Cross-binding antibodies capable of neutralising diverse hantaviruses are produced in response to Puumala virus infection

Summary

Background

Orthohantaviruses (hantaviruses) are emerging rodent-borne pathogens that can cause severe human disease. They are present on multiple continents and are responsible for thousands of human cases per year. Despite this, no licenced therapeutics are available, vaccines for most strains are lacking, and the immunological response to infection is poorly characterised. This study aimed to analyse the humoral immune response to Puumala virus (PUUV) infection to inform future studies focussing on the production of therapeutic monoclonal antibodies and vaccination strategies.

Methods

Serum was obtained from a cohort of 24 patients hospitalised with PUUV infection at four time points, covering the early acute, late acute, early convalescent, and late convalescent stages of the disease. The humoral immune responses at each time point were quantified, and cross-binding, cross-neutralising antibody responses were investigated. Serum cytokine levels were also interrogated, and expression was correlated with humoral outputs.

Findings

PUUV infection elicited a robust anti-PUUV neutralising antibody response. However, cross-reactive antibodies that were capable of binding diverse hantaviruses were also induced in late convalescence. Modulations in the abundance of IgG subclasses were evident following infection, with significant differences present months after infection.

Interpretation

This study demonstrates that broadly reactive anti-hantavirus antibodies are produced in response to Old-World hantavirus infection, but predominantly months after recovery. As this is concomitant with changes in IgG subtypes, our results suggest that PUUV infection promotes prolonged class-switching and somatic hypermutation, favouring conserved epitopes long after exposure.

Funding

Work at Mount Sinai was supported by Institutional Funds, work at the Medical University of Vienna was supported by Institutional Funds. The study was supported in part by the Styrian government, Austria (project no. ABT12-106729/2022-13) and the Austrian Science Fund (FWF) (number J 4737-B).

Research in context.

Evidence before this study

We searched PubMed for articles published up to January 1, 2025, using the terms (“Puumala virus” OR “PUUV”) AND (“antibody”) AND (“cytokine”), including only English-language publications. Puumala virus (PUUV) is endemic in central Europe, particularly Germany and Finland, causing hundreds of cases annually. Although case fatality rates are low (∼0.1%), infection can result in hospitalisation and long-term sequelae. While PUUV elicits a strong IgG response, it has not been thoroughly investigated. This study examines humoral responses to PUUV infection using a cohort of patients from which longitudinal serum samples were taken covering the early acute, late acute, early convalescent, and late convalescent stages of disease. We also analysed serum cytokine levels and compared moderate and severe disease cases.

Added value of this study

Our study shows that PUUV infection triggers a potent, neutralising IgG response early, which remains stable for six months. Broadly neutralising antibodies, effective against multiple orthohantaviruses, emerge mainly in late convalescence, six months post-infection. IgG subtypes showed changes over time, with IgG1 increasing steadily, IgG3 decreasing, and IgG4 rising at six months. These shifts were associated with reduced antibody effector function. Severe PUUV infection correlated with increased serum levels of Th2/Th17-associated cytokines (IL-23, IL-22, IL-4, IL-13, IL-5), suggesting immune dysregulation.

Implications of all the available evidence

Our findings suggest that affinity maturation still occurs six months post-infection, favouring conserved epitopes among different orthohantaviruses. The rise in broadly neutralising antibodies coincided with IgG class switching, implying continued antigen presentation months after infection, possibly due to persistent infection in immune-privileged tissues. These results are relevant for future monoclonal antibody research from disease survivors. Additionally, Th2/Th17 cytokines may serve as biomarkers for severe PUUV infection.

Introduction

Orthohantaviruses (Hantaviruses, family: Hantaviridae, order: Bunyavirales) are rodent-borne, negative-sense, segmented RNA viruses that are found worldwide. These viruses are endemic in every continent except Antarctica and can be geographically and phylogenetically divided into New-World and Old-World hantaviruses. Hantaviruses naturally infect rodents, shrews, moles, and bats, resulting in avirulent, presumably lifelong infections. While hantaviruses were once thought to exhibit strict reservoir host restriction, recent evidence suggests that some hantaviruses may productively infect multiple host species.1, 2, 3, 4 Although these reservoir hosts do not exhibit disease signs, they continually shed the virus in their urine, faeces, and saliva. These excreta can then become desiccated and, when disturbed, can be inhaled by humans or directly inoculated into broken skin, mucous membranes, or eyes. When infecting humans, viruses harboured by Old-World mouse and rat species (Muridae), such as Seoul virus (SEOV), Hantaan virus (HTNV), and Dobrava-Belgrade virus (DOBV), can cause haemorrhagic fever with renal syndrome (HFRS). Depending on the hantavirus species responsible, HFRS can result in case fatality rates of 1–10%, and there are approximately 150–200,000 cases per year, predominantly in China.^5,6 New-World hantaviruses such as Andes virus (ANDV) and Sin Nombre virus (SNV) are spread by New-World mouse and rat species (Sigmodontinae & Neotominae), which are found in the Americas. New-World hantaviruses can cause hantavirus cardiopulmonary syndrome (HCPS), a severe disease that has case fatality rates ranging from 30 to 60%. Conversely, infections with hantaviruses that are found in voles, lemmings, and muskrats (Arvicolinae), such as Puumala virus (PUUV) can cause a milder form of HFRS known as nephropathia epidemica (NE), which has a lower case fatality rate of ∼0.1%.^7,8 PUUV is the most common cause of hantavirus infection in Europe and was responsible for 98% of laboratory-confirmed infections in the European Union in 2020. Of the European countries, Finland and Germany exhibit the highest case numbers, accounting for 45% and 29% of the European cases between 2016 and 2020, respectively, while other central European countries, such as Austria, Sweden, Slovenia, and Slovakia, account for approximately 0.1%–12.1% of European cases.⁹ PUUV cases in Europe fluctuate and depend on bank vole populations, which are, in turn, influenced by complex environmental factors such as climate variation and predator populations.9, 10, 11, 12, 13, 14, 15 While PUUV is associated with low case fatality rates, infection can result in debilitating symptoms such as increased vascular permeability, acute kidney injury, and thrombocytopaenia, which often necessitate hospitalisation and potentially intensive care unit admission.16, 17, 18, 19, 20 Although most patients recover without long-term sequelae, rare cases of hypopituitarism have been reported.²¹ Several biomarkers have been investigated in patients suffering from PUUV, including interleukin (IL)-6, C-reactive protein (CRP), pentraxin-3 (PTX3), indoleamine 2,3-dioxygenase (IDO), and soluble urokinase-type plasminogen activator (uPAR), among others.^22,23 While PUUV infections are common in Europe, no targeted treatments are available, and no approved vaccines are currently available for the prevention of hantavirus disease outside of China and South Korea. While it is presumed that PUUV infection elicits lifelong immunity, the humoral immune responses to infection are poorly understood. Previous serology studies have predominantly focused on the nucleoprotein (N), which, while straightforward to produce recombinantly, is not a target of neutralising antibodies. The glycoprotein of PUUV is encoded by the medium (M) genomic segment, which is translated as a single polypeptide. This is post-translationally proteolytically cleaved into two separate proteins, Gn and Gc, which interact to form heterodimers.²⁴ These heterodimers interact with adjoining dimers to form tetrameric spikes on the virion surface, which mediate cell entry,²⁵ and are thus both targets of neutralising antibodies. Recent therapeutic intervention strategies have focused on broadly neutralising monoclonal antibodies elicited by natural infection in humans,²⁶ polyclonal sera derived from vaccinated humanised cattle,^27,28 and DNA-based vaccination platforms.²⁹ We therefore sought to investigate the kinetics and durability of the antibody response to PUUV infection using longitudinal serum sampling.

