrate (Thermo Scientic, 34028) were added. The reaction was stopped with 0.5 M sulfuric acid, and absorbance was measured at 450 nm in a microplate reader.


Introduction
Dengue virus (DENV) is endemic in more than 100 countries in Asia, America, Africa, and the Caribbean 1 , in such a way that the World Health Organization (WHO) has classi ed it as one of the arthropod-borne viruses (arboviruses) with the highest prevalence in the world 2 . Previous reports indicated that there are four different serotypes of this virus (DENV 1-4). These viruses belong to the avivirus genus, which also includes Zika (ZIKV), Yellow Fever (YF), and West Nile viruses (WNV) 3 . Infection caused by DENV can be asymptomatic; however, the symptoms classi cation is: 1) Undifferentiated Fever (UF), 2) Dengue Fever (DF), 3) Dengue Haemorrhagic Fever (DHF), 4) Dengue Shock Syndrome (DSS), 5) Expanded Dengue Syndrome (EDS) and 6) Unusual Dengue (UD) 4 . Severe cases of DENV cause DHF or DSS (generally associated with a secondary DENV infection), resulting in approximately 22,000 deaths per year globally, mainly affecting children 5 . The characteristic of the infection course presents three stages: 1) febrile (1-7 days from symptom onset), 2) critical (from days [8][9], and 3) recovery (after day 10). The symptoms developed are fever, headache, retro-ocular pain, myalgia, arthralgia, dehydration, vomiting, lethargy, abdominal pain, loss of appetite (anorexia), and skin eruptions (rash) 6,7 .
Likewise, the presence of leukopenia and thrombocytopenia has been widely reported 8 . Due to its complex nature and the cocirculation of ZIKV or chikungunya virus (CHIKV) in the same endemic area, it is not easy to obtain the correct diagnosis. Different viral markers can be detected at different stages of infection. However, early identi cation of the virus is pivotal in disease management. Usually, DENV diagnosis considers the clinical symptoms which are con rmed through laboratory analysis. The diagnostic method to be applied depends on the stage of infection; during symptoms, direct methods are carried out: viral isolation, detection of the viral genome (RT-PCR) or viral antigens (NS1 mainly), or a combination of these. Non-structural protein 1 (NS1) is an early marker of infection and its detection in serum matches with detectable viremia [9][10][11] . In the critical stage of recovery, serological tests are implemented for the detection of IgM and IgG 12 . Several research groups have developed different diagnostic methods, (Table 1) with a tendency for Point of Care (PoC), using different biological matrices at the early stage of the infection.  16,17 or single-chain fragments variable (scFv) 18 . However, an important factor is the time and cost of large-scale production and puri cation processes of these antibodies. Another approach uses peptide-display phages as an alternative for the detection of viral diseases 19 . Peptide-display phages present some advantages: can be easily synthesised and their use can reduce the complexity of the development of platforms such as lateral ow assays or biosensor approaches, due to the ease coupled to nanoparticles or biomolecules 20,21 . The improvement of the DENV detection test, especially in the early stage of illness, it is essential to generate peptides or antibodies with no cross-reaction with other arboviruses. The avivirus NS1 protein is a useful infection marker because its accumulation in dengue patients' sera can be found at concentrations up to 50 µg/ml in the febrile stage of illness 22 . The work presented herein aimed to isolate and characterise linear peptides that bind to the NS1 protein of DENV without crossreactivity with ZIKV and CHIKV using serum samples. As proof of concept, we evaluated an ELISA for NS1 detection in the illness's febrile stage. As a perspective, it can be used for the development of a PoC.
A phage library displaying linear peptides of 20 amino acids was challenged for binding to DENV-NS1 protein.
In the rst round of selection, we obtained 2.0x10 6 phage particles with a title increase in the second and third rounds with 4.0x10 9 and 5.0x10 9 , respectively ( Figure 1). After three selection rounds, fteen phage clones were selected for sequencing. Plasmids containing no sequence or repeated sequences were discarded. The sequences of the selected peptides are shown in Table 2. A total of six phages displaying linear peptides were isolated, and ELISA results suggested that three of these linear peptides clones denominated NS1P3, NS1P6, and NS1P8 ( Figure 2A, Figure 2C, Figure 2D) were speci c against DENV-rNS1 as compared to BSA (negative control) with a statistically signi cant difference (p=0.0002, p=0.0002, and p<0.0001, respectively). The commercial antibody used as a positive control had no difference between recognizing rNS1 protein and BSA.
