key: cord-0726820-sbcmk0jr
authors: Ntumvi, Nkom F; Ndze, Valantine Ngum; Gillis, Amethyst; Le Doux Diffo, Joseph; Tamoufe, Ubald; Takuo, Jean-Michel; Mouiche, Moctar M M; Nwobegahay, Julius; LeBreton, Matthew; Rimoin, Anne W; Schneider, Bradley S; Monagin, Corina; McIver, David J; Roy, Sanjit; Ayukekbong, James A; Saylors, Karen E; Joly, Damien O; Wolfe, Nathan D; Rubin, Edward M; Lange, Christian E
title: Wildlife in Cameroon harbor diverse coronaviruses, including many closely related to human coronavirus 229E
date: 2022-01-12
journal: Virus Evol
DOI: 10.1093/ve/veab110
sha: 78c9dcc02de5bc1e6e21defe8d484946bafe3acc
doc_id: 726820
cord_uid: sbcmk0jr

Zoonotic spillover of animal viruses into human populations is a continuous and increasing public health risk. Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) highlights the global impact of emergence. Considering the history and diversity of coronaviruses (CoVs), especially in bats, SARS-CoV-2 will likely not be the last to spillover from animals into human populations. We sampled and tested wildlife in the Central African country Cameroon to determine which CoVs are circulating and how they relate to previously detected human and animal CoVs. We collected animal and ecological data at sampling locations and used family-level consensus PCR combined with amplicon sequencing for virus detection. Between 2003 and 2018, samples were collected from 6,580 animals of several different orders. CoV RNA was detected in 175 bats, a civet, and a shrew. The CoV RNAs detected in the bats represented 17 different genetic clusters, coinciding with alpha (n = 8) and beta (n = 9) CoVs. Sequences resembling human CoV-229E (HCoV-229E) were found in 40 Hipposideridae bats. Phylogenetic analyses place the human-derived HCoV-229E isolates closest to those from camels in terms of the S and N genes but closest to isolates from bats for the envelope, membrane, and RNA-dependent RNA polymerase genes. The CoV RNA positivity rate in bats varied significantly (P < 0.001) between the wet (8.2 per cent) and dry seasons (4.5 per cent). Most sampled species accordingly had a wet season high and dry season low, while for some the opposite was found. Eight of the suspected CoV species of which we detected RNA appear to be entirely novel CoV species, which suggests that CoV diversity in African wildlife is still rather poorly understood. The detection of multiple different variants of HCoV-229E-like viruses supports the bat reservoir hypothesis for this virus, with the phylogenetic results casting some doubt on camels as an intermediate host. The findings also support the previously proposed influence of ecological factors on CoV circulation, indicating a high level of underlying complexity to the viral ecology. These results indicate the importance of investing in surveillance activities among wild animals to detect all potential threats as well as sentinel surveillance among exposed humans to determine emerging threats.

The emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has highlighted the inherent risks and potential consequences of pathogen spillovers from animal reservoirs into the human populations. Animals are known to be the reservoir for many human diseases historically and contemporarily as exemplified by zoonotic diseases, such as rabies, brucellosis, bubonic plague, and trichinellosis (Shope 1982; Weiss et al. 2003; Wolfe et al. 2007; Gottstein et al. 2009; Drancourt and Raoult 2016; Cross et al. 2019) . Humans are often considered an accidental host for zoonotic diseases in this context, even if human-tohuman transmission is possible. This understanding has shifted with the advent, and increased availability, of genetic characterization of pathogens, and humans are now often considered opportunistic rather than accidental hosts. This is an especially apt description for viruses, since we now know that a significant number of pathogens commonly referred to as 'human viruses' did not originally evolve with humans but spilled over from animals more recently and subsequently adapted to humans (Weiss et al. 2003; Wolfe et al. 2007) . The best example for this may be HIV, which originated from multiple non-human primate (NHP) spillover events in Africa during the early 20th century, and it also applies to measles virus, influenza A viruses, and others (Weiss et al. 2003; Sharp and Hahn 2011) . While these and SARS-CoV-2 are some of the most publicized examples, there is a general trend of increasing infectious diseases outbreaks in humans, in particular, viral and zoonotic agents, over the past decades, potentially due to factors such as land use and climate change, population growth, and increased international trade and mobility (Jones et al. 2008; Smith et al. 2014; Allen et al. 2017) .