Methods

Biospecimen sampling

A total of 24 patients suffering from PUUV infection (confirmed by PUUV IgM point-of-care testing, Reagena POC PUUMALA IgM, and PUUV reverse transcription PCR) were enrolled in the study.^20,30 Detailed descriptions of this cohort, including information on patient demographics, comorbidities, symptoms, and laboratory values, can be found elsewhere.²⁰ Of these patients, 20.8% (n = 5) exhibited a severe course of disease (defined by the requirement of oxygen, haemodialysis, or intensive care unit admission), and one patient died. Serum was sampled from these patients at four time points post-hospitalisation which covers the early acute (T1, febrile phase, mean sampling time: six days post-symptom onset (PSO), range: 2–11 days PSO), late acute (T2, hypotensive/oliguric phases, mean sampling time: 12 days PSO, range: 8–20 days PSO), early convalescent (T3, polyuric/convalescent phases, mean sampling time: 28 PSO, range: 21–64 days PSO), and late-convalescent (T4, mean sampling time: 197 days PSO, range: 180–256 days PSO) stages of infection (Fig. 1). Of the 24 patients enrolled in the study, 20 had serum sampled at T1-4, one patient had samples taken at T1-3, one patient had samples taken at T1 and T2, and two patients had sera taken only at T1. This study was approved by the institutional review board of the Medical University of Graz (approval no. 33-329 ex 20/21). Written informed consent was obtained from all participants.

Fig. 1.

Schematic timeline of the symptoms associated with HFRS caused by PUUV infection. Following an incubation period of one to six weeks, HFRS symptoms manifest as five distinct phases: febrile, hypotensive, oliguric, polyuric, and convalescent. The average time for which each stage lasts is shown, in addition to the symptoms associated with each phase. Serum creatinine, a measure of kidney injury, is displayed, as is urine output as a marker of kidney function. IgM and IgG titres are also shown. The four time points at which patient sera were sampled are also shown in relation to each disease phase. Time point one (T1) was sampled six days post-symptom onset (PSO, range: 2–11 days PSO). Time point two (T2) was sampled 12 days PSO (range: 8–20 days PSO). Time point three (T3) was sampled 28 PSO (range: 21–64 days PSO). Time point four (T4) was sampled 197 days PSO (range: 180–256 days PSO). Adapted from Avsic-Zupanc et al.³¹ Created in BioRender. Clark, J. (2025) https://BioRender.com/p36u208.

Cells and viruses

Vero.E6 (American Type Culture Collection, ATCC Cat# CRL-1586, RRID:CVCL_0574, clone E6) and human hepatocarcinoma Huh7.5 cells³² (RRID:CVCL_7927) were cultured in Dulbecco's modified Eagle medium (DMEM; Gibco Cat# 11-995-065), supplemented with Antibiotic-Antimycotic (100 U/mL penicillin–100 μg/mL streptomycin–0.25 μg/mL amphotericin B; Gibco Cat# 15240062), and 10% foetal bovine serum (FBS; Corning Cat# 35-011-CV) in a 37 °C incubator with 5% CO₂. These cell lines were verified as mycoplasma-free using a commercially available PCR kit; however, they have not been authenticated using Short Tandem Repeat (STR) analysis. Recombinant vesicular stomatitis viruses (rVSVs) expressing enhanced green fluorescent protein (eGFP) and the GnGc from HTNV, SEOV, DOBV, and SNV have been previously described,²⁶ as has the rVSV expressing the PUUV GnGc in addition to a phosphoprotein fused to mNeonGreen (mNG).³³ The wild-type (WT) VSV (Indiana strain) was obtained from ATCC (Cat# VR-1238). The rVSV expressing the GnGc of ANDV is also described elsewhere.³⁴ Stocks of rVSV-ANDV, -HTNV, -SEOV, -DOBV, and WT-VSV were generated by infecting confluent monolayers of Vero.E6 cells at a multiplicity of infection (MOI) of 0.01 using 1 × minimal essential medium (MEM; Gibco Cat# 11430030) supplemented with 2% FBS, while stocks of rVSV-PUUV and -SNV were generated utilising Huh7.5 cells and identical methodology. All viruses were grown for three days at 37 °C before clarification via centrifugation at 4000g for 10 minutes at 4 °C. Viral stocks were titred using both the 50% tissue culture infectious dose (TCID₅₀) method and via classic plaque assay and were stored at −80 °C before use. For all infectious assays, Vero.E6 cells were used in conjunction with rVSV-ANDV, -HTNV, -SEOV, -DOBV, and WT-VSV while Huh7.5 cells were used with rVSV-PUUV and -SNV.

Enzyme-linked immunosorbent assay (ELISA)