These results suggest that the phage clones NS1P3, NS1P6, and NS1P8 are speci c for rNS1. The batch production of these clones was used to evaluate the recognition ability of viruses present in serum samples. The remaining three phage clones (NS1P4, NS1P9, NS1P14) were not speci c for rNS1 protein ( Figure 2B, Figure 2E, Figure 2F). These phage clones were discarded for this study and stored for further evaluations.
-Linear peptides can bind to NS1 proteins from all four DENV serotype in serum.
The DENV serotype speci city using the peptide-display phages (NS1P3, NS1P6, and NS1P8) was evaluated in an ELISA sandwich, using sera samples of the four serotypes of DENV. The results revealed the binding a nity of the peptide-display phages against the four serotypes of DENV, where the phage clone NS1P3 ( Figure 3A) and NS1P6 ( Figure 3B) shows a statistically signi cant difference (p <0.0001, p <0.0001 and, p=0.0208, respectively) compared to the negative serum control. The same result is found in recognition of the DENV-2, DENV-3, and DENV-4 serotypes, respectively. However, in the NS1P8 ( Figure 3C), no statistically signi cant difference (p<0.05) was found to recognise DENV-3, and DENV-4 serotypes present in serum, compared to the negative serum control. These results indicated that NS1P3, NS1P6, and NS1P8 showed DENV recognition in general, but they do not show serotype speci city. However, the recognition of the DENV present in serum is relevant, suggesting all three peptide-display phages were reactive to the hexameric isoform of the NS1 protein.
-Linear peptides anti-NS1 have no cross-reactivity with other arboviruses.
Since DENV is phylogenetically related to ZIKV, NS1 protein shares high homology in amino acid sequences; therefore, the selected peptidedisplay phages were tested to recognise native ZIKV NS1 by ELISA assay, using serum samples. We also tested our phage clones against CHIKV serum samples to assure no cross-reactivity. Phage clones NS1P3 ( Figure 4A), NS1P6 ( Figure 4B), and NS1P8 ( Figure 4C) have a statistically signi cant difference (p values are shown on each graph) among the recognition of the DENV present in serum samples versus ZIKV or CHIKV present in positive serums. The result suggests that NS1P3, NS1P6, and NS1P8 did not cross-react with another phylogenetically related arbovirus, such as ZIKV and a non-related arbovirus, like CHIKV and, these peptide-display phages can be used in a detection platform for the recognition of DENV in complex samples.

Discussion
The linear peptides speci c for DENV-NS1 were selected by phage display using rNS1 protein-coated as the capture antigen. During panning progression against NS1 protein, the amounts of eluted phages increase stepwise as expected (Figure 1). From the panning experiment, we isolated three individual phage clones from the third round of selection that were designated as NS1P3, NS1P6, and NS1P8. All these three phage clones bind to rNS1. The phage display technique has been extensively used by several authors to search for interacting peptides against different targets, for example, de la Guardia et al. (2017)  detection; they analysed commercial kits designed to detect DENV-NS1 27 . Their nding revealed that 33% of these commercial kits reacted with ZIKV-NS1 present in supernatants of cell culture. It is well-known that arboviruses may cross-react in the immunodiagnostic tests due to their phylogenetic relationship and antigenic similarities 28 . However, it can be noted that NS1P3, NS1P6, and NS1P8 speci cally bound to DENV-NS1 in the ELISA sandwich (Figure 4), and not to other arboviruses (ZIKV, CHIKV), which represents an advantage over other publications based on the use of mAbs for the detection of NS1, such as Rocha et al. (2017) who generated four monoclonal antibodies against the NS1 of dengue virus serotype 2, and two of them did cross-react to ZIKV 29 . Thus, these preliminary results indicate that the phage-displayed peptides, like NS1P3, NS1P6, and NS1P8, could be particularly useful for developing a PoC test that minimizes the risks associated with false-positive results among ZIKV and/or CHIKV-infected subjects. As a perspective, the linear peptide can be evaluated for the same purpose.