With the global SARS-CoV-2 pandemic, we are witnessing such a post-spillover adaptation in real time, and there is strong evidence that this is not the first time an animal coronavirus (CoV) has gone through this process. While primary attention has previously focused on SARS-CoV, Middle East Respiratory Syndrome (MERS) CoV, and closely related CoVs, it has become clear that the four CoVs that are associated with the common cold (HCoV-NL63, HCoV-229E, HCoV-OC43, and HCoV-HKU1), all likely derived from animal CoVs in the previous decades to centuries (Drosten et al. 2003; Kan et al. 2005; Yip et al. 2009; Drexler et al. 2010; Zaki et al. 2012; Huynh et al. 2012; Corman et al. 2014 Corman et al. , 2015 Corman et al. , 2018 Forni et al. 2017; Cui et al. 2019) . Unlike SARS-CoV-1 and MERS-CoV, these four established themselves in the human population in a process that may have been similar to the current SARS-CoV-2 pandemic, albeit likely much slower. Genetic analysis suggests that the two alpha CoVs HCoV-NL63 and HCoV-229E originated in bats, like SARS-CoVs and MERS-CoV, while the beta CoVs HCoV-OC43 and HCoV-HKU1 likely originated in rodents; in either case with or without a potential intermediate host (Pfefferle et al. 2009; Tao et al. 2017; Zhou et al. 2020) . The fact that we know of three CoVs that spilled over into humans in just the past two decades and that spillovers of the other four can be traced back to recent centuries suggests that these events are common, especially since it is estimated that only a minority of spillover events lead to continued transmission and detection in humans (Glennon et al. 2019; Letko et al. 2020) . CoV spillovers will thus likely continue to occur in the future. This indicates that a better understanding of animal CoVs will be useful to determine spillover risks and the biological mechanisms and drivers for diversification and spillover and to develop appropriate prevention, mitigation, and treatment strategies for future CoV spillovers.

It is generally accepted that there is a direct link between a close genetic relationship of host species and the likelihood of interspecies transmission; however, considering the biology of influenza A viruses, for example, where we see spillover from pigs and birds into humans, or CoVs where bats, rodents, camels, cows, and civets may play a role, it is by no means a hard rule (Kan et al. 2005; Parrish et al. 2008; Dennehy 2017) . With less closely related host species, the risks are more difficult to determine, but much emphasis has recently been placed on host diversity as an indicator for viral diversity, and the role of contact rates (Wolfe et al. 2005; Pike et al. 2010; Maganga et al. 2014; Anthony et al. 2017; Dennehy 2017; Leopardi et al. 2018; Ntumvi et al. 2021) . Host diversity likely results in more viruses circulating in a biome and also provides more opportunities for interspecies transmission, host plasticity, and viral recombination (Dennehy 2017) . Host plasticity in animals may in turn be a predictor for a virus' ability to be transmitted from human to human and hence a major risk factor (Kreuder Johnson et al. 2015) . Consequently, regions with a high biodiversity, such as large parts of Central Africa, South America, and Southeast Asia, may be considered hot spots for spillover (Jones et al. 2008; Allen et al. 2017) . Reports from several African countries suggest that there are many CoVs circulating, primarily in bats, including species related to pathogens such as SARS-CoV-1, MERS-CoV, HCoV-229E, and HCoV-NL63 (Tong et al. 2009; Tao et al. 2012 Tao et al. , 2017 Annan et al. 2013; Geldenhuys et al. 2013 Geldenhuys et al. , 2018 Geldenhuys et al. , 2021 Razanajatovo et al. 2015; Anthony et al. 2017; Markotter et al. 2019; Nziza et al. 2019; Maganga et al. 2020; Kumakamba et al. 2021) .

While the viral genome provides a lot of information about viruses, their origins, and their hosts, other factors also need to be considered when evaluating risks. Direct or indirect humananimal interaction is a prerequisite for zoonotic transmission, but human and animal behaviors and ecologies can play key roles in this process that may involve multiple steps of interspecies transmission and adaptation (Wolfe et al. 2005; Wolfe et al. 2007; Pike et al. 2010; Maganga et al. 2014; Euren et al. 2020; Kumakamba et al. 2021; Ntumvi et al. 2021) . Previous studies have, for example, identified a high host plasticity for certain bat CoVs, indicating that these may potentially pose a higher zoonotic risk than those primarily adapted to a single host (Anthony et al. 2017) . Other factors influencing transmission may include the type of humananimal interface and seasonal fluctuations of CoV circulation as observed in bat populations (Montecino-Latorre et al. 2020; Kumakamba et al. 2021; Grange et al. 2021) . Although important, data on many of these factors are still limited and needs further exploration. Studying CoV diversity in Africa promises rich data that could improve our understanding of their biology, evolutionary history, and risks for humans.

The Central African country Cameroon, which includes some of the northern part of the Congo Basin, where HIV is believed to have spilled over into humans, is rich in biodiversity, with wildlife interaction being common for a large part of the rural population. Increased bushmeat trade and diverse wildlife living in close proximity with human populations makes some of these areas hotspots for high-risk interfaces between animals and humans (Wolfe et al. 2005; Mickleburgh et al. 2009; Saylors 2021) .

Our goal was to determine what CoVs are circulating in wild animals, including rodents, bats, and NHPs, and assess if key ecological factors may influence the rate of CoV detection and thus the exposure risk.

