Serological evidence of zoonotic filovirus exposure among bushmeat hunters in Guinea

Ebolavirus outbreaks in West and Central Africa have increased in frequency and severity, including two WHO-declared PHEICs caused by Zaire ebolavirus (EBOV). EBOV typically spills over from sylvatic reservoirs, with bats widely hypothesised as the principal reservoir based on detection of EBOV RNA or antibodies in multiple bat species and the isolation of Bombali virus (BOMV) from healthy bats. The 2013–2016 West Africa epidemic was the largest EBOV outbreak, with >28,000 cases and >11,000 deaths, and began in Meliandou, Guinea, with suspected bat interactions. EBOV reactivations have since caused outbreaks in Guinea, challenging the single spillover model and highlighting latency and persistence. Marburg virus (MARV) has also emerged in the region with known bat reservoirs and human spillover events. Serology offers a tool to disentangle the ecological dynamics and human exposure history to filoviruses, but cross-reactivity across filoviridae necessitates multi-faceted, well-characterised assays. This study aimed to assess zoonotic exposure to EBOV and related filoviruses among bushmeat hunters and their household members in Macenta Prefecture, Guinea, near historical EBOV index sites, using stepwise serological analyses and to explore environmental correlates of exposure.

Prior investigations suggest bats as likely reservoirs for ebolaviruses, supported by detection of viral RNA/antibodies in diverse bat species and the identification of BOMV in West Africa. Historical seroprevalence studies across Africa reported variable exposure estimates, influenced by differences in assay choice and analysis, and complicated by cross-reactivity among filoviruses. Longitudinal studies of EBOV survivors indicate long-lived, slowly waning IgG responses and durable neutralisation that can persist for decades, enabling retrospective exposure assessment. Cross-reactivity is well documented across EBOV glycoproteins and other antigens (NP, VP40), and MARV sera can cross-react with some EBOV antigens without neutralisation. Evidence of EBOV RNA fragments in local bats, BOMV in Mops condylurus, and recent MARV detection and human case in the region underscore the plausibility of repeated zoonotic exposures in Guinée Forestière and complicate attribution of serological profiles to specific filovirus species.

Design and setting: Cross-sectional seroepidemiological study conducted February–December 2017 in Toma-speaking villages of Macenta Prefecture, Guinea, purposively sampling both villages affected and unaffected by the 2013–16 EBOV epidemic. Participants were bushmeat hunters and their closest household relative (typically spouse). Comparative sera were obtained from PCR-confirmed EBOV Makona survivors and their contacts (2017 sampling) from Guéckédou and Coyah prefectures. Ethics: Approvals from National Ethics Committee for Health Research, Guinea (No. 33/CNERS/15) and UK National Research Ethics Service; additional approval for survivor cohort (N°012/CENRS/2017). Sampling and processing: 5 ml peripheral blood collected; serum separated (2000 g, 10 min) and stored at −20 °C. Serological assays:

• ELISA: Indirect ELISA targeting EBOV GP (Makona) on 96-well plates (antigen 1 µg/ml), serial plasma dilutions starting 1:50; IgG detected with goat anti-human IgG-AP; OD405; 4-parameter logistic fit; titres expressed in IU/ml. Latent profile analysis (LPA; Gaussian finite mixture) classified responses into low, intermediate, and high titre groups.
• Western blot (WB): Recombinant EBOV GP (Makona), NP (partial, Mayinga; aa 488–739), VP40 (Mayinga). SDS-PAGE, transfer to PVDF, blocking, incubation with plasma (1:1000), detection with peroxidase-conjugated anti-human IgG; chemiluminescent development; qualitative binding to GP, NP, and VP40 recorded.
• Live virus neutralisation: BSL-4 microneutralisation of EBOV Mayinga (100 TCID50). Heat-inactivated sera serially diluted; Vero cells added; CPE read at day 7; neutralisation geometric mean titres (GMT) from four replicates.
• Latent class analysis (LCA): Combined outcomes from GP-ELISA and WB (GP, NP, VP40) to classify serological profiles into two latent classes.
• Multiplex microsphere immunoassay (MMIA; Luminex): Measured total binding IgG to GP from EBOV, BDBV, BOMV, RESTV, SUDV, LLOV, MLAV, RAVV. Recombinant GP ectodomains (trimerised with GCN3) expressed in 293-F cells, purified, coupled to magnetic microspheres. Sera diluted 1:400; detection with biotinylated anti-human IgG and streptavidin-PE; MFI reported. Positivity thresholds not defined due to lack of suitable controls and extensive cross-reactivity. Environmental and statistical analyses: Land cover (2017, 100 m resolution) classified into closed forest (>70% canopy), open forest, shrubs, herbaceous, urban. Fragmentation indices (fractal dimension, perimeter-area ratio, shape index) computed within buffers (0.5–20 km) around village centroids. Mixed-effects models (random intercept for village) assessed associations between serological outcomes (continuous ELISA titre; binomial group A; binomial high ELISA) and demographic/environmental predictors. Variable selection addressed collinearity (Pearson r ≥ 0.75) with AIC-guided stepwise selection. Spatial autocorrelation of residuals assessed with Moran’s I. Analyses in R 4.0.4 (mclust, poLCA, spdep).