To generate antigen for use in ELISA, rVSVs were used to infect confluent monolayers of either Vero.E6 or Huh7.5 cells in nine T175 tissue culture flasks (Corning Cat# 353112) at an MOI of 0.01. Media was drained from the flasks and replaced with virus diluted in 10 mL of MEM (Gibco Cat# 11430030) supplemented with 2% FBS. One hour post-infection, the inoculum was removed and replaced with 20 mL MEM (Gibco Cat# 11430030) supplemented with 2% FBS. Three days post-infection, supernatants were harvested and clarified via centrifugation at 4000×g for 10 minutes. A five mL 30% (w/v) sucrose cushion formulated in 1 × NTE buffer (1M NaCl, 100 mM Tris–HCl, and 10 mM ethylenediaminetetraacetic acid, pH 7.4) was then added to 30 mL of supernatant. Supernatants were concentrated via high-speed centrifugation at 25,000 rpm at 4 °C for two hours. The resulting virus pellet was resuspended in phosphate-buffered saline (PBS, Gibco Cat#10010-23, pH 7.4) and the protein quantity was determined via Bradford assay (Bio-Rad Cat#5000205). To generate recombinant Gn proteins, the coding sequences of the PUUV strain Sotkamo (NCBI Reference Sequence: NP_941983.1) and SEOV strain Baltimore (GenBank: AMQ36103.1) Gn ectodomains were cloned into the pFastBacDual (Gibco Cat# 10712024) vector as previously described.³⁵ Recombinant baculoviruses were rescued using a published protocol,³⁶ and purified as described previously.³⁵ Ultracentrifuge-purified rVSV was diluted in PBS to a concentration of 5 μg/mL and coated on Immulon 4 HBX 96-well plates (Thermo Scientific Cat# 3355) using 50 μL per well, while recombinant Gn proteins and PUUV nucleoprotein (NP, abcam, ab74555) were coated at a concentration of 2 μg/mL. 24 hours later, the coating solution was removed, and plates were blocked with PBS containing 0.01% Tween-20 (PBST; Fisher Scientific, Cat# BP337-500) and 3% non-fat milk (Bioworld Cat# 30620074-2). The blocking solution was incubated for one hour before three washes using PBST. Serum samples were serially diluted 1:3 from a starting dilution of 1:40 in PBST supplemented with 1% non-fat milk before addition to the respective plates. Sera dilutions were not added to 16 wells per ELISA plate to serve as a blank reading. After a one hour incubation at room temperature, the plates were washed three times with PBST and, when assaying total IgG, 100 μL of mouse anti-human IgG conjugated to horseradish peroxidase (HRP; Sigma–Aldrich, Cat# A0293-1ML, RRID:AB_257875) diluted 1:3000 in PBST supplemented with 1% non-fat milk was added to each well. When assaying other immunoglobulin subtypes, anti-human IgA HRP (Sigma–Aldrich, Cat# A0295, RRID:AB_257876, 1:3000), anti-human IgM HRP (Southern Biotech, Cat# 2020-05, RRID:AB_2795603, 1:3000), anti-human IgG1 (Southern Biotech, Cat# 9054-05, RRID:AB_2796627, 1:5000), anti-human IgG2 (Southern Biotech, Cat# 9060-05, RRID:AB_2796633, 1:5000), anti-human IgG3 (Southern Biotech, Cat# 9210-09, RRID:AB_2796701, 1:5000), or anti-human IgG4 (Southern Biotech, Cat# 9200-05, RRID:AB_2796691, 1:5000) were used. Plates were incubated for one hour at room temperature, washed three times with PBST, and 100 μL of o-phenylenediamine dihydrochloride (OPD, Sigma–Aldrich; Cat# P4664-100TAB) was added. Plates were developed for 10 minutes, after which the reaction was stopped by adding 50 μL of 3M hydrochloric acid (HCl). The plates were read using a Synergy 4 (BioTek) plate reader at an optical density of 490 nm. The area under the curve (AUC) for each plate was calculated using GraphPad Prism 10, with the average plus three standard deviations of the blank wells serving as the baseline. 11 age- and sex-matched healthy serum controls obtained from the Mount Sinai Health System were utilised as negative controls. The limit of detection for the assay was formulated by taking the average AUC plus three standard deviations of the negative serum samples. An in-house, cross-binding, anti-PUUV human monoclonal antibody, Ab523, was utilised as a positive control. This antibody was cloned from the memory B cell response of a patient infected with PUUV one year after infection, and its cross-binding/neutralising activity was previously confirmed via ELISA and microneutralisation assays using rVSV-PUUV, -SEOV, -HTNV, -DOBV, -ANDV, and -SNV.

Microneutralisation assays

Vero.E6 or Huh7.5 cells were seeded in a 96-well cell culture plate (Corning; Cat# 3340) at a density of 2 × 10⁴ cells per well. Sera was serially diluted 1:3 from a starting dilution of 1:10 to a final dilution of 1:7290 in 1 × MEM (Gibco Cat# 11430030) supplemented with 2% FBS. As with the ELISA, the same 11 age- and sex-matched healthy serum controls, and the in-house antibody Ab523, were utilised as negative and positive controls, respectively. 24 hours later, the rVSV was diluted to 10⁴ tissue culture infectious dose 50 (TCID₅₀)/mL, and 80 μL of virus and 80 μL of sera dilution were combined and incubated together for one hour at room temperature on a separate 96-well plate. After the incubation, 120 μL of virus–antibody inoculum was used to infect cells for one hour at 37 °C. The inoculum was then removed and 100 μL of each corresponding sera dilution was added to the wells in addition to 100 μL of 1× MEM supplemented with 2% FBS. Throughout this process, a total of six wells per plate were incubated with 1xMEM with 2% FBS containing no virus or sera, and another six wells with 1xMEM with 2% FBS containing virus with no sera to serve as control wells for the quantification of % inhibition. Cells were incubated at 37 °C in a 5% CO₂ incubator for two days and fixed with 10% paraformaldehyde (Polysciences Cat# 04018-4) for 24 h at 4 °C. The paraformaldehyde was removed, cells were washed twice with PBS, and permeabilised via the addition of 100 μL of PBS supplemented with 0.1% Triton X-100 (Biorbytorb Cat# 1566728). After 15 minutes, the Triton X-100 solution was removed, and cells were blocked with PBS supplemented with 3% non-fat milk (Bioworld Cat# 30620074-2) for one hour at room temperature. The cells were stained for one hour using a mouse anti-VSV-N antibody (Kerafast, clone 10G4, Cat# EB0009) diluted 1:3000 in PBST 1% milk. Cells were then washed twice with PBS and anti-mouse IgG conjugated to horseradish peroxidase (Rockland, Cat# 610-603-002) diluted 1:3000 in PBST with 1% milk was added to each well. After one hour, the plates were washed twice using PBS and developed via the addition of 100 μL of o-phenylenediamine dihydrochloride (Sigma–Aldrich; OPD). Ten minutes later, 50 μL of 3M hydrochloric acid (HCl) was added to stop the reaction, and the plates were read using a Synergy 4 (BioTek) plate reader at an optical density of 490 nm. The 50% inhibitory dilution (ID₅₀) of each serum sample was calculated using a previously described methodology.³⁷

ADCC in vitro reporter assays

Huh7.5 cells were seeded in white 96-well microplates (Corning Cat# 3917) at a density of 2 × 10⁴ cells per well. The next day, cells were infected with rVSV-PUUV at an MOI of one diluted in 1× MEM supplemented with 2% FBS, 100 μL per well. After one hour, the inoculum was removed and replaced with 100 μL 1× MEM with 2% FBS. 48 hours later, patient sera were serially diluted 1:3 from a starting dilution of 1:10 in 1 × MEM supplemented with 2% FBS to a final dilution of 1:21,870. The same age- and sex-matched healthy serum controls utilised in the ELISAs and microneutralisation assays were again employed as negative controls, and the in-house antibody, Ab523, was again utilised as a positive control. The media in the wells was aspirated and replaced with 25 μL Roswell Park Memorial Institute 1640 (RPMI-1640) media (Thermo Scientific Cat# 22400071) containing 2% low-IgG FBS (Thermo Scientific Cat#A3381901) and 25 μL of sera dilution. A total of 16 wells did not have sera added to them to serve as background luminescence readings. As with the ELISAs and microneutralisation assays, 11 age- and sex-matched healthy serum controls, and the in-house antibody Ab523, were utilised as negative and positive controls, respectively. 3 × 10⁶ ADCC Bioassay Effector Cells expressing the human FcγRIIIa receptor (Promega Cat# G9302) diluted in 25 μL RPMI-1640 with 2% low-IgG FBS were added per well and the cells were incubated at 37 °C with 5% CO₂ for six hours. 75 μL of Bio-Glo luciferase reagent (Promega Cat# G7940) was added to each well and luminescence was quantified using a Synergy 4 (BioTek) plate reader. The 16 wells with no added sera were used to determine a background luminescence reading which was used to calculate the AUC using GraphPad Prism 10.