Sample acquisition methods differed depending on the species and interface. Animals in peridomestic settings were captured and released after sampling (bats, rodents, and shrews only), while samples from the (bushmeat) value chain were collected from freshly killed animals voluntarily provided by local hunters upon their return to the village following hunting or by vendors at markets. Non-invasive fecal samples were collected from free-ranging NHPs, while some NHP samples, such as blood or serum, were collected during routine veterinary exams in zoos and wildlife sanctuaries. To avoid incentivizing hunting, hunters and vendors were not compensated. Identification was done in the field by trained field ecologists, as well as retrospectively based on field guides and other resources, including those by Kingdon and Monadjem (Kingdon 2005; Monadjem et al. 2010 Monadjem et al. , 2015 . Oral and rectal swab samples were collected into individual 2.0-ml screw-top cryotubes containing 1.5 ml of either Universal Viral Transport Medium (BD), RNA later, lysis buffer, or Trizol® (Invitrogen), while pea-sized tissue samples were placed in 1.5ml screw-top cryotubes containing 500 µl of either RNA later or lysis buffer (Qiagen), or without medium. Specimen collection was approved by an Institutional Animal Care and Use Committee of University of California (UC) Davis (#16067, # 17803, and # 19300) , Johns Hopkins University (JHU) (ACR #2007-110-03, #2007-110-02 and #2007-110-11) , and University of California Los Angeles (UCLA) (Protocols # FS03M221 and FS06H205) and the Government of Cameroon. All samples were stored in liquid nitrogen as soon as practical, before being stored long term in freezers at −80 • C. Sample collection staff were trained in safe collection techniques in collaboration with representatives from Ministry of Fisheries Livestock Animal Industries (MINEPIA), the Ministry of Forestry and Wildlife (MINFOF), and the Ministry of Environment Nature Protection and Sustainable Development (MINEPDED) and wore dedicated clothing, N95 masks, nitrile gloves, and protective eyewear during animal capture, handling, and sampling.

All laboratory work was carried out at Cameroon's Military Health Research Center (CRESAR). RNA was extracted either manually using Trizol®, with an Qiagen AllPrep kit (tissue), Qiagen Viral RNA Mini Kit (swabs collected prior to 2014), or with a Zymo Direct-zol RNA kit (swabs collected after 2014) and stored at −80ºC. Afterwards, RNA was converted into cDNA using a GoScript™ Reverse Transcription kit (Promega) and stored at −20 • C until analysis. Two conventional nested broad range PCR assays, designed based on different sequence sets, both targeting different conserved regions within the RNA-dependent RNA polymerase gene (RdRp) were used successfully to screen samples of bats, rodents, and other animals (and humans) for CoV RNA (Anthony et al. 2017; McIver et al. 2020; Kumakamba et al. 2021; Li et al. 2021 ). The first PCR amplifies a product of approximately 286 nucleotides between the primer binding sites and was specifically designed for the detection of a broad range of CoVs (Quan et al. 2010 ). The second PCR was used in two modified versions, with one of them specifically targeting a broad range of CoVs in bats and the second one broadly targeting CoVs of other hosts (Watanabe et al. 2010) . Both versions amplify 387 nucleotides between the primer binding sites. Follow-up PCRs were designed to amplify the 3 ′ -end of the Spike (S), as well as the Envelope (E), Membrane (M), and Nucleoprotein (N) genes of CoVs related to HCoV-229E (Table 1) .

PCR products were subjected to gel electrophoresis on a 1.5 per cent agarose gel to identify amplicons of the expected size. PCR was repeated for all samples where this was the case, and PCR products were excised. Amplified DNA was extracted using either the QIAquick Gel Extraction Kit (Qiagen) or the Wizard SV Gel and PCR Cleanup System (Promega) and sent for commercial Sanger sequencing at either GATC, First Base, or Macrogen. Extracts with low DNA concentrations were cloned prior to sequencing. All results from sequencing were analyzed in the Geneious 7.1 software, and primer trimmed consensus sequences compared to the GenBank database (BLAST N, NCBI). Samples where the PCR results could not be repeated, or where the sequencing did not yield interpretable sequences consistent with a CoV were counted as CoV RNA negative. All viral sequences obtained were deposited in the GenBank (Table 2 , Supplement 1).

To facilitate phylogenetic analysis, select sequences of published complete CoV genomes isolated from humans, bats, and other hosts, as well as partial sequences from CoVs closely related to those detected in this study were included. Novel Cameroonian sequences were included if they differed from others by at least 5 per cent. Multiple sequence alignments were made in Geneious (version 11.1.3, ClustalW Alignment). Bayesian phylogeny of the polymerase gene fragment was inferred using MrBayes (version 3.2) with the following parameters: Datatype = DNA, Nucmodel = 4by4, Nst = 1, Coavion = No, # States = 4, Rates = Equal, 2 runs, 4 chains of 10,000,000 generations. The sequence of an avian Gamma Coronavirus (NC_001451) served as outgroup to root the trees based on the RdRp PCR amplicons, while HCoV-NL63 (AY467487) served as outgroup for sequences related to HCoV-229E. Trees were sampled after every 1,000 steps during the process to monitor phylogenetic convergence, and the final average standard deviation of split frequencies was below the MrBayes-recommended final average <0.01 for all analyses (Ronquist et al. 2012 ). The first 10 per cent of the trees were discarded and the remaining ones combined using TreeAnnotator (version 2.5.1; http://beast.bio.ed.ac.uk) and displayed with FIGTREE (1.4.4; http://tree.bio.ed.ac.uk/) (Bouckaert et al. 2019 ).