• Participants: 499 recruited from 38 villages; viable sera from 498. Fourteen villages (36.8%) had PCR-confirmed EBOV cases during 2013–16; 24 had none by records and key-informant verification. Median age 39 years; 280 males (43.9%); 276 hunters (55.3%).
• ELISA (EBOV GP): Latent profile analysis identified three titre groups: low (n=206), intermediate (n=278), high (n=14). High-titre Macenta group exhibited anti-GP levels comparable to PCR-confirmed EVD survivors.
• Western blot (subsample n=115; high n=14, intermediate n=101): Binding observed to NP in 31/115 (27.0%), VP40 in 16/115 (13.9%), GP in 12/115 (10.4%). Any-target binding in 40/115 (34.8%); multi-target binding in 17/115 (14.8%) across combinations (e.g., GP+NP+VP40 in 2/115). Proportion with any WB positivity was higher in the high ELISA group (11/14, 78.6%) vs intermediate (29/101, 28.7%; p=0.0007).
• Latent class analysis (combined ELISA+WB): Two classes identified. Group A (n=20) showed high GP-ELISA with multi-target WB; Group B (n=95) showed negative or single-target (often NP-only) WB with intermediate ELISA.
• Live EBOV neutralisation (n=62 tested: all group A and 42/95 group B): Strong neutralisation in 5/62 (8.1%; serum dilutions 1:152–1:1218), low in 5/62 (1:10–1:16), absent in 52/62 (<1:10). All strong neutralisers were in group A. Neutralisation correlated nonlinearly with GP-ELISA titre (cubic spline p=0.004 vs linear).
• Multiplex GP binding (MMIA; group A n=20): All neutralisation-positive sera (n=5) showed highest IgG to EBOV GP with strong cross-binding to BDBV GP; cross-reactivity decreased with phylogenetic distance. The remaining 15 showed no appreciable binding to the GP panel at levels similar to EBOV GP; no strong binding to other ebolavirus GPs (e.g., TAFV, SUDV) was detected.
• Spatial epidemiology: High GP-ELISA, group A membership, and strong neutralisation were dispersed across sites with low counts per village; no residual spatial autocorrelation (Moran’s I approximately −0.0032 to 0.0076). All response types occurred in both previously affected and unaffected villages; high GP-ELISA more frequent in affected villages, while group A and strong neutralisation were more evenly distributed.
• Environmental associations: Higher anti-GP titres were inversely associated with fragmentation of closed canopy forest. In multivariable mixed-effects linear regression, the shape index (500 m buffer) for closed forest was negatively associated with log2 anti-GP titres (estimate −0.50; 95% CI −0.95 to −0.06; p=0.03). Similar inverse relationships were found when using group A (binomial) and when restricting to unaffected villages; patterns persisted for high GP-ELISA outcomes with metric/scale differences.
• Overall serological signature prevalence: 20/498 (4.0%) exhibited multi-antigen EBOV responses (group A); 5/20 (25%) had strong live-virus neutralisation consistent with prior EBOV-Makona infection.

The study addressed whether bushmeat hunters and their households in Macenta, Guinea, exhibit serological evidence of zoonotic exposure to EBOV or related filoviruses. A distinct subset (4.0%) displayed multi-antigen EBOV reactivity, including five individuals with strong EBOV neutralisation and GP titres comparable to PCR-confirmed survivors, suggesting prior EBOV-Makona infection—potentially including under-reported cases during 2013–16. However, the larger subset of non-neutralising but multi-antigen responders (n=15) is unlikely to be fully explained by symptomatic or asymptomatic EBOV-Makona infections given the expected high prevalence and persistence of neutralisation in survivors and the observed ratios when accounting for CFRs. Instead, these heterogeneous seroprofiles are consistent with exposure to phylogenetically related filoviruses circulating in the region, as supported by bat detections of EBOV RNA fragments, BOMV genomes, and MARV isolates, and by known cross-reactivity among filoviridae antigens. Spatial dispersion of seropositive phenotypes across villages without residual autocorrelation and associations with intact closed canopy forest align with expectations for sporadic zoonotic spillover and stuttering transmission in overdispersed pathogens. The ecological findings emphasise the role of relatively undisturbed forest habitats in human exposure risk at the wildlife–human interface. Together, the results suggest that both EBOV and non-EBOV filoviruses have contributed to historical and possibly ongoing spillover exposures in Guinée Forestière, underscoring the need for targeted surveillance that integrates serological, ecological, and genomic data in high-risk interfaces like bushmeat hunting communities.

A multi-stage serological approach revealed that 4% of bushmeat hunters and household contacts in Macenta, Guinea, exhibited EBOV-directed immune signatures, with a subset showing strong neutralisation consistent with prior EBOV-Makona infection and others showing multi-antigen, non-neutralising profiles indicative of exposure to related filoviruses. Serological outcomes were spatially dispersed and inversely associated with fragmentation of closed canopy forest, supporting a zoonotic spillover context. These findings provide evidence for unrecognised exposures to EBOV and other filoviruses in the region prior to and possibly after the 2013–2016 epidemic and highlight priorities for surveillance. Future directions include: longitudinal follow-up to assess durability of diverse seroprofiles; expanded, representative community sampling beyond hunter households; incorporation of species-specific and pan-filovirus neutralisation assays; targeted wildlife surveillance in intact forests; and integration of serological, genomic, and environmental data to refine risk mapping and pre-empt outbreak emergence in high-risk locales.

• Sub-sampling for downstream WB and neutralisation focused on intermediate and high ELISA groups may have missed additional responders; the true size of group A could be underestimated.
• Purposive sampling of bushmeat hunters and their households within a limited geography reduces generalisability to broader populations.
• Luminex MMIA lacked population-specific positivity thresholds due to absence of negative and single-infection control sera and extensive cross-reactivity among filoviruses.
• Disentangling exposure aetiology is challenging in settings with prior resistance to public health responses and persistent stigma, increasing uncertainty around undocumented cases.
• Cross-reactivity among filoviridae complicates precise attribution of exposures to specific virus species in the absence of paired genomic data.