Hendra virus (HeV) is a paramyxovirus of the genus Henipavirus responsible for fatal infection in horses and humans in eastern Australia. Flying-foxes of the genus Pteropus have been identified as the reservoir host of HeV through serological surveys [1–3] and the isolation of virus in naturally infected Pteropus poliocephalus and P. alecto [2]. HeV antibodies have been identified in all four Australian mainland Pteropus species; P. alecto, P. poliocephalus, P. scapulatus, and P. conspicillatus. No evidence of HeV infection has been found in non-pteropid bats [1,2,4]. Since its discovery in 1994, HeV has spilled from its natural reservoir host on 51 known occasions to 30 June 2014 [5] resulting in predominantly respiratory and neurological disease (typically fatal) in horses [6]. In a minority of cases, infected horses have transmitted HeV to close-contact humans, causing similar disease profiles with a case fatality rate of 57% [7,8]. There is no evidence that HeV causes clinical disease in either naturally infected [4] or experimentally infected flying-foxes [9–11].

Despite significant research efforts over the past 20 years, the exact mode of HeV transmission between flying-foxes, or from flying-foxes to horses, has not been conclusively determined. Several studies have investigated potential routes of excretion, tissue tropism and transmission of infection in both experimentally and naturally infected flying-foxes [2,9–11]. HeV has been identified in the lung, spleen, liver, and kidney of experimentally infected adult P. alecto [11] and in the kidney, spleen, heart, and vascular tissue of experimentally infected P. poliocephalus [9,10]. HeV has also been detected in naturally infected flying-foxes [12] including the uterine fluid and pooled foetal lung and liver from aborted twin foetuses of one P. poliocephalus, and the foetal lung of one P. alecto [2]. There are a number of biologically plausible modes of HeV excretion in flying-foxes, including infected placental tissue, placental fluids, urine, faeces and oronasal secretions. The predominant hypotheses regarding HeV spillovers involve equine exposure to 1) infected placental and birthing material, or 2) infected excreta (i.e. urine, faeces, spats) or partially-eaten food.

The majority of studies that examine risk factors for spillover events have used serological data or modelling to assess flying-fox infection dynamics over time. Serological surveys of wild P. conspicillatus [13] and P. scapulatus [14] populations have shown increased detection of HeV antibodies associated with late-stage gestation and/or early lactation. Consequently, gestation and birthing seasons in flying-foxes have been broadly accepted as risk factors for increased HeV prevalence, and for spillover risk. Although these studies related to P. scapulatus and P. conspicillatus, this assumption has been extrapolated in risk modelling involving P. alecto and P. poliocephalus [15].

Understanding how natural infection may differ from experimental models, and identifying potential risk factors for HeV infection in its natural host is fundamental to accurately assessing the risk of spillover from flying-foxes to horses. While there has been a tendency for all flying-foxes to be considered equal in terms of their role as the natural HeV host [15], recent spatial modelling of spillover events indicates that the presence of only P. alecto and P. conspicillatus correlate with spillovers [16].

There remain important gaps in our knowledge of HeV infection dynamics at both the individual flying-fox level and the population level, and how they influence the risk of spillover to horses. To date our knowledge of individual flying-fox infection dynamics of HeV is derived from experimental and, to a lesser extent, natural infections from a limited number of species and animals. Population-level infection studies of HeV in flying-foxes primarily utilise seroprevalence as a surrogate for infection. How experimental infection compares with natural infection within the four Pteropus species remains largely unknown.

This study screens an archived collection of bat tissues for molecular evidence of natural HeV infection and analyses the data to determine tissue tropism and risk factors associated with its detection in flying-foxes.

This archived tissue collection has presented a unique opportunity to examine both HeV tissue tropism and risk factors for infection in naturally infected flying-foxes of the four Australian species of the genus Pteropus. HeV was detected in multiple tissues including kidney, spleen, lung, liver, placenta and blood components, consistent with previous studies of P. alecto and P. poliocephalus [9–11]. However, this study represents the first detection of HeV RNA in the tissues of P. conspicillatus, with detections in spleen, kidney, liver and blood components.