Ecological data collected along with the samples obtained from bats were statistically analyzed in relation to the outcome of PCR tests. The variables included species, family, and suborder, age, and sex, and factors such as the interface of exposure with humans, and season (wet/dry) of sampling. The taxonomy information for the analysis was based in the Integrated Taxonomic Information System website (https://www.itis.gov/), age coded as either adult or non-adult, interface categorized as 'value chain' for animal samples obtained at markets or directly from hunters, 'tourism' for samples obtained at zoos and sanctuaries, and 'other peridomestic'. The seasons were defined as switching from wet to dry after November 15th and from dry to wet after March 15th of each year. All statistical data analyses were conducted using the statistical software Statistical Package for the Social Sciences version 26. At the univariate level, frequencies and percentages of the selected variables of interest (i.e. PCR test results, species, family, suborder, age, sex, interface, season, etc.) were generated. At the bivariate level, simple cross-tabulation, chi-square tests have been used to determine the statistical association (which are considered statistically significant at 5 per cent level) between outcome and predictor variables. Here, the PCR test result has been considered as the dependent variable, and species, family, suborder, age, sex, interface, and season have been considered as the predictor/independent variables.

A total of 11,474 samples from 6,580 animals of several different orders were collected between 2003 and 2018, covering all 10 regions of Cameroon (Fig. 1, Supplement 2) . Animals sampled were largely rodents (2,740/41.6 per cent), bats (2,581/39.2 per cent), and primates (1,006/15.3 per cent) representing 28 rodent species, 50 bat species, and 24 primate species. Samples were also collected from 159 Eulipotyphla (3 species), 38 pangolins (1 species), 37 carnivores (3 species), 17 even-toed ungulates (4 species), and 2 hyraxes (1 species) (Supplement 3). Samples were predominantly oral (5,214) and rectal (4,818) swabs but also included tissues, such as spleen (893), liver (93), lung (5), colon (4), small intestine (4), muscle (1), skin (1), and thyroid (1). Other sam- ple types included plasma (377), serum (20), and whole blood (1), as well as feces (36), genital swabs (5), and nasal swabs (1). CoV RNA was detected in at least one sample with at least one PCR assay in 175 individual bats, 1 civet, and 1 shrew ( Table 2) .

Rectal swabs were CoV RNA positive in 129 instances, oral swabs in 71 instances, liver and spleen in 3 instances each, and plasma in 2 instances. The Watanabe PCR protocol produced 173 positive results, while the Quan PCR protocols produced 64 positive results. Bat sampling was conducted in many areas of the country, but 46.7 per cent of the bats were sampled in the South Region (Supplement 4). Male bats were slightly overrepresented (53.5 per cent) compared to females; however, female bats were significantly (P < 0.01) more likely to have a positive CoV test (8.2 per cent) than male bats (5.6 per cent). Differences in CoV RNA-positive rate were also observed between adult (7.4 per cent, n = 1495) and younger bats (10.3 per cent, n = 87), but these were not statistically significant. Samples collected at animal-human interfaces categorized as 'Tourism' (1.5 per cent, n = 136) and 'Value chain' (3.1 per cent, n = 488) were significantly less likely (P < 0.001) to be CoV RNA positive than those collected at 'other peridomestic' interfaces (8.1 per cent, n = 1957) such as in or near human dwellings and temporary settlements or close to crop production.

The CoV RNA detected in an African palm civet (Nandinia binotata) resembles that of an alphacoronavirus and, on the nucleotide level (BLASTN), is most closely related to canine CoV, feline CoV, and porcine CoV (TGEV) with 87 per cent, 86 per cent, and 85 per cent identities, respectively (Table 2) . Phylogenetic analysis places the RNA in a cluster with dog, cat, mink, and pig CoVs (Fig. 2) .

The CoV RNA detected in a Goliath shrew (Crocidura goliath) resembles that of an alphacoronavirus and is on the nucleotide level (BLASTN) closest related to isolates of Wencheng Sm shrew CoV (from a Suncus murinus) and isolates of Coronavirus PREDICT CoV-46 (from a Crocidura sp.), with up to 84 per cent and 83 per cent identities, respectively (Table 2) . Phylogenetic analysis places the RNA in a cluster with other shrew CoVs as part of a basal branch of alphacoronaviruses (Fig. 3) .

The CoV RNA detected in the 175 bats form 17 different genetic clusters, of which 8 coincide with alpha and 9 with beta CoVs. The majority of sequences share identities above 90 per cent with known bat CoVs, while the sequences in up to 6 of the 17 clusters do not (bat CoV clusters 2, 11, 14, 15, 17, and 18) Figure 2. Phylogenetic tree: Maximum likelihood phylogenetic tree of coronavirus sequences presented as a proportional cladogram based on the RdRp region targeted by the PCR by Watanabe et. al. (Watanabe et al. 2010) . The sequences detected during the project are highlighted by red boxes, and numbers in brackets indicate the number of sequences sharing more than 95 per cent nucleotide identities. GenBank accession numbers are listed for previously published sequences, while sequences obtained during the project are identified by cluster names (compare Table 2 ). Numbers at nodes indicate bootstrap support. Quan et. al. (Quan et al. 2010) . The sequences detected during the project are highlighted by red boxes, and numbers in brackets indicate the number of sequences sharing more than 95 per cent nucleotide identities. GenBank accession numbers are listed for previously published sequences, while sequences obtained during the project are identified by cluster names (compare Table 2 ). Red boxes indicate isolates from this study. Numbers at nodes indicate bootstrap support.