Cytokine analysis

The concentrations of IL-18, IL-33, MCP-1, IL-8, IFNα, IFNβ, IFNγ, IL-13, IL-23, IL-1β, IL-22,IL-6, IL-4, IL-5, GM-CSF, IL-12p70, IL-10,TNFα, IL-17F, IL-2 and IL-17A in serum samples were determined using a LEGENDplex bead-based immunoassay (Biolegend, USA) in accordance with the manufacturer's instructions. Briefly, 25 μL of serum was mixed with equal volumes of assay buffer, and beads to bring the final volume to 75 μL per well. A serial dilution of standards was prepared in the assay buffer and further processed in the same way. The samples were incubated in the dark on a plate shaker at 750 rpm for two hours at room temperature. After washing with 200 μL of wash buffer (centrifuged at 1100 rpm for five minutes at 25 °C), the samples were incubated with 25 μL of detection antibodies for one hour on a plate shaker. Following this, an equal volume of Streptavidin-R-phycoerythrin conjugates was added, and the samples were incubated for another 30 minutes with subsequent washing and reconstitution in 100 μL of PBS. Data were collected using the Attune flow cytometer (Thermo Fisher Scientific) and processed with FlowJo_v10.8.1 and LEGENDplex™ data analysis software (https://legendplex.qognit.com).

Phylogenetic analysis

Full-length M segment sequences were downloaded from GenBank. Andes virus (ANDV, accession number AF291703.2), Black Creek Canal virus (BCCV, accession number L39950.1), Choclo virus (CHOV, accession number DQ285047.1), Dobrava-Belgrade virus (DOBV, accession number NC_005234.1), Hantaan virus (HTNV, accession number KT885048.1), Prospect Hill virus (PHV, accession number X55129.1), Puumala virus (PUUV, accession number KT885051.1), Seoul hantavirus (SEOV, accession number M34882.1), Sin Nombre virus (SNV, accession number L25783.1), and Tula virus (TULV, accession number Z69993.1) were included. The coding sequences were extracted and aligned using MUSCLE (version 3.8.425)³⁸ and translated into amino acid sequences within the Geneious Prime software package (version 2022.1.1, https://www.geneious.com). Amino acid sequences were realigned using MUSCLE, and a maximum likelihood phylogenetic tree was compiled using PHYML utilising the Le Gascuel substitution model with 100 bootstrap replicates.³⁹ The final maximum likelihood tree was visualised using FigTree version 1.4.5 (https://github.com/rambaut/figtree/releases).

Statistical analysis

For ELISA, microneutralisation, and ADCC reporter assay data, statistical analysis was performed in GraphPad Prism. AUC and ID₅₀ values were log10-transformed, and statistical significance was assessed via ordinary one-way ANOVA using Tukey's multiple comparisons test. Significance was defined as p < 0.05 and is indicated in graphs where present. Cytokine concentrations in serum samples from patients infected with PUUV at various time points were compared using the pairwise Wilcoxon test. Normality of sample distribution was assessed with the Shapiro–Wilk test. Pearson or Spearman correlation analyses were applied to normally or non-normally distributed variables, respectively, with p-values adjusted for multiple comparisons using the Benjamini-Hochberg false discovery rate method. For heatmap visualisation, a log (p + 1) transformation was applied, followed by min–max normalisation. Ward's clustering algorithm was employed to group individuals based on similar immune response characteristics. Comparisons between groups identified through serum cytokine profiles were conducted using Student's t-test. Statistical significance was defined as p < 0.05. All analyses were performed in RStudio (R version 4.3.3) and Google Colab (Python version 3.10).

Role of funders

The funders of the study had no role in experimental design, data collection, data analysis, data interpretation, or writing of the study.

Results

Infection elicits a strong anti-PUUV IgG response with cross-binding antibodies produced at convalescence

To investigate the humoral immune response to PUUV infection, we utilised a previously described cohort of 24 patients from Graz, Austria, who were hospitalised due to PUUV infection and had longitudinal sera collected during hospitalisation and after discharge (up to 256 days after symptom onset).²⁰ Sera sampling covered the five phases of disease progression (Fig. 1). To assess the IgG response to PUUV, we utilised ultra-centrifuge-purified rVSV expressing the full-length GnGc of PUUV as antigen for use in ELISAs.²⁶ Each of the serum samples was assayed for the presence of anti-PUUV IgG alongside 11 age- and sex-matched healthy serum controls obtained from the Mount Sinai Health System. Due to serum scarcity, some samples were prioritised for microneutralisation assays instead of ELISAs. The samples that were assayed using each technique are shown in Table S1. At T1, 80.9% (17/21) of serum samples exhibited a robust anti-PUUV IgG response which was maintained at T2 and T3 at which point 90.5% of the serum samples (19/21) displayed IgG (Fig. 2A). The IgG titre remained high at T4, 6–7 months after infection at which point all the patients analyzed (20/20) had seroconverted, indicating a strong adaptive immune response to infection. Anti-SEOV IgG was absent from most patients at T1-3. However, titres were detected at T4 for 50% (10/20) of patients (Fig. 2B). Similarly, anti-DOBV titres were low at T1-3 but detectable in 30% (6/20) of serum samples analyzed at T4 (Fig. 2C). Conversely, more patients exhibited detectable anti-HTNV IgG titres at T1-3, with 65% (13/20) patients displaying IgG at T4 (Fig. 2D). Anti-SNV IgG titres were low at T1-3 and detectable only in 10% (2/20) of serum samples at T4 (Fig. 2E). A small number of patients also exhibited anti-ANDV IgG at T1–3, however by T4, 75% (15/20) of the patients had detectable IgG titres (Fig. 2F). This indicates that cross-binding antibodies are elicited by PUUV infection, with the highest proportion of patients at T4 exhibiting cross-reactive IgG, but at early time points this varied according to the hantavirus GnGc tested. To verify that the observed IgG responses are hantavirus-specific and not due to antibodies binding to the rVSV vector, ELISAs were carried out using wild-type VSV. IgG titres were comparable between the PUUV patient sera and control sera, indicating that the observed IgG titres are specific to the hantavirus glycoproteins tested (Fig. 2G).

Fig. 2.

