We captured and sampled 276 bats (141 Rousettus leschenaultii bats, 75 Cynopterus spp. bats, 59 Megaderma lyra bats, and 1 Macroglossus sobrinus bat) during April 2010–March 2011 from the Faridpur, Rajbari, Lalmonirhat, and Comilla Districts in Bangladesh. All bats were identified to species in the field, except Cynopterus spp. bats, because of cryptic diversity in this group; we are awaiting genetic species confirmation. Bats were captured in mist nets near roosts or at feeding sites and were handled in accordance with the Tufts University (Medford, MA, USA) Institutional Animal Care and Use Committee protocol (no. G2011-106).

We collected 50–800 µL of blood from brachial or cephalic veins of each bat, and diluted it 1:4 with phosphate-buffered saline in the field before serum was separated, as described (9). We also collected throat, urine/urogenital, and fecal swab specimens, which were placed in 750 µL of NucliSENS lysis buffer (bioMérieux, Marcy l’Etoile, France). All samples were collected in cryovials, placed in liquid nitrogen in the field, and maintained at −80°C until testing. We recorded morphologic measurements, weight, sex, age, and body condition and collected a wing biopsy specimen before releasing animals at capture sites.

We screened serum samples for IgG against REBOV and ZEBOV by using ELISA and Western blotting at the Commonwealth Scientific and Industrial Research Organisation Australian Animal Health Laboratory Biocontainment Facility (Geelong, Victoria, Australia). To inactivate potentially infectious agents, serum samples were heated at 56°C for 20 min before shipment. All samples were screened by using a 1:1 mixture of purified recombinant nucleoproteins (0.2 mg/mL) of REBOV and ZEBOV (R + Z ELISA), which were expressed in an Escherichia coli vector that contained a histidine tag (10,11).

Potentially positive serum cutoff values were determined to be >0.454 for the R + Z ELISA by using maximum-likelihood estimation, gamma distribution, and 95% risk for error (7). Potentially positive serum samples were tested by ELISA against each nucleoprotein independently to confirm reactivity and by Western blotting against nucleoproteins of Reston and Zaire virus strains as described (10). Serum samples were tested at a dilution of 1:50. Endpoint titrations with an optical density >3× the background reading were determined for serum samples positive against REBOV and ZEBOV antigens individually.

Total nucleic acids were extracted from samples (urine/urogenital, fecal, and throat swab specimens) by using the easyMAG NucliSENS platform (bioMérieux) at Columbia University (New York, NY, USA). Samples were tested for filovirus RNA (RNA polymerase gene) by using a consensus PCR protocol validated to amplify 19 diverse filovirus strains. This PCR has a sensitivity of 50–500 RNA copies with synthetic transcripts and has been further validated with blood samples (12).

Fifteen (11%) of 141 R. lescehnaulti, 6 (8%) of 75 Cynopterus spp., and 4 (7%) of 56 M. lyra bats were potentially positive after initial screening. Five (3.5%) of 141 (95% CI 1.5%–8.0%) R. leschenaultii bats were confirmed as seropositive after testing by ELISAs and Western blotting (Table 1). Bats were sampled during the breeding season; 21 (62%) of 34 sampled female R. leschenaultii bats were pregnant and 8 (23%) of 34 carried pups. We sampled 3× as many males as females; all 5 confirmed virus-positive animals were healthy adult males (Table 2). All 698 throat, urine/urogenital, and fecal samples were virus negative by PCR (Table 2). All confirmed seropositive samples except 1 (April 2010–042) reacted more strongly to Zaire virus antigens than Reston virus antigens (Table 1). Similarly, 2 samples (April 2010–057 and SB0311–059) showed higher reactivity to ZEBOV by Western blotting, and other samples were equally reactive to REBOV.

Our study provides evidence of Ebola virus infection in wildlife from mainland Asia and corroborates the observation that filoviruses are harbored across a much larger geographic range then assumed (2). Preferential reactivity to ZEBOV suggests exposure to an Ebola virus that is distinct from REBOV, the only filovirus currently found in Asia. We consider the likelihood of cross-reactivity with MARV as low because there is only a 35% aa identity between nucleoprotein genes of REBOV/ZEBOV and MARV. However, we cannot rule out co-infection with multiple filoviruses.

Seroprevalence found in this study is consistent with that found in another study (4). However, other studies of Rousettus spp. bats have reported higher values (e.g., 7%–20% and 8% of R. aegyptiacus bats seropositive for MARV and ZEBOV, respectively) (6,7), and 5 (31%) of 16 R. amplexicaudatus bats seropositive for REBOV (8). These differences might have been caused by poor specificity of the assay if this virus is novel, an artifact of low volume of blood collected, the potential that other species may have greater roles as reservoirs than Rousettus spp. in Bangladesh, or timing of sampling. R. leschenaultii bats have a large range (China to India) (13); and more detailed studies of virus ecology and diversity are warranted to better understand their role as a potential reservoir of zoonotic disease agents.

We demonstrated that serologic and virus surveys of bats can be informative for identifying potential virus hosts. Previous studies amplified ZEBOV nucleic acid from bat feces (14). We also screened bat feces to identify potential routes of virus excretion, which is useful when the route of exposure from bats to humans is known. A short interval for Ebola virus shedding by reservoir hosts and an inverse relationship between viremia and antivirus titer probably explain our negative PCR results for seropositive bats. Failure to detect filovirus nucleic acid might reflect our relatively small sample size, low virus prevalence, or use of a PCR that has low sensitivity for filoviruses circulating in Bangladesh.

In Bangladesh, human outbreaks of Nipah virus have been linked to drinking date palm sap contaminated with bat excreta, presumably from Pteropus giganteus bats (15). R. leschenaultii bats and other small fruit bat species visit date palm trees 10× more frequently than Pteropus spp. bats (15). This finding could indicate potential transmission of filoviruses or any other novel viruses that R. leschenaultii bats carry. It also highlights the need for more research to understand this ecologic system and for better implementation of low-cost barriers to reduce bat–human contact during periods of date palm harvesting (15).