( Figs 2 and 3) . In 170 bats, RNA corresponding to a single CoV was detected, with 89 resembling alpha CoVs and 81 beta CoVs. In five bats, RNA corresponding to two different CoVs was found. In two of these cases, RNA of both an alpha and a beta CoV was present; in two bats, the RNA of two different beta CoVs was detected; and in one bat, we found the RNA of two different alpha CoVs (Table 2 ). In all other cases (n = 30), where we detected RNA in more than one sample of the same bat, the RNA amplified by the same PCR differed by less than 10 per cent. In 24 instances, they were 100 per cent identical; in five instances, they differed by less than 1 per cent (as a result of quasispecies replication or sequencing artifacts); and in one instance, the difference was 5.2 per cent (potentially the result of a co-infection).

Sequences resembling HCoV-229E were found in 40 bats from the Hipposideridae family, primarily in the species Hipposideros ruber, and showed differences of up to 6 per cent among each other in the Watanabe amplicons and up to 11 per cent in the Quan amplicons. Additional sequence information of the full or partial S, E, M, and N genes was obtained, and differences in the sequences ranged up to 8 per cent for the E, 12 per cent for the M, 32 per cent for the N, and 26 per cent for the S genes. Phylogenetic clustering of sequences differed depending on the gene, with isolates BtCoV/KW2E-F56/Hip_cf._rub/GHA/2011 (KT253271) and BtCoV/AT1A-F1/Hip_aba/GHA/2010 (KT253272) being consistently the most basal isolates in the CoV-229E branch (Fig. 4 , Supplement 1). The phylogenetic analyses, which involve sequence isolates from bat, camel, and human hosts, place the HCoV-229E isolates that were obtained from humans closest to isolates from camels in case of the S and N genes but closest to isolates from bats for the E, M, and RdRp genes (Fig. 4, Supplement 5 ).

Bat sample collection was focused on seasons and thus varied over the months, with a high of 432 samples collected in March and a low of 48 in August (Supplement 6). Bats sampled in the wet season accounted for 61.9 per cent of total bats compared to 38.1 per cent in the dry season. The proportion of CoV RNA-positive bats varied significantly (P ≤ 0.001) between the wet (8.2 per cent) and dry seasons (4.5 per cent), while the proportion of events (samples collected at the same location on the same day) with at least one positive animal was very similar with 29.4 per cent during the wet season and 28.3 per cent during the dry season. The proportion of CoV RNA-positive individuals for the sampling events with >25 bats (n = 24) fluctuated between 0 per cent and 48 per cent and was at 21.2 per cent for the largest event. Nine events took place during the wet season, and 15 during the dry season. Among the bats sampled during this largest event were 17 different species with varying rates of CoV RNA detection, including Chaerephon pumilus (0 per cent, n = 29), Epomophorus gambianus (54.5 per cent, n = 22), Mops condylurus (26.7 per cent, n = 30), Nycteris hispida (0 per cent, n = 18), Scotophilus dinganii (16.7 per cent, n = 12), and Scotophilus leucogaster (38.7 per cent, n = 31). Seasonal differences in the CoV RNA detection rate were associated with and differed depending on certain taxonomic species, families, and suborders (Table 3) . Both Yinpterochiroptera and Yangochiroptera bats were overall more likely to be CoV RNA positive during the wet season, with the observed difference being more pronounced in the latter. Significant seasonal differences were observed in four species; high rates of CoV RNA detections were associated with the wet season and low rates with the dry season for Eidolon helvum, Hipposideros ruber, and Mops condylurus, while it was the opposite for Macronycteris gigas (Table 3) .

While we sampled and tested roughly equal proportions of individual bats and rodents and a significant number of NHPs in this study, all but 2 of the 177 animals that turned out to be positive for CoV RNA were bats. This is not a surprise, but it rather reinforces the notion that bats are the major reservoir for alpha and beta CoVs and generally for viruses with zoonotic potential (Li et al. 2005; Woo 2009; Annan et al. 2013; O'Shea et al. 2014; Anthony et al. 2017; Geldenhuys et al. 2021) . While it has been suggested that under-sampling of rodents, rather than a particularly high prevalence in bats, may be a reason for this pattern, we did not find any evidence to support such a hypothesis in our study population, despite a high number of sampled rodents. However, considering the high diversity of rodent species, our sample is certainly not representative and biased toward those that thrive in a peridomestic environment or that are being hunted for consumption. A caveat is also that while we sampled approximately 4.1 per cent of all bat species, we only sampled 0.6 per cent of rodent species, which could have significant effects if certain species exhibit a higher prevalence of CoVs than others as it seems to be the case in bats (Geldenhuys et al. 2021) . The detection of novel CoV RNA in a civet is certainly an interesting finding, as civets played a key role as intermediate hosts in the emergence of SARS-CoV-1 (Kan et al. 2005) . While civets are not farmed in Cameroon, they are hunted for consumption, which implies human contact, thus posing a potential risk. The detected RNA suggests a close relationship with other carnivore CoVs and thus potentially a low risk for humans (Spillover Risk score of 54 out of 155); however, in the absence of a full genomic sequence and further characterization experiments, this remains to be determined (Grange et al. 2021) .

