Alphacoronaviruses in New World Bats: Prevalence, Persistence, Phylogeny, and Potential for Interaction with Humans

Bats are reservoirs for many different coronaviruses (CoVs) as well as many other important zoonotic viruses. We sampled feces and/or anal swabs of 1,044 insectivorous bats of 2 families and 17 species from 21 different locations within Colorado from 2007 to 2009. We detected alphacoronavirus RNA in bats of 4 species: big brown bats (Eptesicus fuscus), 10% prevalence; long-legged bats (Myotis volans), 8% prevalence; little brown bats (Myotis lucifugus), 3% prevalence; and western long-eared bats (Myotis evotis), 2% prevalence. Overall, juvenile bats were twice as likely to be positive for CoV RNA as adult bats. At two of the rural sampling sites, CoV RNAs were detected in big brown and long-legged bats during the three sequential summers of this study. CoV RNA was detected in big brown bats in all five of the urban maternity roosts sampled throughout each of the periods tested. Individually tagged big brown bats that were positive for CoV RNA and later sampled again all became CoV RNA negative. Nucleotide sequences in the RdRp gene fell into 3 main clusters, all distinct from those of Old World bats. Similar nucleotide sequences were found in amplicons from gene 1b and the spike gene in both a big-brown and a long-legged bat, indicating that a CoV may be capable of infecting bats of different genera. These data suggest that ongoing evolution of CoVs in bats creates the possibility of a continued threat for emergence into hosts of other species. Alphacoronavirus RNA was detected at a high prevalence in big brown bats in roosts in close proximity to human habitations (10%) and known to have direct contact with people (19%), suggesting that significant potential opportunities exist for cross-species transmission of these viruses. Further CoV surveillance studies in bats throughout the Americas are warranted.


Introduction
Bats play important roles in maintaining and transmitting zoonotic viruses [1,2,3]. More than 99 different viruses have been detected in and/or isolated from bats of diverse species [2] (and C. Calisher, personal communication). Rabies virus and other lyssaviruses infect bats of many species, and Old World fruit bats (family Pteropodidae) are reservoirs for both Hendra and Nipah viruses [4,5,6]. Two newly discovered human reoviruses, Melaka virus and Kampar virus, associated with influenza-like illnesses in humans, may be transmitted from small flying foxes (fruit bats; Pteropus hypomelanus) based on the close phylogenetic relationships of these viruses to Pulau virus, a bat reovirus [7,8]. Egyptian fruit bats (Rousetttus aegyptiacus) are known reservoirs of Marburg and certain ebolaviruses [9,10].
The potential for emergence of zoonotic viruses into the human population depends on the prevalence of the virus in its host species, host range mutations within viral quasispecies, and the degree to which the reservoir host interacts with humans. In 2006, we reported the first detection of alphacoronavirus RNA in feces of North American bats sampled in the Rocky Mountain region of Colorado [17]. Here we describe a much larger and more comprehensive study of coronavirus prevalence, epizootiology, geographic distribution, and persistence, as well as preliminary phylogenetic analysis of CoV genome sequences in bats in Colorado.

Ethics Statement
Capture, marking, and sampling of bats followed guidelines of the American Society of Mammalogists [25] and animal protocols were approved by the Institutional Animal Care and Use Committee of the U.S. Geological Survey, Fort Collins Science Center ('Standard Operating Procedure 01-01 for the Capture, Handling, Marking, Tagging, Biopsy Sampling, and Collection of Bats') and Colorado State University (CSU IACUC number 03-096A). Bats were captured under authority of a scientific collecting license (permit numbers: 07TR738A3, 08TR2010, and 09TR2010) issued by the Colorado Division of Wildlife.