The apparent predilection of HeV for the spleen in naturally-infected flying-foxes is consistent with previous experimental infection studies [11]. These combined findings suggest that the spleen may play a fundamental role in either the maintenance of infection or the processing of viral components in the infected flying-fox. Firstly, detection of HeV RNA in the spleen could indicate that this organ is an active site of viral replication, as live virus has been isolated from respective foetal and adult spleens in two of four experimentally-infected pregnant grey-headed flying-foxes [10]. While IHC staining in the vascular tissue of the spleen was an inconsistent finding in experimentally-infected male P. poliocephalus (two of eight animals) [9], it nonetheless also supports the possibility that the spleen could be an active site of virus replication in infected flying-foxes. Secondly, detection of HeV in the spleen could simply reflect the processing of viral components in this immunologically-active tissue. A third possibility is that the spleen represents a site of latent infection in flying-foxes, and a plausible mechanism for the hypothesized role of recrudescence in the persistence of HeV infection in flying-fox populations [22,23]. It is not possible to differentiate between these hypotheses on the basis of PCR testing alone, and while our negative IHC findings in this study constrain further interpretation, we suggest that the alternative scenarios are not necessarily mutually exclusive. Regardless of whether virus replicates in the spleen or at a distant site, circulating via blood and lymph to sequester and concentrate in the spleen, it is evident that the spleen is the tissue of choice in determining the infection status of an individual. Future immunological studies exploring the structure and function of bat follicular dendritic cells (FDCs) may clarify the role of the spleen.

While viral RNA was detected in tissues, no antigen was detected in corresponding formalin-fixed paraffin embedded tissue sections by IHC. This finding is consistent with that of Halpin et al. (2011), who also reported negative IHC results in PCR-positive tissue samples from experimentally-infected P. alecto (including kidney and bladder samples from a single non-pregnant female bat in which HeV was isolated in urine 12–14 days prior to euthanasia). Thus, the negative IHC results in our study likely reflect the lower diagnostic sensitivity of that assay compared to PCR.

While the exact mode of HeV transmission in flying-foxes is unknown, the detection of HeV RNA in the spleen, liver and lung of naturally-infected flying-foxes demonstrates that systemic infection almost certainly occurs, presumably following local replication in the exposed mucosal surfaces. It is possible that HeV could progressively track from upper (e.g. nasal mucosa) to lower (e.g. lung) respiratory tract tissues without systemic involvement, and the portal vein draining the gastrointestinal tract could theoretically translocate HeV from the gut to the liver without precipitating a truly systemic infection, presuming an oral route of exposure. However it is improbable that detection of HeV in the spleen could be explained through such direct transmission pathways. The detection of HeV RNA in kidney tissue also reflects likely initial mucosal replication at the portal of entry (e.g. oronasal) followed by systemic infection and localization in the kidneys. Though again improbable, the possibility of a retrograde-type urinary tract infection in naturally infected flying-foxes (whereby viral replication in lower urogenital tissues precedes viral spread via urethral, bladder, ureteral and finally, kidney tissues cannot be totally excluded). Local replication at the site of inoculation with subsequent systemic infection has been demonstrated in several experimental HeV and NiV infection trials in Pteropus spp. [9–11,18]. Our findings in naturally infected flying-foxes support this basic model of infection for henipaviruses in their reservoir hosts.

Vertical transmission of HeV has been demonstrated previously in experimentally infected [10] and naturally infected [2,10] flying-foxes. The absence of HeV detection in foetal tissue in all four HeV-positive pregnant animals in this study indicates that in-utero transmission does not always occur. The finding has relevance to both natural HeV infection dynamics in flying-foxes and to the hypothesized role of infected placental and foetal tissues as a source of infection in equine spillovers [2,10]. While limited numbers preclude any meaningful estimate of frequency of vertical transmission, the lack of detections in foetal tissues in our study supports previous contentions (based on seroprevalence patterns in three species) that it is unlikely to be the predominant form of transmission [1,13,14]. In these studies, the low seroprevalence observed in juvenile and immature flying-foxes after maternally-acquired passive immunity had waned was not considered to be consistent with a high frequency of HeV exposure in utero.

