Experimental Inoculation of Egyptian Fruit Bats (Rousettus aegyptiacus) with Ebola Virus

Colonized Egyptian fruit bats (Rousettus aegyptiacus), originating in South Africa, were inoculated subcutaneously with Ebola virus (EBOV). No overt signs of morbidity, mortality, or gross lesions were noted. Bats seroconverted by Day 10–16 post inoculation (p.i.), with the highest mean anti-EBOV IgG level on Day 28 p.i. EBOV RNA was detected in blood from one bat. In 16 other tissues tested, viral RNA distribution was limited and at very low levels. No seroconversion could be demonstrated in any of the control bats up to 28 days after in-contact exposure to subcutaneously-inoculated bats. The control bats were subsequently inoculated intraperitoneally, and intramuscularly with the same dose of EBOV. No mortality, morbidity or gross pathology was observed in these bats. Kinetics of immune response was similar to that in subcutaneously-inoculated bats. Viral RNA was more widely disseminated to multiple tissues and detectable in a higher proportion of individuals, but consistently at very low levels. Irrespective of the route of inoculation, no virus was isolated from tissues which tested positive for EBOV RNA. Viral RNA was not detected in oral, nasal, ocular, vaginal, penile and rectal swabs from any of the experimental groups.


Introduction
The family Filoviridae comprises the Ebolavirus, Marburgvirus, and Cuevavirus genera. The genus Ebolavirus includes five species, each represented by a single virus member: Sudan ebolavirus (Sudan virus, SUDV), Zaire ebolavirus (Ebola virus, EBOV), Bundibugyo ebolavirus (Bundibugyo virus, BDBV), Taï Forest ebolavirus (Taï Forest virus, TAFV), and Reston ebolavirus (Reston virus, RESTV). With the exception of RESTV and Lloviu virus (LLOV, species Lloviucuevavirus), all known filoviruses cause severe hemorrhagic fever in humans [1]. The unprecedented scale of Ebola virus disease (EVD) outbreaks in West Africa in 2013-2015 caused by EBOV [2,3] represent a dramatic expansion of case numbers and introduction of this highly lethal virus into new geographic areas, a frightening account for the prospect of EBOV turning into a formidable public health threat globally. Despite the increase in EVD outbreaks in recent years [4], the natural reservoirs of filoviruses remain elusive [5,6].
Circumstantial evidence for an association between ebolaviruses and bats was reported in the first SUDV outbreak in 1976 [7] and in the TAFV outbreak in chimpanzees in Ivory Coast in 1994 [8].

Virus
The SPU 220/96 isolate of EBOV (4th passage in Vero cells) used to inoculate bats was isolated from the serum of a nurse who contracted a fatal infection from a Gabonese physician admitted to a private hospital in South Africa in 1996 [29].

Experimental Inoculations
Thirty-six EBOV-seronegative bats, aged 4-24 months, were used during the first stage of the experiment. Twenty-four bats were inoculated subcutaneously (s.c.) with 100 µL of tissue culture supernatant containing 10 5.0 50% tissue culture infectious doses/mL (TCID 50 /mL) of EBOV, and 12 control bats were mock-inoculated s.c. with 100 µL of Eagles Minimal Essential Medium (EMEM). Subcutaneous inoculation was administered into the loose skin over the shoulders. Animals were split up into 6 cages, each containing 4 EBOV-infected and 2 in-contact control bats. Sampling of blood and tissues was carried out as previously described [24]. The sampling framework from s.c. inoculated bats is given in Table 1. Blood was collected from in-contact control bats on Days 0, 14, 21 and 28 post inoculation (p.i.).     During the second stage of the experiment, 11 bats that remained seronegative after 28 days in-contact exposure to EBOV-infected bats were inoculated intraperitoneally (i.p.; n = 6), and intramuscularly (i.m; n = 5) with the same dose of EBOV as for s.c. inoculation. Intraperitoneal inoculation was administered in the lower right quadrant of the animal's abdomen, close to the midline above the genitalia. Intramuscular inoculation was given into caudal thigh muscle. Sampling of blood and other tissues from i.p. and i.m. inoculated bats is given in Table 2. Oral, nasal, vaginal, penile and rectal swabs were collected from all the EBOV-inoculated groups at regular intervals (Table 3). Animals were anaesthetized prior to inoculation and specimen collection as previously described [24]. Bats were monitored daily for the development of clinical signs as well as for food intake. Post mortem tissues were processed as previously described [24].  Oral

Serology
Sera from virus and mock inoculated animals were tested by an indirect IgG ELISA, as described in Paweska et al. [24], modified to use recombinant EBOV glycoprotein (GP) (Integrated BioTherapeutics, Gaithersburg, MD) antigen (3.1 mg/mL) diluted 1:6000 in PBS pH 7.2, and positive control serum from the Wahlberg's epauletted fruit bat (Epomophorus wahlbergi) experimentally infected with EBOV [9]. The mean OD ELISA readings were converted to a percentage of positive (PP) control serum as previously described [30].