Sample Collection
Insectivorous bats of the families Vespertilionidae (16 species) and Molossidae (1 species) were sampled at 16 rural sites (sites #1-16, Fig. 1) in the Rocky Mountain region during the summer of 2007. Bats were identified to species based on external morphological characteristics as described in regional faunal manuals [26,27] adopting revised taxonomy for Myotis occultus [28] and Parastrellus hesperus [29]. To determine whether CoVs persist in bat populations over the course of several years, additional bat fecal samples were collected during the summers of 2008 and 2009 at two rural sites in north central and southeastern Colorado. In addition, big brown bats (Eptesicus fuscus) were sampled at 5 different sites (sites #17-21) within a single urban municipality in Northern Colorado (Fort Collins) during the summers of 2007 and 2008. These sites were chosen because they were in close proximity to humans [30]. Site #17 was in a vintage farmhouse that is currently being used as a family visitation center; site #18 was a natural creek surrounded by suburban neighborhoods; site #19 was in the recreation center of a church; site #20 was within an education building, and site #21 was within a picnic pavilion at a public park. Several of these sites had been previously used in rabies ecology studies, and some bats had been tagged with Passive Integrated Transponders (PIT tags) for host demographic analysis [31,32]. This allowed for repeated capture and sampling of known individual bats.
All bats were either captured in mist nets during the night as they drank or foraged near open water, or were caught in mist nets or harp traps as they emerged from roosts. Whenever possible, the species, sex, reproductive status, age (adult or juvenile), date, and location of capture were recorded for each bat sampled. Bats were sampled as previously described [17], typically within 5-10 min of capture, and then released. Anal/rectal swabs or fecal pellets were taken using sterile calcium alginate swabs and stored in RNAlater (Ambion, Austin, TX) and/or M4 viral transport medium (VTM, Remel; Lenexa, KS). All samples were stored at 270uC prior to analysis. Based on sample type and medium results were pooled for analysis of prevalence surveys. In a post hoc analysis we identified differences in the efficacy of different sampling methods (Text S1) such that the data represent minimal estimates of the prevalence of CoV infection in bats.
Bat carcasses submitted to the Colorado Department of Public Health and Environment (CDPHE) that were negative for rabies viruses were sent to our laboratory for detection of CoV RNA. These bats had been submitted from counties throughout Colorado for rabies testing to rule out the need for post-exposure rabies prophylaxis of humans who had had close contact with these animals [30,33]. Intestines were removed from the bats and stored at 270uC prior to analysis. Extracted RNA was eluted in 60 mL of RNase-free water and stored at 280uC. Before RT-PCR, 50 microliters of RNA was treated with Zymo OneStep PCR Inhibitor Removal Kit (Zymo Research, Orange, CA) following the manufacturer's instructions. cDNA was generated by SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA) with random hexamers in a 20 mL reaction using 11 mL of RNA as a template according to the manufacturer's instructions. All samples were analyzed in duplicate. Reverse-transcription products were stored at 220uC.

PCR and Nucleotide Sequencing
All cDNA samples collected from bats at rural sites or during 2007 were screened for CoV RNA by PCR with a pair of pan-CoV consensus primers [13] that amplify a highly conserved region (400 nucleotide amplicon) of the coronavirus RNAdependent RNA polymerase (RdRp) gene as previously described [17] except that we used 2.0 mmol/L of primers and 1 mL of cDNA or PCR product (for hemi-nested reactions). To increase the sensitivity of RNA detection, based on our previously published bat CoV sequences [17] and new data from this study, we designed specific primers within the amplicons of alphacoronaviruses from bats of several species in the genus Myotis and big brown bats (Table S1). All of the specimens collected from longlegged and big brown bats were also tested with these primers.
To obtain longer nucleotide sequences, RT-PCR was performed using consensus degenerate primers from several regions within the RdRp gene in a SuperScript III one-step RT-PCR system with Platinum Taq High Fidelity kit (Invitrogen, San Diego, CA, USA). Similarly, we designed consensus primers that targeted a highly conserved region of the S2 region of the alphacoronavirus spike gene, and made primers from an exact S2 sequence obtained from a big brown bat (Table S1).
To minimize the possibility of contamination, all RT and PCR reactions were prepared in an enclosed acrylic nucleic acid workstation equipped with a UV light (Clone Zone, USA Scientific, Ocala, FL) in a room separate from the main laboratory. Water controls without template included in every RT and PCR experiment gave no false-positive results. Amplicons were analyzed by agarose gel electrophoresis and sequenced on an ABI 3730 DNA sequencer (Applied Biosystems Technologies, Carlsbad, CA) at the University of Colorado School of Medicine Cancer Center DNA Sequencing and Analysis Core. Samples were scored as positive if CoV RNA was detected on two PCR runs. Statistical significance was determined using Fisher's exact test. Phylogenetic analyses were conducted using MEGA version 4, and phylogenetic trees were constructed using the neighbor-joining method [34]. The nucleotide sequences from this study were deposited in GenBank under accession numbers HQ336973-HQ336976 and JF414933-JF414936.  Table 1).