The first two detected HeV spillovers to horses were temporally correlated with the southern hemisphere spring birthing season of three of the four Australian mainland species of Pteropus, leading to postulation of flying-fox birthing as a risk factor for spillover [24]. However, as spillover numbers have accumulated over intervening years, a strong temporal clustering of spillovers during the southern hemisphere winter is increasingly evident [12], now with 35/51 spillovers to 30 June 2014 occurring in June, July and August. This temporal association coincides with mid- to late-gestation in P. alecto, P. conspicillatus and P. poliocephalus rather than the birthing season per se, and while unseasonal births do occur throughout the year, the main birthing pulse in these species occurs in late September to November [25]. Indeed, the described reproductive biology of flying-foxes also argues against infected placental and foetal tissues as a likely source of infection in equine spillovers: Hall & Richards (2000) observed that flying-foxes have diurnal parturition that takes place in the roost with the mother consuming the expelled placenta [25].

When investigating effective risk management strategies for disease spillover from flying-foxes to horses and humans, it is important to understand infection dynamics in the host species. Analysis indicates that flying-fox species is a risk factor for HeV infection, with the two closely related species P. alecto and P. conspicillatus [26] evidently the primary reservoir host of HeV. Notwithstanding the modest sample size in our study, this contention is supported by the absence of HeV detection in P. scapulatus and the detection of HeV in only one P. poliocephalus. These differential species detections provide biological support for a recent modelling study that showed a positive spatial correlation between the combined density of P. alecto and P. conspicillatus, and the location of equine spillovers [16]. Taken together, the two studies suggest that P. alecto and P. conspicillatus play a particularly important role in the cross-species transmission of HeV infection to horses.

The lack of HeV detection in P. scapulatus tissues in this study (95% CI 0–7%) is consistent with the findings of Field et al. (2011) [12] and Edson et al. (in preparation), and suggests that systemic infection is infrequent or lacking in this species, however, all four mainland Australian species of Pteropus, including P. scapulatus, have antibodies for HeV [1,13,14]. We suggest that exposure to HeV in P. scapulatus may result in seroconversion in the absence of systemic replication and/or shedding of virus. However, the possibility of local replication and/or shedding cannot be excluded, given the recent description in a mouse model of HeV infection, characterized by the absence of viremia or systemic involvement [27]. If P. scapulatus are able to effectively clear HeV infection before systemic involvement, it is possible that HeV RNA could be present in tissues (e.g. oral cavity) or secretions (e.g. oronasal secretions) that were unavailable for testing in this study. However, if our contention that P. scapulatus seroconverts in the absence of systemic replication and virus excretion is correct, this would suggest that serology may not be a robust proxy for infection and transmission data when viral excretion has not been demonstrated.

The role of P. poliocephalus in HeV infection dynamics is less clear. While there is demonstrated HeV detection in this species in this and previous studies through detection of RNA, virus isolation, and IHC [2,9,10], the significantly lower infection prevalence found in this study (1%: 95% CI 0.03–7%) suggests P. poliocephalus has limited viral replication and/or excretion. This finding is again supported by the modelling of Smith et al.(2014) [16], and may have implications for experimental HeV infection studies, the majority of which have used P. poliocephalus [9,10], and it may be that P. alecto and P. conspicillatus are better model species.

Our analysis identified species as a significant risk factor for HeV infection, but not reproductive status, season, latitude and year. Seroprevalence studies have previously reported risk factors for HeV as age, pregnancy and lactation in P. conspicillatus [13], and age, sex, pregnancy, lactation, and season in P. scapulatus [14]. Whether this lack of agreement reflects sample size limitations in our study, or the different outcome variables used (HeV detection in this study verses HeV sero-status in the earlier studies) is unclear at this time.

Our study has yielded a number of significant findings including the first detection of HeV in the tissues of P. conspicillatus, providing evidence of systemic infection in this species. The predilection of HeV for the spleen of naturally infected flying-foxes suggests that the spleen may play a vital role in either the maintenance of infection, immunological processing of viral components or a mechanism for recrudescence. Our lack of detection of infection in foetal tissues of HeV positive pregnant females supports placental and foetal tissues not being an important source of equine infection, and reinforces flying-fox urine as a key transmission pathway. Finally, we have shown that species is a risk factor for the detection of HeV in naturally infected flying-foxes, specifically P. alecto and P. conspicillatus, supporting the contention of Smith et al. (2014) that these two species are likely the primary reservoir host of HeV, and play a major role in the cross-species transmission of HeV infection to horses.

Additional research that investigated routes of excretion in naturally infected flying-foxes, and thus the modes of transmission to horses would complement the findings of this study, and enable better targeted HeV exposure risk management advice to the horse industry.