Patient sera exhibit robust anti-PUUV IgG at all time points and cross-binding IgG by T4. ELISAs were carried out using patient sera and plates coated with ultracentrifuge-purified rVSV expressing the GnGc of A) PUUV, B) SEOV, C) DOBV, D) HTNV, E) SNV, or F) ANDV. Wild-type VSV G) was used as a negative control. IgG titres are shown as area under the curve (AUC) with the geometric mean and geometric standard deviation denoted by error bars. 11 age and sex matched healthy control sera were included as negative controls and an in-house, anti-PUUV monoclonal antibody was used as a positive control. AUC values were log10 transformed and statistical significance was interrogated via ordinary one-way ANOVA using Tukey's multiple comparisons test. Only comparisons with p ≤ 0.05 are shown. Limits of detection (LOD) were calculated for each antigen using the average + 3× standard deviations of the AUC of the negative control sera and are shown as dotted lines on each graph. The percentage of serum samples at each sampling time point with AUC values above the LOD are shown as pie charts, with the percentage of positive samples displayed in green. The n for each time point is displayed below each pie chart. H) Maximum likelihood phylogenetic tree of orthohantavirus M segment amino acid sequences. Clades are colour-coded according to the reservoir host species, with blue denoting viruses found in Arvicolinae, red viruses found in Sigmodontinae, and green viruses found in Murinae. Viruses included are: Puumala virus (PUUV), Prospect Hill virus (PHV), Tula virus (TULV), Andes virus (ANDV), Chocclo virus (CHOV), Black Creek Canal virus (BCCV), Sin Nombre virus (SNV), Dobrava-Belgrade virus (DOBV), Seoul virus (SEOV), and Hantaan virus (HTNV). Viruses that are included in this study are shown in bold. The scale bar represents amino acid substitutions per site, and bootstrap consensus support values are denoted at branches if less than 90.

The phylogenetic relationship of the tested hantavirus GnGcs is shown as an amino acid-based phylogenetic tree (Fig. 2H). Most serum samples exhibited a high background AUC as determined by the average plus three times the standard deviation of the negative control samples. To determine if the high background values were due to the use of rVSV pseudotyped viruses, we employed recombinant PUUV and SEOV Gn for comparison with the full-length GnGc expressed by the rVSVs. Titres against PUUV Gn were reduced compared to the full-length GnGc (Fig. S1A), while titres against SEOV Gn were comparable (Fig. S1B). The limit of detection was slightly lower for the recombinant Gn ELISA compared to the GnGc, therefore, more patients exhibited detectable IgG titres at T4 (12/20, 60%) compared to the rVSV-SEOV ELISA (10/20, 50%).

Neutralising activity is detected at early time points and is highest at convalescence

As PUUV infection elicits a strong IgG response, with samples taken at T4 exhibiting cross-binding antibodies, we utilised the rVSV pseudotyped viruses for microneutralisation assays to determine whether these sera also demonstrated neutralising activity. Like the IgG ELISA data, patients exhibited robust anti-PUUV neutralising titres at early time points which were significantly higher at T4 compared to T1 (Fig. 3A). Conversely, despite most patients displaying little detectable IgG at T1-3, some patient sera neutralised rVSV-SEOV at earlier time points, with 95% (19/20) of patients showing neutralising activity at T4 (Fig. 3B). Similarly, although most patients did not exhibit anti-DOBV IgG titres at both acute and convalescent time points, neutralising titres were detected at T1-3 with 95% (19/20) of T4 sera displaying detectable neutralisation (Fig. 3C). A higher proportion of patients displayed neutralising titres against rVSV-HTNV in 75% (18/24) of T1 sera, 86.4% (19/22) at T2, 66.7% (14/21) at T3, and finally 85% (17/20) at T4 (Fig. 3D). Neutralising activity against rVSV-SNV was more muted, with a small minority of patient sera at T1-3 exhibiting neutralising titres which then increased to 20% (4/20) at T4, though to a greater extent than the IgG titres (Fig. E). Again, this is unlike the ELISA data in which a smaller proportion of patients had detectable anti-SNV IgG, which peaked at T4. Finally, rVSV-ANDV neutralising titres were detected in 41.6% (10/24) patients at T1 and 54.2–52.4% (13/24–11/21) at T2-3 before increasing to 75% (15/20) of patients at T4 (Fig. 3F). Again, wild-type VSV was utilised as a negative control to ensure that the observed neutralisation is due to the specific neutralisation of the hantavirus GnGcs expressed by each pseudotyped virus. Like the ELISA data, none of the patient sera neutralised wild-type VSV, indicating that the observed neutralisation is due to anti-GnGc antibodies (Fig. 3G).

Fig. 3.

Some patient sera exhibit cross-neutralisation at acute and early convalescent time points, with the highest proportion of sera with cross-neutralising activity at T4. Microneutralisation assays were carried out using patient sera and rVSV pseudotyped viruses expressing the GnGc of A) PUUV, B) SEOV, C) DOBV, D) HTNV, E) SNV, or F) ANDV. WT-VSV G) was utilised as a negative control. Neutralising titres are shown as 50% inhibitory dilutions (ID₅₀s) with the geometric mean and geometric standard deviation denoted by error bars. 11 age and sex matched healthy control sera were included as negative controls and an in-house, anti-PUUV monoclonal antibody was used as a positive control. ID₅₀ values were log10 transformed and statistical significance was interrogated via ordinary one-way ANOVA using a Tukey's multiple comparisons test. Only comparisons with p ≤ 0.05 are shown. Limits of detection (LOD) are shown as dotted lines on each graph. The percentage of serum samples at each sampling time point with ID₅₀ values above the LOD are shown as pie charts with the percentage of positive samples displayed in green. The n for each time point is displayed below each pie chart.

To determine whether other immunoglobulin subclasses with cross-binding potential may be present in the patient sera, we performed ELISAs utilising anti-IgM and IgA secondary antibodies. We found that PUUV GnGc-specific IgM was present in around 39.1% (9/23) of the patients at T1, 40.9% (9/22) at T2, and 33.3% (7/21) at T3. This significantly decreased by T4, with all patients except one exhibiting titres below the limit of detection at T4 (Fig. 4A). Anti-PUUV IgA was also detectable in 21.7% (5/23) of patient sera at T1, which increased to 40.9% (9/22) at T2 before moderately decreasing to 25% (5/20) by T4 (Fig. 4B). To determine if cross-binding IgM or IgA were elicited in response to infection, we carried out ELISAs using ultracentrifuge-purified rVSV-SEOV. Cross-binding anti-SEOV GnGc IgM was present in 17.4% (4/23) of patient sera at T1, which increased to 27.3% (6/22) and 25% (5/20) patients by T2 and T3 before decreasing below the limit of detection for all but two patients by T4 (Fig. 4C). Anti-SEOV GnGc IgA was present in 29.2% (7/24) of patient sera at T1, and persisted until T2, before reducing below the limit of detection for all but two patients at T3 and T4 (Fig. 4D). This indicates that cross-binding IgM and IgA are present at earlier time points and that these may be responsible for the observed cross-neutralisation of other hantaviruses before cross-binding IgG titres increase at T4.

Fig. 4.