Persistence of CoV RNA in Bat Populations
At site #4, a high-elevation meadow in a mountainous area of north-central Colorado, 76 long-legged bats were sampled during three consecutive summers (2007)(2008)(2009). Although the sampled bats were not individually marked, the consistent capture of large numbers of females soon after sunset at the site indicated that most of the sampled bats likely came from a nearby maternity roost. Female bats often show year-to-year fidelity to maternity roosts [35]. The percentage of long-legged bats that tested positive for CoV RNA at site #4 varied by year from 6% to 31% (Table 2).
At site #5, an arid grassland bisected by canyons in southeastern Colorado [36], 56 bats of eight different species were sampled during two consecutive summers (2008 and 2009). Only big brown bats at site #5 were positive for CoV RNA. Although the number of big brown bats sampled at site #5 was small (4 in 2008 and 14 in 2009), the prevalence of CoV RNA in these bats during these two summers was high (29% to 100%) ( Table 2).
In the five different urban locations (sites #17-21), 465 samples were collected from big brown bats during the summers of 2007

Lack of Persistence of CoV Infections in Individually Tagged Big Brown Bats in Urban Roosts
All of the urban bat sampling sites were part of a previous study of the ecology of rabies in big brown bats that emphasized host demography [31,32], and 113 (24%) of the 465 bats from these sites sampled for this study had been previously individually tagged. Sixteen (14%) of these tagged bats were captured and sampled more than once (14 captured twice, and 2 captured three times). Five (31%) of the 16 repeatedly sampled tagged bats captured in 2008 were positive for CoV RNA, but no CoV RNA could be detected in subsequent samples ( Table 3). Four of the 5 bats became negative for CoV RNA within 6 weeks after they tested positive for CoV RNA. (The fifth bat was not recaptured after turning positive). Thus, in this small group of serially sampled bats, individual bats were not continually shedding detectable amounts of CoV RNA, so did not appear to be persistently infected.

Age and Sex Distribution of Bats Positive for CoV RNA
The age and sex distributions of the 999 (94%) bats sampled for which these data were available and the subset of big brown bats in the urban maternity roosts sampled are shown in Table 4. Juvenile bats were two times more likely to be positive for CoV RNA than adults bats (13% vs. 6%, p = 0.008). In the urban maternity roosts, as expected, the majority of the big brown bats sampled were adult females, but juvenile bats (10 of 52 tested, 19%) were also more than twice as likely to be positive for CoV RNA than adult bats (36 of 413 tested, 9%, p = 0.03).

