During 2011–2016, we collected serum from bats of the 3 aforementioned species in Singapore and screened samples for evidence of exposure to filoviruses. Samples were collected with permission from the National University of Singapore Institutional Animal Care and Use Committee (B01/12) and the National Parks Board (NP/RP11–011–3a). We diluted venous blood 1:10 in phosphate-buffered saline and then centrifuged, recovered, and heat-inactivated the serum at 56°C for 30 minutes and stored it at −80°C.

We developed a Bio-Plex (Bio-Rad, Hercules, CA, USA) bead-based multiplex assay that simultaneously probes serum for immunoglobulins specific to the viral envelope glycoproteins (GPs) from representative strains of all described Ebolavirus and Marburgvirus species (Table 1). A human FreeStyle 293-F stable cell-line expression system was used to produce the Ebolavirus and Marburgvirus spp. GPs as a soluble GP consisting of the entire ectodomain, sGP_(1,2), which retains a native-like oligomeric conformation, as described previously with modifications (10). In brief, each GP_(1,2) coding sequence was truncated at the C-terminus to remove the predicted transmembrane domain and cytoplasmic tail, then appended with the GCN trimerization peptide sequence (10) together with a factor Xa protease cleave site and a Twin-Strep-tag sequence (IBA Lifesciences, Göttingen, Germany). The sGP_(1,2) proteins were produced in serum-free conditions and purified by Strep-Tactin XT technology (IBA Lifesciences). The Twin-Strep-tag was removed by factor Xa enzymatic cleavage; factor Xa was removed by Xarrest Agarose (Merck Millipore, Billerica, MA, USA); sGP_(1,2) was purified further by S-200 size exclusion chromatography, concentrated, and stored frozen. These sGP_(1,2)s were coupled to carboxylated beads (Bio-Rad). Screening was performed on a Bio-Rad Bio-Plex 200.

In the absence of confirmed filovirus-negative bat serum, we used methods developed by Peel et al. to establish a median fluorescence intensity (MFI) cutoff value (11). We confirmed a cutoff value of 200 MFI (Technical Appendix), as was previously used for Eidolon helvum bat serum in a Bio-Plex serologic assay (12). We screened 409 samples with our Ebolavirus and Marburgvirus spp. sGP_(1,2) Bio-Plex assay modified from that described by Bossart et al. (13). Samples were diluted 1:100 and tested in duplicate; the sGP_(1,2)-coupled beads were mixed with individual samples; and a 1:1 combination of recombinant biotinylated-protein A/protein G (1:500) (Pierce, Rockford, IL, USA) was added to the wells, followed by addition of streptavidin-phycoerythrin (1:1,000) (Bio-Rad) and determination of MFI.

Figure 1

Figure 1. Mean fluorescence intensity (MFI) values obtained from Bio-Plex assay (Bio-Rad, Hercules, CA, USA) screening of individual serum samples from bats of 3 species with soluble filovirus glycoproteins. Dashed line indicates the...

Samples were positive for 17 (9.1%) of 186 E. spelaea, 13 (8.5%) of 153 C. brachyotis, and 3 (4.3%) of 70 P. lucasi bats (Figure 1). Positive samples reacted with EBOV, BDBV, SUDV, or TAFV sGP_(1,2). However, no samples were positive for RESTV, MARV, or RAVV sGP_(1,2). We further examined positive samples to determine cross-reactivity between the Ebolavirus spp. sGP_(1,2) (Table 2). Twelve (71%) samples from E. spelaea bats cross-reacted with >2 Ebolavirus spp. sGP_(1,2) (BDBV, EBOV, SUDV, or TAFV). In contrast, 8 (62%) C. brachyotis and 2 (66%) P. lucasi samples were positive for only 1 sGP_(1,2) (BDBV or SUDV).

Figure 2

Figure 2. Western blot results of individual bat serum samples probed against Zaire ebolavirus and Bundibugyo ebolavirus glycoproteins 1 and 2 (GP₁, GP₂). Boldface indicates positivity by Western blot and underlining indicates positivity...

To further determine the cross-reactivity of positive samples and to corroborate Bio-Plex assay results for a selected number of samples, we performed Western blot (WB) assays (Figure 2). The filovirus GP_(1,2) is a trimer of heterodimeric GP₁ and GP₂ subunits. The trimeric-like sGP_(1,2) is the antigen in the multiplex Bio-Plex assay, whereas linearized monomeric sGP₁ and sGP₂ subunits are the antigens in WBs. Reduced and denatured EBOV or BDBV unconjugated sGP_(1,2) was loaded on 8% sodium dodecyl sulfate–polyacrylamide electrophoresis gels, transferred to a polyvinylidene difluoride membrane, and probed with 1:100 dilutions of positive and negative bat serum, as previously determined by the Bio-Plex assay. All 3 E. spelaea bat samples and 2 of 3 C. brachyotis bat samples that were Bio-Plex positive were also positive by WB and displayed reactivity with EBOV and BDBV GP₁ and GP₂ antigens; no P. lucasi bat samples positive by Bio-Plex were positive by WB.

We present evidence of antibodies specific to filoviruses antigenically related to Ebolavirus spp. in 3 species of fruit bats widely distributed throughout Southeast Asia. We detected seroreactivity with Ebolavirus spp. but not Marburgvirus spp. GP. Despite the close relatedness of the viruses, we detected samples reacting with only SUDV, not RESTV, GP. This finding contrasts with previous reports of bat serum cross-reactivity with RESTV nucleoprotein (5,7,14). Possible explanations include 1) the fact that our customized Bio-Plex assay is based on conformational sGP_(1,2), which can differentiate antibody specificity better than the more sequence conserved nucleoprotein, and 2) the lack of evidence of RESTV GP positivity with Cynopterus and Eonycteris bat serum samples, which is in line with previous findings (both species were negative while only Rousettus amplexicaudatus bats were positive) (7). E. spelaea bats were previously predicted to be filovirus hosts (15), and sequences of novel filoviruses have been discovered in E. spelaea bat populations in Yunnan, China (14). Our data provide additional empirical evidence that populations of C. brachyotis, E. spelaea, and P. lucasi bats in Southeast Asia are hosts of filoviruses, which seem antigenically more closely related to EBOV, BDBV, and SUDV than to RESTV.

Examination of cross-reactivity of positive samples from E. spelaea, C. brachyotis, and P. lucasi bats revealed no clear patterns of preferential reactivity with EBOV, BDBV, or SUDV GP. Factors that might contribute to the lack of P. lucasi positivity by WB include sensitivity differences between Bio-Plex and WB assays paired with the change in sGP_(1,2) conformation. Two Bio-Plex EBOV-positive samples (E. spelaea samples 0805149 and 011603) reacted with EBOV sGP₂ and BDBV sGP₁ in the WB. Bio-Plex and WB data strongly suggest the presence of yet-undetected batborne filoviruses, which are antigenically related to but distinct from BDBV, EBOV, and SUDV circulating in local bat populations. Reasons why these filoviruses have remained undetected include their inability to cross the species barrier, the rarity of spillovers into humans or domestic animals, or the fact that spillover events cause mild or no disease. We suggest that a yet-undescribed diversity of filoviruses exists in Southeast Asia bat populations, a hypothesis supported by the recent identification of filovirus sequences in E. spelaea and R. leschenaulti bats in China (14,16). Comprehensive surveillance including serology and detection of viral nucleic acid, along with virus isolation, will help elucidate the characteristics of filoviruses endemic to Asia and identify bat species that function as maintenance populations and reservoirs.