Analysis of the contact map indicates that our novel peptides can bind to similar regions of DENV-NS1 located between 205 and 336 residues.
In this sense, a sandwich of two of these peptides (NS1P6-NS1P8) for DENV-NS1 detection cannot be an option.  2020), where DENV-NS1 was recommended and explored as a molecular target for reducing infectious virus production in cell culture; several peptides of 12 amino acids were able to bind to DENV-NS1 and shows a decrease in the DENV infectious production, making the NS1 protein a promising molecular target for drug design 31 .
Many research groups have reported the use of different formats of antibodies as mAb, VHH, scFv, and Fab for the detection of one or more serotypes of DENV 3,15,26,32-34 however, they inevitably resort to the production of antibodies by recombinant expression in cellular systems, increasing the cost of the production and downstream procedures, as puri cation and refolding if necessary. Instead, our peptides are a cheaper alternative and easy to synthesise. Our results provide the proof of concept for developing a diagnostic test using peptide-display phages as elements for the capture of DENV-NS1. Nevertheless, an essential limitation of this study is that NS1 concentration in the serum samples is not known; therefore, it requires additional validation. Future evaluation of the established assays will involve more massive sets of a case-control study.
In conclusion, we generated three peptide-displaying phages against the DENV-NS1 protein. Our three peptide-displaying phages named NS1P3, NS1P6, and NS1P8 recognised all serotypes of DENV by ELISA and did not cross-react to ZIKV and CHIKV in serum samples. Thus, the selection and characterisation of the peptides, the prediction of potential epitopes, and the speci city of anti-DENV NS1 peptides may contribute to the development of diagnostic tools as LFA, ELISA, peptide-based biosensors or, another immunoassay formats able to differentiate DENV, ZIKV, or CHIKV infections. Due to the current situation caused by SARS-Cov2, it is crucial to consider this virus. Although it is not phylogenetically related to DENV, ZIKV, and CHIKV, it can also cause similar symptoms in the disease's initial stages. In this sense, it would be interesting to develop a PoC test to detect and differentiate among all these viruses simultaneously.

Methods
-Bio panning from peptide library.
The TriCo-20 peptide library (Creative Biolabs) was panned against recombinant protein NS1 (rNS1). The bio panning was done using procedures previously described by the manufacturer. Three rounds of subtracted panning were performed. The rNS1 (Fitzgerald, 80R-4280) was adsorbed onto a 96-well plate at 2.5 µg/mL; next, wells were blocked with blocking buffer (0.1 M NaHCO 3 (pH 8.6), 5 mg/mL BSA, 0.02 % NaN 3 , 0.1 μg/mL streptavidin). The library of phages displaying linear peptides (2x10 11 PFU) in 100 µL of blocking buffer, were added to the blocked wells where no rNS1 was adsorbed, and incubated for 1 h at 37 °C. Next the non-bound phages were transferred to the wells where rNS1 was immobilized, and incubated for 1 h at 37 °C. Subsequently, 10 washes were performed using PBS with 0.05% Tween 20 (PBS-T). The bound phages were eluted from the plate and used to infect E. coli ER2738 cells for 15 min at room temperature without shaking. These infected cells were used to amplify the selected phages. The ampli ed phages were used in the next round of panning, and the titration of the precipitated phages was performed.
-Screening by sequencing. -Reactivity of phage-displayed peptides to different serotypes of Dengue virus.
An ELISA assay for assessing the binding speci city of NS1P3, NS1P6, and NS1P8, against the four DENV serotypes was carried out. RT-PCR validated the positivity of human sera as described before. Three biological replicates were used in this assay. The plate was sensitized with performed with 100 µL of 3 % BSA for one hour at 37 °C. Then 50 µL of positive serum for each DENV serotype was added at 1: 400 dilution in PBS and, healthy serum was used as a negative control. The plate was incubated for two hours at room temperature. Three washes were carried out with 100 µL of PBS-T, with constant agitation for 2 min in each wash. Detection was performed with 50 µL of anti-NS1 Ab (R&D Systems, MAB94441) at 2 µg/mL, respectively, incubating 2 h at room temperature. Three more washes were carried out, and 50 µL of antimouse IgG conjugated with HRP in dilution 1:10,000 (Abcam, Ab97023) were added, incubated for one hour at room temperature. Three nal washes steps were performed, and 50 µL of 1-Step Ultra TMB substrate (Thermo Scienti c, 34028) was added. The absorbance signal was measured at 450 nm after stopping the reaction with 50 µL of 0.5 M sulphuric acid.