Preliminary Phylogenetic Analysis of Rocky Mountain Bat CoVs
From the samples positive for CoV RNA, we obtained nucleotide sequences of amplicons ranging in length from 93-356 nt from the RdRp region of gene 1b. These formed three clusters (.90% nt identity within each cluster). The first cluster (A) included CoV RNAs of big brown bats from sites #5 and #17-21, the one big brown bat from site #4, and two long-legged bats from site #4 that were collected in 2007 and 2009. The sequence of the A cluster (representative bat: RM-Bt-CoV 453/2007 EF) was 96% identical to the same region from a big brown bat (RM-Bt-CoV 65) reported in our previous study [17]. The second cluster (B) (representative bat: RM-Bt-CoV 09-07/2009 MV) was found in 2 long-legged bats (one sampled in 2008 and one in 2009) and one western long-eared bat sampled at site #4. These sequences had .97% identity in this region to CoV RNA obtained from several occult bats (M. occultus; RM-Bt-CoV 6 and 11) reported previously [17]. The third cluster (C) of CoV amplicons (representative bat: RM-Bt-CoV 429/2007 MV) were from other long-legged bats sampled at site #4. These sequences were 96% identical to that from an occult bat (RM-Bt-CoV 3) reported previously (Table 5 and Figure S1). Cluster A had ,65% identity with clusters B and C, whereas clusters B and C had 83% identity to one another.
An 1100 nt sequence encoding the S2 domain of the spike glycoprotein was obtained from a big brown bat collected at site #4 in 2007 (Rocky Mountain Bat-CoV 453/2007 EF). We compared this sequence to S2 sequences of other known coronaviruses (Table 5 and Figure 2) and found that this genome was distantly related to other known alphacoronaviruses in group 1a, with ,67% nucleotide identity to CoVs. We also obtained a 700 nucleotide sequence in the same region of S2 from the long-legged bat (RM-Bat-CoV 433/2007 MV) that had a similar sequence to this big brown bat in the RdRp gene (both in RdRp cluster A). These S2 amplicons had .98% nt sequence identity. The closest bat coronavirus spike sequence to RM-Bt-CoV 453/2007 found in GenBank, was Bt-CoV A701, from an Old World species, Rickett's big-footed bat (Myotis ricketti) sampled in Southeast China in 2005 [14] (65% nucleotide identity, 65% amino acid identity).
An approximately 4000 nt sequence in 2 segments of the RdRp gene was obtained from one of the little brown (RM-Bt-CoV-15/ 2006/ML) and one of the big brown bats (RM-Bt-CoV-61/2007/ EF) that were submitted to the CDPHE. These nt sequences were only 62% identical, indicating that they represented two unique viruses in bats of these two species. These sequences were distantly related (,75% nt identity) to other known alphacoronaviruses,   with ,75% nt identity to CoVs in this group, including all currently available Old World bat CoVs (Table 5 and Figure 3).