The quantity of immunoglobulin subtypes changes between the acute and convalescent phases and is associated with a decrease in Fc effector function. ELISAs were carried out using patient sera to investigate A) anti-rVSV-PUUV IgM, B) anti-rVSV-PUUV IgA, C) anti-rVSV-SEOV IgM, and D) anti-rVSV-SEOV IgA. IgG subtype ELISAs were performed to investigate the quantities of anti-rVSV-PUUV E) IgG1, F) IgG2, G) IgG3, or H) IgG4 and anti-PUUV NP I) IgG1, J) IgG2, K) IgG3, or L) IgG4. M) A schematic of the luciferase-based reporter assay, which was utilised to characterise the effector activity promoted by patient sera. rVSV-PUUV-infected Huh7.5 cells were incubated with patient sera and reporter cells that express the FcγRIIIa receptor, coupled to a gene expression pathway that results in the production of luciferase. N) The FcγRIIIa activity of the sera are shown as AUCs, with the mean and standard deviation denoted by error bars. 11 age and sex matched healthy control sera were included as negative controls, and an in-house, anti-PUUV monoclonal antibody was used as a positive control. AUC values were log10 transformed and statistical significance was interrogated via ordinary one-way ANOVA using a Tukey's multiple comparisons test. Only comparisons with p ≤ 0.05 are shown. Limits of detection (LOD) are shown as dotted lines on each graph. The percentage of serum samples at each sampling time point with AUC values above the LOD are shown as pie charts with the percentage of positive samples displayed in green. The n for each time point is displayed below each pie chart.

The proportion of the IgG subclasses changes post-infection and is associated with a decreased FcγRIIIa activation

As cross-reactive IgG is most abundant in collected sera at T4, we sought to determine whether this increase is associated with changes in the abundance of IgG subclasses. We utilised ultracentrifuge-purified rVSV-PUUV alongside a panel of IgG subclass-specific secondary antibodies to quantify IgG1, IgG2, IgG3, and IgG4 in the patient sera. At T1, anti-PUUV IgG1 was present in 62.5% (15/24) of the patient sera, albeit at low titres (Fig. 4E). The titre significantly increased by T2. At that point, 90.9% (20/22) of patients exhibited detectable anti-PUUV IgG1. These titres again significantly increased by T3 at which point all the patients exhibited a robust IgG1 response, which again increased significantly at T4 (Fig. 4E). Conversely, IgG2 was absent from all patient sera at all time points (Fig. 4F). IgG3 was present in 58.3% (14/24) of the patient sera at T1. It significantly increased at T2 at which point 90.9% (20/22) of patients had detectable titres (Fig. 4G). These titres remained constant at T3, however, at T4 the IgG3 titres significantly decreased and were below the limit of detection in 65% (13/20) of the patients (Fig. 4G). IgG4 was absent from all the patient sera at T1 and was detectable in a single patient at T2 and in two patients at T3 (Fig. 4H). At T4 there was a significant increase in IgG4, with 40% (8/20) of patients displaying low, but detectable titres (Fig. 4H). To determine if the IgG subtype profile of the patients was similar against another viral protein, we repeated these assays using recombinant PUUV NP. Like the anti-GnGc response, we found that anti-NP IgG1 titres were significantly increased at T2, T3, and T4, compared to T1, although the AUC was lower (Fig. 4I). Conversely, unlike the anti-GnGc response, 20% of patients (5/20) exhibited detectable IgG2 antibodies at time point 4 (Fig. 4J). The anti-NP IgG3 response showed a similar pattern to the anti-GnGc response, with a significant increase in titre between T1 and T2, followed by a significant decline by T4, whereupon 55% (11/20) of patients were below the limit of detection (Fig. 4K). The anti-NP IgG4 response was comparable to the GnGc response; however, a larger proportion of patients exhibited detectable IgG4 at T4 (70%, 14/20), and titres were marginally higher (Fig. 4L). As the Fc portions of the IgG subtypes interact with the Fcγ receptors expressed on immune cell populations to different extents, we sought to determine if the sera from each time point could activate FcγIIIa. We therefore carried out antibody-dependent cellular cytotoxicity (ADCC) reporter assays using cells infected with rVSV-PUUV (Fig. 4M). At T1, all the serum samples exhibited robust FcγIIIa activity, which was maintained at T2-3 (Fig. 4N). At T4, however, this activity was significantly reduced by 10-100-fold for most patients, with only 4 patient samples maintaining high levels of activity (Fig. 4N). This indicates that changes in the abundance of the IgG subtypes likely influence the Fc-dependent activity of the sera.

Patients can be segregated based on the serum abundance of Th1 and Th2 associated cytokines

While conducting serology to assess antibody responses, we utilised the same samples to evaluate cytokine responses in the serum of the initial 24 individuals. Concentrations of inflammation type-I/Th1-associated cytokines (IL-1β, IL-6, IL-2, MCP-1, TNFα, IL-8, IL-18, IFNα, IFNβ, IFNγ, IL-12p70), inflammation type-II/Th2-associated cytokines (IL-10, IL-33, IL-13, IL-4, IL-5), and inflammation type-III/Th17-associated cytokines (IL-23, IL-22, GM-CSF, IL-17F, and IL-17A) in serum samples were determined. The concentration of all analysed cytokines increased in the acute stage (T2) were compared to the late convalescent stage (T4). The most significant increases relative to the “baseline” (cytokine concentrations in the convalescent stage) were observed for IFNγ, IL-6, IL-13, IL-10, and IL-17 (Fig. 5a). Patients with severe PUUV infection (as defined by a composite endpoint: ICU admission, need for oxygen supply, and/or haemodialysis) exhibited elevated serum concentrations of IL-23, IL-22, IL-4, IL-13, IL-5, IL-2, and IL-10 (Fig. 5b). Based on principal component analysis (PCA) results, two major patterns of cytokine expression can be identified: one) IL-2, IL-4, IL-5, IL-10, IL-13, IL-22; and two) IL-6, IFNγ, IL-8, IFNα, IFNβ, TNFα, IL-12p70, IL-33, GM-CSF (Fig. 5c). The first set of cytokines is primarily represented by major regulators of the type-II inflammatory response, produced by type 2 innate lymphoid cells (ILC-2) and T helper 2 (Th2) cells, while the second set consists of inflammation type-I-associated cytokines. Interestingly, the type-II-polarised immune response was primarily associated with severe disease cases in the acute stage of the infection. The highest correlation coefficients were shown between the concentrations of IFNα, IFNγ, IL-12p70, IL-23, and IL-33 and HTNV-neutralising antibody titres in the acute phase of the infection (Fig. 5d). Additionally, positive correlations were shown between the levels of neutralising antibodies against rVSV-ANDV, -DOBV, -HTNV, and -SEOV in the convalescent stage and the concentrations of IL-23 in the acute stage of the PUUV infection (Fig. 5e). However, the revealed associations between the analysed variables were not statistically significant according to the results of Spearman correlation analysis (Fig. S2).

Fig. 5.