Real-Time Quantitative RT-PCR (Q-RT-PCR)
Processing of blood, tissues, and swabs for quantitative reverse-transcription PCR (Q-RT-PCR) targeting the L polymerase gene of EBOV [31] was performed as previously described [24]. Primers Filo A2.4 and Filo B were used according to the protocol published by Panning et al. [31] except for using a 0.1µM concentration of the FAMEBOg probe. Samples with C t values ď40 were regarded as positive.
A quantification standard was produced by cloning the EBOV L gene PCR target region into a pCRII-TOPO expression vector (Invitrogen) as described in Paweska et al. [24]. Other internal controls incorporated an extraction control and a template-free control. Copy numbers of EBOV RNA detected per reaction volume were converted to copy numbers per milliliter of plasma or gram of tissue. For converting RNA copy numbers detected in samples to a TCID 50 , a logarithmic titration curve using stock EBOV SPU 220/96 isolate was generated as previously described [24].

Virus Isolation
Processing of specimens for virus isolation was performed as described in Paweska et al. [24]. Virus isolation was attempted on all Q-RT-PCR positive blood and tissues. Cultures were tested for EBOV replication by Q-RT-PCR using tissue culture fluids collected at the time of first visible cytopathic effect (CPE) or at 14 days p.i.

Statistical Analysis
The cut-off value for the ELISA was determined as two times the mean plus 3-fold standard deviations (SD) of ELISA PP values recorded in 36 experimental animals before inoculation.

Results
The ELISA cut-off value was determined as 36.7 PP. Q-RT-PCR performed on RNA extracts from log 10 dilutions of the challenge EBOV in EMEM yielded a linear correlation between RNA copy numbers and TCID 50 (R 2 = 0.9819). One TCID 50 /mL was equivalent to 5ˆ10 3 EBOV viral RNA per mL of blood or g of tissue, corresponding to Q-RT-PCR Ct value of 35. Extracts of RNA from log 10 dilutions of EBOV below one TCID 50 /mL were all negative (Ct values >40, results not shown).
All bats remained clinically well and maintained normal food uptake; no gross pathology was detected. None of the bats sustained apparent skin or mucous wounds or injury for the duration of the experiments.
In EBOV-inoculated bats, the first seroconversion was detected in one out of three bats tested on Day 10 p.i. On Day 14 p.i, all s.c. inoculated animals had seroconverted. The highest mean IgG antibody level was recorded on Day 28 p.i., followed by a slight decrease by Day 37 after s.c. inoculation (Figure 1). Seroconversion was not detected in any of the 12 in-contact bats. In i.p. and i.m. inoculated bats, seroconversion was detected in all but one i.p. inoculated bat on Day 16 p.i. (Figure 2).

Results
The ELISA cut-off value was determined as 36.7 PP. Q-RT-PCR performed on RNA extracts from log10 dilutions of the challenge EBOV in EMEM yielded a linear correlation between RNA copy numbers and TCID50 (R 2 = 0.9819). One TCID50/mL was equivalent to 5 × 10 3 EBOV viral RNA per mL of blood or g of tissue, corresponding to Q-RT-PCR Ct value of 35. Extracts of RNA from log10 dilutions of EBOV below one TCID50/mL were all negative (Ct values >40, results not shown).
All bats remained clinically well and maintained normal food uptake; no gross pathology was detected. None of the bats sustained apparent skin or mucous wounds or injury for the duration of the experiments.
In EBOV-inoculated bats, the first seroconversion was detected in one out of three bats tested on Day 10 p.i. On Day 14 p.i, all s.c. inoculated animals had seroconverted. The highest mean IgG antibody level was recorded on Day 28 p.i., followed by a slight decrease by Day 37 after s.c. inoculation (Figure 1). Seroconversion was not detected in any of the 12 in-contact bats. In i.p. and i.m. inoculated bats, seroconversion was detected in all but one i.p. inoculated bat on Day 16 p.i. (Figure 2).      Except for one bat (No. 106) on Day 3 p.i., the blood of the remaining s.c. inoculated bats tested on Days 3, 5, 7 and 10 p.i. were all qRT-PCR negative (Table 1). EBOV RNA was detected in the blood of i.p. and i.m. inoculated bats on Day 7 p.i., and one out of two i.m. inoculated bats on Day 16 p.i. (Table 2).
Of the 16 other tissues from s.c. inoculated bats, EBOV RNA was detected in the liver and the lung in a single bat on Day 7 p.i. (Table 1). In the i.p. inoculated bats, EBOV was detected in the stomach of two bats and the skin of one bat on Day 5 p.i., and in the spleen and kidney of one bat on Day 7 p.i. In i.m. inoculated bats, EBOV RNA was detected in the stomach of one of two bats on Day 5 p.i., in the liver of one bat on Day 7 p.i., and in the spleen in one of two bats on Day 16 p.i. The Ct values of the Q-RT-PCR ranged from 35.08 to 38.21 in blood, and from 35.88 to 40 in the other tissues, an equivalent of less than 1 TCID 50 /mL of plasma, or 1 TCID50/g of tissue. Irrespective of the inoculation route, all EBOV RNA-positive blood samples and tissues were negative by virus isolation (Tables 1 and 2). Viral RNA was not detected in any of oral, nasal, vaginal, penile or rectal swabs collected from EBOV-inoculated bats on Days 3-10 p.i. (Table 3).