Discussion
This is the first multiyear surveillance project of CoVs in wild bats in North America. CoV RNA was detected in approximately 7% of all bats sampled (likely an underestimate of prevalence, Text S1), comparable to the prevalence of CoV RNA detected in various species of bats reported in other parts of the world (ranging from 2-55%) [14,18,19,21,22,37,38,39]. In our study no CoV RNA was detected in bats in 13 of the 17 species we sampled (also likely biased negatively). Failure to detect CoVs in bats of these species could be related to the smaller numbers sampled. How-  ever, a relatively high prevalence of CoV RNA was detected in bats of 2 species collected at several different sites: 12% for big brown bats and 8% for long-legged bats, and at lower prevalence, 3% in little brown bats and 2% in western long-eared bats.
In marked contrast to the enormous diversity of CoV genomes found in Old World bats [14,24,40], in this and several other CoV surveillance studies of New World bats [17,18,22], all CoVs detected were alphacoronaviruses. Our data indicate that nucleotide sequences of alphacoronaviruses harbored by Colorado bats are distinct from those found in Old World bats. Two recent studies of the bat guano virome using next generation sequence technology also only detected alphacoronaviruses in the New World bats of the species tested, as well as a diverse array of other types of viruses [41,42]. Thus, so far there appears to be much more limited CoV diversity in New World bats of the species tested than in Old World bats.
Betacoronaviruses have only been detected in Old World bat species belonging to the families Pteropodidae (Rousettus spp) and Rhinolophidae (Rhinolophus spp.) which belong to the chiropteran suborder Yinpterochiroptera. Based on available evidence, betacoronaviruses could be restricted to hosts in the suborder Yinpterochiroptera (families Pteropodidae, Rhinolophidae, Megadermatidae, Craseonycteridae, Rhinopomatidae). No bat families of the suborder Yinpterochiroptera occur in the New World. [43]. The finding of only alphacoronaviruses in our study may be because bats of these species are resistant to other CoVs and/or bats from different parts of the New World have yet to be tested for CoV infection, as we sampled bats from only a subset of the hundreds of species that reside in the New World.
These observations also support the hypothesis that coronaviruses may have co-evolved with their bat hosts, as no species of bat is found both in the New World and Old World [44]. To date, however, only a small subset of New World species of bats has been tested for coronavirus infection. As 75% of living genera of all bats worldwide are found in the New Worlds tropics alone, further CoV surveillance in bats of additional species from different regions in the Western hemisphere may reveal hitherto undetected varieties of coronaviruses. The seasonal epidemiology and persistence of New World CoV infections in individual bats and within bat populations has not been elucidated. The most comprehensive epidemiological investigation of CoVs to date in Old World bat populations showed that the prevalence of SARS-Rh-BatCoVs in rhinolophid bats over a four-year period at collection sites in Hong Kong SAR and China peaked in the spring and varied from year to year. We found similar results in New World bats. At site #4 long-legged bats had an alphacoronavirus RNA prevalence of 31% in 2007, 19% in 2009, but only 6% in 2008. In all five of the urban maternity roosts sampled, CoVs persisted in bat roosts throughout the course of the non-hibernating part of the year (spring/summer) and persisted from year to year. We also found that the prevalence of CoV infection in these bat roosts tended to peak in late spring/ early summer. The prevalence of infection with human CoVs also shows significant annual variations [45], possibly depending on environmental conditions and/or fluctuating CoV antibody levels in the population. Possible seasonal variation in CoV infection rates may explain why in our initial 2006 study we found a high prevalence (50%) of alphacoronavirus RNA in occult bats [17], but in 2007 we did not detect any positive individuals (22 tested in the same region).
The majority of the bats sampled in our study were adult females because they were primarily captured from maternity roosts. The highest prevalence of infection was noted in juvenile bats. In Germany, CoV infection was also found to be associated with young age and was more common in female bats from maternity roosts compared to female bats found at foraging or swarming sites [19]. These findings support the hypothesis that younger bats may be more susceptible to CoV infection and may serve to propagate and maintain these viruses within bat colonies.
No overt clinical manifestations of disease were observed in any of the captured bats, including those that were infected with CoVs. In the small subset of bats that were tagged and recaptured, no individual bat remained persistently positive for CoV RNA after 6 weeks. Similar findings were made in rhinolophid bats in Asia that harbor SARs-like-bat-CoVs [37] and in fruit bats experimentally infected with bat CoVs which showed no signs of illness [39]. These data suggest that although CoVs persist within bat populations, individual bats may experience only self-limited infections with CoVs without apparent illness.