The CoV RNA we detected belongs to 19 different genetic clusters, of which 8 might represent novel CoV species-6 of these were found in bats, 1 in a civet, and 1 in a shrew. While resource and logistical constraints prevented us from obtaining large or complete genomic sequences from the isolates, these findings indicate that the CoV diversity in African wildlife species, and particularly in bats, is still poorly understood. This is concerning since spillover events are likely to occur, given the close interactions that arise from human housing conditions, hunting practices, ecotourism, and other human behaviors in Cameroon and other parts of Africa. While the emergence of SARS-CoV-1 and SARS-CoV-2 have put the spotlight on Southeast Asia and China, in particular, it is important to keep in mind that the risks are present worldwide and local efforts should be enhanced across the globe to determine how to mitigate these risks (Wolfe et al. 2007; Jones et al. 2008) . CoV diversity poses a significant challenge for surveillance efforts and requires further exploration, but it also provides great opportunities to learn more about the biology and history of CoVs, including that of non-SARS-CoVs, such as HCoV-229E.

We detected RNA of bat CoVs closely related to HCoV-229E in 40 bats, which all belonged to the Hipposideridae family (3.9 per cent, n = 1,035). This finding, along with the fact that previous detections of HCoV-229E-like bat viruses were also almost exclusively associated with Hipposideros bats, supports the hypothesis that this family of bats constitutes the original host family and reservoir of HCoV-229E (like) viruses (Corman et al. 2015) . Further evidence for this is the high diversity that HCoV-229E-like viruses detected in bats exhibit compared to those found in humans or camels, which suggests a long shared evolutionary history with bats.

The observed rather low host plasticity of HCoV-229E-like viruses among bats is noteworthy since host plasticity has been proposed as a predictor for the likelihood of (successful) spillover into humans (Kreuder Johnson et al. 2015) . However, in the absence of our knowledge of HCoV-229E-like viruses actually infecting camels and humans, one would potentially predict that these alpha CoVs might exhibit a lower risk for zoonotic spillover than more promiscuous bat CoVs, such as Kenya bat coronavirus BtKY56 and Eidolon bat coronavirus/Kenya/KY24 (Kumakamba et al. 2021) . In that respect, there seem to be parallels between HCoV-229E and SARS-CoV-1 and SARS-CoV-2, which we believe to be primarily hosted by Rhinolophus bats (Li et al. 2005; Lau et al. 2005; Yip et al. 2009; Yuan et al. 2010; Hu et al. 2017; Paraskevis et al. 2020) . This counterintuitive observation might be the result of a sampling bias, but considering the comparably high number of sampled animals (2,581 bats in this study alone) it is one that can hardly be ignored. Surveillance and predictions of spillover risks particularly of bat CoVs will likely play an increasing role in the aftermath of the SARS-CoV-2 pandemic but will face the challenge of limited data, despite the advances that have been made since the emergence of SARS-CoV-1.

While the zoonotic origin of globally circulating HCoV-229E is undisputed, the route by which it made its way from bats into humans is not clear. Some evidence, such as deletion patterns in the S gene and the open reading frames 4 and 8, suggests that camels may have served as an intermediate host. The hypothesis that an intermediate host was involved would be concurrent with what we know about the origins of other CoVs, including as MERS-CoV (camel) or SARS-CoV-1 (civet) (Kan et al. 2005; Corman et al. 2014 Corman et al. , 2016 Sabir et al. 2016; Forni et al. 2017) . Phylogenetic evidence, however, especially from the comparison of the more conserved genes, does not necessarily support that hypothesis. While the phylogenetic analyses of the most variable S and the N genes does place human and camel isolates into the same branch, this is not the case for the most conserved RdRp gene and the E and Mgenes (Figs 2-4, Supplement 5 ). This pattern would indicate that the transmission of HCoV-229E-like bat viruses into camels (which resulted in the isolates we know to date) may have been unrelated to the spillover that eventually led to the emergence of HCoV-229E. Such a multiple spillover scenario would not be without parallel since CoVs closely related to SARS-CoV-2 were detected in pangolins but overall clearly represent a different, earlier, and unrelated transmission (Zhang et al. 2020; Lam et al. 2020) . Recombination could be a factor in HCoV-229E evolution as has been proposed for this and other CoVs; however, given the partial character of the sequences we generated, we can neither confirm nor reject that this may be the case (Corman et al. 2015) .

Determining the evolutionary history of HCoV-229E thus remains a challenge, not least due to limited and highly biased data available to date. Only 42 HCoV-229E isolates have been sequenced completely or near completely, with 28 derived from North America, 8 from Europe, and 7 from Southeast Asia-but not a single complete sequence from Africa. Similarly, almost all isolates from camels are derived from the Arabian peninsula and only one from Africa (Corman et al. 2016; Li et al. 2017 ). More research into the diversity and origins of HCoV-229E would be highly advantageous for our understanding of CoV evolution, spillover, and adaptation to humans.

The detection of CoVs RNA in bats has been associated with seasonal differences in the past, and we found statistically significant evidence for such a correlation in our data set as well (Anthony et al. 2017; Montecino-Latorre et al. 2020; Kumakamba et al. 2021) . Much like in a recent study from the Democratic Republic of the Congo, we found that animals were more likely to be found shedding CoV RNA in the wet compared to the dry season but that this is species dependent and may be true for some but be reversed for others (Table 3 ) (Kumakamba et al. 2021) . Aside from this overall trend, the findings regarding seasonality from the two studies do not necessarily match up; however, this may be due to different bat species sampled on the one hand and due to different ecological and climate conditions on the other hand. A potential reason for the seasonal differences may be related to the species' birthing seasons since it has been suggested that CoV transmission spikes in bat populations as juvenile bats become susceptible to infection once maternal antibody levels wane (Maganga et al. 2020; Montecino-Latorre et al. 2020) . The observation that young bats in our data set were more likely to be positive for CoV RNA than adults would support this hypothesis, although our finding was not statistically significant. Given that there is not much known about the behavior and reproductive biology of the respective bat species in the specific study area(s) drawing conclusions remains challenging, although much could be learned if studies exploring the behavior and physiology of the bats of interest were to be undertaken.