Cytokine profile of patients at acute and convalescent disease stages. A) Concentrations of cytokines in serum samples of patients infected with PUUV at acute (T2) versus convalescent (T4) time points. Values were compared using a pairwise Wilcoxon test (∗: p < 0.05). B) Heat maps display the normalised serum concentration of each cytokine in each sample at the acute (T2) and convalescent (T4) phases. Values were normalised using log (p + 1) transformation and min–max normalisation. Severe disease cases are indicated with red labels. C) Principal component analysis (PCA) biplots representing the distribution of patients infected with PUUV by the serum concentration of cytokines at T2 as well as factor loads of analysed variables on the first two principal components. Severe cases are shown in red. D-E) Heatmaps represent Spearman correlation analysis between acute-phase cytokine concentrations and humoral immune response parameters in the acute phase (D) as well as between acute-phase cytokine concentrations and humoral immune response parameters in the convalescent phase (E). Only humoral immune response variables with a seroconversion rate exceeding 50% were included in the correlation analysis. To refine the data, the limit of detection (LOD), calculated as three times the standard deviation (SD) of negative samples, was subtracted from the values used in the analysis. Any resulting negative values were replaced with zero. Log (p + 1) transformation and min–max normalisation were applied to the original data. Benjamini-Hochberg false discovery rate adjustment was used for multiple comparisons.

Discussion

Prior studies focussing on hantavirus serology have predominantly concentrated on anti-NP antibodies. This protein is relatively easy to produce recombinantly; however, it does not harbour any neutralising epitopes and is highly conserved between hantaviruses compared to GnGc. Utilising rVSV pseudotyped viruses that express the full-length GnGc, we undertook a comprehensive analysis of the anti-GnGc humoral immune response to PUUV infection in a longitudinal cohort. We found that PUUV infection resulted in a strong IgG response, which was detectable in the patients at T1, shortly after disease onset. This has been previously reported and is characteristic of the long incubation period associated with PUUV infection, which ranges from two to six weeks.^17,40,41 Anti-PUUV total IgG titres were stable from T1 to T4, at which point all the patients exhibited detectable IgG. Interestingly, cross-binding antibodies capable of binding the GnGc of both Old-World (SEOV, HTNV, DOBV) and New-World hantaviruses (SNV, ANDV) were detectable in a subset of patients, predominantly at T4. A potent anti-PUUV neutralising antibody response was also noted, with sera from all time points exhibiting neutralising activity, which peaked at T4. Unlike the IgG ELISA data, over 50% of patients displayed cross-neutralising activity against rVSV-SEOV, -DOBV, and -ANDV at T1-3, which increased to over 75% of patients at T4. Patient sera showed a more muted cross-neutralising response against rVSV-SNV and an enhanced response against rVSV-HTNV, which more closely matched the observed anti-SNV and anti-HTNV IgG titres. For each of the VSV pseudotyped viruses tested, the observed number of serum samples that displayed cross-neutralising activity was higher than those that demonstrated cross-binding IgG. These results may be partially explained by differences in sensitivity between the ELISA and microneutralisation assay methodologies, particularly due to the high background IgG values noted in the age and sex-matched healthy donors, which were used as negative control samples. This discrepancy at earlier time points could also be explained by the presence of cross-neutralising IgM and IgA. Indeed, cross-binding anti-SEOV IgM was detectable in patients at T1 and peaked, remaining stable at T2-3, before subsiding by T4. In contrast, IgA was detectable in around a third of patients at T1 and T2, before clearing by T3. While immunoglobulin subclasses aside from IgG exert neutralising activity, we are unable to determine which of these subclasses may be responsible for the observed cross-neutralisation utilising only ELISA. This could be investigated further by depleting IgG from the sera before carrying out additional neutralisation assays, however due to serum scarcity, we were unable to utilise this methodology. Additionally, due to the serum scarcity and the difficulty associated with the use of hantaviruses for cell-based assays, we utilised pseudotyped VSV instead of authentic hantaviruses. While the results obtained from pseudotyped virus-based neutralisation assays have been shown to correlate well with those using authentic viruses,⁴² some differences may exist in the extent of neutralisation.

Presently, no therapeutics are available for the treatment of hantavirus infections, and current treatment strategies are limited to the alleviation of disease symptoms. The broad-acting antiviral ribavirin is effective in the treatment of HTNV infection if given soon after symptom onset,^43,44 however, there is limited evidence of its efficacy in the treatment of HCPS,^45,46 and it failed to reduce PUUV viral loads in a clinical trial.⁴⁷ Conversely, convalescent plasma from survivors of ANDV infection is efficacious in the treatment of HCPS.⁴⁸ Therefore, one promising therapeutic avenue is the use of human-derived monoclonal antibodies cloned from the B cells of hantavirus infection survivors. Potently neutralising antibodies cloned from convalescent patients infected with ANDV are protective in the Syrian golden hamster challenge model,⁴⁹ promoting partial protection even when administered 10 days post-infection.⁵⁰ While encouraging, these mAbs were administered to infected animals before the development of clinical signs, and also at a high dose (50 mg/kg).⁵⁰ Broadly neutralising antibodies have also been identified in survivors of SNV infection and PUUV infection.^26,51 Our results indicate that the optimal time point for the cloning and production of broad-binding antibodies is upwards of six months post-infection. The results presented in this study should, therefore, be taken into consideration in future studies focused on the identification of therapeutic, human-derived, monoclonal antibodies.

Interestingly, cross-reactive anti-GnGc IgG titres were highest at T4, 6–7 months after disease onset. These patients are presumably exposed to the virus a single time; however, months after infection, IgG class switching takes place, giving rise to increased IgG1, decreased IgG3, and increased IgG4, concurrent with increasing titres of cross-binding antibodies. Notably, this trend was observed for anti-GnGc IgG subtypes, as well as anti-NP. The development of IgG4 antibodies in response to infection is poorly understood; however, it is associated with repeated or chronic exposure to antigens. Repeated severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) mRNA vaccination has been shown to induce IgG4 responses,⁵² as has repeated exposure to phospholipase from honey bee stings.^53,54 Similar changes in IgG repertoire have also been demonstrated in survivors of Ebola virus (EBOV) infection,⁵⁵ with strong evidence for persistent viral infection in immune-privileged tissues such as the eyes and testes.56, 57, 58, 59, 60 Repeated exposure to antigen appears to be necessary but not sufficient to induce IgG class-switching towards IgG4, as repeated tetanus vaccination has not been found to stimulate an IgG4 response,⁶¹ nor has persistent infection with human cytomegalovirus.⁶² Interestingly, increased IgG4 titres have been identified in one previous publication in which anti-Gn and anti-Gc IgG4 titres were found to be significantly increased in a smaller cohort of patients infected with PUUV two years post-PUUV infection.⁶³ This study also noted an increase in IgG1 titres and a decrease in IgG3 titres, in line with our findings.⁶³ Each IgG subtype interacts with the Fcγ receptors (FcγRs) present on immune cell populations to different degrees, promoting antibody-dependent cellular cytotoxicity, antibody-dependent cellular phagocytosis, and complement protein binding.⁶⁴ Of the IgG subtypes, IgG3 is a potent inducer of effector functions, more so than IgG1 (63). Conversely, IgG4 exhibits the weakest binding to FcγRs and is generally considered a non-inflammatory, poor inducer of effector functions.⁶⁵ The significant decrease in ADCC observed at T4 coincides with a significant decrease in IgG3 titres and a concomitant increase in IgG4 titres, indicating that the abundance of these IgG subtypes likely influences the FcγR activation of patient sera. However, the FcγRIIIa activation assay utilised in this study is a luciferase-based reporter assay that does not measure ADCC activity via the killing of target cells. While the FcγRIIIa receptor is the sole receptor expressed on natural killer cells, other immune cell populations express additional FcγRs; therefore, this assay is unable to recapitulate the complex antibody-FcγR interactions that take place in vivo. Taken together, the observed increases in cross-binding and cross-neutralisation, increased class-switching, and reduced effector function at T4 are indicative of B-cell maturation, which is ongoing 6–7 months post-infection. These observations must be confirmed using fluorescence-activated cell sorting (FACS) and B-cell sequencing, which will be the goal of future studies.