Phylogenetic studies of CoV genomes in Old World bats in Asia and Europe have suggested that some bat CoVs may infect bats of only one species or several closely related species. In Asia and Germany, different species of bats roosting in the same cave were found to host different CoVs, whereas bats of the same species in different locations harbored similar CoVs [14,19]. In Europe, strict associations were found between bat CoV deduced amino acid sequences in an 816 bp fragment of the RdRp gene and their specific bat hosts [40]. In Africa, CoVs found in one species of bat were not detected in bats of different species co-roosting in the same cave [38]. Similarly, our study showed that New World bats of the same species in geographically distinct locations and over the course of several years harbor similar CoVs. In contrast to these findings, in Kenya some CoVs appear to be able to infect Old World bats of several different species [21]. Our preliminary nucleotide sequence data also suggests that we found very closely related CoV nucleotide sequences in New World bats from three different species of Myotis (M. volans, M. evotis, and M. occultus). Furthermore, in site #4, we found similar nucleotide sequences in the spike and replicase genes in CoV RNAs from both a big-brown bat and a long-legged bat, suggesting that at least some New World bat CoVs may be able to infect bats of different genera. These findings are notable, as recent phylogenetic studies of rabies viruses in bats suggest that host species barriers play a key role in cross species transmission of viruses [46].
To assess the potential for zoonotic transmission of bat CoVs, we focused part of this present work on North American bats that have the closest contact with humans and sampled roosts where big brown bats had histories of contact or potential for contact with people [30]. Big brown bats are common inhabitants of buildings in cities and towns in Colorado and across the United States, and are the primary species encountered by humans in terms of potential exposure to disease agents [30,33,47] These bats had a high prevalence of CoV infection, ranging from 0-67% (overall 10%) depending on the site and time of year. Big brown bats submitted to the CDPHE for rabies testing because of known direct contact with humans also had a very high prevalence (19%) of CoV infection. Because bats which have known or potential contact with humans have such a high prevalence of CoV infection, opportunities exist for potential transmission of these viruses to humans.
Following the SARS epidemic, intensive surveillance detected a great diversity of CoVs throughout the animal kingdom. CoVs can undergo a high frequency of RNA recombination, both in vitro and in vivo, which may play an important role in their evolution and virulence [48]. Old World bat CoVs of several different genotypes were found to co-exist in a single bat [49]. Thus recombination between different bat CoVs could potentially occur in vivo, giving rise to new CoV genomes. Two strains of HCoV-HKU1 have recombined to yield a novel HCoV-HKU1 genotype [16], and recombination between different strains of SARS-CoV-like viruses in bats may have given rise to civet SARS-CoV [37]. The great diversity of CoVs, their high frequency of RNA recombination, their ability to persist in bat populations, and the finding that some CoVs can apparently infect bats of divergent genera, suggest that ongoing evolution of CoVs in bats may pose a continuing threat for emergence of novel CoVs into new hosts. Table S1 Primers and RT-PCR programs. A. Consensus primers targeted a highly conserved region of the S2 region of the spike gene and from an exact sequence obtained from one of the big brown bats. PCR was performed under the following conditions: one mL of cDNA was amplified in a 50-mL reaction containing, 0.2 mmol/L deoxynucleoside triphosphates, 1 U of PhusionTaq High-Fidelity DNA Polymerase (Finnzymes, Espoo, Finland), and 2.0 mmol/L primers by the following PCR program: 30 sec at 98uC; 40 cycles for 10 sec at 98uC, 15 sec at 50-52uC (depending on the primer set), and 15 sec at 72uC; and then 10 min at 72uC. B. Primers used for detection of CoV sequence in bat samples. One microliter of cDNA was amplified in a 50-mL reaction containing 1.5 mmol/L MgCl 2 , 0.2 mmol/L deoxynucleoside triphosphates, 2.5 U of HotStarTaq (QIAGEN), and 2.0 mmol/L primers using the following PCR program: 15 min at 95uC; 45 cycles for 1 min at 95uC, 1 min at 48uC for MY-F and MY-R and 50uC for EF-F and EF-R, and 1 min at 72uC; and 10 min at 72uC. C. To obtain additional sequences for phylogenetic analysis, for two of the CDPHE intestinal samples, RT-PCR was performed using consensus degenerate primers from several areas within the RdRp gene in a SuperScript III one-step RT-PCR system with Platinum Taq High Fidelity kit (Invitrogen, San Diego, CA, USA). Primers and protocols were kindly provided by Suxiang Tong, PhD and Ying Tao, PhD of the Centers for Disease Control and Prevention, Atlanta, Georgia, USA.

(DOC)
Text S1 Influence of different sampling and analysis techniques on CoV RNA detection. (DOC)