Discussion
Jones et al. [28] reported the experimental infection of groups of four captive-bred, juvenile Egyptian fruit bats inoculated s.c. with each of the five known ebolaviruses. Irrespective of the virus used, no bat became viremic, there was no evidence for viral shedding, and tissue distribution of viral RNA was limited in all but SUDV-inoculated bats. In a follow-up experiment [28] involving 15 bats inoculated s.c. with SUDV, viral RNA was detected in 8 out of 10 different tissues, but in the absence of detectable viremia or viral shedding. The only tissues in which SUDV levels were high enough to suggest possible viral replication were the inoculation site, at Days 3 and 6 p.i., and the spleen in one bat, on Day3 p.i. However, of the 15 bats, viral RNA was detected only in one tissue in eight individuals (skin or axillary lymph node), and in one bat in two tissues (skin and liver). These results seem to indicate notable differences in individual response to s.c. inoculation with SUDV. In contrast, s.c inoculation of R. aegyptiacus with MARV consistently results in systemic infection, including relatively long viremia, viral replication in multiple tissues, and detection of viral RNA in oral and rectal swabs [24][25][26]28].
We expanded the pilot EBOV experimental work by Jones et al. [28] using colonized R. aegyptiacus bats originating in South Africa in a larger serial euthanasia study. This is the first experimental infection study attempting in-contact transmission of EBOV in bats, as well as comparing tissue tropism, viral shedding, and clinical and gross pathological effects in R. aegyptiacus exposed to EBOV by s.c, i.p. and i.m. routes. We studied the infection kinetics in 16 different tissues from Days 3-37 and in the blood from Days 3-10 after s.c. inoculation with EBOV. In our study, low levels of EBOV RNA was detected in 2 out of 24 bats (0.83%) on Days 3 (blood) and 7 (liver and lung) p.i., with negative Q-RT-PCR results in all tissues sampled on Days 21 and 37 p.i. There was no evidence for viral shedding as demonstrated by negative Q-RT-PCR results in different swabs and lack of horizontal transmission of the virus from EBOV-inoculated to in-contact control bats. Although in our study urine samples were not collected, during handling of animals they consistently urinated, thus vaginal and penile swabs tested were inherently contaminated with urine. None of the immunologically privileged sites tested (brain, testis) were PCR-positive. Our findings strongly suggest that s.c. inoculation of R. aegyptiacus with EBOV does not result in acute or chronic infection. It is unlikely that EBOV can spread among caged bats in the absence of detectable viral replication and shedding. However, it should be taken into account that horizontal virus transmission may be intermittent and slower, and subjected to external factors, including stresses, pregnancy, estrous, and starvation.
It appears that compared to s.c. inoculation, exposure of R. aegyptiacus to EBOV by the i.p. and i.m. routes results in wider tissue distribution of viral RNA and detection in a higher proportion of inoculated bats. EBOV RNA was detected in two out of six (33.3%) i.p. inoculated, and in two out of five (40%) i.m. inoculated bats. In this context, i.m inoculation mimics a possible natural exposure route for virus transmission, e.g., via bite. Irrespective of the route of inoculation and the type of tissues tested in this study, the recorded Ct values were consistently very high (>35), indicating extremely poor replication of EBOV in the major tissues of R. aegyptiacus bats. This has been substantiated further by failure to isolate virus from Q-RT-PCR-positive blood and tissue samples. A direct correlation between Marburg virus RNA levels (viral load) in R. aegyptiacus determined by Q-RT-PCR and the ability to isolate virus was also shown by Towner et al. [21]. Although replication of EBOV in R. aegyptiacus below the detection limit in Vero cells seems to be unlikely, passaging of blood and tissue homogenates which tested low positive by Q-RT-PCR is worth considering in future studies.