-Speci city of selected anti-dengue peptides.
The aim of this study was to isolate linear peptides that bind to the NS1 protein of DENV. For this purpose, an ELISA sandwich assay was designed based on anti-DENV NS1 peptide-display phages NS1P3, NS1P6, and NS1P8 and commercial mAbs, acting as capture and detection elements, respectively. A 96-well plate was coated with the phages NS1P3, NSP6, and NSP8, and incubated overnight at 4 °C. Blocking was performed with 100 µL of 3% BSA for one hour at 37 °C, then 50 µL of each positive serum of DENV, ZIKV, CHIKV, were added and a healthy serum was used as negative control, in 1: 400 dilution in PBS 1X and incubated for 2 h at room temperature. Three washes were performed with 100 µL of PBS-T. The detection was then performed with 50 µL of the anti-ZIKV (Fitzgerald, 10-2714) and anti-CHIKV (Fitzgerald,  antibodies at 1 µg/mL and anti-NS1 (R&D Systems, MAB94441) at 2 µg/mL respectively, incubating 2 h at room temperature. Three washes were performed and 50 µL of mouse anti-IgG antibody, conjugated with HRP in 1: 10,000 dilution (Abcam, Ab97023) were added and incubated for 1 h at room temperature. After washing as described above, 50 µL of 1-Step Ultra TMB substrate (Thermo Scienti c, 34028) were added.
The reaction was stopped with 0.5 M sulfuric acid, and absorbance was measured at 450 nm in a microplate reader.
The binding of peptides to DENV-NS1 was explored by molecular docking using two different algorithms implemented by GalexyPepDock and CABS Dock server. GalaxyPepDock is a protein-peptide docking tool based on interaction similarity and energy optimization, using Python 37 to input the peptide sequence and protein ID PDB: 4OIG (DENV-NS1). Default settings were selected. Furthermore, CABS-Dock was used to con rm the analysis. This server is based on a global docking procedure that includes explicit, fully exible docking simulation and then clustering-based scoring. We selected the CABS-dock simulation runs 10,000 alternative models of the complex, models were ranked and numbered according to their occurrence in docking trajectory (1= most probable result), and ten top-ranked models were reconstructed to allatom representation 38 . The best model was selected for each peptide´s complex, and pairs of protein/peptide residues closer than 3 Å were evaluated.
The differences in recognizing between DENV rNS1 protein and BSA used as a negative control by each peptide-display phage were assessed by one-way ANOVA followed by Turkey's multiple comparison test, the detection of rNS1 with a p-value below 0.001 was considered with a signi cant statistical difference in recognition.
One-way ANOVA was used followed by a Dunnett´s test to compare the recognition of each serotype and healthy serum, the signi cance (pvalue) determined by each DENV serotype was assessed with appropriate arithmetic averaging. Detection of DENV serotype with a p-value below 0.001 was considered differentially recognised for each phage clone. Also, to compare the recognition of the different DENV serotypes a Turkey´s test was used, and detection of DENV serotype with a p-value below 0.05 was considered with a signi cant statistical difference in recognition for each phage clone.
Reactivity of isolated phage clones was assessed against DENV, ZIKV, and CHIKV in serum. To compare ZIKV and CHIKV recognition versus DENV recognition a One-Way ANOVA was performed followed by Turkey's multiple comparison test, where the detection of DENV serotype with a p-value below 0.01 was considered with a signi cant statistical difference. GraphPad Prism software (version 5.03) was used for performing statistical analyses. The number of replicates per experiment is indicated in each gure legend.

Declarations
Competing interests: The authors declare no competing interests.