Interestingly, we found female bats to be significantly more likely to be CoV RNA positive than male bats, which is the opposite of what was found in DRC and what was suspected based on behavioral differences between males and females during breeding, birthing, and breastfeeding seasons (Fayenuwo and Halstead 1974; Kumakamba et al. 2021 ). However, while apparently a contradiction, this might be a reflection of differences in the species composition of the data set and indicate that caution should be used when making generalizations for members of the Chiroptera order, even if climate and habitat are similar.

A key goal of the PREDICT work in general was to collect data about and samples from animals that are actually or potentially in close contact with humans. Part of that objective was to determine what species can be found at the human-animal interface, and consequently the sampling targets were based on animal orders (primarily bats, rodents, and primates) rather than any particular set of species. The sample numbers for individual species are consequently not representative but rather a series of snapshots of what can be found at or near the human-animal interface in Cameroon-a matter that is true for most studies on the CoVs in African wildlife in general (Geldenhuys et al. 2021) . This certainly limits the amount of meaningful statistics that can be done on many of the sampled species, and it would be desirable for future studies that want to investigate connections between ecological or behavioral trades of individual species with CoVs or other viruses to specifically define robust sampling goals for species of interest.

Overall the results of this study on CoVs in African wildlife unveil or highlight several important aspects regarding the risks of future spillover and pandemics: 1) Our knowledge about the diversity of CoVs circulating in wildlife remains limited as exemplified by the fact that almost half of the CoV species we detected had not been described before. Despite having sampled 114 animal species, sample populations for most of them are too small to draw any conclusions about prevalence or risks. The role bats may play as a reservoir is certainly reinforced by the findings, but other species might simply be undersampled. The CoV RNA detection in a civet could be hinting toward a largely undetected CoV circulation in what could be an intermediate host for future spillovers.

2) The evidence for ecological factors such as seasonality driving transmission among bats is increasing. While the evidence remains circumstantial and the mechanisms elusive, it seems to become clear that further studies into this matter could enable smarter surveillance initiatives and mitigation measures, such as limiting access to caves or discouraging hunting during periods of increased virus shedding. 3) The high prevalence and diversity of HCoV-229E-like viruses in African Hipposideridae bats reinforces the notion that HCoV-229E originated in Africa. However, while more and more data supports this origin hypothesis, the question whether or not camels acted as an intermediate host during the spillover into humans remains unclear. Regardless of the intermediate host, or if there was one, HCoV-229E highlights how important it is to not focus on Southeast Asia for CoV surveillance but that CoV pandemics can start in any areas of the world.

Additional data are available in the supplementary materials (Supplements 1-6). All viral sequences obtained are deposited in GenBank. Accession numbers for RdRp amplicons are listed in Table 2 and accession numbers for other partial genomic sequences of HCoV-229E-like viruses (MZ474745-MZ474805) are listed in Supplement 1.

Supplementary data is available at Virus Evolution online.

Global Hotspots and Correlates of Emerging Zoonotic Diseases'

Human Betacoronavirus 2c EMC/2012-related Viruses in Bats, Ghana and Europe

Global Patterns in Coronavirus Diversity

BEAST 2.5: An Advanced Software Platform for Bayesian Evolutionary Analysis

Rooting the Phylogenetic Tree of Middle East Respiratory Syndrome Coronavirus by Characterization of a Conspecific Virus from an African Bat

Zoonoses under Our Noses

Origin and Evolution of Pathogenic Coronaviruses'

Evolutionary Ecology of Virus Emergence

Molecular History of Plague

Genomic Characterization of Severe Acute Respiratory Syndrome-related Coronavirus in European Bats and Classification of Coronaviruses Based on Partial RNAdependent RNA Polymerase Gene Sequences

Identification of a Novel Coronavirus in Patients with Severe Acute Respiratory Syndrome

Human Interactions with Bat Populations in Bombali

Breeding Cycle of Strawcolored Fruit Bat, Eidolon Helvum, at Ile-Ife, Nigeria

Molecular Evolution of Human Coronavirus Genomes

Coronaviruses in South African Bats', Vector-Borne and Zoonotic Diseases

A Metagenomic Viral Discovery Approach Identifies Potential Zoonotic and Novel Mammalian Viruses in Neoromicia Bats within South Africa

Estimating Undetected Ebola Spillovers

Epidemiology, Diagnosis, Treatment, and Control of Trichinellosis

Ranking the Risk of Animal-to-human Spillover for Newly Discovered Viruses

Discovery of a Rich Gene Pool of Bat SARS-Related Coronaviruses Provides New Insights into the Origin of SARS Coronavirus

Evidence Supporting a Zoonotic Origin of Human Coronavirus Strain NL63

Global Trends in Emerging Infectious Diseases'