These data suggest that PUUV may establish persistent or latent infections in humans, or that viral antigen is retained and continuously presented to the immune system. As the abundance of IgG subclass antibodies targeting both anti-GnGc and anti-NP exhibited similar trends following infection, this may hint that a viral reservoir is indeed present in patients infected with PUUV months after recovery. Indeed, neutralising antibody titres against PUUV have been shown to increase in the years after infection, with some patients exhibiting high titres decades later.⁶⁶ This is unlike respiratory pathogens such as influenza A virus (IAV) and SARS-CoV-2, where the antibody titres stimulated by infection or vaccination increase following exposure, wane after 6 months, and then stabilise.67, 68, 69 Additional sera sampling from the cohort described here, one to two years post-infection, would shed further light on this phenomenon. Other evidence for hantavirus persistence in humans comes from studies that have shown the presence of high levels of highly differentiated long-lived CD127^- memory T cells years after ANDV infection,⁷⁰ and the expansion of NK cells that remained functional over 60 days after PUUV infection.⁷¹ In contrast to PUUV infection, IgG1 has been reported as the predominant subclass in convalescent patients infected with Choclo virus (CHOV),⁷² while IgG3 has been identified as dominant in convalescent patients infected with SNV.⁷³ However, in the absence of ongoing class switching, patients infected with ANDV and SNV also exhibit elevated neutralising antibody titres years after exposure.74, 75, 76 In another study, ANDV RNA was detected in the semen of a patient 71 months after exposure, indicating that viral reservoirs may persist in disease survivors for years after infection.⁷⁷ Additional experiments, including post-mortem tissue analysis, are necessary to determine the possible location of viral reservoirs before it can be confirmed that hantaviruses establish persistent infections in humans.

Our cohort of patients infected with PUUV was characterised by a substantial increase in the production of proinflammatory and anti-inflammatory cytokines associated with all major branches of the immune response (Th1, Th2, Th17). The most significant increase was shown for IFNγ, IL-6, IL-13, IL-10, and IL-17. Elevated levels of IL-6, IL-10, and IFN-γ have previously been reported in the sera of acute PUUV patients,⁷⁸ severe HCPS cases,⁷⁹ and PUUV-infected cynomolgus macaques.⁸⁰ Patients with severe PUUV infection were characterised by the elevated production of Th2/Th17-associated cytokines IL-23, IL-22, IL-4, IL-13, and IL-5. This suggests a potentially dysregulated immune response contributing to disease severity. It was shown previously that the protective immune response to hantaviral infections is mediated primarily by Th1 CD4+ and cytotoxic CD8+ T-cells.^81,82 Patients with severe/critical HFRS showed lower amounts of single cytokine (IFN-γ, IL-2, and TNF-α) or dual-cytokine-producing CD4+ Th1 cells. On the other hand, a trend towards higher levels of perforin + CD4+ T cells was observed in the mild/moderate group. Th2 responses usually counterbalance Th1 inflammation and excessive Th2 activation might not effectively control the virus and could lead to immunopathology.⁸³ It has previously been demonstrated that type 2 innate lymphoid cells (ILC2), which are major regulators of type II inflammation and are responsible for Th2 polarisation, play a significant role during the acute phase of PUUV infection.⁸⁴ Viral infection induces the release of alarmins such as IL-25, IL-33, and thymic stromal lymphopoietin (TSLP) which activate ILC2 thereby promoting the secretion of IL-4, IL-5, and IL-13. This increased production of type-II-associated cytokines was also shown to be associated with higher disease severity.⁸⁴ The elevated levels of IL-23 in the sera of acute patients have been observed in other studies, however, we demonstrated a non-statistically significant correlation between the IL-23 levels in the acute stage of infection and the cross-neutralising humoral immune response against various hantaviruses (ANDV, DOBV, HTNV, SEOV) in the convalescent stage. This suggests that, while IL-23 may contribute to acute severity, it also promotes broader antibody-mediated immunity over time. Taken together, these data identify biomarkers of PUUV infection that may have diagnostic utility and may inform the analysis of future hantavirus vaccine design platforms that aim to promote broad antibody responses to protect against multiple hantaviruses.

Contributors

SH and RK performed clinical work in Austria. JJC, KV, RAS, JSY performed assays. EM, EK, and KC provided reagents. RK, VS, and FK supervised the study; JJC, SH, RK, and FK developed the concept. JJC, KV, and SH drafted the manuscript. JJC, KV, SH, VS, RK, and FK performed the data analysis and accessed and verified the underlying data. All authors edited and approved the manuscript.

Data sharing statement

All data is available from ImmPort accession ID: SDY3327.

Declaration of interests

JJC declares that he got Postdoctoral Scholar Travel Award to present this work at the 42nd Annual Meeting of the American Society for Virology. KC declares funding from BARDA, DTRA and NIH and stock and ownership interest in Eitr Biologics, Inc. (not related to current work). All other authors declare no conflict of interest.

Appendix A. Supplementary data

Fig. S1.

Patient sera IgG binding to recombinant PUUV and SEOV Gn protein. ELISAs were carried out using patient sera and plates coated with recombinant truncated Gn of A) PUUV and B) SEOV. IgG titres are shown as area under the curve (AUC) with the geometric mean and geometric standard deviation denoted by error bars. 11 age and sex matched healthy control sera were included as negative controls, and an in-house, anti-PUUV monoclonal antibody was used as a positive control. AUC values were log10 transformed, and statistical significance was interrogated via ordinary one-way ANOVA using Tukey's multiple comparisons test. Only comparisons with p ≤ 0.05 are shown. Limits of detection (LOD) were calculated for each antigen using the average + 3× standard deviations of the AUC of the negative control sera and are shown as dotted lines on each graph. The percentage of serum samples at each sampling timepoint with AUC values above the limit of detection are shown as pie charts, with the percentage of positive samples displayed in green. The n for each time point is displayed below each pie chart.

Fig. S2.

Correlation analysis between cytokine and humoral responses in acute versus convalescent disease phases. Plots show variables with the highest correlation coefficients between the cytokine concentrations in the acute phase of PUUV infection and the humoral immune response parameters in the A) acute (T2) or B) convalescent (T4) phases of the infection.

Table S1

Patient sera samples that were utilised in this study. Individual patients are shown with the sera samples taken at each of the four time points on the y-axis, while each experimental procedure is shown on the x-axis. Green squares denote where a sample was included, and red squares denote where a sample was not included due to serum scarcity.