In the absence of any convincing laboratory evidence for EBOV replication, the easily detectable and relatively strong anti-EBOV IgG responses in bats from all experimental groups are rather intriguing. In the study by Jones et al. [28], terminated at Day 10 p.i., no seroconversion was detected against any of the five ebolaviruses. However, in bats inoculated with SUDV in a serial euthanasia study terminated Day 15 p.i., one out of six surviving bats seroconverted on Day10 p.i., and two out of three had seroconverted on Day 15 p.i. Our serological results are similar, but we also demonstrated that bats inoculated s.c. with EBOV can maintain high antibody levels up to Day 37 p.i. The kinetics of anti-EBOV IgG responses in i.p. and i.m. inoculated bats seem to be similar. Our serological results and those from Jones et al. [28] could be explained by an initial immune response to the viral antigen present in the inoculums and/or to viral antigens of low-level replicating virus at the inoculation site. SUDV viral RNA loads were shown to be highly suggestive of viral replication at the inoculation site where viral antigen was also demonstrated in small numbers of macrophages in the deep subcutis. In one instance the SUDV antigen was also present in an axillary lymph node [28]. We could detect the EBOV RNA in the skin of only one bat but in our study skin specimens were taken from the ventral thorax, not from the dorsal inoculation site.
Interestingly, the in vivo findings by Jones et al. [28] and our study contrast with the in vitro study by Krähling et al. [27] who showed replication of EBOV in an Egyptian fruit bat-derived cell line similar to the replication kinetics observed in Vero E6 cells [27], and morphological changes similar to those noted in EBOV-infected cell cultures originating from monkeys and humans. Despite efficient replication of EBOV in the immortalized fetal cells from the Egyptian fruit bat, the virus did not replicate efficiently in vivo. Thus, caution should be used when inferring host status of certain species based on in vitro results. Our larger serial euthanasia study confirms earlier results in smaller experimental group of bats [28] that R. aegyptiacus is refractory to infection with EBOV. Experimental inoculation was, however, sufficient to induce easily detectable antibody. A similar situation might occur in nature. Thus, serological findings in experimentally inoculated bats are important for the interpretation of epidemiological significance in wild-caught bats. Firstly, it demonstrates that it is impossible to identify bat species as filovirus reservoirs entirely based on seropositivity. On the other hand, the identification of EBOV-seropositive Egyptian rousettes in Gabon [11,12] might indicate their contact with another bat species or other hosts which are infected with and shedding the virus. Communal feeding by different fruit bat species or other mammals on the same fruit-bearing tree might represent a possible route of incidental exposure of Egyptian fruit bats to the actual reservoir species. Natural exposure of R. aegyptiacus to EBOV could occur during fighting with infected bats of different species or inoculation of the virus by bite of infected ectoparasites [5]. If bat ectoparasites do migrate between co-roosting or co-feeding bat species, it could explain why bats of different species test positive for anti-EBOV IgG [6].
Although reasonably low, the passage history of EBOV isolate used in this study may have influenced the outcome of our results in experimentally inoculated R. aegyptiacus, including the severity of infection, level of viral replication, tissue tropism, and shedding pattern. The effect of in vitro EBOV passage history on genetic and biological characteristics of this virus remains to be investigated.

Conclusions
In conclusion, our findings further confirm that R. aegyptiacus bats are unlikely to maintain and perpetuate EBOV in nature. However, natural ports of filovirus entry and the resulting viral replication and shedding patterns in R. aegyptiacus are unknown. In this context, the likelihood of altered EBOV infection profiles following discrete exposure routes warrant further investigations. While R. aegyptiacus might not be a suitable investigational model for studying ebolavirus-reservoir relationships, our results contribute towards establishing an experimental bat-filovirus pathogenesis model and interpretation of EBOV-seropositive results in wild-caught R. aegyptiacus. Understanding of the host mechanisms involved in abortive (following EBOV inoculation) versus systemic but sub-clinical infection (following Marburg virus inoculation) profiles in R. aegyptiacus might assist in a better understanding of filovirus pathogenesis.