Molecular Evolution Analysis and Geographic Investigation of Severe Acute Respiratory Syndrome Coronaviruslike Virus in Palm Civets at an Animal Market and on Farms

The Kingdon Field Guide to African Mammals

Spillover and Pandemic Properties of Zoonotic Viruses with High Host Plasticity

Coronavirus Surveillance in Wildlife from Two Congo Basin Countries Detects RNA of Multiple Species Circulating in Bats and Rodents

Identifying SARS-CoV-2-related Coronaviruses in Malayan Pangolins

Severe Acute Respiratory Syndrome Coronavirus-like Virus in Chinese Horseshoe Bats

Interplay between Co-divergence and Crossspecies Transmission in the Evolutionary History of Bat Coronaviruses

Bat-borne Virus Diversity, Spillover and Emergence

Bats are Natural Reservoirs of SARS-like Coronaviruses

A Novel Potentially Recombinant Rodent Coronavirus with A Polybasic Cleavage Site in the Spike Protein

Identification of Diverse Viruses in Upper Respiratory Samples in Dromedary Camels from United Arab Emirates

Genetic Diversity and Ecology of Coronaviruses Hosted by Cave-dwelling Bats in Gabon

Paramyxo-and Coronaviruses in Rwandan Bats

Coronavirus Surveillance of Wildlife in the Lao People's Democratic Republic Detects Viral RNA in Rodents

Bats as Bushmeat: A Global Review

Bats of Central and Southern Africa: A Biogeographic and Taxonomic Synthesis

Reproduction of East-African Bats May Guide Risk Mitigation for Coronavirus Spillover

Evaluation of Bat Adenoviruses Suggests Co-evolution and Host Roosting Behaviour as Drivers for Diversity

Coronaviruses Detected in Bats in Close Contact with Humans in Rwanda

Bat Flight and Zoonotic Viruses

Full-genome Evolutionary Analysis of the Novel Corona Virus (2019-ncov) T Rejects the Hypothesis of Emergence as a Result of a Recent Recombination Event

Cross-Species Virus Transmission and the Emergence of New Epidemic Diseases

Distant Relatives of Severe Acute Respiratory Syndrome Coronavirus and Close Relatives of Human Coronavirus 229E in Bats

The Origin and Prevention of Pandemics

Identification of a Severe Acute Respiratory Syndrome Coronavirus-like Virus in a Leaf-nosed Bat in Nigeria

Detection of New Genetic Variants of Betacoronaviruses in Endemic Frugivorous Bats of Madagascar

MRBAYES 3.2: Efficient Bayesian Phylogenetic Inference and Model Selection across a Large Model Space

Co-circulation of Three Camel Coronavirus Species and Recombination of MERS-CoVs in Saudi Arabia

Market characteristics and zoonotic disease risk perception in Cameroon bushmeat markets

Origins of HIV and the AIDS Pandemic

Rabies-related Viruses

Global Rise in Human Infectious Disease Outbreaks

Genomic Characterization of Seven Distinct Bat Corona-Viruses in Kenya

Surveillance of Bat Coronaviruses in Kenya Identifies Relatives of Human Coronaviruses NL63 and 229E and Their Recombination History

Detection of Novel SARS-like and Other Coronaviruses in Bats from Kenya

Bat Coronaviruses and Experimental Infection of Bats, the Philippines

Cross-species Infections

Bushmeat Hunting, Deforestation, and Prediction of Zoonotic Disease

Origins of Major Human Infectious Diseases

Coronavirus Diversity, Phylogeny and Interspecies Jumping

Phylogenetic Perspectives on the Epidemiology and Origins of SARS and SARS-like Coronaviruses

Intraspecies Diversity of SARS-like Coronaviruses in Rhinolophus Sinicus and Its Implications for the Origin of SARS Coronaviruses in Humans

Isolation of a Novel Coronavirus from a Man with Pneumonia in Saudi Arabia

Probable Pangolin Origin of SARS-CoV-2 Associated with the COVID-19 Outbreak

A Pneumonia Outbreak Associated with A New Coronavirus of Probable Bat Origin

The authors would like to thank the Government of Cameroon for permission to conduct this study; the staff of Metabiota and CRESAR, who assisted in sample collection and testing; and any other involved members of the PREDICT-1 consortium (https://ohi. sf.ucdavis.edu/programs-projects/predict-project/authorship).The study was undertaken as part of the global United States Agency for International Development (USAID) funded Emerging Pandemic Threats (EPT) PREDICT project, which focuses on enhancing the global capacity for the detection and discovery of potentially zoonotic viruses at the human-animal interface. It was made possible primarily by the generous support of the American people through the USAID Emerging Pandemic Threats PREDICT program (cooperative agreement numbers GHN-A-OO-09-00010-00 and AID-OAA-A-14-00102). The contents are the responsibility of the authors and do not necessarily reflect the views of USAID or the US Government. Some funding also came from the National Institutes of Health (NIH) Director's Pioneer Award Program (grant number DP1-OD000370), the International Research Scientist Development Award from the NIH Fogarty International Center (K01 TW00003-1), Google.org, the Skoll foundation, the US Military HIV Research Program, and the Johns Hopkins Bloomberg School of Public Health, Center for a Livable Future.

None declared.